Distributed feedback laser modification
Customized fabrication modifications to DFB laser cavities address facet-induced optical phase variation, improving yield and single-mode operation by compensating for threshold gain spectrum issues.
Patent Information
- Application Number
- PCT/GB2025/052234
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2024-10-11
- Filing Date
- 2025-10-13
- Publication Date
- 2026-04-16
AI Technical Summary
Variation in optical phase at the facets of DFB laser cavities leads to degraded single-mode operation due to threshold gain spectrum variation, causing yield loss in manufacturing.
On-wafer testing of DFB lasers followed by customized fabrication modifications, such as nonuniform adjustments to laser cavities through segmented electrical contacts, carrier injection, effective refractive index distribution, and stress distribution, to compensate for threshold gain spectrum variation.
Improves yield by aligning single-mode performance with parametric test specifications, reducing yield loss and enhancing manufacturing efficiency.
Smart Images

Figure GB2025052234_16042026_PF_FP_ABST
Abstract
Description
[0001] DISTRIBUTED FEEDBACK LASER MODIFICATION
[0002] The present invention relates to methods of manufacturing distributed feedback (DFB) laser apparatuses and associated distributed feedback (DFB) laser apparatuses.
[0003] Background Art
[0004] DFB lasers provide single mode operation at precisely specified lasing frequencies, which is useful in for example data communication applications.
[0005] In the field of compound-semiconductor DFB laser manufacturing, variation in optical phase in the laser cavity at the facets causes degraded single-mode operation. The underlying mechanism is threshold gain spectrum variation caused by the facet phase variation. This threshold gain spectrum variation leads to yield loss when parametric tests relating to single-mode operation are failed or the test results are out of specification.
[0006] An approach to reducing yield loss in semiconductor manufacturing is post-test modification of devices using laser trimming or cutting. However, this is a slow serial process using specialised equipment.
[0007] Summary of invention
[0008] It is desirable to provide DFB laser apparatuses and methods of their manufacture that overcome the above-identified problems.
[0009] According to a first aspect of the present invention, there is provided a method of manufacturing a distributed feedback laser apparatus, the method comprising the steps: (a) fabricating a plurality of distributed feedback lasers on a wafer substrate, each laser comprising a respective laser cavity;
[0010] (b) on-wafer testing the lasers to generate respective lasers’ parametric test results;
[0011] (c) determining, using the respective lasers’ parametric test results, different respective fabrication modifications customised to respective lasers to adjust each respective laser cavity nonuniformly to compensate for variation in the parametric test results; and
[0012] (d) performing one or more wafer-level fabrication step comprising lithographic patterning on the wafer substrate having the plurality of the lasers fabricated on it, to apply the different respective fabrication modifications to the respective lasers.
[0013] Preferably, the respective fabrication modifications are determined to compensate for threshold gain spectrum variation.
[0014] Preferably, the respective fabrication modifications are determined to adjust each respective laser cavity nonuniformly axially along the cavity.
[0015] Preferably, the respective fabrication modifications are determined to adjust each respective laser cavity nonuniformly transversely across the cavity.
[0016] Preferably, the respective fabrication modifications adjust each respective laser cavity nonuniformly by providing a nonuniform effective refractive index distribution to each respective laser cavity.
[0017] Preferably, the respective fabrication modifications adjust each respective laser cavity nonuniformly by providing a nonuniform carrier injection distribution to each respective laser cavity.
[0018] Preferably, the respective fabrication modifications adjust each respective laser cavity nonuniformly by providing a nonuniform stress distribution to each respective laser cavity. Preferably, each respective laser comprises a segmented electrical contact arranged to inject carriers to the respective laser cavity.
[0019] Preferably, the on-wafer testing is performed by providing a uniform carrier injection distribution to the respective laser cavity.
[0020] Preferably, the on-wafer testing is performed by wiring segments of the segmented electrical contact together to provide the uniform carrier injection distribution to the respective laser cavity.
[0021] Preferably, the respective fabrication modifications comprise hardwiring of segments of the segmented electrical contact to inject carriers nonuniformly to the respective laser cavity, customised for the respective laser.
[0022] Preferably, the fabrication modifications comprise addition of material to the respective laser to provide material nonuniformly to the respective cavity, customised for the respective laser.
[0023] Preferably, the respective fabrication modifications comprise addition of dielectric material, to provide a nonuniform effective refractive index distribution to the respective laser cavity.
[0024] Preferably, the respective fabrication modifications comprise implantation of material, to provide a nonuniform carrier injection distribution to the respective laser cavity.
[0025] Preferably, the respective fabrication modifications comprise addition of material, to provide a nonuniform stress distribution to the respective laser cavity.
[0026] Preferably, the respective fabrication modifications comprise removal of material from the laser to provide material nonuniformly to the respective laser cavity, customised for the respective laser. Preferably, the respective fabrication modifications comprise removal of dielectric material, to provide a nonuniform effective refractive index distribution to the respective laser cavity.
[0027] Preferably, the fabrication modifications comprise removal of material, to provide a nonuniform stress distribution to the respective laser cavity.
[0028] Preferably, the one or more fabrication step to apply the different respective fabrication modifications is performed concurrently for a plurality of the lasers.
[0029] Preferably, the step of performing one or more wafter-level fabrication step to apply the respective fabrication modifications comprises one or more step from the group of fabrication steps consisting of: patterning, deposition, implantation and etching.
[0030] Preferably, the step of performing one or more wafter-level fabrication step to apply the respective fabrication modifications comprises modifying circuitry on a sub-mount and comprises mounting the laser on the sub-mount in an assembly step.
[0031] Preferably, the method further comprises the step of re-testing the lasers having the respective fabrication modifications to generate a revised parametric test result.
[0032] Preferably, the step of re-testing the lasers is performed as a wafer-level testing step on the wafer substrate having the plurality of the lasers with respective fabrication modifications fabricated on it.
[0033] Preferably, the laser is an in-plane laser.
[0034] Preferably, further comprising the step of fabricating a lithographically patterned structure for each respective laser, and the step of on-wafer testing the lasers comprises supplying injection current uniformly along the respective laser cavity via the respective lithographically patterned structure.
[0035] Preferably, the lithographically patterned structure comprises a temporary short configured to supply injection current uniformly along the respective laser cavity by shorting together contact segments of the respective laser, and the method further comprises the step of removing the temporary short after the on-wafer testing.
[0036] Alternatively, the lithographically patterned structure comprises a temporary contact configured to supply injection current uniformly along the respective laser cavity by contacting a semiconductor layer of the respective laser, and the method further comprises the step of removing the temporary contact after the on-wafer testing.
[0037] Alternatively, the lasers are buried heterostructure lasers and the lithographically patterned structure comprises an offset contact that is offset laterally from the respective laser cavity and is configured to supply injection current uniformly along the respective laser cavity by contacting a semiconductor layer of the respective laser.
[0038] Preferably, the offset contact is not annealed before the step of on-wafer testing.
[0039] According to a second aspect of the present invention, there is provided distributed feedback laser wafer, comprising:
[0040] - a wafer substrate having;
[0041] - a plurality of distributed feedback lasers fabricated on the wafer substrate, the lasers each comprising a cavity; and
[0042] - different respective fabrication modifications applied to respective lasers post lasing test by one or more wafer-level fabrication step comprising lithographic patterning on a wafer substrate having the plurality of the lasers fabricated on it, the different respective fabrication modifications customised to respective lasers to adjust the respective laser cavities nonuniformly.
[0043] Preferably, the fabrication modifications adjust the respective lasers’ cavities nonuniformly axially along the cavities.
[0044] Preferably, the fabrication modifications adjust the respective lasers’ cavities nonuniformly transversely across the cavities. Preferably, the fabrication modifications adjust the respective lasers’ cavities nonuniformly by providing different nonuniform effective refractive index distributions to the respective cavities.
[0045] Preferably, the fabrication modifications adjust the respective lasers’ cavities nonuniformly by providing different nonuniform carrier injection distributions to the respective cavities.
[0046] Preferably, the fabrication modifications adjust the respective lasers’ cavities nonuniformly by providing different nonuniform stress distributions to the respective cavities.
[0047] Preferably, the lasers each comprise a segmented electrical contact arranged to inject carriers to their respective cavities, wherein the fabrication modifications comprise hardwiring of segments of the respective segmented electrical contact to inject carriers nonuniformly to the respective cavities, customised for each respective laser.
[0048] Preferably, the fabrication modifications comprise addition of material to the lasers to provide material nonuniformly to the respective cavities, customised for each respective laser.
[0049] Preferably, the fabrication modifications comprise removal of material from the lasers to provide material nonuniformly to the respective cavities, customised for each respective laser.
[0050] Preferably, the one or more wafer-level fabrication step comprise one or more step from the group comprising: patterning, deposition, implantation and etching.
[0051] Preferably, the lasers are in-plane lasers.
[0052] Preferably, the lasers are buried heterostructure lasers and each respective laser comprises a lithographically patterned offset contact that is offset laterally from the respective laser cavity and is configured to supply injection current uniformly along the respective laser cavity by contacting a semiconductor layer of the respective laser.
[0053] Brief description of drawings
[0054] Embodiments of the present invention will now be described, by way of example only, with reference to the drawings, in which:
[0055] Figures 1 a to 1 c illustrate, in schematic form, a known etched-facet distributed feedback (DFB) laser, in orthographic and cross-section views.
[0056] Figure 2a is a graph of photon density versus distance along the cavity for two different facet phase conditions.
[0057] Figure 2b is a graph of carrier density versus distance along the cavity for the same two facet phase conditions.
[0058] Figure 3 is a graph of side-mode suppression ratio (SMSR) versus current density for the two different facet phase conditions.
[0059] Figures 4a and 4b are graphs of threshold gain versus wavelength for the first and second facet phase conditions respectively.
[0060] Figure 5 is a flowchart of a method of manufacturing a DFB laser apparatus, in accordance with an embodiment of the present invention.
[0061] Figures 6a to 6c illustrate, in schematic form, an etched-facet ridge waveguide distributed feedback (DFB) laser, in orthographic and cross-section views, with segmented p-contacts.
[0062] Figures 6d to 6f illustrate, in schematic form, an etched-facet buried heterostructure distributed feedback (DFB) laser, in orthographic and cross-section views, with segmented p-contacts. Figure 7a illustrates, in schematic form, testing a DFB laser on a substrate using a probe card, by probing individual contact segments.
[0063] Figure 7b illustrates, in schematic form, testing a DFB laser on a substrate using a probe card, by temporarily shorting together contact segments.
[0064] Figure 7c illustrates, in schematic form, two DFB lasers on a substrate after a lithographic fabrication step has been performed to apply different fabrication modifications customised to the respective lasers, in accordance with an embodiment of the present invention.
[0065] Figure 7d illustrates, in schematic form, the DFB lasers of Figures 7c on a wafer substrate.
[0066] Figure 8a illustrates, in schematic form, on-wafer testing of a buried heterostructure DFB laser on a substrate using a probe card, by probing a lithographically patterned temporary contact.
[0067] Figure 8b illustrates, in schematic form, the buried heterostructure DFB laser of Figure 8a, after testing and after removal of the temporary contact.
[0068] Figure 8c illustrates, in schematic form, two buried heterostructure DFB lasers on a substrate after a lithographic fabrication step has been performed to apply different fabrication modifications customised to the respective lasers, in accordance with an embodiment of the present invention.
[0069] Figure 8d illustrates, in schematic form, the buried heterostructure DFB lasers of Figures 8c on a wafer substrate.
[0070] Figure 8e illustrates, in schematic form, on-wafer testing of a buried heterostructure DFB laser on a substrate using a probe card, by probing a lithographically patterned offset contact. Figure 8f illustrates, in schematic form, two buried heterostructure DFB lasers on a substrate after a lithographic fabrication step has been performed to apply different fabrication modifications customised to the respective lasers, in accordance with an embodiment of the present invention.
[0071] Figure 8g illustrates, in schematic form, the buried heterostructure DFB lasers of Figures 8f on a wafer substrate.
[0072] Figure 9a illustrates, in schematic form, testing a DFB laser, using a probe card needle.
[0073] Figure 9b illustrates, in schematic form, the DFB laser of Figure 9a after a material deposition fabrication step has been performed to apply the fabrication modification to the laser.
[0074] Figure 9c illustrates, in schematic form, testing a DFB laser having dielectric on the cavity, using a probe card needle.
[0075] Figure 9d illustrates, in schematic form, the DFB laser of Figure 9c after a material etching fabrication step has been performed to apply the fabrication modification to the laser.
[0076] Figure 10a illustrates, in schematic form, testing a DFB laser using a probe card needle, with contact segments wired together.
[0077] Figure 10b illustrates, in schematic form, the DFB laser of Figure 8a after a waferlevel etching fabrication step has been performed to apply a fabrication modification to the laser, in accordance with an embodiment of the present invention.
[0078] Figure 11 is a graph of output power versus current density showing the effect of fabrication modifications provided by embodiments of the present invention. Figure 12 shows a simulation of optical power spectra, showing a single mode wavelength shift before and after fabrication modification by hardwiring of a heater element to provide heating power nonuniformly along the cavity.
[0079] Figure 13 shows a simulation of optical power spectra, showing multimode wavelength shift before and after fabrication modification by hardwiring of segments of the segmented electrical contact to provide a nonuniform carrier injection distribution along the cavity.
[0080] Detailed description
[0081] In this description and claims, the terms optical and optical radiation relate to electromagnetic radiation over a range of wavelengths not limited to visible radiation, such as wavelengths spanning ultraviolet, visible and infrared radiation. An InP distributed feedback (DFB) laser is described as an example of a DFB laser apparatus. Other compound-semiconductor based devices may be used with embodiments. For example photonic devices based on GaAs, GaSb, or GaN, or photonic devices based on other material systems, may be used.
[0082] The examples described herein relate to DFB lasers fabricated with an etched facet and modified post-test on a wafer.
[0083] Figure 1 illustrates a known DFB laser, in this example an etched-facet laser 100, with a compound semiconductor laser on an InP (indium phosphide) substrate.
[0084] Figure 1 a is an orthographic view of the laser 100. Figure 1 b is a cross-section (not to scale) along a-a shown in Figure 1 a. Figure 1 b is thus a cross-section through the waveguide 104 along its propagation direction (length). Figure 1 c is a cross-section (not to scale) along b-b shown in Figure 1 a. Figure 1c is thus a cross-section across the waveguide 104 perpendicular to its propagation direction.
[0085] The structure of the laser 100 is now described in the context of its wafer-scale fabrication. A ridge waveguide 104 is defined by a waveguide etch. A pattern of openings in a hard mask in a lithographic step defines trenches 102, 106 that are etched to define the ridge waveguide 104 in between them. An insulating dielectric material 1 18 covers most of the top surface, and a contact window is opened up in the dielectric along the top of the ridge 104. Subsequently, metal 116 is deposited covering the ridge waveguide and is annealed, making contact through the contact window to the top of the ridge waveguide 104. After annealing of the metal, an alloyed ohmic contact 145 is formed between the metal and semiconductor.
[0086] A pad of the metal 116 at one side of the ridge waveguide is used as an area for soldering or bonding to the metal. In subsequent fabrication steps, a patterned hard mask and facet etch defines front and rear etched facets 110, 108 at either end of the ridge waveguide 104. The facet etch creates an etched surface, which extends either side of the etched facet. The facet etch is deeper than the ridge etch.
[0087] A small horizontal spacing is provided between the ridge trenches 102, 106 and the facet etch features, so that the front and rear etched facets 110, 108 are etched as flat planes rather than having corners with the ridge waveguide, which would etch unevenly and would be detrimental to the smoothness of the facet at the end of the waveguide. This results in a flange structure shaped like a T, with the waveguide being the trunk of the T, and walls 1 12 being the crossbar of the T. The effect of the spacing and resulting T-shaped structure is to ensure that the facet is smooth to provide efficient and reproducible transmission through optical coupling regions, or internal reflection at, the facets.
[0088] After the facet etch, an anti-reflective (AR) coating 138 is applied to one etched facet 1 10 and a high-reflectance (HR) coating 140 (or an AR coating, not shown) is applied to the other etched facet 108 at the other end of the waveguide 104.
[0089] Finally, a metallisation step coats the underside of the wafer with metal 142.
[0090] With reference to Figures 1 a to 1 c, in operation the laser cavity, comprising the ridge waveguide 104 and underlying layers 124-134 bounded by facets 108 and 110 at either end, outputs optical radiation 142 through an optical coupling region 114. With reference to Figures 1 b and 1c, the layer structure will now be described in detail. From the top in Figure 1 b, a p-metal layer 116 extends down through a window in the dielectric layer 1 18. The p-metal layer 116 makes contact to a p-type InGaAs contact layer 120, which is the top epitaxially-grown layer. Below that, a p- type InP cladding layer 122 is followed by a p-type etch stop / grating layer 124. The etch that stops on that layer 124 is the waveguide ridge etch, as illustrated in Figure 1c. Next, a p-type InP spacer layer 126 is followed by a p-type separate confinement heterostructure (SCH) layer 128, an undoped multi-quantum well (MQW) layer 130, and an n-type SCH layer 132. The MQW layers are the optically active layers in the laser.
[0091] The n-type InP buffer layer 134 is the first of the epitaxial layers that is grown on the n-type InP substrate 136.
[0092] In a distributed feedback (DFB) laser, a grating is superimposed on the waveguide to provide optical feedback in the laser cavity. In this example, the grating is made by performing the epilayer growth in two stages and in between the stages patterning the grating. First, a lower epitaxial layer structure 144 is grown on the substrate, starting with the n-type InP buffer layer 134 then the SCH and MQW active layers 125 then the p-type InP spacer layer 126 and the etch stop / grating layer structure 124. Electron beam lithography is used to define a grating pattern 150 (shown in Figure 1 b), which is transferred by etching into the etch stop / grating layer structure 124.
[0093] After the grating patterning, the upper epitaxial layer structure 146 is overgrown on the lower epitaxial layer structure 144, using for example metalorganic vapour-phase epitaxy (MOCVD). The upper epitaxial layer structure 146 includes the p-type InP cladding layer 122 under the p-type InGaAs contact layer 120.
[0094] In this example, the front and rear etched facets 110, 108 are coated with a PECVD- deposited silicon nitride AR coating 138 and an HR coating 140 respectively. The AR coating is selectively removed after deposition to allow bonding to metallic layers. Finally, the n-metal layer 142 is shown. In operation, as shown at the left of Figure 1 b, a beam of optical radiation 142, illustrated bounded with dashed lines, is output from the etched facet 1 10 at the optical coupling region 114. In this example, the optical radiation is output from the optically active layers of the ridge waveguide 128, 130, 132 (collectively labelled 125 in Figure 1c) into the air to the left of the front etched facet 110. For an SOA example (not shown), instead of an HR coating 140 another AR coating is applied to the rear etched facet 108 and radiation is input to the waveguide at the rear etched facet 108.
[0095] With reference to Figure 1 c, the p-metal layer 116 can be seen on top of the dielectric layer 118 as it covers trenches 102, 106 either side of the ridge waveguide 104. The p-metal layer 1 16 contacts the top of the ridge 104 through a window in the dielectric 118. The trenches 102, 106 are etched by the waveguide etch, which selectively stops on the p-type etch stop / grating layer 124.
[0096] The location of the optical coupling region 114 is shown projected along the waveguide from the etched facet 110 onto this cross-section plane b-b. It is centred horizontally with respect to the ridge waveguide 104 and centred vertically with respect to the undoped MQW layer 130.
[0097] With reference to Figure 1 c, a contact is made by the metal layer 116 at the top of ridge 104 to the p-type InGaAs contact layer 120 via the alloyed contacting layer 145. Current spreads outwards from under the ridge 104 towards the active layer 125 because it is not laterally confined when it leaves the ridge downwards towards the active layers. Therefore the whole of the optical mode is supplied with current. The optical mode roughly corresponds to the projected coupling region 114.
[0098] In the drawings of Figure 1 and in subsequent drawings, features labelled with the same numerals correspond to the same features in subsequent drawings. Therefore, a description of a feature in any drawing should also apply to a feature labelled with the same numeral elsewhere in this description.
[0099] Facet phase is the phase of the DFB grating with respect to the end facets. Facet phase variation arises from fabrication processes of the laser. As mentioned above, facet phase variation leads to yield loss in compound-semiconductor DFB laser manufacturing. In more detail, facet phase variation causes variation in reflectivity and transmittivity versus wavelength of the cavity. This variation in reflectivity and transmittivity versus wavelength of the cavity causes increased variation of the photon distribution along the cavity.
[0100] Figure 2a is a graph of photon density nPh versus distance along the cavity z for two different facet phase conditions 201 , 202. The plotted curves represent the photon distributions along the cavity. In this example, it is apparent that there is a higher photon density nPh for the second phase condition 202 at the right-hand side of the graphs (z = 800 to 1000 pm), which is near the front facet of the laser cavity.
[0101] Increased variation of the photon distribution causes increased variation of the carrier distribution ncalong the cavity to form, at antinodes in the standing wave pattern of the longitudinal optical mode, regions having reduced carrier density due to stimulated recombination. This is called spatial hole burning. Figure 2b is a graph of carrier density ncversus length along the cavity z for the same two facet phase conditions of Figure 2a. In this example, the spatial hole burning can be seen where there is lower carrier density ncfor the second phase condition 202 (dashed line) at the right-hand side of the graphs (z = 800 to 1000 urn), near the front facet of the laser cavity.
[0102] The spatial hole burning causes gain saturation and refractive index variation. This gain and refractive index variation has an effect on the optical properties of the laser, with effects on parameters such as lasing wavelength, side-mode suppression ratio (SMSR) and modulation performance.
[0103] Figure 3 is a graph of SMSR (in dB) versus J / Jth, the ratio of the current density J to the threshold current density Jth for the two different facet phase conditions 301 , 302 discussed with reference to Figures 2a and 2b. The J / Jth ratio indicates how far above the threshold the laser is operating. For the second phase condition 302, the SMSR collapses at higher output power, above a J / Jth of about 6. This indicates a transition to multi-mode operation for the second phase condition, which is a failure mode for a single mode laser. Meanwhile for the first phase condition 301 , the SMSR is stable with increasing current density.
[0104] In general, the gain and refractive index variation resulting from the facet phase variation causes unwanted variation in the threshold gain as a function of wavelength. Threshold gain is gain required for lasing, with the gain increasing with J and clamped at the threshold. Unacceptable threshold gain spectrum variation may or may not be accompanied by spatial hole burning. Unwanted multimode behaviour may be occur just above the lasing threshold, but before spatial hole burning occurs.
[0105] Figures 4a and 4b are graphs of threshold gain gth versus wavelength for first and second facet phase conditions.
[0106] With reference to Figure 4a, the threshold gain spectrum 401 of the first facet phase condition has one mode 411 with lowest threshold gain (lasing).
[0107] With reference to Figure 4b, the threshold gain spectrum 402 of the second facet phase condition has two modes 412 with equal or very similar threshold gain. This results in a device with the first facet phase condition (as shown in Figure 4a) having SMSR ~40dB while a device with the second facet phase condition (as shown in Figure 4b) has SMSR of ~20 d B , which would fail a test based on a single mode spec of +35dB. This is an example of unacceptable threshold gain spectrum variation.
[0108] The threshold gain spectrum is therefore sensitive to facet phase. Variation of threshold gain causes large variation in parametric performance of the laser. In embodiments, the harm caused by facet phase variation can be mitigated after parametric testing. The DFB grating structure cannot itself be directly adjusted after laser fabrication, but the laser cavity can be adjusted by post-test fabrication modifications that are nonuniform along and / or across the cavity. Nonuniform distribution of carrier injection, effective refractive index and material can be used to compensate for the threshold gain spectrum variation, because threshold gain is affected by nonuniform cavity adjustments such as these. Figure 5 is a flowchart of a method of manufacturing a DFB laser apparatus, in accordance with an embodiment of the present invention. The method has the steps:
[0109] 510: Specifications for single mode operation are defined, including acceptable limits for parametric test results.
[0110] 512: Fabricating a plurality of distributed-feedback (DFB) lasers on a substrate, each laser comprising a cavity. The lasers are in-plane lasers, in which the radiation propagates along an elongate cavity in a direction parallel to the planar surface of the wafer from which the laser is made. This includes edge-emitting lasers and inplane lasers with vertical emission provided by mirrors or grating couplers, for example.
[0111] The lasers may each comprise a segmented electrical contact (as described with reference to Figures 6a to 8b) arranged to inject carriers to the respective cavity.
[0112] This step may include fabricating a lithographically patterned structure for each respective laser, for on-wafer testing of the lasers by supplying injection current uniformly along the respective laser cavity via the respective lithographically patterned structure.
[0113] The lithographically patterned structure may be a temporary short configured to supply injection current uniformly along the respective laser cavity during on-wafer testing by shorting together contact segments of the respective laser, and the method may then further comprise the step of removing the temporary short after the on-wafer testing.
[0114] The lithographically patterned structure may be a temporary contact configured to supply injection current uniformly along the respective laser cavity during on-wafer testing by contacting a semiconductor layer of the respective laser, and the method may then further comprise the step of removing the temporary contact after the on- wafer testing. When the lasers are buried heterostructure lasers, the lithographically patterned structure may be an offset contact that is offset laterally from the respective laser cavity and is configured to supply injection current uniformly along the respective laser cavity during on-wafer testing by contacting a semiconductor layer of the respective laser.
[0115] The offset contact may be not annealed before the step of on-wafer testing.
[0116] 514, 516: Testing the plurality of lasers to generate parametric test results. This is performed with multiple lasers on a substrate, as an on-wafer test. Electrical power in the form of injection current is supplied to the lasers and measured, and typically output optical radiation is guided to optical test equipment to be measured. The single mode performance is tested. Parametric optical tests include LIV (light- current-voltage) sweeps at multiple temperatures and optical spectra at multiple injection currents and temperatures. These tests can be performed under CW (continuous-wave) and pulsed laser operation. SMSR, lasing frequency and wavelength results can be obtained, and mode hopping can be inferred from the measured data, such as kinks in an LIV sweep. The graphs described with reference to Figures 3, 12 and 13 are examples of parametric test results. Parametric tests can be optical and / or electrical.
[0117] In examples where the lasers each comprise a segmented electrical contact arranged to inject carriers to the cavity, the testing is performed with the carriers injected uniformly along the cavity of each laser. This can be achieved by some or all segments of the segmented electrical contact being wired together to inject the carriers uniformly along each respective laser’s cavity (described, for example, with reference to Figure 7b).
[0118] In examples where the lasers each comprise a heater element arranged to provide heating power to the cavity, the heater element may be wired to provide heating power uniformly along each respective laser’s cavity (described, for example, with reference to Figure 10a). 518, 520: Determining, using respective parametric test results, fabrication modifications to the lasers customised to respective lasers, to adjust each respective laser cavity nonuniformly to compensate for parametric variation and / or parametric test fails in the parametric test results. When the selected parametric tests are sensitive to threshold gain spectrum variation, the fabrication modifications are thus determined to compensate for threshold gain spectrum variation. When the selected parametric tests are sensitive to spatial hole burning, the fabrication modifications are thus determined to compensate for spatial hole burning. The fabrication modifications can be determined and applied to all lasers, to reduce overall variation. Alternatively, the fabrication modifications can be determined and applied to a subset of the lasers, which have unacceptable variation in their parametric test results, including being out of specification, such as parametric test fails.
[0119] The fabrication modifications in the examples described with reference to the Figures are determined to adjust the respective laser cavities nonuniformly axially along their cavities. Additionally or alternatively, the fabrication modifications may be determined to adjust the respective laser cavities nonuniformly transversely across their cavities, for example by segmenting contacts orthogonally to the segments described with reference to Figure 6, such that there is a gap along the top of the ridge between contacts on either side.
[0120] The modifications may be determined using a model, for example an analytical model, a simulation or a machine learning model. Alternatively, the modifications may be determined using a look-up table. The parametric test results are input to the model or algorithm, or look-up table, and a modification is determined or selected to adjust the laser cavity nonuniformly to compensate for the parametric variation and / or parametric test fails and / or threshold gain spectrum variation and / or spatial hole burning.
[0121] In the example where the lasers each comprise a segmented electrical contact, hardwiring of segments provides injection of carriers nonuniformly to the cavities, to provide nonuniform carrier injection distributions along the cavities, customised for each respective laser (described, for example, with reference to Figure 7c). This customised nonuniform carrier injection distribution leads to a customised nonuniform carrier density distribution along the cavity, to compensate for parametric variation and / or parametric test fails and / or threshold gain spectrum variation and / or spatial hole burning. The nonuniform carrier density distribution also affects the nonuniform effective refractive index distribution, to further compensate for parametric variation and / or parametric test fails and / or threshold gain spectrum variation and / or spatial hole burning, as increased carrier density causes increased effective refractive index.
[0122] In examples where the lasers each comprise a heater element, hardwiring of the heater element provides heating power nonuniformly along the cavity, customised for each respective laser (described, for example, with reference to Figure 10b). This customised nonuniform heating power along the cavity leads to a customised nonuniform effective refractive index distribution along the cavity, to compensate for the parametric variation and / or parametric test fails and / or threshold gain spectrum variation and / or spatial hole burning.
[0123] Alternatively or additionally, the fabrication modification may use addition of material to the laser to provide material nonuniformly to the cavity, customised for the laser (described, for example, with reference to Figures 9a and 9b). This may involve addition of dielectric material, to provide a nonuniform effective refractive index distribution to the cavity. The addition may involve implantation of material, for example through the p-metal, to provide a nonuniform carrier injection distribution to the cavity. The addition of material can provide a nonuniform stress distribution to the cavity.
[0124] Alternatively or additionally, the fabrication modification may use removal of material from the laser to provide material nonuniformly to the cavity, customised for the laser (described, for example, with reference to Figures 9c and 9d). This may involve removal of dielectric material, to provide a nonuniform effective refractive index distribution to the cavity. The removal of material of material may provide a nonuniform stress distribution to the cavity.
[0125] Alternatively, a customised nonuniform effective refractive index distribution can be achieved by making fabrication modifications involving addition and / or removal of material to the lasers nonuniformly along the cavity (described, for example, with reference to Figures 9a to 9d). Such fabrication modifications lead to a customised nonuniform refractive index distribution along the cavity, to compensate for the parametric variation and / or parametric test fails and / or threshold gain spectrum variation and / or spatial hole burning.
[0126] 522: Performing one or more fabrication step to apply the fabrication modifications to the respective lasers.
[0127] This step is efficiently performed with multiple lasers on a substrate, as a wafer-level fabrication step, and the subsequent re-testing step 524 may then also be done as a wafer-level testing step.
[0128] The wafer-level fabrication steps may comprise, for example, lithographic patterning, deposition, implantation and etching. Examples are described with reference to Figures 6a to 10b. The skilled person will appreciate that these fabrication steps may be used together in other combinations.
[0129] 524: Re-testing lasers having the fabrication modifications to generate revised parametric test results. Because the fabrication modifications that provide nonuniform carrier injection distributions and / or nonuniform effective refractive index distributions were determined in this example to compensate for the parametric variation and / or parametric test fails and in particular threshold gain spectrum variation, the yield loss due to threshold gain spectrum variation is at least partially mitigated. Upon re-testing, the single mode performance is better aligned to the parametric test specifications, so the parametric test distribution improves and yield improves as there are fewer parametric test fails.
[0130] 526: Assembly of the modified lasers. The lasers can be singulated or diced and known good die are then packaged and / or assembled into the finished products.
[0131] 528: Alternatively, the fabrication step may be performed by modifying circuitry on a sub-mount and completing the fabrication by mounting the respective laser on the sub-mount in an assembly step. Thus, the circuitry on the sub-mount has the fabrication modifications, rather than the laser itself, and the unmodified lasers are assembled on the modified sub-mount. In this case the retesting 530 is performed on the laser after assembly.
[0132] Figures 6a to 6c illustrate an etched-facet distributed feedback (DFB) laser, in orthographic and cross-section views, with segmented p-contacts.
[0133] Figure 6a is an orthographic view of the laser 600. The laser 600 in Figure 6a is shown as a section of a wafer substrate. The features are the same as described with respect to Figures 1 a to 1c, except the p-type contact 116 of Figures 1 a to 1 c is replaced by a segmented contact 116a-116f. In the examples described herein, there are six segments in the segmented contact. However, the skilled person will appreciate that there may be fewer or more segments. With more segments, there will be more gaps between the segments, so the effective contact resistance will be increased and limited by the gap width.
[0134] Figure 6b is a cross-section (not to scale) along a-a shown in Figure 6a. Figure 6b is thus a cross-section through the waveguide 104 and cavity along its length.
[0135] Figure 6c is a cross-section (not to scale) along d-d shown in Figures 6a a and 6b. A cross section made along c-c shown in Figure 6a would be similar to cross-section a- a shown in Figure 1 c, except the metal 116 in Figure 1c is in this case contact segment 1 16c.
[0136] An insulating dielectric material 118 covers most of the top surface, and contact windows are opened up in the dielectric along the top of the ridge 104.
[0137] Subsequently, metal 116a-116f is lithographically patterned and deposited covering the ridge waveguide and making contact through the contact windows to the p-type InGaAs contact layer 120 at the top of the ridge waveguide 104. After annealing of the metal, alloyed ohmic contacts 145 (depicted with a cross-hatch pattern) are formed between the metal 116a-116f and semiconductor 120. As shown in Figure 6b, after the facet etch, an anti-reflective (AR) coating 138 is applied to one etched facet 1 10 and a high-reflectance (HR) coating 140 is applied to the other etched facet 108 at the other end of the waveguide 104.
[0138] Figures 6d to 6f illustrate an etched-facet distributed feedback (DFB) laser with a buried heterostructure waveguide, in orthographic and cross-section views, with segmented p-contacts.
[0139] Figure 6d is an orthographic view of the buried heterostructure laser 650. The laser 650 in Figure 6d is shown as a section of a wafer substrate. The features are the same as described with respect to Figures 6a to 6c, except there is a buried heterostructure waveguide 604 and there is thus no need for insulating dielectric material 1 18. In the examples described herein, there are six segments in the segmented contact. However, the skilled person will appreciate that there may be fewer or more segments and the segments may be of different shapes and sizes and may have gaps between them or may be contiguous. With an increasing number of separated segments, there may be more gaps between the segments, so the effective contact resistance will be increased and limited by the gap width.
[0140] Figure 6e is a cross-section (not to scale) along a-a shown in Figure 6d. Figure 6e is thus a cross-section through the buried heterostructure waveguide 604 and cavity along its length.
[0141] Figure 6f is a cross-section (not to scale) along c-c shown in Figures 6d and 6e. Either side of the buried heterostructure waveguide, grown on the n-type InP substrate 136, there is a p-type InP blocking layer 626 and an n-type InP blocking layer 624, over which the InP upper cladding layer 122 and the p-type InGaAs contact layer 120 are grown. The location of the optical coupling region 114 is shown projected along the waveguide from the etched facet 110 onto this cross-section plane c-c. The buried heterostructure waveguide 604 is generally indicated by the projection of the optical coupling region.
[0142] Metal 116a-116f is lithographically patterned and deposited covering the buried heterostructure waveguide 604 and making contact to the p-type InGaAs contact layer 120 above the buried heterostructure waveguide 604. After annealing of the metal, alloyed ohmic contacts 145 (depicted with a cross-hatch pattern) are formed between the metal 116a-1 16f and semiconductor 120.
[0143] As shown in Figure 6e, after the facet etch, an anti-reflective (AR) coating 138 is applied to one etched facet 1 10 and a high-reflectance (HR) coating 140 is applied to the other etched facet 108 at the other end of the buried heterostructure waveguide 604.
[0144] In another example of a buried heterostructure laser (not shown), like for the ridge laser described with reference to Figures 6a to 6c, an insulating dielectric material 118 covers most of the top surface, and contact windows are opened up in the dielectric above the buried heterostructure waveguide 604. Subsequently, metal 116a-116f is lithographically patterned and deposited over the buried heterostructure waveguide 604 and making contact through the contact windows to the p-type InGaAs contact layer 120 above the buried heterostructure waveguide 604. In cross section along the buried heterostructure waveguide, this example would be the same as illustrated by Figure 6b.
[0145] Ridge waveguide lasers and buried heterostructure lasers are both examples of inplane lasers, in which the radiation propagates along an elongate cavity in a direction parallel to the planar surface of the wafer from which the laser is made.
[0146] Figure 7a illustrates on-wafer testing a DFB laser 750 on a substrate using a probe card, by probing individual contact segments 116a-1 16f . In this diagram, the contact windows in the dielectric layer (1 18 in previous figures) and the trenches (102, 106 in previous figures) are not shown.
[0147] Probe needles 714a-714f mounted on a probe card are used to supply injection current to the cavity that includes ridge waveguide 104 or alternatively buried heterostructure waveguide (not shown) and the underlying layers 124-134, bounded by the facets at each end. During the on-wafer testing, the injection current is supplied through the probe needles 714a-714f uniformly along the cavity, with each contact segment 116a-116f supplying the same injection current to its respective underlying sections of the cavity.
[0148] As an alternative to probing individual contact segments, Figure 7b illustrates testing a DFB laser 752 on a substrate using a probe card with one probe needle 716, by shorting together contact segments 116a-1 16f with a lithographically patterned temporary short 718. Again, the injection current is supplied through the probe needle 716 and via temporary short 718 uniformly along the cavity, with each contact segment 1 16a-116f supplying substantially the same magnitude of injection current to its respective underlying sections of the cavity. In this example, all segments are shorted together at test. Alternatively, some other pattern of shorts could allow testing to be done by a multi-probe on-wafer tester, to provide uniform current injection.
[0149] In operation during the test, the laser cavity outputs optical radiation 142, which is then measured and the generated parametric test results are recorded. Fabrication modifications are then determined as described with reference to Figure 5.
[0150] Figure 7c illustrates two DFB lasers 750, 752 on a section 757’ of wafer substrate (757 in Figure 7d) after a wafer-level lithographic fabrication step has been performed to apply different fabrication modifications customised to the respective lasers, which have unwanted variation in their parametric test results. Different respective fabrication modifications are applied to respective lasers by one or more wafer-level fabrication step, to compensate for the parametric variation and / or parametric test fails and / or threshold gain spectrum variation and / or spatial hole burning. The one or more fabrication step (e.g. patterning, deposition, etching) is performed concurrently for a plurality of the lasers when they are on the same wafer. In this example, where the lasers each comprise a segmented electrical p-contact, the different respective fabrication modifications comprise hardwiring of segments to provide customised nonuniform carrier injection distributions axially along the respective laser cavities.
[0151] For one laser 750, the fabrication modification is a lithographically patterned customised short or busbar 760, that connects together all contact segments 116a, 116c, 116e and 1 16f, except two 116b and 116d . For the other laser 752, the fabrication modification is a lithographically patterned customised short or busbar 762, that connects together all contact segments 116a-1 16d and 116f, except one 116e. The different disconnected contact segments, 116b / 116d and 116e respectively for the two lasers 750,752 cause different nonuniform current injection, thus the customised nonuniform carrier injection distribution along the cavity leads to a customised nonuniform carrier density distribution that is different for each of the two lasers 750, 752 to compensate for their different threshold gain spectrum variation. Upon re-testing, the modifications mean that they will be performing in specification for the single-mode tests. Thus, yield for the wafer is increased.
[0152] Figure 7d illustrates the DFB lasers 750, 752 of Figure 7c on a wafer substrate 757 after the lithographic fabrication step has been performed to apply different fabrication modifications customised to the respective lasers. Each laser is depicted as a rectangle and the electrically connected segments are shown schematically as horizontal lines.
[0153] Figure 7e illustrates two DFB lasers 780, 782 on a section 787’ of a wafer substrate after a bonding step has been performed to apply different fabrication modifications customised to the respective lasers, which have unwanted variation in their parametric test results. In this example, where the lasers again each comprise a segmented contact, the fabrication modifications provide customised nonuniform carrier injection distributions axially along the respective laser cavities.
[0154] For one laser 780, the file tuning modification is a bonding configuration of bond wires 770a and 770c-770f, that connects together all contact segments 116a and 116c-116f, except one 116b. The bond wires are all connected to the same carrier source. No carriers are injected to the cavity under the disconnected contact 116b. For the other laser 782, the fabrication modification is a bonding configuration of bond wires 772a-772d and 772f, that connects together all contact segments 116a- 116d and 116f, except a different one 116e. The different respective disconnected contact segments 1 16b and 116e cause different nonuniform current injection, thus the customised nonuniform carrier injection distribution along the cavity leads to a customised nonuniform carrier density distribution that is different for each of the two lasers 780, 782 to compensate differently for their different threshold gain spectrum variation and / or spatial hole burning. Upon re-testing, the modifications mean that they will be performing in specification for the single-mode parametric tests. Thus, yield for the wafer is increased.
[0155] Figure 8a illustrates on-wafer testing of a buried heterostructure DFB laser 852 on a substrate using a probe card with one probe needle 816, probing a lithographically patterned temporary contact 818. Preferably, the temporary contact is non-annealed for easy removal by a wet etch. Ti / Au is a suitable non-annealed contact metallisation. A non-annealed contact may not have optimum contact resistance, but it may still be used to supply injection current for testing. In this diagram, any contact windows in the dielectric layer, if present, are not shown.
[0156] During the on-wafer testing, the injection current is supplied through the probe needle 816 and via temporary contact 818 uniformly along the cavity corresponding to the buried heterostructure waveguide 604 with its coatings 138, 140 at each end. The temporary contact contacts the p-type InGaAs semiconductor contact layer 120 of the laser. In this example, one temporary contact extends substantially along the whole length of the buried heterostructure waveguide 604. Alternatively, some other pattern of temporary contacts could allow testing to be done by a multi-probe on- wafer tester, with injection current supplied uniformly along the cavity.
[0157] In operation during the test, the laser cavity outputs optical radiation 142, which is then measured and the generated parametric test results are recorded. Fabrication modifications are then determined as described with reference to Figure 5.
[0158] Figure 8b illustrates the buried heterostructure DFB laser 852 of Figure 8a, after testing and after removal, in this example by a wet etch, of the temporary contact 818. The wet etch is applied to the devices as a wafer-level process.
[0159] Figure 8c illustrates two buried heterostructures DFB lasers 850, 852 on a section 857’ of wafer substrate (857 in Figure 8d) after a wafer-level lithographic fabrication step has been performed to apply different fabrication modifications customised to the respective lasers, which have unwanted variation in their parametric test results. Different respective fabrication modifications are applied to respective lasers by one or more wafer-level lithographic patterning fabrication step, to compensate for the parametric variation and / or parametric test fails and / or threshold gain spectrum variation and / or spatial hole burning. The one or more fabrication step (e.g. patterning, deposition, etching) is performed concurrently for a plurality of the lasers when they are on the same wafer. In this example, where the lasers each comprise a differently shaped castellated electrical p-contact, the different respective fabrication modifications comprise patterning and deposition of the shaped contact reaching over the laser cavity to provide customised nonuniform carrier injection distributions axially along the respective laser cavities.
[0160] For one laser 850, the fabrication modification is a lithographically patterned customised shape 860, that reaches over the laser cavity at all but two contact segments 816a, 816c, 816e and 816f. For the other laser 852, the fabrication modification is a lithographically patterned customised shape 862, that reaches over the laser cavity at all but one contact segments 816a-816d and 816f. The different uncontacted segments for the two lasers cause different nonuniform current injection, thus the customised nonuniform carrier injection distribution along the cavity leads to a customised nonuniform carrier density distribution that is different for each of the two lasers 850, 852 to compensate for their different threshold gain spectrum variation. Upon re-testing, the modifications mean that they will be performing in specification for the single-mode tests. Thus, yield for the wafer is increased.
[0161] Figure 8d illustrates the DFB lasers 850, 852 of Figure 8c on a wafer substrate 857 after the wafer-level lithographic fabrication step has been performed to apply different fabrication modifications customised to the respective lasers. Each laser is depicted as a rectangle and the electrically connected segments are shown schematically as horizontal lines.
[0162] Figure 8e illustrates on-wafer testing a buried heterostructure DFB laser 852 on a substrate using a probe card with one probe needle 816, probing a lithographically patterned offset contact 820. The offset contact 820 can be left on the wafer after testing and left on the laser chip after dicing / cleaving of the wafer. Being offset laterally at the side it will not interfere with the subsequent fabrication modifications or the laser operation when manufacturing is complete. The offset contact 820 may be annealed to reduce contact resistance. Alternatively, in another example (not shown) the offset contact may be non-annealed. In that case it may be then easily removed by a wet etch after testing. A non-annealed contact may not have optimum contact resistance, but it may still be used to supply injection current for testing.
[0163] During the on-wafer testing, the injection current is supplied through the probe needle 816 and via offset contact 820 uniformly along the cavity corresponding to the buried heterostructure waveguide 604 with its coatings at each end. The offset contact 820 contacts the p-type InGaAs semiconductor contact layer 120 of the laser, without extending above the buried heterostructure waveguide. In this example, one offset contact extends substantially along (but offset from) the whole length of the buried heterostructure waveguide 604. Alternatively, some other pattern of offset contacts could allow testing to be done by a multi-probe on-wafer tester, with injection current supplied uniformly along the cavity.
[0164] In operation during the test, the laser cavity outputs optical radiation 142, which is then measured and the generated parametric test results are recorded. Fabrication modifications are then determined as described with reference to Figure 5.
[0165] Figure 8f illustrates two buried heterostructures DFB lasers 870, 872 on a section 877’ of wafer substrate (877 in Figure 8g) after a wafer-level lithographic fabrication step has been performed to apply different fabrication modifications customised to the respective lasers, which have unwanted variation in their parametric test results. Different respective fabrication modifications are applied to respective lasers by one or more wafer-level lithographic patterning fabrication step, to compensate for the parametric variation and / or parametric test fails and / or threshold gain spectrum variation and / or spatial hole burning. The one or more fabrication step (e.g. patterning, deposition, etching) is performed concurrently for a plurality of the lasers when they are on the same wafer. In this example, where the lasers each comprise a differently shaped digitated electrical p-contact, the different respective fabrication modifications comprise patterning and deposition of the shaped contact reaching over the laser cavity to provide customised nonuniform carrier injection distributions axially along the respective laser cavities.
[0166] For one laser 870, the fabrication modification is a lithographically patterned customised shape 880, that reaches over the laser cavity at all but two contact segments 816a, 816c, 816e and 816f. For the other laser 872, the fabrication modification is a lithographically patterned customised shape 882, that reaches over the laser cavity at all but one contact segments 816a-816d and 816f. The different uncontacted segments cause different nonuniform current injection, thus the customised nonuniform carrier injection distribution along the cavity leads to a customised nonuniform carrier density distribution that is different for each of the two lasers 870, 872 to compensate for their different threshold gain spectrum variation. Upon re-testing, the modifications mean that they will be performing in specification for the single-mode tests. Thus, yield for the wafer is increased.
[0167] Figure 8d illustrates the DFB lasers 870, 872 of Figure 8f on a wafer substrate 877 after the wafer-level lithographic fabrication step has been performed to apply different fabrication modifications customised to the respective lasers. Each laser is depicted as a rectangle and the electrically connected segments are shown schematically as horizontal lines.
[0168] Figure 9a illustrates testing a DFB laser 952 using a probe card needle. The injection current is supplied through the probe needle 916.
[0169] Any mechanical stress in the materials of the cavity resulting from fabrication steps is distributed uniformly along the cavity. The laser is tested to generate parametric test results and, for lasers that have fails, one or more fabrication steps are performed to apply a fabrication modification determined as described with reference to Figure 5. The fabrication modification is customised to this specific laser, using the parametric test results, to compensate for parametric variation and / or parametric test fails and / or threshold gain spectrum variation and / or spatial hole burning.
[0170] Different respective fabrication modifications are applied to respective lasers by one or more fabrication step, to compensate for parametric variation and / or parametric test fails and / or threshold gain spectrum variation and / or spatial hole burning. In this example, the fabrication modifications comprise addition of material to the laser nonuniformly along the respective cavity that provides a nonuniform effective refractive index distribution along the cavity, customised for each respective laser.
[0171] Figure 9b illustrates the DFB laser of Figure 9a after a material deposition fabrication step has been performed to apply the fabrication modification to the laser. Dielectric 902 is deposited and lithographically patterned. It is placed on the top of the ridge and / or on the sidewalls. The material of the dielectric, and / or stress it causes in the underlying materials of the cavity, is therefore distributed nonuniformly along the cavity. The nonuniformity causes nonuniform refractive index along the cavity, and this leads to a customised nonuniform effective refractive index distribution along the cavity that compensates for threshold gain spectrum variation. The uniform injection current is supplied through the bond 970. Upon re-testing, the modification means that the laser will be performing in specification for the single-mode tests. Thus, yield for the wafer is increased.
[0172] In an alternative example, the fabrication modification comprises implantation. In this example, the area 902 represents an ion implant through the metal 1 16. The implantation species and / or damage causes change in the electrical characteristics of the cavity in the area 902, for example, a reduction in carrier injection. The implantation therefore provides a nonuniform carrier injection distribution, with a similar effect to the disconnection of a segment of the contact, as described with reference to Figure 7c.
[0173] Figure 9c illustrates testing a DFB laser 954 using a probe card needle. The injection current is supplied through the probe needle 916.
[0174] Dielectric 904 is deposited and lithographically patterned uniformly along the cavity. It is placed on the top of the ridge and / or on the sidewalls. The material of the dielectric 904, and / or stress it causes in the underlying materials of the cavity, is therefore distributed uniformly along the cavity. The laser is optically tested to generate parametric test results and, for lasers having fails, one or more fabrication steps are performed to apply a fabrication modification determined as described with reference to Figure 5. The fabrication modification is customised to this specific laser, using the parametric test results, to compensate for parametric variation and / or parametric test fails and / or threshold gain spectrum variation and / or spatial hole burning.
[0175] Different respective fabrication modifications are applied to respective lasers by one or more fabrication step, to compensate for parametric variation and / or parametric test fails and / or threshold gain spectrum variation and / or spatial hole burning. In this example, the fabrication modifications comprise removal of material from the laser nonuniformly along the respective cavity that provides a nonuniform effective refractive index distribution along the cavity, customised for each respective laser.
[0176] Figure 9d illustrates the DFB laser of Figure 9c after a material removal fabrication step has been performed to apply the fabrication modification to the laser. Dielectric 904 is etched to leave a gap 906 using a mask that is lithographically patterned. The material of the dielectric, is therefore now distributed nonuniformly along the cavity. The nonuniformity of the dielectric material causes a nonuniform refractive index distribution along the cavity, and this leads to a customised nonuniform effective refractive index distribution that compensates for threshold gain spectrum variation. The uniform injection current is supplied through the bond 970. Upon re-testing, the modification means that the laser will be performing in specification for the singlemode parametric tests. Thus, yield for the wafer is increased.
[0177] In other examples, the nonuniform material distribution resulting from nonuniform addition / removal can affect the stress in the cavity or heat dissipation from the cavity, depending on the size and shape of the nonuniform material. This can also lead to a customised nonuniform stress or thermal distribution that compensates for threshold gain spectrum variation.
[0178] Figure 10a illustrates testing a DFB laser 1052, with contact segments wired together in a comb shape. The injection current is supplied through the probe needle 1016 and combed contact 1018 uniformly along the cavity, with each contact segment 1016a-1016f supplying the same injection current to its respective underlying sections of the cavity. Figure 10b illustrates the DFB laser 1052 of Figure 10a after a wafer-level etching fabrication step has been performed to apply a fabrication modification to the laser. The fabrication modification is determined as described with reference to Figure 5. The injection current is supplied through the wire bond 1070. Lithographic patterning and etching fabrication steps for the fabrication modification results in an open circuit 1020, that disconnects one contact segment 1016b, while leaving connected together all other contact segments 1016a and 1016c-1016f. The disconnection of contact segment 1016b causes nonuniform current injection, thus the customised nonuniform injection current distribution along the cavity leads to a nonuniform customised carrier density distribution that compensates for threshold gain spectrum variation. Upon re-testing, the modification means that the laser will be performing in specification for the single-mode tests. Thus, yield for the wafer is increased.
[0179] Figure 11 is a graph of output power P (in mW) versus ratio of current density J to threshold current density Jth (J / Jth, referred to here as bias current) showing the effect of fabrication modifications provided by embodiments of the present invention.
[0180] The graph shows simulated optical power versus bias current for one DFB laser with a specific facet-phase condition with three different current injection configurations. The first configuration 1102, 1104 is with uniform current injection as shown for example in Figures 7a, 7b and 8a. For the high bias current J / Jth from approximately 4.3 to 7 the plot is linear, corresponding to single-mode operation. However, it can be seen that with uniform injection this device is multimoded 1 104 for a lower range of the bias current J / Jth up to approximately 4.3, which is a failure mode for a single mode laser. This is yield loss due to threshold gain spectrum variation.
[0181] Because the single-mode parametric test has failed, one or more fabrication steps are simulated to apply a fabrication modification determined as described with reference to Figure 5. The fabrication modification is customised to this specific laser, using simulated parametric test results, to compensate for threshold gain spectrum variation and / or spatial hole burning. The second configuration 1106 simulates the same laser with a p-contact electrical contact with 15 segments. The fabrication modification provides a customised nonuniform injection current distribution along the cavity by disconnecting the second of the 15 segments. For the bias current J / Jth from 1 to 7, the plot 1106 is linear, corresponding to desired single-mode laser operation.
[0182] The third configuration 1108 simulates the same laser with a p-contact electrical contact with 6 segments, as described with reference Figures 6a to 6c. The fabrication modification provides a customised nonuniform injection current distribution along the cavity by disconnecting the second of the 6 segments. This modification is described for example with reference to the first laser 750 in Figure 7c, or the first laser 780 in Figure 7e, or the laser 852 in Figure 8b. For the bias current J / Jth from 1 to 7, the plot 1106 is linear, corresponding to desired singlemode laser operation.
[0183] For the latter two configurations 1106, 1108, because the fabrication modifications were determined to compensate for parametric variation and / or parametric test fails and / or threshold gain spectrum variation and / or spatial hole burning, the yield loss due to threshold gain spectrum variation and / or spatial hole burning is mitigated. Upon re-testing, the single mode parametric performance is within specification, so the yield improves.
[0184] Figure 12 shows a simulation of optical power spectra, showing a single mode wavelength shift before and after fabrication modification by hardwiring of a heater element to provide heating power nonuniformly along the cavity, in accordance with embodiments of the present invention. The horizontal axis is wavelength (nm) and the vertical axis is optical power P (dBm / nm). The spectra before 1202 (dashed line) and after 1204 (solid line) the fabrication modification are shown. The peak lasing frequency 1202 before modification is out of specification. This represents a parametric test fail. It can be seen that the lasing wavelength (i.e. highest peak) shifts to a lower, in-specification, wavelength after the modification. Thus, the fabrication modification to the laser compensates for a parametric test fail caused by the threshold gain spectrum variation and / or spatial hole burning. Figure 13 shows a simulation of optical power spectra, showing multimode wavelength shift before and after fabrication modification by hardwiring of segments of the segmented electrical contact to provide a nonuniform carrier injection distribution along the cavity, in accordance with embodiments of the present invention. The horizontal axis is wavelength (nm) and the vertical axis is optical power P (dBm / nm). The spectrum before 1302 (dashed line) and after 1304 (solid line) the fabrication modification are shown. The peak lasing frequency 1302 before modification is out of specification. It can be seen that the lasing mode (i.e. highest peak) switches to another mode on the other side of the stopband 1306 at a lower, in-specification, wavelength, after the modification. Thus, the fabrication modification to the laser compensates for a parametric test fail caused by the threshold gain spectrum variation and / or spatial hole burning.
[0185] The skilled person, when looking at the modified laser apparatuses described herein, would be able to identify the fabrication modifications using standard methods and tools found in a semiconductor failure analysis laboratory. They would observe different respective fabrication modifications applied nonuniformly to respective lasers. They could perform computer simulations to observe the effect of the modifications on parametric tests and on threshold gain spectrum variation and / or spatial hole burning. They could use standard tools for localised etching and deposition to undo the modifications, then retest the lasers to observe the uncompensated parametric test fails caused by threshold gain spectrum variation and / or spatial hole burning, such as shown by the dashed “before” graphs shown in Figures 12 and 13. By analysing a plurality of lasers in these ways, it would be apparent to the skilled person that the different respective fabrication modifications compensate for parametric variation and / or parametric test fails and / or threshold gain spectrum variation and / or spatial hole burning.
[0186] Embodiments provide improved efficiency and flexibility of post-test modification. As well as the advantages of wafer-level fabrication related to efficiency of processing many lasers in parallel, embodiments provide increased flexibility for the laser process designer in modification methods available for post-test modification, rather than them being limited to processes such as laser trimming and cutting. This is because wafer-level lithographic patterning provides more options for customising the shape of modifications to respective lasers. Furthermore, wafer-level steps such as deposition, implantation and etching, efficiently allow adding as well as removing material to make a wider range of types of modifications.
Claims
Claims1 . A method of manufacturing a distributed feedback laser apparatus, the method comprising the steps:(a) fabricating a plurality of distributed feedback lasers on a wafer substrate, each laser comprising a respective laser cavity;(b) on-wafer testing the lasers to generate respective lasers’ parametric test results;(c) determining, using the respective lasers’ parametric test results, different respective fabrication modifications customised to respective lasers to adjust each respective laser cavity nonuniformly to compensate for variation in the parametric test results; and(d) performing one or more wafer-level fabrication step comprising lithographic patterning on the wafer substrate having the plurality of the lasers fabricated on it, to apply the different respective fabrication modifications to the respective lasers.
2. The method of claim 1 , wherein the respective fabrication modifications are determined to compensate for threshold gain spectrum variation.
3. The method of claim 1 or claim 2, wherein the respective fabrication modifications are determined to adjust each respective laser cavity nonuniformly axially along the cavity.
4. The method of any preceding claim, wherein the respective fabrication modifications are determined to adjust each respective laser cavity nonuniformly transversely across the cavity.
5. The method of any preceding claim, wherein the respective fabrication modifications adjust each respective laser cavity nonuniformly by providing a nonuniform effective refractive index distribution to each respective laser cavity.
6. The method of any preceding claim, wherein the respective fabrication modifications adjust each respective laser cavity nonuniformly by providing a nonuniform carrier injection distribution to each respective laser cavity.
7. The method of any preceding claim, wherein the respective fabrication modifications adjust each respective laser cavity nonuniformly by providing a nonuniform stress distribution to each respective laser cavity.
8. The method of any of claims 1 to 6, wherein each respective laser comprises a segmented electrical contact arranged to inject carriers to the respective laser cavity.
9. The method of claim 8, wherein the on-wafer testing is performed by providing a uniform carrier injection distribution to the respective laser cavity.
10. The method of claim 9, wherein the on-wafer testing is performed by wiring segments of the segmented electrical contact together to provide the uniform carrier injection distribution to the respective laser cavity.1 1 . The method of any of claims 8 to 10, wherein the respective fabrication modifications comprise hardwiring of segments of the segmented electrical contact to inject carriers nonuniformly to the respective laser cavity, customised for the respective laser.
12. The method of any of claims 1 to 7, wherein the fabrication modifications comprise addition of material to the respective laser to provide material nonuniformly to the respective cavity, customised for the respective laser.
13. The method of claim 12, wherein the respective fabrication modifications comprise addition of dielectric material, to provide a nonuniform effective refractive index distribution to the respective laser cavity.
14. The method of claim 12, wherein the respective fabrication modifications comprise implantation of material, to provide a nonuniform carrier injection distribution to the respective laser cavity.
15. The method of claim 12, wherein the respective fabrication modifications comprise addition of material, to provide a nonuniform stress distribution to the respective laser cavity.
16. The method of any of claims 1 to 7, wherein the respective fabrication modifications comprise removal of material from the laser to provide material nonuniformly to the respective laser cavity, customised for the respective laser.
17. The method of claim 16, wherein the respective fabrication modifications comprise removal of dielectric material, to provide a nonuniform effective refractive index distribution to the respective laser cavity.
18. The method of claim 16, wherein the fabrication modifications comprise removal of material, to provide a nonuniform stress distribution to the respective laser cavity.
19. The method of any preceding claim, wherein the one or more fabrication step to apply the different respective fabrication modifications is performed concurrently for a plurality of the lasers.
20. The method of any preceding claim, wherein the step of performing one or more wafter-level fabrication step to apply the respective fabrication modifications comprises one or more step from the group of fabrication steps consisting of: patterning, deposition, implantation and etching.21 . The method of any preceding claim, wherein the step of performing one or more wafter-level fabrication step to apply the respective fabrication modifications comprises modifying circuitry on a sub-mount and comprises mounting the laser on the sub-mount in an assembly step.
22. The method of any preceding claim, further comprising the step of re-testing the lasers having the respective fabrication modifications to generate a revised parametric test result.
23. The method of claim 22, wherein the step of re-testing the lasers is performed as a wafer-level testing step on the wafer substrate having the plurality of the lasers with respective fabrication modifications fabricated on it.
24. The method of any preceding claim, wherein the laser is an in-plane laser.
25. The method of any preceding claim, further comprising the step of fabricating a lithographically patterned structure for each respective laser, and wherein the step of on-wafer testing the lasers comprises supplying injection current uniformly along the respective laser cavity via the respective lithographically patterned structure.
26. The method of claim 25 wherein the lithographically patterned structure comprises a temporary short configured to supply injection current uniformly along the respective laser cavity by shorting together contact segments of the respective laser, and the method further comprises the step of removing the temporary short after the on-wafer testing.
27. The method of claim 25 wherein the lithographically patterned structure comprises a temporary contact configured to supply injection current uniformly along the respective laser cavity by contacting a semiconductor layer of the respective laser, and the method further comprises the step of removing the temporary contact after the on-wafer testing.
28. The method of claim 25 wherein the lasers are buried heterostructure lasers and the lithographically patterned structure comprises an offset contact that is offset laterally from the respective laser cavity and is configured to supply injection current uniformly along the respective laser cavity by contacting a semiconductor layer of the respective laser.
29. The method of claim 28 wherein the offset contact is not annealed before the step of on-wafer testing.
30. A distributed feedback laser apparatus, comprising:- a wafer substrate having;- a plurality of distributed feedback lasers fabricated on the wafer substrate, the lasers each comprising a cavity; and- different respective fabrication modifications applied to respective lasers post lasing test by one or more wafer-level fabrication step comprising lithographic patterning on a wafer substrate having the plurality of the lasers fabricated on it, the different respective fabrication modifications customised to respective lasers to adjust the respective laser cavities nonuniformly.31 . The distributed feedback laser apparatus of claim 30, wherein the fabrication modifications adjust the respective lasers’ cavities nonuniformly axially along the cavities.
32. The distributed feedback laser apparatus of claim 30 or claim 31 , wherein the fabrication modifications adjust the respective lasers’ cavities nonuniformly transversely across the cavities.
33. The distributed feedback laser apparatus of any of claims 30 to 32, wherein the fabrication modifications adjust the respective lasers’ cavities nonuniformly by providing different nonuniform effective refractive index distributions to the respective cavities.
34. The distributed feedback laser apparatus of any of claims 30 to 33, wherein the fabrication modifications adjust the respective lasers’ cavities nonuniformly by providing different nonuniform carrier injection distributions to the respective cavities.
35. The distributed feedback laser apparatus of any of claims 30 to 34, wherein the fabrication modifications adjust the respective lasers’ cavities nonuniformly by providing different nonuniform stress distributions to the respective cavities.
36. The distributed feedback laser apparatus of any of claims 30 to 34, wherein the lasers each comprise a segmented electrical contact arranged to inject carriers to their respective cavities, wherein the fabrication modifications comprise hardwiring of segments of the respective segmented electrical contact to inject carriers nonuniformly to the respective cavities, customised for each respective laser.
37. The distributed feedback laser apparatus of any of claims 30 to 35, wherein the fabrication modifications comprise addition of material to the lasers to provide material nonuniformly to the respective cavities, customised for each respective laser.
38. The distributed feedback laser apparatus of any of claims 30 to 35, wherein the fabrication modifications comprise removal of material from the lasers to provide material nonuniformly to the respective cavities, customised for each respective laser.
39. The distributed feedback laser apparatus of any of claims 30 to 38, wherein the one or more wafer-level fabrication step comprise one or more step from the group comprising: patterning, deposition, implantation and etching.
40. The distributed feedback laser apparatus of any of claims 30 to 39, wherein the lasers are in-plane lasers.41 . The distributed feedback laser apparatus of any of claims 30 to 40, wherein the lasers are buried heterostructure lasers and each respective laser comprises a lithographically patterned offset contact that is offset laterally from the respective laser cavity and is configured to supply injection current uniformly along the respective laser cavity by contacting a semiconductor layer of the respective laser.
Citation Information
Patent Citations
Transmitter photonic integrated circuits (TxPIC) and optical transport networks employing TxPICs
US20030095737A1
Method of tuning optical components integrated on a monolithic chip
US20050013330A1
Surface emitting dfb laser structures for broadband communication systems and array of same
US20050053112A1
Edge-emitting laser chip wafer layout that facilitates on-wafer testing of the lasers
US20160211644A1
Semiconductor Integrated Optics Element and Production Method Therefor
US20210006032A1