Sorption properties of paper treated with silane-modified starch

Bleached softwood Kraft fibers (Södra Black R, Sweden) with average length and width of 2100 and 30.0 μm, respectively, were used as the raw material. The cellulose fibers were characterized by the coarseness of 135 μg/m, ash content of 0.25%, brightness of 89.5%, and pH of 4.8.

Wheat starch (C*Flex 20002, Minneapolis, MN, USA) was used as a functional additive in the study. The starch was characterized by the amylose content of 31.2 ± 0.16%, the lipid content of 0.25 ± 0.02%, and the protein content of 0.30 ± 0.01%. An aqueous solution of starch (concentration of 5%) was stirred at 70 ± 3 °C in the presence of 0.5% sodium hydroxide (CAS no, 1310-73-2, ChemPur, Poland) to reduce the temperature of starch gelatinization (Kolpak et al. 1978; Okano and Sarko 1985). The starch was separately modified with four silane coupling agents, i.e. methyltrimethoxysilane (MTMS) (Sigma Aldrich, CAS no, 1185-55-3, St. Louis, MO, USA), tetraethyl orthosilicate (TEOS) (Sigma Aldrich, CAS no, 78-10-4, St. Louis, MO, USA), n-octyltriethoxysilane (NOTES) (Sigma Aldrich, CAS no, 2943-75-1, St. Louis, MO, USA) and 1H,1H,2H,2H-perfluorooctyltriethoxysilane (WR) (Sigma Aldrich, CAS no, 51851-37-7, St. Louis, MO, USA).

Before paper manufacturing, either as native or in a modified form, cellulose fibers were immersed in deionized water for 24 h for better defibrillation during paper sheet production. A lubricancy of diluted suspension of pulp was measured by the Schopper-Riegler apparatus (Labormex, Poland) according to SIST PN-EN ISO 5267-1:2000/AC: 2003. Pulp suspension in water in terms of the Schopper-Riegler number was 14 ± 0.8. The Rapid-Köthen sheet former (Labormex, Poland) was applied to prepare paper sheets with a diameter of 200 ± 0.1 mm and an average thickness of 0.5 ± 0.02 mm. The paper sheets were vacuum dried in the Rapid-Köthen sheet former for 600 s at a temperature of 95 ± 1 °C. Then the sheets were conditioned at the set temperature of 18 °C and relative humidity of 50%.

The cellulose fibers were modified after forming paper sheets, i.e. the additives were applied at both surfaces of already produced sheets. An aqueous solution of wheat starch (concentration of 5%) was first stirred at 70 ± 5 °C in the presence of 0.5% sodium hydroxide to reduce the starch gelatinization temperature. The starch structure was loosened to make it easier for silanes to penetrate amylose and amylopectin chains. The obtained suspension was cooled down to 25 ± 5 °C, and while mixing, 2.5% concentration of organosilicon compounds (i.e. MTMS, TEOS, NOTES, or WR) was added. The prepared formulations were applied as coatings using 100 g/m² for paper sheets produced according to the Rapid-Köthen sheet former procedure. After modification, sheets were conditioned for 7 days in a laboratory chamber at temperature of 18 ± 1 °C and relative humidity of 50%. Five paper sheets were produced for each option of modification.

The application of the additives resulted in the following options of the investigated materials, i.e. Control—untreated paper, WS—paper treated with wheat starch only, WS + MTMS—paper treated with starch modified with MTMS, WS + TEOS—paper treated with starch modified with TEOS, WS + NOTES—paper treated with starch modified with NOTES, WS + WR—paper treated with starch modified with WR. For each option of materials, 5 samples were prepared.

The samples of dimensions of 0.5 · 50 · 30 mm were used for investigating hygroscopic properties. The dimensions and the number of samples for each option of the examined material resulted from the error analysis based on the total differential method (e.g. Taylor 1997). It was assumed that the relative error of the equilibrium moisture content determination should not exceed 2% for the lowest level of air relative humidity. Before starting sorption experiments, all samples were stored for 3 weeks in closed containers over phosphorus pentoxide to approach the dry state. The experimental set-up consisted of two chambers, as described by Majka and Olek (2007; 2008). The outer chamber encased the inner chamber to stabilize air temperature inside the set-up. The samples for all options of the examined material were placed in the inner chamber, and a fan forced airflow. The air relative humidity was controlled by saturated salt solutions listed in Table 1. The weighing procedure was repeated until the difference between the subsequent mass readings was not exceeding 0.001 g. It indicated that the hygroscopic equilibrium was approached. During the weighing phase of the samples, the upper part of the outer chamber was removed and the balance was placed on the top of the inner chamber. A string was used to link the balance and a pan which was placed inside the inner chamber. Therefore, the air exchange between the inner and outside chambers was minimized and limited to the weighing phase only. The sorption experiments were performed at a temperature of 22 ± 1 °C for both adsorption and desorption modes. The stability of air parameters during the sorption experiments was controlled by measuring temperature and relative humidity with an electronic thermohygrometer (LB 706, LAB EL, Poland). The samples were weighed at least twice at each relative humidity level specified in Table 1. After completing sorption experiments, the samples were placed in a laboratory dryer at a temperature of 103 °C to determine the oven-dry mass. Each equilibrium moisture content value was determined as the average of five observations for each material option.

As obtained for all options of the investigated material, the adsorption and desorption isotherms were parametrized with three structural models. The first equation was the three-parameter Guggenheim, Anderson, and De Boer (GAB) model, still the most frequently applied equation for describing sorption isotherms in wood science. The GAB model was used in the following formula:where EMC; kg/kg—equilibrium moisture content, RH—air relative humidity, M_m; kg/kg—monolayer capacity, C_GAB—equilibrium constant related to the monolayer sorption, K_GAB—equilibrium constant related to the multilayer sorption.

The model was often transformed into another mathematical form in which the ratio of relative humidity and equilibrium moisture content was the parabolic function of relative humidity (e.g., Babiak 2007; Skaar 1988; Thybring et al. 2021):where a, b, c—fitted parameters are functions of the original coefficients of the GAB model.

The GAB equation assumes binding water onto primary sorption sites and forming the monolayer of water, i.e., the primary water. These water molecules are converted into the secondary sorption sites, and additional layers of water are sorbed, i.e., the secondary water. The model assumes that the molecules of the secondary water are less strongly sorbed than the ones of the monolayer water. The application of the GAB model in wood science was questioned due to (a) deviation of empirical data from the parabolic relation (e.g Nakano 2006; Olek et al. 2013), (b) the decrease in the ratio of primary and secondary water contents with temperature (Zelinka et al. 2018), (c) the prediction of too low monolayer capacity and the capacity decreases with increasing temperature (Thybring et al. 2021).

It was already postulated to develop and apply new models describing sorption isotherms (e.g., Zelinka et al. 2020). Therefore, the four-parameter Generalized D’Arcy and Watt (GDW) equation developed by Furmaniak et al. (2007a, b) was utilized as the second model in the present study. The GDW equation assumes the existence of the primary sorption sites on solid surfaces and postulates that each active site can adsorb only one water molecule. The already adsorbed water molecules can be converted into the secondary sorption sites. The most important feature of the GDW model is related to three possible scenarios of the secondary water sorption resulting from the mathematical structure of the equation:where EMC; kg/kg—equilibrium moisture content, RH—air relative humidity, m_GDW; kg/kg—monolayer water content (the maximum content of water bound to the primary sites), K_GDW—kinetic constant related to sorption on the primary sites, k_GDW—kinetic constant related to sorption on the secondary sites, w—ratio of water molecules bound to the primary sites and converted into the secondary sites.

The mentioned three scenarios of the secondary water sorption depend on possible values of the w parameter: (a) w < 1—water molecules bound on the primary sites are not completely converted into the secondary sorption sites (i.e., the number of the secondary sites is lower than the primary sites), (b) w = 1—all primary water molecules are converted into the secondary sorption sites, and the GDW model is reduced to the GAB model (Furmaniak et al. 2007a), (c) w > 1—each primary water molecule is statistically converted into more than one secondary sorption site, and the higher w values are obtained, the more intensive process of water cluster formation occurs.

The third applied equation was the four-parameter model proposed by Yanniotis and Blahovec (2009) and denoted here as the Y-B model. It assumes that the primary water is sorbed according to the Langmuir mechanism. In contrast, the remaining water is less strongly bound due to the solution formation according to the modified Raoult’s law. The following form of the model is given:where a₁, a₂, b₁, b₂ are the coefficients to be fitted. The first component of the sum is recognized as the content of water sorbed on active sorption sites, while the second one involves the content of water forming solution with the solid. The solution formation mechanism is less realistic than the water cluster arrangement according to the GDW model. However, Yanniotis and Blahovec (2009) proposed a method for classifying water vapor sorption isotherms. The fitted coefficients of the model were used to calculate additional parameters, which were originally denoted as D₁₀, R_fi, and X₄. The relations of positive or negative values of the additional parameters were the criteria for the isotherms classification (Table 2).

The sessile droplet contact angle was measured at the standard conditions, i.e., at temperature of 20 ± 1 °C and air relative humidity of 50%, using a Kruss contact angle measuring device (Kruss, Germany). A water droplet of 4.0 µL volume was placed on the surface, and a camera took a picture of the droplet within 30 s. Every sample was evaluated in five repetitions.

The results of the model parametrization are presented in Table 3. The obtained values of the monolayer capacity (M_m), as reported for adsorption, were very low (ca. 0.04 kg/kg) and practically equal for all options of the treated material. It might suggest that the majority of absorbed water molecules were linked to the secondary sorption sites. Such exceptionally low values of the monolayer capacity are contrasted to the availability of sorption sites determined with the deuterium exchange method. Recently, Thybring et al. (2021) analyzed values of the monolayer capacity as predicted from different parabolic models of water sorption isotherms in beech and spruce wood (incl. the GAB equation) and experimentally obtained data on the accessibility of sorption sites. It was found that the measured values of accessibility were 2–5 times higher as compared to the predicted ones. Therefore, it was stated that the monolayer capacity values obtained from the analyzed sorption isotherms models, incl. the GAB equation, are not physically correct. The deuterium exchange method was also applied to quantifying available sorption sites in cellulosic pulps (e.g., Väisänen et al. 2018). The content of the accessible hydroxyl groups in kraft pulp obtained from spruce and pine wood was equal to 11.0 mol/kg. It had to result in the monolayer capacity as high as 0.2 kg/kg and suggests that the Langmuir mechanism governs the majority of sorption, and an only minor portion of water molecules is linked to the secondary sorption sites. Such a scenario will be verified within this study when analyzing the fitting results of the applied four-parameter GDW and Y-B sorption isotherm models.

The values of the C_GAB coefficient of the GAB model are referred to sorption enthalpy of the monolayer water and are used for classifying sorption isotherms. The necessary condition for ranking the isotherms as sigmoidal ones (i.e., type II isotherms) is to obtain values of C_GAB higher than 2. The condition was satisfied for all treatment options and sorption phases (Table 3). However, Lewicki (1997) enhanced the condition and defined the following system of inequalities: 5.57 ≤ C_GAB < ∞ and 0.24 < K_GAB ≤ 1. The additional condition was again satisfied for all treatment options and sorption phases. Therefore, all isotherms were fully classified as sigmoidal ones. The observed values of C_GAB were much higher than K_GAB for all examined treatment options. The relation indicates that water molecules directly bound to active primary sites are much stronger linked to the sorption centers than the secondary water. However, the changes in C_GAB values could not be related to the applied treatment options.

The K_GAB coefficient denotes the difference between the free enthalpy of polymolecular and bulk water (Timmermann 2003). The reported values of the coefficient were somewhat lower than 1 and practically the same for all treatment options and sorption phases (Table 3). It can be concluded that neither the treatment option nor the sorption phase influenced the binding energy of water molecules during polymolecular sorption.

The results of fitting the GDW model to all treatment options and sorption phases are given in Table 4. The predicted maximum water content bound to the primary sites (m_GDW) was always over 0.1 kg/kg for the adsorption mode. Comparing the m_GDW value to registered values of the equilibrium moisture content shows the importance of sorption of the primary water compared to binding water molecules to the secondary sites. The paper treatment with starch resulted in a distinct increase of m_GDW to 0.137 kg/kg in the adsorption phase. It was due to the higher hydrophilicity of starch as compared to cellulose. The application of starch modified with silanes usually reduced m_GDW compared to paper coated with unmodified starch. The only exception was paper coated with starch modified with TEOS, which caused an additional increase of m_GDW.

The ratio of water molecules bound to the primary sites and converted into the secondary ones (w) was analyzed for the adsorption mode only as it refers to water gain. The w parameter was much lower than 1 for all analyzed treatment options. It can be concluded from the results presented in Table 4 that statistically, less than every fifth water molecule bound to the primary sorption site was converted into the secondary site. However, the improvement of paper hygroscopicity should be analyzed for the combined values of the parameters m_GDW and w. It is evident from the results in Table 4 that the best improvement should be expected for the treatment with starch modified with WR. It is consistent with the results presented in Fig. 1d. The reported values of the kinetic constant related to sorption on the primary sites (K_GDW) were consistently higher than those of the kinetic constant related to sorption on the secondary sites (k_GDW). It indicates that water molecules directly sorbed on active sorption sites were more strongly attached to the material than the molecules of secondary water.

Furmaniak et al. (2011) proposed a simplified classification of sorption isotherms based on the fitted GDW model. The type I isotherms with only primary sorption sites occurring in a sorbent were identified for the w parameter being equal or close to zero. However, the last alternative condition seems to be not clearly specified. It is of special importance in light of the values of the w parameter obtained in the present study, i.e. ranging from as small value as 0.13 to the highest of only 0.25. The type II isotherms were originally classified by Furmaniak et al. (2011) for w > 0 and K_GDW > 1, while the type III ones for w > 0 and K_GDW < 1. When strictly applying the listed conditions, the sorption isotherms obtained in the present study were classified as type II. However, the reservations specified on the proposed classification derived from the GDW model induced us to perform more analysis on the sorption isotherms classification.

The estimated coefficients of the Yanniotis and Blahovec model are depicted in Table 5. Unfortunately, the mathematical structure of the model did not allow for direct coefficient analysis. Therefore, the model was only used for classifying sorption isotherms. The additional parameters D₁₀, R_fi and X₄ were calculated for each isotherm. The obtained values of D₁₀ and X₄ were always positive, while the values of R_fi were always negative. Therefore, all sorption isotherms were classified as type IIb, i.e., sigmoid more to the solution-like (Table 5). It supports our earlier statements that the Langmuir mechanism is significant during water sorption in the analyzed materials but not the only one. The secondary water bonding has to be accounted for as the sorption mechanism and regarded almost equivalent to the previous one.

The sorption isotherms obtained for all treatment options were used for quantifying sorption hysteresis. The set of descriptors defined by Majka et al. (2016) was used. The descriptors consisted of sorption hysteresis loop (H), the maximum difference of equilibrium moisture content for desorption and adsorption (∆EMC), and corresponding relative humidity (RH). The estimated descriptors of hysteresis are presented in Table 6. The applied treatment options had no significant influence on sorption hysteresis. It was clearly resulting from the values of all descriptors of sorption hysteresis.

The observed increase of equilibrium moisture content of paper treated with starch modified with MTMS, TEOS, and NOTES was contrary to the findings reported for the interaction with liquid water (Nowak et al. 2022). It was reported that the application of the coatings caused a significant improvement in the static water contact angle (WCA). The values of the WCA recorded by Nowak et al. (2022) started from non-measurable ones for untreated paper (control material), with some improvement for starch-treated paper (contact angle approx. 50°), up to a significant increase over 115° for paper treated with starch modified with MTMS. However, not all options of the paper treatment were equally effective, e.g., a coating made of starch modified with TEOS and NOTES revealed less significant improvement of the angle than unmodified starch. The differences in the properties of paper properties were explained by the starch network’s silane distribution. MTMS molecules were characterized by short side chains (i.e., methyl groups) and relatively easily fixed within the starch structure. In contrast, silanes having longer side chains (e.g., NOTES) primarily covered starch surfaces. The Penetration Dynamics Analysis supported the findings, and the recommendation was made that the paper coating with starch modified with MTMS can be considered an effective method for improving the properties related to the interaction of paper with liquid water.

A different relation was observed for paper coated with starch modified with WR (Table 7). It was the only treatment option causing the reduction in the equilibrium moisture content of the paper sheets. Moreover, the measured static water contact angle was the highest among all studied treatment options and equal to 125°. It clearly shows that the paper treatment with starch modified with WR ensured the best hydrophobization of the material and improved the hygroscopic properties. Similar results on the effectiveness of fluorosilane coating were reported by Szubert et al. (2019), where high hydrophobic properties of surface treated wood were presented (WCA > 120°). The disparate characteristics obtained for the interaction of coated paper with moist air are presumably related to mechanisms blocking active sorption sites within the paper and/or starch.