Innovation plus scale-up needed to tackle the complexities and cost of 450mm

Innovation plus scale-up needed to tackle the complexities and cost of 450mm

by Debra Vogler, Instant Insight Inc.

Suppliers were asked to share their thoughts about tackling the challenges, as well as taking advantage of the opportunities, as they forge ahead with the 450mm transition.

It’s fair to say that the 450mm transition is one of the most expensive and complicated endeavors the industry has had to navigate; some might even say it’s the most complicated undertaking. Though estimates of the total cost have been all over the map, Applied Materials has been using $15-20Bn as the estimate for the industry as a whole. “What we’re also trying to do is ensure that this estimate is wrong and way too high!” said Kirk Hasserjian, corporate VP, Strategic Programs at Applied. “To do that, we’ll need a lot of collaboration and coordination with our customers [that are going to 450].”

It’s clear that the way the transition is implemented will have an affect on how its cost will impact the supplier community in terms of ROI. Hasserjian maintains that a synchronized transition is key. “You can’t really coordinate the [technology] nodes, but if you could get a group of companies to transition to 450mm at about the same time, it would be helpful,” he said. For example, if three or fourKey Factor #1 Image companies built 450mm fabs all within about a year or year-and-a-half of each other, that would provide volume orders for equipment suppliers (See Chart: Synchronized Transition). That volume, in turn, will help enable a reasonable ROI so that R&D expenses - especially those to be incurred over the next six years - can be recouped. A piecemeal transition will result in poor ROI for the equipment suppliers. “In the end, what we’re trying to do is make sure we have a sustainable, vibrant semiconductor industry,” said Hasserjian 

It will take more than money to make the 450mm transition happen, however. Hasserjian breaks the effort up into two major areas. The first is meeting the technology nodes during the transition, which will require innovation in the hardware design, process characteristics, and the materials that are used. This is one consideration that should give the industry pause concerning the level of effort equipment suppliers will need to put forth. Hasserjian pointed out that the industry doesn’t yet understand the requirements that will be needed to get to 10nm and below on a 300mm wafer, let alone on a 450mm wafer. “The uniformity of the thin film layers, the uniformity of dimensions that we are producing, the electrical characteristics across an entire wafer - these will be done at very advanced technology nodes (10nm and below),” noted Hasserjian. “And today, we’re just tackling 20nm technology in manufacturing and working on R&D for 14nm.”

Mark Fissel, VP 450mm Program, at Lam Research Corporation, reiterated concerns about the risks that come with not knowing the technical requirements for sub-10nm processing. “We’ll have to address 300mm and 450mm development at the same time, so we will have dual R&D requirements,” said Fissel. The parallel efforts mean having two sets of teams working on the two wafer sizes - a huge cost factor.  Additionally, Fissel, noted that the parallel efforts will have to go on for a much longer time to bridge the gap between the two wafer sizes. The lack of 450mm wafers, as well as their cost, only add to the risk burden. Still, he acknowledges that this time, the industry is approaching the transition in a more rational way. “There is more collaboration trying to agree on a timeline,” he said. “And standards have come out in a much more rational time frame.”

Regarding tool costs, Hasserjian said that the second major area to address is the innovation required to enable tools to handle the larger wafers without the tool costs going over 2X the cost of 300mm tools. “If we just scale up, the probability of the tool cost being so high that it just doesn’t make economic sense to go to 450, I think, is very real,” said Hasserjian. “So you need innovation along with the scale up.”

As a way to ease the transition to 450mm, TEL has offered the industry use of its open Common Platform. “The hope is that the availability of a common platform will encourage innovative solutions from creative folks in the industry,” noted Akihisa Sekiguchi, VP & GM, Corporate Marketing at TEL. While there is no news about whether or not any suppliers have taken TEL up on the offer, Sekiguchi said that the company has had opportunities to discuss the concept with other tool suppliers.

TEL intends to unify its platforms at 450mm on the Common Platform, providing easy process module integration for future extendibility. The company has set itself a target to be able to complete integration on the platform in one month, which would require a “clear articulation of interfacing details including hardware, software, and utilities,” explained Sekiguchi. “It would not be an easy task, but we have years of experience integrating multiple chamber types to our platforms.” He also said that the company’s development roadmap for the platform is consistent with end users‘ HVM timing. 

TEL’s commitment to standardization is further evidenced by its efforts at reducing tool and chamber installation costs by 50%, as well as reducing tool installation and upgrade times by 25-30%. “The general concept is that the physical layout of the utilities, e.g., where and how the gas lines, power lines, and other utilities interface with the tool, should be standardized,” noted Sekiguchi. “Imagine us having different power sockets for all our household appliances and/or plumbing - the cost would be prohibitive.”

Leveraging pre-competitive efforts; addressing facilities specs

Besides the many opportunities for innovation along with standardization, suppliers are recognizing ways to leverage pre-competitive efforts. For example, Hasserjian observed that there is much that can be done in areas that do not have unique IP in OEM equipment. “We all use gas sticks, valves, power supplies, and pumping systems,” he said. “There are many different components for which we could have much better standardization when going to 450 and we could have better collaboration on R&D to develop these sub-components that are wafer-size dependent.”

Adding to the list of noncompetitive components/subsystems that would most likely benefit from commonality - and reduce tool costs - among the equipment suppliers, Fissel includes examples such as gas box and liquid delivery system components, mechanical pumps, heat exchangers, and standard facilities interfaces. Service commonalities will also be useful to address given the fact that 450mm tool sets will be much larger than those for 300mm. Fissel also observed that common methods of getting components from their shipping crates to the systems will be needed. The real challenge in pursuing all of these efforts will be accomplishing them within the timeframe needed to support the fabs’ requirements. With the earliest IC manufacturers expecting 450mm pilot lines to come up in 2016, followed by volume manufacturing lines in 2018, Fissel says there needs to be movement on the issue of component commonality fairly soon.

Addressing the inefficiencies associated with facilities specifications poses still another opportunity for suppliers (both equipment and facilities suppliers) to ensure that the economics of going to 450mm are not compromised. The chart below (Utility Consumption Drives Investment) illustrates how utility consumption drives investment. According to Allen Ware, VP, Program Utility Consumption SlideExecutive for 450mm, at M+W Group, a 10% increase of equipment power consumption relates to an overall increase of ~2% for base build capex.

Ware further explained that when equipment suppliers draw up their specifications on utilities (electrical, ultra-pure water, gas volume, exhaust, etc.), they build in worst-case values. But then so do the fab owners when all the inputs from all the tool suppliers are analyzed. And then the facilities suppliers build in their own buffer. “So when you start looking at the total impact of taking the worst case along with everybody who touches it, you end up with a facility that costs tens of millions, if not a hundred million more dollars than it needs to be - especially given the size of a 450mm fab,” said Ware. “If we could place instruments on tools in pilot lines and measure such things as idle time, peaks, etc., and come up with a data-driven utility matrix, we could build a more efficient factory.” And though this scenario has not yet occurred even on 300mm fabs, Ware is hopeful that 450mm will be different.

A new facet to the 450 transition is that, unlike previous wafer transitions, facilities developers will now have an opportunity to directly interface with equipment suppliers rather than wonder what attributes tools being delivered to a fab will have when they arrive. “During the 300mm wafer transition, suppliers had to customize products for 15+ unique semiconductor manufacturing companies,” said Ware. “Customization not only drives up fabrication cost and lead time, but also results in multiple variation of spare parts with longer lead time and cost.”

With respect to the difficulty of estimating the costs associated with going to 450mm, Ware noted that, “300mm fab sizes are all over the board, so there’s no direct correlation for transition to 450mm” explained Ware. ”And we have to keep in mind that most 300mm fabs will be retrofitted for 450mm.”   

Figure captions:

Figure 1. Key Factor #1: Synchronized Transition. SOURCE: Applied Materials

Figure 2. Utility consumption drives investment. SOURCE: M+W Group