|Publication number||USRE37477 E1|
|Application number||US 08/532,011|
|Publication date||Dec 18, 2001|
|Filing date||Sep 21, 1995|
|Priority date||Nov 6, 1987|
|Publication number||08532011, 532011, US RE37477 E1, US RE37477E1, US-E1-RE37477, USRE37477 E1, USRE37477E1|
|Inventors||Francois Tailliet, Jacek Kowalski|
|Original Assignee||Sgs-Thomson Microelectronics, Inc.|
|Export Citation||BiBTeX, EndNote, RefMan|
|Patent Citations (15), Non-Patent Citations (7), Referenced by (2), Classifications (6), Legal Events (1)|
|External Links: USPTO, USPTO Assignment, Espacenet|
This application is a continuation of application Ser. No. 08/320,124 filed Oct. 7, 1994, now abandoned, which is a continuation of application Ser. No. 08/149,719 filed Nov. 9, 1993, now abandoned, which is a continuation of application Ser. No. 07/814,400 filed Dec. 23, 1991, now abandoned, which is a Reissue of 07/264,202, now U.S. Pat. No. 4,890,187 filed Oct. 28, 1988.
1. Field of the Invention
The present invention concerns integrated circuits made with MOS (metal-oxide-semiconductor) technology and, more generally, integrated circuits susceptible to risks of malfunctioning or destruction in the presence of electrostatic discharges.
Electrostatic discharges may sometimes occur quite simply because a user touches the input/output terminals of the integrated circuits. This risk will become increasingly frequent as integrated circuits come into widespread use among the public, for example in the form of chip cards which will undergo a lot of handling and in which the output terminal can be directly accessed without protection.
2. Description of the Prior Art
To protect integrated circuits against these risks, it is now standard practice to provide for an element integrated in the circuit, between each of the signal terminals and the common ground terminal. The function of this element is to become suddenly conductive when overvoltage, exceeding a pre-defined threshold, appears between its terminals. This element, of course, has energy absorption or current absorption capacities which are greater than those of the other elements of the integrated circuit and, furthermore, it is designed to become conductive more quickly than the elements to be protected in the integrated circuit.
In a standard way, the protection element is a diode with its anode connected to the ground terminal and its cathode connected to another input/output terminal (assuming, of course, that the ground terminal is a negative supply terminal).
This diode stays normally off for as long as the voltage of the terminal to be protected does not go beyond an avalanche threshold. If the threshold is exceeded, the diode goes into avalanche mode and absorbs the electrostatic discharge current. It must go into avalanche mode before the other junctions, present in the integrated circuit and subject to the same overvoltage. Now, the voltage for triggering the avalanche process is essentially related to technological parameters such as the doping of the semiconducting substrate in which the circuit is formed, and the doping of the various regions diffused in this substrate.
It is difficult to have precise knowledge of the avalanche voltages of the junctions to be protected and to master the making of a protective diode with an avalanche voltage which would be below that of the junctions to be protected while, at the same time, remaining greater than the normal operating voltages of the circuit. For, it would be unacceptable for the protection element to go into a conduction state simply because a signal which has a relatively wide amplitude (but does not represent a dangerous discharge) appears at a terminal of the circuit.
Now, the specifications relating to behaviour, under voltage, of the integrated circuits are relatively strict and are finally fairly close to voltages which might damage or destroy the circuit. For example, specifications may require that the circuit should work normally if it receives signals or supply voltages of up to 15 volts, whereas the destruction threshold would be 22 volts. It must therefore be seen to it that the triggering of the protective element starts from a voltage which is considerably below 22 volts (if it is too close to 22 volts, there remains a risk of destruction because the conduction is not triggered fast enough). But it is also necessary that the protection element should be triggered for a voltage which is sufficiently greater than 15 volts (if not, there is a risk of untimely triggering below 15 volts).
The margin available for choosing the avalanche voltage is therefore narrow and it is all the smaller as technological variations are greater, i.e. it is all the smaller as the dimensions, concentrations, depths of junctions and other parameters are less well controlled in the manufacturing process.
To arrive at protection of the integrated circuits which is as efficient as possible while, at the same time, permitting normal functioning voltages which are as high as possible, the present invention proposes a new approach based on the observation that the rising edge of the overvoltages coming from electrostatic discharges (or of most overvoltages liable to appear and damage the circuits) is far steeper than the rising edge of the normal voltages (which may be relatively high) occurring at the input/output terminals of the circuit.
For, these normal voltages are either supply voltages which may exceed the value stipulated by the specifications but for which there is no reason or possibility that they should vary abruptly, or input-output logic signals which would exceptionally change to a value greater than the maximum permitted by the specifications. The rising edges of these logic signals do not exceed a few volts per nanosecond, while the rising edges of the electrostatic discharges rather have values of a some hundreds of volts per nanosecond.
According to the invention, it is proposed to connect, between two terminals of the integrated circuit, a diode associated with a means to modify the distribution of the equipotential lines at the position where the avalanche conduction of the diode is triggered, said means being connected to the terminal to be protected and being sensitive to the rising slope of the overvoltages appearing at this terminal in such a way that the avalanche triggering voltage is weaker, when the slope is steeper, than it is when the slope is less steep.
The means for modifying the distribution of equipotential lines is preferably an insulated gate surrounding an entire diffused region constituting the cathode (or anode) of the diode, and located in the immediate vicinity of this region.
This gate is connected to the terminal to be protected by an integrating circuit that introduces a time constant in such a way that an overvoltage with a steep edge reaching this terminal is transmitted with a slight delay to the gate.
For example, the gate is connected, firstly, to one end of a resistor, the other end of which is connected to the diffused region and the terminal to be protected and, secondly, to an end of a capacitor, the other end of which is connected to the other terminal of the diode.
In practice, the capacitor does not need to be given the form of a well identified circuit element, for the junction capacitances of the resistor (made by a diffused region in the substrate) and the capacitance between the gate and the semiconducting substrate in which the cathode of the diode is diffused are high enough to play the role expected of them, namely to cause a delay in the arrival of a voltage front at the gate. This delay induces a potential difference between the diffused region and the gate, which is all the greater as the front is steeper.
Other features and advantages of the invention will appear from the following detailed description made with reference to the appended drawings, of which:
FIG. 1 gives a schematic view of an integrated circuit, one input terminal of which is protected by a protection element.
FIG. 2 gives a symbolic view of the protection device according to the invention,
FIG. 3 shows a lateral sectional view, along the line AA of FIG. 4, of a possible structure for the protection element according to the invention,
FIG. 4 shows a top view of the structure of FIG. 3 also showing the constitution of the integration resistor of FIG. 2.
FIG. 1 gives a schematic view of an integrated circuit CI, with two input/output terminals M and B. Of course, the circuit may have other terminals which have not been shown.
The terminal M is a potential reference terminal. It is the most negative or the most positive terminal of the circuit. In the example which shall be described in the rest of the description, it is the most negative terminal, that is to say, in normal operation, the voltage at the terminal B never goes below the potential of the terminal M.
The terminal B is an input/output terminal to be protected. It may be a signal terminal receiving logic signals, or again, a positive supply terminal.
In a standard way, there is provision for a protection element P between the terminals B and M. This element behaves like a practically infinite impedance for as long as the potential difference between the terminals B and M does not exceed a threshold Vs, and then as a voltage limiter with a high current absorption capacity when this threshold is exceeded.
The invention concerns the formation of the protection element.
FIG. 2 gives a schematic view, in symbolic form, of the operating principle of the protection element according to the invention, placed between the terminals B and M of the integrated circuit.
The protection element behaves, firstly, like a diode D, the anode of which is connected to the terminal M and the cathode to the terminal B. This diode is normally reverse biased since, as stated, the potentials present at the terminal B are greater than the reference potential M. In any case, if the potential of the terminal B drops by about 0.6 volts below the reference voltage, the diode would start conducting, preventing this fall from continuing.
When the potential of the terminal B becomes highly positive and exceeds a certain threshold Vs, the reverse biased diode D goes into the avalanche state and starts conducting.
In FIG. 2, a symbolic depiction is given, in the form of a gate G in the vicinity of the junction of the diode D, of a means to cause variation in the threshold Vs for which the diode D goes into avalanche mode. The application to the gate G of a variable potential Vg has the effect of modifying the threshold Vs.
More precisely, depending on the aim sought to be achieved according to the invention, namely the increasing of Vs if the overvoltages are simply “normal” level signals, which are slightly too high and the reducing of Vs if the overvoltages are of electrostatic origin, the gate G is connected to the terminal B by means of an integrating circuit in such a way that the potential Vg follows the potential of the terminal B with a certain delay. This circuit acts in such a way that the potential Vg remains all the closer to the potential of the terminal B as the rising time of the overvoltages applied to the terminal B are slower during the rising of the signal.
The integrating circuit is shown in the form of a resistor R and a capacitor C. The resistor R is connected between the gate and the terminal B. The capacitor C is connected between the gate G and the ground M.
The gate is placed in the vicinity of the junction of the diode D in such a way that it has an influence on the distribution of the equipotential lines inside the anode regions and cathode regions of the diode. This influence is in the following direction: the equipotential lines are more curved and more contracted in the breakdown region of the diode (the region in which the avalanche mode gets initially triggered) when the potential Vg is weaker. They are less curved and more spaced out when the potential Vg is greater.
The result of this is that the breakdown occurs for a higher voltage Vs1 when there is an overvoltage with a relatively slow rising edge and a lower voltage Vs2 when there is an overvoltage with a very steep rising edge. In both cases, the gate potential Vg ad the potential of the overvoltage on the terminal B rise and equipotential lines increase their curvature and get contracted as and when this rise occurs. But, in the latter case, the gate potential Vg is not able to follow the potential on the line B, and the curvature of the equipotential lines is even greater and their contraction is even greater than in the former case during the rising edge. The result of this is that the avalanche threshold is appreciably lower in the case of a very steep rising edge.
In an illustrative, digital example, it can be shown that with a resistance of 2 kilohms and a capacitance of 0.1 picofarad, the potential Vg will follow the potential of the terminal B with a difference of about 1 volt for rising edges of about 5 volts per nanosecond (standard value for logic signals), and a difference very swiftly reaching 40 volts for rising edges of 200 volts per nanosecond (standard value for electrostatic discharges). This difference between the two cases is quite enough to substantially modify the avalanche voltage threshold of the diode.
The invention can be implemented especially with a diffused junction diode flush with the surface of a semiconducting substrate, the surface outcrop of the junction being surrounded by or covered with an insulated conducting gate coming to the immediate vicinity of the junction throughout its periphery.
FIG. 3 shows a lateral cross-section view of a structure of this type and FIG. 4 shows a top view. The section is made along the lines AA of FIG. 4. In FIG. 4, the crosses show contact zones between a conducting metallization and a sub-adjacent region.
The diode D consists of an N+ type region 12 diffused on the surface of a P type semiconducting substrate 10.
The substrate is connected to the ground M either by its rear face or by means of a P+ type region 14, locally diffused at the surface of the substrate and covered with a metallization 16 which connects it with the terminal M.
The region 12 is connected to the terminal B by a metallization 18.
The region 12 is made by local diffusion of N+ type impurities from the surface of the substrate. The junction between the region 12 and the substrate 10 thus has, as is well known, a curvature 20 along its periphery, where it rises towards the upper surface of the substrate.
It is known that it is in the region of this curvature 20 that the avalanche mode is initially triggered when the diode is subjected to a high reverse voltage. The reason for this is that the equipotential lines are highly curved and contracted in the neighbourhood of this curvature.
Sometimes also, the avalanche mode is triggered in the zone where the junction is flush with the surface of the substrate, either because the dopings are higher at the surface or as a result of faults that might exist at the surface outcrop of the junction. In this case too, the breakdown occurs in the zones where the electrical field is the highest with respect to the capacity of these zones to support this electrical field, and consequently, again in the zones where the equipotential lines are the most contracted.
The periphery of the diffused region 12 is completely surrounded by a polycrystaline silicon gate G covering the outcrop of the junction and insulated from the surface of the substrate by a thin insulating layer 22 (silicon oxide). This gate G is made by the same process as the gates of the MOS transistors of the integrated circuit and, similarly, as the gates of the MOS transistors that go slightly beyond the source and drain regions. Similarly, the gate G covers the surface outcrop of the junction in going very slightly beyond the entire periphery of the region 12.
The greater the positive potential applied to the gate G, the more the equipotential lines in the neighbourhood of the curvature of the junction 20 and the surface outcrop of the junction get loosened and, hence, the greater is the voltage Vs needed for application to the terminals of the diode D to put it in avalanche mode.
The general form of the equipotential lines in the presence of a positive voltage at the gate G is shown schematically by dashes in FIG. 3. In the presence of a null potential at the gate G, the equipotential lines remain practically parallel to the curvature 20 and rise to the surface.
The gate G is connected to the terminal B by means of a resistor R which cannot be seen in the section AA of FIG. 3 but is seen in the top view of FIG. 4. This resistor can be made by a slightly doped N type diffusion 24 in the P type substrate. A metallization 26 (FIG. 4) comes into contact in a region 28 with the gate G and, in a region 30, with an end of the diffusion 24. Another end of the diffusion 24 is connected by a contact zone 32 with the metallization 18, hence with the terminal B. The contacts 30 and 32 are formed on top of the small N+ type regions 30′ and 32′ diffused at the ends of the resistor 24.
The capacitance value of C of the diagram of FIG. 2 is simply the existing capacitance between the gate and the substrate P, connected to the ground as well as between the region 24 and the substrate. This capacitance can easily reach 0.1 picofarads because the gate G is separated from the substrate only by a very small thickness of insulating material (a few hundreds of angstroms at the most). It is not necessary then to add a specific capacitance to arrive at the diagram of FIG. 2.
The resistance R may have a value of a few kilohms.
In FIGS. 3 and 4, there is also shown an N+ type diffused surface region 34, laterally spaced from the diode D and connected to the ground M by an extension of the metallization 16.
Thus, between the cathode of the diode (region 12), the substrate 10 and the region 34 there is formed an NPN type lateral bipolar transistor. This transistor makes it possible to maintain a current up to total extinction of the overvoltage (and not only until the voltage falls below the avalanche threshold V1).
|Cited Patent||Filing date||Publication date||Applicant||Title|
|US3855605 *||Jul 27, 1973||Dec 17, 1974||Rca Corp||Carrier injected avalanche device|
|US4139935 *||Mar 29, 1977||Feb 20, 1979||International Business Machines Corporation||Over voltage protective device and circuits for insulated gate transistors|
|US4323942 *||Feb 11, 1980||Apr 6, 1982||Bell Telephone Laboratories, Incorporated||Solid-state protector circuitry using gated diode switch|
|US4595941 *||Jun 7, 1985||Jun 17, 1986||Rca Corporation||Protection circuit for integrated circuit devices|
|US4602268 *||Dec 23, 1981||Jul 22, 1986||At&T Bell Laboratories||High voltage dielectrically isolated dual gate solid-state switch|
|US4712152 *||Dec 12, 1986||Dec 8, 1987||Mitsubishi Denki Kabushiki Kaisha||Semiconductor integrated circuit device|
|US4720737 *||Dec 22, 1986||Jan 19, 1988||Fujitsu Limited||Semiconductor device having a protection circuit with lateral bipolar transistor|
|US4733285 *||Jul 24, 1985||Mar 22, 1988||Nec Corporation||Semiconductor device with input and/or output protective circuit|
|US4763184||Apr 30, 1985||Aug 9, 1988||Waferscale Integration, Inc.||Input circuit for protecting against damage caused by electrostatic discharge|
|US4786956 *||Oct 11, 1985||Nov 22, 1988||North American Philips Corporation, Signetics Division||Input protection device for integrated circuits|
|US4789917 *||Aug 31, 1987||Dec 6, 1988||National Semiconductor Corp.||MOS I/O protection using switched body circuit design|
|US4799100 *||Feb 17, 1987||Jan 17, 1989||Siliconix Incorporated||Method and apparatus for increasing breakdown of a planar junction|
|US4821089 *||Jun 4, 1987||Apr 11, 1989||American Telephone And Telegraph Company, At&T Laboratories||Protection of IGFET integrated circuits from electrostatic discharge|
|US4890187 *||Oct 28, 1988||Dec 26, 1989||Sgs-Thomson Microelectronics, Sa||Integrated circuit protected against electrostatic discharges, with variable protection threshold|
|EP0032018A1 *||Dec 17, 1980||Jul 15, 1981||Fujitsu Limited||Semiconductor integrated circuit device|
|1||1987 IEEE "How to Prevent Circuit Zapping" Robert J. Antinone, BDM Corporation, pp. 34-38.|
|2||1988 IEEE/IRPS "Internal Chip ESD Phenomena Beyond the Protection Circuit," V. Duvvery, et al., pp. 19-25.|
|3||1988 IEEE/IRPS, "Failure in CMOS Circuits Induced by Hot Carriers in Multi-gate Transistors," A. Chatterjee, et al., pp. 26-28.|
|4||IBM Technical Disclosure Bulletin, vol. 22, No. 11, Apr. 1985.*|
|5||IBM Technical Disclosure Bulletin, vol. 23, No. 2, Jul. 1980.*|
|6||IBM Technical Disclosure Bulletin, vol. 27, No. 11, Apr. 1985.*|
|7||Textbook Discussion of ESD, "VLSI Engineering," Prentice Hall, 1988, Thomas E. Dillinger, pp. 508-511.|
|Citing Patent||Filing date||Publication date||Applicant||Title|
|US6611028 *||Oct 8, 2002||Aug 26, 2003||Taiwan Semiconductor Manufacturing Company||Dynamic substrate-coupled electrostatic discharging protection circuit|
|US6780685 *||Aug 12, 2003||Aug 24, 2004||Sharp Kabushiki Kaisha||Semiconductor device and manufacturing method thereof|
|U.S. Classification||361/111, 361/56, 257/356|