Semiconductor & Patent Expert Consulting

Litigation expert consultant and patent expert witness for process, device, and circuit of  Dynamic

 Ram (DRAM), Flash  (NAND, NOR, EEPROM), and Static Ram (SRAM) Memories,

 and Microprocessor, Logic, and Analog Devices

Aviza moves ahead in ALD: Q&A with Subrata Chatterji and Michael Martinez, Aviza Technology
Chris Hall,, Taipei [Tuesday 8 August 2006]

As the semiconductor industry moves further into deep submicron and begins to implement new design techniques and materials such as high-K dielectrics, new deposition techniques are required. Currently, the industry buzz is around atomic layer deposition (ALD), which is projected to be among the fastest growing segments in the semiconductor industry, with a CAGR of 22.8% between 2006 and 2011, according to VLSI Research.

Surface Film Formation Technology (SFFT) specialist Aviza Technology is playing a pioneering role in making ALD technology a reality for the global semiconductor manufacturing industry, and shipments are underway. Aviza recently shipped its Celsior ALD system to Inotera Memories in Taiwan as well as to leading foundries in China and Germany, for example. spoke recently with Subrata Chatterji, VP and general manager of Aviza's ALD Business Unit, and Michael Martinez, VP for Corporate Marketing and Investor Relations, Aviza Technology. Chatterji and Martinez outlined the background to Aviza's emergence as a leading player in ALD, the development of the technology and the challenges that lie ahead.

This is Part I of a four-part interview. Part II follows on 9 August.

Q: Can you sketch in the background to Aviza's recent public listing?

MM: We became a publicly listed company on the NASDAQ in December of 2005. The public offering was very unique, in that Aviza merged, as a privately held company, with Trikon Technologies, a publicly held company, in what is called a reverse merger. The resulting surviving entity was Aviza, and it was done through a process that essentially was an IPO for the company. First we merged, and through that process became a publicly listed company.

Q: Prior to that, you had been the largest of the privately held deposition companies, correct?

MM: Yes. Aviza is an outgrowth of several merged companies - including ASML, SVG, Watkins-Johnson and Thermco - going back to the 1960s. ASML put its thermal division up for sale in 2003, and those assets were bought by a venture-capital firm by the name of VantagePoint Venture Partners. They then brought in a management team, which today includes our President and CEO Jerry Cutini and Executive VP & CFO Patrick O'Connor. Jerry and Pat recruited the rest of a very seasoned management team, one that had a lot of experience both in the industry, and in M&A and growth initiatives. What was formed was a privately held company, and at the time Aviza was believed to be the largest privately held company in the industry.

We also created what we called the Certified Refurbished Equipment Business Unit, for used equipment. We made it clear, publicly, from the outset, that it was our full intention to grow and expand the business, both organically and inorganically, and that we had our eyes open for expansion opportunities.

That opportunity for expansion came our way with Trikon Technologies in the spring of 2005. The relationship with Trikon Technologies began as a joint development agreement to use their proven handling platform and merge it with our ALD processing module. Subsequently, we could see that there was synergy and chemistry between the companies, and we went forward with a merger. So the growth activity, since we formed Aviza in October 2003, has been extremely rapid. We've expanded our product portfolio, obviously, in the process. We've also expanded our global infrastructure, worldwide, and aligned our sales forces, noting that two-thirds of our sales come from Asia. We've done all that while focusing on our core expertise, with a combined 75 years of focus on SFFT technology.

As far as competitive positioning is concerned, speaking specifically about the thermal and deposition markets, as we exited last calendar year we believe we were up in the single digit market share range. Thermal processing wasn't ASML's area of core competency and it didn't fit with their strategic roadmap. Upon establishing the new company, Aviza's priority was to reintroduce ourselves and build customer confidence with our products and services, to regain share that had been lost prior to its formation.

Today, our largest competitors in the thermal and ALD business include Tokyo Electron (TEL), Kokusai and ASMI. We've established a very good foothold and gained very good traction in Asia, Europe and North America, competitively, in the short time that we have been Aviza Technology.

Having become a public company, last December, we certainly feel that we are keeping pace with industry demands and trends. Strategically, we feel we're doing the right things. We are continuing to diversify our markets, particularly with the penetration of what were formerly Trikon Technologies' products, which now form three new business units. We've hired a general manager to oversee the new business units based in Newport, Wales. So far this year, we've been making very significant announcements of doing more business, including the significant announcements weve made, very recently, about our ALD business.

Q: In terms of financial results, I believe you're reporting revenues of US$171 million in fiscal 2005.

MM: Our most recent guidance, provided for FY 2006, was US$155-160 million

Q: Are you able to say anything about profitability at this point?

MM: Not at this point, but we are working toward that end as we expand our commentary during our earnings reporting.

Q: As far as your competitive position goes, you mentioned TEL and Kokusai in Japan. Do you still regard Aviza as competitive with ASMI?

SC: Yes they are amongst our competitions. Some of our newest business has actually come at the expense of the mentioned companies' market share.

Q: I'd be interested in any comments you may have on Aviza's competitive position in the ALD market.

MM: Without going into specifics, let us briefly sketch the evolution of the competitive ALD landscape. There were a few companies that started in ALD, back in 1999. At that point, there were three of them competing in the ALD/DRAM application, for which we were shipping manufacturing tools. In this particular segment today, Aviza has gained increasing market share by penetrating multiple high volume DRAM fabs. This is really a testament to our experience, commitment and entrenchment in the ALD industry.

Certain companies are focusing on logic, and logic manufacturers have not yet adopted ALD. Despite early entry into some logic companies, these ALD equipment manufacturers have not been able to expand their installed base. On the other hand, our initial strategy of focusing on DRAM has worked in our favor. We focused on certain DRAM applications, knowing that DRAM would be the first segment to embrace ALD, and we were correct.

This is Part II of a four-part interview. Part I appeared on 8 August. Part III follows on 10 August.

Q: Does there tend to be any religious divide on batch versus single wafer, as there is with ion implantation, say?

SC: No. There are some single-wafer companies who claim that batch may not see the light of day - probably because they don't have a batch solution. The product strategy for these companies is to claim faster ALD processes on single-wafer platforms. We believe that for certain applications, especially in DRAM, batch platforms make more sense due to lower cost-of-ownership (CoO), while other applications that require sequential processing will continue to prefer single-wafer platforms. So we think there's a market for both kinds of products.

Q: In other words, TEL and Kokusai are obviously developing a single wafer tool?

SC: We cannot comment on other companies' developmental plans but logic would suggest this to be the case.

Q: They're already down to 32 nanometers in ALD?

SC: Some companies have already started researching 32-nm materials selection, because they know that getting R&D materials qualified for 32-nm process will take a longer time and much more effort, so it's better to start early.

So that's the basic competitive background. In terms of thermal processing, the key players remain unchanged.

Q: Could you sketch in the background to the development of ALD technology, and the features that differentiate Aviza's systems, such as the Verano and Celsior?

SC: ALD was said to have been invented around 1976 at the University of Helsinki by Professor Suntola. This led to the birth of a company called Microchemistry, in Finland. Professor Suntola is regarded as the father of ALD. Later in 1997, Samsung decided to apply ALD to their DRAM applications.

Aviza got into the market in 2002, and developed our own patented showerhead to address the challenging needs of the ALD process. Celsior is unique because it was developed based on assimilated input from the customers and market to help create a product that addresses their manufacturing challenges as it relates to faster adaptation of ALD technology in high-volume production. Celsior features an innovative chamber, which offers increased throughput, lower chemical consumption and an extended process window-resulting in lower cost of ownership. Celsior's patented showerhead is designed to meet stringent ALD process needs, with the capability of achieving less than one percent thickness uniformity requirements across a 300mm wafer. The tool leverages a small reaction chamber utilizing Aviza's patent-pending computer modeled gas flow dynamics to reduce the reaction volume and remove empty spaces, reducing the areas available for defect formation resulting in higher die yields. The tool is extendible, allowing the seamless addition of process chambers for adding future capacity or new technology.

This is Part III of a four-part interview. Part I appeared on 8 August and Part II on 9 August. Part IV will follow on 11 August.

Q: My understanding is that ALD implementation poses certain stringent requirements. Can you comment on that aspect of the technology?

SC: One thing we should recognize is that ALD, simply as a technology, is not actually the driver. The driver is the shift in the materials regime. When the semiconductor industry moves to high-K dielectrics for thin films, then chemical vapor deposition (CVD) runs out of steam, and the manufacturers need ALD. That's the key. When you go from polysilicon to metal electrodes, again, that's where ALD comes in because no technology can achieve that kind of step coverage.

Q: What's possible in terms of a relatively low temperature? As I remember, ALD can go down to 200C. Is that considered a low temperature for this kind of deposition?

SC: Usually, temperatures are between 200-400C for ALD. Single-wafer CVD can operate within a range of 500-700C. Single-wafer CVD can manage that, but batch CVD cannot. The lowest temperature batch CVD system in the market, currently, operates at about 550C, and even that is regarded as a revolution for CVD.

Q: What is the importance of having this low-temperature capability?

SC: At low temperature, ALD offers better film purity and higher density compared to other deposition techniques.

Q: And in terms of thickness, we're talking about a density of a few angstroms?

SC: Forty to fifty angstroms.

Q: And if you don't maintain that kind of film thickness?

SC: You get leakage.

Q: Any other characteristics that would suffer?

SC: You'd suffer from a higher etch rate.

Q: And how would you define step coverage?

SC: Basically step coverage is the ratio of thickness of film at the top of a trench side-wall to the thickness of the film at the bottom of the trench side-wall.

Q: So in other words, the thin film has to be applied very uniformly.

SC: Right, high conformality leads to good step coverage.

Q: And as I understand it, the standard uniformity that is quoted is less than 1% uniformity.

SC: Usually, 2 - 3 sigma % within wafer uniformity is industry standard. Those are the usual specs for customers. But the better ALD tools usually can achieve less than 1 sigma % uniformity. We claim we can do much better than that, actually, but the specs usually don't need to be that tight.

Q: How are Aviza's different tool offerings applied, in terms of film characteristics?

SC: For thick films we would sell the Verano. For thin films we would sell Celsior. A very important issue, here, is sequential processing. Let's say you're developing a gate cluster. First you need to do the pre-cleaning, then you would deposit the dielectric, then on the same wafer you would deposit the electrode, and finally some applications might need post-deposition annealing (PDA). So you are talking about a sequence of processes taking place in each of the chambers and you can't do that in a batch system. You have to use a single wafer system for that process.

Q: How significant is Aviza's software, in all this?

SC: Software is extremely important. There are two aspects of the software platform that are important to run any deposition system, including an ALD system. One dictates how a transfer module works (wafer handling), and the other dictates how a process module works - (implementation of the recipe details and so on). We received the software platform for our single wafer system from Trikon Technologies. It is a proven platform that has been used for etch, CVD, PVD, 200, 300 and multiples of that, so it's a very robust platform. Aviza software and process engineers worked with Trikon software engineers to extend the Trikon software platform to support ALD functionality in the Celsior process chamber as part of a JDP prior to the Aviza-Trikon merger.

This is Part IV of a four-part interview. Part I appeared on 8 August, Part II on 9 August and Part III on 10 August.

Q: Do you see the recent sale of Aviza ALD systems to a China foundry as a significant step forward in the development of foundry in China? Will it assist the growth of foundry in China generally?

SC: This rollout is significant to this customer for their ALD manufacturing needs. We hope that other foundries in China will follow a similar path when they are ready to implement ALD in a production-proven platform like ours.

Q: My understanding is that installation and qualification for ALD can be quite extended processes. They take time.

SC: Our customers are using ALD for volume production and have already qualified the process.

MM: We feel that we've now created enough critical mass beneath our wings to start becoming more public and substantiating the success that we're having. In doing that, we're also becoming a spokesperson for the technology itself. It's important not just to sell the product but also to educate the industry about the technology.

Q: How do your customers understand their purchases in terms of cost of ownership (COO)?

SC: The industry regards this as a key issue. If someone is asked, "What is the major challenge with ALD?" the first answer you're going to get is that it's a slow process. It's extremely slow. So in order to lower the cost of ownership, we have done two things. We designed a chamber that will enable a lower consumption of chemicals and enhanced throughput. Second, we also offer a batch ALD system for certain applications. Understandably, batch systems offer better CoO. We were very much focused on cost of ownership when we designed the tools, and quite rightly. I think that one of the key distinctions between ourselves and other companies is that we have been able to show results in achieving lower CoO.

MM: I think productivity and cost of ownership, irrespective of which segments you are talking about in the industry, are very significant. It's a forgone conclusion that these areas need to be addressed. Materials, in the classic Sematech cost of ownership model, have a dramatic effect on the final cost of ownership. So I think it goes without saying that (CoO) is something that always has to be addressed in any ALD product that is delivered to a customer.

Q: So what seems to be the consensus at the moment, on the question of throughput?

SC: Throughput is a concern, in terms of ALD, for all applications, and one solution is to go to batch, for those films that can be processed with batch, and for others, you just have to come up with better chemical injection systems, smarter chamber designs and use faster ALD specialized valves, as we have done.

MM: It's a technology that will evolve, just as CMP evolved in the 90s. In the early days of CMP, you would handle only five or six wafers an hour. But then you got up in the 30, 50, 60 wafers range. The technology grew with experience.





Aviza's Celsior ALD system features an innovative chamber for increased throughput and lower chemical consumption.
Photo: Company

Verano 5500

Verano 5500 variable batch ALD tool from Aviza Technology
: Company