Can MEMS resonators compete with quartz?

Startup companies such as SiTime, Discera, Silicon Clocks and Mobius Microsystems are now going after the $4 billion electronic oscillator market with their innovative alternatives to the traditional quartz based components.  We recently spoke with Kurt Petersen, founder and CEO of SiTime, about the company's latest developments as well as their technological approach, competitive advantages and strategy to tap into this lucrative market.

MEMS Investor Journal:  We last spoke in the early spring of this year.  How has SiTime progressed since then and what milestones were accomplished? 

Kurt Petersen: Well, we had our product launch in April with a press tour in Asia including Beijing, Hong Kong, Tokyo and Taiwan.  We've been sampling our products since then and we have procured a significant number of parts through our manufacturing partners TSMC and Jazz Semiconductor.  We have also inked an agreement with Ecliptek, which is a supplier of traditional quartz crystals, to purchase over 1 million of our devices.  The second big recent win for us has been the deal with Swatch's Microcrystal division -- in this deal, we are going to be developing new types of timing chips together with Swatch.  We think the fact that two established players in the quartz crystal resonator market are already working with us is a great sign that our technology is living up to its potential to be competitive with quartz.

MEMS Investor Journal: In particular, how do your device's cost and reliability compare to quartz?

Kurt Petersen:  Right now we are much more cost effective than quartz.  As a simple comparison, our entire oscillator device which includes both the MEMS resonator and the associated IC, costs about as much as a quartz crystal resonator by itself without the circuitry which can include multiple PLL’s, multiple frequencies, frequency skewing, and other timing functions.  For reliability, recent testing results demonstrated our devices to be about 10 times better than quartz devices in terms of aging specs.

MEMS Investor Journal: How does SiTime plan to structure its business model?

Kurt Petersen: This is a very good question.  Right now the timing industry is mainly composed of two segments -- the $3.5B quartz crystal market and the associated IC market which today totals about $1.5 billion in annual sales.  Thus, when people need a timing solution today, they have to buy a quartz crystal device, an IC  timing chip, and 2-3 passive capacitors.  Our goal is to completely revolutionize the current multi-part timing model by providing a single integrated “timing solution”, with all functions in a single package, based on our MEMS component.  In other words, our chip includes both the resonator plus all the IC electronics in the same package.  This strategy will allow us to have higher margins, still provide savings to customers, while simplifying customers designs by providing the entire “timing solution” in one package. 

MEMS Investor Journal: But on the other hand, since you have a fundamentally new approach, what objections are your facing from customers today?

Kurt Petersen: Yes, our approach is certainly very different and we did anticipate a reluctance from some potential customers.  Quartz has been around for such a long time, about 40 years, that designers are extremely cautious of any proposed alternatives.  At the same time, MEMS based devices have been increasingly proving themselves as reliable products -- MEMS microphones and TI's DLP chip, for example -- and this gives us an advantage to a certain degree to make customers more comfortable.  So the main challenge in this area is that our potential customers want us to go through a number of qualifications and tests to make sure that our devices are reliable and operate to spec.  Of course, this is similar to any other new technologies and with time we will be able to prove to our customers that our devices are, in fact, even more reliable than the standard quartz. 

Quartz actually has a huge reliability problem -- it is a mechanical part which is essentially glued inside of the package and its failure rate is worse than an integrated circuit.  Fundamentally, a MEMS based solution is bound to be more reliable.  Another problem is the fact that the quartz supply chain is more fragmented and parts have to be procured through different vendors and then integrated together by designers.  As such, capacity and procurement issues can be problematic.

MEMS Investor Journal: So, when you speak with customers today, what are their compelling reasons to buy your components and how would you rank them? 

Kurt Petersen:  Well, the first one is cost.  And here I mean the entire cost of ownership including the part itself, the associated electronic components as well as design and procurement.  We can certainly beat quartz in cost.  The second one is reliability -- since we have an integrated timing solution, our component is going to be more reliable.  Another important factor is size.  By 2010, 30% of timing components will be housed on mobile devices so the more expensive and bulky quartz based chips will become increasingly disadvantageous for designers.  Finally, the versatility of our chips to generate multiple frequencies within the same package (i.e. the complete “timing solution”) will also become advantageous for manufacturers who will want to make their devices as versatile and as ubiquitous as possible.

MEMS Investor Journal:  Let's talk about your competitors.  Discera, Silicon Clocks and Mobius Microsystems have also been commercializing their innovative timing solutions.  How does your approach and technology compare?

Kurt Petersen:  Sure, I've been involved with MEMS based resonators for more than 30 years and one of the big challenges has always been that the frequency generated by a MEMS resonator would always drift due to the physical debris, organic compounds, and water vapor which can be trapped inside the package.  Our approach is different from others because we do our MEMS encapsulation at 1100C.  This very high temperature eliminates all water vapor and contamination and the associated frequency drift.  We've licensed this unique bonding process from Robert Bosch and this is one of the big things which currently differentiates us from our competitors.  More specifically, our competitors (including quartz crystal manufacturers) package their devices at ~400C or less and incorporate getters within the package to absorb the water vapor. First, 400C is not sufficient to eliminate all the water vapor; secondly a getter is more of a band-aid solution to an inherent problem. Silicon Clocks technology seems to be based on SiGe (vs. silicon) and they will also need to use the 400C process and getters.  Mobius, like other companies, has been working on an all-IC (non-MEMS) solution.  While Mobius has been reporting better performance out of its chip than previous all-IC oscillators, it is going to be very difficult to hit specs over the typical temperature ranges and to maintain the high Q necessary for many applications.

MEMS Investor Journal: Can your competitors somehow acquire similar packaging technology as yours?

Kurt Petersen: We don't think so.  Most devices are based on materials which cannot be used at 1100C.  SiGe and polysilicon, for example, simply cannot be used in our encapsulation process because of the very high 1100C sealing temperature.  Furthermore, we've licensed our technology from Robert Bosch which has 14 issued patents on this process.  On top of this, we have been issuing our own patents as well.  So, we believe that we are well protected from the IP perspective.

MEMS Investor Journal:  What do you estimate your potential market to be?

Kurt Petersen:  Sure, right now the quartz market can be divided into three segments -- watches, in which cheap 5-10 cent crystals can be used; consumer electronics such as PCs and handheld devices; and high end communications such as reference frequency oscillators for cell phones and base stations.  Roughly, each of these three segments is currently at about $1 billion in terms of annual sales of quartz components.  Five years from now, we believe the combined market will be about $4.5 billion per year.  Based on the history of commercializing other MEMS technologies such microphones, accelerometers and DLP chips, we are targeting about 50% penetration with our devices in 5 years within the consumer segment.  Again, since our chips contain both the resonator and the electronics in the same package, our market also includes the $1.5 billion IC market for timing chips (growing at the same rate).   Thus, overall, we are looking at a total market of over $2 billion for our devices about 5 years from now.

MEMS Investor Journal: What do you think will be your critical success factors to execute effectively and reach this market?

Kurt Petersen: Well, first of all, one chip does not make a company.  We'll need to expand and eventually have an extensive and comprehensive product portfolio as we move forward.  Another factor is going to be communication and education of customers about the advantages of MEMS over quartz.   Of course, reliability and lifetime testing assurance will be important as well.  We'll also need to continue to strengthen our components' reputation with progressively larger design wins. 

MEMS Investor Journal: Are you planning to have another VC round in 2007 and what is your ultimate exit strategy?

Kurt Petersen: Yes, we'll probably have another VC round in second or third quarter of next year and our eventual goal for SiTime is an IPO.