Chapter 6: Nanostructured Metal Oxide Gas Sensors for Air-Quality Monitoring

The development of chemical gas sensors has been the subject of intense academic and industrial research activity for the past 40 years. The increasing need to protect the environment and for more efficient control of industrial processes has stimulated the development of various types of gas sensors. These utilize thick or thin films of metal oxides for the detection of inflammable and toxic gases, solid electrolytes for oxygen sensing and to monitor vehicle emissions and production processes, ceramic materials and organic polymers for humidity detection, and artificial olfactory systems with a wide range of applications for the food industry, and these gas sensors are also utilized for medical diagnosis [1]. Another rapidly growing application area is the use of metal oxide sensors for monitoring air pollutants, such as carbon monoxide, nitrogen dioxide, and ozone.

The first thick-film tin dioxide gas sensors were constructed by Taguchi at the beginning of the 1960s [2] and various types were subsequently marketed commercially. In the following years, research was carried out to improve the performance of these sensors whose operating principle is based on measuring the variation of the electrical resistance of semiconductor metal oxides due to the adsorption of gas molecules on the surface. The materials used in fabrication include zinc oxide for the detection of hydrogen, alcohol, carbon monoxide, and various hydrocarbons; tin dioxide for the detection of inflammable and oxidizing or reducing toxic gases; tungsten trioxide for the detection of hydrogen, hydrazine, ammonia, and hydrogen sulfide; titanium dioxide for the detection of oxygen. Polycrystalline tin dioxide is the material most commonly used in commercial gas-sensor applications because it is stable enough to allow reproducible measurements.

Whether for thick-film or thin-film sensors, the response has generally been optimized empirically, without systematic studies of the parameters that influence performance. Various techniques have been used for the deposition of the gas-sensitive layer. These include ultrahigh vacuum and reactive electron beam evaporation [3–10], chemical vapor deposition (CVD) by thermal, plasma or laser technique [11–13], reactive radio-frequency (RF) or magnetron sputtering [14–26], laser ablation [27–32], the sol–gel technique [32–36], screen printing technology [37, 38], electrochemical deposition [7, 39], and pyrosol methods [40–43]. Various noble metals, such as platinum, palladium, and silver, which due to their catalytic properties improve the sensitivity and selectivity, may in addition be incorporated into the active layer [44, 45].

Future improvements in the response of gas sensors will depend on developing a better understanding of the principles governing their operation and, in particular, the influence of the microstructure and morphology of the semiconducting metal oxide layer on the mechanism of adsorption of gas molecules. The requirement to detect simultaneously the component gases contributing to air pollution has stimulated the integration of multiple sensors in single devices, as well as attempts to improve stability and selectivity. The use of nanocrystalline materials [46] allows further improvement in performance compared with conventional polycrystalline material. Reduction of the grain diameter to values comparable to the Debye length and the extremely high surface-to-volume ratios of the nanocrystalline structure significantly enhance the sensor response.

Nanocrystalline materials are commonly defined as having a mean grain size of less than 100 nm. More generally, nanostructured materials are those, such as a thin film, fiber, or a particle, with at least one dimension below 100 nm or containing atomic domains less than this diameter. Although conventional polycrystalline materials are characterized by grains with dimensions of the order of micrometers containing millions of atoms, a grain of a nanocrystalline material typically contains only a few thousand atoms. Because the surface-to-volume ratio increases rapidly with decreasing diameter, a reduction in grain size results in an increase in the density of grain boundaries and triple points and thus an increase in the fraction of atoms lying in interfaces compared to those at regular lattice positions. For example, in a polycrystalline material with a grain size of approximately 100 nm, only a small percentage of atoms are in grain boundaries, whereas if the grain diameter is reduced to 5 nm, the small percentage of atoms adjacent to grain boundaries becomes similar to that within the grain interior [47]. Such drastic reduction of the grain size causes changes in the physical and chemical properties, such as electrical and electronic, thermodynamic and mechanical behaviors [48, 49].

The sensing mechanism is based on the interaction of gas molecules in the air with the surface grains of the metal oxide layer. The size and the interface structure of the grains influence the sensitivity of the sensor. Detection occurs as a result of complex physical and chemical interactions of the gas molecules with ionized oxygen species adsorbed at the surface of the metal oxide grains. The use of a nanocrystalline material increases the efficiency of detection because of the greater specific surface area available to interact with the gas molecules and the higher density of grain boundaries that provide a network of diffusion paths into the interior of the material.

2. THE GAS-SENSING MECHANISM

A sensor is a device that can detect information from the environment and, by converting the energy associated with that information into another form, provide a usable signal. In the case of metal oxide gas sensors, the presence of a relatively small concentration of gas molecules in the air causes a change in the electrical resistivity of the active layer due to modification of the free carrier density due to exchanges between the conduction band and adsorbed species. For an n-type semiconductor, the presence of a reducing gas, such as carbon monoxide, leads to an increase in the electrical conductivity, while an oxidizing gas, such as nitrogen dioxide, leads to a decrease [2].

Tin dioxide is an n-type wide band gap (3.6 eV) semiconductor with rutile type structure, in which the oxygen vacancies act as free electron donors. The gas detection mechanism is governed by two reactions. The first reaction occurs when oxygen present in the air is chemisorbed on the surface of the oxide as molecular O₂⁻, atomic O⁻, or as hydroxide (OH)⁻ species [3]. The adsorption of oxygen on the surface of a grain causes formation of a depleted region (the space charge region) as a result of transfer of electrons from the conduction band of tin dioxide to the adsorbed oxygen species [1, 4, 50]. This depleted region has a width L, which is dependent on the oxygen species and the electron density in the conduction band and can be determined from [51]

where is the Debye length and ε is the permittivity, n is the electron density, e is the electronic charge, k is the Boltzmann constant, T is the temperature, and V_s is the potential at the surface.

The second reaction occurs between the pollutant gas molecules and the adsorbed oxygen. In the case of a reducing gas like carbon monoxide, this leads to a reduction in the height of the potential barrier at the grain boundaries and a consequent decrease in the electrical resistance of an n-type semiconductor due to neutralization of the oxygen ions adsorbed at the surface. This reaction takes place according to the equations [52]

In the case of an oxidizing gas like nitrogen dioxide, the corresponding equations are

resulting in an increase in the height of the potential barrier and the electrical resistance.

Changes in the resistance of the film due to chemisorption can be directly correlated with the concentration of the gas. The sensor response is defined as R_gas/R_air for an oxidizing gas and R_air/R_gas for a reducing gas, where R_gas is the resistance in the presence of the gas and R_air is the resistance in pure air [53]. The sensitivity S is defined as the percentage change in the resistance so that for an oxidizing gas,

while for a reducing gas,

Both the sensitivity and the selectivity are dependent on the operating temperature and can be modified by doping with a noble metal. A significant increase in sensitivity is obtained when the grain size is reduced to approximately twice the Debye length (~3 nm for tin dioxide) so that the width of the space charge layer is comparable to the grain diameter [6, 54, 55].

The microstructure and the surface structure of the sensing layer exert a significant influence on the sensitivity because they influence the rate of surface reactions and the diffusion of gas into the interior of the material. Göpel and Schierbaum [55] developed models of the dependence of electrical conduction on the grain size of metal oxides by representing the band structure as an electronic equivalent circuit. The grain diameter of polycrystalline tin dioxide is generally greater than the Debye length [6, 45, 56], and if it is more than twice the width of the space charge region, the interior of the grain will make a negligible contribution to the variation in the electrical conductivity of the material, which will be dominated by the grain boundary effects (Fig. 6.1). However, if the grain diameter is comparable or less than twice the Debye length, the space charge region will extend over almost the entire grain and the electrical resistance will, therefore, be more sensitive to the presence of the gas [57].

To be useful, a sensor needs to have a selective response to different gases. The required selectivity can be obtained by varying the operating temperature or the use of catalyst [58]. It must be considered that if the temperature is too low, the reaction between the gas molecules and the oxygen species adsorbed at the surface may be too slow to give an adequate sensitivity. On the other hand, if the temperature is too high, the oxidation reaction will take place so rapidly that the concentration of the gas molecules at the surface becomes diffusion limited. For each individual gas, there is a temperature at which the variation of the resistance is maximized with respect to the resistance measured in pure air. It is thus possible to optimize the sensitivity to a specific gas in a mixture by selecting the appropriate operating temperature.

Figure 6.1 Adsorption of oxygen ions on the surface of SnO₂ grains results in the formation of a space charge zone. Interaction of the adsorbed species with gas molecules changes the height of the potential barrier.

3. EFFECT OF CATALYST AND ELECTRICAL CONTACT MATERIALS

The addition of a noble metal, such as platinum, palladium, gold, or silver, to the structure of the metal oxide increases the sensitivity and the selectivity of the sensor and reduces the optimal operating temperature [58]. Yamazoe [45] has proposed a model to describe the effect of a noble metal catalyst in contact with a semiconducting metal oxide. Both chemical and electronic mechanisms have to be considered. The chemical contribution is due to the so-called spillover effect of the gas molecules on the surface of the oxide. This increases the rate of oxidation of the gas and reduces the concentration of adsorbed oxygen species, while the catalyst dissociates the gas molecules enabling them to react with the adsorbed oxygen species.

An electronic interaction occurs between a noble metal particle and the surface of the tin dioxide grain as a result of the difference between the work function of the metal and the electronic affinity of the semiconductor [59]. This leads to oxidation of the metal and hence a variation in the oxidation state of the tin dioxide and the formation of a depleted region inside the grain. The effect is usually associated with platinum and gold, whereas palladium and silver tend more readily to form the stable compounds palladium oxide and silver oxide [4, 7, 45]. In the case of carbon monoxide, the presence of both platinum and palladium increases the sensitivity, particularly at low temperature (275 °C), and reduces the temperature at the maximum sensitivity from over 350 °C to below 300 °C [7, 33].

The total electrical resistance is the result of the contributions from several components: the semiconductor layer, comprising the surface, bulk, and grain boundary components; the substrate on which the metal oxide film was deposited; the electrical contacts; and any catalyst added to increase the sensitivity and the selectivity. A typical sensor can be represented as an ohmic resistance for which the relation between current and potential is linear. The electrical contacts are metallic and are usually made of platinum or gold. The presence of a metallic contact on the tin dioxide layer causes the formation of a Schottky barrier at the interface between the metal and the semiconductor due to the higher work function of the metal compared to the oxide. The contribution of the contacts can, thus, be represented by a diode with nonlinear current–voltage characteristics.

4. THIN-FILM DEPOSITION METHODS

Two commonly used techniques for the deposition of thin films will be considered: reactive RF sputtering and laser ablation. Reactive sputtering is a method that can be used for many different types of metallic and nonmetallic material [60]. It ensures adhesion between the thin film and substrate and a high degree of control of the composition of the deposited material to alter the stoichiometry as required. The surface of a target (the cathode) is bombarded by argon ions in a vacuum of ~10⁸ torr, forming a plasma containing particles of the target material that are deposited on the substrate (the anode). The energy of the incident ions is transferred to the atoms in the surface of the target by atomic collisions [61, 62]. Ejection of atoms or clusters of atoms from the surface occurs when the energy transferred is greater than the binding energy of the material [63, 64].

The rate of sputtering from the target is proportional to the current for a constant applied potential. The power is typically between 50 and 500W for a potential of 0.5–5 kV. A unique feature of this technique is that the rate of deposition can be kept constant by controlling the argon pressure and the power [65]. There are three different modes of operation: diode sputtering, triode sputtering, and magnetron sputtering, determined by the way in which the potential is applied between the cathode and the anode. However, the present discussion will deal exclusively with magnetron sputtering, which is the system most commonly used in the production of semiconductor devices, integrated circuits, and computer memory chips.

Magnetron sputtering involves the application of a magnetic field perpendicular to the electrical field generated between the anode and the cathode. This magnetic field confines the electrons in the plasma within a region close to the surface of the target. Consequently, the ionic current and the rate of sputtering from the target are increased. A direct current is generally used for conducting materials, while RF sputtering is used for nonconductors and semiconductors. Reactive sputtering occurs when the target material reacts chemically in the presence of a gas in the vacuum chamber. The introduction of a reactive gas, such as oxygen or nitrogen, allows the deposition of thin films of oxides or nitrides. Varying the concentration of the reactive gas allows the stoichiometry of the deposited film to be maintained or modified as desired.

Pulsed laser deposition (PLD) can be used to deposit a wide range of materials [66]. The technique was initially used to produce thin films with CO₂ or Nd:YAG lasers [67], but the rapid development of alternative methods, such as RF sputtering and molecular beam epitaxy (MBE), slowed down further progress in this field. Sputtering became the method of choice for the deposition of metallic and dielectric films, while MBE was extensively applied in the production of semiconductor films for electronic and optical devices. The PLD technique increased in popularity when it was possible to obtain semiconductor thin films of quality comparable to those obtained by MBE. After this, its application for the synthesis of high-temperature superconductors with complex crystal structures marked the beginning of the widespread use of PLD for many different types of metallic, dielectric, semiconductor, ferroelectric, and piezoelectric materials.

The principle of operation of PLD is relatively simple [67, 68]: a laser pulse enters the vacuum chamber via a transparent window and impinges on the material to be deposited. The pulse duration is typically 10–30 nanoseconds with an energy density of 1–10 J/cm. The spot size generally corresponds to an area of approximately 1 mm², and a few tens of nanometers of the surface material are vaporized after each pulse, forming a plume of neutral atoms and molecules, which are deposited on the substrate located at a distance of several centimeters from the target. The deposition rate is typically approximately 0.1 nm per pulse, with a repetition frequency of 1–100 Hz. The plume is highly supersaturated, with a high level of ionization (~50%) and particle energies ranging from a few eV to several keV. This results in a high rate of deposition and a high nucleation density [69].

One of the most important features of the PLD technique is the possibility of depositing films of mixed composition with the same stoichiometry as the target material. Heating of the material surface due to the laser beam causes melting and vaporization, and all the components of the target are simultaneously vaporized [70]. Deposition can be carried out under vacuum at 10⁻⁶–10⁻⁷ torr or in the presence of a reactive gas such as oxygen, which allows the stoichiometry of the deposited film to be precisely controlled and also has an influence on the grain size.

The fundamental parameters for all thin-film production methods are the temperature of the substrate during deposition, the pressure inside the vacuum chamber, and the rate of deposition. Formation of the film takes place by nucleation and grain growth, involving adsorption, surface diffusion, and chemical bonding phenomena. The film is generated from a number of discrete nuclei that grow at the expense of smaller ones. For both RF sputtering and PLD, film formation is due to three-dimensional island growth [70]. Following nucleation on the substrate surface, growth will occur when the arriving atoms or molecules have a greater propensity to form bonds between themselves than with the substrate.

The substrate temperature is extremely important because it has a strong influence on the microstructure of the film. Thornton [71] generalized the models proposed by Sanders [72] and Movchan and Demschishin [73] to relate thin-film microstructures to the substrate temperature and the melting point of the material. The substrate temperature influences the variation of the free energy associated with the formation of nuclei and the surface diffusion of the deposited atoms. A reduction in this temperature leads to a lower bulk free energy with the result that the critical radius needed to obtain a stable nucleus is smaller. A reduction occurs in the coefficient of surface diffusion of the atoms adsorbed on the substrate that affects the rate of nucleation.

The rate of deposition, as well as the temperature, has a strong influence on the film microstructure [74] because it determines the rate of nucleation. For a given substrate temperature, a lower deposition rate will result in a lower rate of nucleation and a longer mean free diffusion time of mobile atoms on the surface. These conditions favor the formation of well-ordered structures that often exhibit preferential growth directions determined by the orientations of the low-energy crystal surfaces. The films are generally not very dense and show pronounced surface roughness. In the case of a higher rate of deposition, growth takes place rapidly by agglomeration at the initial binding sites due to the continual arrival of new atoms. The atoms do not have sufficient time to migrate to occupy the most energetically favorable sites, and the resulting film microstructures are characterized by a very small grain size and a greater number of lattice defects. The films tend to be denser and their surfaces are generally smoother than those deposited at lower deposition rates.

The presence of a reactive gas in the vacuum chamber influences the rate of deposition and hence the grain size. These gas molecules reduce the energy of the flux of particles from the plasma incident on the substrate and thus reduce the deposition rate and the nucleation rate. At higher partial pressures, fewer nuclei of critical size are formed on the substrate, and the growth of these larger nuclei is favored. At low partial pressures, the deposition rate remains relatively high and favors the formation of smaller sized grains. The surface of the substrate itself may, in addition, influence film growth and structure. Lattice defects, such as dislocations and point defects, represent nucleation sites of lower energies. High surface roughness can induce a topographic shadowing effect, impeding the growth in the affected areas of the film [75, 76].

PLD has one major difference with respect to RF sputtering: deposition is intermittent rather than continuous and is less stable and more rapid, so it can be expected to produce less ordered film structures. Another typical feature of thin films deposited by PLD, due to the high degree of supersaturation of the vapor plume and the high rate of deposition, is their rather small grain size. Subcritical size nuclei formed on the substrate surface during a laser pulse will tend to dissociate into mobile species during the interval preceding the next pulse, allowing them to renucleate by forming new agglomerates or to be incorporated into existing nuclei with greater than the critical radius.

The ratio between the duration of the pulse and the interval between successive pulses influences both the nucleation and the growth processes. A long pulse will vaporize a greater amount of material from the target, resulting in the formation of a high number of small-sized grains. However, if the pulse length is short, the amount of vaporized material is limited and produces a small number of critically sized nuclei, which grow by diffusion and agglomeration during the intervals between pulses. The pulse length and repetition frequency must therefore be carefully chosen to obtain the required grain size [66].

5. INFLUENCE OF FILM STRUCTURE ON SENSOR RESPONSE

5.1. Radio-Frequency Sputtering

Figure 6.2 shows an array of gas sensors fabricated by depositing a tin dioxide film of a few hundred nanometers thickness on a polycrystalline alumina substrate by reactive RF magnetron sputtering from a 99.9% pure tin dioxide target at 250 °C [77–80]. The active area of each sensor is approximately 1 cm². Alumina is commonly used as the substrate material for thin-film gas sensors because it is an electrical insulator and, therefore, does not interfere with measurements of the film resistance. Film deposition was carried out at a pressure of 50 Pa (3.75 mtorr) with an argon/oxygen ratio of 9:1. A platinum resistance heater and the electrical contacts were deposited by the same method in a pure argon atmosphere. In some cases, the same procedure was used to deposit a 10–20 nm thick layer of platinum to catalyze the reaction with adsorbed gas molecules to increase the sensitivity and the selectivity. Following deposition, a thermal annealing treatment was performed to improve the crystallinity of the oxide film.

The films have a nodular surface structure, in which it is possible to distinguish small granular particles of a few tens of nanometers in diameter (Fig. 6.3). Their surface morphology replicates that of the underlying alumina substrate, and the resulting surface roughness increases the effective area of the oxide in contact with the air [77]. These films are nanocrystalline with grain diameters ranging from a few nanometers to a few tens of nanometers (Fig. 6.4). The sputtered platinum catalyst has mean grain size of approximately 10 nm [79]. The films deposited at a temperature of 250 °C have a compact columnar structure, in contrast to those formed at room temperature, which have an equiaxed grain structure containing spherical pores of ~5 nm diameter that are especially evident in the first 200 nm or so from the substrate. The deposition temperature appears to influence the grain size less than its influence on the growth structure of the film.

Figure 6.2 Thin-film gas sensors fabricated on a polycrystalline alumina substrate by the RF magnetron sputtering technique: (A) nanocrystalline tin dioxide layer with platinum contacts; (B) platinum heater.

Figure 6.3 Surface structure of a tin dioxide film deposited by reactive sputtering at a substrate temperature of 250 °C.

The mean grain size can be estimated from the peak broadening in a glancing angle X-ray diffraction (GARDX) spectra (Fig. 6.5), by applying the Scherrer equation [81]

where D is the mean crystallite diameter, K is the shape factor (~0.9), λ is the X-ray wavelength, β is the line width at half maximum intensity (FWHM), and θ is the Bragg angle. For the RF sputtered films, this lead to values of 3–10 nm [82], compared with that of approximately 10 nm estimated by dark field transmission electron microscopy [21]. This is reasonably close to the optimal grain diameter of approximately 6 nm imposed by the Debye length criterion. It can be concluded that electrical conduction in these nanocrystalline semiconductor films can be expected to be dominated mainly by grain boundary effects.

Figure 6.4 Nanocrystalline microstructure in RF sputtered tin dioxide: (A) bright-field transmission electron microscopy and selected area electron diffraction pattern; (B) dark-field transmission electron microscopy using the 110 reflection.

Figure 6.5 Glancing angle X-ray diffraction spectrum obtained from a tin dioxide film deposited by reactive RF sputtering.

High-resolution electron microscopy studies (Fig. 6.6) indicate that grain boundaries are generally well defined and that planar defects and dislocations are rarely present [78, 79]. Dissociated and undissociated {101}<101> dislocations have, however, been observed in nanocrystalline tin dioxide [83] produced by electron beam evaporation. The composition at the cores of dissociated dislocations differs from the exact stoichiometric composition and widely separated pairs of them form oxygen-deficient {101} crystallographic shear planes (CSP) that act as sinks for oxygen vacancies. Nonstoichiometric dislocation cores and CSP thus act as traps for free electrons, influencing the electrical conductance of the film. Free carrier densities corresponding to oxygen vacancy concentrations in the range 10¹⁸–10¹⁹ cm⁻³, consistent with a slightly understoichiometric composition, have been determined in nanocrystalline films deposited by the RF sputtering technique by Hall Effect measurements [17]. The free carrier mobility µ_H may vary between 0.1–10 cm²/V/s depending on the film structure. Values of µ_H for the compact columnar films deposited at 250 °C were 2–3 cm²/V/s.

The deposition of platinum, either on the surface of the film or between two layers of tin dioxide, results in a metallic layer with a well-defined interface with the oxide. The nonohmic contact between the metallic layer and the semiconductor leads to the formation of a Schottky barrier at the interface [59]. The presence of platinum, in or on the surface of the oxide film, influences the sensitivity of the sensor as a result of the spillover effect due to the platinum and the existence of the contact potential between the metal and the semiconductor. The difference between the work function of platinum (5.6 eV) and the electron affinity of tin dioxide (4.5 eV) generates a potential barrier that causes nonlinear behavior of the conductivity. There is, therefore, an increase in the conductivity of the material with respect to the oxide alone, particularly in the case of films deposited at higher temperatures. This is a result of the columnar structure, which provides more continuous contact with the platinum grains compared to the relatively porous structure of the films deposited at room temperature [80].

Figure 6.6 High-resolution transmission electron microscopy showing lattice plane contrast and grain boundaries in a nanocrystalline tin dioxide film.

The sensitivity to carbon monoxide is greatest at temperatures between 300 °C and 350 °C for tin dioxide films without platinum, while below 250 °C it is negligible. The response (R_air/R_CO), determined by measuring the resistance while increasing the concentration in pure air, is shown in Fig. 6.7 for a sensor with a 12-nm surface layer of platinum. It can be noted that the response is both rapid and almost completely reversible. The presence of the noble metal increases the response from 2.5 with an oxide film alone to 3.7 with the platinum surface layer, for a carbon monoxide concentration of 20 ppm, and reduces the operating temperature required to attain maximum sensitivity.

Figure 6.7 Response to carbon monoxide in pure air of a tin oxide sensor with a thin platinum surface layer operated at a temperature of 250 °C.

For practical applications, it is essential to understand the effects of interference between the various gases present in the air on the overall sensor response. When adsorbed on the surface of the oxide, carbon monoxide acts as an electron donor, whereas nitrogen dioxide acts as an electron acceptor and they, therefore, have opposite effects on the conductance. The effect of a small concentration (0.1 ppm) of nitrogen dioxide in the air on the response of a tin dioxide sensor without a platinum catalyst layer is shown in Fig. 6.8. By comparing the pairs of measurements made at three different operating temperatures, it can be seen that while at 250 °C the effect due to the nitrogen dioxide dominates the response, the interference decreases with increasing operating temperature and is almost negligible at 350 °C. High selectivity can, thus, be achieved by variation of the operating temperature.

Figure 6.8 Response of a tin dioxide sensor without a platinum catalyst, at three operating temperatures, to different concentrations of carbon monoxide in the presence of 0.1 ppm of nitrogen dioxide.

The results of measurements made with two different sensors over a range of carbon monoxide concentrations in pure air, at an operating temperature of 300 °C, are shown in Fig. 6.9. Deposition of the tin dioxide layers was carried out at 250 °C (S1) and room temperature (S2). The film deposited at 250 °C has noticeably greater sensitivity than that of the film deposited at room temperature. This is due to the superior crystallinity and the columnar structure in the film deposited at higher temperature. The oxide film deposited at 250 °C also had higher electrical resistance in pure air, i.e., approximately 1⁶ Ω compared with approximately 2⁵ Ω for the one deposited at room temperature [78].

The sensitivity of tin dioxide films deposited at room temperature can be significantly improved by the addition of a thin layer of platinum. This effect is most pronounced when the noble metal layer is deposited between two layers of oxide (Fig. 6.10) so that two metal/oxide interfaces are formed. The presence of the platinum generates a higher concentration of adatoms at the interface with the oxide [84, 85] and thus a higher free carrier density in the semiconductor, which leads to an increase in the sensitivity. The platinum acts to catalyze the rate of reaction between the tin oxide and the gas molecules, without altering the work function. The superior performance of the films with an internal layer of platinum confirms that the reaction is not limited solely to the surface of the film but that substantial diffusion of gas molecules also occurs within the oxide.

Figure 6.9 Sensitivities of tin dioxide films deposited at 250 °C (S1) and room temperature (S2) to various concentrations of carbon monoxide at an operating temperature of 300 °C.

Figure 6.10 Sensitivity to different concentrations of carbon monoxide of a tin dioxide film alone (SnO₂), with an internal platinum layer (SnO₂-PtIL), and with a surface platinum layer (SnO₂-PtSL).

5.2. Pulsed Laser Deposition

The pulsed laser technique has been used to deposit tin dioxide thin films on the same type of alumina substrate using a KrF excimer laser operating at a wavelength of 248 nm [86, 87]. A pulse length of 12 ns was used with a repetition frequency of 39 Hz. The substrate temperature was varied between 25 °C and 600 °C at a constant oxygen pressure of 2 kPa (150 mtorr). In addition, the oxygen pressure was varied between 1 and 200 mtorr at a constant temperature of 300 °C to investigate the effect of pressure during deposition. In-situ doping of the oxide layer was carried out by a co-deposition method by attaching strips of platinum to the tin dioxide target. An annealing treatment was performed after deposition as in the case of the RF sputtered films.

The structure of thin films of tin dioxide produced by the PLD technique has a marked dependence on the substrate temperature during deposition. It shows a gradual transition from an equiaxed granular porous structure to a compact columnar one with increasing temperature [86]. At room temperature, the mean grain diameter is approximately 4 nm, whereas at a temperature of 300 °C, it increases to 12 nm, when individual crystallites begin to agglomerate in clusters of similar orientation and the film starts to take on a more columnar appearance with its porosity aligned perpendicularly to the substrate surface (Fig. 6.11). The columnar structure becomes more pronounced at 450 °C, though residual equiaxed grains persist, until the structure of the film becomes completely columnar at a deposition temperature of 600 °C.

Figure 6.11 Cross-sectional scanning electron microscope image of a tin dioxide film deposited on a polycrystalline alumina substrate, at a temperature of 300 °C, by the pulsed laser method. Inset shows a transmission electron microscope bright-field image of the same film for comparison.

In-situ doping with small amounts of platinum by the co-deposition method resulted in small particles of metallic platinum evenly dispersed in the films. The overall appearance of these films was similar to that of the undoped films deposited at the same substrate temperature. However, the films produced by the PLD technique contained a higher density of dislocations and planar defects compared to the same material deposited by sputtering.

A small partial pressure of oxygen in the vacuum chamber during deposition is required to maintain the correct stoichiometry in the growing film. This plays a fundamental role in the nucleation of tin dioxide while preventing the precipitation of pure tin or monoxide [88]. X-ray diffraction of films deposited under vacuum indicates the presence of both polycrystalline tin dioxide and amorphous tin oxide [86]. The latter disappears only at a higher deposition temperature (600 °C) where oxidation of the film can easily occur. The mean grain size remains constant at approximately 4 nm with increasing partial pressure of oxygen up to 50 mtorr but increases to 8 nm at 200 mtorr (Fig. 6.12). Films produced at low partial pressure of oxygen have compact columnar structure, with columns of 10–15 nm width consisting of small equiaxed or slightly elongated nanocrystals. Films deposited at higher partial pressures of oxygen have granular, porous structure and a larger mean grain size, with individual grains tending to form clusters and the porosity perpendicular to the substrate.

The resistivity decreases approximately from 200 to 10 Ω cm for films deposited at room temperature and 300 °C and thereafter remains essentially constant with further increase in substrate temperature [89]. The high resistivity of films deposited at room temperature is due to their low density and highly porous structure. Only by using a deposition temperature above 150 °C it is possible to obtain more conductive films. The electrical resistivity reaches a shallow minimum at a deposition temperature of 450 °C. This is possibly because the hybrid structure of films deposited at this temperature tends to be more homogeneous than that of the completely columnar films produced at higher temperature, in which the voids separating the neighboring columns can be expected to reduce the conductivity due to poor electrical contact.

Comparison of the sensitivity to carbon monoxide of films deposited at 300 °C under an oxygen partial pressure of 150 mtorr with that of in-situ doped films containing small amounts of platinum indicated that 2 at% gave the best performance over a range of gas concentrations from 20 to 500 ppm, at operating temperatures from 150 to 350 °C [86]. The presence of the platinum also reduced significantly the temperature at which the maximum sensitivity was obtained from 300 to 200 °C.

Figure 6.12 Dependence of the mean grain size of films deposited by PLD at 300 °C on the partial pressure of oxygen during deposition.

The variation of the resistivity in pure air with the partial pressure of oxygen is shown in Fig. 6.13. The films deposited at low partial pressure of oxygen are quite good conductors due to their microstructural characteristics. The extremely small mean grain size and the absence of porosity in these films facilitate electrical conduction in the material. In fact, the grain radius (~2 nm) is smaller than twice the Debye length for tin dioxide (~3 nm).

Electrical transport is, therefore, no longer controlled by the grain boundary contribution but by the material in the interior of the grains because the space charge region occupies the entire grain, and the potential barrier at the interface between adjacent grains is rather low. Reduction of the grain size, therefore, allows transport of electrons from one grain to the next with a negligible influence due to the presence of the boundary. The formation of agglomerates with similar crystallographic orientation constitutes an additional factor that facilitates electrical conduction. As the partial pressure of oxygen is increased, the resistivity increases as a result of the larger grain diameter and the intercolumnar porosity within the film itself.

It can, therefore, be concluded that the grain size is a very important parameter influencing the performance of tin oxide gas sensors. A grain diameter of approximately 10 nm appears to be sufficient to preserve the semiconducting properties of the oxide; however, if the grain size is less than 2L, the material begins to behave more like a conductor.

Figure 6.13 Dependence of the resistivity at room temperature of tin oxide films deposited by PLD at 300 °C on the partial pressure of oxygen during deposition.

The presence of water vapor causes an increase in the conductivity of tin oxide due to the adsorption of water molecules on the surface. Adsorption and desorption of (OH)⁻ species occur at temperatures of approximately 150–160 °C [90], while the optimal temperature for the detection of carbon monoxide is 300–350 °C, at which these (OH)⁻ species will be already completely desorbed. It is, therefore, important to take into consideration the effect of humidity in the case of detection of NO₂, where the optimal temperature for detection is less than 200 °C.

6. INTEGRATED SOLID-STATE SENSORS

Microelectronic fabrication techniques can be used in combination with thin-film synthesis methods to produce integrated gas microsensors that are fully compatible with CMOS technology. The advantages include cost reduction, device miniaturization, lower power consumption, and the capability to construct multisensor arrays [91]. The use of silicon micromachining methods enables thermally isolated structures to be fabricated to reduce heat dissipation from the metal oxide sensing elements. Thermal isolation can be achieved by depositing the sensing layer on thin free-standing membranes fabricated by micromachining methods. However, it is essential to optimize the membrane dimensions and heater design to ensure a uniform temperature distribution over the sensitive region and to reduce power requirements.

Figure 6.14 shows an integrated sensor array consisting of eight microsensors on a single silicon chip. The sensing elements have dimensions of 0.5 × 0.5 mm, and the entire device is mounted in a standard TO-8 package. Each one of these microsensors can be operated at a different temperature to make it selective to a specific gas. As has already been discussed, metal oxide gas sensors need to be heated during operation to allow the discrimination of different gases, and precise temperature control is, therefore, required to maximize the selectivity. Finite element calculations were used to optimize the design to achieve the correct mechanical and thermal properties for the heater and other components of the device [92]. Silicon nitride was chosen in preference to silica as the membrane material, despite its greater thermal conductivity, because it allows the use of a thinner membrane, resulting in lower power consumption. The tin oxide thin film was deposited by reactive sputtering because this process is compatible with microelectronic fabrication technology.

Figure 6.14 A solid-state integrated microsensor array consisting of eight tin dioxide sensing elements on a single silicon chip.

Figure 6.15 Schematic diagram of an individual microsensor element showing details of its construction.

The construction of an individual microsensor is illustrated schematically in Fig. 6.15. It consists of a tin oxide sensing layer with platinum electrodes, a silica layer for electrical insulation, and the polysilicon heater, all of which are thermally and electrically isolated from the substrate by a thin silicon nitride membrane. The entire device is fabricated on a 300-µm thick P-type <100> silicon chip. Details of the various technological processing steps involved in the fabrication of this device can be found elsewhere [93]. The power consumption is less than 50 mW at an operating temperature of 350 °C.

A cross-sectional transmission electron micrograph of a microsensor (Fig. 6.16) indicates that the thickness of the nanocrystalline tin dioxide layer is approximately 300 nm in this case and the mean grain size of the material is approximately 10 nm. Evidence is visible of a columnar type growth structure, with columns approximately 100 nm in width. The intermediate silica layer is 450 nm thick and completely amorphous and, acting as an insulator, provides a smooth surface on which to deposit the oxide film. The polysilicon heater is approximately 600 nm thick and is polycrystalline with a grain diameter of approximately 200–400 nm, although the grains are noticeably elongated in the direction parallel to the substrate. These structures are supported on a 300-nm thick amorphous silicon nitride membrane, produced by low-pressure CVD (LPCVD), which has low internal stress and good adherence to the polysilicon layer.

The multisensor array is capable of high sensitivity and fast response times, allowing the detection of low concentrations of aromatic hydrocarbons, carbon monoxide, and nitrogen dioxide. The experimental data shown in Table 6.1 are the results of laboratory tests for the detection of low concentrations of nitrogen dioxide in pure air using sensor elements with tin oxide layers of different thicknesses [94]. The sensitivity increases significantly with increasing film thickness for concentrations of nitrogen dioxide in the 0.5–1.5 ppm range. The times for the sensor to reach equilibrium and for recovery after exposure to the gas were similar and were typically less than 3 min.

Figure 6.16 Cross-sectional transmission electron micrograph through a microsensor element showing the structure of the nanocrystalline tin dioxide layer, the silica insulating layer, the polysilicon heater, and the amorphous silicon nitride membrane.

The performance of metal oxide gas sensors is greatly influenced by the grain size and the surface morphology of the active layer because these factors determine the effective surface area on which the adsorption of gas molecules can occur. Nanocrystalline films with a porous columnar structure have higher sensitivities and lower optimal operating temperatures than those with larger grain sizes [95]. Films deposited by RF sputtering generally have finer grain sizes and a higher degree of crystallinity than those produced by other techniques such as spraying [96] or chemical vapor deposition [97].

Other configurations of integrated tin oxide gas sensors have been developed that are completely compatible with standard CMOS fabrication methods [98, 99]. Procedures such as silicon etching to create the thermally isolated microhotplate and the deposition of the oxide film by RF sputtering are carried out as postprocessing steps. Devices with a 1-nm platinum layer deposited on a 300 nm thick layer of oxide to increase the selectivity were tested for their response to carbon monoxide concentrations of 200–1000 ppm over an operating temperature range 200–300 °C [100]. The response was fastest at operating temperatures of 250–300 °C and was dependent on the gas concentration; in general, higher concentrations gave the shortest response times.

Table 6.1 Sensitivity to low concentrations of nitrogen dioxide in pure air for tin dioxide films of three different thicknesses at an operating temperature of 250 °C

7. THICK-FILM TECHNOLOGY

The original gas sensors were constructed using thick films of tin dioxide. Although thin-film technology has gained in popularity among research workers, due to the ease with which the microstructure can be controlled simply by adjusting the deposition conditions, many commercial gas sensors nevertheless still use thick-film technology based on wet chemistry powder synthesis methods and screen printing technology. Some examples of the application of the older technology in the construction of metal oxide sensors for air-quality monitoring will now be considered.

Nanosized powders of several metal oxides and a mixed oxide were prepared either by a sol–gel process using alkoxide precursors or, in the case of the mixed oxide, by a thermal decomposition of the corresponding hexacyanocomplex [101]. Thick-film sensors were fabricated by screen printing on thin alumina substrates that had a heating element and interdigitated gold electrodes for making the resistance measurements. The thick films made in this way were still extremely porous even after firing at 850 °C, a temperature that was selected to maintain the particle size within the 30–50 nm range.

The performance of sensors fabricated from these synthesized oxide nanoparticles and commercially available nanopowders was tested in extensive field trials [101, 102], in parallel with measurements made conventional air-quality monitoring systems. By using the sensors based on different oxides, it is possible to overcome the problem of interferences in the presence of different gases. The sensors based on tin dioxide and tantalum-doped titanium oxide films were sensitive mainly to carbon monoxide. In contrast, the sensor based on a perovskite-type lanthanum iron oxide, a P-type semiconductor, prepared by thermal decomposition, was sensitive principally to nitrogen dioxide. By comparing the sensor data to measurements made by standard analytical equipment at a conventional monitoring station, calibration curves were generated that allowed the sensor response to be directly related to the concentrations of both these gases.

In the light of the results obtained from these preliminary experiments, a system for air-quality monitoring was developed [103], containing an array of five different sensors based on nanograin films of tin dioxide, tantalum, and niobium-doped titanium dioxide, perovskite-type LaFeO₃, and indium oxide (for the detection of ozone). Addition of the metallic ions to titanium dioxide is necessary to maintain the nanocrystalline structure by inhibiting sintering during firing [104]. The indium oxide film, which was made with a commercially available nanopowder, had a somewhat larger particle size than the others, which was in the 200–400 nm range.

In general, the sensor data corresponded quite closely to the evolution with time of the concentration of pollutant gases recorded by the nearby conventional monitoring stations. Figure 6.17 shows the readings obtained from the tin dioxide sensor superimposed on the variation of the level of carbon monoxide measured by the standard instrumentation. The sensor signal was recorded every minute, but values were averaged over an hour to facilitate comparison with the data obtained using the standard analysis methods. The correspondence between the signal from the LaFeO₃ mixed oxide sensor and the true concentration of nitrogen dioxide was less good because this device was also sensitive to other reducing gases, such as volatile organic compounds.

The main sources of error in the sensor measurements are due to the influence of ambient temperature and humidity. These effects have to be taken into consideration to compensate the sensor response, which is affected by the temperature of the surroundings independently of the operating temperature of the sensor itself [105]. Adsorption of water molecules on the surface of the oxide leads to the formation of hydroxyl groups, which cause changes in the oxygen affinity and that act as donors. At low gas concentrations and low operating temperatures, reactions with these groups will tend to dominate the sensor response [106]. It is, however, the partial pressure of the water vapor rather than the relative humidity that determines the sensor response, and the true gas concentration can be derived from an equation of the form [107]:

Figure 6.17 Response of a tin dioxide sensor compared with the concentration of carbon monoxide measured in air by standard monitoring equipment. Reproduced with permission from Carotta MC et al. [103]

where CO_conc is the concentration of carbon monoxide, p is the water vapor pressure, G is the conductance, G₀ is the conductance in the absence of the gas and water vapor; A, B, C, α, and γ are constants obtained by fitting experimental values for different concentrations of CO and different water vapor partial pressures.

Another thick-film gas sensor developed for air monitoring consists of three thick-film sensors on microhotplates, which are monolithically integrated with digital temperature controllers and the interface circuitry on a single silicon chip [108, 109]. The surfaces of the microhotplates are coated with tin dioxide to create the gas-sensitive layer. A transistor heater was used, rather than a resistive heating element, to reduce the overall power consumption and allows operation at temperatures up to 350 °C. One of the sensors is coated with a pure tin oxide layer, whereas the other two are doped with 0.2 and 3 wt% palladium. The performance of the device was assessed with mixtures of the environmentally relevant gases: carbon monoxide; methane; and nitrogen dioxide. The undoped tin dioxide sensor was selective to nitrogen dioxide, whereas the sensors doped with 0.2 wt% and 3 wt% palladium, respectively, were selective to carbon monoxide and methane.

Extensive testing was carried out [110] to evaluate the calibration accuracy, minimum detection limit, sensitivity, selectivity, long-term stability, cross sensitivity to other gases, and the effects of interference due to humidity. The optimal operation temperatures were 275 °C for the detection of carbon monoxide, 300 °C for the detection of nitrogen dioxide, and 300 °C for the detection of methane. Although the overall performance of the device was generally good, problems were experienced with calibration drift after extended periods of operation. The cross sensitivity to humidity and volatile organic compounds could be compensated, but measurements with binary mixtures of carbon monoxide and methane gave poor results due to the influence of calibration drift. Cross sensitivity to ozone was also a difficulty in measurements of nitrogen dioxide because the sensitivity is higher to ozone than that to nitrogen dioxide. These cross-sensitivity problems could be resolved to some extent by modulation of the operation temperature.

During operation in the field, the complex mixture of gases present in real air can lead to significant systematic and random errors. Standard analytical instruments were, therefore, used to provide a direct calibration of the sensor data because these errors prevented the use of the laboratory calibration. Temperature modulation of the sensor hotplates was not as useful in resolving the problem of analyzing multicomponent gas mixtures as it had been in the laboratory tests. Measurements made at constant temperature gave results in closer agreement with the standard reference methods. Because the conventional analytical instruments provide concentrations that are averaged over 10-min periods compared to readings every second for the sensor system, there were difficulties in correlating the noisy raw data from the sensors to these smoothed averaged data. This problem is even more evident when temperature modulation is used because the rapid variation of the temperature prevents the metal oxide layer from reaching chemical equilibrium, resulting in non-steady state operation.

8. INNOVATIVE METAL OXIDE ARCHITECTURES

As already discussed, the sensitivity of metal oxide gas sensors is dependent on various factors, such as the grain size, porosity, surface morphology, and dopant levels. To some degree, the sensitivity can be increased by tailoring the structure of the metal oxide layer by careful control of the synthesis conditions, particularly when thin-film techniques are used for fabricating the sensor. The interaction of gas molecules with the semiconducting oxide is primarily a surface phenomenon. Further improvements in sensor performance are, therefore, likely to be due to the development of novel film architectures that allow the gas molecules access to much larger surface areas of metal oxide and the integration of these structures in miniaturized integrated devices.

Varghese and Grimes [111] have reviewed some of the currently available routes for the production of metal oxide nanoarchitectures with increased surface areas for gas sensing applications, including mesoporous thin films, nanowires or nanobelts, and nanotubes. Among the techniques considered were mesoporous thin films produced using a sol–gel process, titania nanowires formed using a highly ordered nanoporous alumina membrane as a template, and arrays of titania nanotubes formed by anodization of titanium. The use of nanoporous alumina for humidity sensing and titania nanotubes as hydrogen sensors were considered examples. An excellent response can be achieved, even in the low humidity regime, with a layer of pores of uniform nanodimension in alumina containing adsorbed anionic impurities, while titania nanotubes are highly sensitive to low concentrations of a reducing gas.

Quasi one-dimensional nanobelts of semiconducting oxides were first synthesized by a high-temperature evaporation technique [112]. They are highly pure, monocrystalline, uniformly structured, rectangular in cross section, and generally defect free, with widths of 30–300 nm and lengths of up to several millimeters. This type of structure is ideal for studying dimensionally confined transport in semiconducting metal oxides and offers interesting possibilities for the construction of functional devices. Because of the high specific surface area, surface-related effects, such as catalytic activity and adsorption, are enhanced, and due to the elimination of the grain coalescence and the drift in electrical response typical of polycrystalline materials, they are very promising materials for the development of a new generation of metal oxide sensors with increased stability [113].

The synthesis of metal oxide nanowires can be carried out by thermal decomposition of precursor powders, followed by vapor–solid or vapor–liquid–solid growth. The materials that have been prepared by this method include tin dioxide [114–117], zinc oxide [117], indium oxide [116–119], and tungsten oxide [120]. Figure 6.18 shows a scanning electron micrograph of indium oxide nanobelts that were deposited on an alumina substrate. By varying the temperature of the substrate from 800 to 1100 °C, different crystalline structures could be obtained, ranging from coarse equiaxed grains to nanowires and nanobelts. By seeding the substrate with sputtered particles of pure metallic indium, controlled growth of the nanowires can be achieved along the <100> crystal directions.

Figure 6.18 Indium oxide nanobelts.
Reproduced with permission from Bianchi et al. [118]

Measurements of the electrical characteristics of tin dioxide and indium oxide nanowires at a temperature of 400 °C, in the presence of ozone concentrations of a few hundred ppb, indicated an appreciable sensitivity, whereas the response of zinc oxide nanowires was far poorer [117]. At a temperature of 200 °C, indium oxide nanowires with widths less than 100 nm also showed much better response to nitrogen dioxide concentrations of a few ppm than similar wires of width 500 nm [118]. This behavior can be attributed to their reduced lateral dimension that allows completely depletion of carriers. Three-dimensional orthogonal networks of tungsten oxide nanowires synthesized by thermal evaporation of tungsten powder in oxygen at 1400–1450 °C [120] showed a high sensitivity to nitrogen dioxide, for measurements made at 300 °C over a concentration range 50–1000 ppb, but had only a rather weak response to carbon monoxide at the same temperature. This is, therefore, a promising material for the development of nitrogen dioxide sensors with low cross sensitivity.

Attaching contact electrodes to nanowires is a major technical challenge, and to overcome this difficulty, it has been proposed to use changes in the visible photoluminescence of metal dioxide nanowires due to the presence of nitrogen dioxide, rather than the electrical properties [117, 121]. The adsorbed surface species generate surface states that quench the photoluminescence, and the magnitude of the effect depends on the size and shape of the nanostructure, the crystal surface, and the surface oxygen vacancies. It was possible experimentally to discriminate nitrogen dioxide concentrations in the range 0.1–20 ppm using tin dioxide nanowires at 120 °C. Interferences due to water vapor and reducing gases are negligible, and for zinc oxide, the quenching is maximum at room temperature. This leads to the possibility of applying metal oxide nanowires for the construction of low-power selective optical gas sensors operating at room temperature.

9. SENSOR NETWORKS FOR AIR MONITORING

The integration of metal oxide sensors in CMOS-compatible components allows the construction of inexpensive, portable air-quality monitoring systems based on gas-sensor arrays. A prototype of such a system with high precision and a wide dynamic range has been designed [122] using an array of metal oxide thin-film sensors on micromachined substrates, an integrated front-end with an analog-to-digital converter and temperature control unit, and a data processing system to allow information extraction from acquired data. The sensors are fabricated by spin coating a silicon wafer that contains hundreds of miniature devices, with a 100 nm layer of tin dioxide nanopowder prepared by the sol–gel technique. The gas concentration is directly proportional to the variation in resistance of the sensor. Tests of the system carried out with various concentrations of several volatile organic compounds demonstrated its effectiveness and precision.

Such sensor systems could act as nodes in sensor networks for real-time air monitoring with fast response capability and high spatial resolution. The use of solid-state sensors and appropriate signal processing techniques, in combination with data transmission via the existing telecommunications infrastructure, would enable the creation of intelligent sensor networks able to provide highly localized information on air pollution. A system for monitoring nitrogen dioxide was developed based on a commercial tin dioxide gas sensor [123], and its performance was evaluated by comparing the sensor readings with the concentrations measured at a nearby fixed monitoring station. The network connectivity can also be used for automatic calibration of the sensors, as well as for data collection. Nitrogen dioxide concentration can be determined with reasonable accuracy in this way provided that adequate compensation for temperature and humidity is applied. Among various factors that influence the sensor response, such as interferences from other gases and humidity, baseline drift is the major source of error. This means that the system has to be recalibrated as often as necessary to maintain its accuracy.

Installation of a sensor network in a densely populated urban area can allow the spatial and temporal variation of air pollutants to be determined and correlation of this data with other types of information, such as health effects. Solid-state sensors can generate a data point every second for each sensor, creating problems in processing and analyzing such large quantities of data. Distributed wireless sensor networks can be used together with grid computing technology to resolve the data handling challenge to acquire real-time data on air pollution in a cost-effective manner [124]. This makes it feasible to use sensor networks integrated with telecommunications and informatics systems to monitor, model, and predict the evolution of air pollutants by means of fixed roadside and mobile vehicle-mounted sensors [125, 126].

References

[1] Yamazoe N. Chemical sensors R&D in Japan. Sens Actuators B 1992;6:9–15.

[2] Ikohura K, Watson J. The stannic oxide gas sensor – principles and applications. Boca Raton (FL): CRC Press; 1994.

[3] Schierbaum KD. Engineering of oxide surfaces and metal/oxide interfaces for chemical sensors: recent trends. Sens Actuators B 1995;24–25:239–47.

[4] Schierbaum KD, Geiger G, Weimar U, Göpel W. Specific palladium and platinum doping for SnO₂-based thin film sensor arrays. Sens Actuators B 1993;13–14:143–7.

[5] Bârsan N, Schweizer-Berberich M, Göpel W. Fundamental and practical aspects in the design of nanoscaled SnO₂ gas sensor: a status report. Fresenius J Anal Chem 1999;365: 287–304.

[6] Ogawa H, Abe A, Nishikawa M, Hayakawa S. Electrical properties of tin oxide ultrafine particle films. J Electrochem Soc 1981;128:2020–5.

[7] Schweizer-Berberich M, Zheng JG, Weimar U, Bârsan N, Pentia E, Tomescu A. The effect of Pt and Pd surface doping on the response of nanocrystalline tin dioxide gas sensors to CO. Sens Actuators B 1996;31:71–5.

[8] Sberveglieri G, Groppelli S, Nelli P, Camanzi A. Bismuth-doped tin oxide thin-film gas sensors. Sens Actuators B 1991;3:183–9.

[9] Sberveglieri G, Groppelli S, Nelli P. Highly sensitive and selective NO_x and NO₂ sensor based on Cd-doped SnO₂ thin films. Sens Actuators B 1991;4:457–61.

[10] He YS, Campbell JC, Murphy RC, Arendt MF, Swinnea JS. Electrical and optical characterisation of Sb:SnO₂. J Mater Res 1993;8:3131–4.

[11] Bruno L, Pijolat C, Lalauze R. Tin dioxide thin film gas sensor prepared by chemical vapour deposition: Influence of grain size and thickness on the electrical properties. Sens Actuators B 1994;18–19:195–9.

[12] Davazoglou D. Optical properties of SnO₂ thin films grown by atmospheric pressure chemical vapour deposition oxidizing SnCl₄. Thin Solid Films 1997;302:204–13.

[13] Kanamori M, Suzuki K, Ohya Y, Takahashi Y. Analysis of the change in the carrier concentration of SnO₂ thin film gas sensor. Jpn J Appl Phys 1994;33:6680–3.

[14] Hassan AK, Gould RD, Keeling AG, Williams EW. Electronic conduction processes in Pt-doped tin oxide thin films prepared by RF magnetron sputtering. J Mater Sci: Mater Electron 1994;5:310–4.

[15] Kissine VV, Sysoev V, Voroshilov SA. Individual and collective effects of oxygen and ethanol on the conductance of SnO₂ thin films. Appl Phys Lett 2000;76:2391–3.

[16] Chang SC. Thin-film semiconductor NO sensor. IEEE Trans Electron Devices 1979;ED-26:1875–80.

[17] Gutiérrez FJ, Arés L, Robla JI, Getino JM, Horrillo MC, Sayago I, et al. Hall coefficient measurements for SnO₂ doped sensors, as a function of temperature and atmosphere. Sens Actuators B 1993;15–16:98–104.

[18] Barbi GB, Santos Blanco J. Structure of tin oxide layers and operating temperature as factors determining the sensitivity performances to NO_x. Sens Actuators B 1993; 15–16:372–8.

[19] Vlachos DS, Papadopoulos CA, Avaritsiotis JN. Dependence of sensitivity of SnO_x thin-film gas sensors on vacancy defects. J Appl Phys 1996;80:6050–4.

[20] Capehart TW, Chang SC. The interaction of tin oxide films with O₂, H₂, NO, and H₂S. J Vac Sci Technol 1981;18:393–7.

[21] Rickerby DG, Horrillo MC. Crystallite size distributions and lattice defects in r.f. sputtered nanograin TiO₂ and SnO₂ films. J Nanostruct Mater 1998;10:357–63.

[22] Cricenti A, Generosi R, Scarselli MA, Perfetti P, Siciliano P, Serra A, et al. Pt:SnO₂ thin films for gas sensor characterized by atomic force microscopy and x-ray photoemission spectromicroscopy. J Vac Sci Technol B 1996;14:1527–30.

[23] Cirilli F, Kaciulis S, Mattogno G, Galdikas A, Mironas A, Senuliene D, et al. Influence of Cu overlayer on the properties of SnO₂-based gas sensors. Thin Solid Films 1998;315:310–5.

[24] Sberveglieri G, Faglia G, Groppelli S, Nelli P, Perego C. Oxygen gas sensing properties of undoped and Li-doped SnO₂ thin films. Sens Actuators B 1993;13–14: 117–20.

[25] Kaciulis S, Mattogno G, Galdikas A, Mironas A, Setkus A. Influence of surface oxygen on chemoresistance of tin oxide film. J Vac Sci Technol A 1996;14:3164–8.

[26] Chung YS, Hubenko A, Meyering L, Schade M, Zimmer J. Morphology and phase of tin oxide thin films during their growth from metallic tin. J Vac Sci Technol A 1997;15:1108–12.

[27] Williams G, Coles GSV. Gas sensing properties of nanocrystalline metal oxide powders produced by a laser evaporation technique. J Mater Chem 1998;8:1657–64.

[28] Hu WS, Liu ZG, Wu ZC, Feng D. Comparative study of laser ablation techniques for fabricating nanocrystalline SnO₂ thin films for sensors. Mater Lett 1996;28:369–72.

[29] Hu WS, Liu XG, Zheng JG, Hu XB, Guo XL, Göpel W. Preparation of nanocrystalline SnO₂ thin films used in chemisorption sensors by pulsed laser reactive ablation. J Mater Sci: Mater Electron 1997;8:155–8.

[30] Koinkar VN, Ogale SB. Pulsed excimer laser processing of optical thin films. Thin Solid Films 1991;206:259–63.

[31] Lal R, Grover R, Vispute RD, Viswanathan R, Godbole VP, Ogale SB. Sensor activity in pulsed laser deposited and ion implanted tin oxide thin films. Thin Solid Films 1991;206:88–93.

[32] Philips HM, Li Y, Bi Z, Zhang B. Reactive pulsed laser deposition and laser induced crystallization of SnO₂ transparent conducting thin films. Appl Phys A 1996;63:347–51.

[33] Cabot A, Arbiol J, Morante JR, Weimar U, Bârsan N, Göpel W. Analysis of the noble metal catalytic additives introduced by impregnation of as obtained SnO₂” sol-gel nanocrystals for gas sensors. Sens Actuators B 2000;70:87–100.

[34] Wu NL, Wang SY, Rusakova IA. Inhibition of crystallite growth in the sol-gel synthesis of nanocrystalline metal oxides. Science 1999;285:1375–7.

[35] Wu NL, Wu LF, Yang YC, Huang SJ. Spontaneous solution-sol-gel process for preparing tin oxide monolith. J Mater Res 1996;11:813–20.

[36] Cao X, Cao L, Yao W, Ye X. Influences of dopants on the electronic structure of SnO₂ thin films. Thin Solid Films 1998;317:443–5.

[37] Martinelli G, Carotta MC. Sensitivity to reducing gas as a function of energy barrier in SnO₂ thick-film gas sensor. Sens Actuators B 1992;7:717–20.

[38] Kappler J, Bârsan N, Weimar U, Diéguez A, Alay JL, Romano-Rodriguez A, et al. Correlation between XPS, Raman and TEM measurements and the gas sensitivity of Pt and Pd doped SnO₂ based gas sensors. Fresenius J Anal Chem 1998;361:110–4.

[39] Matsuo Y, Hashimoto K, Sugie Y, Nakamura T, Tsunoda J. Micromachining of tin oxide film by electrochemical reduction process. J Electrochem Soc 1998;145:3067–9.

[40] Rumyantseva M, Labeau M, Delabouglise G, Ryabova L, Kutsenok I, Gaskov A. Copper and nickel doping effect on interaction of SnO₂ films with H₂S. J Mater Chem 1997;7:1785–90.

[41] Brinzari V, Korotcenkov G, Golovanov G. Factors influencing the gas sensing characteristics of tin dioxide films deposited by spray pyrolysis: understanding and possibilities for control. Thin Solid Films 2001;391:167–75.

[42] Yagi I, Hagiwara Y, Murakami K, Kaneko S. Growth of highly oriented SnO₂ thin films on glass substrate from tetra-n-butyltin by the spray pyrolysis technique. J Mater Res 1993;8:1481–3.

[43] Vallet-Regì M, Ragel V, Román J, Martínez JL, Labeau M, González-Calbet JM. Texture evolution of SnO₂ synthesized by pyrolysis of an aerosol. J Mater Res 1993;8:138–44.

[44] Sberveglieri G. Classical and novel techniques for the preparation of SnO₂ thin-film gas sensors. Sens Actuators B 1992;6:239–47.

[45] Yamazoe N. New approaches for improving semiconductor gas sensors. Sens Actuators B 1991;5:7–19.

[46] Siegel RW. Nanostructured materials – mind over matter. Nanostruct Mater 1993;3:1–18.

[47] Hahn H. Gas phase synthesis of nanocrystalline materials. Nanostruct Mater 1997;9:3–12.

[48] Gleiter H. Materials with ultrafine structure: retrospectives and perspectives. Nanostruct Mater 1992;1:1–19.

[49] Suryanarayana C. Nanocrystalline materials. Int Mater Rev 1995;40:41–64.

[50] Göpel W. Solid-state chemical sensors: atomistic models and research trends. Sens Actuators 1989;16:167–75.

[51] Horrillo MC, Gutiérrez FJ, Arés L, Robla JI, Sayago I, Getino J, Agapito JA. Hall effect measurements to calculate the conduction control in semiconductor films of SnO₂. Sens Actuators A 1994;41–42:619–21.

[52] Mosely PT. Materials selection for semiconductor gas sensors. Sens Actuators B 1992;6:149–56.

[53] Ogawa H, Nishikawa M, Abe A. Hall measurement studies and an electrical conduction model of tin oxide ultrafine particle films. J Appl Phys 1982;53:4448–55.

[54] Williams G, Coles GSV. The gas-sensing potential of nanocrystalline tin dioxide produced by a laser ablation technique. MRS Bull 1999;24(6):25–9.

[55] Göpel W, Schierbaum KD. SnO₂ sensors: current status and future prospects. Sens Actuators B 1995;26–27:1–12.

[56] Xu C, Tamaki J, Miura N, Yamazoe N. Grain size effects on gas sensitivity of porous SnO₂-based elements. Sens Actuators B 1991;3:147–55.

[57] Barbi GB, Santos JP, Serrini P, Gibson PN, Horrillo MC, Manes L. Ultrafine grain-size tin-oxide films for carbon monoxide monitoring in urban environments. Sens Actuators B 1995;24–25:559–63.

[58] Morrison SR. In: Hagenmuller P, editor. Proceedings of the second international meeting on chemical sensors. Bordeaux (France): Imprimerie Biscaye; 1986. p. 39–48.

[59] Madou MJ, Morrison SR. Chemical sensing and solid state devices. New York: Academic Press; 1989.

[60] Aaronson AJ. The thin film book of basics. 5th ed. New York: Materials Research Corporation; 1988.

[61] Petrov NN. Ejection of electrons from metals by ions. Soviet Phys Solid State 1960;2: 1182–8.

[62] Rol PK, Onderdelinden D, Kistemaker J. In: Adam H, editor. Transactions of the third international vacuum congress. Vol. 1. Oxford: Pergamon Press; 1966. p. 75–82.

[63] Woodyard JR, Cooper CB. Mass spectrometric study of neutral particles sputtered from Cu by 0–100 eV Ar ions. J Appl Phys 1964;35:1107–17.

[64] Wehner GK, Laegreid N. Sputtering yields of metals for Ar⁺ and Ne⁺ ions with energies from 50 to 600 eV. J Appl Phys 1961;32:365–9.

[65] Horrillo MC. Estudio y Realización de Sensores para CO basados en la Modulación de la Conductividad Eléctrica del Semiconductor SnO₂. PhD Thesis, Universidad Complutense de Madrid; 1992.

[66] Chrisey DB, Hubler GK. Pulsed laser deposition of thin films. New York: John Wiley; 1994.

[67] Cheung J, Horowitz J. Pulsed laser deposition history and laser-target interactions. MRS Bull 1992;17:30–6.

[68] Hubler GK. Pulsed laser deposition. MRS Bull 1992;17:26–7.

[69] Metev S, Meteva K. Nucleation and growth of laser-plasma deposited thin films. Appl Surf Sci 1989;43:402–8.

[70] Ozegowski M, Meteva K, Metev S, Sepold G. Pulsed laser deposition of multicomponent metal and oxide films. Appl Surf Sci 1999;138–139:68–74.

[71] Thornton JA. High rate thick film growth. Ann Rev Mater Sci 1977;7:239–60.

[72] Sanders JV. Structure of evaporated metal films. In: Anderson JR, editor. Chemisorption and reactions on metallic films. Vol. 1. New York: Academic Press 1971; p. 1–38.

[73] Movchan BA, Demschishin AV. Study of the structure and properties of thick vacuum condensates of nickel, titanium, tungsten, aluminum oxide, and zirconium dioxide. Phys Met Metallogr 1969;28:83–90.

[74] Norton MG, Carter CB. On the optimization of the laser ablation process for the deposition of YBa₂Cu₃O_7-δ thin films. Physica 1990;C172:47–56.

[75] Dew SK, Smy T, Tait RN, Brett MJ. Modeling bias sputter planarization of metal films using SIMBAD. J Vac Sci Technol A 1991;9:519–23.

[76] Dew SK, Smy T, Brett MJ. Simulation of elevated temperature aluminum metallization using SIMBAD. IEEE Trans Electron Devices 1992;39:1599–606.

[77] Rickerby DG, Horrillo MC, Santos JP, Serrini P. Microstructural characterization of nanograin tin oxide gas sensors. Nanostruct Mater 1997;9:43–52.

[78] Horrillo MC, Serventi A, Rickerby DG, Gutiérrez J. Influence of tin oxide microstructure on the sensitivity to reductor gases. Sens Actuators B 1999;58: 474–7.

[79] Serventi AM, Rickerby DG, Horrillo MC, Saint-Jacques RG, Gutiérrez J. Transmission electron microscopy investigation of SnO₂ thin films for sensor devices. Nanostruct Mater 1999;11:813–9.

[80] Serventi AM, Rickerby DG, Horrillo MC, Saint-Jacques RG. Transmission electron microscopy investigation of the effect of deposition conditions and a platinum layer in gas-sensitive r.f.-sputtered SnO₂ films. Thin Solid Films 2003;445:38–47.

[81] Patterson AL. The Scherrer formula for x-ray particle size determination. Phys Rev 1939;56:978–82.

[82] Rickerby DG, Horrillo MC. In: Gardener JW, Persaud KC, editors. Olfaction and electronic noses 2000. London: Institute of Physics; 2000. p. 49–54.

[83] Zheng JG, Pan X, Schweizer M, Weimar U, Göpel W, Rühle M. Dislocations in nanocrystalline SnO₂ thin films. Philos Mag Lett 1996;73:93–100.

[84] Gland JL, Sexton BA, Fisher GB. Oxygen interactions with the Pt(111) surface. Surf Sci 1980;95:587–602.

[85] Yamazoe N, Tamaki J, Miura N. Role of hetero-junctions in oxide semiconductor gas sensors. J Mater Sci Eng B 1996;41:178–81.

[86] El Khakani MA, Dolbec R, Serventi A, Horrillo MC, Trudeau M, Saint-Jacques RG, et al. Pulsed laser deposition of nanostructured tin oxide films for gas sensing applications. Sens Actuators B 2001;77:383–8.

[87] Serventi AM, Dolbec R, El Khakani MA, Saint-Jacques RG, Rickerby DG. Highresolution transmission electron microscopy investigation of the nanostructure of undoped and Pt-doped nanocrystalline pulsed laser deposited SnO₂ thin films. J Phys Chem Solids 2003;64:2097–103.

[88] Godbole VP, Vispute RD, Chaudhari SM, Kanetkar SM, Ogale SB. Dependence of the properties of laser deposited tin oxide films on process variables. J Mater Res 1990;5:372–7.

[89] Dolbec R, El Khakani MA, Serventi AM, Trudeau M, Saint-Jacques RG. Microstructure and physical properties of nanostructured tin oxide thin films grown by means of pulsed laser deposition. Thin Solid Films 2002;419:230–6.

[90] Horrillo MC, Serrini P, Santos J, Manes L. Influence of the deposition conditions of SnO₂ thin films by reactive sputtering on the sensitivity to urban pollutants. Sens Actuators B 1997;45:193–8.

[91] Götz A, Gràcia I, Cané C, Lora-Tamayo E, Horrillo MC, Getino J, et al. A micromachined solid state integrated gas sensor for the detection of aromatic hydrocarbons. Sens Actuators B 1997;44:483–7.

[92] Götz A, Gràcia I, Cané C, Lora-Tamayo E. Thermal and mechanical aspects for designing micromachined low-power gas sensors. J Micromech Microeng 1997;7:247–9.

[93] Horrillo MC, Sayago I, Arés L, Rodrigo J, Gutiérrez J, Götz A, et al. Detection of low NO₂ concentrations with low power micromachined tin oxide gas sensors. Sens Actuators B 1999;58:325–9.

[94] Rickerby DG, Waechter N, Horrillo MC, Gutiérrez J, Gracià I, Cané C. Structural and dimensional control in micromachined integrated solid state gas sensors. Sens Actuators B 2000;69:314–9.

[95] Zhu Y, Lu H, Lu Y, Pan X. Characterization of SnO₂ films deposited by d.c. gas discharge activating reaction evaporation onto amorphous and crystalline substrates. Thin Solid Films 1993;244:82–6.

[96] Melsheimer J, Tesche B. Electron microscopy studies of sprayed thin tin dioxide films. Thin Solid Films 1986;138:71–8.

[97] Sundaram KB, Bhagavat GK. X-ray and electron diffraction studies of chemically vapour-deposited tin oxide films. Thin Solid Films 1981;78:35–40.

[98] Tang Z, Fung SKH, Wong DTW, Chan PCH, Sin JKO, Cheung PW. An integrated gas sensor based on tin oxide thin-film and improved micro-hotplate. Sens Actuators B 1998;46:174–9.

[99] Sheng L-Y, Tang Z, Wu J, Chan PCH, Sin JKO. A low-power CMOS compatible integrated gas sensor using maskless tin oxide sputtering. Sens Actuators B 1998;49:81–7.

[100] Sharma RK, Tang Z, Chan PCH, Sin JKO, Hsing I-M. Compatibility of CO gas sensitive SnO₂/Pt thin film with silicon integrated circuit processing. Sens Actuators B 2000;64: 49–53.

[101] Carotta MC, Martinelli G, Crema L, Gallana M, Merli M, Ghiotti G, et al. Array of thick film sensors for atmospheric pollutant monitoring. Sens Actuators B 2000; 68:1–8.

[102] Traversa E, Sadaoka Y, Carotta MC, Martinelli G. Environmental monitoring field tests using screen-printed thick-film sensors based on semiconducting oxides. Sens Actuators B 2000;65:181–5.

[103] Carotta MC, Martinelli G, Crema L, Malagù C, Merli M, Ghiotti G, et al. Nanostructured thick-film gas sensors for atmospheric pollutant monitoring: quantitative analysis on field tests. Sens Actuators B 2001;76:336–42.

[104] Carotta MC, Ferroni M, Gnani D, Guidi V, Merli M, Martinelli G, et al. Nanostructured pure and Nb-doped TiO₂ as thick film gas sensor for environmental monitoring. Sens Actuators B 1999;58:310–7.

[105] Carotta MC, Benetti M, Ferrari E, Giberti A, Malagù C, Nagliati M, et al. Basic interpretation of thick film gas sensors for atmospheric application. Sens Actuators B 2007;126:672–7.

[106] Bârsan N, Weimar U. Understanding the fundamental principles of metal oxide based gas sensors the example of CO sensing with SnO₂ sensors in the presence of humidity. J Phys Condens Mater 2003;15:R813–39.

[107] Carotta MC, Ferrari E, Giberti A, Malagù C, Nagliati M, Gherardi S, et al. Semiconductor gas sensors for environmental monitoring. Adv Sci Technol 2006;45:1818–27.

[108] Graf M, Frey U, Reichel P, Taschini S, Bârsan N, Weimar U, et al. Monitoring of environmentally relevant gases by a digital monolithic metal-oxide microsensor array. Sensors, Proc IEEE 2004;24–27:776–9.

[109] Graf M, Gurlo A, Bârsan N, Weimar U, Hierlemann A. Microfabricated gas sensor systems with nanocrystalline metal-oxide sensitive films. J Nanopart Res 2006;8: 834–9.

[110] Reichel P. Development of a chemical gas sensor system. PhD Thesis, University of Tübingen; 2005.

[111] Varghese OK, Grimes CA. Metal oxide nanoarchitectures for environmental sensing. J Nanosci Nanotech 2003;3:277–93.

[112] Pan ZW, Dai ZR, Wang ZL. Nanobelts of semiconducting oxides. Science 2001;291: 1947–9.

[113] Comini E, Baratto C, Faglia G, Ferroni M, Vomiero A, Sberveglieri G. Quasi-one dimensional metal oxide semiconductors: preparation, characterization and application as chemical sensors. Prog Mater Sci 2009;54:1–67.

[114] Calestani D, Zha M, Salviati G, Lazzarini L, Zanotti L, Comini E, et al. Nucleation and growth of SnO₂ nanowires. J Cryst Growth 2005;275:2083–7.

[115] Comini E, Bianchi S, Faglia G, Ferroni M, Vomiero A, Sberveglieri G. Functional nanowires of tin oxide. Appl Phys A 2007;89:73–6.

[116] Vomiero A, Ferroni M, Comini E, Faglia G, Sberveglieri G. Preparation of radial and longitudinal nanosized heterostructures of In₂O₃, and SnO₂. Nano Lett 2007;7: 3553–8.

[117] Sberveglieri G, Baratto C, Comini E, Faglia G, Ferroni M, Ponzoni A, et al. Synthesis and characterization of semiconducting nanowires for gas sensing. Sens Actuators B 2007;121:208–13.

[118] Bianchi S, Comini E, Ferroni M, Faglia G, Vomiero A, Sberveglieri G. Indium oxide quasi-monodimensional low temperature gas sensor. Sens Actuators B 2006;118: 204–7.

[119] Vomiero A, Bianchi S, Comini E, Faglia G, Ferroni M, Sberveglieri G. Controlled growth and sensing properties of In₂O₃ nanowires. Cryst Growth Des 2007;7: 2500–4.

[120] Ponzoni A, Comini E, Sberveglieri G, Zhou J, Deng SZ, Xu NS, et al. Ultrasensitive and highly selective gas sensors using three-dimensional tungsten oxide nanowire networks. Appl Phys Lett 2006;88:203101.

[121] Faglia G, Baratto C, Sberveglieri G, Zha M, Zappettini A. Adsorption effects of NO₂ at ppm level on visible photoluminescence response of SnO₂ nanobelts. Appl Phys Lett 2005;86:011923.

[122] Baschirotto A, Capone S, D’Amico A, Di Natale C, Ferragina V, Ferri G, et al. A portable integrated wide-range gas sensing system with smart A/D front-end. Sens Actuators B 2008;130:164–74.

[123] Tsujita W, Yoshino A, Ishida H, Moriizumi T. Gas sensor network for air-pollution monitoring. Sens Actuators B 2005;110:304–11.

[124] Richards M, Ghanem M, Osmond M, Guo Y, Hassard J. Grid-based analysis of air pollution data. Ecol Model 2006;194:274–86.

[125] Ma Y, Richards M, Ghanem M, Guo Y, Hassard J. Air pollution monitoring and mining based on sensor grid in London. Sensors 2008;8:3601–23.

[126] Völgyesi P, Nádas A, Koutsoukos X, Lédeczi Á. In: Kaiser WJ, editor. Proceedings of the 7th international conference on information processing in sensor networks. Washington (DC): IEEE Computer Society; 2008. p. 529–30.

..................Content has been hidden....................

You can't read the all page of ebook, please click here login for view all page.

Table of Contents for Chapter 6: Nanostructured Metal Oxide Gas Sensors for Air-Quality Monitoring

Create new playlist

Sign In

Sign Up

CHAPTER 6

1. INTRODUCTION