INDUSTRY NEWS

THE MICRO INTERVIEW:

MEMC's Nabeel Gareeb

When Nabeel Gareeb took over as president and CEO of MEMC in April 2002, the semiconductor industry was mired in one of its worst downturns ever. The wafer supplier itself teetered on the brink of bankruptcy. More than two years later, equipment and materials companies are riding the wave of the long-awaited upturn cycle. During that same time, MEMC has righted its ship under Gareeb's leadership, experiencing several consecutive quarters of profitability, something nearly unheard-of in the narrow-margin world of silicon suppliers.

After I heard Gareeb's compelling keynote address on what he calls "the new S- curve" for semiconductor materials suppliers at the Industry Strategy Symposium (ISS) in January, I knew I wanted to line up an interview with him. Before joining MEMC, he worked at International Rectifier for 10 years, where his last position was chief operating officer. His ISS presentation incorporated some of his chipmaking experiences, laying out his vision of generating a better return on assets through a combination of cutting costs, limiting capital expenditures while yielding more output for the money invested, targeting investment in new materials and wafer diameters, and partnering with customers.

We touched on many aspects of the S-curve model during our nearly hour-long conversation at Semicon West. An intense and energetic guy, Gareeb answered my questions with enthusiasm and focus. Among the topics we discussed were the progress that MEMC has made under his leadership and what else remains to be done, relations with suppliers, the challenges of emerging alternative materials such as SOI and strained silicon, the changing attitudes of customers, and the obstacles to precompetitive cooperation within the wafermaking community. —TC

MICRO: What kind of progress have you made on the goals you had set for yourself or set for the company when you first started on the job?

GAREEB: I typically had fiscal goals for the company because that's the easiest thing to measure and I'm very much of a numbers kind of guy. If you can't measure it, how are you going to improve it? So you've got to have some level of measure. We're doing well with those goals. I think, however, on a scale of 0 to 10, we're just at a 4. And I think we've got a ways to go to get to a 10. So if you look from the external side, people think that we're doing well financially. When you talk about 20-plus points of operating profit, that's a pretty rarified atmosphere, especially for a company that was almost bankrupt two years ago [laughs]. So I think we're doing well, but I think there's a lot more room for improvement, and we're just starting to gain traction.

MICRO: Where do you see the greatest room for improvement: The operations side, the manufacturing side, or the structural side?

GAREEB: I'm a parallel kind of guy. I like to have 50 things going in parallel with significant intensity. If I've got 12 people working for me and each of them can do five things in parallel, well, that's 60 things, and I can keep track of 60 things, OK? So if we can do that in parallel, that's awesome. On the operational front, there's cost reduction, quality improvement. On R&D, there's capability improvement for the customers, performance improvement, and device reduction. There's another interesting element we'll come back to on R&D, which is extension of asset life. I talked about it at the ISS a little bit, and it is an important thing coming up for our customers. On the sales and marketing front: partnering with specific customers, focusing on the right ones, especially in this time of shortages. As wafer [supplies] get short, you want to make sure you supply the guys that you want to partner with for the long haul. So all of those have to occur in parallel. You can't serialize this. If you serialize this, the thing falls off the tracks.

MICRO: What do you see as some of your key six-month to two-year goals from this point forward, more on the manufacturing and R&D side?

GAREEB: Let me start with a little generic philosophy and then I'll give you some specific examples of how that comes about. For example in R&D: I'm very focused on the fact that you have to solve a problem for a customer. You can't just say the classic, "I've got this mousetrap, so where's the problem?" R&D has to be focused on solving customer problems, which means we do three things for customers: either we improve the performance of their device, or we reduce the cost of their device, or we extend the life of their asset. Those are really the only three things that we can fundamentally influence with the customer. So some examples may be, "Well, we want to extend the life of the asset of a 200-mm wafer fab. We can come up with the next generation of silicon, but what does that take? How does that look? What is that?" Strained silicon is an example. SOI is an example. I think those issues represent bigger and better things to come.

Reducing the cost: For example, we have a product called MDZ. We didn't promote it, we didn't focus it, we didn't target it for specific applications. Now, as we're targeting it, it gives yield improvement to our customers and their fabs. If our wafer is 5% of their cost of goods to manufacture, maybe it's 10%— it's in that range—and they get a three-percentage-point yield improvement, that's huge. So our goals for R&D are to make it more targeted, and the second is, obviously, to speed up its development. Time to market here, because it's materials science, is slower than devices, etc., but it used to be that three years was OK to develop a platform. That's a disaster in my world, so we said, "No, we need to get this down to six to nine months, and I'd like to get to that target." We've cut it in half, but we want to cut it in half again.

Manufacturing: We need continuous cost reduction and asset-intensity improvement, those two parallel paths. So you need to reduce cash costs, but you also need to improve the revenue-per-dollar asset on new investment. If you do both, then you can win on operational efficiency.

MICRO: How do you achieve that balance? On the one hand, you have your state-of-the-art part of the plan, which is doing the 300 mm and some 200 mm. On the other hand, you still have a large base of customers who want the smaller wafer sizes, so you have to maximize this old equipment. You don't really want to go buy any new equipment if you can help it. Could you offer some insight into that process.

GAREEB: We look at demand and supply monthly, at a minimum. We look at long-range forecasts monthly and say, "Where is the demand coming from, what types of products is it in, what types of market segments is it in." And you need to make sure that you understand intuitively why it's occurring, so you understand the drivers behind it as opposed to, well, gee, marketing said or R&D said or sales said.

Once you understand that, then yes, it's a little bit of a gamble on where you invest. But we're investing in both. We're investing in 200-mm debottlenecking, not a lot, but because 200 mm's tight, people are running out of wafers, so we're investing in debottlenecking to add basically what I call effective capacity—that to improve yields, you add effective capacity because you produce more wafers. For the same input, you produce more wafers.

MICRO: So you're not adding tools, you're improving the performance of the existing tool sets.

GAREEB: That's right. And you may add some tools here and there for debottlenecking. Second, now you'll have to improve capability as well, because people are trying to take 200 mm below 100 nm, right? So that migration occurs while the same equipment can't do it. It's incapable. OK, so you go from single-side polish to edge and backside buffs, then you go to double-side polish, etc. You have to have that platform strategy so you spend incrementally an appropriate amount of funds, yet give the capability to the customer. That's the trick, that's where the mad secret sauce comes in on how you spend x-dollars to get y-dollars of revenue. You do the same thing on 300 mm as well, because you can spend hundreds of millions of dollars and get less than what you spent in capital on revenue. If you do that, then you're in inverted asset turns and it's game over. We're doing both, and that's really the secret sauce to the operational efficiency we've demonstrated over the last two years.

MICRO: How do you crack the whip on your supply chain in order to get them to be in sync with you and partner with you to achieve that balancing act?

GAREEB: As for our suppliers, there's some uniqueness in equipment and tool-set capability. Some of them are on the same page as we are. There are three different dimensions. One is, how do we get them to be on the same page as us in terms of aggressiveness—what we're trying to do as a company. Second, there is a split on where the demand is coming from, on how aggressive they can be both on pricing as well as on flexibility. We want to ramp up, want to ramp down, we want to slow this down, speed this up. And third is what I call attitude about what we're going to do, how long we're going to be around to make sure we're the leader as opposed to somebody else, whether it's upgrades or what have you.

On all three fronts we are working with our suppliers to make sure that this occurs. Our financial performance has significantly helped, because now suppliers say, "Oh my god, this is a company I want to supply to because they're going to be around for a while. Hell, they might even turn out to be the leader over the long haul." Then the question is, now that you've got that attraction factor going, you need to talk to them about, "OK, you can't just supply to me, you've got to partner with me." And to me that means you've got to be on the same cost-reduction curve, you've got to be on the same flexibility curve; you can't say, "I've got a nine-month equipment lead time, call me nine months in advance." Maybe I wanted a four-month equipment lead time. How do we do that on a consistent basis? You have to work to see if you can change the rules of the game, because we're changing the rules at MEMC. We've got to help change it elsewhere.

MICRO: As for the good and the not-so-good responders, is there any correlation with company size or breadth of product line or experience?

GAREEB: Yeah, there is.

MICRO: One of your key suppliers is KLA-Tencor and they're a big, billion-dollar company. But I've also heard in the past from some wafer suppliers that sometimes they don't get the attention they think they deserve from a company that is so focused on the fab.

GAREEB: That's absolutely right. With the companies that overlap with the fabs, sometimes you get lost in the shuffle and you have to be cautious. The good news is that some of those companies have realized it. KLA is actually a perfect example of that. We went through a conversation, I think it was last year, a gentleman from there called me up and said, "You know, Nabeel, we'd like to do more business with you, I'm not sure how, etc." I said, let me ask my guys. I asked them, they said, "With KLA, the bottom line is the tool has lower throughput, the price is more, and the attention isn't there. What's the point of having the conversation?"

The good news is I gave that feedback directly to this gentleman. They talked about it. The gentleman came out [to see us] but I said, "Don't sit down with me. I'm not going to make this decision, my guys are going to make this decision. You sit down with this product engineer, with this guy, my head of R&D, etc., and you convince them that you can do something different." And over the last year, they have. So they've got more orders, they've responded, and guess what? We got what we wanted, they got what they wanted, they're starting to penetrate the market and that's excellent.

MICRO: Let's talk a little bit more about the R&D side. For instance, you have basically two competing types of SOI. How can you pay for development of both? Because if the market is going to be X, and you pour the usual tens of millions into the R&D but the market's not going to be much more than that, where's your ROI? What are some of the approaches that need to happen?

GAREEB: I'll just draw it here [starts sketching x-y diagram]. There's this fragmentation of the market where you have the diameters on this dimension, and then you have all the application specificity on this dimension, right? You get into SOI, you get into strained silicon, you get into perfect silicon, you get into strained SiGe. So you're compounding the investment requirement of this dimension with this other dimension. The trick is, each customer wants each supplier to be that full-service house. Which I like. And the reason I like that is because the people who have figured out how to do the asset-efficient solution on this dimension have also figured out how to do the asset-efficient solution on the other dimension. So far, we think we're the leader on that asset-efficient solution...we've been demonstrating it to some degree on 300, and we're going to demonstrate it even more over the next couple of years.

So how do you pour the dollars? We're finding ways to do that. We started it in a not-so phenomenally great way. . . with small amounts on 300. We've become a decent-sized player on 300. On SOI, we took our own sweet time doing some evaluations, but we basically licensed the bonded processes. We've got them manufacturing in-house, we're delivering samples, and we think this is going to result in unique capability both for surface uniformities and BOX (buried-oxide) layer uniformities, as well as commercialization price points. We're doing the same thing with strained silicon, with the process we licensed from IMEC. There are ways to do this within existing budgets to evaluate it. Now can you go out and develop full-fledged platforms for Simox and bonded? Not for very long. At some point you have to make a choice. [MEMC has made a choice on SOI. In late July, the company announced it is going with bonded SOI and abandoning the Simox approach.]

MICRO: In terms of these application-specific materials solutions, what are some areas with the greatest challenges? Is it the physical, the electrical, or the integration area?

GAREEB: Rather than challenges, let's talk about opportunities for our customers, right? They can go for cost reduction, but the biggest bang for the buck can come from integration. Because integration is not just about integration of functionality under a single chip, it's also a classic space-time constraint. If they can do that and get the classic heat equation resolved, that's awesome. That's the ultimate solution. That's where we can also add the most value. If you can improve the mobility, if you can improve the heat dissipation, if you can reduce the leakage characteristics, if you can do all of those things, that's where I think the greatest opportunity is, and that's where SOI and strained silicon apply.

The other element is obviously cost reduction. You get things like MDZ that reduce the cost. You get things like perfect silicon that allow you to create zero COP (crystal-originated particles) or zero-defect silicon. Therefore you don't have to use epi, you don't have to use annealed wafers, you can use a perfect silicon wafer instead. And if you have a deep-trench device, the perfect silicon doesn't bother the device.

So I don't see these solutions as either-or's, I see them as parallel and orthogonal solutions that combine. For example, SOI doesn't need anything else. Guess what? As you get into thinner and thinner layers of top silicon, above the BOX [layer], what is that? It needs to be perfect. It needs to have zero defects. Guess what? We already have perfect silicon. We can combine it into our SOI offering. It's complementary.

MICRO: You talked about partnering with the customer as well as with your vendors and also the research community à la IMEC. Are you seeing the customers coming around to that point of view?

GAREEB: I think customers are starting to. Part of the problem is that shortages of wafers cause the "partnership mentality" to come out [laughs]. But people are also starting to see that a wafer manufacturer can offer these types of solutions as opposed to just respond to a request. I think they see, "You can add value because you understand silicon better than we do. We didn't think you did." The reason for that was because for 20 years, all we did was bulk or epi. There wasn't anything for us to apply creativity to. MEMC invented perfect silicon six years ago. Why? Well, it was great, but guess what, its benefits don't really kick in until below 100 nm. So for four or five years our customer community didn't really care. They said, "Oh, this is an interesting solution you've got, but what's the problem it's solving?" Now they realize the problem and so the solution's getting more pull.

MICRO: So you were ahead of the curve at first but now the curve's caught up to you.

GAREEB: Exactly. We have to make sure that we've targeted the problem it solves. Once you target the problem, the solution and its pull become evident. Customers are starting to recognize, "Yes, these guys can help us solve our problems rather than just give us product." So it's starting to happen, but it's not going to happen overnight.

MICRO: What about work with fab OEMs who need your wafers and are using your wafers in a partially or fully processed manner? Is that kind of work involved in some of the partnering you're doing?

GAREEB: Not yet. We haven't moved to that level. We're looking at creative wafer origination solutions as opposed to postprocessing solutions.

MICRO: Is that something you see as a possibility down the line?

GAREEB: It would be a great possibility, but I think it's further down the evolutionary curve.

MICRO: What about the wafer's connection to reliability, beyond yield? Have you had any data flow on that yet?

GAREEB: No. We haven't had any data flow, and that's the toughest part.... The good news is that as customers start to realize it, we feel good, they feel good about it. On the other hand, unfortunately there's this almost natural process built in: "Well, if I'm adding too much value and I get too much of that feedback, well then, gee, I could bake it into the pricing where it would be disadvantageous to them."

I don't have a fab, so I don't really get that direct feedback. I've gone to foundries to pay them to build devices so I can get the feedback on our new products. That's really part of this evolution: because as these values are added, the customer understands it's not just about price. That's OK. We'll take a little bit of it on price, but they will get a lot more value added for it. And then we do it on one example, then we do it on the second example, and the third example, and a fourth. By the fourth one, the relationship begins to build where you can really call it a partnership. It's beyond demand and supply, beyond pricing, and beyond the first couple of semistrategic materials solutions. It's now really, "Hey, you can give me ideas on how to solve these problems on a continuous basis as opposed to one-off solutions I may need."

MICRO: That's a recurring theme throughout the supply chain. The suppliers don't get enough feedback from the customers. It seems like such an old-fashioned way of doing business and yet it still goes on. You, as key suppliers, should be trusted to maintain confidentiality because everyone's going to benefit from more of a feedforward/ feedback process.

GAREEB: We are still part of that exact same loop. You give wafers and you get feedback six months later on the wafers that were started, and maybe you get feedback if it's bad. You may not get feedback if it's good, yet you need to know what to repeat if it's good.... For the customers that figure out who they can trust as a supplier and really partner with them, I think there are benefits to be gained that cannot be quantified. Maybe they can be quantified but they are exponential in nature. They're significant. They're orders of magnitude higher than what our customers are getting today. I see opportunities every day for us to add value to the customers, but sometimes we can't get that traction or that conversation going, and that's a wasted opportunity for the customer. We could give them this value, but we're not being heard.

MICRO:: Can you give an example of a customer with whom you feel the relationship is close to where it should be, where you really feel that they're getting it?

GAREEB: I think they're all moving, especially the larger customers. They're starting to understand. Because again, MEMC and the silicon world took directives more than anything else, never offered up solutions. Now that we're offering solutions and we're starting to talk about solutions to problems, they're saying, "Well, maybe these guys have something." If we give a sample of this or test this out, etc, that's really important. We're making progress on all fronts, but they're all in different states of evolution. On a scale of 0 to 10, I want it to be a 10; I don't think any of them are above a 5. And some of them are at a 2. So there's a long way to go.

MICRO: In terms of the starting materials roadmap in the ITRS, how has it played a role in the company? What do you think of the roadmapping process in general?

GAREEB: Again, it's [about] ownership by the wafer manufacturers. We kinda sorta helped establish roadmaps, but we really didn't. We just sat there and said, "That's the roadmap from the device guys, which are our customers, so yes sir, we'll deliver." That's OK. But on top of that, let's look at what other solutions we have in our toolkit. Because one thing I know, whether your device guy's selling to a systems maker or a wafer guy's selling to a device guy, the customers always know what they want, right? But I guarantee you they don't know all the possibilities. And so, de facto, they will define the solution based on their sphere of knowledge. If you can add insight into that sphere of knowledge and provide additional opportunities, they're willing to learn. But the onus is on you to do that.

We didn't really have a chance or capability to do that [before] because we just talked about bulk or epi, bulk or epi...Now we've hit a wall. There's also a compelling event for the device guys to say, "Hey, help us! We're hitting a wall. Whether it's scaling, whether its mobility, whether it's heat dissipation, whatever it is, please help us!" OK, now that you bring solutions to the forefront, they want to listen because there's an issue that can't be solved in the traditional way.

MICRO:: Have you encouraged your people to get more active then in the roadmapping activities?

GAREEB: Absolutely.

MICRO: Do you see that from your competitors as well?

GAREEB: It's kind of a chicken-and-egg thing: How much is on our roadmap that we want to share and provide the solutions to our customers, and how much do we share with the industry? So I think the chicken and egg will get sorted out over the next year or so as these materials come to the forefront. Because once one supplier does it, the customer wants it from two sources, since the fabs are too big just to have it from one source.

MICRO: Do you see any other areas for precompetitive cooperation with your brethren in the wafer world?

GAREEB: I think there are, it's just difficult because you get into all this collusionary antitrust blah-blah-blah mentality. It's been such a dog-eat-dog world for 20 years that there's not necessarily a lot of trust.

MICRO: The mind-set is so locked in.

GAREEB: Yes, the mind-set: "Why are you talking to me about it?" There are areas that have had good cooperation and other areas that can lead to mistrust. There's a lot of space in between that needs to be explored and can be explored and probably will be explored.

MICRO: How about a consortia or association approach?

GAREEB: Exactly. SEMI provides a vehicle and a tool . . . to be able to move that to the next step. We have to build the level of trust where we can say, "Yeah, we're doing this for our collective good, we're not doing this for our own good." I think that it will evolve, but it will take a separating of the current suppliers into two tiers.