Sound absorption properties of wood-based pulp fibre foams

Industrial hardwood (HW) and softwood (SW) Kraft and dissolving pulps were provided by Stora Enso Oyj. Kraft pulp samples were obtained from three different stages of the pulping process: before and after oxygen delignification, and after bleaching (ECF bleaching sequence, \(\hbox {D}_{{0}}\)-\(\hbox {E}_{{op}}\)-\(\hbox {D}_{{1}}\)-P). Dissolving pulps were produced by prehydrolysis Kraft (PHK) process. The pulps were either used as never-dried or as machine-dried as described in Table 1.

The materials investigated were prepared using a foam-forming technique following the method described by (Pöhler et al. 2016). Pulp, water and sodium dodecyl sulphate (SDS) were axially agitated at a rotational speed ca. 3000 rpm in a cylindrical container of dimensions 67 cm length and 27 cm diameter. The mixer was built according to (Lappalainen and Lehmonen 2012). Fibre consistency was 2.5 % (250 g of dry pulp in 10 L of tap water) and the dosage of SDS was 0.5 g/L. The dry pulps were soaked in 10 L of water for 1 day at room temperature. Before the addition of SDS, the suspension of pulp in water was mixed for 5 minutes to disperse fibres. Then SDS was added and the foaming continued until the initial volume of the mixture doubled. The fibre foam was poured in a square mould of dimensions 40 cm × 40 cm × 50 cm with two stainless steel nets located at the bottom, and then left to gravity drain for over 15 minutes. The top net had an opening mesh of 0.16 mm, while the net below had a mesh opening of 5 mm. The latter served as a support for the top net to avoid curvature of the bottom surface of the samples being produced. The fibre foam was dried with heated air flow at 40–50 \(\hbox {C}^{\circ }\) for three days. Circular samples of 50 mm thickness and 29 mm diameter were cut from the produced foamed materials with the help of a bench drill, using a 32 mm circular cutter/drill. The produced samples had variable inner density, with a higher density at the bottom. The samples were stored in ambient conditions before measurements.

In order to attain more homogeneous samples in terms of density, a second and a third set of samples were produced by removing layers of fibres from both sides of the samples with the help of a sharp blade. The thicknesses of the samples of the second and third sets were 38 mm and 28 mm, respectively. The densities of the three set of samples are presented in Table 2. It is noted that two specimens were produced from each of the studied pulps. The densities reported in Table 2 are the average of the two specimens.

Pulp carbohydrates and lignin were determined according to standard NREL/TP-510-42618 (Sluiter et al. 2012). The monosaccharides were analyzed by high performance anion exchange chromatography with pulsed amperometric detection (HPAEC-PAD) in a Dionex ICS-3000 system, equipped with a CarboPac PA20 column. The content of cellulose and hemicelluloses were determined applying the Janson formula (Janson 1970). Soluble lignin was measured in a UV-Visible spectrophotometer (UV-2550, Shimadzu).

Fibre length, width, kink, curl, fines, fibres per unit weight, and cell wall thickness (CWT) were characterised using Kajaani FiberLab optical fibre analyser (Metso automation, Finland).

Specific surface area was estimated by \(\hbox {N}_{{2}}\) sorption technique for pulp fibres before and after the foaming process. Prior the measurement, the wood pulp fibers were dried by critical point drying (CPD) to minimize hornification and consequently to reduce the collapse of the pores (Lovikka et al. 2016). The drying was performed as in (Ceccherini and Maloney 2019). The foamed fibres were not critical point dried but degasified under \(\hbox {N}_{{2}}\) flow to avoid hornification. The specific surface area was calculated using the Brunauer–Emmett–Teller (BET) equation (Brunauer et al. 1938), while the pore size using the Barrett–Joyner–Halenda (BJH) equation (Barrett et al. 1951). The adsorption branch of the isotherms was used for the determination of pore size, and pores were assumed to be cylindrical.

The normal incidence sound absorption coefficients of the samples were determined using the impedance tube transfer-function method defined in the standard (ISO 10534-2 1998). The impedance tube used was a Brüel & Kjaer type 4206. All the samples were measured in the small impedance tube (diameter of 29 mm) providing results in the frequency range of 500–6000 Hz. The excitation signal used in the measurements was a logarithmic sweep covering the frequency range 20 Hz–20 kHz. Impulse responses were computed as explained in (Farina 2000). The sound absorption coefficients were computed from the impulse responses following the standard (ISO 10534-2 1998). The excitation signal was amplified using a power amplifier of the type Yamaha MX70. Two 1/4”-pressure microphones of the type Brüel & Kjaer 2670 were used to measure pressures at each microphone location. The measured signals were amplified using a pre-amplifier of the type Brüel & Kjaer Nexus. The audio interface used was the UltraLite-mk3 Hybrid. All the signal processing was performed in Matlab. Sound absorption coefficients reported here are given by the average of the results obtained for two samples.

Table 3 presents the chemical composition of the pulps before going through the foam forming process. The results are in accordance with previous findings reported in the literature (Sixta 2006; Dahlman et al. 2003; Duan et al. 2015). As expected, large amount of xylose is detected for the Kraft HWs, whereas greater proportions of mannose are found in the SWs pulps. Despite the significant difference in xylan content between HW and SW pulps, the remaining xylan in the dissolving pulps is almost equal. Residual hemicelluloses in the bleached and dissolving pulps are suggested to be present as co-aggregates with cellulose fibrils aggregates, which increases the complexity of their removal (Teleman et al. 2001). Residual hemicelluloses may be located in the inner layers of the cell wall structure with lower degree of accessibility (Le Moigne and Navard 2010; Dahlman et al. 2003). Additionally, xylan may be also found on the surface of the fibres due to readsorption ocurring during the last stages of Kraft pulping (Ribe et al. 2010; Pönni et al. 2014).

Table 4 presents morphological properties of the pulps before foam forming. It is noted that the process of foam-forming may induce deformations and create kinks and curl to the fibres. Moreover, the content of fines in the pulp after foam forming is reduced as some fibre fraction goes through the mesh after pouring the foam in the mould. The cell wall thickness decreases with the removal of lignin and hemicelluloses, especially for the SW pulp fibres which have a thicker cell wall. The width of HWs slightly differs between bleached and unbleached pulp fibres. More significant differences in width are seen for the SWs. The greatest reduction in average fibre length is seen for the dissolving pulps with approximately 24 % length reduction for the \(\hbox {HW}_\text {diss}\) and 18 % for \(\hbox {SW}_\text {diss}\) fibres with respect to the length of the unbleached pulps. The greatest reduction in fibre length of dissolving pulp fibres is attributed to the weakening experienced by the fibres as larger amount of hemicelluloses is removed from the cell wall (Risén et al. 2004).

As expected, further processing of pulp results in increasing kink index, microcompressions, and curling of fibres. This increase in fibre defects creates pores in the cell wall (Grönqvist et al. 2014) and decreases fibre stiffness (Page et al. 1980). It can also be seen from Table 4 that the number of fibres per unit weight is greater for HW pulps than for SWs. Especially high is the value obtained for \(\hbox {HW}_\text {diss}\) pulp. The number of fibres per unit weight may influence the tortuosity of fibrous materials.

Figure 1a, b show the \(\hbox {N}_2\) sorption isotherms of the unfoamed and foamed air-dried pulps, respectively. Both sets of pulps present isotherms of type II according to the IUPAC classification (Thommes et al. 2015). H3 type-hysteresis is only present in the isotherms of the unfoamed pulps. The drop in the desorption branch occurring at 0.5–0.42 P/\(\hbox {P}_0\) is typical from materials containing mesopores of diameter 3–5 nm (Kimura et al. 2016). Such drops have been attributed to cavitation effects, which emerge during the desorption process due to the material containing mesopores that are connected to the external surface by a neck diameter smaller than 5–6 nm in the case of nitrogen at 77 K (Thommes et al. 2015). The isotherms of the machine dried \(\hbox {SW}_\text {blch}\), \(\hbox {SW}_\text {diss}\) and \(\hbox {HW}_\text {diss}\) pulps, as opposed to the rest of the pulps, did not show any steep increase at higher relative pressure which indicates that the fibres have almost no macro-pores in the range 50–100 nm, which is in agreement with the results reported by (Kimura et al. 2013).

The resulting SSAs and mean pore sizes are reported in Table 4. The SSAs determined for the unfoamed samples were in agreement with the expected results according to the drying history of the pulps. The measured SSAs of the foamed and air dried pulps were remarkably reduced to lower than 2 \(\hbox {m}^2\)/g. The main reason for the collapse of the mesopores can be explained by hornification effects caused by air drying (Lovikka et al. 2016). It is also possible that due to the vigorous agitation of the pulp in water, hemicelluloses may rearrange in the cell wall filling some of the larger mesopores (Liang et al. 2020; Kimura et al. 2013). Further removal of hemicellulose would increase hornification, resulting in further collapse of mesopores as shown in (Oksanen et al. 1997).

The mean pore size measured for the unfoamed pulps was in the range 7.4–9.2 nm. These values were reduced for the foamed and air dried pulps to 6.5–8.2 nm, which indicates the collapse of larger mesopores. It can be observed that the never-dried \(\hbox {HW}_\text {unblch}\) and \(\hbox {HW}_\text {ox}\) pulps show the maximum pore volume at a pore diameter smaller than that of \(\hbox {SW}_\text {unblch}\) and \(\hbox {SW}_\text {ox}\). This difference is attributed to the difference in lignin content which is greater for SW pulp fibres (Kimura et al. 2016).

The material production process used in this study inherently produces samples that have variable inner density. The densest layer of fibres (1–2 mm thick) is found at the side of the samples that was in contact with the mesh in the draining phase. Figure 2 illustrates the effect of variable inner density on sound absorption. It can be seen that when the densest side of the samples is exposed to the sound source, the first peak of the sound absorption curve is found at lower frequencies. Furthermore, when the densest layer of fibres is removed from the samples (1–2 mm thick layer), the first peak of the sound absorption curves moves towards higher frequencies. The displacement in frequency of the first peak is remarkable for all the samples except for the \(\hbox {HW}_\text {blch}\) pulp. The more efficient sound absorption provided by the denser layer of fibres may be explained by the addition of mass that this layer adds to the system. Similar effects have been reported for the covers used to protect some mineral wool panels from releasing fibres (Cox and d’Antonio 2009). Figure 2 also shows that when the densest side of the sample is exposed to the sound source, the sound absorption coefficients are weakened at the frequency where the first dip occurs. Such dips are typical from impedance tube measurements, and they are caused due the fact that the impedance tube method considers only sound waves arriving at the sample in the perpendicular direction (Cucharero et al. 2020). Those dips tend to vanish in measurements of sound absorption coefficients using the reverberation room method (ISO 354 2003), which assumes sound waves arriving at the specimen from all directions.

As demonstrated in Fig. 3, the difference in sound absorption between the two sides of the samples diminishes as the variable inner density is reduced by removing layers of fibres from both sides of the samples. The effect of variable inner density is also seen at frequencies above 2000 Hz, where higher sound absorption coefficients are obtained for the samples measured from the top side. This is attributed to greater impedance matching between the two mediums of sound propagation, air and the less dense side of the samples, which results in lower rate of reflected sound energy from the sample surface.

Figure 4 illustrates the influence of the morphological properties of HW and SW fibres on sound absorption. The sound absorption coefficients of a glass-wool acoustic panel used for room acoustic purposes has also been plotted for comparison. The static air flow resistivity of the glass wool material was estimated to be approx. 23000 \(\hbox {Nsm}^{-4}\). This estimation was obtained though an inverse technique using the Delany & Bazley sound absorption prediction model (Delany and Bazley 1970) and the measured sound absorption coefficients as the known parameters. In general, HW fibres provide greater sound absorption than SW. As shown before, the density of the samples is a relevant parameter determining sound absorption, and it is the case here, that the density of \(\hbox {HW}_\text {unblch}\), \(\hbox {HW}_\text {ox}\) and \(\hbox {HW}_\text {blch}\) is in general greater than that of their counterpart SWs. On the contrary, the density of \(\hbox {HW}_\text {diss}\) is smaller than that of \(\hbox {SW}_\text {diss}\), however, the first peak of the sound absorption curves is found at considerably lower frequencies for the \(\hbox {HW}_\text {diss}\) than for \(\hbox {SW}_\text {diss}\).

It is confirmed here that, as a general trend for fibrous sound absorbers, shorter and thinner fibres move the first peak of the sound absorption curves towards lower frequencies, as reported earlier by other authors (Koizumi et al. 2003; Arenas and Asdrubali 2018). This effect has been attributed to the fact that the number of fibres to fill a unit volume increases as the fibre dimensions decreases, which results in more tortuous paths and surfaces for sound energy to dissipate inside the material. The number of fibres in each of the samples has been calculated from FiberLab measurements and it is illustrated in Table 5. It can be seen that the number of fibres in the HW samples is over twice the number of fibres in the SW samples for all the pulps.

However, smaller fibre dimensions do not always lead to materials with greater sound absorption. As an example, \(\hbox {SW}_\text {dis}\) fibres are wider and larger than \(\hbox {HW}_\text {unblch}\), and as opposed to expected, \(\hbox {SW}_\text {dis}\) foam samples provide similar sound absorption than \(\hbox {HW}_\text {unblch}\). This indicates that some other parameters influence the sound absorption properties of pulp fibre foams. In addition to fibre dimensions, it can also be observed from Fig. 4 a trend for further processed pulp fibres to provide greater sound absorption, especially when comparing unbleached and dissolving pulp fibre foams. Bleached and oxygen delignified pulp fibre foams do not fully follow this trend.

Significant differences in specific surface area were found between the different unfoamed pulp fibres. However, the differences were lost after foam forming process and air drying of the materials. As a consequence, the effect of fibre porosity on sound absorption could not be analysed in this study. Furthermore, the pulp fibre foams were produced to have approximately the same air content. However, equal air content does not guarantee even bubble size distribution for all the fibre foams (Al-Qararah et al. 2012, 2013). The hydrophilicity and flexibility of the different fibres studied in this work influence bubble size distribution of the foams (Lappalainen and Lehmonen 2012; Hou and Wang 2018), and therefore, it may also influence the sound absorption properties of the materials. In future research, the air bubble size distribution in the foam mixture containing different pulp fibres, and the changes in the inter-fibre pore structure caused by the drying process needs to be studied. Further research includes modification of the material manufacturing process to avoid hornification. A potential improvement for this purpose would be using foam forming technology but utilizing non- or less polar solvents instead of water to prevent closure of pores. The use of non- or less-polar solvents would prevent the formation of hydrogen bonds between fibres, thus reducing the mechanical strength of the material. Fibre foams, however, can be formed by adding small amounts of soluble adhesives to provide the inter-fibre bonding.

In general, all the different pulp fibre foams studied in this work provided sound absorption properties comparable to that of conventional synthetic sound absorbers as shown in Fig. 4. Understanding what properties of pulp fibres influence the most the ability of materials to dissipate sound energy is essential for the optimization of bio-based sound absorbers. Optimized wood-fibre based sound absorbers offer excellent acoustic properties while reducing \(\hbox {CO}_2\) from the atmosphere by binding \(\hbox {CO}_2\) into the building structure during the material operating life.