JP3797797B2 - Manufacturing method of semiconductor light emitting device - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a method for manufacturing a semiconductor light emitting device, particularly a semiconductor laser. The manufacturing method of the present invention is used for applications that require both high output and long life, such as an excitation light source for an optical fiber amplifier and a light source for optical information processing. It is suitably used for a semiconductor laser. Further, an LED such as a super luminescent diode having a light emitting end formed by an end face, or application to a surface emitting laser or the like is also possible.
[0002]
[Prior art]
Recent progress in optical information processing technology and optical communication technology is remarkable. For example, there is no spare time for high-density recording using a magneto-optical disk and bi-directional communication using an optical fiber network.
For example, in the communication field, Er as an amplifier for signal amplification that has flexibility for the transmission system as well as a large-capacity optical fiber transmission line that will fully support the future multimedia age.³⁺Research on optical fiber amplifiers (EDFA) doped with rare earths such as these has been actively conducted in various fields. Development of a semiconductor laser for a high-efficiency excitation light source, which is an indispensable element as an EDFA component, is awaited.
[0003]
In principle, there are three types of oscillation wavelengths of excitation light sources that can be used for EDFA applications: 800 nm, 980 nm, and 1480 nm. Among these, it is known that excitation at 980 nm is most desirable in view of the characteristics of the amplifier in consideration of gain, noise, and the like. Such a laser having an oscillation wavelength of 980 nm is desired to satisfy the conflicting characteristics of high output and long life as an excitation light source. Further, in the vicinity of this wavelength, for example, 890 to 1150 nm, there is also a demand as a heat source for an SHG light source and a laser printer, and development of a high output and highly reliable laser is straddling in various other applications.
[0004]
Also, in the information processing field, semiconductor lasers have been increased in output and wavelength for the purpose of high-density recording, short-time writing, and reading, and high output is strongly desired for conventional LDs with an oscillation wavelength of 780 nm. In addition, development of LDs in the 630-680 nm band has been vigorously performed in various directions.
As an approach for achieving both high output and high reliability, which are indispensable for realizing lasers, for example, the band gap of the active layer region in the vicinity of the end surface is made transparent with respect to the oscillation wavelength, and the vicinity of the end surface described above is used. Various methods have been proposed for suppressing light absorption in the light source. These lasers are generally called window lasers or NAM (non Absorbing Mirror) lasers, and are very effective when high power is required.
[0005]
On the other hand, a problem solving method as disclosed in JP-A-3-101183 has been proposed. According to this, a contamination-free end face is formed, and a substance that does not cause diffusion or reaction with the semiconductor end face is formed on the passivation layer or a part thereof. The manufacturing method is said to be effective.
Further, as publicly known documents similar to the above-mentioned JP-A-3-101183, LWTu et al., (In-vacuum cleaving and coating of semiconductor laser facets using silicon and a dielectric, J. Appl. Phys. 80 (11) 1 DEC. 1996). According to this, Si / AlO_xIt is described that when the structure is coated on the laser end face, cleaving in a vacuum slows the recombination rate of carriers on the cleaved face and increases the initial COD level.
[0006]
Further, in order to reduce the electric field strength at the light emitting end face of the semiconductor laser, Si is added to the interface between the coating film and the semiconductor so that the antinode part of the standing wave existing in the cavity direction does not coincide with the end face part. A technique for inserting / 4 wavelengths is also known.
[0007]
[Problems to be solved by the invention]
For example, with a semiconductor laser near 980 nm, a semiconductor laser that can withstand continuous use for about two years at a light output of about 50-100 mW and a manufacturing method thereof have already been developed. Happiness and reliability are insufficient. This is the same for the 780 nm band and 630-680 nm band LDs, and ensuring the reliability at the time of high output is a problem of the entire semiconductor laser using a GaAs substrate.
[0008]
One of the causes is due to deterioration of the emission end face of the laser light that is exposed to a very high light density. As is well known for GaAs / AlGaAs semiconductor lasers, there are many surface levels in the vicinity of the end face. However, since these levels absorb laser light, the temperature in the vicinity of the end face is generally the laser. It is described that the temperature rises higher than the internal temperature, and that this temperature rise further narrows the band gap in the vicinity of the end face and further facilitates the absorption of laser light. This phenomenon is known as end-of-face destruction so-called COD (catastrophic optical damage) observed when a large current is passed instantaneously. In addition, many semiconductors have abrupt deterioration due to a decrease in COD level during a long-term energization test. This is a common problem in laser elements. Although attempts to solve these problems have been made energetically as described above, they are still technically insufficient.
[0009]
Considering the case of an LD having a window structure, for example, there is a manufacturing method in which a semiconductor material transparent to an oscillation wavelength is epitaxially grown on a laser end face. In this method, since the laser is brought into a so-called bar state and epitaxial growth is performed on the end face, the electrode process performed thereafter becomes very complicated.
Various methods for disordering the active layer have also been proposed by intentionally thermal diffusion or ion implantation of Zn or Si as impurities into the active layer near the laser end face. Examples of these known documents include JP-A-2-45992, JP-A-3-31083, and JP-A-6-302906.
[0010]
However, since impurity diffusion generally performed in the LD manufacturing process is performed from the epitaxial direction of the laser element toward the substrate direction, there is a problem in the controllability of the diffusion depth and the controllability of the lateral diffusion with respect to the resonator direction. It is difficult to make.
Further, in the case of ion implantation, since high energy ions are introduced from the end face, damage tends to remain on the LD end face even after annealing. Further, an increase in reactive current accompanying a decrease in resistance in a region where impurities are introduced has problems such as an increase in laser threshold current and drive current.
[0011]
On the other hand, for example, a non-contaminated end surface disclosed in Japanese Patent Application Laid-Open No. 3-101183 is formed, and a substance which does not cause diffusion or reaction with the semiconductor end surface itself and does not contain oxygen is passivated. The technical problems of the manufacturing method formed as a layer or a part thereof are as follows.
In general, there is no effect of suppressing generation of non-radiative recombination centers such as Ga-O and As-O generated at the end face at the time of cleavage depending on work in a clean room such as in the atmosphere. As described in the first claim, it is indispensable to form an inactivation layer in the cleaved spot as described in the first claim. The environment is only cleavage in a vacuum as in the tenth claim. However, this requires much more complicated equipment and work than general cleavage in the atmosphere. In addition, the end face generally formed by dry etching disclosed in the 11th to 14th claims forms more non-radiative recombination centers than the end face formed by cleaving, and has a long life. It is not suitable as a required LD manufacturing method.
[0012]
Further, Si or amorphous Si is cited as the most suitable passivation layer, but generally there is no substance that does not cause any diffusion, and it is assumed that it is driven at high output and high temperature for a long time. In such a semiconductor laser, there is a concern about the diffusion of the passivation material disclosed in the patent.
Also, in the above-mentioned L.W.Tu et al., (In-vacuum cleaving and coating of semiconductor laser facets using silicon and a dielectric, J. Appl. Phys. 80 (11) 1 DEC. 1996), Si / AlO_xCleaving in vacuum when coating the structure on the laser end face slows the recombination rate of carriers on the cleavage plane and increases the initial COD level, but there is no long-term reliability statement, There is no mention of the relationship of the LD structure.
[0013]
Further, in order to reduce the electric field strength at the light emitting end face of the semiconductor laser, Si is added to the interface between the coating film and the semiconductor so that the antinode part of the standing wave existing in the cavity direction does not coincide with the end face part. In the technology for inserting / 4 wavelength, Si itself acts as an optical absorber in a wavelength band where a general semiconductor laser is realized, particularly in a range of 400 to 1600 nm where a high output LD is desired. There is a possibility that the temperature rise at the end face accelerates the deterioration of the device.
[0014]
The present invention has been made to solve such a problem, and the object thereof is to stably suppress the interface state density at the end face of a semiconductor light emitting element such as a semiconductor laser over a long period of time, and to provide an inactive layer. This is the realization of a simple manufacturing method that provides a semiconductor laser that operates stably even when diffusion occurs during LD driving. This means that high output and long life with reduced degradation at the end face are achieved. It is none other than the provision of high-performance semiconductor lasers that are compatible with each other.
[0015]
[Means for Solving the Problems]
  As a result of intensive studies to solve the above-mentioned problems, the present inventors have formed a compound semiconductor layer including a first conductivity type cladding layer, an active layer, and a second conductivity type cladding layer on a substrate, resulting in resonance. In a method for manufacturing a semiconductor light emitting device having a resonator structure, a compound semiconductor layer is sequentially grown on a substrate, a resonator end face is formed by cleavage, and then the substrate, first conductivity type cladding layer exposed on the end face are formed The elements constituting the end face such as the active layer and the second conductivity type cladding layer, in particular, the plasma irradiation is performed in the vicinity of the active layer in a vacuum, and the plasma irradiated surface is subsequently continued in the vacuum.Made of Si, Ge, S and / or SeAlthough the semiconductor light emitting device manufactured by the manufacturing method for forming the passivation layer is a simple manufacturing method, it has been found that both high output and long life can be achieved at a level far exceeding the conventional technology. Reached.
[0016]
  That is, the gist of the present invention is that a compound semiconductor layer including a first conductivity type cladding layer, an active layer, and a second conductivity type cladding layer is formed on a substrate, and a method for manufacturing a semiconductor light emitting device having a resonator structure. Then, after sequentially growing the compound semiconductor layer on the substrate, a cavity end face is formed by cleavage, and then the end face is irradiated with plasma in a vacuum, and the plasma irradiated face is subsequently applied in a vacuum.Made of Si, Ge, S and / or SeIt exists in the manufacturing method of the semiconductor light-emitting device characterized by forming an inactivation layer.
[0017]
DETAILED DESCRIPTION OF THE INVENTION
  The present invention is described in detail below. The method for manufacturing a semiconductor light-emitting device according to the present invention includes at least a first conductivity type cladding layer, an active layer, and a second conductivity type cladding layer stacked on a substrate, and a resonator structure formed by two opposing end faces. A method for manufacturing a compound semiconductor light emitting device, comprising sequentially growing a compound semiconductor layer on a substrate, forming a resonator end face by cleavage, and then irradiating the end face with plasma in a vacuum, In the vacuumMade of Si, Ge, S and / or SeThe structure of the semiconductor light emitting device is not particularly limited as long as it is a manufacturing method of a semiconductor light emitting device characterized by forming an inactivation layer. However, as an example of a specific manufacturing method, a refractive index waveguide structure is described below. The second conductivity type cladding layer is divided into two layers of the first and second, and the second conductivity type second cladding layer and the current blocking layer form a current injection region, and further contact with the electrode A case will be described in which a semiconductor laser having a contact layer for lowering resistance is fabricated and the present invention is applied.
[0018]
The basic epitaxial structure of such a laser is equivalent to, for example, the laser described in Japanese Patent Laid-Open No. 8-130344 by Horie et al. This type of laser is a light source for an optical fiber amplifier used for optical communication, and information It is used as a pick-up light source for a large-scale magneto-optical memory for processing, and can be applied to various uses depending on the layer configuration such as the active layer and the cladding layer, and the material configuration.
[0019]
FIG. 2 is a schematic example in which a groove type semiconductor laser is constructed as an example of an epitaxial structure in the semiconductor laser of the present invention.
As the substrate (1), InP, GaAs, GaN, InGaAs, Al from the viewpoint of desired oscillation wavelength, lattice matching, strain intentionally introduced into the active layer, etc.₂O_ThreeA single crystal substrate such as is used. Al₂O_ThreeIn some cases, a dielectric substrate can also be used. As an embodiment implemented in the present invention, an InP substrate or a GaAs substrate is desirable from the viewpoint of lattice matching with respect to a III-V group semiconductor light emitting device containing As, P, etc. as a V group, and includes As as a V group. Most preferably, a GaAs substrate is used.
[0020]
Al₂O_ThreeSuch a dielectric substrate may be used for a material containing nitrogen or the like as a group V among the group III-V semiconductor light emitting devices.
As the substrate, not only a so-called just substrate but also a so-called miss oriented substrate can be used from the viewpoint of improving crystallinity during epitaxial growth. This has the effect of promoting good crystal growth in step flow mode and is widely used. The off-substrate having an inclination of about 0.5 to 2 degrees is widely used, but depending on the material system constituting the quantum well structure, the inclination may be about 10 degrees.
[0021]
In some cases, the substrate is subjected to chemical etching, heat treatment, or the like as preparation for making a light-emitting element using a crystal growth technique such as MBE or MOCVD.
The buffer layer (2) is preferably grown on the substrate to alleviate imperfections in the substrate bulk crystal and facilitate the formation of an epitaxial thin film having the same crystal axis. The buffer layer (2) is preferably composed of the same compound as the substrate (1). When the substrate is GaAs, GaAs is usually used. However, the use of a superlattice layer as a buffer layer is also widely performed and may not be formed of the same compound. On the other hand, when a dielectric substrate is used, it is not necessarily the same substance as the substrate, and a material different from the substrate may be selected as appropriate from the desired emission wavelength and the structure of the entire device.
[0022]
The first conductivity type cladding layer (3) is generally made of a material having a refractive index smaller than the average refractive index of the active layer (4), and is a substrate (1 prepared for realizing a desired oscillation wavelength). ), The buffer layer (2), the active layer (4), etc., the material is appropriately defined. For example, GaAs is used as the substrate (1), and when the buffer layer (2) is also GaAs, an AlGaAs-based material, an InGaAs-based material, an AlGaInP-based material, an InGaP-based material, or the like is used. In some cases, the entire cladding layer can have a superlattice structure.
[0023]
The effect of the present invention is effective regardless of the conductivity type, material, structure, etc. of the active layer (4). First, the conductivity type is preferably P-type. As the P-type dopant, known ones such as Zn, C, Be, and Mg may be appropriately selected in view of the base material and the growth method, preferably Be and / or C, and most preferably Be.
In general, as described in, for example, Heterostructure Lasers (H.C.Casey, Jr. M.B.Panish, Academic Press 1978 P.157), the band gap of a semiconductor tends to decrease as the hole concentration increases. For example, in the case of GaAs, the band gap is Eg (eV), and the P-type carrier concentration is P (cm^-3As
Eg = 1.424-1.6 × 10^-8× P^1/3
It is shown that. One feature of the present invention is that when a substance inserted as an inactivation layer (14) into the laser end face and the dielectric interface diffuses during long-term laser driving, particularly during high-power operation, n-type impurities Is gradually introduced into the active layer (4) and the like, and causes an effective decrease in hole concentration due to the compensation effect. This means that in the vicinity of the semiconductor end face, in other words, the passivation layer (14) increases the band gap of the part where the constituent elements are diffused together with the LD driving, and absorbs light at the end face. It is expected to act to suppress. In this respect, a P-type active layer is desirable.
[0024]
From the viewpoint of material selection, the active layer (4) is preferably a system containing In and / or Ga, more preferably a system containing In, and most preferably a system containing In and Ga. This is a material system in which so-called ordering is likely to occur during crystal growth, and a substance inserted as an inactivation layer (14) at the interface between the laser end face and the dielectric is as described above during long-term laser driving. This is because it is also expected to cause disorder in the vicinity of the end face when diffusing into the surface. In general, disordering of the material leads to an increase in the band gap, and this, in combination with the carrier compensation effect, suppresses light absorption at the end face for a long period of time.
[0025]
From these viewpoints, the material of the active layer (4) is specifically an AlGaAs-based material, InGaAs-based material, InGaP-based material, AlGaInP-based material, etc._XGa_1-xAs (0 <x <1) or (Al_xGa_1-x)_yIn_1-yP (0 <x, y <1) is desirably included, and in particular, a quantum well structure is desirable from the viewpoint of disordering. The selection of these materials is usually dictated by the desired oscillation wavelength.
[0026]
The oscillation wavelength λ (nm) of the compound semiconductor light emitting element is substantially determined by the material selection of the active layer, and the oscillation wavelength λ (nm) is preferably shorter than the Si absorption edge.
The active layer (4) may be a normal bulk active layer consisting of a single layer as a structure, but a single quantum well (SQW) structure, a double quantum well (DQW) structure, a multiple quantum (MQW) structure. The quantum well structure such as is adopted according to the purpose. In the quantum well structure, an optical guide layer is usually used in combination, and a barrier layer is used in combination for separating the quantum well as necessary. The structure of the active layer includes a structure in which light guide layers are provided on both sides of the quantum well (SCH structure), and a structure in which the refractive index is continuously changed by gradually changing the composition of the light guide layer (GRIN-SCH). Structure) or the like. The material of the light guide layer can be selected according to the active layer, such as an AlGaAs material, an InGaAs material, an InGaP material, or an AlGaInP material.
[0027]
The light guide layer may be a superlattice in which the above materials are combined.
[0028]
On the other hand, Si is desirable as a specific material for the passivation layer (14). This is because, in general, it can become an n-type dopant by diffusion from the (110) plane that forms the resonator in an edge-emitting device or from the (100) plane that becomes the exit end of a surface-emitting laser or the like. Further, it is possible to disorder the AlGaAs-based material, InGaAs-based material, InGaP-based, and AlGaInP-based material that are preferably used as the material of the active layer (4).
[0029]
The second conductivity type first and second clad layers (5) and (8), like the first conductivity type clad layer (3), generally have a refractive index smaller than the average refractive index of the active layer (4). The material is appropriately defined by the substrate (1), the buffer layer (2), the active layer (4), and the like. For example, GaAs is used as the substrate (1), and when the buffer layer (2) is also GaAs, an AlGaAs-based material, an InGaAs-based material, an AlGaInP-based material, an InGaP-based material, or the like is used.
[0030]
FIG. 2 describes two types of etch stop layers (6), (7) and cap layer (10), which are employed in the preferred embodiment of the present invention to create a current injection region. This is effective for precise and easy embedding.
The second etch stop layer (6) is, for example, Al_aGa_1-aWhen composed of an As (0 ≦ a ≦ 1) material, GaAs is usually preferably used. This is because the second conductivity type second clad layer (8) and the like can be stacked with good crystallinity, particularly when regrown in an AlGaAs system by MOCVD or the like. The thickness of the second etching stop layer (6) is usually preferably 2 nm or more.
[0031]
The first etching stop layer (7) is made of In_bGa_1-b A layer represented by P (0 ≦ b ≦ 1) is suitable, and when GaAs is used as a substrate as in the present invention, b = 0.5 is usually used in a system without strain. The thickness of the first etching blocking layer (7) is usually 5 nm or more, preferably 10 nm or more. If it is less than 5 nm, there is a possibility that etching cannot be prevented due to film thickness disturbance or the like. On the other hand, a strain system can be used depending on the film thickness, and b = 0, b = 1, and the like can also be used.
[0032]
The cap layer (10) is used as a protective layer for the current blocking layer (9) in the first growth and at the same time used to facilitate the growth of the second conductivity type second cladding layer (8). Some or all are removed before obtaining.
As the current blocking layer (9), it is necessary to literally block the current so that it does not flow substantially. Therefore, the conductivity type is preferably the same as that of the first conductivity type cladding layer (3) or undoped. For example, if the current blocking layer (9) is formed of AlGaAs, Al_yGa_1-yThe refractive index is preferably smaller than that of the second conductivity type second cladding layer (8) made of As (0 <y ≦ 1). That is, the current blocking layer (9) is Al_zGa_1-zIf As (0 ≦ z ≦ 1), therefore, the mixed crystal ratio is preferably z> y.
[0033]
The refractive index of the second conductivity type second cladding layer (8) is usually set to be equal to or lower than the refractive index of the active layer (4). The second conductivity type second cladding layer (8) is usually the same as the first conductivity type cladding layer (3) and the second conductivity type first cladding layer (5). Further, as one of preferred embodiments of the present invention, the second conductive type first cladding layer (5), the second conductive type second cladding layer (8) and the current blocking layer (9) are all made of the same material system having the same composition. It is mentioned to comprise. In that case, an effective refractive index difference is formed by the first etching stop layer (7), and when the cap layer (10) is not completely removed, the cap layer (7) is added in addition to the first etching layer (7). 10) also forms an effective refractive index difference. By adopting such a layer configuration, it is possible to avoid various problems caused by the mismatch of materials or compositions at the respective interfaces of the second conductivity type second cladding layer (8) and the current blocking layer (9), Highly preferred.
[0034]
A contact layer (11) is preferably provided on the second conductivity type second cladding layer (8) for the purpose of, for example, reducing the contact resistivity with the electrode (12). The contact layer (11) is usually made of a GaAs material. This layer is usually made to have a higher carrier concentration than other layers in order to lower the contact resistivity with the electrode.
In general, the buffer layer (2) has a thickness of 0.1 to 3 [mu] m, the first conductivity type cladding layer (3) has a thickness of 0.5 to 3 [mu] m, and the active layer (4) has a quantum well structure. In this case, 0.0005 to 0.02 μm per layer, the thickness of the second conductivity type first cladding layer (5) is 0.05 to 0.3 μm, and the thickness of the second conductivity type second cladding layer (8) is 0. The thickness of the cap layer (10) is selected from the range of 0.005 to 0.5 μm, and the thickness of the current blocking layer (9) is selected from the range of 0.3 to 2 μm.
[0035]
The semiconductor light emitting device shown in FIG. 2 is configured by further forming electrodes (12) and (13). In the case of the p-type, the electrode (12) is formed by, for example, sequentially depositing Ti / Pt / Au on the surface of the contact layer (11) and then performing an alloy process. On the other hand, the electrode (13) is formed on the surface of the substrate (1). In the case of an n-type electrode, for example, AuGe / Ni / Au is sequentially deposited and then formed by alloying.
[0036]
Although the semiconductor wafer thus formed is cleaved to form a so-called laser bar, the present invention does not necessarily require a complicated cleaving process in a vacuum. The first conductivity type clad layer (3), the active layer (4), and the second conductivity type clad layer (5) which become non-radiative recombination centers at the end face even in the cleavage at atmospheric pressure or in the nitrogen atmosphere. ), (8), and further, elements located in the vicinity of the resonator mirror among the constituent elements such as the substrate (1), the current blocking layer (9), and the contact layer (11) exposed at the end face This is because at least one oxide and / or nitride, etc., particularly in the oxide state, can be removed by the following treatment.
[0037]
The technique for this is plasma irradiation to the end face. In particular, plasma irradiation in which a rare gas is ionized, particularly Ar plasma irradiation is effective. More preferably, by using this Ar plasma irradiation energy in a low energy range of about 25 eV to 300 eV, more preferably 25 eV to 100 eV, it becomes a non-radiative recombination center without excessively damaging the semiconductor end face. Removal of these oxides, nitrides and the like can be realized. This process is, for example, an extremely low energy process compared to conventional techniques such as ion implantation of impurities, so that it can be said that the process can be performed while suppressing the introduction of damage to the end face. From the viewpoint of removing the oxide from the end face, it is particularly effective for removing As-O. Further, Ga-O and the like have the same effect.
[0038]
An example of a method for analyzing whether or not at least one of constituent elements is present in the form of an oxide is XPS (X-ray Photo-electron Spectroscopy X-ray photoelectron spectroscopy). This is a very useful means for knowing the chemical bonding state of each element. The laser end face is irradiated with X-rays with a size of about 100 μm × 100 μm, and the resulting photoelectrons are subjected to energy spectroscopy. By doing so, the chemical bonding state of each element constituting the laser end face can be confirmed. At this time, only the information in the vicinity of the surface can be easily obtained by changing the angle formed with the sample surface of the photoelectron detector. In addition, since a general laser is coated on the end face with a dielectric or a pair of dielectric and semiconductor as described later, various etching methods are used before the XPS measurement to obtain a thickness suitable for analysis. Usually, the coating film is thinned. In addition, regarding a laser on which a thin coating film of about 2 nm is formed, it is possible to analyze the end face of the semiconductor laser without performing such processing as etching.
[0039]
Further, the irradiation effect of the low energy Ar plasma is not limited to the removal of the oxide or the nitride depending on the case. As described above, as a desirable form of the active layer, a quantum well is formed. In particular, when the material system is an AlGaAs-based material, an InGaAs-based material, an InGaP-based material, an AlGaInP-based material, or the like, a deactivation layer ( 14) Diffusion, for example, Si diffusion, has an effect that the end face vicinity can be disordered during the laser manufacturing process, and an end face having high resistance can be formed. This means an increase in the band gap near the end face at the initial stage of laser fabrication, an increase in resistance near the end face, that is, the suppression of light absorption at the end face, and the suppression of current injection into the end face, which tends to break down. Longer life expectancy can be expected.
[0040]
The plasma irradiation also has an effect of widening the band gap in the vicinity of the end face and making it transparent with respect to the oscillation wavelength λ in the LD manufacturing process, and further, while the inactivation layer (14) diffuses during the LD driving, Combined with the effect of widening the band gap in the vicinity, it realizes a high output, long life device.
The method of the present invention is characterized by plasma irradiation to the resonator end face, that is, the cleaved face. More specifically, the first conductivity type cladding layer, the active layer, and the second clad that are exposed on the end face and form the end face. The layer is preferably irradiated with plasma at least, and usually the plasma is irradiated on the entire end face including other layers. Plasma irradiation is 10^-3Torr or less, preferably 10^-FourLess than Torr, most preferably 10^-FivePerform in a vacuum of about Torr or less.
[0041]
Cleavage is preferably used to form the end face. This is widely used in the case of an edge-emitting laser, but a resonator may be fabricated in the crystal growth process like a surface-emitting laser.
[0042]
The end face formed by cleavage differs depending on the orientation of the substrate used. For example, when an element such as an edge-emitting laser is formed using a substrate having a crystallographically equivalent surface as nominally (100) (nominally (100)), (110) or The crystallographically equivalent plane is the plane that forms the resonator, but when the aforementioned off oriented substrate is used, the end face depends on the relationship between the tilted direction and the resonator direction. May not make 90 ° with the resonator direction. For example, when a substrate having an angle of 2 ° from the (100) substrate toward the (1-10) direction is used, the end face is also inclined by 2 °.
[0043]
In the present invention, the passivation layer is formed on the end face of the compound semiconductor light emitting device, and forms a non-radiative recombination center when an element constituting the end face is bonded, for example, a chemical reaction with an element such as oxygen. It is a layer to prevent. The passivation layer is formed so as to cover at least the first conductivity type cladding layer, the active layer, and the second conductivity type cladding layer that form the end face, but is usually formed so as to cover the entire end face. .
[0044]
  In the present invention, the passivation layer is formed in the vacuum after the plasma irradiation on the end face in vacuum, that is, 10^-3In a vacuum of about Torr or less, preferably 10^-6Torr or less, most preferably 10^-7It is formed in a high vacuum of about Torr or less. As the material of the passivation layer, Si, Ge, S, Se isAmong them, it is preferable to include Si, and it is preferable to include 50 atomic% or more of Si in the passivation layer.
[0045]
The element deposited as the passivation layer (14) on the semiconductor end face is preferably an n-type impurity when diffused into the active layer, and Si is preferably used as described above. Although the structure and characteristics of Si differ crystallographically depending on the production method, the effect is recognized in any case of single crystal, polycrystal, and amorphous.
Particularly preferably, amorphous Si formed at a low film formation rate in a high vacuum is used. In general, the absorption edge of Si varies depending on the film quality, but it is transparent for wavelengths of about 2 μm or more, and it is considered that there is no absorption. Conversely, for wavelengths shorter than about 2 μm, the refractive index N of Si is N = n + ik, where n is the real part of the refractive index, k is the extinction coefficient, and n is about 3.5.
[0046]
Generally the thickness T of the passivation layer (14)_pIt is desirable that (nm) is thicker than 0.2 (nm). However, an extremely thick film thickness such as 100 nm may not be suitable. The desirable thickness of the passivation layer (14) is defined by the requirement that the lower limit itself exists as a film, and the upper limit is a balance with the effect that light emitted from the active layer is absorbed by Si. It is determined. In other words, both the requirement that the end face is covered with the entire passivation film and the effect of both the temperature rise of the end face due to the absorption of Si are considered when Si is deposited on the end face. From these experimental results,
0.2 (nm) <T_p(Nm) <λ / 8n (nm) (I)
(In the formula (I), λ represents the oscillation wavelength of the semiconductor light emitting device, and n represents the real part of the refractive index at the wavelength λ of the passivation layer.)
It is confirmed that. However, the effect has been confirmed even when the thickness is 0.2 nm or less.
[0047]
Furthermore, it is important to have a coating layer (15) (16) made of a laminated dielectric or a combination of dielectric and semiconductor on a passivation layer (14) constructed on the exposed semiconductor end face. is there. Particularly preferably, the ion irradiation to the end face, the formation of the passivation layer (14), and the formation of the coating layers (15) and (16) are continuously performed in a vacuum. This is mainly performed for the purpose of increasing the light extraction efficiency from the semiconductor laser and for the purpose of further protecting the end face. In particular, in order to obtain a high output, an asymmetric coating is widely used in which a coating having a low reflectance with respect to the oscillation wavelength is applied to the front end face and a high reflectance is applied to the rear end face.
[0048]
Various materials can be used for this coating, AlO_xTiO_x, SiO_x, SiN, Si and ZnS are preferably selected from the group consisting of one or two or more kinds, but as a low reflection coating, AlO_xTiO_x, SiO_xAlO as a highly reflective coating_x/ Si multilayer, TiO_x/ SiO_xA multilayer film or the like is used. Each film thickness is adjusted to achieve a desired reflectance. However, in general, as a low reflection coating, AlO_xTiO_x, SiO_xIs generally adjusted to a film thickness in the vicinity of λ / 4n, where n is the real part of the bending ratio at the wavelength λ. In addition, it is preferable to adjust the highly reflective multilayer film so that each material constituting the highly reflective multilayer film is in the vicinity of λ / 4n, and further laminate this pair according to the purpose.
[0049]
A so-called IAD (Ion Assisted Deposition) method is preferably used in the method of manufacturing the coating layers (15) and (16). This is a method of irradiating ions having a certain energy simultaneously with vacuum deposition of a coating material, and ion irradiation with a rare gas is particularly suitable. Furthermore, among rare gases, IAD using Ar ions has a great effect on improving the film quality of the coating material. In particular, the optimum condition for Ar ion irradiation is to use in a low energy range of about 25 eV to 300 eV, and more preferably about 50 eV to 200 eV, thereby enabling coating without damaging the semiconductor end face. . 10^-3Torr or less, more preferably 10^-FourLess than Torr, most preferably 10^-FiveIt is good to carry out in a vacuum of about Torr or less.
[0050]
【Example】
EXAMPLES Hereinafter, although an Example demonstrates this invention still in detail, this invention is not limited to a following example, unless the summary is exceeded.
(Example 1)
The groove type laser element shown in FIG. 2 was manufactured as follows.
[0051]
Carrier concentration 1 × 10¹⁸cm^-3On the (100) surface of the n-type GaAs substrate (1), a carrier concentration of 1 × 10 1 as a buffer layer (2) is formed by MBE.¹⁸cm^-3The n-type GaAs layer and the first conductivity type cladding layer (3) have a carrier concentration of 1 × 10 5 having a thickness of 1.5 μm.¹⁸cm^-3N-type Al_0.35Ga_0.656 nm thick undoped In on the As layer, then 24 nm thick undoped GaAs optical guide layer_0.2Ga_0.8An As single quantum well (SQW), an active layer (4) having an undoped GaAs optical guide layer having a thickness of 24 nm thereon, a second conductivity type first cladding layer (5) having a thickness of 0.1 μm, a carrier Concentration 1 × 10¹⁸cm^-3P-type Al_0.35Ga_0.65As layer, second etch stop layer (6) 10 nm thick, carrier concentration 1 × 10¹⁸cm^-3P-type GaAs layer, first etching stop layer (7) having a thickness of 20 nm and a carrier concentration of 5 × 10¹⁷cm^-3N-type In_0.49 Ga_0.51P layer, current blocking layer (9) thickness 0.5 μm, carrier concentration 5 × 10¹⁷cm^-3N-type Al_0.39Ga_0.61As layer and cap layer (10) 10 nm thick, carrier concentration 1 × 10¹⁸cm^-3N-type GaAs layers were sequentially stacked.
[0052]
Next, a mask of silicon nitride was provided in a portion excluding the uppermost current injection region portion. In this case, the width of the opening of the silicon nitride mask was 1.5 μm. Etching was performed using the first etching stop layer as an etching stop layer, and the cap layer (10) and the current blocking layer (9) in the current injection region portion were removed. The etchant used at this time was a mixture of sulfuric acid (98 wt%), hydrogen peroxide (30 wt% aqueous solution) and water in a volume ratio of 1: 1: 5, and was performed at 25 ° C. for 30 seconds.
[0053]
Then HF (49%) and NH_FourThe silicon nitride layer is removed by immersion in an etchant mixed with F (40%) 1: 6 for 2 minutes and 30 seconds, and the second etching stop layer is used as an etching stop layer to stop the first etching in the current injection region. The layer was etched away. The etchant used at this time was a mixture of hydrochloric acid (35 wt%) and water in a ratio of 2: 1, the temperature was 25 ° C., and the time was 2 minutes.
[0054]
Thereafter, the carrier concentration is 1 × 10 as the second conductivity type second cladding layer (8) by MOCVD.¹⁸cm^-3P-type Al_0.35Ga_0.65The As layer is grown to a thickness of 1.5 μm at the buried portion (current injection region portion), and finally has a thickness of 7 μm and a carrier concentration of 1 as a contact layer (11) for maintaining good contact with the electrode. × 10¹⁹cm^-3A p-type GaAs layer was grown to form a laser element. The width W of the current injection region of this laser element, that is, the width of the second conductivity type second cladding layer at the interface with the second etching stop layer was 2.2 μm.
[0055]
On this wafer, AuGeNi / Au is deposited as an n-type electrode (13) on the substrate side, and Ti / Pt / Au is deposited on the p-side electrode (12), and alloyed at 400 ° C. for 5 minutes to form a wafer. Completed.
Subsequently, in the atmosphere, a laser bar having a cavity length of 700 μm was cleaved to expose the (110) plane, and the laser bar was placed in a vacuum chamber having an Ar plasma generator. 3 × 10^-FiveIn Torr vacuum, average energy 60eV, current density 150μA / cm²The end surface (cleavage surface) was irradiated with Ar plasma for 1 minute. 2 × 10 continuously^-7In the Torr vacuum, amorphous Si was deposited on the end face of 2 nm by using an ordinary electron beam evaporation method to form a Si passivation layer (14). Further continuously AlO_xA film was formed at 165 nm so that the reflectance of the front end face was 2.5% at an oscillation wavelength of 980 nm, and a coating layer (15) was formed. AlO_x4 × 10 during film formation^-FiveAverage energy 120eV, current density 200μA / cm by IAD method in Torr vacuum²Ar plasma of AlO_xIrradiated simultaneously with the supply to the end face of
[0056]
It has been confirmed that the real part of the refractive index of amorphous Si at 980 nm is about 3.4.
[0057]
Further, the laser bar was once taken out from the vacuum chamber in order to perform the processing on the rear end face side. On the rear end face side, Ar plasma irradiation and Si passivation layer (14) are formed in exactly the same manner as the front end face side._x170nm film / Amorphous Si film 60nm / AlO_xFour coating layers (16) having a film thickness of 170 nm / amorphous Si of 60 nm were formed, and a rear end face with a reflectance of 92% was produced. AlO_xThe film was formed by the same IAD method as that of the front end face side, and the amorphous Si film was formed by the same electron beam evaporation method as that of the front end face side.
One sample of this laser bar was subjected to XPS measurement for end face analysis. At this time, the extraction angle of the photoelectrons was set to 75 degrees, and the state of the semiconductor laser end face was observed. As a result, neither Ga—O nor As—O present on the GaAs (110) surface, which was once exposed to the normal atmosphere, was detected.
[0058]
Also, one device is taken out from this laser bar as a sample for analysis, and the front end face of AlO_xThe layer and the Si layer were removed with a hydrofluoric acid-based etchant and then put into a vacuum analyzer, and the band gap in the vicinity of the front end face of the active layer was measured using electron energy loss spectroscopy. Electron energy loss spectroscopy is an analysis method that obtains information only in the vicinity of the sample surface (maximum analysis depth of about 1.5 nm), so it is effective in measuring the band gap of the laser end face without being affected by physical properties in the bulk region. It is a technique. An electron beam of 1000 eV focused to about 100 nmφ was irradiated to the vicinity of the active layer on the laser end face, and the energy of the lost electrons diffracted from the region 1 nm deep from the end face of the semiconductor itself was analyzed in the back of the surface oxide layer. From the loss peak caused by the inter-transition, the band gap near the end face of the InGaAs quantum well layer was measured to be 1.5 eV, and the band gap near the end face of the GaAs optical guide layer was measured to be 1.65 eV. The energy gap between the quantum levels of the InGaAs quantum well active layer at room temperature obtained from photoluminescence measurement is 1.29 eV, and the band gap of GaAs is 1.41 eV. And the end face was confirmed to be transparent to the oscillation wavelength.
[0059]
Ten devices from this laser bar were placed on a heat dissipating submount and packaged in a nitrogen atmosphere. As an average initial characteristic of the device, a threshold current was 23 mA at 25 ° C., and kinks were observed at 350 mA and 250 mW. A life test was performed on this population. As a result of an accelerated test at 200 mW and 50 ° C., as shown in FIG. 3, there was no sudden death when 2000 hrs passed, and a stable operation was confirmed.
[0060]
One of the fabricated devices was processed into a sample for observation with a transmission electron microscope (TEM), and the state of the bulk near the active layer near the end face was compared. As a result, it was confirmed that the crystallinity in the vicinity of the active layer was broken in the vicinity of the end face irradiated with Ar plasma and disordered.
[0061]
(Example 2)
SiO formed by coating the rear end face coating layer by the same IAD method as the front end face side_xThe film is 200 nm / TiO_x120nm / SiO_xThe film is 200 nm / TiO_x120nm / SiO_xThe film is 200 nm / TiO_xA laser bar was fabricated in exactly the same manner as in Example 1 except that the thickness was 6 layers continuous with 120 nm, and the rear end face had a reflectance of 88%.
[0062]
When one sample of this laser bar was used for end face analysis and XPS measurement was performed in the same manner as in Example 1, neither Ga—O nor As—O was detected.
Further, when one device was taken out from the laser bar as an analysis sample and the energy of the lost electrons was analyzed in the same manner as in Example 1, the band gap near the end face of the InGaAs quantum well layer was found from the loss peak caused by the interband transition. The band gap in the vicinity of the end face of the GaAs optical guide layer was measured to be 1.5 eV and 1.65 eV. The energy gap between the quantum levels of the InGaAs quantum well active layer at room temperature obtained from photoluminescence measurement is 1.29 eV, and the band gap of GaAs is 1.41 eV. And the end face was confirmed to be transparent to the oscillation wavelength.
[0063]
Further, five devices were taken out from the laser bar as a life test sample separately from the analysis sample, placed on a heat dissipating submount, and packaged in a nitrogen atmosphere. As an average initial characteristic of the device, a threshold current was 25 mA at 25 ° C., and kinks were observed at 359 mA and 240 mW. As a result of the life test at 200 mW and 50 ° C. performed on this group, as shown in FIG. 4, there was no sudden death when 2000 hrs passed and stable operation was confirmed.
[0064]
When one of the fabricated devices was observed with a TEM in exactly the same manner as in Example 1, it was confirmed that the crystallinity near the active layer was broken near the end surface irradiated with Ar plasma, and disordering occurred. .
[0065]
(Comparative Example 1)
For both the front end face and the rear end face, the formation of the Si passivation layer (14) and the Ar plasma irradiation prior to that are not performed, and the coating layers (15) and (16) are formed without using the IAD method. Except for the case where the beam evaporation method was used, the same as the above example, the average initial characteristic of the device was 25 ° C., the threshold current was 23 mA, and the device was kink at 350 mA and 250 mW. However, in the life test, all 10 devices suddenly died as shown in FIG. 5 before 100 hours passed.
[0066]
Further, XPS analysis, electron energy loss spectroscopic analysis, and TEM observation were performed in exactly the same manner as in Example 1. The presence of Ga—O in the end face, and the band gaps of the quantum well layer and the light guide layer were bulky even in the vicinity of the end face. It was confirmed that it was the same as the region, and furthermore, the appearance of the crystal near the end face was the same as the single crystal in the bulk region.
(Comparative Example 2)
When the front end face and the rear end face were exactly the same as those in Example 1 except that the Si passivation layer was not formed, the average initial characteristic of the device was 25 mA at 25 ° C. and the threshold current was 350 mA. Kinks were observed at 250 mW. This was the same result as in Example 1. However, in the life test (200 mW, 50 ° C.) performed on the group of 10 devices, 6 devices were confirmed to have sudden death when 2000 hrs passed as shown in FIG. In addition, the deterioration rate was larger than that in Example 1.
[0067]
(Comparative Example 3)
Except that the Ar plasma irradiation was not performed on the front end face and the rear end face prior to the formation of the Si passivating layer, the same as in Example 1 was performed. Was 23 mA, and kinks were observed at 350 mA and 250 mW. This was the same result as in Example 1. However, in the life test (200 mW, 50 ° C.) performed on the same number of 10 device groups as in Example 1, all devices suddenly died when 250 hours passed as shown in FIG.
[0068]
Further, XPS analysis, electron energy loss spectroscopic analysis, and TEM observation were performed in exactly the same manner as in Example 1. The presence of Ga—O in the end face, and the band gaps of the quantum well layer and the light guide layer were bulky even in the vicinity of the end face. It was confirmed that it was the same as the region, and furthermore, the appearance of the crystal near the end face was the same as the single crystal in the bulk region.
(Comparative Example 4)
The InGaAs quantum well active layer measured by electron energy loss spectroscopy was used in the same manner as in Example 2 except that the front end face and the rear end face were not subjected to the formation of the Si passivation layer and the Ar plasma irradiation prior thereto. The band gap of the end face was 1.28 eV, which was almost the same as the bulk region value measured by photoluminescence. As shown in FIG. 8, the results of the life test showed that all five devices suddenly died when 100 hrs had elapsed.
[0069]
Further, XPS analysis and TEM observation were performed in exactly the same manner as in Example 1, and it was confirmed that Ga—O was present on the end face and the appearance of the crystal near the end face was the same as that of the single crystal in the bulk region.
[0070]
【The invention's effect】
The present invention stably suppresses the interface state density at the end face of a semiconductor light emitting device such as a semiconductor laser over a long period of time, and operates stably even when the passivation layer (14) diffuses. Can be realized by a simple manufacturing method, and provides a great industrial advantage.
[Brief description of the drawings]
FIG. 1 is a perspective view of a semiconductor light emitting device of the present invention.
FIG. 2 is a cross-sectional explanatory view of the semiconductor laser according to the first embodiment of the present invention as viewed from the resonator direction.
FIG. 3 is a result of a life test (200 mW constant output, 50 ° C.) of the semiconductor laser of Example 1 of the present invention.
FIG. 4 is a result of a life test (200 mW constant output, 50 ° C.) of the semiconductor laser of Example 2 of the present invention.
FIG. 5 is a result of a life test (200 mW constant output, 50 ° C.) of the semiconductor laser of Comparative Example 1 of the present invention.
6 is a result of a life test (a constant output of 200 mW, 50 ° C.) of a semiconductor laser of Comparative Example 2 of the present invention. FIG.
7 is a result of a life test (200 mW constant output, 50 ° C.) of a semiconductor laser of Comparative Example 3 of the present invention. FIG.
FIG. 8 is a result of a life test (200 mW constant output, 50 ° C.) of a semiconductor laser of Comparative Example 4 of the present invention.
[Explanation of symbols]
1: Substrate 2: Buffer layer 3: First conductivity type cladding layer 4: Active layer 5: Second conductivity type first cladding layer 6: Second etching prevention layer 7: First etching prevention layer 8: Second conductivity type first Bi-cladding layer 9: Current blocking layer 10: Cap layer 11: Contact layer 12: Electrode 13: Electrode 14: Deactivation layer 15: Coating layer 16: Coating layer

Claims (21)

A compound semiconductor layer including a first conductive type cladding layer, an active layer, and a second conductive type clad layer is formed on a substrate. Then, a cavity facet is formed by cleaving in the air or in a nitrogen atmosphere.Then, the facet is irradiated with plasma in a vacuum, and the plasma-irradiated face is continuously subjected to Si, Ge, A method of manufacturing a semiconductor light emitting device, comprising forming an inactive layer made of S and / or Se.  The method of manufacturing a semiconductor light emitting device according to claim 1, wherein the cleavage is performed at normal pressure. The method of manufacturing a semiconductor light emitting device according to claim 1 or 2, characterized in that to form the passivation layer in a vacuum of 10 ^-3 Torr. The method of manufacturing a semiconductor light-emitting device according to any one of claims 1 to 3, characterized in that forming the passivation layer of amorphous Si. The front passivation layer, its thickness Tp (nm) is a semiconductor light emitting according to any one of claims 1 and forming As the range represented by the following formula (I) 4 Device manufacturing method.
0.2 (nm) <Tp (nm) <λ / 8n (nm) (I)
(In the formula (I), λ represents the oscillation wavelength of the semiconductor light emitting device, and n represents the real part of the refractive index at the wavelength λ of the passivation layer.) The method of manufacturing a semiconductor light-emitting device according to any one of claims 1 to 5, wherein the plasma irradiation is a plasma irradiation of Group 18 element. The method for manufacturing a semiconductor light emitting device according to claim 6, wherein the group 18 element is Ar. 8. The method of manufacturing a semiconductor light emitting element according to claim 7 , wherein the energy of Ar plasma to be irradiated is 25 eV or more and 300 eV or less. After formation of the passivation layer, the semiconductor light-emitting device according to any one of claims 1 to 8 and forming a coating layer further comprising a dielectric or a dielectric and semiconductor combinations on their surface Manufacturing method. The method of manufacturing a semiconductor light emitting element according to claim 9 , wherein the coating layer is formed in vacuum after the formation of the passivation layer. When forming the coating layer, in a vacuum, a method of manufacturing a semiconductor light emitting device according to claim 9 or 10, wherein the performing raw material supply and the plasma irradiation of the coating layer to the end surface at the same time. 12. The method of manufacturing a semiconductor light emitting element according to claim 11, wherein the plasma irradiation is a group 18 element plasma irradiation. The method for manufacturing a semiconductor light emitting device according to claim 12, wherein the group 18 element is Ar. 14. The method of manufacturing a semiconductor light emitting element according to claim 13 , wherein the energy of Ar plasma to be irradiated is 25 eV or more and 300 eV or less. 15. The coating layer according to any one of claims 11 to 14 , wherein the coating layer is composed of one or a combination of two or more selected from the group consisting of AlOx, TiOx, SiOx, SiN, Si, and ZnS. A method for manufacturing a semiconductor light emitting device. 16. The method of manufacturing a semiconductor light emitting device according to claim 11 , wherein the coating layer includes a low reflection coating layer and a high reflection coating layer. 17. The method of manufacturing a semiconductor light emitting device according to claim 16, wherein the low reflection coating layer includes AlOx, and the high reflection coating layer includes AlOx and Si. The method of manufacturing a semiconductor light-emitting device according to any one of claims 1 to 17 wherein the active layer is characterized by containing In and / or Ga. The active layer is Al _x Ga _1-x As (0 ≦ x ≦ 1), In _x Ga _1-x As (0 ≦ x ≦ 1) or (Al _x Ga _1-x ) _y In _1-y P (0 19. The method of manufacturing a semiconductor light-emitting element according to claim 18 , wherein ≦ x, y ≦ 1). The method of manufacturing a semiconductor light-emitting device according to any one of claims 1 to 19, characterized in that the conductivity type of the active layer and p-type. The method of manufacturing a semiconductor light-emitting device according to any one of claims 1 to 20, characterized in that the resonator end face is a (110) plane or crystallographically equivalent plane.

Images