ETH Library
Quantitative determination of
bound water diffusion in multilayer
boards by means of neutron
imaging
Journal Article
Author(s):
Sonderegger, Walter; Hering, Stefan; Mannes, David; Vontobel, Peter; Lehmann, Eberhard; Niemz, Peter
Publication date:
2010
Permanent link:
https://doi.org/10.3929/ethz-b-000157725
Rights / license:
In Copyright - Non-Commercial Use Permitted
Originally published in:
European Journal of Wood and Wood Products 68(3), https://doi.org/10.1007/s00107-010-0463-5
This page was generated automatically upon download from the ETH Zurich Research Collection.
For more information, please consult the Terms of use.
Eur. J. Wood Prod. (2010) 68: 341–350
DOI 10.1007/s00107-010-0463-5
ORIGINALS ORIGINALARBEITEN
Quantitative determination of bound water diffusion in multilayer
boards by means of neutron imaging
Walter Sonderegger ·Stefan Hering ·David Mannes ·
Peter Vontobel ·Eberhard Lehmann ·Peter Niemz
Received: 6 April 2010 / Published online: 10 July 2010
© Springer-Verlag 2010
Abstract Diffusion processes into multilayered samples of
Norway spruce (Picea abies [L.] Karst.) exposed to a differ-
entiating climate (dry side/wet side) were determined and
quantified by means of neutron imaging (NI). The experi-
ments were carried out at the neutron imaging facility NEU-
TRA at the Paul Scherrer Institute (PSI) in Villigen (Switzer-
land).
With NI the influence of different adhesives (polyvinyl
acetate (PVAc), urea formaldehyde resin (UF), epoxy resin
(EP), one-component polyurethane (1C PUR)) on the dif-
fusion process could be determined by varying the layer
number and the thickness of adhesive joints of the samples.
Thereby, neutron transmission images were used to measure
time dependent water profiles in the diffusion direction. Us-
ing Fick’s second law, diffusion coefficients for radial and
tangential water transport in spruce wood and in the adhe-
sive joints were calculated depending on moisture content
(MC). It was found that the diffusion coefficients of the ad-
hesives (1C PUR, EP at high MC) were up to three orders
of magnitude lower than those of spruce wood. PVAc and
UF had a smaller barrier effect compared to wood, which in
contrast to 1C PUR and EP, clearly depends on the MC.
This article is dedicated to Gerd Wegener on the occasion of his
retirement as professor at the Technische Universität München.
W. Sonderegger (
) · S. Hering · P. N iem z
Department of Civil, Environmental and Geomatic Engineering,
Institute for Building Materials, ETH Zurich, 8093 Zurich,
Switzerland
e-mail: wsonderegger@ethz.ch
D. Mannes · P. Vontobel · E. Lehmann
Spallation Neutron Source (ASQ), Paul Scherrer Institute (PSI),
5232 Villigen, Switzerland
Quantitative Bestimmung der Diffusion von
gebundenem Wasser in mehrlagigen Brettlamellen
mittels Neutronenradiographie
Zusammenfassung Es wurden Diffusionsprozesse an
mehrlagigen Proben von Fichte (Picea abies [L.] Karst.),
welche einem Differenzklima (trocken/feucht) ausgesetzt
waren, mittels Neutronenradiographie untersucht und quan-
tifiziert. Die Experimente wurden an der Radiographie-
strahllinie NEUTRA am Paul Scherrer Institut (PSI) in Vil-
ligen (Schweiz) durchgeführt.
Mittels Neutronenradiographie konnte der Einfluss ver-
schiedener Klebstoffe (Polyvinylacetat (PVAc), Harnstoff-
harz (UF), Epoxidharz (EP) und Einkomponenten-Poly-
urethan (1K-PUR)) auf den Diffusionsprozess bestimmt
werden, indem die Anzahl und die Dicke der Klebfugen va-
riiert wurden. Dabei wurden Neutronen-Transmissionsbilder
verwendet, womit zeitabhängige Profile in Diffusionsrich-
tung gemessen werden konnten. Anhand des zweiten
Fick’schen Gesetzes konnten die Diffusionskoeffizienten für
die Klebstoffe sowie für Fichte in radialer und tangentia-
ler Richtung in Abhängigkeit der Feuchte berechnet wer-
den. Dabei wiesen die Klebstoffe (1K-PUR, EP bei hohen
Feuchten) bis zu drei Zehnerpotenzen niedrigere Diffusions-
koeffizienten als Fichtenholz auf. Bei PVAc und UF war die
Sperrwirkung gegenüber dem Holz geringer und es zeigte
sich im Gegensatz zu 1K-PUR und EP eine deutliche Ab-
hängigkeit von der Holzfeuchte.
1 Introduction
In recent years, glued solid wood materials such as glued
laminated timber (glulam) used as girder or multilayered
solid wood panels used for walls and ceilings have gained
342 Eur. J. Wood Prod. (2010) 68: 341–350
in importance. In this context, the diffusion of water vapour
within wood and the barrier effect of adhesive joints are
of high relevance to the field of building physics: this par-
ticularly is the case regarding the hygroscopic behaviour
and the water transfer in building elements and the humid-
ity exchange with the room. While diffusion processes of
wood are subject of many investigations (e.g. Stamm 1959a,
1959b; Kollmann and Côté 1968; Skaar 1988;Siau1995;
Frandsen et al. 2007), the influence of adhesive joints on the
diffusion processes has scarcely been studied. One of the
few studies done on this topic was carried out by Frühwald
(1973). He investigated in detail the influence of bond lines
of a phenolic resin between the veneers of beech plywood
by varying the number of bond lines and films per line. Re-
cently, Foglia (2006) investigated the influence of bond lines
of a one-component polyurethane adhesive between spruce
wood lamellas on the water vapour resistance using the cup
method according to ISO 12572 (2001). Based on these data
(resistance factors of spruce wood without bond lines and
with 2, 3 and 5 bond lines), the water vapour resistance fac-
tor of the bond lines could be calculated according to the
Glaser method (Glaser 1959). This resulted in a mean wa-
ter vapour resistance factor of 3900 for the wet cup test and
8700 for the dry cup test. This indicates a high dependence
of the adhesive’s water vapour resistance on the moisture
content (MC) of the adjacent wood, as already stated by
Frühwald (1973) for a phenolic resin.
The goal of this study was to investigate the phenomenon
of the moisture-depending diffusion resistance for differ-
ent adhesives: polyvinyl acetate (PVAc), urea formalde-
hyde resin (UF), epoxy resin (EP) and one-component
polyurethane (1C PUR). Neutron radiography was used to
detect the water content within wood (according to Mannes
et al. 2009b). The advantage of this non-destructive method
is its high sensitivity to hydrogen allowing visualisation
of time-dependent water diffusion processes in wood. In
the recent past, several investigations with neutron radio-
graphy were carried out in order to analyse the capillary
water absorption by partial immersion of wood specimen
into water (Lehmann et al. 2001a; Niemz et al. 2002;
Mannes et al. 2006). Mannes et al. (2009b) demonstrated
that even small amounts of water absorbed from air mois-
ture can be detected and quantified.
2 Material and methods
2.1 Material
20 samples from Norway spruce (Picea abies [L.] Karst.) of
approximately 120 mm (length) ×15 mm (width) ×30 mm
(height) were prepared for testing bound water diffusion.
The diffusion process was run from the top to the bottom
of the sample. Two of the samples were used for determina-
tion of the radial and tangential diffusion. The other 18 sam-
ples were added with one to five bond lines at half height or
rather regularly distributed over the height in order to deter-
mine the influence of bond lines on the diffusion processes.
For these samples, the diffusion in the spruce wood occurs
in the tangential direction.
The samples were tested in two series. Series 1 included
the two samples without adhesives and samples with 1, 3
or 5 bond lines of UF and 1C PUR. The applied adhesives
are commonly used for load-bearing timber structures, UF
(cold gluing) in dry conditions and 1C PUR even in humid
conditions. Series 2 includes samples with bond lines of EP,
PVAc, UF and 1C PUR; the thickness of the bond lines was
varied (0.1 mm, 0.5 mm, 1.0 mm). EP is a two component
adhesive of high strength, PVAc allows many general ap-
plications in the wooden industry and 1C PUR—in contrast
to that one of series 1—is an elastic adhesive for parquet
flooring. The used UF of series 2 is the same as in series
1. All adhesive joints were prepared at 20 °C and 65% rela-
tive humidity (RH). Table 1 gives an overview of all tested
samples.
2.2 Methods
2.2.1 Experimental setup
The experiments were performed at the neutron radiogra-
phy facility NEUTRA of the spallation neutron source SINQ
at the Paul Scherrer Institute (PSI) in Villigen, Switzerland
(Lehmann et al. 2001b). The samples were tested in two
series over a period of about ten weeks. The detailed test
conditions for the two series are described in Table 2.Dur-
ing this time, the preliminarily oven-dried samples were ex-
posed to a differentiating climate (series 1: 20 °C/85% RH
to almost 0% RH; series 2: 20 °C/90% RH to almost 0%
RH); the moisture of the samples was measured with neu-
tron radiation after defined time intervals in order to follow
the diffusion process into the sample.
Thereby, at the beginning of the tests the samples (dried
at 103 °C until weight constancy) are insulated with alu-
minium tape on the four edges, leaving only the top and the
bottom planes (the orthogonal planes of the required diffu-
sion direction) unsealed. Each sample is then fixed on a cup
filled with silica gel that is closed with a slotted aluminium
plate. The sample sits above the opening of the slot, which
has the same length and width as the sample (Fig. 1). The
first neutron radiation measurement is still made for the dry
sample so as to obtain a reference image.
For the measurement, the cup with the sample is posi-
tioned in front of the neutron detector (a combination of
a neutron sensitive
6
Li doped ZnS scintillator and a CCD-
camera) so that the tangential-longitudinal plane (and in one
Eur. J. Wood Prod. (2010) 68: 341–350 343
Table 1 Overview of the tested samples and the applied adhesives at the beginning of the measurements (oven-dry) and the mean moisture content
(MC
mean.end
) and water concentration (C
mean.end
), respectively at the end of the measurements after 70 days (series 1) or 74 days (series 2) exposed
to a differentiating climate (series 1: 20 °C/85% RH to 0% RH; series 2: 20°C/90% RH to 0% RH)
Tab. 1 Überblick über die untersuchten Proben und die verwendeten Klebstoffe zu Beginn der Messungen (darrtrocken) sowie deren mittlerer
Feuchtegehalt (MC
mean.end
) bzw. Wasserkonzentration (C
mean.end
) am Ende der Messungen, nachdem die Proben 70 Tage (Serie 1) bzw. 74 Tage
(Serie 2) einem Differenzklima ausgesetzt worden waren (Serie 1: 20 °C/85 % rLF zu 0 % rLF; Serie 2: 20 °C/90 % rLF zu 0 % rLF)
Material/adhesive No. Direction of
diffusion
Bond line
No.
(–)
Thickness of
the bond line
(mm)
Oven-dry
height
(mm)
Oven-dry
density
(kg/m
3
)
MC
mean.end
(%)
C
mean.end
(g/m
3
)
Series 1
Spruce wood
1 Tangential – – 29.1 418 12.8 0.054
2 Radial – – 29.9 402 12.5 0.050
Urea-1
3 Tangential 1 0.1 29.0 398 12.5 0.050
4 Tangential 3 0.1 29.0 441 10.6 0.047
5 Tangential 5 0.1 29.3 457 9.6 0.044
PUR-1
6 Tangential 1 0.1 29.0 403 12.0 0.048
7 Tangential 3 0.1 29.3 430 7.9 0.034
8 Tangential 5 0.1 29.7 476 6.1 0.029
Series 2
Epoxy
9 Tangential 1 0.1 29.3 429 14.0 0.060
10 Tangential 1 0.5 29.5 586 11.0 0.064
11 Tangential 1 1.0 30.1 523 11.0 0.058
PVAc
12 Tangential 1 0.1 29.3 407 14.8 0.060
13 Tangential 1 0.5 29.5 389 15.3 0.060
14 Tangential 1 1.0 29.8 351 14.6 0.051
Urea-2
15 Tangential 1 0.1 29.5 420 15.0 0.063
16 Tangential 1 0.5 29.5 419 15.3 0.064
17 Tangential 1 1.0 29.8 417 14.9 0.062
PUR-2
18 Tangential 1 0.1 29.4 458 14.2 0.065
19 Tangential 1 0.5 29.8 403 13.5 0.054
20 Tangential 1 1.0 30.5 452 12.4 0.056
Table 2 Test conditions
Tab. 2 Prüfbedingungen
Series 1 Series 2
Samples No. 1–8 No. 9–20
Measurement times after 0, 1, 3, 7, 23, 29, 59, 74 days 0, 1, 3, 7, 14, 21, 34, 49, 70 days
Climatic chamber: climate 20/85 20/90
Scintillator 200 µm Li6 200 µm Li6
Field of view 130 ×130 mm 279 ×279 mm
Pixel size 0.12695 ×0.12695 mm 0.27246 ×0.27246 mm
Exposure time 210 s 21–30 s
Object-detector distance 90 mm 67 mm
Proton flow ∼1.4 mA 1.07–1.40 mA
case the radial-longitudinal plane) of the sample is exposed
to a parallel beam of neutron radiation for a defined time
(Table 2). The neutron scintillator converts the neutron sig-
nal into visible light, which is led via a mirror onto a cooled
16 bit CCD camera (resolution: 1024 ×1024 pixels, field of
view and pixel size as given in Table 2).
After the measurement, the cup with the sample is put
into a climatic chamber with a climate of 20 °C and 85%
(series 1) and 90% (series 2) relative humidity, respectively.
Further measurements take place in an analogous manner
after defined time intervals (Table 2). The samples are re-
moved from the climatic chamber only for the few minutes
necessary for the radiography measurement.
2.2.2 Data evaluation
The attenuation of the neutron radiation within a sample can
be described by the exponential law of radiation attenuation
344 Eur. J. Wood Prod. (2010) 68: 341–350
Fig. 1 Test setup of a sample with 5 bond lines. 1 = Sample; 2 = Sil-
ica gel; 3 =Aluminium tape; 4 =Glass vessel; 5 = Slotted aluminium
plate
Abb. 1 Prüfanordnung einer Probe mit 5 Klebfugen. 1 = Probe; 2 =
Silikagel; 3 = Aluminiumband; 4 = Glasgefäß; 5 = geschlitzte Alu-
miniumplatte
(Beer-Lambert law). For samples with a known thickness
this can be used to derive net moisture content in the samples
((1)–(4)).
T
w
=
I
w
I
ow
=e
−(
w
t
w
+
d
t
d
)
(1)
T
d
=
I
d
I
od
=e
−
d
t
d
(2)
T
w
T
d
=e
−
w
t
w
(3)
or
−ln
T
w
T
d
=
w
t
w
(4)
Thereby the raw grey-level images, I
w
for the wet sample
and I
d
for the dry sample, were converted to normalised
transmission images (T
w
for the wet sample and T
d
for the
dry sample); I
ow
and I
od
correspond to the intensity of the
respective incident neutron beam,
w
and
d
represent the
attenuation coefficients and t
w
and t
d
the layer thicknesses
of water and wood, respectively.
Since the interaction between neutrons and hydrogenous
materials, such as wood, occurs partially as scattering, for
quantitative analyses scattering corrections are necessary
(Mannes et al. 2009a). Thus, for the evaluation of the ex-
perimental data, the raw images were additionally corrected
with standard procedures using the scattering correction tool
QNI (Hassanein 2006). QNI includes:
• compensation for the offset caused by the background
noise of the CCD camera,
• “flat field” correction equalising inhomogeneities of the
beam intensity and the scintillator,
• corrections due to sample scattering depending on the ma-
terial and the distance of the sample to the scintillator and
• corrections due to background scattering where the neu-
trons are scattered by the experimental facility.
Thereby, the percentage of the background scattering cor-
rection (determined with a “black body”) was varied by ad-
justing the images to the dry part of the sample and the area
of the aluminium plate (if the sample has already taken up
moisture over the whole height, only the area of the alu-
minium plate was used for adjusting).
The goal of the measurements was to detect the distribu-
tion of the water content within the sample in the diffusion
direction. First, a profile over the entire height of the dry
sample was generated, averaging the values over almost the
whole sample width. Then the profiles of the following im-
ages (after the defined time intervals of the diffusion exper-
iment) were referenced against the profile gained from the
initial (oven-dry) image by dividing the wet profile by the
dry profile (3). Due to swelling of the sample during the dif-
fusion process and displacement of the position, the profiles
from the following images had to be first adjusted by shifting
and linear compressing in order to guarantee that the posi-
tion of the bond lines coincide with the profile of the initial
dry image. The differences between the measurements can
be shown via division of the profiles. Application of (4) il-
lustrates the change in water content. To adjust the water
content determined via neutron radiation, the mean water
content of the last image taken was referenced on the water
content obtained by weighing the sample after the measure-
ment was finished. Table 1 shows the mean water content of
the samples as MC and as water concentration.
2.2.3 Determination of the diffusion coefficient
Evaluation of the diffusion coefficient follows the method
described in Mannes et al. (2009b) using a 2nd order lin-
ear partial differential equation (PDE) of the diffusion cor-
responding to Fick’s second law. Owing to varying densities
in the different wooden layers of a sample due to anatomi-
cal properties such as early and latewood, Fick’s second law
was used in a modified version calculating the diffusion co-
efficient D depending on MC (cf. Olek and Weres 2007):
∂M
∂t
=
∂
∂x
D
∂M
∂x
(5)
where M is the MC in percent, t the time, x the moisture
transport direction and D the MC-dependent diffusion coef-
ficient, given in (6):
D =D
0
·e
α·M
(6)
