JPH11214789A - Gain-switched semiconductor laser - Google Patents

Abstract

PROBLEM TO BE SOLVED: To obtain a gain-switched semiconductor laser, which is directly driven by a signal current and used for converting electrical signals into ultra-short light pulses, wherein the semiconductor laser is used in an optical communication system in which light signals are transmitted through a long distance at a high speed. SOLUTION: A gain-switched semiconductor laser is a semiconductor layer where an asymmetrical double quantum well structure 1 composed of a narrow quantum well on a p-side electrode across a barrier wall and a wide quantum well on an n-side electrode is made to serve as an active region. In this case, the thickness of the quantum well and the composition of waveguide layers 2 which sandwich in the asymmetrical double quantum well structure 1 between them are properly selected, so as to enable only a lowest order sub-band to be present in the conduction band of the asymmetrical double quantum well structure 1. A composite structure where a multi-quantum well structure 7 in which a barrier layer or a barrier layer and a well layer are doped with an acceptor to be increased in confinement coefficient is arranged on a p-side electrode side and the asymmetrical double quantum well structure 1 is arranged on an n-side electrode coming into contact with the multi-quantum well structure 7 is made to serve as an active region.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a semiconductor laser used in an optical communication system capable of high-speed and long-distance transmission over an optical fiber, and more particularly to a semiconductor laser directly driven by a signal current to convert an electric signal to an ultrashort optical pulse signal. The present invention relates to a gain-switched semiconductor laser used for conversion into a laser.

[0002]

2. Description of the Related Art A gain switch method for directly driving a semiconductor laser by a current pulse has a simple structure, so that it can be easily manufactured and a highly reliable light source. In this method, at present, it is necessary to reduce the spectrum spread by a Fabry-Perot resonator and further amplify by an erbium-doped fiber amplifier, but the time width of 17 ps and 0.21 n
Transform-limited light pulses with a spectral spread of m have been obtained (Nakazawa, M., Suzuki, K.
and Kimura, Y., “Transform-limited pulse generati
on in the gigaherz region from a gain switched dis
tributed-feedbacklaser diode using spectral window
ing ”, Opt. Lett., vol. 15, no. 12, pp. 715-717, Jun. 199
0). However, in order to achieve a transmission rate of 100 Gbps or more, it is necessary to further reduce the pulse width and reduce the spectrum width by reducing the chirping.

The gain switch is a method of obtaining a light pulse shorter than a current pulse by driving a semiconductor laser with a pulse current. When the semiconductor laser is driven by a pulse having a high current density, the carrier density in the active region increases. Stimulated emission does not occur at the beginning of the injection because there are few photons in the active region. Therefore, the carrier density increases beyond the threshold. However, when a high population inversion is formed in this way, a photon is generated by natural light emission, which is coupled to the mode of the resonator. Next, stimulated emission is rapidly generated by the electric field, and a laser beam is obtained. Due to this stimulated emission, the carrier density decreases, becomes lower than the oscillation threshold, and emission stops. In this way, short light pulses are obtained.

[0004] Lau, the rate width of the light pulse by the model equations show that can be shortened by increasing the, product Aτ _p n _'i of the following parameters (Lau,
K., Y., “Gain Switching of Semiconductor Injection
Lasers ”, Appl. Phys. Lett., Vol. 52 (4), no. 25, pp. 257
-259, Jan. 1988.). Here, A is the differential gain coefficient, τ _p is the photon lifetime, and n ′ _i is the difference between the peak carrier density and its emission threshold. This result, shown by Lau, shows that A
This means that a short optical pulse can be obtained if a high n ′ _i is obtained by driving the laser at a high current density when the values of τ _p and τ _p are determined.

[0005]

Normally, the current density is limited by the current pulse generator used. In this case how will be able to increase the product Aτ _p n _'i by designing the semiconductor laser. n ′ _i is A or τ _p
Because it is not independent of the method that maximizes the product Aτ _p n _'i is not necessarily clear. For example, it becomes low threshold carrier density n _th Larger A, the density of carriers accumulated it becomes photons are easily generated at a low stage, to increase the n _'i Since stimulated emission is generated at an early stage There is a problem that can not be.

The present invention has been made in order to solve the above-mentioned problems, and relates to an improvement of an asymmetric double quantum well structure. 1. Disclose a structure that provides a more significant non-linear increase in gain; In addition, a high gain can be obtained at a low carrier density even when the confinement coefficient is considered.

[0007]

[Table 1]

[0008]

The gain-switched semiconductor laser according to the first aspect of the present invention has an asymmetric structure comprising a narrow quantum well on the p-electrode side and a wide quantum well on the n-electrode side with one barrier layer interposed therebetween. In a semiconductor laser with a double quantum well structure as an active region, the carrier density dependence of gain can be made highly nonlinear by allowing only the lowest order subband to exist in the conduction band of the asymmetric double quantum well. And a short light pulse can be generated.

In the gain-switched semiconductor laser according to the second configuration of the present invention, the barrier layer or the barrier layer and the well layer are doped with an acceptor in contact with the p-electrode-side quantum well of the asymmetric double quantum well structure. A composite structure in which a multiple quantum well structure is arranged, and a structure in which the emission peak energies of the multiple quantum well and the asymmetric double quantum well coincide with each other is used as an active region. Both carrier density dependence of the gain and a high confinement factor and thus a high mode gain can be obtained.

[0010]

DESCRIPTION OF THE PREFERRED EMBODIMENTS As a method of improving the conventional problems of the gain switch method, there is an asymmetric double quantum well structure according to the same applicant in the preceding stage of the present invention. In an asymmetric double quantum well, the active region is composed of a single quantum well layer on the p-electrode side and an n-electrode side with one barrier layer interposed therebetween, and the p-side quantum well layer is compared with the n-side quantum well layer. It is narrow. With this configuration,
In this semiconductor laser, a non-linear increase in gain is obtained in the process of injecting carriers, and a short optical pulse is obtained. Although the asymmetric double quantum well structure is manufactured by an epitaxial growth technique, the quality of the formed quantum well structure varies depending on the material system. Al _x Ga with GaAs as substrate
The allocation of a _{1 over x} As / GaAs heterojunction structure and InP to In 1 _{over x} Ga _x As _y P 1 conduction band energy discontinuity amount in heterojunction heterojunction structure by _{chromatography y} and the valence band to the substrate different. In the GaAs system, the band discontinuity in the conduction band is large, so that many electron levels are formed in the quantum well. On the other hand, in the InP system, since the amount of energy discontinuity in the conduction band is small, the number of levels formed in the conduction band tends to decrease. This difference also affects the mechanism of developing the nonlinear enhancement effect of the asymmetric double quantum well.

The gain-switched semiconductor laser according to the present invention has a semiconductor having an asymmetric double quantum well structure comprising a narrow quantum well on the p-electrode side and a wide quantum well on the n-electrode side with one barrier layer interposed therebetween. In the laser, the thickness of the quantum well and the composition of the waveguide layer sandwiching the asymmetric double quantum well are appropriately selected so that only the lowest order subband exists in the conduction band of the asymmetric double quantum well. By doing so, a nonlinear increase process of the gain becomes remarkable in the process of injecting carriers. The reason is described below.

[0012] To illustrate, an In _{1 over} as shown in Table _{1 x} Ga x As _y P 1 _{over y} based materials SCH (SEPAR
ate Definition Heterostr
Consider the asymmetric double quantum well structures (a) and (b). In Table 1, all the epitaxy layers have compositions that lattice-match with InP. In this case, the composition x of Ga is approximately as follows as the composition y of As. x
= 0.4526y / (1-0.031y). Also, As
The band gap energy for the composition y of
(EV) = obtained by the 1.35-0.72y + 0.12y ^2. The energy discontinuity at the heterojunction interface is distributed to the conduction band at a rate of 31% and the valence band at a rate of 69%. These relationships are shown in the following literature.
Agrawel, G, P. and Dutta, NK, “Long-Waveiength Se
miconductor Lasers ”, New York: VanNostrand Renhol
d, 1986. Hereinafter, it is shown that a high differential gain is obtained in the structure (a). For comparison, the structure (b) in which the difference in band gap energy between the quantum well layer and the waveguide layer is large and two levels are formed in the conduction band is also considered. These structures are chosen to lattice match the InP substrate. Here, special values such as film thickness and composition are selected,
The structurally important points are as follows. 1. Double quantum well structures of different widths. The narrow quantum well is on the p-side and the wide quantum well is on the n-side. 2. The two quantum wells are separated by a high barrier. Therefore, the movement of carriers between wells is dominated by the resonance tunnel. 3. In the structure (a) according to the present invention, since the band discontinuity between the waveguide layer and the well layer is small, only a single level is formed for electrons. In the structure (b) considered for comparison, since the difference in band gap between the quantum well and the waveguide layer is increased, two levels may exist in the conduction band.

[Relationship between carrier density and gain of gain-switched semiconductor laser according to the present invention] The carrier density and gain coefficient when a forward bias is quasi-statically applied to a semiconductor laser having a structure as shown in Table 1 are shown below. To calculate. At this time, the following assumptions are made. 1. It is assumed that a voltage corresponding to the difference between the built-in potential and the applied voltage is induced at both ends of the quantum well structure, and that the electric field therebetween is constant. 2. The quasi-Fermi level of the conduction band in the quantum well structure is n
Assume that it is consistent with the Fermi level of the type cladding region. It is assumed that the maximum carrier density can be obtained statically in the quantum well. This assumption is reasonable because we are considering on a time scale that is sufficiently slower than the time for carrier accumulation and faster than the spontaneous recombination rate. The quasi-Fermi level of the valence band is selected so that the neutral condition of charge is satisfied. Under the above assumption, the gain for the TE mode when the bias voltage Vb is applied is obtained by the following equation. This equation is derived by obtaining the transition probability between bands by the kp perturbation method.

[0014]

(Equation 1)

In the formula (1), D is a constant including only a universal constant and a constant specific to a substance, and does not depend on the structure. That is,

[0016]

(Equation 2)

## EQU1 ## Where e is the unit charge, c is the speed of light, ε ₀ is the dielectric constant of vacuum, n bar is the actual refractive index of the medium,
h bars obtained by dividing Planck's constant h by 2 [pi, E _G band gap energy of the quantum well layer, delta is the split-off band gap energy. Further, _mc is the effective mass of electrons in the conduction band. Further, equation (1)
, L _z is the thickness of the quantum well region, and E is the energy of light. v = hh and v = lh represent the contribution of heavy holes and light holes, respectively. Further, m _v ^r is the reduced mass of the effective mass m _v of holes in effective mass m _c and valence band electrons in the conduction band. That is,

[0018]

(Equation 3)

Here, ψ _ci (ψ _vj ) represents the wave function of the i (j) -th level in the conduction band (valence band). E _c , _i (E _v , _j ) represents the i (j) -th sub-energy in the conduction band (valence band). E ′ _G is the difference between the bottom of the conduction band and the top of the valence band in the quantum well structure,
That is, the actual band gap energy. The wave function and the level energy are obtained by solving the following Schrodinger equation for each band structure.

[0020]

(Equation 4)

Here, the potential V is determined by the structure and the applied voltage. Fermi distribution function f _l in the formula (1) used quasi-Fermi energy F _l,

[0022]

(Equation 5)

## EQU2 ## k is the Boltzmann constant and T is the absolute temperature. The carrier density n is obtained by the following equation.

[0024]

(Equation 6)

FIG. 1 and FIG. 2 show the results of calculating the peak value of the gain spectrum and the derivative of the gain (differential gain) as a function of the carrier density for the structures (a) and (b). In FIG. 1, the carrier density is 1.08 × 10 ¹⁸ c
The gain increases sharply as it increases from m ^-3 . The differential gain has reached 6.4 × 10 ⁻¹⁴ cm ² . This value is 2.5 × 10 ⁻¹⁶ c, which is the value of a typical long wavelength semiconductor laser.
One digit or more larger than m ² . In contrast, in FIG.
Carrier density from 0.95 × 10 ¹⁸ cm ^-3 to 1.7 × 1
It increases almost linearly in the range of 0 ¹⁸ cm ^-3 , and then decreases conversely. In this case, the differential gain is of the same order as in the case of a normal long wavelength semiconductor laser. As shown in FIG. 1, the characteristic in which the gain changes abruptly is a characteristic expected to obtain a short optical pulse. Since the value of the differential gain is large, it can be expected that the chirping becomes smaller than that of a normal laser when the gain switch is operated in the case of the structure (a). This can be predicted from the following well-known relationship: DFB (Distributed Feedbac)
k) The wavelength spread of the laser is represented by (1 + α) ^1/2
(Koyama, F. and Suematsu, Y., “Analysis of Dynamic
Spectral Width of Dynamic-Single-Mode (DSM) Lasers
and Related Transmission Bandwidth of Single-Mode
Fibers ”, IEEE J. Quantum Electron., Vol. QE-21, n
o.4, pp. 292-297, April 1985.). Where the so-called α
The parameter is a quantity proportional to (dn bar / dn) / (dg / dn) (Henry, CH, “Theory of the Linewi
dth of Semiconductor Lasers ”, IEEE J. Quantum Elec
tron., vol. QE-18, no.2, pp.259-264, Feb. 1982.).
Here, n is the carrier density and dg / dn corresponds to the differential gain.

The difference between the magnitudes of the gain changes in FIGS. 1 and 2 is that the structure (a) in Table 1 has only one level formed in the conduction band and the structure (b) has two levels in the case of structure (b). Is formed. To more clearly illustrate this point, consider structure (c). This structure is shown in Table 1 as shown in Table 2.
The other structure is the same as (b) except that the composition of the waveguide layer is different. In the structure (c), only one level is formed in the conduction band because the band interval of the waveguide layer is reduced.
Further, when the relationship between the carrier density and the gain is calculated for the structure (c) in the same manner as in the structures (a) and (c), the high differential gain is larger than that in the normal case by one order of magnitude as in the case of the structure (a). Is obtained.

[0027]

[Table 2]

The reason why the carrier density dependence of the gain differs between the structure (c) and the structure (b) can be intuitively understood from FIGS. FIG. 4 shows the case where one level is formed in the conduction band (structure (c)) and FIG. 3 shows the case where the bias voltage _Vb is changed in the case where two levels are formed in the conduction band (structure (b)). The change of the band edge of the conduction band and the valence band and the change of the wave function of the electron and the heavy hole are shown.
FIG. 4 shows a bias of 1.285 V, 1.290 V, 1.2 B for the structure (c) in which only the first level is formed in the conduction band.
This shows the change of the wave function of the first level of electrons and heavy holes when the voltage is changed to 95V. In this bias voltage range, the electron wave function hardly changes, whereas the heavy hole wave function changes significantly. Especially,
The peak position of the first-level wave function moves from a narrow quantum well to a wide quantum well. The gain is given by equation (1)
Since it depends on the overlap integral of the electron and the hole (Equation (2)), the overlap integral C _ij changes greatly and a sudden change in the gain occurs. On the other hand, FIG. 3 shows that the bias is 1.280 for the structure (b) in which two levels are formed in the conduction band.
The graph shows changes in the wave functions of the first level of electrons and heavy holes when V, 1.285 V, and 1.290 V are changed. In this case, the wave function of the heavy hole moves from a quantum well having a narrow peak position to a wide quantum well as in the structure (c) of FIG. On the other hand, the two levels of the electrons are coupled and spread in both wells, so the overlap integral C _ij ^v
The change in the gain due to the change in the magnitude is not so large as in the structure (c) of FIG.

The above discussion does not refer to higher levels of holes. This is because the higher level of the hole has a relatively small contribution to the gain. When the charge neutral condition is satisfied, the height of the quasi-Fermi level from the band edge is higher in the conduction band and lower in the valence band. This is because the state density of heavy holes is larger than that of electrons. Therefore, considering the case where the carrier density is such that the gain spectrum peak becomes positive, even in the conduction band, even higher-order levels contribute to the gain because they are lower than the pseudo-Fermi level, but the valence-band pseudo-Fermi level Is almost the same height as the lowest level, so that the contribution of the higher level of the hole to the gain is small.

The same applies to carriers having higher energy than the band edge of the confinement layer. Here, their contribution to gain is ignored. Regarding the actual movement of holes between wells, in the structure of FIG. 4C, the first level and the second level of a heavy hole are coupled near V _b = 1.29 V, and a resonance tunnel occurs. ing.
Therefore, it can be considered that holes propagate from a narrow quantum well to a wide quantum well with the period of the resonance tunnel.
The resonance tunnel period Δt = h / 2ΔE estimated from the energy difference ΔE between the first level and the second level is approximately 4 ps in this case. In a situation where resonance tunneling occurs between two states of a heavy hole, the energy density of the two levels is small, so that the density of states of the two states contributes to the height of the gain spectrum peak. This point is also a factor of a sharp increase in gain. In a GaAs-based material system, the band discontinuity is larger in the conduction band than in the valence band. In such a case, many levels may be formed also in the conduction band. In such a case, the non-linear gain increase is due exclusively to this mechanism.

Next, the meaning that the structure according to the present invention is asymmetric will be described. As described above, a sudden change in gain is caused by the combination of the lowest two levels of heavy holes. Considering the case where two levels of heavy holes are coupled in a symmetric quantum well structure, two or more levels in a conduction band will be coupled in a wide electric field range. Therefore, the gain increases almost linearly in a wide carrier density range as in the case of the structure (b). Actually, when the differential gain is calculated for a symmetric structure composed of three quantum wells having a well thickness of 7.5 nm, only a substantially similar structure to that of the structure (b) can be obtained.

[Increase in Gain Considering Confinement Coefficient] The asymmetric double quantum well structure has been described above. The gains mentioned above are for photons in the quantum well. Since the guided light in the actual laser is wider than the quantum well structure, the photons are not amplified outside the quantum well. The ratio of the gain region to the spatial distribution of the guided light intensity is called the confinement coefficient. If this is written as Γ, the net gain is Γg. Generally, the spread of guided light is much wider than the gain region, and Γ is on the order of several percent. Therefore, in order to increase the confinement coefficient Γ, a multiple quantum well structure may be provided and a layer having a gain may be added. The concept of the confinement coefficient described above is, for example, Agrawal, GP and D
utta, NK, “Long-Wavelength Semiconductor Laser
s ", New York: Van Nostrand Renhold, 1986., page 48.

Embodiment 1 FIG. 5 is a sectional view showing a configuration of the gain switch semiconductor laser according to the first embodiment. In the figure, reference numeral 1 denotes an active region in which carriers recombine to generate light. The active region has an asymmetric double quantum well structure. By selecting the thickness of the quantum well and the composition of the confinement layer, a single band is formed in the conduction band. Only places are formed. What is an asymmetric double quantum well?
It has a single quantum well layer on the p-electrode side and the n-electrode side with one barrier layer 1b interposed therebetween, and has a narrow quantum well 1c on the p-side and a wide quantum well 1a on the n-side. Reference numeral 2 denotes a waveguide layer for confining and guiding the generated light, reference numeral 3 denotes a cladding layer sandwiching the waveguide layer 2, and reference numeral 4 denotes a diffraction grating for feeding back the guided light, and constitutes a laser resonator. 5
Is a contact layer, and 6 is an electrode.

The active region 1 of the semiconductor laser, the waveguide layer 2,
Each layer such as the cladding layer 3 is produced by epitaxially growing a thin film crystal on a substrate. Using InP as a substrate, In _x
The respective layers by mixed crystal of Ga _{1 -x} As _y P _{1 over} mixed crystal or In the _{y x} Al _{1 over} x As. In the first embodiment, the composition and thickness of the active region are set as shown in Table 1 (a) so that a gain peak is obtained at a wavelength of 1.55 μm where the transmission loss of the optical fiber is the lowest. The current confinement and the lateral confinement structure of the guided light are provided by performing etching and burying growth.

Next, the gain switching operation of the semiconductor laser according to the present embodiment will be described. In the semiconductor laser according to the present embodiment, the active region has an asymmetric double quantum well structure. In this asymmetric double quantum well, only one level is formed in the conduction band. When such a semiconductor laser is operated as a gain switch, the gain is low at the stage of injecting carriers into the quantum well, and the two levels of holes are combined at the stage of sufficient accumulation of carriers, and thereafter the gain rapidly increases. . Therefore, an optical pulse shorter than the normal gain switch operation can be obtained. Further, in this structure, a high differential gain is obtained, so that chirping is small.

Embodiment 2 In the first embodiment, the case where the active region of the semiconductor laser has only the asymmetric double quantum well structure has been described. However, in order to increase the confinement coefficient, the multiple quantum well in which the acceptor is doped in the barrier layer or the barrier layer and the well layer is used. A high gain can be obtained with a low carrier density by using a composite structure in which the well structure is arranged on the p-electrode side and the asymmetric double quantum well structure is arranged on the n-electrode side in contact with the multiple quantum well structure, In addition, a non-linear increase in gain is obtained.

The structure of the second embodiment is more specifically shown in Table 3 and FIG. Table 1 shows the asymmetric double quantum well structure.
This is the same as the structure (a) shown in FIG. 1 except that a multiple quantum well is arranged in contact with the structure (a). The gain peaks of the asymmetric double quantum well and the multiple quantum well coincide with each other and are selected to be 1.55 μm. In this structure, the potential of the multiple quantum well region is almost constant even in the implantation process because the acceptor is doped. In the region of the asymmetric double quantum well, the gain changes as shown in FIG. 1 as the forward bias increases. Since electrons can tunnel through the thin barrier of the asymmetric double quantum well to reach the multiple quantum well region, the multiple quantum well reaches almost the same pseudo-Fermi level (conduction band) as the asymmetric double quantum well. Gain is obtained. As a result, the overlap between the guided light field and the gain region increases, and the gain in consideration of the confinement coefficient increases.

[0038]

[Table 3]

In the layered structure of the second embodiment shown in Table 3, the confinement coefficient is 12.5%. Further, in the case of the first embodiment, that is, in the structure in which the thickness of the waveguide layer is set to 80 nm to maximize the confinement coefficient and the asymmetric double quantum well is interposed between the cladding layers of 1.5 μm, the confinement coefficient is set to 3.
7%.

[0040]

According to the gain-switched semiconductor laser according to the first configuration of the present invention, an asymmetric two-layer structure comprising a narrow quantum well on the p-electrode side and a wide quantum well on the n-electrode side with one barrier layer interposed therebetween. To make the carrier density dependence of gain highly nonlinear by allowing only the lowest order subband to exist in the conduction band of an asymmetric double quantum well in a semiconductor laser with a quantum well structure as the active region. And a short light pulse can be generated.

According to the gain-switched semiconductor laser according to the second configuration of the present invention, the barrier layer or the barrier layer and the well layer are doped with the acceptor in contact with the p-electrode side quantum well of the asymmetric double quantum well structure. Since the active region is a composite structure in which the multiple well structure is arranged and the emission peak energies of the multiple quantum well and the asymmetric double quantum well coincide with each other, a carrier having a non-linear gain even at a low carrier density is used. Since both the density dependence and the high confinement factor, and thus the high mode gain, are obtained, a high gain can be obtained at a low carrier density, a nonlinear increase can be obtained, and a short optical pulse can be obtained.

[Brief description of the drawings]

FIG. 1 is a diagram showing the carrier density dependence of the gain coefficient (solid line) and differential gain (dotted line) of the quantum well structure (structure (a)) shown in Table 1.

FIG. 2 is a diagram showing the carrier density dependence of the gain coefficient (solid line) and differential gain (dotted line) of the quantum well structure (structure (b)) shown in Table 1.

FIG. 3 shows a case where two levels are formed in a conduction band (b)
FIG. 6 is a diagram showing band edges of a conduction band and a valence band, and changes in wave functions of electrons and heavy holes when the bias voltage _Vb is changed.

FIG. 4 shows a case where one level is formed in a conduction band (c).
FIG. 6 is a diagram showing band edges of a conduction band and a valence band, and changes in wave functions of electrons and heavy holes when the bias voltage _Vb is changed.

FIG. 5 is a cross-sectional view illustrating a configuration of the gain-switched semiconductor laser according to the first embodiment of the present invention.

FIG. 6 is a cross-sectional view showing a configuration of a gain-switched semiconductor laser having a composite structure of an asymmetric double quantum well and a multiple quantum well according to a second embodiment of the present invention as an active region.

[Explanation of symbols]

1 asymmetric double quantum well, 1a wide quantum well, 1b
Barrier layer, 1c narrow quantum well, 2 waveguide layers, 3 cladding layers, 4 diffraction grating of feedback mechanism, 5 contact layers, 6 electrodes, 7 multiple quantum wells.

Claims (2)

[Claims] 1. A semiconductor laser having an asymmetric double quantum well structure comprising a narrow quantum well on a p-electrode side and a wide quantum well on an n-electrode side with one barrier layer therebetween as an active region. A gain switch semiconductor characterized in that the carrier density dependence of gain has high nonlinearity by allowing only the lowest-order subband to exist in the conduction band of the well, and a short optical pulse can be generated. laser. 2. A composite structure in which a barrier layer or a multiple quantum well structure in which a barrier layer and a well layer are doped with an acceptor is disposed in contact with a quantum well on the p-electrode side of the asymmetric double quantum well structure. By making the structure in which the emission peak energies of the quantum well and the asymmetric double quantum well coincide with each other as the active region, the carrier density dependence of the nonlinear gain and the high confinement coefficient and therefore the high mode even at a low carrier density. 2. The gain-switched semiconductor laser according to claim 1, wherein both gains are obtained.