Friday, September 14, 2012

#MATERIALS: "2.5-D Chip to Precede Full 3-D"

Integration of full 3-D semiconductors into the global market will take a back seat to silicon interposers, often called 2.5-D, which will grow to become a $1.6 billion market by 2017, according to Yole Development: R. Colin Johnson

Here is what Yole says about 3D silicon: After meeting with swift commercial success on a few initial applications, including MEMS, sensors and power amplifiers, 3D integration has been on everyone’s mind for the past five years. However, once the initial euphoria faded, and despite technical developments which assured most observers that mass adoption of 3D was not out of reach, some unanticipated technical and supply chain hurdles were revealed that were higher than anticipated. It was then that 2.5D integration by means of 3D glass or silicon interposers was revealed by experts as a necessary stepping-stone to full 3D integration. Our first report on 3D interposers and 2.5D integration was in 2010; at that time, Yole Développement listed the various applications of this technology trend and its drivers, and showed that glass and silicon interposers were expected to become high-volume necessities, rather than just high-performance solutions for a few niche applications.

In this 2012 edition of that report, Yole Développement team provides more evidence of its findings from two years ago: after refining the applications and drivers of 3D interposers and 2.5D integration with the use of detailed forecasts, Yole Développement estimates that far from being a stepping-stone technology to full 3D integration, 3D interposers and 2.5D integration is emerging as a mass volume, long-lasting trend in the semiconductor industry.

The business generated by the 2.5D interposer substrate will grow rapidly, to an expected total value of $1.6B in 2017

Glass & silicon 2.5D interposers are already a commercial reality in MEMS, Analog, RF & LED applications on 150mm / 200mm, supported by the relatively ‘exotic’ infrastructures of MEMS players such as IMT-MEMS, Silex Microsystems, DNP, and DALSA / Teledyne, and structured glass substrate suppliers like HOYA, PlanOptik, NEC / Schott, and tecnisco. On 300mm, the infrastructure and market for 2.5D/3D interposers has hardly emerged as of 2012, but nevertheless Yole Développement analysts expect that in 2017, over 2 million 300mm wafers will be produced in that year alone. They also expect that the silicon or glass type of 2.5D interposer substrate will impact more than 16% of the traditionally ‘organic-made’ IC package substrate business by 2017, with almost $1.6B revenues generated by then.

Strong digital drivers will shift technology and supply chain paradigms
As technology developments progress, the industry will discover clear advantages to using 2.5D interposers for new applications and supply chain possibilities. Throughout this 2012 report, Yole
Yole Développement – Le Quartz – 75 cours Emile Zola – 69100 Lyon-Villeurbanne - France
Développement details these new lead applications, as well as the relevant needs and challenges.
“Also, we show evidence that this emerging infrastructure, which was initially focused on MEMS and sensors, is shifting paradigms to logic modules driven by stringent electrical and thermal performance requirements. As a result, the demand for interposers is shifting to fine-pitch 300mm diameter silicon wafers and high-accuracy flip chip micro-bumping and assembly,” explains Jérôme Baron, Business Unit Manager, Advanced Packaging at Yole Développement.

Graphical Processor Units for gaming and computing and high-performance ASICs and FPGAs are paving the way, with high volumes first expected in 2013. As these drivers increasingly appear as must-haves to serve the ever-increasing need for larger electrical bandwidths imposed by graphical sophistication, cloud computing and many more end uses, leading companies are busy creating the appropriate infrastructure.

The semiconductor supply chain is adapting to these significant in substrate technologies.
Wafer foundries appear to be the most able entities to offer manufacturing solutions on the open market, both technically and in terms of capex investment capabilities. But their ambition extends far beyond the manufacturing of wafers, and into assembly and test services as well.
Concurrently, some of the major IDMs are preparing to exploit their wide capabilities and to enter the open foundry and assembly services side for 2.5D and 3D integration based on such new type of IC package substrate technologies.

Is cost really an issue in the long term?
Significant investments began in 2012, with more than $150M capex expected and driven by both wafer foundries (TSMC, Global Foundries) and OSATs (Amkor, ASE). No one, especially in Taiwan, wants to be left behind in this high-growth story, as it clearly appears to be a central piece of the increasing middle-end business and infrastructure, halfway between the front-end silicon foundries and the back-end assembly & test facilities.

The question now is: “can anyone build a profitable business case to support the growth of 2.5D/3D interposers”? In other words, how long will it take for investing companies to be paid back, while offering affordable prices to their customers? Yole Développement expects the expansion model of this new technology trend to follow a traditional path: first, high-value modules are expected to use the technology to offer unprecedented high performance, followed by higher volume applications.
“The nice thing about 2.5D interposers is that they do not only allow for unprecedented performance: they can do so for a much lower cost than any competing technology. Through a few cost cases in this report, we demonstrate that cost can be a strong adoption driver too. No, silicon and glass interposers are not “additional dead pieces of hardware in the package” - on the contrary, they are among the top five key elements of the semiconductor roadmap for the decade 2010- 2020,” adds Jean-Marc Yannou, Senior Analyst, Advanced Packaging at Yole Développement.
Further Reading