Moulson A.J., Herbert J.M. Electroceramics: Materials, Properties, Applications

Подождите немного. Документ загружается.

It is necessary to maintain close control over temperature because of its eﬀect

upon the base resistance of the semiconductor. With a platinum heater this can

be eﬀected to 0.1 8C through monitoring the resistance of the heater itself

using a Wheatstone bridge. The relatively high temperature coeﬃcient of

resistivity of RuO

complicates the control system.

A sensor (Fig. 4.49) under development for monitoring and activating control

over the air quality in automobile cabins well illustrates the status of present-day

technology, and how thick-ﬁlm circuit fabrication technology is exploited. The

target gases are CO and NO

emitted from petrol and diesel engines respectively.

The SnO

is doped with Fe

to increase NO

sensitivity and with V

and Pd

to enhance sensitivity to CO. The heater pad is RuO

. The two sensing areas have

a covering of porous silica to maintain cleanliness. The measured sensitivity to

is 0.5 p.p.m. and to CO 30 p.p.m., and the long-term stability is good.

212 CERAMIC CONDUCTORS

(a)

(b)

Fig. 4.48 (a) The classic ‘Taguchi sensor’; (b) the typical screen-printed planar format.

TEAMFLY

Team-Fly

There is a wide variety of chemical sensor systems developed for speciﬁcity

to certain gases and vapours. For example WO

-based sensors for O

and

,In

/SnO

for NO and SnO

/Al

for H

S. A comprehensive list is

given in [28].

Apart from chemical factors, the detailed physical form of the sensor can have

an important inﬂuence on selectivity. It is apparent from Fig. 4.48(b) that the

target gas has to diﬀuse through the microporous sensor material towards the

sensing electrodes. The gas ﬂow will depend upon the mean free path of the gas

molecules in relation to the diameter of the channels in the solid. The mean free

path of gas molecules usually present in a functioning sensor will be of the same

size order as the channel sizes making the gas interdiﬀusion process a diﬃcult

one to model mathematically. There will certainly be gas concentration gradients

across the sensor thickness, and diﬀerent gradients for diﬀerent gases. A graded

microporosity, readily accomplished via screen-printing is a route to inﬂuencing

selectivity.

The eﬀective thickness of sensing material overlying the electrodes will depend

upon the electrode spacing, the electric ﬁeld between the electrodes penetrating

more deeply into the sensor layer the more widely spaced the electrodes. Electrode

spacing is, therefore, an important design parameter which can inﬂuence

selectivity. Indeed, varying the electrode spacing can be used as a research tool

to probe the processes ongoing through the depth of the sensor layer.

Water vapour is an inadvertent and very variable ‘contaminant’ in most

sensing situations. The operating temperature of the sensor eliminates the risk of

the involvement of water molecules at the semiconductor surface but not that of

hydroxyl groups. In the case of the SnO

-based sensor it seems that OH



behaves

in much the same way as O



so that the eﬀect of the presence of water vapour

can be large. The eﬀects in the case of the Cr

/TiO

(p-type) sensor system are

far less marked, for reasons yet to be elucidated [30].

CERAMICS-BASED CHEMICAL SENSORS 213

Fig. 4.49 Twin tin oxide-based sensors for monitoring NO

and CO in automobile passenger

compartments. NB: alumina substrate size approximately 22.60.254 mm. (Adapted from

Oto, K. et al. (2001), Sensors and Actuators, B77 , 525–8.)

It is apparent that with so many variables the approach to designing eﬀective

sensors has to be essentially empirical but, nevertheless, one founded on a

working understanding of the underlying basic sciences.

At the research level there is interest in thin ﬁlm sensors. Thin ﬁlms, typically

100 nm thick, of doped tin oxide can be deposited by sputtering, evaporation,

sol–gel, etc. when subsequent annealing leads to the development of a

nanocrystalline porous structure. Although ‘thin’ ﬁlm technology, in comparison

with thick ﬁlm technology, would appear to oﬀer the advantage of relatively

lower production costs because of the avoidance of powder and screen-printing

paste processing steps, there are drawbacks. Thin ﬁlms oﬀer a much smaller

reactive area than sintered thick ﬁlm ceramics and so the sensitivity of sensors

fabricated from them is correspondingly lower; for the same reason they are

more susceptible to surface contamination.

The established and potential applications of ceramic gas sensors are many

and varied. Principal among them is CO sensing because of the highly toxic

nature of the gas. It is also the gas given oﬀ in the early stages of a ﬁre and so its

detection can give warning of an impending ﬁre in, for example, computer areas

and television sets, and of smouldering cables in electrical equipment generally.

The installation of gas sensors to warn of gas leaks and the attendant risk of

explosion is increasing in the developed world, especially in situations where

bottled gas is stored, for example in caravans, trailers and boats. In the UK it is

becoming commonplace to install them in homes along with smoke alarms.

By responding to fumes in cooking areas, car parks, laboratories and similar

places, gas sensors can be used to control ventilation fans. In industrial situations

the sensor can monitor concentrations of carbon monoxide, ammonia, solvent

vapours, hydrocarbon gases, ozone etc., and because of the growing conscious-

ness of environmental pollution and the safety aspects of industrial processes the

applications will multiply.

Gas sensors are being employed to characterise the ‘noses’ of perfumes, fruits,

wines and other foodstuﬀs and to measure the eﬀectiveness of deodorants. The

range of sensor types and applications is expanding at a remarkable rate with the

term ‘electronic noses’ introduced to describe the technology.

Semiconducting lambda-sensors based on titania have been developed and

used for car engine management. However for a variety of reasons, one being

sensitivity to surface contamination, the electrolytic sensor dominates.

A variety of commercial gas sensors is illustrated in Fig. 4.50.

4.6.3 Humidity sensors

Almost all ceramics that contain interconnected porosity exhibit a fall in resistivity

when exposed to humid atmospheres at normal ambient temperatures. Water is

adsorbed from low-humidity atmospheres by most oxide surfaces when it

214 CERAMIC CONDUCTORS

dissociates, with OH



attaching itself to a surface cation and the hydroxonium ion

attaching itself to a surface O

2

ion. Movement of the ions to alternative

sites under the inﬂuence of a ﬁeld gives an initial reduction in resistivity. With

increased humidity further layers of water are adsorbed and the high electrostatic

ﬁelds resulting from the surface states lead to a far greater degree of dissociation

than in liquid water and therefore to a further reduction in resistivity. At high

humidities liquid water may condense in the pores as a result of the curvature of

the liquid surface induced in the conﬁned space by surface tension.

The formation of liquid depends on pore diameter, surface tension and the

contact angle between the solid, liquid and gas phases. The last factor is diﬃcult to

estimate in general but is likely to be close to zero between water, humid air and a

surface consisting of adsorbed water. It can be deduced that, when the contact

angle is zero, the vapour pressure p in equilibrium with a concave liquid meniscus

relative to the vapour pressure p

in equilibrium with a ﬂat surface is given by





2gV

ð4:51Þ

where g is the surface tension, V

is the molar volume of the liquid, R

is the gas

constant, T is the temperature and r is the radius of curvature of the meniscus.

Since the humidity h is 100p=p

and the diameter d of a pore in which water will

condense is given by d ¼ 2r, the relation between pore diameter and humidity

leading to condensation is

d ¼

4gV

T lnð100=hÞ

ð4:52Þ

CERAMICS-BASED CHEMICAL SENSORS 215

Fig. 4.50 A selection of commercial gas sensors. The alumina tile carrying the sensor is

approximately 33mm

. An activated charcoal ﬁlter may be mounted inside the top of the

housing. (Courtesy of City Technology Ltd.)

If g ¼ 0:073 N m

1

, V

¼ 0:018  10

3

, T ¼ 293 K and R ¼ 8:315 J K

1

mol

1

d is approximately 3 nm when h ¼ 50% and approximately 42 nm when h ¼ 95%.

It is therefore evident that condensation will occur only at the narrowest points in

the microstructure, i.e. in the ‘neck’ regions.

The speed of response of a humidity element will depend on the rate of

diﬀusion of the ambient atmosphere into the pores of the sensing element. The

element must therefore be thin in section and may contain a system of larger

pores to reduce the length of the ﬁner pore pathways into which gas must

penetrate. Electrodes covering major surfaces must be porous.

The response of the surfaces of the element to moisture must remain constant,

and in practice this is one of the most diﬃcult conditions to meet. Over a period

of time the adsorbed molecules are thermally activated to the lowest-energy sites

and the response to a change in humidity may then be either sluggish or

permanently altered. Furthermore, molecules other than water may be adsorbed

increasingly with the passage of time, modifying the response to moisture. One of

the most eﬀective solutions to this problem is periodically to heat the sensor to a

temperature of  500 8C so that the surface is restored to a reproducible state.

One of the advantages of ceramic over organic elements is that they can be

repeatedly reconditioned in this way.

A sensor developed to detect the moisture content of the atmosphere in a

microwave oven and so indicate the onset of the cooking process serves to

illustrate the principles. The sensing element consists of a solid solution of TiO

in MgCr

with a minor phase of MgTi

. It is formed by mixing the

constituent oxides, pressing them into compacts 0.25 mm thick at the relatively

low pressure of about 70 MPa (about 10

p.s.i.) and ﬁring for several hours at

about 1350 8C to yield a sintered body with, typically, 35% porosity. The

electrodes are RuO

with the minimum amount of added glaze for adhesion so

that the layer is porous.

MgCr

is a normal spinel with Mg

2þ

ions on tetrahedral sites and Cr

3þ

ions

on octahedral sites. 30 mol.% TiO

can be incorporated without losing the spinel

structure. It is thought that Ti

4þ

and Mg

2þ

ions occupy neighbouring octahedral

sites, and the excess charge is compensated by the formation of Cr

2þ

ions which

occupy tetrahedral sites. It is a p-type semiconductor, although this does not

appear to be relevant to its behaviour as a humidity sensor, in contrast with its

behaviour as a gas sensor discussed above. Typical resistance–humidity

characteristics are shown in Fig. 4.51. It is believed that the high charge on

the Cr

3þ

ions exposed in the pore surfaces promotes the adsorption and

dissociation of water molecules according to

O Ð H

þ OH



ð4:53Þ

leading to the conduction mechanism outlined above. The resistance increases

irreversibly when the element is exposed to high humidities for prolonged

216 CERAMIC CONDUCTORS

periods. This may be due to reduced dissociation in the adsorbed layer as the

ions reach lower energy sites.

In practice the sensor is a plate 4-mm square mounted within a heater coil, and

its characteristic is periodically restored to the standard state by heating to

450 8C for a few seconds.

4.7 High Transition Temperature Superconductors

4.7.1 Overview

The following extract from a paper [31] published in 1968 makes interesting

reading in the light of subsequent developments.

Industrial interest in a physical phenomenon lies in a ‘quality’ or ‘outstanding

feature’ of that phenomenon. And in fact, the absence of electrical resistance

unlocks a wealth of new possibilities. Although the highest transition temperature

so far found is still as low as 20.05 K, the possibility of using underground cooled

superconducting cables for electricity supply in certain densely populated areas

such as London is already being seriously examined to see what advantages it

might oﬀer. There is also interest in the use of superconducting elements for logic

circuits (‘cryotrons’). Some very interesting possible uses include superconducting

magnet coils which can be used to give very high magnetic ﬁelds, sometimes with

ﬂux densities greater than 10 Wb/m

(100 000 gauss) over volumes large enough to

HIGH TRANSITION TEMPERATURE SUPERCONDUCTORS 217

Fig. 4.51 Resistance versus humidity for MgCr

þ 30 mol.% TiO

be of interest in technical applications. One wonders how many other even more

impressive changes in electrical technology are to be expected if superconducting

materials with a transition temperature of 50 or 100 K are ever produced . . .

Nevertheless, it is to be expected that, even if superconductors with transition

temperatures well above 20 K are never found, superconductivity will still ﬁnd its

applications in technology.

Since then there has been the totally unexpected breakthrough with the

superconducting ceramics operating comfortably at liquid nitrogen temperature

(77 K), and stimulating intensive R&D eﬀorts in the US, Japan and Europe.

Probably the most obvious application for superconductors is in the

transmission of electrical power and it seems that that this is close to being a

reality. There are many other potential applications in the power electrical

engineering ﬁeld, for example power generators, motors, transformers, leads and

fault current-limiters.

An important and established application is for electromagnets for nuclear

magnetic resonance (NMR) widely exploited as a research tool, and for its

medical counterpart, magnetic resonance imaging (MRI), a powerful medical

diagnostic tool. Strong magnetic ﬁelds are also essential for high-energy particle

accelerators for nuclear research.

Until superconducting magnets became available the generation of magnetic

ﬁelds of 10 MA m

1

(approximately 12 T in air) involved power dissipation of

order 1 MW in the windings. Not only did such a magnet require a large power

source (a large power station generator delivers about 600 MW) but most is

dissipated as heat which had to be removed by coolants. Superconducting

magnets exploiting Nb–Ti and other intermetallic compounds have been used for

many years, and it is now commonplace to produce ﬁelds of 12 T in quite

compact and easy to run electromagnets.

Exploiting the magnetic ﬁelds generated by superconducting coils, and also

those resulting from trapped magnetic induction within a superconducting bulk

material, is a very active R&D ﬁeld. Frictionless bearings and magnetic levitation

for transport systems are among the many objectives.

For ‘superconducting electronics’ interest is focussed on materials in thin ﬁlm

form, the major objectives being to achieve faster computers and improved

communications systems. Signal transit times are reduced by using super-

conducting interconnects, and the Josephson junction is the basis of a bistable

logic device. In communications technology superconducting ﬁlters and

antennae oﬀer signiﬁcantly improved selectivity and sensitivity.

4.7.2 The phenomenon of superconductivity

The following summary of the essentials of the phenomenon and theory of

superconductivity serves to introduce terminology and permits the ‘high

transition temperatures (T

) superconductors’ (HTSs) to be seen in a historical

218 CERAMIC CONDUCTORS

perspective. The texts by G. Vidali [32] and V.Z. Kresin and S.A. Wolf [33] are

recommended for supplementary reading.

The history of superconductivity has its origins in the experiments of H.

Kamerlingh Onnes who in 1908 at the University of Leiden succeeded in

liquefying helium and reaching temperatures down to 1 K. In 1911, while

pursuing the study of the electrical resistivity of metals at these very low

temperatures, he discovered superconductivity in mercury. As the temperature

was reduced the resistivity fell sharply at about 4 K and at the T

, fell to zero as

shown in Fig. 4.52. A current induced in a superconducting loop circuit has been

observed to persist without decay for over 2 years, setting an upper limit on the

resistivity of 10

28

O m; for practical purposes this is zero resistivity.

Since 1911 superconductivity has been observed in many metallic elements and

intermetallics, principal among them being tin (3.7 K), tantalum (4.5 K), lead

(7.2 K), niobium (9.3 K), Nb

Sn (18 K) and Nb

Ge (20.9 K).

There are two types of superconductor, Type I and Type II, characterized by

the way in which they respond to an applied magnetic ﬁeld.

If a Type I superconductor such as lead is placed in a small magnetic ﬁeld (e.g.

a few mT) and cooled, then at T

the magnetic ﬁeld is expelled from the interior

of the specimen. This is the Meissner eﬀect, which is fundamental to the

superconducting state; it is not simply characteristic of a material which happens

to be a (ﬁctitious) perfect conductor. The total absence of an electric ﬁeld in a

HIGH TRANSITION TEMPERATURE SUPERCONDUCTORS 219

Fig. 4.52 Resistance of mercury versus temperature. (After H. Kammerlingh Onnes (1911)

Akad. van Wetenschappen, Amsterdam, 14, 113, 818. Taken from Introduction to Solid State

Physics, C. Kittel, John Wiley, Chichester, 1986, p. 318.)

material having zero resistivity only demands that the magnetic induction (B)is

constant, not necessarily zero.

Because in the presence of an applied magnetic ﬁeld the interior of the

superconductor has zero magnetic induction it can formally be described as a

perfect diamagnet (see Section 9.1.5). The ‘diamagnetic eﬀect’ is of course not

due to a distribution of diamagnetism throughout the volume, but arises because

of ‘screening currents’ which ﬂow in the very near surface region of the

superconductor. The externally applied magnetic ﬁeld penetrates the super-

conductor to a small depth l, characteristic of the particular superconductor.

Below T

the applied magnetic ﬁeld can be increased up to a critical value B

when the superconductor changes to the normal state. This behaviour is

characteristic of a Type I superconductor.

In the case of a Type II superconductor below its T

, the applied magnetic

induction is totally excluded until a critical value B

is reached when some

ﬂux penetration occurs through ﬁlaments known as ﬂux vortices. The ﬁlaments

are of normal, that is non-superconducting material, and carry a magnetic

ﬂux. The ﬂux is quantized, the single quantum, a ﬂuxon, being of magnitude

h/2e ¼2.067810

15

. The ﬁlament itself is surrounded by screening currents

as a magnetic ﬁeld along the axis of a solenoid is surrounded by circular currents

(see Section 9.1.2). The material is therefore in a ‘mixed’, superconducting/

normal state. As the applied ﬂux density is increased so the penetration increases

until at an applied ﬂux density B

the material as a whole becomes normal.

The diﬀerent behaviours are illustrated in Fig. 4.53(a) and (b). Some of the

alloys, for example Nb

Sn (18 K), and all the high temperature superconductors

are Type II.

Because useful currents carried by a Type I superconductor generate magnetic

ﬁelds it follows that there are also critical current densities J

, corresponding

to the critical applied magnetic ﬁelds. In a Type II superconductor the

220 CERAMIC CONDUCTORS

Fig. 4.53 The induced magnetic moment (m

M) as a function of applied ﬁeld for (a) a Type I

superconductor and (b) a Type II superconductor.

situation is complicated by the extent of ﬂux ‘pinning’ discussed in Section 4.7.4

below.

These are the essential phenomenological characteristics of superconductivity.

The ﬁrst signiﬁcant step in a theoretical interpretation was taken in 1950 when H.

Fro

hlich and J. Bardeen showed that electron–phonon interactions are capable

of coupling two electrons together as if there is a direct attractive force between

them.

In 1956 L.N. Cooper published a theoretical analysis describing the addition

of a single electron pair to a normal metal at 0 K. This showed that if there is an

attraction between the two electrons, however weak, their combined energy is

always less than that of normal conduction electrons in their highest energy state,

the Fermi energy. In 1957 J. Bardeen, L.N. Cooper and J.R. Schrieﬀer extended

Cooper’s treatment to a large population of interacting electrons and showed

that they would form an assembly of so-called ‘Cooper pairs’. This has become

known as the ‘BCS theory’.

Cooper pairs move in the ‘classic’ superconductors in such a way that at

equilibrium their combined momenta is unchanged; whatever momentum change

one of the pair suﬀers due to interaction with a phonon, the change produced in

the partner is equal and opposite. Therefore phonon scattering has no net eﬀect

on the forward momentum of the electron population accelerating in an electric

ﬁeld. Because electron–phonon interactions are also responsible for electrical

resistance at normal temperatures, it is not surprising that relatively stable

Cooper pairs are found in poorly conducting metals with corresponding

relatively high values of T

. The energy required to split up a Cooper pair at

0 K is of the order of kT

(actually 1.8 kT

Fro

hlich also predicted the so-called ‘isotope eﬀect’ which associates a higher

with light isotopes of a metal than with heavier isotopes. This seems

reasonable since the lighter the atom the more easily it is moved and therefore

able to couple to the electron motions. The BCS theory explains the isotope eﬀect

in terms of phonon frequencies.

Although the BCS theory oﬀers an explanation for the behaviour of

superconductors known prior to 1986 it is generally not applicable to the high

temperature ceramic superconductors. In fact the BCS theory predicts a

maximum T

of approximately 30 K and this has now been comfortably passed.

It is surprising that, despite intensive eﬀorts over more than 15 years, there is still

no generally accepted theory to explain the mechanism of high-temperature

superconductivity. However, it is believed that whatever theory is ﬁnally

accepted it will involve paired electrons, although the pairing mechanism may

well diﬀer from that invoked in the BCS theory.

W.L. Ginzburg and L. Landau developed a phenomenological theory, the ‘GL

theory’, which makes no assumptions concerning the basic mechanism

responsible for superconductivity. In this respect the approach is similar to the

Devonshire phenomenological theory as applied to ferroelectrics (see Section

HIGH TRANSITION TEMPERATURE SUPERCONDUCTORS 221