Mitigating Supply Chain Obsolescence: The Semiconductor Manufacturing Move Towards QFN and DFN Assemblies

In part one of our exploration of the manufacturing puzzle, we reviewed the history of how and why semiconductor assemblies have moved away from classic lead frame assembly. For a recap, these require expensive trim and form tooling, such as PDIP, PLCC, PQUAD, and PGA, and the market has moved toward substrate BGA (Ball Grid Array) and QFN (Quad Flat No-lead Package) or DFN (Dual Flat No-lead Package) type assemblies. Here, we direct our focus on QFN and DFN assemblies, which are on the lower end of complexity and cost but will have a high impact on the future of SOIC and low pin-count PLCC assemblies.

Why did the industry move toward QFN and DFN assemblies for lower pin counts and what impact will this have on component obsolescence?

Previously we explored why classic lead frame technologies are being eliminated and how the biggest cost in that assembly type is trim and form tooling. To review, the profit margins for higher volume classic lead frame assemblies were being pushed toward single digits by the year 2000, as this became a specialty item that few semiconductor manufacturers were supporting. One single size PLCC package can cost over $300,000 for just the trim and form tooling. However, these packages peaked in volume in the 1990s where the assembly cost was significantly less at higher volumes and included die attach, wire bond, mold, and trim and form.

Let’s dive into the details as to why QFN makes sense going forward. While QFN assemblies are lead frame based, they do not need trim and form tooling. The lead frame for a QFN is an X x’s Y matrix, like a chocolate bar with squares, where the X and Y dimensions of individual QFNs are flexible. The mold tooling dimensions, and outside lead frame dimensions are the same for many QFN final assembly dimensions. Common dimensions of the individual QFN assemblies are 3x3mm, 4x4mm, 5x5mm, and many other sizes. QFN and DFN lead frames are molded all at once and then sawn to form the individual QFN assemblies. Dimensions for DFN assemblies are more varied but always at the smallest of pin count as compared to QFN. The mold tooling can be the same for most dimensions of both QFNs and DFNs. 

After the saw operation, the assembly is mostly complete. The result is a molded QFN/DFN assembly needing no expensive trim and form operation, needing only one mold tool for many different QFN/DFN sizes. The throughput is much faster without trim and form operations as well. Without a trim and form operation, there is a higher yield of a QFN/DFN assembly than the equivalent pin count package needing trim and form. Less physical space, faster throughput, and higher yields spell the ultimate obsolescence of classic lead frame assemblies that require trim and form operations.

QFN/DFN will cause the demise of equivalent pin count lead frame assemblies that need trim and form operation. This has already happened with classic DIP packages. While not as expensive for trim and form operations, DIPs have been around for more than 50 years, and through-hole assembly technology is no longer driving volume. One could argue that DIPs have already been replaced by SOIC packages, however, not so for the support timelines required by long-lifecycle systems.

SOIC packages will ultimately be replaced by QFN and DFN type assemblies. We have already seen both shortages in SOIC assemblies as well as outright obsolescence of the SOIC version of a product while keeping the QFN version active. If you examine any common logic parts offered today, you will see that they typically sold in both QFN and SOIC versions side by side. These do require different board layouts, given that QFN assemblies have been offered at the classic square QFN dimensions mentioned earlier. At Rochester, we believe that offering the flexibility to keep existing SOIC board layouts for signal, while providing a QFN assembly, is the best path forward to support long lifecycle systems.

When higher reliability solder joints are needed with the circuit board, “wettable flank” technology is employed. A typical QFN only has solder on the bottom of the leads, leaving exposed copper on the sides of the package. This makes inspection of the solder joint to the package difficult. “Wettable flank” allows the QFN/DFN manufacturer to plate solder up on the side of the exposed lead frame. This both covers some or all the exposed copper while also giving more surface area for post assembly inspection of the solder joint. The wettable flank does typically cause more assembly processing and an increased manufacturing cost.

Today, Rochester Electronics offers customers QFN footprints compatible with existing SOIC or small pin count PLCC footprints. This can be accomplished with a simple board modification for the paddle area under the QFN. The modification to the board is required for SOIC equivalency in shock and vibration environments. Unless the QFN paddle is soldered down, trim and formed lead frame assemblies will perform better in these environments as compared to a QFN.

Rochester Electronics anticipated these market trends and has invested in QFN/DFN assemblies. We offer the classic square QFN assemblies in volume production today, but also the flexibility that long-term system companies desire when migrating from other assemblies, by additionally offering non-square QFN assemblies with footprint compatibility and minor board changes. Rochester is solving another part of the obsolescence puzzle for our global customers.