EP0435963A4

EP0435963A4 - Pyrometric thermoelectric sensor

Info

Publication number: EP0435963A4
Application number: EP19900903305
Authority: EP
Inventors: Noel Arthur Burley
Original assignee: NICROBELL Pty Ltd
Current assignee: NICROBELL Pty Ltd
Priority date: 1989-02-17
Filing date: 1990-02-19
Publication date: 1991-08-28
Also published as: EP0435963A1; WO1990009682A1; GB2246019B; AU628936B2; GB2246019A; AU5106190A; GB9115265D0

Description

PYROMETRIC THERMOELECTRIC SENSOR This invention relates to thermocouples, thermocouple structures, thermocouple sheathing, and to protection devices containing a thermocouple. In one embodiment this invention relates to a novel improved design and structure of a thermocouple intended for application at high temperatures above, say, 1000°C. In another embodiment this invention relates to improved metallic alloy materials for use in the manufacture of thermocouple structures. In a further embodiment this invention relates to novel improved thermocouple structures intended specifically for use in the measurement of the temperatures of hot molten glass in the manufacture of a wide range of glass materials, products and components. The thermocouples of the invention are of the mineral-insulated metal-sheathed (MIMS) structure.

The measurement of the temperature of hot molten glass presents singular technical and economic difficulties due to a range of diverse factors. These factors include the very high industrial temperatures involved, the high viscosity and abrasiveness of molten glass at these temperatures, the chemical reactivity of both the glass itself and also of the combustion atmosphere in which it is heated, and the high cost of the rare-metal materials of construction of the conventional thermocouple sensor structures presently employed.

No satisfactory solution to this temperature measurement problem, which at the same time is both technically feasible and economically acceptable, has hitherto been found.

The term "glass" can be used to describe many substances which possess the physical characteristics of a liquid but the rigidity of a solid. Glass, like most liquids, has a random molecular structure particularly when hot. When most liquids solidify or "freeze" these molecules normally are regimented into precise crystallographic arrays. In the manufacture of glass, during subsequent cooling, this freezing does not take place; the viscosity simply rises with falling temperature to a stage where the molecules of the super-cooled liquid cannot move to form a regular crystal structure. Even when still very hot, say at 1200°C, the viscosity of the flowing molten mass remains relatively quite high, and it can exert a significant force on any object located in its path. This includes elongated cylindrical objects such as sheaths housing thermocouples. Thus a metallic thermocouple sheath is normally essential to resist the high bending moment forces exerted by the hot flowing glass.

In all sectors of the glass manufacturing industry, temperature measurement and control is of the utmost importance. With temperature change, glass viscosity and hence flow rate can vary significantly. When such temperature changes are neither intended nor apprehended, subsequent control of glass forming and processing operations is impaired. Gob size and homogeneity, for example, are of prime importance particularly where crystal glass is involved but also in the case of common bottle glass.

The use of standard alumina-sheathed rare-metal thermocouples of the platinum-rhodium versus platinum (Pt-Rh/Pt) variety, designated type R (i3wt-%Rh) or type S (10wt-%Rh) by the Instrument Society of America, for measuring furnace crown temperatures has for some time been common practice. However such types R and S thermocouple assemblies show inherent shortcomings because of several factors, such as

(i) They show inherent lack of strength, due to the relative brittleness of the alumina sheath, particularly where transverse forces are involved. (ii) The hot furnace atmosphere above the molten glass contains exhaust gases from the combustion of the fuel and vapour phases from the glass, both of which can chemically attack the alumina sheath.

In both cases above, penetration of the sheath by glass and/or gas, either when fractured as in (i) or corroded as in (ii), can cause contamination of and consequent intolerable loss of measurement accuracy in the Pt-Rh Pt thermocouple.

(iii) The relatively high cost of rare-metal thermocouples compared to base-metal thermocouples. The above problems are being increasingly exacerbated by the burgeoning number of special refractory materials being used in glass furnace linings and by some of the more exotic glass additions, both of which can be chemically incompatible with alumina and platinum and its alloys.

The more recent use of platinum or a zirconia grain-stabilised platinum-rhodium alloy as a sheathing material for alumina-sheathed Pt-Rh/Pt thermocouples, in the form of an elongated thimble, has proved to be a fairly successful move in overcoming some of the technical problems described above. However, such thimbles to be effective must be of the order of 0.5 to 0.8 mm thick and consequently very expensive. A typical such thermocouple is illustrated in

Figure 1, in which the following features are identified:

1.1. Thermocouple headcover

1.2. Connector head

1.3. Conductor terminals 1.4. "Fish-spine" insulators

1.5. Inconel (INCO Group trade mark) heat-shield tube

1.6. Pure recrystallized alumina thermocouple sheath Thermocouple conductor wires

Platinum sheath ("thimble") Heat-resistant rope Heat-resistant washers Six-bore recrystallized alumina thermocouple insulator tube

Adaptor Cable gland

Thermocouple extension wires Top measuring thermojunction (of "tri-level" group of three) M. Middle ther ojunction B. Bottom thermojunction

The number of failures of alumina-sheathed rare-metal thermocouples, due to mechanical stress and/or chemical contamination, is reportedly reduced by the use of such thimbles. This has been the case particularly where the temperature measurement and control arrangements involve the use of thermocouples actually immersed in the hot molten glass in the furnace, forehearth and feeder channels. Such sensor assemblies may contain multiple thermocouples having measuring-thermojunctions so arranged as to enable temperatures to be measured at various depths in the glass (see Figure 1) . The high temperatures and aggressive glass conditions make it necessary for alumina-sheathed Pt-Rh/Pt thermocouples to be protected by Pt or Pt-alloy thimbles below the glass line.

The glass thermocouple assemblies described above, whilst being the best technical solution so far developed to the problem of accurately knowing hot glass temperatures, yet have a number of residual shortcomings which remain problematic in glass manufacturing technology. These problems arise because of factors such as - (a) The Pt-Rh/Pt thermocouples and, particularly, the Pt-alloy thimbles are becoming increasingly and prohibitively more expensive.

(b) Even though alloying platinum by rhodium addition makes for a stronger thimble, rhodium likewise is most expensive and tends to unacceptably colour the glass under certain conditions.

(c) It is essential to avoid this colouring problem, (b) above, by omitting rhodium from the thimble, where high quality optical or crystal glasses are concerned. Unfortunately this introduces a further problem that molten glass "wets" the surface of pure platinum. This wetting phenomenon causes glass to adhere to the metal, thus further decreasing the already poor flow characteristics of the hot glass past the thermocouple sensor probe; it also reduces the thermal response sensitivity of the sensor.

(d) Avoiding this wetting problem, (c) above, which can be done by alloying platinum by gold addition, only produces a thimble of lower melting point and lesser strength.

(e) All the above platinum group metals and alloys exhibit softening (and thus reduced erosion resistance by thimbles to the flowing glass) with the effluxion of time at high temperature. This causes consequent reduction in strength due to recrystallization and grain growth. In addition, in the case of both thermocouple conductor wires and thimbles, unacceptable deformation occurs by high-temperature creep. The resultant limitation of life of the expensive thimbles can be reduced by the use of zirconia grain-stabilized Pt-base alloys, but only with the significant penalty of a further increase in an already prohibitive cost.

(f) To minimise the high cost of glass thermocouples incorporating Pt-alloy thimbles and thermocouples, it is common practice to attach a shortened platinum thimble to a co-linear extension tube fabricated from a conventional base-metal alloy like Inconel (see Figure 1) . This extension tube passes up through the combustion gas space above the glass line and through the wall or ceiling of the furnace to the cooler ambient environment outside. Conventional base-metal alloys (like Inconel), characterized by an optimized combination of lower cost and oxidation resistance, often are inadequate. The extension tube may even fail prematurely by high-temperature corrosion, particularly near the junction with the rare-metal thimble. Thus the life of the whole thermocouple assembly can be terminated prematurely.

For all these reasons, in particular the prohibitive cost of the conventional glass thermocouple described above, it is essential that novel thermocouple concepts, designs and structures be introduced. The new concepts, in basic form, feature relatively inexpensive base-metal alloys both for the thermocouple conductors and for the protecting sheath tube. The sheath alloys chosen show an adequate and optimum combination of strength, resistance to corrosion and oxidation, and longevity at the temperatures involved and under the environmental conditions prevailing within the glass furnace during the manufacturing processes involved. Furthermore the base-metal thermocouple incorporated in the design and structure of the novel thermocouple sensor shows ultra-high thermoelectric stability such as is exhibited by rare-metal Pt-alloy thermocouples over the range of temperatures involved.

OBJECTS AND SUMMARY OF INVENTION The novel glass-temperature thermocouple of this invention is of the integrally metal-sheathed mineral-insulated (MIMS) format and structure. The MIMS format of this invention, which features improved materials and structure, enables optimal achievement of the necessary performance characteristics.

As is well known in the art, the manufacture of MIMS cable or of individual MIMS thermocouple sensor structures begins with matched thermocouple wires surrounded by non-compacted mineral oxide powder held within a metal tube. By rolling, drawing, swageing, or other mechanical reduction processes, the tube may be reduced in diameter by the required amount and the insulation is compacted around the wires. The conventional product of MIMS structure is illustrated diagrammatically in Figure 2. In Figure 2, which shows a sectional view of the conventional materials of construction, 2.1. is the integral sheath, usually of stainless steel or Inconel; 2.2. is the mineral insulation, usually a mixture of mineral oxides essentially of about 96 wt.-% MgO and 4.wt.-% Si0₂; and 2.3. are the thermoelement conductor wires usually of the ISA (Instrument Society of America) type K variety. Unfortunately, the conventional design concept of the MIMS thermocouple is not suited to the measurement of hot molten glass temperatures in the manufacture of glass products. This is because -

(i) The conventional sheath materials, particularly the stainless steels, will not withstand exposure in molten glass and in the associated combustion atmosphere for extended periods of time at the temperatures involved (up to about 1220°C).

(ii) The conventional thermocouple conductor wires - ISA type K - will likewise not withstand the highest temperatures and longest times encountered in the glass industry.

In the cases of both (i) and (ii) above excessive high-temperature corrosion, mainly oxidation causing premature failure, is the reason for the unsuitability of the alloys.

(iii) The thermoelement conductor wires can be contaminated by chemical elements which thermally diffuse through the compacted insulant material from dissimilar sheath alloys. The resultant changes in the chemical compositions of the conductor alloys can cause substantial changes in their thermoelectromotive forces. Such changes in thermal emf are analogous with and algebraically additive to those caused by the high-temperature oxidation of these alloys. (iv) The thermoelement conductor wires, particularly the negative wire, may fail mechanically because of substantial alternating strains imposed during thermal cycling. These strains are caused primarily by longitudinal stresses which arise because of substantially different temperature coefficients of linear expansion of the thermoelements and of dissimilar sheath materials.

It is clear from the foregoing that the principal problem in the measurement of high temperatures using a thermocouple of conventional MIMS construction is thermoelectric instability, hence measurement uncertainty.

It is equally clear that this thermal emf instability results primarily from the use of dissimilar and unsuitable alloys for both sheath and thermoelement conductors. This problem has arisen because sheath and thermoelement materials have hitherto been chosen independently of each other to match, respectively, the environment of exposure and existing pyrometric instrumentation.

In contrast to the prior art, the MIMS thermocouple of the present invention has been designed as a truly integral system. The choice of materials for its principal components, sheath, thermocouple, insulant, and filling gas - has been made only after a proper consideration of the inter-related properties of all of them. A discussion of how the problems which plague MIMS thermocouples of conventional design can best be overcome has been given by one of the present inventors in Australian Patent Specifications 80105/87 of 23rd October, 1987, and 12149/88 of 19th February, 1988. The above Specification claims a novel MIMS cable incorporating - (a) novel thermoelement conductor alloys known as NIOBELL-P (positive) and NIOBELL-N (NIOBELL is a trade mark of BELL-IRH Limited) or, alternatively, thermoelement conductors which are ISA type N alloys. (b) a thermal passivation process designed to enhance the already demonstrably ultra-high thermoelectric stabilities of the NIOBELL and type N alloys, and

(c) a novel series of sheath alloys known as NICROBELL (trade mark of BELL-IRH Limited) of improved thermomechanical and thermochemical properties over conventional MIMS sheath alloys.

The NICROBELL-sheathed MgO-insulated NIOBELL P/N (or type N) MIMS thermocouple cable is the claimed subject of the above-mentioned Patent Specifications 80105/87 and 12149/88. These specifications set out the conceptual nature and inventive rationale of the above NICROBELL/NIOBELL and NICROBELL/type N MIMS systems; it does not, however, make any reference to the specific thermocouple structures nor the thermocouple sheathing and protection devices for molten glass temperatures which are claimed in the present specification.

The MIMS thermocouple system and specific sensor structures of the present specification are very well suited to glass-temperature thermocouple sensors. It is this present specification which claims specific concepts, designs and structures of novel glass-temperature thermocouples. It is conceivable that, under the most extreme conditions of high temperatures and corrosive atmospheres that can be encountered in a glass furnace, the NIOBELL and type N thermocouple alloys and the protective sheath alloys NICROBELL may not provide, per se, the very high environmental stability and longevity that would be demanded for the longest of the glass furnace compaigns that can eventuate.

In this case we revert to the use of Standard ISA types R or S rare-metal thermocouples, and/or even further improved protective sheath alloys showing the exceptionally high resistance to high-temperature corrosion that would be demanded. In the case of the thermocouple, a further advance in the art is the use of the newer ISA type B system (Pt-30wt.-%Rh/Pt-6wt.-%Rh) which has improved emf-stability and strength over types R and S. In the case of the sheath, such improved alloys also show the relevant values of thermomechanical properties required to properly facilitate fabrication and high-temperature utilization.

The optimum combination of thermomechanical properties (e.g. room-temperature ductility, high-temperature strength) and thermochemical properties (e.g. high-temperature oxidation resistance) can be obtained in such alloys, for instance, if they have an M-Cr-Al base composition (where M signifies Ni, Fe, or Co, singly or in certain concentration combinations). This is in contrast to the Ni-Cr-Si base of NICROBELL. The M-Cr-Al base alloys in question are those compositions which would form a protective scale consisting essentially of the aluminium oxide alpha alumina, α-Al₂0,. This is in contrast to the protective scale which forms on NICROBELL, namely an oxide system comprising a film of the chromium oxide, chromia, Cr-O-*. overlaying thinner discrete films of the silicon oxide, alpha silica, α-Si0₂ and magnesium silicate, MgSiO,, which form at the metal-scale interface. For temperatures up to about 1200°C, the Cr-C /SiO^/MgSiO, protective system is most efficaceous, but at higher temperatures the volatility of Cr₂0₃, as vapour CrO, is a distinct problem. Thus for longer times at temperatures higher than about 1200°C, it is preferable that oxide scales involving A1₂0₃ be relied upon for protection of the alloy. The reasons for this preference include ⁽i^{) A}^2°₃ ^{nas re}l^ati^vely i°^w dissociation-pressures and vapour-pressures in the temperature range of interest up to 1300°C. Thus the protective alumina film will not be lost by decomposition and/or volatilization. ⁽ii^{) A1}2°3 "*"^{s t}^^ermody^nami^cal*^Ly more stable than Cr₂0₃, and thus it forms preferentially to, and persists longer than, Cr₂0₃.

(iii) Counter-diffusion of metal and oxygen ions, which react to form oxides, is much slower in A1₂0^ than in Cr₂0₃« Because of this, continuous layers of

A1₂0₃, formed by the selective oxidation of aluminium in the alloy, can ensure better oxidation resistance than corresponding layers of Cr₂0₃.

In M-Cr-Al alloys, it is possible to form continuous surface layers of A1₂0₃ with aluminium concentration of less than about 10wt.-%.

The conditions under which continuous layers of

A1₂0₃ form on the surface of the M-Cr-Al alloys depend primarily on the composition of the alloy and the temperature and duration of the oxidation process. It is possible to plot ternary isothermal graphs which show the composition of the oxide layers which form as functions of component concentrations. Figure 3 is, for example, a typical such plot for 1000°C in which: In region (1) aluminium is selectively oxidized to form an 1₂0₃ layer on the surface of the alloy.

In region (2) the oxide layer is composed of

Cr₂0₃ and, in addition, aluminium is oxidized internally within the alloys, and In region (3) the scale layer comprises NiO plus spinels based on Ni-Cr (NiCr₂0₄) and Ni-Al (NiAl₂0₄); in addition the alloy undergoes internal oxidation. The compositional regime of greatest interest for the preferred improved alloys for protective sheaths for a thermocouple for molten glass are chosen from region (1) .

Similar isothermal ternary plots for other M-Cr-Al alloy systems (where M signifies Fe or Co) can be prepared. As a general rule, the extent of the formation of protective layers of A1₂0₃ in the different alloys is in the order - Fe-Cr-Al> Ni-Cr-Al > Co-Cr-Al. For particular compositions, aluminium becomes more selectively oxidized with increasing temperatues.

Preferred M-Cr-Al alloys of this invention may also contain mechanically added oxide dispersions such as the thorium oxide, thoria, Th0₂ and the yttrium oxide, yttria, Y₂°3 ^{wnich are} known to have extremely beneficial effects in enhancing oxidation properties and strength. The fundamental mechanisms of the improvement in oxidation resistance are complex and not particularly well understood. However it is believed that the enhancement processes include promotion of selective external oxidation and improvement in scale adherence.

The principal role of chromium in the preferred M-Cr-Al alloys is to increase the strength of the iron, nickel, or cobalt bases. While there are a number of candidate solute elements which cari perform this role, chromium has the essential advantage over virtually all the others that it also greatly enhances the oxidation resistance over a wide range of solid-solution alloys.

For the purposes of the present patent specification, the preferred alloys of this invention can be given the following generic nomenclature MECRALY general

FECRALY if Fe-based

NICRALY if Ni-based

COCRALY if Co-based

The nominal chemical composition tolerances

(percentages by weight) for the preferred FECRALY alloy of this invention are -

Element Composition

Cr 20 + 1.0 Al 4.5 + 0.5

Y₂0₅ or Th0₂ 0.5 + 0.1

Fe balance

Alloys similar in composition to the preferred alloy are available commercially, e.g. INCO alloy MA 956. The said preferred alloy is designed to have an

Fe-Cr-Al base which is highly resistant to high-temperature corrosion by such processes as oxidation. The presence of the oxide dispersoid, Y C or Th0₂, which remains stable up to the melting point of the alloy, together with the presence of an optional 0.5 + 0.1 wt.-% concentration of titanium, produces exceptionally enhanced strength, hardness and consequently corrosion and erosion resistance, including at high temperatures. Experimental data show that the cyclic oxidation resistance of such a FECRALY alloy is greatly superior to that of the alloy Inconel which is used for the conventional glass thermocouple extension tube. The following table contrasts the total weight gain of each alloy, due to oxidation scaling when samples are heated in air in cycles of 1200 s at 1200°C followed by cooling in rbom-temperature air for 600 s.

The desired effects of improving oxidation resistance and strength are in fact efficacious, to a greater or lesser degree, over the fairly wide range of the respective solid solubilities of chromium and aluminium in bases of iron, nickel, or cobalt. Thus the concentration ranges of these elements in solid solution in Fe, Ni or Co can be broadened while still facilitating the aims and objectives of this invention. The preferred embodiments of a range of optional alloys is set down in Table 1.

Table 1

The integral compacted MIMS thermocouple sensor of this example is fabricated using existing manufacturing procedures. They begin with thermoelectrically matched thermoelement wires fabricated in the form of a tri-level thermocouple (of the NIOBELL or ISA type N variety) surrounded by non-compacted ceramic oxide insulating powder held within a metallic alloy tube of NICROBELL, FECRALY, NICRALY, or COCRALY of the appropriate composition. By rolling, swageing or other suitable mechanical reduction processes the alloy tube may be reduced in diameter until the insulation powder is compacted around the thermocouple wires. The manufacturing process parameters are adjusted so that the ratios of sheath diameter to wire-size and to sheath-wall thickness offer an optimal balance between minimum wall-thickness for adequate life and strength and suitable insulation spacing for effective insulation resistance at elevated temperatures. A most important feature of the fabrication process is that considerable attention is given to the initial cleanliness and chemical purity of the components and retention of a high degree of cleanliness and dryness particularly ^"of the insulant throughout fabrication.

A feature of the embodiment which is this example is a suitably attached optional collar made of the same alloy as the sheath, or of another suitable alloy, and located so as to pass through the molten glass/combustion gas interface, where the environment is particularly aggressive.

A diagrammatic conceptual illustration of this embodiment is given in Figure 4. In Figure 4 the following features are identified:

4.1. Thermocouple headcover, connector head, conductor terminals, etc. as individually identified in Figure 1. 4.2. Tri-level thermocouple assembly of the

NIOBELL or type N variety.

4.3. Sensor sheath of NICROBELL or MECRALY alloy.

4.4. Compacted ceramic oxide insulant powder. 4.5. Molten glass/gas interface. 4.6. Optional collar of NICROBELL, MECRALY, or other suitable alloy. T. Top measuring-thermojunction (of "tri-level" group of three). M. Middle thermojunction.

B. Bottom thermojunction.

Example 2

In this example, the mode of manufacture and design of the thermocouple sensor is the same as in Example 1.

In this case, however, which is germane to the most extreme conditions of high temperatures and corrosive atmospheres that can be encountered in a glass furnace, the NICROBELL or MECRALY alloy sheaths may not provide the very high environmental stability and longevity that would be demanded for the longest of the glass furnace manufacturing compaigns that can eventuate. This problem is overcome by applying a suitable relatively thin but strongly adherent coating of a more refractory metal, alloy, or chemical compound which will satisfactorily withstand the extreme conditions described above. Suitable coatings include a optimal thickness of a rare metal, a rare metal alloy, a refractory metal oxide such as alumina, or a metal aluminide, etc. Such coatings can be applied singly or in combination, e.g. a combination such as a rare metal or rare-metal alloy applied as an overcoat to a refractory metal oxide previously applied directly to the sensor sheath alloy described at line 32 on Page 17. Such single or combined coatings can also be applied to other components used in the molten glass in association with the temperature sensor, e.g. mechanical stirrers and other structural devices. Such a combined coating is illustrated in Figure 5. In Figure 5 the following features are identified - 5.1 Tri-level thermocouple, as in Fig. 6.

5.2 Sensor sheath of NICROBELL or FECRALY alloy

5.3 Inner layer of a refractory metal oxide or compound

5.4 Deposit of platinum alloy

The thickness of such metal coatings is much less, of the order of one tenth, than that required for the corresponding thimble of the prior art. It will be clearly understood that this aspect of the invention is not limited by the specific examples given of protective coatings that would be applicable. It is proposed that certain other types of coatings would, in addition, also be suitable. In this example, the methods of coating deposition would include electrodeposition from aqueous solutions or fused salts, thermal plasma or other thermal spraying, physical or chemical vapour deposition, etc. It will be clearly understood that this aspect of the invention is not limited by the specific examples given of suitable deposition processes.

Example 3

In this example the general mode of manufacture and design of the thermocouple sensor is the same as in Example 1.

In this case, however, which is germane to the most extreme conditions of high temperatures and extended production compaigns that can be encountered in a glass furnace, it is necessary to revert to the use of standard ISA types B, R or S thermocouples. Here it is not feasible to compact the mineral insulant by a tube reduction process, but rather a process such as tamping is employed to avoid damage to the rare-metal thermocouple assembly. It is well known in the art that if a rare-metal thermocouple needs to be supported in a base-metal sheath tube, then it must be separated from the base metal by an impervious ceramic sheath of very high purity. This is to prevent contamination of the rare metals by ions emanating from the base metal either by contact or by vapour-phase transfer at high temperatures. Such contamination, which takes place by adsorption and dissolution, can lead to substantial changes in composition, hence changes in thermoelectromotive force and measurement accuracy of the rare-metal thermocouple.

It is a specific embodiment of this invention that, when rare-metal thermocouples are used in this example, the impervious ceramic sheath can be eliminated (thus increasing the thermal response sensitivity of the sensor). Two factors make this possible:

(i) The alumina film which forms on the inside of the preferred FECRALY sheath at high temperatures is, as described above, highly stable having in particular very low vapour- and dissociation-pressures and very high degrees of stoichiometry. Therefore base-metal ions are unable to diffuse freely through this protective oxide film, and aluminium ions can evaporate from the surface of the film only at a minute rate.

Contamination of the rare-metal wires inside the FECRALY sheath is therefore minimal.

This has been tested in laboratory experiments in which a bare type R rare-metal thermocouple was exposed for several thousand hours at 1100°C in the air inside an open-ended FECRALY tube. The expected negative drift in the thermal-emf output of the rare-metal thermocouple, due to base-ion contamination, was negligible not exceeding the equivalent of 1°C. (ii) As a precautionary measure, a particular design of multi-bore insulating ceramic rod is utilized in this example. This design, which is illustrated in Figure 6, features rare-metal thermocouple measuring-junctions which are embedded in aluminous refractory cement. This arrangement ensures that, for any low concentration of base-metal ions that exists, no contact between them and the rare-metal thermocouples can occur. This structure, in the form of the assembly shown in Figure 6, has been the subject of exhaustive laboratory testing by exposure inside a FECRALY tube for over 3000 h at 1100°C, and this has proven its effectiveness. In Figure 6, the following features are identified -

6.1. Three rare-metal thermocouples of the

ISA type B, R or S variety.

6.2. Six-hole insulating "rod" carrying the rare-metal thermocouples.

6.3. Measuring-thermojunction cavity plugged with alumina cement.

T. Top measuring-thermojunction (of

"tri-level" group of three). M. Middle thermojunction. B. Bottom thermojunction.

Example 4

In this example the general mode of manufacture and design of the thermocouple sensor is the same as in Example 1, with the exception of the absence of the optional collar. In this case, a second thin outer or condominium sheath is swaged onto the main sheath tube, or otherwise attached to it, to form a double or lamella sheath structure. This structure is illustrated in Figure 7. In Figure 7, which shows only the sheath structure, the following features are identified -

7.1. Inner main sheath tube of NICROBELL, or other suitable alloy.

7.2. Thin outer or condominium sheath of MECRALY, or other suitable alloy.

In this example the preferred inner sheath of NICROBELL preserves the highly desirable maximum thermal compatibility between sheath alloy (NICROBELL) and thermocouple conductor wires (NIOBELL or type N) separated only by mineral oxide insulant.

As fully described in the abovementioned Australian Patent Specification 80105/87 and 12149/88, this arrangement makes for maximum thermoelectric stability and temperature measurement accuracy. It will be clearly understood that, in this example, the use of ISA type B, R, or S rare-metal thermocouples is not precluded if the extreme operating conditions described above preclude the use of base-metal thermocouples. In an overall sense, it will be clearly understood that the invention in its general aspects is not limited to the specific details referred to hereinbefore.

Claims

THE CLAIMS DEFINING THE INVENTION ARE AS FOLLOWS:

1. A MIMS thermocouple suitable for use at high temperatures in a hostile environment such as molten glass, comprising thermocouple wires surrounded by mineral insulation held within a sensor sheath, characterised in that the sensor sheath is composed of an alloy chosen from the group consisting of NICROBELL, FECRALY, NICRALY an COCRALY as hereinbefore defined.

2. A thermocouple according to Claim 1 in which the sheath is composed of an alloy of the following σompostion:

Element (per cent by weight)

10 to 35

0.05 to 15

0.05 to 2

0.05 to 2 balance

balance

3. A thermocouple according to Claim 1 in which the sheath is composed of an alloy of the following composition:

Element (per cent by weight)

Cr 20 + 1.0

Al 4.5 + 0.5

Y₂0₅ or Th0₂ 0.5 + 0.1

Fe balance

4. A thermocouple according to Claim 1 in which the sheath is coated with a refractory metal oxide or compound which is overcoated with a deposit of plantinum or plantinum alloy. 5. A thermocouple according to Claim 1 suitable for use under extreme conditions in which the thermocouple wires are composed of rare metals as in ISA types B, R, or S thermocouples and the sheath is composed of an alloy of the following composition:

Element (per cent by weight)

Cr 20 + 1.0

Al 4.5 + 0.5

Y₂0₅ or Th0₂ 0.5 + 0.1

Fe balance

6. Novel alloys useful in the manufacture of sensor sheaths for MIMS thermocouples as defined in Claim 1, said alloys consisting essentially of:

Element (per cent by weight)

Cr 10 to 35

^' Al 0.05 to 15

Ti 0.05 to 2

Y₂0₅ 0.05 to 2

ME balance in which ME is chosen from the group consisting of iron, nickel, and cobalt.

7. A novel alloy useful in the manufacture of sensor sheaths for MIMS thermocouples as defined in Claim 1, said alloy consisting essentially of:

Element (per cent by weight) Cr 20 + 1.0

Al 4.5 + 0.5

Y₂0₅ or Th0₂ 0.5 + 0.1

Fe balance