WO1999046821A2

WO1999046821A2 - Semiconductor switch devices and their manufacture

Info

Publication number: WO1999046821A2
Application number: PCT/IB1999/000202
Authority: WO
Inventors: Holger Schligtenhorst; Godefridus A. M. Hurkx; Andrew M. Warwick
Original assignee: Koninklijke Philips Electronics N.V.; Philips Ab
Priority date: 1998-02-28
Filing date: 1999-02-04
Publication date: 1999-09-16
Also published as: DE69940757D1; GB9804177D0; WO1999046821A3; EP0979530A2; JP2001523401A; US20010054747A1; EP0979530B1; US6355971B2

Abstract

In a semiconductor switch device such as an NPN transistor (T) or a power switching diode (D), a multiple-zone first region (1) of one conductivity type forms a switchable p-n junction (12) with a second region (2) of opposite conductivity type. In accordance with the invention, this first region (1) includes three distinct zones, namely a low-doped zone (23), a high-doped zone (25), and an intermediate additional zone (24). The low-doped zone (23) is provided by a semiconductor body portion (11) having a substantially uniform p-type doping concentration (P-) and forms the p-n junction (12) with the second region (2). The distinct additional zone (24) is present between the low-doped zone (23) and the high-doped zone (25). This triple-zone formation for the first region (1) permits an improvement in switching behaviour, e.g. in terms of fall-time and energy dissipation during turn-off of the device (T, D). A very low doping (P-) can be used for low-doped zone (23) so that, in the off-state of the device (T, D), this zone (23) and also the additional zone (24) can be fully depleted. The additional zone (24) having its additional doping concentration provides a path for extracting residual charge carriers from the low-doped zone (23) when the device (T, D) is being switched off.

Description

DESCRIPTION

SEMICONDUCTOR SWITCH DEVICES AND THEIR MANUFACTURE

This invention relates to semiconductor switch devices (e.g. an NPN switching transistor or a power switching diode) having a switchable reverse- biased p-n junction between a first region (such as a p-type base of the transistor or of the diode) of a first conductivity type and a second region (such as a collector or cathode) of an opposite second conductivity type. The invention further relates to methods of manufacturing such a semiconductor switch device.

United States patent specifications US-A-3,507,714 and US-A-3,710,203 disclose respectively a power switching transistor and a power switching diode, each of which has a p-type base ( "first" region) forming a switchable p-n junction with a second region of an opposite second conductivity type. The second region is an n-type collector region in US-A-3,507,714 and an n-type cathode region in US-A-3,710,203. In both the transistor and the diode, the first region includes a high-doped zone having a higher doping concentration of the first conductivity type than a low-doped zone of the first region. In the manufacture of these switch devices, a semiconductor body portion having a substantially uniform doping concentration of the first conductivity type is provided to form the low-doped zone adjacent to the p-n junction, the body portion having a surface located opposite the p-n junction; and a doping step is carried out by doping the body portion over a part of its thickness with dopant characteristic of p-type conductivity through the surface so as to provide the high-doped zone with a doping concentration which decreases towards the low-doped zone. The whole contents of US-A-3,507,714 and US-A-3,710,203 are hereby incorporated herein, as reference material.

In the method disclosed in US-A-3,507,714 the NPN transistor has its base region formed by a uniformly low-doped p-type body portion (wafer) into which the p-type high-doped zone (and also n-type emitter and collector regions) are diffused. This method dates to the late 1960s, i.e. several decades ago. Nowadays it is more conventional to form an NPN transistor from a uniformly low-doped n-type body portion (epitaxial layer) into which the p-type base region and n-type emitter region are diffused. United States patent specification US-A-4, 805, 004 discloses a variant NPN transistor in which the p-type base region comprises a uniformly low-doped p-type body portion (epitaxial layer) on a uniformly low-doped n-type body portion (epitaxial layer) of the collector region. The whole contents of US-A-4,805,004 are also hereby incorporated herein, as reference material.

It is an aim of the present invention to change the design and manufacture of semiconductor switch devices (such as, e.g. NPN switching transistors and power switching diodes) so as to permit an improvement of their switching behaviour, e.g. in terms of fall-time and energy dissipation during turn-off of the device.

According to one aspect of the present invention, there is provided a semiconductor switch device as set out in Claim 1.

According to another aspect of the present invention, there is provided a method of manufacture as set out in Claim 7. As described hereafter in more detail, the inventors have discovered that the switching behaviour can be significantly improved by providing an additional zone with a doping concentration in accordance with the present invention, between the low-doped and high-doped zones of the first region. When the semiconductor switch device is being switched off, it appears that this additional zone provides a low-resistance path for extracting the remaining plasma which is mainly present in the middle of the uniformly low-doped zone towards the end of the discharge period. A significant reduction in fall-time and energy dissipation can be achieved by this means.

The first and second regions may be respective anode and cathode regions of a switching diode. Thus, the high-doped zone of the first region may be provided with an anode contact at the surface, and the second region may be provided with a cathode contact at a surface of the cathode region opposite the anode region.

The first and second regions may be respective base and collector regions of a bipolar transistor. Thus, the body portion of the first conductivity type may be overdoped over a part of its thickness with dopant characteristic of the second conductivity type adjacent to the surface to provide the transistor with an emitter region which forms a p-n junction with the high-doped zone of the base region. The emitter region and high-doped zone of the base region may be provided with respective emitter and base contacts at the surface. The second region may be provided with a collector contact at a surface of the collector region opposite the emitter and base regions.

The present invention is particularly advangeous for so-called "p-base" switching devices, i.e. in which the conductivity type of the first region (and hence its low-doped zone and additional zone) is p-type. The minority charge carriers in a p-type region are electrons which have a high mobility, and so a p-base device constructed in accordance with the invention can have fast switching characteristics. Holes have a much lower mobility than electrons, and so the provision of the low-doped zone on the p-type side of the p-n junction and the provision of the additional zone between the low-doped and high-doped zones of the p-type region permits efficient removal of the holes when switching off the p-base device. Thus, the first region is preferably p-type, and the body portion may be typically of p-type silicon. Boron is a well-established p-type dopant in silicon, and several known boron doping technologies are suitable for providing the desired doping profiles of the low-doped zone, the additional zone and the high-doped zone. Phosphorus and/or arsenic are suitable n-type dopants for regions of opposite conductivity type in silicon. The starting material may be a uniformally boron doped silicon substrate. The switch device may be manufactured without a need for epitaxial growth, by locally doping such a substrate material (a remaining portion of which forms the uniformally low-doped zone) with the respective dopants to form the various regions and zones.

Typically the maximum doping concentration of the additional zone of the first region is one or two orders of magnitude lower than the doping concentration of the high-doped zone. It may be at least an order of magnitude higher than the doping concentration of the low-doped zone. Generally the low-doped zone of the first region has such a low doping concentration as to be depleted by a depletion layer which extends, when the switch device is operated in an off state, from the reverse-biased p-n junction across the thickness of the low-doped zone and into the additional zone of the first region. Indeed, both the additional zone and the low-doped zone may be fully depleted when the switch device is operated in its off state close to its breakdown voltage. In this manner a high blocking voltage can be obtained in the off state of the switch device, while still retaining fast turn-off due to the inclusion of the additional zone. The thickness of the additional zone with its additional doping concentration may be maximised to reduce the voltage drop therein, and the thickness of the low-doped zone may be maximised to increase the breakdown voltage together with the decrease in doping concentration of the additional zone towards the low-doped zone. Typically the thickness of the low-doped zone and the additional zone are of the same order and typically greater than that of the higher-doped zone. Depending on the type of switching device, the conductivity type determining dopant concentration of the high-doped zone may typically exceed 10¹⁷ cm ³, whereas that of the additional zone may be between 10¹⁴ cm^"3 and 5x10¹⁵ cm^"3. Typically, the conductivity type determining dopant concentration of the uniformally low-doped zone may be at most 5x10¹⁴ cm^"3 and is preferably less than 10¹⁴ cm^'3.

The multiple zone structure of the first region in accordance with the present invention may be combined with various measures to control the electric field in a peripheral area of the device, in the off-state of the device. Thus, for example, extension zones may be provided at both sides of the reverse-biased p-n junction so as to extend towards each other at a surface of the body in a manner similar to that disclosed in US-A-4,805,004. An advantageous different measure is described below, in which the p-n junction is extended to the surface adjoined by the first region, and an annular field-relief region of the second conductivity type is present in the low-doped zone of the first region at a location between the additional zone of the first region and the extension of the p-n junction at said surface, so as to control the spread of the depletion layer from the p-n junction.

These and other features in accordance with the present invention will now be illustrated in embodiments of the present invention, now to be described with reference to the accompanying drawings, in which:

Figure 1 is a cross-sectional view of an active and peripheral area of one particular example of a NPN bipolar switching transistor in accordance with the invention, manufactured by a method in accordance with the invention;

Figure 2 is a plot of doping profile (N++. P++. P+. P-, N++) and hole dissipation (at times tO, t1 , t2, t3 and t4) in cm^"3 as a function of depth x in μm (micrometres) on the section line ll-ll in Figure 1, vertically through the active area of the emitter, base and collector regions of the transistor;

Figure 3 is a plot of energy dissipation E in microJoules (μJ) as a function of base current l_B0 in Amperes (A) for the transistor of Figure 1 (case C) and for three comparative prior-art NPN transistor structures (cases A, B and D);

Figure 4 is a schematic representation of the dc output characteristics of such a transistor, in terms of the variation of its collector current l_c with collector- emitter voltage V_CE , for three values of base current l_B1, l_B2, and l_B3 ;

Figure 5 is a circuit diagram of the NPN bipolar switching transistor and a HV (high voltage) switching diode in a horizontal deflection circuit for a television display, for simulating the switching off of the transistor; Figure 6 is a plot of the switching characteristics of the NPN bipolar switching transistor of Figure 2 in terms of the variation of collector and base currents l_c and l_B in Amperes (A) and collector-emitter voltage V_CE in volts (V), with time t in μs (microseconds) in the Figure 5 circuit;

Figure 7 is a plot (similar to Figure 2) of a doping profile (N++. P++. P+. N-, N++) and hole dissipation (at times tO, t1 , t2, t3 and t4) in cm^"3 as a function of depth x in μm (micrometres) for a prior-art variant of the Figure 1 transistor, corresponding to case B of Figure 3;

Figure 8 is a plot (similar to Figure 6) of the switching characteristics of the prior-art variant transistor of Figure 7 (case B of Figure 3) in terms of current variations l_c and l_B in Amperes (A) with time t in μs (microseconds) in the Figure 5 circuit; and

Figure 9 is a cross-sectional view of the transistor body of Figure 1 , at a stage during manufacture by a method in accordance with the present invention.

It should be noted that Figures 1 , 4 and 9 are diagrammatic and not drawn to scale. Relative dimensions and proportions of parts of Figures 1 and 9 have been shown exaggerated or reduced in size, for the sake of clarity and convenience in the drawings. The plots of Figures 2,3, 6, 7 and 8 are derived from a computer model for NPN bipolar switching transistors. The same reference signs are generally used to refer to corresponding or similar features in different embodiments.

The NPN bipolar switching transistor T of Figures 1 and 2 is an example of a semiconductor switch device having a switchable reverse-biased p-n junction 12 in a monocrystalline silicon semiconductor body 10, between a multiple-zone first region 1 (23,24,25) of p-type conductivity type and a second region 2 (20) of n-type (i.e. of an opposite second conductivity type). In the transistor embodiment of Figures 1 and 2, the p-n junction 12 is a collector junction, the first region 1 (23,24,25) is a p-type base region, and the second region 2 (20) is an n-type collector region of the transistor T. An n-type emitter region 3 is provided in the base region 1 to form an emitter junction 13 of the transistor. The base region 1 and emitter region 3 are contacted by base and emitter contacts 31 and 33 respectively at a major surface 18 of the body 10. In a specific examplary embodiment, the emitter region 3 may have a known finger-type layout geometry, with the base and emitter contacts 31 and 33 comprising interdigited fingers. The p-n junction 12 extends parallel to the surface 18, and so the junction 12 and surface 18 are located opposite each other across the base region 1. The collector region 2 is a high-doped n-type region 20 which adjoins the opposite major surface 19 of the body 10 where it is contacted by a collector contact 32.

The fundamentals for design, characteristics and operation of NPN bipolar switching transistors were established long ago. An early but convenient overview is given in the Basic Transistor Engineering - Switching Characteristics of Transistors section 4-8 of Integrated Circuits Design Principles and Fabrication, Editor: Raymond M. Warner, published 1965 by the McGraw-Hill Book Company, and the whole contents of which are hereby incorporated herein as reference material. The transistor T is switched on and off by a control signal applied to its base contact 31. In the off-state, both the emitter and collector junctions 13 and 12 are reverse biased by this control signal. When the transistor is switched into saturation in the on-state, both the emitter and collector junctions 13 and 12 of T become forward biased, due to a high density of minority carriers (electrons) injected into the p-type base region 1. In accordance with the present invention, the base region 1 of the NPN transistor of Figures 1 and 2 includes three distinct zones 23,24,25, namely: a low-doped zone 23, a high-doped zone 25, and an additional zone 24. The low-doped zone 23 is provided by a semiconductor body portion 11 having a substantially uniform p-type doping concentration P-. This zone 23 forms the p-n junction 12 with the collector region 2(20). The other distinct zones 24 and 25 are sequentially present between the low-doped zone 23 and the major surface 18, in the active area of the device. The high-doped zone 25 has a p-type doping concentration P++ which is higher than that of the low-doped zone 23 and which decreases towards the low-doped zone 23 over a part of the thickness of the body 10. The distinct additional zone 24 has an additional p-type doping concentration P+ which is lower than the p-type doping concentration P++ of the high-doped zone 25 and which decreases towards the low-doped zone 23.

This triple-zone formation of the base region 1 permits the use of a very low doping P- for low-doped zone 23 so that, in the off-state of the transistor T, the zone 23 can be fully depleted by a depletion layer which extends from the reverse-biased collector junction 12 across the thickness of the low-doped zone 23 and into the additional zone 24 of the base region 1. Preferably the zone 24 is also fully depleted by this depletion layer, when the transistor T is operated near breakdown. Thus, in the off-state of T, the depletion layer from the reverse-biased collector junction 12 (which also extends marginally into the more highly doped collector region 2) may be substantially co-extensive with the zones 23 and 24 in Figure 1 ; for this reason the outline of this depletion layer is not depicted in Figure 1 , but the zones 23 and 24 are depicted without hatching in Figure 1. As described hereinafter, the additional zone 24 having its additional doping concentration provides a path for extracting residual charge carriers from the low-doped zone 23 when the transistor switch is being switched off.

Figure 5 is a circuit diagram for simulating the switching behaviour of the PNP transistor T in a horizontal deflection circuit for a television display. The deflection coil of the display is coupled to the collector (contact 32) of T, and this coil is represented by inductor L_c in Figure 5. The transistor T in a common-emitter configuration is switched by a pulse which is applied to its base (contact 31). This base pulse is represented for the Figure 5 simulation by switching between a base current l_B0 in the on-state of T and a bias voltage V_BB in the off-state of T. The switch together with an inductor L_B represents the base drive circuit of T in the simulation.

The circuit produces a sawtooth type of current l_c through the deflection coil L_c. The transistor T conducts the positive current l_c , and the negative part of the current is conducted by a PN switching diode D which is coupled between the base and emitter ( i.e. between contacts 31 and 33) of T. Capacitor C forms an LC oscillator with L_c when both T and D are in the OFF state. This LC oscillator drives V_CE to negative values such that D is switched on. For simulating the switching behaviour of T, the various components of the Figure 5 circuit have the following values: V_cc = 150V

C = 13nF

L_B = 6μH

V_BB = -4 V

The diode D is in the off state when the transistor T is switched off, and so the characteristics of D need not be taken into account in simulating the switching off of T.

The behaviour of the base current l_B, collector current l_c and collector voltage V_CE of T during turn off is shown in Figure 6. The time interval t_s (the so-called "storage time") is the interval from when l_B starts to decrease at tO until l_c starts to decrease from its maximum value l_CM. This interval t_s is related to charge-storage phenomenon in the base and collector regions 1 and 2. The time interval t_f is the so-called "fall time" and is the subsequent time interval over which l_c decreases until T enters the off-state. Typically t_f may be defined as the interval during which l_CM falls from 90% of l_CM to 10% of l_CM. The slope of l_c during t_s is determined by the voltage drop across the collector inductor L_c, which initially is virtually equal to V_CE . The slope of l_c during t, is dependent on the base-collector region structure, as illustrated later in the comparison of Figures 6 and 8.

The time sequence of Figure 6 can be understood as follows: • at t = tO, the base is switched from l_B0 to V_BB , and the base starts being discharged; the base current l_B starts to fall. at t = tO + t_s , excess carriers are removed from the base region 1 , and the emitter junction 13 is no longer forward-biased, and so the collector current l_c is at l_CM. Figure 6 (and Figure 2) indicate the slightly later time t1 , when l_c is starting to decrease from l_CM. Due to base inductance L_B, the emitter junction 13 is entering avalanche breakdown. The base current l_B falls to a minimum (reached at t2), but after t = t2: part of the base current is now provided by the avalanche process, and so l_B now increases. The slope of the now-increasing l_B is determined by the base-emitter breakdown voltage BV_EB0, i.e. dl_B/dt = (BV_EB0 - V_BB)/L_B. V_CE is now rising as the collector current l_c falls (the base-emitter junction 13 being no longer forward biased). The behaviour of l_c and V_CE in this region is determined by L_c and capacitor C. the fall time t_f is determined by the stored excess carriers at t=(t0 + t_s) in the base and collector regions 1 and 2, and particularly the holes with their lower mobility. because V_CE increases rapidly during the fall of l_c , power is dissipated during this switching cycle.

This power dissipation during t_f is a main issue which the present invention addresses. For faster switching circuits, the power dissipation would rise if the energy dissipation during switching were to remain the same. Thus, in order to keep the power dissipation low for faster circuits, the energy dissipation per circle has to be reduced.

The simulated energy dissipation during turn-off of a transistor T in the Figure 5 circuit is indicated in the plots of Figure 3 for different NPN transistor structures, for which l_CM is the same (l_CM = 4.5A) in each case A, B C and D. Due to its shape, this type of plot is called a "bath-tub" curve. The increase in energy dissipation to the right-hand side (higher l_B0 ) of the minimum is caused by the discharge process of excess carriers during the fall time t_f. The responsible effects contributing to the energy dissipation in the different regions of the bath-tub curve can be explained as follows, with reference also to the three l_c - V_CE curves of Figure 4: the left-hand side of the bath-tub curve ( for values of l_B0 which are lower than l_B0 at minimum dissipation) relates to switching the transistor from an under-driven on-state in which the base current l_B0 is low (such as from l_B1 in Figure 4). In this case, the under-driven transistor T is out of saturation (region ST of Figure 4) before switching off (i.e. t<t_s ); a high value of V_CE is needed to reach a required maximum collector current l_CM. This leads to a high dissipation before t = tO + t_s. The required value of V_CE ( and hence the dissipation) increases with decreasing l_B0 and with increasing l_CM. At these high values of V_CE the transistor is out of saturation and quasi-saturation, and few excess carriers are present in the base and collector regions 1 and 2. Hence the fall time t, is short and the energy dissipation which occurs during that time is low. the right-hand side of the bath-tub curve ( for values of l_B0 which are higher than l_B0 at minimum dissipation) relates to switching the transistor T from an over-driven on-state in which it is heavily in saturation (such as from curve l_B3 in Figure 4). In this case, the over-driven transistor T does not come out of saturation (region ST of Figure 4) to reach the required l_CM. Hence very many excess carriers are present in the base and collector regions 1 and 2, and so t_f is long and the energy dissipation which occurs during that time is high. The turn-off time at a given l_CM increases with increasing l_B0, because for higher l_B0 the transistor is in even stronger saturation. For a given l_B0, V_CE is lower for lower l_CM, and so the transistor is in stronger saturation for lower l_CM and, hence, t_f is larger. Thus, contrary to the situation at low l_B0, the dissipation at a given l_B0 increases with decreasing l_CM. the minimum of the bath-tub curve lies at a value for l_B0 which is just high enough to bring the transistor out of saturation at l_c = l_CM (i.e. around curve l_B2 in Figure 4). In this case, no high V_CE is needed to reach l_CM, and t_f is short because the transistor is just out of saturation. The best turn-off performance (i.e. with least energy dissipation) for a

NPN transistor is obtained by forming the transistor as in Figures 1 and 2 in accordance with the present invention, on uniformly p-type doped substrate material 11 (for base zone 23) together with the two higher doped (double-diffused) base zones 24 and 25, i.e. case C in Figure 3. The variation in doping profile for this case is illustrated by the particular example shown in Figure 2 for two boron diffusions in accordance with Figure 9. The base region 1 comprises three distinct zones 23, 24 and 25, and so its doping profile exhibits distinct transitions in doping gradient from zone 25 to zone 24 and from zone 24 to zone 23. Thus, in the example of Figure 2, the boron doping P++ of zone 25 decreases progressively from about 3x10¹⁷ cm^"3 at the emitter junction 13, the boron doping P+ of zone 24 decreases progressively from about 1.5x10¹⁵ cm^"3 at its interface with zone 25, and zone 23 has a uniform boron doping of about 4x10¹³ cm^"3. In the particular example illustrated in Figure 2 and forming the basis for the simulation model, the emitter junction 13 is at a depth x of about 9 μm, the interface between zones 25 and 24 is at a depth x of about 25 μm, the interface between zones 25 and 24 is at a depth x of about 55 μm, and the interface between zone 23 and region 2 (i.e. the collector junction 12 of Figure 1) is at a depth x of about 140 μm from the surface 18.

The Figure 1 transistor structure having this Figure 2 doping profile in accordance with the invention is contrasted with three prior-art types of NPN transistor structure, namely cases A, B and D. In case A, the additional zone 24 is omitted and so the p-type high-doped zone 25 interfaces directly with the uniformly p-type doped zone 23. In case B, a uniformly n-type doped (4x10¹³ cm ³) substrate zone 22 is adopted instead of the p-type substrate zone 23 of Figure 1 , i.e. the transistor base consists of p-type zones 24 and 25 and the transistor collector consists of n-type zones 22 and 20. In case D, the transistor base consists only of p-type zone 25, and the transistor collector consists of n-type zones 22 and 20 (i.e. the same collector as in case B in which a uniformly n-type doped (4x10¹³ cm ³) substrate zone 22 is adopted instead of the p-type substrate zone 23 of Figure 1). The buried collector junction depth in each case A, B, C, and D is so chosen that the transistors in all four cases have the same BV_CB0.

From Figure 3 it can be seen that the turn-off behaviour in terms of energy dissipation is significantly improved by having the triple base zones 23, 24 and 25 in accordance with the invention (case C in Figure 3), as compared with all the other cases A, B and case D. The minimum of the energy dissipation curve for case C is found to be less than for the prior-art structures (cases A and B) and is shifted somewhat to higher base currents l_B0. The dissipation curve for case C is also clearly widened for high base currents l_B0, which means that, for high base currents, the energy dissipation during one switching cycle is less dependent on the actual magnitude of the base current. For the Figure 1 transistor in accordance with the invention, i.e. for case C of Figure 3, Figure 2 shows (in addition to the doping profile) the hole concentrations at the different times tO, t1 , t2, t3 and t4 of Figure 6. It is important to note that the base zones 23 and 24 are fully flooded with electron-hole plasma when the transistor is in its on-state. Thus, these zones 23 and 24 should not usually be made thicker than needed to meet the specified breakdown voltage, because an increase in their thickness increases the amount of stored charge. Another important point is that because the holes have a much lower mobility than electrons, the voltage drop is built up at the side of the low-doped zone nearest the p-type contact zone 25. This is a general property of the discharge of power p-n junction devices with a low-doped zone (e.g. 23 and/or 22) adjacent to the p-n junction. Furthermore the hole current is highest around t = (tO + t_s) and decreases rapidly to zero between t = (tO + ts) and t = (tO + t_s + t_f) , as can be seen from Figures 2 and 6. After the emitter junction 13 has become reverse biased, the hole current then equals the collector current. The difference between the hole extraction current and the collector current is delivered by avalanche multiplication in the emitter junction 13.

Figures 2 and 6 relate to the invention case C, but similar plots are given in Figures 7 and 8 for the prior-art case B. In both cases the transistor was switched off from a base current of l_B=3A. At t=tO, the stored charge is of the same magnitude for both cases B and C. The carrier distributions for both cases are also still similar at t=t1 , comparing zones 23 and 24 of case C with zones 22 and 24 of case B.

The discharge process for both cases B and C can be followed from Figures 2 and 7. As long as the p-type base region is being discharged, the associated voltage drop is given by the multiple of the hole current and the series resistance of the empty part of the base. Because the extraction current is highest at t = (tO + t_s) and shortly afterwards, the voltage drop is kept low by making the doping level of base zone 24 to have its high value near the base contact zone 25. When the extraction process progresses the extraction current decreases, and so the doping level can decrease towards the zone 23 in case C (and towards zone 22 in case B). This decreasing doping level is advantageous in terms of the depletion of both zones 23 and 24 in the off-state of the transistor, i.e. the breakdown voltage.

After time t2 the left-hand edge of the electron-hole plasma reaches the N- zone 22 in the prior-art case B transistor. Further discharge of the case B transistor now separates the electron-hole plasma from the p-type base region 24,25, and so a depletion layer is built up at its collector-base junction 12'. The case B transistor therefore has the problem that the remaining holes have to be removed through this depletion layer. The associated voltage drop (e.g. at times t3 and t4) in the case B transistor is given by the width of this depletion layer (which increases as the discharge progresses) and the space charge in the depletion layer. In the case C transistor in accordance with the invention, the high specific resistance of the P- zone 23 causes a voltage drop when the left-hand side of the electron-hole plasma enters the zone 23 after t2, because the holes are removed through the neutral part of the zone 23. Although it increases with further discharge, this voltage drop for the case C transistor is much lower than for the case B transistor for a given location of the left-hand edge of the plasma from zone 24, i.e. for a given amount of remaining holes.

This major difference in the voltage drop between cases B and C can be understood in terms of the ionized donor dopant in the depletion layer of case B having a positive charge which adds to the positive charge of the holes to be extracted, whereas the ionized acceptor dopant in the depletion layer of case C has a negative charge which compensates the positive charge of the holes to be extracted. This results in a shorter recovery time and a reduced energy dissipation for the case C transistor in accordance with the invention. A comparison of the fall-off of l_c in Figures 8 and 6 shows the much longer fall time t_f (longer than t4) for the prior-art case B transistor as compared with the case C transistor in accordance with the invention.

The discharge process in the prior-art transistor cases A and D can be followed as respective modifications of Figures 2 and 7, while omitting the role of the additional zone 24 in their respective base regions 25,23 and 25. The double diffused base profile (zones 24 and 25) which is adopted for transistors in accordance with the present invention (together with a low-doped p-type P- substrate base zone 23) is extremely beneficial in terms of its influence on the base series-resistance vertically across the thickness of the base region 1. Thus, the somewhat higher doping arising from the inclusion of zone 24 of the double-diffused base profile (zones 24 and 25) reduces the vertical series- resistance, when the remaining plasma during the discharge process is mainly present in the middle of the low-doped base zone 23 (at t2 and afterwards), i.e. when the discharge current has to be maintained by charge carriers in the outer zones 24 and 25 of the base region 1.

By comparing the energy dissipation curves in Figure 3, it can be seen that the combination of a double-diffused p-type base profile (zones 24 and 25) with a low-doped p-type (P-) substrate zone 23 (i.e. case C in accordance with the invention) is beneficial for the transistor switching behaviour. The improvement over prior-art cases A and D (as well as that over case B) can also be seen from the bath-tub curves of Figure 3.

Case A in Figure 3 shows that a comparable transistor performance cannot be achieved with a low-doped p-type (P-) substrate (zone 23) without the additional zone 24 of the double-diffused base profile. In case A the zone 24 is omitted, and so the base region 1 consists only of the two zones 23 and 25. Case D in Figure 3 shows that a comparable transistor performance cannot be achieved with a low-doped n-type (N-) substrate (zone 22) without the additional zone 24 of the double-diffused base profile. In case D the zone 24 is omitted, and so the base region 1 consists only of the zone 25. When switching from high l_B0 , the energy dissipation for prior-art case A based on a P- substrate is comparable to the prior-art case B transistor based on an N- substrate. However, cases A and B differ at low l_B0 and at the minimum dissipation. The vertical base resistance for the case A transistor (without the additional base zone 24) is higher than for the double-diffused base cases B and C. Furthermore Figure 3 compares curves for the prior-art case A transistor on P- substrate with a prior-art transistor of case D which has an N- substrate with a shallow P++ base region 25 only. It can be seen from this comparison that the transistor behaviour is improved by simply using a p-type (P-) substrate instead of an n-type (N-) substrate. However, the case C curve shows how this improvement is surpassed by adopting a transistor structure in accordance with the invention, with a double-diffused base structure 24,25 on a p-type (P-) substrate 23.

Figure 9 illustrates base-doping steps in the manufacture of the NPN transistor of Figures 1 and 2. Preferably, the starting material for this Figure 9 embodiment is a silicon semiconductor body in the form of a very lowly doped p- type substrate 10' having, for example, a uniform boron doping of the order of 10¹³cm^"3. At least a part of the collector region 2 is formed by overdoping a portion of the p-type substrate 10' to form the p-n junction 12 with the remaining p-type body portion 11. This permits the low-doped zone 23 of the base region 1 to be formed by a remaining very lowly doped p-type body portion 11 of the substrate 10' Thus, phosphorus or arsenic may be diffused into the whole of the back surface 19 of this p-type substrate 10' to provide the highly doped n-type region 20 for the collector region 2 of the transistor. This n-type dopant diffusion may be carried out at opposite major surfaces of a semiconductor wafer which is subsequently cut (for example by sawing and then polishing) parallel to these major surfaces so as to form two substrates 10' each having the high-doped n-type region 20 at one major surface 19 and the remaining low-doped p-type body portion 11 (for zone 23) at its newly-exposed, opposite major surface 18. Thus, no epitaxial growth process is necessary.

As illustrated schematically by arrows 44 and 45 in Figure 4, localised doping of the remaining low-doped p-type body portion is now effected through the major surface 18, over a part of the thickness of the body portion 11. This localised doping consists of two doping steps, one to form the high-doped zone 25 and the other to form the intermediately-doped, deeper zone 24. These two doping steps (each of which may use boron as the dopant) can be carried out by known techniques of implantation and/or deposition and then thermal diffusion. Both doping steps can be carried out in known manner using, for example, a silicon dioxide layer pattern 41 as a mask on the major surface 18. The doping 45 for the zone 25 may be done before or after the doping 44 for the zone 24, and/or both may be diffused simultaneously from previous implants. The zones 23 and 24 are so constructed that together they have a width and doping level such that both are fully depleted at voltages approaching breakdown. While avoiding an excess thickness (which would unnecessarily increase the stored charge and the resistance of the current path), the zone 24 with its decreasing doping concentration is made as thick as possible. Typically the thicknesses of zones 23 and 24 are of the same order. The maximum doping level of zone 24 is typically one or two orders of magnitude lower than the doping concentration arising from the adjacent high-doped zone 25 and one order of magnitude higher than the doping concentration of zone 23.

In order to passivate the periphery of the collector junction 12 at the major surface 18, a groove 40 may be etched into the p-type body portion from the surface 18, with an annular geometry so that the groove 40 surrounds the base region 1. The p-type body portion 11 (which forms the zone 23) may then be peripherally overdoped along the side-walls of the groove 40 with n-type dopant (for example arsenic) to form an n-type annular region 42 which surrounds the p-type base zone 23 and reaches to the collector region 2 so as to extend the p-n junction 12 to the surface 18. The same doping step that is used to form the annular region 42 may be used to provide simultaneously the emitter region 3 which forms the p-n junction 13 with the high-doped zone 25 of the base region 1.

Simultaneously with the formation of the regions 3 and 42, this doping step may also be used to provide one or more annular field-relief regions 43 in the low-doped zone 23 at a location between the zone 24 and the collector junction extension at the suface 18. These field-relief regions 43, spaced from the zone 24, may be of an annular configuration surrounding the zone 24 at the surface 18, and they are located within the spread of the depletion layer in the zone 23 from the extended collector junction 12. The major surface 18 and the side-walls of the groove 40 can be passivated subsequently with an insulating layer 48 of, for example, silicon dioxide. The emitter region 3 and the high- doped zone 25 of the base region 1 are contacted at windows in the insulating layer 48 by a metallisation pattern (for example of aluminium) to provide the base and emitter contacts 31 and 33. The annular groove 40 surrounds a large active area of the device where the transistor structure is present. Figures 1 and 4 illustrate only a small, peripheral area of the transistor T. Generally a power switching transistor T has a large number of emitter regions 3 (for example in a matrix or finger pattern) diffused in one or more common base regions 1 and may have, for example, interdigitated emitter and base contacts 33 and 31. In accordance with the present invention the or each common base region 1 of such a transistor T can comprise the vertical multiple-zone structure 23,24,25 of Figures 1 and 2.

Many modifications and variations are possible within the scope of the present invention. Thus, for example, epitaxial growth could be used. Instead of starting with a p-type substrate 10', the low-doped p-type body portion 11 (for zone 23) may be provided by a p-type epitaxial layer which is deposited on the collector region 2. In the embodiment of Figure 1 , the collector region 2 consists solely of the high-doped n-type region 20. However, an NPN transistor in accordance with the present invention may comprise a two-part collector region 2 which comprises, for example, a low-doped n-type zone 22 between the low- doped p-type zone 23 of the base region 1 and the high-doped n-type region 20 of the collector region 2. Thus, for example, the low-doped base zone 23 may be a p-type epitaxial layer 11' deposited on an n-type epitaxial layer 22' deposited on a high-doped n-type substrate 20. There is then a degree of freedom in the choice of the relative thicknesses (widths) W_p and W_n of the respective p-type and n-type low-doped zones 23 and 22 provided by the two epitaxial layers. However, because the minimum of the energy dissipation curve C in Figure 3 is lowest for the transistor case C with the p-type (P-) substrate zone 23 (rather than for case B with the n-type (N-) substrate zone 22), it is expected that the minimum energy dissipation for a correct optimisation will be found for a NPN transistor as in Figures 1 and 2, in which W_p is maximised and W_n is zero.

The device embodiment of Figure 1 is an NPN switching transistor. However, the present invention may be used to form other types of semiconductor switch device, for example a HV power switching diode. In this case, the p-type and n-type regions 1 and 2 may be respective anode and cathode regions of the diode. The anode region 1 then comprises a high-doped zone 25, a distinct intermediately-doped zone 24, and a low-doped base zone 23, all of p-type conductivity. The switchable p-n junction 12 is formed between the low-doped p-type base region 23 (formed by the body portion 11) and the n-type cathode region 2. The high-doped zone 25 is provided with an anode contact 31 at the major surface 18, whereas the n-type (second) region 2 is provided with a cathode contact 32 at the major surface 19 of the cathode region 2 opposite the anode region 1. The diode may, for example, have a structure similar to that of Figure 1 , except that the region 3 and contact 33 are absent in the diode. Such a high-voltage switching diode having a doping profile in accordance with the present invention may form, for example, the switching diode D in the Figure 5 circuit.

In the embodiments so far described, the region 1 is of p-type conductivity. However, embodiments of the present invention are also possible in which the region 1 (with its distinct zones 23, 24 and 25) is of n-type conductivity, and the region 2 is of p-type conductivity. A PNP transistor embodiment is also possible. In the NPN transistor embodiment of Figures 1 and 2, the p-type zones 23, 24, 25 are comprised in the base region 1. In a PNP transistor embodiment, p-type zones 23', 24', 25' in accordance with the present invention may be comprised in the collector region 2'. In this case the uniformally low doped p-type zone 23' forms the collector junction 12" with the n-type base region 1', the high-doped p-type zone 25' is a collector contact zone, and the distinct additional p-type zone 24' (of an intermediate doping concentration which decreases towards zone 23') is present between zones 25' and 23'. Both zones 23' and 24' may be depleted by the depletion layer from the junction 12" in the off-state of this PNP transistor. From reading the present disclosure, other modifications and variations will be apparent to persons skilled in the art. Such modifications and variations may involve equivalent features and other features which are already known in the art and which may be used instead of or in addition to features already disclosed herein. Although claims have been formulated in this Application to particular combinations of features, it should be understood that the scope of the disclosure of the present application includes any and every novel feature or any novel combination of features disclosed herein either explicitly or implicitly and any generalisation thereof, whether or not it relates to the same invention as presently claimed in any Claim and whether or not it mitigates any or all of the same technical problems as does the present invention. The Applicants hereby give notice that new claims may be formulated to such features and/or combinations of such features during prosecution of the present application or of any further application derived therefrom.

Claims

1. A semiconductor switch device comprising a semiconductor body in which a switchable reverse-biased p-n junction is present between a first region of a first conductivity type and a second region of an opposite second conductivity type, wherein the first region includes a high-doped zone having a higher doping concentration of the first conductivity type than a low-doped zone of the first region, the low-doped zone comprises a body portion having a substantially uniform doping concentration of the first conductivity type which forms the p-n junction with the second region, and the doping concentration of the high-doped zone decreases towards the low-doped zone over a part of the thickness of the body, characterised in that the first region further comprises, between the low-doped and high-doped zones, a distinct additional zone having an additional doping concentration of the first conductivity type which is lower than the doping concentration of the high-doped zone and which decreases towards the low-doped zone, the additional zone providing a path for extracting charge carriers from the low-doped zone when the switch device is being switched off.

2. A device as claimed in Claim 1 , further characterised in that the first and second regions are respective base and collector regions of a bipolar transistor, the transistor also comprises an emitter region which forms a p-n junction with the high-doped zone of the base region, and the emitter region and high-doped zone of the base region are provided with respective emitter and base contacts at a surface of the body.

3. A device as claimed in Claim 1 , further characterised in that the first and second regions are respective anode and cathode regions of a switching diode, the high-doped zone of the first region is provided with an anode contact at a surface of the body, and the second region is provided with a cathode contact at a surface of the cathode region opposite the anode region.

4. A device as claimed in any one of the preceding Claims, further characterised in that the low-doped zone of the first region has such a low doping concentration as to be depleted, when the switch device is operated in an off state, from the reverse-biased p-n junction across the thickness of the low-doped zone and into the additional zone of the first region.

5. A device as claimed in any one of the preceding Claims, further characterised in that the maximum doping concentration of the additional zone of the first region is at least an order of magnitude lower than the doping concentration of the high-doped zone and at least an order of magnitude higher than the doping concentration of the low-doped zone.

6. A device as claimed in any one of the preceding Claims, further characterised in that the p-n junction is extended to a surface of the body which is adjoined by the first region, and an annular field-relief region of the second conductivity type is present in the low-doped zone of the first region between the additional zone of the first region and the extension of the p-n junction at said surface.

7. A method of manufacturing a semiconductor switch device having a switchable reverse-biased p-n junction between a first region of a first conductivity type and a second region of an opposite second conductivity type, which first region includes a high-doped zone having a higher doping concentration of the first conductivity type than a low-doped zone of the first region, wherein there is provided a semiconductor body portion having a substantially uniform doping concentration of the first conductivity type to form the low-doped zone adjacent to the p-n junction, the body portion having a surface located opposite the p-n junction, and wherein a doping step is carried out by doping the body portion over a part of its thickness with dopant characteristic of the first conductivity type through the surface so as to provide the high-doped zone with a doping concentration which decreases towards the low-doped zone, characterised in that an additional doping step is carried out to form an additional zone of the first region, between the low-doped and high-doped zones, by additionally doping the semiconductor body portion with dopant characteristic of the first conductivity type through the surface to a greater depth than the high-doped zone and with a lower doping concentration than the high-doped zone to provide the additional zone with a doping concentration which decreases towards the low-doped zone.

8. A method as claimed in Claim 7, further characterised in that at least a part of the second region of the second conductivity type is formed by overdoping a part of the semiconductor body portion to form the p-n junction with the first region.

9. A method as claimed in Claim 7 or Claim 8, further characterised in that a groove is etched into the body portion from the surface to surround the first region, and the body portion of the first conductivity type is overdoped along side-walls of the groove with dopant characteristic of the second conductivity type to form an annular region of the second conductivity type which reaches to the second region of the second conductivity type so as to extend the p-n junction to said surface.

10. A method as claimed in Claim 9, further characterised in that the first and second regions are respective base and collector regions of a bipolar transistor, and the body portion of the first conductivity type is overdoped with the dopant characteristic of the second conductivity type to form simultaneously the annular region and an emitter region of the transistor, the emitter region forming a p-n junction with the high-doped zone of the base region.