Uncovering lead formate crystallization in oil-based paintings

Article abstract of DOI:10.1039/d0dt00327a

Lead carboxylates are an extensive group of compounds studied for their promising industrial applications and for their risky behavior when they are formed in oil paintings as corrosion products of lead-based pigments, leading to serious deterioration of paintings. Although the processes leading to the formation of aggregates, protrusions or inclusions, affecting undesirably the appearance of paintings, are assumed to be long term, neo-formed lead carboxylates are detectable in the early stage of paint drying. To uncover the chemical changes in lead pigments during the drying of oil paint films, model systems consisting of minium (Pb₃O₄) and four common drying oils were studied by X-ray powder diffraction (XRPD), ¹³C and ²⁰⁷Pb solid state NMR (ssNMR) spectroscopy and Fourier-transformed infrared spectroscopy (FTIR). For the first time, a degradation mechanism of Pb₃O₄via the crystallization of lead formate (Pb(HCOO)₂), at the end of oxidative polymerization of oil paint films, was uncovered. The formation of formic acid in oils was proved by gas chromatography-mass spectrometry (GC-MS). Vapor experiments evidenced the susceptibility of Pb₃O₄ to react with volatile formic acid released during the autoxidation of oils comparably to the direct pigment-binder interactions in paint films. The investigation of the local environment of lead atoms in the paint film by ²⁰⁷Pb WURST-CPMG NMR spectroscopy showed that Pb(ii) atoms reacted with linseed oil preferentially to form highly crystalline Pb(HCOO)₂, while the local chemical environment of Pb(iv) atoms did not change. The results proved the co-existence of (i) highly crystalline Pb(HCOO)₂, (ii) a highly mobile amorphous phase corresponding to free carboxylic acids or a nascent lead soap phase and (iii) the remaining Pb₃O₄ in the polymeric/ionomeric network. Pb(HCOO)₂ is assumed to be an intermediate for the conversion of Pb₃O₄ to lead soaps and/or lead carbonates.

Full text of DOI:10.1039/d0dt00327a

View Article Online
View Journal
Dalton
Transactions
An international journal of inorganic chemistry
Accepted Manuscript
This article can be cited before page numbers have been issued, to do this please use: S. Švarcová, E.
Koí, P. Bezdika, S. Garrappa, L. Kobera, J. Plocek, J. Brus, M. Stastny and D. Hradil, Dalton Trans., 2020,
DOI: 10.1039/D0DT00327A.
Volume 46
Number 10
14 March 2017
Pages 3073-3412
This is an Accepted Manuscript, which has been through the
Royal Society of Chemistry peer review process and has been
accepted for publication.
Dalton
Transactions
An international journal of inorganic chemistry
Accepted Manuscripts are published online shortly after acceptance,
before technical editing, formatting and proof reading. Using this free
rsc.li/dalton
service, authors can make their results available to the community, in
citable form, before we publish the edited article. We will replace this
Accepted Manuscript with the edited and formatted Advance Article as
soon as it is available.
You can find more information about Accepted Manuscripts in the
Information for Authors.
Please note that technical editing may introduce minor changes to the
text and/or graphics, which may alter content. The journal’s standard
Terms & Conditions and the Ethical guidelines still apply. In no event
ISSN 0306-0012
COMMUNICATION
Douglas W. Stephen et al.
shall the Royal Society of Chemistry be held responsible for any errors
N-Heterocyclic carbene stabilized parent sulfenyl, selenenyl,
and tellurenyl cations (XH+, X = S, Se, Te)
or omissions in this Accepted Manuscript or any consequences arising
from the use of any information it contains.
rsc.li/dalton

Page 1 of 15
Dalton Transactions
View Article Online
DOI: 10.1039/D0DT00327A
Uncovering the Lead Formate Crystallization in Oil-Based Paintings
Silvie Švarcováa*, Eva Kočía, Petr Bezdičkaa, Silvia Garrappaa, Libor Koberab, Jiří Ploceka, Jiří Brusb, Martin Šťastnýa and David
Hradila,c
aInstitute of Inorganic Chemistry of the Czech Academy of Sciences, Husinec-Řež 1001, 250 68 Husinec-Řež, Czech Republic
bInstitute of Macromolecular Chemistry of the Czech Academy of Sciences, Heyrovského nám. 2, 162 06 Praha 6, Czech
Republic
cAcademy of Fine Arts in Prague, ALMA Laboratory, U Akademie 4, 170 22, Prague 7, Czech Republic
e-mails: svarcova@iic.cas.cz, koci@iic.cas.cz, bezdicka@iic.cas.cz, garrappa@iic.cas.cz, kobera@imc.cas.cz,
plocek@iic.cas.cz, brus@imc.cas.cz, stastny@iic.cas.cz, hradil@iic.cas.cz
*corresponding author: e-mail: svarcova@iic.cas.cz, phone number: +420 311 236 937
Keywords
Lead oxide, lead formate, oil paintings, pigment-binder interactions, XRPD, ssNMR
Abstract
Lead carboxylates are an extensive group of compounds studied for their promising industrial applications as well as for
their risky behavior when they are formed in oil paintings as corrosion products of lead-based pigments, leading to serious
deterioration of paintings. Although the processes leading to the formation of aggregates, protrusions or inclusions,
affecting undesirably the appearance of paintings, are assumed to be long-term, the initial stages of the neo-formed lead
carboxylates are detectable in the early state of paints’ drying. To uncover the chemical changes of lead pigments during
the drying of the oil paint films, model systems consisting of minium (Pb3O4) and four common drying oils were studied by
X-ray powder diffraction (XRPD), 13C and 207Pb solid state NMR (ssNMR) and Fourier-transformed infrared spectroscopy
(FTIR). For the first time, a degradation mechanism of Pb3O4 via the crystallization of lead formate (Pb(HCOO)2), in the end
of the oxidation stage of polymerization of oil paint film, was uncovered. The formation of formic acid in oils was proved by
gas chromatography-mass spectroscopy (GC-MS). The vapor experiments evidenced the susceptibility of Pb3O4 to react
with volatile formic acid released during autoxidation of oils comparably to the direct pigment-binder interactions in paint
films. The investigation of the local environment of lead atoms in the paint film by 207Pb WURST-CPMG NMR spectroscopy
showed that Pb(II) atoms reacted with linseed oil preferentially to form highly crystalline Pb(HCOO)2, while the local
chemical environment of Pb(IV) atoms did not change. The results proved the co-existence of (i) highly crystalline
Pb(HCOO)2, (ii) highly mobile amorphous phase corresponding to free carboxylic acids or nascent lead soap phase and (iii)
remaining Pb3O4 in the polymeric/ionomeric network. Pb(HCOO)2 is assumed to be an intermediate for conversion of Pb3O4
to lead soaps and/or lead carbonates.
Introduction
Historical paintings have been chemically changing since their creation. While some of these changes only reduce the
aesthetic value of the artwork, many of them cause serious damage. Since paint layers represent complicated
heterogeneous mixtures of organic and inorganic components, it is difficult to uncover the mechanisms of the degradation
processes. These undesirable phenomena, often poorly understood, are influenced by both internal and external factors
(e.g., susceptibility of components to environmental conditions). The attention is given especially to the degradation of
lead-based pigments, representing on one hand the widespread and extensively used colorants in the history of European
art, while on the other hand, being susceptible to numerous reactions with salts or other components in paint layers, they
lead to undesirable changes within the paintings.1–6
An example of such phenomena that seriously endangers mechanical cohesion of paint layers is the formation of metal
soaps – saponification. Metal carboxylates, being secondary products of reaction between metal ions (typically Pb or Zn
from commonly used inorganic pigments) and free fatty acids (FA, released from hydrolyzed triglycerides of oil-based
binders), were widely documented in numerous paintings from various historical periods.1

Dalton Transactions
Page 2 of 15
View Article Online
Resulting in blistering, protruding, spalling or color changes of oil paintings, saponification-related phenomena represent a
particularly topical problem studied in various scientific fields. The reports contributing to the description ofDmOeI:t1a0l.1039/D0DT00327A
carboxylates’ transport through paint layers,3,4 evaluating the composition and/or crystal structure of metal soaps,5
developing new analytical approaches for reliable detection and identification of metal soaps4 or constructing
computational models for prediction of processes taking part during polymerization of paint films6 all attempt to describe
this complicated degradation mechanism with the aim to find a suitable conservation, prevention and protection strategy
for precious artworks.
In a number of experimental studies, model systems were considerably simplified in order to (i) examine the role of
particular factors and components and/or (ii) to obtain experimental results in a reasonable amount of time since the actual
processes are supposed to be long-term.2,6 However, since the drying of oil paints is a complex collection of individual
reactions7 some unpredicted processes may be overlooked.
Within this study, we used natural drying oils instead of pure individual FA for mock-ups. The binary model paints were
prepared by mixing minium (Pb3O4) with linseed, walnut, poppy-seed and standoil representing the commonly used oil
binding media. One of the unexpected results was a newly crystallized lead formate, Pb(HCOO)2, identified unambiguously
by XRPD and confirmed also by 13C and 207Pb ssNMR.
In the past, various metal formates were observed as corrosion products on metal artifacts consisting of lead, copper or
their respective alloys, but also on glass or calcareous objects.8–11 In case of paintings, the only reported example is the
detection of zinc formate dihydrate (Zn(HCOO)2·2H2O).12 It was identified together with Zn(CH3COO)2·2H2O and ZnS as a
degradation product of zinc white (ZnO) on the surface of Dalí’s paintings Couple with Clouds in their Heads (1936).
Regardless of the source of the metal cation, formic acid emitted from wooden frames, storage boxes or showcases was
supposed to be the corrosion cause. However, here presented findings clearly demonstrate that drying oils alone represent
another source of formic acid for the crystallization of Pb(HCOO)2 during the formation of paint film. Lead formate has
never been reported for oil-based paints, as it does not form in overly simplified systems and is probably no longer present
in aged paintings or mock-ups. The role of this intermediate phase in the process is now going to be clarified.
Experimental section
Model experiments
Model paints The model paints were prepared as binary mixtures consisting of a lead-based pigment minium (Pb3O4; Sigma
Aldrich, p.a.) and drying oil in the weight ratio 1:3, respectively. We used the following drying oils: linseed oil (LO; Kremer
Pigmente), stand oil (SO; pre-polymerized linseed oil; Kremer Pigmente), walnut oil (WO; Kremer Pigmente) and poppy-
seed oil (PO; Kremer Pigmente). After thorough mixing, the mixtures were applied on a support and the freshly prepared
films were analyzed immediately and then monitored regularly during their hardening (up to 2 months) without any
additional treatment. For comparison, a paint prepared from the pre-reacted suspension of minium and linseed oil was
analyzed. The suspension was placed into a close vial and shaken constantly on a laboratory shaker for one month. After
sampling, the suspension was applied on a support as the freshly prepared paints and analyzed during drying. The samples
were stored under laboratory conditions (T = 23 ± 2 °C, relative humidity (RH) was 40 %). For the XRPD measurements in a
transmission mode, the paint films were prepared on the Mylar foil (polyester, 6 micrometers thick). For FTIR
measurements in the reflection mode, the thin paints were applied on a laboratory glass slides covered by an aluminium
foil in order to achieve the maximal reflectivity of the specimens and to avoid deformations of acquired spectra. The
starting pigment was characterized by XRPD including the Rietveld refinement (Fig. S1, S2 in ESI) and the distribution of the
particle size was determined by the laser diffraction analysis and the scanning electron microscopy (Fig. S3, S4 in ESI). The
experimental details of this characterization are given also in ESI.
Exposure to oil vapors A thin layer of minium powder was applied on a surface of a Si zero background sample holder (Si-
plate). The specimen was placed on a ceramic support above the appropriate oil into a closed glass vessel, avoiding any
contact between the powder and the oil similarly as reported by Ghiara et al.9 This setup was prepared separately for LO,
PO and WO. The samples were quickly analyzed by powder XRPD every 7 days.
Synthesis of lead formate
Lead(II) formate (Pb(HCOO)2) was synthesized by the reaction of basic lead(II) carbonate ((PbCO3)2·Pb(OH)2; Sigma-Aldrich)
with formic acid (HCOOH, Lachema a.s., p.a.). The basic lead carbonate was heated with an excess of formic acid (i.e., thirty-
times of stoichiometric amount) under reflux for one week. Then the reaction mixture was evaporated to one third of the

Page 3 of 15
Dalton Transactions
View Article Online
initial volume and left to precipitate at room temperature. Solid product was filtered out, washed with isopropylalcohol and
DOI: 10.1039/D0DT00327A
dried at 65 °C. Yield was 92,5 %. Phase purity was confirmed by X-ray powder diffraction.
X-ray powder diffraction (XRPD)
Model paints The model paints were monitored by XRPD according the following time schedule:
Time of hardening (days)
0 – 5
Interval of measurements
12 hours
8 – 51
7 days
Diffraction patterns of model paints were collected using a PANalytical X´Pert PRO diffractometer equipped with
a conventional X-ray tube (CuK 40 kV, 30 mA, line focus) in transmission mode. An elliptic focusing mirror, a divergence slit
0.5°, an anti-scatter slit 0.5° and a Soller slit of 0.02 rad were used in the primary beam. A fast linear position sensitive
detector PIXcel with an anti-scatter shield and a Soller slit of 0.02 rad were used in the diffracted beam. All patterns were
collected in the range of 1 to 40 deg. 2theta with the step of 0.013 deg and 200 sec/step producing a scan of about 47
minutes. The calculation of current concentration of Pb-phases in model systems based on XRPD-data is described in ESI.
Exposure to oil vapors Diffraction patterns of minium exposure to oil vapors were collected with the PANalytical X´Pert PRO
diffractometer equipped with a conventional X-ray tube (Cu K radiation, 40 kV, 30 mA) and a linear position sensitive
detector PIXcel with an anti-scatter shield. A programmable divergence slit set to a fixed value of 0.5 deg, incident anti-
scatter slit of 1 deg, Soller slit of 0.04 rad and mask of 15 mm were used in the primary beam. A programmable anti-scatter
slit set to fixed value of 0.5 deg., Soller slit of 0.04 rad and Ni beta-filter were used in the diffracted beam. Data were
collected in the range of 8 - 60 deg 2theta with the step of 0.0131 deg and 200s / step producing a scan of about 1 hour.
XRPD data were evaluated using HighScorePlus software package (4.8.0) and JCPDS PDF-4 database.13
Fourier transformed infrared spectroscopy (FTIR)
The model paint film consisting of minium and linseed oil was monitored by FTIR spectroscopy according the following time
schedule: every hour during the first day, then each 24 hours until the fifth day, and then once a week until the end of the
first month and finally after 2 months of drying. The measurements were carried out with a Bruker Alpha II spectrometer
equipped with an external reflection module which allows a fast, contactless and non-destructive FTIR analysis of large
samples on top or in front of the spectrometer. Spectra were acquired in the middle infrared region (from 400 to 4000 cm-
1), at a resolution of 4 cm-1 and by summing 32 scans. Alpha II was chosen in order to follow changes in the chemical
composition during the hardening of the film avoiding any sample preparation. Qualitative evaluation of data acquired was
performed with OPUS Spectroscopy Software.
Solid state NMR (ssNMR)
The ssNMR spectra were measured at 11.7 T using a Bruker AVANCE III HD 500 WB/US NMR spectrometer. The 3.2-mm and
4-mm cross-polarization magic angle spinning (CP/MAS) probes were used for 13C and 207Pb experiments at Larmor
frequencies of ν(13C) = 125.783 MHz and ν(207Pb) = 104.609 MHz, respectively. The 13C ssNMR spectra were collected at 15
kHz spinning speed and 207Pb WURST-CPMG (Wideband Uniform Rate Smooth Truncation excitation with the Carr-Purcell
Meiboom-Gill method) NMR at static conditions. The 13C NMR isotropic chemical shift was calibrated using α-glycine
(176.03 ppm; carbonyl signal) as external standards. The 13C MAS NMR spectra of were measured with recycle delay of 20 s
at the 3584 number of scans. The 13C CP/MAS NMR spectrum of Pb(HCOO)2 was measured with recycle delay of 5 s. The
number of scans was 256 at 1 ms spin lock. The 13C-1H PILGRIM/MAS NMR spectrum14 (Phase-Inverted Lee−Goldburg
Recoupling under MAS) was measured at 15 kHz spinning frequency with the number of scans 256 per increment and
recycle delay of 5 s. The nutation frequency of the B1(1H) field for on-resonance cross-polarization and LG (Lee−Goldburg)
irradiation in SEMA (Spin Exchange at the Magic Angle) recoupling sequence was ω1/2πꢀ=ꢀ75 kHz with resonance offset +53
kHz.
The 207Pb WURST-CPMG NMR experiments15 were carried out using 50 μs CT-selective WURST pulse16,17 using 500 kHz
sweep width, with 75 loops, the recycle delay was 7 s. The applied experimental parameters were adapted from literature
data.4 207Pb WURST-CPMG NMR experiment was combined with variable offset cumulative spectroscopy (VOCS)
technique.18 The final 207Pb WURST-CPMG NMR spectra are the sum of three sub-spectra with 178 kHz step at offsets from -
178 kHz to 178 kHz. Each sub-spectrum of model materials, Pb(HCOO)2 and Pb3O4, were collected at 512 scans. In case of
the prepared system (Pb3O4 + linseed oil) 8192 scan accumulation for each sub-spectrum were conducted. The 207Pb NMR
spectra were referenced to the external standard Pb(NO3)2 (δiso = -3473.6 ppm). High-power 1H decoupling (SPINAL64 -
small phase incremental alternation with 64 steps - for MAS experiments and CW - continuous wave - for static
experiments) was used to eliminate heteronuclear dipolar couplings in all measurements. The NMR experiments were

Dalton Transactions
Page 4 of 15
View Article Online
performed at a temperature of 303 K, and temperature calibration was performed to compensate for the frictional heating
of the samples.19 The all NMR spectra were processed and simulated using Top Spin 3.5 pl7 software packagDeO. I: 10.1039/D0DT00327A
207Pb ssNMR signals were attributed based on spectra of the reference materials (Pb3O4, Pb(HCOO)2; SI) and the obtained
NMR parameters are listed in Table S1. The two signals in the 207Pb ssNMR spectrum of minium were assigned and fitted
based on literature data.20,21 NMR parameters of the observed signal of Pb(HCOO)2 was fitted based on DFT/ZORA (density
functional theory with the zeroth order regular approximation) calculation22 of the structural model published by Lanza et
al.23 (CCCD deposit number 925259), computational details are listed in ESI (Table S1).
Analysis of oils and oil vapors
Determination of fatty acids in drying oils The fatty acid profiles of the drying oils used were analyzed by gas
chromatography with flame ionization detector (GC-FID) according to a certified procedure in the Metrological and Testing
Laboratory of University of Chemistry and Technology Prague (Testing laboratory 1316.2 accredited by the CAI according to
the EN ISO/IEC 17025:2005). The determined fatty acids are given in Table S2.
GC-MS analysis of oil vapors To detect low molecular polar compounds (i.e., water and/or methanol-derivatives) the
following procedure was tailored. 2 ml of each oils were placed into 4 ml vials with septum. Empty space of each vial was
filled with an inert atmosphere (Argon 4.6 purity). Vials were then placed into a programmable rotator-mixer (Starlab Inc.)
for a continuous mixing of 7 days. Then the samples of as formed gas over the oil were taken through the septum with gas-
tight Hamilton injection syringe and immediately analyzed by a simple headspace gas chromatography (GC) method
coupled with mass spectrometry (MS). Analyses were performed on gas chromatograph Agilent 6890N coupled with mass
spectrometer JEOL JMS-Q1000GC. The gas chromatograph was equipped with a single injection connected with column HP-
5 19091J-413 (5%-Phenyl-methylpolysiloxane, 30 m x 0.32 mm I.D. x 0.25 µm).24 Helium (purity 99.9999 %) was employed
as a carrier gas. The chromatographic conditions and instrument parameters are given in Table S3.
Results and discussion
Detection of lead formate
In the time-dependent XRPD patterns recorded on freshly prepared oil pints (Fig. 1), there is a gradual formation of two
new phases. While the low angle region of the XRPD patterns has background with a broad hump which suggests the
formation of lead ionomer contributing to an amorphous lead phase,3 in addition, a well crystalline phase identified
unambiguously as lead formate – Pb(HCOO)2, was formed as a secondary product in the model paints during their drying. In
order to elucidate the formation of this unexpected lead phase, additional experiments and measurements were carried
out.
Fig. 1 Time-dependent XRPD patterns of model paint consisting of minium and linseed (LO), poppy-seed (PO), walnut oil (WO) and standoil
(SO). The XRPD patterns between 0 hours (0H) and 5 days (5D) were collected every 12 hours and, further on, every week. The detected
phases: F – lead formate, Pb(HCOO)2, PDF 01-082-8113; M – minium – Pb3O4, PDF 00-041-1493; S – Pb-soap/ionomer.13

Page 5 of 15
Dalton Transactions
13
View Article Online
DOI: 10.1039/D0DT00327A
C ssNMR spectroscopy involving direct-polarization, cross polarization and PILGRIM experiment (Fig.2) was applied to
detect various organic-based phases in the system with linseed oil where lead formate was detected already after 12 hours
of drying and where its portion was the highest one within the studied model paints (Fig. 1). The signal assignments of 13
C
ssNMR spectra are based on literature data.25 Comparing fresh and polymerized linseed oil (without pigment), significant
decrease of signals corresponding to vinyl carbons from the unsaturated acyl chains (around 130 ppm) indicates terminated
oxidation and evidence of broadening of all signals confirms formation of a polymeric network as expected in case of drying
oils.26 Furthermore, a new carbonyl signal at 177.1 ppm appearing in the 13C MAS NMR spectrum (Fig. 2b) suggests the
presence of two distinct types of carbonyls with different local environments and mobilities in the polymerized oil network.
The relative narrower signal at 177.1 ppm reflects a fraction with enhanced mobility and is attributed to carboxylic groups
either corresponding to free carboxylic acids (released from triacylglycerides during polymerization of oil) and/or to
carboxylic functionalities attached to the polymeric network, while the fraction reflected by a broader signal at ca. 175 ppm
corresponds to ester bonds in the polymeric network. On the other hand, in the paint film with minium, the missing signal
at 177.1 ppm suggests formation of lead ionomer where lead ions are bound to carboxylic groups along the polymer
backbone chain.3
Additionally, experiments using the 13C CP/MAS NMR technique revealed two other carbonyl signals, which – based on the
analysis of the 1H-13C PILGRIM/MAS NMR spectrum (Fig. 2d) – are attributed to Pb(HCOO)2. Specifically, this follows from
the determined dipolar splitting which reaches to ca. 11.5 kHz. Such a high value corresponds to the one-bond 1H-13C
dipolar coupling constant of a rigid segment in which the distance between proton and carbon is ca. 1 Å; only formyl groups
containing a carbonyl unit directly bonded to hydrogen atom (H-C=O) can be detected. Other carboxyl groups are non-
protonated therefore provide significantly smaller dipolar splitting in 2D 1H-13C PILGRIM/MAS NMR spectra. Consequently,
detection of a large 1H-13C dipolar splitting for carbonyl units identifies the presence of formyl segments (H-C=O). This
assignment is further supported by the 13C NMR chemical shift recorded for the reference sample – lead formate (175.1 and
171.4 ppm) and by 207Pb NMR chemical shift (-2500 ppm) as discussed (vide infra). The two well-separated dipolar doublets
resonating at 175.1 and 171.3 ppm clearly show the presence of two inequivalent formyl groups. Both these groups are
quite rigid with negligible internal dynamics. The corresponding narrow 13C resonance and the observed large 1H-13C dipolar
splitting confirm a regular arrangement of both these units in one crystalline domain. The two-component character of the
investigated system is further reflected by the distinctly different dipolar profiles determined for aliphatic CH2 segments.
The outer doublets with the splitting of ca. 6-7 kHz reflect relatively rigid fraction, whereas the central signal with the
linewidth of ca. 3-4 kHz indicate amorphous phase with partly released segmental motions. All these findings indicate co-
existence of several phases and domains including (i) a highly rigid crystalline fraction of lead formate, (ii) immobilized
amorphous polymeric/ionomeric network with dissolved Pb3O4 and (iii) relatively mobile organic fraction, e.g., free
carboxylic acids or non-crystalline soaps.
The local environment of lead atoms in the LO-based paint was investigated by 207Pb WURST-CPMG NMR spectroscopy
(Fig. 3), which was combined with variable offset cumulative spectroscopy technique. This combination is the most suitable
approach dealing with the large chemical shift anisotropy and extremely wide chemical shift range of the 207Pb nuclei. The
obtained NMR parameters for the model paint as well as the reference Pb3O4 and Pb(HCOO)2 are listed in Table 1. Signal
processing and attribution is given in ESI. Apparently, lead(II) atoms (arranged in the square pyramid-coordinated PbO4) in
Pb3O4 react with LO preferentially during reaction to form highly crystalline Pb(HCOO)2, while the local chemical
environment of lead(IV) atoms does not change during the corrosion process, indicating that there is no redox reaction
involving Pb3O4. The 207Pb NMR spectra of Pb(HCOO)2 show chemical-shift tensors with smaller span (Ω = 720) indicating
the holodirected coordination of lead atoms with 8 neighboring oxygen atoms, which was reported also for long-chain lead
carboxylates (soaps) or PbCO3.2,4,22
The lead (IV) atoms probably remained in the form of nanocrystalline PbO2 species dispersed in the polymeric matrix. This
assumption could explain both, the sharp peak attributed to PbO2 in the ssNMR spectra (Fig. 3, Table 1), indicating its
crystalline nature, and on the other hand, the absent signal in XRPD patterns, indicating that the coherent domain (≈
crystallite size) remains under the detection limit of XRPD. The partial decomposition of minium accompanied by the
formation of new Pb-phases (i.e., Pb-formate, Pb-soap/ionomer or PbO2) did not lead to any observable color changes (we
performed only visual inspection by naked eye) of the studied paint which remained orange-red.

Dalton Transactions
Page 6 of 15
View Article Online
DOI: 10.1039/D0DT00327A
Fig. 2 The 13C MAS NMR spectra of a) fresh linseed oil, b) dried linseed oil, c) prepared Pb3O4+linseed oil system and d) experimental 2D 1H-
13C PILGRIM/MAS NMR and 13C CP/MAS NMR spectra of prepared Pb3O4+linseed oil system. The 13C NMR isotropic chemical shifts were
determined with the experimental error of ca. ±0.1 ppm; and the 1H-13C dipolar couplings with ca. ±1 kHz. These uncertainties are
negligible and cannot affect interpretation of the recorded spectra.

Page 7 of 15
Dalton Transactions
View Article Online
DOI: 10.1039/D0DT00327A
Fig. 3 Experimental 207Pb WURST-CPMG NMR spectra (black solid line), simulations of the individual lead atoms (blue dashed lines) and
their sums (red solid lines) of investigated samples.
Table 1. Experimentally measured 207Pb chemical shift tensor parameters of investigated samples
Sample
Site
δiso (ppm) [a]
Ω (ppm) [b]
κ
mol % (± 10 %)
Pb(HCOO)2
Pb3O4
Pb2+
PbO
-2567±20
804±20
720±20
0.75±0.1
0.72±0.1
100
75
3088±20
PbO2
PbO2
-1091±20
-1091±20
170±20
170±20
-0.53±0.1
-0.53±0.1
25
20
Pb3O4+Linseed oil
Pb(HCOO)2
-2567±20
720±20
0.75±0.1
80
[a] The data are presented using the Maryland notation, where δiso = (δ11+δ22+δ33)/3. [b] The span is calculated by Ω ≈ δ11-δ33 (Ω ≥ 0) and the
skew is given by κ = 3(δ22-δiso)/Ω; (-1 ≤ κ ≤ +1).27
The effect of oxygen
Monitoring of the processes in the model paints by XRPD was complemented with time-dependent FTIR measurements
(Fig. 4). Comparing results from both methods, it is obvious that Pb(HCOO)2 starts to crystallize in the end of autoxidation
reactions. These are characterized by changes in the FTIR spectrum related to isomerization and, finally, to saturation (i.e.,
decrease of the signal corresponding to C=C stretch of unconjugated cis-double bonds at 3010 cm-1 and 724 cm-1 and
simultaneous arising of signals corresponding to conjugated double bonds at 1630 cm-1 and 988 cm-1). Once the oxidation
processes are completed, a signal of C=O stretch in lead carboxylate (1533 cm-1) occurs. Its broad character is in agreement
with XRPD and ssNMR, where a significant signal contribution of lead ionomer prevents the detection of Pb(HCOO)2 due to
signal overlap3 (see Fig. S5 in ESI). On the other hand, the crystalline Pb(HCOO)2 is detected unambiguously by XRPD at the
corresponding time of drying (i.e., 12-24 hours for Pb3O4 in LO).

Dalton Transactions
Page 8 of 15
View Article Online
DOI: 10.1039/D0DT00327A
Fig. 4 Time-dependent FTIR spectra of model paint consisting of minium and LO. Highlighted bands correspond to oxidation and formation
of carboxylates.
The necessity of atmospheric oxygen for formation of Pb(HCOO)2 is also illustrated with XRPD patterns recorded on the
paints prepared from (i) freshly mixed LO and minium and (ii) pre-reacted mixture of minium with LO (shaken in a closed
vial for 1 month), Fig. 5. In both cases, the amount of Pb(HCOO)2 increases with the time of drying. The minute amount of
Pb(HCOO)2 in the pre-reacted mixture (see XRPD pattern in Fig. 5, day 0) can be attributed to the surface oxidation of oil
suspension during shaking due to the presence of air in the vial. XRPD patterns reflect also the different character of lead
soap phases in both systems. In the freshly prepared paint, it is apparent that a rather Pb-ionomer phase was formed, while
in the pre-reacted mixture, structural ordering with diffraction maxima in the low angles region indicate partial
crystallization of lead soaps. However, detailed description of this phenomenon requires additional experimental work and
is beyond the scope of this paper.

Page 9 of 15
Dalton Transactions
View Article Online
DOI: 10.1039/D0DT00327A
Fig. 5 Time-dependent XRPD patterns of model paints prepared of pre-reacted mixture of minium and linseed oil, which was in form of
suspension shaken for 1 month (a) and paints of freshly prepared mixture of minium and linseed oil (b). Both paints were measured by
XRPD immediately after application (0D) up to two weeks during drying without any treatment or sampling. The detected crystalline
phases: lead formate – Pb(HCOO)2 (F), minium – Pb3O4 (M).
Further evidence of the oxygen effect is clearly demonstrated by the experimental exposition of minium powder to oil
vapors (LO, PO and WO) without any contact of the pigment and the oil (Fig. 6). Formic acid evaporates from drying oils
during their autoxidation and, reaching the minium particles, reacts to form lead formate (documented by both the
increasing intensities of diffraction lines of lead formate as well as the decreasing intensities of minium). Interestingly, the
time of first detectable formation of lead formate differs for the vapors of the different oils. This corresponds to the same
trend found in model oil paints (Fig.1), where in the LO paint, the lead formate crystallizes promptly, while in the other oils
the crystallization is slower. This indicates the effect of oil composition, which differ in the character and abundance of
(poly)unsaturated fatty acids, especially in the content of α-linolenic acid with 3 isolated double bonds in the chain
(Table S2).

Dalton Transactions
Page 10 of 15
View Article Online
DOI: 10.1039/D0DT00327A
Fig. 6 XRPD patterns collected on minium specimens exposed to vapors of linseed (LO), walnut (WO) and poppy-seed oil (PO) for 21 days
(a). The detected crystalline phases: lead formate – Pb(HCOO)2 (F), minium – Pb3O4 (M). The indices denote the diffraction lines used for
semi-quantitative analysis. Profiles of the relative current concentration of Pb(HCOO)2 (b) and Pb3O4 (c) in specimens exposed to oil vapors
represented by relative diffraction intensities of Pb-phases detected in the samples. The trend lines serve as a guide for eyes only.
Formic acid and the effect of unsaturation in oils
Autoxidation of drying oils is the initial step of hardening of oil films.26 This involves numerous and often simultaneous
reactions, representing a somewhat complicated process reflecting the complex composition of drying oils (i.e., glyceryl
esters of various (poly)unsaturated and saturated fatty acids, antioxidants, etc.). In general, the reactions to be considered
are the formation of unsaturated ester hydroperoxides, their decomposition and subsequent cross-linking.26 The
decomposition can occur via several routes, namely (i) radical recombination, (ii) radical addition and (iii) β-scission and
oxidation.26,28 β-scission and secondary oxidation generate volatile decomposition products such as aldehydes, ketones,
hydrocarbons, alcohols, carboxylic acids and other products with low molecular weight. Among the volatile organic
compounds (VOCs) originating from oils, various aldehydes, typically those with 5-10 carbons, are the most abundant, while
carboxylic acids belong to minor VOCs.7,26,28–30
Reports on the presence of formic acids in oils containing polyunsaturated fatty acids are rather sporadic, which can reflect
the fact that the analyses of edible oils focus on the detection of potentially toxic products of oxidation30 or on compounds
affecting the aroma, flavor or rancid odor.28,31,32 In addition, the choice of the method or the experimental setup can cause
that some polar low weight molecules may be omitted in the analysis of the complicated mixtures of oils or their oxidation
products. For example, the type of the chromatographic column as well as the temperature gradient strongly affects the
possibility to detect methanol derivatives.33

Page 11 of 15
Dalton Transactions
View Article Online
DOI: 10.1039/D0DT00327A
In order to detect the formic acid in vapors of the studied oils, sample preparation and GC/MS procedure were tailored for
the qualitative analysis of methanol-related compounds and/or water in oil vapors, thus other VOCs were not detected.34
Formic acid, together with formaldehyde, were detected in vapors of all investigated oils except poppy-seed oil, where no
substance of interest was found (Fig. S6, S7). It is noteworthy that the formic acid predominates in the vapors of linseed oil
from the bottle which was opened ca. 1 year ago, in comparison to linseed oil from a newly opened bottle (where
formaldehyde was also detected), illustrating both the necessity of oxygen action and the promptness of linolenic acid to be
oxidized leading to the production of formic acid.
Formic acid as a by-product of oxidation of unsaturated aldehydes, which were produced by decomposition of primary
unsaturated ester hydroperoxides by autoxidation of oil triacylglycerides, is reported, e.g., by E.N. Frankel, who studied the
volatile lipid oxidation products related to edible oils.28 In addition, lower aldehydes formed from a cascade of oxidation
reactions can be further oxidized, thus resulting in the corresponding carboxylic acid, as reported by H. Wexler.26 This
reaction mechanism is illustrated by the scheme in Fig. S8 in ESI. In addition, Wexler described a pathway based on the
decomposition of intermediate ozonides during the polymerization of drying oils. Although unassigned, a signal was also
reported at 8.27 ppm in the 1H NMR spectra of linseed oil after 3 days of oxidation30 attributable to C=O bond in formic
acid.35
Our results indicate (Fig. 1, 5, 6) that the rate of oxidation reactions resulting in the final formation of lead formate is
different for different oils. Comparing the relative integral intensities of the most intense XRPD line of Pb(HCOO)2
positioned at 16.9° 2θ Cu Kα, it is obvious that LO is prone to form Pb(HCOO)2 more readily than other investigated oils in
mixtures with Pb3O4 (Fig. 7).
Fig. 7 Comparison of the profiles of the relative current concentration of Pb(HCOO)2 (a), Pb3O4 (b) and Pb-soap phase (c) in model paints
represented by relative diffraction intensities of Pb-phases in the mixtures. The trend lines serve as a guide for eyes only.
LO consists predominantly of α-linolenic acid (C18:3n3), while the main fatty acid in both WO and PO is the linoleic one
(C18:2n6c) and pre-polymerized SO contains mainly oleic acid (C18:1n9c) (Table S2). In the traditional oxidation mechanism,

Dalton Transactions
Page 12 of 15
View Article Online
DOI: 10.1039/D0DT00327A
the rate-limiting step in the reaction is abstraction of the hydrogen atom occurring at the bis-allylic positions (CH=CH-CH2-
CH=CH) of polyunsaturated acids (PUFA), implying that the susceptibility of PUFA to oxidation depends on the availability of
the bis-allylic hydrogens.36 In other words, the oxidizability of PUFA is proportional to the number of bis-allylic positions in
molecules and their position.26,30,37,38 Therefore, three isolated double bonds as well as their position near the methyl end
in α-linolenic acid affect both the rapid oxidation29 and the formation of formic acid via decomposition of secondary
oxidation products of the oil, resulting in rapid crystallization of Pb(HCOO)2 in the LO-based-paint. Guillén and Ruíz, who
studied the oxidation of unsaturated oils by proton NMR, concluded that the oxidation and polymerization of the linseed oil
was quick due to both the rapid appearance of hydroperoxides and their rapid subsequent degradation.30 The course of
Pb(HCOO)2 crystallization in WO and PO paints shows a similar trend, reflecting the close composition of UFAs in both oils.
The increased amount of α-linolenic acid in WO is probably responsible for the earlier formation of Pb(HCOO)2, however,
the antagonistic effect of a relatively higher antioxidant activity of the oil36,39 probably reduces the amount of the evolved
formic acid.
Lead formate as an intermediate
Raychaudhuri et al.40 identified lead formates together with lead carbonates as corrosion products of lead under action of
formic acid. Thermodynamic calculations indicated that lead carbonates are the final products of the conversion of lead
formates depending on the variations of partial pressure of CO2 and formic acid in the environment. It implies the probable
role of lead formate as an intermediate of the binder-induced conversion of minium to lead carbonate in here-presented
experiments. In the analysis of micro-samples from actual paintings, it is not easy to differentiate the primarily applied lead
carbonates (i.e., lead white consisting of PbCO3 and 2PbCO3·Pb(OH)2) from the secondary ones. In some cases, particle size
and morphology may be indicative – lead carbonates found in the core of neo-formed aggregates were explained as
products of mineralization of lead soaps.5,6 On the other hand, if the grains are randomly distributed in the paint layer, no
definitive answer can be given.
Changes in relative concentrations of Pb3O4 and neo-formed lead phases in time are illustrated by plotting the relative
integral intensities of their most intense diffraction lines or, in case of Pb-soaps, humps (Fig. 7). Apparently, the most
pronounced consumption of Pb3O4 occurs in the PO paint. In the PO and the WO paint as well, the amount of Pb(HCOO)2
decreases in time, indicating its partial transformation probably to an ionomer and/or nascent non-crystalline lead soap
phase, which is indicated by the gradually increasing hump in the 1-10°2θ Cu Kα interval in the XRPD patterns (Fig. 1). While
the relative amount of soaps in model paints is proportional to the content of SFA in the studied non-polymerized oils
(Table S2), the variations of the Pb(HCOO)2 content indicate its conversion to lead soaps in PO and WO systems in the
investigated time horizon since no other lead-based phase was identified in the model paints.
The assumption that Pb(HCOO)2 plays a role as an intermediate either for lead carbonates or lead soaps could explain its
absence in paint layers obtained from artworks. However, it is necessary to admit that such substance can be easily omitted
in the routine analyses of complicated paint mixtures. Therefore, an extended research of lead formate crystallization in
paint systems should (i) shed light on its role and fate in oil paints in the long-term scale, (ii) examine the effect of
environmental factors on its stability and (iii) evaluate how other lead-based pigments are prone to Pb(HCOO)2
crystallization.
Conclusion
Lead formate was identified unambiguously by powder XRPD as a secondary product of the interaction between minium
pigment and four different drying oils during the curing of the model paints.
The 13C CP/MAS NMR and the 1H-13C PILGRIM/MAS NMR techniques revealed the co-existence of crystalline lead formate
within the immobilized amorphous polymeric/ionomeric network with dissolved Pb3O4 and relatively mobile organic
fraction.
207Pb WURST-CPMG NMR spectroscopy combined with variable offset cumulative spectroscopy technique was applied to
investigate the local environment of lead atoms in the LO-based paint, providing the NMR parameters for Pb3O4 and
Pb(HCOO)2 and showing that lead(II) atoms in Pb3O4 react with LO preferentially during reaction to form highly crystalline
Pb(HCOO)2, while lead(IV), having the unchanged chemical environment, is not reduced to Pb(II).
The evaporation experiments and the GC/MS procedure of oil vapors tailored for the detection of polar low molecular
volatile compounds proved the emission of formic acid from the drying oils during their autoxidation.
The different rate of the formation of lead formate in the drying oils depends on the number and position of double bond
in the fatty acid chains, i.e., on the availability of bis-allylic hydrogen for the hydrogen abstraction representing the rate-
limiting step of the oxidation of oils.

Page 13 of 15
Dalton Transactions
View Article Online
DOI: 10.1039/D0DT00327A
Lead formate is assumed to be an intermediate product contributing either to the formation of lead carbonates and/or to
lead soaps. It is important to realize that the rapid formation of lead formate during the initial stages of paint drying might
be a source for further degradation of paint layers induced by a conservation treatment. Thus an incorporation of
monitoring of the chemical changes during the treatment would be beneficial to control the process.
Acknowledgements
The study was supported by the Czech Science Foundation, project no. 17-15621S and 19-05259S. The authors thank Prof.
Karel Dušek and Dr. Miroslava Dušková-Smrčková for their support and advice.
References
1
F. Casadio, K. Keune, P. Noble, A. van Loon, E. Hendriks, S. A. Centeno and G. Osmond, Eds., Metal Soaps in Art,
Springer, 2019.
2
3
4
E. Kočí, J. Rohlíček, L. Kobera, J. Plocek, S. Švarcová and P. Bezdička, Dalt. Trans., 2019, 48, 12531–12540.
J. J. Hermans, K. Keune, A. Van Loon, R. W. Corkery and P. D. Iedema, RSC Adv., 2016, 6, 93363–93369.
J. Catalano, A. Murphy, Y. Yao, G. P. A. Yap, N. Zumbulyadis, A. Centeno and C. Dybowski, Dalt. Trans., 2015, 44,
2340–2347.
5
6
7
8
9
C. L. Higgitt, M. Spring and D. R. Saunders, Natl. Gall. Tech. Bull., 2003, 24, 75–95.
K. Keune, A. Van Loon and J. J. Boon, Microsc. Microanal., 2011, 17, 696–701.
L. H. Oakley, F. Casadio, K. R. Shull and L. J. Broadbelt, Angew. Chemie - Int. Ed., 2018, 57, 7413–7417.
L. Nodari, L. Tresin, A. Benedetti, M. Katia and P. Tomasin, J. Cult. Herit., 2019, 35, 288–296.
G. Ghiara, S. Campodonico, P. Piccardo, C. Martini, P. Storme and M. Maddalena, J. Raman Spectrosc., 2014, 45,
1093–1102.
10
11
12
L. T. Gibson and C. M. Watt, Corros. Sci., 2010, 52, 172–178.
A. Fischer, G. Eggert, R. Dinnebier, T. Runčevski and A. Fischer, Stud. Conserv., 2017, 63, 342–355.
K. Keune and G. Boevé-Jones, in Issues in Contemporary Oil Paint, eds. Jk. J. van den Berg, A. Burnstock, M. de
Keijzer, J. Krueger, T. Learner, A. de Tagle and G. Heydenreich, 2014, pp. 283–294.
13
14
15
16
17
18
JCPDS PDF-4 database, International Centre for Diffraction Data, Newtown Square, PA, U.S.A., 2020.
M. Hong, X. Yao, K. Jakes and D. Huster, J. Phys. Chem. B, 2002, 106, 7355–7364.
A. W. Macgregor, L. A. O’Dell and R. W. Schurko, J. Magn. Reson., 2011, 208, 103–113.
E. Kupce and R. Freeman, J. Magn. Reson. Ser. A, 1995, 115, 273–276.
R. Bhattacharyya and L. Frydman, J. Chem. Phys., 2007, 127, 2007.
D. Massiot, I. Farnan, N. Gautier, D. Trumeau, A. Trokiner and J. Pierre, Solid State Nucl. Magn. Reson., 1995, 4,
241–248.
19
20
J. Brus, Solid State Nucl. Magn. Reson., 2000, 16, 151–160.
F. Fayon, I. Farnan, C. Bessada, J. Coutures, D. Massiot, J. P. Coutures, A. V. D. R. Scientifique, R. V October, V. Re,
M. Recei and V. May, J. Am. Chem. Soc., 1997, 207, 6837–6843.
21
S. P. Gabuda, S. G. Kozlova, V. V. Terskikh, C. Dybowski, D. L. Perry and G. Neue, Solid State Nucl. Magn. Reson.,
1999, 15, 103–107.
22
23
24
25
26
F. Alkan and C. Dybowski, J. Chem. Phys. Chem. A, 2016, 120, 161–168.
A. Lanza, A. Gambirasi and M. Favaro, Acta Crystallogr. Sect. C, 2013, C69, 41–43.
A. Bartošová and T. Štefko, Reasearch Pap. Fac. Mater. Sci. Technol. Trnava, 2017, 25, 73–81.
M. A. Brescia and A. Sacco, in Modern Magnetic Resonance, Springer, 2006, pp. 1615–1621.
H. Wexler, Chem. Rev., 1964, 64, 591–611.

Dalton Transactions
Page 14 of 15
View Article Online
DOI: 10.1039/D0DT00327A
27
28
29
30
31
32
J. Mason, Solid State Nucl. Magn. Reson., 1993, 2, 285–288.
E. N. Frankel, Prog. Lipid res., 1982, 22, 1–33.
W. Grosch, in Autoxidation of unsaturated lipids, ed. H. W.-S. Chan, Academic Press, 1987.
D. M. Guillén and A. Ruiz, Eur. J. Lipid Sci. Technol., 2005, 107, 36–47.
S. Krist, G. Stuebiger, S. Bail and H. Unterweger, Eur. J. Lipid Sci. Technol., 2006, 108, 48–60.
H. W. Gardner, in Xenobiotics in Foods and Feeds, eds. J. W. Finley and D. E. Schwass, American Chemical Society,
Washington D.C., 1983, pp. 63–83.
33
34
35
36
37
38
39
40
S. K. S. Patel, P. Mardina, D. Kim, S. Kim, V. C. Kalia, I. Kim and J. Lee, Bioresour. Technol., 2016, 218, 202–208.
A. López-Islas, M. Colín-García and A. Negrón-Mendoza, Int. J. Astrobiol., 2019, 18, 420–425.
O. Steinhof, É. J. Kibrik and H. Hasse, Magn. Reson. Chem., 2014, 52, 138–162.
K. Miyashita, Eur. J. Lipid Sci. Technol., 2014, 116, 1268–1279.
R. A. Hancock, N. J. Leeves and P. F. Nicks, Prog. Oraganic Coatings, 1989, 17, 321–336.
J. P. Cosgrove, D. F. Church and W. A. Pryor, Lipids, 1987, 22, 299–304.
P. Gornas, A. Siger, K. Juhn, L. Gunars, E. Sne and D. Seglina, Eur. J. Lipid Sci. Technol., 2014, 116, 563–570.
M. R. Raychaudhuri and P. Brimblecombe, Stud. Conserv., 2000, 45, 226–232.

Page 15 of 15
Dalton Transactions
View Article Online
DOI: 10.1039/D0DT00327A
Lead formate was found in the oil paint system, crystallizing in the early stage of polymerization of oil paint and playing role
as an intermediate in the degradation of lead pigment.