Layout technique for C3MOS inductive broadbanding

Description

CROSS-REFERENCES TO RELATED APPLICATIONS

This application is a continuation of U.S. patent application Ser. No. 09/861,143 filed May 17, 2001 now U.S. Pat. No. 6,864,558 which is related to U.S. patent application Ser. No. 09/610,905, filed Jul. 6, 2000, entitled CURRENT-CONTROLLED CMOS CIRCUITS WITH INDUCTIVE BROADBAND, the disclosures of which are hereby incorporated by reference herein.

BACKGROUND OF THE INVENTION

The design of an integrated circuit (IC) requires that a layout be designed which specifies the arrangement of the various circuit components on the major surface of a semiconductor substrate, for example a silicon crystal.

Since many circuit elements are repeatedly utilized, these circuit elements are reduced to cells. The layout can be generated by arranging the cells and connecting them using conductive interconnects. Layout is usually performed utilizing sophisticated software tools well-known to persons of skill in the art.

The layout of the interconnects is a complex geometrical problem. However, in high frequency ICs the layout must also account for electromagnetic effects, which cause parasitic resistance and capacitance which can degrade the performance of the IC.

FIG. 1 depicts the standard design of a cell 10, a layout of two cells, and the interconnection of the cells where each cell 10 includes an inductor region 12, a resistor region 14, and a transistor region 16. An example of such a circuit is disclosed in the aforementioned application.

In FIG. 1 a first set of conductive lines 18 couples the inductor region 12 to the resistor region 14 and a second set of conductive lines 20 connects the resistor region 14 to the transistor region 16. A set of cell to cell signal interconnects 22 couples the output nodes of the first cell to the inputs of the transistor region of the second cell. All the regions 12, 14, and 16 are rectangular and have a characteristic lateral dimension: DI for the inductor region 12, DR for the resistor region 14, and DT for the transistor region. As is apparent from FIG. 1, the lateral dimension of the cell 10 is about equal to the dimension of the largest circuit element, in this case the inductor, and is about equal to DI. Because of the symmetrical design of the inductor, resistor regions, and transistor regions, the regions tend to be aligned and the length of the conductive lines connecting the regions minimized. The length of the cell-to-cell interconnects, which transmit high frequency signals, is thus very long because of the large lateral dimension of the inductor compared to the other regions.

In very high frequency applications, the interconnect parasitic resistors and capacitors form an RC network that plays a very important role. This RC network attenuates the high frequency clock and creates Inter-Symbol Interference (ISI) jitter on the data. These effects become even more important for C3MOS cells with inductive broadbanding. (The presence of the inductors in these cases changes the RC networks to RCL networks). As described above and in the referenced patent, the load here includes an inductor in series with a resistor. Since the physical size of the inductor is typically an order of magnitude bigger than the physical size of the resistor, these cells require a small area for transistors and resistors and a very large area for inductors. As depicted in FIG. 1, this makes the cell-to-cell interconnects 22 very long. Since the length of the interconnects is directly proportional to their parasitic resistance and capacitance, whatever speed improvement is gained through inductive broadbanding can be lost due to these additional parasitic effects if the layout is not done carefully. Moreover, if the inductors are close to metal or active areas, magnetic coupling further degrades the speed improvement.

BRIEF SUMMARY OF THE INVENTION

According to one aspect of the invention, an improved cell layout for a C3MOS circuit with inductive broadbanding effectively isolates the inductor region from metal and active layers and reduces the length of cell-to-cell interconnects.

According to another aspect of the invention, first and second cells have a common boundary. The inductor, resistor, and transistor regions of each cell are aligned near the common boundary to reduce the length of the cell-to-cell interconnect.

According to another aspect of the invention, the length of the conductive lines connecting the inductor region to the resistor region is greater than the length of the conductive lines connecting the resistor region to the transistor region to isolate the inductors from metalizations and active areas.

According to another aspect of the region, the parasitic capacitance of the lines connecting the inductor and resistor regions is less than 20% of the load capacitance, thereby improving circuit performance.

Other features and advantages of the invention will be apparent from the following detailed description and appended drawings.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a block diagram depicting a standard cell layout;

FIG. 2 is a C3MOS buffer circuit with inductive broadbanding; and

FIG. 3 depicts an embodiment of a layout cell.

DESCRIPTION OF THE SPECIFIC EMBODIMENTS

In one embodiment of the invention, the circuit elements are fabricated utilizing ultra-high-speed logic circuitry implemented in silicon complementary metal-oxide-semiconductor (CMOS) process technology. A distinction is made herein between the terminology “CMOS process technology” and “CMOS logic.” CMOS process technology as used herein refers generally to a variety of well established CMOS fabrication processes that form a field-effect transistor over a silicon substrate with a gate terminal typically made of polysilicon material disposed on top of an insulating material such as silicon dioxide.

FIG. 2 depicts a buffer circuit fabricated utilizing, by way of example, not limitation, C3MOS technology which is described in detail in the above-referenced patent application. It is to be understood that the present invention is useful in many contexts and is not limited to particular circuit designs.

FIG. 2 is a schematic diagram of a buffer circuit utilizing inductive broadbanding, illustrates the basic C3MOS buffer 200 with shunt inductors L, and load capacitors C_L. A pair of n-channel MOSFETs 202 and 204 receive differential logic signals V_in+ and V_in−, at their gate terminals, respectively. Resistive loads 206 and 207 in series with shunt inductors 208 and 209 connect the drain terminals of MOSFETs 202 and 204, respectively, to the power supply VDD. Drain terminals of MOSFETs 202 and 204 form the outputs V_out− and V_out+ of the differential pair, respectively. In a preferred embodiment, the shunt inductors 208 and 209 are spiral inductors coupled to the substrate utilizing standard techniques. Resistive loads 206 and 207 may be made up of either p-channel MOSFETs operating in their linear region, or resistors made up of, for example, polysilicon material. In a preferred embodiment, polysilicon resistors are used to implement resistive loads 206 and 207, which maximizes the speed of buffer 200. The source terminals of n-channel MOSFETs 202 and 204 connect at node 210. A current-source n-channel MOSFET 212 connects node 210 to ground (or negative power supply). A bias voltage VB drives the gate terminal of current-source MOSFET 212 and sets up the amount of current I that flows through buffer 200.

In FIG. 2 a first pair of nodes, A and AB, and a second pair of nodes, B and BB are depicted. The first pair of nodes are coupled to the outputs and are thus sensitive to parasitic series resistance and shunt capacitance. Accordingly, it is desirable to reduce the length of cell-to-cell interconnects coupled to the first pair of nodes as much as possible.

On the other hand, the second pair of nodes are not very sensitive to parasitic resistance and shunt capacitance so that the conductive lines between the resistor region 14 the inductor region 12 can be increased in length to increase the isolation of the inductors from other circuit elements.

FIG. 3 depicts an embodiment of a cell layout with first and second cells 10a and 10b having a common boundary 30. In FIG. 3, referring to the first cell 10a, the inductor region, resistor region, and transistor regions all have their right edges aligned so that the right edge of each region is disposed near the right edge of the cell. Similarly, referring to the second cell 10b, the inductor region, resistor region, and transistor region are all aligned so that the left edge of each region is disposed near the left edge of the cell. Because of this alignment, the signal lines connecting the output nodes of the first cell to the input nodes of the transistor region in the second cell have a reduced length compared to the signal interconnects of the standard cell in FIG. 1. This reduced length reduces the parasitic resistance and capacitance of the signal lines.

Additionally, the length of the second set of conductive lines between the resistor and transistor regions is kept small while the length of the first set of conductive lines between the inductor and resistor regions is increased to increase the isolation of the inductor region from the signal lines. For these first interconnect lines the parasitic resistance is added to the load resistance of the resistor region without degrading the bandwidth of the cell. Thus, the parasitic burden is shifted to the relatively insensitive first set of conductive lines from the highly sensitive signal interconnects.

Further, the inventors have discovered that overall performance can be improved over the case where there is no parasitic capacitance This can be explained as follows. The basis of the inductive shunt-peaking technique is to add a pair of inductors to the circuit so that the circuit's natural frequencies are changed in such a way that the circuit's transient response becomes faster. When a moderate parasitic capacitance (less than 20% of the load capacitance) is added to the first set of conductive lines, the circuit's natural frequencies change in a similar way, such that the circuit's transient response is further improved. The improvement ceases, however, if the parasitic capacitance becomes larger than 20% of the load capacitance.

The invention has now been described with reference to the preferred embodiments. Alternatives and substitutions will now be apparent to persons of skill in the art. For example, other fabrication technologies can be utilized instead of CMOS processing technology. Further, although a C3MOS buffer has been used as an exemplary embodiment, the principles of the invention are extendable to other circuits such as flip-flops, latches, etc. that include an inductor in series with a resistor. Accordingly, it is not intended to limit the invention except as provided by the appended claims.