1

Printed Chemical Sensors: from ScreenPrinting to Microprinting∗

Alexey A. Tomchenko

∗

The presented text is a revised and updated version for the article published in “Encyclopedia of Sensors” (Alexey A. Tomchenko, “Thick-Film Semiconductor Chemical Sensors”, in Encyclopedia of Sensors, Ed. Craig A. Grimes, Elizabeth C. Dickey, and Michael V. Pishko, American Scientific Publishers, Stevenson Ranch, California, USA, 2006, V. 10, 279-290)

2

CONTENTS 1.

Overview

2.

Technologies

3.

Principle of operation

4.

Designs

5.

Conclusion Glossary References

Afterword: More about microprinting and printed microsensors

1. OVERVIEW Chemical sensors are microelectronic devices providing immediate feedback on variations in chemical composition of the environment (aquatic or gaseous). In recent years, research and development in chemical sensors have acquired unprecedented scope, which has been a reflection of the broad societal interest in these devices. Various chemical sensors have already been used in emission monitoring, industrial hygiene, home safety alarms, food and water quality controls, clinical diagnostics, process controls, and homeland security. Evidently, the proliferation will continue, and it can be foreseen with great deal of confidence that the future postindustrial world will be inundated with various chemical microsensors making human life safer and more comfortable. A chemical sensor (or a chemosensor) could be defined as “a small device that as the result of chemical interaction or process between the analyte gas and the sensor device, transforms chemical or biochemical information of quantitative or qualitative type into an analytically useful signal” [1]. In its turn, a semiconductor chemical sensor is a chemical sensor with a sensitive element made of semiconductor whose electrical properties (conductivity, work function etc.) vary with the

concentration of a detected chemical agent. In principle, any semiconductor could be considered as a potential material for chemical sensor sensitive elements, since any semiconductor changes its electrical properties upon variations in the chemical composition of the environment. The point is the magnitude of the electrical response, and because modern sensors should be responsive to very low concentrations of the detected chemical agents as a rule (for example, these are one part per million concentrations for the exhaust monitoring, and one part per billion concentrations for the chemical warfare agent detection), the family of sensor semiconductors is limited to the materials that demonstrate the heightened sensitivity towards chemical composition of the environment in general and to a detected substance in particular. These materials are wide-bandgap metal oxide semiconductors, which are applied in modern sensors in the form of polycrystalline thick or thin films. The terms “thick films” and “thin films” do not relate so much to the thickness of the films but more to the method of deposition. Thick films are fabricated using specially formulated metal-oxide suspensions, which are deposited onto substrates by screen-printing [2, 3], drop-coating (microprinting) [4-7], spin-coating [8-14], dip-coating [15-18], or micromolding in capillaries (MIMIC) [19]. Thin films are deposited by sputtering [20-24], chemical vapor deposition [25-28], or laser ablating [29-31]. Thick-film semiconductor chemical sensors have a special status in the realm of Chemosensorics. Owing to their robustness, diminutiveness, simplicity in production, and relative cheapness, the devices are widely used in today’s detectors of toxic and explosive gases. It has been established that thick films based on nano-scaled powders offer better sensing properties than thin films [32-37]. When powder synthesis routes are used, a better control is possible over grain size, phase composition, dopant stoichiometry and other characteristics influencing gas sensing performance of the devices [38]. The advancement of the sensors originates from the outstanding progress in material

3

science and microelectronics achieved in the last decades of the 20th century. The profound understanding of the physical-chemical processes occurring upon gas adsorption-desorption on semiconductor surface, the development of the reliable methods for the synthesis of nanocrystalline semiconductor metal oxide (SMO) powders, and innovations in technology of thick films – these are the main conditions predetermined the rapid evolution of the sensors from laboratory freaks to advanced selective devices and sensor arrays. Thick-film sensors make up the lion’s share of the modern market of semiconductor chemical sensors. The leading producers in chemical sensor industry, such as Figaro Engineering Inc. and City Technology Ltd., extensively use thick-film technologies in the manufacture of their newest chemical sensors [39, 40]. The article below covers the principles and applications of a variety of semiconductor thick-film sensors, including their technological basics, principles of operation, design, constructions, and gives practical examples of where they used today and where they may be used in the near future.

2. THICK FILM TECHNOLOGIES Until recently screen-printing has been considered as only practical method for manufacturing thick films. For decades, these two matters have been inseparable: when someone talked about thick films, the one meant unambiguously that the films had been produced by screen-printing. The method has been optimized for the fabrication of hybrid integrated circuits. These microelectronic devices comprising thick-film elements and add-on miniature discrete components are widely used in areas such as televisions, telecommunications, automotive electronics, military and space technologies. Recently, screen-printing has been successfully applied for the production of various physical and chemical sensors [41]. The demand for

new technological approaches to the thick-film deposition has emerged with the advent of micromachined sensor platforms or microhotplates [42-44]. Semiconductor chemical sensors usually operate at elevated temperatures – at 150°C and above, and up to 1000°C. The sensors based on traditional platforms, i.e. on ceramic substrates equipped with resistance heaters, typically consume too much power to be used in portable or handheld detectors. The modern microhotplate platforms on silicon substrates demonstrate a significant reduction in sensor power consumption, but they are too fragile to survive the traditional screen-printing. As alternative, several new thick-film deposition methods have been advanced recently: special stencil screen-printing [45, 46], microprinting [4-7, 47], and MIMICS [19].

2.1. Screen-Printing Screen-printing in its traditional form – silk screenprinting – has been widely used in reproduction, printing and graphic arts for centuries. It was quite natural when in the late 1950s this technology capable of large-scale producing fine-line print geometry was first sampled in electronics as a probable alternative for the printed circuit board process. However, the real widespread use of screen-printing in electronic industry was started after the development of thick-film hybrid integrated circuits (HIC) in the 1960s [3, 48]. In principle, it was the same approach in design as the one used in printed circuit boards but realized at the new level of miniaturization and on the basis of new materials and new technologies. It was shown that some of functional electronic modules could be reproduced as miniature hybrid integrated circuits. These microcircuits comprised a ceramic substrate with thick-film interconnections (a transformation of a printed circuit board), integrated thick-film passive elements, and add-on diminutive discrete devices – components. Hybrid integrated circuits have found wide application in televisions, telecommunications, automotive electronics, military and space technologies. They are compact, robust and

4

relatively inexpensive. The essence of the HIC technology has not changed for the past decades. Thick-film interconnections and thick-film elements of modern HICs are formed by the screen-printing method, layer after layer; every print operation is followed by drying and sintering of the deposited film. Add-on active components and, if need be, add-on passive components are attached to the substrate with the finished thick-film elements by either soldering or bonding. Finally, the assembled circuits are packed in appropriate packages and tested.

substrate), and a squeegee to force the paste through the screen. Fig.1 shows the schematic arrangement of the aforenamed parts. The screen is held at a distance of around 0.5 mm from the substrate. The thick-film paste is poured onto the top surface of the screen, and the squeegee moves across the screen, pressing the mesh down into contact with the substrate and forcing the paste through the open areas. Directly behind the squeegee the mesh peels away from the substrate and leaves a deposit of the paste, having the required pattern, on the substrate surface.

With the advent of the new materials, which, on the one hand, demonstrated sensitivities to various physical and chemical stimuli, and, on the other hand, were capable to be both printed and fired, screen-printing came into wide use as the method for fabricating of neoteric microelectronic devices – thick-film sensors. The spectrum of thick-film sensors presented on the today market is really huge. The diversity could be organized in five common domains according to the input energy converted in the output electrical signal, namely: mechanical, magnetic, radiant, thermal, and chemical [3, 41]. The domain of chemical sensors includes electrochemical cells, semiconductor sensors, and biosensors [49, 2, 50].

The printing is followed by drying and sintering of the film. Pastes for thick-film microelectronics (also often referred to as inks) contain three ingredients, as a rule: an organic vehicle, a functional material which defines electrical and sensing properties of the film, and either a glass frit or an oxide admixture, which serves as a binding agent of the film to the substrate [51].

The deposition process adopted for the thick-film production is quite similar to that used for the silk screen-printing on plates, plastic cards, T-shirts etc. However, the screen materials and degree of sophistication of the printing machines are somewhat different in microelectronics [48]. A typical thick film screen is finely woven mesh of stainless steel or polyester mounted under tension on a metal frame. The mesh is coated with UV-sensitive emulsion onto which the pattern of the desired thickfilm layer is formed photographically. The finished screen has open mesh areas through which the pattern can be printed. Four constituent parts are essential in order to screen print. These are a printing medium (thick film paste), a screen, a surface onto which the print will be made (i.e. a

Figure 1. The basic screen-printing process

5

The organic vehicle gives the paste the desired viscosity for screen-printing. After printing this ingredient has to be removed out of the film body. Drying that typically takes place at temperatures between 100°C and 150°C removes most of the organic vehicle and makes the film relatively immune to smudging. This operation is performed in intensively ventilated either a conventional box oven or an infrared belt drier. The next stage of the process is to sinter the film. During the sintering (annealing) the remaining organic binder is removed, the glass frit melts, the fine powder of the functional material sinters, and the overall film becomes a solid composite material. The glass frit - or, alternatively, chemically active oxide - bonds the film to the substrate and binds the functional material particles together. Typically annealing is performed in a belt furnace. The electrical properties of thick films are extremely sensitive to the annealing conditions, and, for this reason, accurate control over the processing parameters is required. Modern belt furnaces provide control over a variety of parameters including throughout speed, peak temperature, and dwell time. The processed thick-film structures are mounted on the metal belt that slowly transport them through the heated furnace: their temperature slowly ramps up during the travel towards the hot zone, then remains constant in the hot zone for a given time, and then slowly ramps down to the room temperature. Annealing is a final operation of the thick-film fabrication process. After annealing, the structure is ready for the deposition of the next layer. Thick-film layers are formed in an orderly sequence. In order to protect finished films against destructive overheating during the posterior heat treatments, the materials having high annealing temperatures (for example, Pt conductors) are deposited first. The sequence is closed with materials specially formulated to have low annealing temperature – these are various overglazes or sealants, as a rule. Finally the prepared thick-film structure must be diced (if need be), wire bonded, mounted in the appropriate package and tested.

The technology outlined above is widely exploited in semiconductor sensor fabrication. There is some specificity in chemical sensor manufacture, however, which is conditioned by peculiarities both in the design of the devices and in the principle of their operation. The sensors are generally heated during their operation, and hence they need an integrated heater that is most often formed on the rear side of the sensor substrate. This feature predetermines the necessity of the precise coordinated positioning of the screen-printed patterns located on the opposite sides of the substrate. Sintered metal oxide films often have high resistivity, and therefore the fine interdigitated metal structures are typically used as electrodes to sensor sensitive elements. The feature dimensions of the electrodes are typically 150-200 µm that is close to the lower limit of the screen-printing with the conventional screen-printing machines. This proximity imposes especially strict requirements on both quality of applied screens and quality of machine adjustment. There are some good signs, however, that the fabrications of the fine geometry of the sensor IDT electrodes will not be such a difficult problem in the nearest future. Recently it has been reported on the successful try-outs of the new screens facilitated printing down to 50 µm [52], and even to 25 µm [53]. The former ones - the so-called µ-Screens – have been presented by the ERA Technology Ltd. The active central area of the µ-Screen is a thin foil of stainless steel into which the ink feeder holes are etched. The holes are etched only where printing is required. The unique combination of the precisely patterned ultra-fine mesh and the robust stainless steel foil basis permits 50 µm line and space printing. µScreens provide not only higher printing resolution but also significantly higher edge definition [52]. The latter of the aforementioned advanced screens have been developed by the DEK. The DEK’s screens capable to produce 25 µm patterns are realized using more traditional approach, i.e. with a new emulsion technology and new ultra-thin high-strength wire mesh.

6

One more specific problem that sometimes emerged in the manufacture of thick-film metal oxide sensors is the low adhesion of some types of sensitive films to a substrate. In order to maintain high sensitivity of the sensors, developers often use reduced amount of glass frit or even do not use frit at all [54]. As a result, such weak-sintered films suffer from low adhesion and reduced durability. In this case there is a risk of the film destruction at any accidental mechanical impact. The problem is accentuated if a mechanical treatment (e.g., dicing with diamond dicing saw) is an inevitable operation following the deposition of the sensitive film. The good method to obviate the risk of the destruction is the application of porous inorganic protectors. As demonstrated in [55, 56], the chemically inert porous coatings can effectively protect sensitive elements.

2.2. Deposition onto Micromachined Platforms Early in the 1990s the invention of the silicon-based micromachined chemical sensor platforms (or microhotplates) [42, 43] opened up opportunities for the development of portable and even handheld chemical detectors based on semiconductor sensors. A variety of the platform designs have been implemented since [34]. The most outstanding feature of microhotplates is the low amount of energy required to heat the active sensor layer up to its sensing temperature - in the range of 5-100 mW. A microhotplate is a suspended multilayer membrane (or micro-bridge), which is typically made of silicon dioxide. It has an incorporated heating element, which is usually made of polycrystalline silicon, and then, from bottom to top, an insulating layer of silicon dioxide, a metal heat-distribution plate, another SiO2 insulating layer, and, finally, metal electrodes for a sensing element (Fig. 2). A freestanding membrane has a very low thermal mass and is thermally isolated from the surrounding frame (i.e. from the silicon substrate). It means that the frame itself stays nearly at ambient temperature even though the membrane is heated up to a very high operating temperature – 400°C and higher.

Microhotplates are widely used in production of individual sensors [45, 46, 57, 58]. What is more important, they have become an excellent base for the fabrication of the chemical sensors of the new generation, namely, integrated multi-element sensor arrays and miniature e-noses [44, 34, 59-63]. There successfully advanced several methods for depositing thick films onto microhotplates. The most popular among them – and the very well developed one – is drop-coating or microprinting [4-7, 47]. As any one of the thick-film fabrication processes, the process of microprinting is based on the use of a paste prepared by mixing functional materials and an organic binder. A glass frit is also added, if necessary. The paste is deposited onto membranes using a microinjector coupled with a micropipette that is positioned over the sensor platform by means of a micromanipulator. The process is performed under the microscope. Obviously, a thorough control over the parameters of the paste, precise positioning mechanisms, and a careful control over micro-injector operation, are necessary, in order to obtain the film of the desired characteristics (size, porosity, thickness, adhesion etc.). After the deposition, the formed thick-film sensitive elements are annealed at an appropriate temperature. As a rule, integrated thin-film heaters are used for performing this operation.

Figure 2. Micromachined gas sensor: (a) top view; (b) crosssectional view

7

Typically, pastes for the microprinting are less viscous then the pastes used in screen-printing [48]. To provide better control over the droplet injection, the paste has to be loaded into micropipette avoiding any air introduction. Air bubbles in microchannels impart compressibility to the paste, which results in the loss of the control over the paste delivery. The deposition starts with obtaining the precise vertical positioning of the micropipette filled with the paste against the micromembrane. Applying overpressure to the fluid the meniscus of the paste is formed on the tip of the pipette. The micropipette is moved towards the sensor until the meniscus touches the membrane. Because of the overpressure evacuation during the contact, the paste goes from the micropipette to the substrate. The process is stopped moving away the micropipette. The amount of the deposited paste is controlled by the time of the contact. The advantage of the above deposition method is the absence of any direct contact of the instrument with sensor: the risk of the destruction of micromachined membranes is minimal in this case. A limitation of microprinting is that the areas of the coatings are quite large. The smallest droplets deposited by this method are about 100 µm across [5, 64]. It means that the dimensions of the membrane have to be larger than 100 µm. Obviously, the accidental ingress of the paste into the undermembrane cavity will destroy thermo-isolation of the sensor and cause its failure. As alternative to microprinting, Heule M. and Gauckler L. J. [19] developed and successfully applied micromolding-incapillaries technique (MIMIC) for fabricating their thick-film sensitive elements. MIMIC (also referred to as a soft lithography) is a microfabrication technique based on elastomeric molds of silicon rubber, polydimethylsiloxane (PDMS), for pattern transfer. Micropatterns are transferred by casting of PDMS against a positive relief structure (master) containing photoresist lines. PDMS is then peeled off, cut and used as a mold. Being placed on a substrate, a mold forms microcapillaries, which can be filled with a SMO suspension. The master structures may be reused many times to cast PDMS molds. Performing of this technique does not require

clean room conditions. Heule and Gauckler succeeded in fabricating tin oxide microlines of just 10 µm wide. Together with platinum electrodes these lines formed an array of 12 tin oxide gas sensors on a single microhotplate. Using this approach, i.e. integrating more than one sensor on a single hotplate, allowed the authors to cut the energy consumption to 1/12 compared to operating a 12hotplate sensor array. The single sensor was shrunk on a minimal area of 10 µm × 30 µm. A schematic diagram of the MIMIC coating process is shown in Fig. 3.

Figure 3. Micromolding-in-capillaries process: (a) PDMS mold is aligned and placed directly on a microhotplate chip. PDMS establishes a conformal contact; the microchannels are sealed. (b) A suspension droplet at the side outside of the membrane is applied; capillary filling begins. (c) Wait for the suspension to dry; microlines of metal oxide are deposited inside microchannels. (d) Gently remove PDMS mold to release the structure [99].

8

The PDMS mold was aligned perpendicular to the electrode pattern of the microhotplate chip and brought into contact. The tin oxide suspension was dispensed at the entrance of the mold. After the capillary filling, the sample was left to dry at room temperature. After 30 minutes, the mold was gently lifted. Annealing was carried out on-chip at 290300°C. One more approach in thick-film deposition onto microhotplates is based on stencil screen-printing. As mentioned above, the traditional screen-printing technology cannot directly be applied for fabricating thick films on micromembranes: the mechanical stresses typical for the process destroy fragile architecture of the devices. Nevertheless, D. Vincenzi et al. [45, 46] has demonstrated that screenprinting on micromachined sensor platforms is possible in principle provided that certain modifications in screens and in platform design have been accomplished. The authors have successfully applied their method for the formation of 250 µm × 350 µm × 40 µm sensitive elements on 2.5 mm × 2.5 mm membranes. It is mandatory to apply negligible forces to membranes during the printing in order to avoid cracks. To meet this requirement, the screenprinting machine has been equipped with a stencil with chamfered edges. Stencils are much stiffer with respect to conventional screens based on stainless steel mesh. They operate in contact with substrate. Conventional screens distribute the squeegee pressure over a line that scans the substrate area (see Fig. 1). In contrast to screens, stencils distribute the pressure over a large surface thus limiting the probability of cracks. Moreover, the absence of sharp edges in the apertures minimizes adhesion of the paste to the stencil, which allows one to significantly improve printing resolution.

3. PRINCIPLE OF OPERATION A typical sensitive element of a modern thick-film semiconductor chemical sensor is a gently sintered

porous 10-100 µm thick film prepared on the basis of a nanocrystalline metal-oxide semiconductor powder precursor (Fig. 4). Amongst the materials most frequently used for the sensors are SnO2, ZnO, TiO2, In2O3, WO3, Ga2O3, CuO and Fe2O3. Gas sensitivity of many other metal oxides are under investigation at present: the obvious progress in the field, as well as a myriad of chemicals to detect from the one hand, and a myriad of metal oxide semiconductors from the other hand, stimulate the extensive search for the new more sensitive and more selective materials.

Figure 4. SEM image of the surface of a WO3 thick-film sensitive element [100].

Similar to other electrical chemosensors, as well as electrochemical, magnetic, thermometric, and masssensitive ones, thick-film semiconductor chemosensors generate electrical signals in response to changes in chemical composition of the surrounding atmosphere. The feature, which sets semiconductor chemosensors from the others, is the

9

specific and quite complex mechanisms of the signal generation. Semiconductor thick-film sensors can be divided in two big classes, depending on their principle of operation, namely: sensors for the detection of minority gases and oxygen sensors [2, 66].

Physical adsorption is assumed to be a surface binding caused by Van der Waals interaction. Chemical adsorption is a chemical bonding of molecules with adsorbent. The chemical adsorption is caused by covalent forces with plausible involvement of ionic interactions.

Sensors for the detection of minority gases are used in atmospheres with high and relatively stable oxygen partial pressure (in air) to detect minor concentrations of hazardous gases. These devices usually operate at 150-500°C. The physical phenomenon underlying the operation of the sensors is chemical adsorption of the detected particles on the semiconductor surface. Chemical adsorption changes electrophysical properties of semiconductors, which is observed experimentally as the electrical signal with the magnitude depending on the amount of the adsorbed particles and, correspondingly, on the concentration of the particles in atmosphere. The idea about the influence of chemical adsorption on the electrical properties of semiconductor adsorbents was formulated in the 1930s by Yoffe, Roginsky, Wagner and others and was theoretically elaborated in the 1940-1960s by Wolkenstein, Hauffe, Gray, Heiland, Myasnikov, Weisz and some other authors (see book [67] and the reference list therein). The validity of considering adsorbed particles as surface defects was demonstrated. It was also demonstrated that, depending on the nature of the adsorbed particle and the nature of the adsorbent, the particle could be a surface localization centre of either free electrons or holes. In other words, the particle can act as either an acceptor or donor of electrons. Under thermodynamic equilibrium conditions the population of the arising localized electron levels obeys the Fermi statistics. Adsorption is usually defined as “a process by which molecules are taken up on the surface of a solid by chemical or physical action” [68]. Traditionally two types of adsorption are distinguished (Fig. 5).

Figure 5. Various forms of adsorption: 1 – physical adsorption is a surface binding caused by polarization dipole-dipole Van der Waals interaction; 2 and 3 – chemical adsorption stems from covalent forces with plausible involvement of ionic interactions.

The chemisorbed particle (adsorbate, adparticle) and the adsorbent form a unified quantum mechanical system. In its turn, there are two different forms of chemical adsorption. Weak chemisorption is realized without free current-carriers of the adsorbent lattice, and the adsorbed particle remains neutral. Strong chemisorption takes place, if the free lattice electron (or the free hole) is localized near the chemisorbed particle. The strong bond is conditioned by the exchange interaction between free carriers and adparticles. This is the charged form of chemisorption. Upon strong chemisorption the

10

value of the surface charge localized on the adsorbent surface changes. This inevitably leads to the change in the depth of the space-charge region and to the alteration in the height of the surface potential barrier, and - as a result - to the significant modification of the electrical properties of the adsorbent [67]. There exists one-to-one correspondence between the nature of strongly chemisorbed particles and the form of variations in electrical characteristics of the semiconductor adsorbent caused by the chemisorption. In the case of the n-type semiconductor, the adsorption of acceptor particles (e.g., O2, NO2, Cl2) results in formation of the electron-depleted surface layer, which, in its turn, leads to a decrease in conductivity and to an increase in work function of the semiconductor. Conversely, adsorption of donor particles (e.g., atoms H, Na, Ag) on the n-type semiconductor surface leads to an increase in the conductivity and to a decrease in the work function. In the case of the p-type semiconductor, adsorption of chemical particles influences its electrical conductivity and work function in the opposite manner. The barrier energy eVs on the semiconductor surface may be described by the Schottky equation. In the case of the n-type semiconductor and acceptor adparticles,

e 2 N t2 eVs = 2εε 0 N d

,

(1)

where Nt is the surface density of strongly chemisorbed particles (which is equal to the density of electrons trapped on the surface as a result of adsorption), εε0 is the permittivity of the semiconductor, and Nd is the volumetric density of the electron donors within the semiconductor [69]. A thick-film sensitive element of a semiconductor chemical sensor is an aggregate of nanosized metal

oxide crystallites. Adjacent crystallites can either be connected by metal oxide necks or simply touch each other (point contacts). The neck can or cannot be a barrier for current-carriers depending on the neck thickness, defect concentration, and the value of the surface charge localized at its surface. If the neck is thin enough (s<2LD, where s is the thickness of the neck, and LD is the Debye length) and the value of the charge on its surface is big enough, the depletion regions overlap within the neck. This overlapping results in a high potential barrier for current-curriers. By commonly recognized terminology [70, 71], necks with barriers more than kT are necks of closed type, and necks without such barriers are referred to as necks of open type. Point contacts between crystallites are equivalent to double Schottky barriers. As a rule, either necks of closed type or point contacts dominate in nanocrystalline thick-film sensor elements. Electrical characteristics of the elements strongly depend on the heights of the intercrystalline potential barriers: variations in the value of the surface charge caused by chemical adsorption result in the barrier variations, which inevitably affect conductance of the elements. According to Morrison [72], the conductance G of a sensor element with predominated point intercrystalline contacts may be described by

G = G0 exp(−eVs / kT ) ,

(2)

where eVs is the surface potential barrier between crystallites and G0 is the factor that includes the bulk intracrystalline conductance. The equation (2) (with minor reservations) can also be extended to the films with a predominance of closed-neck type contacts [72]. Sensors for minority gas detection operate in atmospheric air, i.e. the heated surface of the sensors is already densely populated with adsorbed oxygen (acceptor-type particles). This determinates electrical characteristics of the sensors in pure air, and this is the precondition for the electrical response

11

generation in the presence of the detected gas. Depending on the surface reaction dominating during gas exposures, it is possible to divide minority gases into two big classes (Table 1). Reducing gases actively react with surface preadsorbed oxygen particles (i.e., molecular oxygen, atomic oxygen). As a result, the surface concentration of the adsorbed oxygen particles decreases, a number of electrons is delocalized, and the electrical resistance of the semiconductor sensitive element changes (as a consequence of an increment in the concentration of free electrons within the semiconductor and of a variation in the surface potential barrier height, Equations 1, 2). For example, in the presence of carbon monoxide, the reaction may occur

CO (ads ) + O (−ads) → CO 2 ↑ + e − , where

O (−ads )

takes place, where

NO 2−( ads)

is the complex of an

adsorbed nitrogen dioxide molecule with a localized electron from the conduction band. As a result of the reaction, the value of localized electrons increases, and the electrical resistance of the semiconductor sensitive element changes: it increases in the case of n-type semiconductor and decreases in the case of ptype semiconductor (Table 1).

(3)

is the complex of an adsorbed oxygen

atom with a localized electron from the conduction band, CO2 (carbonic gas) is the product of the surface reaction of CO with oxygen, and e- is a free electron in the conduction band. In the case of an n-type semiconductor sensitive element, the reaction (3) will result in an electrical resistance decrease (Fig. 6). In the case of a p-type element, the result will be opposite: the decrease in majority charge-carrier (hole) concentration will cause an increase in the resistance of the element. Oxidizing gases, such as chlorine and nitrogen dioxide, act as acceptors of electrons. Upon exposure to these gases, i.e. upon adsorption of the molecules possessing a higher electron affinity compared to the molecule of oxygen, the total amount of acceptors on the semiconductor surface increases. For example, in the presence of nitrogen dioxide, the reaction

NO 2 (ads) + e − → NO −2( ads)

,

(4)

Figure 6. Schematic representation of the chemisorption processes on the surface of an n-type semiconductor thick film

Elevated temperature is a crucial factor for the operation of semiconductor chemical sensors. The physicochemical process inducing an electrical response of a metal oxide semiconductor to a stimulus gas, i.e. strong chemisorption, intensively proceeds at high temperatures (at 150-500°C). The elevated temperature is necessary: first, to free enough carriers for the charge transport within semiconductor crystallites; and secondly, to impart enough energy to the free carriers for escaping surface energy barriers. The operational temperature

12

could be an instrument for the improvement of the sensors’ selectivity. The above discussion has demonstrated that semiconductor sensors are nonselective in principle. They react to various gases, both reducing and oxidizing. The proper choice of the work temperature for semiconductor sensors could radically narrow the range of the gases capable to activate the sensor’s electrical response. In additions to the transport function and the function of free electron delivery to adsorbate, temperature regulates lifetime of adsorbates on the semiconductor surface and could intensify or, conversely, depress the sensitivity of the sensor to one or another gas.

Resistance responses of semiconductor towards minority gases in air

Table 1 sensors

Semiconductor

Reducing gases

Oxidizing gases

n-type

Resistance decrease

Resistance increase

p-type

Resistance increase

Resistance decrease

gas indicates precisely how complete the combustion of the air/fuel mixture is in the combustion zone, thus the oxygen concentration is the best reference point for controlling the air/fuel ratio (λ-ratio). The sensors operate at high temperatures – at 500°C and above. Thick films of TiO2, SrTiO3, CeO2, or ZnO are in the most common use as sensitive elements of the sensors [76, 77]. The conductivity of the oxides is highly determined by the concentration of lattice defects. The concentration of the defects, in its part, depends on the environmental conditions in general and on atmospheric oxygen partial pressure in particular. The atmospheric oxygen dependence becomes stronger when a metal oxide is at elevated temperature: at 500°C and above a dynamic equilibrium is established between the oxygen concentration in the atmosphere and the bulk stoichiometry of the oxide. Variations in oxygen partial pressure change stoichiometry of the oxide, and as a result, vary the conductivity of the material. The relationship between oxygen partial pressure and the electrical conductivity of a metal-oxide sensor may be written as

 − E A  1n  ⋅ pO2 ,  kT 

σ = Α ⋅ exp

(5)

where σ is the electrical conductivity, A is a constant, The other most common methods for enhancing selectivity of semiconductor chemical sensors are: use of catalysts and promoters, use of surface additives promoting a specific adsorption, and use of various molecular filters [73, 74]. The implementation of SMO sensor arrays combined with appropriate pattern recognition tools is also considered as a promising approach to compensate for the lack of selectivity and to provide coverage for multiple types of gases [75]. Oxygen sensors (also referred to as “lambda sensors”) respond to changes in oxygen partial pressure and are typically used for monitoring exhaust gases. The amount of oxygen in the exhaust

EA is the activation energy for conduction,

pO2

is

oxygen partial pressure, and n is a constant determined by the dominant type of the bulk defects involved in the equilibrium between atmospheric oxygen and the sensitive element [2, 78]. Provided that doubly charged oxygen vacancies are the dominating defects in the sensitive element, the oxygen equilibrium between oxide and atmosphere can be written as

Ο Οx ↔ VΟ•• + 1 2Ο 2 + 2e −

,

(6)

13

where

Ο Οx

is an oxygen atom in its site of the oxide

lattice, and

VΟ••

is a doubly charged oxygen

vacancy. The equilibrium constant is

[ ]

[ ]

K = VΟ•• ⋅ [e] ⋅ pO2 2

1

2

,

(7)

where [e] is the concentration of free electrons.

[V ] = 2[e] , the equation •• Ο

Taking into account that (3) can be rewritten as

[ ]

K = 2[e] ⋅ [e] ⋅ p O2 2

hence [e] ∼

complex for realization than the deposition and processing of planar-type elements.

1

2

,

(8)

Screen-printed thick-film semiconductor chemical sensors have been commercially available since the mid 1990s through Figaro Engineering Inc. [39], Capteur Sensors and Analysers Ltd [40], and UST Umweltsensortechnik GmbH [82]. The marketed sensors are offered for the applications in environmental monitoring, as well as in air quality, automotive, medical, industrial, aerospace, and similar gas phase monitoring. Gases that can be sensed effectively include ammonia, hydrogen, carbon monoxide, nitric oxides, ozone, sulphur dioxide, hydrogen sulphide, methane, propane, and many others.

−1

pΟ 2 6 , and n = -6.

4. DESIGNS Semiconductor chemical sensors have been marketed since 1968. The sensors as an industrial product were first developed by Naoyoshi Taguchi [79-81]. Founded by Taguchi in 1962, Figaro Engineering Inc. began to mass-produce and sell semiconductor gas sensors based on n-type metal oxides in October 1968. Named for the inventor, Taguchi Gas Sensors (TGS) are still on the market [39]. A typical example of Taguchi-type sensor design is shown in Fig. 7. The embodiment consists in a ceramic tube that contains a heater coil inside. A semiconductor material layer and electrodes are formed on the tube. TGS sensor can be mounted in TO-like socket. There are two obvious drawbacks to this design. The first is the significant loss of power due to the poor contact of the heater coil with the ceramic tube, and hence there will be always excessive power consumption by the sensor. Another drawback is the geometry of the device: the film deposition onto cylindrical surfaces, as well as the following processing of cylindrical elements, are always more

Figure 7. Taguchi-type sensor. The heater is embedded in a ceramic tube and the semiconductor material is mounted on the tube with two pre-printed electrodes

A thick-film chemical sensor is a layered thick-film structure formed on a ceramic substrate. Alumina is typically used as a substrate material. This classical substrate material, which is widely applied in HIC production, has also become a good basis for the thick-film chemical sensor structures. Alumina is cheap and combines suitable for the sensors mechanical characteristics, electrophysical

14

properties, and chemical inertness. What is most important is that alumina is able to withstand high temperatures for a long time, i.e. all of the above characteristics – mechanical, electrical, chemical – remain satisfactory during the long-term exploitation of the substrates at elevated temperatures typical for active chemical sensors (150-700°C). The thickness and shape of the substrate vary from sensor to sensor depending on the design and application. The thickness ranges from 200 to 600 µm. As for the shape, different approaches have been considered. Square is the most common shape for sensor substrates (Fig. 8). Some applications however – in harsh and vibrating environment of a vehicle exhaust system, for example – need rigid encapsulations. For this use elongated sensors have been developed [83, 84]. The elongated sensors based on the substrates with side ratios about 1 to 10 are manufactured by Siemens AG [85] and Robert Bosch GmbH [86] for automotive applications.

Figure 8. Designs of thick-film semiconductor chemical sensors. (a) 2-d design. (b) 1-d design: the layer of the insulator (typically Al2O3) separates the SMO film from the metal layer of the heater and contacts; the electrical contacts to the heater and contacts are provided through the windows in the insulator

The need to sustain stable and high operating temperatures dictates the embedding of various

heaters in chemical sensors. Typically the heater and the sensitive element are formed on the opposite sides of the substrate. The design practically eliminates the possibility of the electrical crosstalk between the heater and the sensitive elements, makes it feasible to operate chemical sensors at very high temperatures. However, the necessity to conduct heat through the comparatively bulky substrate inevitably leads to an excessive power consumption of the device. The giant - for microelectronics power consumption (from 200 mW) hampers the application of the sensors in the field. The design with the location of the heater and the sensitive element on the same side of the substrate but isolated with the reliable dielectric layer looks more attractive from the standpoint of energy saving. The variations of this sensor design are actively used in the newest developments from Figaro Engineering Inc. and UST Umweltsensortechnik GmbH [39, 82]. In spite of the proximity of the heater and the sensitive element, operating temperatures up to 600°C are obtainable for this sensor architecture owing to the reliable thermally stable thick-film isolators [82]. The material typically used for fabricating of the thick-film heaters is platinum. Pt films meet many stringent requirements imposed on sensor heaters - heaters based on platinum demonstrate an excellent long-term stability even operating in aggressive environments and at extremely high temperatures. Some manufacturers also use the hybrid design approach in their chemical sensor developments: it has become common practice to apply Pt thin films as heaters and contacts, whereas sensitive elements are formed by thick-film methods (screen-printing or drop-deposition) [82]. Thick films of RuO2 are also widely applied as heaters [39, 56, 87, 88]. The shape of sensor heaters varies from sensor to sensor and may be either an ordinary rectangle or a sophisticated meander (Fig. 8). In order to control operating temperature, many developers use integrated resistance thermometers based on gold [89] or platinum [90, 91], however the dominating tendency in the modern chemical sensor design is the use of heaters as thermometers [39, 40,

15

57, 82, 92-94]. Applying simple temperature-control feedback circuits [95-97], it is possible to maintain constant heater temperature with the accuracy and stability satisfactory for the most of chemical sensor applications. Sensor sensitive elements are deposited onto preformed gold or platinum electrodes. Either thickfilm or thin-film structures are used as electrodes. Interdigitated geometry is employed most often (Fig. 8); the feature dimensions of these electrodes are typically 150-200 µm. The electrodes of a simple rectangular geometry are also in common use [82]. Marketed chemical sensors are encapsulated in various packages (Fig. 9). Similar to other microelectronic devices, the main objective of the package is to protect the sensor from mechanical damage. But in contrast to the other devices purposely isolated from atmosphere - chemical sensors have to freely interact with environment.

Figure 9. Packaged thick-film semiconductor chemical sensors

Therefore the sensor cases, made either of plastic (e.g., Nylon 66 [39]) or plated steel (e.g., Ni-plated steel, NiCu-plated steel [39, 82]), have openings covered with several layers of stainless steel gauze. The flameproof gauze allows gases to reach the sensor while ensuring that the hot sensitive element of the sensor cannot ignite an explosive gas mixture.

5. CONCLUSIONS Thick-film technology has become an ideal basis for deploying manufacture of various chemical microsensors. The application of the technology in semiconductor chemical sensors has been especially successful. Thick-film polycrystalline chemoresistors fabricated on the base of metal-oxide suspensions demonstrate remarkable sensitivity towards chemical composition of atmosphere, fast response times, and ability to be tuned for the detection of either a single agent or a group of chemical agents. It has been established that using powder synthesis routes (which are an initial and basic phase for any thick-film technique), a precise control is possible over grain size, phase composition, dopant stoichiometry and other characteristics influencing performance of the sensitive materials for semiconductor chemical sensors. Screen-printing which is a primary method for the fabrication of thick films in microelectronics has been successfully adapted for the production of various thick-film sensors. The adaptation of the method for the manufacturing of miniature chemical sensors has involved further improvement in both quality of applied screens and quality of printing machine adjustment. Over the past decade, screen-printed thick-film semiconductor chemical sensors have experienced significant growth and market acceptance, driven largely by their robustness, diminutiveness, simplicity in production, and relative cheapness. Screen-printed devices have a serious drawback that hampers their applications in portable or handheld devices. They usually operate at elevated temperatures – at 150°C and above, and up to 1000°C. Sensors based on traditional platforms, i.e. on ceramic substrates equipped with resistance heaters, typically consume too much power to be used in the field instruments. Early in the 1990th silicon-based micromachined chemical sensors platforms (microhotplates) were invented. These platforms reduced power consumption by a factor of

16

10, compared to the thick-film sensors on alumina substrates. The development of the chemical sensors consuming less than 100 mW was made possible. However, it has taken some time to comprehend that the combination of silicon microhotplates with sensitive metal-oxide thick films could be extremely profitable. The following effort in hybridization of two technologies, undertaken by various research groups in Germany, Spain, Italy, Switzerland, and the USA, have resulted in the development of the diversity of methods to deposit thick films onto microhotplates. Nowadays at least two companies – MicroChemical Systems SA (MiCS) [98] and AppliedSensor [7] – market semiconductor gas sensors that combine silicon microstructures with nanostructured thick films. This approach in the development of semiconductor chemical sensors is likely to be dominating for the next several years.

GLOSSARY Chemical sensor “A small device that as the result of chemical interaction or process between the analyte gas and the sensor device transforms chemical or biochemical information of quantitative or qualitative type into an analytically useful signal” [1]. Chemical sensors are subdivided into six classes depending on the nature of the response upon the interaction: electrochemical, electrical, optical, magnetic, thermometric, and mass-sensitive. In their part, electrical chemical sensors subdivided into semiconductor chemical sensors and chemically sensitive field effect transistors (CHEMFET). Semiconductor chemical sensor A chemical sensor with a sensitive element made of semiconductor whose electrical properties (conductivity, work function etc.) vary with the concentration of a detected chemical agent. Thick film A polycrystalline film deposited onto a supporting material (a substrate) as a suspension (paste, ink) and solidified by an appropriate heat

treatment. The thickness of the films typically lies within the range from 1 to 100 µm. Screen-printing A technique for reproduction of a text or a pattern by forcing printing suspension (a paste, ink, or dye) through screen openings onto the printing surface. A screen for screen-printing is a polymer, silk, or metal mesh coated in the spacing areas with a layer impenetrable for the printing substance. In modern microelectronics, the technique is the main method for the deposition of thick films. Drop-coating (in semiconductor chemosensorics) is a technique for depositing droplets of metal-oxide suspensions onto sensor platforms. The deposition is realized using a precisely manipulated micropipette coupled with a microinjector regulating delivery of the deposited suspension. The technique is also referred to as microprinting or microdeposition. Sensor platform A basic microelectronic module for chemical sensor fabrication; either a substrate or a membrane (microhotplate) with an integrated heater and contact layers. Microhotplate A miniature sensor platform; a suspended three dimensional structure that is fabricated on silicon using standard complimentary metal-oxide semiconductor processing and is suspended from the bulk silicon by either surface or bulk micromachining. MIMIC. Micromolding in capillaries. MIMIC (also referred to as soft lithography) is a microfabrication technique for production of a pattern from a liquid material by filling microchannels of an elastomeric mold placed on the patterning surface. The popular elastomer for MIMIC is silicon rubber, polydimethylsiloxane (PDMS). The technique allows shrinking the size of semiconductor sensing element to 10 µm and hence forming several chemical sensors on a single microhotplate.

17

Hybrid integrated circuit (HIC) A microelectronic device comprising thick-film elements on the ceramic substrate and add-on miniature discrete components. The devices are widely used in areas such as televisions, telecommunications, automotive electronics, military and space technologies. Element (of HIC) A thick-film microdevice, an integral part of HIC, fabricated by screen-printing technique. Thick-film elements could be either passive elements (e.g., thick-film resistors, thick-film capacitors) or active elements (e.g., thick-film sensitive elements in sensors). Component (of HIC) A discrete microdevice, a constituent of HIC, attached to the substrate with pre-formed thick-film interconnections and thickfilm elements by either soldering or bonding. Components of HIC could be either passive components (resistors, capacitors etc.) or active components (transistors, diodes, integrated microcircuits etc.) Thick-film suspension (paste, ink) A suspension for screen-printing thick films; a disperse system containing particles of film-forming inorganic materials (disperse phase) suspended in an organic binder (disperse medium). The disperse phase contains a functional material and a binding agent (either a glass frit or an oxide). Organic binder or Organic vehicle A disperse medium in thick-film suspensions. Organic binders are removed out of thick films after deposition by heat treatment in an intensive airflow. Functional material (in thick-film sensor technology) An ingredient of the thick-film suspension, which dominates the electrical and sensing properties of the thick film. Glass frit An ingredient of the thick-film suspension; a glass powder; a vitreous substance serving as binding agent both for grains of the functional

material within the thick film and of the whole fired thick film to its substrate. µ-Screen A screen for precise screen-printing (down to 50 µm lines and spaces) that combines an active central area made of stainless steel foil in which a pseudo-mesh of the required pattern is etched and a conventional stainless steel mesh as a frame and support for the active area. Adsorption “A process by which particles are taken up on the surface of a solid by chemical or physical action”[68]. Adsorbent is a solid that takes up the gas (vapor). Adsorbate is a gas (vapor) that is taken up on the surface of the adsorbent. Physical adsorption Adsorption caused by forces of electrostatic nature (e.g., by Van der Waals forces). Chemical adsorption Adsorption caused by covalent forces with plausible involvement of ionic interactions. The chemisorbed particle and the adsorbent together form a unified quantum mechanical system. Weak chemical adsorption is realized without free current-carriers of the adsorbent lattice, and the adsorbed particle remains neutral. Strong chemical adsorption takes place, if the free lattice electron (or the free hole) is localized near the chemisorbed particle. The strong bond is conditioned by the exchange interaction between free carriers and adparticles. A charged form of chemisorption.

REFERENCES 1.

J. R. Stetter, W. R. Penrose, S. Yao, J. Electrochem. Soc., 150 (2), S11 (2003).

18

2.

3. 4.

5.

6. 7. 8.

9. 10. 11.

12.

13.

14.

15.

16. 17.

P. T. Moseley, Thick Film Semiconductor Gas Sensors in Handbook of Sensors and Actuators, V. 1, Thick Film Sensors edited M. Prudenziati, Elsevier, Amsterdam – Lausanne - New York – Oxford – Shannon Tokyo (1994), p. 289 N. M. White and J. D. Turner, Meas. Sci. Technol., 8, 1 (1997). D. Briand, A. Krauss, B. van der Schoot, U. Weimar, N. Barsan, W. Göpel, and N. F. de Rooij, Sens. Actuators, B, 68, 223 (2000). J. Cerdà, A. Cirera, A. Vilà, A. Cornet and J. R. Morante, Thin Solid Films, 391, 2, 265 (2001). Thomas K. H. Starke, Gary S. V. Coles, Hans Ferkel, Sens. Actuators, B, 85, 239 (2002). Tom Aiken, Sensors, 21, 40 (2004). Xusheng Wang, go Sakai, Kengo Shimanoe, Norio Miura, Noboru Yamazoe, Sens. Actuators, B, 45, 141 (1997). Guojia Fang, Zuli Liu, Chuanqing Liu, Kailun Yao, Sens. Actuators, B, 66, 46 (2000). N. S. Baik, G. Sakai, N. Miura and N. Yamazoe, Sens. Actuators, B, 63, 74 (2000). R. E. Cavicchi, R. M. Walton, M. AquinoClass, J. D. Allen and B. Panchapakesan, Sens. Actuators, B, 77, 145 (2001). Warren S. Dabney, Nicholas E. Antolino, Brian S. Luisi, Annette P. Richard, Doren D. Edwards, Thin Solid Films, 411, 192 (2002). K. Galatsis, Y. X. Li, W. Wlodarski, E. Comini, G. Sbeveglieri, Sens. Actuators, B, 83, 276 (2002). C. Garzella, E. Bontempi, L. E. Depero, A. Vomiero, G. Della Mea and G. Sbeverglieri, Sens. Actuators, B, 93, 495 (2003). Jun Tamaki, Yoshihiro Yamada, Yoshifumi Yamamoto, Masao Matsuoka and Isao Ota, Sens. Actuators, B, 66, 70 (2000). Lu Xuchen, Xu Tingxian and Dong Xianghong, Sens. Actuators, B, 67, 24 (2000). J. Shieh, H. M. Feng, M. H. Hon, H. Y. Juang, Sens. Actuators, B, 86, 75 (2002).

18. S. Shukla, S. Patil, S. C. Kuiry, Z. Rahman, T. Du, L. Ludwig, C. Parish and S. Seal, Sens. Actuators, B, 96, 343 (2003). 19. Martin Heule, L. J. Gauckler, Sens. Actuators, B, 93, 100 (2001). 20. Giorgio Sberveglieri, Sens. Actuators, B, 6, 239 (1992). 21. Lie-yi Sheng, Zhenan Tang, Jian Wu, Philip C. H. Chan and Johnny K. O. Sin, Sens. Actuators, B, 49, 81 (1998). 22. Shinichiro Tanaka and Takao Esaka, Materials Research Bulletin, 35, 2491 (2000). 23. E. Comini, G. Sberveglieri and V. Guidi, Sens. Actuators, B, 70, 108 (2000). 24. A. Vomiero, G. Della Mea, M. Ferroni, G. Martinelli, G. Roncarati, V. Guidi, E. Comini and G. Sberveglieri, Material Science and Engineering, B, 101, 1-3, 216 (2003). 25. L. Bruno, C. Pijolat and Lalauze, Sens. Actuators, B, 18, 195 (1994). 26. S. Semancik, R. E. Cavicchi, K. G. Kreider, J. S. Suehle and P. Chaparala, Sens. Actuators, B, 34, 209 (1996). 27. C. C. Chai, J. Peng and B. P. Yan, Sens. Actuators, B, 34, 412 (1996). 28. Sanjeev Majoo, J. L. Gland, K. D. Wise and J. W. Schwank, Sens. Actuators, B, 36, 312 (1996). 29. M. A. El Khakani, R. Dolbec, A. M. Serventi, M. C. Horrillo, M. Trudeau, R. G. SaintJacques, D. G. Rickerby, I. Sayago, Sens. Actuators, B, 77, 383 (2001). 30. R. Dolbec, M. A. El Khakani, A. M. Serveni, M. Trudeau and R. G. Saint-Jacques, Thin Solid Films, B, 419, 230 (2002). 31. R. Dolbec, M. A. Khakani, A. M. Serveni, R. G. Saint-Jacques, Sens. Actuators, B, 93, 566 (2003). 32. N. Barsan, M. Schweizer-Berberich, W. Göpel, Fresenius J. Anal. Chem., 365, 287 (1999). 33. Sang Woo Lee, Ping Ping Tsai, Haydn Chen, Sens. Actuators, B, 67, 122 (2000).

19

34. Isolde Simon, Nicolae Bârsan, Michael Bauer, Udo Weimar, Sens. Actuators, B, 73, 1 (2001). 35. Th. Becker, S. Ahlers, Chr. Boschv.Braunmühl, O. Kiesewetter, Sens. Actuators, B, 77, 56 (2001). 36. Céline Nayral, Eric Viala, Vincent Collière, Pierre Fau, François Senocq, André Maisoonnat, Bruno Chaudret, Applied Surface Science, 164, 219 (2000). 37. B. Marquis, A. Tomchenko, J. Allen, Abstract Book of the 9th International Meeting on chemical Sensors, Boston, USA, p. 206 (2002) 38. A. Chiorino, G. Ghiotty, F. Prinetto, M. C. Carotta, D. Gnani, G. Martinelly, Sens. Actuators, B, 58, 338 (1999). 39. Figaro an ISO9001 company. The information is available via http://www.figarosensor.com. 40. City Technology. The Power Behind Gas Detection. The information is available via http://www.citytech.com. 41. Thick Film Sensors edited M. Prudenziati, Elsevier, Amsterdam – Lausanne - New York – Oxford – Shannon - Tokyo (1994). 42. S. Semancik, R. E. Cavicchi, M. Gaitan, J. S. Suehle, US Patent No. 5,345,213 (issued 1994). 43. M. Gaitan, J. S. Suehle, S. Semancik, R. E. Cavicchi, US Patent No. 5,464,966 (issued 1995) 44. S. Semancik, R. E. Cavicchi, M. C. Wheeler, J. E. Tiffany, G. E. Poirier, R. M. Walton, J. S. Suehle, B. Panchapakesan, D. L. DeVoe, Sens. Actuators, B, 77, 579 (2001). 45. D. Vincenzi, M. A. Butturi, V. Guidi, M. C. Carotta, G. Martinelli, V. Guarnieri, S. Brida, B. Margesin, F. Giacomozzi, M. Zen, G. U. Pignatel, A. A. Vasiliev, A. V. Pisliakov, Sens. Actuators, B, 77, 95 (2001). 46. D. Vincenzi, M. A. Butturi, M. Stefancich, C. Malagù, V. Guidi, M. C. Carotta, G. Martinelli, V. Guarnieri, S. Brida, B. Margesin, F. Giacomozzi, M. Zen, A. A.

47.

48.

49.

50.

51.

52.

53. 54.

55. 56.

Vasiliev, A. V. Pisliakov, Thin Solid films, 391, 288 (2001). J. Puigcorbé, A. Vilà, J. Cerdà, A. Cirera, I. Gràcia, C. Cané, J. R. Morante, Sens. Actuators, A, 97-98, 379 (2002). Albert Cirera, New technologies and their characterization for nanostructured SnO2 gas sensor devices, Doctoral Thesis, Universitat de Barcelona, Spain (2000), p.132. N. M. White, Thick Film Technology in Handbook of Sensors and actuators, V. 1, Thick Film Sensors edited M. Prudenziati, Elsevier, Amsterdam – Lausanne - New York – Oxford – Shannon - Tokyo (1994), p. 3. V. Leonhard, Gas Sensors Based on Solid Ionic Conductors in Handbook of Sensors and actuators, V. 1, Thick Film Sensors edited M. Prudenziati, Elsevier, Amsterdam – Lausanne - New York – Oxford – Shannon Tokyo (1994), p. 261. N. Hampp, C. Eppelsheim, and A. Silber, Thick-film Biosensors in Handbook of Sensors and actuators, V. 1, Thick Film Sensors edited M. Prudenziati, Elsevier, Amsterdam – Lausanne - New York – Oxford – Shannon Tokyo (1994), p. 341. M. Prudenziati, Pastes, Inks and Slurries in Handbook of Sensors and actuators, V. 1, Thick Film Sensors edited M. Prudenziati, Elsevier, Amsterdam – Lausanne - New York – Oxford – Shannon - Tokyo (1994), p. 113. Chris Robertson, R. D. Shipton, D. R. Gray, Sensor Review, 19, N1, 33 (1999). “Tomorrow’s screen technology available today” 18 March, 2004, the information is available via http://www.dek.com/homepage.nsf/press_re lease_list!openform A. Tomchenko, Sens. Actuators, B, 68, 48 (2000). A. A. Tomchenko, I. L. Emelianov, Tungsten trioxide based thick-film NO sensors: design and investigation in Eurosensors XII edited N. M.

20

57.

58.

59.

60.

61.

62.

63.

64. 65. 66. 67.

68.

69.

White, Institute of Physics, Bristol (1998), V1, p.253. H. Debéda, P. Massok, C. Lucat, F. Ménil and J. L. Aucouturier, Meas. Sci. Tecnol, 8, 99 (1997). Nak Jin Choi, Yun Su Lee, Jun Hyuk Kwak, Joon Shik Park, Kwang Bum Park, Kyu-sik shin, Hyo Derk Park, Jae Chang Kim, Jeung Soo Huh, Duk Dong Lee, Chem. Sensors, 20, Supplement B, 814 (2004). Takuya Suzuki, Kenji Kunihara, Mitsuo Kobayashi, Soichi Tabata, Katsuki Higaki, and Hisao Ohnishi, Chem. Sensors, 20, Supplement B, 310 (2004). Kraig D. Mitzner, Jason Sternhagen, David W. Galipeau, Sens. Actuators, B, 93, 92 (2003). Zvi Boger, Douglas C. Meier, Richard E. Cavicchi, and Steve Semancik, Sensor Letters, 1, 86 (2003). D. C. Meier, R. E. Cavicchi, M. W. Ellzy, Z. Boger, S. Semancik, Chem Sensors, 20, Supplement B, 308 (2004). S. Semancik, R. E. Cavicchi, D. C. Meier, K. D. Benkstein, G. Li, M. A. Ryan, M. L. Holmer, K. Manatt, A. Kisor, and C. J. Taylor, Chem Sensors, 20, Supplement B, 314 (2004). Ping Ping Tsai, I-Cherng Chen and Chao-Jen Ho, Sens. Actuators, B, 76, 380 (2001). A. A. Tomchenko, V. V. Khatko, I. L. Emelianov, Sens. Actuators, B, 46, 8 (1998). P. T. Moseley, Sens. Actuators, B, 149 (1992). V. Ya. Sukharev, Physical and Chemical Basics of the Method of Semiconductor Sensor, in Handbook of Sensors and Actuators, V. 4, Semiconductor Sensors in Physico-Chemical Studies edited L. Yu. Kuprianov, Elsevier, Amsterdam – Lausanne - New York – Oxford – Shannon - Tokyo (1996), pp. 10, 25. Grolier New Webster’s Dictionary, Grolier Incorporated, Danbury, Connecticut, USA (1992), p. 6. V. Lantto, P. Romppainen and S. Leppävuori, Sens. Actuators, 14, 149 (1988).

70. J. F. McAleer, P. T. Moseley, J. O. W. Norris and D. E. Williams, J. Chem. Soc., Faraday Trans. I, 83, 1323 (1987). 71. N. M. Beekmans, J. Chem. Soc., Faraday Trans. I, 74, 31 (1978). 72. S. Roy Morrison, Sens. Actuators, 2, 329 (1982). 73. S. Roy Morrison, Sens. Actuators, 12, 425 (1987). 74. V. Ya. Sukharev, Physical and Chemical Basics of the Method of Semiconductor Sensor, in Handbook of Sensors and Actuators, V. 4, Semiconductor Sensors in Physico-Chemical Studies edited L. Yu. Kuprianov, Elsevier, Amsterdam – Lausanne - New York – Oxford – Shannon - Tokyo (1996), p. 8. 75. T. C. Pearce, S. S. Schiffman, H. T. Nagle, J. W. Gardner, Handbook of Machine Olfaction: Electronic Nose Technology, Wiley-VCH, Weinheim (2003), pp. 134, 347. 76. Yulong Xu, Xiaohua Zhou, O. Toft Sorensen, Sens. Actuators, B, 65, 2 (2000). 77. P. I. Gouma, S. A. Akbar, and M. J. Mills, JOM-e, 50, No 11, (1998) available via: http://www.tms.org/pubs/journals/JOM/981 1/Gouma/Gouma-9811.html#Ref45. 78. P. Kofstad, Non-Stoichiometry, Diffusion and Electrical Conductivity in Binary Metal Oxides, Wiley-Interscience, New York (1972). 79. S. Taguchi, US Patent No. 3,625,756 (issued 1971). 80. S. Taguchi, US Patent No. 3,695,848 (issued 1972). 81. S. Taguchi, US Patent No. 3,631,436 (issued 1971). 82. UST, Umwelt Sensor Technik. High Tech Sensorics – made for your success. The information is available via http://www.umweltsensortechnik.de/ . 83. Maximilian Fleischer and Hans Meixner, Sens. Actuators, 43, 1 (1997). 84. Ralf Moos, Ralf Müller, Carsten Plog, Aleksandar Knezevic, Holger Leye, Eckard Irion, Tillmann Braun, Klaus-Jürgen

21

85.

86.

87. 88.

89. 90.

91.

92. 93.

94.

95. 96. 97.

98.

99.

Marquardt, Klaus Binder, Sens. Actuators, B, 83, 181 (2002). Siemens VDO Automotive. NOx sensor: Equipped for future standards The information is available via http://www.siemens.com/index.jsp Bosch Premium Planar Type Sensors. Bosch ceramic technology affords best-in-class performance. The information is available via http://www.boschusa.com V. K. Singh, R. Dwivedi, S. K. Srivastava, Microelectronics Journal, 27, 531 (1996). K. Oto, A. Shinobe, M. Manabe, H. Kakuuchi, Y. Yoshida, T. Nakahara, Sens. Actuators, B, 77, 525 (2001). Radu Ionescu, Eduard Llobet, Sens. Actuators, B, 81, 289 (2002). Guiliano Martinelli, M. Cristina Carotta, Enrico Traversa, and Giovanna Ghiotti, MRS Bulletin, No 6, 30 (1999). J. Hubálek, K. Malysz, J. Prášek, X. Vilanova, P. Ivanov, E. Llobet, J. Brexmes, X. Correig, Z. Svěrák, Sens. Actuators, B, 101, 277 (2004). Jong-In Yang, H. Lim, Sang-Do Han, Sens. Actuators, B, 60, 71 (1999). Alexey A. Tomchenko, Gregory P. Harmer, Brent T. Marquis, John W. Allen, Sens. Actuators, B, 93, 126 (2003). S.R. Aliwell, J. F. Halsall, K. F. E. Pratt, J. O’Sullivan, R. L. Jones, R. A. Cox, S. R. Utembe, G. M. Hansford and D. E. Williams, Meas. Sci. Technol., 12, 684 (2001). J. Watson, Sensor Review, 14, No 1, 20 (1994). Peter McGeehin, Sensor Review, 20, No 2, 106 (2000). J. W. Allen, B. T. Marquis, D. J. Smith, Sensors 2002, Proceedings of IEEE, 2, 1260 (2002). MicroChemical Systems. Semiconductor gas sensors for improved air quality. The information is available via http://www.microchemical.com/ Adapted from Sensors and Actuators B: Chemical, Vol. 93, No 1-3, Martin Heule, L. J.

Gauckler, “Miniaturised arrays of tin oxide gas sensors on single microhotplate substrates fabricated by micromolding in capillaries”, pp. 100-106, © 2003, with permission from Elsevier. 100. Reprinted from Sensors and Actuators B: Chemical, Vol. 46, A. A. Tomchenko, V. V. Khatko, I. L. Emelianov, “WO3 thick-film gas sensors”, pp. 8-14, © 1998, with permission from Elsevier.

AFTERWORD: MORE ABOUT MICROPRINTING AND PRINTED MICROSENSORS The above text was written in the mid of 2004. Ironically, several months later, in December of 2004, the author found himself actively implementing the forecast he had made in his Conclusions – the one on the inevitable merging of two microelectronic technologies - silicon technology and thick-film technology - into a newest hybrid process which would let begin manufacturing truly miniature and low-power consuming chemical sensors. The author was provided with SiO2 microhotplate platforms and was requested to develop a reliable method for transforming nanoparticle powders into ‘on-theplatform nano-crystalline gas-sensitive films’. As always, the time frames were tight – three months at most. Also, a) everything had to be reasonably priced (read, cheap), and b) the deposition equipment had to be easy for operating and servicing. In the end, the excellent reproduction in the characteristics for the manufactured sensors would have to be demonstrated. Under those circumstances, the author had to promptly reevaluate his list of the techniques for depositing suspensions onto microplatforms (see above) and to pick the most practical one as a base for the development. (Non-suspension deposition methods, i.e., thin-film methods, were excluded by the requester ab initio).

22

After careful consideration, microprinting had been chosen as a base technology. The method satisfied almost all of the above requirements. Though at the time, it wasn’t clear whether it would be possible to apply the method in depositing on the provided platforms - as small as 100 micrometers and as thin as 3.5 micrometers. Originally microprinting had been developed for dispensing suspensions onto significantly larger and more robust platforms [see 5, 47, and 48 in References above]. It had taken two months of experimenting, before it was eventually realized that the precise printing was possible. In the end of January 2005 the first batch of chemical microsensor arrays was fabricated. The key element of the process proved to be the optimal combination of the deposition parameters and the printed suspensions’ characteristics. The most important precondition had been formulated for the successful, reliable and reproducible microprinting on to 100µmplatforms: powder precursors have to be truly nanometric – no blocks of nanocrystals, no agglomeration.

microhotplate platforms was equipped with interdigitated platinum sensor contacts. Gassensitive films were microprinted onto the top of the platforms. Each of the provided devices – which, in fact, were prototypes of future microsensor arrays – integrated sixteen microhotplate platforms (Fig. A1). Microprinting was performed using a MINJ-D Tritech micro-injector (Tritech Research Inc.) coupled with micropipette positioned over the µHPs by means of a Narishige MN-151 micromanipulator (Fig. A2). The process was performed under a microscope. Metal-oxide suspensions were prepared by mixing nanocrystalline oxide powder precursors with propylene glycol. The thin organic binder provided easy delivery of the nanomaterials through the micron orifices of the micropipettes and was easily removed from the deposits by drying at room temperature. Narishige PC-10 puller was used for the fabrication of the pipettes (Fig. A3); 1mm × 90mm glass capillaries were used for the fabrication.

The method had been reported in MRS Proceedings [1] and at the 34th Northeastern Regional Meeting of ACS, (Binghamton, USA, October 2006). The below text summarizes the results of the author’s experiments in microprinting. The provided microhotplate platforms had a design outlined in the above subsection 2.2 - with only one exception: they did not have heat-distributing plates - which are just an additional metal layer incorporated into SiO2 over the heaters. The design modification had further reduced thickness of the platforms, reduced power consumption, and improved control over the sensitive element’s temperature. (Well, it also had slightly increased temperature gradient from the centre to periphery of the platform - I realized that). In other words, the author dealt with 100µm × 100µm SiO2 microplatforms with incorporated poly-Si heaters only. The platforms had been suspended on four corner SiO2 microbridges over the Si-substrate by front-side micromachining. The top surface of the

Figure A1. Optical micrograph of a microhotplate array (shows four platforms of a 16-platform array)

After microprinting, the samples (i.e., the µHP arrays) were dried at room temperature, and then were placed into a test-chamber where the deposited films were annealed using the integrated microplatform heaters in airflow of 200cm3 min-1.

23

A trapezoid temperature profile with heating and cooling ramps of 5°C min-1 and a 10-minute exposure at a peak temperature was used for annealing.

Figure A2. a) Photograph of a microprinting station: metaloxide suspensions are delivered onto µHPs through glass micropipette; micropipette is moved using XYZmicromanipulator; b) Schematic representation of µHP platform; each microplatform measures 100 µm × 100 µm; c) “Nanosnow”: nanoparticles chaotically move inside of the deposited droplet of the suspension, and gradually, as the solvent (organic binder) evaporates, they settle down and form the “proto-film” on the platform.

Figure A3. Microscopic image of pipettes for microprinting: the pipettes touch surface of the ceramic substrate with 50µm platinum interdigitated contacts

As underlined above, the most challenging objective for the research was the development of the reliable procedure for the precise delivery of the sensitive materials onto the microhotplate platforms without destroying the fragile µHP structures (about 3.5 µm thick) and avoiding any accidental ingress of the deposited suspension into the under-platform cavities. The deposited droplet of the suspension had to be the only substance contacting the surface of the platform, and for that the droplet had to be formed on the tip of the pipette prior the moment of microprinting. Such microprinting regimes had eventually been found: the author had successfully identified the appropriate combinations of the operating parameters for the microinjector system for each deposited material. The combinations varied in dependence of a) properties of a precursor powder and b) suspension composition (i.e., ratio precursor/binder). The most influencing parameters were the holding pressure of the microinjector and the size of the micropipette’s orifice. Fortunately, the deposited SMO materials – nanopowders of SnO2, WO3, In2O3 and CuO – had similar granulometric composition (20-60nm), and about the same density. As a result their suspensions demonstrated similar characteristics - as materials for microprinting. The similarity significantly simplified the search for the optimal parameters, which had eventually taken the author to the pressures from 0 to 1psi (0-7kPa, roughly) for 10wt. % suspensions. Precise adjustment for the pressure (in frames of the above range) was performed prior to printing - since every material was delivered through the individual pipette, and the pipettes had some scatter in the orifice sizes. Fig. A4 shows the microscopic image of a µHP array after printing and annealing.

24

In (1) Np is the total number of the SMO particles in the delivered droplet

Np = n×

πR 2 S p + 2S v

,

(2)

mp is the mass of the SMO nanoparticle (3.49×10-7 ng for WO3). n is the number of nanolayers in the film, which is easily converted into thickness of the film D. R is the radius of the droplet (50µm), Sp is the crosssection area for the SMO particle, and Sv is the area for the cross-section of the related void. Figure A4. Microscopic image of a µHP array after printing and annealing

Experimenting with the suspensions the author came to the simple method to control thickness of microprinted thin films by varying suspensions’ compositions [A1]. The idea was pretty straightforward: the thinner suspension, the less SMO material is delivered to the platform, the thinner the film formed on the surface. Taking into account the fact that the largest droplet safely deposited onto the µHP had the diameter defined by the geometry of the platform, and making some reasonable simplifications, the formulas for the approximate calculation of the film thickness vs. suspension composition was derived. The following assumptions had been made: a) the nanoparticles have the ideal spherical shape, b) the particles fall in close packing, layer after layer, i.e., for each particle there are two voids in the layer; c) the deposited droplet of the suspension have a semispherical shape with radius of 50µm. Using these assumptions, the mass of the delivered SMO was calculated as

m SMO = N p × m p .

(1)

The results of the calculations for the investigated WO3 films are shown in Table A1. The mass of binder, mb, was found as

 m mb = ρ b ×  4 6 πR 3 − SMO ρ SMO 

  . 

(3)

Table A1 Thickness of the microprinted WO3 films vs. suspension composition*

∗ ∗ ∗

dWO3 = 45.3nm; density of the particle ρWO3 = 7.16×10-12 ng/nm3; density of the binder ρb = 1.036×10-12 ng/nm3

As mentioned above, the reproduction in fabricated sensors’ characteristics was the main criterion in evaluating applicability of the method. The microprinting turned out an excellent deposition technique from this point of view. It was observed

25

remarkable reproduction in electrical/gas-sensitive properties for the fabricated microsensors (Figs A5, A6).

Figure A5. Gas-sensitive characteristics for 12 microsensors formed at the same µHP-chip: the twelve ZnFe2O4 sensors have the same baselines and act identically during the gas tests (left plot)

Figure A5. Gas-sensitive characteristics for 16 WO3 microsensors formed at the same µHP-chip

REFERENCES TO AFTERWORD A1. Alexey Tomchenko, and Brent Marquis, “Nanoparticle Metal-Oxide Films on Microhotplate Platforms: Fabrication and Gas-Sensitive Properties”, in Nanostructured Materials and Hybrid Composites for Gas Sensors and Biomedical Applications, edited by P.I. Gouma, D.

Kubinski, E. Comini, V. Guidi (Mater. Res. Soc. Symp. Proc. 915, Warrendale, PA, 2006), 275-281.