2015 Emmersonnjphd

Heritage wrought iron: towards the
development of evidence based

standards for coating
Nicola J. Emmerson
Thesis submitted to Cardiff University in candidature

for the degree of PhD
September 2015
i
Summary
Effective management of heritage assets relies on decision-making which is
underpinned by empirical evidence of impact of treatments on long term survival
prospects of materials. Historic wrought iron presents a particular problem for
conservation. It occupies a niche position between heritage and engineering, is
frequently exposed to outdoor atmospheric corrosion and, in the case of bridges,
gates and similar structures, may be required to perform a distinct function. Sector
guidance to direct practices is based on anecdotal evidence and established
methods. British Standards relate to modern steels hence application to historic
ferrous metals is complicated by differences in metallurgy and lack of concession to
conservation ethics.
This study generates empirical evidence of the effects of five surface preparation
methods and three protective coating systems on the corrosion rate of historic
wrought iron samples. Immersion in sodium hydroxide solution and blasting with
crushed walnut shells are found to reduce corrosion rates of uncoated wrought iron.
Aluminium oxide and glass beads blasting increase corrosion rate but offer removal
of contaminants and a keyed surface for coating adhesion. Flame cleaning increases
corrosion rate by almost four times the uncleaned wrought iron corrosion rate.
A two-pack epoxy resin coating system with polyurethane topcoat applied over
substrate surfaces blasted to Sa2.5 (near white metal) and a surface tolerant single-
pack alkyd coating applied over coherent oxide layers successfully prevented
corrosion for almost two years in high static relative humidity environments. An alkyd
system applied over Sa2.5 blasted surface does not significantly reduce corrosion
rate of the uncoated substrate. A cost benefit approach to interpreting the empirical
results in relation to practicalities of applying the treatments is advocated.
The methods developed for standardising historic sample material and measuring
oxygen consumption of coated samples as proxy corrosion rate offer scope for
further work in this area. A standardised approach to testing permits correlation of
test data between workers in this area to generate a database of empirical data to
inform decision-making.
i
Acknowledgements
Many thanks to Historic Scotland for funding this study, in particular Craig Kennedy
(now Herriot Watt University), David Mitchell, Alick Leslie and all of the Technical
Conservation team.
This practitioner focused study could not have taken place without the participation
of conservation and heritage ironwork practitioners. Thanks to Pete Meehan for
advice, sample material and access to blasting shops and foundries, to Eura
Conservation for coatings, to Alex Coode for his insight and to all practitioners
involved in the flame cleaning. Thanks also to Dave Thickett at Historic England for
his support and advice.
At Cardiff University, thanks to the Conservation team of Jane Henderson, Yiota

Manti and especially to Phil Parkes for his training on the SEM. Thanks to Paul
Nicholson and John Hines for their support during the PhD annual reviews and to
Helen Szewczyk for her caring and efficient administration of postgraduate study. At
Amgueddfa Cymru, I am grateful to Tom Cotterell for his assistance with XRD.
Within the Cardiff Ferrous Metals Research Group, thanks to Melanie Rimmer for her
invaluable advice and assistance and to Eric Nordgren for his companionship,
experience and anecdotes. Thanks to Amber Lawson for sharing the images of paint
layer cross sections. The biggest debt of gratitude is owed to David Watkinson for his
supervision and friendship over the course of this PhD.
Finally, thanks to my family for their unfailing support in all things.
ii
Dedication
For James Donald Emmerson, Christopher John Mannion and Reuben Jung Karki.
iii
Contents
1 Introduction ......................................................................................................... 1
1.1 The nature of historic wrought iron .............................................................. 1
1.1.1 Production and standardisation............................................................. 1
1.1.2 Composition and properties .................................................................. 3
1.1.3 Form and function .................................................................................. 5
1.2 Research in heritage wrought iron ................................................................ 7
1.3 The study in context ...................................................................................... 9
1.4 Aim and objectives ...................................................................................... 10
1.5 Thesis structure ........................................................................................... 10
2 Corrosion Principles: Heritage wrought iron and its contexts ........................... 12
2.1 Principles of corrosion ................................................................................. 12
2.1.1 The corrosion process .......................................................................... 12
2.1.2 Corrosion of iron: an overview ............................................................ 14
2.2 Corrosion prevention by protective coatings.............................................. 18
2.2.1 Mechanisms of protection ................................................................... 19
2.2.2 Factors affecting coating performance and failure ............................. 21
2.3 Corrosion variables ...................................................................................... 25
2.3.1 Water.................................................................................................... 25
2.3.2 Climate ................................................................................................. 26
2.3.3 Oxygen.................................................................................................. 28
2.3.4 Intrinsic factors..................................................................................... 33
2.3.5 Pollutants and contaminants ............................................................... 35
2.4 Corrosion products and their properties .................................................... 38
iv
2.4.1 The importance of corrosion products ................................................ 38
2.4.2 αFeOOH (Goethite) .............................................................................. 39
2.4.3 βFeOOH (Akaganeite) .......................................................................... 39
2.4.4 γFeOOH (Lepidocrocite) ....................................................................... 41
2.4.5 Fe3O4 (Magnetite) ................................................................................ 41
2.4.6 α-Fe2O3 (Haematite) ............................................................................. 41
2.4.7 ƴFe2O3 (Maghemite) ............................................................................ 42
2.4.8 Ferrous and ferric chlorides ................................................................. 42
2.5 Corrosion context ........................................................................................ 43
3 Corrosion Prevention Treatments ..................................................................... 48
3.1 Traditional treatment of wrought iron ........................................................ 48
3.2 Current practice........................................................................................... 50
3.2.1 Recording and dismantling .................................................................. 51
3.2.2 Cleaning and surface preparation ........................................................ 51
3.2.3 Repairs .................................................................................................. 53
3.2.4 Protective coatings............................................................................... 54
3.2.5 Application of protective coatings ....................................................... 55
3.2.6 Maintenance ........................................................................................ 55
3.3 Evidence from paint layers in section ......................................................... 56
3.4 Decision-making in conservation of historic wrought iron ......................... 60
3.4.1 Decisions and decision-makers ............................................................ 60
3.4.2 Guidance .............................................................................................. 64
3.4.3 Use of standards in heritage contexts ................................................. 65
4 Experimental ...................................................................................................... 69
4.1 Ethos and rationale ..................................................................................... 69
4.1.1 Research ethos ..................................................................................... 69
v
4.1.2 Related work ........................................................................................ 69
4.1.3 Rationale for methods and parameters ............................................... 71
4.2 Investigating the impact of surface preparation method on corrosion of

historic wrought iron.............................................................................................. 76
4.2.1 Aims and Objectives ............................................................................. 76
4.2.2 Characterisation of sample material .................................................... 76
4.2.3 Surface preparation methods .............................................................. 78
4.2.4 Examination of resultant surfaces ....................................................... 80
4.2.5 Measuring oxygen consumption and corrosion rates ......................... 81
4.2.6 Examining corrosion rates.................................................................... 82
4.2.7 Identification of corrosion products .................................................... 83
4.2.8 Flame cleaning ..................................................................................... 84
4.3 Investigating protective coatings for historic wrought iron........................ 85
4.3.1 Aims and objectives ............................................................................. 85
4.3.2 Sample material ................................................................................... 85
4.3.3 Protective coatings............................................................................... 86
4.3.4 Sample surface preparation ................................................................. 87
4.3.5 Coating application .............................................................................. 87
4.3.6 Oxygen consumption and corrosion rates ........................................... 88
4.3.7 Control samples ................................................................................... 88
4.3.8 Practical observations .......................................................................... 88
4.3.9 Properties of cured coatings ................................................................ 89
5 Results ................................................................................................................ 91
5.1 Investigating the impact of surface preparation method on corrosion of

historic wrought iron.............................................................................................. 91
5.1.1 Railing sample characterisation ........................................................... 91
vi
5.1.2 Preparation of surfaces ........................................................................ 96
5.1.3 Oxygen consumption and corrosion rates of prepared samples ......... 97
5.1.4 Visual sample surfaces post high RH environment ............................ 106
5.1.5 XRD of corrosion products formed during high RH exposure ........... 107
5.1.6 Practitioner flame cleaned samples................................................... 110
5.1.7 Scanning Electron Microscopy ........................................................... 113
5.2 Investigating protective coatings for historic wrought iron...................... 115
5.2.1 Gasometer sample characterisation .................................................. 115
5.2.2 Prepared samples............................................................................... 120
5.2.3 Coating properties.............................................................................. 122
5.2.4 Oxygen consumption and corrosion rates of coated samples .......... 136
5.3 Comparing sample materials ..................................................................... 149
6 Discussion ......................................................................................................... 151
6.1 Surface Preparation Methods ................................................................... 151
6.1.1 Grading rust by pictorial standards within heritage contexts ........... 151
6.1.2 Classification and comparison of prepared surfaces ......................... 152
6.1.3 Morphological interpretations ........................................................... 153
6.1.4 Suitability of surfaces for application of protective coatings ............ 154
6.1.5 Practical considerations in application of preparation methods:

heritage context ............................................................................................... 156
6.1.6 Success and practicality criteria ......................................................... 158
6.1.7 Oxygen consumption rates of prepared surfaces .............................. 159
6.1.8 Flame cleaning in practice.................................................................. 166
6.1.9 Interpolation to develop context ....................................................... 167
6.2 Protective Coatings.................................................................................... 168
6.2.1 Cost..................................................................................................... 171
vii
6.2.2 Availability .......................................................................................... 172
6.2.3 Volumes of coating and numbers of coats ........................................ 172
6.2.4 Primers ............................................................................................... 174
6.2.5 Required surface preparation ............................................................ 174
6.2.6 Application ......................................................................................... 174
6.2.7 Testing quality of application ............................................................. 175
6.2.8 Adhesion of coatings .......................................................................... 176
6.2.9 Aesthetics ........................................................................................... 178
6.2.10 Practicalities ....................................................................................... 182
6.2.11 Recommendations and endorsements .............................................. 183
6.2.12 Lifetime to maintenance and durability ............................................ 184
6.2.13 Corrosion prevention ......................................................................... 185
6.2.14 Sample material ................................................................................. 186
6.2.15 Standards for corrosion protection by protective coatings ............... 190
6.2.16 Heritage coating standards ................................................................ 192
6.3 Standardising test methods for heritage: scope, limitations and further work
195
6.3.1 Scope of method for heritage sector studies .................................... 195
6.3.2 Limitations and further work ............................................................. 197
7 Conclusions ...................................................................................................... 199
7.1 Cost benefit ............................................................................................... 199
7.2 Method ...................................................................................................... 200
8 Reference List ................................................................................................... 202
9 Appendices ....................................................................................................... 224
9.1 Appendix 1: Recommended surface preparation methods ...................... 224
9.2 Appendix 2: Table of coating details ......................................................... 229
viii
9.3 Appendix 3: Leighs Metagard L574 product technical data...................... 231
9.4 Appendix 4: Leighs Epigrip C400V3 product technical data...................... 233
9.5 Appendix 5: Leighs Resistex C137V2 product technical data.................... 235
9.6 Appendix 6: Hammerite Red Oxide Primer datasheet .............................. 237
9.7 Appendix 7: Hammerite Garage Door datasheet ...................................... 238
9.8 Appendix 8: Hammerite Direct to Rust datasheet .................................... 239
ix
List of Figures
Figure 1.1 Polished section of railing wrought iron sample material used in this study..
...................................................................................................................................... 3
Figure 1.2 Wrought iron gates and railings a) Davies Brothers gates at Chirk Castle b)
Tijou Screens at Hampton Court Palace....................................................................... 5
Figure 1.3 Wrought iron ship ss Great Britain a) returning to its dry dock in Bristol
1970 b) the hull in sealed dry dock with desiccation plant to control humidity c) epoxy
resin coating to minimise atmospheric corrosion of the freeboard. ........................... 6
Figure 2.1 Electrolytic corrosion of iron. .................................................................... 13
Figure 2.2 Corrosion process during wet/dry cycle phases. ...................................... 17
Figure 2.3 Development of blisters in poorly adherent paint films owing to osmotic
pressure...................................................................................................................... 23
Figure 2.4 Suggested sequence of blister formation ................................................. 23
Figure 2.5 Wrought iron gates at the Palace of Holyroodhouse showing coating failure
and laminating corrosion of underlying substrate..................................................... 24
Figure 2.6 Oxygen starvation in a water drop on the surface of iron. ....................... 29
Figure 2.7 Possible consequences of coating damage............................................... 31
Figure 2.8 Schematic of filiform corrosion ................................................................. 33
Figure 2.9 Lamellar corrosion of early wrought iron railway wedge. ........................ 34
Figure 2.10 Potential pH diagram for the iron water system .................................... 38
Figure 2.11 Wrought iron brine tank at Lion Salt works Cheshire. ............................ 44
Figure 2.12 Oxygen starvation cell. ............................................................................ 44
Figure 2.13 (i) Lamellar corrosion (ii) pitting corrosion ............................................. 45
Figure 2.14 Brunel’s ss Great Britain paint loss and lamellar corrosion. ................... 46
Figure 3.1 Bituminous coating on a brine tank in Cheshire. ...................................... 49
Figure 3.2 Polished section of paint layers from: a) the Tijou screen at Hampton Court
Palace; b) railings at Dumbarton Castle ..................................................................... 50
Figure 3.3 Interrelating factors in selection of surface preparation method. ........... 62
Figure 3.4 Interrelating factors in coating system selection...................................... 62
Figure 3.5 Diagram indicating the complexity of the decision making process for
treatment of historic wrought involving many interrelated factors ......................... 63
x
Figure 4.1 a) working principle behind optical oxygen electrodes. ........................... 72
Figure 4.2 Iron railing from which sample material was cut. .................................... 77
Figure 4.3 Sample of wrought iron railing as cut to form samples. ........................... 78
Figure 5.1 SEM image of polished section of railing wrought iron showing slag
inclusions distributed in direction of rolling. ............................................................. 91
Figure 5.2 Backscatter electron SEM image showing detail of two phase slag
inclusions with typically rounded wüstite (FeO). ....................................................... 92
Figure 5.3 Diffraction patterns of corrosion sample and matching compounds: the
corrosion product sample; lepidocrocite (01-074-1877); magnetite (01-085-1436);
goethite (01-081-0462). ............................................................................................. 92
Figure 5.4 Backscattered electron image showing location of analyses of paint layers
from the sample material. ......................................................................................... 93
Figure 5.5 Spectrum 1 of SEM analysis of paint layers on sample material showing
major peaks for sulfur and barium with minor peaks for copper, cobalt, strontium
and iron. ..................................................................................................................... 93
major peaks for sulfur and barium and minor peaks for copper, cobalt and strontium.
.................................................................................................................................... 94
major peaks for lead and calcium and minor peaks for aluminium, barium and copper
.................................................................................................................................... 94
major peaks for lead and minor peaks for copper, zinc and iron. ............................. 95
Figure 5.9 Morphology of metal surface post-preparation shown in macroscopic
(upper) and secondary electron SEM (lower) images for each surface preparation
method. ...................................................................................................................... 96
Figure 5.10 Oxygen consumption of glass beads blasted samples at 90% RH. ......... 99
Figure 5.11 Oxygen consumption of aluminium oxide blasted samples at 90% RH. 99
Figure 5.12 Oxygen consumption of crushed walnut blasted samples at 90% RH.. 100
Figure 5.13 Oxygen consumptiom of sodium hydroxide and wire brush cleaned
samples at 90% RH. .................................................................................................. 100
xi
Figure 5.14 Oxygen consumption of flame and wire brush cleaned samples at 90%RH.
.................................................................................................................................. 101
Figure 5.15 Oxygen consumption of un-cleaned railing samples at 90% RH. ......... 101
Figure 5.16 Average oxygen consumption of samples prepared by each preparation
method and control samples. .................................................................................. 103
Figure 5.17 Change in gradient of trendline (i.e. rate) for average oxygen
consumption of samples by surface preparation method ...................................... 103
Figure 5.18 Boxplot showing oxygen consumed by samples prepared by each method
(mol/year/mm2 x 10-8) as averaged over test period. ............................................. 104
Figure 5.19 Comparison of prepared and post-high RH exposure surfaces. Images
show sample surfaces of 30mm x 40mm................................................................. 106
Figure 5.20 a) Diffractogram of corrosion products from glass beads cleaned sample;
b) patterns of identified corrosion products (uppermost pattern) indicating presence
of lepidocrocite (00-044-1415 (middle pattern)) and goethite (00-029-0713 (lowest
pattern)). .................................................................................................................. 108
Figure 5.21 a) Diffractogram of corrosion products from flame cleaned sample; b)
patterns of identified corrosion products (uppermost pattern) indicating presence of
hematite (01-089-0598 (middle pattern)) and goethite (01-081-0464 (lowest
pattern)). .................................................................................................................. 109
Figure 5.22 Oxygen consumption of samples flame cleaned by Practitioner A and
subsequently exposed to 90% RH. ........................................................................... 110
Figure 5.23 Oxygen consumption of samples flame cleaned by Practitioner B and
Figure 5.24 Oxygen consumption of samples flame cleaned by Practitioner C and
Figure 5.25 Boxplot showing oxygen consumption (mol/year/mm2 x 10-8) of samples
flame cleaned by three heritage practitioners (PA, PB and PC) and the wire brushed
gasometer substrate ................................................................................................ 112
Figure 5.26 Boxplot showing theoretical metallic iron converted to FeOOH
(mol/year/mm2 x 10-8) of samples flame cleaned by three heritage practitioners (PA,
PB and PC) and wire brushed gasometer substrate ................................................ 113
xii
Figure 5.27 Backscatter SEM image (x150 magnification) of a polished cross section
of gasometer wrought iron showing characteristic slag inclusions and corrosion
product layers.. ........................................................................................................ 114
of gasometer wrought iron after flame cleaning showing fragmentation of corrosion
product layers .......................................................................................................... 114
of gasometer wrought iron showing characteristic slag inclusions. ........................ 115
Figure 5.30 Backscatter SEM image showing location of spectra for analysis of
gasometer wrought iron. ......................................................................................... 116
Figure 5.31 a) Diffractogram of corrosion products from gasometer sample material;
b) patterns of identified corrosion product (uppermost pattern) indicating presence
of lepidocrocite (00-044-1415 (lower pattern)). ..................................................... 117
of goethite (00-029-0713 (lower pattern)). ............................................................. 118
of magnetite (01-080-0390 (lower pattern)). .......................................................... 119
Figure 5.34 Prepared samples showing uncoated substrates and samples after
application of each coating constituting the system ............................................... 121
Figure 5.35 Sample HSII08 showing bubbles in System A primer after curing ........ 126
Figure 5.36 Sample HSII04 showing pinhole holidays in cured System A build coating
.................................................................................................................................. 126
Figure 5.37 Sample HSII01 showing brushstrokes in cured System A build coating
.................................................................................................................................. 127
Figure 5.38 Sample HSII08 showing pinholes in cured System A topcoat ............... 127
Figure 5.39 Sample HSII16 showing pinhole holidays in cured System B primer .... 128
Figure 5.40 Sample HSII13 showing brushstrokes in cured System B primer ......... 128
Figure 5.41 Sample HSII18 showing peaks run back of System B topcoat from peaks
revealing red primer................................................................................................. 129
xiii
Figure 5.42 Sample HSII11 showing run back of System B topcoat from peaks and
edges. ....................................................................................................................... 129
Figure 5.43 Sample HSII26 showing pinholes in Coating C after curing .................. 130
Figure 5.44 Sample HSII28 showing run back of cured Coating C from edges. ....... 130
Figure 5.45 Sample HSII22 showing areas missed during application of Coating C to
sample edge ............................................................................................................. 130
Figure 5.46 Boxplot showing the dry film thicknesses of the cured coatings ......... 132
Figure 5.47 Boxplot showing adhesion pull-off values for coatings.. ...................... 134
Figure 5.48 Pull-off dollies (left) and corresponding areas of coated samples (right)
showing modes of failure. Dollies and pull-off areas have a diameter of 20mm. a), b)
System A with failure cohesive within the primer; c), d) System B with failure mostly
adhesive between primer and topcoat; e), f) Coating C with failure cohesive within
corrosion products. .................................................................................................. 135
Figure 5.49 Oxygen consumption of wrought iron samples coated with coating
System A and subsequently exposed to 90% RH. .................................................... 137
System B and subsequently exposed to 90% RH. .................................................... 138
Figure 5.51 Oxygen consumption of wrought iron samples coated with Coating C and
Figure 5.52 Oxygen consumption of glass control samples coated with coating System
A and subsequently exposed to 90% RH.................................................................. 139
B and subsequently exposed to 90% RH. ................................................................. 139
Figure 5.54 Oxygen consumption of glass control samples coated with Coating C and
Figure 5.55 Boxplot showing oxygen consumed by coated and uncoated samples
(mol/year/mm2 x 10-8) as averaged over test period. Oxygen consumption of the
coatings has been calculated using the surface area of coating applied ................ 142
Figure 5.56 Boxplot showing theoretical metallic iron converted to FeOOH by coated
and uncoated samples (mol/year/mm2 x 10-8) as averaged over test period. Oxygen
consumption of the coatings has been calculated using the surface area of coating
applied ...................................................................................................................... 143
xiv
coating is calculated using the mass of coating applied .......................................... 145
consumption of the coatings is calculated using the mass of coating applied ....... 146
Figure 5.59 Sample HSII02 after exposure to 90%RH for 540 days showing damage to
corners but otherwise unchanged appearance. ...................................................... 147
Figure 5.60 Sample HSII02 showing damage to topcoat of System A after 90%RH
exposure ................................................................................................................... 147
Figure 5.61 Sample HSII11 showing adhesion of particulates to the surface of System
B samples ................................................................................................................. 148
Figure 5.62 Sample HSII17 after exposure to 90%RH showing particulate pickup and
run back of top coat from edges and peaks. ........................................................... 148
Figure 5.63 Sample HSII26 after exposure to 90%RH for 540 days showing no visible
corrosion even where coating build is very low on edges. ...................................... 148
Figure 5.64 Boxplot showing oxygen consumed by railing and gasometer wrought
iron samples (mol/year/mm2 x 10-8) as averaged over test period......................... 149
Figure 5.65 Boxplot showing theoretical metallic iron converted to FeOOH by railing
and gasometer wrought iron samples (mol/year/mm2 x 10-8) as averaged over test
period ....................................................................................................................... 150
Figure 6.1 Oxypropane torch used by practitioner PA. ........................................... 163
Figure 6.2 Dull red glow of sample flame cleaned by practitioner PA. ................... 163
Figure 6.3 Position of the flame relative to the sample during flame cleaning by
practitioner PB. ........................................................................................................ 164
Figure 6.4 Luminescence of coating and oxide vestiges under flame (practitioner PB).
.................................................................................................................................. 164
Figure 6.5 Wire brushing of flame cleaned sample by practitioner PB. .................. 164
Figure 6.6 Position of the oxyacetylene torch relative to the sample during flame
cleaning by practitioner PC. ..................................................................................... 165
Figure 6.7 Dull red glow of wrought iron during flame cleaning by practitioner PC.
.................................................................................................................................. 165
xv
Figure 6.8 Characteristic bright red oxides forming developing 1-2 minutes after
flame cleaning on all samples (uncleaned sample top left for comparison). .......... 165
Figure 6.9 Showing corrosion from pinholes in a sprayed coating. ......................... 175
Figure 6.10 a-c Showing corrosion on mild steel structural elements in areas missed
during coating application. ...................................................................................... 179
Figure 6.11 Failure of system similar to System A within 12 months of reinstatement
in coastal environment. Corrosion staining evident on pale coloured topcoat. ..... 181
Figure 6.12 Showing an example of the effects of preparing two-pack coating in
incorrect ratio of resin and hardener. ..................................................................... 182
xvi
List of Tables
Table 1.1 A representative analysis range of wrought iron ......................................... 3
Table 2.1 Components of paint coatings ................................................................... 19
Table 2.2 Corrosivity categories of exterior environments according to BS EN ISO
12944-2:1998 ............................................................................................................. 47
Table 3.1 Summary of some non-destructive surface cleaning methods, their
applications, advantages and disadvantages after Childs (1985, 26). ....................... 52
Table 3.2 Suggested schedule of maintenance of exterior wrought ironwork (after
Godfraind et al. 2012, 80). ......................................................................................... 56
Table 3.3 Microscopy images of polished sections of coatings from wrought and cast
iron and mild steel railings and gates from Historic Scotland properties (courtesy of
Amber Lawson) highlighting aspects of application, performance and failure. ........ 59
Table 4.1 Regional humidity averages 1981 – 2010. ................................................. 73
Table 4.2 Contextual scenarios for coating selection. ............................................... 75
Table 4.3 Parameters of flame cleaning by practitioners. ......................................... 84
Table 4.4 Overview of coatings investigated. ............................................................ 87
Table 5.1 Results of SEM analysis of the wrought iron giving compositions of three
areas analysed. ........................................................................................................... 95
Table 5.2 Classification of surface preparation outcomes using BS EN ISO 8501-
1:2007......................................................................................................................... 97
Table 5.3 Oxygen consumption averaged over test period (mol/year/mm2x10-8) and
theoretical conversion of metallic iron to FeOOH (mol/year/mm2x10-8) of all samples
giving averages by preparation method .................................................................. 102
Table 5.4 Significantly different oxygen consumption per year per mm2 of iron
between surface preparation methods (calculated by Kruskal-Wallis and with a
significance level of 0.05). ........................................................................................ 105
Table 5.5 Results of quantitative SEM analysis of gasometer wrought iron. .......... 116
Table 5.6 Properties of coatings during application. ............................................... 123
Table 5.7 Properties of cured coatings. ................................................................... 125
Table 5.8 Dry film thicknesses of cured coatings. .................................................... 131
Table 5.9 Adhesion pull-off values (MPa) and modes of failure. ............................. 133
xvii
Table 5.10 Results of oxygen consumption testing of coated wrought iron samples
giving moles of oxygen consumed and theoretical moles of Fe converted to FeOOH
per year per mm2 x10-8.. .......................................................................................... 141
giving moles of oxygen consumed and moles of metallic iron converted to FeOOH per
year per mm2 x10-8.. ................................................................................................. 144
Table 5.12 Significantly different oxygen consumption and theoretical metallic iron
conversion to FeOOH (mol/year/per mm2 of iron) between coated samples and
uncoated substrates (calculated by Kruskal-Wallis and with a significance level of
0.05). ........................................................................................................................ 146
Table 6.1 Compliance of surface preparation methods with practitioner/specifier
criteria. ..................................................................................................................... 159
Table 6.2 Summary of features of coating systems tested...................................... 170
Table 6.3 Showing trendline gradients for first 75 days of oxygen consumption testing
at 90% RH for uncleaned railing sample material.................................................... 188
Table 6.4 Descriptive statistics for railing and gasometer sample material blasted to
Sa2.5 and exposed to 90%RH for 338 and 351 days respectively. .......................... 189
Table 6.5 Further work test variables for corrosion rate investigations. ................ 198
xviii
1 Introduction
1.1 The nature of historic wrought iron

1.1.1 Production and standardisation
Modern production processes are designed to deliver material outputs that are
standardised, uniform in appearance and composition, predictable in properties and
responses, mass producible and guaranteed to perform over a specified and
predetermined period. Many are ultimately disposable and intended to require
minimal maintenance during their useful lifetime. This approach fits with societal
expectations of predictability, ease of use, fit-for-purpose, minimum intervention
until wholesale replacement and a money back guarantee in case of early failure.
Industry and engineering also work in this manner with components engineered for
minimum maintenance and periodic replacement.
Having as its raison d’être the preservation of archaeological and historic objects in
perpetuity, conservation cannot buy into this disposable culture. Failures of materials
must be prevented and deterioration rates reduced to a bare minimum. A running
battle against agents of decay is fought with little financial support for research into
appropriate materials, treatments and environments. All this must be achieved
within the constraints of professional ethics. The process is made all the more
difficult by the nature of the wide range of historic materials it deals with which are
not standardised and predictable in the manner of their modern counterparts.
The production of archaeological and historic wrought iron is a case in point.

Numerous histories, chronologies and studies detail developments up to the 20 th
century (Tylecote 1976; Starley 1999; O’Sullivan and Swailes 2009; Merriman 1958;
Hooper et al. 2003; McDonnell 1989; Charlton et al. 2012; Rehren et al. 2013; Aston
and Story 1939; Wallis and Bussell 2008; Lyon 2010, 1695; Scott and Eggert 2009)
highlighting changing methods with technological advances and resultant change in
material composition and properties. Reduction of iron from its ore was originally a
small scale batch process for domestic, utilitarian objects. Furnace temperatures well
below the melting point of iron (1540oC) produced a solid bloom of iron, siliceous
1
slag and other waste materials. Hand working removed much of the slag with the
remainder distributed in stringers. Growing in scale to meet the requirements of the
Roman military and a broader range of domestic and civic objects, it remained a
batch process with smelt halted to remove the bloom until the development of the
blast furnace in the late 16th/ early 17th centuries. Increased fuel: ore ratios created
more reducing atmospheres carburising the iron, reducing its melting point to 1200oC
and permitting liquid iron to be tapped from the bottom of the furnace with the slag
as pig iron. Wrought iron was produced from the pig by puddling or finery in an
oxidising furnace. The quantity and uniformity of iron produced by hand puddling
was limited but mechanisation reduced quality which relied on the skill of the
puddler. Fuel use changed in the 18th century from charcoal to coke. Ore type was
variable by region and over time with depletion of deposits. Eventually, invention of
the Bessemer Converter in the later 19th century producing mild steel at a lower cost
led to its replacement of wrought iron by the 20th century.
The picture of wrought iron production is of temporal and geographic diversity of

methods, batch processes and influence of individual operators, a range of ores and
changing fuel. A heterogeneous material results with composition varying:
 microscopically between bulk of iron and slag inclusions and within both;
 across objects and structures forged from multiple blooms or by joining
sections;
 over time with changing furnace technology and fuel type;
 regionally with ore.
It was not until 1910 that there was any regulation of production when British
Standard 51 was introduced to test and certify new wrought iron for use in railways
and 1939 that this was extended to cover wrought iron for general engineering
purposes (British Standards Institute 1910; British Standards Institute 1939). The
previous variability in composition greatly affected the properties and quality of the
material.
2
1.1.2 Composition and properties
Wrought iron is essentially a composite material of ferrite, small amounts of carbon
and entrapped slag stringers (Scott and Eggert 2009, 165). Slag can form up to 3wt%
of the total mass of wrought iron with thickness of inclusions ranging from
microscopic to 3mm (O’Sullivan and Swailes 2009, 261). Figure 1.1 is a Nital etched
polished section of wrought iron showing the characteristic variable grain sizes and
two phase slag (typically globular wüstite, FeO, dendrites in a glassy siliceous matrix)
(Scott 1991, 89). A representative elemental analysis range for wrought iron is given
Table 1.1.
Figure 1.1 Polished

section of railing wrought
iron sample material
(4.2.2) used in this study.
Etch: Nital. Variable grain
size and two phase
(typically rounded
wüstite (FeO) dendrites
in a glassy matrix) slag as
small particles and larger
fragments are
characteristic of wrought
iron (Scott 1991, 89).
Element Content in wrought iron (%)
Carbon 0.02 – 0.03

Silicon 0.02 – 0.10
Sulfur 0.008 – 0.02
Manganese 0.00 – 0.02
Phosphorus 0.05 – 0.25
Slag 0.05 – 1.50
Iron Balance
Table 1.1 A representative analysis range of wrought iron (Higgins 1957, 36).
3
Improving techniques and furnace technology increased the scope and efficiency of
wrought iron production. Associated compositional changes result, for example a
decrease in phosphorus content of Romano British wrought iron relative to earlier
material. Phosphorus content is determined by ore and fuel, sulfur by ore (until later
coal fired furnaces), and nickel and copper content by ore (Tylecote 1976, 32).
Phosphorous, manganese, sulfur and silicon may indicate furnace temperatures by
their degree of reduction into the iron (Starley 1999, 1128). Carbon content
evidences smelting technique and influences properties; malleability and ductility are
reduced by increasing carbon content (Tylecote 1976, 32; O’Sullivan and Swailes
2009, 261).
Rolling of puddled iron elongates the masses of slag in the direction of rolling and
they are visible as streaks in longitudinal section and dots in transverse (Higgins 1957,
36; Greaves and Wrighton 1967, 82; O’Sullivan and Swailes 2009, 261) (Fig. 5.1 &
5.29). Amount and distribution of slag depends on extent of working during forging
and influences properties of the wrought iron. Tensile strength of plate wrought iron
is greater along the grain (direction of rolling) than across it and is more resistant to
bending about the axis perpendicular to the grain than parallel to it (Greaves and
Wrighton 1967, 83). When load is across the grain, slag filaments running
perpendicular to the load path act as voids for propagation of internal cracks
(O’Sullivan and Swailes 2009, 266). With slag central to mechanical properties of
wrought iron, variability in its mass and distribution affects its behaviour. Within a
wrought iron structure, sections will have undergone different degrees of working
causing an inhomogeneity of properties across the structure. Testing of historic
wrought iron properties must be on large samples relative to the slag inclusions to
ensure representation of the bulk material (Hooper et al. 2003, 78).
This inherent variability in wrought iron offers challenges to conservation in

predicting its corrosion behaviour and mechanical properties. A one-size-fits-all
approach cannot be expected to succeed, particularly given the nature of historic
wrought iron objects and structures.
4
1.1.3 Form and function
Early ironwork took the form of domestic and military items which become
archaeological or museum objects with corrosion controllable by manipulation of the
internal environment. By the 18th century wrought iron was in use for gates and
railings (Tijou screens at Hampton Court Palace and the Davies Brothers’ gates at
Chirk Castle) (Fig. 1.2), garden ornamentation (Thomas Bakewell garden arbour) and
street furniture (Hayman 2011).
a) b)
Figure 1.2 Wrought iron gates and railings a) Davies Brothers gates at Chirk Castle
(http://www.ntprints.com/image/374510/the-magnificent-wrought-iron-gate-at-
chirk-castle) b) Tijou Screens at Hampton Court Palace
(http://www.geograph.org.uk/photo/2657762)
An array of historically significant industrial ironwork exists as bridges, mining and

other industrial plant, equipment and infrastructure embodying industrial and
technological progress, period and society (Kreislova et al. 2013, 311-314). High
tensile strength allowed wrought iron plate to become the basic device of
construction as riveted rolled sections could span required distances and carry
sufficient loads for railway bridges, unlike cast iron (Topp 2010). Joining techniques
for wrought iron included welding but also mortice and tenon joints, rivets and collars
influenced by carpentry (Taylor and Suff 2010). Much industrial ironwork takes the
form of trusses, tubes and piping systems which are part of larger objects or sheets
(Kreislova et al. 2013, 330). 1843 saw the first extensive use of wrought iron for the
5
hull of a ship with the launch of the ss Great Britain (Fig. 1.3) (Watkinson and Tanner
2008; Watkinson et al. 2006).
a)
b) c)
Figure 1.3 Wrought iron ship ss Great Britain a) returning to its dry dock in Bristol
1970 b) the hull in sealed dry dock with desiccation plant to control humidity c) epoxy
resin coating to minimise atmospheric corrosion of the freeboard. Images ss Great
Britain Trust.
The diversity of wrought iron objects and structures in heritage contexts means
responsibility for their preservation falls to individuals who have varying degrees of
treatment experience. Objects may be in museum collections curated and conserved
6
by specialists, items of street furniture the responsibility of local councils, gates and
railings on historic properties managed by the likes of The National Trust, Cadw or
Historic Scotland with or without dedicated or peripatetic conservation staff or could
equally be in the private sector. Its condition may be anywhere from well-maintained
with corrosion limited to closely adherent oxides overlying a substantial metal core
to extremely corroded with laminating corrosion products, significant loss of section
and minimal metal core remaining. The challenges in dealing with its preservation
are significant. The decision-making process for individuals specifying and carrying
out corrosion protection treatment of exterior historic wrought iron and the
guidance available to underpin decisions are discussed in 3.4.
1.2 Research in heritage wrought iron

Heritage wrought iron may be on display or in store as part of museum collections
but frequently occurs outdoors exposed to prevailing conditions. Whatever its
context and despite variability of available finance, resources and expertise,
management of iron objects and structures aims to reduce corrosion rates to a bare
minimum for maximum longevity of object lifetimes. Achieving this requires decision-
making on treatments and appropriate environments that is based on empirical
evidence of resultant corrosion rates and proven success in their reduction.
Storage and display in low humidity environments has been studied extensively for
objects and structures, small and large (Watkinson and Lewis 2004; Rimmer and
Watkinson 2010). This research has been driven by big budget projects such as the
development of an award winning, innovative conservation strategy for the ss Great
Britain and by the needs of large institutions housing vast iron collections which test
management protocols (Watkinson et al. 2005). In contrast, conservation of historic
wrought iron exposed externally and above ground has received little attention yet
it offers a significant challenge to specifiers and practitioners responsible for its
survival and treatment.
7
Outside conservation, wrought iron has been studied by workers concerned with its
metallography (Bates 1984; Blakelock et al. 2009), technology and history of
production (Light 2000; Sutherland 2009; Charlton et al. 2012) or unusual examples
(Kumar and Balasubramaniam 1998) and corrosion mechanisms (Neff et al. 2006;
Degrigny et al. 2007; Maréchal et al. 2007; Monnier et al. 2007; Bouchar et al. 2013).
Industrial and engineering research deals with atmospheric corrosion and protective
coatings but focuses on modern steel substrates limiting direct applicability to
heritage. There has been no holistic overview bringing these studies together and
relating them to exterior historic wrought iron exposed to atmospheric conditions
and requiring corrosion prevention conservation treatment. Further relating
conservation products and methods and introducing empirical evidence of
effectiveness of treatments whilst considering the unique nature of this conservation
context and the diverse range of individuals responsible for the wellbeing of this
material has never been carried out. As a result, the limited guidance available to
underpin practice in this area is anecdotal and unstandardised. Recourse to British
Standards is unsatisfactory as they are designed for engineering and industry and are
not fit-for-purpose for heritage contexts, historic materials and related ethical
concerns.
This PhD takes the approach of integrating metallurgy, corrosion theory,

conservation and industrial practice and experimental study as befits a material
occupying a niche position between heritage, industry and engineering. Practitioner
methods for wrought iron treatment are studied to develop testing protocols.
Quantitative and qualitative evidence of effectiveness of surface preparation
methods and protective coatings is generated and will be used to produce guidance
tailored to the needs of specifiers and practitioners directly involved in developing
and implementing treatments. Focus is on delivering the required empirical data to
underpin effective decision-making for heritage.
8
1.3 The study in context
This study has been commissioned by Historic Scotland to extend the largely
qualitative testing of conservation methods previously carried out for historic cast
iron. Having a rich cast and wrought iron heritage and with the traditional built
environment and heritage tourism important economically (Ecorys 2013), examining
treatments for traditional building materials is a strategic concern. Selection of
methods for ironwork preservation is currently based on practitioner experience and
anecdotal reports. Empirical evidence will allow decisions to be made with a strong
underpinning rationale developed from understanding of treatment effects on
materials and their long term survival prospects. Consideration of practicalities,
resource implications and maintenance regimes is integral to examination of
methods in this study and enriches the decision-making conversation. Results will
contribute to informing future treatments specified by Historic Scotland and
strengthen their position for offering evidence based advice to others in this field.
Conservation at Cardiff University has developed an international reputation for

delivering the empirical evidence essential for effective management of heritage
assets, particularly ferrous metals. This PhD follows successful determination of ‘safe’
relative humidity values for chloride infested wrought iron (Watkinson and Al Zahrani
2008; Watkinson and Lewis 2004; Watkinson and Lewis 2005a; Watkinson and Lewis
2005b; Watkinson and Tanner 2008) and efficiency of chloride extraction treatments
for archaeological iron (Watkinson and Al Zahrani 2008; Rimmer and Watkinson
2010; Watkinson et al. 2013). Sector guidelines for the storage of archaeological iron
resulted which direct practices in this area nationally and internationally (Rimmer et
al. 2013). The current ferrous metals conservation research at Cardiff includes further
investigation of desalination treatments for archaeological iron and study of clear
coatings for armoured steels alongside this project, all of which share the focus on
empirical evidence of treatment successes and generation of sector guidance for
effective decision-making and management.
9
1.4 Aim and objectives
The study aims to perform a cost benefit analysis of corrosion prevention treatments
for historic wrought iron including surface preparation methods and protective
coatings systems.
This will be achieved by:
 Engagement with practitioner methods, materials and decision-making

processes to tailor laboratory testing to produce fit-for-purpose, standardised
test methods.
 Generation of empirical data on the impact of surface preparation methods
and coating systems on high humidity corrosion rates of historic wrought iron.
 Relating practicalities to empirical data on treatment performance to
underpin effective and holistic sector guidance.
1.5 Thesis structure

The breadth of this study dictates its background context and underpinning theory is
equally broad and drawn from a range of disciplines. It is not the intention of the
thesis to explore in great detail all aspects of atmospheric corrosion or mechanisms
of protection by protective coatings. A selective approach is taken with corrosion and
coatings theory being reported and discussed as relevant to the context of historic
wrought iron corrosion and treatments. Chapter 2 presents an overview of the
principles of corrosion and specific mechanisms of atmospheric corrosion of iron and
offers an introduction to corrosion prevention by protective coatings. These
background themes give the context for the conservation problem presented by
wrought iron in uncontrolled exterior conditions.
Chapter 3 explores traditional and current practice in corrosion prevention for

historic wrought iron. Practitioner methods of surface preparation and application of
protective coatings link to theory in Chapter 2 and the critical sector problem of lack
of standardised guidance and empirical evidence to direct practice is explained. This
presents the context and rationale for the experimental study of surface preparation
10
methods and protective coatings which is detailed in Chapter 4. Central to this
chapter is the development of a method which not only answers the research
questions proposed here but is also standardised and applicable to other studies thus
encouraging further generation of empirical data. Results are offered in Chapter 5
and discussed in Chapter 6.
The discussion is intended to examine the empirical data generated and to fit that
into decision-making in heritage sector treatment of wrought iron. Empirical data is
balanced against practicalities in the context of specification, tendering, practice and
conservation ethics. Limitations of the study and scope for further work to enhance
the dataset are proposed. Conclusions are drawn in Chapter 7 regarding the insight
afforded by the results of this study and the contribution it makes to developing
evidence based management through data production and delivery of a standardised
method which could be used in further work and by other workers to increase the
breadth and depth of empirical data.
11
2 Corrosion Principles: Heritage wrought iron and its
contexts
This chapter is intended to introduce the concepts of corrosion and corrosion

prevention which dictate the problems and offer potential solutions in preservation
of historic wrought iron. An overview of corrosion and coatings theory is presented
followed by a discussion of the extrinsic and intrinsic factors that drive corrosion of
exterior wrought iron and impact on coating performance. These are contextualised
with examples of heritage objects and structures.
2.1 Principles of corrosion

2.1.1 The corrosion process
Metal extraction involves large amounts of energy to separate the metal from its
compounds within its ore. This elevates it to a higher energy state where it is
thermodynamically unstable relative to its lowest energy state, which is as an ion
combined with other elements (Trethewey and Chamberlain 1995, 69). Therefore
the natural tendency is for metals to oxidise, combining primarily with oxygen and
elements such as sulfur and carbon. Metallic corrosion occurs either as an
electrochemical process involving water and dissolved ions (Scully 1990, 55) or as
direct oxidation at high temperatures where moisture is not involved (Jones 1992,
407). Each metal has its own energy value, which is recorded in the electrochemical
series as a positive or negative potential relative to the standard hydrogen electrode
(SHE), with all readings taken in standard conditions of temperature, pressure and
ionic concentration (Jones 1992, 41). While this offers a guide to their
thermodynamic reactivity, with more negative potentials being more prone to
oxidation, it does not take into account kinetic factors such as solubility, conductivity
and the coherence of corrosion layers they develop in their service life.
Since the research reported here is based on wrought iron this is used to exemplify
the electrolytic corrosion process (Fig. 2.1). Electrolytic corrosion is a redox reaction.
12
Solid phase contact and electronic conductivity must occur between dissimilar metals
or areas of differing energy within the same metal to create a potential difference
that transfers electrons from high to low energy areas (Trethewey and Chamberlain
1995, 41). Electrons are generated by oxidation of the metal to its ionic form at an
anode and these are passed in the solid phase to a cathode, where they are
consumed by reduction reactions. Charge balance between anode and cathode is
completed by movement of ions in the aqueous phase.
The mix and concentration of ions in solution, pH and ambient gasses capable of
dissolving in the electrolyte dictate the range of corrosion products that can form.
These may be solids formed by precipitation or dissolved ionic species according to
the thermodynamic stability fields of the products. Corrosion products can have a
major impact on corrosion rate due to kinetic factors. Electrolytic corrosion is an all
or nothing process; if one part of the reaction sequence is prevented the whole
reaction ceases and this forms the basis of methodologies for corrosion prevention.
Thus the role of coatings is to block access of oxygen and moisture to the metal
surface and offer electrical insulation.
Figure 2.1 Electrolytic corrosion of iron.
13
2.1.2 Corrosion of iron: an overview
For iron corrosion the anode reaction produces ferrous ions, releasing two electrons
[1] and occurs readily with a standard electrode potential of -0.44V (Cornell and
Schwertmann 2003, 491). In aerated aqueous solutions over pH 4 to 14 the favoured
cathode reaction is the reduction of oxygen to produce hydroxyl ions [2] (Stratmann
and Mϋller 1994; Jones 1992, 7; Scully 1990, 55).
Anode Fe → Fe2+ + 2e [1]
Cathode O2 + 2H2O + 4e → 4OH – [2]
The ferrous and hydroxyl ions combine to form ferrous hydroxide [3] and further
oxidation will lead to the production of iron oxides. These have many forms and the
particular oxide produced will depend on factors such as oxygen availability, pH and
the range and concentration of dissolved ions in the electrolyte (Cornell and
Schwertmann 2003). This influences ongoing corrosion rate as thickness, coherence,
conductivity and density of corrosion products will influence oxygen access to the
metal surface and movement of ions away from cathode and anode sites. Coherent
layers can lead to insufficient oxygen ingress to support cathode reactions, charge
build up that reduces corrosion rate and the production of passivating layers on the
surface of the iron.
Fe2+ + 2OH – → Fe(OH)2 [3]
The abundance of hydrogen ions at acidic pH values makes the reduction of H + to

hydrogen gas possible [4] (Trethewey and Chamberlain 1995, 76). Reaction [2]
continues in such conditions but as oxygen concentration in the electrolyte falls the
high H+ concentration allows reaction [4] to dominate as the main reaction below pH
4 (Stratmann and Müller 1994; Scully 1990, 55). However, in most atmospheric
environments reduction of oxygen is the favoured cathode reaction unless
electrolytes are stagnant, oxygen access is restricted and hydrolysis of cations lowers
pH.
2H+ + 2e → H2 [4]
14
In terms of corrosion product build up delivering passivity to the metal,
predominance of reaction [4] is particularly damaging as many oxides are soluble at
low pH values (Cornell and Schwertmann 2003). This prevents the formation of any
passive oxide layer. Reduction to hydrogen must occur on a bare metal surface and
will cease if a coherent non-conducting oxide layer forms and prevents this access.
Additionally, the hydrolysis of Fe2+ ions [5] can lead to low pH values in the vicinity of
the anode, which will hinder the precipitation of solid corrosion products. Hydrolysis
reactions involving more oxidised Fe3+ are possible and these will also lower pH. In
pitting of stainless steel in 0.5M NaCl a pH of 4.8 has been recorded at anodic sites
(Suzuki et al. 1983) and, considering equilibrium constants, Turgoose showed this
was due to Fe2+ hydrolysis (Turgoose 1982a). Potential, diffusion within electrolytes,
delocalisation and precipitation of Fe2+ will influence how low pH values can become,
with stagnant environments inside corrosion pits offering ideal conditions for low pH
values.
Fe2+ + H2O → Fe(OH)+ + H+ [5]
Corrosion rate will relate to the nature, integrity and morphology of the corrosion
layers produced on surfaces and whether they exist as solids, since this influences
transfer of reactants to anodes sited on the metal surface and the movement of ions
from anodes to cathodes. Voluminous corrosion products will lack coherence and be
more permeable to gases and vapours than dense more evenly formed layers that
offer better oxygen and moisture ingress. Examining potential/pH diagrams reveals
that for many metals their corrosion products are soluble at low pH and therefore
only ionic species are the stable phases and no solid corrosion layers form to hinder
corrosion (Krauskopf 1982, 204). Kinetic properties of corrosion products influence
corrosion, as conducting corrosion products such as Fe3O4 can establish cathodes
remotely from the metal surface and support corrosion (Cornell and Schwertmann
2003, 115).
Corrosion may take the form of uniform or general corrosion, where a flux of energy
over the metal surface creates constantly changing anodes and cathodes, delivering
an even loss of metal over the entire surface (Jones 1992, 11), or fixed cathodes that
15
produce local corrosion, which has several forms including pitting (Trethewey and
Chamberlain 1995, 168) and crevice corrosion (Jones 1992, 12). Galvanic corrosion is
a specific form of corrosion where the energy between two differing metals drives
the process and the greater the potential difference between them the faster the
corrosion rate (Trethewey and Chamberlain 1995, 130; de Rooij 1989). The more
electronegative metal acts as the anode and is oxidised, while the less energetic
metal receives electrons and acts as the cathode which protects it from corrosion.
Galvanic corrosion appears to be ongoing and destructive but it ceases when a build-
up of insoluble non-conducting corrosion products is formed between the two
metals, as this prevents solid phase electron transfer. Both metals then corrode
individually, provided the conditions for electrolytic corrosion exist. There are
multiple causes of general and local corrosion including intrinsic factors such as
alloying and inhomogeneity, as well as external factors such as oxygen concentration
within electrolytes, these and other factors are addressed later in this chapter.
Atmospheric corrosion of clean metal surfaces is initially fast due to access of oxygen
and moisture to the metal surface feeding electrolytic corrosion but later slows if
insoluble corrosion products form that hinder access to the metal surface creating
passivity (Kucera and Mattsson 1987). Products from reactions [1] and [2] can react
with each other and other ambient ions to produce corrosion products, however
corrosion mechanisms are more complex than would appear.
The mechanism of rusting in the atmosphere caused by rain events is different to

that occurring in bulk aqueous systems, due to its three phase system (Dillmann et
al. 2004; Hoerlé et al. 2004). This involves a range of steps from initial reduction of
corrosion products through the anodic dissolution of iron to reoxidation of the
reduced iron species (Fig. 2.2) (Dillmann et al. 2004; Hoerlé et al. 2004; Maréchal et
al. 2007).
16
Figure 2.2 Corrosion process during wet/dry cycle phases (Maréchal et al. 2007).
In the wetting phase anodic dissolution of iron is balanced by reduction of ferric

species to less oxidised phases such as Fe3O4 (magnetite Fe2O3.FeO) and little oxygen
is consumed, as evidenced by the higher consumption of iron as compared to oxygen
in terms of balance between anode reaction [1] and cathode reaction [2] (Stratmann
and Streckel 1990). Once ferric reduction is complete, oxygen reduction becomes the
cathode reaction during the wet phase and balances iron oxidation. Finally, during
the drying phase, oxygen oxidises the reduced species formed in the first phase and
any new oxidisable species produced by corrosion to produce FeOOH. Drying
eventually stops corrosion. Reduction of γFeOOH in the wetting phase depends on
potential; Fe3O4 is formed from αFeOOH at potentials lower than -0.9V (SHE) and
γFeOOH at -0.5V (SHE) (Dunnwald and Otto 1989). The reduction of γFeOOH begins
at the metal surface and progresses outwards along pore walls (Cox and Lyon 1994).
Atmospheric corrosion typically produces a rust layer that contains some Fe3O4
within its inner layer and the more oxidised FeOOH (Fe2O3.2H2O) polymorphs form
its outer layer, with spaces within the magnetite layer being ‘plugged’ by the
formation of corrosion products produced by water and oxygen ingress through
cracked corrosion layers (Jones 1992, 404). This profile is almost universally observed
and fits a more detailed description of a dense inner layer of Fe3O4, amorphous
17
FeOOH overlaid by a looser crystalline layer of αFeOOH (goethite), γFeOOH
(lepidocrocite) and γFe2O3 (maghemeite) (Leygraf and Graedel 2000, 285). Overall, it
also aligns well with the mechanism occurring during wetting and drying cycles
described by Hoerlé et al. (2004). Green rusts ([Fe2IIFeIIIOX(OH)y](7-2x-y)+ and
[Fe2IIFeIIIOX(OH)y](5-2x-y)+) also occur within corrosion layers but these Fe(II) – Fe(III)
compounds are unstable and are intermediates that transform to more oxidised
species when exposed to oxygen (Refait et al. 1998; Leygraf and Graedel 2000, 284;
Cornell and Schwertmann 2003). Characterisation of medieval wrought iron
corrosion products on wrought iron bars corroded inside Amiens cathedral has
identified ferrihydrite (5Fe2O39H2O; nominal formula) and feroxyhydrite (δFeOOH),
which are less crystallized phases (Monnier et al. 2007). Many other corrosion
products can occur but this overview records the most commonly occurring products.
Specific environments such as sulphate and chloride rich contexts will influence
corrosion product formation and are discussed later.
While a simple stoichiometric equation for rusting can be offered [6] not all oxygen
consumed can be directly linked to loss of iron as the wet/dry reaction phases involve
reduction of oxides and their later re-oxidation. However, given that the reactions
occurring on rusting iron surfaces will follow similar reaction routes, the principle of
using oxygen consumption for quantitative comparisons between corrosion rates is
viable. This is described in 4.1.3, 4.2.5 and 4.2.6. Corrosion mechanisms and routes
are examined in more detail in section 2.3.
4Fe + 3O2 + 2H2O → 4FeOOH [6]
2.2 Corrosion prevention by protective coatings

Reducing the rate of atmospheric corrosion of metals is achievable by modifying the
environment to make it less corrosive, by exploiting electrochemical principles to
suppress anodic or cathodic reactions or by separating the metal from the corrosive
environment and introducing a large resistance to impede ion transport between
anode and cathode (Scully 1990, 130; Mayne 1954, 286). It is clear from practitioner
18
guidance and feedback that application of organic coatings, effectively separating the
wrought iron from its exposure environment and exploiting electrochemical
principles, is the most common approach.
Protective coatings are a complex combination of materials. Coating systems are

formed of one or more layers (coats): the primer is applied to bare metal; the last
coat is the finish or topcoat; any coats applied between are intermediate or build
coats (Trethewey and Chamberlain 1995, 304).
Component Purpose
Vehicle Liquid giving fluidity during application and drying or curing to

a solid film (see Binder) by evaporation of solvent, chemical
change (e.g. oxidation) or polymerisation of vehicle and curing
agent (two-pack coatings).
Pigment Solid particles suspended in the vehicle. They may inhibit
corrosion or corrode in preference to the substrate metal
when in primers. In topcoats inert pigments lengthen diffusion
pathways for oxygen/water to delay corrosion and slow
reaction rates.
Additives/fillers Accelerate drying/curing of the coating or impart desirable
properties to the paint film.
Binder Solid portion of the vehicle once dried/cured. Holds pigments,
keys paint film to the substrate and functions as a barrier to
oxygen, water and aggressive ions.
Table 2.1 Components of paint coatings (Trethewey and Chamberlain 1995, 304-
305).
2.2.1 Mechanisms of protection

Corrosion prevention is one of the least well understood and most unpredictable
properties of organic coatings but there are three main mechanisms by which it
occurs (Bierwagen 1996, 44):
i) Barrier protection (transport of aggressive species to the metal surface is

impeded by coatings with low permeability for gases, liquids and ions);
19
ii) Cathodic/sacrificial protection (sacrificial corrosion of an
electrochemically less stable metal);
iii) Inhibitive/passivation protection (by a chemical conversion layer or
inhibitive pigments within the coating).
Barrier coatings typically contain inert pigment (titanium dioxide, micaceous iron
oxide or glass flakes) at lower pigment volume concentrations which result in dense,
cohesive coatings displaying very low permeability to aggressive species (Hare 1979).
It has long been recognised (Bacon et al. 1948; Kittelberger and Elm 1952; Dickie and
Smith 1980; Thomas 1991) that the mechanism of protection by barrier coatings is
via ionic impermeability rather than inhibition of moisture permeation. Common
electrolyte ions are Cl-, SO42- and a mixture of cations such as Na+, K+, NH4+ and Ca2+
(Bierwagen and Huovinen 2010, 2647). Ionic impermeability ensures moisture at the
metal surface has a high electrical resistance to minimise corrosion (Sørensen et al.
2009, 142). A typical barrier coating is a cross-linked epoxy matrix with aromatic
groups forming an effective barrier to water, oxygen and ions and strong adhesion
to the metal substrate promoted by many secondary hydroxyl groups (Rouw 1998,
181). In the majority of coating systems the topcoat is the layer whose primary
function is to act as a barrier and the best performers in this role are hydrophobic
and resistant to photooxidation (Bierwagen and Huovinen 2010, 2467-2468).
Sacrificial coatings are applied as primers. The sacrificial, more electrochemically

active metal must be in electrical contact with the substrate. Zinc-rich primers offer
sacrificial protection as anodic zinc will corrode in preference to cathodic iron as long
as the system is conductive and sufficient zinc remains within the primer layer
(Sørensen et al. 2009, 143). Penetration of water through the binder corrodes zinc to
zinc hydroxide which further reacts with carbon dioxide to form zinc carbonate, filling
pores in the coating and producing an impervious, compact and adherent protective
layer (Kalendová et al. 2015; Trethewey and Chamberlain 1995, 305). If the primer
layer is damaged, zinc corrosion products seal the pores between zinc particles, the
system becomes electrically non-conductive and the corrosion products act as a
barrier (Feliu et al. 1989a; Feliu et al. 1989b). Adhesive and cohesive strength of zinc-
rich primers may be reduced by high volume concentrations of zinc pigment, and
20
consequently lower percentages of binder, required to provide electrical
conductivity (Kalendová et al. 2015; Greenfield and Scantlebury 2000).
Inhibitive coatings applied as primers passivate the substrate and build up a

protective layer of insoluble metallic complexes which are a barrier to transport of
aggressive species to the substrate (Sørensen et al. 2009, 144). Inhibitive pigments
are inorganic salts (e.g. phosphates, chromates, molybdates, nitrates, borates and
silicates) which are slightly soluble and carried to the substrate surface by
permeating water where they react with the metal to form a passivating product
which inhibits corrosion (Rammelt and Reinhart 1994, 304). Efficiency of inhibitive
pigments depends on barrier properties of the coating; a degree of coating
permeability triggers the effects of inhibitive pigments whereas with low
permeability coatings barrier pigment effects will be predominant (Liu 1998, 576).
Coatings are frequently applied as systems. Careful selection of properties for each
coating within the system maximises corrosion prevention performance.
Hydrophobicity and UV resistance are desirable for topcoats, high cross-link density
and adhesion to substrate and each other for intermediate/build coats and primers
(Bierwagen 1996, 45).
2.2.2 Factors affecting coating performance and failure

Protective coatings fail when they no longer prevent corrosion of the coated
substrate, when coating damage exposes substrates to uninhibited corrosion and
when coating appearance is degraded to an unsatisfactory level by corrosion and
weathering (Bierwagen 1996, 44). A failure of the coating will concentrate corrosion
at that point, set up differential aeration effects beneath the paint film and further
exacerbate corrosion (Trethewey and Chamberlain 1995, 307).
The main features necessary for successful protective coating are (Bierwagen 1996,
45; Sørensen et al. 2009, 137; Bierwagen and Huovinen 2010, 2646):
 Wet/dry adhesion which maintains the coating contact with the substrate and
prevents damage by osmosis;
 Low ion (Cl-, SO42-, Na+, K+, NH4+, Ca2+), water and oxygen permeability;
21
 Low conductivity preventing flow of current in local corrosion cells and ion
and electron motion in the film;
 Stability of the polymer within the environment (hydrolysis, UV radiation,
extremes of temperature);
 Strong absorption of coating polymer/substrate interface and good wetting
of the substrate surface by the coating;
 Pigmentation for barrier and sacrificial protection mechanisms;
 Flexibility and toughness to withstand impact and cracking;
 Maintenance of appearance under stress, swelling, mechanical forces and
weathering.
The range of factors affecting protective coating performance is vast and includes the
composition of the coating (binders, pigments, solvents, extenders and additives),
substrate types, substrate pretreatment or preparation level, curing parameters,
coating thickness, adhesion of the coating to the substrate surface and various
environmental factors (Sørensen et al. 2009, 137). Almost all paint media exhibit a
degree of permeability to water and oxygen but the introduction of solid pigment
and filler lengthens the diffusion pathway and makes this more convoluted (Ross
1977, 28). Increasing thickness of barrier coatings has been shown to reduce the
delamination of defect free and intentionally damaged coatings on hot rolled steel in
long term exposure tests whereas extent of surface preparation was found to have a
surprisingly limited effect (Keane et al. 1969, 372; Steinsmo and Skari 1994, 934)
despite assertions to the contrary (British Standards Institute 2007; Leighs 2008a;
Ross 1977, 29). Pinholes in a coating (holidays) are particularly problematic as the
coated surface presents a large cathode to a small anode (exposed substrate at the
holiday) causing high corrosion rates (Trethewey and Chamberlain 1995, 308).
Adhesion is a key attribute for successful performance of a protective coating and
can be maximised by surface preparation with clean metal surfaces or metal oxides
ideal for wetting due to their high surface energies (Bierwagen and Huovinen 2010,
2655).
Environmental conditions during application affect coating success. Moisture and

low temperatures interfere with adhesion and completeness of cover and moisture
22
under the coating initiates corrosion (Ross 1977, 31). Where coating/substrate
adhesion is poor, salt solutions can form between water diffusing through the coating
and soluble corrosion products or compounds from the coating; subsequent osmotic
pressure drives faster water diffusion and causes blistering and delamination with
pressure rupturing the substrate/coating bond (Fig. 2.3 and 2.4.) (Trethewey and
Chamberlain 1995, 321; Greenfield and Scantlebury 2000).
Figure 2.3 Development of

blisters in poorly adherent
paint films owing to osmotic
pressure (after Trethewey and
Chamberlain 1995, 321).
Sufficent liquid
Alkalinity at
containing Osmotic or
Corrosion cathodes reduces
chloride ions endosmotic
Paint film imbibes proceeds at coating adhesion
passes to passage of water
water, possibly anodic sites with and produces
substrate metal through coating
with dissolved consequent osmotically active
initiating corrosion from exterior
salts. hydroxyl ion build- substances at the
at the environment to
up at cathodes. metal/coating
metal/coating substrate.
interface.
interface.
Figure 2.4 Suggested sequence of blister formation (Greenfield and Scantlebury

2000).
23
Small failures in coatings may lead to exponential corrosion rates. A small blister in a
coating or corrosion at a pinhole holiday causes a small area of coating failure which
allows further ingress of oxygen, water and ions. A small anode and large cathode
leads to aggressive corrosion, further spalling of paint layers, ingress of corrosive
species, undermining of coatings and exposure of more substrate material for
corrosion.
The effects of coating failure are seen clearly in Figure 2.5. Blistering, disbondment
and lifting of coating layers has opened channels for the ingress of water, oxygen and
aggressive ions. The effects of the coating delamination and exposure of the
underlying substrate to uncontrolled atmospheric corrosion can be seen.
Laminations of corrosion products open yet more channels for ingress of corrosion
accelerants and this further corrosion results in devastating loss of metal section.
Techniques preparing the surface of these gates for recoating will remove loose
corrosion products and the extent of loss of original material will become clear.
Figure 2.5 Wrought iron gates at the Palace of Holyroodhouse showing coating
failure and laminating corrosion of underlying substrate.
24
2.3 Corrosion variables
The interdependent nature of the processes involved in electrochemical corrosion
mean that if one part of the reaction ceases the whole process stops; if there is no
water there cannot be movement of ions in the aqueous phase; loss of solid phase
electron conducting contact between anode and cathode prevents transfer of
electrons. Ambient environment controls availability of the aqueous phase and
factors such as the build-up of non-conducting corrosion products between anode
and cathode control electron movement. The variables that facilitate, drive and
control corrosion are reviewed to offer an overview of corrosion, how coatings can
prevent this and how they begin to fail in this task.
2.3.1 Water
Moisture in the atmosphere is measured as relative humidity (RH), which is the ratio
of moisture in the atmosphere to the amount of water it can hold before dew point
occurs (Thomson 1986, 68). This is a temperature dependent function; warmer air
can hold more moisture before dew point. At sufficiently high RH, termed critical RH,
enough moisture is adsorbed onto a metal surface to provide a continuous layer that
can support electrolytic corrosion with early work determining this to be in the region
of 60%-70% RH (Vernon 1935).
Electrolytes are formed by soluble ionic contaminants on the metal surface or

gaseous pollutants that dissolve in the film. Generally corrosion is considered to be
negligible where monolayers are only 1 to 2 layers thick but adsorbed water films are
2 to 5 layers thick at 60% RH and corrosion begins to escalate and becomes significant
at 80% RH, where the layers are 6 to 10 layers thick (Leygraf and Graedel 2000, 11;
de Rooij 1989). At the critical humidity of 60% RH, water layers are likely to be
clustered rather than continuous, which will influence the location of corrosion
(Leygraf and Graedel 2000, 283). Water layers begin to form on polished metal
surfaces around 55% RH (Garverick 1994, 5). Capillary condensation has a part to play
in corrosion. Small capillaries of 1.5nm will condense water at 50% RH (Garverick
1994). Capillary condensation occurs in pores at a vapour pressure below 100% RH
and occurs for water when capillaries approach 10-5 cm (Bikerman 1970, 309).
25
Consequently, corrosion product layers and dust particles can influence the
development of water films on metals and lower the critical humidity for corrosion
to commence. This could be particularly influential for corroded surfaces beneath
failed paint films if capillary size within corrosion layers is small and soluble ions,
hygroscopic salts and hydrated oxides are present. Corrosion couples between
differing metals vary considerably according to RH; Cu/Fe couples (µAcm-2) are six
times greater at 100% RH as compared to 75% (de Rooij 1989). Time of wetness on
metal surfaces has long been known to influence occurrence of corrosion (Vernon
1931).
If corrosion is to occur in adsorbed water layers dissolved salts are necessary to form
electrolytes. Uncontaminated surfaces of shiny clean iron were found not to corrode
even at 95% RH (Cai and Lyon 2005). A lower RH corrosion threshold value can be
achieved in the presence of hygroscopic soluble salts on the metal surface, as they
attract water to support electrolytic corrosion and at higher RH can deliquesce to
provide strong electrolytes (Chandler 1966). Sodium chloride (Evans and Taylor 1974)
and ferrous chloride (Turgoose 1982b) are two typical salts with the ability to lower
the critical RH for corrosion. Solids such as carbon and ash can contain impurities
such as sulfur and also increase the thickness of water layers and promote corrosion
at lower RH values than might be expected (Scully 1990, 103). Chandler (1966) found
that marked corrosion of steel occurred at 52% RH in the presence of chlorides or
sulphates but was inappreciable below 42% RH, with concentration of dissolved ions
determining the rate of corrosion rather than the amount of water uptake by the
metal. This is below the accepted 60% critical humidity for corrosion of ferrous
metals. Clearly, both RH and dissolved salts are important factors when considering
the corrosion of metals.
2.3.2 Climate
Climate has a strong part to play in any corrosion cycle for iron or any other metal
(Sereda 1974; Hayrie 1982); outdoors by direct action and indoors by influencing
temperature and RH. Annual humidity patterns, location, temperature and pollution
all play a part in the corrosion process and this is reflected in the choice of test site
location across the world and the data sets produced at them (Hudson and Stanners
26
1953). Rain events produce wetting and drying cycles on the metal surface. These
involve; an initial wetting phase forming a thin water film quickly followed by a
saturation stage that supports a thicker film of water, which remains in place during
the rain event and shortly after, then finally a drying phase with a thinning water film
until it finally disappears (Hoerlé et al. 2004). Corrosion is fastest in the drying cycle,
as the surface of the metal retains a thin water layer for some time and this supports
the ready solvation of atmospheric oxygen and its diffusion through the layer to
cathodes [2] (Leygraf and Graedel 2000, 13). In contrast, during the saturation phase
the diffusion path for oxygen is longer due to the thick continuous water films and
will affect corrosion rate (Leygraf and Graedel 2000, 286). The impact of climate on
corrosion is clear, as many short rain showers with high humidity, low wind and
limited sun will produce multiple drying cycles with slow drying that prolongs the
aggressive final drying phase. Temperature mostly influences corrosion by its impact
on RH values and thus availability of water (Sereda 1974).
Pollutants and contaminants can play an important role in the corrosivity of rain.
Corrosion of the cables of the Williamsburg Bridge in New York City was credited to
the impact of acid rain that was then prevalent in the New York atmosphere and to
chloride originating from the river and de-icing salts thrown into the air by vehicular
traffic (Eiselstein and Caliguri 1988). Additionally, intrinsic factors were also
contributors, as the graphite contained in a slushing oil intended to protect the cables
of the bridge was cathodic to the steel cable and offered a vast reaction area leading
to local corrosion where the coating failed to cover the cable. This integrated attack
on the metal reveals the complexity of predicting corrosion as climate, location,
function, coating choice and contractor standards all influenced the corrosion
pattern and rate. Similarly, rusting steels designed to develop protective rust coatings
performed well in dry environments and in even wet/dry cycles but in stagnant wet
conditions the formation of non-protective rust layers allowed high humidity,
condensation and chloride contaminants to develop severe corrosion beneath them
(Raman 1988).
27
2.3.3 Oxygen
Oxygen solubility and its diffusion to cathode sites is important in controlling rate of
reaction. Solubility of oxygen in water at atmospheric pressure and 25oC is 2 x 10-4
mol l-1 in equilibrium with pO2 of 0.2 atm-1 (Stratmann and Müller 1994), which
decreases with increasing temperature until freezing prevents corrosion, as there is
no electrolyte. A strong electrolyte may enable corrosion to occur at low
temperatures by its ‘de-icing’ effect of the dissolved salts lowering the freezing point
of water. The porosity of corrosion product layers or protective coatings can be
influenced by saturation with water, as this reduces the diffusion rate of oxygen to
anode sites on the metal surface and likely favours the formation of larger quantities
of less oxidised corrosion products such as Fe3O4.
Consequently, extended rain events may not be as negative as might be expected. As

well as fewer drying cycles the micro-pores within coatings will be saturated with
water and oxygen has to both dissolve and diffuse within this system, whereas dry
coatings allow for better gas and vapour permeation and multiple short rain events
offer scope for wet/drying cycles that are aggressive. High humidity may also lead to
pores being saturated by capillary condensation. A worst case scenario is likely the
retention of moisture beneath a dried and damaged coating, allowing easy oxygen
ingress to feed the cathode reaction. Atmospheric humidity is important for the
delivery of electrolyte and diurnal cycles of temperature can have a significant
influence on RH and thus availability of electrolyte. Night temperatures normally
raise RH and cool metal surfaces producing a microclimate in their vicinity, which
may lead to condensation when dew point is reached promoting corrosion and
offering media for better solvation of atmospheric gases.
2.3.3.1 Oxygen starvation corrosion: rain and condensation

Rain events can also lead to rapid local corrosion on iron and develop pitting
(Trethewey and Chamberlain 1995, 169). This occurs by oxygen starvation occurring
in the centre of a water drop created by the high surface tension of water (Fig. 2.6).
Initially the spot is oxygenated throughout but the flux of energy over the metal
surface under the spot leads to general corrosion and multiple cathodes rapidly
deplete oxygen. Since oxygen replenishment within the water drop is from its
28
exterior this develops an oxygen rich outer layer leaving an oxygen starved inner
layer. Oxygen starvation produces an anode at the centre of the spot [1], with
cathodes sited at the oxygen rich outer layer that use up oxygen before it can diffuse
into the drop centre [2]. Hydrolysis reactions, such as that of Fe2+ at the anode [5],
lead to low pH in the centre of the spot hindering formation of solid corrosion
products. Migration of Fe2+ from the anode leads to their reaction with cathodically
produced OH- precipitating Fe(OH)2 [3] in the vicinity of the cathode, further
oxidation of this hydroxide produces iron oxides. This creates a ring like corrosion
pattern with an anode at its centre leading to pitting corrosion, which can produce
quite large and wide pits that can merge into a more general corrosion layer.
Figure 2.6 Oxygen starvation in a water drop on the surface of iron.
2.3.3.2 Oxygen starvation corrosion: pitting

Localised pitting corrosion can occur when inhomogeneity or particulate deposition
creates fixed cathodes. Typically, it involves a corrosion pit containing water with a
low oxygen concentration in its base and a low pH due to hydrolysis of metal ions
(Galvele 1976) such as Fe2+ hydrolysis [5]. The low pH prevents formation of solid
corrosion products and provides a large positive charge in the pit base. To
29
counterbalance this, anions are pulled into the pit with chloride being the most
common ion due to its high solubility and mobility. Electrons from the anode travel
in solid phase to the pit mouth where an abundance of oxygen and availability of
electrolyte creates anodes and oxygen is reduced there [2]. Migration of Fe 2+ and
Fe(OH)+ towards the pit mouth oxidise to produce ferric ions [7] and [8], which can
further oxidise to produce solid corrosion products whose oxidation state depends
on the amount of oxygen present. In less oxidising conditions magnetite (Fe 3O4)
forms [9] and in more oxidising environments ferric oxyhydroxide (FeOOH) is the
product [10]. A mixture of both at the pit mouth can be expected and this explains
the appearance of rust blisters there.
Fe2+ + ½O2 + 2H+ → Fe3+ + H2O [7]
2Fe(OH)+ + ½O2 + 2H+ → 2Fe(OH)2+ + H2O [8]
2Fe(OH)2+ + Fe2+ + 2H2O → Fe3O4 + 6H+ [9]
Fe(OH)2+ + OH- → FeOOH + H2O [10]
Acidity is exacerbated by the vast coastline of Britain, as chloride will be readily

available in many areas of the country. This produces an Fe2+/Cl- solution with
hydrolysis of Fe2+ [5] producing low pH values and association of H+ with Cl- producing
HCl (Leygraf and Graedel 2000, 46) to keep corrosion products in a soluble form and
maintain a low oxygen environment. The aqueous system within the pit allows Fe 3+
to form complexes with Cl- and mixed complexes containing Cl- and OH-, also drying
in oxidising conditions with high Cl- concentrations will lead to precipitation of basic
oxychlorides such as β-Fe2(OH)3Cl, which is a precursor to the formation of βFeOOH
(Refait and Génin 1997). Once the pit dries out corrosion ceases but can be
reactivated later by water producing electrolyte and with a low pH as the ingredients
for this are in place.
Pitting can occur where pinhole breaks and holidays are present in coatings and can
be particularly damaging if the integrity of an iron sheet is breached by small pits
penetrating its thickness. Within heritage contexts, this may or may not be a
preferable form of corrosion depending on circumstance; disastrously, a functioning
30
tank will lose its gas pressure or water tightness and be unable to fulfil its purpose,
whereas localised loss of an historic paint layer by pitting may be preferred to overall
loss of large areas by general corrosion, where evidence of the paint layer and surface
technology or decoration would disappear entirely. Clearly, context influences the
impact of the corrosion type. Further to the loss of the historical record are the
aesthetic considerations for heritage objects, which can create a dichotomy of
interest when the aim is to develop optimum preservation strategies.
2.3.3.3 Coatings: cathodic disbondment and oxide lifting

Since this dissertation is centred on surface preparation and coating performance on
heritage wrought iron, cathodic disbondment and oxide lifting are important
problems to consider, as they are associated with corrosion through damaged films
(Jones 1992, 487) (Fig. 2.7). Cathodic disbondment involves hydroxyl ions generated
by the cathode reaction [2] beneath damaged paint films being solvated and reacting
with particular types of paint films, such as oils, to saponify them. This causes
separation at the metal/paint interface at the macroscopic or microscopic level,
which aids further oxygen and moisture ingress to support corrosion.
Figure 2.7 Possible consequences of coating damage (Greenfield and Scantlebury

2000).
31
Oxide lifting is summarised by Jones (1992) as lifting and undercutting of paint films
by compacted oxides that have dried and formed a lamellar structure. The oxide
layers have a large volume. Fully oxidised iron as FeOOH forms the outer layer, while
next to the metal surface ferrous species Fe2+ and Fe(OH)2 are favoured in the low
oxygen environment there, as is the formation of the partially reduced oxide
magnetite (Fe3O4) and this is overlaid by the FeOOH forming a more oxidised outer
layer due to a supply of incoming oxygen (Jones 1992, 489). These layers are formed
by alternate wetting and drying cycles which impact on hydration of oxides, oxygen
availability and the volume of the corrosion products. The Fe3O4 can be reoxidsed by
ingress of oxygen to form FeOOH and the differing densities between these products
means volume changes are large and frequent as a function of rain events and access
to the metal surface. This is essentially the same as the atmospheric corrosion
mechanism in oxygen rich environments (Dillmann et al. 2004; Hoerlé et al. 2004),
where lepidocrocite reduction is the cathode reaction in the stage 1 wetting phase
of the reaction process. Additionally Fe3O4 can act as a cathode due to its conductivity
(Cornell and Schwertmann 2003, 115) and this facilitates further corrosion when the
corrosion products are in a wetting cycle. In the drying phase corrosion products are
deposited and these do not redissolve with a new wetting phase, thereby building
lamellar corrosion which is familiar on corroded iron.
2.3.3.4 Coatings: filiform corrosion

Filiform corrosion is a characteristic form of corrosion occurring beneath coatings on
iron but it results in only limited metal loss. The mechanism involves ingress of
oxygen and moisture at a pinhole break in a coating and corrosion begins beneath,
with an active head that is anodic to the metal, due to it being de-aerated by using
up oxygen that cannot be easily replenished. The head produces Fe2+ ions [1] and this
is followed by an aerated tail where the cathode reaction occurs [2], producing an
inactive thin corrosion product trail as the head moves forward due to surface
tension effects from the OH- ions. A cracked and ruptured paint film results (Fig. 2.8)
(Jones 1992, 226). Secondary problems arise from damaged films, as ingress of
moisture and pollutants through the damaged film may set up more permanent
corrosion centres that eat into the iron. The general degradation of the protective
32
properties of the film and the corrosion that may result from this are much more
problematic than the filiform corrosion itself, which is unsightly rather than severely
damaging. Reducing the protective properties of the coating opens pathways for
further corrosion.
Figure 2.8 Schematic of filiform corrosion (Jones 1992, 226).
2.3.4 Intrinsic factors

The intrinsic properties of wrought iron will also influence its corrosion resistance
and the nature of the corrosion that occurs. Alloying, grain structure and working of
metals impact on occurrence of corrosion type and rate. Manipulation of modern
ferrous alloy composition combats corrosion in alloys such as weathering steels,
where intentional corrosion produces a self-healing rust layer (de Meybaum and
Ayllon 1982; Knotkova-Cermakova et al. 1982). For wrought iron, alloying is less
important than the quantity and distribution of heterogeneous inclusions (Dillmann
et al. 2004) such as slag impurities, as the only alloying element is a small amount of
carbon (0.03% maximum) and the grain structure is normally pure ferrite (Rawdon
2013, 3; Dillmann et al. 2004). Slag inclusions in ancient wrought iron are mainly
faylite (Fe2SiO4), wüstite (FeO) and glass like inclusions (Neff et al. 2005). The amount
of slag varies widely but it can act as a cathode and imperfections at slag/iron
boundaries facilitate movement of ions (Neff et al. 2005). Piling during manufacture
can cause lamellar corrosion that follows slag planes present from manufacturing
33
(Fig. 2.9) (Chilton and Evans 1955; Higgins 1957, 36; Greaves and Wrighton 1967, 82-
83; O’Sullivan and Swailes 2009, 261 and 266). Both pitting and general corrosion are
evident on the wrought iron plate in Figure 2.13.
Wrought iron has a reputation for good corrosion resistance and there is support for
this in given contexts and relative to the orientation of corrosion in the iron (Chilton
and Evans 1955). Overall general conclusions cannot be drawn on performance, as
the quality of the wrought iron is crucial to its corrosion resistance. Historically it
varies. At the height of wrought iron technology metal from the American Civil war
ironclad USS Monitor was mediocre quality comprising a low carbon, high
phosphorous ferrite with 4.8% silicate slag comprising phosphor and olivine, glass,
wϋstite and silica (Boesenberg 2006). These considerations need to be taken into
account for experimental design and conclusions in this dissertation.
Wrought iron may also contain manganese, sulfur and considerable amounts of
phosphorous (Rawdon 2013, 9). Manganese sulphide inclusions can occur in steels
and these are strongly cathodic and will lead to local corrosion and pitting (Trethewey
and Chamberlain 1995, 172). This potentially has other consequences, as acidity from
hydrolysis [5] and [9] can dissolve the MnS and resulting S2- and HS- decrease
activation polarisation on the iron, increasing its corrosion, and the low pH may
support hydrogen reduction as the cathode reaction [4] with evolution of H2
damaging the overlying oxide layer.
Figure 2.9 Lamellar corrosion of early wrought iron railway wedge.

34
2.3.5 Pollutants and contaminants
2.3.5.1 Sulfur dioxide pollutants
Sulfur dioxide can cause very significant corrosion of iron by the formation of sulfur
nests (Jones 1992, 406; Leygraf and Graedel 2000, 289; Evans and Taylor 1972). These
arise from the deposition of sulfur dioxide and its oxidation to sulphate and sulfuric
acid lowering pH on the surface of metals. SO2 is soluble in water with about 120g/Kg
at 20oC falling to just over 50g/Kg at 40 oC and this facilitates many deposition routes.
This enables SO2 to dissolve in the damp atmosphere or adsorb onto aerosol particles
allowing it to deposit in a dissolved state, often oxidised to SO42-, also dry deposition
by solvation in adsorbed water on the metal surface is possible and SO2 reactivity
with the hydroxyl radical means gas phase deposition is possible as sulfuric acid
(Leygraf and Graedel 2000, 45). Water availability clearly has an important role to
play in the impact of SO2 on corrosion and correlations between SO2 and water have
been identified (Sydberger and Vannerberg 1972). Average corrosion rates for steels
in the atmosphere, as a function of average SO2 concentrations at three sites, clearly
showed increasing corrosion as a function of both RH and SO2 concentrations and it
is suggested that corrosion in the presence of SO2 should not be a problem when RH
is below 30% (Hayrie and Upham 1974).
Winter is a more aggressive period for SO2 as it is more abundant from fossil fuel use,
has higher solubility at lower temperatures and winter climatic factors influence its
distribution (Leygraf and Graedel 2000, 288; Mayne 1959). While sulfur was a major
problem in the industrialised world, clean air policies and reduced heavy industry in
Western Europe has led to reduced outputs since the late 1960’s. For industrial
heritage wrought iron the presence of sulfur may also be related to function, where
historically equipment involved in processes such as smelting are exposed to high
levels of SO2 from the fuels involved.
The sulfuric acid produced can dissolve oxides opening up the corrosion layer to
oxygen ingress and forming FeSO4 which is soluble and hydrolyses to reduce pH
further and open up the oxide layer more (Jones 1992, 405). Hygroscopicity of solid
FeSO4 and its multiple hydration states means RH corrosion thresholds for iron can
35
fall (Rimmer and Watkinson 2010) and rust layers containing SO42- have been shown
to have reduced critical humidity for corrosion to 42% RH (Chandler 1966)
The mechanism for corrosion of iron by sulfur has been studied and an
electrochemical cyclic corrosion process gradually deposits rust (Evans and Taylor
1972; Evans 1981, 109; Jones 1992, 405) [10], [11] and [12]. Fe2+ is generated at the
anode [10] and since the pores of the corrosion layer are full of electrolyte, Fe3O4 is
generated on the outer surface of the existing Fe3O4 fed by oxygen [11] (Jones 1992,
405). Cathodic reaction involves reduction of FeOOH to Fe3O4 [12] and this is later re-
oxidised by inward migration of atmospheric oxygen [13] (Evans and Taylor 1972).
Density changes will cause disruption to the corrosion product layer (FeOOH 4.26 and
Fe3O4 5.18) (Cornell and Schwertmann 2003, 5). Some insoluble sulphate precipitates
and reaction will eventually cease if there is no SO2 replenishment. The mechanism
does not require diffusion of oxygen to the metal surface and gives rise to linear
corrosion rates, as diffusion through oxides and electrolyte is not necessary (Jones
1992, 400). Leygraf and Graedel (2000, 289) support this reaction process and
indicate that a range of mixed oxidation state hydroxysulphates can also exist in the
base of the nest.
Anode Fe → Fe2+ + 2e [10]
3Fe2+ + 2OH- + 1.5O2 → Fe3O4 + H2O [11]
Cathode 8FeOOH + Fe2+ + 2e → 3Fe3O4 + 4H2O [12]
3Fe3O4 + 0.75O2 + 4.5H2O → 9FeOOH [13]
An acid regeneration cycle was suggested before the electrochemical model above
(Schikorr 1963). This may also take place as it involves loss of sulphate from the
reaction solution and this is seen to occur in small amounts as the electrochemical
process continues [14].
2FeSO4 + ½ O2 + 3H2O → 2FeOOH + 2H2SO4 [14]
Sulfur dioxide also influences performance of protective coating systems. It has been
suggested (Mayne 1959) that soluble deposits of ferrous sulphate short-circuit
36
resistance of coatings and that given the permeability of coatings to water and
oxygen the ferrous sulphate becomes oxidised and hydrolysed to voluminous
corrosion products which rupture the coating film.
2.3.5.2 Chloride
Chloride is a major contaminant for buried archaeological wrought iron (Watkinson
2010; Turgoose 1982a) but it can also be deposited from the atmosphere. It is highly
soluble and has a high transport number in water making it an ideal anion for
electrolytes. It travels inwards to anodes at the metal surface to counter balance the
charge produced by the Fe2+ ions, where its high solubility makes it a good electrolyte
and corrosion rate can be expected to increase as chloride concentration rises
(Chandler 1966). Additionally, depending on its concentration it can dictate corrosion
product formation and produce corrosion products such as βFeOOH that promote
corrosion due to the aqueous mobility of its adsorbed surface chloride (Turgoose
1982a and b; Watkinson and Lewis 2005a). Other chloride compounds can form and
are precursors to the formation of βFeOOH and these include Fe2(OH)3Cl (Neff et al.
2007). The coastal nature of the UK and location of towns and cities along the coast
mean a considerable amount of heritage wrought iron will be exposed to elevated
levels of chloride from marine sources. Chloride concentration falls rapidly moving
inward from the coast but severe storms may take it 15km inland (Syed 2006) and
lower RH facilitates further ingress of sea salt particulates from coastal areas
(Gustaffson and Franzén 2000).
Hygroscopic salts can lower the corrosion threshold below the RH at which water
films are deposited on metal surfaces. By attracting water to themselves they can
provide water for electrolyte activity and their solubility can provide the ions for this
to take place. Corrosion in the presence of sea salt is well documented (Evans and
Taylor 1974) and hygroscopic salts such as magnesium chloride or corrosion products
such as ferrous chloride can dramatically lower corrosion thresholds.
37
2.4 Corrosion products and their properties
2.4.1 The importance of corrosion products
Corrosion products and the mechanism of their formation can be wide ranging and
complex, since many factors have the potential to influence which products form
including; oxygen availability, moisture levels, acidity, alkalinity, temperature and
availability of other ions. Their growth on iron can reduce corrosion rates to a level
considered to be passive but this relies upon their continuity, morphology, porosity,
solubility and whether they can conduct ions or electrons. Conducting corrosion
products can support remote cathode reactions and facilitate anodic dissolution
where oxygen does not access the metal surface. Pourbaix or potential/pH diagrams
offer a thermodynamic guide to the formation of corrosion products by plotting their
stability fields for specific concentrations of dissolved species, pH and redox potential
(Fig. 2.10) (Pourbaix 1977). Introducing other ions such as chloride and elements such
as sulfur or gases such as carbon dioxide to the aqueous system will influence
formation of corrosion products. A brief selective review of the properties of the
most common atmospheric corrosion products relevant to this study is offered.
Figure 2.10 Potential pH diagram for the iron water system at 25oC, 1 atmosphere
pressure and 10-6 dissolved species. (Licensed under CC BY-SA 2.5 via Commons -
https://commons.wikimedia.org/wiki/File:Pourbaix_Diagram_of_Iron.svg#/medi
a/File:Pourbaix_Diagram_of_Iron.svg )
38
2.4.2 αFeOOH (Goethite)
Generally the range of oxide hydroxide polymorphs have low solubility and large
surface areas that make them good sorbents for gases and ions (Cornell and
Schwertmann 2003, 3). αFeOOH is the most geologically stable iron oxide, is
considered un-reactive and the β, γ and δ FeOOH polymorphs are unstable relative
to it. It will form the greater part of the rust layer in most instances of atmospheric
corrosion unless oxygen availability is limited (Hoerlé et al. 2004). It can adsorb up to
0.5% chloride onto its surface (Turgoose 1982b). It has been widely reported as
occurring as a corrosion product from atmospheric corrosion (Maréchal et al. 2007;
Maeda et al. 1992; Dillmann et al. 2004; Monnier et al. 2007)
2.4.3 βFeOOH (Akaganeite)

βFeOOH was first identified in 1935 (Weisser and Milligan 1935) as forming from
conditions with high molarities of chloride and low pH either from solutions (Atkinson
et al. 1977; Rémazeilles and Refait 2007) or corrosion of iron in high chloride
concentrations and high RH (Turgoose 1982b). These conditions can arise on iron
with significant chloride contamination, typically in pits and in the vicinity of the sea.
Atmospheric drying cycles are when it is likely to form, since this is when chloride and
acidity concentrate within the surface electrolyte and pits. Formation of βFeOOH
from FeCl2/NaOH solutions depended on [Cl-]/[Fe2+] ratios and it did not occur at all
for Cl- below 1.6M but was the sole product above 3.2M, whereas high [Cl-] but low
[Fe2+] produced only α and γFeOOH (Rémazeilles and Refait 2007). Its formation is
preceded by intermediates such as β-Fe2(OH)3Cl (Rémazeilles and Refait 2007;
Dillmann et al. 2004) and green rusts, which is likely why low [Fe2+] prevents its
occurrence as the necessary intermediates cannot form. This highlights the
complexity of βFeOOH formation.
It has chloride occluded in its tunnel like crystal structure and adsorbed on its surface
(Stahl et al. 2003), with total w/w chloride varying from as little as 0.3% to as high as
18% for βFeOOH formed in marine contexts (Childs et al. 1980; Ishikawa and Inouye
1975; Keller 1970; Watkinson and Lewis 2005a) and Thickett recorded chloride
contents between 3% to 14.8% in βFeOOH naturally formed on archaeological
objects post excavation (Thickett and Odlyha 2014). Recent study suggests up to 6%
39
chloride is occluded in tunnels and only 1% mass reduction of tunnel chloride was
achieved in aqueous washing procedures, leading to the conclusion that the residual
chloride posed no corrosion threat as it was effectively locked in the crystal structure
(Réguer et al. 2009).
Surface adsorbed chloride is readily removed by washing (Watkinson and Lewis

2005b; North 1982) making it available to support corrosion (Turgoose 1982b). It has
been shown to be hygroscopic adsorbing water onto its surface (Kaneko and Inouye
1979; Watkinson and Lewis 2005b) and corrodes iron in contact with it down to 16%
RH (Watkinson and Lewis 2004) and Thickett and Odlyha (2014) detected corrosion
at 13% RH. Reactions at these RH values are very slow and only begin to accelerate
appreciably around 40% RH (Watkinson and Lewis 2004b). As with most electrolyte
driven reactions on iron, corrosion rate begins to increase significantly around 55% -
60% RH, which is in line with critical humidity for corrosion. βFeOOH is insoluble and
will remain on the metal surface but rain events can be expected to wash off
adsorbed chloride. Hydrolysis of mobile surface adsorbed chloride may produce
volatile HCl, as at 92% RH iron in an enclosed desiccator containing βFeOOH corroded
remotely (Watkinson and Lewis 2004b).
It can be expected to occur in rust when iron is corroding in chloride rich contexts;
again emphasising the importance of geography, climate and surface preparation
before application of coatings. Analysis of two objects from coastal sites detected
βFeOOH alongside αFeOOH and γFeOOH which are both expected as the more
commonly occurring rust components in atmospheric conditions (Maeda et al. 1992).
It is both electrochemically and physically dangerous, promoting oxidation of iron at
low humidity and its tower-like growth causing physical damage to overlying
corrosion layers and coatings. At low RH over long time periods its formation and
presence under coatings could support slow corrosion that promotes coating
disruption. This highlights the importance of surface preparation to remove chloride
prior to coating. It has been detected in many heritage contexts; within iron bars
inside Amiens cathedral (Monnier et al. 2007); on the iron hull of Brunel’s ss Great
Britain (Watkinson and Lewis 2004); and archaeological objects (Refait et al. 1992;
Réguer et al. 2006).
40
2.4.4 γFeOOH (Lepidocrocite)
γFeOOH is a common corrosion product in rust layers and has been widely reported
on outdoor iron (Graedel and Frankenthal 1990) and indoors (Monnier et al. 2007).
In rapidly rusting iron it is an oxidiser in the wetting phase of wetting/drying corrosion
cycles, when oxygen is less readily available as it has to access the surface through
pores in the rust layer (Hoerlé et al. 2004) (see 2.1). The reduced form of γFeOOH is
an electronic conductor. It is unstable with respect to αFeOOH but remains in rust
layers for some time.
2.4.5 Fe3O4 (Magnetite)

This black mixed oxidation state compound (Fe2O3.FeO) occurs when there is a
limited supply of oxygen in the reaction environment (Fig. 2.10). Its occurrence under
coatings can be expected where oxygen ingress is limited. Fe3O4 often occurs in
tandem with αFeOOH, as slight variation in the supply of oxygen will lead to the
stability of one or other of these compounds (Fig. 2.10). It can occur in a powdery
voluminous form or as a dense layer, which is more common on buried
archaeological material (Neff et al. 2007). It is a conducting oxide that ‘almost
displays metallic properties’ (Cornell and Schwertmann 2003, 115), which allows it to
act as a surface for cathode reactions remote from anode sites (Jones 1992, 405). It
can also adsorb chloride onto its surface and may act as a source of mobile chloride
in aqueous systems (Ardizzone et al. 1983). It can reduce oxygen more than the metal
and residues of it beneath paint layers may lead to corrosion continuing under the
film, since it can act as a remote cathode to inward diffusing oxygen (Cornell and
Schwertmann 2003, 500).
2.4.6 α-Fe2O3 (Haematite)

Haematite occurs in high temperature corrosion and may be a likely transformation
product when iron oxides are heated at high temperatures (Cornell and
Schwertmann 2003). Dehydroxylation of αFeOOH to α-Fe2O3 can commence at 85oC
(Cornell and Giovanoli 1990) and can occur as a solid state thermal transformation at
around 270oC (Cudennec and Lecerf 2005). It is not expected to occur during
atmospheric corrosion but flame cleaning of surfaces carried out prior to application
of coatings may offer opportunities for its formation, although this will depend on
41
duration of heating and temperature. It is a semiconductor (Cornell and
Schwertmann 2003, 115) and is also stable in acid down to pH 4 (Cornell and
Schwertmann 2003, 117).
2.4.7 ƴFe2O3 (Maghemite)

This is not commonly recorded as a corrosion product occurring during the
atmospheric corrosion of iron as magnetite is the more commonly identified product.
It has been detected on atmospherically corroded historical iron in its more lightly
corroded areas (Maréchal et al. 2007) and several times on archaeological objects
(Neff et al. 2007). ƴFeOOH may occur in wet dry cycles (Lyon 2010; Hoerlé et al. 2004)
as it is a preferred route for Fe3O4 and may be produced in the drying phase as
reoxidation of Fe3O4. It is relatively resistant to reduction and becomes a more stable
part of the rust layer (Lyon 2010).
2.4.8 Ferrous and ferric chlorides

Ferrous chloride solutions can exist in moisture layers on iron highly contaminated
with chloride and solid FeCl2.4H2O has been detected in pits during post excavation
corrosion on archaeological iron, which normally has a high chloride content derived
from ingress of Cl- as a counter ion in burial contexts (Thickett 2011). It forms acidic
solutions and hydrolyses to produce low pH values. It is hygroscopic and as
FeCl2.4H2O promotes corrosion of iron down to 20% RH (Watkinson and Lewis 2005b;
Turgoose 1982b), with corrosion ceasing below this RH as FeCl2.2H2O becomes the
stable phase (Watkinson and Lewis 2005b). So called ‘weeping’ on archaeological
iron involves aqueous blisters of ferrous chloride/ferric chloride solution that form a
skin of βFeOOH on their surface where abundant oxygen oxidises the solution
forming hollow shells when the water within them evaporates (Selwyn and
Argrylopoulous 2005). Ferrous chloride solutions can oxidise to form βFeOOH in the
atmosphere at high humidity, which will bind up some chloride in its tunnel structure
but will present corrosion risks to iron (see 2.3.3). Overall, these soluble chlorides are
dangerous corrosion products and can disrupt coatings appreciably.
42
2.5 Corrosion context
The outcomes of differing corrosion routes and the challenges presented by heritage
objects are evident when examining the brine tank from the Lion Salt works Cheshire
(Fig. 2.11). For historical, ethical and visual aesthetic reasons preservation of the
remaining paint may be a desirable conservation option, while applying a new
protective coating to the areas where the metal is exposed. This presents challenges
for surface preparation to receive the coating and development of strategies to deal
with the existing paint layers, where an industrial approach would be to strip to bare
metal or even replace the metal panels where corrosion is extreme.
A range of corrosion has occurred on the tank. Design features such as overlapping
plates and riveted joints in conjunction with damaged coatings can lead to oxygen
starvation cells being set up (Fig. 2.12) (see section 2.1). In this instance, coating
damage at an overlap allows moisture to enter between plates of wrought iron and
general corrosion ensures. This depletes oxygen and creates an anode by oxygen
starvation. Oxygen replenishment to the interface region is via the overlap creating
a highly oxygenated electrolyte at the edge of the overlap and this supports the
cathode reaction. This will be activated by wet dry cycles and lamellar corrosion
results at the edge of the overlap, due to transport of the Fe 2+ ions from the anode
to the cathode where they react with cathodically generated OH - ions (Fig. 2.12).
Where the protective coating fails, capillarity will draw water into the gap between
the plates and corrosion products hydrate thereby maintaining the electrolyte
solution. The disruption this produces is evident in Figure 2.11 and 2.13, where whole
lines of rivets and plate overlaps show loss of paint and corrosion. Rain events
influence oxygen starvation cells such as the one described here, as once the gap
between the plates dries corrosion will cease. For the brine tank, object function may
also influence corrosion. It contained brine and loss of paint at the overlap may have
resulted from seepage of the brine from the inside of the tank providing an
electrolyte in the overlap (Fig. 2.13).
43
Figure 2.11 Wrought iron brine tank at Lion Salt works Cheshire.
Figure 2.12 Oxygen starvation cell.
44
Figure 2.13 (i) Lamellar corrosion (ii) pitting corrosion
Lamellar corrosion is a common feature of wrought iron corrosion. When coatings

are in place it develops by oxygen disbondment beneath the paint layer (see section
2.1). Brunel’s 1843 wrought iron ship ss Great Britain was stored for many years open
to the elements in a dry dock in Bristol. Attempts to protect it from corrosion using
flame cleaning and coatings (Keystone 1999) were ineffective and flaking of the paint
layers revealed lamellar corrosion. This layer contained magnetite due to the low
oxygen levels beneath it during the corrosion process (Fig. 2.14). Corrosion would
have been exacerbated by chloride from the marine environment the ship operated
within being present beneath the paint layer, where it offered an excellent
electrolyte (Watkinson and Lewis 2004). The impact of extensive wetting and drying
cycles from rain events in the uncovered dry dock provided the aqueous phase
necessary for corrosion via water running down the hull and as driving rain. Corrosion
is now controlled by desiccation within a protective envelope which meets aesthetic,
ethical and financial goals by providing excellent visitor access (Watkinson and
Tanner 2008).
Solving the problem of hull corrosion above the waterline was directed more towards
commercial practices with cleaning to Sa2.5 and two pack epoxy resin with a
45
polyurethane (PU) overcoat. This may be considered less ethical and more intrusive
an approach but cost-benefit calculations and pragmatic decisions on maximising the
lifespan of the hull and its maintenance regime are of equal importance within
conservation strategies. Rationalising such decisions can only occur with confidence
in the degree of predictability and for this to occur an evidence base is required. This
dissertation sets out to contribute to developing such an evidence base for use of
coatings on heritage wrought iron.
Figure 2.14 Brunel’s ss Great Britain paint loss and lamellar corrosion.
Selection of coatings for ferrous metals is complicated by the range of conditions to

which the ironwork will be exposed (Table 2.2). Clear differences could be expected
between inland and marine contexts but even on the same site corrosivity conditions
may change dramatically, for example with the introduction of de-icing salts on roads
in winter. Effective decision-making in selecting a fit-for-purpose coating system
must therefore be based on nuanced understanding of corrosion parameters and the
specifics of the material and site.
46
Corrosivity Loss of low-carbon steel after first Typical exterior
category year of exposure environments in
Mass loss Thickness loss temperate climates
2
(g/m ) (μm)
C1 very low ≤ 10 ≤ 1.3 -
C2 low > 10 to 200 > 1.3 to 25 Atmospheres with low
levels of pollution. Mostly
rural areas.
C3 medium > 200 to 400 > 25 to 50 Urban and industrial
atmospheres, moderate
sulfur dioxide pollution.
Coastal areas with low
salinity.
C4 high > 400 to 650 > 80 to 200 Industrial and coastal
areas with moderate
salinity.
C5-1 very high > 650 to 1500 > 80 to 200 Industrial areas with high
(industrial) humidity and aggressive
atmosphere.
C5-M very > 650 to 1500 > 80 to 200 Coastal and offshore
high (marine) areas with high humidity.
Table 2.2 Corrosivity categories of exterior environments according to BS EN ISO

12944-2:1998 (British Standards Institute 1998).
Coating performance and corrosion of iron are interlinked and involve multiple
variables that make predicting service life difficult and unreliable. The impact of this
on management and best use of available budgets is significant and delivering a
message of limited-predictability to the heritage sector managers, when they want
quantitative answers, is a major challenge when developing standards for heritage
contexts. Predicting coating performance on ‘polished factory perfect un-corroded’
iron surfaces has potential for error, given the range of variables involved, for
heritage coatings covered in oxide this unpredictability escalates. For these reasons,
this dissertation focuses initial surface preparation methods on genuine heritage iron
samples before addressing coating performance.
47
3 Corrosion Prevention Treatments
3.1 Traditional treatment of wrought iron

Recognition that corrosion of iron degrades its useful properties leading to failures
of structures and objects is long standing with treatments applied to wrought iron
historically to prolong its useful lifespan. Surface preparation would involve scraping,
pickling and chipping to remove scale and contaminants then application of linseed
oil, heating of the iron and thorough working of a beeswax and boiled linseed oil
mixture into the metal surface (Kreislova et al. 2013, 333; Ashurst and Ashurst 1988).
As now, corrosion prevention commonly involved application of protective coating

systems. Traditionally, topcoats were of oil based paints with turpentine solvent,
linseed oil binder, white lead pigment (carbonates and sulphates) plus additional
pigments and driers (Blackney 2010). Corrosion inhibiting primers in the 17th and 18th
centuries contained red ochre, white lead and occasionally the red lead that would
become the first choice of anticorrosive pigment in the late 18 th and early 19th
centuries (Kreislova et al. 2013, 333; Blackney 2010). Red lead has a strong affinity
for linseed oil and a tendency to form lead soaps in a tough, elastic and impermeable
film (Gayle et al. 1992, 137). Pitch, coal tars and bitumen were alternative coatings
for larger structures such as bridges, piers and aqueducts (Blackney 2010). Waxes
and oils may have been used on bare metal and working machinery parts. Bitumen
was identified as the protective coating on a brine tank from a salt works in Cheshire
(Fig. 3.1).
48
Figure 3.1 Bituminous coating on a brine tank in Cheshire.
A rich resource for examining historic coating practices exists on wrought iron
structures. Where surface preparation before recoating has been minimal, a
chronology of paint layers remains beneath more recently applied coatings. Even
prepared surfaces retain vestiges of historic coatings in inaccessible areas. When
examined in cross section, these coating layers allow analysis of compositional and
aesthetic changes over time. Taking colour as an example, a relatively recent
preponderance of black coatings on wrought iron belies earlier trends towards
greater colour diversity. Combining study of historic sources and paint layer sections
allows identification of early 18th century regional trends for blue, green, grey or
stone colours and a dominance of green by the end of the century (Catt 1995, 19).
When selecting colours for wrought iron coatings, inspection of paint sections
permits restoration of an original or historic aesthetic. Traditional green railings can
be seen at the English Heritage property Apsley House in London. Figure 3.2 shows
multiple paint layers resulting from many overcoating episodes on the Tijou screens
at Hampton Court Palace and on railings at Dunbar Castle. Between 20 and 30
49
coatings can be identified on the Tijou paint section and 16 on the Dunbar section.
Striking colour variation is evident in the Tijou section until the uppermost layers of
black coatings. The bright yellow and red layers near the top of the section are likely
primers. The potential of such sections to inform on practicalities and failure of
coatings is explored in 3.3.
a) b)
Figure 3.2 Polished section of paint layers from: a) the Tijou screen at Hampton
Court Palace (http://patrickbaty.co.uk/2010/11/25/hampton-court-palace-2); b)
railings at Dumbarton Castle (courtesy of Amber Lawson).
3.2 Current practice

Modern techniques for corrosion prevention of historic wrought iron continue to
centre on application of protective coatings following a degree of surface
preparation. Obtaining a clean substrate surface is thought critical for successful
corrosion prevention by coatings (Ross 1977, 29; Trethewey and Chamberlain 1995,
320; Leighs 2008a) but what constitutes adequate and appropriate preparation for
historic material is a grey area. Industry based standards are the main guide but their
translation to heritage contexts and materials is not simple (3.4.3). Heritage ironwork
practitioners, having arguably the greatest experience of treating historic wrought
iron, have outlined common practice in online guidance (Blackney 2010; Meehan
50
2010; Taylor and Suff 2010; Topp 2010) and in personal communications. Methods
are reported here. Where discussion is supplemented by other references, these are
given.
3.2.1 Recording and dismantling

Where possible and necessary, structures are recorded and dismantled for treatment
off site. This facilitates full condition assessment and thorough cleaning. Recording
includes drawing with measurements, photographs and details of repair methods
employed. Mechanical joints which are simple to remove may be difficult to reinstate
hence no more dismantling than strictly necessary is recommended.
3.2.2 Cleaning and surface preparation

There is much debate over necessary preparation levels for surface tolerant paints
(Higgins et al. 2010, 2484). Reports insist even minor contamination by oil or grease
prevents successful adhesion and hygroscopic pollutants (chlorides and sulphates)
draw moisture through holidays and pores in coatings (Godfraind et al. 2012, 180).
For non-surface tolerant coatings a surface free of oxides, contaminants and grease
is demanded by manufacturers for successful coating performance (Leighs 2008a).
Removal of existing surface coatings and corrosion products is therefore common
practice prior to repair and recoating with all coating types. This also facilitates
repairs, removes hazardous residues of lead based paints and reveals detail obscured
by coatings.
Difficulties arise where historic paint layers are viewed as valuable evidence and
original material. Ethical conservation practice may not permit removal of such
coatings. A trade-off exists between increased efficacy of coatings applied to
thoroughly prepared surfaces and the ethics of removing historic material as coatings
and surface oxides which may also be protective. A range of minimally-destructive
cleaning techniques (Table 3.1) may prepare surfaces adequately for successful
application of surface tolerant coatings by removal of (Childs 1985, 26):
1) Mud and disfiguring deposits which trap moisture;

2) Soluble salts which encourage electrolytic corrosion;
3) Grease/oils which stain coatings and inhibit adhesion.
51
Material Removes Advantages Do not use on
Emulsifiable Thick oil, grease Can be used on Associated organic

solvent systems and dirt metals and most materials – textiles,
(white spirit & paints rubber
emulsifier)
Aqueous alkaline/ Oil grease and Effective on bare Aluminium or tinplate

detergent dirt ferrous metals elements, galvanised
(sodium iron, organics
hydroxide type)
Sugar soap Light coatings of Inexpensive and Some paint films

oil, grease, dirt easy to use become softened
Detergents Light coatings of Effective on most Some paint films or

oil, grease, dirt surfaces where frothing is a
problem
Organic solvents Oil, grease but Good degreasing Organic materials or

(paraffin, white not dirt efficiently action but may areas difficult to rinse
spirit) smear oil over wider
area
Table 3.1 Summary of some non-destructive surface cleaning methods, their

applications, advantages and disadvantages after Childs (1985, 26).
Where more destructive surface preparation is permissible/desirable, methods

include hand and power tools (scrapers, wire brushes, needle guns), chemical
stripping, flame cleaning, air abrasion (shot and grit blasting) and high pressure water
blasting. Grit blasting to Sa 2.5 or ‘near white metal’ (British Standards Institute 1998)
is common for removal of existing coatings and oxides but damages iron surfaces and
removes manufacturing and working evidence. Consequently, chemical removal of
paint may be advocated. Paint type dictates the chemical used which must be
thoroughly rinsed after cleaning and constant widening of health and safety
regulations reduces options for removal. Heritage ironwork practitioners
recommend removal of corrosion products by application of heat (flame cleaning).
52
This exploits the differential in expansion of iron and iron oxides to remove corrosion
products from iron surfaces.
Practical considerations in selecting surface preparation methods include health and

safety concerns, cost, equipment, expertise and preparation for waste management
(spent blast media and coatings, residual chemicals from stripping baths, debris
containing toxic heavy metals such as lead from coatings). Appendix 1 summarises
guidance on surface preparation methods from heritage sector texts (Wilson et al.
2008; Childs 1985; Ashurst and Ashurst 1988; Godfraind et al. 2012). Although there
is broad agreement between authors, suitability of blasting with ‘hard’ media is
recommended by one author and not by two others. There are slightly divergent
parameters for success according to the priorities of the authors. This encourages
treatment specifiers to consider a range of factors but confuses decision-making
processes. Guidance is neither clear nor standardised.
3.2.3 Repairs
Repair using the same material and techniques as employed for original manufacture
of the object or structure are deemed desirable. Dominant considerations when
selecting repair and replacement material are durability, price, corrosion resistance,
strength, ease of working, compatibility with adjacent materials and availability. In
reality, price of a job likely plays the biggest role in decision-making, particularly since
councils and government departments are often the specifiers. Tendering for
contracts demands competitive pricing which narrows the options for materials and
processes.
The availability of wrought iron for conservation is due to its recyclability. Early
charcoal iron is a limited resource but 19th century material is abundant and can be
reforged for repairs. Much derives from dismantled historic bridges and similar
structures. The variable quality of historic wrought iron should be considered when
selecting material for repairs. Some practitioners advocate use of pure iron which is
argued to permit higher quality welds and conforms to ethical requirements that
repair material be distinguishable from original. Mild steel is often used but it
performs differently to wrought iron, does not lend itself to the same forging
53
techniques (necessitating arc welding) and rolled sections are uniform and easily
recognisable from hand forged wrought iron. This may not be at odds with
conservation ethics but delivers an undesirable aesthetic. Malleability permits
deformations in original material to be straightened which avoids replacing sections.
Bolts, rivets, collars and traditional fire or forge welding are recommended and
resetting of iron fittings in stone using molten lead in preference to historically
inaccurate resins. Areas acting as water traps may be modified to enhance drainage,
flooded with paint or packed with red lead putty or pitch. Modern fillers (acetic acid
free RTV silicone rubbers, polysulphide mastics and epoxy or polyester resins) are
compatible with most modern protective coatings.
3.2.4 Protective coatings

Application of surface coatings to minimise corrosion is advocated and the range in
use is briefly summarised here. Traditional oil based systems (or modern alkyd
equivalents) are recommended with two-pack epoxy systems with polyurethane
topcoats to be used on exposed sites or where maximum longevity is required (up to
25 years). Lead based paints can be licensed for use on Grade I and II* listed buildings
in England and Wales and category A in Scotland (Historic England 2015). Traditional
oil paints had become unpopular due to labour intensive application of many thin,
slow drying coats. A trend towards traditional and sympathetic materials and historic
paint schemes renewed their popularity. They offer advantages of compatibility with
existing coatings and tolerance of surface oxides. Coatings with alkyd oil binders
largely replace traditional oil paints having advantages of higher film thicknesses
(permitting application of fewer coats) and faster drying times. They do not offer as
long a service life as epoxy resin coatings but maintenance is simple and coatings
widely available. Other single pack coatings with increased service life are solvent
drying vinyls, chlorinated rubbers and moisture-curing polyurethanes. Waxes and oils
conform to minimum intervention and reversibility conservation ethics but are not
recommended for exterior wrought iron.
Galvanising is considered inappropriate for historic wrought iron as the necessary

acid dipping etches the surface, can damage delicate elements and is difficult to rinse
54
thoroughly. Galvanising is clearly very difficult to reverse. Zinc or aluminium sprayed
coatings are occasionally applied to cathodically protect grit blasted wrought iron but
spraying in situ presents difficulties.
3.2.5 Application of protective coatings

Methods should follow those stipulated on manufacturer data sheets with strict
adherence to conditions of temperature and humidity. Roller and spray application
are faster but brush application likely results in more attention to joints, crevices and
inaccessible areas. Film thicknesses should be carefully controlled to conform to
manufacturer instructions and breaks or pinholes in the coating (holidays) be
avoided.
3.2.6 Maintenance
Conservation of exterior wrought ironwork must be an ongoing process not a discrete
event and this should be identified in management plans. Systematic maintenance
regimes are required involving periodic inspections, written repair schedules,
dedicated and adequate budgets and trained staff. Inadequate maintenance, often
due to lack of time, funding and long term vision, eventually necessitates costly and
highly interventive conservation and restoration projects which could be avoided.
Alternatively, the iron is lost from the heritage pool. A suggested maintenance
schedule is given in Table 3.2.
55
Twice yearly Biannually Occasionally
(spring/autumn)
General inspection Minor repairs Renewal of coatings –

timescales dependent on
Record areas of concern Inspect seals and renew
coating, atmospheric
as necessary
Touch up damaged corrosivity, use of the
coating Check iron fittings in structure, extent and
masonry for corrosion quality of surface
Clear gutters and
rainwater goods Repair and renew preparation
coatings as necessary
Remove debris
Check drains and
soakaways
Secure loose components
Clean and lubricate
working parts
Remove snow/ice/de-
icing salts
Table 3.2 Suggested schedule of maintenance of exterior wrought ironwork (after

Godfraind et al. 2012, 80).
3.3 Evidence from paint layers in section

Examining paint layers from historic ironwork offers insight into application methods,
coating performance and modes of failure. Table 3.3 gives examples of polished
sections of coatings from wrought and cast iron and mild steel gates and railings at
Historic Scotland properties. Samples were gathered for a parallel project by Amber
Lawson examining coating performance on steel substrates but are relevant for this
study.
56
Coating cross section Comments
a) Dunstaffnage Castle chapel gate
Primer application on top of previous topcoat.
b) Dumbarton Castle stair handrail
Primer ingress into laminations of substrate.

Break-up of the primer where crevice expanding?
c) Dumbarton Castle railing
Air bubble evident in thick grey coating from which

a crack in the coating has propagated which
eventually follows the line of the grey/black
coating interface and causes deadhesion. Lack of
adhesion evident between most grey and black
coating layers and black and red layers.
d) Dunglass Church gate
Large air bubble likely formed during coating

application. Air bubbles present in a surprising
number of sections and are likely not particularly
visible during application.
57
e) Fort George railing
Primer application over topcoat.
Coating failure/damage with the loss of several

layers which are then overcoated.
Variation in thickness of layers evident.
f) Broughty Castle railings
Cracking and lifting of coatings infilled by next

coating application. Demonstrates importance of
regular maintenance, touching up to prevent
further delamination of cracked coatings.
g) Arbroath Abbey railings
Break up of black coating layer indicates

embrittlement? Caused by environmental factors
or movement of underlying substrate?
h) Palace of Holyroodhouse railing water leaf
Delamination between the black and white

coating layers. Alternating dark and light colours of
coatings permits identification of holidays and
missed areas during application.
i) St Andrews Cathedral gate
Consolidation of fragmentary coating layers by

subsequent coating application. Corrosion
products also consolidated and coating ingress
into crevices in substrate is evident where air
bubbles have formed.
58
Table 3.3 Microscopy images of polished sections of coatings from wrought and cast
iron and mild steel railings and gates from Historic Scotland properties (courtesy of
Amber Lawson) highlighting aspects of application, performance and failure. Scale
bars are 500μm.
Discussion of samples a)-i) offers insight into the range or absence of surface
preparation and the evidently ad hoc approach to recoating. Polymeric coatings are
applied to a dynamic substrate. Expansion and contraction of the iron with
temperature fluctuations tests the physical properties of coatings. Sections f), g) and
i) show cracking of coating layers possibly resulting from substrate movement
coupled with polymer weathering and embrittlement. Break-up of the black layers in
i) may be caused by corrosion product growth and associated volume changes below
the coating. In section b) widening of the crevice in the substrate has fractured and
de-adhered primer from within the crevice leaving cracks permitting ingress of water.
Surface preparation of i) prior to application of the grey coating (likely primer) was
limited as evidenced by fragmentary coatings and corrosion products below it. The
newly applied primer layer might seem to be sitting above these as a layer yet the
primer was sufficiently low viscosity to be drawn into crevices and make contact with
the metal substrate. It also has a consolidating effect on the fragmentary layers.
Air bubbles in coatings are evident in sections c), d), e), i). These are sites of weakness
with cracking propagating from a large bubble in c) eventually causing de-adhesion
between layers. Air bubbles are likely invisible to practitioners applying coatings but
potentially undermine protection. Two pack coatings may be more susceptible to air
bubbles given extensive stirring to mix two parts. Single packs are stirred to a lesser
extent before application to counter settlement of solids and separation of coatings.
This may also cause bubbles. Application of more, thinner layers may mitigate the
problem of large bubbles.
Sections a), c), e), f), h) and i) show minimum surface preparation leading to
application of primer over pre-existing coating systems. If these primers protect
substrates by sacrificial or inhibitive mechanisms, contact with the substrate is
59
essential for performance. They cannot be expected to protect effectively when
applied over existing coatings.
Use of contrasting colours for consecutive layers in a system is evident in these

sections. This is advantageous for assessing successful overcoating and is clearly
common practice whether through deliberate selection or manufacturer design.
Sections offer insight into practicalities and performance of coatings and application
methods which cannot be seen in plan view. These are not generally available to
practitioners to inspect their own work but could be a useful tool for generation of
illustrated best-practice guidelines for coating of historic wrought iron. A striking
observation from these cross sections is the range of coatings and apparent cavalier
attitude to recoating. This lack of surface preparation means that the rheology of the
coating system and the ambient environment (RH, temperature, precipitation) will
be important for any successful outcome. There will also be no degree of
predictability attached to the coating performances in the long term.
3.4 Decision-making in conservation of historic wrought iron

3.4.1 Decisions and decision-makers
Historic wrought iron is ubiquitous in Western Europe and structures of this material
may be set apart from other museum-type object by their scale, complexity and
exposure to uncontrolled, exterior atmospheric conditions. Costs associated with
treatment and maintenance can be prohibitive and procedures are frequently more
interventive and ad hoc than conservation ethics advocate. Mechanisms of corrosion
and corrosion prevention by protective coatings are complex and not fully
understood even within the sector dedicated to its research. Responsibility for
planning or specifying its treatment belongs to a range of individuals who may or may
not have experience of heritage iron or a working knowledge of conservation
principles. This includes home owners, museum professionals, house managers and
property surveyors. Realpolitik often dictates that there is no budget for specialist
consultation prior to specifying conservation work for historic wrought iron, yet it
60
involves a large number of variables including choice of surface preparation methods,
coating systems and maintenance regimes whose interrelations complicate decision-
making (Figs 3.3-3.5).
61
Figure 3.3 Interrelating factors (red arrows) in selection of surface preparation
method.
Figure 3.4 Interrelating factors (yellow arrows) in coating system selection.
62
Figure 3.5 Diagram indicating the complexity of the decision making process for
treatment of historic wrought involving many interrelated factors. Arrows indicate a
relationship between factors.
Due to heavy reliance on tendering in this part of the heritage sector, specification
for treatment of wrought iron often falls to a range of individuals whose familiarity
with historic materials and structures and conservation practice and ethics is limited.
The most basic step in the decision-making process, identifying wrought iron from
mild steel or cast iron, can be difficult for the non-specialist especially when sampling
is not feasible and decisions are made on visual appearance, technology and
corrosion patterns. Even where treatment is by conservation specialists, sector
specific research to aid their choice of paint and treatment system is sparse and
quantified data is absent entirely. Negotiating the complex web of treatment
63
variables depicted here and juggling one factor against another for positive outcomes
without essential underpinning evidence is extremely challenging.
3.4.2 Guidance
Written guidance is available in the form of technical bulletins, leaflets, short articles,
book chapters and occasional case studies in specialist journals and conference
proceedings (Ashurst and Ashurst 1988; Barker 2010; Blackney and Martin 1998;
Blackney 2010; Cheltenham Borough Council (online); Davey 2007; Davey 2009;
Meehan 2010; Mitchell 2005; Taylor and Suff 2010; Topp 2010; Schütz and Gehrke
2008; Watkinson et al. 2005; Wilson et al. 2010) published by heritage bodies,
government agencies, local councils and heritage ironworkers.
An overall absence of evidence based heritage standards for the treatment of historic
wrought iron means decision-making is reliant on un-scaled comparisons and, at
worst, guesswork. Although detailed industrial and commercial standards and
guidelines exist (ASTM 2008; British Standards Institute 2000, 2005/2006,
2006/2007, 2012; Corus 2004), their focus on modern materials, without any ethical
constraints of the type found in conservation, limits their direct use in historic
contexts to specific situations.
The heritage sector guidance on surface preparation listed above is well-intentioned

but limited in scope, frequently conflicting and reliant on an evidence base that is
rarely quantified or offers data and methodologies that are difficult to translate into
practical contexts. It is entirely understandable that application of this guidance
becomes anecdotal or experience-based on many occasions. Identifying how to
generate the necessary quantitative evidence based data that will transform practice
will rely on well designed, and preferably co-ordinated, research. Unfortunately,
conservation of large heritage wrought iron assemblages is dominated by contractor-
based private sector activity; understandably, research is not a priority here and
gaining funding to support it is extremely difficult. Similarly, research within
commercial coating companies and national standards bodies into developing
evidence based, dedicated procedures and products for historic material is
hampered by the lack of potential profit in the heritage sector. This leaves the beacon
64
of research to be borne by academia and the heritage institutions themselves, where
it must exist amongst many other priorities and be subject to underfunding. A natural
outcome of this situation is that surface preparation techniques developed for
industrial contexts are employed in the heritage sector, despite a dearth of study into
their suitability for historic wrought iron. The upshot is that the surface preparation
of heritage iron to receive paint is unregulated, ad hoc and unable to support
predictive conservation procedures. Exceptions occur when industrial standards can
be adopted and adhered to; the wrought iron hull of Brunel’s ss Great Britain was
cleaned to Swedish Standard Sa 2.5 (ASTM 2008) prior to painting, as befitted the
ethical constructs in place for a corroded hull that had been cleaned to the metal and
painted periodically during its lifetime (Watkinson et al. 2005).
3.4.3 Use of standards in heritage contexts

Conservation decisions are necessarily being made on the strength of anecdotal
reports of successes and failures. This limited understanding of the effects of
treatments applied to heritage material is in direct contravention of sector codes of
ethics as set out in national standards, which advocate evidence based treatment
techniques delivered within ethical constructs (Canadian Association for the
Conservation of Cultural Property 2000; American Institute for Conservation of
Historic and Artistic Works 1994; International Council of Museums 1984). For
instance, the availability of empirical evidence relating selection of surface treatment
methods and protective coatings to the corrosion rates of historic wrought iron is
non-existent, yet this information is the key to delivering evidence based
management of our ferrous metal heritage. If the limited resources available for
heritage preservation across the globe are to be employed to best effect, informed
and predictive management strategies are essential.
This approach requires heritage specific guidance that can be utilised internationally
across the sector, much as industry uses agreed international standards to dictate
procedure for surface preparation, coating application and coating performance. The
stringent stipulations of coating manufacturers and the existence of international
industrial standards guide practice in the protection of modern steels, producing a
heavily regulated industry where best practice methodologies for surface
65
preparation and application of protective coatings are clear for practitioners
(Standardiseringskommissionen i Sverige, et al. 1980; British Standards Institute
2007). Coating manufacturer datasheets recommend suitable coatings for steel
substrates in various corrosive environments and specify appropriate surface
preparation and application methods according to these standards. The emphasis
here is on strict preparation protocols to support the claimed longevity of coatings.
In contrast to industry, heritage preservation is a poor relation with neither

regulation nor its own industry focused standards. Despite offering a rich source of
information, caveats prevent wholesale application of the international corrosion
prevention standards and guidelines to heritage contexts. First, these international
standards for surface preparation are scaled by the corrosion performance of mild
steel. While this may be applicable to many heritage contexts dating from the
twentieth century, mild steel having replaced wrought iron in many construction
applications, each material exhibits different composition, microstructure and
corrosion mechanisms (L’Héritier et al. 2013; L’Héritier and Dillmann 2005; Gandy
2007; Santarini 2007). Second, industry as relates to modern construction is not
hampered by the ethical and aesthetic constraints applicable to heritage ironwork
practitioners which are set out in national standards. For example, coating
manufacturers regularly specify blasting to Sa2.5 or ‘near white metal’ to produce a
surface profile promoting coating-substrate adhesion. For the heritage industry, loss
of original material associated with blasting to a metal core (where this exists)
conflicts with concepts of ‘limitos’ and preservation of original surface (Bertholon
2001a, 2001b, 2007). The perceived advantages of achieving the most suitable
surface for optimal performance of new protective coatings to ultimately minimise
loss of metal mass through further corrosion, must be weighed against decimation
of information held within historic coating and corrosion product layers. Little
quantitative information exists regarding the impact of aggressive surface
preparation techniques on heritage metals.
A recent extensive survey of references to the use of national and international

standards cited in metal conservation literature, covering the period 1995-2010,
revealed only two instances of citations of surface preparation standards for heritage
66
ferrous metals (Argyropoulos et al. 2013). Why is there a lack of engagement with
existing industrial standards in heritage conservation; is it an indication of a lack of
standardisation in metal conservation practice, an indictment of standards which are
not fit for purpose within heritage contexts or simply that there is limited reportage
of current practices? Whatever the reason, the applicability and usefulness of the
standards employed in the corrosion protection sector to heritage contexts need to
be explored, if they are to be recommended as meeting the particular needs of
ferrous metals in heritage contexts. Research delivering quantitative data is required
to fulfil these goals.
A first step towards achieving this is to use international standards to determine how
surface preparation methods used in these contexts influence corrosion rates of the
ferrous metal substrate. This is fundamental to evaluating the likely efficacy of a
coating system, as transmission of reactants through coatings will result in corrosion
related to the reactivity of the underlying surface. Corrosion will impact on the
deterioration and loss of the overlying coating subject to the input of other variables
such as its adhesion to the prepared surface. Clearly, a wide range of interrelating
factors determine coating performance but surface preparation remains a
fundamental variable within the equation.
Selection of protective coatings from a saturated market of heavily promoted

products is equally unstandardised. Fashions within the protective coating sector and
heritage practitioner circles influence choices but empirical evidence of performance
on historic materials is utterly absent. Two-pack epoxy resins found favour in heritage
after well-publicised projects promoted their credentials. Today, questions are raised
over their suitability for application to intricate and dynamic heritage ironwork
subject to contraction and expansion in different planes, their ‘plastic’ appearance
which jars with the historic aesthetic and high build which disguises fine detail. Low
build oil-based coatings had fallen out of favour thanks to their painstaking
application in many thin coats. A resurgence of interest in these coatings has arisen
out of concerns for compatibility with historic aesthetic and traditional methods.
67
What cannot currently be introduced to balance these arguments is empirical
evidence of corrosion rates, predictions of lifetimes and modes of failure of coating
systems when applied to historic wrought iron exposed to atmospheric corrosion.
These are all factors to aim for and experimental research should be focused on
answering them.
68
4 Experimental
4.1 Ethos and rationale

4.1.1 Research ethos
Decision-making in treatment of historic wrought iron subject to atmospheric
corrosion processes is complicated and evidence of treatment successes and failures
to underpin decisions is entirely anecdotal (3.4). Empirical evidence of treatment
effects is required to facilitate management decisions and produce cost benefit
analysis of options. This approach is applicable throughout the heritage sector which
is a cross-over discipline that must be both derivative and innovative. Heritage draws
heavily on other sectors but must adapt methods and approaches to historic
materials and contexts. The data on which these are based must be both empirical
and fit for purpose. Recognition that treatments will in many cases not be specified
by conservators or specialist heritage practitioners highlights the need for evidence
based guidance that is widely disseminated and accessible both physically and
intellectually. It must be balanced to consider divergent priorities of stakeholders and
offer a range of considered options that can be tailored to meet constraints of
budget, time, resource and available expertise.
This experimental research was designed to produce empirical data of surface

preparation and coating selection effects on corrosion rates of historic wrought iron
which will underpin published guidelines on best practice. The research was funded
by Historic Scotland and builds on earlier work by their Technical Conservation team.
It is contextualised by qualitative reporting of practicalities and relates interplay of
surface preparation methods and successful application of protective coatings.
4.1.2 Related work

An abundance of historic ironwork exists in Scotland from a tradition of wrought and
cast iron production (David Mitchell pers. comm.). Acknowledging the variable state
of preservation of this ironwork and the predominantly anecdotal guidance for its
treatment, Historic Scotland instigated a research programme to assess
systematically optimum treatments for cleaning and repairing these traditional
69
materials (Wilson et al. 2010, 248). Investigation of 25 surface preparation
techniques for cast iron (Wilson et al. 2008; Wilson et al. 2010) assessed methods
according to success criteria including:
 Removal of existing paint;

 Preservation of oxide layers;
 Influence on physical properties (bend strength, micro hardness);
 Resultant surface profile;
 Value for money;
 Operator skill level required.
As a result (see Appendix 1), manual and power tool cleaning were recommended
for indoor cast iron, flame cleaning for in situ exterior ironwork and blasting with
‘soft’ blast media (plastic, walnut shells or dry ice) for workshop procedures.
Chemical cleaning (immersion in NaOH, HCl or H2SO4/H3PO4) was found to require
further investigation. Laser cleaning, rotary abrasive disc (power) and wire brushing
(manual and power) techniques were found to be unsuitable for further investigation
or sole use for cleaning. Whilst offering data (largely qualitative) on effectiveness of
methods, the study was limited to cast iron and did not project to real life exposure
contexts by examining corrosion rates of differently prepared surfaces. This PhD
research was funded by Historic Scotland to address some of these shortcomings.
The research extends the investigation of surface preparation methods to wrought

iron substrates. It builds on the previous findings by examining recommended
procedures of flame cleaning and blasting with ‘soft’ media and further investigating
chemical cleaning and blasting with ‘hard’ media (see 4.1.3.2). Physical effects on
substrate surfaces were quantitatively examined followed by measurement of their
oxygen consumption rate, as proxy for corrosion rate (4.1.3.1), when sealed in
individual reaction vessels at high relative humidity (RH). Investigation of coating
effectiveness for wrought iron could not produce empirical data for a broad range of
coatings given the enormous diversity in use. A contextual approach was taken to
selection of relevant coatings for this study (4.1.3.4).
70
4.1.3 Rationale for methods and parameters
4.1.3.1 Oxygen consumption as proxy corrosion rate
Surface preparation and application of protective coatings aims to prolong lifetimes
of wrought iron structures by minimising or preventing their corrosion. The ultimate
measure of success of the treatment is empirical measurement of corrosion rate.
Quantification of corrosion rate can be achieved by measurement of cathodic

consumption or production of reaction components (Turgoose 1985, 14). The
predominant cathodic reaction in atmospheric corrosion of iron is the reduction of
oxygen (Jones 1992, 7). Despite reduction and reoxidation reactions, corrosion
products that occur during wet/dry cycles in the atmosphere correlate closely with
metallic dissolution at the anode (Hoerlé et al. 2004). Measurements of oxygen
consumption of a metallic sample in a given environment can therefore be used to
estimate comparatively the extent and rate of corrosion of the metal. The advantages
of this method are its non-destructive nature, its relative sensitivity and the lack of
need for specialist knowledge.
Amperometric sensors for monitoring oxygen levels are robust, cheap and widely
available but consume oxygen themselves (Matthiesen 2013, 365). An alternative
method, described by Matthiesen (2007, 272; 2013, 365-371), is based on
luminescence with molecules excited by light at one wavelength emitting energy at
another (Fig. 4.1). Oxygen-sensitive molecules have been developed for which the
presence of oxygen quenches light emission and oxygen concentration can thus be
determined by luminescence decay time or intensity of emitted light. The oxygen
sensitive compound within the sensors is a ruthenium complex excited by light at
505nm with emitted light measured at 600nm. This ruthenium complex is attached
to a polyester base and formed into sensor spots of 5-10mm diameter.
Advantages of such a system include (Matthiesen 2007, 272; 2013, 371):
 No leakage;
 No self-consumption of oxygen;
 High selectivity of measurement for oxygen;
 Versatile reaction vessel dimensions;
71
 Relatively low cost;
 Sensor spots function in air and water;
 Remote measurement through transparent vessel walls is possible.
The system is not without any difficulties. Measurements are temperature

dependent; temperature must be controlled and recorded throughout. Reaction
vessel leakage must be measured and accounted for in interpretation of the results.
Consumption of oxygen by other components within reaction vessels (e.g. silica gel,
dataloggers) must also be measured and subtracted from sample consumption.
There must also be a degree of confidence that the corrosion reaction is the only
reaction of the sample material that is likely to consume oxygen.
Figure 4.1 a) working principle

behind optical oxygen electrodes.
Luminescent compound excited by
light at one wavelength is emitted at
another wavelength (a.1). In the
presence of oxygen the energy of
the excited molecule is transferred
by collision with oxygen rather than
by emission of light (a.2). Oxygen
concentration is correlated to
intensity of emitted light and
luminescence decay time.
b) Diagram showing modulated
light sent from transmitter to sensor
foil via an optical fibre via which the
emitted light is also returned.
Measurement is possible through a
transparent container to sensor
spots inside.
(Modified from www.presens.de
and published in (Matthiesen 2007,
272; Matthiesen 2013, 370).
4.1.3.2 Surface preparation methods

Flame cleaning and blasting with soft media were recommended in Historic Scotland
findings as appropriate for retaining oxide layers. Further testing here examined their
72
suitability for recommendation for wrought iron and investigated corrosion rates
after treatment which had not been looked at previously. Crushed walnut shell was
selected as the soft media (hardness 3-4 Mohs) having been recommended for
cleaning copper alloys (Lins 1992) and having the advantage of biodegradability.
Stipulations by coating manufacturers and in British Standards for surfaces prepared

to near white metal (Sa2.5) to optimise coating performance conflict with heritage
concepts of oxide layer retention for ethical and corrosion protection reasons.
Despite not being recommended by Historic Scotland, blasting with hard media to
Sa2.5 in this study examined resultant surface profiles for coating adhesion and
corrosion rates relative to substrates with oxide layers retained.
Chemical cleaning was found by Historic Scotland to require further testing. Sodium
hydroxide has long been known as a corrosion inhibitor for ferrous metals (Mayne
and Menter 1954; Mayne et al. 1950; Turgoose 1985) and with the probability of its
solvent effect on existing paint layers, was selected for testing here.
4.1.3.3 Humidity
Potentially conflicting priorities exist in laboratory testing of corrosion rates. Aims to
produce measureable corrosion of uncontaminated samples in a relatively short
duration experiments may be incompatible with an ethos of modelling real-life
contexts (Leygraf and Graedel 2000). Fortuitously, climatic conditions in Britain
involve relative humidities above 80%RH as an annual average and in winter months
in the region of 84-86% (Met Office 2014) (Table 4.1).
Region Average Relative Humidities 1981-2010 (%)
Annual Oct – Mar Nov – Feb
East Scotland 81.41 83.53 84.30
Wales 82.14 84.55 85.30
Forth River Basin 81.45 83.51 84.32
Severn River Basin 81.10 84.58 85.67
Table 4.1 Regional humidity averages 1981 – 2010 (Met Office 2014).
73
An RH of 90% within the reaction vessels was deemed likely to promote sample
corrosion beyond the error of the oxygen meter to allow corrosion rate
characterisation in permitted timescales whilst not representing an unreasonable
step up from ambient exterior conditions, especially during the drying phase in
wetting/drying atmospheric corrosion cycles.
4.1.3.4 Protective coatings

The number of coatings investigated was restricted by cost and time. Sensor spots
for oxygen consumption measurements are prohibitively expensive and time to
perform measurements is considerable. The number of coatings investigated was
limited to three permitting appropriate numbers of coated iron samples and controls
to be run.
Coating selection for this research was contextual. With the ubiquity of wrought iron,
contexts for coating and criteria for decision-making will be vast in number. Three
contextual scenarios have been modelled for this study (Table 4.2).
74
Scenario A Scenario B Scenario C
Budget Large Medium Small
Structure Bridge Historic gates Local museum

(example) architectural
fittings
Rationale Treatment to Thorough surface Limited surface
engineering preparation preparation
standards Removal of corrosion Conform to
Thorough surface products to enhance historic aesthetic
preparation coating performance Preserve original
Industrially Conform to historic material
recognised coating aesthetic Easy to maintain
Expertise National Local contractors Non-specialist
contractors
Maintenance Periodic complete Ongoing vigilance Ongoing vigilance
renovation by and touch up by and touch up by
specialist specialist non-specialist
Coating Two-pack epoxy Single-pack alkyd oil- Single-pack,

System resin primer and based primer surface tolerant
intermediate coats Single-pack alkyd oil- alkyd oil-based
Two-pack based top coat coating
polyurethane top
coat
Table 4.2 Contextual scenarios for coating selection.
A system fitting the criteria of scenario A was recommended by a heritage

practitioner and is the same system as applied to the ss Great Britain and other high
profile heritage projects. Epoxy resin polymers are commonly used in protective
coatings thanks to their desirable properties: good wetting of metal/metal oxide
surfaces; good to excellent chemical resistance; easily controlled cross-linking, wide
range of molecular weights; ease of forming pigment dispersions; range of available
cross-linking reactions; and relatively good water resistance (Bierwagen and
Huovinen 2010, 2651). The polyurethane topcoat offers UV resistance to the coating
system (Leighs 2008c). Coatings for scenarios B and C were chosen from household
name manufacturers and were available off the shelf. Alkyd oil coatings are used
75
extensively because of their low cost and wide range of applications but in an
unmodified form are sensitive to hydrolysis (Bierwagen and Huovinen 2010, 2652).
Specifics of each coating system are given in 4.3.3, Table 4.4 and Appendix 2 and 3 –
8.
4.2 Investigating the impact of surface preparation method on

corrosion of historic wrought iron
4.2.1 Aims and Objectives
This study aims to address the current knowledge gap regarding impact of choice of
surface preparation method on effective corrosion prevention for wrought iron by:
 preparing the surfaces of historic wrought iron samples by five methods;

 determining the effect of these preparations on sample surface morphology,
retention of existing coatings and corrosion products;
 quantifying the corrosion rates of the samples via measurement of oxygen
consumption at high relative humidity;
 relating corrosion rate to surface preparation method.
4.2.2 Characterisation of sample material

The sample material, sourced by Historic Scotland, is a wrought iron railing of
unknown provenance (Fig. 4.2). Vestiges of failed coating systems remain on
approximately 40-50% of the surface, adhering closely in some areas and loosely in
others. Corrosion products also cover the entirety of the railing in the form of closely
adhering, coherent oxide layers as well as laminating and powdery corrosion
products and pitting.
76
Figure 4.2 Iron railing from which sample material was cut.
Wrought iron is by nature an inhomogeneous material with local compositional

differences possible within the same piece of metal (Dillmann et al. 2004; O’Sullivan
and Swailes 2009, 260-261; Rawdon 2013). To minimise the chances or extent of
compositional variation between the samples they were cut from a continuous
length of flat bar iron. However, the skill of wrought ironworkers is such that the bar
may have been produced by welding shorter lengths of the metal together without
leaving any macroscopically visible indication (Chris Topp heritage blacksmith pers.
comm.).
Samples were cut from the wrought iron bar using a hand hacksaw with white spirit
as a lubricant to avoid heat from machine sawing. Sample size was 40mm x 30mm x
10mm as dictated by the dimensions of the bar and the diameter of the aperture of
the reaction vessel used for the oxygen consumption tests (Fig. 4.3). Each sample was
weighed and minor discrepancies in overall dimensions, and hence surface area,
were recorded.
77
Figure 4.3 Sample of wrought iron railing as cut to form samples.
Paint layers were examined in profile and elementally analysed using a CamScan
Maxim 2040 scanning electron microscope (SEM) equipped with Oxford Instruments
energy and wavelength dispersive X-ray spectrometers.
Polished sections of the railing were also analysed using the SEM to determine the
composition of the wrought iron. Samples of corrosion products were removed from
exposed areas and beneath exfoliating paint and analysed using a PANalytical X'Pert
Pro (Cu Κα) X-ray powder diffraction.
4.2.3 Surface preparation methods

Five surface preparation methods, determined through prior testing by Historic
Scotland (Wilson et al. 2008), were applied to the historic wrought iron samples:
 airbrasion with glass beads;

 airbrasion with aluminium oxide;
 airbrasion with crushed walnut powder;
 immersion in sodium hydroxide solution followed by wire brushing;
 flame cleaning followed by wire brushing.
78
Five samples of the wrought iron railing were prepared by each method with five
samples left in an as received, untreated condition acting as controls. The
methodology used for each preparation method is described below.
4.2.3.1 Blast cleaning with glass beads, aluminium oxide and crushed walnut
abrasive media
Blast cleaning of samples was carried out using a Texas Airsonics Model AJ-1
airbrasive unit. The airbrasive feed was thoroughly cleaned when changing between
abrasive media to avoid cross contamination. The abrasive media were:
 glass beads (Grade No.9 - 44µm; hardness 6 Mohs);

 aluminium oxide (Grade No. 3 - 53μm; hardness 8-9 Mohs);
 crushed walnut shell (Grade No. 6 - 0.3-0.6mm; hardness 3-4 Mohs).
Parameters standardised during the cleaning were:
 pressure - c.4 bar (increased slightly for crushed walnut to due softness of the
medium);
 powder flow/aim intensity (5);
 distance of nozzle from sample surface (50mm);
 angle of nozzle to surface (45°);
 end point [aim Sa 2.5 (near white metal)].
Post-cleaning samples were blasted with pressurised air to remove loose blast media
from their surfaces.
4.2.3.2 Immersion in sodium hydroxide followed by wire brushing

Samples were immersed in a 1.25M solution of sodium hydroxide in deionised water
for two hours then rinsed in deionised water until pH 7 was attained in two rinses.
Vigorous manual brushing with a stainless steel wire brush removed any remaining
coating and loosely adhering corrosion products until there was no further visible
change in the sample surface. The samples were then rinsed again in deionised
water, blotted to remove excess water and left to dry in a well-ventilated, warm area.
79
4.2.3.3 Flame cleaning followed by wire brushing
A hand held GoSystem Tech Multi Torch MT2055 using propane/butane fuel and
having a 17mm burner, a power output of 1000 Watts and a flame temperature of
1350°C was used. A standard protocol for cleaning was developed ensuring samples
were exposed to the same part of the flame which was constantly moving across the
surfaces. The duration for which each sample was exposed to the flame was similar
but with slight variation due to differing amounts of paint and corrosion product
adhering to surfaces, as would be the case in practice.
Samples were then brushed vigorously with a stainless steel wire brush until there
was no longer any visible change in the surface and blasted with high pressure dry
air to remove any remaining loose corrosion products or paint residues.
4.2.3.4 Determination of end point

In practice, the end point of surface preparation will be guided by a range of factors.
Coating manufacturer instructions may stipulate removal of all oxides and remnants
of previous coatings, heritage ethics or aesthetics may prioritise maximum retention
of coherent original material and availability of techniques limit possible outcomes.
Surfaces would generally be blasted with glass beads and aluminium oxide where the
aim was removal of oxides and existing coatings to bare metal (Sa 2.5). This was the
end point aimed for here. Crushed walnut having a lower hardness than magnetite
limits its use to contexts in which retention of coherent oxides is desirable. A surface
free from loosely adherent oxides and vestiges of coating systems was deemed the
end point for this technique. Immersion in sodium hydroxide solution and flame
cleaning both rely on wire brushing for final removal of loosely adhering corrosion
products and are unlikely to remove coherent oxide layers. End point in these cases
was when continued brushing caused no further visible change in surface
appearance. With all techniques, end point was assessed visually macroscopically as
would be the case for their in situ use.
4.2.4 Examination of resultant surfaces

Prepared surfaces were subjected to macro and microscopic examination and
comparison with untreated control samples. Macroscopic recording was by
80
photography and microscopically by use of a CamScan Maxim 2040 scanning electron
microscope (SEM) equipped with Oxford Instruments energy and wavelength
dispersive X-ray spectrometers using secondary electron imaging to examine the
resultant surface topography. BS EN ISO 8501-1:2007 (British Standards Institute
2007) was used to identify surface preparation grade with respect to the
internationally applied standard.
4.2.5 Measuring oxygen consumption and corrosion rates

Samples were individually enclosed within airtight reaction vessels (250ml Mason
Ball glass jars with plastic coated brass sealing discs tightened with threaded outer
sealing rings of brass which create a seal by deforming a synthetic rubberised ring on
the disc) containing 160g silica gel conditioned to 90% relative humidity (RH). Within
each vessel was an oxygen sensor spot (World Precision Instruments (WPI) part
#503090) adhered to the interior wall of the vessel using silicon adhesive (Radio
Spares RTV silicone rubber compound) and a watch glass to separate the sample from
the silica gel. The reaction vessels with samples were stored in a Binder KB240
climate chamber to control temperature to 20 ± 0.5oC and avoid RH changes within
the vessels which would result from fluctuating temperature. Madgetech RHTemp
101A data loggers monitored the internal environment to ± 3% RH of 28 of the 31
reaction vessels. All vessels maintained humidity between 88 – 93% RH ± 3% (logger
error) throughout the test period.
Using a fibre optic probe and WPI OxyMini oxygen meter (WPI OXY-MINI-AOT with
cable #501644) the oxygen concentration within each vessel was measured at regular
intervals (twice weekly) over an 11 month period (338 days for prepared samples,
257 days for un-cleaned samples). The oxygen within the vessels was replenished by
opening the vessel and resealing when internal oxygen pressure became depleted to
150mbar to ensure the reduced partial pressure of oxygen did not influence
corrosion rates and samples were photographed at this stage. The precision of the
oxygen measurements is 2mbar at atmospheric oxygen pressure of 210mbar and
increases proportionally with decreasing oxygen pressure. Control vessels filled with
nitrogen showed a negligible ingress of oxygen over a two year period indicating very
little leakage of the vessels (Watkinson and Rimmer 2014). A further control vessel
81
containing silica gel, watch glass and sensor spot was measured alongside the sample
vessels to quantify the oxygen consumption of the apparatus. This was found to be
0.03 mbar day-1 which was subtracted from the oxygen consumption of each vessel
when results were analysed.
4.2.6 Examining corrosion rates

The corrosion rate of the samples can be calculated as loss of metal using a simplified
equation for corrosion [6] if conversion to a range of oxidation products is not
considered. Corrosion is at a constant 90% RH and this offers a continuous film of
moisture and unchanging saturation of capillaries, providing a constant barrier.
Oxygen diffusion is a likely rate controlling step and the three phase corrosion
process involved in wet/dry cycle where the initial phase is the cathodic reaction,
reduction of corrosion products (Hoerlé et al. 2004), is unlikely to occur. It is likely
that corrosion will more readily equate to the saturated phase of this process where
oxygen reduction is the cathode reaction. Analysis of corrosion products may reveal
a more complex corrosion outcome but this simplified approach is based on the
reasoned estimation that FeOOH predominates.
4Fe + 3O2 + 2H2O → 4FeOOH ___________ [6]
This can be related to total loss of the metal, assuming corrosion rate does not
change with time, within a calculated number of years in a very humid mid-range
temperature environment of 90% and 20oC to provide some comparative indication
of corrosion rate.
The calculation uses the change in pressure of oxygen in the reaction vessel over the
test period (atmospheres), temperature (maintained at 20oC), volume of gas within
the reaction vessel (litres) and the gas constant (R = 0.08205746) to calculate the
number of moles of oxygen consumed by the corrosion of the sample [15]. The ratio
of oxygen moles to iron moles in the corrosion reaction is given in equation [6] and
is used to calculate the number of moles of iron converted to FeOOH during the test
82
period by assuming that [6] is the only reaction occurring. Changes to oxidation state
of iron in reactive corrosion product phases to balance dissolution of metallic iron
can occur immediately after wetting and do not involve oxygen consumption
(Stratmann and Hoffmann 1989). This would not be detected by the measurement
technique used here but is likely to be insignificant as the samples are constantly at
90%RH. Any contribution from these reactions is not considered in this calculation.
The mass of FeOOH per unit area can be calculated [16] and, using the density of iron
(7.874 g/cm3) and, assuming uniform corrosion, the depth of metallic iron becoming
FeOOH per unit time can be derived using [17].
 Ideal gas law:

𝑃𝑉
PV = nRT or 𝑛 = 𝑅𝑇 ___________ [15]
P = pressure of gas (atm.) R = ideal gas constant

V = volume of the gas (l) T = temperature (K)
n = amount of substance (moles)
 Converting mass to moles:
mass = moles x molar mass ___________ [16]
 Calculating depth of iron converted to FeOOH:
depth = mass loss / density of iron ___________ [17]
4.2.7 Identification of corrosion products

Following completion of oxygen consumption recording, corrosion products formed
on the glass beads, aluminium oxide and flame cleaned samples were mechanically
removed and analysed using a PANalytical X’Pert Pro (Cu) X-ray powder
diffractometer.
83
4.2.8 Flame cleaning
Reporting the corrosion rates of the prepared and uncoated samples to the sector at
the Icon Metals Conference Amazing Technicoloured Dreamcoats: Protective surface
finishes for metals (November 2013, Wallace Collection) generated consternation
amongst practitioners and the laboratory method for flame cleaning was called into
question. To address this, samples of gasometer wrought iron (described 4.3.2) were
flame cleaned by three different practitioners according to their preferred method
for historic wrought iron. Eight samples were prepared by each practitioner. The
specifics of their individual flame cleaning methods were recorded for comparison
(Table 4.3). All practitioners applied the flame to the surfaces of the samples and
intermittently brushed the sample surfaces vigorously with a steel wire brush.
Duration of application of the flame to the samples varied between practitioners. The
temperature of the flames and the metal were not known by the practitioners; the
colour of the metal during cleaning was recorded as an indicator.
Practitioner Torch Estimated Metal Duration of Determine

max. colour flame end point
temperature during application
of the metal cleaning to surface
(approx.)
PA Oxypropane 500°C Dull red 20-30 Visual

(oxygen cutting seconds cleanliness
boost)
PB Oxypropane 100°C Grey <10 seconds Cessation of

(oxygen cutting luminescence
boost) of coating and
oxide vestiges
under flame
(Fig. 6.4)
PC Oxyacetylene 700°C Cherry red 50-60 No further

(no oxygen seconds corrosion
boost) removable or
coating visible
Table 4.3 Parameters of flame cleaning by practitioners.
84
Prepared samples were photographed and subjected to the oxygen consumption and
corrosion rate measurement method described above. Laboratory flame (method
4.2.3.3) cleaned gasometer sample cross sections were imaged using the SEM in
backscatter mode alongside uncleaned gasometer samples to examine impact on
oxide layers in section.
4.3 Investigating protective coatings for historic wrought iron

4.3.1 Aims and objectives
This study aims to investigate the effectiveness of commonly applied protective
coatings for corrosion prevention of wrought iron by:
 preparing the surfaces of historic wrought iron samples;

 coating the samples with three commonly applied protective coatings (one
epoxy resin system and two alkyd oil based coatings);
 quantifying the corrosion rates of the coated samples via measurement of
oxygen consumption at high relative humidity;
 relating corrosion rate and practical considerations to a cost benefit analysis
of application of these coatings for corrosion prevention of historic wrought
iron.
4.3.2 Sample material
Insufficient sample material from surface preparation investigation remained for use
in coating system testing. The previous success in standardising historic sample
material encouraged further use of historic wrought iron and mid-19th century rolled
plate from the Kings Cross/St Pancras gasometer was sourced (courtesy of Pete
Meehan).
Optical and scanning electron (CamScan Maxim 2040) microscopy of polished

sections confirmed this to be wrought iron due to the presence of stringers of slag.
Wavelength dispersive SEM analysis was used to determine the composition of the
gasometer metal as part of a separate project run jointly with Eric Nordgren. XRD
85
analysis of the corrosion products present using a PANalytical X'Pert Pro (Cu Κα) X-
ray powder diffraction identified magnetite, goethite and lepidocrocite, (Figs 5.31-
5.33). This is consistent with corrosion products reported on iron subjected to
atmospheric corrosion (Bouchar et al., 2013).
The large gasometer plates are a consistent thickness (4mm). Samples were cut (to
30mmx40mm) from one plate by a contractor using a water cooled cutting process
to minimise the temperature increase and associated changes in the microstructure
of the wrought iron. Historic wrought iron is by nature an inhomogeneous material
(Dillmann et al. 2004) and local differences in microstructure and slag distribution are
possible. Samples were cut from the same plate to minimise compositional variation.
Possible removal of chlorides by cooling water was not considered problematic for
this investigation.
4.3.3 Protective coatings
Focus was on the performance of an epoxy resin system (System A) such as has been
used in high profile, big budget conservation projects (e.g. ss Great Britain, Forth Rail
Bridge) against that of a low cost, widely available, household name, alkyd oil-based
system (System B). A context in which the facilities for more extensive surface
preparation are not available is modelled using a surface tolerant alkyd oil-based
coating from the same household name manufacturer (Coating C). A summary of the
coatings is given in Table 4.4 and further details of the coatings and their application
requirements in Appendix 2.
86
Manufacturer Identifier Coating Trade Name Chemistry
Leighs System A Primer Metagard L574 Two-pack epoxy
Primer/ build Epigrip C400V3 Two-pack epoxy –

zinc rich
Topcoat Resistex C137V2 Two-pack acrylic

polyurethane
Hammerite System B Primer Red Oxide Primer Oil-based
Topcoat Garage Door Alkyd oil-based
Coating C Single coat Direct to Rust Metal Alkyd oil-based

system Paint
Table 4.4 Overview of coatings investigated.
4.3.4 Sample surface preparation

Samples were prepared according to coating manufacturer recommendations and in
line with the findings from surface preparation testing in this investigation. System A
and B samples were blasted to Sa2.5 using aluminium oxide powder. The method for
blasting was as detailed above (4.2.3.1). Coating C is surface tolerant (suitable for
application over existing corrosion products) hence surface preparation was limited
to wire brushing to remove non adherent corrosion products. End point of wire
brushing was when no more loose corrosion product was removed. For images of
prepared and coated samples see Figure 5.34.
4.3.5 Coating application

Coating systems were applied as per manufacturer recommendations for brush
application. Brush application was chosen as the most readily available method and
because of suggestions that brush application leads to the most effective adhesion
of coating to substrate (various practitioners pers. comm.). Numbers of applications
of each coating were as recommended for optimal performance of the overall system
in each case. Over-coating intervals were in line with those specified by the
manufacturers given the curing conditions of 22-24oC and 45-50% RH. A full summary
87
of application details from the manufacturer data sheets is given in Appendix 2.
Application was with a 1” Harris synthetic brush, as appropriate for the size of the
samples, with numbers of strokes determined by viscosity and wetting properties of
the coatings. Application aimed to cover the entirety of the sample, minimising the
appearance of holidays and brush marks. Curing conditions were interior and
relatively constant at c.24oC and 45% RH with air extraction.
4.3.6 Oxygen consumption and corrosion rates

The prepared samples and controls were subject to the oxygen consumption and
corrosion rate quantification method described above (4.2.5). Measurements of
oxygen within the vessels were conducted at regular intervals (approximately every
14 days) over a period of 540 days.
4.3.7 Control samples

Changing the sample material necessitated determination of baseline oxygen
consumption of control samples to allow comparison of corrosion rate and
assessment of standardisation. Blasted and wire brushed samples were prepared to
provide background consumption levels for substrates of Systems A/B and Coating C
respectively.
The coating systems themselves were likely to consume oxygen during the corrosion
testing. In order to subtract the oxygen consumption of the coating systems from the
overall oxygen consumption, giving that of the metal substrate only, the coatings
were also applied to inert substrates (glass) of the same surface area as the iron
samples. These were then exposed to the same controlled environment and oxygen
consumption measurement procedure. The oxygen consumption of a control vessel
containing silica gel, watchglass and glass slide was also recorded.
4.3.8 Practical observations

The requirement to approach evaluation of the coating systems pragmatically calls
for consideration of the practicalities of their application. During the coating process
relevant observations were made regarding the properties of the coatings, their
behaviour and any factors which might affect application in the field.
88
4.3.9 Properties of cured coatings
Coatings were assessed following the curing period for visible holidays, brushstrokes,
run back from edges, tack and aesthetics. These properties may affect performance
and suitability of a coating for application to historic material.
4.3.9.1 Dry film thicknesses

Recommended dry film thickness (DFT) values are specified by the manufacturers for
each coating for optimum performance and aim to strike a balance between
slumping caused by application of too great a mass of coating and holidays and lack
of coverage from too little. DFT is argued to be the most important single
measurement made during application and inspection of protective coatings as these
are designed for optimum performance when applied within a narrow range of DFT
as specified by the manufacturer (DeFelsko 2014, 5). These values vary for different
application methods and can directly impact physical properties of the coatings.
Following curing of test samples coated with systems A and B and coating C, the dry
film thicknesses were tested using a DeFelsko PosiTector 6000 equipped with a
Ferrous/Non-Ferrous substrate probe calibrated on uncoated substrate material
using standard calibration shims 10371F, 8944F and 8623F accurate to ±0.43μm. In
its ferrous substrate function the probe uses the magnetic principle to measure the
thickness of non-magnetic coatings applied to ferrous metals with an error of ±1.0μm
at DFT 0-50μm and ± 2.0μm at DFT >50μm (DeFelsko 2014, 3).
4.3.9.2 Coating adhesion

Effective adhesion of the coating to the substrate is a parameter for success of
protective coatings. It was measured for each coating using a DeFelsko PosiTest AT-
A Automatic Adhesion Tester as per the manufacturer instructions (DeFelsko 2011).
Aluminium pull-off dolly surfaces were prepared by rubbing the base of each dolly 5
times across a 3M Scotch-Brite TM General Purpose Hand Pad 7447. Any residue
from this abrasion was removed by wiping with a clean cloth. The surface of the
coating was also prepared by passing the same abrasive pad across the surface 5
times and degreasing with industrial methylated spirits. Prior testing revealed no
negative effects of this small amount of alcohol on the coatings. 10 dollies per coating
system were adhered to the coated samples using Loctite Universal Superglue
89
applied in a uniform film to the dolly base, the dollies pressed onto the coated surface
for 40 seconds and excess adhesive wiped from around the dolly sides. The adhesive
was allowed to cure for 24 hours before the pull-off test was carried out. The
cyanoacrylate superglue was selected as an adhesive due to their use in the field
where a quick-curing adhesive allows pull-off testing to be achieved in a short time.
90
5 Results
5.1 Investigating the impact of surface preparation method on

corrosion of historic wrought iron
5.1.1 Railing sample characterisation
The mean composition of the sample analysed was 99% iron, 0.37% silicon, 0.36%
phosphorus and 0.27% manganese (Table 5.1) which is typical of a wrought iron. Slag
content was moderate and relatively evenly distributed in stringers, although
inclusions varied in size from ˂10μm to >600μm (Figs 5.1, 5.2). Corrosion was
localised and pitting was present (Fig. 4.3). XRD identified goethite (αFeO(OH)
diffraction code 01-081-0462), magnetite (Fe3O4 diffraction code 01-085-1436) and
lepidocrocite (γFeO(OH) diffraction code 01-074-1877) to be present as corrosion
products (Fig. 5.3). There were at least nine layers of paint on the samples and the
compositions of these (Figs 5.4-5.8) indicated the presence of lead based coatings,
those with barium sulphate fillers, cobalt drying agents, copper based pigments and
a suggestion of zinc. Overall the picture is of an ad hoc painting maintenance regime
with a range of pigment and paint types such as might be expected in the protective
coating of an outdoor railing.
Figure 5.1 SEM image of

polished section of railing
wrought iron showing slag
inclusions distributed in
direction of rolling.
91
Figure 5.2
Backscatter
electron SEM
image showing
detail of two
phase slag
inclusions with
typically rounded
wüstite (FeO).
Figure 5.3 Diffraction patterns of corrosion sample and matching compounds (top
to bottom): the corrosion product sample; lepidocrocite (01-074-1877); magnetite
(01-085-1436); goethite (01-081-0462).
92
Figure 5.4 Backscattered electron image showing location (indicated by white
squares) of analyses of paint layers (aim to analyse pigment inclusions) from the
sample material. Spectra 1-4 are given (Figs 5.5-5.8).
Figure 5.5 Spectrum 1 of SEM analysis of paint layers on sample material (location
indicated in Fig. 5.4) showing major peaks for sulfur, and barium with minor peaks
for copper, cobalt, strontium and iron.
93
indicated in Fig. 5.4) showing major peaks for sulfur and barium and minor peaks for
copper, cobalt and strontium.
indicated in Fig. 5.4) showing major peaks for lead and calcium and minor peaks for
aluminium, barium and copper.
94
indicated in Fig. 5.4) showing major peaks for lead and minor peaks for copper, zinc
and iron.
Weight %
Si P Mn Fe Total
Spectrum 1 0.4 0.3 0.28 99.02 100
Spectrum 2 0.36 0.41 0.33 98.91 100
Spectrum 3 0.36 0.36 0.21 99.06 100
Mean 0.37 0.36 0.27 99 100
Std. deviation 0.02 0.05 0.06 0.08
Max. 0.4 0.41 0.33 99.06
Min. 0.36 0.3 0.21 98.91
Table 5.1 Results of SEM analysis of the wrought iron giving compositions of three
areas analysed.
95
5.1.2 Preparation of surfaces
The results of cleaning (Fig. 5.9) reflect the end point goals and limitations of the
methods employed as discussed above. Both aluminium oxide powder and glass
beads blasting exposed the entire metal surface. Crushed walnut blasted, sodium
hydroxide immersed and flame cleaned surfaces were visually similar in their
retention of oxides but differed in colour and extent of powdery corrosion products
evident.
Figure 5.9 Morphology of metal surface post-preparation shown in macroscopic

(upper) and secondary electron SEM (lower) images for each surface preparation
method.
96
Preparation Coherent Powdery Micro- Surface Corresponding
Method oxide corrosion cracks in profile surface
layer products oxide preparation
grade (BS EN ISO
8501-1:2007)
Un-cleaned Yes Yes No Rough Rust Grade D

control
Glass beads No No N/A Undulating D Sa 2.5

blasting
Aluminium No No N/A Rough D Sa 2.5

oxide
blasting
Crushed Yes No No Rough (ice D Sa 1

walnut floes)
blasting
Sodium Yes No No Smooth D St 1

hydroxide/
wire
brushing
Flame Yes Yes Yes Rough (ice Fl (approx.)

cleaning/ floes)
wire
brushing
Table 5.2 Classification of surface preparation outcomes using BS EN ISO 8501-1:2007

(British Standards Institute 2007).
5.1.3 Oxygen consumption and corrosion rates of prepared samples

All samples within all preparation methods, including the unprepared controls,
consumed oxygen during the 350 days of testing (Figs 5.10-5.15). The rates of
consumption differed according to the preparation technique used but patterns of
consumption are similar; an initial fast phase followed by a steady rate. While the
range of oxygen consumption rates within a treatment likely reflects the difficulty of
standardisation and reproducibility for the heritage metal samples, patterns emerge
for oxygen consumption according to preparation method (Figs 5.16, 5.18). Oxygen
consumption is given as a function of surface area (Fig. 5.18, Table 5.3) but it should
be noted that surfaces are not perfectly planar and variation in surface morphology
97
is inherent in the preparation methods (Figs 4.3, 5.9). Flame and wire brush cleaning,
aluminium oxide blasting and glass beads blasting have averages that are faster than
the untreated control samples, whereas both crushed walnut blasting and sodium
hydroxide and wire brush preparations are significantly slower than the untreated
samples. Oxygen consumption values are used to calculate moles of iron theoretically
reacting to form corrosion products (Table 5.3) in accordance with the methodology
discussed in 4.2.6. It is possible to calculate the corresponding loss of depth of
metallic iron across sample surfaces but this is not appropriate for the localised
corrosion seen on these samples (Fig. 5.19).
The data demonstrates broad agreement within preparation methods, except for
two anomalous results; aluminium oxide blasting (HS7) and flame cleaning (HS25)
(Figs 5.10-5.15). Statistically significant differences between corrosion rates of
sodium hydroxide immersed and aluminium oxide blasted samples, sodium
hydroxide immersed and flame cleaned samples and crushed walnut blasted and
flame cleaned samples (Table 5.4).
98
90.00
80.00
Glass Beads Blasted
70.00
Oxygen C onsumed (mbar)
60.00
HS1
50.00
HS2
40.00
HS3
30.00 HS4
20.00 HS5
10.00
0.00
0 50 100 150 200 250 300 350 400
Time (Days)
Figure 5.10 Oxygen consumption of glass beads blasted samples at 90% RH.
160.00
140.00
Aluminium Oxide Blasted
Oxygen Consumed (mbar)
120.00
100.00
HS6
80.00 HS7
HS8
60.00
HS9
40.00
HS10
20.00
0.00
0 50 100 150 200 250 300 350 400
Time (Days)
Figure 5.11 Oxygen consumption of aluminium oxide blasted samples at 90% RH.
99
30.00
Crushed Walnut Blasted
25.00
20.00
HS11
15.00 HS12
HS13
10.00 HS14
HS15
5.00
0.00
0 50 100 150 200 250 300 350 400
Time (Days)
Figure 5.12 Oxygen consumption of crushed walnut blasted samples at 90% RH.
20.00
18.00
Sodium Hydroxide Cleaned
16.00
14.00
12.00 HS16
10.00 HS17
8.00 HS18
6.00 HS19
HS20
4.00
2.00
0.00
0 50 100 150 200 250 300 350 400
Time (Days)
Figure 5.13 Oxygen consumptiom of sodium hydroxide and wire brush cleaned
samples at 90% RH.
100
200.00
180.00 Flame Cleaned

Oxygen Consumed (mbar) 160.00
140.00
120.00 HS21
HS22
100.00
HS23
80.00
HS24
60.00
HS25
40.00
20.00
0.00
0 50 100 150 200 250 300 350 400
Time (Days)
Figure 5.14 Oxygen consumption of flame and wire brush cleaned samples at 90%RH.
50.00
45.00 Un-cleaned
40.00
Oxygen consumed (mbar)
35.00
30.00 HS27
HS28
25.00
HS29
20.00
HS30
15.00 HS31
10.00
5.00
0.00
0 50 100 150 200 250 300
Time (Days)
Figure 5.15 Oxygen consumption of un-cleaned railing samples at 90% RH.
101
Prep Sample O2 Consumption Average O2 Fe Converted Average Fe
2 -8
Method (mol/year/mm x10 ) Consumption to FeOOH Converted to
(mol/year/mm2 (mol/year/mm2 FeOOH
x10-8) x10-8) (mol/year/mm2
x10-8)
HS1 7.76 10.35

HS2 12.16 16.22
HS3 9.16 12.22
Beads
Glass
HS4 8.90 11.87

HS5 14.93 10.58 19.91 14.11
HS6 15.14 20.18
Aluminium
HS7 26.54 15.58 35.38

20.77
HS8 7.64 (12.84 without 10.19 (17.11 without
Oxide
HS9 12.09 anomalous 16.12 anomalous

result)
HS10 16.47 21.97 result)
HS11 1.94 2.58
HS12 3.18 4.24
Crushed
Walnut
HS13 3.03 4.04

HS14 3.17 4.22
HS15 3.98 3.06 5.31 4.08
HS16 3.00 4.00
Hydroxide
HS17 1.67 2.23

Sodium
HS18 1.96 2.61

HS19 1.90 2.53
HS20 2.29 2.16 3.05 2.89
HS21 33.16 44.21
HS22 26.30 26.03 35.06 34.70
HS23 24.86 (28.48 without 33.15 (37.97 without
Flame
HS24 29.60 anomalous 39.47 anomalous

HS25 16.21 result) 21.61 result)
HS27 6.66 8.88
Un-cleaned
HS28 6.57 8.76

HS29 5.93 7.91
HS30 8.06 10.75
HS31 6.30 6.70 8.40 8.94
Table 5.3 Oxygen consumption averaged over test period (mol/year/mm2x10-8) and
theoretical conversion of metallic iron to FeOOH (mol/year/mm2x10-8) of all samples
giving averages by preparation method. All consumption values are given with
consumption of the control vessel (set up without sample) removed. Calculations of
metallic iron conversion assume that all oxygen consumed reacts directly with
metallic iron to form FeOOH and as such is an estimate of the iron consumed since
some oxygen will be involved in redox reactions within the corrosion process.
102
Uncleaned
Flame
Sodium Hydroxide
Crushed Walnut
Aluminium Oxide
Glass Beads
0 0.5 1 1.5 2 2.5

Average Oxygen Consumption (mbar year-1 g-1)
Figure 5.16 Average oxygen consumption of samples prepared by each preparation

method and control samples. The consuption of the control vessel without sample has
been subtracted and the anomalous result for aluminium oxide and flame cleaned
samples.
2.5
2
Gradient of trendline
1.5 Glass Beads

Aluminium Oxide
Crushed Walnut
1 Sodium Hydroxide
Flame
Uncleaned
0.5
0
0 50 100 150 200 250
Days
Figure 5.17 Change in gradient of trendline (i.e. rate) for average oxygen
consumption of samples by surface preparation method. Points denote midpoint of
gradient period.
103
Figure 5.18 Boxplot showing oxygen consumed by samples prepared by each method
(mol/year/mm2 x 10-8) as averaged over test period. The box represents the
interquartile range, the horizontal line within the box denotes the median and the
upper and lower whiskers show the maximum and minimum values. A circle
represents an outlying value (lying between 1.5 and 3 times the interquartile range
from the upper or lower quartile) and an asterisk represents an extreme value (more
than three times the interquartile range from the upper or lower quartile).
104
Significantly different oxygen consumption (mol/year/mm2)
Sodium Hydroxide / Aluminium Oxide
Sodium Hydroxide / Flame
Crushed Walnut / Flame
Table 5.4 Significantly different oxygen consumption per year per mm2 of iron
between surface preparation methods (calculated by Kruskal-Wallis and with a
significance level of 0.05). Sodium Hydroxide / Glass Beads and Flame / Uncleaned
results are close to significantly different but do not fall within the 0.05 significance
level.
105
5.1.4 Visual sample surfaces post high RH environment
Figure 5.19
Prepared Surfaces Post-High RH Exposure Comparison
of prepared
and post-
high RH
GLASS BEADS exposure
BLASTED surfaces.
Images show
sample
surfaces of
30mm x
ALUMINIUM 40mm.
OXIDE
BLASTED
CRUSHED
WALNUT
BLASTED
SODIUM
HYDROXIDE/
WIRE
BRUSHED
FLAME/
WIRE
BRUSHED
UN-CLEANED
106
5.1.5 XRD of corrosion products formed during high RH exposure
All samples subjected to 90% relative humidity testing developed fresh corrosion
products on their surfaces during the oxygen consumption tests (Fig. 5.19). XRD
analysis of samples of the fresh corrosion product identified goethite and
lepidocrocite for all preparation methods and controls except flame cleaning (Figs
5.20 and 5.21). Both goethite and lepidocrocite are products commensurate with
corrosion in the atmosphere (Santarini 2007, 29) while hematite is expected at high
temperature either as a transformation product or as a newly formed corrosion
product (Cornell and Schwertmann 2003). Chloride contamination of samples is
thought to be low as evidenced by an absence of akaganeite in the analysis (Réguer
et al. 2007; Zucci et al. 1977).
107
Counts
HS1 corrosion products
8000
6000
4000
2000
0
10 20 30 40 50 60 70
Position [°2Theta]
a)
Peak List
00-044-1415
00-029-0713
20 30 40 50 60 70 80 90
Position [°2Theta]
b)
Figure 5.20 a) Diffractogram of corrosion products from glass beads cleaned sample;
b) patterns of identified corrosion products (uppermost pattern) indicating presence
of lepidocrocite (00-044-1415 (middle pattern)) and goethite (00-029-0713 (lowest
pattern)).
108
Counts
HS21 corrosion products
8000
6000
4000
2000
0
10 20 30 40 50 60 70
Position [°2Theta]
a)
Peak List
01-089-0598
01-081-0464
20 30 40 50 60 70 80 90
Position [°2Theta]
b)
Figure 5.21 a) Diffractogram of corrosion products from flame cleaned sample; b)

patterns of identified corrosion products (uppermost pattern) indicating presence of
hematite (01-089-0598 (middle pattern)) and goethite (01-081-0464 (lowest
pattern)).
109
5.1.6 Practitioner flame cleaned samples
5.1.6.1 Oxygen consumption
All the practitioner flame cleaned samples consumed oxygen within the test period.
All (Figs 5.22-5.24) show the same general trend of fast initial consumption rate and
then a slowing down evidenced by the levelling off of the consumption graphs similar
to the laboratory prepared samples (Fig. 5.14). Use of different sample material
precludes direct comparison of laboratory and practitioner flame cleaned samples.
When compared to Sa2.5 aluminium oxide blasted samples and wire brushed
samples of the same gasometer material (Figs 5.25, 5.26), the Practitioner C (PC)
corrosion rate is very similar to the wire brushed substrate but Practitioner A (PA)
and Practitioner B (PB) are significantly higher.
Practitioner A (PA)
90.00
80.00
70.00
PA1
60.00
PA2
50.00 PA3
PA4
40.00
PA5
30.00 PA6
PA7
20.00
PA8
10.00
0.00
0 20 40 60 80 100 120
Time (days)
Figure 5.22 Oxygen consumption of samples flame cleaned by Practitioner A and

subsequently exposed to 90% RH.
110
Practitioner B (PB)
90.00
Oxygen Consumed (mbar) 80.00
70.00 PB1
60.00 PB2
50.00 PB3
40.00 PB4
30.00 PB5
20.00 PB6
10.00 PB7
0.00 PB8
0 20 40 60 80 100 120
Time (days)
Figure 5.23 Oxygen consumption of samples flame cleaned by Practitioner B and

Practitioner C (PC)
25.00
20.00
PC1
PC2
15.00
PC3
PC4
10.00
PC5
PC6
5.00
PC7
0.00 PC8
0 20 40 60 80 100 120
Time (days)
Figure 5.24 Oxygen consumption of samples flame cleaned by Practitioner C and

111
Figure 5.25 Boxplot showing oxygen consumption (mol/year/mm2 x 10-8) of samples
flame cleaned by three heritage practitioners (PA, PB and PC) and the wire brushed
gasometer substrate. The box represents the interquartile range, the horizontal line
within the box denotes the median and the upper and lower whiskers show the
maximum and minimum values. A circle represents an outlying value (lying between
1.5 and 3 times the interquartile range from the upper or lower quartile) and an
asterisk represents an extreme value (more than three times the interquartile range
from the upper or lower quartile).
112
Figure 5.26 Boxplot showing theoretical metallic iron converted to FeOOH
(mol/year/mm2 x 10-8) of samples flame cleaned by three heritage practitioners (PA,
PB and PC) and wire brushed gasometer substrate. The box represents the
5.1.7 Scanning Electron Microscopy

Backscatter SEM images of uncleaned and laboratory flame cleaned gasometer
sample material show a distinct fragmentation and de-adhesion of oxide layers
opening channels to the metal surface (Figs 5.27-5.28).
113
of gasometer wrought iron showing characteristic slag inclusions and corrosion
product layers. Scale bar 200µm.
of gasometer wrought iron after flame cleaning showing fragmentation of corrosion
product layers. Scale bar 200µm.
114
5.2 Investigating protective coatings for historic wrought iron
5.2.1 Gasometer sample characterisation
Slag inclusions are visible in transverse and longitudinal section (Fig. 5.29) showing
rolling and piling of the wrought iron during its production. Inclusions are between
10-200µm and distributed in stringers in the direction of rolling. Fewer and smaller
slag inclusions are present than in the railing sample material (Fig. 5.1) indicating a
greater extent of working. Phosphorous is present (Table 5.5) although %
composition is likely skewed by contamination by copper. Sulfur and chloride are
indicated in the oxide layer consistent with atmospheric pollutants in the urban
environment. Corrosion products identified by XRD (Fig. 5.31-5.33) of lepidocrocite,
goethite and magnetite are also consistent with atmospheric corrosion.
of gasometer wrought iron showing characteristic slag inclusions.
115
Figure 5.30 Backscatter SEM image showing location of spectra for analysis of
gasometer wrought iron.
Spectrum O P S Cl Fe Cu Total
Spectrum 1 Alloy 0.55 99.15 0.3 100
Spectrum 2 Oxides 32.57 0.24 0.28 66.56 0.35 100
Spectrum 3 Oxides 30.88 68.7 0.42 100
Spectrum 4 Oxides 29.86 0.29 0.96 0.18 68.26 0.45 100
Max. 32.57 0.55 0.96 0.18 99.15 0.45
Min. 29.86 0.24 0.28 0.18 66.56 0.3
Table 5.5 Results of quantitative SEM analysis of gasometer wrought iron.
116
Counts
MS1 ENNE
30000
20000
10000
0
10 20 30 40 50 60
Position [°2Theta]
a)
Peak List
00-044-1415
10 20 30 40 50 60 70 80 90
Position [°2Theta]
b)

of lepidocrocite (00-044-1415 (lower pattern)).
117
Counts
MS1 ENNE
30000
20000
10000
0
10 20 30 40 50 60
Position [°2Theta]
a)
Peak List
00-029-0713
10 20 30 40 50 60 70 80
Position [°2Theta]
b)

of goethite (00-029-0713 (lower pattern)).
118
Counts
MS1 ENNE
30000
20000
10000
0
10 20 30 40 50 60
Position [°2Theta]
a)
Peak List
01-080-0390
10 20 30 40 50 60 70 80 90
Position [°2Theta]
b)

of magnetite (01-080-0390 (lower pattern)).
119
5.2.2 Prepared samples
Images of samples after surface preparation and addition of each coating are given
(Fig. 5.34). Contrasting primer and build colours are evident allowing completeness
of coverage to be ascertained. The high build of System A obscures the substrate
surface morphology which is retained with Coating C. System B reduces visibility of
morphology through settlement of slow curing topcoat into the pits and troughs of
the surface. System A has a bulky, plastic appearance, System B a very high gloss and
Coating C a matte finish and close conformation to the substrate surface.
120
Figure 5.34 Prepared samples showing uncoated substrates and samples after
application of each coating constituting the system. In the case of Coating C, there is
only one coating applied.
121
5.2.3 Coating properties
5.2.3.1 Coating properties during application
Coating properties during application influence success of the cured coating by
dictating ease of application. The properties of System A (e.g. viscosity and
completeness of curing) are dependent on correct mixing ratio of the two pack
system. Its properties change over time during application as the curing reaction
proceeds. Observations of application properties are given in Table 5.6. Curing of all
coatings within System A was quicker than oil-based System B and Coating C coatings.
122
Coating Viscosity Brush strokes/holidays Curing Comments
Primer Initially low, Significant pooling and run back Longer curing Need to make up in
becoming from edges although less on time than correct ratio of two
more viscous the iron substrate than the suggested by parts – inconvenient
as reaction glass. manufacturer and scope for mistakes.
progresses Longer curing –
problem with two-pack
ratio? Bubbles from
stirring to mix – leave
to stand. Difficult to
clean brushes – time
and expense of
resources.
System A
Primer/ Initially low, Significant pooling and run back Slightly longer As above.
Build becoming from edges- high gloss of cured than primer but
more viscous primer exacerbates problem. much quicker Noticeable settling into
as reaction Coverage and adhesion better than the oil based pits in surface – leaves
progresses on second application – coatings. peaks with less depth of
reaction had progressed coating.
further and polymer was more
viscous. Need to overcoat
within 7 days.
Topcoat Initially low, Very similar colour to Longer than As Primer.

becoming Primer/Build – brush strokes epoxy primer and
more viscous difficult to discern. Not build but much
as reaction significant run back. quicker than the
progresses oil based
coatings.
Primer Low Not pronounced on iron Rel. slow. Min. 6 Low viscosity means
substrate, more so on glass hours between more, thinner coats
despite blasting to provide key. coats. 24 to which may minimise
overcoating. holidays and slumping.
Topcoat Low Brush strokes visible Very slow. Still Bubbles created by
System B
immediately but low viscosity very tacky after a stirring transfer to

and slow curing allow levelling few days. sample – leave to stand
out for good coverage. Covers but not long enough to
primer relatively well but some settle out.
run back from corners and
edges. Second coat adheres
much more effectively despite
high gloss – good adhesion
possibly due to high tack.
Coating C Very low More brush strokes and run Touch dry more Difficult to observe
back on both substrates than quickly than holidays with the
experienced with System B. System B. second coat due to lack
Partly an effect of the surface of contrast in
prep method for the iron appearance between
substrate – wire brushing coats.
creates a polished surface with
little keying. Second coat
adheres more successfully.
Table 5.6 Properties of coatings during application.
123
5.2.3.2 Properties of cured coatings
Properties of the cured coatings influence and offer insight into their protective
performance and aesthetic suitability for heritage contexts. They are summarised in
Table 5.7 and illustrated in Figures 5.35-5.45. Holidays are access sites for ingress of
water and aggressive species and are evident in System A build and topcoat, System
B primer and Coating C. Brushstrokes indicate uneven build and of coatings and
consequent uneven protection in barrier coatings where length of the diffusion
pathway will be greater in peaks and shorter in troughs. These are evident in System
A build and topcoat where they add to the plastic aesthetic and on System B primer
where they are eventually obscured by settlement of slow-curing topcoat. Run back
from edges is evident with Coating C and from edges and peaks with System B. These
areas will be afforded less protection by the coatings.
The high build (Table 5.8) of System A (210 – 406µm) leads to a bulky appearance
which obscures surface morphology of the wrought iron. Although System B build is
lower (31 – 98µm) surface morphology is still obscured by the coating through
settlement of coating into pits. Its appearance is characterised by the high gloss of
the topcoat. Closely following the contours of the substrate, Coating C has the lowest
build (DFT 11 – 64µm) as expected of the coating with fewest coats applied and a
matte appearance.
Dry film thicknesses (DFT) show a wide range (Fig. 5.46) despite frequent
recalibration of the meter to counter drift. This may be due to relative thicknesses of
peaks and troughs of brushstrokes such as evident on System A samples (Fig. 5.37)
which have the greatest range of thicknesses with outlying values. Different
thicknesses of System B may be due to settlement of coating into pits and run back
from peaks which have consequently lower thicknesses (Fig. 5.42). Coating C DFT
measurements were calibrated on the uncoated substrate material with oxide layers
which are measured by the meter as a coating. Variable thickness of the oxides may
account for some of the range of Coating C DFT values.
124
Coating Holidays visible Brushstrokes Runback Aesthetic of
(macroscopic) visible from edges complete
(macroscopic) and peaks system
visible
Primer Bubbles visible No No High gloss

on all samples
(3-10 per cm2) Plastic
particularly in Bulky
pits (Fig. 5.35).
Build No Yes – more No

noticeable on
System A
less pitted
reverse of
samples (Fig.
5.37).
Topcoat Pinholes from Yes (Fig. 5.38) No

burst bubbles
(1-2 per sample
12cm2).
Primer Pinholes visible Yes (Fig. 5.40) No High gloss

on most
samples (1-10 Surface
per cm2) (Fig morphology
5.39). of metal
visible
System B
Build No pinholes. No All samples

Peaks and show run
runback create back from
holidays on edges and
most samples. peaks
revealing
primer (Fig.
5.42).
Coating C Yes – mostly No Noticeable Matte

visible in more thinner
pitted areas and coating at Surface
where a greater edges and morphology
thickness of corners of metal
coating has (Figs 5.44, visible
been applied 5.45).
(Fig. 5.43).
Table 5.7 Properties of cured coatings.
125
Figure 5.35 Sample HSII08 showing bubbles in System A primer after curing (scale of
background squares 10 x 10mm).
Figure 5.36 Sample HSII04 showing pinhole holidays in cured System A build coating
(scale of background squares 10 x 10mm).
126
Figure 5.37 Sample HSII01 showing brushstrokes in cured System A build coating
(scale of background squares 10 x 10mm).
Figure 5.38 Sample HSII08 showing pinholes in cured System A topcoat (scale of
127
Figure 5.39 Sample HSII16 showing pinhole holidays in cured System B primer (scale
of background squares 10 x 10mm).
Figure 5.40 Sample HSII13 showing brushstrokes in cured System B primer (scale of
background square 10 x 10mm).
128
Figure 5.41 Sample HSII18 showing peaks run back of System B topcoat from peaks
revealing red primer (scale of background squares 10 x 10mm).
Figure 5.42 Sample HSII11 showing run back of System B topcoat from peaks and
edges (sample area shown 30mmx4mm).
129
Figure 5.43 Sample HSII26 showing pinholes in Coating C after curing (scale of
Figure 5.44 Sample HSII28 showing run back of cured Coating C from edges (sample
area shown 40mmx4mm).
Figure 5.45 Sample HSII22 showing areas missed during application of Coating C to
sample edge (sample area shown 40mmx4mm).
130
5.2.3.3 Dry film thicknesses (DFT)
Coating
System A System B Coating C
338 72 36
332 31 62
406 98 34
280 23 33
Coating DFT (micron)
392 66 22
322 80 38
210 52 48
316 64 45
346 46 62
340 88 64
300 44 11
318 36 58
Nominal 325 58 43
(average) DFT
(micron)
Table 5.8 Dry film thicknesses of cured coatings.
131
Figure 5.46 Boxplot showing the dry film thicknesses of the cured coatings. The box
represents the interquartile range, the horizontal line within the box denotes the
median and the upper and lower whiskers show the maximum and minimum values.
A circle represents an outlying value (lying between 1.5 and 3 times the interquartile
range from the upper or lower quartile) and an asterisk represents an extreme value
(more than three times the interquartile range from the upper or lower quartile).
5.2.3.4 Pull-off adhesion testing

Adhesion pull-off values are given in Table 5.9 and Figure 5.47 and modes of failure
in Table 5.9 and illustrated in Figure 5.48. For each coating system the mode of failure
was consistent between the test dollies. Coating C values are significantly lower than
Systems A and B and its mode of failure is cohesive within the corrosion product layer
rather than a failure of the coating. The value for adhesion of Coating C to the oxide
layer must exceed these values. System A adhesion values are slightly higher than
System B on the whole although there is an overlap in values. System A failure is
consistently cohesive within the primer layer. System B failure is adhesive between
the primer and the topcoat. With neither system is the substrate exposed after pull-
off hence the adhesion value of the primers to the substrates must exceed these
132
values. There is good agreement between the pull-off values between samples of
each coating.
Adhesion Pull-off Value (MPa)
Dolly System A System B Coating C
1 11.19 9.47 4.77
2 12.50 8.95 2.93
3 11.00 8.90 4.92
4 11.73 8.61 3.61
5 12.81 8.24 4.41
6 10.96 10.93 4.66
7 12.34 10.78 4.41
8 12.09 10.42 4.30
9 10.40 10.25 4.87
10 12.90 10.66 4.26
Mode Cohesive within Adhesive Cohesive within

of primer. between primer corrosion
Failure and topcoat. product layer.
Table 5.9 Adhesion pull-off values (MPa) and modes of failure.
133
Figure 5.47 Boxplot showing adhesion pull-off values for coatings. The box represents
the interquartile range, the horizontal line within the box denotes the median and the
134
a) b)
c) d)
e) f)
Figure 5.48 Pull-off dollies (left) and corresponding areas of coated samples (right)
showing modes of failure. Dollies and pull-off areas have a diameter of 20mm. a), b)
System A with failure cohesive within the primer; c), d) System B with failure mostly
adhesive between primer and topcoat; e), f) Coating C with failure cohesive within
corrosion products.
135
5.2.4 Oxygen consumption and corrosion rates of coated samples
All coated samples consumed oxygen during the test period. Raw consumption
graphs for all coated metal samples are given in Figures 5.49-5.51 and for coated
control samples (glass substrates) in Figures 5.52-5.54. Agreement between the
samples is good indicating a level of standardisation of sample preparation and
coating application. The majority of oxygen consumption of the coated metal
samples is accounted for by the consumption of the coating controls.
Oxygen consumption rates fall off after an initial high consumption rate for all
samples. This is most pronounced with System A epoxy coating. Alkyd oil-based
System B and Coating C exhibit greater oxygen consumption overall as expected in
relation to their chemistry and oxidative cross-linking. Oxygen consumption by the
coatings may be a predominantly surface phenomenon or a bulk process from within
the coating. Below a critical pigment volume concentration, curing of alkyd coatings
proceeds along a microscopic reaction front limited by diffusion of oxygen; above
this concentration, reaction appears to be homogeneous (Erich et al. 2008). As the
pigment volume concentration of these coatings is not available from the
manufacturer, both manners of curing must be considered. Average oxygen
consumption of the controls for each coating system have been subtracted from the
coated wrought iron sample consumptions according to surface area of the coatings
(Table 5.10, Figs 5.55 & 5.56) and by mass of coatings applied (Table 5.11, Figs 5.57
& 5.58) for comparison.
By both calculation methods, Coating C samples consumed no oxygen beyond the

error of the oxygen meter. System A sample results were most affected by the
method of calculating coating consumption due to their greater mass of coating. By
both calculation methods, oxygen consumption was minimal but more so when
calculated by mass of coating. System B sample results show oxygen consumption
beyond that of the coating indicating corrosion of the underlying metal substrate.
When compared to the corrosion rate of the uncoated substrate sample material
(Figs 5.56 & 5.57), whether calculated by mass or surface area, System A and Coating
C significantly reduce corrosion rate of the wrought iron (Table 5.12). System B also
136
reduces the corrosion rate but not to a statistically significant degree. In keeping with
surface preparation testing results (5.13), uncoated samples with retained oxides
exhibited a slower corrosion rate than aluminium oxide Sa2.5 blasted samples but
blasting and applying System A also offers a reduced corrosion rate relative to
uncoated substrate with retained oxides.
45.00
40.00
System A
HSII 01
35.00 HSII 02
Oxygen Consumption (mbar)
30.00 HSII 03
25.00 HSII 04
20.00 HSII 05
15.00 HSII 06
10.00 HSII 07
5.00 HSII 08
0.00
0 100 200 300 400 500 600
Time (days)
System A and subsequently exposed to 90% RH.
137
100.00
90.00
System B
HSII 11
80.00
HSII 12
70.00
HSII 13
60.00
HSII 14
50.00
HSII 15
40.00
HSII 16
30.00
20.00 HSII 17
10.00 HSII 18
0.00
0 100 200 300 400 500 600
Time (days)
System B and subsequently exposed to 90% RH.
70.00
Coating C
60.00 HSII 21
HSII 22
50.00
HSII 23
40.00
HSII 24
30.00 HSII 25
HSII 26
20.00
HSII 27
10.00 HSII 28
0.00
0 100 200 300 400 500 600
Time (days)
Figure 5.51 Oxygen consumption of wrought iron samples coated with Coating C and
138
50.00
45.00
System A Controls
HSII 51
Oxygen Consumption (mbar) 40.00 HSII 52
35.00
HSII 53
30.00
HSII 54
25.00
HSII 55
20.00
HSII 56
15.00
HSII 57
10.00
HSII 58
5.00
0.00
0 100 200 300 400 500 600
Time (days)
A and subsequently exposed to 90% RH.
70.00
System B Controls
60.00
50.00 HSII 61
HSII 62
40.00 HSII 63
HSII 64
30.00
HSII 65
20.00 HSII 66
HSII 67
10.00
HSII 68
0.00
0 100 200 300 400 500 600
Time (days)
B and subsequently exposed to 90% RH.
139
70.00
Coating C Controls
60.00
Oxygen Consumption (mbar) HSII 71

50.00
HSII 72
40.00 HSII 73
HSII 74
30.00
HSII 75
20.00 HSII 76
HSII 77
10.00
HSII 78
0.00
0 100 200 300 400 500 600
Time (days)
Figure 5.54 Oxygen consumption of glass control samples coated with Coating C and
140
Coatings Sample O2 Average O2 Fe Converted to Average Fe
Consumption Consumption FeOOH Converted to
(mol/year/mm2 (mol/year/m (mol/year/mm2 FeOOH
x10-8) m2x10-8) x10-8) (mol/year/mm2
x10-8)
HSII01 0.40 0.53
HSII02 1.42 1.90
HSII03 0.35 0.47
System A
HSII04 0.32 0.43

HSII05 -0.26 -0.35
HSII06 0.62 0.83
HSII07 0.41 0.55
HSII08 0.06 0.45 0.08 0.60
HSII11 1.50 2.00
HSII12 0.60 0.79
HSII13 1.90 2.53
System B
HSII14 0.13 0.17

HSII15 2.70 3.60
HSII16 0.94 1.26
HSII17 2.43 3.24
HSII18 2.07 1.53 2.77 2.05
HSII21 -1.94 -2.59
HSII22 -1.30 -1.73
HSII23 -0.02 -0.03
Coating C
HSII24 -0.12 -0.16

HSII25 -2.64 -3.52
HSII26 -0.79 -1.05
HSII27 -1.25 -1.67
HSII28 -1.69 N/A -2.26 N/A
HSII31 5.44 7.25
Wire Brushed Control
HSII32 5.47 7.30

HSII33 5.07 6.76
(uncoated)
HSII34 5.29 7.06

HSII35 5.72 7.63
HSII36 5.40 7.20
HSII37 7.14 9.52
HSII38 4.59 5.52 6.12 7.36
HSII41 7.30 9.73
HSII42 6.86 9.15
Sa2.5 Control
HSII43 5.52 7.36

(uncoated)
HSII44 3.84 5.12

HSII45 7.26 9.68
HSII46 8.56 11.41
HSII47 6.36 8.48
HSII48 9.256.87 12.34 9.16
giving moles of oxygen consumed and theoretical moles of Fe converted to FeOOH
per year per mm2 x10-8. Values have been calculated by subtracting oxygen
consumption of the control samples of the coatings according to surface area of the
coating applied to the wrought iron samples.
141
coatings has been calculated using the surface area of coating applied. The box
142
consumption of the coatings has been calculated using the surface area of coating
applied. The box represents the interquartile range, the horizontal line within the box
denotes the median and the upper and lower whiskers show the maximum and
minimum values. A circle represents an outlying value (lying between 1.5 and 3 times
the interquartile range from the upper or lower quartile) and an asterisk represents
an extreme value (more than three times the interquartile range from the upper or
lower quartile).
143
Coating Sample O2 Average O2 Fe Converted to Average Fe
Consumption Consumption FeOOH Converted to
(mol/year/ (mol/year/ (mol/year/mm2 FeOOH
mm2x10-8) mm2x10-8) x10-8) (mol/year/
mm2x10-8)
HSII01 -0.33 -0.44
HSII02 0.47 0.63
HSII03 -0.37 -0.49
System A
HSII04 0.40 0.53

HSII05 -0.47 -0.63
HSII06 0.16 0.22
HSII07 0.46 0.62
HSII08 0.09 0.05 0.11 0.07
HSII11 0.53 0.71
HSII12 0.36 0.48
HSII13 1.55 2.07
System B
HSII14 -0.16 -0.22

HSII15 2.56 3.41
HSII16 1.30 1.73
HSII17 2.49 3.32
HSII18 2.72 1.42 3.63 1.89
HSII21 -0.92 -1.22
HSII22 -1.52 -2.03
HSII23 0.55 0.73
Coating C
HSII24 -0.27 -0.37

HSII25 -0.91 -1.22
HSII26 0.31 0.41
HSII27 0.01 0.01
HSII28 -0.33 N/A -0.44 N/A
HSII31 5.44 7.25
HSII32 5.47 7.30
Wire Brushed
(uncoated)
HSII33 5.07 6.76

Control
HSII34 5.29 7.06

HSII35 5.72 7.63
HSII36 5.40 7.20
HSII37 7.14 9.52
HSII38 4.59 5.52 6.12 7.36
HSII41 7.30 9.73
HSII42 6.86 9.15
Sa2.5 Control
HSII43 5.52 7.36

(uncoated)
HSII44 3.84 5.12

HSII45 7.26 9.68
HSII46 8.56 11.41
HSII47 6.36 8.48
HSII48 9.25 6.87 12.34 9.16
giving moles of oxygen consumed and moles of metallic iron converted to FeOOH per
year per mm2 x10-8. Values have been calculated by subtracting oxygen consumption
of the control samples of the coatings according to the mass of coating applied to the
wrought iron samples.
144
coating is calculated using the mass of coating applied. The box represents the
145
consumption of the coatings is calculated using the mass of coating applied. The box
Significantly different oxygen consumption and Fe converted to FeOOH

(mol/year/mm2)
Coating oxygen consumption Coating oxygen consumption

calculated by mass calculated by surface area
Coating C / Uncoated Wire Coating C / Uncoated Wire Brushed

Brushed
System A / Uncoated Sa2.5
System A / Uncoated Sa2.5
System A / Uncoated Wire Brushed
Table 5.12 Significantly different oxygen consumption and theoretical metallic iron
conversion to FeOOH (mol/year/per mm2 of iron) between coated samples and
uncoated substrates (calculated by Kruskal-Wallis and with a significance level of
0.05).
146
Following 90% RH exposure testing, System A samples showed damage to the
polyurethane topcoat caused by small movements of the samples within the reaction
vessels (Figs 5.59, 5.60). Notable on System B samples was particulate pick-up (Fig.
5.61) but no other damage. Coating C samples appeared unchanged even where the
coating had a low build at edges (Fig. 5.63).
Figure 5.59 Sample HSII02 after exposure to 90%RH for 540 days showing damage to
corners but otherwise unchanged appearance.
Figure 5.60 Sample HSII02 showing damage to topcoat of System A after 90%RH
exposure (scale of background scale 10mm between white lines).
147
Figure 5.61 Sample HSII11 showing adhesion of particulates to the surface of System
B samples (scale of background scale 10mm between white lines).
Figure 5.62 Sample HSII17 after exposure to 90%RH showing particulate pickup and
run back of top coat from edges and peaks (sample area shown 30mm x 4mm).
Figure 5.63 Sample HSII26 after exposure to 90%RH for 540 days showing no visible
corrosion even where coating build is very low on edges (sample area shown 40mm x
4mm).
148
5.3 Comparing sample materials
Comparison of railing and gasometer sample material is permitted using the
corrosion rate data for the aluminium oxide Sa2.5 blasted samples of both materials.
The railing wrought iron demonstrates a higher corrosion rate overall and a greater
range of values but ranges overlap between the two materials. Boxplots of oxygen
consumed and iron converted (assuming [6] is the predominant reaction) are given
(Figs 5.64, 5.65).
Figure 5.64 Boxplot showing oxygen consumed by railing and gasometer wrought
iron samples (mol/year/mm2 x 10-8) as averaged over test period. The box represents
the interquartile range, the horizontal line within the box denotes the median and the
149
Figure 5.65 Boxplot showing theoretical metallic iron converted to FeOOH by railing
and gasometer wrought iron samples (mol/year/mm2 x 10-8) as averaged over test
period. The box represents the interquartile range, the horizontal line within the box
denotes the median and the upper and lower whiskers show the maximum and
minimum values. A circle represents an outlying value (lying between 1.5 and 3 times
the interquartile range from the upper or lower quartile) and an asterisk represents
an extreme value (more than three times the interquartile range from the upper or
lower quartile).
150
6 Discussion
6.1 Surface Preparation Methods

6.1.1 Grading rust by pictorial standards within heritage contexts
Within heritage conservation it is challenging to define end points for cleaning
methods that can be applied across the sector for specifying treatment
methodologies to contractors. Swedish Standard SIS 05 59 00
(Standardiseringskommissionen i Sverige et al. 1980) and BS EN ISO 8501-1:2007
(British Standards Institute 2006/2007) specify surface preparations necessary for
the protective performance of coatings by determining the preparation end point
visually using pictorial grading relative to the original condition of the metalwork.
Since they are based on modern steels, pre-treatment surfaces are categorised
according to their degree of pitting corrosion. These standards can be used to classify
test samples since they employ preparation methods that match those used here:
blast-cleaning, hand and power tool cleaning and flame cleaning from four pre-
preparation rust grades. However, their use offers challenges for practitioners
treating historic wrought iron, which is likely to display more advanced corrosion that
is lamellar (Fig. 2.13) rather than the pitting normally evident in the modern steels
used to create the standards. How an untreated rusted surface is graded using the
standard will have a direct impact on judging the surface aesthetic used to decide the
end point of a preparation method (British Standards Institute 2006/2007). While
this undermines the value of the pictorial references provided in the standard,
because of their limited relevance to wrought iron, in practice, application of an
international standard is expected of contractors fulfilling specifications for
treatment of wrought iron within the heritage sector.
The importance of using visual comparators to describe the level of preparation

attained is underlined by the broad similarity in appearance of the samples prepared
here. Differentiation is clearly difficult using only written descriptors therefore an
internationally accepted standard provides a base line that can be used comparably
across the heritage sector. Despite its limited relevance to the wrought iron, it
enables clear specification for contractors and offers comparability between
151
experimental studies carried out within the heritage sector. The grade within the
standard that offers the best match for historic wrought iron is D ‘Steel surface on
which the mill scale has rusted away and on which general pitting is visible under
normal vision’ (British Standards Institute 2006/2007). The severe limitations of using
BS EN ISO 8501-1:2007 illustrate that employing a dedicated classification based on
wrought iron in heritage contexts would be preferable. Developing such a standard
may be a consideration for the future, since the wrought iron heritage bank is
growing and presents ever greater preservation problems that require predictive
management. Generating the standard would be challenging as a corrosion
classification would need to be developed from samples subjected to real-time aging
and this may take tens of years.
6.1.2 Classification and comparison of prepared surfaces

Using the un-cleaned control sample as a comparator to identify the end point of the
surface preparation methods, macro and microscopic investigation of surfaces
reveals significant changes have occurred (Fig. 5.9, Table 5.2). Glass beads and
aluminium oxide blasted surfaces produce the most change by removing the oxide
layer entirely. This correlates to grade D Sa 2.5 ‘very thorough blast-cleaning’ or D Sa
3 ‘blast-cleaning to visually clean steel’ being free of visible oil, grease, dirt, mill scale,
rust, paint coatings and foreign matter and having a more or less uniform metallic
colour (British Standards Institute 2006/2007).
Crushed walnut blasted surfaces are noticeably darker in appearance than the un-
cleaned controls due to exposure of the dense, coherent magnetite (diffraction code
01-080-0390) layer following removal of overlying loosely adhering and powdery
corrosion products and any vestiges of coatings. The outcome corresponds to D Sa 1
‘light blast-cleaning’ being free from visible oil, grease, dirt, poorly adhering mill
scale, rust, paint coatings and foreign matter but retaining the closely adhering mill
scale (British Standards Institute 2006/2007).
Samples prepared by sodium hydroxide immersion followed by manual wire brushing

are markedly more polished and lustrous than the un-cleaned samples. Powdery
corrosion products and remains of coatings have been removed and a coherent
152
corrosion product layer is the end point. The surface preparation level corresponds
to D St 2 ‘thorough hand and power tool cleaning’ being free from visible oil, grease,
dirt, poorly adhering mill scale, rust, paint coatings and foreign matter (British
Standards Institute 2006/2007).
Flame cleaning causes the least change in some respects. No residues of coatings
remain but the surface retains powdery corrosion products. A change in colour is
noticeable; the corrosion products have gained a brighter, orange-red hue. The
preparation level cannot be said to correlate to D Fl as it is not free from mill scale
and rust, possibly due to wire-brushing being manual rather than by the specified
power tool procedure in the standard (British Standards Institute 2006/2007). The
British Standards relate to modern steels and the mill scale referred to is the oxide
layer produced quickly during hot rolling which is reported to be loosely adherent in
the long term and prone to cracking and flaking if the metal flexes (British Standards
Institute 1998). The corrosion products existing on historic wrought iron may have
been produced over long periods and can be closely adhering and relatively stable
physically.
6.1.3 Morphological interpretations

6.1.3.1 Blasting techniques
Spherical glass bead media peen surfaces to produce a satin finish with undulations
that are visible at high magnification (Fig. 5.9). This contrasts with the matte surface
of the aluminium oxide blasted samples caused by the angular roughness that is
evident under high magnification. Both glass beads and aluminium oxide removed all
oxide that was visible to the naked eye. The crushed walnut blasted surfaces retain a
coherent oxide layer but have lost their paint residues. This oxide layer shows a
degree of surface roughness and some angularity which is due to plate-like
laminations rather than a micro roughness produced by the cutting abrasive action
of impinged media, as with aluminium oxide blasting. Since the hardness of glass
beads on the Mohs scale (6) is similar to magnetite (5.5-6.5), corrosion product
removal occurs at a slower rate than with aluminium oxide media, providing less
danger of underlying metal removal. This degree of control is attractive for heritage
practitioners but is less cost effective in terms of operator time and materials. In
153
contrast, the hard, angular aluminium oxide (8-9 on Mohs scale) cuts the surface of
the corrosion products on impact, creating a more roughened, angular surface
profile. This requires control and vigilance to prevent it removing underlying metal
substrate. Despite their angularity, the low hardness (3-4) of crushed walnuts limits
their effect to removal of unsupported laminations, powdery corrosion products and
polymeric coatings. This controlled and less aggressive nature of crushed walnut is
reflected in its industrial use for polymer preparation and in heritage conservation
for surface cleaning of copper alloys (Lins 1992).
6.1.3.2 Sodium hydroxide immersion with manual wire-brushing

The sodium hydroxide solution (1.25M) solvated the polymeric coatings on the
surfaces of the samples but the corrosion products remained insoluble at this pH
(Cornell and Schwertmann 2003). Immersion facilitates the removal of powdery and
loosely adhering oxides by wire-brushing, producing a surface that appears polished
to the naked eye and relatively flat and smooth under high magnification. The areas
of slight roughness bear some similarity to the plate-like laminations visible on the
crushed walnut and flame cleaned samples with minimal disruption of the surface.
6.1.3.3 Flame cleaning with manual wire-brushing

The polymeric coatings adhering to the samples were removed by combustion during
flame cleaning. Removal of large laminations of corrosion products by the abrasive
action of vigorous wire-brushing was possible but both powdery and adherent
corrosion products remain clearly visible. Under high magnification the surface
appears rough in a manner similar to the crushed walnut blasted surfaces. The oxide
layers are fractured, possibly by differential expansion of metal and oxides during
heating leading to loss of laminations. Micro-cracks are visible in some areas and
break-up of the oxide layers is evident in section (Figs 5.9, 5.28).
6.1.4 Suitability of surfaces for application of protective coatings

Removal of corrosion accelerating contaminants and creation of a mechanical key for
adhesion through roughening of the surface are critical for performance of protective
coatings on prepared substrates (Higgins et al. 2010; Funke 1985). Of the treatments
tested here, blasting with glass beads or aluminium oxide either to Sa 2.5 or Sa 3 is
154
likely to advantageously remove the most contaminants, due to the thorough
removal of corrosion products. The angular roughening of the surface profile evident
on aluminium oxide blasted surfaces offers better adhesion relative to the undulating
profile of glass beads peened surfaces.
Pull-off testing of coatings on blasted and un-blasted surfaces report un-blasted

surfaces had the lowest adhesion strengths, with failure of coating adhesion on
blasted surfaces occurring between or within coating layers rather than between
coating and substrate (Momber and Wong 2005) as was found in this study.
However, large peak-to-trough height differentials on surfaces allow rogue peaks to
penetrate the coating leading to holidays (Higgins et al. 2010). Clearly the dry
thickness of a coating and its application rheology will be of importance in relation
to the impact of peak height on coating performance. For aesthetic reasons, the
heritage sector often utilises thin, low concentration and transparent acrylic coatings
to protect surfaces (Siatou et al. 2007), which means the deep keying supplied by
aluminium oxide may prove a challenge to producing a continuous coating with an
even thickness, as relatively peak height will be large in proportion to coating
thickness. This merits further study as it has major impact on matching coatings to
surface preparation and is an area that has received no research in heritage contexts.
Although no impacted particles of blast media were identified on SEM images of

prepared samples, this effect has been noted in unpublished tests commissioned by
Historic Scotland (Lyn Wilson pers. comm.). Blasting angle was modified here to
minimise occurrence of impacted media but were this not possible in practice,
residual blast media may have a negative effect on coating adhesion.
Problems arise with the techniques which retain the oxide layer, as they are
incapable of producing the minimum Sa 2.5 preparation level generally required by
manufacturers of coatings that are not formulated specifically for surface tolerance.
Removal of the coherent oxide layer is not always either ethically or aesthetically
desirable in a heritage context and its retention is not an uncommon requirement.
Of the three techniques retaining oxides within this study, blasting with crushed
walnut and flame cleaning appear to produce surfaces offering more mechanical
155
keying for adhesion, although this effect may be reduced by the powdery oxides
remaining on the flame cleaned samples producing future failure in adherence of the
corrosion layer. The smooth and polished surfaces of the sodium hydroxide
immersed samples have little potential for mechanical keying but removal of
powdery corrosion products may mean fewer contaminants to promote corrosion
under coatings and adherence of the corrosion products subjectively appears to be
better than for flame cleaning. The un-cleaned surfaces of the controls with powdery,
laminating corrosion products and vestiges of previously applied coatings are
inappropriate for direct application of a coating due to a plethora of disadvantages
that include: unevenness, poor intra surface adhesion, residual contaminants,
hydration of oxide layers and ongoing corrosion.
The pictorial standard for flame cleaned surfaces states that flame cleaning must be
followed by power tool wire-brushing to remove the products of flame cleaning, as
hand wire-brushing does not prepare the surface satisfactorily to receive protective
paint coatings (British Standards Institute 2006/2007). For reasons of practicality,
manual wire-brushing after flame cleaning was investigated here, since it has
anecdotally been found to be common practice in heritage ironwork conservation
(see 4.2.8). The presence of powdery corrosion products on surfaces after such
cleaning lends weight to the assertion that this is not satisfactory preparation before
application of paint.
6.1.5 Practical considerations in application of preparation methods: heritage

context
6.1.5.1 Blasting with glass beads, aluminium oxide and crushed walnut shell
These three blasting techniques were similar in their application, equipment needs
and practical considerations. Blasting within a cabinet is a relatively controlled
process with detritus of operation largely contained, as is blasting of large objects in
a workshop. In situ blasting requires a system by which spent media, corrosion
products and paint are collected and cleared from the site. The cost, practical and
financial considerations of in situ blasting often preclude their use in heritage
contexts where resources are often severely limited.
156
Control of parameters such as angle of nozzle to and distance from surfaces is readily
achievable with small, flat samples in the blasting cabinet, making this method ideal
for portable heritage structures. It would be difficult to accomplish in situ for large,
complex structures within heritage, although garnet blasting to Sa2.5 was
successfully used by Eura Conservation to prepare the surface of Brunel’s ship ss
Great Britain (Watkinson and Tanner 2008). Additionally, the morphology of heritage
objects may make it difficult to remove paint layers and corrosion products from
intricate and inaccessible areas of a structure by blasting, although this is equally true
for other preparation methods. The pressures involved have potential to buckle
material of thin section or destroy heavily corroded and weakened heritage iron that
it is normally desirable to retain.
Health and safety concerns exist relating to exposure of the operator to airborne
particulates and to disposal of blasting detritus. Crushed walnut shell has the
potential to cause allergic reaction and the larger, angular particles require use of
heavier weight protective gloves. Considering these factors it appears that blasting is
an ideal preparation method for use on smaller heritage objects, where total removal
of coatings, corrosion and surface finish is acceptable, with execution occurring in
controlled workshop environments. Its use in the field may be suitable or even
essential on large projects such as ss Great Britain but this is at the expense of high
cost and less controlled application, making it appropriate for large, well-funded in
situ projects. Cost of the method may be a movable point when the input of operator
time is considered.
6.1.5.2 Immersion in sodium hydroxide solution and wire-brushing

Immersion of samples in sodium hydroxide is a simple, controlled process in the
laboratory but this technique requires disassembly of structures, removal from site
and access to a sufficiently large dipping tank for immersion. In situ application may
occasionally be possible and techniques employing poultices could be considered but
rinsing to remove sodium hydroxide solution is necessary and must be planned for
from financial, disposal and health and safety viewpoints. Similarly to blasting
techniques, sodium hydroxide seems to be better suited to the controlled
environment of the workshop for heritage contexts.
157
6.1.5.3 Flame cleaning
This is a far less controlled and more operator dependent preparation method when
compared to other techniques examined here. Maintaining surface temperatures,
torch to surface distance and even distribution of heat, along with assessing the end
point of the process, are entirely qualitative decisions with large operating
parameters based on operator experience and interpretation. How reproducible this
preparation technique could be for heritage objects requires further research that
examines how differing temperatures, time of treatment, operator identity, surface
morphology and oxide transformation influence the consistency of the preparation
end point. The importance of form in heritage metalwork and often inherent object
fragility mean flame cleaning is not suitable for iron of section less than 5mm, due to
deformation risks (heritage ironworkers pers. comm.). This is clearly not a technique
to be applied by inexperienced operators. Major advantages for contractors are its
portability for in situ work, low cost and conformation to an international standard
and the assurance that this appears to offer; disadvantages include hazards from
large scale combustion of existing paint layers, especially as lead based paints are
often present within heritage coatings.
6.1.6 Success and practicality criteria

The advantages and limitations of the preparation methods have been discussed but
a useful summary is perhaps to consider how the techniques comply with
practitioner and specifier criteria for practicalities and success (Table 6.1).
158
Criteria Compliant methods
Removal of oxides and Glass beads blasting

contaminants
Aluminium oxide blasting
Retention of oxides Crushed walnut blasting
Sodium hydroxide immersion/wire brushing
Flame cleaning/wire brushing
Keyed surface for coating Aluminium oxide blasting

adhesion
Glass beads blasting (to an extent)
Reduced corrosion rate of Sodium hydroxide immersion/wire brushing

substrate
Crushed walnut blasting
Minimal waste produced Flame cleaning/wire brushing
On site treatment possible Glass beads blasting
Aluminium oxide blasting
Crushed walnut blasting
Flame cleaning
Table 6.1 Compliance of surface preparation methods with practitioner/specifier

criteria.
6.1.7 Oxygen consumption rates of prepared surfaces

All samples consume oxygen and show a decrease in oxygen consumption rate over
time (Figs 5.10-5.15). Oxygen consumption rates can be examined via the changing
gradient of the average trendline for each preparation method (Fig. 5.17). This is
done by using visual analysis of consumption graphs based on dividing the trend line
into an initial period of 0-12 days (0-20 for un-cleaned samples), secondary period of
19-103 days (20-104 days un-cleaned) and final period of 103-338 days (104-257 days
un-cleaned). Un-cleaned sample corrosion rate testing was conducted separately,
hence discrepancies in gradient analysis periods. As might be predicted, all surface
preparation methods show highest average consumption rates during the initial
exposure period when the maximum area of freshly exposed metal is present and
159
available for oxidation. Once oxides form a degree of passivation ensues, reducing
corrosion rate.
6.1.7.1 Total consumption over study period

Clear and significant differences in oxygen consumption rate can be identified
between a number of preparation methods using average oxygen consumption
(mol/year/mm2x10-8) of sample for each preparation method, with the un-cleaned
samples as a baseline for comparison (Table 5.4). No significant difference could be
determined between aluminium oxide and glass beads blasting methods or sodium
hydroxide and crushed walnut blasting methods, which is evident when examining
box plots (Fig. 5.18). Removal of oxide layers to reveal the metal surface, using glass
beads or aluminium oxide blasting, produced higher average oxygen consumption
than the un-cleaned control. This is likely due to the exposure of the metallic surface
and its unevenness offering a relative increase in surface area and a large reaction
platform of exposed metal still populated by corrosion accelerators. In contrast,
there is a closely adhering, coherent oxide layer overlying the metal surface of the
untreated sample but this also retains all corrosion accelerators. Initial high oxygen
consumption of exposed metal surfaces is expected and is followed by slowing as
corrosion products develop (Fig. 5.17). The higher average oxygen consumption of
aluminium oxide blasted samples, relative to those blasted by glass beads, may be
due to a larger surface area produced by the rougher surface finish on the aluminium
oxide samples. Upon termination of the tests both aluminium oxide and glass bead
sample sets had developed new corrosion products concentrated in areas previously
pitted during atmospheric exposure of the railing and along slag planes (Fig. 5.19).
This is likely due to the low pH anodic centres expected within pits and the potential
difference between slag and iron (Scott 1989).
Both crushed walnut blasting and sodium hydroxide immersion with wire brushing
decrease average oxygen consumption relative to the un-cleaned samples average.
They show negligible oxygen consumption beyond 50 days (Figs 5.12, 5.13, 5.18).
Treatment with sodium hydroxide solution has long been reported to inhibit iron
corrosion (Mayne and Menter 1954; Mayne et al. 1950; Turgoose 1985), which may
explain the reduction in corrosion rate recorded. For crushed walnut it may be that
160
the impact of the blast media further compacts the coherent corrosion product layer
visible on the sample surfaces post-cleaning, enhancing protection of underlying
metal substrate and delivering hydrophobic effects from the drying oils in the nut
shells. These oils cross link to produce films (Mills and White 1987), whose
hydrophobicity is evidenced in their traditional use for paint media. Developing even
a thin non continuous drying oil layer offers potential to hinder moisture and oxygen
ingress to anode sites at the metal surface. The dark colour of the surface post-
preparation may be due to oils and disappears post-corrosion due to formation of
new corrosion products (Fig. 5.9).
Despite, or because of, the retention of closely adhering corrosion that must also
retain contaminants beneath it, both these preparation techniques reduce corrosion
rate relative to the un-cleaned control (Fig. 5.18). However, their lower oxygen
consumption rate may simply relate to the arithmetic of there being a smaller
reactive surface area due to retention of coherent corrosion product layers that
hinder ingress of oxygen and moisture. The potential impact of a drying oil layer could
be examined by using aluminium oxide blasting to remove all corrosion products,
then impinging crushed walnut shells onto the surfaces of the samples. Oxygen
consumption rates could then be compared to those of aluminium oxide prepared
samples which were not subsequently subjected to crushed walnut blasting.
Flame cleaned samples show a considerably higher oxygen consumption rate than
the un-cleaned samples (Fig. 5.18). While macroscopically the extent and form of
corrosion product on flame cleaned samples was least changed relative to un-
cleaned surfaces, showing only colour difference, at high magnification a fractured
and micro-cracked surface is evident (Fig. 5.9). SEM images of the cross section of
flame cleaned samples shows fragmentation of the corrosion product layers (Fig.
5.28). While differential expansion of oxides and metal substrate are used to describe
the mechanism by which oxides are removed from the surface, it may also cause
fracturing and cracking, which then offers pathways to oxygen and water ingress and
subsequently corrosion. These imperfections make it difficult to understand the
popularity of flame cleaning and emphasise the need for much more quantified
assessment of flame cleaning as a preparation method. This is further evidenced in
161
the practitioner methodology. Evidence based support for the efficacy and
advantages of flame cleaning are hard to find.
6.1.7.2 Practitioner flame cleaning

Results obtained by the laboratory flame cleaning procedure are reinforced by the
oxygen consumption rates of the practitioner flame cleaned samples. Although the
laboratory method was verified by a practitioner as being representative of the field
procedure, other practitioners understandably questioned the method when results
demonstrated a surprisingly high oxygen consumption/corrosion rate.
There is no significant difference between the corrosion rates of the samples cleaned
by practitioners PA and PB. The range of rates is similar in both cases. Corrosion rates
of the samples prepared by practitioner PC are significantly lower than both PA and
PB. All three practitioners employed slightly different methods with parameters most
likely to impact corrosion rate identified as the nature of the torch (oxypropane or
oxyacetylene), the temperature of the metal during cleaning and the time for which
the flame was applied to the metal (see Table 4.3).
Practitioners PA and PB used oxypropane torches with oxygen boost setting to

increase flame temperature. Practitioner PC used an oxyacetylene torch with no
oxygen boost. Oxyacetylene flame temperature is generally higher than oxypropane.
Flame application time was greatest for PC (50-60 seconds contact time with breaks
to wire brush), slightly less for PA (20-30 seconds) and least for PB (<10 seconds). All
practitioners had similar parts of the flame in contact with the sample surface (Figs
6.3, 6.4, 6.6).
Temperature of the metal during flame cleaning is likely more important than time
of flame application. None of the practitioners could state the metal temperature
during cleaning with confidence. Colour of the metal is given to be a rough measure
of temperature (Uddeholm n.d.) and the samples prepared by PA and PC both
achieved a dull blood red/cherry red state (Table 4.3, Figs 6.2, 6.7) estimated to be
c.550-650oC (Uddeholm n.d.). Samples prepared by practitioner PB remained grey
throughout, likely below 300oC. Practitioner PA estimated a temperature of 500oC
which is likely lower than that actually attained. PC estimated 700oC, probably an
162
overestimation, and PB c.100oC which may well be approximately correct as it was
lower than the temper colour range. There may be corrosion product
transformations occurring at the elevated temperatures. Transformation of goethite
to haematite has been reported at 240 – 250oC (Ruan et al. 2002) and lepidocrocite
to maghemite at c.200oC which is metastable and transforms again to haematite at
c.500OC (Gehring and Hofmeister 1994). Transformation of oxides may explain the
bright red corrosion products seen within a minute or so of flame cleaning a sample
(Fig. 6.8).
Given that the time of application of flame to the surface was different for all
practitioners and temperature of the metal was similar for PA and PC but different
for HI, the parameter which separates PA and PB from PC is the nature of the torch.
Practitioners PA and PB used oxypropane torches (Fig. 6.4). It should also be noted
that this corrosion rate tested an unreal scenario. None of the practitioners would
have flame cleaned wrought iron without immediately applying a coating to the
surface, be that linseed oil, a rust converter or a polymeric coating.
Figure 6.1 Oxypropane torch used by practitioner PA.
Figure 6.2 Dull red glow of sample flame cleaned by practitioner PA.
163
Figure 6.3 Position of the flame relative to the sample during flame cleaning by
practitioner PB.
Figure 6.4 Luminescence of coating and oxide vestiges under flame (practitioner PB).
Figure 6.5 Wire brushing of flame cleaned sample by practitioner PB.
164
Figure 6.6 Position of the oxyacetylene torch relative to the sample during flame
cleaning by practitioner PC.
Figure 6.7 Dull red glow of wrought iron during flame cleaning by practitioner PC.
Figure 6.8 Characteristic bright red oxides forming developing 1-2 minutes after
flame cleaning on all samples (uncleaned sample top left for comparison).
165
6.1.8 Flame cleaning in practice
Practitioner flame cleaning is carried out in an ad hoc manner with a range of
methods employed and different philosophical approaches. Torch choice may be
based on availability and cost. Oxyacetylene is more costly than oxypropane. Use of
the oxygen boost cutting flame with the oxypropane torch was deliberate to increase
flame temperature and burn off paint layers more quickly. Flame application time
must relate to extent of corrosion product and amount and nature of existing
coatings but also to the philosophy of the practitioner. PB was concerned with
conservation ethics, preserving historic evidence within the metallographic structure
and the idea that the metal last attained red heat during forging. This concern led to
the shortest duration of cleaning and lowest metal temperature. Neither PA nor PC
considered negatives of heating the metal in relation to the material or principles of
conservation. This divergence, one practitioner using ethics to control end points and
two others using appearance during the process, apparently ignores any
interpretation or conception of physical or chemical outcome for the metal and
relation to application of coatings. Additionally, and perhaps cynically, it may be that
flame cleaning is carried out to ‘tick a box’ for procedure and evidence base in
relation to coating procedure and any future failure.
A high level of experience and knowledge of the materials and their properties as
related to the task in hand does not prevent a ‘technician approach’ to work. Unless
an agreement is in place for maintenance work by the same practitioner, an
individual may never be in situ to judge long term implications of decisions and
practices in cleaning and coating ironwork. Training directs practitioner thought
processes; the conservation trained individual was more thoughtful about impact of
decisions on retention of historic evidence, for example. It may be through training
that a synergy of practical experience and understanding of chemical and physical
processes can be produced both to improve practice and researcher knowledge.
Practitioners are an invaluable source of information regarding wrought iron, its
properties and responses to treatments.
166
6.1.9 Interpolation to develop context
By linking the morphology of the surfaces produced by the selected preparation
methods to their oxygen consumption rate in high humidity, defined as a proxy for
corrosion, a more informed approach to devising coating systems for heritage iron is
possible. Whilst no tests examining the adhesion of coatings to surfaces were carried
out, it is possible to use surface morphology to predict that aluminium oxide and
glass bead blasting potentially offer the best keying surfaces for coatings. There is
little to choose in terms of keying morphology for the walnut blasted samples and
the flame cleaned samples, with both offering less keying opportunity than either
aluminium oxide or glass bead blasting. Sodium hydroxide appeared to offer the
fewest keying opportunities but this must be balanced against the often powdery
finish on the flame cleaned samples, which is likely to interfere with adhesion.
Similarly, the possible impact of oils delivered by walnut media on adhesion of
coatings must be considered, although it requires further study.
Linked to these predictions of coating performance relative to keying onto surfaces

must be the response of the surfaces themselves to high relative humidity. Should a
coating applied to each surface offer exactly the same transmission of moisture and
oxygen then this study indicates that sodium hydroxide and walnut blasted surfaces
can be expected to corrode more slowly than glass bead blasted surfaces, which will
corrode more slowly than aluminium oxide blasted surfaces with flame cleaned
surfaces offering the worst corrosion rate of the five treatments. It is tempting to
translate this into an estimation that slower corrosion rate of the samples tested here
mean that a coating would remain fit for purpose for a longer time period on the less
reactive substrate surfaces, thus offering better cost benefit. However, there are
multiple variables that relate to coating performance. In this instance, the degree of
disruption to the coating caused by the corrosion forming on the metal surface may
influence its performance and adhesion at an early stage of the corrosion process.
This may cause corrosion rates at the metal surface to diverge from those identified
when testing surfaces without applied coatings. Also, the initial adhesion of the
coating to the surface may have a major influence on how any corrosion impacts on
the coating above it. Only further testing could examine these factors and other
167
variables to add to understanding that could build a clearer picture of the complex
relationships between coatings and surfaces. This study has effectively examined one
fixed condition context, changing the context may well change the order of
comparative performance. It is these multiple scenarios that must be tested which
make development of guidelines and a model so difficult.
6.2 Protective Coatings

The array of factors in decision-making for coating selection was discussed in 3.4.
These are revisited in the light of the coating performance test results to discuss their
impact on decision-making and how the coatings tested here fare in relation to these
concerns. Emphasis is on the thought process behind the decision-making and the
contribution of empirical evidence and qualitative understanding of coating
properties and practicalities (Tables 5.6, 5.7) to guiding decisions. Anti-corrosive
performance of a coating in laboratory testing is not the sole factor in selection.
Ultimately, long term success is likely due to a complex web of factors including
achievability of a defect-free coating in field application, compatibility of the coating
with the substrate and its dynamic responses to environments, its maintainability
and its fit to the historic context in addition to its anti-corrosion credentials.
It is assumed that in many instances the decision maker is not a specialist in coating
ferrous metals and may not be trained in conservation principles and heritage
context. The report of flame cleaning by heritage ironworkers highlights the variation
in understanding of materials and their properties, the influence of treatment
processes on subsequent survival of materials, and the principles of preservation of
evidence and retention of original material. It also underlines the benefits of
experience and anecdotal evidence of successes and failures where much of the
following discussion relies on laboratory testing and manufacturer claims.
This discussion centres on selection of a coating system and is separate from the
discussion of surface preparation methods. In reality, the two are interrelated. The
extent of surface preparation may be predetermined and coating selection must aim
168
to find the most appropriate system for the surface, or vice versa. Table 6.2 offers an
overview of the coating systems tested highlighting features relevant to the
discussion.
169
System A System B Coating C
System Primer (2 coats) Primer (2 coats) Single coating (2

Build (2 coats) Topcoat (2 coats) coats)
Topcoat (2 coats)
Surface Blasting Sa2.5 Blasting Sa2.5 or Wire brushing
Preparation wire brushing and
rubbing with coarse
abrasive paper
Surface prep Yes Yes No
contractor
required?
Coating contractor Likely for best No No

required? results
Cost £12.50/m2 coating £11/m2 coating £7/m2 coating

+ blasting contractor + blasting + wire brush
+ waste disposal contractor + paint brush
+ coating contractor or emery paper
+ waste disposal
+ paint brush
Loss of original Yes Yes Minimal
material?
Availability Online Online/DIY outlet Online/DIY outlet
Application Brush/Roller/Spray Brush Brush/Spray
Appearance Plastic Glossy Matte/satin

High build Medium build Low build – surface
profile evident
Colours Vast range BS 381 6 traditional colours Satin – black/white
and RAL colours Hammered – 8
colours
Smooth – 12
premixed/26 store
mixed colours
Recommendations Forth Rail Bridge Household name Household name
ss Great Britain
Compliance with
British Standards
Railtrack
Ministry of Defence
Highways Agency
Table 6.2 Summary of features of coating systems tested.
170
6.2.1 Cost
This is a critical factor in the planning of many projects yet there is limited difference
between these coatings on face value. System A is more expensive per m 2 being
formed of three coatings each applied in two coats but the cost of the coatings per
litre is slightly less than System B and Coating C. Likewise, there is little to choose
between System B and Coating C based on coating cost per m2 although differences
of a few pounds per square metre become more important when calculated over a
large area.
There are a few main differences in cost between the coating systems when
considered in detail. First, the requirement to blast surfaces to Sa2.5 or near white
metal for optimal performance of System A necessitates use of specialist equipment
and likely employment of a contractor to carry out the work. If the chosen surface
preparation for System B were wire brushing and rubbing down, this could be carried
out by a non-specialist but is likely to be time consuming, adding an employee time
cost. The minimal surface preparation for Coating C likely incurs the lowest cost.
Second, the application of the coating has cost implications. If spraying were chosen
as the application method for System A, further equipment and contractor costs
could be expected. System B and Coating C are readily applicable with care by non-
specialists. Brushes can be cleaned readily with white spirit and reused. The two-pack
nature of System A coatings (the need to make up in a correct ratio and the changing
properties of the coatings as they react in the pot (see 5.2.3)) make application more
difficult and may necessitate using a specialist contractor. If System A is applied by
brush by a non-specialist, painting equipment (e.g. brushes and pots) should be
considered disposable, adding to the cost.
A cost implication arises from tin sizes available for each coating. System B coatings
and Coating C are available in volumes starting from 250ml. The smallest readily
available for System A coatings is 5 litre. If the area to be coated is small, the large
tins of System A may be a disincentive. Given the lifetimes to maintenance discussed
below, it is unlikely that remainders of the unused coatings would retain their quality
until required for maintenance coating.
171
6.2.2 Availability
System B and Coating C are produced by a household name manufacturer and readily
available off the shelf at DIY outlets. This offers the benefits of browsing products on
a shelf, asking an assistant for advice and easy access should more coating be
required unexpectedly and in a hurry. Availability at well-known outlets lends a
feeling of popularity and ‘tried-and-tested’ to these products. Alternatively, the
‘household’ nature of the manufacturer may seem insufficiently technical and
suitable only for domestic scenarios.
System A is available to order online but difficult to source from other outlets.
Browsing requires scrolling through internet pages rather than examining tins on a
shelf. The tins themselves are clinical, without attractive embellishment or logos. This
may be less or more convincing depending on decision-maker mind-set. A technical
coating sold on merit not packaging may imply greater efficacy in performance.
Advisors are available via a telephone service but there is limited option for face to
face discussion. Internet ordering may be ideal big projects requiring delivery of large
volumes but difficulties or delays may arise if underestimation means more coating
is required to finish the work.
6.2.3 Volumes of coating and numbers of coats

This is standalone but strongly linked to cost considerations discussed above. The
greater the mass of coating added, the greater the cost, the longer the time in
application and the more disfiguring to fine detail of a surface or structure (Figs 5.34,
5.37). Alternatively, the perception may be that a greater mass of coating affords
more protection to the substrate. Either way, there is a clear financial distinction
between System A (6 coats applied in total) and Coating C (2 coats applied) and this
may be important in selecting between very interventive (System A) and minimally
interventive (Coating C) approaches. In performance terms, this links to the
protective nature of the coherent and closely adhering oxides retained under Coating
C. The surface tolerant coating exploits protective oxides (and any sound existing
coatings) which increase the thickness of the effective barrier to water and pollutants
driving corrosion without the necessity of multiple coating applications.
172
System B comprises two coats each of two coatings, offering a compromise with the
extra protection of a primer under the topcoat but no build coat. Crucially, any
protective oxides are removed before coating. Despite appearing to offer greater
protection through an increased number of coats, some protective effect may be lost
with the oxide layers relative to Coating C.
Application of each coat is an investment of time and money. If coating is carried out
by staff there are salary costs and commitment of time. If carried out by a contractor
there is a direct cost of contractor time and inconvenience of contractor presence on
site for a longer period. Manufacturers stipulate over-coating intervals for optimum
coating cure between coats. A long over-coating interval can delay or extend a
project. Coating compatibility may dictate timings; there is an over-coating window
after which the polyurethane topcoat of System A will not adhere successfully to the
intermediate epoxy build coat (Leighs Paints 2008b). The topcoat must be applied
within 7 days of application of the build coat. A coating system requiring multiple
numbers of coats may be viewed unfavourably, especially if slow curing or
introducing time constraints for application schedules.
Measurements of coating thickness are given in Table 5.8. Overall system build
ranges are 210-406μm (System A), 23-98μm (System B) and 11-64μm (Coating C).
Ranges are increased by pitted substrate surfaces with greater coating thickness in
pits and reduced thickness on peaks due to run back of coating (Fig. 5.42). BS EN ISO
12944-5:1998 gives expected durability of coatings based on surface preparation
grade, coating type, number of coats and dry film thicknesses (DFT) of coats and
complete systems (British Standards Institute 1998). How DFT relates to corrosion
rate of substrates is not explained. With barrier coatings, conceivably the greater the
DFT the greater the diffusion pathway for moisture and ions for a given coating. No
option correlates exactly to System A but the stated durability of the closest system
in urban and industrial atmospheres (category C3 medium) is High which equates to
more than 15 years. The closest to System B in the same environment has a Low
durability (2-5 years) although none of the DFT values are as low as those recorded
with System B. No paint system in the standard has as few coats as Coating C or DFT
values lower than 120μm so this must also have a Low durability.
173
6.2.4 Primers
This links to number of coats and to perceptions. The necessity of priming a surface
is a disincentive for some, increasing amount of coating needed and number of coats
with cost and time implications as discussed. For others, a primer is attractive
implying more adequate preparation and greater overall efficacy of the system
relative to a single coating such as Coating C. The term ‘primer’ indicates that the
product improves the readiness of the iron to receive the final coating. Implicit is an
extra level of care, attention to detail and a better overall finish. It may also improve
aesthetic properties such as depth of colour if chosen to complement or enhance the
colour of the topcoat. Primers can only be applied with suitable surface preparation;
by definition it must be applied to the substrate surface not over existing coatings.
6.2.5 Required surface preparation

Discussed above in relation to cost and time implications, there is also a crucial
ethical consideration. Blasting to Sa2.5 (near white metal) produces an effective
keying surface for adhesion of coatings and removes corrosion accelerating
contaminants but it also removes the oxides which represent original historic
material and may be protective. Evidence retained within corrosion layers, such as
mill scale, will be lost as the original surface of the metal is removed. Blasting also
diminishes metal section and this thinning of the material may reduce structural
strength. Although System A replaces some of this with bulk of coating, it is not
integral to the core material.
Wire brushing for Coating C does not offer such effective keying for adhesion of the
coating but is appropriate for the surface tolerance of the coating, likely preserves
much original material and fits with a minimal intervention conservation strategy.
Wire brushing followed by coarse abrasive paper rubbing could be a preparation
strategy for System B, removing more original material than wire brushing alone but
much less than blasting to Sa2.5.
6.2.6 Application
System A is applied by brush, roller or spray. Brush application requires no specialist
equipment and is more effective for coating hard to reach areas and convoluted
174
details. Application of System A coatings should be at temperatures of 10 oC and
above and humidity <90% RH (Leighs 2008 a,b,c) which has implications for in situ
application to exterior ironwork during the winter months or wet periods even if the
ironwork itself is under shelter. Spray application by a non-specialist is problematic
with regions of insufficient coverage likely to have numerous holidays (holes in the
coating) (Fig. 6.9) and areas of a structure sheltered by other parts being difficult to
reach with the spray. Even with the close contact and attention to detail possible
with brush application, areas may be missed and consequent low build afford
reduced protection (Fig. 5.45). System B and Coating C can be applied by brush with
System B primer also applicable by roller and Coating C by spray. The same
considerations apply as with System A. The oil-based coatings are more temperature
tolerant at the lower end (being applicable down to 8oC) but have an upper
recommended limit of 25oC.
Figure 6.9 Showing corrosion from pinholes in a sprayed coating.
6.2.7 Testing quality of application

Paint test equipment can be bought, hired or borrowed to check the quality of the
coating application. Dry film thicknesses (DFT) can be tested with a thickness meter
such as the DeFelsko PosiTector 6000 to assess compliance with manufacturer
specified DFT values. Difficulties arise with pitted and uneven historic wrought iron
175
surfaces leading to regional differences in coating thickness between the peaks and
troughs of pitted surfaces.
BS EN ISO 12944-5: 1998 uses nominal DFT to assess thickness (British Standards
Institute 1998). This takes an average of the thickness readings across the sampled
area. For historic material with extensive pitting and uneven surfaces, this is a
misleading measure. Table 5.8 gives the nominal DFT for System B (run-back from
peaks present Figs 5.41, 5.42) as 58μm but the lowest DFT measured was 31μm,
nearly half the nominal value. For Coating C this is more pronounced with a nominal
value of 43μm yet lowest DFT value is 11μm, a quarter of the nominal. Settlement of
coatings into troughs and run back from peaks may cause this variable DFT across a
surface. Nominal values are dangerous if they mask low values in areas where lack of
build means susceptibility to corrosion. Unrealistic expectations of durability and
lifetimes to maintenance may result based on comparison of nominal DFT values to
manufacturer guidance which assumes flat prepared surfaces. Thus the build and
rheology parameters of a coating are likely to have a very significant impact on
coating performances on heritage wrought iron.
6.2.8 Adhesion of coatings

Adhesion is taken to correlate to effectiveness of protection by the coating. It is
tested by the procedure described in 4.3.9.2 according to the standard BS EN
ISO4624:2003 (British Standards Institute 2003). The method is straight forward but
interpretation of the data is less so. There is no clear consensus on what constitutes
a ‘good’ adhesion reading. Generally, the higher the value the better the
performance of the coating is assumed to be but this idea has been challenged
(O’Donoghue et al. 2010; heritage practitioners pers. comm.). As important as the
pull-off value is the mode of failure of the system. Possible failure modes are:
 adhesive between primer and substrate;

 adhesive between coatings;
 cohesive within coatings;
 adhesive between dolly adhesive and topcoat;
 cohesive within dolly adhesive;
176
 adhesive between dolly and dolly adhesive.
Each mode of failure informs on the weakest point within the coating system and the
dolly. The value of the pull-off test is the strength of the weakest adhesive or cohesive
bond within the system.
Results of pull-off tests for coatings examined here are given in Table 5.9. There is
good agreement between the measured values for each coating system. Where
values are lower within a group, this correlates with dollies which had fewer points
of adhesion between the cyanoacrylate on the dolly and the surface of the coating
system. This is a particular problem on the pitted and laminating surfaces common
in heritage. Values for System A were higher than System B which were higher than
Coating C. This is unsurprising as greater adhesion is expected on blasted surfaces
with their improved profile for adhesion (Momber and Wong 2005). Modes of failure
for both systems on blasted substrates mean actual values for primer to substrate
adhesion must be higher than the pull-off value as there was no failure between the
primer and the substrate in either case. System A failure was cohesive within the
primer layer. This could be at the interface between successively applied coats of
primer but this cannot be determined. High solids, zinc-rich primers contain lower
than ideal percentages of binder which reduces cohesive strength (Sørensen et al.
2009, 144). Within System B, failure was adhesive between primer and topcoat. With
Coating C, failure was cohesive within the corrosion products. The value for adhesion
of coating to corrosion products is higher than the pull-off values. Adhesion values
for systems A and B are likely relatively reproducible on different substrate material
prepared in the same manner but the values for Coating C must depend on corrosion
product nature and will vary between substrate materials.
Holiday testing allows detection of pinholes in a coating which are unprotected areas.
The test was not applied to the coatings in this study due to unavailability of a holiday
tester. Pinholes are visible on the coated samples (Figs 5.36, 5.39, 5.43) but with a
high build system such as System A it is unlikely that pinholes in the topcoat would
be detected. The value of multiple applications of coats is that defects are unlikely to
177
overlie one another which reduces the chance of any faults extending from coating
surface to metal substrate.
6.2.9 Aesthetics
6.2.9.1 Colour range
Coating C is available in satin, smooth or hammered finishes. Satin was chosen here
as most likely to conform to the historic aesthetic but this version is currently only
available in black or white. The hammered finish is available in 8 colours and the
smooth in 12 premixed or 26 colours mixed in store. System A topcoat is available in
a vast range of British Standard 381 colours and RAL colours. Colour could be tailored
perfectly to the particular job and may offset the plastic appearance of the system.
System B topcoat is available in six traditional colours allowing some scope for
matching to historic paint schemes but not such a tailored approach as System A. The
importance of this factor will depend upon expected adherence to historic aesthetic
in the project and whether nuances of colour are more valuable than a traditional oil
paint finish.
Colour selection also relates to ability to judge quality of coating application. Coating
shade should be sufficiently different to the preceding coating in a system to allow
visual identification of pinholes and missed areas (Norsok 1999). The number of coats
of the topcoat must then account for the need to obscure the colour of the primer
or build coat. The colour of the primer relative to the substrate material should also
be sufficiently different. Figure 6.10 shows areas of mild steel structural elements
missed during spraying of the primer/build and subsequently corroding. Although the
missed areas are evident now due to the corrosion products, the contrast between
the steel blasted to Sa2.5 and the grey coating may have prevented easy
identification during application.
178
a.
b.
c.
Figure 6.10 a-c Showing corrosion on mild steel structural elements in areas missed
during coating application.
179
Pigment volume concentration and type linked to coating colour may also impact on
performance by affecting barrier properties, for example. Manufacturers are
notoriously tight lipped regarding formulations of coatings so the effects of this are
difficult to predict but could be quantified by testing different colours of the same
coating.
6.2.9.2 Appearance and change over time

The appearance of the coatings as applied to wrought iron can be seen in Figure 5.34.
System A is characterised by a high build which obscures surface morphology and
may obscure delicate features within a structure. The overall appearance is
somewhat ‘plastic’ which is undesirable if conformation to the historic aesthetic is a
priority. For the public, the concept of ‘original appearance’ is important as part of
the visitor experience. A modern finish spells ‘replica’ to many. Where systems such
as this have failed, corrosion has caused extensive staining of lighter colour shades
such as that tested here (Fig. 6.11). Epoxy resins are susceptible to ‘blooming’ with
exposure to UV light and the polyurethane topcoat applied here aims to protect the
epoxy build coat from this. The post 90% RH exposure images (Figs 5.59, 5.60) show
damage to the topcoat caused by slight movements of samples within reaction
vessels during oxygen measurements. Susceptibility to impact damage due to
embrittlement of the polyurethane coating is a well-known drawback (various
heritage ironworkers, pers. comm.). If ironwork coated with a similar system is to be
transported after coating, it is expected to require retouching of the polyurethane
before assembly to repair impact damage to the topcoat (P. Meehan 2015, pers.
comm).
180
Figure 6.11 Failure of system
similar to System A within 12
months of reinstatement in
coastal environment.
Corrosion staining evident on
pale coloured topcoat.
Iron responds dimensionally to thermal changes, expanding with increasing

temperature and contracting with decreasing temperature. Epoxy resins cure by
reaction creating cross linking between the resin and hardener (Carraher 2011, 128).
This forms a rigid three dimensional structure with little dimensional response to
changing temperature. Consequently, epoxy resin coatings on iron and steel are
known to crack, particularly at joints where thermal expansion and contraction
effects are greatest (Raedel and Buecker 2014; heritage ironwork practitioners pers.
comm.). Cracks in coatings allow ingress of water and pollutants and become sites
for corrosion of substrate metal. This may then cause the unsightly staining.
181
The high gloss of System B is its most notable feature although this may dull over
time with exposure to an exterior environment with weathering and abrasion effects.
The morphology of the iron surface can be seen, possibly due to the relatively low
build of the primer coating (Fig. 5.39). Oil based coatings are more mobile than epoxy
resins to accommodate the dimensional changes of metallic substrates more readily
(Horovitz 1986). This minimises likelihood of cracking. Coating C has a low build and
allows the substrate morphology to be seen readily. This may be an advantage as it
is unlikely to obscure detail or may be undesirable in revealing the extent of pitting
on a surface. The satin finish conforms to a historic aesthetic which may be a priority.
6.2.10 Practicalities
Much of the success of protective coatings relates to the quality of its application.
Assuming no carelessness, the more challenging the application procedure the more
likely are problems with the finished coating. System A requires the application of
two coats of three two-pack coatings. Two-pack coatings involve mixing both parts
of the coating in the correct ratio either by mass or volume to deliver optimum
properties of the coating. Too little hardener and the coating will not cure fully, too
much and the excess will remain as a sticky residue on the surface (Fig. 6.12).
Figure 6.12 Showing an example of the effects of preparing two-pack coating in

incorrect ratio of resin and hardener.
Properties of the coatings change during application as reaction curing begins in the
pot. Viscosity may be very low initially resulting in low build and run back from edges
but gradually increases during application until it becomes stringy, bulky and difficult
to attain a smooth finish with. The point at which the coating becomes too viscous
182
and no longer produces a satisfactory finish must be judged. Non-uniform application
across the structure or object may lead to more successful performance in some
areas than in others. The rate of change in properties relates to cure time which in
turn relies on environmental conditions (Leighs 2008a,b). Elevated temperatures
accelerate curing and lower temperatures retard it. Additionally, two-part mixing can
be wasteful of materials since without extensive experience it is difficult to predict
the volumes required.
System B topcoat was notable for its long drying time and persistent tackiness. Were
ironwork to be coated off site, based on this experience it would be at least two
weeks before the object could be handled and transported without damaging the
coating. Were the coating applied on site, particulate pickup and damage from
passers-by, animals and so on would negatively impact on coating appearance and
continuity. The slow drying time is likely responsible for extensive run-back from
peaks (Figs 5.41, 5.42) and consequent low DFT in these areas.
6.2.11 Recommendations and endorsements

System B and Coating C are manufactured by a household name brand. This may be
seen as an endorsement or a negative depending upon the perception of the decision
maker. The product range manufactured by the brand suggests preparation of
different formulations to meet the needs of particular situations (e.g. galvanised and
high temperature surfaces) which may inspire confidence. Anecdotal evidence for
and against use of these products can be found on many DIY and automotive
websites and discussion forums. Although unlikely to offer balanced and insightful
discussion in all cases, they may present useful experience of the difficulty or ease of
use of the products by non-specialists and reports of successes or failures over time.
Perhaps a subjective review of such feedback would offer a useful starting point for
choosing coatings.
The manufacturer of System A is not a household name but is well known within
engineering and coating sectors. Their products are used extensively in aggressive
marine environments and for high profile projects such as the heavily publicised
coating of the Forth Rail Bridge (Sherwin-Williams 2013) which ended the previous
183
continuous painting cycle. The manufacturer advertises its credentials with an
endorsements section on the product datasheets (Leighs 2008 a, b, c) including from
large, private companies and national bodies and compliance with British Standards.
Within heritage, coatings produced by this manufacturer have been used for small
and large projects (Watkinson et al. 2005; Peter Meehan pers. comm.; Eura
Conservation pers. comm.) with a consequent word-of-mouth effect which has likely
increased their use. A non-specialist, for example the manager of a historic property
with no conservation experience, may not have heard of this manufacturer and is
unlikely to come across it without online research.
6.2.12 Lifetime to maintenance and durability

It is clear from manufacturer data sheets and British Standards that there is no
guaranteed lifetime of a coating (British Standards Institute 2007; Leighs 2008 a, b,
c). It is easy to see why. With so many variables (environment, corrodibility of
substrate material, thoroughness of surface preparation, quality of coating
application, adherence to stated over-coating intervals, DFT, holidays and more) it is
impossible to guarantee the lifetime unless the substrate and the application are
perfectly controlled. Figure 6.11 shows failure within 12 months of a coating system
almost identical to System A applied by non-specialists over a substrate which was
likely not desalinated thoroughly before coating and reassembly in its coastal
environment (C4 or C5-M atmospheric corrosivity category). According to BS EN ISO
12944-5:1998 this system in a C4 environment should have had a High durability with
first major maintenance after 15 years or more (British Standards Institute 1998). In
a C5-M environment (aggressive coastal) the expected durability would be 5-15
years. The coating failed long before the predicted time.
The expected durability (time to first major maintenance) of System A is 15 years or

more in an urban/industrial environment. A lifetime of 8 years is advertised by the
manufacturer of Coating C. Lifetime of the primer for System B is not stated although
it is advertised as ‘Long Lasting’ and no lifetime is given for the topcoat. According to
BS EN ISO 12944-5:1998 (British Standards Institute 1998) the expected durability for
such a system in an urban/industrial environment would be less than 5 years.
184
These predictions do not offer a helpful guide for scheduling maintenance. If a system
is expected to require maintenance within five years, should this be scheduled at one
year or four? The answer to this must be greater resolution of timescales through
wider testing of substrates and environments on the part of the manufacturer or
Standards Institute and ongoing vigilance and inspection of coated ironwork at
intervals of a few months. Truly effective management of heritage preservation
cannot occur with the level of uncertainty presented by the current standards.
On the basis of advertised lifetimes and predictions from British Standards, it would
appear that System A offers the greatest lifetime to maintenance, followed by
Coating C and System B. This cannot be inferred from short term, static RH testing
carried out here with no concession to weathering processes or pollutants. Long term
testing with cycling conditions of RH and temperature and introduction of corrosion
promoting species consistent with atmospheric pollutants could offer this
understanding of performance. British Standards and manufacturer predictions are
based on accelerated testing. Proposed here is a programme of real-time testing on
historic material which is time consuming and labour intensive but essential for the
heritage sector to build the necessary evidence base for effective practice. The
prospect of developing such a programme is discussed in 6.2.16 and 6.3.
6.2.13 Corrosion prevention

Oxygen consumption reported in 5.2.4, Figs 5.49-5.58 and Tables 5.10 and 5.11
indicates no corrosion of System A and Coating C iron samples as the consumption
of the coated samples does not exceed that of the control coating samples. Small
amounts of oxygen consumption may relate to discrepancies in surface area of
coating applied which was difficult to standardise due to pitting of the sample
surfaces. There is a significant difference (Table 5.12) between the corrosion rates of
the uncoated substrate material and System A and Coating C samples underlining the
significant corrosion reduction by the coating.
Oxygen consumption by the wrought iron samples was greater than by the coating
controls for System B. It is assumed that this oxygen was consumed by corrosion
reactions. Figure 5.57 shows a distinction between corrosion rate of System B coated
185
samples and the Sa2.5 blasted samples which represent their uncoated substrates.
System B does reduce the corrosion rate of the substrate but not as successfully as
System A.
Based on predicted durability from the British Standards, alkyd coatings are not
expected to perform as well as epoxy resin coatings so results for Systems A and B
are not surprising. BS EN ISO 12944-5:1998 states that the same alkyd system has
greater durability when applied to Sa2.5 blasted substrates than those retaining
oxides (British Standards Institute 1998) so System B and Coating C results might be
considered surprising. As the coatings themselves are not directly comparable, no
conclusion can be drawn on this but the retained oxides of Coating C samples may
be protective. It would be desirable to repeat the test with System B applied over a
wire brushed substrate. This preparation method is advocated for this system but it
is also recommended for use on bare steel surfaces hence testing on a blasted surface
here. Much lower adhesion values with cohesion of oxides the failure point for
Coating C suggests an element of caution is required when advocating coating
application to corrosion products. This must be investigated further before firm
conclusions are drawn. The very nature of oxide coatings varies greatly. The oxides
here were fairly coherent and compact, in other contexts they may be lamellar and
flaking (Figs 2.9, 2.11).
These results only relate to the conditions to which these samples were exposed, a
high static RH with no wetting/drying cycles, atmospheric pollutants, wind, rain or
ultraviolet radiation. Coating performance beyond this limited environment and time
frame cannot be predicted from the results of this study.
6.2.14 Sample material

The challenge of assessing heritage conservation methods experimentally lies in the
nature of the sample material and the method of quantification adopted. The
individual nature of heritage metals whose corrosion layers have developed over
many years, often in unknown and variable environmental conditions, means that
the production of analogues to represent them in experimental study can be
challenging if results are expected to reflect the reality of treating heritage objects.
186
There are specific contexts in which attempts to deliver standardised analogues that
represent heritage objects are useful for reasons of reproducibility. Degrigny (2010)
offers a clear methodology for generating standardised chloride-containing corrosion
layers on copper alloys and iron on which to test the performance of protective
coatings for corroded heritage metals. This worked effectively for producing
comparative data between partners in the PROMET project (Argyropoulos et al.
2007). Nevertheless, there is inevitably a compromise in this approach between the
imperfect representations of naturally generated corrosion layers and those grown
in accelerated corrosion contexts, although this must be balanced against good
reproducibility of analogue samples and the positive impact of this on data quality.
Uniformity provided by analogues offers potential for ranking that can be fed back to
real life scenarios by extrapolation using the context of ‘expected performance’.
The data revealed that it is possible to use the wrought iron railing to produce
comparative test samples for determining the effect of selected surface cleaning
techniques on the corrosion rate of historical iron. The corrosion rate of the samples
was sufficiently consistent to allow comparison between surface treatments that
would not be influenced by the nature of the sample. The graph of five uncleaned
railing samples shows a good agreement between the oxygen consumption rates of
the samples, demonstrated by the clustering of the points over the first 70 days (Fig.
5.15). Rate agreement between uncleaned samples 1 and 2 means their points on
the graph overlie each other throughout the test period. The oxygen consumption
rates do not form straight lines and to increase clarity no trendlines are shown. The
trendline gradient for each sample over the first 75 days is given in Table 6.3. The
slight falling off of the rate seen for each sample may be due to the limiting effect of
diminishing oxygen concentration within the reaction vessels or a consequence of a
build-up of newly formed corrosion products. It may also be that full adsorption of
water into capillaries (see 2.3.1) reduces the rate of oxygen diffusion to the anodes
on the metal surface.
187
Sample No. Gradient
1 0.2246
2 0.2049
3 0.1866
4 0.2378
5 0.1702
Table 6.3 Showing trendline gradients for first 75 days of oxygen consumption testing
at 90% RH for uncleaned railing sample material.
It is surprising that iron retaining vestiges of paint in a random survival pattern should
show good agreement of corrosion rate for samples of similar nominal surface area.
Minor differences in sample mass appear to have no influence on corrosion. Overall,
the homogeneity of the slag distribution (see 5.1.1) likely provides for even corrosion
patterns over the metal surface more akin to general corrosion than extensive
localised pitting. The actual surface area of the samples clearly differs from
calculations related to sample dimensions as surfaces are uneven and pocked and it
may be this, rather than the mass of the samples, that creates differences in
corrosion rate. This point is reinforced when the data for the railing and gasometer
samples are considered (Table 6.4).
The railing and gasometer samples differ in extent of corrosion product and vestiges
of existing corrosion products. As such, the only samples permitting direct
comparison between sample materials are those aluminium oxide blasted to Sa2.5.
Given the fall off in oxygen consumption rate over time evident with both sample
materials, comparison is made by truncating the gasometer data (to 351 days) to
match approximately the number of days oxygen consumption at 90% RH recorded
for the railing material (338).
188
Statistic Railing Gasometer
(mol/mm2/year x 10-8) Mean 15.58 9.29
Median 15.14 9.82
Range 18.90 5.99

O2 consumed
Interquartile Range 11.64 1.61
Standard Deviation 7.01 1.77
Mean 20.77 12.38

(mol/mm2/year x 10-8)
Median 20.18 13.09
Range 25.20 8.00

Fe converted
Interquartile Range 15.20 2.16
Standard Deviation 9.34 2.36
Table 6.4 Descriptive statistics for railing and gasometer sample material blasted to
Sa2.5 and exposed to 90%RH for 338 and 351 days respectively.
The skewness values demonstrate that the data is not normally distributed therefore
median and interquartile range are more appropriate measures of central tendency
than mean and standard deviation. The median values for the railing wrought iron
are higher than for the gasometer material signifying a greater corrosion rate. This is
not unexpected given that the extent of corrosion on the railing prior to sampling
was much greater than that on the gasometer despite both having been exposed to
outdoor atmospheric corrosion over hundreds of years. The railing wrought iron has
a greater slag content (Figs 5.1, 5.29) which may contribute to its higher corrosion
rate. The range and interquartile range of the data sets indicate extent of sample
standardisation. These values for the gasometer are smaller than for the railing
suggesting greater standardisation and reproducibility. Again, this is not unexpected
and is likely due to more extensive pitting of the railing material which leads to
greater differences in surface area available for reaction between the samples.
189
Although it is difficult to draw conclusions from small sample populations, results
indicate that standardisation of historic sample material is possible. It is nearly
impossible to generate perfectly standardised, analogous samples with corrosion
products accurately replicating those formed over many years of atmospheric
corrosion and subject to the same corrosion mechanisms as historic wrought iron. It
would seem from these results that historic sample material can be standardised
sufficiently when care is taken to minimise chances of compositional differences,
variation in surface area and extent of contamination with corrosive pollutants.
While standardisation of analogues is possible, transferring results to heritage

contexts is difficult. The reality is that datasets can only ever be related to the test
sample context. Extrapolating interpretation and contextualisation will always be
necessary.
6.2.15 Standards for corrosion protection by protective coatings

6.2.15.1 Existing standards and application to historic contexts
The question of how suitable British Standards for coating ferrous metals are for
application to heritage sector treatments must be considered. BS 5493:1977 Code of
practice for: Protective coating of iron and steel structures against corrosion (British
Standards Institute 1977) has been partially superseded by BS EN ISO 12944:1998
Paints and varnishes – Corrosion protection of steel structures by protective paint
systems (British Standards Institute 1998). Both are considered here.
Employing these standards when devising coating strategies for historic wrought iron
is problematic. The standards apply to modern steels which themselves conform to
further British Standards of production and composition. Historic wrought iron might
have been produced at any point over a one thousand years plus time span, is
unlikely to have been standardised in its production and will exhibit a range of
compositions (see 1.1). The standards BS 5493 and BS EN ISO 12944 assume that the
paint specification is prepared prior to erection of a structure. Coating application is
expected to be on new milled steel with corrosion products limited to mill scale
formed during hot rolling and flash rusting from less than ideal storage conditions.
This is not the case with historic wrought iron. Corrosion product layers built up over
190
centuries may have caused extensive pitting of surfaces and may include deep seated
corrosion-driving chlorides which are difficult to remove from pits and deep
laminations. Structural members may have suffered loss of section through corrosion
which prohibits blasting to deliver the expected level of surface preparation.
BS EN ISO 12944-3:1998 deals with measures to effectively design out of the

structure any elements which make corrosion more likely or coating of surfaces
difficult to achieve (British Standards Institute 1998). Access for maintenance work is
planned at the design stage and any element inaccessible after construction is to be
sufficiently protected to survive the expected lifetime of the structure without
maintenance. The anticipated functional nature of the structures is underlined by the
instruction that designs should be simple, avoiding complexity and shapes which
promote corrosion. With historic ironwork the design of the structure is
predetermined, likely to be decorative and convoluted and its expected lifetime
would be survival in perpetuity according to conservation professional standards and
ethics.
Descriptions of maintenance procedures are the closest that these standards come
to conservation treatments in the first instance. Crucially, the standards expect
maintenance to have been built into the initial design of an iron or steel structure. It
is assumed that corrosion prevention measures have been maximised in the initial
construction phase, that parts of a structure can be replaced, that structures have
limited lifetimes and that facilitating maintenance was part of the original design. Re-
blasting a structure to bare metal enabling full recoating is expected to be acceptable
and feasible practice. Loss of original material, provided sufficient section remains to
ensure structural stability, is not a concern.
Methods for testing coating properties are difficult to apply to and interpret for
heritage materials. The drawbacks of nominal dry film thickness values are discussed
in 6.2.7. Adhesion testing using pneumatic pull-off dollies is a standard test
procedure for characterising coating properties and performance outlined in BS EN
ISO 4624:2003 (British Standards Institute 2003). Standardisation requirements are
that the coating be applied to uniform thickness on steel panels of uniform surface
191
texture. The test is necessarily destructive as it assesses the failure point and mode
of the coating adhesion to the substrate surface. Problems for the heritage sector in
conducting this test arise from unevenly corroded and pitted surfaces which limit
stability of the dollies during adhesion and prevent even coating application with
pooling in pits and run back from peaks. A wider range of results is likely on
inhomogeneous historic wrought iron surfaces which presents challenges for
interpretation and raises questions over relevance of data.
Without doubt, British Standards can be useful to heritage ironwork practitioners

despite the engineering/industrial focus that limits their relevance for and
application to historic material in heritage contexts. However, it is critical to
recognise their limitations and practitioners wishing to conform and offer an
evidence base for their protocols and decision-making may not do this. Standardised
and evidence based approaches are crucial to effective conservation practice.
Practitioner flame cleaning highlighted variation between individual methods which
are based on anecdotal evidence in the main and relate strongly to extent of training
in conservation ethics. There is a wealth of experience of surface preparation and
coating of historic wrought iron dispersed throughout the heritage ironwork
practitioner community but these are individuals who do not publish widely and the
benefit of this experience is not felt across conservation. Similarly, unless a
blacksmith undertakes training in conservation he/she is unlikely to consider all
facets of conservation ethics and operate with attention to concepts of original
surface and appreciation of the line between conservation and restoration. The
treatment of historic wrought iron is a cross-over discipline in this respect. Ethical
principles apply but structures must often function safely which necessitates more
extensive repair and replacement than might otherwise be ethical. The lack of linkage
between anecdotal and quantitative evidence leads to uninformed decision-making
and practice, dissatisfaction from those contracting work and ultimately, the concept
that heritage is badly managed.
6.2.16 Heritage coating standards

The difficulties inherent in applying the existing British Standards directly to historic
material have been discussed. The ideal would perhaps be to generate evidence
192
based heritage standards to guide practice but this is not without challenges. The
Drafting Committee list for the 1977 BS 5493 (British Standards Committee 1977) has
approximately 50 names of individuals from public and private institutions who
contributed to its content. The extent of this list and the sum total of knowledge and
experience allows the document to cover a wide range of materials and processes in
sufficient depth. It could be argued that given the variation in historic wrought iron
and the vast array of structures and objects produced over time and still in existence,
as much or more input would be needed from experts in the heritage sector to draw
up fit-for-purpose guidelines. Producing an empirical evidence base for the standards
would require an immense amount of research from the few heritage scientists
working in this area. Testing a coating is not a simple process as the many
permutations of a system (surface preparation level, number of coatings, number of
coats of each coating and so on) must all be tested individually to produce guidance
approaching the level of detail found in BS 5493:1977 or BS EN ISO 12944:1998
(British Standards Institute 1998). Development of a standardised method for testing
would be essential for comparative data. Added to this is the requirement to update
standards regularly with results of tests on the new coatings produced at a
bewildering rate by manufacturers and all claiming to be the new wonder coating. If
empirical-evidence based standards were required, the input of time and money
would be unsupportable in such a small and poorly funded sector.
If clear limitations are accepted, there is perhaps scope for a set of standards offering
a synergy of empirical evidence generated in laboratory contexts and anecdotal
experience of successes, failures and practical and ethical considerations. This data
could be built over time by as many individuals as are willing to contribute and in an
ideal world this would lead to a culture of sharing experiences to improve practice
for all. In this era of social media and instant communication this could be achieved
readily with some coordination. Evidence may be in the form of treatment records
and site photos showing modes of failure of coating systems or evidencing successes.
Many practitioners make periodic return visits to site and managers of properties
with treated ironwork could be encouraged to contribute. The commercial sensitivity
of the data could be mitigated by careful moderation of the contributions.
193
A major benefit of this semi-anecdotal approach to data gathering would be the
ability to survey coatings over many years. This move away from the accelerated
testing advocated in industrial and engineering contexts may offer better resolution
of data regarding times and modes of failure of coatings which could inform coating
selection and maintenance scheduling. Clear criteria would underpin assessment of
application and short and long term performance. BS EN ISO 12944:1998 currently
offers categories of ‘low’ 2-5 years, ‘medium’ 5-15 years and ‘high’ more than 15
years for protective coating performance in a given environment (British Standards
Institute 1998). Evidence from practitioners and managers could narrow these
categories or, alternatively, suggest that coatings do not fit neatly into these
categories and evidence the need for further investigation of failures. It would
require the input of environmental data.
Although it may not be possible to provide empirical evidence for the performance
of every coating type from each manufacturer, it would be possible to draw
information from a range of researchers investigating mechanisms of wrought iron
corrosion. For example, there is a trade-off between optimal performance of coatings
on an Sa2.5 blasted surface and the possible reduction of corrosion rate through
retention of the coherent oxide layer present on historic iron objects which have
been subject to centuries of atmospheric corrosion. If the mechanisms of protection
were more clearly understood, it might be possible for practitioners to characterise
and retain protective oxides and limit surface preparation to wire brushing and
perhaps rubbing with emery paper to improve keying and adhesion of coatings. This
would satisfy the need to reduce corrosion rates whilst conforming to conservation
ethics. This study has begun the cost benefit analysis of surface tolerant coatings
relative to greater surface preparation but more work is required to assess long term
protection.
194
6.3 Standardising test methods for heritage: scope, limitations
and further work
6.3.1 Scope of method for heritage sector studies
This study aimed to develop fit-for-purpose test methods for treatment of historic
wrought iron. The results support the methods employed here to standardise surface
preparation and application of protective coatings, the use of appropriate control
samples and the generation of quantitative corrosion rate data via oxygen
consumption measurements. The ability to standardise historic wrought iron
samples was demonstrated, encouraging its use in other studies. The method can be
applied readily to wrought iron of all periods but also to cast iron and mild steel
sample material and to investigating the range of variables required to underpin
predictive guidelines for practitioners.
Coated and uncoated samples show good agreement of results for samples treated
in the same manner (Figs 5.10-5.15, 5.49-5.51). This implies a good level of
standardisation and reproducibility even with samples of historic wrought iron. The
protocol for running sufficient numbers of appropriate control samples to determine
their oxygen consumption should allow corrosion rates and treatment effects to be
compared across results from workers using different sample materials. A database
of corrosion rates of untreated samples of wrought iron from different sources would
be illuminating. Variability in corrosion rates of wrought iron of different periods and
compositions would indicate the extent to which a one-size-fits-all approach to
treatment of historic wrought iron is appropriate. This data will be generated as part
of continuing work at Cardiff as more sample material is sourced. Other workers
using the same method would be encouraged to pool their data on uncoated
wrought iron for comparison alongside details of metallurgy.
Flame cleaning by practitioners demonstrated that supporting laboratory methods

and results with samples prepared by practitioners using their preferred techniques
allows verification of legitimacy of laboratory testing. This good practice could be
extended in future to other techniques examined in this study. It would also permit
close investigation of the diversity of practice and methods within the sector and
impact of differing protocols on success or failure of treatments. This would feed
195
directly into guidance for practitioners which would be entirely focused on and
relevant to their practices. Extending this yet further to coating application,
practitioner methods of spray coating could be compared to brush application of the
same coating. The importance of this being carried out in the field or shop is that
application methods must be fit-for-purpose for the situation in which they will be
employed. The practicalities of using that method in the field are integral to its
success. Laboratory testing does not allow challenging of the techniques to this
extent.
Most standardised laboratory testing of protective coating systems for ferrous

metals aims to determine whether the coatings have the properties in given
environments which suggest that they will offer a satisfactory level of protection.
Testing is normally accelerated by exposure of samples to more aggressive
environments than would be expected. The properties of the coatings after these
accelerated artificial-ageing exposures are examined and predicted performance
extrapolated from the results.
The real-time method employed for examining coating performance here is not
related to determination of properties after ageing but to measurement of corrosion
rate of coated wrought iron substrates during exposure to a given environment, in
this case a static high humidity. The difference between this and the approach
outlined in the standard is that one level of extrapolation is removed by the
measurement of corrosion rate from oxygen consumption. Running sufficient
numbers of control samples gives a degree of confidence in measurement of the
oxygen consumption of the coatings themselves and allows the determination of the
oxygen consumption by the substrate material which is expected to be due to
corrosion. This is therefore a direct measure of the protective performance of the
coating in the test environment. Likewise, oxygen consumption of prepared,
uncoated wrought iron allows corrosion rate in the test environment to be
quantitatively measured.
196
6.3.2 Limitations and further work
Although successful, the methods developed here are not without their limitations.
Laboratory preparation of samples permits a degree of control and attention to detail
through small scale actions in a managed setting that is not achievable in the field or
shop. Blasting samples using laboratory airbrasive units with glove boxes, tailored
lighting and magnification to increase visibility of surfaces allows assessment of
tightly controlled end points of cleaning which is not possible for practitioners. The
blast media are more akin to fine powders than the grits commonly used for blasting
structures in practice. Resultant surface profiles will therefore be coarser than those
attained in this study which may have some impact on corrosion rate tested here but
would likely be more noticeable as differences in adhesion and numbers of rogue
peaks with low build coatings.
Correlation of the test environment to real-life conditions in exterior exposure of

historic wrought iron is limited in this case, being static 90% RH at 20 oC. This extreme
of humidity is not unlikely in Britain (Met Office 2014) but static RH and temperature
represents unreal conditions. There is no concession to the influence of pollutants,
ultraviolet radiation, wetting and drying cycles, extremes of temperature and other
factors in this study. The method of corrosion rate measurement relies on the
samples being sealed in the reaction vessels for continuous periods of fairly long
duration in the case of samples which are consuming oxygen slowly. It would degrade
the quality of the data were the reaction vessels opened frequently to change the
conditions inside, for example to cycle RH. The variable that can be altered readily
within the climatic chambers is temperature. This could be programmed to reflect
diurnal and seasonal changes. The effect of cycling temperature on the RH within the
reaction vessels with silica gel must be examined. The extent to which conditioned
silica gel will buffer the change in RH with temperature is not known. Addressing
these limitations will be a focus for further work in this area building on the methods
used in this study. Potential foci for continued research are outlined in Table 6.5.
197
Variable Focus
Relative Humidity Wet/dry cycles

Corrosion rate measurement via
Ultra Violet Radiation Coating breakdown
Temperature Diurnal and seasonal

oxygen consumption
Coating Damage Scribed samples
Corrosion Accelerants Chlorides/seawater
Joints and Crevices Coating performance
Table 6.5 Further work test variables for corrosion rate investigations.
A short term goal and integral to further work is dissemination of results and
engagement with practitioners. This experimental investigation was developed to
meet a sector need which was evident in communication with heritage ironwork
practitioners and specifiers of treatment programmes for wrought iron and in the
commissioning of this project by Historic Scotland. It is essential that results are fed
back to the sector to justify investment of time and money in the project by
individuals and organisations. It is only by reporting results that benefit can be felt by
the sector and researchers can hope to influence practices and encourage evidence
based management. Publication of results in heritage sector and corrosion journals
will disseminate findings to a portion of the heritage community but it is unlikely that
most practitioners have access to corrosion journals. The next step is the production
of clear, freely accessible sector based guidance that links results to methods and
balances corrosion rates and expected lifetimes to practicalities and costs.
198
7 Conclusions
7.1 Cost benefit

This study had as its aim to perform a cost benefit analysis of surface preparation
methods and protective coatings for wrought iron. In reality, this is very complex
equation. It is possible to enter empirical data of corrosion rates not as distinct
values, given differences in substrate composition, corrosivity of environments and
a range of other factors, but as comparatives. Compared to the uncleaned wrought
iron, crushed walnut blasting and sodium hydroxide immersion reduce corrosion
rates. Blasting to Sa 2.5 with aluminium oxide and glass beads increases corrosion
rate. Flame cleaning greatly increases corrosion rate (approximately by a factor of 4).
This may seem to imply that crushed walnut blasting or sodium hydroxide immersion
should be the surface preparation methods of choice yet, practically, immersion in
chemical solutions relies on being able to disassemble a structure and neither
produce surfaces with keying for adhesion of coatings.
This introduces another complication. Surface preparation is a precursor to coating.

Whilst corrosion rates of uncoated substrates might indicate relative risk to the
underlying metal if coatings are damaged, surface preparation must also aim to
maximise performance of the protective coatings. Methods retaining oxides (for
example, the crushed walnut and sodium hydroxide techniques which performed
well in uncoated testing) do not comply with recommendations for surface
preparation prior to coating with two-pack epoxy resin coatings or indeed most non-
surface-tolerant coatings. Coating performance may be compromised with these
techniques. They do, however, comply with conservation ethics. A major factor on
the ‘cost’ side of the equation is loss of original material associated with surface
preparation to Sa2.5. Aluminium oxide and glass beads blasting applied to Sa2.5 level
may remove corrosion products and contaminants which potentially undermine
coating performance but they also remove original material and associated evidence.
Cost benefit analysis of coatings is subject to similar challenges. What can be said
with confidence is that there is no benefit to applying System B over a substrate
199
surface blasted to Sa2.5. Reduction of corrosion rate of the uncoated Sa2.5 substrate
is much less than with the epoxy resin System A and loss of original material is
extensive as compared with Coating C. System A and Coating C offer the benefit of a
reduction in corrosion rate of Sa2.5 blasted and wire brushed substrates respectively.
With System A, this is countered by the cost of loss of original material, practical
difficulties of coating application, financial implications of necessary contractors and
maintenance regimes requiring full re-blasting and re-coating. Coating C has the
benefits of reducing corrosion rates, conforming to ethics of retention of original
material, simple maintenance regimes and low financial cost. The drawbacks are the
unknown long term implications of retention of oxides on performance of the
coating. Low adhesion values of oxides to substrate are a concern.
Despite difficulties in performing true cost benefit analysis of the surface

preparations and coatings tested here, there are clear indications that this will be
possible with a larger dataset gathered over longer timescales. It also requires input
of specifics for each project such as budget, availability of expertise, required
lifetimes to maintenance, importance of historic aesthetic and so forth. A very
welcome outcome of this study is the suggestion that there is potentially a successful
surface preparation and alkyd coating combination for low budget, minimal
intervention projects (Coating C) which conforms to ethical and aesthetic
requirements and performs as well (in relatively short term testing at static high RH)
as the highly interventive, higher cost option of blasting and applying an epoxy resin
system (System A). This horses-for-courses approach appears to offer options to suit
a range of treatment contexts and specifier priorities.
7.2 Method
This study has developed a protocol which allows an oxygen consumption method
previously used on uncoated archaeological objects to be employed as proxy
corrosion rate measurement for coated iron samples. Consumption of oxygen can be
accounted for by this method through measurement of control samples. The same
method can be applied to coated and uncoated samples and by other workers and
200
standardised by use of controls. A method for processing the data of the oxygen
consumption measurements is also proposed which can be applied to work out the
number of moles of iron converted for the oxygen consumed assuming the
predominant reaction is the formation of FeOOH. This can again be utilised (with
caveats) by other workers to build a database of complementary results which would
generate a rich resource for theoretical study and practical application of treatments
for historic wrought iron heritage.
201
8 Reference List
American Institute for Conservation of Historic and Artistic Works. 1994. Code of ethics and
guidelines for practice. Available at: http://www.conservation-us.org/about-us/core-
documents/code-of-ethics-and-guidelines-for-practice/code-of-ethics-and-
guidelines-for-practice#.VdoWs_lViko (accessed 08/2015)
American Society for Testing Materials (ASTM). 2008. Standard Practice for Use of Pictorial
Surface Preparation Standards and Guides for Painting Steel Surfaces. ASTM
Ardizzone, S., Biagiotti, R. and Formaro, L. 1983. Interactions of Cl- ions with Fe3O4. Journal
of the Electroanalytical Chemistry and Analytical Electrochemistry 147 (1-2), pp 301
– 305
Argyropoulos, V., Boyatzis, S., Giannoulaki, M. and Polikreti, K. 2013. The role of standards
in conservation methods for metals in cultural heritage. In P. Dillmann, D. Watkinson,
E. Angelini and A. Adriens (Eds) Corrosion and conservation of cultural heritage
metallic artefacts European Federation of Corrosion Publications 65, pp 478 – 517.
Cambridge: Woodhead Publishing Ltd
Argyropoulos, C., Hein, A., and Abdel Harith, M. (Eds). 2007. Strategies for Saving our Cultural
Heritage. Proceedings of the International Conference on Conservation Strategies for
Saving Indoor Metallic Collections, Cairo. Athens: Technological Educational Institute
of Athens
Ashurst, J. and Ashurst, N. 1988. Practical Building Conservation Vol. 4: Metals. English
Heritage Practical Handbook. Aldershot: Gower Technical Press
Aston, J. and Story, E. B. 1939. Wrought Iron: Its Manufacture, Characteristics and
Applications. Pittsburgh: A. M. Byers Company
Atkinson, R. J., Posner, A. M. and Quirk J. P. 1977. Crystal Nucleation and Growth in
Hydrolysing Iron (III) Chloride Solutions Clay and Clay Minerals 25, pp 49 – 56
202
Bacon, C. R., Smith, J. J. and Rugg, F. G. 1948. Electrolytic Resistance in Evaluating Protective
Merit of Coatings on Metals. Industrial and Engineering Chemistry 40, pp 161 – 167
Barker, D. 2010. Restoration of Antique Wrought Ironwork. Cathedral

Communications:http://www.buildingconservation.com/articles/antiquewrought/a
ntiquewrought.htm (accessed 06/2012)
Bates, W. 1984. Historical Structural Steel Handbook: Properties of U.K. and European Cast
Iron, Wrought Iron and Steel Sections in Design, Load and Stress Data since the Mid
19th Century. London: The British Constructional Steelwork Association Ltd
Bertholon, R. 2001a. The location of the original surface: a review of the conservation
literature. In I. D. MacLeod, J. M. Theile and C. Degrigny (Eds.) Metal 01: Proceedings
of the International Conference on Metals Conservation, Santiago, Chile. Perth:
Western Australian Museum
Bertholon, R. 2001b. To get rid of the crust or not: emergence of the idea of original surface
in the conservation of metal archaeological objects during the first half of the XXe.
In: Oddy, A., Smith, S. (eds) Past Practice-Future Prospect. Pp 5-11. London: British
Museum Press
Bertholon, R. 2007. Archaeological metal artefacts and conservation issues: long-term

corrosion studies. In P. Dillmann, G. Béranger, P. Piccardo and H. Matthiesen (Eds)
Corrosion of Metallic Heritage Artefacts: Investigation, conservation and prediction
for long-term behaviour. Pp 31 – 40. Cambridge: Woodhead
Bierwagen, G. P. 1996. Reflections on corrosion control by organic coatings. Progress in

Organic Coatings 28, pp 43 – 48
Bierwagen, G. P. and Huovinen, A. M. 2010. Paint Formulation. In: B. Cottis, M. Graham, R.

Lindsay, S. Lyon, T. Richardson, D. Scantlebury and H. Stott (Eds). Shreir’s Corrosion.
4 (2643 – 2665) Amsterdam: Elsevier
Blackney, K. 2010. Painting Historic Ironwork. Cathedral Communications Ltd.

http://www.buildingconservation.com/articles/paintingiron/paintingiron.htm
203
(accessed 11/2011)
Blackney, K. and Martin, B. 1998. Development of long-term testing of methods to clean and
coat architectural wrought ironwork located in a marine environment: the
maintenance of railings at the Garrison Church, Portsmouth. In J. M. Teutonico
English Heritage Research Transactions Volume 1: Metals. Pp 103 – 116
Blakelock, E., Martinón-Torres, M., Veldhuijzen, H. A., and Young, T. 2009. Slag inclusions in
iron objects and the quest for provenance: an experiment and a case study. Journal
of Archaeological Science 36, pp 1745 – 1757
Boesenberg, J. S. 2006. Wrought iron from the USS Monitor: mineralogy, petrology and
metallography. Archaeometry 48 (4), pp 613 – 631
Bouchar, M., Dillmann, P. and Neff, D. 2014. Atmospheric corrosion of iron-based

reinforcement of gothic cathedrals – overview of possible corrosion systems and
influence of the structure and composition of the corrosion layers on the corrosion
diagnosis. In Hyslop, E., Gonzalez, V., Troalen, L. and Wilson, L., (Eds.) Metal 2013
Edinburgh, Scotland. Interim Meeting of the International Council of Museums
Committee for Conservation Metal Working Group, 16th -20th September 2013. Pp
215 – 221
British Standards Institute. 1910. BS 51:1910 Wrought iron for use in railway rolling stock.
Best Yorkshire and grades A, B and C. London: British Standards Institute
British Standards Institute. 1939. BS 51:1939 Wrought iron for general engineering purposes
(grades A, B, C and D). London: British Standards Institute
British Standards Institute. 1977. BS 5493 Code of practice for protective coating of iron and
steel structures against corrosion. London: British Standards Institute
British Standards Institute. 1998. BS EN ISO 12944 Paints and Varnishes – Corrosion
protection of steel structure by protective paint systems. London: British Standards
Institute
204
British Standards Institute. 2000. BS EN ISO 8504 Preparation of steel substrates before
application of paints and related products. Surface preparation methods. London:
British Standards Institute
British Standards Institute. 2003. BS EN ISO 4624 Paints and varnishes – Pull-off test for
adhesion. London: British Standards Institute
British Standards Institute. 2005/2006. BS EN ISO 8503 Preparation of steel substrates before
application of paints and related products. Test for the assessment of surface
cleanliness. London: British Standards Institute
British Standards Institute. 2006/2007. BS EN ISO 8501 Preparation of steel substrates before
application of paints and related products. Visual assessment of cleanliness. London:
British Standards Institute
British Standards Institute. 2007. BS EN ISO 12944 Paints and Varnishes. Corrosion protection
of steel structures by protective paint systems. London: British Standards Institute
British Standards Institute. 2012. BS EN ISO 8503 Preparation of steel substrates before
application of paints and related products. Surface roughness characteristics of blast-
cleaned steel substrates. London: British Standards Institute
Cai, J. and Lyon, S. B. 2005. A mechanistic study of initial atmospheric corrosion kinetics using
electrical resistance sensors. Corrosion Science 47 (12), pp 2956 – 2973
Canadian Association for the Conservation of Cultural Property. 2000. Code of ethics and
Guidance for Practice 3rd edition. Ontario: CAC/CAPC
Carraher, C. E. 2011. Carraher’s Polymer Chemistry Eighth Edition. Florida: Taylor and Francis
Catt, R. 1995. Railings. Structural Survey 13 (3), pp 16 – 21
Chandler, K. A. 1966. The influence of salts in rusts on the corrosion of the underlying steel.
British Corrosion Journal 1, pp 264 – 266
205
Charlton, M. F., Blakelock, E., Martinón-Torres, M. and Young, T. 2012. Investigating the
production provenance of iron artefacts with multivariate methods. Journal of
Archaeological Science 39, pp 2280 – 2293
Cheltenham Borough Council. The Conservation and Renewal of Historic Ironwork.

Cheltenham: Cheltenham Borough Council Built Environment Division (accessed
23/3/2013):
http://www.cheltenham.gov.uk/info/200023/conservation/752/advice_leaflets_an
d_useful_links
Childs, R. E. 1985. Protective treatments in industrial archaeology. In S. Keene (Ed.) Corrosion

Inhibitors in Conservation UKIC Occasional Papers 4, pp 26 – 30
Childs, C. W., Goodman, B. A., Paterson, E. and Woodhams, F. W. D. 1980. The nature of iron
in Akaganéite (β-FeOOH). Australian Journal of Chemistry 33 (1), pp 15 – 26
Chilton, J. P. and Evans, U. R. 1955. The corrosion resistance of wrought iron. Journal of iron
and Steel Institute, pp 113 – 122
Cornell, R. M. and Giovanoli R. 1990. Transformation of akaganeite into goethite and

hematite in alkaline media. Clays and Clay minerals 36 (5), pp 469 – 476
Cornell, R. M. and Schwertmann, U. 2003. The Iron Oxides: Structure, properties, reactions,
occurrences and uses (2nd Ed). Weinheim: Wiley-VCH
Corus Construction and Industrial. 2004. A Corrosion Protection Guide For Steelwork Exposed
to Atmospheric Environments. Scunthorpe: Corus Group
Cox, A. and Lyon, S. B. 1994. An electrochemical study of the atmospheric corrosion of mild
steel—III. The effect of sulfur dioxide. Corrosion Science 36 (7), pp 1193 – 1199
Cudennec, Y. and Lecerf A. 2005. Topotactic transformations of goethite and lepidocrocite

into hematite and maghemite. Solid State Sciences 7 (5) pp 520 – 529
Davey, A. 2007. INFORM The Maintenance of Iron Gates and Railings. Edinburgh: Historic
Scotland Technical Conservation, Research and Education Group
206
Davey, A. 2009. Iron: The repair of wrought and cast ironwork. Dublin: Stationery Office
DeFelsko Corporation. 2011. PosiTest Pull-Off Adhesion Tester: Instruction Manual v.4.0.
http://www.defelsko.com/manuals/download/at/PosiTestAT-v4.0.pdf (accessed
01/2015)
DeFelsko Corporation. 2014. Positector 6000 Coating Thickness Gages: Full Guide v.7.4.
Available at http://www.defelsko.com/p6000/p6000_manuals.htm [Accessed
January 2015]
Degrigny, C. 2010. Use of artificial metal coupons to test new protection systems on cultural
heritage objects: manufacturing and validation. Corrosion Engineering, Science and
Technology 45 (5), pp 367 – 374
Degrigny, C., Vella, D., Golfomitsou, S. and Crawford, J. 2007. Characterisation of Corrosion
Product Layers on Atmospherically Corroded Historic Ferrous Objects: Application to
the Armour of the Palace Armoury, Valletta, Malta. In: V. Argyropoulos, A. Hein, M.
Abdel Harith (Eds) Strategies for Saving our Cultural Heritage. Proceedings of the
International Conference on Conservation Strategies for Saving Indoor Metallic
Collections, Cairo. pp 31 -39. Athens: Technological Educational Institute of Athens
de Maybaum, B. R. and Ayllon, E. S, 1982. Atmospheric marine corrosion of structural steels.

In W. H. Ailor (Ed) Atmospheric Corrosion. Corrosion Monograph Series. J Wiley, pp
423 – 430
de Rooij A. 1989. Bimetallic Compatible Couples. ESA Journal, 13, p199-209
Dickie, R. A. 1994. Paint adhesion, corrosion protection, and interfacial chemistry. Progress
in Organic Coatings 25 (1), pp 3 – 22
Dickie, R. A. and Smith, A. G. 1980. How paint arrests rust. Chemtech 10, pp 31 – 35
Dillmann, P., Mazaudier, F. and Hœrlé, S. 2004. Advances in understanding atmospheric

corrosion of iron. I. Rust characterisation of ancient ferrous artefacts exposed to
indoor atmospheric corrosion. Corrosion Science 46 (6), pp 1401 – 1429
207
Dunnwald, J. and Otto, A. 1989. An investigation of phase transitions in rust layers using
Raman spectroscopy. Corrosion Science 29 (9), pp 1167 – 1176
Ecorys. 2013. Review of Estimates of the Economic Impact of Scotland’s Historic Environment:
A report to Historic Scotland. Ecorys UK http://www.historic-
scotland.gov.uk/reconomiimpact-ecorys2013.pdf (accessed 08/2015)
Eiselstein, L. E. and Caliguri, R. D. 1988. Atmospheric Corrosion of cables on the Williamsburg

Bridge. In: S. W. Dean and T. S. Lee (Eds) Degradation of Metals in the Atmosphere,
ASRM STP 965. Philadelphia: American Society for Testing and Materials, pp 78 – 95
Emmerson N. and Watkinson D. 2014. Preparing historic wrought iron for protective
coatings: quantitative assessment to produce evidence based protocols. In E. Hyslop,
V. Gonzalez, L. Troalen and L. Wilson (Eds.) Metal 2013 Edinburgh, Scotland. Interim
Meeting of the International Council of Museums Committee for Conservation Metal
Working Group, 16th -20th September 2013. Pp 119-128
Erich, S. J. F., Huinink, H. P., Adan, O. G. C, Laven, J. and Esteves, A. C. 2008. The influence of
the pigment volume concentration on the curing of alkyd coatings: A 1D MRI depth
profiling study. Progress in Organic Coatings 63, pp 399 – 404
Evans, U. R. 1981. An Introduction to Metallic Corrosion 3rd Edition. London: Arnold
Evans, U. R. and Taylor, C. A. 1972. Mechanisms of Atmospheric Rusting. Corrosion Science

12, pp 227 – 246
Evans, U. R. and Taylor, C. A. J. 1974. Critical humidity for rusting in the presence of sea salt.
British Corrosion Journal 1, pp 26 – 28
Feliu, S., Barajas, R., Bastidas, J. M. and Morcillo, M. 1989a. Mechanisms of cathodic
protection of zinc-rich paints by electrochemical impedance spectroscopy. I: Galvanic
stage. Journal of Coating Technology 61, pp 63 – 69
Feliu, S., Barajas, R., Bastidas, J. M. and Morcillo, M. 1989b. Mechanisms of cathodic
protection of zinc-rich paints by electrochemical impedance spectroscopy. II: Barrier
stage. Journal of Coating Technology 61, pp 71 – 76
208
Funke, W. 1985. The role of adhesion in corrosion protection by organic coatings. J. Oil Col.
Chem. Assoc. 68, 229 – 232
Galvele, J. R. 1976. Transport process and the mechanism of pitting in metals. Journal of the
Electrochemical Society 123 (4), pp 464 – 474
Gandy, D. 2007. The Carbon Steel Handbook. Palo Alto, USA: Electric Power Research
Institute
Garverick, L. 1994. Corrosion in the Petrochemical Industry (2nd Edition). ASM International
– The materials information society
Gayle, M., Look, D. W. and Waite, J. G. 1992. Metal in America’s Historic Buildings: Uses and
preservation treatments. Washington: US Dept of Interior
Gehring, A. U. and Hofmeister, A. M. 1994. The transformation of lepidocrocite during

heating: a magnetic and spectroscopic study. Clay and Clay Minerals 42 (4), pp 409 –
415
Godfraind, S., Pender, R. and Martin, B. (Eds). 2012. English Heritage Practical Building
Conservation: Metals. Farnham: Ashgate Publishing Limited
Graedel, T. E. and Frankenthal, R. P. 1990. Corrosion mechanisms for iron and low alloy steels
exposed to the atmosphere. Journal of the Electrochemical Society 137, pp 2385 –
2394
Greaves, R. H. and Wrighton, H. 1967. Practical Microscopical Metallography. London:

Chapman and Hall and Science Paperbacks
Greenfield, D. and Scantlebury, D. 2000. The Protective Action of Organic Coatings on Steel:
A Review. The Journal of Corrosion Science and Engineering 3 (5)
http://www.jcse.org/volume3/paper5/v3p5.php Accessed 8/10/2014
Gustafsson, M. E. R. and Franzén, L. G. 2000. Inland transport of marine aerosols in southern

Sweden. Atmospheric Environment 34 (2), pp 313 – 325
209
Hare, C. H. 1979. Anti-corrosive barriers and inhibitive primers. Federation Series on Coatings
Technology, Unit 27. Philadelphia: Federation of Societies for Coating Technology
Hayman, R. 2011. Wrought Iron. Oxford: Shire
Haynie, F. H. 1982. Evaluation of the effects of microclimate differences on corrosion.

Atmospheric Corrosion of Metals. ASTM STP 767 American Society for Testing
Materials, pp 286 – 308
Haynie, F. H. and Upham, J. B. 1974. Correlation between corrosion behaviour of steel and
atmospheric pollution data. Corrosion in Natural Environments, ASTM STP 558
American Society for Testing Materials, pp 33 – 51
Higgins, G. L., Hullcoop, R. S., Turgoose, S. and Bullough, W. 2010. Surface Pretreatment. In:
B. Cottis, M. Graham, R. Lindsay, S. Lyon, T. Richardson, D. Scantlebury and H. Stott
(Eds). Shreir’s Corrosion. 4 (2483 – 2493) Amsterdam: Elsevier
Higgins, R. A. 1957. Engineering Metallurgy Part 1: Applied Physical Metallurgy. London:

English Universities Press
Historic England. 2015. Sourcing lead based paint. (Accessed 09/2015)

https://historicengland.org.uk/advice/technical-advice/buildings/building-
materials-for-historic-buildings/sourcing-lead-paint/
Hœrlé, S., Mazaudier, F., Dillmann, Ph. and Santarini, G. 2004. Advances in understanding
atmospheric corrosion of iron. II. Mechanistic modelling of wet-dry cycles. Corrosion
Science 46, pp 1431 – 1465
Hooper, J. J., Foecke, T., Graham, L., and Weihs, T. P. 2003. Metallurgical analysis of wrought
iron from the RMS Titanic. Marine Technology 40 (2), pp 73 – 81
Horovitz, I. 1986. Paintings on copper supports: techniques, deterioration and conservation.

The Conservator 10, pp 44 – 48
Hudson, J. C. and Stanners, J. F. 1953. The effect of climate and atmospheric pollution on
corrosion. Journal of Applied Chemistry 3, pp 86 – 96
210
International Council of Museums. 1984. The Conservator Restorer: A Definition of the
Profession. Available at (accessed 05/2015) http://www.icom-cc.org/47/history-of-
icom-cc/definition-of-profession-1984/#.Vgkuvfmqqko
Ishikawa, T. and Inouye, K. 1975. Role of Chlorine in β-FeOOH on Its Thermal Change and
Reactivity to Sulphur Dioxide. Bulletin of the Chemical Society of Japan, 48, pp 1580
– 1584
Jones, D. 1992. Principles and Prevention of Corrosion. New York: Macmillan
Kalendová, A., Veselý, D., Kohl, M. and Stejskal, J. 2015. Anticorrosion efficiency of zinc-filled
epoxy coatings containing polymers and pigments. Progress in Organic Coatings 78,
pp 1 – 20
Kaneko, K. and Inouye, K. 1979. Adsorption of water on FeOOH as studied by electrical

conductivity measurements. Bulletin of the Chemical Society of Japan 52 (2), pp 315
– 320
Keane, J. D., Wettach, W. and Bosh, C. 1969. Minimum Paint Thickness for Economical
Protection of Hot-Rolled Steel Against Corrosion. Journal of Paint Technology 41, pp
372 – 382
Keller, P. 1970. Eigenshaften von (Cl,F,OH)<2Fe2(O.OH)16 und Akaganéite. Neues Jahrbuch für
Mineralogie Abhandlungen 113, pp 29 – 49
Keystone 1999. Conservation plan for the Great Western Steamship Company Dockyard and
the ss Great Britain. ss Great Britain Project
http://www.ssgreatbritain.org/sites/default/files/kcfinder/files/brunel-
institute/ssGBConservationPlanVol1.pdf
Kittelberger, W. W. and Elm, A. C. 1952. Diffusion of chloride through various paint systems.
Industrial and Engineering Chemistry Research 44 (2), pp 326 – 329
Knotkova-Cermakova, D, Vlckova, J. and Honzak, J. 1982. Atmospheric corrosion of

weathering steels. Atmospheric Corrosion of Metals. ASTM STP 767 American Society
for Testing Materials, pp 7 – 44
211
Krauskopf, K. B. 1982. Introduction to Geochemistry 2nd Edition. Singapore: McGraw-Hill
Kreislova, K., Knotkova, D. and Geiplova, H. 2013. Atmospheric corrosion of historical

industrial structures. In P. Dillmann, D. Watkinson, E. Angelini and A. Adriens (Eds)
Corrosion and conservation of cultural heritage metallic artefacts European
Federation of Corrosion Publications 65, pp 311 – 343. Cambridge: Woodhead
Publishing Ltd
Kucera, V. and Mattsson, E. 1987. Atmopsheric Corrosion. In: F. Mansfeld (Ed.) Corrosion
mechanisms. Chemical Industries 28. New York: Marcel Dekker, Inc., pp 211 – 284
Kumar, A. V. R. and Balasubramaniam, R. 1998. Corrosion product analysis of corrosion

resistant ancient Indian iron. Corrosion Science 40 (7), pp 1169 - 1178
Leighs Paints. 2008a. Metagard L574 Product Technical Data. Available (Accessed June
2012): https://www.promain.co.uk/specsheets/TL574.pdf
Leighs Paints. 2008b. Epigrip C400V3 Product Technical Data. Available (Accessed June
2012): https://www.promain.co.uk/specsheets/TC400V3.pdf
Leighs Paints. 2008c. Resistex C137V2 Product Technical Data. Available (Accessed June
2012): https://www.promain.co.uk/specsheets/TC137V2.pdf
Leygraf, C. and Graedel, T. 2000. Atmospheric Corrosion. Chichester: Wiley
L’Héritier, M. and Dillmann, P. 2005. Fer ou acier? Charactérisation des alliages ferreux
utilises dans la construction des églises gothiques au Moyen Âge et à la période
modern L’exemple de Troyes et de Rouen. L’acier en Europe avant Bessemer.
Toulouse: CNRS and Université de Toulouse Le Mirail
L’Héritier, M., Dillmann, P., Aumard, S. and Fluzin, P. 2013. Iron? Which iron? Methodologies
for metallographic and slag inclusion studies applied to ferrous reinforcements from
Auxerre Cathedral. The World of Iron. London: Archetype
Light, J. D. 2000. Observations concerning the hand forging of wrought iron. Materials
Characterization 45, pp 327 – 340
212
Lins, A. 1992. The Cleaning of Weathered Bronze Monuments: A Review and Comparison of
Current Corrosion Removal Techniques. In: T. Drayman-Weisser (Ed) Dialogue/89-
The Conservation of Bronze Sculpture in the Outdoor Environment: A Dialogue Among
Conservators, Curators, Environmental Scientists, and Corrosion Engineers. Pp 209 –
230 Houston: NACE
Liu, W. M. 1998. Efficiency of barrier-effect and inhibitive anti-corrosion pigments in primers.

Materials and Corrosion 49, pp 576 – 584
Lutenegger, A. J. 2008. Historic Iron Bridges: Adaptive Use Bridge Project, University of
Massachusetts-Amhurst. In: Adeli, H. (Ed). Historic Bridges: Evaluation, Preservation
and Management. Pp 205 – 217. Boca Raton, Florida: Taylor & Francis
Lyon, S. B. 2010. Corrosion of Carbon and Low-Alloy Steels. In Cottis et al. (Eds). Shreir’s
Corrosion. Vol.3 pp 1693-1736. Amsterdam: Elsevier
McDonnell, G. 1989. Iron and its alloys in the fifth to eleventh centuries AD in England. World
Archaeology 20 (3), pp 373 – 382
Maeda, Y., Matsuo, Y., Sugihari, S., Momshima, N. and Takashima, Y. 1992. Mossbauer
studies of first stage corrosion products on iron powder and corrosion products on
highly corroded nails. Corrosion Science 33 (10), 1557 – 1567
Maréchal, L., Perrin, S., Dillman, P., and Santarini, G. 2007. Study of the atmospheric
corrosion of iron by ageing historical artefacts and contemporary low-alloy steel in a
climatic chamber: comparison with mechanistic modelling. In: Dillmann, P.,
Béranger, G., Picardo, P., and Matthiesen, H. (Eds). Conservation of metallic heritage
artefacts: Investigation, conservation, prediction. European Federation of Corrosion
Publications Number 48. Pp 131 – 151. Abington: Woodhead Publishing
Matthiesen, H. 2007. A Novel Method to Determine Oxidation Rates of Heritage Materials

in vitro and in Situ. Studies in Conservation 52, pp 271 – 180
Matthiesen, H. 2013. Oxygen monitoring in the corrosion and preservation of metallic

heritage artefacts. In P. Dillmann, D. Watkinson, E. Angelini and A. Adriens (Eds)
213
Corrosion and conservation of cultural heritage metallic artefacts European
Federation of Corrosion Publications 65, pp 368 – 391. Cambridge: Woodhead
Publishing Ltd
Mayne, J. E. O. 1954. How paints prevent corrosion. Anti-Corrosion Methods and Materials
1 (8), pp 286 - 290
Mayne, J, E, O. 1959. The problem of painting rusty steel. Journal of Applied Chemistry 9, pp
673 – 680
Mayne, J. E. O. and Menter, J. W. 1954. The Mechanism of Inhibition of the Corrosion of Iron
by Sodium Hydroxide Solution Part II. Journal of the Chemical Society (Resumed). Pp
99 – 103
Mayne, J. E. O., Menter, J. W., and Pryor, M. J. 1950. The Mechanism of Inhibition of the
Corrosion of Iron by Sodium Hydroxide Solution. Journal of the Chemical Society
(Resumed), Pp 3229 – 3236
Meehan, P. 2010. Historic Railings: Conservation and Repair. Cathedral Communications Ltd.
Available at (accessed 7/11/2011):
http://www.buildingconservation.com/articles/historicrailings/historicrailings.htm
Met Office. 2014. Regional Relative Humidity Averages 1981 – 2010. Available at:
http://www.metoffice.gov.uk/climatechange/science/monitoring/data/Regional_av
erages/RelativeHumidity_regional_1981-2010_LTA.txt (Accessed 23/09/2014)
Merriman, A. D. 1958. A Dictionary of Metallurgy. London: Macdonald and Evans
Mills, J. and White, R. 1987. The Organic Chemistry of Museum Objects. Oxford: Butterworth-
Heinemann
Mitchell, D. S. 2005. INFORM Boundary Ironwork: A guide to re-instatement. Edinburgh:

Historic Scotland Technical Conservation, Research and Education Group
Momber, A. W. and Wong, Y. C. 2005. Overblasting Effects on Surface Properties of Low-

Carbon Steel. Journal of Coatings Technology and Research. 2 (6), 453 – 461
214
Monnier, J., Bellot-Gurlet, L., Legrand, L., Dillmann, P., Réguer, D., Neff, D. and Guillot, I.
2007. The long term indoor atmospheric corrosion of iron: rust layer
characterisation. In C. Degrigny, R. van Langh, I. Joosten and B. Ankersmit (Eds) Metal
07: Interim Meeting of the ICOM-CC Metal Working Group Vol. 2: Innovative
Investigation of Metal Artefacts, 17-21 September 2007, Amsterdam. Pp 47 – 54
Amsterdam: ICOM-CC Metal Working Group.
Monnier, J., Réguer, S., Foy, E., Testemale, D., Mirambet, F., Saheb, M., Dillmann, P. and
Guillot, I. 2014. XAS and XRD in situ characterisation of reduction and reoxidation
processes of iron corrosion products involved in atmospheric corrosion. Corrosion
Science 78, pp 293 – 303
Morcillo, M., D. de la Fuente, I. Diaz and H. Cano. 2011. Atmospheric Corrosion of Mild Steel.
Revista De Metalurgia 47 (5), pp 426 – 444
Neff, D. P., Dillmann, P., Bellot-Gurlet, L. and Beranger, G. 2005. Corrosion of iron
archaeological artefacts in soil: characterisation of the corrosion system. Corrosion
Science 47 (2), pp 515 – 535
Neff, D., Bellot-Gurlet, L., Dillmann, P., Reguer, S., and Legrand, L. 2006. Raman imaging of
ancient rust scales on archaeological iron artefacts for long-term atmospheric
corrosion mechanisms study. Journal of Raman Spectroscopy 37, pp 1228 - 1237
Neff, D., Vega, E., Dillmann, P. and Descostes, M. 2007. Contribution of iron archaeological
artefacts to the estimation of average corrosion rates and the long-term corrosion
mechanisms of low-carbon steel buried in soil. In: P. Dillmann, G. Béranger, P.
Piccardo and H. Matthiesen (Eds) Corrosion of metallic cultural heritage artefacts:
Investigation, conservation and prediction for long-term behaviour. European
Federation of Corrosion Publications 48. Cambridge: Woodhead
Norsok. 1999. Norsok Standard M-501 Surface Preparation and Protective Coating. Oslo:
Norwegian Technology Standards Institute
North, N. 1982. Corrosion products on marine iron. Studies in Conservation 27, pp 72 – 83
215
O’Donoghue, M., Datta, V. J., Winter, M. and Reed, C. 2010. The Dark Side of Misreading the
Relevance of Coating Testing. Journal of Protective Coatings and Linings May 2010,
pp 30 – 45
O’Sullivan, M. and Swailes, T. 2009. A study of historical test data for better informed
assessment of wrought iron structures. International Journal of Architectural
Heritage 3, pp 260 – 275
Pourbaix, M. 1977. Electrochemical corrosion and reduction. Corrosion and Metal Artifacts:
A Dialogue Between Conservators and Archaeologists and Corrosion Scientists.
National Bureau of Standards Special Publication 479. Washington DC: US
Government Printing Office, pp 1 – 16
Raedel, M. and Buecker, M. 2014. Testing the durability of modified corrosion protection
systems for iron monuments conservation. Oral presentation: EuroCorr 2014, Pisa,
9-12 September
Raman, A. 1988. Atmospheric corrosion problems with weathering steels in Louisianna

Bridges. In: S. W. Dean and T. S. Lee (Eds) Degradation of Metals in the Atmosphere,
ASRM STP 965. Philadelphia: American Society for Testing and Materials, pp 16 – 29
Rammelt, U. and Reinhart, G. 1994. Characterization of active pigments in damage of organic

coatings on steel by means of electrochemical impedance spectroscopy. Progress in
Rawdon, H. S. 2013. Some Unusual Features in the Microstructure of Wrought Iron:

Technological Papers of the Bureau of Standards 97. BiblioBazaar
Refait, P. and Genin, J. M. R. 1997. The mechanisms of oxidation of ferrous hydroxychloride

β-Fe2(OH)3Cl in aqueous solution: the formation of akaganeite versus goethite.
Corrosion Science 39 (3), pp 539 – 553
Refait, Ph., Abdelmoula, M. and Génin, J. M.-R. 1998. Mechanisms of formation and
structure of green rust one in aqueous corrosion of iron in the presence of chloride
ions. Corrosion Science 40 (9), pp 1547 – 1560
216
Refait, Ph., Ouahman, R., Forrières, C. and Génin, J.-M. R. 1992. The role of Cl- ions in the
oxidation of iron artifacts from chlorinated archaeological environmental. Hyperfine
Interactions 70, pp 997 – 1000
Réguer, S., Dillmann, P., Mirambet, F., Susini, J. and Lagarde, P. 2006. Investigation of Cl
corrosion products of iron archaeological artefacts using micro-focused synchrotron
X-ray absorption spectroscopy. Applied Physics A: Materials Science & Processing 83
(2), pp 189 – 193
Réguer, S., Dillmann, P., and Mirambet, F. 2007. Buried iron archaeological artefacts:
Corrosion mechanisms related to the presence of Cl-containing phases. Corrosion
Science 49, pp 2726 – 2744
Réguer, S., Mirambet, F., Dooryhee, E., Hodeau, J.-L., Dillmann, P. and Lagarde, P. 2009.
Structural evidence for the desalination of akaganeite in the preservation of iron
archaeological objects, using synchrotron X-ray powder diffraction and absorption
spectroscopy. Corrosion Science 51, pp 2795 – 2802
Rehren, T., Belgya, T., Jambon, A., Káli, G., Kasztovszky, Z., Kis, Z., Kovács, I., Maróti, B.,
Martinón-Torres, M., Minici, G., Pigott, V.C., Radivojevic, M., Rosta, L., Szentmiklósi,
L. and Szokefalvi-Nagy, Z. 2013. 5000 years old Egyptian iron beads made from
hammered meteoric iron. Journal of Archaeological Science 40, 4785 – 4792
Rémazeilles, C. and Refait, P. 2007. On the formation of β-FeOOH (akaganéite) in chloride-

containing environments. Corrosion Science 49, pp 844 – 857
Rimmer, M. 2010. Investigation the treatment of chloride-infested archaeological iron

objects. PhD Thesis, Cardiff University.
Rimmer, M. and Watkinson, D. 2010. Residues of alkaline sulphite treatment and their
effects on the corrosion of archaeological iron objects. In: P. Mardikian, C. Chemello,
C. Watters and P. Hull (Eds) METAL 2010, Proceedings of the Interim Meeting of the
ICOM-CC Metal Working Group, Charleston, South Carolina 11-15 October 2010,
Clemson University, pp 16 – 22
217
Rimmer, M., Thickett, D., Watkinson, D. and Ganiaris, H. 2013. Guidelines for the Storage and
Display of Archaeological Metalwork. Swindon: English Heritage
Ross, T. K. 1977. Metal Corrosion. Engineering Design Guides 21. Oxford: Oxford University
Press
Rouw, A. C. 1998. Model epoxy powder coatings and their adhesion to steel. Progress in
Ruan, H. D., Frost, R. L., Kloprogge, J. T. and Duong, L. 2002. Infrared spectroscopy of goethite
dehydroxylation: III. FT-IR microscopy of in situ study of the thermal transformation
of goethite to hematite. Spectrochimica Acta Part A 58, pp 967 – 981
Santarini, G. 2007. Corrosion behaviour of low-alloy steel: from ancient past to far future. In:
Dillmann, P., Béranger, G., Picardo, P., and Matthiesen, H. (Eds). Conservation of
metallic heritage artefacts: Investigation, conservation, prediction. European
Federation of Corrosion Publications Number 48. Pp 18 – 30. Abington: Woodhead
Publishing
Schütz, A. and Gehrke, J. 2008. Modern corrosion protection on historic buildings. Materials
and Corrosion 59 (3), pp 228 – 231
Scott, B. 1989. The retrieval of technological information from corrosion products on early
wrought iron artefacts. In Janaway, R. and Scott, B. (Eds) Evidence Preserved in
Corrosion Products: New fields in artefact studies. Occasional Paper 8. Pp 8-14.
London: UKIC
Scott, D. A. 1991. Metallography and Microstructure of Ancient Metals: Marina del Rey,
California: The J. Paul Getty Trust
Scott, D. A. and Eggert, G. 2009. Iron and Steel in Art. London: Archetype
Scully, J. C. 1990. The Fundamentals of Corrosion 3rd Edition. Oxford: Pergamon Press
218
Selwyn, L. S. and Argyropoulos, V. 2005. Removal of chloride and iron ions from
archaeological wrought iron with sodium hydroxide and ethylenediamine solutions.
Studies in Conservation 50 (2), pp 81 – 100
Sereda, P. J. 1974. Weather factors affecting corrosion of metals. Corrosion in Natural

Environments, ASTM STP 558 American Society for Testing Materials, pp 7 – 22
Sherwin-Williams. 2013. Case Study: Forth Rail Bridge United Kingdom.

http://protectiveemea.sherwinwilliams.com/Documents/LiteratureLibrary/Case%2
0Studies/en/Forth%20Rail%20Bridge%20Case%20Study.pdf (accessed 09/2015)
Schikorr, G. 1963. Werkstoffe und Korrosion 14, pp 69
Siatou, A., Argyropoulos, V., Charalambous, D., Polikreti, K. and Kaminari, A. 2007. Testing
New Coating Systems for the Protection of Metal Collections Exposed in Uncontrolled
Museum Environments. Strategies for saving our cultural heritage. Proceedings of
the International Conference on Conservation Strategies for Saving Indoor Metallic
Collections, Cairo pp 115-120
Sørensen, P. A., Kiil, S., Dam-Johansen, K. and Weinell, C. E. 2009. Anticorrosive coatings: a
review. Journal of Coatings Technology and Research 6 (2), pp 135 – 176
Stahl, K., Nielson, N., Jiang, J. Z., Lebech, B., Hanson, J.B., Norby, P., and van Langschott, J.
2003. On the Akaganeite transformations and possible role in post-excavation
corrosion of iron artifacts. Corrosion Science 45 (11), pp 2563 – 2575
Standardiseringskommissionen i Sverige, Steel Structures Painting Council, American Society

for Testing and Materials. 1980. Svensk Standard SIS 05 59 00-1967: Pictorial Surface
Preparation Standards for Painting Steel Surfaces. Stockholm: Swedish Standards
Institution
Starley, D. 1999. Determining the Technological Origins of Iron and Steel. Journal of
Archaeological Science 26, pp 1127 – 1133
Steinsmo, U. and Skari, J. I. 1994. Factors Influencing the Rate of Cathodic Disbonding of
Coatings. Corrosion Science 50, pp 934 – 939
219
Stratmann, M. and Hoffmann, K. 1989. In situ Mößbauer spectroscopic study of reactions
within rust layers. Corrosion Science 29, pp 1329 - 1352
Stratmann, M. and Müller, J. 1994. The mechanism of the oxygen reduction on rust-covered
metal substrates. Corrosion Science 36 (2), pp 327 – 359
Stratmann, M. and Streckel, H. 1990. On the atmospheric corrosion of metals which are
covered with thin electrolyte layers—II. Experimental results. 30 (6-7), pp 697 – 714
Sutherland, J. 2009. The Birth of Prestressing? Iron Bridges for Railways 1830 to 1850.
International Journal for the History of Engineering and Technology 79 (1), pp 113 –
130
Suzuki, T., Yamabe, M. and Kitamua, Y. 1973. Composition of anolyte within pit anode of
austenitic stainless steels in chloride solution. Corrosion 29 (3), pp 18 – 22
Sydberger, T. and Vannerberg, N. G. 1972. The influence of the relative humidity and
corrosion products on the adsorption. Corrosion Science 12 (10) pp 775 – 784
Syed, S. 2006. Atmospheric corrosion of materials. Emirates Journal for Engineering Research
11 (1), pp 1 – 24
Taylor, J. and Suff, S. 2010. Wrought Ironwork. Cathedral Communications Ltd.

http://www.buildingconservation.com/articles/wroughtiron/wrought2000.htm
(accessed 11/2012)
Thickett, D. 2011. Post Excavation Changes and Preventive Conservation of Archaeological

Iron. PhD Thesis: University College London
Thickett, D. and Odlyha, M. 2014. The formation and transformation of akaganeite. In

Hyslop, E., Gonzalez, V., Troalen, L. and Wilson, L., (Eds.) Metal 2013 Edinburgh,
Scotland. Interim Meeting of the International Council of Museums Committee for
Conservation Metal Working Group, 16th -20th September 2013. Pp 103 – 109
Thomas, N. L. 1991. The barrier properties of paint coatings. Progress in Organic Coatings
19, pp 101 – 121
220
Thomson, G. 1986. The Museum Environment 2nd Edition. London: Butterworth-Heinemann
Topp, C. 2010. Wrought Iron and Conservation. Cathedral Communications Ltd.

www.buildingconservation.com/articles/wrought/wrought.htm (accessed 11/2011)
Trethewey, K.R. and Chamberlain, J. 1995. Corrosion for Science and Engineering 2nd Edition.
Harlow: Longman
Turgoose, S. 1982a. The corrosion of archaeological iron during burial and treatment. Studies
in Conservation 30, pp 13 – 18
Turgoose, S. 1982b. The nature of surviving iron objects. In R. W. Clarke and S. M. Blackshaw
(Eds) Conservation of Iron Maritime Monographs and Reports 53, pp 1 – 7
Turgoose, S. 1985. Corrosion inhibitors in conservation. In S. Keene (Ed.) Corrosion Inhibitors

in Conservation UKIC Occasional Papers 4, pp 13 – 17
Tylecote, R. F. 1976. A History of Metallurgy. London: The Metals Society
Uddeholm. n.d. Temperature Guide. (Accessed 09/2015)

http://www.uddeholm.com/files/Temperature_guide.pdf
Vernon, W. H. J. 1931. A laboratory study of the atmospheric corrosion of metals. Part I. -

The corrosion of copper in certain synthetic atmospheres, with particular reference
to the influence of sulphur dioxide in air of different relative humidities. Transactions
of the Faraday Society 27, pp 255 – 277
Vernon, W. H. J. 1935. A laboratory study of the atmospheric corrosion of metals. Part II.—
Iron: the primary oxide film. Part III.—The secondary product or rust (influence of
sulphur dioxide, carbon dioxide, and suspended particles on the rusting of iron).
Transactions of the Faraday Society 31, pp1668 – 1700
Wallis, G. and Bussell, M. 2008. Cast iron, wrought iron and steel. In M. Forsyth (Ed.)
Materials and skills for historic building conservation. Pp123 – 159. Oxford: Blackwell
221
Watkinson, D. 2010. Measuring the effectiveness of chloride extraction methods for
corrosion control of heritage iron: problems and challenges. Corrosion Engineering
Science and Technology, 45 (5), pp 400 – 406
Watkinson, D. and Al Zahrani, A. 2008. Towards quantified assessment of aqueous chloride

extraction methods for archaeological iron; de-oxygenated treatment environment.
The Conservator 31, pp 75 – 86
Watkinson, D. and Lewis, M. 2004. ss Great Britain iron hull: modelling corrosion to define
storage relative humidity. In J. Ashton and D. Hallam (Eds) Metal 04: Proceedings of
the International Conference on Metals Conservation Canberra, Australia 4-8 October
2004. Pp 88 - 103. Canberra: National Museum of Australia
Watkinson, D. and Lewis, M. R. T. 2005a. Desiccated storage of chloride-contaminated

archaeological iron objects. Studies in Conservation 50, pp 241 – 252
Watkinson, D. and Lewis, M. R. T. 2005b. The Role of beta-FeOOH in the Corrosion of

Archaeological Iron. In P. B. Vandiver, J. L. Mass, and A. Murray Material Issues in Art
and Archaeology VII. Pp 103 – 114. Warrendale, PA: Materials Research Society of
America Symposium
Watkinson, D. and Rimmer, M. 2014. Quantifying effectiveness of chloride desalination

treatments for archaeological iron using oxygen measurement. In Hyslop, E.,
Gonzalez, V., Troalen, L. and Wilson, L., (Eds.) Metal 2013 Edinburgh, Scotland.
Interim Meeting of the International Council of Museums Committee for
Conservation Metal Working Group, 16th -20th September 2013. Pp 119-128.
Watkinson, D., Tanner, M., Turner, R. and Lewis, M. 2005. ss Great Britain: teamwork as a
platform for innovative conservation. The Conservator 29, pp 73 – 86
Watkinson, D. and Tanner, M. 2008. ss Great Britain: Conservation and Access – Synergy and
Cost. In D. Saunders, J. H. Townsend, and S. Woodcock Conservation and Access; IIC
London Congress 15-19 September 2008. Pp 109 – 114. London: International
Institute for Conservation of Historic and Artistic Works
222
Watkinson D., Rimmer M. and Kergourlay F. 2013. Alkaline desalination techniques for
archaeological iron. In: P. Dillmann, D. Watkinson, E. Angelini, A. Adriaens (Eds)
Corrosion and Conservation, of Cultural Heritage metallic Artifacts. European
Federation of Corrosion (EFC) Series No. 65, pp. 407 – 433. Cambridge: Woodhead
Weisser, H. B. and Milligan, W. O. 1935. X-ray studies on the hydrous oxides. V. Beta ferric
oxide monohydrate. Journal of the American Chemical Society 57, pp 238 – 241
Wilson, L., Mitchell, D. S. and Davey, A. 2008. Ferrous Metals Cleaning Report Part 1: Cast
Iron. Edinburgh: Historic Scotland Technical Conservation Group
Wilson, L., Davey, A., Mitchell, D. S., and Davidson, A. 2010. Traditional Architectural
Ironwork: Scientific Approaches to Determining Best Conservation Practice and the
Bute Canopy Study. Metal 2010: Proceedings of the Interim Meeting of the ICOM-CC
Metal Working Group, Charleston, South Carolina 11-15 October 2010. pp 248-255.
Charleston: Clemson University
Zucci, F., Morigi, G. and Bertolasi, V. 1977. Beta iron oxide hydroxide formation in localized
active corrosion of iron artifacts. Corrosion and Metal Artifacts: A Dialogue Between
Conservators and Archaeologists. National Bureau of Standards Special Publication
479.
223
9 Appendices
9.1 Appendix 1: Recommended surface preparation methods

Table of recommendations of surface preparation techniques in heritage sector
publications after: Wilson et al. 2008, 5; Childs 1985, 27; Ashurst and Ashurst 1988;
Godfraind et al. 2012
Nb: Test on cast iron but results likely useful when considering wrought iron surface
preparation.
Nb: Aim to retain closely adherent oxide layers influences perceived

advantages/disadvantages.
* Relates to sandblasting, the use of which is now prohibited in Britain and

elsewhere
224
Cleaning Description Advantages Disadvantages Recommendations
method
Manual & Chipping, scraping and Easy Slow Recommended

power tool brushing to remove Inexpensive Labour intensive
cleaning surface coverings Wire brush less damaging to oxide Not completely effective at paint (wire brushing
layer than other manual and power removal recommended in
tool methods Rotary abrasive disc and needle conjunction with
Cheap gun (high pressure) led to other methods such
Easy fragmentation of oxide layer as flame cleaning or
Safe Only removes loose rust; blast cleaning with
Effective on light rust burnishes underlying rust (wire ‘soft’ blast media)
Simple techniques brushing) Recommended if
Useful as first stage in cleaning Manual techniques ineffective applied with care
process Scoring of surface and loss of
Relatively good control and visibility detail possible
with manual techniques Difficult to reach some areas of
structure
Restrictions on use of vibrating
tools (e.g. needlegun)
Laser Laser beam (532nm or Expensive Not recommended
cleaning 1064nm) blast paint from ----- Ineffective
object surface. Impractical
Technique designed for
stone cleaning.
Flame Flame (oxyacetylene or Fast and effective Must be carried out by a specialist Recommended
cleaning oxypropane) passed over Oxide layer well preserved
225
metal surface to remove Fairly inexpensive Health & safety concerns – (use in conjunction
corrosion products and Removes moisture vaporisation of toxic paints with manual wire
loose mill scale. Burns off part scale Buckles thin metal brushing preferred)
Remaining loose Effective removal of paint, mill scale Not very effective
corrosion products can be and corrosion products Slow on large structures (under controlled
removed by wire brushing Can apply in situ Warping of thin sections conditions)
Considered particularly suitable for Fire hazard Thought to be more
wrought iron as only loose mill scale appropriate for
and rust removed wrought than cast
iron
Aqueous Includes cold-water Aqueous methods are effective at

washing pressure washing, warm removal of soluble salts
methods water washing, high High pressure steam removes oils,
pressure steam waxes and grease
Chemical Corrosion product and No chemical ingress into metal Dichloromethane may have Not recommended
cleaning coating removal usually microstructure noticed damaged surface profile
requiring immersion of Phosphoric acid may have helped in HCl and caustic soda completely (proposed further
the object in a chemical retention of surface oxide layer removed oxide layer research of
bath followed by Inexpensive – can be carried out by Health & safety concerns – phosphoric acid
thorough rinsing homeowner (dichloromethane) or proposed European Council ban treatment)
Can include the use of specialist on use of dichloromethane (now
mild detergents and acid Effective in force)
cleaners Controllable Dangerous; needs thorough
Can leave inhibited surface rinsing and drying
Attacks tin and zinc coatings
226
Phosphorus containing solutions can Cannot be carried out in situ
form protective oxides Bath dimensions dictate size of
2-4% phosphoric acid (pickling acid) objects/ structures treated
useful for removal of heavy rust Dips only suitable for dismantled
staining structures
Blast Airborne abrasive Inexpensive Increased strength noticed in 3- Not recommended

cleaning particles blast iron surface Effective* point bend test – possible
(‘hard’ producing roughened and No moisture* decrease in toughness Generally considered
media – clean surface. Fine abrasive removes some surface Danger of galvanic corrosion if appropriate for cast
alumina, Control: type and contaminants* metal blast medium used and wrought iron.
chilled iron, diameter of grit, blasting Fast Blast media particles and other Not recommended
copper pressure, blasting Effective debris impacted into iron surface
slag, distance, blasting time, Quickest and most effective method Oxide layer not retained Softest abrasive and
olivine, angle between blasting Resultant surface maximises coating Must be carried out by a specialist smallest particle size
garnet, nozzle and surface – these adhesion Expensive* should be chosen
glass grit, factors often dictated in Limited removal of contaminants Dusty* - must be removed before
glass practice by coating
beads, object/structure and Unpleasant for operatives*
Sponge Jet blasting equipment. Can lead to loss of detail on
30, Sponge wrought iron
Jet 120) Hazardous
Aggressive
Risk of damage to fine, decorative
detail
Poor visibility
227
Use limited to structural iron and
steelwork
Waste collection and disposal
costly especially with lead coating
detritus
Blast As above. Fast More expensive Recommended
cleaning Highly effective Must be carried out by a specialist
(‘soft’ Oxide layer well preserved (under controlled
media - No increase in strength of substrate
walnut, or impacted blast media conditions)
plastic, dry Dry ice blasting leaves no residue
ice)
Wet blast Effective Immediate post-cleaning corrosion -----
cleaning Removes soluble contaminants Oxide layer generally not retained
(with glass No dust Must be carried out by a specialist
grit) Preferable to dry blasting where lead Drying problems cause re-rusting
based paints are present Possible penetration of water at
Effective washing of iron chlorides junctions
and sulphates from corrosion pits Slow
Effective – less so than dry blasting Poor visibility
but more controllable Run off can stain surrounding
building material
Disposal of resultant slurry difficult
Nb: Test on cast iron but results likely useful when considering wrought iron surface preparation.
Nb: Aim to retain closely adherent oxide layers influences perceived advantages/disadvantages.
* Relates to sandblasting, the use of which is now prohibited in Britain and elsewhere
228
9.2 Appendix 2: Table of coating details
Details of coating chemistry, surface preparation requirements, application methods and curing and overcoating intervals.
Manuf Coating Chemistry Surface Preparation Application Coats Application Overcoating Interval
Conditions
Primer Metagard Two-pack Sa2.5. Primer for blast cleaned substrates. Average Brush DFT brush: Ambient 10oC + Recoatable:
L574 epoxy surface profile in the range 30-50 microns. Roller 20µm/coat Substrate 3oC 15°C 23°C 35°C
Spray above dew 4hrs 3hrs 2hrs
point
Primer/ Epigrip Two-pack Sa2.5. Average surface profile in the range 50-75 Brush DFT brush: Ambient 10oC + Recoatable:
build C400V3 epoxy – zinc microns. Roller 65µm/coat Substrate 3oC 15°C 23°C 35°C
Leighs
rich Ensure surfaces to be coated are clean, dry and free Spray above dew 5 hrs 3½ hrs 2 hrs
from all surface contamination. Manually prepared point Resistex within 7
surfaces to min. St3. Application to such surfaces RH≤90% days.
should be by brush or roller where the mechanical
action will aid adhesion.
Topcoat Resistex Two-pack Ensure surfaces to be coated are clean, dry and free Brush DFT brush: Ambient 10oC + Recoatable:
C137V2 acrylic from all surface contamination. Roller 25-50 Substrate 3oC 15°C 23°C 35°C
polyurethane Spray µm/coat* above dew 8 hrs 6 hrs 4 hrs
point
Primer Red Oxide Oil-based Surfaces clean and dry. Remove loose rust, scale and Brush 2 Min. 6 hours
Hammeri
Primer paint with wire brush then rub down with coarse Roller between coats.
abrasive paper to provide firm surface. Abrade bare Topcoat best results
te
steel with coarse abrasive paper. Degrease all after 24 hours.

surfaces with Hammerite Brush Cleaner and Thinners
and allow to dry.
229
Primer No. 1 Rust Oil-based Remove loose rust, scale and paint by wire brushing Brush 2 8-25oC Touch dry 2 hours.
Beater or rubbing down with coarse abrasive paper. Abrade Min. 6 hours
clean steel with emery paper to provide a key. between coats.
Degrease where necessary with white spirit, rinse Topcoat best results
with clean water and allow to dry. after 24 hours.
Topcoat Garage Oil-based Surface dry and free from loose rust and mill scale, Brush 1 or 2 8-25oC Touch dry 4 hours.
Door paint, oil and grease. Lightly abrade painted surfaces Recoatable 16 hours.
with suitable abrasive paper. Wire brush to remove
rust if necessary. Degrease where necessary with
white spirit.
Single Direct to Oil-based Abrade smooth surfaces and remove all loose rust Brush 1 (prev. 8-25oC Touch dry 1-2 hours.
coat Rust Metal and flaking paint with a wire brush. Remove dirt, Spray coated) or Do not over Recoatable 4 hours.
system Paint dust and grease with diluted detergent. Rinse with 2 (bare, apply – sagging.
clean water and allow time to dry. rusty)
* The actual thickness within the quoted range will depend on many variables including ambient conditions, type of brush or roller used and operator expertise
230
9.3 Appendix 3: Leighs Metagard L574 product technical data
Leighs Metagard primer manufacturer datasheet.
231
232
9.4 Appendix 4: Leighs Epigrip C400V3 product technical data
Leighs Epigrip build coat manufacturer datasheet.
233
234
9.5 Appendix 5: Leighs Resistex C137V2 product technical data
Leighs Resistex topcoat manufacturer datasheet.
235
236
9.6 Appendix 6: Hammerite Red Oxide Primer datasheet
Hammerite Red Oxide Primer manufacturer datasheet.
http://www.hammerite.co.uk/guide/red_oxide_primer.jsp
237
9.7 Appendix 7: Hammerite Garage Door datasheet
Hammerite Garage Door manufacturer datasheet.
http://www.hammerite.co.uk/guide/garage_door_paint.jsp
238
9.8 Appendix 8: Hammerite Direct to Rust datasheet
Hammerite Direct to Rust manufacturer datasheet.
http://www.hammerite.co.uk/guide/direct_to_rust_metal_paint_satin_finish.jsp
239

2015 Emmersonnjphd

Uploaded by

Copyright:

Available Formats

2015 Emmersonnjphd

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

2015 Emmersonnjphd

Uploaded by

Copyright:

Available Formats

Heritage wrought iron: towards the

development of evidence based

Thesis submitted to Cardiff University in candidature

At Cardiff University, thanks to the Conservation team of Jane Henderson, Yiota

Finally, thanks to my family for their unfailing support in all things.

1.1 The nature of historic wrought iron .............................................................. 1

1.1.1 Production and standardisation............................................................. 1

1.1.2 Composition and properties .................................................................. 3

1.1.3 Form and function .................................................................................. 5

1.2 Research in heritage wrought iron ................................................................ 7

1.3 The study in context ...................................................................................... 9

1.4 Aim and objectives ...................................................................................... 10

1.5 Thesis structure ........................................................................................... 10

2 Corrosion Principles: Heritage wrought iron and its contexts ........................... 12

2.1 Principles of corrosion ................................................................................. 12

2.1.1 The corrosion process .......................................................................... 12

2.1.2 Corrosion of iron: an overview ............................................................ 14

2.2 Corrosion prevention by protective coatings.............................................. 18

2.2.1 Mechanisms of protection ................................................................... 19

2.2.2 Factors affecting coating performance and failure ............................. 21

2.3 Corrosion variables ...................................................................................... 25

2.3.2 Climate ................................................................................................. 26

2.3.4 Intrinsic factors..................................................................................... 33

2.3.5 Pollutants and contaminants ............................................................... 35

2.4 Corrosion products and their properties .................................................... 38

2.4.2 αFeOOH (Goethite) .............................................................................. 39

2.4.3 βFeOOH (Akaganeite) .......................................................................... 39

2.4.4 γFeOOH (Lepidocrocite) ....................................................................... 41

2.4.5 Fe3O4 (Magnetite) ................................................................................ 41

2.4.6 α-Fe2O3 (Haematite) ............................................................................. 41

2.4.7 ƴFe2O3 (Maghemite) ............................................................................ 42

2.4.8 Ferrous and ferric chlorides ................................................................. 42

2.5 Corrosion context ........................................................................................ 43

3 Corrosion Prevention Treatments ..................................................................... 48

3.1 Traditional treatment of wrought iron ........................................................ 48

3.2 Current practice........................................................................................... 50

3.2.1 Recording and dismantling .................................................................. 51

3.2.2 Cleaning and surface preparation ........................................................ 51

3.2.3 Repairs .................................................................................................. 53

3.2.4 Protective coatings............................................................................... 54

3.2.5 Application of protective coatings ....................................................... 55

3.2.6 Maintenance ........................................................................................ 55

3.3 Evidence from paint layers in section ......................................................... 56

3.4 Decision-making in conservation of historic wrought iron ......................... 60

3.4.1 Decisions and decision-makers ............................................................ 60

3.4.2 Guidance .............................................................................................. 64

3.4.3 Use of standards in heritage contexts ................................................. 65

4.1 Ethos and rationale ..................................................................................... 69

4.1.1 Research ethos ..................................................................................... 69

4.1.3 Rationale for methods and parameters ............................................... 71

4.2 Investigating the impact of surface preparation method on corrosion of

4.2.1 Aims and Objectives ............................................................................. 76

4.2.2 Characterisation of sample material .................................................... 76

4.2.3 Surface preparation methods .............................................................. 78

4.2.4 Examination of resultant surfaces ....................................................... 80

4.2.5 Measuring oxygen consumption and corrosion rates ......................... 81

4.2.6 Examining corrosion rates.................................................................... 82

4.2.7 Identification of corrosion products .................................................... 83

4.2.8 Flame cleaning ..................................................................................... 84