Isotropic conductive adhesive joints
09/01/2001
Improving mechanical properties by adding carbon fibers
BY M. KEIL, B. BJARNASON, B. WICKSTRÖM, L. OLSSON
Epoxy-based electrically conductive adhesives represent an intrinsically clean and simple alternative to solders for today's high-density electrical interconnects. In most electronic packaging applications, however, adhesive joints exhibit weaker mechanical strength than solder joints. A simple method to overcome this disadvantage relies upon the addition of micron-sized carbon fibers to produce a more efficient distribution of mechanical forces within the joint. A mechanical shear test investigation demonstrates that a significant internal strength improvement can be achieved with the addition of low weight percentages of carbon fibers into a commercially available conductive adhesive.
Isotropic Conductive Adhesives
Epoxy-based isotropic conductive adhesives (ICAs) for component mounting in electronic assembly applications have several advantages compared to traditional soldering technologies.1 For instance, the use of ICAs instead of soldering in electronics packaging is more environmentally friendly because ICA joints are lead-free and do not require fluxes and flux cleaning. ICAs can be cured at lower temperatures than required for solder reflow so they are less destructive for thermally sensitive components. ICAs also make assemblies much lighter (the weight of an ICA joint is between 5 and 10 percent that of a solder joint). ICAs can also be applied to non-solderable substrates, and they are less sensitive to thermal fatigue and require fewer process steps in production.2 However, existing ICA materials have a lower electrical conductivity and, in most cases, poorer mechanical strengths than soldered joints.3
 Figure 1. E-pad shaped signal layers used as contact points for ICA joints. |
To analyze the mechanical strengths of adhesive joints, a number of tests can be performed, including mechanical shock, shear and bend tests, temperature shock, temperature cycling, and humidity tests.3 However, not all of these tests are well-suited to investigate the reliability of all electronic assemblies. For example, ICAs generally do not pass the American drop test standard criteria of the National Center for Manufacturing Science (NCMS),4 but they resist cracking from vibration and shock better than solders in cases of highly flexible formulation joints, such as circuits on thin plastics or glasses.5
With respect to these destructive mechanical tests, potential fracture points in ICA joints can be divided into three classes:
I. The interface between the adhesive and the printed circuit board (PCB).
II. The interface between the adhesive and the component lead.
III. The bulk material of the adhesive itself.
In the development of mechanically stronger ICA joints, all of these fracture points can be regarded separately. Fractures of the first and second types can be overcome by changes in construction. The PCB/ICA interface (class I) could be improved in several applications by the introduction of E-shaped gold pads (E-pads) instead of the traditionally shaped rectangular gold pads (Figure 1). The "E" shape allows the ICA to adhere to the board material at the E interspaces and not only to the gold layer, as in the case of the traditional pads. The adhesion between ICA and board material (most often an epoxy-based polymer) is higher than that between ICA and gold pad, which leads to a significant improvement of the internal strength on the board side instead of epoxy/gold. Further advantages of the pad structure are related to its capacity to transfer any excess amount of adhesive from the component center (Figure 1) and the ability to ensure a certain minimum thickness of the adhesive in the E interspaces to absorb forces created by the mismatch of thermal expansion coefficients between the PCB and the component leads.
At the component/ICA interface (class II) the shape of the leads can be optimized, as well, by substituting plated component leads for the commonly used J-leads. The plated shape ensures large interacting surfaces at the bottom sides, which are terminated by sharp rectangular edges giving rise to good mechanical strength. Additionally, the interfacial adhesion properties on both the board signal layer and the component lead can be improved chemically. The fracture is predominantly located on the weakest part for all classes, but it is possible for fractures to occur in other places, as well.
Addition of Carbon Fibers
To improve the internal mechanical strength within the bulk material of the ICA (class III), carbon fibers have been used as an addition to commercial ICA. Carbon fibers were chosen as a filler material for several reasons:
- The high tensile strengths (some suppliers have claimed 2.7 times higher than those of high tensile steel) together with their tube-like shapes (6 µm in diameter, 60 to 300 µm in length) give them the possibility to distribute mechanical forces over the whole joint when shear and bend forces are applied to the component leads and PCB.
- The very low specific density of 1.75 g/cm3 makes them desirable for low weight assemblies.
- Their graphite-like chemical structure shows that they are chemically inert and do not change the chemical properties of the ICA (such as cure characteristics).
- They possess a moderate electrical conductivity of 1000 (Ω*cm)-1, which can be improved by metallization of the fiber surfaces or charge doping of the fiber bulks.
- Finally, they are disposable, making them environmentally friendly.
Experimental Details
Two electrically conductive, fast-curing ICAs, labeled X and Y, were used for this investigation. Control samples with no carbon fibers, as well as samples containing fiber concentrations of 1, 2 and 3 weight percent (equal to approximately 5, 10, and 15 volume percent, respectively) were tested. The fiber lengths vary in each sample between 60 and 300 µm. The samples were cured at 150°C for 30 minutes (to ensure complete curing).
Components were mounted applying either a stencil screen-printing technique using a 0.15 mm thick stencil or dispensing of the adhesive directly onto the board. Two different mechanical test procedures were performed at least 24 hours after sample preparation: 60-inch drop tests using the NCMS test standard4 and shear tests performed with the help of a self-constructed, pneumatic-operated piston, applying a shear force to one particular component edge. PLCC44 components with Sn/Sb J-leads were mounted onto a copper-coated FR4 board for the drop tests. The shear tests were performed with PLCC44, SO8, SOT23 and 1206 components - the latter with plated lead shapes - mounted onto self-designed test boards possessing both traditional gold pads and E-pads.
Accompanying scanning electron microscopy (SEM), light microscopy and electron spectroscopy for chemical analysis (ESCA) investigations of the fracture surfaces were performed. The ESCA measurements were carried out to analyze the chemical composition of the interfaces.
60-inch Drop Tests
A total of 24 samples of PLCC44 components with J-leads, assembled onto copper-coated surfaces using screen printed adhesive, were subjected to the drop test: three samples each of the ICAs X and Y with 0, 1, 2 and 3 weight percent carbon fibers. One sample of adhesive Y with 2 percent carbon fibers passed one drop from the 60-inch height, but all others failed after a single drop. Figure 2 shows a typical light microscopic image of a component lead. It can be observed that the ICA predominantly breaks at the component lead/ICA interface, resulting in an almost uncovered component lead. However, when the same component is mounted by dispensing onto the same board, the performance of the drop tests was much more successful. Two samples of ICA Y with 2 weight percent carbon fibers passed one drop and 11 drops, respectively, and one sample of ICA Y with 1 weight percent carbon fibers was still fixed on the board after 20 drops. The fracture points are also predominantly located on the ICA/lead interface.
Shear Tests
Components mounted on traditional gold pads: A series of shear tests were performed using various components (PLCC44, SO8, SOT23), with the adhesive dispensed onto traditional gold pads of test boards. Light microscopy shows that the fracture point is predominantly located at the gold layer/ICA interface, which is demonstrated by the flat fractures at the component side (Figure 3).
It should be noted that the selected tests are very unsuitable to characterize any internal mechanical strength improvement of the ICA bulk material related to the addition of carbon fibers. This is because the positions of the breaking points are predominantly located at the ICA/component lead interface (drop test) or at the ICA/board interface (shear test). In the case of the drop test done using screen printed ICA, the unfavorable shape of J-leads when they are used in ICA joints leads to failure. When the ICA is dispensed, the J-lead can be forced deeper into the dispensed ICA, making a fracture at the ICA/component lead interface alone less likely and increasing the drop test success rate. The ICA film thicknesses were approximately 110 µm for screen printed and 300 µm meters for dispensed samples. The problem during the shear tests was that the epoxy groups do not adhere well to the gold surfaces of the signal layer on the PCB.
Components mounted on E-pads: To overcome the experimental restrictions of the shear tests done using conventional pads, the shear tests were repeated. The second shear test group used the more suitable E-pads as well as components possessing leads with plated shapes (1206). As shown in the light microscopic image of the PCB side (Figure 4), the fracture point is predominantly located within the ICA bulk material, with just a small part of the fracture area located at the gold interface. Figure 5 displays the result of the shear tests of the ICA Y with 0 and 2 weight percent carbon fibers. In each case, 28 samples were observed. The two curves represent Gaussian fits onto the results to the experimental shear tests, in which the number of experiments is plotted with respect to the shear pressure applied to the piston. The maximum of the Gauss curve for pristine ICA shifts from 3.92 bar to 5.42 bars after adding of 2 weight percent fibers. This corresponds to a major shear force improvement of 38 percent.
ESCA Measurements
The ESCA investigation demonstrates an unexpected low silver concentration within the first 20 nm on both sides of the ICA layer, on its surface and on its interface to the PCB (the results are not shown here). Surprisingly, at the ICA/board interface almost no silver was found. On the ICA surface, the silver concentration was just approximately 17 weight percent (corresponding to 2.5 volume percent), being substantially lower than the interior concentration of the bulk (80 weight percent). Consequently, these interface regions are strongly dominated by the epoxy adhesive and are characterized by an almost absence of the silver fillers. This opens a substantial opportunity to perform interfacial chemistry in order to improve the adhesion properties of the epoxy adhesive to both the board and the component lead. Additionally, it gives rise to a new discussion concerning the conductivity mechanisms through ICA joints.
Conclusion and Outlook
The addition of low weight percentages of carbon fibers can be a simple and elegant method to improve the internal strength of conducting adhesives, because carbon fibers possess abilities to distribute mechanical forces over the whole joint. Strength improvements of nearly 40 percent were observed, but only in conjunction with E-pads. This design ensures space under the component terminals because of openings in the connection pads, giving rise to disoriented fibers. In cases when the component is mounted using a very small conductive adhesive thickness, the fiber implant can also lead to a reduction instead of an improvement of the strengths. In cases when the component is mounted using a very small conductive adhesive thickness (using .1 mm stencils), the fiber implant can also lead to poor strength, rather than improved strength. Here, the fibers show enhanced tendencies to lie flat on the surface, resulting in lower capacity to distribute mechanical forces. Further optimization of the fiber lengths would lead to more stochastically distributed orientations of the fibers within the joints, which would considerably enhance the internal strength. In the future, efforts should be focused in the optimization of the relevant parameters, as well as perhaps studying electrical conductivity.
AP
M. Keil, Ph.D., research and development, B. Bjarnason, Ph.D., research and development, B. Wickström, vice president, and L. Olsson, C.T.O., can be contacted at Obducat A.B., P.O. BOX 580, 201 25 Malmö, Sweden; 46-40-36-21-00, Fax: 46-40-36-21-50; [email protected].
References
- M. Zwolinski et al, "Electrically Conductive Adhesives for Surface Mount Solder Replacement," IEEE Tran. Comp. Pack. Man. Tech, Vol. C19, 1996, p. 241.
- J. C. Jagt, "Conductive Adhesives for Electronics Packaging," Conductive Adhesives for Electronics Packaging, ed. J. Lui, Bristol, England, 1999, p. 272.
- K. Orthmann, Electrical and Mechanical Properties of Conductive Adhesive Bonds in Comparison with Soldering in PCB Technology, H. Vogel, Münich, Germany, 1991 (in German).
- S. L. McCarthy, "New Test Methods for Evaluating Electrically Conductive Adhesives" SMT, 1996, p. 19.
- D. J. Small, P. Biocca, "Lead-free Solders vs. Conductive Adhesives," Advanced Packaging, Vol. 9(9), October 2000, pp. 45-49.