Breakthrough in Conformal Self-Shielding of Electronic Packages
By Scott Morris and Milind Shah, RFMD®

Like it or not, electromagnetic shielding is a common need in radio frequency/microwave applications and, in an increasing number of cases, a requirement from the customer. Commonly, it is an afterthought in the design of electronic components. This way of thinking can potentially pose a large consequence to the design in terms of both form factor and performance. For example, when a shield is put in place at the customer, the originally intended performance can be significantly altered due to the electromagnetic coupling between the design’s components and the shield. Multiple design iterations with the customer are often needed to arrive at the final application solution. Therefore, integrating the shield in the product, prior to the customer application, is extremely valuable to both the designer and customer. This yields a more robust product. The development and implementation of this type of self-shielding technique, package-level plating, has been effectively proven. The process will be further discussed in this article.

The purpose of any electromagnetic shield is to eliminate (reduce) the coupling of electromagnetic fields between two or more locations. In most applications, there are three different effects that need to be addressed. First, there is the prevention of outside interference affecting the devices’ performance. Second, there is the minimization of signals emanating from one component interfering with other components present within the design. An example of this is a multi-chip module. Lastly, it is used to prevent cross-talk of components within the customer’s application, namely a printed circuit board (PCB). One of the best ways to implement electromagnetic shielding is at the component level by the use a conformal self-shield. There are numerous methods by which the shields can be applied, but care must be taken in the grounding scheme implemented.

In the case of a large manufacturer, one needs to pay attention to the high-volume manufacturing (HVM) of all self-shielding applications. The processes required for adding a shield to a product must fit into current production procedures. If additional process steps are necessary for full implementation, they must be kept to a minimum. This is done primarily to keep the final product cost as low as possible. In the manufacturing of modules for RFMD®, laminate strips are typically designed with a ground ring located on the periphery of the module outline. During assembly, the die attach, wire bond and molding processes take place first in what is commonly known as the “front of the line.” After completion of these processes, the shield can be integrated with the assembled module. Since the module’s ground connection has already been defined under the mold compound, all that is needed to connect this to the shield is to expose the ground prior to the application of the conformal coat. This shield ground can be exposed to air using standard dicing equipment or other removal mechanisms. This process allows for a direct ground connection of the shield to the package I/Os.

The conformal conductive materials can be applied using any number of materials and/or methods, including plated, painted, discrete cans or sputtered coatings. As is the case for most things, each of these technologies has its pros and cons. The ease of implementation, cost and performance are the main drivers for the down selection of the materials to be used for production. A critical aspect of the down selection is the development of a shield process integrated at strip level (batch) to lower the cost of shielding.

In order to select the best material to use for shielding, a baseline vehicle was identified. This chosen passive vehicle allowed for a wide range of measurements in a controlled fashion. The measurements did not rely on the narrow radiation of a single part, but on an average of multiple parts. In this manner, statistically valid results can be gathered with a single test. Since a passive vehicle was used, great attention to detail was required in the test setup. Care was taken to ensure that no spurious noise was present in the test setup. The evaluation PCB was shielded so there was copper over the entire surface. This copper was then connected to ground. All connections to the board were also required to be shielded. This unintended radiation would skew the results and render them ineffective.

The test setup itself was rather simple. All that was needed was a signal generator, a GTEM test chamber and an EMC analyzer. The signal generator was used to create a known signal. The EMC analyzer read the signal so the amount of radiation could be detected. The purpose of the GTEM test chamber was to keep out errant signals from other sources. An off-the-shelf software system was used to run the measurement system.

Using this test setup, measurements were made on a passive test vehicle comparing plating, metal impregnated paints, sputtering and discrete cans. The baseline device measured was chosen to be an unshielded part. The shielded parts were then measured and compared against this reference point. The difference was defined as the shield attenuation. The tests were made in a frequency span of 400 MHz to 12 GHz in 200 MHz increments. The results are presented in Figure 1. As one will see from the data, on average the plated shield yielded the best overall performance.

Another critical parameter requiring determination is the package reliability. As a device’s footprint shrinks, the need for robust packaging solutions is paramount. To meet RFMD requirements, the integrated shield must withstand various Joint Electron Device Engineering Council (JEDEC) standards package reliability tests. These tests are indicated in Chart 1. Incidentally, these reliability tests can also be used to help fine-tune the production process. Multiple experiments were performed to find the “sweet spot” for each material process, thus defining the limit for the given process. For high-volume manufacturing, these process limits must have been known. The only material that was able to pass all manufacturing and reliability tests was plating. The use of cans was omitted from the study due to the overall cost restrictions and an undesirable increase in the size of the package footprint.

The competing technologies in the marketplace have a larger footprint, an increase in cost, an addition of headroom and decrease in performance. The biggest competition to the integrated shield is the discrete shield. Since most customers have experience with them, they are well established as the industry standard. The discrete shield’s main drawbacks are the need to maintain additional inventory, the additional cost of application and the increase in the final size of the board area and height.

There are many positive aspects to the self-shield, but the strongest driver is the ability to fix the final design of the part with the shield already in place. The modules are then built to specification at the original design house, and there is no need for iterative tuning requiring customer involvement. This differs from most customer applications in that the typical design involves application of the shield post-design, and therefore needs multiple PCB (customer/design house) design spins to arrive at a final solution. This adds critical time to the ever-shrinking product cycles. The customer is also able to reduce their final solution size. By eliminating the discrete shield, the customer realizes a 20 percent to 30 percent savings in PCB real estate. There is also a potential height savings, since the self-shield adds only roughly 10 microns to the module height. This is compared to the approximately 2mm stand-off required from the highest component for discrete shields. The final advantage is the simplification of rework of the module. Since there is no shield to remove, rework becomes much easier.

RFMD®’s development of conformal self-shielding is one of the most revolutionary module technologies to be introduced in the last 10 years. It offers a true competitive advantage for customers wanting to reduce their final solution size. It also has a direct benefit to the bill of material costs since fewer components are needed per application. This technology is currently being ramped for product release in Q1 of ‘08.

Acknowledgements
The authors would like to thank Leonard Reynolds, Don Leahy, Al Hatcher, Kenney Edney, Mark Held, Steve Parker, Vic Steel, Kermit Law, Brian Calhoun, Waite Warren and others who were instrumental in helping to develop and test the various processes examined.

We would also like to thank Dave Halchin for his time and effort in preparing this article.

About the Authors
Scott Morris is a graduate of NC State University and UNC Charlotte. He has been working for RFMD for the last 10 years. Currently, he is in the corporate R&D group as a staff packaging engineer. Milind Shah is a graduate of Florida International University and Bangalore University. He has been working for RFMD for the last 9 years. Currently, he is in the corporate R&D group as a packaging engineer manager.

RFMD®
www.rfmd.com
TXTLINX.COM150
Email this article to a friend!