ENCAPSULATED GaAs-BASED LASER DEVICES
REFERENCE TO RELATED APPLICATION This application is based on and claims priority to US Provisional Application
No. 60/524,478, hereby incorporated by reference.
FIELD OF INVENTION
This invention relates generally to optical non-hermetic packaging, and, more specifically, to non-hermetic packaging of gallium arsenide (GaAs) substrate semiconductors such as vertical cavity surface emitting lasers (VCSELs).
BACKGROUND
An important component of an optical telecommunication system is the opto- electric interface which converts signals between the optical domain and the electrical domain. Typically, the opto-electric interface comprises some kind of opto-electric device (OED) for either transmitting or receiving optical signals.
OEDs are commonly packaged in transistor outline packages or "TO cans." A TO can comprises a semiconductor laser device mounted on a header and sealed with a cap to form a hermetic package.
Although extensively used, TO cans suffer a number of drawbacks including relatively high cost. Pressure to reduce costs has mandated that manufacturers review and consider alternative packaging for OEDs other than the relatively costly and difficult to handle TO cans. One promising alternative is silicon-based platforms which not only support the active component, but also provide a substrate for interconnecting active devices and electronics. The need for lower cost components has also lead to the increased use of vertical cavity surface emitting lasers (herein
"VCSELs"). VCSELs can be manufactured using standard microelectronic fabrication
method allowing them to be integrated on-board with other components without requiring pre-packaging. Of particular interest are 850nm GaAs-based VCSELs which are particularly well suited for data transceivers. Such VCSELs typically have an oxide aperture structure, in which a native oxide is formed on a thin AlAs layer in the structure to provide current confinement.
These alternative packaging approaches are used on a number of products offered by Tyco Electronics Corporation (Harrisburg, PA). For example, referring to Fig.4a, its LC transceiver product uses an optical sub-assembly (OS A) 400 for providing the optical interface between the OEDs and the optical fiber. The OSA 400 comprises an optical block 410 which is optical grade plastic molded to define a connector interface 401 for an optical connector fiber connector and a cavity 402 for receiving an OED. The OSA 400 in this embodiment is designed to utilize surface mount lead frame 403, 404 style packages, which provide easy and fast techniques for assembling surface emitting devices. The lead frame sub-assemblies 403, 404 are easily aligned and set in place utilizing UV cure epoxy. The resulting optical assembly provides a compact and optically stable package with environmental protection for the enclosed components.
The initial design concept for this packaging approach centered upon the utilization of VCSELs having enhanced moisture resistance. This moisture resistance was effected through the application of an auxiliary protective coating covering the
VCSEL emitter area. In addition, components would be screened at the wafer level for a low failure rate when exposed to a high moisture environment. This approach was implemented for transceiver production and showed initial success. Over time, however, VCSEL failures in the transceivers became problematic. Therefore, there is a need to package GaAs-based VCSELs in a way which extends their useful life. The present invention fulfills this need among others.
SUMMARY OF INVENTION
The present invention involves the applicant's discovery of the problem contributing to the long-term degradation of GaAs-based VCSELs, likely causes for the problem, and a solution targeted at the causes of the problem. The applicant discovered through failure analysis that VCSEL degradation was occurring due to moisture effects. While the semi-hermetic plastic package was protecting the components from gross exposure to moisture, long term exposure was apparently allowing enough moisture into the assembly interior to cause the VCSEL to degrade. This type of failure was also demonstrated in subsequent Bellcore 85/85 testing, an aggressive accelerated mode of environmental testing. The conclusion was that the plastic package approach was not adequate to protect the VCSEL in these environments, and that the VCSEL coating alone was not a sufficient moisture barrier.
The applicant has identified several causes for the failure in VCSELs. First, the failure of VCSEL lasers in a high moisture environment is due in part to the undesirable growth of an oxide film over the light-emitting active area of the device.
This additional uncontrolled oxide film results as the ambient moisture chemically reacts with the active area. As the oxide layer grows, it imparts mechanical stress in the VCSEL, causing multiple defects in the device crystal structure. Eventually the high stresses result in the degradation and failure of the device. More specifically, unlike InP substrates in which the relatively large In atoms serve to halt the propagation of point defects, in a GaAs substrate, the defects are relatively free to propagate until the device fails. It has also been found that, when the VCSEL is operating with an applied voltage, electro-chemical reactions at the device active area can also occur while moisture is present. This is the component version of electro-galvanic corrosion. This effect causes ionic migration paths to be produced, delaminating the oxide or nitride interfaces and even corroding the wire bond to sub mount interface. This effect will also cause oxidation as well as other reactions that will stress the device and lead to premature failure.
Applicant has also determined that prior art attempts at protecting the oxide layer with a nitride layer are ineffective over time because a typical nitride layer contains microscopic defects that provide access to the oxide layer. Over time, moisture reaches the oxide layer through these defects and precipitates the failure mechanisms described above.
Another approach for protecting GaAs-based laser devices from moisture damage is therefore needed. To this end, the present invention provides a cost- effective approach for moisture protection of the laser device through encapsulation. In accordance with this technique, a GaAs-based laser device is covered with a protective coating such that moisture cannot reach the oxide layer. h a preferred embodiment, the coating is a silicone gel. The silicone gel encapsulation prevents high levels of water vapor from contacting not only the active area of the device, but also the oxide aperture, which, as mentioned above, is where point failures of a GaAs-based laser device are likely to initiate. Specifically, the oxidation and chemical reactions, which contribute to premature failure, do not occur as the moisture is effectively prevented from contacting the oxide layer. In addition, the silicone is an excellent electrical insulator. Thus, the device is not only isolated from moisture, but also is coated with an electrically insulating barrier which inhibits electro-chemical reactions. The ability to prevent these unwanted reactions, while providing a stress free, insulating, and optically transparent environment makes encapsulation a highly attractive approach.
One aspect of the present invention is a GaAs-based device encapsulated with silicon gel. In a preferred embodiment, the device comprises: (a) a GaAs substrate; (b) an active region defined in or on the GaAs substrate; (c) a layer of oxide covering the active region to define an aperture through which light generated in the active region is emitted from the device; and (d) a coating encapsulating the oxide layer.
Another aspect of the present invention is an optical package comprising a coated GaAs-based device. In a preferred embodiment, the optical subassembly
comprises a GaAs-based device as described above and a non-hermetic housing containing the device.
Yet another aspect of the present invention is a transceiver comprising the optical subassembly described above, a prefeπed embodiment, the transceiver comprises the optical' subassembly described above connected to a circuit board for electrically interfacing to a hqst device.
BRIEF DESCRIPTION OF FIGURES
Fig.l show a schematic of a preferred embodiment of the invention. Fig. 2 shows a prior art VCSEL.
Figs. 3a and 3b show experimental results of the encapsulated laser device compared to a non-encapsulated equivalent laser device.
Figs. 4a and 4b show lead frames disposed in an optical block in a non-potted and potted state respectively. Fig. 5 shows an exploded view of an optically-aligned sub-assembly.
Figs. 6 (a)-(d) are perspective views of a preferred embodiment of an optical block.
Fig. 7 shows a cross section of a transceiver having the optical subassembly shown in Fig. 5. DETAILED DESCRIPTION OF PREFERRED EMBODIMENT
Referring to Fig. 1, a semiconductor device 10 of the present invention is shown. As shown, the, device comprises a GaAs substrate 11. An active region 12 is defined in or on the GaAs substrate 11 and a layer of oxide 13 covers the active region 12 to define an aperture 13 a, through which light generated in the active region is
emitted from the device. A coating 14 encapsulates the oxide layer 13 to prevent moisture from reaching and degrading it and causing the eventual failure of the device.
The GaAs-based device may be any known device having a GaAs and an aperture which is defined by an oxide layer. Preferably, the GaAs-based device is a surface emitting laser (SEL), and, even more preferably, it is a VCSEL. Such devices are well known. A VCSEL is described, for example, in U.S. Patent No. 5,359,447. Referring to Fig. 2, the manufacture of a VCSEL is considered. The VCSEL is grown on an n+GaAs (gallium arsenide) substrate 31. A bottom output minor, for example 18.5 pairs of n-doped GaAs/ALAs (gallium arsenide/aluminum arsenide) quarter- wave layers (generally designated 33 in the drawing), is epitaxially grown on the substrate
31. The interface between the layers is graded using an AlAs/GaAs/Al03Gao 7As variable duty cycle short period superlattice ("SPSL"). The SPSL reduces any heterojunction band discontinuities at the GaAs/AlAs interface. The doping level is lxlO18 cm"3 in uniform regions and 3x IO18 cm"3 in graded regions. For simplicity only a few of the 18.5 pairs of layers are shown in the figure. The reflectivity of the bottom mirror 33 is 98.9%.
Next an optical cavity structure 35 is grown. The cavity structure includes an n-cladding layer 37, a quantum well 39, and a p-cladding layer 41. The cladding layers 37 and 41 comprise Al03Gao7As doped to 1x10 1018 cm"3 reduced to 5xl017 cm"3 adjacent the quantum well 39. The quantum well 39 comprises 3 MQW of strained
I 2Gao 8As (indium gallium arsenide) having a thickness of about 80 A, with GaAs barriers having a thickness of 100 A.
Above the quantum well 35 is a highly-reflective top mirror 43. The reflectivity of the top minor is greater than 99.96%. The top mirror 43 comprises, for example, 15 pairs of GaAs/AlAs quarter wave layers (generally designated 45), a phase matching layer 47, and an Au (gold) layer 49. A proton isolation region 51 surrounds the perimeter of the quarter wave layers 45. As with the bottom mirror 33, only a few of the quarter wave layers 45 are actually shown in Fig. 2. The interfaces between the quarter wave layers are graded in a manner generally similar to the grading of the
interfaces in the bottom mirror 33. The doping levels are lxl 018 cm"3 in uniform regions and 5xl018 cm"3 in graded regions.
The phase matching layer 47, which is GaAs, compensates for phase delays that result from finite penetration of the optical field into the Au layer. The Au layer 49 is about 2000 A thick and is fabricated, after MBE growth of the underlying structure, as follows. First a 2000 A layer of Au is deposited on the GaAs phase matching layer 47. Then a thick (more than lOμm) Au button is plated on top to serve as a mask for proton isolation. The wafer is then proton implanted. Crystal structure damage that results from the proton implantation provides for cuπent confinement and therefore gain guiding. Then another thick Au button 53 with a diameter of about
300 m is plated on top. This button 53 is used for solder/die attachment of the completed device to a heat sink. The wafer is then lapped and polished to a diameter of 125μm and an annular electrode 55 is patterned on the bottom. A quarter-wave anti- reflection coating 57 of SiO2 (silicon dioxide) is deposited in the open region of the electrode 55.
Upon the oxide layer of the VCSEL, the moisture-resistant coating 14 is applied. The coating must be sufficient to prevent water from reaching the oxide layer of the GaAs-based laser. To this end, it is preferable to use a coating which has fluid properties so the coating may flow over the device and into it various contours. To remain situated on the GaAs-based laser, the coating should also be curable. Suitable coatings include commonly utilized include epoxies, polyimides, silicones, acetates, acrylates with various advantages and disadvantages for each material. Preferably, the coating is a gel, which is any convenient gel or polymer which is curable, generally at or near ambient temperature and with a minimum generation of heat. Typical curable gels which are utilized for this purpose include silicone gels, cellulose butyrate acetate, poly methyl methacrylate, cyanoacrylate, etc.
Through investigation of the various coatings and their applicability to the VCSEL optical subassembhes, silicone gel is preferred. Silicone gel encapsulants have
a number of unique properties that are advantageous for VCSEL encapsulation besides just moisture resistance, including:
• Coatings are optically transparent. • Ultra high purity for semiconductor compatibility. • Gel properties which cause no additional mechanical stress on device junctions or wire bonds. • Self healing if mechanically violated. • Extreme thermal stability. • Properties combining the best of a liquid and a solid-soft but will not run. It has been observed that silicone gels tend to absorb a certain amount of moisture before becoming saturated and not allowing any other moisture through. The amount of moisture the gel can absorb can vary although it tends to be about 0.1%. At this level, the moisture has very little effect on the oxide layer and thus on the overall performance of the device. The amount of coating to apply to the device can vary although generally it is desirable to apply as thick a layer as possible without diminishing the transmittance of the device. Typically this layer ranges form about 0.01 to about 1mm.
The GaAs-based device of the invention provides for reliable use of non- hermetic or semi-hermetic subassembhes. Of particular interest is its use in an optical subassembly as described, for example, in US Application No. 09/901,293 which is hereby incorporated by reference in its entirety.
Referring to Figs. 4a and 4b, an optical subassembly 400 comprising the GaAs- based device is shown. Specifically, Fig. 4a shows the optical subassembly 400 which comprises a connector interface 401 and cavities 402 for receiving lead frames 403, 404. This particular connector interface is adapted for mating with an LC optical connector, although other connector interfaces are possible (see, e.g., Fig. 5 which shows an MTRJ optical interface). The GaAs-based device is mounted on one or both
of the lead frames 403, 404. Fig. 4a shows the lead frames in the cavities 402 in an unpotted state, while Fig 4b shows them in a potted state.
Referring to Figs 5 and 6, the optical subassembly and its manufacture are considered in greater detail. In a preferred embodiment, an optically-aligned subassembly is used in conjunction with a preferred assembly method to maintain critical optical alignments and to minimize stress along the optical paths. The concept behind the optically-aligned subassembly is to provide a robust package of all the components which effect the optical coupling from the fiber assembly to the OEDs.
Referring to Figure 5, an exploded view of the optically-aligned subassembly 500 is shown. Central to the optically-aligned subassembly 500 is the optical coupling
511, which is shown in its prefeπed embodiment as optical block 516. The optical block 516 is described below in detail with respect to Figs. 6a through 6d. Connected to the optical block 516 are a number of other components including, the connector interface 512, clam shell connectors 521a, 521b to secure the connector interface 512 to the optical block 516, and opto-electric devices 513 operatively connected to the optical block 516.
Preferably, the OED is manufactured using lead frame technology. It is an advantage of the present invention that the use of a lead frame structure allows for the simultaneous fabrication of a large number of opto-electronic packages, such as transmitters, receivers or transceivers. Similar to conventional electronic integrated circuit processing, a plurality of integrated circuits may be simultaneously attached and wire bonded to the lead frame. In accordance with the known manufacturing techniques, an associated plurality of optical devices are coupled to lead frames and the combination of electronics and optics encapsulated using a molding process (e.g., transfer molding) to form the final packaged assembly. When the molding operation is completed, lead frames may be severed from one another to form a plurality of final package assemblies.
In the transceiver module 700 shown in Fig. 7, the OEDs 513 are mounted to the surface 718a of a substrate 718. As the term is used herein, "substrate" refers to an electronic component having electronic circuit elements mounted thereto or forming part thereof. The substrate may include, for example, a plurality of integrated chips. Such chips may represent, for example, a pre-amplifier or post-amplifier and additional electronic circuits. The type and nature of such circuit elements, and the techniques and methods for mounting such elements to the substrate 718 are well known in the art and do not form part of the present invention. In typical embodiments, the substrate 18 comprises a printed circuit board (PCB), printed wiring board (PWB) and/or similar substrates well known in the art. The substrate 718 has an electrical interface in the form of connecting pins 719 depending therefrom and is adapted to mate with another substrate (not shown) of the host system. According to preferred embodiments, substrate facilitates a dual-inline package (DIP) adapted to be mounted to the motherboard or some other system board of a host system. Alternatively, rather than pins 719, the substrate 718 may have contacts aπanged on its side to facilitate card edge connections. Such an embodiment is preferred for pluggable modules and is well known in the art. In addition to pins and edge contacts, any other known means for interfacing the substrate with the host system may be used within the scope of the present invention. Figures 6a through 6d show a preferred embodiment of the optical block 516 in various stages of assembly. The optical block 516 couples light between the multi- fiber array of a connector assembly and a plurality of OEDs 513 in a module. The optical block 516 preferably comprises a unitary structure of an optically-clear moldable material comprising at least the following features: (a) a plurality of first lenses 660a, 660b adapted for interfacing with a multi-fiber array of the connector assembly, each first lens 660a, 660b corresponding to a fiber in the multi-fiber array; (b) a plurality of second lenses 667a, 667b adapted to cooperate optically with the OEDs 13, wherein each second lens 667a, 667b corresponds to a first lens 60a, 60b; and (c) one or more reflective surfaces 615a, 615b. The first lenses 660a, 660b, the seconds lens 667a, 667b and the reflective surfaces 615a, 615b, respectively, are
configured to provide two optical paths in the optical block, each optical path comprising a first section between a particular first lens and its coπesponding reflective surface, and a second section between the reflective surface and the corresponding second lens. The first sections preferably are parallel. The two optical paths defined by the optical block 516 pertain to a transmitting path and a receiving path since the optical block 516 is configured as a transceiver module. A detailed discussion of the lenses and reflective surfaces that effect these optical paths is set forth below, although it should be understood that the optical block of the present invention is not limited to just two optical paths nor is it limited to any particular combination of transmitting/receiving paths.
The lenses and reflective surfaces of the optical block are configured to effect the optical paths of the present invention. Since the first lenses 660a, 660b preferably are arranged non-axially to the second lenses 667a, 667b, and, in the embodiment shown in Fig. 6, are arranged at substantially a 90° angle in the y,z plane, some kind of light bending mechanism preferably is used. According to preferred embodiments, the light bending mechanism comprises means for altering the direction of a substantial portion of the light emitted by the light emitting device such that a substantial portion of emitted light is received by the light receiving device. The particular structure of the light bending means may vary widely, depending upon such factors as the particular emitting and receiving device being coupled, the portion of the light whose direction is to be altered, and the relative positions of the OED and the x,y fiber array. In general, however, it is preferred that the light bending means comprises reflecting means in operative optical association with the OED and with the x,y array for reflecting at least a portion of light emitted by the light emitting device onto the light receiving device. It will be appreciated by those skilled in the art that numerous structures are capable of performing this function. For example, one or more reflective means disposed at the appropriate angle relative to the operative axis of the OED and the light transmission axis may be used to achieve this result. According to preferred
embodiments described in more detail hereinafter, such reflective means comprises a reflective surface, such as a prism having an internal surface disposed at the appropriate angle with respect to the axes of the OED and the fiber optic transmission line. In order to minimize signal loss associated with the present coupling devices, it is preferred that the reflective means comprises a total internal reflection (TIR) prism. Since the path of light travel is reversible, the same light bending means may be used for embodiments in which the fiber optic transmission line is the light emitting device.
The same light bending means may also be used for multiple optical paths. More specifically, rather than using an individual reflective surface for each optical path, a single reflective surface may be used to bend a plurality of optical paths. Embodiments favoring the use of a single reflective surface for a plurality of optical paths include those in which the bending occurs at approximately the same location along the z axis so that the reflective surface may be a simple planer surface. One skilled in the art will be able to assess the need for collimating the light beam in the optical path or otherwise focusing the beam depending on the divergence from the light source, which may be, for example, an OED, a fiber end of the multi-fiber array, or a surface of the optical block. For example, light emitted from the active area of an OED may be in the form of a beam of substantially parallel light rays centered on and substantially parallel to the operative axis of the device. In such embodiments, the need for lensing is minimal, and the optical block preferably comprises the light reflecting means positioned in the path of the beam with no collimating lens. On the other hand, the OED may be a light emitting device which produces a substantially divergent source of light, such as a VCSEL or the end of a fiber. In such embodiments, it is preferred that the optical block 516 include one or more collimating elements in operative optical association with the divergent light source. The principal purpose of the collimating element of the present invention is to reduce the degree of divergence of the rays emitted from the opto-electronic device or the fiber optic cable. Such a collimating element is
preferably operatively associated with the light emitting source by aligning the optical axis of the lens with the operative axis of the light emitting device.
Since the collimating elements functions to focus divergent light form a divergent light source, it should be appreciated that one optical path may comprise a plurality of such elements. For example, in a typical optical path, where light is coupled between a fiber end and an OED, collimating elements may be disposed at the interface between the fiber end and the optical block, such an element is herein referred to as a "first lens", at the interface between the optical block and the OED, such an element is herein referred to as a "second lens", and at any point in between where the light encounters a surface interface, such as the cavity used to accommodate the reflective surface 666 for feedback.
The collimating element may comprise any device capable of focusing light from a divergent source. Preferably, the collimating element comprises an optical power surface, such as a positive, aspheric lens. The lens may be discrete from or integral to the optical block. In embodiments in which the lens is discrete from the optical block, the lens may or may not comprise the same material as the optical block. For example, the lens may comprise glass or a different grade of optically- clear plastic, and it may be coated according to known techniques. Preferably, however, the collimating elements is internally molded to the optical block. Such an embodiment is advantageous since the collimating element and optical block can be formed in a single molding operation. Furthermore, an integrally-molded lens avoids the need for assembly and the alignment steps therefor. Indeed, as mentioned above, a principal advantage of using an integrally-molded optical block is fixing in a single component many of the critical alignments along the optical path.
EXAMPLES
Example 1 - Comparative testing
Before fully implementing gel encapsulation for production components, an accelerated test program was started to ensure that silicone gel would indeed provide the protection required. A group of 66 production VCSELs were received to use as test components. These parts were assembled and tested using the normal production processes, including "burn-in" and performance screening. The VCSELs were die bonded onto "airstrate" lead frames, along with the monitor photodiode, and wirebonded as usual. The parts were divided into three groups of 22 each: gel coated, uncoated, and control. The gel coated components were coated with silicone gel. The uncoated and control groups were left "as received".
Both the gel coated and uncoated groups were placed in an environmental chamber and exposed directly to a Bellcore 85/85 environment. ( 85C, 85% humidity ) This is an extremely aggressive test environment, as the component is significantly less exposed while packaged within the optical block assembly. The components were powered using 1.5 Volts. This voltage level will provide enough current to activate the device and cause electro-galvanic activity, but will not cause enough component heating to drive moisture away from the active area. The control group was placed in a dry N2 environment at 23 °C, as a check to see if there would be VCSEL failures without any environmental stress. It was also powered to 1.5 Volts.
All VCSELs were individually tested using an HP 8152A Optical Power Meter and an Agilent 81002FF Integrating Sphere/ 81521B Detector at a drive current of 6 mA.
After 2000hrs of testing, four of the twenty two uncoated VCSELs failed. All twenty-two gel coated components continued to operate normally with no reduction in output. All the control group components also operated normally.
Results of this testing is shown in Figs. 3a and 3b. The results of this aggressive test provide a high degree of confidence encapsulation provides a functional moisture barrier for VCSEL components.
Example 2— Functional Testing of Encapsulated Components in Transceivers
Twelve production qualified lead frames were received and the components were encapsulated with silicone gel. The lead frames were assembled into LCMM transceivers and subjected to formal Bellcore MVT qualification tests under Mil. Spec. 883. This testing was conducted to ensure that there would be no problems with the performance of a gel coated component when installed into production transceivers. Mil. Spec. 883 Testing included: 'Shock and Vibration: Passed
•Temp. Cycle/Moisture: Passed
•Temp. Shock: Passed
•Moisture Resistance: Passe
•Four Corner Performance Test: Passed The above positive test results indicates that the gel encapsulation approach produces a robust and reliable transceiver product.