IEEE TRANSACTIONS ON Parts, Hybrids, and Packaging, vol. PHP-8, no. 3, September 1972

Explosive Bonding of Electrical Interconnections B. H. CRANSTON

Abstract-Primary explosive systems, as described. are now being used as a safe and predictable energy source to bond metals on a size scale comparable to microcircuit electrical interconnections.

Basic primary explosive reaction characteristics have bean analyzed and related to effective bond mechanisms for solid-phase explosive .bonding of several metal combinations. Methods and materials systems were devised to accurately deposit a specific quantity and configuration of explosive with predetermined reaction characteristics. This has permitted a significant advance in use of explosive bonding on a microscale. The theoretical explosive bonding mechanism is supported by experimental data. Included in considerations are surface jetting characteristics which influence interface structures and are responsible for removing surface oxides and contaminants.

A large variety of similar and dissimilar metal pairs was joined reliably and repeatedly with bond strengths greater than those of the parent metals. Explosive bonding advantages and disadvantages con- sidering materials and bonding equipment requirements are cited.

Introduction

Solid-phase joining of electrical leads to contact areas is often preferred. ‘Keeping these materials below their melting points minimizes complex metallurgical problems. Among the popular bonding techniques being used extensively are thermo- compressioh and ultrasonic. Though the application of energy in the variousprocesses differs to achieve a metallurgical bond, they all meet the same basic requirement, that is, to bring the metal atoms into intimate contact. To accomplish a metal- lurgical bond, two things must happen: the materials must deform in order to conform at the mating surfaces and there must be a method of dispersing surface films (oxides, nitrides, and absorbed gases).

Currently, a new technique of solid-phase joining using explosive energy has become popular. Thus far, the process has been used exclusively for joining large metal pieceparts. Liter- ature searches reveal conflicting reports as to the origin of the process, but the most logical explanations relate to both the accidental bonding in metal forming operations and adherence resulting from bullet and shrapnel impact on metal surfaces [ I] . However, recent reports concerning explosive joining

.express positive results. Each solid-phase bonding technique has limitations dictated

by the materials to be joined. The more familiar processes are governed by the plastic properties of the materials. Explosive joining, however, imposes a completely different set of param- eters. The high velocity collisions that produce intermetallic bonds can be explained in terms similar to those defining kinetic energy. Fig. 1 shows two metals being explosively joined. To formulate reliable bond parameters, the bulk sonic

Manuscript received May 17, 1972; revised July 3, 1972. This paper was presented at the Electronic Components Conference, Washington, D. C., May 15-17, 1972.

The author is with Western Electric Company, Inc., Princeton, N. J. 08540.

Fig. 1. Explosive bonding, top view: glowing tungsten filament at left initiates detonation producing explosive reaction; resulting impact bonds an 0.005~in-thick copper strip to a piece of nickel.

velocity of the materials to be joined is more important than the- yield or melting points. Materials usually regarded as difficult to join, such as titanium and tantalum, are readily joined to each other or dissimilar metals.

Explosive Materials

The dynamics and thermochemistry of a large number of explosives are well understood. Their performances are con- trollable, reliable, and completely predictable [21, 131. Explosives can be grouped into two distinct types based on their reaction rates: high explosives and low explosives. The low explosives are characterized by slower rates of reaction which do not result in severe disturbances unless confined. In contrast, the high explosives yield a complex high velocity reaction with rates of 5000 to 25 000 ft/s [2]. This is defined as detonation as opposed to deflagration. Detonation occurs in a stable manner at a constant velocity characteristic of the chemical properties and density of the explosive material.

High explosives are further divided into two categories: primary and secondary. The secondary explosives (examples are TNT, dynamite, and PETN) are basically stable, requiring intense shock to detonate them. They are also characterized by high critical masses, that is, the mass necessary to support detonation. The primary or initiating explosives have slower reaction rates but are still considered high velocity detonating explosives. Due to the physical size of the work pieces involved #with microbonding, secondary explosives requiring detonators would be impractical. The unusual approach taken in this project is to use primary explosives as a direct energy source. They can easily be detonated mechanically or

28 IEEE TRANSACTIONS ON Parts, Hybrids, and Packaging, September 1972

thermally. In fact, their sensitivity warrants a certain degree of caution when handling. However, methods of handling these materials have been developed which are safe and reliable.

Controlled Deposition of Explosive Materials

To investigate the potential of explosively joining electrical interconnections, a system was devised to accurately deposit a specified quantity of explosive material ,in a designed pattern. The shape of the explosive, in all dimensions, determines the resulting pressure profile. Assuming the quantity of explosives could be controlled by a microdot application, the configura- tion of the charge could not be defined.

Primary explosive materials, as commercially available, are in a powder form and cannot be applied with accurate control. Therefore, a potentially attractive approach is a screening technique. This requires the explosive to be suspended in a screenable vehicle. The screening approach offers another benefit in that the normally sensitive primary explosives are nonsensitive in a “wet” state.

Though there is a wide variety of primary explosives, lead azide (PbNe) has been chosen because it has a relatively high detonation velocity and no lower critical mass. Other than chemical compatibility, the following considerations were of prime importance in obtaining a usable screening compound.

1) Screenability. Due to the precise quantity and configura- tion requirements, the resultant mixture must screen ac- curately.

2) Adherence to Surface, Explosive materials are screened on the pieceparts to be joined or buffer sheets. Buffer sheets are carriers which are used between the explosive material and the pieceparts to be joined.

3) Consistency of Blend. To provide a well-distributed pressure disturbance, the explosive particles must be homo- geneous through the total mass. The particles should remain in suspension until the system dries. The homogenization of the constituents can be estimated after detonation by either a buffer deformation pattern or velocity measurements.

4) Effect on Detonation Velocity. Volatility of the de- veloped mediums is less than the vehicles used commercially. Therefore, a definite dampening is realized depending on the mixture. When regulation of the detonation velocity is re- quired, the screening medium can act as a control, provided the resultant mixture is within screenable limits.

,5) Effect on Sensitivity. Methods of detonating lead azide can be mechanical or thermal. The present method utilizes resistance heating or infrared heating for initiation. Therefore, the surface composition of the screened material must react similarly to the basic explosive.

Bond Mechanism

Under certain conditions, a bond can be formed between colliding metals. The velocity at impact must be extremely high, so that the pressure exerted at the collision point exceeds the elastic yield limit of the materials being joined. One of the most common arrangements for using explosives to accelerate metal pieces together is illustrated in Fig. 2.

The extremely high acceleration of the upper plate is a direct function of the actual detonation velocity of the

I LOWER PLATE I

Fig. 2. Parallel plate arrangement for explosive bonding.

explosive. The resultant pressure experienced at the interface is very complex and a function of impinging velocity, material mass and density, and surface flow of the materials. The large transient pressures at the collision point produce severe surface strain and deformation resulting in intimate contact between the metals. Experimental evidence indicates the most favorable bond conditions are realized when “jetting” occurs between the metal plates. The jetting phenomenon, which will be discussed further, is the formation of a metallic jet between the impacting plates.

There are basically three types of metallurgical bonds that can result from high velocity collisions [61 : 1) a straight solid-phase metal-to-metal bond; 2) a molten layer between the metals; and 3) a wave or ripple pattern which is basically a solid-phase bond that may, under certain conditions, exhibit small molten zones. The ripple interface is preferred in most commercial cladding applications for physical strength. This wavy interface condition, unique to explosive joining, is a function of the materials being joined and the velocity of the impinging metals.

; Detonation Velocity Components

For practical purposes, the controlling factor, in terms of either energy or pressure applied at the point of collision, is a direct function of the detonation velocity. Though mass and material density affect the total system, parameter control can be achieved by velocity adjustments. To better understand the dynamics, many authorities chose to divide the velocity into three separate components which can be varied as a function of the geometrical arrangements (see Fig. 3). The three components described are I) detonation velocity VD, a

Fig. 3. Parallel plate velocity components.

Cranston: Explosive Bonding 29

property of the explosive material; 2) collision point velocity Vcp, the velocity with which the point of collision moves in the direction of the detonation velocity; and 3) plate velocity Vp, a component normal to the angle of inclination and a function of the ratio of the explosive mass to the plate mass and detonation velocity.

The parallel plate arrangement, as previously illustrated, is the most straightforward system and is generally used for the reported bonding experiments. This can be analyzed vec- torially where fl is the dynamic bend angle at impact (see Fig. 3). In this geometric configuration,

Therefore.

VD = v,,.

sin0 = VPor VP vcP VD

and

VP = Vcp sin 0. (I)

As previously mentioned, VP is a function of the ratio of the mass of explosive c to the mass of the plate m being accelerated ‘c/m.

Gurney obtained from the conservation of.momentum the following equation to define the plate velocity (see,Chadwick [71):

VP= VD (0.612 6)

2+$ (2)

Calculated results using the Gurney formula agree closely with data obtained experimentally with large cladding systems.

Literature searches reveal a number of geometric arrange- ments that have been employed to change velocity compo- nents in the impact region. Considering the detonation veloc- ity of the explosives used in the following experiments, the parallel standoff appears to be sufficient. Ideal standoff dis- tance or spacing at this time has not been determined. Experimentation has proven that some space is required to effect a bond, even on a microscale. A rough or knurled surface may provide adequate spacing. For instance, when bonding gold ribbon to metallized ceramic, the surface rough- ness of the ceramic appears to provide sufficient spacing. Thus far,,very flat surfaces, such as metallized glass, appear to be unbondable without spacing. Accurate space control may affect the dynamics, but sound bonds have been formed with minimal control effort.

Jetting hwface Disruption)

Two conditions are required to produce jetting: 1) the velocity ,of the induced pressure wave in the bond material (that is, the sonic velocity) must exceed the detonation velocity of the explosive; and 2) the pressure generated by the explosive ahead of the collision point must exceed a critical value. During bonding, extreme pressure exceeding the dy- namic elastic limit in both plates causes the surfaces to plastically deform and behave like nonviscous fluids for a short interval, forcing them to spurt out between the plates (see Fig. 4). This jetting phenomenon removes surface contaminants

such as oxides, nitrides, or absorbed gases, bringing the underlying metals into direct contact.

Fig. 4. High velocity jet emanating from collision point.

The configuration may vary from a fine metallic spray to a concentrated molten metal jet, depending on the hydro- dynamics, mass, and momentum of the whole system. The configuration may be effectively altered by the impinging velocities and collision angles.’

The single system that may be joined in the absence of jetting would be clean gold-to-gold because there are no interfering surface oxides. Gold-to-gold bonds have been achieved experimentally without apparent jetting; however, the bonds would likely be of superior quality if jetting did occur.

The three types of interfaces previously mentioned are a function of the jet behavior. The straight interface, direct solid-phase bond, is formed when the jet remains stable ahead of the collision point. This may be the least common case and is reportedly difficult to accomplish repeatedly. The second case is a melt zone between the parts being joined. This is the result of jet entrapment between the plates which causes a highly localized heat. These interfaces are melted and solid- ified zones. The thickness of this layer varies with both the bonding parameters and the materials joined.

The third and most complex case arises when the jet is unstable in the collision region. This results in a solid-phase bond and a wavy interface unique to explosive bonding. Sometimes a very small molten zone may appear in the wave vortices. Formation of these waves lacks a satisfactory the- oretical description. It is thought that most of the jet energy is absorbed in this wave formation as it oscillates 151. The wavelength is usually associated with the collision point velocity, while the amplitude ise function of the plate velocity components. In unions requiring high strength, there are advantages in the increased surface area at the wave interface. Experimentally it js not difficult to produce this type of interface.

The theory of the bond mechanism as discussed above has been published in [I], [3], and [4] - [7].

Buffers

It is often advantageous to use an intermediate material between the reaction and pieceparts to be joined. This practice is common in large metal joining applications and is intended to maintain the integrity of metal surfaces exposed to secon- dary explosives. Advantages that can be realized in micro- joining with primary explosives are 1) a pattern is more easily screened on a flat surface and the buffer sheet, then placed

30 IEEE TRANSACTIONS ON Parts, Hybrids, and Packaging, September 1972

over a bonding array, 2) the buffer protects surfaces from explosive By-products, 3) pressure disturbances could be dampened when required. Buffer sheets are required when the shape of the part to be joined is not flat (for example, round wire leads).

Materials used experimentally were either polyimide sheets or metal foils ranging from 2 to 10 mil thick. Of prime consideration is the material compatibility with wet lead azide. For example, copper and zinc are rapidly attacked by moist lead azides, while Monel, chrome-nickel, and lnconel are not. This would introduce a time (shelf-life) factor if lead azide were screened directly to copper surfaces. Surface hardness of the buffer, as experimentally concluded, is also a considera- tion. Softer metal materials will tend to adhere to the pieceparts, while harder surfaced materials do not. No real data have been generated related to the degree of hardness required; however, it would be a function of the size of explosive charge. Sound bonds have been formed through buffers with no welding to the buffer.

Experimental Results and Interface Studies

The metal combinations described in this section are typical of the systems studied. Many of the metals joined have widely different physical properties and would be difficult, if not impossible, to join by other techniques. Pull strength is not reported as related to the specific examples; however, the combinations cited exhibited bond strengths above that of the parent metals.

In general, the molten layers are not preferred because of the inherent weaknesses associated with cast structures [3] and possible problems resulting from brittle intermetallics.‘The three main sources of heat affecting the structure of the interface, as defined by Crossland and Williams [I], are 1) heat of explosive detonation, 2) internal heat generated in metals subjected to high strain rates, and 3) adiabatic heat of compressed gases between the plates. Reduction of heat in the impact region could be accomplished by both minimizing the magnitude of strain and in vacuum operations.

Bonded Materials

Photomicrographs of explosively bonded materials are shown in Figs. 5 through 9. An analysis of the system characteristics for the Ni-Cu bond using the Gurney formula is given in the following paragraph as an example of the

Fig. 5. Right: ye - by l-in strip of copper (0.005in-tfiick) explosively bonded to */e-in2 (0.030-in-thick) piece of nickel. Left: 250 x magnification of Cu-Ni interface.

application of the basic theoretical analysis. The values of VP and P obtained are in close agreement with those required for secondary explosive cladding of large plates. The wavy inter- face (Fig. 5) is the result of an unstable jet in the collision region. This is the preferred condition because the increased surface area produces a stronger bond.

The Gurney formula applied to Cu-Ni system to determine theoretical plate velocity and bend angle parameters is as follows:

parallel spacing 0.003 in measured detonation velocity 12 890 ft/s explosive mass c 0.01712 g accelerated plate mass m 0.02688 g applying (2) VP = 1905 ftls as defined (Fig. 3) /3 = 8-12“.

The explosive bonding technique has been applied to various metal combinations; see Figs. 6 and 7.

Fig. 6. Stainless steel (0.002.in-thick) bonded to copper (0.005~in- thick) at 100 x magnifications.

Fig. 7. Gold-plated copper (0.005in) to tantalum (0.010~in) at 100 x magnification with blowup of interface at 250 x magnification.

Figs. 8 and 9 are an illustration of the parameter control that is available through velocity adjustments. The parameters are cited in the figure captions.

A topographical view, shown in Fig. 10, clearly illustrates the pattern of wave propagation at the interface of a copper- to-nickel bond. The interface in cross section would be as shown in Fig. 5. To expose the interface, the copper was preferentially etched, leaving the total wave pattern on nickel. The point of detonation (right center) and time lag to initiation of rippling is evident and illustrated by the semicircular region.

Some valuable information can be gained from even a

Cranston: Explosive Bonding

Fig. 8. Copper (0.005~in-thick) to nickel (0.030-in-thick) at 500 x magnification. VP = 985 ftls; VD = 10 800 ft/s; mass ratio c/m = 0.3503.

Fig. 9. Copper (0.005~in-thick) to nickel (0.030 in thick) at 500 x magnification. VP, = 3037 ft/s VD = 13 000 ftls; mass ratio c/m = 1.2351.

Fig. 10 Topographical view of an explosively generated interface wave DroDaaation at 25 x maanification.

31

casual observance, and the technique will yield a better understanding of intentional directional detonation as opposed to random detonation. Edge effects are apparent as the magn/t$$e of rippling increases at the borders. This effect is related to the escapement of compressed gases between the plates.

Advantages

Summarized, the advantages are as follows. 1) Required equipment or tooling for large area or multiple

lead bonding applications is minimized. ’ 2) The highly localized energy applications do not affect surrounding areas.

3) The bonding cycle is very short. 4) A potential method for joining difficult-to-join materials

is presented. 5) The process is easily adaptable to short-run projects that

would require expensive tooling. 6) Precious metals (Au) are not required to facilitate

bonding. 7) The need for ultraclean surfaces required for most

solid-phase bonding systems is minimized.

Disadvantages

Summarized, the disadvantages are as follows. 1) Caution is required when using explosives. 2) Accurate control of explosive material application is

necessary. 3) Contamination could possibly result from by-products of

the reaction. 4) New techniques such as this will require training of the

potential users. 5) Manufacturing systems may require exhaust facilities for

removal of explosion fumes. Although this would not present a pollution or cleaning problem, the fumes should not be directly inhaled.

Hazards are minimized due to the stability of the explosive when mixed in carrier vehicles. In a wet condition, they will not detonate. The accurate screening techniques that were developed iprovide the required control. Further, the successful use of buffers prevents contamination of bond regions.

Cbnclusions

An analysis has been made of the reaction characteristics of certain primary explosives. These materials are now considered safe and predictable energy sources. Methods were devised to accurately deposit a specific quantity and configuration of explosive mixtures with predetermined reaction velocities. The degree of control has permitted a significant reduction in the state of the art as currently being practiced and presents a new solid-phase joining technique applicable to electrical type connections or microassemblies.

Acknowledgment

The author wishes to express his appreciation to the following: D. A. Machusak who has contributed through all phases of the project and in assembly of the information presented herein; J. L. Edwards whose extended efforts and

IEEE TRANSAdTlONS ON Parts, Hybrids, and Packaging, September 1972 32

enlightening discussions have led to a better understanding of the .theoretical concepts; R. L. Whalin and G. K. Grechus for supervision, direction, and support of the project; and T. G. Steele for suggestions for, screening media and advice on screening tectinology which made explosive material deposition controllable.

References

[I] B. Crossland and J. D.. Williams, “Explosive welding,” Met& Mat&., vol. 4,July 1970.

[2] J. S. Rinehart and J: Pearson, Explosive Welding of Metals. New York: ‘iMacMillan, 1963.

[3] J. Taylor, Detonation in Condensed Explosives. New York: Oxford, 1952.

[4] H. H. .Holtzman ’ and G. R. Cowan, “Bonding of metals with explosives,” Weld. Res. Count. Bull., no. 104, Apr. 1965.

[5] R. F. Tylecote, The Solid Phase Welding of Metals. New York: St. Martin’s, 1968.

[61 J. F. Kowalick and D. R. Hay, “Metallographic mea- surement of explosive welding parameters,” Explosive Welding Inst., broc. Select Conf., 1968.

[7] M. D. Chadwick, “Some aspects of explosive welding in different geometries,” Explosive Welding Inst.,’ Proc. Select Conf,, 1968.