PLACE IN RETURN Box to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE | DATE DUE DATE DUE

 

£12633“ ‘°

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 c:lCIRCJDateDue.p65—p.1s

THE IDENTIFICATION AND CHARACTERIZATION OF ARTISTS’ DYES
AND PIGMENTS USING LASER DESORP’l‘ION/IONIZATION MASS
SPECTROMETRY
By

Donna M. Grim

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirement
for the degree of
Doctor of Philosophy
Department of Chemistry

2003

ABSTRACT
THE IDENTIFICATION AND CHARACTERIZATION OF ARTISTS’ DYES
AND PIGMENTS USING LASER DESORPTION/IONIZATION MASS
SPECTROMETRY ‘
By

Donna M. Grim

Laser desorption mass spectrometry (LDMS) was evaluated as an analytical tOOl
for the characterization of dyes and pigments found in ink and in artists’ paints.
Experimentation revealed that colorants can be easily detected on a variety Of surfaces.
Samples included ink from modern ballpoint pens, and ink found on stamps and on
newspapers. However, the primary focus Of this work rested in the LD mass
spectrometric analysis of standard watercolor and Oil paints.

Artists’ paints are typically three component systems that include a colorant,
vehicle, and resin. Currently, there are numerous analytical techniques available and
used by art conservators to analyze these components individually, and a limited number
are available for their analysis as mixtures. In the specific area Of the identification and
characterization of colorants, scientists have cited significant weaknesses in these
methods. Consequently, there is a need for the development Of new methods having the
ability to identify more than one colorant in an impure mixture, and tO provide molecular
level information on both organic and inorganic colorants, simultaneously. Laser
desorption mass spectrometry (LDMS) is an ideal candidate.

The LD mass spectra of numerous colorants will be presented, which demonstrate

the potential for the development Of LDMS as a sensitive analytical technique to be used

in art conservation laboratories. The molecular level information generated in LD mass
spectra can lead to the unambiguous identification of specific colorants present in an
impure paint mixture. For example, the LD mass spectra Of Colorants used tO create an
illuminated page of the Koran from the 17‘h century, were obtained and the identified
colorants offer insight intO the authenticity of the document. Furthermore, sample size
considerations and challenges typically encountered in the analysis Of whole documents
and paintings were identified and addressed. Proposed solutions were tested and the
results suggest that the sampling challenges will not limit the use Of LDMS as a tOOl in

the field Of art conservation.

TO my husband, Mike, for his patience, understanding, and relentless love and support.

iv

ACKNOWLEDGEMENTS

I would like to thank my advisor, Dr. John Allison, fOr all of his guidance and
support throughout my graduate school career. You are a wonderful mentor and a

positive influence on all Of your students. Thank you for everything!

I would like to thank my committee members, Dr. Jay Siegel, Dr. Simon Garrett,

Dr. James Jackson, and Dr. Gary Blanchard for helping my project along the way.

I would also like to thank Dr. Karen Trentelman and Valerie Baas, from the

Detroit Institute of Arts, for their insightful conversations and sample donations.

I would like tO thank Dr. Susan J. Masten, from the Environmental Engineering
Department, Michigan State University, for the use Of her facilities to perform ozone

experiments.

On a personal note, I would like to thank Mike, J asminka, Richie, Eric, Nicole,
and Paul for their friendship and helping to make my short stay in Michigan a little bit

brighter. Also, thank you to my family for always believing in me.

TABLE OF CONTENTS
List of Tables .......................................................... xi
List of Figures .................................... i .................... xiii
Chapter One: Introduction
Project Overview .................................................. 1
The Analyte: Dyes and Pigments ..................................... 7

The Technique: Laser Desorption/Ionization Mass Spectrometry

Overview Of Mass Spectrometric Techniques ..................... 10
Development Of LDMS ....................................... 12
Instrumentation ............................................. 16
Calibration ................................................. 17
Sample Analysis ............................................ 18
TheExperiment: LDMS OfCOlorants................... ........... 18
What to Expect ................................................... 21
Isotopic Profiling ................................................. 26
References ....................................................... 27

Chapter Two: Watercolors

Introduction ...................................................... 3 1
Experimental ..................................................... 36
Summary of Methyl Violet Degradation Study .......................... 38

Identification Of Watercolor Dyes (Unknown Dye Composition)
Cobalt Blue ................................................ 49

Prussian Blue .............................................. 53

vi

Lemon Yellow ............................................. 55

Viridian (Green) ............................................ 60
Carmine (Red) ......................... i .................... 63
Vermilion (Red) ............................................ 69
Identification Of Watercolor Dyes (Known Dye Composition) .............. 71
Cadmium Red Hue 095 ....................................... 72
Mauve (Dark Red) .......................................... 76
Preliminary Light and Environmental Studies ........................... 80
Summary ........................................................ 87
References ....................................................... 88

Chapter Three: Lake Pigments

Introduction ...................................................... 92
Carmine Lake .................................................... 93
Rose Madder ..................................................... 98
Summary ....................................................... 104
References ...................................................... 105

Chapter Four: Inorganic Pigments

Introduction ..................................................... 106
The Case Of the Two Prussian Blues ................................. 110
“G13” ......................................................... 122
The LDMS Analysis Of Two Illuminated Manuscripts ................... 131
Example 1 ................................................ 134
Example 2 ................................................ 135

vii

Gold ..................................................... 143
Black .................................................... 146
“Green” .................................................. 154
Hebrew Manuscript ............................................... 160
A New MALDI Matrix? ........................................... 169
References ...................................................... 180

Chapter Five: Stamps

Introduction ..................................................... 185
Experimental .................................................... 1 88
The Carmine Stamp .............................................. 189
The French Spy Stamps ........................................... 191
ThePbCrO48tamp..................................... .......... 198
Summary ....................................................... 204
References ...................................................... 205

Chapter Six: Studies of the Interactions of H28 with Colorants

Introduction ..................................................... 206
Experimental .................................................... 209
Preliminary Work ................................................ 210
Designing the H28 Exposure Experiments ............................. 211
What to Expect? ................................................. 213
Uncontrolled H28 Studies .......................................... 219
Controlled H28 Exposure Studies .................................... 228

viii

Summary ....................................................... 23 1
References ...................................................... 232

Chapter Seven: Newspaper Ink

History Of the Printing Press ........................................ 233
Printing Methods ................................................. 234
Newspaper Ink .................................................. 236
Experimental .................................................... 238
Results
A Current Newspaper ....................................... 238
Aged Newspaper Ink ....................................... 242
Peak Assignments .......................................... 245
Weathering ............................ . ................... 247
Scientific American Journal ....................... ‘ ........... 248
References ...................................................... 252

Chapter Eight: Practical Experimental Aspects

The Problem .................................................... 254
“GO Big” ....................................................... 255
“GO Small”
The Analysis Of Questioned Documents ......................... 258
The Analysis Of Paintings .................................... 260
References ...................................................... 268
Chapter Nine: Conclusions and Future Directions ......................... 269
References ...................................................... 272

ix

Appendix ............................................................ 273

List of Tables
Chapter One: Introduction

Table 1.1: Summary Of analytical instrumentation available at the Getty Conservation

Institute, Los Angeles , CA ........................................ 4
Table 1.2: Summary Of MOLART molecular level studies Of artists’ pigments ........ 5
Table 1.3: Summary Of allowed transition states ................................. 8
Table 1.4: Desorption/ionization methods used in mass spectrometry ............... 12
Table 1.5: Manual calculation for the isotopic distribution Of C12 .................. 27

Chapter Two: Water Colors

Table 2.1: A summary of the ASTM lightfastness scale ......................... 36
Table 2.2: Artist-grade watercolors subjected to LDMS analysis .................. 37
Chapter Three: Lake Pigments

Table 3.1: A list of the identified positive ions for carrnine lake ................... 94
Chapter Four: Inorganic Pigments

Table 4.1: Summary of the pigments discussed in the three volume series Artists’
Pigments: A History of Their Manufacture and Characteristics .......... 108

Table 4.2: A list Of the identified negative ions formed by laser desorption from Prussian
blue (aqueous) on paper, and the corresponding oxidation states Of Fe. . . . 113

Table 4.3: A list of the identified positive and negative ions for G13 in linseed Oil. . . 126

Table 4.4: A summary of possible molecular weights Of the components of gallotannic
acid ......................................................... 149

Chapter Five: Stamps

Table 5.1: A list of colorants used in contemporary postage stamp ink formulations. . 186

xi

Appendix

Table 1: Summary of Positive Ions Generated by Pb-compounds in the LDMS
Experiment................................_ .................... 274

Table 2: Summary Of Negative Ions Generated by Pb-compounds in the LDMS
Experiment .................................................... 275

Table 3: Summary Of Positive and Negative Ions Generated in the Uncontrolled
H28 (g) Exposure Experiments .................................... 276

Table 4: Summary of Positive and Negative Ions Generated in the Uncontrolled
H28 (g) Exposure Experiments Followed by Treatment with H202 ........ 277

xii

List of Figures
Chapter One: Introduction

Figure 1.1: Matrix-assisted laser desorption time-of—flight mass spectrometer
(Voyager - DE mass spectrometer) ................................ 15

Figure 1.2: Summary Of the steps involved in the desorption/ionization of an inorganic
compound .................................................... 20

Figure 1.3: a) Positive ion LD mass spectrum of Ink A on paper; b) Negative ion LD
mass spectrum of Ink B on paper .................................. 22

Figure 1.4: a) Positive ion LD mass spectrum of copper phthalocyanine (Aldrich) on
paper; b) Negative ion LD mass spectrum of copper phthalocyanine on

paper ........................................................ 25

Figure 1.5: Theoretical isotopic pattern for a) C]; b) C12 ......................... 26
Chapter Two: Watercolors

Figure 2.1: The structure of methyl violet and designations for its degradation
Products ..................................................... 39

Figure 2.2: Preliminary UV-accelerated aging study: Bic ballpoint pen ink on
paper ........................................................ 40

Figure 2.3: A portion of the positive ion LD mass spectrum Of Bic black ballpoint pen
ink-on-paper; a) ink from a new document; b) a 38-month Old controlled,
naturally aged document; c) ink from a document irradiated for 6.25 hours
with UV light ................................................. 43

Figure 2.4: UV accelerated aging study data for new Bic black ballpoint pen ink on
printer paper: a plot Of the relative intensity of the m/z 372, 358, and 344
(x2) peaks versus irradiation time .................................. 46

Figure 2.5: UV accelerated aging study data for new Bic black ballpoint pen ink on
printer paper: a plot Of the average molecular weight of the dye, methyl
violet, versus minutes Of UV irradiation ............................ 48

Figure 2.6: Controlled, natural aging study data for Bic black ballpoint pen ink on printer
paper: a plot of the average molecular weight of the dye, methyl violet,
versus the age of the document .................................... 48

Figure 2.7: Cobalt Blue (N iji Watercolors): a) the partial positive ion LD mass spectrum;
b) the partial negative ion LD mass spectrum ........................ 51

xiii

Figure 2.8: Copper phthalocyanine (Aldrich): a) the partial positive ion LD mass

spectrum; b) the partial negative ion mass spectrum ................... 52

Figure 2.9: Prussian blue (Niji Watercolors): a) the partial pOsitive ion LD mass

Figure 2.10:

Figure 2.11:

Figure 2.12:

Figure 2.13:

Figure 2.14:

Figure 2.15:

Figure 2.16:

Figure 2.17:

Figure 2.18:

Figure 2.19:

Figure 2.20:

Figure 2.21:

spectrum; b) the partial negative ion LD mass spectrum ............... 54
Lemon Yellow (Niji Watercolors): a) the partial positive ion LD mass
spectrum; b) the partial negative ion LD mass spectrum ............... 58
Viridian (Niji Watercolors): a) the partial positive ion LD mass
spectrum; b) the negative ion LD mass spectrum .................... 62
Carmine (Niji Watercolors): a) the partial positive ion LD mass
spectrum; b) the negative ion LD mass spectrum .................... 66
Vermilion (N iji Watercolors): a) the partial positive ion LD mass
spectrum; b) the negative ion LD mass spectrum .................... 70
Structures of: a) PR 149; b) PR 255 .............................. 72
Cadmium Red Hue O95 (Winsor & Newton): a) the positive ion LD mass
spectrum; b) the negative ion LD mass spectrum .................... 73
Enlarged regions Of the Cadmium Red Hue 095 LD mass spectra: a) + ion,
m/z 365-395; b) + ion, m/z 460-520; c) — ion, m/z 470-500 ............ 75
Structures of: a) PR 122; b) PV23RS ............................. 78
Mauve 398 (Winsor & Newton): a) the partial positive ion LD mass
spectrum; b) the partial negative ion LD mass spectrum ............... 79
Vermilion (Niji Watercolors) exposed to 24 hours UV-light: a) the partial
positive ion LD mass spectrum; b) the partial negative ion LD mass
spectrum ................................................... 82
Vermilion (Niji Watercolors) exposed to 10 hours light from a Hg-arc lamp:
a) the partial positive ion LD mass spectrum; b) the partial negative ion LD
mass spectrum .............................................. 82
Vermilion (N iji Watercolors) exposed 3 months tO natural sunlight: a) the

partial positive ion LD mass spectrum; b) the partial negative ion LD mass
spectrum ................................................... 84

xiv

Figure 2.22: Mauve 398 (Winsor & Newton) exposed 12.5 minutes to O3 (375 ppm):

a) the partial positive ion LD mass spectrum; b) the partial negative ion LD
mass spectrum ............................................... 86

Chapter Three: Lake Pigments

Figure 3.1:

Figure 3.2:

Figure 3.3:
Figure 3.4:
Figure 3.5:

Figure 3.6:

Figure 3.7:

Figure 3.8:

The LD mass spectra of carmine lake in linseed Oil on paper: a) the partial
positive ion mass spectrum and structure Of carmine alum lake; b) the
negative ion mass spectrum ...................................... 95

The partial negative ion LD mass spectrum Of carminic acid in linseed Oil on

paper ........................................................ 96
The structure of carminic acid .................................... 96
Structure of a) alizarin; b) purpurin; and c) pseudopurpurin ............. 99
Structure of alizarin lake, PR 83 ................................. 100
The positive ion LD mass spectra Of lake pigments in water on paper:

a) Rose Madder and b) R37 ..................................... 101
The negative ion LD mass spectra of lake pigments in water on paper:

a) Rose Madder and b) R37 ..................................... 102
The structure of Rhodamine 6G .................................. 103

Chapter Four: Inorganic Pigments

Figure 4.1:

Figure 4.2:

Figure 4.3:

Figure 4.4:

Figure 4.5:

 

The structure of ferric ferrocyanide ............................... 110

The LD mass spectra of Prussian blue (Sinopia) from an aqueous solution of
paper: a) the partial positive ion mass spectrum; b) the negative ion mass
spectrum. ................................................... 115

The LD mass spectra of ferric ferrocyanide (Aldrich) from an aqueous
solution of paper: a) the positive ion mass spectrum; b) the partial negative
ion mass spectrum ............................................ 117

The LD mass spectra of Prussian blue (Niji) from an aqueous solution on
paper: a) the partial positive ion mass spectrum; b) the partial negative ion
mass spectrum ............................................... 118

The LD mass spectra of paper (Hammermill Fore MP): a) the positive ion
mass spectrum; b) the negative ion mass spectrum ................... 121

XV

Figure 4.6: Examples Of pigment vials from the case of pigments found in the Michigan

State University’s archives ...................................... 122
Figure 4.7: The LD mass spectrum Of G13 in linseed oil on _a gold sample plate: a) the
partial positive ion mass spectrum; b) the partial negative ion mass
spectrum .................................................... 123
Figure 4.8: Enlarged portions of the positive and negative ion LD mass spectra of G13 in
linseed Oil on a gold plate: (a, c, e) experimental data; (b, d, f) theoretical
mass spectra ................................................. 124
Figure 4.9: An itemized list of the contents of the box .......................... 127
Figure 4.10: A portion Of the list describing the contents Of the pigment vials found in
the box Of pigments .......................................... 128
Figure 4.11: Energy dispersive x-ray spectra: a) Prussian blue; b) chrome green. . . . 130
Figure 4.12: A page of the Koran, allegedly from the 17‘h century ................ 132
Figure 4.13: A page from a Hebrew manuscript, allegedly from the 18003 ......... 133
Figure 4.14: The positive ion LDMS spectrum of rhodamine 6G. The structure Of the
dye 1s shown. The singly charged dye molecule has a monoisotopic
molecular mass Of 443 ........................................ 135
Figure 4.15: LDMS spectra of the ions formed by UV laser irradiation Of a gold surface:
a) positive ion LDMS spectrum; b) negative ion LDMS spectrum. . . . .137
Figure 4.16: A portion of the page of the Koran used in this study ................ 138
Figure 4.17: a) Negative ion LD mass spectrum Of the red ink/dye region Of the Koran
sample. b) Expanded view of the higher m/z portion Of the spectrum. . . .140
Figure 4.18: a) The partial positive ion and b) the partial negative ion LD mass spectra of
the gold region of the Koran sample .............................. 144
Figure 4.19: Gall nuts ................................................... 147
Figure 4.20: Structures Of gall nut components ............................... 149
Figure 4.21: a) The partial positive ion and b) the negative ion LD mass spectra of
homemade iron gallotannate ink on paper ......................... 150
Figure 4.22: a) The partial positive ion and b) the negative ion LD mass spectra Of the

black ink Of the Koran sample .................................. 153

xvi

Figure 4.23:

Figure 4.24:

Figure 4.25:

Figure 4.26:

Figure 4.27:

Figure 4.28:

Figure 4.29:

Figure 4.30:

Figure 4.31:

Figure 4.32:

Figure 4.33:

Figure 4.34:

Figure 4.35:

Figure 4.36:

Figure 4.37:

a—c) Negative ion LD mass spectra Of the “green dot” taken from the outer
edge Of the dot, continuing tO the center Of the dot .................. 156
a) The partial positive ion and b) the negative iOn LD mass spectra Of
malachite (Sinopia) suspended in linseed Oil and applied to paper ...... 157
a) The partial positive ion and b) the partial negative ion LD mass spectra Of
verdigris (Sinopia) suspended in linseed Oil and applied to paper ....... 159
Portion Of the 1800s page of an Arabic manuscript that was evaluated. . .161
Postive ion spectra of a) the blue framing line, and b) the black framing line
Of the Hebrew manuscript page ................................. 163
The partial positive ion LD mass spectrum of the blue paint on the Hebrew
manuscript .................................................. 164
a) The partial positive ion and b) the partial negative ion LD mass spectra of
the orange paint on the Hebrew manuscript ....................... 166
a) The partial positive ion and b) the partial negative ion LD mass spectra Of
the green paint on the Hebrew manuscript ........................ 168
A 16th century Icelandic illuminated manuscript ......... ' ........... 170
The negative ion LD mass spectrum Of red ochre in linseed oil on

paper ...................................................... 172
The structures Of the primary unsaturated acids composing linseed
Oil ........................................................ 173
Summary Of LDMS experiments involving Fe-containing pigments
suspended in linseed oil ....................................... 173
Illustration Of Tanaka’s et a1. “ultra fine metal + liquid matrix”

method .................................................... 174
Partial negative ion LD mass spectrum of Fe203 suspended in walnut Oil
applied to paper .............................................. 178
Partial negative ion LD mass spectrum of R203 suspended in poppy oil
applied to paper .............................................. 179

xvii

Chapter Five: Stamps

Figure 5.1:

Figure 5.2:

Figure 5.3:

Figure 5.4:

Figure 5.5:

Figure 5.6:

The negative ion LD mass spectrum of the 2¢ carmine stamp .......... 189

The spy stamps: a) the original French stamp; b) the forged British
stamp ...................................................... 191

The LD mass spectra of the original French stamp: a) positive ions;
b) negative ions .............................................. 192

The LD mass spectra of the forged British stamp: a) positive ions;
b) negative ions .............................................. 193

a) The general structure of a B-napthol pigment; b) the general structure
Of a naphthol AS pigment ....................................... 195

Proposed structures for a) m/z 157 and b) m/z 172 ................... 196

Figure 5.7: The Gold Star Mothers Stamps: a) Exposed to H28 (g), followed by treatment

Figure 5.8:

Figure 5.9:

with H202 (l); b) Exposed tO H28 (g); c) original (untreated) ........... 199

The positive ion LD mass spectra of the PbCrO4 stamps: a) the original
stamp; b) the stamp exposed tO H28 (g); c) the stamp exposes to H28 (g)
followed by treatment with H202 (l) .............................. 201

The negative ion LD mass spectra Of the PbCrO4 stamps: a) the original
stamp; b) the stamp exposed tO H28 (g); c) the stamp exposes tO H28 (g)
followed by treatment with H202 (l) .............................. 202

Chapter Six: Studies of the Interactions of H28 with Colorants

Figure 6.1:

Figure 6.2:

Figure 6.3:

Figure 6.4:

The LD mass spectra Of G13 in linseed Oil; a) the partial positive ion mass
spectrum; b) the partial negative ion mass spectrum .................. 212

The LD mass spectra Of PbS in linseed Oil on paper (sample 1): a) the partial
positive ion mass spectrum; b) the partial negative ion mass spectrum. . . 215

The LD mass spectra of PbS in linseed Oil on paper (sample 2): a) the partial
positive ion mass spectrum; b) the partial negative ion mass spectrum. . . 216

The LD mass spectra Of PbS in linseed Oil on paper treated with H202 (l):

a) the partial positive ion mass spectrum; b) the partial negative ion mass
spectrum .................................................... 218

xviii

Figure 6.5:

Figure 6.6:

Figure 6.7:

Figure 6.8:

Figure 6.9:

The partial positive ion LD mass spectra Of the G13 pigment in linseed Oil on
paper samples exposed to 2 minutes H28 (g) created using an HCl solution
diluted; a) 75%; b) 50% ........................................ 221

The partial positive ion LD mass spectra Of the G13 pigment in linseed Oil on
paper samples exposed to 2 minutes H28 (g) created using an HCl solution
diluted; a) 25%; b) 0% ......................................... 222

The partial negative ion LD mass spectra Of the G13 pigment in linseed Oil
on paper samples exposed to 2 minutes H28 (g) created using an HCl
solution diluted; a) 75%; b) 50% ................................. 224

The partial negative ion LD mass spectra of the G13 pigment in linseed Oil
on paper samples exposed to 2 minutes H28 (g) created using an HCl
solution diluted; a) 25%; b) 0% .................................. 225

The LD mass spectra Of G13 in linseed Oil on paper exposed to 15 minutes
H28 (g) and treated with H202 (l): a) the partial positive ion mass spectrum;
b) the partial negative ion mass spectrum .......................... 227

Figure 6.10: The LD mass spectra Of PbCrO4 in linseed Oil on paper exposed to 65 hours

H28 (g); a) the partial positive ion mass spectrum; b) the partial negative
ion mass spectrum ........................................... 230

Chapter Seven: Newspaper Ink

Figure 7.1:

Figure 7.2:

Figure 7.3:

Figure 7.4:

Figure 7.5:

Figure 7.6:
Figure 7.7:

Figure 7.8:

The partial positive ion LD mass spectra of a recent issue Of the State News:

a) the black ink; b) the paper .................................... 240
L2MS spectrum from a petroleum pitch sample ..................... 241
The partial positive ion LD mass spectra of black newspaper ink: a) a 1986

sample; b) a 1963 sample ....................................... 243
The structure Of coronene ....................................... 246

An enlarged portion of the positive ion LD mass spectrum of the 1986 ink
sample ..................................................... 246

The positive ion LD mass spectrum Of the Scientific American Journal. . .250
The positive ion LD mass spectrum of yellow newspaper ink on paper. . . 251

A proposed structure for the compound represented at m/z 250 ......... 251

xix

Chapter Eight: Practical Experimental Aspects

Figure 8.1:

Figure 8.2:

Figure 8.3:

Figure 8.4:

Figure 8.5:
Figure 8.6:
Figure 8.7:

Figure 8.8:

Figure 8.9:

Illustration of the sample size problem. Samples which can be analyzed

using LDMS are limited to a 6 in2 sample plate. .................... 254
A MALDI MS similar to the instrument in our laboratory ............. 255
A cartoon Of a MALDI instrument with an enlarged ion source housing to

accommodate large samples, such as a painting ..................... 256
A 14‘h century Spanish ciborium ................................. 257

Examples Of two punches used by questioned document examiners to remove
microplug samples from documents in question ..................... 259

Illustration Of the microplug experiment. The negative ion mass spectrum Of
the red ink on the 17th century Koran presented in Chapter 4 ........... 260

A cartoon Of the paint chip in linseed Oil experiments ................. 261
The LD mass spectra Of the Detroit Institute of Arts paint chip sample
in linseed Oil on a gold sample plate: a) the positive ion mass spectrum;

b) the negative ion mass spectrum ................................ 262

The positive ion mass spectra Of an early 20th century painting: a) green
paint chip; b) white paint Chip; c) red paint chip ..................... 264

Figure 8.10: The positive ion LD mass spectrum of the red paint ground up, in linseed

Oil, and analyzed on a gold sample plate .......................... 265

Figure 8.11: The negative ion LD mass spectrum of the red paint ground up in linseed

Oil and analyzed on a gold sample plate: a) the first spot irradiated; b) the
second spot irradiated ........................................ 266

XX

Chapter One: Introduction

In 2001, the art community received a shock, when the Smithsonian Institute’s art
conservation laboratory (Washington, DC), one of the most well-known and productive
art conservation laboratories in the nation, was in danger of being shut down, as a result
of governmental funding cutbacks. The chemistry community responded immediately,
and the chemistry of art became a common theme in journal articles and at conferences.
For instance the July 31, 2001 issue of Chemical and Engineering News featured a cover
story entitled, “Chemistry and Art: Helping to Conserve Our Nation’s Treasures” (1).
The article highlighted the achievements of art conservators at the National Gallery of
Art, also located in Washington, DC. As a result of the overwhelming support from both
the art and chemistry communities, as well as the general public, the Smithsonian
Institute’s art conservation laboratory is still active today. Even though‘the chemistry of
art is a narrow component of the field of chemistry, the research is still valued and serves

a significant role in preserving and maintaining the history and cultures of the world.

Project Overview

A paint is a complex system, which can be broken down into three components:
the colorant (dyes or pigments), the vehicle (solvent system), and the resin (binder).
Once the paint is applied to a substrate, a varnish layer (thin, protective coating) may be
added. Scientists work to explain the role of each component in the mixture and how
they interact with each other to create what our eyes perceive as art. This work draws the

attention of art historians, art conservators, and forensic scientists. To narrow the field,

let us consider the chemistry of the colorant. One job of art historians is to decipher the
artist’s ever-changing palette, not only to gain a clear understanding of the artist’s intent
with the colors he selected, but also to learn the provenance. of colorants, which provides
information on pigment trade routes and availability. For example, during the High
Renaissance period (16005) in Venice, the political climate influenced the way in which
Venetian artists portrayed religious figures. Traditionally, artists placed religious
subjects such as the Madonna in the center of the painting. During this time period,
artists began focusing on landscape details; however, at this time and place, artists were
forbidden from choosing landscapes as the primary focus of their paintings. One may
wonder why artists from this period consistently selected autumnal colors for landscape
features in their paintings. The golden features are considered to be almost exaggerated —
what were the artists trying to convey by this consistent choice? The answer to the
question is chemical in nature, since the pigment commonly used throughout Italy to
paint green trees, copper resinate, darkened and turned golden as time passed (2). This
leads into the area of art conservation.

Art conservators are more interested in learning the mechanisms behind colorant
reactions with their environment, including gases in the atmosphere and the paint mixture
itself. Once the mechanism has been confirmed, art conservators can focus on
developing preventative measures, such as paint additives or protective coatings. They
can also work on finding methods to reverse a damaging chemical reaction, or at least
make the color change less apparent.

In an unrelated field, forensic scientists are interested in the chemistry of art, in

regards to the determination of the authenticity of art objects. Simply identifying the

colorants used to create a work of art can sometimes prove that it is a forgery. For
example, the controversial Vinland Map was allegedly created in the early 15‘h century.
Predating Christopher Columbus’ discovery of America (1492), the map depicted regions
of the “New World”. Carbon dating of the parchment was consistent with the alleged
time of creation, however the chemical analysis of the colorants used to create the map
indicated the presence of anastase TiO2, which had not been discovered until the early
20th century (3,4). The Vinland Map was obviously a modern day forgery.

Here we demonstrate the technique of laser desorption/ionization mass spectrometry
(LDMS) as a versatile tool for the analysis Of colorants. Colorants encountered in works
of art and items of historical interest are most Often organic dyes, inorganic pigments, or
lakes (organic dyes on an inorganic support). Based on the interests in art chemistry,
there were two general directions in which the project could proceed. The first and more
basic, is the development of LDMS as a useful analytical tool for the identification of
artists’ colorants. Table 1.1 lists all of the analytical instrumentation available to art
conservators at the Getty Conservation Institute (Los Angeles, CA). It is important to
note that this is an extreme example, considering that the majority of art conservation
laboratories are significantly smaller, and may be limited to only a few and in some
cases, none of these techniques. Considering the list in Table 1.1, it appears as though
artists are currently well-equipped for the identification of colorants. Therefore, the
second and more promising direction was the development of LDMS for the
characterization of the chemical reactions of colorants in response to their environment.
For example, as its name implies, Ultramarine blue is a greenish/blue pigment, which is

known to change into a blotchy-grayish color over time. This transformation is well-

 

known and is called “Ultramarine Sickness”, however the mechanism behind the reaction

is poorly understood (5)

 

 

 

 

 

 

Chemical Analysis Microscopy
UV/vis Spectrosccmy Electron Microprobe
X-ray Fluorescence Environmental scanning electron microscopy
X-ray Diffraction Polarizing light microscopy
Infrared Spectroscopy
GC/MS
LC/MS

 

 

 

 

 

Table 1.1: Summary of analytical instrumentation available at the Getty Conservation
Institute, Los Angeles, CA

On February 1, 1995, the Netherlands Organization for Scientific Research
funded an intense five year research project entitled MOLART (Molecular Aspects of
Aging in Painted Works of Art). The primary goal of MOLART was to contribute to the
development of a scientific framework for the conservation of painted art on the
molecular level. Numerous graduate students participated in the project, and earned their
doctoral degrees by researching a specific area of art conservation. The project focused
on molecular level studies of specific artist materials, falling into three broad categories:
resins and varnishes, polymer networks, and pigments. A detailed description of the
research projects initiated, as well as progress reports for each of the studies, including

listings of presentations and publications, is available (6).

 

Project Title Primary Researcher(s)

 

 

 

 

Copper green glazes Klaas Jan van den Berg
Indigo used in easel paintings Margriet van Eikema Hommes &
Ella Hendriks
LD-ITMS 1n the analysrs of N. Wyplosz
natural organic plgments
Orplment, deterioration of Arie Walle rt

arsenic sulfide pigments

 

J. R. J. van Asperen de Boer &

 

 

 

 

Smalt A. Wallert
Cadmium pigments J. R. J. van Ageren de Boer
Secondary Ion Mass Spectromery Klaas Jan van den Berg &
on paint cross sections Ron Heeren

 

 

Table 1.2: Summary of MOLART molecular level studies of artists’ pigments.

Considering that the focus of our project is the analysis of artists’ pigments, the
projects falling under MOLART’S broad category of the molecular analysis of pigments
will be highlighted. Table 1.2 summarizes the title of each project with the names of the
scientists conducting the research. Each of the colorants listed in Table 1.2 share a
common trait. They all eventually change color, either by reacting with pollutants in the
atmosphere, the vehicle, or with other colorants present in the paint mixture. For
instance, copper resinate is known to change from a green color to a brown color. Each
group proposed a series of questions to be answered. For example, the research team
involved in the indigo project sought to answer the following questions.

1) What are the physical and chemical changes that take place as indigo paint discolors?
2) What are the causes of discoloration?

3) At what rate does this change occur and is it a linear or exponential process?

4) Can the results of aging tests be used to estimate the original appearance of indigo

paint areas that have discolored on paintings?

Several different mass spectrometric techniques were commonly employed to answer
these questions along with many others in additional projects. For example, as the title Of
the project indicates, laser desorption ion trap mass spectrometry (LD-ITMS) was used to
characterize natural organic pigments. Pure reference chemicals, including flavonoids,
anthraquinones, and indigo, as well as the extract of plants (specifically weld, dyer’s
broom, and madder) were analyzed. This project was Of particular interest to us, since
these natural organic colorants have historically been used to prepare lake pigments,
which are known to be extremely light fugitive. Direct LDMS, LD followed by post
electron ionization, and traditional MALDI (Nd:YAG laser, 335 nm) were used to
analyze the colorants. Notably, indigo was detected on the surface of a wool fiber using a
nitrogen laser (337 nm).

The diversity and success Of the MOLART project in the field of art conservation,
helped to formulate our project. Furthermore, their initiative to gain a mOlecular level
understanding of the numerous chemical changes which occur in paintings over time, and
their frequent decision to use mass spectrometry to achieve this goal, was inspiring.
However, a discussion with local art conservators at the Detroit Institute of Arts, helped
to finalize the direction of our project.

During a discussion with art conservators, they surprisingly expressed
dissatisfaction with the methods listed in Table 1.1. For example, x-ray fluorescence
provides elemental information, as opposed to more informative molecular level
information. Suppose a paint mixture is composed of ultramarine blue and red ochre
(Fe2O3). Ultramarine blue does not contain iron, but physically resembles Prussian blue

[Fe4(Fe3CN6)], which does. Complications may arise if the paint mixture is analyzed

using a technique, such as XRF. Iron will be detected due to the presence of the red
ochre, but due to the similar appearance of the two blue pigments, a false positive
identification for Prussian blue may result (7). Additionally, other techniques, such as
infrared spectroscopy, require a pure sample for analysis. The numerous components of
a paint mixture make it difficult to obtain a pure sample from a painting.

Another weakness involves the characterization of lakes. Many of the techniques
available to art conservatorss require the separation of the organic dye from the inorganic
support, which are then analyzed separately. Consequently, art conservators are seeking
the development of new analytical techniques which would allow them to analyze a lake
pigment as a unit. While both project directions were pursued, the primary focus of the
project became to develop LDMS to a point at which it could be a useful analytical

technique in art conservation laboratories for the identification of artists’ colorants.

The Analyte: Dyes and Pigments

Color is the result of molecules absorbing light from the visible region of the
electromagnetic spectrum (450 - 625 nm wavelength). Light consists of oscillating
electric and magnetic fields. The energy of the photon (light) is indirectly proportional to
the wavelength (A) of the photon, by the relationship, E = hc/A, where h = Planck’s
constant and c = the speed of light in a vacuum. When the frequency of the light
oscillation and the frequency of the electron or molecular “transition motion” are the
same, an atom or molecule can absorb energy. Consequently, an electron may be
promoted to a higher energy antibonding orbital by the absorption of light. A transition is

possible for all valence electrons in a molecule. Table 1.3 summarizes the allowed

transition states for sigma (0), pi (7t), and nonbonding (n) electron pairs. The UV laser
used in the LDMS experiments here, emits pulses of light at 337 nm wavelength, thus we
are interested in the n 9 11* and 1t 9 7t* transitions. Molar absorptivity (a ) reflects the
efficiency of light absorption. Thus, a large 8 indicates a high efficiency of light
absorption, which increases the probability of the electronic transition occurring. Typical
molar absorptivity ranges for the n 9 7t* and 1t 9 7t* transitions are 10 — 100 UcmOmol
and 1000 — 10,000 L/cm-mol, respectively (8).

Colorant molecules are generally termed dyes and pigments. Dyes are typically
organic aromatic compounds and are known to be partially or completely soluble in their
“vehicle” (solvent system). The presence of multiple chromophores in a colorant results
in an increased molar absorptivity value, but does not affect Am”. An increase in
conjugation also increases 8, resulting from the additive effect of conjugated double

bonds, but also shifts A"... to a longer wavelength (9).

 

 

 

 

 

 

 

Electronic Transition Wavelength
o 9 0* A < 185 nm (vacuum UV)
n9o* 150<k<250nm
n91t*,7t91t* 200<A<700nm

 

Table 1.3: Summary of allowed transition states.

Inorganic colorants typically fall into the category of pigments. The color of
inorganic pigments may arise from ligand field transitions, charge transfer transitions,
intervalence charge transfer transitions, or metal reflectance (in the case of Au or Ag).

The attachment of specific ligands to metals greatly influences the wavelength of

maximum absorbance. Many transition metal ions are colored in solution due to their
incomplete d-orbitals. The absorption of relatively low energy visible light will promote
an electron to an excited state. Since the d—orbital is the outermost shell, d-electron
excitations are highly influenced by external forces. Consequently, solvent and ligand
effects are significant. 4f and 5f metals (lanthinides and actinides) absorb light because
Of f electron excitation. Ligand and solvent effects are minor considering that the f
electrons are not the outermost electron, and are thus shielded. Charge transfer
absorption involves a complex, consisting of an electron donor and an electron acceptor
(a metal and a complex anion). Absorption of light causes the transfer of an electron
from the donor orbital to the acceptor orbital, resulting in an excited state where an
oxidation/reduction reaction has occurred. The metal ion in the complex serves as the
electron acceptor (10).

Between an organic dye and an inorganic pigment is a “lake”. A lake is a unique
type of pigment having both organic and inorganic components, since it is prepared by
precipitating an organic dye onto an inorganic substrate. By doing so, a dye is given
pigment properties. It is unclear whether or not a lake should be considered as a separate
compound or simply a dye which is chemically adsorbed onto the inorganic surface. In

either case, the color of lakes originates from the adsorbed organic dye absorbing light.

The Technique: Laser Desorption/Ionization Mass Spectrometry

Overview of Mass Spectrometric Techniques
A mass spectrometer is a very powerful detector following separation by gas

chromatography (GC). In the GCMS experiment, volatile components of a mixture, such

as ink, are separated, then ionized using electron impact ionization (EI) and subsequently
analyzed by the mass spectrometer. Desorption/ionization (DI) methods that are used
with mass spectrometry have the advantage of generating iOns for MS analysis from
nonvolatile/thermally labile analytes. Currently there are a number of D1 methods
available. These are summarized in Table 1.4. For example, field desorption (FD) can
be used to generate ions from analyte molecules with molecular weights greater than
1,000 (11). In this method, analyte molecules are deposited onto a FD “emitter”, which
employs a carbon surface. When the surface is heated in the presence of a high electric
field, ions can be generated from the analyte. FD has been used recently for the analysis
of ballpoint pen inks (12). Secondary ion mass spectrometry (SIMS) and FAB are two
very similar techniques (13). In a SIMS experiment, the analyte is deposited onto a metal
surface, which is inserted into the ion source of the mass spectrometer. This target is
continuously bombarded with ions having several keV of kinetic energy. Secondary ions
representative of species on the target surface are generated and analyzed. In the FAB
experiment, the analyte is dissolved in a liquid matrix such as glycerol. This viscous
solution is deposited on a metal surface, and inserted into the ion source of the mass
spectrometer. This target is continuously bombarded using fast (keV) Xe atoms,
typically. Both are used with magnetic sector mass analyzers. We note that both FAB
and SIMS have been used to obtain spectra of nonvolatile organic dyes (14-16), and there
has been one report of using SIMS to analyze colorants on paper (17).

Matrix-assisted laser desorption/ionization (MALDI) (18) and LD (19) are very

similar experiments differing only in the presence of a crystalline matrix used to absorb

10

the laser radiation in MALDI. In both techniques, a pulsed laser ionizes a solid analyte.
The ions generated are separated using time-of-flight mass spectrometry (TOP-MS).
Very recently, the characterization of colorants in ballpoint pen inks was
accomplished using electrospray ionization mass spectrometry (ESIMS), which is another
desorption/ionization method (20). In this experiment, a liquid sample is forced through
a needle held at a high potential, resulting in the formation of multiply-charged droplets.
In the vacuum system, solvent evaporates from the droplets and a point is reached when a
“coulombic explosion” occurs. The original droplet explodes into numerous smaller
droplets, carrying multiple charges. ESI is considered a soft ionization method.
Consequently, ions of characteristic m/z values will appear in the mass spectrum, with
minimal or no fragmentation (21). Negative ion E81 MS has also been used for the

analysis of sulfonated azo dyes (22).

ll

 

 

 

 

 

 

 

 

 

 

 

Analyte . Mass Ionization
M
ethod Form Mechanisms Analyzer Method
Thermal
Field Desorption Adsorbed on excrtatron & Magnetic .
a carbon electric-field Continuous
(FD) . sector
surface Induced
emission
Fast ions
Secondary Ion Mass Adsorgggl on collide with Magnetic Con t' s
Spectrometry (SIMS) surface the surface and sector 1nuou
deposit energy
Fast atoms
Fast Atom Dissolved in a collide with Magnetic Continuo
Bombardment (FAB) liquid matrix the surface and sector us
deposit energy
Matrlx ASSISted Embedded in Absorb laser .
Laser matrix 1i ht from a Tme'0f' Pulsed
Desorption/Ionization cr stals gls d 1 Flight MS
(M ALDI) y pu e aser
Laser Desorption Adsorbed on Absorb laser Time-o f.
a metal light from a . Pulsed
(LD) Flight MS
surface pulsed laser

 

Table 1.4: Desorption/ionization methods used in mass spectrometry

 

Development of LDMS

In 1970, Vastola, Mumma, and Pirone performed what is considered to be one of
the first successful laser desorption/ionization mass spectrometric experiments (23). In
their historic experiment, they were able to produce molecular ion species of several
alkali hexylsulfonate salts, which were known to decompose under analysis using typical
gas phase ionization techniques (EI, CI). Specifically, they observed peaks in their mass
spectrum representing cationized salts and dimers. No fragmentation was noted,

classifying the technique as a “soft ionization” method. In the early developmental stages

12

 

of LDMS, a variety of lasers were employed, having wavelengths ranging in the IR
region to the far UV (24).

Initially, LDMS was well-suited for the analysis of relatively small (< m/z 1000),
nonvolatile, thermally labile, polar molecules. This was in contrast to existing gas phase
ionization techniques, such as E1 or Cl, which required the derivatization of polar groups
prior to analysis, to create a volatile compound. LDMS was further developed and
applied to the analysis of larger, nonvolatile bio-organic molecules. In 1978, Kistemaker
and coworkers experimented with a continuous-wave (cw) CO2 laser (10.6 urn), as well
as a Nd-YAG laser (265 nm) and were able to generate pseudomolecular ions originating
from several oligosaccharides, glycosides, and oligopeptides, by alkali attachment (25).
Considering that LDMS is a pulsed experiment, a time-of-flight (TOF) mass spectrometer
is commonly employed, since it has the ability to generate a complete mass spectrum
from a single laser pulse. However, Stoll and ROllgen used a cw-CO2 laser ion source,
coupled with a quadrupole mass spectrometer to produce molecular ions of carboxylic
acids and tetra-n-butylammonium salts (26).

There have been several proposed mechanisms describing the desorption and
ionization processes (27-31). Evidence has been presented suggesting a thermal process.
In general, absorption of the laser light results in a rapid increase in temperature over a
broad, unfocused area of the substrate, followed by exponential decay of the temperature
over several us. Vastola and Pirone have reported continued emission of ions for several
hundred us following the duration of the laser pulse (30). This results in both a spatial
distribution and a kinetic energy distribution of ions, which have detrimental effects on

resolution in a TOF experiment. Ions of the same m/z value desorbed at different times

13

will have different starting points (spatial distribution). Contrary to laser microprobe
experiments, “unfocused” lasers are typically employed in laser desorption experiments.
The result is the formation of “hot spots”. Consequently, ions Of the same m/z value may
be desorbed with different kinetic energies. To increase resolution in a TOF experiment,
a delay time (~ 100-250 ns) between each laser pulse and the application of an ion-
extraction potential is used (32). This allows time for all of the ions to reach the “starting
point” and reduces the energy spread of desorbed ions to thermal values. Thus, all ions

are theoretically accelerated to the same kinetic energy.

14

 

Linear

   

 

 

 

 

  
  
 

 

 

4 detector
l
l .. __ _..,
| Ion path
I ...... >
I Laser path
I
l
l .
Beam I Filght
guide tu 6
wire '
l
I
I
|
Laser 1 |
1 Video
I camera
Laser — ;— | . A erture
attenuator 1 GP (grounded)
' round grid
Prism V ~ 2 ‘ I

§
\

Variable-voltage ‘ ‘ . .
grid .32.

Sample plate . Main
source chamber

   

Sample
loading
chamber

 
   
 

Figure 1.1: Matrix-assisted laser desorption time-of-flight mass spectrometer
(Voyager-DE mass spectrometer)

15

Instrumentation

Positive and negative ion mass spectra of ink and paint samples were obtained
using a Perseptive Biosystems Voyager delayed extraction time-of-flight (TOF) mass
spectrometer (Framingham, MA). This instrument, illustrated in Figure 1.1, is typically
designed to perform matrix-assisted laser desorption ionization (MALDI) experiments,
however direct LDMS experiments were performed in this work. In the current MALDI
technology, sample plates containing 100-400 “wells” are introduced into the instrument.
One microliter of a sample solution can be placed in a well. The solvent evaporates,
leaving a solid sample target in the well for analysis. The planar target is moved, using
mechanical X-Y devices, so that light from a pulsed laser can be focused onto various
positions. For our purposes, this plate can be easily modified such that a piece of paper
containing ink or paint is introduced, and the paper is moved so that the laser irradiates
one or more locations where ink is present, generating ions for subsequent MS analysis.

Once the sample is introduced into the ion source, the instrument utilizes a pulsed
nitrogen laser (337 nm, 3 ns, 3 Hz) to desorb and ionize the sample. The ionized
molecules are accelerated and analyzed using a linear time-of-flight (TOF) mass
spectrometer. The guide wire located in the TOF tube attracts the ions to keep them on a
focused route to the detector. In the TOF tube, ions of different masses accelerated to the
same kinetic energy will travel at different velocities. As a result, ions with different m/z
values will reach the detector at different times. This allows for separation of the ions
according to velocity. An electron multiplier is employed as the detector.

The electron multiplier is made of a copper-beryllium alloy, which has the ability

to sputter electrons when bombarded with ions. For example, if one fast ion collides with

16

 

the copper-beryllium surface, 10 electrons may be sputtered off. The 10 ejected electrons
are then accelerated to another copper-beryllium surface, which is held at a slightly larger
potential, where each one of the 10 electrons sputters off 10 mOre electrons. The process
continues, until the signal is amplified 105-106 times (33).

The following user-selected parameters were employed: for analysis of positive
(negative) ions formed by LD, the sample plate was held at a voltage of 20,000 V
(-15,000 V), an intermediate acceleration grid was held at 94.5% (94.5%) of the plate
voltage, and a delay time ranging between 100 — 250 ns (dependent on the sample) was
used between laser irradiation and ion acceleration. The manufacturer supplies several
different sample plates, which were employed in these experiments. For example, one
type of sample plate does not contain wells, but is machined such that a polyacrylamide
gel can be attached. When paper is taped onto this metal sample plate, and spectra are
generated at a rate of 3 Hz, full resolution of the instrument is realized using the

conditions cited here. Care must be taken to maintain a flat target.

Calibration

For TOF MS experiments, ion flight times are measured, and m/z values must be
computed. To calibrate LD spectra, a spectrum containing peaks with known m/z values
is first obtained. In most experiments, a saturated solution of CsI (99.9%; Aldrich,
Milwaukee, WI) was pipetted onto paper and allowed to dry. Laser irradiation of this
target yields positive ions including Cs+, and ions with the formula Cs(n+1)ln+, such as
CS2I+. These were used for calibrating the instrument, so that flight times for other ions

could be converted into m/z values in the mass spectra generated. Similarly, for neat ink

17

analysis, 2uL of the same CsI solution were pipetted onto a gold plate containing 100

wells, allowed to dry, and used for generating spectra for calibration.

Sample Analysis

As a result of the different types of samples analyzed in this work, including paint
and ink on paper, stamps, currency, coins, medallions, etc., a variety of different sample
preparation procedures were employed. Consequently, each method will be discussed

when appropriate.

The Experiment: LDMS of Colorants

Why does this experiment work? In theory, it is a relatively easy task to
predeterrnine if one can obtain a UV- laser desorption mass spectrum of an organic
compound. One simply needs to prepare a dilute solution of the organic cOmpound and
obtain a UV/vis spectrum to see if the compound absorbs strongly at 337 am. If so, one
can expect to see a peak in the mass spectrum representing molecular ions, or either
protonated or deprotonated molecules. If not, one can simply employ an organic matrix.
Inorganic compounds are not as simple, since it is much more difficult to determine
whether or not the solid material is going to absorb energy from the UV laser.
Additionally, inorganic compounds exist as crystal lattices, and it is even more difficult
to predict how they will form ions if they are able to absorb the energy from the laser. In
order to try to predict, we need to consider the energy involved for each of the three steps

in the LDMS experiment: 1) absorption, 2) desorption, 3) ionization.

18

Consider the generic crystal lattice shown in figure 1.2. M represents the metal,
and X indicates the ligand. The initial step in the LDMS experiment is irradiation of the
sample using a pulsed N2 laser. If the sample can not absorb photons at 337 nm, the
experiment has ended. Limited absorbance data is available for inorganic compounds in
the solid state. Assuming that the compound absorbs energy, the question becomes how
much energy has been absorbed? The efficiency of light absorption is reflected in the
molar absorptivity of the analyte under the specified conditions. Again, without UV/vis
data, these values are usually difficult to obtain and/or not available. Assuming that the
compound absorbs at this wavelength, the crystal lattice heats up rapidly, rupturing bonds
in the lattice. But which bonds? This remains unknown until the mass spectrum is
obtained. Assuming that a MaxiJ fragment breaks loose from the lattice, will it form an

ion?

19

Desorption/Ionization

 

 

 

 

 

 

 

 

 

(M+X)' (M+X)+
3) Ionization
EA
Eavailable
(M+X) (g)
AHvap 2) Desorption
X..M—X ------ » M+X (S)

 

 

 

| l l
M - X - M 1) Absorption, 337 nm

1 l I
X—M-X

Figure 1.2: Summary of the steps involved in the desorption/ionization of an
inorganic compound.

The first step is a deposition of energy, since the following steps require energy.
The next step is desorption, which involves the sublimation of the analyte in a condensed
phase to the gas phase. Typically, heat of vaporation values may be available for the pure
metal, however the data are not available for fragments of a particular inorganic lattice.

The final step is ionization, which can be a competing process between the
formation of positive and negative ions. In the formation of positive ions, ionization
potentials and proton affinity values of gas phase molecules are considered. In the

formation of negative ions, electron affinities of gas phase molecules are considered.

20

Obviously, ions observed in a mass spectrum are determined by the polarity of the
experiment performed. If sufficient energy is absorbed in the initial step of the
experiment, for both the desorption and ionization steps to occur, ions are observed.
Though we can not predict if and how a particular pigment will generate ions in a LDMS
experiment, we know that the experiment frequently works, because we can obtain

spectra of numerous organic and inorganic pigments.

What to Expect

In regards to mass spectrometric analysis, dyes may be broken down into two
broad classes: ionic and neutral. If the dye is ionic, it is in the form of a salt (C+A'),
containing a cationic portion (C) and an anionic portion (A'). Methyl violet is an
example of a cationic dye and its structure is shown in Figure 1.3a. The dye molecule is
a triphenyl methane dye, with five methyl groups attached to three nitrogen atoms (358
Da). A CT serves as its counterion. Solvent black 29, shown in Figure 1.3b, is an
example of an anionic dye. The structure consists of a Cr3+ center with two identical
ligands, each carrying a 2- charge attached to it, giving the overall structure a 1- charge.
The dye has a metal cation as its counterion. Since these two examples already exist as
ions, they only require enough energy to be desorbed in a LDMS experiment.
Consequently, a peak at m/z 358 representing the intact C” ion would be expected in the
positive ion mass spectrum of methyl violet, and a peak at m/z 666 representing the intact

A" ion would be expected in the negative ion mass spectrum of Solvent black 29.

21

 

372 a)

358

Relative Intensity

 

+NCH ° CI-

 

 

 

 

 

 

Relative Intensrty
!/
e2 \
o

 

 

 

 

N
500 1000 1500 2000

m/z
Figure 1.3: a) Positive ion LD mass spectrum of Ink A on paper; b) Negative ion LD

mass spectrum of Ink B on

 

paper.

22

 

Two industrial ink jet inks were provided to us, with known compositions. Ink A
contains the cationic dye methyl violet and its positive ion LD mass spectrum is shown in
Figure 1.3a (35). Dyes are frequently sold as impure mixtures. In the case of methyl
violet, the hexamethylated form (m/z 372) is more abundant than the pentamethylated
form (m/z 358). This becomes evident when examining the ink’s positive ion LD mass
spectrum (Figure 1.3a). The base peak (largest peak) is seen at m/z 372, which
corresponds to the hexamethylated structure (crystal violet), whereas the peak at m/z 358
represents methyl violet. Note, the manufacturer used methyl violet, but the mass
spectrum shows it was really crystal violet.

Ink B contains the anionic dye Solvent Black 29. The negative ion LD mass
spectrum of Ink B on paper is shown in Figure 1.3b (35). As expected, a peak is present
at m/z 666 representing the intact anionic portion of the dye salt. The structure of
Solvent Black 29 is very stable, and therefore no degradation products were observed in
LD mass spectra.

Neutral dyes are very different than ionic dyes in that they must be desorbed and
ionized to be detected in a LDMS experiment, resulting in a few different expectations in
the mass spectrum. In a positive ion LD mass spectrum of a neutral dye such as copper
phthalocyanine (structure shown in Figure 1.4a), a peak representing the molecular ion is
observed and a peak representing the deprotonated molecule is seen in negative ion mode
(1.4b) (36). In other instances (discussed in Chapter 2) peaks representing both the
molecular ion (M+) and protonated molecule [(M + H)‘] are observed simultaneously in
positive ion mode. Likewise, in the negative ion mass spectrum, peaks representing both

the molecular ion (M') and deprotonated molecule [(M — H)'] were present. The LD mass

23

 

spectra of these examples will be shown and discussed in more detail in later chapters,
however at this point the dyes are presented to gain an understanding of what to expect in
a mass spectrum of similar dye compounds.

Inorganic pigments exist as crystal lattices, and it is therefore much harder to
predict what ions will be observed in a mass spectrum. Therefore, much of our work
involving the LDMS analysis of inorganic pigments has been a learning process, in
discovering how the colorants form ions in this experiment. The analysis of numerous
inorganic pigments containing metals such as Fe, Pb, Cr, As, and Hg will be presented.
Unlike organic dyes, inorganic pigments possess metals having unique isotopic patterns,
making their appearance in a mass spectrum very obvious. Generally, these patterns
allow for an easy, positive identification of a compound. However, there are times when
the patterns become quite complex, requiring skill and knowledge of how the patterns
change with the addition of different elements. The following is an explanation of how

isotopic pattern analysis is performed.

24

575

 

Relative Intensity

 

 

 

 

sis 600

 

574
b) 574

576

 

 

Relative Intensity

   

 

 

 

 

Relative Intensity

 

 

I
100 200 300 m/z 400 500 600

Figure 1.4: a) Positive ion LD mass spectrum of copper phthalocyanine (Aldrich) on
paper; b) Negative ion LD mass spectrum of copper pthalocyanine on

paper.

25

Isotopic Profiling

Certain elements exist in nature in more than one isotopic form. Isotopes are
stable atoms having the same number of protons, but differing in the number of neutrons
present in the nucleus. Isotopes occur naturally in different abundances. The most
abundant is designated as the “A” form. The atomic weights of the elements listed on the
Periodic Table are the average masses. The average mass is a weighted sum of all of the
isotopes of a specific element. In a negative ion mass spectrum of chlorine, a single peak
at m/z 35 representing Cl' ions will not be observed. Rather, a pattern resembling that of
Figure 1.5a will be present. Figure 1.5a represents the theoretical isotopic distribution for
C1. C135 occurs much more commonly in nature than Cl37 (“A+2”). The ratio is roughly
3:1. This pattern is unique only to Cl, and therefore may be used to confirm the presence
of an ion containing C1 in a mass spectrum. Considering that the relative abundances for
most isotopes are relatively constant, the isotopic pattern for samples containing more

than one element having distinct isotopes can be predicted.

 

 

 

 

 

 

 

a b
) 35 ) 70
100 100
72
50 50
37
74
1 f 1 I
HassiC’harge Hasleharge

Figure 1.5: Theoretical isotopic pattern for a) Cl; b) C12.

26

What does the isotopic pattern for C12 look like? Consider a container containing
hundreds of chlorine atoms. The exact number of atoms present in the container is of no
importance, since it is already known that the relative amounts of the atoms are in a 3:1

ratio. Thus, the probability of choosing a C13 5

from the container will always be 66.67%
and 33.33% for C137. Table 1.5 lists all of the possible combinations in which the Cl
atoms could be chosen, with their corresponding probabilities. The peak appearing at
m/z 72 has two combinations with equal probabilities. These values are summed,
yielding a final ratio of 9:6:1 for the peaks at m/z 70, 72, and 74, respectively. This
corresponds with Figure 1.5b, representing the theoretical isotopic distribution for two Cl
atoms. While this calculation was relatively simple and not very time consuming, the
situation becomes much more complicated when compounds containing multiple

elements with distinctive isotopic patterns are analyzed, as in the case of artists’

pigments.

 

 

 

 

 

 

C11 C12 Mvvsicglfitar Probability
35 35 70 3 x 3 = 9
35 37 72 1 x 3 = 3
37 35 72 3 x 1 = 3
37 37 74 1 x 1 = 1

 

 

 

 

 

Table 1.5: Manual calculation for the isotopic distribution of C12.

27

References

l) Ember LR. Chemistry and art: helping to conserve the nation’s treasure. Chemical and
Engineering News 2001; July 31: 51-59.

2) Howarth E. Crash course in art. New York: Barnes & Noble, 1994.

3) Brown KL, Clark RJH. Analysis of pigmentary materials on the Vinland Map and
Tartar Relation by Raman microprobe spectroscopy. Anal Chem 2002;74:3658-3661.

4) McCrone WC. The Vinland Map. Anal Chem 1988;60:1009—1018.

5) Plesters J. Ultramarine blue, natural and artificial. In: Roy A, editor. Artists’
pigments: a handbook of their history and characteristics, vol. 2. New York: Oxford
University Press, 1993;37-65.

6) http://www.amolf.nl/research/biomacromolecular mass spectrometry/molart/molart.html

7) Berrie BH. Prussian blue. In: Fitzhugh EW, editor. Artists' pigments: a handbook of
their history and characteristics, vol. 3. New York: Oxford University Press,l997;
191-217.

8) Skoog DA, Holler FJ, Nieman TA. Principles of instrumental analysis, 5‘h ed.. USA:
Brooks/Cole, 1997.

9) Zollinger H. Color chemistry: syntheses, properties and applications of organic
pigments, 2nd ed. USA: John Wiley & Sons, 1991.

10) Schenk GH. Inorganic and organic charge transfer spectra in solution and their
analytical implications. Record of Chemical Progress 1967;28:135—148.

11) Giessman U, Rollgen FW. Electrodynamic effects in field desorption mass-
spectrometry. Int J Mass Spectrom Ion Phys 1981;38:267—79.

12) Sakayanagi M, Komuro J, Konda Y, Watanabe K, Harigaya Y. Analysis of ballpoint
pen inks by field desorption mass spectrometry. J Forensic Sci 1999;44:1204-1214.

l3) Lyon PA, editor. Desorption mass spectrometry — are SIMS and FAB the same?
Washington, DC: American Chemical Society, 1985.

14) Monaghan JJ, Barber M, Bordoli RS, Sedgewick RD, Tyler AN. Fast atom
bombardment mass spectra of involatile sulphonated azo dyestuffs. Org Mass
Spectrom 1982;17:569-74.

15) Scheifers SM, Verma S, Cooks RG. Characterization of organic dyes by secondary
ion mass spectrometry. Anal Chem 1983;55:2260-5.

28

16) Detter-Hoskin LD, Busch KL. SIMS: secondary ion mass spectrometry. In: Conners
TE, Banerjee, editors. Surface analysis of paper. New York: CRC Press, 1995;206-

34.

17) Pachuta SJ, Staral J S. Nondestructive analysis of colorants on paper by time-of-flight
secondary ion mass spectrometry. Anal Chem 1994;66:276-84.

18) Kara M, Bachman D, Hillenkamp F. Influence of the wavelength in high-irradiance
ultraviolet-laser desorption mass-spectrometry of organic molecules. Anal Chem
1985;57:2935-9.

l9) Denoyer E, Van Grieken R, Adams F, Natusch DFS. Laser microprobe mass
Spectrometry 1: Basic principles and performance characteristics. Anal Chem
1982;54:26—41.

20) N g L-K, Lafontaine P, Brazeau L. Ballpoint pen inks: characterization by positive
and negative ion-electrospray ionization mass spectrometry for the forensic
examination of writing inks. J Forensic Sci 2002;47:1-10.

21) McCloskey JA, editor. Methods in enzymology, vol. 193. New York: Academic
Press, 1990.

22) Lee ED, Muck W,Henion JD. Determination of sulfonated azo dyes by liquid
Chromatography/atmospheric pressure ionization mass spectrometry. Anal Chem

1987;59:2647-52.
23) Vastola FJ, Mumma RO, Pirone AJ. Organic Mass Spectrometry 1970;3:101.

24) Conzemius RJ, Capellen JM. A review of the applications to solids of the laser ion
Source in mass spectrometry. Int J Mass Spectrom Ion Physics 1980;34:197-271.

25) Posthumus MA, Kistmaker PG, Meuzelaar HLC. Laser desorption mass spectrometry
0f polar nonvolatile bio-organic molecules. Anal Chem 1978;50:985-991.

26) Stoll R, Rollgen FW. Laser desorption mass spectrometry of thermally labile
cOmpounds using a continuous wave CO2 laser. Organic Mass Spectrometry
1979;14:642-645.

27) Cotter RJ, Yugey AL. Thermally produced ions in desorption mass spectrometry.
Anal Chem 1981;53:1306-1307.

28) Benzazouz M, Hakim B, Debrum JL, Strivay D, Weber G. Study of the mechanism

of direct laser desorption/ionization for some small organic molecules (m < 400
Daltons). Rapid Commun Mass Spectrom 1999;13:2302-2304.

29

29) Hillenkamp F. Laser desorption techniques of nonvolatile organic substances. Int J
Mass Spectrom Ion Physics. 1982;45:305-313.

30) Cotter RJ, Tabet J-C. Laser desorption ms for nonvolatile organic molecules.
American Laboratory 1984;16:86-99.

31) Tabet J-C, Cotter RJ. Laser desorption time-of-flight mass spectrometry of high mass
molecules. Anal Chem 194;56: 1662-1667.

32) van Breeman RB, Snow M, Cotter RJ. Time-resolved laser desorption mass

spectrometry. l. Desorption of preformed ions. Int J Mass Spectrom Ion Physics
1983;49:35-50.

33) Watson JT. Introduction to mass spectrometry. Philadelphia: Lippincott-Raven,
1997.

34) Lias S, Bartmess JE, Liebman IF, Holmes JL, Levin RD, Mallard G. Gas-phase ion
and neutral thermochemistry. J Physical and Chemical Reference Data
1988;17:649,794-95.

35) Grim DM, Siegel J, Allison .1. Evaluation of desorption/ionization mass
spectrometric methods in the forensic applications of the analysis of inks on paper. J
Forensic Sci 2001 :46: 141 1-20.

36) Grim DM, Allison J. Identification of colorants used in watercolor and oil paintings

by UV laser desorption mass spectrometry. International J Mass Spectrometry
2003;222:85-99.

30

Chapter Two: Water Colors

Introduction

Art conservators devote much time and effort to preserving our artistic heritage.
Museums frequently lack Open windows, to keep out the sun’s damaging UV-rays. Art
objects may only be displayed for limited time periods, under minimal lighting (1).
Typical air pollution concentrations in museums have also been determined (2-6).
Preventative measures have been employed in museums, which include the use of
activated carbon air-filtration or alternative absorption or washing systems, use of
reduced ventilation rates (to hinder the circulation of harmful gases), enclosure of art
objects within display cases, and protection behind glass or varnish layers (7-9). The job
of art conservators is made much more difficult by the fugitive and unstable nature of
artist’s colorants. Watercolor paints are infamous for fading and/or shifting hue, as a
response to environmental conditions (ie. light, pollutant gases) (10,11). In particular,
azo dyes have been reported to be extremely susceptible to light (12,13). Consequently,
several kinetic studies have been performed to characterize the degradation process
(14,15). The effects of air pollutants which have been detected in museums, specifically
03 (16-23), N02 (24), HN03 (25,26), peroxyacetyl nitate (PAN) (27), and carbonyls such
as formaldehyde (28), on organic colorants have been well researched. The Cass
research group, at the California Institute of Technology, has performed extensive studies
in this area over the past decade. A summary of their work follows.

Ozone is a highly oxidizing air pollutant present in relatively low concentrations,
however in regions of high concentration of photochemical smog, the concentration can

reach between 0.3 and 0.4 ppm (18). In their experiments, modern watercolors (ie.

31

 

triphenylmethanes, acridones, and alizarins) and natural organic colorants (i.e. indigo and
curcumin) were subjected to ozone exposure, equivalent to the accumulated dose during
less than ten years in a typical air conditioned museum. They concluded that the ozone—
induced fading of natural colorants typically proceeds via destruction of the
chromophore, yielding colorless reaction products and hence a faded sample of the same
hue (16,18,20,22).

Nitrogen oxides are formed in the atmosphere from the nitric oxide emissions of
fuel combustion sources. These include N02, HN03 and PAN. Studies completed at the
California Institute of Technology indicated that the nitrogen oxides were not as
damaging as ozone (required longer exposure times to induce fading or color shifts) (24-
28), however they noted that specific chemical classes of dyes were more susceptible to
particular gases than others. For instance, lakes are more susceptible to N02, whereas
triphenylmethane dyes were more susceptible to HNO3 than to O3 (24). Whitmore and
Cass have concluded that unsaturated organic compounds are particularly vulnerable to
oxidation reactions, resulting in the discoloration of the colorant. They also noted that
N02 can result in the addition of nitro (-NO2) or nitroso (-ONO2) groups to organic
molecules, which then tend to yellow due to the N02 chromophore (24).

The Cass group employed a Diano Match Scan II spectrophotometer to measure
the diffuse reflectance spectra of colorants before and after exposure to different pollutant
gases. From the spectral reflectance data, tristirnulus values (X, Y, Z) and color
differences (AE, from the CIE 1976 L*a*b formula) were calculated for CIE Illuminated
C (24,29). Munsell notatations, which break down the color change into changes in the

visual perception of hue, value (lightness), and chroma (saturation), were subsequently

32

computed from these tristimulus values. AE gives the magnitude of the color change and
provides a basis for comparing the effect of gaseous pollutant exposure on the colorant
systems. Consequently, the results of their experiments are reported in the form of a
measure of the color change, as opposed to actual chemical information describing the
degradation mechanisms. One study is noted (exposure to atmospheric nitric acid) in
which, Grossjean, Salmon, and Cass employed chemical ionization mass spectrometry
(CIMS) to identify some (but not all) of the degradation products (25). The following
reasons were provided to explain why not all of the products were detected: 1) formation
of volatile products that were no longer present on the Teflon filter at the completion of
nitric acid exposure experiment, 2) products formed in low yields of <1%, since the mass
spectrorneter’s data acquisition systems did not record mass fragments whose abundances
were less than 1% of the base peak, and 3) products of very low vapor pressure, of which
no detectable amount could be introduced into the reaction chamber of the instrument’s
ion source even at high probe temperature. From this study, Grossjean et al. were able to
propose degradation pathways for alizarin, acridone, quinacridone, and indigo (25).

The Cass research group has touched upon the potential of mass spectrometry for
the characterization of dye degradation mechanisms. Here, laser desorption mass
spectrometry is being explored as a potential analytical technique for the detection of
organic watercolor dye degradation products. Potential obstacles are acknowledged,
including the formation of small, volatile, non-absorbing degradation products, which are
non-ideal analytes for the LDMS experiment. In a previous project, briefly discussed in
the next section, we were able to effectively characterize the degradation process for

methyl violet, a common organic toner used in modern ink formulations. Demethylated

33

degradation products were easily detected (30-32). Preliminary experiments were
designed and performed based on the results of the LDMS analysis of methyl violet.

The goals of the current project were to I) obtain positive and negative ion LD
mass spectra of commonly used watercolor paints; 2) identify the colorants in the mass
spectra; 3) artificially age the samples using a variety of light sources and expose the
colorants to different environmental pollutants (03, N03, etc.); 4) identify and quantify
degradation/reaction products; 5) determine if there is a relation between the age of the
paint and the extent of degradation/reaction products present in the mass spectra.
Preliminary data suggested that goals 1 and 2 were obtainable. Many of the dyes found
in the watercolor paints selected produced very intense, well-resolved peaks
representative of the intact dye molecule, indicating that the dye molecules are readily
desorbed and ionized when irradiated with a 337 nm laser. Additionally, specific dye
information was frequently provided on the labels of the watercolor tubes, making the
identification process relatively easy. However, a lack of dye information on the
watercolor tubes was not a deterrent. Identification of unknown samples was completed
using mass spectral interpretation and the information readily available in the literature.
Problems were encountered when trying to “age” and react the watercolors.

A box of Niji watercolors (Yasutomo & Co., San Francisco, CA) was purchased
containing the colorants labeled carmine, vermilion, lemon yellow, chrome green,
Viridian, cobalt blue, Prussian blue, yellow ochre, burnt sienna, burnt umber, and ivory
black. The names of the colorants drew interest, since several of them are typically
associated with inorganic compounds. However, their LD mass spectra prove different.

Positive and negative ion LD mass spectra were obtained for all of the samples, dissolved

34

in water and applied to paper. Following long exposure to various light sources,
including a UV-lamp, a Hg-arc lamp, and natural sun light, only a few of the colors had
faded to a noticeable extent.

Additional artist grade pigments were obtained (Cadmium Red Hue 095 and
Cadmium Orange Hue, Winsor & Newton), which listed the dye compositions on their
labels. Mass spectra of these colorants confirmed the information provided on the label,
however these paints also failed to change color following long exposure to light.
Consulting the literature, it was found that the structures of contemporary organic
watercolor dyes have been modified, significantly improving their resistance to light and
environmental pollutants. While this is wonderful for contemporary watercolor artists,
artists 100 years ago did not have access to the new and improved watercolors.
Consequently, their masterpieces were created with colorants known to disappear over
time, if not protected properly. The challenge became to find a supply of the older “bad”
watercolors, since they are no longer manufactured. The search was unsuccessful,
therefore we resorted to choosing the least stable pigments available based on ASTM
(American Standards for Testing Materials) light-fastness ratings. A detailed analysis of
these colorants follows.

Watercolor paints are composed of one or more organic dyes, which are known to
be fugitive (fade easily on exposure to light) and reactive with gases in their environment.
Lightfastness is a term generally used to describe the ability of a colorant to resist fading
on exposure to light. However, lightfastness also encompasses a colorant’s
impermanence to additional damaging environmental factors, including weak acids,

alkalis and impurities in the atmosphere. The ASTM has established a well-defined

35

 

testing procedure to which each colorant is subjected (33-35). They have developed a

scale (ASTM I-V), rating each colorant according to its lightfastness. A summary of the

AS TM scale is provided in Table 2.1.

 

 

 

 

 

 

 

Rating Meaning
ASTM I Excellent lightfastness.
ASTM 11 Very good lightfastness.
A STM 111 Not sufficiently lightfast to be used in paints conforming to the
specification. Such colors can fade rather badly, particularly the tints.
L ASTM IV Pigments falling into this category will fade rapidly.
L ASTM V Pigments will bleach very quickly.

 

Table 2.1. A summary of the ASTM lightfastness scale.

Colorants have been named and categorized based on the Color Index.
Throughout this chapter, the common and color index names will be given for each
COlorant discussed, and the names may be used interchangeably. The chemical class of
each colorant will also be provided. Generally, the common name gives Some kind of

indication of the chemical class, but not always.

Experimental

A box of Niji watercolors (Yasutomo & CO., San Francisco, CA) was purchased
Containing colorants labeled carmine, vermilion, lemon yellow, chrome green, Viridian,
CObalt blue, Prussian blue, yellow ochre, burnt sienna, burnt umber, and ivory black.
Additional colorants listed in Table 2.2 were purchased from a local artist’s supply store.
All of the samples were dissolved in water and applied to paper (Hammermill Fore MP).

After the samples were dry, they were subjected to positive and negative ion LDMS

analysis, as described in Chapter 1.

36

 

UV-accelerating aging studies were performed on dye-on-paper samples using a
UV—lamp (254 nm, 760 microwatts/cmz, UVP Inc., San Gabriel, CA model UVGL-58),
as well as a Hg-arc lamp (Oriel, Stratford, CT), and analyzed at various intervals.
Controlled ozone exposure experiments were performed in a reaction chamber (0.308 L)
at the Environmental Engineering Department (Michigan State University, East Lansing,
MI). Samples were exposed to 375 ppm 03, at a flow rate of 2.57 mL/min, for 12.5
minutes. Exposure to concentrated HN03 vapors, was accomplished by suspending the

sample (dye—on-paper) over a HNO3 bath in a covered beaker. Exposure times varied.

 

 

 

 

 

 

 

 

 

 

 

Colorant (Brand) Dye Composition Vehicle
Cadmium Orange Hue 090 Diarylide yellow (PY 83) Gum
(Winsor & Newton) Perinone orange (PO43) Arabic/dextrin
Cadmium Red Hue 095 Perylene (PR 149) Gum
(Winsor & Newton) Pyrrole (PR 255) Arabic/dextrin
Purple Lake 544 Cu-phthalocyanine (PB 15) Gum
(Winsor & Newton) Rhodamine/PMTA (PV 2) Arabic/dextrin
Mauve 398 Quinacridone (PR 122) ' Gum
(Winsor & Newton) Carbazole dioxazine (PV 23) Arabic/dextrin
Alizarin Crimson Hue 003 Quinaccridene-pyrrolidone, Gum
(Winsor & Newton) Quinacridone (PR 206) Arabic/dextrin
§?::0¥Zd‘§; 323]) 1,2 dihydroxyanthraquinone lake (PR 83) Arabgyé‘e‘xtrin
Hooker’s Green Dark 312 Chlorinated cu-phthalocyanine (PG7) Gum
(Winsor & Newton) Quinacridone (P0 49) Arabic/dextrin
Cu-phthalocyanine (PB 15)
. Amorphous carbon (PBk 7)
(Wiri:c:ilg8:)131:\zvton) Complex silicate of Na & A1 with 8 (PB 29) Arabilldzxtrin
Cu-phthalocyanine (PB 15)
Opera (Ho‘bem Am“ 5 PR 122, BV 10 (not provided)
Watercolor)

 

 

 

Table 2.2. Artist-grade watercolors subjected to LDMS analysis.

37

 

Summary of Methyl Violet Degradation Study

As mentioned previously, methyl violet and its degradation mechanism was
studied extensively, as part of a separate ink dating project (30-32). Important concepts
were learned in regards to the ability to detect organic colorants off of paper (a
nontraditional surface for an LDMS experiment). Additionally, a method for artificially
accelerating the age of organic colorants using UV light was developed. This project
served as a “springboard” to a much more extensive project involving the LDMS analysis
of several organic watercolor dyes. The results of the methyl violet study relevant to this
project are summarized in this section. Furthermore, it will become necessary to
recognize the difference between the mass spectra of new and degraded methyl violet, in
order to appreciate the LDMS analysis of an illuminated manuscript discussed in a later
chapter.

Methyl violet is known to be extremely light fugitive, and fades very easily when
exposed to natural and artificial light sources. The structure of methyl violet is shown in
Figure 2.1. It is a pentamethylated triphenyl methane cationic dye. The structure, as
shown in Figure 2.1, has a molecular weight of 358 Da. As was discussed in Chapter 1,
dyes are typically sold as impure mixtures and in the case of methyl violet, the
hexamethylated homolog (crystal violet) is the most abundant component in the mixture.
As an ink containing methyl violet ages, the dye structure degrades via an oxidative

demethylation process (35), which is evident in mass spectra of aged ink (Figure 2.2).

38

mlz Structure
H30

\N+—CH3 372 . C+(Me)6
358 Ct Me)5H
H3C 344 C+ Me 4H2

(
\ ( )
"ac/N O 330 C+(Me)3H3
( )
(

 

 

 

 

 

316 0+ Me
C+(Me)5H 302 0+ Me)1H5

 

 

288 c+HO

 

 

 

 

Figure 2.1. The structure of methyl violet and designations for its degradation products.

Figure 2.2 shows the positive ion LD mass spectra of a preliminary accelerated
aging study performed on black Bic ballpoint pen ink-on-paper. The ink was applied
uniformly to a piece of 4 in2 paper. An initial positive ion LD mass spectrum was
obtained. Approximately one third of the sample was covered and the entire sample was
irradiated using a UV lamp (760 microwatts/cmz) , placed directly on top of the sample
for 12 hours. Another third of the sample was masked off and the sample was irradiated
for another 12 hours. This experiment demonstrated that significant fading was noted
following 12 or more hours of intense UV irradiation. Also, it showed that by using
masks, an array of UV aged ink on a single piece of paper could be easily generated,

introduced as a single sample into the mass spectrometer, and spectra for each region

39

obtained. We acknowledge that there is currently no accepted irradiation/age correlation

for this experiment.

 

 

 

 

Initial 372
284 358
a L _ _
12 Hrs 330 344358372
31 6
302
288
344
24 Hrs 330
358
316 372
302
288
m_ _ LL_ L A _.- 4‘ AL A _4_
250 300 350 400 j 450
mlz
Figure 2.2. Preliminary UV—accelerated aging study: Bic ballpoint pen ink on
paper.

The preliminary UV accelerated aging data are presented in Figure 2.2. Positive
ion LD mass spectra are shown of methyl violet dye as originally deposited by the
ballpoint pen onto paper, and the same portion of the mass spectrum is shown following
12 and 24 hours of UV irradiation. After irradiating the sample for 12 hours, degradation
Products (summarized in Figure 2.1) are clearly present — a series of peaks separated by
14 amu. Each methyl group (15 arnu) is replaced with a H atom (l amu), accounting for

the 14 mass unit difference. The m/z 372 peak remains the most intense peak. The

40

spectrum shows that six new compounds, lower—mass forms of methyl violet, are now in
abundance. Irradiating the sample for another 12 hours causes the initial base peak to
further decrease in intensity, such that the degradation product peaks dominate.

Andrasko noted similar results using high performance liquid chromatography (HPLC) to
study ballpoint pen inks stored under various light conditions (36).

When comparing these results to naturally aged samples (Figure 2.3b), the same
degradation products were generated. However, we realized that we had artificially aged
the samples past that which we had observed in naturally aged samples. Aginsky stated
the importance of ensuring that the accelerated aging method mimics natural aging as
closely as possible, meaning that the technique should not induce any chemical changes
that would not occur naturally (37). Consequently, the experimental design was adjusted
by elevating the UV lamp 6 cm above the sample, and by monitoring the irradiated
sample at much shorter time intervals (every 15 minutes). This allowed for us to
visualize the dye degradation process more clearly, as well as to mimic the rate of the
natural aging process more effectively. The developed method was applied to UV-
accelerated aging experiments performed on watercolors.

UV-accelerated aging mimics natural aging from a dye perspective, and can be
characterized by LDMS. Figure 2.3a represents a portion of the positive ion LD mass
spectrum of new black Bic ballpoint pen in on paper. New ink is characterized by a large
peak at mlz 372, representing the non-degraded dye molecule (C+Me6), with a very small
peak at m/z 358 (C+Me5H1). Figure 2.3b is the positive ion LD mass spectrum of a 38-
month old naturally aged Bic black ballpoint pen ink on printer paper. As an ink ages,

lower mass peaks appear in the spectrum representing the molecules that are referred to

41

as degradation products, accompanied by a decrease in the relative intensity of the
original m/z 372 base peak, representing the original intact dye molecule. The number
and amount of degradation products present is a function of the age of the ink, but if this
was ink on a questioned document, how could the age be determined from the spectrum?
As in other ink dating methods, there must either be spectra from naturally aged samples
for comparison purposes, or there must be a calibrated methods for accelerating the aging
of a new sample of similar ink that can be use to create a sample that yields the same
spectrum. Irradiating Bic black ballpoint pen ink-on-paper, for 6.25 hours with UV light,
produces a very similar mass spectrum (Figure 2.3c) as that of the naturally aged 38-
month old document (aged in the dark). Based on this data alone, a calibration for the
UV method can be estimated. Irradiation for 6.25 hours produces the same extent of
degradation as what occurs naturally over a period of 38 months. Thus, every hour of
UV irradiation accelerates the aging by approximately 182 days. If a different sample
was in question, the ink could be irradiated until the same extent of degradation was
observed, and from the irradiation time required, the corresponding natural age could be
calculated. This approach, along with other insights into the variables that influence the

rates of dye degradation, were studied in this work.

42

 

 

 

 

Relative Intensity

 

 

 

372
a
358
b 372
358
344
330
C 372
358
344
330
l I i ‘
285 300 31 5 330 345 360 375

mlz

Figure 2.3. A portion of the positive ion LD mass spectrum of Bic black ballpoint
pen ink-on-paper: a) ink from a new document; b) a 38-month old
controlled, naturally aged document; c) ink from a document irradiated
for 6.25 hours with UV light

43

Methyl violet can be efficiently degraded on paper with UV irradiation. Ink
samples on paper were subjected to UV radiation, followed by LDMS analysis, for
periods of up to 8 hours. In these time periods, more than 50% of the dye molecules can
be converted into degradation products. Figure 2.4 is a plot of the normalized relative
intensities for each of the three mass spectral peaks representing the dye or degradation
product as a function of time for the UV accelerated aging study of Bic black ballpoint
pen ink-on-paper. The relative intensity is a measure of the concentration of a particular
species in comparison to the other components detected in the sample mixture. In Figure
2.4, the relative abundance of the intact dye molecule (m/z 372) decreases as the relative
abundances of the degradation products (m/z 358, 344) increase with irradiation time,
which is an indication of the dye degrading. How can the data be used to develop a
determination of an ink’s age? There are many options. For instance, the age correlation
may involve the relative intensity of the mlz 372 peak, revealing the amount of dye
remaining in its original form Another possibility would be to take a ratio of two peaks
that are always present, and seem to experience the most significant changes as the ink
ages. Ideally, a function would be identified, in which a single value could be computed
that incorporates information on all of the degradation products present. One way to
accomplish this is to compute the average molecular weight for the dye at each time
interval. Figure 2.5 shows the plot of the average molecular weight versus the time the
document was irradiated in the UV accelerated aging study of Bic black ballpoint pen ink
on paper. The average molecular weight (MWavg) was calculated by multiplying the
normalized intensity of each peak by the nominal mass of that peak, summing all of the

products, and dividing by the sum of the relative intensities of all of the peaks:

MW__m = [RIIm/z 372) x 3721+ [RI(m/z 358) x 3581 + lRI(m/z 344) x 3441+
RI(m/z 372) + RI(m/z 358) + Rl(m/z 344) +

This value was computed for each of the five spectra acquired per sample, and the five
molecular weights were averaged. Scatter in the data is noted, however a distinct trend is
established in which an increase in the degradation of the dye, i.e., a decrease in the
average molecular weight, occurs over time. The use of MWavg lends some physical
meaning to the experiment. For instance, when the dye is initially deposited on paper, its
average molecular weight would be nearly 372 Daltons, since it is essentially only crystal
violet. If a sample is analyzed that shows an average molecular weight of 364 Daltons, it
is understood that the degradation products are dominating the spectrum, indicating that
the dye has degraded. The lower MWavg limit for this experiment is 288 Daltons, which
would occur if all of the C+Me6 were converted to C+H6. We have not yet found any
naturally aged samples in which degradation has been this extensive. Figure 2.5 shows
that the average molecular weight falls to 361 Daltons following 450 minutes of UV

irradiation.

45

 

1.2

mlz 372 (R’=o.9289)
\ ll. ,_,.__ L_ L__,_L _ , ____- , ., ________.i,,

1+
.990 ,
o
o. .

.0
co

 

Relative Intensity
l
l
l
l
l
l
l
l
l
l

 

 

 

 
 

 

 

 

0.4 <——-————— —
. A
0.2
A
mlz344(x2) (82:03287)
0 .. a —. s e . T . . . . r v .
0 50 100 150 200 250 300 350 400 450 500

UV Irradiation 11m (minutes)
Figure 2.4. UV accelerated aging study data for new Bic black ballpoint pen ink on

printer paper: a plot of the relative intensity of the m/z 372, 358, and
344 (x2) peaks versus irradiation time

Similar dye degradations occur naturally. In order to relate the UV accelerated
aging curve in Figure 2.5 to aging which occurs naturally, a natural ink aging curve
(Figure 2.6) was constructed using a set of controlled ink library samples from Speckin
Forensic Laboratories, Okemos, MI. The samples were written with the same pen, on the
same paper, and stored under the same conditions, in the dark. Again, from the LDMS
spectra, the average molecular weights were computed. According to the UV accelerated
aging curve prepared under the outlined experimental conditions, 500 minutes (8.3 hrs) of
UV irradiation degraded methyl violet to an average molecular weight of approximately
360 Daltons. Referring to the straight line fit of the controlled natural aging data (Figure

2.6), an average molecular weight of 360 Daltons is equivalent to 52 months of natural

46

aging for this particular ink and paper. By combining this information in Figures 2.5 and
2.6, it appears that 500 minutes of UV irradiation (at the conditions used here) would age
a document by approximately 52 months (~1560 days). Therefore, in this particular
study, for every hour of UV irradiation, the document was aged roughly 187 days. This
compares favorably with the initial estimate of 182 days per hour determined from the
initial data shown in Figure 2.3.

A comment should be made concerning the linearity of the two best fit lines in
Figures 2.5 and 2.6. The linear least squares fit to the data in Figure 2.5 has an R2 value
of 0.8845. The R2 value is a measure of the strength of the linear relationship. The
closer the R2 value is to unity, the stronger the relation. The linear regression line or the
data in Figure 2.6 has an R2 value of 0.6331, indicating that the data generated during the
UV accelerated aging study correlated better to a straight line fit than the data
accumulated from the controlled natural aging study. This is not unexpected considering
the time frame of the experiments. The UV accelerated aging study was completed in the
matter of hours, whereas the controlled natural aging study occurred over ten years.
During the UV accelerated aging study, variables affecting the aging process, such as
temperature, lighting, and humidity fluctuations, were minimal, resulting in a more

constant rate of aging.

47

 

 

 

 

 

 

 

 

 

 

  

 

 

 

Average Molecular Weight (Da)

 

 

 

 

 

 

 

 

 

sea-L———————— #

361

359

357 T . . . . . . . . ,
0 so 100 150 200 250 300 350 400 450 500

UV irradiation Time (minutes)

Figure 2.5: UV accelerated aging study data for new Bic black ballpoint pen ink on
printer paper: a plot of the average molecular weight of the dye,
methyl violet, versus minutes of UV irradiation (32).

 

380 ——k—-—--—,-——------— -—— ,___ ~——# —~-——- --
«r i 82:0.6331
375 .._...._ _ _, -fif , *_ _‘__._____. “E

 

 

 

 

(a)

‘1

o
I

 

 

 

 

 

iii
l
l
l
l

 

 

 

 

 

 

83
Oi

 

 

Average Molecqu Weight (De)
8
O

 

 

a
l

3
m
i—OJA

 

 

 

 

340 T r i l l
0 20 4O 60 80 100 120
Age (months)
Figure 2.6: Controlled, natural aging study data for Bic black ballpoint pen ink
on printer paper: a plot of the average molecular weight of the dye,
methyl violet, versus the age of the document (32).

48

Identification of Watercolor Dyes (Unknown Dye Composition)

Artists’ watercolor paints are historically known to fade, which is an indication of
dyes degrading. Consequently, we had hoped to be able to apply the experimental
methods developed in the analysis of methyl violet in ink, to the analysis of organic dyes
in paint. A box of Niji watercolors (Yasutomo & Co., San Francisco, CA) was purchased
containing the colorants labeled white, carmine, vermilion, lemon yellow, chrome green,
Viridian, cobalt blue, Prussian blue, yellow ochre, burnt sienna, burnt umber, and ivory
black. The laser desorption mass spectra of a select number of these colorants (lemon
yellow, Viridian, chrome green, Prussian blue, carmine, and vermilion) will be shown and
discussed. The dye compositions of these colorants were unknown. The results of
accelerated aging and environmental studies performed on each colorant will also be

discussed.

Cobalt Blue

The positive and negative ion mass spectra of cobalt blue are shown in Figure 2.7.
The peak at m/z 575 in the positive ion mass spectrum represents the molecular ion, M+,
for copper phthalocyanine (pigment blue 15 — PB 15). Valued by artists for its bright
blue-green color, copper phthalocyanine has a relatively high ASTM rating of II.
Shankai et al. have previously reported the use of MALDI to analyze metal
phthalocyanines, and do show that M+ ions can be formed (38). Furthermore, Conneely
et al. employed MALDI MS and E81 M8 to analyze sulfonated copper phthalocyanine
dyes in negative ion mode (39). The isotopic distribution is consistent with the formula

C32N3H16Cu, and is dominated by the presence of a copper atom, which has abundant

49

”Cu ill
ballpolr
and ani
rcspons
of ii hit

rim 5'

cop-re

in F19

63 Cu (100%) and 65 Cu (44.57%) isotopes. In our experience with organic dyes used in
ballpoint pen inks, cationic dyes such as methyl violet are detected in positive ion mode
and anionic dyes, such as solvent black 29, are detected in negative ion mode (30). A
response in both positive and negative ion modes indicates the presence of a neutral dye,
of which copper phthalocyanine is an example. It forms an M+ peak in positive ion mode
(m/z 575), and an [M-H]’ peak in negative ion mode (m/z 574).

Figure 2.8 shows positive and negative ion spectra of an aqueous solution of
copper phthalocyanine (Aldrich), applied to paper. The spectra confirm the assignments

in Figure 2.7.

50

 

575

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

mlz

Figure 2.7 : Cobalt Blue (Niji Watercolors): a) the partial positive ion LD mass
spectrum; b) the partial negative ion LD mass spectrum

51

a)
0 ~ 0
/ / \
5» N N
'5 / \ / \
8 C\ N
5 \r/ N 577
\ \
, Q ~ \Q
:1:
PB 15
ii 579
- _,/\AA~ E ,E J , - I A
550 560 570 580 590 600
mlz
574(M-H)'
b)
3‘
'2
3 576
.5
0
>
’U
.9.
G)
a:
1 iii:
550 “M ’560 5704 ' 580 4' 530 3' N600

 

575

 

 

 

 

 

 

 

 

 

 

 

 

 

 

a)
5*
'2
8
.5
G)
>
ti
a
E
at
520
I T” ' A i “ T ‘ i w i ”T
500 525 550 575 600
m/z 574
b)
574
Z?
5
% e-I
3" E. 76
g s
.3 ‘__
.5 134 g
g 577
:e
3 .-_- - A - E u.“
9 I
m 550 575 600
269 412
‘ |
100 200 300 m/z 400 500 600

Figure 2.8: Copper phthalocyanine (Aldrich): a) the partial positive ion LD mass
spectrum; b) the partial negative ion LD mass spectrum

52

Pnnshni

Vihfieir
PhihdiO.‘
masssp
intheri
Raina
respec1
pOxll l\
SUggcs
reach:
lilt‘ pr
15 km
beifil

mm

Prussian Blue

Historically, Prussian blue refers to the inorganic colorant ferric ferrocyanide.
While iron blues were used since the early 17008, until about .1970, copper
phthalocyanine blue is now often used in their place (40). The positive and negative ion
mass spectra of Prussian blue (Niji) are shown in Figure 2.9. Evidence of PB 15 appears
in the positive ion mass spectrum at m/z 575 (Figure 2.9a) and at m/z 574 in the negative
ion mass spectrum (Figure 2%). The cluster of peaks represent the M+ and [M-H]' ions,
respectively. The additional higher mass peaks in Figure 2.9 are noteworthy. In the
positive ion mass spectrum, the peak at m/z 609 is 34 mass units higher than mlz 575,
suggesting that a H atom has been replaced by a Cl atom. The same conclusion is
reached for the next cluster. Negative ion m/z values and isotopic distributions confirm
the presence of a mixture of copper phthalocyanine and copper chlorophthalocyanines. It
is known that such compounds can be modified in a number of ways. Sulfate groups can
be added to change the color (41). Additionally, chlorination of the compound will vary

the color (produce a greener shade), as well as increase the dye’s solubility (41).

53

   

 

575

 

 

 

 

 

 

 

 

 

 

 

 

 

3)
577
5‘
e
55.
=
fl
0
>
'U
£2
0
a:
609
J a -3 . Min- . - iii..-
550 575 600 mm 625 650 675
574

b)
5‘
'51
§
.5 576
Q
'5 610
i? 608’
a:

645
675
k 711
550“ 5185*“; 620 A ' 655 690 725
m/z

Figure 2.9: Prussian Blue (Niji Watercolors): a) the partial positive ion LD mass

spectrum; b) the partial negative ion LD mass spectrum

54

Lemon Yellow

If an artist, a century ago, had selected a lemon yellow colorant, he/she would
have chosen either a barium or strontium chromate pigment. Both Ba and Sr have unique
isotopic patterns, which do not appear in the positive or negative ion LD mass spectra of
this lemon yellow (Figure 2.10). Many more peaks are observed in the mass spectra,
when compared to the spectra of Cobalt blue and Prussian blue. This may indicate the
presence of multiple colorants in the paint mixture, or perhaps the presence of fragment
ions. Fortunately, there are clues present in both the positive (Figure 2.10a) and negative
ion (Figure 2.10b) mass spectra, which help to narrow down the large number of yellow
dye possibilities.

In mass spectrometry, there are a few general steps to follow when interpreting a
mass spectrum. For example, the peak with the largest m/z value is typically the
molecular ion (or protonated molecule), unless isotopic peaks are present. Thus, in the
positive ion mass spectrum (Figure 2.10a), the cluster of peaks around m/z 395 may
possibly be representing the intact dye molecule. The next step is to analyze the pattern,
to determine if it is consistent with any known isotopic patterns. The cluster of peaks at
m/z 395 is not completely resolved, however there appears to be three peaks present, each
separated by two mass units, in a 9:621 ratio. As was discussed in the Isotopic Pattern
Analysis section in Chapter One, this pattern is characteristic of a molecule containing
two Cl atoms. Additionally, there are numerous lower mass peak clusters which have
two peaks present, separated by two mass units, in a 3:1 ratio, indicating the presence Of

only one Cl atom. Therefore, the lower mass peaks potentially represent fragment ions.

55

What is known thus far about the yellow dye? If the dye has a molecular weight
of approximately 395 mass units, it must contain a few aromatic rings, in order to absorb
UV light. The dye molecule has two Cl atoms present (70 mass units). It is also noted
that the peaks at m/z 395 are not the most intense peaks in the mass spectrum.
Consequently, if the lower mass peaks are indeed fragments ions, the structure of this dye
molecule is not very stable (in contrast to copper phthalocyanine). Therefore, unstable
functional groups, such as azo linkages (N=N), may be present. At this point, it is
difficult to derive the exact dye structure from the information available in the mass
spectrum. However, all of these clues can be used to perform a more efficient literature
search for the dye’s structure.

In the positive ion mass spectrum (Figure 2.10a), the cluster of peaks at m/z 395
have been identified as representing pigment yellow 3 (PY 3). The structure for PY 3 is
shown in the inset of Figure 2.10b. It is commonly known as arylide yellow 10G, but
may also be found listed as Hansa Yellow Light. It has been assigned an ASTM rating of
II, and is included on the ASTM approved pigments for watercolors (42).

Like copper phthalocyanine, PY 3 is a neutral dye and has a molecular weight of
394 Da. However, a different trend is observed in the mass spectra, meaning that we do
not observe simply the M+ and [M-H]' ions in positive and negative ion modes,
respectively. Instead, there is an overlap of peaks representing the M+ (mlz 394) and
[M+H]+ (m/z 395) ions in positive ion mode. The presence of sodium and potassium
adduct ions at m/z 417 [M+Na]+ and 433 [M+K]+, lend support to these assignments.
Likewise, in the negative ion mass spectrum (Figure 2.10b), an overlap of peaks is

observed representing M' (mlz 394) and [M-H]' (mlz 393) ions. Returning to the positive

56

ion mass spectrum (Figure 2.10a), a number of the remaining peaks have been identified
as fragment ions, suggesting the presence of only one dye. The positive fragment ion
peak assignments are listed in Table 2.3. Only one fragmention (m/z 171) is present in
the negative ion mass spectrum. The peaks have a pattern consistent with the analyte
containing one Cl atom. At this point it is unclear whether it is related to the fragment
ion at m/z 170 in the positive ion mass spectrum. The remaining, unidentified peaks in
the positive ion mass spectrum may possibly be products of rearrangement reactions or
they may simply be due to impurities (such as reactants in the synthesis of PY 3) in the
mixture. Considering the structure of the dye, the appearance of several fragment ions is
not surprising. Unlike copper phthalocyanine, PY 3 is not composed primarily of fixed
aromatic rings. Instead, it has a number of single bonds, which could be easily broken.
Additionally, PY 3 is a different kind of molecule than methyl violet, and is expected to

be less stable.

57

Illlilll

 

.Xzzteu:— 4.5:-«.9:

\I‘Nrill

 

 

268

 

 

 

 

 

 

 

 

3)
(M+H)+
395
3‘
'Z’
3 397
5 127
Q)
E
3 270
E 173 »
198 335 l
170 349 .
154 312 37/ 377
240 L (M+Na)+
100 170 240 310 380 450
mlz
(M-H)‘ 393 _
b)
or
395
E‘
8
er N02
"' PY3
.5
a) N
> /
'a N/ CI
2
0)
9‘ H3C NH
1 1
7 o O
. .4: L .L a . - a a -. A
150 205 260 mu 315 370 425

Figure 2.10: Lemon Yellow (N iji Watercolors): a) the partial positive ion LD mass
spectrum; b) the partial negative ion LD mass spectrum

58

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Fragment Ion I mlz
r Cl \
HZN +
b 127
K J
( CI \
O NH
Y O 154
K J
Cl
r \ +
170
K Non
HN
r C' N.
+ HN
" O 268
J
[M— N02 — CH3 + H]H 335
[M- NOle" 349

 

Table 2.3: PY 3 positive fragment ions.

 

Viridian (Green)

Viridian is a green colored pigment, historically identified as a hydrated oxide of
chromium (Cr2O3-2H2O) (43). However, the positive and negative ion LD mass spectra
of this Viridian, shown in Figure 2.11, suggest the presence of organic colorants. Clusters
of peaks in the positive ion mass spectrum (Figure 2.11a) at m/z 395, 268, 154, etc. and at
mlz 393 and 171 in the negative ion mass spectrum (Figure 2.11b), confirm the presence
of PY 3 in the mixture. Green colorants may either be a pure green compound or a
mixture of blue and yellow colorants. Since there is a yellow dye present, there may be
evidence of a blue pigment somewhere in the mass spectra. Additional peaks are noted in
the positive ion mass spectrum at m/z 1129, 1092, and 1059, which did not appear in the
lemon yellow spectra. These peaks are separated by 34 mass units, suggesting that a
series of Cl atoms are being replaced by H atoms. The dye has been identified as
phthalocyanine green (pigment green 7 - PG 7) and the structure is shown in the inset of
Figure 2.11a. PG 7 has the same base structure as PB 15, however it is completely
chlorinated, resulting in the green color. Both e' donating and e' withdrawing groups can
be added to Cu-phthalocyanine to shift the color to either the red or the blue. The
structure is stable and fragment ions are not expected. The dye is known for its high
tinting strength, meaning that only a small quantity of the colorant is required to produce
a bright green shade, and has earned an ASTM rating ofI (42).

Peaks representative of deprotonated PG 7 molecules are noted in the negative ion
mass spectrum. There was no evidence of a blue dye in the paint mixture, suggesting that
PY 3 was included to alter the green shade of PG 7. However, it is important to note that

(CuxCly) ions were identified in negative ion mode, based on their isotopic patterns. An

60

aql

HO

aqueous solution of [Cu(H2O)4]2+ ions (blue) and [CuCl4]2' ions (yellow) are green (44).

However, this combination is not typically classified as an artist’s water colorant.

61

Relative Intensity

Relative Intensity

268

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

a) CI or Cl CI PG 7
Cl Cl M+
N 1129
/ / \
PY 3 Cl N N Cl
(M+H)+ N/ \CU/ \N
395
\F/ \N
Cl \ Cl 1133
270 \
397 N \
127 CI CI
Cl C
173 3 5 Cl I Cl W
J17 575 1093
.L__.J ..- - L'J‘J _ .f L _._a A L‘- ‘ .
100 330 560 mlz 790 1020 1250
FY 3 393 393
b) (M-H)°
395
395
C1. W 397
35 ‘— , J
375 383 391 ml 399 307 415
PG 7 1122
171 M'
(CuCl2)‘ (C112Cl3)'
135 233

J L .ALllVL-A_ILL r *m'4_. . u, - A :4 “L-
0 240 480 mlz 720 960 1200

Figure 2.11: Viridian (Niji Watercolors): a) the partial positive ion LD mass
spectrum; b) the negative ion LD mass spectrum

62

C1

Carmine (Red)

True carmine is associated with a lake pigment. As described earlier, a lake is a
colorant prepared by precipitating a transparent organic dye onto an inorganic support
(45). In the case of carmine lake (Chapter 3), the dye is carminic acid, and the inorganic
base is typically alumina (A1203). The positive and negative ion LD mass spectra of
carmine (N iji) are shown in Figure 2.12 and are not suggestive of real car.mine. Thus far,
the exact dye composition of this carmine (Niji) has not been determined. Fortunately,
the mass spectra provide interesting clues, which help to characterize the dyes.

In the positive ion mass spectrum (Figure 2.12a), there are peaks present in the
higher mass region (~ m/z 1129) indicative of PG 7. The assignment is confirmed with
the presence of complimentary peaks in the negative ion mass spectrum (Figure 2.12b).
The presence of PG 7 in the paint mixture was somewhat unexpected. The Wilcox Guide
to the Best Watercolor Paints is a comprehensive source which characterizes and rates a
large number of modern day watercolors produced by thirty different domestic and
foreign colorant manufacturers (42). The book is divided into chapters based on color
(i.e. red, yellow, blue, etc.). Each chapter is broken down into sections based on the
generic name of the colorant (i.e. carmine, vermilion, etc.). The dye compositions of the
watercolor paints were provided, when available. Although Niji Watercolors were not
included in the text, we were able to develop an idea of which colorants were more
commonly used in the paint formulations. In the specific case of carmine, only red dyes
were employed. Occasionally, pigment violet 19 (PV 19) was included in a paint mixture

to give the paint a more pinkish tone. Of the 28 different brands of carmine included in

63

the Wilcox Guide, not one was created using a green colorant, which is why the presence
of PG 7 was unexpected.

Based on a literature search, we were looking for evidence in the mass spectra of
any of the following pigment red colorants: PR 5, 9, 48:4, 83, 88, 112, 146, 170, 176,
178, and 264. The structures of a select number of these colorants are shown in Table 4,
which illustrate the different classes of these dyes. Several of the PR dyes listed have the
same basic structure, but differ in their substituents. Therefore, if we can determine how
certain classes of dyes “behave” in the LDMS experiment, meaning whether they form
molecular ions or fragment ions, we can establish patterns that will ultimately make the
mass spectral interpretation process much easier. Unfortunately, none of the peaks in
both positive and negative ion modes appear to have originated from any of the expected
red dyes. However, the mass spectra are still informative.

The isotopic patterns of the peaks in the positive ion mass spectrum (Figure
2.12a) are not consistent with the analyte containing Cl, whereas some of the peaks in the
negative ion mass spectrum are. In the positive ion mass spectrum, there are intense
peaks present at m/z 284 and 368. It is unclear whether or not these peaks are
representative of a colorant, or a component of either the colorant’s vehicle or the paper.
The peaks appear consistently in the spectra of carmine, however they have been noted
sporadically in the spectra of other N iji watercolors (ie. Prussian blue). As will be
discussed shortly, the peaks at m/z 284 and 368 become more apparent in the spectra of
faded colorants. This may suggest that they are originating from the paper. Positive and
negative ion mass spectra were obtained of the paper by itself (spectra not shown) (46),

and do not contain peaks at m/z 284 or 368. However, a peak at m/z 284 was noted in the

positive ion mass spectrum of a different paper sample from a previous study (30).
Consequently, the results of the paper study suggest that the peaks at m/z 284 and 368
may be originating from a vehicle component common to other Niji water colorants.
Additionally, the peaks at mlz 284 and 368 appear to be the beginning and the end of a
series of peaks separated by 28 mass units. The mass difference may be the result of the

loss of consecutive C2H4 groups.

65

hawtteap: 9a: :3»—

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

368 a)
284
PG7
5, 11129
'12
§
.5
Q
'3 340
.5.
SE
232 312
1093
212
(J 1059 l
150 ' 362 574 mlz 786 'L 998 ' 1210
205
b)
79
g; 220
.3
.5
3
a C1 207
r. 35 185
M
26 42 72 96 222
84 141
265 294
l
0 60 120 mh 180 240 300

Figure 2.12: Carmine (N iji Watercolors): a) the partial positive ion LD mass
spectrum; b) the negative ion LD mass spectrum

66

 

Structure Name (MW)

 

PR 88
(434)

 

PR 112
(484)

 

PR 176
(573)

 

 

PR 264
(440)

 

 

 

Table 2.4: Structures of Pigment Red colorants typically found in carmine watercolors.

67

 

The clusters of peaks at m/z 205 and 220 in the negative ion mass spectrum
(Figure 2.12b), have patterns consistent with a molecule containing one Cl atom, whereas
the peaks at m/z 294 are characteristic of a molecule containing two C] atoms. The peak
at mlz 185 may be related to the ions represented at m/z 220. The peak is 35 mass units
less than m/z 220, indicating the loss of a Cl atom, and lacks an isotopic pattern
characteristic of C], which supports this claim. A cluster of peaks at m/z 463 (not shown)
may represent ions containing 3 Cl atoms, however the pattern is not exact. The pattern
may be affected by noise, considering the peaks are of such low intensity (~ 400 counts).
Peaks representing Cl' ions are noted at m/z 35 and 37.

Due to the noncomplirnentary positive and negative ion mass spectra of carmine,
there maybe more than one dye present in the carmine paint mixture, besides PG 7. The
positive ion mass spectrum possibly suggests the presence of a non-chlorinated cationic
dye, whereas the negative ion mass spectrum clearly shows evidence of a Cl-containing
compound. Certainly, not all of the PR colorants were considered in this example, which
may explain our inability to identify all of the dyes present. As will be explained in the
case of vermilion, yellow and orange pigments were frequently used in combination with
the red pigments. Consequently, the dye possibilities are numerous, and it was not our
intention to devote the entire project to colorant identification, but to complete a survey at

this point.

68

 

limit}

impoi

posili
and c

Guid

Dem
619
I68“

PIC

\X‘i

Vermilion (Red)

True vermilion will be discussed in detail in Chapter 4, however at this point it is
important to know that the name refers to the compound mercuric sulfide (HgS). The
positive and negative ion LD mass spectra of vermilion (Niji) are shown in Figure 2.13,
and certainly do not show any evidence for the presence of Hg. Referring to the Wilcox
Guide, the organic watercolor vermilion is typically composed of a mixture of red,
orange, and yellow dyes. Both spectra are rich with information, but as in the case of
carmine, the dyes remain unidentified. Complimentary peaks in both positive and
negative ion modes are noted, indicating the presence of at least one (M 1) or two (M|,M2)
neutral dyes. For instance a peak at m/z 620 appears in Figure 2.13a and a peak at mlz
619 is present in Figure 2.13b. The peaks may be representing M+ and [M-H]’ ions,
respectively. The same is noted for m/z 431 (+) /m/z 430 (-), which indicates the
presence of a second neutral dye (M2). Additionally, due to the lack of Cl' ions (m/z 35
and 37) in the negative ion mass spectrum, the dyes are apparently non-chlorinated,

which significantly decreases the number of colorant possibilities.

69

Relative Intensity

 

Relative Intensity

26

 

1

0

 

 

 

 

 

 

M+2 431

 

 

 

 

 

243
268
241
213 25 310
184 284 368 386
- ll 1
250 350 m,z 450
189 (MZ-H) 430
227
212 243 415
145
69 175 455
269
130 260 ' h 390
mlz

550

a)

my 620

 

650

(Ml-H)- 619

M.

520

b)

 

 

650

Figure 2.13: Vermilion (Niji Watercolors): a) the partial positive ion LD mass

spectrum; b) the negative ion LD mass spectrum

70

ldenti

dye c
be to
Appl
Deco
{lest
cont

deli

Identification of Watercolor Dyes (Known Dye Composition)

Fortunately, many modern day artists’ watercolor paints are sold in tubes with the
dye composition provided on the label. The majority of thestructures of these dyes can
be found in Herbst and Hunger’s Industrial Organic Pigments: Production, Properties,
Applications (41). Consequently, the mass spectral interpretation process, in most cases,
becomes much easier. As will be demonstrated in the following examples, the laser
desorption mass spectra of colorants does not always correlate with the chemical
composition found on the labels. In some instances only one of two components is
detected. This does not necessarily mean that the second component is not present, it
might simply mean that it is not detectable (does not absorb light efficiently at 337 nm) in
the LDMS experiment, or it is present at a very low level. Other times, unexplainable
peaks are generated which strongly suggest the presence of additional colorants not listed
on the label. This may be an intentional omission to protect trade secrets.

The spectra of several of the colorants in Table 2.3 will be presented and
discussed in this section. As stated in the introduction, these colorants were chosen, since
they were known to be unstable and fade easily overtime. The colorants which will be
discussed illustrate key points. Additionally, a few of the colorants not presented here,
such as Hooker’s Green and indigo, contained pigments such as PB 15 and PG 7, which
were already presented and discussed in detail in the Niji watercolor examples. Rose
madder lake will be presented in the following chapter on lakes, since the mass spectra of

the colorant parallels the spectra of some of the lakes presented in the chapter.

71

Cadmium Red Hue 095

Cadmium Red Hue 095 (Winsor & Newton) is a mixture of PR 149 (Figure
2.14a) and PR 255 (Figure 2.14b). Perylene Red BL (PR 149) is an anthraquinone dye
having a molecular weight of 598 Da, whereas Pyrrole Scarlet (PR 255) is categorized as
an aminoketone diketopypyrolopyrrole, having a molecular weight of 288 Da. PR 149
has not been subjected to the ASTM lighfastness test, however PR 255 earned an
excellent ASTM rating ofI (42). The positive and negative ion LD mass spectra of
Cadmium Red Hue 095 are shown in Figure 2.15. Theoretically, the mass spectral

interpretation process should be easy, since the colorants are already known.

H30 0 0 CH3
5 NWN 8 E
H30 0 O CH3

PR 149
(598 Da) H

  
  

PR 255
(288 Da)

0

Figure 2.14: Structures of: a) PR 149; b) PR 255.

72

 

 

289

 

289

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

3)
290
288
oz. 262 311
g .1 I 1.
,3 39 255 290 325
.5
2
’4': 378
£2
0
a: 23
311 486
262 i 1508
.-_J‘..,_L_u11 IAAATA “.4 2.4 1-- i. . A “A TEL L ‘3_.'m4 A
0 200 mlz 400 600
287
b)
287 485
..

>. 32’
3" E a
8 a w
3 - _
E .7. a. .1. a .1. 300
Q mlz
.2
E
Q
a:

“ALI-J... . . - 11 .MfLJLL ... .. .4- L...” L

0 275 mlz 550

Figure 2.15: Cadmium Red Hue 095 (Winsor & Newton): a) the positive ion LD
mass spectrum; b) the negative ion LD mass spectrum

73

In the positive ion mass spectrum, peaks are present in the m/z 288 region,
representing the characteristic neutral dye overlap of the M+ (mlz 288) and [M+H]+ (m/z
289) peaks, for PR 255. The same is noted in the negative ion mass spectrum at m/z 287
([M-H])’ and m/z 288 (M'), also representing PR 255. The peaks are enlarged in the
insets of the spectra, to observe the characteristic overlap. Intense clusters of peaks are
noted at m/z 378 and 486 in the positive ion mass spectrum (Figures 2.16a, b) and at m/z
485 in the negative ion mass spectrum (Figures 2.16c). The isotopic patterns of these
clusters suggest the presence of halogenated compounds, which neither PR 149 nor PR
255 are. Additionally, PR 149 has a molecular weight of 598 Da, and there are no peaks
representing the intact dye molecule in either positive or negative ion modes.
Considering the stable anthraquinone structure of PR 149, fragment ions are also not
expected. Therefore, this is a specific example in which the mass spectra of a colorant
does not coincide with the dye information provided on the label. There appears to be no
evidence of PR 149 present in the mass spectra, however there are peaks possibly

representing a second colorant not listed on the label.

74

378

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

. 376 a)
+ 1011
3’
’5
1:.-
3
.5
a) 380
.2
E W
G)
a:
J 382
l T ‘ M ‘ A l ’
365 380 395
486
b)
+ ion 484

3.:
'71

:1
8
.5

0
.2.
.. 1 .

er 506
m 466

460 mlz 490 520

485
e)
, 484

5 - 1011 483 436

i

E 87

.E’

E

a J

l ‘ F‘ l A“ h“ ‘ A ‘
470 480 mlz 490 500

Figure 2.16: Enlarged regions of the Cadmium Red Hue 095 LD mass spectra:
a) + ion, m/z 365-395; b) + ion, m/z 460-520; c) — ion, mlz 470-

500).

75

Mauve (Dark Red)

Winsor & Newton’s Mauve 398 contains the colorants PR 122 and PV 23RS, the
structures of which are shown in Figure 2.17. PR 122 goesby the common name
quinacridone magenta and is known to be initially resistant to light exposure. However,
PR 122 is known to eventually fade, and has therefore earned an ASTM lightfastness
rating of only 111. Dioxazine purple (PV 23) is a carbazole dioxazine dye, and may take
on either a blue (BS) or red shade (R8). The chemical distinction between the two foams
is not apparent, however they have distinctly different ASTM lightfastness ratings. For
instance, PV 23RS earns an ASTM rating of I, II, and III, when in a vehicle of oil,
acrylic, and water, respectively. However, PV 2388 has been rated a IV, as a watercolor
(42). The red shade is observed here.

Both of the colorants are represented in the positive (Figure 2.18a) and negative
ion (Figure 2.18b) LD mass spectra of Mauve 398. PR 122 (340 Da — M1) demonstrates
the classic trend of the neutral dye, generating molecular ions, as well as protonated and
deprotonated molecules in both the positive and negative ion mass spectra. Peaks
representing the neutral losses of one and two methyl groups are noted at m/z 326
[M-CH3]+ and 311 [M-2CH3]+ are formed in both positive and negative ion mode. PV
23RS (M2) behaves a little differently, though.

Peaks of relatively low intensity are present in the positive ion mass spectrum
representing the molecular ions of PV 23RS (mlz 588). The PV 23RS dye molecule
contains two chlorine atoms, which is reflected in the isotopic pattern observed at m/z
588 (enlarged in the inset of Figure 2.18a). The ratio is not exactly 9:6: 1, however the

presence of multiple elements skews the pattern. Additionally, the peaks are at such low

76

intensity, that noise may also be affecting the pattern. In the negative ion mass spectrum,
peaks representing the loss of one (m/z 559) and two ethyl groups (m/z 530) are noted.
These peaks maintain the 9:6:1 ratio, indicating that there was no loss of the Cl atoms.
There is no evidence of the intact molecule.

The higher mass peaks at m/z 663 and 661 in the positive and negative ion mass
spectra, respectively, are a mystery. Tentative assignments have been made. The
isotopic patterns of these peaks do not indicate the presence of Cl, therefore it was
assumed that they were due to some sort of polymerization, or reaction products
involving PR 122. Possibly, two PR 122 molecules joined via an 0 bridge, accompanied
by the loss of water. The resulting neutral species formed makes its presence known in a
LDMS experiment by becoming protonated (+ ion mode) or deprotonated (- ion mode).
If these assignments are incorrect, then there may simply be an additional colorant
present in the paint mixture with a molecular weight of approximately 660 Da. The
compound makes it presence known in both positive and negative ion modes, suggesting

that it is a neutral species.

77

 
 
  

PR 122
(340 Da)

PV 23RS
(588 Da)

 

Figure 2.17 : Structures of: a) PR 122; b) PV23RS.

78

Relative Intensity

Relative Intensity

588

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

(M.+H)+
341
590 a)
5 2
575 582 589 596 603 610
ml! (2M1-H20+H)+
663
(Ml-i-Na)+ \
(MI-CH3\+H)+ 363 M;
326 588
‘ L'ALA 3.1., ' LIV 'IIA L _: iLhtlLL
275 365 455 545 635 725
mlz
M,’ 340 b)
(Ml‘H)-
339
[M -( )l'
(Nil-(3113445)~ 2 9:1st _H OH).
326 [Mz-(CZH5)2]' / 1 6621
530
. -1 , at}? , , 617,717
250 350 450 mlz 550 650 750

Figure 2.18: Mauve 398 (Winsor & Newton): a) the partial positive ion LD mass
spectrum; b) the partial negative ion LD mass spectrum

79

Preliminary Light and Environmental Studies

The N iji watercolors were exposed to a variety of natural and artificial light
conditions to induce fading. These included natural sunlight, UV light (254 nm), and
light from a Hg lamp. In general, the colorants required longer exposure times to the
light sources, as compared to those used to degrade methyl violet, before fading was
evident. This indicated a higher degree of stability of the watercolor dyes. In most cases,
such as Prussian blue, cobalt blue, lemon yellow, and Viridian, the spectra obtained of the
samples following light exposure (not shown) were the same as the initial spectra.
Occasionally, low intensity, lower mass peaks appeared, but not consistently to suggest
the presence of degradation products. Thus, they were assumed to be impurities in the
sample. The most significant changes involved the red colorants. The vermilion
watercolor will be discussed in detail.

The positive ion mass spectra of UV - aged vermilion 24 hours (Figure 2.19a) and
aged with the Hg-lamp 10 hours (Figure 2.20a) appear to be relatively unaffected. One
might question whether or not the peaks at m/z 310 and 368 are becoming more intense
relative to the other peaks in the mass spectrum, however this was not a consistent
observation when compared to other spectra obtained from the same sample. These
peaks were also present in the positive ion mass spectrum of carmine. If they are truly
becoming more intense in the mass spectrum as the colorant fades, it may suggest that the
peaks are originating from the transparent vehicle, and are therefore not affected by light.

The positive ion mass spectrum of the sample that was faded naturally with
sunlight (Figure 2.21a) is significantly different than the initial spectrum. Peaks

representing the initial colorant have disappeared almost completely and the spectrum is

80

populated with peaks that seem to be unrelated to the dye. The peak at m/z 575
represents copper phthalocyanine, which is a component of the paper itself (explained in
more detail in a later chapter). More work would need to be devoted to determining the
degradation mechanism to identify these new peaks.

More changes are observed in the negative ion mass spectra. In both of the UV-
exposed (Figure 2.1%) and Hg-lamp exposed (Figure 2.20b) samples, all but the peak at
mlz 188 noticeably decreased in intensity, but relative to each other. The spectra are also
noisier. The sample exposed to natural sunlight produced a negative ion mass spectrum
in which the peaks in the m/z 430 and 619 regions decreased significantly, however the
remainder of the peaks remained intense. The spectrum contains more noise, since the
laser power needed to be increased significantly to obtain the spectrum.

From performing these fading experiments on the organic watercolor dyes, it
seemed as though, as a dye fades (degrades), its degradation products are either too small
and volatile, that they are simply removed by the vacuum system, or they are not able to
absorb the energy of the laser (337 nm). The fact that some of the peaks decreased in
intensity is an indication of the dye degrading. But what is happening to the dye? The
chemistry of these reactions is poorly understood and the lack of identified degradation
products confuses the issue. Based on these results, we decided to try different
environmental conditions. Here we subjected the Winsor & Newton watercolors to

pollutant gases, which included ozone, N02 and 802.

81

.-l.\\l|lliili..i

1" 1 i\11

heat-urea:— O>5§T¢-

.Ia.

 

 

 

 

 

 

 

 

  

  

   

431

 

 

 

 

a)
5‘
5 243
a)
>
3:
g 620
a 268
368
310
284
213 386
4 l L . I . - h ‘ 1 . .
l
50 250 350 mlz 450 550
189
b)
619
'8
g 227 430
0 200
>
'5 212
'3
m 145 181
\ 43 415
269
. 1.1-11.... Luluua.11.......1...1.-.1.1......1.....a....~...1fl.1 -.....11u.1..1..-......-...-..-,. .4“...-
125 230 335 440 545

mlz

Figure 2.19: Vermilion (Niji watercolors) exposed to 24 hours UV-light: a) the
partial positive ion LD mass spectrum; b) the partial negative ion LD
mass spectrum

82

650

 

 

650

ill-l.)

>.- 31:415.: 0)? «:0:

Relative Intensity

Relative Intensity

213

 

150

161
175

 

 

125

243

 

268

30

 

227

212

 

 

230

368
386

431

 

45

 

.1- .

335 mlz 440
Figure 2.20: Vermilion (Niji watercolors) exposed to 10 hours light from a Hg-arc
lamp: a) the partial positive ion LD mass spectrum; b) the partial
negative ion LD mass spectrum

83

 

 

a)
620
550
b)
619
574
545

650

 

 

650

111. -l fillili
>=a=3~=~ 0>mu~w~ne-

 

Relative Intensity

Relative Intensity

165

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

     

 

3)
265
128
100 189 210 320 mlz 430 540 650
b)
227
162
145
185
175 430 619
243
415
353
456 574
125 ' 230 335 mlz 440 -- 545 - I 650

Figure 2.21: Vermilion (Niji watercolors) exposed 3 months to natural sunlight
lamp: a) the partial positive ion LD mass spectrum; b) the partial
negative ion LD mass spectrum

84

majorilj
CllVll’OI‘.
ilauie

020118 1

puxr
liigu1
ltpre:
disap

(leer.

con
1&1)

C0

m

As in the case of the light experiments concerning the N iji watercolors, the
majority of the colorants remained unaffected when exposed to the different
environmental conditions. Some colorants faded, but their spectra remained the same.
Mauve 398 is an example which the spectra did not, and the results of the controlled
ozone exposure study will be summarized here.

Following exposure for 12.5 minutes to 03 (375 ppm), the Mauve 398 colorant on
paper faded, but did not appear to change color. In the positive ion mass spectrum
(Figure 2.223), there was only evidence of the intact PR 122 colorant (m/z 340). Peaks
representing PV 23RS and the tentatively identified compound at m/z 663 have
disappeared. In the negative ion mass spectrum (Figure 2.22b), the peak at mlz 661
decreased significantly and it is questionable whether PV 23RS is still present. If PV
23RS degraded completely, one would expect a color change along with the fading.
However, returning to the initial LD mass spectra of mauve (Figure 2.18), the
concentration of PV 23RS is minimal relative to the concentration of PR 122, as is
reflected in the relative intensities of the peaks representing the intact dye molecules.
Consequently, the presence of PV 23RS may have had more of an impact on the full-
strength color, however the absence of PV 23RS in the faded colorant did produce an
obvious shift in hue.

As with the light studies, either degradation or reaction products formed did not
seem to be easily detectable. Preliminary experiments exposing the samples to 802 and
N 02 were inconclusive (spectra not shown) and the experiments were difficult to control
with the equipment available in the lab. Consequently, we decided to change research

directions.

85

 

 

IL.)

.41 (MrrH)+

 

 

 

 

 

 

 

a)
5‘
'61
§
.5
0
>
3:
S
Q)
a:
275 365 455 545 635 725
mlz
340 M,“
13)
>3
.2; (Ml-H)
a
.5
O
>
'5
2
Q
a:
(2Ml-H20'H)-
.L y. val unit. _. . v“ A- ‘4 . u. ' a- '6.?1 _
250 350 450 550 650 750
mlz

Figure 2.22: Mauve 398 (Winsor & Newton) exposed 12.5 minutes to O3 (375 ppm):
a) the partial positive ion LD mass spectrum; b) the partial negative ion

LD mass spectrum.

86

Summary

The preliminary experiments performed concerning the LDMS analysis of water
colorants suggest that LDMS may be used to generate “fingerprint” spectra of organic
watercolor dyes, but may not be suitable for the analysis of their degradation products. If
the degradation products formed are volatile, they may not be detected in this experiment.
In the case that the products are nonabsorbing at 337 nm, traditional MALDI could be
employed. Considering that standard MALDI matrices are also of low molecular weight,
analyte peaks may get “lost” in matrix background peaks. Based on these experiments,
we decided to change research directions and look for colorants that were known to react
and change color over time in response to their environment, as opposed to fading.
Numerous inorganic artist pigments were known to do so. Consequently, the direction of

the project changed to the analysis of inorganic colorants.

87

References:
1) Museum Lighting Protocol, http://wwwncptt.nps.gov/notes/32/2.stm.

2) Brimblecombe P. The composition of museum atmospheres. Atmospheric
Environment 1990;25B: 1-8.

3) Hackney S. The distribution of gaseous air pollution within museums. Studies in
Conservation 1984;29:105-1 16.

4) Thompson G. The Museum Environment. London: Butterworths, 197 8.

5) Hisham MWM, Grosjean D. Air pollution in southern California museums: indoor and
outdoor levels of nitrogen dioxide, peroxyacetyl nitrate, nitric acid, and chlorinated
hydrocarbons. Environmental Science and Technology 1991;25:857-862.

6) Hisham MWM, Grosjean D. Sulfur dioxide, hydrogen sulfide, total reduced sulfur,
chlorinated hydrocarbons, and photochemical oxidants in southern California
museums. Atmospheric Environment 1991;25A: 1497-1505.

7) Eastlake CT, Faraday M, Russel W. Report on protection by glass of the pictures in the
National Gallery. London: House of Commons, 1850.

8) Grosjean D, Parmar SS. Removal of air pollutant mixtures from museum display
cases. Studies in Conservation 1991 :36: 129- 141.

9) Parmar SS, Grosjean D. Sorbent removal of air pollutants from museum display cases.
Environment International 1991 ;7 :39-50.

10) Hemphill JE, Norton JE, ijor CA, Stone RL. Color fastness to light and atmospheric
contaminants. Textile Chemist and Colorist 1976;8z60-62.

11) Williams EL, 11, Grosjean E, Grosjean D. Exposure of artists’ colorants to air
pollutants: reflectance spectra of exposed and unexposed samples. Final Report to
the Getty Conservation Institute, DGA, Inc., CA, 1991.

12) Csepregi Zs, Aranyosi P, Rusznak I, TOke L, Vig A. The light stability of azo dyes
and azo dyeings II. Perspiration - light stability of dyeings with reactive and non-

reactive derivatives, respectively, of two selected azochromophores. Dyes and
Pigments 1998;37:15-31.

13) Aranyosi P, Csepregi Zs, Rusznak I, TOke L, Vig A. The light stability of azo dyes
and azo dyeings III. The effect of artificial perspiration on the light stability of
creactive and non-reactive derivatives of two selected azo chromophores in aqueous
solution. Dyes and Pigments 1998;37:33-45.

88

19)

l4) Sirbiladze KJ, Vig A, Anyisimov M, Anyisimova OM, Krichevskiy GE, Rusznak 1.
Determination of the lightfastness of dyestuffs by kinetic characteristics. Dyes and
Pigments 1990;14:23-34.

15) Aranyosi P, Czilik M, Rémi E, Parlagh G, Vig A, Rusznak I. The light stability of
azo dyes and azo dyeings IV. Kinetic studies of the role of dissolved oxygen in the
photofading of two heterobifunctional azo reactive dyes in aqueous solution. Dyes
and Pigments 1999;43:173-182.

16) Grosjean D, Whitmore PM, Cass GR, Druzik JR. Ozone fading of natural organic
colorants: mechanisms and products of the reaction of ozone with indigos.
Environmental Science and Technology 1988;22:292-298.

17) Grosjean D, Whitmore PM, Cass GR, Druzik JR. Ozone fading of triphenylmethane
colorants: reaction products and mechanisms. Environmental Science and
Technology 1989;23:1164-1167.

18) Grosjean D, Whitmore PM, DeMoor CP, Cass GR, Druzik JR. Fading of alizarin and
related artists’ pigments by atmospheric ozone: reaction products and mechanisms.
Environmental Science and Technology 1987;21:635-643.

19) Grosjean D, Whitmore PM, DeMoor CP, Cass GR, Druzik JR. Ozone fading of
organic colorants: products and mechanisms of the reaction of ozone with curcumin.
Environmental Science and Technology 1988;22:1357-1361.

20) Shaver CL, Cass GR, Druzik JR. Ozone and the deterioration of works of art.
Environmental Science and Technology 1983;17:748-752.

21) Whitmore PM, Cass GR. The ozone fading of traditional Japanese colorants. Studies
in Conservation 1988;33:29-40.

22) Whitmore PM, Cass GR, Druzik JR. The ozone fading of traditional natural organic
colorants on paper. Journal of the American Institute for Conservation 1987;26:45-58.

23) Salvin VS. Ozone fading of dyes. American Association of Textile Chemists and
Colorists 1969;1z22-28.

24) Whitmore PM, Cass GR. The fading of artists’ colorants by exposure to atmospheric
nitrogen dioxide. Studies in Conservation 1989;34:85-97.

25) Grosjean D, Salmon L, Cass GR. Fading of artists’ colorants by atmospheric nitric
acid: reaction products and mechanisms. Environmental Science and Technology
1992;26:952-959.

26) Salmon LG, Cass GR. The fading of artists’ colorants by exposure to atmospheric
nitric acid. Studies in Conservation 1993;38:73-91.

89

'4)
D

Cth

jslshan
Phlha

Mari.
asSISi€(
2084,

27) Williams EL, 11, Grosjean E, Groshean D. Exposure of artists’ colorants to
peroxyacetyl nitrate. Journal of the American Institute for Conservation 1993;32:59-
79.

28) Williams EL, 11, Grosjean E, Grosjean D. Exposure of artists’ colorants to airborne
formaldehyde. Studies in Conservation 1992;37:201-210.

29) ASTM Method D 2244-93 Standard test method for calculation of color difference
from instrumentally measured color coordinates. American Society for Testing and
Materials. West Conshocken, PA, 2001.

30) Grim DM, Siegel J, Allison J. Evaluation of desorption/ionization mass spectrometric
methods in the forensic applications of the analysis of inks on paper. J Forensic Sci
2001;46:1411-1419.

31) Grim DM, Siegel J, Allison J. Does ink age inside of a pen cartridge? J Forensic Sci
2002 :47: 1294- 1297.

32) Grim DM, Siegel J, Allison J. Evaluation of laser desorption mass spectrometry and
UV accelerated aging of dyes on paper as tools for the evaluation of a questioned
document. J Forensic Sci 2002;47:1265-1273.

33) ASTM Method D 4303-99 Standard test methods for lighfastness of pigments used in
artists’ paints, May 10, 1999. American Society for Testing and Materials. West
Conshocken, PA, 2001.

34) ASTM Specification D 5067—99 Standard specification for artists; watercolor paints.
American Society for Testing and Materials. West Conshocken, PA, 2001.

35) Personal communication with Dr. James N. Jackson, Chemistry Department,
Michigan State University, East Lansing, MI.

36) Adrasko J. HPLC analysis of ballpoint pen inks stored at different light conditions. J
Forensic Sci 2001;46:21-30.

37) Aginsky VN. Dating and characterizing writing, stamp pad and jet printer inks by gas
chromatography/mass spectrometry. Int J Forensic Doc Examiners 1996;2:103-116.

38) Shankai Z, Feng Z, Weide H, Zhongping Y, Zhang W, Chait T. J Porphyrins
Phthalocyanines 1995;9z230.

39) Conneely S, McClean S, Smyth F, McMullan G. Study of the mass spectrometric
behaviour of phthalocyanine and azo dyes using electrospray ionisation and matrix-
assisted laser desorption/ionisation. Rapid Commun Mass Spectrom 2001;15:2076-
2084.

90

4;
' I.

461

40) Berrie BH. Prussian blue. In: Fitzhugh EW, editor. Artists' pigments: a handbook of
their history and characteristics, vol. 3. New York: Oxford University Press,l997;
191-217.

41) Herbst W, Hunger K. Industrial Organic Pigments. Production, Properties,
Applications, 2nd ed. Weinheim: VCH, 1997.

42) Wilcox M. The Wilcox guide to the best watercolor paints, 2001-2 ed. USA: School
of Colour Publications, 2001.

43) Kuhn H. Verdigris and copper resinate. In: Roy A, editor. Artists’ pigments: a
handbook of their history and characteristics, vol. 2. New York: Oxford University
Press, 1993;131-158.

44) Douglas B, McDaniel D, Alexander J. Concepts and models of inorganic chemistry,
3rd ed. New York: John Wiley & Sons, 1994.

45) Delamare F, Guineau B. Colors. The Story of Dyes and Pigments. New York: Harry
N. Abrams, 2000.

46) Grim DM, Allison J. Identification of colorants used in watercolor and oil paintings
by UV laser desorption mass spectrometry. International J Mass Spectrometry
2003 ;222:85-99.

91

Elf.

del

PIC:

Chapter Three: Lake Pigments

Introduction:

A unique class of colorants used in art is the class knOwn as lakes. Such dye-
based paints have been used since ancient times. Medieval painters mostly used mineral
pigments. Dyes were also available for dyeing fabrics. These dyes were attractive but
were chemically inappropriate for painting. The solution was to chemically bind the dyes
to mineral substances to make the colors insoluble. The resulting pigment was a called a
lake. A true lake is defined as a transparent color precipitated on a transparent base (1).
A lake is prepared by dissolving the organic colorant in water and then precipitating the
colorant on a base by introducing a solution of an appropriate metal salt or oxide. The
introduction of lakes greatly increased the range of pigment colors available. In 1809,
some pigments were discovered in a shop in the excavations at Pompeii. Following the
analysis of the pigments, Chaptal determined that a particular pink pigment was a madder
alumina lake (2). Madder lakes use coloring substances extracted from madder roots. If
a museum laboratory is fortunate to have XRF available as a tool, they may be able to
determine that an inorganic material such as alumina (A1203), tin oxide, or zinc oxide is
present as the support, but they could not detect the organic dye.

As discussed in the introductory chapter, art conservators are in need of new
analytical techniques to help them identify and characterize lake pigments. As a result of
their unique composition, lakes require a minimum of two analytical techniques to be
characterized. One method is used to identify the organic portion, and the second is used
to characterize the inorganic portion. Before this is done, a separation of the two

components is required, which is not an easy task. Laser desorption mass spectrometry is

92

ca
col

COS

Oil p
CXpC

their

a versatile tool, which is able to provide molecular information on both organic and
inorganic compounds. LDMS was investigated to determine if it would be a beneficial
addition to the analytical techniques currently available to art conservators. The results

of the analyses of a few lakes will be presented and discussed.

Carmine Lake

Carminic acid has been used as a colorant since the 1500s. The natural organic
dye is made from the dried bodies of female beetles, coccus cacti, that live on cactus
plants in Mexico and in Central and South America. They were brought to Europe soon
after the discovery of these countries. The scarlet red colorant immediately gained
popularity, and was commonly used to dye fabric and cosmetics. However, the origin of
the dye was a well-kept secret, simply because the ladies of this time period would have
been horrified to discover that a component of their lipstick originated from a beetle.
Eventually, the secret was revealed, and controversy ensued. Animal rights groups of
this time period objected to the use of the colorant, because they believed that fashion
was an inexcusable reason to sacrifice the lives of millions of beetles. The use of
carminic acid subsequently tapered off, however it is still commonly used today. The
colorant is identified as Natural Red No. 4 by the color index, and is still used to dye
cosmetics (lipstick, blush, and eyeshadow) and even Cherry Coke (2,3).

Carminic acid and carmine lake are violet-red colorants that continue to be sold as
oil paints, even though they are considered to be fugitive, meaning that extended
exposure to sunlight can lead to bleaching of the lake. Carmine lakes are known to lose

their color when mixed with specific pigments, as well. Consequently, the lake is

93

 

freque
narura

carmir

alum 1
Shown
alumin
indicat
the ger
dominz
precipi;

Organic

frequently considered to be unsuitable for most artistic use. Unlike madder lakes, most
natural organic lakes will change over time when exposed to light. In many uses,
carmine lake is being replaced by the more stable alizarin lake (4).

Positive and negative ion LD mass spectra (Figure 3.1) were obtained for carmine
alum lake suspended in linseed oil and applied to paper. The structure of carmine lake is
shown in the inset of Figure 3.1a. As the name implies, the inorganic substrate is
alumina (A1203). In the positive ion LD mass spectrum (Figure 3.1a), the major ions are
indicative of the aluminum oxide support. These ions are listed in Table 3.1, and have
the general formula AIXOYH2+. The negative ion spectrum (Figure 3.1b) appears to be
dominated by peaks representing an organic compound. A lake is prepared by
precipitating an organic dye onto an inorganic substrate. In the case of carrrrine lake, the

organic dye is carminic acid.

Carmine Lake: Positive Ions

 

 

 

 

 

 

 

 

 

m/z .
Assrgnment
60 A102H+
74 16.120+
104 Al203H+
131 A1303H2+
236 A1506H5+
365 AlsogHs"

 

 

Table 3.1: A list of the identified positive ions for carmine lake.

94

ante—5a:— atom—oz

5.55:5 9:33“

Figu rt

A1,o,H,+

 

 

 

 

 

 

 
   

 

 

 

 

 

 
 

104
Q “a E- 0.. a)
,. :CUC 0..
3‘
ca 1-120-4>M- -0H
5
a HOCH,
{:3 COO
.2 86 0 CH; 3
*5 74 A13O3H2’r
Ta 131
a: 285 365
301
60 184 236 323
|
50 150 m/z 250 350
327 b)
>.
J‘:
(D
s:
3
.5
o
.2
E 298
52 269 447 491
42 97 675
473
_ 1 . . .
0 180 360 mlz 540 720 900

Figure 3.1: The LD mass spectra of carmine alum lake in linseed oil on paper; a) the
partial positive ion mass spectrum and structure of carmine alum lake; b)
the negative ion mass spectrum.

95

[M-Hl'

 

 

 

 

491
[M-H-Co,]- ‘
s? 447
m
E
8
5 [(A-H)-H1-
E 327 473
g _ _. [M-H-H20]'
a: 299 353
371 418
200 300 400 500 600 ‘
mlz
Figure 3.2: The partial negative ion LD mass spectrum of carminic acid in linseed oil on

paper.

 

Figure 3.3: The structure of carminic acid.

96

negat

appli

11611
BC 11'
De

3C1

To further investigate the ions observed from the lake in linseed oil, positive and
negative ion LD mass spectra were obtained of carminic acid suspended in linseed oil and
applied to paper. The negative ion spectrum of carminic acid in linseed oil is shown in
Figure 3.2. Several peaks were identified in the negative ion spectrum derived from the
organic dye. Neutral carminic acid weighs 492 Da (M) (Figure 3.3). The peak at mlz
491 represents the pseudomolecular [M-H]' ions. Peaks at mlz 473 and 447 represent the
neutral losses of H20 and C02 from the deprotonated molecule, respectively. Carminic
acid is composed of an anthraquinone base (A), with a glucopyranosyl sugar group (G).
Designating the compound as G-A, a reaction occurs in which the sugar group is lost,
accompanied by a H shift, reaction:

GA 9 (G+H) + (A-H)
This leads to the [(A-H)-H]' ions at m/z 327. This could be either a fragment ion peak or
the pseudomolecular ion of an anthraquinone impurity. The same peaks are observed in
the negative ion mass spectrum of carmine lake (Figure 3.1b), but with varying
intensities. This is not unexpected since in the lake, carminic acid is present as a dianion.
Additionally, when the alumina support is taken away, all of the peaks listed in Table 3.1
disappear, which is evident in the positive ion mass spectrum of carminic acid in linseed
oil on paper (not shown). Thus, the presence of both the organic and inorganic
components of the lake could be detected using both positive and negative ion modes.

As a side note, when carminic acid is applied to paper from an aqueous solution,
the sample instantly turns from red to black. Positive and negative ion LD mass spectra

(spectra not shown) were obtained of the black sample. Few peaks were observed in the

97

SUI

R0:

r0015

1

3313)

llth

positive ion spectrum and the negative ion spectrum is dominated by C.{ clusters (n = 2-
13). Formation of the lake clearly stabilizes the molecule so it can be used as a colorant.
In the carmine lake example, the positive ion mass spectrum provided information
on the inorganic support (alumina) and the negative ion mass spectrum provided
structural information on the organic dye (carminic acid). Ideally, we had hoped to
identify peaks in either mass spectrum representing carminic acid-alumina complex ions.
There is one peak at mlz 675 in the negative ion mass spectrum of carmine lake (Figure
3.1a), which is not present in the negative ion mass spectrum of carminic acid. This peak
has been tentatively assigned as [M+A1503]'. To confirm this assignment, one could
perform a post source decay (PSD) LDMS experiment, to gain structural information on
the ion. However, a potential problem could be that an insufficient number of ions at m/z
675 are initially formed. On average, an ion count of at least 10,000 is needed in a
standard LD or MALDI experiment to perform a PSD analysis. Another option is to
switch instruments. When available, a FT MS can be used to perform extreme high
resolution experiments, which can be used to confirm the exact mass of an ion. If the
assignment is correct, it is one piece of evidence of the organic dye and the inorganic

support intact, which would be unique only to carmine lake.

Rose Madder

Historically, madder lake was manufactured using the dyestuffs found in madder
roots. These coloring substances include alizarin (1,2 dihydroxyanthraquinone, Figure
3.4a), purpurin (1,2,4-trihydroxyanthraquinone, Figure 3.4b), and pseudopurpurin (1,2,4-

trihydroxyanthraquinone—3-carboxylic acid, Figure 3.4c). The color of madder lake

98

depei
SUPP‘
cont
Ono

synt

m5

depends on the concentrations of the organic colorants present, as well as the inorganic
support the colorant is precipitated on. For example, a madder lake with a high
concentration of purpurin will have more of a pinkish tone rather than a scarlet tone.
Once alizarin was synthesized, natural madder lakes were replaced by the less costly,

synthetic alizarin lakes (5).

a) 0 b) 0 ”0

lllllnllliil OH

0 0H 0 OH

OH

COOH

OH
0 0H

Figure 3.4: Structures of a) alizarin; b) purpurin; and c) pseudopurpurin.

99

Rose Madder 035 (Winsor & Newton), mentioned in Chapter Two, was analyzed.
According to the label on the watercolor tube, 1,2 dihydroxyanthraquinone lake (PR 83)
was the only dye used. The structure of PR 83 is shown in Figure 3.5. Like carmine
lake, the organic component PR 83 is a dianion. The watercolor was dissolved in water
and applied to paper. Positive and negative ion LD mass spectra were obtained and are
shown in Figures 3.6a and 3.7a, respectively. The lake (dianionic) portion of PR 83 (M1)
has a molecular weight of 520 amu. The second most abundant peak in the positive ion
mass spectrum occurs at m/z 543. This peak has been tentatively assigned as (M1-
0H+Ca)+. M1 carries a -2 charge. When the hydroxyl group is removed from the Al
atom, the Al atom is left with a +1 charge, yielding a net -1 charge for (Ml-0H). A peak
representing this species (m/z 503) is noted in the negative ion mass spectrum of Rose

Madder (Figure 3.7a). When a Ca2+ ion attaches to the lake, the species now carries a net

 

+1 charge.
0
/ l‘ \
\ /
o-
O 0
/Al 0H Ca2+
O O
.0 O-
0

Figure 3.5: Structure of alizarin lake, PR 83.

100

Relative Intensity

 

Relative Intensity

 

+

0

39

 

 

 

39

 

1.1.1.-..

 

 

 

 

 

 

 

M; 443
(Ml-OH-I-Ca)+
543
(M,-C,H5)+
415 821
278
0—9
400 800 z 1200
543
821
115 288
1104
400 800 mlz 1200

279 1648
e——v
1369

 

1425
281 1102 I I 279 1704
L4. 1. . 1 m . .

 

 

T600

 

1648
1369
1425 1704
.l 1 ll 1
1600

 

1.1
2000

Figure 3.6: The positive ion LD mass spectra of lake pigments in water on paper;
a) Rose Madder and b) R37.

101

>....z:3:~ 01:45.01

Figure

 

 

 

 

 

 

 

 

 

 

 

(M-OH)‘ 503
2!)
>:
H
'5
=
d)
H
.5
Q)
.2
H
.53
v 781
m
278
¢——>
1606
223 281
$239 a———*1062 1383
11111 .111. w1l111fi1k11 . 11L 1 . . 11.11
0 400 800 mlz 1200 1600 2000
503
b)
>3
fl
'5
:'
G)
H
.5
Q)
.E:
H
E
0
m
781
223
AL 1 111L_1 v 11111 1 1 .44 1 . 11 Y 1 111 '1 1
0 400 800 mlz 1200 1600 2000

Figure 3.7: The negative ion LD mass spectra of lake pigments in water on paper;
a) Rose Madder and b) R37.

102

The most intense peak in the positive ion mass spectrum occurs 100 mass units
lower at m/z 443. From experience, the peak was easily identified as the cationic red
dye, Rhodamine 60 (M2). The structure of the dye is shown in Figure 3.8. Rhodamine
6G is a commonly used dye in modem red ballpoint ink formulas and is discussed in
more detail in the following chapter. This dye was not listed in the chemical composition

of the watercolor Rose Madder.

‘3
@0011ch3
CI ‘
CH3 \ CH3
4.
CHBCHzNH o 011101126113

Figure 3.8: The structure of Rhodamine 6G.

The higher mass regions of both the positive (Figure 3.6a) and negative (Figure
3.7a) ion mass spectra are not as easily explained. In both spectra, there are a series of
peaks separated by approximately 280 mass units. In the m/z 1300-1700 range of the
positive ion mass spectrum, there even appears to be a second series occurring, 56 mass
units higher than the first set. To help determine what these peaks are representing, we
considered a second lake pigment, which generated very similar LD mass spectra.

The second lake pigment was labeled “R37” (Tuscan red). The pigment has a
unique origin, which will be addressed in Chapter 4 (“013”). R37 has a chemical

composition of 45% ferric oxide, 17% alizarin lake, and 38% calcium carbonate. The

103

USc’

iftl

€pr

positive and negative ion LD mass spectra of R37 suspended in water and applied to
paper are shown in Figures 3.6b and 3.7b, respectively. Neglecting the peaks
representing Rhodamine 6G in the positive ion mass spectrum, the spectra are nearly
identical. Originally, the mass difference (~ 280 mass units) was thought to be caused by
the attachment of linseed oil molecules (discussed in greater detail in Chapter 4). The
reasoning behind this explanation was based on the fact that linseed oil is commonly
ground in with pigments during their manufacture. Considering what both lakes have in
common (alizarin lake), a second explanation might be the adduct of a

1,2-dihydroxyanthraquinone molecule and a Ca atom.

Summaty

From these few examples, we have learned that not all lakes generate the same
type of ions in the LDMS experiment. For example, carmine lake generated ions
indicative of the organic dye and inorganic support as separate entities. There was no
evidence in either mass spectrum of the intact lake molecule as shown in Figure 3.1a.
The Rose Madder watercolor was different. Peaks were identified in both the positive
and negative ion mass spectra representing the intact lake molecule (with the exception of
a hydroxyl group). The results of the Rose Madder experiment are ideal, simply because
the peaks are unique only to alizarin lake. The results of the carmine lake are also quite
useful, considering that carminic acid is not typically used alone as a pigment. Therefore,
if the dye is present, the pigment is most likely carmine lake. More lakes would need to
be analyzed to fully realize the potential of LDMS in this area. The results of these

experiments, suggest that it has considerable potential.

104

References

l) Delamare F, Guineau B. Colors. The story of dyes and pigments. New York: Harry N.
Abrams, Inc., 2000.

2) Schweppe H, Winter J. Madder and Alizarin. In: Fitzhugh EW, editor. Artists'
pigments: a handbook of their history and characteristics, vol. 3. New York: Oxford
University Press, 1997; 109-142.

3) Finlay V. Color: a natural history of the palette. New York: Ballantine Books, 2002.

4) Weber FW, Artists’ pigments: their chemical and physical properties. New York:
D. Van Nostrand Company, 1923.

5) Schweppe H, Roosen-Runge H. Carmine - cochineal carmine and kermes carmine. In:

Feller RL, editor. Artists' pigments: a handbook of their history and characteristics,
vol. 1. New York: Cambridge University Press, 1986; 255-283.

105

St;

9111
C112
Sim

“Sig

Chapter Four: Inorganic pigments

Introduction

Nearly fifty years ago, Rutherford J. Gettens of the Freer Gallery of Art had a
vision of a comprehensive source on artists’ materials. His project began after realizing
that limited information was available on pigments in the literature. In 1961, with the
support of the International Institute for Conservation (HC) of Historic Artistic Works,
Gettens proposed an international effort to accumulate a series of pigment monographs of
the highest quality, each becoming its own chapter in the completed handbook. The goal
was to provide artists, conservators, scientists, and art historians with an up-to-date
reference on the history of pigments’ manufacture and occurrence in paintings, as well as
on the detailed results from the analytical techniques that have been used to characterize
and identify them. Much to Gettens’ dismay, only nine monographs were completed and
ready for publication between the years 1966 and 1974. Limited funding and slow output
prompted Gettens to appeal to the National Gallery of Art to take over the project. The
absence of a formal publication agreement between Gettens and J. Carter Brown, the
director of the Gallery at that time, and the death of Gettens in 1974, resulted in a
standstill for the project.

The project was soon revitalized by the new director of the National Gallery of
Art, Robert L. Feller. Ten more pigment monographs were produced, and were
published as Volume I of Artists’ Pigments: A Handbook of Their History and
Characteristics (1). By this time, the original nine monographs previously published in
Studies in Conservation, the journal for the IIC, were outdated. As a tribute to Gettens’

vision, these monographs were revised to include more recent historical and scientific

106

developments, and republished in Volume II of the Artists’ Pigments Handbook, edited
by Ashok Roy (2). Notably, x-ray crystallographic and scanning electron microscopic
(SEM) data were now available and included. Historical aspects and notable occurrences
were also researched more thoroughly and expanded upon. Volume III (3) has since been
published, edited by Elisabeth West Fitzhugh, discussing an additional ten pigments.
Unfortunately, these three volumes present only 29 out of the 50 to 75 pigments Gettens’
had originally hoped to discuss. Additionally, the authors of the monographs were
thorough not only in providing information on what is known about the pigments, but
what still remains unknown, such as determining the detailed mechanisms for specific
color changes that occur over time. These two factors indicate that there is still much
research to be done.

Table 4.1 lists the pigments covered in Volumes I—III of the Artists’ Pigment
handbooks (1-3). Based on the history of the handbook, we believe that the individual
chapters cite and discuss all of the analytical chemistry and spectroscopy that has been
used to characterize and identify the pigments. While numerous analytical techniques
have been employed by art experts and conservators to characterize pigments, including
microchemical spot tests, microscopy, X-ray fluorescence (XRF) and infrared, RAMAN,
and UV spectroscopies, each method has its limitations. It is rare that the analytical
laboratory in a museum, if present at all, would have an extensive arsenal of tools for the
analysis of colorants. XRF is a popular tool when available and provides useful atomic
information. It may be used to determine that a pigment contains metals such as Pb and
Cr, however this is insufficient to identify a compound. Also, XRF cannot be used to

detect elements lighter than F (4), an obvious limitation when the colorant is organic. In

107

such cases, the absence of a metal may be the most useful clue. Other methods, such as

IR spectroscopy, produce a fingerprint spectrum of the pigment, which can be matched

with a standard spectrum. However, the presence of binding media and other pigments

can complicate interpretation. Frequently, spot tests are the most direct way to determine

the presence of specific chemical species, but such methods are ultimately destructive.

 

 

 

 

 

 

 

 

 

Chromate Pigments

Verditer

Volume 1 Volume II Volume III
Indian Yellow Azurite and Blue Verditer Egyptian Blue
. Ultramarine Blue, Natural .
Cobalt Yellow (Aureolm) and Artificial 0rp1ment and Realgar
Barium Sulfate, Natural Lead White Indigo and Woad
and S ynthetic
gadrruum Yellows, Lead-Tin Yellow Madder and Alizarin
ranges, and Reds
Red Lead and Minium Smalt Gamboge
Green Earth Verdigris and Copper Vandyke Brown
Resmate
Zinc White Vermilion and Cinnabar Prussian Blue
Chrome Yellow and Other Malachite and Green Emerald Green and

Scheele’s Green

 

Lead Antimonate Yellow

Calcium Carbonate Whites

Chromium Oxide Greens

 

Carmine

 

 

 

Titanium Dioxide Whites

 

Table 4.1: Summary of the pigments discussed in the three volume series Artists’
Pigments: A History of Their Manufacture and Characteristics.

We have investigated laser desorption/ionization mass spectrometry (LDMS) as a

minimally invasive method for confirming the presence of a specific pigment.

Preliminary work has shown that both positive and negative ion LDMS spectra of

inorganic pigments suspended in linseed oil may be easily obtained. The spectra contain

molecular level information, which could allow for the distinction between two pigments

108

 

of similar composition. Once fully developed, we believe that this method will be a
significant contribution to Gettens’ work, as a means to identify pigments.

As discussed in Chapter One, the primary deficiencies of current analytical
instrumentation used to identify and characterize artists’ pigments include:

1) the inability to detect organic and inorganic components simultaneously
2) the inability to generate molecular information concerning a colorant
3) the inability to detect multiple colorants in a paint mixture

LDMS excels in all of these areas. The ability of LDMS to generate positive and
negative ions indicative of both the organic colorant and inorganic support, in the specific
case of lake pigments, was demonstrated in the previous chapter. In this chapter, the
results of the LDMS analysis of two additional pigments (Prussian blue and “013”), will
be presented and discussed. These examples illustrate how LDMS can be used to solve
problems 2 and 3, listed above.

Once the usefulness of LDMS in the identification and characterization of
inorganic pigments has been established, a specific application will be presented. Two
old illuminated (decorated) manuscripts were analyzed using LDMS. The first document
is purportedly a page of a Koran from the 17Lh century. The primary source of the second
document is unknown, however it is allegedly created in the 1800’s, and is written in
Hebrew. LD mass spectra of the colorants were obtained and analyzed to help determine

the authenticity of the documents.

109

The Case of the Two Prussian Blues:

Prussian blue is an inorganic pigment, introduced in the early 17008. It is
considered the first of the modern pigments. Prussian blue, is a hydrated iron
hexacyanoferrate complex, and the generic term “iron blue” refers to both ferric
ferrocyanide, Fe4[Fe(CN)6]3 ° xH20, and the more soluble KFe[Fe(CN)6] - xH2O ("alkali
ferric ferrocyanide"). Ammonium and sodium salts are also used, with x = 14-16 (5).
The pigment became instantly popular worldwide for artists' palettes, interior painting,
wallpapers, as well as for dying fabrics such as silk (6-8). The structure of ferric
ferrocyanide is shown in Figure 4.1. Prussian blue was a key material used for making
cyanotypes, a primitive photograph and precursor to the blueprint (9,10). It has also been
incorporated into ink, cosmetics, and automotive paint formulations (5). Artists have
used Prussian blue as both a watercolor and an oil paint pigment. When the medium is
water-based, alcohol is frequently added, improving the pigment’s solubility and
preventing flocculation. It was popular, not only as a blue colorant with extremely high
"tinting strength" (1 1), but also as a compatible pigment to be used in mixtures with

pigments such as lead chromate, to produce the chrome green pigment (5,12).

 

 

 

Figure 4.1: The structure of ferric ferrocyanide.

llO

Studies of Prussian blue by Mossbauer spectroscopy (13), photoelectron
spectroscopy (16), IR (17-19), neutron activation (20), Raman spectroscopy (21), and
scanning electron microscopy (5) have been reported. In powder form, Prussian blue is
of fine particle size, typically 0.01 to 0.2 microns in diameter. Dispersed Prussian blue
is transparent under optical microscopy, and may appear as a blue streak. Aggregates
are opaque, resembling indigo. Complications in identifying Prussian blue using optical
microscopy have been discussed (5,22,23). As a result of its high tinting strength,
Prussian blue is frequently present in only small amounts in oil paints, where impurities,
bulking agents and extenders can further confuse identification. Consequently, it is
often not detectable using methods such as X-ray powder diffraction analysis (5).
Another problem in identifying Prussian blue involves confirming its presence in a
mixture of pigments. For instance, ultramarine blue is an inorganic pigment physically
resembling Prussian blue in a painting. Ultramarine blue does not contain Fe; however
when the pigment is in a mixture of other pigments that do, a false positive identification
for Prussian blue may result (5).

Consider the situation in which two watercolor artists have painted with a color in
their palette specifically identified as Prussian blue. Figure 4.2 shows the positive and
negative ion laser desorption mass spectra of Prussian blue (from the supplier Sinopia,
San Francisco, CA) of the first artist, directly from paper. The positive ion spectrum,
Figure 4.2a, is relatively uninformative, with the exception of the interesting peak at m/z
575, which will be discussed shortly. Some peaks are present, but do not appear to be
related to the known structure of Prussian blue. The negative ion spectrum, Figure 4.2b,

contains a number of peaks that can be rationalized as evolving from an iron cyanide.

111

Assignments were made by considering the masses of Fe (56 amu) and cyanide (CN, 26
amu), and by taking advantage of the characteristic isotopic profile of Fe which, notably,
has an (A-2) isotope at m/z 54 (6.43%). Table 1 lists all of the peaks in the negative ion
mass spectrum that have been identified thus far. The most intense peak is identified as
Fe(CN)3' at m/z 134. All of the peaks listed in Table 4.2 had isotopic satellites consistent
with the presence of one or more iron atoms. For the majority of the ions listed in Table
1, the metals are in the +1 or +2 oxidation states. For example, mlz 134 has an isotopic
signature consistent with the presence of one Fe atom, which would be in a formal
oxidation state of +2, complexed with three anionic ligands, to yield a singly charged
anion. A number of peaks 16 mass units above Fex(CN)y' peaks were observed,
suggesting the incorporation of oxygen. These could be incorporated into the ligands.
Consider mlz 166, corresponding to [Fe(CN)3 + 20]'. If both oxygen atoms were directly
attached to the metal, it would be in, formally, a +8 oxidation state. Consequently, we
tentatively assign it as [Fe(OCN)2(CN)]'. Also, when oxygen is present in these ions,
there are always fewer O atoms present than CN groups, so direct Fe-O bonding need not

be invoked.

112

Prussian Blue: Negative Ions

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

mlz Assignment Fe-Oxidation States
108 Fe(CN)2' +1

124 Fe(OCN)(CN)' +1

134 Fe(CN)3' +2

150 Fe(OCN)(CN)2' +2

166 Fe(OCN)2(CN)' +2

182 Fe(0CN)3' +2

190 Fe2(CN)3’ (+l,+1) or (0, +2)
206 F62(OCN)(CN)2- (+1,+1) 01' (0, +2)
216 Fe2(CN)4' (+1,+2) or (0, +3)
222 Fe2(OCN)2(CN) ' (+1,+1) or (0, +2)
242 Fe2(CN)5' (+2,+2) or (+1,+3)
258 Fe2(0CN)(CN)4' (+2,+2) or (+l,+3)
274 F62(OCN)2(CN)3- (+2,+2) OI‘ (+1,+3)

 

Table 4.2: A list of the identified negative ions formed by laser desorption from
Prussian blue (aqueous) on paper, and the corresponding oxidation states

ofFe.

Evaluation of the ions presented in Table 4.2 offers insight into the types of ions
to expect, and just as importantly, into the ions which would not be found in a LD mass
spectrum of similar materials. For example, it appears as though Fe prefers lower
oxidation states, despite the fact that there are both Fe2+ and Fe3+ in the original crystal
structure (Figure 4.1). In the negative ion mass spectrum, there are signals at m/z 108
and 134, representing the Fe(CN)2' and Fe(CN)3' ions, requiring Fe to be in the +1 and +2
oxidation states, respectively. However, the absence of a signal at m/z 160 representing
Fe(CN)4' is noted, which would require Fe to be in the +3 oxidation state. Furthermore,
the most dominant peaks in the mass spectrum, occurring at m/z 134, 150, and 166, all

represent ions in which Fe would have to be in the Fe2+ state. Additionally, there are

113

peaks representing Fe(OCN)3', but not Fe(OCN)4', and peaks representing Fe2(CN)3',
Fe2(CN)4’, and Fe2(CN)5‘, but not Fe2(CN)6’. For ions such as Fe(0CN)4' and Fe2(CN)6',
at least one Fe would have to be in the +3 oxidation state. In a case such as Fe2(CN)4', in

which mixed oxidation states are possible, the presence of Fe3+ need not be invoked.

114

.I‘o-U:Qh=~ Q>o~h~w~mz

figure 4.

104

575

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

a)
>1 2‘
3:1 3 ‘
a 88 s
o i
H i
E i 578
Q j L111 A
> 565 5‘70 575'”,
0:
£3 57
4’ 120
m 72 213 261
1 368 429
MW“ ‘1 “ ““““T‘
50 150 250 mlz 350 450 550
Fe(CN)3' b)
134
Fe(0CN)(CN)2'
150
>5
.2
(I)
G
8 Fe(OCN)2(CN)'
.5 166
0 F 82(CN)5'
3 242
*5 Fe2(OCN)(CN)4'
a, 258
CN' 108 206 Fe2(OCN)2(CN)3'
26 96 ‘ 274
l l I .
0 69 137 mlz 204 272 340

Figure 4.2: The LD mass spectra of Prussian blue (Sinopia) from an aqueous solution
on paper; a) the partial positive ion mass spectrum; b) the negative ion mass

spectrum.

115

fér
H111

C011

mass

Prese

In these spectra, there are certainly peaks that remain unassigned. However,
during the manufacturing of artist’s pigments, extenders and bulking agents such as alum,
CaCO3, CaSO4, starch, alumina, magnesia, and clay are often added to provide the
pigment with specific properties or to reduce the cost of production. As a result, what
might be reported to be ferric ferrocyanide may in fact be a complex mixture. In
summary, the negative ion spectrum provides ample evidence that the pigment is an iron
compound, and that it is specifically an iron cyanide. The molecular information is
certainly superior to what would be determined by a method such as XRF, which would
only indicate the presence of iron. Clearly, mass spectral peaks containing both Fe and
CN are present, which could be used to differentiate Prussian blue from other blue or
iron-containing pigments.

Figure 4.3 shows the positive and negative ion LD mass spectra of authentic ferric
ferrocyanide that was purchased from Aldrich, suspended in water and applied to
HammerMill Fore MP paper. Characteristic peaks which may be indicative of the
compound in some way, but do not contain iron, are again observed at mlz 72, 88, and
104 in the positive ion mass spectrum. Again, note the presence of the relatively high
mass peak at m/z 575. Negative ion LDMS is clearly the method for determining the

presence of Prussian blue, with the same ions observed as in Figure 4.2b.

116

l

k.» Fen-m5: 0 m .. 1..
O >~ 1 m
N oflmw~mvm :0“

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

a)
39
>, 23
.4:
1’1"
3 245
.5
Q)
.2
33' 129
g 56
88 173
261 337 413
l I “I l | L j 575
1‘“ _.f.-_.11 1111 11111
0 118 236 mlz 354 472 590
134 242 b)
>5
4: *—
3 x4
8
.5 150 258
Q)
.2
E 216
Q
a 206 274
108 166
11 1 Ll ,1 1 .
100 150 200 250 300
mlz

Figure 4.3: The LD mass spectra of ferric ferrocyanide (Aldrich) from an aqueous
solution on paper; a) the positive ion mass spectrum; b) the partial negative
ion mass spectrum.

117

. 5_.m:3:— «Zea—e:

4 ufis

>1. 15:1... =~ a; .a :31

. .1.4 .‘s

Fig

575

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 4.4: The LD mass spectra of Prussian blue (Niji) from an aqueous solution
on paper; a) the partial positive ion mass spectrum; b) the partial
negative ion mass spectrum.

118

 

3)
577

3::
'7.

fl
8

fl
_

G)

>
'5
3 ll

4)
M
609
64
1119111111,,,,1.1?.,-
550 575 600 mlz 625 650 675
574
b)
3*
.5 576
O
'5 610
8 60.
a:
645
m 675
711
. . 1.11 . . .
550 585 620 655 690 725
mlz

Wat
disc
The
cop;
C 3;."
of a

posit

SPEC]
deso;
for d

Pruss

idem
of fer
50me
One u
SPGClr
the an
L0 m
have s

The second artist also employed Prussian blue, but a different brand (Niji
Watercolors). The positive and negative ion spectra are shown in Figure 4.4 and were
discussed in detail in Chapter two. Obviously the spectrum is not of ferric ferrocyanide.
The peak at m/z 575 in the positive ion spectrum represents the molecular ion, M", for
copper phthalocyanine. The isotope distribution is consistent with the formula
C32N8H16Cu. A response in both positive and negative ion modes indicates the presence
of a neutral dye, of which copper phthalocyanine is an example. It forms an M+' peak in
positive ion mode, and an [M-H]' peak in negative ion mode (m/z 574).

Again, if a laboratory was equipped with only the capability of atomic
spectroscopy, identification of copper phthalocyanine would have been difficult. The
desorption and ionization of the intact dye provides the required molecular information
for distinguishing these two watercolors that were both sold under the same name,
Prussian blue.

Now that the peak at m/z 575 in Figures 4.2a and 4.3a has been positively
identified as copper phthalocyanine, one must question its presence in the mass spectrum
of ferric ferrocyanide. While it is plausible that artist pigment manufacturers might blend
some copper phthalocyanine with ferric ferrocyanide to alter the pigment’s appearance,
one would not expect a chemical company to do so. Positive and negative ion mass
spectra of the paper (Figures 4.5a and 4.5b), reveal that the cationic peak at mlz 575 and
the anion species at m/z 574 originate from the paper and not from the sample. The paper
LD mass spectra were more complex than initially anticipated, however experiments
have shown that all of the peaks, excluding the peaks representing copper

phthalocyanine, are usually masked by the paint sample covering the paper. The m/z 574

119

 

the .
emp
been
pulp

samp

that 11'
53mph
This is

detecti.

an Oil p
using a;
the Micl
aPPTOXiI
the Vials
box 3131)::

unkIlOWn

peak in negative ion mode was masked by the sample in Figures 4.2b and 4.3b, however
the peak was observed in other spectra not presented here. The paper industry commonly
employs copper phthalocyanine in the paper manufacturing process. The pigment has
been used in paper mass coloration, paper surface treatment, and as an additive to paper
pulp (23). Thus, the appearance of the peaks representing copper phthalocyanine in the
sample of pigments on paper, is not unexpected.

An extremely dilute aqueous solution of copper phthalocyanine was prepared, so
that the solution was visually clear. The spectra (not shown) obtained from luL of this
sample, dried on a gold plate, revealed that the pigment was still detectable using LDMS.
This is encouraging, since it demonstrates that the sensitivity of LDMS allows for the
detection of the pigment, even when its presence is not apparent.

When questions arise concerning colorant composition, the subject is more likely
an oil painting than a watercolor painting. In the next example, an oil paint was prepared
using an “old pigment” and linseed oil. The pigment was found in a box of pigments in
the Michigan State University Chemistry Department archives. The box consisted of
approximately 100 vials of old inorganic pigments, in a dry powder form. Examples of
the vials are shown in Figure 4.6. Images in this dissertation are presented in color. The
box appears to be a pigment distributor’s product case. The age of the pigments remains

unknown, however there were clues indicating that the box is from the early 19008.

120

.3523:— 102.53%

 

 

4». a 1:35 93:50“

Plgllr

39

23

Relative Intensity

 

 

 

Relative Intensity

26

 

0

 

 

 

 

 

 

 

a)
575
63 191
91 372
13
I- 35131--. 1111111 1
200 m/z 400 600
115
b)
204
293 574
513
200 m/z 400 600

Figure 4.5: The LD mass spectra of paper (HammerMill Fore MP); a) the positive ion
mass spectrum; b) the negative ion mass spectrum.

12]

“613"

green ;
to dry.

these a
suspen.
suppon
(Figure
iron. T
ferrocy.
that [he
numero

are rela'

Fig

“613”

A green pigment, designated “013”, was selected for LDMS analyses. The dark
green powder was suspended in linseed oil, spotted on a gold sample plate, and allowed
to dry. Very informative positive and negative ion spectra of G13 were obtained and
these are shown in Figure 4.7. Similar spectra were obtained from the pigment
suspended in linseed oil and applied to paper. Linseed oil alone, on both paper and metal
supports, does not yield ions in the UV LDMS experiment. In the negative ion spectrum
(Figure 4.7b), a group of familiar peaks are noted at m/z 134, 150, and 166, containing
iron. The isotopic distributions are consistent with our previous assignments for ferric
ferrocyanide. Considering that ferric ferrocyanide is a blue pigment, one might presume
that there is also a yellow pigment present in the mixture. Fortunately, there are
numerous other peaks with characteristic isotopic patterns present in the spectra, which

are related to a yellow pigment.

   

Figure 4.6: Examples of pigment vial from the case of pigments found in the
Michigan State University’s archives.

122

.IiuQF-AH‘F-h anximh-u—Ahz

 

43.523 =~ 4.13.550“

104

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

3)
88
5
8 120 PbK
3 247
Pb0K+
'5 Pb“ 263
° 208
g 72
7, 430
9‘ 447
all. 530 754
111111121 J i 11111111 1 A; i J.—
100 300 m/z 500 700
200
b)
>, 100
3:
m
G
3
.5
Q)
.2
E
Q)
.. if”
1 ' ;WL
50 250

 

Figure 4.7: The LD mass spectra of G13 in linseed oil on a gold sample plate; a) the
partial positive ion mass spectrum; b) the partial negative ion mass spectrum.

123

208
100

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Pb+
a) PbK+ b)
E Pb+ PEEK 50 30‘ 207
g 204 J
I Pb0+ 1 J
3: .
200 225 ml: 250 275 100 530 532
C) m d) szC r0; 531
b 532
g f 5” i so 523
g 528
1: m L 526 527 I 5i” 53;
m l J L , 1
522 A 5211 m 584 A 540
100
e) 100 D 1” Cr03
g 1
g so
5
a 101
II: as m
A l J\ 9' T1 102
04 1110 106 1 l J
mlz

Figure 4.8: Enlarged portions of the positive and negative ion LD mass spectra of G13
in linseed oil on a gold plate; (a, c, e) experimental data; (b, d, f) theoretical
mass spectra.

In the positive ion spectrum, Figure 4.7a, there are numerous isotopic patterns
consistent with the presence of lead. An enlarged portion of the spectrum is shown in
Figure 4.83. The series begins at m/z 298, representing Pb+ ions. The isotopic pattern
compares well with the theoretical distribution of Pb isotopes (Figure 4.8b). The clusters
of peaks 16 and 39 mass units higher are the result of additions of an O atom and a K

atom, respectively. The additions do not appreciably alter the isotopic patterns. Lead

124

chromate is a very popular yellow pigment. Analysis of higher mass peaks reveals the
presence of ions containing both lead and chromate. The peaks at m/z 530 represent
Pb2CrO4+ ions (Figure 4.8c). The additional Pb atom, as well as the presence of 8 Cr
atom, produces a predictable and characteristic isotopic pattern for this ion (Figure 4.8d).
The identification of these peaks, along with those at m/z 754 representing Pb3Cr05+
ions, were confirmed by comparison with theoretical isotopic patterns. This allows for a
confident identification of the ions formed. The remaining peak identifications are listed
in Table 4.3. Additionally, confirmatory ions for the presence-of the chromate salt can be
found mixed in with the ferric ferrocyanide negative ions (Figure 4.7b). In fact,
chromium oxide ions appear as the most intense peaks in the mass spectrum, with peaks
at m/z 100, 200, and 300 representing 003', Cr206', and 0302, respectively. An
enlarged portion of the mass spectrum displaying the CrO3' isotopic peaks is shown in
Figure 4.86. The pattern compares well with the theoretical isotopic pattern in Figure
4.8f. An itemized list (author unknown, shown in Figure 4.9) and a description of the
contents of the vials (Figure 4.10) accompanied the box, in which this pigment was
found. The contents of 013 were listed as 66.7% PbCr04 and 33.3% Fe4[Fe3(CN6)]. The
mass spectra obtained were consistent with the information provided, that the pigment

was a blend of two pigments.

125

G13: Positive and Negative Ions

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

*m/z Assignment
208 Pl)+
224 P130+
247 PbK+
263 PhD K+
308 PbCrO3+
347 PbCrO3 K+
363 PbCrO4 K+
430 Pb20+
447 Pb202H+
5 30 13b2CI‘O4+
754 Pb3CI‘05+
100 Cr03'
1 34 Fe(CN)3'
150 Fe(OCN)(CN)2'
166 Fe(OCN)2(CN) '
1 82 FC(OCN)3'
184 (31205-
200 Cl’206-
284 C1308-
300 CF309-

 

* m/z value of most abundant isotopic form

Table 4.3: A list of the identified positive and negative ions for G13 in linseed oil.

126

I046 04 P4 Green ‘
#948 is;
. 7947 ' I '

We; Y01.Btr.

‘ ‘1 . .'

“use Ink!

Dutch ' ' :18 ‘
Y21125ernn ne
Carmine W ‘3?
,2 Corner. 1’2" ‘Ilnx

119 de rpee P'. P;
Sihaa ore flaunt '

Bronze I1"

I)
’ =\.1§W5:'
J ‘11. r.

5g. 1... we Meg-‘3

 

Figure 4.9: An itemized list of the contents of the box.

127

 

Figure 4.10: A portion of the list describing the contents of the pigment vials found in
the box of pigments.

When comparing the negative ion spectra of Prussian blue and 613, differences in
a few of the isotopic peak patterns are evident. For instance, the peaks at m/z 134 in
Figure 4.7b resemble that expected for the presence of two Cr atoms. The peaks could
actually be an overlap of isotopic peaks for Fe(CN)3' and CerO'. The peaks at m/z 184
are also unique. Again, this may be an overlap of isotopic peaks for two ions:
Fe(OCN)3' (m/z 182) and 0205' (m/z 184). The peak at m/z 180 is still unidentified,
however it appears to represent another Fe-containing ion.

The analysis of the G] 3 pigment was very different from the analysis of Prussian
blue, in that the pigment was suspended in a transparent medium (dried linseed oil) on a
gold sample plate, as opposed to a dry watercolor paint sample on paper. This
demonstrates the versatility of LDMS to detect different types of paint samples on a

variety of surfaces, and in a complex matrix. Additionally, the Prussian blue pigment

128

was a single component, whereas 613 was a mixture of two pigments. Even though the
613 pigment is primarily composed of lead chromate, which would be expected due to
the high tinting strength of Prussian blue, both components are still easily detected. This
demonstrates that LDMS may be used to detect the presence of multiple pigments in a
mixture, which as stated earlier, can be a limitation for other analytical methods in the
analysis of a complex paint. For example, Figure 4.11 shows the energy dispersive x-ray
(EDX) spectra of Prussian blue and chrome green, which is a mixture of Prussian blue
and lead chromate (5). In the EDX spectrum of Prussian blue (Figure 4.11a), there are
peaks present representative of Fe, however there is no evidence that it is an iron cyanide
pigment. In the EDX spectrum of chrome green (Figure 11.b) peaks representative of Pb
and Cr are noted, with some Ti and Si impurities. However, the absence of Fe is noted.
This is a specific example which illustrates a case in which the concentration of Prussian

blue was so low, that it did not meet the detection limit of this instrument.

129

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 4.11: Energy dispersive x-ray spectra; a) Prussian blue; b) chrome green.

130

The l

is use

idem

an il

4.12

M?

Y0

ma

Yc

at

$2

The LDMS analysis of two illuminated manuscripts

LDMS is introduced here as a potential tool for the detection of art forgeries, and
is used to determine if the two documents in question could be authentic, based on the
identification of the colorants present. Samples analyzed using this technique included
an illuminated page of the Koran, reported to be approximately 400 years old (Figure
4.12), and a decorated page of a Hebrew manuscript allegedly from the 18003 (Figure
4.13). Both were obtained from the Sadigh Gallery (303 Fifth Avenue, Suite 1603, New
York, NY). Pigment standards that were analyzed in this work included vermilion,
malachite, verdigris (Sinopia, San Francisco, CA), and orpiment (Kremer Pigments, New
York). The vermilion sample was suspended in a 50/50 solution of water/gum Arabic,
applied to standard printer paper, and dried. The malachite, verdigris, and orpiment
samples were suspended in linseed oil, applied to paper, and dried. Positive and negative
ion mass spectra were also obtained of the standard gold 100-well plate to serve as an

example of authentic gold.

131

~me «ml» ‘94P?» ~.
I.“ IVW“,~‘QQ\»“MV~ “I
We w~~<~r~ mm 4’“
(.._ ‘ --'r~t‘~s~w%\“fn
‘ . iflgm‘v-M g‘tifixafiu}

Month»

”4:1,..‘9337 . .

 

Figure 4.12: A page of the Koran, allegedly from the 17“ century.

132

 

Figure 4.13: A page from a Hebrew manuscript, allegedly from the 18003.

133

Example 1

A pen stroke from a red Bic® pen, on paper, was introduced into the instrument
and irradiated by the nitrogen laser. The mass spectrum shown in Figure 4.14 was
obtained. This represents the positive ions formed by laser irradiation. The peak at m/z
443 represents the intact dye molecule Rhodamine 6G. This is a positively-charged
(cationic) dye molecule with the formula C23H31N203+. With each C weighing 12 atomic
mass units (amu), H weighing 1 amu, etc., Rhodamine 6G molecular ions weigh 443 amu.
This is a very commonly-encountered red dye. Since it is naturally positively-charged, a
strong peak representing the intact molecule is detected in positive ion LDMS. In the
negative ion LDMS spectrum, no peak is present representing the dye (spectrum not
shown). In our experience, we have seen that negatively-charged dyes only yield

negative ions, and neutral dye molecules yield both positively and negatively charged
forms of the intact dye (24,25). In Figure 4.14, the peak at mlz 23 represents N a+ ions

and that at mlz 39 represents K+ ions, commonly found in most samples.

134

443

   
  
  
  

 

 

 

 

 

 

 

39
‘3.
3, C—OCHQCH3
'2
3 CH3 CH3
5 +
g 23 CH3CH2NH 0 HCH2CH3
H
.2
G)
a:
415
A M~~ :.-u,' HALL -4‘7 A .3, i #44 - M. .: LJJL a
0 100 200 300 400

Figure 4.14: The positive ion LDMS spectrum of rhodamine 6G. The structure of the
dye is shown. The singly charged dye molecule has a monoisotopic
molecular mass of 443.

Example 2

In contrast, if a metallic gold (Au) surface, an inorganic sample, is irradiated with
the UV laser, both positive and negative ions are formed. The positive and negative ion
LDMS spectra for gold are shown in Figures 4.15a and b, respectively. In Figure 4.15a,
there is a peak at m/z 197, which corresponds to the m/z value for a singly charged gold
atom. Peaks representing clusters of two and three gold atoms, as singly-charged

chemical species, are also observed at twice and three times the mass of atomic gold.

Similar ions are observed in the negative ion spectrum (Au' through Au3'). Unlike an

135

orga
repr:
proc
met.

golt

I q
{A

SUC

ICE

St:

organic dye, which exists as a collection of individual, discrete molecules, a gold surface
represents a three-dimensional array of atoms. Apparently during the laser desorption
process, not only atomic ions but small ionic clusters can easily be formed. Not all
metals can form stable negative ions. Chemists are more familiar with positively-charged
gold atoms than negatively charged gold atoms. Gold atoms have an electron affinity of
2.3 eV, indicating that they can stabilize a negative charge (extra electron). Some metals,
such as titanium, have much larger electron affinities (7.9 eV), and therefore may more

readily form negative ions. This is not the case with all metals. For example, scandium

(Sc) has an electron affinity close to zero (0.18 eV) indicating that Sc' may not be a
stable species in the gas phase (26). Using such considerations, LDMS spectra can
provide information on species present in any type of target which can 1) absorb the UV
light fiom the laser and 2) be desorbed and ionized, to yield singly-charged gas phase

ions for subsequent MS analysis, in this case, by a time-of-flight mass spectrometer.

136

llllllllllll
human-9:: 03.5.9:

 

. 394 (Aug)

 

 

 

 

 

 

 

 

 

 

197 (Au+) a)

3’
E 591 (Aug)
E

G)

>
a:

a
'3
a:
L “"‘“‘1i“ ““i A M i“ i “ i
150 250 350 m/z 450 550 650
197 (Ad)
b)

3'

'5":

§

.5

d)

>

‘3

g 394 (Auz')

591 (Aug)
‘ ‘ li‘ i ' ‘ i ‘ - r ‘ l
150 250 350 mlz 450 550 650

Figure 4.15: LDMS spectra of the ions formed by UV laser irradiation of a gold
surface. a) positive ion LDMS spectrum; b) negative ion LDMS

spectrum.

137

Figure 4.16 shows a portion of the illuminated page of the Koran from the 16003
that was analyzed. The text is written in black ink with red diacritical marks, and the
page is illuminated using gold, red, and black inks. For the analysis, both positive and

negative ion LDMS spectra were obtained for each of the colors used in the document.

  
  
 
 
 

Gold

' 0.‘

?
“Green” 1511. )

a“,
-\
. “I ,3

3‘. WW;

Figure 4.16: A portion of the page of the Koran used in this study.

Red

The first color on the Koran sample analyzed was the red pigment. Information
sufficient to establish its identity was present in the negative ion spectrum (Figure 4.17a),
so that will be the focus of this discussion. A series of peaks separated by 32 mass units
(m/z 32, 64, 96, 128, 160, 192) dominate the spectrum. In mass spectrometry, a mass of

32 amu most likely represents two oxygen atoms (16 amu each) or one sulfur atom. The

latter is more likely, thus the ions from m/z 32 to 192 represent sulfur cluster ions from S'

138

through S6". Sulfur does not suffice as a colorant alone, however its dominance in the

spectrum indicates that it is a major elemental component of the pigment. Seventeenth
century artists used a variety of red pigments, including red lead, several red lakes, and
vermilion (HgS) (27). We have previously analyzed lead-containing pigments such as

PbCrO4; lead and lead-containing ions are commonly formed from such salts (25). None

are observed here.
Lakes are a special type of pigment having both an organic and an inorganic
component. For example, when the organic dye carminic acid is precipitated onto

alumina, A1203, the red pigment carmine lake is produced. We have studied this lake

previously, and its laser desorption mass spectrum has been reported (25). Vermilion, or
the naturally occurring mineral analog, Cinnabar, is a natural possibility to investigate,
since it is a metal sulfide. A closer look at Figure 4.17a reveals a number of peaks that
are formed with m/z values in the 300—600 range of the spectrum, with relatively low
intensities. This region of the spectrum, in expanded form, is shown in Figure 4.17b.
The clusters of peaks are very important, because they represent isotopic variants of the

ionic species formed.

139

£285 95.23.

5.55: =~ 9» .5 5.9%

Fig

 

96 S3-

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

100 202 Hg 3)
200
1”
50 201
198
5: 204
'53
5 S4. 1
.5 128
Q)
.2
E
Q)
a:
82'
8'64 85' S6'
32 l 160 192 363 426 490
.i..l.L.._ .1 tr- 3 'LTLLLA IE
0 200 m/z 400 600
363 HgSS' 426 HgS7'
. b)
Hgs9'
490
3‘
'55
:1
8
.5
Q
0:
a - H S '
.33 H gS6' HgSs 532210
‘z 458
. 394 I
l l j; '
illllil .ln‘ Illl’I; I“ ll. ll
350 400 450 500

 

Figure 4.17: a) Negative ion LD mass spectrum of the red ink/dye region of the Koran
sample. b) Expanded view of the higher mlz portion of the spectrum.

140

 

 

In mass spectrometry, all of the isotopic forms that exist in nature for each
element are realized. For example, in Figure 4.17a, all of the Sn' ion peaks have an
additional peak 2 atomic mass units higher. These are due to the fact that, while most
sulfur atoms exist with atomic masses of 32 amu, 5.4% of all sulfur atoms exist as 34S,

weighing 2 amu more than the most abundant isotope, 328. This is seen in the mass

spectrum. The distribution of isotopes for mercury is much more complex - 6 isotopic

forms, from 198Hg through 20A’Hg, are found in nature. The theoretical distribution is
shown in the inset in Figure 4.17a for one mercury atom. The number and pattern of the
relative abundances of the isotopes clearly indicate that the peak at m/z 363 contains a

mercury atom. The m/z value is consistent with the peaks around m/z 363 representing
the ionic species HgSS'. Ionic species representing HgSS' through HgSlO' are observed
in this spectrum. In some cases, the clusters of peaks do not exactly match up with ionic

formula of the type ngS '. The probable cause is the fact that there are also species

y
with very similar mass being formed, containing a single hydrogen atom. For example,

the set of peaks around m/z 426 probably represent HgS7' and a small amount of
HgS7H'. This is not unexpected. Bumin and BelBruno (28) reported similar

observations in the formation of anSsz+ clusters from laser ablation of ZnS. The key

point is that the dominant ions in the negative ion LDMS spectrum of the red ink
indicated that the pigment present is a sulfide. Less intense peaks clearly indicate that it

is mercury sulfide. While it was not anticipated that the compound would form species

141

such as HgS7', this is part of what is learned in LDMS experiments - how components of

a pigment combine to form stable gas phase ions that can be detected.

In addition to the information that the mass spectrum'provides, a vermilion
sample was obtained from an artist pigment distributor, Sinopia. The vermilion was
suspended in a solution containing water and gum arabic, "painted" onto paper, and
analyzed. The same mass spectrum was obtained (not shown), confirming the
assignment.

Current methods for the identification of vermilion in art work include
microchemical tests (29-32), x-ray diffraction (33,34), emission spectroscopy (35),
infrared spectrophotometry (36—38), x-ray fluorescence (31), electron probe
microanalysis (31), and x-radiography (39). While microchemical tests are ideal for
indicating the presence of Hg and S, they are destructive. Additionally, x-ray diffraction
analysis can lead to a positive identification for HgS, however the sample must be pure,
as in many of the other techniques. Here, LDMS excels in that it is well-suited for the
analysis of mixtures, such as those found on a painting. Similar spectra (not shown) of
the naturally occurring HgS pigment (Cinnabar, Sinopia) were also obtained. It does not
appear that LDMS can distinguish between the Cinnabar and the synthetically
manufactured vermilion.

In summary, the major ions of the spectrum shown in Figure 4.17a indicate that
the red colorant is a sulfide, less intense peaks clearly indicate that it is a mercury sulfide.
Comparison of the spectrum with that of a known mercury sulfide sample identifies the
colorant as HgS, vermilion. This colorant has a long and extensive history, and its

presence is not unexpected for a manuscript from the 16003. Figure 4.17a clearly shows

142

that the red colorant is not a modern organic dye such as a red Rhodamine dye (Figure

4.14).

Gold

If the laser is moved to irradiate a gold portion of the manuscript page, the
spectrum changes. The positive and negative ion mass spectra obtained from this region
are shown in Figures 4.18a and 4.18b. Clearly, the spectra do not suggest the presence of
elemental gold (Figure 4.15). This is not unexpected since chrysography (painting in
gold) was largely phased out in the 15‘h century, due to the high cost of gold in any form
(27). Thus, some other pigment or dye is present to give a golden, metallic color.

The positive ion LD mass spectrum, Figure 4.18a, shows four peaks separated by
32 mass units, again suggesting the presence of a sulfide. In the negative ion mass
spectrum, differences of 32 (e.g., m/z 138 to 170) and 75 (e.g., m/z 170 to 245) are noted.
An atomic mass of 75 corresponds to the element arsenic, As. In the 16003 a variety of
yellow pigments were being used as a cheap alternative to gold. These include lead tin
yellow, lead antimonate, and a variety of yellow lake pigments such as saffron lake (27).
The pigment orpiment has been cited as a popular choice of artists to serve as a substitute
for real gold (27). The name orpiment is derived from the latin auripigmentum, meaning
the color of gold. Alchemists described the pigment as a "handsome yellow more closely

resembling gold than any other color", and at one time it was believed to contain gold

(27,40,41). Orpiment is arsenic trisulfide, A3283.

143

2&2;

 

 

 

 

 

 

 

 

 

 

 

a)
3
5 AS383 A538;
.3 320-“ 352.65
3::
.53
i’ .
AS382
288.67
250 300 mlz 350
A382'
138.82 b)
E
.5
.g AS384-
3 352.65
a A883. . AS385.
17°79 A3283 340.62 384.63
245.71 A5254 A5383.
. 1 277376 .3294” .
I r l I" l
135 185 235 m/z 285 335 385

Figure 4.18: a) The partial positive ion and b) the partial negative ion LD mass spectra
of the gold region of the Koran sample.

144

In mass spectrometry, identifications are made based on the isotopic composition
of gas phase ions as determined by 1) the m/z value of the ions, 2) the isotopic variants of
that ion, and occasionally 3) the accurate mass of the peaks. For many kinds of mass
spectrometry, the m/z value can only be determined to the nearest m/z value ("unit mass
resolution"). Both As and S atoms contain negative mass defects. That is, while the

nominal mass of an A3 atom is 75, the exact mass is 74.9216 amu — a difference of

-0.0784 amu. A similar situation exists for S. The 328 atom has an atomic mass of
31.9721 mass units — a difference of -0.0279 amu. When several of these atoms are
combined, the deviation from the calculated nominal mass becomes significant. For
example, consider the ion A33S+. Using the nominal masses of As and S, its molecular
mass is 257 amu. In the accurately calibrated mass spectrum (Figure 4.18a), there is a
peak at mlz 256.66 (theoretical, m/z 256.74) representing these ions. The difference
between the nominal and experimental values (-0.3 amu) can be accounted for when the
exact masses of As and S are considered.

In both the positive and negative ion mass spectra, the mlz values are consistent
with the presence of As and S. In the positive ion spectrum, ions with the general

formula A33Sn+ (n = 1-4) are generated. In the negative ion spectrum, ions with the
general formula AsmSn' are detected. To confirm that the pigment used here is actually

orpiment, positive and negative ion LD mass spectra were taken of the actual pigment
obtained from Kremer Pigments. Again, the spectra of the unknown match the spectra of
the standard (not shown).

The appearance of orpiment on the manuscript is not surprising. It has been

identified in Armenian, Arab and Persian manuscripts from the period in question (42).

145

Again, the laser desorption mass spectra clearly indicate that the compound is a sulfide,

and that it is arsenic sulfide.

Black

Black inks are traditionally very difficult to chemically identify, since (prior to the
introduction of modern pen inks) the colorants used were very complex mixtures of
chemical species, not single compounds. One example is carbon black, 3 complex
mixture of carbon and hydrocarbon particles. A second example is iron gallotannate ink,
also a mixture of many compounds, since it is made from an extract of gall nuts, which
are shown in Figure 4.19 (43). Galls are created by parasites (aphids, flies, wasps) to
house their eggs, in various types of vegetation. They contain a high concentration of
gallotannic acid (penta-gallolylglucose), which is extracted to make Fe-gall ink.
Gallotannic acid is a glucose molecule (Figure 4.20a) with up to 5 gallic (3,4,5-
trihydroxy-benzoic acid, 170 amu) (Figure 4.20c) and/or digallic acid groups (322 amu)
(Figure 4.20b) attached. The gallotannic acid is dissolved in a solution of water and gum
Arabic. Gum Arabic is a water soluble vegetable gum from the Acacia tree, which serves
as suspension agent for insoluble pigment particles. Gum Arabic is also chosen to adjust
the ink’s viscosity (42).

Upon the addition of vitriol (FeSO4) to the mixture, a water soluble ferrous
tannate complex is formed. The solubility factor is crucial, since it allows the ink to
penetrate the paper’s surface. The initial ferrous tannate complex is colorless.
Consequently, a natural or synthetic dye (logwood, indigo, aniline dye) may be included

in the ink mixture in order to increase the visibility of a freshly prepared ink. Once the

146

ink is applied to paper, it is exposed to oxygen in the air. An insoluble ferric tannate
pigment is precipitated, resulting in the dark blue/black permanent Fe—gallotannate ink
(44)-

There is some dispute over the true structure of the ferric tannate pigment. The
chemical structure was investigated by C. H. Wfinderlich and C. Krekel (45-47). They
agree that the Fe 11 ions of the vitriol react with both gallic acid and gallotannic acid.
However, Wiinderlich believes Fe reacts with the 3 hydroxyl groups of gallic acid and
with the carboxyl group, creating a three dimensional structure. Thus, the color
formation in the ferric tannate complex is due to the benzene ring and oxygen/iron bonds
(45,46). Krekel’s research suggests that the black pigment is an iron pyrogallol complex
instead of an iron gallic acid complex (47). The exact nature of the ferric tannate

pigment has not been confirmed.

 

Figure 4.19: Gall nuts.

147

The ingredients for making iron gallotannate ink were obtained, and a batch of
iron gallotannate ink was prepared in our laboratory. Positive and negative ion LD mass
spectra (Figure 4.21) were obtained of the freshly prepared ink on paper. Table 4.4 lists
all of the different possible combinations in which gallic and digallic acid molecules
could attach to gallotannic acid. The tables provide the m/z values that we were
expecting in the mass spectra. These values do not take into account that there may be Fe
atoms attached, simply because the smallest possible ion formed would have a m/z value
of 390 (336 amu + 54 amu), and peaks of significant intensity were not observed in either
the positive or negative ion mass spectra.

Only a few peaks have been identified thus far, which are related to gallic acid. In
the negative ion mass spectrum (Figure 4.21b), a peak is present at m/z 169, representing
deprotonated gallic acid molecules. The peaks at mlz 152 and 124, represent neutral
losses of a hydroxyl group and a carboxylic acid group from the deprotonated molecule,
respectively. The negative ion mass spectrum (not shown) of gallic acid (Aldrich)
suspended in gum Arabic and analyzed on a gold sample, coincided with these
assignments. Additionally, there is a relatively small peak at m/z 321, which may
represent deprotonated tannic acid molecules. The remainder of the peaks in the spectra
of the homemade iron gallotannate ink remain unassigned. At this point, the spectra will
be used as a fingerprint to compare with the spectra obtained of the black ink on the page

of the Koran.

148

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

a) H b) C)
HO HO I 00
H ——-«l— OH
RO-—I H O ( (:0 no 0H
H———4 CR “0 0 or
H
CHZOH com
0 O O O 0 1i A i
(Gallotanmc Acnd) (Dlgallic Ac1d) (Gal C c d)
Figure 4.20: Structures of gall nut components.
MW
R1 R2 R3 R4 R5 (D3) . .
G = Gallic ACld
G G G G G 940 . . .
D = Dlgalllc Ac1d
G G G G D 1092
G G G D D 1244 l 1 MW
R R R R R
G G D D D 1396 1 2 3 4 5 (Da)
G G D D D 1548 D D D D H 1397
D D D D D 1700 D D D H H 1112
G G G G H 789 D D H H H 827
G G G H H 638 D H H H H 542
G G H H H 487 G G G D H 941
G H H H H 336 G G D D H 1093
G D D D H 1245
G G D H H 790
G D D H H 942
G D H H H 639

Table 4.4: A summary of possible molecular weights of the components of

gallotannic acid.

 

 

 

 

 

 

 

 

149

 

57

 

 

 

 

 

 

 

a)
245
>5
H
'r'Z
E
0
H
E
E 191 261
H
.2
Q)
n:
164 209
65
96111 149 175
343 367 413
Mm... I. i .
50 130 210 290 370 450
(MM), mlz
169
(M-H-COOH-)' 301
If“
E,“ 279
o- 325
2 297
o 343
H
.5 277
Q 48 261
a: 303 321
:3 259 263 274 281
m 69 ll .1l 1.11L.. l _
250 270 290 310 330
36 . 152 r A a
108 191 301
279 325
. I" I i ‘ i 1.. . 41. L- . 1 1
0 100 200 mlz 300 400 500

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 4.21: a) The partial positive ion and b) the negative ion LD mass spectra of
homemade iron gallotannate ink on paper.

150

The positive and negative ion mass spectra of the black writing ink on the page of
the Koran are shown in Figures 4.22a and 4.22b. No identifications have been made in
the positive ion mass spectrum. Comparing the spectra in Figure 4.21 to the spectra in

Figure 4.22, the ink does not appear to be iron gallotannate ink. In the negative ion mass

spectrum, the lower mass range is dominated by anionic carbon cluster peaks such as C4‘

(m/z 48), C5' (m/z 60), C6- (mlz 72), etc., separated by 12 atomic mass units. While this

may appear to be convincing evidence for the presence of carbon black, it is not
necessarily the case. While infrequent, some aromatic organic compounds very easily

char when exposed to high power UV light, forming essentially a Cn skeleton. This

could be the case here. The carbon cluster anionic peaks may even evolve from a
component of the paper itself. Peaks are present in the spectrum shown in Figure 4.22b
which have been identified as CuClz', CuzCl3‘, and Cu3Cl4', based on the m/z values and
isotopic distribution analysis. Returning to the negative ion mass spectrum of the gold
pigment on the Koran (Figure 4.18b), the presence of Cu is also suggested by the peak at
m/z 341. This peak cluster has been tentatively assigned as CuA32S4'. Copper is not a
commonly used ink component. However, copper resinate has been cited as a pigment
used on manuscripts during this time period (48-50). This observation has not been well-
supported (51). Additionally, the incompatibility of orpiment with copper-containing
pigments was well known by this time (52), which could mean that a copper salt may
have simply been an impurity in the ink. The only conclusion which can be made for
certain is that the ink is not modern.

As a side note, a visual examination is sometimes employed to help distinguish

between carbon black and iron gallotannate ink. Carbon black ink is known to be

151

permanent and noncorrosive to the paper it is applied to. However, it is known to
smudge easily before it is completely dry. Iron gallotannate ink, on the other hand, does
not smudge, but is known to turn yellow along the edges. This is in fact due to the
deterioration of paper caused by the ink. When examining the black used to create the
manuscript, one might argue that both were used. When analyzing the text, there are no
signs of the paper degrading or yellowing around the ink, which suggests the presence of
carbon black. However, the black ink in the medallion of the page has caused serious
deterioration of the paper, which is evident on the back side of the page (shown in Figure
4.12). This was the only area in which paper damage resulting from the ink was noted,
which makes one wonder if a different type of ink was used to decorate the manuscript,
and another used to write the document. It is not known to be a common practice,

however it would explain the observation.

152

.323.

now:- 0> =3 Tav—

104

 

 

 

 

 

 

5‘
i’.‘
8
.5
3 120 155
a
5
Ta
9‘ 161
197
100 200
C6'
71:
C4'
48
84

3‘
E
3
.5

‘l
g 44F
'5
.9;
0
a: 26

36

 

 

 

 

 

0

96

108

 

 

257
284
321
429
300 400
m/z
Cule
133 C0203.
115 233

 

 

 

 

l
l

125
mlz

I A A A A; k I I I

 

 

512

Sill

 

375

Figure 4.22: a) The partial positive ion and b) the negative ion LD mass spectra of the
black ink of the Koran sample.

153

“Green”

Throughout the manuscript, there are several instances in which the character, as
shown in the portion labeled “green” in Figure 4.16, appears. A vertical line is written in
black ink and topped with a large dot of yellow ink. Sometimes the yellow dot appears to
be outlined in black ink or was applied over black ink. Occasionally, the dot appears
green. To determine if a fourth color was used, positive and negative ion LD mass
spectra were obtained of these “green dots”, beginning at the outer edge of the dot and
working in towards the center of the dot. The negative ion results (shown in Figures 4.23
a —c) were informative and suggest several possibilities concerning their origin.

In Figure 4.23a, the negative ion mass spectrum of the outermost edge of the dot,
peaks appearing at mlz 139, 171, 246, etc. strongly suggested the presence of the yellow
pigment orpiment. What is interesting is that a few additional peaks (mlz 234, 309, and
373) are noted as containing both Cu and As. The peak at m/z 341 has been assigned as
CuA32S4', and was the only Cu-As containing peak which also appeared in the negative
ion orpiment mass spectrum (Figure 4.23b). Moving the laser closer to the center of the
dot, the spectrum in Figure 4.23b is obtained. The orpiment peaks have disappeared, and
the spectrum contains peaks representing ions containing Cu and C1 atoms. Peaks
representing CuxCly' ions (mlz 133, 231, and 331), which were also present in the black
ink negative ion mass spectrum (Figure 4.22b), are noted. Additionally, dominant new
peaks appear at m/z 115, 124, 147, and 204. These peaks have not been positively
identified, however their isotopic patterns are consistent with the presence of Cu and Cl
atoms. Finally, when the laser is focused directly on the center of the green dot, the

spectrum in Figure 4230 is generated. The pure CuxCly' peaks disappear. The isotopic

154

patterns of the peaks present in the mass spectrum clearly represent ions containing
multiple Cu and Cl atoms, however the exact compound is not evident.

There are several green Cu-containing pigments, including copper resinate,
verdigris (Cu(CH3COO)2°[Cu(OH)2]3-2H20), malachite (CuCO30Cu(OH)2), and emerald
green (3Cu(Ast)2°Cu(CH3COO)2). The pigment emerald green contains Cu and As,
however the pigment was not discovered until the very late 18‘h century/early 19th century
(53), and therefore would not have been available at the time this manuscript was created.
One possibility is that the green dots were painted with a Cu-containing green pigment.
The appearance of the orpiment peaks may suggest that a green pigment was painted over
a yellow dot, however if this is so, it was not consistently done. Cu- and S-containing
pigments should not be mixed. The darkened or discolored dots could have been due to
the formation of black CuS. The green dots remaining may not have reacted at all or

reacted to a lesser extent.

155

Ass,-

 

 

 

 

 

 

 

 

 

1 39
a)
Cums.-
Ass
1 71 ’ 341
1:3" ..,.,. a...
”382. 246 cunts; ”as.' 353 ”33.4
2“ CuAss \309321 teams?”
234
Hg .1 13,. ., 1.1 3131.
1 1
b)

>3
H
0 Fl!
m
S
1:
1—1
0.)
>
'1:
Cd
—-4
82

011,01;
1
327 2.25. :‘L - -
1 15 ' _ f C)
1 31
1 47
1 2
1
1 O3
1 79
. 11,1. 1.
1 oo 1 60 220 200 340 400

 

 

 

mlz

Figure 4.23: a-c) Negative ion LD mass spectra of the “green dot” taken from the outer
edge of the dot, continuing to the center of the dot.

156

63 Cu+

 

 

 

 

 

 

 

125
>.
H
’17:
1:
d)
H
.5
G)
>
0:
.2
u
m
165
50 _
79
>4
1‘:
8
o
H
5 63
G)
>
0:
.2
e
m
97
I... ll -1
0

 

 

 

 

 

 

 

 

 

 

 

 

 

3)
902
4 Cu
368
3 Cu
303
327 880
448
227 641
240 430 mlz 620 810 1000
281 b)
279
61
277
35
346 357 383
270 296 322 348 374 400
281 627
279
181
221 361
707
.L . 993
200 400 600 800 1000

mlz

Figure 4.24: a) The partial positive ion and b) the negative ion LD mass spectra of
malachite (Sinopia) suspended in linseed oil and applied to paper.

157

Based on the presence of CuxCly' ions in the black ink negative ion mass spectrum
(Figure 4.22b), the discolored dots may have been the result of a reaction between the
black and yellow ink. This is most likely because occasional green spots are only
observed where both gold and black are in contact with each other. Positive and negative
ion LD mass spectra were obtained of the Cu-containing green pigments, malachite
(Figure 4.24) and verdigris (Figure 4.25) (Sinopia), suspended in linseed oil and applied
to paper. Peak assignments were made, when possible. Except for Cu+ ions, no peaks
were identified in the malachite spectra. However, isotopic pattern analysis was used to
determine the number of Cu atoms present in the ions represented at m/z 303 (3 Cu
atoms) and m/z 368 (4 Cu atoms). The important point here is that the spectra do not
match that of the green dot, lending further support to the latter explanation. Some
questions will remain unanswered, however the sensitivity and versatility of the LDMS

technique is demonstrated.

158

 

 

 

 

 

 

 

 

 

 

 

 

63 Cl!+ a)
[Cu2(CI-I3COO)2]+
244
3' 246
'51
r:
g [Cu3O(CH3COO)3]+
a; \ [Cu..0(CH3C00)..]n
1.: 65 [Cu3(CH3COO)3]+
a 384
§ 8.3 [C“(CH3C00)21N8* \368 [Cu,(cn.comsl*
Cu * 366 [CHICHfiOOLP
12% \ 6 l I 427 486506
99 204 248
I | . . 307 l ,
“‘41.. ‘ ‘ ' ‘l
50 145 240 m/z 335 430 525
181 [Cu(CH3COO)2]'
>.
H
“1'3
1:
G)
H
E
o 183
.2
*3 [Cu(CH,c00),]-
g 240 tCu,(CH,co0),]-
303
242
.1 h-.. f. 4 - 11111.1.-- . LF_ Y.1 ' .. ‘. 11.. 1 11..r
150 185 220 m/z 255 290 325

Figure 4.25: a) The partial positive ion and b) the partial negative ion LD mass spectra

159

of verdigris (Sinopia) suspended in linseed oil and applied to paper.

Hebrew Manuscript

Figure 4.26 shows a portion of the illuminated page of the Hebrew manuscript
from the 18005 that was analyzed. Our initial interest in this second example, a
manuscript leaf reportedly from the 18003, arose in part due to the claim that it was
illuminated in gold. Our analyses indicated once again that “real gold” was not used on
the document. However, unlike the Koran sample, many bright colors, including orange,
blue, and green, were also used to decorate the miniature. The text is framed with four
straight lines of different color: blue, orange, gold, and black. The authenticity of the
manuscript was suspect because of the brightness of some of the colors, and the very
different extent to which the various inks spread on the paper.

The analysis presented here focuses on the "framing lines". The positive ion mass
spectrum of the blue line is shown in Figure 4.27a. The peak at m/z 372 has been
identified in previous work as representing the cationic dye crystal violet (24,54). The
structure is shown in Figure 4.27a. The positive ion LD mass spectrum of the black line
is shown in Figure 4.27b. Again, the peak at mlz 372 is indicative of the dye crystal
violet. Crystal violet is a very common component of modern blue and black pen inks.
Should crystal violet be present at all? The information provided with the sample only
indicated that it was from the 18003. It has been reported that crystal violet was
introduced in the 18605 (55), and was used to create manuscripts in the 18705 (56), so its
presence alone is not inconsistent with its age. Of interest are the peaks at mlz values
lower than m/z 372, separated by 14 amu - m/z 358, 344, 330, and 316. These represent
degradation products of crystal violet, which have been extensively characterized

(54,57,58) and discussed in Chapter 2. Crystal violet degrades over time, and its

160

degradation products can be detected using LDMS. There are very few degradation
products present in Figure 4.27a, in contrast to Figure 4.27b, where they are abundant.
This suggests that the blue line has only been on the page for a relatively short period of
time, probably from a modern pen, while the black ink is much older. The different
degrees of spreading of the two lines, and the straightness of the blue line may be

consistent with it being added after the page was originally created.

Blue

., .
‘0» ‘

‘F;“ ‘ V gr,“

Gold
Black

‘3.

,.
,3'

White
Orange

Green

Brown

 

Figure 4.26: Portion of the 1800s page of an Arabic manuscript that was evaluated.

161

Several spectra of the black writing ink were also obtained (not shown). The
peaks representing crystal violet were present in the positive ion mass spectra at various
stages of decomposition. The peaks at m/z 107 and 109, representing Ag“ ions (also seen
in Figure 4.27b) were also present. Silver has been documented as an ingredient in old
ink recipes (44). Solutions of silver salts turn black upon contact with paper (reduction).
Since a silver nitrate solution used would have been initially colorless, a dye was
traditionally added to aid in writing.

There were many other aspects of the analysis of this document which suggested
that, while it may have been created in the 1800s, it had subsequently been altered. For
example, in a light blue "paint" used in the illustration, the dye copper phthalocyanine
was positively identified (Figure 4.28). The copper phthalocyanine dye "Monastral Fast
Blue" was developed by the chemical company ICI between 1935 and 1937. A green,
chlorinated copper phthalocyanine was developed and became an important dye in the
19503 (27). These are still widely used in modern oil paints. Thus, many aspects of the

second manuscript page support the proposal that the original work was “enhanced”.

162

372

 

 

 

 

 

 

 

 

 

 

 

CH3 a)
I _.
+NCH3 ~C|
a U
E
3
g |
.2
g NHCH3
358
. 34“ ,Li
100 160 220 mlz 280 340 400
372 b)
358
>2
“
'53
fl
3 344
.5
a) 322
.2
‘5
'6 330
9‘ 107/109
136 160
316
100 160”" ‘L‘L‘flz‘iii‘flg/‘imiziow 340 400

Figure 4.27: Positive ion spectra of a) the blue framing line, and b) the black framing
line of the Hebrew manuscript page.

163

575 575

 

 

 

 

 

 

 

 

 

>5
H
'5
= 2
8 l
s H.
4) i
.2 L 610
H LA. .
a a I r
T) 555 575 m/z 595 615 635
K 93 153
168 248
107 268 341 385
610
‘ " ‘ 1 u M 1 ~ L J‘
50 150 250 350 450 550

Figure 4.28: The partial positive ion LD mass spectrum of the blue paint on the Hebrew
manuscript.

164

The positive and negative ion LD mass spectra of the orange colorant are shown
in Figure 4.29. Identical spectra were obtained of the orange framing line and the orange
paint in the picture. The exact pigment has not been identified, however we can use the
information available in the mass spectrum to help characterize the colorants. The
spectrum is similar to some of the red organic watercolor pigments discussed in Chapter
2. There are numerous peaks present in the spectra, indicating either the presence of
multiple dyes (perhaps a red and yellow, to produce an orange shade) or the presence of
an orange dye with multiple azo linkages. The azo linkages have proven to be relatively
unstable, resulting in the formation of numerous fragment ions in a mass spectrum.
There appears to be at least one neutral dye (M1) present having a molecular weight of
370 amu. The characteristic overlap of the MK / (M 1+H)+ peaks for a neutral dye is
observed in the positive ion mass spectrum (Figure 4.29a). The peaks at m/z 393
[(M1+Na)+] and m/z 409 [(M1+K)+] lend support to this assignment. Additionally, a
complimentary peak at m/z 369 representing (Ml-H)’ ions is present in the negative ion
mass spectrum (Figure 4.2%). A second neutral dye (M2) may be present at m/z 340,
with Na and K adducts present at m/z 363 and 379, respectively. The protonated
molecule (mlz 341) is labeled in Figure 4.29a. However, it is unclear whether or not the
dye represented at m/z 337 in the negative ion mass spectrum (Figure 4.2%) is related.
Again, this is an indication that the document might be significantly newer than
proposed, simply because modern day organic colorants were not as readily available in
the 18003 as they are today. We may have expected to find peaks with unique isotopic

patterns indicating the presence of an inorganic colorant.

165

 

 

 

 

 

 

 

 

 

 

 

370 Mg

 

 

 

 

    

 

 

 

 

 

 

a)
107 168
>5
::
a 264
l
5 311
g (M2+H)+
:3 341
7, (M,+Na)+
9‘ 393
93 153 130 (M,+K)+
220 409
.. l
4_ i 1- -1
75 155 235 m/z 315 395 475
337 369(M1-H)'
>5
5:
g
H
.5
G)
>
O:
.9.
Q)
m
167
228 355
l u L
- 1+ u . . .A . A. , . A.
125 185 245 m/z 305 365 425

Figure 4.29: a) The partial positive ion and b) the partial negative ion LD mass spectra
of the orange paint on the Hebrew manuscript.

166

The positive and negative ion LD mass spectra of the light green paint in the
Hebrew manuscript are shown in Figure 4.30. Again the colorants have not been
identified, however there appears to be at least two neutral dyes present (M 1 and M2). M1
has a molecular weight of 341 (M 1+). The deprotonated molecule appears in the negative
ion mass spectrum (Figure 4.30b) at m/z 340. The second dye (M2) has a molecular
weight of 395 amu, and has an isotopic pattern consistent with containing two Cl atoms.
The deprotonated molecule appears in the negative ion mass spectrum at m/z 394. A
fragment ion is noted at m/z 171 in Figure 4.30b, which has an isotopic pattern indicating
the presence of only 1 Cl atom, most likely originating from M2. In summary, the
colorants are clearly organic modern day dyes, which sheds some doubt on the

authenticity of the document.

167

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

3)
.2.» M“
'5; 341
=
3 575
.5
v 268
>
O:
.‘S
33 M;
248 395
127153
93 335
1 3198 37 97
L 4.17 l
so 160 ' 27o mlz 380 * 490 F 600
(Mg-HT 394 b)
96
>6
1:
8
3 (Ml-H)-
.E 340
Q)
.2
*5; 171
'5
m
173
‘1‘ 1
. V.-L-L,_.tiv .‘ v.4. ‘nz
125 185 245 mlz 305 365 425

Figure 4.30: a) The partial positive ion and b) the partial negative ion LD mass spectra
of the green paint on the Hebrew manuscript.

168

A New MALDI Matrix?

Currently, the most popular method for identifying colorants on illuminated
manuscripts is Raman spectroscopy (59-62). Raman spectroscopy has the advantage of
being a nondestructive analytical technique, which can perform the analysis directly on
the document, without having to remove or extract a sample. However, organic
compounds are not easily detected in this experiment. Consequently, Raman
spectroscopy is limited to the analysis of inorganic colorants, for this application. A 16‘h
century Icelandic illuminated manuscript is shown in Figure 4.31 (59). Best and
coworkers have been able to generate Raman spectra of several colorants, some of which
have been presented here, including vermilion and orpiment. As with IR spectra, Raman
spectra are typically used as fingerprints, and therefore require a library of standard
spectra for comparison and identification of unknown samples. LDMS generates
interpretable spectra, which do not require a library of reference spectra for comparison.

Let us consider a pigment, which has not been discussed yet, such as Fe203 (red
ochre). Positive (not shown) and negative ion (Figure 4.32) LD mass spectra were
obtained of Fe203 suspended in linseed oil and applied to paper. There was no evidence
of iron or iron oxide ions in either of the spectra. However, the negative ion mass

spectrum contains important information.

169

 

Inttnlttv —)

 

lnunulv ——)

Vermilion

 

 

 

 

 

«w an aw
r w \ c.
'3
r
.. t-
'3.-
1
. ,"i
(k R
t
'1' 1“.
.
l .
.
' fl Orpiment
45— no an 1w

Intensity —)

 

 

 

 

;;;;;;;;;MvA/m
WWW
400 300

 

 

 

 

 

Realgar

1
AW 1.
Mi m

Intensuv —)

 

200
Azurite
S H
3
E .
E
coo 700 600 500 400 300 200
.y
.
,
e
.1 ‘V
‘r.’
Q":
\‘.
.¢,
. ,2;
Kg. ,i‘
‘ -
b
Bone NH 1 t8
7
3
E
C
o
E
130 I?” II” 1000 M I)

 

 

 

Figure 4.31: A 16‘h century Icelandic illuminated manuscript.

I70

Up to this point, there has been no mention of the vehicle (linseed oil, in this
case), simply because it never made its presence known in a LD mass spectrum.
Additionally, we did not necessarily expect for it to be able to generate ions in this
experiment. Linseed oil is transparent, and consists primarily of three unsaturated acids.
These include Iinolenic (278 amu), linoleic (280 amu), and oleic acid (282 amu), having
three, two, and one degrees of unsaturation, respectively. The structures of the acids are
provided in Figure 4.33. The remainder of the mixture is composed of saturated acids,
which include stearic acid (284 amu), palmitic acid (256 amu), and myrtistic acid (228
amu). Returning to the negative ion mass spectrum of Fe203, there is an interesting
distribution of peaks around m/z 280. We believe that the peaks in the mass spectrum
represent the deprotonated unsaturated acids of linseed oil (m/z 281 - oleic acid, m/z 279
— linoleic acid, m/z 277 — Iinolenic acid). The peaks in the higher mass region at m/z
896, may represent a polymerization product, since the m/z values are equivalent to the
sum of approximately three of the unsaturated acids.

Let us now consider the two experiments which have been presented in this
chapter involving Fe-containing pigments (ferric ferrocyanide and red ochre), which are
illustrated in Figure 4.34. In the first experiment, Prussian blue (ferric ferrocyanide) was
suspended in water and analyzed on paper. Several FexCNy' ions were observed in the
negative ion mass spectrum. The same results were observed when ferric ferrocyanide
was suspended in linseed oil and analyzed on paper (spectra not shown). There was no
evidence of linseed oil. In the second experiment, Fe203 was suspended in linseed oil
and analyzed on paper. Ions representing linseed oil were generated in the negative ion

mass spectrum, however there was no evidence of the colorant. At this point in time, we

I71

can not explain why we obtain such different results for these two experiments. Factors
such as particle size, ionization potentials, and heats of formation values were considered,
however there is no clear cut explanation why in one instance, colorant ions are

generated, and in the other, vehicle ions are generated, but not both at the same time.

 

 

 

 

 

 

 

 

 

 

 

2:81
896
281

>3

H
O-

E 279

a)
E

Q 1

>
O:

a 282
T) 277

°‘ MW

270 275 280 285 790 M 295

0 400 800 :1/2 1200 ' 1600 " 2000

Figure 4.32: The negative ion LD mass spectrum of red ochre in linseed oil on paper.

172

Linseed Oil:

ll
CH2 (CH2)5 cr-I2 —C—OH

< H. g 0 Linolenic Acid: 278 Da
CH3 c

Linoleic Acid: 280 Da

||
CH3(CH2)30H2(CH= CH CH2)2 CH2 (CH2)4 CH2 —-C—-OH

O

H . .
CH3(CH2)5CH2\C_C,CH2 (CH2)5 CH2 ”C‘OH OICIC ACld: 282 Da
H’ _ ‘H

Figure 4.33: The structures of the primary unsaturated acids composing linseed oil.

Experiment 1 Fe (CN) -
x y

 

 

Ferric FerroCyanide

Experiment 2 (L-H)‘

  
 

o Ferric FerroCyanide

I Iron Oxide
L Linseed Oil

        
   

.L'L' L
L I I

IL-

L I

Iron Oxide

Figure 4.34: Summary of LDMS experiments involving Fe-containing pigments
suspended in linseed oil.

173

Let us focus on the second experiment a little more closely. In this experiment, a
fine particle solid (Fe203) which efficiently absorbs UV light is in a mixture with a
nonabsorbing analyte (linseed oil). The experiment resembles traditional MALDI, which
employs an organic matrix. In fact, in the early days of MALDI (1988), Tanaka and his
research group first introduced the concept of using a matrix to assist in the LDMS
experiment (63). In their experiment (illustrated in Figure 4.35), they suspended cobalt
particles (30 nm diameter) and the analyte (protein) in glycerol and irradiated the sample
using a pulsed N2 laser (337nm, 15 ns pulse width). Ions formed were separated in a
TOF mass analyzer. Monomeric and multimeric ions of the protein, lysozyme
(~ 100,000 Da), were observed in the mass spectrum. Tanaka and coworkers
characterized the cobalt nanoparticles as having a high photoabsorption, a low heat
capacity, and a large surface area per particle. These physical properties were necessary
for the experiment to be effective, meaning that the Co particles allow for a rapid heating
to occur, which is necessary to desorb and ionize analyte molecules without thermal

decomposition.

“Ultra Fine Metal + Liquid Matrix Method”
(w )+

   
 

Co particle
9 Peptide

Glycerol
..o'."'.o.... 0.... ,:

Figure 4.35: Illustration of Tanaka’s et al. “ultra fine metal + liquid matrix method”.

174

In contrast, Sunner et al. demonstrated that surface-assisted laser
desorption/ionization (SALDI) using graphite particles and glycerol is useful for ionizing
proteins and peptides (64). In their experiments, 10-150 um diameter graphite particles
and a 337 nm pulsed N2 laser were used to ionize proteins and peptides. The particles
were a thousand times larger in size than the 30 nm diameter Co particles that Tanaka and
coworkers used. The results indicated that ionization might occur through a bulk
desorption process like laser ablation. More recently (2000), Kinumi er al. performed an
extensive study in which 11 metals and metal oxides (Al, Mn, Mo, Si, Sn, SnO2, TiO2,
W, W03, Zn, and ZnO) were evaluated as potential inorganic matrix additives, for the
detection of poly(ethylene glycol) 200 (PEG 200) and methyl stearate (65). As with
traditional organic matrices used in MALDI experiments, the results varied according to
the matrix used. Frequently, the experiment worked, meaning ions representative of the
analyte were observed in the mass spectrum. Other times, matrix ions were formed. In
some instances, both analyte and matrix ions were formed. In order to understand these
differences, Kinumi and coworkers researched the ionization potentials and the standard
heat of formation (AHf°(g)) for each inorganic matrix studied. Typically in MALDI
experiments, heat of vaporation values for atoms or compounds are considered.
Considering the relationship,

AHvap° = AHf°(g) - AHf°(s), where AHf°(s) = 0,

AHvap° = AHf°(g).

No correlation was made concerning the performance of a matrix and either its ionization
potential or energy required to sublime the matrix molecules. However, when analyzing

the AHf°(g) values, a trend appeared concerning whether or not matrix ions were formed.

175

They concluded that the matrix with the lower AHf°( g) value will be easier to desorb and
ionize in the MALDI experiment. Consequently, if one were analyzing a relatively low
molecular weight analyte and wanted to suppress background peaks originating from the
matrix, he should select a matrix additive, such as W, which has a large AHf°(g) value
(849 KJ/mol). However, this is only half of the story. The value Kinumi and coworkers
should have been calculating and comparing is the sum of the ionization potential and
AHf°(g) value. For example, Ne requires a very small amount of energy to go from the
solid to the gas phase. However, Ne would never be seen in an LD mass spectrum,
simply because too much energy is required to ionize Ne atoms. The results are
applicable to our experiments.

Unfortunately, Fe and R203 were not included in Kinumi and coworkers study,
however their work helps to characterize our two experiments. When first trying to
explain why in one experiment colorant ions were generated and in the second, linseed
oil ions were formed, LD mass spectra of numerous Fe-containing compounds in linseed
oil were obtained. The point in doing so, was to determine if a trend could be established
with the type of ligand attached to the Fe and the peaks observed in the mass spectra.
This was unsuccessful.

We next considered factors such as particle size, ionization potentials, and heats
of formation values of ferric ferrocyanide and iron oxide. In terms of particle size,
smaller particles have larger surface area-to—volume ratios, which may assist in the
absorption process. The study of nanopanicle properties is a currently a well-researched,
but poorly understood field of chemistry (66,67). Additionally, the work of Sunner et al.

suggests that the performance of an inorganic matrix is not a function of its size (64).

176

Fe(s) has a relatively low ionization potential (7.8 eV), however the ionization potentials
and heat of formation values for ferricferrocyanide and F6203 are unavailable. We can
conclude that ferric ferrocyanide should not be used as a matrix (at least for the analysis
of artists’ colorant vehicles).

It should be noted that the linseed oil peaks appeared in a few other pigment
spectra. For example, linseed oil peaks also appeared in the negative ion mass spectrum
of malachite (Figure 4.24b). An enlarged portion of the mass spectrum (mlz 270-400) is
shown in the inset. It appears that there may be a related set of peaks at m/z 361. It is
unclear whether or not they represent linseed oil-pigment complex ions or simply a
linseed oil polymerization product. The fact that these ions do not appear in the Fe203
spectrum, suggests that the ions are unique to malachite.

To further investigate the utility of using Fe203 in terms of generating information
on the vehicle component of a paint, Fe203 was suspended in other drying oils, which
included walnut oil and poppy oil. Walnut oil and poppy oil contain the same
unsaturated acids as linseed oil, but in different concentrations. Linseed oil “dries”
through a cross-linking process at the sites of unsaturation (68). Consequently, the fastest
drying oil will have the highest concentration of linolenic acid (three degrees of
unsaturation). Walnut and poppy oils have a lower concentration of linolenic acid,
compared to linseed oil, and therefore take longer to dry. The negative ion LD mass
spectra of Fe2O3 suspended in walnut oil and poppy oil, and analyzed on paper, are
shown in Figures 4.36 and 4.37, respectively. It is important to note that the relative
intensities of the distribution of peaks representing the three unsaturated acids present in

the oils has changed. The peak at m/z 279 is now the most intense peak. Based on the

177

compositions of the oils, it does not appear that their LD mass spectra may be used for
quantitative analysis. However, they may be used to determine if linseed oil was used as
the vehicle, as opposed to either walnut or poppy oil. Also, three examples were
evaluated here. A more extensive study would be required. This is encouraging
information, considering that even though we are not particularly interested in the vehicle

component, there are certainly other research groups that are.

 

 

 

 

 

 

 

279
>.
r:
m 281
s:
Q)
E
o 255
>
0:
cu
T»
m
277
a“ - Ag. 2 -' - JLA -A: - . ufi-fnju- k ._-
225 245 265 mlz 285 305 325

Figure 4.36: Partial negative ion LD mass spectrum of Fe203 suspended in walnut oil
applied to paper.

178

Relative Intensity

255

 

 

A A A “4‘ A L LA-JA .
v

279

277‘

281

 

LA

— A “‘4‘ LA
f

 

 

225 245

Figure 4.37 : Partial negative ion LD mass spectrum of Fe203 suspended in poppy oil

applied to paper.

mlz

179

285

305

325

References

1) Feller RL, editor. Artists’ pigments: a handbook of their history and characteristics,
vol. 1. New York: Cambridge University Press, 1986.

2) Roy A, editor. Artists’ pigments: a handbook of their history and characteristics, vol.
2. New York: Oxford University Press, 1993.

3) Fitzhugh EW, editor. Artists’ pigments: a handbook of their history and
characteristics, vol. 3. New York: Oxford University Press, 1997.

4) Carlson JH, Krill J. Pigment analysis of early American watercolors and fraktur. J Am
Institute for Conservation 1978;18:19-32.

5) Berrie BH. Prussian blue. In: Fitzhugh EW, editor. Artists' pigments: a handbook of
their history and characteristics, vol. 3. New York: Oxford University Press,l997;
191-217.

6) Batcheler PH. Historic Structure Report, Old North Church, National Park Service,
Denver Science Center;l93-194.

7) Vitalis J B. Grundiss der Farberei 1842;239:242.

8) Brunello F. The art of dyeing in the history of mankind, trans. Hickey B,
Vicenza, 1973;230.

9) Crawford W. The keepers of light. New York: Morgan & Morgan, 1979;67-68.

10) Latimer WM, Hildebrand JH. Reference book of inorganic chemistry, 3rd
ed. New York: MacMillan Company, 1951;417.

11) Hurst GH. The painter’s laboratory guide: a handbook on paints, colours, and
varnishes for students. London, 1902;28-39.

l2) Kuhn H, Curran M. Chrome yellow and other chromate pigments. In: Feller RL,
editor. Artists’ pigments: a handbook of their history and characteristics, vol. 1. New
York: Oxford University Press, 1986;187-218.

13) Fluck E, Kerler W, Neuwirth W. Angew Chem 1963;6z277-332.

l4) Wertheirn GK, Rosencwaig A. Characterization of inequivalent iron sites in Prussian
blue by photoelectron spectroscopy. .1 Chem Phys 1971;54:3235.

15) Nakamoto K. Infrared spectra of inorganic and coordination compounds, 2"‘1
ed. New York: Wiley-Interscience, 1970;178-186.

180

l6) Newman R. Some applications of infrared spectroscopy in the examination of
painting materials. J Am Institute for Conservation 1979;19:42-62.

17) Wilde RE, Ghosh SN, Marshall B]. Prussian blues. Inorg Chem 1970;922512—2516.

18) Lechtman H. Thesis. New York City (NY): New York University, 1966.

19) Burgio L, Clark RJH. Library of FT-Raman spectra of pigments, minerals, pigment
media and varnishes, and supplement to existing library of Raman spectra of
pigments with visible excitation. Spectrochim Acta Part A 2001;57:1491-1521.

20) De Wild AM. The scientific examination of pictures. London, 1929.

21) Welsh FS. Particle characteristics of Prussian blue in a historical oil paint. J Am
Institute for Conservation 1988;27:55-63.

22) Townsend JH. The materials of J .M.W. Turner: pigments. Studies in Conservation
1993;38:231-254.

23) Herbst W, Hunger K. Industrial Organic Pigments. Production, Properties,
Applications, 2"d Ed., VCH, Weinheim, 1997.

24) Grim DM, Siegel J, Allison J. Evaluation of desorption/ionization mass
spectrometric methods in the forensic applications of the analysis of inks on paper,
J Forensic Sci 2001;46:1411-20.

25) Grim DM, Allison J. Identification of colorants as used in watercolor and oil
paintings by uv laser desorption mass spectrometry. Int J Mass Spectrom
2003;222:85-99.

26) Lias S, Bartmess JE, Liebman JF, Holmes JL, Levin RD, Mallard G. Gas-phase ion
and neutral thermochemistry. J Physical and Chemical Reference Data
1988;17:649,794-95.

27) Ball P. Bright earth. New York: Farrar, Strauss, and Giroux, 2001.

28) Bumin A, BelBruno JJ. ZnnSm+ cluster production by laser ablation, Chemical
Physical Letters 2002;362:341-48.

29) Benedetti-Pichler AA. Microchemical analysis of pigments used in the fossae of the
incisions of Chinese oracle bones. Industrial and Engineering Chemistry, Analytical

Edition l937;9:149-52.

30) Feigl F. Qualitative analysis by spot tests. New York, 1946;227—228.

181

31) Gettens RJ, Feller RL, Chase WT. Vermilion and Cinnabar. In: Roy A, editor.
Artists’ pigments: a handbook of their history and characteristics, vol. 2. New York:
Oxford University Press, 1993;159-182.

32) Plesters J. Cross-sections and chemical analysis of paint samples. Studies in
Conservation 1955-1956;2: I 10-57.

33) Aurivillius KL. On the crystal structure of Cinnabar. Acta Chemica Scandinavica
1950;4zl413-16.

34) Swanson HE, Fuyat RK, Ugrinic GM. 1955, Standard x-ray diffraction powder
patterns in National Bureau of Standards Circular 539 Washington, DC.

35) Waring CL, Annell CS. Semiquantitative spectrographic method for analysis of
minerals, rocks and ores. Analytical Chemistry 1953;25:1174-79.

36) Karr C, Kovach JJ. Far-infrared spectroscopy of minerals and inorganics. Applied
Spectroscopy 1969;23:220-221.

37) van Asperen de Boer JRJ. Reflectography of paintings using an infrared
Vidicon television system. Studies in Conservation 1969;14:96-118.

38) van Asperen de Boer JRJ. Infrared reflectography. Amsterdam, 1970.

39) Rees Jones S. Physics and painting. Bulletin of the Institute of Physics 1960;June:
157-65.

40) Bailey CK. The elder Pliny’s chapters on chemical subjects, part 1. London, 1929.

41) Palache C, Berman H, Frondel C. Dana’s system of mineralogy, vol. 1,
7th ed. New York, 1952;254.

42) Fitzhugh EW. In: Fitzhugh EW, editor. Artists' pigments: a handbook of their history
and characteristics, vol. 3. New York: Oxford University Press,l997;47-79.

43) http://www.knaw.nl.ecpa/ink/htm1/ink.html
44) Lehner S. Manufacture of ink. London: Henry Carey Baird & Co, 1892.

45) Wunderlich CH. Uber Eisengallustinte. Zeitung der Anorganischen Algemeinen
Chemie 1991;598-5992371-76.

46) Wunderlich CH. Geschichte und chernie der eisengallustinte — rezepte, reaktionen
und schadwirkungen. Restauro 1994; 100:414-42 1.

47) Krekel C. Chemische untersuchungen an eisengallustinten und anwendung der
ergebnisse bei der begutachtung rnittelaltericherhandschrifte [thesis]. Institut fiir

182

Anorganische Chemie der Georg August Universitat Gottingen, 1990.

48) Flieder F. Mise au point des techniques d’identification des pigments et des liants
inclus dans la couche picturale des enluminures de manuscripts. Studies in
Conservation 1968; 13:49-86.

49) Laurie AP. The pigments and mediums of the old masters. London: Macmillan,
1914.

50) Roosen-Runge H. Farbgebung und technik fruhmittelalterlicher buchmalerei, 2 vols.
Berlin, 1967.

51) Ktlhn H.Verdigris and copper resinate. In: Roy A, editor. Artists’ pigments: a
handbook of their history and characteristics, vol. 2. New York: Oxford University
Press, 1993;131-158.

52) Thompson DV, Jr. The materials and techniques of medieval painting, reprint of 1936
ed. New York, 1956;169.

53) Fielder I, Bayard M. Emerald green and scheele’s green. In: Fitzhugh EW, editor.
Artists' pigments: a handbook of their history and characteristics, vol. 3. New York:
Oxford University Press, 1997;219-271.

54) Grim DM, Siegel J, Allison J. Evaluation of laser desorption mass spectrometry and
UV accelerated aging of dyes on paper as tools for the evaluation of a questioned
document. J Forensic Sci 2002;47: 1265-73.

55) Druding SC. Dye history from 2600 BC to the 20'h century,
httpzllwww.straw.com/sig/dyehisthtml

56) Dube L. The copying pencil: composition, history, and conservation implications,
httpfl/aic.stanford.edu/conspec/bpg/annual/v17/bpl7-05.htm

57) Andrasko J. High pressure liquid chromatography analysis of ballpoint pen inks
stored at different light conditions. J Forensic Sci 2001;46:21-30.

58) Grim DM, Siegel J, Allison J. Does ink age inside of a pen cartridge? J Forensic Sci
2002 ;47: 1294-97.

59) Best SP, Clark RJH, Daniels MAM, Porter CA, Withnall R. Identification by Raman
microscopy and visible reflectance spectroscopy of pigments on an Icelandic
manuscript, Studies in Conservation 1995;40:31-40.

60) Burgio L, Ciomartan DA, Clark RJH. Pigment identification on medieval

manuscripts, paintings and other artefacts by Raman microscopy: applications to the
study of three German manuscripts. J Molecular Structure 1997;405:1-11.

183

61) Burgio L, Ciomartan DA, Clark RJH. Raman microscopy study of the pigments on
three illuminated mediaeval Latin manuscripts. J Raman Spectroscopy
1997;28:79-83.

62) Clark RJH, Gibbs PJ, Seddon KR, Brovenko NM, Petrosyan YA. Non-destructive in
situ identification of cinnabar on ancient Chinese manuscripts. J Raman
Spectroscopy 1997;28:91-94.

63) Tanaka K, Waki H, Ido Y, Akita S, Yoshida Y, Yoshida T. Protein and polymer
analyses up to m/z 100,000 by laser ionization time-of-flight mass spectrometry.
Rapid Communications in Mass Spectrometry 1988;22151-153.

64) Sunner J, Dratz E, Chen Y-C. Graphite surface-assisted laser desorption/ionization
time-of-flight mass spectrometry of peptides and proteins from liquid solutions.
Analytical Chemistry 1995;67:4335-4342.

65) Kinumi T, Saisu T, Takayama M, Niwa H. Matrix-assisted laser
desorption/ionization time-of-flight mass spectrometry using an inorganic particle

matrix for small molecule analysis. J Mass Spectrometry 2000;35:417-422.

66) Henglein A. Small-particle research: physicochemical properties of extremely small
colloidal metal and semiconductor particles. Chem Rev 1989;89:1861-1873.

67) Alivisatos AP. Perspectives on the physical chemistry of semiconductor nanocrystals.
J Phys Chem 1996;100:13226-13239.

68) Frankel EN. Lipid oxidation. Prog Lipid Res 1980;19:1-22.

184

Chapter Five: Stamps
Introduction

A postage stamp can mean many things to different people. For some, a stamp is
merely an inconvenient tax required to send their correspondence. For others, a stamp is
considered a work of art and an important part of history, which needs proper care and
attention to preserve the stamp in its original form. Throughout history, stamps have
honored famous people, places, and important events. Thus, philately, the collection and
study of postage stamps, has developed into a big business, as well as a hobby for some.
Consequently, forging and faking stamps has also become a big business. Experts can
distinguish between an original and a fake stamp, by studying both its physical
appearance and chemical composition.

The general composition of a postage stamp consists of a colorant (dyes or
pigments), phosphors, paper, and an adhesive. Limited information is available in
regards to colorants used, to maintain the security of the postage stamp ink formulations,
for obvious reasons. The earliest stamps were limited to grayish shades, since only
carbon black was used. Slowly, colors were introduced into the printing industry, and
low chroma reds and browns appeared. These were produced using natural plant (indigo)
and animal (cochineal) dyes. Nineteenth century postage stamps were characterized by
the use of mineral pigments. These colorants included yellow lead nitrate (Pb(N03)2),
red and black iron oxide (Fe203), white lead (PbO), Prussian blue, green chromium oxide
(Cr203), and ultramarine blue (CaNa7A168i6024SgsO4) (1). Today, the colors available
are substantial, due to the numerous synthetic organic dyes and pigments available.

Table 5.1 lists some of the colorants recently used in postage stamp ink formulations (1).

185

The list is dated 1990. Typically, linseed oil or a refined petroleum oil is used as the

colorant carrier.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Yellow Pigments Blue Pigments
Diarylide yellow (dichlorobenzidene) Copper phthalocyanine blue — GS and RS
Hansa yellow G Alkali blue — RS and GS triphenylmethane
type
Iron oxide yellow Iron blue (Milori blue)
Orange Pigments Green Pigments
Dinitroanaline orange Phthalocyanine green
Dianisidine orange Malachite green
Red giginents Black Pigments
Red lake “C” — calcium salt Carbon blacks - oil and furnace blacks
Lithor rubine - calcium salt Iron oxide black
Red 23 - calcium and barium salts White Pigments
Violet Pigments Titanium dioxide — opaque
Methyl violet Calcium carbonate - transparent
Carbozole violet Barite (barium sulfate) - transparent

 

 

Table 5.1: A list of colorants used in contemporary postage stamp ink formulations.

Phosphors are included in postage stamp ink formulations to serve as a tagging
system. Post offices employ facer/canceller machines to locate and cancel stamps, by
detecting the components which phosphoresce. A phosphorescent material is preferred
over one which fluoresces, simply because the paper envelope typically contains a
fluorescent material, which is added as a brightening agent. Tagging of airmail stamps
was first achieved in 1963, using a calcium silicate compound (CaSiOg), which glows an
orange-red color when exposed to short-wavelength UV light. A zinc orthosilicate
(Zn2SiO4) is used to tag first class stamps, and glows a green color (1).

An adhesive is applied to the back of the stamp, so that it may be affixed to an

envelope. The earliest and most primitive adhesive used was gum Arabic, which comes

186

from the acacia tree found in tropical areas. Gum Arabic has a variety of uses. It is used
as a stabilizing agent and viscosity adjuster in the food industry, in addition to the
printing and adhesive industries. Gum Arabic is composed of a mixture of Ca, Mg and K
salts of Arabic acid, which is a branched polysaccharide containing galactose, rhammose,
glucuronic acid, and arabinose residues. The molecular weight of gum Arabic can range
between 260,000 and 1,160,000 Da (1). Gum Arabic has the problem of drying-out,
yellowing, and cracking. Consequently, a switch was made to dextrin.

Dextrin is derived from food grain corn starch and consists of a mixture of low
molecular weight polysaccharides. As gum Arabic, dextrin is also used as a thickening
agent in food and ink formulations (1). One drawback of the use of dextrin as an
adhesive, is that it is known to deteriorate over time. Unfortunately, this devalues the
stamp and alternatives needed to be considered.

The search for the ideal adhesive ended when polyvinyl alcohol (PVA) was
tested. Besides its adhesive uses, PVA is also used as a sizing agent in the manufacture
of textiles, paper, and plastics. It is a water-soluble resin with the formula
(-CH2CHOH-)n. The degradation processes noted with gum Arabic and dextrin are not
observed when PVA is used, due to the inability of PVA to absorb moisture (1).

Additional ingredients are frequently included in the adhesive mixture to modify
the adhesive’s properties. These ingredients may include glycerin, corn syrup, glycols,
urea, sodium silicate, and emulsified waxes. The addition of these ingredients affects
properties such as flexibility, spreading quality, and ease of re-wetting. Sodium

benzoate, quaternary ammonium salts, and phenol derivatives may also be found in the

187

mixture, as preservatives. Lastly, oils of Wintergreen, lemon, grape, or anise are
frequently added as scenting and flavoring components (1).

Overall, there is a lot of chemistry involved in the production of a postage stamp.
As with all of our other studies, the focus of this project involves the colorant. Three
separate cases will be presented. The first and most simple, is the analysis of a stamp
whose color is claimed to be carmine. Having experience with the carmine lake colorant,
we were interested in whether or not the colorant was truly carmine. The second case is
an interesting story involving stamps which were used by French and British spys during
WW II. A slight physical difference is noted between the original French stamp and the
forged British spy stamp. However, we were interested in whether or not they could be
distinguished chemically. The final case involves a “Changeling”. In the early history of
stamps, stamps which were found to change color on exposure to light or gases in the
environment were called “changelings” (2). The stamp studied here was’printed using
PbCrO4, which is known to darken upon exposure to sulfur-containing gases in the

atmosphere (1-5). This reaction was investigated.

Experimental

A mask was prepared, by cutting a 2 x 2 in2 square from a manila folder. A 1 x 1
in2 square was removed from the center of the larger square. A stamp was placed in the
center of the modified sample plate and the mask was placed on top. The edges of the
mask were taped to the sample introduction plate, so that the stamp was left completely
unharmed. The user parameters of the MALDI instrument, as cited in Chapter One, were

employed.

I88

The Carmine Stamp

As was demonstrated in Chapter Two: Watercolors, the names of colorants may
often times be misleading. Reds are frequently referred to as. vermilion, cadium sulfide,
or carmine, but are they chemically what these names imply? Let us consider a 2¢
carmine stamp. Red lakes were commonly used throughout history as colorants on
postage stamps, therefore we were expecting to generate LD mass spectra indicative of
carmine lake (Chapter Four). Red lakes were known to fade when exposed to light or
water, so there was a phasing out of their use. Positive and negative ion LD mass spectra
were obtained of the stamp. Only Na" and K+ ion peaks were generated in the positive

ion mass spectrum (not shown). The negative ion mass spectrum is shown in Figure 5.1.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

222
207 222
207
377 221

z, 80
g 26 m
'8 208
i 97 2“ 209 2%
.2 194 J}
‘5 140 -LLWL-.- . . .
T) 190 200 210 220 230 240
m

35 267 638 736

62 721
496 59‘ j 892
L. ‘ t . ...-L_ has- _.__.-.-_L._A._L_A-LA A .A -.._A- . A ,
0 190 380 570 760 950

mlz

Figure 5.1: The negative ion LD mass spectrum of the 2¢ carmine stamp.

189

The mass spectrum is clearly not of carmine lake. A portion of the lower mass
region is enlarged in the inset of Figure 5.1. It is difficult to determine whether or not the
peaks at m/z 207 and 222 are related. The pattern of the peaks resemble that of a neutral
dye, in which both the molecular ion (M') and deprotonated molecule (M-H)' are
observed. However, one would expect to see a similar pattern in the positive ion mass
spectrum, which was not observed. This suggests that the colorant is anionic. The
pattern of peaks at m/z 222 does not resemble an isotopic pattern for a known element,
which leads us in the direction of an organic colorant. The difference between m/z 207
and m/z 222 is 15 mass units. This could represent the loss of an O atom or an —NH2
group, which would be replaced by a H atom, accounting for the 15 mass unit difference.
Another possibility is that the peak at m/z 207 represents a fragment ion, resulting from
the loss of a ~CH3 group. These are not typical losses observed in a mass spectrum,
which suggests that the peaks represent separate colorants (components). The colorants
remain unidentified, however the point here is that carmine lake was not used to color the

stamp.

190

The French Spy Stamps

During WW 11, communication between the French and the British was frequently
intercepted by the Germans and plots were foiled. The French and British underground
devised the brilliant idea of using stamps to mark important messages. Figure 5.2 shows
two stamps. The stamp on the left is the original 1939 French stamp. The stamp on the
right, is the British forgery, which has a slight physical defect, noticed only by those who
were looking for the defect. The physical difference is indicated with arrows. Important
messages were marked using the British forged stamp. Only a limited number of these
valued stamps are available in mint condition. Currently, philatelists only relied on the
slight physical difference to differentiate between the two stamps. We wanted to take it
one step further and determine whether or not one could distinguish the two stamps by
their chemical composition. The LD mass spectrometric analysis of the two stamps

follows.

 

Figure 5.2: The spy stamps: a) the original French stamp; b) the forged British stamp.

I91

 

 

>5
fl
'5
G
4)
H
E 138
Q)
.2
H
.33.
Q)
9‘ 109
107
“in“. .. J.Mu.__.-.
75 175
340 (M')
>5
.*.:.'
:11
fl
4)
H
.5
Q)
,E (M-COOH')
5, 295
Q)
m
267
250

 

175

 

 

 

 

321

 

 

 

 

332
294 316
361
276 333
.1 21.1 .J I

 

45

mlz 535

521

4 0
524553

475 '

b)

    

Figure 5.3: The LD mass spectra of the original French stamp: a) positive ions;

b) negative ions.

192

575

 

 

 

 

330

 

 

 

 

 

 

 

 

 

 

3)
>3
«0.)
'5
t:
3
.5
a; 157
.5 332
.2
o
a:
159
431
-1 . kill“. 1 . 1 5170' 2 1
125 245 365 mlz 485 605 725
329
1))
>3
H
'5
=
o
H
5
a)
.2
in)
.2
Q
m 35 172
487
w 2 ,1
0 105 210 315 420 525

mlz

Figure 5.4: The LD mass spectra of the forged British stamp: a) positive ions;

b) negative ions.

193

The positive and negative ion LD mass spectra of the original French stamp and
the British forged stamp are shown in Figures 5.3 and 5.4, respectively. The stamps were
clearly printed with inks containing different dye compositions. The identities of the
dyes remain unknown, however we can characterize them. Three red colorants were
cited in the introduction of this chapter, which were used to color postage stamps. These
included Red Lake “C”, Lithor rubine, and Red B. No information was found on Lithor
rubine (calcium salt), however there was a Lithol Red discovered in 1899 by Julius
(BASF). This colorant was considered the first of the azo pigment lakes. The colorant
was prepared by an indirect diazotization method employing 2-napthylamine-l-sulfonic
acid as a diazonium compound. The organic colorant was initially precipitated onto an
inorganic support, which formed the lake. However, it was eventually determined that
the inorganic material did not have much of an effect on the color or properties of the
lake, thus the organic colorant was often times used alone. Lake Red C pigments were
discovered shortly after the introduction of Lithol Red in 1902, by Meister Lucius &
Brtining (now Hoechst AG) (6). The exact structures for these pigments were not found,
however the general structure for these pigments is shown in Figure 5.5a. The pigments
were classified as B-naphthol pigments. The structure in Figure 5.5b is a related pigment
classified as a napthol AS pigment. These structures will become important shortly. No

information was found on “Red B”.

194

a
) R3 b)
M++
N\
\ N
N \N
HO OH R;
NH
RD: 803‘, CH3, OCH3, Cl 0

RD: 803‘, CH3, C2H5, C1 RK: SO33 CH3
m:1-3 m: 1-3, n:1,2
M: Ca, Ba, Sr, Mn, Mg M; Ca, Ba

* With at least 1 803' group

Figure 5.5: a) the general structure of a B-naphthol pigment; b) the general structure of
a naphthol AS pigment.

Focusing on the French stamp first (Figure 5.3), it appears as though a few dyes
were used. The colorants are non-chlorinated. The negative ion mass spectrum suffers
from poor resolution, however it appears that there is a set of related peaks stemming
from the peak at m/z 340, which potentially represents a molecular ion (M'). The peaks
at m/z 295, 267, and 250 could represent the neutral consecutive losses of —COOH, -
C2H5, and —OH, respectively.

The British spy stamp, on the other hand, appears to contain only one, chlorinated
colorant. At first, it may seem as though the peak in the positive ion mass spectrum

(Figure 5.4a) at m/z 330 represents the (M+) ion and the peak in the negative ion mass

195

spectrum (Figure 5.4b) at m/z 329 represents the (M-H)' ion. However, if the structures
in Figure 5.5 are considered, the peaks may actually represent fragment ions. Beginning
with the smallest fragment ion at m/z 157, appearing in both the positive and negative ion
mass spectra, what are the possibilities? In order to absorb in the UV, the intact structure
probably contains an aromatic ring. The isotopic pattern also suggests that one C] atom is
present. Proposed structures for m/z 157 and mlz 172 are shown in Figure 5.6. The
assignments were made based on the possible structures in Figure 5.5. Additional
assignments were not made, however the remaining peaks are believed to represent
fragments of the intact colorant molecule. The reason for this lies in the fact that all of
the peaks maintain the isotopic pattern of one CI atom, and there are mass differences
between peaks which can be accounted for. For instance, the mass difference of 80
between mlz 329 and m/z 410, may be due to the loss of a sulfite group (-SO3). These
assignments have not been confirmed, however a variety of modern day, red pigments

were purchased and tested. The spectra did not match either of the spy stamps.

a) b)
Cl H2N CI
HZN OCH 3 H2N OCH 3
mlz 157 mlz 172

Figure 5.6: Proposed structures for a) m/z 157 and b) m/z 172.

196

Another approach is artificial aging. Modern day red rhodamine dyes have been
studied extensively in our lab (7). It has been shown that one redink containing
rhodamine B and a second red ink containing rhodamine 6Gproduce the same positive
ion mass spectrum, when the ink is freshly applied to paper. The reason for this is that
both dyes have the same molecular weight, but differ structurally. When the inks age
(degrade), they produce different degradation products, and unique peaks appear in the
“aged” mass spectra. Thus, it has been demonstrated that UV light (among other light
sources) can be used to artificially accelerate the age of the dye to obtain more structural
information.

The French and British spy stamps studied here, are rare and highly valued.
Therefore, the owner did not want them faded (aged). Numerous French and British
stamps were purchased from a local stamp collector’s shop. LD mass spectra were
obtained of all of them in the hope to find a few samples with the same dyes as the spy
stamps. One British stamp was found to have the same dye as the British spy stamp.
This stamp was subjected to UV irradiation for several hours. Fading of the colorant was
not observed, and the mass spectra remained the same. This approach was unsuccessful
here.

Ideally, we had hoped to identify the exact dye composition used on the stamps,
however the results are still interesting and useful. By completing this project, some
important concepts were learned about philately. For example, it was learned that
philatelists are more concerned with the physical appearance of stamps, rather than their
chemical composition, in regards to the detection of forgeries. Perhaps, publishing these

results in a philatelist journal (i.e. Linn’s Stamp News), would peak the interest of avid

197

stamp collectors to seek more sophisticated analytical techniques to help characterize
their stamps. LDMS is definitely an attractive option, considering it is a noninvasive
technique, which leaves the stamp virtually unharmed. Until then, philatelists will be

limited to microscopy techniques.

The PbCr04 Stamp

As was mentioned in the introduction, specific stamps have been known to
change color throughout history (1,2). Initially, the color changes were thought to be the
result of some sort of error in formulation. Eventually, it was determined that the
colorants reacted in response to either light or gases in the environment. Red lakes were
especially known to be fugitive and faded on exposure to light. Other metal containing
pigments were known to darken in the presence of pollutant gases, such as H2S.
Fortunately, there are means to chemically reverse the color change. The focus of this
study is on a yellow stamp (Gold Star Mothers) which contained the pigment PbCr04.
Three samples were provided (shown in Figure 5.7): an original (c), one that was exposed
to H28 (g) and had blackened (b), and one that had been exposed to H2S (g) followed by
treatment with H202 (a). The stamps were analyzed using LDMS to detect and identify

reaction products to characterize the reaction mechanism more effectively.

I98

 

Figure 5.7: The Gold Star Mothers Stamps: a) Exposed to H2S (g), followed by treatment
with H202 (l); b) Exposed to H2S (g); c) original (untreated).

The chemistry of this reaction is known (2-4). On exposure to H2S (g), some
black PbS(s) is formed. Following treatment with H202 (1), white PbSO4 (3) forms and
the original yellow color of unreacted PbCrO4 returns. 13 the reaction really this simple?
It has been reported that Pb-containing pigments darken naturally on exposure to light,
due to a change in crystal structure (8). Others have cited the formation of brown Pb02(s)
resulting from bacterial deterioration as the cause of the darkening (9). Additionally,
there are two metals present: Pb and Cr. Does Cr042' react with H2S (g) to form 0283 (a
dark brown/black powder) or another product (10)? These are a few of the questions we
were hoping to answer at the end of this study.

The positive ion LD mass spectra of all three stamp samples are shown in Figure
5.8, so that they may easily be compared. Likewise, the negative ion LD mass spectra of
all three stamps are shown in Figure 5.9. Returning to the positive ion mass spectra,
Figure 5.8a shows the mass spectrum of the stamp in its original form. Characteristic
PbCrO4 peaks are noted (appendix), which were discussed in Chapter 4. For example,

the peaks at m/z 208 represent Pb+ ions and the peaks at m/z 430 represent Pb2O+ ions.

199

Following exposure of the stamp to H2S (g), the positive ion mass spectrum changed
dramatically (Figure 5.8b). The colorants’ vehicle is unknown. Therefore, positive and
negative ion LD mass spectra were obtained of PbS in a variety of vehicles. The carriers
included linseed oil, water and gum Arabic. The spectra (not shown, see Chapter Six)
were not identical in every case. A few characteristic peaks were noted, which have not
appeared in other Pb-containing pigment spectra. In the positive ion mode, a cluster of
peaks at m/z 483 was noted and has been assigned Pb2ClS+. These peaks are not seen in
Figure 5.8b. An observation which is consistent with the PbS spectra is the absence of
the typical Pb—containing ions. This was noted in the spectra obtained of PbS in linseed
oil and in gum Arabic. Unfortunately, the new ions formed, do not have unique isotopic
patterns to aid in their identification. No peaks were observed having an isotopic pattern

consistent with Cr, suggesting that Cr2S3 or -any other Cr-containing product was formed.

200

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

39
a)
23
208
206
86 430
s 263 432
22
til -1- - . r48 131 7:14
200 7 .
>5
H
O-
m
5
160
E 1 17
Q 147
E
N 62 80
III!
é) s 9 232
184 317 70
LJTL- n J_ *4 I J
39
23
e)
208
247
206
57 368 430
22 263
447 550
413 495 754
. W_-,-L_.‘J.L_+_M_4._..L-LA4L.__ L‘“ AA‘“ “ .
155 310 mlz 465 620 775

 

 

Figure 5. 8: The positive ion LD mass spectra of the PbCrO4 stamps: a) the original
stamp; b) the stamp exposed to H2S (g), c) the stamp exposes to H2S (g)

followed by treatment with H202 (1).

201

 

100

 

 

 

 

 

 

 

 

 

 

 

 

 

>:
J“

g mo

9 117

H
'5 147

0
-E
.5 62 80

Q)
m 5 96

184
200
100
o ' 105

 

 

210

232

317

L1L_ u

420

mlz

b)

170

 

32:

Figure 5.9: The negative ion LD mass spectra of the PbCrO4 stamps: a) the original
stamp; b) the stamp exposed to H28 (g); c) the stamp exposes to H28 (g)
followed by treatment with H202 (l).

202

 

 

The sample was then treated with H202 (l) to restore the original yellow color.
PbCrO4 is not reformed. Rather, some of the PbS is converted into white PbO. This
transformation gives the appearance that the yellow color has returned. In actuality, the
color never left, it was simply masked by the inhomogeneous layer of black PbS formed
on the surface. The positive ion LD mass spectrum of this sample is shown in Figure
5.8c. The spectrum has changed again. The characteristic Pb—containing ions (m/z 208,
247, 430, etc.) have returned. A new peak appeared at m/z 368, which again does not
have any unique isotopes to help characterize the compound, indicating that it may
simply be an impurity.

Switching to negative ion mode, the spectra (Figure 5.9) of the three samples
remain relatively stable. In the mass spectrum of the original sample (Figure 5.9a), peaks
representing characteristic PbCrO4 ions (i.e. m/z 100, 200, and 300), as were discussed in
chapter four, are observed. Once the stamp is exposed to H28 (g), the spectrum (Figure
5.9b) changes slightly. The intensities of Cery' ion peaks decrease in the m/z 300
region. Additionally, two new peaks are noted at m/z 232 and m/z 470. There was no
evidence of any Pbey' or related ions formed

The sample was treated with H202 (l), and the negative ion mass spectrum of this
sample is shown in Figure 5.9c. The intensities of the peaks at m/z 300 have increased,
as compared to the H2S (g) exposed sample. The peak at m/z 470 has disappeared.

Was this experiment successful? The sample studied had changed color in
response to its environment. This was confirmed in both the positive and negative ion
mass spectra. Unique PbS ions were not formed, however the absence of certain Pb-

containing ions were noted, which is characteristic of the presence of PbS. Following

203

treatment with H202 (l), the spectra changed once again to resemble the original spectra.
Peaks representing CrxSy ions were not detected, which supports the accepted explanation
for this reaction, in that only Pb and not Cr is reacting with the H2S(g). The “new” peaks
which appear sporadically in some of the spectra were not identified, however the
conditions of the experiments preparing the stamps may not have been controlled,
allowing for the presence of impurities. Overall, the experiment was successful in that

we were able to map the chemical change of the stamp.

Summary

By completing this project, we have demonstrated the usefulness of LDMS in the
area of philately. First, we showed how the color name of a stamp (carmine) may be
misleading of its true composition. Next, we demonstrated how forged stamps may differ
chemically, in addition to physically. We successfully used LDMS to determine that the
famous French and British stamps were not chemically the same. Lastly, we used LDMS
to detect and analyze the products formed from chemical reactions taking place on the
stamps. The PbCrO4 exposed to H2S (g) reaction will be explored further in the

following chapter.

204

References

l) Sharkey J. Chemistry of postage stamps: dyes, phosphors, adhesives. The United
States Specialist 1990 Augustz469-482.

2) Cabeen RM. Standard handbook of stamp collecting. New York: Thomas Y. Crowell
Company, 1965.

3) Clay HF, Watson V. Some important properties of chromes. Journal of the Oil and
Colour Chemists’ Association. 1944;27:3—18.

4) Torlaschi S, Caggiati L, Zorzella G. Chemical and physical effects of sulphur dioxide
on inorganic pigments and paints. 207-215.

5) Erkens LJH, Hamers H, Herrnans RJM, Claeys E, Bijnens M. Lead chromates: a
review of the state of the art in 2000. Surface Coatings International Part B: Coatings
Transactions 2001 ;84: 169-176.

6) Herbst W, Hunger K. Industrial organic pigments: production, properties, applications,
2nd ed Weinheim VCH,1997.

7) Dunn JD, Siegel J, Allison J . Photodegradation and laser desorption mass spectrometry
for the characterization of dyes used in red pen inks. .J Forensic Sci 2003;48:652-657.

8) Watson V, Clay HF. The light- -fastness of lead chrome pigments. Journal of the Oil
and Colour Chemists’ Association 1955; 38. 167-177.

9) Petushkova JP, Lyalikova NN. Microbiological degradation of lead-containing
pigments in mural paintings. Studies in Conservation 1986;31:65-69.

10) Saleh FY, Parkerton TF, Lewis RV, Huang JH, Dickson KL. Kinetics of chromium
transformations in the environment. The Science of the Total Environment
1989;86:25—41.

205

Chapter Six: Studies of the Interactions of H2S with Colorants

Introduction

Before the hazardous nature of lead compounds was established, Pb—containing
pigments were frequently found in artists’ palettes. Some of these pigments included
chrome yellow (PbCrO4), lead white (PbCO3 0 Pb(OH)2), and lead antimonate
(Pb3(SbO4)2). Throughout history, Pb-containing pigments were known to darken or
change color over time in response to their environment. Researchers have cited several
different mechanisms for these transformations, dependent on the type of Pb-pigment
used and the environment it is in. The mechanisms of these reactions are at least partially
understood and will be explained here.

In the 19508, Watson and Clay investigated the crystal structures of lead chromes
and how they affected the darkening of the lead chrome pigments in the presence of light
(1-2). They proposed that a photochemical reduction was responsible for the blackening
of PbCrO4. Watson and Clay concluded that the polymorphic nature of lead chromate
influenced the pigment’s sensitivity to light, citing the monoclinic form to be more stable
than the orthorhombic form. They were not able to detect any Pb-containing products
formed or remaining, however they measured the pressure of gases released during the
transformation. Watson and Clay explained that the darkening was caused by the
formation of the greenish-black lead chromite.

Several research groups have investigated the darkening of lead white and
proposed solutions to reverse the transformation (3-5). Petushkova and Lyalikova

performed an interesting study involving bacterial deterioration of lead white, massicot

206

(PbO), and minium (Pb304) on mural paintings. Microorganisms frequently infest
paintings resulting in detrimental effects on the pigments and the underlying material.
Petushkova and Lyalikova discovered through X-ray examination that the darkened
brown spots on the mural painting studied were due to the formation of PbOz. They
explained that heterotrophic bacteria generate H202, which resultedin the oxidation of

bivalent lead. Equation (1) summarizes the overall reaction:
2PbCO3 ° Pb(OH)2 + 3H202 —'> 3Pb02 + 2H2CO3 + 2H20 (I)

Giovannoni and coworkers investigated a similar reaction involving the discoloration of
the lead white pigment in wall paintings (4). They proposed a similar mechanism, but
also noted that the alkalinity of the environment and moisture have a significant influence
on the rate of transformation. A slightly acidic H202 solution was cited as a means to
reverse the reaction (4,5). The following 3-step mechanism (equations 2-4) describes the

reconversion to lead white.

Pb02 + H202 + 2CH3COOH '9 Pb(CH3COO)2 + 02 + 2H20 (2)
3Pb(CH3COO)2 + 2H20 '9 2Pb(CH3COO)2 ' Pb(OH)2 + 2CH3COOH (3)

2Pb(CH3COO)2 0 Pb(OH)2 + 2C02 + 2H20 -) 2PbC03 ' Pb(OH)2 + 4CH3COOH (4)

Petushkova and Lyalikova also noted the transformation of the Pb-containing
pigment to black PbS. When bacteria use S-containing amino acids, they generate H28.

Again, H202 is cited as the appropriate treatment to use in order to reverse the reaction.

207

However, in this case, the product of the reconversion is PbSO4, as opposed to lead
white. The reaction between H2S and Pb—containing pigments was noted in Chapter 5
and is investigated more in depth here.

In Chapter 5, the lead chromate stamps analyzed had been previously exposed to
H28 (g) and treated with H202, by another researcher. We wanted to recreate the
experiment in our laboratory to learn more about the transformation and how LDMS
could be used to help characterize the reaction. For instance, were there any additional
products formed? Also, the transformation of PbCrO4 to PbS is not complete, meaning
that there is always yellow pigment remaining. Considering that PbS is not a commonly
used black pigment by artists, as compared to the more commonly used carbon black
pigments, the detection of PbS in a painting offers insight into the original pigments used
and where the painting was stored (e. g. in a highly industrial area having a high
concentration of S-containing gases). We also wanted to test whether or not the extent of
deterioration could be determined by comparing the intensity of the peaks of the reactants
(PbCrO4) and the products (PbS) in the mass spectrum.

One other direction we were interested in pursuing is the treatment of H28-
exposed samples with H202, to determine if reconversion products could be detected.
Hypothetically, the reaction product is PbSO4, however the samples analyzed are impure
mixtures, and there may be additional products formed. This information could
potentially be very useful to art conservatorss, since the detection of a reconversion
product would reveal that a painting had previously been treated to cover up a discoloring

reaction.

208

Experimental

Numerous Pb-containing pigments were obtained, including PbO, PbOz, PbCrO4,
PbS, Pb3(SbO4)2, PbMoO4, and PbSO4. All of the compounds were suspended in linseed
oil and applied to paper. The samples were taped to the modified sample introduction
plate and subjected to positive and negative ion LD mass spectral analysis.

The green 613 pigment from the pigment distributor’s carrying case, discussed in
Chapter 4, was used as the focus of this study. A second pigment, labeled Y12, was
selected from the pigment set. Yl2 is a yellow pigment composed of PbCrO4_ The
second pigment was added to see if there was any difference in the results when PbCrO4
was by itself (Y12) or in a mixture (G13, with Prussian blue). The pigments were
suspended in linseed oil and applied to paper. Once the samples “dried”, they were
exposed to H28 gas.

In the first experiments performed, H28 (g) was made in our lab. Approximately
5 mL of HCl (1) was added to a beaker containing approximately 5 g Na2S (s). A stick
was taped across the top of the beaker, so that samples (~ 1 in2 paint-on-paper) could
hang freely in the center of the beaker. In the original experimental design, the samples
were taped directly to the sides of the beaker. However, at the end of experiments, the
samples were wet due to condensation on the walls of the beaker. To reduce
condensation, the beakers were placed in an ice bath. The ice bath was also employed to
reduce the HCl from vaporizing and reacting with the samples. The beaker was covered
with a watch glass following the addition of HCl to the Na2S in the beaker. The samples
were exposed to H2S (g) for varying time increments. The exposed samples were taped

to a modified sample plate holder and subjected to LDMS analysis. The H28 exposed

209

samples were then treated with H202 (1) until the darkened spots disappeared and the
sample appeared to have returned to its original color. The samples were then taped to
the LDMS plate and subjected to LDMS analysis once again.

The design is very primitive and not effectively controlled, however it provided
us with data suggesting that it was possible to detect the chemical changes involved in the
reaction, using LDMS. The results of these experiments led to a set of controlled
experiments. A canister of H2S (g) was purchased and connected to a vacuum manifold.
Samples (613 and pure PbCrO4 in linseed oil on paper) were exposed to approximately
65 Torr H2S (g) for a variety of time increments. Following H2S exposure, the manifold
and sample reaction chamber was flushed with air, and the samples were removed and

analyzed using LDMS.

Preliminary Work

As a pigment, lead chromate is prepared in many ways. Other Pb-containing
pigments, such as PbSO4, PbMoO4, and Pb3(SbO4)2, are frequently precipitated with
PbCrO4 in order to alter its shade (6). Numerous Pb—containing compounds were
obtained, which were cited in the literature as either possible impurities in lead chrome
pigments or a reaction product resulting from exposure to H28 gas. Positive and negative
ion LDMS spectra were obtained of each of the samples, suspended in linseed oil and,
applied to paper. The purpose in completing this task was to catalog all of the possible
ions which might be observed in our HzS exposure experiments involving lead chrome
pigments. The results of these experiments are summarized in the Appendix. While

there are some ions, which are generated from nearly every Pb-containing compound

210

(e. g. Pb+, PbK", szOzH+), each compound usually produces at least one or two ions
which help to distinguish it from others. The data in the Appendix will be referred to

throughout this chapter.

Designing the H 25 Exposure Experiments

In the first experimental design, concentrated HCl was added to NaZS in a beaker
containing the sample, and the reaction proceeded at room temperature. Black spots
appeared on both the G13 and Y12 samples during exposure, however the green 613
sample was turning an aqua, more bluish-color, while the yellow of the Y12 sample was
disappearing. The 613 pigment is a mixture of blue ferric ferrocyanide and yellow
PbCrO4. The Y12 pigment is strictly PbCrO4. The black spots were expected due to the
formation of PbS. The Cery' ions noted in the initial negative ion mass spectrum of the
G13 sample (Figure 6. lb, Y12 spectrum not shown) also disappeared, supporting the
hypothesis that the chromate chromophore was destroyed.

By cooling the reaction chamber, the black spots still formed upon exposure to
H25 (g), however the sample remained the appropriate color. This suggested that
moisture played a role in the destruction of the chromate chromophore. Koller and
coworkers had noted that moisture greatly accelerated the conversion of lead white to

lead dioxide (5). Once the samples were removed from the cooled reaction chambers, the

211

104

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

88 a)
5‘
g 120 PbK
3 247
PbOK+
5 Pb+
a) 0 263
1% 72 2 8
'5 430
m 447
J 530 754
LLLL- i #LJJ iJL'i Ax i 1
100 300 mlz 500 700
200
b)
>, 100
J‘:
m
fl
3
.5
o
.2
E
o
9‘ 284 300
I 1 l A ‘L‘L—h‘J
50 250 300

 

Figure 6.1: The LD mass spectra of G13 in linseed oil; a) the partial positive ion mass
spectrum; b) the partial negative ion mass spectrum.

212

 

yellow PbCrO4 pigment would gradually disappear and the aqua color of the G13
pigment returned. Thus, samples were analyzed immediately after exposure to H28 (g)
before any side reactions could take place.

What was causing the destruction of the chromate chromophore? It has
previously been determined that oxygen can have detrimental effects on Pb-containing
pigments (8). Specifically considering PbCrO4, if left undisturbed, oxygen does not have
any effect on the pigment. Pb has a high affinity for S. If a pollutant, such as H2S (g), is
introduced into the environment, Pb will bond with S to form PbS, leaving the chromate
free to react. Linseed oil (and other drying oils) extracts 0 from CrO4, ultimately
destroying the chromophore (8). An oxygen rich environment accelerates this process.
Lead may also react with linseed oil, forming Pb-linoleate (lead soap), which may be

washed away by sulfurous acid (8).

What to Expect?

The proposed reaction between PbCrO4 and H2S (g) results in the formation of
PbS. To return the pigment to its original color, the PbS product is treated with H202 (l)
to form PbSO4. Positive and negative ion LD mass spectra were obtained of PbS
suspended in linseed oil and applied to paper. Two samples prepared at separate times
produced different spectra. The spectra of the first sample are shown in Figure 6.2 and
the spectra of the second sample are shown in Figure 6.3. In the positive ion mass
spectrum of the first sample (Figure 6.2a) a small peak at mlz 208 is noted, representing
Pb+ ions. In the middle mass region, the peaks at mlz 240, 268, 304, 332, and 360 have

not been identified, however they originate from the paper. In the higher mass region of

213

the mass spectrum, numerous peaks have been identified as representing PbeyCl;r ions.
The negative ion mass spectrum of the first sample (Figure 6.2b) is dominated by peaks
representing PbClg’ ions. This is an important observation. ,As will soon be revealed,
PbCl3’ were detected in samples that were exposed to H2S (g) created using HCl.
Initially, we were concerned that our experiment was creating conditions which resulted
in the formation of products that did form naturally. From these experiments, it appears
as though, if Cl (even as an impurity) and PbS are present, PbCl3' ions readily form.
PbCl3' ions have only been observed when PbS is present in the sample, and therefore

appear to be and indicator that PbS is present.

214

Relative Intensity

208

 

 

240

 

 

332

 

360

 

 

 

 

 

 

 

 

 

775

 

 

150 275 400 mlz 525 {so
PbCls'
313
b)
315
5‘
'5:
E
E
H
G)
>
15
"3 311
O .
a:
317
3.3. . . . l -. 1 .
275 290 305 mlz 320 335 350

Figure 6.2: The LD mass spectra of PbS in linseed oil on paper (sample 1); a) the partial
positive ion mass spectrum; b) the partial negative ion mass spectrum.

215

208

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

3)
263 +
>. 165 1:3?
a 247
E
G)
>
3U
:3
Q)
a:
432 Pbsszt
686
414 480
377 A“ _ A! _ _ 6 4
150 260 370 ml 480 590 700
82-64 1
80 b)
5‘
a 53' 185
.5 96 137 PbSCI-
6; 275
’5
.‘3
S2 C1-
35
240 304 339 Pb232C"
5 5
ALL - 3., - LL 1 , l
0 110 220 mh 330 440 550

Figure 6.3: The LD mass spectra of PbS in linseed oil on paper (sample 2); a) the partial
positive ion mass spectrum; b) the partial negative ion mass spectrum.

216

The mass spectra of the second PbS in linseed oil on paper sample (Figure 6.3)
are very different. The positive ion mass spectrum (Figure 6.3a) is abundant with the Pb-
containing peaks (m/z 208, 225, 247, 263, etc) characteristic of the positive ion mass
spectrum of nearly every Pb-containing compound. It is important to point out the cluster
of peaks at m/z 446, representing szS+ ions. These peaks are indicative of PbS. Other
Pb-containing pigments generate peaks at mlz 447, representing Pb202H+ ions. It is also
interesting to note that Cl' ions are represented in the negative ion mass spectrum (Figure
6.3b) at m/z 35, but there were no PbeyClz+ ions identified in the positive ion mass
spectrum. Sulfur cluster ions (m/z 64, 96) are also noted in the negative ion mass
spectrum, which do not consistently appear in every negative ion mass spectrum of PbS.
The peaks at m/z 313, representing PbClg' ions, are not observed in Figure 6.3b, however
there is a set of peaks at m/z 275 representing PbSCl' ions. In results to be discussed, the
peaks at mlz 275 and 313 often times occur together. While it is discouraging that the
mass spectra of PbS are not reproducible, variations could easily be due to differences in
sample loading, for example. Both spectra provide unique peaks which may be
potentially generated by the compound. These unique peaks have been identified in the

spectra of H28 exposed PbCrO4 samples.

217

208

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

a)
szoa+
4457
483

3*

g Pb202H+

E 447

in:

G)

>

'5

£3

32

430 Pb3SCl+
243 689 Pb,sc10+
263 301 39
150 320
PbCl3'
313
315 . b)

3.:

E

.5

0

>

’3 311

'5

a:

317
PbSCl'
275 L 339
. 11 pl“ - 1 . , “1“ . v
250 275 300 mm 325 350 375

Figure 6.4: The LD mass spectra of PbS in linseed oil on paper treated with H202 (l);
a) the partial positive ion mass spectrum; b) the partial negative ion mass

spectrum.

218

The PbS in linseed oil on paper samples were then treated with enough H202 to
give the samples a white “frosted” appearance. Theoretically, the spectra of this new
sample would be that of PbSO4. Positive and negative ion LD mass spectra (not shown)
were obtained of PbSO4 (Aldrich). Considering that PbSO4 does not generate any unique
peaks when compared to the other Pb—compounds, we will not be able to tell for certain if
PbSO4 is the product. Perhaps, the disappearance or a decrease in the intensities of PbS
peaks will be observed.

The positive and negative ion LD mass spectra of the first PbS sample treated
with H202 are shown in Figure 6.4. The positive ion mass spectrum (Figure 6.4a)
changed dramatically. The results are summarized in Appendix B. Notably, the clusters
of peaks at m/z 447 (szOzH+) have replaced the peaks at m/z 446 (Pb2S+). It should be
noted that these experiments seem to be laser power dependent. A higher laser power
allows for the ions with larger m/z values to be seen in the mass spectrum. The negative
ion mass spectrum (Figure 6.4b) did not change dramatically. A new peak appeared at
m/z 339, which has not been identified but contains Pb and Cl, based on its isotopic
pattern. From this experiment, it appears as though we may expect for Pb+ and PbO+ ions

[0 return.

Uncontrolled H 2S Studies

Preliminary experiments using the cooled reaction chambers and chemically
generated H2S (g) indicated that the spectra of the lead chrome pigments changed
following H28 (g) exposure, and PbS ions could be detected. Using this method, the

reaction occurred very quickly (within seconds). As in the methyl violet degradation

219

study, we wanted to determine if we could effectively monitor this transformation.
Several experiments were designed. One involved a timed study, in which samples were
taken at half minute increments up to 2 minutes and then every two minutes there after up
to 8 minutes. The reaction occurred too quickly. A second experiment was designed in
which separate samples were exposed for 2 minute intervals to H2S (g) created using a
series of diluted HCl solutions. This approach appeared to be more effective and will be
discussed here.

Approximately 5 g NazS were placed in four beakers. A set of solutions was
prepared, diluting the concentrated HC125%, 50%, and 75% with water. Approximately
5 mL of each solution was added to a separate beaker, with SmL of the full strength HCl
solution added to the fourth beaker. The purpose in doing so, was to vary the
concentration of the H2S (g) the samples were exposed to. Thus, Beaker 1 theoretically
contained the least concentrated H2S environment, since the least concentrated HCl
solution was added. Four samples were prepared for each pigment (G13 and Y12). The
samples were suspended in the beakers for 2 minutes. The samples were removed and
analyzed immediately. We had hoped that exposing the samples for the same time
period, to varying concentrations of H28 (g), would result in a set of samples with a range
of discoloration, resulting from the formation of PbS.

The results of the Y12 samples were comparable to the G13 samples. For
simplicity sake, only the G13 results will be presented. The results of 613 were chosen,

since the pigment was previously discussed in Chapter 4.

220

208

 

 

247

Relative Intensity

25 263

 

 

175 208 250

247

263

 

Relative Intensity

225

 

 

 

 

175 . 250

 

432
szozm
447
308 j
l
325 mm 400 475
332
mom
4:447
i
l . -
325 mu 400 475

L

530

 

550

b)

530

 

A

550

Figure 6.5: The partial positive ion LD mass spectra of the G13 pigment in linseed oil
on paper samples exposed to 2 minutes H23 (g) created using an HCl
solution diluted; a) 75%; b) 50%.

221

332

 

 

 

 

 

 

 

 

 

 

 

 

 

 

a)
208
g. 263
3 szsc1+
'5 Pb28+ 483
.5 446
g 301
a: 240 360
., P112320“ Pb3s2C1+
515 I
|‘ ' .ita I...“ l l' Ji . t ' ..
175 290 405 m/z 520 635 750
332
b)
29
.5
Q)
'5 240
GB
'5
a:
304 483
208 386
268 515
448 . 721
l.-.L. .i ... , JLL_1 , - - .. L..-‘ . .2 A .2 . _ _ -1- A .....
175 290 405 mm 520 635 750

Figure 6.6: The partial positive ion LD mass spectra of the 613 pigment in linseed oil on
paper samples exposed to 2 minutes HzS (g) created using an HCl solution

diluted; 3) 25%; b) 0%.

222

The positive ion LD mass spectra of the 613 samples exposed to the less
concentrated H28 environments (Beakers 1 and 2) are shown in Figure 6.5. The LD mass
spectra of the 013 samples exposed to the more concentrated H2S environments (Beakers
3 and 4) are shown in Figure 6.6. The spectra in Figure 6.5 resemble the positive ion
mass spectrum of the initial G13 sample in Figure 6.1a. No evidence of PbS is noted,
however the paper peaks are becoming more evident. Evidence of PbS only appeared in
the positive ion LD mass spectra of the G13 samples exposed to the more concentrated
H2S (g) environments. Peaks at m/z 446 representing szs+ replaced the peaks at m/z
447 representing szOzH+ ions. Although it was not demonstrated in this example, in
other studies, one may argue that there is an overlap of the two sets of peaks. This is
more feasible. Additional evidence of PbS appeared at m/z 483, 515, and 721.

In the negative ion mass spectra of these samples, there appears to be a more
gradual appearance of the ions representative of PbS. The negative ion LD mass spectra
of the G13 samples exposed to the less concentrated H2S (g) environment (Beakers l and
2) are shown in Figure 6.7 and the samples exposed to the more concentrated H2S (g)
environment (Beakers 3 and 4) are shown in Figure 6.8. The spectra in Figure 6.7 are
similar to the negative ion mass spectrum of the nonexposed G13 sample in Figure 6. lb.
The peaks representing Cer,' are the most abundant peaks in both of the mass spectra,
however there is the appearance of a small peak at m/z 313, representing PbCl3' ions in
Figure 6.7a. The peak appears to be increasing in intensity in Figure 6.7b. As the
concentration of H28 (g) increases, the spectra change more dramatically, which is
demonstrated in the negative ion mass spectra shown in Figure 6.8. The Cer,’ ions are

no longer generated by the sample, and the peaks representing PbCl3‘ ions are the most

223

 

 

200

 

 

 

 

 

 

 

 

 

3)
3. 100
'2
3
.5
0
>
2:!
2
o
a:
117 “’05
l 135 134 ll
.1L , .L i 1 , - 29,3903}
75 130 185 m” 240 295 350
200
b)
e»
E
O
.. 100
.5
0
>
’5
.53
0
9‘ I
117 135 l PbCI3-
' I 313
~ 2 284
L 68 30° ,
75 130 185 m/z 240 295 350

Figure 6.7: The partial negative ion LD mass spectra of the 613 pigment in linseed oil

on paper samples exposed to 2 minutes H2S (g) created using and HCl

solution diluted; a) 75%; b) 50%.

224

 

 

 

PbCl3-

 

 

 

 

 

 

 

 

 

313
315
5*
E. 31
=
u
2
'3 145 317
1:
a:
157
131
93 161
iL 183
A 11 ' .. 1. .ALL “A -f Y A.“ #1 .1...
75 130 185 mlz 240 295
PbCl3'
313
315
2.:
.5
3
'5
55 311
a:
157
116 317
139
93 145 1 0
82
3sz 3. 371 7,215 , Jr
75 130 185 mlz 240 295

336

336

 

 

 

350

Figure 6.8: The partial negative ion LD mass spectra of the 613 pigment in linseed oil
on paper samples exposed to 2 minutes H2S (g) created using and HCl
solution diluted; a) 25%; b) 0%.

225

intense peaks in the mass spectra. The peaks in the m/z 130-185 range have not been
identified, but their isotopic patterns suggest that they are chlorinated products.

One more 613, suspended in linseed and applied to paper, sample was subjected
to Beaker 4’s conditions for 15 minutes. The darkened sample was then treated with
enough H202 to return the sample to its original green color. Positve and negatitve ion
LD mass spectra were obtained of the sample and are shown in Figure 6.9. In the
positive ion mass spectrum (Figure 6.9a), peaks representing PbS ions have disappeared
and the spectrum is abundant with peaks characteristic of a Pb-containing pigment not
exposed to H28 (g). The peaks at mlz 447, representative of szOzH" ions, have replaced
the peaks representative of szS" ions at m/z 446. This is one example in which the
isotopic pattern is consistent with an overlap of the two sets of peaks. In the negative ion
mass spectrum (Figure 6%), the Cery' ions have returned, and the peaks representing
PbCl3' ions are still present, but have significantly decreased in intensity.

Like the uncontrolled H28 study, in which the HCl solution was diluted to
decrease the concentration of H2S gas in the environment, an experiment was designed
using diluted H202 solutions. Equal amounts of a range of diluted samples were applied
to a set of 613 in linseed oil on paper samples, exposed to H2S (g) for 15 minutes. The
hope was to see a gradual transformation to the spectra in Figure 6.9. This was not

successful.

226

208

 

 

247 a)
i?
3
.5".
Q, 430
.>
‘3 432
“a 332
a:

 

 

 

 

 

 

 

 

 

 

 

175 250 325 mlz 400 475 550
200
b)
5‘
E 100
=
fl
0
>
3:
a:
'5
a:
117 135 184 PbCl3'
I , l I 284 313
75 130 185 mlz 240 295 350

Figure 6.9: The LD mass spectra of G13 in linseed oil on paper exposed to 15
minutes H2S (g) and treated with H202 (l); a) the partial positive ion mass
spectrum; b) the partial negative ion mass spectrum.

227

By completing this experiment, it does not appear that we can effectively monitor
the transformations using LDMS analysis. Additionally, without being able to obtain LD
mass spectra of a real sample that has undergone this transformation, it is difficult to
determine if we were effectively recreating the reaction. Part of the reason this
experiment is referred to as the “uncontrolled H2S study” is simply because there are too
many variables involved. As will be demonstrated in the following section, the presence
of some of the H28 (g) reactants (HCl and H20) may have a greater influence on the rate
and type of reaction occurring, beyond our understanding. For instance, did the increased
HCl concentration have a greater effect on the increased intensities of the peaks
representing PbClg' ions, than the increased concentration of H25 (g)? For these reasons,

a new controlled method was sought.

Controlled H 2S Exposure Studies

One way to eliminate the majority of the variables involved in the “uncontrolled
H28 studies” was to purchase H2S (g), as opposed to making it in the lab. A lecture bottle
of H2S (g) (AGA) was connected to a vacuum manifold using tubing. A tank of
compressed air was also attached. The vacuum manifold was connected to a rotary
pump, which decreased the pressure in the manifold to approximately 300 mTorr. The
samples were placed in a round flask, which was then attached to the manifold. The
valve connected to the H25 (g) was opened, allowing approximately 65 Torr H2S (g) into
the manifold. The valve connected to the flask holding the sample was then closed, and

the manifold was flushed with air. Once the sample had reacted, the valve to the flask

228

was opened and flushed with air to halt the reaction. Samples were analyzed
immediately using LDMS.

The G13 pigment and pure PbCrO4 were used in this study. Contrary to the
previous uncontrolled experiments, a significant amount of time was required for a
visually-detectable reaction to occur. Approximately 40 hours elapsed before the PbCrO4
sample acquired a darkened appearance. Black spots did not appear on the green 613
sample until about 65 hours had gone by. Both samples were analyzed following 65
hours exposure to H2S (g), however only the PbCrO4 results will be presented here.

Positive and negative in LD mass spectra were obtained of PbCrO4 in linseed oil
on paper, exposed to H2S (g) for 65 hours, and are shown in Figure 6.10. There is very
little evidence of PbS in both the positive and negative ion mass spectra. In positive ion
mode (Figure 6.10a), an overlap of the peaks representing Pb202H+ and Pb23+ ions is
noted. In the negative ion mass spectrum (Figure 6.10b), a very small set of peaks is
noted at m/z 275, representing PbSCl' ions. Complications with the vacuum manifold
set-up and the pressure gauge contributed to the end of this study.

Which set of conditions is more appropriate — the controlled or uncontrolled? Is it
practical to consider the reaction taking place in the absence of O; or other contaminants
(i.e. Cl)? Was the uncontrolled environment close to what occurs naturally? The
equipment available in the laboratory was not sufficient to recreate natural conditions a

painting may be subjected to.

229

208 447

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

449 a)
e .. ..
.5 I 463
W... *J ’
.5 433 A 443 45; 463 :73 483
o
.5 430
2
a.)
9‘ 73
447
113 225
147 263 .
388 465 5 2 654 754
L‘_ . l . . . i 1 r
50 195 340 mlz 485 630 775
100 m
284 b)
200
269
r? 285
E»
,5 270
a) 2“ 275 m
.3, 184 ill Miami 282 m
a
Ila 117 A. .“L 14M LANA“ -
M 260 267 274 281 288 29s
35 135 268 284
275 367
‘ LL 4'4 .1 .LLL 4.27% L . A A
0 105 210 ml 315 420 525
Z

Figure 6.10: The LD mass spectra of PbCrO4 in linseed oil on paper exposed to 65 hours
H28 (g); a) the partial positive ion mass spectrum; b) the partial negative
ion mass spectrum.

230

Summary

The purpose in pursuing this project was to determine if LDMS could be used to
characterize the products formed, when pigments in a paint mixture react with their
environment (gases or other pigments). Was the project successful? We successfully
produced darkened samples by exposing them to H25 (g). Peaks were positively
identified which are characteristic of PbS, the expected product. The darkened
appearance was reversed upon the addition of H202. Again, the mass spectra changed,
providing evidence that the unwanted PbS product had been converted to a white Pb
pigment. So, the study was a success. The question is whether or not we recreated what
occurs naturally in a painting exposed to H2S (g). A donated sample from a painting
which has suffered from H2S (g) exposure over several years would answer this question.
Ultimately, the H28 studies were abandoned, since the focus of the project changed to

using LDMS for the identification of pigments.

231

References

1) Clay HF, Watson V. The crystal structure of lead chromes. Journal of the Oil and
Colour Chemist’s Association 1948;31:418-422.

2) Watson V, Clay HF. The light-fastness of lead chrome pigments. Journal of the Oil
and Colour Chemist’s Association 1955;38:167-177.

3) Petushkova JP, Lyalikova NN. Microbiological degradation of lead-containing
pigments in mural paintings. Studies in Conservation 1986;31:65-69.

4) Giovannoni S, Matteini M, Moles A. Studies and developments concerning the
problem of altered lead pigments in wall painting. Studies in Conservation 1990;35:21-
25.

5) Koller M, Leitner H, Paschinger H. Reconversion of altered lead pigments in alpine
mural paintings. Studies in Conservation 1990;35:15-20.

6) Erkens LJH, Hamers H, Hermans RJM, Claeys E, Bijnens M. Lead chromates: a
review of the state of the art in 2000. Surface Coatings International Part B; Coatings
Transactions 2001;84: 169- 176.

7) Sharkey J. Chemistry of postage stamps: dyes, phosphors, adhesives. The United
States Specialist 1990 Augustz469-482.

8) Standage HC. The artist’s manual of pigments. Philadelphia: Janentzky & Weber
1886.

232

Chapter Seven: Newspaper Ink

History of the Printing Press

In early Egyptian, Greek, and Roman civilizations, handwritten copies of
important manuscripts were extremely common. A number of factors escalated the need
to develop a method to mass produce copies of written material in a timely manner.
Religion was the main contributor for this need. The rise of the prosperous literate class
during the Renaissance period, followed by the religious controversies resulting from
Martin Luther’s Reformation, placed a high demand on the printing press.
Pamphleteering in particular was a key component in spreading religious propaganda
during this time period (1).

The Chinese are credited for inventing the first printing press as early as 200 AD.
They employed a wooden block technique, using a movable type. Letters and pictures
were Cut in relief in wooden blocks and arranged in sequence. This method had been
used previously for printing on textiles. Much of their success was based on the type of
substrate available. The invention of paper by the Chinese in 195 AD. proved to be a
main contributing factor for their success. In the western world, writing substrates were
limited to papyrus and vellum (animal skin), which were expensive and unsuitable for
printing on. Early printing inks in the east were strictly aqueous (1).

By the 12th century, papermaking had found its way to the west and became
popularized in the 13th and 14th centuries, paving the way for the development of the
printing press in the west. Officially, Johann Gutenberg, of the German city of Mainz is

considered the inventor of Western printing in 1450. A movable metal type cast was

233

employed and the design of the press modeled that of a winepress. Unlike the east, the
west used oil-based inks from the start. Juan Pablos is given credit for bringing the
printing press to the New World in 1539, when he established a printing press in Mexico
City. Later, in 1628, Stephen Day assisted in the establishment of the Cambridge Press in
Massachusetts Bay (1).

Improvements to the original design of the printing press have been made
throughout the years leading to four main types of high speed printing techniques still
used today. These include the letterpress and flexography, which employ a raised

surface, and lithography and direct lithography, which employ a flat surface.

Printing Methods

The letterpress is a more simple and primitive technique which employs a rotary
press with an elevated image area printing plate. Rollers transfer ink onto the raised
surfaces and the image is transferred onto the newsprint using an impression cylinder.
The cylinder consists of a hard rubber material, which is made to withstand numerous
impressions. This method is simple and effective, however one disadvantage is that only
one side can be printed at a time (1,2).

Flexography is another example of a raised surface printing technique, resembling
the letterpress in many aspects. With this method a softer image plate is employed than
that used in the letterpress. Consequently, an aqueous ink is used which is distributed
uniformly and in the proper amount onto the printing plate using an anilox cylinder. The
anilox cylinder is equipped with a scraper blade which cleans up excess ink applied to the

plate, and recycles the ink back to the ink fountain. As in the letterpress, only one side of

234

the newsprint may be printed at a time, however the image is transferred directly from the
printing plate to the newsprint, eliminating the need for an image cylinder (1,2).

With the phasing out of raised-surface mechanical printing techniques,
lithography (offset) has surfaced as the most common type of printing press used today.
Aloys Senefelder, a German map inspector, developed the underlying principles behind
lithography printing, which he termed “chemical printing”, in the late 18‘h century.
Modern day lithographic offset printing presses are much more technical, using
computers to automate and generate the image process (1,2).

The design of a lithographic printing press employs a flat printing plate, with two
chemically different areas. The non-printing area is hydrophilic, while the image area is
hydrophobic. The printing plate is typically aluminum, with a thin surface coating of a
photopolymer. The light sensitive material undergoes a solubility change when exposed
to an intense source of blue or UV light. The image is transferred directly to the surface
by exposing the plates directly, as in a graphic arts camera or by a computer controlled
laser beam. During the printing process, an oil based ink and an aqueous fountain
solution are applied to the plate. The ink is applied in a thin layer using rollers, and wets
the image area. The fountain solution wets the non-image area. The solution may be
sprayed directly onto the plate, transferred from a high speed brush, or by a molleton or
sock roller (1,2).

The chemistry occurring on the plate is crucial to the quality of the completed
copy. Oil and water do not mix, but here they must interact. Thus, the ink must possess
specific properties which enable it to emulsify the fountain solution so it can wet the

image area. This ensures uniform transfer of the ink. Additionally, the fountain solution

235

must possess detergent properties enabling it to wash away any ink deposited in the non-
image area. Thus, formulating the ink and fountain solutions is an exact science (1).

Once the ink is properly applied to the printing plate, the ink is transferred to a
blanket cylinder. The cylinder is covered with a compressible material, allowing it to
conform to the printing surface. The image is then transferred to the newsprint. With
this technique, double sided pages may be printed simultaneously (1).

Direct lithography is the same as lithography printing, except as the name implies,
the blanket cylinder is missing, allowing for the direct transfer of the image from the
printing plate to the newsprint. However, the absence of the blanket cylinder limits this
method to strictly paper. The presence of the blanket extends the printing substrates to

plastics and metals (1,2).

Newspaper Ink

In general, newspaper inks consist of pigments, resins, an oil or other carrier, and
additives. The resin acts as both a dispersing agent and a binder for the pigment. The oil
used is typically treated napthenic petroleum or poppyseed oil (3). These are non-drying
oils, meaning that their purpose is strictly to carry the ink onto the paper. All printing
methods, discussed thus far in this chapter, dry by absorption of the oil into the paper.
No heat is applied and no other volatile components are added. Excluding flexography
printing, the ink is not cleaned off of the press in between copies. Inclusion of volatiles
in the ink formulations would result in the ink drying out directly on the rollers,
ultimately causing printing problems. Other additives are included to control pigment

wetting and dispersion, viscosity and flow characteristics, and provide proper ink/water

236

balance (3). It should be noted that in 1979 the board of directors of the American
Newspaper Publishers Association ordered the development of a soy-based newspaper
ink, since it has a lower VOC (volatile organic compound).count (4—6). Soy ink is
composed of non-toxic soybean oil, extracted from soybeans produced domestically.
Soybean oil is the same oil found in edible items, such as salad dressings and
mayonnaise. However, the addition of petroleum-based pigments and resins creates a
non-edible soy ink (4). The ink was not introduced until 1987. Approximately 90% of all
US. daily newspapers are currently printed using soy-based ink, when printing in color.

Carbon black is commonly the primary pigment found in black newspaper ink. It
is produced by cracking oil in a continuous furnace. The process is highly controlled in
order to produce a specific grade of pigment, having a specific particle size and structure.
When manufactured, carbon black is dry and packed together forming aggregates or
agglomerates, resulting in the need for a dispersion agent, or resin, to form the
appropriate pigment particles. The resin is the most expensive component of the ink,
however it has the greatest effect on rub resistance of the ink on the final product. Thus,
the manufacturer must find a balance between cost and quality, since rub resistance
increases with increasing resin content, but so does cost. Other factors affecting rub
resistance which can be used to offset the resin content include viscosity, ink film
thickness, paper quality, humidity, and the time after printing (3).

Although only mass spectral data will be presented for black newspaper ink in
this chapter, color appears in newspaper ink quite frequently. The three primary colors
(cyan, magenta, and yellow) are used to produce all of the colors seen in a newspaper.

The colors are synthetic organic pigments. Typical pigments used for each color include

237

phthalocyanine blue, Lithol Rubin, and Diarylide yellow (3). Black, cyan, magenta, and
yellow colored newspaper inks were obtained from the Lansing State Journal. Positive
and negative ion LD mass spectra were obtained of all four samples on paper, however

the primary focus of this side project is black newspaper ink.

Experimental

Positive and negative LD mass spectra were obtained of the black ink on
numerous newspapers dated between 1923 and 2002. This chapter will focus on the
positive ion results of a current newspaper (2002), a 1986 newspaper, a 1963 Weekly
Reader, and an issue of the Scientific American Journal, allegedly from the late 19‘h
century. Spectra were obtained of the ink directly on the paper substrate, and of the paper
by itself as a control. A l in2 section was removed from each sample and taped to the
modified sample plate. The user parameters as cited in the instrumental section were

employed.

Results
A Current Newspaper

Beginning with the most recent sample, a current copy of the Michigan State
University’s State News was obtained and analyzed. The paper had been freshly printed
the night before. The positive ion LD mass spectrum of the ink-on-paper sample is
shown in Figure 7.1a. In the positive ion mass spectrum, there are notably two clusters of
peaks present. The first cluster is extremely complex and is located between

approximately m/z 240 to mlz 450. The most intense peaks in the cluster are separated

238

by 14 mass units, suggesting that the ions may be a homologous series. The complexity
and distribution of peaks is characteristic of a mass spectrum of a polycyclic aromatic
hyrdrocarbon (PAH), which one would expect to find in either a petroleum or soy-based
product. For example, petroleum oil consists of three major types of hydrocarbons,
which include paraffins, olefins, and aromatics (7). Several of these compounds, having
varying vaporization and boiling points, constitute the oil, resulting in an array of peaks
in a mass spectrum. Many of these compounds may only differ in chain length and
according to the number of degrees of unsaturation, resulting in peak separations of 14

and 2 mass units, respectively.

239

291

 

 

 

 

 

277 305
319
E? 333
’53
::
£3 263
£5
a; 359
I:
53 .
o , ‘
a: 258 j ‘ j 13 l 374
‘ ‘ j j | l 387
l i i‘ ii‘ , ‘
244 ‘ , j ‘ ‘ W1 j 401
. L‘ ‘ ‘ i ‘ M
l .j i i “‘ i it" “i ' “1“ ‘ ‘
, a mi.idhhhmJwLJMWMaunmhuunmmhli
225 295 365 435
nn/z
>.
a:
U}
c:
O
H
£5
6)
>
0:
£3
0)
a:

 

250 320 390

rng

460

a)
550
522
. A
505
522
550 b)

 

 

 

530

 

575

 

600

Figure 7.1: The partial positive ion LD mass spectra of a recent issue of the State News

a) the black ink; b) the paper.

240

Nearly a decade ago, Zenobi and his research group investigated the used of
LDMS for identifying polycyclic aromatic hydrocarbons (PAHs) in a petroleum pitch
sample (8). A small percentage of the PAHs are volatile and may be detected in a GCMS
experiment, which is a commonly employed technique for characterizing the volatile
hydrocarbons of gasoline and other fuels. Here, Zenobi employed a two-step laser
desorption technique to detect the nonvolatile, higher molecular weight species. In this
technique, desorption of the PAHs was achieved using an IR (C02) laser, followed by
ionization using a UV laser (308 nm). They designated this method as “L2MS”.

Figure 7.2 shows the positive ion mass spectrum of the petroleum pitch sample
generated using the L2MS method. The mass spectrum is similar to the mass spectra
obtained of newspaper ink. In Figure 7.2, Zenobi and coworkers identified a PAH
skeleton beginning at m/z 302. The peak at m/z 302 was identified as dibenzo[a,h]
pyrene. The higher mass peaks in the mass spectrum were assigned, based on the simpler
dibenzo[a,h] pyrene structure. The higher mass compounds were the-result of alkylation
([M+14] for methylation, [M+28] for double methylation or ethylation), as well as the
addition of ethylene bridges (M+24), ethyl bridges (M+26), or benzo groups (M+50).

We will apply their method of assigning peaks to our own results.

0.30- 326
0.25- 352 Mioniz) =- 308 nm

302
0.204

o.15~ 426
0.104 /

0.05.1 216

Signal (V)

0.00« i ‘ «~—
200 250 300 350 400 450 soc
Mass (amu)

 

Figure 7.2: L2MS spectrum from a petroleum pitch sample.

241

A second notable set of peaks separated by 28 mass units occurs at m/z 494, 522,
and 550. While these peaks remain unassigned, they were initially suspected to represent
a component(s) of the paper. The positive ion LD mass spectrum of the paper is shown
in Figure 7.1b. The same cluster of peaks at m/z 494, 522, and 550 are noted, lending
support to the hypothesis that the ions are generated from the paper. It is also important
to note here that the peaks in the lower mass region of the spectrum, believed to represent
the petroleum oil carrier, are significantly more intense than the peaks in the higher mass

region of the spectrum, at this point believed to represent components of the paper.

Aged Newspaper Ink

Having had experience with aged modern ballpoint pen ink samples (9-11), we
wanted to investigate if the mass spectra of newspaper ink changed over time. To do so,
a newspaper dated 1986 and a “Weekly Reader” dated 1963 were analyzed. The positive
ion LD mass spectra of the black ink on the two documents are shown in Figures 7.3a and
7.3b, respectively. Again, two distinct clusters of peaks are observed in both spectra.
The lower mass clusters appear to have shifted to a higher mass, and the distribution of
peaks has become more narrow and less intense as compared to the higher mass cluster of
peaks around m/z 550. The effect is most obvious in the mass spectrum of the 1963

newspaper ink (Figure 7.3b) and will be discussed shortly.

242

 

 
 

 

 

 

 

 

 

3)
550
CZSHM” C25st+
.4: C24H12+ | C26H30+
E 314 328
‘H 356— C28Hzo+
,5 342 494
0 384
>
.5 38
.‘2
Q)
a:
200 275 350 mlz 425 500 575
522 550

13)
>3
H
'2
a, 494
H
.5
Q
.2
H
E.
0
fi

298 12 466
245
760
225 345 465 m/z 585 705 82s

 

 

 

 

Figure 7.3: The partial positive ion LD mass spectra of black newspaper ink a) a 1986

sample; b) a 1963 sample.

243

The positive ion LD mass spectra of all three newspaper ink-on-paper samples all
contain the higher mass group of peaks at m/z 494, 522, and 550. However, the peaks are
absent in the positive ion control LD mass spectra of the 1986 and 1963 samples (spectra
not shown). This suggests that these peaks may represent another component of the ink,
such as the resin. Newspaper ink “dries” by the mechanism of absorption, meaning that
components of the ink are continuously migrating into the paper over time. Let us
consider the current newspaper ink sample. The ink was fresh and beginning to “dry”.
The carrier component (oil and resin) began seeping into the paper away from the black
pigment deposited on the paper. The control spectra were taken adjacent to the ink
samples. In the older samples, the resin may harden or become trapped in the complex
cellulose structure of paper, making desorption and ionization of these molecules very
difficult. In the case of the current newspaper sample, the ink is still “wet”, perhaps
making it possible to observe other components of the ink, such as the resin.

Another important, and somewhat unexpected observation is that the mlz values
of the most intense peaks in the positive ion mass spectrum of the current newspaper ink
are odd and those of the 1986 sample are even. Aromatic hydrocarbons have the ability
to generate either molecular ions (M‘") or protonated molecules (M+H)+. If the mlz value
of a peak is even, it represents the M’”. If the value is odd, there exists a couple of
possibilities. The first is that the peak represents the protonated molecule. The second is
that the compound is not a pure PAH, but actually contains an odd number of nitrogen
atoms. Assuming that there are only PAHs present, potential peak assignments can be

made.

244

Peak Assignments

In regards to assigning peaks, let us focus on Figure 7.3a. Looking at the wide
distribution of peaks in the positive ion mass spectrum of the 1986 newspaper ink, there
appears to be a series of related peaks separated by 14 mass units, beginning at m/z 300.
If the compound was a saturated alkane, it would have the formula C22H36. Considering
that alkanes do not absorb in the UV, this assignment is improbable. Aromatic
compounds tend to absorb efficiently in the UV region. For m/z 300, one possibility
would be C24H12. A proposed structure, illustrated in Figure 7.4, represents coronene.
The next possible compound would be Csto. Considering compounds containing a
lower number of C atoms, the next possible compound would be C23H24, followed by
C22H36 (the alkane).

Figure 7.5 shows an enlarged section of Figure 7.3a in the m/z 300 region. The
structure in Figure 7.4 can not lose any additional H atoms. The peaks greater than m/z
300 can be accounted for by saturating some of the double bonds of coronene. For
example, the peak at m/z 302 would represent C24H14. Odd m/z values are difficult to
assign. In this case, m/z 301 could represent either the protonated coronene molecule or
a fragment ion, if the ion’s composition is limited to only C and H atoms. The other

possibility is that the ion contains an odd number of N atoms (C23NH11).

245

Figure 7.4: The structure of coronene.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

314
300
2
,3? 287 302 31
U)
a 298 301
3 286 l n ”9... ’1’
.5
Q
.3 303 j
53 296 305
4)
°‘ U U WV”
285 ' 291 ' 297 mlz 303 ' 309 ' 315

Figure 7.5: An enlarged portion of the positive ion LD mass spectrum of the 1986
ink sample.

246

Additional assignments were made in Figure 7.3a, however they have not been
confirmed. It is also difficult to propose a structure for these compounds, considering the
possibilities are large. Consulting the literature, the mlz values in the newspaper ink
spectra we generated do not match any of the previously observed gas phase PAH ions
documented (12). Identifying PAHs in petroleum based products has proven to be a
difficult task by other research groups as well (13-15). In fact, the complexity of the
petroleum mixtures has prompted one research group to identify the field as

“Petroleomics”, in comparison to the more well-known field of Proteomics ( l6).

Weathering

Returning to the observation noted earlier concerning the relative intensities of the
peaks in the lower and higher mass clusters, we propose an explanation. It appeared as
though, as a newspaper ink “aged”, the distribution of peaks shifted to a higher mass,
became more narrow, and decreased in intensity relative to the “paper peaks”. There is a
phenomenon associated with drying oils, termed “weathering”. Weathering is used to
describe the processes which alter an oil’s characteristic fingerprint. These processes
include, evaporation, dissolution, photochemical oxidation, and biodegradation, to name
a few (17). Evaporation and dissolution are the main contributors to the weathering
process (18). Although the two processes are in direct competition with each other, the
initial rate of weathering is proportional to the concentration of smaller, more volatile
saturated or aromatic hydrocarbons present, and results from evaporation. Larger
components are less volatile, but also less soluble, resulting in dissolution. These

components are not absorbed in the paper, and are “washed” away by physical processes.

247

Consequently, in a mass spectrum, the fingerprint distribution of peaks shift to higher
masses and become more narrow, resulting from the evaporation and dissolution
processes. The peak intensities would also become less intense, since the concentration
of the carrier would decrease as a result of evaporation, dissolution, and absorption into
the paper (19).

Initially, it appeared as though the spectra of the three newspaper inks supported
the weathering theory. However, more samples needed to be analyzed to test this theory.
The spectra of these samples (not shown) did not always follow the same trend. The
possible PAH distribution was present in each spectrum, however a pattern could not be
established, which is not all that surprising. As with writing inks, each manufacturer has
its own formula, so we would not expect the mass spectrum of every newspaper ink
freshly applied to paper (never mind aged ink) to be identical. We also need to keep in
mind the switch to a soy-based ink in the late 19808, early 19908. A more extensive
study could be performed to first determine if LD mass spectral dataicould provide a
“fingerprint” of the different types of newspaper inks. If so, then other factors, such as
the type of paper the ink is printed on and storage conditions, can be considered to

determine if the aging of newspaper ink can be characterized.

Scientific American Journal

In the initial stages of the newspaper ink project, it appeared as though the
weathering process of complex drying oils, such as petroleum oil, could be illustrated
using LD mass spectral analysis. The assumption was based on the results of the 2002,

1986, and 1963 samples discussed in greater detail in a previous section. To test this

248

aging picture, an issue of the American Scientific Journal, dated 1886, was analyzed.
The authenticity of the document was immediately suspect, based on the perfect
condition of the document. The document was folded, therefore different portions of the
ink-on-paper were analyzed. The spectra of the different sections sampled were
predominantly the same, therefore a representative positive ion mass spectrum was
chosen and shown in Figure 7.6. The spectrum reflects significant differences in the
newsprint ink formula encountered thus far in the study.

In the positive ion mass spectrum, the absence of the standard PAH peaks is
noted. Rather, the spectrum shows evidence of the cationic dye, methyl violet, at mlz
372. Methyl violet is documented as a commonly used toner in modern day ballpoint pen
and ink jet printer inks. Additional peaks, at m/z 250 and 275, indicate that more than
one colorant has been included in the ink mixture. This sample was not the first time
these peaks (m/z 250 and 275) appeared in the positive ion mass spectrum, with a
complimentary peak in negative ion mode at m/z 250 (spectrum not shown). The peaks
frequently appeared in yellow colored samples. In fact, the peaks appear to represent the
yellow colorant in the soy—based yellow ink used to print the Lansing State Journal

(Figure 7.7).

249

 

372

 

 

 

 

 

 

 

 

>.
H
’r'n"
r:
G)
H
.5
3
'4: 250275
.2
Q)
m 358
M..-..._...... .._ . L .L - 4.. ...k.‘.. .. h _
100 200 300 400 500 600
mlz

Figure 7.6: The positive ion LD mass spectrum of the Scientific American Journal.

Figure 7.7 shows the positive ion mass spectrum of the yellow ink applied to
paper, which was supplied by the Lansing State Journal. Much time and effort has been
devoted to determining the identity of this colorant. A structure has been proposed
(Figure 7.8), based on the m/z value of the ion, as well as its isotopic pattern. However,
this compound is categorized as a rare chemical in the Aldrich catalog. It is doubtful that
a rare chemical would be employed in this application. The isotopic distribution is
consistent with the compound containing either 2 Cl atoms or 1 Cu atom. The colorant
remains unidentified, however the important point here is that it is most likely a modern
day yellow pigment. In contemporary ink jet printer inks, it is common to have a mixture

of yellow and violet dyes to produce black ink. The chemical analysis of the document

250

using LD mass spectral data confirmed our suspicions that the document is a modern day

copy and not made from carbon black newsprint ink.

 

 

 

 

 

 

 

 

 

  

 

 

 

 

 

52.2
43 250 550
275
>. 93
3:
v: 438
5 67
H
.5
01>.” 494
H
g 105
Q
m
81 119 628
i I . i i k 1 I LL21. '._ 1L
0 130 260 390 520 650

mlz
Figure 7.7: The positive ion LD mass spectrum of yellow newspaper ink on paper.

Cl
N
\\N

Cl

Figure 7.8: A proposed structure for the compound represented at mlz 250.

251

References

1) The history of printing. http://www.usink.com/history_printing.html
2) How is a newspaper printed, vol XXIV. US Ink, 1998, 1-5.

3) What is ink? http://www.usink.com/what is ink.html

4) Soy ink talking points. http://www.soyink.com/inktalking.html

5) For printers. Kohler’s comer - soy ink article.
httg/lwww.ippaper.com/gettips l__(c soy.html

6) Brown MS. Printers shift to environmentally friendly soybean ink. Memphis Business
Journal 1999.

7) Wigger JW, Torkelson BE. Petroleum hydrocarbon fingerprinting —- numerical
interpretation developments.
httpzllwww.bioremediationgroup.org/Bioreferencesfl‘ier1Papers/petroleumhtm

8) Zenobi R. Advances in surface analysis and mass spectrometry using laser desorption
methods. Chimia 1994;48:64—7 1.

9) Grim DM, Siegel J, Allison J. Evaluation of desorption/ionization mass spectrometric
methods in the forensic applications of the analysis of inks on paper. J Forensic Sci
2001;46:1411-1419.

10) Grim DM, Siegel J, Allison J. Does ink age inside of a pen cartridge? J Forensic Sci
2002;47:1294-1297.

11) Grim DM, Siegel J, Allison J. Evaluation of laser desorption mass spectrometry and
UV accelerated aging of dyes on paper as tools for the evaluation of a questioned
document. J Forensic Sci 2002;47:1265-1273.

12) Lias S, Bartmess JE, Liebman JF, Holmes JL, Levin RD, Mallard G. Gas-phase ion
and neutral thermochemistry. J Physical and Chemical Reference Data
1988;17:649,794-95.

l3) Balasanmugam K, Viswanadham SK, Hercules DM. Characterization of polycyclic
aromatic hydrocarbons by laser mass spectrometry. Anal Chem 1986;58:1102-1108.

l4) Rodgers RP, Lazar AC, Reilly PTA, Whitten WB, Ramsey JM. Direct determination

of soil surface-bound polycyclic aromatic hydrocarbons in petroleum-contaminated
soils by real-time aerosol mass spectrometry. Anal Chem 2000;72:5040-5046.

252

15) Herod AA, Stokes BJ, Hancock P, Kandiyoti R, Parker JE, Johnson CAF, John P,
Smith G. Laser-desorption mass spectrometry of standard polynuclear aromatic
hydrocarbons and fullerenes. J Chem Soc Perkin Trans 1994;22499-506.

16) Rodgers RP, Klein GC, Hendrickson CL, Marshall AG. Petroleomics: environmental
forensic applications of high resolution FT-ICR mass spectrometry. In: Proceedings
of the 55‘h American Academy of Forensic Sciences Annual Meeting. Chicago, IL
February 17-22, 2003.

17) Forensic oil analysis (fingerprinting).
http://www.rdc.uscg.gov/mslpages/method.html#source.pfitr

18) Smith CL, MacIntyre WG. Initial aging of fuel oil films on sea water. In: Proceedings
of Joint Conference on Prevention and Control of Oil Spills. Sheraton Park Hotel,
Washington, DC. June15-17, 1971.

19) Rodgers R, Blumer EN, Freitas MA, Marshall AG. Complete compositional
monitoring of the weathering of transportation fuels based on elemental compositions
from fourier transform ion cyclotron resonance mass spectrometry. Environ Sci
Technol 2000;34: 1671- 1678.

253

Chapter Eight: Practical Experimental Aspects

The Problem

The usefulness of LDMS for the identification and characterization of colorants
found in inks and paints, has been well demonstrated thus far. However, what use is the
technique if it is not practical for art conservators and forensic scientists? One major

problem we encounter is sample size. The problem is illustrated in Figure 8.1.

 

Figure 8.1: Illustration of the sample size problem. Samples which can be analyzed
using LDMS are limited to a 6 in2 sample plate.

The figure shows the modified sample plate, which was machined to hold
polyacrylamide gels. Unless the piece of art or questioned document is a stamp or a post-

it-note, we can not introduce the sample into the instrument. There are a couple of

254

solutions to this problem. These include “going big’ and “going small”. The two

solutions will be dealt with separately.

“Go Big”

A MALDI mass spectrometer similar to the instrument in our lab, is shown in
Figure 8.2 (1). One option is to physically increase the size of the ion source housing to
accommodate a large painting. This design is illustrated in Figure 8.3. There are no
limitations in doing so, it simply has rarely been done before. The design could be
modified to hold a wide range of art objects, including pottery and vases. There are some
concerns with this solution. The LDMS experiment is performed at an extremely low
pressure (1045 Torr). Art objects contain volatile components, and are not made to
withstand vacuum conditions. We are concerned that placing an art object in a high
vacuum system could have detrimental effects on the object. These effects include

dehydration and eruption of the samples.

 

Figure 8.2: A MALDI MS similar to the instrument in our laboratory.

255

 

o'. ‘ - a
- . I
.
q 2
. (If-fr
I. .

MS

 

 

/

 

 

 

 

 

 

Figure 8.3: A cartoon of a MALDI instrument with an enlarged ion source housing to
accommodate large samples, such as a painting.

Dehydration is a concept which deals with items such as pottery and vases, in
which water plays an integral structural role. In a specific example, Barbara Berrie, an
inorganic chemist at the National Gallery of Art, Washington, DC, was analyzing a 14th
century Spanish ciborium (a container for the Holy wafer during communion services)
shown in Figure 8.4. The ciborium is made of gilded copper and champléve enamel.
Specific areas of the ciborium, which contained a turquoise glass enamel, were corroded.
Analysis of the turquoise glass enamel revealed a high lead content. Berrie explained
that potassium glass is not stable, and is susceptible to slight fluctuations in the relative
humidity of the surrounding atmosphere. The ciborium was drying out (as would occur
in a high vacuum system). As water from the ciborium evaporated, the remaining
potassium became potassium hydroxide. The hydroxide reacted with CO2 in the air, and

potassium carbonate was formed. The environment around the copper metal became

256

alkaline. Consequently, copper carbonate corrosion resulted. Once the corrosion was
removed, the ciborium was placed in an environment of constant relative humidity, which
was not permitted to fall below 40% (2). The case of the Spanish ciborium is just one

example, which would pose a great risk to analyze using LDMS.

 

Figure 8.4: A 14‘'1 century Spanish ciborium.

Eruption is a concept which deals moreso with paintings. When a painting is
created, air, moisture, and volatile components become trapped under the protective
varnish layer. Chemical changes in the underlying paint layer may result in the
mechanical expansion and protrusions at the surface of the painting. Over time, these
protrusions naturally erupt (3,4). We are concerned that subjecting a painting to high
vacuum conditions, will ultimately accelerate the inevitable eruption process. The goal
of the project is to develop a technique which provides useful data, without harming the

sample. An alternative direction would be to “go small”.

257

“Go Small”

The Analysis of Questioned Documents

In the field of forensic science, questioned document examiners are asked to
analyze the ink on a document, to provide data on either its source or its age. The
document in question is often times admitted in a court of law, and therefore maintaining F
the integrity of the document in its original form is essential. In the field of questioned
document examination, a method for removing ink samples from a document, resulting in

minimal damage to the document, has been approved. Figure 8.5 shows a couple of

 

tools, which are used by questioned document examiners to remove small punches from a
sample. The tool in the top portion of Figure 8.5 is actually used at a local questioned
document laboratory (Speckin Forensic Laboratories). The tool in the bottom portion of
the figure is a more primitive model. We experimented with the Speckin tool, and
removed several punches from a black ballpoint pen ink-on-paper sample, and also from
the red ink on the Koran discussed in Chapter 4. The experiment is illustrated in Figure
8.6. The appropriate mass spectrum was generated in both cases, as when a much larger
sample was employed.

One drawback is that resolution tends to suffer as the sample size decreases.
Often times, this problem may be rectified by tweeking the user parameters slightly. The
important point here is that the removal of small punches is an acceptable sampling
method in the field of questioned document examination, and we have shown that it is
possible to generate an informative mass spectrum from a single punch. It should be

noted that the most commonly used methods by document examiners to analyze ink, such

258

as HPLC, require the removal of 10 to 15 punches to perform one analysis (5,6). With

LDMS, we only need one.

Tools

    

Figure 8.5: Examples of two punches used by questioned document examiners to
remove microplug samples from documents in questions.

259

  

Document

 

 

 

 

>3
:2
U)
=
0
1d
5
d)
>
.5
N
713 4-
°: [28
32 492
64 428 L
460 522
.l . . 3.. 396 I r.
0 110 220 mlz 330 440 550

Figure 8.6: Illustration of the microplug experiment. The negative ion mass spectrum
of the red ink on the 17'h century Koran presented in Chapter 4.

The Analysis of Paintings

In the area of painting analysis, we could “go small” in a couple of ways. At the
Walter McCrone Institute in Chicago, IL, it is common to remove microplug samples
from a painting and analyze the cross-sections using microcopy (7,8). It is stated that if
the diameter of the microplug is less than 0.01 mm, the “damage” to the painting can not
be seen by the naked eye. The other possibility is to remove small paint chips from the
painting.

Dr. Karen Trentelman from the Detroit Institute of Arts, Detroit, MI, was kind
enough to supply us with two paint chips from a painting at the Institute. The sample we

analyzed had the dimensions of approximately 75 x 30 am. The challenge became, how

260

we were going to introduce the sample into the instrument without it being swept away in
the vacuum system. We could not simply place it on a regular gold sample plate and
double stick tape seemed too risky. The solution was to suspend the paint chip in linseed
oil, and deposited the mixture in a standard sample plate well (illustrated in Figure 8.7).
Positive and negative ion LD mass spectra (Figure 8.8) were successfully obtained of the

sample.

Sample Linseed Oil

75 um x 30 W
- .— Sample Plate

Figure 8.7: A cartoon of the paint chip in linseed oil experiment.

In the positive ion mass spectrum (Figure 8.8a), numerous peaks have been
identified containing the element Pb. Due to the absence of any unique Pb-containing
peaks, such as PbCrO4+ in the mass spectrum of the lead chromate pigment, we can not
positively identify the pigment present. Again, note the decrease in resolution as the

sample size decreases.

Switching to the negative ion mass spectrum (Figure 8.8b), linseed oil peaks, as
discussed in Chapter 4, are noted. This could suggest a couple of possibilities. The first
is that the paint mixture contains a pigment, such as red ochre, which does not appear in
the mass spectrum itself, but has a “matrix” quality, which allows us to see the vehicle
component. The other explanation is that the ability to generate linseed oil peaks in the
LDMS experiment is a function of particle size. This observation surfaces again, very
shortly. Overall, we were not discouraged by the fact that we could not identify the

pigments present. Our goal here was to overcome the sampling size problem.

261

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

208 Pb+ a)
Pb,+
5 414
0-
§ Pb02+
E 429
G)
>
0:
g 39 P o + Pb,o_,+
g 69 f2 2 /Pb302+
81 225 436 652
. 709 892
108 ,
. 62° * 874 1113
1.... . .I.L .i I .i IJ 1‘ A
0 400 800 mlz 1200 1600 2000
281 1))
279 28‘
279
>.
H
0-
m s
1::
4)
5H
.5 277
‘5’
-.-.- 27%-
T) 254
m A.
71 250 260 5 270 280 290 300
35 25
1-1-5. - 'l - - .A ,_A'- 1 A- T A A
0 300 600 m/z 900 1200 1500

Figure 8.8: The LD mass spectra of the Detroit Institute of Arts paint chip sample
suspended in linseed oil on a gold sample plate a) the positive ion mass

spectrum; b) the negative ion mass spectrum.

262

To explore the paint-chip in linseed oil method a little more in depth, a second
painting was provided, from which we were able to obtain our own paint chip samples.
The painting was created in the early 19008. White, green and red paint chips were
removed from the painting. Positive and negative ion LD mass spectra were obtained of
the samples. All three paint chips produced the same spectra, as can be seen in Figure 8.9
a-c, which shows the positive ion mass spectra. What is happening here? There are a
few possibilities.

The first explanation is simply the colorants do not absorb at this wavelength. A
solution to this problem, is to revert to traditional MALDI MS and add a matrix. Several
matrices were tested, including alpha hydroxyl cinnaminic acid and dihydroxy benzoic
acid, however the matrix did not improve the spectra. The second possibility may have
been that the painting was protected by a varnish layer, which was trapping the colorants
underneath. Varnish removal treatments were researched and attempted. Specifically the
Pettenkofer’s process was employed (9). The technique was rather simple. The paint
chips were exposed to ethanol vapors in an enclosed container, and then brushed with
distilled water. Spectra of the paint chips following the varnish removal process were the
same as for the pretreated samples.

The last possibility involved the vehicle. Linseed and other drying oils dry by a
crosslinking mechanism (3,10). Hence, it is quite plausible that over 100 years time (the
estimated age of the painting), the drying oil present in the painting formed a protective
polymer layer, trapping the colorants underneath. The solution to this problem was to
grind up the pigment and suspend the particles in linseed oil on a gold sample plate. In a

sense, we were creating a “fresh” paint.

263

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

39 r
a)
23
57
522
I 550
l 'A__.A- v ' ALA; - ‘194L-fi1
39
23
522
I 550
A v LA .L'J ' - ‘424A Av L A
e)
23
43
57
l 522 550
AAA ' 4 A A ' 424 L 3L
0 200 400 600

 

 

mlz
Figure 8.9: The positive ion mass spectra of an early 20‘h century painting; a) green

paint chip; b) white paint chip; c) red paint chip.

264

The solution was successful and unique positive and negative ion LD mass

spectra of the three different color paint chips were obtained. Focusing on the red paint

chip, the positive ion mass spectrum is shown in Figure 8.10. Numerous lead and lead
oxide ions were identified based on their isotopic patterns. Considering these ions are
common to many lead-containing pigments, the identity of the red colorant can not be

confirmed. However, there are two likely possibilities. The first is, the pigment is red

lead, which has the formula Pb304. The second possibility is that the peaks represent the

pigment lead white (Pb(OH)2°2PbC03). In the painting, there is clearly a layer of white

paint underneath the red layer.

 

 

 

 

 

Pb+
208
>5
H
O-
U)
I:
Q)
3: 20
G)
>’
0:
.53
g; 39
247
263
0 190

 

 

 

 

szo+
430
' 1)Red Lead: Pb3O4
2)Lead White:
Pb(OH)2-2PbCO3
Pb O H+
sz+ 447 2 2
41 Pb3,02+
465 654
Pb403+
i - - Li” . - .815
380 m/z 570 760 950

Figure 8.10: The positive ion LD mass spectrum of the red paint ground up, in
linseed oil, and analyzed on a gold sample plate.

265

Relative Intensity

 

Relative Intensity

 

0

 

79
a)
71
63
35
70
b)
s,-
64
26
79
70

 

PbClO'
259

 

 

 

 

‘_ PbCl3'
PbCl2 m
278
PbCIZO" 315
257 295
241
243 261 280 29.,
311
23’ 27 293
238 £50 A I “270 mlz 290 310 380
259
99
183
140 210 280 850
. m/z
S3
I96
Vermilion: HgS
Red Ochre: Fe203
S4'
128
254
281
1 214 277/
I .... -.. L- LA“
140 m/z 210 280 350

 

 

 

 

 

 

 

Figure 8.11: The negative ion mass spectra of the red paint ground up in linseed oil
and analyzed on a gold sample plate; a) the first spot irradiated; b) the

second spot irradiated.

266

 

The negative ion mass spectrum of the red paint chips (Figure 8.11a), also
supports the presence of a lead-containing pigment in the paint mixture. Various lead
chloride ions were identified, which have only appeared thus far in aging studies (Chapter
5). Considering that the paint mixture was an inhomogeneous mixture, different spectra
were generated as the sample was moved so that the laser could irradiate new spots.
Figure 8.11b shows another negative ion mass spectrum of the ground-up red paint chip
sample. Sulfide ions are noted in the lower mass region of the spectrum. This is
encouraging, since there are red sulfur-containing pigments, such as vermilion. In the
higher mass region, a set of peaks is noted of relatively low intensity around m/z 281,
resembling the linseed oil peaks discussed in Chapter 4. If these are linseed oil peaks, the
red pigment may actually be red ochre. Again, the identity of the pigments remain
unknown for certain, however we have shown that it is possible to obtain mass spectra

from extremely small samples.

267

References
1) httpzl/chemistry.caltech.edu/resources/massspec.html

2) Ember LR. Chemistry and art: helping to conserve the nation’s treasure. Chemical and
Engineering News 2001; July 31: 51-59. '

3) The drying of oils and oil paint. Smithsonian Center for Materials Research and
Education. http://www.si.edu/scmre/takingcare/paint_cracking.htm

4) Boon JJ. Analytical mass spectrometry: MOLART painting studies.
http://amolf.nI/publications/annual.

5) Andrasko J. HPLC analysis of ballpoint pen inks stored at different light conditions. J
Forensic Sci 2001;46:21-30.

6) Brunelle RL, Cantu AA. A critical evaluation of current ink dating techniques. J
Forensic Sci 1987;32:1522-1536.

7) McCrone WC. Authenticity study of a possible Manet painting. The Microscope
1987;35:173-195.

8) McCrone WC. Microscopial examination of art and archaeological objects.
Proceedings of the IS&T/SPIE Conference on Scientific Detection of Fakery in
Art;1998 Jan;San Jose, California.

9) Schmitt S. Effects of regeneration methods on paintlayers — with special reference to
the effects of solvent vapour and copaiba balsam.

http://www.amolf.nl/research/biomacromolecule/Pettenkoferhtm

10) Turner GPA. Introduction to paint chemistry and principles of paint technology. 3rd
ed. New York: Chapman and Hall.

268

Chapter Nine: Conclusions and Future Directions

Laser desorption mass spectrometry was evaluated as an analytical tool for the
analysis of colorants on art objects. The results of this project can be used by scientists in
the fields of art conservation and forensic science. The ability of LDMS to provide
molecular level information for both organic and inorganic components of a paint
mixture, simultaneously, makes the technique a more attractive choice over some
currently used methods. Additionally, the sensitivity of the LDMS instrument was
demonstrated.

Completing this project has provided a survey of the various forms of artwork,
which can be analyzed using LDMS. For example, the technique is well-suited for the
analysis of both watercolor and oil paints on a variety of substrates. Other pieces of
‘”,art including ancient Chinese coins and an old cross allegedly from the Crusades were
analyzed. The results of these experiments were not presented, however the experiments
were quite challenging. Standard samples in our studies included either ink-on-paper or
paint-on-paper samples. In other words, they were flat. In the cases of the Chinese coins
and the cross, the samples were three-dimensional. Using a special sample plate holder,
with the disposable sample plate removed, we were able to introduce the samples into
mass spectrometer and generate mass spectra.

Practical aspects of the LDMS experiment in regards to sample size were also
addressed. It was shown that even though increasing the size of the ion source housing to
accommodate larger, three-dimensional samples is feasible, it may not be best (safest)

solution. Currently, there are acceptable sample removal procedures in the fields of art

269

 

conservation and questioned document examination. We have shown that the sensitivity
of the LDMS experiment can accommodate these small sample sizes, making this option
more appealing.

One other solution, which was not discussed previously, is atmospheric pressure
laser desorption mass spectrometry (AP LDMS). AP MALDI and AP LD are not novel
techniques (1,2). In this technique, analyte molecules are desorbed and ionized at
atmospheric conditions outside of the mass spectrometer. A stream of N 2 (g) transports
the formed ions into the mass spectrometer for analysis. This is not an efficient method.
Consequently, AP MALDI requires a larger quantity of sample. Considering the typical
sample sizes encountered in the fields of art conservation and forensic science, there may
not be enough sample to perform the experiment. However, the design of the instrument
could be manufactured to accommodate a document or a large painting, so that the laser
could be moved to irradiate different portions of the sample directly, negating the need
for an extraction of a sample. The development of an AP LDMS method for this type of
application would, of course, fall under the category of future work.

In conclusion, laser desorption mass spectrometry has considerable potential in
the analysis of colorants used in various works of art. If funds are limited to purchase an
LD mass spectrometer, which is often the case for art museums and especially forensic
laboratories, there are numerous mass spectrometry facilities nationwide, such as the one
here at Michigan State University, to which samples could be submitted. Of course, paint
chips would not qualify as traditional samples submitted, however accommodations

could be made. The point is that our work has demonstrated the benefits of LDMS in the

270

analysis of colorants, compared with traditional instrumentation currently used. The

technique is worth developing further.

271

References

l) Laiko VV, Baldwin MA, Burlingame AL. Atmospheric pressure matrix-assisted laser
desorption/ionization mass spectrometry. Analytical Chemistry 2000;72:652—657.

2) Laiko VV, Taranenko NI, Berkout VD, Musselman BD, Doroshenko VM.

Atmospheric pressure laser desorption/ionization on porous silicon. Rapid
Communications in Mass Spectrometry 2002;16:1737-1742.

272

 

APPENDIX

273

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Ion m/z PbO PbOz PbMoO4 Pb3(SbO4)2 PbSO4 PbS PbCrO4
Pb+ 208 :8 =1: :1: a: at: =I<
PbOH+ 225 *

PbK+ 247 *

PbOK+ 263

PbCrOz+ 293

PbCr03+ 308

PbSbOz+ 361 *

Pb; 414 * *
Pb20+ 430 * =l< :1: :1: :1: at:
PbQS" 446

Pb202H+ 447 * * * *
Pb28Cl+ 483 *
szCrOf 514
szCrO4+ 530
Pb2Sb03+ 585 *

Pb302+ 654 *
Pb303H+ 671

Pb3S2’ 686

Pb3S2Cl+ 721
Pb3Cr05+ 754 *
Pb3$bO4" 807 *

Pb4O3+ 878 * *

Pb404H+ 89S * *

Pb504+ 1102 * *

Pb505H+ 1119 * *

 

 

 

 

 

 

 

 

 

 

Table 1: Summary of Positive Ions Generated by Pb-compounds in the LDMS

Experiment.

274

 

Ion m/z PbO PbOz PbMoO4 Pb3 Sb04 PbSO4 PbS
C ' 100
M004- 162 *
Cr ' l 84
Cr206° 200
PbO' 224
PbOzH' 241
Cr307' 268
PbClS’ 275
C1303- 284
M ' 288
Cr309' 300
M0207- 304
PbCl3‘ 3 13
Cf409- 352
Cr4O ’ 368
sz' 414
M ' 432
Pb202' 446
szO3H' 463
2' 480
Pb232Cl’ 5 15
' 622
Pb303' 670
04H- 687
JPb405' 910
Pb506' l 134
' 1340
Pb6O7H' 1357

   

Table 2: Summary of Negative Ions Generated by Pb—compounds in the LDMS
Experiment.

275

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Sam le 1 Sam 1e 2 Sam le 3 Sam le 4
1°“ mlz “’5 (25%pHCl) (50%pHCl) (75%pHCl) (100‘7cleCl)
Pb+ 208 =1: :1: >1: * *

PbOH+ 225 * *
PbK+ 247 * * *
PbOK+ 263 * * * *
Pb; 414 *
Pb20+ 430 * * *
PbS+ 446 * * *
szOzH+ 447 * *
PngCl+ 483 *
Pb282C1+ 515
szCrO4+ 530 * *
Pb382+ 686
Pb3S2Cl+ 721 * * *
CrOg' 100
Cl‘zOs- 184
Cl'206- 200
Cr307’ 268
PbClS' 275 * *
CrgOg' 284 * a:
CF309- 300 * *
PbCl3' 313 *
? 336
CF409- 352
CI'4010- 368
szsz' 480
szszcr 515

 

 

 

 

 

 

 

Table 3: Summary of Positive and Negative Ions Generated in the Uncontrolled H28 (g)
Exposure Experiments.

276

 

 

 

Sample 4

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Ion m/z PbS/11202 (exposed 15minutes)
my 208 * *
PbOH+ 225 * *
pb02H+ 241 *
PbCl" 243 * Eh
PbK+ 247 *
PbOK‘ 263 *
PbCle" 279 *
pb; 414 *
szO+ 430 *
szOzH+ 447 *
szocr 465 *
szscr 483 *
szCI‘O4+ 530
pb3()2+ 654 *
Pb380+ 670 *
131335; 686 * *
Pb3SCl+ 689 *
Pb3SClO+ 705 *
*

 

Pb382Cl+ 721
Pb3Cr05+ 754 *
Pb48ClO+ 913
Pb4SC102+ 929
Pb4SC 103+ 961

 

 

**

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

CrO3' 100 *
CI'205- 184 *
0205 200 *
CI‘3O7- 268 *
PbSCl’ 275 *
CI‘303- 284 *
PbCl3' 313

 

 

Table 4: Summary of Positive and Negative Ions Generated in the Uncontrolled H25 (g)
Exposure Experiments Followed by Treatment with H202.

277

IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII

IIIIIIIIIIJIIIIIIIIIIIISIIBII

3 024