|
If you looked inside any electronics system manufactured in recent decades, you would expect to find a green printed wiring board populated with components on one or both surfaces. But a new method of housing components has recently begun to emerge that places the components inside the board. Peeking inside a system made in this fashion would reveal only an apparently bare board made of a material that doesnot resemble FR4 material. It is not quite correct to say that the components are embedded in the board. Instead, the components are placed on some type of thin core, and the board is built up around the components. Techniques for placing components inside the board have been suggested many times in the past- even back in the 1960s - but two somewhat different methods have recently been introduced that appear to have advantages that will permit them to be used in place of conventional FR4 boards. One of the two methods is already being used in production on a modest scale. The other is at an earlier stage of development, but is intended to largely or totally replace FR4, surface mount, and probably through-hole as well. What advantages does the embedding of components give? First, no solder needs to be used in assembly; if there is no solder, there is no reflow. The numerous defects directly or indirectly related to solder and to reflow temperatures, will vanish. Second, performance is enhanced because interconnection distances are shorter. In most designs, there are no bond wires; instead, components are connected by microvias and traces. Components can also be placed on top of each other. Third, the finished board should be inherently rugged, able to withstand much more shock and vibration than conventional boards. A few prototype boards have already passed the informal “throw test” (throw the board at the floor), which has replaced the “drop test” that conventional boards are expected to pass. The Occam Process
This method has been described by Joe Fjelstad, the President of Verdant Electronics (www.verdantelectronics. com). Verdant is promoting the Occam Process and is providing guidance to several firms in the US and other countries as they begin building prototypes. Development of this method is still at a relatively early stage. Few prototypes have been built, and there are no test results as yet. A sticky substrate, which may be made of any suitable material, is used. The components of choice are burnedin and tested chip-scale packages, which are placed onto the substrate with their leads or lands facing downward. Once the components are in place, they are overmolded with a suitable build-up material (figure 1). Fjelstad prefers not to specify materials for the core or for build-up materials because multiple materials are available or will be developed in the near future. At this point the assembly is turned upside down - the leads or pads on the components are now pointing upward - and vias are laser-drilled down to the contact points on the components. Next the contact points are electrolessly plated with a seed coat of copper. Electroless plating is generally a lengthy process, but Fjelsltad has long experience with the process and estimates that the seed layer can be put down in about 10 seconds. The vias are then electroplated with copper. A resist layer is patterned on top, and the traces are formed by etching. If needed, additional build-up layers can be put on top. Chip in Polymer
A consortium of several European manufacturers, the Technical University of Berlin and Fraunhofer IZM began nearly 10 years ago to develop economically feasible assembly techniques for embedding components. The project was named “Hiding Dies” (www.hidingdies.net), and the method that the consortium developed is called Chip in Polymer. The development philosophy included using only currently available assembly equipment. Fraunhofer physicist Andreas Ostmann explains that Chip in Polymer is intended for fairly small assemblies, probably with not more than four integrated circuits with their associated passive components. The Chip in Polymer process begins with the wafers, which are thinned to 50 microns, and in some cases less than 50 microns (figure 2). The bond pads on the die are modified to make them suitable for PCB metallization, and the wafers are diced. The substrate for the die is a board core, which may be as thin as 150 microns. The die is adhesively bonded to the boardcore, with particular attention to the flatness of the die. A build-up layer is added next. The material for this layer is Resin-Coated Copper (RCC), in which the resin is 70 microns thick and the copper 5 microns thick. During processing, the resin flows around the 50 micron tall die, and the copper surface is perfectly flat. At this point the assembly is about 225 microns thick, or less than a quarter of a millimeter. Microvias are then laser-drilled down to the contact points on the now-buried components. The microvias are plated with copper, a patterned resist is put onto the copper layer, and etching produces the traces Comparing the two systems
Chip in Polymer is intended to be used only on small boards. This size limitation exists because the developers of the system think that larger boards would have unacceptably low yield. Testing of boards at Fraunhofer has given good results: 1000 hours of standard humidity testing without failures, and 6000 cycles of thermal cycling (-55°C to +125°C) without failures. Researchers at Fraunhofer have focused on making boards as thin as possible. In one project, they boosted the performance of a BGA by embedding it along with specially made ultra-thin ceramic chip capacitors. In recent years, ceramic chip capacitors have become smaller and thinner, and while the smallest ones are not yet down to the roughly 50 micron thickness that would be needed to be used in a Fraunhofer design, they could be used in somewhat thicker embedded-component designs. One German company is already successfully embedding components as thick as 1mm, which permits the use of many types of passive components as well as unthinned silicon die. Another German firm has a design that starts with a full-thickness flip chip that is mounted onto an interposer. This little assembly is then flipped over and placed on its substrate so that the interposer is now on top, and a prepreg is laid down as the build-up layer. Vias are then drilled down to the contact points on the interposer. Fraunhofer physicist Ostmann notes that the resulting board is relatively thick, but that it is assembled using conventional equipment. Fraunhofer, Bosch and other partners are currently using the Chip in Polymer approach to develop an automotive radar system that will operate at 77 GHz. The system will use 100 micron pitch SiGe chips to handle transmission and reception for the radar system. Chip in Polymer was selected in order to provide high performance, low cost, and ruggedness. Fraunhofer recently announced that a European consortium will shortly begin work on a project to move Chip in Polymer into volume production, with emphasis on refining supply line issues. The Occam Process is evolving somewhat differently because the firms working with the process have fewer guidelines concerning dimensions or materials. Work began only last year, and there are no test results or commercial applications yet. But Joe Fjelstad anticipates that wellengineered prototypes will be announced in mid-2008. A Brazilian company is using the Occam Process to develop demonstration boards for the Argentine Space Agency. An early board has already passed the “throw test.” The ruggedness and high performance of embedded boards is likely to appeal to military and aerospace users. Joe Fjelstad believes that one of the great strengths of the Occam Process is the elimination of both solder and reflow. That same is true, of course, of Chip In Polymer. But Fjelstad thinks that if tested and burned-in components are used, and if solder and its accompanying defects are removed, then defects that can cause electrical failures should be almost non-existent. He thinks it is possible that new types of failure mechanisms will emerge - cracks in copper, for example, if all-copper systems are designed. He believes that the Occam Process can be used to assemble large system with numerous components. He points out that individual boards can, if desired, be encased in metal, and that these “bricks” (as he likes to call them) can in turn be assembled into more complex and powerful systems. He suggests, for example, that it might be possible to use bricks to assemble a supercomputer having a volume of one cubic foot. Both Chip in Polymer and the Occam Process make rework either very difficult or impossible. Andreas Ostmann thinks that rework is essentially impossible. Joe Fjelstad, on the other hand, thinks that rework, while certainly not easy, could be achieved in various ways. It is far too early to predict to what extent or in what markets these two methods will be adopted. Both methods have very attractive features, but only time and continued evolution will reveal the degree to which they may replace current technology. Tom Adams, Contributing Writer Figure 1 Figure 2 Tom Adams can be reached at tom100adams@comcast.net. | 
