ADHESIVE BONDING OF POLYMERIC MATERIALS FOR AUTOMOTIVE APPLICATIONS

Prepared for the Roceedings of the 1994 Annual Automotive Technology Development Contractors Coordination Meeting

24-27 October, 1994 Dearborn, Michigan

Charles David Warren

Raymond Gerard Boeman

Felix Leonard Paulauskas

Oak Ridge National Laboratory*

Oak Ridge National Laboratory*

Oak Ridge National Laboratory*

'The submitted manuscript has been authored by a contractor of the U.S. Government under contract DE- AC05-840R21400. Accordingly, the U.S. Government retains a non-exclusive, royalty-free license to publish or reproduce the published form of this contribution, or d o w others to do so, for U.S. Government purposes."

*Oak Ridge National Laboratory is managed by Martin Marietta Energy Systems, Inc., for the U.S. Department of Energy under contract DE-AC05-840R2 1400.

DISCLAIMER

This report was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government nor any agency thereof, nor any of their employees, make any warranty, express or implied, or assumes any legal liabili- ty or responsibility for the accuracy, completeness, or usefulness of any information, appa- ratus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessar- ily state or reflect those of the United States Government or any agency thereof.

ABSTRACT a joint effort between the ACC Joining Group and ORNL staff members.

In 1992, the Oak Ridge National Laboratory (ORNL) began a cooperative research program with the Automotive Composites Consortium (ACC) to develop technologies that would overcome obstacles to the adhesive bonding of current and future automotive materials. This effort is part of a larger Department of Energy (DOE) program to promote the use of lighter weight materials in automotive structures. By reducing the weight of current automobiles, greater fuel economy and reduced emissions can be achieved. The bonding of similar and dissimilar materials was identified as being of primary importance shce this enabling technology gives designers the freedom to choose from an expanded menu of low-mass materials for structural component weight reduction.

Early in the project's conception, five key areas were identified as being of primary importance to the automotive industry. (1)The development of industry appropriate methods for determining the mechanical, physical and chemical properties of the adherends and adhesives independent of one another. (2) The determination of accurate, highly standardized fracture test methods for quantifying, not just qualifying, an adhesive\adherend system's resistance to crack growth due to shear, peel and mixed-mode stresses. (3)Modeling of joints so that designers would be able to examine the effects of minor design changes without entering into an expanded test program. (4)Non-destructive inspection of production bonds either during the bond formation, after adhesive cure or after unit completion. (5)Mechanisms for rapidly curing adhesives in a production environment.

This paper provides an overview of the entire program outlined above. It examines in more depth the details of developing accurate, highly standardized fracture test methods, particularly for mode I fracture. It takes an in-depth look at the advantages and disadvantages of the microwave curing of adhesives for high production rate environments. Additionally, issues and resolutions are considered that are of concern when using adhesives to join structural automotive composites.

This program is completing its second year. f i e tasks under this program are being performed by industry, university and government researchers. The tasks are being managed in

BACKGROUND

Automobiles of the future will be forced to travel further on a tank of fuel while discharging lower levels of pollutants [ 11. Currently, automobiles account for just under two-thirds of the nation's petroleum usage, and about one-third of the total energy consumption of the United States. By improving automotive fuel efficiency, the United States can lessen the impact that foreign oil prices have on our economy and our lives. In addition, decreased emissions from reduced fuel consumption will provide a cleaner environment for future generations.

A significant reduction in fuel consumption by domestic automobiles can be achieved by one of three primary means: (1) improving engine and drivetrain efficiency; (2) reducing automotive component mass and thus vehicle weight; or (3) reducing the size and thus weight of an automobile. Engine efficiency improvements are being studied by a wide variety of industry and government organizations and great strides have been and are being made in this area. Vehicle down- sizing has been undertaken since the early 70s and is still occurring, however, consumers are becoming more reluctant to purchase smaller and smaller vehicles because their transportation requirements dictate the need for a family sized car. Reducing component mass and thus vehicle weight, while not sacrificing vehicle size, reducing safety or increasing vehicle cost, can be accomplished by the use of high strength steels, low mass metals and other alternate, lighter weight materials. The goal of this project is to provide one enabling technology, adhesive bonding, which will allow for the use of alternate materials, particularly fiber reinforced polymer composites.

The commercial application of composites has an extensive history in the marine, aerospace and construction industries but has evolved relatively slowly in the automotive industry during the past 20 years [2,3]. Composite use in automobiles has historically been limited to secondary structures such as appearance panels and dash boards. As the evolution of the

. automobile continues, however, fiber-reinforced polymers are being considered for weight reduction in future automotive load-bearing structures and components [4]. A critical aspect of using these materials is the manner in which they are joined. Adhesive bonding has the potential to be both an economical and a structurally sound means of joining reinforced polymers and other alternative automotive materials. It may overcome a significant obstacle to the incorporation of lighter weight materials into automobiles including not only polymer composites but also aluminum alloys and reinforced metal composites.

The aerospace and military indusmes have expended tremendous resources in developing test m e t h d and test standards for material evaluation and selection. Due to the high performance environments that structural composites were subjected to in these applications and the low factors of safety, the test methods were often highly involved (and thus expensive) and highly specific to the end use application of the material under evaluation. As a result, a survey of current aerospace and military industry standards makes it apparent that there are so many individual standards for arriving at a specific material property that it is fair to say that there are few "standardized standards.

In the early part of this century the metals industries were forced into adopting a single set of standards. This was due primarily to the limited number of steel producers and the size of their industries. When those producers decided to use a set of standards (ASTh4 standards) for reporting data, the rest of the industry had to follow suit. The structural composites industry has not developed with the same limited number of mega-producers.

High production rate indusmes, such as the automotive industry Cannot bear the mts of testing like the aerospace and defense oriented industries. They cannot afford the time required for full-scale, multi-year prototype testing of each material before making material selection and moving into production. High production rate consumer goods industries also have a greater variability in material properties from one batch of material to another or from one location in a component to another due to the increased rate of productivity and the need to use less expensive fibers and resins as their base materials. All of these factors point to the need for standards that cater to these industries and can be performed at a a t and schedule that is within acceptable limits.

After .consultation with members of the domestic automotive industry, it was determined that standardized and automated test methods need to be developed for the evaluation of coniposites joined by adhesive bonding. The single-lap shear strength values that are currently employed yield a qualitative comparison between adhesively bonded joints but do not produce specific material property values that an engineer can use in designing the structural components of 'an automobile. One charter of this research project is to develop and validate those test methods.

Emphasis is p l a d on being able to derive designer usable test data and models from industry ready standardized test methods. Since methodologies and not simply specific materials for specific applications are being developed, only a few materials were selected and the entire method and process development is being accomplished with those materials. After

- completion of this step, the processes, methods and standards developed will then be verified using other materials. The materials used for the initial phase of this program are: one urethane adhesive, one epoxy adhesive, one structural reaction injection molded (SRIM) glass-fiber reinforced urethane composite, a standard E-coated steel, and a standard aluminum alloy. The adhesives are experimental and are being developed and refmed by two industry suppliers. The composite utilizes an experimental resin developed by a supplier and the steel and aluminum are standard industry stock

PROGRAM ORGANIZATION

In 1992, The Department of Energfs Ofice of Transportation Materials completed a comprehensive program plan entitled, The Lightweight Materials CWM) Multi-Year Program Plan, for the development of technologies aimed at reducing vehicle mass [5J. The plan identifies potential applications, technology needs, and R&D priorities. The goal of the Lightweight Materials Program is to develop materials and primary processing methods for the fabrication of lighter weight components which can be incorporated into automotive systems. These technologies are intended to reduce vehicle weight, increase fuel efficiency and decrease emissions.

The Lightweight Materials p r o m is jointly managed by the Department of Energy (DOE) and the United States Automotive Materials Partnership (USAMP). The adhesive bonding project is executed by members of the Automotive Composites Consortium Joining Working Group and Oak Ridge National Laboratory. The goals that are to be accomplished in this program were developed by members of the ACC. Plans of action and the division of tasks to reach each goal were then determined in a joint effort between ACC and O W L staff members. The research to be accomplished through this project is being conducted by automotive companies, automotive suppliers, DOE national laboratories, universities and consultants. Funding is coming from both the U.S. government and from the U.S. automotive industry.

The adhesive bonding project is divided into five tasks: (1)The bulk materials task is designed to characterize the properties of the adhesives and adherends independent of one another. (2)The goal of the fracture mechanics task is to develop industry standards for determining the fracture toughness for adhesive\adherend systems. (3)The modeling\ characterization task is intended to develop finite element models for the three test geometries used in the fracture mechanics test development. (4)The goal of the non- destructive evaluation task is to develop plant scale inspection methods for assuring adhesive bonds are not only "stucK' but that they are also strong. It includes process control, laboratory scale sample evaluation and new plant scale technologies. (5)The alternate processing task is geared toward the identification and development of new technologies for reducing the cure time of adhesives.

BULK MATERIALS

The bulk materials task of this project is focused on the characterization of the properties of each of the joint constituents independent of one another or interfacial

- influences. Mechanical, chemical, physical, electrid and environmental properties are being corrsidered and tests conducted for those properties that may influence adhesive bond integrity. Tbe work in this task is being spearheaded by the ACC and the near-term goal is the development of standardized test procedures for the evaluatioa of potential adhesives and adberends independent of one another. These procedures may be adopted by the "Big Three" domestic automotive companies and wed as national industry standards. One of the long range goals of this project task is to be able to correlate these bulk properties with the properties of the resultant adhesive joints which will permit more rapid evaluation of candidate adhesives and adherends. A second, longex range goal of this task is to develop the property and performance relationships which would allow the development of computer models that will aid in the characterization and design of adhesive pints.

As previously noted by the authors [a, when the first composite samples wexe bonded using the epoxy based adbesive, the composite blistered and the adhesive "blew" out of the joint during the adbesive curing cycle yielding warped samples and joints with little adhesive on the interior. Upon heating at 150°C (the adhesive m e temperature), absorbed moisture in the composite was constrained from escaping due to micrcscopic, localized, thermal constriction of the voids and capillaries in the composite. This allowed sufficient pressure to build inside the composite to produce blistering. Similarly, the thixotropic adhesive was constraining the surface and s u b d a w ! moisture from eskaping due u, its high Viscosity. As heating progressed, the adhesive's viscosity went down and the steam pressure increased until the adbesive was literally blown Out of the pint by the escaping gas.

After an extensive series of tests, it was determined that a 48 bout, 101°C pre-drying treatment would remove more than 95% of the absorbed moisture. Twelve inch square composite plaques were then bonded using this pre-drying treatment prior to application of the adhesive. After the 45 minute, 150°C adbesive cure cycle, no bond or material problems were noted. In later efforts to reduce the drying time by boasting the drying temperature, it was determined that 125°C was the higbest temperature that the compasite could be subjected to for extended periods of time (>4 hours) witbout suffering &gradation

The evaluation of the effect of drying time, drying temperature and moisture content on the mechanical strength of adhesive joints was the final step in this evaluation. Single lap shear samples were used to obtain an idea of the relative quality of adhesive joints prepared by pre-drying at different times and temperatures. Samples were prepared by pre-drying one batch of samples at 101°C and a second batch of samples at 125°C. Drying times for each batch of material were 1, 2, 3,4,8, 16,24, 36 and 48 hours. After drying, single lap shear plates Were bonded wing the epoxy adhesive (30 mil bndLine thickness) and cured for 45 minutes at 150°C. For comparison, a third set of samples were prepared that had undergone no pre-drying. Next, the plates were sectioned into one inch wide lap shear samples which were tested in a conventional Instron using a crasshead speed of 0.05 in/min. All saniples faild by composite fiber pullout and fiber t>ar.

Interpretation of the strength data from these tests is difficult. The raw data looks like typical load vs. crosshead displacement curves from single material tensile tests but cannot be converted into a true stress - strain curve due to the specimen geometry and the resulting complex stres state present in the joint- In our analysis, we used terms similar to that for typical tensile specimen stress - strain curves but attached the adjective "apparent" in front of the noun. Due to the multiple components in this system that could contribute differing amounts to the total &formation, no attempt was made at deftniag strain values but instead the data is presented as "deformation" (i.e. crosshead displacement).

Figures 1,2 and 3 show typical load vs. displacement curves for samples dried for 3, 16 and 48 hours, respectively. From these curves, it is apparent that the slope of the "elastic" @olymm are not truly elastic) curve is approximately the same regardless of the drying treatment. Drying the composite at 101°C tends to produce a slight decrease in the apparent "yield strength" of the joints when compared to samples not dried. Increasing the temperature further to 125°C produces and even greater decrease in the apparent "yield strength".

A

0 0025 0050 0075 0.10 0125 01% ~ a d l Ctoss.iiead o~rplacsment a ma system. 8 cnchu)

Figure 1. Laad a. Displacement for Compasite~dhesive Single Lap Shear Samples he-Dried for 3 Hours.

While decxeases in apparent "yield strength" are naed with increasing drying temperatures, the opposite effect is seen on the "ultimate strength" of the samples. Drying the samples at 101°C produces an increase in "ultimate tensile strength" of the joint and b a s t i n g the drying temperature to 125°C further increases this system property. The total crosshead displacement and thus system deformation of the joint was approximately the Same between samples dried at 101°C and hose not dried. Samples dried at 125°C had a significantly increased plastic range which indicates that the composite may have teen annealed by the drying treatment.

Two types of scanning electron microscopy (SEM) were conducted on the bonded composite joints. The first t ~ > e of

0 0025 00% 0.10 o m 0155 Total Cross.Head Cisplacement CX Th? System. b (inches)

A p e 2. Load vs. Displaament for Compositewdhesive Single Lap Shear Samples he-Dried for 16 Hours.

0 Ooir 0050 0075 010 0125 0 TOW OW Head D ~ ~ p r a c e m e n i o( The System I (nchar)

0

Figure 3. h d vs. Displacement for Composite~dhesive Single Lap Shear Samples he-Dried for 48 Hours.

evaluation involved examination in the as-bonded state. Bonded sandwich specimens were cross-sectioned at 90 and 45 degrees, polished using diamond paste 2nd examined Referring to Figure 4, it was found that t h e was good penetration of the resin around the fibers in the f i h r bundles. It was also observed that the composite bad a very significant void density. The voids were not limited to regions between fibers in the fiber bundles, as is often seen, but were also in abundance at the fiber bundle\compasite resin interface and

within the bulk of the composite resin. This analysis also showed that a very uniform adhesive thickness was achieved while no voids were found along the adhesive\compmite intcrface regardless of whether the composite interface was compriscd of composite resin or glass fibers. Additionally, no voids were observed in the adhesive layer.

Figure 4. scanning electron mimgmph of a crm-seaioa of the bonded samples showing high void &nsity in the composite. The adhesive and the adhesive\composite interface are void free.

Tbe second evaluation was an SEM analysis of the as- fractured surface of the specimens. CompasiteWhesive\ composite sandwich specimens, 1/2 inch wide by 10 inches long, were fractured using the Height Tapered Double Cantilever Beam (HTDCB) m c e p t which is later addressed in this paper. The fracture mode was pure peel (mode I) with a teflon film insert in the center of the adhesive layer acting as a stress concentrator. Analysis of these specimens showed that the crack began to grow at the tip of the teflon film but quickly went into the composite. Once this bad taken place, the crack would grow through the composite and not through the adbesive\composite interface or the adhesive. This was noted even though the crack would "jump" acrms the adhesive layer to the other composite layer of the sandwich specimen.

It was observed (Figure s) that tbe crack would sanetimes grow through fiber bundles in the pre-form plane and in other cases nm around fiber bundles. The fractures left exposed fiber bundles in which the fibers were clean of any boaded resin. Additionally, evidence was Seen (Figure 6) where fibers had been stressed along their longitudinal axis. The fibers were broken but it was noted that the fiberbesin interface had debonded witbout cracking of the resin around the fiber. The conclusion of this evidence is that the weakest component of the composite and the compositebdhesive system is the fibex\resin interface. This indicates that there may be problems with the binder being used in this composite. Closer examination revealed that the fracture surface showed signs of significant micrwracldng (Figure 7). Along the fracture path, the micro-cracks coalesced forming the crack which was responsible for specimen failure. It was noted that these micro-cracks tended to originate at fiber\resin interfaces within

Figure 5. Scanning electron micrograph showing the fracture surface of the bonded composite sample. Surface is composite with the adhesive being the lighter region below the surface. (A) Note that the crack ran through the fiber bundle on the right and around the fiber bundle on the left. (B) Separation of the surfaces often occurred at the fiberkesin interface leaving the fiber "footprint" behind.

Figure 6. Scanning electron micrograph showing the fracture surface of the composite. Separation is visible at the fiber\resin interface.

Figure 7. Scanning electron micrograph showing micre cracking along the fracture surface. Micrcxracks tended to start at and flow away from the fiber\resin interface. The lighter colored region is the adhesive.

the composite which again points to that interface being the weakest component of the composite system.

SEM analysis of the bonded but unfractured samples revealed that there was a very uniform, void free interface between the adhesive and the compasite. Micrographs of the adhesive\composite interface after fracture testing showed that there was no evidence of micro-cracking or debonding along the bondline interface of the two materials.

FRACTURE MECHANICS

The overall goal of the fracture mechanics task is to develop standardized and automated test procedures for characterizing the fracture toughness of joints for adhesively- bonded automotive materials. The resultant tests will be used by automotive companies and their suppliers to generate

material property data to be incorporated into design codes that will predict the performance of banded joints. Performance refers not only to the initial strength, but also to the continued strength of joints subjected to long-term aggressive environments and variable load histories. While tests incorporating in-service environments and repeated load conditions are planned, this test metbod development is currently limited to static tests d u c t e d under standard laboratory conditions. To successfully achieve the overall goal of this task, several objectives were established (1) Resolve many tbeoretical and experimental issues dealing with specimen design, load introduction and data reduction scbemes. (2) Develop test methods that will be valid for a wide range of automotive adherend and adbesive materials using standardized geometries, sizes, fixtures and procedures. (3) Establish and incorporate the most repeatable and accurate data rduction schemes. (4) Automate the test methods using commercial products fa export to industry. (5) Publish and issue a test manual to potential suppliers.

Comprehensive f r a m e toughness characterization of a material requires determining its resistance to crack propagation for three m& of deformatim mode I (cleavage or opening), mode I1 (forward shear or sliding) and mode III ( transverse shear or edge tearing.) Additionally, the combinatioa of these modes (mixed-mode) must be considered. The following discussion will be restricted to mode I testing.

MODE I TEST DEVELOPMENT - Mode I fracture toughness is a mechanical property that defines a material's resistance to crack propa-gation for a crack acted upon by tensile forces directed normal to the crack surface. The typical test specimen for adhesively bonded joints, the uniform double cantilever beam (UDCB or DCB), is the subject of ASTM Standard Practice D3433. [7l The standard was developed for testing adhesive joints with metallic adherends, but has gained broader acceptance including the determination of the fracture toughness for laminated composites. It has been demonstrated to work quite well for aerospace-grade composites.

Of interest here however, are Wed joints in which the adherends are an automotive-grade composite. Specifically, the adherend is a SRIM glass fiber reinforced urethane composite. ?his material has a higher void content and a lower fiber volume fraction than typical aerospace-grade composites (Figure 5). Furthermore, due to the randomness of the fiber placement, the uniformity is significantly less than aerospace-grade composites resulting in random zones of high- fiber content and resin-rich pockets.

The applicability of DCB testing pracths, as typically found in the literature, was investigated with specimens made by bonding two 0.125 inch thick SRIM panels with an epoxy to form a 30 mil thick bondline. Specimens, 1.0 inches by 9.5 inches, were machined from the bonded panels after the adhesive was cured for 1 hour at 150°C. Hinges for load introduction were bonded to the sample with Hysol XEA 9359.3 structural adhesive. The speciniens were loaded in a loo0 pound electromechanical testing machine with a cross- head speed of 0.2 in./mia

Clack extension in tte adhesive was preempted by damage accumulation in tbe composite adherends resulting in one of the specimen arms failing prematurely due to bending as shown in F i v e 8. As a rcsult, fracture toughness

'

measurements were not possible, and it was determined that the standard DCB geometry was pot appropriate for these materials.

p t n specimen arm failure due to bending \

Figure 8. SRIM composite adherends failed prematurely due to excessive bending during traditional DCB tests.

In order to conduct a successful hcture toughness test for bonded joints with these SRIM composite adherends, a modified specimen is required. 'Since the adherends fail prematurely due to excessive bending, it was concluded that stiffening the adherends by bonding on "backing-beams" would be beneficial. A contoured shape developed by Mcstovoy and colleagues [S-101 was employed for the backing-beams as shown in Figure 9. Tbe Mmtovoy specimen, the height- tapered double cantilever beam (HTDCB) is also the subject of ASTM D3433. Employing backing-beams with the Mcstovoy contour has advantages for the following reasons:

metallic backing-beams

Figure 9. Backing Beam Concept using the Mmtovoy Contour.

Small Disdaccments - In many applications of the DCB, large displacements of the cantilever ends are encountered. This introduces two primary error sources that must be

accounted for in the analysis of the results. Firstly, large deflections cause an effective shortening of the cantilever. Secondly, if end blocks (rather than hinges) are used to introduce the load and if deflection is measured at the load-line then end block rotation reduces the deflectioa Correction factors can be applied to account for these effects. As a practical testing matter, the correction factors are troublesome, but correction factors can be circumvented by incorporating the backing-beam amcept. With this concept &e deflections are governed by the stiffer backing beam thereby limiting the deflectioa to acceptably-small values. In addition, since tbe backing beams provide the majority of the overall stiffness, the deflections from tests with a wide range of adherend stiffnesses will exhibit a much narrower range avoiding the need to change the test setup for the variety of different adherends of interest to the automotive industry.

such that the determination of the strain energy release rate is independent of the crack length. (It is only necessary to measure the load required to drive the crack) In tbe present case the stiffness of the backing beam is modified by the adherend, and thus crack length independence is lost. If however, the stiffness of the backing beam dominates the overall stiffness, then the toughness should become only weakly dependent on the crack length. Thus the sensitivity of the experimental results to errors in the measurement of the crack length has been minimized. This is a particularly desirable feature when the crack length varies through the width of the specimen.

Anticlastic Curvatures - It bas been reported [ 111 that thin @erpendicular to the crack surface) adherends develop anticlastic curvatures. As a result, strong width-variations of the strain energy release rate develop. By bonding the backing beams to the specimen, it is expected that the Curvature and the subsequent variation in the strain energy can be significantly diminisbed It is further believed that this would result in crack lengths that are more d o r m through the thickness.

EXPERIMENTS - Backing-beams, 0.5 inch wide by 10 inches long with a contour parameter m - 90 l/inch, were machined from 17-4PH stainless steel. SRIM panels (approximately 1/8 inch thick) were bonded with an experimental epoxy to form a 30 mil bondline with an inserted Teflon film to serve as a crack initiator. Composite\epoxy\ composite specimens, 0.5 inch by 9.5 inch, were machined from the panel after the adhesive was cured at 150°C for approximately 1 hour. The backing beams were then bonded to the joint with 3M AF-163-2 film adhesive. When cured, the specimens were loaded in an electromechanical testing machine under displacement control with a crm-head speed of 0.1 inch/min. Data acquisition equipment was used to collect and process the data in real-time.

Under these conditions this adhesive exhibited the "marrest" response indicative of rate-sensitive adhesives as shown in Figure 10. Neglecting the stiffness of the composite, the initiation toughness G , , was 5.67 in-lb/in*, whereas the average arrest toughness, GI,, was 2.69 in-lb/in2. The adhesion to the adherend is excellent (Figure 4) and that the failure generally takes place in the composite. This often exposes the glass fiber in areas of high fiber content near the adbesive\composite interface indicating that the fiberbatrix

C k k L e d M m m t - ? b e H T D C B testis designed

interface may dominate the response. Figures 5a and 5b illustrate that the failure predominately follows the fiber- bundle\matrix interface. Previously measured values of the fracture toughness of the adhesive with metallic adherends exceed those quoted above. This supports the conclusion that the adhesive is tougher that the composite. It also has been observed that in some specimens the crack location changes from be near interface in one adherend to the near interface in the other, It is hypothesized that this is because the crack follows a path that takes it to the weakest interface that is the interface with the highest local concentration of fibers near the interface. It is also quite probable that the distributions of voids, particularly near the high fiber content zones, affects the path of the crack

EXTENSION (In) O W 0 0 9 0 0 4 006 OM 0 1 0 0 1 2

1.751 I I I I 1 I

f 3 a 0 4

I so - I 2s - 1.00 -

0.75 -

0 00

I I I I I I - -sa -0 2s

- 0 6 P O O S I O I S 2 0 2 S 3 0 5 5 EXiENSION hd

Figure 10. Load deflection curve for compasitebdhesive\ composite specimen tested with backing beams. The saw-tooth nature of the curve indicates a nm-arrest behavior.

FUTURE WORK - Ln the experiments described in the previous section, the contribution of the composite adherends to the specimen compliance was neglected requiring only the load to be known to calculate the toughnesses GI, and GI,. This is an approximation. In future work the crack length and compliance as a function of the crack length will be measured and used to determine GI, and G,,. Tests will also be conducted on specimens where the adherends are an E-Coated steel. The entire test method will be automated including the crack length measurement. Then the complete process will be repeated for mode II (shear mode) and mixed-mode (opening and shear) test development. Throughout the process, analytical and numerical studies will be conducted to assess the advantages of the backing-beam concept and to define optimal configurations.

MODELING\CHARACTERIZATION

The modeling\characterization task of this project was charged with analyzing a wide variety of material properties of adhesively bonded joints and developing non-linear finite element analysis models that accurately simulate the fracture behavior of three test geometries. (1)The double cantilever beam configuration which loads the specimen in the crack opening mode (mode I). (2)The end-notched flexure test which applies a shearing load to the crack (mode II). (3)The

. mixed-mode flexure (mode I & II) test configuration which delivers a load that is a combination of both mode I and mode 11 and is the prevalent fracture niode in the "real world. (Note: In homogeneous materials cracks will often turn to become mode I.) This work is being conducted at the University of Texas.

In order to develop accurate models, the structural responses of the adhesives and adherends, independent of one another, are first being evaluated. Properties such as Young's moduli, Poisson's ratios, shear moduli, shear relaxation behavior, etc., are being determined under near static conditions. In addition, dynamic properties such as shear moduli, Poisson's ratios and tensile moduli are being determined using three different load frequencies.

Prelunimy work with the SRIM composite has shown, as expected, that the material has a higher strength in the direction perpendicular to the roll direction. Surprisingly, however, this composite has low strain-rate sensitivity and low temperature dependency in either tension or compression. This glass-urethane composite is only slightly non-linear in it's "elastic" behavior. Similar tests on cast samples of the adhesives have shown that the epoxy based adhesive is also only slightly non-linear with limited strain-rate sensitivity. Quite differently, however, the urethane based adhesive is extremely non-linear with a high degree of strain rate sensitivity.

During the next phase of this project, basic material characterization tests will be repeated for bonded composite\adhesive\composite sandwich samples. Information will be sought on the interfacial fracture properties, compliances, strain energy release rates, stress distribution, relaxation moduli and visco-elastic behavior. Characterization of the plastic zone will also be conducted including crack tip displacement measurements. The effects of anti-clastic bending and changing the initial location of the crack tip on the fracture behavior will be examined. After completion of material testing, finite element models will be constructed for each of the three test geometries and verified using different adherends and adhesives. This will include a determination of the fracture envelope for each adherend\adhesive pair.

The goal of this program is to develop an analytical method for predicting the fracture response of "real world joints. Succesr;fid completion of that goal will require interaction between the modeling\characterization, fracture mechanics and bulk materials researchers. This entire development process can then be repeated to develop an understanding of the fatigue, creep and creep fracture behavior

' of composite joints.

NON-DESTRUCTIVE EVALUATION

The non-destructive evaluation phase of this project was designed to develop plant scale inspection methods for assuring adhesive bonds are both "stuck" and strong. Many of the inspection methods currently in use can identify voids in an adhesive joint, however, they cannot identify "kissing bonds". These are bonds in which the adhesivekdhesive or adhesive\adherend interface does not leave a void but the two surfaces are not chemically bonded to one another. Similarly, joints with bonds that are weak are also very difficult to

identify. Non-destructive evaluation methods for adhesives and adhesive joints must be developed which can reliably identify voids, kissing bonds and weakened bonds.

In addition to the problems listed above, inspection techniques for the automotive industry must be able to evaluate joints that are covered by one or more assemblies that restrict the ability of quality control personnel and equipment to have access to the joint. This is a problem that is going to require the ingenuity of manufacturing specialists, the creativity of joint designers and the development of alternate evaluation techniques.

Despite the monumental size of the previously mentioned technical issues, the largest hurdles involved in non-destructive evaluation of adhesive joints for the automotive industry are the speed and affordability of any proposed technique. The f d evaluation methods must not slow down the production line and cannot add excessive COst to the f d product. Due to the manufacturing issues involved in this project task, the ACC is spearheading this phase of the project.

This evaluation task has been approached from two different angles. First is the application of process control which can be used to provide quality assurance and minimize waste and cost. The second approach is to use non-destructive evaluation to validate the proper assembly of the manufactured structure. Two methods are currently being examined: thermography and shearography. Both techniques are rapid, full-field methods. After initial evaluation on a "real" automotive structure, shearography appears to be the most suitable to large structures. It picked-up more defects and was easier to read than the results from thermography.

ALTERNATE PROCESSING

The intent of this task was to demonstrate the feasibility of curing adhesive bonds through the application of microwave radiation. The microwave processed specimens were evaluated for mechanical acceptance and subsequently compared with conventionally (thermally) cured samples. This alternate processing method was judged as successful since it yielded a substantial reduction in the requited curing time for the adhesive while producing joints with equivalent physical and mechanical characteristics when compared to thermally cured samples.

In this study, p p glass, polyethylene (LDPE), plexiglass (PMMA), polycarbonate (pc), and a composite were used as adherends. First the coupling characteristics of the neat adhesive resin, independent of any substrates, was studied. To achieve this goal, glass substrates were used to eliminate the variables represented by the absorption of microwave energy by differing substrates. In spite of the original intention to study only pure (epoxy based) adhesive, it was decided to extend the study to include adhesive additives. These enhanced the coupling efficiency to microwave radiation and improved the kinetic reaction rate during curing of the adhesive. Carbon black powder, which was evaluated in this program, is a good example of a microwave radiation coupling enhancer.

The microwave system used was a Cober SF6 power supply which provided up to 5.5 kw of 2.45 GHz radiation into the 2ft x 2ft x 2ft multimode cavity. In order to maintain

* . uiiiforni bond line diichi~ss g I a s k a d s of 30 niii diameter were embedded in the adhesive at the joint. Sukcqucndy, all samples were e x p s c d to varying p w e r ICVCIS of microwave radiation. At selc.ctc.d time intervals during the prcxessing, substrates were inspected to dctcmiine the drlgree of plynieri7~tion (crosslinkiiig). Substrate type, adhesive type, expsure time, forward power, and reflectcd power were recorded for each tr ial in tllis cxpcriniental program.

Figure 1 1 shows data obtained for the pyrex glass\epxy adhesive system without additives. Cure times ranged from less than one minute to twelve minutes. This is very attractive when conipared to a conventional adhesive cure tinie of forty- five minutes.

SUBSTRATL: Class ADHESIVE WITHOUT ADDITIVES

4l

n % - . Conventbnil ly Cured

~-6Ommln/lZl-l lPC I I I I I I *

c

-

. Conventbnil ly Cured ~-6Ommln/lZl-l lPC

I I I I I I * c

10 15 m z 30 Curing Time (mln)

Figure 11. Power Level vs. Curing Time for an Epoxy Adhesive subject to 2.45 GHz Microwave Radiation.

Figure 12. Microwave Processed Samples Showing Adhesive Bubbling from Overheating. High Microwave Energy Rate Deposition.

As expected from theory, for each kind of SubstrateQdhesive system studied, the required curinp time decreased with an increase in the level of the microwave input power, and with an increase in the concentration of the active

additive in the adhesive system that responded pcsitivcly to the niicrowavc energy dcpasition. It was observed in all of the exp.xinicntal work that an upper threshold power level (cnergy dcpi t ion rate) for curing of the adhesive existed such that any input power level above this limit resulted in bubbles k ing generated in the adhesive (Figure 12). TIUS upper liniit may be due to the presence of high local electric fields resulting in the decomposition (and eventual vaporization) of the adhesive. This limitation is found in all the systems (substratesbidhesives) studied, but the initiation p i n t of this negative affect varied from system to system. The bubbles and voids in the adhesive could not have originated from trapped moisture in the substrate material because the adherends were preconditioned through an established drying cycle. This experimental phenomena may be altered through changes in the chemical formulation of the adhesive to produce an improvement in thermal stabilization and a further reduction of the required crosslinking time.

Based 011 the current results and prior experimental studies of microwave cured polymers, any reformulation of the adhesive which reduces the conventional cure time should also result in a corresponding reduction in the microwave cure time. Our current knowledge indicates that microwave curing requires 1/3 to 1/4 of the conventional cure time. For this epoxy adhesive, crosslinking required less than 15 minutes by microwave radiation compared to 45 minutes thermally.

Joint bond strength was determined using single lap shear samples. Tbe test used is an extension for polymer composites of the an ASTM-Standard for metal sheets. A standard Instron tensile machine was used to perform the tests. It was noted that failure in satisfactorily cured samples occurred by near bondline fiber tear of the urethane composite (Figure 13).

Figure 13. Microwave Processed Samples Showing Satisfactory Fracture of the Substrate in the Near Interface Region.

Table I is a comparison of conventional and microwave cured saniples. The data demonstrates that slightly higher maximum ultinlate tensile strengths (aJ were consistently obtained for the microwave processed samples when compared

with the thermally cured samples. This was observed with and without additives to the adhesive. In all cases, the range of uB was within experimental error. The total craishead displacement at uB (TCHD) for the conventionally processed samples showed a significantly lower value than those obtained for microwave processed saniples. The greatest difference in TCHD occurred when comparing conventionally cured samples and microwave processed samples containing carbon as an additive to the adhesive. This indicates that the conventionally processed samples show a greater stiffness and lower ductility.

Conventionally Microwave cured h s s e d 45 min w/o 0.1 W% c 1.0 W% c

0 IWC Additive Additive Additive ( M R C ) (Met) (rana e) ( M R C )

Max Stress 2736 .2845 2802 2755 0 , @si) (2560-2840) (2600-3100) (2600-3100) (2600-2930)

TCHD 0 a, C h ) 0.0546 0.0652 0.0790 0.0819

TCHD 8 1000 Ib load (inch) 0.0333 0.0368 0.0457 0.0495

Table I. Comparison of Single Lap Shear Test Results: Conventionally Cured Samples vs. Microwave Processed Samples with and without additives.

Figure 14. Single Lap Shear Test Comparison of Microwave and Conventionally Processed Samples.

Figure 14 represents dam obtained in single lzpshear tcsts for conventionally cured saniplcs and nucrowave processed saniples cured at short processing times (10 to 12.5 niinutes). The data represents Inqd and not stress. It is clear that conventionally cured samples demonstrated a slightly higher rigidity than the microwave processed saniples. In ths type of

test, the contribution of each systeni component and its processing history, to the fracture behavior of the system can not be differentiated accurately. It was noted that the higher the concentration of carbon in the adhesive, the larger is the deforniation that the system can tolerate and that the carbon content affects the tolerable state of deformation.

In conclusion, the application of microwave technology for the joining of substrates using epoxy based adhesives can significantly reduce the required curing tinie to 1/3 of the conventional cure time. This is accomplished while maintaining equal or slightly higher ultimate tensile strengths as measured by the single lapshear test. The coupling characteristic (microwave energy deposition) of the adhesive is enhanced by inclusion of an additive such as carbon black Within the scope of this work, it appears that a recognizible threshold in the rate of energy deposition (input power level) into the adhesive exists above which the generation of bubbles and expelling of adhesive from the joint results. This consequently generates a lower cure h i e limit and an upper input power threshold.

ACKNOWLEDGEMENTS

The authors wish to thank the following individuals for their diligent effort in assisting with this project: Fahniy Haggag and Ronny Lomax of Oak Ridge National Laboratory; Professor 'Ihomas Meek, University of Te.nnessm and the entire crew of the ACCs Joining Group.

This project is sponsored by the U. S . Department of Energy, Office of Transportation Materials, Lightweight Materials Project. Oak Ridge National Laboratory is operated by Martin Marietta Energy Systems hc. under contract DE- AC05-8mR2 1 4 0 .

REFERENCES

1. Baxter, Donald F., "Plastics Beat the Heat in Underhood Components", Advanced Materials & Processes, May (1990) pp. 36-41.

2. Reindl, John C. "Comniercial and Automotive Applications." pp. 832-835.

3. Beardmore, P., "Automotive Components: Fabrication", pp. 24-31.

4. McConnelI, Vicki P., "In the Fast Track Composites in Race Cars", Advanced C o m m i t s , March/April (1991) pp. 23-35.

5. Office of Transportation Materials, "Materials for Lightweight Vehicles Program Plan" (Draft), United States Departnient of Energy, July (1992).

6. Warren, C. D., E h m a n , R. G . and Paulauskas, F. L., "Adhesive Bonding of Polymeric Materials for Automotive Applications", Proceedings of the Annual Ailtonlot ive Technology Dew lopnie nt Contractors' Conrdirution hleeting, Held 18-21 Octolxr 1993, Dearborn, Michigan. pp. 1-11.

7. "St'anhd Practice for Fracture Strength in Clcavage of Adhesives in Bonded Joints," D 3433-75 (Reapproved 198.5), Annud Bonk of ASTM Stnnthrds, American Swiety for Testing and Matcrials, Philadelphia.

~ - *

' 8. Mostovoy, S., E. J. Ripling, and C. F. Bench, "Fracture Toughness of Adhesive Joints," Journal of Adhesion,

9. Riphg, E. J., S. Mostovoy, and H. T. Corten, "Fracture Mechanics: A tool for Evaluating Structural Adhesives," Journal of Adhesion, Vol. 3, 1971, pp. 107-123.

10. Mostovoy, S., P. B. Crosley, and E. J. Ripling, "Use of Crack-Line-Loaded Specimens for Measuring Plain- Strain Fracture Toughness," Journal of Materials, Vol. 2, No. 3, September 1967, pp. 661-681.

"Strain Energy Release Rate Distributions for Double Cantilever Beam Specimens," AIAA Journal, October 1991, pp. 16861691.

VO~. 3, 1971, pp. 125-144.

11. Crews, J. H., Jr., K. N. Shivakumar, and I. S. Raju,