CROSS-REFERENCE TO RELATED APPLICATIONS This application is a divisional of application Ser. No. 09/708,932, filed Nov. 8, 2000, pending, which is a divisional of application Ser. No. 09/389,316, filed Sep. 2, 1999, now U.S. Pat. No. 6,295,730, issued Oct. 2, 2001.
BACKGROUND OF THE INVENTION The present invention relates generally to forming contacts on a semiconductor substrate and, more specifically, to the formation of metal bump contacts or connectors on a semiconductor substrate using micromachining techniques.
Recent advances in data processing devices and memory circuits have resulted in the implementation of very large scale integrated circuits (VLSI) and even ultra large scale integrated circuits (ULSI). These VLSI and ULSI circuits are fabricated on semiconductor chips that include integrated circuits and other electrical parts. In order to mount a semiconductor chip to a carrier substrate, such as a printed circuit board or a ceramic substrate, solder bumps are arranged onto one of the semiconductor chips and the carrier substrate so that the semiconductor chip can be mechanically and electrically connected via metallurgical processes by melting the solder bumps.
One approach to applying and forming solder bumps and a carrier substrate is to use a solder paste. The solder paste is printed onto the carrier substrate and leads extending from the semiconductor chip are placed on the solder paste on the carrier substrate. The structure is then heated to cause the solder in the solder paste to melt so that the semiconductor chip can be mechanically and electrically connected to the carrier substrate. To place the solder paste onto the carrier substrate, a metal mask with predetermined openings is typically used. The solder paste is applied to the surface of the metal mask and a wiper is moved across the surface of the mask, thus pushing the solder paste through the openings of the metal mask onto the surface of the carrier substrate. Such masks are typically referred to as stencils.
Unfortunately, as the critical dimensions of the integrated circuits become smaller and smaller, the amount of solder paste that can be pressed through a given stencil becomes smaller and the placement of the solder paste becomes even more difficult. Additionally, with the smaller critical dimensions, the stencil mask becomes even more difficult to clean for a subsequent solder paste application as well as being subject to high rates of wear because of the constant placement of the stencil, application of the paste to the stencil, and removal and cleaning of the stencil.
Another method of placing conductive contacts for connecting the semiconductor chip to the carrier substrate has been to use preformed solder balls that are placed directly upon either the carrier substrate or the semiconductor chip with precisely controlled placement. Once the solder balls are in place, the solder balls are subjected to heat to cause a partial reflow so that the solder balls adhere to the solder pad. Unfortunately, in this process, as the critical dimensions of the features on the semiconductor chip tend to decrease, significant disadvantages become apparent in using this type of technique. One disadvantage is that the processing costs due to the limited process reliability and the speed of the pick and place nature of the transfer process become more evident. Another disadvantage is that the physical handling and placement of the solder balls by the machine dictates the minimum spacing allowed between solder bumps on a semiconductor chip or carrier substrate and, thus, requires a semiconductor chip that would be larger than otherwise necessary for the desired VLSI or ULSI circuitry.
Additional problems involve the uniformity of the preformed solder balls. At smaller and smaller ball sizes, the average diameter of the preformed solder ball may vary greatly from the desired diameter of the preformed solder ball. This wide discrepancy in uniformity can lead to several problems. Preformed solder balls not only cannot be applied where desired, but when a too large or too small preformed solder ball is placed upon a pad, after the formation of a connection using such a preformed solder ball, typically the location will be noted as either having several bad connections surrounding a ball that is too large or having a defective connection where a ball is too small. Large diameter preformed solder balls tend to prevent adjacent acceptable preformed solder balls from mechanically and electrically connecting between the carrier substrate and the semiconductor chip. Small diameter preformed solder balls are not large enough in diameter to connect to either of the two structures since the adjacent acceptable preformed solder balls are larger in diameter than the smaller ball, which can only touch one of the two surfaces.
Yet another technique has been developed that uses a method for forming solder balls on a semiconductor plate having apertures. One such technique is described in U.S. Pat. No. 5,643,831, (the '831 patent) entitled “Process for Forming Solder Balls on a Plate Having Apertures Using Solder Paste and Transferring the Solder Ball to Semiconductor Device”, issued Jul. 1, 1997. The '831 patent discloses a method for fabricating a semiconductor device using a solder ball-forming plate having cavities. Solder paste is placed in the cavities using a solder paste application, such as a squeegee. Once the cavities are filled with solder paste, the solder ball-forming plate is heated to form solder balls in the cavities while the plate is in an inclined position. The solder balls are then transferred from the plate to a semiconductor chip.
The solder ball-forming plate is fabricated from a semiconductor material such as silicon, according to the following method. Initially, a substantially uniform flat surface is formed on the plate. Next, a plurality of cavities is formed in the flat surface of the plate. The cavities are formed by etching the semiconductor materials after a mask has been formed on the flat surface, each cavity having the shape of a precisely formed rhombus or parallelogram.
Yet another example of using a solder ball-forming plate is disclosed in U.S. Pat. No. 5,607,099, (the '099 patent) entitled “Solder Bump Transfer Device for Flip-Chip Integrated Circuit Devices,” issued Mar. 4, 1997. The '099 patent discloses a carrier device that has cavities formed in its surface for receiving and retaining solder material. The solder material can then be transferred to a flip-chip as solder bumps. The cavities are located on the surface of the carrier device such that the location of the solder material corresponds to the desired solder bump locations on the flip-chip when the carrier device is placed in alignment with the flip chip. The size of the cavities can be controlled in order to deliver a precise quantity of solder material to the flip-chip. Further, in the '099 patent, the apertures are fabricated so that they have a width of about 300 μm at the surface of the die and a width of about 125 μm at its base surface. Meanwhile, in the '831 patent, the rhombus-shaped cavities are designed to produce a ball size of about 100 μm in diameter. Unfortunately, both of these structures cannot yet produce a ball size for a solder ball that approaches the dimensions currently required in placing a semiconductor chip upon a carrier substrate using the flip-chip technology. Additionally, the solder ball-forming cavities are limited in shape.
Accordingly, it would be advantageous to overcome the problems of producing and using solder balls having uniform sizes as have been shown in the prior art approaches of utilizing preformed solder balls or to use metal masks or stencils to apply solder paste for reflow into solder balls. Additionally, it would be advantageous to make even smaller, more precisely formed solder balls than is possible in the prior art as well as to fabricate metal traces during the same step as that of forming solder balls using a solder ball-forming plate.
Not only would it be advantageous to overcome the problems of producing uniform solder ball sizes for use in connecting a device to a substrate, but it would also be beneficial to provide a way of greatly improving the precision with which solder connections are made in alignment.
BRIEF SUMMARY OF THE INVENTION According to the present invention, metal traces and solder bump pads are formed on a semiconductor substrate by way of a semiconductor template that has been micromachined to receive solder paste material. The solder paste material is then formed into precisely controlled ball shapes and metal trace geometries. First, a semiconductor substrate is covered with a mask material for protecting selected surfaces of the substrate that are not to be etched. Next, a mask is applied in order to anisotropically etch the substrate surface below. Solder ball sites and metal trace channels are formed at this time. A solder nonwettable material is applied to the exposed surfaces of the solder ball sites and the metal trace channels. A solder paste can then be applied uniformly across the surface of the substrate, thus filling in any sites and channels, or both, that are used to form the desired balls. The semiconductor template is then applied solder side to a second substrate so that the solder balls and traces can be applied directly on the second substrate, the solder balls being subsequently formed on the second substrate by the heating thereof to form the solder paste into a solder ball.
BRIEF DESCRIPTION OF THE SEVERAL VIEWS OF THE DRAWINGS FIGS.1A-D illustrate a cross-sectional view of steps used in forming solder-receiving holes and channels in a substrate mold according to the present invention;
FIG. 2 depicts a surface of the substrate mold having a plurality of cavities formed therein;
FIG. 3 illustrates the application of solder paste to the cavities and traces of the substrate mold ofFIG. 2;
FIG. 4 depicts the formation of solder bumps in the first substrate mold as mated to a second substrate;
FIG. 5 depicts the second substrate having metal bumps and traces before final reflow;
FIG. 6 illustrates the formation of metal balls on the second substrate after reflow;
FIG. 7 illustrates a schematic diagram of a mold system using the solder mold according to the present invention;
FIG. 8 depicts a surface of a second embodiment of the substrate mold of the present invention having a plurality of hemispherical cross-sectional shaped cavities formed therein prior to the removal of the resist coating on the surface of the substrate mold;
FIG. 9 depicts the substrate mold ofFIG. 8 having solder paste in the cavities formed therein in contact with a second substrate;
FIG. 10 depicts the second substrate having the solder paste applied thereto after the second embodiment of the substrate mold of the present invention ofFIG. 8 is removed;
FIG. 11 depicts the second substrate ofFIG. 10 after the solder paste has been heated to form solder balls thereon;
FIG. 12 depicts a surface of a third embodiment of the substrate mold of the present invention having a plurality of rectangular cross-sectional shaped cavities formed therein;
FIG. 13 depicts the substrate mold ofFIG. 12 having solder paste in the rectangular cavities in contact with a second substrate;
FIG. 14 depicts the second substrate ofFIG. 13 having the rectangularly shaped solder paste thereon after having been removed from the substrate of the third embodiment of the invention by the heating thereof;
FIG. 15 depicts the second substrate after the heating of the solder paste thereon to form solder balls;
FIG. 16 depicts a fourth embodiment of a substrate mold of the present invention having a plurality of cavities in a surface thereof and a plurality of heating elements on the other surface thereof;
FIG. 17 depicts the substrate mold ofFIG. 16 having solder paste in the cavities formed in a surface thereof; and
FIG. 18 depicts the other side of the substrate mold ofFIG. 16 illustrating the plurality of heating elements thereon along section line18-18 of drawingFIG. 17.
DETAILED DESCRIPTION OF THE INVENTION Illustrated in drawingFIGS. 1A-1D is a method for fabricating the semiconductor substrate to form metal bumps or metal traces, or both, on the surface of a secondary substrate. A semiconductor substrate, typically a flat planar substrate having a flat planar upper surface, a flat planar lower surface, and a plurality of planar sides forming the periphery of the substrate, is selected to serve as a bump-formingsubstrate mold10. The semiconductor substrate may be of any desired size and geometric shape suitable for use with an associated semiconductor device. The semiconductor substrate is selected from a semiconductor base material such as silicon, gallium arsenide, silicon on insulator, which may include silicon on glass or sapphire, or other well-known semiconductor substrate materials, as well as other similar types of materials, which are capable of being precisely micromachined and having a coefficient of thermal expansion (CTE) similar to that of the semiconductor materials. In this particular application, it is preferred that a silicon substrate is used forsubstrate mold10, although any of the other base materials may be freely substituted therefor. The silicon substrate is aligned such that the flat, planarupper surface12 ofsubstrate mold10 defines the <100> plane of thesubstrate mold10 which mates with a semiconductor device (not shown). As is shown in drawingFIG. 1 A, the flat, planarupper surface12 ofsubstrate mold10 has a firstprotective mask layer14 located thereon. The firstprotective mask layer14 serves to protect the surface ofsubstrate mold10 when a subsequent etch is performed to make the cavities or apertures in the flat, planarupper surface12. Firstprotective mask layer14 may be selected from particular etch-resistant materials such as nitride, oxide, or a hardened polymer spin-on mask.Substrate mold10 typically has a thickness of about 25 to 28 mils.
Next, in drawingFIG. 1B, aphotoresist16 is applied over the surface ofmask layer14 and then exposed through a mask to define openings exposing the selected cavity locations to be formed in flat, planarupper surface12. Then, as shown inFIG. 1C, a sufficient amount of semiconductor material is removed by an anisotropic etching from the exposed portion of the flat, planarupper surface12 after penetration of the exposed portion of firstprotective mask layer14, thereby forming at least onecavity18. Using an anisotropic etching process, thecavity18 has walls sloped at 54° relative to the <100> plane of thesubstrate mold10. The anisotropic etchant may be, for example, KOH, or other etchant materials well known to those skilled in the art. Further, if straight walls are desired, a dry etch using a plasma etch apparatus may be used to formcavity18.
After the formation ofcavity18, firstprotective mask layer14 is removed using a dry-etch process that is selective to removing firstprotective mask layer14 only and not removing any of the underlying silicon either in thecavity18 thus formed or on the flat, planarupper surface12 ofsubstrate mold10. For example, if firstprotective mask layer14 is silicon dioxide, a removal substance such as phosphoric acid may be used. After the removal of the firstprotective mask layer14, arelease layer20 is formed over the entire flat, planarupper surface12 ofsubstrate mold10, particularly coveringcavity18.Release layer20 is selected from a material that is relatively nonwettable to metal solder. Such materials include silicon dioxide or silicon nitride, which can be applied using a chemical vapor deposition process. Other materials that are relatively nonwettable to metal solder may also be used, such as nonwettable polymers or the like. The resulting structure is depicted in drawingFIG. 1D.
Although drawing FIGS.1A-D illustrate only asingle cavity18, it is intended that a plurality of cavities be formed in an array acrosssubstrate mold10. An example of a solder ball forming mold orsubstrate mold10 that has such a plurality ofcavities18 is depicted in drawingFIG. 2.Release layer20 is applied and utilized to minimize the wetting of solder paste on thesubstrate mold10 when the assembly is heated in order to transfer the solder onto the bumps of the secondary surface.
Solder paste is applied, as shown in drawingFIG. 3, by use of anapplicator22, such as a squeegee, that is passed across the surface ofsubstrate mold10, pressing ametal solder paste24 into the plurality ofcavities18 and wiping the excess paste away. Thesolder paste24fills cavities18, thus forming frustoconically shaped solder bumps26 (shown inFIGS. 3 and 4).
Various types of metal solder may be used. The most widely employed types include a lead-tin combination. Other types of metal solder may include, but are not limited to, lead-silver, lead-tin-silver, lead-tin-indium, indium-tin, indium-lead, or any paste using copper or gold in combination with the lead or tin. For example, a lead-tin solder paste having a 63/37 weight ratio has a eutectic temperature of 183° C. Another type of lead-tin paste that has a 95/5 weight ratio has a eutectic temperature of about 350° C.
Once thesolder paste24 is applied to flat, planarupper surface12 ofsubstrate mold10, the entire assembly is heated to a temperature sufficient enough to slightly melt the metal solder paste in order to begin the formation of the solder bumps to be transferred. As shown in drawingFIG. 4, after this partially melted solder state has been reached,substrate mold10 is inverted and applied to the surface of acarrier substrate28, which may comprise a semiconductor device (die), wafer, or flexible substrate, such as a flex tape. The assembly of thesubstrate mold10 andcarrier substrate28 is heated to a sufficiently high enough temperature to cause solder bumps26 to slightly reflow and release from therelease layer20 formed onsubstrate mold10.Substrate mold10 is then removed and solder bumps26 adhere to bond pads, terminal pads or other conductive, solderwettable sites30 oncarrier substrate28, as shown in drawingFIG. 5. Next, an additional reflow step may be performed that causes solder bumps26 to form into approximately spherically shapedsolder balls32 attached to conductive solderwettable sites30 as depicted in drawingFIG. 6.
Because of the generally trapezoidal shape of solder bumps26, the solder paste, upon heating reflow, draws into a substantially spherical shape and is held together by the surface tension of the solder material to form approximately spherically shapedsolder ball32 or a truncated spherical ball (not shown).
Although it has been depicted how solder balls or bumps32 are formed in drawingFIG. 4, it is also possible to form metal traces usingsubstrate mold10. The same type of patterning and etch steps as described with respect toFIGS. 1A-1B would be followed, but would include a layout that would form metal traces or channels.
A solder mold system is depicted in drawingFIG. 7 which incorporates thesubstrate mold10 shown in drawingFIGS. 1-6. The mold system includessolder applicator22 for spreadingmetal solder paste24 as dispensed bymetal paste dispenser52. Once the paste is sufficiently in place within thecavities18, thesubstrate mold10 is mated to a secondary substrate, as shown in drawingFIG. 4, and then placed in a low-temperature metalpaste reflow oven54 to melt the paste to a sufficient enough consistency to form self-supported bumps and has sufficient enough tackiness to wet the conductive gates on thecarrier substrate28.
Referring to drawingFIG. 8, an alternative embodiment of asubstrate mold40 of the present invention is illustrated. Thesubstrate mold40 is similar to thesubstrate mold10 described hereinbefore as to construction and methods of construction except that thecavities18 formed therein are hemispherically shaped. As illustrated, the firstprotective mask layer14 used to form the plurality ofcavities18 is present on portions of the flat, planarupper surface42 of thesubstrate mold40. As with thesubstrate mold10, thesubstrate mold40 may include arelease layer20 to aid in the release of the solder paste contained within thehemispherical cavities18.
Referring to drawingFIG. 9, once thesolder paste24 is applied to flat, planarupper surface42 ofsubstrate mold40, as described herein with respect tosubstrate mold10 illustrated in drawingFIG. 3, the entire assembly of thesubstrate mold40 andcarrier substrate28 having conductive sites orbond pads30 located thereon for thesolder paste24 to be applied is heated to a temperature sufficient enough to slightly melt themetal solder paste24 in order to begin the formation of the solder bumps to be transferred.
As shown in drawingFIG. 9, after this partially melted solder state has been reached, the assembly of thesubstrate mold40 and thecarrier substrate28 is inverted so that thesolder paste24 incavities18 is applied to theconductive sites30 on the surface of thecarrier substrate28, which may comprise a semiconductor device (die), wafer, or flexible substrate, such as a flex tape. The assembly of thesubstrate mold40 andcarrier substrate28 is heated to a sufficiently high enough temperature to cause solder bumps26 to slightly reflow and release from therelease layer20 formed onsubstrate mold40.Substrate mold40 is then removed and solder bumps26 adhere to the conductive sites, bond pads, terminal pads or other conductive, solderwettable sites30 oncarrier substrate28, as shown in drawingFIG. 10. Next, an additional reflow step may be performed that causes solder bumps26 to form into approximately spherically shapedsolder balls32 attached toconductive sites30 as depicted in drawingFIG. 11.
Because of the generally hemispherical shape of solder bumps26, the solder paste, upon heating reflow, draws into a substantially spherical shape and is held together by the surface tension of the solder material to form approximately spherically shapedsolder balls32 or truncated spheres.
Referring to drawingFIG. 12, an alternative embodiment of asubstrate mold50 of the present invention is illustrated. Thesubstrate mold50 is similar to thesubstrate molds10 and40 described hereinbefore as to construction and methods of construction except that thecavities18 formed therein are generally rectangular, or square shaped (shown in dashed lines). The firstprotective mask layer14 used to form the plurality ofcavities18 present on portions of the flat, planarupper surface42 of thesubstrate mold50 is not illustrated. As with thesubstrate mold10, thesubstrate mold50 may include arelease layer20 to aid in the release of the solder paste contained within thehemispherical cavities18. Referring to drawingFIG. 13, once thesolder paste24 is applied to flat, planarupper surface42 ofsubstrate mold50, as described herein with respect tosubstrate mold10 illustrated in drawingFIG. 3, the entire assembly of thesubstrate mold50 andcarrier substrate28 having conductive sites orbond pads30 located thereon for thesolder paste24 to be applied is heated to a temperature sufficiently high enough to slightly melt themetal solder paste24 in order to begin the formation of the solder bumps to be transferred.
As shown in drawingFIG. 13, after this partially melted solder state has been reached, the assembly of thesubstrate mold50 and thecarrier substrate28 is inverted so that thesolder paste24 is applied to theconductive sites30 on the surface of thecarrier substrate28, which may comprise a semiconductor device (die), wafer, or flexible substrate, such as a flex tape. The assembly of thesubstrate mold50 andcarrier substrate28 is heated to a sufficiently high enough temperature to cause solder bumps26 to slightly reflow and release from therelease layer20 formed onsubstrate mold50.Substrate mold50 is then removed and solder bumps26 adhere to the conductive sites, bond pads, terminal pads or other conductive, solderwettable sites30 oncarrier substrate28, as shown in drawingFIG. 14. Next, an additional reflow step may be performed that causes solder bumps26 to form into approximately spherically shapedsolder balls32 attached toconductive sites30 as depicted in drawingFIG. 15.
Because of the generally rectangular shape of solder bumps26, the solder paste, upon heating reflow, draws into a substantially spherical shape and is held together by the surface tension of the solder material to form approximately spherically shapedsolder balls32.
Referring to drawingFIG. 16, another embodiment of thesubstrate mold100 of the present invention is illustrated. Thesubstrate mold100 is similar to thesubstrate molds10,40, and50 described hereinbefore. Thesubstrate mold100 includescavities18 having any desired shape as described herein in the flat, planarupper surface12 and includes electrical resistance heating strips66 located on the bottom thereof for the heating of thesubstrate mold100 withelectrical conductor68 connected thereto. The bottom surface of thesubstrate mold100 includes acoating62 thereon to electrically insulate the heating strips66 from thesubstrate mold100. The heating strips66 may be of any desired geometrical configuration to cover the bottom surface of thesubstrate mold100 to uniformly heat themold100 and thesolder paste24 located in thecavities18 thereof. Theelectrical conductor68 may be any desired shape and have any desired location for connection to the heating strips66. Theelectrical conductor68 is covered with aninsulation layer70 located thereover. In areas or portions of the bottom surface of thesubstrate mold100 not having aheating strip66 located thereon, aninsulative coating64 of any suitable type is provided.
Referring to drawingFIG. 17, thesubstrate mold100 is illustrated havingsolder paste24 located incavities18 havingrelease layer20 therein. After thesolder paste24 is placed in thecavities18, a carrier substrate28 (seeFIG. 4) is applied to thesubstrate mold100, the assembly of thesubstrate mold100 andcarrier substrate28 inverted, and the electrical resistance heating strips66 on thesubstrate mold100 actuated to heat thesolder paste24 to transfer the same to thecarrier substrate28. After thesolder paste24 is transferred to thecarrier substrate28, thecarrier substrate28 is further heated to cause the solder paste to adhere to theconductive sites30 on thecarrier substrate28 to substantially formsolder balls32 thereon.
Referring to drawingFIG. 18, the electrical resistance heating strips66 andelectrical conductor68 are illustrated. The heating strips66 may be of any desired shape to substantially uniformly heat thesubstrate mold100. Similarly, theelectrical conductor68 may be any desired shape to electrically connect to the heating strips66. Further, any desired connector may be used to electrically connect theelectrical conductor68 to a source of electrical power.
Substrate molds10,40,50 and100 described herein are useful in forming contact bumps for many applications. One application is the formation of flexible connecting tape that requires bumps for interconnection of traces on the tape to a die or other element. The micromachining ofsubstrate mold10 provides a much more accurate means for placing the solder ball shaped bumps over the prior art methods of merely placing bumps on top of a screen and then having the screen place the bumps in a proper alignment. Further, the solder ball shaped bumps have a more uniform volume and shape as the cavity dimensions in the semiconductor mold provide a substantially precise control over the formation of the solder ball shaped bumps. By contrast, in the prior art, the uniformity of solder balls has always been a problem, especially at the smaller diameter dimensions that are now being used. Another application for the present invention is for the direct placement of the solder ball shaped bumps on a semiconductor device or die for attachment. Yet another application includes placing the solder ball shaped bumps on a wafer-scale device for interconnection. This allows multiple devices placed on the same substrate to be interconnected using the precision of the solder ball shaped bumps. For example, the solder ball shaped bump application is useful in chip scale packages (CSP) or in fine ball grid array (FBGA) packages. The in situ electrical resistance heating strip allows for selecting which balls need to be transferred by selectively heating only those electrical resistance heating strips66.
The applications of providing interconnect and bump contacts are numerous. For example, the metal trace interconnect and the bump contact may be used in any type of semiconductor device such as a memory storage device. These memory storage devices can range from read-only memory (ROM) and random access memory (RAM) to exotic types of memory such as video memory and the memory used in computer systems. Additionally, the application of this metal trace interconnect and bump contact structure can be utilized in micro-processor packages that are used in computer systems as well as in other types of systems, and other types of single processing devices and support chips normally used in electronic devices. These electronic devices range from cellular phones to microwave systems, to automobiles and even to programmable wrist watches.
Although the present invention has been described with reference to a particular embodiment, the invention is not limited to this described embodiment. The invention is limited only by the appended claims, which include within their scope all equivalent devices or methods which operate according to the principles of the invention as described herein.