Received: 21 April 2020 Revised: 22 May 2020 Accepted: 25 May 2020

DOI: 10.1002/mdp2.189

REPORT

Application of the Guggenheim, Anderson, de Boer

equation to study the impact of sealing treatments on pine
wood sorption characteristics

Chiara Bertolin1 | Lavinia de Ferri1 | Marcin Strojecki2

1
Department of Mechanical and Industrial
Engineering, Norwegian University of
Abstract
Science and Technology, Trondheim, Surface treatments can modify the materials behavior toward water sorption
Norway phenomena. The Guggenheim–Andersen–de Boer sorption equation, already
2
Polish Academy of Sciences, Jerzy Haber
used in the past to study various wood species, is here proposed to investigate
Institute of Catalysis and Surface
Chemistry, Kraków, Poland the sorption behavior of pine wood samples treated with different sealing
material. Particularly, the paper aims to explore the modification of the
Correspondence
Chiara Bertolin, Department of
k parameter with the applied coating. Measurements of sorption–desorption
Mechanical and Industrial Engineering, branches permitted to clearly distinguish the behavior of treated samples, the
Norwegian University of Science and relative humidity (RH) conditions over which they are more dimensionally sta-
Technology, Trondheim, Gløshaugen,
Norway.
ble, as well as their tendency to swell and shrink. The proposed study permit-
Email: [email protected] ted to evaluate the usefulness of the k parameter as an indicator of the
presence of chemical treatments on wood at high RH values. Particularly, the
Funding information
Norges Forskningsråd, Grant/Award k value has been found to change as a function of chemicals surface treatments
Number: 274749; ICSC PAS; Norwegian limiting in different ways the access and expulsion of water moisture.
Research Council, Grant/Award Number:
274749 Highlights
• Surface sealed pine wood samples were analyzed by vacuum microbalance.
• The Guggenheim–Andersen–de Boer equation has been applied to water
sorption–desorption data.
• The equilibrium moisture content and its first derivative explained samples
swelling/shrinking tendency.
• Assessment of the usefulness of k as indicator of the wood treatment at high
relative humidity values.
• Assessment of k modifications in relation to the type of sealing treatment.

KEYWORDS
GAB equation, pine wood, sealing treatments, microbalance, water sorption, high RH

This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided
the original work is properly cited.
© 2020 The Authors. Material Design & Processing Communications published by John Wiley & Sons Ltd

Mat Design Process Comm. 2020;e189. wileyonlinelibrary.com/journal/mdp2 1 of 12

https://doi.org/10.1002/mdp2.189
2 of 12 BERTOLIN ET AL.

1 | INTRODUCTION

Modification in ambient temperature and relative humidity (RH) affects the equilibrium moisture content (EMC) of
hygroscopic materials as wood. The EMC can be considered as material parameter describing the binding potential of
water to this adsorbent material. An excess or a pauperization of water vapor adsorbed or desorbed by the material
modifies its EMC as well as its swelling or shrinking rate and consequently its dimensions. On the long-term wood kept
at high, low conditions of RH or exposed to RH fluctuations will be subjected to a high level of risk of decay. Therefore,
an accurate estimation of wood response to moisture variation can permit to determine RH conditions with lower risk
of environmentally induced damage. Similarly, a deep understanding of the wood–water relationship after surface treat-
ments can further help in assessing the service life of conservation treatments or to plan additional maintenance for
materials treated with short-lasting treatments under the same or similar environmental RH conditions.
Many different chemicals are currently used to preserve wood surfaces as well as to modify its properties, generally
depending on the final application foreseen for the material. In particular, the chemical modification of wood has been
mainly attempted in order to improve its dimensional stability resulting from changes in moisture content. Among
many treatment methods, the most popular are heating,1,2 impregnation with monomers polymerizing inside the mate-
rial, or functionalization.3 It is generally possible to distinguish improved wood or densified wood, modified by
densification/resin impregnation, from materials submitted to cell-modification processes, occurring by reactive
chemicals.3 Only in this case it is possible to define the process as a chemical modification because chemical reactions
occur between reactive parts in the wood and a reagent.
Different approaches have been proposed depending on the property to be modified: for water repellency, a reduc-
tion of the hydrophilic nature of wood is usually carried out by bonding on hydrophobic groups. The improvement of
dimensional stability can be achieved by bulking cell walls with bonded chemicals, or using cross-linking polymers lim-
iting cell wall expansion, or bounding functional group that can increase hydrophobicity. Another often desirable effect
is flame retardance achievable by bonding retardants or fame suppressants functional groups to the cell walls. Similarly,
ultraviolet (UV) resistance can be obtained by bonding UV blockers or absorbers to lignin.4
As stated above, most of the applied treatments are commonly aimed to limit variations in the material moisture
content, and this can be done acting on the mechanism of water absorption. Moisture sorption process in wood has
been described by the Hailwood and Horrobin equation that in turn has been demonstrated to be equivalent to the
Guggenheim–Andersen–de Boer (GAB) sorption equation.5,6
The latter equation can describe the full shape of type II isotherm5,7 covering an activity range (a0) much wider than
that of models such as the Brunauer–Emmett–Teller (BET) equation.7 This is due to the fact that the theoretical basis
for the GAB sorption isotherm fits with the vapor adsorption in porous bodies as wood because of the bonds forming
between water molecules and wood hydroxyl groups.8 When additional water layer is deposited, the interactions with
the sorbent surface are assumed to be weaker until reaching the energy level of liquid water.9–12 In addition to the BET
equation, the GAB one introduces a further parameter indicated as k, related to the difference of the standard chemical
potential between molecules creating the additional adsorbed layers and those in the liquid state.7
This work focuses on the application of GAB equation as an alternative to other water sorption-based models avail-
able in literature13 (e.g., the Langmuir, the BET, the Oswin, and the Halsey models). In particular, the model has been
applied to interpret the sorption data obtained from pine wood samples, which tangential–radial (TR) surfaces were
coated with different sealing materials. The RH-dependence differences in the moisture content of the tested samples
(one standard—without any treatment—and five treated) has been explained, and the GAB water sorption isotherms
have been used to
1 investigate how the GAB parameters can change when fitting water vapor sorption data for the six tested pine
samples,
2 understand the moisture-induced response of coated wood in general (Sections 3.1 and 3.2),
3 study the behavior of the GAB k constant obtained for all tested pine samples as it has never been studied in detail in
literature.

2 | MATERIALS AND METHODS

During this study, the adsorption branches of the water vapor isotherms were measured for six small samples cut from
pine wood slices. The original pine log—constituting the material for the experiments—was grew, cut, and naturally air
BERTOLIN ET AL. 3 of 12

dried in the Trøndelag area over a 3-year period. Then from the log, radially cut pine slices (dimensions of
20 × 10 × 2 cm) were prepared, and five of them were treated on the TR surfaces with the following chemicals:

• Paraloid B72 40% (w/v) (P40) deposited using a brush,

• Cellulose sand seal (CS) spray (Chestnut),
• Paraloid B72 40% (w/v) deposited over a dried cellulose sand seal layer (CS + P40),
• Renaissance microcrystalline wax (R) deposited with a spatula,
• Epoxy resin (E), containing bisphenol A diglycidyl ether as epoxy component (the hardener is not specified) depos-
ited with a spatula.

As reported in other works14 currently in press, chemicals were tested in order to select the most suitable treatment
for further experiments involving the stress development induced by moisture gradients appearing in coated wood
under high RH changes (50%) and further investigations of its damage induced by such stress dynamics.
The last pine sample was instead kept clean, that is, not treated with chemicals on the surface to maintain a stan-
dard reference during the tests. Measurements of the water vapor sorption on the six samples were carried out following
the dynamic gravimetric method at room conditions (i.e., 24
C) with the aid of a vacuum microbalance (CI Electronics
Ltd.) through a continuous measuring procedure. During the process, water vapor was introduced (or removed) at a
slow constant rate of 0.01 Torr/min to keep quasi-equilibrium conditions while the changes in weight were detected
within the full range of water vapor relative pressures. Samples were considered to have reached “equilibrium” only
when the weight changes were less than 0.05% over 40 min. In the application of the gravimetric technique, special
attention has been devoted to reduce at minimum the sources of potential errors, mainly caused by unprecise measure-
ments of pressure (i.e., higher than ±10 Pa), lack of determination of outgassed weight of the adsorbent either before or
after the adsorption measurements, or low attention in controlling measurement of the adsorbent temperature. For
details on the experimental procedure and time using the microbalance with wooden samples, see Bertolin et al.15 From
the experiments, the Mads (i.e., adsorption mass) values were calculated based on the initial weight of the outgassed
samples. Then these outcomes were further elaborated to calculate the volume of adsorbed water (i.e., Vads [cm3/g of
wooddry]) in relation to the water activity or RH (%) using the GAB equation (Equations 1 and 2).

RH
V m Ck 100
V ads =
 , ð1Þ
1 − kRH
100 1 + ðC −1Þ 100
kRH

p RH
= , ð2Þ
p0 100

where Vm is the monolayer capacity, C and k are the adsorption GAB constants, and p and p0 (Equation 2) are respec-
tively the partial water vapor experimental pressure and its value at saturation. The GAB constants C and k help in
identifying the sigmoidal shape of the isotherm functions as reported in Section 3. The mathematical model behind this
type of isotherm, besides describing in detail the sorption behavior at RH < 93%, has a strong theoretical background
with parameters that may explain the sorption processes (further refinement of the Langmuir and BET models of physi-
cal adsorption). The GAB mathematical model is representative of microporous materials where unrestricted
monolayer–multilayer adsorption at the internal surface of a material exists. The state of the sorbate molecules in the
second layer is identical to that one in superior multilayers but different from that of the monolayer below (directly
bound to the wood hydroxyls) or of the liquid state molecules on the top of the multilayers. Therefore, as stated in
Andrade et al.,13 CGAB coefficient represents the water primary layer binding strength (with respect to the multi-
layers15,16) at the sorption sites at a certain T value (24
C = 297.15 K). The correcting factor kGAB, with pn being the par-
tial water vapor pressure of multilayers, describes a less structured state of the sorbate in multilayers (on the top of the
monolayer) than in the sorbate's pure bulk liquid state7,15,17:

p0
k GAB = < 1: ð3Þ
pn

Details concerning the physical meaning of the C and k parameters are reported in Data S1.
4 of 12 BERTOLIN ET AL.

3 | R E S UL T S A ND D I S CUS S I O N

A preliminary characterization of the coated surfaces has been described in previous works14 currently in press and
involved the evaluation of (a) surface homogeneity, penetration of chemicals, and treatment thickness using both opti-
cal and electron microscopy; (b) color variation of the samples' surfaces caused by the coating with a spectrocolorimeter
and a spectrophotometric scanner; (c) surface wettability by means of static contact angle measurements.
This work, instead, analyzes and discusses in details the obtained sorption–desorption data, evaluating the type of
adsorption isotherms and hysteresis measured for the six tested samples, the way water molecules are bound in the
wood structure at different RH conditions, and finally discussing the possible reasons, which explain the observed
effects, comparing our findings with those currently in press and available in literature.5,18–20

3.1 | Adsorption isotherms types and GAB constants of the analyzed samples

In 1940, Brunauer et al.21 was the first who classified sorption isotherms in five types based both on their shape and on
their physico-chemical processes influencing these shapes. In this study, the analysis of experimental data (reported in
Table 1) compared with literature results22–24 reports that all the samples show isotherms of type II21 with GAB-like sig-
moidal shape with overall determination coefficient R2 > 0.999 because CGAB > 5.7623 (except epoxy resin) and
0.24 < kGAB < 1.
Table 1 reports the Vm coefficients calculated from the GAB fit per each tested sample. It is representative of the
minimum water monolayer (or minimum moisture content) covering the active hydrophilic binding sites on the sample
under examination. It provides information on possible risk of decay caused by variation (increase) of moisture content
in the material or, in our case study, on the material service life after the treatments. This is mainly linked to the possi-
bility of having available reactive sites on the sample surface, a characteristic that can be modified by the presence of
surface treatments. Generally, Vm values decrease with higher temperature or due to a change on physico-chemical
characteristics as in this study. Treated pine samples show, in fact, a Vm decrease of up to approximately 22% with
respect to the reference clean sample. These findings indicate a lower availability of active hydrophilic bonding site on
treated samples, except in the case of epoxy resin (E) with instead shows a Vm value comparable with the untreated
pine sample. In particular: P40 and P40 + CS show similar decreasing effects highlighting the same service life horizon,
basically because the final surface to be considered is P40 based for both samples; the wood surface treated with the
sprayed cellulose sand spray (CS) displays the maximum decrease in water accessing to primary sorption sites; an inter-
mediate decrease is observed for the sample treated with the Renaissance microcrystalline wax (R).
In Figure 1 the treatments effects on CGAB and kGAB for adsorption branch of isotherms are reported. The effect of
the treatment on CGAB can be expressed by Equations 4 and 5

CGAB,treated = CGAB,untreated  CGAB,chemical : ð4Þ

 I 
C GAB,treated H 0 −H In
= C GAB,chemical = C 0,chemical exp : ð5Þ
CGAB,untreated RT

TABLE 1 Guggenheim–Andersen–de Boer (GAB) parameters (Vm, Ads, CGAB and kGAB) and their determination coefficient (R2) as
obtained by experiments with the microbalance on the five treated samples (i.e., P40, CS, R, E, and P40 + CS) and the untreated reference
standard (clean)

Treatment Vm (cm3/g) CGAB kGAB R2

Clean 0.061 ± 0.001 8.530 ± 0.530 0.758 ± 0.005 0.999
P40 + CS 0.057 ± 0.001 7.680 ± 0.330 0.734 ± 0.004 0.999
P40 0.056 ± 0.001 7.630 ± 0.170 0.722 ± 0.002 0.999
CS 0.048 ± 0.001 6.810 ± 0.510 0.816 ± 0.004 0.999
R 0.052 ± 0.002 5.980 ± 0.640 0.791 ± 0.008 0.999
E 0.062 ± 0.003 4.280 ± 0.340 0.644 ± 0.014 0.999
BERTOLIN ET AL. 5 of 12

F I G U R E 1 The Guggenheim–Andersen–de Boer (GAB)

constants C and kGAB of the five treated (i.e., P40, CS, R, E, and
P40 + CS) samples and of the untreated (clean) sample

Therefore, from Table 1 and Figure 1 (blue dataset), comparing CGAB of the treated samples with the untreated one,
a damping factor can be clearly visible. Except for the correction provided by the entropic accommodation factor
(i.e., C0,chemical), the overall observed effect on the measurements represents the modification of the molar sorption
enthalpies of multilayers (on top of the monolayer) with respect to that of the monolayer of water, in this specific case
caused by the application of chemicals. This is due to a change, generally a weakening, of the bonds formed between
water and the primary sorption sites on treated surfaces with respect to what happens on clean wood. Consequently, a
decrease in the CGAB can be interpreted that treated surfaces are adsorbing water molecules less strongly that untreated
wood. In other words, it can give indication on how and how strongly molecules creating the very first water layer
bound to the sample surface are in turn bound to molecules creating the successive water layers. Modifications of
values reported in Table 1 can be interpreted considering the strength of chemical bonds forming between the samples
surfaces and the first water monolayer.
In untreated wood, hydrogen bond forms between the material hydroxyl groups and water molecules. In this case,
dipoles form because the electronegative oxygen atoms attract hydrogen valence electrons acquiring a partial negative
charge (Oδ−) and leaving partial positive charges on such atoms (Hδ+). At the same time, these Hδ+ are attracted by
Oδ− of other water molecules creating the H bonds whose strength is comparable with that of bonds formed with the
primary sites on wood. In case of treatments, when water molecules interact with materials having CH2/CH3 terminal
groups, that is, when a polar liquid is deposited on a nonpolar surface, only weak interactions are possible due to the
extremely weak dipole-induced dipole attractions. A slight shift in the center of gravity of the negative partial charge on
the water molecules creating the first water layer occurs; thus, consequently, bond formed with water molecules in the
following layers become weaker than normally expected. In other words, after treatments, the monolayer of water binds
more weakly to primary sorption sites20,25 and consequently less energy (heat) is necessary to break such weak bonds
and remove water molecules.
More specifically, the values found for P40 and P40 + CS show a variation of approximately −10% with respect to
untreated wood, while the strength of bonds formed on the surfaces of samples treated with cellulose sand spray (CS),
Renaissance microcrystalline wax (R), and epoxy resin seem to gradually decrease.
Similarly, the effect of the treatment on kGAB can be expressed by Equations 6 and 7.

k GAB, treated = k GAB,untreated  kGAB, chemical, ð6Þ

 I 
kGAB,treated H 0 −H In
= k GAB,chemical = k 0,chemical exp , ð7Þ
kGAB,untreated RT

and it is associated to the relation existing between the already formed multilayer and water deposited entirely in its liq-
uid phase. CGAB is found to be an order of magnitude higher than kGAB describing a higher sorption heat of the
6 of 12 BERTOLIN ET AL.

monolayer compared with the multilayer.26 The change in kGAB coefficient on treated samples (green dataset in
Figure 1) shows two different tendencies depending on the type of applied chemicals:

1 A decrease, indicating a less favorable situation to surface saturation, means the amount of sorbed water vapor is
lower than the “idealized” case at RH = 100%. This is the case observed with Paraloid-coated surfaces (P40 and
P40 + CS) and at the maximum extension with the application of epoxy resin (E) treatment. Such behavior was inter-
preted by Timmermann in 20037 as a demonstration of less stable and structured state of the multilayers.
2 An increase in the dewpoint temperature (TDP), which reduces the dew point spread, that is, the difference between
TDP and the T, creating a more favorable situation to saturation. This was observed for samples treated with Renais-
sance wax (R) or the cellulosed sand layer sprayed alone (CS) that highlights a possibly more structured state of
water multilayers.

3.2 | Analysis of EMC variation and hysteresis caused by the treatments

As widely reported (see, e.g., Almeida et al.27) the water sorption in wood, relates to two types of water, that is, the
bound and free water. The first is adsorbed on the wood cell walls, while the second mainly fills in the wood lumen.27
The relative proportions between these types of water are in relation with the EMC of wood that is sensitive to changes
in p/p0 (and T). Figure 2 reports the isotherms of six tested samples (i.e., p/p0 vs. EMC adsorption and desorption bra-
nches at T = 24
C). As generally observed (top panel), both adsorption branches and desorption branches of treated
samples have lower EMC than the clean sample. A possible reason of that, as reported in Broda et al.,20 could be the dif-
ferent saturation pressures of the used chemicals lowering the EMC values expected for the treated wood samples.

TABLE 2 Parameters (Vm, Ads) as obtained by experiments with the microbalance on the five treated samples (i.e., P40, CS, R, E, and
P40 + CS) and the untreated reference standard (clean) and as calculated from the first derivative of equilibrium moisture content with
respect to p/p0 (±10%) as reported in Figure 2
  =0
∂EMC
V m,ads −V ads at min
Treatment EMCads at ∂ pp Min p/p0 (%) Vm,ads − Vads at min V m,ads (%) Vads at p/p0 min
0

Clean 14.66 ± 1.47 20 0.019 ± 0.01 30.2 ± 0.02 0.043 ± 0.01

R 12.40 ± 1.24 40 0.004 ± 0.01 5.94 ± 0.06 0.049 ± 0.01
P40 + CS 11.53 ± 1.53 50 −0.017 ± 0.01 −30.2 ± 0.02 0.074 ± 0.01
P40 11.48 ± 1.15 40 −0.0040 ± 0.01 −6.59 ± 0.02 0.060 ± 0.01
CS 10.88 ± 1.09 40 −0.007 ± 0.01 −11.26 ± 0.02 0.055 ± 0.01
E 10.43 ± 1.04 30 0.023 ± 0.01 38.14 ± 0.07 0.038 ± 0.02

F I G U R E 2 Top: equilibrium moisture content (EMC) for the

isotherms (adsorption/desorption branches) of 6 tested samples:
clean (black), CS (sealed spray—blue), R (Renaissance
microcrystalline wax – orange), P40 + CS (Paraloid above the
cellulose sand seal spray layer—light green), P40 (Paraloid—Red)
and E (epoxy resin—violet). Bottom: First EMC derivative respect to
p/p0 for the same datasets. Solid lines: sorption branches; dashed
lines: desorption branches
BERTOLIN ET AL. 7 of 12

The results reported in Table 2 helps in the analysis of the wood–water interactions as follows.
In the clean wood, the monolayer saturation moisture content in adsorption (Vm,Ads) is equal to 0.061 cm3/g. This is
the maximum amount of water taken by the environment on the sample via monolayer sorption. The point after which
multilayer sorption starts to prevail is highlighted in Figure 2 (bottom panel) by the minimum of the derivative of the
EMC in relation to p/p0 and corresponds to the value VAds = 0.043 cm3/g, occurring at 20% RH (min p/p0 in Table 2).
This indicates that from 20% to 100%, the water adsorbed on the untreated sample via monolayer sorption is approxi-
mately 0.01858 (or 30%). The minimum of the derivative falls at 30% RH for E, displaying a water adsorption via mono-
layer of about 12%, and at 50% for CS + P40, while for the other samples, the minimum of the curve is reached at 40%
RH. In the case of the microcrystalline wax treatment (R), a behavior similar to that of clean wood is observed with
about 6% of water adsorbed on sample via monolayer.
At the opposite, P40 and CS deposited alone and together, does not show any additional uptake of water via mono-
layer but rather that already within the inflection point (i.e., 40% and 50% RH) the multilayer sorption is the prevailing
phenomenon.
In addition, Figure 2 (bottom panel) reports interesting information on the first derivative that is the rate of EMC
change with RH depending on the treatment. The first derivative can be defined as a sort of moisture coefficient, which
exemplifies the propension to contraction (negative slope of the first derivative in Figure 2—bottom panel—at low RH
levels) or to expansion (positive slope of the first derivative in Figure 2—bottom panel—at higher RH levels) of the
tested samples when the RH is changing. When considering the adsorption branch, it highlights that—in expansion—
the cellulose sand seal spray layer (CS) has the fastest EMC change rate, that is, 0.58% each 1% in RH while the Renais-
sance wax 0.48%; the three samples (i.e., clean, CS, and P40-coated samples) have a similar rate, that is, 0.37%, 0.33%,
and 0.30%; and finally, the epoxy resin has the slowest rate of EMC change with 0.17% each 1% in RH.
While in contraction, the fastest EMC change rate is −0.12% each 1% in RH for Paraloid coupled with cellulose sand
seal layer (P40 + CS), followed by the group of untreated, treated with P40 and CS samples with −0.09% each 1% in
RH. The most stable sample in shrinkage is that treated with the epoxy resin. The minimum response of the samples is
generally within 30% < RH < 40%.
Looking at the desorption and adsorption branches of the isotherms, the most remarkable alterations of the hystere-
sis have been calculated per each tested sample with Equation 8:

ΔEMCmax = ðEMCdes − EMCads Þmax , ð8Þ

and related to the corresponding p/p0 (i.e., RH) condition. ΔEMC, over the whole range of partial pressures used in the
experiments, is shown in Figure 3A, and Table 3 reports the results as follows:

• The untreated pine, the sample treated with the seal spray (CS) alone and coupled with Paraloid (P40 + CS), and the
microcrystalline wax (R) have the largest hysteresis at 70% RH and range from 3.41 to 3.68 g/kg going from the

F I G U R E 3 (A) The alteration of the hysteresis (ΔEMC) and (B) the hysteresis coefficient for the six tested samples: clean (black), CS
(sealed spray—blue), R (Renaissance microwax—orange), P40 + CS (Paraloid above a cellulose layer—light green), P40 (Paraloid—red), and
E (epoxy resin—violet)
8 of 12 BERTOLIN ET AL.

TABLE 3 Parameters to assess the hysteresis alteration as maximum difference between desorption and adsorption branches and the
corresponding p/p0 (±10%) value.

Treatment ΔEMC (g/kg) p/p0 (%)

Clean 3.42 ± 2.68 70
CS 3.49 ± 2.18 70
P40 3.06 ± 2.30 60
P40 + CS 2.80 ± 2.4 70
R 3.69 ± 2.48 70
E 1.98 ± 2.08 50

untreated to the R sample, while the combination of Paraloid and seal spray shows a difference in the hysteresis of
2.80 g/kg.
• The Paraloid alone (P40) at 60% shows a similar hysteresis of 3.06 g/kg.
• The epoxy resin (E) at 50% the lowest one with 1.98 g/kg.

In Figure 3B, the hysteresis coefficient (Equation 9) is reported:

EMC ads
hysteresis coeff = : ð9Þ
EMC des

Hysteresis are similar for epoxy resin (E) and Paraloid (P40) treated samples, while wood treated with the seal spray
(CS) has some similarities with that coupled with Paraloid (CS + P40); a halfway behavior between the two groups is
shown by the R sample.

3.3 | Water sorption in pine at low and high value of RH and discrimination of chemical
treatments on the surfaces

In literature, the water vapor sorption has been used to estimate specific surface area of materials by adaptation of the
approaches exploiting the advantages available in modeling methods.28–32 Similarly, in this section, the GAB model has
been used to further analyze the water sorption behavior on treated and untreated wood basing on the calculation of
the water film thickness tGAB (Equation 10), that is, the thickness of the water layer adsorbed by wood:

V ads
t GAB = , ð10Þ
SGAB

where Vads is obtained using the whole water vapor sorption isotherms fitted to the GAB model and SGAB is the specific
surface area calculated per each sample from Equation (11):

σ
SGAB = V m ρN ð11Þ
M

with N, the Avogadro number, ρ, the water density, σ, the average area occupied by one water molecule in the complete
monolayer (0.114 nm2), and M, the water molar mass of Timmermann and Almeida et al.7,27 Similarly, the direct mea-
sures obtained with the help of microbalance (i.e., the Vads − p/p0 isotherms) can be expressed in terms of water film
thickness with a simple linear plot as it results from Equation 11 and is shown in Figure 4. The variation of SGAB value
is linked to the distance between clusters of water molecules on the treated surface: The more the distance increases,
the more the SGAB value decreases.
Looking at the set of values of the specific surface area of the tested samples obtained from the GAB model applica-
tion reported in Table 4, the lowest values were found for CS and R surfaces: observing also the results previously
BERTOLIN ET AL. 9 of 12

FIGURE 4 Vads–t relationship for the tested pine samples: clean (black), R (Renaissance microcrystalline wax—orange), E (epoxy
resin—violet), P40 (Paraloid—red), CS (sealed spray—blue), and P40 + CS (Paraloid and cellulose sand seal layer—green)

TABLE 4 Specific surface area of untreated and treated wood (i.e., clean, CS, R, P40 + CS, P40, and E) as obtained from the
Guggenheim–Andersen–de Boer analysis of sorption data

Treatments SGAB (m2/g)

Clean 233.4 ± 0.2
E 234.9 ± 0.7
P40 + CS 216.6 ± 0.2
P40 213.2 ± 0.1
R 198.2 ± 0.3
CS 184.0 ± 0.2

reported in other publications currently in press.14 These treatments displayed the highest static contact angle values, so
it seems reliable to hypothesize that the deposited coatings promote the formation of smaller water droplets that tend
to remain more distant from each other. At the opposite, as the epoxy resin is a hydrophilic substrate, water molecules
tend to form bigger and closer clusters. In the first case, cohesion forces within water molecules will prevail on the
adhesion forces between water molecules and the substrate. Such behavior of SGAB is inversely proportional to kGAB:
The highest SGAB values correspond to the lowest kGAB ones. Linking kGAB to the strength of bonds existing between
the water multilayer and final layer of water in the liquid phase; the connection with the surface wettability properties
seems to be supported.
Figure 5 highlights as on the final part of the isotherm (i.e., p/p0 > 0.8 or RH > 80%) the use of the GAB
model better explains as the water vapor accessible on surface layer of treated samples is already grouped together
in multilayers as the Pearson's coefficient of the linear fit RGAB is equal to 0.99 at RH = 90%. (right side). At the
opposite, the GAB model underestimates the water thickness of the monolayer at low RH, that is, <30% especially
10 of 12 BERTOLIN ET AL.

F I G U R E 5 kGAB–t relationship for water thickness layer

obtained from the Guggenheim–Andersen–de Boer (GAB) model
fitted on the sorption data at experimental condition of 10% (black
thick line), 30% (black dash thick line), 40% (black short dash thick
line), 70% (black thin line), 80% (dark gray thin line), and 90% (dark
gray thin dash line) of RH. Samples: clean (black), R (Renaissance
microcrystalline wax—orange), E (epoxy resin—violet), P40
(Paraloid—red), CS (sealed spray—blue), and P40 + CS (Paraloid
and cellulose sand seal layer—green)

for the samples treated with microcrystalline wax (R) and the cellulose sand seal (CS) on the sample surfaces are
chemically less grouped together.32

4 | C ON C L US I ON S

The presented study proposed the application of the GAB equation to investigate the monolayer water content as well
as specific surface area of wood from water sorption of untreated pine wood used as standard reference in comparison
with samples coated with different materials able to limit the penetration/expulsion of moisture through the radial sur-
face. In particular, a cellulose sand seal spray alone or coupled with Paraloid B72 at 40%(w/v), a microcrystalline wax
and an epoxy resin were tested.
Analysis carried out by means of a vacuum microbalance highlighted differences in terms of water content on
the samples modified at the surface with chemicals. In particular, the modification of the kGAB parameter as well
as of Vads, CGAB, and SGAB were here carefully evaluated. Differences observed in case of coated samples with
respect to the untreated one could be explained in terms of lower availability of active hydrophilic bonding sites,
strength modifications of the chemical bonds formed between the surface and first water layer, and consequently
between it and the following ones up to the last deposited layer entirely considered as water in the liquid phase.
At the same way, the chemistry of modified surfaces strongly affected their wettability and the capability of water
molecules to aggregate in clusters more or less distant from each other. Then, the analysis carried out on EMC
variation and hysteresis of isotherms on samples reported the RH ranges within which treated wood could addi-
tionally uptake water via monolayer as well as their different rates in shrinkage and swelling at different RH
conditions.
Finally, results clearly highlights as, working at partial pressure condition higher than 0.8 (the last part of the iso-
therm, that is, p/p0 > 0.8 or RH > 80%), the corrective GAB factor kGAB results particularly useful in discriminating the
water vapor uptake on the material surface (expressed as thickness of the water film). This is shown by the kGAB–t rela-
tionship that clearly distinguishes the type of chemicals in relation to their rate of moisture uptake, indicative of their
propensity for grouping in multilayers and of their binding strengths.
As evident, the reported results clearly showed for the first time how the GAB parameters can be used in the study
of wooden materials in order to obtain useful information about their behavior toward water in both its vapor state and
its liquid state. Additionally, it has been demonstrated how modification in these values can help discriminating if
wooden materials have been chemically treated or not and what class of chemical was used.

ACK NO WLE DGE MEN TS

This study was funded by the Norwegian Research Council within the framework of the “SyMBoL – Sustainable Man-
agement of Heritage Building in a Long-term Perspective” Project (Project No. 274749). This research was funded in
part by the statutory research fund of ICSC PAS.
BERTOLIN ET AL. 11 of 12

CONFLICT OF INTEREST
On behalf of all authors, the corresponding author states that there is no conflict of interest.

A U T H O R C ON T R I B U T I O NS
C. Bertolin is responsible for data elaboration—interpretation, writing, and acquisition of financing. L. de Ferri is
responsible for sample preparation, data interpretation, and writing. M. Strojecki is responsible for data acquisition—
elaboration and review of the content.

ORCID
Chiara Bertolin https://orcid.org/0000-0002-0684-8980
Lavinia de Ferri https://orcid.org/0000-0001-8904-4779
Marcin Strojecki https://orcid.org/0000-0001-6411-1306

R EF E RE N C E S
1. Esteves BM, Pereira HM. Wood modification by heat: a review. Bioresources. 2009;4:370-404.
2. Hakkou M, Petrissans M, Zoulalian A, Gerardin P. Investigation of wood wettability changes during heat treatment on the basis of
chemical analysis. Polym Degrad Stab. 2005;89:1-5.
3. Kumar S. Chemical modification of wood. Wood Fiber Sci. 1994;26:270-280.
4. Rowell RM. Chemical modification of wood. In: Rowell RM, ed. Handbook of Wood Chemistry and Wood Composites. 2nd ed. CRC Press,
Taylor and Francis group; 2013:538-597.
5. Bratasz Ł, Kozłowska A, Kozłowski R. Analysis of water adsorption by wood using the Guggenheim-Anderson-de Boer equation. Eur J
Wood Prod. 2012;70:445-451.
6. Hailwood AJ, Horrobin S. Absorption of water by polymers: analysis in terms of a simple model. Trans Faraday Soc. 1946;42B:84-92.
7. Timmermann EO. Multilayer sorption parameters: BET or GAB values? Colloid Surface A. 2003;220:235-260.
8. Camuffo D. Microclimate for cultural heritage. In: Measurement, Risk Assessment, Conservation, Restoration, and Maintenance of Indoor
and Outdoor Monuments. 3rd ed. Elsevier; 2019.
9. Van den Berg C. Vapour Sorption Equilibria and Other Water Starch Interactions; a Physicochemical Approach, PhD thesis. Wageningen
Agricultural University; 1981.
10. McMinn WAM, Magee TRA. Moisture sorption characteristics of starch materials. Drying Technol. 1997;15(5):1527-1551.
11. Rizvi SSH. Thermodynamic properties of foods in dehydration. In: Rao MA, Rizvi SSH, eds. Engineering Properties of Foods. New York:
Marcel Dekker; 1986:133-214.
12. Vertucci CW, Leopold AC. Bound water in soybean seed and its relation to respiration and imbibitional damage. Plant Physiol. 1984;75:114-117.
13. Andrade RD, Lemus R, Perez CE. Models of sorption isotherms for food: uses and limitations. Vitae, revista de la facultad de quimica
farmace. 2011;18:325-334.
14. de Ferri L, Strojecki M, Bertolin C. Preliminary results on surface treatments on wood. IOP Conf Ser Mat Sci. 2020. Accepted 1–12.
15. Bertolin C, de Ferri L, Grottesi G, Strojecki M. Study on the conservation state of wooden historical structures by means of acoustic
attenuation and vacuum microbalance. Wood Sci Technol. 2020;54:203-226.
16. Quirijns EJ, van Boxtel AJB, van Loon WKP, van Straten G. Sorption isotherms, GAB parameters and isosteric heat of sorption. J Sci
Food Agric. 2005;85:1805-1814.
17. Wolf W, Spiess WEL, Jung G, Weisser H, Bizot H, Duckworth RB. The water-vapour sorption isotherms of microcrystalline cellulose
(MCC) and of purified potato starch. Results of a collaborative study. J Food Eng. 1984;3:51-73.
18. Lesar B, Gorisek Z, Humar M. Sorption properties of wood impregnated with boron compounds. Sodium Chloride and Glucose Dry
Technol. 2009;27:94-102.
19. Jang X, Li H, Ramaswamy HS, Zhu S, Yu Y. Moisture sorption isotherms and isosteric heats of sorption of high-pressure treated Paulow-
nia wood under different storage conditions. Transac the ASABE. 2019;62:1-10.
20. Broda M, Majka J, Olek W, Mazela B. Dimensional stability and hygroscopic properties of waterlogged archeological wood treated with
alkoxysilanes. Int Biodeter Biodegr. 2018;133:34-41.
21. Brunauer S, Deming LS, Teller E. On a theory of van der Waals adsorption of gases. J Am Chem Soc. 1940;62:1723-1732.
22. Van den Berg C. Description of water activity of foods for engineering purposes by means of the G.A.B. model of sorption. In:
McKenna BM, ed. Engineering and Foods. Vol.1 New York: Elsevier; 1984:311-320.
23. Lewicki PP. The applicability of the GAB model to food water sorption isotherms. Int J Food Sci Technol. 1997;32:553-557.
24. PeasanthaBDR. Prediction of moisture adsorption characteristics of dehydrated fruits using the GAB isotherm model. Ann Agric Crop
Sci. 2018;3:1036–1039.
25. Timmermann EO, Chirife J, Iglesias HA. Water sorption isotherms of food and foodstuffs: BET or GAB parameters? J Food Eng. 2001;
48:19-31.
26. de Oliveira GHH, Correa PC, de Oliveira APLR, Devilla IA. Application of GAB model for water desorption isotherms and thermody-
namic analysis of sugar beet seeds. J Food Process Eng. 2017;40:1–8. e12278.
12 of 12 BERTOLIN ET AL.

27. AlmeidaG, Gagné S, Hernández RE. A NMR study of water distribution in hardwoods at several equilibrium moisture contents. Wood
Sci Technol. 2007;41:293–307.
28. TullerM, Or D. Water films and scaling of soil characteristic curves at low water contents. Water Resour Res. 2005;41:W09403.1–6.
29. Arthur E, Tuller M, Moldrup P, et al. Soil specific surface area and non-singularity of soil-water retention at low saturations. Soil Sci Soc
Am J. 2013;77:43-53.
30. Akin ID, Likos WJ. Specific surface area of clay using water vapor and EGME sorption methods. Geotech Test J. 2014;37:1016-1027.
31. Leao TP, Tuller M. Relating soil specific surface area, water film thickness, and water vapor adsorption. Water Resour Res. 2014;50:7873-
7885.
32. Arthur E, Tuller M, Moldrup P, Greve MH, Knadel M, de Jonge LW. Applicability of the Guggenheim-Anderson-Boer water vapour
sorption model for estimation of soil specific surface area. Eur J Soil Sci. 2018;69:245-255.

S UP PO RT ING IN FOR MAT ION

Additional supporting information may be found online in the Supporting Information section at the end of this article.

How to cite this article: Bertolin C, de Ferri L, Strojecki M. Application of the Guggenheim, Anderson, de Boer
equation to study the impact of sealing treatments on pine wood sorption characteristics. Mat Design Process
Comm. 2020;e189. https://doi.org/10.1002/mdp2.189