JP2007208300A - Semiconductor laser, and method of manufacturing same - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a semiconductor laser where, a threshold current density for a short wavelength semiconductor laser using a nitride compound semiconductor can be reduced, and to provide a method of manufacturing the same. <P>SOLUTION: In an electron block layer disposed at the p-side of the active layer of a semiconductor laser using a nitride compound semiconductor, the Mg concentration thereof is set equal to or greater than 7×10<SP>19</SP>cm<SP>-3</SP>. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

The present invention relates to a semiconductor laser and a method for manufacturing the same, and more particularly to a semiconductor laser characterized by a configuration for reducing the threshold current density _Jth in a semiconductor laser using a nitride compound semiconductor and a method for manufacturing the same. It is about.

  Conventionally, short-wavelength semiconductor lasers have been used as light sources for optical disks, DVDs, etc., but since the recording density of optical disks is inversely proportional to the square of the wavelength of the laser light, shorter-wavelength semiconductor lasers have been demanded. The shortest-wavelength semiconductor laser that has been commercialized is a red semiconductor laser having a wavelength in the vicinity of 630 to 650 nm, and is used in the DVD released last year.

  However, in order to further increase the recording density, it is necessary to further shorten the wavelength. For example, in order to record a moving image on an optical disk for 2 hours, a blue semiconductor laser having a wavelength of around 400 nm is indispensable. As a light source for an optical disk, a short wavelength semiconductor laser having a wavelength in a blue region has been actively developed.

  As such blue semiconductor laser materials, IISe-VI group compound semiconductor ZnSe series and III-V group compound semiconductor GaN series have been studied. Of these, ZnSe series has been proven as a high-quality substrate. It is considered that the ZnSe system is more advantageous for a long time because it is lattice-matched to GaAs, which has a high level of GaAs. Most of the researchers all over the world have been engaged in research on this ZnSe system. Regarding the research, the ZnSe system is ahead.

  As for this ZnSe system, room temperature continuous oscillation by injection excitation has already been reported, but since it is a material that is inherently easily deteriorated, reliability is a problem, and it has not yet been put into practical use.

  On the other hand, in the case of the GaN system, GaN with excellent environmental resistance has been reconsidered with respect to the reliability that has become a bottleneck in the ZnSe system after the announcement of GaN high-brightness LED by Nichia Chemical at the end of 1993. We are seeing a large increase in researchers.

  Next, in early December 1995, Nichia also reported the success of pulsed laser oscillation, and research has progressed rapidly. In room temperature continuous oscillation (CW oscillation), a 35-hour oscillation duration was reported. Since then, an oscillation duration of an estimated 10,000 hours has been reported in an accelerated test.

Here, a conventional short wavelength semiconductor light emitting device will be described with reference to FIGS. 9 and 10. FIG. 9A is a schematic cross-sectional view perpendicular to the optical axis of a conventional short wavelength semiconductor laser. 9 (b) is a schematic sectional view of a short wavelength light emitting diode, and FIG. 10 is a schematic sectional view perpendicular to the optical axis of a short wavelength semiconductor laser having a different buffer layer structure.
9A. First, an n-type GaN buffer layer 813, an n-type In _0.1 Ga _0.9 N layer 814, and an n-type are disposed on a sapphire substrate 811 having a (0001) plane as a main surface via a GaN buffer layer 812. Al _0.15 Ga _0.85 N clad layer 815, n-type GaN light guide layer 816, InGaN MQW active layer 817, p-type Al _0.2 Ga _0.8 N layer 818, p-type GaN light guide layer 819, p-type Al _0.15 Ga _0.85 N clad layer 820 Then, the p-type GaN contact layer 821 is epitaxially grown by the MOVPE method (metal organic vapor phase epitaxy).

  Next, a part of the n-type GaN buffer layer 813 is exposed by dry etching, an n-side electrode 822 made of Ti / Au is provided, and a p-side electrode made of Ni / Au is formed on the p-type GaN contact layer 821. After providing 823, dry etching is further performed to form a pair of parallel end faces, and this end face is used as a resonator face to succeed in pulsed laser oscillation (see, for example, Non-Patent Document 1). .

In the case of a light emitting diode, an n-type GaN layer 824, an n-type or p-type In _0.15 Ga _0.85 N active layer 825, and a GaN buffer layer 812 are provided on a sapphire substrate 811. Then, the p-type GaN layer 826 is epitaxially grown by the MOVPE method.

In this case, in order to obtain light emission luminance practical as a light emitting diode operating with low injection, the Si concentration or Zn concentration of the In _0.15 Ga _0.85 N active layer 825 is set to 1 × 10 ¹⁷ to 1 × 10 ²¹ cm ⁻³ . In addition, the layer thickness of the In _0.15 Ga _0.85 N active layer 825 needs to be 1 to 500 nm, and more preferably 10 to 100 nm (see, for example, Patent Document 1 or Patent Document 2).

FIG. 10 is a cross-sectional view perpendicular to the optical axis of another conventional short wavelength semiconductor laser. First, on a sapphire substrate 831 having a (0001) plane as a main surface, a GaN buffer layer 832 is interposed. n-type GaN intermediate layer 833, n-type Al _0.09 Ga _0.91 N cladding layer 834, n-type GaN light guide layer 835, MQW active layer 836, p-type Al _0.18 Ga _0.82 N overflow prevention layer 837, p-type GaN light guide layer 838 Then, the p-type Al _0.09 Ga _0.91 N clad layer 839 and the p-type GaN contact layer 840 are sequentially epitaxially grown by the MOVPE method.

Next, as in the case of FIG. 9A, the p-type GaN contact layer 840 and the p-type Al _0.09 Ga _0.91 N clad layer 839 are mesa-etched by dry etching, and a part of the n-type GaN intermediate layer 833 is removed. is exposed, provided with an n-side electrode 841 made of Ti / Au on the exposed portion of the n-type GaN intermediate layer 833, is formed on p-type GaN contact layer 840 through the SiO ₂ film 842 having a stripe-shaped opening Ni / A p-side electrode 843 made of Au is provided, and then dry etching is performed to form a pair of parallel end faces serving as a resonator surface.

It has also been proposed to provide an overflow prevention layer, that is, a carrier stopper layer on the n-type layer side (see, for example, Patent Document 3). In this case, the n-type impurity concentration is 1 × 10 ¹⁸ cm ^{−. 3} Si-doped n-type Al _0.15 Ga _0.85 N layer as a hole stopper layer, and Mg-doped p-type Al _0.15 Ga _0.85 N layer with a p-type impurity concentration of 5 × 10 ¹⁹ cm ⁻³ as an electron stopper layer The growth temperature is 1100 ° C., which is a normal growth temperature for growing GaN or AlGaN.
Japanese Patent Laid-Open No. 06-260682 Japanese Patent Laid-Open No. 06-260683 Japanese Patent Laid-Open No. 10-056236 S. Nakamura et al. , Japan Journal of Applied Physics, vol. 35, p. L74, 1996

However, in the case of the conventional short wavelength semiconductor laser, there is a problem that the threshold current density is about 3.6 kA / cm ² , which is very large. This is because a material such as a nitride compound semiconductor, that is, a nitride compound semiconductor, essentially requires a large carrier density in order to generate an optical gain.

  In other words, the semiconductor lasers that have been put into practical use use III-V group compound semiconductors with zinc blende type crystal structures such as AlGaAs and AlGaInP, whereas nitride compound semiconductors are extremely This is because it has a hexagonal wurtzite structure with a large forbidden band width, and has completely different physical properties from the zinc blende type crystal material.

  The major physical properties of such nitride compound semiconductors are hexagonal crystals and the presence of anisotropy in crystals, large forbidden band width and large effective mass, small spin-orbit interaction, and valence electrons. There are three bands in which three bands of HH (Heavy Hole), LH (Light Hole), and CH are close to each other.

More specifically, first, as a feature due to the large effective mass of holes, first,
a. The effective mass of the hole is large, so that the quasi-Fermi level E _{Fp of} the valence band is difficult to increase, and b. Since the effective mass of the hole is large and LO (longitudinal optical) phonon scattering is large, the mobility of the hole is small,
c. Since the effective mass m of the hole is large, the average hole velocity at the temperature T, that is, the heat velocity v _{p of the} hole is
From (m / 2) v _p ² = (3/2) kT,
v _p ∝m ^-1/2
It is mentioned that the heat velocity v _{p of the} hole is small.

Second, there are three bands of HH, LH, and CH close to the valence band, and the two effective masses of them are large, so that the quasi-Fermi level E _Fp for the hole is difficult to increase. Cause
d. Increase in quasi-Fermi level E _Fn of the conduction band required to achieve the population inversion is larger than the conventional material, and the barrier layer and the well layer, or, GaN constituting the active layer and the optical guide layer / The ratio between the energy discontinuity ΔE _V in the valence band and the difference ΔE _g of the forbidden band width in the heterojunction of InGaN or In _x Ga _1-x N / In _y Ga _1-y N, that is, ΔE _V / ΔE _g is as large as about 0.7 compared to about 0.4 in the conventional material system, and the band offset at the interface is 3: 7 and is biased toward the valence band side.

For these reasons, holes are not efficiently injected from the p-type layer side into the active layer. In particular, when the active layer has a multiple quantum well (MQW) structure, the poor hole injection efficiency is The hole density is uneven between the well layers (well layers), and electrons that are not effectively consumed in the active layer overflow to the p-side and overflow current to the p-side light guide layer or p-type cladding layer. Since the present inventors have found through simulation, this situation will be described with reference to FIGS.
In addition, about the simulation of FIG. 11 thru | or FIG. 15, it simulates by the structure which does not provide an electron block layer, ie, an overflow prevention layer.

FIG. 11 shows the change of the Fermi level in the vicinity of the active layer. As shown in the enlarged circle in the figure, the pseudo Fermi level E _Fp for the hole in the valence band is expressed in the active layer. It can be seen that it is on the lower energy side than the p-side light guide layer.

That is, in a normal material system, a pseudo-equilibrium state is reached by hole injection, and the quasi-Fermi level E _Fp of the active layer and the p-side light guide layer substantially coincide with each other. The mismatch between the quasi-Fermi level E _Fp between the active layer and the p-side light guide layer is very large, indicating that holes are not effectively injected from the p-side light guide layer into the active layer.

FIG. 12 is a diagram showing the layer position dependence of the hole current in the MQW structure short wavelength semiconductor laser having five well layers. The simulation results of where the injected holes disappear due to recombination are shown. As is apparent from the figure, the hole current injected from the p-type cladding layer is consumed by 4 kA / cm ² in the p-side light guide layer before reaching the active layer.

This consumed current is a reactive current that does not contribute to laser oscillation and leads to an increase in the threshold current density _Jth . Thus, the reason why recombination in the p-side light guide layer is large is as follows. This is probably because the efficiency of hole injection from the p-type cladding layer to the active layer is poor.

In addition, when the active layer has an MQW structure, the poor hole injection efficiency results in non-uniform hole density between the quantum well layers (well layers), which makes laser oscillation inefficient. This will be described with reference to FIGS.
FIG. 13 is a diagram showing a simulation result of the hole density distribution in the element thickness direction in the state of FIG. 11, and as is clear from the figure, the hole density in the MQW active layer is close to the p-side light guide layer. It is understood that it is so large and non-uniform.

FIG. 14 is a diagram showing a simulation result of the electron density distribution in the element film thickness direction in the state of FIG. 11 as well. As is apparent from FIG. 14, the MQW active layer injected from the n-side light guide layer side It is understood that the electron density in the region also becomes large and non-uniform as it goes toward the p-side light guide layer. This is due to the above-described non-uniform injection of holes, and the electrons are attracted to the holes to satisfy the charge neutrality condition. It is a result.

In this way, it is expected that the generation of optical gain in the multiple quantum well structure will be remarkably non-uniform due to the occurrence of the same non-uniformity in both holes and electrons. I will explain.
FIG. 15 is an explanatory diagram of the optical gain distribution in the multiple quantum well of the MQW semiconductor laser using the above-mentioned nitride compound semiconductor, and in the first quantum well from the p-type cladding layer side, FIG. Since the supply of holes from the type cladding layer is large, it has a large optical gain. However, the optical gain decreases toward the n-type cladding layer side, and the two quantum wells on the n-type cladding layer side are optical. This is different from the conventional laser using a zincblende crystal structure semiconductor that not only generates gain but also causes light loss.

  That is, in conventional semiconductor lasers using a zincblende crystal structure semiconductor, it is common knowledge that carriers are uniformly injected when the MQW structure is composed of about five quantum well layers. The long wavelength laser for optical communication uses about 5 to 10 layers, and the red laser for DVD uses about 5 layers, but the optical gain is uniformly generated.

  The generation of the light absorption layer due to the non-uniformity between the quantum well layers having such an optical gain has two adverse effects. First, even in the two quantum well layers on the n side serving as the light absorption layer, As apparent from FIGS. 13 and 14, since the carriers exist at high density, it means that the recombination current is large, and the three quantum well layers on the p side reach the threshold Fermi level of laser oscillation. As a result, the two quantum well layers on the n side increase the amount of current.

Second, since the two quantum well layers on the n side are light absorption layers, the internal loss to be overcome for laser oscillation increases, and the threshold Fermi level E ^F _th itself increases. This will cause an adverse effect.

FIG. 16 is a diagram showing optical output-current characteristics of an MQW semiconductor laser actually manufactured by changing the number of quantum well layers in the multiple quantum well active layer. In this case, the entire optical confinement is constant. However, the MQW semiconductor laser provided with the active layer consisting of the five quantum well layers is different from the MQW semiconductor laser provided with the active layer consisting of the three quantum well layers. It can be seen that the threshold current density _Jth is high and the efficiency after laser oscillation is also poor.

  This is because the efficiency of the semiconductor laser is determined by the internal quantum efficiency and the internal loss, and both the internal quantum efficiency and the internal loss are degraded by the fact that the two quantum well layers on the n side are light absorption layers. This is considered to be the cause.

  Further, in the conventional MQW structure semiconductor laser, the intensity distribution (radiation intensity distribution) of the radiated photoelectric magnetic field has a symmetric structure so that the maximum intensity position comes to the center position of the active layer as shown in FIG. There is a problem that optical confinement is not effectively performed because the quantum well of the first layer that generates optical gain and the maximum intensity position do not match.

  That is, the substantial gain that contributes to laser oscillation is the optical gain multiplied by the radiated light intensity distribution. Even if the optical gain is large, the radiated light intensity distribution must be present at the position where the optical gain is generated. It cannot contribute to oscillation.

  Actually, the distribution of the radiated light to the active layer is about 3% of the total light intensity even if all the layers are summed, so that even a small light distribution is maximum as shown in FIG. It is a big problem that the layer that generates the optical gain does not become large.

In addition, in the case of MQW lasers with 10 to 20 quantum well layers, which are currently reported, even if the entire quantum well structure has sufficient optical confinement, an optical gain is substantially generated. There is a problem that light confinement in the first quantum well from the p-type cladding layer side becomes considerably small and the threshold current density _Jth increases.

Furthermore, for the reason of d described above, about 70% of the influence of the forbidden band width difference ΔE _g appears on the valence band side, the energy discontinuity ΔE _C on the conduction band side becomes small, and electron overflow becomes a problem. Therefore, in the conventional short wavelength semiconductor laser, an overflow prevention layer or a carrier stopper layer is provided, but this results in an asymmetric structure in which the emitted light intensity distribution is further shifted to the n side, and a p-type cladding layer having a large optical gain. There is a problem that light confinement in the first quantum well from the side is further reduced and the threshold current density _Jth is increased.

Therefore, in order to improve the characteristics of an MQW semiconductor laser using a nitride compound semiconductor, it is necessary to reduce the threshold current density _Jth . For this purpose, the above-described non-uniform carrier injection is performed. It is effective to improve.

  However, in the case of MQW semiconductor lasers using nitride compound semiconductors as described above, there are problems peculiar to nitride compound semiconductors, and simply applying the common sense of conventional zincblende crystal structure semiconductors. An excellent solution cannot be found, and it is necessary to examine whether the nitride compound semiconductor has a non-essential configuration in the MQW semiconductor laser using the nitride compound semiconductor. become.

Therefore, in general, in order to improve the hole injection efficiency, it is conceivable to increase the hole concentration by making the p-side light guide layer a p-type layer. In this case, however, There exists a problem that injection | pouring efficiency falls.
That is, the main cause of the low hole injection efficiency is that the mobility of holes in the p-side light guide layer is small, but the scattering increases due to p-type doping, which further reduces the mobility of holes.
Furthermore, even if the doping concentration of the p-type impurity is increased, since the activation rate of the impurity is small, the hole concentration does not increase easily.

  In general, in order to improve the non-uniform injection of carriers in the MQW semiconductor laser, the thickness of the well layer in the multiple quantum well structure, the thickness of the barrier layer, and the barrier layer It is considered effective to reduce the height, that is, to reduce the forbidden bandwidth.

  Among these, the film thickness of the well layer greatly affects the optical gain characteristics of the laser and is the most important item in laser design, so there is a problem that it is difficult to change it independently, and the nitride MQW blue semiconductor In the laser, reducing the forbidden band width of the barrier layer means increasing the In composition in the barrier layer, which leads to an increase in strain and deteriorates the crystallinity. Since this is not preferable, this situation will be described with reference to FIG.

FIG. 17 is a diagram showing experimental results on the dependence of the luminous efficiency on the In composition ratio x when In _x Ga _1-x N used as the barrier layer is used. Since the strain applied to the active layer is increased and the light emission efficiency is decreased, the In composition ratio x of the barrier layer cannot be increased. Therefore, the barrier of the quantum well structure cannot be decreased to increase the injection efficiency.

  On the other hand, with regard to the film thickness of the barrier layer, in the conventional laser using a zinc blende type crystal structure semiconductor, if the film thickness is thin, the interaction due to the oozing of the wave function between the quantum well layers cannot be ignored, and the stepped shape The optical gain distribution that should have been reduced and the generation of the optical gain per fixed carrier density is reduced, so the thickness is set to 5 nm or more. This structure is also applied to an MQW semiconductor laser using a nitride compound semiconductor as it is. Adopted.

  However, in the MQW semiconductor laser using a nitride compound semiconductor, since the effective mass of the carrier is large as described above, the wave function oozes out from the quantum well and the band gap is not uniform. For this reason, the stepwise optical gain distribution has been slightly reduced from the beginning, and therefore, it has been concluded that a film thickness of 5 nm or more is not an essential requirement.

  In addition, as described above, electrons that are not effectively consumed in the active layer overflow to the p-side and become an overflow current to the p-side light guide layer or the p-type cladding layer, and holes from the p-side light guide layer to the active layer. Due to the poor injection efficiency, the holes accumulated in the p-side light guide layer attract electrons to the p-side layer, thereby increasing the overflow.

Furthermore, because of the above-mentioned d, since the band offset at the heterojunction interface is 3: 7, the band is biased toward the valence band side, and the energy discontinuity ΔE _C on the conduction band side becomes small. has become a problem, in the semiconductor laser of conventional short wavelength, is provided with the overflow preventing layer or the carrier stopper layer, nevertheless, there is a problem that the overflow easily occurs because the threshold carrier density N _th is high, Thus, in a nitride-based semiconductor, an overflow of electrons becomes an essential problem as compared with other semiconductors.

That is, the threshold current density J _th of the semiconductor laser is as follows: τ _s is the electron lifetime, d is the thickness of the active layer, e is the elementary charge, and N _th is the threshold carrier density.
J _th = N _th · d · e / τ _s
The threshold Fermi level E ^F _th is a value at the threshold carrier density N _th of the Fermi level E _F depending on the carrier density N, that is,
E ^F _th = E _F (N _th )
It is represented by

The threshold carrier density N _th is a carrier density at which G _m (modal gain), which is a function of the carrier density N, exceeds the cavity loss to start laser oscillation, and the threshold carrier density N _th In order to reduce the value, it is necessary to increase G _m .

This G _m has Γ as an optical confinement factor, and G as a gain determined by the composition of the active layer, carrier density, etc.
G _m = Γ · G
Therefore, if the thickness of the active layer is thin and the optical confinement is insufficient, G _m becomes small, and accordingly, the threshold carrier density N _th also becomes large. The Fermi level E ^F _th is likely to rise.
If this threshold Fermi level E ^F _th is also increased, the number of electrons in a high energy state is increased, so that the overflow of electrons becomes a problem also from this point.

In addition, the present inventors have found through simulation that the leakage current due to the overflow of electrons to the p-type cladding layer is very large when the overflow prevention layer is not provided mainly due to the large effective mass. The situation will be described with reference to FIG.
FIG. 18 shows a simulation of the dependence of the overflow current of electrons on the total current amount in an MQW structure short wavelength semiconductor laser in which five well layers of In _0.15 Ga _0.85 N are sandwiched between In _0.05 Ga _0.95 N barrier layers. The results are shown by changing the Al composition of the AlGaN cladding layer. When the Al composition ratio of the cladding layer is 0.05, that is, when an Al _0.05 Ga _0.95 N layer is used, the low current region is shown. Thus, it can be seen that the leakage current starts to increase, and at 20 kA / cm ² , more than half of the total current leaks.

Such a leakage current is a reactive current that does not contribute to laser oscillation, which itself leads to an increase in the threshold current density _Jth and also causes heat generation, which further makes laser oscillation more difficult. .

When an Al _0.05 Ga _0.95 N layer having an Al composition ratio of 0.05 is used as the cladding layer, the difference in the forbidden band width from the active layer is 500 meV, which is a sufficient difference in the conventional material system. In the nitride system, since the overflow current is not negligible, an Al _0.15 Ga _0.85 N layer or the like having a large Al composition ratio in the cladding layer is used to reduce the leakage current due to overflow. .

The present inventor has found through simulation that the overflow current also depends on the element temperature. This situation will be described with reference to FIG.
FIG. 19 shows the simulation result of the total current amount dependence of the electron overflow current when the Al _0.1 Ga _0.9 N clad layer is used, with the element temperature changed, which is apparent from the figure. Thus, it can be seen that the overflow increases in the high current region as the element temperature rises.

  In an actual device, the overflowed current reaches the p-side electrode and heat is generated to raise the device temperature. This rise in device temperature creates a vicious circle in which the overflow current further increases, impeding laser oscillation. It is thought to do.

On the other hand, when an overflow prevention layer such as a p-type Al _0.15 Ga _0.85 N layer or an Al _0.18 Ga _0.82 N layer is provided to prevent an overflow of electrons, this Al _0.15 Ga _0.85 N layer or Al _0.18 Ga _0.82 N layer is provided. Has a lower refractive index than that of the p-side light guide layer, so that there is a problem that the optical confinement necessary for obtaining laser oscillation is greatly reduced, and an energy spike is formed at the heterojunction interface to prevent hole injection. There is a problem that becomes a barrier.

Further, as shown in FIG. 10 described above, in a conventional MQW semiconductor laser using a nitride compound semiconductor, a p-type having a large forbidden band between the MQW active layer 836 and the p-type GaN light guide layer 838. In the case where the Al _0.18 Ga _0.82 N overflow prevention layer 837 is provided, there is a problem that the increase of internal loss is increased and the potential barrier against holes is increased to increase the driving voltage, and also due to the difference in electron affinity. Since there is a problem that the drive voltage rises due to the potential barrier, this situation will be described with reference to FIG.

See FIG. 20A. FIG. 20A is a band diagram of a conventional short wavelength semiconductor laser. As is clear from FIG. 20, the forbidden band width of the p-type Al _0.18 Ga _0.82 N overflow prevention layer 837 is large. A barrier formed between the p-type GaN light guide layer 838, that is, between the band edge of the valence band when no voltage is applied as indicated by a solid line and the band edge 844 of the valence band when a voltage is applied as indicated by a broken line. Therefore, the applied voltage V for injecting holes into the MQW active layer 836 is increased.

See FIG. 20B. FIG. 20B is a diagram schematically showing a band diagram on the valence band side in the vicinity of the p-type Al _0.18 Ga _0.82 N overflow prevention layer 837. As shown in FIG. Due to the difference in electron affinity at the interface between the active layer 836 and the p-type Al _0.18 Ga _0.82 N overflow prevention layer 837 and at the interface between the p-type Al _0.18 Ga _0.82 N overflow prevention layer 837 and the p-type GaN light guide layer 838 Thus, notches 845 and 846 are formed, and the notches 845 and 846 serve as a potential barrier against hole injection, and the hole injection efficiency is lowered.

  In addition, other factors that promote this overflow include a high specific resistance of the p-type cladding layer and a short non-emission lifetime in the p-type cladding layer. These factors are not essential. Therefore, it is conceivable to reduce overflow by improving them.

  However, a p-type cladding layer having sufficient crystal quality has not been obtained at present. For example, regarding the specific resistance, doping to the p-type cladding layer is difficult and insufficient, but the Al composition ratio is low. As doping increases, doping becomes more difficult. Therefore, when a p-type cladding layer having a large Al composition ratio is used to reduce overflow, it is not easy to reduce the specific resistance.

  Further, regarding the non-emission lifetime, even in an undoped crystal, the non-emission lifetime is about 1 ns (nanosecond), which is shorter than that of the conventional material. There is a problem.

That is, the crystal quality characteristics of the nitride-based compound semiconductor include a very high dislocation density. In particular, when a sapphire substrate is used as the growth substrate, it does not sufficiently lattice match with the growth layer. The crystallinity of the growth layer is poor. For example, it has been reported that the normal dislocation density reaches 10 ¹⁰ cm ⁻² , and there is a problem that the non-emission lifetime is as fast as about 1 ns (nanosecond).

In addition, the p-type impurity doping deteriorates the crystallinity of the p-side light guide layer, the non-emission lifetime is further shortened to about 0.1 ns (= 100 ps), and the amount of non-emission recombination increases. As a result, the threshold current density J _th for laser oscillation further increases.

  For this reason, in a nitride-based semiconductor where it is not easy to increase the hole concentration even if Mg is doped, the effect of increasing the hole concentration of the p-side light guide layer is due to a decrease in mobility associated with p-type doping, It is considered that the deterioration of crystallinity becomes a problem.

On the other hand, regarding the dislocation density, the value of 10 ¹⁰ cm ⁻² is about 1 million times that of the conventional zinc blende type crystal structure semiconductor whose dislocation density is 10 ⁴ cm ⁻² or less. In compound semiconductors, it is said that dislocations do not form non-luminescent centers and thus do not affect device characteristics.Therefore, it is not necessary to reduce the dislocation density in order to reduce non-luminescent centers. Semiconductor lasers have been realized in the presence of a high density of dislocation density.

InGaN, which is usually used as an active layer of a nitride semiconductor laser, has properties that are completely different from those of conventional materials, in addition to the physical characteristics common to the nitride compound semiconductor.
That is, InGaN is a mixed crystal of InN and GaN, while the forbidden band width of InN is 1.9 eV, whereas the forbidden band width of GaN is very different from 3.4 eV. Also, while InN is around 600 ° C., GaN is significantly different from around 1000 ° C.

  For these reasons, it is known that this InGaN mixed crystal is very difficult to mix, and even in a region where the non-mixed crystal region is large and the In composition ratio is as small as 0.2 or less, the In composition ratio is low. There is a problem that the proportion of composition separation increases with the increase.

  As a result, in the InGaN layer with an In composition ratio of about 0.15, the half-value width of the photoluminescence (PL) spectrum seen in a macro region of about 200 μm is very large, reflecting in-crystal nonuniformity due to composition separation. Even a good crystal is 150 meV.

  This is because the half-width of the PL spectrum, which should increase when the semiconductor is thinned with a conventional zinc blende crystal structure, becomes smaller with an InGaN mixed crystal. This phenomenon is reduced even when cooled to a very low temperature. This is due to a property that is completely different from the conventional material, that is, there is almost no change in the value range.

  The characteristics of these InGaN do not depend on the type of substrate used as a growth substrate such as a sapphire substrate, SiC substrate, or spinel substrate, and also include a reduced pressure MOVPE method (a reduced pressure metal organic vapor phase growth method), a normal pressure MOVPE method. Or, it does not depend on the growth method such as MBE method (molecular beam epitaxial growth method), and also appears without depending on the crystal structure such as hexagonal crystal or cubic crystal. Features are considered to be inevitable specialities that should be tolerated to some extent.

  In this way, conventional short-wavelength semiconductor lasers and light-emitting diodes (LEDs) that use InGaN as an active layer have been developed with such in-crystal composition non-uniformity intact. It is thought that commercialization is carried out with a large compositional non-uniformity.

  Even in short-wavelength semiconductor lasers, although laser oscillation has been successful as described above, there has been no appropriate evaluation means so far. It is not known whether the compositional nonuniformity is occurring, and it is not known how the compositional nonuniformity affects the device characteristics. Therefore, how much crystal quality is necessary for laser oscillation. It was completely unknown whether the crystal of such quality can be obtained with good reproducibility when grown.

  Therefore, the present inventors recently performed photoluminescence (PL) measurement with a small spot diameter of 1 μm on an InGaN mixed crystal serving as an active layer of a short-wavelength semiconductor laser, thereby achieving non-uniform composition in the InGaN mixed crystal. A method to quantitatively evaluate the degree was developed.

As a result of such an evaluation, it has been found that the PL peak wavelength of the InGaN MQW (multiple quantum well) active layer has a very large distribution in the crystal, which will be described with reference to FIGS.
The details of the conventional short-wavelength semiconductor laser that performed this measurement have not been published.

FIG. 21A shows the PL peak wavelength and the PL light at each measurement point when the PL spectrum was measured at 2500 points every 2 μm in the region of 10,000 μm ^{2 in} the element that did not oscillate. It shows the correlation of intensity, and both PL peak wavelength and PL light intensity are widely distributed over a range of 396 nm (≈3.11 eV) to 416 nm (≈2.980 eV), and the PL peak wavelength distribution range is 151 meV, That is, it was about 150 meV.

  Incidentally, as a result of the same evaluation of the InGaAs-based active layer used as a semiconductor laser for optical communication, the PL peak wavelength distribution is about 5 meV, that is, about 1/30 of the InGaN mixed crystal. From this result, it was found that the InGaN system is a very special material system, and conventional common sense is not applicable.

In this InGaN-based semiconductor laser, the threshold current density J _th is essentially high for reasons of physical properties, and the nonuniform composition of the active layer is fatal in achieving laser oscillation. This is a drawback. From the above evaluation, it is obtained that the laser oscillation does not occur when the PL peak wavelength distribution is 150 meV or more.

Refer to FIG. 21B. FIG. 21B shows the PL peak wavelength and the PL light intensity at each measurement point when the PL spectrum is measured at 2500 points every 2 μm in the region of 10000 μm ² in the laser-oscillated device. The PL peak wavelength distribution range is 91 meV, that is, about 90 meV over the range from 400 nm (≈3.100 eV) to 412 nm (≈3.009 eV), and the PL light intensity distribution is also It was a small one.

See FIG. 22A. FIG. 22A is a diagram showing current-light output characteristics (IL characteristics) of an InGaN-based semiconductor laser having the PL peak wavelength distribution as described above, and has a wavelength of 414.3 nm. At room temperature, pulse oscillation was achieved, but kink clearly appears as the current is increased.
Note that the PL wavelength and the laser oscillation wavelength are slightly different from each other.

See FIG. 22B. FIG. 22B shows the same InGaN-based semiconductor laser as in FIG. 22A, 1.1 times, 1.2 times, or 1.3 times the threshold current density J _th. The multi-wavelength oscillation occurs as the current increases, and this multi-wavelength oscillation causes a kink in the IL characteristic in FIG. 22 (a). I understand that.

Conventionally, in InGaN-based semiconductor lasers, it is known that the oscillation wavelength is multi-wavelength or changes greatly depending on the injection current, which reflects the formation of quantum dots (quantum boxes) in the active layer. It was thought that.
That is, in the past, it has been considered that multiwavelength oscillation is caused by the quantum effect (see Japan Journal of Applied Physics, vol. 35, 1996, p. 217, if necessary). It is reported that dots exist (if necessary, see Applied Physics Letters, vol. 70, 1997, p. 981).

  However, as a result of the inventor's research, the spatial distribution of the PL peak wavelength caused by the compositional non-uniformity in the active layer as described above has been considered to be the cause of multiwavelength oscillation. Will be described with reference to FIG.

23 is a histogram of the PL peak wavelength in the measurement results shown in FIG. 21 (b), while the line graph is the result of optical excitation of the semiconductor laser actually oscillated as shown in FIG. This shows the intensity distribution of the oscillation spectrum, and when they are overlapped, a very good agreement is obtained. From this, the multiwavelength oscillation is caused by the PL peak wavelength distribution, that is, the compositional non-uniformity in the active layer. It is believed that there is.

Then, such a multi-wavelength oscillation is only or kink in the I-L characteristic, because it leads to deterioration of the optical properties of the deterioration of the near field pattern and far field pattern to increase the threshold current density J _th, In an InGaN-based semiconductor laser as an optical device light source, it is important to suppress the composition distribution of the active layer.

  As described above, as a result of experiments by the present inventors, it is necessary to set the PL peak wavelength distribution of 150 meV or less, that is, within the distribution range of ± 0.03 in the In composition ratio for laser oscillation. It was also found that the PL peak wavelength distribution must be 90 meV or less, that is, the In composition ratio must be ± 0.018 or less in order to suppress multiwavelength oscillation. , Preferably 50 meV or less, more preferably 20 meV or less.

  Furthermore, since the short wavelength semiconductor light emitting device in FIG. 9B is essentially a light emitting diode, there are descriptions on the impurity concentration or the layer thickness, but under any conditions when a semiconductor laser is used. There is no suggestion about whether laser oscillation occurs at a low threshold current density with high efficiency.

  Accordingly, an object of the present invention is to reduce the threshold current density of a short wavelength semiconductor laser using a nitride compound semiconductor.

Here, means for solving the problem will be described. FIG. 1 is an explanatory view of the principle configuration of the present invention.
See FIG. 1. (1) The present invention is a semiconductor laser using a nitride compound semiconductor, and the Mg concentration of the electron block layer provided on the p side of the active layer is 7 × 10 ¹⁹ cm ⁻³ or more. It is characterized by that.

Thus, by setting the Mg concentration of the electron block layer provided on the p side of the active layer to 7 × 10 ¹⁹ cm ⁻³ or more, carrier overflow can be effectively suppressed from the evaluation of the emission spectrum.

  The reason for this is not necessarily clear, but deep impurity levels are formed at a high density on the valence band side of the electron block layer. Impurity conduction or hopping conduction or tunnel conduction through this impurity level. This is because holes are injected into the active layer due to the above, and the hole injection efficiency is improved.

  (2) Further, the present invention is characterized in that, in the method of manufacturing a semiconductor laser using a nitride compound semiconductor, the growth temperature of the electron block layer provided on the p side of the active layer is 600 ° C. to 900 ° C. .

  (3) Further, the present invention is characterized in that, in the above (2), the growth temperature of the electron block layer is the same as the growth temperature of the active layer.

  Thus, the emission temperature in the active layer is increased by setting the growth temperature of the electron block layer to 600 ° C. to 900 ° C., which is the same as the growth temperature of the active layer, lower than the conventional growth temperature of about 1100 ° C. In addition, light emission in the p-type light guide layer can be reduced.

(4) Further, the present invention is characterized in that, in the above (2) or (3), Mg is doped so that the Mg concentration of the electron block layer is 7 × 10 ¹⁹ cm ⁻³ or more.

Thus, by doping Mg so that the Mg concentration of the electron block layer provided on the p side of the active layer is 7 × 10 ¹⁹ cm ⁻³ or more, it is possible to effectively prevent carrier overflow from the evaluation of the emission spectrum. Can be suppressed.

According to the present invention, since the p-type electron block layer inserted into the semiconductor laser made of a nitride compound semiconductor is constituted by the high concentration layer having an Mg concentration of 7 × 10 ¹⁹ cm ⁻³ or more, the electrons overflow. Can be suppressed almost completely, thereby reducing the threshold current density _{Jth and} reducing the power consumption. As a light source for an optical information recording apparatus or the like, it contributes to increasing the density.

In the present invention, when an electron block layer provided on the p side of an active layer of a semiconductor laser using a nitride compound semiconductor is provided, the growth temperature is set to 600 ° C. to 900 ° C., and the Mg concentration is set to 7 × 10 ¹⁹ cm ^−3. That's it.

Here, referring to FIG. 2 to FIG. 8, the impurity concentration of the electron block layer is set to 7 × 10 ¹⁹ cm ⁻³ or more and the growth temperature of the electron block layer is set to 600 ° C. to 900 ° C. Embodiment 1 of the present invention will be described that prevents the overflow of the current and thereby reduces the threshold current density.
First, with reference to FIG. 2, the manufacturing process of the short wavelength semiconductor laser of Example 1 of this invention is demonstrated.
See FIG. 2. First, n-type SiC having a carrier concentration of 4 × 10 ¹⁸ cm ^{−3 made} of hexagonal 6H—SiC having a (0001) plane, that is, a c-plane as a principal plane, is grown in bulk by the modified Rayleigh method. On the substrate 711, the growth pressure is set to 70 to 760 Torr, for example, 100 Torr, and the growth temperature is set to 800 Torr by MOVPE using TMGa, TMAl, ammonia, SiH ₄ as a dopant source, and hydrogen as a carrier gas as a growth gas. ˜1200 ° C., for example, 1100 ° C., thickness 50 nm to 5 μm, eg 350 nm, impurity concentration 5 × 10 ¹⁷ to 1 × 10 ¹⁹ cm ⁻³ , eg 8 × 10 ¹⁸ cm ^-3 n-type Al _0.09 Ga _0.91 N buffer layer 712 is grown.

Subsequently, using TMAl, TMGa, ammonia, SiH ₄ as a dopant, and hydrogen as a carrier gas, the growth pressure is set to 70 to 760 Torr, for example, 100 Torr, and the growth temperature is set to 800 to 1200 ° C., for example, 1100 ° C. In this state, the thickness is 0.1 to 2.0 μm, for example, 0.55 μm, and the impurity concentration is 1.0 × 10 ^{17 to} 1.0 × 10 ²⁰ cm ⁻³ , for example, 2.0 × 10 ¹⁸ cm. ^-3 n-type Al _0.09 Ga _0.91 N cladding layer 713 is grown.

Subsequently, using TMGa, ammonia, SiH ₄ as a dopant, and hydrogen as a carrier gas, the growth pressure is set to 70 to 760 Torr, for example, 100 Torr, and the growth temperature is set to 800 to 1200 ° C., for example, 1100 ° C. Then, an n-type GaN light guide layer 714 having a thickness of 10 to 300 nm, for example, 100 nm, and an impurity concentration of 5 × 10 ¹⁷ to 1 × 10 ¹⁹ cm ⁻³ , for example, 2 × 10 ¹⁸ cm ⁻³ is grown.

Subsequently, using TMGa, TMIn, ammonia, and N ₂ as a carrier gas, the growth pressure is set to 70 to 760 Torr, for example, 100 Torr, and the growth temperature is set to 600 to 900 ° C., for example, 780 ° C. 2 to 10 layers, for example, 3 layers of undoped In _0.15 Ga _0.85 N well layer having a thickness of 3 to 10 nm, for example, 4 nm, separated by an undoped In _0.03 Ga _0.97 N barrier layer of 1 nm to 10 nm, for example, 5 nm An MQW active layer 715 is formed by growth.

Subsequently, using TMAl, TMGa, ammonia, biscyclopentadienylmagnesium, and N ₂ as a carrier gas, the growth pressure is set to 70 to 760 Torr, for example, 100 Torr, and the growth temperature is set to 600 to 900 ° C., for example, A p ⁺ type Al _0.18 Ga _0.82 N electron block having a thickness of 5 to 30 nm, for example, 20 nm, and an impurity concentration of 7 × 10 ¹⁹ cm ⁻³ or more, for example, 1 × 10 ²⁰ cm ⁻³ at 780 ° C. Layer 716 is grown.

Subsequently, using TMGa, ammonia, biscyclopentadienylmagnesium, and N ₂ as a carrier gas, the growth pressure is set to 70 to 760 Torr, for example, 100 Torr, and the growth temperature is set to 800 to 1200 ° C., for example, 1100 ° C. P-type having a thickness of 10 to 300 nm, for example, 100 nm, and an impurity concentration of 1.0 × 10 ^{17 to} 1.0 × 10 ²⁰ cm ⁻³ , for example, 1.5 × 10 ¹⁹ cm ⁻³ . A GaN light guide layer 717 is grown.

Subsequently, using TMAl, TMGa, ammonia, biscyclopentadienylmagnesium, and N ₂ as a carrier gas, the growth pressure is set to 70 to 760 Torr, for example, 100 Torr, and the growth temperature is set to 800 to 1200 ° C., for example, In a state of 1100 ° C., the thickness is 0.1 to 2.0 μm, for example, 0.55 μm, and the impurity concentration is 1.0 × 10 ^{17 to} 1.0 × 10 ²⁰ cm ⁻³ , for example, 1.5 × A p-type Al _0.09 Ga _0.91 N clad layer 718 of 10 ¹⁹ cm ⁻³ is grown.

Subsequently, using TMGa, ammonia, biscyclopentadienylmagnesium, and N ₂ as a carrier gas, the growth pressure is set to 70 to 760 Torr, for example, 100 Torr, and the growth temperature is set to 800 to 1200 ° C., for example, 1100 ° C. In this state, the thickness is 0.1 to 2.0 μm, for example, 0.1 μm, and the impurity concentration is 1.0 × 10 ^{17 to} 1.0 × 10 ²⁰ cm ⁻³ , for example, 1.5 × 10 ^19. A p-type GaN first contact layer 719 of cm ⁻³ is grown, and subsequently, the impurity concentration is 5.0 × 10 ¹⁹ to 5.0 × 10 ²⁰ cm ^{− under the} same conditions as the p-type GaN first contact layer 719. ^3. A p ⁺ -type GaN second contact layer 720 having a thickness of, for example, 1.5 × 10 ²⁰ cm ⁻³ and a thickness of 5 to 50 nm, for example, 20 nm is grown.
The growth rate in this case is 2 μm / hr for the n-type layers 712 to 714, 0.3 μm / hr for the MQW active layer 715, and about the p ⁺ -type Al _0.18 Ga _0.82 N electron blocking layer 716. Is 0.9 μm / hr, and the p-type layers 717 to 720 are 2.6 μm / hr.

Next, after polishing the back surface of the n-type SiC substrate 711 to reduce the total thickness to about 100 μm, the p ⁺ -type GaN second contact layer 720 to the p-type Al _0.09 Ga _0.91 N clad layer 718 are formed by dry etching. Then, for example, a stripe mesa having a width of 4 μm and a height of 0.5 μm is formed.

Next, an n-side electrode 722 made of Ni / Ti / Au is provided on the back surface of the n-type SiC substrate 711, and a stripe-shaped opening having a width of 2 μm, for example, is formed on the p ⁺ -type GaN second contact layer 720. An MQW semiconductor laser is completed by providing a p-side electrode 723 made of Ni / Ti / Au via the SiO ₂ film 721 and dividing the element so that the resonator length L becomes 700 μm.
The stripe direction is the <1-100> direction, and the cleavage plane is the (1-100) plane.

FIG. 3 shows the current-light output characteristics of the element having the smallest threshold current among the results of measuring the current-light output characteristics of the MQW semiconductor laser of Example 1 of the present invention. As is apparent from the figure, the minimum value of the threshold current was 380 mA.
Although not shown, the maximum value measured this time is 600 mA, the average is 500 mA, and the threshold current is much larger than the previous threshold currents of 650 to 1600 mA and the average value of 900 mA. Reduced.

In order to confirm the effect of using such a p ⁺ -type electron block layer, a surface-emitting LED was fabricated in the same crystal growth process as the above semiconductor laser, and the emission spectrum was measured. 4 and FIG.
4A and 4B. FIG. 4B is a schematic cross-sectional view of a surface-emitting LED manufactured for measurement, and FIG. 4A is a top view. ) Plane, that is, on the n-type SiC substrate 731 having a carrier concentration of 4 × 10 ¹⁸ cm ^{−3 made} of hexagonal 6H—SiC with the c-plane as the principal plane, the thickness is 0.35 μm and the impurity concentration is n-type Al _0.09 Ga _0.91 n buffer layer 732 of impurity concentration 8 × 10 ¹⁸ cm ^-3, a thickness of 0.55 .mu.m, the impurity concentration is 2.0 × 10 ¹⁸ cm ^-3 n-type Al _0.09 Ga _0.91 n Clad layer 733, n-type GaN light guide layer 734 with a thickness of 100 nm and impurity concentration of 2 × 10 ¹⁸ cm ⁻³ , 4 nm thick undoped separated by an undoped In _0.03 Ga _0.97 N barrier layer with a thickness of 5 nm In _0.15 Ga _0.85 MQW active layer and the N-well layer is grown three-layer 35, the thickness is 20nm of p-type Al _0.18 Ga _0.82 N electron blocking layer 736, a thickness of 100 nm, p-type GaN optical guide layer 737 of impurity concentration is 5.0 × 10 ¹⁹ cm ^-3, thickness 0 in .2μm, p-type Al _0.09 Ga _0.91 N cladding layer 738 of impurity concentration is 5.0 × 10 ¹⁹ cm ^-3, a thickness of 0.1 [mu] m, the impurity concentration is 5.0 × 10 ¹⁹ cm ^-3 p A p ⁺ -type GaN second contact layer 740 having a thickness of 20 nm and an impurity concentration of 1.5 × 10 ²⁰ cm ⁻³ is sequentially deposited.

Next, an n-side electrode 741 made of Ni / Ti / Au is provided on the back surface of the n-type SiC substrate 731 and an SiO ₂ film 742 having a substantially square opening on the surface of the p ⁺ -type GaN second contact layer 740. A p-side electrode is formed by providing a translucent electrode 743 made of Ni / Au with a bonding pad 744 made of Ni / Au around it.
Note that the p ⁺ -type GaN second contact layer 740 visible through the semitransparent electrode 743 has a 37 μm square dimension and an element dimension of 300 μm square.

In such a surface emitting LED, the Mg concentration of the p-type Al _0.18 Ga _0.82 N electron block layer 736 is changed in the range of 0 to 2 × 10 ²⁰ cm ⁻³ , and the p-type Al _0.18 Ga _0.82 N electron block is changed. Crystal growth was performed by setting the growth temperature of the layer 736 to two temperatures of 780 ° C. and 1100 ° C.

  Next, a 100 mA pulse current with a width of 100 μs and a frequency of 1 kHz is applied to the surface-emitting LED at room temperature in the forward direction, and the light emitted from the translucent electrode 743 is condensed by a lens, and then dispersed through a grating. The measurement was performed by detecting with a photomultiplier tube.

FIG. 5 shows a surface-emitting LED in which a conventional electron blocking layer has a Mg concentration of 5 × 10 ¹⁹ cm ⁻³ and a growth temperature of 1100 ° C. The measurement results were compared with a sample having a Mg concentration of 1 × 10 ²⁰ cm ⁻³ and a growth temperature of 780 ° C.
As is apparent from the figure, the emission intensity in the vicinity of 400 nm which is the emission center wavelength of the MQW active layer 735 made of InGaN is remarkably increased under the new conditions, which is about 10 times the intensity ratio.

  In addition, in the conventional condition LED, the p-type GaN light guide layer 737 has a light emission peaking at a wavelength of 363 nm. However, in the new condition LED, the light emission in this wavelength band is hardly seen. It is considered that the electrons overflowing into the p-type GaN light guide layer 737 have disappeared.

See FIG. 6A. FIG. 6A is a diagram showing the Mg concentration dependence of the emission intensity in the MQW active layer 735 made of InGaN. As is clear from the figure, the vicinity of 7 × 10 ¹⁹ cm ⁻³ is shown. The light emission intensity suddenly increases at the boundary, and increases by about three orders of magnitude compared with the conventional condition.
Further, in the case of the same Mg concentration, the emission intensity increases by about two orders of magnitude or more under the new condition where the growth temperature is 780 ° C.

Refer to FIG.
FIG. 6B is a diagram showing the Mg concentration dependency of the emission intensity in the p-type GaN light guide layer 737. As is clear from the figure, the emission intensity is around 5 × 10 ¹⁹ cm ^−3. Although it decreases, in the case of the new condition where the growth temperature is 780 ° C., light emission is hardly observed when the Mg concentration is 7 × 10 ¹⁹ cm ⁻³ or more.

From the measurement results of FIG. 5 and FIG. 6, the Mg concentration of the p ⁺ -type Al _0.18 Ga _0.82 N electron block layer 716 is more preferably 7 × 10 ¹⁹ cm ⁻³ or more, as in Example 1 of the present invention. The electron overflow can be prevented by setting it to 1 × 10 ²⁰ cm ⁻³ or more, and the recombination is efficiently performed in the MQW active layer 715, so that the threshold current density J _th can be reduced. Become.

In particular, when the crystal growth temperature of the p ⁺ -type Al _0.18 Ga _0.82 N electron block layer 716 is set to 600 ° C. to 900 ° C., for example, 780 ° C., in the same range as the MQW active layer 715, the effect becomes remarkable.
The lower limit of 600 ° C. is the lower limit of the temperature at which InGaN single crystal growth is possible, and the upper limit of 900 ° C. is the substantial upper limit of the InGaN growth temperature.

Thus, the reason why the overflow of electrons is almost completely suppressed by setting the Mg concentration of the p ⁺ -type Al _0.18 Ga _0.82 N electron block layer 716 to 7 × 10 ¹⁹ cm ⁻³ or more is unknown. Since 1.0 × 10 ²⁰ cm ⁻³ is a high impurity concentration higher than the concentration limit at which Mg is activated, it has a high density in the forbidden band on the valence band side of the p ⁺ -type Al _0.18 Ga _0.82 N electron blocking layer 716. Impurity levels are formed, and this impurity level helps to improve the hole injection efficiency. As a result, it is considered that the overflow of electrons is suppressed. This situation will be described with reference to FIG.

FIG. 7 is a band diagram in the vicinity of the MQW active layer 715. Holes in the p-type GaN light guide layer 717 are formed by the MQW active layer 715 by tunnel conduction through the impurity level 724 or impurity conduction through the impurity level 724. As a result, the efficiency of hole injection is improved. As a result, electrons that have been attracted by the electric field of the holes staying in the p-type GaN light guide layer 717 can no longer be attracted. The threshold current density _Jth can be lowered.

See FIG. 7 and FIG. 20A. When this MQW semiconductor laser is driven, the applied voltage V may be set so that the band edge of the valence band of the p-type GaN light guide layer 717 reaches the impurity level 724. As is apparent from the comparison between FIG. 7 and FIG. 20A, the drive voltage can be lowered as compared with the conventional MQW semiconductor laser.

Next, the reason why the Mg concentration in the p-type electron block layer of the conventional short wavelength semiconductor laser, that is, the p-type overflow prevention layer is 5 × 10 ¹⁹ cm ⁻³ or less will be examined with reference to FIG.
FIG. 8A is a diagram in which the p-type carrier concentration in the p-type GaN layer, that is, the Mg concentration dependence of the hole concentration is examined. When the Mg concentration is low, the p-type carrier concentration is Although it increases with the Mg concentration, it becomes maximum when the Mg concentration is about 5 × 10 ¹⁹ cm ⁻³ , and decreases when the Mg concentration is higher than that.

Such a dependency of the p-type carrier concentration in the p-type GaN layer on the Mg concentration also holds true for the p-type AlGaN layer. Therefore, even in the p ⁺ -type Al _0.18 Ga _0.82 N electron block layer 716, 5 × It is considered that the p-type carrier concentration decreases at 10 ¹⁹ cm ⁻³ as a boundary.

As described above, the phenomenon in which the carrier concentration is saturated or lowered at a dopant concentration higher than a certain dopant concentration is a phenomenon that is generally observed in other compound semiconductors. In a conventional short wavelength semiconductor laser, It is estimated that the value of 5 × 10 ¹⁹ cm ⁻³ is used as the Mg concentration of the p-type electron block layer due to the above-described circumstances.

Therefore, setting the Mg concentration of the p-type electron block layer to 7 × 10 ¹⁹ cm ⁻³ or more as in Example 1 of the present invention is not at all effective as a means for increasing the p-type carrier concentration. Since it is disadvantageous, it is a configuration that cannot be adopted without new knowledge by the present inventor that the overflow of electrons is suppressed by setting the Mg concentration to 7 × 10 ¹⁹ cm ⁻³ or more.

In addition, in a high doping concentration region where the carrier concentration is saturated, a phenomenon of crystallinity deterioration due to overdoping is generally observed, and the mobility of the carrier decreases as the crystallinity decreases. This is a technique in which excessive doping is not employed unless there are other special circumstances. From this point of view, it is unpredictable that the Mg concentration of the p-type electron block layer is 7 × 10 ¹⁹ cm ⁻³ or more. is there.

FIG. 8B is a diagram in which the growth temperature dependence of the p-type carrier concentration in the p-type GaN layer grown by the MOVPE method is examined. As the growth temperature increases, the p-type carrier concentration increases. To increase. That is, it can be seen that when the growth temperature is lowered, the p-type carrier concentration does not increase but decreases.

  The growth temperature dependence of the p-type carrier concentration in such a p-type GaN layer is also established for the p-type AlGaN layer. Therefore, the growth temperature of the p-type electron block layer is set to 600 ° C. as in the present invention. ˜900 ° C. is adopted without new knowledge by the present inventors that the overflow of electrons is suppressed by setting the temperature to 600 ° C. to 900 ° C., more preferably 730 to 830 ° C., for example, 780 ° C. It is a configuration that cannot be done.

  Further, when the growth temperature of the GaN layer or AlGaN layer is lowered to 900 ° C. or lower, the surface morphology is remarkably deteriorated. Therefore, the crystal growth temperature of the p-type GaN layer or p-type AlGaN layer is usually in the range of 900 ° C. to 1200 ° C. For example, a high temperature of 1100 ° C. is adopted, and from this point, it is unpredictable that the crystal growth temperature of the p-type electron block layer is 600 ° C. to 900 ° C.

As described above, the first embodiment of the present invention has been described. However, the present invention is not limited to the configuration described in the embodiment, and various modifications are possible. For example, a p-type electron block layer is formed of Al _0.18 Ga _0.82. Although it is composed of N, it is not limited to such a composition ratio, AlGaN with other composition ratios may be used, and Al _x Ga _y In ₁ depending on the composition of the active layer and the p-type cladding layer. _-xy N those may be varied within the scope of (0 <x <1,0 <y <1).

In the description of the first embodiment, the MQW active layer having three well layers is used as the active layer. However, the MQW active layer having another structure, for example, six layers of In _0.03 Ga having a thickness of 5 nm, is used. _An MQW active layer or the like in which five layers of In _0.15 Ga _0.85 N well layers with a thickness of 2.5 nm are alternately sandwiched between _0.97 N barrier layers may be used. Furthermore, an SQW active layer may be used. is there.

  In the description of the first embodiment, the p-side light guide layer is a p-type layer and the n-side light guide layer is an n-type layer, but at least one of them is an undoped layer. Is also good.

It is explanatory drawing of the fundamental structure of this invention. It is explanatory drawing of the MQW semiconductor laser of Example 1 of this invention. It is explanatory drawing of the current-light output characteristic of the MQW semiconductor laser of Example 1 of this invention. It is explanatory drawing of the surface emitting LED produced in order to confirm the effect in Example 1 of this invention. It is explanatory drawing of the structural condition dependence of the emission spectrum intensity | strength of surface emitting LED. It is explanatory drawing of Mg density | concentration dependence of the emitted light intensity in surface emitting type LED. 2 is a band diagram near the MQW active layer of the MQW semiconductor laser according to Example 1 of the present invention. It is explanatory drawing of the structural condition dependence of the p-type carrier density | concentration in a p-type GaN layer. It is explanatory drawing of the conventional short wavelength semiconductor light-emitting device. It is explanatory drawing of the conventional short wavelength semiconductor laser. It is a band diagram at the time of the oscillation of the conventional MQW semiconductor laser. It is explanatory drawing of the layer position dependence of the hole current in the conventional MQW structure short wavelength semiconductor laser. It is explanatory drawing of the hole density distribution at the time of the oscillation of the conventional MQW semiconductor laser. It is explanatory drawing of the electron density distribution at the time of the oscillation of the conventional MQW semiconductor laser. It is explanatory drawing of the optical gain and radiation intensity distribution in the conventional MQW short wavelength semiconductor laser. It is explanatory drawing of the optical output-current characteristic of the conventional MQW semiconductor laser. A In _x Ga _1-x N In composition ratio of the barrier layer x dependency of illustration of the light emission efficiency of the quantum well structure active layer. It is explanatory drawing of the clad layer composition dependence of an overflow current. It is explanatory drawing of the element temperature dependence of the overflow current. It is explanatory drawing of the band diagram of the conventional short wavelength semiconductor laser. It is explanatory drawing of PL peak wavelength distribution of the conventional short wavelength semiconductor laser. It is explanatory drawing of the optical output characteristic in the conventional short wavelength semiconductor laser. It is explanatory drawing of the correlation of the histogram of PL peak wavelength, and light intensity in the conventional short wavelength semiconductor laser.

Explanation of symbols

711 n-type SiC substrate 712 n-type Al _0.09 Ga _0.91 N buffer layer 713 n-type Al _0.09 Ga _0.91 N cladding layer 714 undoped GaN light guide layer 715 MQW active layer 716 p ⁺ type Al _0.18 Ga _0.82 N electron block layer 717 p-type GaN light guide layer 718 p-type Al _0.09 Ga _0.91 N clad layer 719 p-type GaN first contact layer 720 p ⁺ -type GaN second contact layer 721 SiO ₂ film 722 n-side electrode 723 p-side electrode 724 impurity level 731 n-type SiC substrate 732 n-type Al _0.09 Ga _0.91 N buffer layer 733 n-type Al _0.09 Ga _0.91 N cladding layer 734 n-type GaN light guide layer 735 MQW active layer 736 p-type Al _0.18 Ga _0.82 N electron block layer 737 p-type GaN light guide Layer 738 p-type Al _0.09 Ga _0.91 N clad Layer 739 p-type GaN first contact layer 740 p ⁺ -type GaN second contact layer 741 n-side electrode 742 SiO ₂ film 743 translucent electrode 744 bonding pad 811 sapphire substrate 812 GaN buffer layer 813 n-type GaN buffer layer 814 n-type In _0.1 Ga _0.9 N layer 815 n-type Al _0.15 Ga _0.85 N cladding layer 816 n-type GaN light guide layer 817 InGaN MQW active layer 818 p-type Al _0.2 Ga _0.8 N layer 819 p-type GaN light guide layer 820 p-type Al _0.15 Ga _0.85 N Cladding layer 821 p-type GaN contact layer 822 n-side electrode 823 p-side electrode 824 n-type GaN layer 825 In _0.15 Ga _0.85 N active layer 826 p-type GaN layer 831 sapphire substrate 832 GaN buffer layer 833 n-type GaN intermediate layer 834 n-type Al _0.09 Ga _0.91 N Clad layer 835 n-type GaN light guide layer 836 MQW active layer 837 p-type Al _0.18 Ga _0.82 N overflow prevention layer 838 p-type GaN light guide layer 839 p-type Al _0.09 Ga _0.91 N clad layer 840 p-type GaN contact layer 841 n side Electrode 842 SiO ₂ film 843 p-side electrode 844 Band edge of valence band when voltage is applied 845 notch 846 notch

Claims (4)

A semiconductor laser using a nitride compound semiconductor, wherein the electron block layer provided on the p side of the active layer has a Mg concentration of 7 × 10 ¹⁹ cm ⁻³ or more. A method for manufacturing a semiconductor laser using a nitride compound semiconductor, wherein the growth temperature of an electron block layer provided on the p side of the active layer is set to 600 ° C. to 900 ° C. 3. The method of manufacturing a semiconductor laser according to claim 2, wherein the growth temperature of the electron block layer is the same as the growth temperature of the active layer. 4. The method of manufacturing a semiconductor laser according to claim 2, wherein Mg is doped so that the Mg concentration of the electron block layer is 7 × 10 ¹⁹ cm ⁻³ or more.

Images