JPH036677B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a three-terminal double heterojunction injection laser.

Conventional double heterojunction injection lasers not only reduce the oscillation starting current density but also stabilize the oscillation state by introducing a stripe structure, etc., but electrically they are all diodes. Direct modulation requires an external modulation transistor or the like.

As a prior art, an application has been filed for a semiconductor device in which a semiconductor laser and a bipolar transistor are integrally stacked, with the aim of driving a semiconductor laser with low power and increasing input impedance (Japanese Patent Laid-Open No. 1986-1999). No. 14692). However, in a structure in which a bipolar transistor and a semiconductor laser are stacked, it is possible to drive the semiconductor laser with low power due to the amplification function of the bipolar transistor, but the bipolar transistor is not suitable for the high-speed modulation characteristics required of a semiconductor laser. Since the modulation characteristics of the device are determined by the high frequency characteristics of the bipolar transistor, a faster driving element is required. In addition, since no consideration is given to current concentration in the active layer, the threshold current does not decrease, the light emitting position becomes unstable, and the optical output is inevitably unstable, making it impossible to obtain modulation of the optical output according to the modulation signal. It has the disadvantage of

Furthermore, no consideration is given to axial mode stability, and a high-quality light source with good monochromaticity cannot be provided.

Furthermore, as a prior art, a semiconductor laser device in which a semiconductor laser and a FET are integrally formed has been proposed for the purpose of high-speed modulation of optical output and transverse mode stability (Japanese Patent Laid-Open No. 104488/1983). However, the modulation characteristics are limited by the high frequency characteristics of the FET, and although transverse mode stability is improved by the depletion layer extending from the gate of the FET, axial mode stability is not improved at all. .

The purpose of the present invention is to eliminate these drawbacks, to efficiently inject carriers into the active layer of a semiconductor laser in order to control laser output efficiently, stably, and at high speed, and to achieve low threshold current and high speed control. It is an object of the present invention to provide a semiconductor laser device that can be controlled in an axial mode and is stabilized in an axial mode.

FIG. 1 is a perspective view of a semiconductor laser in which an electrostatic induction transistor and a semiconductor laser are integrated for high-speed modulation. For example, the material is GaAs.
Shown for Ga _1-X Al _X As. On p ⁺ -GaAs substrate 1, p ⁺ -GaAlAs layer 2, active layer n-GaAs 3, n
The structure consists of a -GaAlAs layer 4, an n ⁺ GaAs layer 5, a p ⁺ region 6 surrounded by a line 12, and electrodes 7, 8, 9. As for the device, electrode 8 is an anode, electrode 7 is a cathode, and electrode 9 is a gate.

A region 10 surrounded by a dotted line 11 in FIG. 1 indicates a depleted region. As shown in FIG. 1, the depletion layers 10 extending from the p ⁺ gate regions on both sides are in complete contact with each other in the carrier path (channel) at zero gate voltage, and the channel is in a pinch-off state at so-called zero gate bias. is being done. It is a normally-off type static induction transistor in which no current flows in the zero gate bias state. When the channel is not in a pinch-off state with zero gate bias, the depletion layers 10 are not in contact with each other, and only the width of the current path is changed,
By incorporating a normally-on static induction transistor in which current flows even with zero gate bias, current always flows even with zero gate bias, resulting in wasteful power consumption. FIG. 2 shows the potential distribution of the gate and channel portions when cut in the AA' direction of FIG. 1. Taking Ga _0.7 Al _0.3 As as an example, when the channel is sufficiently pinched off with zero gate bias, if positive voltages of approximately 0.4V, 0.8V, 1.2V, and 1.6V are applied to the gate, the potential distribution of the channel will change. changes like 22, 23, 24, 25. A dotted line 26 indicates the cathode potential. As the forward gate voltage increases, the potential barrier in the channel gradually lowers, and electrons are injected into the channel from the cathode region 5, allowing current to flow. Since the p ⁺ gate is deeply biased in the forward direction, holes are injected from the gate into the channel, and the channel has almost equal amounts of electrons and holes.

FIG. 3 is a potential diagram in the BB' direction of FIG. 1. Gate voltage is zero. Area 1, 2,
3, 4, and 5 are the same as the areas in FIG. 1, and the vertical axis indicates the potential. In the figure, region 27 is a potential barrier generated in the channel due to p ⁺ n diffusion potential,
It becomes a barrier to carrier (electronic) 31. 32 indicates a hole. The gate voltage increases the barrier height to the second level.
The cathode n ⁺ region 5 changes as explained in the figure.
Electrons are injected into the n ^- channel region 4 from then on, and the amount of electron injection into the region of the active layer 3 changes.

FIG. 3a is a distribution diagram when the voltage between the cathode and the anode is zero. FIGS. 3b and 3c show the potential distribution between the cathode and the anode when a positive voltage is applied to the anode. FIG. 3b shows the potential distribution when the depletion layer extending from the p ⁺ gate has almost reached the active layer, and FIG. 3c shows the potential distribution when the active layer is also almost depleted. It is somewhat difficult to see because of the appearance of heterojunctions, but the first
When viewed as an electronic device, the device having the structure shown in the figure is an n-channel electrostatic induction thyristor.
When viewed in the direction of line B-B' in FIG. 1, this device has an n ⁺ np ⁺ diode structure. The mechanism by which current does not flow even when a forward voltage is applied to this diode and the current is maintained at zero can be explained as follows. Electron injection from the n ⁺ cathode region 5 is inhibited by an electron barrier 27 created in the channel by the p ⁺ gate, and hole injection from the p ⁺ anode region 2 is inhibited by the p ⁺ n
This is because it is suppressed by the diffusion potential. Third
When the active layer 3 starts to be depleted and becomes completely depleted as shown in Figure c, the barrier potential due to the p ⁺ n diffusion potential is lowered, holes are injected from the p ⁺ anode region, and the current increases. It starts to flow. Of course, when a heterojunction is incorporated as in the structure shown in FIG. 1, even if the active layer is depleted, the current flowing at that time is sufficiently smaller than that of a homojunction structure. In any case, in order to create a normally-off device in which no current flows even if a certain amount of positive voltage is applied to the anode as long as the gate voltage is kept at zero, it ^is necessary to
It is essential that a sufficiently high potential barrier be created in the channel by the n ^-channel diffusion potential.

The impurity density of the n-region 4 in the channel section surrounded by the gate with a gate-to-gate spacing 2a (cm) is N _D (cm
^-3 ), the channel length determined by the length of the p ⁺ gate in the anode/cathode direction is L (cm). first,
In order for the depletion layers extending from the p ⁺ gate region to completely contact each other, completely pinching off the channel and creating a potential barrier, N _D (2a) ² <4.5×10 ⁷ (cm ^-1 ). , at least required.

If 2a=1μm, N _D <4.5×10 ¹⁵ cm ^-3 ,
If 2a = 2 μm, then N _D <1.1×10 ¹⁵ cm ^-3 . Furthermore, even if a certain voltage is applied to the anode, the presence of a potential barrier in the channel requires 2a
On the other hand, the channel length L must be long to some extent. If L/2a is not at least greater than 0.5, there will be no potential barrier in the channel when a voltage of 1 to 2 V is applied to the anode. To further improve normally-off characteristics,
It is desirable that L/2a is 0.7 or more. Therefore,
N _D (2a) ² <4.5×10 ⁷ (cm ^-1 ), more preferably N _D
(2a) ^{Unless 2} <2×10 ⁷ (cm ^-1 ) and L/2a>0.5, more preferably L/2a>0.7, do not hold true at the same time, it will not become a normally-off device. Regarding the relationship between the two, if the height of the potential barrier in the channel is to be kept equal, the smaller N _D (2a) ² becomes, the smaller the value of L/2a can be selected. Normally, to obtain good normally-off characteristics, N _D is 10 ¹⁴ cm ^-3 or
A value of the order of 10 ¹³ cm ^-3 is chosen. That is,
The impurity density of the n ^- GaAlAs layer 4 is selected to be within this range. Even with such a low impurity density in the channel, when a deep forward bias is applied to the p ⁺ gate, the density of electrons injected into the channel from the cathode exceeds 1×10 ¹⁷ cm ⁻³ . Electrons injected from the cathode flow into the active layer 3, as shown in FIG. 3b, for example. Electrons flowing into the active layer 3 cannot flow out to the p ⁺ region 2 due to the barrier potential, and are accumulated in the active layer 3. Since electrons are accumulated, the active layer 3 is negatively charged. The junction between p ⁺ region 2 and n region 3 is region 3
becomes forward biased as it becomes negatively charged. That is, holes are injected from p ⁺ region 2 into active layer 3 . Due to the potential barrier between regions 3 and 4, holes hardly flow out to n region 4. That is, both electrons and holes are accumulated in the active layer 3. When the density of electrons and holes increases to a certain degree, laser oscillation occurs. That is,
The device having the structure shown in FIG. 1 has a structure similar to a static induction thyristor when viewed as an electronic device. An active layer with a narrow forbidden band is provided adjacent to the anode of the electrostatic induction thyristor, and the structure is such that laser oscillation occurs there. moreover,
The flow of electrons from the cathode to the active layer is narrowed by the barrier of the depletion layer extending from the gate.
The current flows within the channel region and flows into the adjacent active layer to emit light, but at this time, the current can be concentrated in the direction parallel to the active layer junction surface by the gate barrier. Furthermore, considering the spread of current between the bottom of the gate region and the active layer,
If the thickness is 3 μm or less, the current concentration due to the gate will reach the active layer and emit light without expanding greatly, and the stability of the light emitting position and transverse mode stability will be significantly improved.

To cut off the current and stop laser oscillation,
Just return the forward bias of the p ⁺ gate to zero.
In order to further shorten the turn-off time, it is sufficient to apply several volts to the gate in the opposite direction when shutting off. Very fast shutoff is possible. As mentioned earlier, the structure shown in Figure 1 is basically similar to a static induction thyristor, but most of the holes injected from the anode remain in the active layer and reach the p ⁺ gate. It will never be reached. Further, the turn-off process is a process in which holes injected from the p ⁺ gate during the channel are sucked out to the p ⁺ gate again.
If the channel width is narrow, it will be extremely fast.

An outline of the structure shown in FIG. 1 will be described below. p ⁺ substrate 1
Impurity density: 1~2×10 ¹⁹ cm ^-3 , p ⁺ Ga _{1-X Al}
The thickness and impurity density of the As layer 2 are about 1 to 5 μm and about 1×10 ¹⁸ to 1×10 ¹⁹ cm ^-3 , and the thickness and impurity density of the GaAs active layer are about 0.05 to 1 μm and 1×
The thickness and impurity density of the nGa _1-X ^Al _X As layer ⁴ are approximately 0.5 to ³ μm and 1×10 ¹³
The impurity density of the n ⁺ GaAs layer is about 1 x 10 ¹⁸ - 1 x 10 ¹⁹ cm ^-3 , and ^the impurity density of the p ⁺ gate region 6 is about 1 x ^{10 17 -} 5 x 10 It is about ¹⁹ cm ^-3 . Gate-gate spacing (channel width) 0.3~
It is about 3 μm. The channel width is twice the hole diffusion distance because when a forward voltage is applied to the p ⁺ gate, the holes injected from the gate to the n-channel region increase the amount of electrons injected from the n ⁺ cathode. Shorter is preferable. It is also possible to widen the channel width and use an operation that allows a certain amount of current to flow even at zero gate bias.
To reduce the current to zero, a certain amount of reverse bias is applied to the gate.

As is well known, a semiconductor laser does not start emitting laser light immediately after a current begins to flow therein. No light is emitted unless the carrier density of the active layer exceeds a certain threshold. Assuming that the threshold current density J _th , the unit charge q, the active layer thickness d, and the lifetime τ due to spontaneous emission of kiloria within the active layer, the threshold carrier density is approximately given by τJ _th /qd. Therefore, in order to obtain laser emission in response to the signal voltage applied to the gate, a current close to the threshold current is normally allowed to flow. You can use a normally-on structure so that a current close to J _th flows even with zero gate bias, or you can use a normally-off structure and apply a certain amount of forward bias to the gate so that a current close to J _th flows. It is also possible to allow a current close to that to flow.

The structure in Figure 1 consists of layers 2 ^, 3,
4 and 5 are continuously provided by epitaxial growth,
p ⁺ gate region by Be ion implantation 1~5× ¹⁰¹⁹
It can be constructed by implanting about cm ^-3 and annealing at 800 to 900°C in an AsH ₃ atmosphere of about 1 to 3 Torr.

In the structure shown in FIG. 1, layer 5 is GaAs, layer 4 is, for example, Ga _0.7 Al _0.3 As, and region 6 is a p ⁺ region, so when the gate voltage approaches the diffusion potential of GaAs, than the Ga _0.7 Al _0.3 As channel.
A forward current flows between p ⁺ GaAs and n ⁺ GaAs in layer 5, and the current gain defined by the ratio of anode current to gate current drops. Therefore, the gate region 6
A high resistance region 52 is formed between the cathode region 51 under the cathode electrode 7 by implanting ions such as protons. Figure 4 is an example. This prevents carrier injection from the gate from the gate region 6 to the cathode region 51, increasing the current gain. Further, if a high resistance region 53 is provided next to the gate region as shown in region 53 in FIG. 4, the capacitance of the gate is reduced, the modulation frequency is increased, and the current gain is also improved. As another structure for suppressing carrier injection from the gate region to the cathode region, the layer 5 may be partially removed to separate the cathode and the gate as shown in FIG. 5, or the layer 5 may be completely removed as shown in FIG. 5
The characteristics can be improved even if it is formed without the . Area 5
1 is the n ⁺ area. In GaAs and GaAlAs double heterostructure semiconductor lasers, the layer 5 is formed solely for the purpose of lowering the ohmic resistance, and is not due to any essential problem in laser operation. In double heterostructure semiconductor lasers such as InP and InGaPAs, InP,
It has a three-layer structure of InGaPAs and InP, so
The structure is similar to that shown in Figure 6.

What has been described so far has been the case where the gate region is formed from the surface, but the gate region may also be buried. FIG. 7 is an example. In the case of a buried gate as well as in the case of FIG. 7, holes are injected from the gate region 6 to the cathode region, so further improvement can be achieved. FIG. 8 is an example. If the portion of layer 5 in the figure adjacent to p ⁺ gate region 6 is made into a semi-insulating region 62 and only the region 61 where electrons are injected is made into n ⁺ region, the current gain is improved. Also, layer 5
If GaAlAs is used, which is the same material as layer 4, it will become a completely buried gate type. Since the current control portion of the semiconductor laser described above is similar to the operation of a bipolar mode static induction transistor, this operation will be considered.

Figure 9 shows the current-voltage characteristics of BSIT made of GaAs. Gate-to-gate spacing 1.5 μm, p ⁺ gate depth 2 μm, distance from n ⁺ source region to n ⁺ drain region 2.5 μm, channel stripe length
Impurity density of 200μm channel 5×10 ¹³ cm ^-3
This is the current-voltage characteristics of GaAsBSIT. The vertical axis is the drain current, and the horizontal axis is the drain voltage. Vg in the diagram
indicates the gate voltage. The current increases as the gate voltage Vg increases in the forward direction.
In BSIT, even though the channel is composed of a region with extremely low impurity density, carriers are directly injected into the channel region from the source region, so it is easy for more than 1×10 ¹⁷ cm ^-3 of electrons to enter the channel. Injected. Furthermore, the electron mobility is 8000~
Since it is as large as 9000 cm ³ /Vsec, a very large current density can be obtained, and a current density of 1×10 ⁴ A/m ² or more can be achieved extremely easily.

FIG. 10 shows the current-voltage characteristics of the structure shown in FIG. The vertical axis is the anode current, and the horizontal axis is the anode voltage. In the figure, Ig indicates the gate current at a positive gate voltage. When the structure shown in FIG. 1 is adopted, the current-voltage characteristics change from those shown in FIG. 9, and in the conductive state, they almost become as shown in FIG. 10. The structure shown in FIG. 1 is made of a GaAs-Ga _0.7 Al _0.3 As system, and the dimensions are the same as those from which the characteristics shown in FIG. 9 were obtained. The thickness of the active layer is approximately 0.5 μm.
Laser oscillation has been observed at 20-30 mA.

Although all of the above-mentioned examples are single-channel type semiconductor lasers, multi-channel type semiconductor lasers can also be easily implemented. FIG. 11 shows a cross-sectional structure of a multi-channel semiconductor laser in a direction perpendicular to the laser resonator direction. As shown in FIG. 11, if a multi-channel type is simply used, the laser beams of each light emitting area 101 emit light with completely different phases. However, in the design, the laser emitting region 1
For example, when using a GaAlAs or GaAs laser, layer 4 has a thickness of about 2 μm and a stripe width of 3 μm.
Then, if the distance between the stripe centers is set to approximately 10 μm or less and some of the stripes are combined,
Each laser emission spot is synchronized,
Emit laser light with uniform phase. When the phases of multiple spots are aligned, the light emission directionality becomes narrower and more sensitive than that determined by a single spot. The interval between the laser emitting regions depends on the carrier diffusion distance and the seepage distance of the laser beam, and varies depending on the material, structure, etc. If all the light emitting regions 101 emit light in the same phase, it will be extremely effective as a high output semiconductor laser light source. On the other hand, if you want to control each laser emission region 101 independently by gate voltage, if you provide a high resistance region 111 between every other gate,
Each element can be controlled independently by the gate voltage of each element. FIG. 12 shows an example thereof. Further, the regions 111 may be separated into elements by etching or the like. However, the spacing between the light emitting regions 101 has the problem described above. When the intervals between the light emitting regions 101 are narrow, they interact with each other and do not operate independently. Therefore, when operating each element independently, use a Ga _0.7 Al _0.3 As or GaAs laser, for example, with a layer 2: 2 μm thick. , layer 3: 0.1-0.3 μm, layer 4: 2 μm
When the stripe width is about 3 μm in m, the stripe interval is preferably about 20 μm. As shown in FIG. 12, if a plurality of electrostatic induction thyristor type semiconductor lasers are manufactured on the same substrate and used for optical communication, for example, it becomes possible to simultaneously transmit information through multiple channels. Able to process large amounts of information at high speed. In addition, the directivity of the laser beam can be changed by making the interval between the light emitting regions of the semiconductor laser 10 μm or less and sequentially switching from the first laser. becomes possible. In addition, the directivity can be improved by initially applying current to every other stripe with a spacing of about 10 μm and operating each strip independently, and by applying current to the stripes in between to match the phase of each light. , the light output can also be controlled. It goes without saying that the structures shown in FIGS. 4 to 8 may be applied to the multichannel semiconductor lasers shown in FIGS. 11 and 12. In addition, the material is GaAs, GaAlAs, InP,
Needless to say, the present invention is not limited to InGaPAs and can be applied to other materials.

The example described above has a structure that can modulate light at high speed and improve transverse mode stability, and if a multi-stripe structure is adopted, high output light can be obtained.
No consideration was given to axial mode stability. The semiconductor laser of the present invention, which was created in view of the above drawbacks, can realize a semiconductor laser with a single emission wavelength and high output. FIG. 13 shows an example thereof. FIG. 13a is a surface view, and FIG. 13b is a sectional view taken along line D-D' in the figure. Each region will be explained using an InP-InGaAsP semiconductor laser as an example. 111:
p ⁺ InP substrate, 102: p ⁺ InP growth layer, 103:
nInGaAsP active layer, 104: n ^- InP growth layer, 10
5: n ⁺ cathode region, 106: p ⁺ gate region,
107: cathode electrode, 108: Si ₃ N ₄ , SiO ₂ ,
Insulating layer such as AlN, 109: Mirror surface, 11
0: Anode electrode. The period of the laser emission direction of the p ⁺ gate is λ/2n (λ: emission wavelength, n:
(approximately the refractive index of the active layer). for example,
If λ=1.5μm and n=3.3, p ⁺ gate 106
The period of is approximately 0.23 μm. For example, the length of the p ⁺ gate region 106 in the laser emission direction is
If it is 0.13 μm, the gate-to-gate spacing is 0.1 μm. The impurity density of the n ^- region 104,
The p ⁺ gate depth (channel length) is designed under the conditions described above. Since the gate-to-gate spacing is quite narrow, the impurity density in the n ^- region can be as high as about 1×10 ¹⁵ cm ^-3 and still be normally off. If the channel length is 0.1 to 0.2 μm or more, the normally-off type can be obtained. In the semiconductor laser shown in Fig. 13, in which the current flows periodically only in the vicinity where the electric field strength of the axial standing wave of laser emission is maximum, the axial mode tends to oscillate in a single mode. ,
Moreover, it operates in a single axial mode even when the current value becomes considerably large compared to the threshold current. This semiconductor laser is configured such that the standing wave of the laser emission and the gain distribution are synchronized, and the semiconductor laser has a single axial mode.
Operates up to very large outputs. Additionally, there are transverse modes and vertical modes.
If the mode) is also made single, complete single wavelength, single frequency operation can be achieved. To make the transverse mode single, the length in the cross-sectional direction surrounded by the p ⁺ gate region 106 should be about several μm, and for the longitudinal mode, the active layer thickness should be about 0.5 μm or less. good. These values vary depending on the difference in refractive index between the active layer and the layers outside it. In order to make the axial mode single, it is also effective to shorten the resonator length.

In this way, a semiconductor laser that operates completely in a single mode in axial mode, longitudinal mode, and transverse mode maintains single wavelength operation even when direct modulation is performed to modulate the injection current of the semiconductor laser. When used for communications, etc., the entire system is stable and extremely effective.

In FIG. 13, if the gate-to-gate interval in the direction of the laser resonator is narrowed, the current flows only in the vicinity of literally the maximum electric field strength of the laser light standing wave, making it easier to unify the axial mode. One reason why the axial mode becomes multiple modes is that carriers injected into the active layer diffuse in the lateral direction.
Therefore, in order to suppress this lateral diffusion of carriers, it is desirable that the active layer be thinner. More preferably, the thickness of the active layer is thinner than the period λ/2n of the standing wave of the laser beam. Therefore, in the case of a semiconductor laser with a single axial mode, the difference in refractive index between the active layer and the outer layers is set to be large enough to ensure that light is sufficiently confined even if the active layer is thinned to some extent. It is desirable to be present.

In order to suppress the lateral spread of carriers, it is of course preferable that the p ⁺ 106 region has a depth that is almost close to contacting the active layer 103. In order to increase the ^current gain, the 10
4 regions are made into semi-insulating regions by proton irradiation or the like.

It goes without saying that the structure of the semiconductor laser of the present invention is not limited to that described here. The conductivity types may be completely opposite. Furthermore, in order to shorten the channel length and improve normally-off characteristics in the structure shown in Figure 13, it is effective to insert a thin p layer with relatively low impurity density between the n ⁺ cathode region and the active layer. . The laser emitting part also has the structure explained here, as well as the BH (Buried
Heterostructure), CSP (Channel Substrate)
Planar), PCW (Plano−Couvex Waveguide)
It goes without saying that structures etc. can be introduced. The material is also not limited to GaAs-GaAlAs and InP-InGaPAs.

Further, the present invention can also be applied to a semiconductor laser having a multi-stripe structure as shown in FIG.

As described above, the present invention has a small threshold current,
Laser output can be easily controlled by gate voltage, high-speed modulation can be performed, and the range of applications is further expanded by making it multi-channel.
It can also be realized when direct modulation is performed to emit light of a single wavelength and single frequency, making it extremely effective in the field of communications.

[Brief explanation of the drawing]

Figure 1 is a perspective view of a semiconductor laser, and Figure 2 is a perspective view of a semiconductor laser.
The potential distribution in the A-A' cross section in the figure, FIG. 3 is the potential distribution in the B-B' cross section in FIG. 1, FIGS. 4 to 8 show other embodiments of the semiconductor laser, and FIG.
Current-voltage characteristics of BSIT made of GaAs, Part 1
0 shows the current-voltage characteristics of a semiconductor laser, FIGS. 11 and 12 show an example of a multi-channel semiconductor laser, and FIG. 13 shows an embodiment of the present invention in which the axial mode is stabilized.

Claims (1)

[Scope of Claims] 1. A first conductivity type having a high impurity density region having a wider forbidden band width than the active layer on the main surface of the active layer; a first conductivity type high impurity density region that is in contact with the active layer and has almost the same forbidden band width as the active layer that is to become an anode; a channel region made of a high resistance region; a cathode region of a second conductivity type opposite to the first conductivity type made of a high impurity density region at one end of the channel region; An electrostatic induction type gate region consisting of a high impurity density region is provided around the cathode region so as not to block the channel region, and an anode electrode, a cathode electrode formed on the anode, cathode, and gate, It has a gate electrode, and by controlling a depletion layer extending from the gate region, charges of the second conductivity type flowing into the active layer from the cathode region through the channel region are concentrated so as to reduce the threshold current, and the channel region is The relationship between the thickness L of the region and the distance 2a between the gate regions is L/2a > 0.5, the distance between the shortest part of the gate region and the active layer is 3 μm or less, and the gate region is 1. A semiconductor laser characterized in that the semiconductor laser is formed periodically with a period of approximately λ/2·n (λ: oscillation wavelength, n: active layer refractive index) in the direction of the emission axis. 2. A plurality of sets of the cathode region and the gate region are provided in a direction perpendicular to the laser emission axis and parallel to the active layer, and a multi-channel structure with two or more laser emission spots is formed. A semiconductor laser according to range 1.