JP4637503B2 - Manufacturing method of nitride semiconductor laser device - Google Patents

Description

The present invention relates to a nitride semiconductor laser device using a nitride semiconductor (In _X Al _Y Ga _1-XY N, 0 ≦ X, 0 ≦ Y, X + Y ≦ 1).

  In general, when a semiconductor is grown on a substrate, it is known that if a substrate lattice-matched with the semiconductor to be grown is used, the crystal defects of the semiconductor are reduced and the crystallinity is improved. However, nitride semiconductors are generally grown on heterogeneous substrates that do not lattice match with nitride semiconductors such as sapphire, spinel, and silicon carbide because there is no lattice-matched substrate in the world.

  On the other hand, attempts to fabricate a GaN bulk crystal that is lattice-matched with a nitride semiconductor have been made by various research institutions, but only a few millimeters have been reported so far. Is far away.

  As a technique for producing a GaN substrate, for example, in JP-A-7-202265 and JP-A-7-165498, a buffer layer made of ZnO is formed on a sapphire substrate, and a nitride semiconductor is formed on the buffer layer. A technique for dissolving and removing the buffer layer after growth is described. However, the crystallinity of the ZnO buffer layer grown on the sapphire substrate is poor, and it is difficult to obtain a good quality nitride semiconductor crystal even if a nitride semiconductor is grown on the buffer layer. Furthermore, it is difficult to continuously grow a nitride semiconductor having a thick film as a substrate on a buffer layer made of thin ZnO.

JP-A-7-202265 JP 7-165498 A

If a substrate made of a nitride semiconductor can be manufactured when manufacturing an LD element, a new nitride semiconductor can be grown on the substrate, and a nitride semiconductor with few lattice defects can be grown. Crystallinity is dramatically improved, and an element that has not been realized can be realized.
Accordingly, an object of the present invention is to provide a new method for manufacturing a nitride semiconductor laser device based on a growth method capable of growing a nitride semiconductor having good crystallinity.

That is, in the method for manufacturing a nitride semiconductor laser device according to the present invention, a nitride semiconductor layer doped with an n-type impurity is laterally grown on a heterogeneous substrate having a main surface with an off angle of 1 ° or less. The nitride semiconductor layer is grown to 70 to 500 μm, and then the heterogeneous substrate is removed to form a nitride semiconductor substrate. The nitride semiconductor substrate has a ridge stripe, and the surface of the ridge stripe A laser device structure in which a p-electrode is formed on the entire surface is formed so that a (11-00) plane of the nitride semiconductor layer becomes a resonance surface, an electrode is formed on the back surface of the nitride semiconductor substrate, and the laser device A dielectric film is formed on the resonance surface of the structure.

Since the nitride semiconductor device of the present invention is formed by laminating a nitride semiconductor serving as a device structure on a first nitride semiconductor with few crystal defects obtained by lateral growth , the crystallinity of the device structure As a result, a nitride semiconductor laser device having good performance is obtained.

Previously nitride semiconductor Despite currently being evaluated as an ideal semiconductor, for the nitride semiconductor substrate is not present, is practiced in grown many lattice defects nitride semiconductor device on a foreign substrate was Tei. For this reason, when a device such as a laser element in which crystal defects immediately affect the lifetime is realized, the element is damaged in several tens of hours. However, according to this onset bright, for realizing a nitride semiconductor substrate that could not be conventionally grown by lateral growth, on the nitride semiconductor substrate, laminating a nitride semiconductor layer made of an element structure, lattice defects Thus, a nitride semiconductor laser device with a very small amount can be realized , and a device that has almost reached a practical level can be obtained. Thus, the fact that the nitride semiconductor laser element which has not been conventionally possible can be obtained by the present invention has a great industrial utility value.

  FIG. 13 specifically shows an embodiment of the nitride semiconductor laser element of the present invention. As shown in FIG. 13, the nitride semiconductor laser device according to the present invention includes a nitride semiconductor substrate 4, a laser device structure 43-49 having a ridge stripe formed on the nitride semiconductor substrate 4, and a nitride. A nitride semiconductor laser device having an electrode 53 formed on the back surface of a semiconductor substrate 4 and a dielectric film formed on a resonance surface of the laser device structure, wherein the nitride semiconductor substrate 4 crosses over a heterogeneous substrate. The nitride semiconductor layer is formed while growing in the direction, and the (11-00) plane of the nitride semiconductor layer is a resonance plane.

  In the present embodiment, the nitride semiconductor substrate for forming the laser element structure preferably has the following configurations (1) to (8).
(1) A first protective film is partially formed on a nitride semiconductor grown on a different substrate made of a material different from that of the nitride semiconductor, and the first protective film is formed on the first protective film. The nitride semiconductor is grown, and the nitride semiconductor that forms the device structure is stacked on the first nitride semiconductor.
(2) The first protective film is formed to have a surface area larger than the surface area of the portion where the first protective film is not formed.
(3) A first protective film is partially formed on a dissimilar substrate made of a material different from a nitride semiconductor and has a surface area larger than a surface area of a portion where the first protective film is not formed. A first nitride semiconductor is grown on the first protective film, and a nitride semiconductor that forms an element structure is stacked on the first nitride semiconductor.
(4) The first nitride semiconductor has a total film thickness of 70 μm or more, and the heterogeneous substrate is removed.
(5) The first nitride semiconductor is grown on a first protective film having a stripe shape with a window width of 5 μm or less.
(6) The nitride semiconductor having the element structure has an n-side nitride semiconductor having a superlattice structure.
(7) An n-electrode is formed on the n-side nitride semiconductor having the superlattice structure.
(8) The ratio Ws / Ww of the width (Ww) of the window and the width (Ws) of the protective film is 1 to 20.

  When the first protective film is formed by laminating the element structure on the first nitride semiconductor formed having a surface area larger than the surface area of the portion where the first protective film is not formed, Since there are few crystal defects appearing on the surface of the first nitride semiconductor, a nitride semiconductor laser device having even better performance can be obtained, which is preferable.

  Furthermore, it is preferable that the first protective film has a stripe shape with a window portion having a width of 5 μm or less because crystal defects of the first nitride semiconductor are further reduced, and device performance is further improved. In addition, an n-side nitride semiconductor having a superlattice structure is formed as a nitride semiconductor serving as an element structure, an n-electrode is formed on the n-side nitride semiconductor having the superlattice structure, and the width of the window (Ww ) And the protective film width (Ws) ratio Ws / Ww is 1 to 20, the effect of the present invention can be more easily obtained.

  The nitride semiconductor laser element according to the present embodiment has the following structure on nitride semiconductor substrate 4.
(Buffer layer 3)
  The second nitride semiconductor layer 4 manufactured by the lateral growth method described later is polished and removed from the sapphire substrate, and the wafer having the second nitride semiconductor layer 4 (Si-doped GaN) as the main surface is reacted by the MOVPE apparatus. A third buffer layer 41 made of GaN doped with Si is grown on a surface opposite to the surface exposed by removing the sapphire substrate 1 and the like set in the container. The third buffer layer 41 is preferably a strained superlattice layer formed by stacking nitride semiconductors having different compositions, each having a thickness of 100 angstroms or less, more preferably 70 angstroms or less, and most preferably 50 angstroms or less. When the strained superlattice layer is used, the crystallinity of the single nitride semiconductor layer is improved, so that a high-power laser element can be realized.

(Crack prevention layer 42)
  Next, a crack prevention layer 42 made of InGaN doped with Si is grown. The crack prevention layer 42 can be prevented from being cracked in the nitride semiconductor layer containing Al by growing it with an n-type nitride semiconductor containing In, preferably InGaN. The crack prevention layer is preferably grown with a film thickness of 100 Å or more and 0.5 μm or less. If it is thinner than 100 angstroms, it is difficult to act as a crack prevention as described above, and if it is thicker than 0.5 μm, the crystal itself tends to turn black. The crack prevention layer 42 can be omitted.

(N-side cladding layer 43)
  Next, the n-side cladding layer 43 functions as a carrier confinement layer and an optical confinement layer, and is preferably a nitride semiconductor containing Al, preferably a superlattice layer containing AlGaN. It is desirable to grow the entire superlattice layer with a thickness of 100 Å or more and 2 μm or less, more preferably 500 Å or more and 1 μm or less. When the superlattice layer is used, a carrier confinement layer having good crystallinity without cracks can be formed.

(N-side light guide layer 44)
  Next, an n-type light guide layer 44 made of n-type GaN doped with Si is grown. The n-side light guide layer 44 acts as a light guide layer of the active layer, and it is desirable to grow GaN and InGaN. Usually, the n-side light guide layer 44 is grown to a thickness of 100 Å to 5 μm, more preferably 200 Å to 1 μm. desirable. The n-side light guide layer 44 is usually doped with an n-type impurity such as Si or Ge so as to have an n-type conductivity type, but can be particularly undoped. In the case of a superlattice, at least one of the first layer and the second layer may be doped with an n-type impurity or may be undoped.

(Active layer 45)
  Next, undoped In _0.2 Ga _0.8 N well layer, 25 Å, and undoped In _0.01 Ga _0.95 An active layer 45 of a multiple quantum well structure (MQW) having a total film thickness of 175 Å formed by alternately laminating N barrier layers and 50 Å is grown. The well layer and / or the barrier layer may be doped with Si, and doping with Si is preferable because the threshold value is lowered.

(P-side cap layer 46)
  Next, Mg-doped p-type Al having a band gap energy larger than that of the p-side light guide layer 47 and larger than that of the active layer 45. _0.3 Ga _0.9 A p-side cap layer 46 made of N is grown. The p-side cap layer 46 is p-type, but since it is thin, it may be i-type in which carriers are compensated by doping n-type impurities, or undoped, and most preferably a layer doped with p-type impurities. And The film thickness of the p-side cap layer 17 is adjusted to 0.1 μm or less, more preferably 500 angstroms or less, and most preferably 300 angstroms or less. This is because if the film is grown to a thickness greater than 0.1 μm, cracks are likely to occur in the p-type cap layer 46, and a nitride semiconductor layer with good crystallinity is difficult to grow. When the AlGaN having a larger Al composition ratio is formed thinner, the LD element tends to oscillate. For example, Al with a Y value of 0.2 or more _Y Ga _1-Y If N, it is desirable to adjust to 500 angstroms or less. Although the lower limit of the film thickness of the p-side cap layer 46 is not particularly limited, it is desirable to form the p-side cap layer 46 with a film thickness of 10 angstroms or more.

(P-side light guide layer 47)
  Next, a p-side light guide layer 47 made of p-type GaN doped with Mg having a band gap energy smaller than that of the p-side cap layer 46 is grown. This layer acts as a light guide layer of the active layer, and is preferably grown of GaN and InGaN as with the n-side light guide layer 44. This layer also functions as a buffer layer when the p-side cladding layer 48 is grown, and functions as a preferable light guide layer by growing it at a film thickness of 100 angstroms to 5 μm, more preferably 200 angstroms to 1 μm. . This p-side light guide layer is usually doped with a p-type impurity such as Mg to have a p-type conductivity, but it is not particularly necessary to dope the impurity. The p-type light guide layer can be a superlattice layer. In the case of a superlattice layer, at least one of the first layer and the second layer may be doped with a p-type impurity or may be undoped.

(P-side cladding layer 48)
  Next, a p-side cladding layer 48 made of a superlattice layer is formed. This layer acts as a carrier confinement layer like the n-side cladding layer 43, and acts as a layer for reducing the resistivity on the p-type layer side by adopting a superlattice structure. The film thickness of the p-side cladding layer 48 is not particularly limited, but it is desirable that the p-side cladding layer 48 be grown at a thickness of 100 angstroms to 2 μm, more preferably 500 angstroms to 1 μm. Although the superlattice layer is also provided on the n-side cladding layer side here, the resistance value of the p-layer tends to decrease when the superlattice layer is provided on the p-side layer side rather than the n-side cladding layer side. Therefore, it is preferable in reducing Vf.

  In the case of a nitride semiconductor element having a double hetero structure having an active layer 45 having a quantum well layer, Al having a film thickness of 0.1 μm or less having a band gap energy larger than that of the active layer 45 is in contact with the active layer 45. A cap layer 46 made of a nitride semiconductor is provided, and a p-side light guide layer 47 having a bad gap energy smaller than that of the cap layer 46 is provided at a position farther from the active layer than the cap layer 46. The p-side light guide It is very preferable to provide a p-side cladding layer 48 made of a superlattice layer containing a nitride semiconductor containing Al having a band gap larger than that of the p-side light guide layer 47 at a position farther from the active layer than the layer 47. In addition, since the electrons injected from the n layer, which has a large band gap energy of the p-side cap layer 46, are blocked by the cap layer 46, the electrons do not overflow the active layer, so that the leakage current of the element is small. Become.

(P-side contact layer 49)
  Finally, a p-side contact layer 49 made of p-type GaN doped with Mg is grown. The thickness of the p-side contact layer is adjusted to 500 angstroms or less, more preferably 400 angstroms or less, and 20 angstroms or more.

  As shown in FIG. 13, the uppermost p-type contact layer 20 and the p-type cladding layer 19 are etched to form a ridge shape having a stripe width of 4 μm. An electrode 51 is formed. Further, as shown in FIG. 13, an insulating film 50 made of SiO 2 is formed on the surface of the p-side cladding layer 48 and the contact layer 49 excluding the p electrode 51, and is electrically connected to the p electrode 51 through the insulating film 50. A p-pad electrode 52 is formed.

  Further, an n-electrode 53 made of Ti / Al is formed to a thickness of 0.5 μm on the entire surface of the second nitride semiconductor layer 4 that has not been exposed by polishing and removing the sapphire substrate 1 and the like of the wafer. A thin film made of Au / Sn is formed thereon for metallization with a heat sink.

  Then, scribing is performed from the n-electrode side 53, and the second nitride semiconductor layer 4 is cleaved at the M plane of the second nitride semiconductor layer 4 (11-00, the plane corresponding to the side of the hexagonal column in FIG. 9). A resonant surface is produced. Resonant surface has SiO ₂ And TiO ₂ A dielectric multilayer film is formed, and a bar is cut in a direction parallel to the p-electrode to form a laser chip. If this chip is placed face-up (with the substrate and the heat sink facing each other) on the heat sink and the p-pad electrode 52 is wire-bonded, continuous laser oscillation can be obtained at room temperature and a lifetime of 1000 hours or more is exhibited.

  The element structure of the nitride semiconductor element of the present invention is not particularly limited, such as the layer configuration, shape, electrode, etc. shown in FIG. 13, and any of these may be used in combination. It is preferable that an n-side nitride semiconductor having a superlattice structure is formed as the n-side nitride semiconductor of the element structure. Such a superlattice layer is preferable because the device performance can be improved. In addition, it is preferable to form the n electrode in the superlattice layer, and even if the superlattice layer is doped with an n-type impurity to reduce the contact resistance with the n electrode, the superlattice layer improves the crystallinity. This is preferable.
  Furthermore, as a preferable layer structure of the element constituting the nitride semiconductor element, for example, having an active layer having a quantum well structure containing In, an active layer sandwiched between cladding layers having different band gap energies, luminous efficiency, lifetime characteristics, etc. This is preferable in terms of improving the performance of the element.
  An element structure having such a layer structure is preferably formed on a nitride semiconductor with few crystal defects obtained by a lateral growth method, because the element performance is further improved.

  Next, in the laser device of the present invention, in order to obtain a nitride semiconductor substrate in which a nitride semiconductor layer is grown in the lateral direction, the following configurations (1) to (13) are preferable.
  (1) a first step of partially forming a first protective film on a nitride semiconductor grown on a different substrate made of a material different from that of the nitride semiconductor;
  After the first step, the first nitride semiconductor is grown on the nitride semiconductor, and includes a second step of growing the first nitride semiconductor onto the first protective film. Semiconductor growth method.
  (2) The nitride semiconductor according to (1), wherein the first protective film has a surface area larger than a surface area of a portion where the first protective film is not formed. Growth method.
  (3) A first protective film is partially formed on a dissimilar substrate made of a material different from that of a nitride semiconductor, with a surface area larger than a surface area of a portion where the first protective film is not formed. A first step;
  And a second step of growing a first nitride semiconductor on the heterogeneous substrate and growing on the first protective film after the first step. Growth method.
  (4) The first protective film has a stripe shape, and a width of a portion (window portion) where the adjacent stripe-shaped first protective film is not formed is 5 μm or less. The method for growing a nitride semiconductor according to any one of (1) to (3) above.
  (5) The ratio (Ws / Ww) between the width (Ww) of the window portion and the width (Ws) of the stripe-shaped first protective film is 1 to 20, wherein (1) to (4) The method for growing a nitride semiconductor according to any one of the above.

  (6) The method for growing a nitride semiconductor according to any one of (1) to (5), wherein the heterogeneous substrate has a main surface that is off-angled from a main surface of the substrate.
  (7) The method for growing a nitride semiconductor according to (6), wherein the heterogeneous substrate is off-angled stepwise.
  (8) The heterogeneous substrate is sapphire whose principal surface is (0001) plane = (C plane), and the first protective film is perpendicular to the (112-0) plane = (A plane) of the sapphire. The method for growing a nitride semiconductor according to any one of (1) to (7), wherein the nitride semiconductor has a stripe shape.
  (9) The heterogeneous substrate is sapphire whose principal surface is (112-0) plane = (A plane), and the first protective film is (11-02) plane = (R plane) of the sapphire. The nitride semiconductor growth method according to any one of (1) to (7), wherein the nitride semiconductor has a vertical stripe shape.
  (10) The heterogeneous substrate is a spinel having a (111) plane as a main surface, and the first protective film has a stripe shape perpendicular to the (110) plane of the spinel. The method for growing a nitride semiconductor according to any one of (1) to (7).

  (11) After the second step, a third step of partially forming a second protective film on the first nitride semiconductor;
  And a fourth step of growing a second nitride semiconductor on the first nitride semiconductor and growing on the second protective film after the third step. The method for growing a nitride semiconductor according to any one of (1) to (10).
  (12) The method for growing a nitride semiconductor according to (11), wherein the second protective film is formed on a crystal defect appearing on a surface of the first nitride semiconductor.
  (13) The method for growing a nitride semiconductor according to (11) or (12), wherein the second protective film has a stripe shape parallel to the first protective film.

  A first protective film is formed on a nitride semiconductor grown on a different substrate (the method of the first embodiment), and the first nitride semiconductor is grown, so that crystallinity with few crystal defects is good. A nitride semiconductor can be obtained. Furthermore, the first protective film grown on the nitride semiconductor is formed to have a surface area larger than the surface area of the portion (window part) where the first protective film is not formed. This is preferable because crystal defects are reduced and the first nitride semiconductor is easily grown.
  In another growth method, a first protective film is formed on a heterogeneous substrate (the method of the second embodiment), and the surface area of the first protective film is a window portion where the protective film is not formed. By growing the first nitride semiconductor on the first protective film having a surface area larger than the surface area, a nitride semiconductor with good crystallinity with few crystal defects can be obtained. In this growth method, the first protective film is formed directly on the different substrate, and the first nitride semiconductor layer is grown from the exposed different substrate surface (window portion where the protective film is not formed). Although many crystal defects are generated, the first nitride semiconductor layer with few crystal defects can be obtained by adjusting the surface area of the first protective film and the surface area of the window portion. When the surface area is further adjusted, the first nitride semiconductor starts to selectively grow on the window portion.

  Furthermore, the dissimilar substrate has a main surface that is off-angled (inclined) from the main surface of the dissimilar substrate, so that it is easy to obtain a nitride semiconductor with a small number of crystal defects, and the off-angle is continuously formed. It is preferable to form in a step shape rather than to reduce crystal defects.

  Furthermore, it is preferable that the number of crystal defects can be further reduced by specifying the plane orientation of the heterogeneous substrate and / or the shape of the first protective film.

  Furthermore, if the first protective film is formed in a stripe shape with a window width (distance between the protective film and the protective film) of 5 μm or less, crystal defects generated at the interface between the heterogeneous substrate and the nitride semiconductor However, dislocation is less likely to occur in the surface direction of the first nitride semiconductor, and crystal defects appearing on the surface of the first nitride semiconductor are preferably reduced. Furthermore, by adjusting the ratio Ws / Ww of the width (Ww) of the window portion and the width (Ws) of the protective film to be 1 to 20, a tendency to further reduce dislocations of crystal defects is observed. The crystal defects appearing on the surface of the nitride semiconductor are preferably reduced.

  Preferably, after the second step, a second protective film is partially formed on the first nitride semiconductor, and the second nitride semiconductor is formed thereon on the second protective film. It is preferable that the crystal defect dislocation is suppressed more satisfactorily and the crystal defects appearing on the surface of the second nitride semiconductor are further reduced. Furthermore, it is preferable that the second protective film is formed on the crystal defects appearing on the surface of the first nitride semiconductor so as to cover the crystal defects, so that dislocation of the crystal defects can be further prevented.

  Furthermore, it is preferable that the second protective film has a stripe shape parallel to the first protective film in order to obtain the effects of the present invention. This is because, for example, when the first protective film is provided in a direction perpendicular to the sapphire A surface, in a direction perpendicular to the sapphire R surface, and in a direction perpendicular to the spinel (110) surface. Means that the stripes of the first protective film and the stripes of the second protective film are provided in the same parallel direction.

For the first example the specific schematic diagram in the form of a lateral growth method used in the present invention will be described in more detail the growth method used in the present invention.
The growth method according to the first mode and the growth method according to the second mode of the lateral growth method used in the present invention are such that the surface on which the first protective film is formed in the first step is a nitride semiconductor on a different substrate. It is different whether the surface is a surface (first form) on which a layer (for example, a buffer layer, a GaN layer, etc.) is grown or a heterogeneous substrate surface (second form). In the case of forming a protective film, the difference is that the surface area of the first protective film and the surface area of the window portion are adjusted in advance so that the first nitride semiconductor layer is easy to grow. Others are almost the same. Also in the first embodiment, the first nitride semiconductor can be grown better by adjusting the surface area of the first protective film and the width of the window.
First, description will be given of a first embodiment of the lateral growth method used in the present invention, a description will be given of a second embodiment of the growth method in the following.

1 to 6 are schematic cross-sectional views showing the structure of a nitride semiconductor wafer obtained in each step of the first mode of the lateral growth method used in the present invention . In the figure, 1 is a heterogeneous substrate, 2 is a nitride semiconductor layer, 3 is a first nitride semiconductor layer, 4 is a second nitride semiconductor layer, 11 is a first protective film, and 12 is a second protective film. The membrane is shown.

In the first mode of the lateral growth method used in the present invention , in the first step, the nitride semiconductor layer 2 is grown on the buffer layer grown on the heterogeneous substrate 1 as shown in FIG. A first protective film 11 is partially formed on the nitride semiconductor layer 2.
The heterogeneous substrate 1 that can be used in the present invention may be any substrate made of a material different from a nitride semiconductor. For example, in addition to the sapphire C surface, the R surface and the A surface are the main surfaces. Insulating substrates such as sapphire, spinel (MgA12O4), SiC (including 6H, 4H, 3C), ZnS, ZnO, GaAs, Si, and oxide substrates lattice-matched with nitride semiconductors are conventionally known. A substrate material different from the nitride semiconductor can be used.
Furthermore, a substrate in which the main surface of the substrate material is off-angle, more preferably a substrate in which the main surface is off-angled in a step shape can be used. Thus, it is preferable that the main surface of the different substrate is off-angle because crystal defects are reduced.

  As the buffer layer 2 formed on the heterogeneous substrate 1 shown in FIG. 1, for example, AlN, GaN, AlGaN, InGaN or the like is grown at a temperature of 900 ° C. or less at a film thickness of several tens of angstroms to several hundreds of angstroms. It will be. This buffer layer is formed to mitigate the irregularity of the lattice constant between the heterogeneous substrate 1 and the nitride semiconductor layer 2, but may be omitted depending on the growth method of the nitride semiconductor, the type of the substrate, and the like. In addition, the buffer layer is preferable for alleviating the lattice constant irregularity between the heterogeneous substrate and the nitride semiconductor layer 2 and preventing the occurrence of crystal defects.

In the first embodiment of the lateral growth method used in the present invention, the nitride semiconductor layer 2 grown after the first step is doped with undoped (undoped impurity), GaN, or n-type impurity. GaN or GaN doped with Si can be used.
The nitride semiconductor 2 is grown on a heterogeneous substrate at a high temperature, specifically 900 ° C. to 1100 ° C., preferably 1050 ° C., and the film thickness is not particularly limited, but for example 1 to 20 μm, preferably 2 to 10 μm. is there. When the thickness of the nitride semiconductor layer 2 is in the above range, the total thickness of the nitride semiconductor layer 2 and the first nitride semiconductor can be suppressed, and the warpage of the wafer (warpage in a state having a different substrate) can be prevented. .

As the material of the first protective film 11, a material having a property that a nitride semiconductor does not grow or hardly grow on the surface of the protective film is preferably selected. For example, silicon oxide (SiO _x ), silicon nitride (Si _x N) _Y ), titanium oxide (TiO _x ), zirconium oxide (ZrO _x ) and other oxides, nitrides, multilayer films, and metals having a melting point of 1200 ° C. or higher can be used. These protective film materials can withstand the nitride semiconductor growth temperature of 600 ° C. to 1100 ° C., and the nitride semiconductor does not grow or hardly grow on the surface thereof. In order to form the protective film material on the surface of the nitride semiconductor, for example, vapor deposition techniques such as vapor deposition, sputtering, and CVD can be used. Further, in order to form partially (selectively), a photomask having a predetermined shape is produced by using a photolithography technique, and the material is vapor-deposited through the photomask. The first protective film 11 having a predetermined shape can be formed. The shape of the first protective film 11 is not particularly limited. For example, the first protective film 11 can be formed in the shape of dots, stripes, and a grid surface. . Further, it is preferable that the surface area of the first protective film 11 is larger than the surface area of the window portion because the first nitride semiconductor 3 with few lattice defects is easily obtained.

It is preferable that the first protective film 11 is formed so that the surface area of the first protective film is larger than the surface area of a portion (window) where the first protective film is not formed. When the surface area of the first protective film is increased in this way, dislocations of crystal defects generated at the interface between the heterogeneous substrate and the nitride semiconductor are suppressed by the first protective film, and further, the crystal defects dislocated from the window portion are dislocated in the middle. Is preferable because it is easy to interrupt. The adjustment of the surface area of the first protective film and the surface area of the window portion varies depending on the shape of the protective film. For example, when the protective film has a stripe shape, the width of the stripe of the protective film and the width of the window portion are adjusted. Can be done.
The size of the first protective film 11 is not particularly limited. For example, when the first protective film 11 is formed in a stripe shape, the preferred stripe width is preferably 0.5 to 100 μm, more preferably about 1 μm to 50 μm. The pitch is desirably narrower than the stripe width. That is, when the surface area of the protective film is made larger than that of the window, a nitride semiconductor layer with fewer crystal defects can be obtained.

Furthermore, as a preferable form for adjusting the surface area of the protective film and the window part, the first protective film 11 is formed in a stripe shape, and the width of the window part is preferably adjusted to 5 μm or less, and more preferably the width of the window part ( (Ww) and the width (Ws) of the stripe-shaped first protective film, Ws / Ww is adjusted so as to be 1 to 20. In this way, when the first nitride semiconductor is grown by adjusting the width of the first protective film 11 and the window portion and Ws / Ww, a nitride semiconductor having very good crystallinity with very few crystal defects can be obtained. Can do.
The preferable value of the width of the window is 3 μm or less, more preferably 1 μm or less, and the lower limit is 0.1 μm or more. By adjusting the window width in this way, a nitride semiconductor layer with fewer crystal defects can be obtained.
The width of the stripe-shaped protective film is within the above range, but is particularly 2 to 30 μm, preferably 5 to 20 μm, more preferably 5 to 15 μm when the width of the window portion is 5 μm or less. . Within this range, a nitride semiconductor layer with few crystal defects is obtained, which is preferable. The thickness of the stripe-shaped protective film is not particularly limited, but is, for example, 0.01 to 5 μm, preferably 0.1 to 3 μm, and more preferably 0.1 to 2 μm. This range is preferable for obtaining the effect. Moreover, ratio Ws / Ww of the width | variety (Ww) of a window part, and the width | variety (Ws) of a protective film is 1-20, Preferably it is 1-10. If it is 1 or less, crystal defects are likely to occur on the window portion and the protective film, and if it is 20 or more, the first nitride semiconductor grown on the protective film does not completely stick to it and a cavity is formed on the protective film. It becomes easy.

Next, in the second step, as shown in FIG. 2, the first nitride semiconductor 3 is grown on the nitride semiconductor layer 2 on which the first protective film 11 is formed.
The first nitride semiconductor 3 grown on the nitride semiconductor layer 2 is not particularly limited, but preferably includes undoped (undoped state, undope) GaN or GaN doped with n-type impurities. It is done.

  As shown in FIG. 2, when the first nitride semiconductor layer 3 is grown on the nitride semiconductor layer 2 on which the first protective film 11 is formed, the first protective film 11 is formed on the first protective film 11. The nitride semiconductor layer 3 does not grow, and the first nitride semiconductor layer 3 is selectively grown on the exposed nitride semiconductor layer 2. As the growth continues further, the first nitride semiconductor layer 3 grows laterally on the first protective film 11 and is connected to the adjacent first nitride semiconductor layers 3 as shown in FIG. In addition, the state is as if the first nitride semiconductor layer 3 has grown on the first protective film 11.

  The crystal defects (threading dislocations) appearing on the surface of the first nitride semiconductor layer 3 grown in this way are very small compared to the conventional one. However, the number of crystal defects in the upper portion of the window portion and the upper portion of the first protective film 11 in the initial growth of the first nitride semiconductor layer 3 is significantly different. In other words, the first nitride semiconductor layer 3 in the initial stage of growth grown in a portion (window) where the first protective film 11 is not formed on the different substrate 1 includes the different substrate 1 and the nitride semiconductor layer. Although there is a tendency that crystal defects are generated from the interface with the interface (for example, the buffer layer in the case of FIG. 2) and dislocation tends to occur in the vertical direction, the first nitriding in the initial stage of growth grown on the first protective film 11 The physical semiconductor layer 3 has almost no crystal defects dislocation in the vertical direction.

  For example, as shown in a schematic diagram of threading dislocations due to crystal defects of the nitride semiconductor crystal of the wafer shown in FIG. 3, it is indicated by a plurality of thin lines from the heterogeneous substrate 1 toward the surface of the first nitride semiconductor layer 3. Such crystal defects are considered to be generated and dislocations. The crystal defects in the window portion shown in FIG. 3 occur very often at the interface between the heterogeneous substrate 1 and the nitride semiconductor due to the mismatch of the lattice constant between the heterogeneous substrate 1 and the nitride semiconductor. Most of the crystal defects in the window part are dislocated from the interface between the heterogeneous substrate 1 and the nitride semiconductor toward the surface during the growth of the first nitride semiconductor layer 3. However, as shown in FIG. 3, the crystal defects generated from the window portion continue to dislocation at the initial growth stage of the first nitride semiconductor layer 3, but the first nitride semiconductor layer 3 grows. In the course of continuing, the number of crystal defects dislocations in the surface direction tends to decrease drastically, and the number of crystal defects dislocations to the surface of the first nitride semiconductor layer 3 becomes very small. Also, the first nitride semiconductor layer 3 formed on the first protective film 11 is not grown from the heterogeneous substrate 1 but is formed by connecting the adjacent first nitride semiconductor layers 3 during the growth. Therefore, compared with the portion of the first nitride semiconductor layer 3 grown on the nitride semiconductor layer 2 grown from the substrate, there are very few crystal defects from the beginning of the growth. As a result, the surface of the first nitride semiconductor layer 3 (the upper portion of the protective film and the upper portion of the window) after the growth has very few dislocation crystal defects, or the upper surface of the protective film is observed by a transmission electron microscope. Almost disappear. By using the first nitride semiconductor layer 3 having very few crystal defects as a nitride semiconductor growth substrate having an element structure, a nitride semiconductor element having better crystallinity than the conventional one can be realized. Further, since the generation of crystal defects and the tendency of dislocations due to the growth of GaN of the present invention as described above are observed, it is preferable to make the surface area of the window portion smaller than the surface area of the protective film.

  In addition, the first nitride semiconductor layer grown on the upper part of the window part having a large number of crystal defects in the initial stage of growth, although the crystal defect is reduced in both the window part on the surface of the first nitride semiconductor layer 3 and the upper part of the protective film. The surface of 3 tends to have slightly more crystal defects than those grown on top of the protective film. This is probably because dislocations of many crystal defects stopped during the growth of the first nitride semiconductor layer 3, but crystal defects that continue to dislocation tend to dislocation almost directly above the window portion. It is thought that there is.

When the number of crystal defects due to the difference in dislocations of such crystal defects is observed by a cross-sectional TEM, dislocations are observed only in the upper part of the window and almost no defects are observed in the upper part of the protective film. In a preferred form, the crystal defect density of the window top, approximately 10 ⁶ / cm ² or less, in a preferred condition is 10 ⁵ / cm ² or less, the protective film upper, approximately 10 ⁵ cells / cm ² or less The preferable condition is 10 ⁴ pieces / cm ² or less. Crystal defects can be measured, for example, by measuring the number of etch pits that appear on the etched surface when a nitride semiconductor is dry etched.

Although the film thickness of the first nitride semiconductor layer 3 varies depending on the film thickness and size of the first protective film formed earlier, the first nitride semiconductor layer is formed so as to cover the surface of the protective film. In order to grow it, it is desirable to grow it at a film thickness of at least 10 times, more preferably 50 times or more the thickness of the protective film. Further, the film thickness of the first nitride semiconductor tends to decrease the dislocations of crystal defects because the dislocations of crystal defects tend to decrease drastically during the growth of the first nitride semiconductor as described above. It is preferable to adjust to the above.
Furthermore, the first nitride semiconductor serves as a substrate for growing a nitride semiconductor having an element structure thereon. In order to form the element structure, the dissimilar substrate, the protective film, etc. are removed in advance. There are cases where the process is performed after only one nitride semiconductor is used, and cases where the process is performed while leaving a heterogeneous substrate or the like. Further, the removal of the heterogeneous substrate or the like may be performed after the element structure is formed. When forming an element structure on the first nitride semiconductor, the thickness of the first nitride semiconductor affects the ease of forming the element structure depending on the presence or absence of a heterogeneous substrate. The thickness of the physical semiconductor covers the first protective film, reduces dislocations of crystal defects, and further differs in manufacturing processes such as forming an element structure with or without removing a heterogeneous substrate. It is desirable to adjust in consideration of the above.

When removing a heterogeneous substrate, a protective film, etc., the first nitride semiconductor is grown to a thickness of more than 50 μm and about 1 mm or less, for example, preferably 70 to 500 μm, more preferably 100 to 300 μm, Preferably it is 100-250 micrometers. Within this range, it is preferable in terms of the growth of the nitride semiconductor to be an element structure, and even if the underlying layer and the protective film are removed by polishing, the first nitride semiconductor is not easily cracked and is easy to handle. . Further, it is preferable to remove the heterogeneous substrate because the nitride semiconductor substrate is not warped when the element structure is formed, and an element structure with good crystallinity is easily obtained.
When the heterogeneous substrate or the protective film is left, the thickness of the first nitride semiconductor is 1 to 50 μm, preferably 2 to 40 μm, more preferably 5 to 30 μm, and most preferably 10 to 20 μm. Within this range, the wafer can be prevented from warping due to the difference in thermal expansion coefficient between the dissimilar substrate and the nitride semiconductor, and the nitride semiconductor as the element structure can be satisfactorily grown on the first nitride semiconductor as the element substrate. be able to.

When a nitride semiconductor is grown on a heterogeneous substrate, it differs depending on the type of the heterogeneous substrate, but due to the difference in thermal expansion coefficient with the heterogeneous substrate, the entire wafer tends to warp after growth, and the warpage is caused by thickening the nitride semiconductor It tends to grow as it grows. When a nitride semiconductor layer of a wafer having a different kind of substrate is subjected to various processes to obtain an operating structure, it is difficult to process the nitride semiconductor when the wafer is warped. Therefore, when a nitride semiconductor device having a different substrate is used, the first nitride semiconductor layer has a film thickness that is easy to process with the different substrate attached even if the wafer is warped, that is, a film thickness of 50 μm or less. desirable. 1 μm indicates a limit value at which a nitride semiconductor can be grown on the protective film. In the case of a nitride semiconductor element formed by leaving the heterogeneous substrate as it is and forming an element structure, the first nitride semiconductor layer grown on the protective film is 1 × 10 ¹⁶ / cm ^{3 to} 5 × 10 ¹⁹ / cm ³ . Most preferred is GaN doped with n-type impurities.

In the case of a nitride semiconductor element obtained by removing a different substrate, in order to prevent the first nitride semiconductor from being cracked or chipped when the different substrate is removed, and to improve handling in the device process. The film thickness of the first nitride semiconductor layer is preferably greater than 50 μm. If the first nitride semiconductor has such a film thickness, the warpage of the wafer tends to increase as described above. Therefore, the first nitride semiconductor layer is removed after removing the dissimilar substrate. It is preferable to form an element structure on top. When the heterogeneous substrate is removed as described above, the warp of the first nitride semiconductor that becomes the substrate for forming the element structure is eliminated, and the nitride semiconductor that becomes the element structure is easily formed easily. Moreover, the upper limit of the film thickness of the first nitride semiconductor layer is not particularly limited, but if it is too thick, a film thickness of 1 mm or less is preferable from the viewpoint of taking too much growth time.
In a nitride semiconductor device formed by removing a heterogeneous substrate or the like, the heterogeneous substrate may be removed before or after the device structure is formed on the first nitride semiconductor, and preferably, the heterogeneous substrate is removed. An element structure is formed using the first nitride semiconductor layer as a nitride semiconductor substrate (GaN substrate). As a method for removing the heterogeneous substrate, polishing, etching, or the like is used. Further, when the second protective film is formed, the element structure is formed on the second nitride semiconductor by removing only the second protective film and removing only the second nitride semiconductor when removing the heterogeneous substrate. Alternatively, the element structure may be formed using the first nitride semiconductor and the second nitride semiconductor as a GaN substrate without removing the second protective film. Further, when growing a nitride semiconductor to be an element structure on a nitride semiconductor substrate from which a heterogeneous substrate or the like has been removed, the element structure is grown on a surface opposite to the surface exposed by removing the heterogeneous substrate or the like. It is preferable that an element with good crystallinity can be obtained.

  In the present invention, the thickness of the first nitride semiconductor layer is the same as the thickness of the first nitride semiconductor layer 3 alone shown in FIGS. 6 and 8, or the first nitride semiconductor layer 3 and the first nitride semiconductor layer 3. 2 indicates the total film thickness of the nitride semiconductor layer 4. That is, it refers to the film thickness of the nitride semiconductor layer grown on the first protective film 11 grown first on the substrate.

Next, as a preferable step, the nitride semiconductor substrate having an element structure can be obtained with good crystallinity by performing the third step and the fourth step after the second step.
First, in the third step of the present invention, as shown in FIG. 4, a crystal defect is likely to appear on the surface of the first nitride semiconductor layer 3, for example, on the top of the window, and on the surface. A new protective film (second protective film 12) is provided so as to cover the crystal defects. In the present invention, the formation position of the second protective film 12 is not particularly limited, and appears partially on the surface of the first nitride semiconductor layer 3, preferably on the surface of the first nitride semiconductor layer 3. It is formed so as to cover the crystal defects, and more preferably the upper part of the window part where the crystal defects exist at the initial growth stage of the first nitride semiconductor layer 3.
Providing the second protective film 12 in this way can prevent further dislocations of crystal defects dislocated to the surface of the first nitride semiconductor layer 3, and further dislocations at the upper part of the window after forming the element structure. There is a possibility that the interrupted crystal defects may be re-dislocated to the active layer or the like during the operation of the laser element or the like.

  In FIG. 4, the first nitride semiconductor layer 3 grown in FIG. 3 is polished to have a flat surface in order to reduce unevenness, but the first nitride semiconductor layer is not polished and is left as it is. The second protective film 12 may be formed on the surface 3.

  Preferably, the surface area of the second protective film 12 is not particularly limited, but is desirably formed so as to cover crystal defects appearing on the surface of the first nitride semiconductor with the second protective film. For example, when the first protective film 11 has a stripe shape, a stripe-shaped second protective film 12 having a width larger than the width of the window portion is formed on the upper portion of the window portion. As the material of the second protective film 12, the same material as the first protective film can be used.

  Next, in the fourth step of the present invention, the second nitride semiconductor layer 4 is grown on the first nitride semiconductor layer 3 on which the second protective film 12 is formed. As shown in FIG. 5, at first, the second nitride semiconductor layer 4 does not grow on the second protective film 12 as in the case of the first nitride semiconductor layer 3, but the first nitride semiconductor layer 3 does not grow. It grows selectively only on the physical semiconductor layer 3. The second nitride semiconductor layer 4 grown on the first nitride semiconductor layer 3 is the same nitride semiconductor and is grown on the first nitride semiconductor layer 3 with few crystal defects. Therefore, it is difficult for crystal defects due to lattice constant mismatch. Furthermore, since the first nitride semiconductor layer 3 that is the base layer of the second nitride semiconductor layer 4 has few crystal defects, the number of crystal defects that dislocations in the initial growth stage of the second nitride semiconductor layer 4 is also reduced. .

As the growth continues further, as shown in FIG. 6, adjacent second nitride semiconductor layers 4 are connected to each other at the upper part of the second protective film 12 so as to cover the second protective film 12. . Since the second nitride semiconductor layer 4 grown in this way grows with the first nitride semiconductor layer 3 having few crystal defects as an underlayer, it becomes a nitride semiconductor having very few crystal defects. By using the second nitride semiconductor 4 as a nitride semiconductor growth substrate having an element structure, a nitride semiconductor element having extremely excellent crystallinity can be realized.
Since the second protective film 12 is formed so as to cover the crystal defects appearing on the surface of the first nitride semiconductor 3, dislocation of the crystal defects can be suppressed. Even if a slight crystal defect continues dislocation at the beginning of the growth of the second nitride semiconductor 4, as in the case of the growth of the first nitride semiconductor 3, As the growth continues, dislocations of crystal defects tend to stop, and crystal defects appearing on the surface of the second nitride semiconductor 4 are reduced.
Since the second nitride semiconductor layer 4 obtained in this way has fewer crystal defects than the first nitride semiconductor layer 3, if an element structure is formed on the second nitride semiconductor layer 4, the crystallinity of the second nitride semiconductor layer 4 is reduced. Good elements can be obtained more easily.

  Next, a second embodiment of the lateral growth method used in the present invention will be described. 7 and 8 are schematic cross-sectional views showing the structure of a wafer in a part of the process according to the second embodiment of the lateral growth method used in the present invention.

As shown in FIG. 7, the second mode of the lateral growth method used in the present invention is to form a first protective film 11 and a first protective film 11 on a heterogeneous substrate 1 in the first step. Forming partly with a surface area larger than the surface area of the part (window part) that has not been formed, and subsequently forming a buffer layer on the dissimilar substrate 1 where the window part is exposed in the second step; From this, the first nitride semiconductor layer 3 is grown as shown in FIG. In the growth method of the second mode, the buffer layer may be omitted depending on circumstances.
In the growth method of the second mode, the first nitride semiconductor layer 3 selectively starts growing from the window and covers the first protective film 11 as in the growth method of the first mode. As shown in FIG. 8, the first nitride semiconductor layer 3 is formed. Further, as shown in FIG. 8, crystal defects generated at the interface between the heterogeneous substrate 1 and the nitride semiconductor (the buffer layer in the growth method of the second embodiment) are the same as those of the growth method of the first embodiment. In addition, although dislocations are generated from the window portion, the dislocations drastically decrease as the growth of the first nitride semiconductor layer 3 continues, and further, no crystal defects are generated on the upper portion of the first protective film 11 and the grown first The number of crystal defects on the surface of the nitride semiconductor layer 3 is reduced.
Further, in the growth method of the second embodiment, as in the case of the growth method of the first embodiment, if the adjustment of the width of the protective film and the width of the window portion, the plane orientation of the different substrate, etc. are specified, the better first This nitride semiconductor is obtained.
Examples of the heterogeneous substrate, the protective film, the buffer layer, and the first nitride semiconductor layer 3 that are used in the growth method of the second embodiment are the same as those used in the growth method of the first embodiment.

  In the second embodiment of the lateral growth method used in the present invention, the buffer layer and the first nitride semiconductor layer 3 grown on the buffer layer are also referred to as the first nitride semiconductor layer. That is, in the second step of claim 3, it is continuously grown on the surface of the heterogeneous substrate 1 or continuously with the nitride semiconductor grown on the heterogeneous substrate, and is formed on the first protective film 11. All of the nitride semiconductors grown up to 1 are defined as the first nitride semiconductor layer 3.

Furthermore, in the second mode of the lateral growth method used in the present invention, preferably, as in the case of the growth method of the first mode, as the third step, the surface of the first nitride semiconductor layer 3 is used. The second protection is provided on the portion where the crystal defect is likely to appear so as to cover the crystal defect appearing in FIG. 5 and on the upper part of the window where the crystal defect is dislocated at the initial stage of the growth of the first nitride semiconductor layer 3. The film 12 is formed, and then, as a fourth step, the second nitride semiconductor layer 4 is grown on the first nitride semiconductor layer 3 on which the second protective film 12 is formed.
When the second protective film 12 is formed in this way, as in the growth method of the first embodiment, further dislocation suppression of crystal defects appearing on the surface of the first nitride semiconductor layer 3 and discontinuation of dislocations are performed. Thus, the second nitride semiconductor 4 can be obtained as a nitride semiconductor substrate capable of preventing re-dislocation of the crystal defects and forming a highly reliable element with good crystallinity. For example, as shown in FIG. 8, when the second protective film is formed on the upper part of the window where the crystal defects are dislocated at the initial growth stage of the first nitride semiconductor layer 3, the relocation of the crystal defects where the dislocation is interrupted. Even if this occurs, dislocation to the second nitride semiconductor layer 4 can be prevented by the second protective film 12.

  In the growth methods of the first and second embodiments of the present invention, the third step and the fourth step can be repeated. That is, it is preferable to form a new protective film on the portion of the nitride semiconductor layer where crystal defects are exposed, and to grow a new nitride semiconductor on the protective film.

Next, a preferable embodiment of the different type substrate used in the present invention will be described.
FIG. 9 is a unit cell diagram showing a crystal structure of a nitride semiconductor. Although a nitride semiconductor has a rhombohedral structure precisely, it can be approximated in a hexagonal system in this way. In the method of the present invention, sapphire having a C-plane = (0001) plane as a main surface is preferably used, and the first protective film has a stripe shape perpendicular to the sapphire A-plane = (112-0) plane. For example, FIG. 10 is a plan view of a sapphire substrate on the main surface side. In this figure, the sapphire C surface is the main surface, and the orientation flat (orientation flat) surface is the A surface. As shown in this figure, stripes of the first protective film are formed in a direction perpendicular to the A plane and parallel to each other. As shown in FIG. 10, when a nitride semiconductor is selectively grown on the sapphire C surface, the nitride semiconductor tends to grow in a direction parallel to the A plane in the plane and hardly grow in a vertical direction. It is in. Therefore, if a stripe is provided in a direction perpendicular to the A plane, the nitride semiconductor between the stripes is connected and easily grown, and crystal growth as shown in FIGS. Become.

  Similarly, in the case of using a sapphire substrate having the A plane as the main surface, for example, when the orientation flat surface is the R plane = (11-02) plane, stripes parallel to each other are formed in the direction perpendicular to the R plane. By doing so, since the nitride semiconductor tends to grow in the stripe width direction, a nitride semiconductor layer with few crystal defects can be grown.

Also, spinel (MgAl ₂ O ₄ ) has anisotropy in the growth of the nitride semiconductor. If the growth surface of the nitride semiconductor is the (111) plane and the orientation flat surface is the (110) plane, the nitride semiconductor is nitrided. A physical semiconductor tends to grow in a direction parallel to the (110) plane. Accordingly, when a stripe is formed in a direction perpendicular to the (110) plane, the nitride semiconductor layer and the adjacent nitride semiconductor are connected to each other at the upper portion of the protective film, so that a crystal with few crystal defects can be grown. The above description is about the first protective film. Similarly, when the second protective film is formed, a stripe parallel to the first protective film may be formed on the surface of the second nitride semiconductor layer. desirable. Spinel is not particularly shown because it is tetragonal.

  Next, a heterogeneous substrate having a main surface off-angled from the main surface of the heterogeneous substrate will be described with reference to FIG. FIG. 15 is a schematic view showing an enlarged cross section of the sapphire substrate. The substrate off-angled in a step shape shown in FIG. 15 has a substantially horizontal terrace portion A and a stepped portion B. The surface unevenness of the terrace portion A is adjusted to about 0.5 angstroms on average and about 2 angstroms at maximum, and is formed almost regularly. On the other hand, the height of the step portion is adjusted to about 15 angstroms. Although the off angle θ is exaggerated, it is inclined only 0.13 ° with respect to the horizontal plane of the growth surface. Such a step-like portion having an off angle is desirably formed continuously over the entire substrate, but may be formed partially in particular. As shown in FIG. 15, the off angle θ refers to an angle between a straight line connecting the bottoms of a plurality of steps and a horizontal plane of the uppermost step. The step difference is 30 angstroms or less, more preferably 25 angstroms or less, and most preferably 20 angstroms or less. The lower limit is preferably 2 angstroms or more. In particular, when a sapphire C-plane is used for the substrate, the off-angle θ from the C-plane is adjusted to within 1 degree, preferably 0.8 degrees or less, and more preferably 0.6 degrees or less. In this embodiment, a step-like off-substrate is used, but a normal off-substrate may be used instead of a step-like substrate. By using an appropriately off-angled dissimilar substrate, the interatomic distance between the nitride semiconductor and the dissimilar substrate approaches, and a GaN substrate with fewer crystal defects can be obtained.

  In the present invention, a method for growing a nitride semiconductor is not particularly limited, but MOVPE (metal organic vapor phase epitaxy), HVPE (halide vapor phase epitaxy), MBE (molecular beam vapor phase epitaxy), etc. All methods known to grow physical semiconductors can be applied. A preferred growth method is the MOVPE method, which allows crystals to grow neatly. However, since the MOVPE method takes time, when the film thickness is thick, it is preferable to perform the method using a short time.

[Example 1 regarding growth of nitride semiconductor substrate ] (Growth Method of First Mode)
The present embodiment shows MOVPE (metal organic vapor phase epitaxy), but is not limited to the M OVPE method, for example, HVPE (halide vapor phase epitaxy), MBE (molecular beam vapor phase epitaxy), etc. All known methods for growing nitride semiconductors can be applied.

1 to 6 are schematic cross-sectional views of a wafer in each step showing the first embodiment. A sapphire substrate 1 with the C plane as the main plane and the orientation flat plane as the A plane is set in the reaction vessel, the temperature is set to 510 ° C., hydrogen is used as the carrier gas, and ammonia and TMG (trimethylgallium) are used as the source gas. Then, a buffer layer made of GaN is grown on the sapphire substrate 1 to a thickness of 200 Å. After the buffer layer growth, only TMG is stopped, the temperature is raised to 1050 ° C., and when it reaches 1050 ° C., TMG, ammonia, and silane gas are used as source gases, and the undoped GaN layer 2 is grown to a thickness of 5 μm. A stripe-shaped photomask is formed on the GaN layer 2 of the wafer in which the buffer layer and the GaN layer 2 are laminated, and a first protective film 11 made of SiO _{2 having a} stripe width of 10 μm and a window portion of 8 μm is formed by a CVD apparatus. Is formed with a film thickness of 0.1 μm (FIG. 1). The stripe direction of the first protective film 11 is a direction perpendicular to the sapphire A plane.

  After the formation of the first protective film 11, the wafer is transferred to a reaction vessel, and a first nitride semiconductor layer made of undoped GaN is grown to a thickness of 100 μm at 1050 ° C. using TMG and ammonia as source gases ( 2 and 3).

  Next, the wafer is taken out of the reaction vessel, and the surface of the first nitride semiconductor layer 3 is lapped to make a mirror surface, which is similar to the formation of the first protective film 11. A second protective film 12 made of Si3N4 with a stripe width of 12 .mu.m and an interval of 6 .mu.m is formed on the surface so as to cover the crystal defects with a thickness of 0.1 .mu.m (FIG. 4).

  After the second protective film 12 is formed, the wafer is returned to the reaction vessel again, and the second nitride semiconductor layer 4 made of undoped GaN is grown to a thickness of 150 μm using TMG and ammonia as source gases. After the second nitride semiconductor layer 4 was grown, the wafer was taken out of the reaction vessel and the surface was mirror-polished.

(Comparative example)
On the other hand, for comparison, a GaN buffer layer of 200 angstroms is directly grown on a sapphire substrate having a C surface as a main surface and an A surface as an orientation flat surface, without forming the first protective film 11, and a Si GaN buffer layer is grown thereon. GaN doped with 1 × 10 ¹⁸ / cm ³ is grown to 100 μm.

When the number of lattice defects per unit area of the second nitride semiconductor layer 4 obtained in Example 1 and the GaN layer obtained in the comparative example is observed and compared with a cross-sectional TEM, the nitride semiconductor of the present invention is compared. Was reduced to 1/200 or less as compared with the comparative example.
Further, when the surface of the first nitride semiconductor layer 3 was mirror-polished and the number of crystal defects was observed without growing the second protective film 12 and the second nitride semiconductor layer 4, the first The number of crystal defects in the nitride semiconductor layer 3 was reduced to 1/100 or less with respect to the number of crystal defects in the GaN layer of the comparative example.

[Example 2 regarding growth of nitride semiconductor substrate ] (Growth Method of Second Embodiment)
A striped photomask is formed on a sapphire substrate 1 having a 2 inch φ, C-plane as the main surface and an orientation flat surface as the A-plane, and is made of SiO 2 having a stripe width of 10 μm and a stripe interval (window portion) of 6 μm by a CVD apparatus. The first protective film 11 is formed with a thickness of 0.1 μm (FIG. 7). The stripe direction is formed in a direction perpendicular to the orientation flat surface as shown in FIG.

  After forming the protective film, the substrate is set in the reaction vessel, the temperature is set to 510 ° C., hydrogen is used as the carrier gas, ammonia and TMG (trimethylgallium) are used as the source gas, and the first protective film 11 is formed. A buffer layer made of GaN is grown on 1 to a thickness of about 200 angstroms. (Fig. 7)

After growing the buffer layer, only TMG is stopped and the temperature is raised to 1050 ° C. When the temperature reaches 1050 ° C., the first nitride semiconductor layer 3 made of GaN doped with Si at 1 × 10 ¹⁸ / cm ³ is grown to a thickness of 100 μm using TMG, ammonia, and silane gas as source gases (FIG. 8). reference).

  After the growth of the first nitride semiconductor layer 3, the wafer is taken out from the reaction vessel, and the surface of the first nitride semiconductor layer 3 is lapped into a mirror surface to obtain a nitride semiconductor substrate made of Si-doped GaN.

  The number of lattice defects per unit area of the GaN layer obtained in Example 2 and the GaN layer obtained in the comparative example shown in Example 1 was observed and compared with a cross-sectional TEM. The layer was reduced to 1/10 or less as compared with the comparative example.

[Example 3 regarding growth of nitride semiconductor substrate ]
A striped mask is formed on the surface of the first nitride semiconductor layer 3 obtained in Example 2, and a second protective film 12 made of Si3N4 having a stripe width of 10 .mu.m and a window portion of 6 .mu.m is formed by a CVD apparatus. It is formed with a film thickness of 1 μm (FIG. 8). As shown in FIG. 8, the mask of the second protective film 12 is aligned so that the 10 μm stripe of the second protective film 12 is on the crystal defect so as to cover the crystal defect. At the same time, stripes parallel to the first protective film 11 are formed.
After the second protective film 12 is formed, the wafer is returned to the reaction vessel again, and the second nitride semiconductor layer 4 made of GaN doped with Si at 1 × 10 ¹⁸ / cm ³ using TMG, ammonia, and silane gas as source gases. Is grown to a film thickness of 150 μm (FIG. 8). After the growth of the second nitride semiconductor layer 4, the wafer is taken out from the reaction vessel, and the surface is mirror-polished in the same manner as in Example 2 to determine the number of lattice defects per unit area with the GaN layer of the comparative example. When compared, the product of the present invention was reduced to 1/100 or less.

[Example 4 regarding growth of nitride semiconductor substrate ]
In Example 2, sapphire having an A plane as a main surface and an orientation flat surface as an R plane is used for the substrate 1. On the sapphire substrate 1, the same first protective film 11 as in the second embodiment is formed. The shape of the first protective film 11 is a stripe perpendicular to the R plane. After that, when the first nitride semiconductor layer 3 made of Si-doped GaN was grown to a thickness of 100 μm in the same manner as in Example 2, a nitride semiconductor layer having almost the same crystal defects as in Example 2 was grown. did it.

[Example 5 regarding growth of nitride semiconductor substrate ]
In Example 5, the first nitride semiconductor layer 3 is grown by the HVPE method. First, a substrate 1 made of a 1 inch φ spinel having a (111) plane as a main surface and an orientation flat surface as a (110) plane is prepared. A photomask is formed on the surface of the spinel substrate 1 in the same manner as in the second embodiment, and the first protective film 11 made of SiO2 is formed in a stripe shape perpendicular to the orientation flat surface. The stripe width is 12 μm and the stripe interval is 6 μm.

  In the HVPE apparatus, a quartz boat containing Ga metal is installed inside a reaction vessel tube made of quartz. Further, the above-described substrate 1 tilted obliquely is installed at a position away from the quartz boat. A halogen gas supply pipe is provided at a position close to the Ga metal in the reaction vessel, and an N source supply pipe is provided at a position close to the substrate separately from the halogen gas supply pipe.

Nitrogen carrier gas and mainly HCl gas are introduced from the halogen gas pipe. At this time, the Ga metal boat is heated to 900 ° C., and the spinel substrate side is heated to 1050 ° C. Then, HCl gas and Ga are reacted to generate GaCl3, and from the N source supply pipe close to the spinel substrate side, ammonia gas is supplied mainly with nitrogen carrier gas, and further, silane gas is supplied together with halogen gas, Growth is carried out for 3 hours at a growth rate of 50 μm / hr to grow GaN doped with 1 × 10 ¹⁸ / cm ³ of Si having a thickness of 150 μm.

  After the growth, the wafer was taken out of the reaction vessel, the GaN layer was lapped to remove the surface irregularities, and the lattice defects were measured. As a result, a nitride semiconductor layer substantially equivalent to that of Example 2 was obtained.

[Example 6 regarding growth of nitride semiconductor substrate ]
In Example 2, the stripe width 10μm on a sapphire substrate 1, a stripe interval (window) in addition to forming a protective film 11 made of SiO ₂ of 5μm in thickness of 1μm in the same manner, 1 Si × 10 ¹⁸ / A first nitride semiconductor layer 3 made of GaN doped with cm ³ is grown to a thickness of 100 μm. After the growth of the first nitride semiconductor layer 3, the wafer is taken out from the reaction vessel, and the surface of the first nitride semiconductor layer 3 is lapped into a mirror surface to obtain a nitride semiconductor substrate made of Si-doped GaN.

On the wafer of the GaN layer obtained in Example 6 and the GaN layer obtained in the comparative example shown in Example 1 above, a range of 10 × 15 μm is arbitrarily selected at nine locations, and the number of crystal defects per unit area Was observed with a cross-sectional TEM, and the number of crystal defects was measured. The crystal defect is measured by first etching the GaN substrate by about 1 μm by dry etching, and then observing the cross-sectional TEM to count the crystal defects.
As a result, in the present invention, the number of crystal defects is about 1.3 × 10 ⁶ pieces / cm ² , the comparative example is about 2.4 × 10 ⁷ pieces / cm ² , Compared to the example, it was reduced to 1/10 or less. In addition, the number of crystal defects was further reduced as compared with Example 2.

[Example 7 regarding growth of nitride semiconductor substrate ]
In the first embodiment, the first nitride semiconductor layer 3 is formed to a thickness of 100 μm in the same manner except that a protective film 11 made of SiO 2 having a stripe width of 10 μm and a window portion of 3 μm is formed on the undoped GaN layer 2. Grow with thickness. After the growth of the first nitride semiconductor layer 3, the wafer is taken out from the reaction vessel, the back surface of the first nitride semiconductor layer 3 is lapped to remove the sapphire substrate to form a mirror surface, and a nitride semiconductor made of Si-doped GaN. Get the substrate.
The number of crystal defects per unit area was measured in the same manner as in Example 6. As a result, the number of crystal defects was 1 × 10 ³ / cm ² compared to Example 6, and the element substrate had very good crystallinity with almost no crystal defects. As a result, a nitride semiconductor substrate was obtained.
In addition, the number of crystal defects in Example 7 was further reduced as compared with the first nitride semiconductor layer 3 of Example 1.

[Embodiment 8 regarding growth of nitride semiconductor substrate ]
In Example 6, the first protective film 11 is formed on the sapphire substrate 1 in the same manner except that sapphire having the A surface as the main surface and the orientation flat surface as the R surface is formed on the substrate 1. A first nitride semiconductor layer 3 made of Si-doped GaN is grown to a thickness of 100 μm. The shape of the first protective film 11 is a stripe perpendicular to the R plane. As a result, a nitride semiconductor layer with very few crystal defects almost the same as in Example 6 could be grown.

[Example 9 regarding growth of nitride semiconductor substrate ]
In Example 5, the first protective film 11 made of SiO ₂ was formed in the same manner except that the stripe width was 10 μm, the window portion was 3 μm, and the thickness was 1 μm, and Si having a thickness of 150 μm was 1 × 10 ¹⁸ / Grow cm ³ doped GaN. After the growth, the wafer was taken out of the reaction vessel, the spinel substrate was lapped and removed, and the number of crystal defects was measured. As a result, a nitride semiconductor layer having very little crystal defects almost equal to or more than that of Example 5 was obtained. was gotten.

[Example 10: Reference example ]
FIG. 11 is a schematic cross-sectional view showing the structure of one LED element using a nitride semiconductor layer obtained by the growth method of the present invention as a substrate. Hereinafter, Example 10 is demonstrated based on FIG.

The surface of the first nitride semiconductor layer 3 is polished and removed by removing the sapphire substrate 1, the buffer layer, the first protective film 11, and a part of the first nitride semiconductor layer 3 of the wafer obtained in Example 2. To expose only the first nitride semiconductor layer 3. A wafer mainly having the first nitride semiconductor layer 3 (Si-doped GaN) is set in a reaction vessel of the MOVPE apparatus, and the heterogeneous substrate 1 and the like of the first nitride semiconductor layer 3 are removed at 1050 ° C. A second buffer layer 31 made of GaN doped with Si at 1 × 10 ¹⁸ / cm ³ is grown on the surface opposite to the exposed surface. This second buffer layer 31 is a nitride semiconductor single crystal layer grown at a high temperature of usually 900 ° C. or higher, and is distinguished from the buffer layer 2 grown at a low temperature for relaxing the lattice mismatch with the previous substrate. The

Further, an active layer 32 made of In _0.4 Ga _0.6 N having a single quantum well structure on the second buffer layer 31 and a p-side made of Mg-doped Al _0.2 Ga _0.8 N having a thickness of 0.3 μm. A cladding layer 33 and a p-side contact layer 34 made of Mg-doped GaN with a thickness of 0.5 μm are grown in this order.

  After the growth of the second buffer layer 31 to the p-side contact layer 34 to be an element structure, the wafer is taken out of the reaction vessel and annealed at 600 ° C. in a nitrogen atmosphere to make the p-side cladding layer 33 and the p-side contact layer 34 low resistance To. Thereafter, etching is performed from the p-side contact layer 34 side to expose the surface of the first nitride semiconductor layer 3. Thus, by exposing the nitride semiconductor layer below the active layer by etching and providing a “cutting edge” at the time of cutting the chip, it becomes difficult to give an impact to the pn junction surface at the time of cutting. An improved and highly reliable device can be obtained.

  After the etching, a light-transmitting p-electrode 35 made of Ni / Au is formed on the almost entire surface of the p-side contact layer 34 with a film thickness of 200 angstroms, and a bonding pad electrode 36 is formed on the p-electrode 35. Is formed with a film thickness of 0.5 μm. FIG. 12 shows a plan view of the chip after forming the p-electrode (viewed from the pad electrode 36 side).

  After forming the p-side electrode, an n-electrode 37 is formed to a thickness of 0.5 μm on the entire surface of the first nitride semiconductor layer 3 exposed by removing the sapphire substrate 1 and the like.

  Thereafter, scribing is performed from the n-electrode side, and cleavage is performed at the M plane (101-0) of the first nitride semiconductor layer 3 and a plane perpendicular to the M plane, thereby obtaining a 300 μm square LED chip. This LED emits green light of 520 nm at 20 mA, the output is more than twice that of a conventional sapphire substrate grown with a nitride semiconductor device structure, and the electrostatic withstand voltage is more than twice as much. Excellent properties were shown.

[Example 11: Example of the present case ]
FIG. 13 is a schematic cross-sectional view showing the structure of one laser device of the present invention. Hereinafter, Example 11 is demonstrated based on FIG.

Part of the sapphire substrate 1, buffer layer, first protective film 11, first nitride semiconductor layer 3, second protective film 12, and second nitride semiconductor layer 4 of the wafer obtained in Example 3 Then, a wafer having the second nitride semiconductor layer 4 (Si-doped GaN) as the main surface is set in a reaction vessel of the MOVPE apparatus, and the sapphire substrate 1 of the second nitride semiconductor layer is formed at 1050 ° C. A third buffer layer 41 made of GaN doped with Si at 1 × 10 ¹⁸ / cm ³ is grown on the surface opposite to the surface exposed by removing the like. The third buffer layer 41 is a nitride semiconductor single crystal layer that is grown at a high temperature of 900 ° C. or higher, as in the tenth embodiment, and is used to alleviate lattice mismatch between the conventionally grown substrate and the nitride semiconductor. A distinction is made from buffer layers grown at low temperatures. In the case of manufacturing a laser device, the third buffer layer 41 is a strained superlattice formed by stacking nitride semiconductors having different compositions from each other with a film thickness of 100 angstroms or less, more preferably 70 angstroms or less, most preferably 50 angstroms or less. A layer is preferred. When the strained superlattice layer is used, the crystallinity of the single nitride semiconductor layer is improved, so that a high-power laser element can be realized. A strained superlattice layer may be applied to the cladding layer of the LED element.

(Crack prevention layer 42)
Next, a crack preventing layer 42 made of In _0.1 Ga _0.9 N doped with 5 × 10 ¹⁸ / cm ^{3 of} Si is grown to a thickness of 500 Å. The crack prevention layer 42 can be prevented from being cracked in the nitride semiconductor layer containing Al by growing it with an n-type nitride semiconductor containing In, preferably InGaN. The crack prevention layer is preferably grown with a film thickness of 100 Å or more and 0.5 μm or less. If it is thinner than 100 angstroms, it is difficult to act as a crack prevention as described above, and if it is thicker than 0.5 μm, the crystal itself tends to turn black. The crack prevention layer 42 can be omitted.

(N-side cladding layer 43)
Next, a first layer made of n-type Al _0.2 Ga _0.8 N doped with Si 5 × 10 ¹⁸ / cm ³ , 20 angstroms, and a second layer made of undoped GaN, alternating with 20 angstroms A superlattice structure having a total film thickness of 0.4 μm is formed by laminating 100 layers. The n-side cladding layer 43 functions as a carrier confinement layer and an optical confinement layer, and is desirably a nitride semiconductor containing Al, preferably a superlattice layer containing AlGaN, and the total thickness of the superlattice layer is 100 angstroms or more. It is desirable to grow at 2 μm or less, more preferably 500 Å or more and 1 μm or less. When the superlattice layer is used, a carrier confinement layer having good crystallinity without cracks can be formed.

(N-side light guide layer 44)
Subsequently, an n-type light guide layer 44 made of n-type GaN doped with Si at 5 × 10 ¹⁸ / cm ³ is grown to a thickness of 0.1 μm. The n-side light guide layer 44 acts as a light guide layer of the active layer, and it is desirable to grow GaN and InGaN. Usually, the n-side light guide layer 44 is grown to a thickness of 100 Å to 5 μm, more preferably 200 Å to 1 μm. desirable. The n-side light guide layer 44 is usually doped with an n-type impurity such as Si or Ge so as to have an n-type conductivity type, but can be particularly undoped. In the case of a superlattice, at least one of the first layer and the second layer may be doped with an n-type impurity or may be undoped.

(Active layer 45)
Next, a well layer made of undoped In _0.2 Ga _0.8 N, 25 Å, a barrier layer made of undoped In _0.01 Ga _0.95 N, and a multi-quantum well structure having a total thickness of 175 Å formed by alternately stacking 50 Å The active layer 45 of MQW) is grown. The well layer and / or the barrier layer may be doped with Si. Doping with Si is preferable because the threshold value is lowered.

(P-side cap layer 46)
Next, the p-side cap layer 46 made of p-type Al _0.3 Ga _0.9 N doped with 1 × 10 ²⁰ / cm ^{3 of} Mg and having a band gap energy larger than that of the p-side light guide layer 47 and larger than that of the active layer 45. Is grown to a film thickness of 300 Å. The p-side cap layer 46 is p-type, but since it is thin, it may be i-type in which carriers are compensated by doping n-type impurities, or undoped, and most preferably a layer doped with p-type impurities. And The film thickness of the p-side cap layer 17 is adjusted to 0.1 μm or less, more preferably 500 angstroms or less, and most preferably 300 angstroms or less. This is because if the film is grown to a thickness greater than 0.1 μm, cracks are likely to occur in the p-type cap layer 46, and a nitride semiconductor layer with good crystallinity is difficult to grow. When the AlGaN having a larger Al composition ratio is formed thinner, the LD element tends to oscillate. For example, if the Y value is Al _Y Ga _1-Y N of 0.2 or more, it is desirable to adjust it to 500 angstroms or less. Although the lower limit of the film thickness of the p-side cap layer 46 is not particularly limited, it is desirable to form the p-side cap layer 46 with a film thickness of 10 angstroms or more.

(P-side light guide layer 47)
Next, a p-side light guide layer 47 made of p-type GaN doped with Mg at 1 × 10 ²⁰ / cm ³ and having a band gap energy smaller than that of the p-side cap layer 46 is grown to a thickness of 0.1 μm. This layer acts as a light guide layer of the active layer, and is preferably grown of GaN and InGaN as with the n-side light guide layer 44. This layer also functions as a buffer layer when the p-side cladding layer 48 is grown, and functions as a preferable light guide layer by growing it at a film thickness of 100 angstroms to 5 μm, more preferably 200 angstroms to 1 μm. . This p-side light guide layer is usually doped with a p-type impurity such as Mg to have a p-type conductivity, but it is not particularly necessary to dope the impurity. The p-type light guide layer can be a superlattice layer. In the case of a superlattice layer, at least one of the first layer and the second layer may be doped with a p-type impurity or may be undoped.

(P-side cladding layer 48)
Next, a first layer made of p-type Al _0.2 Ga _0.8 N doped with Mg 1 × 10 ²⁰ / cm ³ , 20 Å, and a second layer made of p-type GaN doped with Mg 1 × 10 ²⁰ / cm ³ A p-side cladding layer 48 made of a superlattice layer having a total thickness of 0.4 μm is formed by alternately laminating these layers and 20 angstroms. This layer acts as a carrier confinement layer like the n-side cladding layer 43, and acts as a layer for reducing the resistivity on the p-type layer side by adopting a superlattice structure. The thickness of the p-side cladding layer 48 is not particularly limited, but it is desirable that the p-side cladding layer 48 be grown at a thickness of 100 angstroms or more and 2 μm or less, more preferably 500 angstroms or more and 1 μm or less. In this embodiment, the superlattice layer is also provided on the n-side cladding layer side, but the resistance value of the p-layer tends to decrease when the superlattice layer is provided on the p-side layer side rather than the n-side cladding layer side. Therefore, it is preferable in reducing Vf.

  In the case of a nitride semiconductor element having a double hetero structure having an active layer 45 having a quantum well layer, Al having a film thickness of 0.1 μm or less having a band gap energy larger than that of the active layer 45 is in contact with the active layer 45. A cap layer 46 made of a nitride semiconductor is provided, and a p-side light guide layer 47 having a bad gap energy smaller than that of the cap layer 46 is provided at a position farther from the active layer than the cap layer 46. The p-side light guide It is very preferable to provide a p-side cladding layer 48 made of a superlattice layer containing a nitride semiconductor containing Al having a band gap larger than that of the p-side light guide layer 47 at a position farther from the active layer than the layer 47. In addition, since the electrons injected from the n layer, which has a large band gap energy of the p-side cap layer 46, are blocked by the cap layer 46, the electrons do not overflow the active layer, so that the leakage current of the element is small. Become.

(P-side contact layer 49)
Finally, a p-side contact layer 49 made of p-type GaN doped with 2 × 10 ²⁰ / cm ^{3 of} Mg is grown to a thickness of 150 Å. The thickness of the p-side contact layer is adjusted to 500 angstroms or less, more preferably 400 angstroms or less, and 20 angstroms or more.

  After the completion of the reaction, the wafer is annealed in a reaction vessel at 700 ° C. in a nitrogen atmosphere to further reduce the resistance of the p-type layer. After annealing, the wafer is taken out of the reaction vessel, and as shown in FIG. 13, the uppermost p-type contact layer 20 and p-type cladding layer 19 are etched by an RIE apparatus to form a ridge shape having a stripe width of 4 μm. Then, a p-electrode 51 made of Ni / Au is formed on the entire surface of the ridge. Next, as shown in FIG. 13, an insulating film 50 made of SiO 2 is formed on the surfaces of the p-side cladding layer 48 and the contact layer 49 excluding the p-electrode 51, and electrically connected to the p-electrode 51 through the insulating film 50. A connected p-pad electrode 52 is formed.

  After the formation of the p-side electrode, the n-electrode 53 made of Ti / Al is formed on the entire surface of the second nitride semiconductor layer 4 having no exposed device structure by polishing and removing the sapphire substrate 1 and the like of the wafer. A thin film made of Au / Sn is formed thereon for metallization with a heat sink.

Thereafter, scribing is performed from the n-electrode side 53, and the second nitride semiconductor layer 4 is cleaved at the M plane of the second nitride semiconductor layer 4 (11-00, the plane corresponding to the side face of the hexagonal column in FIG. 9). A resonant surface is produced. A dielectric multilayer film made of SiO ₂ and TiO ₂ was formed on the resonance surface, and finally a bar was cut in a direction parallel to the p-electrode to form a laser chip. Next, the chip was placed face-up (with the substrate and heat sink facing each other) on the heat sink, the p-pad electrode 52 was wire-bonded, and laser oscillation was attempted at room temperature. At room temperature, the threshold current density was 2.0 kA. / Cm ² and a threshold voltage of 4.0 V, continuous oscillation with an oscillation wavelength of 405 nm was confirmed, and a lifetime of 1000 hours or more was exhibited.

[Example 12: Reference example ]
FIG. 14 is a schematic cross-sectional view showing the structure of one LED device using the nitride semiconductor layer obtained by the growth method of the present invention as a substrate, and as an element structure above the substrate made of the second nitride semiconductor layer 4. Has the same structure as the LED element of Example 10.
As the nitride semiconductor substrate of the LED element of Example 12, the film thickness of the first nitride semiconductor layer 3 made of GaN doped with Si in Example 3 is 25 μm, and the second nitride semiconductor layer 4 is An undoped GaN film is used in the same manner except that the film thickness is 25 μm.
On the second nitride semiconductor layer 4 made of undoped GaN thus obtained, a second buffer layer 31 made of GaN doped with Si of 1 × 10 ¹⁸ / cm ³ , a film thickness of 20 Å, An active layer 32 made of In _0.4 Ga _0.6 N having a single quantum well structure, a p-side cladding layer 33 made of Mg-doped Al _0.2 Ga _0.8 N with a thickness of 0.3 μm, and a p-side made of Mg-doped GaN with a thickness of 0.5 μm The contact layer 34 has a structure in which the p-side contact layer 34 is laminated in order, and a translucent p-electrode 35 is formed on almost the entire surface of the p-side contact layer 34, and a bonding pad electrode 36 is formed on the p-electrode 35. Has been.
Note that all of the substrate 1, the buffer layer 2, the first nitride semiconductor layer 3, the first protective film 11, the second protective film 12, and a part of the second nitride semiconductor layer 4 are in Example 10. In the same manner as in the present embodiment, the n-electrode and the p-electrode can be provided on the same surface side.

This element is different from the element of Example 10 in that the element is formed on the second nitride semiconductor layer 4 having better crystallinity than the first nitride semiconductor layer 3 used as the nitride semiconductor substrate in Example 10. A structure is formed, and a p-electrode 35 and a negative electrode 37 are provided on the same surface side.
In the nitride semiconductor device having a structure in which a nitride semiconductor layer doped with an n-type impurity (second buffer layer 31) is stacked on the second nitride semiconductor layer 4 made of undoped GaN, on the n-type layer side. When the n electrode is provided, an LED element having a low _Vf and high luminous efficiency tends to be easily obtained when the n electrode is provided on the nitride semiconductor layer doped with the n-type impurity. In addition, this LED element improved the output about 1.5 times and the electrostatic withstand voltage about 1.5 times compared with the LED element of Example 10.

[Example 13: Reference example ]
As in Example 1, a wafer was prepared by growing a buffer layer made of GaN on a sapphire substrate having a sapphire C surface as a main surface and an orientation flat surface as an A surface, and an undoped GaN layer 2 of 4 μm, Using a CVD apparatus, a wafer in which a first protective film made of SiO 2 having a stripe width of 20 μm and a window part of 5 μm is formed on the undoped GaN layer 2 to a thickness of 0.1 μm is transferred to the MOVPE apparatus. On the undoped GaN layer 2 and the first protective film, a first nitride semiconductor layer made of GaN doped with Si at 1 × 10 ¹⁹ / cm ³ is grown to a thickness of 15 μm to grow an element structure. A nitride semiconductor substrate is formed.

Thereafter, in the same manner as in Example 12, a second buffer layer made of GaN doped with Si at 1 × 10 ¹⁸ / cm ³ on the first nitride semiconductor layer, a film thickness of 20 Å, a single quantum well An active layer made of In _0.4 Ga _0.6 N, a p-side cladding layer made of Mg-doped Al _0.2 Ga _0.8 N with a thickness of 0.3 μm, and a p-side contact layer made of Mg-doped GaN with a thickness of 0.5 μm are sequentially stacked. To do. Etching is performed from the p-layer side to expose the surface of the first nitride semiconductor layer 3 having a high impurity concentration to form an n-electrode, while a light-transmitting p-electrode is formed on the entire surface of the p-side contact layer, A pad electrode for bonding is formed on the p electrode, and an n electrode and a p electrode are provided from the same surface side as shown in FIG. Finally, the thickness of the sapphire substrate is polished to about 50 μm and thinned, and then the polished surface side is scribed to form a 350 μm square element. In this LED element, the first protective film is formed on the GaN layer, and the first nitride semiconductor layer is formed with the window portion of the first protective film being 5 μm. Compared with the LED element of Example 10 formed by forming a protective film and forming a first nitride semiconductor layer with a window portion of 6 μm, better characteristics were exhibited.

[Example 14: Reference example ]
A 2-inch φ sapphire substrate having an off-angle angle θ = 0.13 ° from the C-plane, a step of about 15 Å, a terrace width W of about 56 Å, and an orientation flat surface as the A-plane is prepared. FIG. 15 is a schematic view showing an enlarged cross section of the sapphire substrate. The substrate off-angled in a step shape shown in FIG. 15 has a substantially horizontal terrace portion A and a stepped portion B. The surface unevenness of the terrace portion A is adjusted to about 0.5 angstroms on average and about 2 angstroms at maximum, and is formed almost regularly. On the other hand, the height of the step portion is adjusted to about 15 angstroms. Although the off angle θ is exaggerated, it is inclined only 0.13 ° with respect to the horizontal plane of the growth surface. Such a step-like portion having an off angle is desirably formed continuously over the entire substrate, but may be formed partially in particular. As shown in FIG. 15, the off angle θ refers to an angle between a straight line connecting the bottoms of a plurality of steps and a horizontal plane of the uppermost step. The step difference is 30 angstroms or less, more preferably 25 angstroms or less, and most preferably 20 angstroms or less. The lower limit is preferably 2 angstroms or more. In particular, when a sapphire C-plane is used for the substrate, the off-angle θ from the C-plane is adjusted to within 1 degree, preferably 0.8 degrees or less, and more preferably 0.6 degrees or less. In this embodiment, a step-like off-substrate is used, but a normal off-substrate may be used instead of a step-like substrate. By using an appropriately off-angled heterogeneous substrate, the interatomic distance between the nitride semiconductor and the heterogeneous substrate becomes close, step growth is possible, and a GaN substrate with fewer crystal defects can be obtained.

  As in Example 13, a GaN buffer layer was grown to 200 Å and an undoped GaN layer was grown to 4 μm on the off-angle surface of the sapphire substrate, and then a stripe width was formed on the undoped GaN layer using a CVD apparatus. A first protective film made of SiO 2 having a thickness of 25 μm and a window portion of 5 μm is formed to a thickness of 0.1 μm. Similarly, the stripe direction of the first protective film is perpendicular to the A plane.

Next, this wafer is transferred to a MOVPE apparatus, and a first nitride semiconductor layer made of GaN doped with Si at 1 × 10 ¹⁹ / cm ³ is formed on the undoped GaN layer and the first protective film with a thickness of 10 μm. A second buffer layer made of GaN doped with Si at 1 × 10 ¹⁸ / cm ³ , having a thickness of 20 Å, and having a single quantum well structure, In _0.4 is grown on the first nitride semiconductor layer. An active layer made of Ga _0.6 N, a p-side cladding layer made of Mg-doped Al _0.2 Ga _0.8 N with a thickness of 0.3 μm, and a p-side contact layer made of Mg-doped GaN with a thickness of 0.5 μm are laminated in this order. Thereafter, in the same manner as in Example 13, the surface of the first nitride semiconductor layer is exposed by etching, and an n electrode and a p electrode are provided from the same side as shown in FIG. Then, after the thickness of the sapphire substrate is polished down to about 50 μm and thinned, a 350 μm square element is obtained. This LED element was improved by about 5% in output compared with the LED element of Example 13.

[Example 15: Reference example ]
In Example 13, after growing the first nitride semiconductor layer 10 μm of GaN doped with Si at 1 × 10 ¹⁹ / cm ³ , the wafer is taken out of the reaction vessel, and the position corresponding to the window portion of the first protective film Then, a second protective film having a stripe width of 15 μm is formed to a thickness of 0.1 μm. Then, the wafer is transferred again to the MOVPE apparatus, and a second nitride semiconductor layer made of GaN doped with Si at 1 × 10 ¹⁹ / cm ³ is formed on the first nitride semiconductor layer and the second protective film. Growing with a film thickness of 15 μm.

Thereafter, in the same manner as in Example 13, on the second nitride semiconductor layer, a second buffer layer made of GaN doped with Si at 1 × 10 ¹⁸ / cm ³ , a film thickness of 20 Å, a single quantum well An active layer made of In _0.4 Ga _0.6 N, a p-side cladding layer made of Mg-doped Al _0.2 Ga _0.8 N with a thickness of 0.3 μm, and a p-side contact layer made of Mg-doped GaN with a thickness of 0.5 μm are sequentially stacked. Then, an n-electrode is formed by exposing the surface of the second nitride semiconductor layer, a translucent p-electrode and a pad electrode are formed on almost the entire surface of the p-side contact layer, and an n-electrode is formed from the same surface side. A p-electrode is provided. Finally, the thickness of the sapphire substrate is polished to about 50 μm and thinned, and then the polished surface side is scribed to form a 350 μm square element. This LED element showed better characteristics than the LED element of Example 12.

[Example 16: Reference example ]
In Example 10, the first nitride semiconductor layer 3 was grown on the nitride semiconductor substrate forming the element structure in the same manner as in Example 6, and the sapphire substrate 1, buffer layer, protective film 11 and the like of this wafer were formed. The LED chip is obtained in the same manner except that the surface of the first nitride semiconductor layer 3 is exposed by polishing and removing, and only the first nitride semiconductor layer 3 is used as a nitride semiconductor substrate.
This LED showed very good characteristics as in Example 10, but Example 16 was better than Example 10 of the present invention.

[Example 17: Example of the present case ]
Hereinafter, Example 17 is demonstrated based on FIG. FIG. 16 is a schematic cross-sectional view showing the structure of one laser device using the nitride semiconductor layer obtained by the growth method of the present invention as a substrate.

  The sapphire substrate 1, the buffer layer 2, and the protective film 11 of the wafer obtained in Example 6 are polished and removed, the surface of the first nitride semiconductor layer 3 is exposed, and only the first nitride semiconductor layer 3 is exposed. To do.

  Next, a wafer having the first nitride semiconductor layer 3 (Si-doped GaN) as a main surface is set in a reaction vessel of the MOVPE apparatus, and the heterogeneous substrate of the first nitride semiconductor layer 3 is removed. The following layers are formed on the surface opposite to the exposed surface.

(N-side cladding layer 43)
Next, a first layer made of n-type Al _0.2 Ga _0.8 N doped with Si of 1 × 10 ¹⁹ / cm ³ , 20 angstroms, and a second layer made of undoped GaN, alternating with 20 angstroms A superlattice structure having a total film thickness of 0.4 μm is formed by laminating 100 layers.

(N-side light guide layer 44)
Subsequently, an n-type light guide layer 44 made of n-type GaN doped with Si at 1 × 10 ¹⁷ / cm ³ is grown to a thickness of 0.1 μm.

(Active layer 45)
Next, a well layer made of In _0.2 Ga _0.8 N doped with Si of 1 × 10 ¹⁷ / cm ³ , 25 Å, and a barrier layer made of In _0.01 Ga _0.95 N doped with Si of 1 × 10 ¹⁷ / cm ³ , 50 An active layer 45 of a multiple quantum well structure (MQW) having a total film thickness of 175 Å formed by alternately stacking angstroms is grown.

(P-side cap layer 46)
Next, the p-side cap layer 46 made of p-type Al _0.3 Ga _0.9 N doped with 1 × 10 ²⁰ / cm ^{3 of} Mg and having a band gap energy larger than that of the p-side light guide layer 47 and larger than that of the active layer 45. Is grown to a film thickness of 300 Å.

(P-side light guide layer 47)
Next, a p-side light guide layer 47 made of p-type GaN doped with Mg at 1 × 10 ¹⁸ / cm ³ and having a band gap energy smaller than that of the p-side cap layer 46 is grown to a thickness of 0.1 μm.

(P-side cladding layer 48)
Next, a first layer made of p-type Al _0.2 Ga _0.8 N doped with Mg 1 × 10 ²⁰ / cm ³ , 20 Å, and a second layer made of p-type GaN doped with Mg 1 × 10 ²⁰ / cm ³ A p-side cladding layer 48 made of a superlattice layer having a total thickness of 0.4 μm is formed by alternately laminating these layers and 20 angstroms.

(P-side contact layer 49)
Finally, a p-side contact layer 49 made of p-type GaN doped with 2 × 10 ²⁰ / cm ^{3 of} Mg is grown to a thickness of 150 Å.

After the completion of the reaction, the wafer is annealed in a reaction vessel at 700 ° C. in a nitrogen atmosphere to further reduce the resistance of the p-type layer. After annealing, the wafer is taken out of the reaction vessel, and as shown in FIG. 16, the uppermost p-type contact layer 20 and p-type cladding layer 19 are etched by an RIE apparatus to form a ridge shape having a stripe width of 4 μm. Then, a p-electrode 51 made of Ni / Au is formed on the entire surface of the ridge. Next, as shown in FIG. 16, an insulating film 50 made of SiO ₂ is formed on the surfaces of the p-side cladding layer 48 and the contact layer 49 except for the p-electrode 51, and the p-electrode 51 is electrically connected via the insulating film 50. A p-pad electrode 52 connected to is formed.

  After forming the p-side electrode, an n-electrode 53 made of Ti / Al is formed to a thickness of 0.5 μm on the entire surface of the first nitride semiconductor layer 3 where the element structure is not formed, and a heat sink and A thin film made of Au / Sn is formed for metallization.

Thereafter, scribing is performed from the n-electrode side 53, and the first nitride semiconductor layer 3 is cleaved at the M plane of the first nitride semiconductor layer 3 (11-00, the plane corresponding to the side face of the hexagonal column in FIG. 9). A resonant surface is produced. A dielectric multilayer film made of SiO ₂ and TiO ₂ was formed on both or one of the resonance surfaces, and finally a bar was cut in a direction parallel to the p-electrode to form a laser chip. Next, the chip was placed face-up (with the substrate and heat sink facing each other) on the heat sink, the p-pad electrode 52 was wire-bonded, and laser oscillation was attempted at room temperature. At room temperature, the threshold current density was 2.0 kA. / Cm ² and a threshold voltage of 4.0 V, continuous oscillation with an oscillation wavelength of 405 nm was confirmed, and a lifetime of 1000 hours or more was exhibited.

[Example 18: Reference example ]
FIG. 17 is a schematic cross-sectional view showing the structure of one LED device using a nitride semiconductor layer obtained by the growth method of the present invention as a substrate. The element structure above the active layer 32 has the same structure as the LED element of Example 16. Further, as the nitride semiconductor substrate of the LED element of Example 18, the first nitride semiconductor layer 3 made of undoped GaN was grown in the same manner as in Example 7, and the sapphire substrate 1, buffer layer, and nitride of this wafer were grown. The semiconductor layer 2, the first protective film 11, etc. are removed to use only the first nitride semiconductor layer 3. An n-side cladding layer 51 having the following superlattice layer is formed on the surface of the first nitride semiconductor layer 3 opposite to the surface exposed by removing the sapphire substrate 1 and the first protective film 11. Grow.
(N-side cladding layer 51)
100 first alternating layers of n-type Al _0.2 Ga _0.8 N doped with 1 × 10 ¹⁹ / cm ^{3 of} Si, 20 angstroms, and second layer of undoped GaN, 20 angstroms A superlattice structure having a total thickness of 0.4 μm is formed. By using a superlattice layer, a carrier confinement clad layer having good crystallinity without cracks can be formed.

  Next, the active layer 32, the p-side cladding layer 33, and the p-side contact layer 34 are sequentially stacked on the formed cladding layer 51 in the same manner as in the sixteenth embodiment. Etching is performed from the p-layer side to expose the surface of the n-side cladding layer 51 having a high impurity concentration to form an n-electrode. On the other hand, a translucent p-electrode is formed on almost the entire surface of the p-side contact layer, and the p-electrode. A pad electrode for bonding is formed thereon, and an n electrode and a p electrode are provided from the same surface side as shown in FIG. Finally, the thickness of the sapphire substrate is polished to about 50 μm and thinned, and then the polished surface side is scribed to form a 350 μm square element.

  The obtained LED element showed good characteristics, and further, the output was improved about 1.5 times and the electrostatic withstand voltage was about 1.5 times compared with the LED element of Example 16.

[Example 19: Reference example ]
In Example 7, the first nitride semiconductor layer 3 to be a nitride semiconductor substrate is grown in the same manner except that the thickness of the first nitride semiconductor layer 3 made of undoped GaN is set to 15 μm. An element structure is formed on the first nitride semiconductor layer 3 in the same manner as in Example 18 to obtain an LED element.
The obtained LED element showed the favorable characteristic similarly to the LED element of Example 18.

[Example 20: Reference example ]
In Example 19, an LED element is obtained in the same manner as in Example 14 except that a sapphire substrate that is off-angled stepwise is used as in the case of Example 14.
This LED element improved by about 5% in output compared with the LED element of Example 19.

[Example 21 regarding growth of nitride semiconductor substrate ]
Three types of nitride semiconductor substrates were formed in the same manner as in Example 7, except that the width of the window was set to 5 μm, 3 μm, and 1 μm. As a result, the number of crystal defects was reduced by about 20% when the width of the window portion was 5 μm and when the width was 3 μm and 1 μm.

The typical sectional view showing the structure of the nitride semiconductor wafer obtained in each process of the lateral growth method used for the present invention . The typical sectional view showing the structure of the nitride semiconductor wafer obtained in each process of the lateral growth method used for the present invention . The typical sectional view showing the structure of the nitride semiconductor wafer obtained in each process of the lateral growth method used for the present invention . The typical sectional view showing the structure of the nitride semiconductor wafer obtained in each process of the lateral growth method used for the present invention . The typical sectional view showing the structure of the nitride semiconductor wafer obtained in each process of the lateral growth method used for the present invention . The typical sectional view showing the structure of the nitride semiconductor wafer obtained in each process of the lateral growth method used for the present invention . The typical sectional view showing the structure of the nitride semiconductor wafer obtained in each process of the lateral growth method used for the present invention . The typical sectional view showing the structure of the nitride semiconductor wafer obtained in each process of the lateral growth method used for the present invention . The unit cell figure which shows the surface orientation of sapphire. The top view of the board | substrate main surface side for demonstrating the stripe direction of a protective film. The schematic cross section which shows one structure of the nitride semiconductor LED element using the board | substrate by the lateral growth method used for this invention . The top view which looked at the element of FIG. 11 from the p electrode side. 1 is a schematic cross-sectional view showing one structure of a nitride semiconductor LD element of the present invention . The schematic cross section which shows one structure of the nitride semiconductor LED element using the board | substrate by the lateral growth method used for this invention . The schematic cross section which shows the partial shape of one different board | substrate which carried out the off angle. It is a schematic sectional view showing one structure of a nitride compound semiconductor LD device of the present invention. It is a schematic cross section which shows one structure of the nitride semiconductor LED element using the board | substrate by the lateral growth method used for this invention .

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 ... substrate 2 ... buffer layer 3 ... 1st nitride semiconductor layer 4 ... 2nd nitride semiconductor layer 11 ... 1st protective film 12 ... ..Second protective film

Claims (3)

After a nitride semiconductor layer doped with an n-type impurity is laterally grown on a heterogeneous substrate having a main surface with an off angle of 1 ° or less at an off angle , the nitride semiconductor layer is grown to 70 to 500 μm, Forming the nitride semiconductor substrate by removing the heterogeneous substrate ;
A laser element structure having a ridge stripe on the nitride semiconductor substrate and having a p-electrode formed on the entire surface of the ridge stripe is formed so that the (11-00) plane of the nitride semiconductor layer becomes a resonance plane. And
Forming an electrode on the back surface of the nitride semiconductor substrate;
A method of manufacturing a nitride semiconductor laser element, wherein a dielectric film is formed on a resonance surface of the laser element structure. 2. The method of manufacturing a nitride semiconductor laser device according to claim 1 , wherein the n-type impurity is Si. Manufacturing method of the nitride semiconductor laser device according to claim 1 or 2, characterized in that the crystal defect density of the nitride semiconductor layer is 10 ⁶ / cm ² or less.

Images