Will new ‘Bear’ emerge from the bear market?
China has been impacted by the global recession, just as other regions have, and recently lowered its GDP estimates from 7.5% to 6.5%. However, it is the response to the global recession and what happens as China emerges that may forever change the face of the country and its role in the world economy, not to mention in the electronics industry.
Like other governments around the world, China has adopted stimulus programs. Last November, it announced a $585.2 billion package, which included plans to invest $290 billion on railway network expansion by 2011. China also instituted a program called “home electronics down to the countryside.” This is a remarkable notion, given the nation’s size, population and rural makeup. (Rural China has an estimated 320,000 villages and population of 720 million.) China’s farmer subsidy program offers a 13% rebate to rural households on 10 types of Chinese-produced electronics, including selected brands of TVs, refrigerators, mobile phone handsets and PCs. In many cases, the goods are from domestic companies. According to Goldman Sachs, sales in each subsidized product category have increased 40% year-over-year and 30% above the national average. The subsidized brands represent 80% to 90% of total sales in the category, indicating cannibalization of nonsubsidized brands. PCs were added in February.
Some analysts do not see the farmer’s subsidy program as having an immediate, large impact on semiconductors, but over time it may be more important. While semiconductor companies are reporting increased foundry sales, Goldman estimates Chinese farmers consumed only 2% of foundry sales in 2008, and orders are probably the result of discrepancies in inventory stocking in thousands of sales outlets across China. Much of the sales are for home appliances, not PCs or handsets. In April, Goldman analysts reported that China’s rural electronics sales might not add much to its PC and handset sales in 2009. Several reasons were given, including poor Internet access in rural areas, the determining factor in how useful a PC might be. Learning to use PCs will also take time. The analysts noted the handset price for the subsidized products is not much lower than the “village” models.
Over the next 10 years, however, stimulating domestic demand may change China’s status as a net exporter and spur growth in the domestic electronics industry. According to CCID Consulting, exports account for more than 70% of the sales volume in the domestic IC industry, and exports directly determine industry performance. China’s IC industry saw significant changes in 2008: The export and development rate braked, and the quarter-by-quarter basis slowed to 5% versus 24.3% in 2007. Enter the stimulus. In the past five years, the Chinese government influenced the investment of about $7 billion in new fabs, SEMI reports. In the next five years, another $20 billion to $25 billion will be invested throughout the country. Going forward, the central government will invest up to $30 billion in the industry by 2020.
Many companies have established IC packaging and assembly facilities in China. Years ago, the main products were leadframes. Today, companies have expanded into advanced packages such as BGAs and CSPs. Solder bumping capacity now makes up about 5% of the world market. With the display industry growth, gold bumping and assembly have expanded, the former to 4.8% in 2008. Companies with gold bumping facilities in China include JCAP (based on bumping technology from APS) and Chipmore.
Much of the IC industry is concentrated near Shanghai. Amkor’s China assembly sites are located in the Waigaoqiao Free Trade Zone in Pudong, Shanghai. Others in the area include Jiangyin Changdian Advance Package Co., Ltd. (JCAP), Millennium STATSChipPAC; ASE; ChipMOS; and United Test and Assembly Center. Carsem has a plant in Suzhou, as do China Wafer Level CSP, Ltd. and EEMS. ASAT has a manufacturing facility in Dongguan.
Flip chip assembly and bumping is also expanding in China. Unisem has an assembly and test facility in Chengdu. UTAC also has a JV with SMIC in Chengdu.
Besides the chip investment, China’s economy has bright spots. Boosted by government initiatives and green spending, its IT market will be minimally affected by the global economic slowdown, according to a report from Springboard Research. IT spending in the country will reach $51.2 billion in 2009, up 11% year-on-year.1
Government stimuli provide promise for mobile handsets, white goods, telecom infrastructure and semiconductors.
As China tries to create domestic demand, its economy could turn from export-oriented to consumer-based. While this will take many years, today’s actions may have positive long-term ramifications. As many organizations in regions around the world find it difficult to increase R&D spending, China’s emphasis on science and technology may prove an important development to its domestic electronics industry.
References
1. ZDNet Asia, “China IT Spending in the Country is Expected to Grow by 11% Y/Y to $51.2 billion in 2009,” Feb. 6, 2009.
E. Jan Vardaman is president of TechSearch International (techsearchinc.com); jan@techsearch
inc.com. Her column appears bimonthly.
How gluttonous OEMs are ruining an industry.
The fifth deadly sin in our series is that of Gluttony, which typically refers to eating too much food. Because most electronics manufacturers are not in the habit of eating printed circuit boards, how does that translate to outsourcing?
In contrast to the sins of Greed or Avarice, Gluttony focuses on the destructive consequences that occur when we consume blindly without appreciation. It’s the difference between a gourmet who savors and a gourmand who gobbles: The latter seeks quantity, not quality, piggishly exhausting resources and looking around for more. An OEM that believes in unlimited EMS manufacturing capacity, and jumps from one geography to another chasing penny savings in labor costs, is blind to the value of manufacturing. That OEM disregards the decades of tribal knowledge the EMS supplier has acquired, and the complexity of the skills required to build electronics.
Consider a child looking through a cardboard tube at a magnificent Gothic stained glass cathedral window. He might see one small blue piece of glass, or a green one, but it is impossible for him to see the whole triumphant scene. If he doesn’t like green, or that particular shade of blue, he will get bored and think stained glass is stupid, but is he qualified through experience or background to judge? The electronics outsourcing strategy now has a three-decade history. Some OEM operations managers have limited experience manufacturing electronics. Are they looking at the scene through a cardboard tube? Some have the attitude that “only losers are in manufacturing.” They treat EMS suppliers like we treat car rentals: The prevailing mentality is to run over the curb and slam on the brakes. Who cares? When an EMS supplier’s services are seen as an interchangeable commodity, reproducible easily in any geography, it’s not surprising that the quality of electronic products deteriorates.
Some of our firm’s leading indicators result from tracking electronics outsourcing programs that fail. The failure rate of products coming out of China is quite high, and many programs, especially among low-volume/high-mix products, have moved to other geographies, closer to the end-consumer. One reason for these failures is up to 30% of the components used in products manufactured in China are counterfeit, by some estimates. The problems are well documented. Is anyone worried that certain medical electronics are now manufactured in China? Many of these failures were woefully predictable: When the wrong type of program goes to the wrong geography, failure is virtually guaranteed. The outcome is as obvious as a car that won’t start because of a missing engine. And moving from one EMS supplier to another racks up astounding waste. Millions of dollars are spent chasing miniscule labor-cost savings. This waste is an example of the consequence of Gluttony, of mindless consumption of manufacturing services while constantly beating down suppliers on cost.
Since the economic crisis began, there are some who ask the question, Is American manufacturing coming back? As millions of service sector jobs evaporate, some long for the days when we made things. In the case of the electronics industry, the answer to the question of when manufacturing will return to US soil probably is “never.” To say the playing field today in the global marketplace is not level is a colossal understatement. It is not only unlevel, it is replete with mountains, canyons and valleys. The manufacturing geographies are coming to the market from a different socioeconomic worldview, and governments and citizens have different attitudes and objectives regarding manufacturing. Free market or not is irrelevant; conventional wisdom says we don’t want to manufacture anymore. It’s too 20th Century. This was made especially clear by a recent Kurt Andersen essay in Time:
“ ‘History doesn’t repeat itself, but it rhymes,’ Mark Twain is supposed to have said. If the 21st century rhymed, China would be the new us – feverish with individual and national drive, manufacturer to the world, growing like crazy, bigger and much more populous than the reigning superpower. ... [But] Muscular industrialism gets you only so far.”
In other words, manufacturing is somehow lower on the evolutionary scale than other activities. The shift in manufacturing to Asia and other low labor-cost geographies may soon be viewed as one of the biggest transfers of wealth in history, not to mention the tribal wisdom, engineering expertise and ability to leverage high-tech manufacturing for domestic purposes. We no longer have the political will or collective consciousness to maintain that capability, and unless something changes radically in our education system and culture, it is not likely to be revived.
But the upside of the current state of the EMS industry is that any OEM organization that operates with temperance and an appreciation of the complexity of the manufacturing process stands out like a shining beacon to the EMS community. Suppliers will rush to do business with these companies, and a little self-restraint will translate into loyal EMS suppliers that do everything in their power to help the OEM succeed. That’s the upside. It should be a viable incentive because a combination like that would be a powerful competitive differentiator.
Jennifer Read is cofounder of Charlie Barnhart and Associates (charliebarnhart.com); jennifer@charliebarnhart.com.
In the SMT world, a whole lot! But know the right attack angle, too.
If we compare the stencil printer to an automobile, squeegee blades are analogous to where the rubber meets the road. You would never consider putting low-performance tires on a high-performance sports car; that would diminish the purpose. The same idea applies to a stencil printer. To fully use the capability of a high-performance printer, choose the squeegee blade carefully. How do you do that? Hopefully this discussion will offer a good starting point in choosing the right blade for the right application.
In stencil printing, we see primarily two types of squeegee blade materials: metal and polyurethane. Polyurethane blades with a high durometer rating (90-110) have shown success in many applications, but as boards get denser and components get smaller, the only legitimate choice is to use a metal blade. The primary reason is that metal squeegee blades permit a more controlled and consistent print height across the entire board area, as compared to poly blades. Hence, this discussion is restricted to metal blades.
We are at a point where the line between SMT and semiconductor packaging is becoming blurry. Miniature components such as 01005 passives and 0.3 mm CSPs/BGAs demand the accuracy and precise deposition of solder paste volume, as do wafer bumping and other semiconductor processes. It is well known that stencil printing is a complex process, influenced by a number of variables that include hardware, software, materials and process factors. Squeegee blade assembly happens to be an element of printing that can have a significant effect on the print quality. Print quality is defined here by quantity of solder paste transferred, paste deposition profile, bridging and insufficients. Studies show most of the aforementioned qualities are affected by squeegee blade type and attack angle of the blade. Material hardness and surface characteristics not only affect deposition quality, but also affect the amount of paste wasted by adhering to the blade. The smoother the blade surface, the less the material will adhere to it. We also have seen, through experimental work, a blade hardness of approximately 75-80 on the Rockwell A scale is highly effective in regard to providing a clean stencil and desirable deflection angle. The effect of the squeegee blade’s attack angle will be discussed later.
Metal blade printing only has two print parameters that typically can be controlled: squeegee speed and downward squeegee pressure. The speed cannot be set so high that the paste does not roll as it moves across the stencil or so low that the print cycle time does not keep up with the manufacturing line. The blade pressure is usually set so that no paste remains on the stencil behind the squeegee. Excessive pressure will damage the stencil by either coining the image edges or breaking the fine webs between small pitch apertures. Higher pressures will also shear thin the paste to such an extent that the flux will separate from the metal and problems such as lack of tack at placement or solderability will occur down the assembly line. Figure 1 shows a typical printing process.
Contrary to popular belief, the paste does not fill the aperture until the paste bead has traveled at least 75% beyond the start of the aperture and fills from the front-to-back. Here the attack angle of the blade becomes critical.
Most squeegee blade assemblies are designed to provide a fixed contact angle between the blade and stencil. This contact angle changes as the print process begins due to application of print pressure. The angle between blade and stencil at the time of print process (with print pressure and speed in active mode) is known as the attack angle. This is the parameter to control to obtain optimum print quality. What is the optimum attack angle and how is it obtained? Clearly, this will depend on various process factors such as paste type, board finish, component type, etc. Studies show, on average, an attack angle of 50°-60° provides a more repeatable and accurate paste deposition compared to an attack angle of 30°-45°.
There are different ways to get to this attack angle. One option is to design a blade holder that can be fixed at a certain attack angle (Figure 2a). Another option is to induce this contact angle by manipulating the blade thickness. What actually controls the attack angle of the blade is the deflection of the blade under pressure. The thinner the blade material, the more compliant the blade becomes, which means that, with a fixed squeegee holder angle, a different contact angle is possible. Figure 2b shows this effect.
The figures demonstrate the influence of blade thickness on the ultimate attack angle. A thicker blade material remains rigid and maintains the fixed angle of the squeegee holder. Whereas the thinner blade (Figure 2b) provides a flexible contact angle (based on the blade type, thickness and squeegee pressure) by exhibiting more of a “leaf spring” effect during printing. This effect produces a better pumping action to fill an aperture, thereby permitting the paste to start filling an aperture earlier than the 75% beyond the beginning of an aperture.
What we can conclude, then, is there is more than one way to achieve the correct attack angle for a process. A process engineer should keep in mind there is no free lunch. While attack angle is important in providing a well formed, repeatable paste deposit, it should not come with the price tag of high print pressure and low print speed. A process engineer needs to fully understand the process needed to determine the correct squeegee blade assembly, and more specifically, the squeegee blade type.
Rita Mohanty, Ph.D., is director advanced development at Speedline Technologies (speedlinetech.com); rmohanty@speedlinetech.com.
I’m beginning to really dislike Greenpeace.
The organization was funnier as a target of light mocking in the hit TV show Seinfeld, as the landing spot for the fictional president of NBC who, on a whim to win the affections of Seinfeld co-star Elaine, foregoes his high-profile, high-paying job for a spot on a skiff running interference with whaling boats in the Pacific.
But when Greenpeace starts influencing the decision-making of major electronics OEMs – and not for the better – I start to get a little testy.
Last month, the environmental group issued its dubiously named Greenpeace Greener Electronics Guide. The contents were utterly predictable. As is its wont, Greenpeace took to task several PC makers for having failed to thus far eliminate vinyl plastic and brominated flame-retardants. “The phase-out of toxic substances is an urgent priority to help tackle the growing tide of e-waste,” the group wrote.
Wrong, all wrong. E-waste and toxic substances are two very different things. Yes, products should be designed to be recycled. But it doesn’t follow that discarded devices are necessarily poisoning the planet.
I firmly believe electronics has a leading role to play in the world, and not just as means for entertainment. In Hot, Flat, and Crowded, the latest book by Thomas Friedman (whose exhortations in The World is Flat formed the basis of the argument used to defend the en masse migration of manufacturing to China), he writes that simple makeovers often draw resources and attention away from the more important, and more needed, revolutions.
“I don’t think the leverage points now are in more consciousness-raising. … The danger is you think that if you change your light bulbs, you’ve solved the problem. My motto is change your leaders, not your light bulbs. … Without scale change right now, in terms of climate we’re really cooked.”
As Friedman thus implies, Greenpeace is wrong both in its stance and its methods. The greening of electronics cannot be accomplished in small steps such as the purge of PVCs and BFRs, and besides, the criteria for what’s “green” are rarely clear and always moving.
Still, Greenpeace is turning electronics OEMs into tobacco companies, and in turn we are playing into its hands. I’d offer Greenpeace’s leaders a very public deal: Want to yank BFR’s out of electronics? We’ll do it today, provided you underwrite the insurance policy for any consumer who gets injured as a result of the substitute materials – and for any environmental mitigation that may be needed – for infinity. (And while you’re at it, find an eco-friendly substitute for gas for your whale-watching boats.)
The sad truth is the technology does not exist for an electronics product to be “green.” Yes, some can be more energy efficient than others, but it’s a relative scale, not an absolute. Even if an electronics device itself could be made to be functionally energy neutral, the act of mining for the materials that compose the end-product is horribly ecologically unfriendly.
It’s also politically untenable. As I’ve noted in this space, the decade-long war in the equally dubious Democratic Republic of Congo is pulling in electronics manufacturers, which purchase thousands of tons of raw materials from Congo mines. The revenues are used to arm renegades, and at least 5.4 million people have died from the fighting, sparking campaigns by various groups for bans on buying minerals from Congolese companies.
The battle is making its way to US shores, and could all come to a head as domestic legislators prepare bills requiring companies to disclose – and prove – the origins of the minerals they buy. Critics of the proposed bans – and there are some – say such response would put the miners out of work, and with little other indigenous industry, drive them into lives of crime.
We can question government’s wisdom, but few lawmakers would be sufficiently vapid to promulgate a bill that would cost their constituents money, while simultaneously fueling the problem it proposes to solve.
Oh, wait…
Speaking of vapid, Apex next year begins immediately following Easter, which means much of the setup will take place on what at least two billion people consider the most holy day of the year (see our recap of this year's show, pp. 22-24). IPC should pull a rabbit out of its hat – and move the dates.
Photovoltaics, displays and sensors could lead an industry revolution.
In 2007, the iNEMI Roadmap included (for the first time) a chapter on Organic and Printed Electronics. For the 2009 edition, the scope of this chapter has been expanded to include the variety of technology platforms being investigated in this emerging field. Also, the name of the chapter was changed to Large Area Flexible Electronics to reflect this broader technical scope, and to be “agnostic” to materials (organic, inorganic, hybrid) and manufacturing processes (printing, vacuum, lithography).
Large area flexible electronics is expected to become a multi-billion dollar clean-tech industry that will revolutionize how society interacts with intelligent and responsive electronics-based systems. The unique properties of these systems – flexibility, ultra-thin profiles, light weight, potential for low cost, and high reliability – could have enormous impact on consumer electronics, aviation and space electronics, life sciences, lighting, power, military applications and telecommunications. They will enable a broad range of electronics components and products not possible today, such as: 1) smart medical bandages that sense the presence of infection, alert medical staff to changes in a patient’s vital signs, and deliver needed medication; 2) lightweight, foldable, rugged solar panels for terrestrial and space applications; 3) clothing with integrated distributed sensors and displays for physiological monitoring; 4) phased array antennae for distributed mobile communications networks; and 5) low-profile, ultra-efficient lighting solutions.
Industry analysts, as well as trade associations, have identified a broad range of possible applications and components (Table 1). The systems listed will be composed of organic, inorganic or hybrid materials, and fabricated via large area mass manufacturing processes.
Several factors are leading this emerging industry, including:
- Development of higher-performance solution-processable semiconductor and photoactive materials (e.g., organic small molecule, inorganic precursors, and inorganic nanoscale material suspensions).
- Commercialization of large area processing equipment.
- Qualification of large area flexible electronics manufacturing practices by adopting practices from the microelectronics, semiconductor and roll-to-roll processing industries such as printing.
- Advances in the design and layout tools necessary for product development.
- The roadmap working group identified three near-term opportunities for large area flexible electronics: displays, photovoltaics and sensors.
Displays
Displays are expected to offer the first commercialization opportunity for large area electronics, including point-of-purchase signage, electronic books, PDAs, GPS/electronic maps, automotive displays and instrumentation panels, smart cards and more.
The display module consists of an electronic control element (backplane) and an optoelectronic visual element (front-plane). Several companies are currently developing large area processes to fabricate active matrix backplanes consisting of arrays of organic semiconductor-based transistors, with hopes for commercial launch in early 2010. The first product offerings are expected to be mobile devices having large area flexible displays.
Long-term, the vision is to offer a large area flexible color display for viewing topographic maps, streaming news, magazines, books, etc., which, when not in use, can be folded and placed into one’s pocket. To achieve this long-term vision, several technology features must be developed or enhanced to ensure continued commercial launches of new products integrating large area flexible displays and greater product acceptance by the mass market, as discussed below.
Improved wireless functionality. The ability to offer pseudo-streaming will provide greater product flexibility and new products.
Availability of vivid color gamut. In the past, filters have been used to demonstrate color, but this approach reduces resolution by approximately one-third. An alternative approach is necessary.
Increased flexibility of large area displays. The majority of products that will be introduced in late 2009 rely on a multilayer display module structure that reduces the display flexibility.3 The introduction of a truly rollable/foldable display will provide designers freedom in developing products with novel form factors.
Improved refresh rate. Screens typically refresh at rates slightly faster than one frame per second. Several leaders in this field have begun to develop novel materials and driving schemes to provide video rates (24 to 30 frames per second) without flicker.4
Photovoltaics
Another near-term large area electronics opportunity is fabrication of photovoltaic modules. Several companies have well-staffed activities to develop processes (e.g., printing, roll coating) for the deposition of photoactive materials in an effort to establish a reel-to-reel manufacturing platform.
There are three major classes of photovoltaic materials (Table 2), and to differing degrees, they are all compatible with large area processing and manufacturing platforms. However, organic photoactive materials are currently the most promising for large area processing, as well as increased mechanical flexibility.
Two key technical attributes must be improved for broader commercialization of organic photovoltaic technology: efficiency/cost and stability/lifetime. Currently, the modest efficiencies and shorter lifetime of this technology confine it to the same cost-versus-performance curve as conventional first-generation crystalline silicon. A few of the efforts underway:
Efficiency/cost.
- Improved nanoscale morphology control of the organic photovoltaic bulk heterojunctions via self-assembly/self-organization.5
- Development of solution-processable and transparent conductive materials for flexible electrodes.
- More research into solution-processable hybrid organic-inorganic systems.
- Theoretical modeling of exciton physics at the bulk heterojunction interface.
- Development of polymers with optical harvesting properties (i.e., absorption of many photons by a dendritic polymer and fast intramolecular transfer of the excitons to a splitting site).
Stability/lifetime.
- Fundamental studies to determine the origins of aging-based performance loss in bulk heterojunctions.
- Development of conjugated polymers and electron acceptors with higher stability.
- Development of low-cost, roll-to-roll encapsulation processes.
Sensors
Sensors represent another family of active devices that provide a near-term opportunity for large area flexible electronics. In general, sensors under development today feature simple architectures for monitoring environmental and processing conditions. They include, for example, devices that change electrical or optical characteristics when subjected to an environmental change, such as temperature or humidity. Once observed, this information can be sent wirelessly (or via wireline) to control units for further action.
The potential for low-cost production, lower environmental impact during disposal, and distributed sensing ad-hoc systems make large area flexible electronics a practical solution for many sensor applications, such as the first-generation classes of sensors discussed below.
M-commerce sensors. These devices detect when an upper or lower limit has been reached. Examples include pressure and strain color-changing indicators/sensors. These types of sensors do not require an energy source and have limited reliability attributes.
Ambient assisted-living sensors. These sensors detect the instance when an event has been triggered, and have the ability to store that information in memory (e.g., temperature and humidity sensors). They require an energy source for operation (e.g., continuous calibration) and to store the triggered event in memory. This functionality provides for greater accuracy and reliability compared to M-commerce sensors. Also, the sensor design has wireless readout functionality based on industry standards.
E-health sensors. This class of sensors has the functionality to record the instance of a triggered event and sufficient memory to record a stream of historical data (e.g., medical sensors). These sensors provide high accuracy and repeatability functionality via an external power source. Most of these sensing solutions follow communications protocols that align with available near-field communication-enabled consumer electronics devices. Long-term, it is envisioned these types of sensor architectures will be embedded into objects to create a distributed ad-hoc sensor network.
The most-often mentioned needs for commercialization and wide adoption of large area flexible sensors are:
- Development of nano-material-based sensing devices with improved dynamic calibration to provide higher accuracy.
- Development of printed RAM for storing event history.
- Deployment of standardized readout technology and infrastructure.
- Development of disposable environmentally preferred power sources for active sensing.
Numerous technology and infrastructure concerns must be addressed to bring large area flexible electronics-based products to market. Functional inks, substrates, packaging and barriers, manufacturing platforms and processing equipment, testing and quality control, and standards are all areas where development is needed. The iNEMI Roadmap identifies the following gaps and showstoppers:
- Commercialization rate of materials, manufacturing/processing equipment and inline/offline characterization tools is slow and may not meet the cost/performance/utility demands to enable near-term product launches.
- Rate of development of large area flexible electronic products must accelerate; otherwise, an opportunity may be created for a disruptor to commercialize an attribute-competitive product.
- Lack of design and simulation tools will slow the diffusion and acceptance of large area electronics into the market.
- Lack of a well-developed supply chain and deep infrastructure will lead to products that are not cost-competitive.
Despite the challenges to be addressed, large area flexible electronic technologies will cause a paradigm shift in electronics. They will offer unlimited product design freedom, increased production throughput, and lower cost for higher utility applications offering voice, video, and data packaged in novel product form factors. Large area electronics is an emerging field attracting increased attention within electronics, as demonstrated by greater participation in the 2009 Roadmap. Nearly twice as many people participated in the development of the 2009 chapter as in 2007. Participation from industry, academia and government labs was more diverse and included strong representation from international experts in graphic arts and semiconductors. Large area flexible electronics is expected to be a growth segment over the next 10 years, offering attractive alternatives to “business as usual” electronics manufacturing. The 2009 iNEMI Roadmap is available at inemi.org/cms/roadmapping/roadmaporder.html.
References
1. Nanomarkets, Printable Electronics: Roadmaps, Markets and Opportunities, Sept. 19, 2006, nanomarkets.net/products/prod_detail.cfm?prod=6&id=212.
2. IDTechEx, Organic Electronics Forecasts, Players, Opportunities 2005-2025, 2005, idtechex.com/products/en/view.asp?productcategoryid=82.
3. EInk Corp., eink.com/technology/howitworks.html.
4. Yanko Design, yankodesign.com/2008/01/04/high-tech-napkins/.
5. J.K. J. van Duren, et al, Adv. Func. Mat., vol. 14, no. 425, 2004.
Daniel Gamota is president of Printovate Inc. (printovate.com), and chair of the iNEMI Large Area Flexible Electronics TWG (Technology Working Group) for the 2009 Roadmap; dan.gamota@printovate.com.
Out of favor for decades, the reflow process is undergoing a renaissance.
Vapor phase was the original and principal mass reflow method for a few years in the early 1980s. Somewhat displaced by IR and, eventually, convection dominant reflow, it has remained in limited use, primarily for prototypes, lower-volume production and thermally complex assemblies. In the past few years, vapor phase has experienced renewed interest primarily because of its advantages in achieving the small process window for peak temperatures with Pb-free assembly.
The first vapor phase reflow systems used the same configuration as a traditional vapor degreaser: a vertical, open-topped tank with a heater at the bottom and a cooling coil around the top opening. In the original single-batch versions of vapor phase, a large vat, so to speak, contained a fluid that had a specific boiling point – usually around 210°C, not coincidentally the full liquidus temperature of SnPb37. A heater boiled the fluid, generating vapor. Because the vapor was significantly heavier than air, it remained at the bottom of the tank. As more vapor was generated, a region of saturated vapor called a “blanket” expanded upward, pushing air out the top opening. The cooling coils defined the top of the blanket, preventing (or at least minimizing) vapor losses from the top opening. Assemblies were lowered into this vapor blanket and kept there until reflow was complete. Heat from the transition of vapor to liquid was transferred to all surfaces of PCB and components continuously until they reached saturation (i.e., boiling) temperature. These systems were expensive to operate because of continuous loss of vapor through the open top, and they offered no provision for recovery of fluid condensed on the products.
The first economical systems for commercial production incorporated a “secondary” vapor blanket of a different fluid, which floated on top of the primary vapor, minimizing vapor losses and providing a zone for vapor recovery. Originally Freon R-113 was used because it was low cost, has a low boiling point (48°C), has a vapor density between those of air and the primary vapor, and is inert. Later, more environmentally friendly secondary fluids were developed.
The single-vapor inline machine was developed for high-volume production. It used a single (primary) vapor in a closed top chamber with a continuous inclined conveyor (down on inlet – up on exit) entering and exiting through long chilled ducts. These ducts minimized vapor loss, and the exit duct provided vapor recovery.
‘A complete soldering process.’ The dual vapor and inline single vapor systems made the vapor phase reflow economical, but did not provide a complete soldering process. In dual vapor systems, the Freon secondary blanket was a solvent to the solder paste chemistry. Therefore, the paste on the assemblies had to be “dried” prior to vapor phase, usually in a batch oven. This provided an acceptable process, and also was used with single vapor inline systems. However, in both machines, high defect levels of tombstones, solder ball, and solder wicking indicated that preheating immediately before entering the vapor was required. This was part of the general increase in understanding of the total process of reflowing solder paste: evaporating solvents, activating flux chemistries and permitting time for the reduction of oxides on soldering surfaces, controlling heating rates for critical materials within the assembly, etc.
Inline IR preheat was incorporated into single and dual-vapor systems along with active cooling/vapor recovery, providing for the first time the complete reflow process in a single machine; this configuration remains available. However, at the same time, convection IR ovens demonstrated the capability for complete reflow with adequate uniformity at a much lower cost. Reflow capability was significantly advanced with the development of reliable convection dominant ovens, which are the principal technology in use today.
Vapor recovery started simply as dwell time within a chamber with chilled surfaces. Latent heat within the assembly evaporated (hopefully all) condensate from external, principally horizontal, surfaces. The vapors produced were recondensed on chilled surfaces and captured. More advanced systems added active air circulation (usually also providing cooling to the assembly) coupled with filters, condensers, stacked columns, and so on, to capture fluid more effectively.
As mentioned, vapor phase is experiencing something of a renaissance because of its advantages in achieving the small process window for peak temperatures with Pb-free assembly. The vapor phase process offers some distinct advantages over the convection dominant methodology that may make its use in Pb-free reflow more feasible.
Vapor phase offers absolute maximum control of temperature. As noted, the entire assembly contacts only vapor, which cannot be hotter than the boiling point of the fluid (e.g., 215°C for SnPb or 235°C for SAC). There is no possibility of overheating small components or critical materials. Heating is fast and relatively uniform. As an assembly enters the vapor zone, fluid condenses on all external PCB surfaces, and components, independent of size, shape, location IR or convection ovens, must use higher temperatures to achieve rapid heat transfer, which opens the door to potential overheating of small components.
While IR and convection dominant (forced convection) ovens aggressively have been attempting to reach lower levels of oxygen, vapor phase is inherently inert. (Indeed, another moniker attached to the process a few years ago was condensation inert soldering.) Fluids used in vapor phase reflow are completely chemically inert both in the liquid and vapor states. The vapors are significantly heavier than air and, therefore, exclude air/oxygen from the heating zone. In some instances, an inert reflow atmosphere may provide the advantage of increased wetting for Pb-free solders.
So with all these advantages, how did vapor phase succumb to IR convection and ultimately convection dominant reflow ovens? Why doesn’t everyone use it? In short, there are some severe limitations to the vapor phase process.
The greatest limitation of the vapor phase reflow process is the difficulty in controlling the heating rate (slope) throughout the process. If an unsoldered assembly at ambient temperature is immersed directly into saturated vapor, extremely fast heating occurs, which is uncontrollable, as long as there is adequate vapor present. Condensation proceeds, transferring heat to all surfaces until saturation temperature is reached. Condensation heat transfer has been measured to be as high as 10x that of convection, resulting in board/component heating rates as high as 50˚C/sec. Heating this quickly may not permit adequate time for fluxing actions, and can result in material damage and significantly increase defects such as tombstones, solder balls and solder wicking.
Extended time above liquidus (TAL) can be a drawback of vapor phase reflow. Because condensation heats so effectively, more heat may be stored in the assembly materials, especially thick PCBs and high-mass components. This can slow the cooling process, extending the total TAL. Some vapor phase machine cycles do not start active cooling until after all condensate is re-evaporated, which can further increase TAL.
Expensive fluids. Vapor phase is not inexpensive. Because of their special characteristics, the fluids are very expensive. Fluid for Pb-free reflow with a boiling temperature of 230˚C can cost well over $1,000 per gallon! It is vital to the economic viability of the process to avoid fluid losses, either as liquid or vapor. There is no inherent consumption of fluid during reflow, and the fluid does not degrade and require replacement if proper filtering is performed to remove contaminants, principally flux chemicals. Any vapor phase reflow machine should incorporate a fluid filtration system and be designed and fabricated to comprehensively prevent leaks.
In any vapor phase system, at the end of the reflow heating cycle, the entire product is completely covered with condensed liquid: all surfaces, under all components, inside all cavities! Drag-out of this condensed fluid on the completed assemblies exiting the system constitutes a continuous (potentially high) cost and must be prevented. Recovery of condensed fluid is accomplished by re-evaporation after reflow, in some systems as part of the cooling stage.
Previously, complete vapor phase reflow systems have been available that used IR preheat and convection-based vapor recovery/cooling. Newer systems feature vapor-based preheat, and advanced recovery sometimes uses vacuum systems.
Frankly, we find it surprising that most published information on currently available vapor phase reflow systems provides very little detail on the techniques and performance of the vapor recovery methods employed to prevent “drag-out” of condensed fluid on assemblies exiting the system after reflow. Given the fluid expense, this factor is vital to the economic operation of any system in actual production.
Today’s systems. Most systems available today offer some form of preheat before entering vapor phase reflow. As noted, both batch and inline systems are available with traditional IR preheaters. These configurations have been successful for decades, but IR heating’s limitations are well known, including a lack of speed and uniformity, especially on thermally complex assemblies.
Preheat using vapor is offered on several newer machines. The exact techniques appear to be unique to each manufacturer, some protected by patents. But they appear to fall into a few generic categories.
In dual vapor systems, assemblies dwell within the secondary vapor blanket for controllable intervals, providing some level of preheat before being lowered into the primary vapor zone for reflow. Some single vapor machines use the primary vapor to accomplish preheating through varying strategies of timed exposure to a zone of carefully controlled, non-saturated vapor. This provides a zone of lower temperature and energy density, permitting lower heating rates while maintaining uniformity across product surfaces. The result is the capacity to generate the complete heating profile shapes required for high-quality paste reflow, including low heating slopes, soak sections, etc., while maintaining good thermal uniformity, even across complex assemblies. Again, exact strategies differ among equipment manufacturers. Some use timed immersion into the non-saturated vapor zone, created directly above the saturated vapor zone (reflow section). Other systems enclose assemblies in a sealed chamber and inject measured amounts of primary vapor at programmable intervals to achieve the desired heating profile.
All these systems have produced test results demonstrating typical ramp-soak-spike profiles with low ΔTs. Satisfied users in actual production environments have validated this performance.1 However, long-term experience has shown that, in direct contrast to the saturated reflow zones, non-saturated vapor zones tend to be unstable and difficult to control. Therefore, as with any piece of production equipment, a potential user should validate the temperature profile on their own product, attaching their own thermocouples and confirming repeatability of the process.
Several vapor phase systems come with vacuum systems to enhance processing capabilities. The principal benefit claimed is the reduction or elimination of voids within the solder joint.2,3 By applying some level of vacuum to an assembly following reflow, while the solder joints are still molten, voids formed during the heating/melting stages can be drawn out and eliminated.3 Power electronics are said to benefit specifically because of the improved heat dissipation through solid joints with no voids.
The application of vacuum after reflow should also offer more capability and speed in recovering condensed primary fluid from product surfaces. Relying totally on the stored heat to completely evaporate all condensate can be problematic for component geometries that tend to trap volumes of liquid such as connectors, which may contain upward facing “cups” that trap deep pockets of liquid. Vacuum “drying” to remove trace amounts of liquid is a well-established technique used throughout process industries and should provide more reliable economies for complex assemblies.
Vapor phase will see some level of continuing growth in the next few years, based on its capacity to reflow the most complex Pb-free (or SnPb) assemblies with no chance of high temperature damage to components or materials. The extent of this growth will depend on the ability of the new breed of systems to compete on price-performance (cost of ownership-performance) with well-established families of convection dominant ovens, which already have been proven to provide acceptable reflow for a wide range of Pb-free products. Through it all, vapor phase will always provide the “ultimate reflow zone.”
References
1. Chris Munroe, “Beating the RoHS Heat,” Circuits Assembly, March 2008.
2. Fraunhofer IZM, izm.fraunhofer.de, January 2008.
3. John Bashe, A New Approach to Vapor Phase Reflow Soldering, SMT, November 2006.
Jim Hall is principal consultant and Six-Sigma Blackbelt with ITM Consulting, and a pioneer in vapor phase soldering. Phil Zarrow is president and a principal consultant at ITM Consulting (itmconsulting.org); phil_zarrow@itmconsulting.org.
