This invention relates to an integrated circuit comprising at least one pair of capacitor assemblies whose capacitance values are designed to have a specific ratio.

A number of known circuit configurations require high-ratio-accuracy capacitors. For example, in an article entitled "All-MOS Charge Redistribution Analog-to- Digital Conversion Technigues-Part-I," Journal of Solid- State Circuits, Volume SC-10, No. 6, pages 371-379, December 1975, the use of a binary-weighted capacitor array to perform high-speed successive approximation conversion is described. As stated on page 373 of the article, "the optimization of the ratio accuracy [of the capacitors] in the array is a prime consideration."

Additionally, high-ratio-accuracy capacitor geometries are required for so-called switched capacitor filter units. (See "Analog Sample-Data Filters," IEEE Journal of Solid-State Circuits, Volume SC-7, No. 4, pages 302-304, August 1972, for a description of such units.) In filters of this type, capacitor ratio accuracies of less than + 1% are frequently specified. Fabrication of matched-ratio capacitors in integrated-circuit form typically involves a processing sequence that includes various etching steps. During such a sequence, deviations from an optimum specified ratio frequently occur. Thus, for example, under- or over- etching during the capacitor fabrication process can cause unacceptable deviations from the specified ratio.

Ratio errors arising from variations in the etching steps utilized to fabricate integrated circuit capacitors can be reduced, for example, by interconnecting identical discrete capacitor units in parallel (see FIG; 6 on page 374 of the first above-cited article) . This interconnected array forms a relatively large capacitance-value unit. In turn, the relatively

small capacitance-value unit of a matched pair of ratio capacitors comprises one or more such identical units. In that way, a matched pair of units whose values are related by an integral ratio is provided. The capacitor units of the pair have respective perimeters and areas which are also each related by the specified ratio. Significantly, the prescribed capacitance ratio of such a pair of units is substantially insensitive to etching variations that occur during the capacitor fabrication process.

However, the foregoing method of fabricating ratio capacitors in integrated circuit form has several disadvantages. First, the yield and therefore the cost of such an array suffer from the requirement that, for each capacitor unit in the interconnected array, a separate pair of microminiature contact windows must be formed. For high-ratio arrays, this obviously becomes a burdensome requirement. Second, especially when a relatively high ratio of capacitance values is specified, the parasitic capacitance of the array tends to become unacceptably large and unpredictable. Third, the interconnected array approach does not lend itself to processing-insensitive fabrication of capacitor units having nonintegral ratios. In view of the above, continuing efforts have been made to attempt to devise improved ratio-capacitor geometries for implementation in integrated-circuit form. It was recognized that such improved geometries, if available, would improve the cost and performance of the overall circuit configurations of which the capacitors are important constituent elements.

In accordance with the invention, these problems a solved in an integrated circuit having a pair of capacitor assemblies which are designed to have a ratio k, the smaller-capacitance assembly having an effective area A _s and an effective perimeter P _s , and the larger-capacitance assembly having an effective area A^ and an effective perimeter P _L , and wherein

^'

In one specific illustrative embodiment of the principles of the present invention, these relationships are satisfied by utilizing a so-called H-section geometry to form the ratio capacitors. In the drawings

FIG. 1, which is a top view of one plate (and of only a tab portion of the other plate) of the smaller one of two ratio capacitors, shows a specific illustrative plate geometry that embodies the principles of the present invention;

FIG. 2, which is a top view of one plate (and of only a tab portion of the other plate) of the lower one of two ratio capacitors, also shows a specific illustrative plate geometry that embodies the principles of this invention;

FIG. 3 is a cross-sectional side view of a particular integrated-circuit embodiment of an entire capacitor that includes plates of the type shown in FIG. 1; and

FIG. 4 is a cross-sectional side view of a particular integrated-circuit embodiment of an entire capacitor that includes plates of the type shown in FIG. 2. In accordance with the principles of the present invention, ratio capacitors are made in integrated-circuit form utilizing a unique geometry. In one specific embodiment of the invention, each capacitor of a pair of ratio capacitors includes only two specially configured plates. Thus, only four contact windows need be formed to establish electrical connections to the pair of capacitors. Moreover, relatively large capactior ratios can be thereby realized with relatively low and predictable parasitic capacitances. Significantly, the geometry of the plates is such that a predesigned capacitance ratio is maintained with high precision despite etching variations that occur during the fabrication process. And this is so even if

OMPI

nonintegral ratios are specified.

The specific illustrative upper capacitor plate 10 shown in FIG. 1 comprises a generally rectangular configuration with two equal-sized portions respectively removed from the top and bottom thereof. Hereinafter this specific plate, which is configured similar to a capital letter H, will be referred to as having an H-section geometry, and the removal portions will be referred to as "bites". The upper plate 10 of FIG. 1 also comprises a relatively small-area contact tab portion 12 which protrudes from the right-hand part of the depicted upper plate. As will be evident from the description hereinbelow, an electrical connection to the plate 10 is made by contacting the tab portion 12 with a conductive material.

Disposed beneath the upper capacitor plate 10 of FIG. 1 in spaced-apart registry therewith is a lower capacitor plate. Only a contact tab portion 14 of the lower plate is shown in FIG. 1. Illustratively, if the contact tab portions 12 and 14 are ignored, the geometry of the lower plate is substantially identical to that of the upper plate. Accordingly, the lower plate will also be referred to herein as having an H-section geometry. The capacitance of the specific illustrative unit represented in FIG. 1 is determined essentially only by the overlapping portions of the depicted plates. In other words, the contact tab portions 12 and 14 of the plates do not contribute signif cantly to the capacitance of the unit. Hence, it is not a matter of major concern if the actual configurations of the portions 12 and 14 vary slightly from a preassigned ideal design. Any such variation that occurs during fabrication will have an insignificant- effect on the capacitance value of the integrated circuit embodiment of the unit. Accordingly, hereinafter the areas and perimeters of the contact tab portions will not be included as a part of the overall areas and perimeters of the plates. In determining the

overall areas and perimeters of the plates shown in the drawing, it will be assumed in effect that the various depicted tabs have been severed from their respective plates. As so modified, the plates will be referred to herein as having so-called effective areas and perimeters. The height of the plate 10 depicted in FIG. 1 is W _s , the width of the plate is L and the portions (bites) removed therefrom each have a height b and a width c. The contact tab portions 12 and 14 each have a height e and a width f. In one particular illustrative • embodiment made in integrated-circuit form in accordance with the principles of the present invention, _g = 40 micrometers (μ m) , L = 20 μm, b = 5 μ , c = 10 μm, a = d = 5 μm and e = f = 6 μm. In that particular -embodiment, the plate 10 is advantageously made of a 5,000-Angstrom-unit thick layer of doped polysilicon having a resistivity of about 50 ohms per square. In turn, the plate 10 (which is also shown in FIG. 3) is spaced apart from the substantially identical lower plate 16 (FIG. 3). Illustratively, as indicated in FIG. 3, a layer of silicon dioxide (about 1000 Angstrom units thick) is interposed between the plates 10 and 16. The capacitance of one illustrative such capacitor unit made in accordance with the principles of this invention measured 0.5 picofarads.

As a practical matter, both plates of a capacitor unit made in accordance with this invention need not be identical. As indicated above, the capacitance of each unit is determined essentially only by the overlapping portions of the plates. Thus, in accordance with one basic aspect of this invention, only one plate of each unit need have an H-section geometry of the unique type specified herein. The other plate of each unit may, for example, have only a simple rectangular configuration that is overlapped by the

H-section plate. But, on the other hand, for purposes of modularity in design and fabrication, and/or to reduce parasitic capacitances in the structure, it is often

^■ OMPI . , W w Y/I.POU- ^

advantageous to make the plates of each unit identical or substantially identical. In practice, it is feasible, for example, to construct a unit in which the upper plate thereof has a prescribed H-section geometry and in which the spaced-apart lower plate also has an H-section geometry but one that is slightly larger in size. Thus, illustratively, in one particular embodiment of such a unit wherein the upper plate has the dimensions specified in the paragraph immediately above, the lower plate of the unit has the following dimensions: _s = 44 μm, L = 24 μm, b = 5 μm, c = 6 μ , a = d = 9 μ . In this particular example, the dimensions of the upper plate are relatively critical and -are essentially determinative of the capacitance of the unit. The dimensions of the larger lower plate are not critical. Moreover, registration ■ between two such differently sized plates is not critical, the only requirement being that all of the upper plate overlap portions of the lower plate.

Herein, for purposes of a specific illustrative example in the detailed description, idealized units each having matched identical plates (except for the contact tabs thereof) will be assumed. In practice, in those cases wherein the plates of a unit are not identical, the effective perimeters and areas specified in the relationships set forth below are those of only the critical H-section plate of each unit.

A cross-sectional side view of an entire specific illustrative capacitor unit, including the plates represented in FIG. 1, is shown in FIG. 3. FIG. 3 is a cross-section viewed at the line 3-3 indicated in FIG. 1. The capacitor unit shown in FIG. 3 includes the aforedescribed plates 10 and 16 including respective contact tab portions 12 and 14. The depicted unit further includes a silicon substrate 18 having a layer 20 of silicon dioxide thereon. In one specific illustrative embodiment, the thickness of the layer 20 between the top of the substrate 18 and the bottom of the lower plate 16 was about 9000 Angstroms units. In that embodiment, the

thickness of each of the plates 10 and 16 was about 5000 Angstrom units and the thickness of the silicon dioxide region 22 between the overlapping portion of the plates 10 and 16 was about 1000 Angstrom units, which are the same illustrative values mentioned earlier above. Disposed on top of the upper plate 10 is a layer 24 of silicon dioxide about 8000 Angstrom units thick.

Individual electrical connections to the plates 10 and 16 are also shown in FIG. 3. These connections comprise, for example, conductive regions 26 and 28 deposited in small-area contact windows formed in the silicon dioxide in any one of various standard ways known in the art. These windows are respectively made in aligned registry with the contact tab portions 12 and 14. Accordingly, when a conductive material such as aluminum is deposited on the top surface of the silicon dioxide region 24, portions of the aluminum extend into the windows and establish electrical connections to the respective tab portions of the plates 10 and 16. Subsequently, the aluminum layer deposited on the entire top surface of the unit is patterned in a standard way to form individual spaced-apart conductors. In turn, these conductors are utilized to connect the indicated capacitor unit to other components included in an overall circuit array. A portion of a specific illustrative larger- capacitance-value capacitor unit made in accordance with the principles of the present invention is shown in FIG. 2. The entire upper plate 30, including a contact tab portion 32, and only a contact tab portion 34 of the lower plate of this unit are represented in FIG. 2. The two plates are assumed to be aligned in exact overlapping registry. Again, for the same reasons discussed above in connection with FIG. 1, the tab portions will in effect be disregarded when the areas and perimeters of the plates are considered.

The advantages of the present invention are particularly evident when relatively large nonintegral capacitor ratios are specified. Thus, for example,

advantageous matched units having a capacitance ratio of 15:1 + 0,5% have been built and successfully tested. But, so as not to unduly complicate the drawing, a relatively small-ratio pair of capacitor units is depicted in the figures herein. Thus, the particular illustrative unit represented in FIGS. 2 and 4 is assumed to have a capacitance value 3.1 times that of the unit represented in FIGS. 1 and 3.

In the course of fabricating ratio-capacitor units, it is advantageous to design the units to exhibit a generally modular structure. Thus, for example, fabrication of the larger unit partially shown in FIG. 2 is facilitated by designing the unit to comprise constituent parts each of which conforms in width and general geometry to the corresponding characteristics of the smaller unit partially shown in FIG. 1. Thus, in accordance with an aspect of the principles of the present invention, each of the plates depicted in FIG. 2 comprises, for example, three main H-section parts _j each L units in width. Further, each such plate includes an additional fractional section having a width d _j . The capacitance attributable to the overlapping fractional sections comprises the nonintegral portion of the specified ratio. Thus, for the particular example assumed above, the spaced-apart fractional sections are designed to exhibit a capacitance that is 0.1 times that of the unit partially shown in FIG. 1. Each of the three main parts of FIG. 2 is designed to exhibit a capacitance that is the same as that of the FIG. 1 assembly. Accordingly, overall, the units of FIGS. 1 and 2 are related by a capacitance ratio of 3.1:1. Herein, the specified ratio is designated by the letter k_.

An entire illustrative capacitor unit that includes the plates represented in FIG. 2 is shown in FIG. 4. Except for the inclusion therein of larger-area and differently configured plates, the capacitor unit of FIG. 4 is advantageously identical to the unit depicted in FIG. 3 and described earlier above.

To make the indicated capacitor ratio relatively

- O

insensitive to etching variations, several design criteria must be ^{^} embodied in the depicted capacitor structures. First, the effective' area _L of each plate of the larger capacitor unit must be 1_ times the effective area A _s of each plate of the smaller unit (that is, _L = kA _g ) .

Second, the effective perimeters and areas of the plates must be related by the expression

^A S _ ^A L P _S ^{= p} L ' where Pg is the effective perimeter of each plate of the smaller unit and P _L is the effective perimeter of each plate of the larger unit.

Additionally, it is known that corner rounding effects occur during the fabrication of ratio-capacitor units in integrated circuit form. These effects are typically different in the case of so-called 90-degree and 270-degree corners. (In FIG. 2, reference numeral 36 indicates a typical 90-degree corner and numeral 38 indicates a typical 270-degree corner.) Ideally, the number of each type corner in the larger-capacitance unit should be m times the corresponding number included in the smaller-capacitance unit, where m is the integer portion of the ratio .k. As is evident from FIGS. 1 and 2, the number of

270-degree corners included in the larger plate 30 is exactly 3 times the number of such corners in the smaller plate 10. The number of 90-degree corners in the larger plate 30 is, however, only twice the number of such corners in the smaller plate 10. This is not ideal, but the relationship between the indicated 90-degree corners approaches m more closely as _k increases.

In the particular illustrative capacitor units represented in the drawing, the bites removed from each of the larger plates (such as the plate 30) are each advantageously also £ units wide. But, in accordance with the principles of the present invention, each such bite is h units high. The dimension h_ is selected, in accorda

with the systematic procedure set forth below, to ensure that the aforespecified area and perimeter relationships are satisfied.

If the larger-capacitance unit of a pair of ratio capacitors were formed by, in effect, simply butting together a number of identical FIG. 1-type sections, it is apparent that the area relationship specified earlier above would thereby be satisfied. But, of course, the specified perimeter relationship would not thereby be met. Hence, in accordance with one specific aspect of the principles of the present invention, area is removed from the center portion of each butted-together section and added to the legs of the multiple-section unit. In that way, the specified area ratio of the two capacitor units is preserved while at the same time the perimeter of the larger-capacitance unit is selectively augmented to achieve exactly the specified perimeter ratio. With respect therefore to the basic H geometry of the illustrative smaller-capacitance unit, the larger unit is seen to comprise plural modified H-geometry sections.

The desired aforespecified perimeter relationship between two capacitor units of a ratio pair of the type shown in the drawing can be expressed as follows: n ² [8b ² (2m+l)]

n [8b ² (m - k)+4bW _s (2m-k+2)+4bL(m-k) 3 + (1)

[2W _s L(m-k)+4bW _s (m-k)+2w|(l-k)+2A _f ] = 0

where n - number of additional b-high bites removed from each bite region of larger capacitor W _s = height of smaller capacitor L = width of smaller capacitor b = height of bite in smaller capacitor = number of complete H sections in larger capacitor

k = total capacitance ratio

h _ψ = area of fractional section of larger capacitor.

Expression (1) above can be rewritten as follows:

An ² + Bn + C = 0 (2)

where

A = 8b ² (2m+l)

B = 8b ² (m-k)+4bW _s (2m-k+2)+4bL(m-k)

C = 2W _s L(m-k)+4bW _s (m-k)+2W (l-k)+2A _f ,

Therefore,

n _a -B ÷y B ² -4AC_ ₍₃₎

2A Assume a particular illustrative capacitor pair of the specific type shown in the drawing, in which W _s = 40.02 μm, L = 20 μm, b = 5.0 μm, c = 10 μm, = 3, k = 3.1 and A _f = 70.04 μm ² . Then A = 1400.0, B = 3861.7, C = -6826.0, and n = 1.224. Accordingly, since W _L = W _s +2nb, h = 2nb+b and _c = Wg-2(n+l)b, _L = 52.26 μ , h - 17.24 μm, W<-. = 17.78 μm and d _j = 1.34 μm.

For the particular illustrative values set forth above, the area _s of a plate of the smaller depicted capacitor is 700.35 μm ² . The area A _L of a plate of the larger depicted capacitor is 2171.09 μm which is exactly 3.1 times Ag. The perimeter Pg of a plate of the smaller capacitor is 140.035 μm, and the perimeter P _L of a plate of the larger capacitor is 434.109 μm. Accordingly, it is

Aς Ar evident. that the relationshi —- = —— is in fact realized

^P S ^P L in the depicted structures.

Finally, it is to be understood that the above- descr.ibed arrangements are. only illustrative of the principles of the present invention. In accordance with these principles, numerous modifications and alternatives may be devised by those skilled in the art without departing from the spirit and scope of the invention.

example, although particular emphasis herein has been directed to capacitor assemblies that include H-section geometries, it is apparent that alternative geometries that satisfy the above-specified area and perimeter relationships are feasible. In each such alternative case, the larger-capacitance assembly of a ratio pair comprises only two continuous spaced-apart plates.