G Model

CULHER-2815; No. of Pages 12 ARTICLE IN PRESS

Journal of Cultural Heritage xxx (2013) xxx–xxx

Available online at

ScienceDirect
www.sciencedirect.com

Original article

Artwork conservation materials and Hansen solubility parameters:

A novel methodology towards critical solvent selection
Theodora Fardi a , Emmanuel Stefanis a , Costas Panayiotou a,∗ ,
Steven Abbott b , Sander van Loon c
a
Department of Chemical Engineering, Aristotle University of Thessaloniki, 54124 Thessaloniki, Greece
b
TCNF Ltd, Ipswich, UK
c
Chemical Innovations BV, Science Park 408, 1098 XH Amsterdam, The Netherlands

a r t i c l e i n f o a b s t r a c t

Article history: In research and in actual conservation practice, conservators have to choose adequate methodologies for
Received 7 May 2013 carrying out treatments successfully, while respecting the integrity of artworks. Besides the knowledge
Accepted 13 November 2013 of the authentic materials present in each artwork, conservators must be able to choose appropriate
Available online xxx
conservation materials and methods. Solvents are widely used in cleaning, but solubility issues are also
of high importance in consolidation treatments as well as in protective coating applications. The central
Keywords: aim of this work is to critically re-assess the potential of Hansen solubility parameters (HSPs) for a
Hansen solubility parameters
reliable use in the field of artwork conservation. An effort was made to develop an efficient methodology
Critical solvent selection
Coatings
for critical solvent selection (CSS). For this purpose, two different methods were used for the estimation
Adhesives of various artwork conservation materials. A group-contribution method (GC), based on the chemical
Consolidants composition of materials, was applied for the prediction of HSPs of egg yolk, pine resin and seven red
Organic colorants organic colorants (Mexican, Polish and Armenian cochineal, kermes, madder, lac dye and dragon’s blood),
Binders traditionally used in paintings, textiles and illuminated manuscripts. Additionally, an experimental set
up was used for testing the solubility of the commercial products of synthetic conservation materials,
Primal AC-532K, Beva gel 371 a and b (old and new formula) and a commercial matt varnish made of
dammar and wax. With the direct use of Hansen solubility parameters and the relative energy difference
(RED) between various materials it was made possible to carry out ad hoc “virtual” solubility tests that
may apply to real and complex systems such as cultural heritage artworks.
© 2013 Elsevier Masson SAS. All rights reserved.

1. Research aims and conservation materials. Through the proposed methodology it

The aim of this research is to critically re-assess the use of
Hansen solubility parameters (HSPs) and their potential applica-
tion in the field of artwork conservation. For this reason, a set of
materials was chosen, including binding media, protective coatings,
adhesives and consolidants, and their partial solubility parame-
ters were estimated using two different methods. The first one is
a group-contribution method (GC), which is based on the chem-
ical composition of materials. The second one is experimental,
based on solubility tests with a set of solvents and mixtures. Using
these two complementary methods, it was possible to supplement
the existing databases for HSP of art and conservation materials
(supplementary material). Furthermore, a novel methodology is
proposed for critical solvent selection (CSS), with the direct use of
HSPs and the relative energy difference (RED) between various art
∗ Corresponding author. Tel.: +30 2310 996323; fax: +30 2310 996323.
E-mail address: cpanayio@auth.gr (C. Panayiotou).
is possible to carry out ad hoc “virtual” solubility tests in order to
facilitate hands on laboratory conservation practice and bring new
perspectives in the treatment of complex and often multi-layered
structures, like the cultural heritage artworks.
2. Introduction
In the field of cultural heritage conservation or conservation
science, solvents and solvent formulations are widely used in the
context of academic research as well as in the conservator’s stu-
dio practice. Thus, the concept of solubility, the judicious selection
of the appropriate solvent for the specific solute, and the use of
rational solvent selection schemes, such as the partial or Hansen
solubility parameters (HSPs), are particularly prominent. Since the
theoretical background of the partial solubility parameters has
been extensively discussed elsewhere [1,2], a brief mention of the
relevant basics will be made here.
The starting point of the solubility parameter approach is due to
Hildebrand [3], who introduced the total solubility parameter ıtot ,

1296-2074/$ – see front matter © 2013 Elsevier Masson SAS. All rights reserved.
http://dx.doi.org/10.1016/j.culher.2013.11.006

Please cite this article in press as: T. Fardi, et al., Artwork conservation materials and Hansen solubility parameters: A novel methodology
towards critical solvent selection, Journal of Cultural Heritage (2013), http://dx.doi.org/10.1016/j.culher.2013.11.006
G Model
CULHER-2815; No. of Pages 12 ARTICLE IN PRESS
2 T. Fardi et al. / Journal of Cultural Heritage xxx (2013) xxx–xxx
or cohesive energy density (CED), as the square root of the ratio of
the molar cohesion or potential energy, Ecoh over the molar volume
V:
 Ra 2 = 4(ıd1 − ıd2 )2 + (ıp1 − ıp2 )2 + (ıhb1 − ıhb2 )2
Ecoh
␦tot (CED) =
V where 1 and 2 refer to the solute and the solvent respectively, or
In other words, the cohesive energy density is the total energy
required to eliminate all intermolecular forces that keep molecules
together and, as such, the following equation has been suggested
for vaporizable compounds:
Ecoh − Hvap − RT
where, Hvap is the molar enthalpy of vaporization at temperature
T. Combining equations (1) and (2), we get the following alternative
definition of total solubility parameter:
 gressively higher RED numbers indicate less effective solvents for
H v − RT
␦tot =
V
Hansen [1,2], in the effort of improving the applicability of the
total solubility parameter to compounds whose cohesive energy
derives from dispersive forces but also polar and hydrogen-bonding
forces, has divided the Hildebrand total solubility parameter into
three partial components ıd , ıp and ıhb , respectively. The sum of
squares of these three partial components is the total solubility
parameter, or:
 solubility triangle”, through which it is possible to plot Teas graphs
2 2 2 for several art and conservation materials and get their solubility
␦tot = ␦d + ␦p + ␦hb
An alternative scheme was developed by Teas in 1968 [4], who
used fractional parameters derived from the Hansen solubility
parameters, as shown by the following equations:
ıd
fd =  
ıd + ıp + ıhb
ıp
fp =  
ıd + ıp + ıhb
ıhb
fhb =  
ıd + ıp + ıhb
and
fd + fp + fhb = 1
The partial solubility parameters of solvent mixtures can be eas-
ily obtained by calculating their mole fraction averages. This applies
for both Hansen and Teas or fractional solubility parameters.
For the visualization of solubility data, two different systems
exist: a two-dimensional (triangular) representation scheme or
Teas chart, developed by Teas [4], and a three dimensional (sol-
ubility sphere) scheme as found in reference [5] and in [1,2]. In the
first one, all solvents may be accommodated in a triangular graph
as spots, on the basis of equations equations 5 to 8. Based on exper-
imental data, the solubility windows of materials may be produced
as contours embracing certain “good” solvents while excluding
some “bad” others. The second system is a 3D representation and
every solute – target is considered as a sphere, whose center is
defined by its ıd , ıp and ıhb values. The radius of the sphere, Ro , is
the “interaction” radius considered to embrace the “good” solvents
and exclude the “bad” ones. Of course, this method of depiction
can also be reduced to 2D plots of cross-sections through the cen-
ter of the solubility sphere on a graph perpendicular to each axis
and resulting in three two-dimensional graphs [5]. Coming back
to the 3D representation, the distance between the solute and any
Please cite this article in press as: T. Fardi, et al., Artwork conservation materials and Hansen solubility parameters: A novel methodology
towards critical solvent selection, Journal of Cultural Heritage (2013), http://dx.doi.org/10.1016/j.culher.2013.11.006
solvent – or more generally, between any two materials – is given,
Ra , the solubility parameter distance
(9)
(1)
any two different materials under study. The smaller the Ra , the
better is the solvent for the solute. Another very useful parameter
is the RED number, or the relative energy difference, given by the
equation:
RED = R a /R o (10)
(2)
Ro equals to the maximum value of Ra that still gives a “good”
solvent.
RED becomes 0, if there is no energy difference. RED numbers
less than 1.0 indicate high cohesive energy affinity, while pro-
(3) the particular solute or target. The radius of a solute’s sphere is
defined through experimental solubility results from tests with a
large number of solvents [1,2,5,6].
There is plenty of literature dedicated to the use of solubil-
ity parameters with relevant applications in the field of cultural
heritage. These have focused mainly to the use of Teas chart, the
fractional parameters concentrating, mainly, on varnish removal
from painted artworks [7–15]. Coladonato and Scarpitti developed
a software application called “interactive triangle of solvent and
(4)
windows [16]. The Teas chart system was criticized for its limita-
tions and inaccuracy [5,17,18]. More recently, the direct use the
Hansen solubility parameters was discussed in the field of artwork
conservation [19,20].
The present work reports the first results of a research study
that was carried out regarding the solubility of art and conserva-
(5)
tion materials. The focus is on the direct use of the Hansen solubility
parameters and the possibility to carry out ad hoc “virtual” solubil-
ity tests for the removal or application of conservation materials,
(6)
paying attention to the safety of original constitutive materials of
the artwork as well as the safety of the conservator.
(7)
3. Materials and methods
3.1. HSP estimation methods
(8)
3.1.1. Group-contribution method
The group-contribution method is computational/theoretical
and no experimental data is needed, besides the molecular struc-
ture of the materials, in order to calculate their partial solubility
parameters. The description of the molecular structure is done
using two kinds of functional groups: the first-order groups, which
are used to describe the basic molecular structure, and the second-
order groups that are based on the conjugation theory [21]. The
latter groups are based on the ABC framework, developed by
Mavrovouniotis et al. [22]. The basic equation for calculating the
property values, as derived from the molecular structure of a given
molecular structure is the following:
 
f (x) = Ni Ci + W Mj Dj (MPa)(1/2) (11)
i i
where, Ci is the first-order group-contribution of type i appearing
Ni times and Dj is the second-order group of type j appearing Mj
times. W is a constant, which equals 0 if no second-order groups are
found and 1 if any second-order groups are found in a given com-
pound. The determination of the group-contributions is described
G Model
CULHER-2815; No. of Pages 12 ARTICLE IN PRESS
T. Fardi et al. / Journal of Cultural Heritage xxx (2013) xxx–xxx 3

Table 1
Composition synthetic commercial products used for experimental solubility tests.

Commercial product name Composition pH MFFT Solid content (%)

Primal AC 532K (Rohm & Haas) Acrylic emulsion 8.5–9.5 5–10 ◦ C 46.5

Beva gel 371a old formula EVA copolymers (DuPont Elvax 150 and Allied A-C – 60
Copolymer 400), cyclohexanone resin (Laropal K80),
phthalate ester of hydroabietyl alcohol (Cellolyn 21)
(40%)/naptha and toluene (60%)
Beva gel 371b new formula (2010) As in Beva 371a, aldehyde-ketone resin instead of – – –
cyclohexanone
Clear Matt Varnish (Lefranc & Bourgois) Dammar and wax – – 33

thoroughly in reference [21]. The estimation of the three partial After defining partial solubility parameters, we proceeded to
solubility parameters is done using the following equations: the actual ad hoc “virtual” solubility test. The methodology pro-
 0.4126 posed here adapts the RED number, as defined above by equation
  (10), in order to classify various materials in terms of their cohe-
ıd = Ni Ci + Mj Dj + 959.11 (MPa)(1/2) (12) sive energy difference and then proceed to the critical selection of
i i the potentially effective solvents. For reasons of abbreviation, the
  term ‘target’ will be used hereafter for the solutes, be they mate-
 
ıp = Ni Ci + Mj Dj + 7.6134 (MPa)(1/2) (13) rials for coating, consolidation or adhesion. “Anti-targets” will be
the constitutive materials that should not be affected during treat-
i i
  ment, like binding media and organic colorants. The solvent grid
  used for the “virtual” solubility tests in the present study com-
ıhb = Ni Ci + Mj Dj + 7.7003 (MPa)(1/2) (14) prises of pure solvents and binary mixtures of acetone, ligroin
i i and ethanol (supplementary material, Table SII). This was chosen
When the polar and hydrogen-bonding solubility parameters mainly because it has been widely used for performing cleaning
are less than 3 MPa(1/2) , the following equations are used for these tests in conservation laboratory practice, initially proposed by Cre-
components: monesi [11].
  First, are chosen solvents and mixtures with RED greater than
  1 for the anti-targets. Out of them are chosen, subsequently, the
ıp = Ni Ci + Mj Dj + 2.6560 (MPa)(1/2) (15) solvents with RED lower than 1 for the target material. The solvents
i i and mixtures that fulfil the above requirements both for target(s)
 
 
ıhb = Ni Ci + Mj Dj + 1.3720 (MPa)(1/2) (16)
i i

We have selected this method of estimating HSP believing it to

be reliable and accurate, however there are a number of alternative
procedures for such estimations. References [5,23,24] for example
provide computations for these.

3.1.2. Experimental method

The experimental solubility test method was carried out for the
determination of the Hansen solubility parameters of four commer-
cial synthetic products: acrylic emulsion Primal AC-532K, Beva gel
371a (open pot), 371b (old and new formula) and a matt varnish
by Lefranc Bourgeois made of dammar and wax (Table 1). Materi-
als were applied in liquid form on glass slides and were left to dry
for three months at ambient conditions. They were scraped with
a scalpel and 10 g of each sample were collected. Solubility tests
were carried out using a total of 32 solvents and mixtures for each
solute. The selected solvents make a grid using binary mixtures of
preset proportions of five solvents: methyl isobutyl ketone (MIBK),
toluene, dimethyl sulfoxide (DMSO), propylene carbonate (Pc) and
1,4-Dioxane. These solvents cover a wide space in the HSPiP 3D
representation (supplementary material, Table SI). The solid sam-
ples were mixed with each solvent or blend in concentrations of
5% w/v for Primal AC-532K and 2.5% for the two Beva gel 371
and the matt varnish. The mixing was carried out through auto-
mated standardized method using Chemspeed’s Formax® system,
a software-driven robotic platform.

3.2. Critical solvent selection methodology (CSS)

In Fig. 1, a flowchart of the methodological approach for critical Fig. 1. Flow chart of the steps followed for the critical solvent selection
solvent selection (CSS) is presented. methodology.

Please cite this article in press as: T. Fardi, et al., Artwork conservation materials and Hansen solubility parameters: A novel methodology
towards critical solvent selection, Journal of Cultural Heritage (2013), http://dx.doi.org/10.1016/j.culher.2013.11.006
G Model
CULHER-2815; No. of Pages 12 ARTICLE IN PRESS
4 T. Fardi et al. / Journal of Cultural Heritage xxx (2013) xxx–xxx

and anti-target(s) will be defined as the actual solvents proposed
to be further tested. Moreover, when it is deemed necessary, an
optimization of the chosen actual solvent or solvent mixture can
be made for HSPs, Ra and RED numbers.
It must be made clear that, in every case, Ra remains always the
same for two materials with fixed ı parameters. As mentioned ear-
lier, Ro is defined via experimental data and computer-based data
correlations, and it may vary depending on the set of solvents used
for the solubility tests and the interpretation of results (the crite-
ria that one uses to define a solvent or non-solvent for a particular
solute). Furthermore, when plotting data for RED numbers, these
may vary, depending on the given value of Ro in each case. In other
words, when increasing Ro values RED numbers decrease, but the
sequence of solvents with increasing/decreasing REDs, for the same
target and anti-target, remains the same.
The software HSPiP (4th edition, version 4.0.05) was used for the
calculations of HSPs of solvent mixtures and RED numbers between
the various targets and anti-targets, the plotting of 2D and 3D dia-
grams, and the CSS optimization [5].
4. Results and discussion
4.1. Estimation of Hansen solubility parameters (HSPs)
4.1.1. Group-contribution method
4.1.1.1. Egg yolk. Egg yolk is a traditional binding medium in paint-
ings, especially in byzantine and post-byzantine icon paintings and
in Italian painting before 16th century [25,26]. It is a composite mix-
ture of proteins (∼16,5%) and lipids (∼32,5%), cholesterol (∼5,5%),
carbohydrates (∼1%), water and other inorganic elements including

Fig. 2. Chemical formulas of the of the amino acid and fatty acid components of egg yolk.

Please cite this article in press as: T. Fardi, et al., Artwork conservation materials and Hansen solubility parameters: A novel methodology
towards critical solvent selection, Journal of Cultural Heritage (2013), http://dx.doi.org/10.1016/j.culher.2013.11.006
G Model
CULHER-2815; No. of Pages 12 ARTICLE IN PRESS
T. Fardi et al. / Journal of Cultural Heritage xxx (2013) xxx–xxx 5

phosphorous, potassium and calcium [27]. After total evaporation
of water, the solid fraction is composed of approximately 34% of
proteins and 66% of lipids. Upon drying of paint film with egg yolk
as binding medium, the lipids tend to reach the surface and the
volatile components evaporate.
For the estimation of the partial solubility parameters of the
dried egg yolk in paint films, the contributions of first-order and
second-order groups of the specific amino acids and fatty acids
that are found in egg yolk were obtained and their respective par-
tial solubility parameters were estimated (Fig. 2). The respective
solubility parameters of the dried egg yolk were calculated by the
contributions of the proteinaceous 33% and lipidic 66% components
[27,28]. In Table 2, the first and second-order groups of contribution
for each amino acid and fatty acid are presented. Table 3 summa-

Table 2
First-order and second-order groups of contribution of the fatty acid and amino acid components of egg yolk.

Compound First-order groups Second-order groups First-order groups Second-order groups

Amino acids
Glycine COOH 1 NHm CHn COOH 1 Threonine CH3 1 NHm CHn COOH 1
CH2 NH2 1 CH 1
OH 1
CHNH2 1
COOH 1

Alanine COOH 1 NHm CHn COOH 1 Cystine COOH 2 NHm CHn COOH 1
CHNH2 1 CHNH2 2
CH3 1 CH2 S 2

Valine CHNH2 1 NHm CHn COOH 1 Methionine COOH 1 NHm CHn COOH 1
COOH 1 CHNH2 1
CH< 1 CH2 1
CH3 2 CH3 1
CH2 S 1

Leucine COOH 1 NHm CHn COOH 1 Serine COOH 1 NHm CHn COOH 1
CH2 NH2 1 CHNH2 1
CH2 1 CH2 1
CH3 2 OH 1
CH< 1

Isoleucine CH3 2 NHm CHn COOH 1 Hystidine CH N 1

CH2 1 NH (except as above) 1
CH< 1 >C CH 1
CHNH2 1 CH2 1
COOH 1 CHNH2 1
COOH 1

Proline COOH 1 NHm CHn COOH 1 Lysine CH2 3

CH< 1 COOH 1
CH2 1 CH2 NH2 1
CH2 NH CHNH2 1

Phenylalanine ACH 5 NHm CHn COOH 1 Aspartic acid COOH 2 NHm CHn COOH 1
ACCH2 1 CH2 1
COOH 1 CHNH2 1
CHNH2 1

Tyrosine ACOH 1 NHm CHn COOH 1 Glutamic acid CH2 2 NHm CHn COOH 1
AcH 4 COOH 2
AcCH2 1 CHNH2 1
CHNH2 1
COOH 1
*
Arginine

Fatty acids
a-linolenic acid (or COOH 1 Oleic acid COOH 1
linolenic acid)
>CH2 1 >CH2 14
CH CH 0 CH3 1
CH3 3 CH CH 1
1

Palmitic acid COOH 1 Stearic acid COOH 1

>CH2 1 >CH2 16
CH3 4 CH3 1
1

Linoleic acid COOH 1

>CH2 1
CH3 2
CH CH 1
2
*
1st order group >CO NH is not included in the method.

Please cite this article in press as: T. Fardi, et al., Artwork conservation materials and Hansen solubility parameters: A novel methodology
towards critical solvent selection, Journal of Cultural Heritage (2013), http://dx.doi.org/10.1016/j.culher.2013.11.006
G Model
CULHER-2815; No. of Pages 12 ARTICLE IN PRESS
6 T. Fardi et al. / Journal of Cultural Heritage xxx (2013) xxx–xxx

Table 3
HSPs of the fatty acids and amino acids and the resultant parameters of egg yolk, according to its components’ percentage.

Amino acids % %* ␦d ␦p ␦hb

Glycine 3.5 3.7 16.8 6.9 12.1

Alanine 5.6 5.9 15.9 5.3 11.4
Valine 8.3 8.8 15.5 4.2 10.4
Leucine 9.2 9.7 15 5 12.2
Isoleucine 5.1 5.4 15.6 3.9 10
Proline 4.5 4.8 18.4 6.9 12
Phenylalanine 3.9 4 18.4 4.8 10.1
Tyrosine 2.8 2.9 18.7 6.4 17.5
Serine 9.1 9.5 16.6 7.6 19.1
Threonine 5.6 5.9 16.3 6.9 18.5
Cystine 1.9 2 19.1 7.3 16.3
Methionine 2.3 2.4 17.1 5.5 10.9
Lysine 5.5 – – – –
Hystidine 2.4 2,5 18.4 8.1 12.6
Lysine 5.7 6 16.4 5.6 13.7
Aspartic acid 11.5 12 16.5 7.3 15.6
Glutaminic acid 13.9 14.5 16.5 7 15.2
Sum of amino acids 100 16.52 6.17 13.95

Fatty acids % %a ıd ıp ıhb

Palmitic acid 25–27 27.3 15.8 2.9 5.4

Stearic acid 9–12 12.1 15.9 2.7 4.6
Oleic acid 38–44 44.4 16.1 2.8 5.2
Linoleic acid 13–15 15.2 16.3 2.9 5.9
Linolenic acid 0–1 1 16.5 3 6.5
Sum of fatty acids 100 16.03 2.83 5.3

% ıd ıp ıhb

Fatty acids 66 16.03 2.83 5.3

Amino acids 34 16.52 6.17 13.95
Egg yolk 100 16.2 3.97 8.24

HSPs: Hansen solubility parameters.

*
Percentage contributions adjusted to a total of 100%.

major product of gum rosin (70–80% w/w). As a pine resin, it con-

Fig. 3. Chemical formula of dehydroabietic acid, major component of aged pine
resin.
rizes the calculation process for the estimation of egg yolk’s Hansen
solubility parameters, as calculated from the fractional contribu-
tion of its lipidic and proteinaceous components.
4.1.1.2. Colophony. Colophony, a diterpenic resin, has been used
since long time as a protective varnish in paintings and is the
tains 90–95% w/w abietic and pimaric acids, when fresh, whose
proportions may vary from species to species. Upon distillation,
boiling and ageing, the composition of the resin undergoes changes
through isomerization of the double bonds and oxidation. In aged
samples extracted from paintings, mainly dehydroabietic acid is
usually identified, a product from the oxidative dehydrogenation
of adietadienic acids (Fig. 3). 7-oxodehydroabietic acid may also be
traced as an oxidation product in paint samples [28,29]. Here, the
HSPs estimation of pine resin was based on the group-contributions
of dehydroabietic acid only.
Colophony is an oil-based varnish and it was usually mixed and
boiled with linseed oil for the preparation of paint varnishes. For
the HSPs calculation a mixture of pine resin with polymerized lin-
seed oil was considered, following the recipe that Dionysius from
Fourna suggests in his treatise [30]. Table 4 presents the results

Table 4
Groups of contribution for dehydroabietic acid and HSPs of pine resin, polymerized linseed oil and the their mixture.

First-order groups Second-order groups

CH3 4 (CH3 )2 CH 1
AcH 3 Ring of 6 carbons 2
Ac 3
CH2 5
CH< 2
>C< 2
COOH 1

ıd ıp ıhb

Pine resin 18.7 1.6 4.5

Polymerized linseed oil 16 6 7
Pine resin and polymerized linseed oil (43:57) 17.2 4.1 5.9

HSPs: Hansen solubility parameters.

Please cite this article in press as: T. Fardi, et al., Artwork conservation materials and Hansen solubility parameters: A novel methodology
towards critical solvent selection, Journal of Cultural Heritage (2013), http://dx.doi.org/10.1016/j.culher.2013.11.006
G Model
CULHER-2815; No. of Pages 12 ARTICLE IN PRESS
T. Fardi et al. / Journal of Cultural Heritage xxx (2013) xxx–xxx 7

Table 5
Groups of contribution of nine chromophores, constituents of 7 red organic
colorants.

Chromophore First-order groups Second-order groups

Coccid dyes
(Kermes, Armenian, Polish and Mexican cochineal, lac dye)
Carminic acid COOH 1 C(cyclic) O 2
AcCH3 1 AcCOOH 1
>C O“except” 2
AcOH 3
AcH 2
Ac 5
AcCH 1
>O 1
CH< 3
CH2 1
OH 4

Kermesic acid COOH 1 C(cyclic) O 2

AcCH3 1 AcCOOH 1
AcOH 4
AcH 2
>C O “except” 2
Ac 5

Flavokermesic acid COOH 1 C(cyclic) O 2

AcOH 2 AcCOOH 1
AcH 3
AcCH3 1
>C O “except” 2
Ac 5

Laccaic acid A COOH 2 C(cyclic) O 2

AcOH 5 AcCOOH 2
AcH 4
Ac 8
>C O “except” 2
AcCH2 1
CH2 NH 1
CH3CO 1

Laccaic acid B COOH 2 C(cyclic) O 2

AcOH 5 AcCOOH 2
AcH 4
Ac 8
>C O “except” 2
AcCH2 1
CH2 1
OH 1

Dragon’s blood
Dracorubin >O 2 C(cyclic) O 1
CH3 O 1
CH3 1
>C O 1
AcH 10
>C C< 4
CH C< 3
>CHO 1
Ac 2
CH2 2

Dracorhodin AcH 5 C(cyclic) O 1

Ac 1
CH3 1
CH3 O 1
>C C< 1
>C O 1
CH C< 3
>O“except” 1

Fig. 4. Chemical formulas of the chromophoric components of the seven red organic Madder
colorants. Alizarin AcH 6 C(cyclic) O 2
Ac 4
AcOH 2
>C O “except” 2

Purpurin AcH 5 C(cyclic) O 2

Ac 4
AcOH 3
>C O “except” 2

Please cite this article in press as: T. Fardi, et al., Artwork conservation materials and Hansen solubility parameters: A novel methodology
towards critical solvent selection, Journal of Cultural Heritage (2013), http://dx.doi.org/10.1016/j.culher.2013.11.006
G Model
CULHER-2815; No. of Pages 12 ARTICLE IN PRESS
8 T. Fardi et al. / Journal of Cultural Heritage xxx (2013) xxx–xxx

Table 6
HSPs of the seven red organic colorants, water and ethanol.

Colorant Chromophores % ıd ıp ıhb

Madder Alizarin/purpurin 50/50 19.5 12.10 25.15

Kermes Kermesic acid/flavokermesic acid 88/12 18.74 14.77 37.17

Armenian cochineal Carminic acid/flavokermesic acid 97/3 21.23 21.95 59.32

Polish cochineal Carminic acid/kermesic acid and flavokermesic acid (50/50) 75/25 20.69 20.24 53.23

Mexican cochineal Carminic acid/Kermesic acid and flavokermesic acid (50:50) 99/1 21.28 22.12 60.11

Lac dye Laccaic acid A/laccaic acid B/flavokermesic acid and kermesic acid (50/50) 84/10/6 23.38 21.79 49.16

Dragon’s blood Dracorubin/dracorhodin 50/50 20.1 4.2 1.7

Solvents
Water 15.5 16 42.3
Ethanol 15.8 8.8 19.4

HSPs: Hansen solubility parameters.

derived from identification of the characteristic groups for dehy-
droabietic acid as the major component of aged pine resin and the
calculation of relevant HSPs of a mixture of aged pine resin with
polymerized linseed oil. The partial solubility parameters of the
latter were obtained from reference 2.
4.1.1.3. Natural red organic colorants. As in the case of egg yolk, the
first and second-order characteristic groups for each chromophore
were identified and the respective HSPs of the seven colorants
were calculated based on the percentage contribution of each frac-
tional chromophoric component (Fig. 4). The estimation of HSPs of
coccid dyes was based on experimental data from literature [31].
Due to lack of specific experimental data for the precise contri-
bution of their respective chromophoric components of madder
and dragon’s blood, the relevant percentages were considered as
50/50%. Tables 5 and 6 summarize the results.
4.1.2. Experimental method
The interpretation of the solubility tests was done using the 1–6
grading method and by eye observations, according to the swelling
and/or solvation of the solutes to every component of the solvent
grid [5,6]. For Beva gel 371a, a fine-tuning process was carried out
with extra solubility tests including toluene and hexane while, for
Beva gel 371b, xylene was also tested. These solvents are found at
the borders of the solutes’ sphere and by knowing if they are “good”
or “bad” solvents, a refinement of the sphere was possible. The HSPs
that were estimated through the computer-based correlation of the
experimental data are presented in Table 7.
4.2. RED data correlation
4.2.1. Miscibility and solvation of common art and conservation
materials
In an effort to explore the miscibility and solvation of art
and conservation materials, a small data base was created with
RED numbers of selected couples of organic adhesives or consol-
idants and binding media or colorants and is presented in Table 8
(supplementary material, Table SIII).
As shown in Table 8, egg yolk, linseed oil, and their 50/50 mix-
ture show high relative cohesive energy affinity to some synthetic
consolidants and adhesives. More precisely, these three binding
media exhibit low RED numbers (< 1) versus Primal 532K, Las-
caux 360HV, Aquazol diluted in water (5%), Paraloid B-72 pure or
diluted in acetone (10%). Also, quite low RED numbers (RED < 1.1)
are detected for the same media versus Beva 371b diluted in toluene
(30%) and Paraloid B-72 in toluene (10%). Linseed oil mixed with egg
yolk exhibits high affinity to cyclododecane diluted with toluene
in proportions 30/70, pure Aquazol and Klucel (RED < 1.2). Gelatin
exhibits considerably low affinity to all materials.
Regarding organic colorants, they seem to have quite high
differences in relative cohesive energy from almost all selected
consolidants and adhesives. Exceptions seem to be dragon’s blood
and kermes. The former seems to be susceptible to dissolution in
the case of Primal AC532K diluted (5%) in water, Beva gel 371b
in toluene (70/30) and cyclododecane diluted in ligroin (30/70),
pure Paraloid B-72 and Primal AC-532K. RED numbers decrease
for Primal AC-532K when diluted in water and for cyclododecane
when used in pure form. Paraloid B-72 exhibits higher RED num-
ber if diluted in acetone and lower if diluted in toluene. Moreover,
kermes exhibits quite high affinity (REDs < 1.2) to the three acrylic
emulsions when diluted with water.
The susceptibility of organic colorants to alteration and solubil-
ity is of the most crucial issues in conservation practice. Aiming
to present the correlation of REDs with two solvents widely used
in conservation practice, namely, water and ethanol, plots were
carried out using Ro equal to 5 and 10. As regards HSPs of water,
we have chosen a set of data, which appear to be most widely
used. Other set of data may be also applicable, with varying val-
ues depending on the method used for their estimation [2]. It must
be stressed also that the present study on organic colorants is

Table 7
HSPs and Ro , as predicted using the experimental solubility test.

Solute Solvent grid ıd ıp ıhb Ro

L & B Clear Matt Varnish (Dammar and wax) Standard grid 17.36 3.22 3.65 3.0

Primal AC-532K Standard grid 17.84 8.34 7.19 8.7

Beva gel 371 a Standard grid 17.98 2.56 3.45 1.9

Toluene/hexane fine-tune 17.95 1.27 2.49 0.6

Beva gel 371 b Standard grid 18.02 2.47 3.44 1.9

Toluene/hexane fine-tune 17.85 1.7 2.35 0.7
Xylene fine-tune 17.76 1.39 2.38 0.5

HSPs: Hansen solubility parameters; L & B: Lefranc & Bourgois.

Please cite this article in press as: T. Fardi, et al., Artwork conservation materials and Hansen solubility parameters: A novel methodology
towards critical solvent selection, Journal of Cultural Heritage (2013), http://dx.doi.org/10.1016/j.culher.2013.11.006
 CULHER-2815; No. of Pages 12
G Model
towards critical solvent selection, Journal of Cultural Heritage (2013), http://dx.doi.org/10.1016/j.culher.2013.11.006
Please cite this article in press as: T. Fardi, et al., Artwork conservation materials and Hansen solubility parameters: A novel methodology

Table 8
RED numbers of various adhesives and consolidants (pure and diluted in solvents) versus organic binding media and organic colorants.

Synthetic consolidants and adhesives Organic binding media (Ro :7) Organic pigments (Ro :7)

Egg yolk Linseed oil Tempera Gelatin Madder Kermes Mexican Polish Armenian Lac dye Dragon’s

ARTICLE IN PRESS
grassa cochineal cochineal cochineal blood

T. Fardi et al. / Journal of Cultural Heritage xxx (2013) xxx–xxx

Aquazol Poly (2-Ethyl - 2-Oxazoline) 1.31 1.24 1.19 3.01 2.08 3.79 7.23 6.20 7.12 5.83 1.83

Aquazol 5% in water 0.78 0.76 0.81 4.67 3.54 5.16 8.67 7.64 8.55 7.39 1.41

Beva gel 371 a 4.12 4.00 4.23 6.57 6.07 7.08 10.29 9.34 10.18 9.44 5.08
Beva gel 371 b 1.05

Beva gel 371b toluene (70/30) 1.09 1.09 1.00 4.83 3.62 5.34 8.82 7.79 8.71 7.46 0.70

Cyclododecane 1.37 1.32 1.34 5.25 4.08 5.75 9.26 8.23 9.14 7.94 1.24

Cyclododecane 30% in ligroine 1.22 1.21 1.16 5.04 3.83 5.53 9.02 7.99 8.90 7.67 0.91

HPC (Klucel) 1.31 1.24 1.19 3.01 2.08 3.79 7.23 6.20 7.12 5.83 1.83

HPC 1% in water 5.06 5.20 5.14 1.99 2.72 1.17 3.19 2.27 3.09 2.61 6.14

Lascaux 360Hv (BA(55)/MMA (45)) 0.87 0.78 0.68 3.78 2.80 4.50 7.99 6.96 7.87 6.61 1.26

Lascaux 360Hv (45) 5% in water 4.84 4.98 4.92 1.81 2.51 1.02 3.36 2.42 3.26 2.69 5.92

Paraloid B-72 (PMMA/EMMA) 0.71 0.54 0.50 3.99 2.92 4.64 8.13 7.10 8.02 6.76 1.01

Paraloid B72 10% in acetone 0.91 0.59 0.82 3.70 2.85 4.46 7.95 6.92 7.84 6.64 1.72

Paraloid B72 10% in toluene 1.09 1.08 1.04 4.83 3.62 5.34 8.82 7.79 8.71 7.46 0.70

Primal AC-532K 0.74 0.62 0.55 3.73 2.66 4.39 7.87 6.84 7.76 6.49 1.18

Primal AC-532K 5% in water 4.83 4.97 4.90 1.84 2.50 1.02 3.38 2.43 3.27 2.71 5.91

RED: relative energy difference.

9
G Model
CULHER-2815; No. of Pages 12 ARTICLE IN PRESS
10 T. Fardi et al. / Journal of Cultural Heritage xxx (2013) xxx–xxx

Table 9
RED numbers of red organic colorants versus ethanol and water.

Solvents Madder Kermes Mexican cochineal Polish cochineal Armenian cochineal Lac dye Dragon’s blood

Water (Ro :5) 3.86 1.67 4.42 3.13 4.27 3.63 8.65

Water (Ro :10) 1.93 0.84 2.21 1.57 2.14 1.81 4.33

Ethanol (Ro :5) 1.99 3.93 8.84 7.41 8.68 7.17 4.04

Ethanol (Ro :10) 0.99 1.96 4.42 3.70 4.34 3.58 2.02

RED: relative energy difference.

Fig. 5. Double-Shere 3D plot of madder and kermes (targets) versus water and ethanol. The rest of the organic colorants are also included in the plot: madder (1), kermes
(2), Mexican (3), Polish (4), Armenian (5) cochineal, lac dye (6) dragon’s blood (7), water (a) and ethanol (b). The hollow circles indicate the colorants that are found outside
the spheres and the solid ones indicate the colorants found inside the spheres, with Ro defined as 10.

primarily theoretical. For a comprehensive study, extensive exper- anti-target’s sphere (linseed oil) with RED 0.94, while it exhibits
imental work would be needed for determining Ros . quite low RED versus T (0.52).
The corresponding RED numbers of the seven red colorants ver- The above results show that the only “good” pure solvent from
sus water and ethanol are shown in Table 9. It is evident that all the selected solvent grid is ligroin, which exhibits quite high RED for
colorants show quite high REDs versus water and ethanol, if plot- A1 (1.19) and relatively low RED (0.79) for T1 . Additionally, ligroin
ting with Ro = 5 for both solvents. By increasing Ro to 10, REDs mixed with acetone in proportions 90/10 gives RED < 1 (0.63) for T1
decrease for all colorants, but their sequence does not alter. In and RED > 1 (1.03) for A1 . Through solvent optimization, other sol-
the latter case madder shows RED = 0.99 versus ethanol while vents that may satisfy the CSS criteria include solvesso 150, benzyl
kermes RED = 0.84 versus water. Thus, in the case of cleaning of benzoate, toluene and an 80/20 mixture of benzyl benzoate with
paintings and textiles or paper deacidification, special attention cyclohexanol (Table 10). Fig. 6 illustrates the position and relative
should be paid when madder or kermes is present in the artwork distances in solvent grid and the two spheres of targeted materials
under treatment. Representative 2D and 3D plots are shown on in 2D and 3D plots.
Fig. 5. 4.2.2.1.2. Targeting mastic varnish (T2 ). Targeting mastic var-
nish (T2 ) versus egg yolk mixed with polymerized linseed oil (A2 )
4.2.2. Critical solvent selection (CSS) in proportions 50/50 and the selected solvent grid, it appears that
4.2.2.1. Coatings and binding media. Three representative exam- five solvents and mixtures have high cohesive energy to T2 while
ples of the critical solvent selection methodology are presented in being at a considerable distance from A2 (binding medium). Ethanol
this subsection and discussed. Targets will be referred to as T and shows the highest RED number versus A2 (1.76). Furthermore, two
anti-targets as A. Refer to Tables 3 and 4 and 8 and in supplementary mixtures of ethanol with acetone (25–50%) and two other mixtures
material, Tables III–IV for HSPs and REDs. with ligroin (10–20%) show REDs between 1.05 and 1.48 versus A2 .
4.2.2.1.1. Aged pine resin. Aged pine resin mixed with poly-
merized linseed oil, in proportions 47/53%, was used as the target Table 10
material (T1 ) and was plotted versus pure polymerized linseed oil, Selected solvents for aged linseed oil-pine resin mixture (T1 ) and polymerized lin-
the anti-target (A1 ), presuming that the latter makes the binding seed oil (A1 ).
medium of an underlying paint layer. High cohesive energy affinity Solvent/mixture (% v/v) Linseed oil Pine resin and
between these two materials is expected, thus the selection of a (A1 ) (Ro :7) linseed oil
“good” solvent or a mixture that would dissolve T1 without affect- (T1 ) (Ro :7)
ing A1 . Plotting therefore mentioned materials with Ro equal to 7 1. Solvesso 150 1.22 0.83
and following the CSS steps one can quickly eliminate most of the
2. Ligroine 1.19 0.79
solvents and mixtures. The 50/50 mixture of ligroin and ethanol
has RED 0.68 for the oil-resin but also 0.60 for the oil, so it would 3. Benzyl benzoate 1.18 0.82

not be suggested as an adequate cleaning agent. The same applies 4. Toluene 1.13 0.72
for the 30/70 mixture of ligroin and acetone, with REDs 0.57 and 5. Ligroine/acetone 90/10 1.03 0.63
0.36 versus the oil-resin and the pure oil, respectively. The mixture
6. Benzyl benzoate/cyclohexanol 80/20 1.01 0.68
of ligroin and ethanol 90/10 is situated near the borderline of the

Please cite this article in press as: T. Fardi, et al., Artwork conservation materials and Hansen solubility parameters: A novel methodology
towards critical solvent selection, Journal of Cultural Heritage (2013), http://dx.doi.org/10.1016/j.culher.2013.11.006
G Model
CULHER-2815; No. of Pages 12 ARTICLE IN PRESS
T. Fardi et al. / Journal of Cultural Heritage xxx (2013) xxx–xxx 11

Fig. 6. Double – sphere 2D (a) and 3D (b) plots of aged pine resin mixed with linseed
oil (T1 ) and linseed oil (A1 ). The solvents/mixtures that are found outside the anti-
target’s sphere are indicated by hollow circles and the ones that are found inside
it are indicated by solid circles. The selected solvents are indicated by numbers
according to Table 10. Fig. 7. Double – sphere 2D (a) and 3D (b) plots of mastic resin (T2 ) and egg yolk mixed
with linseed oil (A2 ) versus the solvent grid. The solvents/mixtures that are found
outside the anti-target’s sphere are indicated by hollow circles and the ones that are
Nevertheless, all these solvents exhibit quite high RED numbers found inside it are indicated by solid circles. The selected solvents are indicated by
versus T2 , ranging between 0.83–0.87. In an effort to optimize CSS, it numbers according to Table 11.
was found that mixtures of benzyl benzoate with various solvents,
including water, ethylene glycol, tetrahydrofurfuryl alcohol and
caprolactone, seem to be more effective than the previously men-
tioned ones. Some of the selected mixtures are shown in Table 11.
Representative 2D and 3D plots are shown on Fig. 7.
4.2.2.1.3. Cyclohexanone (T3 ). Cyclohexanone (T3 ) was tar-
geted versus egg yolk (A3 ) and the selected solvent grid. Only three
mixtures are selected through CSS process. Pure ligroin alone and
mixed with acetone (10%) show RED between 0.64–0.65, while 20%
of acetone mixed with ligroin is positioned marginally inside the
sphere of T3 , with RED 0.99 (Fig. 8). Additionally, a mixture of ben-
zyl benzoate and benzyl alcohol (93/7) was found to have RED 0.35
versus T3 and 1.18 versus A3 (Table 12).

Table 11
Selected solvents for mastic resin (T2 ) and egg yolk - polymerized linseed oil mixture
(A2 ).

Solvent/mixture (% v/v) Egg yolk and Mastic

linseed oil 50:50 (T2 ) (Ro :17.9)
(A2 ) (Ro :7)

1. Ethanol 1.76 0.92

2. Ligroin/ethanol 10/90 1.48 0.87

3. Aceton/ethanol 25/75 1.38 0.88

4. Ligroin/ethanol 20/80 1.21 0.83

5. Benzyl alcohol/propylene 1.17 0.65

carbonate/ethanol/ethylene
glycol/acetone
25/25/20/15/15
Fig. 8. Double – sphere 2D (a) and 3D (b) plots of cyclohexanone (T3 ) and egg yolk
6. Benzyl ben- 1.10 0.48
(A3 ) versus the selected solvent grid. The solvents/mixtures that are found outside
zoate/caprolactone/water
the anti-target’s sphere are indicated by hollow circles and the ones that are found
53/32/15
inside it are indicated by solid circles. The selected solvents are indicated by numbers
7. Acetone/ethanol 50/50 1.05 0.80
according to Table 12.

Please cite this article in press as: T. Fardi, et al., Artwork conservation materials and Hansen solubility parameters: A novel methodology
towards critical solvent selection, Journal of Cultural Heritage (2013), http://dx.doi.org/10.1016/j.culher.2013.11.006
G Model
CULHER-2815; No. of Pages 12 ARTICLE IN PRESS
12 T. Fardi et al. / Journal of Cultural Heritage xxx (2013) xxx–xxx
Table 12
Selected solvents for cyclohexanone resin (T3 ) and egg yolk (A3 ).
Solvent/mixture (% v/v) Egg yolk
(A3 ) (Ro :7)
1. Naphtha (high-flash) 1.18
2. Ligroine/ethanol 20/80 1.12
3. Benzyl benzoate/benzyl alcohol 93/7 1.14
4. Ligroin/acetone 90/10 1.05
5. Further discussion and conclusions
As shown, the numerous and important application options of
partial solubility parameters in conservation practice seem not
to have been exhausted yet. By simulating case studies that may
occur in conservation practice, it was made possible to extract
encouraging results regarding solvent selection through correla-
tion of solubility data and relative cohesive energy differences
(RED) between materials and solvents.
The methods used for the HSP prediction seem to be comple-
mentary, since the group-contribution method can be widely used
for the prediction of nearly any material with a given chemical
structure and geometry while the experimental one can be applied
on any material available in a relatively small quantity, even if
its composition is unknown, like in the case of some commercial
conservation products.
As regards the CSS process, it must be stressed that there is no
pre-fixed solution. The actual solubility tests are, without doubt,
the final decision step in conservation treatment. As discussed pre-
viously, one basic reason for this is the relative character of RED
numbers. These may vary or change depending on the set of used
experimental data. Another reason is, of course, the undisputed
fact that artworks often comprise systems of extreme complexity
and many parameters must be taken into consideration in order to
proceed to actual treatments. Some of these parameters are:
• the state of preservation of organic components found in art-
works (ageing and denaturation of organic materials, oxidation
of resins, etc.);
• the past interventions;
• the method of application of the actual cleaning treatment (time
of solvent retention, mechanical attrition, etc.);
• the influence of inorganic materials in the cleaning process.
The CSS methodology presented here enables the conservator to
build up on case studies – provided that the identity of the mate-
rials present in the artwork is known. This may facilitate studio
practice and improve interventions related to cleaning or consoli-
dation treatments. Critical solvent selection can be carried out using
RED data alone, but the additional use of 2D and 3D plots seems
to be a very useful method for visualizing results with enhanced
accuracy compared to the traditional TEAS chart.
It is hoped that the proposed methodology will form a basis
and promote discussions concerning solubility issues in conserva-
tion science. Of course, much work needs to be done as regards
the evaluation of the various available HSP prediction methods. A
greater variety of conservation materials need to be studied. The
integration of other physicochemical properties of polymers and
solvents, such as diffusivity, molar volume, vapor pressure, etc.
must be considered along with the shifting of HSP values during
ageing of materials. Furthermore, the division of the hydrogen-
bonding parameter into its acidic and basic components may enable
us to further explore the key properties of artwork materials and
incorporate also aqueous cleaning agents and other chemical clean-
ing systems. In this respect, work is underway in our laboratory
Please cite this article in press as: T. Fardi, et al., Artwork conservation materials and Hansen solubility parameters: A novel methodology
towards critical solvent selection, Journal of Cultural Heritage (2013), http://dx.doi.org/10.1016/j.culher.2013.11.006
by exploring the recently proposed partial solvation parameter
approach [32], whose molecular descriptors are based on quantum
Cyclohexanone chemical calculations.
(T3 ) (Ro :14.9)
0.65 Appendix A. Supplementary material
0.99
Supplementary material associated with this article can be
0.40
found, in the online version, at http://dx.doi.org/10.1016/j.culher.
0.64 2013.11.006.
References
[1] C.M. Hansen, The Three Dimensional Solubility Parameter and Solvent Diffusion
Coefficient, Danish Technical Press, Copenhagen, 1967.
[2] C.M. Hansen (Ed.), Hansen Solubility Parameters. A User’s Handbook, Second
ed., CRC Press–Taylor and Francis Group, Boca Raton, 2007.
[3] J. Hildebrand, R.L. Scott, Regular Solutions, Prentice-Hall, Englewood Cliffs, New
Jersey, 1962.
[4] J.P. Teas, Graphic analysis of resin solubilities, J. Paint Technol. 40 (516) (1968)
19–25.
[5] S. Abbott, C.M. Hansen, H. Yamamoto, Hansen solubility parameters in practice,
Hansen-Solubility.com, 2013.
[6] C.M. Hansen, 50 years with solubility parameters – past and future, Prog. Org.
Coat. 51 (2004) 77–84.
[7] G. Hedley, Solubility parameters and varnish removal: a survey, Conservator 4
(1980) 12–18.
[8] J. Burke, Solubility parameters: theory and application, AIC Book Pap. Group
Ann. 3 (1984) 13–58.
[9] R.L. Feller, N. Stolow, E.H. Jones, On Picture Varnishes and their Solvents,
National Gallery of Art, Washington DC, 1985.
[10] P. Cremonesi, Materiali e Metodi per la Pulitura di Opere Policrome, Phase,
Firenze, 1997.
[11] P. Cremonesi, L’uso dei solvent organic nella pulitura di opera policrome, Il
Prato, Padova, 2004.
[12] V. Horie, Materials for Conservation. Organic Consolidants, Adhesives and
Coatings, Second ed., Butterworth-Heinemann, Oxford, 2010.
[13] G. Dale Smith, R. Johnson, Strip ‘teas’–solubility data for the removal (and appli-
cation) of low molecular weight synthetic resins used as in painting media,
WAAC Newsl. 30 (1) (2008) 11–19.
[14] A. Phenix, The swelling of artists’ paints in organic solvents. Part 2, comparative
swelling powers of selected organic solvents and solvent mixtures, J. Am. Inst.
Conserv. 41 (2002) 61–90.
[15] C. Stavroudis, T. Doherty, A novel approach to cleaning II: extending the mod-
ular cleaning program to solvent gels and free solvents. Part 1, WAAC Newsl.
29 (3) (2007) 9–15.
[16] http://www.icr.beniculturali.it/flash/progetti/TriSolv/TriSolv.html. [last
accessed on 24/03/2013].
[17] C. Stavroudis, S. Blank, Solvents & sensibility, WAAC Newsl. 11 (2) (1989) 2–10.
[18] C. McGlinchey, Boundaries of the teas solubility concept, WAAC Newsl. 22 (2)
(2002) 17–19.
[19] E. Stefanis, Physical and Chemical Parameters of Paper Conservation, Aris-
totle University of Thessaloniki, Thessaloniki, 2007 (unpublished PhD thesis
[in Greek]).
[20] E. Stefanis, C. Panayiotou, Selection of solvents for the conservation of art-
works, paper and textiles, in: Proceedings of the Symposium “Works of Art
and Conservation Science Today”, Thessaloniki, Greece, November 26–28, 2010
(Proceedings CD-ROM).
[21] E. Stefanis, C. Panayiotou, Prediction of Hansen solubility parameters with a
new group-contribution method, Int. J. Thermophys. 29 (2008) 568–585.
[22] M.L. Mavrovouniotis, Estimation of properties from conjugate forms of molec-
ular structures: the ABC approach, Ind. Eng. Chem. Res. 29 (1990) 1943–1953.
[23] J. Bicerano, Prediction of Polymer Properties, Third ed. revised and expanded,
Marcel Dekker Inc., New York, 2002.
[24] D.W. Krevelen, K. te Nijenhuis, Properties of Polymers. Their Correlation with
Chemical Structure; Their Numerical Estimation and Prediction from additive
Group-contributions, 4th completely revised ed., Elsevier, Amsterdam, 2009.
[25] D. Bomford, J. Dunkerton, D. Gordon, R. Ashok, Art in the Making. Italian Painting
Before 1400, National Gallery of Art. Yale University Press, London, 1989.
[26] R. Gettens, G. Stout, Painting Materials, A Short Encyclopaedia, Dover Publica-
tions, New York, 1966.
[27] W.J. Stadelman, O. Cotterill (Eds.), Egg Science and Technology, Haworth Press,
New York, 1995.
[28] M.P. Colombini, F. Modugno (Eds.), Organic Mass Spectrometry in Art and
Archaeology, John Willey and Sons Publications, West Sussex, 2009.
[29] J.S. Mills, R. White, The Organic Chemistry of Museum Objects, Butterworth-
Heinemann, London, 1996.
[30] D. tou ek Fourna, Hermeneia tes zographikes technes, Hermeneia Publ., Agion,
Oros, 2007.
[31] J.H. Hofenk de Graaft, The Colourful Past, Origins, Chemistry and Identification
of Natural Dyestuffs, Archetype Publications, London, 2004.
[32] C. Panayiotou, Partial solvation parameters and mixture thermodynamics, J.
Phys. Chem. B 116 (2012) 7302–7321.