Patent Application
20080272503
Disclosed herein is a semiconductor device and a method for making a semiconductor device. The method relates generally to manufacturing electronic circuit assemblies, and in particular to the molding process of a molded matrix array package.
In the manufacture of semiconductor devices, dice are arranged on a substrate in a matrix array. Subsequently, the dice in the matrix array may be encapsulated in a curable mold compound. After the molding process, the individual dice may be singulated into packages. The molding process may involve placing a mold over the array of dice, filling the space between the mold, the dice and the substrate which supports the dice with the mold compound and after the mold compound has cured, removing the mold. A mold compound usually used in the process comprises a resin material with filler particles evenly distributed therein.
During the molding process, migration of the filler particles may occur. The mold compound enters the mold through an inlet and is forced to an outlet of the mold which primarily serves to enable discharge of the air which is displaced by the inflowing mold compound. Thus, the mold and the substrate with the dice arranged on its surface form walls of a flow channel through which the mold compound flows during the molding process. Due to the friction between the mold compound and these walls, the velocity of the mold compound shows an increasing gradient towards the walls, i.e., the mold compound moves significantly slower in close proximity to the walls than it does in the middle between two walls. This decline of velocity, observed in the cross-sectional direction of the mold flow from center to edge, i.e., towards a wall of the flow channel, leads to an increased probability of collision of filler particles with each other which in turn leads to migration of filler particles away from the walls of the flow channel and towards the forefront of the innermost streak of the flowing mold compound. Obviously, this effect is amplified by the fact that the flow channel is wider between the dice than it is above the dice as the mold compound can flow significantly faster in those wider areas.
The result of this is that the originally evenly distributed filler particles concentrate in certain areas of the stream, thereby creating other areas of significantly lower concentration of filler particles in the resin material. After the mold compound has cured, the mold is removed from the substrate. The cured mold compound now constitutes a housing for the semiconductor device.
In a subsequent step, performed either before or after the individual dice are separated into semiconductor devices, each of the devices is marked, usually by employment of a laser beam which is directed over the surface of the housing so as to write the manufacturer's name, the device type, etc., onto that surface by changing certain properties of the material due to the heat transferred.
However, because the laser light absorption is different for the resin material and the filler particles of the mold compound, the characters written in surface areas of high filler particle concentration will suffer from low contrast which results in poor legibility. This is a serious quality issue and may lead to a significant yield loss.
A transfer mold process for encapsulation of a matrix array package of dice on a substrate is proposed wherein the flow of the mold compound between dice is obstructed. In other words, the flow velocity of the mold compound between dice is constrained with the goal of approximating it to the flow velocity above the dice. It is to be understood that every limitation of the flow velocity between the dice, even if it does not result in equal or uniform velocity throughout the cross-sectional area, will bring about a positive effect in terms of reducing the clustering of filler particles in certain areas of the mold compound. One way to achieve a slower mold flow in those areas is to reduce the cross-sectional area of the flow channel between dice, i.e., to reduce the clearance of the mold over the substrate. The result of this is that the mold compound cannot flow substantially faster between the dice than above the dice. Thereby, the segregation of filler particles in the mold compound is reduced.
In one embodiment, a flow blocker is disposed between dice on the substrate. The flow blocker may, for instance, consist of substrate material. Alternatively, the flow blocker may consist of a printable material, as for instance a polymer resin, die attach, etc. Both alternatives involve operations being performed on the matrix array package before the mold is placed over the substrate.
For instance, flow blockers built from substrate material may be obtained by grinding or etching the surface of the substrate so as to create grooves or recesses inside which a die or dice can be mounted. The substrate material surrounding the grooves or recesses then constitutes a flow blocker. The depth of the grooves or recesses may be selected so as to be substantially equal to the height of a die or, as the case may be, the combined height of a die and a layer of die attach material underneath the die, the combined height of the die and solder bumps underneath the die, and so on. The width of the grooves or recesses may be selected so as to leave a passage for the mold compound on either side of the die mounted therein whose width is substantially equal to the height of the passage above the die.
Flow blockers made of a printable material may be printed on the substrate's surface prior to installing the dice. The thickness of the flow blockers printed on the substrate may be selected so as to be substantially equal to the height of a die, the combined height of the die and a layer of die attach material or solder bumps underneath the die, and so on. The width of the printed flow blockers may be selected so as to leave a passage for the mold compound on either side of the die mounted therebetween whose width is substantially equal to the height of the passage above the die.
Both alternatives of this embodiment provide for the use of a mold which has a substantially flat inner surface. It is, of course, also possible to make the flow blockers so as to fill the space between dice entirely, without departing from the basic idea of this disclosure. For instance, a squeegee or wiper may be used to fill the space between dies with a curable compound prior to placing a mold with a flat inner surface over the matrix array package.
In another embodiment, the height of the package is reduced in the areas between dice. This may, for instance, be achieved by using a mold whose inner surface is rippled, i.e., the surface has ridges which are arranged between dice when the mold is placed over the substrate. The height of the ridges may be selected so as to be substantially equal to the height of the die, the combined height of the die and a layer of die attach material or solder bumps underneath the die, and so on. The width of the ridges may be selected so as to leave a passage for the mold compound on either side of the die mounted therebetween on the substrate whose width is substantially equal to the height of the passage above the die. Each of these ridges narrows the space between two dice in a similar manner as the flow blockers described above. There is, however, a difference in that the ridges protrude downwards from the mold while the flow blockers protrude upwards from the substrate. Either way, the width of the diverse segments of the flow channel (those above the die and those beside the die) is approximated or even equalized. After singulation, the semiconductor devices have flattened edges, i.e., margin areas of their housings have a relatively small height, compared to the rest of the housing. In other words, the thickness of the housing is reduced in at least one margin area of the semiconductor device.
For a more complete understanding of the present invention, and the advantages thereof, reference is now made to the following descriptions taken in conjunction with the accompanying drawings, in which:
In
As can be seen from
A first embodiment of the proposed method and the semiconductor device thus produced is illustrated in
The width of the recess 7 is selected so as to be equal to the sum of the width of the die 1 installed therein plus twice the distance of the mold tool 4 from the upper surface of the die 1. In other words, on either side of the die 1 a flow channel is formed whose width is equal to the height of the flow channel above a die 1 so that every section of the flow channel provided between the mold tool 4 and the matrix array of dice 1 on the substrate 3 is equally wide. This provides for a uniform mold compound flow.
In
In the embodiment illustrated in
According to the conventional method described with respect to
