SIA ISSUE BACKGROUNDERS

INTERNATIONAL TECHNOLOGY ROADMAP FOR SEMICONDUCTORS

Issue: The International Technology Roadmap for Semiconductors (ITRS) identifies the technical capabilities that need to be developed for the industry to continue trends such as increased scaling (e.g. "Moore's Law" -- the industry's ability to double the number of transistors on a chip every two years). By identifying principal technology needs, ITRS can guide shared research by industry, universities, and national labs. ITRS is especially valuable in highlighting those technical areas where there are no "known manufacturable solutions" to continued scaling. It is in these areas that breakthroughs in research are needed.

Importance: The ITRS has two key findings:

• In order to maintain the progress of Moore's Law, the latest ITRS envisions more aggressive scaling than projected in prior roadmaps (2003 Renewal, 2004 Update - view online at http://public.itrs.net). For example, dynamic random access memory chips will feature critical dimensions of 90 nanometers in 2004, which is both smaller and sooner than the 100 nanometers projected for 2005 in the roadmap published just four years ago. Similarly, microprocessor transistor gate lengths - a critical dimension that affects the processor's speed -- will be just 25 nanometers in 2007, six years sooner than expected in the 1999 version of the roadmap. (Note: a nanometer is one-billionth of a meter. A human hair is 100,000 nanometers in width, and a red blood cell is 5,000 nanometers in width.)
  
  
• We are beginning to reach the fundamental limits of the materials used in the planar CMOS process, the process that has been the basis for the semiconductor industry for the past 30 years. By introducing new materials into the basic CMOS structure and devising new CMOS structures, further improvements in the CMOS process can continue for the next ten to fifteen years, at which time it becomes evident that most of the known technological capabilities of the CMOS device structure will approach or have reached their limits. In order to continue to drive information technology advances, it becomes necessary to investigate new devices that may provide a more cost-effective alternative to planar CMOS in this timeframe.

Taken together, the two findings of the ITRS lead to the conclusion that we will continue to accelerate the rate of technical progress, and therefore will reach the fundamental limits of our current process that much sooner. The roadmap highlights in red those technical areas where there are no known manufacturable solutions. Consequently, the point at which there are no known solutions for most technical areas has been called "the red brick wall", which serves to highlight the need for research breakthroughs in those areas. According to the 2003 ITRS, we will reach this red brick wall between 2006 and 2007.

• How are the ITRS conclusions reached?

The ITRS is a result of a worldwide consensus building process involving over 800 experts from Europe, Japan, Korea, Taiwan, and the United States. A majority of the experts come from semiconductor makers, with the balance coming from equipment/materials suppliers, universities, and research consortia or institutes. Each region's semiconductor association collects the opinions of technical experts in a dozen fields such as lithography and metrology. These views are then pulled together by the Roadmap Coordinating Committee to develop the ITRS.

• What are examples of the difficult challenges that may impede progress in the near term (through 2009)?

There are dozens of challenges outlined in the ITRS. One example is the difficulty of making the mask used to transfer the integrated circuit layout designs onto future chips. As the layout designs require manufacturing transistors with finer line-widths it becomes progressively more difficult to accurately control the line-widths (e.g. a difference of a few nanometers separates success from failure.) Another difficult challenge is the prompt development of new metrology tools to accurately perform critical measurements as new materials, processes, and device structures are introduced. A third example is the difficulties associated with depositing metal into deep and narrow holes etched into the chip. Billions of these holes must be built on each chip to interconnect the billions of individual transistors on the chip.

• What are examples of the long range (2010-2018) challenges?

Again, the ITRS lists dozens of technical barriers that must be overcome. For example, optical lithography falls short of meeting the tough requirements expected after 2010, requiring the introduction of next generation lithography tools such as extreme ultraviolet lithography and electron projection lithography. These new tools will require development of a totally new infrastructure. Breakthroughs are also needed to interconnect all of the transistors required on a single chip. These breakthroughs might include optical or wireless connections within a chip rather than today's electronic/metal interconnect.

• What are the benefits of achieving the roadmap's targets?

The ITRS ties performance progress to key technology nodes, as indicated on the table below. The DRAM half pitch correlates to the average of the width, plus the space in between, metal lines connecting DRAM bit cells -- the smaller the half pitch, the more DRAM bit cells can fit in a given area. The Microprocessor "physical gate length" refers to the length of the gate that controls the flow of mobile charges in the underlying silicon. When a voltage larger than a threshold level is applied to the gate, the gate is switched "on" and charges flow between the two regions adjacent to the gate, from source to drain in the underlying silicon. When the voltage is lower than a the threshold level, no flow of charges occurs and the switch is "off." The shorter the gate length, the faster the switch time becomes.

ITRS Technology Nodes and Chip Capabilities

	2004	2007	2010	2018
DRAM Half-Pitch (nanometers)	90	65	45	18
DRAM Memory Size (mega or gigabits)	1G	2G	4G	32G
DRAM Cost/Bit (micro-cents)	2.7	0.96	0.34	0.021
Microprocessor Physical Gate Length (nanometers)	37	25	18	7
Microprocessor Speeds (MHz)	4.2	9.3	15	53

The table indicates the DRAM density and function cost, and the microprocessor speeds, associated with given technology nodes. The benefits to achieving the ITRS technology targets are readily apparent. By the end of the decade, the cost of memory declines to 1/8 today's cost, and microprocessors are three times faster. By 2018, the end of the ITRS timeframe, memory costs is less than 1/100 today's, and microprocessors are 12 times faster.

Smaller, faster, denser, and cheaper semiconductors have helped propel the information revolution, resulting in faster U.S. economic growth, greater productivity, higher federal budget surpluses, and the creation of high-tech, high wage jobs. Semiconductors have also contributed to our national security, and will increasingly contribute to our homeland security. Continuation of these gains is at risk as we approach 2006 when the ITRS projects that progress will stall without research breakthroughs in most technical areas.

SIA Position/Action: Increasing support for university research has become SIA's top public policy priority.

In considering the approaching limits of planar CMOS device technology, it is worth noting that this technology ultimately resulted from technical investigations initiated in the 1940's. These early studies did not lead to the start of the semiconductor industry, as we know it today, until the late 1960's. It would be difficult for any single company to support the progressively increasing R&D investments necessary to evolve the technology leading to a set of new devices and manufacturing paradigms, usable beyond the limits of CMOS.

Finding solutions to these challenges will require increased understanding of the fundamental device physics and properties of materials, as well as breakthrough approaches to technical problems, coming from long-term, pre-competitive, university-based research. Increasing support for university research has thus become SIA's top public policy priority.

There has been an alarming decline in federal spending in information technology related fields, leading to fewer professors and students and thus a diminishing pool of knowledge and experts in critical areas. From 1992 to 2000, federal funding for key disciplines declined or remained stagnant, with significant increases starting only in 2000. Federal funding trends during this period included:

• Physics declined nearly 7%;
• Chemistry dropped by 9% through 1999, until boosted in 2000;
• Although overall engineering rose, key fields declined:
  • Chemical engineering fell by 34%
  • Mechanical engineering was down by over 14%
  • Electrical engineering dropped 2%
    

While there has been progress in reversing these trends, such as the 8 percent increase in the NSF appropriation for 2002 and 14% in 2003, the total budgets remain inadequate to maintain the IT productivity trends on which our economy has come to rely. Looking forward, SIA supports significant increases in research in a variety of IT fields, including the Defense Department's GICUR (Government Industry Cooperative University Research) program, the Networking and the Information Technology Research Initiative, and National Nanotechnology Initiative. SIA also encourages agencies such as the National Science Foundation (NSF) and the National Institute of Standards and Technology (NIST) to monitor the ITRS and allocate the resources required for university and national lab research to help meet the roadmap's long-range targets. SIA is pleased that the NSF and Semiconductor Research Corporation have formed a partnership, Silicon Nanoelectronics and Beyond, to support research needs identified in the ITRS and other challenges at the nanoscale.