BACKGROUND OF THE INVENTION

Field of the Invention

This invention relates to circuit protection devices.

Introduction to the Invention

Many circuits require protection from voltage transients which will damage one or more components of the circuit. This is a particular problem with large scale integrated circuits, which require protection from transients of relatively low voltage. It is known to provide such protection by connecting the circuit to ground through a protection device which has a very high resistance in its normal, unswitched state, but which, when exposed to the voltage transient, is very rapidly converted into a switched state in which it has a very low resistance. Many such devices comprise two metal electrodes which are separated by an insulating medium, and are switched by electrical breakdown of the insulating medium. Such devices are sometimes referred to as foldback switching devices; examples include gas tubes, Zener diodes, and devices comprising chalcogenide glasses. Some of these devices are designed so that the switching takes place as a result of electrical discharge through a gaseous insulating medium and along the surface of a solid dielectric which lies between the electrodes. Such discharge is referred to herein as surface-mediated discharge.

A problem of the known devices, especially those which are switched by surface- mediated discharge, is that they fail in the low resistance state after a relatively small number of switching cycles.

SUMMARY OF THE INVENTION

We have discovered that this problem can be substantially reduced by replacing at least one of the metal electrodes by an electrode whose active region comprises a conductive material which, if it is dislodged from the electrode during the switching, is dislodged as a heated fluid which, when cooled, becomes a solid having a relatively high resistivity. The term "active region" of an electrode is used herein to denote that part of an electrode from which discharge takes place during switching of the device, and which is, therefore, liable to electrical erosion. The term "dislodged" is used to include partial

or complete removal of the material from the active region of the electrode. When we say that the dislodged material, when cooled, has a relatively high resistivity, we include dislodged material which, when deposited on a solid dielectric between the electrodes, reacts with the solid dielectric and becomes resistive as a result of such reaction.

We believe that such materials result in improved performance because they do not (as metals do) assist in the formation of conductive pathways between the electrodes as a result of redeposition of material dislodged from the electrode during switching. A preferred such material is a doped semiconductor which sublimes when heated and which on sublimation loses its dopant and thereby becomes resistive.

We have found that, through use of this invention, it is possible to make, at relatively low cost, novel devices which have unexpectedly superior combinations of electrical properties such as switching threshold stability and cycle switching stability. More specifically these devices can exhibit low capacitance; high resistance in the unswitched state; fast response; fast recovery; higher energy handling capability for their size than prior art devices, especially on-chip devices; low impulse ratio (i.e. a low ratio of pulse firing threshold voltage to DC withstand voltage); and essentially temperature independent device characteristics.

in one aspect, this invention provides a circuit protection device which comprises

a first electrode,

a second electrode which is spaced apart from the first electrode, and

an insulating medium between the first and second electrodes;

the device

(a) if it does not form part of a circuit or if it forms part of a circuit which is in a normal operating condition, being in an unswitched state in which the electrodes are electrically separated from each other, and

( b ) if it forms part of a circuit which is initially in a normal operating condition and is then subjected to a condition against which the device provides protection, is converted from the unswitched state into a switched state in which the electrodes become electrically connected to

each other as a result of electrical breakdown of the insulating medium; and

the active regions (as hereinbefore defined) of the first electrode comprising a first conductive material which, if it is dislodged from the first electrode when the device is converted from its unswitched state to its switched state, is dislodged as a heated fluid which, when cooled, has a relatively high resistivity.

In another aspect, this invention provides an electrical circuit having a ground line which contains a device as defined above.

In another aspect, this invention provides an electrical system in which a device as defined above is used as a trigger for another switching system.

In another aspect, this invention provides a process for making a device as defined above, or a set of such devices, which process comprises

( a ) providing a semiconductor wafer;

( b ) forming on the wafer a plurality of spaced-apart dielectric segments;

( c ) forming a second electrode on each of the dielectric segments; and

( d ) dividing the product of step (c) into individual devices or into sets of devices.

In developing procedures for characterizing and testing switching devices of the invention, we have discovered novel characterizations and quality control procedures for circuit protection devices which are applicable to such devices in general as well as to devices as defined above.

BRIEF DESCRIPTION Q THE DRAWING

The invention is illustrated by the drawings in which Figures 1 , 5, and 7 show plan views of devices of the invention, Figures 2, 3, 4, 6, 8, 11, 13, 15 and 19 show cross-sectional views of devices of the invention, Figure 9 shows a perspective view of devices of the invention, Figures 10, 12, 14, 16, 17, and 18 show electrical circuits and electrical systems comprising devices of the invention, and Figure 20 shows a plot of

device survival in number of cycles as a function of pulse energy for devices of the invention.

In each of the figures, 1 indicates the first electrode, 2 is the second electrode, and 3 is the dielectric material. In some of the figures, additional features are indicated and are designated as follows: 4 indicates an aperture through the second electrode, 5 and 51 are inactive regions suitable for wire attachment, 6 is a second dielectric layer, 7 is a device of the invention, e.g. a fast foldback device, 8 is a power line or a signal line, 9 is a diode, 10 is a region of n-type semiconductor material in diode 11, 11 is a diode composed of a layer of p-type material and a layer of n-type material, 12 is an electrical switching mechanism, 13 is an electrical circuit, 14 is a current control means, 15 is an impedance, 16 is an electrically insulating wall of the switching mechanism 12, 17 is a second electrode of the switching mechanism 12, 18 is a plasma generator, 19 is a variable DC voltage source, 21 is a blocking capacitor, 22 is a pulse generator, 23 is a chip comprising a number of individual devices 7, 25 is a telephone line, 26 is the end of the telephone line 25 connected to the subscriber (i.e. the telephone subscriber line), and 27 is the end of the telephone line 25 connected to the telephone exchange.

DETAILED DESCRIPTION QFTHE INVENTION

The First Electrode

The active region of the first electrode is composed of a first conductive material which, when the device is switched, can be dislodged from the electrode as a heated fluid, i.e. as a gas, vapor or liquid. The dislodged material can be completely dislodged, e.g. by sublimation, evaporation or liquefactions, or can be only partially dislodged, e.g. by liquefaction. The dislodged material is such that when it solidifies, it has a comparatively high resistivity so that it cannot provide, or help to provide, a low resistance path between the electrodes which will prevent the device from operating effectively. The first material preferably has a resistivity at 23°C of less than 10 ohm-cm, e.g. less than 1 ohm-cm, particularly less than 0.05 ohm-cm, especially less than 0.005 ohm-cm. When a thin film of the first material is evaporated under vacuum and the vapors condensed, the condensate preferably has a resistivity at 23°C of at least 1000 ohm-cm, particularly at least 10,000 ohm-cm, especially at least 100,000 ohm-cm.

The first material preferably comprises an insulating matrix and a dopant which renders the material conductive and which is removed, e.g. sublimed or evaporated, or otherwise rendered ineffective when the first material is heated. Particularly preferred are doped inorganic semiconductors, in particular doped polycrystalline materials, especially n-type or p-type polycrystalline silicon which has been doped to a resistivity of less than 10 ohm-cm, e.g. 0.001 to 0.01 ohm-cm.

The active region of the first electrode preferably consists essentially of a first material as defined, and it is often convenient for the whole of the electrode to consist of the first material. The first material can be a mixture of two or more first materials as defined.

The thickness of the active region of the first electrode is preferably about 0.15 to about 10 micron. The electrode may have at least one area of greater thickness for attachment of an electrical lead thereto.

The Second Electrode

The active region of the second electrode can also comprise a first material as defined. For example the first and second electrodes can be identical. However, it is generally preferred that the active region of the second electrode should comprise, preferably consist of, a metal or metal-like material, with the device being used so that the metallic electrode is connected to ground or the negative pole of the circuit so that switching of the device does not cause metal to be dislodged from the electrode. Preferred metals are aluminum, nickel, gold and tungsten; copper, molybdenum and tin can also be used. Alloys, intermetallic compounds, and mixtures of these and other metals can also be used, as can certain metal nitrides, suicides and oxides, e.g. titanium nitride, titanium suicide, and indium tin oxide, which are referred to as narrow bandgap compounds. The second electrode can also be composed of carbon or graphite, particularly pyrolytically-deposited graphite. The second electrode is preferably composed of a material which when molten, will not melt the solid dielectric (if there is one) between the electrodes.

The thickness of the active region of the second electrode is preferably about 0.15 to about 10 microns. Preferably the electrode has at least one area of greater thickness for attachment of an electrical lead, e.g. a wire bond, thereto. This area of greater thickness may be referred to as an inactive region.

The Solid Dielectric

The devices of the invention preferably comprise a solid dielectric which lies between the electrodes so that the device is switched by electrical discharge through an insulating medium along a surface of the dielectric. Preferably at least part of the electrical discharge takes place through a gaseous insulating medium. The dielectric has a resistivity at 23°C which is generally at least 10 ⁶ ohm-cm, preferably at least 10 ⁸ ohm-cm, particularly at least 10 ⁹ ohm-cm. The dielectric should not create conductive debris when the surface-mediated discharge takes place along its surface. Suitable inorganic dielectrics include silicon dioxide, aluminum oxide and mica. Suitable organic dielectrics include methacrylate esters, e.g. poly(methylmethacrylate), and fluoropolymers, e.g. homopolymers and copoiymers (including terpolymers) of tetrafluoroethylene, e.g. copoiymers of tetrafluoroethylene and ethylene or propylene or both.

Arrangements of the Electrodes and the Solid Dielectric

A wide variety of arrangements of the electrodes and the solid dielectric can be used. The distance between the active region of the first electrode and the active region of the second electrode, measured along the surface of the dielectric which mediates the discharge, is preferably about 0.1 to about 10 microns, particularly about 0.5 to about 4 microns, especially about 0.5 to about 2 microns.

Generally, it is convenient for the dielectric to be in the form of a laminar, preferably planar, slab. The slab can have a regular geometric shape in plan view, such as a rectangle or a circle, The device shown in plan view in Figure 1 has a rectangular shape. The cross-sectional view in Figure 2 taken along line 2-2 of Figure 1 shows that the dielectric layer 3 is positioned between the first electrode 1 and the second electrode 2. A second dielectric layer 6 partially covers the second electrode, leaving a circular area 5 (an inactive area from which discharge does not take place during switching of the device) available for making electrical connection to other components.

In one class of devices, one electrode is on one side of the slab and the other electrode is on the other side of the slab, with one of the electrodes, preferably the first electrode, extending outwardly from an edge of the slab, and the other electrode being flush with that edge or set back from that edge, as shown in Figure 3. The electrodes can be otherwise flush with, or set back from the edge of the dielectric slab. The side of the other electrode (i.e. the one which does not extend outwardly from the dielectric) can be

shaped to produce desired effects. For example, it can have a stepped configuration so that it is thinner near the operating edge, as in Figure 4, or it can be chamfered. The edge of the slab along which the discharge takes place can be at right angles to the slab, or it can be shaped, e.g. chamfered at an angle to the plane of the slab so as to increase the effective distance between the electrodes. The edge of the slab along which discharge takes place is usually an external edge, particularly a straight external edge. However, it can be an internal edge provided by an aperture in the slab, e.g. a round hole which extends at least part of the way through the slab.

When the edge is an internal edge provided by an aperture which extends only a part of the way through the dielectric slab, this is an example of a class of devices of the invention in which at least the first switching of the device takes place as a result of electrical discharge partly along a surface of the dielectric and partly through the dielectric. The first switching, it is believed, creates holes in the dielectric through which electrical discharge takes place during subsequent switching. Figures 5 and 6 are examples of this aspect of the invention.

For some devices, it is desirable that the ratio of the area of the slab to the length of its perimeter be as small as possible in order to maximize the amount of edge. Devices of this type can have one or more reentrant portions as shown in Figures 7 and 8.

In another class of devices, both electrodes are secured to the same surface of a solid dielectric. A device of this type is shown in Figure 9. Shown in this figure are two inactive areas 5, 51, suitable for electrical connection.

Use of the Devices of the Invention

The devices of the invention generally have electrical characteristics which are dependent upon whether the first electrode or the second electrode is connected to a negative pole (or to ground). Connection of the first electrode to the positive pole is denoted herein as a normal polarity connection, while connection of the first electrode to the negative pole is denoted herein as a reverse polarity connection. When the second electrode is metallic, normal polarity connection generally results in a lower threshold voltage than reverse polarity connection. To protect a circuit comprising two conductors from a fault voltage which may be imposed on either of the conductors, two substantially identical devices can be connected between the conductors, one in its normal polarity connection, the other in its reverse polarity connection. Examples of this are shown in

Figures 10 and 11. In Figure 10, two devices of the invention 7 are inserted in a circuit between a telephone line 25 connecting the telephone exchange (or central office) 27 to the telephone subscriber line 26. in Figure 11, the two devices are inserted across a line so that one has a normal polarity connection and the other has a reverse polarity connection.

The devices of the invention generally have at least in a normal polarity connection and preferably also in a reverse polarity connection, a threshold voltage of less than 1000 volts, e.g. less than 500 volts, preferably less than 300 volts, and a switching time of less than one nanosecond when subjected to a voltage 10% above the threshold voltage.

A device of the invention can be connected in series with a diode as shown in Figure 12 in order to provide a combination which will be switched only by a voltage transient having a particular polarity. As shown in Figure 13, the diode 11 can be provided by an n-type semiconductor component 10 which is secured to a first electrode composed of a p-type semiconductor. Two such combinations, preferably of identical components, can be connected in parallel with each other and with opposite polarities across a pair of conductors.

Devices of the invention are particularly useful as part of a ground line connected to a circuit in order to protect the circuit from voltage transients which switch the device and which are thus bled off to ground before they can damage the circuit.

A device of the invention can also be used as a trigger for another electrical switching system, thus substantially reducing the response time of that other switching system. A system of this type is shown in Figures 14, 15, and 16. An impedance 15 and/or a current control means 14 can be connected in series with the device 7 to control and/or limit the current passed by the device through the circuit 13. For example, the device can act as a plasma generator 18 within a second electrical switch which switches by electrical discharge between two electrodes. The second switch can be a conventional switch, or another device of the invention. As shown in Figure 14, one of the electrodes of the device of the invention, preferably the first electrode, can also provide one of the electrodes for the second switch. The device can also be connected in series with a plasma generator which forms part of a second electrical switch.

A device of the invention can also be integrated with a semiconductor chip to form a composite electronic device, with one of the electrodes of the device, preferably the first electrode, providing or being connected to, the ground plane of the composite device.

Figure 17 shows a circuit used to test the devices of the invention by applying normal polarity pulses. A device 7 is inserted across a line and power (which may be a bias voltage) is supplied by a variable DC voltage source 19. Preferably the DC voltage is provided by a source having an internal impedance of less than about 10 ohms and/or the product of the internal resistance (in ohms) of the DC voltage source multiplied by the internal capacitance (in farads) of the DC voltage source is at least 20 microseconds.

The invention is illustrated in the following Example.

EXA E E

A chip containing 20 devices of the invention was prepared. The chip is illustrated in Figures 18 and 19. Figure 19 is a partial cross-section taken on line 19- 19 of Figure 18. In each device, the electrode 2 was composed of aluminum and was 0.5 micron thick; and the dielectric 3 was composed of silicon dioxide, with its dimensions being 0.014 by 0.026 inch (0.036 by 0.066 cm), with a thickness of 2.5 microns in the center section and 1.3 micron in the peripheral section. The electrode 1 , which is physically shared by all the devices, is composed of a p-type silicon having a resistivity of 0.001 to 0.01 ohm-cm.

This chip was made as follows. A doped silicon wafer was polished on one side and cleaned with a mixture of H2SO4 and H2O2. The wafer was oxidized at 1050°C in wet oxygen to form a 2.3-micron-thick Siθ2 layer on it. Using positive photoresist techniques, the Siθ2 on the polished side of the wafer was selectively etched into rectangular islands having dimensions of about 0.014 to 0.026 inch (0.036 to 0.066 cm) and about 2.3 microns thick. The etched wafer was again oxidized at 1050°C in wet oxygen to form an Siθ2 layer 1.3 micron thick on the exposed parts of the wafer and to increase the thickness of the existing Siθ2 islands to about 2.5 microns. A 0.5-micron- thick layer of aluminum was sputtered over the Siθ2 layer on the polished side of the wafer. After applying a photoresist over the aluminum, the Siθ2 layer was removed from the unpolished side of the wafer. Using photoresist techniques, the aluminum and Siθ2 on the polished side of the wafer were selectively etched into rectangular islands about 14 by 26 mile.s (0.355 by 0.660 mm). The wafer was then divided into a number of chips as shown, each containing 20 devices.

The chips could be again divided into individual devices, but were retained in the form shown for testing. For the tests, the chips were die-bonded inside 24-lead ceramic flat packs using a conductive epoxy resin to provide a conductive common path to the silicon electrode 1. Wires were then ultrasonically bonded at one end to lead pads inside the package and at the other end to the centers of the aluminum electrodes 2. The packs were then hermetically sealed. The tests were carried out using the circuit illustrated in Figure 17. Each test comprised subjecting 10 devices (half the devices on a chip) to repeated normal polarity test pulses (that is, a pulse of negative amplitude with respect to the metal electrode) of a given energy (measured in coulombs) until failure of all the devices to perform their intended function occurred. Failure is defined as one or more of the following: (1) the device resistance in the high resistance condition falls below 1 x 10 ⁶ ohms, (2) the device fails to switch, and (3) the device threshold voltage increases by more than 100 volts above its initial value. The pulse energy (i.e. the peak current (I) times the pulse half-width (T)), in coulombs, was varied between about 10" ⁶ coulombs and about 2 x 10" ⁴ coulombs by varying the peak current (I) and/or the half width (T) of the pulse. The number of pulses which each device withstood before failure occurred was noted. Figure 20 illustrates the results of this test and shows the mean number and also the minimum and maximum number of pulses to failure for each sample set of 10 devices.