IRG/WP 05-50224:5

THE INTERNATIONAL RESEARCH GROUP ON WOOD PROTECTION

Section 5 Environmental aspects

Diffusion modeling of inorganic wood preservative leaching in

service

Levi Waldron, Paul Cooper, Tony Ung

Faculty of Forestry, University of Toronto,
33 Willcocks St., Toronto, M5S 3B3 Ontario, Canada
p.cooper@utoronto.ca

Paper prepared for the 6th International Symposium

“Environment and Wood Preservation”
Cannes-Mandelieu, France
7-8 February 2005

IRG SECRETARIAT
SE-100 44 Stockholm
Sweden
www.irg-wp.com
Diffusion modeling of inorganic wood preservative leaching in
service
Levi Waldron, Paul Cooper and Tony Ung
Faculty of Forestry, University of Toronto,
33 Willcocks St., Toronto, M5S 3B3 Ontario, Canada
p.cooper@utoronto.ca
Tel 416-946-5078
Fax: 416-978-3834

Abstract:
To evaluate the potential environmental and health implications of leaching of inorganic
wood preservatives in service under different conditions, there is a need for a predictive
model that provides estimates of the rate and extent of leaching over a wide range of
product dimensions and exposure conditions. In this paper, we show that the leaching
behavior of inorganic preservative components from wood in continuous water contact
can be characterized by three easily measured parameters: total leachable component
(Le) based on intensive leaching of fine ground material; amount of dissolved or
dissociated component (Di) in water saturated wood; preservative component diffusion
coefficients (Dt,l) in the transverse and longitudinal directions. Use of the applicable D
and Di or Le in a diffusion model allows the prediction of total amount leached and
emission rate at different times of exposure. Both D and Di increase somewhat with
increasing ambient temperature. Laboratory determined parameters for alkaline copper
quaternary (ACQ), copper azole (CA), chromated copper arsenate (CCA) and borate
(DOT) wood preservatives are used to predict leaching rates from larger lumber samples.
Preliminary comparisons of predicted leaching with measured leaching of larger samples
in laboratory and natural rain exposure indicate that the approach is surprisingly
effective at predicting leaching performance but some model refinements are needed for
some components such as copper in CCA to account for the slow dissolving of a
component of the preservative available for leaching.

Keywords: ACQ, borate, CCA, copper azole, diffusion, dissociation, leaching, risk

1. Introduction

The leaching characteristics of wood preservatives from treated wood exposed to

different environmental conditions affects both the long term efficacy of the treatment
against target organisms and the environmental impacts of the treatment. In the risk
assessment of new and currently in use wood preservatives, it is important to have
representative estimates of the leaching rates of the preservative components under
different exposure scenarios and to understand how these leaching rates change with
time and other variables. This information, in combination with data on dilution and fate
of substances in the environment, permits the estimation of predicted environmental
concentrations (PEC) at any time in service (Homan and van Oosten 1999). These PEC
values can be compared to predicted no effect concentrations (PNEC) to assess their
potential environmental impacts. There are a number of initiatives through the OECD,
EU, AWPA and other organizations and associations (OECD 2003, Hingston et al. 2001;
Lebow 1996; Morsing 2003; Willeitner and Peek 1998) to devise appropriate leaching
procedures that can predict both expected long term efficacy of the treatment and
potential impacts of leachates on human health and the environment. Generally
inorganic preservative components are easier to model than organic components
(Schoknecht et al 2003). Willeitner and Peek (1998) discussed the difficulty of devising
a practical test that is cheap, fast and reproducible while giving a realistic indication of
short term and long term impacts.

To predict long term performance of preservatives against target organisms, their

leaching characteristics are usually expressed as % of total preservative lost when
standard samples are exposed to a defined leaching protocol. The potential
environmental impact of wood preservative contaminants is best characterised by the
rate of preservative emission from the treated wood, expressed as amount released per
unit surface area per unit time or flux (e.g., mg/m2/day). It has been shown that flux
decreases with time in service and that the most acute environmental and health exposure
to leached substances occurs in the period soon after installation (Brooks 2001) while
long term accumulations in soil and sediments result from the lifetime exposure to ever-
declining emissions. Previous studies (Brooks 2001; Van Eetvelde et al.1994; Van
Eetvelde et al. 1998; Hingston et al. 2001) have modeled this declining flux rate, usually
with exponential equations. These models can be applied to risk assessment scenarios to
predict releases at any given time; long term accumulations in water columns, soils and
sediments can be estimated by determining the area under the curve by integration or by
use of a “growth model” as suggested by Homan and van Oosten (1999).

It is presumed that leaching of inorganic wood preservatives from wood involves a

number of physical phenomena depending on the exposure conditions. For example,
leaching of a preservative from a treated product exposed to rain above ground should
consists of the following concurrent and consecutive phenomena, each of which is
affected by a number of wood product, preservative and exposure conditions:
1. Wetting of wood and penetration of free water into the wood void space. This
factor depends on wood permeability, surface wetting properties, orientation to
the weather, type of exposure etc.
2. Dissolving/dissociation of precipitated or reacted preservative components into
free water in the cell lumens to the limit of their solubility under the test
conditions (depends on preservative, temperature, MC, etc.). This creates a
concentration gradient between the wood surface and its interior.
3. Diffusion of preservative components to the wood surface, under the
concentration gradient. The diffusion rate will depend on a number of factors
such as wood permeability, direction of movement in wood, moisture content,
temperature, dimensions, the nature of the diffusing species and the concentration
gradient established.
4. Drying of wood after rain exposure and possibly, wicking of dissolved
components to the wood surface (depends on ambient T, RH, air flow
conditions).

Morsing and Lindegaard (2004) demonstrated the difficulty in predicting above

ground natural exposure leaching using laboratory tests.
For wet wood in continuous contact with moisture i.e. not affected by the
conditions 1 and 4 above, it seems appropriate to model leaching emissions using a
diffusion model. Leaching of wood components from two opposing surfaces can be
described by one-dimensional Fickian diffusion.
 (1)

For leaching of a preservative component from wood, C is the concentration of

preservative in wood at any time t that is available for leaching. D is the diffusion
coefficient (cm2/s); x is the diffusion direction. The above equation is solved for
boundary conditions of (i) Uniform initial concentration in the wood at the start of
leaching, ie, the initial concentration C(x,0) is a constant Co, this assumes that all of the
available solute is dissolved in the cell lumen water and no additional preservative
dissociates. (ii) The component concentration is 0 at the wood surface/water interface
i.e., dissolved components are washed off the wood surface or greatly diluted; (iii) There
is continuous leaching in one direction only (ends and edges are coated or not
considered).
In this case, the mass of material Mt leached as a function of time, relative to the
ultimate amount available (M∞) is given by Equation 2. l is the sample length in the
direction of movement of leached component if leached from only one side, or half the
sample length if the two opposing faces are leached. For preservative systems with low
solubility reaction products, M∞ is the amount of preservative that is available for
leaching rather than the amount of preservative in the wood.

Mt ∞
8  − D(2n − 1)2 π 2 t 
=1− ∑ exp 

n = 0 (2n + 1) π
2 2 2
M∞  4l  (2)

Equation 2 relates cumulative loss (Mt/M∞) to elapsed (leaching) time t (s), specimen
dimension in the diffusion direction (2 l cm for diffusion from two opposing surfaces)
and diffusion coefficient (D).
Equation 2 can be solved in several ways including numerical solution using sum-of-
squares chi-square optimization or an analytical solution, for example calculating the first
6-8 terms of Equation 2 using a spread sheet. For a less precise but simpler estimation,
Equation 3 can be applied up to the point where the concentration of preservative
dissolved in the void water starts to be depleted at the centre of the sample (up to about
33% removal of leachable component):

πl 2 (M t / M ∞ )2
D= (3)
4t

Using either equation 2 or 3, D values can be estimated from controlled leaching studies.
Once D is known, Mt/M∞ can be estimated for different dimension specimens and
leaching times and presented in terms of % of total leached if the amount available for
leaching (M∞) is known and expressed as a % of total of preservative retained in the
wood (Ret):
 Mt
% Leached = 100(M ∞ / Re t )(M t / M ∞ ) = 100 (4)
Re t
assuming that Mt, M∞ and Ret are expressed in the same units (e.g., µg, % m/m, kg/m3
etc.).

Alternatively, estimates can be converted to flux estimates (e.g., mg/m2/d) at any elapsed
time (t) by computing Mt (in mg) at two consecutive time intervals and dividing the
change in estimate of amount leached over the interval by the exposed sample area (m2)
and the time interval in days:

M t 2 − M t1
Flux =
A(t2 − t1 ) (5)

All of the variables in the above equations are unambiguous except for M∞. For
preservative systems that are not chemically reacted with wood with highly soluble
components (such as borates) M∞ can be assumed to be the total retention (Ret) in wood.
For low solubility (fixed) systems, M∞ could be taken as the amount that is dissolved or
dissociated (Di) when the wood is wetted, since this concentration creates the diffusion
concentration gradient. However, in cases where this is not identical to the amount of
readily soluble component in the wood, the situation is more complex and the above
boundary conditions are not met. In this case total leachable amount (Le) may be a
better estimate for M∞ and it is likely that a more complex equation would have to be
used which considers the rate of solution of the available but undissolved component.

In consideration of the above, we suggest that the leaching behaviour of a given

inorganic preservative product under continuous moisture exposure conditions can be
characterised by three fairly simple-to-measure factors:
1. Equilibrium dissociation of preservative components in wood;
2. Percentage of preservative components available for leaching;
3. Directional diffusion coefficients for movement of preservative components out
of wood.

These results can then be applied to more general systems to predict long term leaching
of components from wood products of different geometries.

2. Experimental approach
2.1 Dissociation
To evaluate the equilibrium dissociation and rate of dissociation of preservative
components, solid treated wood samples (25 mm cubes) at a moisture content of
approximately 10 % were cut from southern pine (Pinus sp.) 38mm X 140 mm boards
commercially treated with chromated copper arsenate (CCA-C), alkaline copper
quaternary (ACQ), copper azole (CA) and disodium octaborate tetrahydrate (DOT).
The retention of the inorganic preservative component in ground samples was
determined (% and kg/m3) by digesting a 1 g sample in H2O2/Nitric Acid (AWPA 1993)
and analyzed by ICP spectroscopy for copper (CCA, ACQ, CA), chromium (CCA)
arsenic (CCA) and borate (ACQ and DOT). The wafers were vacuum/pressure
impregnated with water at room temperature and maintained at 22°C. The mass uptake
of water was determined and assumed to represent the amount of water in the wood
containing dissolved components. The rates of change in amounts of preservative
components in the free water in the cell lumens was determined by squeezing the blocks
in a press to express solution from matched samples at different times. Some samples
were evaluated at 5°C and 34°C to determine the temperature effect on this variable. The
expressed solution was analysed for the above inorganic components by ICP and the
mass of dissolved component (Di) expressed as a percentage of total content of the
wood:

Di = Water uptake (g) x concentration in expressate (µg/g) x 100 % (6)

Retention in wood (µg/g) X sample mass (g)

2.2 Ultimate amount available for leaching

The percentage of preservative components readily available for leaching was
estimated from the leaching performance of finely ground wood. Cross-section samples
of the same wood samples as above were ground to pass a 1 mm screen. Five grams of
the ground samples were leached in 50 mL water at 22ºC and the water exchanged
regularly over a 12-week period. Cumulative percent leaching was determined by
analysing the leachate by ICP.

2.3 Directional diffusion coefficients

Small blocks (20mmX20mmX40mm long) were cut from the southern pine boards
treated with each preservative system so that the transverse faces were perfect
radial/tangential and the longer dimension was parallel to the wood grain. To ensure
uniform initial preservative component concentration at the start of leaching, the blocks
were vacuum treated with water and stored for six days to allow full dissociation of the
component(s) before they were subjected to leaching. Continuous leaching in one
direction only was achieved by sealing all except one face with silicone sealant.
Samples were placed individually in 300 ml water and the water removed for analysis
and replaced with fresh water periodically. The leachate samples were analyzed by ICP
and the % of total preservative leached plotted vs. leaching time. For some
preservatives, samples were leached at different temperatures (5°C, 22°C, 34°C) to
compare T effect on leaching rate. The leaching curves were fit with the diffusion model
(Equation 2) to estimate D values. For preliminary studies, calculations were performed
using a FORTRAN program to determine best fit diffusion coefficients by minimizing
the sum of square of differences between experimental and calculated leachate
quantities. For the T dependence studies, Equation 3 was applied.

The maximum amount of preservative available for leaching (M∞) should be the total
leachable amount (Le) determined as above but the quantity of dissociated preservative
at equilibrium (Di) establishes the concentration gradient that controls the diffusion
process. Unless these quantities are identical, use of Le will over-estimate the rate of
leaching while using Di will under-estimate the cumulative amount of leaching that can
occur. In this study, diffusion values are estimated using the amount of dissociated
preservative as the first estimate of M∞. As discussed later this is appropriate for some
preservative systems but Le is a better value for others and in some cases, the model will
have to be refined further.
3. Results and discussion
3.1 Dissociation
The rates of dissociation of typical samples of treated wood are shown in Figure
1 (CCA), Figure 2 (ACQ) and Figure 3 (CA) along with the leaching of fine sawdust
properties discussed below. The equilibrium dissociation amounts for the preservatives
are shown in Tables 1-2. The dissolving or dissociation process requires some time to
reach equilibrium (Table 1). These times are relatively short for the copper amine and
borate treatments (1-3 days) but are up to several weeks for CCA components (especially
arsenic). This has implications for the leaching performance of these preservatives under
wetting and drying conditions, as it affects the time to reach maximum free component
concentration in the wood after the start of a rain event. There are also very significant
differences among the preservative systems in the extents that components dissolve in
the free water. The percent dissociation is low for CCA components, moderately low
for copper azole, higher for ACQ and very high (almost all of the preservative present)
for borate treatment (Figures 1 - 3 and Tables 1-2). This is obviously due to the
solubility properties of the less strongly fixed components of the formulations. CCA
reaction products have very low solubility in the almost neutral pH environment in wood
while in the ACQ treatment most of the poorly fixed copper can dissolve in the water in
the wood, possibly due to the presence of unreacted ethanolamine. Essentially all of the
borate in the wood can dissolve in the free water in wood as long as the temperature and
retention allow for complete solubility in the amount of available water.

40
As Cu Cr
35 As Leached Cu Leached Cr Leached

30
% Dissociation/Leached

25

20

15

10

0
0 48 96 144 192 240 288 336 384 432 480 528 576 624 672
Time, h

Figure 1 Comparison of rates and amounts dissociated (solid lines) with amounts
leached from sawdust (broken lines) for CCA-C components in southern pine – 4.3 kg/m3
CCA
Table 1: Summary of average leaching, dissociation and diffusion parameters (assuming M∞ =
Di) for southern pine treated commercially with a number of wood preservatives (standard
deviations in brackets)
Preserv- Retention Total leached Total Time to eq’m DL DT
ative (kg/m3) from sawdust Dissociated dissociation (10-6cm2/s) (10-6 cm2/s)
Le (%) Di (%) (days)
ACQ 1.82 CuO 34.0 (3.52) 28.1 (2.50) 2 1.9 0.15
CuA 1.21 CuO 26.2 (2.16) 9.3 (0.92) 1-2 1.4 0.12
DOT 2.28 B2O3 ~100 91.0 (4.2) 6-8 8.6 0.58
CCA-As 1.44 As2O5 36.0 (2.5) 6.6 (2.2) >30 0.60 0.05
CCA-Cu 0.86 CuO 19.8 (1.5) 5.1 (1.3) 15 0.43 0.043
CCA-Cr 2.01 CrO3 7.1 (0.44) 2.6 (0.74) 15-20 0.21 0.08

100
Cu B B Leached Cu Leached
90
80
% Dissociation/Leached

70
60
50
40
30
20
10
0
0 48 96 144 192 240 288 336 384 432 480 528 576 624 672 720

Time, h

Figure 2 Comparison of rates and amounts dissociated (solid lines) with amounts
leached from sawdust (broken lines) for Cu and B components of ACQ in southern pine -
1.82 kg/m3 CuO

40
Cu Cu Leached
35
% Dissociation/Leached

30

25

20

15

10

0
0 48 96 144 192 240 288 336 384 432 480 528 576 624 672 720
Time, h

Figure 3 Comparison of rates and amounts dissociated (solid line) with amounts
leached from sawdust (broken line) for Cu component of CA in southern pine - 1.21 kg/m3
CuO
 For most preservative components there is a moderate temperature effect on dissociation
with an increase in the amount of dissolved or dissociated preservative with increasing
temperature (Figure 4, Table 2).

Table 2: Temperature effects on dissociation and unidirectional diffusion coefficients in

southern pine
Preservative Retention T Total DL DRad. DTan Quality of
3 -7 2 -7 2 -7 2
(kg/m ) (°C) Dissociated (10 cm /s) (10 cm /s) (10 cm /s) fit
(%)
L R T
CCA-As 1.44 As2O5 5 4.4 7.5 0.43 0.08 P P P
22 7.0 11 1.8 0.19 M G M
34 8.7 54 1.8 0.09 G M G
CCA-Cu 0.86 CuO 5 4.2 1.1 * 0.03 M - M
22 3.7 4.4 0.29 0.009 M P G
34 3.9 1.1 0.24 0.034 G G P
CCA-Cr 2.01 CrO3 5 1.5 1.4 * 0.001 P - P
22 1.8 3.4 0.7 0.001 G P G
34 2.8 0.54 0.94 0.0002 G G G
ACQ 3.14 CuO 5 20 3.8 0.37 0.093 G G M
22 22 9.2 1.7 0.20 G M M
34 16 3.6 0.41 0.043 G P P
CuA 1.94 CuO 5 8 3.6 0.78 0.05 G G G
22 10 6.0 1.2 0.07 G G M
34 12 5.9 1.1 0.13 G G M
DOT 2.28 B2O3 5 75 3.5 1.1 1.3 G M M
22 100 7.2 3.2 1.8 M G G
34 100 5.0 2.3 2.1 G G M
* Coating failure M - Moderate; P - Poor; G – Good;

3.2 Leaching of fine sawdust

For all preservatives, the finely ground treated wood exposed to water lost a proportion of each
component rapidly, to an asymptotically approached maximum indicating that the remaining
component was highly leach-resistant (Figures 1-3 and Table 1) (DOT was an exception since
virtually all borate could be extracted). In practice, it is unlikely that leaching in service will
progress beyond this asymptote and its value provides an upper limit to the amount potentially
lost by leaching.
For CCA (Figure 1, Tables 2-3) it is evident that the amounts of the three components available
for leaching are much greater than the amounts dissociated. In this case, the concentration
gradient for diffusion (concentration of dissociated component) will be low resulting in a reduced
leaching rate, but there will be a “reservoir” of available preservative that will replace leached
components. The net effect should be characteristic slow leaching of CCA components over a
very long period of time.
 14

12
% Dissociation

10

6
10oC
4
20oC
2 30oC
0
0 48 96 144 192 240 288
Time, h

Figure 4: Effect of temperature on copper dissociation in CA treated southern pine -

1.94 kg/m3 CuO

As confirmed by numerous studies, the chromium was the least soluble component
and these results show that the ultimate amount of chromium available for leaching is
also relatively low (7.4 % of total at this retention). The copper component reached a
leaching equilibrium more rapidly than arsenic suggesting that at the retention evaluated
here, it will initially leach faster than arsenic, but the rate of leaching should decline
faster than for arsenic. The arsenic component appears to dissolve and leach slowly but
for a long time. The availability of arsenic to leaching is highly dependent on retention
in wood (Jin and Preston 1993) and species (Stevanovic-Janesic et al. 2001) so these
properties will depend greatly on these and other variables.
For ACQ, at the retention evaluated in Figure 2, both the copper and borate
components had amounts available for leaching almost identical to the amounts
dissolved in the free void space at equilibrium. This means that most of the readily
soluble components in the treated wood can dissolve in the free water present in the
wood. This creates a steeper concentration gradient during leaching, compared to CCA
components. This was also more or less confirmed for a range of species and ACQ
retentions (Table 1, last two columns). This high copper solubility compared to copper
in CCA is likely due to free amine dissolved in the cell lumen water.
At the CA retentions evaluated, the copper behaved more like the copper in CCA than
the copper in ACQ (Table 1-3 Figure 3). A much larger amount was available for
leaching than was dissolved in the free void space at equilibrium. This is surprising
since it is also a copper amine system. The difference is attributed to the lower copper
amine concentration compared to the ACQ system, and in fact when higher CA
retentions were evaluated, we observed a higher % dissociated (results not shown).
For the DOT treatments, virtually all of the borate was available both for leaching and
dissolved in the free water in the wood void space. However, the fraction of total B
dissolved in the wood cell lumens may be restricted at lower temperatures as seen in
Table 2 and at high DOT retentions. This creates a very steep concentration gradient to
drive diffusion of boron out of wood during leaching.
 It is likely that for borate and ACQ systems, any portion of the leachable component
that is not dissociated at any time will dissolve readily as soon as the wood is wetted and
some of the free component is leached out. For this reason we believe that for these
systems, the most appropriate estimate for M∞ in diffusion modelling is Le, the amount
of leachable component, which in these cases is approximately equal to Di, the
dissociated amount. In contrast, for CCA, the potentially leachable component is made
available very slowly. In a study not described here, we found that repeatedly
impregnating CCA treated wood with water and extracting the free solution by
centrifuge did not remove more than the original dissociated amount even after 6
extractions with a combined dissociation period of 16 days. For this reason we believe
that for most applications, the most appropriate estimate for M∞ in diffusion modelling
with CCA is Di, the equilibrium dissociated amount, except for cases of aggressive
leaching of small samples and for very long leaching exposures.

3.3 Unidirectional diffusion modelling

Examples of the rate of leaching of 20 mm samples in different directions and the best
analytical solution for the diffusion coefficient D assuming M∞ is the equilibrium
dissociated amount are shown in Figures 5-10. Estimated diffusion coefficients are
summarized in Tables 2 and 3.
The diffusion model describes leaching data very well for leaching of Cu, Cr and As
from CCA treated wood in the transverse direction (Figures 5, 6 and 7). At the end of
the leaching test, the amount of component leached is still much lower than the amount
dissociated in the wood so the amount dissociated is a good short term estimate of M∞.

Figure 5 Comparison of leaching values with best fit of diffusion model for copper
leaching from CCA treated southern pine – Transverse direction
Figure 6 Comparison of leaching values with best fit of diffusion model for chromium
leaching from CCA treated southern pine – Transverse direction

Figure 7 Comparison of leaching values with best fit of diffusion model for arsenic
leaching from CCA treated southern pine – Transverse direction

When more vigorous leaching occurs as for arsenic leaching along the grain in CCA
treated samples (Figure 8) the model underestimated the amount of leaching indicating
that use of Di for M∞ was too conservative and in such cases, the model has to take into
account the reservoir effect. For borate treated samples (Figures 9 and 10) the diffusion
model fits experimental values relatively well for both longitudinal and transverse
diffusion because the amounts dissociated are approximately equal to the amounts
available for leaching (100 % of retention).
Figure 8 Comparison of leaching values with best fit of diffusion model for arsenic
leaching from CCA treated southern pine – Longitudinal direction

Figure 9 Comparison of leaching values with best fit of diffusion model for borate
leaching from DOT treated southern pine – Radial direction

Figure 10 Comparison of leaching values with best fit of diffusion model for borate
leaching from DOT treated southern pine – Longitudinal direction
 The rate of leaching is known to depend on both temperature (Van Eetvelde et al.
1995a; Van Eetvelde et al. 1995b) and diffusion direction (Ra et al. 2001). Preliminary
results shown in Table 2 confirm this. D generally increases with temperature as
expected, although the effects are inconsistent and repeated studies are needed to
develop the temperature dependence relationship for each preservative type. Leaching is
much greater along the grain than transversely and in southern pine, leaching in the
radial direction is much greater than in the tangential direction e.g. Figure 11.

These results show some differences in the inherent diffusion characteristics of the
different preservative components and systems and more study and analysis is needed to
better understand the formulation and other variables that affect the inherent diffusion
coefficient.

For borate samples, D values were in the range observed by Ra et al. (2000) for boric
acid diffusion into southern pine wood.

Leaching of Cu from CA treated wood with time at

ambient condition.
14 Radial
Tangential
12
End
10
% Loss

8
6
4
2
0
0 100 200 300 400 500 600

Time, h

Figure 11: Example of directional effects on leaching rate for copper azole treated
southern pine (22°C)

4. Application
Application of an appropriate diffusion model to the loss of preservative from wood
by leaching offers the following advantages:
1. Diffusion models are science based and should describe the physical process
more consistently than empirical models. Extrapolation of results based on diffusion
theory should be more reliable than extrapolation of empirical models; furthermore, they
provide a theoretical basis to assist with the design of appropriate test methods.

2. The diffusion model corrects for effects of specimen dimensions. It should be

possible to use small scale short term laboratory tests to predict long term emission
characteristics.

3. Diffusion coefficients can be determined for variables of interest such as wood

direction (longitudinal vs. transverse), temperature, wood species, preservative
component and formulation and relationships defined (for example T dependence) that
would allow these variables to be mathematically formulated in a more comprehensive
model..

4. Diffusion equations can be solved for virtually any product geometry (pole,
board, block) and can be solved in three dimensions to describe practical leaching
scenarios. It may also be applied to shell treatments through use of the penetration depth
as the diffusion distance rather than the half thickness of the sample.

As an example of application of this model to a more real-to-life situation, the above

parameters (Table 1) were applied to estimate copper and borate leaching (Figure 12)
and the corresponding calculated emission flux rates (Figure 13) from the top surface of
southern pine decking (38mm) treated with CCA, ACQ and DOT and continuously
exposed to rainfall for a 1 year period. For this comparison, total leachable amounts of
copper and borate were used as estimates for M∞ although for Cu in CCA, leaching was
also predicted for Di = M∞ for comparison. The model predicts that virtually all of the
DOT will be lost after 1 year and that about 60% of the available copper will be lost
from ACQ and CA treated boards. The rate of leaching and flux of copper from CCA
treated wood is relatively low and expected to fall somewhere between the two estimates
based on Le and Di respectively.

Cu - ACQ Cu - CA
B - DOT Cu - CCA Le = 19.8%
Cu CCA Di = 5.1%
100
90
Predicted leached (%)

80
70
60
50
40
30
20
10
0
0 100 200 300 400
Time (days)

Figure 12: Predicted leaching rate and flux rate from the top surface of 38mm deck
board continuously exposed to rainfall
 100

80 ▲B
♦ Cu - ACQ
Flux (mg/m2/day

■ Cu - CA
60
- Cu - CCA (Le)
● Cu - CCA (Di)
40

20

0
0 100 200 300 400
Time (days)

Figure 13: Predicted preservative flux (mg/m2/day) from top surface of 38mm deck board
continuously exposed to rainfall. Assumed retentions: ACQ 1.45% m/m Cu; CA 0.97 % m/m Cu;
DOT 0.71 % m/m B; CCA 0.69 % m/m Cu.

5. Preliminary validation
At this point, we have not validated this approach using large scale samples and
different geometries exposed to continuous water exposure. However data were
available on the laboratory shower leaching of full size specimens of southern pine from
which the above small scale specimens were cut. In addition, natural exposure leaching
results were available for southern pine lumber samples CCA-treated in the laboratory to
similar retentions. While the defined boundary conditions for the above approach apply
to leaching from wood that is saturated with water and in continuous water contact, it
was of interest to evaluate the “robustness” of the approach by applying it to these larger
scale leaching tests.
In the first comparison, all of the above tested preservative systems were evaluated for
leaching characteristics of 38 X 140 X 1000 mm boards exposed on edge to alternating
misting water spray (5 cycles of: 48 hours alternating one hour spray with 3 hours rest,
(12 hours water exposure) followed by 48 hours freezing then 72h drying at 50°C).
Twenty liters of water were recirculated during each cycle at the rate of about 2.5 L/m.
To apply the model, it was assumed that the D and Di or Le values determined above
applied, the cumulative water spray time was equivalent to continuous soaking and all
leaching was in the transverse direction from the top and bottom of boards only (edge
and end effects ignored).
When M∞ was assumed to be Di, the amount dissociated, the diffusion model under-
estimated copper losses from both copper amine based treatments and especially from
CA (Figures 14-15). However, when M∞ was assumed to be Le, the amount ultimately
available for leaching, the fit was much better for both systems, suggesting that
undissolved but leachable copper in these systems readily dissociates as soon as some of
the copper is leached out. The model significantly underestimated borate leaching from
 DOT treated wood (Figure 16) which cannot be explained since it was assumed that Le =
Di = 100%; either the D values was underestimated or the effects of edge and end
leaching or repeated wetting and drying were much greater with this preservative.

Predicted Leach 1 (0.42%Cu) Predicted Leach 1 (0.42%Cu)

Leach 2 (0.37% Cu) Leach 3 (0.31% Cu)
Leach 2 (0.37% Cu) Leach 3 (0.31% Cu)
4.5
5
4

% Cu Leached
3.5 4
% Cu Leached

3
3
2.5
2 2
1.5
1
1
0.5 0
0 0 12 24 36 48 60
0 12 24 36 48 60
Leaching Exposure (h)
Leaching Exposure (h)

(a) (b)
Figure 14: Predicted vs. measured leaching from 38 X 140 X 1000 mm ACQ treated
boards in laboratory shower test – (a) M∞ = Di (28%) (b) M∞ = Le (35%)

Predicted Leach 1 (0.31% Cu) Predicted Leach 1 (0.31% Cu)

3 3
2.5 2.5
% Cu Leached

% Cu Leached

2 2

1.5 1.5
1 1
0.5 0.5
0 0
0 12 24 36 48 60 0 12 24 36 48 60
Leaching Exposure (h) Leaching Exposure (h)

(a) (b)
Figure 15: Predicted vs. measured leaching from 38 X 140 X 1000 mm CA treated boards
in laboratory shower test – (a) M∞ = Di (10%) (b) M∞ = Le (26%)
 Predicted Leach 1 (0.095% B)
Leach 2 (0.39% B) Leach 3 (0.37% B)

70
60
% B Leached

50
40
30
20
10
0
0 20 40 60 80 100 120

Leaching Exposure (h)

Figure 16: Predicted vs. measured leaching from 38 X 140 X 1000 mm DOT treated
boards in laboratory shower test – M∞ = Di = Le = 100%

In the case of the copper component in CCA (Figure 17), the model under-estimated
copper leaching when M∞ was assumed to be Di the amount dissociated but over-
estimated leaching when M∞ was assumed to be Le, the amount ultimately available for
leaching. This suggests that the dissolving of copper during the leaching process is slow
and its kinetics must be considered in the leaching model for this system.

Predicted Leach 1 (0.18% Cu)

Leach 2 (0.13% Cu) Predicted Leach 1 (0.18% Cu) Leach 2 (0.13% Cu)

1.2
0.7
0.6 1
% Cu Leached
% Cu Leached

0.5 0.8
0.4
0.6
0.3
0.2 0.4

0.1 0.2
0
0
0 12 24 36 48 60
0 12 24 36 48 60
Leaching Exposure (h) Leaching Exposure (h)
(a) (b)

Figure 17: Predicted vs. measured Copper leaching from 38 X 140 X 1000 mm CCA
treated boards in laboratory shower test – (a) M∞ = Di = 4 % (b) M∞ = Le
= 20 %

Arsenic losses were in the order of the predicted value and chromium losses were over-
estimated by the model when M∞ was assumed to be Di (Figures 18-19). It appears that
the slow dissociation of these components during the rewetting phase of each cycle
reduced leaching.
 Predicted Leach 1 (0.24% As) Leach 2 (0.19% As)

0.5

0.4

% As Leached
0.3

0.2

0.1

0
0 12 24 36 48 60

Leaching Exposure (h)

Figure 18: Predicted vs. measured Arsenic leaching from 38 X 140 X 1000 mm CCA
treated boards in laboratory shower test - M∞ = Di = 5.5 %

Predicted Leach 1 (0.26% Cr) Leach 2 (0.20% Cr)

0.14

0.12
% Cr Leached

0.1

0.08

0.06

0.04

0.02

0
0 12 24 36 48 60

Leaching Exposure (h)

Figure 19: Predicted vs. measured Chromium leaching from 38 X 140 X 1000 mm
CCA treated boards in laboratory shower test - M∞ = Di = 1.8%

Thus, application of this model to a laboratory leaching test that involved wetting and
drying and other deviations from the model boundary conditions resulted in relatively
good estimates of leaching in some cases but discrepancies in others. These variances
from predicted values are not consistent among preservative components, suggesting that
there may be additional preservative specific phenomena that affect leaching which are
not considered in the model. It will be important to identify these to have a better
understanding of the leaching phenomenon.
For a second comparison, the leaching results for horizontal 38 X 140 mm end-sealed
southern pine boards treated with 1% CCA and exposed to natural weathering in
Toronto, Canada for one year (Taylor 2001) were compared to model predictions. In this
case, in addition to cumulative rainfall amount, the duration of all rain events was
recorded with a tipping bucket device. For application of the model, cumulative rain
exposure time was used in the model. It was also assumed that all leaching was across
the grain from the top surface (edge leaching and leaching from the bottom was ignored.
The effects of ambient T, UV degradation and drying between rain events were ignored.
These results are shown for three replicate boards in Figures 20 – 22. As expected,
large differences between predicted and actual losses are seen. Copper losses were
significantly under-estimated when M∞ was assumed to be Di, but closer conformance
was seen when M∞ was assumed to be Le. Chromium estimates were reasonably close
to measured values and arsenic levels (using M∞ = Di) were consistently under-
estimated. Figure 22 demonstrates how other events (in this case the splitting of the
board) can have a large and unpredictable effect on leaching.

Cr Actual As Actual Cu Actual

Cr Predicted As Predicted Cu Predicted (Di)
Cu Predicted (Le)
1.60

1.20
% Leached

0.80

0.40

0.00
0 100 200 300 400
Cumulative rain (h)

Figure 20: Predicted vs. measured CCA component losses for southern pine exposed
to natural weathering in Toronto for one year – Replicate 1

Cr Actual As Actual Cu Actual

Cr Predicted As Predicted Cu Predicted (Di)
Cu Predicted (Le)
1.60

1.20
% Leached

0.80

0.40

0.00
0 100 200 300 400
Cumulative rain (h)

Figure 21: Predicted vs. measured CCA component losses for southern pine exposed
to natural weathering in Toronto for one year – Replicate 2
 Cr Actual As Actual Cu Actual
Cr Predicted As Predicted Cu Predicted (Di)
Cu Predicted (Le)

3.00
2.50
% Leached

2.00
1.50
1.00
0.50
0.00
0 50 100 150 200 250 300 350 400 450
Cumulative rain (h)

Figure 22: Predicted vs. measured CCA component losses for southern pine exposed
to natural weathering in Toronto for one year – Replicate 3

6. Some Implications and Limitations of this Approach:

1. It appears that the dissociation characteristics have several important
implications:
a. The concentration gradient between the dissolved component in the cell
lumens and the reduced concentration at the surface during a leaching
event provides the concentration gradient driving the diffusion of
components to the wood surface where they are leached away.
b. In some cases, equilibrium dissociated concentration in the cell lumens is
an indicator of total available preservative component for leaching (see
below).
c. Rate of dissolving may be a significant factor in intermittent water
exposure conditions where wood dries between exposures.
2. The models assume that wood is fully saturated with water (dissociation is
complete) and the surfaces are in continuous contact with water. Refinements to
the model are needed to account for wetting and drying during natural exposure
and the effects of moisture content changes on rates of solution of components.
3. At this time, the effect of low solubility of some components of CCA
components relative to their ultimate availability is not taken into account in the
simple diffusion models presented here and the long term leaching losses of these
components may be under-estimated if the dissociated amount is used in the
analysis and the rate of leaching may be over-estimated if the total availability is
used.
4. If the above principles are accepted, it is clear that prediction of emissions should
be based on time of rain and other water exposure rather than intensity of
exposure. This is validated by studies that show that more leaching occurs per
 unit rainfall by slow steady rainfall than by an intense down-pour (Taylor 2001)
and that the amount of leaching occurring over a fixed leaching time is
practically independent of intensity of rainfall (Klipp et al. 1991). Unfortunately,
it is more common to record meteorological information as mm rainfall rather
than duration of actual precipitation.
5. As generally assumed, leaching is much faster longitudinally than transversely.
These results show that at least for southern pine, radial diffusion (leaching) is
much higher than tangential leaching. This means that flat sawn boards will
leach at a greater rate than quarter-sawn boards and may explain some of the well
known variability in leaching rates between boards of the same treatment and
species.
6. The procedure should also be applicable to leaching of organic wood
preservatives but it does not predict losses from volatilization or bleeding and
these factors would have to be considered in a model for organic substances.
7. No approach, including this one can predict the effects of natural exposure
conditions that may affect leaching in an unusual way such as UV exposure,
exposure to organic and other acids and other aggressive leaching media, effects
of coatings, water repellents etc.

7. Summary and Conclusions

An approach is shown to estimate leaching impacts from inorganic preservative

treated wood based on laboratory estimation of the amount of preservative component
available for leaching, the equilibrium dissociation of preservative into free water in
wood and diffusion coefficients in different wood directions. Combining this
information with a simple diffusion equation allows the estimation of potential risk from
leaching over a wide range of preservatives, product dimensions and specified
conditions.
The results predict that ACQ and CA will have higher initial leaching rates compared
to CCA but the latter preservative will continue to leach copper and arsenic at a
measurable rate for a much longer time. As expected, leaching is much faster
longitudinally than transversely and in pine, radial diffusion (leaching) is much higher
than tangential leaching. Higher temperature increases both amount of preservative
dissociated and diffusion rates, but these effects are relatively minor.
More work is needed to investigate the applicability of this approach with particular
emphasis on:
• Generating more reliable and consistent values for directional diffusion
coefficients in most commonly treated species and defining the temperature dependence
more precisely.
• Investigating the rate of replenishment of leached material by continued
dissociation and integrating this into a diffusion model;
• Evaluating effects of rates of wetting and drying of wood between water
exposures and their impacts on leaching;
• Developing models for three-dimensional diffusion in rectangular and pole
shaped samples.

References
Brooks, K.M. (2001) “Literature review and assessment of the environmental risks
associated with the use of ACQ treated wood products in aquatic environments”. Report
prepared for: WWPI, Vancouver, WA.
Hingston, J.A., C.D. Collins, C.D., Murphy, R.J. and Lester, J.N. (2001) “Leaching of
chromated copper arsenate wood preservatives: a review”. Environmental Pollution 111,
53-66

Homan, W.J. and van Oosten, A.L. (1999) “Statistically stable models for determination
of PEC”. . Int. Res. Group on Wood Preservation, Doc. IRG/WP 99-50135.

Lebow, S. (1996) “Leaching of wood preservative components and their mobility in the
environment. Summary of pertinent literature”. General Technical Report. Forest
Products Laboratory, USDA Forest Service, Madison Wisconsin. FPL-GTR-93:36.

Jin, L. and Preston, A.F. (1993) “Depletion of preservative from treated wood: Results
from laboratory, fungus cellar and field tests”. Int. Res. Group Wood Preserv. Doc.
IRG/WP 93-50001.

Klipp, H., H. Willeitner, K. Brandt and A. Muller-Grimm. (1991) “Migration of active

ingredients from treated timber into fresh water”. Int. Res. Group on Wood
Preservation, Doc. IRG/WP/3669.

Morsing, N. (2003) “Leaching of active components from preservative treated timber.

Status after approx. 4 months out-door exposure”. Int. Res. Group on Wood
Preservation, Doc. IRG/WP 03-20276.

Morsing, N. and B. Lindegaard (2004) “Leaching of active components from

preservative treated timber. Stages 2 & 3: Laboratory testing and comparison with semi-
field testing”. Int. Res. Group on Wood Preservation, Doc. IRG/WP 04-20303
OECD. 2003. OECD series on emission scenario documents, Number 2. Emission
Scenario Document for Wood Preservatives. PART 3. Organisation for Economic Co-
operation and Development, Paris, France.

Ra, J.B. Barnes, H.M. and Conners, T.E. (2001) “Determination of boron diffusion
coefficients in wood”. Wood and Fiber Science. 33(1), 90-103.

Schoknecht U, Wegner R, Horn W, Jann O. 2003. Emission of biocides from treated

materials: test procedures for water and air. Environ Sci Pollut Res Int. 10(3):154-61.

Schoknecht, U., R. Wegner and E. Melcher. 2004. Experiences with the OECD guideline
proposals for the estimation of emissions from preservative treated wood in the
environment Int. Res. Group on Wood Protection. Doc. IRG/WP/04-50209.

Stevanovic-Janesic, T., Cooper, P.A. and Ung, Y.T. (2001) “Chromated copper arsenate
treatment of North American hardwoods. Part 2. CCA leaching performance”.
Holzforschung. 55(1):7-12.

Taylor, J. (2001) “Natural leaching of CCA from treated lumber”. M.Sc.F. Thesis,
Faculty of Forestry, University of Toronto.

Van Eetvelde, G., Stevens, M. and Vander Mijnsbrugge, L. (1994) “Comparative study
on leaching of CCA from treated timber: Modeling of emission data”. Int. Res. Group
Wood Preserv. Doc. IRG/WP 04-50027.
Van Eetvelde, G., Stevens, M., Mahieu, F., Wegen, H.-W., Platen, A. (1998) “An
appraisal of methods for environmental testing of leachates from salt-treated wood; part
1” The International Research Group on Wood Preservation, IRG/WP 98-50115.

Van Eetvelde, G., Orsler, R., Holland, G., Stevens, M. (1995a) “Effect of leaching
temperature and water acidity on the loss of metal elements from CCA treated timber in
aquatic applications. Part 1. Laboratory scale investigation”. Int. Res. Group on Wood
Preservation. Doc. IRG/WP/95-50046.

Van Eetvelde, G., Homan, W.J., Militz, H., Stevens, M. (1995b) “Effect of leaching
temperature and water acidity on the loss of metal elements from CCA treated timber in
aquatic conditions. Part 2: semi-industrial investigation”. Int. Res. Group on Wood
Preservation. Doc. IRG/WP 95-50040.

Willeitner, H., and Peek, R.-D. (1998) “How to determine what is a realistic emission
from treated wood”. The International Research Group on Wood Preservation.
Stockholm, Sweden. Document IRG/WP/98-50105.