Electrical fuse for integrated circuits

1. Field of the Invention

The invention relates to programmable integrated circuits, and in particular, to an electrically programmable fuse.

2. Related Art

It is often beneficial for an integrated circuit (IC) to include one or more one-time programmable elements (fuses). For example, static random access memories (SRAMs) often include metal fuses that can be blown open with a laser to activate redundancy circuitry, thereby improving production yields. As another example, programmable logic devices (PLDs), such as a field programmable gate array (FPGA), may also include fuses for activating redundant resources, storing encryption or security keys, or other purposes. A PLD may include configurable resources, such as configurable logic, a configurable interconnect structure, programmable input/output blocks, memories, transceivers, and processors. Fuses in a PLD may be used to control some or all of the configurable resources.

However, since an external programming tool (e.g., a laser) can sometimes be cumbersome, it is often desirable for fuses to be electrically programmable. An electrically programmable fuse is typically a device that undergoes a permanent change in electrical characteristics in response to a high voltage or current.

For example, commonly owned U.S. Pat. No. 6,496,416, issued Dec. 17, 2002 to Kevin T. Look, describes a metal-oxide-semiconductor (MOS) device that includes a gate heating element. When a programming voltage is applied across the gate heating element, the gate heating element raises the temperature of the device in the channel region, which results in dopant migration. This dopant migration changes the threshold voltage of the device, thereby allowing programmed and unprogrammed devices to be differentiated.

Because this type of electrically programmable fuse can be formed using standard CMOS process steps, it can be readily integrated into an IC. However, the high temperature provided by the gate heating element can sometimes cause a break in the gate heating element prior to sufficient dopant migration to program the fuse. Reducing the temperature of the gate heating element to prevent this failure can slow dopant migration significantly, which in turn increases both the time required for programming the fuse and the testing costs for the IC that includes the fuse.

Accordingly, it is desirable to provide a reliable fuse that can be efficiently programmed, and that can be manufactured using standard metal-oxide-semiconductor (MOS) transistor structures and semiconductor processes.

According to an embodiment of the invention, an electrically programmable fuse is formed by coupling a regulating element between the source and gate of a programmable transistor for lowering snapback voltage of the transistor below its nominal snapback voltage. According to an embodiment of the invention, the programmable transistor comprises a first n-type metal-oxide-semiconductor (NMOS) transistor, in which case the regulating element can comprise a resistive element such as a second NMOS transistor. The gate of the first NMOS (programmable) transistor is connected to the drain of the second (regulating element) NMOS transistor, while the source of the first transistor, the source of the second transistor, and the gate of the second transistor are all connected to ground. Implementing the regulating element using a transistor can beneficially reduce the layout area requirements of the fuse.

According to another embodiment of the invention, the programmable transistor comprises a first p-type metal-oxide-semiconductor (PMOS) transistor, in which case the regulating element can comprise a second PMOS transistor. The gate of the first PMOS transistor is connected to the drain of the second PMOS transistor, while the source of the first PMOS transistor, the source of the second PMOS transistor, and the gate of the second PMOS transistor are commonly coupled. Once again, implementing the regulating element as a transistor can beneficially reduce the layout area requirements for the fuse.

If the programmable transistor is unprogrammed, the regulating element keeps the source and gate of the programmable transistor at essentially the same voltage, thereby ensuring that a minimal current flows when a read voltage is applied across the programmable transistor (i.e., the programmable transistor remains essentially turned off). Typically, the read voltage is the nominal operating voltage of the programmable transistor (e.g., the voltage for which the programmable transistor was designed).

The programmable transistor can be “blown” (programmed) by applying an elevated programming voltage across the programmable transistor. The programming voltage must be large enough to place the programmable transistor in snapback mode. Once the programmable transistor enters snapback mode, the current flow through the programmable transistor will rapidly increase until thermal breakdown of the programmable transistor occurs. This shorts out the programmable transistor, so that any subsequent read operations will detect the programmable transistor as being turned on.

During the programming operation, the regulating element causes the programmable transistor to turn on slightly, thereby reducing the magnitude of the required programming voltage. Specifically, the resistive element allows a gate-source voltage differential to be created due to capacitive coupling between the drain and gate of the programmable transistor. This gate-source voltage differential turns on the programmable transistor slightly, which reduces the snapback voltage of the programmable transistor (i.e., the drain-source voltage required to place the programmable transistor in snapback mode).

The invention will be more fully understood in view of the following description of the exemplary embodiments and the drawings thereof.

FIGS. 1A and 1B are schematic diagrams of an electrically programmable fuse in accordance with embodiments of the invention.

FIG. 2 is a graph of current-voltage curves for grounded gate and active gate MOS transistors.

FIGS. 3A and 3B are schematic diagrams of an electrically programmable fuse in accordance with various other embodiments of the invention.

FIG. 1A shows an electrically programmable fuse 100 A in accordance with an embodiment of the invention. Fuse 100 A includes an NMOS programmable transistor 110 and a resistive element 121 . Resistive element 121 can comprise any structure or device that provides a desired resistance (e.g., a polysilicon resistor). The drain of programmable transistor 110 is connected to an input terminal 101 , while the source of programmable transistor 110 is connected to ground. (Note that as used herein, “ground” refers to the lower supply voltage of the circuit that includes the fuse of the invention.)

Meanwhile, resistive element 121 is connected between the gate and source of programmable transistor 110 , thereby coupling the gate of programmable transistor 110 to ground. As a result, when fuse 100 A is in an unprogrammed state, programmable transistor 110 is essentially off—i.e., when an input voltage V_IN at input terminal 101 is equal to a read voltage (typically the nominal operating voltage of transistor 110 ), the resulting read current will be negligible.

However, when input voltage V_IN is equal to a programming voltage that is sufficiently greater than the nominal operating voltage of programmable transistor 110 (typically the upper supply voltage VDD of the IC that includes fuse 100 A), programmable transistor 110 enters its snapback mode and its current flow increases dramatically. This increasing current flow heats up transistor 110 until thermal breakdown occurs and transistor 110 is shorted (i.e., transistor 110 remains on regardless of its gate voltage). Any subsequent read operations will result in a large read current flow through transistor 110 .

By coupling the gate of programmable transistor 110 to ground by resistive element 121 , the invention reduces the programming voltage required to place transistor 110 into its snapback mode. As is known in the art, as the gate voltage of an NMOS transistor increases, its snapback voltage (i.e., the drain-source voltage at which the transistor enters its snapback mode) decreases.

FIG. 2 shows a graph of sample current (drain-source current IDS) versus voltage (drain-source voltage VDS) response curves for a typical NMOS transistor (e.g., programmable transistor 110 shown in FIG. 1A ). Curve C( 1 ) (dotted line) represents a typical response curve that could be generated when the gate and source of the NMOS transistor are connected directly to ground (i.e., “grounded-gate” configuration). Curve C( 2 ) (solid line) represents a typical response curve that would be generated when the gate of the NMOS transistor is raised above ground (i.e., “active-gate” configuration).

As indicated by curve C( 1 ), the current flow through a grounded-gate NMOS transistor is minimal until a nominal snapback voltage V(SB 1 ) is reached. At this point, transistor 110 enters its snapback mode, and current flow increases rapidly. When the current reaches a breakdown current I(BR), thermal breakdown occurs and transistor 110 is shorted. Thus, a grounded-gate NMOS transistor remains off until nominal snapback voltage V(SB 1 ) is reached. Note that in general, the nominal snapback voltage of a MOS transistor is the drain-source voltage at which the MOS transistor enters snapback mode when the gate and source of the transistor are held at the same voltage.

In contrast, as indicated by curve C( 2 ), an active-gate NMOS transistor exhibits current flow before its active-gate snapback voltage V(SB 2 ) is reached. At this point, the active-gate NMOS transistor enters its snapback mode, and the current flow through the transistor increases rapidly until thermal breakdown occurs at breakdown current I(BR). Note that active-gate snapback voltage V(SB 2 ) is significantly lower than grounded-gate snapback voltage V(SB 1 ). As is known in the art, even if a transistor is barely turned on, its (active-gate) snapback voltage will be significantly lower than its grounded-gate snapback voltage.

Returning to FIG. 1A , because the gate of programmable transistor 110 is coupled to ground by resistive element 121 , transistor 110 behaves like an active-gate transistor during programming. Specifically, resistive element 121 allows capacitive coupling between the source and gate of transistor 110 to generate an elevated gate voltage for transistor 110 , thereby turning on transistor 110 . The larger the resistance of resistive element 121 , the larger the resulting gate voltage, and the lower the programming voltage must be to place transistor 110 in snapback mode. In this manner, resistive element 121 acts as a regulating element that lowers the snapback voltage of programmable transistor 110 relative to the nominal (grounded-gate) snapback voltage of programmable transistor 110 .

Note that according to an embodiment of the invention, an appropriate programming voltage V_PROG can be provided to input terminal 101 of fuse 100 via an optional high-voltage control transistor 130 (indicated by the dotted line). When a control signal V_CTRL is asserted (LOW) at the gate of control transistor 130 , input terminal 101 is connected to programming voltage V_PROG and programmable transistor 110 is “blown” (i.e., shorted). Note further that according to another embodiment of the invention, control transistor 130 can be part of an optional control circuit 131 (indicated by the dashed line) that selectably applies either the read voltage or programming voltage to input terminal 101 .

According to an embodiment of the invention, programmable transistor 110 can be a “high performance” transistor (i.e., a transistor with a short channel length and thin gate oxide), while control transistor 130 can be a “high voltage” transistor (i.e., a transistor having operating voltage and current limits that are at least double those of a high performance transistor), thereby ensuring that the elevated programming voltage only affects programmable transistor 110 . For example, for a 1.2V process, programmable transistor 110 may have a gate aspect ratio (width/length) of 0.5 um/0.08 um while control transistor 130 may have a gate aspect ratio of 40 um/0.24 um. A 3.3V programming voltage will then push programmable transistor 110 into snapback mode without doing the same for control transistor 130 , and the resulting high current flow that shorts out (programs) transistor 110 will not reach the breakdown current level of transistor 130 .

FIG. 1B shows an electrically programmable fuse 100 B in accordance with another embodiment of the invention. Fuse 100 B is substantially similar to fuse 100 A shown in FIG. 1A , with resistive element 121 of fuse 100 A being implemented as a grounded-gate NMOS transistor 122 in fuse 100 B. Grounded-gate transistor 122 provides a very high resistance path (e.g., in the mega-Ohm range) between the gate of programmable transistor 110 and ground. Typically, grounded gate transistor 122 can be implemented much more easily and efficiently than a conventional semiconductor resistor, and therefore can reduce the cost and space requirements of fuse 100 B.

Fuse 100 B operates in a manner substantially similar to that described with respect to fuse 100 A. Transistor 122 weakly grounds the gate of programmable transistor 110 to ensure that a low read current is maintained while transistor 110 is unprogrammed. The application of an elevated programming voltage to input terminal 101 results in capacitive coupling between the drain and gate of transistor 110 . Due to the high resistance of grounded-gate transistor 122 , this capacitive coupling results in a gate voltage that is greater than ground, which begins to turn on transistor 110 . Consequently, the snap-back voltage of transistor 110 is reduced, thereby ensuring that transistor 110 is placed in snapback mode by the programming voltage and eventually “blows”. Once programmed in this manner, transistor 110 provides a high read current at input terminal 101 .

FIG. 3A shows an electrically programmable fuse 300 A in accordance with another embodiment of the invention. Fuse 300 A includes a PMOS programmable transistor 310 and a resistive element 321 . Resistive element 321 can comprise any structure or device that provides a desired resistance (e.g., a polysilicon resistor). The source of programmable transistor 310 is connected to a high voltage input terminal 301 , while the drain of transistor 310 is coupled to ground by an NMOS control transistor 330 . Meanwhile, resistive element 321 is connected between the gate and source of programmable transistor 310 , thereby coupling the gate of programmable transistor 310 to input terminal 301 .

To perform a read operation, a control signal V_CTRL is asserted (HIGH) at the gate of control transistor 330 to complete the fuse circuit, and an input voltage V_IN provided at input terminal 301 is set equal to the nominal operating voltage of programmable transistor 310 (e.g., the upper supply voltage VDD). If transistor 310 is in an unprogrammed state, the read voltage provided to its gate by resistive element 321 ensures that transistor 310 is always off—i.e., the read current is negligible.

To program fuse 300 A, control transistor 330 is turned on, and input voltage V_IN at input terminal 301 is set equal to an elevated programming voltage, which causes programmable transistor 310 to enter its snapback mode. Note that according to an embodiment of the invention, the read and programming voltages can be selectably provided to input terminal 301 by an optional control circuit 331 . Once transistor 310 enters its snapback mode, the current flow through transistor 310 increases rapidly. When this current flow reaches the thermal breakdown current of transistor 310 , transistor 310 is permanently shorted. Any subsequent read operations will result in a large read current flow through transistor 310 .

Note that for reasons similar to those described with respect to fuse 100 A in FIG. 1A , coupling the gate of transistor 310 to input terminal 301 via resistive element 321 allows capacitive coupling between the drain and gate of transistor 310 to turn on transistor 310 during programming. Specifically, the low voltage at the drain of transistor 310 is capacitively coupled to the gate of transistor 310 , which drops the gate voltage of transistor 310 below the programming voltage provided at input terminal 301 . As a result, transistor 310 is turned on, which in turn reduces its snapback voltage (as described with respect to FIG. 2 ). Once again, resistive element 321 acts as a regulating element that lowers the snapback voltage of programmable transistor 310 below the nominal snapback voltage of transistor 310 , thereby minimizing the required programming voltage for fuse 300 A.

According to an embodiment of the invention, programmable transistor 310 can be a high performance transistor, while control transistor 330 can be a high voltage transistor, thereby ensuring that the elevated programming voltage only affects programmable transistor 310 . For example, for a 1.2V process, programmable transistor 310 may have a gate aspect ratio of 0.5 um/0.08 um while control transistor 130 may have a gate aspect ratio of 20 um/0.24 um. A 3.3V programming voltage then places programmable transistor 310 in snapback mode without doing the same for control transistor 330 , and the resulting high current flow that shorts out transistor 310 will not reach the breakdown current level of transistor 330 .

FIG. 3B shows an electrically programmable fuse 300 B in accordance with another embodiment of the invention. Fuse 300 B is substantially similar to fuse 300 A shown in FIG. 3A , with resistive element 321 of fuse 100 A being implemented as a gate-source coupled PMOS transistor 322 in fuse 300 B. Gate-source coupled transistor 322 provides a very high resistance path (e.g., in the mega-Ohm range) between the gate of programmable transistor 310 and the high voltage input terminal 301 . Typically, gate-source coupled transistor 322 can be implemented much more easily and efficiently than a conventional semiconductor resistor, and therefore can reduce the cost and space requirements of fuse 300 B.

Fuse 300 B operates in a manner substantially similar to that described with respect to fuse 300 A. Transistor 322 weakly pulls the gate of programmable transistor 310 to a high voltage to ensure that a low read current is maintained while transistor 310 is unprogrammed. The application of an elevated programming voltage to input terminal 301 results in capacitive coupling between the drain and gate of transistor 310 . Due to the high resistance of grounded-gate transistor 322 , this capacitive coupling results in a gate voltage that is less than the programming voltage, which begins to turn on transistor 310 . Consequently, the snap-back voltage of transistor 310 is reduced, thereby ensuring that transistor 310 is placed into snapback mode by the programming voltage and eventually “blows”. Once programmed in this manner, transistor 310 will always provide a high read current at input terminal 301 .

The various embodiments of the structures and methods of this invention that are described above are illustrative only of the principles of this invention and are not intended to limit the scope of the invention to the particular embodiments described. Thus, the invention is limited only by the following claims and their equivalents.