Driven by production adoption, happy days are here at last.
In an era of slower growth and limited investment, wafer-level packages are bucking the trend, with unit
volumes expected to top 12.5 billion die this year. What is driving WLP shipments in what is for many package types an otherwise slow year? And what is the impact on major IC package contract assemblers?
As end-users continue to strongly prefer small form factor, low-profile consumer products, small packages such as WLPs meet the requirements. Demand for thin packages and greater functionality in smaller spaces drives the increased adoption of WLPs in mobile phones. Cellphone makers are increasing WLP use. Under the most pessimistic scenario, more than a billion handsets are expected to ship this year. Many Japanese DoCoMo phones contain four to nine WLPs per handset. Shipments are growing as many companies convert leaded packages into WLPs in the search for low-profile packages. In addition, the perception is WLPs are less expensive because they have no substrate or wire bonds. Many companies are designing wireless devices in WLPs, resulting in growth for 300 mm wafers, whereas many early WLPs were fabricated on 200 mm or smaller wafers. However, demand for 200 mm wafers is also increasing.
WLPs are also found in an increased number of consumer products (PDAs, watches, MP3 players, cameras, and digital camcorders), and are expected to see greater use in laptops.
Issues and improvements. Wafer-level packaging requires a good material set and process, as well as test and backend processes such as wafer thinning and sawing. This can be challenging with copper low-k wafers at the 65 and 40 nm nodes. For this reason, companies are looking at improvements in wafer-thinning technology and use of laser singulation to prevent chipping and cracking.
WLPs, often with underfill, are increasingly able to pass the drop test (one of the most important tests for cellphone makers). New processes to make WLPs more reliable are key to greater adoption. Process improvements include increased dielectric thickness, new materials with improved properties, and improved designs. Unfortunately, the wide range of Pb-free alloys for WLP solder balls makes it difficult to predict reliability from supplier to supplier.
Traditionally, WLPs have been used for a variety of devices with low pin counts and small die sizes. WLPs are used for power MOSFETs, power management, controllers, ringtones for mobile phones, battery management devices, integrated passives, DC-DC converters and some memory devices. RF components such as Bluetooth and wireless LAN devices, as well as diodes, EMI filters, and devices for ESD protection are packaged in WLPs. Many of today’s cellphones contain CMOS image sensors packaged in WLPs using ShellCase technology (now owned by Tessera). The latest image sensors are being fabricated with TSV technology to reduce the size of the camera module for cellphones.
Fan-out WLPs. With increased interest in using WLPs for larger die with higher pin counts, an increasing number of companies are looking at fan-out WLPs. Casio Micronics’ copper post technology has been used for years in cellphones and consumer products such as watches. Infineon developed a “molded reconfigured wafer” technology for its fan-out WLP solution, called eWLB (Figure 1).
The chip backside and edges are covered with a mold compound and array interconnect solder balls attached to the active side of the die. This permits use of a fan-out redistribution. The technology is targeted at medium-to-high-I/O count (<300) devices such as baseband processors and multiband transceivers. Several companies have licensed eWLB and production lines have been installed at ASE and STATS ChipPAC. Infineon, STATS ChipPAC and STMicroelectronics are proposing eWLB as a standard for future-generation products.
Capacity expansion. WLP capacity, especially for 300 mm wafers, is in short supply at several major IC package subcontractors, especially in certain geographies. The transport time and logistics to move wafers from the fab to the WLP processing operation cuts into the tight time-to-market that many semiconductor makers are experiencing. Frequent flyer miles are no substitute for WLP stocks. In response, some subcontract assemblers have placed customers on allocation. The announced capacity additions are a welcome signal for advanced packaging equipment suppliers, as they benefit from investment. Happy days are here again – at least in advanced packaging. CA
E. Jan Vardaman is president of TechSearch International, (techsearchinc.com); jan@techsearchinc.com. Her column appears bimonthly.
The supply chain works best when diligence and consistency rule the relationships.
With this writing we come to the last in our series on the Seven Deadly Sins (and their counterparts, the Seven Heavenly Virtues) of Outsourcing. We have enjoyed putting these articles together and want to
thank Circuits Assembly readers for their indulgence. The Seventh Sin is Acecia, which translates as Sloth. Its outsourcing equivalent is Apathy, and its Virtuous counterpart is Diligence and Consistency.
Before embarking on long-winded analysis of Diligence versus Apathy, we would remind readers of the legend of the Gordian knot. The way the story goes is shortly after the citizens of Athens were told by the gods that whoever rode to town on an ox-driven wagon should be crowned king, a gentleman named Gordius (or his son Midas, some accounts say) pulled into the center of town on exactly the right ride! After being crowned king, Gordius decided to show his gratitude to the Athenians by tying his famous wagon to the Acropolis (with a particularly impressive knot) and proclaiming that whoever could untie it would become his successor.
However, Gordius wasn’t as generous as it might seem, as the knot was actually impossible to untie: It was made from a piece of rope whose two ends had been spliced together (arguably a good analogy to the outsourcing model in place today). Ultimately, Alexander the Great solved the problem with a sword and thereby was handed the crown (and Asia) in 331 BC. (Some say this was cheating. But if the knot had no ends, wasn’t cutting it the correct solution?)
Back to our sin du jour. As we have said often, those on the front lines of the outsourcing relationship are some of the hardest-working, least-appreciated workers in the electronics industry. Day in and day out, they perform the tactical, task-based activities required to maintain an outsourcing program (at an OEM or from within an EMS), which unfortunately also means they ultimately get blamed when something goes wrong (and inevitably it does).
Why is it inevitable that things will go wrong? Because the task is as impossible as the Gordian knot. But move away from the day-to-day grind of managing outsourcing and get assigned to anything “strategic” and you’re on Easy Street. In other words, in today’s business climate (or perhaps the entire Western World), being “tactical” equates to career suicide, while being “strategic” equates to flying high. But what of the needs of the company? A successful outsourcing relationship is hard work. There isn’t an Easy Button. Manufacturing electronics products is not an exact science: It’s messy, and things go wrong. It can seem like a Gordian knot to those charged with implementing it.
This is especially the case when you try to solve manufacturing problems a half-dozen (or more) time zones away. Plus, supply solutions increasingly have become interdependent; planning is based on forecasts shared up and down the supply chain. When the OEM forecast is off (as it usually is), small problems quickly become large ones, and those on the front lines find themselves in the intolerable position of having to put out fires, while simultaneously dodging bullets from coworkers and management. A Chinese Foxconn worker recently committed suicide after being questioned over a next-generation Apple iPhone prototype that went missing from his possession. It’s an extreme case, but it gave many in the industry pause, as we consider the pressures he must have been under.
Is the sword the only solution to outsourcing as currently practiced? Should outsourcing be abandoned and everything brought back in-house? Is our industry so plagued by Sloth and Apathy that we are no longer able (willing?) to do the heavy lifting and accomplish the tactical, tasked-based activities so critical to a successful outsourcing relationship?
Perhaps in some cases OEMs should consider changing direction – a few have started or are thinking about doing so – but in most cases, we suggest not. We still see companies that manage outsourcing extremely well: OEMs and their EMS suppliers able to work together by practicing Diligence and Consistency, day-in and day-out. There is nothing strategic about it – just plain hard work by two parties that have a job to do and are committed to excellence.
Charlie Barnhart is cofounder and principal of Charlie Barnhart & Associates (charliebarnhart.com); charlie@charliebarnhart.com.
In May, after its Electronics Assembly Systems business had lost tens of millions of euros over the past few quarters, parent Siemens reclassified the unit for “disposal.” With more than 22,000 installments worldwide, and a huge market share in Europe, the unit might be considered a prize to some. However, the current
outlook for electronics equipment is slow, and several competitors are on the block as well. Yet SEAS pushes on, introducing a new placement head (see Equipment Advances, March 2009) and the SX placement platform (see http://www.circuitsassembly.com/cms/component/content/article/120/8242-siemens-rolls-out-siplace-sx-placement-machine). It was against this backdrop that SEAS CEO Guenter Lauber spoke with Mike Buetow in July. Excerpts.
CA: According to Siemens, SEAS could be integrated into other Siemens segments, divested, rolled into a joint venture, or closed. Which of these options is most likely?
GL: What is most likely is a partnership to go forward and continue our successful way to grow the business. Health care, automation and infrastructure, and energy are Siemens’ three cores. With our setup and customer structure, we don’t fit those core strategies. Additionally, we don’t fit in the Siemens organization with the Siemens regional setup anymore; we wouldn’t get the opportunity to develop the Siplace business in the best possible way; thus the carve-out of the Siplace business was decided some time ago.
Since Jan 1, 2009, we are a separate legal entity. We are 100% owned by Siemens, but we report not to any of the three core sectors, but directly to the Siemens board as a separate company. This helps us in terms of controlling our business in a specific way. All our regional companies with the regional and local sales and service teams are directly controlled, supported and managed by Siplace management. In doing so, we set up our business in seven clusters. Before, we were part of the Siemens business in each region, which meant we had a far bigger structure than was eventually needed, with a lot of overhead positions and processes, and a lot of different reporting structures, which could make things a bit complicated from time to time. Now, we can decide to be only where the customers are. In doing so, we could save significant money. For example, we don’t have to pay for marketing costs for dual efforts. A lot of those examples helped us reduce our costs, and we can better support our customers.
CA: What are those clusters?
GL: The Americas – North and South; China; Rest of Asia; Germany; Northwest Europe (Scandinavia, UK); Southwest (France, Italy, Spain, Portugal); Central/Eastern Europe (Austria and the rest of Eastern Europe). There is no general manager for each region, just a sales manager in each cluster running the sales and service organization.
CA: What would the ideal partner look like?
GL: There are several options: One would be a strong partner in the industry with similar products, same or different but with synergies. Other possibilities would be a sale or a joint venture, or even remaining an individual company under the Siemens roof.
CA: What do you think the SEAS unit’s major appeal would be to a prospective partner?
GL: Definitely our strength in innovation and technology. We are an innovation driver. That would be a real appeal, with our know-how and products. Of all technology and equipment suppliers, we come closest to the ideal of real on-demand production, thanks in part to our strengths in software development and integration.
CA: What would be the risks of acquiring the unit?
GL: With the new SEAS, with new setup and restructuring done for the future, and with a global team like this, I personally don’t see a risk that could jeopardize the success of any future arrangement.
CA: One might point to the forecasts for equipment sales over the next few years and say the outlook poses a risk.
GL: Electronics equipment suppliers have always been part of a very volatile market. So downturns are not really new to us. Consequently, it also means the market will recover, at which level we cannot say right now, of course. But be assured, SEAS is in very good shape, not only in its setup, but also with its new and innovative products.
CA: Is an employee or management buyout of the unit a consideration?
GL: Again, why should we exclude any options and limit our flexibility? We will analyze each realistic option that can make us more competitive and successful.
CA: You appear confident of SEAS’ future.
GL: We went through this carve-out and it helped us to save some costs. We are an equipment supplier and solution provider, and now, we are more acting like a mid-size company – flexible, and reacting very quickly to customer demands. We feel very comfortable under the Siemens umbrella. It makes me very happy about our future. Also, looking into our product portfolio, the SX placement machine has been launched and we’re already bringing it into the market. This gives us a bright and strong future.
SEAS has installed more than 22,000 placement machines around the world. In each region we are one of the leading equipment suppliers, and in Europe we are by far No. 1. Within our service business, the share of classic support and after-sales services is, in relative terms, on the decline. On the other hand, the share of ambitious, customer-specific process improvement projects is rising steadily, because they enable customers to benefit from double-digit efficiency improvements and significant cost reductions. These start from new built-to-order concepts, which eliminate rigid manufacturing concepts and inflexible SMT lines. This means new technologies are required that make the flexible line reality.
Two recently unveiled innovations – the MultiStar CPP head and the modular gantry SX-Line, which separates the investment of feeder capacity and placement performance – bring us a lot closer to the holy grail of on-demand production. This is the future of electronics production. And rest assured, at Productronica we have something really exciting in our pipeline.
From Jim Collins to Michael Porter, the latest generation of management gurus argues companies must focus on core competencies and shed all other activities.
Just what makes a “core competency,” however, is always in flux. And as electronics companies see sales plunging like cliff divers, they are quickly redefining the terms.
With its August launch of the Nokia Booklet 3G, Nokia, long synonymous with mobile phones, has now officially entered the netbook market. This makes Nokia just the latest in a string of high-profile OEMs that are trying to jump-start their revenues by going after what are increasingly commodity markets.
To wit:
- Dell, in conjunction with China Mobile, is said to be looking at jumping in the mobile phone wars. In doing so, the world’s No. 2 computer maker would join Hewlett-Packard, Acer and Asustek as PC OEMs that either have launched or are planning to debut smartphones.
- Meanwhile, China Mobile, AT&T and Far EasTone Telecommunications are among the mobile providers now pitching netbooks, the small-sized, low-priced mini PCs.
In The Wall Street Journal on Aug. 24, Roger Yuen, vice president of Acer’s Asia-Pacific smart handheld business group, explained the fascination. “[I]t is relatively easy for PC makers to make smartphones because the two devices share similar components and software.”
Which makes sense to analysts, I suppose, but is something of an insider’s joke in electronics manufacturing. After all, what today doesn’t have Intel Inside?
The moves are highly questionable. As the WSJ notes, “Analysts say PC makers are unlikely to reap significant benefits in the near term, as they need to develop better relationships with mobile operators to sell their products. It will also take time to develop differentiated products and market their own brands in a segment where consumers already have many choices.”
The WSJ hedges, adding, “[M]any agree that longer-term, PC makers have a chance to gain share, which would generate a new source of revenue growth and improve overall profitability.”
I don’t see it. These are extraordinarily competitive markets, flush with big-name brands with deep pockets. Lenovo. Nokia. Samsung. Dell. H-P. The list goes on. None is going to give in without a (very expensive) fight.
Meanwhile, the broader markets are showing some signs of leveling: Worldwide mobile phone sales fell 6.1% year-over-year to 286.1 million units during the second quarter. And the battle for the niche markets – like smartphones – may already be over. Nokia holds a 47% share of that market, and RIM has been entrenched in second place.
New players have found the going bumpy. Take for example, Apple’s much-ballyhooed entry, the iPhone. Measured in terms of style and buzz, it has performed exceedingly well. In terms of units sold, it’s another story. Apple shipped 5.4 million units in the second quarter, says research firm Gartner, good for 2.4% market share. It is almost as if Apple makes the equipment as a medium to sell its highly profitable catalog of digital music.
Given Apple’s lagging market share even in the face of its marketing and design savvy, it’s hard to fathom why other, less sophisticated companies would risk dominance in one market to attempt to conquer such foreboding – and possibly worthless – terrain.
It brings to mind one more business truism: that the grass – and the profit – is always greener on the other side of the fence.
PCB West. Speaking of green, one way to earn it is by knowing more about your field than anyone else. To that end, I strongly recommend readers attend PCB West later this month. This Silicon Valley-based show, now in its 18th year, is an outstanding forum for the entire PCB supply chain. (Disclosure: PCB West is sponsored by UP Media Group, the parent of Circuits Assembly.) PCB West will be held Sept. 14-18, at the Santa Clara (CA) Marriott. Show details are at pcbwest.com. Please support the companies that bring you this magazine each month by stopping by.
SPI market leader Koh Young is shrugging off the recession and setting its sights on Europe.
On the subject of expansion, every business case study says the best time to do it is during a downturn. Take advantage of the market lull to grab market share, the researchers say.
Few do. It’s much safer to protect cash and other assets than to risk failure (and one’s job). In other words, follow the herd.
Following the herd isn’t what Koh Young Technology (kohyoung.com) is about. The vision machine maker in May opened the doors on its pristine new demo and training center in Alzenau, Germany, about 45 minutes from Frankfurt’s international airport. (The event coincided with a less auspicious debut of its Ireland office, which will handle sales, support and administrative functions.)
The new offices are set to serve continental Europe, the UK and Ireland, where the company through May had sold 230 of its 940 installations worldwide.
In making the announcement, founder and chief executive Dr. Kwangill Koh said, “Today we celebrate a significant milestone in the growth of Koh Young as a global company. The European market a key growth area, and the establishment a strong presence here, in conjunction with the introduction of our revolutionary new 3-D AOI technology, will fuel that growth.”
That “revolutionary new 3-D AOI technology,” of course, refers to the Zenith machine, which debuted at Apex in April and has been making media waves ever since.
Why Europe, and why now?
The stage for Koh Young’s migration to Europe was set by Pieter Stins, a veteran of Vitronics-Soltec and Nutek, who saw a Koh Young SPI machine at Productronica and, though on the cusp of retirement, decided to buy one and build a new distribution company around it. (That company, PPT, is now Koh Young’s distributor in Germany, Austria, Switzerland and parts of Eastern Central Europe.) Meanwhile, in Asia, Koh Young was going gangbusters, and looking to extend its reach abroad.
Having gone public on Korea’s KOSDAQ exchange in June of last year, and hit $35 million in sales in fiscal 2008, Dr. Koh embarked on a five-year plan with the goal to reach $150 million in annual sales, which would make it by far the largest OEM of electronics assembly inspection equipment (Table 1).
Seeking to make its mark on the West, Koh Young turned to two well-known industry veterans: Harald Eppinger and Thorsten Niermeyer. Niermeyer, who previously worked for Agilent and MVP, was named global sales director, based in Ireland, while Eppinger runs the Germany office as European sales manager. It also brought aboard André Myny as global marketing director, a role he previously handled at Vitronics-Soltec.
Its staff in place, the company was ready to make its move. After some discussion and due diligence, it settled on Alzenau, which offered ready access to the key German market, which still dominates electronics manufacturing in Western Europe, but was centrally located and accessible to both the traditional pockets in Italy and France, and the emerging markets of Eastern Europe.
The new center has in place the full range of the Koh Young platform, including the previously released 8030-2 and 8030-3 SPI machines, on which the company built its name and reputation; a KY-3020T, which is the semiautomatic tabletop version; and the aSPIre-2 SPI, the four-way light projection system that also debuted at Apex.
The new site also features an Ekra screen printer for hands-on process training with customers.
The highlight, however, is the new automated optical inspection system, now named Zenith, which is the first to feature 3-D inspection. The machine senses and measures the z-axis profilometry of whole assembled PCB surfaces, including components, solder joints, patterns, holes, and foreign material. The system, which features eight-way projection in order to compensate for the taller solder joints and paste, can run conventional 2-D AOI as well.
Zenith is designed for pre- or post-reflow, but as of the May open house had not yet been tested on mixed-technology boards. (Most AOI traditionally have struggled with plated through-holes.) Still, the distributors on hand at the May open house, including Danutek (danutek.com), LifeTek (lifetek.it), Amtech (amtech.cz) and several others, some of which brought boards to test on it, were visibly impressed.
“I do believe the Zenith is a unique approach to AOI and will guide industries and competitors to a new era of defect detection and measured process control,” said Giovanni Scotece of Meda, Italy-based LifeTek.
With the opening of the European offices, the company hopes to build an AOI expert center, expand its service organization, and create an innovation center for new products, which could include solar, although the executives were a bit coy about it.
More expansion is ahead, Dr. Koh said. The company in December will move to a new headquarters in western Seoul, which will include an R&D center and manufacturing facility. The new site will be 77,000 sq. ft., or 89% larger than the company’s current digs. It will house most of the firm’s 120 staff, seven of whom have a Ph.D., and 32 of whom have a master’s degree.
Koh Young has in its sights the inspection markets for PCB, wafer, substrate and even solar cell markets. It’s an ambitious plan, but the company hasn’t missed its targets yet.
Mike Buetow is editor in chief of Circuits Assembly; mbuetow@upmediagroup.com.
Understanding how mechanical anomalies can cause electrical failure in MLCCs.
When engineers go through the task of pinning down the cause of one or more field failures, the cause often turns out to be a damaged or failed multilayer ceramic chip capacitor. The nature of a field failure caused by a bad capacitor may range from total system failure to a failure of one or more system functions, to an intermittent failure, the latter of which can mimic a software problem.
Here, we look at the mechanical causes of failures in MLCCs from the fabrication of the capacitor through system assembly. Much of the detailed information comes from Sonoscan’s applications laboratories, which have imaged acoustically and analyzed tens of millions of MLCCs.
An understanding of how mechanical anomalies can cause electrical failure in MLCCs is important for three reasons:
1) End-of-line electrical testing usually does not detect a mechanical anomaly because the electrical signature of the anomaly is minute or absent at this time.
2) A mechanical anomaly may endure weeks or months of service use before it changes or expands enough to reveal itself as an electrical failure.
3) The number of field failures resulting from a single type of mechanical anomaly can be large.
Defects introduced during manufacture. MLCCs are made by laying down alternate layers of dielectric and electrode materials, and then firing the capacitors. Of the three most frequent types of internal damage (voids, delaminations and cracks), two – voids and delaminations – can form during the manufacturing process. Cracks as a result of manufacturing processes were frequent a decade or two ago, but have been made rare by more precise control over the processing. Voids are most often tiny air bubbles trapped within the capacitor. A void can be very much flattened – its width might be 100 times its height – but dielectric material will be missing from one or more layers. A delamination is simply a non-bonded area between layers, without loss of dielectric material. When an MLCC is cross-sectioned, a void is typically lens-shaped (because it is a bubble flattened by the pressure of overlying layers), while a delamination is simply a thin horizontal gap.
Voids and delaminations can be the precursors of cracks. Cracks also can be caused by variable porosity in the dielectric layers, a condition in which irregularly distributed microscopic air bubbles weaken the ceramic.
The photo on the cover of this month's issue shows the acoustic image of an MLCC that contains a single void. In performing acoustic imaging of an MLCC, the scanning ultrasonic transducer of the acoustic microscope pulses ultrasound into the MLCC several thousand times a second as it moves back and forth across the MLCC. A fraction of a microsecond after pulsing, it also registers the return echoes from within the MLCC. System software typically accepts return echoes from just below the top surface of the MLCC to just above the bottom surface, a technique known as bulk imaging. A defect-free MLCC will send back no return echoes, but an MLCC having any gap-type defect such as a void, delamination or crack will send back very high amplitude echoes from the defect. (Here, the void appears red, which is highest level on the color map at left, because the solid-to-air interface reflects nearly all of the ultrasound.) The amplitude is very high because of the extreme difference in acoustic properties between the solid material of the capacitor and the air inside the defect. It is this interface between a solid and gas that reflects virtually 100% of the pulsed ultrasound and produces an acoustic image of the anomaly.
Voids such as the one shown on the cover are important because they can cause long-term failures in MLCCs. For example, a void may be located in the middle of one layer of dielectric. This may seem like a harmless location, and in some capacitors it is harmless. But the electric field between two electrodes may cause metal to migrate and to plate the inner surface of the void. Eventually, a weak current may begin to flow between the electrodes. Ultimately – weeks or months after the beginning of field service, perhaps – the metal plating becomes substantial enough to cause a short. Many of the MLCCs imaged in Sonoscan’s laboratory are destined for aerospace or military applications, where a slowly developing short or other internal anomaly can have catastrophic results.
A crack, like a void, also can experience metal migration and become plated. Under some conditions, the metal plating within the void or crack may diminish or disappear. This is likely to happen if the capacitor is exposed to infrequent higher voltages. If a capacitor is suspected of having a short, it may intentionally be exposed to a high-voltage current. This crude method may repair the capacitor temporarily, although the resumption of normal lower voltages may restart the plating process. Intermittent failures can occur in this manner (but see the information on cracks below). A crack also can cause a short with no metal migration at all. If the crack extends through multiple electrode layers, thermal or mechanical forces can move layers until electrodes having different polarities come in contact with each other.
Less frequent types of anomalies also can occur during fabrication of an MLCC. Very occasionally, airborne organic particles can find their way into the layers. The small white spots in the acoustic image of the capacitor in Figure 1 are ordinary, microscopic dust particles trapped during fabrication and burned off during firing of the capacitor. They leave tiny empty voids that are strong reflectors of ultrasound. The longer feature (left of center) is a microscopic fiber of organic material, which, when burned off, leaves a characteristically elongate void that, like any other void, has the potential to cause an eventual short.
When adjacent layers in the MLCC are not bonded during manufacture, the result is thin air-filled delamination. Delaminations can be large in area, and a single MLCC may have multiple delaminations at different depths. When the MLCC is imaged acoustically from above, the delaminations may overlap (Figure 2), where essentially the entire area of the MLCC has a delamination at least at one depth, and where the yellow edges (marked by arrows) of some of the individual delaminations can be made out.
Delaminations are generally very thin, but they can be imaged acoustically even if the vertical extent of the gap is as few as 0.01 µm. Unlike some voids, delaminations are not typically in a location where they can breach a dielectric layer, yet they can cause shorts, presumably because delaminations may make it easier for a crack to form.
Defects in the unmounted MLCCs that flow into Sonoscan’s applications laboratories for imaging are most likely to be voids or delaminations. Cracks are occasionally seen, but are more likely to result from handling rather than manufacture. Special techniques are sometimes needed for imaging. Some newer MLCCs have acquired additional layers in order to increase capacitance without increasing footprint; as a result, the MLCC has become square in end view, and the operator of the acoustic microscope cannot tell which side is up. A delamination that is obvious when horizontal is hard to image when it becomes a vertical knife-edge. To avoid the tedious labor of turning square MLCCs by 90° and imaging them twice, Sonoscan has developed a technique (Figure 3) that images cracks and delaminations, regardless of orientation.
Defects introduced during assembly. Damage to MLCCs during the early stage of assembly is likely to be caused by handling. The MLCC is picked up by a vacuum tool or tweezers and is either placed on bond pads in preparation for reflow, or glued to the board in preparation for wave soldering. In either case, damage – usually in the form of a crack – is possible.
Damage also can occur during reflow and wave soldering, when the MLCC receives a thermal shock that can aggravate existing internal stresses until a crack forms. MLCCs are much less susceptible to moisture-related damage than plastic-packaged ICs. In theory, moisture can collect within a capacitor and fill an existing void (with destructive results when the moisture flashes into steam), but such events are far less common in MLCCs than they are in plastic-packaged ICs. The very small dimensions of some MLCCs also may make it harder for moisture-related damage to occur because the much higher ratio of surface area to volume lets moisture escape more rapidly.
Whether a damaged MLCC can be identified during end-of-line electrical testing depends first on the extent of the damage. A small crack or other anomaly that has not (yet) created contact between adjacent electrodes will not be identified, even though it may expand and cause an electrical failure later. An anomaly that has created a leakage current within the MLCC might be found if the MLCC serves an essential function within the signal path. But an MLCC in a decoupling role can be found only by examining the noise level in the supply lines.
These constraints make acoustic imaging of MLCCs after reflow or wave soldering an important tool. Finding cracks in a significant number of MLCCs, or in MLCCs at specific locations on the board, gives the opportunity to change process parameters and remove the stresses generating the cracks. Acoustic imaging is often carried out during R&D or during pilot production, but also may be used periodically during full production to eliminate the possibility of field failures.
During assembly, large panels are in some fashion separated into individual printed wiring boards. Cracks can form in MLCCs during the separation process. These cracks are most likely induced when the panel sections are snapped apart, but they can occur with other methods of separation and are sometimes more frequent near the edges of the board, where mechanical stresses are presumably higher. They are generally more likely when Pb-free solders are used, because Pb-free solders are quite rigid. SnPb solder is more plastic and better at absorbing mechanical stresses. Separation-related cracks are likely to be near the terminations and are likely to be vertical, and are most easily found with the method for vertical crack detectiondeveloped at Sonoscan.
When cracks caused by the panel separation process are suspected of causing field failures, their role can be clarified by first performing acoustic imaging on unmounted MLCCs, preferably from the same lot. Unmounted MLCCs with no internal defects go through assembly and are imaged acoustically again after reflow or wave soldering, to remove those steps as causes of the cracks. The panel is then separated by the same method used for the MLCCs that caused field failures, and the boards examined acoustically for telltale vertical cracks near the terminations.
Tom Adams is a consultant at Sonoscan, Inc. (sonoscan.com); tom100adams@comcast.net.
