What’s Next For DNA Chips?

CMOS-based DNA sequencing chips, and the competing technologies, face a number of cost and technical hurdles.


The field of genomic personalized medicine is an emerging and promising area. Using lab systems called DNA sequencers, these units could provide vital information about an individual’s genetic makeup.

In this area, several companies are developing next-generation DNA sequencers, built around CMOS-based semiconductor technology. But CMOS-based DNA sequencing chips, and the competing technologies, face a number of cost and technical hurdles.

At present, the industry’s goal is to sequence a human genome for $1,000 or less. At that cost, insurance companies may one day reimburse patients. Eventually, the goal is to reach $100. But the actual cost to sequence a human genome is roughly $4,000, according to the National Human Genome Research Institute (NHGRI). The cost could be substantially higher for complex whole-genome sequencing.

DNA is a molecule that carries the genetic instructions in living things. A complete set of DNA is called the genome. DNA sequencing is a technique that determines the order of DNA bases, which is key for genetic research and possible therapies. “At its core, cancer is a disease of the DNA,” said Andy Felton, head of product management for genetic analysis at Thermo Fisher Scientific. “It’s a mutation in the DNA that causes the cell to get out of control. That’s the promise of sequencing of cancer genes.”

So, in any case, the big question is clear—Can CMOS lower the cost, and pave the way towards more breakthroughs, in DNA sequencing? In this area, OEMs are not scaling traditional CMOS transistors. Instead, they are scaling the functions of each new chip. But like the IC industry, DNA sequencing chips are made in a CMOS-based fab flow using lithography, CVD, etch and other process steps.

Still, there are several challenges. “The sequencing by synthesis chemistry, which Illumina uses, is the most successful (DNA sequencing technology) by a wide margin,” said Ross Muken, an analyst at Evercore ISI. “Semiconductor-based sequencing, while still in its infancy, has shown to have some significant disadvantages. They have the homopolymer issue, where it’s tough to read sequential bases of the same letter. And then you have accuracy issues as you shrink the detector on the chip. You get signal-to-noise. This basically means the smaller the detector is, and the closer it is, you get cross talk. Nanopore-based technology, which is also relatively new, has similar issues in terms of stabilization and scaling.”

What’s next?
DNA chip developers are moving to overcome these challenges. So where is the technology heading? To get a sneak preview of the future, one could look at two separate efforts in the arena. The first is from Life Technologies. A separate and more futuristic technology is being developed by a team from IBM and Arizona State University (ASU).

Life Technologies, part of Thermo Fisher Scientific, continues to advance its CMOS-based chip for DNA sequencing. The chip, dubbed the Ion Torrent, is an array of transistor pH (pHFET) sensors. Tiny microwell structures are situated over the sensor arrays. This, in turn, localizes the DNA fragments being sequenced near specific sensors. “As you change the pH inside the little well on the surface of the transistor itself, that transistor stack registers that pH change and converts it into a voltage signal. That voltage signal is used to measure the incorporation of a nucleotide,” Thermo Fisher Scientific’s Felton said.

The company’s goal is to scale the pHFET sensors for each new chip, which promises to lower the cost and increase the simultaneous sequencing reactions performed in the device. Rolled out in 2011, the first Ion Torrent chip had up to 11 million wells on the same chip. Then, two years ago, the company developed a new chip, which has 165 million microwells at 1.5-micron geometries. That chip, dubbed the Proton 1 (P1), has 61,824 analog-to-digital converters (ADCs) running in parallel with 24 high-speed links at 2.5-Gbs.

Then, at last year’s IEEE International Electron Devices Meeting (IEDM), Life Technologies described its next-generation chip–dubbed the Proton 2 (P2), which will be incorporated in future DNA sequencers. Based on 0.5-micron feature sizes, the P2 device boasts 660 million microwells on the same chip. The P2 has the same amount of ADCs, but the data rate per link is 5-Gbs, according to the IEDM paper. “Making the chip at that density is not the challenge,” Felton said in a recent interview. “It’s some of the other areas, such as making it run at the right speed, and making sure we get enough signal out of it that we can measure it accurately.”

What’s next? The company is also developing the P3, which will consist of 1.2 billon tiny wells on the same device. The 0.5-micron device will have a 0.7-micron pitch, compared to 0.84-micron for the P2 and 1.68-micron for the current P1 chip. The P3 will have a die size of 463mm square, which is the same as the P1 and P2.

Meanwhile, in a more futuristic approach, ASU and IBM’s T.J. Watson Research Center have developed a prototype DNA reader. “Our goal is to put cheap, simple and powerful DNA and protein diagnostic devices into every single doctor’s office,” said Stuart Lindsay, an ASU physics professor and director of its Biodesign’s Center for Single Molecule Biophysics.

Researchers devised a tiny DNA-reading device, which appears to be a combination of CMOS and nanopore technology. “It is a solid-state device that uses electron tunneling to sense the individual bases,” Lindsay said. “The goal is to put thousands or more on a final device. The smallest critical dimension is 2nm.”

The solid-state device resembles a tiny sandwich. The device consists of multiple layers. The device is composed of two metal electrodes separated by 2nm thick insulating layers. The layers were deposited using atomic layer deposition (ALD). “It is built using standard CMOS type steps,” he said.

Then, a hole is cut through the structure. DNA bases are read as they pass through the gap between the metal layers. “(With this technology) some applications could occur quite soon. For example, there is a possibility that it would make a super-sensitive protein detector,” he said. “DNA and protein sequencing pose formidable challenges associated with putting a fine nanopore through the structure without damaging the junction. Controlling the flow of molecules past the junction is also a challenge. We are working on these problems.”

Over time, researchers hope to commercialize the technology. ASU is collaborating with Roche on DNA sequencing. Meanwhile, an ASU spinout company, dubbed Recognition Analytix (RAX), hopes to develop a way to sequence single protein molecules. “RAX has an exclusive option on IP for sequencing proteins and oligosaccharides, and is in the process of raising funds to commercialize the technology for those applications. The nucleic acid sequencing applications are exclusively licensed to Roche,” he said. This work was supported by Roche, NIH and NHGRI.