The hitrrnatiowl Journal of Nautical Archaeology (1996) 25.

1: 3 8 4 5

Technical Communication
An improved tannin-based corrosion inhibitor-coating
system for ferrous artefacts

Worth Carlin and Donald H. Keith

Ships of Discovery, Corpus Christi Museum, 1900 N. Chaparral St. Corpus Christi, TX 78401,
USA

Introduction ered, most of the wrought-iron artefacts

Artefacts recovered from the sea in the latewere heavily encrusted with calcareous
1960s and early 1970s were re-examined in marine deposits. Following mechanical
1993 to determine their present condition, removal of the encrustation, they were
cleaned electrolytically to remove corrosive
to evaluate the effectiveness of the original
treatment, and to re-treat those exhibiting chlorides, then washed with deionized
symptoms of active corrosion. Originally water, dehydrated in alcohol, air-dried and
conserved by mechanical cleaning, electro- immersed in molten microcrystalline wax
lytic reduction, and coating in micro- (Hamilton, 1976). To the authors’ knowl-
edge, this is the oldest, largest, collection
crystalline wax, after 20 years the artefacts
were in good condition, but the micro- of wrought-iron artefacts conserved by
crystalline wax coating was thin and no electrolytic reduction.
longer impermeable to air penetration. In An inspection in 1993 determined that
some places, slow corrosion was taking most of them were in good condition after
place. More than 350 16th-century 20 years, with less than 5% of the collection
wrought-iron items were re-treated with showing any trace of ‘rust’. The overall
the experimental tannic acid solution de- excellent condition of the collection attests
scribed below followed by a tannin-derived to the effectiveness of this method of con-
coating which is impervious to air and servation. Investigators who have been
moisture. convinced that it is preferable to strip iron
to bare metal, removing all corrosion
The Padre Island 1554 Plate Fleet products, should reconsider the advan-
collection tages of treatments which strive to preserve
A collection of 16th-century artefacts sal- and pacify an artefact’s ‘rind’ as well as its
vaged from two ships wrecked in the Gulf metallic core.
of Mexico in 1554 is currently housed in During the passage of more than 20
the Corpus Christi Museum of Science and years, many of the objects had been ex-
History (Arnold, 1992). They were recov- posed to a variety of less-than-optimum
ered at intervals between 1967 and 1973, environmental conditions; part of the col-
and conserved at the Texas Archaeological lection was used in a travelling exhibit
Research Laboratory at the University of and other artefacts were loaned to other
Texas at Austin from 1969 to 1976 (Arnold museums for temporary exhibit and some
& Weddle, 1979; Olds, 1975). When recov- changes had taken place. The original
I057 2414/96/010038 + 08 $18.0010 (01996 The Nautical Archaeology Society
W.CARLIN & D.H.KEITH:CORROSION INHIBITOR-COATINGSYSTEM

microcrystalline wax coating was very thin, preserving the stable ‘rind’ of corrosion
and susceptible to water vapour penetra- products, which although derived from
tion and some new corrosion was begin- the core has become an entirely different
ning. Each artefact was examined closely material. An additional complication is
and it was noted that a few of the smaller that this rind prevents any coating applied
items, such as bolts, had developed small to the exterior from reaching the still-
cracks perpendicular to their long or reactive metallic core and leaves the
wrought dimension. The cracks indicated artefact susceptible to on-going corrosion.
that a dimensional change was occurring Because the 1554 Plate Fleet artefacts
consistent with continuing oxidation. still have their outer corrosion layers in-
In order to quantify the physical differ- tact, the decision was made to re-treat the
ence between surface and core, samples collection with a corrosion inhibitor and to
were taken from a bolt (41WY3-1583.0).A provide a coating to minimize the penetra-
dental pick was used to separate a flake tion of air and moisture. The decision to
from the outer oxidized surface of a treat the entire collection with tannic acid
bolt, removing a fragment approximately was made because it would be relatively
0.32 cm thick by 1.0 cm wide by 2.0 cm in simple to accomplish--provided an effec-
length. A sample of the underlying metal tive tannic acidsurfactant system and a
was obtained by drilling into the bolt removable vapour barrier coating could be
below where the flake had been removed. developed.
These samples were analysed by X-ray
powder diffraction (XRPD) to determine Tannic acid-based corrosion inhibitors
their composition. On a comparative basis Tannic acid and its derivatives have been
the surface sample (Fig. la) contained less used as corrosion inhibitors for ferrous
than one-third of the elemental iron found materials for many years (Knowles &
in the underlying material (Fig. lb), with White, 1958; Matamala et al., 1990) al-
corresponding increases in the oxide con- though recently some investigators have
tent of the surface material. In this study, raised questions about its effectiveness
samples taken for XRPD analysis were (Morcillo et al., 1992; Ashton, 1993: 20).
also analysed by X-ray fluorescence (XRF) The chemistry of tannic acid and its
in the atomic number range 9-92. No derivatives is complex; the material itself
chloride (C1- ) concentration greater than varies greatly from one source to another.
0.001% was detected among the trace Generally speaking, ‘tannic acid’ refers to
elements found. It was deduced that corro- hydrolysable tannins which are essentially
sion was continuing, as the cracks had galloyl esters of glucose. The hydrolysis of
developed since the last inspection. tannic acid yields primarily gallic acid and
While the most expedient and direct glucose. Tannic acid’s corrosion inhibiting
method of cleaning and stabilizing iron mechanism has not been determined due
artefacts from the sea is to remove to the complexity of the large tannic acid
mechanically every trace of oxidized ma- molecule. Some investigators have sug-
terial, leaving only metallic iron behind, gested that anodic polarization of the metal
this can be akin to ‘throwing the baby out occurs, forming a passive film (Parkins &
with the bath water’ if the outermost layer Pearce, 1966: 649-650; Rosenberg, 1987:
of corrosion products preserves evidence 15). Considering the complexity of the
of manufacture and use, as well as unique system, ‘anodic polarizatiodfilm-forming’
markings. The resulting dilemma for the is a generality that can be accepted because
archaeological conservator is to pacify there are ample data to support the claim
the metallic core while at the same time that tannic acid is a corrosion inhibitor.
39
NAUTICAL ARCHAEOLOGY,25.1

File: Z04638.RAW Scan: 2-75/.02/1/#3651, Anode: CU Zero = 0.0

(a) Magnetite
750

500
rn
0
c
s
0" Maghemite 2

250 1 Iron

20 30 40 50 60 70
2-Theta
5le: Z04639.RAW Scan: 2-75/.02/1/#3651, Anode: CU Zero = 0.0
(b)

750 Magnetite

Iron
500
m
U
C
0

Maghemite
250

0
20 30 40 50 60 70
2-Theta
Figure 1. A comparison between the composition of samples taken from the metallic core (b) and rind
of a wrought-iron forelock bolt (a) as determined by XRPD analysis. Peak heights reflect the
proportion of the element or compound present. The graph shows that the amount of iron in the
metallic core is about three times that of the rind, while the amount of magnetite/maghemite in the rind
increases accordingly. X-axis numbers are arbitrary. The repetition of peaks is a reflection of different
energy levels. Maghemite= 1, magnetite=2, iron=3.
40
W. CARLIN & D. H . KEITH: CORROSION INHIBITOR-COATING SYSTEM

Sixteen to eighteen percent aqueous/ compatible with tannic-acid solutions but

alcohol solutions of tannic acid have been the presence of chlorides preclude their
used to coat ferrous artefacts. However consideration. The properties of 35 com-
tannic acid solutions have poor metal mercial surfactants were reviewed and
wetting properties, particularly on arte- seven chloride-free materials were selected
facts such as those in the 1554 Plate Fleet for testing. The surfactants tested were
collection which have been coated with products of PPG Industries, Specialty
microcrystalline wax. Theoretically, the Chemicals, Chemicals Group. Although
wax can be removed. But in practice, the several manufacturers produce similar sur-
object must be immersed in a hot volatile factants these were selected because of the
solvent (such as Stoddard’s solvent). For investigators’ familiarity with their product
small artefacts this could be accomplished line. The initial compatibility test was to
safely enough in a well-vented hood with determine which chloride-free surfactants
ample protection from sources of ignition. were soluble in 18% aqueous tannic acid.
For large artefacts the inhalation and In this simple test, 0.25 g of each surfactant
flammability hazards of microcrystalline were mixed with 100 gm of aqueous ( 1 8%)
wax removal are too serious to be tannic acid solution and the degree of
contemplated. solubility was observed.
Corrosion of the objects indicated that As revealed in Table 1, only one of the
the protective wax coating was no longer tested surfactants, Mazon 40@, was com-
patible with the tannic acid solution. To
fully effective. There was a possibility that
they could be treated without removing the ensure that the surfactant incompatibility
microcrystalline wax coating, but this demonstrated in Table 1 was not a func-
made it imperative to design a tannic-acid tion of tannin-acid concentration, the test
inhibitor with good penetration and wet- procedure was repeated with 8% tannic-
ting properties. Although the surface was acid concentrations. The results were the
hydrophobic and poorly wetted by either same. To determine the effectiveness of the
water or the aqueous/alcohol solution of Mazon 40@ surfactant on the ‘wetting’
tannic acid, when tested with an aqueous properties of aqueous 18% tannic-acid
solution containing a suitable surfactant, solutions, a series of simple tests was per-
the wax-treated artefact surface was suffi- formed by adding known quantities of the
ciently porous to allow penetration of the surfactant to the tannic-acid solution and
tannic acid. Other tests demonstrated that then placing a 0-2-ml drop of the mixture
a fresh, thick coat of microcrystalline wax on a rusted flat iron plate. The spread or
is completely impermeable, even to the coverage of the solution on the surface of
surfactant solution. the plate is a reliable measure of the effec-
tiveness of the surfactant’s performance.
A new tannic acid coating The tannic-acid solution contained 18%
Prior to determining if a surfactant could tannic acid (technical grade, water
increase the ‘wetting’ of aqueous tannic soluble), 11% denatured ethanol with
acid it was necessary to determine what the remainder distilled water. To this
types of surfactant were compatible with solution Mazon 40@ was added at various
concentrated tannic acid solutions. A se- concentrations.
cond mandate was that the surfactant The data in Table 2 reveal that the
should contain no chloride ion due to the addition of the surfactant to the aqueous
inherent corrosive nature of that molecule. tannic acid solution increases the surface
For example, surfactants derived from coverage and the penetration of the inhibi-
alkali metal sulphates or sulphonates are tor into the oxide-metal interface. With
41
NAUTICAL ARCHAEOLOGY, 25.1

Table 1. Compatibility of various chloride-free surfactants with 18% aqueous

solutions of tannic acid

Trade name* Chemical type Results

Macol NPG@ POE (6) nonyl phenol Insoluble

Macol OP-10 PS@ POE octyl phenol ether Reacts
Macol OP-30 (70)@ POE (30) octyl phenol Reacts
Mazawet 30° Nonionic surfactant Reacts
M u o n LDA@ Lauramine oxide Reacts
Mafo 13@ Amphoteric surfactant Reacts
Mazon 40@ Caustic coupling agent Soluble

*PPG Industries, Inc. Speciality Chemicals, Chemicals Group, 3938 Porett

Drive, Gurnee, IL 60031.

Table 2. Area of rusty iron plate covered by 0.2 ml Table 3. Area of rusty iron plate covered by 0.2 ml
of tannic-acid solution in square centimetres (cm2) of tannin test solution, in square centimetres (cm2)

Surfactant concentration (wt %) 18% tannic

18% tannic Fertan@ acid w/l.5%
acid surfactant

Test No. 1 3.1 3.8 11.9 12.8 Test No. 1 3.5 4.5 13.4
Test No. 2 2.8 3.6 12.3 14.0 Test No. 2 2.8 4.1 14.2
Test No. 3 3.3 3.7 12.6 13.4 Test No. 3 3.3 4.6 14.5

heavily-corroded artefacts where a large

intention was to paint the tannic acid
percentage of the particles are iron oxides,
it is necessary to have the corrosion in- solution on the artefacts, applying one or
hibitor in contact with the metal/oxide more coats as necessary. It quickly be-
came apparent that due to the presence in
interface to prevent or impede further cor-
the collection of many tubular objects
rosion. Therefore, it is essential to have an
inhibitor solution that thoroughly pen- (cannons, breech chambers) a better
etrates the corrosion products to inhibit method of application would be to im-
and stabilize the corrosion process. A merse them in the solution. For treat-
comparison of the ‘wetting’ properties of ment, they were grouped by size in order
tannic acid, tannic acid with surfactant to minimize the quantity of solution
added, and Fertan (a widely-used commer- necessary. Small artefacts were treated
cial tannin-based inhibitor/coating), are simultaneously in small tanks, while large
presented in Table 3. ones were treated individually in larger
tanks. The immersion tanks were kept
Artefact treatment covered to minimize evaporation and
The collection contained objects of vari- oxidation of the solution. The fact that
ous shapes and size ranging from small they absorbed a surprising quantity of
nails and bolts to wrought-iron, breech- solution was indicated by an incremental
loading cannons 0-15-0.20 m in diameter lowering of the solution level each time
and up to 2-6m in length. The initial an artefact was withdrawn.
42
W. CARLIN & D. H. KEITH: CORROSION INHIBITOR-COATING SYSTEM

A new vapour-barrier coating Repeat the previous step at least three

The new vapour-barrier coating, known times. After discarding the final wash,
as MOP-30, developed at the Corpus let the resin dry (by air or under a heat
Christi Museum, is derived from a lamp) until it becomes a hard, wax-like
combination of the surfactant MACOL substance. This material, dissolved in
OP-30 (PPG Industries) and tannic acid. three parts (by weight) of denatured
MACOL OP-30 is an ethoxylated octyl ethanol, is the protective coating,
phenol containing thirty ethylene oxide MOP-30. The first decantation of the
units. The composition of MOP-30 has reaction mix, which had been retained,
not been studied; presumably it is an ester is performed after 48 hours. The recov-
or similar condensation product. It is a ered resin is added to future prepara-
good coating for artefacts that will not be tions. A batch with 180 g tannic acid
exposed to the elements because its high should yield approximately 171 g of
ethylene oxide content provides a good MOP-30 product. The MOP-30 is
oxygen barrier, a good moisture barrier applied to the artefact which has been
and excellent compatibility with tannate previously covered with at least two
films. Easily removed with alcohol, the coats of tannic acid, making sure that
coating will not withstand prolonged each coating is dry before applying the
immersion in water. The effectiveness of next.
any tannic acid treatment is counteracted ( 2 ) Mix the tannic acid-surfactant solu-
and eventually negated by exposure to tion. Solution number 1: To 3080 ml
water, but the addition of a vapour- of deionized water add 340ml of
barrier coating can enhance the pro- (denatured) ethanol. Mix in 758 g of a
tection offered by tannic acid. Judging commercial grade of completely water/
from the literature, those investigators alcohol soluble tannic acid. Note:
who have denigrated the usefulness of tannic acid is very dusty, the weighing
tannic acid as a corrosion inhibitor were and mixing should be performed in
really questioning its effectiveness in wet a well-ventilated area with the techni-
environments. cian wearing a dust mask and eye
protection.
Treatment procedure and formulas Solution number 2: To 380ml of
The treatment procedure is as follows: deionized water add 38 g of PPG
(1) Synthesize the protective coating Industries Mazon 40°, a thick, viscous
(MOP-30) well in advance due to the alkyl glucoside. The surfactant dis-
difficulty of preparing it in large solves slowly with moderate stirring.
quantities. Combine Solutions number 1 and
Dissolve 180 g of tannic acid in number 2. Agitate gently, as the mix-
720 ml distilled water. To the resulting ture has a tendency to create foam.
solution, slowly add 92g of MACOL (3) Clean the artefact with a stiff nylon
OP-30 with minimum stirring. Let the brush to remove dust or loose
reaction product, a brown resinous particles.
syrup, separate from the solution for (4) Immerse the artefact in the tannic acid
at least one hour. Slowly decant the solution for 25-35 minutes. Tubular
supernate and set aside for further objects such as guns or breech cham-
processing. Add a volume of deionized bers should be gently rocked to prevent
water equal to the resinous mass, stir the formation of air pockets.
gently and let sit for one hour. Decant (5) Allow artefact to air dry for 24-48
and discard the aqueous supernate. hours.
43
NAUTICAL ARCHAEOLOGY, 25.1

(8) Artefacts destined for display should

be given an additional coat consisting
of a mixture of 40% powdered graphite
and the MOP-30 coating. Best results
are obtained when the coating is
applied with a small sponge roller,
which imparts a soft, dull finish
(Fig. 2).

Conclusion
This coating system for the protection of
Figitre 2. Wrought-iron breech-loading guns on ferrous artefacts not exposed to the ele-
display in the Corpus Christi Museum following
treatment.
ments consists of three components: (a)
A tannic acid solution with increased
penetration properties (Fig. 3). (b) A clear
(6) Vigorously brush the dry, coated arte- coating derived from tannic acid that pro-
fact with a nylon-bristled vegetable vides good compatibility when applied
brush to remove any loose oxidized over a tannic-acid inhibitor coating. Ident-
tannate residue. ified as MOP-30, this coating has good
(7) Apply the protective MOP-30 coating vapour-barrier properties. (c) The addition
with a brush or roller. Two coats is of graphite pigment into the MOP-30 coat-
normally sufficient. The coating dries ing produces a gray matt appearance.
quickly, leaving a glossy black surface Applied over the clear protective coating,
on the artefact. this mixture imparts an aesthetically-

Figure 3. Schematic comparison of the wetting characteristics of a drop of a normal 18% aqueous tannic
acid solution and a drop of tannic acid solution with surfactant three seconds after contact with the
surface of an artefact. The rind is composed of stable corrosion products.
44
W. CARLIN & D. H. KEITH: CORROSION INHIBITOR-COATING SYSTEM

pleasing finish. Like MOP-30, the coatingmission to research, develop and apply
can be removed in an alcohol bath. this treatment to artefacts from the Padre
The three-coating system is based on aIsland 1554 Plate Fleet collection. Special
time-proven inhibitor (tannin). It is revers-
thanks go to Linda Zitting of the Corpus
ible, and it penetrates permeable, thin Christi Museum history section, who is
layers of microcrystalline wax. Although undoubtedly more familiar with the 1554
the coating system is suitable for ferrous
collection than any other person. Messrs
artefacts having a porous structure, such as
Charles Holifield and Michael Feeney
cast or wrought iron, it is not recom- obtained the XRPD analysis for us
mended for artefacts that will be exposedgratis. With regard to the artefacts them-
to the elements. selves, we would like to thank Dr D. L.
Hamilton and all the people involved in
Acknowledgments the treatment of this collection more than
The authors would like to thank Rick 25 years ago. The excellent condition of
Stryker, Director of the Corpus Christi the hundreds of artefacts are a testa-
Museum and J. Barto Arnold 111, Texas ment to their skill, inventiveness, and
State Underwater Archaeologist, for per- determination.

References
Arnold 111, J. Barto & Weddle, Robert S., 1969, The Nautical Archaeology of Padre Island. New York.
Arnold 111, J . Barto. 1992, Shipwreck!-the 1554flota exhibit. IJNA, 21.4: 343--355.
Ashton, J., 1993, Article Review: Tannic acid: does it work? Australian Institutefor the Conservation of
Cultural Murerials h’ationul Newsletter, 47.
Hamilton, D. L.. 1976, Conservation of Metal Objects From Underwater Sites: A Study in Methods.
Austin.
Knowles, E., & White, T., 1958, The protection of metals with tannins. Journal of the Oil and Color
Chemists Association, 41: 10-23.
Logan, J. A., 1989, Tannic acid treatment. CCI Notes 915. Ottawa.
Matamala, G., Smeltzer, W. & Benavente, R., 1990, Pine tannin rust converter for steel protection by
painting. Surjuce Modijicmtiori Technologies 111. Ottawa.
Morcillo, M . et al. 1992. Corrosion of rusted steel in aqueous solutions of tannic acid. The Journal of
Science and Engineering-Corrosion, 48.1 2: 1032-1039.
Olds, Doris, 1974, Texas Legacy ,from the G u v Austin.
Parkins, R. N. & Pearce, A. S., 1966, Inhibition of corrosion of mild steel by tannins. National
Association (?f Corrosion Engineering, Second International Congress on Metal Corrosion, 646-650.
Pelikan, J. B., 1966, Conservation of iron with tannin. Studies in Conservation, 11: 109-1 14.
Rosenberg, S. P., 1987, The inhibition of aqueous corrosion of iron by gallic acid. Corrosion
Australasia, 11-15.
Ross, T. K. & Francis, R . A,, 1978, The treatment of rusted steel with mimosa tannin. Corrosion
Science, 18: 35 1-36 1.
Shreir, L. L., 1964, Tannins to control corrosion. New Scientist, 4 0 3 332-333.

45