While the rest of the industry anticipates the coming of 3D ICs, and along with it the long anticipated return on investment that goes with it, one small Silicon Valley fabless has been chugging along, already reaping the benefits of interposer integration and 3D IC technologies. I first heard Farhang Yazdani, CEO, BroadPak present during a GSA 3DIC Working Group meeting on 3D IC readiness, and then again at the European 3D TSV Summit, January 19-21, 2015 in Grenoble. During his talk, Yazdani addressed both the advantages and challenges of designing for interposer integration and 3D. Intrigued to learn more about his work and his company, I asked him to share with me his and his company’s view of the 3D space.
Yazdani first launched BroadPak in 2007. Since day one, BroadPak has focused on providing expertise in interposer integration (2.5D) and 3D technologies and services, believing them to be the way things would go in the future. The company provides state of the art co-design and signal integrity services for customer’s packaging and design needs as well as manufacturing and assembly through its ecosystem of partners. “As of today, we offer total solutions for customers who want to launch 2.5D and 3D products,” he said. Customers include system companies, IP developers, ASICs; basically anyone involved in IC design looking for package design solutions. “One of the big challenges startups have is how fast can they can get a test chip on the table,” said Yazdani. “We can provide engineering sample in a few weeks, which is important for the time-to-market window. Customers want to test the die, and decide whether to re-spin it or not. The sooner we show the data to the customer, the better.”
Although many were skeptical about the future of 3D ICs back in 2007, Yazdani was already a believer. Early on, he recognized the benefits for high-end computing, as well as start-up companies needing help leveraging design IP. “One of the most interesting drivers for 2.5D has been IP reuse,” he explained “When there is a lack of IP for a certain process node, and the customer wants to go to market ASAP, 2.5D provides fast entry into the market because you can mix IP on an interposer,” he explained. “60% of our work is coming up with multiple integration scenarios and solutions, that once established, become mainstream.”
Yazdani said many BroadPak customers are at the R&D stage, exploring options and prototypes to see the feasibility. They come to BroadPak for the whole solution. They provide the package and integration design, manufacturing and assembly through their ecosystem. The only thing they don’t do is design the silicon. Examples of their work include a configurable 2.5D/3D silicon interposer platform that simultaneously develops prototypes and production-ready solutions that they launched in January of 2014. They also worked with Silex Microsystems to develop its silicon interposer solution in 2012.
Are Designers Ready to Design in 3D?
As it’s clear that the technology processes are ready for the big time, I asked Yazdani for his perspective on why the design community seems to be reluctant to take the leap. “For designers, 3D is still a new area for them to explore. It takes time for them to come up with a flow, and there are risks that comes with new technology,” he explained. “(Questions like) what part of the die are we going to partition? What’s going to come off the interposer?” His team works with chip designers on the trade-off, cost, and routability until it makes sense. They figure out how much they can save on the power. It all becomes part of the co-design. “2.5D and 3D is the co-design we’ve been talking about for years,” he said.
Design Advantages and Challenges
Helping the customers to understand the benefits of not having all the IP in the same process node is a challenge, says Yazdani. They are afraid of the risk. “They need to better understand the benefits 3D has to offer,” he said. “IP reuse is the most important benefit. If IP is available at other process nodes, we can put that together.”
One of the biggest challenges from the design perspective is the amplified security risk of stacking, either on an interposer or another die. “Security risk is nothing new. It’s been around for 2D SoC in the form of piracy, counterfeiting, over builds, malicious circuits; the list goes on… “ He said. “In stacking, the impact is amplified, because partitioned die can come from different sources around the globe. It’s more challenging to detect problems and put them together.” Yazdani said this issue had to be addressed from the packaging perspective, and as such BroadPak introduced a tamper-resistant, security–enabled interposer and 3D package in July, 2014.
Low cost interposers are on everyone’s mind. One of the most recent innovations from BroadPak is a rigid Si interposer that can be directly integrated onto the circuit board (Figure 1). Eliminating the organic package substrate also eliminates the need for an ultra-thin interposer that needs to be handled via temporary carrier. Additionally, there is no via fill or via reveal steps, thereby significantly reducing the overall cost of the interposer, explained Yazdani. They have also addressed the CTE mismatch between the Si and the board, and the ‘rigid interposer’ is qualified and ready for production. An additional bonus – performance, in terms of signal and power integrity, is improved due to the elimination of the organic substrate.
3D ICs in Mainstream Consumer Applications
As is general consensus, Yazdani says 3D ICs will start in high-end computing applications such as CPUs, GPUs, and networking. But he thinks that it will slowly perforate into the mainstream because of miniaturization, and the demand for low power.
While historically OSATS need high volumes to be profitable, this is not the case for BroadPak. “One of the things many companies come to us for is prototypes and small volumes that big guys won’t consider,” he explained. “We have the ecosystem and partners in place to make that happen. We can take a bunch of small volumes and place a larger order.”
The Success Question
While many measure the success of 3D IC by finding it in a smartphone or tablet, Yazdani says he thinks that finding its way into high end computing applications should be considered a success. For BroadPak, interposer integration and 3D is already a success. They are a profitable organization with customers in high-end computing who work with them for that type of integration. The key value BroadPak brings to its customers is the solution for bringing it all together. It requires a deep understanding of chip, packaging, process, cost, and cost-effective solutions.
“Design is so important. In many cases package is more expensive than the chip itself. Design has a direct impact on cost and performance,” said Yazdani. “Good design can make you stand out as success. Success is a great package design.” ~ F.v.T.