Phosphoric Anodize for bonded joints 4

blakmax... far more to this subject than meets-the-eye.

To understand fatigue aspects of adhesive bonded structures You must first understand the permutations of the subject, starting with basic adhesive-bonding principles [materials, preparation, etc], affects of the environment, structural geometry, loading, etc... such as...

Aluminum alloys/tempers; bare or clad?

Adhesive bonding process/materials selection...

PAA process... ASTM D3933? SAE ARP1524? Corporate? Other?

Test adhesive bonding primer and application process over PAA [... or NONE].

Test adhesive and application/curing/handling process variables [cure temperatures, pressures, time, etc].

Bonded test specimen configuration? coupon or structure? Special shaping/tapering of the metallic elements? Metal-to-metal? Metal-honeycomb/core-metal? 100% load transfer or shared load transfer? constant amplitude or spectrum loading; Fasten-bonded; test environment [dry, seacoast] and high-temp/low-temp or spectrum-temp; added effects of peel, eccentricity, creep, etc?

Etc...

As far as Your basic question... 'learning tools' I've ever had on the subject of adhesive bonded-structure strength and durability are as follows [readily available on-line]...

Douglas PABST study... Primary Adhesive Bonded Structures Technology... done in the 1970s into the early 1980s. A comprehensive look at every aspect of structural [mostly aluminum structures] adhesive bonding and the influence of every aspect on strength and durability. Most of the important elements of PABST documentation [many volumes] are available on the DTIC website with some hard searching.

NASA and WR-lab reports, many done by companies or universities under contract, this subject.

MIL-A-83376 ADHESIVE BONDED METAL FACED, SANDWICH STRUCTURES, ACCEPTANCE CRITERIA

MIL-HDBK-337 ADHESIVE BONDED AEROSPACE STRUCTURE REPAIR [elements CX]

MIL-HDBK-349 MANUFACTURE AND INSPECTION OF ADHESIVE BONDED, ALUMINUM HONEYCOMB SANDWICH ASSEMBLIES FOR AIRCRAFT

MIL-HDBK-691 ADHESIVE BONDING [CX]

MIL-HDBK-725 ADHESIVES: A GUIDE TO THEIR PROPERTIES AND USES AS DESCRIBED BY FEDERAL AND MILITARY SPECIFICATIONS

MIL-HDBK-83377 ADHESIVE BONDING (STRUCTURAL) FOR AEROSPACE AND OTHER SYSTEMS, REQUIREMENTS FOR

AMS3920 Acceptance Criteria Adhesive-Bonded Metal-Faced Sandwich Structures

SAE ARP1524 Surface Preparation and Priming of Aluminum Alloy Parts for High Durability Structural Adhesive Bonding

If You work for any of the 'big-box aerospace companies', there are likely to be many documents/studies/processes available that are corporate proprietary.




Regards, Wil Taylor

o Trust - But Verify!
o We believe to be true what we prefer to be true. [Unknown]
o For those who believe, no proof is required; for those who cannot believe, no proof is possible. [variation,Stuart Chase]
o Unfortunately, in science what You 'believe' is irrelevant. ["Orion", Homebuiltairplanes.com forum]

Thanks Will

I was well aware of most of these publications but they are of little use for the specific issue I have been asked about, and that is Does PAA reduce the fatigue life of structures compared to FPL etch.

There are real issues about fatigue testing of bonded joints. Many people use short overlap shear test specimens without realising that the test specimen itself skews the results. As the adherend material approaches its fatigue limit the adherend suffers ductile deformation adjacent to the joint and that leads to unrealistic shear strains as the material deforms. The increase shear strains result in premature failure of the bond. It is the specimen that drives this failure mode, not the properties of the adhesive.

In contrast, if the overlap is longer, then the failure always occurs outside the joint for thin adherends so you are not measuring any adhesive properties. My personal opinion is that the way to avoid fatigue is to design with longer overlaps then it does not matter if the surface is PAA or FPL with regard to fatigue. (The long term bond durability differences are another issue and that is not fatigue related.)

Thanks

Blakmax

I'd've thought that PAA was superior to FPL. These are mostly corrosion treatments, not fatigue critical (like shotpeening). they become fatigue critical when they break down, which is why I think PAA is superior, because the process control is much better.

of course the folks over at the "lazy B" ranch have the answers, but they aren't telling.

another day in paradise, or is paradise one day closer ?

Ahhhh... blakmax... the deeper question. hope this make sense...

The PABST contract-testing evaluated all of the well-established adhesive bond preps known at that time.

As I recall many popular processes were rigorously tested in-detail for initial properties and long-term durability in various environments. These included 'known inorganic metal treatments’, combined with various organic primers/adhesive methods/materials . The following methods [tested with process controls for consistent application and important variables] were given close consideration.

Clean/scuff-sand/prime [CSSP]

Forrest Products Lab [FPL] etch [Pasa-Jell 105 is a well-known commercialized version]

Chromic Acid Anodize [CAA, MIL-A-8625 Type I, sealed and unsealed]

Sulfuric Acid Anodize [SAA, MIL-A-8625 Type II, sealed and unsealed]

and almost as an afterthought...

Phosphoric Acid Anodize [PAA], sealed and unsealed]

When all was said-and-done, PAA (unsealed, with certain process controls) + thin-film adhesive epoxy primer [with added corrosion protection] and toughened epoxy adhesives [for peel] stood-out as the undisputed ‘best practice’ for aluminum adhesive bonding. The long-term strength/durability of PAA, under environmental stress and static/fatigue loading was superior to all-other combinations for metal-metal and metal-core-metal assemblies. Other methods had certain advantages for initial strength and fatigue durability… but typically fared poorer for long-term environmental durability. The technical reasons for this result would take considerable time to explain… best left to PABST reports. I believe, that SOME of the advantages of PAA were also due to anodic processes that were less sensitive to [production process] variables that proved to be statistically important.

CAA and SAA also worked ‘OK’ but generally had poorer long-term benefits and were more process sensitive for certain aspects. Also, CAA and SAA anodic baths are far more dangerous and environmentally unfriendly to be around than PAA baths.

The FPL etch was surprisingly good when combined with appropriate surface-sanding preparation [to develop a ‘roughened’ metal surface for better adhesion just before application of the PFL etch]... and over-coating with adhesive-bonding epoxy primers. Notably, this process lacked the long-term benefits of anodic coating surface features that promoted greater adhesive strength and longer environmental durability.

CSSP was found to be particularly poor and ineffective for anything more-than urgent temporary non-structural repairs [if that].

NOTE. In this study, the latent effect of exposed bare-metal and adhesive bondlines... at holes, countersinks, trimmed-edges, lap-steps, exposed adhesive flash/fillets and bondlines, etc... was also found to be particularly important for long-term environmental degradation of the assembly. Protective measures for these features were found to be especially important as they eliminated ‘entry paths’ for the abusive environment. There was added emphasis on the application of protective over-coatings for these features, including: proper deburring and smoothing of metal surfaces/edges/holes/countersinks, etc [‘can’t paint a sharp edge’]; application of chromate conversion coatings [CCC] to bare aluminum; application of corrosion-protective [chromated] epoxy primers over the CCC’ed details AND onto all outer surfaces exposed to the environment [uniform/constant epoxy primer film encasing the finished part]; and installation of the adhesive bonded assemblies/details with corrosion-protective [chromated] pressure/environmental sealants [‘wet sealant’ fay assembly and fastener installations***].
*** except install rivets wet with epoxy primer.

Long-after the PABST program ended, newer technology coatings have evolved that provide ~80--90% of the advantages of the PAA + adhesive primer. These coatings… applied directly to metallurgically-bare metal… were found to provide amazing adhesion and environmental durability when combined with epoxy primers… but that is another discussion.


Regards, Wil Taylor

o Trust - But Verify!
o We believe to be true what we prefer to be true. [Unknown]
o For those who believe, no proof is required; for those who cannot believe, no proof is possible. [variation,Stuart Chase]
o Unfortunately, in science what You 'believe' is irrelevant. ["Orion", Homebuiltairplanes.com forum]

Blakmax,
The test results that Wil is referring to are summarized in a neat table in MIL-HDBK-691, (Table L, page 375 of the PDF on the ASSIST website).
Looking at this table, I can't determine how it compares with the other method from Forest Labs, but are other clues in this 469 page document...
Maybe Table LIV is what you need.

STF

Blakmax... a few last comments before moving-on…

When I was an in-the-field service engineer we were forced to use a combination/sequence of treatments to get a decent quality FPL bond. As I recall here is how a typical repair went-down on Al-Al or Al-HCC-Al...

Hand-sanded bonding metal surfaces [or top of HC-core] to smoothly-roughen the existing/surrounding skin/patch surfaces to a uniform 'matte' appearance [be careful not to sand abusively on thin sheets].

Applied wet IPA process wipe/wipe-dry to clean-off sanding debris.

Applied Pasa-Jell 105 [or in-a pinch P-J 102] surface treatment per PPG [was SEMCO] application guidance [see attached sheet].

Applied a min-dwell-time application of chromate conversion coating [CCC]. NOTE: IF the applied CCC went-down smoothly it added the necessary chromates for improved bondline corrosion resistance; but if the CCC failed to apply smoothly/evenly [exhibited POOR color or streaking or water-break], it indicated that the surface needed further cleaning and Pasa-Jell treatment.

Applied a thin coating of adhesive bonding primer… or a very thin coating [wipe-on/off] of epoxy laminating resin [Typ EA956]… [<0.0005 DFT] that was ‘warm dried/cured’.

Mix/apply RT cure paste adhesives with bondline thickness control [typ 0.005--0.007 nylon thread, woven scrim-cloth or ‘fresh’ glass beads]. Apply paste-mix [~0.010 thick] to BOTH the mating part surfaces. Fill top of exposed HC core with the paste mixture, min 0.12 deep.

Roll-down the mating parts into position’ [center-first then-towards the edges if possible] by hand pressure. Then press surfaces together ‘hard’ and a [slight] squirming movement to spread adhesive contact and eliminate large voids; and ensure squeeze-out around all edges. Follow-up vacuum bag vacuum-bag pressure [with bleeder cloth] ensured max squeeze-out [flattening-down] and relatively uniform adhesive contact thru the bondline during adhesive cure.

After adhesive cure, typical tap-testing, followed by discrete void filling if needed [drill a pattern of 0.063 Dia holes into the voided area then inject-fill-bleed thru these holes with WARM laminating resin from a syringe]. NOTE: by warming the laminating resin in a syringe to 120—140F** [usually a heat lamp] the initial viscosity dropped dramatically, which made syringe injection feasible [higher viscosity of RT resin often made needle injection almost impossible]; and allowed the resin to initiate cure [cross-linking]. Had to be careful to ensure that resin was never over-heated or other damage could occur.
**140F—150F is the temp range that becomes ‘too-hot to touch/hold’ for any length of time, most people.

After adhesive cure and void repairs done, then the surrounding resin ‘flash/fillet’ was sanded-smooth [and bubbles/voids filled by wiping with warm laminating resin].

The repair was then over-coated [sealed] with at least 2-coats of corrosion-protective epoxy primer onto adjacent surfaces [0.0016-DFT min primer film]. NOTE: any exposed bare metal [due to sanding, cutting, drilling, countersinking, etc] was treated with CCC before application of the primer over-coating. A top-coat of polyurethane was applied IF/AS needed. And all fasteners were installed wet with sealant or epoxy primer.

NOTE. I always specified solvent-borne epoxy primer… NOT water-borne epoxy primer… in these type repairs for many reasons.

NOTE. Configuration of mating parts edges was also a critical step. I often taper edges [and sine-wave the edge profile] of thicker patches/doublers, etc to attain a relatively thin perspective, then rounded-them off, to reduce ‘edge-peel’ loading stress and ensure the coating system would ‘wrap-around’ the edge. Reasons for mating edge treatments like this are obvious once You understand the phenomena of how edges of laminations factor into bondline durability.

CAUTION. A hard lesson learned was the ‘human factors’ of adhesive bonding. The technician doing the job MUST understand the details and ‘why-it-has-to-be-done-this-way’ nuances to ensure the work is done cleanly, consistently and uniformly. How did I learn this??? : My Depot unit used a high temp film- adhesive for critical depot repair of bonded [F-15] metal parts. This film-adhesive was VERY expensive… but came in 50-Yd or 100-yd rolls that were life limited to 6-months in a freezer… or up-to 9-months if certain controls/testing could verify quality. One adhesive roll [about 3/4ths unused] reached 9-months and the production supervisor was begging to extend it to 12-months. I agreed to ‘give-it-a-shot’. Several lap-shear test coupons were assembled by a technician I did not know very well: they performed very erratically… over 1300-PSI variation ‘low’ from the 4000-PSI-LS requirement. I condemned the material. A week later the supervisor asked for ‘lap-shear re-test’ of the film-adhesive roll, with the coupons assembled by a [another] technician that I knew fairly-well. This time the results were remarkable: all the coupons tested fell in the range of 3800—3950-PSI-LS. THAT is when I had a long conversation with the tech and discovered he ‘knew-his-stuff’… and had ‘expressed concerns about the quality of the previous testing’ [IE: the tech who did that job]. I couldn’t re-qualify the adhesive… but I was able to ‘disqualify’ the original [sloppy] tech and enthusiastically ‘qualify’ the second tech!


Regards, Wil Taylor

o Trust - But Verify!
o We believe to be true what we prefer to be true. [Unknown]
o For those who believe, no proof is required; for those who cannot believe, no proof is possible. [variation,Stuart Chase]
o Unfortunately, in science what You 'believe' is irrelevant. ["Orion", Homebuiltairplanes.com forum]

Will, you mention using glass beads for bondline control. I know this is a fairly common technique but i have observed some problems with it.

It is also well known that to get good bondline properties that the bondline thickness needs to be with a certain range. Too thin and the bondline has very little strain capacity to help distribute loads evenly. That is why beads are used. However, where the bead is, the bondline thickness is essentially zero. The bead has a very high modulus relative to the adhesive and will not shrink with the adhesive when it cures. I have observed disbonding at the beads by bonding two glass microscope slides together as an experiment. You can see interference fringes at each bead caused by the disbond.

Does anyone have any comments on this issue?

Thanks to all who have responded, but what I am after is fatigue testing, not bond longevity testing. My friend who works for a large company is trying to justify a change from FPL to PAA but hits a brick wall because of a mantra that exists within a customer organisation that strongly believes that PAA is prone to fatigue at high loads. I personally hate these mantras that once embedded on (in some cases) spurious test results then become "law". Often the original test data and test method become lost, but the mantra lives on. A classic example is the removal of cladding from bonded repairs because of a perceived fatigue weakness in the cladding. That mantra was justified by ASTM D1002 specimens tested under fatigue, and yes there was a difference in specimen life. However the lap shear test is exceptionally bad for testing under fatigue because as the adherend itself approaches its fatigue life the adherend exhibits plastic deformation adjacent to the joint. That plastic deformation greatly increases the shear strains at the ends of the joint and leads to premature failure either of the cladding or the adhesive. This failure is driven by the test method, not any property of the adhesive or the cladding. To demonstrate this, we undertook long overlap fatigue tests and in every case the failure occurred outside the joint. There was no failure of the adherend or the cladding. Hence for realistic overlap lengths fatigue of the cladding is not an issue.

From a personal perspective, I have absolutely no faith in FPL or CCC for repairs based on wedge test ASTM D3762 results and literally hundreds of failed repairs on F-111 which were performed using Pasajel. I am also aware that one organisation found that the only way to get Pasajel 105 to pass wedge tests was to heat the structure to 140-180F using heat lamps. My response was that there was no way I would load up a brush with a carcinogenic acid and reach past screaming hot heat lamps to apply it. By changing from Pasajel to a grit blast and silane process (no primer) we changed the repeat repair rate from 43% to 0.06%.

Will, I am in total agreement with you on technician reliability. The 0.06% of failures we had over fifteen years were as a direct result of technician malfunction.

With regard to the glass sphere discussion, I don;t have much experience with paste adhesives. Most of our repairs were on high speed aircraft and paste adhesives simply don't have an adequate Tg for those applications.

Regards

Blakmax

I think you'll need to do your own fatigue test, particularly since the opinion is so specific.

It is "odd" since I think it's universal that PAA is a superior bond to FPL. How your QC on the FPL bond ? If QC is very good FPL can be a good bond, but if you take your eye off the ball the quality drops quickly.

another day in paradise, or is paradise one day closer ?

A few minor points before I fade away...

1. I learned repair configuration design from the basic PABST document dedicated specifically to 'adhesive bond design'. I tried really hard to adhere to those design principles.

2. Composite pro...

I think the rational for use of glass beads in paste adhesives for bondline thickness control has to be clarified.

The general rule I use is 2-to-4% glass beads by mass/weight. For a 100-gr adhesive, adding 3% glass beads by weight [3-gr], thoroughly mixed-in, results in a fairly 'sparse' bead population.

Also, I actually prefer glass or microspheres ['bubbles'] which are more typically added to commercial potting compounds for density control. For my purposes, adding 2--4%-by weight microspheres [ILO glass beads] added to a paste mixture, will have a much larger population [quantity] per unit-area than solid glass beads [often specified 3M Glass bubbles]. However, the 'availability/scarcity' of the small size microspheres needed for bondline control typically drive us to call-out solid glass beads/shot which are commonly used for peening and cleaning [MIL-PRF-9954, AMS2451/6, etc].

For a long-time I specified a fused-nylon-filament low-density cross-weave scrim cloth that was made by a vendor in the USA-N/E. This scrim was applied to an adhesive 'buttered' part just before assembly [trimmed slightly oversized so threads protruded all-around the added part]... however that became difficult to acquire due to low quantity use in-the-field. I have also used clean nylon-monofilament [fishing] line for bondline control: weaving the line back-forth across patches/doublers with threads taken outside of the patch/doubler planform. In both cases the nylon threads had to protrude beyond edge of bonded part to ensure edge crush did NOT thin-out/the adhesive locally around the edge [a bad situation]. The protruding nylon threads then required special consideration: nylon wicks moisture, so the residual threads ‘bits’ had to be buried by epoxy and sealant coatings to ensure they weren’t exposed to moisture: requiring a lot of attention to detail by the repair tech.

I hope You can understand the circular-rational for current emphasis to use glass beads for bondline control.

3. One last 'trick I learned somewhere in the stacks of documents. After application of Pasa-Jell/CCC, it was discovered that warming the parts [120-140F by heat lamp] did (4) very important/subtle 'things' which helped form a MUCH tougher primer bond to the metal than straight primer application methods.

(1) the heat drove-out excess moisture from the surface [etched/anodized-tooth] microstructure.

(2) the initial heat-shock helped initiate/drive primer [epoxy-resin] cure in very thin coatings.

(3) epoxy resin coatings [IE: catalyzed adhesive bond primers] viscosity dropped slightly improving 'flow/wetting'.

(4) and as the part cooled to RT with the wet-primer coat over the surface, the [lowered viscosity] primer was 'sucked/wicked' into the surface microstructure for a more-effective primer-surface-mechanical bond. This is due to capillary wicking action that is assisted by 'cooling pressure' differential... air-pressure forced the wet-resin into the porous surface microstructure as the metal cools. CAUTION: there is a upper temperature limit to this practice for RT cure primers

OH YEAH... 'Grit-blast/silane' and 'grit-blast or sanding/SolGel [Boegel?]' pretreatments have made a huge improvement in in adhesive bond surface prep reliability over FPL or PANTA surface preps.

Revealing my 'secrets' here...ahhhh-well... gotta-go back-to-work.


Regards, Wil Taylor

o Trust - But Verify!
o We believe to be true what we prefer to be true. [Unknown]
o For those who believe, no proof is required; for those who cannot believe, no proof is possible. [variation,Stuart Chase]
o Unfortunately, in science what You 'believe' is irrelevant. ["Orion", Homebuiltairplanes.com forum]

Wil,
Your "secrets" go into my back pocket, and get passed off as my own more often than I admit. I will continue doing this until you publish something copyrighted, and force me to give you all the credit you rightly deserve.

STF

I also have a question to CompositePro or Wil (whoever dares to answer):
Is there an advantage to specifying a scrim cloth layer in the joint, over the glass beads, for bondline control?
This cloth is available and convenient at this facility (0.005" thick). It also is easier to prove that it was done, when checking that the process was done properly by the tech. No need to look for strands sticking out of the bond, one can check that the roll was taken out of storage to have evidence that this step was not skipped. The beads are clearly much easier to mix into the adhesive, and I'm thinking of changing my habits.

STF

I was involved with the manufacture of aerospace film adhesives. Most used a spun bonded polyester carrier or a nylon knit. I believe that the knit carrier provided better peel properties because of the 3-D structure of the knit, which tended to interrupt crack propagation. These were both polymer fibers so the CTE would match the adhesive more closely than glass beads and also the spacing was due to several fibers bridging over each other in an area rather than one hard spot. These carriers also helped provide a path for air removal by vacuum and would also break air bubbles into smaller ones.
Some high temperature adhesives used a woven glass fabric carrier because nylon or polyester would melt or degrade. High temperature adhesives generally do not have high peel strength.
So, I would say that scrims are superior to beads, but beads are better than nothing.