New materials - new business model?
08/01/2005
Solid State Technology asked industry experts to comment on the pressures being exerted on the materials business model.
As industry grows, new business models needed
Phil Dembowski, Dow Corning, Midland, Michigan
 Phil Dembowski |
Ever since IBM introduced copper interconnects in 1997, suppliers and chipmakers alike have been keenly aware that device scaling cannot continue without materials innovation. That’s why the industry plunged headfirst into low-k materials development. Yet most suppliers who invested in low-k early have not seen any revenue, and even those who have seen some return have not recovered their investments in full.
The low-k experience is just the first example of the business challenge facing the industry; the International Technology Roadmap for Semiconductors now calls for the introduction of one to two materials per technology node for the foreseeable future.
The high risks in materials development discourage innovation and investor enthusiasm. Most major materials suppliers today are multimarket companies with activities in industries beyond semiconductors, such as plastics, pharmaceuticals, or consumer products. They evaluate business opportunities against a minimum return on investment (ROI) benchmark. When business directors and CFOs at these companies don’t see a realistic opportunity to achieve that ROI, they are more likely to gravitate toward the less risky business propositions in their portfolio of opportunities.
Suppliers’ experience with low-k materials development suggests that sales-based business models - the predominant method of doing business in the chip industry - afford unattractive project economics; these models are a poor fit for the industry’s ongoing materials development needs. It’s time for the industry to find new models that make the business of materials development feasible.
Collaborate early. Any new business model must be based on early collaborations between end users and suppliers. Successful collaborations allow materials suppliers and IC manufacturers to have an open and honest dialogue regarding performance issues, alternative approaches, and the realistic prospects for adopting a material.
Focus on what you do best. There are two main problems with the typical approach to new materials integration whereby end users expect a complete solution:
- Each customer’s process and integration scheme is very different, and solving the puzzle one time does not guarantee success the next time
- Most materials suppliers are not experts in integration, device design, or many of the other areas it takes to make an IC work from top to bottom.
New business models should be designed to let each participating company focus on what it does best. By working together and sharing expertise, IC manufacturers, equipment makers, and materials suppliers can find answers faster and (likely) more economically than if each party tries to do it all alone.
Be willing to share in risk and reward. The development effort for a new material typically needs to start 5-8 years before the technology is introduced into manufacturing. The supplier faces a huge technical risk: Can it develop the right product with the right performance for the application? There are also major market risks: Will the technology still be relevant, and will end users delay implementation?
A better approach is when the risk and reward are clearly defined and balanced among the companies involved, giving everyone a fair shot at achieving their desired ROI. This can be achieved in many ways: technology licensing, precompetitive R&D using pooled resources, contract manufacturing agreements, partnering and royalty deals, outright sharing of development costs up front, joint development programs, and pay-for-play programs (e.g., where chipmakers pay for materials samples).
At Dow Corning, we’ve begun to implement some of these new approaches. The initial reaction from end users has been mixed. Some are open to new ideas; others are cautious. Change comes slowly, but the outlook is promising. More materials suppliers are ready to drive partnerships with chipmakers, toolmakers, and other suppliers, according to a survey conducted by Kline and Co., Little Falls, NJ.
Innovation is what fuels the semiconductor industry engine, but individual companies that support this industry also need to focus on the bottom line and ensure that they are choosing attractive investment opportunities. If we don’t, we miss an opportunity to safeguard the pace of innovation - and the long-term health of this industry.
For more information, contact Phil Dembowski, global market manager, semiconductor fabrication materials, at Dow Corning, 2200 W. Salzburg Rd., PO Box 994, Midland, MI 48686-0994; e-mail [email protected].
New materials: Is the risk worth the reward?
L. Shon-Roy, K. Holland, S. Holland, J. Housley, Techcet Group LLC
 Lita Shon-Roy |
The drive toward more compact and complex electronics systems and devices has caused a move to smaller and more integrated devices. Since the 1980s, key materials developments have widened process windows, allowing designers to use more wired transistors per unit of silicon. To enable the continuation of Moore’s Law through shrinking design rules, the industry must focus on materials innovations. Numerous technologists believe following Moore’s Law past the 13nm node will not be device shrink-related, but rather will be based on new materials and technologies that will allow further increases in functionality per dollar.
With each new device generation, chip design size and functionality have challenged the ability of traditional materials to meet electrical and physical (mechanical, thermal, etc.) performance requirements. It now appears that the number of new assimilated materials will have to increase in order to meet the technology demands. As an example, over the four-year period 2001-2004, 10 materials were introduced into semiconductor production, including those for copper CMP slurry, Cu electroplating solutions, Ta/TaN sputtering targets, GeH4 for strained silicon, ceria for STI CMP, TMCTs as a low-k precursor, and H2O for immersion lithography.
 Karey Holland |
Development costs may outweigh profits. The development cost of most new materials is often in the tens of millions of dollars. Some materials were in development by material manufacturers for years, only to be rejected by the end users, yet some chipmakers have stated they expect materials manufacturers/suppliers to develop the new materials and process integration, and even test for reliability. Major corporations that have been through this investment and disappointment cycle may hesitate to embrace yet another IC industry material opportunity. Corporations and startups, however, are always coming up with new ideas. The potential revenues for a successful material in the IC market tend to attract brainpower and working capital. Thus, while some suppliers may be unwilling to repeatedly take this risk after a failure, there are always companies willing to bring new technology to the marketplace.
Market size is huge and growing. The semiconductor materials market is estimated to be ~$29 billion in 2004 and is expected to grow an average of 8% annually, to total more than $36 billion in 2007. Certain advanced materials are expected to grow in excess of 25%/year through 2010, although they represent a relatively small portion of the materials market. They include low- and high-k dielectric materials, low-temperature nitride sources, electro/electroless plated metals, and photoresist/ancillary materials. In particular, the market for advanced dielectrics is expected to have grown from ~$50 million in 2003 to more than $200 million by 2007, with a compound annual growth rate higher than 45% from 2003-2007. Interconnect dielectric materials, such as SiF4 and TMS (trimethylsilane), account for more than 60% of these revenues, with frontend dielectrics such as BTBAS (bis(tert-butylamino)silane) and future Hf-based products making up the rest. The markets for new photoresist and related ancillary materials are expected to grow just as rapidly, as new materials are put forth to support future lithography such as 193nm and EUV. Consequently, while there is risk involved in the development and commercialization of new materials, the rewards are still quite promising.
65nm and 45nm materials. For 65nm devices, lower-k materials (for interconnect) in combination with new Cu barrier and/or capping materials are being developed. The first 65nm production is expected in 2006. Among the new process materials being looked at for 65nm include cobalt-tungsten-phosphide, as a barrier or capping material, ruthenium for the gate metal for MRAMs, spin-on SiO2 as pre-M1 dielectric in RAMs, and possibly OMCTS (octomethylcyclotetrasiloxane) for low-k dielectrics.
The 45nm node is scheduled for 2010 and myriad new materials will be required, including resists/ancillaries; high-k gate dielectric materials such as hafnium nitride precursors; lower-temperature nitrides for sidewall spacers; and even lower-k interlayer dielectrics, in addition to a complementary set of barrier and conductive metals.
Semiconductor process materials represent a huge business, and their development is critical to continued device scaling. As devices shrink, more materials innovations will be required to deal with higher aspect-ratio designs and thinner and more densely packed films. There are many technical challenges for each new material that defines the pathway to the future of semiconductor technology; if successful, the potential revenues for materials suppliers are substantial.
For more information, contact Lita Shon-Roy, partner and senior market analyst, at Techcet Group LLC; [email protected].