UvA-DARE (Digital Academic Repository)

Coating strategies for the protection of outdoor bronze art and ornamentation

Brostoff, L.B.

Publication date
2003
Document Version
Other version

Link to publication

Citation for published version (APA):

Brostoff, L. B. (2003). Coating strategies for the protection of outdoor bronze art and
ornamentation.

General rights
It is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s)
and/or copyright holder(s), other than for strictly personal, individual use, unless the work is under an open
content license (like Creative Commons).

Disclaimer/Complaints regulations
If you believe that digital publication of certain material infringes any of your rights or (privacy) interests, please
let the Library know, stating your reasons. In case of a legitimate complaint, the Library will make the material
inaccessible and/or remove it from the website. Please Ask the Library: https://uba.uva.nl/en/contact, or a letter
to: Library of the University of Amsterdam, Secretariat, Singel 425, 1012 WP Amsterdam, The Netherlands. You
will be contacted as soon as possible.

UvA-DARE is a service provided by the library of the University of Amsterdam (https://dare.uva.nl)

Download date:06 nov 2022
C O A T I N GG STRATEGIES
forr the
PROTECTIONN of OUTDOOR I
BRONZEE ART and ORNAMENTATION
byy LYNN B. BROSTOFF
 Coatingg Strategies
forr the
Protectionn of
Outdoorr Bronze Art
andd Ornamentation
-- i J I M * I U M a L 22*53 3
 Coatingg Strategies for the Protection of
Outdoorr Bronze Art and Ornamentation

ACADEMISCHH PROEFSCHRIFT

terr verkrijging van de graad van doctor

aann de Universiteit van Amsterdam

opp gezag van de Rector Magnificus

prof.. mr. P. F. van der Heijden

tenn overstaan van een door het college voor promoties ingestelde

commissie,, in het openbaar te verdedigen in de Aula der Universiteit

opp donderdag 24 april 2003, te 12:00 uur

door r

Lynnn Beth Brostoff

geborenn te Pittsburgh (USA)

Promotiecommissie e

Promotor:: prof. dr. E. R. de la Rie

Overigee commissieleden: prof. dr. H. Schenk

prof.. dr. J. W. Verhoeven

prof.. dr. N. H. Tennent

prof.. dr. R. van der Linde

prof.. dr. O. Chiantore

prof.. dr. G. P. Bierwagen

dr,, H. A. Ankersmit

Faculteitt der Natuurwetenschappen, Wiskunde en Informatica

Coverr illustration: Reclining Figure, Henry Moore, 1957, Carnegie Museum

off Art, Pittsburgh, PA.

Coverr design: Lynn Brostoff and Terrie Churchin

 Thee research in this work was carried out at the National Gallery of Art,
Washington,, DC, and sponsored through funding and support from the US
Departmentt of the Interior, National Park Service, and National Center for
Preservationn Technology and Training (NCPTT). Other funding sponsors include
thee J. Paul Getty Foundation, and generous matching funds from the National
Galleryy of Art. Prior to the work at the National Gallery or Art, investigation into
thee interaction of benzotriazole with copper corrosion minerals was conducted at
thee Sherman Fairchild Center for Objects Conservation, The Metropolitan Museum
off Art, New York, NY, with the generous support of the L.W. Frohlich
Foundation.. Any opinions, findings, conclusions and recommendations expressed
inn this paper are those of the author and do not reflect the views of the National
Galleryy of Art, the US Department of the Interior, the National Park Service, the
Nationall Center for Preservation Technology and Training, or the Frohlich
Foundation. .

Materiall in this thesis has been included in the following presentations and
publications: :

L.. B. Brostoff, "Investigation into the Interaction of Benzotriazole with Copper

Corrosionn Minerals and Surfaces" in Metal 95. Proceedings of the International
ConferenceConference on Metals Conservation, ICOM-CC Metals Working Group, 25-28
Septemberr 1997, Semur-en-Auxois, France, ed. Ian D. MacLeod, et al. (James &
James,, Ltd., London, 1997), 99-108.

L.. B. Brostoff and E. René de la Rie, "Research into Protective Coatings Systems
forr Outdoor Bronze Sculpture and Ornamentation" in Metal 95. Proceedings of the
InternationalInternational Conference on Metals Conservation, ICOM-CC Metals Working
Group,, 25-28 September 1997, Semur-en-Auxois, France, ed. Ian D. MacLeod, et
al.. (James & James, Ltd., London, 1997), 242-244.

Lynnn Brostoff and E. René de la Rie, "Conservation Treatments: Methods, Options

andd Research," Paper Presentation at The Conservation of Outdoor and Indoor
Sculpturee and Monuments Through a Conservator's Eye, Brookgreen Gardens,
Murrellss Inlet, South Carolina, 21-23 August 1997.

Lynnn Brostoff and E. René de la Rie, "Research into Protective Coating Systems
forr Outdoor Bronze Sculpture and Ornamentation, Phase I" (PTTPublications No.
1997-03,, NCPTT, Natchitoches, Louisiana, 1997).

Lynnn B. Brostoff and E. René de la Rie, "Chemical Characterization of

Metal/Coatingg Interfaces from Model Samples for Outdoor Bronzes by Reflection-
Absorptionn Infrared Spectroscopy (RAIR) and Attenuated Total Reflection
Spectroscopyy (ATR)" in Metal 98. Proceedings of the International Conference on
MetalsMetals Conservation, ICOM-CC Metals Working Group, 27-29 May 1998,

ll
Draguignan,, France, ed. W. Mourey and L. Robbiola (James & James, London,
1999),, 320-328.

L.. Brostoff, T. Shedlosky and R. de la Rie, "Research into Protective Coating

Systemss for Outdoor Bronze Sculpture and Ornamentation, Phase II"
(PTTPublicationss No. 1997-03, NCPTT, Natchitoches, Louisiana, 1999).

L.. Brostoff, T. Shedlosky and R. de la Rie, "Research into Protective Coating

Systemss for Outdoor Bronze Sculpture and Ornamentation, Phase IN"
(PTTPublicationss No. 1997-03, NCPTT, Natchitoches, Louisiana, 2000).

Lisaa Ellingson, Tara Shedlosky, Lynn Brostoff, René de la Rie, and Gordon P.
Bierwagen,, "Evaluation of Coating Systems for Protecting Bronze Using
Electrochemicall Impedance and Accelerated Test Methods," Poster Presentation at
Electrochemicall Society Meeting, Session Zl, Toronto, Canada, May 2000.

Taraa J. Shedlosky and Lynn B. Brostoff, "The Application of Digital Image

Analysiss to Performance Assessment of Coatings for Outdoor Bronze and
Copper,"" Abstracts, AIC 28lh Annual Meeting, 8-13 June 2000, Philadelphia, PA,
12-13. .

Lynnn B. Brostoff, Tara J. Shedlosky and E. René de la Rie, "External Reflection

Studyy of Copper-Benzotriazole Films on Bronze in Relation to Pretreatments of
Coatedd Outdoor Bronzes," in Tradition and Innovation: Advances in Conservation,
Preprints,, IIC 18th International Congress, 10-14 Oct. 2000, Melbourne, Australia
(IIC,, London, 2000), 29-33.

uu
 Thiss dissertation is dedicated to Anita Brostoff, the one
personn above all others who has been my inspiration and
modell and endless source of support; her life force
permeatess all my endeavors.

WhatWhat we call the beginning is often the end

AndAnd to make an end is to make a beginning.

T.S.T.S. Eliot

111 1
 ??

IV V

^^ s
- --;;-'-' J*1"" i il rJT' im"M
Contents s
page e
1.. The problem of protecting bronze in outdoor exposures
Abstract t
1.11 Introduction 1
1.22 The effect of outdoor environments on bronzes 2
1.33 The engineering problem 4
1.44 The conservation problem: current attitudes, practices, and research 7
1.55 The research problem 12
1.66 Conclusions 15
Referencess 16

2.. Performance of 29 coatings on two types of copper alloy

substrates s
Abstract t
2.11 Introduction 21
2.22 Experimental methods 23
2.2.11 Description of the coatings 23
2.2.22 Sample preparation 26
2.2.33 Weathering 27
2.2.44 Methods of evaluation 29
2.2.55 Sources of error 30
2.33 Results and discussion 31
2.3.11 Substrate characterization 31
2.3.22 The coatings 34
2.3.33 Coating quality 35
2.3.44 Corrosion analysis after weathering 36
2.3.55 Coating performance on bronze, accelerated weathering 38
2.3.66 Coating performance on bronze, natural outdoor weathering 40
2.3.77 Coating performance on copper roof, accelerated weathering 41
2.3.88 Coating performance on copper roof, outdoor weathering 43
2.3.99 The influence of thickness and adhesion on coating
performancee 44
2.44 Conclusions 49
Referencess 51
Platee I 55
3.. Performance of five coatings on four types of copper alloy
substrates s
Abstract t
3.11 Introduction 57
3.22 Experimental methods 58
3.33 Results and discussion 63
3.3.11 Substrate characterization 63
3.3.22 Corrosion on uncoated substrates after accelerated weathering 67
3.3.33 Coating quality 68
3.3.44 Coating performance 70
3.3.55 Influence of dry film thickness and adhesion on coating
performancee 73
3.44 Conclusions 78
Referencess 80
Platess II-V 82

4.. Electrochemical Impedance Spectroscopy (EIS) of select

coatingss on bronze
Abstract t
4.11 Introduction 85
4.22 Experimental methods 88
4.33 Results and discussion 89
4.3.11 Phase I samples 89
4.3.22 Phase II samples, set A 90
4.3.33 Phase II samples, set B (after accelerated weathering) 96
4.44 Conclusions 98
Referencess 99

5.. Chemical characterization of the bulk coating and the

metal/coatingg interface
Abstract t
5.11 Introduction 101
5.22 Experimental methods 103
5.33 Results and discussion 106
5.3.11 Acrylics 106
5.3.22 Acrylic urethanes 116
5.3.33 Waterborne acrylic urethanes 127
5.3.44 Microcrystalline wax blend 131

VI I
5.44 Conclusions 132
Referencess 135

6.. The role of benzotriazole (BTA) in bronze protection

Abstract t
6.11 Introduction 137
6.22 Background 139
6.33 Experimental methods 142
6.44 Results and discussion 145
6.4.11 Investigation into the interaction of benzotriazole with copper
corrosionn minerals and surfaces 145
6.4.1.11 Identification of CuBTA derivatives by FTIR 145
6.4.1.22 Identification of copper-BTA powder reaction products 146
6.4.1.33 Reactions of BTA on copper/copper salt surfaces 151
6.4.1.44 Implications for BTA reactions with copper corrosion
mineralss and surfaces 154
6.4.22 External reflection study of copper-benzotriazole films on
bronzee in relation to pretreatments of coated outdoor bronzes 155
6.4.2.11 Reflection-absorption infrared spectroscopy 155
6.4.2.22 Film thickness 157
6.4.2.33 Film growth 158
6.4.2.44 EIS 161
6.4.2.55 Implications for coated outdoor bronzes 161
6.55 Conclusions 162
Referencess 164

7.. Summary and concluding remarks

Abstract t
7.11 Summary of results 167
7.22 Low maintenance vs. high maintenance coatings 169
7.2.11 The wax question 169
7.2.22 Waterborne acrylic urethanes 170
7.2.33 Acrylics 171
7.2.44 Acrylic urethanes and other coatings 172
7.2.55 Multi-part coatings 173
7.33 Suggestions for future work 173
Referencess 174

Acknowledgmentss 175
Samenvattingg 177

Vil l
 aa

VH1 1

M * * * ^ f f i ee .^.1
 11
Thee problem of protecting bronze in
outdoorr exposures

Abstract Abstract
Thiss chapter defines the problem of protecting bronze art and
ornamentationn in outdoor exposures according to engineering, conservation, and
conservationn research points of view. The discussion includes a brief overview of
atmosphericc corrosion of bronze.

LÏ.LÏ. Introduction
Outdoorr bronze sculpture and ornamentation are a seamless part of our
urbann landscape. Without the imposition of museum walls, these objects are a vital
partt of our everyday experience, subtly influencing the way we appreciate our
artisticc and cultural heritage. Hidden to the commuter and passerby, however, a
dailyy micro-drama is being played out on the metal surface: materials struggling to
maintainn their identity against the forces of nature. And so the sustained presence
off bronze sculpture and ornamentation in our environment is anything but
effortless.. In fact, the maintenance of outdoor bronze art and ornamentation
requiress an enormous amount of consideration and energy on the part of the
caretakerss of this heritage. It is to this effort that the present research is addressed.
Accordingg to surveys by the public organization SOS! (Save Outdoor
Sculpture!),, tens of thousands of outdoor bronze monuments in the United States
aree in need of attention [1], This represents only a small portion of bronze
sculpturee and ornamentation on an international scale that require preservation.
Besidess vandalism and unintentional damage by the public, pollution in the
environment,, especially acid rain, has been linked to detrimental and disfiguring
corrosionn on bronzes globally [2,3,4,5,6,7]. That this poses a serious threat to our
artisticc patrimony is without question. The most common approach to these
problemss is to apply a coating, the most important class of which is clear, organic
coatings.. While many traditional coatings, such as wax, are adequate in certain
situations,, decreasing funding to institutions responsible for the maintenance of

11
ChapterChapter I

outdoorr sculpture instills a pressing need for new, low maintenance coating
solutions. .
Thee research presented in this dissertation approaches the problem of
protectivee organic coatings for outdoor bronzes in terms of coating design and
systemm failure. These concepts are considered within the framework of current
practicess and needs in the conservation field. In this study, new and old coating
systemss have been tested on various model substrates in simulated accelerated
weatheringg and natural outdoor weathering conditions. The model samples were
evaluatedd by visual and chemical means. Results of this research may lay the
groundworkk for field trials and optimization of some new coatings for practical use
byy conservators of outdoor sculpture. Ultimately, it is hoped that this work will be
aa guide to conservators in the development of new strategies for the protection of
outdoorr bronze sculpture and ornamentation.

1.2.1.2. The effect of outdoor environments on bronzes

Byy the third millenium BC it was discovered that addition of certain ores,
specificallyy tin and arsenic, to naturally occurring copper not only hardens the
metal,, but minimizes the release of oxygen bubbles during casting and allows
difficultt castings to succeed. Stretching from the Bronze Age to the present, this
amazingg and valuable material has been associated not only with functional tools
andd vessels but with artwork. Due to developing metallurgical technology and
changess in the availability of different metals over the centuries, many
compositionss of bronze alloys have historically been used to produce artwork. It is
thereforee difficult to classify bronze as one material, but bronzes do have
significantt properties in common, foremost of which is the predominance of
copper—andd the noble behavior of copper—in the alloy composition. In the world
off art and ornamentation, besides relatively pure copper materials, common types
off bronzes in the 19* century include copper-lead-tin-zinc alloys, and, increasingly
inn the 20th century, silicon bronze alloys. This study utilizes a copper-lead-tin-zinc
bronzee that is broadly representative of 19th century alloys.
Itt should be recognized that not only techniques of alloying and casting
havee changed over time, but also the impact of outdoor exposure for bronze objects
hass changed dramatically since the industrial revolution. The major factors in
environmentall exposure before this era consisted of oxygen, water, and, near
oceans,, chloride salts. Exposure of copper alloys to air immediately results in the
formationn of an oxide layer. Alloying alters the oxidation rate of copper, generally
slowingg it in the case of tin addition, and speeding up oxidation in the case of
nickell addition. A slow-growing, natural oxide film is normally thin, smooth, and
compact,, and may protect the metal surface to a fair degree from non-aggressive
environments.. The oxidized bronze surface may appear golden to reddish to
blackishh brown, depending on the alloy composition, and is commonly referred to
inn artistic circles as a patina. Green copper carbonate minerals have only rarely
beenn reported in patinas from natural atmospheric exposure. Exposure to
chlorides,, on the other hand, readily causes the localized formation of various

22
 TheThe problem of protecting bronze

greenn colored copper minerals and is known to have a marked destabilizing effect
onn copper alloy patinas [8,9,10,11,12,13,14,15].
Sincee the advent of urban-industrial environments, outdoor exposure of
materialss has entailed contact with corrosive pollutants in the atmosphere, such as
sulfurr dioxide, a by-product of burning of fossil fuels. In recent decades, the urban
environmentt has been associated with so-called acid rain, a precipitation cocktail
off sulfates, nitrogen oxides, and ozone, as well as more minor quantities of
ammonia,, ammonium sulfate, other organics, and soot [7,16]. Records indicate
thatt sulfur dioxide emissions in Europe and the United States peaked in the 1970s
andd have since been declining; in Europe nitrates and ozone concentration continue
too increase, while a general decrease in these emissions has occurred in the United
Statess since the 1980s [16,17]. In general, however, levels of SO2 and NOx
emissionss remain high today throughout the world, and there are indications that
thee current political climate in the United States will reverse this trend and allow
increasedd emissions again. Chlorides have also been found to be a significant
contaminantt in urban environments, a major source of them being the steadily
increasingg winter salting practices [17,18].
Althoughh the complex reaction mechanisms between metal surfaces and
acidd rain remain incompletely understood, the basics of this process may be
summarized.. Acidity is introduced to the metal/patina surface via condensation
and/orr rain deposition, forming a layer of water that contains conductive salts,
includingg corrosive species. It is also known that hydrolysis and oxidation of SO2
andd NOx occur in the atmosphere, yielding sulfuric and nitric acids directly on the
bronzee surface. The acidity of modern precipitation is normally less than pH 4.0.
Ass the precipitation brew evaporates on the metal surface and from within the
patina,, the mixture becomes increasingly concentrated, and, by extension,
increasinglyy acidic. In combination with oxidizing pollutants, the acid
precipitationn induces corrosion at significant rates at the metal surface [19].
Duringg extended wetting/washing cycles and wind exposure, there is a high
potentiall for dissolution of the metal along with erosion, dissolution, and migration
off select components of the patina. New corrosion minerals are precipitated in
pits,, on the surface, and underneath existing minerals. In this way a patina that is
generallyy porous, permeable, and soluble to different degrees in the environment
formss and reforms. This process thus effectively disarms the metal's natural
abilityy to form a protective mineral layer [5,7,20,21].
Thee question of whether patinas are protective is complex. The
corrodibilityy of the metal surface in any one environment depends upon multiple
factors,, including alloy composition, manufacture and exposure history.
Comparedd to copper roof material, Weil relates the apparently greater corrodibility
off many cast bronze monuments directly to a cast vs. a wrought structure, rather
thann to alloying [21]. In particular, the quality of the bronze casting is of crucial
importance,, since inhomogeneous and porous casting greatly increases
susceptibilityy to corrosion. The patinated surface itself introduces additional
factorss into the complex electrochemistry of environmental interactions. These
factorss include the chemical stability, solubility, and hygroscopic behavior of the
patinaa salts in the environment, as well as the physical structure of the patina.

33
ChapterChapter 1

Importantt physical properties of the patina layer include the density, porosity,
homogeneity,, and coverage [3,5,17,22].
Studiess and general observations have established that bronze corrosion
patinass created by acid rain are not only disfiguring in terms of loss of detail and a
unifiedd surface, but are also unstable. The dominant corrosion minerals found in
thee patinas of outdoor copper and its alloys are: cuprous oxide (cuprite); copper
hydroxyy sulfates, primarily brochantite and antierite; and copper hydroxy
chlorides,, primarily paratacamite and atacamite [18,22]. While the copper sulfates
renderr bronze surfaces light green overall or in patches, complex cycles of wind,
rainn washing, and particulate deposition often result in unsightly black crusting and
streakingg over the surface. The existence of black copper sulfides in the dark
crustss has not been established, although it is often assumed.
Thee formation of raised black areas adjacent to green areas on outdoor
bronzess suggests to some corrosion scientists that carbonaceous and other
particulatee material in the black areas play a crucial role in terms of creating local
cathodicc protection, which accelerates dissolution and corrosion in the green areas
[2,17,21],, A study of bronze corrosion on the Capitoline Hill in Rome showed that
theree was a large difference in the relatively low corrosion rates observed in
compactt green, green-brown, or black areas on bronzes vs. relatively high rates in
porous,, pale white-green areas. These researchers concluded that the pale green
corrosionn crust does not significantly contribute to the corrosion resistance of metal
[23].. In addition, chloride corrosion is most often visible as severe light green
pittingg or pockmarking overall. Corrosion rates are normally quite high in this
typee of pitting corrosion. Chloride salt layers have also been shown to exist
underneathh the sulfate patina [20],
Thee manufacture of bronze artworks since the 19th century has usually
includedd the production of an artificial patina on the metal surface through the
applicationn of chemicals to obtain certain coloristic effects [13,24]. These mineral
patinass vary greatly in nature, have generally not been investigated, and may or
mayy not have protective qualities. For more detailed discussions of bronze
corrosionn in outdoor settings, see excellent treatments of this subject in the
literaturee [2,5,16,17,18,22,25].

1.3.1.3. The engineering problem

Thee problem of bronze and copper alloy corrosion in the outdoor
environmentt is at base defined by the electrochemical conversion of metal into
metall ions. For an engineer, the problem is fundamentally a metallurgical one:
howw to limit the rate of overall and localized corrosion of the pure metal, whether
thiss involves bridge ornamentation, a roof, or a car. That is, corrosion is primarily
aa structural problem, as in the case of intergranular corrosion of metals, which
weakenss the integral fabric of the material. Corrosion of metal surfaces,
particularlyy in the form of pitting, has structural as well as aesthetic repercussions.

44
 TheThe problem ofprotecting bronze

Aestheticc issues have importance commercially for consumer goods like cars, but
thesee issues are mostly limited to maintaining a new appearance.
Onee approach to achieving corrosion resistance in outdoor environments is
throughh protective organic coatings. The industrial importance of protective
coatingss has given rise to the field of coating science, to which the paint and
automotivee industries have devoted much energy. The following is a brief outline
off some principles and current thinking in this field [17,26,27,28,29,30,31].

1.3.1.1.3.1. Surface preparation and coating adhesion

Itt is an axiom in the coatings industry that good surface preparation is the
essentiall starting point for good coating performance. This means a clean surface,
ass well as one with good wetting and adhesive properties for the intended coating.
Itt is well known that contaminants such as oil, dirt, or corrosion products on the
metall surface compromise coating adhesion. In addition, both dirt and corrosion
mineralss on the metal surface are hygroscopic. Residual corrosion salts at a
coating/metall interface are an important source of soluble salts in solutions that
mayy form in this region once moisture penetrates through the coating; the salts may
thenn initiate corrosion beneath the coating. Scrupulous washing and degreasing
mayy remove these contaminants; in cases of corroded surfaces, industrial practice
recommendss removal of a layer of metal through an abrasive procedure such as
sandblasting. .
Inn order to achieve sufficient wetting during application of a coating, the
surfacee tension of the coating must be lower than that of the metal surface.
Cleaningg of the metal raises the surface tension so that good wetting can occur.
Goodd wetting leads to good contact and spreading, resulting in superior adhesion
andd film quality. Roughening of the surface increases surface area and may
improvee adhesion of the coating, although a too-rough surface may leave voids
underneathh a coating with imperfect wetting characteristics [32]. Blasting
techniquess employed for cleaning off corrosion layers also have the potential to
leavee behind stresses and deformations in the metal to various degrees. These
stressess may also induce corrosion. As always, a balance must be struck.
Chemicall means of surface preparation also exist and include: etching with acidic
mediums;; reaction with chemicals such as chromates and phosphates, the salts of
whichh may passivate the surface; cathodic protection by a zinc primer film;
treatmentt with adhesion promoters such as silanes; and treatment with corrosion
inhibitorss such as benzotriazole. These are all important industrial practices.
Goodd adhesion of a coating to the metal is of course a fundamental aspect
off protection of the underlying metal surface from contact with corrosive media.
Adhesionn has often been differentiated as dry or wet adhesion [33,34], Dry
adhesionn may be defined as a net phenomenon, equal to the force necessary to de-
adheree a defined area of coating under standard room conditions. Wet adhesion
mayy be defined more particularly as the force necessary for water to displace the
coatingg at the metal interface. This displacement is dependent on the dominant
typee of bonding here, i.e., strong chemical bonding or relatively weak physical
bonding.. Poor wet adhesion is significant in all instances of moisture, oxygen, and

55
ChapterChapter 1

saltt penetration to the metal/coating interface. This is, of course, the case with all
organicc coatings, which are permeable to water and oxygen to some degree, and
aree imperfect structures that may become damaged or deteriorated [35]. It should
bee remembered that although we can attempt to optimize adhesion in a system,
chancee events that occur during coating application, such as dirt deposition, air
bubblee formation, or change in humidity, can compromise good adhesion as well
ass affect measured adhesion values.
Thee displacement forces involved in wet adhesion are difficult to measure
directly.. Dry adhesion is thus a starting point for assessment of an intact coating
systemm with the potential for corrosion protection. Adhesion is also dynamic, i.e.,
mayy change over time depending upon extraneous forces within the system, such
ass stress and the presence of moisture. Measurement of dry adhesion after
weatheringg provides an indication of disruption of coating to metal adhesive forces
thatt may have taken place, and thus indirectly reflects wet adhesion properties of a
systemm [36].

1.3.2.1.3.2. Coating quality and durability

Goodd coating quality, i.e., the relative lack of defects, is another essential
buildingg block in the corrosion protection of a coating system. Coating defects not
onlyy affect aesthetic appearance, but also mitigate performance in aggressive
environments.. This is obvious in the case of gross defects, such as bubble holes
andd cracks. Another clear example is pinholing, a common defect that may arise
fromm solvent popping/air entrapment during coating application. These small air
holess allow air and moisture ingress and may result in local corrosion pits. The
defectt commonly known as orange peel is a phenomenon arising from poor flow
comparedd to relatively fast solvent evaporation characteristics. This sets up
tensionss that result in a bumpy surface upon drying, which may be clearly seen in
reflectedd light. The orange peel surface translates to thickness variation on a
micro-scale.. This fluctuation has important bearing on performance since it will
sett up local differences in permeability as well as in electrochemical and osmotic
pressuree gradients. Other common coating defects include seeds and cratering:
littlee wells and dimples created by specks of dust and other contaminants on the
surfacee during coating application. These present similar problems in terms of
locall thickness fluctuation [27].
Inn order to achieve good coating quality, scrupulous practices and technical
skilll are of utmost importance. Nevertheless, normal spray and brush application
off coatings inevitably results in a range of defects. Multiple, thin layers of
coatings,, applied at right angles if possible, are usually recommended to offset
defects.. A thin wax topcoat may be quite beneficial in this respect, while not
appreciablyy adding to overall thickness. However, operators must be careful that
layerss do not exaggerate thickness variations nor leave residues that can lead to
interlayerr delamination. Layering may also aggravate residual stresses built up
duringg film formation, and may be associated with adhesion loss and/or cracking
andd crazing [37,38]. It is interesting to speculate whether orange peel may also be
associatedd with stress-related loss of adhesion or cohesion.

66
 TheThe problem ofprotecting bronze

Assumingg that a coating is applied in an optimal manner, coatings vary

greatlyy in terms of their inherent protective properties and stability in outdoor
environments.. Several coating properties that are desirable for good outdoor
durabilityy include: 1) good ultraviolet light (UV) resistance, 2) sufficient flexibility
too withstand thermal stresses, 3) chemical compatibility with the metal surface in
orderr to achieve good adhesion and good chemical resistance to the environment,
andd 4) relatively low oxygen and water permeability. In the first case, superior UV
protectionn may be provided foremost by pigments; for clear coatings, UV
protectionn is provided by innate chemical saturation as well as UV absorbers and
hinderedd amine light stabilizers. Minimum flexibility is not a well defined
quantity,, but some flexibility should be built into the chemical and/or physical
structuree of coatings in order to avoid stress build up during application and
service,, which may result in debonding and cracking. Chemical compatibility with
thee metal may be promoted, for example, by the inclusion of polar hydroxy! and
carboxyll groups in a polymer that are attracted to the oxides and hydroxides on the
metall surface. These same groups may increase water permeability, however,
throughh a solvency effect. Low permeability, which is perhaps most important, is
bestt achieved in a coating by high density, such as in crosslinked polymers.
Ideally,, crosslink density should be as homogeneous as possible. However, very
highh crosslink density may also contribute to stress buildup in the film.
Unfortunately,, it is not easy or even possible to engineer all desirable
qualitiess into one coating material. It is more feasible to design a coating system
thatt combines desirable qualities, such as in 3-part systems where the primer
coatingg promotes good adhesion to the metal, the middle coat provides durability
andd other desired qualities, and a top coat aids in low permeability [39]. This
commonn industrial approach was utilized in the research design of the National
Galleryy of Art's project.

1.4.1.4. The conservation problem: current attitudes, practices,

andand research

1.4.1.1.4.1. Aesthetics vs. protection

Everyy outdoor bronze work of art is a unique case scenario, with its own
historyy in terms of materials, manufacture, and outdoor exposure. Artists, art
historians,, connoisseurs, and conservators have developed keen sensitivity to the
uniquenesss of each work of art, and yet no consensus exists about what limits are
acceptablee for the changes in appearance that inevitably occur in outdoor settings.
Heree we enter the domain of aesthetics, where the ground shifts with changing
tastess in different societies and different eras. This is not the territory of the
scientistt or, theoretically, the conservator, although decisions regarding
conservationn treatments that intimately involve aesthetics inevitably fall to the
conservator.. Ideally, treatment and aesthetic decisions for works of art go hand in
hand,, and should be arrived at jointly by conservators, curators, the artist (if alive),

77
ChapterChapter 1

andd art custodians. This does happen, but a lack of consensus about the aesthetics
off outdoor sculpture continues to fuel controversy surrounding treatment options
[40]. .
AA camp of opinion still exists that supports non-intervention. This is a
disputedd approach, especially given the obvious effects of man-made pollution on
copperr alloys. In many cases, the entire form of an artwork may become obscured
underr a veil of contrasting streaks and dots as corrosion progresses in outdoor
environments.. In a sense, this is nature's graffiti, clearly overwriting the artist's
intent.. Indeed, in the vast majority of these cases, such changes, which may
penetratee quite deeply into the metal, are completely unforeseen by the artist.
Inn less aggressive environmental conditions, the overall surface texture and
colorationn of a work may corrode evenly, and change may be viewed in some
circless as an acceptable course of events. On the other hand, an historical view of
outdoorr bronzes, such as that presented by P.D. Weil, asserts that any value
ascribedd to matte green, mineralized outdoor bronze surfaces is misplaced, unless
thatt surface was purposely created by an artist, as it may be on some modern
sculptures.. Weil maintains that, artistically and aesthetically, the original lustrous,
semi-transs lucent browns and green-browns that are natural to bronzes should be
prizedd and protected by proper maintenance [41]. Furthermore, we must consider
changess in appearance that inevitably result from protective treatments themselves,
suchh as color changes from repatination, darkening from saturating coatings, and
shininesss from some synthetic polymer coatings.
Realistically,, the rates of loss suffered by bronzes in outdoor urban
environmentss in terms of surface texture and detail are unacceptable by any
standards.. This is because exposure to even mildly polluted outdoor urban
environmentss subjects these materials to chemical and electrochemical instability,
i.e.,, progressive deterioration. Therefore, the conservation approach to metal
objectss in outdoor exposures should more properly be based on a different set of
criteriaa than that relied upon for objects protected in indoor museums. Traditional
notionss of non-invasive or minimal intervention may have to be abandoned, or in
anyy case redefined, if objects are to be properly protected and preserved.
Ultimately,, decisions concerning surface preparation, coating choice, and
maintenancee planning will be based on subjective as well as objective criteria.
Withh curatorial and scientific input, conservators must find a difficult middle
groundd of an aesthetically acceptable option that does not compromise the object in
termss of stability in the harsh setting of outdoor urban environments. It seems
obviouss that it is unrealistic to expect that one solution may exist to universally
satisfyy the aesthetic requirements and protective function of coatings for outdoor
bronzes.. Yet, these expectations are often encountered in the field. The coupling
off corrosion science, polymer science, and interfacial chemistry, along with
aesthetics,, renders the subject of coatings for outdoor bronze sculpture and
ornamentationn complex and difficult to approach for conservators and scientists
alike.. Nevertheless, it is hoped that a combined approach may help practitioners to
identifyy and reach a better understanding of the relative importance of various
factorss in outdoor coating appearance and durability appropriate to individual
situations. .

88
 TheThe problem of protecting bronze

1.4.2.1.4.2. Current methods in the treatment of outdoor bronze

Conservationn applications may be distinguished from industrial
applicationss by aesthetic considerations, as discussed above, and perhaps more
importantlyy by physical and chemical differences. While industrial coatings are
designedd for fresh metal surfaces that are stripped, sandblasted, and primed, the
surfacess of artworks and historical objects generally cannot be prepared in this
fashionn without unacceptable damage. Thus, conservation applications are
handicappedd from the outset by typically requiring coatings to perform on
inhomogeneous,, contaminated and naturally or artificially corroded surfaces.
Becausee of this handicap, it is difficult to draw up a list of desirable
propertiess of coatings for outdoor bronzes, since the requirements vary with each
individuall work of art. In general, coatings should be durable within a proposed
maintenancee period, and should have a viable and safe method for removal once
theyy start to fail, i.e., must be reversible by some method. Although it is desirable
nott to alter the original surface of an object, this is a controversial point, since the
originall surface is not strictly retrievable once oxidation and corrosion occur. In
orderr to retain—or regain—original appearance and surface detail as much as
possible,, clear coatings are usually applied relatively thinly; they should thus be
UVV stable and capable of providing protection well below the usual manufacturer's
recommendedd thickness. It should be noted that pigmented coatings are
increasinglyy tolerated for use over existing corrosion crusts; for this application,
conservatorss most often prefer to tint clear coatings themselves. The following
discussionn outlines current approaches and methods in thee conservation of outdoor
bronzee [41,42,43].

1.4.2.1.1.4.2.1. Cleaning

Approachess to cleaning already weathered objects before coating vary

greatlyy among practitioners, both historically and geographically. This stems from
availablee resources and differing aesthetic approaches, as well as an incomplete
understandingg of the relative stability of the surface and corrosion patina, both
beforee and after treatments. In the United States, the most common method for
cleaningg weathered bronze and copper alloys entails washing with an aqueous
detergent,, usually followed by light to heavy scrubbing with bristle brushes or
nylonn pads to remove loose corrosion, plus degreasing in an organic solvent.
Steamm cleaning and/or solvent cleaning are often necessary to remove old wax and
otherr organic coatings, which may be deeply embedded.
Moree aggressive cleaning is also common and may include abrasion with
bronzee wool or a form of air-abrasive blasting to remove all or a portion of the
corrosionn patina [44]. One blasting medium currently favored by North American
professionalss is ground walnut shells. This method is usually employed so that the
loosee and pale green corrosion are removed, leaving behind a dark green mineral
surfacee [45]. Other blasting media in use in similar manners include low- to high-
pressuree water [46], ground maize, and wheat starch. Less favored at this time, but
stilll in use internationally, are low pressure, fine mesh glass beads, which are
normallyy used to clean down to the metal surface [47]. Other media that have been

99
ChapterChapter 1

proposedd for blasting but have found little application and development include
C0 2 ,, ice blasting, plastic beads, and baking soda [47,48]. The use of poultices and
alkalii cleaning agents to remove old green and black patinas has also been reported
[49].. Recently, reports and studies of laser cleaning of corroded bronze may be
foundd in the literature [50].
Afterr cleaning procedures, the surface is often re-colored through
applicationn of chemicals, i.e., the work is repatinated, as previously mentioned. It
iss known, for example, that application of an ammonium sulfide solution to an
existingg green patina adjusts the color to a darker, more yellow-brown. Potassium
permanganate,, applied with heat, turns a green patina to dark brown or even black.
Thesee techniques chemically alter the surface; they have arisen out of foundry
practicess for the most part, and have not been examined as a chemical component
off the coating system.

1.4.2.2.1.4.2.2. Coating

Coatingg technologies in current use in conservation have generally been

borrowedd from traditional and industrial applications without accompanying
researchh into adaptation and optimization of materials and methods. There is,
however,, well-voiced accord among conservators that these materials and
techniquess often fail to meet the full range of requirements and conditions
encounteredd in the field.
Traditionall coatings that continue to be popular for use on outdoor bronzes
include:: (1) oils, such as lemon oil, paraffin oil, linseed oil and castor oil; and (2)
tintedd or clear natural waxes, such as camauba and beeswax mixtures [51]. These
typess of coating treatments easily saturate existing patinas, causing darkening.
Dryingg oils are reported to last up to one year and are increasingly insoluble with
age.. Waxes require frequent maintenance, ideally about every six months, and
varyy widely in makeup and formulation. Commercial paste waxes favored on
outdoorr sculpture, such as Butcher's Paste Wax, are basically natural wax mixtures
withh high camauba content and some synthetic wax and solvent added. Carnauba
iss a vegetable wax that imparts desirable hardness and luster to the wax. The
naturall waxes, particularly animal waxes, may contain some free acids which can
potentiallyy attack the metal surface, and are also subject to acid hydrolysis at the
esterr group [18,52,53].
Manyy wax recipes favored by professional conservators in North America
andd elsewhere since the 1980s are based on microcrystalline waxes. These
syntheticc waxes are straight chain, cyclic, and branched products that are refined
fromm petroleum. The small crystal structure inherent in these materials allows the
waxx to approach an amorphous state. Individual conservators often make up their
ownn wax formulation based on microcrystalline wax, and typically mix in other
syntheticc wax ingredients, such as polyethylene wax, to achieve a relatively high
meltingg point while retaining good application properties.
Typically,, wax coatings are either brushed or sprayed on and buffed. In
eitherr case, the wax may be applied either hot, i.e., after heating of the metal
surface,, or cold. Hot waxing allows the wax to penetrate more deeply into the

10 0
 TheThe problem of protecting bronze

surface,, especially in cases of high porosity or existing patinas. In both cases, but
particularlyy hot waxing, conservators have ample experience that shows wax
coatingss are not entirely reversible [54,55,56]. Wax residue is inevitably left
behindd to some degree, and complicates future treatments that are incompatible
withh wax. This has also been demonstrated in some analytical studies [18,52].
However,, waxes remain the coatings of choice in many situations. This is due in
largee part to their accessibility and ease of use. While waxing application is often
perceivedd to require only minimal training, this is also a controversial point.
Addingg pigments to wax, particularly earth browns, is a very common practice,
mainlyy for aesthetic reasons.
Cellulosee nitrate, sold as an Agateen product, continues to have some
popularityy although it has been commonly observed to have poor outdoor
durability.. By far the most widely used modern coating material on outdoor
bronzess is an acrylic lacquer coating called Incralac®*. The International Copper
Researchh Association (INCRA) developed this unpatented coating in 1964 for
generall use on copper alloys. It is also common to use pigmented acrylic inpaints
inn conjunction with acrylic coatings to unify color variations in an existing
corrosionn patina [57].
Incralacc has been reported to generally last two to three years on outdoor
bronzes,, and up to nine years with topcoats of wax and regular maintenance
[58,59].. The most common problem associated with Incralac in normal application
iss orange peel [60], The shiny appearance of Incralac, which is often felt to be
undesirable,, can be moderated by the addition of matting agents and/or wax
topcoats.. Other problems commonly noted with Incralac may include eventual
embrittlementt of the film, resulting in delamination and spalling.
Off great concern with Incralac are inconsistent reports of difficulty with its
removall over time. There has been one report of yellowing and crosslinking of an
Incralacc film over a gilded bronze statue [61]. However, many conservators say
theyy have not experienced this problem. Related to this perceived problem,
conservationn professionals have sometimes voiced concern about batch
inconsistency.. It is unclear whether scattered reports of reversibility problems with
Incralacc coatings are inherent in the polymer or are related to particular conditions
off its use, such as application over certain artificial patinas, inclusion of pigments
inn the coating, and use over other metals such as gold or iron. This will be
exploredd during further discussions, particularly in Chapter 5.
Theree is one final point of consensus among conservation professionals: the
needd for adequate maintenance planning along with any treatment of outdoor
bronzess [42]. The lack of planning and funding in place or earmarked for future
maintenancee may well dictate the type of coating that is appropriate for
consideration.. In addition, the inconsistent hiring of professionals and workers to
maintainn an individual work may compromise adequate coating maintenance.
Theree is a great need in this field to educate local government appointees or hired
architecturall firms, who are often responsible for decisions about maintenance,

** Hereafter referred to simply as Incralac.

11 1
ChapterChapter 1

regardingg the importance of having local advisory committees with conservators to

aidd in planning of care for outdoor monuments. Such committees are necessary
alsoo to ensure that proper bidding and hiring procedures are adhered to, so that well
trainedd professionals with proper credentials are given the responsibility to treat
outdoorr works, which is often not the case.

1.5.1.5. The research problem

1.5.1.1.5.1. History of the National Gallery of Art research

Inn November 1994, in preparation for launching this research project,
relevantt issues and conditions governing the conservation application of protective
coatingss for outdoor sculpture were brought into focus at an international meeting
organizedd by the National Gallery of Art in Washington, DC. The meeting
includedd conservators, scientists, and representatives of Save Outdoor Sculpture!
(SOS!).. Several issues were highlighted in this forum, as well as in subsequent
consultationss with leading figures in the fields of outdoor sculpture conservation,
corrosion,, and coating science.
Themess that echoed throughout these discussions may be summarized as
follows.. While high maintenance coatings such as annually maintained waxes are
reasonablyy successful in many applications, there exists a very real need to develop
loww maintenance coatings, i.e., coatings with improved durability and corrosion
protection.. Furthermore, volatile organic compounds (VOC) laws, Occupational
Safetyy and Health Administration (OSHA) restrictions on solvent use, and safety
issuess of public exposure to harmful substances, increasingly affect conservation
treatments,, especially in terms of the removal of old coatings. It is thus
worthwhilee to look to industry for new, more durable coatings or coating systems.
Itt is also important to devise a method for testing of coatings that has general
validityy for outdoor sculpture. It was generally acknowledged that the limited
researchh on coatings for outdoor metal sculpture and ornamentation has been
inconclusivee and gives little basis for forming coating strategies.
Att the same time, continued research into wax and acrylic resin coatings
suchh as Incralac is valuable to better understand the necessary conditions for their
optimization.. These common coatings remain the standards against which new
coatingss can be assessed. It is clear that evaluation of coatings prior to this
researchh project has most often been based on visual inspection of samples
followingg accelerated testing, so that reasons for coatings failures are unknown.

1.5.2.1.5.2. Aims and design of the National Gallery of Art research

Researchh priorities and problems were identified from the above
discussionss and from a review of current literature. The ultimate goal of the
Nationall Gallery's research project was to impartially evaluate a series of model
systemss and to develop recommendations for better coatings and coating methods

12 2
 TheThe problem ofprotecting bronze

basedd on a materials science approach. To this end, the investigation aimed at

identifyingg reasons why protective organic coatings work and fail on pristine and
patinatedd bronze surfaces, as well as designing improved coating systems for
possiblee trial in the field. The study paid particular attention to the interplay of
adhesion,, thickness, coating quality, and pretreatment with the corrosion inhibitor
BTAA in overall performance. Other factors, such as coating composition, were
alsoo considered.
Withh the aim of identifying and testing new coating systems with high
performancee potential for the conservation of outdoor bronzes, both old and new
materials,, and issues of surface preparation and removability were considered. The
specificc selection of coatings and substrates arose from the belief that the
successfull development of new coating strategies in conservation will ultimately
arisee from a tailored design approach that includes the combination of materials
intoo one coating system to achieve desired properties, and the selection of materials
basedd on an understanding of physical and chemical interactions between the
particularr bronze substrate at hand and the selected coating. Therefore, different
typess of model coatings were chosen, and each was varied by surface preparations,
additivess and combination with other coatings. The different coating systems were
thenn examined on different types of copper alloy substrates.
Thee research project was divided into two phases. In Phase I [62], a total of
twenty-ninee coating systems were weathered on two types of substrates: polished,
castt bronze and 50-year-old copper roof. These are referred to throughout as
"Phasee I samples." In Phase II [63], five of the coating systems deemed most
viablee were chosen for further study in a new round of weathering on an expanded
rangee of substrates ("Phase II samples"). Methods used in this phase sought to
repeatt as well as to clarify results of Phase I. Weathering in Phases I and II was
bothh natural, on the roof of the National Gallery of Art, and accelerated in the
laboratoryy using UV exposure and an acid rain-type solution, coupled with cyclic,
broadd range exposures of temperature and humidity. This type of accelerated
weatheringg was modeled on methods used in the automotive industry.
Experimentall details and results are presented for each phase of the research in
theirr respective chapters.
Thee results described in the following chapters provide information which
mayy help to select and tailor real coating systems, as well as roughly predict their
relativee performance. In particular, measurements of thickness and adhesion
propertiess of the samples provide useful data, and warnings, about coating
propertiess and imperfections resulting from application methods. Broad
correlationn of these results with visual and chemical analyses of bulk coatings and
aa variety of interfacial surfaces from weathered, model coating systems provides
insightt into the complex mechanisms involved in coating failure on outdoor
bronzes. .

1,5.2.1.1,5.2.1. Choice of substrate

Noo one substrate can represent a general situation encountered in outdoor

bronzee conservation. This is true in terms of the metal alloy composition, as

13 3
ChapterChapter I

discussedd above, as well as its manufacture and condition. For experimental

purposes,, however, it is of utmost importance to limit and define important
variabless that occur in the field as much as possible. For this reason, model
systemss must be designed that offer reasonable correlation to the actual world as
welll as enough control over variables to allow conclusions to be made from the
study. .
Thee model systems designed for use in Phase I of the study consisted of
twoo types of substrates, which were chosen to represent typical and, at the same
time,, extremes of surface conditions. One model substrate was a polished, cast
bronzee alloy of Cu <85%), Pb (5%), Sn (5%), Zn (5%). This alloy composition is
fairlyy representative of 19th century bronze alloys and is a good general example of
bronzee as encountered in monuments found in North America [18]. The substrate
wass cast in a mould in order to reproduce a normal level of material stresses and
microstructuree characteristics. This type of manufacture carries a high degree of
imperfectionss which could not be eliminated, such as a certain level of porosity.
Samples,, which appeared extraordinarily porous were rejected, but due to the cost
off the bronze, most material was utilized in the study, imperfections included. The
relativelyy large size of the samples allowed averaging of results across the surfaces.
Thee bronze plaques were polished to mirror finish in the foundry.
Althoughh this is a less typical finish for outdoor monuments, the polished surface
offeredd certain experimental advantages. First, the polished surface allowed an
evenn base from which to judge the coating appearance and performance. Second,
thiss surface preparation allowed coating thickness measurement with reasonable
accuracy,, as well as the possibility of adhesion testing and delamination for
interfaciall examination. A polished surface is probably the least forgiving in terms
off coating application and performance, since adhesion is difficult and initial
corrosionn potential is high. In the field, these disadvantages are in fact often
encountered.. Experimentally, these are aids to quicker differentiation of coating
performance. .
Thee copper roof substrate was chosen as an excellent example of a natural,
brochantitee patina. These patinas cannot be reproduced exactly in a laboratory,
eitherr compositionally or morphologically, but are very similar to those formed on
bronzess in terms of the corrosion products and impurities [22]. In addition, the
patinaa formed on pure copper is relatively even, which is necessary for
experimentall purposes.
Inn Phase II, this repertoire of substrates was expanded to include an
artificiallyy patinated cast bronze, and a walnut shell-blasted copper roof patina.
Thee chemical patina was chosen to represent a typical foundry patina. The cleaned
copperr roof patina was included as a typical substrate produced by conservation
treatmentt as favored at the present time. By studying a small number of coatings
onn four representative substrates, Phase II of the research focused on the
importancee of substrate preparation and substrate composition in the choice of a
protectivee coating.

14 4
 TheThe problem ofprotecting bronze

1.5.2.2.1.5.2.2. Choice of coatings

Inn addition to coating materials in current use, new materials that appeared
too meet general conservation criteria were chosen for study. Both thermoset and
thermoplasticc polymeric materials were considered, since mechanical means of
coatingg removal remain viable, if underdeveloped, options for coatings that cannot
bee dissolved. Wax and Incralac coatings included in the study serve as
benchmarks,, and were varied with different surface preparations in order to test
theirr effect. The wax coating chosen for this study is broadly representative of
microcrystallinee wax preparations that are used by conservators. This choice was
alsoo based on results of a preliminary study by Stromberg, which indicated that
twoo microcrystalline wax mixtures typically used in the United States had superior
performancee characteristics in outdoor weathering on bronze compared to two
naturall wax mixtures [64]. The wax was also used as a topcoat in some systems in
orderr to gauge its effect on overall performance and aesthetic qualities. Topcoats
off wax are typically used over Incralac.
Neww model coating systems were designed according to coatings industry
principless of three-part systems, in which an adhesion-promoting layer is first
applied,, followed by a main and topcoat. The design of coating systems for
outdoorr bronzes with optimal adhesion characteristics to the bronze/corrosion
patinaa substrate was considered essential for maximizing the protection a coating
cann afford. The main coat of model systems was chosen with consideration of
impermeability,, durability, removability, and appropriate physical and aesthetic
properties.. Coupling agents, corrosion inhibitors, such as benzotriazole, and other
appropriatee materials were considered for the adhesion-promoting layer.

1.6.1.6. Conclusion
Itt should be recognized that a "safe" coating choice for all outdoor bronze
situationss will necessarily remain the elusive wish of conservators and caretakers
off art. A more realistic approach to the complex problem of protecting outdoor
bronzess should include a basic understanding of the complexity of environmental
andd metallurgical phenomena in aggressive chemical environments, coupled with
thought,, expertise, research, and, above all, decision-making appropriate for each
individuall work of art. It is hoped that the research methodology presented here
willl serve as an aid in understanding and pinpointing "weak links" in coated
bronzee systems in outdoor exposures, and thus in making informed decisions about
appropriatee coating strategies.

15 5
ChapterChapter 1

References References

11 Cf. Heritage Preservation Update (Fall 1999), 14.

22 Andrew Lins and Tracy Power, "The Corrosion of Bronze Monuments in Polluted
Urbann Sites: A Report on the Stability of Copper Mineral Species at Different pH
Levels,"" in Ancient and Historic Metals, ed. D. A. Scott, et al. {Getty
Conservationn Institute, 1994), 119-151.
33 T. E, Graedel, "The Corrosivity of the Atmosphere: Past, Present, and Future," in
DiaIogue/89-TheDiaIogue/89-The Conservation of Bronze Sculpture in the Outdoor Environment,
ed.. Terry Drayman-Weisser (NACE, Houston, TX, 1992), 13-32.
44 Arthur Beale, "Conservation of Outdoor Sculpture: An Overview" in
Dialogue/89-TheDialogue/89-The Conservation of Bronze Sculpture in the Outdoor Environment,
ed.. Terry Drayman-Weisser (NACE, Houston, TX, 1992), 3-12.
55 Luc Robbiola, Christian Fiaud, and Stèphane Pennec, "New Model of Outdoor
Bronzee Corrosion and Its Implications for Conservation," Preprints, The
InternationalInternational Council of Museums Committee for Conservation (1COM-CC)
10th10th Triennial Meeting, Vol. II, Working Group 22, 22-27 August 1993,
Washington,, DC (Paris, 1993), 796-802.
66 Maurizio Marabelli, "The Monument of Marcus Aurelius: Facts and Comments,"
inn Ancient and Historic Metals, ed. D. A. Scott, et al. (Getty Conservation
Institute,, 1994), 1-20.
77 T. E. Graedel, "Copper patinas formed in the atmosphere—I-III," Corrosion
ScienceScience 27, 7 (1987), 639-658; 721-740; 741-69.
88 R. J. Gettins, "Patina: noble and vile," in Art and Technology: A Symposium on
ClassicalClassical Bronzes, ed. S. Doeringer, et al. (MIT Press, Cambridge, MA, 1970),
57-72. .
99 Marcel Pourbaix, "Electrochemical Corrosion and Reduction," in Corrosion and
MetalMetal Artifacts—A Dialogue Between Conservators and Archaeologists and
CorrosionCorrosion Scientists, ed. B. Floyd Brown, et al, (National Bureau of Standards
Speciall Publication 479, July 1977), 1-16.
100 T. Stambolov, "The Corrosion and Conservation of Metallic Antiquities and
Workss of Art," Central Research Laboratory for Objects of Art and Science
(Amsterdam,, N.D.).
111 Tom Chase, "Chinese Bronzes: Casting, Finishing, Patination, and Corrosion," in
AncientAncient and Historic Metals, ed. D. A. Scott, et al. (Getty Conservation Institute,
1994),, 85-118.
122 D. J. G. Ives and A. E. Rawson, "Copper Corrosion, I-IV," J. Electrochem. Soc.
109,, 6 (1962), 447-451; 452-457; 458-462; 462-466.
133 Phoebe Dent-Weil, "A Review of the History and Practice of Patination," in
CorrosionCorrosion and Metal Artifacts—A Dialogue Between Conservators and
ArchaeologistsArchaeologists and Corrosion Scientists, ed. B. Floyd Brown, et al. (National
Bureauu of Standards Special Publication 479, July 1977), 77-92.

16 6
 TheThe problem ofprotecting bronze

144 Phoebe Dent-Weil, "Patina: Historical Perspective on Artistic Intent and

Subsequentt Effects of Time, Nature, and Man," in Sculptural Monuments in an
OutdoorOutdoor Environment, ed. V.N. Naudé (Philadelphia Academy of Fine Arts,
Philadelphia,, PA, 1985), 21 ff.
155 Helena Strandberg and Lars-Gunnar Johansson, "Some Aspects of the
Atmosphericc Corrosion of Copper in the Presence of Sodium Chloride," J.
Electrochem.Electrochem. Soc. 145, 4 (April 1998), 1093-1100.
166 Helena Strandberg, Perspectives on Bronze Sculpture Conservation: Modelling
CopperCopper and Bronze Corrosion, Doctoral dissertation, Góteborg University,
Swedenn (1997).
177 Robert Baboian, "A Corrosion Scientist's Summary of Dialogue 89,"
Dialogue/89-TheDialogue/89-The Conservation of Bronze Sculpture in the Outdoor Environment,
ed.. Terry Drayman-Weisser (NACE, Houston, TX, 1992), 379-386.
188 L. S. Selwyn, N. E. Binnie, J. Poitras, M. E. Laver and D. A. Downham,
"Outdoorr bronze statues: analysis of metal and surface samples," Studies in
ConservationConservation 41, 4 (1996), 205-228.
199 K. Nassau, A. E. Miller and T. E. Graedel, "The reaction of simulated rain with
copper,, copper patina, and some copper compounds," Corrosion Science 27, 7
(1987),703-720. .
200 Ira Block, N. Jensen, S. M. Sommer and W. F. Buckingham, "Chloride Leaching
fromm Outdoor Bronzes," Preprints, ICOM-CC 8lh Triennial Meeting, Vol. Ill, 6-
111 September, 1987, Sydney, Australia (The Getty Conservation Institute, 1987),
1057-1062. .
211 Phoebe Dent-Weil, et al., "The Corrosive Deterioration of Outdoor Bronze
Sculpture,"" Science and Technology in the Service of Conservation, Preprints of
thee International Institute for Conservation of Historic and Artistic Works (IIC)
Washingtonn Congress, 3-9 Sept. 1982, ed. N.S. Brommelle and Garry Thomson
(IIC,, London, 1982), 130-134.
222 "Copper Patina Formation," Special Issue, Corrosion Science 27, 7 (1987).
233 C. Bartuli, R. Cigna and O. Fumei, "Prediction of Durability for Outdoor Exposed
Bronzes:: Estimation of the Corrosivity of the Atmospheric Environment of the
Capitolinee Hill in Rome," Studies in Conservation 44 (1999), 245-252.
244 R. Hughes and M. Rowe, The Colouring, Bronzing and Patination of Metals
(London,, 1982).
255 E. Mattson, "Corrosion of Copper and Brass: Practical Experience in Relation to
Basicc Data," Brit. Corrosion Journal 15 (1980), 6-13.
266 Zeno W. Wicks, Jr., Corrosion Protection by Coatings (Federation of Societies
forr Coatings Technology, Philadelphia, PA, Feb. 1987).
277 Percy E. Pierce and Clifford K, Schoff, Coating Film Defects (Federation of
Societiess for Coatings Technology, Philadelphia, PA, Feb. 1987).

17 7
ChapterChapter J

288 Sidney B. Levinson, Application of Paints and Coatings (Federation of Societies

forr Coatings Technology, Philadelphia, PA, Feb. 1987).
299 Henry Leidheiser, Jr., "A Primer on Corrosion Protective Organic Coatings," in
Dialogue/89-TheDialogue/89-The Conservation of Bronze Sculpture in the Outdoor Environment,
ed.. Terry Drayman-Weisser (NACE, Houston, TX, 1992), 257-274.
300 David R. Bauer, "Chemical Criteria for Durable Automotive Topcoats," J.
CoatingsCoatings Tech. 66, 835 (August 1994), 57-65.
311 CHve H. Hare, Protective Coatings: Fundamentals of Chemistry and Composition
(Technologyy Publishing Co., Pittsburgh, PA, 1994).
322 D. E. Packham, 'The Adhesion of Polymers to Metals: The Role of Surface
Topology"" in Symposium on Adhesion Aspects of Polymeric Coatings, ed. K. L.
Mittall (New York, NY, 1983), 19-44.
333 W. Funke, U. Zorll, and B. G. K. Murthy, "Interfacial Effects in Solid Paint Films
Relatedd to Some Film Properties," J. Paint Technology 41, 530 (March 1969),
210-221. .
344 H. Leidheiser, Jr. and W. Funke, JOCCA 70, 5 (1987), 121 ff.
355 E. L. Koehler, "Corrosion Processes Through and Under Organic Coatings," in
Proceedings,Proceedings, 4th International Congress on Metallic Corrosion, ed. Norman E.
Hamnerr (NACE, Houston, TX, 1972), 736-741.
366 S. J. Spadafora and H. Leidheiser, Jr., "Water disbondment characterization of
polymerr coating/metal substrate systems," JOCCA 71,9 (Sept. 1988), 276-285.
377 CHve H. Hare, "Internal Stress-Related Coating System Failures," J, Protective
CoatingsCoatings and Linings 13, 10 (Oct. 1996), 99-135.
388 S. G. Croll, "The Origin of Residual Internal Stress in Solvent-Cast Thermoplastic
Coatings,"// Applied Polymer Sci. 23 (1979), 847-858.
399 Lloyd Smith, "Engineering Design of Coating Systems," Dialogue/94: Coatings
forfor Outdoor Metals Used in Artistic and Historic Works, ed. Terry Drayman-
Weisserr (NACE, Houston, TX, 1994), 14.
400 Glen Wharton, "Sharing the Responsibility for Outdoor Sculpture: Roles of the
Conservator,, Curator, Art Administrator and Maintenance Personnel,"
MaintenanceMaintenance of Outdoor Sculpture: Whose Job is it?, Symposium at the 20th
Annuall Meeting of the American Institute for Conservation of Historic and
Artisticc Works (AIC), 2-3 June 1992, Buffalo, NY, section 3.
411 Phoebe Dent-Weil, "The Conservation of Outdoor Bronze Sculpture: A Review
off Modern Theory and Practice," Preprints, AIC H'h Annual Meeting, 22-25 May
1980,, San Francisco, CA, 129-140.
422 Various authors, Maintenance of Outdoor Sculpture: Whose Job is it?,
Symposiumm at the 20th Annual Meeting of the AIC, 2-3 June 1992, Buffalo, NY.
433 Virginia N. Naudé and Glenn Wharton, Guide to the Maintenance of Outdoor
SculptureSculpture (AIC, Washington, DC, 1993).

18 8
 TheThe problem ofprotecting bronze

444 Andrew Lins, "The Cleaning of Weathered Bronze Monuments: A Review and
Comparisonn of Current Corrosion Removal Techniques," in Dialogue/89-The
ConservationConservation of Bronze Sculpture in the Outdoor Environment, ed. Terry
Drayman-Weisserr (NACE, Houston, TX, 1992), 209-230.
455 W. T. Chase and Nicolas F. Veloz, "Some Considerations in Surface Treatment of
Outdoorr Metal Sculptures," Preprints, AIC 13'h Annual Meeting, 22-26 May
1985,, Washington, DC, 23-35.
466 Linda Merk-Gould, R. Herskovitz, and C. Wilson, "Field Tests on Removing
Corrosionn from Outdoor Bronze Sculptures Using Medium Pressure Water,"
Preprints,Preprints, I(fh Triennial Meeting, ICOM-CC, Vol. II, Working Group 22, 22-27
Augustt 1993, Washington, DC (ICOM, Paris, 1993), 772-778.
477 P. D. Weil, "The Use of Glass Bead Peening to Clean Large-Scale Outdoor
Bronzee Sculpture," Bulletin of the American Institute for Conservation 15 (1974),
51-58. .
488 Gary C. Neumeister, "Advanced Coating Removal Methods," in Dialogue/94:
CoatingsCoatings for Outdoor Metals Used in Artistic and Historic Works, ed. Terry
Drayman-Weisserr (NACE, Houston, TX, 1994), 18.
499 Unpublished notes from non-ferrous metal discussion group, Maintenance of
OutdoorOutdoor Sculpture: Whose Job is it?, Symposium at the 20th Annual Meeting of
thee AIC, 2-3 June 1992, Buffalo, NY.
500 S. Siano and R. Salimbeni, "The Gate of Paradise: Physical Optimization of the
Laserr Cleaning Approach," Studies in Conservation 46 (2001), 269-281.
511 Copper Development Association (CDA), Copper Brass Bronze Design
Handbook:Handbook: Architectural Applications (New York, 1977), 22-23.
522 Andreas Burmester and J. Koller, "Known and New Corrosion Products on
Bronzes:: Their Identification and Assessment, Particularly in Relation to Organic
Protectivee Coatings," in Recent Advances in the Analysis of Artifacts, ed. Black
(London,, 1987), 97-104.
533 Justin C. Bolger, "Acid Base Interactions Between Oxide Surfaces and Polar
Organicc Compounds," in Symposium on Adhesion Aspects of Polymeric Coatings,
ed.. K.L. Mittal (New York, 1983), 3-18.
544 R. Johnson, "The removal of microcry stal line wax from archaeological
ironwork,"" in Adhesives and Consolidants, ed. N.S. Brommelle, et al. (London,
Internationall Institute for Conservation of Historic and Artistic Works (IIC),
1984),, 107-109.
555 Rika Smith and Arthur Beale, "An Evaluation of the Effectiveness of Various
Plasticc and Wax Coatings in Protecting Outdoor Bronze Sculpture Exposed to
Acidd Deposition," Conservation of Metal Statuary and Architectural Decoration
inin Open-air Exposure, ICCROM Symposium, 6-8 Oct. 1986, Paris, France
(Rome,, 1987), 99-124.
566 Dana L. Moffett, "Wax Coatings on Ethnographic Metal Objects: Justifications
forr Allowing a Tradition to Wane," JAJC 35, I (Spring 1996), 1-8.

19 9
ChapterChapter 1

577 Arthur Beale, "A Low Cost Reversible Alternative for Compensating Damaged
Patinass of Outdoor Bronze Sculpture," Objects Specialty Group Abstracts, AIC
1515thth Annual Meeting, 20-24 May 1987, Vancouver, British Columbia, Canada,
209. .
588 Phoebe Dent-Weil, "The Approximate Two-Year Lifetime of Incralac on Outdoor
Bronzee Sculpture," Preprints, ICOM-CC 4th Triennial Meeting, 13-18 Oct. 1975,
Venice,, Italy, 22/2-1-4.
599 Clifford Craine, Kent Severson, Scott Merritt, "Prospects for the Long Term
Maintenancee of Outdoor Bronze Sculpture: The Shaw Memorial Nine Years After
Cleaningg and Coating," Abstracts, AIC 2tfh Annual Meeting, 2-7 June 1992,
Buffalo,, NY, 30.
600 International Copper Research Association (INCRA), "Project 83: Development
off an Improved Incralac," unpublished report.
611 David Ehrhardt, et al., "The Durability of Incralac: Examination of a Ten Year
Oldd Treatment," Preprints, ICOM-CC 7th Triennial Meeting, Vol. II, 10-14 Sept.
1984,, Copenhagen, Denmark, 84.22.1-84.22.3.
622 Lynn Brostoffand E. René de la Rie, "Research into Protective Coating Systems
forr Outdoor Bronze Sculpture and Ornamentation, Phase I" (PTTPublications No.
1997-03,, NCPTT, Natchitoches, Louisiana, 1997).
633 Incorporates Phase II and III, as reported to NCPTT: L. Brostoff, T. Shedlosky
andd R. de la Rie, "Research into Protective Coating Systems for Outdoor Bronze
Sculpturee and Ornamentation" (PTTPublications No. 1997-03, NCPTT,
Natchitoches,, Louisiana, 1999); and L. Brostoff, T. Shedlosky and R. de la Rie,
"Researchh into Protective Coating Systems for Outdoor Bronze Sculpture and
Ornamentation,, Phase III" (PTTPublications No. 1997-03, NCPTT, Natchitoches,
Louisiana,, 2000).
644 Constance Stromberg, "A Comparison of Wax Treatments used in the
Conservationn of Outdoor Bronze Sculpture," Student Papers Presented, at the Art
ConservationConservation Training Program I}'h Annual Conference, 2-3 May 1985, Univ. of
Delawaree and The Winterthur Museum, Newark, DE, 120-139.

20 0
 22
Performancee of 29 coatings on two
typess of copper alloy substrates

Abstract Abstract
Inn these experiments, comprising Phase I of the research, a wide range of
neww and traditional protective coatings were tested on two typical substrates: a)
mirror-finished,, cast "architectural" bronze, and b) the natural brochantite patina of
50-yearr old copper roof panels. The coatings and substrates were characterized by
X-rayy diffraction (XRD), X-ray fluorescence spectrometry (XRF), infrared
spectroscopyy and scanning electron microscopy/energy dispersive spectrometry
(SEM/EDS).. Coating thickness was measured, and adhesion was assessed before
andd after weathering. The coatings were examined for defects, and coating
performancee was evaluated and ranked after both accelerated simulated weathering
andd natural outdoor weathering. Results lay the groundwork for selection of
coatingss for further testing, and provided insight into the complex interplay of
forcess at work during weathering.

2.1.2.1. Introduction
Currentt methods of coating outdoor bronze art and ornamentation remain
rootedd in traditional practices, such as waxing, lacquering and oiling. Some newer
materialss and methods have trickled down from industrial technologies, but amid
thee plethora of available industrial coatings, including acrylics, acrylic urethanes,
andd silicones, the means of properly adopting these coatings has generally been
lacking. .
Theree are some exceptions, including the acrylic lacquer Incralac, which
wass specially developed in 1964 for use on copper alloys under funding from the
Internationall Copper Research Association (INCRA). Recent efforts include the
Europeann Commission's funding of a collaborative project with several European
Union'ss partners, which was co-coordinated by the Fraunhofer Institut fur
Silicatforschungg in Würzburg, Germany. This project sought to apply a new class
off material called Ormocer® (organically modified ceramic) to outdoor bronze
protection.. Ormocer®s are heteropolysiloxanes, which are basically organic

21 1
ChapterChapter 2

polymerss modified with silicones and silicates; they remain experimental today
[1].. Otherwise, industrial coatings have occasionally been tried on outdoor
bronzes,, but have been difficult for conservators to use and adapt, and therefore
havee received little attention. As discussed in Chapter 1, conservation applications
mayy be distinguished from industrial applications by aesthetic considerations, and
perhapss more importantly by physical and chemical differences, so that adaptation
off new technologies is no simple matter.
Thee experiments described in this chapter comprise Phase I of the research
projectt conducted at the National Gallery of Art, Washington, DC from 1994 to
19977 [2]. A total of 29 protective coating systems were tested on two types of
substrates:: a) mirror-finish, cast architectural bronze, and b) natural copper sulfate
(brochantite)) patina, formed over about 50 years on copper roof panels from the
Libraryy of Congress in Washington, DC. The natural patina on the copper roof is
essentiallyy equivalent to that found on naturally weathered bronze, and for
experimentall purposes represents a naturally weathered bronze surface [3,4], The
twoo basic substrates used here may be seen to represent extremes, i.e., a new,
polished,, bare metal surface, and an old, naturally weathered, patinated surface.
Single-- and multiple-layer model coatings were applied to the substrates in
mannerss consistent with standard conservation practices. Thermoplastic acrylics
comprisee a fair portion of the coatings in this study due to their use in the field and
theirr well-known properties of good adhesion, flexibility and fair to good outdoor
durability,, especially in terms of resistance to UV degradation. Thermoset
coatingss were included with the idea that coating removal by mechanical blasting
techniquess is a feasible option both now and in the future. In particular,
crosslinked,, solvent-borne acrylic urethanes were chosen as good candidates for
testingg on outdoor bronzes due to superior outdoor durability and extensive
development,, especially by the auto-refinishing industry. Several coating systems
weree designed according to industrial principles of three-part coatings, in which an
adhesion-promotingg layer is applied, followed by protective coating materials.
One-- and two-part waterborne polyurethanes were also included in the study.
Thesee coatings are in high demand due to low VOC requirements and the desire
forr environmental- and operator-friendly materials.
Ann automotive-type accelerated weathering program was conducted in
tandemm with natural outdoor weathering on the roof of the National Gallery. The
automotivee industry has expended singular effort in the development of cyclic,
simulatedd weathering methods in order to improve correlation to typical urban
environmentall effects compared to that seen with traditional salt fog testing [5,6].
Thee results of those efforts were utilized to design weathering methods in this
studyy that would similarly improve correlation to natural outdoor weathering of
bronzee art and ornamentation. However, it is recognized that weathering
conditionss vary widely for typical outdoor bronzes and are unpredictable, so that
anyy simulated weathering program can only be expected to correlate reasonably
welll to a defined cross-section of cases.
Phasee I focused on properties and appearance of the coatings before and
afterr weathering. Physical testing included dry film thickness measurement and

22 2
 PerformancePerformance of 29 coatings

cross-cutt adhesion testing. Tensile strength, permeability, and other inherent

mechanicall properties of the films, as they are known from the literature, were
takenn into account, but not tested. Relative corrosion development and coating
failuree were assessed visually. These results, along with limited chemical
analyses,, provided a basis for pinpointing key factors in overall coating
performance,, and for the selection of coatings for further testing and evaluation.

2.2.2.2. Experimental methods

Phasee 1 samples consisted of ninety cast bronze ("Set I") and ninety
naturallyy patinated copper roof coupons ("Set II") prepared with one of 29 coating
systemss or left uncoated as a control. Both sets were divided into three series,
whichh underwent either A) no weathering (controls), B) about 1140 hours of
acceleratedd weathering, or C) approximately 1.5 years outdoor weathering on the
rooff of the National Gallery of Art. In addition, several coupons of rolled copper
sheet,, as well as glass slides, were coated with the basic coating types and included
inn the accelerated weathering tests.

2.2.1.2.2.1. Description of the coatings

AA list of the coatings and coating systems is shown in Table I. The
descriptionss are supplemented by more detailed chemical analysis in Chapter 5.
Coatingss #1-13 are acrylic lacquers, #1-8 of which are based on Incralac. The sole
manufacturerr of Incralac in the United States is StanChem, Inc., which reportedly
producess the coating according to the original formulation, using Paraloid B-44* (a
methyll methacrylate copolymer from Rohm and Haas, Inc.), a silicone oil, and a
UVV absorber. This series is varied by combining the coating with a substrate
pretreatmentt and/or a wax topcoat. In addition, the coating formulation is
supplementedd in two cases by adding Tinuvin 292, a UV absorber (#5) or yAPS
silanee (#6).
Coatingss #9-13 are based on other acrylic resins, including Paraloid B-48
(Rohmm and Haas, Inc.), a higher molecular weight methyl methacrylate copolymer
withh reportedly better adhesion to metals. The manufacturer's literature states that
B-488 is more flexible than B-44, and recommends that it be used instead of B-44 in
thee Incralac formulation. Coatings #10-11 were formulated from B-48 for this
studyy by Cape Cod Research, Inc. (CCR). The Nikolas 11565 Outdoor Lacquer
(#12)) is a commercial coating designed for exterior brass and bronze and supplied
readyy for use. The manufacturer's literature states that it is a "modified acrylic"
containingg BTA as well as UV absorbers, and that, "[b]ecause of its excellent
flowingg capabilities, orange peel is virtually eliminated" at thinner coatings [7].
Analysiss indicated that it is based on B-48 (see Chapter 5).

** Paraloid resins were formerly sold under the name of "Acryloid" in the United States.

23 3
ChapterChapter 2

Urethanee coatings include three basic varieties: two-part solvent-bome

systemss (coatings #19-22), two-part waterborne systems (#24), and one-part
waterbomee systems (#26-27). The solvent-borne coatings are based on acrylic
polyolss (or hydroxyl functional acrylics), which are reacted with an aliphatic
polyisocyanatee crosslinking agent. Additives such as catalysts, flow aids such as
silicones,, UV stabilizers, and defoamers are typically included in these systems. In
thiss study, particular use is made of the Nikolas 9778 Exterior Uralac, a room
temperaturee cure, two-component acrylic urethane developed for exterior lighting,
exteriorr hardware, automotive trim, exterior signage, and sporting goods, including
gold-plated,, brass, and silver-plated metals. Experimental sets of acrylic/acrylic
urethanee (ac/au) multi-part coating systems using the Nikolas coatings are included
inn the study (#15-18), as well as one other such system using coatings from PPG,
Inc.. (#14). The purpose of these coatings was to combine the good adhesion
characteristicss of an acrylic with the superior outdoor durability of the urethane
[8].. This type of system also demonstrated a unique method for coating removal,
consistingg of self-peeling after immersion in or exposure to solvent.
Waterbornee systems in this study are dispersions, moisture curing, or a
combinationn of both; they incorporate chemical systems that are similar to the
solvent-bornee acrylic urethanes, as well as substantial quantities of surfactants and
otherr additives. Because of the way in which they are engineered, these coatings
tendd to have ultra-high molecular weight polymers, and are therefore difficult to
remove.. Coatings in this category include the StanChem one-part waterborne
acrylicc urethane (#26), which has been marketed in the conservation field as
"water-basedd Incralac." Chemically, the coating is quite distinct from Incralac,
however,, because of the urethane cross-linkage component. The Nikolas 11650
Eco-bomee Brass Lacquer (#27), a one-component, waterbome acrylic emulsion
withh a urethane dispersion component, was designed for the interior metal
maintenancee market, in particular for elevators, but also has been used for limited
exteriorr applications. The coating is reported to be reversible in a specially made
stripperr from Nikolas. The manufacturer states that the coating contains BTA, but
noo additional UV absorbers [9]. Also in this category is the CCR two-part
waterbornee polyurethane (#24), which was formulated for the National Gallery of
Artt study.
Thee silicone alkyd (#23) was also formulated for the National Gallery of
Artt by CCR. It is made by condensation co-polymerization of a silicone resin and
ann organic resin and blended with organic corrosion inhibitors and UV absorber.
CCRR coating formulations are known, but are undisclosed as part of a secrecy
agreement. .
Thee microcrystalline wax blend coating (#28-29) is a synthetic wax mixture
off about 75% Bareco Victory Wax (microcrystalline, low melting point wax), plus
Barecoo Polywax 2000 and 500 (polyethylene microcrystalline waxes), and
Petronaubaa C, an oxidized polyethylene wax. This mixture substitutes Petronauba
CC for natural carnauba wax in order to eliminate any possible free acid
components,, and does not contain BTA [10]. It is a good example of synthetic
blendd currently used by American conservators, and was also used for topcoats.

24 4
 PerformancePerformance of 29 coatings

TABLEE I: List of Coatings in Phase I

DESCRIPTION N MAINN INGREDIENT BTA A SILANE E WAX X

pretreat-- pretreat-- top--
ment t ment t coat t
1.. Incralac Paraloidd B-44 (R&H)
2.. Incralac Paraloidd B-44 VV
3.. Incralac Paraloidd B-44 VV
4,, Incralac Paraloidd B-44 VV VV
5.. Incralac w/ 2% Paraloidd B-44
Tinuvinn 292
6.. Incralac w/ 2% silane Paraloidd B^I4
7.. Incralac Paraloidd B-44 VV
8.. thick Incralac (5 coats) Paraloidd B-44
9.. B-48Sf Paraloidd B-48 (R&H)
10.. CCR acrylic Paraloidd B-48
11.. CCR acrylic Paraloidd B-48 VV
12.. NK 11565 acrylicc lacquer
13.. PPGDCA468 acrylicc lacquer
14.. PPG DCA468 + acrylicc lacquer/urethane
DAU75 5
15.. NK11565+ NK 9778 acrylicc lacquer/urethane
16.. NK11565+ NK 9778 acrylicc lacquer/urethane V V
17.. NK 11565+ NK 9778 acrylicc lacquer/urethane VV
18.. NK11565+ NK 9778 acrylicc lacquer/urethane VV
19.. NK9778 acrylicc urethane
20.. PPG DAU75 acrylicc urethane
21.. PPG pt + PPG DAU75 acrylicc urethane
22.. NK 9778 acrylicc urethane VV
23.. CCR silicone alkyd siliconee alkyd resin
24.. CCR polyurethane 2-partt wb polyurethane
25.. silane pretreatment SilquestA-1100®® yAPS VV
only y Silanee (OSi)
26.. StanChem wbb acrylic urethane
"waterbornee Incralac"
27.. NK 11650 wbb acrylic emulsion-
urethanee dispersion
28.. Microcrystalline wax syntheticc wax mixture
29.. Microcrystalline wax syntheticc wax mixture VV
30.. uncoated (control)

KEY:: BTA=benzotriazole; R&H= Rohm & Haas; CCR = Cape Cod Research, Inc., NK =
G.J.. Nikolas & Co., Inc.; PPG = PPG Industries, Inc.; OSi= OSi Specialties, Crompton
Corp.;; wb=waterborne

ff B-48S is supplied as 45% solids in toluene solution; B-48N, as designated in other

studiess and supply catalogues, is in the form of solid beads, but the resins are the same.

25 5
ChapterChapter 2

2.2.2.2.2.2. Sample preparation

Thee bronze substrates were investment wax cast as plaques measuring 15"
xx 6" x 1/8" and polished to a mirror finish by Art Research and Technologies, Inc.
Thee Architect of the Capitol generously donated the copper roof samples. They
weree obtained from the Library of Congress in Washington, DC, where the roof
wass undergoing replacement at the time. The copper sheets had been exposed at
ann incline of about 30 degrees, due south, for approximately 50 years. Several
adjacentt sheets were sheared into 5" x 6" coupons. The coupons chosen for the
coatingg study had a fairly even, light green patina with black spots that were either
finelyy distributed or large and raised; there were also light yellow-orange spots on
manyy samples.
Alll the bronze plaques were solvent cleaned by immersion, wiping, and
rinsingg in a series of solvents of different polarity in order to remove fatty
polishingg residues on the surface and in the pores, and to generally reduce surface
tensionn for better wetting of the coatings. The solvents were benzine, xylene, and
ethanol.. Copper roof samples were cleaned with light scrubbing in deionized
waterr and Triton X-100, a non-ionic surfactant, and air-dried. Selected samples
alsoo underwent the following pretreatments: a) immersion in a corrosion inhibitor,
1.5%% benzotriazole (BTA)/ethanol, b) brushing with a coupling agent, 2% solution
off y-aminopropyltriethoxysilane (yAPS) in propanol/water [11], or c) the PPG
(manufacturer'ss recommended) pretreatment: wiping with a degreaser (Acryli-
Clean®,, DX330), and light etch cleaning with a phosphoric acid/chromic acid
solutionn (DX501). The silane was used as an adhesion promoter, and the latter
pretreatmentt was used to expose a new metal surface and promote good adhesion.
Mostt coatings were applied with an airless sprayer, following
manufacturers'' recommendations as closely as possible, by an experienced outdoor
sculpturee conservator [12]. Incralac coatings were sprayed in two coats after
reductionn in xylene in the ratio 60:40. Several samples were brush-coated,
includingg the silicone alkyd and waterborne polyurethane, due to high viscosity.
Topcoatss of the microcrystalline wax blend were applied by cold brushing and
buffing.. The pure wax coatings were applied by a spraying and reheating
techniquee [13],
Afterr coating, the bronze plaques were cut on a band saw into three equal
pieces,, resulting in 5" x 6" coupons. Series B and C of each set, destined for
weatheringg cycles, were scribed with a 2" vertical line in the bottom middle of the
samplee in order to encourage and enable controlled assessment of creepage
corrosion,, i.e., corrosion which radiates out from a cut, underneath the coating.
Thee edges of all panels were waxed in order to exclude or minimize extraneous
creepagee corrosion around the panel edges, in preparation for measurement of
overalll corrosion. All samples were photo-documented in their before-weathering
statee and stored in the dark in polyethylene bags in a controlled environmental
roomm at 23 °C and 50% RH when not otherwise undergoing weathering.

26 6
 PerformancePerformance of 29 coatings

2.1.1.2.1.1. Weathering
Ann automotive-type accelerated weathering program, modified according
too appropriate ASTM standard test procedures and SAE standards (ASTM D3459-
988 [14], ASTM D2243-95 [15], SAE J1960 [16,17]) for the National Gallery's
equipment,, was conducted in tandem with natural outdoor weathering on the roof
off the National Gallery (ASTM G7 [18]). The initial accelerated weathering
program,, which may be called a "modified Volvo test," lasted about 1140 hours
andd consisted of a) exposure to light (outdoor simulation) and dark cycles with
humidityy and temperature cycling in an Atlas Ci65a xenon-arc weatherometer, and
b)b) exposure to humidity and freeze-thaw temperature cycling with intermittent
"acidd rain1" spraying twice per week in a Tenney Thirty temperature/humidity test
chamber.. The spray solution, recommended by Dr. Richard Granada, consisted of
aa 1:1 mixture of 1% NaCl (wt) and H2SO4/HNO3 "acid rain" solution. The final
solutionn pH was 4.0 (adjusted with the addition of NaOH), and the average ion
concentrationn was, in meq/1: [S042"] = 18.73, [NO3] = 5.60, [CI] = 140.84, and
[Na*]] = 234.49.
Simulatedd weathering conditions were subsequently modified for an
additionall six months of accelerated weathering in order to try and improve the
testingg procedure and better conform to ASTM G26 standards [19]. Modifications
too the programs included a higher irradiance level in the weatherometer, more
cyclingg with the elimination of conditioning cycles, and intermittent "acid rain"
sprayingg in the Tenney five times per week. The "acid rain solution" was also
alteredd in order to reduce the chloride concentration to realistic levels, and, based
onn experimental work of Nassau, et al. [20], to better mimic relative concentrations
off ions in naturally occurring precipitation. The final solution pH was 3.0; average
sulfatee and nitrate concentrations were 3.84 and 0.54 meq/1, respectively, and
chloridee and sodium concentrations were 0.69 and 2.64 meq/1, respectively. This
artificiall rain solution represents roughly 25x the maximum concentrations of ions
inn a high sulfate type of precipitation [20]. Average accelerated conditions and
programss are shown schematically in Figures 1 and 2.
Inn the weatherometer, a black metal panel is situated in the sample rack and
attachedd to a thermocouple. This temperature was taken as a measure of radiant
heatt and as the closest equivalent temperature of the metal panels during the light
cycle,, although in reality it is expected to be somewhat higher than the sample
temperature.. During dark cycles, black panel and dry bulb temperatures remain
aboutt the same and either may be used to gauge the sample temperature. The
differencee between wet bulb and dry bulb temperatures was used to calculate
relativee humidity.
Forr outdoor weathering of the samples, painted wood racks were
constructedd and positioned on the roof of the National Gallery of Art, West
Building,, so that samples were exposed due south at a 45° angle, according to
ASTMM G7 [18]. Outdoor weathering lasted a total of 1.5 years.

27 7
ChapterChapter 2

100 0
900 +
800 -
OO
oo
FF 700 -
oo 33 DARK K
XX 60 0 CYCLE E
33
03 3 Q) ) 50 0 LIGHTT CYCLE
O) ) >> 0.355 W/m2 (o) 340 nm
OO ro ro 40 0
F Fa> > 300 f
h-- rr r
<D D

20 0
10 0
00
22 3
T i m ee (hours)

•temp,, (black panel, ) yy (%RH;

Figuree 1 Xenon-arc weatherometer accelerated testing cycle, average conditi

120 0
110 0 FREEZE-THAWW CYCLING HUMIDITYY CYCLING
100 0
90 0
80 0
OO !2 70 0 ii \ \\
5-- E 60 0
CDD 3
50 0
40 0
22 > 30 0
20 0
EE i» 10 0
JJ—öv —ö
I-- $5 00
-10 0 repeatt x 10
-20 0
-30 0 repeatjj/8 8
-40 0 HH h
88 10 12 14 16 18 20 22 24 26
T i m ee ( h o u r s )

•temp.. ) ee humidity (%) - A uncontrolled humidity

Figuree 2 Tenney Thirty chamber program for accelerated weathering, used with
intermittentintermittent "acid rain " spraying.

28 8
 PerformancePerformance of 29 coatings

2.2.4.2.2.4. Methods of evaluation

Metall substrates were analyzed by secondary emission energy dispersive
X-rayy fluorescence spectrometry (XRF) using a Kevex 0750A spectrometer
equippedd with a BaCb secondary target and 6 mm collimators. The anode voltage
andd current were 60 kV and 0.4 mA, respectively. Live accumulation time was
2000 seconds. Elemental weight percentages were calculated with EXACT (energy
dispersivee X-ray analysis computation technique), a software program provided by
Kevex.. Metal substrates were also characterized by scanning electron microscopy
(SEM)) and energy dispersive spectrometry (EDS) using a JEOL 6300 scanning
electronn microscope equipped with a Link Super ATW Si(Li) detector and a Link
eXLL II spectrometer. The accelerating voltage was 10-20 kV, depending on the
sample.. Selected corrosion products were analyzed by powder X-ray diffraction
(XRD)) using a Philips X-ray generator 3100 equipped with a copper target and
nickell filter to provide Cu Ka radiation, 45 kV anode voltage, and 25 mA current.
Coatingss and the copper roof patina were analyzed by Fourier-transform
infraredd spectroscopy (FTIR) using a Bio-Rad FTS-60A spectrometer equipped
withh an external UMA 300A microscope. Chemical analysis of the coatings is
moree fully discussed in Chapter 5. After application, coatings were characterized
underr low magnification for overall film quality in terms of the presence of seeds,
orangee peel, air entrapment, solvent popping, hazing or whitening, bumps and
sinks,, cratering, and crawling/dewetting, according to common industrial
definitionss [21]. Dry film thickness and its standard deviation (ASTM D1400
[22])) were determined on the unweathered control samples (set A, or 1/3 of the
originall coated plaque) with an Elcometer 345 dry film thickness gauge. Between
500 and 70 readings were taken on each sample. The bronze coating measurements
weree subject to an error of 0.1 mils (2.5 urn), and thus coatings less than about
0.22 mils (5.1 urn) could not be measured accurately. The copper roof samples had
aa large margin of error, which could not be determined due to the extremely rough
andd uneven patina.
Coatingg adhesion was evaluated by the cross-cut adhesion tape test (ASTM
D33599 [23]) on sample groups A, B, and C, i.e., before and after weathering. In
thiss method cross-hatch cuts or X's are cut through the coating with a specified
tool.. A piece of tape is then pressed onto the cuts and pulled off, and the
remainingg coating is rated on a scale of 0-5 for adhesion. This test provides a
roughh estimate of adhesion, and is capable of differentiating adhesion between
coatings.. The method does not, however, distinguish between types of adhesive
failure,, including cohesive and interlamellar failure in multi-part coating systems,
andd results were therefore somewhat deceptive in these cases. Tensile adhesion
testingg using a portable tester was attempted but was found to be too inaccurate
andd insensitive to produce meaningful results.
Samplee groups B and C samples were evaluated visually after initial
weathering.. Creepage corrosion was measured perpendicular to the edge of the
scribemarkk (ASTM D1654 [24]). Following both extended accelerated weathering
(aboutt 7 1/2 months total), and 1.5 years outdoor weathering, samples were re-

29 9
ChapterChapter 2

evaluatedd and "total failure ratings" were compiled based on ratings for a) coating
performance,, b) creepage corrosion at the scribemark, and c) overall corrosion.
Ratingss were based on the following categories: a) texture change (including
surfacee spotting/etching), b) flaking/peeling, c) hazing or whitening, d)
blistering/cracking/checkingg (ASTM D714, D660 [25,26]), e) color change, f)
overalll corrosion (ASTM 610 [27]), and g) creepage corrosion at the scribemark..
Inn each category, ratings were given on a scale of 0-5, where: 0=not present;
l=barelyy noticeable; 2=slightly noticeable; 3=noticeable; 4=very noticeable;
5=dramatic.. Category " f included pitting and overall corrosion. For this rating,
thee ASTM 610 standard was used as a basis for evaluation, but the designated
scalee was adjusted to include overall oxidation and normalized to a 0-5 scale
commensuratee with categories (a) to (e). The resulting scale was defined in terms
off overall percent corrosion of the entire surface, where: 0 = <0.03%; 0.25-1 =
0.03-1.00 %; 2 = 1.0-3.0%; 3 = 3-10%; 4 = 10-30 %; and 5 - >30 %. In this rating,
greaterr than about 30% area corrosion was felt to represent failure by normal
conservationn standards.
Totall failure ratings were derived by combining ratings for overall coating
condition,, overall corrosion, and mean creepage corrosion, after normalizing each
off the three overall ratings so that they represented one-third of the total failure, as
follows:: 1.32*(a + b + c + d + e) + 6.6*(f) + 3.3*(g), for a total maximum failure
off 99 [6]. This formula was empirically derived, and was found to correspond
reasonablyy well to empirical rankings. The scale of this total failure rating,
however,, is large compared to an acceptable level of failure that may be tolerated
inn conservation. This will be discussed in the Results and Discussion section,
below.. The total failure ratings are shown as bar graphs on a scale of 0-60, since
noo failure was seen beyond this. Although 60% failure may be a normal industrial
standardd for coating failure, in museum and conservation circles, failure would
moree likely be defined closer to about 30%. Even greater than 20% failure may
nott be acceptable according to some opinions.
Itt was found that the majority of set IIC samples (copper roof, outdoor
weathering)) did not exhibit high degrees of failure in any category, and could not
bee meaningfully distinguished by the above method. Therefore, samples IIC were
groupedd into rough categories of better and worse performance.

2.2.5.2.2.5. Sources of error

Becausee of the subjective nature of the visual evaluation, samples were
judgedd on more than one occasion by both the principal investigator and project
researchh assistant. In addition, it is recognized that the method of combining
differentt ratings is somewhat arbitrary, since the assumption that all factors
involvedd in performance rating have equal weight is by no means true. However,
inn the absence of any evidence or set criteria for weighting various factors, in the
finalfinal judgment it was felt that equal weighting was preferable [28]. According to
empiricall observations, however, overall corrosion appeared to be under weighted
inn the total failure ratings. Therefore, the failure ratings contain a significant

30 0
 PerformancePerformance of 29 coatings

amountt of variability and uncertainty, so that the margin of error was estimated to
bee about 5%.
Otherr sources of error include variability in both simulated and natural
weathering.. Although sample placement was varied randomly in the Tenney
chamber,, for example, it is possible that sample position could have played a small
rolee in results. Also, instrumental shut downs occasionally occurred during the
coursee of accelerated weathering, which created uncontrolled conditions for
uncontrolledd lengths of time, and some differences between weathering conditions
forr sets I and II. Over the course of 7 V2 months, however, these unplanned events
remainedd relatively minor influences in the results. During natural weathering,
however,, unplanned events were of a more serious nature, including the blowing
overr of samples for various lengths of time and even the loss of samples in some
cases.. These events built more variability into the natural weathering results, as
wouldd occur in reality.
Thee other important source of error margin in the performance ratings was
samplee variability. The substrates themselves had many non-uniformities, either in
thee form of casting flaws or natural patina development. In addition, coating
applicationn was done by hand, vs. machine, so that the normal imperfections and
non-uniformitiess which present themselves in actual practice, although kept to a
minimumm as much as possible, were not eliminated. This is discussed more fully
inn the results section.
Whatt the total failure ratings convey is a basis for comparison of relative
performancee behavior, and, above all, a guide to trends of the coatings during
weathering.. Given these sources of variability and error, it would therefore be
meaninglesss to project a correlation factor between accelerated weathering results
andd real time weathering of coatings. For this reason, Incralac and wax coating
servee as types of internal standards in terms of real time behavior and maintenance
requirements. .

2.3.2.3. Results and discussion

2.3.1.2.3.1. Substrate characterization

XRFF analysis showed that the actual bronze alloy composition (% weight)
wass closer to Cu (87%), Pb (2%), Sn (5%), Zn (4%), with a small nickel
contaminantt (about 0.6%). The SEM backscattered image shown in Figure 3b
clearlyy reveals the metal microstructure: it consists of a three-phase dendritic
structure,, containing a Cu-Sn-Zn and a similar Sn-rich Cu-Sn-Zn phase, in addition
too isolated globules of Pb, which are fairly evenly spread throughout. The SEM
normall image (Figure 3a) of the bronze surface reveals numerous pits as well as
randomm groupings of dark spots. Such imperfections are typical of cast, polished
bronze.. EDS analysis of the pits showed the presence of aluminum and oxygen,

31 1
ChapterChapter 2

HH

Figuree 3 A) SEMphotograph of the surface of the polished, cast bronze, and B) SEM
backscatteredbackscattered image of the bronze surface showing a three-phase dendritic structure
(250x). (250x).

32 2
 PerformancePerformance of 29 coatings

Figuree 4 SEMphotograph of 50-year-old copper roof patina (600x).

Tablee II: XRD Identification of Copper Roof Patina Mineral Phases

Corrosion Corrosion
MajorMajor Phase MinorMinor Phase/ ReferenceReference PDF
Sample Sample File* (A)
UnidentifiedUnidentified Lines File*

green/red d copper r 4-0836 6

mixture e cuprite e 5-0667 7
brochantitee (monoclinic & 43-1458,, 13-
orthorhombicc mixture) 0398,3-0282 2
anterlitee (?) 7-0407 7
copperr sulfides (?) 33-0490,23--
0959,23-0958 8
black k copper r 4-0836 6
from m cuprite e 5-0667 7
black k brochantitee (monoclinic & 43-1458,, 13-
spot t orthorhombicc mixture) 0398,3-0282 2
copperr sulfides (?) 33-0490,, 23-
0959.23-0958 8
</=3.32,, 1.265 (?) )
** formerly JCPDS-International Centre for Diffraction Data files

33 3
ChapterChapter 2

mostt likely from grinding compounds and/or oxidation products. EDS of the dark
spotss showed the presence of carbon, but not in the surrounding areas, indicating
thatt the spots visible in Figure 3b are most likely fatty polishing residue from the
so-calledd jeweler's rouge. This residue persisted despite meticulous cleaning
efforts,, pointing out the difficulty of completely removing fatty deposits from a
polishedd surface. Deposits such as these will necessarily affect the surface tension,
andd thus the film formation and adhesion of coatings.
XRFF analysis of the copper roof metal indicated it had a purity of 99.93%.
XRFF analysis of rolled copper sheet, which was locally obtained and used for a
smalll amount of additional samples, showed a purity of 99.99% Cu. XRD analysis
off the copper roof patina (Table II) provided positive identification of the
followingg crystalline phases: copper, cuprite (CU2O), brochantite (Cu4S04(OH)6,
monoclinicc and orthorhombic mixture), and possibly copper sulfide(s). No tenorite
(CuO)) or copper chlorides were detected.
Thee SEM image shown in Figure 4 of the copper roof patina shows a very
roughh and porous morphology. EDS analysis of the green patina revealed the
presencee of Al, Si, Fe and P contaminants in addition to Cu, S, and O. The
presencee of phosphorous in outdoor bronze and copper patinas has been
sporadicallyy noted, particularly in connection with black spots on copper patinas,
andd has been variously attributed to airborne fertilizer-containing particles or
pigeonn droppings [29,30]. The other contaminants are commonly found in
airbornee dirt particles. No carbon was detected. (Nitrogen cannot be detected
fromm this detector array.)
Mostt interestingly, SEM/EDS analysis of black spots in the patina revealed
thee presence of Cl, Ca, and V. No explanation can be offered for the presence of
vanadium.. The presence of silicon as well as aluminum also appeared relatively
higherr in this spot. XRD of material from these spots (see Table II) revealed an
almostt identical pattern to the green patina, and contained cuprite, brochantite, and
possiblyy copper sulfide, but also faint new lines at d spacings of 3.32 and 1.265 A;
thesee were not identified. No differences were detected between the IR spectra of
greenn and black patina components (not shown), both of which showed excellent
agreementt with the spectrum of pure brochantite [31]. These analyses also
confirmm the very similar nature of natural patinas that are formed on copper and
bronze,, although minor amounts of tin and/or zinc oxides are absent from a copper
patinaa [4], This should have little impact on cleaning and coating strategies,
however. .

2.3.2.2.3.2. The coatings

Forr chemical analysis of coatings that received primary attention in this
study,, see Chapter 5. The list of coatings is shown in Table I, and a brief
descriptionn of the chemical character of coating systems chosen for study is
providedd in the experimental section.

34 4
 PerformancePerformance of 29 coatings

2.3.3.2.3.3. Coating quality

Thee aesthetic quality of the coatings varied greatly. On the polished bronze
substrates,, appearance change due to coating application generally ranged within
aestheticallyy appropriate limits, i.e., did not alter the character of the metal surface
significantly,, with a few exceptions. First, the thickly sprayed wax coatings (#28,
29)) appeared matte, opaque and uneven. Under magnification, "grains" could be
seenn in the coating, with many gaps in between. It was apparent that this
applicationn technique required further development on this type of polished
substrate.. Second, the silicone alkyd (#23) was viscous and orange-brown in color
beforee application, and clearly unacceptable on aesthetic grounds. CCR stated that
thiss batch might have been unstable, resulting in viscosity changes and color
instability.. The CCR waterborne polyurethane (#24) was also somewhat dark and
viscouss in the can. As a result, both CCR coatings had to be brushed instead of
sprayed,, resulting in uneven, streaky appearances, and films full of solvent
popping.. Although none of the abovementioned coatings was excluded from the
study,, their aesthetic drawbacks should be taken well into account next to
durabilityy performance. Otherwise, the coatings generally added extra "shininess"
too the samples, but this was not considered to be a handicap since shine may be
tonedd down by a wax topcoat, or even incorporation of matting agents like fumed
silicaa in real practice. In addition, the silane pretreatment produced streakiness on
thee bronze, which was attributed to its application in too high concentration. This
wass apparent through the coatings over this pretreatment. In contrast, the BTA and
thee PPG-recommended pretreatments were invisible.
Severall coatings were marred by somewhat poor film quality on the bronze,
includingg the thick Incralac (#8), which developed severe orange peel, and the
StanChemm waterborne acrylic urethane (#26), which developed dense pinholing in
thee second coat. It was noted that the acrylic coatings in general were subject to
densee pinholing and/or solvent/air popping, including most of the Incralac
coatings.. Coatings that developed light orange peel or other texture included:
Incralacc with silane formulation (#6), BTA pretreatment plus Incralac (#2), and
silanee pretreatment plus Nikolas acrylic/acrylic urethane (#17). These defects not
onlyy affect aesthetic criteria, but also must be borne in mind as potential factors in
coatingg failure.
Thee quality and aesthetics of coatings on the copper roof samples were far
moree varied, especially with regard to color effects. Generally, it can be stated that
mostt coatings resulted in poor aesthetic results, although this is of course a
subjectivee judgment. More objectively, it can be said that an unchanged
appearancee was very difficult to achieve on the mineralized surfaces: only B-48
(#9)) and the Nikolas waterborne coating (#27) boasted this achievement. Incralac
(#1-8)) appeared dark, shiny and produced welling and pooling around surface
imperfectionss on the mineralized surfaces. The other acrylics (#10-13) went on
somewhatt more evenly. The urethanes (#19-22) appeared very shiny and
saturated,, resulting in a dark, muddy green color. Topcoat waxing generally
improvedd appearances, due to a matting effect. It should be noted, as stated above,
thatt matting agents are available for use with the acrylics and acrylic urethanes,

35 5
ChapterChapter 2

whichh would help minimize their shininess. The effect of matting agents was not
addressedd in this study, and therefore these agents were not included. The wax
coatingss (#28, 29) were dark, matte, muddy and uneven-looking.
Coatingg quality on the copper roof samples was similar to that on bronze,
butt defects were somewhat masked by the rough, patinated surface. In particular,
itt was noted that the acrylic coatings had peculiar wetting properties on the
patinatedd surface, and produced a microscopic "shrink-wrap" appearance upon
drying.. Silane pretreatment, as well as the BTA and the PPG recommended
pretreatments,, were invisible on the copper roof panels.
Ass already mentioned, the coating defects noted above not only impair
aestheticc quality, but may impede good coating performance. The potential effect
off pinholes or uneven thickness, for example, should not be underestimated, and
mustt be blamed for a degree of variability in the performance evaluation. This
complicatess the picture represented by the final ratings, but is realistic in the sense
thatt no laboratory prediction can hold out the promise of absolute reproducibility
withoutt strict control over quality, which is in fact unachievable in the field.

2.3.4.2.3.4. Corrosion analysis after weathering

Thee corrosion resulting from accelerated and outdoor weathering on the
sampless was quite varied in appearance. Accelerated weathering in particular
producedd different corrosion patterns and colors at the scribes, indicating that
corrosionn mechanisms may vary in association with different coatings. On bronze,
mostt areas of corrosion appeared to begin with thin films of orange-red-brown
cupritee formation, often etched in appearance, followed by accumulation of light
greenn and sometimes black spots on top of the red, as well as isolated thick green
crustedd areas. Samples pretreated with silanes or containing silanes in the
formulationn showed brown or gold spots radiating out from the scribes. Sample
coatingss were also water-spotted after accelerated weathering. On bronze samples
exposedd to outdoor weathering, orange-brown areas appeared etched and often
speckledd with light green, green, red, and black corrosion. Some type of biological
attackk was also noted on these samples in the form of filaments, which were easily
wipedd off. Outdoor weathered samples generally were also covered with dirt,
particularlyy the waxed coatings. The samples treated only with silane appeared
similarr after both indoor and outdoor weathering, with variegated areas of matte
brown-orangee and shiny gold metal.
Copperr roof panels exposed to accelerated weathering often exhibited
blackishh corrosion in the scribes and under intact coatings near the scribes. This
wass not noted on samples exposed outdoors. New light green corrosion and white
accretionss were also noted in the scribes of samples after accelerated weathering.
Thee control and samples treated only with silane or BTA showed areas of dark
greenn toward the bottom and around the edges where more wetting took place. As
expected,, samples exposed outdoors did not exhibit much change: scribes most
oftenn appeared orange-brown, presumably from cuprite formation, with some new

36 6
 PerformancePerformance of 29 coatings

TABLEE III:
XRDD Identification of Phases Present After Accelerated Weathering

Sample Sample PhasesPhases Identified/

Corrosion Corrosion ReferenceReference PDF
Color(s) Color(s)
UnidentifiedUnidentified Lines (A) and File* File*
TentativelyTentatively Associated Phases
I,, 30B: orange-- cuprite e 5-0667 7
uncoated d brown n paratacamitee and 25-1427,, 19-
bronze e green n otherr copper chloride hydroxides 0389,, 2-0146,
(includingg botallackite) 25-0269,4-0193 3
brochantite e 43-1458,, 13-
0398,3-0282 2
</=3.70,, 3.42, 3.08,3.01-3.03, 1.89:
copperr sulfate hydrate (?) 21-0269 9
zincc oxide sulfate (?) 32-1476 6
l,, IB: Incralac powdery, , cuprite e 5-0667 7
onn bronze, light t paratacamite;; botallackite; atacamite 25-1427,, 19-
scribe e green n 0389,4-0193, ,
corrosion n 23-0948 8
NaCl l 5-0628 8
[I,, 30B: darkk green samee phases as before weathering (seee Table II)
uncoated d black k paratacamite e 25-1427 7
copperr roof orange-red d cuprite e 5-0667 7
11,, IB: Incralac green n paratacamitee or other copper 25-1427,25-1427, 19-
onn copper hydroxyy chlorides 0389,23-0948 8
roof,, scribe cuprite e 5-0667 7
corrosion n NaCl l 5-0628 8
11,, 15B,NK green n paratacamitee + other copper hydroxy 25-1427,25-1427, 19-
acrylic/ / greenish-- chlorides s 0389,2-0146 6
acrylicc ure- black k cuprite e 5-0667 7
thanee on red d d=A.l\,d=A.l\, 3.42, 2.32:
copperr roof, copperr chloride hydroxide hydrate 23-0950 0
scribe e (?) )
corrosion n
^formerlyy JCPDS-International Centre for Diffraction Data files

lightt green corrosion visible to various degrees. The waterbome acrylic urethane
coatingss after both types of weathering were unique in the appearance of a tide-line
off corrosion outside the scribe, apparently from water penetration and spreading
underr the coatings.
XRDD results for corrosion scraped from selected samples after accelerated
weatheringg are shown in Table III. These results show that corrosion formed
duringg accelerated weathering consisted mainly of cuprite and copper hydroxy
chlorides,, in particular paratacamite (Cu2(OH)3Cl), although brochantite was
detectedd as a minor phase on the uncoated bronze substrate. Paratacamite is listed
inn the PDF International X-ray Diffraction Files as greenish-black as well as light

37 7
ChapterChapter 2

green,, and this is most likely the identity of the new black corrosion formed on
copperr roof panels during accelerated weathering. The identity of new black
corrosionn on the outdoor bronze samples was not clear, however. It is possible that
cupritee at a certain thickness has a black interference color. Unidentified lines in
thee patterns from substrates coated with acrylics suggested the presence of
additionall products, which may have been related to the individual coatings and
theirr additives. In addition, it may be significant in terms of corrosion mechanisms
thatt the NaCl remained present on the Incralac-coated substrates.
Althoughh XRD results indicated that the accelerated weathering regime did
nott perfectly mimic the formation of typical copper sulfate patinas, all corrosion
productss found here have been reported to occur naturally in typical outdoor
environmentss [30, 32]. In particular, it has been found that chlorides are
omnipresentt in corrosion products of street-level bronze and copper objects. The
predominancee of chlorides over sulfates on the samples after this accelerated
weatheringg program appears unbalanced, however, compared to typical
occurrences.. This may be attributed to too much NaCl in the initial rain solution
andd the relatively slow natural formation of brochantite.

2,3.5.2,3.5. Coating performance on bronze, accelerated weathering

Coatingg performance was evaluated after both the initial and extended
periodss of accelerated weathering. Following initial weathering, relatively few
sampless showed signs of overall corrosion and/or general coating failure. The
Nikolass acrylic/acrylic urethane (#15-18) series and single acrylic urethane
coatingss (#19, 22) were performing quite well, followed closely by most of the
Incralacc series (#1-4, 6, 8) coatings. Several coatings did stand out as poor
performerss at this stage. The B-48 coating (#9) showed signs of fairly advanced
creepagee corrosion at the scribemark, and both the PPG acrylic urethane (#20) and
Incralacc with Tinuvin 292 coating (#5) showed marked signs of wear. Other
coatingss that showed poor to intermediate performance included, in order from
worstt to best: the CCR acrylics (#10, 11), the StanChem waterborne coating (#26),
thee wax coating without BTA pretreatment (#28), PPG acrylic/urethane (#14), and
thee CCR polyurethane (#24).
Figuree 5 shows the total failure ratings after extended weathering,
determinedd as outlined in the Experimental section. Results illustrate that the poor
andd intermediate performance trends noted above generally continued through
extendedd weathering. The ratings do show some new trends, though. After
extendedd weathering, the Incralac coatings (#1-8) appeared to be lagging
somewhatt behind the Nikolas group (#15-18, 22), which continued to show
excellentt performance characteristics. Three of the Incralac coatings, Incralac with
waxx (#3), BTA + Incralac with wax (#4), and thick Incralac (#8), appeared to be
doingg somewhat better within that group. Results thus suggest a benefit from the
waxx topcoat over Incralac, which was equal to or better than the benefit of
applyingg a thicker coating. The addition of silane to the Incralac may have caused
somee crosslinking in the coating, which could have boosted performance [33].

38 8
 PerformancePerformance of 29 coalings

However,, this was not confirmed. No clear benefit of a BTA or silane

prctreatmentt was indicated in this or any other series of coatings.

11 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29

Coatingg Number (sample)

•• overall corrosion 1 scribe creepage corrosion coating rating

Figuree 5 Relative failure of coatings on polished, cast bronze after extended accelerated
weathering. weathering.

Comparisonn of the Nikolas acrylic (#12), acrylic/acrylic urethane systems

(#15-18),, and acrylic urethane alone (#19, 22) after extended weathering suggests
thatt the acrylic urethane layer was most responsible for boosting performance, with
orr without the acrylic underlayer. The presence of the acrylic undercoat in the
multi-partt coatings had the added function of providing a method of removal
throughh solvent swelling of the underlayer and subsequent peeling of the coating.
Thee remaining acrylic coatings (#9-11, 13) did not show improved performance
overr Incralac after either initial or extended weathering. The Nikolas acrylic
urethanee (#19, 22) appeared superior to the PPG acrylic urethane (#20, 21), the
latterr of which did show a strong benefit from the manufacturer's recommended
pretreatmentt (see adhesion discussion, below).

Thee extremely poor performance of the B-48 resin (#9), which nonetheless
appearedd to stabilize during extended weathering, has no simple explanation other
thann that the coating was applied too thinly. This points to the importance of using

39 9
ChapterChapter 2

aa well-formulated coating, rather than the resin alone. The mediocre performance
off the CCR acrylic coatings may be explained with similar reasoning, i.e., that they
weree experimental formulations, for which the kinks were not worked out. These
coatingss resulted in rather poor quality films, as discussed above, which clearly
compromisedd performance. The relatively poor performance of the Incralac +
Tinuvinn 292 (#5) may have been due to improper mixing of the Tinuvin into the
Incralac,, and may represent a warning against adulterating already-formulated
coatings.. In addition, the startlingly poor performance of PPG acrylic/acrylic
urethanee (#14) vs. the similar Nikolas systems illustrates well that coatings can
varyy greatly within the same coating class according to manufacturer and
application-friendlyy formulations.
Otherr coatings which showed increased rates of failure during extended
weatheringg include the waterborne coatings (#24, 26, 27) and the wax coatings
(#28,, 29). These stood out from the other coatings, but for different reasons, as
shownn by the breakdown of ratings in the bar graphs. The wax coatings did little
too protect the metal against corrosion, as did the silane pretreatment. Waterborne
coatingss are generally known to have problems with water sensitivity, and this was
bornee out in the study. The StanChem waterborne acrylic urethane (#26) yellowed
markedlyy and appeared to be unstable to UV exposure. The Nikolas waterborne
acrylic/urethanee (#27) was also applied very thinly, which appeared to exacerbate
suchh problems. Photographs of these samples are shown in Plate I.

2.3.6.2.3.6. Coating performance on bronze, natural outdoor weathering

Afterr about 3 months of outdoor weathering, the majority of coated bronze
sampless exhibited little corrosion or failure. Under magnification, however,
brown-greenn dendritic corrosion showed through beneath the wax coatings. The
bestt performers after the preliminary outdoor exposure included some of the
Incralacc coatings, followed by most of the Nikolas acrylic/acrylic urethane
coatings. .
Totall failure ratings for sample set IC after roughly 1.5 years of weathering
onn the roof of the National Gallery of Art in Washington, DC were calculated and
aree shown in Figure 6. At this point in the weathering, most of the samples were
generallyy performing well, with the exception of the wax coatings (#28, 29), some
acrylicss (#11-14), and the waterborne coatings (#24, 26, 27), although performance
ratingss for the latter two groups were still reasonable. Generally, the results appear
lesss internally consistent, so that it is not possible to distinguish the coatings more
thann this. Results for some samples were unreliable, since several mishaps
occurredd in which samples were knocked or blown over for uncertain periods of
time.. Samples that performed noticeably better in natural exposure than in
acceleratedd weathering included Incralac with Tinuvin 292 (#5), B-48 (#9), and the
StanChemm waterborne acrylic urethane (#27). Perhaps these coatings were
particularlyy sensitive to chlorides. Otherwise, results were consistent with
acceleratedd weathering results, which appear to project performance in these
samples. .

40 0
 PerformancePerformance of 29 coatings

1 22 3 4 5 6 7 8 9 1011121314151617181920212223242526272829
Coatingg Number (sample)

•• overall corrosion Ascribe creepage corrosion Dcoating rating

Figuree 6 Relative failure of coatings on polished, cast bronze after 1.5 years natural
outdooroutdoor weathering.

Inn summary, the accelerated and natural outdoor weathering results helped
too eliminate several coatings as unworthy of further study. Of the acrylics, no
coatingg tested with or without prctreatments clearly usurped Incralac's place as a
triedd and true performer with medium performance potential. Out of the Incralac
series,, #3, with a wax topcoat, appeared to be the slight frontrunner; results were
ambivalentt regarding any benefit from BTA pretreatment. In terms of new coating
options,, the Nikolas acrylic/acrylic urethanes and straight acrylic urethane
appearedd to warrant further study. The benefit of prctreatments and wax topcoats
wass in this case even less clear, however. Results underscored that the wax and
waterbornee coatings were generally in a much lower performance category, as
testedd in this study.

2.3.7.2.3.7. Coating performance on copper roof, accelerated weathering

Thee coated copper roof samples were examined after both initial and
extendedd accelerated weathering for overall signs of corrosion and coating failure,
andd for corrosion at a two-inch vertical scribe at the bottom of each sample. After
initiall weathering, signs of failure were not very prominent. One notable

41 1
ChapterChapter 2

exceptionn was the StanChem waterbome acrylic urethane (#26), which showed
markedd yellowing and embrittlement on the copper sulfate patina after a short
time.. Visual examination of the other samples revealed different amounts of
corrosionn in the scribemarks themselves. Some samples appeared to be almost
protectedd from corrosion in the exposed scribe, while others appeared to have
acceleratedd corrosion in the scribemark, as would be expected from the creation of
localizedd potential differences in coated and uncoated areas.
Otherr than by examination of the scribes, new corrosion on the samples
wass often difficult to distinguish from the pre-existing patina. As previously
discussed,, much less corrosion was expected on the partially passivated copper
rooff substrates than on the polished bronze. Coatings which were performing
noticeablyy better on copper roof than on bronze after initial accelerated weathering
includedd Incralac with Tinuvin 292 formulation (#5), straight B-48 (#9), PPG
acrylicc urethane coatings (#20, 21), CCR acrylics (#10, 11), which were however
appliedd more thickly than on bronze, CCR waterborne polyurethane (#24), and the
waxess (#28, 29). The manufacturer's pretreatment for the PPG acrylic urethane
(#21),, which resulted in improved performance on bronze, actually fared worse
afterr pretreatment on the copper roof panel. Nikolas aery lie/aery lie urethane
systemss (#15-18) did not perform as well on copper roof panels as on the cast
bronze. .
Thee good initial appearance of the scribemarks and lack of obvious
creepagee corrosion on the copper roof panels with waxed coatings was partly
attributablee to unavoidable spreading of the wax into the cut during scribing. This
iss one reason why the copper roof panels with pure wax or wax topcoats received
higherr ratings after initial accelerated weathering. It is not clear, however, why
thiss effect was not as marked on bronze, nor whether the relatively better
performancee of pure wax coatings on copper roof panels was real or due to
difficultyy in assessing the condition of copper roof panels after the short exposure.
Noo improvement from either BTA or silane pretreatment was seen at this point for
anyy coatings on the copper roof panels. The presence of BTA on a treated copper
rooff panel was confirmed by EDS analysis.
Inn general, most coatings continued to perform reasonably well after
extendedd accelerated weathering. Little or no light green pitting corrosion
occurredd on the substrates during this weathering regime, and the appearance of
new,, dark green or red corrosion remained difficult to judge from the top surface.
Forr these reasons, failure ratings were more difficult to obtain than for the cast
bronzee samples. Those compiled for coatings on copper roof substrates after
extendedd accelerated weathering are shown in Figure 7, Results show only the two
waterbornee coatings (#26, 27) stand out from the field in terms of poor
performance.. As noted above, the StanChem waterborne acrylic urethane (#26)
failedd relatively early in the weathering cycle. Otherwise, the ratings indicate
fairlyy equivalent performance by the various coatings, albeit with significant
marginn of variability and error due to the nature of these substrates. Differences in
thee coatings remained more an aesthetic issue, as previously discussed.

42 2
 PerformancePerformance of 29 coalings

-o-o 60
cc
50 0
oo
oo
cc
II I 40 0
oo

£ £ 30 0
Tc c
QL QL
o> > 20 0
__
33
10 0

la a mmm mmm nn annnnnnn n

11 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 26 27 28 29
Coatingg Number (sample)

•• overall corrosion W scribe creepage corrosion D coating rating

Figuree 7 Relative failure of coatings on 50-year-old copper roof after extended

acceleratedaccelerated weathering.

2.3.8.2.3.8. Coating performance on copper roof, outdoor weathering

Sampless were examined after about 3 months, 10 months, and roughly 1.5
yearss of natural, outdoor weathering on the roof of the National Gallery of Art, but
thee majority showed relatively little change over this time period. Again, only the
StanChemm waterborne acrylic urethane (#26) clearly failed. As in accelerated
weathering,, this coating turned markedly yellow and exhibited flaking, all-over
cracking,, and dark green mottling under the coating. A small group of coatings
showedd some noticeable change, including, in order of best to worst: the B-48
resinn (#9); the silicone alkyd (#23); and the Nikolas waterborne coating (#27). The
B-488 coating appeared slightly mottled, with a bit more corrosion at the bottom
edgee where the sample was touching the wood rack. The Nikolas waterborne
coatingg had a slightly mottled appearance, with some corrosion at the bottom.
Notee that both of these samples had very thin coating applications (see discussion
below).. The silicone alkyd coating, which was thick and had poor appearance
qualitiess initially, developed whitish areas at the peak of brushmarks in the
coating,, but showed no signs of corrosion.
Inn addition, both of the waxed copper roof samples (#28, 29) had a
comparativelyy poor appearance: mottled, with dark spotting, in addition to severe
accumulationn of dirt and soot embedded in the coating. Sample #28. without BTA
pretreatment.. appeared darker and slightly worse than that with the BTA
prctreatment.. Under magnification, the surfaces of the coatings no longer

43 3
ChapterChapter 2

appearedd smooth, but rough, with opaque patches of bloom. Some flaking was
alsoo evident. On sample #28, at least one pit of light green corrosion was visible.
Inn general, the results of both accelerated and natural weathering for
coatingss on copper roof stand apart from those for bronze. Results show only a
feww coatings out of the 29 tested that failed outright. In terms of visible corrosion,
Incralacc and the thick wax coatings ranked about the same as many of the newer
coatingss tested. However, there was significant aesthetic variation among samples,
inn part due to varying degrees of penetration of the coatings into the thick copper
patinas.. Coatings that saturated patinas created a dark, muddy appearance. In the
casee of waterborne acrylics and B-48, poor wetting characteristics and presumably
thee high molecular weight of these polymers resulted in less penetration into the
greenn patina, and thus in less change in appearance [34]. Even without significant
saturationn of the patina, the wax coatings looked opaque and muddy, and darkened
withh weathering.

2.3.9.2.3.9. The influence of thickness and adhesion on coating

performance performance

2.3.9.1.2.3.9.1. Thickness

Inn the complex interplay of factors that contribute to coating performance

onn outdoor bronzes and copper alloys, coating thickness and adhesion to the
substratee may be singled out as essential links in the performance of the system. It
seemss obvious that thicker coatings will provide more protection, and this is true to
aa point. The creation of a very thick film can also aggravate certain defects in the
coatingg such as orange peel. In order to avoid orange peel, in fact, operators often
applyy coatings much thinner than recommended by the manufacturer, less than
aboutt 2.0 mils (51 urn). In addition, a thick film may have a tendency to peel or
liftt off the substrate if adhesion and flexibility are not great enough to offset
internall stresses. On the other hand, a thick film will provide more protection
againstt oxygen and moisture diffusion to the metal interface. In addition, layered
coatings,, especially when applied at 90 degree angles to each other, minimize the
effectt of certain other defects, such as pinholes, seeds, and craters. Thus, thickness
andd adhesion may be more property considered together as assets or weaknesses in
anyy system.
Thicknesss of coatings is thus an important aspect to monitor. In fact, this is
quitee easy to do, and a common understanding of actual thickness would
undoubtedlyy benefit coating application practices in conservation. The mean dry
filmfilm thickness on the two sample substrates was calculated as described in the
experimentall section, and is shown in Figure 8. Results illustrate that film
thickness,, as applied in common practice, varies a great deal from coating to
coating,, and often does not meet the manufacturer's recommendations. This is
true,, for example, of the Incralac coatings, most of which were about half of the

44 4
 PerformancePerformance of 29 coatings

1-lncralac c
2-BTA+lnc. .
3-lnc+wax x
4-BTA+lnc+wax x
5-lnc.. w/Tinuvin
6-lnc.. w/silane
7-silane+lnc. .
8-thickk Inc.
9-B-48 8
*10-CCRacrylic c
*11-sil.+CCRacrylic c
12-NKacrylic c
13-PPGacrylic c
14-PPGacr/au u
O. O. 15-NKacr/au u
EE 16-BTA+NKacr/au u
TO TO
17-sil+NKacr/au u
18-NKacr/au/wax x
19-NKau u
20-PPGau u
21-man.. pt+PPGau
22-sil.+NKau u
23-SiAlkyd d
24-CCRpolyur. .
25-silane e
26-SCwbau u
27-NKwbau u
28-wax x
29-BTA+wax x
30-uncoated d

0.00 0 0.50 0 1.00 0 1.500 2.00 2.50 0 3.00 0 3.5 5

Averagee Dry Film Thickness (mils)

*coatingg applied m o r e thickly to copper roof than bronze

Figuree 8 Average dry film thickness and variation of coatings on cast bronze and 50-
year-oldyear-old copper roof panels before weathering.

manufacturer'ss recommended thickness. When Incralac was applied close to

specificationss (#8), the result was aggravated orange peel and a poor appearance.
Forr this reason, the coating is generally applied quite thinly. However, results
showedd less corrosion occurred under the thicker coating on polished bronze after
weathering.. B-48 (#9) and the Nikolas waterborne coating (#27) were particularly
thinn coatings as applied, and appeared to suffer on polished bronze in accelerated

45 5
ChapterChapter 2

weatheringg as a consequence. The two wax coatings (#28, 29) were applied by an
experimentall spray technique, and resulted in much thicker films than can be
obtainedd by normal brushing and buffing procedures. Both aesthetic and
performancee results on the polished bronze were unacceptable, however. Thus, the
rotee of coating thickness in performance on bronze was not clear cut by itself
Onn copper roof patina, the partial saturation and thickness of the wax
coatingg appeared to significantly aid performance. Figure 8 illustrates patina
saturationn by the differences between bronze and copper roof coating film
thicknesss as compared to the thickness of the patina itself on the uncoated control.
Mostt coatings penetrated to some degree into the mineral layer, in effect creating a
polymer/minerall matrix. This appeared to help performance in general.
Ass shown in Figure 8, standard deviations (a) of coatings on the polished
bronzee indicate that most were fairly even. As applied by a skilled conservator,
standardd deviations were typically in the range of about 0.1-0.2 mils (2,5-5.1 um),
withh the exception of the brushed coatings (#23, 24), the StanChem waterborne
acrylicc urethane (#26), and the microcrystalline wax blend coatings (#28, 29).

2.3.9.2.2.3.9.2. Adhesion

Whilee dry film thickness is easily measured, adhesion of a coating to a

substratee is a relatively obscure quantity, which is as difficult to define as to
measure,, as discussed in Chapter 1. Results of cross-cut and X-cut adhesion tests
givee us a rough idea of net adhesion forces at work in a system. Indirectly, we get
ann idea of the irreversible effect of moisture penetration on adhesion at the
interfacee by measuring dry adhesion, i.e., at room conditions, after exposure to
moisturee in the weathering protocols.
2.3.9.2.12.3.9.2.1 Adhesion before weathering
Adhesionn test results are illustrated in Figures 9 and 10. General
comparisonn of the two sets of results shows that initial adhesion of coatings on
copperr roof panels was equal to or better than that on bronze for most samples,
withh a few exceptions. In contrast to polished bronze, the rough and porous
surfacee of the copper roof patina represents vastly increased surface area and
allowss for increased mechanical adhesion. For the solvent-borne coatings in
particular,, patina saturation also translated into better initial adhesion. On the
otherr hand, poor adhesion was particularly marked for all three waterborne
coatingss (#24, 26, 27) on copper roof patina. This may be ascribed to surface
tensionn effects, resulting in poor wetting.
Adhesionn on the bronze substrate was mixed. The B-48 coating series (#9-
12)) and silicone alkyd (#23) coatings showed excellent initial adhesion to both
bronzee and copper roof patina. The PPG acrylic (#13) and acrylic urethane (#20)
showedd particularly poor adhesion to bronze without any pretreatments, but
excellentt adhesion in the latter case after the PPG pretreatment (#21). This effect

46 6
 PerformancePerformance of 29 coatings

cc
0) )
4
**
uu
XX
LU U
II I 3 ii
in n

en n

ID D 22 -
CU CU
cc
oo
'(/)
7Z7Z) 1
CD D
sz sz
<<00 111 1
( M M - ï i n i D N C O C n OO
™™
unCDr^-CO^-cNjc\JCN(NCNCNJCMCMM

Coatingg Number (sample

DD no weathering
NB:: values of 0 reported as 0.1
DD after accelerated weathering
*#15-188 showed cohesive failure; hard to rate
after outdoor weathering

Figuree 9 Cross-cut adhesion ratings of coatings on polished, cast bronze. Phase I. before
andand after weathering.

(NnTintoNcooioo
Coatingg Number (sample)

NB:: values of 0 reported as 0.1 Dnoo weathering

*dataa not available for 11C, after outdoor DD after accelerated weathering
weathering g after outdoor weathering

Figuree 10 X-cut adhesion ratings of coatings on 50-year-old copper roof Phase I. before
andand after weathering.

4~ ~
ChapterChapter 2

wass not seen on the copper roof samples, and was most likely due to an observed
etchingg effect of the polished surface, which would both clean and roughen the
surfacee and allow greater adhesion. Results also show that the silane pretreatment
dramaticallyy improved adhesion of Incralac (#6) and the CCR acrylic (#11) to
bronze,, but had little effect on adhesion to the copper roof panels. The silane
pretreatmentt apparently prevented coatings from penetrating into the copper roof
patina,, as demonstrated by the relative thickness measurements (Figure 9). The
effectt of silane pretreatment could not be properly assessed by the cross-cut
adhesionn test on multi-component systems, which had mixed interlamellar and
wholee system failure. BTA showed no detectable effect on adhesion by this test
method,, including no apparent effect on patina penetration.

2.3.9.2.22.3.9.2.2 Adhesion after weathering

Cross-cutt and X-cut adhesion results for Phase I samples after weathering
(Figuress 9 and 10) show a fair degree of change on the polished bronze substrates
inn most cases, but relatively little change on the copper roof substrates. On both
substrates,, the worst case scenario was the StanChem waterborne acrylic urethane
(#26),, which fell below measurable levels after accelerated and outdoor
weathering.. In a few cases, adhesion actually increased on polished, cast bronze,
suchh as with the CCR waterborne polyurethane (#24), CCR acrylic (#10) and NK
acrylic/acrylicc urethane coatings (#15-18). The Nikolas two-part acrylic/acrylic
urethanee coatings showed some interlamellar adhesive failure, primarily before
weathering,, which decreased the initial ratings. Adhesion testing of an
unweatheredd sample coated with the CCR waterborne polyurethane (#24) also
showedd an increase in adhesion during storage. The adhesion in this case initially
measuredd zero, but over time, without exposure to weathering tests, changed to a
ratingg of five. It was observed that the polyurethane films became extremely
tough,, so that it was hard to cut through them with a scalpel, and therefore the test
wass difficult to conduct properly. For a discussion of apparently delayed curing
effectss in some coatings, see Chapter 5.
Adhesivee failure in the wax coatings on bronze (#28, 29) was somewhat
differentt in nature from the other coatings. Here failure was primarily cohesive,
i.e.,, in the coating layer itself vs. between the coating and substrate, complicating
interpretationn and inspection of the test results. After weathering, the wax coatings
alsoo showed a tendency toward increase in adhesion, most likely due to hardening
off the wax.
Thee Incralac coatings on bronze, except for Incralac with Tinuvin 292 (#5),
showedd some decreased adhesion after weathering, rather dramatically in some
casess of outdoor weathering. Adhesion of the thick Incralac coating (#8) showed
thee largest drop in magnitude after accelerated weathering. This may be attributed
too poorer mechanical response to thermal shocks during freeze-thaw cycling or to
residuall stress in the coating. It is interesting that these samples show a larger
decreasee in adhesion after outdoor vs. accelerated weathering. Adhesion loss was
alsoo shown to occur in B-48-based coatings, except #10, which showed odd and
unexplainablee behavior. The Nikolas acrylic retained adhesion values better than

48 8
 PerformancePerformance of 29 coatings

most.. Otherwise adhesion loss in the acrylic coatings on polished bronze appeared
too forecast their limited lifetime in outdoor exposures.
Contraryy to expectations, however, results show that the relationship of
coatingg thickness and adhesion to coating performance is not clear cut. The
correlationn between poor adhesion and poor performance in the StanChem
waterbomee acrylic urethane (#26) is obvious, despite the thick coating. On the
otherr hand, with B-48 (#9) on bronze, good adhesion did not translate into good
performance.. Here the weak link may have been the thinness of the film. Results
doo suggest partial correlations, as discussed above, but also highlight exceptions
thatt cannot be explained by coating thickness or adhesion. This was the case, for
example,, with the less than stellar performance of PPG acrylic urethane after the
manufacturer'ss pretreatment (#21), despite excellent adhesion, adequate thickness,
andd good overall film quality.
Onn copper roof patina, the acrylic coatings showed much improved
adhesionn characteristics, presumably due to mechanical adhesion contributions, as
discussedd above. In the case of B-48 (#9), however, adhesion virtually dropped to
aa rating of zero after outdoor weathering. Here, as well as with the waterbome
acrylicc urethane coatings, adhesion loss appears to be related to poor penetration
intoo the patina and/or poor wetting of the patina itself. In these cases, poor
adhesionn and poor performance show a direct correlation. Thus, the superior
performancee of most coatings on the naturally patinated copper roof compared to
thatt observed on polished bronze may be attributed to the formation of a mineral-
polymerr matrix with good mechanical adhesion properties and general
reinforcementt of the passivating corrosion layer. It must be emphasized, however,
thatt appearance of many of these saturated and/or glossy green patinas would be
deemedd unacceptable from an aesthetic point of view.

2.4.2.4. Conclusions
Thee results described in this chapter first point out different physical
attributess of coatings on the bronze and copper substrates. Simple comparison of
sampless after coating application emphasized that aesthetic qualities of coatings
varyy widely across the board, particularly when applied over an existing patina.
Forr this study, a good aesthetic judgment was associated with as little visual
changee as possible to the substrate and coating. The different colors resulting on
coated,, patinated surfaces relate to varying degrees of penetration of the coating
intoo the patina, where darker colors correspond to saturation of the patina. This
alsoo corresponds to formation of a polymer/mineral matrix, which improved the
adhesivee and weathering properties of most solvent-borne coating systems during
weathering. .
Resultss also underscore that within broad classes of coatings, any two
commerciall coatings may differ in terms of performance. This may occur due to
additives,, as well as the properties of the base resin and quality of manufacture. In
particular,, it was clear that formulations and properties varied significantly

49 9
ChapterChapter 2

betweenn commercial acrylic urethanes, in some cases to the detriment of

performancee under the test conditions. Similarly, although the silane pretreatment
usedd in this study failed, silanes are a broad class of chemicals, and it is clear that
suchh a pretreatment must be tailored to any one application in order to gauge its
truee potential in the field.
Inn general, this survey of coatings does not reveal dominant trends that can
mapp out property-performance relationships as a rule. Rather, results point to
coatingg performance as a complex puzzle of interdependent factors, including
coatingg thickness, water and oxygen permeability (related to thickness), adhesion,
coatingg stability, and the presence of defects in the coating film. In addition, the
studyy points out the difficulties in obtaining good correlation between accelerated
weatheringg and outdoor weathering testing. Although it is not possible to make a
specificc correlation between accelerated and natural weathering in this study, it
wouldd not be unreasonable based on the performance of Incralac, which is
expectedd to perform well outdoors from 3-5 years, to estimate that the accelerated
weatheringg program was approximately equivalent to five to seven years of
outdoorr weathering.
Phasee I helped to establish the guidelines for further investigations
presentedd in the following chapters. Several model coating systems were singled
outt as candidates for future study, including: 1} acrylic/acrylic urethane coatings;
2)) Incralac plus a wax topcoat; 3) a thicker version of the Nikolas waterborne
coating;; and 4) commercial acrylic urethanes. Although the development of an
appropriatee silane coupling agent was deemed outside the scope and resources of
thiss study, unsolved questions surrounding the role of BTA in coating performance
weree targeted for further study. These investigations are described in Chapter 4.

50 0
 PerformancePerformance of 29 coatings

References References

11 H. RÖmich, et al., "New conservation methods for outdoor bronze sculptures,"

ProtectionProtection and conservation of the European cultural heritage. Research report
No.No. 3 (EUR 16637 EN, European Commission, Belgium, 1996).
22 L. Brostoff and R. de la Rie, "Research into Protective Coating Systems for
Outdoorr Bronze Sculpture and Ornamentation, Phase I" (PTTPublications No.
1997-03,, NCPTT, Natchitoches, Louisiana, 1997).
33 W. H. Finkeldey, "Report of Subcommittee VI on Atmosphere Corrosion of Non-
ferrouss Metals and Alloys," Proceedings of the American Society for Testing and
MaterialsMaterials (ASTM, 1933), 234-251.
44 Various authors, "Special Issue, Copper Patina Formation," Corrosion Science
27,7(1987). .
55 C. H. Simpson, et al., "Accelerated Corrosion Testing of Industrial Maintenance
Paintss Using a Cyclic Corrosion Weathering Method," Journal of Protective
CoatingsCoatings & Linings 8, 5 (May 1991), 29-36.
66 Jeff Andrews, et al., "Correlation of Accelerated Exposure Testing and Exterior
Exposuree Sites," Journal of Coatings Technology 66, 837 (Oct. 1994), 49-67.
77 Technical Data Bulletin, 11565 OD Lacquer RFU, G.J. Nikolas & Co., Inc.,
5/24/96. .
88 Concept first suggested by Andrew Lins, unpublished notes, roundtable
discussionn at the National Gallery of Art, Washington, DC, November 1994.
99 Private communication from Mr. Buck Nikolas, G.J. Nikolas & Co., 1996.
100 This wax recipe was formulated and generously supplied by Nicolas Veloz,
privatee conservator. The formula is similar to one used by private conservator
Stevee Tatti, and is generally popular among U.S. conservators.
111 Silquest A-l 100® Silane, recommended by and purchased from OSi Industries,
currentlyy OSi Specialties, part of Crompton Corp., South Charleston, WV.
122 Applied by Andrew Baxter, private conservator (Bronze, et al.).
133 Technique developed and applied by Nicolas Veloz, private conservator.
144 "Standard Test Method for Humid-Dry Cycling for Coatings on Wood and Wood
Products,"" in 1990 Annual Book of ASTM Standards, Vol. 6.02 (American
Societyy for Testing and Materials, 1990), 178-179.
155 "Standard Test Method for Freeze-Thaw Resistance of Water-Bome Coatings," in
19901990 Annual Book of ASTM Standards, Vol. 6.02 (American Society for Testing
andd Materials, 1990), 82-83.
166 "Accelerated Exposure of Automotive Exterior Materials Using a Controlled
Irradiancee Water Cooled Xenon Arc Apparatus," (Society of American Engineers
(SAE),Junee 1989).
177 Conversation with Dr. Richard Granada, Lehigh University, 1995.

51 1
ChapterChapter 2

188 "Standard Practice for Atmospheric Environmental Exposure Testing of

Nonmetallicc Materials," in 1993 Annual Book ofASTM Standards, Vol. 6.01, ed.
Paulaa C. Fazio, et al. (American Society for Testing and Materials, Philadelphia,
PA,, 1993), 1008-1011.
199 "Standard Practice for Operating Light-Exposure Apparatus (Xenon-Arc) With
andd Without Water for Exposure of Nonmetallic Materials," in 1993 Annual Book
ofof ASTM Standards, Vol. 6.01, ed. Paula C. Fazio, et al. (American Society for
Testingg and Materials, Philadelphia, PA, 1993), 1033-1041.
200 K. Nassau, A. E. Miller and T. E. Graedel, "The Reaction of Simulated Rain with
Copper,, Copper Patina, and some Copper Compounds," Corrosion Science 27, 7
(1987),, 703-71.
211 Percy E. Pierce and Clifford K. Schoff, Coating Film Defects (Federation Series
off Coatings Technology, Philadelphia, PA, January 1988).
222 "Standard Test Method for Nondestructive Measurement of Dry Film Thickness
off Nonconductive Coatings Applied to a Nonferrous Metal Base," in 1993 Annual
BookBook of ASTM Standards, Vol. 6.01, ed. Paula C. Fazio, et al. (American Society
forr Testing and Materials, Philadelphia, PA, 1993), 222-223.
233 "Standard Test Methods for Measuring Adhesion by Tape Test," in 1993 Annual
BookBook of ASTM Standards, Vol. 6.01, ed. Paula C. Fazio, et al. (American Society
forr Testing and Materials, Philadelphia, PA, 1993), 447-450.
244 "Standard Test Method for Evaluation of Painted or Coated Specimens Subjected
too Corrosive Environments," in 1993 Annual Book ofASTM Standards, Vol. 6.01,
ed.. Paula C. Fazio, et al. (American Society for Testing and Materials,
Philadelphia,, PA, 1993), 271-273.
255 "Standard Test Method for Evaluating Degree of Blistering of Paints," in 1993
AnnualAnnual Book ofASTM Standards, Vol. 6.01, ed. Paula C. Fazio, et al. (American
Societyy for Testing and Materials, Philadelphia, PA, 1993), 121-126.
266 "Standard Test Method for Evaluating Degree of Checking of Exterior Paints," in
19931993 Annual Book of ASTM Standards, Vol. 6.01, ed. Paula C. Fazio, et al.
(Americann Society for Testing and Materials, Philadelphia, PA, 1993), 107-116.
277 "Standard Test Method for Evaluating Degree of Rusting on Painted Steel
Surfaces,"" in 1999 Annual Book ofASTM Standards, Vol. 6.02, ed. Robert F.
Allen,, et al. (American Society for Testing and Materials, Philadelphia, PA,
1999),, 13-15.
288 Note that there is no sound scientific basis for weighting factors related to
mechanismss of corrosion and coating failure. Cf. also Edward C. Ferlauto, et al.,
"Selectionn of Corrosion Test Methods Based on Mechanism Principles," J.
CoatingsCoatings Technology 66, 835 (August, 1994), 85-94.
299 T. E. Graedel, K. Nassau and J. P. Franey, "Copper Patinas Formed in the
Atmosphere-I,"" Corrosion Science 27, 7 (1987), 639-657.
300 L. S. Selwyn, et al., "Outdoor Bronze Statues: Analysis of Metal and Surface
Samples,"" Studies in Conservation 41 (1996), 205-228.

52 2
 PerformancePerformance of 29 coatings

311 K. Nassau and P. K, Gallagher, "The Characterization of Patina Components by

X-rayy Diffraction and Evolved Gas Analysis," Corrosion Science 27, 7 (1987),
669-684. .
322 Andrew Lins and Tracy Power, "The Corrosion of Bronze Monuments in Polluted
Urbann Sites: A Report on the Stability of Copper Mineral Species at Different pH
Levels,"" in Ancient and Historic Metals, ed. D. A. Scott, et al. (Getty
Conservationn Institute, 1994), 119-151.
333 Private communication from Mr. Herb Petty at OSi Specialties, 1996, who
revealedd that this silane also commonly works and is sold as a crosslinking agent
inn coatings and is known to cause yellowing. It therefore was not an appropriate
choicee in this study despite OSi's previous recommendation for use in this study.
344 Molecular weight of the dispersion polymers may be assumed to be very high, as
iss typical in these types of preparations; Mw of B-48 has been reported to be
184,000,, compared to 105,00 for B-44, cf. M. Lazzri, O. Chiantore, "Thermal-
ageingg of paraloid acrylic protective polymers," Polymer 41 (2000), 6447-6455.

53 3
ChapterChapter 2

54 4
 PerformancePerformance of 29 coatings

Platee I Phase I, Set IB: polished, cast bronze with (reading left to right) coatings
#1-29,#1-29, plus uncoated control, after extended accelerated weathering.

55 5
 ChapterChapter 2

II 56

^TttriniHr^7*^'5^ ^
Hit^^^mÊiïB*BÈUt^^^^^a*m^ÊiiiammBÊÈSmSB^ Hit^^^mÊiïB*BÈUt^^^^^a*m^ÊiiiammBÊÈSmSB^
 33
Performancee of five coatings on four types of
copperr alloy substrates

Abstract Abstract
Coatingg analysis undertaken in Phase I of the project provided a systematic
evaluationn of 29 coatings applied to bronze and naturally patinated copper
substrates.. In Phase II of the project, described in this chapter, five of the most
effectivee and relevant coatings were chosen for further study. The coating systems
weree modified in design as deemed appropriate, and were: 1) Incralac with a wax
topcoat;; 2) benzotriazole (BTA) pretreatment plus wax; 3) Nikolas 11565 acrylic
undercoat,, with Nikolas 9778 acrylic urethane middle coat, and a wax topcoat; 4)
BTAA pretreatment plus BASF 923-85 acrylic urethane, with a wax topcoat; and 5)
two-layerr Nikolas 11650 waterborne acrylic emulsion/urethane dispersion with
waxx topcoat. The model substrates were expanded and included: I) satin-finished,
castt monumental bronze, II) artificially patinated, cast monumental bronze, III)
naturallyy patinated 50-year-old copper roof, and IV) pressure-blasted, naturally
patinatedd 50-year-old copper roof. Samples underwent accelerated and natural,
outdoorr weathering testing. Corrosion on uncoated samples was analyzed, and
overalll failure ratings, film thickness, and adhesion ratings were compiled as in
Phasee I with the aid of digital image analysis.

3.1.3.1. Introduction
Resultss of Phase I, presented in the previous chapter, established the
criteriaa for selection of five types of coatings for further study in Phase II of the
researchh [1]. These coating types showed the most promise or relevance in
practicall conservation use and were singled out with the aim of further studying
andd comparing their performance on an expanded range of representative model
substrates,, as follows: 1) satin-finished, cast monumental bronze; II) artificially
patinated,, cast monumental bronze; III) brochantite patina that had formed over
roughlyy 50 years on copper roof; and IV) the same copper roof material after
walnutt shell-blasting. The five coating systems, applied in accordance with field
practices,, were: 1) two coats of Incralac with a wax topcoat, 2) benzotriazole

57 7
ChapterChapter 3

(BTA)) pretreatment plus two coats of a microcrystalline wax blend, 3) an

acrylic/acrylicc urethane/wax layered system (G.J. Nikolas & Co. coatings), 4) BTA
pretreatmentt plus BASF acrylic urethane and wax topcoat, and 5) a two-coat
Nikolass waterborne acrylic emulsion/urethane dispersion plus wax topcoat. The
BASFF acrylic urethane was included in Phase II in order to vary manufacturers
amongg the five coatings, and as a representative, viable acrylic urethane coating
thatt has been used in the field with some success.
Byy choosing a limited number of coatings for study on more substrates,
Phasee II of the research focused more closely on the role of substrate in coating
performance.. Experiments were also designed to repeat and clarify some of the
questionss raised in Phase I, including whether applying some coatings more
thicklyy could significantly boost performance. Samples were divided into three
groupss in order to follow both accelerated and natural, outdoor weathering against
controls.. The automotive-type, cyclic accelerated weathering regimen used in
Phasee I was somewhat harsher in Phase II in order to increase the acceleration
factor.. Sample analysis methodology in Phase II was additionally aided by digital
imagee analysis.

3.2.3.2. Experimental methods

Analyticall methods are generally the same as those used in Phase I, and are
describedd in Chapter 2. Particulars relating to Phase II only are as follows.

3.2.1.3.2.1. Substrates
Thee bronze substrates were a "monumental" bronze alloy (analysis below),
castt in 9" x 5" plaques by Bedi-Makky Foundry in Brooklyn, NY. Half of the
plaquess were polished at the foundry to a satin finish, using a water medium in
orderr to avoid fatty residue from jewelers' rouge polishing. The remaining plaques
weree wheel-polished to 120 grit, then artificially patinated to a warm, medium
brownn color. The patina was commissioned as a typical potassium sulfide (liver of
sulfur)/ferricc nitrate patina, but was later reported to have been produced by a one-
stepp process with heat using a dilute solution of ferric nitrate in tap water with one
orr two drops of cupric nitrate, and no rinsing. In order to accommodate all
sampless together in the weatherometer and Tenney chamber during testing,
sampless were cut into 3" x 5" plaques. The polished and artificially patinated
bronzee samples constituted Sets I and II.
Thee remaining sample substrates were coupons of naturally patinated, 50-
year-oldd copper from the Library of Congress in Washington, DC. These
substratess were also used in Phase I and are described in Chapter 2. Half of the
copperr roof samples were pressure-blasted with Agrashell ground English walnut
shells,, AD-10.5B mesh, using about 40 psi at a slight angle, roughly 8-10 inches
fromm the surface [2], This is consistent with normal conservation practice in the
Unitedd States. Agrashell reports that the shell oil content is 0.5% by weight.

58 8
 PerformancePerformance offivecoatings on four substrate

Thee number of substrates included in Phase II totaled 122. The four

substrates,, Sets I-IV, were divided into three groups each, where A were used as
controls,, group B was exposed to accelerated weathering, and group C was
exposedd to natural, outdoor weathering. Enough samples were commissioned to
conductt accelerated testing in triplicate, so that there were 30 samples in each set,
i.e.,, five samples for each coating system, and five uncoated samples. Two extra
sampless were added to Set IB; these are described below but were not included in
performancee results.
Thee substrates were examined by scanning electron microscopy (SEM) and
energyy dispersive spectrometry (EDS), and selected corrosion products and patinas
weree analyzed by powder X-ray diffraction (XRD), as described in Chapter 2.
Additionall open architecture XRD analysis was run by Dr. George Wheeler, the
Metropolitann Museum of Art, New York, NY, on a Philips 1710 open-architecture
diffractometerr with Cu Ka radiation acquired at 30 mA, 40 kV, 0.02 °/sec. In the
latterr system, peaks were recorded and marked from the 1710 by Sietronic
Softwaree and transferred to the Fein-Marquart software program.
Gass chromatography (GC) was used to analyze derivatized extracts from
threee samples: size AD 3B English walnut shells, the copper patina, and the walnut
shell-blastedd copper patina. Each of the powdered samples was immersed
overnightt in methylene chloride to extract the fatty material. The solvent extract
wass filtered through a 0.45 um filter and reduced under a nitrogen stream to an
appropriatee concentration. According to the method of White and Pile, 25 ul of a
solutionn of 5% TMTFTH ((m-trifluoromethylphenyl)trimethylamrnonium
hydroxide)) in methanol was added to the extract in order to convert the fatty acids,
inn both free acid and ester-bound forms, into quaternary ammonium salts, which
decomposee into methyl esters upon heating [3,4]. An aliquot of the solution was
directlyy injected into a Perkin Elmer Autosystem GC, in splitless mode, with a
Restekk Rtx-1 column and flame ionization detector. The temperature program
was:: 50 °C for 0.5 minutes, increased to 100 °C at a rate of 25 °C/min., and
increasedd to 280 °C at 6°/min., and held there for 5 minutes. The injector
temperaturee was 300 °C, and the detector temperature was 325 °C. Data was
processedd using Turbochrome software.

3.2.2.3.2.2. Substrate preparation

Alll substrates underwent cleaning prior to coating. The polished, cast
bronzee substrates were immersed in a xylene bath, then wiped with xylene. This
wass followed by an acetone bath, then alternating wipes with acetone, ethanol and
AA cry li-clean®,* until the substrate passed the water drop break test [5], The
artificiallyy patinated bronze coupons received only brief acetone baths, followed
byy gentle wiping with acetone and ethanol, since a large amount of loose brown
patinaa was otherwise removed. Unblasted copper roof substrates were bathed and

** Acryli-clean® wax and grease remover DX330 is a commercial solvent from PPG
Industries,, Inc. and is made specifically for metal degreasing before lacquering.

59 9
ChapterChapter 3

lightlyy scrubbed in distilled water with a small amount of Triton-X surfactant

added,, then rinsed and air-dried. Both blasted and unblasted copper roof samples
weree additionally soaked for about 3 minutes in a xylene bath, then washed briefly
inn acetone, followed by wiping with acetone and ethanol. All solvents were
reagentt grade.

3.2.3.3.2.3. Coating application

Coatingss are described in Chapter 2, except for #4, the BASF Glassodur-
MSS Top Clear 923-85. This is a two-component, high solid acrylic urethane
formulatedd from a proprietary acrylic copolymer crosslinked with a trimer of
hexamethylenediisocyanatee (HDI). The manufacturer reports that this coating
containss BTA, although this was not confirmed. The coating was developed for
thee automotive refinishing industry, and has also seen limited application in the
conservationn of outdoor metal objects.
Inn preparation for coating, substrates 1-IV were each divided into six
subsets:: one subset for each coating and one uncoated subset. Two of the subsets
weree pretreated with BTA by applying several brush coats of 1.5% (wt)
BTA/ethanoll solution, followed by air-drying and rinsing with ethanol. The five
sampless in each subset were lined up vertically and spray-coated together in order
too achieve uniformity within groups A, B, and C. Experienced outdoor sculpture
conservatorss applied all coatings [6]. The spray pressure used was about 35 psi in
alll cases. Consistent with normal practices in the field, Incralac coatings were
appliedd after 60:40 dilution by volume in xylene. In order to achieve good
appearancee on the patinated substrates, the base coat of Incralac was brush coated,
andd followed by a second spray coat. On the polished bronze, both coats of
Incralacc were spray-applied.
Alll other coatings were spray-applied as per manufacturer
recommendations.. The Nikolas acrylic was applied as one coat without reduction,
andd the Nikolas 9778 was mixed parts A:B in a ratio of 9:1, reduced 50% in
speciall thinner, and applied in a single coat over the acrylic after 30 minutes. The
BASFF acrylic urethane was mixed parts A:B in a 2:1 ratio, reduced 50%, and
spray-appliedd after a final wipe on the substrates with the BASF reducer. The
Nikolass 11650 was sprayed as a single coat after reducing 5% in distilled,
deionizedd water; a second coat, reduced 5% in distilled, deionized water, was
brush-coatedd on top about one week later.
Afterr sufficient curing for 1 to 2 weeks, all samples were cold-waxed.
Consistentt with normal practice in the field, wax was applied in two brush coats
andd followed by light buffing. This method was also used for the wax coating
(#2).. One extra polished bronze sample was coated with Incralac and left unwaxed
forr comparison; this was included in accelerated weathering. Another extra
polishedd bronze sample was treated with an experimental silane by W.J. van Ooij
att the University of Cincinnati, Ohio, and also included in accelerated weathering.
Thee silane-coatcd sample did not do well in testing and was omitted from the
results. .

60 0
 PerformancePerformance of five coatings on four substrates

Beforee weathering, coatings were characterized under tow magnification

forr overall film quality in terms of the presence of seeds, orange peel, air
entrapment,, solvent popping, hazing or whitening, bumps and sinks, cratering, and
crawling/dewetting,, according to common industrial definitions [7], Dry film
thicknesss and its standard deviation (ASTM D1400 [8]) were determined on
samplee groups A (unweathered) and B (after accelerated weathering) using an
Elcometerr 345 dry film thickness gauge. Ten to twenty readings were taken on
eachh sample. The bronze coating measurements were subject to an error of 0.1
mill (2.5 microns), and thus coatings or patinas less than about 0.2 mils (5.1 um)
couldd not be measured accurately. The copper roof samples had a large margin of
error,, which could not be determined due to the extremely rough and uneven
patina. .
Beforee weathering, sample groups B and C were scribed through the
coatingss on the bottom half of the sample with an "X" in preparation for creepage
corrosionn measurement. Sample group A was evaluated for coating adhesion by
thee ASTM D3359 [9] crosscut or X-cut adhesion tape test, as were sample groups
BB and C after weathering. Results of the control group A were used to represent
adhesionn before weathering, since the test cuts would interfere with weathering
results.. For all adhesion tests, a small area of wax was cleaned off in order to test
adhesionn only of the underlying coating system.
AA small group of additional samples were prepared by draw-down coating
applicationn on solvent-cleaned, rolled bronze coupons. Because of the varying
viscosityy of test coatings, uniform thickness between coatings was difficult to
achieve.. Additional samples of coated glass slides and aluminized mirror slides
weree also prepared for accelerated weathering. These samples all experienced
rapidd deterioration during accelerated weathering and were removed after 46 days
duee to poor condition.

3.2.4.3.2.4. Weathering
Acceleratedd weathering of sample group B consisted of an automotive-
type,, accelerated weathering program, modified from ASTM standard test
proceduress and SAE standards [ 10,11,12] for an Atlas Ci65a xenon-arc
weatherometerr and a Tenney Thirty temperature/humidity test chamber. This
weatheringg regimen was modified from that used in Phase I (Chapter 2) in order to
increasee the acceleration factor. Total accelerated weathering time was 120 days,
consistingg of two types of exposure: 1) constant light and intermittent back- or
specimen-sprayingg in the weatherometer, and 2) alternating humidity cycling and
freeze-thaww cycling in the Tenney, including daily, manual spraying during the
weekdayss with a concentrated "rain" solution. Programming cycles in the two test
chamberss were similar to those used in Phase I, as shown schematically in Chapter
2,, Figures 1 and 2. In order to increase "time of wetness" following spraying with
thee rain solution in the Tenney chamber, conditions in Phase II weathering were
heldd at 23 °C and 80% RH for a period of one to three hours. Otherwise,
conditionss in the Tenney were cycled between 20 and 85 % RH, or between (-15)
°CC and 60 °C, in four hour segments.

61 1
ChapterChapter 3

Inn the first 27 days of Phase II testing, samples were alternated between
Tenneyy and weatherometer exposures, and the spray solution average ion
concentrationn was, in meq/1: [S04=]=12.16; [N0 3 ]=4.81; [NH4+j=14.65; [CI"
]=34.38;; [Na"]=34.34. The pH of this solution was 4.2. The subsequent 37 days
off weathering consisted of weatherometer exposure only, with deionized water
spraying.. Average conditions in the weatherometer were: black panel 60.5 °C,
relativee humidity 23-44 %, irradiance 0.55 W/rrf @ 340 nm. The final 56 days of
weatheringg were Tenney exposure only. The medium chloride rain solution used
heree was reformulated to represent approximately lOOx the maximum
concentrationn of common ions reported for typical acid rain [13,14], but with a
lowerr chloride content than is normally used in accelerated weathering studies
[15].. Total average ion concentration was, in meq/1: [S04~ ]=15.30; [NO3~]=4.80;
[NH4+]=4.18;; [Cl>22.07; [Na+]=35.37. The average pH of this solution was 3.64.
Sampless were thoroughly rinsed with room temperature supermillipore water after
weatheringg and conditioned to room temperature and humidity.
Phasee II sample group C was naturally weathered on the roof of the
Nationall Gallery for just over one year. Time constraints did not allow longer
outdoorr weathering. The samples were placed facing due south at a 45° angle
accordingg to ASTM G7 [16], as in Phase I.

3.2.5.3.2.5. Performance assessment

Ass in Phase I, performance was broken down into three categories: coating
condition,, overall corrosion, and mean linear creepage corrosion at a scribemark.
Eachh assessment category contributed to one-third of the performance rating. The
"totall failure ratings" were calculated by the same method used in Phase I and are
describedd in the experimental section of Chapter 2.
Inn Phase II, some improvements were made in the evaluation methods for
obtainingg individual ratings. Coating performance was assessed by the naked eye,
ass well as by a HunterLab Ultra Scan XE spectrophotometer. However, color
changee during weathering of a clear coating over metal involves change in the
coatingg (usually yellowing), as well as the change in the metal surface from
oxidation.. Therefore interpretation of the color data was not straightforward, and
visuall evaluation of color change in the coating layer was used in the final ratings.
Forr the overall corrosion category rating, digital imaging and analysis were
usedd to aid visual evaluation of overall pitting and creepage corrosion at the
scribemark.. Photographs of the bronze and copper samples (group A, and groups
BB and C after weathering) were captured and scanned as 32-bit images. All images
includedd a reference color strip and ruler. The digital images were analyzed with
Scanalytics,, Inc. IPLab software for Macintosh, version 3.2. For overall pitting
corrosionn assessment, a set area on the weathered samples was defined according
too value, hue, intensity, saturation, or the color parameters red, green, blue, yellow,
magenta,, and cyan. This software allows only three parameters to be combined at
onee time in order to define a specific color, so that the range of defined colors was
quitee limited. This resulted in poor differentiation of corrosion, which was not
uniformm in color. In addition, the patinated surfaces were mottled in coloring

62 2
 PerformancePerformance of five coatings on four substrate

similarr to the corrosion itself. Although individual algorithms for each of the four
typess of bronze and copper roof substrates were written, this method of digital
analysiss was successful only for the polished bronze substrates, which were
uniformm in color apart from the corrosion.
Digitall images of the remaining patinated substrates could be examined
convenientlyy by zooming in and out, and were used as checks for obtaining the
ASTMM D610-defined value for percent overall pitting corrosion [17], translated
intoo a 0-5 scale. This value was adjusted to include overall change from surface
oxidation,, as necessary. This combined method led to improved consistency over
thee estimation method used in Phase I.
Inn contrast to overall corrosion evaluation, the digital image analysis
softwaree unequivocally improved the precision in mean linear creepage corrosion
measurementt at a scribemark (ASTM D1654 [18]). For this analysis, a one-pixel
linee was superimposed on the image of each side of the "X" scribe. A "measure
length1'' function was used to draw perpendicular lines from the scribe mark to the
edgee of the corrosion which radiates out along the scribe. This was done on either
sidee of the scribe line at eleven evenly spaced positions, for a total of 88 readings
perr sample. The lengths of the lines were calculated in millimeters and averaged.
Thee maximum mean linear creepage corrosion for all samples was 3.76 mm,
althoughh individual values ranged up to about 10 mm where areas of corrosion
mergedd near the centers of the "X" mark. Thus millimeters of creepage corrosion
naturallyy lent itself to a 0-10 scale.
Althoughh the maximum possible combined failure rating by this method is
99,, actual values only approached 60. As in Phase I, creepage corrosion ratings
tendedd to contribute about 10 % of total failure rather than 33 %, keeping the total
ratingg values low. In addition, the coating system tended to contribute to less than
aa third of the total failure, since failure was not seen in every coating category.
Resultss in Phase II were thus consistent with Phase I in that "good performance" in
conservationn terms, i.e., almost no visible change, remained at a total value of
aboutt 10 or less.

33.33. Results and discussion

33.1.33.1. Substrate characterization

SEMM imaging of the polished, cast bronze surface, as seen in Figure la,
showss a fair amount of scratches and pits. XRD open architecture analysis
detectedd copper tin/copper zinc alloy phases and lead phases separately. This
concurss with the backscattering SEM image of the bronze (not shown), which
revealedd the lead precipitated in small globs throughout a two-part matrix, similar
too the bronze used in Phase I. XRD also identified the presence of two tin oxides
(PDFF file #33-1374, 25-1259).

63 3
ChapterChapter 3

SEMM of the artificial patina on bronze samples (Figure lb) revealed a

patchy,, finely cracked mineral layer. This fissuring was finer in some areas than in
others.. EDS analysis of the artificial patina showed the presence of iron spread
unevenlyy across the surface. No chlorine or potassium was detected. The presence
off sulfur could not be determined due to overlapping of the metal peaks.
Aluminumm and silicon were also detected; these are common contaminants, most
likelyy found in the foundry environment. The average thickness of the patina was
<< 0.2 mils (5.1 (am), which is less than experimental error for measurement using
ann Elcometer.
XRDD open architecture analysis of the artificial patina, sampled by scraping
offf the metal, yielded a clear pattern which could be reasonably identified through
thee International PDF files with two phases: ammonium iron sulfate
[(NH4bFe(S04h]] (# 3-0043) and copper nitrate hydroxide [Cu2(OH)3N03] (#15-
0014).. These minerals account for most, but not all the peaks in the pattern. Iron
oxide/hydroxidess (reddish brown) were expected, but could not be convincingly
matched,, although this does not rule out their presence in hydrated or poorly
crystallizedd forms. These identifications imply that ammonium sulfide, a common
patinizingg agent, was used or present as a contaminant in the solution or the brush.
Analysiss and imaging of the copper roof substrate is described in Chapter 2
andd shown again in Figure 2a. These substrates appeared very changed after
blastingg with walnut shells. The color of the remaining patina was a much darker
greenn overall; the yellowish spots and tiny black dots from the original patina still
remained.. SEM of the walnut shell-blasted copper roof (Figure 2b) shows a
surfacee in which the voids in the top mineral layer now appear exaggerated. The
minerall layer also appears denser and more compact or matted down, with fine
crackingg visible in spots. EDS results for blasted copper roof coupons, both with
andd without subsequent solvent cleaning, matched those for the unblasted surface
withh the notable exception of a carbon peak in both spectra. This evidence appears
too confirm suspicions that walnut shell blasting leaves oily deposits which arc not
easilyy removed from the surface of treated metals.
Additionall support for the presence of residue from blasting was difficult to
detect,, however. No walnut oil was detected either by FTIR of the patina or by GC
off patina extracts. While GC positively identified walnut oil from extractions of
thee shells used for blasting, results from patina extractions both before and after
blastingg contained many acids, so that positive identification of walnut oil was not
possiblee by this method.
Surprisingly,, there was no measurable difference in the average patina
thicknesss on substrates with or without walnut shell blasting using the Elcometer
gauge.. For both types of copper roof substrates the average patina thickness
readingg was 0.70 mils (17.8 p.m) 0.1. However, open architecture XRD analysis
off the patinas on the respective substrates detected not only brochantite, but also a
substantiall amount of copper and cuprite in the case of the blasted copper roof
Thiss implies that the patina was somewhat thinner on the blasted surface, enabling
thee X-rays to penetrate more deeply into the metal.

64 4
 PerformancePerformance of five coatings on four substrates

Figuree 1 SEM photographs, normal image of A) polished, cast bronze (500x), and
B)) artificially patinated bronze (600x).

65 5
ChapterChapter 3

Figuree 2 SEM photographs, normal image of A) 50-year-old copper roof patina (600.x).
andand li) 50-year-old copper roof patina, walnut shell-blasted (300.x)**.

Notee that charging effect on this substrate did not allow higher magnification.

66 6
 PerformancePerformance of five coatings on four substrates

3.3.2.3.3.2. Corrosion on uncoated substrates after accelerated weathering

XRDD analysis of corrosion products on the various substrates after
acceleratedd weathering is summarized in Table I and described below. In general,
corrosionn products on the polished bronze and natural brochantite patina were as
expected,, with a few exceptions. Differences from Phase I results appeared to
reflectt changes in the accelerated weathering rain solution in Phase II. Results
variedd slightly on the other two substrates, and also indicated that BTA treatment
providedd limited protection, in particular against chloride corrosion.

3.3.2.1.3.3.2.1. Polished, cast bronze

Brown,, black and green corrosion which formed on uncoated, polished cast
bronzee samples after only 21 days of accelerated weathering was analyzed by XRD
powderr diffraction. The pattern provided matches for the following phases: cuprite
(CU2O),, tenorite (CuO), copper hydroxy chlorides (atacamite, paratacamite,
botallackite),, copper hydroxy sulfates (including brochantite-M, brochantite-O,
andd possibly posnjakite), and nantokite (CuCl). Lines remaining in the pattern
suggestedd the presence of various hydrated copper chloride sulfate hydroxides,
zincc sulfate hydroxides (white), and/or copper zinc sulfates, without firm
identifications.. The peak at 6.90 A, along with other lines in this pattern, provided
aa fairly good match with copper nitrate hydroxide (green). FTIR of the corrosion
(nott shown) confirmed the presence of atacamite, as well as sulfates and nitrates.
Thesee analyses do not rule out either the presence of other copper chlorides,
includingg zincian paratacamite, nor copper sulfides. The latter, which are mostly
blackk or brown, do not crystallize well and are difficult to identify.
Afterr completion of accelerated weathering, XRD open architecture
analysiss of the intact sample and of corrosion scraped from the surface showed
somee differences from earlier powder diffraction results. Identifiable phases
included:: cuprite (red); tenorite (black); and chlorides, including
atacamite/paratacamitee and botallackite (green/black), and, tentatively, zinc
chloridee sulfate hydroxide hydrate (white). Copper sulfates were not positively
identifiedd at this point in the weathering. The substrate treated with BTA appeared
lesss corroded after accelerated weathering, notably in terms of black corrosion.
However,, open architecture XRD analysis of this bronze coupon yielded almost
identicall results, but with additional peaks at d spacing = 6.54 and 3.22 A. On the
unweatheredd bronze, the latter peak was attributed to tin oxide; this supports visual
evidencee that a thinner corrosion layer formed during weathering on BTA-treated
bronze. .

3.3.2.2.3.3.2.2. Artificially patinated, cast bronze

Analysiss of the patinated bronze after accelerated weathering showed that

thee patina had changed dramatically in terms of chemical makeup as well as color,
havingg largely turned a gray-black color with a purplish reflection and light green
spots.. The original brown remained at the sample edges only, and neither the
originall copper nitrate hydroxide nor the ammonium iron sulfate was detected by

67 7
ChapterChapter 3

openn architecture XRD (whole coupon, no BTA). The pattern obtained for this
substratee had much in common with that of the unpolished bronze, except for the
absencee of tenorite and a peak near rf=4.34 A. The latter peak may have been
associatedd with the patina or ammonium sulfate. The metallic character of the
blackk color also suggested the presence of sulfides. The sample treated with BTA
againn appeared somewhat less corroded, with more brown remaining. Analysis
confirmedd the continued presence of copper nitrate hydroxide, but revealed new,
unassignedd peaks at d spacing = 12.59 and 4.36 A, as well as the absence of
botallackite. .

3.3.2.3.3.3.2.3. 50-year-old copper roof, natural and walnut shell-blasted patinas

Openn architecture XRD analysis of an uncoated, natural copper roof patina

couponn after accelerated weathering showed several new peaks and changes in
peakk intensities. These results indicated new cuprite and brochantite formation, as
welll as some copper chlorides, including atacamite and possibly ammonium
copperr chloride hydrate (blue). The BTA-treated coupon appeared less corroded
inn terms of black products, and no chlorides were detected.
Thee uncoated, walnut shell-blasted copper roof substrates appeared
somewhatt different from the unblasted samples after accelerated weathering:
Corrosionn was generally spottier, and there were more areas of bright green. XRD
analysiss showed evidence of new cuprite and brochantite formation, along with a
neww peak consistent with the presence of paratacamite or other copper hydroxy
chlorides.. Copper chloride sulfate hydroxide hydrate was also tentatively
identified.. On the BTA-treated substrate, chloride products were again not
detected. .

3.3.3.3.3.3. Coating quality

Coatingss were examined for defects, including seeds (particulate
contaminants),, cratering, orange peel, and solvent popping/air entrapment
(commonlyy described as pinholing). Incralac had medium orange peel on the
bronzee substrates, and a fine, bumpy appearance on the copper roof substrates,
despitee efforts to avoid these defects. Wax coatings, which were applied very
thinlyy by brushing and buffing, generally looked good, and also achieved a matting
effectt on top of the shiny coatings. The wax-coated, blasted copper roof substrate
wass particularly saturated in appearance, achieving a matte, fairly brown
coloration.. The Nikolas acrylic/acrylic urethane coating showed light orange peel
onn the cast bronze substrates, although this was largely masked by the wax topcoat.
Muchh practice and experimentation were necessary to avoid orange peel in the
BASFF acrylic urethane coatings, and, as reported above, a 50% reduced coating
wass generally successful for this purpose. A small amount of air entrapment was
notedd on the patinated bronze substrates. Seeding, visible as individual dust
particless surrounded by small wells, was also a problem in the BASF coatings.
Thee Nikolas waterborne coating exhibited severe solvent popping and/or air
entrapment,, visible mostly on the polished bronze and unblasted copper roof

68 8
 PerformancePerformance of five coatings on four substrates

Tablee 1:
XRDD Identification of Mineral Phases on Uncoated Substrates
Afterr Accelerated Weathering

Sample Sample IdentifiedIdentified Phases + Residual Lines (A)ReferenceReference PDF

FileFile *
brown,, black, & cuprite e 5-0667 7
greenn corrosion tenorite e 5-0661,41-0254 4
on n atacamitee (major); paratacamite; botallackite 23-0948;25--
polished,, cast 1427,4-0913 3
bronze e brochantitee (after 21 days only) 43-1458,, 13-
0398,, 3-0282
nantokitcc (minor) 6-0344 4
tinn oxide** 25-1259,4-0550 0
rf=13.O0-13.19,rf=13.O0-13.19, 10.25-10.33,
Residual Residual 7.60-7.80, 6.90,
LinesLines (A) 6.54**,, 2.33-2.42:
and and
tentatively tentatively
posnjakite e 43-0670 0
associated associated
zincc chloride sulfate hydroxide hydrate (?) 41-1421 1
phases phasescopperr zinc sulfate hydrate + other 13-0309,35--
copper/zincc sulfate/chloride hydroxy 0538,35-0561 1
hydratess (?) 29-0578,, 23-
0962,, 1-0086
copperr nitrate hydroxide (?) 15-0014 4
sulfidess (?) 26-0575,, 29-
0578,23-0958 8
gray-black,, light cuprite e 5-0667 7
green,, & brown atacamite;; paratacamite; botallackite 23-0948;25--
corrosionn on 1427;4-0913 3
patinated,, cast copperr nitrate hydroxide** 15-0014 4
bronze e copperr hydroxy sulfate hydrate (?)** 20-0357 7
t/=13.15**.. 12.89.6.54,4.34-4.36:
Residual Residual
ammoniumm sulfate (?)
LinesLines (A) 41-0621 1
gray-black, , cuprite e 5-0667 7
lightt green, & brochantite e 43-1458,, 13-
brownn corrosion on 0398,3-0282 2
50-year-oldd copper nantokitee (?) 6-0344 4
roof f ammoniumm copper chloride hydrate (?) 25-0262 2
blackk & bright green copper r 4-0836 6
corrosionn on cuprite e 5-0667 7
walnutt shell- brochantite e 43-1458,13--
blastedd copper 0398,, 3-0282
roof f paratacamite;; other copper chloride 25-1427,19--
hydroxides s 0389,, 23-0953
copperr chloride sulfate hydroxide hydrate (?) 8-0135 5

*formerlyy JCPDS-International Centre for Diffraction Data

*** detected only on BTA-pretreated substrate

69 9
ChapterChapter 3

substrates,, despite following the manufacturer's recommendations for methods to

avoidd these defects. Cratering and seeding were also visible in this coating.
Inn a few cases, coatings showed some change in appearance after eight
monthss of storage in controlled museum conditions. The wax coatings showed
somee bloom, i.e., small opaque patches. The uncoated, artificial patinated
substratess noticeably darkened and became patchy. This latter change is not
unexpected,, according to common observation, especially for patinas formed with
sulfidee compounds.

3.3.4.3.3.4. Coating performance

Totall failure ratings after 120 days of accelerated simulated weathering
(groupp B) are presented in Figure 3, and ratings after about one year of natural
outdoorr weathering (group C) are shown in Figure 4. Note that three samples
placedd on the roof unfortunately disappeared during the course of the experiment
andd are therefore not included. See also plates II-V
Ass described in Chapter 2, the failure ratings for sample groups B and C
aree a product of three main components: a) coating condition, b) overall percent
corrosionn (area corrosion), and c) mean linear creepage corrosion at the
scribemark.. These are shown separately in the rating bars. As in Phase I, the
ratingss should be viewed as sensitive to conservation standards, in that any
perceptiblee visual change is significant. Thus a total failure rating of 10-20
indicatess perceptible failure in at least one category, while a failure rating of 30-60
(orr more) represents an unacceptable degree of failure. Good performance
maintenancee should remain around 10 or less. The failure ratings thus obtained for
eachh set of samples first of all illustrate that coating performance was highly
dependentt on the type of substrate. This was particularly true of two substrates:
thee artificially patinated bronze (II) and the walnut shell-blasted copper roof (IV).
Onn these substrates, the coatings generally failed earlier and more definitively in
acceleratedd weathering; outdoor weathering also supported these trends.
Onn the artificially patinated bronze (II), coating ratings reflect above
averagee change in the coating layer itself, particularly in terms of tiny blisters and
peelingg in the Incralac coating (#1), NK acrylic/ urethane/wax (#3), and
BTA/BASFF acrylic urethane/wax (#4). This may have been due to a degradative
effectt of iron and nitrate salts from the patina, which XRD results implied were
partlyy soluble. In addition, area and scribe corrosion ratings are relatively
advancedd after exposure to both weathering regimes. This appears to be partly
relatedd to poor adhesion (see discussion below). BTA pretreatment, as in coatings
#22 and 4, had no apparent inhibiting effect, despite the positive effect of this
treatmentt on uncoated substrates shown in XRD analysis. The overall ratings of
thee coatings on this substrate after both accelerated and outdoor weathering are
internallyy consistent in showing that failure was surprisingly rapid compared to
resultss on the other substrates. Results thus confirm that the artificial patina
surfacee itself lent instability to the coating systems.

70 0
 PerformancePerformance of Jive coatings on four substrales

01 1
cc
03 3
-C C
OO
oo
cc
II I

ai i
_c c
"*--
TO TO
D£ £
<D D
i--
_3 3
'ra a
LL L
"ra a +
J?J? A ^ > rrA
&&&&&& ^
jrjrjrjr ^ ~~ ^ V 1 ^" , #
jr^r^prp
oo X^>VVV* / / / / /
J
/•&&** •/*&&-*
,s>W> W y^ ^^ ^i ^ y*&&
I.. Polished, Cast II. Patinated, Cast II.. 50-yr.-old IV.. Walnut shell
Bronzee Bronze Cuu Roof blastedd Cu

areaa corrosion A s c r i b e creepage corrosion D c o a t i n g rating

Figuree 3 Total failure ratings for coatings on bronze and copper substrates after 120
daysdays of accelerated simulated weathering.

CD D
D) )
CC
ra ra
oo
oo
cc
II I
oo
Dl l
_c c
"ra a
KK
oo
uu
33

r£>efrr£>efr &
N^ N^ dr r
<r>' <r>'
N' '
SJV
^^
V iSjy y "S. "S. <o' <o'
VV
I.. Polished, II.. Patinated, I.. 50-yr-old IV.. Walnut shell-
Castt Bronze Castt Bronze Cuu Roof blastedd Cu Roof

II area corrosion II scribe creepage corrosion D c o a t i n g rating

Figuree 4 Total failure ratings for coatings on bronze and copper substrates after about
oneone Year of natural outdoor weathering.

71 1
ChapterChapter 3

Ratingss show that Incralac performed rather poorly on both the patinated
bronzee and walnut shell-blasted copper roof substrates. After accelerated
weatheringg on the blasted copper roof, the poor performance of both Incralac and
thee waterborne coating is associated with pronounced scribe corrosion, despite
goodd adhesion ratings (see Figure 8). The scribes appeared to have dark green
haloes,, suggesting reprecipitation of the brochantite from moisture/electrolyte
ingresss [19]. In contrast, failure in Incralac on the unblasted copper roof substrate
iss extremely low in accelerated weathering tests. These results imply that moisture
moree effectively penetrated into the blasted copper roof patina and probably
lingeredd beneath the coatings. As discussed above, SEM/EDS of the uncoated
walnutt shell-blasted surface showed aggravated porosity and non-homogeneity in
thee patina layer, as well as markedly more pitting and/or mottling after accelerated
weatheringg than the natural patina. This evidence indicates that the patina was less
stablee after the blasting procedure, and provides an explanation for the trend
towardd greater overall corrosion and failure in Incralac. and most coatings, on this
substrate.. The failure ratings after outdoor weathering, while not showing
advancedd failure, appear to support this trend.
Apartt from differences in the substrates, ratings show that the BTA/wax
coatingg failed outright on all substrates after accelerated testing. Failure was
observedd as overall darkening and severe pitting in the form of light green,
powderyy spots. The latter was especially noticeable on the walnut shell-blasted
copperr roof substrate. Results after about one year of outdoor weathering show
clearr failure in the waxed, patinated bronze sample, and suggest that samples with
waxx coatings on the copper roof substrates were beginning to overtake most other
coatingss in terms of failure. However at this one year mark, ratings show that the
waxx coating on unblasted copper roof still bordered on acceptable.
Thee waterborne coating was applied more thickly and given a wax topcoat
inn Phase II in the expectation that performance would be boosted. Ratings show
thatt the coating did perform better than in Phase 1 accelerated testing, but still only
inn a mediocre to poor fashion. The coating failed dramatically on the 50-year-old
brochantitee patina after one year of natural weathering, i.e., at an even faster rate
thann predicted by accelerated tests. The coating looked very good on the other
substratess at this point in the outdoor weathering, however. It is interesting to note
thatt the waterborne coating has the best ranking of the five coatings on the
patinatedd bronze substrate. No explanation can be offered for this, other than a
typee of consolidation effect of the patina with this particular coating. While the 3-
coatt application clearly benefited performance of this coating, results also point to
ann inherent weakness in the coating itself, most likely due to water sensitivity and
difficultyy in achieving a high quality film in application (see also Chapter 5).
Totall failure ratings for the solvent-borne acrylic and acrylic urethane
coatingss are mixed. Results for Incralac/wax and the Nikolas 3-part system on
polishedd bronze and natural brochantite patina generally confirm trends observed
inn Phase I. Both coatings performed extremely well on the unblasted, 50-year-old
copperr roof substrate in accelerated and outdoor testing. The Incralac coating on
polishedd bronze had clear signs of failure after accelerated weathering, however,
whilee the Nikolas acrylic/urethane coating was very yellowed but maintained good

72 2
 PerformancePerformance of five coatings on four substrates

protectionn on the polished bronze. An unwaxed sample of Incralac on bronze (not

includedd in the ratings) showed slightly more advanced corrosion after accelerated
weathering.. The BASF coating also has a slightly better rating than Incralac on the
polishedd bronze, despite mediocre ratings on the other substrates after accelerated
weathering.. This coating has an excellent rating on polished bronze after one year
off outdoor exposure. The somewhat inconsistent performance in this coating may
bee partly related to problems with film quality, as noted above. Ratings further
showw that while Incralac plus wax, and the Nikolas and BASF coatings rate poorly
afterr accelerated weathering on the walnut shell-blasted copper roof, Incralac is the
onlyy one of these three coatings to show signs of failure after one year of outdoor
exposuree on the same substrate. Thus, ratings show a tendency toward superior
performancee in the solvent-borne, acrylic urethane-containing coatings. However,
noo coating among those tested unequivocally improved on the general performance
characteristicss of Incralac. These coatings will be further discussed in Chapter 5.

3.3.5.3.3.5. Influence of dry film thickness and adhesion on coating

performance performance

3.3.5.1.3.3.5.1. Coating thickness

Averagee dry film thickness and thickness variation for Phase II
unweatheredd samples (group A) are shown in Figures 5 and 6. Coating thickness
readingss for the weathered sample group B showed no significant difference and
aree not reported. As noted in Chapter 2, the large standard deviation of coatings on
thee copper roof substrates is due mainly to the uneven patina. Standard deviation
inn coating thickness is also noticeably increased by brush coating.
Resultss show that although attempts were made to keep most coatings in
thee same thickness range, some differences in coating thickness could not be
avoided.. This was in part due to employment of common practices, as well as the
aestheticc requirements of the coatings, as previously discussed. For example, the
waxx coating is significantly thinner than the other coatings and also compared to
thee wax coatings tested in Phase I. The Nikolas acrylic/acrylic urethane/wax is
thickerr than the other coatings. However, this coating is in the recommended
thicknesss range and had a satisfactory appearance on the substrates. As previously
noted,, the recommended film thickness for most coatings is 1.5-2.0 mil (38.1-50.8
urn).. In the case of Incralac/wax and the BASF acrylic urethane/wax, much
thinnerr coatings, around 0.6-1.0 mil (15.2-25.4 urn), resulted from efforts to avoid
orangee peel and a glossy, "plastic" appearance.
Comparisonn of film thickness measurements in Figures 5 and 6 with the
totall failure ratings in Figures 3 and 4 shows an apparent correspondence between
poorr performance and the thinness of the BTA/wax coating. Measured at about
0.22 mils (5.1 |um) or less, this dry film thickness is typical of normal wax
applicationn in conservation practice and clearly represents a major drawback, since
itss performance characteristics do not appear to make up for the thin coating.

73 3
Chapter's Chapter's

2.0 0
polished bronze
i2 2 1.8 8 patinated bronze
1</> > 11.6 6 Dsd d

V) V) 1.4 4
O) )
cc 1.2 2
oo
1.0 0
!c c
EE 0.8 8
ill 0.6 6
QQ 0.4 4
di i
> > 0.22 -I
<<
0.0 0 11 •1 1
1-lncralacc + 2-BTAA + 3-NK K 4-BTAA + 5-NK K 6-no o
wax x wax x acrylicc + BASF F waterborne e coating g
acrylic c acrylic c acrylic c (patina a
urethanee + urethanee + urethanee + only) )
wax x wax x wax x

Sample e
Figuree 5 Coating + patina average dry film thickness and standard Deviation (sd) on
unweatheredunweathered bronze substrates.

natural patina
walnut shell-blasted
'i 'i •• sd

cc
oo

QQ

>>
<<
1-lncralacc +
wax x
II
2-BTAA +
wax x
3-NK K
acrylicc +
11
4-BTAA +
BASF F
5-NK K
waterborne e
tt
6-no o
coating g
acrylic c acrylic c acrylic c (patina a
urethanee + urethanee + urethanee + only) )
wax x wax x wax x

Sample e
Figuree 6 Coating + patina average dry film thickness and standard deviation (sd) on
unweatheredunweathered 50-year-old copper roof substrates.

74 4
 PerformancePerformance offivecoatings on four substrates

However,, results in Phase I showed that even a thick coating of wax is no

guaranteee of good performance (see Chapter 2 and [20]). Examination of the
waxed,, polished bronzes in Phase I under magnification showed etching beneath
thee wax after relatively short exposures in both accelerated and outdoor conditions.

Performancee was relatively better after outdoor weathering on the copper

rooff substrates in Phase II, where the coating penetrated the patina to a degree and
thuss had a greater effective thickness. Results after accelerated weathering
indicatedd that this advantage is time limited, however. These results are consistent
withh the notion that the wax coatings in Phase II were not only too thin to provide
adequatee protection, but are also highly permeable coatings.
Onn the other hand, the greater thickness of the Nikolas acrylic/acrylic
urethane/waxx coating most likely gave this system an advantage in performance
relativee to the other coatings. It may be noted, however, that Incralac applied on
polishedd bronze in an equivalent thickness in Phase I still did not perform as well
ass the Nikolas series of 3-part coatings in accelerated weathering. Thus, any boost
inn performance that is related to the thickness of the Nikolas 3-part coating seems
minimal.. The remaining three coatings are in a comparable thickness range, from
0.66 to 1.0 mil (15.2-25.4 urn), which represents a typical sprayed coating thickness
inn conservation practice. Coating thickness was therefore determined to be a major
factorr only in the poor performance of the BT A/wax coating.

3.3.5.2.3.3.5.2. Adhesion before and after weathering

Resultss of cross-cut and X-cut adhesion tests for Phase 11 samples, before
andd after accelerated weathering, are shown in Figures 7 and 8. Results for
unweatheredd and naturally weathered samples were obtained from one sample in
eachh category, while results for samples that underwent accelerated weathering
weree averaged over three samples. The averaged values reflect some variability in
thee results. Generally, differences within one unit of adhesion are of questionable
significance.. Results for Incralac, wax, and the Nikolas 3-part coatings are again
nott shown after outdoor weathering due to loss of the actual samples during the
experiment. .
Ratingss for the BT A/wax coating (#2) on all four substrates are deceptively
highh before and after weathering, as seen in Phase I. This is due to difficulty in
observingg cohesive failure within the very thin wax layer, as was observed in the
thickerr coatings in Phase I and presumably predominates in wax coatings. Also,
thee wax does not adhere well to the tape. Comparison of the total failure ratings to
adhesionn tests shows no correlation in this case, and further supports the contention
thatt adhesion in wax coatings cannot be measured by this method and is not a
significantt factor. It may be assumed that while wax has good mechanical
adhesionn on rough surfaces and can penetrate patinas, this is not sufficient to offset
itss inherent permeability.

75 5
 ChapterChapter 3

5
cc
Dpolished. .
no o
88 4 weathering g
xx
00 Dpolished, ,
II I
to o 33 accelerated d
weathering g
cc
JJ 2 HI Dpolished. .
outdoor r
or r weathering g
cc
o o 11 - Dpatinated. .
<D D
no o
-C C 0 0 weathering g
<< 1-- Incralac 2-- BTA + 3-- NK 4-- BTA + 5-- NK patinated.
++ wax* wax* * acrylicc + B A S F F w a t e r b o r n e e accelerated d
acrylic c acrylic c acrylic c weathering g
u r e t h a n ee + u r e t h a n e + urethane +
NB:: values of 0 reported Elpatinated. .
ass 0.1 wax x wax x outdoor r
'dataa not available for weathering g
polishedd bronze,outdoor Coatingss on Polished and
weathering,, # 1 , 2, & 3 Patinatedd Cast Bronze

Figuree 7 Cross-cut adhesion rating of coatings on cast bronze, before and after 120 days
acceleratedaccelerated and 1 year natural outdoor weathering.

5
cc Dunblasted, ,
jU U
no o
"Ö Ö weathering g
oo 4
XX
0) )
II I Dunblasted, ,
in n accelerated d
33
CD D weathering g
CC

To o 2 2 Dunblasted, ,
KK outdoor r
cc weathering g
oo 11
0) 0) Dblasted,, no
££ weathering g
<<
1-- Incralac 2-- BTA + 3-- NK 4-- BTA + 5-- NK Iblasted, ,
++ wax wax x acrylicc + BASF F waterborne e accelerated d
acrylic c acrylic c acrylic c weathering g
urethanee + urethanee + urethanee +
wax x wax x wax x Iblasted, ,
outdoor r
weathering g

C o a t i n g ss o n U n b l a s t e d a n d
W a l n u t - s h e l ll B l a s t e d C o p p e r R o o f

Figuree 8 X-Cut Adhesion ratings of coatings on 50-year-old copper roof, before and
afterafter 120 days accelerated weathering and 1 year natural outdoor weathering.

76 6
 PerformancePerformance of five coatings on four substrates

AA marked feature of the adhesion results is the general failure apparent on

thee artificially patinated bronze surface (Figure 7). Initial adhesion of Incralac and
thee BASF coating on the patinated bronze is good, but is poor for both Nikolas
coatings.. After accelerated weathering, only the Incralac retains even fair adhesion
onn this substrate, and Incralac completely loses its adhesion on this substrate after
outdoorr weathering. As previously discussed, the artificial patina was only loosely
adheredd to the bronze and appears to have been chemically unstable. Cohesive
failuree within the patina layer was apparent in areas that delaminated from the
bronze.. Here, poor coating adhesion corresponds well to coatings failure,
suggestingg that adhesion was a major factor as well as a symptom in coating failure
onn the artificially patinated bronze.
Incralac/wax,, the Nikolas 3-part coating, and the BASF coating show good
too excellent initial adhesion (with no weathering) on all other substrates. Better
initiall adhesion on polished bronze than seen in Phase I may be attributed to the
slightlyy rougher finish and the use of water polishing to produce cleaner surfaces.
However,, Incralac and the Nikolas coating exhibit large drops in adhesion on the
bronzee surface after accelerated weathering. This is consistent with Phase I results
forr Incralac, but not the Nikolas coating. The discrepancy in the latter case
appearss to relate to difficulty in evaluating results when failure occurs between the
coatingg layers. This also points to poorer coating application of the Nikolas
systemm in Phase II than in Phase I. The BASF coating shows excellent retention of
adhesion,, or increase in adhesion, throughout the weathering regimes on polished
bronze. .
Adhesionn testing results for these three coatings thus do not go hand in
handd with their relatively good performance ratings on polished bronze. Loss of
adhesionn in Phase II may, however, be reflected in the higher contribution of
corrosionn in the failure rating of Incralac (Figure 3), and may ultimately forecast
pendingg failure for both acrylic-based coatings. Good adhesion of the BASF
coatingg appears to confirm relatively good performance after outdoor weathering,
butt does not correlate well to the fair performance of this coating in accelerated
testing.. These coatings are further discussed in Chapter 5.
Initiall adhesion ratings for the coatings on natural and walnut-shell-blasted
copperr roof patinas (Figure 8} are similar to those on bronze, and generally high,
exceptt for the waterborne coating. As discussed in Phase I, this reflects good
mechanicall adhesion resulting from wetting and penetration into the patinas (see
alsoo Figure 6). Incralac/wax, the Nikolas acrylic/acrylic urethane/wax, and the
BTA/BASFF acrylic urethane/wax also retain good adhesion on the natural copper
rooff after weathering. Here, adhesion results appear to justify performance on the
unblastedd substrate reasonably well in terms of reinforcement of the partially
protective,, natural patina. Results on the blasted copper roof are less consistent,
however,, including some loss in adhesion in the Nikolas coating after accelerated
weathering,, and increase in adhesion in the BASF coating after weathering. Here
noo obvious correlation to performance results exists, except in the sense of being
unpredictable.. Other factors besides adhesion apparently dominated in poorer
performancee on the blasted copper roof substrate, as discussed above.

77 7
ChapterChapter 3

Thee almost non-existent adhesion of the waterborne coating on the

unblastedd copper roof samples was attributed to poorer wetting of the substrate by
thee coating. This alone appears to predict the dramatic failure observed on the
copperr roof sample during natural weathering, as well as in accelerated
weathering.. On the other hand, it is interesting that the waterborne coating
actuallyy gained adhesive strength to a small degree or even a reasonable level after
weatheringg on the bronzes and the blasted copper roof. This did not obviously
affectt performance, except possibly in the case of the patinated bronze, where the
coatingg rated highest after weathering, but still poorly. An increase in adhesion
wass also noted in the BASF acrylic urethane on blasted copper roof. These results
suggestt continued curing of the coatings during weathering, and show that a
correspondencee between adhesion and performance cannot be drawn alone.
Investigationn of delayed curing is discussed in Chapter 5.

3.4.3.4. Conclusions
Totall failure ratings compiled in Phase II of the study for five model
coatingg systems on four model bronze and copper substrates highlight significant
trendss in coating performance that are relevant to conservation practice. Results
firstfirst of all illustrate that coating performance can be highly dependent on the
substratee and its preparation. Two of the sample substrates, which were treated
eitherr by adding an artificial patina or partially reducing a natural patina with a
blastingg medium, showed overall instability that translated into poor coating
performancee across the board in accelerated weathering tests. In the case of a
foundry-appliedd patina, made with ferric nitrate and, most likely, ammonium
sulfidee on cast bronze, the substrate appeared particularly unstable in terms of
oxidation,, soluble salts, and poor adhesion characteristics. On this patina, all
coatingss in the study failed rapidly in both accelerated and outdoor testing.
Inn the second case, substrate preparation or treatment closely modeled a
currentt conservation practice and entailed walnut shell-blasting of a naturally
formedd brochantite patina on copper roofing. Substrate preparation also included
washingg and degreasing, consistent with practice. Here, coating performance was
alsoo noticeably worse than on polished, cast bronze or unblasted, natural copper
rooff patina in accelerated testing. However, outdoor testing results were less
decidedd in showing a trend toward decreased performance characteristics in the
coatings.. SEM and XRD analysis of the uncoated, blasted copper roof substrate
beforee and after weathering supported the conclusion that the substrate was more
pronee to corrosion after blasting. This appeared to be the result of increased
porosityy and fissuring in the patina structure itself, as well as possible oily residue
leftt from the blasting medium. These results support a picture of coating
performancee which was compromised by slightly increased susceptibility to
corrosionn on the blasted copper roof, and greatly increased instability on the
artificiallyy patinated bronze. Accelerated simulated weathering methodology
exaggeratedd and emphasized these results, and was interpreted to give a predictive
picturee of coating performance on these types of substrates.

78 8
 PerformancePerformance of five coatings on four substrates

Failuree ratings of the coatings also showed trends within coating types that
aree relatively independent of the substrates. The microcrystalline wax blend
coatingg applied thinly over a BTA-treated substrate, according to common
practice,, showed markedly inferior performance characteristics in general. In this
case,, coating performance clearly suffered from the thin application. Accordingly,
thee coating showed better performance after one year of outdoor weathering when
itt saturated an existing patina and effectively worked in tandem with the semi-
protectivee mineral crust. In accelerated weathering, this wax/patina advantage was
lost,, though. Building on results from a thicker wax coating used in Phase I, it can
bee concluded that the wax coating is a poor barrier to acid rain-type exposure at
differentt thickness, and does not provide significant protection in outdoor
exposuress except as a patina reinforcement for a limited period of time.
Thee other coating that exhibited general problems on all of the substrates
wass the waterborne acrylic urethane dispersion. In this case, adhesion testing on
mostt substrates revealed significant problems that were not solved by the thicker
applicationn and wax topcoat (compared to Phase I). This was most clearly
demonstratedd on the natural brochantite patina, on which the coating showed very
poorr adhesion and also dramatic failure in accelerated testing after only one year in
outdoorr exposure. In this case, poor adhesion may be due to surface tension
effects,, and poor wetting and film formation characteristics, within the partially
saturatedd mineral phase. The appearance of the samples after weathering also
indicatedd that moisture effectively penetrated beneath the coating. It is interesting
thatt these problems were not as marked on the blasted copper roof substrate
subjectedd to outdoor weathering. However, accelerated testing results appeared to
confirmm that the coating has inherent properties that lead to ineffective protection
fromm acid-rain type exposure on all of the model substrates in this study.
Whilee analysis of average dry film thickness and adhesion greatly aids in
explainingg coating behavior (and in designing improved behavior), it is clear that
thesee factors do not predict coating performance by themselves. This is simply
becausee so many factors come into play in coating performance. In Phase II of this
study,, the combination of thickness measurement, adhesion testing, and combined
failuree rating of model coating systems, with and without weathering exposures,
allowedd assessment of five coating systems for use in the outdoor bronze
conservationn field. This assessment shows close relative rankings, within a margin
off variability, for three coatings on most substrates, where: Nikolas acrylic/acrylic
urethanee + wax > Incratac + wax « BTA + BASF acrylic urethane + wax. It
remainss to be seen whether other methods of application would allow the
inherentlyy superior durability properties of the acrylic urethane-containing coatings
too be translated into much better system performance than was witnessed with the
Incralac/waxx coating. It is clear, however, that these three coatings belong to a
mediumm to high performance category, while the wax and waterborne coatings
belongg to a separate, comparatively low performance/high maintenance category.
Thiss study also suggests that with further coating development and use in the field,
thee two solvent-borne acrylic urethanes are viable low maintenance coatings for
usee where Incralac or wax do not meet performance criteria.

79 9
ChapterChapter 3

References References

11 Includes Phases II and III as reported to NCPTT: L. Brostoff, T, Shedlosky

andd R. de la Rie, "Research into Protective Coating Systems for Outdoor
Bronzee Sculpture and Ornamentation. Phase 11" (PTTPublications No. 1999-
23,, NCPTT, Natchitoches, Louisiana, 1999); L. Brostoff, T. Shedlosky and R.
dee la Rie, "Research into Protective Coating Systems for Outdoor Bronze
Sculpturee and Ornamentation. Phase III" (PTTPublications No. 1999-23,
NCPTT,, Natchitoches, Louisiana, 2000).
22 Carried out by Cameron Wilson, private conservator (Wilson Conservation).
33 R White, and J. Pile, "Analyses of paint media," National Gallery Technical
BulletinBulletin 17 (1996), 91-103.
44 Kenneth R. Sutherland, "Solvent extractable components of oil paint films,"
Ph.D.. Thesis, University of Amsterdam, Netherlands (April 2001), 139-158.
55 Bruno M. Perfetti, Metal Surface Characteristics Affecting Organic Coatings
(Federationn Series of Coatings Technology, Blue Bell, PA, August 1994), 63.
66 Coatings applied by Cameron Wilson, private conservator (Wilson
Conservation);; second application of Nikolas 11650 coating, and wax coatings
appliedd by Andrew Baxter, private conservator (Bronze, et al.).
77 Percy E. Pierce and Clifford K. Schoff, Coating Film Defects (Federation
Seriess of Coatings Technology, Philadelphia, PA, January 1988), 11-20.
88 "Standard Test Method for Nondestructive Measurement of Dry Film
Thicknesss of Nonconductive Coatings Applied to a Nonferrous Metal Base,"
inn 1993 Annual Book of ASTM Standards, Vol. 6.01, ed. Paula C. Fazio, et al.
(Americann Society for Testing and Materials, Philadelphia, PA, 1993), 222-
223. .
99 "Standard Test Methods for Measuring Adhesion by Tape Test," in 1993
AnnualAnnual Book of ASTM Standards, Vol. 6.01, ed. Paula C. Fazio, et al.
(Americann Society for Testing and Materials, Philadelphia, PA, 1993), 447-
450. .
100 Society of American Engineers, "Accelerated Exposure of Automotive
Exteriorr Materials Using a Controlled Irradiance Water Cooled Xenon Arc
Apparatus,"" SAEJ1960 (June 1989).
111 "Standard Test Method for Humid-Dry Cycling for Coatings on Wood and
Woodd Products," 1990 Annual Book of ASTM Standards, Vol. 6.02 (American
Societyy for Testing and Materials, 1990), 178-179.
122 "Standard Test Method for Freeze-Thaw Resistance of Water-Borne
Coatings,"" 1990 Annual Book of ASTM Standards, Vol. 6,02 (American
Societyy for Testing and Materials, 1990), 82-83.
133 K. Nassau, A. E. Miller and T. E. Graedel, "The reaction of simulated rain
withh copper, copper patina, and some copper compounds," in Special Issue:

80 0
 PerformancePerformance of five coatings on four substrates

CopperCopper Patina Formation Corrosion, Corrosion Science 27, 7 (1987), 703-

720. .
144 T. E. Graedel, "Copper patinas formed in the atmosphere-II. A qualitative
assessmentt of mechanisms," in Special Issue: Copper Patina Formation
Corrosion,Corrosion, Corrosion Science 27, 7 (1987), 741-770.
155 S. Suga and S. Suga, "Development of Simulated Acid Rain Test Using CCT
Method,"" in Accelerated and Outdoor Durability Testing of Organic
Materials,Materials, ASTM STP 1202, eds. Warren D. Ketola and Douglas Gorssman
(Americann Society for Testing and Materials, Philadelphia, PA, 1994), 247-
262. .
166 "Standard Practice for Atmospheric Environmental Exposure Testing of
Nonmetallicc Materials," in 1993 Annual Book of ASTM Standards, Vol. 6.01,
ed.. Paula C. Fazio, et al. (American Society for Testing and Materials,
Philadelphia,, PA, 1993), 1008-1011.
177 "Standard Test Method for Evaluating Degree of Rusting on Painted Steel
Surfaces,"" in 1999 Annual Book of ASTM Standards, Vol. 6.02, ed. Robert F.
Allen,, et al. {American Society for Testing and Materials, Philadelphia, PA,
1999),, 13-15.
188 "Standard Test Method for Evaluation of Painted or Coated Specimens
Subjectedd to Corrosive Environments," in 1993 Annual Book of ASTM
Standards,Standards, Vol. 6.01, ed. Paula C. Fazio, et al. (American Society for Testing
andd Materials, Philadelphia, PA, 1993), 271-273.
199 Concluded following conversations with Dr. Helena Strandberg, November
1999. .
200 Lynn Brostoff and E. René de la Rie, "Research into Protective Coating
Systemss for Outdoor Bronze Sculpture and Ornamentation. Phase I"
(PTTPublicationss No. 1999-23, NCPTT, Natchitoches, Louisiana, 1997).

81 1
ChapterChapter 3

afterr accelerated weatherin

Platee II Phase II, polished bronze Set I A turnsfathered, left) and IB, after accelerated
weathering,weathering, with coatings #1-5, and uncoated controls.

Platee III Phase II. patinated. cast bronze. Set IIA (unweathered. left) and IIB, after
acceleratedaccelerated weathering, with coatings #1-5, and uncoated controls.

HI HI
 PerformancePerformance of five coatings on four substrates

Platee IV Phase II, 50-year-old copper roof Set III A (unweathered, left) and IIIB. after
acceleratedaccelerated weathering, with coatings #1-5, and uncoated controls.

Platee V Phase II. walnut shell-blasted copper roof Set IVA (unweathered. left) and IVB,
afterafter accelerated weathering, with coatings #1-5, and uncoated controls.

S3 3
UU
 44
Electrochemicall Impedance Spectroscopy (EIS)
off select coatings on bronze

Abstract Abstract
Electrochemicall impedance spectroscopy (EIS) analysis was undertaken in
orderr to examine the coated, polished bronze samples prepared and studied in
Phasess I and II by a parallel, quantitative method. Chapters 2 and 3 describe
failuree ratings of coatings on bronze as determined by visual means. This chapter
describess EIS of select coatings on bronze from Phase I, as well as a complete set
off unweathered coatings on polished bronze from Phase II, and provides rankings
off these coatings by purely electrochemical means. The unweathered samples
weree followed by EIS during a simulated accelerated weathering program. Phase
III samples that had already undergone separate accelerated weathering were also
examinedd by EIS for purposes of comparison. Results support the findings of
coatingg performance on bronze as reported in Chapters 2 and 3.

4.1.4.1. Introduction
Electrochemicall test methods have recently moved center stage as
importantt tools for the quantitative characterization of coatings on metal substrates
[1].. In particular, electrochemical impedance spectroscopy (EIS) has become
knownn as a valuable method for the rapid ranking of coatings and prediction of
futuree performance [2,3,4,5,6,7]. In order to compare EIS analysis with visually-
basedd performance assessment as described in Chapters 2 and 3, EIS was initially
performedd on a small group of coated samples from Phase I [8], Following this
preliminaryy analysis, the complete set of unweathered samples on polished bronze
fromfrom Phase II were monitored by EIS during an accelerated weathering regime
consistingg of cyclic salt fog/UV exposure (see Experimental Methods, below).
Thesee results were compared to EIS values obtained from a second set of Phase II
coatedd bronze samples that had already been weathered under the National Gallery
off Art accelerated weathering protocol, as described in Chapter 3.

85 5
ChapterChapter 4

EISS is based upon the creation of an electrochemical cell; this consists of a

coatedd metal sample, which functions as the working electrode, in contact with an
electrolyte,, into which are immersed a counter electrode and reference electrode.
Thee setup is shown schematically in Figure 1 [9].

COUNTERR ELECTRODE REFERENCE

(PLATINUM)) ELECTRODE
\\ /

POTENTIOSTAT T

T"-"f f
1 1!! ;;
0, ,
3 3 --
•• I I .-m::-m\;

mm VV - - ;-' ; : ' .'

ff /
:-:-:-:-:-:-:--------

ff f lAlA WORKING ELECTRODE

X X=j j i f ))

Figuree 1 Schematic representation of electrochemical impedance experimental setup.

Ann alternating voltage (V, volts) is applied to the working electrode, and the
currentt response (I. amps) is measured over a range of frequencies (co) and time.
Thee measured time lag. or phase shift (0). between the excitation (V(t)) and
responsee signal (l(t)) is held to arise from characteristic types of frequency
absorptionn or response to the applied voltage and the electrolyte [10]. This
principlee is illustrated in Figure 2 [9].

86 6
 EISEIS of select coalings

Appliedd Potential ( E )
Phasee Shift
EE (t) = Eo sin (wt)

II Sinusoidal Current Response ( I )

I ( t )) = I o s i n ( w t + 0 )

Figuree 2 Theoretical current (I) vs. time sinusoidal waves for excitation and response
signalssignals in EIS.

Simplistically,, Ohm's Law, V/I=R, describes the EIS cell, so that the
workingg electrode may be defined in terms of its property of resistance (R) in the
chosenn electrolyte and setup. For a coated metal sample, the experimental
electrochemicall cell is much more complicated, however, and may be better
understoodd as analogous to an electronic circuit, with numerous resistive and
capacitivee elements. Therefore we define the ratio E/I (where E=electrode
potentiall (V)) as the impedance, Z, of the sample, which is a measure of the total
oppositionn to current flow in the alternating current circuit, such that
Z=E/I=[Eosin(tot)]/[Iosin(cot-0)].. Impedance is a complex quantity and is made up
off two components, real and imaginary, so that impedance modulus, I Z |, is
definedd as shown in equation (1).

z|| =v (zyinn phase ' ar outt (I) (I)

arof phase
Forr purposes of interpretation, log impedance modulus, | Z I (ohms (ClJ), is
oftenn plotted as a function of log frequency, (o (Hz), in a so-called Bode plot. In
thiss representation, the low frequency impedance modulus values, e.g., at 0.1 Hz.
aree held to be a measure of the coating's corrosion protection of the metal; these
valuess may be used to rank coating performance, where higher impedance values
correspondd to better protection [11]. In the system used for Phase I sample testing,
impedancee modulus was reported as a function of exposed surface area (Q cm").

87 7
ChapterChapter 4

Quantitativee values at 0.1 Hz were used for coating ranking as follows: good to
excellentt = values of IO9 to greater than 1012 D. cm2 maintained after two weeks
immersion;; fair to good = values of 106 to greater than 109 Q cm2 maintained after
22 weeks immersion; poor = values of less than 106 Q cm2 that do not maintain even
thiss level of protection in immersion.
Inn the testing system used with Phase II samples, impedance modulus was
reportedd strictly in ohms; coatings with impedance modulus values of |Z| o.i Hz >
1077 Q were considered to be performing well under the cyclic weathering protocol.
Absolutee failure was considered to be the point at which a coated sample had an
impedancee modulus value of |Z| o.i Hz = 103 Q, which was close to the value
obtainedd for the uncoated bronze. Although visual changes during testing were not
takenn into account in the rankings, failure was obvious either visually or from the
EISS data in some samples at values greater than that of the uncoated bronze, e.g., at
|Z|| o.i Hz < 106 Q. This is discussed in greater detail below.
AA complicating factor in the interpretation of performance based solely on
onee impedance modulus value is that impedance may actually be increased during
weatheringg by the formation of new corrosion or a passivating layer beneath the
coating.. Obviously the coating has failed if this occurs, and data can be easily
misinterpreted.. Therefore, EIS measurements of a system undergoing weathering
aree particularly valuable for tracking such events. Interpretation of Bode plots may
yieldd other information about dynamic changes taking place in the coating/metal
systemm during progressive weathering as well. Whereas low frequency impedance
iss thought to arise mainly from resistance to wetting at the metal surface beneath
thee coating, decreases or flattening in the slope of the curve are thought to arise
fromm increasing non-homogeneity in the coating, such as the formation of
micc roc racking. Decreases in the impedance modulus at the high frequency end
mayy indicate increasing porosity in the coating [12].

4.2.4.2. Experimental methods

Thee experimental EIS setup consisted of a glass cylinder clamped with an
o-ringg to a bronze sample, which acted as the working electrode. The glass
cylinderr was filled with dilute Harrison's electrolyte (0.35 wt.% (NFU^SCM and
0.055 wt.% NaCl in H2O), and a saturated calomel reference electrode and a
platinumm counter electrode were immersed in the solution. The area of exposure
onn Phase I samples was 12.56 cm , and on Phase II samples was 7.07 cm . For
Phasee I samples, a 10 mV RMS sinusoidal potential (E) was applied to the cell
withh respect to the open circuit potential, and the current (I) response was scanned
overr a frequency (00) range of 0.1 to 5,000 Hz, over a period of time (t). For Phase
III samples, the amplitude of the applied wave was 5 mV, and the frequency range
wass 0.1 to 10,000 Hz. Instrumentation included a Gamry Instruments PC-3
Potentiostatt controlled by Gamry CMS 100 software.
Priorr to initial EIS measurements all coatings were free of any corrosion.
Thee most evenly coated areas were chosen for measurement. For Phase I samples,

88 8
 EISEIS of select coatings

repeatedd EIS measurements were taken while the samples were in continual
contactt with the Harrison's electrolyte. For Phase II, set A samples, EIS
measurementss were taken weekly between cycles of simulated, accelerated
weathering.. The weathering protocol at the North Dakota State University was
fashionedd according to ASTM D 5894-96 "Standard Practice for Cyclic Salt
Fog/UVV Exposure of Painted Metal (Alternating Exposures in a Fog/Dry Cabinet
andd a UV/Condensation Cabinet)." In the first week, the samples were placed in a
QUV®® cabinet, where they were exposed to four hours of exposure to 340 nm UV-
AA at 60 °C, followed by 4 hours of condensation at 50 °C. In the second week, the
sampless were placed into a Prohesion® Chamber that cycled between one hour of
saltt fog at 25 °C and one hour of no fog at 35 °C. The salt fog used for weathering
wass the dilute Harrison's electrolyte.

43.43. Results and discussion

43.1.43.1. Phase I samples

Preliminaryy EIS studies were conducted on five polished, cast bronze
sampless from Phase I (see Chapter 2). The five coatings tested were: StanChem
waterbornee acrylic urethane (#26)), Incralac (#1), BTA pretreatment + Incralac
(#2),, Incralac + wax (#3), and BTA pretreatment + wax (#29) [13]. Initial readings
indicatedd that StanChem waterborne acrylic urethane was the best coating, with an
impedancee modulus at 0.1 Hz equal to about 8* 107 Q cm2, followed by, in order of
performance,, Incralac (1*107 Q cm2), Incralac + wax (4*106 Q cm2), BTA +
Incralacc (4* 106 H cm2), and BTA + wax (7* 105 Q cm2).
Afterr continual contact with the electrolyte for two days, a thin, foggy film
formedd over the samples pretreated with BTA, and 0.1 Hz impedance values for
BTAA pretreatment + Incralac (#2) and BTA pretreatment + wax (#29) increased, by
aboutt two or one order(s) of magnitude, respectively. This foggy film did not
occurr in outdoor or accelerated weathering tests, however, and appears to relate to
thee immersion-like conditions used in testing. As previously mentioned, a rise in
impedancee modulus during testing typically indicates that a corrosion patina layer
hass formed at the metal/coating interface. This added layer would offer increased
resistancee in the cell. The opaque film may thus be interpreted as an artifact of the
testing,, i.e., does not reflect natural outdoor weathering. Therefore, these results
weree discarded. At this point, the StanChem waterborne acrylic urethane coating
alsoo appeared to have fogged, possibly affecting its continued high ranking. No
visuall change was observed for the samples coated with Incralac (#1) or Incralac +
waxx (#3).
Afterr 15 days of contact with the electrolyte, EIS rankings from |z| 0 .i m
weree as follows: StanChem waterborne acrylic urethane (9*105 Q. cm2), Incralac +
waxx (9*105 D. cm ), and Incralac (3*105 Q cm2). Some changes in the appearance
off coatings occurred. StanChem waterborne acrylic urethane (#26) was fogged
andd heavily pitted, seriously altering the appearance of the coating. Incralac + wax

89 9
ChapterChapter 4

hadd two small pits. The sample coated with Incralac alone had become cloudy.
Thee impedance moduli of the Incralac coatings had both fallen in the poor
performancee range. The presence of BTA in the coating formulation appeared to
showw little or no effect on the coating system by EIS. The ranking order also
continuedd to show a slight benefit from the addition of a wax topcoat. These
resultss support similar conclusions in Chapter 2.
Afterr 22 days of continual contact with the electrolyte, all coatings
degradedd slightly, but still held the same ranking from best to worst coatings.
Afterr 29 days, however, the StanChem waterborne acrylic urethane was omitted
fromm EIS measurements due to its performance value far below what is considered
too be a poor coating. Also, its appearance was full of pits, with highly corroded
areas.. These results confirm the extremely poor performance of the StanChem
waterbornee acrylic urethane (#26) observed in Phase I (see Chapter 2).
Measurementss of Izlo.i Hz for the Incralac coatings were as follows: Incralac +
wax,, 3.1* 105 H cm2; and Incralac, 1.2* 105 O cm2.

4.3.2.4.3.2. Phase II samples, set A

Furtherr EIS analysis was coupled with a more realistic accelerated
weatheringg protocol on Phase II coated, polished bronze samples (set A, controls)
untill failure or up to 238 days, whichever came first (see Experimental Methods).
Afterr an initial set of readings, periodic measurements were taken on the same
samplee spot between cycles of salt fog in a Prohesion® chamber and QUV®
exposure,, i.e., during progressive, artificial weathering, as described in the
experimentall section.
Bodee plots for Phase II sample set A are shown in Figures 3-5.
Measurementss are shown up to 238 days of weathering; the number of days shown
inn thee legends with asterisks indicates this reading followed salt fog exposure. The
rankingg as determined by these results, in order of worst to best, was as follows:
baree bronze (#6) < BTA + wax (#2) « waterborne acrylic urethane + wax (#5) <
Incralacc + wax (#1) < BTA + BASF acrylic urethane + wax (#4) < Nikolas
acrylic/acrylicc urethane/wax (#3). This ranking is identical to the overall failure
ratingg rankings on both polished bronze and the blasted copper roof substrates after
acceleratedd weathering at the National Gallery of Art. Even the relative separation
off these rankings after accelerated weathering seems comparable.
Thee Bode plot of the BTA + wax coating on polished bronze in Figure 3b
showss that the initial impedance value characterizes this coating as a poor
performerr from the outset. After one day, the modulus actually rose; by 8 days of
simulatedd weathering, the modulus fell to the value of the bare bronze (Figure 3a).
Althoughh there is not a clear explanation for the almost immediate rise in modulus,
itt should be noted that the decrease in impedance occurred not only at low
frequencies,, but also at high frequencies, suggesting significant increase in the
porosityy of the film. Results thus support performance evaluations in Phases I and
II,, and underscore the fact that this wax coating is in a different performance
categoryy from the other coatings tested in this study.

90 0
 EISEIS of select coatings

1E+010 0
(( • Initial l
AA dayy 1
1E+009 9 day8 8
dayy 14
+ dayy 18
1E+0088 ' VV
1E+0077 '

—— 1E+006 '
10 0
.3 3
33 1E+005 r
"O O
oo
EE 1E+004
NN
1E+003 3
?»•» »
1E+002 2 ^ S S H J Ü K W ** mi;;;;;;;;;;;ajus;
+++-+ +
1E+001 1
0.1 1 100 100 1000 0 10000 0
F r e q u e n c yy (Hz)

Figuree 3a EIS of bare bronze (Phase II. 6A).

1E+010 0

Initial l
AA
1E+0099
»» dayl l
TT dayy 8
1E+0088 ' ++ dayy 14* Failure
JJ
|lE+007 7
si si
-^IE+0066 r

31E+0055 '
OO
E1E+004 4

1E+003 3 t'f t'f " » » T TT

" ' " T T ? ,,
++
'++ + + + +++ l+m * .. 5 w i i M M » » * t * É É t t t t É É
1E+002 2

1E+001 1
0.1 1 100 100 1000 0 10000 0

F r e q u e n c yy ( H z )

Figuree 3b EISofBTA pretreatment + Wax on bronze (Phase II, 2A).

91 1
ChapterChapter 4

1E+010 0
// • Initiall ^

1E+009 9
»» dayy 1
TT dayy 8
dayy 14*
1E+008 8 KiUu KiUu
«i i
dayy 21
(/i i
i l l»X X •• dayy 28-
dayy 35
E11 E+007 =5», ,
.c c •22a a •• dayy 42-

—1E+006 6
•fci i TT dayy 49
(/> > i i2 + dayy 56* Failure
2SS VV
_3 3 ' O
o 0o0
31E+005 5 <<
aa ^ J l X I ' t T T T T T » » . .. °P0
oo
E 1EE 04:
NN * S 1.. } „«« ""••••» •

1E+002 2

1E+001 1
0.1 1 100 100 1000 0 10000 0
F r e q u e n c yy ( H z )

Figuree 4a EIS ofWaterborne Acrylic Urethane + Wax on bronze (Phase II, 5A).

1E+010 0
Initial l
dayl l
11 E+009
day8 8
dayy 14*
1E+008 8 dayy 28*
dayy 42*
££ 1 E+007
•• Ba,,
• , dayy 70'
-C C »*••• day84' '
day98" "
•2-11 E+006
l/i i •-.. . dayy 112*
_g g
33 IE-
'•iiiiitiiis::;;:::: -•••!!!"*. dayy 126*
TS S 0055 dayy 140 Failure^
OO - • « « « « : : hij..
EE 1E+004
NN
++
-+++ ( + + _ + + , + + +
«•/I I
!!
1E+003 3

11 E+002

1E+001 1
0.1 1 100 100 1000 0 10000 0

F r e q u e n c yy ( H z ]

Figuree 4b ELS of Incralac + Wax on bronze (Phase II. I A).

92 2
 EISEIS of select coatings

EISS results for the waterborne acrylic urethane coating (Figure 4a) show
thatt the IZ I 0.1 Hz value dropped precipitously at day 14 of weathering, and by day
499 the impedance modulus could no longer recover its resistance after a drying
period.. Complete failure was imminent at that point. This rather sudden decline
appearss to predict the dramatic failure of the waterborne coating that was observed
onn the naturally weathered polished bronze and 50-year-old copper roof in Phases I
andd II of the National Gallery of Art study. EIS of this coating also shows a
decreasee in impedance with weathering at the high frequency end of the graph.
Thiss feature supports findings in Phases I and II that the coating tended to lose
adhesion,, and may also indicate increased porosity and/or cracking.
Lesss dramatically, the | Z | o i Hz value for Incralac + wax (Figure 4b) also
exhibitedd a rapid drop from over 108 to under 106 O between days 28 and 42 of
weathering.. By day 70, the low frequency impedance modulus indicated rather
poorr performance, which continued over the next 70 days until complete failure.
Thiss mimics the rapid advance of failure noted between initial accelerated
weatheringg of Incralac coatings in Phase I at the National Gallery of Art after 47.5
days,, and the extended weathering after 7.5 months. Normal applications of
Incralacc coatings are expected to last 3-5 years in outdoor environments. On this
basis,, as well as results for the wax coating, it is tempting to correlate 15 days of
thee simulated weathering protocol used with EIS to about 1 year of natural outdoor
weathering. .
EISS analysis of the urethane coatings showed a different kind of picture
fromm that of the other coatings (Figure 5a,b). Measurements indicate that the
systemss maintained initial |z|o.i Hz values for close to 98 days of accelerated
weathering.. During this period, the Nikolas acrylic/urethane exhibited almost no
changee in the curve of the Bode plot. The BASF acrylic urethane did begin to fan
downwardd somewhat, although it also maintained a good performance level.
Graduall decease in low frequency impedance began to occur in the Nikolas coating
aboutt day 98 of weathering and both coatings were showing some signs of pitting
aroundd 238 days. Results thus support the conclusion in Phases I and II that the
Nikolass acrylic/urethane coating on polished bronze belongs to a "high
performance"" category compared to the other coatings. Similarly, EIS supports
conclusionss in Phase II that the BASF acrylic urethane coating appears to straddle
highh and medium performance categories.
Iff we examine the same data for sample set A in a different form, i.e.,
plottedd as impedance modulus jZ|o.i Hz VS. time of weathering (Figure 6), we can
discernn some of the weathering trends more clearly. This graph illustrates first of
alll that once the impedance of the coatings started to fall, the decline was not
steadyy with time, but oscillated up and down. These oscillations correspond to
whetherr the samples were measured immediately following UV exposure or salt
fogg cycling. After Prohesion* chamber exposure, i.e., every 14 days, if the coating
hadd imbibed a lot of moisture, the metal beneath the pores would be wetted,
conductivityy would be increased, and this would cause the impedance to drop.
Afterr QUV® exposure, samples would dry out, inhibiting conductivity and allowing
impedancee to rise again.

93 3
ChapterChapter 4

1E+010 0
Initial l
1E+009 9 dayl l
day42' '
day49 9
1E+008 8 day70* *
["^vVV.-.-SVv!*!! !
day84' '
j== 1E+007 >••••••• • &!», , day91 1
-C C
•2-1E+006 6
!l», , day98* *
day104 4
«/) ) dayy 112*
_a a
33 1E+005 ++++ I-H-+++++H H day119 9
III.. dayy 126*
oo day133 3
day140" "
EE 1E+004
dayy 154'
dayy 196*
1E+003 3 day238'. .

1E+002 2
1E+001 1
0.1 1 100 100 1000 10000 0
Frequencyy (Hz)

Figuree 5a EIS of BTA + BASF Acrylic Urethane + wax on bronze (Phase II, 4A).

1E+010 0 Initiall >

dayy 1
1E+0099 | dayy 42"
ii dayy 49
"aso.
1E+0088 ' dayy 70-
,, dayy 84-
(/> > .77 -"••*„ s
! dayy 91
E1E+0077 i**4
t dayy 98-
§§ -«••t:::::::r "t!!:!f:'iis iB dayy 104
- 1 E + 0 0 66 { • • • • • • * l M t » « « « * • • • • • ***é*2gffB dayy 112-

u l E + 0 0 55 '
suunilUl! ! dayy 119
dayy 126-
"O O Nil l dayy 133
OO
E11 E+004 nn dayy 140*
dayy 154-
NN
dayy 196-
1 E + 0 0 33 \ dayy 238',

1E+002 2
1E+001 1
0.1 1 100 100 1000 0 10000 0
Frequencyy (Hz)

Figuree 5b EIS of Nikolas Acrylic + Acrylic Urethane + Wax on bronze (Phase II, 3A).

94 4
 EISEIS of select coatings

8C3 3

f33 3

013 3

961 1
!! s
<DD ™
381. . CC 5
rara +
-CC O)

- ** *-
891-- oo i>

VQl VQl
I! !
J?? fr
oo o
:== ra

Off I rara S

931-- «J J r o o i o o CD D NN 3

<JW W
^^
(Ü Ü \ \
11
++
E E1 11 11 1
II 22 Si

•33 Ö

"aa Li
86 6 22 o
XX

f88 55
+ai +
i ^JJ ^c:
cc
Oi i
(1) )
33

99 9 oo

XX
ra raU--
3f f 55 ss
+o o+< <+ +
ra raCD D *-< <
Q. .
83 3 i_ _ <<
oo h-- 1 --
CD Dm m
ftt

\\
C J T - O O C O I —— t o m a t o e s * -
T ¥¥ + ? ? ? ? ? ? ? ? ?
U J U J U J U l L U U l l l l U l l l l l l l L i l l UU
o o o o o o o o o o o oo

(Ü) Z HK)| Z | |

95 5
ChapterChapter 4

Afterr 14 days of weathering, i.e., after completion of thee first salt fog cycle,
alll coatings except the BASF acrylic urethane showed a fall in impedance. Again,
explanationn for the one exception at this early stage is not readily apparent. The
goodd physical condition of the sample and subsequent good performance suggest
thatt the BASF coating underwent continued curing during these initial weathering
cycles.. This was thought to occur on the blasted copper roof sample in Phase II
experiments,, where a definite increase in adhesion was measured after accelerated
weathering. .
Duringg subsequent weathering, both the Nikolas acrylic/urethane coating
andd the BASF acrylic urethane coating held fairly steady for a period of time,
whereass the Incralac and the waterbome acrylic urethane coatings exhibited large
oscillationss in impedance almost immediately upon weathering cycles. This
illustratess well that the latter two coatings, and especially the waterbome coating,
weree initially susceptible to moisture/electrolyte ingress. While the waterbome
coatingg failed fairly rapidly, by day 56, the Incralac coating showed small
oscillationss in modulus, then larger oscillations starting at about 42 days, then
stabilizationn for a period starting at about 91 days. This suggests the formation of
microcrackss and/or loss of adhesion relatively early on, followed by medium
performancee until a coherent passivating corrosion layer formed underneath the
coating.. This hypothetical, new patina layer appeared to have subsequently broken
downn before ultimate failure at 140 days. However, as this graph shows, 91 days
correspondss more truly to real failure in the Incralac coating.
Thee behavior of the remaining two top performing coatings, the Nikolas
acrylic/urethanee and BASF acrylic urethane, is thus clearly distinguishable from
thee others in this analysis. The initially long, steady impedance behavior of these
coatings,, 77 days in the case of the Nikolas coating and 61 days for the BASF
coating,, marks them as high performance coatings. The Nikolas coating also
showss itself to be the slight leader in this race. Thus, the information as plotted in
Figuree 6 lends more support to the findings in Chapter 3 regarding relative
performance. .

4.3.3.4.3.3. Phase II samples, set B (after accelerated weathering)

Furtherr testing was conducted on a set of coated, polished bronze samples
thatt had already undergone accelerated weathering at the National Gallery of Art
(Phasee II, set B). Results are shown in the Bode plot in Figure 7. EIS rankings
are,, from worst to best: BTA pretreatment + wax (#2)< uncoated bronze «
Nikolass waterbome acrylic urethane + wax (#5) < BTA pretreatment + BASF
acrylicc urethane + wax (#4) < Nikolas acrylic + Nikolas acrylic urethane + wax
(#3)) < Incralac + wax (#1). The results are consistent with both the EIS results for
Phasee II set A samples, and with the overall failure ratings shown in Chapter 3.
Thee difference seen here is that the EIS results do not distinguish the three highest
rankedd coatings from each other, as they are in the above experiment or in the
failuree ratings (see Chapter 3). A disturbing result shown here is that the waxed
sampless actually fell behind the uncoated sample, which was at this point covered
byy a coherent layer of corrosion, i.e., patina.

96 6
 EISEIS of select coatings

1E+010 0

1E+009 9

1E+008 8

1"lE+007 7
.c c -- - . . "-DCtJDcn n
-2-1E+006 6!! ! ......
(/> >
33 1E+005
oo
OO
'-v*:-. .
***?»? ?
,§§ 1E+004
,,
,vv
1E+003 3 0< <° O o 0 oo vvv, v, v v
00 0 0 0 0 0 0
1E+002 2

1E+001 1
0.11 1 10 100 1000 10000
Frequencyy (Hz)

Figuree 7 EIS of Phase II, sample set B, after accelerated weathering at the National
GalleryGallery of Art.

Thee fair correspondence between EIS results of the two sets of Phase II
sampless indicates that the weathering protocols followed in the laboratories in
Northh Dakota and the National Gallery of Art were basically comparable, although
weatheringg at the North Dakota laboratory appears to have been somewhat harsher
andd more accelerated. EIS may thus predict behavior further ahead than the
acceleratedd weathering conducted at the National Gallery.*. The increased
accelerationn factor in the EIS weathering protocol, as compared to accelerated
methodss used at the National Gallery of Art, may be ascribed in large part to a
greatlyy increased "time of wetness" in the salt fog equipment. In the absence of
automatedd fogging or spraying equipment at the National Gallery, approximately 5
minutess of hand spraying followed by 4 hours of 85% relative humidity did not
equall the wetness achieved by 4 hours of direct fogging. Other differences in the
acceleratedd weathering methods, such as the composition of the "acid rain'*
solutionn and duration of other cycles, may also have significance. On the other
hand,, the EIS experiments do not take visual changes in the coatings into account.
Itt must be emphasized that no method of accelerated weathering or performance
evaluationn will ever reproduce outdoor exposure exactly, which in any case is site
specific.. This underscores the importance of comparing results of different
methodss of weathering as well as different methods of analysis in the general
evaluationn of coating performance.

97 7
ChapterChapter 4

4.4.4.4. Conclusions
Preliminaryy EIS work showed visual changes to the samples during testing,
ass well as increasing failure in the coatings. Visual evidence of a "foggy film"
developingg in the samples pretreated with BTA suggested a passivating layer
formedd as an artifact of the testing procedure, and these results were discarded.
Thee ranking order from EIS for the remaining three coating systems tested on
bronzee from Phase I samples indicated a slight benefit from the addition of a wax
topcoatt to Incralac, although the performance of this coating was generally poor.
Thee StanChem waterbome acrylic urethane on polished bronze (#26) performed
fairlyy well for a time period, and then failed dramatically. This also concurs with
performancee evaluations for this coating on patinated substrates as described in
Chapterr 2, but strongly predicts dramatic failure for this coating on polished
bronze,, which was only slightly indicated in the Phase I study.
Additionall EIS testing of Phase II polished bronze samples provided
rankingss for the five coatings. These results concurred with findings of the study
describedd in Chapter 3, where the Nikolas acrylic/urethane + wax showed superior
performancee on polished bronze, and BTA + wax showed very inferior
performancee characteristics. Rankings also concurred with a mediocre
performancee level from Incralac + wax on the bronze, but EIS rankings indicated
slightlyy better performance from the BTA + BASF acrylic urethane + wax coating
onn polished bronze than were observed in Phase II. Examination of the Bode plots
andd impedance modulus vs. time of weathering data yielded additional information
aboutt the coating performance that was interpreted in terms of susceptibility to
porosityy and/or microcracking, as well as the formation of corrosion layers
underneathh the coating. EIS results also appear to predict some behavior not yet
seenn in the natural outdoor or accelerated weathering conducted at the National
Galleryy of Art, in particular that of the complete breakdown of Incralac. In
general,, results provide strong support for the findings in Chapters 2 and 3.

98 8
 EISEIS of select coatings

References References

11 John N. Murray, "Electrochemical test methods for evaluating organic

coatingss on metals: Part I. Introduction and generalities regarding
electrochemicall testing of organic coatings," Progress in Organic Coatings 30
(1997),, 2225-2233.
22 F. Mansfeld, "Electrochemical Impedance Spectroscopy (EIS) as a new tool
forr investigating methods of corrosion protection," Electrochim. Acta, 35
(1990),, 1533-1544.
33 U. Rammelt and G. Reinhard, "Application of electrochemical impedance
spectroscopyy (EIS) for characterizing the corrosion-protective performance of
organicc coatings on metals," Progress in Organic Coatings 21 (1992), 205-
226. .
44 Gordon P. Bierwagen, "Recent Developments in Coatings Science in
Corrosionn Control and Durability: Implications for Art Conservation,"
Abstracts,Abstracts, AIC 25,h Annual Meeting, 9-15 June 1997, San Diego, CA, 7-9.
55 C. Price, D. Hallam, G. Heath, D. Creagh, J. Ashton, "An Electrochemical
Studyy of Waxes for Bronze Sculpture," in Metal 95. Proceedings of the
InternationalInternational Conference on Metals Conservation, ICOM-CC Metals Working
Group,, Sept. 1997, Semur-en-Auxois, France, ed. Ian D. MacLeod, et al.
(Jamess & James Ltd., 1997), 233-241.
66 V. Otieno-Alego, G. Heath, D. Hallam and D. Creagh, "Electrochemical
evaluationn of the anti-corrosion performance of waxy coatings for outdoor
bronzee conservation," in Metal 98. Proceedings of the International
ConferenceConference on Metals Conservation, ICOM-CC Metals Working Group, May
1998,, Draguignan, France, ed. W. Mourey and L. Robbiola (James & James,
Ltd.,, London, 1999), 309-314.
77 Paola Letardi, "Electrochemical Impedance Measurements in the Conservation
off Metals," in Radiation in Art and Archeometry, ed. D. C. Creagh and D. A.
Bradleyy (Elsevier Science, 2000), 15-39.
88 Laboratory of Dr. Gordon P. Bierwagen, Dept. of Polymers and Coatings,
Northh Dakota State University, Fargo, ND.
99 Lisa Ellingson, Tara Shedlosky, Lynn Brostoff, E. René de la Rie, and Gordon
P.. Bierwagen, "Evaluation of Coating Systems for Protecting Bronze Using
Electrochemicall Impedance and Accelerated Test Methods," Poster
Presentationn at Electrochemical Society Meeting, Session Zl, May 2000,
Toronto,, Canada.
100 John M. Murray, "Electrochemical test methods for evaluating organic
coatingss on metals: an update. Part I," Progress in Organic Coatings 30, 4
(1997),, 225-233; John M. Murray, "Electrochemical test methods for
evaluatingg organic coatings on metals: an update. Part II: Single test parameter
measurements,"" Progress in Organic Coatings 31,3 (1997), 255-264; John M.

99 9
ChapterChapter 4

Murray,, "Electrochemical test methods for evaluating organic coatings on

metals:: an update. Part III: Multiple test parameter measurements/' Progress
inin Organic Coatings 31, 4 (1997), 375-391.
111 C. Gabrielli and M. Keddam, "Review of Applications of Impedance and
Noisee Analysis to Uniform and Localized Corrosion," Corrosion 48, 10
(Octoberr 1992), 794-811.
122 A. Mixzczyk and K. Darowicki, Corrosion Science, 40, 4/5 (1980), 663-672.
133 Internal report prepared for National Gallery of Art, Washington, DC, by
Jamiee Welton, laboratory of Dr. Gordon Bierwagen, Dept. of Polymers and
Coatings,, North Dakota State University, Fargo, ND, 1997.

100 0
 55
Chemicall characterization of the bulk coating
andd the metal/coating interface

Abstract Abstract
Selectedd coatings were analyzed for bulk characteristics by various
chemicall techniques, including Fourier-transform infrared spectroscopy and
pyrolysiss gas chromatography/mass spectroscopy (Py-GC/MS). In addition,
reflection-absorptionn infrared spectroscopy and attenuated total reflection
spectroscopyy were used to investigate the interfacial regions formed by selected
coatingss on different substrates, including glass, rolled copper, and rolled and
polishedd bronze substrates. Interfacial regions were investigated before and after
acceleratedd weathering where possible. These results are compared to infrared
spectraa and Py-GC/MS data of the bulk polymers. In the context of observations
aboutt coating behavior on bronze and copper substrates made in Phases I and II of
thee research, these results provide insight into reactions at the metal/coating
interfaces,, as well as rationale for some performance characteristics of the
coatings. .

5.7.. Introduction
Physicall evaluation of weathered coatings on bronze and copper roof
sampless in Phases I and II did not show dominant trends that correlate overall
performancee and any one property such as thickness or adhesion. As discussed in
Chapterss 2 and 3, these results point to coating performance being contingent on a
complexx play of interdependent factors, including coating thickness, coating
quality,, adhesion, and inherent coating qualities such as water and oxygen
permeability,, flexibility, effect of additives, and chemical stability. An important
avenuee of investigation into coating failure is chemical characterization of the
metal/coatingg interface, combined with comparative analysis of the bulk coating.
Studiess of polymer/metal systems have shown that at the metal surface,
interfaciall regions that have different chemical compositions and properties than
thee bulk polymer are often formed [1,2]. The chemistry of these interfaces has
profoundd effect upon coating bond strength and durability, either positively or

101 1
ChapterChapter 5

negativelyy [3]. Infrared reflection techniques, including reflection-absorption

infraredd spectroscopy (RAIR) and attenuated total reflection spectroscopy (ATR).
havee been used by researchers such as Chan and Allara [4] to chemically
distinguishh metal/coating interfaces from the bulk of a coating. In both RAIR and
ATR.. detection may be limited to several microns or less of the surface under
appropriatee conditions. In the case of RAIR (Figure 1). reflection is external, at a
glancingg angle, through thin polymer films on metal (<1000 A). RAIR offers the
advantagee of providing information about an intact interface. In the case of ATR
(Figuree 2), reflection is internal through a crystal pressed against a polymer film
thatt has been separated from the substrate. Although the actual interface is
disturbedd in this technique, ATR of delaminated coatings from metal allows direct
examinationn of the polymer side of the interface.

AIRR (n,)
hi' '
Li
ar ir "COATINGG (n,)
M E T A LL (n 3 ) (thicknesss <10üÖ A)

aa = 75-85'

Figuree 1 Optical scheme of reflection-absorption infrared spectroscopy (RAIR)

(ii/,(ii/, //_>, and n< are refractive indexes).

INTERNALL REFLECTION
E L E M E N TT (IRE) (n,)
h i '' ..

^^
=r r SAMPLEE COATING (n2)

Figuree 2 Optical scheme of attenuated total reflection spectroscopy (ATR) (n/, and n:, are
refractiverefractive indexes).

102 2
 ChemicalChemical characterization of the bulk and interface

Inn order to increase our understanding of the complex chemical interactions

betweenn individual coatings and substrates, selected coating/metal interfaces from
thee model samples were analyzed by ATR and, in some cases, RAIR. Results of
spectroscopicc analyses are presented here for Incralac, three solvent-borne acrylic
urethanes,, two waterbome acrylic urethanes, and a wax coating. Limited
discussionn of B-48 and the Nikolas acrylic coating is also included. Results are
discussedd in light of analysis of the bulk coating and observations on coating
performance. .

5.2.5.2. Experimental methods

5.2.1.5.2.1. Sample preparation

Selectt samples for interfacial analysis were prepared in Phase I; these
sampless are described fully in Chapter 2. For comparative chemical analyses,
sampless of the coatings were also applied onto glass microscope slides, aluminized
mirrorr slides, rolled copper (99.99% Cu), and rolled bronze coupons. The
aluminizedd mirror slides were purchased from Aldrich. The rolled bronze,
purchasedd as sheet from Lubaloy Co., was spring-tempered, 425 bronze, 0.016
gauge,, with an alloy composition of 88.5% Cu, 9.5% Zn, 2.0% Sn. Rolled copper
andd rolled bronze were polished with a series of Micro-mesh cloths from either
24000 or 6000, to 12000 mesh. The polished samples were solvent cleaned by
wipingg and rinsing with alternating polar and non-polar solvents until they passed
thee water drop break test [5], rinsed thoroughly with ethanol, and air-dried.
Mirrors,, glass, and copper substrates were about 1 x 3 inches, and rolled bronze
couponss were roughly 3 x 5 inches. Coatings were applied onto the rolled bronze
withh a 2 mil (51 um) draw-down bar during Phase II.
Sampless designated as "weathered" in this chapter underwent accelerated
weatheringg as conducted in Phase I or II. Experimental details of the Phase I
automotive-typee accelerated weathering program are found in Chapter 2. RAIR
analysiss of a model sample of Incralac on rolled copper also entailed accelerated
weathering,, which in this case consisted of immersion in the acid rain solution,
alternatedd with humidity and temperature cycling. Accelerated weathering of
coatingss on glass and rolled copper was conducted alongside Phase I samples, and
weatheringg of additional coatings on glass, aluminized mirror slides and rolled
bronzee coupons was conducted alongside Phase II samples and is described in
Chapterr 3; these samples weathered more quickly, and were removed after 46 days
duee to their advanced deterioration at this stage.

5.2.2.5.2.2. Bulk chemical analysis

Variouss techniques of Fourier-transform infrared spectroscopy (FTIR) were
usedd to characterize bulk coatings with and without exposure to accelerated
weathering.. These primarily included microtransmission with a diamond cell
accessory,, and specular reflectance from aluminized mirrors or rolled bronze using

103 3
ChapterChapter 5

aa Harrick fixed angle specular attachment, with 64 scans per spectrum. All FTIR
spectraa were collected in absorbance at a resolution of 4 cm"1 with a Bio-rad FTS-
60AA equipped with a UMA300A microscope, and normalized to appropriate bands
forr direct comparison of concentration of various functional groups, according to
thee Beer-Lambert Law, where absorbance is directly proportional to concentration.
Thee bulk coatings were also analyzed by pyrolyis-gas
chromatography/masss spectrometry (Py-GC/MS) using a CDS 2000 Pyroprobe
mountedd directly onto the injector of a Varian 3500 capillary gas chromatograph
equippedd with a RTX-1 column (32mm ID, 30 m, 0.25 pm film thickness) and
interfacedd to a Finnigan 800 Ion Trap Detector (ITD). Pyrolysis was carried out at
6000 °C for 10 sec; the Py-GC interface was held at 300 °C; the GC oven was
programmedd with an initial temperature of 40 °C (held for 5 min.) and increased to
300°CC (held for 10 min.) at a rate of 8 °C/min. The column was interfaced directly
too the ITD; the transfer line was at 250°C. The scan range was 35-550 amu, scan
timee one sec, and data analysis was with Finnigan ITD 4.10. The split/splitless
injectorr was at 300 °C in the split mode, with a split ration of about 100:1; the
carrierr gas was helium.
Removabilityy tests were conducted according to a standard protocol [6],
usingg varying mixtures of cyclohexane and toluene. Approximate solubility of the
filmfilm was determined under magnification as the point at which the coating began to
adheree to several cotton fibers that were gently trailed across a film surface.
Becausee Incralac is highly soluble in toluene, the starting solvent for this test was 4
mll of cyclohexane, and toluene was added in 0.1 ml increments up to 4 ml or until
solubilityy was detected. A topcoat of wax, where present, was first removed from
thee test spot with cyclohexane. The solvent mixture at which dissolution of the
coatingg was first visible was reported in terms of percent toluene.
Att a later date, gel content was analyzed in several coating or resin films.
Smalll samples were dissolved in toluene, mixed for 24 hours on a laboratory
shaker,, and injected through 0.45 p Gelman syringe filters. The amount of
insolublee material present was determined gravimetrically from the injection
volumee and weights of the sample, solution, and filter before and after filtration.
Thee sample vial and syringe were also washed with solvent, which was passed
throughh Whatman #40 filter paper or the Gelman filter in order to trap undissolved
residue.. The Gelman filters were preconditioned before use by injection with
solvent,, air dried, and oven dried for several hours at 34 °C. After filtration, the
filtersfilters were again air-dried, oven-dried, and conditioned in a controlled
temperaturee and humidity room. Filter weights were determined on a Mettler
H51ARR balance.
HPLCC analysis was conducted with a Waters 2695 Separation Module,
Waterss 2996 Photodiode Array Detector from 210-400 nm, and an Xterra® RP18
3.55 pm, 4.6 x 50 mm column on line; the flow rate was 1.0 ml/min. The HPLC
gradientt eluent consisted of 1 min. hold (a 90% H2O/10% methanol, 3 min. ramp
too 100% methanol, 2 min. hold % 100% methanol, and 10 sec. drop back to 90%
H 2 00 /10% methanol. Samples were dissolved in THF or methanol mixtures, and
weree injected in 5 or 10 pi aliquots by an autosampler. All solvents were HPLC-

104 4
 ChemicalChemical characterization of the bulk and interface

grade;; THF was also stabilized. The BTA reference had a purity of 99%. Data
wass processed by Empower software.

5.2.3.5.2.3. Interfacial analysis

Forr investigation of polymer interfaces, RAIR and ATR were used. RAIR
wass performed using a Harrick Versatile Reflection Attachment for center-focused
beamm fitted with a Retro-Mirror Accessory and a wire-grid polarizer set for parallel
beamm polarization. The angle of incidence was 78° or 85°, and 1024 scans were
collectedd for good resolution. For ATR, coatings were delaminated from metal
substratess by immersion in liquid nitrogen, or simply peeled off when possible
afterr accelerated weathering. Coatings on rolled bronze coupons were in many
casess easily peeled off a curled strip of coated metal that was created by cutting
withh a metal snip.
ATRR spectroscopy was performed with a Harrick 4x beam condenser fitted
withh a delrin holder for SPP crystals with faces measuring 10 x 5 x 1 mm. The
IREE crystals were KRS-5 45°, KRS-5 60°, and Ge 45°. Calculated sampling
depthss of the evanescent wave for these crystals were, respectively, 2.92, 1.56, and
1.533 um. Depth of penetration, dp, was calculated as follows:

dp=Xi/{27r[sin2a-(n2/n,)2]l/2}, ,

wheree a is the angle of incidence, X\ the wavelength of radiation in the optically

denserr medium, and ni and n2 are the refractive indexes [7]. For samples that were
smallerr than the crystals, exposed areas were masked with aluminum foil. ATR
spectraa were collected with 1024 scans and corrected for wavelength dependency
usingg the "atrcorr.ab" function in the Grams 32 or Win-IR software.

105 5
ChapterChapter 5

5.3.5.3. Results and Discussion

Chemicall investigation of the coatings was conducted throughout Phases I
andd II of the research project; specific results are presented below.

5.3.1.5.3.1. Acrylics

5.3.1.1.5.3.1.1. Incralac

Thee most widely used polymer coating in the field of outdoor bronze
conservationn is Incralac. According to the manufacturer, it is formulated from
Paraloidd B-44*, a methyl methacrylate (MMA) copolymer, and includes
benzotriazolee (BTA), silicone oil, and an unspecified UV absorber [8], Py-GC/MS
analysiss of Incralac showed that the major depolymerization product after
pyrolysiss is MMA, as is expected for this polymer. There were also two smaller
fractions,, identified as ethyl acrylate (EA) and ethyl methacrylate (EMA). Both of
thesee fractions were also identified in Py-GC/MS of a sample of B-44 resin,
indicatingg the resin and coating are copolymers principally of these three units. In
addition,, a very small fraction of butyl methacrylate (BMA) was identified in
Incralac,, but was not found in B-44. These results differ slightly from Chiantore
andd Lazzari, who estimated the copolymer composition of B-44 resin to be about
1%% BMA, 28% EA, and 70.3% MMA (mol %) [9]. While the two sets of results
aree not in strict conflict, they lend some credence to occasional speculation by
conservatorss that batch differences may exist in Incralac [10].
Thee presence of BTA in Incralac was confirmed by HPLC. As shown in
Figuree 3, the HPLC chromatogram of BTA shows a single, strong peak at 3.12-
3.144 minutes in the gradient elution. This peak has a distinct UV spectrum, with a
maximumm at 255.1 nm in THF or 259.9 nm in methanol (as shown), in agreement
withh the literature [11]. HPLC chromatograms of Incralac from films on glass,
whichh were either fresh, stored in the laboratory, or weathered and stored in the
laboratory,, all eluted a peak due to BTA. It was noted that BTA was more difficult
too detect in weathered Incralac by this method. Results confirm a difference in the
ratioss of peak area of BTA to the total area of several unidentified peaks attributed
too the coating at 3.650, 4.146, and 4.604 minutes. In the fresh Incralac film, this
ratioo was calculated as 1:6, and in the aged Incralac film, the ratio was 1:4. This
impliess some loss of BTA from the coating during weathering.
Bulkk unweathered Incralac coatings on glass, aluminized mirrors, and
polished,, rolled bronze and copper were examined by FTIR microtransmission and
specularr reflectance. The various spectra correlated well to each other and to
publishedd spectra [12]. A representative spectrum of the bulk unweathered film is
shownn in Figure 4a. The spectrum is dominated by the ester carbonyl stretching
vibrationn centered near 1738 cm ' in specular reflectance. The profile of this peak
includess a very weak shoulder near 1686 cm', which is also apparent in published

** All Paraloid resins were formerly sold under the name of Acryloid resins in the United
Statess by Rohm and Haas, Inc.

106 6
 ChemicalChemical characterization of the hulk and interface

referencee spectra [12] and indicates the presence of monomer. The prominent
peakss near 1235 and 1145 cm"1 may be assigned to ester CO stretching vibrations
[7,9].. After weathering, no changes were detected in FTIR spectra of the bulk
films. .

II n c r a l a c
(THF))

AA --
W e a t h e r e dd I n c r a l a c
(THF/MeOH))

BB AA
BTAA
(THF))

CC
T H FF

DD JL L
O.OOO 0.50 LOO 1.50 2 . 0 00 2.50 3 . 0 0 3 . 5 00 4.00 4.50 5.00
Minutes s

228.00 0

SampleNamee BTA Retention Time 3.137 SampleNamee incralac-thf Retention Time 3.14!

Figuree 3 HPLC chromatograms (above) extracted at 254 nm for A) Incralac in TUF, B)

weatheredweathered Incralac in THF/methanol, C) BTA in methanol, and D) THF, shown with UV
spectraspectra (below) of peak eluted at 3.14 minutes from BTA in methanol (left) and Incralac in
THFTHF (right).

107 7
 ChapterChapter 5

1.6 6--
1.4 4 --
co o
1.2 2--
CM M

CO O

11
in n
.8 8

.6 6 coo II
CM/ /
.4 4 88
en en
c\i i
.2 2

II I
A. . A Aspecularr reflectance
JJ
/ A\ \A B BATR, ,KRS-55 45
ii i ii ii JJ
3500 0 3000 0 25000 2000 1500 0 1000 0
Wave-numberr (cm-1)

Figuree 4 FTIR spectra of unweathered Incralac: A) bulk film on rolled bronze, specular
reflectance,reflectance, and B) surface of free cast on glass. ATR (KRS-5 45°). Spectra are
normalizednormalized to CI I stretch bands near 2950 cm'.

5.3.1.1.1.5.3.1.1.1. The metal/coating interface

ATRR spectroscopy was used to compare weathered and unweathered
interfaciall surfaces of Incralac delaminatcd from glass, copper, rolled bronze, and
castt bronze. Figure 4b shows the ATR spectrum of unweathered Incralac from the
coating/glasss interface. Note that ATR spectra of Incralac generally showed a shift
inn the carbonyl position to about 1728 cm'1, as well as slight differences in relative
peakk intensities toward either end of the spectrum. These differences may be
attributedd to artifacts of the techniques and the ATR correction function, and were
nott included in spectral interpretations. Otherwise, the ATR spectra of Incralac
correspondedd well to transmission and specular reflectance spectra.
Whilee changes were not detected in the spectra of glass or air interfaces
afterr weathering, slight changes were revealed in spectra of the weathered polymer
interfacess delaminated from bronze. Figure 5 shows details of several ATR spectra
(KRS-55 45°) of thin and thick films of Incralac, with and without weathering,
comparedd to the film on glass in Figure 4b. A spectrum of the thin Incralac film on
castt bronze after weathering, taken with a KRS-5 60° IRE for less depth of
penetrationn into the interfacial region, i.e.. 1.56 urn vs. 2.92 urn, is also shown.
Spectraa are normalized to the CH stretch bands near 2950 cm"1 for direct
comparison. .

108 8
 ChemicalChemical characterization of the bulk and interface

.45 5
.4 4

JJ 5

33
.25 5
oo

11 -15
<< .1
.05 5
00
-.05 5
-.1 1
18000 1750 1700 1650 1600 1550 1500 1450 140(
VVavenumberr (cm-1)

Figuree 5 Detail of ATR spectra (KRS-5 45°) of Incralac: A) free film, no weathering
(wide(wide solid line); B) coating on rolled bronze, no weathering, C) #8a, thick coating on cast
bronze,bronze, no weathering; C) U8b, thick coating on cast bronze, weathered; D) #lb coating
onon cast bronze, weathered; and E) (KRS-5 60°) #lb coating on cast bronze, weathered.
SpectraSpectra are normalized to CI I stretch bands near 2956 cm' (not shown).

Closee inspection of the carbonyl regions in these spectra illustrates that

subtle,, consistent differences exist in the relative size and shape of the peak
shoulder,, depending on the substrate and aging conditions. The ATR spectrum of
thee coating/ glass interface (wide solid line) shows a very small, weak shoulder
nearr 1686 cm"1, as noted in the bulk film. Increased absorbance is observed here in
spectraa of polymer interfaces from either unweathered rolled bronze, a thick
Incralacc coating on bronze (Phase I, #8) before weathering, or the latter after
weathering.. These results suggest that relatively more unsaturated species, such as
monomer,, and possibly limited amounts of carboxylic acids, are trapped near or
migratee to the bronze surfaces in cast and sprayed coatings. It is not clear whether
thiss also indicates some type of limited interaction between the polymer and the
metall surface, as might be expected with carboxylic acids.
Resultss are unambiguous, however, in two respects: 1) the species
absorbingg in this region of ATR spectra from a thick coating on bronze were little
affectedd by the weathering regime, and 2) marked change occurred in this region in
aa weathered, thin Incralac coating on cast bronze (Phase I. #lb). ATR spectra from
interfacess of the thin Incralac on bronze (Figures 5d and 5e) show the development
off a markedly higher and broader shoulder on the carbonyl band, especially from
aboutt 1660-1600 cm' 1 . This points on the one hand to limited degradation through
scissionn reactions, resulting in increased amounts of unsaturated and conjugated

109 9
ChapterChapter 5

species.. This agrees with Chiantore et al., who reported various degrees of
broadeningg of the carbonyl band near 1638-1640 cm"1, which were associated with
chainn scission reactions during thermal and photo-oxidative aging of different
acrylicc resins [20].
Onn the other hand, the observed carbonyl broadening in Figures 5d and e is
alsoo consistent with the formation of copper carboxylate salts at the metal
interface.. Chan and Allara, among others, have attributed growth near 1640, 1600
andd 1430 cm" in RAIR spectra of oxidizing hydrocarbon polymers in contact with
copperr surfaces to carboxylate salt formation and its subsequent catalysis of
degradationn reactions [4]. Copper ions and many copper salts are well known to
actt as catalysts in degradation reactions [13,14,15]. Results thus provide evidence
thatt the metal/coating interface in bronze/Incralac systems serves as an important
sitee for the initiation of degradation. This effect is apparently delayed in thicker
coatings. .
RAIRR was used to extract further information about the intact coating/metal
interfacee of a thin film of Incralac on polished, rolled copper, which was subjected
too an acid rain immersion/ humidity cycling weathering regime. RAIR spectra
weree taken at different points during weathering, and are shown in Figure 6.

copper r
hydroxy y
chloride e
Cu,0 0

3500 0 25000 2000 1500

Wavenumberr (cm-1)

Figuree 6: RAIR spectra of very thin film of Incralac on rolled copper with simulated
acceleratedaccelerated weathering for: A) 0 days, B) 4 days, C) 7 days, and D) 13 days. Spectra are
normalizednormalized to CI I hands near 2956 cm'.

110 0
 ChemicalChemical characterization of the bulk and interface

Resultss clearly illustrate the development of copper corrosion minerals

beneathh the intact polymer film. Reading up the figure, we see the progressive
increasee of peaks near 645 cm'1, arising from cuprous oxide, and near 3449, 3358,
andd between 800 and 1000 cm"1, indicating the formation of copper hydroxy
chloridee salts. Appearance of a shoulder near 1100 cm"1 in Figure 6d suggests the
beginningss of a copper hydroxy sulfate, although this may also be part of a general
broadeningg in this region due to oxidation. Unfortunately, atmospheric moisture at
thiss level of absorbance could not be eliminated, and interfered with a reading of
thee carbonyl region. Still, the RAIR spectrum of the polymer appears little
changed,, with the exception of evidence for hydroxyl species from about 3600-
25000 cm"1. This may be associated with the corrosion minerals on the metal
surface,, as well as carboxylic acids on the polymer surface. What these spectra
bestt illustrate is how corrosion may displace the coating at the metal interface
duringg weathering. This will inevitably cause de-adhesion of Incralac to the metal
surface. .
Importantt information about the Incralac coating/metal interface with
respectt to BTA was further obtained by RAIR. The function of BTA as a coating
ingredientt has been the subject of much speculation over the years. In particular,
theree has been debate about whether BTA functions primarily as a UV absorber or
aa corrosion inhibitor. Chapter 6 contains a detailed discussion of this experiment,
whichh showed that copper-BTA (CuBTA) remains on the copper alloy surface
afterr removal of an Incralac coating. To summarize, after immersing the coupon in
Incralacc for 2 minutes, followed by solvent rinsing of the deposited film, RAIR
clearlyy revealed the presence of the insoluble CuBTA complex.
Sincee the absorption bands due to CuBTA (or BTA) are normally swamped
outt in spectra of Incralac, RAIR is used here to ascertain that application of
Incralacc to copper, and by straightforward assumption to bronze, deposits a
corrosion-inhibitingg film at the metal/coating interface. Most of the commercial
coatingss tested in this study purported to have BTA as an ingredient, and this was
confirmedd for Incralac and the Nikolas coatings. It can therefore be assumed that a
CuBTAA film forms at the metal/coating interfaces on copper alloys during
applicationn of other coatings as well.
Inn this light, results shown in Figure 6 also verify that the CuBTA film did
nott inhibit corrosion effectively under this set of experimental conditions. It
shouldd also be noted that no difference was detected in ATR spectra of various
coatingss from metal substrates that had been pretreated with BTA, compared to
spectraa of coatings without the BTA pretreatment. It may be speculated that the
existencee of a thin CuBTA film between the metal and the coating serves to offset
thee increased likelihood of oxidation initiation at the metal interface for both the
metall and the coating to some degree, but only in a very limited capacity. The role
off BTA on copper alloys is further discussed in Chapter 6.

Ill l
ChapterChapter 5

5.3.1.1.2.5.3.1.1.2. Solubility
Removabilityy tests, as described in the Experimental section, give an
indicationn of the solubility of the major fraction in an organic film. Removability
off a coating may not change appreciably until a fair amount of chemical change
hass taken place. Removability testing of Incralac indicated that 11-13% toluene in
aa toluene/cyclohexane mixture was necessary to remove the Incralac from all of
thee bronze or copper samples, either before or after accelerated weathering. In
otherr words, no significant increase in the polarity of the solvent necessary for
coatingg removal occurred in any of the cases. These results concur with a thermal
agingg study by Lazzari and Chiantore [16], who reported that B-44 resin remained
completelyy soluble (in chloroform) after isothermal aging at 110 and 135 °C for up
too 200 hours. These authors found that during thermal degradation, polymer
fragmentationn competes with crosslinking reactions, and the former prevails.
Itt is relevant to this discussion that the films weathered on glass and copper
alloyss were fairly brittle and could only be removed in small fragments for ATR
sampling.. The fragments themselves were quite tough, and could not be
compressedd in a diamond cell to an appropriate film thickness for
microtransmission.. In addition, the unweathered film on glass could not be
removedd as a coherent film, only scraped. Thus, despite continued solubility,
physicall changes in Incralac coatings after weathering were apparent.
Afterr several years of storage, some sample films were subjected to HPLC
analysis,, at which time they were also observed to have become extremely brittle
andd have poor solubility in both THF and toluene. Therefore, gel content was
quantitativelyy determined by filtration of toluene solutions of several storage-aged
filmsfilms remaining from Phase I, using 0.45 u filters (as described in the
Experimentall section). Results are shown in Figure 7. The films included:
unweatheredd B-44 and B-48 on glass slides, Incralac on cast bronze (Phase I,
samplee #1), and weathered B-48 and Incralac on glass slides. Among these
samples,, the unweathered films were clear and firmly adhered to the substrates.
Thee Incralac-coated bronze appeared shiny and unchanged; the average film
thicknesss of this sample (as reported in Chapter 2) was 0.68 mils (17.27 u). The
weatheredd B-48 film (2.7 mils, 68.58 u) appeared clear and somewhat brittle; it
retainedd cohesion and was lifting off the glass. The weathered Incralac film (0.60
mils,, 15.24 \x) appeared slightly yellowed, and could only be scraped off in small
shards. .
Resultss indicate that both Incralac and B-48 are susceptible to gel formation
duringg the combination of simulated, accelerated weathering and subsequent,
extendedd storage; presumably this arose from chemical crosslinking. Noticeable
pressuree during filtration of the aged film/toluene solutions confirmed this result.
Noo pressure was apparent during filtration of the other solutions. The relative
differencee in solubility between the weathered and unweathered films appears
significant.. In addition, the slightly higher amount of insoluble material detected
inn the toluene solution of Incralac on cast bronze may support the contention that
Incralacc has an increased tendency to undergo degradative chemical change in

112 2
 ChemicalChemical characterization of the bulk and interface

contactt with copper alloy substrates, Thiss difference may not be statistically
significant,, however.

20.000 -,
filmss on glass or weatheredd films
bronze,, + lab onn glass, + lab
-- 15.00
storagee b years storagee t> years

aa 1000 0
cc
oo
O O 5.00 0
0) )

0.000
Blankk Incralac, B-44 B-48 Incralac B-48 Incralac
solventt new film on
(Al)) bronze
Coating g
Figuree 7 Gel content (weight percent) in various coatings or resins in toluene solutions.

Thee development of chemical changes on a micro-scale, followed by stress

build-upp in the coating during weathering, may have only minor effect on the
solubilityy or removability of a coating. Insoluble gel at the level of 10 % does not
renderr the test samples completely insoluble. However, these results reinforce one
reportt of crosslinking in the literature [17], as well as scattered reports in the field
thatt Incralac often becomes brittle and difficult to remove after many years of
naturall outdoor weathering [18]. The fact that no change in solubility was
measurablee by removability tests earlier in the study relates to the lower sensitivity
off removability tests, but also, it seems, to continued aging of the films during
aboutt five years in storage.
Inn this regard, it is worth mentioning that the minor copolymer units of EA
orr BMA in B-44 and B-48, respectively, are known to be quite susceptible to both
thermal-- and photo-induced oxidative degradation [19,20]. Typically, for example,
BMAA polymers have long induction periods in which little change is observable,
followedd by periods of rapid change in properties. As minor fractions incorporated
intoo more stable copolymer units, such as MA, EA, or MMA. crosslinking
reactionss are not thought to compete effectively with scission reactions during
aging,, so that the copolymers do not become insoluble; this has been reported for
ParaToidss B-72 (MA/EA/BMA copolymer) and B-82, (MMA/EA copolymer) [21].
Itt is also feasible, however, that small amounts of less stable chemical units in B-
44.. Incralac or B-48 form scattered centers within the copolymer matrix, as well as
att the metal surface, where degradation may be initiated. Even before change in
solubilitvv is evident, this could manifest itself as film toimhenin<j or embrittlement.

113 3
ChapterChapter 5

i.e.,, as a form a kind of physical crosslinking, which may even aid protection of the
metall up to the point at which cracking and disbondment begin to occur. In this
light,, results of interfacial analysis suggest that the copper ions near the metal
surfacee may activate less stable centers of the copolymer and serve as an important
initiationn site for oxidative degradation in an otherwise stable polymer.

5.3.1.2.5.3.1.2. Other acrylics

Paraloidd B-48 (Röhm and Haas, Inc.) is a higher molecular weight MMA
copolymer,, which, according to the manufacturer's literature, has "unique"
adhesionn to treated or untreated metals, is more flexible than B-44, and is
recommendedd for use in the Incralac formulation instead of B-44 [22]. The
weight-averagee molecular weight (Mw) of B-48 has been reported to be 184,000,
comparedd to a Mw of 105,000 for B-44 [9], This would largely account for
increasedd flexibility in the copolymer. It is also known from the supplier's
literaturee that B-48 has a slightly lower glass transition temperature (Tg) than B-44
(50°Cvs.. 60°C).
Py-GC/MSS analysis of Paraloid B-48 revealed three large depolymerization
products,, identified as MMA, butyl acrylate (BA), and BMA. There were also two
unidentifiedd components. Chiantore and Lazzari reported the copolymer
compositionn of B-48 as 74.5 mol % MMA and 25.5 mol % BMA units [9]. The
FTIRR spectrum of bulk B-48 (not shown) showed only minor differences from that
off Incralac (Figure 4), including relatively increased methylene absorbance in the
CHH stretch region from 3000 to 2800 cm"1, and differences in the so-called
fingerprintfingerprint region, i.e., 1100-500 cm"1. No indications of chemical modifications
thatt might be responsible for increased adhesion to metals, such as inclusion of
acidd functionalities, were detected by these methods.
Solubilityy of B-48 films is shown in Figure 7 and discussed briefly above.
Thee greater gel content measured in films of B-48 vs. B-44 after about 5 years of
storagee does not appear to be statistically significant. However, a weathered film
off B-48 clearly exhibited change in solubility and gel formation after the same
periodd of prolonged storage, as did Incralac. The film also exhibited physical
changes,, such as peeling, cracking, and loss of flexibility.
Inn Phase I of the research, two coatings (#10 and 11) were specially
formulatedd based on B-48 resin (Cape Cod Research, Inc.). The other acrylic
coatingg used extensively in the study was the Nikolas 11565 Outdoor Lacquer.
Thee manufacturer states that this is a "modified acrylic" lacquer designed for
exteriorr brass and bronze, which is supplied ready for use and contains BTA as a
"chelatingg agent," as well as UV absorbers; in addition,"[b]ecause of its excellent
flowingg capabilities, orange peel is virtually eliminated" at thinner coatings [23],
Thee coating was observed to have good application properties, but was also noted
too have slight coloration before aging.
Py-GC/MSS of the Nikolas acrylic coating confirmed that the major polymer
fractionn is MMA, with additional fractions of butyl acrylate and BMA, indicating
thee base resin is similar to Paraloid B-48. HPLC analysis showed a peak identified
ass BTA eluting at 3.125 minutes. The chromatogram (Figure 8, top) is similar to

114 4
 ChemicalChemical characterization of the bulk and interface

thatt of Incralac, except that the relative quantity of BTA in the coating appears
muchh higher, given the same injection volume and about equal solution
concentrations. .

NKK 1 1 5 6 5
(THF) )
AA
AA
NKK 9 7 7 8 , P a r t A
(THF) )
BB
NKK 1 1 6 5 0
(THF) )
cc W_ _
Incralac c
(THF) )
DD - A --
BTA A
(THF) )
EE
T H F --

0.000 0.50 LOO 1.50 2 . 0 00 2 . 5 0 3 . 0 0 3 . 5 00 4.00 4.50 5.00

Minutes s

Figuree 8 HPLC chromatograms extracted at 254 nm for A) Nikolas 11565 acrylic lacquer
inin THF, B) Nikolas 9778, part A acrylic polyol in THF, C) Nikolas 11650 waterhorne
acrylicacrylic urethane, methanol extract, D) Incralac in THF, E) BTA in THF, and F) THF.

FTIRR microtransmission and specular reflectance spectra of bulk Nikolas

115655 lacquer showed good correspondence on either glass, rolled bronze, or
aluminizedd mirror. The latter is shown in Figure 9 (wide line). These spectra
confirmedd a match with Paraloid B-48, with the exception of two weak absorbance
bandss near 3533 and 3441 cm'1, and a distinct shoulder near 1721 cm'. These
spectrall features suggest that the Nikolas acrylic is slightly modified in terms of
hydroxyll and/or acid functionalities. Such modification could be incorporated
specificallyy in order to improve adhesion to metals, although this was not reflected
inn our measurements in Phase I (see Chapter 2).
Afterr accelerated weathering for 46 days, slight broadening of the carbonyl
peakk and relative reduction in absorbance near 1150 cm" arc observed in the
spectrumm of the bulk film (Figure 9, thin line). These changes suggest limited
developmentt of unsaturation, as well as the loss of some ester groups, and support
thee notion that scission-produced conjugated double bonds, i.e.. chromophores,
developedd in the coating during accelerated weathering.

115 5
ChapterChapter 5

1.4 4

1.2 2

11

.8 8

.6 6

.4 4

COLD D
.2 2

(1 1 KP--
3500 0 3000 0 25000 2000 1500 0 1000 0
Wavenumberr (cm-1)

Figuree 9 FTIR specular reflectance of Nikolas 11565 Outdoor Lacquer, bulk film on
aluminizedaluminized mirrors, A) unweathered (wide solid line), and B) weathered (thin solid line).
SpectraSpectra are normalized to CH stretch bands near 2950 cm '.

Resultss of accelerated and natural outdoor weathering showed that no other

acrylicc coatings outperformed the Incralac series coatings, although the Nikolas
acrylicc was about equivalent to Incralac without a wax topcoat in accelerated
weatheringg tests. After outdoor weathering, however, the coatings based on B-48
resin,, as well as the single Nikolas acrylic, all lagged behind the Incralac coatings
(seee Chapter 2). Based on the analysis of B-48 and the Nikolas acrylic, it appears
thatt increased flexibility in B-48-based coatings does not make up for a slightly
decreasedd stability in simulated and natural outdoor exposures. Furthermore, we
mayy assume that lower Tg renders the film softer and more moisture-permeable.
Ass previously mentioned, the presence of BMA in coatings based on B-48 may be
significantt in terms of slightly increased susceptibility to degradation on copper
alloyss under the extreme conditions of outdoor exposure [16,19].

5.3.2.5.3.2. A cry lie ur ethanes

Solvent-bornee acrylic urcthanes represent a large and important class of
highh performance coatings on the market today. Physical properties can be
engineeredd into these systems by selection of type and functionality of the acrylic
portion,, and control of the crosslink density. Accordingly, properties vary greatly
fromm product to product. While high crosslink density generally correlates to
excellentt weathering properties, a balance must be struck between crosslinking,
flexibility,flexibility, and other application properties.
Thee solvent-borne, two-component acrylic urethane systems tested in this
studyy were developed for the metal maintenance and auto refmishing markets.

116 6
 ChemicalChemical characterization of the hulk and inter/ace

Theyy are based on hydroxyl-functional acrylic polyols (R2-OH), reacted with an

aliphaticc polyisocyanate (R|-N=C=0), to form urethane crosslinks (Ri-NH-CO-
R2),, as shown in the generalized scheme below (Figure 10).

R2 2O HH to * i - N HH „0-R2

acrylicacrylic polyol unit

isocyanateisocyanate unit OO
urethaneurethane crosslink unit
++ H 2 0

''

R.-NI-LL „OH __ R--NH- + co 2 t t

11 x
c amineamine unit ^ s ^
00
R^NH^NH-R, ,
carbamiccarbamic acid (unstable)
ÖÖ
disubstituteddisubstituted urea unit

Figuree 10 Representation of general reaction scheme for the isocyanate functional group
withwith acrylic polyol resin or atmospheric moisture.

Diisoocyanatess are typically trimerized into stable isocyanurate ring

structuress with high crosslink functionality for use in these coatings. Side products
suchh as amines, disubstituted ureas and biurets, and amides are often formed
duringg curing by reaction with moisture and other reactions products [24,25]. In
particular,, urea and amide linkages are known to be more light sensitive and have
poorerr weathering resistance than urethanes, and are therefore less desirable
products.. The formation of amine and carboxyl salts is also possible in coatings on
outdoorr metal objects, and may be undesirable in terms of moisture sensitivity.
Catalysts,, flow aids such as silicones, UV absorbers, including BTA, and
defoamerss are typically added to these coatings in relatively large quantities.

5.3.2.1.5.3.2.1. Nikolas 9778 Exterior Uralac

5.3.2.1.1.5.3.2.1.1. The bulk coating

Thee Nikolas 9778 Exterior Uralac acrylic urethane belongs to the category
off a two-part acrylic polyol/poly-isocyanate system. Py-GC/MS of the
unweatheredd coating showed that components of the uncrosslinked acrylic polyol
(partt A) consist of a copolymer of polystyrene, MMA, BA, and BMA. As shown
inn Figure 9, the presence of BTA in part A was confirmed by HPLC elution and the
matchingg UV spectrum of a peak at 3.12 minutes. Compared to Incralac and the

117 7
ChapterChapter 5

Nikolass 11565 acrylic, the 9778 part A coating contains much more BTA (relative
too the solvent peak in a solution of similar concentration).
Thiss coating was applied as a medium thick film (about 38 pm) in Phase I
(#19,, 22), as well as a middle coat over the Nikolas 11565 acrylic in Phase I (#15-
18)) and in Phase II. In all cases, the coating was a very good performer, especially
inn accelerated weathering tests. Measured adhesion of the urethane coating to
polishedd bronze was mediocre, as expected for this class of coatings (see Chapters
22 and 3).
Figuress I la and lie show FTIR microtransmission spectra of the auto-
polymerizedd crosslinking agent (part B) and the uncrosslinked acrylic polymer
(partt A), respectively. These spectra allow identification of peaks from the
isocyanuratee and its reaction products with atmospheric moisture in the first case,
andd those which belong to the acrylic in the second. Based on differences between
thesee figures and spectra of three cured resin films (Figure 11 b-d), tentative
assignmentss for various functional groups in this type of acrylic urethane were
made,, as listed in Table I [7,25].
Comparedd to Figure 11a, no new peaks that may be uniquely assigned to
urethanee linkages are obvious in the spectra of the crosslinked coatings. This
illustratess that the various functional groups of interest in the cured resin give rise
too overlapping absorbance bands. However, assuming that samples of the cured
coatingg show infrared bands predominantly due to the acrylic reaction products,
andd less from side reaction products, several features appear useful in identifying
relativee urethane crosslink density: 1) increased intensity in the C=0 peak near
17277 cm"1 (position depending on the resin); 2) the position and breadth of the peak
nearr 1521 cm"1; and 3) relative proportion of OH and NH stretching near 3520 and
33800 cm"1. Vibrations involving CN and NH stretching and deformation in side
reactionn products, including ureas and amines, appear to absorb principally near
1540-555 cm"' and 1640 cm"1. The band near 1689 cm"1 may be principally assigned
too C=0 stretching in isocyanurates, but apparently also has contributions from
urethanee and/or urea groups.
Figuree lib shows the ATR spectrum of a freshly exposed surface from
materiall that remained after spray coating and solidified in the beaker. Figures
11 lc-d show microtransmission spectra of a film cured on glass and on polished,
castt bronze (Phase I, #19), respectively. Results offer a contrasting picture of a
masss of polymer vs. a cast or spray-coated film. The FTIR spectrum of the bulk
polymerr in Figure lib is dominated by bands assigned to the isocyanurate and
urethanee functionalities, but also includes relatively fewer bands from side reaction
products.. The resin shows little similarity to the acrylic by FTIR. This indicates a
highlyy crosslinked resin. Further evidence of extensive curing is seen in the
absencee of absorption bands due to OH stretching near 3522 cm'. Thus, we can
concludee that the material cured under these conditions had an excess of
crosslinkingg agent, possibly from poor solubility in the resin and settling. The fact
thatt no residual isocyanate is detected indicates that it reacted fully.
Thesee results also imply that the other part of the batch used for spray
coatingg the metal coupons was deficient in crosslinking agent. This is borne out in

118 8
 ChemicalChemical characterization of the bulk and interface

thee spectrum from the coating on polished, cast bronze, shown in Figure 1 Id. Here
thee increased intensity of the peak near 1727 cm", and peaks near 3400 and 1521
cm',, provide evidence of urethane crosslinking, although in relatively lower
concentrationn than seen in the beaker-cured resin. The spectrum otherwise appears
similarr to that of the acrylic part A. including the continuing presence of
absorptionn bands near 3522 cm" . This indicates that crosslinking did not take
placee at all hydroxy! sites.
Thee FTIR spectrum of a thicker 9778 film applied to glass (Figure lie)
revealss an intermediate level of curing, and a predominance of urethane linkages,
ass seen in the relatively intense bands at 1727 and 1521 cm' . This spectrum
appearss to show a properly cured coating, with a higher ratio of the bands
3388:35200 cm" . In general, the differences between the FTIR spectra of these
threee cured coatings or materials point out the difficulty of working with two-
componentt curing materials and obtaining properly cured coatings.
Afterr accelerated weathering of the 9778 coating on various substrates,
FTIRR spectra of the bulk films showed little or no noticeable change, with the
possiblee exception of a slight decrease and broadening in absorbance near 1521
cm"" . Thus the bulk coating characteristics of the 9778 acrylic urethane, as applied
onn samples in Phases I and II, appear to show low-medium crosslinked character,
withh fairly good chemical stability.

35000 3000 2500 2000 1500 1000

Wavee number (cm-1)

Figuree 11 FTIR spectra of bulk, unweathered 9778 acrylic urethane (G.J. Nikolas &
Co.),Co.), A) 9778 part B, auto-polymerized, microtransmission, B) solidified material from
beaker,beaker, ATR (KRS-5 45°), C) film on glass, microtransmission, D) film on polished bronze.
(Phase(Phase I, #19), microtransmission, and E) 9778 part A resin, microtransmission. Spectra
B-EB-E are normalized to acrylic band near 845 cm ; A is normalized to 1765 cm' in B.

119 9
ChapterChapter 5

TABLEE I:
Frequenciess and Tentative Assignments of Typical Infrared Absorption Bands
inn Acrylic Polyol/Polyisocyanate Urethanes

Bandd Frequencies (cm1) Acrylic c Urethane e Tentativee Assignment^

portion n portion n
3530-34400 (broad) XX C-OH/HH OH str
3500-33000 (broad) XX NH/NH22 str; N HOH; NH OH2
str r
3378/33400 (sharp apex, XX urethane/ureaa NH str
broad) )
3080-30244 (weak) XX aryll C-H str
2987-28500 (medium-strong) XX CH2/CH33 str
2270-2230 0 XX isocyanatee N=C=0 out of phase
str r
17855 (shoulder) XX isocyanuratee C=0 str
1732-17177 (strong) XX XX esterr C=0 str + urethane C=0
1700-16855 (strong) XX isocyanuratee C=0 str/urea C=0
str/carboxylicc acid C=0 str
1660-1625 5 XX ureaa C O str/NH2 def/C=C
1617-16100 (weak) NH22 def/carboxyl salts C=0
1601-1600(weak) ) XX aryll breathing
1585-1580 0 XX ureaa CNH str-bend /NH2+ salt
def f
1556-1530 0 XX ureaa CNH str-bend
1535-1518 8 XX urethane/ureaa CNH str-bend
14933 (sharp) XX
1472-14633 (medium-strong) XX XX CH22 def (isocyanurate,/acrylic)
1456-14300 (medium-strong) XX NCOO str/CH2 def
13844 (medium) XX CH2/CH,, def/C-OH (?)
1369 9 XX XX CH2/CH33 def;C-N; NH def
1250-12100 (medium-strong) XX XX CNN (isocyanurate/ urethane/
urea); ;
C-OO ester
1212-1172 2 XX XX C-OH H
1192-11611 (medium-strong) XX C-00 (?)
1146-1139 9 XX XX C-OO ester/alcohol; CNC str.
1080-1071 1 XX XX C-OO alcohol/NH2
8455 (weak) XX aromaticc CH wag (?)
767 7 XX isocyanuratee CCO in phase str;
NHH wag ; CH2 rocking
7600 (medium) XX C-C-00 str
702-7000 (medium) XX XX aryll vibration

tt Based on IR spectra of acrylic part A resin and auto-polymerized part B crosslinking

agents,, and assignments by Colthup, et al. [7] and Urban, et al. [25],
Key:: str=stretching; def=deformation

120 0
 ChemicalChemical characterization of the hulk and interface

5.3.2.1.2.5.3.2.1.2. The metal/coating interface

Figuress 12a-c represent the polymer interfaces from cast bronze (#19,
Phasee I), with and without weathering. Sampling for ATR in this instance was
particularlyy difficult, since the coating tended to shatter into small fragments when
immersedd in liquid nitrogen. ATR results show that the unweathered coating near
thee metal interface (Figure 12a) is very similar to the bulk (Figure 11 b and d),
indicatingg good homogeneity within the coating. In a weathered sample, the ATR
spectrumm of the polymer at the metal interface appears little changed (Figure 12b),
withh the exception of a perceptible reduction of the peak near 1521 cm" , and a
moree notable increase in the OH stretching region near 3500 cm" . With approach
too the metal surface in the weathered sample (Figure 12c), we see a more
pronouncedd increase in broad OH/NH stretching band absorbance, plus
pronouncedd shifts in this region. Although poor signal to noise ratio makes
interpretationn less definitive in the region between 1700 and 1500 cm" the CNH
vibrationn near 1521 cm"' appears slightly reduced, and the carbonyl band appears
broadenedd at the base. These changes suggest increased moisture sensitivity and
acceleratedd oxidative degradation at the bronze surface, which may include
carboxyll and/or amine salt formation. Absorption near 2850 (shoulder) and 2700
cm"11 in Figure 13c is consistent with the presence of acidic hydrogen from these
functionall groups.

35000 3000 2500 2000 1500 1000

Wavenumberr (cra-1)

Figuree 12 ATR spectra of (A-Q Nikolas 9778 acrylic urethane polymer interfaces from
polished,polished, cast bronze (#19): A) before weathering (KRS-5 45°), B) after weathering
(KRS-5(KRS-5 45°), and C) after weathering (KRS-5 60°); and D) ATR spectrum of Nikolas
1156511565 acrylic/9778 acrylic urethane interface from polished, cast bronze ((#16), after
weatheringweathering (KRS-5 45 °). Spectra are normalized to bands near 2950, 845 and 760 cm' .

121 1
ChapterChapter 5

Resultss thus suggest that incomplete curing of the hydroxylated acrylic

resinn plays a role in moisture sensitivity, and that the polymer is more susceptible
too degradation at the bronze interface. While homogeneity and chemical stability
inn the bulk polymer provide rationale for good performance in this coating,
reactionss at the metal/coating interface undoubtedly forecast complete loss of
adhesionn and development of corrosion not yet witnessed during Phase I coating
evaluationn (see Chapter 2).
5.3.2.1.3.5.3.2.1.3. The metal/coating interface of Nikolas acrylic/acrylic urethane coating
ATRR spectra were obtained from the acrylic side of the metal/coating
interfacee of Nikolas acrylic/acrylic urethane two-part systems. Weathered and
unweatheredd samples were taken from Phase I, #16, i.e., from polished, cast bronze
pretreatedd with BTA, and from rolled bronze samples in Phase II. Delamination of
thee coating in all cases was difficult, and spectra were rather poor quality. After
acceleratedd weathering, delaminated coatings were slightly less difficult to obtain,
andd a legible spectrum was obtained, as shown in Figure 12d. ATR of the first
2.922 um of the weathered polymer from the polished bronze interface shows
surprisingg dissimilarity between the metal/polymer interface and the bulk acrylic
coatingg (Figure 9). The spectrum reveals that the interfacial region is in fact a
compositee of the acrylic and the acrylic urethane. In particular, peaks near 3350,
2920,, and between 1723 and 1450 cm"1, show that the interfacial region has much
off the polyurea character. A broad, single absorbance centered near 3350 cm"1
indicatess the strong presence of NH functionalities. Note that peaks at 2919 and
28477 cm also appear to have contributions from wax contamination.
Spectraa of this coating delaminated from rolled bronze similarly showed
bandss related to ureas and the isocyanurate, as well as urethanes, supporting
conclusionss that the first coating layer was in all cases quite inhomogeneous, and
thatt the layered structure of the coating was diffused. One explanation is that the
isocyanatee migrated through the acrylic layer, which was made soluble during
applicationn of the second coating system. In the absence of hydroxyl functional
groupss in this region, the isocyanate would have self-polymerized. This is a
disappointingg result in that it shows the intended composition of the layered
coatingg was altered. It is not clear how this affected performance of the system,
butt most likely caused undercuring in the top urethane layer, thereby having a
negativee impact on weathering properties.

5.3.2.2.5.3.2.2. PPGDAÖ75

DAU755 solvent-borne acrylic urethane (PPG Industries, Inc.) is another

examplee of a two-part acrylic polyol/isocyanate coating system. It was applied in
Phasee I and performed below expectations on the cast bronze, but fairly well on the
copperr roof substrates in accelerated weathering tests (see Chapter 2). This was
ascribedd in large part to the poor adhesion on cast bronze, since use of the
manufacturer'ss pretreatment on bronze before coating application (sample #21)
markedlyy improved both adhesion and coating performance.

122 2
 ChemicalChemical characterization of the bulk and interface

Py-GC/MSS of the coating showed evidence for a styrene/MMA/BMA

terpolymer.. The FTIR transmission spectrum of the bulk unweathered coating,
shownn in Figure 13b, is characterized by a small, broad, double absorption near
35200 cm" , aromatic vibrations near 3026, 1602, and 1500 cm" , and a sharp
carbonyll vibration near 1730 cm' 1 . (The region from about 2200-1800 cm'1 cannot
bee read due to absorbance from the diamond cell accessory here.)

35000 3000 2500 2000 1500 1000

VVavenumberr (cm-1)

Figuree 13 FTIR spectra of unweathered DAU75 acrylic urethane (PPG Industries, Inc.):
A)A) polymer interface from polished, cast bronze (Phase I, #20), APR (KRS-5 45 °), and B)
freefree film, microtransmission. Spectra are normalized to acrylic bands near 2927 and 845
cm'cm' .

Thee ATR spectrum of the unweathered PPG acrylic urethane coating after
delaminationn from polished bronze (Phase I, #20) shows a dramatic difference
betweenn the coating/metal interface and the bulk polymer. As seen in Figure 13a,
urethane,, urea, and isocyanurate groups, identified by vibrations near 3388, 1689,
1727,, 1525, 1430, and 1335 cm" , are concentrated at the metal interface. These
featuress indicate that there is a much higher degree of cure at the metal/coating
interfacee than exists in the bulk film. However, the lack of relative intensity in the
carbonyll peak near 1727 cm"1, and strong presence of OH stretching bands near
35000 cm" , indicate that the cure products are dominated by ureas and amines,
ratherr than urethanes, and that some crosslinking sites still remain on the acrylic.
Conversely,, the FTIR spectrum of the bulk film has the appearance largely of an
uncured,, hydroxylated acrylic resin.
Inn addition, a weak peak at 2234 cm"' in Figure 13a indicates the
persistencee of a small portion of unreacted isocyanate groups, suggesting that

123 3
ChapterChapter 5

isocyanuratee segregated out and massed at the bronze surface. This would explain
thee higher degree of isocyanate self-polymerization products in this region, and the
existencee of an interfacial region that is chemically distinct from the bulk film.
Thee poor adhesion of this coating on the cast bronze suggests that isocyanate
groupss are not complexed to the metal surface. However, its persistence here
raisess the question of whether the isocyanurate may weakly interact with a CuBTA
filmm at the surface. In either case, results suggest that not only poor adhesion, but
alsoo lack of chemical homogeneity between the bulk and the interfacial region had
significantt impact on coating performance during Phase I.

5.3.2.3.5.3.2.3. BASF 923-85

5.3.2.3.1.5.3.2.3.1. The bulk coating

Thee BASF coating is a two-component, solvent-borne acrylic urethane,
similarr in type to the PPG and Nikolas coatings described above. According to the
manufacturer,, the crosslinking agent (part B) is a trimer of HDMI,
hexadimethyleneisocyanatee (an isocyanurate), and the resin in part A is an acrylic
polyol.. In addition, the manufacturer has stated that the coating contains BTA, is
nott expected to have good adhesion to metal, and can be removed, even after
aging,, by prolonged contact with n-methyl pyrolidone-soaked rags. They also
recommendd that the coating be applied at least 2 mils (51 um) thick [26]. This
acrylicc urethane was applied in Phase II of the study on BTA-pretreated substrates,
andd given a wax topcoat after about one week. Performance on cast bronze after
acceleratedd weathering was superior to Incralac and second to the Nikolas
acrylic/acrylicc urethane. On the copper roof substrate, the coating performed
somewhatt worse, and on the blasted copper roof barely surpassed its competitors,
althoughh none did very well (see Chapter 3).
Investigationn by FTIR first of all revealed that the coating cured much more
slowlyy than expected at room temperature, with or without a wax topcoat. Figure
144 (wide solid line) shows a typical FTIR spectrum of the unweathered, bulk
BASFF 923-85 coating on bronze, with or without BTA pretreatment. Here we see
thee persistence of a small peak near 2259 cm"1 due to unreacted isocyanate groups.
Residuall isocyanate was detected in films of different thickness on rolled bronze,
rolledd copper, and aluminized mirrors.
Thee FTIR spectrum of the unweathered bulk coating on rolled bronze
(Figuree 14, wide solid line) also shows the following features: 1) a somewhat
broadd absorption band near 1527 cm"1, attributable to urethane and urea linkages;
2)) a very strong peak near 1691 cm"1 due to the isocyanate and other species,
coupledd with a rather weak carbonyl peak at 1723 cm'; and 3) a relatively small
doublee OH/NH peak with a maximum near 3383 cm"1. This spectrum is similar to
thatt of PPG DAU75 and shows somewhat limited urethane crosslink formation.
Despitee this evidence of incomplete curing, the coating did pass a solvent
testt for a cured film, i.e., was not marred by rubbing with a bit of the reducer.
Afterr heating at 100 °C for 3 days, the isocyanate peak disappeared. Similarly,
afterr 46 days of accelerated weathering, isocyanate peaks disappeared from the
specularr reflectance spectra of the films applied to aluminized mirrors and copper

124 4
 ChemicalChemical characterization of the bulk and interface

alloys.. This is exemplified in Figure 14b (thin solid line), and was presumably
causedd by heating conditions during the weathering program.

3500 0 3000 0 25000 2000 1500 0 1000 0

VVavenumberr (cra-1)

Figuree 14 FTIR specular reflection spectra of bulk 923-85 acrylic methane (BASF, Inc.)
onon rolled bronze: A) unweathered (wide solid line) and B) weathered (thin solid line).
SpectraSpectra are normalized to aromatic CH stretch bands from 3100-3000 cm

Weatheringg produced other subtle changes in the bulk film, including

relativelyy increased peak intensities at both 1723 and 1691 cm"1, broadening
betweenn 1690 and 1500 cm', and an increase near 1155 cm"1. The OH/NH
stretchingg region appears relatively unchanged, however. These changes indicate
delayedd curing. Thus, the resulting polymer resin appears to have increased
amountss of amines and urea linkages, as may be expected from curing in the dried
film. .
5.3.2.3.2.5.3.2.3.2. The metal/coating interf ace
Examinationn of the BASF metal/coating interfaces showed some striking
differencess in the interfacial region of this acrylic urethane coating on metal.
Comparedd to the bulk film spectrum in Figure 14a, the ATR spectrum of the
unweatheredd film from the rolled bronze interface (Figure 15a) shows a relatively
higherr proportion of isocyanate species at 2259 cm" , as well as increased
absorbancee near 3500 and 3373 cm'. In particular, the ratio of the
isocyanurate/ureaa C=0 peak to ester/urethane C=0 groups appears shifted toward
thee former.
Afterr weathering, as shown in Figure 15b, more changes are evident at the
metall coating interface. Although this spectrum is relatively poor in quality, it is
apparentt that the isocyanate peak has disappeared. Moreover, no new urea or
urethanee functionalities are obvious. Rather, the effect of weathering resides

125 5
ChapterChapter 5

mainlyy in further increase in OH/NH stretching bands centered near 3350 cm"1, and
reductionn in concentration of carbonyls. This not only indicates moisture
sensitivityy at the interface, but also suggests degradation of the polymer.

-.144 -

-.1455 -

35000 3000 2500 2000 1500 1000

Wavenumberr (cm-1)

Figuree 15 ATR spectra (KRS-5 45°) of 923-85 acrylic urethane (BASF, Inc.) from the
rolledrolled bronze/coating interface, A) unweathered, and B) weathered. Spectra are
normalizednormalized to CH bands, 3100-2800 cm'1.

Ass applied in Phase II, this coating thus does not appear to have had a high
densityy of urethane crosslinks. As in the case of the PPG acrylic urethane, low
crosslinkk density appears to be related to segregation, and possibly aggregation, of
thee isocyanate, particularly on copper alloy substrates. Unreacted isocyanates at
thee metal interface apparently do not enhance adhesion to the bare metal, since
adhesionn testing values were generally low on polished bronze for the acrylic
urcthanes.. Results showed some chemical differences exist between the bulk and
metall interfacial region, although much less than in the PPG acrylic urethane. The
Nikolass acrylic urethane appeared more homogeneous, but also undercured as
appliedd in this study. The presence of residual hydroxyl species within the coating
andd at the metal interface near 3520-3300 cm"1 in all three urethanes appears to be
involvedd in increased moisture sensitivity and susceptibility to degradation at the
copperr alloy interface. These results provide rationale for performance that was
lowerr than expected for this system.
Inn summary, the difference between the weathered and unweathered BASF
acrylicc urethane, as well as the Nikolas 9778, generally appears to depend upon
degreee and type of crosslinking more than the weathering itself. Results of
infraredd spectroscopy suggested that crosslinking was incomplete in all coatings,
possiblyy due to a tendency of the crosslinking agent to segregate out of solution.
Inn the PPG and BASF acrylic urethanes, based on evidence of excess isocyanate,
thee ratio of isocyanate:hydroxyl functional groups seems large. This resulted.

126 6
 ChemicalChemical characterization of the bulk and interface

however,, in unreacted residue at the metal surface, and delayed curing effects,
whichh include a disproportionate amount of urea and amine groups. In the Nikolas
acrylicc urethane, no isocyanate residue was detected, corresponding to more
homogeneouss but also somewhat undercured coating. This imparted some
moisturee sensitivity to the coating.
Resultss also imply that application of the solvent-borne, two component
acrylicc urethanes onto patinated metal surfaces, where cure reactions could involve
minerall salts, may be even more complicated, and may result in unreliable
productss in terms of coating performance.

5.3.3.5.3.3. Waterborne acrylic urethanes

Waterbornee coatings are increasingly sought, due to less toxicity and
stringentt VOC requirements in many states in the USA. One-part waterborne
urethaness may either be moisture curing resins or dispersions. The former are
preparedd first by reaction of excess diisocyanates with a hydrogen-donating
functionall polymer to give an isocyanate-terminated product that can be used
subsequentlyy for crosslinking in water. The films then typically cure by the
reactionn of residual diisocyanates with atmospheric moisture to form disubstituted
ureaa and biuret-linked polymers, as shown in the schematic in Figure 10. The
productionn of carbon dioxide gas during curing may be difficult to control,
especiallyy in thicker coatings. Dispersions can be fully reacted, high molecular
weightt urethanes, blends, or copolymer formulations, often with acrylic
components.. Small amounts of crosslinking agents may also be built into
dispersions.. Both types of water-bome coatings typically have substantial amounts
off surfactants, additives, and stabilizers [24,27].

5.3.3.1.5.3.3.1. StanChem waterborne acrylic urethane

Thee commercial coating marketed by StanChem, Inc. as 96X0049 "water-

basedd Incralac," and chosen for study in Phase I, is classified as a waterborne
acrylicc urethane. It was applied as a thick film (about 70 urn), and had severe
pinholing,, a gross defect which could not be overcome despite attempts to the
contrary.. The coating was a poor performer on polished bronze and exhibited
dramaticc failure on the natural brochantite patina of the copper roof. It also
showedd very poor adhesion to both substrates, and was unstable to ultraviolet
radiationn on the metallic substrates, particularly the copper roof, as evidenced by
rapidd yellowing during accelerated exposure. A sample brushed onto glass and
similarlyy exposed did not noticeably yellow. Peeling and corrosion occurred on
bothh types of copper alloy substrates, along with severe cracking on the copper
roof.. (See Chapter 2 for more details.)
Thee exact composition of this coating is not known. Two components were
identifiedd by Py-GC/MS: polystyrene and possibly butyl acrylate. The acrylic
portionn of the polymer therefore seems to have little in common with Incralac.
FTIRR spectra in Figures 16a-b represent, respectively, the unweathered, bulk
coatingg cast on glass (microtransmission), and the weathered glass/polymer

127 7
ChapterChapter 5

interfacee (ATR). ATR spectra of the weathered coating delaminated from bronze
andd copper roof (Phase I, #26) are shown in Figure 16c-d.

35000 3000 2500 2000 1500 1000 50

VVavenumberr (cm-1)

Figuree 16 FTIR spectra of "water-based Incralac" (StanChem, Inc.). A) imweathered

bulkbulk polymer cast on glass (microtransmission), and B-D) A TR spectra of weathered
interfaces:interfaces: B) on glass (KRS-5 45), C) on polished, cast bronze (Phase I. U26) (KRS-5 60).
andand D) on copper roof (KRS-5 60). Spectra are normalized to CH bands 3100-2800 cm'.

Thee imweathered bulk coating (16a) exhibits the general character of an

acrylicc urethane. Major absorption bands are at 3500-2700 (very broad, weak).
3320,, 1731 (strong), 1700 (weak), 1602, 1540 (broad, weak), 1238, and 1169 cm'.
Thesee features indicate a mixture of urethane and urea linkages, as well as the
presencee of carboxylic acids (see Table I). In particular, the position of the CNH
stretch-bendd vibration near 1540 cm"1 indicates that urea structures predominate
[25].. Noting the difficulty with bubbles during application of this clear coating
material,, results confirm that the StanChem coating is a type of moisture-curing
acrylicc urethane.
Notably,, the spectra in Figure 16 all show a peak at 2239 cm"1, in various
concentrations,, due to residual isocyanate. At the bronze/coating interface the
isocyanatee peak is relatively large; in the bulk and at the coating/glass and
coating/brochantitcc interfaces the peaks are small. This provides strong evidence
forr preferential adsorption of the isocyanate groups at the bronze surface, which,
wee can assume, is covered with a thin CuBTA film. It is also seen that the
isocyanatee peak is not shifted from one spectrum to another, indicating that this
functionall group does not complex with the metal or mineral surface, possibly due

128 8
 ChemicalChemical characterization of the bulk and interface

too the separating CuBTA layer. Poor adhesion measured on the sample substrates
supportss this interpretation.
Thee FTIR spectra also show that the residual isocyanate groups are stable
too high temperatures, and have a lower frequency of absorption than isocyanates in
thee solvent-borne acrylic urethanes. This tentatively suggests a resonance-
stabilizedd aromatic polyisocyanate structure [7]. The poor light stability observed
inn the coating would support this interpretation. In any case, the unchanged
appearancee of the isocyanate peak after weathering of the film on glass suggests
thatt this species is present as a functional group on high molecular weight
polymerss and inhibited towards further crosslinking reactions in the dried film,
perhapss due to steric hindrance and lack of mobility. It is unclear, however, how
thee isocyanate groups are evidently stabilized toward reaction with atmospheric
moisturee to produce amines.
ATRR spectra of the weathered coatings (Figure 16c-d) show significantly
greaterr change occurred in the coating's chemical structure at the metal interfacial
regionss than at the glass interface. This picture is somewhat complicated by the
presencee of a particularly wide range of products, indicated in multiple peaks
betweenn 1800 and 1200 cm"'. Inspection of the carbonyl region of the interfacial
regionn from the coating aged against bronze (Figure 16c) shows sharpening of the
mainn carbonyl peak at 1727 cm"1 accompanied by marked broadening and increase
inn absorbance peaks at its base, from about 1750-1500 cm"1. New peaks are
centeredd near 1615 and 1550 cm"1. This suggests formation of ureas, as well as
carboxylatee and amine salts. The latter is supported by weak absorbance near 2500
cm"" .
Inn the coating weathered on copper roof (Figure 16d), the region between
17000 and 1500 cm appears even more changed. It is now a broad hump of
overlappingg peaks, suggesting the presence of complex mixture of degradation
products,, including acids and amine salts and/or carboxyl salts, in lieu of peaks
arisingg from polymer crosslinks.
Readingg up in Figure 16, we also see markedly higher concentrations of
OH/NHH stretch vibrations (near 3377 and 3448 cm"1, respectively) after
weathering,, depending on the substrate. Absorbance due primarily to NH
functionalitiess is increased at the coating/glass interface after weathering. On
bronzee or copper roof, the concentration of both OH and NH stretching vibrations
appearss even more significant after weathering, although this region overlaps with
absorptionn peaks from minerals adhered to the delaminated coating in the latter
spectrumm (Figure 16d).
Degradationn and susceptibility to moisture-related changes thus appear
inherentt in the coating, even without the complication of pinholing. Furthermore,
resultss show that coating instability is exaggerated by contact with the bronze and
copperr mineral salts. This is particularly evident on the brochantite patina surface,
wheree polymer degradation appears to predominate over curing-type reactions
duringg accelerated weathering. These results provide strong chemical rationale for
thee coating's poor performance on these substrates in Phase I.

129 9
ChapterChapter 5

5.3.3.2.5.3.3.2. Nikolas waterborne acrylic/urethane

Thee 11650 Eco-borne Brass Lacquer (G.J. Nikolas & Co.) is an acrylic
emulsionn with a urethane dispersion, primarily made for indoor metal maintenance
applicationss [28]. Py-GC/MS analyses showed that the main components of the
acrylicc resin are styrene, butyl methacrylate, and an alcohol. A large amount of
BTAA was detected in the coating from HPLC analysis of methanol extract of a
driedd film; this could not be quantified in the present method. Results are shown in
Figuree 8, where a 10 |al injection of the methanol extract (vs. 5 ul injections of the
otherr resins) produced a very large peak compared to solvent peaks.

35000 3000 2500 2000 1500 1000

Wavenumberr (cm-1)

Figuree 17 FTIR spectra of Nikolas 11650 waterborne acrylic/urethane: A) un weathered

bulkbulk coating on aluminized mirror, specular reflectance, B) unweathered bulk coating on
rolledrolled bronze, specular reflection, C) unweathered coating interface from rolled bronze.
ATRATR (KRS-5 45°), D) weathered coating interface from rolled bronze, ATR (KRS-5 45°),
andand E) weathered coating interface from polished, cast bronze (Phase I, #27), ATR (KRS-5
6060 °). Spectra are normalized to CH stretch bands near 2920 cm'.

Ass shown in Figure 17, FTIR spectra of the bulk coating and metal/coating
interfacess are marked on the whole by their uniformity, and a lack of residual
isocyanates,, as expected from a dispersion. This confirms that the coating
undergoess coalescence during drying, rather than moisture curing, and largely
accountss for homogeneity between the bulk and interfaces, as indicated by FTIR.
Thee spectra also show relatively large proportions of NH/OH vibrations near 3338
cm"11 and urea/urethane vibrations near 1680, 1653, and 1539 cm"1. Bands in this
absorptionn region, as well as a weak, broad absorbance hump underneath the NH
andd CH stretching bands, may also indicate the presence of carboxylic acids and/or
aminee salts.

130 0
 ChemicalChemical characterization of the bulk and interface

Infraredd analysis after weathering (Figure 17a,b) does not show changes in
thee bulk films, suggesting that the coating has surprisingly good chemical stability.
ATRR spectra of the coating from the rolled bronze interface after accelerated
weatheringg appear unchanged as well (Figure 17c,d). This was also true for ATR
spectraa of the coating delaminated from the polymer/polished bronze interface
fromm Phase 1 (#27) (Figure 17e), except for broadening in the OH/NH stretch
regionn from about 3600 to 2500 wavenumbers. This coating was thinly applied in
aa single coat in Phase I, and in two layers on the rolled bronze in Phase II, which
mayy account for the increase in hydrogen bonding detected in Figure 17e.
However,, even the unweathered coatings on rolled bronze show a hump in this
region,, which is relatively larger at the metal interface. These results confirm that
thee coating allows passage of moisture through the film, and suggest that
carboxylicc acids, and probably carboxyl salts, form at the metal surface. This
moisturee sensitivity appears to be the coating's Achilles' heel and largely accounts
forr its fairly poor performance (see Chapters 2 and 3).

5.3.4.5.3.4. Microcrystalline wax blend

Thee microcrystalline wax blend coating is a mixture of about 75% Bareco
Victoryy Wax (microcrystalline, synthetic, low melting point wax), Bareco
Polywaxess 2000 and 500 (polyethylene microcrystalline waxes), and Petronauba
C,, an oxidized polyethylene wax. No BTA is added to the formulation [29]. This
formulationn is consistent with specular reflectance spectra of the wax on
aluminizedd mirrors or rolled bronze (Figure 18a), which exhibit only the prominent
mcthyl/methylenee vibrations, plus the small doublet near 750 cm"1 attributed to
crystall splitting from the microcrystalline wax. No change in the wax after 46 days
acceleratedd weathering was indicated by specular reflectance, although little
remainedd on the mirror after this amount of time.

35000 3000 2500 2000 1500 1000

Wavenumber(cm-I) )

Figuree 18 FTIR specular reflection spectra of microcrystalline wax blend on rolled

bronze.bronze. A) unweathered and B) weathered for 46 days. Spectra are normalized to CH
stretchstretch bands near 2916 cm'.

131 1
 ChapterChapter 5

Afterr weathering the wax coating on rolled bronze for 46 days (Figure
18b),, the spectrum of the wax is similarly unchanged, and indicates the coating is
stilll present. However new peaks attributed to the formation of corrosion beneath
thee wax are clearly visible. Peaks at 3446 and 3337 cm~\ as well as between 800
andd 900 cm" , indicate that the corrosion is a hydroxy mineral, most likely copper
hydroxyy chloride. Results thus confirm the poor protection provided by the wax to
bronzee exposed to simulated acid rain weathering (see Chapters 2 and 3).

5.4.5.4. Conclusions
ATRR and RAIR were used to investigate interfacial regions in four types of
coatingg systems on copper or bronze, before and after accelerated weathering:
acrylics,, solvent-borne, two-component acrylic urethanes, waterborne urethanes,
andd a wax coating. Chemical information about the polymer interfaces against
glass,, rolled copper, patinated copper, rolled bronze, and/or cast bronze, gathered
byy these methods was compared to FTIR microtransmission spectra, specular
reflectance,, Py-GC/MS, and HPLC of the bulk polymers.
Chemicall characterization of coating/metal interfaces showed that
significantt differences exist between the metal/coating interface and the bulk in
somee coatings. This tendency was seen before weathering mostly in coatings that
curee in-situ vs. lacquers or dispersions. After weathering, differences between
metal/coatingg interfaces and the bulk in all of the coatings appeared to different
degrees,, and generally indicated a catalytic effect of copper and copper salts on
degradationn reactions in the presence of moisture and/or air. The presence of a
CuBTAA corrosion inhibition film at the metal interface apparently does not
completelyy inhibit either this effect or corrosion of the metal. Whether degradation
reactionss are impeded to some degree was not rigorously shown by the data, but is
nott supported in the performance ratings. While CuBTA films may inhibit passage
off moisture to the metal surface for a limited time, they are not good barriers for
copperr ions themselves, which can easily migrate out once an electrolytic solution
formss in the interfacial region. In the absence of a reservoir of BTA in this
solution,, these ions are free to form corrosion salts underneath the coating.
Inn acrylics, the chemical character of the bulk film and the metal/coating
interfacee on copper alloys provides rationale for their fairly good performance in
termss of good chemical homogeneity, stability and barrier properties. However,
resultss also suggest that greater susceptibility to degradation exists at the
metal/coatingg interface, especially when film thickness is not adequate enough to
providee an effective diffusion gradient to oxygen and moisture. This supports
evidencee of loss of adhesion during weathering (see Chapter 2). In addition, the
formationn of small amounts of gel and the onset of embrittlement may occur in the
coatingg before gross chemical changes are detected in the bulk. Microcracking is
likelyy to result from embrittlement and stress build-up, and lead to a dramatic
changee in moisture ingress. Corrosion and degradation at the interface may
proceedd rapidly after this type of induction period, and may be catalyzed by copper
salts.. In this light, the addition of a wax topcoat may benefit these systems not

132 2
 ChemicalChemical characterization of the hulk and interface

onlyy by increasing thickness and impermeability to moisture, but also by initially

fillingfilling microcracks as they develop. Investigation of Paraloid B-48-based acrylic
coatingss suggested that they have slightly reduced oxidative stability compared to
Incralac,, which is based on Paraloid B-44 resin. This is supported by performance
ratingss in this study.
Chemicall investigation of the bulk and interfacial regions of several acrylic
urethaness revealed curing problems and suggested that coating properties and
performancee are probably contingent on proper curing. FTIR analysis showed that
thiss was not easy to achieve: evidence of unreacted functionalities remaining in the
coatingss was rife. In general, differences between the acrylic urethane coatings
appearedd to hinge on two factors: 1) homogeneity between the bulk and
metal/coatingg interface, and 2) the degree and type of crosslinking. These
chemicall characteristics are related to the existence of residual crosslinking agents.
FTIRR analysis of the bulk coatings and the metal/coating interfaces showed in
severall cases that excess isocyanates tend to congregate and persist at the metal
surface.. This not only causes chemically distinct regions to form in the coating,
butt may also result in delayed curing and the formation of less stable cure
products. .
Onn the other hand, undercuring was also observed to occur in the coatings.
Thiss was evidenced in FTIR spectra by the presence of unreacted hydroxyl
functionall groups, and in the relative proportions of bands assigned to crosslink
functionall groups. Resulting moisture sensitivity in these coatings was apparent at
thee metal/coating interfaces of weathered samples, which tended to show increases
inn hydrogen-bonding and suggested conversion of some species to amine and
carboxyll salts. Results also suggested that these reactions were accelerated in
proximityy to the copper alloy surfaces. While good homogeneity in a coating
withoutt excess isocyanate, such as in the Nikolas 9778, appears to aid in good
performance,, a less highly crosslinked resin may be correlated to less durability.
Thiss undoubtedly impinges on the system's lifetime prediction, and appears to be
reflectedd in performance ratings which did not meet the very high expectations
heldd for acrylic urethane coatings in general. While the performance of solvent-
bornee acrylic urethanes remained fairly high in this study, results of chemical and
FTIRR analysis suggest that better and more consistent performance may be
achievedd with more optimal curing.
Thee one-part waterborne systems do not offer many avenues for better
controll of curing, and therefore performance is unlikely to be improved through
applicationn development in these cases. This is clearly the case with the StanChem
waterbornee acrylic urethane. Spectroscopic analysis of this coating after
weatheringg showed inherent chemical instability, which was accelerated on the
metall and mineral substrates. This provided an explanation for poor performance
inn Phase I of the study. The Nikolas waterborne acrylic urethane is a type of
dispersionn coating. FTIR spectra indicated that the bulk coating was quite stable
onn polished bronze during accelerated weathering. However, infrared analysis of
thee interfacial regions provided evidence that the coating allows moisture to
accumulatee near the interface, resulting in degradative reactions.

133 3
ChapterChapter 5

Infraredd analysis of the wax coating on rolled bronze similarly showed

goodd chemical stability, but an ineffective barrier to moisture and air. In this case,
FTIRR provided direct evidence of the early occurrence of corrosion underneath the
coatingg during accelerated weathering. This failure far preceded similar evidence
inn any other coating studied.
Thee chemical characterization of bulk and interfacial regions in coatings on
modell samples described in this chapter, in the context of their observed behavior
inn accelerated and outdoor weathering tests, gives a valuable picture of processes
thatt occur in the coatings. These small puzzle pieces help to construct a common
picturee of coating failure for individual coatings as well as for groups of coatings.
Thee results provide rationale for the performance observed on model samples after
acceleratedd weathering in Phases I and II, and furthermore provide evidence for
changess at the polymer/metal interfaces which may predict performance problems
nott yet seen in our weathering program.

134 4
 ChemicalChemical characterization of the bulk and interface

References References

11 W. J. van Ooij, et al., "Application of Surface Analysis Techniques to the

Studyy of Paint/Metal Interfaces Related to Adhesion and Corrosion
Performance,"" Surface and Interface Analysis 17 (1991), 403-420.
22 R. A. Dickie, "Paint adhesion, corrosion protection, and interfacial chemistry,"
ProgressProgress in Organic Coatings 25 (1994), 3-22.
33 W. Brockmann, et al., Int. J. Adhesion and Adhesives 6 (1986), 127.
44 M. G. Chan and D. L. Allara, "Infrared Reflection Studies of Metal-Polymer
Interfaces,"" Polymer Engineering and Science 14, 1 (Jan. 1974).
55 Bruno M. Perfetti, Metal Surface Characteristics Affecting Organic Coatings
(Federationn Series of Coatings Technology, Blue Bell, PA, August 1994), 63.
66 E. René de la Rie, Suzanne Quinlen Lomax, Michael Palmer, Lisha Deming
Glinsmann and Christopher A. Maines, "An investigation of the photochemical
stabilityy of urea-aldehyde resin retouching paints: removability tests and
colourr spectroscopy," Tradition and Innovation: Advances in Conservation,
Preprints,, IIC 18th International Congress, 10-14 Oct. 2000, Melbourne,
Australiaa (IIC, London, 2000), 51-59.
77 N. B. Colthup, L. H. Daly and S. E. Wiberley, Introduction to Infrared and
RamanRaman Spectroscopy (Academic Press, Inc., Boston,, MA, 1990).
88 Formulation information provided by StanChem, Inc.
99 O. Chiantore and M. Lazzari, "Characterization of Acrylic Resins," Int. J.
PolymerPolymer Analysis & Characterization 2 (1996), 395-408.
100 Unpublished notes from non-ferrous metal discussion group, Maintenance of
OutdoorOutdoor Sculpture: Whose Job is it?, Symposium at the 20th Annual Meeting
off the AIC, 2-3 June 1992, Buffalo, NY.
111 John E. Fagel, Jr. and Galen W. Ewing, "Ultraviolet absorption of
benzotriazole,"" J. Am. Chem. Soc. 73 (1951), 4360-2.
122 The Infrared Spectroscopy Committee of the Chicago Society for Coatings
Technology,, An Infrared Spectroscopy Atlas for the Coatings Industry
(Federationn of Societies for Coatings Technology, Philadelphia, PA, 1980),
116. .
133 Anthony Davis and David Sims, Weathering of Polymers (London, 1983).
144 D. L. Allara, C. W. White, and R. L. Meek, "Mechanism of Oxidation at a
Copper-Polyethylenee Matrix. II. Penetration of Copper Ions in the
Polyethylenee Matrix,"/ Poly. Sci. 14 (1976), 101 ff.
155 Norman Grassie and Gerald Scott, Polymer Degradation and Stabilization
(Neww York, NY, 1985), 89.
166 M. Lazzari and O. Chiantore, "Thermal-ageing of paraloid acrylic protective
polymers,"" Polymer 41 (2000), 6447-6455.

135 5
ChapterChapter 5

177 D. Erhardt, W. Hopwood, and T. Padfield, "The Durability of Incralac:

Examinationn of a Ten Year Old Treatment," Preprints to the 7th Triennial
MeetingMeeting of the ICOM Conservation Committee, 1984, Copenhagen, Denmark,
22.1-22.3! !
188 Unpublished notes, roundtable discussion, National Gallery of Art,
Washington,, DC, December 1994.
199 Robert L. Feller, Mary Curran and Catherine Bailie, "Photochemical Studies
off Methacrylate Coatings for the Conservation of Museum Objects," ACS
SymposiumSymposium Series Number 151, ed. S.P. Pappas and F.H. Winslow (American
Chemicall Society, Washington, DC, 1980), 183-196.
200 O. Chiantore, L. Trossarelli, and M. Lazzari, "Photooxidative degradation of
acrylicc andmethacrylic polymers," Polymer 41 (2000), 1657-1668.
211 O. Chiantore, M. Lazzari, "Photo-oxidative stability of paraloid acrylic
protectivee polymers," Polymer 42 (2001), 17-27.
222 Rohm and Haas technical literature.
233 Technical Data Bulletin, 11565 OD Lacquer RFU, G.J. Nikolas & Co., Inc.,
5/24/96. .
244 Clive H. Hare, Protective Coatings: Fundamentals of Chemistry and
CompositionComposition (Technology Publishing Co., Pittsburgh, PA, 1994).
255 M.W. Urban, et at., "Interfacial Studies of Crosslinked Urethanes: Part III.
Structure-Propertyy Relationships in Polyester Waterborne Polyurethanes," J.
CoatingsCoatings Technology 7'1 (Jan. 1999), 75-85.
266 Private Communication, Mr. Mike Cameron, technical support staff scientist,
BASF,, Inc., 1997.
277 David R. Bauer, "Chemical Criteria for Durable Automotive Topcoats,"
JournalJournal of Coating Technology, 66, 835 (August 1994), 252.
288 Private communication, G.J. Nikolas & Co.
299 Information provided by the recipe's proprietor and supplier, Mr. Nicolas
Veloz,, private conservator, Washington, DC.

136 6
 66
Thee role of benzotriazole (BTA)
inn bronze protection

Abstract Abstract
Inn a first set of experiments, copper-benzotriazole (CuBTA) complexes
weree made under varying conditions from solutions of benzotriazole (BTA) with
copperr mineral powders commonly associated with bronze corrosion, as well as by
immersionn of simulated copper corrosion surfaces in BTA solutions. Complexes
weree characterized as polymeric cuprous (Cu(I)) and/or cupric (Cu(II)) BTA
complexess in all systems investigated. It was found that the CuBTA complexes
varyy in terms of bonding, composition, physical structure, thickness and
uniformity,, depending upon the reactants and the conditions of reaction. In a
separatee series of experiments, CuBTA film growth on rolled bronze was
investigatedd by ellipsometry and external reflection, also known as reflection-
absorptionn infrared spectroscopy. Results indicated that CuBTA film growth in
near-neutrall conditions is dependent on time, BTA concentration, and solvent.
Evidencee related to film thickness also suggested that Cu(I)BTA films develop in a
step-wisee manner. Some properties of the Cu(I)BTA films, including heat
stability,, oxidative stability, and ability to form on unclean surfaces or under the
coatingg Incralac, were also investigated. These results are discussed within the
contextt of BTA as a pretreatment for coated outdoor bronzes.

6.1.6.1. Introduction
lH-benzotriazolee (BTAH or, generally, BTA) has been in use as a
corrosionn inhibitor for the conservation treatment of bronzes since the late 1960s
[1],, Despite the search for improved inhibitors, BTA remains one of the most
effectivee and acceptable anti-corrosion treatments available for copper alloys today
[2].. Although there has been concern to the contrary, recent articles have asserted
thatt BTA is safe for use in conservation [3], and conservators in the fields of
archaeologicall and outdoor sculpture conservation continue to rely on BTA for
treatmentt of archaeological and outdoor copper alloys and bronze art and
ornamentation.. However, the uneven success of this agent in treatments for bronze
diseasee and bronze corrosion in outdoor exposures remains an important subject of

137 7
ChapterChapter 6

scrutinyy and scientific research. In particular, a scientific basis for the choice of
treatmentt parameters has been lacking.
Ass a result of this discrepancy in research and application, there is a gap in
knowledgee between the theoretical usefulness of BTA as a corrosion inhibitor for
bronzee in general, and the observed inconsistency of its success. The first set of
experimentss presented in this chapter investigates the interaction of BTA with
typicall copper oxide and chloride corrosion products associated with bronzes, as
welll as with model copper corrosion surfaces. Results shed light on several factors
whichh influence the corrosion protection properties of CuBTA films as formed on
heterogeneous,, corroded copper alloy surfaces during typical treatment conditions.
Inn the context of coatings for outdoor bronzes, the role of CuBTA films
underr thicker protective coatings was also probed. As outlined in Chapters 2 and
3,, samples of cast bronze and 50-year-old, naturally corroded copper samples were
givenn various surface preparations, including brush-coating with 1.5%
BTA/ethanoll before applying the protective coatings; the samples were then
exposedd to either accelerated indoor or natural outdoor weathering. As discussed,
visuall inspection and x-ray diffraction (XRD) measurements of corrosion on the
uncoatedd substrates treated with BTA showed that this treatment by itself has little
inhibitingg effect on the formation of chlorides and black corrosion in accelerated,
simulatedd outdoor weathering and natural outdoor weathering. Furthermore, BTA
treatmentt prior to the application of a coating did not show any clear performance
benefitt in these testing regimes. These observations are supported by reports in the
literaturee that air-dried Cu(I)BTA films have vastly decreased corrosion protection
inn NaCl solutions compared to films that form and remain in a NaCl/BTA solution
[4,5],, It is also known that CuBTA protection in sulfate solutions is superior to
thatt in chloride solutions [6]. It is thus unclear whether it is possible to optimize
BTAA pretreatment of outdoor bronzes and, in so doing, unambiguously boost the
protectionn of a coating system.
Inn order to gain a better understanding of these observations, a further set of
experimentss was designed in which CuBTA film formation on rolled bronze was
studiedd by external reflection, also known as reflection-absorption infrared
spectroscopyy (RAIR). RAIR was used to investigate the thickness of CuBTA films
formedd on rolled bronze under various conditions of BTA concentration and
solvent,, as well as to gather chemical information about the films themselves. The
formationn of a CuBTA film from contact with an organic coating such as Incralac,
whichh contains BTA, was also investigated. Chemical aspects of the coatings were
discussedd in Chapter 5. Ellipsometry and electrochemical impedance spectroscopy
(EIS)) were also utilized to confirm trends observed by RAIR.

138 8
 TheThe role ofbenzotriazole

6.2.6.2. Background
Itt is well established that BTA complexes with copper, copper alloys, and
cuprouss oxide surfaces to form a cuprous film (Cu(l)BTA)*. The complex is
usuallyy identified as polymeric [Cu(I)BTA] (Figure 1) [7,8,9,10,11,12,13,14,15].
Inn this structure, BTA acts as a unidentate bridging ligand in a linear chain of
alternatingg Cu'~ and resonance-stabilized BTA molecules. [Cu(I)BTA] is reported
too be amorphous, insoluble, and to be stable up to about 200 °C, depending on the
atmospheree [7,9,16]. A related cuprous film complex, consisting of the polymeric
linearr chain [Cu2BTACl], has been proposed for complexes prepared with high
concentrationss of chlorides [17].

HH

A)) 1-Hbenzolriazole B) benzolriazole anion

C)) Cu(I)BTA

Figuree 1 Proposed structures for A) l-II benzotriazole (BTA); B) BTA anion; and C)
Cu(I)BTA. Cu(I)BTA.

BTAA may also form cupric complexes (Cu(II)BTA)*, the structures of

whichh are less well defined, partly because they have received less attention.
Cottonn [7] proposed the polymeric network structure of [Cu(II)(BTA)2] as shown
inn Figure 2. In this structure, the Cu2" ions act as crosslinking sites between BTA

*Thee complexes are referred to without specific compositions unless otherwise noted.

139 9
 ChapterChapter 6

ligandss in a square planar geometry. The inordinately long Cu-N bonds drawn in
Cotton'ss structure suggest a fundamentally different, weaker type of metal-ligand
interactionn than that described for [Cu(I)BTA]. Based on evidence of non-ideal
stoichiometryy in some cases. Cotton proposed the existence of another, undefined
cupricc BTA derivative whose composition is influenced by preparation in the
presencee of chlorides. Roberts [18] also proposed an alternative structure of
Cu(II)BTAA with distorted octahedral geometry for the cupric complexes, with the
inclusionn of oxide and/or water ligands. In addition, Caramazza et al. [19] reported
aa non-polymeric, cupric chloride-BTA derivative with the composition of
Cu2(BTA)2BTAHCl2,, although it is unclear what the structure of such a complex
wouldd be. Cupric complexes have also been reported to be insoluble, to have
slightlyy lower thermal stability than [Cu(I)BTA], and to be crystalline when
preparedd with excess chlorides [7].

Figuree 2 Proposed structure for Cu(II)BTA (Cotton [7]).

Itt is not known whether Cu(II)BTA derivatives inherently impart corrosion

protectionn equal to that of thin, highly polymerized films of [Cu(I)BTA] formed on
cleann copper surfaces. However, the effectiveness of any film in corrosion
protectionn depends largely on a combination of its adhesive and film properties.
Thee corrosion control which BTA treatments have often been observed to impart to
archaeologicall bronze objects may lie to some extent in the excellent coverage and

140 0
 TheThe role of benzotriazole

adhesionn of CuBTA films, which form primary chemical bonds to underlying

copperr mineral surfaces.
Otherr important protective properties of CuBTA films are largely
determinedd by the chemical and physical structure of CuBTA films, as well as their
growthh mechanism. The majority of investigations into the inhibition effect of
BTAA on copper and copper alloys advance the theory that CuBTA films protect the
metall from corrosion by means of a highly impermeable physical barrier layer to
chloridee and sulfate ions, rather than by chemical or electrochemical means. Good
barrierr protection in films generally arises from high ionic resistance, crosslink
density,, and degree of polymerization. On the other hand, permeability should be
low;; the effect of crystallinity on protective properties may be variable. Film
uniformityy itself is also an important property in order to avoid the creation or
aggravationn of electrochemical potential differences or osmotic pressure gradients,
whichh are driving forces for corrosion [20].
Theree is overwhelming evidence in the literature that the method of
preparation,, including condition and oxidation state of the reacting surface, pH,
temperature,, and chloride and oxygen concentration, strongly influence Cu-BTA
reactions.. These variables have been implicated in reported differences in key film
properties,, such as thickness, oxygen permeability, degree of polymerization,
crystallinity,, resistance, and thermal stability. For example, film formation under
non-corrosivee conditions has been found to be self-limiting, leveling off as the
thicknesss effectively reinforces the underlying oxide and blocks the metal surface.
Underr acidic conditions, i.e., pH < 4.0, the underlying oxide or patina tends to
dissolvee and be replaced by thick CuBTA films, reported to grow up to 5000 A. In
thiss case, the CuBTA film formation is accompanied by simultaneous corrosion
beneathh the CuBTA film, so that the combined processes result in nearly unlimited
growth.. However, these CuBTA films are known to be much less protective
[4,9,14,20,21,22]. .
Inn this light, CuBTA inhibition should be highly dependent on thickness, as
welll as properties such as porosity, uniformity and degree of polymerization.
Investigationss into the dependence of CuBTA thickness on BTA concentration and
exposuree time are not unanimous and have typically examined copper maintained
inn BTA solutions at concentrations much lower than are used in conservation
applicationss [9,23]. Optimization of these film properties for conservation
applications,, as well as an understanding of corresponding chemical and physical
structure,, are thus of key importance in the field and have remained a subject of
speculationn and trial and error.
Thee purpose of the experiments described in this chapter is to gain an
understandingg of the structure of CuBTA films produced by typical BTA
treatmentss of bronze, as well as the effect of some treatment parameters on film
properties.. This research focused primarily on the parameters of BTA
concentration,, time of immersion, pH and solvent. Evidence presented in this
paperr concurs with the literature in that different methods of CuBTA film
preparation,, such as found in typical conservation treatments, may have significant
influencee on the CuBTA films, especially in terms of thickness and uniformity.

141 1
ChapterChapter 6

6.3.6.3. Experimental methods

6.3.1.6.3.1. BTA-copper mineral reaction products

Modell systems that mimic natural bronze reaction products were made for
study.. CuBTA complexes were formed by reacting solutions of 1-H BTA
(Aldrich,, 99.99%) and Cu 2 0 (cuprite), CuO (tenorite), CuCl (nantokite), CuCl2,
Cu2(OHhCll (paratacamite), Cu2(OH)2C03 (malachite), or copper powders, under
varyingg conditions of solvent, pH, and molar ratio of BTA to Cu. Commercially
availablee mineral powders were reagent grade. Paratacamite was synthesized
accordingg to the method of Tennent and Antonio [24] and its identity was
confirmedd by x-ray diffraction (XRD) and Fourier transform infrared spectroscopy
(FTIR).. The additional presence of lower hydroxy chlorides and/or hydroxides of
copperr was suggested by non-ideal stoichiometry found in x-ray photoelectron
spectroscopyy (XPS) elemental analysis. The XPS O(ls) spectrum of the
paratacamitee also indicated that a minor amount of cuprous oxide was present.
Alll CuBTA reaction products were gravity filtered, rinsed repeatedly with
reagentt alcohol to remove excess BTA, and dried at room temperature under
vacuum.. Solvents were double distilled deionized water, HPLC grade reagent
alcohol,, and acetone. Aqueous solutions below pH 5.0 were adjusted by the
additionn of HC1, while basic solutions above pH 7 were adjusted by the addition of
NaOH.. Reaction products were analyzed by FTIR as ground powders in oven-
driedd KBr pellets with a Digilab FTS-40 spectrometer equipped with a DTGS
detector,, and by XRD on a Philips 1710 open-architecture diffractometer, and, in
onee case, a Philips 1840 powder camera diffractometer. The x-ray sources were
Cuu Ka, and spectra were acquired at instrumental settings of 30 mA, 40 kV, and
0.022 °/sec. Peaks were recorded by Sietronic Software and transferred to the Fein-
Marquartt software program for analysis. Several of these reaction products were
analyzedd by XPS on a V.G. ESCA III Mark II spectrometer, using Mg Ka
radiationn with a photon energy of 1253.6 eV (400 W), ambient temperature,
chamberr pressure below 10"8 Torr, and a take-off angle of 45°.
CuBTAA films were formed on simulated copper corrosion surfaces by
immersionn in 3% (wt/vol) solutions of BTA in alcohol or water, followed by
rinsing.. Three types of substrates were used: 1) mechanically polished, 2 mm thick
copperr coupons (Aldrich, 99.999%), on which selective corrosion films were
formedd either in air (cuprous oxide), electrochemically, or by adsorption of mineral
solidss on the surface; 2) pellets of compressed mineral powders; and 3) rolled,
oxidizedd copper sheet coupons which had been used for paratacamite synthesis and
hadd corrosion patinas formed during this process. CuBTA films were analyzed
mainlyy by FTIR microreflection spectroscopy, using a UMA-300 FTIR microscope
attachmentt equipped with an MCT detector, and by XRD. All FTIR spectra were
recordedd in absorbance. Three mineral pellets immersed 17 hours in BTA/alcohol
weree examined by XPS using a Perkin-Elmer 5300 XPS spectrometer, using
MgKaa radiation, with a photon energy of 1253.6 eV (400 W), ambient
temperature,, chamber pressure below 10"* Torr, and a take-off angle of 45°.

142 2
 TheThe role of benzotriazole

6.3.2.6.3.2. CuBTA films on bronze

RAIRR of thin Cu(I)BTA films formed on rolled bronze by immersion in a)
1.55 wt.% BTA/ethanol, b) 3.0 wt.% BTA/ethanol, or c) 1.5 wt.% BTA/water was
performedd with a Bio-Rad Digilab FTS-60A Fourier transform infrared
spectrometerr fitted with a Harrick versatile reflection attachment for center-
focusedd beam and a retro-mirror accessory, plus a wire-grid polarizer set for
parallell beam polarization. The angle of incidence in all cases was 78 degrees.
Sampless were purged with dry air for 10-20 minutes to minimize the presence of
atmosphericc moisture. The final reflection spectra were obtained by digitally
subtractingg the cleaned bronze spectrum, placing each sample with the same
geometryy in the accessory before and after treatment. In some cases of very long
immersionn in BTA solutions, sample spectra were produced by subtracting that of
aa new, freshly scanned, untreated rolled bronze coupon. Absorbance at 745 cm
wass calculated with the WIN-IR software from the maximum height of this peak to
aa baseline drawn between 835 to 729 cm" .
Solutionss of BTAH (Aldrich, 99%) were made with HPLC-grade ethanol or
millipore,, distilled and deionized water. The pH of aqueous BTAH solutions was
5.30.. The rolled bronze (Lubaloy Co.) was spring-tempered, 425 bronze, 0.016
gauge,, with an alloy composition of 88.5% Cu, 9.5% Zn, 2% Tin. Rolled bronze
sampless were polished with a series of Micro-mesh cloths from either 2400 or
6000,, to 12000 mesh. The polished samples were solvent cleaned by wiping and
rinsingrinsing with alternating polar and non-polar solvents until they passed the water-
breakk test [25], immersed into the appropriate BTA solution, rinsed thoroughly
withh ethanol, and air dried.
Ellipsometryy was performed with a Rudolph Research Inc. thin film
ellipsometer,, type 43603-200E at the University of Cincinnati. Film thickness was
calculatedd from measurements of the delta and psi parameters using a film
refractivee index of 1.6 [9,26]. An uncoated bronze coupon served as the
background,, so that the thickness of the natural cuprous oxide layer was subtracted
fromm CuBTA thickness measurements.
Electrochemicall impedance spectroscopy (EIS) was conducted at North
Dakotaa State University on a set of polished rolled bronze samples that had been
immersedd in 3% BTA/ethanol solutions for 4 minutes, 1046 minutes, or 10,047
minutes.. EIS measurements were taken with a Gamry Instruments PC-3
Potentiostat,, using a removable 1.5 inch (outer diameter) glass tube clamped to the
bronzee coupon by an o-ring. The glass tube was filled with the dilute Harrison's
solutionn (0.35 wt. % (NH^SC^, and 0.05 wt. % NaCl in H 2 0). A saturated
calomell reference electrode and a platinum counter electrode were also immersed
inn the solution. The amplitude applied to the system was 5 mV, and the frequency
rangee was from 0.1 Hz to 10,000 Hz. Impedance modulus is reported as an
averagee of three spots on each sample.

143 3
 ChapterChapter 6

TABLEE I: Frequencies and Tentative Assignments of

Infraredd Absorption Bands in Benzotriazole and CuBTA Complexes

Benzoo triazole' (cm'1) Cu(I)BTA3(cm-') ) CuaiJBTAfcm') ) Tentativee Assignment

3346/32599 (broad) 3420-34355 (broad) HOHH stretching
(adsorbedd water;
waterr of
crystallization) )
3500-32000 (m) NN HOH; NH OH2
stretching g
3146 6 freefree N-H stretching
3076/30366 (w) 3076/3059/3040 0 3057-3097(w) ) aryll C-H stretching
(w) )
2991/2958/2908; ; NHH N stretching
2850-24000 (m) (intermolecular) )
16400 (w) liquidd H 2 0
1621 1 1612(vw) ) 1612-16199 (m) aryll ring stretching
15955 (w) 15800 (w) 1577(m) )
15066 (w)
14600 (m) 1490-14922 (m) 1490-14922 (w-m) aryll + triazo (?)
stretching g
14199 (m) 1445(m) ) 1445(m) ) aryll + triazo (?)
stretching g
13822 <m) 1388-13900 (m) 1385/13955 (vw) aryll + triazo (?)
stretching g
13055 (w) 12999 (w) 12988 (w)
1265 5 12700 (w) 1271-12755 (m)
12077 (s) 1211-12233 (m) triazoo ring stretching
11955 (w-m)
1149-11522 (m-s) (1150-1152)(w) ) triazoo ring stretching
1143(w) ) 1132(vw) )
11199 (w-m)
10755 (w) aryll ring mode (?)
1046 6 1028-10300 (w)
1006 6 N-HH in-plane bending
(?) )
9822 (m) 9911 (w-m) 9966 (w)
8766 (w-m)
7766 (m) 787-7899 (m) 788-7899 (m) aryll out-of-plane 4
743-7455 (s) 7411 (s) 7455 (s) adjacentt C-H wag
7077 (m) N-HH wagging (?)
6822 (w)
641-6433 (w-m)
607 7 triazoo ring torsion
561-5644 (w-m)
4355 (w-m)

'basedd on assignments for benzimidazole by Colthup [19] and Cordes and Walter [20],
2
basedd on assignments by Poling [4] and Suetaka and Morito [7].

144 4
W&$HÈ W&$HÈ