Historical linseed oil/colophony varnishes formulations: Study of their
molecular composition with micro-chemical chromatographic techniques

Sophie Tirat, Ilaria Degano, Jean-Philippe Echard, Agnès Lattuati-Derieux,

Anna Lluveras-Tenorio, Arul Marie, Stéphane Serfaty, Jean-Yves Le Huerou

PII: S0026-265X(15)00325-2
DOI: doi: 10.1016/j.microc.2015.11.045
Reference: MICROC 2344

To appear in: Microchemical Journal

Received date: 15 August 2015

Revised date: 24 November 2015
Accepted date: 25 November 2015

Please cite this article as: Sophie Tirat, Ilaria Degano, Jean-Philippe Echard, Agnès
Lattuati-Derieux, Anna Lluveras-Tenorio, Arul Marie, Stéphane Serfaty, Jean-Yves Le
Huerou, Historical linseed oil/colophony varnishes formulations: Study of their molecu-
lar composition with micro-chemical chromatographic techniques, Microchemical Journal
(2015), doi: 10.1016/j.microc.2015.11.045

This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.
 ACCEPTED MANUSCRIPT

Historical linseed oil/colophony varnishes formulations: study of their

molecular composition with micro-chemical chromatographic techniques
Tirat Sophie*a,b,c, Degano Ilariad, Echard Jean-Philippec, Lattuati-Derieux Agnèsa, Lluveras-Tenorio Annad, Marie
Arule, Serfaty Stéphaneb, Le Huerou Jean-Yvesb

T
* corresponding author. E-mail address: stirat@cite-musique.fr

IP
a Centre de Recherche sur la Conservation des Collections, CNRS/MNHN/MCC USR 3224, Muséum national

d’histoire naturelle, Sorbonne Universités, 36 rue Geoffroy Saint-Hilaire 75005 Paris, France

R
b Laboratoire Systèmes et Applications des Technologies de l’Information et de l’Energie (SATIE), CNRS UMR

8029, Université de Cergy-Pontoise, 5 mail Gay Lussac, 95031 Neuville sur Oise, France
c Equipe Conservation Recherche du Musée de la musique, CNRS/MNHN/MCC USR 3224, Cité de la musique-

SC
Philharmonie de Paris, 221 avenue Jean Jaurès, F-75019 Paris, France.
d University of Pisa, Department of Chemistry and Industrial Chemistry, Via Moruzzi, 13, I-56124 Pisa, Italy
e Unité Molécules de communications et adaptations des micro-organismes, CNRS/MNHN UMR 7245, Muséum

National Histoire Naturelle, Sorbonne Universités, 57 rue Cuvier, 75005 Paris, France

NU
MA
D
P TE
CE
AC
 ACCEPTED MANUSCRIPT

Abstract
Mixtures of siccative oil and Pinaceae resin were among the most widespread formulations used for
varnishing in the 16th-18th centuries. The aim of this research is to study the molecular composition of
different formulations of varnishes based on linseed oil and colophony. The mixtures were prepared

T
by cooking both raw materials in different proportions, following the outlines of historical recipes

IP
gathered in treatises and manuscripts from that period. Size exclusion chromatography (SEC)
evidenced the presence of numerous oligomeric compounds from linseed oil and colophony. Analysis

R
using Electrospray-Quadrupole-Time of Flight mass spectrometer (ESI-Q-ToF) allowed the

SC
characterization of oligomeric compounds from colophony. In particular, structures of dimeric
diterpenic acids units attached by ester, ether, anhydride or keto linkage are proposed. Concerning

NU
linseed oil, oxidized and chain-shortened triglycerides were characterized with High Performance
Liquid Chromatography coupled with Electrospray-Quadrupole-Time of Flight mass spectrometer
(HPLC-ESI-Q-ToF). SEC quantitation of compounds allowed us to assess the influence of the oil to
MA
colophony ratio on the molecular composition of the varnishes: oil polymerization appears to be
minimized in favor of its hydrolysis in presence of colophony, whereas linseed oil seems to inhibit the
formation of diterpenic dimers and trimers. In addition, analyses evidence the reaction of diterpenic
D

acids with linseed oil compounds. In particular, we identified hybrid molecules formed from a
TE

diglyceride and a diterpenic acid. Adducts of a triglyceride and a diterpenic acid may also be present in
the mixtures. These results bring new information about the molecular composition of varnishes
P

made from linseed oil and colophony, and will improve the knowledge of historical varnishes. To this
CE

purpose, all analytical protocols were implemented on micro-samples so that they could be applied
later on ancient artifacts.
AC

1. Introduction

Coatings formulated with linseed oil and colophony were commonly used for varnishing before the
introduction of synthetic varnishes in the 1900s [1]. In particular, they were used to protect and
ornament musical instruments in the 16th-18th C. [2–4]. Such coatings were also frequently used for
furniture, easel paintings and in decorative arts [5,6]. Though reactivity and ageing mechanisms of
linseed oil and colophony are the topics of numerous studies [7–10], the chemical behavior of such
mixtures is still poorly known. Moreover, very few information is available on the molecular
composition of the “fresh” mixture, before any drying occurs.

Heat treatment is necessary to mix colophony with linseed oil. The effect of temperature on the
diterpenic acids composing colophony has been previously studied [11–13]. Diterpenic acids undergo
isomerization, dehydration, oxidation, polymerization and decarboxylation reactions under heat. It
has been shown that after few hours of heating, isomerization of abietane compounds leads to the
 ACCEPTED MANUSCRIPT

vanishing of laevopimaric acid, and the predominance of abietic acid in resin [14,15]. In addition,
abietane-type acids dehydrate into dehydroabietic acid [16,17]. Thermo-oxidation reactions of
abietane-type acids lead to various oxidation products, mainly alcohols and ketones. Compounds such
as 7-oxo-dehydroabietic acid, 7-oxo-abietic acid, 15-hydroxy-dehydroabietic acid and 15-hydroxy-7-
oxo-dehydroabietic acid were evidenced in pitch-an organic substance obtained by combustion of

T
resin exudates or resinous wood [18,19]. These oxidation products are also formed upon natural

IP
ageing of diterpenic acid on a longer time-scale [20,21]. When submitted to temperatures above 210

R
°C, diterpenic acids can undergo decarboxylation reactions [22]. Decarboxylated and demethylated
compounds were evidenced in pitch [18,19,23]. It has been suggested that polymerization of

SC
diterpenic acids submitted to heat treatment in inert atmosphere mainly leads to the formation of
ester linkages [11]. Under air, oxidative polymerization could also lead to ether or anhydride linkages

NU
[13]. Formation of oligomers with molecular weight (MW) up to 2200 Da (approximately heptamers)
was observed in artificially aged colophony [8]. Dimers of colophony with C-C linkages were
MA
synthetized in acidic conditions and under high temperatures [24,25].

Linseed oil is mainly composed of unsaturated triglycerides (TAGs), the most abundants being
formed with linolenyl, linoleyl and oleyl substituants. The effect of heat treatment on linseed oil has
D

been previously studied [7,10]. We can also gather information from numerous studies concerning
TE

other vegetable oils [26–31]. Oxidation mechanisms occurring in vegetable oils submitted to
temperatures from 150 to 300 °C initiate with the formation of radicals on (poly)unsaturated acyl
P

chains, followed by cis-trans isomerization of the double bonds. Reaction of oxygen with the radicals
forms hydroperoxides. These primary oxidation products head to stable oxidized TAGs as well as to
CE

short-chain glycerol-bound oxidation products. Analyses of heated triolein and trilinolein showed that
ketones, alcohols and epoxides are the major groups found in oxidized monomeric TAGs [26,32,33].
AC

These compounds may be formed either by hydroperoxides cleavage or by addition of external

hydroperoxides on a double bond [34]. β-scission of hydroperoxides leads to low MW volatile
compounds and to shortened TAGs. The shortened branch can consist of an aldehyde, a carboxylic
acid or an alkane moiety [34,35]. Radical polymerization of TAGs forms mainly carbon-carbon and
ether bonds between fatty acid chains [30,36,37]. Peroxy links are favored in low temperature
processes but readily decompose at high temperatures of heating [38]. Diels-Alder mechanisms,
occurring in stand oils, have also been postulated for linseed oil heated under air [7]. Oligomers up to
tetramers [30,38,39] and hexamers [7] have been evidenced in heated oils or pure standards of
triglycerides.

We have recently shown that the amount of ageing markers in oil/colophony varnish films, after a
few months drying, strongly depends on the oil/resin proportion, which indicates competition
between different mechanisms and/or different drying kinetics [40]. In particular, results highlighted
 ACCEPTED MANUSCRIPT

the higher state of oxidation of colophony when mixed with high amounts of linseed oil, as
postulated earlier [41]. Rivenc and co-workers [42] referenced that the presence of colophony may
promote oil hydrolysis and suggested the reaction of molecules from linseed oil and colophony with
each other. Such works underlined possible interactions between oil and colophony compounds, and
highlighted that the conservation state of varnishes seems to be linked to the proportion of the

T
mixture. Having the knowledge to what extent proportions affect the mixture’s reactivity would be

IP
essential for interpreting analytical data on historical varnishes’ composition as well as for

R
conservation strategies, in particular, to adjust the solvents used for cleaning/removing the varnish to
its state of ageing. In this context, our approach includes the reconstruction of ancient varnishes

SC
recipes[43], the study of their molecular compositions and the monitoring of their curing and ageing
under natural conditions. Conventional techniques such as infrared spectroscopy and Pyrolysis-

NU
GCMS have shown to be of interest for this topic [8,9,40–42]. However, characterization of
triglycerides and high molecular weight oligomeric compounds requires more powerful analytical
MA
techniques, such as liquid chromatography and tandem mass spectrometry, as shown in [44].

In this work, we present the analyses of linseed oil and colophony varnishes prepared by heat
treatment in different proportions. The characterisation of colophony and linseed oil derivatives was
D

carried out using size exclusion chromatography with a refractive index detector (SEC-RI) and two
TE

mass spectrometric techniques: HPLC-ESI-Q-ToF and ESI-Q-ToF. In addition, quantitation of the

major types of compounds observed with SEC-RI analyses was performed. The reactivity of linseed
P

oil/colophony mixture depending on the initial proportions is discussed, and in particular the
identification of a new “hybrid” compound is presented.
CE

2. Material and Methods

AC

2.1 Varnishes preparation

Varnishes preparation protocol was elaborated based on ancient treatises, and with the supervision of
a contemporary luthier. Numerous varnishes recipes from the 15th-18th C. mention the use of linseed
oil and colophony, often along with other resinous or gum material [1]. We studied in particular those
in which the only ingredients mentioned were Pinaceae resin and linseed oil. In these recipes, oil and
colophony are heated together, until the desired homogeneity/viscosity of the mixture is obtained.
Different orders in the addition of the components as well as different oil/resin ratio (from 2:1 to
1:3) are indicated. Specifications regarding time and temperature of heating are most of the time
qualitative (“boil until the mixture thrown in the fire will burn without crackle”[45] or “boil the
mixture softly until it is viscous”[46]).
 ACCEPTED MANUSCRIPT

Based on this short review, we elaborated the following protocol, in which the same amount of
linseed oil is pre-heated for each preparation, before adding different amounts of colophony. This
procedure ensures that linseed oil always undergoes the same heat treatment. In a previous study,
time and temperature of heating were selected in order to give mixtures with suitable application
properties, usable for varnishing [43].

T
IP
To prepare linseed oil/colophony mixtures, 5 g of cold pressed linseed oil (Huilerie de l’Orme Creux,
France), maintained under constant stirring, was heated in a glass beaker for 30 min, until it reached

R
230 °C, and kept for 30 min at this temperature. A weighed amount of grounded colophony

SC
(Domaines & Patrimoine, France) was then added slowly to the hot oil. During this mixing step,
heating was constantly adjusted so that the temperature remained at 230 +/- 5 °C. After 100 minutes,
the heating was stopped and the mixture was allowed to cool down until it reached room

NU
temperature. The mixtures were transferred to closed 10 ml vials, flushed with nitrogen and kept in
the dark at room temperature (20-23 °C).
MA
To prepare the 100 wt% oil varnish, 5 g of cold pressed linseed oil, maintained under constant
stirring, was heated for 30 min, until it reached 230 °C, and kept for 130 min at this temperature,
totaling 160 min of heat treatment. Heating was then stopped and the oil was allowed to cool down
D

until it reached room temperature. Cooked oil was transferred to a closed 10 ml vial, flushed with
TE

nitrogen, and kept in the dark at room temperature.

100 wt% colophony varnish was prepared by directly pouring 7 g of grounded colophony, into a glass
P

beaker heated at 230 °C. Heating was maintained for 100 min, then, colophony was allowed to cool
CE

down for about 10 minutes, before it was transferred in a closed 10 ml vial, flushed with nitrogen and
kept in the dark at room temperature.
AC

These procedures ensure that linseed oil and colophony were submitted to equivalent heat
treatments, regardless of whether they were cooked alone or in the mixtures.

SEC-RI analyses
For SEC experiments, a weighed amount of sample (5 to 10 mg) was dissolved in 5 ml tetrahydrofuran
(VWR, U.S.A.), to obtain an approximate concentration of 1 mg/ml. Solutions were placed in a
Branson 2510 ultrasonic bath (Branson Ultrasonics, U.S.A.) at ambient temperature for 10 min, then
filtered on a 0.45 µm PTFE filter (VWR, U.S.A.) before 50 µg of the solution was injected in the
chromatographic system, consisting of a Waters 2695 separation module (Waters, U.S.A.) with an
external column heater. Separation was achieved on three Agilent PLgel columns connected in series
(7.5 mm × 300 mm, 5 µm - pores size 500 , 500 and 100 .), with an Agilent PLgel guard column
(7.5 mm × 50 mm, 5 μm). Detection was performed with a Waters 2414 Refractive Index Detector
 ACCEPTED MANUSCRIPT

(Waters, U.S.A.). The system was operated at a temperature of 40 °C (detector and column furnaces)
with a flow rate of 1 ml/min of tetrahydrofuran. Calibration was performed using seven polystyrene
(PS) standards (Agilent Technologies, U.S.A.) with molecular weights (MWs) ranging from 370 to
6940 Da.

T
The relative error percentage between the apparent MW of diterpenic standards (that is to say the

IP
MW assigned by PS calibration), and their real MW, is in the acceptable range (7% for abietic acid
stands and 1% for 7-oxo-dehydroabietic acid standard). Indeed, diterpenes and PS are both polycyclic

R
molecules with a small hydrodynamic volume. However, PS calibration cannot be used to retrieve

SC
MWs of glyceridic compounds, without considerable error in mass assignment [28,47]. Four lipid
standards ranging from 282 to 885 Da (purified triolein, dioleoylglycerol, mono-oleoylglycerol, and
oleic acid, Sigma Chemical Company, U.S.A.) were used to construct a second calibration curve,

NU
which is distinctly separate, but parallel to the PS one. Therefore, apparent MWs assigned using the
PS calibration will be proportional to those assigned using the lipid calibration. Since the PS curve
MA
was established with seven standards covering a wider range of MWs, the random error in mass
assignment is lower. Therefore, the PS calibration was used to assign apparent MWs for glyceridic
compounds, knowing that the proportionality of MWs is conserved.
D

The detection limit was calculated from the calibration curves of triolein and dehydroabietic acid to
TE

0.062 mg/ml (62 ppm) and 0.050 mg/ml (50 ppm) respectively. Quantitation was performed by
dividing the peak area by the mass of the introduced sample. All compounds eluting over 27.3
P

minutes were integrated together.

CE

Data acquisition and integration of the chromatograms were performed with Empower 2.2 software.

This protocol was successfully applied to varnish micro-samples by dissolving 0.2 to 0.5 mg of sample
AC

into 400 L tetrahydrofuran.

2.2 ESI-Q-ToF mass spectrometric analyses

Two analytical procedures and two different instruments were used for ESI-Q-ToF mass
spectrometric analyses. Protocol 1 (HPLC-ESI-Q-ToF and flow injection analyses) was performed to
analyze the extracted lipidic fraction of the varnishes. Protocol 2 (ESI-Q-ToF analyses) was
implemented to analyze colophony compounds in the 100 wt% colophony varnish.
Protocol 1:
A weighted amount of sample (1 to 3 mg) was diluted in 300 μL chloroform/hexane (3/2, vol/vol) ,
then submitted to extraction assisted by a microwave Ethos One (Milestone, U.S.A.) at 80 °C for 60
min (power 200 W), under inert N2 atmosphere. Additional 300 μL of chloroform/hexane (3/2,
vol/vol) were added to dilute the extracted fraction, before sampling 200 μL of it. This extract was
 ACCEPTED MANUSCRIPT

dried under a N2 stream, diluted with the methanol and iso-propanol (90/10, vol/vol) elution mixture
to 1 ml and filtered on a 0.45 μm PTFE filter (Grace Davison Discovery Sciences, U.S.A.). Then, 10
μL of the mixture was diluted in 500 μL methanol and iso-propanol (90/10, vol/vol) just before
injection.

T
Analyses were carried out using a 1200 Infinity HPLC (Agilent Technologies, USA) coupled to a Jet

IP
Stream ESI interface with a Quadrupole-Time of Flight tandem mass spectrometer 6530 Infinity Q-
ToF (Agilent Technologies, U.S.A.). Chromatographic separation was achieved with an Agilent

R
Poroshell 120 EC-C18 column (3.0 mm × 50 mm, 2.7 µm) with a Zorbax eclipse plus C-18 guard

SC
column (4.6 mm × 12.5 mm, 5 μm). The injection volume was 1 to 3 µL, depending on the extract’s
concentration. Elution conditions, ESI operating conditions, mass axis calibration method and
MS/MS acquisition parameters have previously been described by some of the authors [44], though
two precursors were acquired per cycle.
NU
Flow injection analyses were carried out with the same HPLC and mass spectrometer devices.
MA
Eluents were methanol and iso-propanol (90/10, vol/vol); flow rate was set to 0.2 ml/min and
injection volume was 1 µL. The ESI and MS operating conditions were the same as for the HPLC
analyses.
D

Identification of triglycerides was achieved by comparing their retention time and their tandem mass
TE

spectra with published data acquired using the same method [44,48]. Interpretation of the tandem
mass spectra of glyceridic compounds, in which the fragments derive from the elimination of a fatty
P

acid branch, is based on the assumption that the loss of the acyl chain in sn-1 and sn-3 position is
CE

energetically more favored than the loss in sn-2 position and leads to the most abundant ions [49].
The following abbreviations are used to indicate acyl substituents in glyceridic compounds: Ln:
AC

linolenyl (C18:3); L: linoleyl (C18:2); O: oleyl (C18:1); S: stearyl (C18:0); P: palmityl (C16:0).

Mass spectrometer control, data acquisition, and data analysis were performed with MassHunter
Workstation Software (B.04.00).

Protocol 2:

A weighed amount (3.5 mg) of the 100 wt% colophony varnish sample was dissolved in 1 ml
dichloromethane and placed in Branson 2510 ultrasonic bath (Branson Ultrasonics, U.S.A.) at
ambient temperature for 15 minutes to ensure dissolution. Before injection, 4 µL of the sample was
diluted 1 to 5 in 20 µL of the acetonitrile/water (70/30, vol/vol) elution mixture, before adding 4µL
of a 10 mM ammonium acetate solution. Samples were analyzed by direct injection in an ESI-Q-ToF
mass spectrometer (Qstar, pulsar I, PE-Sciex, Canada) using ACN and 0.1% formic acid in water
(70:30, vol/vol) at a flow rate of 40 µL/min. Injection volume varied between 1 to 3 µL, depending
on the concentration of the sample. The spectra were acquired in positive ion mode using a potential
 ACCEPTED MANUSCRIPT

of 5.2 kV, between m/z 230 and m/z 1200. MS/MS spectra were acquired by varying the collision
energy between 35 V to 50V depending on the selected precursor ion and the spectra were recorded
in the range m/z 30-1200. The spectra were treated using Analyst software (version 1.1).

To calculate the most probable formula for fragments observed on tandem mass spectra, the online

T
calculator ChemCalc [50] was used.

IP
3. Results

R
3.1 SEC analyses of raw materials and varnishes

SC
Five varnishes with different oil/colophony proportions (100/0, 80/20, 66.5/33.5, 50/5, 42/58,
33.5/66.5 and 0/100, wt%/wt%) were submitted to SEC analyses, along with crude linseed oil and

NU
colophony (Figure 1). To allow a proper comparison, intensity of the chromatograms is corrected by
the sample weight.
MA
Chromatograms of the 100 wt% oil and 100 wt% colophony varnishes, show that glycerides and
related species elute in the range 15-22 min, whereas diterpenic acids and their derivatives elute
between 20 and 25 minutes. The chromatogram of crude oil displays a unique peak at 20.6 min,
D

attributed to triglycerides. Free fatty acids are absent or below the detection limit. Instead, the
TE

chromatograms corresponding to the oil/colophony mixtures show numerous glyceridic compounds

in the range 15-22 min.
P

The apparent MW attributed to TAGs with the PS calibration curve is 1468 Da. Since apparent MWs
CE

are proportional to real MWs, an apparent MW of (2 × 1468) Da is attributed to a compound having

twice the MW of a TAG. Calculated apparent MWs of 2, 3, 4, 6 and 8 TAGs are placed on the PS
AC

scale (Figure 1).

The peak at 19.2 min in the varnishes’ chromatograms matches with the apparent MW of two TAG
units, and is therefore ascribed to dimers of triglycerides. Trimers are also present, although, based on
MW scale, the compounds represented by the two peaks maxima at 18.3 and 17.9 min are not
constituted of three intact TAG units. Compounds eluting at 18.3 min could be shortened trimers in
which one of the fatty acid branch has undergone chain scission/hydrolysis reaction. Compounds
eluting at 17.9 min could be trimers attached to an extra fatty acid chain and/or shortened tetramers.
These two types of triglycerides trimers will be referred as “low MW trimers” and “high MW
trimers”. Oligomers up to heptamers are evidenced. According to the pure glyceride standards, the
shoulder at 21.4 min on the varnishes’ chromatograms can be ascribed to diglycerides. In the 100
wt% oil varnish sample, no glyceridic compound with a lower MW than diglycerides is detected,
probably because low MW compounds formed under heat (short-chain alkanes and aldehydes, free
fatty acids, monoglycerides) are volatile at 230 °C.
 ACCEPTED MANUSCRIPT

The peak at 24.3 minutes can be ascribed to non-oxidized and oxidized diterpenic acids containing up
to three extra oxygen atoms. The width of this peak is slightly larger for the varnishes than for crude
colophony, which may be attributed to the presence, in the varnishes, of oxidized compounds with
higher MW and decarboxylated or demethylated compounds with lower MW. According to the PS
calibration, an approximate molecular weight of 420 Da is ascribed to compounds eluting at 23.5

T
minutes. Similar compounds, having intermediate MW between diterpenic acids monomers and

IP
dimers, were evidenced by other authors [51]. The peak at 22.8 minutes, particularly intense in the

R
chromatogram of the 100 wt% colophony varnish, corresponds to a molecular weight of 590 Da,
which matches the MW of a dimeric compound formed from two diterpenic acids units. The

SC
shoulder at 21.8 minutes, observed only in the 100 wt% colophony varnish’s chromatogram can be
tentatively attributed to trimers of diterpenic acids, with an approximate molecular weight of 915 Da.

NU
In addition to the compounds previously identified, two unidentified peaks at 20.9 and 20.0 min are
noticeable on the chromatograms of the mixtures.
MA
3.2 MS and MS/MS characterization of varnishes’ compounds
The oil/colophony mixtures were submitted to ESI-Q-ToF analyses. We focused on the
D

characterization of diterpenes derivatives eluting in the range 21-23.5 min and glyceridic compounds
TE

eluting in the range 19.5-20.5 min. We also attempted to characterize the compounds corresponding
to the two unidentified peaks on the chromatograms.
P

3.1.i Compounds from colophony

CE

ESI-Q-ToF analyses were performed on the 100 wt% colophony varnish in order to confirm the
AC

chromatographic analyses’ attributions and propose structures for the evidenced compounds. The
fingerprint mass spectrum is reported in Figure 2. Three clusters corresponding to monomers, dimers
and trimers of diterpenic acids are visible, thus confirming the attribution of the chromatograms’
peaks. Theoretically, it cannot be excluded that the polymeric material observed with ESI-q-ToF
consists of non-covalently bound complexes that are formed during the electrospray ionization
process [7]. Therefore, it is relevant to perform both ESI-q-ToF and SEC analyses, the last one in
order to ascertain the presence of covalently bound oligomers.

The most intense cluster visible in Figure 2 in the range m/z 300-400 corresponds to typical
diterpenic acids, dehydrated diterpenic acids and oxidized diterpenic acids: dehydro-dehydroabietic
acid (m/z 299.201, [M+H]+), dehydroabietic acid (m/z 301.218, [M+H]+), abietic acid (m/z 303.299,
[M+H]+, 7-oxo-dehydroabietic acid (m/z 315.200, [M+H]+ and m/z 332.226 [M+NH4]+), hydroxy-
dehydroabietic acid (m/z 317.216, [M+Na]+) and hydroxy-abietic acid (m/z 319.233, [M+H]+).
 ACCEPTED MANUSCRIPT

Compounds responsible for the second cluster in the range m/z 550-700 may be dimer formed upon
two diterpenic acids. For example, adducts at m/z 616.429 and m/z 632.438 correspond to molecules
having 40 carbon atoms, thus formed from two 20-carbon atoms diterpenic acids. In both cases, the
measured masses agree well with the theoretical values (relative error is 11 ppm). As for the adduct
m/z 588.400, the only corresponding formula with less than 15 ppm error in mass assignment, is

T
[C38H50O4+NH4]+. Such formula implies the loss of two carbon atoms, most probably by

IP
decarboxylation or demethylation. Decarboxylation can also lead to the formation of ketone-type

R
dimers [52], such as the one represented in Figure 4c. The cluster in the ranges m/z 850-1000 may
correspond to trimers of diterpenic acids.

SC
Tandem mass spectra of two dimers adducts belonging to the two most intense clusters, centered
around m/z 615 and m/z 630, were collected (Figure 3,a,b). In addition, the tandem mass spectrum

NU
of a dimer having a lower mass was acquired in order to have a better understanding of this type of
structure (Figure 3c). The different series of fragments observed in each spectrum suggest that
MA
parallel and consecutive fragmentation pathways occur in the gas phase, which might be due to the
presence of isobaric compounds. In Figure 4, we consequently depict several structures for each
possible precursor ion (without pretending to be exhaustive). Since it has been generally admitted that
D

abietane are more prone to oxidative polymerization than pimarane, we decided to privilege
TE

oligomers formed upon abietane units, but polymerization of pimarane was also postulated [8].
According to the literature, ester, anhydride and ether linkage can be formed between two diterpenic
P

acids. It is generally considered that the carbon in position 7 is highly reactive and can undergo
addition reactions on the double bond leading to ether or ester linkages [11,22,25]. Dimers connected
CE

by two C-C bonds forming a 6-membered ring with no extra oxygen atoms, as postulated by some
authors [24], do not match with the fragments observed. Though C-C linked dimers cannot be
AC

excluded, given the strong thermo-oxidative conditions used, we believe that it is more probable that
incorporation of oxygen would take part in the dimer formation.

3.1.ii Glyceridic compounds

HPLC-ESI-Q-ToF and flow injection analyses were performed on the extracted non-polymerized
fraction of varnishes samples. Identification of DAGs and TAGs present in the chromatograms was
carried out by comparing both retention time and m/z of the [M+Na]+ adduct with previously
reported data acquired using the same protocol [48]. Other compounds were identified through the
interpretation of their tandem mass spectra. The HPLC chromatogram and the corresponding
fingerprint mass spectrum of the 100 wt% oil varnish are presented in Figure 5a and Figure 5b.

The first portion of the HPLC chromatograms is characterized by the presence of diglycerides,
oxidized TAGs and chain-shortened TAGs. DAGs with two C18-substituents elute in this order:
 ACCEPTED MANUSCRIPT

LnLn (m/z 635.4666, [M+Na]+), LnL (m/z 637.4858, [M+Na]+), LL/LnO (m/z 639.4982,
[M+Na]+), LO/LnS (m/z 641.5095, [M+Na]+) and OO/LS (m/z 643.5309, [M+Na]+). DAGs
having a palmityl substituent, such as LnP (m/z 613.4820, [M+Na]+), LP(m/z 615.4945, [M+Na]+)
and OP(m/z 617.5066, [M+Na]+), are present in lower amounts. TAGs oxidation products elute in
the range 5-8 min. Shortened TAGs are present in the range 3-8 min, overlapping DAG and TAGs

T
oxidation products.

IP
The second portion of the chromatogram is characterized by the presence of TAGs: LnLnLn (m/z

R
895.6743, [M+Na]+), LnLnL (m/z 897.6973, [M+Na]+), LnLnP (873.6906, [M+Na]+),

SC
LnLL/LnLnO (m/z 899.7098, [M+Na]+), LnLP (875.7108, [M+Na]+), LnLO/ LLL (m/z 901.7228,
[M+Na]+), LnLnS (m/z 901.7228, [M+Na]+), LnOP/LLP (m/z 877.7266, [M+Na]+), LnOO/LLO
(m/z 903.7433, [M+Na]+), LOP (m/z 879.7449, [M+Na]+), LOO (m/z 905.7567, [M+Na]+),

NU
LnOS/LLS (m/z 905.7556, [M+Na]+), OOP (m/z 881.7534, [M+Na]+), OOO (m/z 907.7730,
[M+Na]+), OOS (m/z 909.7875, [M+Na]+) elute in this order. As expected in linseed oil, linolenyl,
MA
linoleyl and oleyl are the main acyl substituents. Since heat treatment can lead to isomerisation of double
bonds (positional and configurational), the abbreviations “Ln”, “L” and “O” refer here generally to C18:3,
i.e. linolenic acid or an isomer, and likewise for linoleic and oleic acids.
D

Tandem mass spectra of some of the characterized oxidized TAG are reported in Figure 6a. Their
TE

identification relies in particular on the fragments at m/z 319.223, which correspond to a sodium
adduct of C18 moiety containing one extra oxygen atom and two insaturations. According to the
P

other publications, the main oxidation products formed upon heating of pure standards of
CE

triglycerides are ketonic, hydroxyl and epoxide compounds. ESI-fragmentation scheme of epoxides
usually shows adduct ions corresponding to the loss of H2O by dehydration of the epoxidized DAG
fragments, as well as fragment ions in the range m/z 400-500 resulting from scission reactions of the
AC

epoxidized DAG fragments [53,54]. Such fragments ions are not observed here, suggesting the
characterized compounds are not epoxides.

Tandem mass spectra of some of the characterized shortened TAGs are presented in Figure 6b.
According to previously published works regarding heated triolein and trilinolein standards [26,32,34],
scission of fatty acid branches leads to shortened TAGs. The shortened branch can consist of
unmodified hydrocarbon chain as well as unconjugated and conjugated aldehydes. It is then very
likely that the oxidized fragments at m/z 195.0980, m/z 207.1036, m/z 221.1112 and m/z 247.1315
correspond to unconjugated and conjugated aldehydes moieties.

Diglycerides (m/z 635 to 645), shortened TAGs (m/z 750 to 850), TAGs (m/z 873 to 909) and
oxidized TAGs (m/z 915 to 970), are present in the fingerprint mass spectra of the 100 wt% oil
(Figure 5b). The two TAGs clusters, in the ranges m/z 873-881 and m/z 895-907 respectively,
correspond to triglycerides with 55 carbon atoms (containing one palmityl substituent) and
 ACCEPTED MANUSCRIPT

triglycerides with 57 carbon atoms. Flow injection analyses point out the wide range of oxidized
compounds present in the samples, evidencing up to four clusters corresponding to TAGs, in which
one to four extra oxygen atoms have been incorporated. Compounds with more than one extra
oxygen atom do not elute in the conditions used for the HPLC experiments, and thus, no MS/MS
data was acquired. However, the measured masses of [M+Na]+ adducts agree very well with the

T
theoretical mass of triglycerides bearing extra oxygen atoms (see Table 1).

IP
Table 1 summarizes the oxidized TAGs and shortened TAGs identified with mass spectrometric

R
experiments. The position of the oxidized/shortened branch relatively to non-oxidized acyl chains is

SC
undetermined. The measured masses agree well with the theoretical values (maximum of relative error
in mass assignment is 9 ppm).

Abbreviation Molecular
formula
Monoisotopic
mass
NUTheoretical
mass
Experimental
mass
Fragment ions
MA
[M+Na]+ [M+Na]+
LnLnC8:0 C47H78O6 738.5799 761.5696 761.5657 617.4374, 483.3461

LnOC9:1(O) C48H82O7 770.6061 793.5958 793.5948 599.5109, 515.3799,

D

511.3236, 493.3898,
195.0980
TE

LnLnC10:2(O) C49H78O7 778.5748 801.5645 801.5621 617.4453, 523.3319,

207.0980
P

LnOC10:2(O) C49H82O7 782.6061 805.5959 805.5920 621.5015, 599.5085,

527.3661, 523.3475,
CE

207.1036
LnLnC11:1(O) C50H82O7 794.6061 817.5958 817.5934 617.4475, 539.3623,
223.1284
AC

LnOC11:2(O) C50H84O7 796.6217 819.6115 819.6139 621.4763, 541.3751,

537.3569, 221.1112
LnLnC13:3(O) C52H82O7 818.6061 841.5959 841.5981 617.4462, 563.3670,
247.1315
LnLnC18:3(O) C57H92O7 888.6843 911.6741 911.6679 633.4421, 617.4477;
317.2083
LnLnC18:2(O) C57H94O7 890.7000 913.6898 913.6818 635.4645; 617.4481;
595.4722; 319.2232
LnLC18:2(O) C57H96O7 892.7156 915.7054 915.7003 637.4742; 635.4599;
619.4694; 597.4774;
319.2236
LnOC18:2(O) C57H98O7 894.7313 917.7211 917.7211 639.4925; 635.4617;
621.4776; 319.225
TAG+2O C57H96O8 908.7106 931.7003 931.6963 /
TAG+2O C57H100O8 912.7419 935.7316 935.7243 /
TAG+3O C57H100O9 924.7055 947.6952 947.6985 /
 ACCEPTED MANUSCRIPT

TAG+4O C57H100O9 940.7004 963.6902 963.6972 /

Table 1: Compounds characterized with tandem mass spectra experiments. Acyl substituent abbreviations: Ln,
linolenyl (C18:3); L, linoleyl (C18:2); O, oleyl (C18:1). Subscript «O» refers to the presence of an oxygen atom on the acyl
substituent (as a hydroxy, ketone or epoxy function). For Cx:y, x represents the number of carbon units and y the total
number of double bonds, which might be both C=O or C=C.

Oxidized and chain shortened-TAGs were evidenced by many authors in oxidized trilinolenin,

T
trilinolein and triolein standards and in edible oils (see the review of [55] and [35]). However, few

IP
studies regarding cooked linseed oil exist, and to our knowledge tandem mass spectra characterization

R
of chain shortened-TAGs in linseed oil was never performed. Concerning oxidized TAGs, our study
complements the one of [44], by characterizing additional oxidized products.

SC
3.1.iii Newly identified hybrid compounds

NU
The fingerprint mass spectrum of the 50/50 (wt%/wt%) oil/colophony varnish, which is
representative of the oil/colophony mixtures in various proportions, is reported in Figure 7. As for
MA
the 100 wt% oil varnish (Figure 5b), TAGs (m/z 873 to 909) shortened TAGs (m/z 750 to 850), and
DAGs (m/z 635-645) are present. However, relative intensity of shortened TAGs, respectively to
TAGs, is much lower. Moreover, adducts in the range m/z 910-1050 have a different profile, the
D

cluster centered at m/z 935.58 being the most intense. Finally, a series of compounds in the range
TE

m/z 1150-1300 are observed. These compounds were not evidenced in the 100 wt% oil varnish.

A closer look at the cluster centered at m/z 935.58 highlights that the measured masses (931.6609,
P

933.6676, 935.6757, 937.6883) are not equivalent to those observed in the 100 wt% oil varnish. The
CE

corresponding compounds elute in the HPLC chromatograms between 7 and 10 min (data not
shown), confirming that they are different from TAGs with two extra oxygen atoms (which did not
elute with the chromatographic conditions used). Tandem mass spectra of two of these adducts are
AC

presented in Figure 8a. In both cases, a fragment ion at m/z 337.18 is present. This MW matches the
molecular formula C20H26O3, a 20 carbon atoms molecule with 7 insaturations. It is then conceivable
that the fragment ion at m/z 337.18 has a diterpenic structure, and that adducts at m/z 931.6609 and
m/z 935.6757 correspond to DAG linked to a dehydroabietic acid. Possible structures for adducts
with m/z 931.6609 are proposed in Figure 8b. Two types of bonds are considered: ester linkage
between the hydroxyl function of DAG and the carboxyl function of diterpenes, and ether linkage on
the carbon in position 7 of the diterpenic acid.

It was not possible to characterize the low intensity adducts in the range m/z 1150-1300. Since these
compounds were only observed on the extracted lipidic fraction of oil/colophony mixtures but are
absent from the 100 wt% oil varnish, it is worth considering that they also may be hybrid molecules
formed from a TAG and a diterpenic molecule. For instance, a hybrid compound formed with a
 ACCEPTED MANUSCRIPT

TAG such as LnLnL and a dehydroabietic acid would have a molar mass of 1197.90, which fits in the
range m/z 1150-1300.

These two different types of supposed hybrid compounds will be referred as H1 (adducts in the range
m/z 930-940) and H2 (adducts in the range m/z 1150-1300).

T
3.3 SEC quantitation of characterized compounds

IP
After attributions were confirmed with ESI-Q-ToF experiments, the characterized compounds were

R
quantitated. Compounds eluting over 27.3 minutes, corresponding to TAGs’ polymers from

SC
tetramers to heptamers (4- to 7-mers), were quantitated together. In order to account for the relative
quantity of the different compounds forming the oil fraction of the varnishes, we divided the

NU
corrected areas by the oil proportion in each composition. The values obtained for diglycerides,
triglycerides, triglycerides dimers, the two types of triglycerides trimers and 4- to 7-mers are presented
in Figure 9, as function of oil proportion.
MA
Triglyceride relative abundance corrected by oil proportion is not equal in all varnishes, but reaches a
maximum for the 50/50 (oil/colophony, wt%/wt%) proportion (Figure 9b). Abundance of dimers
D

reaches a maximum for the 42/58 (oil/colophony, wt%/wt%) varnish, and then decreases with oil
proportion (Figure 9c). The same trend is observed for the low MW trimers (Figure 9d). Concerning
TE

the three varnishes containing the highest proportions of oil (100 wt%, 80 wt% and 66 wt%), the
decrease in the abundance of dimers and low MW trimers can be explained by subsequent
P

oligomerisation of these compounds. Indeed, high MW trimers and 4- to 7-mers are present in these
CE

varnishes (Figure 9e and Figure 9f). In the case of the 100 wt% oil varnish, the slightly lower
abundances of 4- to 7-mers compared to the 80/20 (oil/colophony, wt%/wt%) mixture is probably
AC

due to extended polymerization, giving longer oligomers eluting in the void volume. These data set
forth that, the less colophony there is in the mixture, the more linseed oil is polymerized. Concerning
diglycerides, their relative abundance decreases with oil proportion.

We applied the same approach for colophony compounds and divided their corrected area by the
colophony proportion in each mixture (Figure 10). Colophony trimers are not represented, because
the intensity of the corresponding peak was too low to allow us to define integration limits.

The relative abundance of diterpenes monomers decreases with oil proportion, in particular for
proportions above 50 wt% (Figure 10a). The abundance of dimers is two to three times higher in the
100 wt% colophony varnish than in the mixtures, and slightly decreases with oil proportion (Figure
10b). Trimers (not quantified) are nearly absent in the mixtures. Polymerization is thus far more
extended in the 100 wt% colophony varnish than in the oil/colophony mixtures. Moreover,
abundance of all diterpenic compounds is lower in the mixtures than in the 100 wt% colophony
 ACCEPTED MANUSCRIPT

varnish. Since polymerization cannot account for this phenomenon, these data strongly suggest that
colophony compounds are reacting with linseed oil.

We can hypothesize that the unidentified compounds eluting at 20.9 min on the SEC chromatograms
(Figure 1) correspond to compounds H1, characterized by MS/MS experiments as hybrid molecules

T
formed from a DAG and a diterpenic acid. Indeed, such compounds would elute right after TAGs,

IP
since their hydrodynamic volume is lower, because of the spatially compact diterpenic moiety.
Compounds eluting at 20.0 min could correspond to the non-identified adducts H2, whose masses

R
range from 1150 to 1250 Da and could be formed by a TAG and a diterpenic molecule. In Figure 11

SC
(a and b), the peak areas of these two types of supposed hybrid compounds, H1 and H2, are
represented as a function of oil proportion. The two types of compounds are both more abundant in
the 50-50 mixture (oil/colophony, wt%/wt%), which is consistent with the assumption that they are
hybrid molecules.

4. Discussion
NU
MA
Linseed oil and colophony mixtures prepared by heat treatment are constituted of chain-shortened
TAGs, oxidized TAGs, and polymerization products, covering a MW range from 200 to 7000 Da.
D

Size-Exclusion Chromatography proved to be the method of choice to separate the different

TE

polymeric compounds of oil and colophony.

MS and MS/MS experiments allowed us to identify dimers and trimers of diterpenes, which can be
P

linked with ester, ether, anhydride or ketone linkages. It has been postulated that ester linkage was
CE

formed by carboxylic addition on C-C double bond, and ether linkage by radical addition reactions
[13]. Esterification between a hydroxyl function in position 3 or 7 and the carboxylic
AC

function/anhydride can also be postulated, considering reactivity of colophony with phenolic

compounds [22,25]. However, further study would be necessary to establish with more certainty the
structure of these compounds.

As for linseed oil, SEC experiments show the presence of oligomers up to heptamers. HPLC-MS/MS
experiments focused on the non-polymerized fraction and highlighted the presence of shortened and
oxidized TAGs’ in the 100 wt% linseed oil varnish. Our technique does not discriminate in which
form (keto, hydroxy, epoxy, aldehyde) oxygen is incorporated in the molecules. Relying on previously
published works, we can however assume that oxidized TAGs are more likely to be ketonic and
hydroxyl compounds than epoxides, given their fragmentation [53,54]. A shortened TAG in which
the shortened branch consists of a saturated C8 alkane was identified, as well as TAGs containing one
extra oxygen atom on the shortened branch, very probably as aldehyde moiety. Indeed, short-chain
glycerol-bound aldehydes were identified in heated trioleine and trilinolenin standards [26,32,34].
 ACCEPTED MANUSCRIPT

Two unidentified compounds were present in the SEC chromatograms of oil/colophony mixtures,
eluting right before and after TAGs. Compounds eluting after TAGs were identified by MS/MS.
They originate from the reaction of a DAG with a diterpenic acid. SEC results show that DAGs’
relative abundance increases with colophony proportion, which is consistent with the statement of
[42] that colophony promotes oil hydrolysis. This phenomenon could be explained by the fact that, in

T
the varnishes’ preparation process, moist colophony is added to hot oil: the addition of water to oil

IP
promotes TAG hydrolysis leading to DAGs, which then react with diterpenic compounds. In the

R
case of frying oils, it has been shown that water evaporating from the food enhance oil hydrolysis
[56,57]. The scheme of fragmentation does not allow identifying with certainty the type of linkage

SC
between the DAG and the diterpenic molecule: ether and ester linkage are both conceivable. An ester
of a triterpenoid and a fatty acid was evidenced in pitch tar [58].

NU
Compounds eluting before TAGs probably correspond to the adducts observed in the mass spectra
between m/z 1050 and m/z 1300. Their structure was not identified. Though the measured MWs
MA
match with hybrid addition products of a TAG and a diterpene, it cannot be excluded that they are
compounds formed by addition of acyl chain subunits on intact TAGs, which were indeed
characterized in heated triolein standards [26,59]. However, such addition products are not observed
D

in the fingerprint mass spectrum of the 100 wt% linseed oil varnish. Therefore, it is likely that they
TE

are not formed in the oil/colophony mixtures either. Moreover, these unidentified compounds are
only present in oil/colophony mixtures and are more abundant in the 50/50 (oil/colophony,
P

wt%/wt%) proportion, which reinforces the assumption that they are hybrid molecules.
CE

The relative amounts of linseed oil and colophony oligomers strongly depend on the proportions of
the mixtures. The lesser colophony is in the mixture, the larger is the extent of polymerization in
linseed oil, suggesting that the presence of colophony affects the polymerization mechanisms of
AC

glycerides. Indeed, the reaction of diterpenic acids with triglycerides can be competitive with oil
polymerization. The presence of colophony compounds in high amounts can also result in steric
hindrance, diminishing the probability of two TAGs to react and form oligomeric compounds.

Finally, in the same way that colophony inhibits oil polymerization, the presence of linseed oil inhibits
the formation of diterpenic dimers and trimers. Diterpenic acids are however reacting since the more
linseed oil is in the mixtures, the less colophony “monomers” are present. This result confirms that
diterpenic acids are reacting with linseed oil compounds.

5. Conclusions

This work is part of a fundamental chemistry study dealing with the analysis of markers in mixture of
a drying oil with a Pinaceae resin, that are frequently encountered in coating of musical instruments,
furniture, decorative art objects and easel paintings. The first stage of this work focuses on the
 ACCEPTED MANUSCRIPT

molecular analyses of oil/colophony mixtures prepared by heat treatment in different proportions.

We highlighted that in thermo-oxidative conditions, numerous oxidised and polymerized compounds
of triglycerides and diterpenic acids are formed. Linseed oil’s TAGs also undergo chain-scission
reactions. It is evidenced that the reactivity of linseed oil and colophony strongly depends on the
proportions of the mixture. Moreover, linseed oil and colophony compounds are most probably

T
reacting together to form hybrid adducts. To our knowledge such a phenomenon was never observed

IP
before in scientific publications. The presence of cross-linked diterpenes in heat-treated 20th c.

R
oil/resin paints was recently assessed [60]. Based on our analytical results, it may be suggested that
these compounds consist of oligomeric diterpenes and/or hybrid glyceride-diterpene adducts.

SC
The heat treatment is very probably responsible for the formation of these hybrid addition products;
this type of compounds may therefore be proposed as a marker to determine the processing methods

NU
of varnishes used on art and historic artefacts. However further works should be carried out to assess
the specificity of such a marker.
MA
Before any ageing reaction occurs, the initial abundance of oxidized and polymerized compounds in
oil/colophony mixtures can be extremely different, depending on the original proportion of the
mixture. This single factor is discussed in this work, but linseed oil/colophony varnishes composition
D

is also affected by heat treatment parameters (temperature, time of heating). All these aspects should
TE

be considered in future studies to better understand the molecular ageing pathway of these types of
coatings. The results presented here strongly suggest that conservation state of varnishes, often
P

evaluated through the presence of oxidised markers, is closely linked to the initial proportion of resin
CE

in the mixture.

6. Acknowledgments
AC

The authors sincerely acknowledge Dr. Jacopo La Nasa for his help with performing the HPLC-ESI-Q-
ToF experiments at the University of Pisa. We are grateful to Tony Echavidre, violin-maker (Bordeaux,
France), for his guidance with the preparation of varnishes and to Kremina Kirilova for providing
language help. We would like thank Olivier Segouin and Florian Hodelin (Domaines & Patrimoine,
Besançon, France) for kindly providing us the colophony. We are very grateful to Stéphane Vaiedelich
(Cité de la Musique, Paris, France) and Bertrand Lavédrine (Centre de Recherche sur la Conservation,
Paris, France) for their support along this work. This research was partially supported by La Fondation
des Sciences du Patrimoine (Patrima, Université de Cergy-Pontoise, Cergy-Pontoise, France).

7. References

[1] J.-P. Echard, V. Malecki, Oil- Pinaceae resin varnish recipes in 15th – 18th-century written
sources, in: H. Dubois, J.H. Townsend, J. Nadolny, S. Eyb-Green, S. Kroustallis, S. Neven (Eds.),
 ACCEPTED MANUSCRIPT

Making and Transforming Art: Technology and Interpretation, Archetype, London, 2014: pp.
131–132.

[2] J.-P. Echard, L. Bertrand, B. Lavédrine, S. Vaiedelich, The Nature of the Extraordinary Finish of
Stradivari’s Instruments, Angew. Chemie Int. Ed. 49 (2010) 1–7. doi:10.1002/anie.200905131.

[3] L. Bertrand, L. Robinet, S.X. Cohen, C. Sandt, A.S. Le Hô, B. Soulier, et al., Identification of the

T
finishing technique of an early eighteenth century musical instrument using FTIR

IP
spectromicroscopy, in: Anal. Bioanal. Chem., 2011: pp. 3025–3032. doi:10.1007/s00216-010-4288-
1.

R
[4] G. Chiavari, S. Montalbani, V. Otero, Characterisation of varnishes used in violins by pyrolysis-gas
chromatography/mass spectrometry, Rapid Commun. Mass Spectrom. 22 (2008) 3711–3718.

SC
[5] C. Daher, L. Drieu, L. Bellot-Gurlet, A. Percot, C. Paris, A.-S. Le Hô, Combined approach of FT-
Raman, SERS and IR micro-ATR spectroscopies to enlighten ancient technologies of painted and

NU
varnished works of art, J. Raman Spectrosc. 45 (2014) 1207–1214. doi:10.1002/jrs.4565.

[6] L. Osete-Cortina, M.T. Doménech-Carbó, Analytical characterization of diterpenoid resins present

in pictorial varnishes using pyrolysis–gas chromatography–mass spectrometry with on line
MA
trimethylsilylation, J. Chromatogr. A. 1065 (2005) 265–278. doi:10.1016/j.chroma.2004.12.078.

[7] J.D.J. Van den Berg, N.D. Vermist, L. Carlyle, M. Holcapek, J.J. Boon, Effects of traditional
processing methods of linseed oil on the composition of its triacylglycerols, J. Sep. Sci. 27 (2004)
181–199. doi:10.1002/jssc.200301610.
D
TE

[8] D. Scalarone, M. Lazzari, O. Chiantore, Ageing behavior and pyrolytic characterization of

diterpenic resins used as art materials: colophony and Venice turpentine, J. Anal. Appl. Pyrolysis.
64 (2002) 345–361. doi:10.1016/S0165-2370(02)00046-3.
P

[9] J. Mallégol, J.-L. Gardette, J. Lemaire, Long-term behavior of oil-based varnishes and paints I.
CE

Spectroscopic analysis of curing drying oils, J. Am. Oil Chem. Soc. 76 (1999) 967–976.

[10] I. Bonaduce, L. Carlyle, M.P. Colombini, C. Duce, C. Ferrari, E. Ribechini, et al., New insights into
the ageing of linseed oil paint binder: a qualitative and quantitative analytical study., PLoS One. 7
AC

(2012). doi:10.1371/journal.pone.0049333.

[11] B.A. Parkin, W.H. Schuller, R. V. Lawrence, Thermal Dimerization of Rosin, Ind. Eng. Chem.
Prod. Res. Dev. 8 (1969) 304–306. doi:10.1021/i360031a019.

[12] W.H. Schuller, C.M. Conrad, Thermal Behavior of Certain Resin Acids, J. Chem. Eng. Data. 11
(1966) 89–91. doi:10.1021/je60028a024.

[13] I. Artaki, U. Ray, H.M. Gordon, M.S. Gervasio, Thermal degradation solder reflow of rosin during
high temperature solder reflow, Thermochim. Acta. 198 (1992) 7–20.

[14] V.M. Loeblich, R. V. Lawrence, Inhibition of Resin Acid Isomerization, Ind. Eng. Chem. 50
(1958) 619–620. doi:10.1021/ie50580a031.

[15] I. Portugal, J. Vital, L.S. Lobo, Isomerization of resin Acids During Distillation of Pine Oleoresin,
Chem. Eng. Sci. 51 (1996) 2577–2582.

[16] J.S. Mills, R. White, Natural resins of art and archaeology. Their sources, chemistry, and
identification, Stud. Conserv. 22 (1977) 12 – 31.
 ACCEPTED MANUSCRIPT

[17] H. Takeda, H. Kanno, W.H. Schuller, R. V. Laurence, Effect of Temperature on Various Rosins
and Pine Gum, Ind. Eng. Chem. Prod. Res. Dev. 7 (1968) 186–189. doi:10.1021/i360027a006.

[18] F.C. Izzo, E. Zendri, A. Bernardi, E. Balliana, M. Sgobbi, The study of pitch via gas
chromatography-mass spectrometry and Fourier-transformed infrared spectroscopy: The case of
the Roman amphoras from Monte Poro, Calabria (Italy), J. Archaeol. Sci. 40 (2013) 595–600.
doi:10.1016/j.jas.2012.06.017.

T
IP
[19] M.P. Colombini, F. Modugno, E. Ribechini, Direct exposure electron ionization mass
spectrometry and gas chromatography/mass spectrometry techniques to study organic coatings on
archaeological amphorae., J. Mass Spectrom. 40 (2005) 675–87. doi:10.1002/jms.841.

R
[20] K.J. Van den Berg, I. Pastorova, L. Spetter, J.J. Boon, State of oxidation of diterpenoid Pinaceae

SC
resins in varnish, wax lining material, 18th century resin oil paint, and a recent copper resinate
glaze, in: Prepr. 11th Trienn. Meet. ICOM Comm. Conserv. Edinburgh, Scotland, 1-6 Sept. 1996,
Edinburgh, 1996: pp. 930–937.

NU
[21] K.J. Van den Berg, J. Boon, Mass spectrometric methodology for the analysis of highly oxidized
diterpenoid acids in Old Master paintings, J. Mass Spectrom. 35 (2000) 512–533.
MA
[22] G.-F. Chen, Developments in the field of rosin chemistry and its implications in coatings, Prog.
Org. Coatings. 20 (1992) 139–167. doi:10.1016/0033-0655(92)80002-E.

[23] J.S. Mills, R. White, The Organic Chemistry of Museum Objects, First Edit, Butterworth-
Heinemann, London, 1987. doi:10.1016/B978-0-408-11810-1.50008-8.
D
TE

[24] R. Fujii, K. Arimoto, D.F. Zinkel, Dimeric components from the dimerization of abietic acid, J.
Am. Oil Chem. Soc. 64 (1987) 1144–1149. doi:10.1007/BF02612991.

[25] A.J.D. Silvestre, A. Gandini, Rosin: Major Sources, Properties and Applications, in: M.N.
P

Belgacem, A. Gandini (Eds.), Monomers , Polymers and Composites from Renewable Resources,
CE

Elsevier Ltd., Oxford, 2008: pp. 39–66.

[26] W.C. Byrdwell, W.E. Neff, Non-volatile products of triolein produced at frying temperatures
characterized using liquid chromatography with online mass spectrometric detection, J.
AC

Chromatogr. A. 852 (1999) 417–432.

[27] E.P. Kalogianni, C. Karastogiannidou, T.D. Karapantsios, Effect of the Presence and Absence of
Potatoes under Repeated Frying Conditions on the Composition of Palm Oil, J. Am. Oil Chem.
Soc. 86 (2009) 561–571. doi:10.1007/s11746-009-1380-9.

[28] J.D. Caldwell, B.S. Cooke, M.K. Greer, High Performance Liquid Chromatography–Size
Exclusion Chromatography for Rapid Analysis of Total Polar Compounds in Used Frying Oils, J.
Am. Oil Chem. Soc. 88 (2011) 1669–1674. doi:10.1007/s11746-011-1845-5.

[29] A.I. Hopia, A.M. Lampi, V.I. Piironen, L.E.T. Hyvönen, P.E. Koivistoinen, Application of high-
performance size-exclusion chromatography to study the autoxidation of unsaturated
triacylglycerols, J. Am. Oil Chem. Soc. 70 (1993) 779–784. doi:10.1007/BF02542600.

[30] W.C. Byrdwell, W.E. Neff, Electrospray Ionization MS of High M. W. TAG Oligomers, J. Am. Oil
Chem. Soc. 81 (2004).

[31] A. Zeb, M. Murkovic, Analysis of triacylglycerols in refined edible oils by isocratic HPLC-ESI-MS,
Eur. J. Lipid Sci. Technol. 112 (2010) 844–851. doi:10.1002/ejlt.201000064.
 ACCEPTED MANUSCRIPT

[32] K. Warner, W.E. Neff, W. Craig Byrdwell, H.W. Gardner, Effect of oleic and linoleic acids on the
production of deep-fried odor in heated triolein and trilinolein, J. Agric. Food Chem. 49 (2001)
899–905. doi:10.1021/jf000822f.

[33] W.E. Neff, W.C. Byrdwell, (triolein, trilinolein and trilinolenin) autoxidation products via high-
performance liquid chromatography coupled with atmospheric pressure chemical ionization mass,
J. Chromatogr. A. 818 (1998) 169–186.

T
IP
[34] O. Berdeaux, J. Velasco, G. Márquez-Ruiz, C. Dobarganes, Evolution of short-chain glycerol-
bound compounds during thermoxidation of FAME and monoacid TAG, J. Am. Oil Chem. Soc.
79 (2002) 279–285. doi:10.1007/s11746-002-0474-x.

R
[35] J. Velasco, S. Marmesat, G. Márquez-Ruiz, M.C. Dobarganes, Formation of short-chain glycerol-

SC
bound oxidation products and oxidised monomeric triacylglycerols during deep-frying and
occurrence in used frying fats, Eur. J. Lipid Sci. Technol. 106 (2004) 728–735.
doi:10.1002/ejlt.200401032.

NU
[36] E.N. Frankel, Lipid Oxidation, Woodhead Publishing Ltd, Cambridge, 2014.

[37] G. Picariello, A. Paduano, R. Sacchi, F. Addeo, MALDI-TOF mass spectrometry profiling of polar
MA
and nonpolar fractions in heated vegetable oils, J. Agric. Food Chem. 57 (2009) 5391–5400.
doi:10.1021/jf9008795.

[38] W. Muizebelt, M. Nielen, Oxidative crosslinking of unsaturated fatty acids studied with mass
spectrometry, J. Mass Spectrom. 31 (1996) 545–554.
D
TE

[39] S.L. Hansen, M.R. Myers, W.E. Artz, Nonvolatile components produced in triolein during deep-fat
frying, J. Am. Oil Chem. Soc. 71 (1994) 1239–1243. doi:10.1007/BF02540544.

[40] A. Lattuati-Derieux, S. Gomes, S. Tirat, S. Thao-heu, J.-P. Echard, New insights into molecular
P

evolution of oil/colophony varnishes : towards pyrolysis-gas chromatography/mass

CE

spectrometry-based quantitation, E-Preservation Sci. 11 (2014) 76–85.

[41] C. Daher, Analyse par spectroscopies Raman et infrarouge de matériaux naturels organiques issus
d’objets du patrimoine : méthodologies et applications, PhD Thesis, Université Pierre et Marie
AC

Curie, Paris, 2012.

[42] R. Rivenc, A. Phenix, B. Singer, N. Balcar, L’élimination des vernis huile-résine. Première Partie :
huile de lin et colophane, in: J.-P. Echard, S. Vaiedelich (Eds.), De la Peinture de Chevalet à
l’instrument de Musique : Vernis, Liants et Couleurs, Cité de la musique, Paris, 2008: pp. 89–101.

[43] S. Tirat, J.-P. Echard, A. Lattuati-Derieux, J.-Y. Le Huerou, S. Serfaty, Reconstructing historical
recipes of linseed oil/colophony varnishes: influence of preparation processes on application
properties, unpublished results, submitted to Journal of Cultural Heritage (Manuscript Ref.:
CULHER-D-15-00077).

[44] J. La Nasa, I. Degano, F. Modugno, M.P. Colombini, Alkyd paints in art: characterization using
integrated mass spectrometry., Anal. Chim. Acta. 797 (2013) 64–80. doi:10.1016/j.aca.2013.08.021.

[45] Theodore Turget De Mayerne, Pictoria Sculptoria et quae subalternarum artium présenté M.
Faidutti et Camille Versini, Lyon, 1977.
 ACCEPTED MANUSCRIPT

[46] A.M.A. Guidotti, Metodo facil performare qualunque sia sorte di vernici: della Cina, e del
Giappone : practico in Francia, ed in Inghliterra : secondo gli sperimenti di accreditati
professori, Rimini, 1784.

[47] G.M. Languri, Molecular studies of Asphalt , Mummy and Kassel earth pigments - their
characterisation , identification and effect on the drying of traditional oil paint, PhD Thesis,
University of Amsterdam, 2004.

T
IP
[48] J. La Nasa, Development of innovative analytical tools based on chromatography and mass
spectrometry for the characterization and ageing studies of modern materials in artworks, PhD
Thesis, University of Pisa, 2015.

R
[49] L.C. Herrera, M.A. Potvin, J.E. Melanson, Quantitative analysis of positional isomers of

SC
triacylglycerols via electrospray ionization tandem mass spectrometry of sodiated adducts., Rapid
Commun. Mass Spectrom. 24 (2010) 2745–52. doi:10.1002/rcm.4700.

NU
[50] L. Patiny, A. Borel, ChemCalc: a building block for tomorrow’s chemical infrastructure., J. Chem.
Inf. Model. 53 (2013) 1223–8. doi:10.1021/ci300563h.

[51] S. Maiti, S.S. Ray, A.K. Kundu, Rosin: a renewable resource for polymers and polymer chemicals,
MA
Prog. Polym. Sci. 14 (1989) 297–338. doi:10.1016/0079-6700(89)90005-1.

[52] I. Bicu, Allylic polymers from resin acids, Die Angew. Makromol. Chemie. 255 (1998) 45–51.
D

[53] F. Giuffrida, F. Destaillats, L.H. Skibsted, F. Dionisi, Structural analysis of hydroperoxy- and
epoxy-triacylglycerols by liquid chromatography mass spectrometry., Chem. Phys. Lipids. 131
TE

(2004) 41–9. doi:10.1016/j.chemphyslip.2004.03.008.

[54] A. Zeb, Triacylglycerols composition, oxidation and oxidation compounds in camellia oil using
P

liquid chromatography-mass spectrometry, Chem. Phys. Lipids. 165 (2012) 608–614.

doi:10.1016/j.chemphyslip.2012.03.004.
CE

[55] A. Zeb, Chemistry and liquid chromatography methods for the analyses of primary oxidation
products of triacylglycerols, Free Radic. Res. 49 (2015) 549–564.
doi:10.3109/10715762.2015.1022540.
AC

[56] M. Shimizu, J. Moriwaki, T. Nishide, Y. Nakajima, Thermal Deterioration of Diacylglycerol and

Triacylglycerol Oils During Deep-Frying, J. Am. Oil Chem. Soc. 81 (2004) 571–576.

[57] E.P. Kalogianni, C. Karastogiannidou, T.D. Karapantsios, Effect of potato presence on the
degradation of extra virgin olive oil during frying, Int. J. Food Sci. Technol. 45 (2010) 765–775.
doi:10.1111/j.1365-2621.2010.02188.x.

[58] S.N. Dudd, R.P. Evershed, Unusual Triterpenoid Fatty Acyl Ester Components of Archaeological
Birch Bark Tars, Tetrahedron Lett. 40 (1999) 359–362.

[59] W.C. Byrdwell, W.E. Neff, Dual parallel electrospray ionization and atmospheric pressure chemical
ionization mass spectrometry (MS), MS/MS and MS/MS/MS for the analysis of triacylglycerols
and triacylglycerol oxidation products, Rapid Commun. Mass Spectrom. 16 (2002) 300–319.
doi:10.1002/rcm.581.

[60] M. Kokkori, K. Sutherland, J. Boon, F. Casadio, M. Vermeulen, Synergistic use of Py–THM–

GCMS, DTMS, and ESI–MS for the characterization of the organic fraction of modern enamel
paints, Herit. Sci. 3 (2015) 1–14. doi:10.1186/s40494-015-0058-x.
 ACCEPTED MANUSCRIPT

Figure 1

T
IP
R
SC
NU
MA
D
PTE
CE
AC
 ACCEPTED MANUSCRIPT

Figure 2

T
IP
R
SC
NU
MA
D
PTE
CE
AC
 ACCEPTED MANUSCRIPT

Figure 3

T
IP
R
SC
NU
MA
D
PTE
CE
AC
 ACCEPTED MANUSCRIPT

Figure 4

T
IP
R
SC
NU
MA
D
PTE
CE
AC
 ACCEPTED MANUSCRIPT

Figure 5

T
IP
R
SC
NU
MA
D
PTE
CE
AC
 ACCEPTED MANUSCRIPT

Figure 6

T
IP
R
SC
NU
MA
D
PTE
CE
AC
 ACCEPTED MANUSCRIPT

Figure 7

T
IP
R
SC
NU
MA
D
PTE
CE
AC
 ACCEPTED MANUSCRIPT

Figure 8

T
IP
R
SC
NU
MA
D
PTE
CE
AC
 ACCEPTED MANUSCRIPT

Figure 9

T
IP
R
SC
NU
MA
D
PTE
CE
AC
 ACCEPTED MANUSCRIPT

Figure 10

T
IP
R
SC
NU
MA
D
PTE
CE
AC
 ACCEPTED MANUSCRIPT

Figure 11

T
IP
R
SC
NU
MA
D
PTE
CE
AC
 ACCEPTED MANUSCRIPT

Figure 1: SEC chromatograms of crude oil (CO), oil/colophony varnishes and crude colophony (CC).
From down to up the oil/colophony proportions of the varnishes are 100/0, 80/20, 66/33, 50/50,
42/58, 33/66 and 0/100, wt%/wt%. Intensity of the chromatograms is corrected by the sample
weight

T
Figure 2: ESI-Q-ToF mass spectra of 100 wt% colophony varnish

IP
Figure 3: Tandem mass spectra of adducts at m/z 632.41 (a), 616.39 m/z (b) and m/z 588.41 (c)

R
Figure 4: Attributed formula, theoretical mass and proposed structures for [M,NH4]+ adducts at m/z
632.41 (a), 616.39 m/z (b) and m/z 588.41 (c).

SC
Figure 5 : HPLC-ESI-Q-ToF chromatogram (a) and FIA-ESI-q-TOF (b) spectra of 100 wt% oil
varnish. Acyl substituent abbreviations: Ln, linolenyl (C18:3); L, linoleyl (C18:2); O, oleyl (C18:1); S,

NU
stearyl (C18:0); P, palmityl (C16:0), S. Subscript «O» refers to the presence of an oxygen atom on the
acyl substituent (as a hydroxy, ketone or epoxy function). For Cx:y, x represents the number of
MA
carbon units and y the total number of double bonds, which might be both C=O or C=C. On the
HPLC-ESI-Q-ToF chromatogram, the range 3-8 min is amplified. As DAGs, oxidized TAGs and
chain scission products are overlapping in this range, peaks corresponding to DAGs and oxidized
D

TAGs are presented on a shifted x-scale. On the FIA-ESI-Q-ToF spectra peaks labeled (*) are
TE

contaminants.

Figure 6 : Tandem mass spectra of TAG oxidation (a) and chain-scission products (b). Acyl
P

substituent abbreviations: Ln, linolenyl (C18:3); L, linoleyl (C18:2); O, oleyl (C18:1). Subscript «O»
CE

refers to the presence of an oxygen atom on the acyl substituent (as an hydroxyl, keto or epoxy
function). For Cx:y, x represents the number of carbon units and y the total number of double bonds,
which might be both C=O or C=C.
AC

Figure 7: FIA-ESI-Q-ToF spectra of 50/50 (wt%/wt%) oil/colophony varnish.

Figure 8: Tandem mass spectra of adducts at m/z 931.6396 and m/z 935.6747 (c) and proposed
structures corresponding to m/z 931.6396 (b). Acyl substituent abbreviations: Ln, linolenyl (C18:3);
O, oleyl (C18:1).

Figure 9: Corrected SEC peak area of compounds from linseed oil present in oil/colophony
varnishes, as a function of oil proportion. Each bar represents one replicate.

Figure 10: Corrected SEC peak area of compounds from colophony present in oil/colophony
varnishes, as a function of colophony proportion. Each bar represents one replicate.

Figure 11: SEC peak area of identified compounds present in oil/colophony varnishes, as a function
of oil proportion. Each bar represents one replicate. H1 represent the hybrid compounds eluting at
20.9 min and H2 stands for the second type of supposed hybrid compounds eluting at 20.0 min.
 ACCEPTED MANUSCRIPT

Highlights

- We analyse experimental reconstructions of ancient oil/resin varnish formulations

- We implement analytical protocols applicable on micro-samples

- Numerous oxidised and polymerized compounds are formed during varnish preparation

T
- The reactivity of linseed oil/colophony varnishes depends on the mixtures’ proportions

IP
- Linseed oil and colophony compounds react together to form hybrid adducts

R
SC
NU
MA
D
P TE
CE
AC