UMC News

40 nm production line comes fully into operation, putting the company’s true value to the test


Michiari Kawano, Director and Corporate Senior Vice President

It has been a little over a year since Mie Fujitsu Semiconductor Limited (MIFS) was established at the end of 2014 as a pure-play foundry company. At the beginning of 2016, a production line dedicated to 40 nm process technology will come in full operation. For the company, 2016 will be a year that its true value as a foundry company will be put to the test. Director and Corporate Senior Vice President Michiari Kawano talks about the company’s future business strategies, which center around this 40 nm production line.

40 nm products to start shipping in 2016

It’s been a year since your new company started, hasn’t it?

Michiari Kawano, Director and Corporate Senior Vice President

We started out as a wholly-owned subsidiary of Fujitsu Semiconductor (FSL) in December 2014, and have been building the systems and functions we needed. In the first half of 2015, we moved into the black. This trend continued during the second half of the financial year, and we expect our first year’s performance to exceed the plan we made at the start of the year.
To become an independent pure-play foundry company, it was determined that we would receive a total of 10 billion yens in funding from UMC, a Taiwanese company. Based on the contract for this funding, we received the first 5 billion yen in March 2015 and the remaining 5 billion yen at the end of December that year. This means that that UMC’s investment accounted for approximately 16% of MIFS’s entire funding.
Apart from the funding we received from UMC, we also received 40 nm process technology from them and introduced production equipment at our Mie plant’s Fab B2. Until then, MIFS’ production facility was able to handle up to 55 nm processes, so this is the first time we have installed a production line for 40 nm. Products from this line will start shipping in 2016.

Some of your competitors in the market are large-scale foundries, aren’t they?

Out of all areas in the semiconductor industry, foundries are one area from which you can expect a high growth rate. We think that our choice to go into this business was the right one. The Mie Plant had undertaken foundry projects since the time we were part of FSL, and our foundry business accounted for a large percentage of the plant’s work. We decided to become independent as a pure-play company, because we had the resources to continue with our foundry business.
Basically, we will be focusing on semiconductors for IoT and automotive semiconductor devices. In order to do this, we have developed and built up expertise in ultra-low power process technology, embedded flash memory technology and radio frequency (RF) technology, including the millimeter band. We would like to offer these three technologies to our clients as a set to improve our value as a company.

Pushing forward with ultra-low power technology for the IoT

In April 2015, you acquired from SuVoltain the USA the IP rights for the low power CMOS technology PowerShrink. What was the intent behind this?

During the time with FSL, we obtained a technology license from SuVolta for PowerShrink, which we had implemented for our 55 nm process technology. PowerShrink is an ultra-low power CMOS technology that features DDC (Deeply Depleted Channel) CMOS transistors with suppressed threshold voltage (Vt) variations and heightened carrier mobility. By using this technology, we can cut down operational power consumption of ICs by up to 50% without compromising performance.
In the future, lower voltage and lower leakage devices will be key for IoT applications. We have developed new DDC transistors with the leakage current reduced by two orders of magnitude. This DDC transistor is perfect for sub-threshold circuits, contributing to the lower power consumption of IoT devices.
We will be applying this advanced DDC technology to the 40 nm process. We are enhancing it with embedded flash memory (2.44 MB ). Although this is a new challenge for us, we have already completed the key parts of this development.

40 nm line coming into full operation in 2016

What is the present status of the 40 nm process production line, and what lies in the future for it?

The Mie Plant which installed the 40 nm production line With the technology from UMC, we introduced a compatible 40 nm production line at Fab B2 in our Mie plant. The building has two floors of cleanroom space. We moved the immersion exposure tool to the upper floor in December 2014 to manufacture prototype chips. For mass production, we have started installing equipment in a new cleanroom that takes up half of the first floor of the building.
The monthly production capacity (the number of 300 mm wafers processed) of the new line will be approximately 3,000 in 2016, and we plan to increase that to 5,000 by the end of 2017. As with our 55 nm process products, we will use UMC’s fabs as backup when necessary. Depending on the business outlook, we will consider new investments to expand our production capacity.
As for the 40 nm technology, we are currently verifying its reliability. Commercial production will start at the beginning of 2016, with mass production starting in the second half of the year. We already have contracts for multiple projects for consumer electronics and industrial devices. These contracts are mostly with clients we have known since our time with FSL, when we were working on 55 nm processes. We have strong relationships with these clients, who trust us greatly, and they fully understand our Mie plant’s performance.
There are increasing demands for the use of 40 nm process technology with embedded flash memory. We started out late as a 40 nm foundry, but many clients from the United States, China and Korea are making such demands. As for IoT, we are currently in talks with startups from the United States. We will continue to develop new clients.

Saving power with the new air-conditioning system with an easy-to-change layout

Tell me about the new air-conditioning system you introduced into your new cleanroom.

Concept design of SWIT (Swirling Induction Type HVAC System) We made a full-scale introduction of Takasago Thermal Engineering’s ” SWIT (Swirling Induction Type HVAC System)”.SWIT causes the thermal stratification of air within the cleanroom, carrying airborne fine particles into the upper part of the room with the rising warm airflow from heat sources such as production equipment. It can reduce the amount of air supplied compared to conventional down-flow systems. We have another test area, away from the new cleanroom, where we piloted this SWIT system to collect various data. These data revealed that we can reduce transport power by approximately 47% and heat-source power by approximately 32%, meaning this system is cost-competitive, energy-saving and thermally efficient. The cleanroom space can also be easily expanded by simply installing units by the wall. Installation is relatively simple and the layout can be changed easily. The costs involved in introducing this system are also low.

Are you thinking about adopting 28 nm process technology?

Not at the moment. When we move into the 28 nm generation, there will be many technological hurdles like High-K/metal gates. We will need new production facilities and will need to increase our investments. Also, ICs that will require production techniques for the 28 nm generation and beyond have a limited scope of use. In reality, ICs manufactured with the 0.18 µm process technology are being used for a variety of purposes. We think that most things can be done with ICs produced with process technologies of 40 nm and below.

Enhancing manufacturing capacity of existing lines

What is the operation of your existing lines like?

They have been in full operation for about a year and a half now, backed up also by the recent trend of yen depreciation. As I mentioned earlier, our profit has exceeded the levels we had forecast at the time of the company’s establishment. As the lines are at full capacity, we are faced with some challenges. We could inconvenience our clients when it comes to delivery, for example. If there is a production line problem, it would be extremely difficult to us to make up for the lost production.
We overly focused our management resources on developing the new 40 nm process, so we misinterpreted the supply-demand balance for 55/90 nm products and lagged behind when making production-expansion decisions. That is something we have to review thoroughly at the management level. We have already completed investment to eliminate production line bottlenecks and to improve production efficiency. The production capacity of the 55/90 nm process lines is scheduled to expand from the beginning of 2016, so we expect to resolve our delivery issues by spring 2016.

Actively advancing porting projects

How do you plan to acquire new clients?

For embedded flash memory technology, an area where we would like to differentiate from our competitors, we have developed Plug-In Flash. Plug-In Flash is logic process compatible and can be manufactured with little impact on the characteristics of existing devices. Because it doesn’t require many additional process steps, its manufacturing costs are low. It also enables customers to use existing libraries available in the baseline logic technology. We would like to combine this newly developed flash memory technology with ultra-low power DDC and RF technologies as a platform to acquire new clients. Our other business is porting projects, where we port the client’s technology to the production line of our Mie plant for manufacturing. We have already closed a major deal and have started installing a dedicated production line. We are tuning the production equipment to deal with the client’s specific analogue and high voltage process technologies.
We have carried out individual porting projects before per clients’ requests. In the future, we will actively develop this as a business.

How is your partnership with UMC(2.02 MB ) going?

We are currently considering adding value to UMC’s 40 nm process technology (40LP) by integrating it with ultra-low power CMOS DDC technology, for example, and giving feedback to UMC. We are also good at adaptations to automotive specifications. We will consider all these possibilities and improve the 40 nm process technology./改善に取り組んでいく。
UMC is building a new plant in Xiamen, China. For clients needing ICs made in China, we can cooperate with UMC and respond to client needs at this new Chinese plant.

Combining the expertise of an IDM and of a pure-play manufacturer

Are there any differences with regard to plant operation between you and UMC?

We are benchmarking with UMC concerning production lines. There are a lot of things to learn about operational stability and quality maintenance that we would like to introduce at our Mie plant.
On the other hand, for many years, MIFS (FSL) has been in business as an integrated device manufacturer (IDM). We understand the difficulties faced by upstream circuit designers, and we can also give appropriate advice to our clients. We would like to maintain these strengths while building on the knowledge about plant operation and production management from UMC./生産管理に関する部分を強化していきたい。

What are your main management challenges for 2016?

Michiari Kawano, Director and Corporate Senior Vice PresidentWe would like to establish a robust production line for 40 nm process technology to deliver products to our clients. We will resolve the supply issues with the existing 55 nm production line. Currently, we are still not up to par with regard to certain features and systems when compared to competing foundries. We would like to establish a solid business base as a foundry company as early as possible, based on production quality and secure delivery systems.
To achieve wider recognition as a foundry company, we have joined GSA (the Global Semiconductor Alliance), which includes 400 semiconductor value-chain companies from around the world. At GSA we would lobby to acquire new clients and actively interact with other companies.

Do you have any unique services to differentiate yourselves from your competitors?

Well, this is just one example, but RF circuit design requires extensive engineering knowledge and know-how. You cannot just get the necessary libraries ready; you need to understand and confirm how your clients want to design the circuit, and if they so request, offer them extensive support with the technologies and design techniques required to build the circuit. We are also collaborating with companies from the Fujitsu Group and other developer organizations in and outside of Japan to start manufacturing millimeter band MMICs (Monolithic Microwave Integrated Circuits) with CMOS technology in 2016.

It is also important to amass unique IP cores and libraries, isn’t it?

You can load up to 8 Mb of memory in the Plug-In Flash currently being developed, so it will be used in products like IoT devices where sensor data need to be saved. ICs using this technology will start being manufactured in 2016 utilizing 55 nm process technology. Products using 40 nm process technology will be at the center of our business, and we will start to make preparations for mass-production in 2017. We are also planning to develop chips with embedded flash memory on logic circuits using DDC technology.

A+ A-