Static, low-voltage fuse-based cell with high-voltage programming

1. Field

This disclosure relates to fuse-based cells, more particularly to those fuse-based cells programmable with high voltages.

2. Background

Fuse-based cells have many applications in semiconductor device manufacture. The cells can permanently store data, for example, where a burned fuse represents a “1” and an unburned fuse represents a “0.” Similarly, fuses can form permanent connections on integrated circuits after they are manufactured. A sensing circuit typically determines the state of the fuse. These sensing circuits typically detect a change in the resistance of the fuse device from a low value for unburned fuses to a high value for burned fuses.

However, advances in semiconductor manufacturing technology have decreased the resistance of fuses after they have been burned, referred to as post-burn resistance. Similarly, the operating voltages of most semiconductor devices have been reduced in pursuit of lower power consumption. While higher burning voltages move the post-burn resistance to a high level, the higher voltages may damage semiconductor devices designed to operate at lower voltages.

Therefore, it would be useful to have the ability to use higher burning voltages for fuse-based devices, while providing protection for devices that operate at lower voltages.

The invention may be best understood by reading the disclosure with reference to the drawings, wherein:

FIG. 1 shows an embodiment of a programming circuit for a fuse-based device, in accordance with the prior art.

FIG. 2 shows a complimentary embodiment of a programming circuit for a fuse-based device, in accordance with the prior art.

FIG. 3 shows an embodiment of a high-voltage programming circuit for a fuse-based device, in accordance with the invention.

FIG. 4 shows an embodiment of a differential high-voltage programming circuit for a fuse-based device, in accordance with the invention.

FIG. 1 shows an embodiment of a programming circuit for a fuse-based device, in accordance with the prior art as discussed in U.S. Pat. No. 5,959,445 issued Sep. 28, 1999. The example shown is a ‘differential’ embodiment, where the difference is detected based upon the programming of one of the two fuses. In FIG. 1 , a programming transistor PT 1 is a p-channel transistor. The source is coupled to Vcc, which is the voltage used in this example to program either the set fuse 10 or the clear fuse 12 . The programming device and circuitry for programming the clear fuse 12 is not shown in this drawing.

Logic circuitry, such as the NAND gate shown, is couple to the gate of the transistor PT 1 to control the programming circuit. If only one input is used, the logic circuitry is not necessary, as the input line could be coupled directly to the gate of the transistor. Any input received, either through the logic circuit or directly, that cause the signal at the gate of the transistor to be low turns the transistor PT 1 on. This will cause current to flow through the fuse 10 , coupled to the output of PT 1 . This programs the set fuse. A similar operation may be implemented to program the clear fuse 12 instead of the set fuse 10 .

Once programmed, the state of the fuse must be determined. The sensing circuit includes a sensing device S 1 , with its source couple to the terminal of the set fuse 10 . The drain of S 1 is coupled to the drain of load device L 1 and the drain of clamping device C 1 . In addition, the output node 18 , which is used to determine the final result of the fuse state, is coupled to the drain of S 1 .

The two sensing devices S 1 and S 2 have their gates coupled together. Sensing device S 2 also has its drain coupled to a reference voltage output VREF. This establishes a current mirror between the first and second branches of this circuit. The sensing device S 1 , clamping device C 1 , load device L 1 , programming device PT 1 and set fuse 10 are all included in the first, or set, branch of the circuit. The sensing device S 2 , clamping device C 2 , load device L 2 , a programming device not shown and the clear fuse 12 are all included in the second, or clear, branch of the circuit.

When the sense enable signal 16 is high, the two load devices L 1 and L 2 are off, and there is no current flow between the two branches. The clamping devices C 1 and C 2 turn on, and cause the output voltage on the output node 18 to clamp to Vss. This brings the output to a known voltage when the circuit is off. When sense enable 16 is low, L 1 and L 2 turn on and a current path between the supply voltage Vcc and S 2 is created. The gate of S 2 is pulled up, causing it to conduct. This affects the voltage at VREF.

For purposes of this description, a logical ‘1’ will be assumed if the set fuse is burned and a logical ‘0’ will be assumed if the clear fuse is burned. The voltage level at output node 18 determines the logic state. If the output voltage is within one range, it is deemed to be a 1, if within another range, it is deemed to be a 0. The ranges are determined by the nature of the devices used.

If the set fuse is burned, it will have higher resistance than the clear fuse. There is essentially no current flow in the set branch of the circuit, including none through the sensing device S 1 . Load device L 1 then pulls the output node up to a high state, approximately equal to Vcc. If the post-burn resistance is close to that of the unburned clear fuse, it is possible that some current may flow through S 1 . However, the gates of S 1 and S 2 are coupled together and there is a higher potential on the source of S 1 because of the burned fuse. This causes the range of the gate-to-source voltage Vgs on S 1 to be very close, which reduces the current through S 1 . L 1 will then operate to pull the output node voltage high.

Similarly, if the clear fuse is burned, there is essentially no current flow through the clear branch. The reference voltage VREF is pulled high by the load device L 2 . The gate of S 1 rises to Vcc. The current in the first branch is therefore increased and the voltage on the output node drops low, to approximately Vss.

FIG. 2 shows the opposite complimentary metal oxide semiconductor (CMOS) arrangement for a single-ended version of the circuit of FIG. 1 . In FIG. 1 , p-channel devices performed the programming and n-channel devices performed the sensing. In FIG. 2 , n-channel programming and p-channel sensing has an advantage because the n-channel programming transistors have a decreased cell size. In addition, FIG. 1 shows a differential circuit, while FIG. 2 is a non-differential, or single-ended, circuit. This has advantages when comparing embodiments of the invention to the prior art.

However, as discussed previously, this circuit suffers from the same disadvantage of the output being marginal, where the output is not clearly either a 1 or a 0. This marginal output results from a small difference between the post-burn resistance and the unburned resistance of the fuses. Increasing the programming voltage applied to the fuse may increase this difference, but the higher voltages may damage other parts of the circuit.

An embodiment of a single-ended, or ‘non-differential’ fuse-based circuit is shown in FIG. 3 . This embodiment uses VdNMOS devices, designated by a circle, for programming and protection of the non-high-voltage devices, although any other transistors that allows higher than nominal voltage could also be used. In one embodiment, these high-voltage devices are also asymmetric, in that their sources and drains cannot be reversed. Because of the high-voltage devices, the circuit has two levels of voltage supply, Vcc and VccHF. In this embodiment, the VdNMOS devices are programming transistor PT 1 , load transistors L 1 and L 2 , and pass transistors PA 2 and PA 3 . Sense transistors S 1 , S 2 and clamping transistors C 2 and C 3 are not high voltage devices and require protection from the high voltage.

For ease of understanding of the embodiments of the invention, the operation of the circuit will be discussed in terms of one of three phases: the pre-programming phase; the programming phase; and the sensing phase. In the pre-programming phase, for example, the voltage supply Vcc is at a nominal voltage, such as 1.3 volts. The high-voltage supply VccHF is at a high voltage, typically two times Vcc, in this case 2.6 volts. In one embodiment, the load transistors L 1 and L 2 are matched high-voltage devices that can handle double the usual voltage, hence the doubling of Vcc to reach the value for VccHF.

In the pre-programming phase, the pass transistors PA 2 and PA 3 are off. The programming circuit and transistor PT 1 drift up to VccHF. PT 1 is also off, and its drain drifts up to VccHF. The pass transistors guarantee that the gate voltage of the clamp and sense transistors is at a neutral voltage and the bulk, source and drain are allowed to drift to VccHF. This does not damage these transistors, as the gate voltage is clamped to a safe voltage of Vcc, thereby protecting the sense and clamp transistors from the high voltage.

Using a mix of high-voltage transistors and nominal-voltage transistors allows for higher post-burn resistance in the fuse(s) due to the high-voltage burning of the fuse, while also providing lower power consumption. The combination of high and nominal voltage devices provides high sensitivity and lower power consumption than using all high-voltage devices.

In the non-sensing phases of programming and pre-programming, clamp transistor C 3 clamp the gate of the output driver to a predictable, known value when not sensing. This ensures that there is no additional leakage current. The circuit shown in this embodiment does not support simultaneous sensing and programming.

In the programming phase, the AND gate inputs 14 are manipulated such that PT 1 turns on, pulling the negative terminal of the fuse 10 to ground, which causes the fuse 10 to be burned. At the end of the burning event, the negative terminal of the fuse 10 is at or near Vss. Subsequently, the source of the sensing transistor S 2 will also be at or near Vss. Additionally, the bulk of S 2 , electrically coupled to its source, is also pulled to a voltage at or near Vss.

During programming, claiming device C 2 pulls the gate of S 2 to nominal Vcc, the gate to drain and gate to source voltages are held to safe levels. Tying the bulk terminal of S 2 to the source additionally guarantees that the device voltages stay within safe limits. As the source of S 2 goes towards Vss, the gate is clamped to ground by clamping device C 2 and the voltage across the device will never exceed 1.3 volts, or the value of Vcc. Similar precautions are taken with sensing device S 1 to protect it from the high voltages.

In the sensing phase, VccHF and Vcc are set to the same voltage. This will typically equal the Vcc voltage used in the pre-programming and programming phases, but may vary as needed. The sense enable signal 16 is high, which turns on load devices L 1 and L 2 . The pass transistors PA 2 and PA 3 turn on. This causes the output of PA 3 , pulled high by L 2 to be clamped to Vcc and produce a voltage in the appropriate range for a logical 1. This causes the circuit to operate similar to that shown in the prior art, with a logical 1 being produced if the fuse if burned, and a logical 0 being produced if it is not burned.

FIG. 4 shows an embodiment of a differential fuse-based circuit. It is analogous to that shown in FIG. 1 , but includes the high-voltage VdNMOS devices for the programming transistors PT 1 and PT 2 , the pass transistors PA 1 , PA 2 and PA 3 and the load transistors L 1 and L 2 . These devices operate to protect the non-high-voltage devices from the high programming voltages.

In FIG. 3 , the protection circuit of PA 3 and C 3 were necessary to protect the sensing transistor S 2 . The reference 12 in FIG. 3 never had its negative terminal pulled toward Vss, therefore no protection circuit was necessary. However, in FIG. 4 , the reference fuse 12 will have its negative terminal pulled towards Vss, and therefore the protection circuit of PA 1 and C 1 is necessary. In addition, a second programming transistor PT 2 is necessary to control programming of the clear, or reference, fuse 12 .

The protection from the high voltages used in programming the fuse or fuses comes from these high-voltage devices configured to block the non-high voltage devices. Additionally, tying the bulk of the sensing devices S 1 and S 2 to their sources also operate to protect them from the high voltages.

Thus, although there has been described to this point a particular embodiment for a method and apparatus for a programming circuit for fuse-based cells, it is not intended that such specific references be considered as limitations upon the scope of this invention except in-so-far as set forth in the following claims.