Full text of DOI:10.1557/mrs2001.19

www.mrs.org/publications/bulletin
screw-driven tensile testers. The average
strain rate was a very slow 0.0005 mm/mm
per second, to minimize time-dependent
behavior. Temperature and relative humid-
T
raditional Oil Paints: ity were kept constant at 22ꢀC and 50%,
respectively.
Rates of Oxygen Uptake for
The Effects of
Pigmented, Cold-Pressed
Linseed Oil
Most of the literature describes the dry-
ing of oil paints as a process involving an
uptake of oxygen, an intermediate forma-
tion of hydroperoxides, and, finally, poly-
merization. This entire free-radical process
is called autoxidation. Oxygen absorption
is easily monitored by weighing paint
samples over time, the weight change re-
flecting oxygen uptake minus the loss of
some volatile compounds. The weight loss
due to evaporation of volatiles is small
compared with the oxygen uptake during
the first year of drying in a pigmented
film. The term “drying” has many mean-
ings, but is taken here to mean stabilization
of the oxidative and physical processes
over time. Paints made with different pig-
ments absorb differing amounts of oxygen,
largely as a result of the amount of oil used
in making the paint. Some paints have high
pigment-volume concentrations (PVCs) and
low oil-volume concentrations (OVCs) and
other paints have the reverse. At the same
time, the rate of oxygen absorption varies
from paint to paint and is a result of the
pigment used in the paint. Over time, un-
pigmented cold-pressed linseed oil in-
creases its weight by about 14% of its
initial weight, due to this oxygen absorp-
tion. Even after two years, the pure oil is
rubbery, tacky, and difficult to handle,
suggesting that the drying process is slow.
Paints appear to be even slower driers,
even though they feel stiffer and less
tacky. Chemically, the drying process goes
on for years. Much of this apparent per-
ception of “dryness” is a result of the pig-
mentation of the paint, as the pigment is
considerably stiffer than the medium.
Lead carbonate and cold-pressed linseed
oil were used to make a traditional white-
lead paint having a pastelike consistency
typical of artists’ paints. The resulting
paint has a 44.4% PVC and a 55.6% OVC.
The percentage weight of oil for this paint
is 15.9%, and the oil component alone is
expected to increase in weight by 14% in
the drying process. The total weight of the
paint sample should increase by 2.25%
over a similar drying time. However, the
white-lead paint gained only 1.1% in weight
after 1.5 yr, as illustrated in Figure 1. Fur-
thermore, after an initial period of about
one month, the rate of weight increase set-
tles into a natural log increase, as shown in
Long-Term Chemical
and Mechanical
Properties on
Restoration Efforts
Marion F. Mecklenburg and Charles S.Tumosa
Introduction
Some of the most important cultural
question held the most interest, in that
some conservation research has been done
on modern paints up to 10 or so years old
in the hope of developing a better under-
standing of the conservation treatment of
paintings and painted surfaces that are
icons in the world are oil paintings. Pre-
serving them for future generations requires
a fundamental understanding of the long-
term chemical, mechanical, and physical
behavior of their components. If the prop-
erties are understood, modeling and even
predicting the effects of exposure to changes
in temperature, relative humidity, shock,
and vibration are possible.1 Furthermore, if
the chemistry and rate of the drying of oil
paints is understood, predicting the effects
of both structural and cosmetic (cleaning)
conservation treatments is possible.
3
–6
hundreds of years old. The principal
criticism of these experiments was that the
paint samples used by these researchers
were not old enough to truly represent old
paint films. This paint research program
was undertaken to study the effects of
pigments on the behavior of oils and to
determine when the chemical and physi-
cal processes of drying reached equilib-
rium. The determination of the time frame
for chemical (oxidation) and physical (me-
chanical) processes is reported here.
,2
In the late 1980s, the Smithsonian Cen-
ter for Materials Research and Education
SCMRE), formerly the Conservation
(
Analytical Laboratory (CAL), purchased
custom-manufactured oil paints for a re-
search program examining several aspects
of oil paint behavior. Thirty-nine different
pigments and six different oils were used
in mixing the paints. No other fillers, ex-
tenders, or driers were added to the paints,
which is in marked contrast to commercial
artists’ oil paints currently manufactured.
This restriction was imposed so that the
custom-made paints would more closely
resemble those used in past centuries. The
only consideration was the interaction of
pigment and oil. Two of the most pressing
questions were how do pigments affect
the drying of oil paints, and how long
does it take for paint to dry? The latter
Sample Preparation andTesting
The paints were prepared for drying
by uniformly applying them to clear poly-
ester sheets so that, after a period of time,
“dry” paint films could be cut and the
polyester film backing peeled away. The
thickness of a typical paint sample was
0.25 mm. In this way, an unsupported
paint sample was capable of being tested
for chemical, physical, and mechanical
properties. Several hundred paint samples
were prepared and tested over the past
10 years. Specimens were weighed peri-
odically with a Mettler balance to 0.1 mg.
Tensile tests were performed on miniature
MRS BULLETIN/JANUARY 2001
51

Traditional Oil Paints:The Effects of Long-Term Chemical and Mechanical Properties
Figure 2. When the data are extrapolated
into the future, they indicate that the paint
will reach the theoretical weight limit only
after several centuries. With the exception
of a paint made with cobalt-blue pigment
and cold-pressed linseed oil, other paints
follow a similar behavior as the white
lead. The cobalt-blue paint gains weight
much more rapidly and appears to reach
its maximum limit within decades rather
than centuries, as shown in Figures 3 and
of the paint is from the oil component. The
theoretical maximum weight gain for the
cobalt-blue paint was 8.48%. It is impor-
tant to note that the data suggest that the
maximum weight gain will be ultimately
reached.
Clearly, different pigments can modify
the chemistry of drying oil paints, and this
can in part explain the lower-than-
expected weight gain of the white-lead
paint. On the other hand, it can be readily
shown that thin surface films form on the
drying paints, and these films act as bar-
riers to further oxygen absorption. They
also give the illusion that the paint has
dried throughout. A slowly but ever-
thickening surface film gives rise to
slower diffusion and produces the natural
log absorption rate observed. Regardless
of the explanation, the results of these
measurements strongly suggest that some,
if not most, oil paints take a very long time
to dry (i.e., for their bulk properties to
reach equilibrium). If applied paint is very
thin, then the paint will give the impres-
sion of drying very quickly.
4
. This paint has a 14.7% PVC and an
5.3% OVC, and 60.6% of the total weight
8
Figure 1. The weight gain of white-lead oil paint with time. There
is an initial rapid weight gain due to oxygen absorption for about
one month, after which the weight increase is very slow. The
theoretical maximum weight gain is the expected weight increase
of the oil component alone.
Figure 3. The weight gain of cobalt-blue oil paint with time. There
is an initial rapid weight gain up to about one month, after which
the weight increase slows but is much more rapid than that of
white-lead paint. The theoretical maximum weight gain represents
the expected weight increase of the oil component alone.
Figure 2. The weight gain of white-lead oil paint plotted against
the natural log of the drying time. Once the early initial and rapid
weight gain is completed, the rate follows the natural log plot quite
linearly. The dashed lines indicating the points of 50 yr and 100 yr
are included to provide references to the length of time required
for the paint to completely gain the weight expected. The data
indicate that centuries are required for the paint to completely
gain the maximum weight.
Figure 4. The weight gain of cobalt-blue oil paint plotted against
the natural log of the drying time. Once the early initial and rapid
gain is completed, the rate follows the natural log plot quite
linearly. The dashed lines indicating a point at 25 yr is included
to provide reference to the length of time required for the paint
to reach the weight expected. In contrast to the white-lead paint,
the data here indicate that it will take a little less than 25 yr for
the paint to reach the maximum weight.
52
MRS BULLETIN/JANUARY 2001

Traditional Oil Paints:The Effects of Long-Term Chemical and Mechanical Properties
MechanicalTests and
on a time versus stress plot, as shown in
Figure 6. In this figure, while the stress is
shown to increase over time for any given
strain, the rate of this stress increase di-
minishes with time. Also in this figure are
the values of the breaking stresses over
time, which initially rise, but then start
to level out. This suggests that there is a
maximum stress level for the paint that is
reached before the paint is completely dry.
The rate of stress increase or stiffening of
the paint with time can be more clearly
shown when the natural log (ln) of time is
used instead of linear time for the ordinate
of the plot.
Figure 7 shows that all of the data be-
yond the very early stages of drying (three
months) plot as linear functions of the ln
of time. Even more useful is the potential
to extrapolate the data (dashed lines) to
extended periods of drying time and ex-
amine the possible long-term results of
drying. All of the vertical lines on Figure 7
represent lines of constant drying times.
The stresses indicated by the intersection
of the lines at constant drying times and
the lines of constant strain can be used to
construct the expected stress–strain plots
of the paint film in the future.
paints are shown to be less, since it is
known that the stiffer (the older) the paint,
the more prone to fracture it will be. Fig-
ure 8 suggests that while there is a signifi-
cant difference between the 1-yr-old paint
and the 50-yr-old paint (which is about
2.6 times stiffer), there is only predicted to
be about a 20% increase in stiffness be-
tween the 50- and 250-yr-old paints.
Observations overTime
As a paint polymerizes during drying,
it becomes stiffer and stronger. The stiff-
ness and strength of the paint can be
measured using mechanical testing meth-
ods. Increased stiffness means that the paint
sample requires more force (stress) for the
same amount of elongation (strain). The
strength of the paint is the highest stress
reached, and in this case, this occurs when
the sample breaks. Since the paint samples
are unsupported films (not bonded to sub-
strates), tensile testing yields the strength
of this paint directly. The rate of any in-
crease in stiffness can be used to obtain a
further and independent measure of the
rate of drying of paints. The same white-
lead paint described earlier in this article
will be used to illustrate some points, al-
though the same behavior was observed
in several other paints, including titanium
white in cold-pressed linseed oil and some
commercially prepared white paints con-
taining different oils.
If the projected stress–strain behavior
shown in Figure 8 has validity, it sug-
gests—at least for white lead in cold-
pressed linseed oil—that the drying time
is very long, and the paint continues to
stiffen and perhaps become brittle after
several centuries. The projection also sug-
gests that a paint that does not have con-
siderable flexibility at the 1–2-yr drying
stage will in fact become very brittle over
time. Solvent tests conducted on paints
less than 20–25 yr old leach out a consid-
erable amount of low-molecular-weight
materials, embrittling the paint film. This
should not be confused with aging or dry-
ing effects. Another result is that attribut-
ing the increases in stiffening of the paint
to chemical degradation processes may
also have little meaning, since an old paint
gets stiffer naturally without additional
chemical interactions. What remains to be
determined is the natural rate of processes
such as hydrolysis and what impact these
processes have on the mechanical proper-
ties of paint.
The tests were conducted after the paint
was allowed to dry for 0.184 yr, 0.269 yr,
0
.98 yr, and 10 yr in a controlled indoor
environment typical for artists’ paintings.
Figure 5 illustrates the results of the ten-
sile tests of the white-lead paint and the
age of each test specimen. Also in this fig-
ure are vertical dashed lines (lines of con-
stant strain) that intersect the stress–strain
plots. As can be seen, for any value of
strain, the stress increases as the paint
samples age.
The information provided in Figure 7
was used to reconstruct the stress–strain
plots of white-lead paint at 1 yr and 10 yr,
and this is shown in Figure 8. Also shown
in Figure 8 are the predicted stress–strain
plots for the paint at 50 yr, 100 yr, and
250 yr. The theoretical strength of the
50-yr-old paint sample is estimated to be
at a maximum at approximately 3.7 MPa.
The strengths of the 100- and 250-yr-old
On the other hand, if the paint drying
process stops sooner than the projections
suggest, then the paints will dry to more
flexible films than anticipated in the long
run. In many respects, the projected plots
The information shown in Figure 5—
time, stress, and strain—can be re-plotted
Figure 6. The plots of the paint stresses at constant strains plotted
against the drying time of the paint. After 10 yr, while the rate of
the stress increase is diminishing, the stresses are clearly still
rising. This paint has not dried after 10 yr. The breaking or failure
stress is still rising, but at a lower rate of increase than the
stresses at the fixed strains. This suggests that the strength of
the paint will reach a maximum before the paint is fully dried.
Figure 5. The stress–strain results of white-lead oil paint tested
at different time intervals and under identical test conditions. The
vertical dashed lines are lines of constant strain, and the stress of
the paint at these constant strains increases with time.This indicates
that the oil polymerization is continuing and is not completed
during this test period; that is, the paint is not “dry” after 10 yr.
MRS BULLETIN/JANUARY 2001
53

Traditional Oil Paints:The Effects of Long-Term Chemical and Mechanical Properties
Figure 8. The stress–strain plots for white-lead paint at 1 yr and
1
1
0 yr and the predicted stress–strain plots for the future (50 yr,
00 yr, and 250 yr). The plots suggest that the paint continues
Figure 7. The plots of the paint stresses at constant strains plotted
against the natural log of the drying time of the paint. After an
initial drying period of about three months, the data show a linear
stress increase. This suggests that the drying of the paint is an
ongoing process with ever-diminishing stress increases. The data
also suggest that the behavior of the paint can be projected into
the future, and the stresses are clearly still rising. This paint has
not dried, even after 10 yr. This information allows one to predict
the stress–strain behavior of this paint into the future.
to stiffen over long periods of time as a result of the natural drying
processes alone.
represent a conservative approach, in
the sense that this is the most severe
expectation.
conservation treatments and confuses the
issue of whether natural aging or previous
treatment is responsible for the present
condition of many paintings.
Technology and Practice for Easel Paintings and
Polychrome Sculpture, edited by J. S. Mills and
P. Smith, Preprints of the Contributions to the
Brussels Congress (International Institute for
Conservation of Historic and Artistic Works,
London, 1990) p. 98.
4. M.R. Shilling, H.P. Khanjian, and D.M. Carson,
Techne 3 (1997) p. 71.
5. J.J. Boon, S.L. Peulve, O.F. Van den Brink,
M.C. Duursma, and D. Rainford, in Early Italian
Paintings: Techniques and Analysis, edited by T.
Bakkenist, R. Hoppenbrouwers, and H. Dubois
(Linburg Conservation Institute, The Nether-
lands, 1997) p. 35.
Clearly the data indicate that chemical
and physical processes are going on that
will take decades, if not centuries, to com-
plete. A lack of understanding of these
chemical and mechanical drying proc-
esses of traditional oil paints has led to
potentially damaging conservation treat-
ments, particularly when cleaning solvents
are used. The results of this research pro-
gram suggest that great care must be taken
in the modeling of cleaning methods, con-
servation treatments, and environmental
effects if recently prepared paint samples
are used in experimental testing. This
presents difficulties in forming models for
References
1
. M.F. Mecklenburg and C.S. Tumosa, in Art
in Transit: Studies in the Transport of Paintings,
edited by M.F. Mecklenburg (National Gallery
of Art, Washington, DC, 1991) p. 173.
2. M.F. Mecklenburg and C.S. Tumosa, in Art
in Transit: Studies in the Transport of Paintings,
edited by M.F. Mecklenburg (National Gallery
of Art, Washington, DC, 1991) p. 137.
6. A. Phenix, “The Science and Technology of
the Cleaning of Pictures: Past, Present and
Future,” 25 Years School of Conservation: Preprints
of the Jubilee Symposium, edited by K. Borchersen
(Det Kongelige Danske Kunstakademi, Copen-
3
. G. Hedley, M. Odylha, A. Burnstock, J.
Tillinghast, and C. Husband, “A Study of the
Mechanical and Surface Properties of Oil Paint
Films Treated with Organic Solvents and
Water,” in Cleaning, Retouching and Coatings:
hagen, 1998) p. 109.
ꢀ
.
..for information available via the internet that is of
Advertisers in This Issue
Page No.
18
interest to the Materials Research Community...
Chemat Technology, Inc.
High Voltage Engineering
Inside front cover
www.mrs.org/gateway/
Huntington Mechanical
Laboratories, Inc.
Outside back cover
Materials Information
includes
JY Horiba
37
55
Society of Vacuum Coaters
ꢀ
ꢀ
ꢀ
ꢀ
Patents—Intellectual Property Network from IBM
Thermionics Vacuum Products
64
Atomic and Plasma physics databases from the Weizmann Institute
ENGnetBASE—Engineering Database Online
For free information about the products
and services offered in this issue, check
www.mrs.org/publications/bulletin/
advertisers or fill out and fax the inside
back cover to 724-779-4397.
3-D Molecular Structures Library at New York University
54
www.mrs.org/publications/bulletin
MRS BULLETIN/JANUARY 2001