Why was one of the major technological success factors behind ASML, the electric table, not an integral part of the deal between ASM International and Philips?
To this day, the electric table, the superior alignment technology and Zeiss’ lenses and EUV mirror optics are ASML’s technological bedrock. Its super-fast, super-precise stage and alignment have been unique selling points for decades, the reasons why chipmakers buy the Dutch lithography machines. The electric table makes ASML’s steppers the most productive in the world, and the alignment system ensures that despite that high speed, everything runs at ultra-precision.
This makes it all the more remarkable that the joint-venture deal between ASM and Philips in 1984 didn’t include the electric table. Instead, the deal was based on the PAS 2000 wafer stepper, which was unsellable due to its oil-based drive system. Why did Philips’ engineers cling to the old technology for so long, while the electric stage was available as a prototype at Philips’ Natuurkundig Laboratorium (Natlab)? That may be the greatest mystery from ASML’s early days. The seventeen hydraulic clunkers that were transferred to Veldhoven were in fact white elephants – and Philips S&I knew that. ASM was being taken for a ride, but to be fair, it didn’t inspect the deal closely enough, either.
Liability
In 1979, S&I business unit director Wim Troost sent the edict to build an oil-based machine. Attempts to construct a mechanical stage together with Philips CFT had failed and he wanted to save the delivery to chip heavyweight IBM Burlington by making a ‘Chinese copy’ of the wafer stepper using the original hydraulic stage that was developed at Natlab. With the caveat: we need that electric table, but we’ll do it later. Delivery to IBM was of higher priority.
The oil-driven stage is a great example of an early advantage turning into a liability. When Ad Bouwer built the first stepper stage at Natlab in 1973, Natlab’s oil-based precision technology had an excellent reputation. Frits Klostermann had used it to make his photorepeater, a device that was far ahead of its time. Philips’ fabs had manufactured chips with it for years. Why would S&I’s engineers develop a new drive system when they had something that had worked well for years?
Device lifecycle management for fleets of IoT devices
Microchip gives insight on device management, what exactly is it, how to implement it and how to roll over the device management during the roll out phase when the products are in the field. Read more. .
Wim van der Hoek and Rien Koster, both luminaries in the Dutch construction and precision world, had devised an alternative wafer table based on spindles at Philips CFT, but that development had failed miserably. The stage “rang like a bell,” which isn’t a positive thing in precision technology – resonating vibrations make it impossible to image microstructures. The result: everyone clung to the oil-based drive system.
Call it a lack of realism, tunnel vision or conservatism; the fact is, this decision meant ASML had no income for its first two years. But the company was saved by a happy accident. In late 1984, a deep crisis in the chip industry brought progress to a standstill. If the machinery market for chips had thundered on, ASML wouldn’t have been able to sell anything in a growing market and it would have been dead as a doornail.
No mention
After ASM and Philips had decided to partner in the spring of 1983, S&I cranked out several bulky reports. They all ended up on the desk of Arthur del Prado, ASM’s founder and CEO, as he was preparing for his litho partnership with Philips.
In June, a fresh litho group leader, Jacques de Vos, diplomatically wrote that the “hydraulic motors are an expensive part of the machine” and that the hydraulic table was to be replaced by a table driven by linear motors in 1984. This would allow them to ax the power-guzzling generator required to keep the oil pressure constant. “With the removal of the bulky hydraulics, the machine will also take up less floor space, which is a huge plus in cleanrooms, where every square foot is valuable,” he stated.
De Vos’ report didn’t mention that hydraulics are unacceptable to chip manufacturers, nor did the business plan that reached Del Prado three months later. In that document, S&I’s Ger Janssen said almost nothing about the downsides of hydraulics. By then, the disadvantages of oil had been known for years, but the document implied that S&I didn’t see it as an insurmountable problem.
One might think that this fact needed to be kept hidden from ASM, but that doesn’t seem to be the case. S&I’s engineers simply weren’t looking outside their own building; they’d never heard of product marketing, they never talked with customers and suffered from a disease that was widespread at Philips in those days: the illusion that superior technology sells itself.
The 1983 business plan did say that hydraulics wouldn’t be able to meet the required precision in the future, mentioning the development of a linear magnetic table as an upgrade to the PAS 2000 slated for 1984. As reasons to look into a new wafer stage, it cited the hydraulic system’s high cost and the extra floor space the electric generator took up. It didn’t mention that oil was unacceptable in chip fabs and would make the system unsellable – if only because the hydraulic pressure pumps made so much noise that they’d need a soundproofing room around them that was bigger than the stepper itself.
The necessity of replacing the hydraulics tout suite apparently didn’t register for Arthur del Prado and his technical right-hand man, Willem de Leeuw, who had just arrived from the aerospace industry and didn’t have a clue about chip lithography. The 1983 business plan listed the electric table purely as a future upgrade. The joint-venture contract they signed six months later noted that the electric table had been shelved at Natlab. Yet, the component fell outside the agreement.
No idea
It seems that ASM didn’t realize the importance of this technology – and didn’t want to. In their policy meetings with Philips, Del Prado and De Leeuw talked with George de Kruiff mostly about ramping up the sales and marketing efforts. Troost agreed but warned that ASML would then be emphasizing the sale of an oil-based machine with a dubious lens (from French Cerco; Zeiss didn’t support the Dutch litho developers at that time) and an insufficiently strong service organization.
ASM was the joint-venture partner that enjoyed bragging about its sales abilities. But in 1983, the company was completely blind to the fact that oil was a showstopper in the chip industry. Steppers were hot that year, and unaware of the issues, Del Prado and De Leeuw piled up the pressure to manufacture oil-based steppers. The CEO, especially, should have known better. There’s nothing that indicates that ASM did any research into chip manufacturers’ acceptance of hydraulic steppers. They relied on Philips, but their partner didn’t dive into customer needs at all.
At S&I, no one had taken the trouble to talk with customers in the preceding years. Faith in its own technology reigned supreme. As far as marketing and sales were concerned, steppers to them were no different from light bulbs. In 1983, high-precision positioning systems based on oil were still superior. But it’s also fair to say that even IBM’s chip fab in Burlington, which took S&I’s first and only machine for in-depth evaluation, did highlight oil as a problem but not as a show stopper in its evaluation reports.
S&I’s failure at that time: after receiving IBM’s review, the stepper team didn’t contact Burlington to discuss Big Blue’s wish list. The Eindhoven stepper team had a beer party, returned to their factory and waited for the orders.
In 1982 and 1983, S&I’s engineers did ask, time and again, for Natlab’s electric table, but there was no money to continue its development. It didn’t help that the lab’s scientific egos clashed with the practical egos in the product division.
In late 1983, Troost, who had the steppers in his S&I portfolio, reported to his management his doubts about ASM’s rush to ramp up production of the oil-based steppers. But his boss De Kruiff, who knew little about chips himself, was certain that the veteran supplier to the industry knew what it was doing. In truth, ASM had no idea.
No sense of urgency
Shortly after arriving at the joint venture ASML, the first CEO, Gjalt Smit, had his first discussions in March 1984 with the senior engineers who would be joining his team from Philips. It quickly became clear to him that a hydraulic table was unacceptable. Still, the management team thought that chip manufacturers would want to purchase the hydraulic stepper to develop their next-generation chip processes. In those first months, Smit didn’t get the information to give him the needed sense of urgency to completely ban oil from the ASML systems.
ASML management team members Richard George and Joop van Kessel did point out to their director that there was a promising electric table at Natlab, but it was still no more than a proof of concept. No one knew its value, whether the electric motors were precise enough or whether they could be mass-produced. Strikingly, the H-table (originally designed with a hydraulic drive system) had been patented, whereas its successor with the electromagnetic linear motors had not – Philips’ patent department had set that proposal aside.
In Natlab’s optical group in the 1970s, the stepper had to compete with the compact disc for the patent attorneys’ attention. Philips had assigned a much higher priority to the consumer audio system. That was the right choice at the time, but that focus continued well into the nineties and left ASML with an initially weak patent portfolio. Nikon seized that opportunity in 2000 to start a patent war. In contrast, ASML’s other competitive advantage, the alignment technology, had already been patented to the max in the early seventies.
Smit’s lacking sense of urgency becomes evident from the business plan he drew up in his first months and presented to his management team and supervisory board in August 1984. “The present PAS 2000 has still unrivaled basis characteristics. Less fundamental disadvantages with respect to competition (oil table, 6”, LFL) do make it unsellable in larger quantities,” he wrote, putting the oil drive in the same category as the size of the stage, which had to accommodate 6-inch wafers. “There exists still a narrow but exploitable time window to correct these crucial drawbacks.”
In the same plan, Smit even listed twenty oil-based steppers as a sales goal for the following year. Around that time, ASML also ordered materials so it could manufacture ten machines on time, albeit with an option to cancel on 1 January 1985. Smit’s business plan did, however, clearly state: “No PAS 2000 sales after 1985.”
The illusion that the PAS 2000 could still be sold was possibly reinforced by the purchase of two of these steppers by Philips’ fabs in Nijmegen and Hamburg. Despite complaints, the two sites were actually using the machines, and ASML regularly dispatched service engineers.
Clear as day
In November of 1984, Smit got a ruthless wake-up call when he paid a visit to potential customers in the US. They told him that hydraulic stages were a total no-go. He rushed back home to inform his team that oil was definitely off-limits.
All those involved finally realized the choice facing them. Only then did the curtain decisively fall on hydraulics. Only then did Frans Klaassen get the assignment to develop an electric table like a bat out of hell – a task upon which ASML’s continued existence would hinge. Only then did Frits van Hout, the young engineer who was given the R&D responsibility for the oil-based PAS 2000 in the company’s first months, hear that his project had truly reached its end. Evert Polak, dressed as Santa, told him the news at the company’s Christmas party: “Frits, your oil days are over! It’s finally been decided.”
Once the electric drive technology showed what it was capable of in 1985, it was clear as day: this was the way to go. In the years that followed, the electric table gained a massive reputation. It was the force that drove the high throughput: semiconductor manufacturers could make more chips using ASML’s steppers than using Nikon and Canon machines.