JPH01105589A - Semiconductor raman laser - Google Patents

Abstract

PURPOSE:To oscillate with an extremely low CW exciting optical power by introducing an exciting light from an incident window to a Raman active layer, and enclosing the light inside a clad layer to multiplex reflecting it. CONSTITUTION:A GaP Raman active layer 3 is formed in a stripe shape having an extremely smooth side face, and its sectional periphery is completely surrounded by an AlxGa1-xP clad layer 2. Further, a hole 8 sufficiently smaller than the end face of the active layer is formed at a high reflectivity deposited face on only part of the end face, an injected semiconductor laser light 7 is condensed, and incident from the hole. Since the layers 3, 2 have sufficiently low carrier density, less absorption, and the end face has high reflectivity, the incident light 7 is broadened and guided in the layer 3. Accordingly, it becomes substantially equal to the optical intensity of the incident window of the incident light on the whole part of the active layer, and the oscillation of a stokes light is performed on the whole active layer having an end face sufficiently larger than the incident window.

Description

DETAILED DESCRIPTION OF THE INVENTION (Field of Industrial Application) The present invention relates to a semiconductor Raman laser, and particularly to a semiconductor Raman laser used for optical communication and spectroscopic measurement. In optical communications, it is particularly used as a means for generating light waves in the far infrared region using optical heterodyne detection, and for generating carrier light waves with extremely high wavelength purity and stability.

[Conventional technology]

Semiconductor Raman lasers have been realized by the present inventor using GaP single crystals and epitaxial crystals, but
The bombing light has mainly been pulsed by a YAG laser or a high-output semiconductor laser. To use it for optical communication, it is a low-power semiconductor laser that must be excited with continuous wave (GW), and for use in spectroscopic measurements,
For example, CW dye laser, in any case from 1W to 10
A semiconductor Raman laser that operates with excitation laser light with an extremely low power of mW is required. The power density of the optical excitation laser beam is already 1 to 0.7×10 W/Ce.
It has been confirmed that Raman oscillation occurs. This means that if the incident light can be concentrated in a 1μ square area, it will only be 10m
In principle, this means that it oscillates with the power of W excitation light. On the other hand, the present inventor has published IEE Proceedings Vol.
134, Pt, J, N.

, Il, Auaust 1987, pages 215 to 220, published a Raman laser having a structure as shown in FIG. In Fig. 1, 1 is a Gap substrate, 2 is Aj, XGa, -xP-th cladding layer, and 3 is Ga.
P Raman activity diagram, 4 indicates At-Ga, -Xp second cladding layer, 5.6 indicates high reflectance dielectric film, and 7 indicates incident excitation light. The evaporated film on the incident side is formed in a stripe shape using a evaporation mask, has a width of about 111-, and a resonator length of 3 to 4n+*. The excitation light is incident obliquely forward from the end of the deposited film as shown at 7 and excites the resonator region constituted by 5.6. As a result, the Stokes light is amplified and confined in the vertical direction by the ALXGa, XP cladding layer with a low refractive index, and although there is no particular confinement effect in the lateral direction, the width of the resonator region is as wide as 111, making it easy to oscillate. do.

[Problem that the invention seeks to solve]

In order to make the excitation light incident using the method described above, the beam cross section of the excitation light must have a certain width as shown in Figure 1.
In other words, it must have a width of 100μ or more, and since the Stokes light to be oscillated also spreads in the lateral direction, it usually requires a pulsed excitation light of 10W or more, and the intensity of the excitation light must be set to 1W or less for oscillation. was difficult.

An object of the present invention is to provide a semiconductor Raman laser that oscillates with an extremely low incident power of about 1 W to 10 mW. That is, an object of the present invention is to provide a semiconductor Raman laser that oscillates using a CW output injection type semiconductor laser for optical communication or a CW dye laser for spectroscopic measurement as an excitation laser.

[Means for solving problems]

As shown in FIGS. 2 to 8 of the embodiment drawings of the present invention, a GaP Raman active layer is formed in a stripe shape with extremely smooth sides, and the periphery of the cross section is completely covered with A, L, Ga,
- Surrounded by XP cladding layer. Furthermore, a hole that is sufficiently small compared to the end surface of the active layer is provided in the high-reflectance vapor-deposited surface only in a portion of the end surface, and the injection type semiconductor laser light or dye laser light is focused with a lens or fiber and is emitted from the hole portion, that is, the entrance window. incident. Since the active layer and cladding layer have sufficiently low carrier density, absorption is low, and the end face has high reflectance, the incident excitation light spreads within the active layer and is guided, and is eventually lost through the input window. Repeat multiple reflections until Therefore, the light intensity of the incident light at all parts of the Raman active layer is approximately equal to the light intensity at the entrance window, and Stokes light oscillation can be realized in the entire active layer having a sufficiently large end face compared to the entrance window. The size of the window should be the maximum size that can be focused by the lens.
In this part, the threshold power density of oscillation is 1 x 10”W/am
It can be understood from the above that it is sufficient to reach JL. For example, when concentrating light with a lens, it is possible to condense light down to 1μφ by using a combination of lenses, and even a simple lens can narrow down the light to 5μφ, so in the former case, it is possible to focus light down to 1μφ.
W, the latter can oscillate with a pumping light intensity of about 2On+W, and a semiconductor Raman laser that oscillates when excited by a normal CW output injection type semiconductor laser can be obtained. A portion of the Stokes light to be amplified is lost through the entrance window, but the stripe width is selected with respect to the entrance window and with respect to the laser length so that the proportion of the collimated beam reaching the window due to diffraction is sufficiently small. In other words, although the input 111y/l is smaller than the end face of the active layer, it still causes loss of Stokes light. In order to reduce this, the first step is to make the stripe width large enough to reduce the bending loss compared to the size of the entrance window, and to the extent that absorption loss due to multiple reflections of the excitation light inside the active layer does not become a problem. must be designed to be small. This diffraction effect depends on the length of the laser. As a further improved method, the active layer stripes may be sloped as shown in FIG. 4 to reduce diffraction loss, or an excitation light introduction layer may be provided separately as shown in FIG. 5. The refractive index of the Raman active layer cladding layer and the excitation light introduction layer are n, , n, respectively.
,,n, and if nt>n, >n, the excitation light spreads to the Raman active layer, but is confined by the cladding layer, while the Stokes light is confined only within the active layer and enters the entrance window. Diffraction loss can be eliminated.

(Function) 1y+ As mentioned above, both the excitation light and Stokes light focused on the entrance window are confined within the stripe, and the excitation light is multiple-reflected, so that loss due to absorption does not become a problem, and the entire active layer is covered by the entrance window. The power density of the excitation light reaches the same level as the power density of the excitation light in the CW, and the loss of the Stokes light to be oscillated through the entrance window is virtually no problem, making it possible to oscillate with an extremely low CW excitation light power. .

The semiconductor Raman laser thus formed is extremely suitable for optical heterogain detection and far-infrared light generation, which have already been proposed. Furthermore, since the semiconductor Raman laser of the present invention is entirely composed of semiconductors, including the excitation laser, it has a long life, high reliability, and is compact. Furthermore, since the semiconductor Raman laser of the present invention is composed of an optical resonator with extremely low reflection, absorption, and diffraction losses, that is, with a significantly high Q, the oscillation frequency accommodation is higher than that of an injection type semiconductor laser. It's extremely expensive. In other words, the Q value is about 70 times that of a semiconductor injection laser compared to the end face reflectance, and the frequency purity is correspondingly higher. Also, the half-value width of Raman scattering is the gain band, which is about 30 GHz in the case of GaP, and when the diameter of the Daiken laser is about 1 μ, the corresponding spectral width is 0.
It is about 5A. In other words, it does not matter if the excitation light has this degree of fluctuation. On the other hand, the longitudinal mode spacing determined from the cavity length of the semiconductor Raman laser is 15 GHz (length 4+u
e), the longitudinal mode is a single mode.

In addition, since Raman scattering is not an absorption phenomenon, internal heat generation can be ignored, and there is no frequency instability due to heat as in injection lasers. Therefore, it has extremely high frequency stability as a carrier light wave for optical communications, making it ideal for super-multiplex communications.

(Example) The present invention will be described in detail below based on the example shown in the drawings. - (Example 1) FIG. 2 shows a semiconductor Raman laser showing an example of the present invention. In this figure, 1 is a GaP crystal substrate, 2 is an At
xGa, -XP first cladding layer, 3 is a GaP Raman active layer formed in a stripe shape, 4 is an AlXGa, -XP second cladding layer covering the top and side surfaces of the Raman active layer, 9 is a GaP buffer It is a layer. 5 is a dielectric high reflection film on the incident side and has a reflectance of 99% or more.

Reference numeral 8 denotes an excitation light entrance window, which is an extremely small hole made in a part of the end face of the Raman active layer using photoresist technology. 6 is a high reflectance film on the output side and has a reflectance of 99%, which is also a high value.

Reference numeral 7 denotes excitation laser light that is focused by a lens or optical fiber and enters the stripe region from 8.

The length of the Raman active layer is 3m1Il~4mm, and the thickness of each layer is A, L, G, 11, -, P Clara 11M1~5.
The stripe width of the GaP Raman active layer is about 100 μm to 10 μm. The window for incident light is formed by removing the dielectric film deposited on the entire end face in stripes as shown in FIG. If the thickness of the GaP Raman active layer is 5 μm and the width of the entrance window is 5 μm, the area of the entrance window is 25 μm, and unless the light that enters through this is lost due to absorption or residual transmittance of the dielectric, it will not be absorbed into the active layer. Multiple reflections are repeated until the overall intensity is approximately the same.

Then, the threshold power density for GaP is approximately I x 10 W/C, or 10 mW per 1μ square, so the required pumping light power is 250 mW, which is C with an output of 1 W or less.
It can be excited by a W dye laser, a wavelength of 0.6 to 1.degree. 1 .mu.m, and a high output injection type CW semiconductor laser (wavelength of 0.8 to 0.85 .mu.m). The stripe width of the active layer at this time varies depending on the carrier density of the active layer. In other words, the semiconductor Raman lede can only oscillate in an extremely transparent medium such as GaP, but even then there is residual absorption that is approximately proportional to the carrier density, and in the case of GaP, 1
In the μ band, the absorption coefficient α is α, where N is the carrier density.
/N=Q.

05cm /1'Ocm'! It is Iia. Therefore, when the carrier density is lX10c+e and the resonator length is 411IIll, the intensity will be cut to about 1/2 by absorption after about 17 reciprocations. Therefore, the stripe width is approximately 17 times the width of the 5μ hole, or 8
If it is set to about 5μ, the intensity of the excitation light inside the active layer will be about the same as the intensity of the excitation light focused at the entrance window. If the stripe width is made larger than this, the light intensity inside the active layer decreases and the incident power must be increased, which is not preferable. Furthermore, if the stripe width is made smaller than this, the conversion efficiency from excitation light to Stokes light will decrease accordingly. Furthermore, the intensity distribution of the Stokes light is as shown in FIG. 3, and it decreases on the entrance window side, but since the Stokes light is lost from the entrance window due to diffraction, the narrower the stripe width, the more the diffraction loss increases. That is, the stripe width is 81 wavelengths λ,
If the refractive index is n and the resonator length is, then the spread width δ of the Stokes light towards the entrance window due to diffraction can be estimated by approximately λ~1μ, n
-1, 1, 1 to 4 mm, it becomes δ)7°6 μm.

That is, this portion reaches the entrance window through diffraction and exits from there, resulting in a loss.

Therefore, this diffraction loss is given by l) = δ/S and is approximately 9%, which cannot be ignored compared to the sum of the loss due to the reflective film, R, and absorption loss, which is approximately 6%. The above diffraction loss is an approximation by Fraunhoof diffraction, and since it is actually Fresnel diffraction, the diffraction loss is reduced to 2 to 3%. However, if the stripe width is made narrower, the breakage loss obviously increases, and it can no longer be ignored and the incident power must be increased.

When the carrier density of the active layer is 2×10 cn+−3, the stripe width of the active layer must be 1/2 of the above value, that is, about 40 μm or less.

Therefore, the diffraction loss increases, and the intensity required as the excitation light must be greater than 25 mW.

The difference in refractive index between ALP and GaP is not precisely known, but ALP has a smaller refractive index, approximately 80≦0.
In the above example, A L, Ga t-
The value of χ of z pH is suitably between 0.3 and 0.07, and X->Q, l is optimal. Methods for forming stripes, especially stripes with smooth sides, are summarized at the end of the examples.

(Example 2) In Example 1, the Stokes light is a nearly parallel beam, but the intensity and intensity decrease at the entrance window part, resulting in an intensity distribution as shown in Figure 3. Since the light reaches the window, the loss from the entrance window becomes large. Therefore, the shape of the Mann active layer stripe viewed from above is made to have an inclination toward the entrance window as shown in Figure 4, thereby reducing the effect of Stokes light entering the window and reaching the window through diffraction and being lost to the outside. let In other words, Stokes light travels along the dotted line in the figure. As in Example 1, when the laser length is 4IllI11, the width S of the active layer stripe is 85μ, and the width W of the entrance window is 5μ,
If the length L' of the sloped part is 100μ and the stripe width on the incident side is 90μ, the loss due to diffraction at the part L' can be reduced to 1/40 or less by replacing the equation (1) with L'.
It can be completely ignored. If L' is reduced by a small amount, the diffraction loss will be reduced accordingly, but since the angle of incidence of the incident light will increase, the loss due to reflection on the stripe side will increase, and in extreme cases, there will be a confinement effect due to the difference in refractive index for the excitation light. is lost. However, since the incident light focused by a lens etc. spreads inside the crystal, the inclination angle θ is about the same as the spread angle.
8-! W/J,' may also be included. As an example of type Ill, the spread angle of the excitation light inside the crystal is 0.0.
1, and the slope L' is 50 uni! 'It is allowed up to a degree.

(Example 3) FIG. 5(a) is a top view, and FIG. 5(b) is an end view as seen from the incident side. The Raman active layer 3 is in the form of a stripe with a width S, and on top of it is a stripe with a width S' equal to or less than the active layer width.
The excitation light introduction layer 10 having the following properties is formed. The latter is AAGaP
This is the y 1-7 layer, which is χ×χ.

If the refractive indices of the active layer, cladding layer, and excitation light introduction layer are nl, n, and n, respectively, then n, >n, and >n2,
The excitation light spreads throughout the active layer, but the Stokes light is confined only to the active layer. The 〇'l window is the end face 10 itself in the excitation light introduction diagram in Fig. 1(b). This is formed by removing a portion 11 of the deposited film 5 by photolithography.

Examples are χ-0,15, Y-0,1, Ai.

Ga, -y P layer width 2μ, thickness 2μ, active layer width 2
It is 0μ to 50μ. The length of the excitation light introduction layer does not need to span the entire length of the resonator. That is, since the incident light that enters through the entrance window spreads over the entire layer including the active layer, a structure may be adopted in which the light narrows in the middle and disappears as shown in FIG. Since the area of the entrance window is 4 μm 1, it can be excited with an injection type CW semiconductor laser of about 40 mW.

(Embodiment 4) The stripe of the active layer is narrowed through a tapered part in most of the middle part of the total length of 4 cm, and only the vicinity of the input end and the output end have the dimensions and structure shown in Example 2. In the top view of the stripe shown in FIG. 7, if the stripe width in the wide part is S1, the average stripe width in the narrow part is 8z, and the length is L, then the excitation light intensity in the narrow part, and therefore the gain, is 8. /82 times, so compared to the third embodiment, the excitation light power can be reduced by the factor given by. L-4nu
e, 2-3mm, S, -85μ, S 2 =
If the light is 10 μ, this factor becomes 0.15, and the required excitation light power is 38 mW compared to 250 mW in Example 2.
W will do.

The shape of the taper may be arbitrary as long as it is gentle. As a typical example, the length of each tapered portion is 250 μm1
The length of each wide portion is also 250 μm.

(Example 5) It goes without saying that stripes having a completely similar structure are also possible in the structure of Example 3. For example, in Example 3, if the width S of the active layer is 20μm and the narrow part is the same width as the excitation light introduction layer, that is, 5t−2μm, the factor ``is 0.13, and the excitation light power is changed from 40mW. 5mW
can be reduced to. By narrowing the stripes by the tapered portions in this way, not only can the power of the excitation light be lowered, but also the intensity of the light near the end facets can be lowered, thereby making it possible to avoid the surface breakdown phenomenon caused by the strong electric field of the light.

(Embodiment 6) In the above, the width of the stripe is narrowed, but of course the thickness may be gradually narrowed in the portions excluding the vicinity of the end face, as shown in FIG. FIG. 8 is a side view and is schematically shown with its length shortened. Moreover, both may be performed simultaneously. For example, in Example 4, the thickness dl near the end face is 5 μ, the average thickness d near the center is 1 μ, and the length of that portion is 3 μ.
If the total length of Il111 is 4+es, the factor f is (2
), it is 0°25, and the required excitation power is 9.5mW from 38111W in Example 4.
decreases to .

The manufacturing method of Example 1.2.4 is as follows.

Impurity-free ALxGa, -xP-th cladding layer and GaP were grown on a GaP substrate by vA differential liquid phase growth.
Grow the active layer. Next, a striped resist film is formed on the GaP active layer by lithography, and using this as a mask, reactive ion etching (RIE) is performed using PCl gas.

PCl, a gas, is particularly effective for GaP. This creates a nearly vertical and smooth side surface.

Since the gain of the Raman laser is small, even if there is a few percent scuffle loss on the side, the threshold will decisively increase, but PCl,
When etched with a gas, the side surfaces become sufficiently smooth and scattering loss does not substantially occur.

The active layer stripes are relatively deep, from 2 to 10 microns, and must be etched nearly vertically to prevent light scattering. PCJL, the gas is introduced into a vacuum chamber with counter electrodes and 3.2 M at a pressure of approximately 0.05 Torr.
Hz, by applying high frequency power of 1200W.
Etching can be performed at a high speed of μIIl/win. An ordinary positive photoresist may be used as the etching mask. After the stripe formation process, A
A Ga, -xP second cladding layer and a GaP buffer layer are grown.

In the case of Example 3.5, first the first cladding layer, the active layer, and the excitation light introduction layer are grown. Next, using a mask for the excitation light introduction layer, stripes are formed only in this layer by the above-mentioned RIE process.

Next, using a mask for the active layer, a resist group is applied to the width of the active layer, and the same RIE process is performed again. After that, a second cladding layer and seven buff layers are grown.

In Example 6, in which the thickness is gradually changed, after the active layer is grown, a resist film having an opening near the center of the stripe (3+u++ length part) is used, and this part is etched to a depth of 1 μm by RIE process.Next A new resist film with openings about 100μ wide on both sides, that is, an opening width of 3.2mg, is formed and etched again by 1μ.If the openings are expanded in the same way and the RIE process is performed four times, the thickness of the narrow part becomes 1μ. The length of each tapered part is 3
It becomes 00μ. The second RIE will smooth out the step caused by the first etching, but to make it even smoother, normal wet etching can be added at the end.

〔Effect of the invention〕

The semiconductor Raman laser of the present invention has the above-described configuration, and operates with a CW excitation laser having a much lower power than conventional ones, that is, 250 mW to 10 mW or less. . In particular, since a CW fishing injection type semiconductor laser is operated as an excitation laser, a CW fishing injection semiconductor Raman laser is provided which is composed entirely of semiconductors including the excitation source. Since it does not require a discharge tube or a large Ti source, it has an extremely long life, high reliability, and is compact. The present inventor has already proposed that semiconductor Raman lasers can be used for optical heterodyne detection, far-infrared light wave generation, and carrier light wave generation with extremely high wavelength purity stability, but the CW fishing operation is relatively low. Since it operates with a power injection type semiconductor laser, current injection type semiconductor lasers for optical communication can be used as is, and it provides a means for ultra-wideband optical communication, that is, ultra-multiplexing. Lol! is figure s6ae

Claims (5)

[Claims] (1) A first cladding layer of Al_xGa_1_-_xP formed by epitaxial growth on a GaP substrate crystal, a Raman active layer of GaP formed in a stripe shape on the first cladding layer, and a Raman active layer formed on the Raman active layer. Al_xG surrounding the Raman active layer and its sides
A dielectric film having a high reflectance is formed on two opposing end faces of the a_1_-_xP second cladding layer, and the focused excitation light is introduced into a part of the dielectric film into the Raman active layer. The entrance window is opened, and the excitation light incident from the entrance window is confined inside the cladding layer and undergoes multiple reflections, so that its intensity almost reaches the intensity at the entrance window, and the loss of Stokes light from the entrance window A semiconductor Raman laser characterized in that the area of the entrance window is designed to be small so that the area of the incident window is sufficiently small. (2) The semiconductor according to claim 1, wherein the stripes of the Raman active layer have an inclination near the entrance window to reduce loss of Stokes light from the entrance window. Raman laser. (3) An excitation light introduction layer Al_yGa_1_-_yP layer is provided inside the cladding layer in contact with the Raman active layer, and the refractive index of the Raman active layer and the cladding excitation light introduction layer are n_1, n_2, and n_3, respectively. When n_1>n
The semiconductor Raman laser according to claim 1, characterized in that _3>n_2. (4) The semiconductor Raman laser according to any one of claims 1 to 3, wherein the stripe of the Raman active layer is at least partially narrowed through a tapered portion. (5) The semiconductor Raman laser according to any one of claims 1 to 4, wherein the stripes of the Raman active layer are formed by etching with PCl_3 gas.