 Original article
 Published:
Solute diffusion into cell walls in solutionimpregnated wood under conditioning process III: effect of relative humidity schedule on solute diffusion into shrinking cell walls
Journal of Wood Science volume 63, pages 263–270 (2017)
Abstract
This study has focused on solute diffusing into cell walls in solutionimpregnated wood under conditioning, process of evaporating solvent. The purpose of this paper was to clarify the RH (relative humidity) schedule that promotes the solute diffusion into shrinking cell walls during conditioning. The wood samples impregnated with a 20 mass% aqueous solution of polyethylene glycol (PEG1540) was conditioned with a temperature of 40 °C to the equilibrium point at the RH where the samples swelled maximally. The samples were subsequently conditioned at 40 °C under the schedules including four ways of RHdecrease steps where the cell walls shrunk. The amount of solute (PEGs) diffused into cell walls during the conditioning logarithmically increased with increasing the number of the RHdecrease steps. This was well explained by the theoretical model that describes the solute diffusion into shrinking cell walls. It is clarified from the model that the RH, or moisture content of the sample, should be decreased as gradually as possible to increase the total amount of diffused solute into shrinking cell walls, and that the amount of diffused solute is smaller for the lower moisture content. The model also suggests that effect of change in RH schedule on change in total amount of diffused solute does not depend on solute diffusivity in the sample under drying in a vacuum over phosphorous pentoxide, and that impregnated wood should be conditioned under natural convection rather than forced convection for promoting the diffusion into shrinking cell walls.
Introduction
Chemical treatment of wood is one of the major methods to change its properties such as biodegradability, combustibility, and dimensional instability (for example, [1]). In the treatments, the chemical substance has been developed and introduced into wood in order to control these properties that disturb the intended usage of wood [2].
For the effective improvement especially of the dimensional stability, the chemical substance is required to exist in wood cell walls rather than in cell cavities (for example, [3]). In the general treatment process, wood is impregnated with a solution of the chemical substance and is subsequently conditioned in the atmosphere to evaporate the solvent from the impregnated wood. Our previous study has shown that the amount of chemical substance, or solute, in cell walls after the impregnation process is smaller than that after the conditioning process [4].
The solute diffusion into cell walls was suggested to be driven by the higher solute concentration in cell cavities than that in cell walls, which is caused by the higher evaporation rate of solvent from cell cavities than that from cell walls [5]. The amount of the solute diffusion is considered to be increased by increasing the following two factors. The first factor is the difference in solute concentration between cell cavities and cell walls. The second factor is the solute diffusivity into cell walls [6], which is considered to increase with increasing moisture content [7] and relative swelling [6] of the impregnated wood. It should be noticed that both the concentration difference and the solute diffusivity may be affected by the atmospheric state such as relative humidity (RH) and temperature during the conditioning.
The objective of this study was to determine the atmospheric states that maximize the amount of solute diffused into cell walls during conditioning process. In the previous studies [6–8], we have examined the effect of RH during the conditioning on the solute diffusion into cell walls. In these studies, crosscut wood samples were used since they are favorable for causing the soluteconcentration difference between cell walls and cell cavities. As solute and solvent in these studies, polyethylene glycols (PEGs) and water, respectively, were employed, because the amount of water or PEGs in cell walls and cell cavities can be related to mass or dimension, respectively, of the impregnated wood [9–21]. Most of PEGs are confirmed to diffuse into cell walls during conditioning [4], though the cell walls impregnated with aqueousPEG solution absorb far smaller amount of PEGs [22]. The RH schedule in our previous study [6–8] is exemplified in Fig. 1. In the first term, the impregnated wood was conditioned at several levels of RHs. The solute (PEGs) was clarified to diffuse maximally into cell walls during the conditioning at the RH causing the cell walls to swell maximally. This is considered to have its origin in the following two relations. The first relation is the increase in the solute diffusivity with increasing the region into which solute can diffuse and replace water that diffuses into cell cavities. The second relation is the increase in this region with swelling of cell walls. If the cell walls swelled maximally in the first term, however, the RH schedule that promotes solute diffusion into shrinking cell walls after the first term (Fig. 1) has not yet been examined.
The purpose of this paper was to clarify the RH schedule that promotes solute diffusion into shrinking cell walls after the first term (Fig. 1). The solute diffusivity was surmised to decrease with both shrinking of cell walls and decreasing of solvent (water) around the solute. In addition, it may take time for the solute to diffuse into cell walls. Therefore, the solute diffusion was expected to be promoted if the water evaporation, or RHdecrease, was as slow as possible so that the solute gained time to diffuse. It was, however, difficult to control experimentally and to analyze theoretically the continuous decrease in RH while monitoring the solute diffusion. The RH, therefore, should be decreased stepwise in the schedule.
In this paper, crosscut wood samples were impregnated with a solution including PEGs and water as a solute and solvent, respectively. These samples were conditioned at five levels of RH to confirm the RH causing the samples to swell maximally. The other groups of aqueousPEG solutionimpregnated wood samples were conditioned at the confirmed RH to the equilibrium point. These were subsequently conditioned in four schedules including 1, 2, 3, or 5 steps of RHdecrease. The theoretical model that describes solute diffusion into shrinking cell walls when RH decreases stepwise is constructed. The model was used to predict qualitatively the relation between the number of RHdecrease steps and the amount of diffused solute during conditioning. The model was also used to discuss the effect of RH schedule in general, such as continuous decrease in RH, on solute diffusion into shrinking cell walls.
Theory
The model that describes solute diffusion into shrinking cell walls is constructed here to predict qualitatively the relation between the number of RHdecrease steps, n, and the amount of diffused solute during conditioning. The model also can be used to discuss effect of RH schedule in general, such as continuous decrease in RH, on the solute diffusion. The solutionimpregnated wood under conditioning swells with decreasing RH down to a level, and subsequently shrinks with decreasing RH to the further lower level, when PEG and water are employed as solute and solvent, respectively [6]. In this paper, the periods of shrinking are called the ith terms in the schedule including the n times of RHdecrease (i = 2, ···, n + 1).
Assumption
To simplify the phenomenon, the following assumptions, compared to the experiment in “Results and discussion,” are employed.

1.
Moisture content of impregnated wood under conditioning, M, affected by stepwise decrease in the RH, approximately shows stepwise decrease in order of \({{M}_{\text{1}}},{{M}_{\text{2}}},\ldots ,{{M}_{i}},\ldots\), and M _{ n+1} (= 0).

2.
Solute diffuses from a point representing cell cavities to another point representing cell walls.

3.
The amount of solute diffused during a unit time at the ithe term, dA _{ i }/dt, is approximately equal to the product of the solute diffusivity into cell walls, k _{ i }, and the soluteconcentration difference between cell cavities and cell walls, Δx _{ i }, or the approximate equation of dA _{ i }/dt = k _{ i }Δx _{ i } consists.

4.
The diffusivity, k _{ i }, is approximately in direct proportion to M _{ i } + μ, where M _{ i } represents moisture content at the ithe term and μ is a constant directly proportional to the solute diffusivity in impregnated wood with moisture content of M _{ n+1} (= 0).

5.
The soluteconcentration difference just after the beginning of the ith term, Δx _{ i }, is approximately in direct proportion to the decrease in the moisture content during the same term, ΔM _{ i } (= M _{ i−1} − M _{ i }).

6.
The difference, Δx _{ i }, approximately shows exponential decrease after the beginning of the ith term, or Δx _{ i } is approximately in direct proportion to exp(–t/τ), where τ represents relaxation time independent on i.

7.
The solute diffusion is almost finished till the end of the ith term so that Δx _{ i } decreases to be close to zero.
Theoretical equation deduced from assumption
On the basis of the assumptions 1–6, the amount of solute diffused from cell cavities to cell walls during a unit time at the ithe term (i = 2, ···, n + 1), dA _{ i }/dt, is formulated as follows:
where C represents normalizing coefficient.
The amount of solute diffused during the ithe term, ΔA _{ i }, is obtained by time integration of the Eq. (1) as follows:
where Δt _{ i } represents the duration time of the ith term, and can be regarded as infinity based on the assumption 7, leading to the following equation.
If the impregnated wood is placed under the RH schedule causing its moisture decrease as shown in Fig. 2 in the conditioning (\(n\,=\,\text{2},\text{ 3},\ldots\)), the total amount of solute diffused after the first term, A _{ n }, is formulated as follows:
where M _{ n + 1} and \(\sum\nolimits_{i=2}^{n+1}{\Delta {{M}_{i}}}\) are equal to zero and M _{1}, respectively, on the basis of the assumption 1.
On the other hand, if the impregnated wood just after the firstterm conditioning is placed in vacuumdried state (n = 1), ΔM _{ i } (\(i\,=\,\text{2},\ldots ,n\)) is equal to zero in the Eq. (4), on the basis of the assumption 1. Thus, the total amount of the diffusion, A _{1}, is represented as follows:
Qualitative prediction using theoretical equation
For qualitative prediction, the decrease in moisture content during each term is assumed to be equal to each other. The decrease in moisture content, ΔM _{ i }, and equilibrium moisture content, M _{ i }, during the ith term (i = 2, ···, n + 1, and n = 1, 2, ···) are thereby represented as follows:
The amount of solute diffused during the ithe term, ΔA′ _{ i }, and their total amount after the first term are predicted using the Eqs. (3), (4), (6), and (7) as follows:
To normalize the Eqs. (3), (4), (8), and (9), the coefficient, C, is determined as follows:
Figure 3 shows the relation between the number of RHdecrease steps, n, and the total amount of diffused solute during conditioning, Aʹ_{ n }, predicted using the Eqs. (9) and (10). With increasing the number of steps, n, the total amount, Aʹ_{ n }, is qualitatively predicted to increase logarithmically.
Materials and methods
Sample preparation and impregnation process
Twentyseven crosscut samples, with dimensions of 5 mm × 25 mm × 25 mm in longitudinal (L), radial (R), tangential (T) directions, respectively, were subsequently prepared from a block of hinoki (Chamaecyparis obtusa). The samples were impregnated with deionized water and left in it for 2 months, and their mass and area in RT cross section (m _{A} and s _{A}, respectively) were measured. They were conditioned at 60% RH (20 °C) for over 2 weeks and subsequently dried at 105 °C to relatively constant mass, m _{O}, their RT crosssectional area, s _{O}, and volume, v _{O}, and their average ovendry density was 0.379 ± 0.003 g/cm^{3}. The dried samples were impregnated with an aqueous solution of polyethylene glycol (PEG1540, E.P., Wako) with concentration of 20 mass%, and left in the solution for 5 months.
Conditioning process
The closed containers with a diameter of 88 mm and a height of 109 mm inside were prepared in the chamber (Standard industrial oven, PV211, Espec Co.) with a temperature inside controlled at 40 °C. The temperature in this study was different from that in the previous studies, 35 °C [4, 6–8]. This was because the chamber employed in this study was changed from that in the previous studies, and because the temperature in the chamber in this study was more stable when it was controlled at 40 °C than at 35 °C under a room temperature of 20–25 °C. Atmosphere inside the container was controlled at RH of 11, 32, 53, 66, 75, or 89% with a temperature of 40 °C using saturated aqueous solution of lithium chloride (LiCl), magnesium chloride (MgCl_{2}), sodium bromide (NaBr), potassium iodide (KI), sodium chloride (NaCl), or potassium nitrate (KNO_{3}), respectively [23], over which the impregnated samples can be hanged over to be conditioned.
Three impregnated samples were conditioned at the RH of 11, 32, 53, 75, or 89% (40 °C) for 2.17 × 10^{3} h, followed by drying in a vacuum over phosphorous pentoxide (P_{2}O_{5}). The RT crosssectional area, s(t), as a function of conditioning time, t, was measured, and the relative swelling, r(t), was calculated using the following equation.
Mass of the samples after dried in a vacuumed state, m _{V}, was measured to calculate weight percent gain (WPG) of the dried samples using the following equation.
The temporal variability of r(t) is shown in Fig. 4. The sample was confirmed to show maximum swelling when conditioned at RH of 75%. The values of W _{V} were 39.9 ± 0.3, 40.1 ± 0.2, 39.4 ± 0.3, 39.3 ± 0.4, and 39.5 ± 0.6% for RH of 11, 32, 53, 75, and 89%, respectively. This confirms that the total amount of PEGs in each sample after the conditioning is not so different among all RHs.
To clarify the effect of number of steps of RHdecrease when impregnated wood was conditioned during a unified period, the other sets of twelve impregnated samples were conditioned at the confirmed RH of 75% (40 °C) for 2.18 × 10^{3} h, and subsequently conditioned at 40 °C in four schedules including 1, 2, 3, or 5 steps of RHdecrease, called schedule A, B, C, or, D, respectively (Fig. 5). The schedule A contained RH of 75% for 1.28 × 10^{3} h, B RH of 53% for 1.28 × 10^{3} h, C RH of 66% for 4.33 × 10^{2} h and 32% for 8.43 × 10^{2} h, and D RH of 66% for 1.00 × 10^{2} h, 53% for 3.34 × 10^{2} h, 32% for 4.82 × 10^{2} h, and 11% for 3.61 × 10^{2} h, where the final step in all the schedules was drying in a vacuum over P_{2}O_{5}. The mass and RT crosssectional area during the conditioning, m(t) and s(t), respectively, and after the conditioning, mv and s _{V}, respectively, were measured. The relative swelling, r(t), and the moisture content, M(t), of the impregnated samples under conditioning are calculated using the Eq. (11) and the following equation, respectively.
The relative swelling and the WPG after dried in a vacuumed state, r _{V} and W _{V}, were calculated using the following equation and Eq. (12), respectively.
The values of W _{V} were 39.8 ± 0.2, 39.6 ± 0.4, 40.1 ± 0.1, and 39.9 ± 0.7% for schedules A, B, C, and D, respectively. This confirms that the total amount of PEGs in each sample after the conditioning is not so different among all schedules.
Results and discussion
Effect of number of RHdecrease steps on total solute diffusion
Figure 6 shows the relation between the number of RHdecrease steps, n, and the relative swelling of the samples after the conditioning, r _{V}. The value of r _{V}, which is proportional to the amount of solute (PEGs) diffused into cell walls during conditioning [7], logarithmically increased with increasing the number of steps, n. This trend showed good agreement with the qualitative prediction (Fig. 3).
Relationship between theory and experiment
To compare the assumption 1 to the experiment, temporal variability of moisture content of the sample during conditioning, M(t), is shown in Fig. 7. The value of M(t) decreased stepwise after the stepwise decrease in the RH (Fig. 5). This indicates that the moisture content approximately shows stepwise decrease, affected by the stepwise decrease in RH. This suggests the assumption 1.
The assumptions 2, 3, and 5 have already been compared to the experiments in our previous studies [4, 6–8].
To compare the assumptions 6 and 7 to the experiment, temporal variability of relative swelling of the sample during conditioning, r(t), is shown in Fig. 8. The value of r(t) is known to increase with the total amount of substance in cell walls [9–21], and thus the variation of r(t) implies the migration of substance into and out of cell walls. The value of r(t) decreased after the RH decreased for each term. This value reached to the equilibrium point after the moisture content, M(t), reached to the equilibrium point for each term (Figs. 7, 8). These findings indicate that it takes time for the soluteconcentration difference caused by the water evaporation to disappear, and that the difference disappears to become zero till the end of each term. This suggests the assumptions 6 and 7.
The assumption 4 is related to the mobility of the solute molecules, which increase with increasing the amount of solvent around solute and with increasing the dimension of the cell walls or the samples [6, 7]. In the conditioning after the first term, the samples did not swell but shrunk (Fig. 8), indicating the solute diffusivity decrease with decreasing the moisture content. This suggests the assumption 4. At the beginning of the term of 11% RH conditioning around 3100 h for schedule D (Fig. 8), the samples shrunk rapidly followed by a moderate swelling. In contrast, the samples did not swell during the term of RH over 11%. This was probably because it took longer time for the solute under the 11% RH term than the other terms to diffuse into cell walls after the solvent exudation from cell walls, where the solvent exudation from and solute diffusion into cell walls caused the shrinkage and swelling, respectively, of the samples. This supports the assumption 4.
To compare the theoretical equations to the experiment, the total amount of diffused solute during conditioning for each number of RHdecrease steps, n, was simulated using the Eqs. (4) and (5), and the equilibrium moisture content, M _{ i }, and moisture decrease, ΔM _{ i }, during the ith term (Table 1) estimated using the moisture content in Fig. 7. Figure 9 shows the relation between the simulated total amount of solute diffused into cell walls, A _{ n }, and the relative swelling, r _{V}, which is proportional to the amount of diffused solute during conditioning after the first term. The simulated value, A _{ n }, strongly correlated with the experimental value, r _{V} (R ^{2} = 0.988). This indicates that the theoretical equations can explain the total amount of solute diffused into cell walls after the first term.
Suggestions obtained from theoretical model
The theoretical model represented by Eqs. (4) and (5) can simulate the total amount of diffused solute using the moisture content of the sample. This suggests that the total amount of the diffused solute during conditioning can be simulated by the temporal variability of moisture content of impregnated wood.
In the simulated total amount of diffused solute, A _{ n } (n = 1, 2,...), in Eqs. (4) and (5), the term as a function of the number of RHdecrease, n, does not contain the constant μ, which is directly proportional to the solute diffusivity in impregnated wood dried to zero moisture content. This suggests that effect of change in RH schedule on change in total amount of diffused solute does not depend on solute diffusivity in the sample under drying in a vacuum over P_{2}O_{5}.
The relation of the amount of diffused solute, ΔA _{ i }, to equilibrium moisture content, M _{ i }, during the ith term, is represented as Eq. (15), if the decrease in moisture content during each term is assumed to be equal to each other, or the Eqs. (7), (8), and (10) consist.
This equation suggests that the amount of diffused solute is smaller for the lower moisture content.
To satisfy the assumption 7, the soluteconcentration difference is required to disappear due to the solute diffusion till the end of each term. If the RH decreases continuously in the schedule, the quasistatic state in which the solute diffusion constantly finishes is required to be maintained during conditioning to maximize the total amount of the solute diffusion. Although the quasistatic state is hardly realized, this is close to the state in which the time until the moisture content reaches to the equilibrium point is far longer than the relaxation time of the concentration difference, τ, on the basis of the assumption 6. This suggests that moisture content of sample, or RH, should be decreased as gradually as possible to increase the total amount of diffused solute into cell walls.
The convection type as well as the RH is an important factor as the atmospheric state of conditioning, because the evaporation rate of water under forced convection is usually higher than that of natural convection, affecting the solute diffusivity. In the first term, the forced convection is confirmed in our previous study [8] to promote the solute diffusion into swelling cell walls. After the first term, the natural convection is suggested to promote the solute diffusion into cell walls, because the gradual evaporation of water promotes the diffusion as suggested above.
Conclusion
The purpose of this paper was to clarify the RH schedule that promotes solute diffusion into shrinking cell walls after the first term.
Crosscut wood samples impregnated with a 20 mass% aqueous solution of PEG1540 were conditioned to the equilibrium point at RH of 75% (40 °C) where the samples swelled maximally. The samples were subsequently conditioned at 40 °C under the RHs scheduled in four ways of RHdecrease steps where the cell walls shrunk.
The amount of solute (PEGs) diffused into cell walls during the conditioning, estimated using the relative swelling of the conditioned samples, logarithmically increased with increasing the number of the RHdecrease steps. This was well explained by the theoretical model that describes the solute diffusion into shrinking cell walls. The assumptions in the model were compared to the experiments. The following suggestions were obtained from the model when impregnated wood shrinks during conditioning after the first term.

1.
Moisture content of impregnated wood under conditioning, or RH, should be decreased as gradually as possible to increase the total amount of diffused solute into shrinking cell walls.

2.
The amount of the diffused solute is smaller for the lower moisture content.

3.
Effect of change in RH schedule on change in total amount of diffused solute does not depend on solute diffusivity in the sample under drying in a vacuum over phosphorous pentoxide.

4.
For promoting solute diffusion into shrinking cell walls, impregnated wood should be conditioned under natural convection rather than forced convection.
To investigate the applicability of the proposed model is the important challenge to be solved.
References
Inoue M, Ogata S, Nishikawa M, Otsuka Y, Kawai S, Norimoto M (1993) Dimensional stability, mechanical properties, and color changes of low molecular weight melamineformaldehyde resin impregnated wood (in Japanese). Mokuzai Gakkaishi 39:181–189
Norimoto M, Grill J (1993) Structure and properties of chemically treated wood. In: Shiraishi N, Kajita H, Norimoto M (eds) Recent research on wood and woodbased materials, Current Japanese materials research, vol 11, Elsevier Applied Science, London, pp 135–154
Furuno T, Goto T (1978) Structure of the interface between wood and synthetic polymer. XI. The role of polymer in the cell wall on the dimensional stability of wood polymer composite (WPC). Mokuzai Gakkaishi 24:287–293
Tanaka S, Seki M, Miki T, Shigematsu I, Kanayama K (2016) Solute diffusion into cell walls in solutionimpregnated wood under conditioning process II: effect of solution concentration on solute diffusion. J Wood Sci 62:146–155
Stamm AJ (1956) Dimensional stabilization of wood with carbon waxes. For Prod J 6:201–204
Tanaka S, Miki T, Seki M, Shigematsu I, Kanayama K (2015) Mechanism verification of solute diffusion into cell walls in solution impregnated wood under conditioning process: effect of relative humidity on swelling and shrinkage behavior of wood impregnated with an aqueous solution of polyethylene glycol (in Japanese). J Soc Mater Sci Jpn 64:369–374
Tanaka S, Seki M, Miki T, Shigematsu I, Kanayama K (2015) Solute diffusion into cell walls in solutionimpregnated wood under conditioning process I: effect of relative humidity on solute diffusivity. J Wood Sci 61:543–551
Tanaka S, Seki M, Miki T, Shigematsu I, Umemura K, Kanayama K (2016) Effect of solvent evaporation rate on solute diffusion into cell walls in solutionimpregnated wood under conditioning process (in Japanese). J Soc Mater Sci Jpn 65:359–364
Stamm AJ (1964) Dimensional stabilization, wood and cellulose science. Ronald Press, New York, pp 312–342
Ishimaru Y (1993) PEG to mokuzai no hanashi (in Japanese). Mokuzaihozon 19:204–218
Jeremic D, Cooper P, Heyd D (2007) PEG bulking of wood cell walls as affected by moisture content and nature of solvent. Wood Sci Technol 41:597–606
Jeremic D, Cooper P (2009) PEG quantification and examination of molecular weight distribution in wood cell walls. Wood Sci Technol 43:317–329
Jeremic D, QuijanoSolis C, Cooper P (2009) Diffusion rate of polyethylene glycol into cell walls of red pine following vacuum impregnation. Cellulose 16:339–348
Kitani Y, Ohsawa J, Nakato K (1970) Adsorption of polyethylene glycol on waterswollen wood versus molecular weight (in Japanese). Mokuzai Gakkaishi 16:326–333
Sadoh T, Hashihira M (1973) A note on the swelling of wood with polyethylene glycols (in Japanese). Bull Kyoto Univ For 45:227–237
Schneider VA (1969) Contribution on the dimensional stabilization of wood with polyethylene glycol—part 1: basic Investigation on the dimensional stabilization of wood with polyethylene glycol (in German). Holz Roh Werkst 27:209–224
Schneider VA (1970) Contribution on the dimensional stabilization of wood with polyethylene glycol—Part 2: investigation of changes of wood properties by PEG impregnation and on the effectiveness of various impregnation processes (in German). Holz Roh Werkst 28:20–34
Stamm AJ (1959) The dimensional stability of wood. For Prod J 9:375–381
Stamm AJ (1964) Factors affecting the bulking and dimensional stabilization of wood with polyethylene glycols. For Prod J 14:403–408
Tarkow H, Feist WC, Southerland CF (1966) Interaction of wood with polymeric materials penetration versus molecular size. For Prod J 16:61–65
Sadoh T (1967) The dimensional changes of wood during polyethylene glycol treatment and some elastic properties of the treated wood. Mokuzai Gakkaishi 13:41–45
Yamaguchi T, Ishimaru Y, Urakami H (1999) Effect of temperature on dimensional stability of wood with polyethylene glycol II. Temperature dependence of PEG adsorption and mechanical properties of treated wood (in Japanese). Mokuzai Gakkaishi 45:441–447
Greenspan L (1977) Humidity fixed points of binary saturated aqueous solutions. J Res Natl Bureau Stand Sect A Physics Chem 81A:89–96
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Tanaka, S., Seki, M., Miki, T. et al. Solute diffusion into cell walls in solutionimpregnated wood under conditioning process III: effect of relative humidity schedule on solute diffusion into shrinking cell walls. J Wood Sci 63, 263–270 (2017). https://doi.org/10.1007/s1008601716138
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s1008601716138