Integrated Circuit With MOS Capacitor for Improved ESD Protection

1. Field of the Invention

The present invention relates to an integrated circuit having improved electrostatic discharge (ESD) protection.

2. Description of the Prior Art

In the design of integrated circuits, the problem of ESD protection has received considerable attention, especially as dimensions shrink below the 1.5 micron level. In addition, the lightly doped drain (LDD) type transistor structure is suspected of reducing the voltage threshold at which ESD failure occurs. An excessively high ESD voltage conducted from a package terminal to the integrated circuit bondpad can easily damage input or output circuitry, unless protection techniques are adopted. Some approaches to improving ESD performance rely on the use of voltage-clamping diodes or transistors. For example, referring to FIG. 1, an input/output bondpad 100 is connected to a package terminal. An output stage (transistors 101, 102) and/or an input stage (transistors 108, 109) are coupled to the bonded. Typically, protection devices 103 and 104 limit the voltage on the bondpad. The input resistor 105 and additional diodes 106, 107 are also optionally provided in some cases, to further protect the input transistors. When the n-channel output transistor 102 is present, experience indicates that it often breaks down at a lower voltage than the protection circuitry, due to its inherent bipolar action through the source/substrate/drain path.

The voltage-clamping techniques are especially desirable to protect the input transistors, where breakdown of the thin gate oxide is a very significant source of failure due to ESD events. Various voltage clamping device structures have been adopted to improve ESD protection; see, for example, U.S. Pat. Nos. 4,806,999 and 4,821,089 co-assigned herewith.

Another approach to ESD protection is to supply a capacitance across the power supply voltage conductors (V.sub.DD and V.sub.SS); see "The Dynamics of Electrostatic Discharge Prior to Bipolar Action Related Snap-Back"; G. Krieger, 1989 EOS/ESD Symposium Proceedings. This is said to provide increased protection by slowing down the increase of the voltage on the input/output bondpad during an ESD event. That is said to provide additional time for the ESD protection circuitry to react, thereby protecting the thin gate oxides that are very susceptible to ESD damage. However, the capacitance relied on for this purpose is the n-well to PG,3 p-substrate junction capacitance that is formed by the n-well in which p-channel transistors are formed. It is also noted in the Krieger article that the capacitance is charged via a forward biased p+ drain to n-well diode, being diode D.sub.p shown in FIG. 2 of Krieger.

I have invented an improved integrated circuit ESD protection technique. A MOS capacitor connected across the power supply voltage conductors provides for improved ESD protection. The capacitor operates in the accumulation mode when positive ESD voltages are present.

FIG. 1 shows an illustrative embodiment of input/output circuitry protected by the present technique.

FIG. 2 shows an illustrative embodiment of a MOS capacitor that implements the present invention for n-type semiconductor substrates.

FIG. 3 shows an illustrative embodiment of a MOS capacitor that implements the present invention for p-type semiconductor substrates.

The following detailed description relates to an integrated circuit having improved ESD protection. Referring to FIG. 1, an illustrative embodiment of circuitry that implements the present technique is shown, with various other protection circuits being possible. The junction capacitance noted above in the Krieger article is indicated as capacitor 110, and is an inherent part of CMOS integrated circuits formed in tubs. In the present invention, an additional capacitor 111 is included between the positive and negative power supply conductors. I have determined that when constructed according to the technique described below, this additional capacitor is beneficial in increasing the ESD failure threshold voltage, thereby reducing the likelihood of failure due to an ESD event. According to the inventive technique, the additional capacitor 111 is a MOS type formed from an overlying conductor, a dielectric, and an underlying doped semiconductor region.

This type of capacitor operates in the accumulation mode during a positive-voltage ESD event, and is known per se in the integrated circuit art for various small-signal applications. However, I have surprisingly discovered that the use of such a capacitor can significantly increase ESD failure threshold voltages. This capacitor may be formed with the same dielectric as the gates of the MOS transistors formed on the integrated circuit. Surprisingly, this relatively thin capacitor dielectric does not prematurely break down during an ESD event, as might be expected. That is, the capacitor increases the ESD failure threshold, rather than decrease it, as might be expected from the use of an additional device with a thin oxide gate dielectric that is susceptible to ESD breakdown when used as the gate oxide for input transistors.

Referring to FIG. 2, an illustrative embodiment of the capacitor 111 is shown in the case wherein the integrated circuit is formed in an n-type substrate region 200. A p-tub region 201 is formed in the n-region 200. This may be accomplished using the so-called twin tub technique, as shown for example in U.S. Pat. No. 4,435,896, co-assigned herewith. In that case, tubs 202, 203 of the opposite conductivity type are formed adjacent to tub 201. However, other tub-forming techniques, including single-tub (also referred to as single-well) processes are possible for use with the present invention. The capacitor according to this embodiment of the inventive technique includes the p-tub region 201 as one plate (the "lower" plate) of the capacitor. This plate is connected to the positive power supply conductor V.sub.DD by at least one tub-tie region, with two p+ doped tub-tie regions 204 and 205 being shown in FIG. 2. The capacitor dielectric is formed by dielectric region 206, being typically silicon dioxide that may optionally include other dielectric layers. This capacitor dielectric is typically a gate-level dielectric that is formed by the same process as forms the gate dielectrics of the MOS transistors of the integrated circuit.

The other plate of the capacitor (the "upper" plate) is formed by conductor 207, being typically formed from the same doped polysilicon layer that forms the gate electrodes of the MOS transistors. Note that a silicide layer 212 (e.g., titanium silicide, cobalt silicide, tungsten silicide, molybdenum silicide, tantalum silicide, etc.) may optionally be included as part of this conductor, according to techniques known in the art. However, the use of other conductor levels is possible, including a second (or third) polysilicon conductor level, for example. In that case, the dielectric layer 206 is formed at a different time than the gate dielectric layer. This upper plate is connected to the negative power supply conductor, V.sub.SS, in this embodiment. The edges of the upper capacitor plate are optionally isolated by field oxide regions 208 and 209 from the adjacent regions tub-tie regions 204 and 205, respectively. This isolation is desirable in the case wherein a silicide layer is formed by the self-aligned silicide ("salicide") process as part of the upper conductor and tub-tie contact. This reduces the likelihood of shorts between this plate and the tub-ties, which would short out the capacitor. The surface region of tub 201 may be spaced from that of the adjacent tubs by field oxide regions 210 and 211 if desired.

Note that in operation, positive charges accumulate immediately under the capacitor dielectric 206, due to the polarity of the voltages applied to the capacitor plates, and the fact that the lower plate is formed in a p-type region. During a positive ESD event, still more positive charges are drawn to this region, and hence the capacitor is referred to as operating in the accumulation mode. This contrasts to the inversion-type MOS capacitor, wherein in operation the surface immediately under the gate oxide inverts to the opposite conductivity as the semiconductor region in which it is formed.

A protective circuit according to FIG. 1, was constructed, including a capacitor 111 constructed according to the above-embodiment, in an integrated circuit formed in 0.9 micron technology. The total capacitor area was 2.23.times.10.sup.6 square microns, which comprised a multiplicity of capacitor elements in parallel. The dielectric layer was a gate oxide having a thickness of 15 nanometers (150 angstroms), producing a capacitance of 5.6 nanofarads. The p-tub 201 had a surface doping level of 4 to 5.times.10.sup.16 /cm.sup.3, formed in a substrate 200 having a doping of 5.times.10.sup.15 /cm.sup.3. The conductor 207 included doped polysilicon layer approximately 420 nanometers (4200 angstroms) thick, and including a layer of titanium silicide formed thereon by the salicide process. Accordingly, the tub-tie regions 204 and 205 also had a layer of titanium silicide formed thereon. An ESD test was conducted using the "human body model" that is well-known in the art. It was found that the ESD failure threshold was 2000 volts. By effectively removing the capacitor 111 from the circuit (by disconnecting the upper plate from V.sub.SS and connecting it to V.sub.DD), the ESD failure threshold was found to be 1000 volts. Hence, a significant improvement in ESD failure threshold was obtained by the inventive technique.

It is presently believed that the diode 103 aids in the operation of the inventive technique, by providing a conduction path from the bondpad to the capacitor 111 during a positive ESD event, in a comparable manner as diode D.sub.p of Krieger, noted above. However, in other designs, a suitable conduction path may be provided by other circuit elements, in lieu of, or in addition to, the diode 103. For example, a p-channel output transistor (in lieu of the n-channel transistor 101 shown) may provide this conduction path. This is accomplished via the drain-to-tub junction, with the n-tub being connected to V.sub.DD, and hence the lower capacitor plate, and the p+ drain being connected to the bondpad. In some cases, the drain is connected to the bondpad through a resistor, typically deposited polysilicon or silicide, as shown in U.S. Pat. No. 4,990,802 co-assigned herewith.

Although the above-embodiment has shown the accumulation-type capacitor formed in a p-tub, it may alternatively be formed in an n-tub, as shown in FIG. 3. In that case, the p-type substrate in which the n-tub is formed is connected to the negative power supply conductor, V.sub.SS. Also, the upper capacitor plate is then connected to the positive power supply conductor, V.sub.DD. When operating voltages are present, and when a high positive ESD voltage is present on the bondpad, the negative charges accumulate immediately under the capacitor dielectric region. In either embodiment, I recommend that a protective ESD capacitor according to the inventive technique have a capacitance of at least 100 picofarads. For this purpose, the capacitor dielectric is desirably relatively thin, and I recommend a thickness of less than 25 nanometers (250 angstroms).