«
I
SYSTEM FOR INTERCONNECTING A SEMICONDUCTOR DIE TO OTHER DEVICES
BACKGROUND OF THE INVENTION 5 This invention relates in general to semiconductors and in particular, to a system for interconnecting a semiconductor die to other devices.
In conventional semiconductor chip packages, one of the most frequently used interconnect structures
10 to connect circuits on semiconductor dies to other devices is by means of bonding wires and lead frames. One end of the bonding wire is welded to a bonding pad on the die and the other end of the wire is similarly connected to a lead of the lead frame. As is known to
15 those skilled in the art, the length of the bonding wires used should not exceed a certain value since wires that are too long will sag, which increases stress at joints with bond pads and leads. This imposes many limitations on the packaging process. Furthermore, the
20 process for welding the wires to the bond pads on the dies requires that the bond pad pitch cannot be too small; this imposes a lower limit on the bond pad pitch. This is undesirable in view of recent trends in downsizing of dies and increasing the density of
25 connections to dies.
To overcome the above difficulties, tape automated bonding or TAB has been increasingly used. In TAB, conductive traces are provided on a substrate, usually a flexible tape. Since the conductive traces
30 are supported by a substrate, the traces do not need to be limited in length as in the case of bonding wires. Furthermore, the TAB process permits a finer bond pad
pitch to be used on semiconductor dies. Nevertheless, there are still some drawbacks which remain with TAB type interconnect structures, as explained below.
In wire bond technology, the interconnect structure cannot extend over circuits on the dies to locations on the semiconductor dies; otherwise, the wires may contact such circuits to cause unintentional shorts. In TAB, while the conductive traces on tape can be made over circuits to internal parts of the die, it is difficult to determine whether the bonds so made are satisfactory. For these reasons, in both TAB and wire bond technology, the connections between semiconductor dies and the interconnect structure can only be made at the edges of the dies. With the advent of submicron technology and the continual downsizing of semiconductor dies, limiting the connections to the edges of the dies greatly limits the number of electrical connections that can be made. It is therefore desirable to provide an improved interconnect structure which permits the interconnect structure to overlap with surfaces of the die containing circuits, thereby opening up much of the die surface for connections and increasing the number of electrical connections that can be made between the semiconductor dies and other devices. Similarly, direct interconnect to the power supply bus also reduces current, crowding and noise.
In both wire bond and TAB technology, the wires or conductive traces are soldered or welded to bond pads on the semiconductor dies. In the soldering or welding process, the portions of the dies underneath the bond pads are subjected to intense pressure and heat, which may cause damage to the dies such as cratering. It is therefore desirable to provide an improved interconnect structure with reduced probability of damage to the dies when the structure is connected to the die.
Yet another conventional interconnect structure for connecting semiconductor dies to other devices is the flip chip-type system as disclosed in "Active Silicon Chip Carrier" by Bodendorf et al., IBM Technical Disclosure Bulletin, Vol. 15, No. 2, July 1972, pp. 656-657, and U.S. Patent No. 4,870,224. In the flip-chip type system, conductive traces are provided on a substrate and the surface of the semiconductor die containing circuits are placed in a face-down position facing the traces. The bond pads on the die are then soldered to the conductive traces directly, without the use of bonding wires or a tape with conductive traces thereon. The substrate may be made of a ceramic material or silicon. Typically solder bumps are provided on the traces on the substrate surface and the die correspondingly. When bumps on the die are aligned with such bumps on the substrate, the solder is heated until the solder reflows in order to connect the traces to the bond pads on the die. The use of solder bumps again sets a lower limit to bond pad pitch so that the bonding pads on the die cannot be placed too close together. For this reason there is again a limit to the number of connections that can be made between the die and other devices. It is also difficult to determine whether the solder bonds so formed are of satisfactory quality.
None of the above-described conventional interconnect systems is entirely satisfactory. It is therefore desirable to provide an improved interconnect system whereby the above-described difficulties are alleviated or avoided altogether.
SUMMARY OF THE INVENTION This invention is based on the observation that the process in forming layers, and the masking and etching processes in fabricating the semiconductor die may also be used to construct an interconnect structure to avoid the above-described difficulties of conventional interconnect systems. The process of forming insulating and conductive layers, masking certain areas of these layers and exposing the rest for etching in semiconductor fabrication has already been refined to a high degree so that the accuracy and the density of electrical connections that are possible are much higher than conventional packaging techniques such as wire bonding, TAB or flip chip-type interconnect systems. Furthermore, the use of such wafer fabrication techniques permits connections to be made on top of circuits on the die surface, so that electrical connections may be made to a much larger area of the die surface than in conventional interconnect systems. Moreover, the semiconductor die is not subjected to soldering or welding so that the chances of damage to the die during the interconnect process is much reduced. One aspect of the invention is directed towards an interconnect structure for connecting a semiconductor die to other devices. The die has a surface with circuits thereon. The structure comprises a first layer of electrically insulating material on a portion of said die surface and a second layer of electrically conductive layer in contact with at least one portion of said surface and separated from a portion of the die by said insulating layer, to connect the circuits to other devices.
Another aspect of the invention is directed towards a method for making an interconnect structure for connecting a semiconductor die to other devices where the die has a surface with circuits thereon. The
method comprises providing a first layer of electrically insulating material on a portion of said surface and providing a second layer of electrically conductive layer in contact with at least one portion of said surface and separated from a portion of the die by said insulating layer, to connect the circuits other devices.
BRIEF DESCRIPTION OF THE DRAWINGS
Figs. 1-5 are cross-sectional views of semiconductor dies and different layers for forming an interconnect structure to illustrate a method for connecting the dies to an insulating layer, copper traces and a support structure.
Figs. 6-8 are cross-sectional views of a semiconductor die and portions of various layers for forming a conductive layer which is connected to selected portions of the die but separated from the other areas of the die by the insulating layer shown in
Figs. 1-5, by the use of a patterned photoresist layer.
Figs. 9-11 are cross-sectional views of a semiconductor die and portions of an interconnect structure to illustrate further steps for adding additional alternating insulating and conductive layers for forming additional connections between the semiconductor die and other devices by etching vias and by adding alternating insulating and conductive layers by the use of patterned photoresist layers.
Figs. 12 and 13 are cross-sectional views of a semiconductor die and the interconnect structure illustrating further steps for completing the manufacture of the interconnect structure of Figs. 1-11.
Figs. 14, 15 are cross-sectional views of a completed interconnect structure electrically connecting portions of the semiconductor die to other devices through leads and of the steps in completing the packaging process.
Figs. 16-18 are cross-sectional views of a semiconductor die and interconnect structures to illustrate alternative steps for forming an electrically conducting layer on top of the insulating layers of Figs. 1-7 to illustrate an alternative method of the invention.
DETAILED DESCRIPTION OF THE INVENTION Figs. 1-5 are cross-sectional views of various structures for attaching two semiconductor devices to an insulating substrate, copper traces and a support structure. As shown in Fig. 1, a support frame 20 is attached removably to a polyamide base 22. The polyamide base 22 will form the insulating layer attached to semiconductor devices as described below. Support frame 20 is used primarily because the polyamide base 22 is flexible so that a support frame is required to support the base in subsequent processing steps. The support frame may be removed by cutting or punching the base 22 along a line adjacent to the frame. An adhesive layer such as a polyamide adhesive
24 is spun onto base 22 as shown in Fig. 2. Then a second support ring 26, copper trace 28 and two semiconductor devices 30 are attached to base 22 by means of the adhesive 24, all as shown in Fig. 3. The support frame 20 is then removed and support ring 26 then provides the necessary support to base 22. It will be noted that where base 22 is flexible, support frame 20 is necessary. Without frame 20, one would have to first attach support ring 26 to base 22 and then spin on the adhesive layer 24. The presence of ring 26, however, will hinder the spinning process. For this reason support frame 20 is preferably provided before ring 26 is attached to base 22 so that a surface of base 22 is unobstructed in the spinning process.
In reference to Figs. 3 and 4, surfaces 30a of devices 30 contain circuits; these surfaces are attached to base 22 by means of adhesive 24. The structure of Fig. 4 is then flipped over as shown in Fig. 5 so that the bottom surface of base 22 is now the top surface. A photoresist layer 32 is patterned on select portions of the polyamide base as shown in Fig. 6, exposing selected areas of base 22. The exposed portions of base 22 and the adhesive layer 24 immediately below the exposed portions of base 22 are etched away as shown in Fig. 7, exposing selected portions of surface 30a of devices 30 and of copper traces 28. The photoresist layer 32 is then removed and a conductive layer 34 is sputtered onto the exposed portions of surfaces 30a, the copper traces and on top of base 22 as shown in Fig. 8. Another photoresist layer 36 is patterned on top of the conductor layer 34, covering part of and exposing the remaining parts of the conductive layer. The exposed portion of the conductive layer 34 is then etched away. The photoresist layer 36 is removed and a polyamide isolator layer 38 is then formed (such as by spinning) on top of the structure and yet another photoresist layer 40 is formed on selected areas of the isolated layer 38 as shown in Fig. 9, exposing selected areas of the isolator area 38. Vias are then etched into the exposed areas of isolator layer 38 and another conductor layer 42 is sputtered onto the structure as shown in Fig. 10 and yet another photoresist layer 44 is patterned on top thereof, exposing selected areas of the conductor layer 42. The exposed portions of conductor layer 42 is etched away and another polyamide insulator layer 46 is spun on top thereof, selected areas of which are covered by yet another patterned photoresist layer 48 as shown in Fig. 11. The above-described process of adding alternating insulating and conductive layers can
be repeated to add additional electrical connections as required.
The exposed areas of the insulator layer 46 and all the intervening layers (42, 38, 34, 22, 24) are etched through to expose portions of the copper traces 28 as shown in Fig. 12. The support ring 26 and the remaining structure on top thereof are removed to yield a semiconductor die and interconnect assembly 60 as shown in Fig. 13. Structure 60 is then attached to a lead frame 62 as shown in Fig. 14. Structure 16 and portions of the lead frame 62 are then enclosed by an inert material 64 in an injection molding or transfer molding process to form package 70 as shown in Fig. 15. This completes the packaging process. From the above, it is evident that the objectives of the invention are achieved. Different from wire bonding and TAB technology, semiconductor wafer fabrication techniques are used for fabricating the interconnect structures 60; wafer fabrication techniques have been refined to the extent that much finer bond pad pitches on devices 30 and much finer spacings between conductive traces can be achieved compared to conventional packaging techniques such as wire bonding and TAB. Furthermore, since surface 30a of the devices 30 are covered by an insulating layer, and electrical connections are made through vias formed by etching through the insulating layer, electrical connections may be made to the dies 30 substantially at all locations of surface 30a, not just at the edges of the surface. This permits many more electrical connections to be made than conventional interconnect techniques, and also improves electrical performance of the product.
Figs. 16-18 illustrate an alternative method for forming the conductor layers such as layer 34 in
Fig. 8. The method illustrated in Fig. 16 is performed
after steps 1 through 7 illustrated in Figs. 1-7. Thus instead of removing the photoresist layer 32 as discussed above in reference to Figs. 7 and 8, selected portions of the photoresist layer 32 is removed to expose selected portions of base 22 as shown in Fig. 16. Then a conductor layer 102 is sputtered on top of the entire structure in Fig. 16, resulting in the structure shown in Fig. 17. Then the remaining portions of the photoresist layer 32 and portions of the conductor layer 102 immediately above layer 32 are removed. A mechanical and chemical cleaning system may be used for removing these layers, such as by freezing the structure, thawing and then washing it with solvent. This results in a structure 110 as shown in Fig. 18. Thus as illustrated in Figs. 16-18, in order to form a conductor layer at desired locations, instead of forming a photoresist pattern over the previously formed conductor layer and then etching the exposed portions of the conductor layer, it is possible to first pattern the photoresist layer on top of the insulating layer, sputter the conductor layer on top of the photoresist, and then remove all of the patterned portions of the photoresist layer together with the portions of the conductor layer immediately on top of the photoresist layer. This method may be used for forming the conductor layers 34 and 42 as well. However, it is also possible to continue the processing of structure 110 using the process steps described in reference to Figs. 9-15 to complete the packaging process. In such event, layer 102 could be treated as layer 34 in Fig. 9.
The invention has been described by reference to preferred structures and processes described above. It will be understood, however, that various modifications may be made without departing from the scope of the invention which is to be limited only by the appended claims.