ASIC prototyping gives product teams a way to evaluate custom silicon before committing to larger manufacturing runs. For small and medium-sized companies, research groups and system developers, the challenge is not only designing a chip: it is also gaining access to foundry processes, libraries, fabrication capacity, packaging, test and the expertise needed to move from first silicon toward production.
A coordinated service model can bring these activities together while allowing customers to use only the support they need. The practical value lies in reducing barriers to early fabrication and providing a path through the engineering and manufacturing steps that follow.
Why ASIC prototyping can be difficult to access
Custom chips involve non-recurring engineering (NRE) costs as well as design, mask, fabrication and test work. Those costs can be challenging for a project with limited initial demand, particularly when a customer does not have the resources or experience to manage every part of the semiconductor supply chain independently.
Access agreements with foundries and library providers can also involve technical and administrative steps. A service that coordinates access to design rules, process information, cell libraries and design kits can help customers begin design work without establishing each relationship from scratch. Technical support can also help teams address matters such as technology selection, test planning and package requirements.
How multi-project wafer runs share prototype costs
A multi-project wafer (MPW) run combines designs from several customers on a shared mask set and fabrication run. Instead of one customer bearing the full cost of a dedicated mask set, participating projects share some of the associated expense. This makes MPW runs particularly useful for early silicon evaluation when a design needs a limited number of prototypes.
Before fabrication, submitted layouts undergo checks such as design rule checking (DRC) and electrical rule checking (ERC). These checks help identify issues against process requirements before the design is sent to the foundry. Depending on the project, prototypes may be delivered as bare dies or assembled devices, enabling engineers to evaluate functionality on the bench.
Another approach described for reducing mask expense is multi-level mask (MLM) processing. In the service model outlined in the source material, mask area is divided so that a mask can be used for multiple design layers. This approach was presented as an option for certain technologies and circumstances, including projects needing more prototypes or a larger design area than may suit a scheduled shared run. Availability and suitability depend on the process and manufacturing arrangements.
ASIC prototyping is one stage in a broader design flow
Fabrication is only one part of ASIC development. The work typically starts with front-end activities, such as defining specifications, evaluating feasibility, entering schematics or describing logic in VHDL or Verilog, simulation, synthesis and scan insertion. Customers may perform this work in-house or use an external design partner.
Once a netlist is ready, back-end design prepares the physical implementation. This can include layout generation, place and route, timing analysis and iterative optimization. Deep-submicron projects may require additional attention to areas such as clock behavior, power distribution, signal integrity, process variation and the integration of intellectual property blocks.
Before tape-out, the layout is checked against the schematic and manufacturing rules. DRC and layout-versus-schematic (LVS) verification help confirm that the final GDS-II database is ready for fabrication. The design flow also needs to consider design for testability (DFT): planning how the chip will be tested can influence both its implementation and the later development of production test programs.
Choosing a technology for the application
The smallest process node is not automatically the right choice for every ASIC. Technology selection depends on the application’s functional, electrical and manufacturing needs. The source material identifies 0.18-micron CMOS, including mixed-signal, radio-frequency, high-voltage and embedded-flash options, and 65-nanometer CMOS as technology areas of interest for a range of differentiated applications. It also describes access to more advanced nodes, including 28 nanometers, in the period covered by the document.
For engineering teams, the key consideration is to match process characteristics and available design resources to the intended circuit. A technology portfolio may include logic, mixed-signal, RF, high-voltage, silicon-germanium and photonics options. The appropriate selection should be made in conjunction with the project’s design requirements and the process documentation, libraries and support available for that technology.
Moving from prototypes toward production
Successful bench evaluation does not, by itself, establish production readiness. A production transition can involve developing an automatic test equipment (ATE) solution, characterizing the device, assessing variation and completing qualification activities suited to the application.
Characterization checks whether the ASIC meets its specifications, including across relevant operating conditions. Statistical analysis can help assess sensitivity to process variation. Qualification may include reliability and environmental testing, with the specific flow determined by the product and applicable requirements. For space applications, the source describes support aligned with ESCC 9000, including lot acceptance activities and radiation testing.
As production ramps, wafer and assembly data can be reviewed to identify yield concerns. Supply-chain coordination can connect foundry fabrication with packaging, test and logistics, while providing a single interface across suppliers. This can be especially useful for organizations adopting a customer-own-tooling model but lacking extensive experience in areas such as test, packaging and yield analysis.
Practical implications for engineering teams
When planning a custom chip, teams should evaluate more than the prototype price or process node. Consider how the design will be checked before tape-out, how prototype devices will be assembled and tested, what production test development may be needed, and which qualification requirements apply. It is also important to understand the expected path from prototype quantities to any initial or continuing production volumes.
The service model described in the source has supported both academic and commercial projects. Its historical results include 544 ASIC designs prototyped on MPW runs in 2014; this figure reflects that period, not a current activity level. The broader lesson is that shared fabrication and coordinated technical support can make custom silicon development more accessible to teams that do not need large volumes at the outset.
Conclusion
ASIC prototyping is most effective when it is planned as part of a complete development path, from technology and design choices through fabrication, bench evaluation, test, qualification and production ramp. MPW runs can help distribute prototype costs, while design and supply-chain support can address expertise and coordination gaps. For engineers and decision-makers, aligning those steps early helps clarify what is required to move a custom IC from initial silicon toward a manufacturable product.