By Tom Morrow
John Smythe, advanced technology lead for Micron’s Advanced Materials Technology Group, kicked off the 2012 Strategic Materials Conference (SMC) on Oct. 23, with a comprehensive overview of the materials challenges to continued scaling in memory, including a status update on several novel and emerging materials sets that may have potential at sub-20nm.
How the industry will address these challenges through an evolution (or radical transformation) of the traditional collaboration models between equipment supplier, materials supplier and device maker also became a key topic during the two-day conference. Trends, challenges and collaboration strategies in other semiconductor devices, OLED displays, PV, and LEDs were also discussed at the conference, which highlighted the synergies and trends in advanced materials application and development. SMC is organized by the Chemical and Gas Manufacturers Group (CGMG) special interest group at SEMI.
Smythe outlined the scaling challenges facing memory cell technology going forward. Unlike the recent past, now “electrostatics are beginning to dominate,” requiring innovative approaches involving new and novel material sets that are complex and not well understood. More than 120 different material combinations have been tested in the industry over the past few years and their impact on etch, clean, and deposition sources are just beginning to be evaluated.
“If you really don’t understand the interactions, you’re just guessing,” he said. “We are only at the beginning of the learning curve.”
Smythe discussed how emerging memory cell material systems lack scaled solutions. Fundamental principles of electrostatics and dielectric leakage limit dimensional scaling of charge-based memory systems. Dry etch, CMP, cleans and cell material sources/methods will be challenged beyond normal evolutionary paths. Future memory cell technology must rely on characteristic changes of alternate state variables related to more complex cell material and electrode systems. This will significantly increase complexity in the form of additional atoms and hence additional aspects of control and interaction risk. Many of the candidate systems are likely to drive new requirements in the materials/chemistry supply chain.
Micron currently evaluates new memory technology by the parameters of data retention, bit density, endurance, power per bit, and manufacturability (see chart). Each of the potential new memory technologies, such as FeRAM, MRAM, and molecular switches, has associated risks with these attributes that can only be determined through costly analysis. Academic research can help guide that analysis to a point, but often focus on non-critical parameters that are not important for commercial development. According to Smythe, “Did they (academic researchers) test for the really hard things, or did they test for the easy things in order to get published.”
Micron Emerging Memory Strategy
The challenge for Micron and other manufacturers is aligning research and development priorities around the complexities of new materials evaluation. The rapid expansion of elements used in semiconductor manufacturing over the past decade—commonly illustrated with a color coded periodic elements table — only begins to illustrate the complexities involved. Each new element reflects a new molecular set. Tellurium, for example, is found in a variety of applications, including organotellurium compounds such as dimethyl telluride, diethyl telluride, diisopropyl telluride, diallyl telluride and methyl allyl telluride are used as precursors for metalorganic vapor phase epitaxy growth. Each new molecule has unique characteristics that can drive materials/chemistry related issues. According to Smythe, a key differentiating factor is generally the choice of subtractive versus damascene patterning pathway.
To illustrate the complex landscape of materials research today, Smythe discussed the challenges of a number of the most promising materials sets in memory technology today, including:
- Electrolyte CBRAM: Ag-doped GeX, Cu-GeSe, CuTe;
- Mixed Valence Oxide: CMOx, Pt/PrCaMnO;
- Stackable PCM with OTS selector;
- “Simple” HfOx, and
- STT-RAM: dry etch and clean pathways.
In addition to front-end materials, new memory cell solutions present novel packaging issues, especially with advanced packaging concepts leading to the question, “die to what?” Thermal and stress management issues will become more critical in the future; chip density and performance needs may drive the development of nano composites; anisotropic character tuning may become critical.
Collaboration Models Stressed, Debated
Within this landscape of exploding materials and device complexity, traditional collaboration models between device makers, tool suppliers and materials suppliers are undergoing rapid evolution. In a lively panel discussion on the hot issue, Smythe was joined by representatives from IBM, Applied Materials and Air Liquide to debate the means, mechanisms and requirements for orderly development of next-generation processes.
Jean-Marc Girard, PhD. and CTO of Air Liquide Electronics, began the discussion outlining his view of the requirements for effective collaboration in this new era. He pointed out that device makers need new films and processes on time for their roadmap, as well as a reliable and cost-effective material supply chain before ramp up. They also need fast screening capabilities (synthesis and films) and an understanding of the process CoO anticipation for new molecules already in the R&D stage. But if a precursor is not already available as feedstock, or if it cannot be extracted from a non-isolated intermediate, then a complex investment decision is required to synthesize a new material, with associated equipment-related process dependencies.
How this risk profile is understood and evaluated is increasingly complex as the success rate for new CVD/ALD materials is very low. Complicating the risk assessment is IP protection. Innovations in chemistry are at the core of new process development and IP positions today are increasingly held across the supply chain.
Girard wants to see “a clear collaboration framework in which IP ownership and IP rights are negotiated in advance.” The system should be “virtuous,” favoring disclosure over secrecy with tracking of IP throughout the development process and fair IP recognition and evaluation. Smythe, however, illustrated the difficulty in IP valuation, noting that “Relative Value” should be based on “what can be defined by the fractional content of the end product that results from implementation.” In memory, the silicon portion of the technology is generally more than 400 value-added steps, involving design, product, test, packaging, and more, each with additional degrees of freedom. There is a significant challenge to identify the fractional benefit given the significant infrastructure and relative degree of risk, according to Smythe, and “models with percent of revenue just don’t make sense.” He makes a case for “cooperative development,” where feedback can be assigned increasing equivalent value as the level of detail is increased.
Dr. David Thompson, technology manager for process chemistry at Applied Materials, also emphasized relativity, preferring to focus on the “big problems” and the significant contributions to cost reductions in dollars per wafer pass and cost of ownership. He sees the need for a new role for “process chemists” to bridge the gap between process engineers and synthetic chemists. He also sees a greater need for coupon evaluations to focus resources on full-scale production while increasing the probability of success by evaluating more options earlier. While Thompson emphasizes the need for clear “ground rules and scope” of collaborations, he reinforced the need for “trust.”
Dr. Dale C. McHerron, project manager for process research and strategy at IBM, also summarized the necessity for R&D collaboration. Early and deep collaboration on process, materials and equipment directions are essential to align development roadmaps earlier in the technology development cycle. Advanced technology development must leverage expertise of process equipment and materials suppliers to gain broader perspective on process R&D activities and enable efficient, collaborative R&D projects across the supply chain. But McHerron recognizes that three-way collaborations, while essential, are extremely difficult to arrange and manage. The framework that IBM utilizes, in conjunction with The College of Nanoscale Science and Engineering (CNSE) of the University at Albany, focuses on three levels of collaborations: 1) Joint development projects based on rigorous contractual agreements on IP, cost and return; 2) Beta evaluations with “limited exchange” contractual agreements; and 3) Demos based on “black box” CDAs.
With the exponential growth of new materials in advanced ICs, the complexity of joint collaborations in new process development is rapidly escalating, challenging all players in the supply chain. Other presentations at SMC from academia, other industry sectors, and from adjacent industries such as PV, LED and OLEDs, reinforced the expectation that materials-driven complexity and new collaboration models will be increasing. Key players in the industry recognize the need for rigorous contractual agreements to protect IP and assure fair return, but seem to understand the need for trust, flexibility and shared commitment to industry progress—factors not easily assimilated into legally-binding contracts.
The CGMG holds and sponsors a number of meeting events during the course of the year, including materials sessions at various SEMICON exhibitions and the Strategic Materials Conference. For more information, click here or e-mail: cgmg@semi.org
—Tom Morrow is the executive vice president for emerging markets at SEMI.
