dear Moslem Najafi

if it is possible, i suggest you to use a specific coupling agent to create a chemical bond on the interface fiber/polyester, this may avoid delamination.

Dear Moslem,

in Bonding Technology there are serveral possibilities to optimize the bonding quality.

1. Preparation of bonding surfaces: cleaning, roughening, etc.

2. As Oussama suggested: coupling agents

3. A proper choice of resin or additional adhesive...

You should contact a bonding specialist. I can recomment the company Delo in Germany. They also develop Al/GFK-Bonding systems for the automotive industry: http://www.delo.de/en/

BR, Johannes.

To a bond to be good both parts shall be effectively be clean an free from fragile layers. Aluminum tends to form Bohemita, fragile and with low resistance by oxidation, if you paint and loads are low, cleaning is enough. But if you want a good adhesive bondint, it is better to use a fpl ecth, sulfuric acid, sodium dicromate, and some copper sulfate, or even better to make a complet phosphoric anodizind with the highest possible tension and the lowest possible temperature without any sealing, the adhesive if appropriate will fill the pores and maximum interfacial resistance achieved. Epoxi adhesives and adhesives with lower surface energy are good to this. Another action is to effectively clean the polymer, and eventually raise chemically its surface energy. Abrasive plus solvent cleaning is adequate if the polymer to be applied is the base polymer of the plate, if a good evaporation of the solvent is achieved and at least the polymer absorve the sovent in its voids and dwell. Part of the solvent will be in the plate grid, part in the new layer grid and the extra polymer material will achieve the old polymer material surface. Flory-Hugges theory and Hansen solubility parameters can  help the solvent choice. If the polymer is dissolved, we have dimensional changes.

The Hysol bonding manual is a good reading to bonding preparation.

As aluminium forms a low strength oxyde layer at the surface, surface preparation is key for a good bond. Look for adequate etching agent (as per L. Bambace recommendation).

Also make sure that if the treatment you select is not stable in time, you make sure to proceed to bonding as fast as possible after the surface preparation.

Laminate preparation is also important. After manufacturing, a fraction of the mold release agent is likely to remain on the composite part surface. It is therefore important to proceed to a thorough cleaning with a solvent that is safe for the resin, but that will remove the layer of release agent. The manufacturer of the release agent you are usign might have suggestion of suitable cleaners. Another option is to place peel-ply on the surface to be bonded during manufacturing and leave it there until you are ready to bond the part. Just before bonding, you peel of the peel-ply fabric leaving a clean and textured surface and removing any surface contaminants. Just make sure you do not use a coated peel-ply or it might leave anti-adhesive residues.

Best regards.

Completing the answer, people may care if the main load is due to thermal dilatation or  other mechanical load. For instance, FR$ printed circuit boards have exactly the thermal dilatation of copper and tends to have small themal loads and the copper layer may be directly bonded to it with a strong and high Young modulus adhesive. It there's a mismatch thermal stresses may be higher of any other stresses.

If both parts are thin its is easy to calculate the thermal stresses even to big parts. Reissener equations may be adapted. In a thin bar, the compression stress of one part, the aluminum or the composite plate is the derivative of the axial displacement with the position paralell to the direction of  its axis, the second derivative the stress variation. The stress variation shall be the shear stress promoted by the adhesive, that is the difference in displacement of both parts divided by the bonding thickness, and multiplied by the shear modulus of the adhesive. This to both parts, being the shear term identical in modulus and with oposite sign to both pieces. These 2 second order differential equations may be conveter to a 4th order differential equation and solved analiticaly. The bondary conditions are zero displacement and derivative at the middle,  no normal stress at the free ends ( that is the bondary is free to dilatate), solving this the peak normal stress is in the middle and the peak adhesive stress in the end of the bond zone, where the plate mismatch is biggest. The plate mismatch goes to an asyntotic value. The adhesive or its interface with one of the substrates will crack if sooner or later if the stress is bigger than the peak adhesive stress, or its asymetric fatigue strees or its endurance stress.

An adhesive may fail in cyclic loads with time, a fatigue similar to metals, but also with a static stress high enough to promot its desorption, Endurance Limit. That is found with exponential fit between load and failure time. The Endurance LImit (EL) is not an adhesive property, but a interface property and depends on the preparation and substrate, and also bubbles that may promote stress concentration at interface.

Regarding durability, if the parts founds in its usage any substance that releases more energy if absorbed than the energy that is made free by adhesive absorption it will displace the adhesive. One usual solution is using a chlorosilane that reacts with the surface with an organic part identical to the applied adhesive. In this case there's no chance of such displacement.

Poliurethane clhorosilane primers are common, in the past we used Lord AP131 for many applications. Poliurethane are flexible and able to acomodate thermal displacements but less strong than epoxies. The nature of the load and the dimensions of the bonding zones and adhesive bond shall be carefuly chose to a good result.