JP2015109365A - Semiconductor optical amplifier - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a semiconductor optical amplifier (SOA) capable of optical amplification of polarization independence across a wide gain band in a wavelength range for performing optical communication.SOLUTION: A semiconductor optical amplifier comprises: an active layer 14 of a quantum well which has barrier layers 142 and composite quantum well layers 141 each composed of a plurality of semiconductor layers and in which the barrier layers 142 and the composite quantum well layers 141 are alternately stacked. The plurality of semiconductor layers of the composite quantum well layer 141 are composed of at least one layer of a quantum level control layer 143 for controlling a quantum level in the composite quantum well layer 141 and at least one layer of main quantum well layer 144 having bandgap smaller than that of the quantum level control layer 143. At least one layer of the quantum level control layer 143 and the main quantum well layer 144 has tensile strain against the substrate.

Description

  The present invention relates to a semiconductor optical amplifier, and more particularly to a semiconductor optical amplifier in which an active layer has a strained quantum well structure.

  2. Description of the Related Art Conventionally, a semiconductor optical amplifier (SOA) used in an optical communication system is known that includes an active layer having a quantum well structure including a well layer and a barrier layer. Since such a quantum well active layer has a reduced density of states compared to an active layer made of a bulk crystal without quantum confinement, the wavelength band of optical gain is wide. In the quantum well active layer, the heavy hole band and the light hole band are separated from each other due to the quantum confinement effect.

  Here, when the entire quantum well active layer is lattice-matched with the substrate, the ground quantum level of heavy holes is higher than that of light holes (when viewed in terms of electron energy). Because they exist, the holes are concentrated at the ground level of heavy holes. At the band edge of this heavy hole, the square of the absolute value of the dipole moment matrix element responds only to the TE mode, and therefore the TE mode gain coefficient is larger than the TM mode gain coefficient.

  In addition, when the crystal growth direction is the vertical direction, since the quantum well active layer has a larger width in the horizontal direction than the thickness in the vertical direction, the optical confinement factor is larger in the TE mode than in the TM mode. The TE mode optical gain will exceed the TM mode optical gain.

  However, in general, the state of the optical fiber and various optical elements constituting the optical communication system changes every moment due to external factors (environmental temperature, external force, etc.), and the optical signal transmitted from the optical communication system due to this changes. The polarization direction of is changing every moment. For this reason, an SOA used in an optical communication system is required to be independent of polarization in order to always realize a constant optical gain for an optical signal in an arbitrary polarization direction.

  Therefore, in order to realize polarization independence, by applying an extension strain to the well layer of the quantum well active layer, the heavy hole band and the light hole band separated by the quantum confinement effect are corrected, and the TE mode and the TM mode are corrected. An SOA in which the optical gains are matched is proposed (see, for example, Non-Patent Document 1).

  On the other hand, a strained multiple quantum well (MQW) structure that realizes polarization independence in the optical wavelength band of 1550 nm by applying an extension strain to the barrier layer of the quantum well active layer has also been proposed (for example, Non-patent document 2).

  In Non-Patent Document 2, by applying an extension strain to the barrier layer, a structure in which the ground quantum level of a light hole is shifted to a higher energy side than the ground quantum level of a heavy hole over the entire MQW structure. It has become. With this structure, TM mode gain is improved, and excellent polarization-independent amplification operation characteristics are realized.

P. Koonath, Sangin Kim, Woon-Jo Cho, A. Gopinath, "Polarization-insensitive quantum-well semiconductor optical amplifiers", IEEE Journal of Quantum Electronics, Volume 38, Issue 9, Sep 2002, p. 1282 -1290 K. Magari, M. Okamoto, Y. Noguchi, "1.55 μm polarization-insensitive high-gain tensile-strained-barrier MQW optical amplifier", IEEE Photonics Technology Letters, Volume 3, Issue11, Nov 1991, p. 998 -1000 J. Barrau, B. Brousseau, M. Brousseau, R. J. Simes and L. Goldstein, "Novel principle of confinement in quantum-well structures", Electronics Letters, Volume 28, Number 8, pp.786-788, 1992 C. P. Seltzer, S. D. Perrin, M. C. Tatham, D. M. Cooper, "Zero-net-strain multiquantum well lasers", Electronics Letters, Volume 27, Number 14, pp.1268-1270, 1991

  However, when a polarization-independent SOA is realized by uniformly applying a tensile strain to the well layer as in the configuration disclosed in Non-Patent Document 1, the heterogeneity of electrons is increased by increasing the tensile strain as described below. Non-Patent Document 3 discloses that the problem of a reduction in barriers occurs.

  An effective heterobarrier for electrons in the conduction band is a heterobarrier height ΔE which is an energy difference between the ground quantum level of electrons in the well layer 541 and the conduction band edge of the barrier layer 542 shown in FIG. As shown in FIGS. 14A to 14B, when the extension strain of the well layer 541 is increased, the hetero barrier height ΔE is decreased, and electrons overflow from the well layer 541 to the barrier layer 542 and the cladding layer. In addition, a wide gain spectrum cannot be obtained, and it is difficult to reduce the noise figure (NF).

  In Non-Patent Document 3, it is claimed that electrons are spatially localized due to the electrostatic shielding phenomenon and a quantum level is generated in the conduction band. However, it is difficult to design a desired SOA gain wavelength band under such a phenomenon. In addition, for example, when the signal light intensity incident on the SOA changes greatly at high speed, the carrier density in the SOA changes at high speed, so that the quantum level formed by the electrostatic shielding phenomenon also changes, and the signal light There may be a problem such as a decrease in gain.

  In addition, as in the configuration disclosed in Non-Patent Document 2, there is the following problem regarding the configuration in which the tensile strain is applied to the barrier layer. In the case of a 1310 nm band super luminescent diode (SLD), when the barrier layer having a tensile strain of 1% or more is formed over 10 nm or more, the applicant of the present application We observe the phenomenon that the flatness of the heterointerface is lost. Therefore, it is presumed that applying the extension strain as described above to the barrier layer leads to deterioration of optical gain characteristics and difficulty in manufacturing in a 1310 nm band SOA, for example.

  The present invention has been made to solve such a conventional problem, and is a semiconductor optical amplifier capable of polarization-independent optical amplification over a wide gain band in a wavelength band for optical communication ( The object is to provide SOA).

  In order to solve the above problems, a semiconductor optical amplifier according to claim 1 of the present invention has a barrier layer and a composite quantum well layer composed of a plurality of semiconductor layers, and the barrier layer and the composite quantum well layer are alternately arranged. In the semiconductor optical amplifier in which the quantum well active layer formed on the substrate is formed on the substrate, the plurality of semiconductor layers of the composite quantum well layer include at least one for controlling a quantum level in the composite quantum well layer. Each of the quantum level control layer and at least one main quantum well layer having a smaller band gap than the quantum level control layer, and at least one of the quantum level control layer and the main quantum well layer. The layer has a configuration having a tensile strain with respect to the substrate.

  Here, the band gap refers to an energy difference between the valence band edge having the highest energy in the valence band and the conduction band edge in terms of electron energy.

  With this configuration, the valence band structure of the layer with tensile strain in the composite quantum well layer shifts the quantum level of light holes to the high energy side, so that the quantum levels of heavy and light holes can be brought closer. it can.

  In such a valence band structure, the effective mass in the in-plane direction of the quantum well increases the density of states because light holes are larger than heavy holes, and the TE mode optical confinement factor as described above is superior to the TM mode. It acts so as to cancel out, and the optical gain of the TE mode and the TM mode can be matched to realize polarization independence.

  In addition, since the band gap of the main quantum well layer is the smallest in the composite quantum well, the main quantum well layer generates a region having a deep potential in the composite quantum well, thereby reducing the electron quantum level energy. For this reason, since the effective hetero barrier height for electrons in the conduction band can be sufficiently increased to reduce the overflow of electrons, the gain band can be expanded.

  The semiconductor optical amplifier according to claim 2 of the present invention has a configuration in which the plurality of semiconductor layers of the composite quantum well layer include one main quantum well layer and one quantum level control layer. ing.

  With this configuration, since the number of layers constituting the composite quantum well layer in the semiconductor optical amplifier of the present invention is minimized, the productivity is improved.

  In the semiconductor optical amplifier according to claim 3 of the present invention, the plurality of semiconductor layers of the composite quantum well layer include a plurality of quantum level control layers and at least one main quantum well layer. At least two or more quantum level control layers are stacked so as to sandwich at least one or more main quantum well layers from both sides.

  With this configuration, the band gap and the amount of strain of each of the plurality of quantum level control layers and the main quantum well layer can be set independently, so that the wavelength range and gain band of the optical gain of the semiconductor optical amplifier of the present invention can be designed. The degree of freedom can be expanded.

  In addition, in this configuration, the conduction band and valence band potential can be made symmetric in the composite quantum well layer, so that the overlap integral value of the envelope wave function between the electron and heavy holes and the envelope between the electron and light holes can be obtained. Since the overlap integral value of the wave function is improved, recombination efficiency can be increased, and further improvement in optical gain can be realized.

  In the semiconductor optical amplifier according to claim 4 of the present invention, the plurality of semiconductor layers of the composite quantum well layer include a plurality of quantum level control layers and at least one or more main quantum well layers. In addition, at least one main quantum well layer is stacked between the barrier layer and the plurality of quantum level control layers.

  In the semiconductor optical amplifier according to claim 5 of the present invention, the main quantum well layer has a configuration having compressive strain with respect to the substrate.

  In this configuration, a desired quantum level can be formed for electrons, light holes, and heavy holes by increasing the amount of tensile strain applied to the quantum level control layer.

  That is, the main quantum well layer having compressive strain increases the quantum level energy of heavy holes and decreases the quantum level of light holes. On the other hand, a quantum level control layer having an extensional strain lowers the quantum level energy of heavy holes and increases the quantum level of light holes. In addition, because of the confinement effect of holes due to the quantum well potential, the quantum level of light holes is greatly reduced compared to heavy holes. By increasing the tensile strain applied to the quantum level control layer, And the ground quantum level of light holes can be brought close to the desired spacing. A polarization independent operation SOA can be realized by an active layer having such quantum levels.

  The semiconductor optical amplifier according to claim 6 of the present invention has a configuration in which the main quantum well layer is unstrained with respect to the substrate.

  In this configuration, the absolute value of the amount of extension strain of the quantum level control layer can be set to less than 1% by setting the layer thickness of the quantum level control layer having extension strain to the critical thickness or less. Thereby, since stress at the interface of each layer constituting the composite quantum well can be reduced, generation of crystal defects can be suppressed, and desired quantum levels can be formed with respect to electrons, light holes, and heavy holes.

  The semiconductor optical amplifier according to claim 7 of the present invention has a configuration in which the main quantum well layer has a tensile strain with respect to the substrate.

  In this configuration, a desired quantum level can be formed with respect to electrons, light holes, and heavy holes by applying compressive strain to the quantum level control layer.

  That is, the main quantum well layer having an extension strain increases the quantum level energy of light holes and decreases the quantum level of heavy holes. On the other hand, a quantum level control layer having compressive strain decreases the quantum level energy of light holes and increases the quantum level of heavy holes. In addition, the effect of confining holes by the quantum well potential is that the quantum level of light holes is significantly lower than that of heavy holes. Therefore, by applying a relatively small compressive strain to the quantum level control layer, The ground quantum level of a light hole can be brought close to a desired interval. A polarization independent operation SOA can be realized by an active layer having such quantum levels.

  The semiconductor optical amplifier according to claim 8 of the present invention has a configuration in which the quantum level control layer has a tensile strain with respect to the substrate.

  In this configuration, a desired quantum level can be formed with respect to electrons, light holes, and heavy holes by applying an extension strain different from that of the main quantum well layer to the quantum level control layer.

  That is, the main quantum well layer having an extension strain increases the quantum level energy of light holes and decreases the quantum level of heavy holes. On the other hand, the quantum level control layer having an extension strain also raises the quantum level of light holes and lowers the quantum level energy of heavy holes. In addition, because of the confinement effect of holes due to the potential of the barrier layer, the main quantum well layer, and the quantum level control layer, the quantum level of light holes is significantly lower than that of heavy holes. By appropriately setting the extension strain applied to the level control layer, the base quantum levels of heavy holes and light holes can be brought close to a desired interval. A polarization independent operation SOA can be realized by an active layer having such quantum levels.

  Also, with such a layer configuration, for example, the strain distortion of the quantum level control layer can be reduced as compared with the configuration described in claim 6, so that the average strain amount of the composite quantum well layer can be suppressed. Thereby, the crystallinity of the active layer having the composite quantum well layer can be improved.

  In the semiconductor optical amplifier according to claim 9 of the present invention, the composite quantum well layer is made of InGaAsP which is a III-V group mixed crystal.

  With this configuration, by adjusting the composition of the InGaAsP mixed crystal, it is possible to form a potential structure in which the light hole band edge and the heavy hole band edge are shifted to a desired energy level as a semiconductor optical amplifier in the optical communication wavelength band. .

  In the present invention, a tensile strain is applied to at least one of a plurality of semiconductor layers constituting a composite quantum well layer structure, so that polarization is independent over a wide gain band in a wavelength band for optical communication. A semiconductor optical amplifier capable of optical amplification is provided.

It is a perspective view which shows the structure of the semiconductor optical amplifier as embodiment of this invention. It is an expanded sectional view showing the composition of the principal part of the semiconductor optical amplifier as a 1st embodiment of the present invention. 1 is a schematic diagram showing an example of a band structure of a semiconductor optical amplifier as a first embodiment of the present invention. It is a schematic diagram which expands and shows the conduction band side of the band structure of FIG. It is a schematic diagram which expands and shows the valence band side of the band structure of FIG. It is an expanded sectional view which shows the structure of the principal part of the semiconductor optical amplifier as 2nd-5th embodiment of this invention. It is a schematic diagram which shows an example of the band structure of the semiconductor optical amplifier as the 2nd Embodiment of this invention. It is a schematic diagram which shows an example of the band structure of the semiconductor optical amplifier as the 3rd Embodiment of this invention. It is a schematic diagram which shows an example of the band structure of the semiconductor optical amplifier as the 4th Embodiment of this invention. It is a figure which shows the structural conditions of the semiconductor optical amplifier of this invention. It is a schematic diagram which shows an example of the band structure of the semiconductor optical amplifier as the 5th Embodiment of this invention. It is a graph which shows an example of the optical gain characteristic with respect to the drive current of the semiconductor optical amplifier of this invention. It is a schematic diagram which shows the structural example of the distortion compensation structure in the semiconductor optical amplifier of this invention. It is a schematic diagram explaining the relationship between the effective hetero barrier with respect to the electron of the conduction band of a quantum well structure, and an extension strain.

Hereinafter, embodiments of a semiconductor optical amplifier (SOA) according to the present invention will be described with reference to the drawings. In addition, the dimensional ratio of each structure on each drawing does not necessarily correspond with the actual dimensional ratio. In addition, in the band structure diagrams shown in FIGS. 3 to 5, 7 to 9 and 11, the conduction band edge (E _c ) is represented by a thick solid line, and the heavy hole band edge (E _hh ) and the light hole band edge on the valence band side. (E _lh ) is represented by a thick solid line and a thick broken line, respectively.

(First embodiment)
First, the configuration of the SOA 1 as the first embodiment of the present invention will be described. FIG. 1 is a perspective view showing the configuration of the SOA 1, and FIG. 2 is an enlarged cross-sectional view showing the structure of the main part of the SOA 1.

  That is, as shown in FIG. 1, the SOA 1 includes, for example, an n-type semiconductor substrate 11 made of n-type InP (indium phosphorus) and an n-type cladding made of n-type InP formed on the n-type semiconductor substrate 11. A layer 12, an optical separation and confinement (SCH) layer 13 made of InGaAsP (indium, gallium, arsenic, phosphorus) formed on the n-type cladding layer 12, and the SCH layer 13. An active layer 14 made of InGaAsP, an SCH layer 15 made of InGaAsP formed on the active layer 14, and a p-type cladding layer 18 made of p-type InP formed on the SCH layer 15 are provided. In the configuration in which the semiconductor substrate 11 is p-type InP, the n-type cladding layer 12 may be replaced with p-type InP, and the p-type cladding layer 18 may be replaced with n-type InP.

  Note that the n-type cladding layer 12, the SCH layer 13, the active layer 14, the SCH layer 15, and the p-type cladding layer 18 constitute a mesa optical waveguide, and p is formed on both sides of the mesa optical waveguide. A lower buried layer 16 made of type InP and an upper buried layer 17 made of n type InP are formed.

  Further, when the direction in which the fundamental transverse mode light propagates in the mesa optical waveguide is the main axis direction of the waveguide, the normal direction of the end face is reduced with the purpose of reducing the light returning from the SOA end face to the mesa optical waveguide. The mesa-type optical waveguide configuration in which the direction of the main axis is tilted, or the mesa-type optical waveguide in the vicinity of the end face is given a radius of curvature so that the optical waveguide loss is sufficiently small in the horizontal plane, and the guided light is incident on the end face obliquely. It is good also as composition to do.

  A p-type cladding layer 18 made of p-type InP is formed on the upper side of the SCH layer 15 and on the upper surface of the upper buried layer 17. A p-type contact layer 19 made of p-type InGaAsP is formed on the upper surface of the p-type cladding layer 18. Is formed. Further, a p-type metal electrode 20 is provided on the upper surface of the p-type contact layer 19. An n-type metal electrode 21 is provided on the lower surface of the n-type semiconductor substrate 11.

  Further, an antireflection film (not shown) is applied to each of the light incident end face and the light exit end face formed by cleaving the SOA 1.

  As shown in the enlarged sectional view of FIG. 2, the active layer 14 has a multiple quantum well (MQW) structure in which a composite quantum well layer 141 and a barrier layer 142 made of a plurality of semiconductor layers are repeatedly and alternately stacked. Have Alternatively, the active layer 14 may have a single quantum well (SQW) structure in which one composite quantum well layer 141 is sandwiched between two barrier layers 142. In addition, the active layer 14 may have an SQW structure in which the composite quantum well layer 141 is directly sandwiched between the SCH layer 13 and the SCH layer 15 without using the barrier layer 142.

  The composite quantum well layer 141 includes a quantum level control layer (strain application layer) 143 that has an extensional strain with respect to the n-type semiconductor substrate 11, and a main quantum well layer 144 that has a smaller band gap than the quantum level control layer 143. ,including.

  The barrier layer 142 has a larger band gap than the quantum level control layer 143 and the main quantum well layer 144. The band gap is the smallest energy difference among the energy differences between the conduction band edge and the valence band edge unless otherwise specified.

  Between the two barrier layers 142, one pair in which the quantum level control layer 143 and the main quantum well layer 144 constituting the composite quantum well layer 141 are stacked adjacent to each other is disposed. The main quantum well layer 144 is unstrained or nearly unstrained with respect to the n-type semiconductor substrate 11. Alternatively, the main quantum well layer 144 may have an extension strain or a compression strain with respect to the n-type semiconductor substrate 11.

  As illustrated in FIG. 2, the barrier layer 142, the main quantum well layer 144, the quantum level control layer 143, and the barrier layer 142 are stacked in this order from the n-type cladding layer 12 side. Alternatively, the layers may be stacked in the order of the barrier layer 142, the quantum level control layer 143, the main quantum well layer 144, and the barrier layer 142 from the n-type cladding layer 12 side.

  Further, the SCH layer 13 made of InGaAsP has a band gap in the range between the n-type cladding layer 12 and the barrier layer 142. Similarly, the band gap of the SCH layer 15 is in the range between the p-type cladding layer 18 and the barrier layer 142.

  In the SOA 1 configured as described above, when incident light propagates through the active layer 14 with a drive current injected between the p-type metal electrode 20 and the n-type metal electrode 21 in the forward direction, the composite quantum well The TM mode optical gain is mainly obtained by stimulated emission by recombination of electrons and light holes in the layer 141. At the same time, the TE mode optical gain is mainly obtained by stimulated emission by recombination of electrons and heavy holes in the composite quantum well layer 141.

  FIG. 3 is a diagram schematically showing the main part of the band structure of SOA1. In FIG. 3, the band structure is shown so that the upper part has higher energy for electrons.

FIG. 3 shows an example of a band structure in which the quantum level control layer 143 is designed to have a layer thickness larger than that of the main quantum well layer 144. In the quantum level control layer 143, the degeneration of the light hole band edge E _lh and the heavy hole band edge E _hh of the valence band is solved by the extension strain. Further, since the main quantum well layer 144 is _unstrained , the light hole band edge E _lh and the heavy hole band edge E _hh are degenerated in the main quantum well layer 144.

More specifically, the light hole band edge E _lh quantum level control layer 143 is shifted to the higher energy side than the valence band edge E _V of primary quantum well layer 144, heavy quantum level control layer 143 hole band end E _hh is shifted to the low energy side than the valence band edge E _V of primary quantum well layer 144. The main reason the quantum well layer 144 has a small band gap than the quantum level control layer 143, a quantum level higher energy side than the conduction band edge E _C of the control layer 143 the conduction band edge E _C is primary quantum well layer 144 in Exists.

  With the above configuration, light holes spread throughout the composite quantum well layer 141, so that the energy of the ground quantum level of the light holes is increased. On the other hand, since heavy holes tend to localize in the vicinity of the main quantum well layer 144, the energy of the ground quantum level of the heavy holes decreases.

Furthermore, since the light hole band edge E _lh is located on the higher energy side than the heavy hole band edge E _hh , the ground quantum level of the light hole shifts to the high energy side and becomes the ground quantum level of the heavy hole. Get closer. That is, it can be said that the quantum level control layer 143 having an extensional strain is controlled so that the base quantum level of light holes in the composite quantum well layer 141 is close to the base quantum level of heavy holes. As a result, the TE mode optical gain and the TM mode optical gain can be matched to achieve polarization independence.

  Such a configuration can be applied to, for example, InP / AlGaInAs, GaAs / InGaAsP, and GaAs / AlGaInP, which can produce a strained quantum well structure.

In particular, as is remarkable in the example of FIG. 3, since the shape of the potential for electrons and the potential for heavy holes is asymmetric in the composite quantum well layer 141, the existence probability density distribution of electrons and heavy holes is the composite quantum well layer 141. Biased from the center of Here, the existence probability density distribution of electrons (or holes) corresponds to the square of the absolute value of the wave function ψ (x) of electrons (or holes) | ψ (x) | ² .

  On the other hand, the potential for light holes is defined by the layer thickness of the quantum level control layer 143 whose rectangular width is sufficiently larger than the layer thickness of the main quantum well layer 144. The distribution is almost symmetric in the composite quantum well layer 141.

  By such potential design, the ground quantum levels of heavy holes and light holes can be brought close enough. In addition, the recombination rate between electrons and light holes whose main optical gain is in the TM mode is larger than the recombination rate between electrons and heavy holes whose main optical gain is in the TE mode.

  As described above, since the general slab type quantum well active layer is wide in the horizontal direction, the optical confinement coefficient in the TE mode is larger than that in the TM mode. It can also be expected that the problem of the TE mode optical gain becoming relatively large can be reduced. This facilitates the realization of optical gain polarization independence.

  Further, in the band structure shown in FIG. 3, the main quantum well layer 144 that constitutes a part of the composite quantum well layer 141 has a smaller amount of tensile strain than the quantum level control layer 143, so that it is uniform over the entire well layer. Compared to the conventional configuration in which an extension strain is applied, the ground quantum level of electrons in the conduction band of the composite quantum well layer 141 can be suppressed to a low level. As a result, it is possible to avoid that the hetero barrier height ΔE, which is the energy difference between the electron ground quantum level of the composite quantum well layer 141 and the conduction band edge of the barrier layer 142, is significantly reduced. Electron overflow to the barrier layer 142 can be suppressed.

  4 is an enlarged view showing the conduction band side of the band structure of FIG. In the following, the effective heterobarrier height ΔE ′ of the active layer in the conventional configuration in which a uniform extension strain is applied to the InGaAsP well layer described in Non-Patent Document 1, and the SOA 1 that is the configuration of the present embodiment. The result of comparing the effective hetero barrier height ΔE of the active layer 14 in FIG.

  As a result of the simulation, the hetero barrier height ΔE ′ in the conventional configuration composed of the InGaAsP / InP system shown in Non-Patent Document 1 is 30.8 meV, whereas the hetero barrier height in the SOA 1 of the present embodiment. ΔE was 53.6 meV, and it was confirmed that the SOA 1 of this embodiment has a high effect of preventing the overflow of electrons.

FIG. 5 is an enlarged view showing the valence band side of the band structure of FIG. In the band structure shown in FIG. 5, the light hole band edge E _lh has a convex shape. By such potential design, the ground level difference between the ground quantum level of heavy holes and the ground quantum level of light holes can be made narrower than the energy width of in-band relaxation (about 6.6 meV). It becomes possible to bring the two ground quantum levels close to a level at which polarization-independent amplification of signal light can be realized.

Hereinafter, an example of the manufacturing method of SOA1 which concerns on this invention is demonstrated.
First, an n-type cladding layer 12 made of n-type InP and an SCH layer 13 made of InGaAsP are epitaxially grown on an n-type semiconductor substrate 11 made of n-type InP using a metal organic chemical vapor deposition (MOVPE) method. . The band gap of the SCH layer 13 is in the range between the n-type cladding layer 12 and the barrier layer 142.

  Next, on the SCH layer 13, a barrier layer 142 made of InGaAsP, a main quantum well layer 144 made of InGaAsP that is unstrained or nearly unstrained with respect to the n-type semiconductor substrate 11, and stretched with respect to the n-type semiconductor substrate 11. The MQW structure or SQW structure active layer 14 is formed by stacking the strained quantum level control layer 143 made of InGaAsP and the barrier layer 142 made of InGaAsP in this order. Note that the distortion amount of the main quantum well layer 144 is not limited to the distortion of the n-type semiconductor substrate 11 as described above due to the gain wavelength band required for the semiconductor optical amplifier of the present invention. An extension strain can also be set.

  On the active layer 14 thus formed, the SCH layer 15 having a band gap in the range between the p-type cladding layer 18 and the barrier layer 142 is formed.

Next, a lower layer portion of a p-type cladding layer 18 made of p-type InP is grown on the SCH layer 15. Further, an insulating film made of SiN _x film or SiO ₂ film is laminated on the upper surface of the lower layer portion of the p-type cladding layer 18 by using a plasma CVD method, a resist is applied, and a striped mask pattern is exposed by photolithography. And develop.

  Using this striped mask pattern, both sides of the insulating film are removed by etching with hydrofluoric acid. Subsequently, the remaining resist is peeled and removed to form an insulating film etching mask to which the shape of the stripe-shaped mask pattern is transferred.

  Then, using the etching mask for the insulating film set as described above and an etchant composed of a mixture of hydrochloric acid, hydrogen peroxide solution, and water, the lower layer portion of the p-type cladding layer 18, the SCH layer 15, the active layer 14 The SCH layer 13 and the n-type cladding layer 12 are wet-etched to form mesa stripes.

  Next, the MOVPE method is used for the portion removed by the wet etching, and the lower buried layer 16 made of p-type InP and the upper buried layer 17 made of n-type InP are formed using the insulating film etching mask as a growth inhibition mask. Sequentially stack and embed.

  Next, the etching mask of the insulating film is removed with hydrofluoric acid, the upper surface of the mesa stripe is exposed, and a buried layer made of p-type InP having the same composition as the lower layer portion of the p-type cladding layer 18 is laminated to form a p-type cladding. The layer 18 is completed, and a p-type contact layer 19 made of p-type InGaAsP is stacked thereon by the MOVPE method.

  Then, a p-type metal electrode 20 is formed on the p-type contact layer 19 and an n-type metal electrode 21 is formed on the bottom surface of the n-type semiconductor substrate 11 by vapor deposition, and an alloy and plating process is performed to complete a semiconductor wafer.

  Next, the semiconductor wafer is separated into individual chips by cleaving or dicing at a predetermined position. Further, an antireflection film (not shown) is formed on the light incident end face and the light exit end face of the separated chip. Thus, the SOA 1 of the present embodiment is completed.

  As described above, the SOA 1 according to the present embodiment can realize the polarization independence by bringing the quantum levels of the heavy hole and the light hole close to each other to match the optical gains of the TE mode and the TM mode.

  In addition, the SOA 1 of the present embodiment can sufficiently increase the effective heterobarrier height ΔE related to the electrons in the conduction band and reduce the electrons leaking from the composite quantum well layer 141 to the barrier layer 142.

  Thereby, the reactive current when operating the semiconductor optical amplifier of the present invention can be suppressed. In addition, since the carrier density in the composite quantum well layer 141 can be increased, it is possible to improve the optical gain, expand the optical gain wavelength band, and reduce the noise figure (NF).

  In addition, when the SOA 1 of the present embodiment is applied to the 1310 nm wavelength band, the barrier layer 142 and the composite quantum well layer 141 are particularly configured by setting the barrier layer 142 to be unstrained with respect to the n-type semiconductor substrate 11. Since the flatness of the hetero interface can be sufficiently maintained, good optical gain characteristics and easy manufacturing can be realized. As a result, the SOA can be easily manufactured without deteriorating the optical gain characteristics. The 1310 nm band includes a wavelength of 1280 nm to 1340 nm.

  Further, the SOA 1 of the present embodiment effectively suppresses the amount of strain of the entire active layer 14 and effectively combines the quantum well layer when the main quantum well layer 144 is unstrained with respect to the n-type semiconductor substrate 11. A desired potential can be formed in the 141.

  In order to fabricate the SOA 1 of the present embodiment by simulating the relationship between the layer thickness of the quantum level control layer 143 and the strain amount while changing the band gap wavelength and the layer thickness of the main quantum well layer 144 as described below. The structural conditions of were investigated. The result is shown in FIG.

In FIG. 10, the horizontal axis represents the layer thickness (d _QLC ) of the quantum level control layer 143 expressed in nanometers (nm), and the vertical axis represents the strain of the quantum level control layer 143 expressed in percent (%) units. The quantity (ε _QLC ) is used.

For example, when the SOA 1 of the present embodiment is applied to the 1300 nm wavelength band, the hatched area in FIG. 10A, that is, the straight line f1 (d _QLC ) represented by the following expressions (1) and (2): It was found that SOA1 can be fabricated in a region sandwiched between f2 (d _QLC ) and −1.5% ≦ ε _QLC <0.
f1 (d _QLC ) = 0.24 × d _QLC −1.9 (1)
f2 (d _QLC ) = 0.24 × d _QLC −2.95 (2)

Equation (1) is a limit line of the gain wavelength band, and the gain wavelength band shifts from 1300 nm in a region where ε _QLC is larger and negative than this straight line. Further, in the region where ε _QLC is smaller and negative than the straight line represented by the formula (2) and in the region where ε _QLC <−1.5%, heterogeneity is present at the interface between the quantum level control layer 143 and the barrier layer 142. The barrier becomes a Type-II type, resulting in a band structure that hinders electron transport.

For example, when the SOA 1 of the present embodiment is applied to the 1500 nm wavelength band, the hatched area in FIG. 10B, that is, the straight line f3 (d) expressed by the following expressions (3) and (4) _QLC) and f4 _(in the region sandwiched _{d QLC),} and, under the condition of -1.5% ≦ ε _QLC <0, was found to be manufactured SOAl.
f3 ( _dQLC ) = 0.25 × _dQLC−2.0 (3)
f4 (d _QLC ) = 0.2 × d _QLC −4.0 (4)

Note that equation (3) becomes a limit line of the mixed crystal composition of the quantum level control layer 143, and the quantum level control layer cannot be grown in a region where ε _QLC is larger than this line and negative. On the other hand, in the region where ε _QLC is smaller and negative than the straight line represented by the equation (4), the layer thickness of the quantum level control layer 143 exceeds the critical film thickness, making it difficult to achieve good crystal growth.

In the region where ε _QLC <−1.5%, the _{heterobarrier} becomes a Type-II type at the interface between the quantum level control layer 143 and the barrier layer 142, and a band structure that hinders electron transport is formed.

(Second Embodiment)
Next, the SOA 2 as the second embodiment of the present invention will be described with reference to the drawings. Note that the description of the same configuration and operation as in the first embodiment will be omitted as appropriate.

  As shown in the enlarged cross-sectional view of FIG. 6, the active layer 24 has an MQW structure in which a composite quantum well layer 241 and a barrier layer 142 made of a plurality of semiconductor layers are alternately and repeatedly stacked. Alternatively, the active layer 24 may have an SQW structure in which one composite quantum well layer 241 is sandwiched between two barrier layers 142. The active layer 24 may have an SQW structure in which the composite quantum well layer 241 is directly sandwiched between the SCH layer 13 and the SCH layer 15 without the barrier layer 142 interposed therebetween.

  In the present embodiment, the composite quantum well layer 241 includes a plurality of quantum level control layers 143 and a main quantum well layer 144 having a band gap smaller than that of the plurality of quantum level control layers 143. The potential control layers 143 and the main quantum well layers 144 are alternately stacked. Here, it is assumed that the plurality of quantum level control layers 143 have the same layer thickness and band gap.

  Hereinafter, a structure in which one main quantum well layer 144 is sandwiched between two quantum level control layers 143 will be described as an example.

FIG. 7 is a diagram schematically showing the main part of the band structure of SOA2. In the example shown in FIG. 7, in the quantum level control layer 143, the degeneration of the light hole band edge E _lh and the heavy hole band edge E _hh of the valence band is solved by the extension strain, and the light hole band edge E _lh is the heavy hole band edge. _Ehh shifts to a higher energy side. Further, in the main quantum well layer 144, the degeneracy of the valence band is solved by shifting the light hole band edge E _lh to the lower energy side from the heavy hole band edge E _hh due to the compressive strain.

More specifically, the light hole band edge E _lh quantum level control layer 143 is present on a higher energy side than the light hole band edge E _lh of primary quantum well layer 144, heavy quantum level control layer 143 hole band end E _hh is located on the lower energy side than the heavier holes band edge E _hh of primary quantum well layer 144.

  That is, in the composite quantum well layer 241, the quantum level control layer 143 having the extension strain cancels the shift amount of the light hole base quantum level toward the low energy side by the main quantum well layer 144 having the compressive strain. Thus, it can be said that the light is controlled so that the ground quantum level of a light hole approaches that of a heavy hole.

  In the SOA 2 described here, the potentials for electrons, light holes, and heavy holes all have a symmetrical shape in the composite quantum well layer 241. Therefore, the existence probability density distributions of electrons, light holes, and heavy holes are all present. The distribution is symmetric within the composite quantum well layer 241.

  Further, since the layer thickness per layer of the quantum level control layer 143 having the extension strain is reduced as compared with the SOA 1 of the first embodiment, a more favorable heterointerface can be formed.

(Third embodiment)
FIG. 8 is a diagram schematically showing the main part of the band structure of SOA3. In the example shown in FIG. 8, in the quantum level control layer 143, the degeneration of the light hole band edge E _lh and the heavy hole band edge E _hh of the valence band is solved by the extension strain, and the light hole band edge E _lh is the heavy hole band edge. _Ehh shifts to a higher energy side. Further, since the main quantum well layer 144 is _unstrained , the light hole band edge E _lh and the heavy hole band edge E _hh are degenerated in the main quantum well layer 144.

More specifically, the light hole band edge E _lh quantum level control layer 143 is shifted to the higher energy side than the valence band edge E _V of primary quantum well layer 144, heavy quantum level control layer 143 hole band end E _hh is shifted to the low energy side than the valence band edge E _V of primary quantum well layer 144.

  That is, it can be said that the quantum level control layer 143 having an extensional strain is controlled so that the base quantum level of light holes in the composite quantum well layer 241 is close to the base quantum level of heavy holes.

  In the SOA 3 described here, the potentials for electrons, light holes, and heavy holes all have a symmetrical shape in the composite quantum well layer 241. Therefore, the existence probability density distributions of electrons, light holes, and heavy holes are all present. The distribution is symmetric within the composite quantum well layer 241.

(Fourth embodiment)
FIG. 9 is a diagram schematically showing the main part of the band structure of SOA4. In the example shown in FIG. 9, in the quantum level control layer 143, the degeneration of the light hole band edge E _lh and the heavy hole band edge E _hh of the valence band is solved by the extension strain, and the light hole band edge E _hh is the heavy hole band edge. Shifted to higher energy side than E _lh . Also in the main quantum well layer 144, the degeneracy of the valence band is solved by shifting the light hole band edge E _lh to the higher energy side than the heavy hole band edge E _hh due to the extension strain.

More specifically, the light hole band edge E _lh quantum level control layer 143 is present on a higher energy side than the light hole band edge E _lh of primary quantum well layer 144, heavy quantum level control layer 143 hole than the band edge E _hh primarily quantum well layer 144 heavy hole band edge E _hh with located on the lower energy side.

  That is, in the composite quantum well layer 241, the main quantum well layer 144 and the quantum level control layer 143 having an extension strain are light due to the quantum size effect due to the potential formed by the valence band of the composite quantum well layer. This suppresses the phenomenon that the ground quantum level of the hole shifts to the low energy side.

  With this action, it can be said that the SOA 4 controls the base quantum level of a light hole to be close to the base quantum level of a heavy hole. Further, with such a layer configuration, for example, the extension strain of the quantum level control layer 143 can be reduced as compared with the SOA 3. For this reason, since the average strain amount of the composite quantum well layer 241 can be suppressed, there is an advantage that the crystallinity of the composite quantum well layer 241 can be improved.

  In the SOA 4 described here, the potentials for electrons, light holes, and heavy holes all have a symmetrical shape in the composite quantum well layer 241. Therefore, the existence probability density distributions of electrons, light holes, and heavy holes are all present. The distribution is symmetric within the composite quantum well layer 241.

  As will be described below, in the configuration example in which the quantum level control layer 143 has an extensional strain, the relationship between the layer thickness of the quantum level control layer 143 and the strain amount while changing the band gap wavelength and the layer thickness of the main quantum well layer 144. The structural conditions for fabricating SOA4 from SOA2 of this embodiment were investigated by simulating the above. The result is shown in FIG.

In FIG. 10, the horizontal axis represents the layer thickness (d _QLC ) of the quantum level control layer 143 expressed in nanometer (nm) units, and the vertical axis represents the quantum level control layer 143 expressed in percent (%) units. The amount of strain (ε _QLC ).

For example, when SOA2 to SOA4 of the present embodiment are applied to the 1300 nm wavelength band, the hatched region in FIG. 10C, that is, the straight line f5 (d expressed by the following equations (5) and (6) _QLC) and in the region sandwiched f6 _{(d QLC),} and, under the condition of epsilon _QLC <0 and _{d QLC} ≧ 2 nm, was able to be manufactured SOA4 from SOA 2.
f5 (d _QLC ) = 0.06 × d _QLC −0.67 (5)
f6 (d _QLC ) = 0.04 × d _QLC -1.38 (6)

Equation (5) is a limit line of the gain wavelength band, and the gain wavelength band shifts from 1300 nm in a region where ε _QLC is larger and negative than this straight line. On the other hand, in the region where ε _QLC is smaller and negative than the straight line represented by the equation (6), the _{heterobarrier} becomes Type-II type at the interface between the quantum level control layer 143 and the barrier layer 142, and the band that prevents the electron transport. It becomes a structure. Further, _under the condition of d _QLC <2 nm, it becomes difficult to control the quantum level by the quantum level control layer 143.

For example, when SOA2 to SOA4 of the present embodiment are applied to the 1500 nm wavelength band, a hatched area in FIG. 10D, that is, a straight line f7 represented by the following expressions (7) and (8): _{(d QLC)} and in the region flanked by f8 _{(d QLC),} and, under the condition of -1.5% ≦ ε _QLC <0, was found to be manufactured SOA4 from SOA 2.
f7 (d _QLC ) = 0.432 × d _QLC −2.32 (7)
f8 (d _QLC ) = 0.2 × d _QLC −4.0 (8)

Equation (7) is a limit line of the mixed crystal composition of the quantum level control layer 143, and the quantum level control layer cannot be grown in a region where ε _QLC is larger than this line and negative. On the other hand, in the region where ε _QLC is smaller and negative than the straight line represented by the equation (8), the layer thickness of the quantum level control layer 143 exceeds the critical film thickness, making it difficult to achieve good crystal growth.

In the region where ε _QLC <−1.5%, the _{heterobarrier} becomes a Type-II type at the interface between the quantum level control layer 143 and the barrier layer 142, and a band structure that hinders electron transport is formed.

(Fifth embodiment)
FIG. 11 is a diagram schematically showing the main part of the band structure of the SOA 5 as the fifth embodiment. This configuration is a modified structure of the fourth embodiment. In the example shown in FIG. 11, in the quantum level control layer 143, the degeneration of the light hole band edge E _lh and the heavy hole band edge E _hh of the valence band is solved by the compressive strain, and the heavy hole band edge E _hh is the light hole band edge. Shifted to higher energy side than E _lh . Further, in the main quantum well layer 144, the degeneracy of the valence band is solved by shifting the light hole band edge E _lh to the higher energy side from the heavy hole band edge E _hh due to the extension strain.

More specifically, the light hole band edge E _lh quantum level control layer 143 is present on the lower energy side than the light hole band edge E _lh of primary quantum well layer 144, heavy quantum level control layer 143 hole also band edge E _hh located lower energy side than the main heavy-hole band edge E _hh quantum well layer 144.

  That is, in the composite quantum well layer 341, the quantum level control layer 143 having compressive strain suppresses the shift amount of the ground quantum level of light holes to the high energy side by the main quantum well layer 144 having extension strain. Thus, it can be said that the light is controlled so that the ground quantum level of a light hole approaches that of a heavy hole.

  In the SOA 5 described here, the potentials for electrons, light holes, and heavy holes all have a symmetrical shape in the composite quantum well layer 341. Therefore, the existence probability density distributions of electrons, light holes, and heavy holes are all present. The composite quantum well layer 341 has a symmetric distribution.

  With the potential design as described for SOA2, SOA3, SOA4, and SOA5, polarization independence of the optical gain can be realized, compared with the composite quantum well structure having the asymmetric potential described in the first embodiment. Thus, the recombination rate between electrons and heavy holes and the recombination rate between electrons and light holes can be promoted, and an optical gain can be improved.

  FIG. 12 is a graph showing an example of TE mode and TM mode optical gain characteristics with respect to the drive current injected in the forward direction through the p-type metal electrode 20 and the n-type metal electrode 21 for the SOA according to the present invention. is there. From this graph, it can be confirmed that the SOA according to the present invention realizes an amplification operation characteristic almost independent of polarization.

  In the above description, the structure in which the composite quantum well layer is composed of two and three semiconductor layers has been described. However, in the present invention, the composite quantum well layer may be composed of a larger number of semiconductor layers.

  Further, in a structure in which the composite quantum well layer is composed of more than three semiconductor layers, for example, as shown in FIG. 13, a quantum level control layer 443a having an extension strain and a quantum level control layer 443b having a compression strain are provided. By arranging them alternately, the composite quantum well layer 441 in the present invention can have a strain compensation structure as shown in Non-Patent Document 4, for example.

  By adopting such a structure, the generation of crystal defects in the composite quantum well layer can be suppressed, and the operational reliability of the SOA can be further improved. Note that the composite quantum well layer 441 may have a configuration in which the arrangement of the quantum level control layer 443a having extension strain and the quantum level control layer 443b having compression strain illustrated in FIG. 13 are interchanged.

  On the other hand, in a structure in which the composite quantum well layer is composed of two and three semiconductor layers, in a structure in which the quantum level control layer has an extensional strain, by applying a compressive strain to the barrier layer 142, the composite quantum well layer is A distortion compensation structure can be obtained. Note that the structure in which compressive strain is applied to the barrier layer 142 may be applied to a structure in which the composite quantum well layer includes more semiconductor layers than three layers.

1-5 Semiconductor optical amplifier (SOA)
11 n-type semiconductor substrate (substrate)
14, 24 Active layer 141, 241, 341, 441 Composite quantum well layer 142 Barrier layer 143, 443a, 443b Quantum level control layer 144 Main quantum well layer

Claims (9)

A quantum well having a barrier layer (142) and a composite quantum well layer (141, 241, 341, 441) composed of a plurality of semiconductor layers, wherein the barrier layer and the composite quantum well layer are alternately stacked; In the semiconductor optical amplifier (1-5) in which the active layers (14, 24) are formed on the substrate (11),
The plurality of semiconductor layers of the composite quantum well layer include at least one quantum level control layer (143, 443a, 443b) for controlling a quantum level in the composite quantum well layer, and the quantum level control layer And at least one main quantum well layer (144) having a smaller band gap, and at least one of the quantum level control layer and the main quantum well layer has a tensile strain with respect to the substrate. A semiconductor optical amplifier characterized by the above.   2. The semiconductor optical amplifier according to claim 1, wherein the plurality of semiconductor layers of the composite quantum well layer includes one main quantum well layer and one quantum level control layer.   The plurality of semiconductor layers of the composite quantum well layer includes a plurality of quantum level control layers and at least one or more main quantum well layers, and at least two or more quantum level control layers include: 2. The semiconductor optical amplifier according to claim 1, wherein at least one main quantum well layer is stacked so as to sandwich the main quantum well layer from both sides.   The plurality of semiconductor layers of the composite quantum well layer include a plurality of quantum level control layers and at least one or more main quantum well layers, and the barrier layers and the plurality of quantum level control layers 2. The semiconductor optical amplifier according to claim 1, wherein the semiconductor optical amplifier is laminated so that at least one main quantum well layer is disposed therebetween.   The semiconductor optical amplifier according to claim 1, wherein the main quantum well layer has a compressive strain with respect to the substrate.   5. The semiconductor optical amplifier according to claim 1, wherein the main quantum well layer is unstrained with respect to the substrate.   The semiconductor optical amplifier according to claim 1, wherein the main quantum well layer has a tensile strain with respect to the substrate.   8. The semiconductor optical amplifier according to claim 7, wherein the quantum level control layer has a tensile strain with respect to the substrate.   9. The semiconductor optical amplifier according to claim 1, wherein the composite quantum well layer is made of InGaAsP which is a III-V group mixed crystal.

Images