This application is a continuation of application Ser. No. 12/138,453, filed on Jun. 13, 2008, now pending.
BACKGROUND OF THE INVENTION(1) Field of the Invention
The invention relates to the manufacturing of high performance Integrated Circuit (IC's), and more specifically to methods of achieving high performance of the Integrated Circuits by reducing the parasitic capacitance and resistance of interconnecting wiring on chip.
(2) Description of the Prior Art
When the geometric dimensions of the Integrated Circuits are scaled down, the cost per die is decreased while some aspects of performance are improved. The metal connections which connect the Integrated Circuit to other circuit or system components become of relative more importance and have, with the further miniaturization of the IC, an increasingly negative impact on the circuit performance. The parasitic capacitance and resistance of the metal interconnections increase, which degrades the chip performance significantly. Of most concern in this respect is the voltage drop along the power and ground buses and the RC delay of the critical signal paths. Attempts to reduce the resistance by using wider metal lines result in higher capacitance of these wires.
To solve this problem, the approach has been taken to develop low resistance metal (such as copper) for the wires while low dielectric materials are used in between signal lines. Increased Input-Output (IO) combined with increased demands for high performance IC's has led to the development of Flip Chip Packages. Flip-chip technology fabricates bumps (typically Pb/Sn solders) on Al pads on chip and interconnect the bumps directly to the package media, which are usually ceramic or plastic based. The flip-chip is bonded face down to the package medium through the shortest path. These technologies can be applied not only to single-chip packaging, but also to higher or integrated levels of packaging in which the packages are larger and to more sophisticated substrates that accommodate several chips to form larger functional units.
The flip-chip technique, using an area array, has the advantage of achieving the highest density of interconnection to the device and a very low inductance interconnection to the package. However, pre-testability, post-bonding visual inspection, and TCE (Temperature Coefficient of Expansion) matching to avoid solder bump fatigue are still challenges. In mounting several packages together, such as surface mounting a ceramic package to a plastic board, the TCE mismatch can cause a large thermal stress on the solder-lead joints that can lead to joint breakage caused by solder fatigue from temperature cycling operations.
U.S. Pat. No. 5,212,403 (Nakanishi) shows a method of forming wiring connections both inside and outside (in a wiring substrate over the chip) for a logic circuit depending on the length of the wire connections.
U.S. Pat. No. 5,501,006 (Gehman, Jr. et al.) shows a structure with an insulating layer between the integrated circuit (IC) and the wiring substrate. A distribution lead connects the bonding pads of the IC to the bonding pads of the substrate.
U.S. Pat. No. 5,055,907 (Jacobs) discloses an extended integration semiconductor structure that allows manufacturers to integrate circuitry beyond the chip boundaries by forming a thin film multi-layer wiring decal on the support substrate and over the chip. However, this reference differs from the invention.
U.S. Pat. No. 5,106,461 (Volfson et al.) teaches a multi layer interconnect structure of alternating polyimide (dielectric) and metal layers over an IC in a TAB structure.
U.S. Pat. No. 5,635,767 (Wenzel et al.) teaches a method for reducing RC delay by a PBGA that separates multiple metal layers.
U.S. Pat. No. 5,686,764 (Fulcher) shows a flip chip substrate that reduces RC delay by separating the power and I/O traces.
SUMMARY OF THE INVENTIONIt is the primary objective of the present invention to improve the performance of High Performance Integrated Circuits.
Another objective of the present invention is to reduce resistive voltage drop of the power supply lines that connect the IC to surrounding circuitry or circuit components.
Another objective of the present invention is to reduce the RC delay constant of the signal paths of high performance IC's.
Yet another objective of the present invention is to facilitate the application of IC's of reduced size and increased circuit density.
Yet another objective of the present invention is to further facilitate and enhance the application of low resistor conductor metals.
Yet another objective of the present invention is to allow for increased I/O pin count for the use of high performance IC's.
Yet another objective of the present invention is to simplify chip assembly by reducing the need for re-distribution of I/O chip connections.
Yet another objective of the present invention is to facilitate the connection of high-performance IC's to power buses.
Yet another objective of the present invention is to facilitate the connection of high-performance IC's to clock distribution networks.
Yet another objective of the present invention is to reduce IC manufacturing costs by allowing or facilitating the use of less expensive process equipment and by accommodating less strict application of clean room requirements, this as compared to sub-micron manufacturing requirements.
Yet another objective of the present invention is to be a driving force and stimulus for future system-on-chip designs since the present invention allows ready and cost effective interconnection between functional circuits that are positioned at relatively large distances from each other on the chip.
Yet another objective of the present design is to form the basis for a computer based routing tool that automatically routes interconnections that exceed a pre-determined length in accordance with the type of interconnection that needs to be established.
The present invention adds one or more thick layers of polymer dielectric and one or more layers of thick, wide metal lines on top of the finished device wafer passivation. The thick layer of dielectric can, for example, be of polyimide or benzocyclobutene (BCB) with a thickness of over, for example, 3 micrometers. The wide metal lines can, for instance, be of electroplated copper or gold. These layers of dielectric and metal lines are of primary benefit for long signal paths and can also be used for power buses or power planes, clock distribution networks, critical signal, re-distribution of I/O pads for flip chip applications. Single, dual and triple damascene techniques, or combinations thereof, are used for forming the metal lines and via fill.
Furthermore, a method for forming a post-passivation, top metallization system for high performance integrated circuits is provided. An integrated circuit is provided, having devices formed in and on a semiconductor substrate. An overlaying fine line interconnecting metallization structure with first metal lines is connected to the devices, and has a passivation layer formed thereover, with first openings in the passivation layer to contact pads connected to the first metal lines. A top metallization system is formed above the passivation layer, connected to the interconnecting metallization structure, wherein the top metallization system has top metal lines, in one or more layers, having a width substantially greater than the first metal lines, and wherein the top metallization system connects portions of the interconnecting metallization structure to other portions of the interconnecting metallization structure.
BRIEF DESCRIPTION OF THE DRAWINGSFIGS. 1a-1bshow a cross section of the interconnection scheme of the present invention.
FIG. 2 shows a cross section of the present invention in a more complex circuit configuration.
FIG. 3ashows the top view of a combination power and X-signal plane using the present invention.
FIG. 3bshows the top view of a combination power and Y-signal plane using the present invention.
FIG. 4 shows the top view of solder bump arrangement using the present invention and is an expanded view of a portion ofFIG. 5.
FIG. 5 shows the top view of an example of power/ground pads combined with signal pad using the present invention.
FIG. 6 shows a basic integrated circuit (IC) interconnect scheme of the invention.
FIG. 7 shows an extension of the basic IC interconnect scheme by adding power, ground and signal distribution capabilities.
FIG. 8 shows an approach of how to transition from sub-micron metal to wide metal interconnects.
FIG. 9 shows detail regarding BGA device fan out using the invention.
FIG. 10 shows detail regarding BGA device pad relocation using the invention.
FIG. 11 shows detail regarding the usage of common power, ground and signal pads for BGA devices using the invention.
FIGS. 12a-12hshows a method for transitioning from a fine-line interconnection to the post passivation interconnection of the invention, and one method for forming the post passivation interconnection.
FIGS. 13-21 show a method for forming the thick, wide metal lines, and via fill, of the invention, using a dual damascene technique.
FIGS. 22-27 show a method for forming the thick, wide metal lines, and via fill, of the invention, using a triple damascene technique.
FIG. 28 shows an embodiment of the invention in which a first metal layer of the invention is deposited directly on the passivation layer.
FIGS. 29-35 show another embodiment of the invention in which the top metal system of the invention is formed by deposition and etching.
DETAILED DESCRIPTION OF THE INVENTIONThe present invention teaches an Integrated Circuit structure where key re-distribution and interconnection metal layers and dielectric layers are added over a conventional IC. These re-distribution and interconnection layers allow for wider buses and reduce conventional RC delay.
FIG. 1ashows a cross-sectional representation of a general view of the invention.Devices2 are formed in and on asemiconductor substrate1, and metallization is accomplished in one or more layers ofIC Interconnection3, above the device layer. The IC interconnection connects the devices to one another to form operational circuits, and also has in its top layer of metal points of electrical contact (such as bond pads), which provide connections from the IC interconnection layer to outside of the IC. Apassivation layer4 covers the IC interconnection scheme, while providing openings to the electrical contact points.
In a key aspect of the invention, the passivation openings can be as small as 0.1 micrometers. In another critical aspect of the invention, various methods are used to form the PostPassivation Technology segment80, in which metal lines which are formed substantially thicker and wider than those in the IC Interconnection layer. More detail is provided below.
Referring now more specifically toFIG. 1b, there is shown a cross section of one implementation of the present invention. Asilicon substrate1 has transistors and other devices, typically formed of polysilicon, covered by adielectric layer2 deposited over the devices and the substrate.Layer3 indicates the totality of metal layers and dielectric layers that are typically created on top of thedevice layer2. Points ofcontact6, such as bonding pads known in the semiconductor art, are in the top surface oflayers3 and are part oflayer3. These points ofcontact6 are points within the IC arrangement that need to be further connected to surrounding circuitry, that is to power lines or to signal lines. Apassivation layer4, formed of for example silicon nitride, is deposited on top oflayer3, as is known in the art for protecting underlying layers from moisture, contamination, etc.
The key steps of the invention begin with the deposition of athick layer5 of a polymer. Apattern7 is exposed and etched through thepolymer layer5 and thepassivation layer4 where thepattern7 is the same as the pattern of the contact points6. This opens the contact points6 up to thesurface8 of thepolymer layer5.
In one important aspect of the current invention, referring now toFIGS. 12a-12h, and specificallyFIG. 12a,openings7 in thepolymer layer5 may be larger thanopenings7′ in thepassivation layer4.Openings7′ may be formed to as small as 0.1 micrometers, and may range in size from between about 0.1 and 50 micrometers. Thesesmall passivation vias7′ are advantageous for the following reasons:
(1) small passivation vias only need small underlying metal pads (or dog bone structures); and these small metal pads will not block the routing capability of the top layer metal in the IC fine line interconnection scheme.
(2) Since the thickness of the inter-metal-dielectric (IMD) in the IC fine line interconnection is thin, a small metal pad is needed to provide small capacitance.
Electrical contact with the contact points6 can now be established by filling the openings7 (and7′) with a conductor. Simultaneously with filling of theopenings7, a first interconnect metal layer may be formed, as shown inFIG. 12a. This thick metal layer is formed by first sputtering anadhesion layer200. The adhesion layer is formed of titanium tungsten (TiW), chromium (Cr), titanium (Ti), palladium (Pd), nickel (Ni) or titanium nitride (TiN), and is deposited to a thickness of between about 0.01 and 3 microns. Anelectroplating seed layer202 is then deposited by sputtering, the seed layer material being copper (Cu), gold (Au), palladium (Pd) or nickel (Ni), formed to a thickness of between about 0.05 and 3 microns, as shown inFIGS. 12band12c. Pd is used as the seed layer when gold is to be electroplated, and Ni used as the seed layer for plating nickel. Athick photoresist203, as depicted inFIG. 12d, of between about 2 and 100 microns thickness, is next deposited and patterned over the seed layer. A thick layer of metal, such as copper (Cu), gold (Au), palladium (Pd) or nickel (Ni), is then electroplated to a thickness of between about 2 and 100 microns, as shown inFIG. 12e, to formthick metal interconnections204 and to fillopenings7. Referring now toFIGS. 12fand12g, the photoresist is then stripped, and portions of the seed metal and adhesion metal removed, using the thick metal as an etch mask.
Where Cu is used for electroplating to form the structure ofFIG. 12g, a nickel cap layer (not shown) may be used to prevent copper corrosion and to prevent interaction of the copper with the surrounding polymer.
Subsequent metal layers may be formed in a similar manner to that shown for the first metal layer inFIGS. 12a-g. For example, referring toFIG. 12h, another thick polymer layer is deposited over theinterconnect line204 and anopening223 formed for connection of the next metal layer to the first metal layer. Adhesion and electroplating seed layers are then sputtered, a thick photoresist deposited and the next thick metal layer electroplated, etc.
Thethick metal204 of the post passivation process of the invention is thicker then the typical fine-line metal layers3 by a ratio of between about 2 and 1000 times. The thick metal layers are formed into interconnecting lines that also are wider than the fine-line metal by a ratio of between about 2 and 1000 times. Thicker, wider metal in the post-passivation process of the invention reduces the resistance of these interconnections.
Alternately, theopening7 may be filled with a metal plug, formed of, for example, tungsten, and then thick metal formed to contact the via plug, using the above described electroplating process.
In one embodiment of the invention,polymer layer5 may be omitted, with the thick metal layer formed directly onpassivation layer4 and connecting to the underlying metal pads.
In one variation of the above-described process for forming the thick metal layers, damascene processes may be used, as shown inFIGS. 13-27. Please refer first toFIGS. 13-21, in which a dual damascene technique for forming 2 vias, followed by a single damascene method for forming interconnecting lines, is shown.Openings7 and7′ are formed as previously described, inpolymer layer5 andpassivation layer4, respectively.Opening7, and subsequent openings in the polymer layers of the top metallization system, can be formed in one of two ways. In a first, preferred method, polymer5 (and subsequent polymer layers) is formed of a photosensitive polymer material, such materials being known in the art. Such photosensitive polymers can be exposed and developed directly using photolithography. Alternatively, a non-photosensitive polymer is used, and patterned using photoresist and known photolithography techniques. The former method is preferred due to the savings in materials cost, from not having to use thick layers of photoresist, which is expensive.
Anadhesion layer200 and anelectroplating seed layer202 are now formed, also as previously described with reference toFIG. 12a, and as shown inFIG. 14. Copper orgold210 is electroplated up fromseed layer202 to fillopenings7′ and7, as well asabove polymer layer5, as depicted inFIG. 15. Chemical mechanical planarization (CMP) is used to remove the platedmetal210,seed layer202, andadhesion layer200, abovepolymer5, stopping onpolymer layer5. This forms viaplugs212 above and connecting to contactpoints6, as shown inFIG. 16. Via plugs212 have a width of between about 1 and 300 micrometers.
Referring now toFIGS. 17-21,polymer214 is deposited and patterned to create interconnectingline opening215, for the purposes of interconnecting two ormore contact points6 of the fine-line metallization system, using a single damascene method. In a similar manner as previously described, an adhesion layer andelectroplating seed layer216 are sputtered.Metal218 is electroplated to fillline opening215, and then planarized back, as shown inFIGS. 18-20. Thus thickinterconnecting metal line221 is formed, connecting two ormore contact points5, and having all the advantages of the invention herein described—thick, wide metal having low resistance and capacitance, and other advantages to be further described below.
Further layers of the upper metal scheme may subsequently be formed. First, anotherpolymer layer222 is deposited and patterned to form via223. Two alternatives may then be used to fill the via and form interconnecting lines. First, a dual damascene method is used, such as shown inFIG. 21b, in which asecond polymer layer224 is deposited and patterned to form interconnectingline opening225. Subsequent processing then takes place, as described above, by formingadhesion layer252, forming a seed layer, platingmetal layer254, and CMP to complete the interconnecting line. The second method is partially shown inFIG. 21c, in which via223 is first filled with adhesion andseed layers226, and then viametal227 is electroplated and planarized using CMP. Thensecond polymer224 is deposited, and patterned as shown, and then filled withadhesion layer256 andmetal258 and planarized, to form interconnecting lines.
The above described damascene techniques provide excellent planarity, and particularly when photosensitive polymer is used, is very cost-effective.
FIGS. 22-27 depict a triple damascene process for forming the first metal system abovepassivation4. Starting from the configuration shown inFIG. 13, a photoresist or otherphotosensitive polymer230 is deposited to fillopenings7 and7′, as well asabove polymer5. This layer is patterned using conventional lithography and etching to form opening232, as shown inFIG. 23. Adhesion layer and electroplating seed layer (shown as a single layer234) are sputtered as previously described, as depicted inFIG. 24, followed by gold or copper plating236, inFIG. 25. CMP is used to planarize back the plated layer, and this simultaneously forms viaplugs212 and thick,wide interconnection221. Subsequent additional layers of metal may then be formed, after deposition ofpolymer layer238 and formation ofopening240, etc, shown inFIG. 27, by using the techniques described above with reference toFIGS. 21b-21c.
In another embodiment of the invention, the first metal lines of the invention top metallization system may be formed directly onpassivation layer4, as partially shown inFIG. 28. A dual damascene technique may be used to fill via7′ andline opening252, withinpolymer layer250. Metallization is performed similarly to that discussed above and so will not be further described in detail.
An alternative to using the above-described damascene techniques for metallization of the top metal system of the invention is to use a metal deposition and etch, as shown inFIGS. 29-35. In one embodiment, pure aluminum Al is blanket sputtered and patterned to form vias and interconnecting lines. Patterning can be by dry or wet etching. Copper Cu, nickel Ni, or gold Au may also be blanket sputtered instead of aluminum Al, using an underlayer of titanium (Ti), titanium tungsten (TiW) or chromium (Cr) (as an adhesion and barrier layer), and then patterned by wet etching. The detailed process is shown starting inFIG. 30, in which theadhesion layer300 is sputtered into via7 and7′, to contact thepad6, and covers (optional)polymer layer5. InFIG. 30, the bulk metal302 (Au, Cu or Ni) is sputtered over the adhesion layer (or in the case of Al, blanket sputtering is used without the adhesion layer), to fillvias7 and7′ and to a thickness sufficient to form metal interconnecting lines, between 2 and 100 microns.
As shown inFIG. 32,photoresist304 is next deposited and patterned to allow for etching of the metal interconnecting lines of the invention,306, as shown inFIG. 33. Etching of thebulk metal302 andunderlying adhesion layer300 can be performed by dry or wet etching. Referring toFIG. 34, thephotoresist304 is stripped. Subsequent metal layers may be formed in a similar manner to that shown for the first metal layer inFIGS. 30-34. For example, referring toFIG. 35, anotherthick polymer layer308 is deposited over theinterconnect line306 and anopening310 formed for connection of the next metal layer to the first metal layer. Adhesion and bulk metal layers are then sputtered as described above, a photoresist deposited and patterned, metal etching performed, etc.
Referring now back toFIG. 1, thetops9 of the top metal conductor can now be used for connection of the IC to its environment, and for further integration into the surrounding electrical circuitry.Pads10,11 and12 are formed on top of thetop9 of themetal conductors7; these pads can be of any design in width and thickness to accommodate specific circuit design requirements. A larger size pad can, for instance, be used as a flip chip pad. A somewhat smaller in size pad can be used for power distribution or as a ground or signal bus. The following connections can, for instance, be made to the pads shown inFIG. 1: pad10 can serve as a flip chip pad,pad11 can serve as a flip chip pad or can be connected to electrical power or to electrical ground or to an electrical signal bus, pad12 can also serve as a flip chip pad. There is no connection between the size of the pads shown inFIG. 1 and the suggested possible electrical connections for which this pad can be used. Pad size and the standard rules and restrictions of electrical circuit design determine the electrical connections to which a given pad lends itself.
The following comments relate to the size and the number of the contact points6,FIG. 1. Because thesecontact points6 are located on top of a thin dielectric (layer3,FIG. 1) the pad size cannot be too large since a large pad size brings with it a large capacitance. In addition, a large pad size will interfere with the routing capability of that layer of metal. It is therefore preferred to keep the size of thepad6 small. The size ofpad6 is however also directly related with the aspect ratio of via7. An aspect ratio of about 5 is acceptable for the consideration of via etching and via filling. Based on these considerations, the size of thecontact pad6 can be in the order of 0.5 micrometers. to 3 micrometers. the exact size being dependent on the thickness oflayers4 and5. The contact points6 can comprise any appropriate contact material, such as but not limited to tungsten, copper (electroplated or electroless), chromium, aluminum, polysilicon, or the like.
The present invention does not impose a limitation on the number of contact pads that can be included in the design; this number is dependent on package design requirements.Layer4 inFIG. 1 can be a typical IC passivation layer.
The most frequently used passivation layer in the present state of the art is plasma enhanced CVD (PECVD) oxide and nitride. In creatinglayer4, a layer of approximately 0.2 micrometers. PECVD oxide is deposited first followed by a layer of approximately 0.7 micrometers. nitride.Passivation layer4 is very important because it protects the device wafer from moisture and foreign ion contamination. The positioning of this layer between the sub-micron process (of the integrated circuit) and the tens-micron process (of the interconnecting metallization structure) is of critical importance since it allows for a cheaper process that possibly has less stringent clean room requirements for the process of creating the interconnecting metallization structure.
In addition to PECVD oxide and PECVD nitride,passivation layer4 may also be formed of silicon oxynitride, phosphosilicate glass (PSG), borosilicate glass (BSG), borophosphosilicate glass (BPSG), or combinations thereof.
Layer5 is a thick polymer dielectric layer (for example polyimide) that has a thickness in excess of 2 micrometers (after curing). The range of polyimide thickness can vary from 2 micrometers. to 50 micrometers. dependent on electrical design requirements. Thepolymer layer5 is thicker than the intermetal dielectric layers in the interconnecting, fine-line, metallization structure by 2 to 500 times.
For the deposition oflayer5 the Hitachi-Dupont polyimide HD 2732 or 2734 can, for example, be used. The polyimide can be spin-on coated and cured. After spin-on coating, the polyimide will be cured at 400 degrees C. for 1 hour in a vacuum or nitrogen ambient. For thicker polyimide, the polyimide film can be multiple coated and cured.
Another material that can be used to createlayer5 is the polymer benzocyclobutene (BCB). This polymer is at this time commercially produced by for instance Dow Chemical and has recently gained acceptance to be used instead of typical polyimide application. Yet other possible materials forlayer5 include a silicone elastomer, paralyne, or parylene.
The dimensions ofopening7 have previously been discussed. The dimension of the opening together with the dielectric thickness determines the aspect ratio of the opening. The aspect ratio challenges the via etch process and the metal filling capability. This leads to a diameter for opening7 in the range of approximately 0.5 micrometers. to 3.0 micrometers. while the height for opening7 can be in the range of approximately 3 micrometers. to 20 micrometers. The aspect ratio ofopening7 is designed such that filling of the via with metal can be accomplished. The via can be filled with CVD metal such as CVD tungsten or CVD copper, with electro-less nickel, with a damascene metal filling process, with electroplating copper, etc.
It must be noted that the use of polyimide films as inter-level dielectrics has been pursued as a technique for providing partial planarization of a dielectric surface. Polyimides offer the following characteristics for such applications:
they produce surfaces in which the step heights of underlying features are reduced, and step slopes are gentle and smooth.
they are available to fill small openings without producing the voids that occur when low-temperature CVD oxide films are deposited.
the cured polyimide films can tolerate temperatures of up to 500 degrees C. without degradation of their dielectric film characteristics.
polyimide films have dielectric breakdowns, which are only slightly lower than that of SiO.sub.2.
the dielectric constant of polyimides is smaller than that of silicon nitride and of SiO.sub.2.
the process used to deposit and pattern polyimide films is relatively simple.
For all of the above characteristics, polyimides are used and recommended within the scope of the present invention.
FIG. 2 shows how the present invention as indicated inFIG. 1 can be further extended to include multiple layers of polyimide and, in so doing, can be adapted to a larger variety of applications. The lower level build up of this cross section is identical to the build up shown inFIG. 1 with asilicon wafer1, thepoly silicon layer2, the metal and dielectric combinedlayer3, thepassivation layer4, thepolyimide layer5 and thepads10 deposited on top oflayer5. The function of the structure that has been described inFIG. 1 can be further extended by depositing another layer ofpolyimide14 on top of the previously depositedlayer5 and overlaying thepads10. Selective etching and metal deposition can further create contact points12. These contact points12 can be connected withpads10 as shown byconnector13. Depositingpads12 on top oflayer14 can thus further extend this process. Thesepads12 can be further customized to a particular application, the indicated extension of multiple layers of polyimides greatly enhances the flexibility and usefulness of the present invention. Additional alternating layers of polyimide and metal lines and/or power or ground planes may be added abovelayers12 and16, as needed. Contact between metal layers formed in the post passivation method of the invention can be made by direct contact between two layers of metal (as shown inFIG. 12), or alternately by metal plugs formed in the openings between metal layers.
The polymer layers14 that are formed between the thick, post-passivation metal lines are formed to a thickness of between about 2 and 30 microns, after curing, and are thicker than the intermetal dielectric layers formed in the typical fine-line metal scheme (layers3) by a ratio of between about 2 and 500. The thicker, organic polymer used in the post-passivation process of the invention reduces capacitance between the thick metal lines. The inorganic materials, such as silicon oxide, used in the fine-line metallization system3, cannot be formed to such thicknesses due to a tendency to crack at these thicknesses.
FIGS. 3aand3bshow a top view of one possible use of the present invention. Interconnecting a number ofpads32 that have been created as described creates signal lines30. Additional contact points such aspoint34 can allow signal lines to pass vertically between layers. The various contact points can, for instance, be created within the surface of a power plane orground plane36. The layers within the interconnecting metallization structure of the present invention can contain signal interconnections in the X-direction, signal interconnections in the Y-direction, signal interconnections between X and or Y directions, interconnections to and/or within power and/or ground buses. The present invention further teaches the interconnection of signal lines, power and ground buses between the connected IC's and the top of the metallization system of the present invention.
FIG. 3ashows signal lines formed in the X-direction,FIG. 3bshows signal lines formed in the Y-direction.
FIG. 4 presents yet another application of the present invention. Shown inFIG. 4 is an exploded view of a part ofFIG. 5 that presents an area array I/O distribution.FIG. 4 shows pads41 (on which solder bumps can be created) and an example of a layout of the redistribution of theperipheral pads41′. The exploded view ofFIG. 4 is taken along the line2-2′ shown inFIG. 5, the redistribution of theperipheral pads41′ (seeFIG. 4) is, for clarity of overview, not shown inFIG. 5. The power or ground connections can be made to any point that is required on the bottom device. Furthermore, the power and ground planes can be connected to the power and ground planes of the package substrates.FIG. 4 shows an example of how to use the topmost metal layer to redistribute theperipheral pads41′ to becomearea array pads41. The solder bumps can then be created onpads41.
FIG. 5 shows the top surface of a plane that contains a design pattern of a combination of power orground pads52 andsignal pads54.FIG. 5 shows the pad openings in the top dielectric layer. It is to be noted that the ground/power pads52 are heavier and larger in design relative to thesignal pads54. The present invention ideally lends itself to meeting these differences in design, as they are required within the art of chip and high performance circuit design. The number of power orground pads52 shown inFIG. 5 can be reduced is there are power and/or ground planes within the chip. From this it is clear that the package number of I/O's can be reduced within the scope of the present invention which leads to a reduction of the package cost by eliminating common signal/power/ground connections within the package. For instance, a 470 I/O count on a BGA chip can, within the scope of the present invention, be reduced to a 256 I/O count using the present invention. This results in considerable savings for the overall package.
FIG. 6 shows a basic design advantage of the invention. This advantage allows for the sub-micron or fine-lines, that run in the immediate vicinity of themetal layers3 and the contact points6, to be extended in an upward direction20 throughmetal interconnect7′. This extension continues in the direction22 in the horizontal plane of themetal interconnect26 and comes back down in the downward direction24 throughmetal interconnect7″. The functions and constructs of thepassivation layer4 and the insulatinglayer5 remain as previously highlighted underFIG. 1. This basic design advantage of the invention is to “elevate” or “fan-out” the fine-line interconnects and to remove these interconnects from the micron and sub-micron level to a metal interconnect level that has considerably larger dimensions and is therefore characterized by smaller resistance and capacitance and is easier and more cost effective to manufacture. This aspect of the invention does not include any aspect of conducting line re-distribution and therefore has an inherent quality of simplicity. It therefore further adds to the importance of the invention in that it makes micron and sub-micron wiring accessible at a wide-metal level. Theinterconnections7′ and7″ interconnect the fine-level metal by going up through the passivation and polymer or polyimide dielectric layers, traverses over a distance on the wide-metal level and continues by descending from the wide-metal level back down to the fine-metal level by again traversing down through the passivation and polymer or polyimide dielectric layers. The extensions that are in this manner accomplished need not to be limited to extending fine-metal interconnect points6 of any particular type, such as signal or power or ground, withwide metal line26. The laws of physics and electronics will impose limitations, if any, as to what type of interconnect can by established in this manner where limiting factors will be the conventional limiting factors of resistance, propagation delay, RC constants and others. Where the invention is of importance is that the invention provides much broader latitude in being able to apply these laws and, in so doing, provides a considerably extended scope of the application and use of Integrated Circuits and the adaptation of these circuits to a wide-metal environment.
FIG. 7 shows how the basic interconnect aspect of the invention can further be extended to now not only elevate the fine-metal to the plane of the wide-metal but to also add power, ground and signal distribution interconnects of power, ground and signal planes at the wide-metal level. The wide-metal interconnect26 ofFIG. 6 is now extended to further include an interconnection with the via21. In typical IC design, some pads may not be positioned in a location from which easy fan-out can be accomplished to a location that is required for the next step of circuit assembly. In those cases, the BGA substrate requires additional layers in the package construction in order to accomplish the required fan-out. The invention teaches an approach that makes additional layers in the assembling of an IC feasible while not unduly increasing the cost of creating such a multi-layer interface.Ball formation28 on the surface ofinterconnect23 indicates how the invention replaces part of the conventional BGA interconnect function, the solder bump provides for flip chip assembly. Thisinterconnect28 now connects the BGA device with surrounding circuitry at the wide-metal level as opposed to previous interconnects of the BGA device at the fine-metal level. The wide-metal interconnect of the BGA has obvious advantages of cost of manufacturing and improved BGA device performance. By being able to readily extend the wide-metal dimensions it also becomes possible to interconnect power, ground and signal lines at a wide-metal level thereby reducing the cost and complexity of performing this function at the fine-metal level. The indication of28 as a ball does not imply that the invention is limited to solder bumps for making interconnects. The invention is equally applicable to wirebonding for making circuit interconnects.
FIG. 8 further shows a cross section wherein the previous linear construction of the metal interconnection running through the passivation layer and the insulation layer is now conical in form. Thesub-micron metal layer60 is covered with apassivation layer62, alayer64 of polyimide or polymer is deposited over thepassivation layer62. Thewide metal level66 is formed on the surface oflayer64. The via70 is shown as having sloping sides, these sloping sides can be achieved by controlling the photolithography process that is used to created the via70. The etching of the polyimide or polymer can for instance be done under an angle of about 75 degrees with the following curing being done under an angle of 45 degrees. Also, a photosensitive polyimide or polymer can be used, the cone shape of the via70 can in that case be achieved by variation of exposure combined with time of exposure combined with angle of exposure. Where non-photosensitive polymer or polyimide is used, a wet etch can be applied that has a gradated faster and longer time etch as the top of the via70 is being approached. The layer of wide-metal pad68 is deposited on the surface of the polymer orpolyimide layer64, the wide-metal pad deposition68 mates with the top surface of the via70 and is centered on top of this surface.
FIGS. 9 through 11 show further detail to demonstrate the concepts of BGA chip ball fan-out, pad relocation and the creation of common ground, power and signal pads.
FIG. 9 shows across section100 of a BGA chip, fiveballs101 through105 are also shown. By using theBGA substrate106 and thewiring107 within thesubstrate106, it is clear thatball101 can be repositioned tolocation111,ball102 tolocation112, etc. for the remaining solder bumps103 through105. It is clear that the separation of contact points111 through115 is considerably larger than the separation of the original solder bumps101 through105. TheBGA substrate106 is the subject of the invention, this substrate allows for spreading the distance between the contact points or balls of the BGA device to a considerable degree.
FIG. 10 shows the concept of pad relocation. BGA pad120 can be any of thecontact balls101 through105. By using theBGA substrate130 and thewiring131 that is provided within the substrate, it is clear that the BGA pads can be arranged in a different and arbitrary sequence that is required for further circuit design or packaging. Forinstance contact point101, which is on the far left side of theBGA device100, is re-routed tolocation121 which is on the second far right of theBGA substrate130. The re-arrangements of the other BGA solder bumps can readily be learned from following thewiring130 within thesubstrate131 and by tracing from solder bump to one of the contact points122 through125 of the BGA substrate.
FIG. 11 shows the interconnecting of BGA device solder bumps into common power, ground or signal pads. TheBGA chip100 is again shown with fivesolder bumps101 through105. TheBGA substrate130 contains a wiring scheme that contains in this example three wiring units, one for each for the power, ground and signal bumps of the BGA device. It is clear fromFIG. 11 thatwire arrangement132 connects BGA device solder bumps101,103 and105 to interconnectpoint138 of theBGA substrate130. It can further be seen that BGAdevice solder bump104 is connected to interconnectpoint140 of the BGA substrate by means of the wire arrangement136, while BGAdevice solder bump102 is connected to interconnectpoint142 of the BGA substrate by means of thewire arrangement134. The number of pins required to interconnect theBGA device100 is in this manner reduced from five to three. It is clear that for more BGA device solder bumps, as is the case for an actual BGA device, the numeric effect of the indicated wiring arrangement is considerably more beneficial.
The concept of pad relocation can be realized using the metal interconnection scheme described in this invention, to replace the function ofBGA substrate130. FromFIGS. 10 and 11 it can be seen that the extended functionality and extended wiring ability that are provided by the interconnect wiring schemes that are typically created in theBGA substrate130 can be created using the method of the invention, ondevice100. Some of the methods and possibilities of interconnect line routing that can be implemented using the method of the invention are highlighted in the following paragraphs.
Fan-out capability can be provided by the invention, using the metal conductors within the openings through the insulating layer and through the passivation layer that connect electrical contact pads of the top metallization structure with contact points of the interconnecting metallization structure. Each of the electrical contact points of the interconnecting metallization structure is connected directly and sequentially with at least one electrical contact point of the top metallization structure. In a fan-out scheme, the distance between electrical contact points of the top metallization structure is larger than the distance between electrical contact points of the interconnecting metallization structure by a measurable amount.
The number of electrical contact pads of the upper metallization structure can exceed the number of contact points of the interconnecting metallization structure by a considerable amount.
Pad relocation may also be accomplished by the method of the invention. Electrical contact points of the top metallization structure are connected with the contact points of the interconnecting metallization structure, directly but not necessarily sequentially, thereby creating a pad relocation effect. In this method, the distance between electrical contact points of the top metallization structure is larger than the distance between the electrical contact point of the interconnecting metallization structure by a measurable amount.
A reduction effect may also be accomplished by the method of the invention, wherein common nodes are connected together. Electrical contact points on a top surface of the top metallization structure are connected with contact points of the interconnecting metallization structure, where fewer contact points are used in the top metallization structure, since functionally equivalent contact points in the interconnecting metallization structure are connected together. That is, the number of contact points for a particular electrical function among the electrical contact points of the top metallization structure is smaller than the number of electrical contact points of the interconnecting metallization structure by a measurable amount.
Some of the advantages of the present invention are:
1) improved speed of the IC interconnections due to the use of wider metal lines (which results in lower resistance) and thicker dielectrics between the interconnecting lines (which results in lower capacitance and reduced RC delay). The improved speed of the IC interconnections results in improved performance of High Performance IC's.
2) an inexpensive manufacturing process since there is no need for expensive equipment that is typically used in sub-micron IC fabrication; there is also no need for the extreme clean room facilities that are typically required for sub-micron manufacturing.
3) reduced packaging costs due to the elimination of the need for redundant I/O and multiple power and ground connection points that are needed in a typical IC packaging.
4) IC's of reduced size can be packaged and inter-connected with other circuit or system components without limiting the performance of the IC's.
5) since dependence on ultra-fine wiring is reduced, the use of low resistance conductor wires is facilitated.
6) structures containing more complicated IC's can be created because the invention allows for increased I/O pin count.
7) more complicated IC's can be created without the need for a significant increase in re-distribution of package I/O connections.
8) power buses and clock distribution networks are easier to integrate within the design of IC's.
9) future system-on-chip designs will benefit from the present invention since it allows ready and cost effective interconnection between functional circuits that are positioned at relatively large distances from each other on the chip.
10) form the basis for a computer based routing tool that automatically routes interconnections that exceed a pre-determined length in accordance with the type of interconnection that needs to be established.
11) provide a means to standardize BGA packaging.
12) be applicable to both solder bumps and wirebonding for making further circuit interconnects.
13) provide a means for BGA device solder bump fan-out thereby facilitating the packing and design of BGA devices.
14) provide a means for BGA device pad relocation thereby providing increased flexibility for the packing and design of BGA devices.
15) provide a means for common BGA device power, ground and signal lines thereby reducing the number of pins required to interconnect the BGA device with the surrounding circuits.
16) provide a means for more relaxed design rules in designing circuit vias by the application of small passivation (0.1 micrometers or more) vias.
17) provide the means for extending a fine-wire interconnect scheme to a wide-wire interconnect scheme without the need to apply a passivation layer over the surface of the fine-wire structure.
Although the preferred embodiment of the present invention has been illustrated, and that form has been described in detail, it will be readily understood by those skilled in the art that various modifications may be made therein without departing from the spirit of the invention or from the scope of the appended claims.