NMOS Driver Circuit for CMOS Circuitry

This invention relates to semiconductive integrated circuits and, more particularly, relates to metal-oxide-semiconductor (MOS) transistor circuitry.

In some applications, an integrated circuit chip that uses only n-channel MOS (NMOS) transistors must communicate with another integrated circuit chip that uses a complementary MOS (CMOS) transistor pair (a PMOS and an NMOS transistor) as its input. Such CMOS circuits typically require a significant input voltage swing, e.g., such as between 0.3 Vdd and 0.7 Vdd, where Vdd is a positive voltage coupled to the source of the PMOS transistor of the input CMOS pair and ground potential is coupled to the source of the NMOS transistor of the CMOS pair whose drain is coupled to the drain of the PMOS transistor, in order to provide a reliable recognizable input change.

One solution to this problem is to include a p-channel enhancement transistor on the same chip as the n-channel transistors and to use it in series with an n-channel transistor. This solution, in effect, results in the chip becoming a CMOS type integrated circuit, which adds considerable complexity to the fabrication process and is undesirable in some applications.

One prior art driver circuit often used as an output driver on an n-channel transistor chip is shown as driver circuit 10 in FIG. 1. Driver circuit 10 essentially comprises a first n-channel enhancement mode MOS transistor 12 and a second n-channel enhancement mode MOS transistor 14 with the source of transistor 12 being connected to the drain of transistor 14 and to an output terminal 18, with the drain of transistor 12 coupled to a positive voltage source Vdd, and with the source of transistor 14 coupled to ground potential. The gate of transistor 12 is coupled to an input terminal 13 and to an input of an inverter 16 which has an output that is connected to the gate of transistor 14 and to a terminal 15. The channels of NMOS transistors 12 and 14 are shown with two "+'s" in each to indicate that a conventional p-type implant was made into the channel of each to raise the threshold voltages and help prevent unwanted inversion of the channels. This implant is important to overcome the natural tendency of the surface of the channel of an NMOS transistor, formed in a typical lightly doped p-type substrate, otherwise to behave as n-type, in which case the transistor can act as a depletion mode transistor with appreciable surface current between the heavily doped source and drain of the transistor, even without any applied gate voltage. The implant is used to set the potential of the surface and thereby to insure that the transistor operate in the desired enhancement mode. Typically, the implant is used to increase the surface doping to about 1.times.10.sup.16 impurities/cm.sup.3 while the bulk of the p-type substrate (not shown) retains its original doping, typically about 1.times.10.sup.15 impurities/cm.sup.3. Inverter 16 is used to insure that opposite logic signals are applied to the gates of transistors 12 and 14.

With a "1" input signal (e.g., a voltage level at or near the voltage level of Vdd) applied to input terminal 13, transistor 12 is biased on (enabled) and transistor 14 is biased off (disabled). These conditions cause output terminal 18 to assume a voltage level of Vdd less the Vth (threshold of transistor 12). Typically, Vth of transistor 12 is about +0.7 volts (plus or minus 0.2 volts) with a zero back gate bias for the example mentioned where the impurity concentration of the p-type substrate (not shown in this figure) on which transistor 12 is fabricated is 1.times.10.sup.15 impurities/cm.sup.3 and the p-type implant into the channel region of transistor 12 result in increasing the impurity concentration in the channel region to about 1.times.10.sup.16 impurities/cm.sup.3. In some applications, including applications in which a CMOS inverter circuit is to be the load connected to output terminal 18, the output voltage representing a "1" is not high enough to reliably switch the load circuit. In such case, a buffer circuit or voltage multiplier circuit must be inserted between the driver circuit and the load for reliable operation.

The above described prior art two serially connected n-channel MOS transistor driver circuit can be modified to include bootstrapping circuitry so as to increase the gate potential of the first transistor.

This modified circuit can generate higher output "1" levels but requires additional components and complexity.

It is desirable to have a circuit driver essentially comprised of a serially connected pair of n-channel enhancement mode MOS transistors in which the output voltages are completely compatible with the input voltages needed reliably to control conventional CMOS circuitry.

The invention basically is a serially coupled pair of enhancement mode n-channel MOS transistors (denoted as an NMOS pair, or as first and second NMOS transistors) that provides an output voltage swing wide enough for it to be effectively coupled directly to the input of a CMOS transistor pair. The first NMOS transistor is conventionally denoted as the pullup transistor or device of the pair and the second NMOS transistor is conventionally denoted as the pulldown transistor or device of the pair. The second NMOS transistor is a conventional enhancement mode NMOS transistor which has had acceptor ions implanted into the surface of its channel between the source and the drain. However, the first NMOS transistor is a "natural" NMOS transistor which is defined as one free of the customary acceptor ion implant into the surface of its channel between the source and the drain.

We have found that the presence of a conventional enhancement mode NMOS transistor in the pulldown position is found to buffer sufficiently any adverse effect of the natural NMOS transistor in the pullup role so that reliable operation becomes feasible.

The invention will be better understood from the following more detailed description taken in conjunction with the accompanying drawing and appended claims.

FIG. 1 is a circuit schematic of a prior art NMOS driver circuit;

FIG. 2 is a circuit schematic of an NMOS driver circuit in accordance with the present invention coupled to a conventional CMOS inverter circuit; and

FIGS. 3A, 3B, 3C, 3D, 3E, 3F, 3G and 3H are cross sections of a portion of a silicon wafer at various stages in its processing to form therein a conventional (normal) enhancement mode NMOS transistor, a natural enhancement mode NMOS transistor and a depletion mode NMOS transistor.

It is to be noted that the drawings are not to scale.

Referring now to FIG. 2, there is shown within the larger dashed line rectangle a driver circuit 20 in accordance with the invention. An output terminal 50 of driver circuit 20 is coupled to gates of a conventional complementary MOS (CMOS) inverter circuit 56 (shown within the smaller dashed line rectangle) which comprises enhancement mode p-channel MOS transistor 38 and an enhancement mode n-channel MOS transistor 40. The drains of transistors 38 and 40 are coupled together to an output terminal 58 of inverter 56. Driver circuit 20 comprises a "natural" enhancement mode n-channel transistor 22, conventional enhancement mode n-channel MOS transistors 24, 26, 28, 30 and 32, and conventional depletion mode n-channel MOS transistors 34 and 36. The source of transistor 22 is coupled to the drain of transistor 24 and to output terminal 50 of driver circuit 20. The combination of transistors 22 and 24 function essentially like transistors 12 and 14 of FIG. 1 except that the high "1" output level generated at output terminal 50 is higher in voltage level than is generated on output terminal 18 of driver circuit 10 of FIG. 1. This, in some applications, allows driver circuit 20 to be directly connected to the CMOS inverter circuit 56 and thus eliminates the need for an interface or voltage amplifier circuit between the two.

Transistor 22 is denoted as a pullup transistor or pullup device since, when it is heavily biased on (enabled), the voltage of output terminal 50 is set (pulled up) to a level that is close to the level of a voltage source Vdd, which is coupled to the drains of transistors 22, 34, 36 and 38 and to a terminal 42. Transistor 24 is denoted as a pulldown transistor or pulldown device since, when it is heavily biased on, the voltage on the output terminal 50 is set (pulled down) to a level that is at or close to ground potential, which is coupled to the sources of transistors 24, 26, 28, 30, 32 and 40 and to a terminal 44. N-channel transistors 24, 26, 28, 30, 32 and 40 each show a double plus (++) in the channel region thereof, which indicates that they are conventional enhancement mode MOS transistors in which a p-type impurity was implanted in the surface of the channel thereof to insure against inadvertent inversion of the channel and to increase the threshold voltage of each transistor typically to 0.7 volts (plus or minus 0.2 volts) at zero back gate bias. Transistor 22 is a "natural" enhancement mode MOS transistor in which the surface of its channel did not have its conductivity changed but is essentially just that of the conductivity of the substrate (not shown in this figure) in which it is formed. The threshold voltage of transistor 22 is typically about 0 volts (plus or minus about 0.2 volts) at zero gate back bias. Transistors 34 and 36 are conventional n-channel depletion mode MOS transistors, which is indicated by the heavy vertical line through the channel of each that joins the drain and source of each. The channel area of each of the transistors 34 and 36 is an n-type region (not shown in this figure) which joins the drain and source regions (not shown in this figure).

The gates of transistors 22 and 28 are coupled to the gate and source of transistor 34, to the drains of transistors 26 and 32 and to a terminal 52. The gate of transistor 24 is coupled to the gate and source of transistor 36, to the drains of transistors 28 and 30 and to a terminal 54. The gate of transistor 26 is coupled to a first input control terminal 46 of driver circuit 20. Transistors 26 and 34 act as a first inverter circuit, and transistors 28 and 36 act as a second inverter circuit. The gates of transistors 30 and 32 are coupled to a second control input terminal 48 of driver circuit 20. Transistors 30 and 32 allow driver circuit 20 to be operated in a tri-state mode in which output terminal 50 is essentially isolated from all other portions of driver circuit 20 and essentially floats in potential or is set in potential by some other circuit (not shown) coupled to terminal 50.

Driver circuit 20 is typically operated as follows: A "1" input signal (a signal having a voltage level at or close to the level of Vdd) applied to terminal 48 biases on both transistors 30 and 32 which then cause terminals 52 and 54 to be set to a voltage level at or near ground potential independent of the voltage of terminal 46. These conditions bias off transistors 22 and 24 and allow terminal 50 to float in potential. This is the tri-state condition. A "0" (a signal having a voltage level at or near ground potential) applied to terminal 48 biases off transistors 30 and 32 and, therefore, allows the potentials of terminals 52 and 54 to be controlled by a voltage signal applied to input terminal 46. With a "0" applied to input terminal 48, a "1" or a "0" can be applied to terminal 46. A "1" applied to terminal 46 biases on transistor 26 which causes the potential of terminal 52 to be pulled down to a voltage near ground potential. Transistor 34 is typically only one-tenth the size of transistor 26 and, thus, even though both are on and conducting, the voltage of terminal 52 is at about +0.2 volts. This results in transistor 22 being only weakly biased on. The +0.2 volts on terminal 52 biases off transistor 28, which results in the potential of terminal 54 assuming the level of Vdd since transistor 36 is always on. This heavily biases on transistor 24 and, thus, the voltage of output terminal 50 is set (pulled down) to a level at or near ground potential.

If the signal applied to terminal 46 is now switched to "0", then transistor 26 is biased off and terminal 52 is set (pulled up) to the voltage level of Vdd. This heavily biases on transistor 22 and biases on transistor 28. Transistor 36 is designed to be about one-tenth of the geometry of transistor 28 and, accordingly, terminal 54 is set to a voltage near ground potential (about +0.2 volts), which biases off transistor 24. Accordingly, the voltage of output terminal 50 is set (pulled up) to a level of Vdd less the threshold voltage (Vth) of transistor 22 (which is about 0 volts) less the body voltage loss associated with transistor 22. The output "1" level of drive circuit 20 is theoretically about 0.7 volts higher than the output "1". voltage level generated by driver circuit 10 of FIG. 1.

Driver circuit 20 has been fabricated in integrated circuit form on a single silicon chip with transistors 22 and 24 having channel width-to-length ratios of 2030/7 and 1330/7, respectively. With Vdd=4.5 and +5.5. volts, the "1" voltage measured at output terminal 50 was +3.988 and +4.645 volts, respectively. CMOS inverter circuit 56 was fabricated on a separate integrated circuit chip and was appropriately connected to Vdd and ground potential.

Driver circuit 10 of FIG. 1 has been fabricated in integrated circuit form on a single silicon chip with transistors 12 and 14 having channel width-to-length ratios of 1740/6 and 1140/6, respectively. With Vdd=+4.5 and +5.5 volts, the "1" voltage measured at output terminal 18 was +3.39 and +4.15 volts, respectively. It is thus apparent that, with these embodiments, the output "1" voltage of driver circuit 20 of FIG. 2 is typically about +0.5 to +0.6 of a volt higher than that of the output "1" voltage of circuit driver 10 of FIG. 1. This greater output voltage level enables circuit driver 20 to directly drive conventional CMOS circuitry.

The preparation of a silicon chip to include a conventional enhancement mode n-channel MOS transistor, a natural enhancement mode n-channel transistor and depletion mode n-channel MOS transistors need involve only mask changes as compared to the conventional processing of a silicon substrate that includes only conventional enhancement mode n-channel MOS transistors and depletion mode n-channel MOS transistors. In particular, the only change that needs be made is add a mask used before the threshold implant step to block the implant into those surface regions of the substrate that will form the desired natural NMOS transistor(s) and the depletion NMOS transistors. The natural n-channel enhancement mode MOS transistor, the conventional n-channel enhancement mode MOS transistors, and the conventional depletion mode n-channel MOS transistors can all be fabricated on a single silicon substrate as is shown below.

Typically, there is processed a relatively large wafer that is eventually diced into a number of chips, each of which includes a large number of transistors appropriately interconnected into an integrated circuit. For the sake of convenience, only so much of the wafer as is needed to house one natural and one conventional enhancement mode n-channel MOS transistor and one depletion mode n-channel MOS transistor is shown in the figures.

Referring to FIG. 3A, there is shown a portion of a monocrystalline silicon wafer 98 that serves as a substrate 100 in the processing which results in the fabrication of both enhancement mode conventional and natural n-channel MOS transistors and conventional depletion mode n-channel MOS transistors. Typically, the substrate 100 is cut so that its top surface 110 corresponds to a <100> crystal plane and is of relatively lightly doped p-type, designated by p-type in the drawing. A typical doping is between 1.times.10.sup.14 and 1.times.10.sup.15 impurities/cm.sup.3.

The illustrative processing begins by forming on the top surface 110 a plurality of thick field oxide regions, typically at least 5000 Angstroms thick, that will serve to isolate laterally the active surface regions of individual transistors. To this end, there is first formed over the surface 110 a relatively thin (e.g., 700 Angstroms) layer 112 of silicon oxide (typically silicon dioxide). It is to be noted that the layers variously denoted herein as of silicon oxide or oxide are predominantly of silicon dioxide, but it is customary in the art to describe them as of silicon oxide or simply oxide.

The formation of the oxide layer 112 is followed by the deposition of a thicker layer 114 (e.g., typically about 1000 Angstroms thick) of silicon nitride that is patterned to expose the underlying silicon oxide layer 110 only where the thick field oxide regions are to be formed. The resultant at this stage is shown in FIG. 3A.

However, before formation of the field oxide regions, it is usual to implant the substrate 100 with acceptor ions to increase the acceptor doping in regions of the substrate 100 that will underlie the field oxide regions. This serves advantageously to reduce any tendency for such regions to experience undesired surface inversion during operation. A typical field implant is a boron implant at a dosage of about 6.2.times.10.sup.12 impurities-cm.sup.2 at an accelerating voltage of about 76 KeV. In conventional fashion, in FIG. 3B the implanted regions are indicated by plus signs and the boron ions by the arrows 115.

After the boron has been implanted, the substrate 100 is heated in an oxidizing ambient in the usual fashion for a time adequate to form the thick field oxide regions, typically about 10,500 Angstroms thick. Then the silicon nitride mask 114 and the thin oxide layer 112 are removed to leave the resultant seen in FIG. 3C in which the surface 110 of substrate 100 is divided by field oxide regions 116.

At this stage, the substrate 100 is ready for the gate oxidation. To this end, the substrate 100 is again heated in an oxidizing ambient for a time to form over the exposed silicon a thin gate oxide layer 119, typically about 650 Angstroms thick, seen in FIG. 3D.

After formation of this gate oxide layer 119, the substrate 100 is ready to receive the accepter ion implant customarily used to set the potential of the silicon surface 110 appropriate for the enhancement mode desired. This implant, termed the threshold implant, is a shallow implant and serves to neutralize the tendency of a silicon surface 110 to behave as n-type, as discussed earlier.

To prepare a chip of the kind that is a feature of the present invention, it is necessary to block the introduction of such acceptor ions into the channel underlying the gate of the natural n-channel MOS transistor, so that it will have the low threshold voltage desired for the practice of the invention. To this end, as shown in FIG. 3D, a photoresist mask 118 is provided over a top surface 119a of the gate oxide layer 119 to block selectively passage of the implanted ions denoted by the arrows 120. A typical dosage is about 3.3.times.10.sup.11 impurities-cm.sup.2 at an accelerating voltage of about 35 KeV. The boron implanted regions are indicated again by plus signs.

After the removal of the photoresist mask 118, further processing is conventional.

First, it is usual to provide a donor implantation into the active surface region of the depletion mode NMOS transistor. To this end, as shown in FIG. 3E, a photoresist mask 123 is provided over the surface of the substrate that exposes only such active surface area. Then the substrate 100 is irradiated with phosphorus ions, typically 5.2.times.10.sup.11 impurities-cm.sup.-2 at an accelerating voltage of about 120 KeV. The implanted donor ions are indicated by minus signs. After the later drive-in, such implanted ions convert a surface layer of the substrate 100 to lightly doped n-type conductivity for use by the depletion NMOS transistor.

The next step is the provision of gate electrodes 124 to the various transistors. Generally, to this end, a polycrystalline silicon layer is first deposited over the surface of the wafer, typically several thousand Angstroms thick. This layer is then patterned to leave portions that are to serve as the gate electrodes 124, as shown in FIG. 3F.

This is then typically followed by an implantation with donor ions to form the desired heavily doped n-type source and drain zones of each of the transistors. An illustrative dosage is about 4 x 15 arsenic ions-cm.sup.2 at an accelerating voltage of about 100 KeV. This implantation also serves to increase the conductivity of the polysilicon electrodes 124 desirably.

In FIG. 3G, the donor implanted regions are denoted by minus signs and the arsenic ion beam by arrows 128.

This generally is followed by a heating step to anneal any damage by the implant and to drive in the implanted arsenic to form the various heavily doped source and drain regions desired. FIG. 3H shows the results. Heavily doped n-type regions (n+) 130, 132, 34, 136, 138 and 140 are formed in the substrate 100. Of these, regions 130 and 132 can serve interchangeably as the source and drain of the natural enhancement mode NMOS transistor 22 shown in FIG. 2; regions 134 and 136 can serve interchangeable as the source and drain of the conventional enhancement mode NMOS transistor 24 shown in FIG. 2; and regions 138 and 140 are interchangeable as the source and drain of the depletion mode NMOS transistor of FIG. 2. Additionally, the implanted phosphorus ions have formed the lightly-doped n-type channel region 142 of the depletion mode NMOS transistors.

Finally, there remains the provision of suitable low resistance contacts to the various sources, drains and gates. Various techniques are known for this purpose and are independent of the present invention.

Typically, to provide the desired contacts, a metal layer, illustratively an aluminum-copper-silicon alloy, is deposited over the substrate and then patterned to leave portions that serve as the source, drain and gate electrode contacts 150 of the various transistors seen in FIG. 3H.

It is to be understood that the specific embodiment described is merely illustrative of the spirit and scope of the invention. Various modifications may be devised without departing from such spirit and scope. For example, the circuitry coupled to terminals 52 and 54 to control the biasing of transistors 22 and 24 can be easily modified in a variety of ways.