Dissimilar materials, thermal expansion, and cracking

[PearlRock]

Hi everybody,

I am new at this, and I'm not sure where to find documentation on it, so I figured I'd ask the community.

I have 2 materials (epoxy and steel) undergoing thermal cycling. I'm getting cracking between the two. Is there a rule of thumb to determine conditions when cracking won't likely occur? I am considering putting a 3rd piece of material between the steel and the epoxy with a thermal expansion coefficient bigger than steal, but smaller than the epoxy. I'd still have to use a bonding agent between the steel and the new material though, so this gets a little tricky.

Other notes:
If possible, I'd prefer not to blindly trial and error this. Even minimum quantity epoxy costs a fortune.
I might be able to get a different epoxy with different thermal expansion coefficient, but I don't believe it will be drastically different enough to solve my problems.
Reference material may be more useful to me than anything.

Thanks for your help!

 

[metengr]

In your example above, cracking is due to the inability of one of the materials being joined to accomodate thermal strain. Material can either absorb strain and deform (if it can permanently deform) or it will crack. You need to either select a material that has a similar linear coefficient of expansion with the mating material or select a material that can accomodate greater strain. You can attempt to sandwich this joint but this can be tricky because more interfaces can create bonding issues.

Are you sure the above is not a delamination problem caused by poor surface preparation?

 

[PearlRock]

I won't be able to mate materials that have precisely equivalent expansion coefficients. Is there a rule of thumb for when the linear expansion coefficient will be close enough? I guess not, because that would depend on the quality of the bond and surface prep as you have suggested. Assuming a perfect surface preparation, could I make any assumptions? What material information would I need to know? Allowable strain? There has to be some tools out there that help find solutions to these problems.

Good question. I did use cleaner on the steel, but I didn't rough the surface or anything. I'll have to ask what prep they recommend for steel/aluminum. Regardless, I know I am going to have cracking in some areas. I need some way to prove this isn't likely to happen in one critical area of the prototype. If I can ensure there won't be cracking there, I may be a made man.

Man this is a really hard problem

 

[rb1957]

i think (possibly wrongly) that the steel is trying to expand more than the epoxy, these there is alot of shear stress at the interface, more than the bond allows (hence the cracking).

i guess you know the thermal expansion co-efficients for both the steel and the epoxy, so you know the difference in the expansion. over the width of the bond, the steel is trying to stretch the epoxy. this'll break either the epoxy (in tension) or the bond (in shear).

 

[desertfox]

Hi

Have a look at this link:-

http://www.scribd.com/doc/43333468/Thermal-Stress

 

[PearlRock]

I thought that to at first. MATWEB says 1018 has CTE of 12e-6 in/in-C. Epoxy datasheet says the expansion coefficient is about 31.


I've filled a steel cup with epoxy. So I believe both materials experience thermal expansion over the same length. Would that not suggest that they epoxy expanded more because the CTE is larger? Maybe I'm thinking about this incorrectly. I would hope the epoxy datasheet is correct, but who knows.

 

[PearlRock]

Hmmmm,


according to that document, I need to determine the max allowable strain. The manufacturer may have that, or I could probably determine it on my own. As long as I use the same prep techniques, and as long as I ensure that thermal strain + elastic strain is less than the allowable strain, I should be in good shape.

Thanks Desertfox. That makes perfect sense. Thanks for the document. Very helpful!

 

[MintJulep]

If this is a critical application then you should use engineering to solve your problem. Not rules of thumb and assumptions.

Is the failure a crack of the material, or a bond failure?

Thermal cycling --> not everything is at the same temperature at the same time --> expansion is not uniform.

Geometry is important, not only material properties.

Stress > strength --> cracks or bond failure

 

[PearlRock]

Hi MintJulep,


This is a bond failure. I'm doing some R&D on potting of electrical stator cores, so no, the expansion will not be uniform, but full thermal simulation with thermal expansion integration is a bit beyond the project scope at the moment. I may try to break the material later. I'll have to see what the bosses say.

What do you mean by geometry importance? I have my own ideas on the importance of geometry, but I'd like to hear what you have to say. Perhaps I can get a new perspective on my design. My geometry is very complex, but there's really only 1 area of critical bond importance which will make or break this research effort. Perhaps I could make some geometry changes to that region and make my life easier.

 

[MiketheEngineer]

Different glue??

 

[rb1957]

did i read right ... that you filled some bores with epoxy and presumably cooled them down, and then they disbonded from steel body ? so like some pucks of epoxy fell out of the plate ??

is that really what you'd do in your final design ? i'd've thought that most of the bore would be filled with some metal and the epoxy would be only a thin-ish band. but then i don't know much about "mounting electrical stator cores".

is the epoxy intended as an electrically inert isolator ?

i the epoxy also doing a structural job ? (ie are you asking alot of the epoxy ?)

 

[PearlRock]

Well, it's not as dramatic as that. A crack formed around the edge is all. Yeah, in most places there will be thinner sections of epoxy, but that's not the part I'm worried about at this time. This test was performed to see if any degradation would happen to the epoxy itself, and so far no significant change. I wasn't sure if debonding around the edges would be a big issue or not, but apparently it is.


I prefer not to comment on those specifics of what we're doing, but I appreciate your willingness to help.

 

[SnTMan]

Difficult to comment on specifics of a solution then.

Regards,

Mike

 

[PearlRock]

That's alright. I think I have the information I need. Now I just need to look at various materials, surface prep and possibly modified geometry.

 

[tomwalz]

We build brazed carbide tools where the difference in CTE is somewhere around 3 to 1 or 4 to 1.

If I understand your situation correctly, you are getting cracks on the edges of the part where the epoxy is separating from the steel.

It really sounds to me like you have an issue with surface prep.

When I go to bond anything, whether with epoxy, braze alloy or anything else I assume that the surface of the material has a skin of some sort. This can be an oxidized layer in titanium or stainless steel, free carbon or oxides on carbide and always incidental shop oils and greases.

Steel is often treated so that it will not rust. We see this with steel tool bodiess all the time. They are sprayed with some sort of a rust protecting that really screws up any sort of a bonding unless you remove it. I do not like solvents for this application. Instead I greatly prefer something that actually saponifies the protectant such as a strong caustic solution.

You might spray your steel part with oven cleaner and see if that helps with the bonding.

A rough and surface will give you greater surface area and thus increase the bonding strength of the total bond.

When brazing carbide to steel, if the parts are over 1 inch in any dimension, we use a tri-metal brazing sandwich which is braze alloy - copper - braze alloy. The copper in Niels to a dead soft condition in helps compensate for the differences in thermal expansion.

Were I you, I would see if I could find a good 3M rep and ask him to come out.


Thomas J. Walz
Carbide Processors, Inc.

http://www.carbideprocessors.com

Good engineering starts with a Grainger Catalog.

 

[Tmoose]

Until the epoxy rep arrives I'd compare my design and mfg/assembly process to these -

Fairly typical surface prep recommendations

http://www.devcon.com/UserFiles/File/Surface_Prep.pdf

tensile strength vs bond line thickness

http://www.specialchem4adhesives.com/home/editorial.aspx?id=1029

 

[tomwalz]

Looks pretty good.,

For oil, grease etc. I like a strong caustic that saponifies rather than a solvent.



Thomas J. Walz
Carbide Processors, Inc.

http://www.carbideprocessors.com

Good engineering starts with a Grainger Catalog.

 

[tomwalz]

Not sure about the scraping step. Depending on the porosity of the underlying material there may be residue left that will not be removed by the following steps.

How thick is your bond line?

Thomas J. Walz
Carbide Processors, Inc.

http://www.carbideprocessors.com

Good engineering starts with a Grainger Catalog.

 

[PearlRock]

Thanks for all the resources everyone, very helpful.

Tomwalz, what exactly do you mean by bond line? Thickness of the epoxy section or bond area?


Just for an update, I am remaking my samples today. I should have results sometime next week. The plan is to try a roughed bond surface. If that's not fruitful, then I'll look into more drastic changes.