This article is one of five papers to be presented exclusively on the web as part of the October 2000 JOM-e the electronic supplement to JOM.
	The following article appears as part of JOM-e, 52 (10) (2000), http://www.tms.org/pubs/journals/JOM/0010/Cosandey/Cosandey-0010.html JOM is a publication of The Minerals, Metals & Materials Society

Materials and Processing Issues in Nanostructured Semiconductor Gas Sensors

TABLE OF CONTENTS
INTRODUCTION MATERIALS ISSUES IN SEMICONDUCTOR OXIDE SENSORS THE SYNTHESIS OF NANOSTRUCTURED THIN FILMS PROSPECTS References

The development of gas sensor devices with optimized selectivity and sensitivity has been gaining prominence in recent years. The use of a semiconductor fabrication line is the preferred manufacturing process because of the potential to reduce cost. However, fundamental materials and processing issues, which are critical for a high-performance gas sensor, need to be addressed. Among the new technologies, a nanocrystalline material offers immense promise for improved sensitivity. This article provides a rationale for using nanocrystalline materials in gas sensors, and outlines the challenges facing commercialization of this new class of material.

INTRODUCTION

In recent years, interest has grown in the development of an electronic "nose," capable of detecting mixed gases and even odors. Instead of analyzing all of the gas constituents by a technique such as gas chromatography, an electronic nose looks for specific patterns or fingerprints of the gas mixture. Such a device generally consists of 4-22 chemical sensors, each one sensitive to a specific gas and a pattern recognition system. Various prototypes, based on organic as well as inorganic sensing materials, have appeared in the market.

Because of their simplicity and low cost, semiconductor metal-oxide gas sensors stand out among the ones used in multi-sensor arrays. A multi-sensor array has been proposed for environmental monitoring,¹ in which gas sensing would be based on changes in the surface or near-surface oxide conductivity. Such conductivity changes are caused by the formation of a space charge region induced by either gas adsorption or by the formation of oxygen vacancies on the surface. Improved efficiency and sensing selectivity of these devices require detailed understanding of the surface and interfacial processes at the atomic level, and their relationship with materials properties and device performance.

A non-exhaustive list of semiconductor oxide materials with targeted selectivity for specific gases is shown in Table I, which lists only materials used for air quality monitoring.

Two of the most important issues in gas sensing devices are gas sensitivity (detection of gas concentrations at the ppm level) and gas selectivity^3,15 (detection of specific gases in a mixed gas environment). Semiconductor oxides suffer mostly from a lack of gas selectivity. For instance, the most commonly used oxide, SnO₂, can be sensitized to different gases by judicious choice of operating temperature, microstructural modification, and by the use of dopants and catalysts.⁴

Nanostructured materials present new opportunities for enhancing the properties and performance of gas sensors because of the much higher surface-to-bulk ratio in nanomaterials compared to coarse micrograined materials.

Table I. Partial List of Semiconductor Oxides with Targeted Selectivity for Specific Gases for Environmental and Air Quality Monitoring
Oxide Type	Detectable Gas	References
SnO2	H2,CO,NO2,H2S, CH4	2-4
TiO2	H2, C2H5OH, O2	5-7
Fe2O3	CO	8
Cr1.8Ti0.2O3	NH3	9,10
WO3	NO2, NH3	11,12
In2O3	O3, NO2,	12,13
LaFeO3	NO2, NOx	14

MATERIALS ISSUES IN SEMICONDUCTOR OXIDE SENSORS

Figure 1. The effect of particle size on gas sensitivity for an SnO2 oxide sensor exposed to CO and H2 gases.15

Figure 2. The effect of In2O3 grain size on sensor sensitivity to 1.0 ppm NO2 at 250°C.13

Establishing sensor selectivity for specific gases is difficult and challenging. Selectivity is dependent on many parameters, such as gas adsorption and co-adsorption mechanisms, surface reaction kinetics, and electron transfer to or from the conduction band of the semiconductor. At present, sensor selectivity remains, for the most part, empirical. In practice, selectivity is achieved by enhancing gas adsorption or promoting specific chemical reactions via catalytic or electronic effects using bulk dopants, surface modification methods, and by the addition of metallic clusters or oxide catalysts.^3,16 For instance, the selectivity of chemical sensors can be strongly influenced by the addition of metal clusters such as platinum and palladium, resulting in an increase in the sensor selectivity to reducing gases, such as CO.¹⁷ An increase in selectivity is believed to be caused by close coupling between the sensing and catalytic properties of the metal/oxide system.

Recently studies indicate that the most noble metal, gold, on transition metal oxides could have important applications for room-temperature catalytic oxidation of CO to CO₂ as well as a selective gas sensor for CO and hydrogen.^18-23 Other metal-oxide systems where selectivity is promoted by gold include Au-WO₃ for NH₃ selectivity and Au-In₂O₃ for ozone and trimethylamine selectivity.¹² These selectivity effects are caused by a yet unknown but unique synergistic effect between gold and the oxides TiO₂, WO₃, or In₂O₃. For chemical sensors, the increased selectivity due to additions of metallic clusters is explained using two mechanisms, chemical sensitization and electronic sensitization.^12,15 In chemical sensitization, the metal particles act as centers for surface-gas adsorption, which is then followed by spill-over onto the oxide surface to react with the negatively charged chemisorbed oxygen. On the other hand, electronic sensitization results in a direct electronic interaction between the oxide surface and metal particles via metal oxidation and reduction processes. This is the case for silver and palladium, which are known to form oxides in the presence of oxygen. In the presence of reducing gases, the oxidized metal particles are reduced, leading to a change in the carrier concentration of the semiconductor oxide substrate. For gold on oxides, however, the mechanisms involved are still not clearly understood.²⁴ Also unknown is how chemisorption, size effects, and interactions with semiconductor oxide substrate relate to chemical sensing.

Sensitivity is generally enhanced either by doping, which modifies the carrier concentration and mobility, or by microstructural changes such as reduction of the oxide particle size to the nanometer scale.²⁵ In recent years, the sensitivity of semiconductor oxide materials has been improved by reducing the particle size, with greatly improved properties reported for sizes in the 5-50 nanometer range.^12,15,25,26 The effect of particle size on the sensitivity of a SnO₂ sensor for H₂ is shown in Figure 1,¹⁵ where an order of magnitude increase in sensitivity occurs when the particle size decreases to below 10 nm. The critical particle size where substantial improvement is observed depends on oxide material, dopant, and processing method. For instance, another study showed that sensitivity for H₂ can be increased tenfold by reducing the average particle size to 22 nm.²⁶ For In₂O₃ a particle size effect¹³ starts to occur below 50 nm with an order of magnitude increase in sensitivity for particles in the 20 nm to 30 nm range (Figure 2).

This particle size effect is due, in part, to an increase in the surface area since, in this size range, a large fraction of the atoms (up to 50%) are present at the surface or the interface region with structure and properties that are different from that of the bulk. However, the main effect is associated with the depth of the surface space charge region affected by gas adsorption in relation to the particle size. In the low-temperature regime, conduction variations arise from surface-controlled processes. For chemisorption of an oxidizing gas such as O₂, the molecule dissociates to O^-1, taking an electron from the lattice. As a result, a surface depletion layer is formed leading to a decrease in near-surface conductivity. The surface depletion layer can be expressed by the Debye length, which is defined as:

LD = (eoKT/noe2)1/2

(1)

where e_o is the static dielectric constant, n_o is the total carrier concentration, e is the carrier charge, K is the Boltzmann constant, and T is the absolute temperature. Maximum sensitivity is achieved whenever the Debye length is about half the particle size. In this conduction regime, the sensitivity (S) expressed as the change in conductivity (DG) is related to the change in the carrier concentration (Dn) and is given by

S = DG/Go = (Dn/no)LD

(2)

Therefore, optimum sensitivity is obtained for small grain size, large Debye length, and relative low carrier concentration. In addition, the values of carrier concentration and mobility are strongly dependent on grain size as well as the electronic structure of the gas-oxide and oxide-oxide interfaces.

THE SYNTHESIS OF NANOSTRUCTURED THIN FILMS

Film processing techniques fall into two categories: thin-film deposition processes such as sputtering, evaporation, and chemical vapor condensation (CVD) for thicknesses between 0.005 mm and 2.0 mm and thick-film deposition processes such as screen printing and tape casting for films thicker than 10 mm. Thermal spraying can be used to deposit coatings of metals, ceramics, and cermets that are thicker than ~50 mm. Figure 3 is a summary of the generic processing routes used for synthesis of gas sensor films.

Figure 3. A summary of the processing routes being used to fabricate thin and thick films for gas sensor applications.

Screen printing involves printing a paste or an ink on a suitable substrate followed by a two-stage heat treatment to form a dense (or porous) layer with the desired structure.^14,27 The paste, which consists of powders mixed with an organic medium and a binder, should have the correct rheological properties. In addition, adherence to the substrate and precise shrinkage characteristics are important to obtain a good film. The technique is routinely used to deposit layers of sensor materials, such as SnO₂, TiO₂, and LaFeO₃.

Recently, screen-printed pastes containing nanoparticles were deposited on 96% alumina substrates.^14,27 The "green" films were 2 mm ´ 2 mm and 250 mm thick. The sensor assembly (i.e., the deposits along with the Pt-100 heating element and interdigitated gold electrical contact) was fired in flowing N₂ or air at various temperatures in the range 650-1,000°C for 1 h. After processing, the dimensions of the films changed to 1.5 mm ´ 1.5 mm and 50 mm in thickness. Williams²⁸ used SnO₂ nanopowders synthesized by laser ablation as starting materials and performed screen printing on an electrode array. Since nanopowders produced by any of the vapor phase techniques have, in general, a low trap density, it was not possible to avoid shrinkage cracks. However, cracking was avoided by deliberately aggregating the particles by first dispersing water and drying, before mixing with an a-terpineol-based vehicle to prepare the paste. Working on similar lines, Carotta et al.²⁹ fabricated pure and niobium-doped nanophase TiO₂ films by screen printing, starting from nanopowders produced by laser pyrolysis.

Much emphasis is being placed on developing chemical solution-based thin-film deposition technologies as an economical alternative to the more expensive chemical vapor deposition and reactive sputtering processes. However, the quality of the films produced by vapor deposition processes has always remained superior. Spray pyrolysis, using an atomizing nozzle as small as 300 mm, has been used recently to deposit SnO₂ films that were 50-300 nm thick. Tin chloride was dissolved in ethanol or deionized water and sprayed at a deposition temperature in the range 300-550°C. Manipulation of the structure of the films was made possible by controlling the deposition parameters precisely.

Recognizing the usefulness of a porous film for gas sensing, Mukhopadhyay et al.³⁰ have developed a modified chemical solution-based technique, where a thin adherent film of tin sulfide is formed on a ceramic substrate by reacting sodium sulfide and tin chloride. Subsequently, the tin sulfide film was reacted in air to produce SnO₂.

Variations of CVD techniques, such as plasma-enhanced (PECVD)³¹ and atmospheric-pressure (APCVD) CVD³² have been used to produce both nanopowders and nanostructured thin films. In addition, physical vapor deposition (PVD) techniques using either an evaporation or a sputter source have also been used extensively.^7,33 In our work,³⁴ we have used a vapor-phase process to directly deposit, in one step, a nanostructured film of gas sensor materials, such as SnO₂ and TiO₂. The process, called low-pressure flame deposition (LPFD), is based on the combustion flame-chemical-vapor condensation process used to produce oxide nanoparticles with minimal aggregation. Both of these processes use a similar experimental configuration. A flat flame burner that can be operated at low pressures of 100 torr is placed parallel to a substrate in a stagnant flow configuration. Under these conditions, the chemical composition, temperature, and residence time are the same across the entire surface of the burner. When the nanoparticles that are formed in the flame are partially quenched, a film develops on the substrate. The nanostructured film is generally porous, although dense films have also been obtained under certain operating conditions. A high-resolution scanning electron microscopy image of a porous SnO₂ deposited on a pre-oxidized silicon substrate is shown in Figure 4. The low-magnification Figure 4a reveals the uniformity of a 2.4 micrometer thick film while the high-magnification Figure 4b shows the porous nature of the LPFD films with particle sizes in the 6 nm to 20 nm range.

Figure 4. A high-resolution scanning electron microscopy image of a porous SnO2 deposit on a pre-oxidized silicon substrate at both (a) low and (b) high magnifications.

The mechanism of film deposition in the LPFD process differs from conventional CVD in two major respects: nanoparticles are deposited on the substrate rather than the precursor being decomposed on a heated substrate and the size of the deposition area corresponds to the size of the burner. Thus, LPFD is a one-step process suitable for the formation of both thin and thick films and eliminates the need for powder processing. It is also a high-rate process, with deposition rates exceeding 1 mm/min.

PROSPECTS

Nanostructured materials are recognized as essential for achieving high gas sensitivity. Numerous processing schemes have been tested successfully, albeit only on a laboratory scale. Processing techniques should be able to provide the desired oxide composition with specific dopant and requiring the least number of processing steps. For this, vapor-based processes seem to be the most promising approach. However, new precursors will be required for synthesizing multi-component oxides with specific dopants. Future trends in sensor miniaturization and integration with electronics will require processing compatibility with silicon-based technologies. In addition to processing nanostructured oxides, more fundamental work is needed to understand the role of nanostructured oxide materials on gas adsorption and conductivity.

An area where greater improvements are necessary is gas selectivity. At present, a fundamental understanding of the basic phenomena associated with metal-oxide interaction in relation to chemical selectivity is lacking. Most of the developments on chemical selectivity have been empirical, and further progress in this area is not possible without an understanding of the processes and interfacial phenomena at the atomic level. Oxide materials currently in use could be made more sensitive by the introduction of dopants with unique gas adsorption characteristics and by the development of new materials having specific catalytic properties for enhancing gas selectivity.

References

1. G. Martinelli et al., "Thick-Film Gas Based Nano-Sized Semiconducting Oxide Powders," MRS Bulletin (June 1999), pp. 30-36.
2. A.M. Azad et al., "Solid-State Gas Sensors: A Review," J. Electrochem. Soc., 139 (1992), pp. 3690-3704.
3. W. Gopel, "New Material and Transducers for Chemical Sensors," Sensors and Actuators B, 18-19 (1994), pp. 1-21.
4. G. Sberveglieri, "Recent Developments in Semiconducting Thin-Film Gas Sensors," Sensors and Actuator B, 23 (1995), pp. 103-109.
5. L.D. Birkefeld et al., "Carbon Monoxide and Hydrogen Detection by Anatase Modification of Titanium Dioxide," J. Am. Ceram. Soc., 75 (1992), pp. 2964-2968.
6. H. Tang et al., "TiO₂ Anatase Thin Films as Gas Sensors," Sensors and Actuators B, 26-27 (1995), pp. 71-75.
7. F. Edelman et al., "E-Gun Sputtered and Reactive Ion Sputtered TiO₂ Thin Films for Gas Sensors," Inst. Electron Technol., 33 (2000), pp. 89-107.
8. T. Kobayashi et al., "Thin Films of Supported Gold Catalysts for CO Detection," Sensors and Actuators B, 1 (1990), pp. 222-225.
9. P. Moseley and D.E. Williams, "A Selective Amonia Sensor," Sensors and Actuators B, 1 (1990), pp. 113-115.
10. V. Jayaraman et al., "Preparation and Characterization of Cr_2-xTi_xO_3^+d and its Sensor Properties," Sensors and Actuators B, 55 (1999), pp. 175-179.
11. D. Manno et al., "Physical and Structural Characterization of Tungsten Oxide in Films for Gas Detection," Thin Solid Films, 324 (1998), pp. 44-51.
12. Y. Shimizu and M. Egashira, MRS Bulletin, June (1999), pp. 18-24.
13. A. Gurlo et al., "In₂O₃ and MoO₃-In₂O₃ Thin Film Semiconductor Sensors: Interaction with NO₂ and O₃," Sensors and Actuactors B, 47 (1998), pp. 92-99.
14. G. Martinelli et al., "Screen-Printed Perovskite-Type Thick Films as gas Sensors for Environmental Monitoring," Sensors and Actuators B, 55 (1999), pp. 99-110.
15. N. Yamazoe, "New Approaches for Improving Semiconductor Gas Sensors," Sensors and Actuators B, 5 (1991), pp. 7-19.
16. V. Demarne and R. Sanjines, Gas Sensors-Principles, Operation and Developments, ed. G. Sberveglieri (Dordrecht, the Netherlands: Kluwer Academic, 1992), pp. 89-116.
17. D. Hohl, "The Role of Noble Metals in the Chemistry of Solid-State Gas Sensors," Sensors and Actuators B, 1 (1990), pp. 158-165.
18. M. Haruta, "Size- and Support-Dependency in the Catalysis of Gold," Catalysis Today, 36 (1997), pp. 153-166.
19. M. Valden, X. Lai, and D.W. Goodman, "Onset of Catalytic Activity of Gold Clusters on Titania with the Appearance of Nonmetallic Properties," Science, 281 (1998), pp. 1647-1650.
20. H. Sakurai et al., "Low-Temperature Water-Gas Shift Reaction Over Gold Deposited on TiO₂," Chem. Commun. (1997), pp. 271-272.
21. S.D. Lin, M. Bollinger, and M.A. Vannice, "Low Temperature CO Oxidation over Au/TiO₂ and Au/SiO₂ Catalysts," Catalysis Letters, 17 (1993), pp. 245-262.
22. M. Haruta et al., "Low-Temperature Oxidation of CO over Gold Supported on TiO₂, a-Fe₂O₃ and Co₃O₄," J. of Catalysis, 144 (1993), pp. 175-192.
23. T. Kobayashi et al., "A Selective CO Sensor Using Ti-Doped a-Fe₂O₃ with Coprecipitated Ultrafine Particles of Gold," Sensors and Actuators, 13 (1988), pp. 339-349.
24. F. Cosandey and T.E. Madey, "Growth, Morphology, Interfacial Effects and Catalytic Properties of Au on TiO₂," submitted for publication in Surface Review and Letters.
25. C. Xu et al., "Grain Size Effects on Gas Sensitivity of Porous SnO₂-Based Elements," Sensors and Actuators B, (1991), pp. 147-155.
26. G. Ansari et al., "Grain Size Effects on H₂ Gas Sensitivity of Thick Film Resistor using SnO₂ Nanoparticles," Thin Solid Films, 295 (1977), pp. 271-276.
27. G. Martinelli et al., "Thick-Film Gas Sensors Based on Nano-sized Semiconducting Oxide Powders," MRS Bulletin, (1999), pp. 30-35.
28. G. Williams and G.S.V. Coles, "The Gas-Sensing Potential of Nanocrystalline Tin Dioxide Produced by a Laser Ablation Technique," MRS Bulletin, June (1999), pp. 25-29.
29. M.C. Carotta et al., "Nanostructured Pure and Nb-doped TiO₂ as Thick Film Gas Sensors for Environmental Monitoring," Sensors and Actuators B, 58 (1999), pp. 310-317.
30. A.K. Mukhopadhyay et al., "A New Method of Preparing Tin Dioxide Thin Film Sensors," J. Mater. Sci. Letters, 17 (1998), pp. 625-627.
31. Y. Liu et al., "Structural and Gas Sensing Properties of Nanometre Tin Oxide Prepared by PECVD," J. Mater. Sci.: Mater. in Electronics, 7 (1996), pp. 279-282.
32. C.C. Chai, J. Peng, and B.P. Yan, "Preparation and Gas Sensing Properties of alpha-Fe₂O₃ Thin Films," J. Electron. Mater., 24 (1995), pp. 799-804.
33. M. Ferroni et al., "Nanosized Thin Films of Tungsten-Titanium Mixed Oxides as Gas Sensors," Sensors and Actuators B, 58 (1999), pp. 289-294.
34. G. Skandan et al., "Low-Pressure Flame Deposition of Nanostructured Oxide Films," J. Am. Ceram. Soc., 81 (1998), pp. 2753-2756.

Frederic Cosandey is with the Department of Ceramic and Materials Engineering at Rutgers University. Ganesh Skandan and Amit Singhal are with Nanopowder Enterprises Inc.

For more information, contact Frederic Cosandey, Rutgers University, Department of Ceramic and Materials Engineering, 607 Taylor Road, Piscataway, NJ 08854-8065; e-mail cosandey@scils.rutgers.edu.