Semiconductor device and method for manufacturing the same

Abstract

A semiconductor device is disclosed in which proton implantation is performed a plurality of times to form a plurality of n-type buffer layers in an n-type drift layer at different depths from a rear surface of a substrate. The depth of the n-type buffer layer, which is provided at the deepest position from the rear surface of the substrate, from the rear surface of the substrate is more than 15 μm. The temperature of a heat treatment which is performed in order to change a proton into a donor and to recover a crystal defect after the proton implantation is equal to or higher than 400° C. In a carrier concentration distribution of the n-type buffer layer, a width from the peak position of carrier concentration to an anode is more than a width from the peak position to a cathode.

Claims

1. A diode comprising: an n-type drift layer provided in an n-type semiconductor substrate; a p-type anode layer provided in a first main surface of the semiconductor substrate; an n-type cathode layer provided in a second main surface of the semiconductor substrate; and a plurality of n-type buffer layers provided between the drift layer and the cathode layer, the buffer layers including hydrogen as a donor, wherein the buffer layers comprise: a shallowest buffer layer formed on a further inner side of the semiconductor substrate than the cathode layer, an intermediate buffer layer formed on a further inner side of the semiconductor substrate than the shallowest buffer layer, and a deepest buffer layer formed in a position deeper than 15 μm from the second main surface of the semiconductor substrate and on a further inner side of the semiconductor substrate than the intermediate buffer layer, wherein a distribution of a carrier concentration of the semiconductor substrate has a peak and a tail, of which the carrier concentration is lower than that of the peak, in each of the shallowest buffer layer, the intermediate buffer layer and the deepest buffer layer, wherein the shallowest buffer layer is in contact with the cathode layer via an interposed region, the interposed region being directly in contact with the cathode layer, wherein the distribution has a flat portion, at which the carrier concentration is substantially constant, between the tail of the shallowest buffer layer and the tail of the intermediate buffer layer, wherein a length of the flat portion is longer than a length from the second main surface of the semiconductor substrate to the peak of the shallowest buffer layer, wherein the carrier concentration between the second main surface of the semiconductor substrate and the deepest buffer layer is higher than the carrier concentration of the drift layer, and wherein the carrier concentration in the interposed region has a first slope, which decreases linearly in a logarithmic scale toward the cathode layer, with a first gradient, within the interposed region, an absolute value of the first gradient being higher than 0.

2. The diode according to claim 1, wherein the flat portion comprises a position of 10 μm from the second main surface of the semiconductor substrate.

3. The diode according to claim 2, wherein the flat portion is a region in which the carrier concentration varies within a range of ±20%.

4. The diode according to claim 3, wherein in each of the buffer layers, the tail is a region between the peak and a point at which a gradient is substantially zero (0).

5. The diode according to claim 4, wherein the flat portion connects the tail of the shallowest buffer layer and the tail of the intermediate buffer layer.

6. The diode according to claim 5, wherein the peak of the shallowest buffer layer is greater than the peak of the intermediate buffer layer, and greater than the peak of the deepest buffer layer, and wherein the peak of the intermediate buffer layer and the peak of the deepest buffer layer are substantially the same.

7. The diode according to claim 6, wherein the intermediate buffer layer is provided in plural.

8. The diode according to claim 7, wherein the carrier concentration between the second main surface of the semiconductor substrate and the peak of the deepest buffer layer is equal to or higher than 1×10.sup.14/cm.sup.3.

9. The diode according to claim 1, wherein the drift layer is provided between the deepest buffer layer and the anode layer.

10. The diode according to claim 1, wherein the carrier concentration in the interposed region is equal to a donor concentration of the cathode layer at an interface between interposed region and the cathode layer.

11. The diode according to claim 1, wherein the flat portion includes hydrogen as a donor.

12. The diode according to claim 1, wherein the carrier concentration of the flat portion is higher than that of the drift layer.

13. The diode according to claim 1, wherein the carrier concentration in the shallowest buffer layer has a second slope, which increases linearly in a logarithmic scale, with a second gradient, from the interposed region to the peak of the shallowest buffer layer, an absolute value of the second gradient being higher than that of the first gradient.

14. The diode according to claim 1, wherein the carrier concentration in the intermediate buffer layer has a third slope at a side of the second main surface of the intermediate buffer layer, the third slope decreasing linearly in a logarithmic scale from the peak of the intermediate buffer layer toward the cathode layer, with a third gradient, an absolute value of the third gradient being lower than that of the first gradient.

15. A diode comprising: an n-type drift layer provided in an n-type semiconductor substrate; a p-type anode layer provided in a first main surface of the semiconductor substrate; an n-type cathode layer provided in a second main surface of the semiconductor substrate; and a plurality of n-type buffer layers provided between the drift layer and the cathode layer, the buffer layers including hydrogen as a donor, wherein the buffer layers comprise: a shallowest buffer layer formed on a further inner side of the semiconductor substrate than the cathode layer, an intermediate buffer layer formed on a further inner side of the semiconductor substrate than the shallowest buffer layer, and a deepest buffer layer formed in a position deeper than 15 μm from the second main surface of the semiconductor substrate and on a further inner side of the semiconductor substrate than the intermediate buffer layer, wherein a distribution of a carrier concentration of the semiconductor substrate has a peak and a tail, of which the carrier concentration is lower than that of the peak, in each of the shallowest buffer layer, the intermediate buffer layer and the deepest buffer layer, wherein the shallowest buffer layer is in contact with the cathode layer via an interposed region, the interposed region being directly in contact with the cathode layer, wherein the distribution has a flat portion, at which the carrier concentration is substantially constant, between the tail of the shallowest buffer layer and the tail of the intermediate buffer layer, wherein a length of the flat portion is equal to or greater than 0.3 times of a length L.sub.AB between the peak of the shallowest buffer layer and the peak of the intermediate buffer layer, wherein the carrier concentration between the second main surface of the semiconductor substrate and the deepest buffer layer is higher than the carrier concentration of the drift layer, and wherein the carrier concentration in the interposed region has a first slope, which decreases linearly in a logarithmic scale toward the cathode layer, with a first gradient, within the interposed region, an absolute value of the first gradient being higher than 0.

16. The diode according to claim 15, wherein the length of the flat portion is equal to or less than 0.7 times of the length L.sub.AB between the peak of the shallowest buffer layer and the peak of the intermediate buffer layer.

17. The diode according to claim 16, wherein the flat portion comprises a position of 10 μm from the second main surface of the semiconductor substrate.

18. The diode according to claim 17, wherein the flat portion is a region in which the carrier concentration varies within a range of ±20%.

19. The diode according to claim 18, wherein in each of the buffer layers, the tail is a region between the peak and a point at which a gradient is substantially zero (0).

20. The diode according to claim 19, wherein the length of the flat portion is longer than a length from the second main surface of the semiconductor substrate to the peak of the shallowest buffer layer.

21. The diode according to claim 20, wherein the flat portion connects the tail of the shallowest buffer layer and the tail of the intermediate buffer layer.

22. The diode according to claim 21, wherein the peak of the shallowest buffer layer is greater than the peak of the intermediate buffer layer, and greater than the peak of the deepest buffer layer, and wherein the peak of the intermediate buffer layer and the peak of the deepest buffer layer are substantially the same.

23. The diode according to claim 22, wherein the intermediate buffer layer is provided in plural.

24. The diode according to claim 23, wherein the carrier concentration between the second main surface of the semiconductor substrate and the peak of the deepest buffer layer is equal to or higher than 1×10.sup.14/cm.sup.3.

25. The diode according to claim 15, wherein the drift layer is provided between the deepest buffer layer and the anode layer.

26. The diode according to claim 15, wherein the carrier concentration in the interposed region is equal to a donor concentration of the cathode layer at an interface between interposed region and the cathode layer.

27. The diode according to claim 15, wherein the flat portion includes hydrogen as a donor.

28. The diode according to claim 15, wherein the carrier concentration of the flat portion is higher than that of the drift layer.

29. The diode according to claim 15, wherein the carrier concentration in the shallowest buffer layer has a second slope, which increases linearly in a logarithmic scale, with a second gradient, from the interposed region to the peak of the shallowest buffer layer, an absolute value of the second gradient being higher than that of the first gradient.

30. The diode according to claim 15, wherein the carrier concentration in the intermediate buffer layer has a third slope at a side of the second main surface of the intermediate buffer layer, the third slope decreasing linearly in a logarithmic scale from the peak of the intermediate buffer layer toward the cathode layer, with a third gradient, an absolute value of the third gradient being lower than that of the first gradient.

31. A method for manufacturing a diode comprising an n-type drift layer provided in an n-type semiconductor substrate; a p-type anode layer provided in a first main surface of the semiconductor substrate; an n-type cathode layer provided in a second main surface of the semiconductor substrate; and a plurality of n-type buffer layers provided between the drift layer and the cathode layer, the buffer layers including hydrogen as a donor, wherein the method comprising: a buffer layer forming process of forming, as the buffer layers, a shallowest buffer layer formed on a further inner side of the semiconductor substrate than the cathode layer, an intermediate buffer layer formed on a further inner side of the semiconductor substrate than the shallowest buffer layer, and a deepest buffer layer formed in a position deeper than 15 μm from the second main surface of the semiconductor substrate and on a further inner side of the semiconductor substrate than the intermediate buffer layer, by implanting hydrogen and performing annealing, such that: a distribution of a carrier concentration of the semiconductor substrate has a peak and a tail, of which the carrier concentration is lower than that of the peak, in each of the shallowest buffer layer, the intermediate buffer layer and the deepest buffer layer, the shallowest buffer layer is in contact with the cathode layer via an interposed region, the interposed region being directly in contact with the cathode layer, the distribution has a flat portion, at which the carrier concentration is substantially constant, between the tail of the shallowest buffer layer and the tail of the intermediate buffer layer, a length of the flat portion is longer than a length from the second main surface of the semiconductor substrate to the peak of the shallowest buffer layer, the carrier concentration between the second main surface of the semiconductor substrate and the deepest buffer layer is higher than the carrier concentration of the drift layer, and the carrier concentration in the interposed region has a first slope, which decreases linearly in a logarithmic scale toward the cathode layer, with a first gradient, within the interposed region, an absolute value of the first gradient being higher than 0.

32. The method according to claim 31, wherein the flat portion is a region in which the carrier concentration varies within a range of ±20%.

33. The method according to claim 32, wherein in each of the buffer layers, the tail is a region between the peak and a point at which a gradient is substantially zero (0).

34. The method according to claim 33, wherein the flat portion connects the tail of the shallowest buffer layer and the tail of the intermediate buffer layer.

35. The method according to claim 34, wherein the peak of the shallowest buffer layer is greater than the peak of the intermediate buffer layer, and greater than the peak of the deepest buffer layer, and the peak of the intermediate buffer layer and the peak of the deepest buffer layer are substantially the same.

36. The method according to claim 35, wherein in the buffer layer forming process, the intermediate buffer layer is formed in plural.

37. The method according to claim 31, wherein the drift layer is provided between the deepest buffer layer and the anode layer.

38. A method for manufacturing a diode comprising an n-type drift layer provided in an n-type semiconductor substrate; a p-type anode layer provided in a first main surface of the semiconductor substrate; an n-type cathode layer provided in a second main surface of the semiconductor substrate; and a plurality of n-type buffer layers provided between the drift layer and the cathode layer, the buffer layers including hydrogen as a donor, wherein the method comprising: a buffer layer forming process of forming, as the buffer layers, a shallowest buffer layer formed on a further inner side of the semiconductor substrate than the cathode layer, an intermediate buffer layer formed on a further inner side of the semiconductor substrate than the shallowest buffer layer, and a deepest buffer layer formed in a position deeper than 15 μm from the second main surface of the semiconductor substrate and on a further inner side of the semiconductor substrate than the intermediate buffer layer, by implanting hydrogen and performing annealing, such that: a distribution of a carrier concentration of the semiconductor substrate has a peak and a tail, of which the carrier concentration is lower than that of the peak, in each of the shallowest buffer layer, the intermediate buffer layer and the deepest buffer layer, the shallowest buffer layer is in contact with the cathode layer via an interposed region, the interposed region being directly in contact with the cathode layer, the distribution has a flat portion, at which the carrier concentration is substantially constant, between the tail of the shallowest buffer layer and the tail of the intermediate buffer layer, a length of the flat portion is equal to or greater than 0.3 times of a length L.sub.AB between the peak of the shallowest buffer layer and the peak of the intermediate buffer layer, the carrier concentration between the second main surface of the semiconductor substrate and the deepest buffer layer is higher than the carrier concentration of the drift layer, and the carrier concentration in the interposed region has a first slope, which decreases linearly in a logarithmic scale toward the cathode layer, with a first gradient, within the interposed region, an absolute value of the first gradient being higher than 0.

39. The method according to claim 38, wherein the length of the flat portion is equal to or less than 0.7 times of the length L.sub.AB between the peak of the shallowest buffer layer and the peak of the intermediate buffer layer.

40. The method according to claim 39, wherein the flat portion is a region in which the carrier concentration varies within a range of ±20%.

41. The method according to claim 40, wherein in each of the buffer layers, the tail is a region between the peak and a point at which a gradient is substantially zero (0).

42. The method according to claim 41, wherein the length of the flat portion is longer than a length from the second main surface of the semiconductor substrate to the peak of the shallowest buffer layer.

43. The method according to claim 42, wherein the flat portion connects the tail of the shallowest buffer layer and the tail of the intermediate buffer layer.

44. The method according to claim 43, wherein the peak of the shallowest buffer layer is greater than the peak of the intermediate buffer layer, and greater than the peak of the deepest buffer layer, and the peak of the intermediate buffer layer and the peak of the deepest buffer layer are substantially the same.

45. The method according to claim 44, wherein in the buffer layer forming process, the intermediate buffer layer is formed in plural.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) The foregoing advantages and features of the invention will become apparent upon reference to the following detailed description and the accompanying drawings, of which:

(2) FIGS. 1A and 1B are cross-sectional views illustrating the structure of a semiconductor device according to Example 1 of the invention;

(3) FIG. 2 is a characteristic diagram illustrating a carrier concentration distribution along the cutting line A-A′ of FIG. 1A;

(4) FIG. 3 is a characteristic diagram illustrating the relation between donor concentration and an oscillating voltage threshold value V.sub.RRO;

(5) FIG. 4 is a cross-sectional view illustrating the state of a semiconductor device according to Example 2 of the invention which is being manufactured;

(6) FIG. 5 is a cross-sectional view illustrating the state of the semiconductor device according to Example 2 of the invention which is being manufactured, following FIG. 4;

(7) FIG. 6 is a cross-sectional view illustrating the state of the semiconductor device according to Example 2 of the invention which is being manufactured, following FIG. 5;

(8) FIG. 7 is a cross-sectional view illustrating the state of the semiconductor device according to Example 2 of the invention which is being manufactured, following FIG. 6;

(9) FIG. 8 is a cross-sectional view illustrating a main portion of a PiN diode 500 including a general n-type buffer layer 55;

(10) FIG. 9 is a characteristic diagram illustrating the net doping concentration of the semiconductor device according to Example 1 of the invention;

(11) FIG. 10 is a characteristic diagram illustrating the relation between a coefficient α and a standardized oscillating voltage threshold value V.sub.RRO;

(12) FIG. 11 is a characteristic diagram illustrating the relation between a coefficient β and a standardized oscillating voltage threshold value V.sub.RRO in a diode with a rated voltage of 1200 V according to Example 1;

(13) FIGS. 12A and 12B are diagrams illustrating the structure of a semiconductor device according to Example 3 of the invention;

(14) FIG. 13 is a schematic diagram illustrating the definition of the reverse recovery waveform and the oscillating voltage threshold value V.sub.RRO of the diode;

(15) FIG. 14 is a characteristic diagram illustrating the relation between the logarithm (log(Rp)) of the range Rp of a proton in proton implantation and the logarithm (log(E)) of the acceleration energy E of the proton in Example 2 of the invention;

(16) FIG. 15 is a table illustrating a distance Y from a rear surface of a substrate to the peak position of the deepest n-type buffer layer 5 at each rated voltage; and

(17) FIGS. 16A and 16B are characteristic diagrams schematically illustrating a carrier concentration distribution with respect to the distance from the rear surface.

DETAILED DESCRIPTION OF SPECIFIC EMBODIMENTS

(18) Hereinafter, a semiconductor device and a method for manufacturing the same according to exemplary embodiments of the invention will be described in detail with reference to the accompanying drawings. In the specification and the accompanying drawings, in the layers or regions having “n” or “p” appended thereto, an electron or a hole means a majority carrier. In the description of the following embodiments and the accompanying drawings, the same components are denoted by the same reference numerals and the description thereof will not be repeated. The embodiments are represented by the following examples.

Example 1

(19) The structure of a semiconductor device according to Example 1 will be described. FIGS. 1A and 1B are cross-sectional views illustrating the structure of the semiconductor device according to Example 1 of the invention. FIG. 1A is a crosssectional view illustrating a main portion and FIG. 1B is an enlarged view illustrating a portion D. FIG. 1B illustrates the depth of n-type buffer layers 5, 6, and 7. A PiN diode 100 is given as an example of the semiconductor device according to Example 1 illustrated in FIGS. 1A and 1B. The PiN diode 100 has a breakdown voltage of 1200 V. The following carrier concentration is calculated from spreading resistance (SR).

(20) The PiN diode 100 includes a p-type anode layer 3 that is provided in a surface layer of a first main surface (front surface) of an n-type silicon substrate 1 with a thickness of about 120 μm and an n-type cathode layer 4 which is provided in a second main surface (rear surface). Three n-type buffer layers 5, 6, and 7 are provided in the n-type silicon substrate 1 at different depths from the n-type cathode layer 4 to the p-type anode layer 3. The impurity concentration of the n-type buffer layers 5, 6, and 7 is more than that of the n-type silicon substrate 1. An anode electrode 8 which is connected to the p-type anode layer 3 is provided on the front surface of the n-type silicon substrate 1. A cathode electrode 9 which is connected to the n-type cathode layer 4 is provided on a rear surface 1a of the n-type silicon substrate 1.

(21) A high-breakdown-voltage junction termination structure 11 is provided outside the p-type anode layer 3 in a direction horizontal to the main surface of the substrate so as to surround an active region in which the p-type anode layer 3 is provided. The high-breakdown-voltage junction termination structure 11 includes a plurality of ring-shaped p-type layers 12 which surround the p-type anode layer 3, termination electrodes 13 which are provided on the p-type layers 12, and an insulating film 14 which is, for example, an oxide film and separates the p-type anode layer 3 from each p-type layer 12.

(22) Among the three n-type buffer layers 5, 6, and 7, the n-type buffer layer (hereinafter, referred to as the deepest n-type buffer layer) 5 which is arranged at the deepest position from the rear surface 1a of the n-type silicon substrate 1 is formed in the vicinity of the middle between the p-type anode layer 3 and the n-type cathode layer 4. Specifically, a peak carrier concentration position 5a of the deepest n-type buffer layer 5 is disposed at a depth P1 from the interface (the rear surface 1a of the n-type silicon substrate 1) between the n-type cathode layer 4 and the cathode electrode 9 and the depth P1 is about 60 μm from the interface between the n-type cathode layer 4 and the cathode electrode 9.

(23) The anode electrode 8 and the cathode electrode 9 are metal electrodes. A region of the n-type silicon substrate 1 between the p-type anode layer 3 and the deepest n-type buffer layer 5 is an n-type drift layer 2. A region interposed between the n-type buffer layers 5 and 6 which are adjacent to each other in the depth direction is a carrier storage region 15 and a region interposed between the n-type buffer layers 6 and 7 which are adjacent to each other in the depth direction is a carrier storage region 16. The entire region, including the carrier storage regions 15 and 16, a region 17 which is interposed between the n-type cathode layer 4 and the n-type buffer layer (hereinafter, referred to as the shallowest n-type buffer layer) 7 that is arranged at the shallowest position from the rear surface 1a of the n-type silicon substrate 1, and the n-type cathode layer 4 is an n-type layer 20. The n-type layer 20 is a region in which protons, which are hydrogen (H) ions (proton) implanted in different ranges from the rear surface of the n-type silicon substrate 1, are changed into donors (hydrogen-related donors).

(24) Next, the carrier concentration distribution of the n-type buffer layers 5, 6, and 7 in the PiN diode 100 will be described. FIG. 2 is a characteristic diagram illustrating the carrier concentration distribution along the line A-A′ of FIG. 1A. FIG. 2 illustrates the carrier concentration distribution of the PiN diode 100 in which the three n-type buffer layers 5, 6, and 7 are formed in the n-type silicon substrate 1. In FIG. 2, the horizontal axis is the depth from the rear surface 1a of the n-type silicon substrate 1 and “0” on the horizontal axis indicates the rear surface 1a of the n-type silicon substrate 1. The diffusion depth T of the n-type cathode layer 4 is about 1 μm (see FIG. 1B for reference numerals T, E1, E2, E3, Q1, Q2, and Q3).

(25) The three n-type buffer layers 5, 6, and 7 are formed in the n-type silicon substrate 1, for example, under the following conditions. A proton implantation dose is, for example, about 1×10.sup.13/cm.sub.2 to 2×10.sup.13/cm.sub.2. Acceleration energy levels (E1, E2, and E3) are, for example, 2.3 MeV, 1.5 MeV, and 0.5 MeV. Ranges (P1, P2, and P3) corresponding to the acceleration energy levels (E1, E2, and E3) are 60 μm, 30 μm, and 6 μm, respectively. A heat treatment is performed at a temperature of 420° C. for 3 hours. In a comparative example represented by a dotted line, the heat treatment temperature was 380° C. In the comparative example, the conditions are the same as those in Example 1 except for the heat treatment temperature.

(26) The depths (Q1, Q2, and Q3) of the n-type buffer layers 5, 6, and 7 from the rear surface 1a (the position of 0 on the horizontal axis in FIG. 2) of the n-type silicon substrate 1 which are represented by the peak carrier concentration positions (5a, 6a, and 7a) of the n-type buffer layers 5, 6, and 7 are 60 μm, 25 μm, and 6 μm, respectively. The depths are equal to proton implantation ranges (P1, P2, and P3). The proton implantation range means the distance from an implantation surface (the rear surface 1a of the n-type silicon substrate 1) to the position (the peak position of a proton distribution) with a peak proton implantation amount. The proton is spread to the front and rear sides of the range (to the anode side and the cathode side in the depth direction). As the range becomes long, the extension area of the proton is widened.

(27) In the carrier concentration distribution of the deepest n-type buffer layer 5, a first width W1 from the peak carrier concentration position 5a of the deepest n-type buffer layer 5 to the anode is more than a second width W2 from the peak carrier concentration position 5a to the cathode (the rear surface 1a of the n-type silicon substrate 1) opposite to the anode (which is reversed in the comparative example). The width of the deepest n-type buffer layer 5 may be the half-width at half-maximum (HWHM) of the peak carrier concentration of the deepest n-type buffer layer 5 or the width of 1/e (e is the base of natural logarithm or a Napier's constant and is about 2.71828) of the peak carrier concentration. It is supposed that the first width W1 is more than the second width W2 because damage by proton implantation and a defect caused by the damage are smaller on the front side (anode side) of the range than on the rear side (cathode side) thereof and many defects remain on the rear side (cathode side) of the range, resulting in a reduction in carrier concentration. A depletion layer which is spread from a pn junction between the p-type anode layer 3 and the n-type drift layer 2 reaches the deepest n-type buffer layer 5 first. In this case, when the first width W1 is more than the second width W2, the gradient of the carrier concentration is gentle on the anode side in the carrier concentration distribution of the deepest n-type buffer layer 5. Then, the spreading of the depletion layer in the region is reduced and an increase in the rate of change of voltage dV/dt is small. When the rate of change of voltage dV/dt increases sharply, voltage oscillation occurs. However, since the increase in the rate of change of voltage dV/dt is small, it is possible to suppress the voltage oscillation. On the other hand, when the first width W1 is less than the second width W2, the spreading of the depletion layer is suddenly stopped at the peak carrier concentration position 5a of the deepest n-type buffer layer 5. Therefore, voltage oscillation is likely to occur.

(28) The carrier concentration distributions of the regions (carrier storage regions) 15 and 16 interposed between the n-type buffer layers 5, 6, and 7 which are adjacent to each other in the depth direction are flat (impurity concentration is uniform in the depth direction). Herein “flat (impurity concentration is uniform in the depth direction)” may be satisfied when the following is established. FIGS. 16A and 16B are characteristic diagrams schematically illustrating a carrier concentration distribution with respect to the distance from the rear surface. The scale of the vertical axis is a common logarithm. In FIG. 16A, it is assumed that an n-type buffer layer A (for example, the n-type buffer layer 5) is provided at a given depth and an n-type buffer layer B (for example, the n-type buffer layer 6) is provided closer to the rear surface of the substrate than the n-type buffer layer A. In addition, it is assumed that the distance between positions (peak positions) X.sub.A and X.sub.B where the n-type buffer layer A and the n-type buffer layer B have the maximum carrier concentration is L.sub.AB. It is considered that a region M with a length (length in the depth direction) of aL.sub.AB is provided between X.sub.A and X.sub.B. Herein “a” is a non-dimensional coefficient and is greater than 0 and less than 1, particularly, equal to or greater than 0.3 and equal to or less than 0.7. That is, the length aL.sub.AB of the region M is in the range of 30% to 70% of the distance LAB between X.sub.A and X.sub.B. In the example illustrated in FIGS. 16A and 16B, for example, the coefficient “α” is 0.4 (the length aL.sub.AB of the region M is 40% of the distance LAB between X.sub.A and X.sub.B). In addition, it is assumed that the region M includes the position where carrier concentration is the minimum between X.sub.A and X.sub.B. For example, the position where carrier concentration is the minimum may be the center of the region M or it may deviate from the center of the region M.

(29) The region M is used to define the term “flat (impurity concentration is uniform in the depth direction)” as follows. When the carrier concentration distribution of the region M is in the range of ±20% of the average carrier concentration N.sub.mean of the region M, that is, in the range of 0.8 N.sub.mean to 1.2 N.sub.mean, the carrier concentration between the n-type buffer layer A and the n-type buffer layer B is “flat (impurity concentration is uniform in the depth direction)”. The average carrier concentration N.sub.mean of the region M is obtained by integrating the carrier concentration in the depth direction in the range of the region M to calculate integrated concentration and dividing the integrated concentration by the length aL.sub.AB of the region M. That is, preferably, the carrier concentration of the region M is in the range of ±20% of the average carrier concentration N.sub.mean of the region M and the region M includes the position where the carrier concentration is the minimum between X.sub.A and X.sub.B. In FIG. 16A, the carrier concentration between the n-type buffer layer A and the n-type buffer layer B is flat. In contrast, in FIG. 16B, the region M includes the position where the carrier concentration is the minimum, but the minimum carrier concentration and the carrier concentration at both ends of the region M are beyond the range of ±20% of the average carrier concentration N.sub.mean of the region M. Therefore, in the example illustrated in FIG. 16B, the carrier concentration between the n-type buffer layer A and the n-type buffer layer B is not flat.

(30) Carriers are stored in the carrier storage regions 15 and 16 when the diode is turned on. It is preferable that the doping concentration of the carrier storage regions 15 and 16 be flat, similarly to the doping concentration distribution of the n-type silicon substrate 1. In addition, the carrier concentration of the carrier storage regions 15 and 16 is less than the peak carrier concentration of the n-type buffer layers 5, 6, and 7 and is, for example, equal to or more than the carrier concentration of the n-type silicon substrate 1 and equal to or less than five times the carrier concentration of the n-type silicon substrate 1. When the carrier concentration of the carrier storage regions 15 and 16 is set to the above-mentioned range, it is possible to increase the remaining number of stored carriers.

(31) In contrast, when the carrier concentration of the carrier storage regions 15 and 16 is less than the carrier concentration of the n-type silicon substrate 1 and is more than five times the carrier concentration of the n-type silicon substrate 1, the following problems occur, which is not preferable. The carrier concentration of the carrier storage regions 15 and 16 which is less than the carrier concentration of the n-type silicon substrate 1 means that the carrier concentration of the carrier storage regions 15 and 16 is not flat and crystal defects are not sufficiently recovered. Therefore, when a reverse bias is applied and the depletion layer which is spread from the pn junction between the p-type anode layer 3 and the n-type drift layer 2 reaches the carrier storage regions 15 and 16, carriers are generated around the residual defects in the carrier storage regions 15 and 16, which results in an increase in leakage current. When the diode is turned on and when the reverse recovery is applied, carriers are recombined around the residual defects in the carrier storage regions 15 and 16. As a result, the carriers are reduced and depleted.

(32) When the carrier concentration of the carrier storage regions 15 and 16 is more than five times the carrier concentration of the n-type silicon substrate 1, net doping concentration is too high due to the donors, neutral conditions are satisfied by electrons and donor ions, and the number of holes is significantly reduced. As a result, the holes are quickly depleted and the voltage and current oscillate during reverse recovery, which causes radiation noise.

(33) For this reason, the carrier concentration of the carrier storage regions 15 and 16 is set to be equal to or more than the carrier concentration of the n-type silicon substrate 1 and equal to or less than five times the carrier concentration of the n-type silicon substrate 1 and the carrier concentration distribution is flat, which makes it possible to delay the depletion of the holes. As a result, it is possible to suppress the oscillation of voltage and current during reverse recovery and thus suppress the generation of radiation noise.

(34) It is preferable that the carrier concentration of the carrier storage regions 15 and 16 interposed between the n-type buffer layers 5, 6, and 7 which are adjacent to each other in the depth direction increase toward the n-type cathode layer 4. That is, the carrier concentration of the carrier storage region 16 interposed between the shallowest n-type buffer layer 7 and the n-type buffer layer (hereinafter, referred to as an intermediate-depth n-type buffer layer) 6 arranged at an intermediate position between the n-type buffer layer 5 and the n-type buffer layer 7 is more than the carrier concentration of the carrier storage region 15 interposed between the deepest n-type buffer layer 5 and the intermediate-depth n-type buffer layer 6 (the carrier concentration of the carrier storage region 16>the carrier concentration of the carrier storage region 15). According to this carrier concentration distribution, the expansion of the depletion layer, which is spread from the pn junction between the p-type anode layer 3 and the n-type drift layer 2, to the cathode is slowly suppressed and the depletion of the holes is delayed. Therefore, the oscillation of voltage and current during reverse recovery is suppressed. As a result, it is possible to suppress the generation of radiation noise.

(35) When the carrier concentration of the carrier storage regions 15 and 16 interposed between the n-type buffer layers 5, 6, and 7 is more than that of the n-type silicon substrate 1, the spreading of the depletion layer from the pn junction between the p-type anode layer 3 and the n-type drift layer 2 is suppressed. Therefore, even when the n-type silicon substrate 1 has a small thickness, it is possible to ensure a breakdown voltage and to reduce generation loss. In addition, the carrier concentration of the carrier storage regions 15 and 16 interposed between the n-type buffer layers 5, 6, and 7 which is more than that of the n-type silicon substrate 1 proves that crystal defects in the carrier storage regions 15 and 16 are sufficiently recovered. Therefore, it is possible to reduce a leakage current.

(36) In the PiN diode 100 having the above-mentioned structure, good soft recovery characteristics are obtained and it is possible to reduce a surge voltage and radiation noise.

(37) Next, the relation between an oscillating voltage threshold value V.sub.RRO and the average donor concentration of the carrier storage regions 15 and 16 will be described. As described above, the average donor concentration may be, for example, the average concentration N.sub.mean of the region M including the position with the minimum carrier concentration in the carrier storage regions 15 and 16. FIG. 3 is a characteristic diagram illustrating the relation between the donor concentration and the oscillating voltage threshold value V.sub.RRO. FIG. 13 is a schematic diagram illustrating the definition of the reverse recovery waveform and the oscillating voltage threshold value V.sub.RRO of the diode. The oscillating voltage threshold value V.sub.RRO is defined as follows. As illustrated in FIG. 13, it is assumed that a forward current I.sub.F is fixed to a predetermined value by a chopper circuit which is an inductance load and the diode is reversely recovered on the basis of a power supply voltage V.sub.CC and the floating inductance L.sub.S of the circuit. In this case, reverse recovery is repeated while the power supply voltage V.sub.CC is gradually increased. When the power supply voltage V.sub.CC is greater than a given value, the voltage and current waveforms of the diode oscillate. The value of the power supply voltage V.sub.CC at which the reverse recovery waveform starts to oscillate is defined as the oscillating voltage threshold value V.sub.RRO. Since the oscillation phenomenon of the diode during reverse recovery is mainly caused by radiation noise, it is possible to suppress the oscillation as much as possible. It is preferable to maximize the oscillating voltage threshold value V.sub.RRO in order to suppress the oscillation during reverse recovery.

(38) As illustrated in FIG. 3, when the concentration of the donors (donor concentration) generated by proton implantation is less than the carrier concentration (electron concentration) of the n-type silicon substrate 1, the crystal defects are not sufficiently recovered (there are many crystal defects). Therefore, as described above, the lifetime of carriers is reduced by the recombination center and the number of stored carriers is reduced. Therefore, the carriers are quickly depleted during reverse recovery. As a result, the voltage and current are likely to oscillate during the reverse recovery of the diode and the oscillating voltage threshold value V.sub.RRO is reduced. In particular, the oscillation phenomenon during reverse recovery is likely to occur when the forward current I.sub.F is less than one-tenth of the rated current.

(39) In contrast, when the donor concentration (electron concentration) is significantly more than the carrier concentration (electron concentration) of the n-type silicon substrate 1, a neutral state is maintained by the electrons (negative charge), which are the majority carriers in the n-type silicon substrate 1, and the donor ions (positive charge) attached to the n-type silicon substrate 1. As a result, the number of holes, which are the minority carriers as the positive charge, is reduced (there are many donor ions). Specifically, when the donor concentration is more than five times the carrier concentration of the n-type silicon substrate 1, the holes are quickly depleted and the oscillating voltage threshold value V.sub.RRO starts to decrease. Therefore, preferably, the carrier concentration of the carrier storage regions 15 and 16 interposed between a plurality of n-type buffer layers 5, 6, and 7 is equal to or more than the carrier concentration of the n-type silicon substrate 1 and equal to or less than five times carrier concentration of the n-type silicon substrate 1.

(40) Next, the range Rp of the protons for forming the deepest n-type buffer layer 5 from the rear surface 1a (that is, a distance Y from the rear surface of the n-type silicon substrate 1 to the peak carrier concentration position 5a of the deepest n-type buffer layer 5) will be described. The deepest n-type buffer layer 5 from the rear surface of the substrate is the n-type buffer layer closest to the p-type anode layer 3. A spatial charge region (the depletion layer in a broad sense) in which the n-type drift layer 2 is spread from the pn junction between the p-type anode layer 3 and the n-type drift layer 2 to the n-type cathode layer 4 during reverse recovery reaches first the deepest n-type buffer layer 5 among the plurality of n-type buffer layers 5, 6, and 7.

(41) The position of the deepest n-type buffer layer 5 is important in order to suppress reverse recovery oscillation.

(42) Next, the net doping concentration distribution of the PiN diode 100 will be described. FIG. 9 is a characteristic diagram illustrating the net doping concentration of the semiconductor device according to Example 1 of the invention. FIG. 9 schematically illustrates the net doping concentration distribution of the cross section of the substrate in the depth direction from the p-type anode layer 3 to the n-type cathode layer 4 of the PiN diode 100. The final thickness (after a manufacturing process is completed) of the n-type silicon substrate 1 from the interface between the anode electrode 8 (not illustrated) and the p-type anode layer 3 which is the front surface of the n-type silicon substrate is simply referred to as the thickness of the substrate and is represented by W.sub.0. The doping concentration of the n-type drift layer 2 is the doping concentration of the n-type silicon substrate 1. The diffusion depth of a pn junction 23 between the p-type anode layer 3 and the n-type drift layer 2 (hereinafter, simply referred to as a pn junction) is represented by xj. The deepest n-type buffer layer 5 is the n-type buffer layer closest to the p-type anode layer 3 and the distance from the pn junction 23 to the maximum concentration position (peak position 5a) of the deepest n-type buffer layer 5 is represented by Z. The rear surface of the n-type silicon substrate 1 is an interface between the n-type cathode layer 4 and the cathode electrode 9 (not illustrated). The depth from the rear surface of the substrate to the peak carrier concentration position 5a of the deepest n-type buffer layer 5 (hereinafter, referred to as a distance from the rear surface of the substrate to the peak position of the deepest n-type buffer layer 5) is represented by Y. In this case, the distance Y from the rear surface of the substrate to the peak position of the deepest n-type buffer layer 5 is represented by Y=W.sub.0−(xj+Z). The distance Y from the rear surface of the substrate to the peak position of the deepest n-type buffer layer 5 is the range Rp when protons are implanted into the rear surface of the substrate.

(43) It is preferable that the distance Y from the rear surface of the substrate to the peak position of the deepest n-type buffer layer 5 be in a numerical range capable of sufficiently increasing the oscillating voltage threshold value V.sub.RRO. First, the distance Z from the pn junction 23 to the peak position of the deepest n-type buffer layer 5 with respect to the thickness W.sub.0 of the substrate is Z=aW.sub.0. Here, a is a non-dimensional coefficient. Then, the relation between the oscillating voltage threshold value V.sub.RRO and Z=aW.sub.0 is examined. FIG. 10 is a characteristic diagram illustrating the relation between the coefficient α and the standardized oscillating voltage threshold value V.sub.RRO. FIG. 10 illustrates the dependence of the oscillating voltage threshold value V.sub.RRO on the coefficient α. In FIG. 10, the vertical axis indicates a value obtained by standardizing the oscillating voltage threshold value V.sub.RRO with a rated voltage V.sub.rate. In the measurement of the oscillating voltage threshold value V.sub.RRO, forward current density J.sub.F was 10 A/cm.sup.2 and the floating inductance L.sub.S of the circuit was 200 nH. In Example 1, the rated voltage V.sub.rate was 1200 V and the rated current density J.sub.rate was 200 A/cm.sup.2. That is, the forward current density J.sub.F is one-twentieth of the rated current density J.sub.rate. The thickness W.sub.0 of the substrate is 120 μm. When the coefficient α is in the range of 0.4 to 0.8, the distance Z from the pn junction 23 to the peak position of the deepest n-type buffer layer 5 is in the range of 40% to 80% of the thickness W.sub.0 of the substrate. As a result, the distance Y from the rear surface of the substrate to the peak position of the deepest n-type buffer layer 5 can be set to the range of 0.2W.sub.0-xj to 0.6W.sub.0-xj by the above-mentioned expression. Therefore, it is possible to suppress the spreading of the spatial charge region at the distance Y from the rear surface of the substrate to the peak position of the deepest n-type buffer layer 5 and at a position which is closer to the rear surface (cathode) of the substrate than the peak position of the deepest n-type buffer layer 5. As a result, the oscillating voltage threshold value V.sub.RRO can be sufficiently close to the rated voltage. In addition, when the coefficient α is in the range of 0.45 to 0.7, preferably, in the range of 0.5 to 0.6, the oscillating voltage threshold value V.sub.RRO is the largest and can be stably increased.

(44) The above-mentioned range of the coefficient α is preferable for the following reason. In general, a power semiconductor element, such as a diode, has a voltage that is about half the rated voltage as the power supply voltage V.sub.CC. When a voltage Vak between both electrodes of the diode (a voltage between the anode and the cathode) is equal to about the power supply voltage V.sub.CC during reverse recovery, the reverse recovery current is the maximum. However, in this case, the width x.sub.0 of the spatial charge region which is spread from the pn junction 23 to the n-type drift layer 2 of the diode is typically about 50% of the thickness W.sub.0 of the substrate. The maximum value of the reverse recovery current also depends on the circuit conditions, but is appropriately equal to the rated current density J.sub.rate. In the spatial charge region, holes are moved to the p-type anode layer 3. Therefore, the speed of the holes is a carrier saturation speed v.sub.sat (about 8×10.sub.6 cm/s) and hole concentration p is substantially constant in the spatial charge region. The average electric field intensity of the spatial charge region is, for example, about 1.5×10.sup.5 V/cm and current density J in the spatial charge region is J=qv.sub.satp (where q is an elementary charge and p is hole concentration).

(45) The average gradient of the electric field intensity E of the spatial charge region is Em/x.sub.0=(q/ss)(p+Nd) from the Poisson's equation (where Em is the maximum electric field intensity and ss is the permittivity of a semiconductor (here, silicon) (11.9×8.sub.0 in silicon; 8.sub.0 is vacuum permittivity and is 8.85×10.sup.−14 F/cm)). Here, Nd is the doping concentration of the n-type silicon substrate 1. As described above, the hole concentration p is substantially constant and is p=J/(qv.sub.sat) from the abovementioned current density J. As described above, the current density J is the maximum value of the reverse recovery current density. Here, for simplicity, the current density J is considered as the rated current density J.sub.rate. In addition, when the reverse recovery current is the maximum, the voltage Vak between both electrodes of the diode is equal to the power supply voltage V.sub.CC. The power supply voltage V.sub.CC is about half the rated voltage. Since the integrated value of the electric field intensity of the spatial charge region is equal to Vak (=V.sub.CC), V.sub.rate/2=(½)Em.Math.x.sub.0.sup.2 is established. When Em is removed from the Poisson's equation, x.sub.0 is represented by the following Equation (3).

(46) $Equation 3$ $\begin{matrix} x_{0} = \sqrt{\frac{{.Math.}_{S} V_{rate}}{(\frac{J_{rate}}{{qv}_{sat}} + N_{d})}} & (2) \end{matrix}$

(47) As described above, x.sub.0 can be referred to as the appropriate width of the spatial charge region during reverse recovery when the rated voltage and the rated current density J.sub.rate are determined. In Example 1 of the invention, x.sub.0 is referred to as a distance index x.sub.0. It is effective to suppress the spreading of the spatial charge region around the distance index x.sub.0 in order to suppress oscillation during reverse recovery. In particular, when the spreading of the spatial charge region is further suppressed at the time the voltage Vak between both electrodes of the diode reaches the power supply voltage V.sub.CC, it is possible to suppress the depletion of the stored carriers (carriers in an electrically neutral region in which the concentration of electrons is substantially equal to that of holes) which are reduced during the subsequent reverse recovery process. As described above, the width of the spatial charge region during reverse recovery is about 50% of the thickness W.sub.0 of the substrate around the distance index x.sub.0. Therefore, the distance Z from the pn junction 23 to the peak position of the deepest n-type buffer layer 5 is set in the range of 0.4W.sub.0 to 0.8W.sub.0 including 0.5W.sub.0, which is half the thickness W.sub.0 of the substrate. That is, the peak position of the deepest n-type buffer layer 5 is disposed in the vicinity of the center of the substrate in the depth direction, or it slightly deviates from the center to the cathode. Therefore, the distance Y from the rear surface of the substrate to the peak position of the deepest n-type buffer layer 5 can be set in the range of (0.2W.sub.0-xj) to (0.6W.sub.0-xj) and it is possible to strongly suppress the oscillation phenomenon.

(48) In addition, the preferred range of the distance Y from the rear surface of the substrate to the peak position of the deepest n-type buffer layer 5 with respect to the distance index x.sub.0 will be described. When the final thickness of the substrate is determined by, for example, grinding from the rear surface of the substrate, protons are implanted into the rear surface in order to form the n-type buffer layer 5. Therefore, the distance Y from the rear surface of the substrate to the peak position of the deepest n-type buffer layer 5 has a great effect on all electrical characteristics of the diode, particularly, switching characteristics such as reverse recovery. First, it is assumed that the distance Z from the pn junction 23 to the peak position of the deepest n-type buffer layer 5 is Z=βx.sub.0. Here, β is a non-dimensional coefficient. FIG. 11 is a characteristic diagram illustrating the relation between the coefficient β and the standardized oscillating voltage threshold value V.sub.RRO in the diode with a rated voltage of 1200 V according to Example 1. FIG. 11 illustrates the dependence of the oscillating voltage threshold value V.sub.RRO on the coefficient β (that is, the dependence of the peak position of the deepest n-type buffer layer 5 from the rear surface of the substrate on the distance index x.sub.0). In FIG. 11, the vertical axis indicates a value obtained by standardizing the oscillating voltage threshold value V.sub.RRO with the rated voltage V.sub.rate. When the coefficient β is in the range of 0.6 to 1.4, the distance Y from the rear surface of the substrate to the peak position of the deepest n-type buffer layer 5 is in the range of 31.8 μm to 80.5 μm and the oscillating voltage threshold value V.sub.RRO has a large value in this range. When the coefficient β is in the range of 0.7 to 1.2, the distance Y from the rear surface of the substrate to the peak position of the deepest n-type buffer layer 5 is in the range of 44.0 μm to 74.4 μm. In the range of the distance Y from the rear surface of the substrate to the peak position of the deepest n-type buffer layer 5, the oscillating voltage threshold value V.sub.RRO can reach 80% or more of the rated voltage V.sub.rate. When the coefficient β is in the range of 0.8 to 1.0, the distance Y from the rear surface of the substrate to the peak position of the deepest n-type buffer layer 5 is in the range of 56.2 μm to 68.3 μm. In the range of the distance Y from the rear surface of the substrate to the peak position of the deepest n-type buffer layer 5, it is possible to stably increase the oscillating voltage threshold value V.sub.RRO. In addition, FIG. 11 illustrates the relation between the coefficient β and the standardized oscillating voltage threshold value V.sub.RRO when the rated voltage is 600 V and 3300 V. When the rated voltage is 600 V and 3300 V, it is possible to stably increase the oscillating voltage threshold value V.sub.RRO in the same range of the coefficient β as that when the rated voltage is 1200 V.

(49) As described above, for the distance Y from the rear surface of the substrate to the peak position of the deepest n-type buffer layer 5, when the coefficient α, which is the ratio of the distance Z from the pn junction 23 to the peak position of the deepest n-type buffer layer 5 to the thickness W.sub.0 of the substrate, or the coefficient β, which is the ratio of the distance Z from the pn junction 23 to the peak position of the deepest n-type buffer layer 5 to the distance index x.sub.0, is in a predetermined range, it is possible to improve the effect of suppressing oscillation. In the above-mentioned example, the coefficient α and the coefficient β seem to be equivalent to the oscillating voltage threshold value V.sub.RRO. However, the coefficient α and the coefficient β are not equivalent to the oscillating voltage threshold value V.sub.RRO and are independent from each other. For example, when the thickness W.sub.0 of the substrate is greater than 120 μm which is the thickness of the substrate in Example 1, in some cases, the oscillating voltage threshold value V.sub.RRO shifts to the value less than a=0.5. The reason is that the dependence of the oscillating voltage threshold value V.sub.RRO on the coefficient β of the distance index x.sub.0 is less than the dependence of the oscillating voltage threshold value V.sub.RRO on the coefficient α and the distance index x.sub.0 is relatively close to the position of the p-type anode layer 3.

(50) With an increase in the thickness of the substrate, the region in which the stored carriers are present expands in the depth direction of the substrate. Therefore, the depletion of the carriers during reverse recovery is further suppressed, but a forward voltage drop or reverse recovery loss increases with the increase in the thickness of the substrate. When the thickness W.sub.0 of the substrate is less than 120 μm, the distance index x.sub.0 is relatively moved to the n-type cathode layer 4 by a value corresponding to the reduction in the thickness W.sub.0 of the substrate. Therefore, the oscillating voltage threshold value V.sub.RRO shifts to a value greater than a=0.5.

(51) In addition to the thickness of the substrate, for the forward current density J.sub.F when reverse recovery starts and the floating inductance L.sub.S of the circuit, the relation between the coefficients α and β and the oscillating voltage threshold value V.sub.RRO is slightly changed. However, when the coefficient α and the coefficient β are in the above-mentioned ranges, it is possible to sufficiently increase the oscillating voltage threshold value V.sub.RRO in the wide range of each of the forward current density J.sub.F and the floating inductance L.sub.S of the circuit and to suppress oscillation.

(52) This numerical range can also be applied to other rated voltages. FIG. 15 illustrates the values of the distance index x.sub.0 and the distance Y from the rear surface of the substrate to the peak position of the deepest n-type buffer layer 5 when the coefficient β is applied to the distance index x.sub.0 at various rated voltages. FIG. 15 is a table illustrating the distance Y from the rear surface of the substrate to the peak position of the deepest n-type buffer layer 5 at each rated voltage. Here, Y=W.sub.0−(βx.sub.0+X.sub.j) and X.sub.j=3.0 μm are established.

(53) It was confirmed that the coefficient β depended on the rated voltage, but it was possible to increase the oscillating voltage threshold value V.sub.RRO at any rated voltage when the coefficient β was in the above-mentioned numerical range, as illustrated in FIG. 11. It is preferable that the coefficient β be in the range of 0.6 to 1.4 in order to maintain the sufficiently large oscillating voltage threshold value V.sub.RRO. When the coefficient β is in the range of 0.7 to 1.2, preferably, in the range of 0.8 to 1.0, it is possible to stably maintain the sufficiently large oscillating voltage threshold value V.sub.RRO close to the rated voltage V.sub.rate at any rated voltage.

(54) In Example 1, three n-type buffer layers 5, 6, and 7 are formed. However, the number of n-type buffer layers is not limited to three, but two n-type buffer layers or four or more n-type buffer layers may be formed. In particular, it is preferable to increase the number of n-type buffer layers in order to increase the breakdown voltage of the element. In this case, preferably, for at least the deepest n-type buffer layer 5, for example, the relation between the first width W1 and the second width W2, the carrier concentration of the carrier storage region 15, the carrier concentration distribution, and the distance Y from the rear surface of the substrate (proton implantation surface) to the deepest n-type buffer layer 5 are in the abovementioned preferred ranges.

(55) For example, in some cases, the oscillating voltage threshold value V.sub.RRO is greater than the rated voltage V.sub.rate, depending on the value of the coefficient α or the coefficient β or circuit conditions. For example, the oscillating voltage threshold value V.sub.RRO may be sufficiently close to the breakdown voltage.

COMPARATIVE EXAMPLE

(56) The comparative example (dotted line) illustrated in FIG. 2 in which a heat treatment is performed at 380° C. after proton implantation will be described. In the comparative example, the ranges of proton implantation for forming three n-type buffer layers from the rear surface of the substrate are about 50 μm, 25 μm, and 10 μm. In addition, the impurity concentration (carrier concentration) of the n-type silicon substrate 1 is less than that in Example 1.

(57) Since proton implantation for forming the deepest n-type buffer layer is performed with high acceleration energy, a large number of crystal defects occur over a wide width (thickness) in the vicinity of a range of about 50 μm from the rear surface 1a of the n-type silicon substrate 1. The crystal defects are not sufficiently recovered by the heat treatment which is performed at 380° C. Therefore, the carrier concentration of the deepest n-type buffer layer is significantly less than the carrier concentration of the n-type silicon substrate 1 on the near side of the peak position of the carrier concentration (between the deepest n-type buffer layer and the n-type buffer layer close to the cathode side of the deepest n-type buffer layer). In addition, a first width W1′ from the peak position of the deepest n-type buffer layer to the anode is less than a second width W2′ from the peak position of the deepest n-type buffer layer to the cathode. That is, in the comparative example, the crystal defects are not sufficiently recovered, which results in an increase in leakage current. In addition, since the lifetime is short and the stored carriers are reduced, holes are quickly depleted. When the spatial charge region reaches the deepest n-type buffer layer, the rate of change of voltage dV/dt increases rapidly. Therefore, since the spreading of the spatial charge region is strongly suppressed, the voltage and current oscillate during reverse recovery and radiation noise is generated.

(58) As described above, according to Example 1, the spreading of the depletion layer is suppressed, the breakdown voltage is ensured, and generation loss is reduced. It is possible to suppress the oscillation of voltage and current during a switching operation. In addition, according to Example 1, since the crystal defects are recovered, it is possible to reduce the amount of leakage current and thus reduce the risk of thermal runaway during a high-temperature operation.

Example 2

(59) Next, a method for manufacturing the semiconductor device illustrated in FIGS. 1A and 1B will be described as Example 2. FIGS. 4 to 7 are cross-sectional views illustrating the state of the semiconductor device according to Example 2 of the invention which is being manufactured. First, as illustrated in FIG. 4, for example, the p-type anode layer 3 and the p-type layer 12 of the high-breakdown-voltage junction termination structure 11 are selectively formed in a surface layer of one main surface of the n-type silicon substrate 1 with a thickness of about 500 μm. Then, the insulating film 14, which is an oxide film, is formed on the surface of a region of the n-type silicon substrate 1 which is interposed between the p-type anode layer 3 and the p-type layer 12 and on the surface of a region interposed between the p-type layers 12. Then, the anode electrode 8 is formed on the p-type anode layer 3 and the termination electrode 13 is formed on the p-type layer 12. Then, the other surface (the rear surface 1b of the n-type silicon substrate 1) of the n-type silicon substrate 1 is ground and polished and the thickness of the n-type silicon substrate 1 is reduced to about 120 μm.

(60) Then, as illustrated in FIG. 5, proton implantation (G1, G2, and G3) is performed three times on the ground surface of the n-type silicon substrate 1 (the rear surface 1a of the n-type silicon substrate 1) with a dose (E1) of about 1×10.sup.13 cm.sup.−2 at acceleration energy levels of 2.2 MeV, 1.5 MeV, and 0.5 MeV. In the proton implantation (G1, G2, and G3), the ranges (P1, P2, and P3) of protons 22 from the implantation surface (the rear surface 1a of the n-type silicon substrate 1) are about 60 μm, 25 μm, and 6 μm.

(61) Then, as illustrated in FIG. 6, the n-type silicon substrate 1 subjected to the proton implantation (G1, G2, and G3) is set in an annealing furnace 18 and a heat treatment is performed, for example, at a temperature of 420° C. for 3 hours. Crystal defects which occur in the passage region of the protons 22 are recovered by the heat treatment and the protons 22 introduced into the n-type silicon substrate 1 are changed into donors. In this way, the n-type buffer layers 5, 6, and 7 illustrated in FIG. 2 are formed.

(62) Then, as illustrated in FIG. 7, the n-type silicon substrate 1 is taken out from the annealing furnace 18 and the n-type cathode layer 4 is formed in a surface layer of the other surface (rear surface 1a) of the n-type silicon substrate 1. Then, the cathode electrode 9 is formed on the n-type cathode layer 4. In this way, the PiN diode 100 illustrated in FIGS. 1A and 1B is completed. Since the heat treatment for forming the n-type cathode layer 4 is performed by, for example, laser annealing, it has no effect on each of the n-type buffer layers 5, 6, and 7 which are formed at positions deeper than the n-type cathode layer 4 from the rear surface 1a of the n-type silicon substrate 1. In addition, when a plurality of n-type buffer layers are formed after the other surface of the n-type silicon substrate 1 is ground to reduce the thickness of the n-type silicon substrate 1 (step illustrated in FIG. 4), the n-type cathode layer 4 may be formed in the surface layer of the other surface (rear surface 1a) of the n-type silicon substrate 1 by proton implantation.

(63) The carrier concentration peak position 5a of the deepest n-type buffer layer 5 is formed at a depth of 60 μm from the rear surface 1a of the n-type silicon substrate 1 by the above-mentioned manufacturing process. In addition, the carrier concentration of the carrier storage regions 15 and 16 interposed between the n-type buffer layers 5, 6, and 7 is more than the carrier concentration of the n-type silicon substrate 1 and is flat. In the carrier concentration distribution of the n-type buffer layers 5, 6, and 7, the first width W1 from the carrier concentration peak position 5a of the deepest n-type buffer layer 5 to the anode is more than the second width W2 from the carrier concentration peak position 5a of the deepest n-type buffer layer 5 to the cathode. The reason is that the acceleration energy of the proton implantation G1 for forming the deepest n-type buffer layer 5 is high and a proton implantation amount distribution extends from the peak position to the anode and the cathode. Since crystal defects can occur in the region in which the proton 22 is present, they are also spread from the peak position of the proton implantation amount to the anode. The crystal defects which are spread to the anode to be deeper than the peak position of the proton implantation amount are recovered. In this region, the protons are also changed into donors. Therefore, in the carrier concentration distribution of the deepest n-type buffer layer 5, the first width W1 from the peak position 5a to the anode is more than the second width W2 from the peak position 5a to the cathode.

(64) A heat treatment was performed at a temperature of about 420° C. after the proton implantation (G1, G2, and G3) and the carrier concentration distribution of the n-type buffer layers 5, 6, and 7 was measured by a known spreading resistance measurement method (SR method). FIG. 2 illustrates the measurement result. When the heat treatment is performed at a temperature of 420° C., the carrier concentration of the carrier storage regions 15 and 16 interposed between the n-type buffer layers 5, 6, and 7 is equal to or more than the carrier concentration of the n-type silicon substrate 1. In the carrier concentration distribution of the deepest n-type buffer layer 5, the first width W1 from the peak position 5a to the anode is more than the second width W2 from the peak position 5a to the cathode.

(65) In the heat treatment after the proton implantations (G1, G2, and G3), the heat treatment temperature may be in the range of 400° C. to 500° C. and the heat treatment time may be in the range of 1 hour to 10 hours. Preferably, the heat treatment temperature is in the range of 420° C. to 450° C. and the heat treatment is in the range of 1 hour to 3 hours.

(66) The inventors found that the relation between the range Rp (the peak position 5a of the deepest n-type buffer layer 5, unit: μm) of protons in silicon and the acceleration energy E (eV) of the protons was represented by the following Equation (4) when log(Rp) was x and log(E) of was y.
y=−0.0047x.sup.4+0.0528x.sup.3−0.2211x.sup.2+0.9923x+5.0474 Equation (4)

(67) FIG. 14 is a graph illustrating the above-mentioned Equation (4). FIG. 14 is a characteristic diagram illustrating the relation between the logarithm (log(E)) of the acceleration energy E of the protons and the logarithm (log(Rp)) of the range Rp of the protons in the proton implantation according to Example 2 of the invention. In FIG. 14, the horizontal axis is log(Rp) and indicates the range Rp (μm) corresponding to the parentheses below the axis value of log(Rp). In addition, in FIG. 14, the vertical axis is log(E) of and indicates the acceleration energy E corresponding to the parentheses on the left side of the axis value of log(E). The above-mentioned Equation (4) is obtained by fitting the value of log(Rp) and the value of the logarithm log(E) of the acceleration energy with a four-order polynomial of x (=log(Rp)).

(68) The use of the above-mentioned Equation 4 makes it possible to calculate the acceleration energy E of the protons required to obtain a desired distance Y from the rear surface of the substrate to the peak position of the deepest n-type buffer layer 5, using the distance Y as the range Rp. When a sample in which protons are implanted with the acceleration energy E that is calculated and set from a predetermined range Rp by the above-mentioned Equation (4) is actually measured by the known spreading resistance measurement method (SR method), the measured value is exactly equal to a predetermined value. In addition, at each rated voltage illustrated in FIG. 15, even when the deepest n-type buffer layer 5 corresponding to the preferred value (range) of the distance Y from the rear surface of the substrate to the peak position of the deepest n-type buffer layer 5 is formed, it is possible to calculate the acceleration energy E using the above-mentioned Equation 4. For example, at each rated voltage illustrated in FIG. 15, the largest distance Y from the rear surface of the substrate to the peak position of the deepest n-type buffer layer 5 is 454.4 μm which corresponds to a rated voltage of 6500 V and β=0.6. The acceleration energy E of the protons corresponding to the distance Y is 7.6 MeV, that is, about 8.0 MeV. The smallest distance Y from the rear surface of the substrate to the peak position of the deepest n-type buffer layer 5 is 17.5 μm which is slightly greater than 15 μm from the rear surface of the substrate and corresponds to, for example, a rated voltage of 600 V and β=1.3. The acceleration energy E of the protons corresponding to the distance Y from the rear surface of the substrate to the peak position of the deepest n-type buffer layer 5 is 1.1 MeV, that is, about 1.0 MeV. Therefore, for example, the acceleration energy E of the protons is preferably equal to or more than 1.0 MeV and equal to or less than 8.0 MeV.

(69) The following relation may be considered between the actual acceleration energy E′ and the actual range Rp′ (the peak position of the carrier concentration of the deepest n-type buffer layer 5) which is measured by, for example, the spreading resistance (SR) measurement method when the above-mentioned fitting equation represented by the above-mentioned Equation (4) is used to set the acceleration energy E of proton implantation from the desired range Rp of the protons and the protons are implanted into silicon with the set acceleration energy. That is, when the actual acceleration energy E′ is in the range of about ±5% of the calculated acceleration energy E, the actual average range Rp′ is within the range of about ±5% of the calculated range Rp and is in a measurement error range. Therefore, the influence of the deviation of the actual average range Rp′ from the calculated range Rp on the characteristics is sufficiently reduced. When the actual acceleration energy E is in the range of ±5% of the calculated acceleration energy E, it is possible to determine that the actual average range Rp′ is substantially equal to the set range Rp. In the actual accelerator, since the acceleration energy E and the average range Rp are both in the above-mentioned measurement error range (±5%), it is considered that the actual acceleration energy E′ and the actual range Rp′ follow the above-mentioned Equation (4) represented by the desired range Rp and the calculated acceleration energy E and no problem occurs.

(70) At each rated voltage illustrated in FIG. 15, the acceleration energy E of proton implantation for forming the deepest n-type buffer layer 5 may be calculated and set by the fitting equation represented by the above-mentioned Equation 4 so as to correspond to the value (range) of the coefficient α or the coefficient β, using the distance Y from the rear surface of the substrate to the peak position of the deepest n-type buffer layer 5 as the range Rp. Specifically, for example, preferably, the acceleration energy E has the following value, on the basis of the range of the coefficient β.

(71) When the rated voltage is 600 V, for example, the acceleration energy E of proton implantation for forming the deepest n-type buffer layer 5 is in the range of 1.1 MeV to 1.8 MeV, preferably, in the range of 1.2 MeV and 1.7 MeV, and more preferably, in the range of 1.4 MeV to 1.6 MeV. When the rated voltage is 1200 V, for example, the acceleration energy E of proton implantation for forming the deepest n-type buffer layer 5 is in the range of 1.6 MeV to 2.8 MeV, preferably, in the range of 1.9 MeV to 2.6 MeV, and more preferably, in the range of 2.2 MeV to 2.5 MeV.

(72) When the rated voltage is 1700 V, for example, the acceleration energy E of proton implantation for forming the deepest n-type buffer layer 5 is in the range of 1.9 MeV to 3.4 MeV, preferably, in the range of 2.4 MeV to 3.3 MeV, and more preferably, in the range of 2.8 MeV to 3.1 MeV. When the rated voltage is 3300 V, for example, the acceleration energy E of proton implantation for forming the deepest n-type buffer layer 5 is in the range of 3.0 MeV to 5.1 MeV, preferably, in the range of 3.6 MeV to 4.8 MeV, and more preferably, in the range of 4.1 MeV to 4.6 MeV.

(73) When the rated voltage is 4500 V, for example, the acceleration energy E of proton implantation for forming the deepest n-type buffer layer 5 is in the range of 3.7 MeV to 6.1 MeV, preferably, in the range of 4.4 MeV to 5.8 MeV, and more preferably, in the range of 5.0 MeV to 5.6 MeV. When the rated voltage is 6500 V, for example, the acceleration energy E of proton implantation for forming the deepest n-type buffer layer 5 is in the range of 4.7 MeV to 7.6 MeV, preferably, in the range of 5.5 MeV to 7.3 MeV, and more preferably, in the range of 6.2 MeV to 6.9 MeV. The rated voltage is not limited to the above-mentioned values. When the rated voltage is, for example, 400 V, 1400 V, and 2500 V other than the abovementioned values, the acceleration energy E may be set from the desired range Rp by the same method as described above.

(74) The use of the above-mentioned Equation 4 makes it possible to calculate the necessary acceleration energy E from the range Rp of proton implantation with high accuracy and to set the acceleration energy during proton implantation. In this way, it is possible to suppress the oscillation of the voltage and current during the reverse recovery of the PiN diode 100. In addition, the spreading of the depletion layer is suppressed and it is possible to ensure the breakdown voltage even when the thickness of the n-type silicon substrate 1 is reduced. It is possible to reduce generation loss. The soft recovery characteristics of the PiN diode 100 are improved and a surge voltage and radiation noise are reduced. In addition, the crystal defects caused by proton implantation are recovered and it is possible to reduce a leakage current. Therefore, it is possible to reduce the risk of thermal runaway during a high-temperature operation.

(75) In Example 2, proton implantation is performed three times. However, the number of proton implantation processes can be set to any value. As described above, according to Example 2, it is possible to obtain the same effect as that in Example 1.

Example 3

(76) Next, the structure of an insulated gate bipolar transistor (IGBT) will be described as an example of the structure of a semiconductor device according to Example 3 of the invention. FIGS. 12A and 12B are diagrams illustrating the structure of the semiconductor device according to Example 3 of the invention. FIG. 12A is a cross-sectional view illustrating a main portion of the semiconductor device according to Example 3. FIG. 12B illustrates a net doping concentration distribution along the cutting line A-A′ of FIG. 12A.

(77) As illustrated in FIGS. 12A and 12B, an n.sup.− drift layer 2 which is an n.sup.− semiconductor substrate is provided and a p-type base layer 33 which has a higher impurity concentration than the n.sup.− drift layer 2 is provided in a surface layer of a front surface of the n.sup.− semiconductor substrate. An n.sup.+ emitter layer 34 which has a higher impurity concentration than the p-type base layer 33 is provided in the p-type base layer 33. A gate electrode 42 is formed so as to face three layers, that is, the n.sup.+ emitter layer 34, the p-type base layer 33, and the n.sup.− drift layer 2, with a gate insulating film 43 interposed therebetween. In this way, for example, a known trench-gate-type MOS gate (metal-oxide film-semiconductor insulating gate) structure is provided. Instead of the trench-gate-type MOS gate structure, a known planar-gate-type MOS gate structure may be provided. An emitter electrode 31 is formed on the front surface of the n.sup.− semiconductor substrate so as to come into contact with the n.sup.+ emitter layer 34 and the p-type base layer 33. The emitter electrode 31 is electrically insulated from the gate electrode 42 by an interlayer insulating film 41.

(78) A p-type collector layer 39 and an n-type field stop layer 38 are formed on the rear surface side of the n.sup.− semiconductor substrate. The n-type field stop layer 38 is provided at a position that is deeper than the p-type collector layer 39 from the rear surface of the substrate so as to come into contact with the p-type collector layer 39. The n-type field stop layer 38 has a function of suppressing holes which mainly flow from the p-type collector layer 39 in an off state to suppress a leakage current. A collector electrode 32 comes into contact with the p-type collector layer 39. Similarly to Example 1, a plurality of n-type buffer layers 35, 36, and 37 are formed in the n.sup.− drift layer 2 at different depths from the rear surface of the substrate by proton implantation and a heat treatment. Carrier storage regions 45, 46, and 47 which have a lower impurity concentration than the plurality of n-type buffer layers 35, 36, and 37 and the n-type field stop layers 38 are formed between the plurality of n-type buffer layers 35, 36, and 37 and the n-type field stop layer 38 which are adjacent to each other in the depth direction. The n-type field stop layer 38 may be doped with n-type impurities. For example, the n-type field stop layer 38 may include impurities such as phosphorus (P), arsenic (As), or proton (hydrogen).

(79) The distance Y of the position of the peak carrier concentration of the deepest n-type buffer layer 35 from the interface between the p-type collector layer 39 and the collector electrode 32 (the distance from the rear surface of the substrate to the peak position of the deepest n-type buffer layer 35) Y is preferably equal to that in Example 1. In the case of the IGBT, the direction of a current in an on state is from the p-type collector layer 39 to the n.sup.+ emitter layer 34, unlike the reverse recovery of the diode. However, when the on-current is turned off, the spatial charge region extends from the pn junction between the p-type base layer 33 close to the front surface of the substrate and the n-type drift layer 2 to the p-type collector layer 39 on the rear surface of the substrate. When the semiconductor device is turned off, holes pass through the spatial charge region in the direction from the rear surface to the front surface of the substrate. The spatial charge region of the diode and the operation of the holes are similar to each other in these two points. Therefore, turn-off oscillation occurs due to the same physical cause as the reverse recovery oscillation in Example 1. In contrast, it is preferable that the distance Y from the rear surface of the substrate to the peak position of the deepest n-type buffer layer 35 be equal to that in Example 1.

(80) As described above, according to Example 3, even when the semiconductor device is an IGBT, it is possible to delay the depletion of holes at the time the semiconductor device is turned off, to suppress the oscillation of the voltage and current at the time the semiconductor device is turned off, and to suppress the generation of radiation noise, similarly to Example 1.

(81) Thus, according to the invention, proton implantation is performed a plurality of times to form a plurality of n-type buffer layers (5, 6, 7) in an n-type drift layer (2) at different depths from a rear surface of a substrate. The depth of the n-type buffer layer (5), which is provided at the deepest position from the rear surface of the substrate, from the rear surface of the substrate is more than 15 pm. The temperature of a heat treatment which is performed in order to change a proton into a donor and to recover a crystal defect after the proton implantation is equal to or higher than 400° C. In a carrier concentration distribution of the n-type buffer layer (5), a width from the peak position (5a) of carrier concentration to an anode is more than a width from the peak position (5a) to a cathode. The carrier concentration of regions (15, 16) interposed between the n-type buffer layers (5, 6, 7) is flat and is equal to or more than the carrier concentration of an n-type silicon substrate (1) and equal to or less than five times the carrier concentration of the n-type silicon substrate (1). Therefore, it is possible to ensure a breakdown voltage and to reduce generation loss. It is possible to suppress the oscillation of voltage and current during a switching operation. In addition, it is possible to recover the crystal defect and to reduce a leakage current. Accordingly, it is possible to reduce the risk of thermal runaway.

(82) The invention is not limited to the above-described embodiments, but various modifications and changes of the invention can be made without departing from the scope and spirit of the invention. As described above, the semiconductor device and the semiconductor device manufacturing method according to the invention are useful for power conversion devices, such as converters, or power supply devices of various industrial machines.

Semiconductor device and method for manufacturing the same

Assignee

Inventors

Cpc classification

Classification Explorer

H01L29/868

ELECTRICITY

Classification Explorer

H01L21/02351

ELECTRICITY

Classification Explorer

H01L29/167

ELECTRICITY

Classification Explorer

H01L21/263

ELECTRICITY

Classification Explorer

H01L29/7395

ELECTRICITY

Classification Explorer

H01L29/404

ELECTRICITY

Classification Explorer

H01L29/1095

ELECTRICITY

Classification Explorer

H01L29/32

ELECTRICITY

Classification Explorer

H01L29/66136

ELECTRICITY

Classification Explorer

H01L21/02304

ELECTRICITY

Classification Explorer

H01L29/365

ELECTRICITY

Classification Explorer

H01L29/7397

ELECTRICITY

Classification Explorer

H01L29/402

ELECTRICITY

Classification Explorer

H01L29/36

ELECTRICITY

Classification Explorer

H01L29/66348

ELECTRICITY

Classification Explorer

H01L29/8611

ELECTRICITY

Classification Explorer

H01L29/66333

ELECTRICITY

Classification Explorer

H01L29/0619

ELECTRICITY

International classification

Classification Explorer

H01L29/36

ELECTRICITY

Classification Explorer

H01L21/02

ELECTRICITY

Classification Explorer

H01L21/263

ELECTRICITY

Classification Explorer

H01L29/06

ELECTRICITY

Classification Explorer

H01L29/10

ELECTRICITY

Classification Explorer

H01L29/167

ELECTRICITY

Classification Explorer

H01L29/32

ELECTRICITY