|This is a 2 x 7 boat die layout for fabricating a traditional flip-chip device. (Fig. 1)|
Flip-chip devices have been important components in electronic circuits for some time, with underfill technology helping to minimize thermal stress between flip chips and the substrates on which they are mounted. The process of adding underfill materials to flip chips and other miniature circuit devices has depended upon dispensers and dispensing techniques and, as flip-chip technology has evolved, so too have underfill dispensers.
|This diagram shows a 14 x 5 strip die layout for a new flip chip product. (Fig. 2)|
Automated fluid dispensing systems have improved with advances in flip-chip technology, supporting an expanding number of applications for mounting flip-chip devices on rigid substrates with increasing speed [units per hour (UPH)] and reliability.
Not many years ago, flip-chip devices were mainly restricted to printed-circuit boards (PCBs) in personal computers (PCs). But as flip-chip devices have evolved, designers have found places for them in many other applications, including radio-frequency (RF) components and circuits and in dynamic-random-access-memory (DRAM) devices. Compared to traditional, older flip-chip devices, these newer flip-chip devices are characterized by different configurations in terms of die size, die thickness, bump numbers, bump pitch, bump gap, substrate size, and other characteristics. Table 1 compares configurations for traditional and newer flip chip devices.
| Table 1: Comparing traditional and newer flip-chip device configurations. |
These differences in flip-chip configurations mean differences in terms of the amount of underfill fluid that will be deposited and how it will be dispensed, such as the number of passes required by the underfill dispenser. The underfill volume is affected not only because die sizes are different, but also because the bump gaps and die thicknesses are different. The bump pitch/bump gap shrink impacts the fluid flow speed: the tighter the gap, the slower the fluid flow speed. Also, when the space between adjacent die is narrower, the fluid must be deposited in a thinner line.
| Table 2: Comparing underfill amount, pass number, and die layouts for traditional and newer underfill methods.|
To achieve a desired/required volume of underfill fluid underneath a flip-chip die, the dispenser deposits a thin amount of fluid that flows under the die, and then the process is repeated, making multiple passes until the desired volume of underfill material is achieved. The narrower or tighter the gap, the less fluid can be applied at each pass, requiring that more passes must be made by the dispenser, resulting in a slower underfill process. The number of passes is also dependent on the height of the die. The volume of underfill deposited next to the die is determined by dividing the total volume of fluid needed by the number of passes.
Different Die Layouts
These newer flip-chip devices also require different die layouts for flip-chip production. A traditional method for placing die is to handle singulated flip-chip devices, where a die is bonded to a substrate. A boat carries about 10 flip-chip devices spaced a given distance apart, determined by substrate size, the number of devices, and the size of the boat. In newer applications, flip-chip die are carried by strips that contain a larger number of devices, typically tens of dies. Instead of the boat-spacing defining the placement, the substrate size determines the distance between the die on the strip. Table 2 compares for traditional and newer underfill methods.
A traditional 2 x 7 flip chip boat die layout consists of 2 even rows of 7 die spaced evenly apart. A strip die layout for flip chips might be 14 x 5. For I-path underfill dispensing, different dispense head movements are needed for traditional and newer flip-chip devices. An I-path underfill dispenser supplies undefill on only one side of a flip-chip die, with multiple passes needed for newer flip chips. For traditional flip chips, where the total volume of fluid needed to underfill a die is deposited in one pass, dispense head movement follows a straight line along the row of die.
For newer flip chips, with multiple passes, the dispense head speed is increased to about 10x the rate of the single-pass dispenser. The fluid flow rate is reduced at this faster head speed, however, to about 1/5 the traditional rate for the newer flip chips because of their thinner die thickness: thin die require thinner amounts of fluid to prevent underfill fluid from coming over the top of the flip-chip die. For balance, the dispense head speed is not 10x the traditional speed but perhaps a few times the traditional speed. The dispense head speed for newer flip-chip applications can be found from the simple formula:
[Original speed (S) x (8/10 mm die side length)]/[(50% down of underfill amount)(6 x pass #)(1/5 flow rate)] = S x (8/10) x 2 x 6 x(1/5) = S x 1.9
For a conventional I-path underfill dispense process, the dispense head ramps up, accelerating before reaching the first die, then moves at constant speed over the die while jetting underfill dots close to the die side. After depositing the underfill material, the head stops once, moves to the next die, stops near the next die, and accelerates again to reach the speed required for dispensing. This cycle continues until all flip-chip dice in the row have been underfilled. Such motion control is convenient because the head can leverage its fastest speed for nondispensing motions between dice, contributing to a UPH increase.
|Backtrack issues are related to the width of the device die and the distance between dice. (Fig.3)|
In this newer flip-chip underfill configuration, the distance between each die is relatively short versus its faster dispensing head speed. Once the dispense head is stopped, it requires longer acceleration and deceleration distances before and after each die to achieve the speed needed to underfill the die and decelerate before starting the process again. The acceleration and deceleration paths begin to overlap between dice. As a result, the dispense head must backtrack after deceleration to provide a long-enough path to accelerate to the required dispense speed without overshooting the dispense start point on the next die.
Backtrack problems can result when the dispensing speed increases. It occurs because of the relationship between dispensing head speed and the distance between dice.
For example, if the distance between dice is far enough, backtracking is not necessary, even when the dispensing head speed increases. But if the distance is not far enough, backtracking will occur even when the speed isn't increasing. Compared to processing traditional flip-chip devices, the configurations presented by newer flip chips are more likely to necessitate underfill backtracking.
|This plot depicts continuous path motion control.|
A software program was recently developed for controlling jetting underfill for newer flip-chip die and die configurations. Known as Continuous Path Motion Control software, it eliminates the need for backtracking, acceleration, and deceleration, and thus saves time, increasing UPH. Instead of stopping between die, backtracking, ramping to speed, dispensing, and de-accelerating, the software maintains the dispense head at a continuous speed and direction throughout the die underfill process. (Fig. 4)
The software saves significant time when spacing between dice is small, by enabling continuous motion at high dispense speeds. Under the control of the software, the dispense head is programmed to jet (dispense) the exact amount of fluid in precisely the correct place. As a result, UPH will increase for underfill dispensing. However, cycle time savings is not an exact reflection of UPH improvement, because the underfill process involves many steps. These include loading/unloading, detecting the fiducials, height sensing, and dispensing (or jetting), which includes both dispensing and nondispensing movements.
As an example, when samples were evaluated with 20 strips of 70 die per strip, comparing continuous-path motion control versus conventional underfill methods (with backtracking), the continuous-path approach delivered about a 27 percent improvement in UPH compared to the conventional approach. This was due to a 40 percent dispensing cycle time savings at 15mg/s flow rate and 95mm/s speed, although loading and unloading were not included in the analysis.
| These data represent cycle-time differences for 20 strips comparing motion control for conventional and continuous paths(Fig. 5).|
The continuous-path approach increases UPH even compared to underfill methods that don't include backtracking because stop-and-go movements between the die are eliminated. In fact, the continuous-path approach saves 23 percent dispensing cycle time at 9mg/s flow rate depending on actual die layout, dispensing conditions, flow rate, and other factors.
Newer applications involving underfilling of flip-chip die and other components that are smaller and closer together, especially where volume production and processing speed are required for manufacturing, are driving manufacturers to seek more efficient and effective methods to apply underfill. By using continuous-path motion control, increased speed can be achieved when applying underfill to thinner, more closely spaced die.
Contact: Nordson ASYMTEK, 2747 Loker Ave. West, Carlsbad, CA 92010 800-279-6835 or 760-431-1919 fax: 760-431-2678 E-mail: firstname.lastname@example.org Web: http://www.nordsonasymtek.com