|
The QFN: Smaller, Faster and Less Expensive
|
By David Comley and Paul Smith, Carsem Inc., Ipoh, Malaysia and Scotts Valley, Calif.
|
This leadless package is gaining wide industry acceptance and represents a significant technology breakthrough. This article discusses the two principal methods of producing the plastic QFN and examines the pros and cons of each process.
|
|
|
This is the singulation station of an in-line CM-MS system. The system measures 19 meters (62 feet) in length.
|
For the past twenty years the objective of the semiconductor assembly industry has been to make plastic packages smaller, thinner, cooler, cheaper and at a faster production rate.
The PDIP (plastic dual inline package), once the dominant package, represents less than 10 percent of the packages used today.1 It has been replaced primarily by the gull-wing SO (small outline) format, which represents about 60 percent of total worldwide volume.
We are seeing a phenomenal growth of a new leadless plastic package family, classified by JEDEC2 as the QFN (quad flat non-leaded), shown in Figure 1. Many people believe this package family will eventually replace the SO format and some of the QFPs (quad flat packages).
|
|
Figure 1. Top and bottom view of a typical QFN package compared to similar QFP and SOIC packages
|
Specifications
Plastic QFN body sizes generally range from 1x2 mm with 3 leads, to 10x10 mm with 68 leads.3 There are currently two package outline versions registered with the JEDEC JC-11 Committee: the MO220 for packages with leads on all four sides, and the MO229, with leads on only two sides.
Compared to the standard leaded package with similar body sizes and lead count, the QFN is significantly smaller and thinner. (Figure 2 compares some typical packages.) For example, when mounted to a PC board, a 48-lead 7x7 mm QFN consumes 40 percent less area than a comparable TQFP (thin quad flat package). The QFN is typically 0.80 to 1.0 mm thick compared to the TQFP, which is 1.0 to 1.2 mm thick.
Removal of Heat Essential
New generations of ICs are adding more functions; they operate at higher speeds and often produce higher device temperatures.
Naturally, the efficient and cost-effective removal of heat is essential. Compared to leaded packages with similar body sizes and lead counts, the QFN offers far superior thermal performance because the leadframe is on the bottom of the package.
|
|
Figure 2. This chart compares QFN body dimensions, maximum die size, PC board area and thermal and electrical data with a similar gullwing package.
|
As a result, the die attach pad is exposed, allowing it to be easily soldered directly to the board.4 In general, the q Ja for the QFN is about half of its leaded counterpart. (Figure 2 compares some typical q Ja values, measured per JESD-51).
A device using a 48-lead, 7x7 mm QFN package, operating at an ambient temperature of 70°C, can dissipate about 2.5W. The equivalent TQFP package is limited to about 1.5W.
At higher operating speeds, the combined inductance of the gold wire and long lead-frame traces will affect device speed performance. The QFN provides excellent electrical characteristics because it eliminates gullwing leads and shortens wire lengths. In general the self-inductance for the QFN is approximately half its leaded counterpart.
Clearly, reducing cost is the biggest challenge. This factor will determine whether the QFN becomes a package for special applications or whether it replaces some of the gullwing packages.
From an end-product standpoint, cheaper also includes the QFN's ability to pack more circuit functions into a smaller area of the PC board.
|
|
Figure 3. Molded strip processed with the CM-MS method (top) and one processed with the AM-SS method
|
Major Assembly Process Methods
There are two major assembly process methods used to manufacture the QFN package. One is an array-molded, saw-singulated (AM-SS) process. The alternate method employs a cavity-molded, mechanically-singulated (CM-MS) process. Figure 3 illustrates both AM-SS and CM-MS leadframe strips post-mold.
Determining what "cheaper" means from an assembly standpoint, requires evaluating the pros and cons of each process.
Both processes use an etched leadframe concept. However, as shown in Figure 3, the design of the leadframe varies, depending on which process is selected. Figure 4 shows the process flows.
The AM-SS Process
The first process step that differs from the assembly of standard plastic packages is the attachment of a polyimide tape to the backside of the leadframe strip.
|
|
Figure 4. Process flows for AM-SS and CM-MS methods.
|
The polyimide tape prevents resins in the mold compound from bleeding onto the back of the package. The tape can either be attached by the leadframe supplier or by the assembler, prior to die attach or after wire bond. Since the tape's adhesive will deteriorate over time, care must be taken in determining when to attach the tape.
Inspection
If there is an inspection after wirebond (3rd optical), then consideration must be given to how rejects are marked and identified at the final pick-and-place process.
The older method of removing a lead using a punch tool is unsuitable, since this method will cause severe peripheral damage. Unit mapping techniques must be implemented, with the ability to transfer data to the laser mark operation. This will allow the auto inspection cameras used at pick-and-place to find the rejects and leave them behind.
Molding
The AM-SS version is generally produced by overmolding three to four sections of a leadframe strip.
Due to technical limitations related to transfer molding, the AM-SS process is typically restricted to overmolding an area of about 50x50 mm. Considering that a single mold shot could involve up to 8,000 units, any power failure would be costly; therefore, a backup power supply is worth considering.
Additionally, each mold shot will almost certainly involve molding different products and lots simultaneously. Each leadframe strip should be bar-coded to assure lot integrity.
|