GB1558886A

GB1558886A - High-frequency thyristor

Info

Publication number: GB1558886A
Application number: GB36834/76A
Authority: GB
Original assignee: BBC Brown Boveri AG Switzerland; BBC Brown Boveri France SA
Current assignee: BBC Brown Boveri AG Switzerland; BBC Brown Boveri France SA
Priority date: 1975-09-09
Filing date: 1976-09-06
Publication date: 1980-01-09
Also published as: FR2324125A1; JPS5235574A; SE416600B; DE2543909A1; CH589942A5; JPS6016104B2; NO763077L; SE7609773L; FR2324125B1

Description

(54) A HIGH-FREQUENCY THYRISTOR (71) We, BBC BROWN, BOVERI AND COMPANY LIMITED, A company organised under the laws of Switzerland, of Baden, Switzerland, do hereby declare the invention, for which we pray that a patent may be granted to us, and the method by which it is to be performed, to be particularly described in and by the following statement: This invention relates to a thyristor and to a method of operating a thyristor.

A thyristor is known for switching frequencies greater than 10 kHz, in the case of forward blocking voltages greater than 500 V. in which the thyristor has an interdigitated emitter-gate structure, a control zone in contact with the gate and a base zone which is less doped than the control zone, the gate being, when the thyristor is being turned off in the forward blocking condition, at a potential whose polarity is opposite to that of the majority carriers in the control zone, and in which the control zone does not have any shorts such as to short-circuit the PN-junction between the control zone and an adjacent emitter zone.

A thyristor of the type described above is disclosed in Int. J. Electronics 36 (1974), pp.

399-416. Furthermore, circuits in which such thyristors can be used are known, for example from Heumann, Stumpe "Thyristors", Teubner-Verlag 1970. pp. 131-133, or Penkowski, Pruzinsky "Fundamentals of a pulse width modulated power circuit" Power Semiconductor Applications, IEEE Press 1972, pp. 266-275. or "Silicon Rectifier Handbook", Aktiengesellschaft Brown, Boveri & Cie., Baden (Switzerland), 1971, pp. 197-201.

The thyristor disclosed in Int. J. Electronics 36(1974), pp. 399-416 has recovery (turn-off) times of less than 2 llsec, a permissible voltage gradient du/dt = 600 V/psec, a forward and reverse blocking voltage of 650 V, and a switching frequency of 100 kHz.

The low recovery (turn-off) time is substantially attained through the above-described control of the gate. The negative (in the case of a P-control zone) potential at the gate causes the charge carriers of the ON-condition, which is to be switched to the OFF-condition, to be cleared from the control-and base-zones more quickly. This principle is also known as 'gate-assisted turn-off" (GATO).However, this principle entails the danger that, in the case of larger diameters of the semiconductor component to be switched to the OFF state, the current may be "crowded" into a channel of small diameter, and cause local over-heating and damage. The known interdigitated emitter-gate structure serves to prevent this, as the negatively polarized gate can rapidly act on all parts of the P-control zone. A high conductivity of the P-control zone is also of assistance in this context. Emitter shorts, which short circuit the PN-junction between the control zone and emitter zone on the cathode side (in the case of a P-control zone). should not be provided in this case, as otherwise the negatively polarised gate would be without effect. Indeed, the said P-N junction must have a sufficiently good blocking ability. as this ability restricts the possible negative potential at the gate.

The interdigitated emitter-gate structure is not only useful for turning the thyristor off, but is also of great importance for turning it to the ON state, as only in this way can firing, proceeding from the gate, sufficiently quickly reach all the active parts of the thyristor.

Otherwise there would be unacceptably high switching losses and too low permissible current rise speeds.

It is true that, in this connection, appreciable limitations are imposed due to the requirement for high firing currents for the purpose of uniformly firing the long cathode edge, and also to the need for very small cathode strip widths, If the whole cathode surface is to be fired within one sec, firing currents of the order of magnitude of the forward current in the ON state of the thyristor would be needed, and cathode strip widths of only about 300 llm. However, this would entail the loss of so much active surface, that this expedient does not appear to be suitable in practice.

However, the electrical circuit layout disclosed in Int. J. Electronics 36 (1974), pp.

399-416, is also not suitable for higher blocking voltages, for example those over 1000 V.

This is because high blocking voltages require a thick, high-ohmic base zone of the thyristor. However, such a base zone prevents sufficiently rapid clearing of the charge carriers from the ON state, if - through the incorporation of recombination centres (for example gold doping) - the lifetime of the change carriers is not to be appreciably shortened. However, again, such recombination centres cause an unacceptably high drop in the maximum blocking voltage.

It is the object of the present invention to provide an improved thyristor which permits, with minimal recovery time and small turn-on losses, high forward blocking voltages and large values of the permissible current and voltage rise speeds.

According to the invention, there is provided a thyristor for high-frequency use which comprises an inter-digited emitter-gate structure, a control zone in contact with the gate, and a base zone adjacent the control zone, the base zone comprising two layers of which the layer adjacent the control zone is doped to a lesser extent that the other layer, said thyristor being constructed in such a manner that, in the forward blocking condition, a forward blocking current is developed across the junction between the base zone and the control zone, which blocking current is drawn to the gate, whereby firing of the thyristor can be initiated by changing the gate potential so that the forward blocking current is switched to the emitter.

A two layer base zone is known per se from IEE Conference Publication 123, London, 3 December 1974, pp. 13-19. By means of this expedient it is possible to apprecjably reduce the thickness of the base zone while just avoiding the development of an axially flowing blocking current, so that the forward blocking voltage remains unreduced. the high-ohmic layer of the base zone is then completely emptied of free charge carries at relatively low voltages, while the low-ohmic layer avoids a further extension of the blocking layer and, thus, a "punch-through" (barrier layer contact breakdown). Thus, control-and base-zones together form an approximately PIN-structure. in which a rectangular field distribution is obtained instead of the triangular field distribution.

If, in the known arrangement, the more highly doped layer of the base zone is directly followed by the highly doped second emitter zone - that is to say the anode layer in the case of the N-doped base zone - then the whole structure is particularly simple to manufacture.

However, this structure is only capable of blocking very low voltages in the backward direction, for example 25 V; however. this is not disadvantageous for circuits used in practice, because the thyristor has. in any case. to be shunted, in the reverse blocking direction, by an antiparallel diode (cf. the literature cited, concerning the known circuits, at the outset).

Preferably, in operation of the thyristor according to the invention, in the forward blocking condition the gate of the thyristor has a potential of polarity opposite to the polarity of the majority of carriers in the control zone, whereby firing g can be initiated by changing the gate potential in the forward blocking condition. The thyristor has à doped emitter zone between the emitter electrode and the control zone. Thus, in the forward blocking condition, the junction between the emitter zone and the control zone is biased by the relative gate and emitter potentials to oppose the i flow of forward blocking current, whereby firing can be initiated by changing the gate potential to remove said blocking current-opposing bias.

In a method of operating the thyristor according to the invention, firing is initiated by changing the gate potential to zero or to a polarity opposite to its polarity in the forward blocking condition, or alternatively simply by opening the thyristor gate circuit.

In accordance with this method the thyristor is not (as has hitherto been usual) fired by applying a potential to the gate, which would otherwise be substantially neutral, to initiate a gate current. Instead, the thyristor - which can be fired by break over at a low voltage, with the gate circuit open - is held blocked, in the forward blocking condition, by way of the gate, and is then caused to fire by opening the gate circuit or by reversing the potential at the gate. In this way not only are the advantages, inherent in the thyristor proposed according to the invention and described above, realised, but. in particular an extremely rapid firing of the whole active part of the thyristor is achieved. This is because firing is not triggered now by laterally flowing gate firing currents. but by the axially flowing blocking current, which-is caused by the presence of the space charge layer in the high-ohmic layer of the base zone. The ease with which this thvristor is fired is further assisted by the absence of emitter short-circuits which, although it would be a disadvantge in other circumstances, is conducive to the thyristor proposed according to this invention.

The efficient operation of the thyristor according to the invention was not at all foreseeable, because the blocking current used for firing the thyristor is markedly temperature-dependent. However, it has been found - and this will be further demonstrated in the following examples - that this temperature dependence is not harmful in practice because the generation current at the PN-junction between the control zone and the contiguous emitter zone has, to a large extent, the same temperature-dependence.

The invention is further explained below with reference to embodiments illustrated in the drawing, and also by means of dimensional examples. In the drawing Figure 1 illustrates a circuit, constituted as a load-controlled inverter for generating high frequency, employing thyristors according to the invention.

Figure 2 shows voltages and currents in a circuit according to Figure 1, Figure 3 is a cross section taken through a thyristor according to the invention, along line s-s of Figure 4, with positive slope of the forward blocking PN-junction, Figure 4 is a plan view of the gate side of a thyristor shown in Figure 3, Figure 5 shows the doping profile for a thyristor having N-doped base zone, and Figure 6 shows the doping profile for a thyristor having P-doped base zone.

Figure 1 illustrates a load-controlled inverter, with series-oscillating circuit load RLC, for example a high-frequency induction furnace L with series capacitance C; Figure 1 also shows the thyristor connections Thl, Th., Th3, Th4 which are shunted, in reverse blocking direction, by antiparallel diodes D1, D2, D3, D4. We are here concerned with an intermediate circuit inverter, the d.c. voltage being fed from the d.c. intermediate circuit art H.

As is clear from Figure 2, the voltage UAK is applied to the main connections of each thyristor Thl-Th4. Control voltage UG is applied to the gate connection G, the pulse form of this voltage being shown in Figure 2 for a thyristor having P-control zone and N-base. UG is initially negative, so that the thyristor is blocked. If. at time point Z, the gate voltage UG is brought to zero or to a positive value, for example at the thyristors of switching connections Thl and Th4, this pair of thyristors is fired, and the current IAK flows through Thl and Th4, this current IAK reversing its sign at time point W. This current IAK must then be guided through the antiparallel diodes D1 and D4 until. at time point T, the thyristor pair of switching connections Th2 and Th3 is fired and conducts the current. At this time the voltage UAK once again rises at the thyristors of switching connections Thl and Tub4. For blocking this voltage, the gates G of Th, and Th4 must be restored to negative potential during the time interval W to T.

In this way rectangular a.c. voltage U, is present at the load L, as is also the alternating current IL, phase shifted with respect to this a.c. voltage.

Through the use of thyristor switching connections as described inverters can be equipped with semi-conductor components at frequencies of a magnitude at which it has only been possible to employ electron valves hitherto.

The thyristor shown in cross-section in Figure 3 comprises: a first emitter zone 1; a control zone 2; a base zone with a high-ohmic layer 3a and a low-ohmic layer 3b; a second emitter zone 4 and a first main electrode 5 : a gate electrode 6. and a second main electrode 7. The PN-junctions J1, J2 j3 lie between the zones 1/2. 2/3a, 3by4, The connection to the main electrode 5 is designated as E1, the connection to the main electrode 7 as E2, and the connection to the gate electrode 6 as G.

As the PN-junction J3 does not have to support a blocking voltage, it is possible to provide, at the forward blocking PN-junction J2, the positive slope which, as is known, is particularly favourable (IEEE Trans.ED-11 (1964) 313).

It can be seen from Figure 4 that the thyristor shown in Figure 3 has an interdigitated emitter-gate structure. The strip width of the emitter zone 1 is indicated as b, conveniently between 0.2 and 4 mm, the strip width of control zone 2 as g, conveniently between 0.05 and 2 mm.

Dimensioning examples Relatively high power (dissipation) losses occur in a continuously operating highfrequency thyristor; an operating temperature of 370 degrees K is therefore used as a basis for dimensioning.

Epitaxy N-base thyristor Figure 5 shows the doping pattern (value of doping concentration N plotted as a function of distance x from the end. surface) for a thyristor which is made, by means of epitaxy, on a P ±substrate. This thyristor has the following lavers: P±substrate (corresponding to the second emitter zone 4 in Figure 3): acceptor concentration greater than 5. 10l8 cm -3, say NA = 1019 cm3, the layer thickness is not important for operation, for example it may be 300 llm and in any case greater than 3em.

N-layer (corresponding to layer 3b in Figure 3): donor concentration ND from 5.10'5 to 5.1017 cm-3, say = 2 . 10 16 cm-3, layer thickness from 10 to 60 m, say 30 m.

N--layer (corresponding to layer 3a in Figure 3): donor concentration less than 5.1014 cm-3 say ND = 1.1014 cm-3, alyer thickness from 30 to 200 m, say 70 m.

P-layer (corresponding to control zone 2 in Figure 3): acceptor concentration from 1.1016 to 2.1017 cm-3, say NA = 4 . 1016 cm-3, layer thickness from 5 to 25 m, say 10 m.

B±layer (corresponding to the first emitter zone 1 in Figure 3) with alloyed metallisation (5 in Figure 3): Donor concentration ND in the range . 1018 - 5 . 1019 cm-3, layer thickness greater than 3 m, say 10-20 m. This N±layer can also be produced by a process of diffusion into a correspondingly thicker p-layer.

For reducing the life time of the injected charge carriers, gold is diffused, in a known manner, into the layers 3a, 3b to a concentration of about 6 . 1013 cm-3, so that a life time of 0.5 and 3 sec, say 0.7 sec results.

The voltage Up, at which the whole N-layer 3a is cleared of free charge carriers is equal to 400 V. The maximum field strength for silicon, of 2. 105 V/cm, is reached at the PN' junction J2 at a forward blocking voltage of UB = 1000 V. The breakdown voltage of the NP±junction J3 in the reverse direction is 35 V. The N+P-junction J1 between control zone 2 and emitter zone 1 can support a maximum of 20 V.

The emitter zone injects electrons into the P-base 2, if-the current density over the N+PPjunction J1 is greater than the generation current at this junction. With increasing blocking voltage at junction J2, the blocking current gradually exceeds the generation current of J1 and, as soon as the forward blocking voltage exceed 130 V, the thyristor can be uniformly fired solely by opening the gate circuit. This value is practically independent of temperature, as the blocking current and the generation current at the N+P-- junction J exhibit almost the same temperature-dependence. At 370 degrees K the minimum firing current density amounts to 3 mAl cm2, and the maximum blocking current density amounts to 20 mA/cm2. at a forward voltage of 1000 V.

If homogeneous current density j from the N--region 3a flows into the P-layer 2, and is drawn off, beneath the emitter strips of width b, to the gate, there will be a mixmum voltage drop of U = Jpb2/8d. In this expression d stands for the thickness and # represents the specific resistnce of the P-layer 2. From this it is clear that, in the case of a negative gate voltage of 10 V and of a strip width of 2 mm, a maximum current density of 4 A/cm- can be drawn off, without the emitter zone 1 injecting electrons. This value is 300 times greater than the maximum blocking current density, so that firing is always reliably controlled by the blocking current. The maximum current density which can be drawn off by way of gate G must be so great that, even where the base 3a/3b is flooded about twenty-fold, a dU/dt of 1000 Vljisec can be allowed. The recovery time of this thyristor then still amounts to 1.5 to 2 rise.

The above-described thyristor may for example be operated on the basis of the following data: Periodic peak blocking voltage 800 V Continuous maximum current 100 A Critical voltage gradient 2000 V/jis Critical current gradient 200 A/jis Switching frequency 50 kHz Electrical potential at the gate when blocking occurs: -3 to -20 V, say -10 :V, and (on firing) 0 to + 1 V P-base thyristor Figure 6 illustrates the doping pattern for a P-base thyristor. Manufacturing starts from a silicon slice slice of about 200 m thick, with a P-doping concentration of 2 . 1015 cm-3 (700 # cm). From one side a P-profile of about 70 m depth and a N±zone are diffused. On the other side a N-layer of about 30 m thickness is epitaxially applied with a doping of 4 . 1016 cm-3, into which layer a P±profile is then diffused. Into the whole sample gold is diffused up to a concentration of 6 . 10l3 cmj, so that a life time of 0.7 jisec results.

The p±layer, the N-layer, the P--layer. the P-layer and the N-layer correspond, in the. same order of sequence., to the zones 1. 2, 3a, 3b. 4 of the thyristor shown in Figure 1. The range of layer thicknesses and doping concentrations previously specified remain generally applicable.

With a forward voltage rises further, this blocking layer will of course extend, by a further distance of up to 20 jim, into the following. P-layer 3b. The maximum forward blocking voltage amounts to 2200 V. As in the previous example, the breakdown voltages of the P+N- junction and of the PN±junction (Jl and J3 respectively) amount to 20 and 35 V respectively.

The minimum forward voltage for breakover firing now amounts to 30 V. This reduction is achieved by the greater distance, conseq-uent upon the lesser degree of doping, of the blocking layer at the PN-junction J2. The maximum blocking current density- at a forward voltage of 2200 V and at an operating temperature of 370 degrees K now amounts to 55 mA/cm.

The conductivity in the 10 jim is almost three times as great as in the P-layer 2 of the foregoing example. The emitter strip width b can now therefore be increased, by a factor 3to b = 3.3 mm.

For the thyristor of this example the recovery time will be about 3 rise.

The operating data of the thyristor of this example are, for example, as follows Periodic peak blocking voltage 1500 v Continuous current 100 A Critical voltage gradient 1000 V/S Critical current gradient 200 A/Cls Switching frequency 30 kHz Potential of the gate when blocking occurs = +3 to + 20 V, say 10 V and, when firing occurs = 0. to -lV.

Claims

WHAT WE CLAIM IS:

1. A thyristor for high-frequency use which comprises an interdigitated emitter-gate structure, a control zone in contact with the gate. and a base zone adjacent the control zone, the base zone comprising two layers of which the layer adjacent the control zone is doped to a lesser extent than the other layer, said thyristor being constructed in such a manner that, and having the lifetime of the charge carriers in the base zone selected such that in the forward blocking condition, a forward blocking current is developed across the junction between the base zone and the control zone, which blocking current is drawn to the gate, whereby firing of the thyristor can be initiated by changing the gate potential so that the forward blocking current is switched to the emitter.

2. A thyristor according to claim 1, wherein, in the forward blocking condition, the gate has a potential of polarity opposite to the polarity of the majority of carriers in the control zone, whereby firing can be initiated by changing the gage potential to zero or to a potential of polarity opposite to that of the gate potential in the forward blocking condition.

3. A thyristor according to claim 1 or claim 2. having a doped emitter zone between the emitter and the control zone, and wherein, in the forward blocking condition, the junction between the emitter zone and the control zone is biased by the relative gate and emitter potentials to oppose the flow of forward blocking current. whereby firing can be initiated by changing the gate potential to remove said blocking current-opposing bias.

4. A thyristor according to claim 3, wherein said control zone does not have any shorts short-circuiting the junction between the emitter and control zones.

5. A thyristor according to any preceding claim, wherein the base zone of the thyristor is N-doped or P-doped, and the layer of the base zone lying adjacent to the control has a layer thickness of 30 to 200 jim and a doping concentration of less than 5 . 1014 cam~3, the other layer of the base zone has a thickness of 10 to 60 jim and a doping concentration of 5. 10l5 to 5. 1017 cm3 or a doping profile which rises to this concentration; the control zone has a layer thickness of 5 to 25 jim and a mean doping concentration of 1 . 1016 to 2. 1017 cam~3 ; the two emitter zones have a layer thickness greater than 3 jim and a doping concentration greater than 5 . 10J8 cam~3 ; the charge carrier lifetime in the first mentioned layer of the base zone is adjusted to a value of 0.5 to 3 Rises; and the emitter-gate structure is of strip-like shape, the emitter strip widths being between 0.2 and 4 mm. and the gate strip widths being between 0.05 and 2 mm.

6. A method of operating a thyristor according to any preceding claim, wherein firing is initiated by changing the gate potential to zero or to a polarity opposite to its polarity in the forward blocking condition.

7. A method of operating a thyristor according to any of claims 1 to 5, wherein firing is initiated by opening the thyristor gate circuit.

8. A method according to claim 6 or claim 7, in which the circuit includes a thyristor with N-doped base zone, and wherein the potential of the gate in the forward blocking condition lies between -3 and -20 V.

9. A method according to claim 6 or claim 7. in which the circuit arrangement includes a thyristor with P-doped base zone, and wherein the potential of the gate in the forward blocking condition lies between +3 and +20 V.

10. A thyristor according to claim 1 substantially as hereinbefore described with reference to and as shown in Figures 2 and 4 of the accompanying drawings.

11. A method according to claim 6 substantially as hereinbefore described with reference to Figures 1 and 2 of the accompanying drawings.

12. A circuit comprising a thyristor as claimed in any of claims 1 to 5, said circuit being adapted to operate the thyristor in accordance with the method of any of claims 6 to 9.