Generally speaking, an Israeli company seeking an exit should eliminate Accretech of Japan from its list of candidates. Accretech does not care for acquisitions of any kind, anywhere.
Accretech, formerly and better known as TSK (Tokyo Seimitsu), was founded in 1949 and focused on developing and marketing measuring devices for vehicle manufacturers. Listed on the Tokyo Stock Exchange, a quarter of the company’s current sales are derived from these devices, with the rest from measuring devices for the semiconductor industry, and the latter’s share is growing.
Historically, Accretech was rather late in joining the Zionist enterprise, opening a branch in Israel only in 1997, a few months after Applied Materials (Nasdaq:AMAT) acquired Israeli company Opal Technologies to establish its development center, and KLA-Tencor (Nasdaq; XETRA:KLAC) followed suit. Accretech Israel general manager Gilad Golan says Accretech had already planned an Israeli venture before Applied Materials’ acquisition, and planned to establish its own development center, without acquiring an Israeli company.
Since then, Accretech has scrupulously taken care not acquire competitors, preferring to invest in in-house development. “I actually consider this a strong strategy,” says Golan. “We expect to complete our current project at half the projected budget, because it is based on previously developed technology, adapted for the new product.” Even so, part of the project was carried out by sub-contractors or based on technologies developed by other companies that lacked intellectual property rights.
“I have a given amount of time to produce the best product possible,” explains Golan. “Obviously, it’s sometimes better to use someone else’s technology and know-how, and I have the authority to look for technologies related to our developments, but the corporate policy is not to seek acquisitions, but technology partnerships.” Accretech established such relationships with DSI - Decision Systems Israel and MicroSpec Technologies (which was acquired by optics manufacturer Carl Zeiss).
“We are not afraid to compete against major companies. KLA-Tencor essentially has a monopoly in the market for which our new product is designed, with over 90% share. Then, up pops Accretech, in cooperation with Carl Zeiss to develop a new product that neither of their competitors – Applied Materials and KLA-Tencor – possessed. One had dark field wafer inspection technology, while the other had light field technology. We were the first to develop light field wafer inspection technology with higher resolution lens capability that could locate defects in specific wafer layers.”
An explanation follows this declaration: Wafers are layers of silicon from which semiconductors are cut. Wafer production is assisted by lithographic process, which stamps the semiconductor’s design onto the wafer. Defects occur throughout the production process, both in design (such as unsuitability between design and raw materials) and during manufacturing.
Finding defects in memory chips is relatively straightforward. Since they are used to store future information, wafer surfaces are even. Finding defects in chips intended for data processing is more complicated, because wafer surfaces are uneven.
Accretech is not readily found in US business magazines, even under its previous name, and despite the fact that all semiconductor manufacturers are potential customers. Golan says that 2001 was a hard year for most of the semiconductor industry, with most customers coming from Japan and Korea, but substantial business is anticipated from the US in 2002. The situation induced Accretech to publish a profit warning in early March, revising downward its November 2001 sales forecast for the year ending in March 2002 from $280 million to $230 million. The worsening business performance also forced the company to cut its staff from 1,100 to 650 last year.
The cost of each inspection device is $3-5 million. They test an average of four wafers an hour, with deviations depending on the size of the defect to be detected, which determines the time needed for inspection. “Due to the length of time needed for inspection, companies do not test every wafer, but only a sample. Assuming a block of 25 wafers, and one hour of inspection time, the manufacturer will test four of the 25 wafers, i.e. less than a sixth. But that can vary between companies and projects,” says Golan.
The beta device currently being tested at Accretech can test wafers with a thickness of 45 nanometers; in physical terms this means a thickness of 100 silicon atoms. Current chip inspection technology can handle 0.13 microns (130 nanometers), which can be drawn out to 0.11 microns. Based on current projections of semiconductor development, Accretech’s devices will only being testing wafers in 2009.
Since its founding, $40 million has been invested in the Israeli center’s flagship product, Wafer Inspection, in addition to $60 million invested in the project in Japan. The software and processing components are developed in Herzliya, while the mechanical and electronic components are developed in Japan. Since the Israeli plant essentially sells its products to its parent company, it is unsurprising that its $17 million in sales in 2000 won the company the Outstanding Exporter to Japan Prize for that year.
The development center is now working on its next project: a next-generation mask inspection device for lithographic technology. Current production methods use “steppers” – the nickname for the large machine that carried out the lithography procedure – and “masks” - high precision quartz or glass substrates imaged with microscopic structures and used to etch patterns onto silicon wafers. The current leaders in stepper manufacture are Nikon (LSE:NKN; XETRA:NKNG) and Canon (NYSE:CAJ; LSE:CNI; XETRA: CANN), known for their cameras, and ASML Holding (Nasdaq; XETRA; AEX:ASML) of the Netherlands.
Over time, the masks wear out and/or become contaminated and are replaced. The masks themselves have to be tested, both during production (to detect manufacturing defects), and during use to test chips (to detect wear and contamination). This is where mask inspection devices come into the picture. “Unlike wafer inspection, which is statistically based on cost and time constraints, there are no short-cuts for mask inspection. Every damaged mask is unusable and has to be thrown out, so they must all be inspected. On the other hand, compared with wafers, there are very few masks.
Technological development and its consequent demand for miniaturization has led the semiconductor industry to consider next generation lithography. “One possibility is to continue using lasers, but at some point, they will reach the end of their capabilities,” says Golan. “Although there are lasers that can etch lines measuring 150-190 nanometers, that is the limit of their capabilities. Moreover, lasers cost $20 million.”
Another option is to use xasers (x-rays lasers). These use focused x-rays at different wavelengths and strengths than ordinary hospital x-rays. However, this is a very costly option. Intel (Nasdaq:INTC), for example, has invested over $1 billion to develop a UV Extreme system, using wavelengths between ultraviolet and x-rays. “So far as we know, mask systems using this technology will cost about $1 million,” asserts Golan. “This may be worthwhile for companies like Intel that manufacture masses of chips, but smaller companies cannot make such an outlay.”
There are also more esoteric methods, such as ion beam systems.
In June 2000, Accretech decided to enter the next-generation stepper field, and established a consortium Low Energy E-beam Proximity Projection Lithography (LEEPL), in which it has 67% ownership. The project is intended to develop steppers using electron beams. Nikon has a similar system, but Golan claims the low energy produced by LEEPL’s beam prevents it penetrating too many wafer layers. Golan says the LEEPL project is already close to producing chips at 50 nanometer technology (0.05 microns), even though demand in 2003 is for 100 nanometer technology. A fringe benefit of the method is that the mask created by the process becomes the printing block for subsequent masks, saving mask production costs. *
Accretech Israel is developing the system’s software and algorithms and hardware imaging processor components. Some of the component integration is also carried out in Israel, while the integration of the production processes is done in Japan.
In contrast to earlier projects handled by the company, the next-generation lithography inspection and manufacturing project will be developed in close cooperation with industry. Accretech will provide only some of the systems that the industry defines. The solution was to establish a consortium, led by Sony (which is expected to be the venture’s largest customer) and 20 other companies, including Sharp, Matsushita and NEC. In addition to being official partners in the consortium, some of the companies are also investing in joint development processes.
Accretech Israel has 40 employees, 20 less than a year ago. $8 million a year is being invested in LEEPL, and the devices are expected to be sold for $9 million each, compared with $25 million for existing inspection devices.
Golan joined Accretech from Opal Technologies about the time it was acquired by Applied Materials. “I have no regrets, and I could go back if I wanted. I don’t know how much freedom Applied Material’s Israeli branch has to operate outside the parent company’s work plan. We are running two projects of our own initiative. One is technology project for the existing market, and the other is seeking a solution for technology developed by others, but which we think can be adapted to the semiconductor industry. We are currently testing the business feasibility,” says Golan. Accretech approved a $1.5 million allocation to its Israeli branch for these projects, which might involve Israeli academic institutions in addition to Accretech Israel itself.
Published by Globes [online] - www.globes.co.il - on April 29, 2002