Epoxy is widely been used as a material to hold the die securely onto the substrate, lead frame which act as means to dissipate the heat energy generate by the chip operations. After Die attach, we need to cure the epoxy under certain temperature and time.
There is a need to control the bond line as there is significant performance contributed by the BLT (Bond Line Thickness)
Die attach media
Epoxy, Silver epoxy, non-conductive adhesive, soft solder, DAF (die attach film),
Different material has different rate of expansion on different temperature.
Copper has a much faster expansion than silicon. And matching of the die and substrate is necessary but not possible to be perfect unless same material is being used. As LED is afraid of heat, 5 degrees lower the LED junction teamperature will able to prolong 5,000 hours of the LED life time, hence Semileds has made copper layer on bottom of the high power LEDs at wafer fab. Stage which can be used on copper substrates for perfect conduction of heat.
Whereas for IC DA process consideration, an optimum of the die material and substrate by means of different type of DA media and the thickness of it is needed.
Effect of the Bond line thickness
Too thick the epoxy will reduce the effect heat conduction from the chip to the substrate. Too thin of the epoxy layer will reduce the cushion effect, in other words, the die and substrates will expand in different rate that the epoxy layer could not absorb the deformation or warp page. Or, under high temperate there is a high chance for the chip to be fully detached and or partly detached from the substrates, thus create hot spots thus the heat dissipation rate is reduced, the operation temperature of the chip will increase and hence shorten life of the chip or even cause premature failure.
Optimum thickness of epoxy is needed for optimizing the bond strength, heat dissipation and to ensure long life time of the package.
More powerful chip is made than before, an integrated circuit of 1mmx1mm contained 1,000 transistors in the 70s' same size of chip nowadays contain 5 millions. However the design of the package material has not much different, nor improvement at all.
Automotive are the toughest test site for an integrated circuit. The temperature of Sahara is 80 degree C in day and -20 at night, the car and their IC package will experience thousand s of temperature cycles every year The requirement for a bond line control as one of the key requirement. Twenty years ago the bond line for general purpose IC package of small power is considered loose. Partly because the available die attach machine was not able to provide such kind of accuracy. Recently, tendency of controlling the BLT is tighter and tighter from CpK 1.33 then 1,67 and even 2.0.
Now there is almost a standard demand for a tight control on the epoxy thickness. Typically SOIC, PBGA or TSSOP, the equipment qualification list has a BLT control of min. 0.4 mil and not more than 0.8 mil with capability index of 2.0 or more. A standard deviation of more than 6 micron will fail to meet this requirement.
Epoxy has an imminent effect on the stability of the BLT, as if Epoxy is expired, not properly thawed, or very short pot life, that will contribution to inconsistency of epoxy dispensed, which in turn will after the volume of the epoxy. Machine wise, a highly consistent epoxy dispensing amount with continuous compensation over the hours is needed.
Some epoxy is very viscous and the die can be used up after the die is left on the epoxy, on the other hand for low viscosity epoxy, it is hard to make a very high BLT as it tends to smear after die is placed on the substrate
Geometry of the Epoxy dispensing nozzle
The factors affecting the performance of a dispensing needle has been investigated. It is found that the geometry of the needle (especially the length) has substantial impact on the characteristic of needles performance. In general, longer needles have a better capability to resist the dripping effect when handling a less viscous epoxy (say 3,500cps). However, a longer needle needs higher driving-pressure in order to dispense the same amount of epoxy when compared to a short needle. Besides, a short needle usually can give a faster response when compared to a long needle. From the past analysis, it seems that the tailing effect is not related to the length of the needle, but actually it is related to the wetting-surface-area on the needle and contact angle of epoxy.
Challenge on Material Variation
Current die bonder does not able to tell the variation of substrates thickness and die thickness; to achieve the BLT of CpK 2.0 is more or less like we throw a dart on the red target with folded eyes.
Challenge on Set-up
The Pick level, ejector level, bond level has interactive effect on the BLT, die tilting and placement (Die Rotation), hence it is a sweating job to achieve certain BLT process window, plus measurement o the bonded units for confirmation, it is really tedious to pass any production requirement on BLT at CpK 2.0
What can we do?
To ensure the high repeatability of machine in maintaining high capability index on epoxy... there is a need to have some means to detect the substrate thickness and at same time die thickness.
In order to have good control over the BLT, the substrates thickness and die thickness has to be known in advance before the attach.
Bond head accuracy
Linear Z travel is needed, soft landing bond Z is an added advantage.
However the bonding Z motion should be precise at micron level, or least there is a high repeatability z motion.
Likely the machine UPH should have been affected in performing the accurate BLT.
All new Die attach equipment should have this precise Control on BLT.