Trailing-edge process nodes running on 8-inch wafers can be profitably applied to a huge number of products.
By Bill Martin & Paul Werbaneth
In April, we all celebrated the 50th anniversary of Moore’s Law. While critical for the industry, sometimes “chasing-the-latest-technology” business model is not the answer. Sometimes veering away from Moore’s Law makes more sense and is a better alternative. But, ever since the 1970’s, many product developer minds were conditioned that smaller was better and homogeneous was the only option.
Recent articles are now showing older technologies being used in new products. For example, RF switches from Peregrine Semiconductor are made in trailing-edge silicon, rather than in the latest silicon process node. The market understands that two factors are pushing us to re-evaluate the technologies that are used for products. The first factor is the cost of the latest technologies, and how advanced process node cost is bending or breaking standard economic models for achieving a positive ROI. The second factor is the array of functions, or functionality, being built in to these latest products.
Whether it be analog components (PHY to support any standard interface, A/D and D/A converters, etc.) or MEMS sensors (motion, chemical, sound, light, temperature, etc.), these new functionalities are not and probably need not be designed or manufactured in the latest silicon technologies. The slogan “good enough tech for the spec” might perfectly exemplify these product strategies: use older manufacturing technologies that are cheap, but that still provide the functionality and performance required.
Removing Assumptions and Asking Why
Fortunately, a few ‘renegades’ decided to remove all assumptions and think anew. Why smaller? Why latest silicon technology? Why homogeneous? New thinking opens up options that were never previously considered.
Over the past year, several articles have been written about older semiconductor foundries shutting down their sites. This was perfectly good (and fully depreciated) equipment, running well-proven, high-yielding fabrication processes with highly experienced personnel producing material that many thought could not compete and make money, since the fabs were running on 8-inch, not 12-inch wafers.
The open question was: Would companies (or entrepreneurs) see the market potential of extending the life of older fabs and acquire or repurpose or extend the life of these sites? And who would those companies be? The answers might surprise you.
It’s SEMI member TSMC! Even while crowing (and rightly so) about new “GIGAFABS” running 28nm, and finer, fabrication technologies, TSMC continues to annually run the world’s largest number of 8-inch wafers. From the TSMC 2013 Annual Report: “In Taiwan, TSMC operates three advanced 12-inch wafer GIGAFABs, four eight-inch wafer fabs, and one six-inch wafer fab. Additional capacity comes from two 8-inch wafer fabs at wholly-owned subsidiaries: WaferTech in the United States and TSMC China Company Limited. In addition, TSMC obtains 8-inch wafer capacity from other companies in which the Company has an equity interest.”
That’s a lot of 8-inch foundry capacity. But how is it being used? TSMC says: “… Advanced technology is not the only requirement for the mobile space and for emerging applications. Enhanced requirements for human-machine interfaces are set to fuel the growth of TSMC’s specialty technology businesses in Embedded Flash, high voltage, MEMS, CIS, and RF. Most of these technologies are now embedded in our mainstream technologies, 40nm, 65nm, 90nm, 0.18μm, 0.15μm, etc. TSMC is rolling out 30 to 50 new specialty technologies each and every year to serve customer needs. … TSMC will also provide the foundry segment’s first gallium-nitride-on-silicon technology for HV soon.”
Where does that lead? Again, per TSMC: “What differentiates TSMC in the specialty technology foundry arena, besides possessing the largest capacity in the broadest scope of technology nodes, is our superior ability to integrate specialty devices. We integrate those specialty devices into our strong CMOS baseline while maintaining our CMOS IP compatibility. We will become the first foundry to offer monolithic CMOS+MEMS in late 2014. Other new integrated specialty solutions featuring low power and small form factor include: MCU+RF; MEMS+Motion Processor; RF/PMIC+Integrated Passives; Analog+DSP; CIS+Image Signal Processor; and Display Driver+Touch Controller.”
The business model for what TSMC has done with depreciated 200mm wafer fabs is in the same vein as what has worked so well for broadcast television: create cutting-edge products (180nm on 200mm wafers for TSMC; Seinfeld for NBC, back when both were fresh) and then “milk them” like crazy via syndication.
For TSMC, technology syndication is the practice of selling already developed foundry process technology, especially to more than one customer, such as an IDM looking for more capacity (STMicro?), a large OEM (Apple?) or large fabless device maker (Qualcomm?), or to a hot startup (Peregrine Semiconductor?).
The same fabrication technology process can be used to serve dozens, or even hundreds, of customers at the same time, and even more over an extended period of time, generating additional incremental revenue everywhere and every time the fab runs it, particularly when the process is trailing-edge technology (180nm process node) running on depreciated 200mm fab equipment.
That “syndication” business model is working great for TSMC, so there’s no reason to think it will change in future, sometime in 2025 say, when the trailing-edge technology is 28nm, or even 16nm FinFET, and the wafers, and depreciated process tools, are 300mm.
What Kind of Future does that Enable for the Consumer?
An amazing one! One where product development economics are reset, unleashing an amazing future of products that will rival the best Sci-Fi ‘toys’ that we only imagine today.
Gaining momentum by resetting and then thinking anew, the ITRS remodeled their roadmap not long ago to V2.0. Version 2.0 now includes heterogeneous technologies that will increase the likelihood of next-generation product success. Rather than a few ‘renegades’ driving new technologies for their own use, the ITRS will help drive these technologies and make them ‘safe’ for many to use, and thereby accelerate adoption.
And with a new focus on these new, cost-effective technologies, the marketing community has been given green fields for creatively envisioning novel products.
“Mix and Match” Technologies: the LEGO Blocks for Product Development?
MEMS functionality can add the differentiation required to make our current products better. These new, differentiated functions are the output of ‘analog’ sensors that can sample non digital and even non electrical inputs and then provide compatible input into the digital system. The digital system is solely to determine what actions are to be taken (processor); be able to store information (memory) and be able to communicate back to the external world (either to LCD screen and/or standard interface such as USB, WiFi, BlueTooth, ZigBee, etc.). In many ways, the digital system could be a platform for various MEMS functionality.
Think of the possible products that you could wear, carry or attach to your various assets. Some products arising from these new ways of thinking are already starting to appear: smart watches that monitor blood pressure, pulse and glucose levels; cell phones with air quality monitors; and monitors that allow doctors to monitor real time EKG and other human-generated electrical signals, or to monitor the chemical components in exhaled breath.
A future medical product arising from these new ways of thinking could be this: single use, battery operated mini labs that can be shipped anywhere and can test for specific diseases, provide results in minutes and transmit information locally or to World Health Organization (WHO) sites. Think of how the mini labs could be used during outbreaks of EBOLA, AIDS, Measles, etc. These products could be used in hospitals as part of their entrance paperwork. Or EMS and Police teams could be outfitted with mini labs to quickly and accurately diagnosis various conditions of immediate importance to public safety, possibly reducing the time before first medications are provided for rapidly-moving diseases, or other biological threats.
Today we think of FedEx and UPS as being able to track packages throughout their infrastructures. But when cheap 28nm CMOS on 300mm wafers becomes available, products could be available to most of us for tracking anything/anywhere. Given the electronic network (cell and WiFi) around the U.S and even the world, most assets can have ‘trackers’ attached allowing owners (and police) to track down where an asset is located. Handling privacy will be an important consideration for the Internet of Things.)
Today, authorities at an airplane crash site quickly try to locate the ‘black box’ to help determine the crash’s cause. But imagine vibration/impact, temperature, altitude, speed, etc. sensors added to every vehicle (cars, trucks, motorcycles, trains, boats, jet skis, etc.), providing streams of data available to be analyzed when accidents occur. These ubiquitous ‘black boxes’ could be used for accident reconstructions, or for determining premiums for insurance, or for establishing factors affecting warranties, etc.
Currently, the above products can be designed, but only with difficulty. Updated and new EDA tools will be required to help design, implement and analyze these products. The tools will need to work across different technology boundaries, enabling path finding, simulation, manufacturing compliance tests, etc.
The thinking may be bold, but the need is great for the semiconductor industry to move along vectors other than classic Moore’s Law scaling. TSMC has already shown a convincing business case for why trailing-edge process nodes running on 8-inch wafers can be profitably applied to a huge number of products TSMC’s customers currently run in high volume; in just a decade, with the same model, similar to TV program syndication, running 28nm processes on 300mm wafers, will result in an amazing future.
SEMICON West 2015 (July 14-16) will feature a TechXPOT on 15 July devoted to “The Evolution of the New 200mm Fab for the Internet of Everything.” As proposed, the Internet of Things encompasses a wide variety of applications and growth sectors, such as Smartphone (Mobile), Healthcare, Wearables, Automotive, Industrial, Security, etc. Many of the devices needed for these applications are not state-of-the-art technology. In most cases, the manufacturing technology has been around since the 1990’s or earlier. With global demand for these products, we see a shortage of 200mm capacity in the foreseeable future. More information about this program, and others, at SEMICON West 2015 is available by visiting www.semiconwest.org.
In addition, SEMICON Taiwan 2015 (September 2-4) offers a Secondary Market Pavilion, and SEMICON Europa 2015 (October 6-8) offers a Secondary Equipment Session.
Secondary Equipment & Applications: SEMI believes it important to support the secondary market to allow greater industry efficiency while protecting buyers and sellers. The SEMI Secondary Equipment and Applications special interest group (SEA) works to ensure global best practices and shared guidelines. Learn more: www.semi.org/en/IndustrySegments/SecondaryMarket or contact Tom Salmon at firstname.lastname@example.org.
Bill Martin is president and vice president of engineering at e-System Design. Paul Werbaneth is contributing editor at 3D InCites.