Adhesion in Industrial Coatings Systems: Standards Literature,

Lab Practice, Field Practice and a Holistic Understanding

William Seawell, Desiree Eason
Induron Coatings LLC
3333 Richard Arrington Jr. Blvd. N. Birmingham, AL USA 35234, +1 (205) 324-9584

Abstract
Here we examine the notion of “adhesion” as it relates to thermosetting organic coatings on metallic substrates. The chemistry
and physics of coatings adhesion is reviewed and considered. Commonly used standard methodologies for adhesion testing
used by paint manufacturers, by field inspectors and others are reviewed and examined. Suggestions are made for recom-
mended field and inspection practice. Adhesion testing’s role in the engineering standard literature is examined.

Philosophy of Coatings Testing in the fog chamber; what about 115ºF? A form of “salt fog perfor-
mance” would still be measured. It would not be ASTM B117 salt
When engineers specify coatings materials for any industrial ap- fog or ISO 9227 salt fog testing though. It would be testing by
plication the process often involves the comparison of various so- some non-standard methodology but still distinctly a form a salt
lutions through the lens of physical and chemical properties of the fog testing and still measuring some form of corrosion resistance
cured coating. Standards organizations such as ASTM Interna- in high salinity environments.
tional (formerly the American Society for Testing and Materials, Choice of precise testing methodology is somewhat arbitrary. This
now having adopted the international moniker), and AMPP (Asso- is not a problem per se. Testers often have a large collection of
ciation for Materials Protection & Performance) produce stand- standardized methodologies to choose from and they may select
ards documents which are widely disseminated and used very ef- one or another for a variety of reasons. The only problem arises
fectively by the entire industry. when “apples to oranges” comparisons are attempted.
There are two ways one may speak of the properties of a com- In the following sections we address such philosophical issues
mercial material. There are the results of an actual standard meth- around the topic of adhesion testing in commonly used organic
odological assessment (a lab test) that may be traced back to a coating systems such as epoxies, urethanes, alkyds and acrylics.
specific version of a specific standard, and there is the actual un- We propose, and at least suggest answers to the following ques-
derlying property of a coating that said methodology is attempting tions: What is adhesion in coatings? How is it measured? What
to measure. As in all scientific measurements, the choice of meth- physical phenomena are at play behind the phenomena of adhe-
odology used to check a particular physical property is somewhat sion and its measurement and how do certain types of measure-
arbitrary. ment best (or worst) capture that? Which form of measurement
Take for example salt fog testing. This is a common corrosion re- and ontological understanding of adhesion best serves the coat-
sistance assay for organic coatings over steel. The test samples ings community and non-expert coatings decision maker?
are carbon steel panels coated with the material in question. The
coated panels are then put inside a “salt fog cabinet” where they Coatings and Adhesion
are exposed to a high humidity, high salinity, warm temperature
environment for extended periods. There are a variety of stand- Adhesion is perhaps the most important property of a coating sys-
ards written that detail procedures one may carry out for such a tem. Without adhesion to substrate, all other properties of the
“salt fog” testing regime. Amond them are ASTM B117 “Standard coating are unhelpful. Simply speaking adhesion is the strength
Practice for Operating a Salt Spray (Fog) Cabinet”[1] and ISO 9227 of a coating’s bond to the underlying substrate. The concept is
“Corrosion Tests in Artificial Atmospheres – Salt Spray Tests”[2]. applicable to almost any combination of coating and substrate.
But what are salt fog tests measuring really? Strictly speaking salt One may speak of the adhesion of an epoxy lining to the carbon
fog tests only measure the coating’s performance in a salt fog steel substrate of a tank interior just as one may speak of the ad-
cabinet. Pragmatically, the testers are often seeking some form of hesion of an overcoat paint job applied to an already painted sub-
information concerning corrosion or creep resistance in the con- strate. Hot dip galvanizing has a certain amount of adhesion to
text of highly corrosive atmospheric services. The 2019 version of raw steel and a coating may later be applied to that same galva-
the standard itself reads (under the heading “significance and nizing and we may inquire about its adhesion as well. The concept
use)[1]: is nearly ubiquitous in coatings.

“3.1 This practice provides a controlled corrosive environment The adhesion of an industrial coatings system may be measured
which has been utilized to produce relative corrosion resistance by a multitude of parties for many commercial reasons. Including
information for specimens of metals and coated metals exposed but not limited to:
in a given test chamber ”
• The paint manufacturer measuring the adhesion of their
Would salt fog testing still serve its purpose if the standard tem- products on various substrates under various applica-
perature at which the test was run was increased one degree tion and curing conditions during the course of re-
Fahrenheit? What about if the salt concentration was just a little search, development and marketing of said coatings
higher than the specified 5 parts NaCl to 95 parts water? Or a little
lower? What if the angle of the panels sitting in the test chamber • An in-field inspector measuring the adhesion of a paint
was shifted one degree? ASTM B117 calls for a 95ºF temperature system installed on-site as part of an applicator quality
control process (either as mandated by a specification,
 or in response to some perceived application issue in-
field during the course of the work) • Dipole Interactions: Molecular substances such as the
polymers that constitute cured paint are inherently com-
• A third party science, engineering or industrial services posed of uneven distributions of charge. Atoms at their
provider vetting technical claims about coatings in the core are composed of discrete units of charge such as
interest of consulting, producing a specification or simi- protons (+1 unit of charge), electrons (-1 charge) and
lar technical writing neutrons (no electric charge). As such, for coatings pol-
ymers there are always large sections of the molecules
• During the course of a failure analysis being conducted in question that posses a “net dipole moment”; or areas
i.e. some problem has occurred in the field and there is of a molecule that have a partially positive or negative
reason to seek the root cause of the failure electric charge. As positive and negative charges are
attracted to each other, and like charges repel, this type
Adhesion may be measured both under laboratory conditions and
under field conditions. The standard literature does not differenti-
ate in any way that necessitates one over the other.
It cannot be reiterated enough, the extent to which nearly all con-
ceivable variables are crucial and relevant in the reporting of coat-
ings adhesion data. The methodology used, the cure time and
conditions [temperature and relative humidity] the coating experi-
enced prior to testing, the substrate used, the substrate surface
preparation method used, and precise details of the test as con-
ducted are all hyper-relevant and may skew results significantly.
For the remainder we restrict our scope to thermosetting [chemi-
cally cured] organic coatings applied to metallic substrates as the
topic of concrete adhesion is complex and an entire topic mostly
unrelated to coatings over metallic substrates. This also limits the
scope in terms of resin technologies discussed to the commonly
used two component or moisture cured industrial coatings types;
namely epoxies, two component urethanes, novolacs, vinyl es-
ters, epoxy poly-siloxanes, moisture cured urethanes and alkyds.

Chemistry & Physics of Coatings Adhesion

The technical details of how organic coatings adhere to substrates
are dense and numerous. Multiple modes of action are at play,
and depending on the precise details of the substrate, the level
Figure 1: Representation of
and type of substrate preparation, and the composition of the
epoxy-like coating interacting
coating some modes of action may be more dominant than others.
with an iron substrate through
Broadly speaking organic coatings develop adhesion to their un-
dipole and Van der Waals in-
derlying substrates during the course of curing. Wet paint applied
teractions. Not all bond angles
directly to a substrate does have some affinity for the underlying
are true to life due to limita-
substrate though liquid adhesion though it is a phenomenon dom-
tions of 2D representations
inated by the rheological behavior of the liquid [thixatropy], gravity,
speed of the cure reaction, surface tension and is not the primary
topic of discussion here (though it will be discussed). The adher-
ence of the curing liquid coating mixture to the substrate is not
easily measured in field conditions and is not relevant when con-
sidering specifications. of columbic force ends up having an appreciable effect
As coatings cure on their substrates they develop adhesion on how a coating interacts with the underlying substrate
through one of the following mechanisms: and therefore how adhesion develops. Dipole interac-
tions are of particular importance in epoxy and other
• “Mechanical Interlocking”: As the liquid coating sits on high polarity resin systems[3].
the surface and time elapses the liquid “wicks” into the
various pores, nooks, pores and surface anomalies on
the substrate. Once the coating has reached a sufficient
degree of cure that the mixtures vitrifies (becomes
solid), the coating material has already penetrated the
substrate in many random and circuitous ways that are
not easily reversed. An often used analogy is a barbed
fish hook. It slides into the fish’s skin readily but is diffi-
cult to remove in the opposite direction. “Velcro” or hook
and loop fasteners are also an illustrative comparison
to make.
 • Hydrogen Bonding Interactions: Any molecule that con- Figure 2: Comparison of paints with extreme surface wet-
tains a hydrogen atom bonded to an oxygen, nitrogen, ting (left) and minimal wetting (right). Coating on left show-
or fluorine may participate in hydrogen bonding. Hydro- ing linseed oil wetting front underneath glass slide
gen bonding is a particularly extreme case of dipole in-
teraction. Coatings that contain -OH, -NH-, or -NH2 moi-
eties are thought to participate in hydrogen bonding in-
teractions with some substrates[4], particularly in the
case of adhesion between coats of paint.

• Van der Waals Interactions: All matter is affected by

Van der Waals forces to some degree. Van der Waals
forces, also known as “induced dipole-induced dipole
interactions” arise as the result of the electron clouds
surrounding all atoms and molecules interacting with
other nearby electron clouds in an electrostatic manner.
For example one electron in a coating resin molecule
may repel an electron near the surface of an iron sub-
strate, this may create a small region of positive charge
in the wake of the displaced electron that was repelled.
This region of positive charge would then go on to at-
tract electron density from the surrounding resin mole-
cules, net strengthening the adhesive interaction at the
Figure 3: Image of a water droplet on the surface of an
interface between substrate and coating. Van der
“elephant ear” houseplant showing very minimal wetting,
Waals forces are a consequence of the quantum me-
high contact angles and the “beading” effect
chanical nature of electrons at this scale and a complete
explanation is far beyond the scope of this work. Van
der Waals Interactions on the whole contribute

• Covalent Bonding: Organic molecules can, in some

cases, form a permanent, irreversible chemical bond to
the substrate in the same manner that atoms bond to
together to form molecules, i.e., the sharing of electron
pairs to form discrete chemical bonds.

The extent to which each adhesion mechanism contributes to the

bulk adhesion of a given coating or class of coatings is debated
and it is unclear [6]. Dipole and hydrogen bond interactions are
thought to be particularly dominant in epoxies due to the resin’s
high polarity and high density of hydrogen bond donor and accep-
tor sites [5] but this is more of a consensus scientific narrative than
an objectively proven fact. Analytical techniques do not exist to
definitively determine the degree to which each adhesion mecha-
nism contributes to a given adhesion value.
The phenomenon known as “wetting” is a crucial step in the pro-
Figure 4: Image of a water droplet on a concrete surface
cess of forming adhesion. Wetting is a necessary but not sufficient
showing a large amount of wetting, low contact angle and
condition for adhesion to be established. Wetting is the process
maximization of contact with the surface
of a liquid maximizing its contact with an underlying solid sub-
strate, displacing gases and “wicking” into the geometry of the
surface. Many factors go into the amount of wetting a coating will
display on a given surface among them; the major factors being
the chemical identity of the coating and substrates themselves
and roughness of the substrate surface at both the nano and mi-
cro scales[7-8]. Wetting may be spoken of in terms of “contact an-
gles”, or the angle a drop of liquid takes on at the solid/liquid in-
terface when placed on a substrate. A droplet of liquid will seek to
minimize its own internal energy by minimizing the number of en-
ergetically unfavorable surface interactions. See figures 3-5 for
visual examples of contact angle.
 Abrasive blasting, hydro-blasting and other destructive surface
Figure 5: Side-view representation of contact angles for a preparation methods drastically alter the roughness of substrate
liquid with A) poor wetting, B) moderate wetting and C) ex- surfaces; increasing the exposed surface area on the order of 2-
cellent wetting 3 times for dry abrasive blasting[9]. This is one reason, among
many, that many industrial coatings require an SSPC SP-6 Com-
mercial Blast or higher level of abrasive blast; often with an addi-
tionally specified 1-4 mils profile depth.

The increased surface area achieved from blasting is a benefit to

mechanical interlocking adhesion but also increases the available
surface area for the various types of chemical adhesion men-
The physical underpinnings of contact angle are rooted in the tioned above.
three phase surface tension relationship between the liquid/vapor
boundary, the solid/liquid boundary, and the solid/vapor bound- Adhesion Development in Pragmatic, “Jobsite” Variables
ary. Theoretically contact angle is determined by the following
relationship: Mechanical interlocking is perhaps the most often discussed ad-
hesion mechanism in the industrial coatings zeitgeist. It is the
mechanism for which paint applicators and jobsite inspectors
have the most control and can exercise the greatest amount of
influence on the final outcome. The degree to which a coating may
Where γlv is the surface tension at the liquid vapor boundary (i.e. develop mechanical interlocking adhesion is thought to be influ-
a liquid droplet on the surface and it’s surface tension with its own enced by the following practical factors:
vapor phase), γsv is the solid/vapor surface tension between the
substrate and droplet vapor phase, γsl is the solid/liquid surface
• Surface Preparation Method Used: Various methods of
tension between the droplet and the solid substrate. In order for
surface preparation produce qualitatively different sub-
significant wetting to occur, the γlv must be sufficiently low or the
strate profile results. Abrasive blast cleaning with a rel-
surface roughness factor, r, must be sufficiently high for a lower-
atively hard [compared to substrate], angular grit tends
ing of the contact angle to cause a net decrease in the energy of
to produce a dense, angular “anchor profile” with a rel-
the system.
atively large actual surface area per unit square area of
metal substrate. Other methods such as the use of nee-
All of the “gamma” surface tension terms have units of [(force) ×
dle guns or wheeled power tools tend to produce a
(length)-1] and can therefore be represented graphically as vector
clean substrate that is not profiled or significantly tex-
quantities. See Figure 6 below.
tured. The results of adhesions tests over these various
surface prep methods vary accordingly. The surface
preparation methods which produce the highest surface
Figure 6: Side-view representation of liquid drop on a sur- area substrates tend to produce more tightly adherent
face showing all surface tensions and their relationship to paint systems. It is known that the presence of metal
contact angle oxides at the coating/substrate interface has an impact
on adhesion[3] and therefore the type of surface prepa-
ration used will likely have an impact on this aspect as
well.

• Surface Profile Depth & Density: The industry stand-

ards literature alone offers a variety of surface prep
standards to select from, SSPC-SP6 Commercial Blast
Cleaning, SSPC-SP10 Near White Metal Blast Clean-
ing, and SSPC SP-5 White Metal Blast Cleaning being
among the most encountered on carbon steel sub-
strates. All of these blasting surface preparation stand-
ards allow different amounts of rust, staining or surface
corrosion products to exist on the surface and also are
likely to result in different surface profile depths as a re-
Contact angles are not immutable chemical properties of a liq-
sult of less or more blasting being required to achieve
uid/solid pairing. Contact angle has a linear dependence on sur-
the specified level of cleanliness.
face roughness given by the simple relationship[6,10]:
• Cure or Drying Speed of the Coating: All other factors
equal, a coating formulated with slower drying thinners
or thinned with slower drying thinners will have more
Where “r” is the surface roughness factor, θm is the experimentally time to develop mechanical interlocking adhesion at the
measured contact angle. The roughness factor scales the theo- substrate / coating interface prior to the film becoming
retical Young contact angle linearly. vitrified enough to case wicking action. This may be
controlled on site by thinning rate or choice of thinner
 within the allowed choices allotted by the coating
Method Scope Notes
manufacturer.
metallic sub-
ASTM D4541 “pull off” adhesion
• Thinning: The addition of solvents, thinners or strates
other viscosity reducing agents into paint may play metallic sub-
ASTM D3359
a very important role in “wetting” and therefore the strates, thin “x-cut” tape test
Method A
development of mechanical interlock adhesion in films
certain scenarios. All other things equal, a lower metallic sub-
ASTM D3359
viscosity liquid coating will tend to penetrate the strates, thin “cross hatch” tape test
Method B
underlying substrate more readily than a more vis- films
cous liquid coating. “Pick back” knife adhe-
ASTM D6677 All substrates
sion
concrete sub-
ASTM D7234 “pull off” adhesion
strates
Loss of Adhesion, Failure Modes
All of the aforementioned methods are destructive testing meth-
A coating may lose its adhesion through one of many failure ods. All result in an endpoint where the coating sample “fails” (is
modes. The root causes for these failure modes are numerous removed from the substrate). These are the testing methodolo-
but all of them fall into one of the following categories based on gies most often used in the United States for industrial coatings
the results of what type of coating or substrate damage is left be- work on carbon steel and as well as non-ferrous substrates such
hind after the failure: as aluminum or hot dip galvanized steel. ASTM D7234 is used for
concrete substrates and comes with its own unique complications.
• Adhesive Failure: The primary mode of failure is at the Concrete adhesion testing will not be addressed here.
coating to substrate interface. The underlying substrate
can be visibly seen showing through the damaged area. All of these methods involve subjecting a coating, on a substrate,
Coating delaminates away from the substrates cleanly. to some force that induces failure of the coating away from the
This indicates that the weakest bond in the system was underlying substrate. The coating samples in question may be la-
adhesive bond at the coating/substrate interface. boratory prepared test panels, or parts of some substrate on an
active job site being tested as part of an inspection or evaluation.
• Cohesive Failure: The primary mode of failure is break-
ing inside of the coating film itself. The underlying sub- “Pull-Off” Adhesion
strate cannot be seen showing through the damaged
area. This indicates that the weakest bond in the system Pull-off adhesion style tests are among the most commonly en-
was the internal cohesion of the film itself. countered in the industry. ASTM D4541 is the most common pull-
off standard on carbon steel in the United States. ISO 4624 is the
• Substrate Failure: The primary mode of failure is break- rough ISO equivalent often encountered in outside the United
ing inside of the substrate itself. Not commonly seen on States.
intact metallic substrates as the vast majority of metallic
substrate materials have tensile strengths greatly ex- The test is conducted by gluing a “dolly”, a small screw or “quick
ceeding both the internal cohesive strength of thermo- connect” fitting onto the surface of a cured coating. After the glue
setting organic coatings and the adhesive strength of cures, the dolly is pulled away at an angle normal to the substrate.
nearly all organic coating / metallic interfaces. However The pulling action may be achieved either mechanically or pneu-
in maintenance painting (painting done on already matically but either way, the value is recorded in pressure units
painted substrates in-field rather than new construction such as pounds per square inch.
painting or in-shop OEM painting), rusted surfaces pre-
pared as per SSPC SP-2 Hand Tool Cleaning, SSPC Pull-off adhesion is notable for being quantitative in a sea of
SP-3 Power Tool Cleaning, or SSPC SP-16 Brush-Off mostly qualitative adhesion tests. All the other methods listed in
Blast of Non-Ferrous Metals are commonly painted. the table above have some degree of objective measurement
These substrates offer opportunities for seemingly scheme, but the methods themselves ultimately involves a cate-
tightly adherent mill scale [in the case of carbon steel] gorization of experimental outputs into some arbitrary category.
or other surface oxides to “fail” away from the bulk sub- This is distinctly not the case for ASTM D4541 and ISO 4642
strate at a lower force than the coating’s bond to the which records a quantitative value for the force required to remove
substrate. the coating from the substrate. That is, pull-off methods may re-
turn any positive real number and that number represents a real
Measuring Coatings Adhesion in Practice physical quantity.

In pragmatic terms adhesion on metallic substrates is measured Pull-off adhesion testing also offers additional information in addi-
by a variety of testing methodologies. A few of which are detailed tion to the adhesion value itself, it also captures the precise mode
below: of failure well; the modes discussed in the prior section. Pull-off
adhesion allows the testing technician to examine the dolly and
substrate after the test and determine which layer is exposed.
When there is coating on the dolly underside, remaining on the
substrate post-pull that indicates that the coating failed cohe- not simple in practice and is subject to many methodological ca-
sively. In cases where bare substrate is exposed there was at veats.
least some degree of adhesive failure at the substrate. See Fig-
ures 7-9 for demonstrative examples of adhesive and cohesive ASTM D4541 allows for the use of five different types of testing
failure on adhesion dollies. apparatus with corresponding methods. The five are enumerated
below:
Figure 7: Side view representation of adhesion dollies
showing complete adhesive failure to substrate (left) • Type I (discontinued) – This type of adhesion tester is a
and cohesive failure within film (right) crank driven dynamometer that mechanically pulls the
dolly from the substrate from turning the crank. These
are an older model of adhesion tester not in common
use today. As of the 2022 version of the standard this
method has been discontinued.

• Type II (Method B) – This device operates by measuring

the displacement of an internal spring. The device is
mechanically operated. This is among the most com-
monly used type of adhesion tester for in-field inspec-
tions.

Figure 8: Image of polyamide epoxy primer, one coat • Type III (Method C) – Tester than pulls the dolly in an
system showing adhesive failure to unprepared alumi- automated fashion with hydraulic force.
num substrate from ASTM D4541 pull
• Type IV (Method D) – These are often referred to at
“patty-style” devices as the pulling action is imparted
onto the dolly by a metallic pancake shaped device con-
nected to an airline. Type IV devices typically have pre-
cise controls for the ramp speed of pulling force based
on the expected value of failure. Type IV adhesion test-
ing devices are often encountered in laboratory environ-
ments and are rarely practical in field.

• Type V (Method E) – Also often used in laboratory en-

vironments, these are similar to Type III testers but uti-
lize a hand crank and controls for ramp speed similar to
a Type IV.

• Type VI (Method F) – Very similar to a Type IV “patty”

Figure 9: Image of ASTM D4541 adhesion results of 25- style tester but more optimized for in-field use.
35 mil 100% solids epoxy lining applied to the same epoxy
polyamide primer depicted in Figure 9. Failure modes All of these types of pull off testing apparatus are useable and
shown are cohesive splitting in the primer; excepting the yield internally comparable results, however it is important to note
bottom right which is a cohesive split in the topcoat. that ASTD D4541 itself reports that fairly large variation may be
expected between types of testing apparatus.

Below, in Figure 10 are reported the average pull values listed in

ASTM D4541’s “round robin” study used to compare the various
pull methods and their statistical properties. It is worth noting that
in all cases but one Type IV testers yielded the highest values.
The one exception being Sample E in which a Type V tester
yielded and higher value closely followed the types VI and IV test-
ers.

It is of the utmost importance to consider the fact that Type IV and

V testers are primarily found in laboratories. This is not to invali-
date numbers taken from those types of testers, but rather to reit-
erate the fact that field numbers may not be comparable to man-
ASTM D4541 allows for two types of tests. One where the sample ufacturer or third party lab numbers for reasons related to how the
is tested until failure and the resultant load value recorded. The test is conducted.
other where a pass/fail criterion is established prior to testing and
the sample is subjected to the agreed upon load (passing if it
doesn’t fail at or below prescribed value). Pull-off testing is very
common and invaluable for formulators but the methodology is
 It should be noted that ASTM D6677 explicitly states it is to only
Figure 10: Results of ASTM D4541 interlaboratory anal- be used on coatings where a cut can be made down to substrate.
ysis of various pulling methods in the standard. Samples This rules out many thick film, or ultra-hard coatings from being
A-F were submitted are consisted of six different types of assessed under this standard.
coatings. Samples A and F both returned values higher
than could be measured by some of the methods. All Due to its qualitative nature ASTM D6677 is often looked over in
clues in pounds per square inch favor ASTM D4541 in the specification literature or reported man-
Type II Type III Type IV Type V Type VI ufacturer performance data; however this method’s simplicity and
Sample B 1195 1974 2458 2210 2070 qualitative nature make it an invaluable tool for inspections and
diagnosing problems in the field.
Sample C 549 1221 1232 1120 1106
Sample D 1212 2110 2707 2481 2368 Pull-off adhesion often suffers from the issue of providing specific
Sample E 1385 2012 2354 2449 2327 quantitative information for which there is no real universally
agreed upon criteria for acceptance and no widespread intuition
for statistical significance. Consider the following rhetorical ques-
Figure 11: Results of ASTM D4541 intralaboratory anal- tion: “What is the minimum acceptable pull off adhesion value for
ysis of various pulling methods in the standard. Showing a moisture cured urethane zinc primer applied to the interior of a
maximum recommended percent difference for two in- potable water tank?” No one in the industry has an agreed upon
tralaboratory results to be considered equivalent answer for that. All that can be known for certain are any specifi-
cation bound minimum values and any information the coatings
Maximum Recommended % manufacturer may provide about their specific product. There is
Intralaboratory
Difference no standards organization; ASTM, AMPP, AWWA, API, or any in-
Method B 67.4 stitution like these setting such standards. The same is true of
ASTM D6677 and other knife pick back methods; however with
Method C 33.8 knife pick-pack methods the experimental data gained from the
Method D 14.8 testing is much more digestible, intuitive and does not overcom-
Method E 27.8 plicate the matter with the illusions of quantitative meaning. That
is not to disparage pull-off adhesion per se. That is just to reiterate
Method F 17.5
that the, only semi-quantitative and simplistic, scoring rubric of 0-
10 used by ASTM D6677 is a feature and not necessarily a hin-
Figure 12: Results of ASTM D4541 interlaboratory anal- derance in all instances. Pull-off methods are more quantitative
ysis of various pulling methods in the standard. Showing and more precise inherently. It’s just that precision is cumbersome
maximum recommended percent difference for two in- when it is not needed and can even actively obfuscate under-
terlaboratory results to be considered equivalent. standing.
Maximum Recommended % The following statement is an opinion of the author and the reader
Interlaboratory
Difference of this paper should take it accordingly; But it would not be risky
Method B 76.0 to hazard a guess that there are no paint failure analyses, cur-
Method C 65.9 rently active specifications or troubleshooting events that suffered
for having used knife adhesion over pneumatic.
Method D 28.4
Method E 34.1
Figure 13: ASTM D6677 knife adhesion results for same
Method F 23.0 coating system pictured in Figure 9

“Pick Back” Knife Adhesion

ASTM D6677 details the method often known as “knife adhesion”.
The test is done by using a sharp knife or box cutter to cut the
coating down to substrate in an “X” pattern. Once the cuts have
been made the knife is then applied to the central intersection
point of the cuts and used to “pick back” the coating in a wrist
flicking motion. The adhesion is scored on a scale of 0 to 10. Each
step on the scale is defined by both the size of the paint fragments
removed and the ease with which they are removed from the un-
derlying substrate.

Knife pick back adhesion is inherently qualitative. However when

proper measurements of the size of the fragments are conducted,
or one person is used to do all the pick-backs in question a sur- Figure 14: Summary of scoring rubric from ASTM
prising degree of consistency can be achieved. D6677.
 Fragment
Score Pick Back Difficulty Size
10 Coating extremely difficult to remove ≤ 0.03125"
0.0625" < x <
8 Coating difficult to remove 0.125"
Coating somewhat difficult to re- 0.125" < x <
6 move 0.25"
Coating somewhat difficult to re-
> 0.25"
4 move
2 Coating easily removed > 0.25"
0 Coating easily peeled > 0.25"

Tape Adhesion
ASTM D3359 is a fairly common standard that comes with two
methods (Method A, and Method B). Both methods involve the
use of some kind of cuts to a film followed by adhering a certain
standard type of masking tape to the cut area. The tape is then
pulled away in a controlled manner and resultant amount of de-
lamination is observed.

Method A is the “X-cut” method where an “X” is cut into the film
with a knife in a manner similar to that described in the pick back Figure 17: Scoring rubric from ASTM D6677 as written.
knife adhesion method ASTM D6677. Here instead of picking at
the central cross of the “X” the tester will apply tape that is com-
pliant with ASTM D3359 [the standard defines the strength and
width of the tape, these tapes are commercially available and usu-
ally labelled “ASTM D3359 Tape” or similar].

Method B is the “cross-hatch” tape method. The cross-hatch is

essentially a six-line by six-line square grid cut into the coating.
After the cross-hatch is cut the same tape is applied and removed.

In both methods the standard explicitly calls for the tape to pulled
back onto itself at a 180º angle. This tends to be the largest source
of error in the method as it is difficult to pull the tape at precisely
180º repeatably and variation in the angle of the tape being pulled
may result in relatively large variation in the compressive and ten-
sile forces experienced at the substrate.

Both methods use a 0-5 rating scale, the letter after the number
denoting the respective methods A and B. Both visual scoring ru-
brics have been reproduced below

Figure 15: ASTM D3359 B cross-hatch adhesion results

for a demonstrative thin film system
 that there be prearranged criteria for acceptance or failure. Ad-
hesion testing conducted outside this arrangement does not fur-
Figure 15: ASTM D3359 B cross-hatch adhesion results
ther the cause of investigation. An adhesion value reported with-
for a demonstrative thin film system
out context, without controls or detailed information about the
methodology cannot be compared to any control data directly and
is of little value in this situation.

A thorough consideration of these criteria for jobsite testing

strongly points toward ASTM D6677 Knife Adhesion as a strong
and robust universal method. The equipment needed to conduct
this test is universal (one hand and one knife or box cutter). The
simple nature of the test only allows for 5 values (0,2,4,6 and 10).
Sometimes people report the odd numbered values “in-between”
such as 3 or 7 and this is technically incorrect but the intention is
obvious. This is part of what makes ASTM D6677 a very robust
method. Its scoring rubric is qualitative and subjective but still
quite reproducible. The scoring rubric is very intuitive for the un-
initiated and is also more context independent than pull-off meth-
ods. That is not to say context and control values do not matter
for knife adhesion; They do. However, the non-expert decision
makers have a much more intuitive feel for a knife and hand pres-
ASTM D3359 admits to the following scope limitation with respect sure than the pounds-per-square-inch pull pressure normal to the
to film thickness: substrate used in pull-off methods.

“1.2 Test Method A is primarily intended to rate the adhesion In the context of paint specification and system selection it is im-
of coatings and coating systems greater than 125 μm (5 mils) in portant to note that adhesion only matters in so far as it is suffi-
total thickness, while Test Method B is primarily intended to cient to not be a hindrance to service life. That is, additional meas-
rate the adhesion of coatings and coating systems less than ured adhesion value beyond a certain value does not enhance
125 μm (5 mils) in total thickness. Test Method B is not performance or increase service life as has been examined thor-
considered suitable for films thicker than 125 μm (5 mils) oughly by Jaimal et Al[9]. In the same work just cited, it is important
unless wider spaced cuts are employed and there is an explicit to note that a correlation was found between corrosion prevention
agreement between the purchaser and seller.” [proxy for service life] and the degree of surface roughness in the
underlying metallic substrates. That would suggest that the con-
Tape adhesion is a robust qualitative method that, similar o knife ditions that lead to long paint service life also tend to lead to higher
adhesion offers a simple, qualitative assay to spot-assess adhe- adhesion values but the adhesion is the dependent variable in this
sion. Like all methods it is imperfect and suffers from a general view and not the causative agent for long service life once a cer-
lack of sensitivity and lack of discrimination between high adhe- tain minimum value is reached.
sion coatings and the standard itself says this explicitly.
Engineers should always use caution when selecting paints for a
Conclusions & Recommendations to Specifying specification. However they should especially use caution when
using paint manufacturer provided adhesion data. Adhesion
Engineers and Painted Asset Owners should not be a highly weighted factor to determine one coating’s
superiority to another for specifications. All major paint manufac-
Adhesion is a complex topic for which there are many variables turers are capable of producing very large pull-off adhesion values
driving the final outcome. When a coating’s adhesion is measured for public consumption due to the surface prep conditions in la-
it must be measured according to some methodology. The choice boratories and their access to Type IV, V and VI testing apparat-
of that methodology may have a large impact on how results area uses. These numbers are real measurements and say something
interpreted and decisions are made. It is crucial, to asset owners, about a coating’s potential maximum adhesion under ideal condi-
inspectors, engineers and paint manufacturers to understand the tions. However these numbers in no way reflect the numbers that
limitations of the testing methods they use. It is of even greater are likely to be measured on a real jobsite connected to the same
importance that results be compared “apples to apples” and re- said specification. This can be attributable to both the work con-
sults are not overinterpreted outside of their window of statistical ditions on site during real large scale surface preparation or vari-
power. ance in the testing methodology used on site versus the control
material laboratories use to test. Either way the adhesion pull data
An adhesion test conducted on a jobsite where coating has been submitted to engineers for specification acceptance is likely to
applied is likely not comparable to any testing conducted by the show much higher numbers than one would expect from the same
paint manufacturer. Both numbers are valid within their respective product in-field.
domains but both values may appear to be in conflict at first
glance. Where adhesion data is used to discriminate between coatings at
the specification level the following particular points should be
In the context of on-jobsite adhesion testing that may lead directly considered:
to decision making by the asset owner or engineer it is imperative
• Whether the specification is for in-shop fabricated new o The adhesion of the epoxy primer to the sub-
construction steel or is for on-site maintenance work is strate
not trivial when considering if reported manufacturer ad-
hesion numbers are comparable. The relative small o The adhesion of the intermediate coat to the
size of test panels virtually guarantees most data re- primer
ported as being taken over an SSPC-SP10 near white
metal blasted substrate was taken over a substrate o The adhesion of the topcoat to the intermedi-
blasted with angular grit to near SSPC SP-5 level. In ate coat.
reality, shop primed new construction steel is blasted to
a very clean level but usually with a mixture of shot and Facts such as say, the adhesion of the topcoat to the
grit rather than pure angular carbon steel or aluminum substrate are not relevant. However, all of this is cap-
oxide grit. This has an appreciable effect on profile tured holistically by pull-off testing of the entire system.
depth, angularity and adhesion to the subsequent coat- The coating will fail at the weakest bond point in the
ing[11]. On-site maintenance work is very unlikely to ever course of the test and as such gives you all limiting in-
achieve the same level of surface cleanliness as labor- formation about the system’s pull-off adhesion with one
atory blast operations or in-shop fabricators. number.

• In the context of ASTM D4541 adhesion pulls, high PVC • The use of different dolly adhesives, the use of scoring
(pigment volume concentration, a measure of the ratio around dollies (which ASTM D4541 allows for) and
of pigment to resin in a paint formulation) coating mate- other methodological variations allowed by the standard
rials inherently show lower numbers. The important ca- may have additional effects on ASTM D4541 or any
veat is that these lower numbers tend to be cohesive other methodology’s results.
failures rather than adhesive failures i.e. cured material
is left behind on the substrate that continues to protect. Closing Remarks
This effect is not negative as long as such materials
pass other reasonable measures of robustness. In fact Those who are involved in the writing of coatings specification lit-
cohesive failure by design is highly preferable to large erature, have input on such committees, and private engineers,
surface area adhesive failures seen in high cohesive consultants and specifiers working in the field actively should con-
strength materials such as elastomers, many water- sider these myriad of aforementioned variables when thinking
borne materials and thick film aromatic urethane/poly- about adhesion in coating’s during the course of their daily work
ureas. High PVC and/or low cohesive strength materi- and problem solving. Contained in the preceding paragraphs
als “breakaway coupling” materials include but are not there is no truly novel information though perhaps this may be the
limited to ceramic epoxies, glass flake epoxies, ethyl sil- first time such facts and principles have been complied into one
icate inorganic zincs and correctly water balanced ce- place for such an audience. The author hopes that by reading you
mentitious epoxy overlay materials. Moisture cured ure- have become more educated on the topic of industrial coatings
thane zincs are notably absent from this class [tending adhesion. Of all the m methodologies discussed, and all the vari-
to fail adhesively rather than cohesively]; speculatively ous ways to look at adhesion it is clear that there is no best meth-
due to their low zinc loadings. odology or absolute best current precise scientific understanding
of the topic.
• Pull-off and tape adhesion are useful methods but do
not simulate any meaningfully likely real world condi-
tion. That is not a criticism of them as methods per se,
but such things should be taken in context. Knife pick-
back adhesion also does represent much of a real world
R E F E R E N C E S:
condition (instances of scoundrels attacking our poor
1) ASTM B117-19 Standard Practice for Operating Salt
steel with bladed weapons are rare) but the method
Spray (Fog) Apparatus; ASTM International, 2019
does inherently approximate some form of gouge or cut
resistance assay and as such speaks to general film ro-
2) ISO 9227:2022; Corrosion tests in artificial atmos-
bustness. ASTM D6677 is explicit that a cut must be
pheres — Salt spray tests; International Organization
made to all the way to substrate and so for samples
for Standardization, 2022
where a cut cannot be made with a knife and hand pres-
sure that offers information about the adhesion of the
3) Epoxy Adhesion to Metals; Schmidt, Bell, Advances in
film in itself.
Polymer Science, 33–37 1975
• Adhesion testing for induvial coatings is interesting but
4) The Critical Humidity Effect in the Adhesion of Epoxy to
not particularly relevant where multi-coat systems are
Glass: Lefebvre, D.R.; Elliker, P.R.; Takasahi K.M.;
specified. Teka for example a typical industrial atmos-
Ragu, V.R.; Kaplan, M.L.; Journal of Adhesion Science
pheric service system, two coats of epoxy at 2-3 mils
& Technology 2012
each followed by a urethane topcoat at 2-3 mils. Only
the following facts area relevant at all with respect to
5) Paint Adhesion, Corrosion Protection & Interfacial
adhesion:
Chemistry; Dickie, R.A.; Ford Research Laboratory
Ford Motor Company; 1994
6) Theory of Adhesion & its Practical Applications, Butt,
M.A.; Chughati, A; Ahmad, J.; Ahmad, R.; Majeed, U.;
Kahn, I.H.; Journal of Faculty of Engineering & Technol-
ogy 2007-2008

7) Soft contact: Measurement and interpretation of Con-

tact Angles; Marmur, A; Soft Matter 2 2006

8) Resistance of solid surfaces to wetting by water; Wen-

zel, R.N.; Industrial and Engineering Chemistry 28 1936

9) Steel surface preparation prior to painting and its impact

on protective performance of organic coating; Jamali, S.
S; Mills, D. J. Progress in Organic Coatings, 77(12)
2014

10) Experimental Study of the Relation between Contact

Angle and Surface Roughness, Tamai, Y; Aratani, K;
Chemical Research Institute of Non-Aqueous Solutions
1971

11) Evaluating Performance Characteristics of Abrasive

Blasat Media, Quatman, C; KTA Tator,
https://kta.com/kta-university/abrasive-media-evalua-
tion/ 2018

12) ASTM D4541-22, ASTM International, 2022

13) ASTM D6677-22, ASTM International, 2022

14) ASTM D3359-22, ASTM International, 2022