US 3739614 A
First workpieces, for example, beam-leaded integrated circuits, and the like, are bonded to second workpieces, for example, metallized ceramic substrates by first depositing a quantity of primary explosive, such as lead azide, onto each beam lead and then detonating the explosive to explosively bond the integrated circuits to the substrate. In another embodiment of the invention, the explosive bonding force is applied through a buffer sheet of plastic or metallic material which protects the surface of the substrate from contamination and which, in addition, dampens the shock of the explosion. In yet another embodiment of the invention, metal conductive paths are explosively bonded directly to a ceramic or glass substrate to form a "printed circuit pattern." The same techniques are used to manufacture resistors, capacitors, inductors, etc. In another embodiment in an explosive metal-working process the explosive charge is applied to the workpiece through apertures in a stencil or through windows of a silk-screen to form a plurality of discrete explosive charges.
Claims available in
Description (OCR text may contain errors)
United States Patent 1 Cranston EXPLOSIVE METAL-WORKING PROCESS  Inventor: Benjamin Howell Cranston, Trenton,
 Assignee: Western Electric Company,
Incorporated, New York, N.Y.
 Filed: Nov. 26, 1971  Appl. No.: 202,304
Related U.S. Application Data  U.S. Cl. 72/56, 29/421, 29/4701,
1 17/38  Int. Cl ..B21d 26/02, B2lj 5/04  Field of Search 72/56; 29/470],
[ 56] References Cited UNITED STATES PATENTS 2,666,008 l/l954 Enslein et al.. 156/230 X 2,984,597 5/1961 Hennes 156/237 X 3,316,627 5/1967 Suzuki et al. 29/4975 X 3,346,946 10/1967 Riegelmoyer 29/470.l 3,372,565 3/1968 Kunsogi et al. 72/56 3,451,239 6/1969 Riegelmoyer et a1 72/56 June 19, 1973 Primary Examiner-.l. Spencer Overholser Assistant ExaminerRonald J. Shore Attorney-W. M. Kain, R. P. Miller and R. C. Winter  ABSTRACT First workpieces, for example, beam-leaded integrated circuits, and the like, are bonded to second workpieces, for example, metallized ceramic substrates by first depositing a quantity of primary explosive, such as lead azide, onto each beam lead and then detonating the explosive to explosively bond the integrated circuits to the substrate. In another embodiment of the invention, the explosive bonding force is applied through a buffer sheet of plastic or metallic material which protects the surface of the substrate from contamination and which, in addition, dampens the shock of the explosion. In yet another embodiment of the invention, metal conductive paths are explosively bonded directly to a ceramic or glass substrate to form a printed circuit pattern." The same techniques are used to manufacture resistors, capacitors, inductors, etc. In another embodiment in an explosive metal-working process the explosive charge is applied to the workpiece through apertures in a stencil or through windows of a silk-screen to form a plurality of discrete explosive charges.
13 Claims, 32 Drawing Figures PMENIKU JUN! mm PATENIED 3.739.614
am 07 I II 7 CONTROL CIRCUIT Slit 08011 PATENIEU JUN 1 9 mm EXPLOSIVE METAL-WORKING PROCESS CROSS REFERENCE TO RELATED APPLICATION This is a division, of application Ser. No. 68,431 filed Aug. 3 l 1970 which is a continuation-in-part of my copending application, Ser. No. 6,829 filed Jan. 29, 1979, now abandoned which is related to the EXPLOSIVE BONDING OF BEAM LEAD-LIKE DEVICES. Said copending application is assigned to the same assignee as the instant application.
BACKGROUND OF THE INVENTION 1. Field of the Invention Broadly speaking, this invention relates to explosive bonding. More particularly, in a preferred embodiment, this invention relates to a method of explosively bonding a first workpiece to a second workpiece.
2. Description of the Prior Art In the manufacture of electronic circuitry, the use of discrete electrical components, such as resistors, capacitors, and transistors, is rapidly becoming obsolete. These discrete components are largely being supplanted by the integrated circuit, a small chip of silicon which, by series of selected masking, etching, and processing steps, can be made to perform all of the functions which may be performed by discrete components when these discrete components are suitably interconnected by conventional or printed wiring to form an operating circuit.
Integrated circuit devices are very small, the dimensions of a typical device being approximately 0.035 X 0.035 inch. While these microscopic dimensions permit a hereto-fore undreamed of degree of miniaturization, there are other reasons why these devices are made as small as they are, one reason being that the microscopic dimensions significantly improve the operating characteristics of circuits which are fabricated on IC devices. For example, the switching speed of gating circuits and the bandwidth of LF. amplifiers, are significantly improved by this miniaturization.
Of course, an integrated circuit cannot operate in vacuo, and must be interconnected to other integrated circuits and to the outside world, for example, to power supplies, input/output devices, and the like. Here, however, the microscopic dimensions are a distinct disadvantage.
Because of improved manufacturing techniques and increased yield, the cost of integrated circuits has dropped drastically in the last decade and now, in many instances, the cost of interconnecting an integrated -circuit to another integrated circuit or to the outside world exceeds the cost of the device itself, a most undesirable situation. I
In one prior art method of interconnecting integrated circuit devices, each device is bonded to the header of a multiterminal, transistor-like base. Fine gold wires are then hand bonded, one at a time, from the terminal portions of the integrated circuit to corresponding terminal pins on the transistor-like base, which pins, of course, extend up through the header for this purpose, in a well-known manner. Interconnection of the device to other devices or to the outside world is then made by plugging the base, with the integrated circuit device attached thereto, into a conventional transistor-like socket which is wired to other similar sockets, or to discrete components, by conventional wiring or by printed circuitry.
Because of the extremely small size of IC devices, and the attendant alignment problems, attempts to automate this uneconomical hand bonding process have not proved to be successful. Further, apart from the economics, the use of plug-in integrated circuit devices vitiates many of the highly desirable properties possessed by such devices, for example, the compactness which may be realized and the improved circuit perfor mance which they are capable of yielding.
For these reasons, circuit designers generally prefer to connect integrated circuits directly to an insulating substrate, such as glass or ceramic, upon which a suitable pattern of metallic, for example, aluminum or gold, conductor paths has been laid down. Unfortunately, most existing techniques for laying down metallic conductor paths on glass or ceramic are expensive and time consuming. Examples of these existing techniques include sputtering or vacuum depositing a thin metallic film on the substrate followed by the application ofa photoresist over the metallic film so deposited. Next, the photoresist is exposed, through an appropriate mark, and developed and the metal film selectively etched away to leave the desired metallic pattern on the substrate. Finally, the metallic pattern is built up to the desired thickness by the electrolytic of electroless deposition technique in which additional metal is deposited onto the existing metallic pattern. An alternate technique, known in the art, for depositing conductive metallic paths on a substrate involves screening a granular suspension of metal particles in a suitable vehicle, such as ethyl cellulose, onto the substrate, in the desired pattern, and then firing the substrate to bind and diffuse the metal granules in the surface of the substrate to thereby create the desired pattern of conductive paths on the substrate. Because of the large number of steps involved, it will be self evident that these prior art techniques are expensive and time consuming.
Returning now to the problems of bonding the devices themselves, the above-described hand-bonding technique for integrated circuit devices may, of course, be used to connect an integrated circuit device to the terminal land areas of a printed conductor pattern. However, techniques which more readily lend themselves to automation have also been developed.
U.S. Pat. No. 3,425,252, for example, which, issued to M. J. Lepselter on Feb. 4, 1969, describes a semiconductor device including a plurality of beam-lead conductors cantilevered outward from the device. To bond such a beam-leaded device to a substrate, the device is first aligned with respect to the terminal land areas of the substrate and then heat and pressure are applied to each of the beam leads, by means of a suitably shaped bonding tool, to simultaneously and automatically bond the beam leads to the substrate.
. Another bonding technique which may be used with beam-lead devices is the compliant bonding technique described in U.S. patent application, Ser.No. 65l,4l l of A. Coucoulas which was filed on July 6, I967. This application described a bonding technique wherein heat and pressure are applied by a bonding tool to the beam leads through a compliant medium, such as a sheet of 2024 aluminum. The heat and pressure which is applied causes the aluminum sheet to flow plastically and to transmit the bonding pressure to the beam leads thereby bonding the beam leads to the substrate.
The above-described techniques successfully permit the simultaneous bonding of all the beam leads of a sin gle device, and, of course, are equally well suited for large area bonding, that is to say, the case where it is desired to simultaneously bond a plurality of beamleaded devices to a single substrate. However, it is somewhat difficult to align a massive, multi-apertured bonding tool (or a plurality of closely spaced, individual bonding tools) with respect to the integrated circuit devices to be bonded. Yet another problem in large area bonding is that, while it is possible to closely control the dimensions of a given IC device and its alignment with respect to a given set of land areas on a substrate, it is very difficult to control the spacing between this set of land areas and another set of land areas at, say, the other end of the substrate. Since there is thus some uncertainty as to the exact location where each integrated circuit device will be found on the substrate, the use of a massive multi-apertured bonding tool (or a plurality of individual bonding tools) becomes difficult because of the variation in device-to-device spacing from one substrate to another.
Another reason why alternative techniques are desirable for use in large area bonding applications is the fact that it is not possible to manufacture large substrates which are substantially flat over the entire surface area of the substrate. There thus exists a substantial degree of nonparallelism between the substrate (and hence the IC devices to be bonded) and the bonding tool (or tools). This lack of parallelism may result in bonding pressures being applied to some IC devices which are far in excess of the maximum permitted pressure, resulting in damage to, or the complete destruction of, the affected devices. Similarly, the lack of parallelism may cause bonding pressures to be applied to other [C devices which are far below the minimum pressures required for satisfactory bonding, resulting in weak or non-existent bonds between the devices and the substrate.
Broadly speaking then, the problem is to find an im proved method of bonding a first workpiece to a second workpiece. In particular, an important aspect of this problem is to find a method of simultaneously bonding the microleads of a plurality of integrated circuit devices to the corresponding land areas of a substrate, after the devices have been aligned with respect to the substrate, without using a bonding tool which must itself be aligned with respect to the devices and/or the substrate or which must be provided with a complicated compensating mechanism to compensate for lack of parallelism between the substrate and the bonding tool.
A second important aspect of this problem is to find a method of forming metallic conductive paths or regions on an insulating substrate, particularly a large area substrate, without subjecting the substrate to numerous expensive and time-consuming processing steps.
I have discovered that explosive bonding provides a highly satisfactory solution to the above-described problems. The use of high explosives for metal-working purposes dates, of course, from the turn of the century; however, serious research into this subject matter was not begun until the late forties and early fifties. Initially, research was concentrated on the use of high explosives to shape massive workpieces which could not be conveniently or economically worked by any other technique. More recently, however, research has been concentrated on explosive welding; the aircraft and aerospace industries, in particular, being extremely active in this area, as explosive welding is highly attractive to these industries because of the exotic nature of the metals and alloys employed therein.
Explosive metal cladding has also proved extremely successful and is used, for example, to produce the blank cupro-nickel/copper stock used by the Govern ment to mint U.S. currency.
When compared to the dimensions of typical substrates and electronic components, the workpieces which are welded or clad by prior art explosive techniques are truly massive. For example, a typical prior art application might be to explosively clad a layer of 14 gauge titanium to the surface of a cylindrical pressure vessel, 15 feet in diameter by 30 feet long, and which is fabricated from 4 inch thick steel. As another example of the massive workpieces handled by the prior art, in the previously discussed explosive cladding of cupro-nickel/copper stock, a 10 foot by 20 foot sheet of cupro-nickel, 9/1 Oths of an inch thick, is explosively clad to a correspondingly dimensioned sheet of copper, 3 inches thick, which in turn is explosively clad to a second 9/l0ths inch thick sheet of cupronickel, to form the finished product.
By way of contrast, the miniature workpieces which are explosively bonded according to the methods of my invention are several magnitudes of order smaller. For example, a typical integrated circuit device may measure only 0.035 by 0.035 inch and the 16 or more beam leads to be bonded to the substrate are cantilevered outward from the device and may each measure only 0.0005 inch thick by 0.002 inch wide by 0.006 inch long. Further, typical ceramic or glass substrates may measure only 4 inch X 2 inch X 20 mils thick.
In prior art explosive bonding techniques, such as above described, the workpieces to be bonded are placed in proximity to each other and a sheet charge of high explosive, such as RDX (cyclotrimethylene trinitramine) is overlaid on the upper surface of one of the workpieces to be bonded. A commercial detonator is then implanted at one end of the sheet explosive, and ignited from a safe distance by means of an electrical spark. The detonator then explodes, setting off in turn an explosion in the sheet charge of RDX. The force created by this latter explosion accelerates thefirst workpiece towards the second workpiece to firmly bond them one to the other.
Because of the massive size of the workpieces used in the prior art, unwanted by-products of the explosion are not of particular concern; neither is contamination of the workpieces or damage to the workpiece surfaces. If a clean surface is required, the workpieces can easily be machined, sanded or buffed to the desired finish. Again by way of contrast, the miniature workpieces to be bonded by the methods of my invention, particu larly electronic components such as integrated circuits, are extremely sensitive to contamination. Further, because of their extremely small size, buffing, sanding or polishing of these workpieces to smooth the surfaces thereof and remove impurities therefrom is impractical, if not impossible. In addition, substrates such as glass and ceramic are extremely brittle and tend to craze or crack when subjected to sudden concentrated stresses.
The use of a buffer layer which is positioned intermediate the sheet charge of explosive and the upper surface of one of the workpieces is known in the prior art.
However, in the prior art this buffer layer is not provided for the purpose of (and indeed would be inoperative for) protecting the surfaces of the workpieces from chemical contamination or reducing stress concentrations in the workpieces. Rather, in the prior art, these buffer layers are provided to modify the characteristics of the secondary explosive material and, in particular, to reduce the velocity of detonation.
In the case of massive workpieces of the type bonded by prior art explosive bonding techniques, as much as several hundred pounds of explosive may be required. Obviously, the explosion must be performed out of doors, under the most carefully controlled safety conditions.
While the exact mechanism by which explosive bonds are formed with workpieces and explosive charges of this size is not fully known, through trial and error, certain formulae have been developed relating the quantity of explosive required to produce a satisfactory bond under given conditions and workpiece dimensions. These formulae are, for the most part empirically derived, and, therefore, do not yield satisfactory results when applied to workpieces which are several orders of magnitude smaller.
An explosive may be defined as a chemical substance which undergoes a rapid chemical reaction, during which large quantities of gaseous by-products and much heat are generated. There are many such chemical compounds and, for convenience, They are divided into two main groups: low explosives, such as gun powder; and high explosives. The latter category may be further subdivided into initiating (or primary) explosives and secondary explosives. Primary explosives are highly sensitive chemical compounds which may easily be detonated by the application of heat, light, pressure, etc. thereto. Examples of primary explosives are the azides and the fulminates. Secondary explosives, on the other hand, generate more energy than primary explosives, when detonated, but are quite stable and relatively insensitive to heat, light, or pressure. In the prior art, primary explosives are used exclusively to initiate detonation in the higher energy, secondary explosives.
Strictly speaking, the difference between a low explosive, such as gun powder, and a high explosive, such as TNT, is in the manner in which the chemical reaction occurs. The fundamental difference is between burning (or deflagration) and detonation, not between the explosive substances themselves. It is quite common to find that an explosive can either deflagrate or detonate according to the method of initiation or the quantity of explosive involved. If the mass of explosive matter is small, thermal ignition thereof, as by an open flame, usually, if not always, leads to deflagration; but if the mass exceeds a certain critical value, it is possible forthe burning to become so rapid that it sets up a shockwave front in the explosive material and detonation ensues. The critical mass'yaries from explosive to explosive, thus, for the primary explosive lead azide, the critical mass is too small to measure, whereas for TNT it is in the order of 2,000 pounds. Thus, the application of an open flame to a mass of TNT of, say, 1,800 pounds would not produce detonation but only deflagration. The application of the same open flame to 2,200 pounds of TNT, however, would produce an immediate detonation. Quantities of secondary explosive, therefore, which are smaller than the critical mass must be detonated by an intense shock, e.g., from the detonation ofa primary explosive such as lead azide and are thus of no value for the bonding of miniature workpieces.
Prior to my invention, then, primary explosives were used exclusively for initiating detonation in secondary explosives such as TNT, dynamite and the like. Because the critical mass of such primary explosives is so small as to be unmeasurable, the empirical equations developed for the use of subcritical masses of secondary explosives are inapplicable. This is primarily due to the difference in the parameters, such as the detonation velocity, of the highly sensitive primary explosives, and the relatively insensitive secondary explosives. The detonation velocity of the primary explosive mercury ful minate, for example, is approximately 2,000 meters per second, whereas the detonation velocities of the secondary explosives TNT and nitroglycerin are approximately*6,000 meters per second and 8,000 meters per second, respectively. A more detailed discussion of the thermochemistry of explosives may be found in the publications entitled, Detonation in Condensed Explosives," by J. Taylor, published by Oxford University Press, London, 1952 and Explosive Working of Metals, by J. S. Rinehart and J. Pearson, published by Macmillan, New York, 1963.
SUMMARY OF THE INVENTION Briefly, my invention comprises, in a first preferred embodiment, a method of bonding a first workpiece to a second workpiece. The method comprises the steps of: placing said first and second workpieces in juxtaposition to each other; and detonating a primary explosive in the region of the desired bond, the force created by the detonation of said primary explosive accelerating at least one of said workpieces towards the other, to thereby form an explosive bond between said workpieces.
Detonation of the explosive material is accomplished, in one embodiment of the invention, by applying heat to the workpiece. In other embodiments of the invention, detonation is accomplished by the application of light, laser, or acoustic energy to the explosive material. In still further embodiments of the invention, detonation is accomplished by means of alpha particles, shock waves, mechanical pressure, an electron beam, alternating magnetic or electric fields, an electric discharge or the provision (or removal) of a chemical atmosphere. In some embodiments of the invention, the bonding force is applied directly to the microcircuits to be bonded; in other embodiments, the bonding force is applied through a protective bonding medium.
Another embodiment of my invention comprises a method of bonding the microleads of at least one beam lead-like device to corresponding regions of a workpiece. The method comprises the steps of placing a charge of explosive material proximate each of the microleads to be bonded in a position to accelerate the microleads towards the workpiece and detonating the explosive material to explosively bond the microleads to corresponding regions of the workpiece. As before, the explosive material may be detonated by heat, light, sound, pressure, etc. and may be applied directly to the workpiece or through a protective buffer medium, such as stainless steel or a polyimide, such as KAPTON.
DESCRIPTION OF THE DRAWINGS FIG. 1 is an isometric view of an apparatus which may be utilized to deposit explosive material on the microleads of a beam lead-like device;
FIG. 2 is a partial top view of a plurality of beam-lead devices, prior to separation, and shows in greater detail the manner in which the explosive material is deposited thereon;
FIG. 3 is an isometric view of a single beam-lead device and shows the location of the explosive material on the microleads thereof in greater detail;
FIG. 4 is a partial, cross-sectional view of a beamlead device prior to the explosive bonding thereof to the land areas of a substrate;
FIG. 5 is a partial, cross-sectional view of the beamlead device shown in FIG. 4 after it has been explosively bonded to the substrate;
FIG. 6 is a partial, cross-sectional view of the beamlead device shown in FIG. 4 illustrating the use of a buffer member positioned intermediate the explosive material and the beam-lead device;
FIG. 7 is a plan view of the buffer member shown in FIG. 6 depicting the location of the explosive charges thereon in greater detail;
FIG. 8 is a partial, cross-sectional view of the beamlead device shown in FIG. 6 after explosive bonding to the substrate has occurred;
FIG. 9 is an isometric view of an apparatus for explosively bonding a plurality of beam-lead devices to a substrate by the application of light thereto;
FIG. 10 is a partially illustrative, partially schematic diagram depicting the use of light from an optical maser to detonate the explosive material;
FIG. 11 is an isometric view of an apparatus for explosively bonding a plurality of beam-lead devices to a substrate by the use of focused light from an incandescent lamp;
FIG. 12 is an isometric view of an apparatus which may be used to explosively bond a plurality of beamlead devices to the land areas of a substrate by the application of heat thereto;
FIG. 13 is a side view of an apparatus which may be used to bond a plurality of beam-lead devices to the land areas of a substrate by the use of radio frequency induction heating;
FIG. 14 is a side view of an apparatus which may be used to bond a plurality of beam-lead devices to the land areas of a substrate by the use of radio frequency dielectric heating;
FIG. 15 is an isometric view of an apparatus which may be used to bond a plurality of beam-lead devices to the land areas of a substrate by the use of acoustical energy;
FIG. 16 is a side view of an apparatus which may be used to bond a plurality of beam-lead devices to the land areas of a substrate by the use of simple mechanical pressure applied through a compliant medium;
FIG. 17 is a side view of an apparatus which may be used to bond a plurality of beam-lead devices to the land areas of a substrate by means of an electrical discharge passing through the explosive material on the beam leads;
FIG. 18 is an isometric view of an apparatus which may be used to bond a plurality of beam-lead devices to the land areas of a substrate by means of a beam of electrons;
FIG. 19A is a cross-sectional view of a beam-lead device illustrating the manner in which the upper surface of the beam leads may be rendered undulating to improve the quality of the bond; and
FIG. 19B is a similar cross-sectional view illustrating the manner in which the upper surface of the beam leads may be castellated to improve the quality of the bond;
FIG. 20 is a partial, cross-sectional view illustrating the manner in which the contact pads of a flip chip" IC device may be explosively bonded to the land areas of a substrate;
FIG. 21 illustrates an alternative embodiment of the invention which may advantageously be used to deposit conductive metal paths on an insulating substrate;
FIG. 22 illustrates the finished appearance of the ap pa'ratus shown in FIG. 21;
FIG. 23 is a side view of another embodiment of the invention in which spacing elements are provided intermediate the workpieces to be bonded to ensure the creation of a strong bond;
FIG. 24 is a side view of the elements depicted in FIG. 23 after an explosive bond has been formed;
FIG. 25 is a side view of a buffer medium having a patterned workpiece fabricated on one side thereof and I a correspondingly patterned explosive charge on the other surface thereof;
FIG. 26 is an isometric view of the buffer medium shown in FIG. 28 positioned over a substrate to which the metallic pattern is to be bonded;
FIG. 27 is an isometric view of the apparatus shown in FIG. 26 after the explosive bond has been formed;
FIG. 28 illustrates yet another embodiment of the invention which may be used to manufacture thin or thick film capacitors by explosive bonding techniques;
FIG. 29 illustrates the embodiment shown in FIG. 28 after the electrode of a capacitor has been explosively bonded to a substrate;
FIG. 30 is another view of the capacitor shown in FIG. 29 illustrating the manner in which a counterelectrode may be explosively bonded thereto; and
FIG. 31 is an isometric view of the capacitor shown in FIG. 30 after the counter-electrode has been explosively bonded thereto.
DETAILED DESCRIPTION OF THE INVENTION FIG. 1 depicts an apparatus which may be used to deposit a small quantity of explosive material on the microleads of a beam-leaded IC device, or the like. As shown, a conventional wax-coated semiconductor carrier plate 30, having a plurality of beam-leaded IC devices 31, temporarily secured thereto, is placed on the bottom surface 32 of a hollow, rectangular container 33. Carrier plate 30 is restrained from movement, and aligned, by means of a plurality of first registration pins 34 which mate with a corresponding plurality of notches 35 in carrier plate 30. A second plurality of registration pins 38 are provided at the four corners of container 33. A rectangular stencil plate 40, having a plurality of orthogonally oriented slot apertures 41 therein, is adapted to fit down inside container 33 so that registration pins 34 and 38 mate with a corresponding plurality of apertures 39 in the stencil plate. When somated, the slot apertures 41 align with the beam leads of the IC devices 31.
Referring now to FIG. 2, as is well known, each of the beam-leaded IC devices 31 is provided with a plurality of gold beam leads 42 cantilevered outward therefrom. In accordance with standard manufacturing techniques for these devices, prior to separation, the beam leads of each device are interdigitated with the beam leads ofits immediate neighbors. Registration pins 34 and 38, FIG. 1, align stencil plate 40 so that the slot apertures 41 therein are positioned intermediate each pair of beam lead devices and cross the interdigitated beam leads 42, FIG. 2, in the region of overlap.
Returning to FIG. 1, a squeegee 43 having a rubber roller 47 is slideably mounted in a frame (not shown) which in turn is attached to the walls of container 33. The rubber roller 47 is adapted to fit within container 33 and to engage the upper surface of stencil plate 40 when the plate is mated with registration pins 34 and 38 and positioned over lC carrier plate 30.
In operation, the carrier plate, bearing the IC devices whose beam leads are to be coated with explosive material, is placed on the bottom surface 32 of container 33 and aligned therewith by means of registration pins 34. Next, stencil plate 40 is fitted over the aligned carrier plate 30 and a metered quantity of explosive material deposited from a suitable container onto the stencil plate. Squeegee 43 is then lowered into engagement with the stencil plate and rolled back and forth to force the explosive material down into slotted apertures 41 and, hence, onto the beam leads of each [C device. When the metered quantity of explosive material has been consumed, the stencil plate and the carrier are removed from container 33 and the explosive material permitted to dry. The individual lC devices are then separated from the carrier by any of several conventional techniques.
It is, of course, necessary to select an explosive which is not so sensitive that the squeegee operation will causepremature detonation thereof. Typically, the explosive material is dissolved in some suitable chemical solution which facilitates the stenciling of the explosive onto the IC device. In addition, the solvent may inhibit premature detonation, at least until the solution has evaporated and the explosive material is dry.
it will be appreciated that a suitably patterned silkscreen (or other equivalent screening device) could be substituted for stencil plate 40. Other analogous printing techniques may, of course, also be used to apply the explosive to the workpiece. It will further be appreciated that this technique for depositing a patterned charge of explosive material onto a workpiece to be explosively bonded is not necessarily restricted to miniature workpieces, such as lC devices or to substrates. The technique may be used, for example, on much larger workpieces. Indeed, a patterned charge ofa conventional, secondary explosive may also be deposited on a workpiece by this technique, provided that the secondary explosive is dissolved in some suitable vehicle to render it sufficiently mobile to pass through the apertures of a stencil or a screen. In this latter event, the stencil plate or silk-screen could be re-used to screen-on the necessary charge of primary explosive required to detonate the secondary explosive.
FIG. 3 illustrates the appearance of a beam-lead device after it has been coated with explosive material and separated from its neighboring devices. As can be seen, a small quantity of explosive material 48 has been deposited on each beam lead 42. It will be apparent that the quantity of explosive deposited, and hence the bonding force produced when the explosive is detonated, may be'controlled by varying the width of the apertures in the stencil plate and/or by altering the thickness of the stencil plate, thereby affecting the amount (i.e., width and height) of explosive material deposited on the beam leads.
For some special applications, it may be desirable to deposit unequal amounts of explosive material on each beam lead. The above-described apparatus can easily accommodate this requirement by a combination of the above-described changes to the apertures of the stencil plate. Further, the apparatus may easily be adapted to handle different [C circuit configurations, or different substrate arrangements, by merely substituting an appropriately configured stencil plate. The apparatus can also handle an individual IC device, if so desired, by the use of a suitably dimensioned holder for the individual device. Advantageously, slotted apertures 41 in stencil 40 are arranged to deposit explosive material onto each beam lead no closer to the main part of the device than one-third of the length of the beam lead and no further from the device than two-thirds of the length of the beam lead. Advantageously, the average distance used in practice is approximately one-half of the length of a beam lead.
As previously discussed, in the bonding of minature workpieces, the conventional use of a secondary high explosive, which is detonated by means of a detonator, is impossible. I have discovered, however, that primary explosives may be used to bond such miniature workpieces. Of the many known primary explosives, the azides and the fulminates are probably the most widely understood, although many other chemical compounds exhibit similar characteristics and may also be used for the explosive bonding of miniature workpieces. The choice of the particular primary explosive to be used in any given bonding application is a function of the amount of explosive force required and/or the manner in which it is desired to initiate detonation. Advantageously, the detonation of the primary explosive, in accordance with my invention, may be accomplished by the application of heat, light, sound, pressure, shock waves and the introduction (or removal) of a suitable chemical atmosphere. For example, if light is employed as the detonating mechanism, then silver nitride (Ag N) or cuprous azide (Cu(N may be used as the primary explosive. Alternatively, if detonation is accomplished by means of mechanical force and pressure, mercury fulminate (C N O Hg) or lead azide (Pb(N may be used as the primary explosive.
Table A, below, lists some of the more common azide compounds, together with their critical detonation temperatures.
TABLE A THE MORE COMMON AZIDE EXPLOSIVES Critical Compound Formula Detonation Temp.
Lead Azide Pb(N;), 350 Silver Azide Ag(N 300 Titanium Azide Ti N 350 Boron Azide B(N Silicon Azide Si(N,), Mercuric Azide Hg(N 460 Copper Azide CU(N: )z 21S Cadmium Azide 'Cd(N l44 Ammonium Azide NH,( N Mercurous Azide Hg,( 3)2 2H) Table B, below, lists some of the more common fulminate compounds, together with their critical detonation temperatures.
TABLE B THE MORE COMMON FULMINATE EXPLOSIVES Critical Compound Formula Detonation Temp.
Mercury Fulminatc l-Ig(ONC) 190 Silver Fulminate Ag(ONC) 170 Copper Fulminate Cu(ONC) Table C, below, lists some additional primary explosive compounds, together with their critical detonation temperatures.
TABLE C MISCELLANEOUS PRIMARY EXPLOSIVES Critical Compound Formula Detonation Temp.
"C Mercuric Acetylide HgC 260 Mercurous Acetylide Hg C 280 Copper Acetylide CuC 280 Silver Acetylide Ag C 200 Lead Styphnate C I-I N O Pb 295 Barium Styphnate C H N O Ba 285 Silver Nitrate Ag N S Tetrazene 200 Diazodinitrophenol HOC HJ NO N( :N) 180 (DDNP) The above three tables are by no means all inclusive. There are many other unstable chemical compounds which may be classified as primary explosives and which, under appropriate conditions of temperature and pressure, might conceivably be utilized for the explosive bonding of miniature workpieces. However, the explosives listed in the above tables are of primary interest in this regard.
Turning now to FIG. 4, there is shown a crosssectional view of integrated circuit device 31 prior to its being bonded to the terminal land areas 50 of a ceramic substrate 52. A thin film 51 of grease, dirt, metal oxide, or other contaminants is shown on the upper surface of land areas 50. A similar film will generally also be present on the surface of beam leads 42 but, for the sake of clarity, this film has been omitted from the drawing.
It will be noted that each beam lead is bent upward away from the substrate to form a small angle a with the plane of the substrate. In order for a bond to form between a beam lead and the corresponding land area of the substrate, the explosive charge 48, when detonated, must accelerate the beam lead downward towards the land area with a sufficiently high impact velocity that the resultant impact pressure is of sufficient magnitude to cause substantial plastic flow of the workpieces to be joined. Thus, the yield points of the materials from which the workpieces are fabricated must be considerably exceeded by the impact pressure.
An important aspect of explosive bonding is the phenomenon known as jetting, that is, the process of material flow which occurs when two metal workpieces strike each other at sufficiently high impact velocity to cause plastic flow of the workpiece metals and the formation of a re-entrant jet" of material between the workpieces, as shown by the arrows 49 in FIG. 4. The formation of this jet of molten material is important to the establishment of a strong bond, as it removes any impurities and oxides which may be present on the surfaces of the workpieces to be bonded and brings freshly exposed, virgin metal surfaces into intimate contact in the high-pressure collision. Notwithstanding the above, some workpiece materials, for example, gold, may be satisfactorily bonded even without the presence of jetting. This is due to the inherently oxide-free surfaces of these materials. In that event, the angle which is formed between the beam lead and the substrate becomes less critical and in some instances even unimportant.
The impact pressure required to bond a beam lead to the corresponding substrate land area may be calculated from the shock Hugoniot data for the workpiece materials. Once the impact pressure required for bonding is known, the impact velocity may be calculated. This in turn yields the necessary ratio of accelerating explosive charge to metal mass (C/M), hence, the quantity of explosive material required for a given bonding operation.
The desirable jetting phenomenon, however, only occurs if the angle of impact, B, at the collision point exceeds a certain critical value. Further, there can exist either a stable jetting condition or an unstable jetting condition, the latter being undesirable as it results in a bond of poor quality.
Stable jetting will occur if the collision point at which the two surfaces first meet, travels along the interface with a velocity equal to or greater than the highest signal velocity in either of the two workpiece materials. Table D, below, lists the velocity of sound in several typical metals and, for comparison, Table E, lists the detonation velocity of several typical primary explosives.
TABLE D TABLE E Velocity of Sound Detonation Velocity in Several of Typical Typical Metals Primary Explosives Detonation Metal Velocity Explosive Velocity (m/sec) (m/sec) Gold 2030 Lead Azide 4000 Silver 2680 Lead Styphnate 5000 Aluminum 5000 Mercury Fulminate 5050 Platinum 2800 DDNP 6800 If the two workpieces to be bonded are positioned parallel to one another, the collision point velocity equals the detonation velocity of the accelerating explosive charge. It will thus be seen that for the typesof metals commonly used for microleads and land areas in the electronics industry, by the choice of an appropriate explosive material, the collision point velocity will always exceed the bulk sonic velocity in the workpiece metals.
Actually, if the collision point velocity substantially exceeds the bulk sonic velocity in the workpiece materials, another undesirable effect is noted. That is, the generation of expansion waves in the workpieces which tend to separate the inner surfaces thereof and destroy or weaken the bond immediately after its formation. The ideal situation is when the collision point velocity slightly exceeds the bulk sonic velocity so that stable jetting occurs, yet undesirable expansion waves do not occur. For parallel geometry, this condition can be achieved by slowing down the detonation velocity of the explosive material, for example, by the addition of inert materials such as liquid paraffin or French Chalk thereto, or by reducing the density of the explosive. For example, the addition of 30 percent liquid paraffin to lead azide will reduce the velocity of detonation from 4,000 m/sec to 500 m/sec, but the mixing process is difficult to control and the results are often unpredictable. For these reasons, other means must be employed to reduce the collision point velocity.
If the workpieces to be bonded are not held parallel, but rather are aligned so that they make a small angle a to one another, the collision point velocity is no longer the same as the detonation velocity of the explosive material, but falls to some fraction thereof. Thus, by varying the geometry of the bonding configuration, the collision point velocity may be adjusted so that it is only slightly more than the bulk sonic velocity in the workpiece materials, which is the optimum condition.
As previously discussed, there is a critical angle of contact B for the collision below which jetting and satisfactory bonding usually will not occur. For parallel geometry, [3 can be increased by increasing the ratio of explosive charge to mass (C/M). However, if this is attempted in nonparallel geometry, such as shown in FIG. 4, it is found that the collision point velocity also increases. There is thus an interaction between changing the impact angle ,8 so that it exceeds the critical angle below which jetting does not occur, and lowering the collision point velocity to approximately the bulk sonic velocity in the workpiece materials. Nevertheless, despite this interaction, for workpieces of the type shown in FIG. 4, and primary explosives of the types listed in Tables A, B, and C, there exists a broad range of orientations, charge densities, and explosive compounds which will simultaneously satisfy all these criteria and produce strong, sound bonds. As an example of a specific bond, which I have produced, according to the methods of this invention, a gold wire measuring 0.002 by 0.0005 inch was bonded to a gold-plated ce ramic substrate by means of from 25 to 40p. grams of lead azide. Detonation was accomplished by an electrical discharge from a 3 volt D.C. source. The wire made an angle of less than 5 to the plane of the substrate. I further discovered that bonding was facilitated if the temperature of the substrate was raised to 175C prior to passing the electrical discharge through the substrate.
FIG. 5 depicts the beam-leaded device shown in FIG. 4 after it has been explosively bonded to the substrate. The beam leads 43 are now, of course, flattened and substantially parallel to the substrate. A small area of discoloration or pitting 53 will be noted on each beam lead in the region priorly occupied by explosive material 48. This discoloration and pitting, however, does not affect the mechanical strength or electrical characteristics of the beam leads to any detectable degree.
In the explosive bonding of massive workpieces, the explosive is laid down upon the upper surface of the upper workpiece as a sheet charge. In the methods of my invention, however, the explosive material is not laid down as a sheet charge, but rather as a point charge. Thus, the region 54 in which bonding actually occurs does not extend over the entire area of the beam lead. This is of no great import, however, as it approximates the geometry which occurs in other satisfactory bonding techniques, such as thermocompression or ultrasonic bonding.
As previously mentioned, because of the size of the workpieces and the extremely large quantities of explosive materials employed, conventional explosive bonding is usually performed out of doors. Thus, the unwanted by-products of the explosion are quickly discharged into the atmosphere. Further, in the prior art, the massive workpieces employed are not particularly sensitive to contamination by these byproducts. This is not necessarily true, however, of the miniature workpieces contemplated by this invention, particularly integrated circuits and the like. Here, the byproducts of the explosion, both gaseous and particulate, pose a very real threat of contamination to the silicon or germanium material from which the active devices in the integrated circuits are fabricated. This contamination may, under certain circumstances, alter the operating characteristics of the devices or, worse, render them totally inoperative. The same is true, to a lesser extent, of thinfilm capacitors and resistors which may also be fabricated upon the same substrate. Fortunately, I have discovered that this contamination can, in part, be prevented by conducting the explosive bonding in a partial vacuum, for example, by the use of a conventional bellshaped vacuum jar. In addition, by removing the air which is normally present between the workpieces, the partial vacuum tends to increase the workpiece acceleration, thereby improving the quality of the bond. As an alternative to the use ofa partial vacuum, the explosive bonding may be effected through an intermediate buffer, such as a layer of plastic, for example the polyimide sold under the registered trademark KAP- TON, of the E. I. DuPont de Nemorus Co. Metallic material, for example, stainless steel, or the like, may also be used for the buffer medium.
FIGS. 6 and 7 illustrate the use of such a buffer layer in an explosive bonding operation. As shown therein, a film of plastic (e.g., a KAPTON film 3 mils thick) or metallic material (e.g., 303 type stainless steel 2 mils thick) 60 having a plurality of apertures 61 therein is positioned over the top surface of beam-lead device 31. The explosive material 48, which priorly was deposited directly onto the beam leads 42, is now deposited on the upper surface of the film 60. Additionally, if film 60 is plastic and, in addition, transparent, alignment of the explosive charges, with respect to the beam leads of the integrated circuit devices, may be facilitated, for exam ple, by use of the alignment technique disclosed in U.S. Patent application, Ser. No. 820,179 of F. .l. Jannett, filed on Apr. 29, 1969.
The explosive charges which are deposited onto the buffer film may be placed there by means of the apparatus illustrated in FIG. 1, or by the use of a patterned silk-screen or printed onto the film, intaglio fashion, by means of a suitable rubber or metallic roller having a raised surface thereon which corresponds to the desired locations of the explosive charges.
FIG. 8 depicts the beam-lead device shown in FIG. 6 after the explosive material 48 has been detonated. As was the case illustrated in FIG. 5, the beam leads 42 are now substantially parallel to substrate 52 and bonded to the land areas 50 of the substrate at locations 54. The buffer film 60 is forced down about device 31 by the explosion, but is not ruptured. As a result, un wanted by-products of the explosion are prevented from reaching the sensitive portions of the substrate, and damage thereto is completely avoided. Although in FIG. 6 buffer sheet 60 is depicted as being apertured so that it may be fitted over the beam-lead devices, it will be appreciated that sheet 60 could be contoured, rather than apertured, and in that event would also serve to protect the IC device from contamination as well as the substrate. After the bonding operation has been satisfactorily performed, buffer film 60 may be peeled off the substrate. If the sheet is fabricated from plastic material, however, no deleterious effects will occur if it is permitted to remain in place.
As previously mentioned, the detonation of the primary explosive, in accordance with my invention, may advantageously be accomplished by exposure to light.
Table F, below, lists some of the primary explosive compounds exhibiting this property, together with the minimum light intensity required to initiate detonation thereof.
TABLE F PHOTOSENSITIVE EXPLOSIVE COMPOUNDS Light Intensity The mechanism which renders these and other similar primary compounds sensitive to detonation by light is not fully understood. One theory is that the light is absorbed in a thin surface layer of the explosive mate rial and within 50p. seconds is degraded into heat; the explosion is then believed to occur by a normal thermal mechanism. Another theory is that the explosion occurs as a result of a direct photochemical decomposition of the explosive matter. Regardless of the theory, however, these compounds may be detonated by the application of light thereto and are useful for the explosive bonding of miniature workpieces.
FIG. 9 illustrates an apparatus which may be used to explosively bond the beam leads of an IC device using light as the detonating mechanism. It will be appreciated that this apparatus may also be used to bond other types of workpieces, for example, to explosively bond conductive metal paths onto a ceramic or glass substrate or to explosively bond the elements of capacitors, resistors, etc. to a substrate. The same is also true for the other apparatus discussed below with reference to FIGS. ll8. The illustrative example of bonding the leads of an IC device to corresponding land areas on a substrate is not intended to be limiting and is only exemplary. The beam leads of the devices 62 to be bonded are coated with a quantity of light-sensitive primary explosive, for example, silver azide, and the devices then aligned with respect to the land areas of the substrate 63 in a conventional manner. If desired, the devices may be temporarily tacked to the substrate by means of a drop of alcohol, or the like. Substrate 63 is then placed within a glass vacuum jar 64, which is exhausted by means of an exhaust pipe 65 and a pump 66. One or more photo flash lamps 67, for example, krypton-filled quartz flash lamps are positioned outside the vacuum jar so that the light which is generated by the tubes will fall upon the photosensitive material on the beam leads. Clearly, vacuum jar 64 must be transparent to the light energy from lamp 67. The vacuum jar may thus be entirely fabricated from glass or quartz or have one or more glass or quartz windows set in the walls thereof. Photo flash lamps 67 are connected via a pair of conductors 68 to a switch 69, thence to a suitable source of energizing potential 70.
In operation, switch 69 is closed to complete a circuit from source 70 to photo flash lamps 67. In a wellknown manner, the lamps fire and generate an intense burst of light which passes through the walls or windows in vacuum jar 64, and strikes the silver azide on each beam lead, detonating it and explosively bonding each of the IC devices 62 to substrate 63.
Silver azide is primarily responsive to light in the ultraviolet range (A 3,500 A units) and krypton-filled photo flash lamps of the type shown in FIG. 9 produce more than enough energy in this ultraviolet range to detonate photosensitive silver azide. The typical duration of the flash from photo flash lamps 67 is approximately 60 .s and explosion of the silver azide usually occurs within 20p.s thereafter. From Table F the critical light intensity required to detonate silver azide is 2.6 joules/cm which corresponds to 8 X 10 calories/mm? This critical light intensity is independent of the mass of explosive material used, at least in the range of from 200 to 1,500 micrograms. Unwanted by-products of the explosion are, as previously discussed, vented from vacuum jar 64 by pump 66. However, in applications where these by-products are not troublesome, the bonding process can, of course, be conducted in a normal atmosphere. The use of a transparent plastic film positioned over the IC devices for alignment purposes is, of course, possible, provided that the intensity of the photo flash is sufficient to compensate for any light energy lost in passing through the transparent film. Further, this method of detonation may also be used with an explosively coated transparent buffer member positioned over the IC device and the substrate.
If the intensity of light from photo flash lamps 67 is not sufficient, additional lamps may be provided or a simple lens system (not shown) may be placed in front of each lamp to focus the light energy therefrom and thereby increase the light intensity above that critical value needed to detonate the explosive.
I have also discovered that a laser beam may be used to detonate the light-sensitive explosive, rather than the photo flash lamp illustrated in FIG. 9. As shown in FIG. 10, light from a pulsed or Q-switched laser 71 is expanded by a pinhole beam expander 72 and collimated by a lens 73. The collimated beam of laser energy is then directed upon the IC devices 62 on substrate 63 detonating the silver azide, or other photosensitive primary explosive, deposited on the beam leads thereof. Again, the substrate and IC devices could be positioned within a transparent vacuum jar, if desired, and the laser energy applied through the walls of the jar to detonate the photosensitive explosive material.
Contrary to what might be expected, the amount of light energy required to initiate detonation of a photosensitive explosive varies inversely with the duration of the flash. Thus, a longer flash, as might be obtained, for example, from a magnesium-filled flash bulb, would have to be several times as intense to produce detonation of the same explosive material. Further, due to thermal lag, if the duration of the flash is too great, the explosive material will deflagrate rather than detonate, regardless of the intensity. Thus, the use of pulsed light sources is, generally speaking, preferable to the use of a continuous light source.
FIG. 11 illustrates the use of a conventional light source to explosively bond a plurality of beam devices to a substrate. As shown, a quartz incandescent lamp 74 is positioned at the focus of an ellipsoidal reflector 75. The filament of lamp 74 is connected by a circuit 76 and a switch 77 to a suitable source of energizing potential 78. When switch 77 is closed and lamp 74 energized, the focused light therefrom impinges upon the surface of substrate 63 and detonates the photosensitive explosive deposited on the beam leads of the integrated circuits to be bonded.