Previous Page  48 / 68 Next Page
Show Menu
Previous Page 48 / 68 Next Page
Page Background


Chip Scale Review May • June • 2019


warpage. With this very simple calculation

method it is easy to get a qualitative

indication of the warpage in advance of

complicated simulations, which makes

the selection of promising EMC types and

package dimensions much faster.

Dynamic clamping

The following sections discuss clamping

dynamics, and the setup and control strategy.

Clamping dynamics.

Transfer molding

starts by closing the mold tool before the

molten EMC is applied to the device. This

makes it possible to clamp gently on the die

and protect it from the EMC, thereby keeping

the top die surface exposed (see

Figure 1


Clamp force at this stage is still relatively

low to protect the (carrier) wafer, dies and

interconnect from damage. Next, the molten

EMC is injected into the mold tool and the

clamping force has to be increased gradually

together with the rise of the EMC pressure, to

prevent the mold tool from opening. Finally,

the complete wafer is covered with the

EMC, and in the compaction phase, the final

transfer pressure is increased to eliminate

potential voids and ensure full adhesion.

Forces can now rise up to 1000KN for a 12”

FOWLP package.

To visualize the balance that needs

to be maintained during this sequence,

Figure 5

displays a simplified force

model of the process. In this model,

the clamping and fluidic pressure

are modeled as a force F1 and F2,

respectively. Z1, Ɵ1, Z2, Z3 represent

the movement and tilt of respectively,

the mold tool, the die, and the

interconnect. K1 is the stiffness of

the foil, K2 the stiffness of the die,

and K3 the stiffness of the die attach

film (DAF) or interconnect in case

of E-MUF. A1, A2 and A3 represent

the contact areas, respectively,

between the mold tool to wafer, the

foil to the die, and that of the die to the

interconnect. The clamp pressures are

represented by P1-3, where P1 and P2

are the sealing pressures on the wafer

edge and die, and P3 is the pressure in

the interconnect. The challenge is to

keep the forces and surface pressures

in balance. In other words, for a

given fluidic pressure F2, F1 should

be adjusted such that there is no

movement in Z1, Ɵ1, Z2 or Z3.

As the clampi ng area A1 is

typically less than 2% of the full area

of the mold, the clamping pressure

P1 to seal the mold cavity to the

wafer is relatively low. A larger contribution

comes from the surface of the dies where

the clamping pressure P2 should be above a

certain minimum to keep the die free from

EMC. At the same instance, the clamping

pressure P3 has a maximum, which depends

very much on the design of the product. For

exposed D2W products, the die and DAF can

generally accommodate a range of force. For

more advanced products such as E-MUF,

or when a soft DAF is used, pressure on the

die must be maintained relatively constant in

order to prevent damage to the interconnect,

or sinking of the die into the DAF. Hence,

there is a more narrow balance window in

which the forces need to be arranged. The

balance between the clamping pressures P1

and P2 is defined mechanically by the mold

tool by using differences in height, which can

be adjusted if needed.

Because flow length is long (300mm)

and relative to that, the mold cap is thin

(down to 150µm), a large pressure gradient

is formed over the wafer going from a high

pressure at the EMC injection point to a zero

pressure at the EMC flow front. In the model,

this pressure is annotated with F2, and can

be substantial depending on the injection

speed and flow properties of the EMC. This

gradient changes the force balance and to

maintain sufficient pressure on the die this

unbalance needs to be controlled. In addition,

the position of F2 starts at the injection point

and moves during the transfer towards the

middle of the wafer. Therefore, F1 needs to

be adapted constantly in magnitude, as well

as in position, to prevent planar opening or tilt

of the mold tool. The next section will discuss

how this is achieved.

Setup and control strategy.

To control

the clamping balance during the injection

state, a setup is created as displayed in

Figure 6

. Visible are a bottom mold tool

onto which a 12” wafer can be centered.

Injection of the molten EMC occurs

from the left-hand side via a separate

construction that is attached on top

of the wafer edge, and that is part of

the clamping edge of the mold tool.

For clamping, the bottom mold tool

is moved upward against a stationary

top mold. The clamping forces thereby

come from below created by a main

clamp cylinder acting on the center of

the mold tool. Two secondary clamp

cylinders are placed at the injection

side (left side) of the mold tool. These

secondary cylinders can deliver up to

15% of the full clamp force, and are

only intended to steer the mold tool

during the injection phase.

To monitor the motion of the

mold tool, four Z-position sensors

are attached to the bottom mold. By

registering the displacement of all four

corners of the mold tool, the motion

vector n0 can be calculated and used

for control purposes. [3]. To monitor

the fluidic pressure in the mold tool,

two pressure sensors are included.

They are placed in the top mold and

are shielded from the EMC flow by a

release foil. The first pressure sensor

measures the EMC pressure at the

injection point (left side) just before it

Table 1:

EMC D gives the lowest calculated value

of the curvature, which is in line with the warpage

measurements displayed in the last row.

Figure 5:

An illustration of the force model of the transfer molding

process, including the stiffness, that needed to be considered.

Figure 6:

Mechanical setup created to control the mold balance during

transfer molding.