This invention relates to the area of optical receivers and more specifically in the area of assembly of optical detector modules.
Optical receiver modules used for receiving high speed—GHz-optical data signals propagating along an optical fiber are known to those of skill in the art. Typically within these optical receiver modules there is an optical detector electrically coupled to an amplifier circuit in such a manner that light from the optical fiber illuminates the optical detector, the optical detector generates photocurrent in response thereto, and the amplifier circuit amplifies this photocurrent. The optical qualities of the optical detector are typically determined at least in part by the material structure of the optical detector. For some ranges of wavelengths, the materials of choice for the optical detector are costly, and as such, semiconductor materials used for manufacturing the amplifier circuit and the optical detector are typically not the same. Thus, the prior art optical detectors must be electrically wired to the amplifier circuits using wires in order to conduct the photo current.
Typically, the amplifier circuit and the optical detector are purchased from third party vendors prior to assembly. Thereafter, the optical detector, the amplifier circuit, decoupling capacitors, and a module housing are assembled to form an optical receiver. Typically, the housing is designed for easy coupling to an optical fiber. Unfortunately, since these modules are used for receiving high speed optical data, the length of bond wires used to connect the optical detector to the amplifier is critical. These wires exhibit inductance and as such, when photocurrent levels are extremely small in the order or microamperes, variations in intensity of the high speed optical data may not be representative of the actual data transmitted due to the effects of these bond wires. Thus, the module may be more or less sensitive depending on an exact configuration and manufacture.
In manufacturing, manufacturers typically are unable consistently to achieve optimal optical operating characteristics for the assembled receiver modules because the wires coupling the detector to the amplifier circuit play an important role in the performance of the receiver module and are known to vary significantly in manufacture.
Furthermore, isolated testing of the amplifier circuit is not economical or effective without the optical receiver coupled thereto due to the frequency range of operation of the device. Thus, even when optimally assembled, the module may fail to meet desired performance characteristics due to amplifier shortcomings.
Finally, the performance of the module or some subset of the entire assembly will also be dependent upon the value, position, and performance of the power supply decoupling capacitors. These capacitors are often integrated into the module by the manufacturer and contribute to the difficulty of designing a manufacturable module.
As a result a need therefore exists to manufacture the receiver module in such a manner that facilitates testing of the receiver module as a complete system in order to eliminate effects that yield undesirable performance prior to selling thereof. Unfortunately, due to the costly nature of many of the optical receiver semiconductor materials, integration of the module into an integrated circuit format is not considered practicable. For example, different material processing systems commonly rely on wafers having different sizes. Thus, a same wafer mask is not usable with the different processes. This greatly increases the design and manufacture costs for implementing a fully integrated or monolithic photodetector with amplifying circuit.
It is therefore an object of the invention to provide an optical receiver module and method of testing thereof that provides improved performance and performance consistency of the optical receiver module finished product.
In accordance with the invention there is provided an optical receiver comprising:
In accordance with an aspect of the invention there is provided a method of testing an optical receiver circuit residing on a semiconductor wafer comprising the steps of:
In accordance with yet another aspect of the invention there is provided an optical receiver comprising:
The invention will now be described with reference to the drawings in which:
a and
In
c illustrates an integrated circuit having bonding pads for receiving a flip-chipped optical detector;
d illustrates the integrated optical detector having a light receiving photosensitive area adjacent two metalized pads;
e illustrates back illumination of the integrated optical detector;
a illustrates a second orientation of the integrated circuit and integrated optical detector direct attached thereon;
b illustrates an on chip DC—DC converter and integrated avalanche photodiode (APD);
b illustrates a plurality of integrated optical receivers on a chip prior to testing;
c illustrates a wafer after testing the plurality of optical receivers on a chip;
In
The prior art optical receiver module (as shown in
Unfortunately, when these components (103, 101, 102) are placed within the TO-46 package, there is very little space left within the package for positioning of these components (103, 101, 102) as well as additional components. The typical placement of these components (103, 101, 102) used in the prior art involves positioning of the optical detector in a geometric center of the housing 100b, with the TiA 101 positioned between two header pins 104c and 104b, and the de-coupling capacitors 103 placed on either side of the optical detector 102 between header pins 104c and 104d, and 104a and 104b. Not to mention that the placement of these components (103, 101, 102), as well as lengths of bond wires used to form connections therebetween, are critical in terms of optical receiver performance. Especially critical is the bond wire thickness and length between the optical detector 102 and the TiA 101. If these bond wires 106 are too thin or too long, then the effects thereof will adversely affect optical receiver performance. Thus, in order to ensure optimal performance of the optical receiver module, component types (103, 101, 102) as well as the types of bond wires 106 used for internal connections, are critical. Therefore, through careful design and component selection optimal performance may be achieved, however this comes at a cost of having to fully assemble the optical receiver module 100 prior to testing. In other words, an assembly that fails testing, represents a loss of all costs used for assembly thereof.
Unfortunately, the optical receiver modules are not tested until they are fully assembled. This significantly affects optical receiver module manufacturing costs. During manufacturing, multiple TiAs 101 are formed on a semiconductor wafer. Due to manufacturing variances, performance of the TiAs varies across the wafer and in some cases the wafer will have some sections that have undesirable performance. Since only the individual components are tested, it is unknown how the manufactured TiA 101, and optical detector 102 wire bonded thereto, will operate until the device is fully assembled in the housing (100a and 100b). Thus, significant manufacturing costs are incurred because of the unknown performance characteristics of the optical receiver until final packaging.
In
The integrated optical detector 202 is mounted to the integrated circuit 201 using a direct attach technique in the form of “flip-chip,” or “bumping.” The terms, flip-chip, or bumping, are known to those of skill in the art and their meaning is clarified hereinbelow for the purposes of this specification and the claims that follow. The connection side (
The metalized pads on the integrated circuit are first gold plated. Then the integrated optical detector 202 is positioned with the upper surface of the integrated circuit (
Direct attach technology advantageously eliminates bond wires therebetween and ensures the integrated optical detector 202 is in close proximity to the integrated circuit 201 with the integrated optical detector 202 only touching the integrated circuit at preferably two points (as seen in
In
A second orientation of the integrated circuit and integrated optical detector direct attached thereon, for optical receiver module 300, is shown in
Advantageously, either the first or the second orientations of the integrated circuit and direct attached integrated optical detector allows for a complete ‘receiver on a chip’ solution that obviates the need for external components. Additionally, vertical stacking of the integrated optical detector and the integrated circuit maximizes space efficiency and allows for a larger die-size that that which was attainable in the prior art. Using a larger die of course decreases the number of devices manufacturable on each wafer, however the benefits that are achieved outweigh the additional costs.
The integrated optical detector is typically fabricated using the most suitable technology in dependence upon a desired wavelength band of operation. Thus, typically for telecommunications purposes the integrated optical detector is manufactured using InP, whereas the integrated circuit is typically manufactured using Si. Thus, direct attaching is highly advantageous since a silicon detector is typically not useable for receiving optical wavelengths used in telecommunications. Of course, to those of skill in the art it would be obvious to eliminate the direct attach process if the integrated detector is manufactured of the same semiconductor material as the integrated circuit. However, this proves problematic due to the limitations imposed within each material process, the costs of the material processes, and the size of the finished integrated circuit.
In
The testing apparatus 500, shown in
c illustrates a scenario where the wafer 420 has been tested and the plurality of optical receivers on a chip 402 have been grouped on the wafer 420 based on their performance in terms of the predetermined criteria. For instance, the optical receivers on a chip 402 that have a first performance characteristic within a range of performances are grouped in area 450, the optical receivers on a chip 402 that have second performance characteristic within a different range of performances are located in area 451 and those that have a third performance characteristic within another different range of performances are located in area 452. For commercial applications this allows for the manufacturers of these optical receivers on a chip 402 to “bin” the optical receivers into groups having known predetermined performance characteristics, thus advantageously allowing for determination of yield and quality prior to dicing of the wafer 420. If for some reason the process used to manufacture these optical receivers on a chip 402 is not adequate, then costs will be avoided because the wafer is scrapped prior to the expensive process of dicing and assembly. Moreover, partial wafer testing provides an early indication of total yield and allows one to determine if further testing is warranted. Thus, a yield loss is seen at the wafer level and not at the stage of integration into the module, advantageously saving manufacturers money and testing time. Once packaged into optical receiver modules, the performance characteristics are known so the components are binned appropriately. Thus, for end user, using the tested optical receiver module is much easier because critical receiver design parameters are assured through the device specifications, which are different for devices in different bins. This provides guaranteed optical performance of the optical receiver sold to the module manufacturer and commands a price premium over the individual components sold to the manufacturer under the prior art methods.
Moreover, price is typically established based on the performance characteristics of the devices allowing for enhanced profit margin for those components having a most advantageous performance characteristic.
Numerous other embodiments may be envisaged without departing from the spirit or scope of the invention.
Number | Name | Date | Kind |
---|---|---|---|
5198684 | Sudo | Mar 1993 | A |
5252852 | Makiuchi et al. | Oct 1993 | A |
5365088 | Myrosznyk | Nov 1994 | A |
5394490 | Kato et al. | Feb 1995 | A |
5576831 | Nikoonahad et al. | Nov 1996 | A |
6396116 | Kelly et al. | May 2002 | B1 |
6649994 | Parsons | Nov 2003 | B2 |
6731122 | Feng | May 2004 | B2 |
6859031 | Pakdaman et al. | Feb 2005 | B2 |
20020135036 | Terano et al. | Sep 2002 | A1 |
Number | Date | Country | |
---|---|---|---|
20040081473 A1 | Apr 2004 | US |