joining technique must be utilized. One such is the use of field-assisted bonding in which a layer of borosilicate glass between the two pieces of silicon permits bonding of the silicon with the borosilicate glass at about 500°C in a vacuum, the silicon pieces being totally attracted to one another by a DC electric field established by on the order of 100 or 200 volts impressed between them, described in NASA Tech Brief B74-10263, January 1975, entitled "Low-Temperature Electro¬ static Silicon-To-Silicon Seals Using Sputtered Borosilicate Glass". The device in the afore¬ mentioned patent also suffers from a very low ratio of variable capacitance to parasitic fixed capa- citance inasmuch as the periphery of the device has conductive surfaces closer to each other than the surfaces of. the deflectable portion of the device.

In order to improve the variable to fixed capacitance ratio, and particularly to mitigate the parasitic fixed capacitance (that part which is not varied as a function of diaphragm flexure in response to pressure changes) , it is necessary to provide topographical shaping, such as moats, pedestals or pistons, to cause the relatively movable capacitive plate portions to be close to each other in contrast with the fixed portions of the conductive body of the device. Additionally, in the event that very small devices are made (such as by large scale integrated circuit processing of wafers to form a plurality of devices per wafer pair) , the small surface area of the opposed capa¬ citive plates requires close spacing in order to

OMPI

- 1 -

Description

Silicon-Glass-Silicon Capacitive Pressure Transducer

Technical Field This invention relates to capacitive pressure transducers of the type in which a pair of silicon pieces are joined together with borosilicate glass by a field-assisted bonding process, to form an evacuated capsule with opposing conductive surfaces forming the plates of a pressure-variable capaci¬ tance, electrical connection to which is made through the bulk of the silicon.

Background Art

Pressure transducers are known to take a variety of forms. One form relies on the pressure- induced deflection of a thin diaphragm. In the case of a capacitive pressure transducer, the diaphragm deflection causes a variation in the distance between a pair of surfaces which form the plates of a variable capacitor. In U.S. Patent No. 3,634,727, there is disclosed a capacitor formed of two wafers of silicon insulated from each other by glass and joined together by a low temperature glass or by brazing thin metal films deposited on the glass. The device disclosed in said patent requires the processing of two wafers of silicon to provide only a single transducer, and the method of joining the two wafers of silicon is cumbersome. In order to process wafers to provide a large number of pressure transducers from a single pair of processed wafers, a suitable

have an adequate capacitance for desired sensitivity to pressure.

In mass production of silicon capacitive pres¬ sure transducers utilizing known microcircuit and thin film technology, particularly where small devices are desired and capacitor plate surface spacing becomes very small, it is necessary that the pro¬ cesses be selected and performed in such a fashion as to control dimensions very accurately. If, for instance, a pedestal or piston is too tall, short circuiting can result in a bad device; if a pedestal or piston is too short, then an inadequate capacity or variable capacitance as a function of pressure may result.

Disclosure of Invention

Objects of the invention include provision of improved dimensional control in the processing of wafers for making miniature silicon capacitive pressure transducers, and reduction of parasitic fixed capacitance in miniature silicon capacitor pressure transducers.

. According to the present invention, silicon capacitive pressure transducers, in which the silicon provides conductivity to opposing capa- citor plates, the distance and therefore the capacitance between which is variable as a result of deflection in response to external pressure changes, the parasitic, fixed capacitance of the nonvariable portion of the device is significantly reduced by spacing conductive portions thereof with borosilicate glass.

In accordance with the present invention, control over the dimensions of topygraphy of one

or more wafers being processed to provide silicon capacitive pressure sensor is achieved by utilizing the depth of sputtered borosilicate glass as the principal dimension defining step, in contrast with the etching of silicon or other material to a time- controlled depth.

According to a first aspect of the present invention, two pieces of silicon, upon which are disposed closely spaced opposing surfaces of a capacitor, the spacing of which varies with ex¬ ternal pressure, are separated about the peripheral fixed capacitive portion by borosilicate glass. In accordance further with the present invention, the borosilicate glass is deposited and etched in pro- cess steps resulting in principal dimensional con¬ trol between the two pieces of silicon so as to provide relatively accurate dimensional control in contrast with etching of silicon or other materials. The invention may be practiced in silicon pressure transducers of a wide variety of types with or without metallization of borosilicate glass or silicon to form capacitive plates, to provide low parasitic capacitance and acceptably accurate dimensional control in the processing thereof.

The invention may be practiced utilizing microcircuit and thin film technology which is well within the skill of the art, in the light of the teachings which follow hereinafter.

The foregoing and other objects, features and advantages of the present invention will become more apparent in the light of the following detailed des¬ cription of exemplary embodiments thereof, as illus- trated in the accompanying drawings.

Brief Description of Drawings

Figs. 1-9 are simplified side elevation, sectional views of one or more silicon wafers being processed to form exemplary silicon capa- citive pressure transducers in accordance with the invention; and

Figs. 10 and 11 are simplified side elevation views of alternative embodiments of silicon pres¬ sure transducers employing the present invention.

Best Mode for Carrying Out the Invention

Referring now to Fig. 1, a wafer 14 of doped silicon has a layer 16 of borosilicate glass depo¬ sited thereon. The silicon may be either N or P type having at least 10 15 impurity per cc so as to achieve something under 1 ohm centimeter con¬ ductivity. The borosilicate glass 16 may, for instance, be Corning 7070 Glass, or other boro¬ silicate glass such as "Pyrex". The glass 16 may be RF sputtered utilizing well known tech- niques so as to achieve a depth on the order of 2.5 microns. In Fig. 2, well known photoresist and etch techniques are utilized to etch the glass layer 16 so as to provide a matrix 18 of glass defining a plurality of circular sites 20 of ex- posed surface of the silicon wafer 14, at each of which a pressure transducer is to be formed as described hereinafter. As shown in Fig. 3, a next step is to deposit additional borosilicate glass in a layer 22 on the order of 6 microns thick. Then, as shown in Fig. 4, conventional photoresist and etch techniques are utilized to etch the layer 22 so as to result in a thickened

fOκEAi

O PI

.Λ,_ WIPO

matrix 18, 21 defining annular moats 24 and cir¬ cular pedestals 26 with centrally disposed aper¬ tures 28 therein, the apertures 28 and moats 24 exposing the upper surface of the silicon wafer 14. The deposition depth of the glass layer 16 and the glass layer 22 provide dimensional control to the thickness of the matrix 18, 21 (Fig. 4) and the circular pedestals 26, in accordance with one aspect of the invention, as described more fully hereinafter.

In Fig. 5, a layer of aluminum, on the order of 5,000 angstroms thick, is RF sputtered or eva¬ porated using electric beam or resistance methodo¬ logy, to provide a mask for further etching of the silicon and to provide a capacitor plate, as described hereinafter. In Fig. 6, utilizing well known photoresist and etch techniques, the alu¬ minum layer 30 is etched away at the bottom of each of the moats 24. Then, the silicon wafer 14 is plasma etched so as to deepen the moats in each of the sites on the wafer to provide a circular moat 24a which extends into the silicon. Notice that the etching of silicon as illustrated in Fig. 7 does not change the dimensional difference between the height of the circular pedestals 26 and the matrix 18, 21. Thus, the particular depth of the etch need concern only leaving sufficient material between the moat 24a and the lower sur¬ face of the silicon wafer 14 for the desired strut- tural strength, and/or the desired pressure respon¬ sive flexure characteristics, as described herein¬ after. As illustrated in Fig. 8, ordinary photoresist and etch techniques are utilized to remove only a portion of the aluminum mask which had been

- JREA

OMPI

deposited in the step illustrated in Fig. 5, so as to provide a conductive surface on the top of the borosilicate glass circular pedestals 26. In this step, the glass in the matrix 18, 21 and the silicon at the base of the annular moats 24a provide etch stops so that there is no dimensional change as a consequence of removing the unwanted aluminum. Fig. 8 illustrates the completion of processing of one of two wafers so as to provide a plurality of sites at which the capacitive pressure transducers can be formed. The next step is to overlay the processed wafer of Fig. 8 with a wafer 32 of conductive silicon, on the order of 250 microns thick, as illustrated in Fig. 9, and bond the wafers together. The field- assisted bonding may take place in a vacuum of about 10 —6 Torr at approximately 500°C, with a voltage, impressed plus to minus from the wafer

32 to the wafer 14, of on the order of 75 to 125 volts. This causes the silicon wafers to be

^• attracted to each other as the glass 18a bonds to the wafer 32, thus ensuring a pressure tight seal at each of the sites of the processed wafer 14. Then, the wafer may be diced appropriately, as shown by the dash lines 36 in Fig. 9, such as by sawing, to provide a plurality of individual pressure transducers 34.

Referring to Fig. 9, because the glass portion 21 around the periphery of the device is formed during the same deposition as the pedestal 26 (the deposition of glass layer 22 in Fig. 3) , regardless of how thick the layer 22 is made, the top of the pedestal 26 will be separated from the bottom of

the wafer 32 by the thickness of the glass portion 18 which in turn is determined by the thickness of the layer 16 (Fig. 1). Thus dimensional control, in providing a prescribed distance between the silicon pedestal 26 and the lower surface of the silicon wafer 32, is maintained simply by proper control of the deposition of layer 16. Of course, the plate-to-plate spacing will be less, as de¬ termined by the thickness of the metal layer 30. On the other hand, the distance between the silicon wafer 14 and the silicon wafer 32 is independently controllable by the thickness of the layer 22 (Fig. 3) and therefore the glass portions 21 and the pedestal 26. The sputtering of glass can be controlled in the thickness to -57o of desired thickness, even in high volume production. This contrasts with etching of silicon (e.g., such as to create the moat 24 and pedestal 26 from a thick wafer) , for which the depth of etching (other than when using a metallic etch stop) can only be con¬ trolled to -15% of the desired depth. This dimen¬ sional control is one aspect of the present inven- ^• tion. It should be noted however, that the amount of glass used (the thickness of the layer 22, in particular) cannot be increased indiscriminately, because the difference in the temperature coeffi¬ cient of expansion of borosilicate glass from that of conductive, single crystal silicon, could re¬ sult in structural flaws occurring as a result of temperature changes in a device when in use. But, for a minimum parasitic capacitance, the glass wall structure 18, 21 should be at least four times as great as the distance between the

OMPI

plates (30 , 32) .

Another embodiment of the present invention is illustrated in Fig. 10. This is substantially the same as that illustrated in Fig. 9 with the exception of the fact that a moat 31a is formed in the upper wafer 32a rather than within the glass 21a in the lower wafer 14a. Obvious varia¬ tions in the procedures set forth in Figs. 1-9 would be utilized to form the device of Fig. 10, the most notable difference being the etching back of the aluminum to form the metallic plate surface and contact 30a, a similar aluminum mask being utilized in the process to form the moat 31a in the wafer 32a. The embodiment of Fig. 10 may be advantageous where a piston 26a having high mobility is desired for extreme sensitivity.

In another embodiment employing one aspect of the invention as illustrated in Fig. 11, the glass material 21b is utilized to define the spacing between the wafer 14b and the wafer 32b, but the distance between the upper surface of a pedestal 26b and the lower surface of the wafer 32b is wholly dependent on the depth of etch of silicon in formulating that which becomes a moat 31b after the glass 21b is disposed and reverse etched thereon. Thus, the embodiment of Fig. 11 does not have the accurate dimension control of the spacing of the capacitor plates, but does employ the aspect of the invention where the two wafers are separated signi- ficantly by the glass portion 21b, thereby to reduce the parasitic static capacitance around the periphery of the device.

O PI

In the embodiments of Figs. 9 and 10, one of the plates of the capacitor is formed of the metal¬ lization 30, 30a, wherein in the embodiment of Fig. 11, both plates of the capacitor are formed by the surfaces of the silicon pieces 14b, 32b. In the embodiments of Figs. 10 and 11, a pedestal 26a, 26b is formed of silicon only, whereas in the embodi¬ ment of Fig. 9 the pedestal 26 is formed at least in part of glass. In all three embodiments, how- ever, the silicon pieces are joined together and separated by glass having a dimension between the two pieces (vertically in the figures herein) which is substantially larger than the distance between the two capacitor plates. In fact, it has been found that the glass wall structure which provides the sidewalls to the evacuated chamber formed between the two pieces of silicon should have a dimension between the pieces of silicon (vertical in the figures) which is at least four times greater than the spacing of the capacitive plates. This provides a substantial reduction in the invariable, parasitic capacitance between the plates around the periphery, while the close spacing between the plates provides for an increase in the dynamic range (the variable capacitance portion of the total capacitance) as a function of pressure. The invention may be practiced in a wide variety of configurations utilizing pistons which are either movable or relatively rigid, with double moats or single moats, and employing other features as well in a wide variety of shapes and sizes. How¬ ever, for best results in the field-assisted bonding step, the glass should all be on one wafer so that

- li ¬

the seal formed during bonding is a silicon/glass seal. Similarly, although the invention has been shown and described with respect to exemplary embodiments thereof, it should be understood by those skilled in the art that the foregoing and various other changes, omissions and additions may be made therein and thereto, without departing from the spirit and the scope of the invention.