Bob Shine, Jerry Bautista, Kevin Sullivan, and Bo Rotoloni
Integration of multiple devices is currently the most effective means of controlling costs and delivering functionality. Hybrid integration optimizes performance and overall yield of components based on planar lightwave circuits (PLCs).
Integrating multiple optical functions into a single device is a key step to lowering the costs of optical networks. When multiple functions are integrated into a single device, the costs of labor, packaging, and testing can all be reduced. Automating processes can reduce costs further. Device performance can also be improved by integration, in the simplest case by reducing the numbers of fiber interconnections and the insertion loss. The primary challenges to integration are finding a material that can perform multiple functions and understanding the impact that concatenating functions has on fabrication yields.
The path to integration most likely will take sequential steps, from packaging discrete elements together in modules through planar hybrid integration of multiple technologies, and eventually will lead to monolithic integration. To follow this evolutionary approach, component vendors must collaborate closely with system designers to understand the benefits and trade-offs of using multiple technologies. Component vendors must also develop strong packaging expertise, as this will be critical to success at each stage of integration.
Traditionally, optical-component vendors have provided discrete components to system integrators, who then retest them and splice them into modules for their systems. Outsourcing subsystem assembly can reduce the total costs for these system integrators. By receiving specifications at a module level, rather than for individual components, a component vendor can increase overall yield, avoid repeated testing steps, and lower labor costs. Since submodules can include proprietary designs or reveal elements of system architecture, however, integrators and component vendors must come to view each other as partners for subsystem integration to occur.
As it is with other steps to achieve a monolithically integrated product, packaging is a key element in design. While many optical component companies have cultivated abilities in micro-optic packaging, integrating discrete components into modules requires different skill sets. Careful thought must be applied to the manufacturing layout to guarantee proper interconnection of the many elements. Proficiency is required in different tasks, including mounting of the discrete elements, fiber routing, mass-fusion splicing, and fiber ribbonization (see Fig. 1). Rapid prototype development can be obtained through the use of stereolithography equipment, allowing testing of form and fit as well as optical function prior to volume manufacturing.
The next step is to combine multiple functions within a single package. This type of integration includes integrated filter and tap couplers; integrated isolator, tap coupler, and WDM filters; and integrated tap coupler and monitor products. With this approach, typically two (at most, three) elements are combined and the assembly still relies heavily on manual labor.
The use of PLC platform technologies such as silica on silicon or indium phosphide requires a much different set of competencies. These technologies allow a great number of discrete elements to be replaced by a single element, and this transition has often been likened to the evolution in the semiconductor industry, mirroring the change from discrete transistors to integrated circuits. The set of competencies required does match many from the semiconductor industry, but two key challenges for the photonics industry are quite different.
First, no single standard material has evolved to satisfy all the different building block needs of the photonics industry. While semiconductors must also perform a wide range of functions, silicon has has allowed equipment manufacturers to create a standard set of tools and processes. A second challenge arises from the complexity of interconnecting to single-mode fiber. Here alignment tolerances are vastly different from wire bonding to semiconductor elements. The result is that packaging of integrated optical components remains a key cost component, whereas in the semiconductor industry, the packaging is a small fraction of the product cost.
Materials and systems used to perform the many optical network functions include indium phosphide, lithium niobate, silica on silicon, microelectromechanical structures, and dielectric thin films.1 Of course, some of these network functions occur at remote locations, and integration is not always required. Dense wavelength-division multiplexing systems also place stringent performance demands on individual components—modulation speed for a modulator, or crosstalk performance on the multiplexer. As a result of these demands, individual device yields can be very difficult to integrate.
Hybrid integration appears to be the best approachto accommodate the performance needs of a system integrator while integrating multiple functions along the signal path. This type of integration includes both a module-level approach, in which elements are interconnected with fiber arrays internal to the product, as well as truly hybrid products with chip-to-chip interconnection. Products based on this approach will likely incorporate multiple channels and functions to offer significant cost, size, and performance advantages.
Improved manufacturability may result from integrating individual elements in a hybrid model, rather than creating a monolithically integrated product (see Fig. 2). Consider the yield impact of integrating three functions—an arrayed waveguide grating, a switch matrix, and a variable optical attenuator array—to create a reconfigurable add/drop module. Each of these functions can be performed using silica on silicon, but if they are integrated on a single wafer, then the cumulative yield is the product of the individual yields (assuming independent yield factors).
Let's assume a 70% yield for each function, a yield that many manufacturers would find acceptable. The combined yield of a monolithically integrated product is 70% x 70% x 70% = 34%. Since the entire device must be scrapped if any individual element fails, monolithic integration should only be done if the yield of each element is very high.
In contrast, hybrid integration allows individual components to be optimized and, with a robust interconnection method, should have minimal impact on overall yield. Additionally, individual components can be tested before integration, which simplifies the testing process.
Although the benefits of hybrid integration have been established, there are significant challenges to overcome. One potential outcome of these challenges may be the integration not just of functions within a product, but the collaboration of companies with complementary technologies to create products.
Packaging is a critical element in optical-component development, and hybrid integration adds a new level of complexity. For this reason, most current integrated packages consist of discrete modules. Individual elements are integrated in a package but still use fiber arrays for their interconnection. This arrangement allows the manufacturer to choose the best-of-breed technologies, resulting in a product with excellent overall performance. The manufacturer can also monitor cost trends and identify when a transition to a hybrid solution is required.
A key element of packaging is the chip-to-fiber connection. Waveguide components with a typical core dimension of 4 to 6 µm must be interconnected to single-mode fiber with an 8-µm core. Misalignment of more than 0.1 µm can result in significant attenuation loss. Fiber arrays allow rapid interconnection to high-channel-count devices. Since the pitches of both the fiber array and the waveguides are precisely defined, alignment is typically done by aligning the first fiber and then rotating the array around that fiber to align the last channel. In this way, a 40-fiber ribbon can be aligned with labor content identical to a two-fiber array, greatly reducing the per-channel cost.
Another challenge in packaging is to reduce the package size. Of course, while almost any integration will save space, the challenge still remains to offer as high a level of complexity in as small a package as possible. The most straightforward integration is a linear concatenation of functions, which can result in a larger box than desired. Often, the size of individual elements is limited by the waveguide properties of the material, specifically the refractive-index difference between the core and the cladding.
- L. Eldada, Opt. Eng. 40 (7), 1165 (July 2001).
Bob Shine is director of marketing, Jerry Bautista is chief technology officer and senior vice president of technology, Kevin Sullivan is vice president of engineering, and Bo Rotoloni is director of planar lightguide circuits and modules at WaveSplitter Technologies, 46430 Fremont Blvd., Fremont, CA 94538. Bob Shine can be reached at email@example.com.