Small Satellites: Past, Present, and Future

Henry Helvajian and Siegfried W. Janson, editors

 

---

Chapter 24: The Generation After Next: Satellites as an Assembly of Mass-Producible Functionalized Modules

Henry Helvajian

Forerun thy peers, thy time, and let
Thy feet, millenniums hence, be set
In midst of knowledge, dream'd not yet.¹

24.1 Introduction

This chapter addresses the practicality of manufacturing satellites as an assembly of mass-producible functional units through approaches currently labeled as mass manufacturing, mass customization, and design for manufacturability. Some concepts presented are natural extensions of the personal computer industry and are actively being pursued by small satellite manufacturers; others are more in the forefront.

The purpose is to investigate manufacturing methodologies to develop a means of mass production and mass customization of satellites. Solutions are sought for the satellite mass class of 1–10 kg (nanosatellite) and by extension mass classes that approach 100 kg (microsatellites) or higher. This treatise does not present the business case for mass production, which will be critical to the viability of the venture. This lack of economic analysis is not the Achilles heel, however, because it is the specifics of the particular mission that justify the business rationale. To meet the requirements for conducting a viable business, the mass manufacturing methodologies must include processes that allow for designs that keep costs low and provide high reliability.

24.1.1 The Power and Limitations of Modularity

In a series of publications in the late 1990s, Baldwin and Clark presented the advantages of modularity in design, using the U.S. computer industry as an example.^2,3 In modular design, elements are separated into modules by a formal architecture that maps function to module. Some of the modules are "hidden," in that design changes to these modules do not affect other modules. Other types of modules must exist to make the hidden modules interoperable; these are the "visible" modules that embody design rules that must be adhered to by the hidden module designers. Modular designs have advantages over interdependent designs because the "hidden" modules can be upgraded continually without incurring the cost of a major system design change. Modularity also engenders commonality, thereby delivering additional cost savings. Finally, modularity also increases the overall system reliability or robustness with respect to maintaining operations over long durations.

To varying degrees, modularity has been applied to many industries, among them consumer electronics, automotive, and aerospace. It is not widespread, partly because of the apparent inherent trade-off between modularity and overall system technical performance.⁴ In standard mechanical systems, an uncoupled or modular system favors the interface with its commensurate losses (e.g., power/signal losses, inefficient packaging), while coupled or interdependent systems favor efficiencies in mass and force transfer to enhance technical performance.⁵ In general, it has been argued that modularity favors "business performance," while more integral architectures favor "technical performance."⁶ In satellite systems, where low mass and low power are desired and technical performance is the primary aim, it is difficult to see how modularity and its demonstrated advantages can ever be wholly adopted. And yet, the limitations of modular architecture have been mitigated somewhat, as first demonstrated in very-large-scale integrated microelectronics and then by the development of microelectromechanical systems (MEMS). In both technologies, modularity in design and manufacturing is used in the context of functional basic devices (e.g., amplifiers, adders, processors, mechanical resonators). However, the inefficiencies that commonly exist at the device interfaces are reduced by the ability to co-fabricate the devices onto a common substrate. The microelectronics industry demonstrated that efficient interfaces can be built to pass current and voltage, and the MEMS industry demonstrated that mechanical force and torque can equally be transferred efficiently across an interface, at least at microdimensions.

Another limitation is the large capital investment needed to convert a nonmodular design and manufacturing process into one that implements modularity. Notwithstanding the advantages of modularity, the cost of this conversion has been deemed too prohibitive in some industries. This would also be true for the satellite manufacturing industry. Perhaps a means to gain a foothold into the manufacturing of satellites as an assembly of modular units is through the nanosatellite concept and in the development of modules that can perform multiple functions. In this approach, the nanosatellite can serve as a modular building block for the manufacture of larger satellite systems. Furthermore, if the nanosatellite served as a multifunctional module, the desire to offer on-orbit mission capabilities for the lowest possible mass would be further realized.

24.1.2 Nanosatellites, Microsatellites, and Functional Modules

Nanosatellites are small space systems having a defined total mass between 1–10 kg. To date, most have been experimental vehicles used for validating specific technologies in space or for educating future space professionals—e.g., the CubeSat⁶ (see chapter 5). No existing operational missions use nanosatellites, but there have been experimental missions that could result in a fully operational mission. For example, the first QuakeSat was an experimental nanosatellite based on a triple-length CubeSat. Launched in 2003, it was designed to detect extremely low-frequency earthquake precursor signals from space⁷ (see chapter 5). Given this example and what has been made clear in a number of articles and ongoing developments, the size and volume of a nanosatellite should not be perceived as limiting⁸ (see chapter 23). Obviously, nanosatellites have limited propellant volume and power; but this volume limitation becomes less of a handicap with regard to packing computational processing capability. The miniaturization trend in microelectronics, along with the development of MEMS and advancements in batteries, enables significant processing capability to be embedded within a 10 × 10 × 10 cm volume. One only needs to view a current cellphone, a personal digital assistant, a camera, and the integration of these systems onto a common platform (e.g., Apple iPhone⁹) to appreciate the possibilities when a systems-level integration is applied. Consequently, there are missions that a nanosatellite can accomplish.

Perhaps those satellite pundits who do not yet perceive the nanosatellite as a viable and practical space vehicle could defer judgment, just for now, and agree to view the nanosatellite as simply a functional module with some level of control and autonomy. Current microsatellite manufacturers have used the concept of functional modules or "drawers" to assemble larger spacecraft (>10 kg).^10,11 The concept has both practical and economic appeal because each module could potentially be manufactured and tested at different locales. The modules are then integrated onto a common scaffolding at final assembly. The concept is a derivative of typical procedure in the instrumentation and electronics packaging industry, where an assortment of multichip modules are integrated into a plug-and-play module, and the integrated system delivers a higher-level function using a common backplane tray (i.e., bus) as the "scaffolding."

Modularity is sure to be accelerated by the recent development and standardization of SpaceWire,¹² which enables high-speed serial communication between connected modules. Still, the common backplane analogy fails to consider the true functional aspects of a satellite. In a computer, the common backplane is only an interface for supply voltages and data lines; by contrast, a space system would consist of an assembly of modules that would require common backplanes for voltage and data lines, radio frequency (RF), fluidics, and thermal control. The problem is further exacerbated by the need for a low-current and a high-current backplane. The solution implemented by most microsatellite manufacturers is to separate the functions (i.e., the backplanes) into individual modules. Each is self-contained and uses only power and data lines from a common satellite bus. In essence, the subsystems are modular and individually tested. There are practical outcomes of this satellite manufacturing approach. These are the decrease in the development time and an increase in overall system reliability; both of these benchmarks have been realized.¹³

Even though a modular subsystem approach is viable, the question arises—using Lego blocks as an analogy—whether there is a need to develop a multifunctional block or module that has elements of multiple subsystems, thereby enabling the assembly of small and large satellites alike. Of course, unlike the Lego block—which is a passive module—the envisioned satellite block would also generate and store power locally for both internal use and external distribution. The modular subsystem may even have propulsion for limited locomotion or for physical attachment to nearby units. It has a primary function but also several secondary functions. Two good examples are the CONRO project and the SuperBot from the University of Southern California's Polymorphic Robotics Laboratory.¹⁴ The SuperBot consists of an assembly of modules, each self-contained and self-sufficient (power, intelligence, control, locomotion) but designed to work together. The advantage of developing such blocks for space applications is that because they are nearly identical, they can be mass produced, and because they can be joined to form larger structures, they enable the assembly of space systems in very complex shapes.

24.1.3 Overview of the Satellite Manufacturing Industry

When a satellite is a manageable size, it becomes feasible to establish a manufacturing process flow whereby the satellite is physically moved through manufacturing stations. This process is an assembly line, though it need not be automated. Assembly line production has increasingly been used for the satellites intended for low Earth orbit (LEO) and medium Earth orbit (MEO; see chapter 14). Modern satellites also contain an increasing number of commercial-off-the-shelf (COTS) components that are put into assemblies, which are typically manufactured by subcontractors. Quality control has shifted from testing all components and systems to testing sample units, though components and systems are still tested when possible. The implementation of these process improvements has reduced satellite delivery times to less than 18 months—a major improvement over the three-year delivery times typically quoted just five years ago. Modern satellites are also expected to do more. There is a drive toward more onboard data processing, increased use of reconfigurable antennas that involve both electronics and robotics, and higher communications bandwidth—for example, very high Ka-band (18–40 GHz) and V-band frequencies (50–75 GHz) or laser communications.

These technological advances drive the need for more satellite power. The objective has been to provide more power without increasing the overall mass. Harnessing power in space—at least as solar energy—is ultimately related to capture area, while power storage is related to volume. Large satellites that serve complex and power-demanding missions have continued to grow in mass.¹⁵ The same fate need not fall on the nanosatellites and microsatellites. Perhaps the satellite shape could be optimized to maximize solar-cell area. Cubic and rhomboid shapes do not capture energy efficiently, but are derivative of the shape of the internally modularized units. Satellites that are designed to get by with only a fraction of the available solar power do have the advantage of not requiring an attitude pointing system. However, the complexity associated with an attitude pointing systems need not be considered a limitation. Small satellite technology has evolved, and sun-pointing attitude control systems can now be included.

In comparison to many terrestrial markets, the space industry is a niche market even though it generates $100 billion in revenue worldwide.¹⁶ Under construction are 100 satellites having an estimated value of $11 billion; another 175–200 new satellites will be required by 2010 to replace the current satellites or to augment service.¹⁷ As transponder requirements grow (estimated requirement is 11,129 transponders by 2010 or more than a 78% increase from 2002), either more geosynchronous (GEO) satellites will be needed or even larger space systems will have to be built. Most of these transponders are designated to support HDTV, but other applications—such as store-and-forward communications or vehicle and container tracking—are being shifted to smaller LEO satellites, which have advantages over GEO satellites. For example, because LEO satellites are 25–40 dB closer, Earth stations can use omnidirectional handheld antennas. However, the primary limitation of LEO satellites is their intermittent service. This limitation can be surmounted by the deployment of large constellations. Several recent commercial attempts to provide nearly continuous coverage via constellations have been less than successful (e.g., Teledesic, Celestri, Ellipso, Globalstar, Odyssey, Iridium), and these failures have rippled through the industry.

Given that existing satellites are lasting longer than their predecessors and the overall demand forecast of nearly 200 satellites over the years 2001–2010 is nearly half the build capacity for the U.S. manufacturers alone, why would the industry consider a change in the manner by which satellites are manufactured?¹⁸ The answer lies in the fact that the models used to make these forecasts do not adequately describe the present state. Information has become a resource, and the value of a deliverable increases with the information content. Satellites working in concert with localized ground sensors can provide far more information worldwide and with more immediacy than is currently possible using any other scheme. Concentrating on celestial rather than terrestrial applications, small satellites in large numbers can also provide a quicker overall view of our solar system and of Earth's space environment than is possible with the current one-vehicle-one-mission approach. The economist Paul Hawken¹⁹ considers information synonymous with traditional mass-based resources and in a concise way has defined the value-of-a-deliverable as the ratio of the information-content it can hold to its physical mass.

The GPS constellation has proved that information content sells. For example, analyzing the total GPS market in North America, the aviation segment grew around 10%, the military and timing markets grew just under 25%, while the land (including marine) market grew 35%.²⁰ In the revenue scale, the land market makes up 62% of the total revenues.²⁰ Those less-than-successful commercial satellite constellation providers knew the value of information content; they failed to implement their vision because their approach to manufacturing was not sufficiently cost-effective.

Bringing revolutionary innovation to the satellite industry is inherently difficult because the financial barrier to entry is high. Using the space imaging market as an example, the entry cost estimates range from $97 million to $497 million.²¹ This includes the satellite, ground segment, launch, and insurance costs. Furthermore, because the industry is heavily regulated by government, it becomes somewhat immune to market forces. The issue should not seem hopeless; indeed, such government involvement may be the saving grace, as governmental sponsorship may be required to significantly alter the existing manufacturing approach.

Government sponsors should be interested in mass manufacturing of satellites for two reasons. First, by implementing mass manufacturing concepts, nonrecurring costs should be reduced. Second, once the infrastructure for mass manufacturing has been established, the cost to design and manufacture a new satellite should also be reduced. These reasons are mere representations of the economies of scale. Two more factors should be of interest to a government sponsor. First, with mass manufacturing, proven quality control methodologies can increase product reliability without incurring additional cost. Examples are easy to find (history of laptop computers, for instance) where a reduction in unit cost also yielded an increase in unit reliability. The key is to reach unit commodity numbers where economies of scale can be a factor and established quality control methodologies can be applied. The satellite industry, as it is constituted today, cannot hope to achieve such numbers if the commodity unit is defined as the complete satellite. However, if the commodity unit is defined as the module or block by which complete satellite systems can be assembled, then economies of scale can be achieved. A second factor that should interest government sponsors is that with the modular approach to satellite manufacturing, customization, adaptability, upgrading, and replacements are easier to implement with the natural consequence that there is "graceful" degradation of a system as opposed to abrupt and catastrophic loss.

The problem is not only with the satellite manufacturing industry but also with the launch industry. The cost of launch to orbit, after 50 years, still hovers around $15,000 per kilogram to polar orbit, and this cost is only realized by using a modest, not small, launcher (e.g., the U.S. Delta II-7920, ~3300 kg to polar orbit).²²

In summary: Because of the large cost of manufacturing, launch, and operations, high reliability has become a paramount concern to enable a reasonable return on investment. Brooding over this issue eventually leads to complete risk aversion. This change in perspective has an ironic component in regard to innovation: As a venture becomes more efficient, it also tends toward more rigidity and can evolve toward a "frozen state" in which substantive innovations are not easily made.²³ Studies show that frozen-state industries are more likely to allow innovative approaches in good economic times but quickly shift toward efficiency mode when the economic outlook is weak or the return on the investment becomes precarious. To mitigate this dilemma and allow innovation back into the aerospace industry, several problems must be tackled simultaneously, and governments have to be involved. These include

• The cost per kilogram to orbit must be reduced by a factor of 2 or 3.
• Smaller launchers must be developed to allow lower total cost to orbit.
• Satellites must be designed as an assembly of mass-producible units.
• Very large space systems must be assembled in space using robotics.
• Modern design methodologies, such as design for manufacturability and concurrent engineering, must be promulgated throughout a space systems design and manufacturing phase.
• Risk must be managed rather than eliminated.
• The total overall system cost must be minimized, rather than the cost of each independent segment (i.e., satellite manufacturing, launch, support, end-of-life operations).

This chapter delineates the required basic criteria and the type of manufacturing technology that must be utilized to economically manufacture a multifunctional module or block.

24.2 Mass Manufacture, Mass Customization, and Build-to-Order

David Anderson, author of Design for Manufacturability & Concurrent Engineering, defines design for manufacturability as "the process of proactively developing products by optimizing all the manufacturing functions" and concurrent engineering as "a proactive practice of designing products to be built on standard processes or concurrently developing new processes while developing the product."²⁴

Both design for manufacturability and concurrent engineering are procedures that need to be implemented prior to and during the design and manufacturing cycle. Furthermore, these procedures become more significant with the sobering fact that 80% of the total system cost (i.e., design, manufacture, operations, service, and repair) of a typical product is committed by the end of the product design phase, before a single deliverable unit has yet to be manufactured. Even more germane is Anderson's conclusion that nearly 60% of a product's cumulative lifetime cost is already committed at the concept/architecture phase, before design.²⁴ Therefore, there should be a strong impetus to get a design right before proceeding to the manufacturing phase. By the time manufacturing has commenced, 95% of the total cost has been committed. Any design changes at this stage is associated with a high-penalty cost.

One approach to ensuring the right design and system architecture is to bring to the table at the outset all the available knowledge, ideas, issues, and concerns, as well as resource allocation. In traditional approaches, the number of people involved is initially very small and continues to grow as the product development proceeds through design, prototype, and release. Various experts are brought in at various stages to deal with the design modifications, customer issues, and, after release, problem solving and servicing. An alternative but counterintuitive approach is to have a very large team at the outset that represents all segments of the product development phases, including the customer, vendors, and marketing. This large team is involved in the architectural design phase, and as the design progresses, is reduced until a core group remains that is responsible for the product release. The first approach initially appears to be more economical but is actually found to be more expensive when viewed from the perspective of the total system cost because changes made as a consequence of expert opinion cost more when applied at the later development stages. The second approach initially appears to be costly but actually provides a more reliable product with an overall reduction in system cost.

24.2.1 Methods of Manufacturing: an Overview

The methods of manufacturing changed in the 20^th century to reduce total system cost. Mass production (i.e., the assembly line), popularized by Henry Ford and the Model T automobile, became the focus. Mass production is capital intensive, with a high proportion of machinery to workers. One characteristic, however, is product uniformity. A key element is that resource inventory must be maintained to sustain the production-line flow at demand levels: Ford wrote that for efficient and cost-effective operations, inventory should be carefully monitored to be just enough for immediate needs.²⁵

In the 1950s, Toyota Corporation explored and then instituted a variant of this method known as "just-in-time" or "kanban" manufacturing. It is an inventory flow process that uses a series of signals that identify key transition points in the production process. These signals ensure that resources are controlled to maintain a balance of no overstock or understock. A subsequent refinement is now called Lean Manufacturing;²⁶ it is associated with the moniker Six-Sigma and includes a series of practices and techniques that were developed by Bill Smith of Motorola²⁷ in the 1980s, but were embraced and applied with success by Jack Welch at General Electric. One key element in Six-Sigma is the continuous effort to reduce process variation by rigorous measurement and the systematic removal of manufacturing defects. The industries where these manufacturing methods have been successful are those that deal with commodities, where the unit production quantities are on the order of thousands to millions. It has not fared as well in industries with a high level of human interaction (i.e., retail), where the production quantities are lower and where some level of customization is considered a feature.

Various other manufacturing methods have also been developed to contain cost and increase quality. One common term is agile manufacturing, which refers to an organization that has established flexible processes that enable it to respond quickly to customer needs and market changes. For example, a machine or tool would be adapted to produce a range of parts or conduct a series of manufacturing processes based on a common platform. Another example is the utilization of several machines that can be altered to perform the same operation and thereby absorb rapid increases in demand.

Rapid Manufacturing is a derivative whereby the sequential delivery of either energy or material to specific places enables the fabrication of a part. The technique is usually ascribed to material-additive fabrication (e.g., inkjet printing of "parts") but can also be utilized, in certain cases, to a material-subtractive process. The key element that differentiates Rapid Manufacturing from other approaches is that it is controlled by a computer and commonly utilizes a mathematical model of the relevant piece part, which is created by another computer. The technique falls under the general heading of CAD/CAM (computer-assisted design and computer-assisted manufacturing). Rapid Manufacturing is inherently a serial process, but if essential elements can be done in parallel (i.e., batch production) it provides a large advantage in speed and cost while offering customization.

One consequence of Rapid Manufacturing is the ease of product customization. Under certain circumstances, flexible manufacturing systems can offer an approach to mass customization, defined as "the technologies and systems to deliver goods and services that meet individual customer's needs with near mass production efficiency."²⁸ Mass customization should not be confused with personalization. Whereas personalization, with its desire to fulfill the customer specifications down to the letter, is an intensive customer-provider process that can work outside the solution space, mass customization works within a fixed and predefined solution space of stable processes to offer a range of customizations. Mass customization has been successfully applied in numerous industries. Examples are in apparel (e.g. Land's End clothing) and in consumer electronics (Dell Computers).

Satellite manufacturing has traditionally adhered to the approach known as build-to-order. Products are only built after a confirmed order is received. It's the oldest style of production and is considered the most appropriate for highly customized or low-volume products. The advantage of build-to-order is the ability to manufacture a product to exact customer specifications and therefore achieve personalization. The disadvantage is that fluctuations in demand can lead to underutilization of manufacturing capacity.

Most governments view space systems manufacturing as a strategic asset, and they have a strong motivation to maintain and support manufacturing capabilities in times of reduced demand. At this point, market forces no longer apply, and manufacturing methods that should be altered are given extended life. The practice leads to an overall added cost to systems procurement, but the additional cost only appears on the government's balance sheet. Measuring quality and reliability may also be more costly. Build-to-order manufacturing personalizes a product; therefore, because there may not be sufficient unit volumes, statistical methods to measure quality may not be possible. Consequently, the manufacturer has to initiate quality control by costly testing of individual components.

24.2.2 Insights Regarding New Manufacturing Approaches for Space Systems Development

To what extent can these management philosophies, manufacturing systems, practices, and techniques be applied to the satellite manufacturing industry, which serves both governments (with emphasis on reliability over innovation) and civilians (with emphasis on innovation over reliability)? Several insights may provide some clues:

• Product uniformity increases reliability.
• High reliability and cost-effective production requires a concerted effort to reduce process variation in manufacturing outputs.

A new approach to the manufacturing of space systems should strive for product uniformity. But how is this to be done in a statistically significant manner when production volumes are small? Furthermore, one wonders how much of satellite manufacturing today still involves component modifications or local "fixes" throughout the manufacturing phases to realize a customer specification. These local fixes or changes normally require innovative solutions, but these solutions may not have the maturity to be labeled as a stable process for reliability verification.

• Manufacturers and government sponsors are more apt to pay for innovative solutions, but are less likely to support process development efforts that transform the innovative solution into a stable manufacturing process.
• If increasing reliability and reducing costs become ends in themselves, they can become counterproductive to innovation.²⁹

It appears that manufacturing can survive on a series of innovative solutions that then make up a patchwork frame of operations instead of defining and developing a viable solution space of stable operations. However, manufacturing via a series of innovative solutions can run counter to processes that are designed to increase reliability and reduce cost. Therefore, innovation must be separated from reliability and cost reduction. Reliability and cost-reduction must be applied only after an innovative solution is converted by investigative characterization into a stable process.

• One must first be able to define and measure a variable in a manufacturing process before it can be controlled.
• Human interaction reduces reliability, whereas computer control increases it.
• Innovation is necessary for product development and enhancement, but it must be documented, and the solution must be represented by the commensurate development of a proven process.

An all-robotic manufacturing line controlled by a network of computers with established processes would progressively lead to better product reliability. But innovation, which is a distinctly human process that enables revolutionary changes, cannot haphazardly be inserted into the manufacturing process. Innovations must either be inserted at the beginning (in the design/architecture development phase) or if it is a concurrent action, only after a reliable, stable process based on that innovation has been realized and configured for computer control.

• Product uniformity enables mass production and a stable market.
• Defining a solution space of stable processes with a limited number of permutations of functionality enables mass customization.
• Mass customization engenders customer loyalty because customers do not pay the full premium for customization.

Given that a major aspect of the satellite industry is driven by customer specification, it is more likely that customization is the norm rather than the instance. The trend is to standardize a satellite bus and limit customization to payloads that can be mounted on the bus. This approach will not realize the production efficiency values commonly associated with mass production because the unit production volumes of the standardized systems are still small. An alternative approach might be to explore a solution whereby the standardized unit is a multifunctional module (MFM), and satellites are developed as an assembly of modules. The actual number of modules would depend on the satellite size and mission. If this conceptual framework were to be made a reality, then defining the manufacturing solution space would entail deriving the stable processes that enable the reliable development of the MFM. Because the MFM would be the common elemental unit in all satellites, a large number of them would have to be manufactured. Mass manufacturing efficiencies would result from the large production volumes. For example, a distributed sparse-aperture array would need hundreds to thousands of units, and a large satellite may need 10,000 units. Therefore, a yearly production quantity close to 100,000 is conceivable. Mass customization would be determined by how many MFMs could be conjoined and the ensuing functional permutations that could be derived. Given these large production quantities, the quality level per unit could be enhanced.

• To increase the quality level of a manufacturing line, piece-part count in a product must be minimized.
• The design philosophy must encompass the total system.

Product quality can be determined by the part quality raised to the exponent of the number of parts in the product. For example, if an MFM is composed of an assembly of the same components, then the product quality value can mathematically be represented as Q_s = (Q_c)ⁿ, where Q_s represents the quality of the total system, Q_c represents the quality of the component and n is the number of components. Assume the quality value for a component, Q_c, is 0.993 (i.e., 99.3% of units pass) and a 100 such components form a system, then the overall system quality factor is ~0.50 (i.e., 50% of system units pass). The obvious conclusion is that to establish a manufacturing process with a desired quality value, the quality values of the fabrication techniques must inherently be much higher.

The goal is then to reduce the number of processing steps that can come about by reducing the materials list, integrating subsystems onto a common platform, or designing subsystems in a manner that permits multiple functions to be derived from the same unit. The first notion realizes an advantage in space applications, where a reduction of the materials list should also reduce problems resulting from differences in thermal expansion. The second notion has the basic advantage of a lower number of interfaces. In the extreme level of integration, the entirety of the device components and functionalities are somehow integrated on a common substrate. In the microelectronics lexicon, it is called wafer-scale integration. Therefore, a substrate material that offers some multiplicity in the types of functions it can support (e.g., structural, electronic, and photonic) would lend itself to the development of a more reliable system because more integration would be possible. Therefore, the choice of the base substrate material and the availability and precision of the processing tools for that material class should be a key element in designing the MFM.

• A serial fabrication or manufacturing process is not as economical as a batch process.
• However, a serial manufacturing process does add flexibility (in essence, customization can be added at a unit level).
• One approach to gaining an agile manufacturing capability economically is to integrate a batch fabrication process into the serial manufacturing process. In this case, customization is applied during the serial process, but it is during the batch process where the effect of the customization is realized.

The microelectronics fabrication industry has proved the value and power of combining lithography and batch chemical etching. During the lithography step, only the pattern information is transferred. Material removal or modification is actually accomplished by chemical etching, ion implantation, thermal processing, and so on, which operates in batch mode. The combination permits some flexibility in the manufacturing (i.e., different patterns, masks). The big advantage is the cost savings in that multiple devices can be manufactured at once with very high reliability.

• The Internet presents a continuous web of connectivity.
• Software "travels" and can engage multiple stations simultaneously worldwide.
• Manufacturing facilities and controllers need not be at the same location.

Computers and the Internet have profoundly changed how business will be done.³⁰ Instead of a manufacturing center where all the tools are collocated, distributed minifactories or processing centers can exist in widely different locations and can conduct near-synchronous operations to provide a virtual manufacturing system. An example is the Modular Integrated Robotized System developed by Pirelli Tires Corporation.³¹ A more subtle variation is the MEMS Exchange,³² a service consisting of hundreds of foundry processes that are physically distributed throughout the United States. In cases where multiple processes must be done to a piece part, the part is shipped from location to location to accommodate the desired processes. It is not inconceivable that MFMs could be manufactured in a similar manner, either using the distributed factory or distributed foundry concepts. The advantage to having distributed operations is the reduced startup costs and the possible utilization of excess capacity in manufacturing foundries. The disadvantage is that the manufacturing solution space representing stable processes must also include reliable processes that seal or encapsulate units for transfer or shipping.

24.3 The MFM and Microelectronics Fabrication, Glass Ceramics, 3-D Patterning

MFM is envisioned as having functions that are implemented by the addition or integration of electronics and photonics (e.g., intelligence, communications), a primary-function device that is aided by a sensor or actuator, and a secondary-function device that can be placed into action in the event of failure of a neighboring MFM. The envisioned MFM also includes ancillary subsystems such as propulsion, energy capture and storage, and attitude control. The energy needed to conduct its primary and secondary functions would be generated by photoconversion and stored in chemical batteries. The MFM has nearly all the attributes of a current satellite, but the extent of its functional capabilities is limited. Units are not designed to operate alone but in concert with others. They must therefore be mass-producible. When they are physically connected, information and power can flow among them as necessary. In more advanced versions, fuel or propellant can also be routed to specific locations. In the simplest example, the MFM is a block that serves as scaffolding with drawers to hold instrument trays. The concept of trays and drawers is already used in the microsatellite industry, and the concept of small blocks, each with space system functionality, is suggestive of CubeSats and picosats. Even though these approaches can result in the building of reliable space systems in modular fashion, it is not what we mean when we speak of a mass producible MFM.

As envisioned for the generation-after-next spacecraft, the MFM is intended to be an active element working in concert with a large number of MFMs. In various sizes and configurations, the ensembles provide different space system capabilities. The metamorphic robot technology described above can provide a viable precedent toward this distributed space systems vision. Within a SuperBot, for example, each module is "self-sufficient and is autonomous with regard to the use of its resources. For example the use of its sensors and actuators."¹⁴ Consequently, a group of these units becomes reconfigurable, and there have been demonstrations of various walking, sliding, and rolling "gaits" implemented by autonomous SuperBots. A recent interesting technological development is the application of a biologically inspired method to establish control and induce self-organization among a module swarm.³³ In this development, a digital algorithm mimics a simple hormone response found in most biological systems, and this is used to control the tasking and execution of program steps among a robot module swarm. The consequence is a significant simplification in the control system and a reduction in the communication bandwidth necessary to conduct an operation.

To enable the development of a reconfigurable space system, it may be useful to explore the manufacturing aspects of an MFM that is first and foremost designed for manufacturability and, second, exists as an integrated system in which the structural material can be used for additional functions beyond that of structural support. The manufacturing and fabrication tools for the MFM must have precision and tolerance over a wide dimensional range because the unit will have macroscopic (1–10 cm), mesoscopic (0.1–1 cm), and microscopic (0.001–0.1 cm) features.

What characteristics should be included in a late-generation MFM? It must have a series of standard attributes and a few series of functional attributes. Some examples of these attributes are given below.

Standard Attributes

• Can be passive or active but can be bypassed.
• Has the innate ability to physically connect with other units.
• Has local intelligence to govern health.
• Can be governed by a master controller (i.e., subordinated to master/slave relationship).
• Can serve as an information router.
• Can serve as a power router.
• Can serve as fluidic router.
• Has sensors to judge local awareness.
• Can share resources with nearby neighbors.
• Can generate some power locally (sun/solar cells).
• When nearing the end of its life, it transfers local information to neighbors, reduces its connections, and decouples from neighbors to enable replacement.

Functional Attributes

• Has particular main task (i.e., primary "payload") that can be performed either as a solitary unit or via concerted effort with others.
• Majority of its resources are focused on main task.

At the conceptual level, the manufacturing approach for the envisioned MFM must at least draw insights from modern manufacturing. Recapitulating, it is clear that product uniformity increases reliability, which dictates a form of mass production. However, it is customization that engenders customer loyalty, and because customization is more akin to the requirements of the satellite industry, mass customization may in fact be the viable manufacturing model. The overall product quality can be increased by either reducing the number of piece parts or fabrication processes. Alternatively, product quality can also be increased through systems integration, or, in the extreme, fabricating integrated components and devices on a common substrate. For a given stable manufacturing process, automation increases reliability, given that the fabrication techniques that are applied have high quality factors. Finally, the microelectronics industry shows that lithography, patterning, and chemical etching can surpass traditional machining approaches both in terms of fabrication quality and cost. Therefore, in laying out the conceptual framework of an MFM manufacturing line, microelectronics manufacturing is a good place to start.

24.3.1 Microelectronics, Metals, and Plastic Fabrication Technologies

Microelectronics fabrication is based primarily on silicon and silicon-dioxide materials. Even though the technology is mature for fabricating electronics and reasonably mature in fabricating actuators, it is not clear that silicon is ideal as a substrate structural material for the envisioned MFMs. First, the MFMs will mostly be used in Earth orbit—and given that the estimated mass of an MFM is between 1–10 kg, large temperature fluctuations is a concern, at least for LEO missions. There are, of course, approaches for encapsulating temperature-sensitive components, and these are currently employed in CubeSats and other small space systems. Still, a conceptual analysis should consider the major effects of the space environment on the chosen structural substrate material. Bulk silicon has high thermal conductivity (149 W/mK),³⁴ which could exacerbate the rate of cooling and heating in LEO missions. Furthermore, many of the useful physical properties in bulk silicon that have been applied for making sensors (e.g., Hall effect, Seebeck effect, and piezoresistance) will have substantial temperature sensitivity. These temperature-dependence effects are typically compensated via engineering the design (e.g., Wheatstone bridge circuits). This is clearly possible if necessary. But is it necessary?

From the perspective of fabrication of structures and systems integration, there are other features of bulk silicon that make it less attractive as a substrate for the envisioned MFM.

• Lithographic patterning in bulk silicon is based on applying a patterned protective layer (e.g., photoresist, oxide, nitride). It would be preferable if the material itself were photosensitive (i.e., photo patternable), and, therefore, there would be no need for applying a protective layer photoresist application step.
• The principal dimension of the MFM is going to be related to the availability of bulk crystalline silicon boule sizes (currently ~300 mm diameter). Consequently, the MFM shape is governed to some extent by the maximum diameter of a boule.
• Because bulk silicon is crystalline by nature, the ability to alter macroscopic shapes smoothly by molding is constrained.
• Bulk silicon processing is based on 2-D mask lithography; therefore, the structures that can nominally be fashioned by chemical etching (e.g., cavities, tunnels, beams) appear as extruded shapes of the lithography mask. It would be preferable to have a material where full 3-D machining or excavating and structuring is possible.
• To fabricate deep undercut structures in bulk silicon, chemical etching anisotropy among the crystalline planes is used.^35,36 For example a <100> cut wafer and a pattern mask can undercut a structure leaving the <111> plane side walls at 54.7 deg relative to the wafer surface.³⁵ These inward sloping walls produce a lot of unused real estate on the surface. It would be more useful if undercut structures could be fabricated without needing wafers with specific crystalline cuts or needing to carefully align masks to the crystallographic axis.
• To fabricate deep wells with near vertical walls by chemical etching in bulk silicon, <110> wafers have to be used. An alternative to chemical etching is to use a deep reactive ion etching (DRIE) tool (1–5 µm/min rates). It would be preferable to rely on standard chemical etching since it is the least costly approach.
• There is an advantage to the integration of systems if the substrate material for the MFM is amenable to implementation of a photonic bus. Silicon becomes optically transparent in the infrared (1–2 µm wavelengths), and therefore, the insertion of a visible to near-infrared photonic bus becomes less practical.

Notwithstanding these issues, bulk silicon does have advantages as a structural material for manufacturing MFMs. These are the maturity of the technology with regard to quality and reliability and the level of detail in material and process characterization that has been conducted worldwide. Furthermore, the transitioning of a microelectronics foundry to an MFM manufacturing line can be done economically if the approach uses outmoded or prior-generation silicon processing tools.

Aside from silicon, other candidate structural materials are metals, plastics, and glass-ceramics, and in each of these material classes, established tooling and manufacturing technology exists. Of these three classes, the metals are the least viable from the perspective of integrating systems on the "wafer-scale." Outside of the desirable mechanical properties (high strength, toughness, noncatastrophic failure, high degree of plasticity) and its ability to withstand space radiation, metals integrated with other materials (glasses, semiconductors, plastics) to produce multifunctional properties is still an area of active basic research. Therefore, an MFM design using metal as the base structural material will likely yield a surface or scaffolding upon which other units are physically attached, which is very close to the current methods used in satellite production. However, for the MFM, there will be a need for high-precision 3-D processing. In this regard, the metals have an advantage over silicon materials. It is now possible to fabricate precise 3-D structures in metal via a number of material additive processes (e.g., EFAB,³⁷ selective laser sintering,³⁸ microstereolithography³⁹) and subtractive processes (EDM⁴⁰). Another approach is via LIGA (from the German for lithography, electroforming, and molding). Using LIGA and injection molding technology, high-precision metal components can be manufactured, but this manufacturing process is not amenable to mass production because the fabrication and assembly processes outside of the LIGA step are strictly serial and expensive. Furthermore, LIGA imparts the added cost of the x-ray lithography.

As a material class, plastics and composites have made significant inroads in the past few decades, often replacing metal. High-resolution injection molding and casting, extrusion, and laser direct-write polymerization have allowed cost-effective fabrication of precision plastic components in 3-D. If cost were the only driver, fabrication with plastics would be hard to beat. Plastics also offer the significant advantage of chemical engineering to modify the material properties and make them suitable for the application. Dense polymers have been developed for armaments (e.g., the Glock polymer-frame gun), and conducting polymers have also been developed⁴¹ along with light-guiding polymers for photonic applications.^43,44

However, for space applications, the current formulations of plastics have drawbacks when considered as a structural material. That is not to say that a space-qualified plastic cannot be developed; but the general properties of plastics are that over long periods, they are not dimensionally stable with the temperature fluctuations encountered in LEO missions. The coefficient of thermal expansion for common non-reinforced plastics is ~25–50 times that of silicon, ~8–20 times that of metal, and ~6–16 times that of glass. For application as a structural material, dimensional stability is an important factor, as all the subsystems will be connected to the base material. Second, in comparison with metals, glass, and silicon, the plastics have a lower temperature of operation. Consequently, plastics would not be the material of choice in high-temperature areas (e.g., combustion chambers and propulsion nozzles). Third, plastics have the potential for outgassing in vacuum, which raises the issue of contamination of nearby optical structures (although this problem can be mitigated by design and by reinforcing the plastic with additives). Fourth, plastics have low yield-strength values: ~2% that of silicon and some metals (e.g., tungsten), and ~13% that of quartz.⁴⁴ This issue would be a factor when MFMs are physically interconnected and the compounding forces generated by neighboring attached units are exerted. Finally, plastics having carbon-carbon bonds tend to show degradation over time, especially in LEO (i.e., in the vicinity of the 7 eV atomic oxygen) and in sun illumination (i.e., UV photochemistry). Historically, the major barrier to space application of plastics has been the poor properties as a radiation barrier for electronics packaging.⁴⁵ However, with additives it is possible to tailor the properties of plastics over a wide range. For example by adding fluorocarbon fibers or glass fibers and whiskers, dimensional stability and material-creep issues can be minimized, and the material can be strengthened and toughened.

The polymer material class is nominally processed either by conventional machining, molding, casting, and extrusion. However, in the past 20 years, laser stereolithography, a technique by which polymeric fluids harden under light excitation, has matured and now offers true 3-D fabrication of polymer structures with near micron resolution.⁴⁶ Laser stereolithography, albeit with its serial fabrication process and its limitation in manufacturing throughput, still offers advantages from the manufacturing perspective; the key is to develop a polymeric material that has ultimate properties useful in space applications.

24.3.2 Glass Ceramics: a New "Old" Material for Space

Glass ceramics are a unique class of materials that combine the special properties of sintered ceramics with the characteristics of glass.⁴⁷ They are manufactured in the amorphous homogeneous glass state and can be transformed to a composite material via heat treatment. The treatment provides the controlled nucleation and crystallization of the constituents. Since their invention nearly 60 years ago, glass ceramics have been used in a wide range of scientific, industrial, and commercial applications. They are particularly well suited for aerospace engineering, biotechnology, and photonics due to their attractive chemical and physical properties (e.g., no porosity, optical transparency, high-temperature stability, limited shrinkage, corrosion resistance, and biocompatibility).⁴⁸ A key advantage is that the constituents that form the glass can be varied, altering the chemical and physical properties of the material. The manufacturing technology is sufficiently mature to allow the variation of select physical properties that can be derived from the controlled growth of crystalline matter within an amorphous glass. These include modifications to the material strength, density, thermal conductivity, maximum temperature of operation, electrical insulation and RF transmission, coloring, transparency in the optical wavelengths, and susceptibility to chemical etching. Consequently, the material has found wide applications. Furthermore, many new exciting applications of glass-ceramic systems are yet possible with the likelihood of incorporating "intelligence" in the form of micro- and opto-electronics.

Glass ceramics can be classified as a phase-change material, and it has been the focus of worldwide research—for example, as insulating materials in the development of optical switches⁴⁹ and metallic films in the development of shape-memory alloys.⁴⁴ Glass ceramics have more in common with the so-called ceramic-matrix composites than with monolithic ceramics (e.g., silicon nitride, silicon carbide). Ceramic-matrix composites include continuous or discontinuous reinforcing materials that reduce the inherent brittleness typical of monolithic ceramics. In the specific class of glass ceramics, the strength reinforcement is derived from phase transformations that occur during thermal treatment. However, these reinforcements affect the physical property globally. In a global transformative process, where everything changes everywhere and to the same degree, the ability to integrate various devices or structures on the same substrate is limited. This is because each device or structure will derive its functionality from the local physical property of the substrate, and there can be little variation of the properties in a globally transformed system. Consequently, the usefulness of glass ceramic as a substrate becomes limiting for the envisioned MFM, where one may want to simultaneously integrate various functionalities (e.g., enhanced IR optical transmittance, reduced RF transmittance, increased electrical conductivity) on the same substrate.

24.3.3 Photosensitive Glass Ceramics: Glass Ceramics that can be Microstructured for Increased Functionality

There is a subclass of the traditional glass-ceramic family known as the photostructurable or photosensitive glass ceramics (PSGC), in which the material phase transformation may not necessarily be global but can be locally controlled.^50,51 Consequently, the issues that limit glass ceramics with regard to integration of multiple functions are somewhat mitigated. The PSGCs differ from glass ceramics by the addition of photosensitive compounds to the base matrix. The physical phase transformation still takes effect during the thermal treatment step, but only in the exposed regions. The material operates much like photoresist except that the patterned material can be controllably transformed into different material phases, each with unique mechanical, thermal, and chemical properties. In one common PSGC composition, the possible phases are a low-temperature (~600°C) semiceramic phase that is soluble in dilute hydrofluoric acid, a high-temperature (~800°C) compatible semiceramic phase that is not soluble in hydrofluoric acid, and an amorphous glass state that is optically transparent and only slightly soluble in hydrofluoric acid.

The three primary processing steps are depicted in Fig. 24.1. Step 1 is the exposure process—fundamentally, a photo-ionization event that generates a trapped electron in the glass. The exposure step has far more versatility when conducted with a pulsed laser in a direct-write mode, as opposed to traditional lithography using a continuous light source. With a pulsed laser, the incident laser light can be focused to a particular spot, and the amount of irradiance within the focal volume can be maintained just above that required for state transformation.⁵² This allows embedded patterns to be made, thereby inducing embedded property changes.⁵³ Step 2 entails a thermal treatment. It begins with the chemical reduction of a constituent metallic ion species by the photo-induced trapped electron. The result is neutral metal atoms. This chemical reduction process that constitutes the "fixing" of the exposure pattern is the fundamental difference between the photostructurable and regular glass ceramics. Subsequently and during the thermal treatment stage, the enhanced mobility of the neutral metal atom induces the formation of metal clusters and then metal nanoparticles. When the particle size has reached ~8 nm for some common PSGC compositions, it can serve as a nucleating site for the growth of a crystalline phase. Step 3 in the figure depicts the third processing stage, and the processing particulars of this stage strongly depend on the thermal treatment protocol chosen in step 2. The equation for the chemical reduction is also depicted in step 3 for one common PSGC composition. If the maximum temperature during the thermal treatment stage reaches higher than 750°C, the resulting patterned material is transformed to a high-temperature glass-ceramic phase. This material is described as an opal—not optically transparent and not chemically soluble in hydrofluoric acid. However, if the maximum temperature in step 2 is less than 750°C or more commonly at ~600°C, then a low-temperature glass-ceramic phase grows in the exposed regions. This material is not optically transparent but is chemically soluble in hydrofluoric acid, with an etch contrast that approaches 50:1 with respect to an unexposed area. However, if the maximum temperature reached in step 2 is ~500°C, then the exposed regions contain trapped metallic nanoparticles that show optical absorption in the UV (< ~300 nm), and under certain exposure conditions it is possible to imbue the patterned area with color.⁵⁰

Fig. 24.1. Primary processing steps for PSGC materials. Step 1 describes the first stage and the exposure process. Step 2 entails a thermal treatment. In step 3 the processing particulars strongly depend on the thermal treatment protocol chosen in step 2.

24.3.4 Three-Dimensional Volumetric Exposure Patterning

The physical property changes as described can be patterned by a laser direct-write tool in a process known as laser volumetric exposure processing (LVEP). Figure 24.2 shows a maskless x-y-z laser direct-write patterning tool that can pattern in 3-D over a 0.5 × 0.5 m area with velocities approaching 0.5 m/sec and overall accuracy of ±0.75 µm (repeatability ±0.1 µm). Figure 24.3 shows a true 3-D microfabricated structure that has been exposed, baked, and etched.

Fig. 24.2. Maskless laser direct write patterning tool. The fabrication process is fully automated from design to manufacturing.

A series of recent experiments has determined that some of the physical property changes depend on the irradiance of the photo-exposure tool.⁵⁴ With a laser direct-write patterning tool, the delivered photon flux can be controllably altered commensurate with the 3-D tool path pattern.⁵⁵ Consequently, the physical property changes can have gradation, which can be used to advantage in making devices or structures on the same material substrate. The physical and chemical properties that currently can be varied over a limited range by controlling the optical excitation include optical transmission in the IR,⁵⁶ material strength/compliance,⁵⁴ optical index,⁵⁷ and chemical solubility in hydrofluoric acid.⁵⁸

Fig. 24.3. Scanning electron microscope (SEM) image of a true 3-D microfabricated double turbine structure that has been exposed, baked, and released. The turbine is a centimeter across with tapered vanes that get as thin as 30 µm. The design is made to look as if it were fabricated by traditional "cut-glass" processes.

Figure 24.4 shows an optical microscope photograph of target patterns showing the gradation of optical transmission that can be patterned. The data below the images show the measured optical transmission near IR telecom wavelengths as a function of the incident irradiance during the exposure step. This dependence allows for metering the optical transmission in the material locally.

Fig. 24.4. Optical photograph of three optical microscope images of target patterns with data showing the measured optical transmission near the IR telecom wavelengths for the structure shown in upper left.

Figure 24.5 also shows an optical photograph of a variegated height structure that has been patterned and chemically etched by just metering the laser irradiance to control locally the chemical etch rate (i.e., solubility in hydrofluoric acid). No masks were used in the exposure or in the chemical etch processing stages.

Fig. 24.5. Optical photograph of a variegated-height structure patterned and chemically etched by metering the laser irradiance. The structure is a step-down waveguide at THz frequencies. Not clearly observable but on the right are a series of slotted and appropriately spaced curved baffles.

The measured data in Fig. 24.6 show the change in chemical etch contrast (as defined by the ratio of chemical etch rates of an exposed region to that of the unexposed region) as a function of laser irradiance. These data show an etch contrast of 30:1. However, more recent results show that a 50:1 etch contrast is possible. The data also show a threshold irradiance that is defined by an etch contrast of 1:1. Clearly, below this value no discernible exposure results in any enhancement of chemical etching in comparison to the unexposed material. Using this threshold irradiance value to advantage, it is possible to pattern embedded fluidic channels using a laser direct-write tool. Figure 24.7 shows two scanning electron microscope pictures of embedded channels that have been patterned, phase-changed by baking, and subsequently etched (note: the process requires vias to transport the chemical etchant to the embedded structure). The exposure process was conducted from above the surface through the material; a focused laser beam was used to provide a high flux only within a small, localized volume around the beam waist. The laser irradiance was set during the patterning step to exceed the exposure threshold only within the embedded volume.

Fig. 24.6. Change in chemical etch contrast as a function of laser irradiance at two laser exposure wavelengths.

Figure 24.7 (left) shows two large-opening reservoirs that are connected by a tunnel (shown with a human hair threaded through). The tunnel is roughly 100 × 100 µm wide and approximately 500 µm long and is etched within a 1 mm thick PSGC wafer. In Fig. 24.7 right, 10 stacked embedded channels are shown also within a 1 mm thick PSGC wafer. These stacked channels were not patterned at a wafer edge, as the photograph can falsely imply; they were patterned away from an edge and through the material in a series of stacked exposures. Connected to the stacked exposures was a pattern called a chemical etch line that itself etched through the material and subsequently provided a chemical transport path to the stacked channels (i.e., the vias for chemical transport). By proper choice of the laser wavelength and the type of pulsed laser, it is possible to confidently pattern structures very near the surface or very deep within a solid (<30 µm deep for 193 nm wavelength, <500 µm deep for 266 nm wavelength, <5 mm deep for 355 nm wavelength, many cm deep for 800 nm wavelength).

Fig. 24.7. Two scanning electron microscope pictures of embedded channels patterned then phase changed and subsequently etched out. The left image shows two large opening reservoirs connected by a tunnel (shown with a human hair threaded through to denote the path). The right image shows 10 stacked embedded channels also within a 1 mm thick PSGC wafer.

An important aspect for space systems is the ability to pattern mechanical compliance where necessary. This is less of an issue for on-orbit operations but becomes paramount during the launch phase. It is expected that the MFM will contain sensors and devices that could be susceptible to vibration and shock. It is possible to vary, over a limited range, the mechanical stiffness of a PSGC substrate. This property arises as a result of the in situ growth of the crystals, which, if not etched away can make the overall structure stiffer. Because the density and size of the grown crystals can be controlled to some extent (i.e., via exposure and thermal treatment), mechanical properties can also be controlled. Figure 24.8 shows two images of mechanically compliant structures fabricated from PSGC. Figure 24.8 (left) shows a scanning electron microscope photo of a turbine blade that has been patterned and released from a wafer. The turbine blade vanes are roughly 155 µm wide (thickest point) and nearly 1 mm tall. Using an air jet, the turbine was spun to 180,000 rpm without failure (the experiment was stopped at this rpm). The turbine survives the initial start and acceleration because the material composition along the length of the blade varies, stiffer at some sections (i.e., more crystalline matter) and more ductile at others (i.e., less crystalline matter). Figure 24.8 (right) shows an optical microscope photograph of a common coil-shaped spring fabricated and released that is overlaid with a human hair for scale. The measured width at the uniform sections is around 10 µm. Other flat piston springs have also been fabricated for providing local compliance to an adjoining mechanically sensitive structure.

Fig. 24.8. Two optical microscope photographs of mechanically compliant structures fabricated from PSGC material. The image on the left shows a turbine blade that has been patterned and released from a wafer. The turbine blade vanes are roughly 155 µm wide and nearly 1 mm tall. The image on the right shows a common hotplate shape, spring fabricated and released, overlaid with a human hair for size compari son. The measured width at the uniform sections is approximately 10 µm.

With the ability to pattern—either by traditional lithography or with higher versatility by laser direct writing—micro- to macro-scale modification of material property changes in 2-D and 3-D becomes possible. The three-step process of controlled optical exposure, thermal treatment, and chemical etching enables the processing of material at more than square-meter scales with near micron resolution. Furthermore, with the current understanding of the photophysical process in PSGCs, there is the hope that additional material properties can similarly be controlled. This would be possible if the base composition is slightly altered during PSGC manufacturing to engender a family of materials that have various functionalities that can be patterned through photo exposure.

24.3.5 PSGC Material Attributes, Space Exposure, and Comparison with Other Fabrication Approaches

There have been at least three commercially prepared PSGCs in the past 30 years, all based on a similar composition. They are: Fotoform, manufactured by the Corning Glass Corporation in Corning, New York, USA; PEG3, synthesized by the Hoya Corporation of Tokyo, Japan; and Foturan, manufactured by the Schott Corporation of Mainz, Germany. Foturan is the only PSGC that is commercially available today. However, extensive data supports the conclusion that other compositional varieties are possible.^59,47 A brief summary of pertinent material characteristics and processing advantages is given below.

• PSGC can be molded/shaped before patterning (e.g., antenna and optical mirror dish shapes can easily be molded and then patterned).
• PSGC can be manufactured in any size or shape and then patterned.
• A PSGC substrate has low thermal conductivity; therefore, it is less affected by on-orbit temperature fluctuations in comparison to aluminum and silicon (at 20°C: Foturan ~2 W/mK; aluminum 250 W/mK; bulk silicon ~149 W/mK).
• PSGC has a moderate coefficient of thermal expansion (at 20–300 K: Foturan 8.6 × 10^-6 to 10.5 × 10^-6 depending on material phase; aluminum 25 × 10^-6; bulk silicon 2.9 × 10^-6).
• PSGC is lighter than most metals (Foturan 2.4 g/cm³; Al-2017 2.8 g/cm³).
• PSGC can be made tough (Knoop hardness: Foturan 530; nickel 560).
• The strength of PSGC can be tailored (Foturan modulus of rupture: 60–150 MPa).
• PSGC has a stiffness-to-weight ratio, E/ρ, (Foturan E/ρ = 35) that is better than that of steel (ASTM-A36 E/ρ = 25), aluminum (Al-1100-H14 E/ρ = 26), and is 42% that of silicon (E/ρ = 83).
• A PSGC substrate is an electrical insulator with a resistance of 8.1 × 10¹² to 5.6 × 10¹⁶ (Ω-cm at 25°C) depending on material phase.
• PSGC has higher emissivity (Foturan ~0.96) than bulk silicon (~0.64), glass (~0.95), and anodized aluminum (0.77).
• PSGC has better RF transparency than metals.
• The visible and IR transparency of PSGC can be locally controlled (~0% to 100%).
• PSGC does not outgas in vacuum (as opposed to sol-gel-based glasses).
• Compositional variation of the base material can act as a radiation shield for gamma and x rays (contains cerium).^60–62
• Zero porosity in ceramic state allows containment of high-pressure gas.
• Four material processing procedures exist for metallization, allowing for patterned high- and low-current lines, embedded lines, and a thin "film" covering on the outside to mitigate space charging effects.

The effects of LEO space environment on Foturan has been examined as part of a Materials International Space Station Experiment (MISSE). Both exposed and unexposed samples together with photostructured samples were flown in a 3-year mission that tested the effects of impinging atomic oxygen (ram direction) and UV light (sun direction). The returned samples show no major changes. Material tests are ongoing.

To assess the feasibility of mass production or mass customization, the LVEP technique for PSGC materials must be compared with other materials processing techniques. Table 24.1 shows a comparison of LVEP with conventional machining and three microfabrication techniques (bulk silicon micromachining, surface silicon micromachining, and LIGA). A few points are worth noting.

First, all the techniques are amenable to automation, which is a necessary requirement for achieving mass throughput efficiencies while maintaining component reliability. For example, the LVEP technique, along with the associated tooling, can be configured for fully electronic CAD/CAM compliance. The laser direct-write tooling shown in Fig. 24.2 operates under full CAD/CAM control.

LVEP can process material with nearly the same feature size, device thickness, and relative tolerance as LIGA and bulk silicon micromachining. Surface micromachining can fabricate a smaller feature size with a commensurate reduction in the range of device thicknesses. As for conventional machining (e.g., milling), it would be too costly and impractical to achieve the small feature sizes that are possible by all the other techniques; however, the relative tolerance of conventional machining supersedes all other techniques because it is possible to process very large linear dimensions with ultrafine diamond milling.

With regard to fabricating structures beyond the base material, LIGA and LVEP share an advantage over bulk and surface micromachining in that they can fabricate precision components in other materials via electroplating or serving as an injection mold tool. Conventional machining accommodates the largest range of materials because it is a direct subtractive process using abrasive force. LVEP shares with conventional machining and surface micromachining the attribute of scalability to processing large areas.

Whereas microelectronics can easily be integrated with surface and bulk silicon micromachining techniques, for LVEP and the current class of available PSGCs, only thin-film electronics integration is possible. However, with some technology development, PSGCs can perhaps be made compatible with low-temperature cofired ceramic (LTCC). LTCC is the electronics and RF substrate that has superseded printed circuit-board technology; it offers the advantage of 3-D electronics packaging on a ceramic substrate. With technology development, there is the hope that the coefficient of thermal expansion of PSGC (e.g., Foturan ~8.6 × 10^-6 at 300°C) can be reduced to match that of LTCC, currently designed to match silicon (~3.4 × 10^-6 at 300°C). If this were feasible, the result would allow direct wafer-scale anodic bonding of LTCC embedded electronics onto PSGC. There is a viable but less elegant solution that mitigates the bonding problem: It is the use of an interlayer that has intermediate thermal properties and can support the mismatch in the thermal expansion. Regardless of how electronics are integrated into PSGCs, the LTCC approach offers a significant advantage because of its maturity and wide use. The LVEP technique has an advantage when it comes to integrating photonics. Leaving the integration of the photonics source aside for the moment, the control and guiding of light within PSGC has already been accomplished.⁶³ Using a variant of the LVEP technique, one can create 3-D optical waveguides within the PSGC glass. Furthermore, optical reflecting surfaces integral to the PSGC wafer have been fabricated.⁶⁴ Figure 24.4 shows that the optical transmission in the IR can be controllably varied. These results (along with other data not presented⁶⁵) suggest that it should be possible to pattern optical filters integral to the device and thereby enable intrasatellite communication via wavelength-division multiplexing. The integration of a photonic source is problematic for the PSGC materials and the LVEP technique. Automated pick-and-place solutions that use a variant of LVEP are being explored,⁶⁶ but the most obvious solution is to integrate the photonic source as a surface device in LTCC. Subsequently, the emitted light would then be coupled and guided within the PSGC.

Table 24.1 also shows that the LVEP technique, like conventional machining, can fashion structures in 3-D with high aspect ratio (~50:1). It also shares with conventional machining the ability to fabricate components without the need of designing, manufacturing, and testing a lithography mask set. In conventional machining, maskless operation leads to a slow serially driven process that can be time consuming and therefore costly—especially when fabricating small feature sizes. Unlike conventional machining, in LVEP, the serial patterning segment can be done quickly (at m/sec velocities with galvanometer motion) because material is not being removed; only the pattern information is being written. The material transformation and material removal steps are done in a parallel or batch process. Serially driven processes have the advantage of not requiring a mask, which means that pattern customization can easily be implemented. For example, in a serial process, it is possible to implement multiple processes concurrently, and the fabrication of structures with variable height in true 3-D can easily be accomplished.

On the other hand, batch or parallel processes have the advantage of process and component uniformity over large areas and consequently low cost when numerous units are fabricated at once. In the LVEP technique, a serial and a batch process are merged with the consequence that advantages from both survive. First, by replacing the mask and lithography procedure with a serially driven direct-write variable-exposure procedure, a sequence of processing steps is immediately removed from the overall set. Figure 24.9 shows a schematic comparing a mask lithography process with that of LVEP, where the laser irradiance is dynamically varied during the patterning to match the desired property changes. The example shown is the fabrication of a shallow well next to a deep well, which requires two exposure masking steps. There are five basic processing steps using the conventional lithography mode (i.e., apply the resist, bake, perform the lithography, etch, and strip the resist) that have to be repeated for each specific etch depth (i.e., mask). Using the LVEP technique, there are only three (i.e., apply all patterns with variable exposure, bake, conduct a single timed etch to fabricate all features). Seven processing steps are saved in this simple example that uses just two masks. With increasing number of masks, N the number of sequential steps in the conventional mode increases as 5N while that for the LVEP remains at three.

Fig. 24.9. Schematic comparing a mask lithography process with that of LVEP where the laser irradiance is dynamically varied during the patterning to match the desired property changes.

A reduction in processing steps is usually commensurate with increasing fabrication quality. Figure 24.10 shows an optical microscope image of a microfabricated structure that would require four masks (i.e., 20 processing steps) using conventional lithography. A cross section of the fabricated part is shown in the adjacent panel in the figure. The structure is a four-level structure with the deepest feature being 600 µm. Plotted on top of the cross-section profile that depicts the design (i.e., calculated etch depth profile) is the measured profile. The variation between the measured and design is less than 10% in this prototype and can be further improved with additional processing controls.

Fig. 24.10. Optical microscope image of a microfabricated structure that would require four masks (i.e., 20 processing steps) to fabricate using conventional lithography (top). A cross section of the part with measured and calculated design depths (bottom).

Finally, as shown in Table 24.1, LVEP is the only technique that offers the ability to vary the substrate material property during the exposure patterning step. This is primarily a consequence of the PSGC material itself. However, these material property changes can only be articulated if there is precision in the delivery of the photons with respect to irradiance (photons/time-area), flux (photons/time; time is per laser pulse), and fluence (energy/area) at the specified volume element (i.e., voxel). A control scheme for pulsed laser direct-write processing has been developed that can deliver scripted photon irradiance, flux, and fluence to each predefined voxel during patterning.⁶⁷ The delivery of this exposure prescript can be done at patterning speeds approaching 0.5 m/sec. Figure 24.2 shows such a tool located at The Aerospace Corporation. In Fig. 24.11 is an optical photograph of a 10 cm diameter PSGC wafer that has been patterned with a complex sequence of both 2-D and 3-D microstructures. The wafer is an integrated system and represents a subsystem of a glass-ceramic micropropulsion prototype unit for a 1 kg nanosatellite. Details of the propulsion prototype unit and its function are described in chapter 18.

24.4 Thoughts on Development of a Mass-Customized MFM Foundry and Nanosatellites in PSGC

Commercially produced PSGCs and their variants can be implemented in a diverse array of new applications in which the material acts as the host substrate or support structure, and the desired "instrument" is fabricated within it or on top of it. Therefore, instruments or devices can be fabricated by direct patterning and local alteration of the material. Alternatively, the PSGC can be a substrate such as that in a multichip module (MCM) and allow the direct attachment of microelectronics, photonics, fluidic MEMS, micro-optoelectronic systems (MOEMS), and high-frequency RF communication systems. In this latter application, the PSGC is used as a multipurpose substrate that goes beyond the capabilities of current substrate materials (e.g., silicon, plastics, and even LTCC). The consequence is the potential to fabricate complex systems such as MFMs. An assembly of MFMs would subsequently be a complete small mass-producible satellite that is constructed almost entirely from glass ceramic, with added advantage that a photonics-based intrasystem communication bus can also be implemented, obviating the need for some electronic vias.

Fig. 24.11. Optical photograph of 10 cm diameter PSGC wafer patterned with a complex sequence of both 2-D and 3-D microstructures. The wafer is an integrated system representing a subsystem of a glass/ceramic micropropulsion prototype unit for a 1 kg nanosatellite. The left photograph shows the exposed and baked wafer before etching. The pattern is of the fluidic lines and the eight thruster nozzles. The right photograph shows two processed wafers integrated with the microvalves.

Figure 24.12 shows a schematic of a possible nanosatellite that is produced primarily of glass ceramic. The figure is intended to illustrate many of the capabilities described in the prior section. It shows a molded component shaped to form the propellant tank. A molded component can also be shaped with negative curvature to become an antenna or, with better surface figure control and smoothness, an optical focusing mirror. A nozzle is shown being fed by an embedded fluidic channel that goes to the propellant tank. This channel can also be routed through a number of stacked "wafers," as in the case of the glass-ceramic schematic prototype described in chapter 18 of this book. Three kinds of vias are also depicted in Fig. 24.12. First, there is a high-power bus—a low-gauge metallized via and data lines that could be manufactured by various metallization schemes described above. Second, there is an RF bus, which could either be an embedded hollow and metalized waveguide or an index-guided RF waveguide. Third, there is an internal optical communication bus that is depicted by a series of line-of-sight systems comprising an optical transceiver chip and a sensor. The line-of-sight optical paths also show beam splitters integrally fabricated to direct and distribute the optical signal. What is not depicted in Fig. 24.12 but relevant to an internal optical bus is the integral fabrication of cutoff filters and the fact that the optical paths need not always run in the line of sight. The inclusion of optical filters allows for wavelength division multiplexing. Using pulsed laser compaction techniques, single-mode optical waveguides can be integrally patterned in 3-D to connect any two points. Also not depicted are integrally fabricated microlenses that focus solar power onto a solar cell or a CCD chip in the case of a camera.

Fig. 24.12. Schematic of a possible primarily glass ceramic satellite that implements many of the capabilities described in the prior section.

A first-order attempt to design a manufacturing line for MFM production would utilize many of the concepts already in use in mass production. Some of these concepts are present in the software used in foundry design.⁶⁸ Furthermore, the first-order design will also have to borrow procedures and approaches currently in use in select small-satellite manufacturing centers where mass customization concepts are being implemented. Because an MFM can essentially become a satellite subsystem, the mission definition and design phase—the basis of most satellite development methodologies—has less relevance. An interesting approach being developed by the Air Force Research Laboratory (AFRL) is the satellite data model.⁶⁹ In this model, satellite devices and applications are developed separately; when there is a need for satellite components appropriate to a particular application or mission, they are selected and added to the "system." One goal is to reduce nonrecurring costs in the design, engineering, and manufacturing phases.⁷⁰ From the manufacturing side, reducing nonrecurring cost essentially means minimizing assembly steps and reducing manufacturing techniques. With regard to assembly, automation should be the aim. However, this does not necessarily argue for robotic pick-and-place machines; ultrasonic consolidation techniques⁷¹ with self-interlocking microstructures should also be explored.

Modeling and simulation tools should be used throughout the design phase, in conjunction with cost-estimating software. Included in the simulation "toolbox" is software for systems integration and its impact on the manufacturing process development. However, it is imperative that the modeling and simulation software have a factual basis that is taken from both ground and space experiments. The current technology demonstrations being conducted with small satellites (e.g., CubeSats, picosatellites, nanosatellites, microsatellites) are key to ensuring the validity of these future simulation tools. The design of the MFM and the subsequent family of derivable satellites would be the goal, addressing mission requirements at the satellite system level. The mission requirements then flow down to the MFM design, as a customization procedure. Furthermore, if the design software (e.g., solid modeling) were to be coupled with the space systems orbital dynamics tools, it could help couple the mission needs to the requirements that are to be levied on the MFM. The modeling and simulation software industry is following a trend in which the software goes beyond CAD/CAM and animation and includes engineering and physics (e.g., thermal modeling, mechanics) and could include space environment specifics (e.g., available power, radiation level).

The design of the MFMs themselves must implement a modular approach where each segment is a separable submodule that can be tested individually using an adopted universal interconnection scheme. To make this design approach feasible, intelligence by way of processors is locally embedded, and computation is distributed to enable plug-and-play assembly and disassembly.⁷² If the onboard computation is distributed, intrinsic in the design is a level of fault tolerance that can be implemented because the MFM can be "operated or tested" via multiple control routing schemes (techniques toward realizing this approach via wireless communications can be found in chapter 19). In subsequent design generations, if the range over which control, via distributed computation protocols, is increasingly expanded, then it could serve as the basis for connecting multiple MFMs to form larger and more complex systems that are designed to work in concert. Ultimately, this process is realized in a functional satellite. Finally, within this distributed control "fabric," neural network analysis software could be used to monitor system trends and highlight functional areas that are out of the nominal range.⁷³ A satellite nervous system is then essentially born.

The MFM manufacturing line will be similar to a microelectronics foundry—in essence, a series of automated cluster tools, pick-and-place machines, and diagnostics instruments in a cleanroom. For a PSGC-based MFM manufacturing line, an abbreviated list of the major processing stations are given below with the understanding that these stations may include multiple processes and that quality-control testing is conducted at each one.

• Shaping and molding layers/wafers for more mission-specific applications.
• Patterning in 2-D and 3-D (RF conduits, optical conduits, fluidic conduits, MEMS sensors and actuators, mechanical devices).
• Batch chemical baking.
• Batch chemical etching.
• Patterning for thin-film electronics.
• Electronics deposition.
• Segment qualification.
• Segment assembly with LTCC electronics.
• Segment packaging/bonding.
• Plug-and-play segment testing.
• MFM as an assembly of plug-and-play segments.
• MFM testing.
• MFM vibrational testing.
• MFM thermal vacuum testing.

24.5 Steps Forward

To realize the mass customization and manufacture of an MFM, several initial steps must be undertaken, starting with a more extensive engineering analysis along with the development of a cost-modeling tool that marshals a rational business case. Government space-policy experts, along with the national laboratories and the major satellite manufacturers, are in a position to evaluate the concept from the perspective of both engineering viability and profitability. Expert opinion should also be drawn from producers of both microelectronics and consumer electronics. This insight will help clarify what is in fact possible with regard to modularity, packaging, automated assembly, and distributed manufacturing. Following the results of this interdisciplinary team analysis, national governments should sponsor small space research centers to explore and demonstrate aspects of the mass-production concepts.

A key element in the proposed concept is the use of glass and ceramic materials as the primary structural material replacing metal; PSGC is the material of choice because it enables lithographic patterning. With lithography comes the consequent batch fabrication technologies that have enabled the microelectronics industry to increase capability without losing reliability. Still, a fundamental problem with PSGC must initially be addressed: Even though the materials have been available for nearly 60 years, issues of nonuniform photosensitivity remain. Given what the material properties of bulk silicon were at the infancy of the microelectronics industry compared with what is now available, it's likely that the glass-ceramics industry will be able to solve this problem. However, far better than refining the composition and uniformity of the existing PSGC materials would be to develop a material specification and requirements list for a class of PSGC that would truly be suitable for the development of an MFM (e.g., rad-hard variants of current PSGC). An argument can then be made to establish a PSGC materials manufacturing line that develops materials with specific and tailored requirements. The estimated required quantities of material will be measured on the order of 10–100 tons per year. In materials manufacturing, there is an inherent cost in terms of the energy expended to mine, refine, and produce a select material. This energy factor (MJ/kg) roughly defines the associated environmental cost for a given material and can be used to compare various MFM structural candidates. Figure 24.13 shows this comparison for a select number of materials.⁷⁴ The PSGCs are manufactured as a glass as opposed to a glass-ceramic composite. Silicon carbide is included as a representative monolithic crystalline material. The data show that glass (~25 MJ/kg) followed by the glass-ceramic composites (~100 MJ/kg) require the lowest amount of energy to produce.

Fig . 24 .13 . Environmental energy cost of material production.

The LVEP approach with PSGC holds advantages that can aid in the design and development of integrated MFMs. Though research-grade fabrication tools do exist, these are far from true manufacturing-quality tools. However, there are no real technological barriers to developing a manufacturing tool that uses lasers and processing techniques as described above. For example, the manufacturing setup shown in Fig. 24.2 incorporates lasers (Spectra-Physics OEM Models J40-BL6-266Q and J40-BL6-355Q) used in other product manufacturing lines, so their reliability has been established. The motion systems are taken directly from the microelectronics manufacturing industry, and the CAD/CAM software tools used are taken from the conventional machining industry. Following the initial exposure step, the processes (thermal treatment and chemical etching) are directly acquired from the microelectronics industry. Regardless of where these technologies originate, a recommendation would be to develop a manufacturing-class LVEP tool that can demonstrate the necessary speed and piece-part fabrication throughput.

The development of a pilot manufacturing plant is not considered a first step. However, following the interdisciplinary team overview, it would be prudent to begin appraising processing capabilities worldwide where subsystems of the MFM can be designed and manufactured for assembly or bonding at a central facility.

An approach toward developing a distributed and agile pilot plant is to enable all the physical machines to "communicate" via a common software interface—for example, the Common Object Request Broker Architecture (CORBA).⁷⁵

The MFM should be developed in stages. Picosatellites, nanosatellites, and CubeSats can serve a role in this development process. Low-cost space testing can be conducted using these small space systems, and as each subsequent MFM generation grows in complexity, so will these small spacecraft grow in capability. In time, there will be a transition from each small spacecraft initially serving as a bus for the MFM package to the spacecraft evolving into an entity that is merely an assembly of MFMs. One positive consequence of this approach is that these new generations of small spacecraft will be mass producible, and large satellites can begin carrying the mass-producible small satellites either as assistants or components of complex distributed missions.

Current producers of small satellites and various research groups worldwide are developing technologies intended to make future space systems more reliable and easier to integrate. Many of these technologies are presented in the pages of this book. Some of them could find direct application to the MFM concept—for example, communication protocols that enhance plug-and-play operation, or a generic and reconfigurable system-on-a-chip to handle onboard and local computation. Other developing technologies that would have to be adapted to the MFM concept include, for example, miniature propulsion and attitude sensors. Finally, some technologies must be developed that would be specific to the MFM and the potential assembly of larger structures in space. Examples include:

• Development of a universal microconnection scheme that can couple fluidics, power, and photonics from one MFM to another.
• Development of proximity sensors that would enable self-assembly in space.
• High-precision propulsion for self assembly only.
• An integral unit that generates power for local and distributed use.
• Use of celestial x-ray sources for attitude sensing. This would allow each MFM with a modified star tracker to have an autonomous sense of location.⁷⁶

24.6 Conclusions

Changing the way we use space can entail more than simply replacing a single large satellite with a counterpart that is just more integrated. On the contrary, satellites as an assembly of mass-producible multifunctional units will have a far more pervasive impact on the aerospace industry by increasing reliability, reducing cost, and enabling new missions that are desired but not possible today. This chapter briefly outlines the idea of mass production of satellites. It describes one approach toward realizing the goal of economies of scale—defining the production unit not as the satellite but as an element of the satellite, the "multifunctional module." The MFM provides the basis for designing and building satellites in a novel manner and permits the adoption of manufacturing procedures and practices that can increase overall reliability. This proposed concept is a far-term vision.

This chapter goes beyond the proposition and suggests the properties that an MFM should have and how it should be developed to acquire the benefits that the microengineeering and microelectronics industries have enjoyed. Presented is the use of glass-ceramics materials, specifically the family of photostructurable glass ceramic, as a candidate structural material and as a substrate for integrating the fluidic, RF, electrical, and possibly photonic systems. Glass and ceramics have an advantage over silicon because they can be manufactured in large quantities and can be macroscopically shaped using cost-effective molding and casting technology. In addition, these materials can further be processed with near micron precision using lithography and batch chemical etching techniques.

It should be possible to leverage the vast resources and technologies of both the glass-ceramics and microelectronics industries and apply them to the development and mass manufacture of MFMs, thereby enabling development of a satellite as an assembly of mass-processed modules. One outcome of this approach is that future spacecraft could have any nominal size and shape.

24.7 Acknowledgments

The author acknowledges The Aerospace Corporation and The Aerospace Press for the opportunity to write this chapter. The author also acknowledges Dr. Howard Schlossberg, Program Manager in the Air Force Office of Scientific Research (AFOSR), who has provided continuous sponsorship of the fundamental investigations and the characterization of the PSGC using laser exposure techniques. The Aerospace Corporation's Independent Research and Development Program is also acknowledged for providing funding to demonstrate the capabilities of the laser material processing technique. Furthermore, the Department of Energy's Jefferson Laboratory is thanked for sponsoring the development of the LVEP tool. These investigations would not have been possible without the help and discussions with Siegfried Janson, Frank Livingston, William Hansen, Ernest Robinson, and Lee Steffeney of The Aerospace Corporation. Finally, the author acknowledges the late Seymour Feuerstein of The Aerospace Corporation, whose belief in the power of miniaturization helped to promote the microengineering of aerospace systems.

24.8 References

1. A. Tennyson, "The Two Voices" (1842).
2. C. Y. Baldwin and K. B. Clark, "Managing in an Age of Modularity," Harvard Business Review, 75 (5), 84–93 (Sept.–Oct. 1997).
3. C. Y. Baldwin and K. B. Clark, Design Rules: The Power of Modularity (MIT Press, 2000).
4. D. E. Whitney, Working paper, ESD-WP-2003-01.03-ESD, Massachusetts Institute of Technology, 2004 (unpublished).
5. K. Hölttä, E. S. Suh, and O. deWeck, "Trade-off between Modularity and Performance for Engineered Systems and Products," ICED 2005: The 15^th International Conference on Engineering Design (Melbourne, Australia, 15–18 Aug. 2005).
6. D. E. Whitney, Working paper, ESD-WP-2003-01.03-ESD, Massachusetts Institute of Technology, 2004 (unpublished).
7. S. Flagg, T. Bleier, C. Dunson, J. Doering, L. DeMartini, P. Clarke, L. Franklin, J. Seelbach, J. Flagg, M. Klenk, and V. Safradin, "Using Nanosats (e.g., QuakeSat) as a Proof of Concept for Space Science Missions," 18^th Annual/USU Conference on Small Satellites (AIAA, Logan, UT, 2004), Vol. SSC04-IX-4.
8. S. W. Janson, in Micro- and Nanotechnology for Space Systems: An Initial Evaluation, edited by H. Helvajian and E. Y. Robinson (The Aerospace Press, El Segundo, 1993); Aerospace Report ATR-93(8349)-1, Second printing AIAA press 1997, pp. 138; S. W. Janson, 30^th AIAA/ASME/SAE/ASEE Joint Propulsion Conference (AIAA, Washington DC, Indianapolis, IN, 1994), Vol. AIAA-94-2998, pp. 14; S. W. Janson, in Microengineering Technology for Space Systems, edited by H. Helvajian (The Aerospace Press, El Segundo, 1995); Aerospace Report ATR-95(8168)-2, Second printing AIAA 1997, pp. 309; S. Janson, H. Helvajian, and E. Y. Robinson, 44^th Congress of the International Astronautics Federation (IAF, Graz, Austria, 1993), Vol. IAF-93-U.5.573, p. 5.
9. iFixit, iPhone Disassembled (IFixit, Atascadero, CA, 2007).
10. M. Sweeting, "25 Years of Space at Surrey–Pioneering Modern Microsatellites" (IAA, Melbourne, Australia, 1998).
11. Rapid Spacecraft Development Office, http://rsdo.gsfc.nasa.gov (23 July 2008).
12. B. M. Cook, C. Paul, and H. Walker, "SpaceWire Plug and Play, An Early Implementation and Lessons Learned," 2007 Conference and Exhibit (AIAA, Rohnert Park, CA, 2007), Vol. AIAA 2007-2930, p. 6.
13. M. Sweating, A. D.-S. Curiel, and W. Sun, "The 'Personal Computer' Revolution in Space," 20^th Annual Small Satellite Conference (AIAA, Ogden, UT, 2006).
14. A. Castano, W-M. Shen, P. Will, "CONRO: Towards Deployable Robots with Inter-Robots Metamorphic Capabilities," Autonomous Robots, 8, 309–324 (2000).
15. S. O. Kennedy Jr, T. Mosher, and Q. Young, "The Rise and Fall of the Capital Asset—An Investigation into the Aerospace Industry Dynamics and Emerging Small Satellite Missions," 20^th Annual Small Satellite Conference (AIAA, Logan, UT, 2006), Vol. SSC06-I-5.
16. ISBC, State of the Space Industry (ISBC, Bethesda, MD, 2005).
17. C. Blalock, S. Busby, W. Chapman, M. Evans, and E. Helton, "Space Industry Study Industrial College of the Armed Forces National Defense University," Report No. A682524 (Ft. Lesley J. McNair, Washington, DC, 2002).
18. J. H. Saleh, "Will We See a Regional or Global Duopoly in the Satellite Manufacturing Industry? The European Perspective," Space Policy, 21 (4), 277–285 (Nov. 2005).
19. P. Hawken, The Next Economy (Holt, Rinehart, and Winston, 1983).
20. Futron Corporation, Industrial College of the Armed Forces (28 March 2002).
21. Y. A. Dehqanzada and A. M. Florini, Secrets for Sale: How Commercial Satellite Imagery Will Change the World (Washington, DC: Carnegie Endowment for International Peace, 2000).
22. Committee on Earth Studies Space Studies Board Commission on Physical Sciences Mathematics Applications, The Role of Small Satellites in NASA and NOAA Earth Observation Programs (National Academy Press, Washington, DC, 2000).
23. W. J. Abernathy and J. M. Utterback, "Patterns of Industrial Innovation," Technology Review, 80, 40–47 (June–July 1978).
24. D. M. Anderson, Design for Manufacturability & Concurrent Engineering (CIM Press, 2004).
25. H. Ford, My Life and Work (Doubleday, Page & Company, New York, 1923).
26. J. P. Womack, D. T. Jones, and D. Roos, The Machine That Changed The World. The Story of Lean Production (Simon & Schuster Ltd., 1990).
27. P. S. Pande, R. P. Neuman, and R. R. Cavanagh, The Six Sigma Way: How GE, Motorola, and Other Top Companies are Honing their Performance (McGraw-Hill Professional, 2000).
28. M. M. Tseng and J. Jiao, Mass Customization (New York: Wiley, 2001).
29. B. Hindo, "At 3M, A Struggle Between Efficiency and Creativity," BusinessWeek (14 Sept. 2007).
30. T. L. Friedman, The World is Flat: A Brief History of the 21^st Century (Farrar, Straus and Giroux, New York, 2006).
31. S. Brusoni and A. Prencipe, "Modularity as an Entry Strategy: The Invasion of New Niches in the LAN Equipment Industry," CESPRI Working Papers 171, (CESPRI, Centre for Research on Innovation and Internationalisation, Universitˆ Bocconi, Milano, Italy, revised July 2005).
32. MEMS and Nanotechnology Exchange, http://www.mems-exchange.org (23 July 2008).
33. W.-M. Shen, P. Will, A. Galstyan, and C.-M. Chuong, "Hormone-Inspired Self-Organization and Distributed Control of Robotic Swarms," Autonomous Robots, 17 (1) pp. 93–105 (2004).
34. J. Fraden, AIP handbook of Modern Sensors, Physics, Design and Applications (American Institute of Physics, New York, 1993).
35. K. E. Petersen, "Silicon as a Mechanical Material," Proceedings IEEE, 70 (5) p. 421 (1982).
36. J. B. Angell, S. C. Terry and P. W. Barth, "Silicon Micromechanical Devices," Scientific America, 248 (4), p. 44 (April 1983).
37. A. Cohen, G. Zhang, F. Tseng, F. Mansfield, U. Frodis, and P. Will, EFAB: "Rapid, Low-Cost Desk Micromachining of High Aspect Ratio of True MEMS," 12^th IEEE International Conference on Micro Electro Mechanical Systems (1999), pp. 244–251.
38. Y. Tang, H. T. Loh, Y. S. Wong, J. Y. H. Fuh, L. Lu, and X. Wang, "Direct Laser Sintering of a Copper-Based Alloy for Creating Three-Dimensional Metal Parts," Journal of Materials Processing Technology, 140 (1–3), pp. 368–372 ( 2003).
39. J. W. Lee, H. Lee, and D.-W. Cho, "Development of Micro-Stereolithography Technology Using Metal Powder," Microelectronic Engineering 83, 1253–1256 (2006).
40. K. P. Rajurkar, "Technology and Research in Electrodischarge and Electrochemical Machining," American Society of Mechanical Engineers, 43, 309 (1990).
41. Handbook of Conducting Polymers, 2^nd edition, edited by T. A. Skotheim, R. L. Elsenbaumer, and J. R. Reynolds (CRC Press, New York, 1998).
42. K.-H. Haas and H. Wolter, Synthesis, Properties, and Applications of Inorganic-Organic Copolymers (Ormocers) (Elsevier, 1999).
43. R. Buestrich, F. Kahlenberg, M. Popall, P. Dannberg, R. Muller-Fiedler, and O. Rosch, "Ormocers for Optical Interconnection Technology," Journal of Sol-Gel Science and Technology, 20 (2), 181 (2001).
44. M. Madou, Fundamentals of Microfabrication (CRC Press, New York, 1997).
45. J. J. Simonne, "A Systems Approach to Microsystems Development," in Microengineering Aerospace Systems, edited by H. Helvajian (AIAA and The Aerospace Press, El Segundo, CA, 1999), pp. 227–258.
46. D. Kochan, "Solid Freeform Manufacturing: Possibilities and Restrictions," Computers in Industry, 20, (2), 133–140 (1992).
47. P. W. McMillan, Glass Ceramics, 2^nd ed. (Academic Press, London, UK, 1979).
48. W. Holland and G. Beall, Glass-Ceramic Technology (American Ceramic Society, Westerville, OH, 2002).
49. S. H. Messaddeq, V. K. Tikhomirov, Y. Messaddeq, D. Lezal, and M. S. Li, "Light-Induced Relief Gratings and a Mechanism of Metastable Light-Induced Expansion in Chalcogenide Glasses," Physical Review B, 63, 224203 (2001).
50. S. D. Stookey, "Photosensitive Glass: A New Photographic Medium," Industrial and Engineering Chemistry, 41, 856 (1949).
51. S. D. Stookey, "Chemical Machining of Photosensitive Glass," Industrial and Engineering Chemistry, 45, 115 (1953).
52. H. Helvajian, "3D Microengineering via Laser Direct-Write Processing Approaches," in Direct-Write Technologies for Rapid Prototyping Applications: Sensors, Electronics, and Integrated Power Sources, edited by A. Pique and D. B. Chrisey (Academic Press, NY, 2002), ch. 14.
53. P. D. Fuqua, D. P. Taylor, H. Helvajian, W. W. Hansen, and M. H. Abraham, in Materials Development for Direct-Write Technologies, edited by D. B. Chrisey, H. Helvajian, and D. P. Taylor (Materials Research Society, Pittsburgh, PA, 2000), Vol. 624, pp. 79.
54. F. E. Livingston and H. Helvajian, "Selective Activation of Material Property Changes in Photostructurable Glass Ceramic Materials by Laser Photophysical Excitation," Journal of Photochemistry and Photobiology A: Chemistry, 182 (3), 310–318 (2006).
55. F. E. Livingston, L. F. Steffeney, and H. Helvajian, "Genotype-Inspired Laser Material Processing: a New Experimental Approach and Potential Applications to Protean Materials," Applied Physics A: Materials Science & Processing, vol. 93, pp. 75–83 (2008).
56. F. E. Livingston and H. Helvajian, "Photophysical Processes that Lead to Ablation-Free Fabrication in Glass-ceramic Materials," in 3D Laser Microfabrication: Principals and Applications, edited by S. J. H. Misawa (Wiley-VCH, Weinheim, 2006), Ch. 11.
57. V. R. Bhardwaj, E. Simova, P. B. Corkum, D. M. Rayner, C. Hnatovsky, R. S. Taylor, B. Schreder, M. Kluge, and J. Zimmer, "Femtosecond Laser-Induced Refractive Index Modification in Multicomponent Glass," Journal of Applied Physics, 97, 083102 (2005).
58. F. Livingston, W. Hansen, A. Huang, and H. Helvajian, "Effect of Laser Parameters on the Exposure and Selective Etch Rate in Photostructurable Glass," Proceedings of the SPIE, 4637, 404 (2002).
59. A. Berezhnoi, Glass-ceramics and Photo-Sitalls (Plenum Press, NY, 1970).
60. P. W. Levy, "The Kinetics of Gamma-ray Induced Coloring of Glass," Journal of the American Ceramic Society, 43 (8), 389–395 (1960).
61. G. S. Monk, "Coloration of Optical Glass by High-Energy Radiation," Nucleonics, 10, 52–55 (1952).
62. S. M. Brkhovskikh, Y. N. Viktorova, Y. L. Grinshteyn, and L. M. Landa, (National Technical Information Service, U.S. Dept. Commerce, 1973).
63. K. Sugioka, Y. Cheng, and K. Midorikawa, in Photon-Based Nanoscience and Nanobiotechnology, edited by J. J. Dubowski and S. Tanev (Springer, Dordrecht, The Netherlands, 2006), Vol. 239.
64. Y. Cheng, K. Sugioka, K. Midorikawa, "Microfluidic Laser Embedded in Glass by Three-Dimensional Femtosecond Laser Microprocessing," Optics Letters, 29 (17) 2007–2009 (2004).
65. F. E. Livingston and H. Helvajian, "Photophysical Processes that Activate Selective Changes in Photostructurable Glass Ceramic Material Properties," in Photon-based Nanoscience and Nanobiotechnology, edited by J. J. Dubowski, and S. Tanev (Springer, Dordrecht, The Netherlands, 2006), pp. 225–266.
66. G. K. I. Zergioti, G. Koundourakis, N. A. Vainos, and C. Fotakis, "Laser-Induced Forward Transfer: An Approach to Single-Step Microfabrication," Direct-Write Technologies for Rapid Prototyping Applications: Sensors, Electronics, and Integrated Power Sources, edited by A. Pique and D. B. Chrisey (Academic Press, New York, 2002), Ch. 16.
67. F. E. Livingston, L. F. Steffeney, and H. Helvajian, "Tailoring Light Pulse Amplitudes for Optimal Laser Processing and Material Modification," Applied Surface Science, 253 (19), 8015–8021 (2007).
68. G. Qiao, R. Lu, and C. McLean, "Flexible Manufacturing System for Mass Customization Manufacturing," International Journal of Mass Customisation, 1 (2/3), 374–393 (2006).
69. K. Sundberg, S. Cannon, T. Hospodarsky, D. Fronterhouse, and J. Lyke, "A Satellite Data Model for the AFRL Responsive Space Initiative," 20^th Annual AIAA/USU Conference on Small Satellites (AIAA, Ogden, UT, 2006), paper number SSC06-II-9.
70. S. Eagleson, K. Sarda, S. Mauthe, T. Tuli, and R. E. Zee, "Adaptable, Multi-Mission Design of CanX Nanosatellites," 20^th Annual AIAA/USU Conference on Small Satellites (AIAA, Ogden, UT, 2006), paper SSC06-VII-3.
71. J. Clements, J. George, and B. E. Strucker, "Rapid Manufacturing of Satellite Structures and Heat Pipes Using Ultrasonic Consolidation," 20^th Annual AIAA/USU Conference on Small Satellites (AIAA, Ogden, UT, 2006), paper SSC06-V-4.
72. M. Wilkinson, "The Changing Paradigms of Satellite Reconnaissance, Creating Opportunities in the Small Satellite Industry," 20^th Annual AIAA/USU Conference on Small Satellites (AIAA, Ogden, UT, 2006), paper number SSC06-III-7.
73. M. Hammond and R. Jobman, "Satellite-as-a-Sensor Neural Network Abnormality Classification Optimization," 20^th Annual AIAA/USU Conference on Small Satellites (AIAA, Ogden UT, 2006), paper SSC06-III-2.
74. U. Cambridge, http://www-materials.eng.cam.ac.uk/mpsite/interactive_charts/energy-cost/IEChart.html (July 2008).
75. C. M. Pancerella and R. A. Whiteside, "Using CORBA to Integrate Manufacturing Cells to a Virtual Enterprise," Proceedings of SPIE: Plug and Play Software for Agile Manufacturing (SPIE, Bellingham, WA), Vol. 2913, 1996, pp. 148–173.
76. D. Beckett, "Overview of the XNAV Program, X-Ray Navigation Using Celestial Sources," 20^th Annual AIAA/USU Conference on Small Satellites (AIAA, Ogden, UT, 2006), paper SSC06-VI-6.

 

---

Order this book

Find more books