Question 1: Tell us about yourself. What is your background, and what projects are you currently engaged in?
My background spans academia, the computer industry, and biotech. I have a Ph.D. in Astrophysics and I taught and did research, mostly at Harvard, for 10 years. After that I was at Digital Equipment Corporation (DEC) in roles that included running their business with some of the process industries (Oil & Gas, Pharmaceuticals), starting an environmental solutions business, and corporate strategy and planning. After moving to California I joined Clinical Micro Sensors (CMS), a startup that developed an electronic DNA detection system. My role there was Vice President of Business Development. While I was at CMS, the CEO and I were approached by three scientists who had been developing a concept for using molecules for information storage. We decided to help them form a company and that became ZettaCore. During the funding process for ZettaCore I became convinced that it was a significant business and technical opportunity and I decided to stay with it full time. I’m very glad I did; we’re making a lot of progress, we’re having fun, and I think we will be able to contribute something important to the electronic devices of the future.
Question 2: Tell us about ZettaCore.
ZettaCore is focused on developing memory devices that use specially-designed molecules for storing information. Like other companies in this area, we are rooted in university research. In our case the founders were three professors from the University of California at Riverside and at North Carolina State University, along with me. We differentiate ourselves in three ways: 1) the quality of the technology, 2) the experience of the science and management team, and 3) our business strategy, which we call “practical molecular electronics.”
On the first point, we use a well characterized charge storage mechanism to store information based on the intrinsic properties of molecules. The molecules we use are designed in a rational way to have known properties, and we couple all of this with thoughtful design for implementing molecular properties based on existing fabrication methods. To the second point, our technical team has years of experience in designing, making, and measuring the properties of molecules, in semiconductor fabrication, and in circuit design. Our management team and investors are recognized experts in a broad range of technology fields. On the third point, we have set as a goal from day one that our products will be manufacturable in a semiconductor fab using common process steps. Our philosophy is to partner with the semiconductor industry.
Question 3: How long do you believe conventional CMOS scaling will continue? Do you believe that the International Technology Roadmap for Semiconductors (ITRS) is accurate?
There is a difference between how long conventional scaling can continue and how long it will continue. While I’m not a fabrication engineer, from what I’ve studied I think it can continue for another several years. But we have to remember Moore’s Second Law, which doesn’t get as much attention as the better known First Law (which is more of a wish than a law by now). The Second Law says that the cost of fabrication plants increases linearly with the decrease in feature size. Because the First Law says that feature size decreases exponentially with time, this implies that fab costs will increase exponentially. This is a “law” that hasn’t quite held, thanks to the ingenuity of lots of clever people. But the trend of increasing fab cost is very real and it’s that cost that will cause a shift, or partial shift, away from a fully CMOS-based process. How long conventional scaling will continue is an economic issue, not a technical one. I would say conventional CMOS will continue only until a viable alternative is found. A viable alternative will have to be a) not incremental, but a major improvement that can scale for a considerable time, b) manufacturable, and c) evolutionary in its use of resources and capital. There is both technical and economic pressure to find alternatives now, which is why companies like ours are being formed.
Regarding the Roadmap, I think it’s conservative relative to the reality we will experience. It’s just that no one knows yet exactly how to get there.
This is a good point to say something about nanotechnology, which I think is often misunderstood. We’re usually classified as a nanotechnology company, and that’s accurate. But nanotechnology is a means, not an end. My advice to investors is to be interested in means, but to invest in ends. We’re basically a semiconductor company, so evaluate us in that industry. Other nanotech companies are medical device companies, or specialty chemical companies, and you have to look at them in those contexts. Being a nanotech company doesn’t confer anything special on any of us. Most broadly, nanotechnology is the capability to do things at very small scales. The value of that is being able to overcome significant barriers to expected progress. In the biomedical area, for example, those barriers concern understanding nucleic acids and proteins and their interactions. In our industry, it’s Moore’s Law. So I view the most interesting parts of nanotechnology as a set of disruptive technologies for industries where growth is already expected but not assured.
Question 4: Tell us about porphyrin molecules.
We use specially designed porphyrin molecules to store information. We do that by adding or removing charge (electrons); in chemical terms we oxidize and reduce them. This is how nature uses porphyrins too. They are part of energy cycles in nature: chlorophyll is a porphyrin, as is part of the hemoglobin molecule.
Our molecules (we have a large library of them that we have created with different characteristics) can have anywhere from one to eight electrons removed and they will remain in that state indefinitely. The amount of voltage required to remove each electron is quantized. That makes it much easier to put molecules in a known state, and to detect that state. We assign each state of the molecule to a bit pattern, which means we can store multiple bits, up to three in some of our molecules. The voltages involved are no more than a few hundred millivolts per electron. That, coupled with the very long retention time, means that it will take much less power to operate a memory device using porphyrin molecules as the storage elements.
Question 5: What is your current timeframe for introducing molecular memory? To what extent will molecular memory supplant conventional silicon based DRAM?
We’re working on prototypes now, and we’re hiring more engineers to accelerate the process. We’re not disclosing our plans in detail, for obvious reasons.
Question 6: How reliable is porphyrin memory? Will it be rugged enough to operate effectively outside of laboratory conditions?
One of the reasons we picked this class of molecules is that they are robust in nature. We’ve done a lot of characterization work on the molecules involving repeated cycling, temperature, voltage, etc. and so far the molecules are at least as rugged as conventional memory.
One of the important things to realize about our approach is that we’re not using individual molecules. We’re putting many, many molecules on each memory element. Depending on the lithography size used it could be as few as several thousand molecules or as many as a million. This provides a high degree of redundancy and excellent signal to noise characteristics.
Question 7: Intel researchers have argued that future chips might have molecular memory, but will never use molecular circuits for logic. These scientists claim that molecular circuits will never be able to exhibit high gain. Do you believe that we will ever have a pure molecular computer?
We aren’t trying to create a fully molecular device. We use a hybrid approach that can be manufactured using process steps common to most semiconductor fabs. Our memory storage locations use molecules to store information, but the I/O circuitry uses conventional CMOS fabrication.
A fully molecular logic device presents much, much more difficult problems. First, it requires individual molecules that have circuit properties. There is progress being reported in this area quite frequently, but many issues remain, and gain is just one of them. A related issue when individual molecules are used is the signal to noise ratio. Second, and much more importantly, no one has yet shown how to assemble groups of individual molecules in a way that can be mass produced. This “wiring problem” has two aspects: one is how to assemble molecules into circuits repeatedly and reliably. The other is how to connect them to the rest of the world. Nobody has even shown how to reliably hook up a three-terminal molecule (other than manipulating individual molecules with an AFM tip – not exactly a solution for manufacturability), let alone how to arrange molecules in circuits in a way that matches the repeatability and economics of lithography. There is interesting academic work being reported in these areas, but what I’ve read hasn’t discussed the kind of practical advances that overcome the manufacturing hurdles and achieve repeatability, stability, predictability, etc. We’ve tried to address those issues from the beginning by working within the constraints of existing fab facilities and using chemical self-assembly of the molecules. Obviously this has meant trade-offs, but I think the trade-offs are worth it to make practical progress that is consistent with the resources and capital deployment of the industry.
The short answer to your question is that it’s very early and a lot of very bright people are working on molecular computers. We shouldn’t dismiss that effort, but in my view it will take a very long time to produce a molecular logic device that isn’t a laboratory toy.
Question 8: Have conventional chipmakers such as Intel and Motorola shown any interest in your research? Are you receiving financial backing from any major semiconductor companies?
We’ve actually declined investment from semiconductor companies (and others) in order to be able to focus on partnering broadly with the industry. At some point we will decide to work more closely with one or more of the large companies in this industry.
Question 9: How does ZettaCore plan on mass-producing these memory devices? How can these devices be deposited on a substrate with sufficient precision?
As I said before, we’ve focused on using common semiconductor fabrication facilities and processes. To be more specific, we create both the I/O circuitry and the memory array using common lithographic processes. The memory array is created in a way that leaves the areas where molecules will attach exposed. The molecules can be attached by dipping the wafer in a solution of molecules, similar to existing protocols. We engineer the molecules to undergo chemical self-assembly, which means that the molecules attach and pack tightly at the correct exposed surfaces (for example, Silicon or an oxide) and don’t attach elsewhere, a process the industry calls “self-aligned.” We can just wash away the molecules that haven’t attached, leaving the rest of the molecules firmly attached where they are supposed to be.
So initially the precision is achieved through lithography. The differences made by the molecules are that we can use much smaller feature sizes for the memory elements, the molecules have quantized energy levels (which makes the memory storage quite stable), and the voltages are reasonably low, usually under a volt.
Question 10: Will molecular memory be able to work without being attached to molecular wires? Has ZettaCore made any attempt to fabricate molecular wires, or to connect these wires to porphyrin molecules?
I hope it’s clear from the previous question that we don’t need molecular wires to achieve significant advances in memory using molecules. Our approach can certainly add capability to an approach using molecular wires, but we don’t require it. Molecular wires are going to have most of the problems of assembly and signal strength I mentioned when we were talking about all-molecular logic devices.
Question 11: The military should be quite interested in potential military applications of porphyrin. Has ZettaCore received any funding from DARPA, or any other military organization?
DARPA has been interested in many applications of molecular technology to electronics. Most of the US-based firms working in molecular electronics arose from the DARPA molecular electronics program that started in 1999. Our scientists were among the first to be funded by that program.
Question 12: What are ZettaCore plans for the next decade? Is there any possibility that ZettaCore could end up as the next Intel?
Of course we could be a major part of the next revolution in the electronics industry. If our technology is successful it could be part of almost every electronic device. I think that’s why we have been able to attract a very strong group of investors, including some well-known firms. But we have to take this one step at a time, and right now that means setting practical goals that make technical sense because they can achieve order of magnitude improvements in one or more key memory parameters and make business sense because they leverage the resources and capital of the semiconductor industry. We’re not trying to be David slaying Goliath. We’re trying to be really good at one thing that people need.
