As the main semiconductor industry marches forward, the push to increase substrate sizes and automate production for high-volume runs continues. Given these trends, the lower volume MEMS applications that are made on smaller substrates sometimes get left behind. Electroplating tools exemplify this trend. Semitool released the first commercial electroplating system for four-inch wafers in the early 1990s and continued to provide tools for small wafers, even as its product line expanded to include 300 mm substrates in the late 2000s. One of the main reasons for Applied Materials’ (AMAT’s) acquisition of Semitool in 2010 was the Raider family of electroplating tools. At that time, and probably to this day, Raiders were among the best automated electroplating tools on the leading edge market, and they were primarily geared for 300 mm substrates.
The provision of 300 mm plating tools is now dominated by three suppliers: Applied Materials, Lam Research, and TEL-NEXX. Applied Materials continues to sell the Raider to high-volume 200 mm and 300 mm manufacturers for a reported price of $3-4 million. Similarly, after the acquisition of NEXX by Tokyo Electron (TEL) and Novellus by Lam Research, commercially available automated plating tools tended to focus on the growth market of wafer level packaging (WLP) for the 300 mm “mega” fabs that produced 5,000, 10,000, or more wafers per month. According to the spokesperson from TEL NEXX, the company’s Stratus tool is currently priced at $2.5-$3.5 million, depending on automated or manual configuration, number of metal reservoirs, bridge tool capability, and other factors.
Applied Materials, TEL, Lam, and other similar tool vendors are very large multinational companies that tend to have as their top customers other very large multinational companies that reside primarily in the 300 mm silicon-based integrated device manufacturing (IDM) market, with some overlap into the 200 mm market. The economics are simple and obvious: top customers want cutting-edge technology performance in plating tools designed for high volume. For example, large wafer fabs needed to run damascene or WLP processes on 300 mm silicon wafers, and that’s what the vendors gave them. Over the years, innovations were made to support 65 to 20 nm node processing, and tool capacity increased, as did tool prices and sizes, to help drive down the overall cost of ownership (CoO) for these large customers. To focus on this mission, small diameter tools and platforms were dropped, phased out, or handed over to secondary equipment suppliers. At this time (from the early 2000s to the early 2010s), the manufacture of MEMS devices was relatively small in volume, and the industry remained in the emergent stage.
Now in 2015, MEMS manufacturing is coming into its own, albeit with a set of technological and economic constraints different than WLP for 20 nm logic or memory devices. Thus, there is a need to adopt cost-reduction technologies, as well as to replace aging processes that may no longer be efficient or desirable.
Today, as MEMS technology advances, new materials and processes that were the mainstay of the main semiconductor market in the mid-2000s are now being adopted. There is a growing need for “Goldilocks” electroplating tools — not too big, not too small, just right! This requires technology more advanced than the legacy tools of 20 years ago, and not necessarily as close to the cutting-edge as the latest-and-greatest 300 mm tools, with price tags and feature sets that fit production and budget requirements. As a consequence, most of the new automated electroplating tools available today (and over the last decade) are not necessarily a good fit for MEMS manufacturing.
“Big Equipment,” small customer
Until recently, advanced automated tools for small volume, or small wafers, were difficult to obtain, or prohibitively expensive when available at all. In terms of sales numbers, most automated plating tools sold over the last five years have been from large companies such as Applied Materials, Lam, and TEL NEXX. It is believed that the majority of these tools were either configured for 300 mm operations or provided on a 300 mm platform outfitted for smaller wafers.
With dominant market shares in their primary industries, and annual sales one or more orders of magnitude larger than most of their customers, these top-tier companies have become the “600-pound gorillas” in the market place, effectively becoming “Big Equipment.” This creates difficulty for manufacturers who run comparatively lower volumes on two-inch, three-inch, or four-inch substrates. This includes not only MEMS but compound semiconductors, automotive parts, sensors, LEDs, and other emerging technologies which form a new growth segment that could be called the “small diameter market”. For these manufacturers, Big Equipment has not always been the best match. In many cases, customers have resorted to using home-built tools, commercially available manual plating benches, or legacy single-wafer equipment.
Recent growth in the small-diameter market portends a need for affordable, commercially available, automated, and appropriately-sized (in terms of substrate and capacity) electroplating tools that provide the level of process performance needed for today’s devices, as well as for the next several device generations.
For small-wafer electroplating, the most sought-after automated plating tool was the Semitool Equinox. This automated plating tool, which originally sold for between $350,000 and $1.2 million, provided solid performance to price value and was available for three-inch and larger substrates, with two, four, or six single-wafer modules in a single, compact frame.
Micralyne, a MEMS foundry based in Edmonton, Canada, uses automated legacy equipment. “We expect modularity in systems, and therefore in pricing,” says Glen Fitzpatrick, Micralyne’s Chief Scientist. Fitzpatrick says that Micralyne uses Semitool Equinox and EEJA equipment for the company’s electroplating capabilities.
The Equinox is not a good fit for everyone, however. It was on the Equinox that the early copper damascene processes were developed and put into production, but it is currently only available from refurbished- and used-equipment vendors. Today, a refurbished Equinox from a secondary tool supplier can be purchased from between $300,000 and $1 million, depending on many factors, and for less, if one is willing to take the risks of an “as-is, where-is” deal. As with all secondary equipment, the challenge is finding a tool that works well and has the right configuration. For electroplating, that means having the right substrate diameter, as well as the correct number of chambers and chemical types needed to perform the desired process. This could mean the tool must have copper, nickel, and tin plating capability, or pre-wet and copper through silicon via (TSV) chambers. Many of the Equinox tools were sold just for gold plating, or just for copper damascene plating. It is for this same reason that Novellus Sabre tools are rarely seen in MEMS manufacturing. These plating tools were the top sellers for the 200 and 300 mm damascene process for many years, and there are likely many available in the secondary market. However, these were mostly single-purpose tools. They were built for one thing: copper damascene. They are not easily reconfigurable, nor are they adaptable to small wafers. EEJA is another smaller player in the market. The company’s POSFER tools were more highly configurable, and EEJA supported some small diameter substrates, and still does. With Intel as the primary customer for 200 mm WLP process, some legacy tools are available to the MEMS market, but they tend to be difficult to find, and they are not easily reconfigured.
All three of these tools, the Equinox, the Sabre, and the POSFER, are based on single-wafer architecture, meaning one substrate at a time is processed in any given plating chamber, of which there can be several. Each chamber is optimized for the performance of that particular process. In a sense, these are cluster tools designed for performance, and they use parallelism to achieve the capacity needed. They deliver the uniformity performance needed to achieve the process results for 90 nm logic interconnect, or 65nm WLP, and were fully automated. However, for simpler electroplating processes which are insensitive to uniformity (such as a ground plane), or for very low levels of capacity (including R&D and development), these types of tools could be an “overkill.”
MEMS process requirements are often different from traditional semiconductor fabrication. For example, there is a technology trend in the MEMS industry towards the use of thicker, more precisely deposited metal layers, which makes electroplating attractive. "Electroplating of precious metals can be an attractive option due to cost. There is very little material wasted during the process,” says Jessica Gomez, Founder and CEO of Rogue Valley Microdevices. “Unlike a sputter or evaporation process where excess metal builds up on the inside of the chamber, and will need to be removed and sent out for reclaim, any metal that is not deposited on the substrate during the electroplating process stays in the solution waiting to be used. With the right equipment setup and process development, electroplating can also be much quicker than other deposition methods, making it attractive for high-volume manufacturing," Gomez adds.
Also, given the cost and complexity of processing, very few MEMS manufacturers are interested in damascene; they just don’t need so many levels of metal for interconnect. As a result, the pool of legacy tools available on the market is smaller. In some cases, an older tool can successfully be repurposed by a secondary supplier. This can be an excellent choice for those who can find such tools; however, given the limited number of these tools, this is not a viable solution for a growing market. So, several questions remain: For how long will proprietary parts be available? At what price? What is the risk of critical parts’ being discontinued? Will the operating expenses of the tool skyrocket with time as the supply base diminishes? What if capacity expansion is needed in the future? Is this a long-term solution?
Batch processing and wet benches
Industrial electroplating was first developed over 100 years ago by immersing substrates into open baths of electroplating chemistry. For many years, this approach was an obvious choice for many metal deposition processes for MEMS and other device manufacturers in the form of wet benches. This type of equipment typically processes multiple substrates simultaneously (hence the moniker “batch”), tends to have a low price point, and has become ubiquitous in development labs, as well as volume manufacturing facilities. In this batch wet bench process, wafers are manually placed into a fixture that has electrical contacts. The fixture is then manually placed into an open bath, and a simple controller runs the process. The wet bench processes have been refined over the years, and now there are several systems which can process multiple substrates in the same bath, improving capital efficiency and lowering the cost of ownership.
“We currently use a manual process for electroplating in our fabrication flow. Volumes are low at this time, so automation is not yet necessary,” says Joe Fragala, Optics Manufacturing Manager at Zeiss X-ray Microscopy (formerly Xradia). Fragala says that he has only investigated manual plating options, mainly whole-wafer plating fixtures. “There are several vendors supplying turnkey benches or drop-in modules for a wet bench, but these are expensive for a small operation if more than one plating solution is used,” he says.
Wet benches suffer from several potential drawbacks. They are operator-intensive and prone to operator error; moreover, the results they obtain are not particularly repeatable. Wet benches also create additional safety hazards, due to exposed chemicals. And because of the need for fixturing, they are not easily automated and tend to take up a large amount of fab space. Nevertheless, they have two significant benefits which make them the tool of choice for non-critical processes: they are inexpensive and flexible. Process quality can be improved by making use of sophisticated fixtures, bath, and anode configurations (thereby increasing the price and CoO), but the underlying issues remain to some degree. In general, wet bench electroplating processes can be expected to deliver 10 to 20% uniformity, as compared to 3 to 5% for single wafer tools. There are always counter-examples, particularly with “home-built” equipment. In these cases, a custom plating tank can be optimized for specific processes that give acceptable results and a lower price point.
Dara Cardwell, Director of Operations and Process Engineering at Nuvotronics, says that she has not been able to find an economical and basic commercial tool for plating thick films on wafers for less than $50,000. “We currently use custom-built electroplaters that require manual loading and unloading. Automatic processing is not necessary in our current operations. One of our custom tools can plate up to six wafers simultaneously, using six independent power supplies,” Cardwell says.
A similar perspective is offered by Brian Stephenson, President and COO at Tronics MEMS, the company’s foundry facility based in the US. “We do not use fully automated equipment at this time. We use custom-built plating equipment. My expectation for a fully automated cassette-to-cassette plating system is on the order of $300,000 per plating cell.”
For low to modest volumes of production, and where engineering resources are available for support, this may be the best choice. But when greater capacity is eventually needed, or when human resources become stretched, a gap appears: there is no easy path to the increased capacity needed for volume production.
In response to demands for reduced operator interaction, and improved repeatability, some vendors have developed automated wet benches which offer dry-to-dry processing. As with manual benches, this can be an appropriate solution for non-critical processes. But as with single wafer tools, automation comes at a price, and automated wet-benches are not an exception. When performance is taken into account, the value of an automated wet bench is diminished when compared to a comparably-priced, automated single-wafer tool.
For most fabrication processes, single-wafer tools typically provide a higher degree of precision than wet benches. The ability to control conditions across a smaller volume results in both enhanced performance and a greater degree of repeatability. Electroplating is no exception. According to Dave Roberts, Principal of Silicon Valley Wafer Plating, there are many options available for wafer electroplating equipment. “A student can use a beaker, stir plate, anode, alligator clips, and a DC current source to get some copper plated, but the uniformity will leave much to be desired,” says Roberts. “On the other hand, a large manufacturing company using a fully-automated fountain-style plater with dedicated engineers and chemists will be able to attain near perfect uniformity on thin copper films.”
Igor Kadija, owner of ECSI FIBRotools, says that the two main challenges for the successful electroplating of MEMS and nanotechnology structures are, first, the control of the cell to maintain a uniform transport of matter at the solid and liquid interface of the substrate and, second, the control of the electric field distribution. Once these two parameters are optimized and set, an appropriately-selected plating current mode can be successfully applied to produce the desired structures.
This phenomenon is readily apparent when one looks at the legacy tools often used in MEMS manufacturing. For example, Semitool (now Applied Materials), Lam (previously Novellus), TEL NEXX, and EEJA offer single-wafer tools exclusively.
Figure 1. Comparison of wet bench and single wafer electroplating tools, including advantages and disadvantages.
Manual or automated operation
The reasons that justify automated equipment are well known (i.e., cost, capacity, and yield), and the degree of automation correlates with manufacturing maturity in all semiconductor-related industries. The tipping point for changing from manual to automated tools is not always as clear. “Our systems are manual, with six-inch wafer capacity greater than 100 wafers per week. We plan to bring in automated systems as demand ramps further,” says Chris Gudeman, VP Process Development at Innovative Micro Technology (IMT), a MEMS foundry based in Santa Barbara, California. Gudeman says that his expectation for automated electroplating tool is $1 million or more. For now, however, IMT will continue to use custom plating equipment with all processes internally developed.
As MEMS devices become more common in the marketplace, the technology will become more pervasive, volumes will increase, and manufacturing processes will mature. The past year has seen significant growth in manufacturing capacity and demand for MEMS devices, and it appears the trend will continue for some time. MEMS manufacturers are increasingly finding themselves facing the question of automation when it comes to adding electroplating tools to their production line.
“We only have manual processes. We have plans to automate the process, but we don't have enough demand that will justify the investment,” says Pilar Herrera-Fierro, a cleanroom supervisor at the Lurie Nanofabrication Facility (LNF) at the University of Michigan, citing one of the benefits that manual equipment provides for low volume manufacturing. Development organizations too can prefer manually-operated systems. “Manual tools are designed to be used for a single plating process at a time,” says Igor Kadija of ECSI FIBRotools. According to Kadija, the goal is typically to establish optimum parameters, including cell geometry, electric field distribution, cell hydrodynamics, and the current mode of operation, which would enable the production of functional single or multiple samples of device structure. “Ultimately, the operator wants to demonstrate the feasibility of their theoretical concept, a process that is difficult and costly to do with automated machines designed for high throughput,” Kadija adds.
For some of the larger foundries though, such as Silex Microsystems in Sweden, automated equipment is indeed the best option. “Given our focus over the last decade on TSVs and wafer level packaging, electroplating has always been an essential part of our core processing capability. We electroplate both six-inch and eight-inch wafers, and currently, the run rate is on the order of 1,000 wafers per month in total,” says Edvard Kalvesten, Silex’s CEO. According to Kalvesten, Silex uses fully automated electroplating tools for throughput and repeatability purposes, and the company’s production volumes can only be done on automated tools, both with regards to plating stability and continuous chemistry analysis. One of the electroplating tools Silex uses is the Semitool Paragon system. The company also recently acquired a tool made by an undisclosed vendor for eight-inch wafer electroplating capability.
"Although simple electrodeposition setups can be home-grown, users with stringent requirements for thickness, alloy composition, microstructure uniformity, repeatability, and throughput should consider production-level systems designed for automated wafer handling, with optimized wafer contact fixturing, uniform current density distribution, automated plating parameter control, uniform and repeatable electrolyte mixing, temperature control, and automated bath monitoring and replenishment," says Al Sidman, the former CTO of Advanced MicroSensors.
Flip Chip International also uses automated electroplating tools. “All of our electroplating applications in the wafer level packaging space have process control and customer quality expectations that make manual process plating unacceptable to most of our customers, especially those requiring processing of high-value add substrates like semiconductor or MEMS wafers,” says Ted Tessier, the company’s CTO. Tessier says that very thick plating applications, such as copper pillar bumping with thicknesses of 75 microns or more, and plating rates of one to two microns per minute, could possibly allow the use of manual processing tools if stringent procedures and controls are implemented, although the general trend in plating is towards more automation and process control.
Figure 2. Comparison of manual and automated electroplating tools, including advantages and disadvantages.
ClassOne’s “Goldilocks” solution
As technology and production control requirements march on, and thick metal deposition becomes more commonplace in the MEMS world, there appears to be a need for an intermediate equipment platform which adequately supports both automated and semi-automated electroplating applications in a price range affordable for typical MEMS manufacturers. One vendor that has made this its focus is ClassOne Technology.
According to Byron Exarcos, President of the ClassOne Group, there is a demand for automated solutions for volume 75 to 200 mm applications, such as MEMS, power semiconductors, biomedical devices, and LEDs. The year 2002 saw the genesis of the ClassOne Group with the formation of ClassOne Equipment, a company that focuses on high-end refurbishment and sales of used equipment. This division of the company is currently estimated to generate $20 million in annual sales and is reportedly growing faster than industry average. By being a part of the used-equipment market, ClassOne has closely engaged with cost-sensitive emerging technology companies, as early as the manual prototyping stage, through the maturation of automated volume production. From these relationships, Exarcos believes that the company has identified unfulfilled needs in the marketplace, in particular for electroplated metal deposition.
“Basically, if you need to do precision electroplating in volume for MEMS applications, your previous options were either the expensive tools from Applied Materials or the legacy Semitool equipment,” says Exarcos. “At this point in time, there is very little Semitool legacy equipment available in the marketplace configured, or configurable for appropriate use for the MEMS market. This observation led to the formation of ClassOne Technology, which designs and manufactures new wet processing tools to specifically serve the small diameter substrate, cost-sensitive market that was originally served by legacy tools.”
“The smallest Solstice plating system, the Solstice® LT was unveiled at SEMICON West in 2013, followed by the largest, the Solstice S8, in 2014. The Solstice S4 was introduced at SEMICON West in 2015, winning Best of The West™ award and completing the product line. Since then, the widespread interest and sales trajectory has validated our analysis” says Kevin Witt, CTO of ClassOne.
Witt says that the company’s Solstice platform is similar to Semitool’s Equinox tool. The S8 platform supports a maximum of eight chambers and is based on a radial robot arrangement, simplifying the automation and constraining costs. Among other electroplating-related positions he held during his ten-year tenure at Semitool, Witt was the Director of Disruptive Technology, and he focused on developing new plating platforms and processes for non-traditional markets. For less than a million dollars, ClassOne’s Solstice S4 platform can be configured with three plating chambers (with one wafer at a time in each chamber), and a single wafer spray pre-wet/spin-rinse-dryer dryer (SRD). So, as an example, for a TSV copper fill process where one wafer needs to be in the bath for 20 to 30 minutes, a manufacturer can run up to 1,500 wafers per week on such a tool. This is enough capacity for most MEMS manufacturers running in high volume, according to Witt, and it is a much better fit for price and performance expectations than the systems developed for the high-volume, 300 mm silicon industry.
Asked about competition, Witt says that the main competition to the Solstice is the perceived value of wet benches, some of which have been “hacked” to be more precise and automatic. “We have seen cases in the industry where some companies have retired their old Semitool automated equipment and, because previously there were no viable alternatives in the marketplace, the customer went back to wet benches for high volume production,” says Witt. This, however, is not precise enough and not scalable, according to Witt. “What the customer gained in cost, they lost in performance. They also had to revisit their process integration in order to be able to use a less capable, manually intensive tool based upon open chemical baths. We are changing the equation for these types of customers and are providing viable and effective alternatives to help mature the process and production line. Automated, high-performance tools are now available in a reasonable price range.”
Exarcos has a big vision for ClassOne Technology – he says that they are on track to become a $100 million company within the next five years. The company appears to be serious about this goal, and it even set up headquarters in Kalispell, Montana, where Semitool was based; and it recently announced the completion of its second capacity expansion. ClassOne has also announced that it has set up an applications lab for customers to test out their samples and develop their processes prior to purchasing equipment. ClassOne offers its basic semi-automated tool, the Solstice LT, for under $450,000. This smaller version of the automated tool has the same hardware, software, and performance and provides the first step along the path to advanced production, as well as a lower-cost option for the development line that feeds production.
Another company offering the same type of “Goldilocks” solution is SEMSYSCO, which is based in Salzburg, Austria. SEMSYSCO’s fountain-plating and high-speed plating tools are available as fully-automated and semi-automated systems for large volume production. The company also offers smaller tool versions for academia, R&D, and small volume foundry production. SEMSYSCO’s spokesperson says that it sells its tools from between $700,000 to $4.5 million, depending on the configuration.
EEJA, based in Hiratsuka, Japan, also sells new manual and automated plating equipment for 150 mm to 300 mm wafers.
The bottom line, for almost all MEMS manufacturers, is economics. To justify the purchase of any type of equipment, there must be sufficient profit. That puts various requirements in place for performance, capacity, and budget, with different constraints occurring at different scales of manufacturing. Having good choices from multiple vendors is important to keep the industry healthy and growing.
According to some industry sources, Applied Materials bought Semitool for the damascene process and wafer-level packaging tools. The company’s Raider automation platform supports up to 24 chambers on a linear robot arrangement and has been used primarily for electroplating applications. Because of AMAT’s size, some industry observers have claimed that the company has focused primarily on the largest semiconductor manufacturers. This trend is echoed by the products available from TEL NEXX and Lam — their highly-automated tools provide cutting-edge performance at very high capacity levels, with commensurately high price tags.
However, Applied Materials maintains that it continues to focus on MEMS and other smaller substrate applications. “We are actively engaged within the MEMS community, with our range of Raider products for electrochemical deposition, or ECD,” says Mike Rosa, the company’s Director of Strategy and Technical Marketing for Emerging Technologies. According to Rosa, electroplating for MEMS applications is experiencing new growth through technologies spurred by the Internet of Things (IoT) and mobile applications.
Figure 3. Competitive landscape of electroplating equipment providers.
For MEMS manufacturers, there is a spectrum of needs — from low-volume high-precision plating to high-volume low-precision plating — as well as all points in between, for a range of materials and substrate sizes. A plating tool’s degree of automation and capacity are good predictors of the price tag it will carry. For used as well as new equipment, not all combinations are available for all substrate sizes. The trend in MEMS manufacturing now seems to be towards a higher degree of automation and precision, as new processes such as TSVs and wafer level packaging are increasingly being explored and adopted. “I think that there are good automated systems available,” says University of Michigan’s Herrera-Fierro. “When we were designing our electroplating equipment, we contacted several vendors, and it seems that there is a wide variety of options in the market.”
The more affordable automated tools may indeed be the best option for most mid-size foundries. “For the price of an electroplating tool, our target would be around $500,000,” says Philippe Azoley, the former Director of Industrial Operations at Tronics Microsystems (Azoley now works for TowerJazz in Israel). Azoley is interested in some of the new offerings though. “ClassOne proposes an interesting solution for us, with a semi-automated machine that allows us to do production at medium volume,” he adds. According to Azoley, less expensive automated or semi-automated tools are useful for companies such as Tronics because homemade equipment cannot be easily scaled for volume production, while big automated equipment is too expensive and requires very high production volumes.
The MEMS market is growing and maturing. Like other industries, it is becoming more focused on operational efficiency and manufacturing cost. The “tried and true” methods that are increasingly being adopted from other industries, such as improved process control and yield management, as well as the adoption of materials and processes, have matured to the point where both costs and risks are low, making them good candidates for MEMS manufacturing. The use of automation production tools has been shown to reduce variability and allow advanced process controls to be used for yield enhancement and cost reduction. When combined with the increased adoption of electroplating, a shift away from repurposed legacy tools and towards the use of new, automated tools will be accelerated, as equipment suppliers bring new choices to the market.
This article is a part of MEMS Journal's ongoing market research project in the area of MEMS manufacturing equipment and technologies. If you would like to receive our comprehensive market research report on this topic, please contact Dr. Mike Pinelis at firstname.lastname@example.org for more information about rates and report contents.
Copyright 2015 MEMS Journal, Inc.