Carbon, Aluminium and Epoxy

[Chapo]

G'day guys,
I am investigating the use of a Carbon-Al suspension system for our car next year and I was wondering if people who already have them could share some hints and tips, advice, opinions etc. I have had a look through the forums already (that magic little button with the magnifying glass next to it) but was only able to find marginal information that came up as an off shoot to other threads.

So here is what I have so far:
-Carbon fibre to begin testing on ID= 14mm, Wall = 1.8mm (If you are interested it is made by a group in Sydney www.cstcomposites.com (http://www.cstcomposites.com) about $80AUD per metre, so with current exchange rates that's $50USD a bargain IMHO).
-Aluminium 7075-T651 got some nice little bearing housings and things drawn up they all look good.
-Epoxy, Devcon HP250 its a structural adhesive that is meant to be able to bond all the different matricis you would find with the composites and also the metals.
-Plan (The most important thing http://fsae.com/groupee_common/emoticons/icon_biggrin.gif) a variety of different tests including static load, until break, fatigue tests and then do the same for damaged bits (rock chips, high speed impacts with cones generally mistreated).

To me this all sounded good, until I came across one fatal flaw, during our first initial load test the bond decided to break at a measly 1600N (160Kgf, 352lbf) when it was calculated to break at around 17000N (3800lbf), a whole order of magnitude out. We talked to our resident composite expert about how to do the bonds and followed his instructions, degrease, scotchbrite, degrease, sandblast, degrease then glue before the Al oxide forms again and sand down the carbon tubing (on the inside) and blow out with compressed air (breathing masks were worn). We also ensured cleanliness of all surfaces, wore gloves the whole time and all that jazz.

So if anyone has any kind of information on how they go about ensuring the quality of the bonds that you use, and also what bonding agents you use that you wouldn't mind sharing with everyone it would be great if you could contribute. Or alternately if you just want to tear holes in the way I am going about all this I'm more than willing to at least listen.

Yours in frustration,

[Mark_W]

What did you use to degrease your parts?

[Chapo]

To degrease the parts I started with Wurth brake cleaner, then did the scotchbrite to remove the oxide, then alcohol (the pure stuff you use to clean for strain gauges), do this a few times even after it comes clean, and then sand blast and then bit more alcohol and then distilled water.

Thanks for the input

[Mark_W]

I believe denatured alcohol is the best stuff. some cleaners can leave a residue but it sounds like what you're doing is ok.

Can i suggest you omit the sand blasting? it will close up all the pores of the 7075. When you anodise stuff they prefer machined surfaces for the same reason. Don't worry too much about oxide, it forms too quickly and shouldn't affect the bond strength too much.

What bond gap are you using? Have you checked the manufacturers recommendation?

You may be able to see on your broken specimen where the glue has bonded and what has failed. Is there glue left on the 7075 piece? If i had to put money on it, i'd say there isn't.

Try this also,
http://www.carbonfibertubeshop.com/cut%20&%20bond.html

[murpia]

There are epoxy primers available that you paint on the aluminium parts to ensure a good adhesive bond and eliminate corrosion risks in service.

Unfortunately I don't know who sells them or what type to specify. I do know that the guys who re-build F1 tubs to fit new (different) engines into them from original, always specify it for their new aluminium inserts and would never attempt such a job without it.

Regards, Ian

[exFSAE]

Don't know who makes em.. but I know there are chemical "kits" for aluminum prep. Degrease and etch or whatever.

This is far from my area of expertise, but I'd think that would be the way to go. I'd be worried with sanding or scotchbright that you might leave some particles behind on the surface, and then your epoxy is bonding to that and not the actual base metal.

Clean surfaces on both the carbon and aluminum are a big deal is all I know. We'd have order-of-magnitude swings like that on failure strength based on how clean the ends were.

[NRBaer]

As exFSAE mentioned, surface prep is one of the most important factors that determines the final strength of your bonds. At Lehigh we use some acid that I am not to familiar with that etches the surface of the aluminum. I would also agree with exFSAE that scotchbrite would leave particles on the surface that can be the cause.

The people who make Loctite also make some two part adhesives that work very well.

I do have one question for you about the test. Did the bond shear and the bung pulled out or did the tube break?

[Ashley Denmead]

If after you've tried the above surface prep suggestions and decide the adhesive itself is the issue you may like to try a methacrylate adhesive such as plexus MA310 (its a more ductile bond from experience rather than a brittle epoxy).....i know it's saved many an fsae bond issue....just ask Deakin, Auckland and Monash....we always carry a spare tube http://fsae.com/groupee_common/emoticons/icon_wink.gif

[RiNaZ]

matthew, what grit sandpaper are you using? And try replacing denatured alcohol with acetone in all your steps. It could also be that you're not cleaning it well enough. Wipe the area till you cant see anything on your rags.

Try doing a test coupon with all your adhesives. I personally feel that Hysol adhesives are much stronger than methacrylate, but also more expensive.

I wonder if the gap between the insert and the tubing isnt big enough for you to have adhesive in there. You could probably be pushing the adhesive away from the bond area when you put the insert inside the tubing.

[Chapo]

G'day again guys,
Thanks for all the help so far.
We are using a bond gap of 0.25mm the recomended gap is 0.254mm (0.010") so we are pretty confident that this should not be the issue. There is glue on both parts in the break and it was the actual glue that failed not the inside of the carbon tube.
We used 320 grit sandpaper. Correct me if I'm wrong here, but I was under the assumption that sndblasting actually opened the grains up by knocking smaller grains out and creating a rougher surface.
Where do the aussie guys get the Plexus glue from? i had a quick search around the web and couldnt find a distributer. This was pretty much the deciding factor in what our glue choice was we had a short list of what we would like and then chose what ever we could get our hands on (Thanks amazon http://fsae.com/groupee_common/emoticons/icon_smile.gif)

Thanks for everything guys,

[Rickertsen2]

We have had good luck with 3M DP series of adhesives. you will find that surface prep, texture and gap make a difference in pull out strength. With some experimentation you should be able to find a combination where your carbon fails before your glue.

[Mark_W]

you could be right about the sandblasting, im only going off what our anodisers told us. anodiser works better on machined surface.

[murpia]

Originally posted by Chapo:
We are using a bond gap of 0.25mm the recomended gap is 0.254mm (0.010") so we are pretty confident that this should not be the issue.
How do you maintain the gap? Wires? Micro-beads? Locating features on your parts? The bonding fixture?

Regards, Ian

[D J Yates]

Originally posted by Chapo:
There is glue on both parts in the break and it was the actual glue that failed not the inside of the carbon tube.

Assuming that you are using the adhesive within its recomended opperating conditions (temperature, moisture, chemical enviroment) the adhesive should have shear strength close to what's on the data sheet. If the joint failed unexpectedly then it is probably due to joint design and analysis method.

Obviously, since you're bonding an insert into a tube you shouldn't be seing any apreciable peel (adhesives have low direct strength compared to shear) so this should not be the cause of the failure either.

I would expect that your failure is due to the non-uniform strain along the length of the joint. You have to consider how the tube and the insert strain along the length of the joint when loaded and what that means in terms of shear strain for the adhesive:

joint under no force

######## insert
######## tube

joint with insert much stiffer than tube under tension

<<########
__#-#--#---#----#-----#------#-------#>>

joint with both insert and tube stiffness decreasing towards unloaded ends, under tension

<<#--#--#--#--#--#--#--#
__#--#--#--#--#--#--#--#>>

Clearly, the joint with a stiff insert sees huge strain towards the right hand side of the joint, whereas the joint with variable stiffness see uniform strain.

If your joint is long and narrow and has constant section and the insert is significantly stiffer than the tube, you'll be seing a massive stress concentration at the end of the joint, quite easily a factor of 10. In general short fat joints are better; those with similar tube and insert stiffness are better still; and if you can reduce the stiffness of the insert as it goes into the tube, you'll be getting towards perfection.

Another thing that can help is if the gap increases with joint length. Can't remember why exactly, but it's all linked to the same theory as above. If i recall correctly i read about this in a fairly standard library book so you shouldn't have to much trouble finding something similar, which is good because i can't remember who wrote it!

BTW, you can apply much of the same logic to the design of threaded (and probably welded) joints. But do your own research because its been a while since i used any of this and i can't be sure i'm 100% right. Hopefully it'll point you in the right direction though.

[RiNaZ]

Matthew, try replacing your #320 with anything lower than #80. I do my polishing with #320 and above.

[Chapo]

G'day again,
I am in the best mood today, we did some more testing on a new peice and it lasted up until 6.25KN (3-4 times as much as the last one) it just makes me happy. The only difference between this peice and the other one was the cure method, 3 days longer and in a naturally hot (25-30 degrees C) room.
So now we have whipped up a few new bits to test follwing some of the advice from people here.
One major things i have noticed about the broken joints is that on some parts of the carbon the glue has not stuck at all, and it is also shiny... can anyone say i didnt get rid of the release agent well enough...
This time we have changed the sand paper to a 60 grit and degreased with acetone, we have also drilled a hole axially through the end plugs to ensure there is no pressure build up due to the changing volume as the plug goes in, or due to any gasses released by the glue.

Ian, we locate our parts through features on the parts and then hold them in V blocks during curing.

Looking foward to the next lot of testing,

[Ppada]

What is the exact curing process that your glue needs??

[Chapo]

All the data about the glue is here,

http://www.chembar.com/~chembar/images/stories/datashee...con/devcon_hp250.pdf (http://www.chembar.com/%7Echembar/images/stories/datasheets/devcon/devcon_hp250.pdf)

Thanks

[RiNaZ]

matthew, i rather go lower than 60 ... maybe 36. And then, make sure you clean it up really good. Keep cleaning until you cant see any dirt on your rag.

If it's a 2 part system, make sure you mix it really good. And when you think it is good, mix for another 2-3 mins. If you dont feel burn in your arms, or have streaks on your mix, then you're not mixing it well enough.

[D J Yates]

With reference to my previous post, the analysis method i was trying to remember was Volkersen's method of single lap joints (tube insert joint is equivelent to a single lap joint) and can be found in Kinloch, "Adhesion and Adhesives" (http://books.google.co.uk/books?id=_QoFkInPQT4C&dq=Adhesion+and+Adhesives+Kinloch&pg=PP1&ots=sjy-N-RsML&source=bn&sig=Mb_-bn1Zq96ccapMMLs8SA0tiPc&hl=en&sa=X&oi=book_result&resnum=4&ct=result#PPP1,M1)

[damonlemmon]

Everyone mentioned good points with regard to the prep of the joint as this is a critical step. Just a few things to add from our teams experience with carbon tube/aluminum bearing housings suspension arms:

1. prep is key, make sure and sand the carbon down to the fibers removing the matrix (however try to not break too many fibers, but this is unavoidable) this provides a good mechanical interface for the adhesive.

2. we used hysol ea9394 2 part epoxy adhesive by loctite, which is one common adhesive in bonding aerospace composites, but a little pricey at $170usd for a quart ( hysol ea9394 (http://www.loctiteaero.com/Images/Datasheet_PDF/Hysol%20EA%209394.pdf) ). for reference we achieved tensile loads of ~27000N (6000lbf) before failure of the joint.

3. bondline thickness critical and there is a lot of literature on it, so go with manufaturers specs, using beads or wire as previously mentioned to set the gap. you can also machine a shoulder into the aluminum insert to help set the gap.

4. phosphoric acid anodizing is the ideal way to prep the aluminum to be bonded (even though we didnt do this due to resources). if the bond gets exposed to moisture it can react with the Al to form an aluminum hydroxide (think this is the compound but dont have my notes handy) which can reduce bond strength by degrading the mechanical interface in the jont. PAA prevents the hydroxide from forming in the bond.

5. when we initially bonded the tubes we also had a much lower failure load than expected which we found was caused by an air bubble formed while inserting the tube adapters into the carbon tube. this pushed out some of the adhesive and after testing we dissected the joint and found that there were large gaps between the Al and carbon tube with no adhesive. our initial design had 2 shoulders on the insert to set the glue gap and this was trapping the air, we tuned one shoulder off the insert on the lathe and tried again and it fixed the problem.

6. with regards to the volkersen model you will find that there is a bathtub shaped stress profile in the joint (symmetric) with the greatest stresses at the ends going roughly to zero in the middle (this is theory, not actual due to imperfections in the bond) so at some point additional length of the joint nets no gains in reducing the stress in the joint. we did some testing as well to find an optimal length for our joint (~1 in.). glue (or spew) fillets also help to lower the stresses at the ends of the joint.

just a few things from our experiences that might be of help.

[CappyUMD]

I agree with much of what has been recommended:

Use 60 grit maximum sandpaper. For adhesive bonding, the objective is to create surface roughness on the order of knurling, not to polish. Don't worry too much about aluminum oxide. As long as all solvent has evaporated and you bond within hours of sanding you should be ok.

Use acetone to degrease aluminum, it contains no water and evaporates fast.

Do not use compressed air from shop lines to remove dust (especially without degreasing afterwards). It contains water from condensation and oil leaks from the pumps. Use a clean rag instead.

Measure resin and hardener precisely and mix thoroughly (reactant ratio and homogeneity are crucial for a complete reaction of epoxy). A post cure (around 120-150 F for hours) may improve the bond of room-temperature cure epoxy. Low viscosity epoxy will not trap air bubbles as much.

Galvanic corrosion of aluminum on carbon is a legitamate concern, at least for aircraft that see decades of service with lots of condensation. As mentioned, PAA (phosphoric acid anodizing) will greatly reduce corrosion of aluminum. I'll let you decide if it's worth the trouble. An alternative is to separate carbon from aluminum with a ply of kevlar or fiblerglass (electrical insulators).

[Markiempje]

comment to Damon,

we also use the 2 shoulder tipe of aluminium inserts. however we have a very simple solution to the filling problem. we just drill 2 small holes in the carbon tube in opposite side and insert the glue on one of the holes. apply preasure and when the glue exits the other hole you know that it is filled properly. we get results of 18.000 N approx. we only sand the carbon and the aluminium inserts we sand and spray with brake cleaner. also we test every rod on a test bench on a lower value before we put it on the car. only a few dont succeed the test. while driving we never had problems with the rods breaking or the insert/carbon failing in any way. we use 16mm od and 13mm id. our costs are about 10 euro per rod on approx 40cm.

good luck on your new car this year

Mark

[vandit]

I am new to this Field of Carbon and Aluminium bonding.
After reading thorugh the discussion i have a question.

What do you do to bond a Rod End and Carbon Fibre? . Do you use a Aluminum insert(having female threads) between Carbon Fibre pushrod/tie rod and Rod End.

[damonlemmon]

We used aluminum lugs that were threaded and bonded into the carbon tube. Here are some shots of what we did:

Rod end lugs back from anodizing

http://i180.photobucket.com/albums/x217/DLEMMON/rod_end_lugs.jpg

Outboard spherical bearing housings

http://i180.photobucket.com/albums/x217/DLEMMON/outboards.jpg

A arms waiting for spherical bearings to be pressed in

http://i180.photobucket.com/albums/x217/DLEMMON/a_arms.jpg

[vandit]

@ Damonlemon

that is like one of the best ever replies i have ever got on this forum ....

[Mark_W]

are they full kevlar?

[CappyUMD]

With the A-arm tubes being placed in tension/compression/bending but not torsion, this would be a great application for unidirectional fiber. The fiber orientation should be parallel to the tube axis.

[jrickert]

How are the "stalks" on the outboard end of the control arms fastened. Did you somehow make that a single piece?

[damonlemmon]

Thanks for the compliment.

Mark_W: They arent full kevlar, just 1 layer of braided kevlar sock laminated around the tube to give a bit of impact resistance from pebbles and such.

CappyUMD: We bought some unidirectional carbon tubes prefabed. A majority of the fibers are uni while there are some wrapped circumferential as well. Sanded them down and laminated the kevlar on. The loading in the a arms is intended to be only tension and compression with no bending moment due to the design and the use of spherical bearings, the axis of the tubes intersect outboard at the sphericals, think pinned truss. But there is probably a bit of bending load in the members due to fabrication being imperfect. We took this into consideration in the design though with emphasis being at the bonded joints.

jrickert: Just started with a piece of billet and whittled it down with a paring knife. Thought about different design to attach stalks separately but in the end I just decided machined it in our shops CNC (not that this was the most efficient way). Attached some pics for clarity.

http://i180.photobucket.com/albums/x217/DLEMMON/first.jpg
http://i180.photobucket.com/albums/x217/DLEMMON/nubs.jpg
http://i180.photobucket.com/albums/x217/DLEMMON/back.jpg

[Gabe]

I'm sorry if this is a newb question, but in this discussion and through my research I understand the importance of the bond gap/thickness. Now when you say use manufacturers specs, did you ask Loctite? I can't seem to find any specs on the datasheets Loctite provides.

Thanks for the help,
Gabe

[Rellis]

min is .002 max can be up to like .030 but I would aim for .005-.010

[Peter7307]

Where do the aussie guys get the Plexus glue from? i had a quick search around the web and couldnt find a distributer. This was pretty much the deciding factor in what our glue choice was we had a short list of what we would like and then chose what ever we could get our hands on.

A quick Google search brought up:
www.fgi.com.au (http://www.fgi.com.au) as the distributor for Plexus products.

Cheers , Pete.

[rollcentre]

Damon-

Are you willing to share what one of those control arms weighs? (with or without the rod ends/sphericals)

[damonlemmon]

Originally posted by rollcentre:
Damon-

Are you willing to share what one of those control arms weighs? (with or without the rod ends/sphericals)

they averaged around .45 lbs a piece without bearings. dont have the numbers of each arm, but our previous steel ones weighed about 1.25lbs so the weight savings was worth the extra work. just have to get in the habit of not setting our helmet on them after a run.

[PBR-Simon]

Originally posted by D J Yates:
With reference to my previous post, the analysis method i was trying to remember was Volkersen's method of single lap joints (tube insert joint is equivelent to a single lap joint) and can be found in Kinloch, "Adhesion and Adhesives" (http://books.google.co.uk/books?id=_QoFkInPQT4C&dq=Adhesion+and+Adhesives+Kinloch&pg=PP1&ots=sjy-N-RsML&source=bn&sig=Mb_-bn1Zq96ccapMMLs8SA0tiPc&hl=en&sa=X&oi=book_result&resnum=4&ct=result#PPP1,M1)

I found this PDF which does a much better job of explaining Volkersen's method:

http://www.virginiadot.org/bus...olymersAdhesives.pdf (http://www.virginiadot.org/business/resources/bridge-02PolymersAdhesives.pdf)

also, the properties of Hysol 9430:

http://www.gluguru.com/Hysol%20Data%20Sheets/9430.htm

[Brian Smith]

Thanks for that link Simon. Quite helpful.

[J-rob68]

Damon,

Thank you for all the great feedback and clarifications. I do have a question when it comes to machining that outboard aluminum insert that you posted pictures for. How were you cutting the cylindrical faces on that part? It seems that if your trying to hold a tolerance between 3-8 thousands that an end mill would have a lot of chatter trying to cut that shape.

For reference I attached a photo with the bonding surface of the cylinder circled. Thanks for the help. 419

[Rory Hourihan]

We had good luck bonding to the OD of the carbon tube on the outboard side, basically an aluminum block with a hole bored to match the OD of the tube. It made the machining of the outboard joints much simpler, but increased the amount of parts we needed.