[0003] Description of the Related Art [0004] Improvements are continually sought in the capa- bilities of various types of sensing and planar heating systems, such as systems for sensing temperature, fluid flow rates and levels, pressure and gaseous environments, self-sensing planar heaters and fast, uniform heaters.

The characteristics sought to be enhanced include faster response time, greater sensitivity, higher temperature capability and low drift.

[0005] Prior art directly related to the present inven- tion includes: [0006] R. Holanda, "Thin-Film Thermocouples on Ceramics", NASA Technical Briefs, March 1997, p. 62: Pt vs PtRh metal thin films are deposited on A1N dies for use as thin-film thermocouples (TCs). The drift of the TC junc- tion vs temperature (to 1500°C) is discussed. <BR> <BR> <P>[0007] Y. H. Chiao et al. ,"Interfacial Bonding in Brazed and Cofired Aluminum Nitride", ISHM'91 Proceedings, pp.

460-468: The reactions for joining interfaces between A1N and several metals, including W, is discussed and com- pared with the joining method (braising or cofiring). A multilayer A1N/W structure is disclosed, in which the in- terface joining is due to interlocking grain-boundaries.

Although not disclosed in the article, such a structure has been used as a heater, but without any mechanism for sensing the actual temperature.

[0008] Patent No. 6,084, 221: Silver and silver alloys on A1N are discussed for planar heater applications.

[0009] Patent No. 6,103, 146: Thick film, screen-printable circuits, comprised of conductive paste compositions which facilitate the application of Au, At, Pt, Pd and Rh mixtures and alloys, are applied directly to A1N sub- strate surfaces.

[0010] Patent No. 6,242, 719: Thick films are described as being deposited on A1N by chemical vapor desposition.

[0011] Patent No. 6,239, 432, issued May 29,2001 in the name of the present inventor: An IR absorbing body of SiC is elec- trically and mechanically connected to an A1N substrate by an electrically conductive mounting layer that includes W, WC or WZ C.

SUMMARY OF THE INVENTION [0012] The present invention seeks to provide a new sen- sor system and method that is capable of achieving a faster response time, greater sensitivity, higher tem- perature capability and lower drift than previous sensor systems.

[0013] In a preferred embodiment, a thin film layer of tungsten is provided on an A1N substrate, with a signal source applying an electrical actuating signal to the tungsten layer, and a sensor sensing the response of the tungsten layer to the actuating signal. Various oxida- tion-resistant protective layers can be provided over the tungsten layer, including gold, B203-SiO2, Au-Pt alloys

(with an optional tungsten or B203-SiO2 layer over the al- loy), or Pt (with an optional B203-SiO2 layer over the Pt). An A1N cap can also be provided over the protective layer.

[0014] The tungsten layer in a preferred embodiment com- prises a plurality of conductive strands distributed on a planar A1N substrate. For substrate shapes such as rec- tangular, the strands are preferably serpentine shaped and parallel. For a circular substrate, the strands preferably extend along respective lines of longitude that merge at opposite poles of the substrate.

[0015] While tungsten on an A1N substrate is preferred, the invention can be generalized to the use of an AlN substrate and a conductive layer on the substrate which, over a predetermined temperature operating range, has an expansion coefficient within 1.00+/-0. 07 of the sub- strate, is substantially non-reactive with the substrate, and exhibits substantially no solid-solubility or inter- diffusivity with the substrate. It can also be general- ized to the use of an insulative substrate, with a tung- sten conductive layer on the substrate which, over a pre- determined temperature operating range, has an expansion coefficient within 1.00+/-0. 07 of the substrate, is sub- stantially non-reactive with the substrate, and exhibits substantially no solid-solubility or interdiffusivity with the substrate.

[0016] Applications for the described material system in- clude planar heaters capable of self-sensing their own temperature; fluid flow rate sensors using only a single W/AlN element, or a pair of such elements spaced apart in the fluid flow path with one heated, the other not heated, and both sensing the fluid temperature at their respective locations ; fluid level sensors capable of sensing whether or not they are immersed in a predeter- mined fluid; pressure sensors in which the voltage/cur-

rent relationship of the sensor is related to the sur- rounding gas pressure; and chemical sensors for environ- ments in which the tungsten layer is subject to an al- teration from the environment which changes its response characteristics.

[0017] These and other features and advantages of the in- vention will be apparent to those skilled in the art from the following detailed description, taken together with the accompanying drawings.

BRIEF DESCRIPTION OF THE DRAWINGS [0018] FIG. 1 is a perspective view of a sensor/heater in accordance with one embodiment of the invention, with a thin film tungsten layer on a A1N substrate; [0019] FIG. 2 is a cut-away perspective view of another embodiment, with a AlN substrate and cap bonded together by a WC layer; [0020] FIG. 3 is a perspective view of another embodi- ment, with a protective layer on the tungsten layer of FIG. 1; [0021] FIG. 4 is a cut-away perspective view of another embodiment, with a protective encapsulation over the structure of FIG. 1; [0022] FIG. 5 is a cut-away perspective view of another embodiment, with an A1N cap over the structure of FIG. 4; [0023] FIG. 6 is a perspective view of a precursor to an- other embodiment, in which an Au-Pt alloy provides a pro- tective cap to the structure shown in FIG. 1; [0024] FIG. 7 is a cut-away perspective view of another embodiment, with a pair of structures as illustrated in FIG. 6 assembled together; [0025] FIG. 8 is a cut-away perspective view of a struc- ture formed by thermally reacting the structure of FIG.

12;

[0026] FIG. 9 is a cut-away perspective view of another embodiment, in which a protective encapsulation is formed over the structure of FIG. 6, after that structure has been thermally reacted; [0027] FIG. 10 is a cut-away perspective view of the structure shown in FIG. 9 with an A1N cap; [0028] FIGs. 11 and 12 are plan views illustrating a rec- tangular planar heater in accordance with an embodiment of the invention, with two different electrode configura- tions; [0029] FIG. 13 is a plan view illustrating a circular planar heater embodiment; [0030] FIGs. 14 and 15 are simplified schematic diagrams illustrating two heater/temperature sensor embodiments of the invention, with voltage and current drives, respec- tively; [0031] FIGs. 16,17, 18,19 and 20 are simplified sche- matic diagrams illustrating the application of the inven- tion to a single element flow rate sensor, dual element flow rate sensor, fluid level sensor, pressure sensor and environmental sensor, respectively; and [0032] FIGs. 21,22 and 23 are tables summarizing the temperature range of various embodiments of the invention for different operating environments, characteristics of temperature sensors using various embodiments of the in- vention compared to prior art temperature sensors, and characteristics of heaters using various embodiments of the invention compared to prior art heaters, respec- tively.

DETAILED DESCRIPTION OF THE INVENTION [0033] The present invention provides a novel system and method for sensing temperature, fluid flow rates, pres- sure and chemical environmental conditions, as well as functioning as a heater capable of sensing its own tem-

perature, with a higher temperature capability, greater sensitivity, faster response time and/or lower drift than previous sensors. In the preferred embodiment, it is based upon a tungsten (W) thin film sensor layer formed on an A1N substrate (a thin film is generally defined as having a thickness of about 100-10,000 angstroms). This combination of materials is particularly advantageous be- cause A1N has a thermal expansion coefficient of approxi- mately 4. 4x10-6/°K at 330° and 5. 3x10-6/°K at 1273°K, while the thermal expansion coefficient for tungsten is ap- proximately 4. 6x10-6/°K at 300°K and 5. 1x10-6/°K at 1273°K.

The thermal expansion coefficients for the two materials are thus very close to each other, allowing for a high degree of structural stability over a broad temperature range. The A1N substrate is insulative, while tungsten is generally conductive with a resistivity that varies in a known fashion with temperature. The temperature-resis- tivity characteristics of tungsten are discussed in American Institute of Physics Handbook, 3d Ed., 1982 Re- issue, pp. 9-41, the contents of which are incorporated herein by reference.

[0034] While W/A1N is the preferred material combination, the material system can be generalized to an A1N sub- strate with a conductive layer on the substrate, or an insulative substrate with a tungsten layer on the sub- strate, with the conductive layer in either case having an expansion coefficient within 1.00+/-0. 07 of the sub- strate over a predetermined temperature operating range, being substantially non-active with the substrate, and exhibiting substantially no solid-solubility or interdiffusivity with the substrate.

[0035] A1N has a thermal conductivity of approximately 1.7-2. 4 W/cm°K, about 10 times higher than ceramic Al203, making it very effective as a heater when it is itself heated by an adjacent tungsten layer. It also exhibits a

desirably high resistance to chemical reaction with met- als such as tungsten. It sublimates at approximately 2500°C and has an upper continuous use temperature from approximately 1150°C to 1800°C, depending upon its envi- ronment, making it useful for high temperature ranges.

Tungsten has a melting temperature of approximately 3410°C and is not known to chemically react with A1N be- low about 1800°C, which also makes it advantageous for high temperature operations in combination with A1N.

[0036] The complete lack, or immeasurably slow, chemical reaction, solid-solubility and interdiffusivity between tungsten and A1N at temperatures to approximately 1880°C in inert environments ensures that the tungsten circuit's cross-section does not decrease due to chemical reaction with the A1N substrate, and that the A1N substrate sur- face does not become electrically conductive. The provi- sion of an AlN substrate surface with crevasses which provide a means for attaching the tungsten circuit, and the closely matched temperature expansion coefficient over the operating temperature range, ensure that the tungsten circuit does not peel away from the A1N sub- strate surface during thermal cycling.

[0037] As described in further detail below, additional circuit layers consisting of Au, Pt or Au-Pt alloys may be provided on the tungsten circuit layer to perform three functions: (1) protect the tungsten circuit from oxidation; (2) bond the circuits on A1N substrates and caps together to form a multilayer circuit with electri- cally insulating A1N surfaces exposed on the top and bot- tom; (3) provide additional cross-sectional area to the circuit path.

[0038] Compatibility requirements between such additional circuit layers and tungsten (or WC, when carbon is re- acted with the tungsten to provide a bonding agent for another layer) include: (1) little or no chemical reac-

tion between them, up to the maximum operating tempera- ture; (2) they bond with limited interdiffusion and lim- ited solid-solubility, thus remaining distinct at and near the opposite facing interfaces; (3) the maximum solid-soluability between them is limited so that they do not form an isomorphic or pseudo-isomorphic phase dia- gram, up to the maximum operating temperature; (4) they do not form compounds with each other; and (5) their melting temperatures exceed the maximum operating tem- perature. Requirements 1-4 ensure that the additional circuit layers do not poison their interface with the tungsten (or WC), and that the combined circuit resis- tance does not drift under operational conditions.

[0039] As also described in further detail below, some embodiments include an encapulation of a borosilicate mixture (B203+SiO2). The borosilicate mixture is applied in unreacted form, and then reacted by heating the struc- ture to at least 1000°C. The reacted mixture is a glass that is bonded to surfaces that can be oxidized, and cov- ers layers that cannot be oxidized. It does not consume the circuit layers with which it is in contact, and re- mains an electrical insulator.

[0040] Electrodes for applying electrical signals to the tungsten circuit layer can be formed from extended area portions of the circuit layer itself, by additional cir- cuit material applied or deposited on and around the edges of an A1N substrate or cap, or by additional cir- cuit material applied or deposited within vias in the substrate or cap.

[0041] Additional electrode layers can be provided on top of the tungsten electrode for lead wire or ribbon attach- ment. Such additional layers can include carbon, plati- num or gold. Carbon provides a thermally activated bond- ing material that bonds W to W or W to Mo when heated above about 700°C, at which carbon is consumed by react-

ing with W and Mo to form a metal-carbide bonding inter- face which remains intact at temperatures greater than 1800°C. Platinum provides a base upon which Pt or Au can be welded; the bond will remain intact at temperatures equal to the melting point of Pt or the Au-Pt alloy formed during welding. Gold provides a base upon which to bond Au or to weld Pt ; the bond will remain intact at temperatures equal to the melting point of Au or the Au- Pt alloy formed during welding.

[0042] The additional electrode layers can also include layered Pt and Au, or Au-Pt alloy. This provides a base upon which to bond Au or weld Pt.

[0043] The thickness of the electrode materials which participate in the braising, bonding or welding process should be at least 0.05 times the diameter of the lead wire or the thickness of the lead ribbon. The electrodes can be exposed, encapsulated, or covered by A1N.

[0044] Process Considerations [0045] AlN-Substrate or Cap Surface in Contact with W: W films hold themselves to the A1N surface by electrostatic forces and by penetrating into crevasses in the A1N sur- face. Though difficult to quantify, observed results in- dicate that good W adhesion is obtained on all ceramic A1N surfaces (roughness average > 2 micro-inches (0.05 pms)). However, the maximum thickness of the W is di- rectly proportional to the A1N surface roughness average.

The maximum thickness of W on an A1N surface is about 100 times the surface roughness average.

[0046] Application to and Forming'on AlN substrate and Cap: W can be applied with the correct stociometry to an A1N surface by several vapor-phase deposition techniques, such as RF/DC sputtering, RF/DC co-sputtering, e-beam evaporation and chemical-vapor-deposition (CVD). The temperature of the A1N surface during W deposition is not

important, since adhesion occurs by physical-bonding, and not by chemical-bonding.

[0047] As deposited, W films will not be of theorectical density unless deposited by CVD. Film density can be in- creased, and grain-boundary area reduced, by thermal an- nealing. When density or grain-boundary area are impor- tant for protecting the W/AlN interface from additional circuit layer metals, the W should be annealed before ap- plication of additional circuit layer metals. The an- nealing temperature range is 800°C to 1400°C, with den- sity and grain-growth depending on time-at-temperature.

The annealing atmosphere should be vacuum or inert (Ar, N2).

[0048] Tungsten films can be partially or completely con- verted to WC, when desired to facilitate the bonding of an upper layer, by"forming". In this process, carbon is applied to as-deposited (preferred) or annealed W films by sputtering, physical vapor deposition or CVD, or physical application of graphite (e. g. , screen-printing).

The W film is transformed to WC by thermally induced dif- fusion ("forming"). The forming temperature range is 800°C to 1400°C, with the higher temperature preferred.

The forming'atmosphere should be vacuum or inert (Ar, N2). For purposes of this invention, references in the claims to"W"also include"WC", although WC has been found to have a lower thermal coefficient of expansion than W and therefore is not as desirable as W, except to bond an overlayer in place.

[0049] The minimum preferred as-deposited W thickness is equal to the roughness average of the A1N surface, if W is to be covered by additional circuit layer metal lay- ers. If W is to be the only film comprising the current path, its minimum thickness is determined by the greater of the following two requirements: (1) the post- processing thickness of the W should be at least 2 times

the roughness average of the A1N surface, or (2) its thickness times its width (cross-sectional area) should be sufficient to provide the current handling capability required by the sensor or heater. While there is no fun- damental limitation on maximum W thickness, low-mass ra- diation heater or circuit applications do not generally require a thickness greater than 10 microns.

[0050] Additional Circuit Layer (ACL) Metals: The as- applied/deposited ACLs can be comprised of one or more- layers, each applied/deposited sequentially. Each layer of the ACL can be applied by painting, screen-printing, electroplating, or vapor deposition (technique depends on material).

[0051] The processed ACL can be a single or multilayer film comprised of element (s) or alloy (s), or graded com- positions of both. The processed ACL can be as- applied/deposited, or it can be thermally processed to redistribute film composition. Melting can occur in one or more (but not all) layers of the ACL during thermal processing, but the resulting alloys must be solids at the same processing temperature. For example, Au-Pt al- loys can be formed by heating an Au/Pt multilayer struc- ture to a temperature in excess of the melting tempera- ture of Au, but below the melting temperature of the de- sired alloy. In this case only the Au melts, whereupon it is quickly consumed by the Pt to form an alloy with a higher melting temperature.

[0052] The thickness of the as-applied/deposited ACL metal (s) should be such that the post-processing thick- ness of the ACL is equal to or greater than the A1N sur- face roughness average. The minimum preferred as- deposited thickness of the ACL is 5xl0-6cm. The minimum thickness of the ACL may be determined by the requirement that the W + the ACL cross-sectional area be sufficient to provide the current handling capability required by

the sensor or heater. The maximum thickness of the ACL is limited by the strain it imparts to the circuit, rela- tive to the A1N, by the expansion coefficient difference between the W and the ACL. Experimental investigations performed to date show the upper thickness limit to be greater than 60 times the W thickness.

[0053] Carbon Reaction Bonding: Carbon provides a ther- mally activated bonding material, which bonds W/Mo wire/ ribbon to W electrodes when heated above about 700°C, with C consumed by reacting with W and Mo to form a WC or Mo-carbide bonding interface between W and W, or W and Mo; the bond will remain intact at temperatures exceeding 1800°C. The thickness of electrode materials which par- ticipate in the reaction bonding process should be suffi- cient to consume all of the C. The bonding process re- quires that C, W and Mo inter-diffuse, so that the rate at which bonding proceeds to completion is directly pro- portional to temperature. The thickness of the electrode materials which participate in brazing, bonding or weld- ing process should be at least 0.05 times the diameter of the wire, or 0.05 times the thickness of the ribbon or flattened wire. This is a minimum thickness requirement for a functional bond, as determined from experimental investigation. However, at this ratio strong bonding is difficult to achieve (low yield). An electrode thickness of at least 0.1 times the wire diameter or at least 0.1 times ribbon thickness is recommended, because yield is higher and the ruggedness of the bond is improved.

[0054] Au and Pt Alloy Bonding: A Pt electrode layer pro- vides a base upon which to weld Pt or weld/bond/braze Au wire/ribbon; the bond formed between the wire/ribbon and the electrode will remain intact at temperatures equal to the melting point of Pt or of the Au-Pt alloy formed dur- ing the bonding process. An Au electrode layer, layered Pt and Au electrode layers, and Au-Pt alloy electrode

layers each provide a base upon which to weld/bond Au wire/ribbon, or to weld/braze Pt wire/ribbon; the bond formed between the wire/ribbon and the electrode will remain intact at temperatures equal to the melting point of Au, or of the Au-Pt alloy formed during the bonding process. The thickness of the electrode materials which participate in the brazing, bonding or welding process is similar to carbon reaction bonding.

[0055] The lead-wires/ribbons should be attached to the electrode pads by bonding, brazing or welding. The ex- pansion coefficient of the lead wire material should be within 2 times that of the composite expansion coeffi- cient of the post-process electrode layers. The portion of the lead-wires to be bonded/welded/brazed to the elec- trodes may be flattened, in which case the lead-wire di- ameter perpendicular to its flattened surface is the ap- propriate diameter to use in determining the minimum thickness of electrode layers.

[0056] FIG. 1 illustrates a sensor in accordance with the invention in which a thin film of tungsten 2 is deposited on an A1N substrate 4, preferably in a generally serpen- tine fashion, with the opposite ends of the tungsten cir- cuit layer expanded in area to form electrodes 6. This structure can serve as a sensor and/or heater for the various applications discussed below.

[0057] FIG. 2 illustrates another embodiment in which a protective A1N cap 8 is secured by applying a thin film carbon layer (not shown separately) over the tungsten layer 2 of FIG. 1, applying a thin film tungsten layer (not shown separately) with a geometry that matches tung- sten layer 2 to the underside of cap 8, placing the cap 8 over the substrate 4 with their respective tungsten and carbon aligned, and thermally reacting the assembly to form a final circuit comprising a WC layer 10 sandwiched

between, and adhering together, the substrate 4 and cap 8.

[0058] In a demonstration of this embodiment, 1000 ang- stroms of tungsten was sputter deposited on both the sub- strate 4 and the cap 8, with 17200 angstroms of carbon deposited onto the substrate tungsten through a shadow mask.

[0059] Another variation is illustrated in FIG. 3. In this embodiment a thin film of gold or Pt 12 is applied over the tungsten layer 2 of FIG. 1 to protect it from oxidation, and also for physical protection.

[0060] In the embodiment of FIG. 4, unreacted B203+SiO2 has been applied over the tungsten circuit 2 and sub- strate 4 of FIG. 1, and thermally reacted to form an en- capsulation 14 that protects the underlying structure and further bonds the tungsten circuit to the substrate.

[0061] The borosilicate mixture consisted of 45wt% B203 + 55wt% Si02. It was reacted in air at 1000°C for 5 min- utes.

[0062] In the embodiment shown in FIG. 5, an A1N cap 16 has been placed on top of the unreacted B203+Si02 14 of FIG. 4, and the resulting structure thermally reacted to bond the A1N cap 16 in place for additional protection.

Both the encapsulation 14 and the cap 16 leaves the elec- trode pads 6 exposed to receive lead wires or ribbons used to apply an electrical signal to the tungsten cir- cuit. The borosilicate was reacted in air at 1000°C for 5 minutes to bond the cap in place.

[0063] In the embodiment shown in FIG. 6, Au and Pt lay- ers 18 and 20 are applied as thin films over the tungsten layer on the completed structure shown in FIG. 1; either Pt or Au can be the top layer. The result is a three- layer circuit on the A1N substrate 4. The assembly is then thermally reacted to form a two-layer circuit, hav- ing a configuration similar to that shown in FIG. 3, with

the lower layer comprising tungsten and the top layer an Au-Pt alloy circuit having a composition determined by the relative thicknesses of the Au and Pt layers.

[0064] Another embodiment with both an ACL and a protec- tive A1N cap is illustrated in FIGs. 7 and 8. Referring first to FIG. 7, a substrate assembly is provided as in FIG. 6, with Au and Pt layers 18 and 20 (not necessarily in that order) over tungsten layer 2 on substrate 4. An A1N cap 22 is provided with a similar three-layer conduc- tor structure, consisting of tungsten layer 2', with Au and Pt (or Pt and Au) layers 18'and 20', all with the same geometry as the layers on substrate 4. The cap 22 is positioned over substrate 4 with the various layers in alignment. The assembly is then thermally reacted to form a three-layer circuit as shown in FIG. 8, with tung- sten layers 2 and 2'adjacent the substrate 4 and cap 24, respectively, and sandwiching a Au-Pt alloy circuit layer 24. In a demonstration, a 100 angstrom thick layer of Au and 1000 angstrom thick layer of Pt were sputter depos- ited on a 1000 angstrom sputter deposited tungsten layer.

[0065] FIG. 9 illustrates the structure resulting from thermal reaction of the structure shown in FIG. 6, with an Au-Pt alloy circuit layer 26 on the tungsten circuit layer 2. A borosilicate encapsulation 28 has been formed over the structure in a manner similar to the encapsula- tion 14 of FIG. 4.

[0066] FIG. 10 illustrates an embodiment in which the structure of FIG. 9 is further protected by an A1N cap 30. With the cap in place, the borosilicate was reacted in air at 1000°C for 5 minutes to bond on the cap.

[0067] The invention can be advantageously used for a planar heater. FIG. 11 illustrates such a heater formed on a rectangular (including square) A1N substrate 32, which serves as a base for a tungsten heating element that can be fabricated as described above, without ACLs.

The tungsten layer is preferably formed as a series of generally parallel, serpentine-shaped strands 34 that are connected in parallel at opposite ends of the substrate by terminal strips 36, with electrodes 38 preferably pro- vided at opposite corners of the array.

[0068] The serpentine shape provides uniform heat-up over the entire A1N substrate area (except for its edges), and causes power dissipation to occur more uniformly during rapid thermal ramping. This allows the temperature to be ramped up very rapidly, at rates in excess of 500°C/sec- ond, without thermally shocking the substrate. If the circuit strands were straight, initial heating would occur most rapidly at the center of each strand and at the 180° turns at their ends, placing the substrate in danger of breakage due to thermal gradients parallel to its surface when heat is ramped very rapidly.

[0069] For a thin (0.01 inch to 0.014 inch (0.025cm- 0.036cm) thickness), low mass A1N substrate, the width of the strands 34 should not exceed 0.1 inch (0.25cm) and the spacing between strands should be uniform and not in excess of 0.07 inch (0.18cm), to ensure that thermal gra- dients across the spacing between strands will not result in substrate breakage.

[0070] The distance between the peak and valley of a ser- pentine strand, perpendicular to the strand conduction path, should not exceed the width of the strand itself.

The strands should be curved, without sharp corners. The distance between the strand peaks and valleys parallel to the strand conduction path should not be more than 1 inch (2.5cm) to prevent substrate cracking during rapid heat- ing, with shorter distances providing more uniform heat- ing over the substrate surface during rapid heat-up.

[0071] The combined resistance of the strands is the re- sistance of a single strand, divided by the total number of strands in the pattern, plus the resistance of one

electrode strip. Parallel conducting strands permits rapid heating to high temperatures, with moderate volt- ages, with thin film strands (100-10,000 angstroms thick).

[0072] With the terminal strips 36 provided in FIG. 11, the tungsten strands are parallel not only geometrically, but also electrically. The terminal strips 36 could be replaced with individual electrodes at the opposite ends of each strand, but this would not be as desirable. Lo- cating the electrodes 38 at opposite corners of the sub- strate keeps any variations in the voltage drop across different strands to a low level.

[0073] In an alternate embodiment illustrated in FIG. 12, multiple lead wire terminals 40 are provided on each electrode strip 36 at positions exactly opposite to each other, relative to the strands they service, with each electrode servicing no more than two strands.

[0074] FIG. 13 illustrates a preferred tungsten strand pattern for a circular substrate 42 that ensures the electric field intensity between oppositely positioned electrodes 44 does not exceed the dielectric breakdown strength of the AlN substrate at elevated temperatures; exceeding the dielectric breakdown strength can cause the substrate to crack or break. The dielectric strength of the A1N substrate decreases with increasing temperature, and is projected to be about 2000 volts/inch (787 volts/cm) at 1300°C.

[0075] The tungsten strands 46 are curved and extend gen- erally along respective lines of longitude that merge at opposite poles which comprise the electrodes 44. The strands are symmetrical about the center line 48 through the polar electrodes. To ensure that power is dissipated uniformly along the length of each strand, the strand widths perpendicular to their conduction paths should be

constant, although the strands can overlap as they emi- nate from an electrode.

[0076] Since strands at different distances from the cen- ter line 48 will have different lengths, the width of each strand relative to the other strands should progres- sively decrease towards the center line so that the power dissipation per unit length of each strand is close to the power dissipation per unit length of the other strands. This ensures that the A1N substrate will be uniformly heated and will not break due to thermal gradi- ents. To ensure that thermal gradients between strands perpendicular to the strand path will not cause a thin, low mass AlN substrate to break, the maximum separation between the mid point of the strands should not exceed 0.07 inch (0.18cm). The separation distance between strands does not have to be identical, but should be sym- metrical about the center line 48. For heater tempera- tures greater than 1300°C, the separation between strand midpoints may need to be less than 0.07 inch (0.18cm).

[0077] FIGs. 14 and 15 illustrate the application of the invention to a temperature sensor, a single-ended heater, or a heater capable of sensing its own temperature.

[0078] In FIG. 14 a voltage source 50 applies a voltage across a tungsten conductive film 2 that is deposited on an A1N substrate 4, as in FIG. 1. An ammeter 52 monitors the current through the tungsten film. When used to sense the temperature of the surrounding environment, a low voltage source is used to apply an actuating signal that does not significantly heat the tungsten film. The film's resistance is thus determined by the temperature of its surrounding environment, in accordance with the known temperature coefficient of resistance for tungsten.

The film's resistance, and thus the temperature of the surrounding environment, can be determined by dividing the applied voltage by the measured current. The re-

sponse of the tungsten film to a known applied voltage, sensed by the ammeter 52, varies with the surrounding temperature. The level of current that can be sustained in the tungsten film without causing the film to heat significantly depends upon several factors, such as the ambient temperature, the film's thickness, surface area and shape, and the heat capacity of the environment. In general, once a heating threshold is reached the film will experience rapid increases in heating with continued increases in the current level. The heating threshold increases as the temperature of the surrounding environ- ment goes up.

[0079] When used as an open-ended heater, a higher volt- age level is applied. The system can be calibrated in advance to at least approximate the amount of heating that will result from a given applied voltage level, and the ammeter 52 is not needed.

[0080] To operate as a heater capable of sensing its own temperature, ammeter 52 is added back to the circuit to sense the current through the tungsten film. The film's temperature can be precisely determined by using the voltage and current levels and the film's known geometry to determine the temperature from the tungsten tempera- ture-resistance curve at which the unit is operating.

[0081] The system of FIG. 15 is similar to that of FIG.

14, but instead of applying a voltage across the tungsten film and sensing the resulting current, a current source 54 drives an actuating current through the film 2, and a voltmeter 56 senses the resulting voltage across the film. The film temperature can again be determined in the same manner as for FIG. 14.

[0082] The invention can also be used to sense gas and liquid flow rates by measuring the change in temperature of a self-heated sensor that is located along the flow path, either immersed in the fluid or in contact with the

flow path wall. In the illustration of FIG. 16, the flow rate of a gas or liquid (indicated by arrow 64) is sensed by measuring the change in temperature of a self-heated sensor 66 of the type described herein that is located along the flow path, either immersed in the flow or in contact with the flow path wall 68. In this and subse- quent embodiments described herein the actuating signal is illustrated as being provided by an adjustable current source 70, with a voltmeter 72 sensing the voltage re- sponse of the sensor 66, which in turn corresponds to the sensor resistance and thus the fluid temperature. How- ever, a voltage could be provided as the actuating signal and the resulting current sensed. By comparing the sensed temperature. to the temperature at zero flow rate, and factoring in the heat capacity of the fluid, the flow rate can be determined.

[0083] In the differential flow rate sensor illustrated in FIG. 17, two or more sensors 74,76 of the type de- scribed herein are positioned in series along a fluid flow path 64. The flow rate is sensed by measuring the temperature difference between the sensors. At zero flow, the upstream sensor 74 is resistively self-heated by a known current from current source 70' (or a known voltage) to a resistance that represents a known tempera- ture. The downstream sensor 76 is biased by a much smaller, non-self-heating current or voltage. The actu- ating currents can be supplied from a common current source, or equivalently from a pair of separate current sources 70'as shown. The fluid flow removes heat from the upstream, self-heated sensor 74 and releases some of that heat to the downstream sensor 76. The change in voltage across each of the sensors is related, through the heat capacity of the gas or liquid, to the flow rate.

This type of differential flow rate sensing can be used

in a conduit that is parallel to the main fluid flow path to determine the flow rate through that path.

[0084] FIG. 18 illustrates an application of the inven- tion to a fluid level sensor, with a sensing element 66 in accordance with the invention disposed at a fixed lo- cation within, or in contact with the outer wall of, a liquid reservoir 78. A liquid 80 within the reservoir rises and falls, causing the sensor 66 to be alternately immersed in and cleared from the liquid 80. The heat ca- pacity of the environment in the immediate vicinity of the sensor varies, depending upon whether or not the liq- uid level has risen sufficiently to immerse the sensor.

With the sensor at a known level in the reservoir and em- ployed as a self-heating temperature sensor as described above, a determination of whether or not the sensor is immersed in the liquid can be made by sensing how rapidly heat is withdrawn from the sensor. While this applica- tion has been described in terms of a liquid sensor, it can also be used to detect the presence or absence of a gas within a closed reservoir, at least above a minimum density level.

[0085] Primary advantages of the invention when applied to the flow rate or fluid level sensing applications de- scribed above are a faster response time (related to the high substrate thermal conductivity), greater sensitivity (also related to the high substrate thermal conductiv- ity), and relatively low or no drift (related to the small expansion coefficient mismatch between the W sens- ing circuit and the A1N substrate).

[0086] In FIG. 19 the sensor 66 of the present invention is used to sense the gas pressure within a closed reser- voir 82. The pressure within the reservoir can be sensed by determining the current required to maintain a speci- fied voltage across the sensor, or the voltage required to maintain a specified current to the sensor. This data

can be related to the pressure/density of a gas with a known heat capacity within the reservoir, which can be a vacuum chamber. The current source 70 shown in FIG. 19 is adjusted to maintain a specified voltage level across the sensor, as sensed by voltmeter 72. The conversion of this current level to a pressure level within the reser- voir is illustrated by a pressure gauge 84 supplied by the current source. The present invention enables a more sensitive voltage/current sensing than do presently available thermocouples, and accordingly a more sensitive pressure sensing, and can also operate at higher tempera- tures than available thermocouples.

[0087] FIG. 20 illustrates the use of the invention to detect the chemical composition of its environment. The sensor 66 has the circuit material exposed to its sur- roundings, as in FIG. 1, such that all or part of the circuit material can be etched or oxidized by the gas to be sensed. A gas is illustrated as being introduced into the reservoir from a gas tank 86, or it can flow through a flow tube. Any etching or oxidation will reduce the cross-sectional area of the sensing circuit, thus causing its resistance for a known current/voltage level to in- crease. This increase can be detected by comparing the level of an applied current to the resultant voltage across the sensor, or the level of an applied voltage to the resultant current through the sensor. Applications include manufacturing and research/development of chemi- cals, pharmaceuticals, cosmetics, plastic and rubber polymers, metals and alloys.

[0088] FIGs. 21-23 summarize demonstrated and projected results of various embodiments described above, and com- pare them to comparable results for prior art sensors and heaters. FIG. 21 summarizes the operational temperature ranges for the embodiments corresponding to the identi- fied figure numbers (the results for FIG. 6 were obtained

after thermally annealing the structure). FIG. 22 summa- rizes advantageous results achieved with temperature sen- sors corresponding to the identified figure numbers, in comparison to the known characteristics of conventional platinum thin-film RTDs and thermistors.

[0089] FIG. 23 summarizes advantageous characteristics and results for planar heaters corresponding to the em- bodiments of the identified figures, in comparison to the following five types of conventional heaters: - Bulk : a conductive substrate material, generally graph- ite, ceramic SiC or ceramic BN, in which the substrate functions as the electrical conductor and heater.

- Foil : a thin metal conductor, mechanically affixed to a supporting insulator sheet or strip such as Kapton@, mica, glass or other ceramic sheet material.

- Rods and Bars: a metallic conductor such as the type used in electric stove tops and toasters.

- Planar Heaters: a metallic conductor, applied in a con- ductor pattern to an insulating substrate (such as A1N, A1203, BN, Si3N4 or BeO), by screen printing. The sub- strate is the heater, and receives its heat from the re- sistively heated circuit metal.

- Tungsten Wire: the resistively heated tungsten wire of the type used in light bulbs and tungsten-halogen lamps.

[0090] It can be seen from these summaries that the pre- sent invention achieves significant improvements in the operating temperature range, precision, and response time for both temperature sensing and heating. Improvements have also been noted in environmental range, sensitivity, lower drift, greater thermal shock resistance, and heat- ing efficiency.

[0091] While several illustrative embodiments of the in- vention have been shown and described, numerous varia- tions and alternate embodiments will occur to those skilled in the art. Such variations and alternate em- bodiments are contemplated, and can be made without de- parting from the spirit and scope of the invention as de- fined in the appended claims.