JPH0764139A - Optical semiconductor logic element - Google Patents

Abstract

PURPOSE:To have tandem type semiconductor lasers turned on by direct input of a light pulse from outside or by amplified light or to change its output so as to stabilize the lasers by adopting the constitution to make switching and logic operation by means for implanting the light within a specific range to the oscillation wavelength of an optical bistable element. CONSTITUTION:This logic element is the optical bistable element consisting of the semiconductor laser controlled in basic mode. The electrodes of a gain region LD1 and absorption region LD2 are electrically separated. The light within the range of +5 to -10nm of the oscillation wavelength of the bistable element is implanted into the waveguide region right under the separating part from a direction parallel with the coupling surface of the lasers in a direction orthogonal with a resonator direction, by which the switching and logic operation are executed. Namely, the interference with the laser beams is weak in case of the wavelengths in the range from +5 to -10nm of the oscillation wavelength of the bistable element. The mode is thus stabilized unless the resonator length is made as short as possible even if the resonator mode of the lasers fluctuate at the coupling point.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an optical semiconductor logic device, and more particularly to an optical logic circuit such as an optical switching circuit and an optical memory.

[0002]

2. Description of the Related Art An optical bistable element is an element in which the output optical power exhibits a hysteresis as the input optical power to the element increases and decreases, and the output power takes two stable states with respect to one input optical power. is there. If this characteristic is utilized, optical signal processing such as an optical switch, an optical memory, and optical waveform shaping becomes possible.

As one method of realizing an optical bistable device, as shown in FIG. 13, the injection current is spatially non-uniform,
There is a tandem electrode semiconductor laser that absorbs light in a low injection region. The operation principle will be described below by using “Semiconductor Laser” (Baifukan) edited by Ryoichi Ito and Michiharu Nakamura.

As shown in FIG. 13, the tandem electrode semiconductor laser has a region A in which light is amplified by sufficient current injection and a region B in which light is absorbed and light is absorbed. In the figure, 1 is an active layer, 2 is a p-side electrode, and 3 is an n-side electrode. Bistability is provided by region B acting as a saturable absorber. That is, if the injection current into the region A is increased while the injection into the region B is kept constant, the absorption in the region B is strong until the oscillation occurs and the threshold current is high, but the oscillation occurs once. Since the absorption in region B saturates, the threshold current decreases, and the output exhibits a hysteresis characteristic with respect to the current injected into region A. Even when light is injected from the outside, bistable optical input characteristics occur due to saturable absorption.

The characteristics of this device are that the carrier densities in the regions A and B are n _A and n _B , and the averaged photon density S in the resonator is S.
Is described by the rate equation. dn _x / dt = n _x / τ _x −g _tx (n _x ) · S + I _x / q _x (1) dS / dt = [(1-h) q _tA (n _A ) + hq _tB (n _B ) − 1 / τ _P ] S = 1 / τ _p (G-1) S (2) where τ _p is a photon lifetime, g _tx is a mode gain coefficient per unit time in each region, that is, g _t = Γ ( c / _ng )
g and G are normalized gains over the entire resonator length. Also, n _g
Is the effective refractive index at the cavity mote, τ _x is the carrier lifetime, d
Is the active layer thickness, Γ is the optical confinement coefficient, q is the electron charge,
C is the speed of light.

For G = [(1-h) q _tA (n _A ) + hq _tB (n _B )] τ _p (3) q _tx (n _x ) the approximation shown in FIG. 14 is performed. Although the gain (absorption) coefficient g _t and the carrier density n _x have a non-linear relationship as shown by a broken line, they are approximated by the following “Equation 1”.

[0007]

[Equation 1] Two straight lines at q _tB in FIG. 14, representing the q _tA.

From the equations (1) and (4), the gain G in the steady state is
It is expressed by the following formula. G = X / (1 + Y) + β / (1 + Y / δ) (5) Here, X and β are the normalized injection rate of each region, Y is the normalized lattice density, and δ is the gain (absorption) of each region. It is a parameter related to saturation. It is given by each of the following equations "equation 2", "equation 3", "equation 4", and "equation 5".

[0009]

[Equation 2]

[0010]

[Equation 3]

[0011]

[Equation 4]

[0012]

[Equation 5]

FIG. 15 is a diagram in which the first term of the equation (5) is drawn by a broken line and the second term is drawn by a one-dot chain line for X = Z, β = −1, and δ = 0.1. When Y = 0 (S = 0), each term takes an unsaturated value X, β, but when Y> 0, stimulated emission,
Saturation due to induced absorption occurs. The solid curves are different A
G for the injection rate X is shown, and each has a maximum value for Y. An oscillation state is established at the intersection of this curve and G = 1. When X is increased from the non-oscillation state, at X = X ₁ , Y = 0 (point A ′) shifts to point A and oscillation starts (turn-on). On the contrary, when X is decreased from the oscillation state, at X = X ₂ , the point shifts from point B to Y = 0 (point B ′), the oscillation stops (turns off), and the injection current to the region A becomes Output shows hysteresis. The solution satisfying G = 1 at X = X ₃ in the hysteresis loop has two points C and C ″, but the point C ″ is an unstable solution and is not realized.

In order to obtain the bistability, it is necessary that G has a maximum value with respect to Y, and for that purpose, the following condition must be satisfied. δ <−β / (1−β), β <0 (10) It is _necessary that S _satB is sufficiently smaller than S _satA , and the absorption saturation of B is more remarkable than the gain saturation of A. .

In the semiconductor laser, the gain (absorption) coefficient is non-linear with respect to the carrier density, as shown in FIG.
dg _tB / dn _A > dg _tA / dn _A , and in the non-doped active layer, τ _s = (Bn) ⁻¹ (B
Is a radiative recombination constant), and τ _B > τ _A , which is convenient for bistable operation.

As a representative example, "Turn-off characteristics of bistable semiconductor lasers" by Tomita et al. (A. Tomita et al "Turn-off characteri", published in the journal of the American Physical Society in 1986).
stics of bistable laser diode "J. Appl. phys, 59
(6) pp1839-1842 (1986)) prototype gain region (I),
(II) DC-PBH with a length of 135 μm and absorption region length of 30 μm
(Double channel Planar Buried Heterostructur
e) The characteristics of the structural tandem LD's are shown in FIG.

Rising current 63 mA, turn-off current 2
It showed a remarkable hysteresis characteristic of 8 mA, and the turn-off time was about 20 nS. Further, the effect of external light injection on the tandem electrode type laser is described by adding the photon injection rate S _in to the right side of the above equation (2), and if the bias current is selected directly below the turn-off current, the optical output vs. the optical input. The fact that bistable characteristics were obtained was shown by "Optical Input / Output Characteristics of Bistable Semiconductor Lasers" published by Kawaguchi in p. 704 from Applied Physics Letter Volume 41, p. 702, American Physical Society (Appl). .ph
ys. Lett, 41.p702-704; Kawaguchi, “Optical
input and output characteristics for bistable sem
iconductor lasers)).

Temperature-controlled 1.55 μm (InGaA) for controlling the wavelength of laser light input to the bistable element
Input the sP / InP) DH laser to the bistable device. The light output is 6 mW. Since the coupling efficiency is 10% or less, the input light is about 0.5 mW to the bistable element. The result is shown in FIG. It shows remarkable bistability at bias just below the turn-off current. The operating time is about 1 nS. When the wavelength of the input laser light is shorter than the wavelength of the bistable element laser, the input optical power forms an electron-hole pair in the bistable element laser as the optical pumping power and is bistable by current injection. It has the same characteristics as the device.

Utilizing the fact that the injection current of the semiconductor laser or the light output with respect to the injection light input is non-linear, FIG.
It is also possible to obtain bistable characteristics by using two semiconductor lasers, one of which is used as a master laser and the other of which is used as an optical amplifier biased to 0.97 times the threshold value, as shown in FIG. In 1983, Nakai et al. Proposed the letter “Optical Bistability in Semiconductor Laser Amplifiers” (TN) proposed as a letter in the European Journal of Applied Physics, Volume 22 pages 310 to 312.
akai, et al “Optical Bistability in a Semicond
uctor Laser Amplifier "Jpn.J.Appl.phy
s, vol.22, pp.L310-L312, 1983) is introduced below.

As shown in FIG. 18, as a semiconductor laser,
A CSP (Channel Stripe Planar Lasers) laser with a wavelength of 0.82 μm, one of which is oscillated,
As a master laser, it is used as an optical input source, and the other is used as a laser amplifier in an LED state of 0.97 times the threshold current. The modulator modulates the master laser light at 2 KHz. The reason why the semiconductor laser amplifier can obtain the bistable characteristic with respect to the optical input is as follows.

If the gain of the semiconductor laser amplifier is Gc, this is given by the following equation. Gc = (1-R) ² G _s / {(1-RG _s ) ² + 4RG _s sin ² φ} (11) where R is the reflectance of the end face and φ is the time during which the light passes through the amplifier once. The amount of phase shift, G _s, is the gain during one passage through the amplifier. Further, Gs can be described as follows.

Gs = G ₀ exp (−APinG _c ) ... (12) where G ₀ is the gain of the single path when the optical input power Pin is as small as possible (Pin → 0), and PinG _c is the optical gain. This is the output of the amplifier.

On the other hand, the phase shift amount φ is φ = φ ₀ + BPinG _c (13) where A and B are parameters, considering that the carrier changes with respect to the optical input and the refractive index changes as a result. Is. Φ ₀ = mπ−π (D / Δλ) (14) Here, D is the difference between the wavelength λ _{s of the} input light and the wavelength λ _r0 of the resonator (amplifier), Δλ is the mode interval, and m is It is an integer.

From equations (11) to (14), the steady-state amplification gain G _c
Is required. FIG. 19 shows that G ₀ = 3.1, A = 0.3 mW ⁻¹ ,
B = 0.38 mW ^-1 , Δλ = 0.29 nm, R = 0.3
The results in the case of are shown in FIG. From this, it can be seen that the tuning power increases as the wavelength shift D increases, and the hysteresis also increases.

[0025]

In the conventional tandem electrode laser, in order to stabilize the injected carrier density and the gain (absorption) characteristics of the gain region and the absorption region in the cavity direction, the temperature control of the entire device is 0.1. Below ℃, wavelength control 1/1
It is necessary to control the spread of light in the horizontal and horizontal directions even when the switch is turned on, in addition to setting it to about 0 nm.

Also in the optical input bistable device, the temperature characteristic is an important parameter and the bias current is 0.1.
It is necessary to control it at mA or less. This is because it is necessary to bias just below the turn-off current in order to draw the hysteresis loop as shown in FIG.

Using two semiconductor lasers, one serving as a master laser and the other serving as a semiconductor laser amplifier,
If used, the master laser light must be guided through an isolator and then to an amplifier or modulator. This is due to the wavelength shift due to the formation of the compound resonator by the return light,
This is because the output is not changed. Further, since the laser is used as an amplifier, the reflected light from the end face forms a loop in the amplifier, and a phase shift and a change in output due to wavelength fluctuation increase. Furthermore, the above fluctuations fluctuate temporally and periodically to cause self-excited oscillation.

The present invention has been made in view of the above circumstances. An optical pulse is directly input from the outside, or the input light is amplified, and the amplified light is used to turn on the tandem semiconductor laser or change the output. Therefore, an object is to provide a stable optical semiconductor logic device.

[0029]

DISCLOSURE OF THE INVENTION The present invention is an optical bistable device composed of a semiconductor laser in which a fundamental mode is controlled, in which electrodes in a gain region and an absorption region are electrically separated from each other. From the direction orthogonal to the cavity direction and parallel to the laser junction surface to the oscillation wavelength of the bistable element from +5 nm to −10 n.
By injecting light in the range of m, switching and logical operation are performed. Further, instead of the bistable element, it is characterized in that the same kind of light is injected from the side surface of the active layer of the surface emitting laser to perform the logical operation.

[0030]

When light is input from the outside in the cavity direction of the tandem semiconductor laser, the phase of the input wave is turned on with wavelength shift or mode conversion in order to match the phase of the propagation state in the cavity direction. There is a need. Further, in order to make the change as small as possible, it is necessary to match the wavelength of the input light with the wavelength in the resonator to control the temperature.

On the other hand, the resonator direction is orthogonal to
Light input from a direction parallel to the laser junction surface (even light amplified by a traveling wave type optical amplifier), and a wavelength in the range from +5 nm to -10 nm with respect to the oscillation wavelength of the bistable element. For example, the interference with the laser light is weak, and the resonator mode of the laser may fluctuate at the coupling point,
The mode is stable unless the resonator length is made as short as possible (for example, about 2 to 3 times the coupling region length).

In the traveling wave type semiconductor optical amplifier, the spread of the spectrum with respect to the input light is within ± 5 nm in the case of a waveguide having the same structure as the bistable element, and the interference is suppressed, and the laser is suppressed. Can be coupled with the waveguide of.

Further, in the direction orthogonal to the laser resonator direction, the reflectance is made low on both the bistable element and the emission surface of the optical amplifier, and the return light in the orthogonal direction and the optical pulse signal due to the formation of the external resonator are generated. The delay of the signal fall time due to the tailing effect is suppressed.

Since the amplification degree of the optical amplifier increases non-linearly like the semiconductor laser when the input light is once passed through the amplifier, the temperature and wavelength are controlled more than the use as the optical amplifier under the threshold current of the semiconductor laser. Will be easier.

Further, in order to monitor the non-linear amplification degree,
A monitor is formed at a position facing the input side with the bistable element interposed therebetween. For this, an ordinary PIN photodetection element of Si or a similar element in which the stripe width of the optical amplifier on the input side is widened at least twice or more can be used as the photodetection element.

When an optical input signal is input to the insulating region of the bistable element from the surface described above, electron and hole pairs are formed on the element surface by high excitation, and then carriers move and recombine in the active layer in the coupling region. Therefore, the excitation pulsed light needs to be at least W class (peak value). On the other hand, in the case of directly injecting light into the active region, the pumping light amount is in the mW class and the operating speed is high.

[0037]

Embodiments of the present invention will be described below with reference to the drawings. (Embodiment 1) This will be described with reference to FIGS.

Reference numeral 21 in the figure is an n-type GaAs substrate.
Tandem semiconductor lasers LD ₁ and LD _{2 in the} region of the substrate
The n-type Al _0.45 Ga _0.55 As clad layer (first clad layer) 22, the active layer 23, the p-type Al _0.45 Ga _0.55
An As clad layer (second clad layer) 24 and a p ⁺ type cap layer 25 are formed. Both sides of the second cladding layer 24 along the light output direction are partially removed, and an n-type In _0.49 Ga _0.51 P current blocking layer (first current blocking layer) 26 is formed in the removed portion.

The length of the coupling region between the LD ₁ and LD ₂ is the carrier diffusion length (up to 5 μm). The stripe width is 2 μm in the real refractive index guided structure in consideration of the mode control in the horizontal lateral direction. Both sides of the mesa are wider than the band gap of the active layer (for example, the oscillation wavelength is 850 nm).
Embedded) by the first current blocking layer 26. A p-side electrode 27 is formed on the cap layer 25 and the first current blocking layer 26. In addition, the LD ₁ ,
A low reflection film 28 is formed on both sides of LD _{2 in} the cavity direction. Further, in the coupling portion (current non-injection region) 29 of LD ₁ and LD ₂ , the cap layer 25 is removed and an insulating film made of SiO ₂ is formed on the surface to suppress current injection from the surface. There is. In addition, the above LD ₁ , LD ₂
For controlling the lateral and lateral modes of the second cladding layer.
The remaining thickness of the bottom of 24 mesas is 0.3 μm, and the refractive index difference between the stripe portion and the current block area is 7 to 10 × 10 ^−3.
And the far-field pattern has a unimodal pattern.

The substrate in the region excluding the LD ₁ and LD ₂
A first cladding layer 22, an active layer 23, a second cladding layer 24, a first current blocking layer 26, an n-type GaAs current blocking layer (second current blocking layer) 30, and a p-side electrode 27 are formed on 21. Has been done. The substrate 21, the first and second cladding layers 22, 24,
A low reflection film 28 is also formed on the side surfaces of the active layer 23 and the first and second current blocking layers 26 and 30.

The optical amplifier A, B, in order to bind from a direction perpendicular to the resonator direction of the tandem-type semiconductor laser LD _1, LD _2, in the form of removal of the second current blocking layer 30 of LD _1, LD _2, Grooves are formed to reach the substrate 21 by dry etching, and each element is separated at 10 μm intervals.
A low reflection film 28 is also formed on the opposite end faces of the optical amplifiers A and B, and the reflectance is at least several percent or less to suppress the end face reflection. Similarly, the reflectance of the input side end surface of the optical amplifier A on the optical input side is also suppressed so that the input signal light is not reflected on the surface. Further, the horizontal and lateral mode control method of the optical amplifier A on the light input side is also the same as the structure of the semiconductor laser, but the second current blocking layer 30 is provided outside the first current blocking layer 26. The points are different. Furthermore, the optical amplifier B on the monitor side has a stripe width of 10 μm.
Therefore, the width is effective for detecting light passing through the tandem semiconductor laser resonator. In addition, 31 in the figure
Is an n-side electrode. The element length of the optical amplifiers A and B is 2
When the optical amplification is required to be 50 μm, the element length of A is lengthened, but in this case, the element of the present invention needs to be assembled on the Peltier element in consideration of heat generation.

FIG. 2 shows an example of the operation of the optical semiconductor logic device having the above structure. In the tandem semiconductor laser, the current in the gain region LD ₁ is made variable and the current in the (saturable) absorption region LD ₂ is made smaller than I ₁ . For example, the current of LD ₁ is fixed to I _b in FIG. Current I _{3 of} optical amplifier A
When the optical input signal does not enter, it is combined with the tandem type LD, and it is effectively the point of I _b ′. At this time, by putting the input signal light, the output of the P _ON turns off. When the input pulse light disappears, the output becomes constant at the point _Pb '. With respect to the signal input, the optical amplifier is turned off by the field back from the monitor side to be in the state of _Ib , and the optical amplifier is turned on again to return to the initial state.

As shown in FIG. 2B, when the bias is turned on and the input wavelength is changed, the amplification degree of the optical amplifier differs depending on the input wavelength. Therefore, the amplified light is tandem semiconductor laser. The optical output P _{out, which} is output in the direction of the resonator in combination with the optical output P _out , responds to the input signal if the signal 1 is equal to or more than half the maximum output regardless of the amplification factor. The response time is as fast as several nanoseconds.

Example 2 Reference is made to FIG. However, in FIG.
The same members as and are denoted by the same reference numerals, and description thereof will be omitted. by the way,
In the optical semiconductor logic device shown in FIG. 1, the light emitted from the optical amplifier A has a beam spot of several μm, and when it goes out into the air, it becomes a point light source and the beam spreads. When the width of the separation groove is narrowed, the coupling efficiency is improved, but it is necessary to provide an interval of about 10 μm from the necessity of coating the opposite surface with low reflection. In order to improve this, an optical semiconductor logic device having the structure shown in FIG. 3 was proposed.

This is due to the p-type Al _0.45 Ga of the bonding portion.
The mesa shape of the _0.55 As clad layer (second clad layer) 24 is expanded in a taper shape in the surface facing the optical amplifier. For example, the width is expanded from 5 μm to 10 μm. Note that reference numeral 32 in the figure denotes a tapered coupling region. By expanding in a tapered shape, the propagation mode in the resonator direction in the coupling portion fluctuates and becomes unstable, but mode conversion occurs again in LD ₂ and the emitted light is a monomodal beam. Since the optical amplifier and the semiconductor laser are integrally formed, there is no center deviation of the beam, and the n-type In _0.49 Ga _0.51 P current blocking layer (first
Current block layer) 26, p-type Al _0.45 Ga _0.55 As clad layer (second clad layer) 24, and n-type Al _0.45 Ga
_0.55 As clad layer (first clad layer) 22 makes active layer
Since 23 is surrounded, it is expected that the coupling from the optical coupling portion toward the resonator will be performed smoothly.

Also, in 1985, Weinard et al., "GaAs /" published in the Upright Physics Letter.
Polarization plane reactivity with strong electric field absorption in AlGaAs quantum well waveguides "(JS Weiner et al," Strong po
larization-sensitive electroabsorption in GaAs
/ AlGaAs quantum well waveguides "Appl.phy
S. Lett, vol47. pp1148-1150, (1985)) that the electric field absorption characteristics of light propagating parallel to the GaAs / AlGaAs quantum well waveguide layer are different from those of the bulk crystal.
It is reported that the absorption edge shifts to the long wavelength region by about 40 meV. Further, in the multiple quantum well laser, the laser oscillation energy becomes smaller by about 20 meV than the absorption edge of the intrinsic material (the wavelength shifts to the long wavelength region) due to the band reduction effect by the injected carriers.

In the quantum well structure, the change in gain obtained with respect to the injected carrier density is larger than that in the bulk crystal due to the quantum effect. In laser oscillation, the change in gain (differential gain) determines the operating injection current density, which is effective in lowering the threshold value.

(Embodiment 3) An explanation will be given with reference to FIG. In the third embodiment, a multi-quantum well structure is introduced as the active layer 41, and the optical waveguides 42 and 43 are provided above and below the active layer to make use of the seeping effect of light to the optical waveguide layer. It reduces the internal absorption coefficient.

For example, as the active layer, the GaAs quantum well width is 10 nm, the Al _0.3 Ga _0.7 As quantum barrier width is 5 n.
A triple quantum well structure of m, with an optical output waveguide layer of A
l _0.3 Ga _{0.7 As} 30 nm multiple quantum well-separated confinement heterojunction structure (MQW-SCH: Multiqua)
ntum well separated Confinement Heterostructur
e) is valid.

The optical amplifier A is a traveling wave type optical amplifier, and there is almost no optical feedback amplification for spontaneous emission light.
It is necessary to amplify the gain coefficient by propagation once in the element.
In this point quantum well structure, the gain coefficient is remarkably larger than that of the bulk, and the energy at the emitting end is about 20 meV lower than that at the band end, so that the absorption coefficient in the propagation region can be suppressed to be small.

Considering the tandem type semiconductor laser, the gain region LD ₁ can be expected to increase the gain for the same reason as the traveling wave type optical amplifier A, and the absorption region LD ₂ has a wavelength dependence of the optical absorption coefficient with respect to the electric field. The absorption coefficient changes by about two orders of magnitude due to the wavelength dependence near the band edge, which is several thousand c
m is ^-1 than several tens of cm ^-1. In this respect, a bistable device and a logic circuit combining a tandem semiconductor laser using a multiple quantum well structure and an optical amplifier are considered to be effective.

(Embodiment 4) Description will be given with reference to FIG. In this Example 4, p-type Al was used from the viewpoint of device reliability.
_{When the} both sides of the mesa of the _0.45 Ga _0.55 As clad layer 24 are filled with the n-type GaInP current blocking layer 26 capable of selectively re-growing, an n-type Ga _0.49 In _0.51 P current blocking layer 51 is formed at the interface.
Is at least 0.1 μm and the lattice constant is In-rich (5
% Or less) It is configured to be embedded. Here, the thickness of the current blocking layer 26 is required to be at least 0.5 μm in consideration of the seepage of the electromagnetic field distribution of light into the cladding layer.

S. S., published in Photonics Technology Letter Volume 8 of the American Institute of Electrical and Electronics Engineers, 1992.
L. Yellen et al., “Reliability of 0.81 μm InAlGaAs strained quantum well lasers” (SL Yellen, et al;
"Reliability of InAlGaAs Strained-Qua
ntum well Lasers Operating at 0.81μm ”IE
EE. photo. Technol. Lett, vol 4, p829-831,
(1992)), a comparison of a quantum well laser of Al _0.03 Ga _0.09 As and a strained quantum well laser of In _0.15 Al _0.13 Ga _0.72 As shows the rapid deterioration in the initial stage observed in the reliability test of the quantum well laser. Is In _0.15 Al _0.13 Ga _0.72
Not seen on As.

The cause of this is that dark lines do not occur in the <100> plane within the laser stripe. This is because the threading dislocations existing in the clad layer surrounding the active layer proceed into the active layer and form a dark line within several hundred hours. This is because the effect on the active layer is suppressed due to the suppression. Especially, in the InGaAsP / InP system, there is no report that the above-mentioned rapid deterioration in the initial stage is observed.

In the fourth embodiment, the cladding layer is made of Al _0.45.
Although the Ga _0.55 As layer is used, the In _0.49 Ga _0.51 P clad layer is effective even if the reliability is improved. The reason for making In rich at the interface is that In is AlGaAs at the interface.
In order to suppress threading dislocations in the layer, they move to move to In and Ga.
This is because it is necessary to increase in advance so that the balance with and does not change. An excess of In5% or less does not cause a large change in the distribution of light bleeding into the cladding layer due to the waveguide structure, but rather suppresses crystal dislocations, vacancies, oxidation, etc. at the interface to produce a clean interface. It is believed that this is more important. From this viewpoint, it is considered that In contributes to suppressing vacancies at the regrowth interface and expansion and generation of dislocations.

(Embodiment 5) Referring to FIG. Optical amplifier A
When the reflectance R of the signal is as close to zero as possible, the phase shift due to the returning light and the traveling light and the spread of the signal light are suppressed, and the speed is increased. Therefore, in the fifth embodiment, as shown in FIG. 6, after propagating through the medium 2 (refractive index n ₂ ) with an inclination of θ _{1 with} respect to the propagation direction of the optical amplifier, a semiconductor laser (refractive index of the medium is n ₁ ) With respect to the direction orthogonal to the center of θ ₂ [= si
n ⁻¹ (n ₁ / n ₂ · sin θ ₁ )] In consideration of the inclination, a mirror is formed by dry etching, and a groove is formed at least up to the substrate. In the fifth embodiment, the mirror is tilted by 5 ° (θ ₁ = 5 °), and the medium 2 is n ₂ = 1 from air.
Therefore, θ ₂ is 18 ° from n _{2 to} 3.5.

The distance between the semiconductor laser and the optical amplifier is 2-3 μm.
If it is about m, the light spread below the active layer does not return. However, when the protective film (low-reflection film) 61 is formed with a distance of 10 μm or more, the substrate surface at the bottom of the groove needs to be in a diffused reflection state. Alternatively, it is necessary to suppress the returning light by attaching a light absorbing material or the like.

Alfons et al., "Mode coupling in tilted-edge semiconductor optical amplifiers and superluminescent diodes," published by Optoelectronics Society Vol. 10, 1992, American Society of Electrical and Electronics Engineers (GA Alphonse et al, "Mod.
e Coupling in Angled Facet Semiconductor O
ptical Amplifiers and Superluminescent Diodes
"IEEE, J, Lightwave Technol, vol10, pp21
5-219 (1992)), the reflectance R when tilted 5 degrees
Becomes less than 10 ⁻⁴ and approaches a non-reflective state.

In the super luminescent diode manufactured as described above, the spectrum spread is about 5 nm,
The output is over 30 nm. In the above Example 5,
As an active layer, GaAs quantum well width 10 nm, Al _0.3 G
a _0.7 As quantum barrier width of 5 nm, a 10-period multiple quantum well active layer is used, and horizontal lateral mode control is performed by embedding a mesa stripe with In _0.49 Ga _0.51 P of a real index guiding type, and a stripe width of 2 to 3 μm. If so, the maximum output is about 80 mW, and the output rising current (threshold current) is 300 μm.
The element length is about 50 to 70 mA.

Actually, since the light is input from the outside of the optical amplifier, the output of the optical amplifier itself is not necessary. However, in the hysteresis loop of the tandem semiconductor laser, the bias current drops below the turn-off current and the input light is input. When the power of 1 is required to be increased, an amplifier with a maximum output of 80 mW can be amplified with a margin. However, since the central wavelength shifts to a long wavelength due to heat generation, which causes a large deviation from the coupling wavelength with the tandem semiconductor laser, in the case of high output, the temperature of the entire Peltier element of the present embodiment is Must be constant. When not propagating in the air, make the angle determined by Snell's law,
The incident light is vertical (orthogonal) at the joint.

Example 6 Reference is made to FIG. However, in FIG.
The same members as and are denoted by the same reference numerals, and description thereof will be omitted. In the sixth embodiment, the n-type distributed reflector 72 is formed on the n-type GaAs substrate 71 by A
l _0.6 Ga _0.4 As / Al _0.3 Ga _0.7 As is formed for 30 cycles with a thickness of λ / 4 cycle. At this time, the reflectance of the distributed reflector as a whole is 99% or more. For example, when the oscillation wavelength λ ₀ is 850 nm, the effective refractive index n _e of the propagating mode is
Is 3.5, an integral multiple of λ ₀ / 2n _e is the resonator length L.
If it does not match with, laser oscillation does not occur. Emitting diameter 5μmφ
To control the basic mode.

On the substrate 71, n-type distributed reflectors 72, n
Type Al _0.45 Ga _0.55 As clad layer (first clad layer)
An active layer 74 is formed via 73. This active layer 74
Is advantageous in that it is thicker in order to repeat the feedback in the thickness direction to increase the gain, but in the sixth embodiment, the active layer thickness of the optical amplifier is also common, and is thicker than the thickness for controlling these fundamental modes. It becomes 0.5 μm or less. For example, if the active layer has a thickness of 0.5 μm, the center of the active layer is the antinode of the standing wave in the surface emitting laser. Standing wave belly and node is λ
Present in the resonator direction at intervals of ₀ / 4n _e. In order to reduce the threshold current of surface-emitting lasers, crystal lattice fluctuations, dislocations, and vacancy defects are present at the interfaces of different layers, so it is ideal to design the interfaces at node positions. .

On the active layer 74, p-type Al _0.45 Ga is formed.
_A p-type distributed reflection layer 76 is formed via a _0.55 As clad layer (second clad layer) 75. This p-type distributed reflection layer 76
Is an Al _0.6 having an In _0.49 Ga _0.51 P layer 77, which is relatively strong against oxidation, for the purpose of forming an electrode on the surface of the optical amplifier.
25 cycles of lambda / 4n _e with Ga _0.4 As, reflectivity 9
It will be about 5%. The surface-emitting laser is biased under a threshold current (about 0.98%), and the same wavelength as the oscillation wavelength is input to the surface-emitting laser via an optical amplifier in a pulse shape, and laser oscillation occurs. . At this time, if the current I _{3 of the} optical amplifier is turned off in response to the input signal to the monitor side, the input light is absorbed in the optical amplifier and the input light is not coupled to the surface emitting laser. When I ₃ flows again, the state returns to the initial state. This relationship matches the operation of FIG.

In the case of the sixth embodiment, when the current of the optical amplifier is changed without input light, the initial low current
The spectrum is obtained in the short wavelength region as compared with the surface emitting laser, which is equivalent to the fact that it is absorbed in the active layer of the surface emitting laser to generate carriers and increase the operating current. Then, if the surface emitting laser is set to about 0.98 times the threshold current, laser oscillation is performed according to the bias pulse current of the optical amplifier, and the external pulse modulation circuit is formed.

[0065] Now, in the distributed reflector, different bandgap greater, to form a wide semiconductor layer band gap at 20-30 cycles lambda / 4n _e intervals than the active layer,
Since the current flowing through this region flows over the hetero barrier, the device resistance becomes as large as 500 to 1 KΩ as compared with about 10Ω of a normal semiconductor laser, and it is important to suppress heat generation and the injection current increases. Therefore, the oscillation wavelength is longer than that of a normal semiconductor laser.

(Embodiment 7) Referring to FIG. In the seventh embodiment, when the n-side electrode 81 of the optical amplifiers A and B is taken from the back surface via the distributed reflector, the bias voltage becomes high. Therefore, as shown in FIG. In
_At least the active layer is removed to the surface of the _0.49 Ga _0.51 P clad layer to form the n-side electrode 81, and the device resistance of the optical amplifier is lowered. Also in the surface emitting laser, the device resistance is lower when the n-side electrode is formed on the clad layer, but since the electrodes are only separated by the low reflection film 28, and the operating current is particularly large, the leakage current It is necessary to consider suppression, and it is on the back side. Regarding the distributed reflector, since the electron mobility is large in the n-side portion, it is unlikely that the element resistance will increase so much, and rather, in the p-side distributed reflector portion, the hole mobility is small, so Resistance increases. Therefore,
P-side Al _0.45 Ga of ring mesa side of surface emitting laser
_0.55 As clad layer 75 and p of the first layer portion of the distributed reflector
A current is injected from the type InGaP layer 77. To reduce the operating current of the surface emitting laser, (1) lower the element resistance,
(2) Leakage current is reduced, (3) Gain coefficient is increased, etc.

(Embodiment 8) Referring to FIG. In order to increase the gain coefficient of the active layer, it is common to introduce a multiple quantum well structure. Also in the optical amplifiers A and B,
When the end face reflectance R is made as close to zero as possible, it is necessary to increase the gain Gc obtained when the light passes through the waveguide once. However, if the number of multiple quantum wells is increased, the carrier density in each well becomes non-uniform and the spread of the spectrum becomes large. In the re-emission laser, since the laser oscillation spectrum is determined by the cavity length, it is necessary to suppress widening of the spectrum width as much as possible. Therefore, a modulation-doped multiple quantum well structure in which the quantum well barrier amount is P-doped is used. In addition, in the active layer of the optical amplifier, since there is a current density at which gain saturation occurs with respect to the injection current per single quantum well, from the viewpoint of controlling the fundamental mode by light emission between the ground levels, It is necessary that the light confinement coefficient is 3% or less, the light confinement coefficient of the entire active layer is 50% or more, and the horizontal and horizontal modes are the basic modes. For example, GaAs quantum well width 1
0 nm, quantum well barrier Al _0.3 Ga _0.7 As 5 nm (P
Dope with an impurity concentration of 7 × 10 ¹⁷ cm ⁻³ ) and 10 cycles. When the active layer of the surface emitting laser is viewed under this condition, the active layer thickness is 0.15 μm, which is not a thick active layer. Therefore, it is necessary to reduce the leak current as much as possible and efficiently inject the current into the active region.

As shown in FIG. 9, Ga ⁺ ion implantation (dose amount 4 × 10 ¹⁴ cm ^-2 ) is performed at the bottom of the mesa of the surface emitting laser,
The active layer 91 has a multiple quantum well structure. The resistance of the p-side Ga ⁺ ion-implanted region returns to the initial value by annealing,
In the n-side Ga ⁺ ion-implanted region, the state where the resistance is increased by 3 to 4 digits is maintained (this region becomes the current non-implanted region).
In the figure, 92 indicates a Ga ⁺ ion implantation region.

At this time, the active layer is disordered to widen the band gap, and the lateral spread of the surface emission beam can be suppressed. Further, also in the optical amplifiers A and B, the same Ga ⁺ ion implantation is performed in the vicinity of the end face of the stripe, so that the end face becomes a current non-injection region and a so-called window region in which the band gap is widened. With respect to the light input from the outside, the absorption coefficient is suppressed to be small, and the coupling efficiency of the surface emitting laser is increased.

(Embodiment 9) Referring to FIG. The ninth embodiment is a modification of FIG. 9, in which two active regions 101 and 102 are formed and the region between the two is adjusted so that laser oscillation occurs when the center of the active layer is at the antinode of the standing wave. However, when the distance between the two is separated in advance by several tens of nm to several nm in terms of wavelength and the light is input by the optical amplifier, the refractive index of the medium changes due to carrier fluctuation, and the position of the antinode is changed. Then, the laser is absorbed in the lower active layer of the laser to generate carriers, and the wavelength of the propagation mode shifts accordingly and laser oscillation occurs.
If the input wavelength and the laser oscillation wavelength match,
The input wave changes the effective refractive index of the active layer and the surrounding clad layer, resulting in laser oscillation. However, the bias current and temperature control of the surface emitting laser must be controlled by the Peltier device. On the other hand, when the input light (signal light) has a long wavelength, after being amplified by the optical amplifier, mode conversion occurs while propagating in the lower active layer of the surface emitting laser, and laser oscillation occurs.

Next, by setting the element length of the tandem semiconductor laser and the optical amplifier to 100 μm or less and making the optical coupling region wider than the stripe width and 10 μm or less, it is possible to generate an optical very short pulsed light.

(Embodiment 10) Referring to FIG. The device length of the gain region LD ₁ of the tandem semiconductor laser is 50 μm.
m, the optical coupling region 29 is 5 μm, the device length of the light absorption region LD ₂ is 30 μm, and the n-type In _0.49 Ga _0.51 P current blocking layer 26 is a mesa stripe laser having a thickness of 0.5 μm. The element length on the input side A of the optical amplifier is 100 μm, the structure is the same as that of the tandem type semiconductor laser, and the reflectance of the end face is as small as possible. Further, both sides left by the dry etching are made to be tammy to facilitate the coupling of the input signal light as a light shield and a light guide (guide).

The active layer is GaAs with a quantum well width of 10 nm.
And a quantum barrier of Al _0.3 Ga _{0.7 As 5} nm with a double quantum well structure of 2 periods, and Al _0.3 Ga _0.7 As30 on the upper and lower sides.
nm-SCH (Multi qua) having an optical waveguide layer of nm
ntum Well Separated Confinement Heterostruc
ture) structure.

When the bias voltage of the light absorption region LD ₂ is changed, the absorption coefficient changes from one digit to two digits due to the Stark effect. For example, when a voltage of 1 V is applied, the coefficient of the absorption region is 1000 to 3000 cm ^-1 , whereas -1 V
When applied, it changes from 10,000 to 15,000 cm ^-1 . Therefore, the LD ₁ current in the gain region is set to 1.0 of the threshold current.
Biased to about 1.2 times (with respect to Vb = 0), the light pulse modulated from 100 MHz to 3 GHz is amplified by the optical amplifier A, and is coupled with the tandem LD in the optical coupling region. Then, even when the absorption region LD ₂ is not biased, the absorption coefficient changes and an extremely short pulse is generated.
At this time, the pulse width can be changed by changing the bias voltage of LD ₂ . However, since there is a bias voltage that causes self-sustained pulsation, that region is excluded in advance.
With this method, a pulse of about several tens pS can be realized.

In addition, the gain region LD _{1 is set} to the threshold current (V
(when b = 0) Biasing back and forth, high frequency pulse 100
It is also possible to superimpose MHz to 1 GHz, apply a reverse bias number of 10 mV to the absorption region LD ₂ and input an optical switching pulse from an optical amplifier to generate a short optical pulse by gain switching.

(Embodiment 11) Referring to FIG. The eleventh embodiment shows an example of a multi-wavelength integrated optical logic device in which light input to an optical amplifier (SLD) is also integrated. Each element is dry-etched at least until it reaches the substrate, and the end face is coated with a low-reflection coating or is composed of an optical amplifier inclined at 5 °. The structure of the semiconductor laser is a multiple quantum well structure active layer, and this active layer is wrapped with an optical guide layer,
This is a so-called separate confinement heterostructure.

The horizontal / horizontal mode control is of the real refractive index guided type, and the coupling surface with the tandem semiconductor laser and the opposing surface on the monitor side are made of Al having a wide bandgap that is not absorbed by the oscillation wavelength and the input signal. _Is formed of a semiconductor layer not containing (for example, an In _0.49 Ga _0.51 As layer).
In the case of a multi-quantum well laser, if the number of wells is 1 or 2, the wavelength can be shortened by about 10 nm by shortening the cavity length. Especially, in the single quantum well laser, the second level oscillation occurs.

In order to increase the amplification degree of the optical amplifier, it is necessary to increase the number of quantum wells from the point of increasing the gain Gs during propagation. However, on the contrary, semiconductor lasers, LD ₄ , L
It is necessary to change the resonator length of D ₅ to be extremely short. For example,
It is necessary to devise such as 100 μm and 800 μm. Since the spectrum of the optical amplifier has a spread of about 5 nm, the input laser light is not amplified.

It is possible to multiplex the switching characteristics and the logic circuit from multiple wavelengths in a time sequence by inputting light input light into the tandem type semiconductor laser in a pulse form. Also in terms of optical axis alignment, the present invention can be integrated and is a powerful means.

In the above-mentioned embodiment, the description is centered on the GaAs system, but GaInAsP / InP system, AlGaInP system.
It is also applicable to the / GaAs type and further to II-VI group semiconductor lasers.

Further, by using the structure of the present invention as an array, a multi-wavelength time division logic element is formed, and by forming the structure of the present invention between chips, it is also used as an optical interconnection element capable of data transmission. It is possible.

[0082]

As described above in detail, according to the present invention,
The wavelength of light input from the outside and the magnitude of its power,
Further, by changing the voltage (current) of the optical amplifier, the amplification factor of the input light changes non-linearly. By inputting this to a tandem type semiconductor laser, a stable bistable logic circuit for a high-speed response of several nS can be realized.

By further shortening the element lengths of the tandem semiconductor laser and the optical amplifier and making the reflectance of the emission end face of the optical amplifier close to zero, it is possible to realize a further high-speed response and an extremely short pulse operation of several pS.

[Brief description of drawings]

FIG. 1 is a schematic perspective view of an optical semiconductor logic device according to a first embodiment of the present invention.

FIG. 2 is a characteristic diagram of an operation of the optical semiconductor logic device of FIG.

FIG. 3 is a schematic perspective view of an optical semiconductor logic device according to a second embodiment of the present invention.

FIG. 4 is a schematic perspective view of an optical semiconductor logic device according to Example 3 of the present invention.

FIG. 5 is a schematic perspective view of an optical semiconductor logic device according to a fourth embodiment of the present invention.

FIG. 6 is a schematic perspective view of an optical semiconductor logic device according to a fifth embodiment of the present invention.

FIG. 7 is a schematic perspective view of an optical semiconductor logic device according to Embodiment 6 of the present invention.

FIG. 8 is a schematic perspective view of an optical semiconductor logic device according to a seventh embodiment of the present invention.

FIG. 9 is a schematic perspective view of an optical semiconductor logic device according to Example 8 of the present invention.

FIG. 10 is a schematic perspective view of an optical semiconductor logic device according to Example 9 of the present invention.

FIG. 11 is a schematic perspective view of an optical semiconductor logic device according to Embodiment 10 of the present invention.

FIG. 12 is a schematic perspective view of an optical semiconductor logic device according to Embodiment 11 of the present invention.

FIG. 13 is an explanatory diagram of a tandem semiconductor laser.

FIG. 14 is a characteristic diagram showing a relationship between a gain absorption coefficient and a carrier density.

FIG. 15 is an explanatory diagram of the principle of bistability of a tandem electrode type laser.

FIG. 16 is a characteristic diagram of light output versus injection current in a tandem electrode laser.

FIG. 17 is a light output light input characteristic diagram of the tandem electrode laser.

FIG. 18 is an explanatory diagram of an experimental example of bistable with a semiconductor laser amplifier.

FIG. 19 is a diagram showing a case where a bistable operation by a semiconductor laser amplifier is turned on.
Explanatory drawing which shows an example of the theoretical value of an output characteristic.

[Explanation of symbols] 21, 71 ... GaAs substrate, 22, 73 ... first clad layer,
23, 74, 91 ... Active layer, 24, 75 ... Second cladding layer, 25 ... Cap layer, 26 ... First current blocking layer, 27 ... P-side electrode,
28 ... Low reflection film, 29 ... Current non-injection region, 30
... second current blocking layer, 31, 81 ... n-side electrode, 72 ... n-type distributed reflector, 92 ... Ga ⁺ ion implantation region.

Claims (1)

[Claims] 1. An optical bistable element, which is fundamentally mode-controlled and is electrically separated and whose electrodes in a gain region and an absorption region are made of a semiconductor laser, and a waveguide region immediately below the separated portion with respect to a resonator direction. And a means for injecting light within a range of +5 nm to -10 nm with respect to the oscillation wavelength of the optical bistable element from a direction orthogonal to the laser beam and a direction parallel to the laser junction surface, thereby performing a logical operation. An optical semiconductor logic device characterized by the above.