Previous Page  49 / 68 Next Page
Information
Show Menu
Previous Page 49 / 68 Next Page
Page Background

47

Chip Scale Review May • June • 2019

[ChipScaleReview.com]

enters the mold cavity. The second pressure

sensor is located at the venting side of the

cavity where the flow of the EMC stops at the

end of the filling process. With the sensors

located at these positions, the injection

pressure can be monitored, as well as when

the EMC flow completely fills the mold

cavity. Secondly, the pressure sensors can be

used to determine the pressure gradient over

the mold tool [4]

Using the set up descr ibed above,

several control strategies were applied

ove r t he i nject ion and compact ion

phase as mentioned in the introduction.

Figure 7

shows a schematic overview of how

the control strategy implementations interact

with the molding process. The left-hand side

represents the inputs for the molding process

consisting of the secondary clamp force, the

main clamp force, and the transfer speed at

which the EMC is injected. Coming from

the molding process are the sensor outputs,

displacement and pressure, which are fed into

the control strategies. Based on the sensor

outputs, these control strategies adjust the

input parameters. The strategies individually

control the inputs, which simplifies the

control strategies. However, as they merge

at the molding process, they are prone to

influencing each other. By defining proper

control parameters it can be ensured that a

stable and workable process is maintained.

Going from top to bottom in

Figure 7

,

the first control strategy that is applied is the

balance control (BC) of the mold tool. As

discussed before, the fluidic pressure in the

first phase of the injection causes a pressure

gradient in the mold tool. As a relative low

clamping force is applied in the beginning of

the molding process, this gradient can open

the mold tool at the injection side, which

needs to be prevented. By using the data from

the distance sensors, the tilt (Ɵ) in the x and

y directions is calculated and accordingly,

the force applied by the secondary clamp

cylinders is adapted. Using this strategy, the

tilt of the mold tool can be minimized while

the total clamp force is not increased.

The second strategy is called dynamic

clamping control (DCC) and controls

the vertical opening of the mold tool. By

once again using the input of the distance

sensors, the opening of the mold tool can be

monitored. This opening typically occurs

when pressures inside the mold tool are

already high. The DCC thereby acts on the

main clamping force. To prevent the DCC

from affecting the BC, or acting earlier than

intended, a threshold value for the mold

opening is included before the DCC increases

the clamp force.

The third strategy is the transfer pressure

limiter, which ensures that the fluidic pressure

stays within a defined window. Without this

limiter, the pressures can increase infinitely,

and there is the risk that the BC or DCC

act too aggressively, or even reach their

operational limit. For very low viscous EMCs

this pressure limit is typically set to 1MPa or

less, but for high viscous EMCs, the limit can

go up to 99% of the final packing pressure.

The transfer pressure limit is part of the

injection (transfer) profile, and is defined per

amount of transferred EMC.

The last strategy is the final transfer

pressure control (FTPC). This control acts

when the mold cavity is completely filled

and replaces the transfer pressure limiter.

At this point the molding process enters the

compaction phase. Using a predefined time in

the molding recipe, gradients are calculated

for the needed increase in clamping force and

packing pressure to reach the programmed

final values. When increasing the clamping

force however, the EMC pressure can also

increase by compression of the mold tool.

During the compaction phase the FTPC

measures the EMC pressure, and adjusts the

transfer motion to ensure the final packing

pressure is maintained. The latter also holds

that in the event of large compaction, the

transfer motion can become negative to

prevent packing pressures from getting too

high. After the final clamping force is reached,

the FTPC controls the packing pressure for

another defined period of time (FTPC final

adjustment) so complete stability is achieved

before the final curing stage starts.

Results

Combining the technology described

above, a demonstrator was made of the

exposed die wafer-level packaging process

via transfer molding. For the purpose of

demonstration, a simple D2W sample was

molded. The sample consisted of a 12’’

800µm-thick glass carrier, and blank Si

dies (

Figure 8

).

For bonding, a soft die attach film is

laminated to the glass carrier on which the

dies are then placed using a standard die

attach machine. In total, 284 dies are placed

in the samples, which are 8.2 x 10.7mm wide,

and 300µm thick. As it will be an exposed

molded product, the latter will also be the

thickness of the EMCmold cap.

As this is a demonstrator, no sample

preparations are done prior to molding.

Actual samples will, however, be prepared

using, for instance, pre-baking to remove

moisture and plasma cleaning to activate

surfaces at and underneath the die in the

case of E-MUF. The EMC chosen for

this demonstration is a low viscous type

of 1.5Pa.s, a filler cut of 20µm, and a gel

time of more than 45 seconds. The latter

properties typically allow a good molded

underfill, but also challenge the bleed

prevention on top of the die. The release foil

is a standard PET foil with a soft adhesive.

Standard transfer molding parameters are

used for the molding process (

Table 2

).

Table 2

also includes the control parameters

mentioned in the previous paragraph.

Figure 7:

Interaction diagram of the applied control strategies.

Figure 8:

Die setup of the molded demonstrator;

300µm thick dies are bonded on a 12” glass carrier

using DAF.