As digital IC engineers, we all know that the tape-out process in chip design is both an exciting and stressful phase — the tape-out itself. Today, let’s discuss the two main options available during tape-out:MPW and Full Mask. This is not just a technical choice, but a strategic decision.
What are the differences between these two methods? To put it in a more straightforward analogy, you can imagine the difference between carpooling and a dedicated car service.
MPW is like a carpool service, where multiple passengers share a vehicle and split the costs.Full Mask, on the other hand, is akin to a dedicated car service, where you enjoy the entire vehicle alone and bear the costs yourself.This analogy aptly reveals the essential differences between the two methods.Now, let’s break this down in detail using this analogy.1. MPW – “Shared” Manufacturing (1) What is MPW?MPW, which stands for Multi-Project Wafer, is a business model that integrates multiple chip designs from different companies and functionalities onto a single wafer for manufacturing. (2) How does it work?The wafer foundries (such as TSMC and SMIC) regularly initiate an MPW tape-out batch. Various chip design companies (like us) act like passengers, “buying tickets” to submit their chip design layouts (GDSII files). The foundry arranges all the “passengers'” designs like a puzzle on a single wafer and produces them all at once. After manufacturing, the wafer is cut, and each company retrieves its respective chip. (3) Why choose it?To minimize costs to the extreme. The most expensive part of chip manufacturing is creating the mask (which can be understood as the “film” for making photos). By “sharing”, all participants collectively bear the exorbitant costs of the mask and wafer manufacturing. For startups, academic research, or small batch trials, this is the only economically viable option. (4) What are the drawbacks?
- No flexibility: You must adhere to the “departure schedule” set by the foundry and cannot tape-out whenever you want.
- Resource limitations: Your chip area is strictly limited because you need to leave space for other “passengers” on the wafer.
- Fixed cycle: The timeline from “boarding” to “receiving the chip” is fixed and cannot be expedited.
2. Full Mask – “Exclusive” Manufacturing (1) What is Full Mask?Full Mask refers to a scenario where a single company independently bears all manufacturing costs and capacity for an entire wafer, with each chip on the wafer being the same design. (2) How does it work?Your company designs a custom complete mask set for your chip and monopolizes an entire production batch at the foundry (meaning a whole batch of wafers is used to manufacture your chip), for which you need to bear all mask and wafer manufacturing costs. (3) Why choose it?To pursue ultimate performance, capacity, and control.
- Complete autonomy: You can freely decide the tape-out time, production pace, and quantity.
- Maximized area: You can design the largest possible chip without sharing wafer space with others.
- Cost optimization: When the production volume is sufficiently large, although the initial investment is huge, the average cost per chip will be much lower than that of the MPW method.
(4) What are the drawbacks?Expensive! Very expensive! The mask and manufacturing costs can reach millions or even tens of millions. This is a gamble; if the chip design is flawed, all that investment will go to waste.
3. A Table to Understand the Differences
|
Characteristic Dimension |
MPW (Shared Manufacturing) |
Full Mask (Exclusive Manufacturing) |
|
Core Analogy |
Carpooling | Dedicated Car Service |
|
Cost |
Very low (shared costs) |
Very high (solely borne) |
|
Flexibility |
Low (fixed cycle) |
High (self-determined schedule) |
|
Chip Area |
Limited (must share wafer) |
Free (can monopolize wafer) |
|
Risk |
Low (low trial and error cost) |
Very high (significant loss if failed) |
|
Applicable Stage |
Prototype validation, small batch, research |
Mass production, commercialization |
Conclusion
In simple terms, MPW is the “prototype” stage of chips, while Full Mask is the “official product” stage.
MPW, the multi-project wafer technology, is an economical “carpooling” model, making it a cost-effective choice in the chip industry.
Full Mask, the full mask tape-out, is an exclusive “dedicated car” service, making it the best choice for chip mass production.
A mature chip product typically follows this development path:
Design → MPW tape-out (function validation) → Back testing → Identify issues → Modify design → Full Mask tape-out (mass production).
This phased approach controls early risks while paving the way for subsequent mass production.
In the chip industry, MPW and Full Mask are not a matter of which is better, but rather tools suitable for different scenarios. Wise engineers know how to make appropriate choices at different stages of a project, validating ideas at minimal costs and pushing them to market in the most reliable way.
Next time you face a tape-out choice, consider asking yourself: What stage is my project currently in? How much risk am I willing to take? The answers will naturally emerge.
I hope this article provides valuable references for fellow engineers. May your chip tape-outs go smoothly, and may you succeed on the first try!