By Michael D. Evans, DESc
Manufacturing Integrated Photonics Or—So You Want to Be an Entrepreneur? (Continued)
Production Is Not Development
The repetition of production processes is typically driven by economics and the confidence of the development cycle, not the next novel feature or cool science. Most development engineers seek to change the product or service and implement new capability or new designs. Production engineers seek to satisfy the minimum requirements necessary to create the most product, at the highest yield, in the least time interval, while meeting quality specifications established by customer requirements or development engineering. Good manufacturing engineers have a sense of profit and loss and know that production velocity can keep inventory and work-in-process, (WIP), low for better financial results. Good development engineers understand the customer requirements and schedules established by marketing and business development along with a deep knowledge of the technology available to improve the function and cost of the product line. One can understand how a fundamental objective to instigate change may conflict with a fundamental objective to maintain consistent levels of quality and control, Figure 1. Process-engineering or technology-transfer specialists often moderate the contention between change and the status quo by introducing improvements to tooling, capital equipment, testing, materials, processes and procedures. A good process engineer works with production engineering and is familiar with the equipment and methods of production but is also embedded in new product design efforts particularly where no existing process is available or a new material is introduced. The process engineer is typically responsible for weaving new product into production processes and handling the product through new offline processes. The process engineer provides in-line training for technical staff and acts as the arbiter of data collected on the quality and capability of the processes in relation to final product performance. A qualified development team often has a technical design lead, process engineer and manufacturing engineer to optimize communication, accelerate product introduction and handle the work required to prove out new product before sales. This team is responsible for prototype production using the exact methods and processes intended for scaled production. The responsibility can rest with fewer engineers, though the process slows and risk is incurred. Manufacturing is a disciplined process not unlike a military campaign, and as with any disciplined regimen some contention between the development team is inevitable and can be constructively managed by the product champion to achieve superior results.
The Role of Contract Manufacturing
New product introduction can explore contract manufacturing where the technical requirements and design specifications are bid to companies that specialize in similar production. Particularly with emerging photonic products, contract manufacturers are unlikely to have the tools and test capabilities required for product delivery and hence will purchase these and build the costs into the contract. The decisions concerning maintenance, process methods and product flow are the responsibility of the contract manufacturer. Captive manufacturing engineers and quality engineers provide oversight and interface. Although the equipment costs are borne by the contractor, they are still paid by the development organization so the decision to contract production is often related to the complexity and quantity of the new product and considers the nature of the contractor. Since the contracted manufacturer also must make a profit, the cost of contract manufacturing often is higher than for a captive concern. Labor costs, material transport and product distribution expenses can factor into the decision along with the character of the development corporation. There are no true contract manufacturers with dominant roles in photonics products, though small shops provide these services. Since the design information is lodged with the contract manufacturer, this information is learned by employees outside of the design organization and can influence control of intellectual property. Similarly, process development by the contract manufacturer is owned by the contractor and suited to the tools and methods developed for the product. Contract manufacturers can provide significant advantage for high-volume or high-mix production where a captive supplier requires additional surge capacity. In such cases, a dedicated production line or specialized dedicated tooling may be economically beneficial for new product release. As an example, specialized substrates were required for advanced photonic assemblies and commercial products were inadequate, but the quantity was attractive for the supplier to install dedicated equipment and processes for their single largest customer, providing attractive costs and essential materials. This path was chosen instead of captive manufacturing of the material since the changes were within the capability of a commercial supplier. Raw materials, semiconductors, optical subassemblies, housings and other components may be candidates for contract manufacturing, not just overall assembly and test. Many companies use contract manufacturing to build circuit card assemblies (CCA) since the operations are relatively similar for power, control and interface boards. The location of operations, quality and methods should be carefully audited and controlled particularly since long-distance supply lines are prone to disruption from transportation, political and economic factors. The cost of disruption can often lead to choices in favor of captive production.
The Role of Captive Manufacturing
The captive supplier typically has a variety of advantages over contract assembly or embedded operations. The manufacturing organization understands the characteristics of the processes, products, materials and methods needed for photonic product integration. This learning and know-how is often trade-secret and can be the basis of profitable and rapid product introduction. A captive supplier has unique knowledge and access to their development organization, and the tolerances and controls of product design can be optimized for rapid deployment. The captive supplier also has control over the work hours and maintenance schedules for equipment and can tailor capacity to demand with clear understanding of the exposure to inventory and work in process. The advantage of captive manufacturing is summarized as: a) intimate understanding of the process, materials, risks and options for affordable production, b) shortest supply chain logistic advantages benefiting product velocity and availability of prototyping using legitimate tools and techniques well-suited to product, c) commonality with tools and methods for any necessary repair or modifications for custom product requirements, d) the ability to tailor testing for variations in customer specifications. To re-emphasize, captive manufacturing provides a faster path for prototypes and familiarity of the processes so that development and manufacturing only need to learn new features once, instead of transferring between suppliers or outside organizations. A more detailed understanding of the operation is possible and can be reflected in the enterprise planning (ERP) system. The downside for captive manufacturing is capacity utilization which risks underused tooling and staff when the market cannot accept production. Each manufacturer must factor their risk of sales in consideration of a manufacturing strategy.
In any case, the methods of accelerated product introduction are built around known tooling, existing processes and development designs that are able to take advantage of tribal knowledge and market anticipation. Taking advantage of alternate capabilities for tooling, substitution of materials, improvement of tolerances for assembly and changes in packaging or interfaces provides ample design space for invention and market dominance. Corporate personality influences the choices for manufacturing preferences since the actions of technology leaders differ from those of fast followers or disruptive market players. A technology leader and disruptive suppliers may require captive facilities to accelerate learning and derive improvements to tools and processes that are capable of meeting the demands of new products. Careful review of industry practice, patent history and academic advancements can influence the choice of manufacturing methods, supply chain structure and choice of captive and contract suppliers. Knowledge of production tools from similar industries can reveal breakthrough capabilities not just for emerging products, but as an example, historical capabilities from the semiconductor industry can favor affordable microphotonic fabrication, integration and test. The telecommunication industry has historically consumed and driven many new photonic derivative products using modifications of existing tooling and practices.
Unique Circumstances for Emerging Photonic Systems
Over the past 20 years, precision assemblies and micro-electromechanical system integration has introduced breakthrough advances in photonic product design and integration. The ever-increasing layers for semiconductor fabrication and through silicon-via (TSV) technology have provided cooler, faster, and easier interconnection that promises new generations of photonic products. The firms that specialize in compound semiconductor heterostructures have further advanced lasers, nonlinear optics, and specialized structures that force re-examination of the processes, materials, and methods available for future products. Techniques like 3D printing have made it possible to prototype some products faster. New composites, materials, optical substrates, and process methods have improved performance to near theoretical limits. Still, advances in thermal design, mechanical stability, optical throughput, and electromagnetic design for speed and quantum-limited systems will continue to challenge the inventors of the 21st century. The micro-optical bench and integration of components draws from a wealth of inventions and know-how to continue to decrease the size and increase the speed of communication products, sensors including chip-scale LADAR, position sensors with optical gyros and accelerometers, and new products in data storage and high-performance vector and scalar processing. Stacked chip and 3D monolithic designs are emerging.
Advances in photonic and microphotonic products will require continued progress in engineering disciplines including:
- Packaging and Environments
- Assembly and Automation
- Automated Testing
- Tolerances and Precision
- Affordability and Repair
Each area is a lengthy topic, but to summarize, we have learned that packaging and environments require critical engineering attention to achieve advancements. We have gained new software and faster processors capable of simulations that speed product introduction and reduce errors and omissions. Although some learning is brought from the semiconductor and MEMS suppliers, new methods are needed for evermore-demanding environments including radiation-hardened systems for space and uncontrolled environments, for remote sensing or automotive systems. The global telecommunication expansion once provided many companies focused on automation for fabrication, alignment, precision assembly and testing, but the reduced size and increased complexity of evolving products will continue to challenge this specialty to provide affordable, reliable systems, particularly for consumer goods. Professional organizations like IEEE have developed standards as have international and other professional and industry trade groups. Custom interfaces are still common, particularly for internal subsystems where designers attempt to prevent external tampering or modification. Sufficient effort is spent on global standards for emerging product introduction to use standard interfaces or comply with accepted quality systems. The interfaces and new designs affect testing. As size changes and complexity increases the challenge of affordable testing will grow more complex and the desire for reliable internal test and self-diagnosis will re-occur as it has for decades. Non-contact test systems and reliable product handlers will often require customization so internal automation teams or an external contractor relationship is highly desirable for the modern photonic manufacturer.
The need for improved precision and better accuracy never diminishes, and as submicron precision is displaced by subnanometer precision the requirements for power and performance will stress the fundamental materials. New materials and new applications of heritage materials are inevitable. Tolerances will continue to decrease, only resisted by practical and intelligent manufacturing engineers with knowledge of the limits of existing tooling and the demands of product compliance, yield and cost. The last summary item is often ignored since it is assumed that no scrap would be generated or at least it would be negligible as a cost of doing business. Manufacturing considers product defects within the spectrum of acceptance from: a) use without consequence and change or b) for waivered use “as-is” after modification or repair. A repair process, similar to designed-in self-test or design-forproduct maintainability, is a product feature that can significantly improve product quality and improve the bottom line of the business.
The 21st century will find increasing numbers of 3D photonic systems, often predicted over the past 50 years. Thanks to the pioneering work of Dr. Philip Russel and his 1991 efforts with hollow-core fiber, a rich field of materials, fabrication and simulation have developed. With the improvements of Multiphysics simulation tools it is only to be expected that foundries will provide design rules for integration and thermal management of future photonic products. Short historical case studies are offered to provide insight into challenges and unique solutions from photonic integration and manufacturing:
Case Study: Integrated Fiber E/O and O/E Switch
The introduction of fiber-optic communication systems provided significant advantages in data throughput, cost of operation and the reduction in component size and cost. Developers at AT&T Bell Laboratories were faced with the challenge of integrating the E/O and O/E transitions between fiber and the switching circuit on or about 1989. Where ceramic had always satisfied using microstrip or stripline designs, fiber data rates introduced interference, crosstalk and unacceptable loss for the emerging network. A team as described in this article was assembled with Dr. Courtland N. Robinson, Dr. David B. Powell, Robert Koehler, Dr. C.P. Wong and others. They concluded that a new composite material, Figure 2, was required to provide electrical and thermal performance along with the assembly, integration and precision demanded by alignment of the fiber cores to laser diodes or receivers. Critical material reviews resulted in a reformulation of a commercial polymer and used commercial fillers that provided the precise dielectric constant, film thickness, uniformity, process consistency and conductor interface without temperature degradation during component assembly and integration. Commercial photoresist processes and flying probe test systems were modified to work with the new material in a rapid learning process that provided field systems within two years from conception. The PolyHIC, or polymer hybrid integrated circuit, unleashed rapid growth and advancement in fiber optic networks.
Case Study: The IBM Silicon Optical Bench—Integration of Fiber and Discrete Components
A well-written paper by Barwicz, et.al. [https://doi.org/ 10.1016/j.yofte.2018.02.019] was published in 2018 and provides an excellent analysis of the challenges and heritage of photonic assembly. The cost of automation and limitations of conventional tooling are described. Silicon photonic optical bench technology is at least 40 years old, but the continuing advances presented by IBM show improvements. The advantage of a silicon optical bench is the means of providing optical waveguides and passive features in the photonic integrated circuit (PIC). Oriented crystals like silicon can be directionally etched along crystal planes to provide precise location of “V” grooves for fiber alignment. The shape and core position of the fiber can provide considerable variability so compliance and adhesives have been traditionally used to optimize optical power throughput using active alignment. Others have used rigid mounts and soldered tubes to confine fibers and control precision location of the fiber core. The advantage of the methods presented by IBM are realized in EO/OE computing applications, telecommunications, and in some sensors. The article discusses costs in a relatively optimistic fashion. Integration with power semiconductors and optical subassemblies requires further complexity based on the IBM methods, though integration of optical systems has long been achieved with variations on the materials and structures in a photonic integrated carrier.
Case Study: Automation and Photonic Assembly
Near the millennium a large variety of precision assembly and test equipment manufacturers offered specialized equipment for photonic assembly including submicron placement accuracy, precision alignment, and bonding of fibers and devices. An example of this sort of automation is found with FICONTEC [https://www.ficontec.com/]. Though other companies offer customized precision systems, a brief description is provided as an example of the complexity and capability of “off-the-shelf” automation. The FICONTEC CUSTOMLINE tools include the ability to handle components of nearly any size and align with submicron precision. Modules are available to handle fiber integration, alignment, and integrated testing. Optical subassemblies can be built with in-line UV epoxy bonding, and the company claims over 900 systems in operation worldwide. In any approach to automation it is wise to take time, carefully consider options and requirements and recognize the value and limitations of trained operators. Complex machine tools tend to have distinct personalities and many companies find out they have to assign an operator for the care and management of the tool, so elimination of labor is not always the objective. Processes depend on the consistency of the raw materials, solders, adhesives, and process control of the tool. Preparation and use of an automated system can take years of development but is often worth the investment for high-mix or high-volume manufacturing.
Summary and Conclusion
The accelerated design, development, and commitment to manufacture new photonic products is sure to offer profit and advantage to emerging corporations. The integrated product team is essential for technology transition and profitable production. A product champion, designer, process and manufacturing engineers, market/product experts, and independent reviewers will accelerate and strengthen the success of new product introduction when their experience combines at the beginning of the development cycle. Contention and problems are likely in every new product transition. At some point, manufacturing will hate development and both will dislike marketing even with the inducements of coffee mugs, t-shirts, logo pens, and the occasional vest or hoodie. The personal relationship between the team buffers people from stagnation or convoluted approaches to problem solving and provides a relationship where each can productively decompress. The functional contentions are typical and somewhat avoided by involving the right team members early and providing communication early and often. Odd as it may sound, development velocity can provide a measure of stability as long as defects and mistakes aren’t introduced. One individual can hold multiple and even changing roles during product development and transition. The key to success is a solid working team that genuinely likes to work hard, together. Good managers look for the elements of high performance teams and often build transition organizations using this approach.
The team won’t succeed without the back office systems to support operations. A disciplined enterprise resource planning system and supply chain management provide on-time performance with acceptable quality. Technologies learned in the manufacturing environment are tools for rapid advancement and new trade secrets. Techniques learned by licensing from tool or competitor efforts are suited to fast followers at noted expense. New products like stacked chip LADAR systems, sensors, integrated active/passive filters, and communication components will continue to find growing markets. As with many businesses, photonics suppliers are often relatively small and diverse so every dollar invested needs to work. This article discusses investment in staff, engineering, research, development, and back-office processes to discipline, regiment, and lead the business. The future is rich for the scientists, inventors, and manufacturers committed to this effort.
A Bit About the Author
I grew up in western Pennsylvania in an era when computers were getting started for military and government programs. Access to computer resources required punch cards, yellow ticker tapes or magnetic disks. Personal computers were a thing for the future, and the costs of calculators were similar to those of the contemporary personal computer. Scientific institutions were abundant in industry, including major laboratories in telecommunications, automotive technology, power electronics, heavy industry including mining and metallurgy, agriculture, defense, aerospace and consumer products. NASA was started with a new mission. Science and engineering were as essential for progress then as now, but corporations recognized the need for investment and sponsored graduate programs, startups and extensions of internal research with tax incentives and market intent. Product development was driven by advances that offered cost savings, improved performance or both. The reputation of the firm was built on quality, reliability and timely performance. I was awarded a bachelor’s degree in chemistry from F&M and received my masters and Doctor of Engineering Science (DESc) in materials science from Columbia University. As a former technology transition leader at AT&T Bell Laboratories as well as small to mid-sized corporations and nonprofit institutes, it is observed that each business focuses on rapidly getting products to market based on inventions and ideas. My work with government advanced technology transitions in consumer and defense segments led international programs through import/export and multiparty international collaboration. In each project there comes a time when an idea is demonstrated as a prototype, and the need to produce a product or offer services is scheduled. These transitions need: finance, reliable design, a resilient supply chain and timely market access. This brief article focused on the significant shift from development to production and considers the challenges that accompany this process.
About This Column This is a regular column that explores business aspects of technology-oriented companies and in particular, the demanding business aspects of photonics startups. The column touches on a broad range of topics such as financing, business plan, product development, program management, hiring and retention, manufacturing, quality, sales methodology and risk management. That is to say, we include all the pains and successes of living the photonics startup life. This column is written sometimes by Daniel Renner, the column editor, and sometimes by invited participants, so that we can share multiple points of view coming from the full spectrum of individuals that have something to say on this topic. At the same time, this is a conversation with you, the reader. We welcome questions, other opinions and suggestions for specific topics to be addressed in the future. Please send us your views and opinions here. The expectation for this column is to provide useful business-related information for those who intend to start, join, improve the operation, fund, acquire or sell a photonic startup, a fascinating area that can provide enormous professional reward to those engaged in it.