JPH06164069A - Semiconductor laser - Google Patents

Abstract

PURPOSE:To compensate for the decrease in the total gain of quantum well active layer by selecting at least one of the composition or the width of layer of quantum well layer at the side of p-type layer to have an effective band gap narrower than the effective band gap of at least one quantum well layer at the side of n-type layer. CONSTITUTION:The semiconductor laser is provided with the active layer of a multiple quantum well structure 4 with at least two layers of InGaAsP quantum well layers 8a-e and a p-type layer 5 and n-type layer 3 putting the active layer between. So as to compensate for the change of peak wavelength of the gain when holes are unevenly injected, at least one of the composition or the width of the layer of quantum well layer 8e at the side of p-type layer is selected to have an effective band gap narrower than at least one of the effective band gap of the quantum well layer 8a at the side of n-type layer. Thus, the potential well for holes becomes deeper than the potential for electrons, resulting in reducing the oscillation threshold value of an MQW laser in which the holes are injected unevenly by the forward energization.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a semiconductor laser,
In particular, InGaAsP having active layer of multiple quantum well structure
/ InP semiconductor laser. In addition, in this specification, "InGaAsP system" shall include InGaAs. The energy difference between the ground energy level in the conduction band and the ground energy level in the valence band in the quantum well layer is called the effective band gap.

Semiconductor lasers are small, lightweight, highly reliable,
It has been widely put into practical use as a key device of optoelectronics for communication and consumer use by taking advantage of its long life.

In particular, an InGaAsP / InP semiconductor laser provided with an InGaAsP active layer on an InP substrate covers an oscillation wavelength band of 1.2 to 1.5 μm, and therefore coincides with the minimum loss region of a silica fiber. , Is regarded as important as a light source for long-distance optical communication.

[0004]

2. Description of the Related Art The characteristics required of a semiconductor laser as a light source for optical communication are wide-ranging such as singleness of oscillation mode, optical output and stability at high speed modulation. In order to expand optical fiber communication from trunk line communication, which is currently being promoted worldwide, to branch systems such as subscriber systems and LAN systems, semiconductor lasers used as light sources are stable over a wide temperature range. It is required to have characteristics that enable low threshold oscillation.

A quantum well laser is attracting attention for the purpose of low threshold oscillation. The layer structure and the band structure of the quantum well active layer are shown in FIG. In addition to the spatial confinement of carriers and light by the conventional double hetero structure, the quantum well laser can increase the gain and lower the oscillation threshold by the energy confinement effect of the carriers. Usually, as shown in the figure, a multiple quantum well structure having a plurality of quantum well layers is often used for the active layer.

FIG. 4A shows a layer structure of a commonly used multiple quantum well laser. On the n-type substrate 21,
N-type cladding layer 22 using an epitaxial growth technique,
An n-type optical confinement layer 23, an i-type (undoped) multiple quantum well layer (active layer) 24, a p-type optical confinement layer 25, a p-type cladding layer 26 and a p-type contact layer 27 are deposited in layers.

The optical confinement layers 23 and 25 are made of SCH (Sepa
Rate Confinement Heterostructure) Also called the layer.
Each growth layer is usually lattice-matched to the n-type substrate 21,
Recently, a strained multiple quantum well structure, which intentionally introduces lattice mismatch, has also been proposed.

As shown in FIG. 4B, the i-type multiple quantum well layer 24 includes a plurality of quantum well layers, for example, three quantum well layers 28a, 28b and 28c and a barrier layer for separating the quantum well layers, For example, it includes two barrier layers 29a and 29b.

The energy band diagram during the current injection is shown in FIG.
It becomes like (C). In the figure, E _c is the energy at the band edge of the conduction band, ΔE _c is the difference between the E _c of the quantum well layer and the barrier layer, E _v is the energy at the band edge of the valence band, and ΔE _v is the quantum well layer. And the barrier layer Ev, and Eg represents the bandgap of the quantum well layer.

The width of the quantum well layer is set to be as thin as the de Broglie wavelength of electrons (usually several to several tens of liters). As a result, the possible levels of electrons are quantized, and recombination radiation is generated between the electrons and holes localized in the quantum levels.

The quantization in the one-dimensional (thickness) direction makes the energy state density stepwise. Therefore, the gain is increased, and oscillation can be performed at a low threshold. The oscillation energy becomes larger than the band gap Eg due to the quantization of the energy level.

[0012]

In order to sufficiently confine carriers in the quantum well layer even under a relatively high temperature environment, the potential well of the quantum well layer is used for electrons and holes. Must also be deep enough.

However, InGa which is a laser for optical communication is used.
In the case of an AsP / InP laser, the depth of the potential well for electrons (ΔE _{c in} FIG. 4B) is shallower than the depth of the potential well for holes (ΔE _v ). Therefore, if ΔE _c is secured with a sufficient size, ΔE _v becomes larger than necessary.

As a result, when carriers are injected into the quantum well layer by forward bias, more holes are distributed in the quantum well closer to the p-type layer, and the hole concentration has a different distribution between the well layers.

With reference to FIGS. 1A and 1B,
It looks like this: Now, in the ideal case in which holes are uniformly injected into each potential well of a quantum well active layer having four layers of potential wells, the gain g of a single quantum well layer is equal in each quantum well layer.

That is, when the potential wells for holes are W _h1 , W _h2 , W _h3 , and W _h4 (however, the numbers are counted from the p layer side), the gain g of each quantum well layer is
A diagram showing the relationship between the wavelength and the wavelength λ is given as in the center diagram of FIG. 1A, and is equal in each quantum well layer.

Therefore, the total gain G of the entire quantum well active layer is simply given by multiplying the gain g of the single quantum well layer by 4, and is given as shown on the right side of FIG. At the gain peak wavelength, the gain is multiplied by the number of layers.

However, when holes are injected into the InGaAsP / InP-based quantum well active layer in which the depth of the potential well for electrons is sufficiently deep, since the potential well for holes is deep, the holes are close to the p-layer side. It is distributed unevenly so that the concentration becomes higher. That is, the hole concentration of each potential well with respect to holes is expressed by p _i (i =
When 1-4), the _{p 1> p 2> p 3} > p 4.

As a result, as shown in FIG. 1B, the gain g in each quantum well layer shows different peak positions with respect to the wavelength, and the peak value g _max itself also differs for each well layer. When the wavelength position of the peak changes, the total gain G of the quantum well active layer, which is the sum of the contributions of the quantum well layers, becomes the peak value G.
_max is smaller than that in the case of FIG.

This leads to an increase in the threshold value of laser oscillation. In order to avoid the problem of non-uniform hole injection, a proposal is made to reduce the thickness of the barrier layer, tunnel-connect between adjacent quantum well layers, and make the hole concentration uniform by tunneling through the barrier layer. There is.

In this case, however, tunneling coupling also occurs between the potential wells of the electrons, and the broadening of the wave function causes a broadening of the electron levels, which reduces the effect of electron localization by the quantum well layer. . As a result, the threshold value increases.

An object of the present invention is to provide an InGaAsP / InP multi quantum well laser capable of compensating for the decrease in the total gain G of the quantum well active layer caused by the non-uniform injection of holes.

[0023]

A semiconductor laser of the present invention is a semiconductor having an active layer having a multiple quantum well structure having at least two InGaAsP-based quantum well layers, a p-type layer and an n-type layer sandwiching the active layer. In the laser, at least one of the composition and layer thickness of the quantum well layer on the p-type layer side is at least one on the n-type layer side so as to compensate the change in the peak wavelength of the gain due to the non-uniform injection of holes. The quantum well layer is characterized by having an effective bandgap narrower than that of the quantum well layer.

[0024]

[Function] For a quantum well layer having a narrow effective band gap,
Due to the non-uniform injection of holes, more holes are injected. Therefore, the recombination radiant energy between electrons and holes increases.

If the increased electron-hole recombination energy becomes equal to the electron-hole recombination radiation energy of the quantum well layer closer to the n-type layer, the peak wavelengths of the gains are aligned. In this way, the change in the peak wavelength of the gain due to the non-uniform injection of holes can be compensated by the change in the effective band gap.

The effective band gap can be changed by adjusting the composition or layer thickness of the quantum well layer.

[0027]

1 is a diagram for explaining the principle of a semiconductor laser of the present invention. Here, it is assumed that the multiple quantum well layer of the semiconductor laser has, for example, four wells. From the characteristics of the InGaAsP-based material, ΔE _c <ΔE _v . The potential well layers for holes are expressed as W _h1 , W _h2 , W _h3 , and W _{h4 in this} order from the p-layer side.

As already described with reference to FIG.
In such a multiple quantum well active layer, when a forward bias is applied, the hole concentration injected from the p layer side is the hole concentration of each potential well with respect to the hole concentration p ₁
When the _{~p 4, p 1> p 2} > p 3> p 4 and unevenly distributed.

In the conventional quantum well layer in which the depths and widths of the potential wells are equal, as a result, the gain g generated in each quantum well layer becomes nonuniform as shown in FIG. Get lower.

In the multi-quantum well laser shown in FIG. 1C, which has a quantum well layer in which the depth of the potential well is not uniform, the peak wavelengths of the gains g of the respective quantum wells can be made uniform.

That is, the InGaAsP / InP system semiconductor
In body lasers, ΔE_c<ΔE _vBut each quantum
Band gap Eg of well layer_(n)Is Eg₍₁₎<Eg₍₂₎<Eg₍₃₎<Eg_(Four) Has been selected.

Such selection is performed by selecting the mixed crystal ratio x, y (composition) of In _1-x Ga _x As _y P _1-y constituting each quantum well layer.
Can be done by changing the. Also,
The effective band gap can also be changed by changing the width of the quantum well layer.

Bandgap Eg of each quantum well layer
_{When (n)} is narrower toward the p-type layer side as shown in FIG. 1C, when uneven hole injection occurs, the difference in gain peak wavelength due to uneven hole concentration is Can be compensated.

[0034] As a result, although the difference of the peak value g _max gain g due to the difference in the hole concentration is present, g _max in the well layers
It is possible to align the wavelengths that give the wavelengths as shown in the center diagram of FIG. Therefore, the total gain G of the multiple quantum well active layer obtained by adding the gains g of the respective quantum well layers is shown in FIG.
It grows like the right edge of.

The magnitude of the total gain G is such that if the wavelengths giving the gain peaks g _max of the respective quantum well layers are completely matched, the uniform hole injection shown at the right end of FIG. Should almost match G in the case of.

When the thickness of each quantum well layer is changed in order to change the effective band gap, the energy state density is different for each well layer, so that the change in the energy distribution of carriers with respect to the change in ambient temperature is changed. Will change for each well layer.

On the other hand, when the composition is changed with the same layer thickness, the energy state density is almost the same in all the well layers, so that the peak wavelengths of the gains can be made uniform even if the temperature changes.

When a multi-quantum well structure is formed on the InP substrate with an InGaAsP-based material, the composition of each quantum well layer may be determined in consideration of the following conditions. In
_{The condition for 1-x} Ga _x As _y P _1-y to be lattice-matched to InP is 0.1894y−0.4184x + 0.0130xy =
Expressed as 0.

InGaAs that is lattice-matched to InP
The band gap energy (Eg) of P is experimentally determined as Eg _(y) = 1.35-0.72y + 0.12y ² . By using these two equations, x and y can be calculated for a desired Eg.

Eg and bandgap wavelength (λg)
Is expressed by λg = hc / Eg (h: blank constant, c: speed of light). For example, a composition that is lattice-matched to InP and has a bandgap wavelength λg = 1.2 μm is In _0.78 G
a _0.22 As _0.48 P _0.52 .

The composition of the quantum well layer is the above InP.
It is not limited to one that is lattice-matched to the substrate. In this case, biaxial strain is applied to the quantum well layer, and the effect of the strain is added to the difference in composition, and the band gap changes. For example, I
When tensile strain is introduced into the quantum well layer of n _1-x Ga _x As ternary mixed crystal, the quantum well layer having a higher tensile strain can increase the energy between the quantum levels of electrons and holes.

By introducing strain into the quantum well layer, a desired gain can be given with a smaller carrier concentration. Therefore, when the active layer having the strained multiple quantum well structure is used, the oscillation threshold can be further lowered.

FIG. 2 shows I according to one embodiment of the present invention.
It is sectional drawing which shows the principal part of the manufacturing process of a nGaAsP / InP multiple quantum well semiconductor laser. An InP substrate is used, and an active layer having a multiple quantum well structure is made of an InGaAsP-based semiconductor.

First, as shown in FIG. 2A, Sn or S-doped n-type I is formed by using a metal organic chemical vapor deposition (MOCVD) method or a gas source molecular beam epitaxy (MBE) method.
On the nP substrate 1, an S-doped n-type InP clad layer 2 having a thickness of 1 μm and an S-doped n-type In _0.85 G having a thickness of 0.17 μm
a _0.15 As _0.33 P _{0.67 An} optical confinement layer 3 is grown, and four layers of In _0.85 Ga _0.15 which are undoped and have a thickness of 10 nm are formed on the optical confinement layer 3.
An i-type InGaAsP multiple quantum well (MQW) active layer 4 is formed by alternately stacking an As _0.33 P _0.67 barrier layer and five undoped 7-nm-thick InGaAsP quantum well layers.

Further, a thickness of 0.17 μm is formed on the active layer 4.
Zn-doped p-type In _0.85 Ga _0.15 As _0.33 P _0.67 optical confinement layer 5 and Zn-doped p-type InP clad layer 6 having a thickness of 2 μm are successively epitaxially grown in this order.

Incidentally, the p-type InP clad layer 6 may be grown only partially here and the rest may be grown later. The composition of each InGaAsP mixed crystal layer is selected so as to be lattice-matched to InP.

5 of i-type InGaAsP MQW active layer 4
The quantum well layers 8a, 8b, 8c, 8d, and 8e of the layers are composed so that the potential well becomes deeper (the band gap becomes smaller) as it is closer to the p-type In _0.85 Ga _0.15 As _0.33 P _0.67 optical confinement layer 5. And an undoped In _0.85 Ga _0.15 As _0.33 P _0.67 barrier layer 9 having a thickness of 10 nm is selected.
It is arranged alternately with a, 9b, 9c and 9d.

In the composition of the quantum well layer, the central 8c is In
_0.54 Ga _0.46 As _0.98 P _0.02 (λg = 1.64 μm), which is 1.55 μ due to the transition between electron and hole quantum levels.
The composition is such that m light is emitted. As described with reference to FIG. 1C, the composition of the other quantum well layers compensates for the influence of the non-uniform injection of holes, and the gain g of each quantum well layer is adjusted.
Is selected so that the peak wavelength is 1.55 μm.

The barrier layer having a thickness of 10 nm is a thickness that can substantially suppress tunneling of carriers due to tunnel coupling between adjacent quantum well layers. Further, ΔE _c of the barrier layer and the quantum well layer is about 0.15 eV and ΔE _v is about 0.22 eV, which is a barrier having a sufficient height to confine carriers in the quantum well layer during driving.

Next, as shown in FIG. 2B, mesa etching is performed to form a stripe structure. First, p
The SiO ₂ film deposited on the surface of the InP clad layer 6 is patterned by photolithography and etching to form a stripe-shaped SiO ₂ mask 10. Next, using the SiO ₂ mask 10 as an etching mask, mesa etching is performed up to the n-type InP clad layer 2.

Next, as shown in FIG. 2C, buried growth on the side surface of the mesa is performed by liquid phase epitaxy (LPE) or the like. As shown in FIG. 2C, first, p-type InP (current)
The confinement layer 40 is grown, and then the n-type InP (current) confinement layer 41 is selectively grown.

The mesa etching of FIG. 2B can be performed by chemical etching, chemical etching and meltback, chemical etching and dry etching, or the like.
In addition, the embedded growth shown in FIG.
It is also possible to use another method such as the CVD method.

As shown in FIG. 2D, after removing the SiO ₂ mask 10, if necessary, the remaining portion of the p-type InP clad layer 6 is grown, and Zn-doped with a thickness of 0.5 μm is formed thereon. Of p-type In _0.72 Ga _0.28 As _0.6 P _0.4 contact layer 7 is grown.

N layer side, p on substrate and contact layer 7
A laser element is completed by forming the layer side electrodes 42 and 43. The n-layer electrode is, for example, an AuGe / Au electrode 42, and the p-layer electrode is, for example, Ti / Pt.
The vapor deposition heat treatment of each / Au electrode 43 should just be carried out. If the end face is cleaved as a resonator, it becomes a Fabry-Perot type laser.

When this laser device is mounted on a stem and is energized in the forward direction, the composition of the quantum well layer is adjusted so as to cancel out the shift of the peak wavelength of the gain due to the non-uniform concentration distribution of holes, and therefore the MQW activity. The overall gain G of the entire layer can be increased. As a result, the size is 1.55 μm as compared with the case where the MQW active layers having the same size and the same composition are used.
The laser oscillation threshold current can be reduced by about 30%.

In the embodiment shown in FIG. 2, the mixed crystal composition of the optical confinement layers 3 and 5 is the same as that of the barrier layer of MQW, and is a single composition. This is a so-called simple SCH (Separate C
onfinement Heterostructure).

However, considering the light confinement efficiency, the region close to the active layer has the highest refractive index, and the refractive index gradually decreases as it approaches the cladding layer, that is, a so-called GRIN-SCH (Graded Index). -SC
H) can also be used. Of course, a heterojunction having another structure may be used.

FIG. 3 is a sectional view showing the layer structure of a semiconductor laser according to another embodiment of the present invention. In this semiconductor laser, an active layer having a strained multiple quantum well structure is used.

On the Sn-doped n-type InP substrate 11, M
1 μm thick S-doped n-type In is formed by using the OCVD method or the like.
P-clad layer 12, 0.17 μm thick S-doped n-type
In _0.85Ga_0.15As_0.33P_0.67SCH layer 13, AND
Three InGaAs quantum well layers 18a, 18b,
18c and two undoped InGaAsP barrier layers 1
I-type InGaAsP / I in which 9a and 19b are alternately laminated
nGaAs multiple quantum well (MQW) active layer 14, thickness
0.17 μm Zn-doped p-type In_0.85Ga_{0. 15}As
_0.33P_0.67SCH layer 15, 2 μm thick Zn-doped p
-Type InP clad layer 16 and Zn layer having a thickness of 0.5 μm
P-type In_0.72Ga_0.28As_0.6P_{0. Four}contact
Layer 17 is deposited in this order.

I-type InGaAsP / InGaAsMQW
The quantum well layers 18a to 18c of the active layer 14 have different compositions and form a strained multiple quantum well structure. The i-type InGaAs quantum well layers 18a to 18c having a thickness of 12 nm are made of InP.
It has a composition with a smaller lattice constant and undergoes tensile strain.

The composition oscillates at 1.53 μm, and In _0.35 Ga _0.65 A is formed in the central quantum well layer 18b so as to compensate the shift of the gain peak wavelength due to the nonuniform injection of holes.
s, the n-layer side quantum well layer 18a has a Ga-rich composition, and the p-layer side quantum well layer 18c has an In-rich composition. The composition of the quantum well layers 18a and 18c is
The gain is adjusted according to the hole concentration ratio of the quantum well layer to maximize the gain.

The barrier layers 19a and 19b have a thickness 10 that can effectively prevent tunneling between adjacent quantum well layers.
nm i-type In _0.85 Ga _0.15 As _0.33 P _0.67 ,
It is lattice-matched to InP. The tensile strain amount ξ of each well layer is
ξ _18a > ξ _18b > ξ _18c . InP lattice matching I
The nGaAs composition is In _0.53 Ga _0.47 As, and the band gap Eg is Eg _18a > Eg _18b > Eg _18c .

In such a ternary mixed crystal, the lattice constant always changes when the composition is changed, and a strained multiple quantum well structure is formed. Since the tensile strain exhibits the same gain effect with a smaller carrier concentration, the threshold is further lowered in the tensile strain MQW active layer.

The layer structure shown in FIG. 3 can also be manufactured by a process similar to the process shown in FIG. When a forward current is applied to this semiconductor laser, the concentration of holes injected into each well layer is unevenly distributed as p _18a <p _18b <p _18c .

However, as described above, the amount of strain is larger on the n-layer side and the band gap is also larger on the n-layer side. That is, if the hole density is the same, a higher gain is obtained on the n-layer side, and the gain peak wavelength is on the shorter wavelength side.

Therefore, the influence of the non-uniform injection of holes is canceled and the gain peak wavelength of each quantum well layer is 1.53.
The gain g at the gain peak wavelength is increased at the same time when the gain is equal to μm. As a result, the total gain G becomes larger, and the threshold current density of laser oscillation further decreases.

In this embodiment, the case where the strain amounts of the InGaAs quantum well layers are different from each other has been described.
Strain may be given to the InGaAsP quantum well layer in the above-mentioned embodiment. In this case, it is possible to adjust the composition so that the applied strain amount is the same in each quantum well layer, or the composition may be adjusted differently.

Further, in the above-described two specific examples, the composition of each quantum well layer is changed so that the band gap becomes Eg _{(n + 1)} > Eg _(n) (where n = 1, 2, ... Although the relationship of the well layer sequence counted from the p layer side is shown), the effective band gap may be changed by changing the width of the well layer. Further, the effective band gap may be changed monotonically, and adjacent quantum well layers may have the same effective band gap.

The present invention has been described above with reference to the embodiments. In the InGaAsP / InP MQW active layer, at least the effective band gap of each quantum well layer monotonically increases from the p-layer side to the n-layer side. The composition may be changed so as to compensate the change in the peak wavelength of the gain due to the non-uniform injection of holes. Furthermore, the threshold can be further reduced by applying strain to the quantum well layer.

The invention is not limited to the embodiments described above. For example, it will be apparent to those skilled in the art that various modifications, improvements, combinations, and the like can be made.

[0071]

As described above, according to the present invention,
The potential well for holes is deeper than the potential for electrons, and it is possible to reduce the oscillation threshold of the InGaAsP / InP-based MQW laser in which holes are nonuniformly injected by forward conduction.

The oscillation threshold can be further reduced by adopting a strained quantum well structure.

[Brief description of drawings]

FIG. 1 is a principle explanatory diagram. FIG. 1A shows an ideal M
Diagram showing gain characteristics of active layer of QW laser, FIG. 1 (B)
FIG. 1 is a diagram showing the gain characteristics of an active layer of an actual InGaAsP / InP MQW laser, and FIG. 1C is a diagram showing the gain characteristics of an active layer having a non-uniform depth quantum well layer of the present invention. is there.

FIG. 2 is a InGaAsP / InP MQ according to one embodiment.
It is sectional drawing which shows the manufacturing process main part of W laser.

FIG. 3 shows an InGaAsP / InP-based M according to another embodiment.
It is a sectional view showing a layer structure section of a QW laser.

FIG. 4 is a diagram showing a configuration of a conventional quantum well laser.

[Explanation of symbols]

1 n-type InP substrate 2 n-type InP clad layer 3 n-type In _0.85 Ga _0.15 As _0.33 P _0.67 optical confinement layer 4 i-type InGaAsP multiple quantum well (MQW) active layer 5 p-type In _0.85 Ga _0.15 As _0.33 P _0.67 optical confinement Layer 6 p-type InP clad layer 7 p-type In _0.72 Ga _0.28 As _0.6 P _0.4 contact layer 8a, 8b, 8c, 8d, 8e quantum well layer 9a, 9b, 9c, 9d barrier layer 10 SiO ₂ mask 11 n-type InP substrate 12 n-type InP clad layer 13 n-type In _0.85 Ga _0.15 As _0.33 P _0.67 SCH layer 14 i-type InGaAsP / InGaAs multiple quantum well (MQW) active layer 15 p-type In _0.85 Ga _0.15 As _0.33 P _0.67 SCH layer 16 p-type InP cladding layer 17 p-type In _0.72 Ga _0.28 As _0.6 P _0.4 contact layer 18a, 18b, 18c i-type InGaAs Sub-well layer 19a, 19b i-type InGaAsP barrier layer 21 n-type substrate 22 n-type cladding layer 23 n-type optical confinement layer 24 i-type multiple quantum well layer (active layer) 25 p-type optical confinement layer 26 p-type cladding layer 27 p Type contact layer 28a, 28b, 28c, 28d quantum well layer 29a, 29b barrier layer 40 p-type InP confinement layer 41 n-type InP confinement layer 42 AuGe / Au electrode 43 Ti / Pt / Au electrode

Claims (3)

[Claims] 1. An active layer of a multiple quantum well structure (4, 14) having at least two InGaAsP-based quantum well layers (8, 18) and a p-type layer (5, 1) sandwiching the active layer.
5) and an n-type layer (3, 13), wherein a quantum well layer (8e, 18c) on the p-type layer side is provided so as to compensate for a change in peak wavelength of gain due to non-uniform injection of holes. At least one of the composition and the layer thickness is selected so as to have an effective bandgap narrower than that of at least one quantum well layer (8a, 18a) on the n-type layer side. And a semiconductor laser. 2. The semiconductor laser according to claim 1, wherein at least one quantum well layer (8a, 18a) on the n-type layer side has a strain. 3. The semiconductor laser according to claim 2, wherein the quantum well layer has a larger amount of strain as it approaches the n-type layer side.