NL8201226A - LIGHT-EMITING SEMICONDUCTOR DEVICE. - Google Patents

Description

-—- * .- * hpw · '' -—1 -, -..- :: .---- - ---------------------- ------ ------------ -J · '' “- · - - ____ ^ J.

t1 * -1-VO 3223

Light-emitting semiconductor device.

The invention relates to a light-emitting semiconductor device, such as a laser and an LED, and more particularly to determining the current in these devices.

Nearly 20 years ago, light emitting semiconductor devices, in particular those having a planar pn junction in a monocrystalline semiconductor body, used large area electrical contacts on opposing major surfaces of the body to apply bias in the forward direction and a pump flow at the junction. With an LED, the resulting radiation recombination of holes and electrons in the active region near the junction caused spontaneous radiation. In essence, a major modification turned the LED into a laser: on the semiconductor body, a cavity resonator was formed by a pair of parallel-split crystal facets perpendicular to the junction. When the pump current exceeded the weld threshold, the spontaneous radiation injected into the LED essentially isotropically from the active region was converted into stimulated radiation, which in the laser as a collimated beam, parallel to the junction and along the centerline of the resonator was emitted. Of course, other design considerations played a role in the transition from LED to laser, but these factors will not be discussed here, as it is currently the intention * only to point to the currently known relationship between p-n junction lasers and LEDs.

The large area contacts * (e.g., 25 µm width) in these devices caused the pump current density at pn junction to be relatively low, which therefore meant that relatively large currents (e.g., several hundred mA in lasers) ) were required to obtain desired radiation energy levels. Large currents, in turn, heat the semiconductor body and require coupling of the device with a suitable hot water sink and / or operation of the device at cryogenic temperatures. The main solution to this problem has been and still is to reduce the area of the p-n junction, which must be pumped, so that for a given current density the amount of pump current required is proportionally less. In realizing this solution, the pump current is forced into a relatively narrow channel ("eg 1 µm width from a major face of the semiconductor body to flow through the active region.

One of the earliest constructions to force current into such a 5 channel was the strip geometry contact, proposed for the first time in semiconductor lasers by R.A.Furnanage and D.K. Wilson in U.S. Patent 3,363-195 - The stripe geometry reduces the threshold current for a laser operation (compared to lasers with large area contacts) and limits the spatial width of the output beam. Since that early proposal, numerous laser configurations have been designed to realize the strip geometry concept: (1) the oxide strip laser; (2) the proton-bombarded laser; (3) the mesa strip laser; (b) the reverse biased p-n junction isolation laser; (5) ribbon waveguide lasers; and (6) buried heterostructures of different types.

The most commonly used configuration over the past eleven years, however, has been the proton-bombarded Ga HetalGaAs dual heterostructure (DH) laser, which has been produced, for example, by ICCasey Jr. and M.B. Panish is described in Heterostructure Lasers, Part B, pages 207-210, Academy 20 Press, Inc., New York, (1978). Despite its various shortcomings, lasers of this type regularly exhibit an intended life of more than 100,000 hours, while some have even exceeded 1,000,000 hours (based on accelerated aging tests). Long life has also been obtained with DH LEDs, using different contact geometries (eg, punctiform or annular configurations), but a similar proton bombardment to capture the flow channel.

Some of the shortcomings of proton-bombarded DH lasers have been discussed by R.W. Dixon and others in The Bell System 30 Technical Journal, vol.59, no.6, pages 975-985, 1980. They experimentally investigated optical non-linearity (presence of "kinks" in the luminous flux- (Ll ) characteristics). and the threshold current distribution of proton bombardment-delineated AlGaAs strip geometry D-Rs as a function of the strip width (5.8 and 12 µm in cases where the protons do and do not penetrate the active layer. They showed that a shallow proton bombardment with appropriately narrow strips (e.g. 5 µm) can lead to a satisfactory optical linearity 8201226-3 (kinks are driven to non-disturbing levels with high current), without the threshold problem This is associated with narrow strip lasers in which the protons penetrate the active layer On the other hand, the lasers with such narrow strips have a statistically significant, though not demonstrably fundamental, decrease in life, and because the protons are not active layer, increases the capacity of the device and therefore reduces the response speed, while additionally increasing the lateral current distribution and therefore the spontaneous emission increases.

In digital systems, the latter requires a larger modulation current * to obtain a predetermined quenching ratio or a lower quenching ratio for a predetermined modulation current.

According to the invention, a satisfactorily good optical linearity, small capacitance and low spontaneous emission levels in a proton bombardment-delineated strip geometry GaAs-AlGaAs-DH laser are obtained by a current determination system, with the current channel at the top at the p-side contact narrow and wider at the bottom of the active layer. The construction applies to other systems, both LEDs and lasers, and to a wide variety of configurations that differ from the DH configuration.

In an illustrative embodiment of a light-emitting semiconductor device according to the invention, a semiconductor body comprises an active region in the body and determining members, through which a current flows from a surface of the body to the active region, whereby a radiation recombination of holes in the active region electrons take place. The determining members are located in the semiconductor body and form a current channel, which is narrow at the top thereof at the surface and wider at the bottom thereof at the active region. In a particular embodiment, the cross-sectional determiners form a trapezoidal channel. In another embodiment, the determining members form a coupled pair of axial channels of different widths, with the narrowest channel located at the surface and the widest channel located at the active region.

In an illustrative embodiment of the invention, the flow sensing means comprises first means having an elongated groove (eg, a V-groove) in the major plane, which form a relatively narrow first channel, which is diseased in the body to a depth. extends a short distance from the active region, and second members (eg proton-bombarded zones), which form a relatively wider second channel, extending from at least that depth into or through the active region. In an alternative embodiment, first members comprise areas of high specific resistance at the major surface which define at least a portion of the beveled sides of the V-groove, i.e., the V-groove extends through these regions. In yet another embodiment, the V-groove is refilled with semiconductor material 10.

Another aspect of the invention is a method by which such a trapezoidal channel device is produced by particle bombardment. In the method, a removable semiconductor layer is first grown on the major surface of the body epi-taxially, and then the layer is exposed to a preferred etchant, which opens up inverted trapezoidal strips in the layer. The remaining parts of the layer form a trapezoidal cushioning mask (in cross section). When the masked surface is subjected to a particle (eg proton-oxygen) bombardment, zones of high specific resistance are formed in the parts of the body between the masks and under the bevels of the trapezium. These zones define the flow channel and give it the desired trapezoidal shape: narrow at the top near the surface and wider at the bottom near the active area. Before the body is metallized to form electrical contacts, the mask is removed. For this purpose, the mask is preferably made of a material different from the portion of the body at the surface, so that a "stop etching method" for removing the mask can be used.

The invention will be explained in more detail below with reference to the drawing. For the sake of clarity, the figures are not drawn to scale herein, while parts common to the various figures are given the same references. In the drawing: Figure 1 shows an isometric view of an embodiment of a light-emitting semiconductor device with a trapezoidal current channel according to the invention; 8201226 - Figure 2 is an end view of another embodiment of a light-emitting semiconductor device with a trapezoidal current channel; Figure 3 is an end view of yet another embodiment 5 of a semiconductor Hi-Lo light-emitting device with a pair of stacked channels according to the invention; Fig. k is an end view of a mask assembly for manufacturing a light-emitting device with a trapezoidal flow channel according to the invention; Figures 5 and 6 show end views of other masks for the manufacture of devices according to the invention by means of a pro- tection attachment; Figure 7 is an isometric view of an embodiment of a trit-finding semiconductor laser or LED; Figure 8 shows a cross-section of another embodiment of a laser or LED according to the invention, wherein the V-groove extends through an area with great specific resistance; and figure 9 shows a cross-section of yet another embodiment. of a laser or LED. according to the invention, wherein the Y-groove is again filled with semiconductor material.

Figure 1 shows a light-emitting semiconductor device (laser or LED) provided with a semiconductor body 11 with an intermediate region 1L. The region 1L, which may have one or more layers, comprises an active region, which emits radiation 22 when a pump current is supplied to this region. Electrode members, indicated as contacts 16 and 18 on body 11 for illustrative purposes, together with a voltage source 20 serve to supply the pump current. Furthermore, the body 11 includes determining members 32 which cause the pump current to flow through the active region through a relatively narrow channel 36 from the top contact 30, after which the current can spread to the bottom contact 18.

Before the invention will be discussed in detail, it is preferable to first discuss the general contributions of a preferred configuration of a light-emitting semiconductor device known as a double heterostructure (Di). As indicated in Figures 1, 2 and 3, includes a DH. first and second semiconductor cladding layers 10 and 10, respectively. 12 with relatively large energy jump and opposite 8201226 4-type jelly type, and in fiber vessel grid, adapted thereto is an intermediate region 1, which is located between and adjacent the layers. The intermediate region 11 + comprises an active layer with a smaller energy jump, which has been indicated here as being coextensive with the region 1U and which may emit many layers of radiation when the fence layers are biased in the transmission direction. From the point of view of quantum efficiency, it is known that the active layer is preferably a direct energy jump semiconductor. The layers 10 ,. 12. and 1U may consist of materials, many of which are selected from a number of systems, such as eg GaAs-AlGaAs or GaAsSb-AlGaAs for small wavelength operation in the range of approximately 0.7-0.9 µm, and InP-InGaAsP or InP-AlGalnAs for wavelength bending, many are greater than approximately 1 µm (e.g., 1.1-1.6 µm}.

The voltage source 20 biases the sternal layers in the pass-through direction and thereby injects carriers into the active layer. These carriers recombine to provide a spontaneous radiation in the case of an LED and a predominantly stimulated radiation in the case of a laser. In any case, however, the radiation has a wavelength, which corresponds to the energy jump of the material. of the active layer. In addition, in the case of a laser or edge-emitting LED, as shown in Figure 1, the radiation 22 is emitted in the form of a beam along the axis 23. In the laser, the beam is collimated and the axis 23 extends perpendicular to a pair resonator mirrors 2k and 26, which are illustratively formed by cleaved crystal facetted and or etched surfaces. These mirrors form optical feedback devices for the stimulated radiation. In other cases, eg integrated optics, refractive grids can be used as a substitute for one or both mirrors.

Although the electrode members indicated in the laser or edge-emitting LED of FIG. 1 comprise contacts 16 and 38 having a large area, it is known that these contacts can be patterned to provide different geometric shapes . For example, in the case of a transversely emitting LED, in which the output light emerges perpendicular to the layers, the contact 36 more particularly consists of a large area surface, but. the contact 18 may consist of an unhealed ring, which stores an unhealed etched opening in one side of the body 13. Where the bottom 8201226 portions ("e.g., the substrate) of the body 11 are absorbent, this etched aperture is used to agitate radiation to an unfinished optical fiber which heys into the aperture.

The conductivity type of the active layer is not critical.

The layer can be of the n-type, p-type, intrinsic type or compensated type, since in certain modes of operation with forward bias the number of injected carriers may be greater than the dopant level of the active layer. In addition, the intermediate region 14 may comprise a number of layers which form an active region, e.g.

10 contiguous layers of the p-type and n-type with the same energy jump forming a p-n homojunction, or with different energy jumps forming a p-n heterojunction. Furthermore, the hetero structure can have configurations other than the simple double hetero structure, including, for example, but not limited to, separate heteroherostructures, such as described in U.S. Patent 3,691. ^ + 76, strip-buried heterostructures of the type disclosed in U.S. Patent 4,190,813 and isotype heterostructures disclosed in U.S. Patent Application Serial No. 050,637.

For a CW laser operation at room temperature, the thickness of the active layer is preferably between approximately f2 and 1.0 µm, the wavelength of the radiation measured in the semiconductor. More specifically, for operation at low thresholds, the thickness is 0.12 to 0.20 µm. However, a thicker active layer, more particularly with a thickness of 2-3 µm, is suitable for LED operation.

In any case, for operation at room temperature, the laser or the LED is more particularly connected to a suitable heat sink, not shown.

In practice, the layers of a double heterostructure 30 are more particularly grown by an epitaxial process, such as a liquid phase epitaxy (LPE), a molecular beam epitaxy (MBE), or a metallo-organic chemical vapor deposition (MO-CVD). Epitaxial growth takes place with a monocrystalline substrate 28, which may be provided with an unimaged buffer layer between the substrate 28 and the first coating layer 10. Furthermore, as shown in Figures 1 and 3, an optional contact-simplifying layer 30 is between the second coating layer 12 and the top contact 16 are present. Opposite i 8201226 -8- ¢. Contact 18 is formed on the underside of the substrate 28.

As already mentioned, in order to flow the current generated by the pump flow 20 through a relatively narrow channel 36 into the active region, determination members 32 are present in the body 11, i.e.

5 that in the semiconductor layers, to illustrate the layers 10, 12, 1k and 30, zones 32 with high specific resistance are formed in a known manner. For example, methods of forming the zones 32 include a proton bombardment, an oxygen bombardment, or an appropriate etching and regrowth of material having high specific resistance. For example, zones 32 have a specific resistance of the order of 10-10 ohm-cm. while channel 36 has a specific resistance of only 0.1 ohm-cm, so that typical specific resistance ratios are in the range of 10 °: 1 to 10: 1.

Trapezoidal channel configurations In an illustrative embodiment according to the invention shown in Figure 1, the current lifting means 32 form a current channel 36 with relatively large conductivity, which is narrow at the top thereof at the major plane (width S ^) and at the bottom thereof at the top. active region (ie the layer I) is wider (width S2) The determination members 32 comprise laterally separated regions 32.1 and 32.2 with high specific resistance, which delimit the channel 36 along its oblique sides ~ 36.1. straight lines are indicated, in practice no linear relationship is necessary and it may also not occur in the actual treatment steps It has been found that the above flow channel shape has an important influence on the operation of the The smaller channel width at the top increases the current density and thereby the energy at which "kinks" occur . The depth of the areas of high specific resistance, which preferably extend through the active area I 30, affects the capacity of the device and the amount of spontaneous emission from a laser. These factors will be discussed in more detail later.

As indicated in Figure 2, the channel 36 formed by the regions 32.1 and 32.2 of high specific resistance need not reach the main plane kk. However, so that the resistance of the device is not too great, a dopant may be diffused into the surface Uk or introduced in some other way therein, in order to avoid 8201226 ....... ...... Provide a highly conductive diffusion front stone k5 extending through channel 36. In this case, the width at the top of the channel 36 is determined by the intersection of the front k-5 and the bevel sides 36.1.

The implementation of the determination members 32 according to the invention need not be limited to configurations in which the channel has a trapezoidal shape. In the Hi-Lo constructions discussed in the next section, the determiners 32 may form a coupled pair of stacked channels.

Further, although the trapezoidal channel 36 of Figure 1 essentially forms a parallelepiped extending parallel to the centerline 23, in the case of a transversely emitting LED, the channel 36 may be in the form of a truncated cone, the centerline of which is perpendicular on the layers.

15 Hi-Lo constructions

In this embodiment of the invention, shown in Figure 3, the current determining means 32 has a two-level or stepped configuration in which a pair of coupled channels 36a and 36b is formed. More specifically, members 32 include first members 32.1a-20 32.2a defining a relatively narrow top channel 36a and second members 32.1b-32.2b defining a relatively wider bottom channel 36b. By way of illustration, the determining members 32 comprise regions 32.1-32.2 with high specific resistance, which delimit channels 36a and 36b with relatively large conductivity. The areas 32 include (1) upper zones 32.1a and 32.2a and lower zones 32.1b and 32.2b. The upper zones are separated by a relatively small distance and extend from the upper major surface W of the body 11 over a depth d1 to a small distance from the active area, thereby defining the narrow upper channel 36a. In contrast, the bottom zones are separated by a relatively greater distance and extend from the depth d in or through the active region, defining the wider bottom channel 36b.

As before, channels 36a and 36b may be approximately parallelepipedas extending perpendicular to the plane of the drawing, such as with a laser or round-emitting LED, while with transversely-emitting LED they may be cylinders , which extend transversely to the layers.

8201226 -1Q- £

When the regions 32 of high specific resistance were manufactured by proton bombardment with GaAs-AlGaAs lasers, this II-Lo construction showed a number of drawbacks. First, the narrow top channel 36a increased the current density in the active region, thereby causing Tkinks'T to be shifted to satisfactorily high current levels outside the range of a typical laser operation, compared to wide range DH lasers (e.g. 12yum). flow geometry. Second, this property also led to lasers with more uniformly distributed and lower laser thresholds, resulting in a higher yield. Third, because the wider bottom channel 36b reduced lateral flow diffusion and spread, less spontaneous radiation was emitted outside the resonator of the laser, allowing for lower minimum modulation currents for predetermined quenching ratios in digital applications. the latter property resulted in reduced device capacity for both lasers and LEDs, allowing for greater operating speed (ie, a greater pulse repetition frequency in digital applications) .-

In order to reduce the capacity of the device, the proton bombardment must penetrate the pn junction, which in a usual DH is located at one of the interfaces between the active layer 1U and the coating layers 10 and 12. However, to avoid the spontaneous emission the protons should preferably penetrate the active region where a recombination takes place.

25 Manufacture of trapezoidal channels

As indicated in Figure k, a way of providing a trapezoidal channel of the type shown in Figure 1 is to epitaxially grow a removable semiconductor layer on the major surface and a known photolithographic and preferred etching method to provide a layer to provide a certain pattern to form inverted trapezoidal openings 5-1, which expose portions of the surface W. Between the openings, the remaining segments * 52 of the removable layer form trapezoidal cushioning masks. For a semiconductive layer consisting of a material of the group 35 HI-V, the oblique side walls 56 of the resistive segments correspond to (11 IA) crystallographic planes, making an angle of about 55 ° with a (100) oriented surface 1Λ.

8201226 -11-

It is also possible to etch the openings in the removable layer as inverted trapezia so that the remaining segments are 52 trapezia. In both cases, therefore, the trapezia and inverted trapezia are complementary.

5 A bombardment of the masked surface UU with particles 50 (eg protons, oxygen) leads to the deepest proton penetration between the segments, no penetration below the central (thickest) parts of the segments and a gradually decreasing penetration under the bevel of the segments. Obviously, a thinner mask segment 10 allows some protons to penetrate below the central parts of the segments, a method which would be useful in realizing the channel configuration of Figure 2.

After the bombardment is completed and before a metallization takes place to form the electrical contacts, the damping masks are removed. For this purpose, it is preferred that the material of the mask 52 be different from the material of the portion of the body 11 located at the surface UU so that a "stop etching method" can be successfully applied . For example, the surface kk more particularly consists of GaAs, in which case the mask '52 may consist of AlGaAs and a known HF etchant or iodine (eg.

113 g KI, β5 g 5 5 cc HgO) as a stop etchant for removing mask 52 can be used. A plasma stop etching treatment can also be used as a substitute for wet chemical procedures. Finally, it is noted that a buffered peroxide solution is also a preferred etchant and can be used to etch the openings forming the mask segments 52.

The formation of the removable layer also provides an edge advantage related to the purity of the epitaxial growth process. When a liquid phase epitaxy is used to manufacture the semiconductor layers of these devices, the last grown layer is more particularly contaminated from various sources, more particularly spheres of the molten metal (eg Ga), which as the source solution. Therefore, this last layer, which is usually the cap layer or the contact-facilitating layer 30 (FIG. 1-3), must be cleaned by etching, a step that requires careful control since the layer 30 is more in particular is very thin · (eg 0.5 µm). However, in the process described here, the last layer grown is the damping mask, which can be much thicker (eg, 3sCymnI and can be easily removed by a stop-etching method, as mentioned above *).

Manufacture of Hi-Lo Structures A number of manufacturing methods can be used to provide a Hi-Lo construction according to the invention. As already mentioned, the high specific resistance regions 32 may be formed by proton bombardment, an oxygen bombardment or an etching and regrowth of high specific resistance material. It will be assumed, however, that these regions are formed by a proton bombardment.

A direct method involves two proton homeshapings and two masks. In the first step, a proton damping mask with a width and protons with an energy (eg.

150 keV) used to delineate the narrow top channel 36a. In the second step, a proton damping mask with a width Sg and protons with an energy Eg ^ (eg Eg = 300 keVj is used to determine the wider lower channel 36b.

A delineation of channels 36a and 36b in a single proton bombardment step is also possible. For this purpose, a composite damping mask with a larger proton damping in the middle and a smaller damping on the sides can be used. Two versions of this type of mask are shown in Figures 5 and 6. In any case, a thick metal body 1 + 0 with a width is formed at the top of a plateau b2, which in turn is formed on the main surface bb, which closest to the active region 1U is formed. The body essentially damps the protons 50 completely, so that no proton damage occurs in the narrow channel 36a, and the plateau b2 only partially damps the protons 50, so that damaged zones 32.1a and 32.2a extend to a depth of 30 d extend a small distance from the active area. Outside the plateau b2, the mask essentially does not provide cushioning either in Figure 5 (because the mask does not extend as far \ or in Figure 6 (because the mask is very thin there). Thus, outside the plateau b2, the areas damaged by protons extend 32.1 b and 32.2b extend over a depth dg and penetrate through the active region 11. Preferably, as indicated, these damaged zones 32.1b and 32.2b extend through the active region 1b. and 6, the body Uo includes galvanized Au for illustrative purposes 8201226 -13. The plateau 1 * 2 in Figure 5 includes layers of Au (1 * 2.1 ,.

Pd or Pt (1 * 2.2 and Ti (1 * 2.3) and in Figure 6 the plateau comprises a mesa of SiOg (1 * 2.1 *) on which are found Ti-Pt layers (1 * 2.5).

The examples below describe in more detail how masks of this type were used to manufacture light-emitting devices. Unless otherwise stated, numerical parameters and different materials are mentioned for illustrative purposes only, and are not intended to limit the scope of protection. In each of the two examples, the semiconductor body 11 included a (100) oriented n-GaAs substrate 28 on which the following epitaxial layers were grown by a standard LPE: an n-GaAs buffer layer (not shown; an n-Alg coating layer 10 having a thickness of about 1.5 µm, a p-Alqqqq active layer 1U with a thickness of about 0.15 µm, and a 15-pq Alq 2 ($ Gaq. Ash existing cladding layer 12 having a thickness of about 1.5 µm, and a highly doped p-GaAs cladding layer 30 having a thickness of 0.5 µm. The finished wafer (body 11 plus epitaxial layers) was treated as follows to prepare light-emitting devices, in particular lasers.

Example I

For fabricating lasers using the composite attenuation mask 1 * 0-1 * 2 of Figure 5, a detachable photoresist mask was applied to the surface 1 * 4 and normal photolithographic methods were used to form an elongated strip-shaped window with a width 25 of 12yum or l8yum perpendicular to the (110) gaps. Ti, Pd, and Au layers 1 * 2.3, 1 * 2.2, and 1 * 2.1 were sequentially applied using a vacuum E gun system. The deposition rate was controlled by a commercially available control system such that the Ti, Pd and Au layers, respectively. a thickness of 1000 $, 1500% and 5000 & 30. had. The total thickness of 0.75 µm for the plateau 1 * 2 was chosen to provide a 50% reduction in the penetration depth of protons 50 of 300 keV. The strip geometry tray 1 * 2 was then formed by known etching methods for removing the photoresist mask.

Subsequently, the body 1 * 0 was also formed in the form of a strip with a width of 5 µm by electro-galvanizing Au to a thickness of about 1-2 µm using a standard 8201226% -1 * 14 - male photolithographic procedure. The Au body 40 essentially provided a full barrier to the high energy protons 50 (300 keV; dose 3x10 cm), thus providing the narrow top channel 36a with a width S ^ = 5 µm and the wider bottom channel 36b with a width 5 S2 = 12 µm or 18 µm was formed. Between and S2, the plateau 42 provided only partial attenuation, so that protons penetrated to a depth d / 1.5 / um. Outside of Sg, no damping mask was present and protons penetrated to a depth dg = 2.8 µm and therefore through the active layer 14.

Example II

To simplify the manufacturing process of Example I, the Ti-Pd-Au plateau 42, as shown in Figure 6, was replaced with a dielectric strip 42.4 (e.g., from SiOg or Si ^ N ^ 1, on which a Ti-Pt coating 42.5. This composite mask was fabricated by depositing 44 SiOg having a thickness of about 1.0-1.2 µm on the surface using a normal vapor phase method. to impart a 50% attenuation to the 300 keV protons 50. Then, the SiOg layer was delineated photolithographically and etched in normal buffered HF etchant to form strips, 12 µm or 18 µm wide. µm perpendicular to the (110) splitting surfaces. After removing the photolithographic mask, the SiOg strip 42.4 and the surface 44 were coated with Ti to a thickness of 1000 lb and then with Pt to a thickness of 1500 lb using a normal vapor deposition process Finally, the body 40 was formed in The shape of a 5 µm strip with a thickness of 1-2 µm using normal photolithographic and electroplating methods. As before, the masked platelets were subjected to 300 keV protons

c - 2 at a dose of 3x10 cm-1 to form the narrow top channel 36a and the wider bottom channel 36b simultaneously. In this case, the layer 42.5 reduced the proton energy such that dg decreased to about 2.3 µm.

In both geyses I and II, after the proton homage was completed, the composite masks 4Q-42 were removed from the surface 44 by an HF etchant. This step also prepared the surface 44 for a subsequent metallization to form the normal p-metal contacts.

8201226 Η - '^ - _ · -15-

Experimental Results - Hi-Lo Constructions

To provide a standard for comparison, half of each wafer "at Fores I and II was processed into 5 µm wide strip lasers using a shallow proton-5 bombardment (150 keV). Any remaining half-plate, as mentioned above, was processed into Hi-Lo lasers using composite masks 1 + 0-42 of three types: type (1) - Au body 40 with a width of 5 µm on a SiOg / Ti Pt plateau.42 with a width of 18 µm (example III); type (2) - Au body 40 with a width of 5 µm on a SiO2 / Ti-Pt-10 plateau 42 with a width of 12 µm (example II) and type (3) - Au body 40 with a width of 5 µm on a Ti-Pd-Au plateau 4θ with a width of 18 µm (example I).

Comparisons, indicated in the table below, were based on a number of parameters: the spontaneous emission power at a drive current of 50 mA; the slope As ^ of the spontaneous emission portion of the L-I curve; the capacity C, measured at 1 MHz (the average C is below specified); and the minimum modulation current MMI, which is determined as the difference in current between upper and lower light energy levels Pg, respectively. P ^, whereby a light intensity-20 quenching ratio E ^ between the IN and OUT states is obtained when the laser is pulsed (the mean MMI is stated below for Ε ^ = 15: 1, P2 = 2.5 mW and P ^ 0.167 mW).

25 1 8201226 * t v> • J6-;

O 3 O CQ

grave i »H f 4 * Ö · -> £»

Kr) 3 3 -333¾ cd S S w Ct · C. - q co ui u) \ o o o 4

«V» fD

ΓΟ Γ0 0 * 3

VO O CD

O H

!> and

Γ0 -P "cf B

- »ui o o S td v # v» 1 PO - * CD t-3

CO ui K.

- Ξ s - * —3 UI CTS O O 4

j y V »CD

onion ro —10 * 3

O 35 * CD

ΓΟ H

onion ro UI -P "O O dr

v »* <J

00 * d

-3 3 = - CD

ro. * -> ro onion —3 + r- O O O 4

• <Λ V * CD

UI "· * O 0 * 3 ro —1 cd

- * H

- * ro.

ro on o o <+ 00 * d ers -p * cd Uï 8201226 is s> -17-

Almost the information indicated in the table has shown that 90% of lasers of type (2) had MMIs in a given MMI of 30 mA with a statistical variation of 2-i-3 mA. Similarly, 75% of type (3) lasers had an MMI within 30 mA, while none of the corresponding lasers had this. These results demonstrate the improved device yield.

It is noted that lasers of the type (2), which are provided with strips with a wide Sg = 12 µm, show the greatest decrease in S1 and the highest yield for an MMI ^ 30 mA, but these 10 advantages only do not necessarily prescribe the use of this strip width. Attention should be paid to the influence on the light energy output level P ^ at which "kinks" occur. Generally, it has been found that "kinks" formation occurred at a higher level in the control lasers than in Hi-Lo lasers, but the latter were still within specifications (ie Ρ ^ Λ 3 m ¥). The lasers of type (1) showed little change in P ^. However, the lasers of type (2} using the narrowest damping masks (Sg = 12 µm), showed a marked reduction of about 50% in P ^ compared to the corresponding control lasers. In contrast, lasers of type (3), where Sg = 18 µm, showed a smaller reduction. of about 35% in P. These data suggest that it is of benefit

xC

it may be that the width Sg is between 12 µm and 18 µm.

The first members defining the narrow top channel 36a can be formed by a groove etched in the top surface UU. Therefore, it is expected that a groove in combination with a wider bottom channel 36b will exhibit features and advantages comparable to those described above. The details of such a construction will be discussed with reference to Figures 7-9.

In Figure 7 there is found a light-emitting semiconductor device (laser or LED), which is analogous to that according to Figure 1-6, provided with a semiconductor body 111 with an intermediate region 11U.

The region 11k, which may comprise one or more layers, comprises an active region, which emits substantially stimulated radiation 122 in the case of a laser or spontaneous radiation in the case of an LED, when a pump current is applied thereto. Electrode members, illustrated on contacts 116 and 118 on body 111, serve to supply the pump current together with a voltage source 120. Body 111 further includes detector members 132-13U, which cause the pump current to flow through a relatively narrow channel 136-138 from the top contact 116 through the active region, whereupon the current can spread to the bottom contact 118.

In an illustrative embodiment of the invention, as shown in Figure 7, the flow determining members 132-13 include first members 13, defining a relatively narrow top channel 136, and second members 132, defining a relatively wider bottom channel 138 . For purposes of illustration, the determining members comprise a V-groove first members 13 ^ extending from the main plane 1 ^ over a depth d, a small distance from the active region, and thereby the relatively narrow upper channel 136 and further laterally separated regions 132 of great specific resistance which define the lower broader channel 138 'and extend from at least the depth d1 approximately to the active region (ie, in or through the active region). As indicated, the separated regions 132 extend from the surface 1UU for illustrative purposes and preferably through the active region. The V-groove 13 ^ is located in the space between the regions 132. However, it is not essential that the regions 132 with great specific resistance actually extend over the entire distance from the main plane 1v. Essentially, for contacting purposes, it may be advantageous that a high conductivity layer is provided between regions 132 and contact 116, as described in U.S. Patent No. 12,826.

Referring to the DH of Figure 7, the V-groove 13 ^ at the major face has a width and a depth d ^, where the groove penetrates the second layer 112, giving the upper channel 136 essentially the same width. In contrast, the regions 132 of high specific resistance are separated by a greater distance S ^ and extend from the plane 1 ^ U over a depth dg · d ^ in and preferably through the active region, thereby extending the wider determine bottom channel 138 with the width.

As shown in Figure 8, the upper channel 136 35 may be further limited by further regions 132.1 of high specific resistance, which define a portion of the bevel sides 13 ^ - 1 of the V-groove 13 ^, thus creating the width S ^ 'of the upper channel according to 8201226 * .............' ............ ë * i -19- figure 8 is smaller than in figure 7 · In in practice, regions 132 and 132.1 (e.g., by a proton bombardment) can reach depths of dg, resp. dj (dg> dj) are executed and then the V-groove 13 ^ can be etched to a depth d ^, to penetrate the regions 132.1 (dj <d ^ <dg).

These V-groove configurations are expected to provide a number of advantages. First, the narrow top channel 136 increases the current density in the active region and thereby causes "kinks" in lasers to be shifted to higher current levels outside the actual operating region. Second, this construction should also lead to more uniformly distributed laser thresholds and lower laser thresholds, thereby obtaining a greater device yield. Third, due to the fact that the wider bottom channel 138 reduces lateral flow diffusion and spread, less spontaneous radiation is emitted outside the resonator of the laser, allowing a smaller modulation current for a predetermined quenching ratio in digital applications. is. Fourth, the latter leads to a reduced capacity of the device for both a laser and an LED, allowing for a higher operating speed (i.e., a greater pulse repetition frequency in digital application).

Other constructions are also possible within the scope of the invention. More specifically, the V-groove 3¾ of Figure 7 or Figure 8 can be refilled with semiconductor material, resulting in devices (more particularly lasers) with a number of useful properties, which will be discussed later. In addition, although the groove has been described as a V-groove, its precise geometric shape is not critical. A V-groove is obtained when semiconductors of the III-V type are subjected to different etchants etching crystallographic surfaces with a particular preference, but a V-groove or rectangular groove can also be obtained with other etchants or by other processes (e.g. ion beam rolling machining or plasma etching).

As shown in Figure 9, the V-groove is filled with semiconductor material 13V and depending on the procedure used for refilling, this may or may not result in the formation of layers 13-2. at the V-groove and / or the main plane lWt · are 8201226 Γ -20-.

located. In addition, depending on the material, on the layer 112 and the type of procedure used, the material 134r may or may not be epitaxial (i.e., monocrystalline).

A number of embodiments arise, depending on the relative dimensions of the energy jumps E of the Dïï layers, • δ with respect to the V-groove material 134 '. Case I: E (13V)> Eg (114}; O * dvz, that the V-groove material 134 'has a greater energy jump than the active layer 114. As a result, laser radiation undergoes the material 134' of the V-groove. groove, reduced absorption compared to Figure 7 · Case II: E (134 ') Ë (112) Λ Ëg (ll4) iss additionally, layers 110 and 112 and active layer 114 all have the same conductivity type, while the material 134 'of the V-groove and the layer 112 have opposite conductivity type. This configuration is a form of isotype laser, the pn junction being located along the oblique planes 134.1. In this case, the material 134r of the V-groove is preferred monocrystalline Case III: E (112.) Eg (134 ').' Eg (114); ie

δ that the material 134 'of the V-groove has a smaller energy jump than the layer 112. but has a larger energy jump than the active layer 114. As a result, the refractive indices n show the relationship n (114) 20 n (134') n ( 112), so that the laser radiation is guided through the refractive index along the V-groove.

t 8201226

Claims (16)

1. Real-emitting semiconductor devices having a semiconductor body having an active region in which optical radiation is generated when a stream flows through this region, and organs within the body to channel current from a surface of the body through a channel by flowing the active region, characterized in that the channel is narrow at the major plane and wider at the active region * 2. Device as claimed in claim 1, characterized in that the determining members are provided with first means for causing the current to flow in a relatively narrow upper channel, which extends from the surface over a depth at a small distance from the active region and second means for causing the current to flow into a relatively wider bottom channel extending from the defined depth to the active region. 3. Device according to claim 2, characterized in that the second members are provided with a pair of second zones with large speci fi c. physical resistance, which limits the bottom channel. k. Device according to claim 3, characterized in that the second zones extend through the active area. Device according to any one of claims 2-¾, characterized in that the first members are provided with a pair of first zones with high specific resistance, which bound the upper channel. Device as claimed in any of the claims 3-5, characterized in that the. zones of high specific resistance comprised by protons' bombarded zones. 7. Device as claimed in any of the claims 2-6, comprising a first coating layer, a second coating layer, which is closer to the surface than the first layer, and the active area comprises an active layer between the coating layers, with the ' characterized in that the top channel extends from the surface to the predetermined depth contained in the second clad layer and the bottom channel extends from the predetermined depth through the active layer. 8. Device as claimed in any of the claims 2-7, characterized in that the top channel has a width of about 5 µm and the width of the bottom channel is between about 12 and 18 µm. Device according to any one of the preceding claims, provided with a laser resonator center line, along which radiation is propagated, 8201226 * -22- V ', characterized in that the determining members define the channels as elongated parallelepipeds, which extend essentially parallel to the center line . 10. Device as claimed in any of the claims 1-8 for use as a light-emitting diode, characterized in that determining members define the channels as cylinders which extend transversely to said area. 11. Device as claimed in any of the claims 1-7, characterized in that the determining members are provided with first elongated groove means for causing a current to flow into the relatively narrow upper channel, which flows from it. surface extends to the defined depth at a distance from the active region, and second means for causing the current to flow into the relatively wider bottom channel, which extends from at least the defined depth to the active region. Device according to claim 11, characterized in that the first members comprise a portion of the surface in which a V-groove is formed, and a pair of laterally separated first regions of high specific resistance, which form at least a portion of define the sloping sides of the V-groove so that the V-groove extends through the first regions. 13. Device according to claim 11 related to claim 7, characterized in that the groove members extend from the surface to the determined depth in the second coating layer and the bottom channel extends through the active layer from at least the determined depth. 1U. The device according to claim 11 or 12, characterized in that the first members comprise a portion of the surface in which the groove is formed and further semiconductor material filling the groove. * Device as claimed in claim 11, characterized in that the material has a greater energy jump than the adjacent parts of the body. "16. Device according to claim 11, characterized in that the material has the same conductivity type as the adjacent parts. 17. Device according to claim 1U, characterized in that the material has a conductivity type opposite to that of the adjacent parts, whereby a pn junction is formed at the sides of the groove, which serves for injecting carriers into the active area. 18. Device according to claim 1, characterized in that the material has a smaller energy jump than the adjacent parts of the body and has a larger energy jump than the active area. Method for manufacturing a device according to claim 1, with a trapezoidal current channel in a semiconductor body, wherein a semiconductor layer is grown on a major surface of the body along epitaxial path, characterized in that a particular layer pattern is given to form trapezoidal openings in the layer, the remaining segments of the layer in cross-section forming complementary trapezoidal masks, which masks have sloping sidewalls, the body is subjected to a particle bombardment in order to pass between the masks and under the slanting sidewalls form zones of great specific resistance, and the masks are removed from the surface. 20. A method according to claim 19, characterized in that the layer and the portion of the body at the surface consist of different semiconductor materials and the masks are removed from the surface by a "stop etching procedure". 21. Method according to any one of claims 19-20, characterized in that the layer and the body consist of materials of the group III-V, the surface has a (100) crystal orientation and when the layer is given a certain pattern the openings are formed by exposing the layer to an etchant, which performs a preferred etching along (111A) crystallographic planes, so that the side walls form the planes. t 8201226