Introduction

There is a long history of using calcium carbonate as an additive in papermaking (Yan et al. 2005). It has been a key component for enhancing multiple attributes of paper. For example, it is often added as a filler to enhance brightness, smoothness, and dimensional stability. The addition of calcium carbonate also helps in reducing the overall cost (Shen et al. 2009). Oftentimes, these enhanced attributes and reduced cost come at the expense of other technical challenges (Hubbe and Gill 2016). For example, the addition of calcium carbonate could lead to a decrease in strength and frequent web breaks. It is widely thought that fillers obstruct the formation of inter-fiber bonds in the paper (Hubbe and Gill 2016). For small particle size fillers, a decrease in filler retention was also noticed. As a result, strategies have been developed to overcome these obstacles. For decreased strength, it has been found that adding cationic starch in the presence of calcium carbonate has improved the mechanical properties of paper (Sang et al. 2012). For decreased retention, the addition of retention aids (such as acrylamide copolymers and polyethylene oxide) has been found effective in increasing fillers retention (Hubbe et al. 2009). Another key role of calcium carbonate as an added filler is the improvement of dimensional stability. Previous studies have shown improved dimensional stability with the increase of filler content of the paper which was attributed to the decrease in inter-fiber bonding (Lyne et al. 1996).

In addition to using calcium carbonate in the pulp and paper industry, it has also been found effective in improving the fire retardancy of wood. The addition of calcium carbonate to wood resulted in a lower heat release rate and the total heat release compared to that of the original wood (without calcium carbonate) indicating an improved fire retardancy (Merk et al. 2016; Huang et al. 2017).

The coexistence of nano/microfibrillated cellulose and calcium carbonate has been mainly focused on the role of CNFs as a retention aid in papermaking. It has been found that the addition of nanocellulose to calcium carbonate-containing paper greatly increased the retention of the filler. There has been, however, controversies over whether or not the addition of CNFs enhances the strength of the paper. Ämmälä et al. (2013) have studied the effect of 2,2,6,6-tetramethylpiperidine-1-oxyl radical (TEMPO) and periodate-chlorite oxidized (DCC) anionic nanofibrils (both at doses of 2% and 4% of sheet weight), on ground calcium carbonate retention (initially added at 15, 30, and 50 wt%), strength, and optical properties of handsheets (Ämmälä et al. 2013). Even though the addition of the chemically treated nanocellulose improved the flocculation and retention properties, it did not result in an improved sheet strength due to the poor formation of the handsheets. To overcome the loss in strength and to induce a better bonding between CNF and precipitated calcium carbonate, He et al. (2016) used cationic starch (He et al. 2016). The authors found that the addition to 2.5% starch (of filler weight) to the CNF (2% of pulp weight)/precipitated calcium carbonate (20, 30, and 40% of pulp weight) improved the strength with acceptable sheet formation. It has been emphasized that chemical modification of CNFs is necessary to increase the electrostatic interactions with particles and other additives which results in increased flocculation and maximum retention (Korhonen and Laine 2014).

In the wood adhesive industry, many of the currently available petroleum-based adhesives (e.g. formaldehyde-based adhesives used in the formulation of wood-based panels) are increasingly recognized as a serious worldwide public health concern (Hemmilä et al. 2017). As a consequence, serious efforts are being directed nowadays towards the development and commercialization of eco- and health-friendly binders for wood-based panels. Although research has been carried out extensively on bio-based adhesives for wood, these alternatives are not without challenges (Frihart 2016). For instance, lignin-based adhesives often require chemical modification due to the inherently low reactivity of lignin (Ferdosian et al. 2017). Additionally, despite the good bonding strength and high level of substitution of starch- and protein-based adhesives, their poor resistance to water and the storage stability are often a challenge. One of the most promising attempts so far has been the application of cellulose nanofibrils (CNFs) as a binder for wood-based panels either as the sole binder (Amini et al. 2017; Leng et al. 2016, 2017; Hunt et al. 2017) or in combination with mycelium-based binders (Sun et al. 2019). Interestingly, these panels were produced in a wet process without the need for prior drying of the CNFs; a typical challenge in CNF applications. The particleboard bonded with cellulose nanofibrils met or exceeded the standard specifications of the American National Standard Institute (ANSI) for low and medium density grades. These attempts have paved the way for further explorations into the CNF-based binders.

The properties of a composite made only of nano/microfibrillated cellulose and calcium carbonate remains to be elucidated. In this study, we first evaluated the properties of nano/microfibrillated cellulose and calcium carbonate composite films. Other characterizations were also conducted on starting materials in the form of suspension. Second, a preliminary assessment to evaluate the possibility of using such a system as a binder for lignocellulosic materials is presented. This study builds on previously published research on using CNFs as a sole binder for particleboard (Amini et al. 2017; Leng et al. 2016, 2017; Hunt et al. 2017). Herein, a composite made of CNFs and FLC (microfibrillated cellulose, MFC, and calcium carbonate CaCO3, 50:50) was characterized in the form of film and suspension.

Little is known about the impact of calcium carbonate when added together with CNFs as a binder for wood-based panels. This combination could lead to improved panel attributes and cost reduction. Thus, if it is found that the presence of calcium carbonate particles does not inhibit the binding properties of CNFs, then a hybrid CNF-based binder could be formulated at a lower cost, leading to better prospects for the commercialization of cellulose nanomaterials.

Materials and methods

Materials

Cellulose nanofibrils (CNFs) at 3 wt% solid content was obtained from the Process Development Center of the University of Maine. FiberLean™ (FLC) was kindly provided by FiberLean® Technologies UK in the form of press cake at 31.6 wt% solid content. The received press cake was diluted to batches of 3 wt% solid content to facilitate the subsequent dilutions during the preparation of the films (to 0.5 wt%). Recycled old corrugated container (OCC) paper (200 g/m2, gsm) was provided by the Process Development Center of the University of Maine. Weldwood® carpenter’s wood glue was purchased for the lap shear experiment.

Methods

The formation of the films

All the films were prepared via a vacuum filtration technique. In total, 5–6 films were prepared from each sample (1 g dry mass each). The vacuum filtration system comprised mainly a vacuum pump, a Büchner funnel (diameter: 115 mm), and a 1 L Erlenmeyer flask. The starting material was diluted to a solid content of 0.5 wt% prior to filtration. The diluted suspension was poured into a Büchner funnel and allowed to filter through a Whatman™ Grade 5 filter paper (2.5 µm pore size). Before applying vacuum, the suspension was kept in the funnel for a few seconds until a few water droplets were drained by gravitational force under atmospheric pressure. The vacuum was then applied (25 inHg = 86 kPa) until the complete removal of free water. The formed film was placed onto a dry (and slightly larger) Whatman™ Grade 5 filter paper (2.5 µm pore size) and placed between two stainless steel disks. The whole assembly was placed in a conventional oven at 75 ± 2 °C for 24 h with a load (~ 2 kg) on top of it to fully dry the films and to avoid wrinkles. Table 1 summarizes the three types of tested samples in this study.

Table 1 Summary of film components (by weight) prepared from CNF and FLC
Full size table

Density and porosity measurements

The density (of films and lap shear test substrates) was measured by first cutting rectangular (50 mm × 30 mm) specimens. The specimens were then conditioned in a conditioning chamber at a relative humidity of 53 ± 2% and temperature of 23 ± 2 °C for 24 h. The mass and volume were determined to calculate the density of the films. The porosity of OCC substrates was calculated according to Eq. 1 assuming that the density of lignocellulosic fibers is 1.5 g/cm3 (Nägele et al. 2002):

$$ Porosity \left( \% \right) = \left( {1 - \left( {\frac{Density\,of\,OCC\,specimen}{1.5}} \right)} \right) \times 100 $$
(1)

Zeta potential (ζ)

The zeta potential, evaluated by Malvern Zetasizer Nano ZS Zen 3600 instrument (Malvern, UK), was determined by measuring the electrophoretic mobility of the sample using a capillary electrophoresis cell. A plastic syringe was used to add the sample into the cell. The cell was placed inside the instrument, equilibrated for 120 s, and the Smoluchowsky formula was used to convert the electrophoretic mobility measurements in zeta potential values (Delgado et al. 2005; Kaszuba et al. 2010).

Particle size analysis

The relative particle size of CNFs and FLC was determined by Malvern Hydro 2000s laser diffraction equipment (Malvern, UK). A sample suspension was added into the sample opening in the equipment and subsequently circulated through the system. Once the laser hit the particles, the intensity and angle of the scattered light were detected. The particle size distribution was derived afterward through an algorithm. The size of the particle represented the diameter of a sphere having the same volume as the particle.

Scanning electron microscopy (SEM)

To assess the morphology of CNFs, FLC, and FLC-CNFs, a Zeiss NVision 40 microscope was used. The film samples were examined from the surface and at the cross-section. The cross-section samples were prepared by cutting the film specimens using a sharp blade. The samples were placed on a sample holder and coated with gold/palladium (15-nm thick) to enable the imaging of samples. All images were taken at a working distance of 4 mm and an accelerating voltage of 3 kV.

Energy dispersive spectroscopy (EDS)

An iXRF Model 550i EDS system (on an AMRay 1820 SEM) was used to produce images of the spatial distribution of elements in the sample. In this study, the distributions of carbon, oxygen, and calcium were obtained. The distribution of calcium carbonate was inferred from the elemental calcium distribution. Similar to SEM, the samples were placed on a sample holder and coated with gold/palladium (15-nm thick) to enable the imaging of samples. All images were taken at a working distance of 15 mm and an accelerating voltage of 20 kV.

X-ray diffraction (XRD)

A Panalytical X’Pert PRO diffractometer (Royston, UK) was used to acquire the X-ray data. The test was conducted using a chromator-filtered CU Kα radiation (λ = 0.15418 nm) at 40 kV and 40 mA. The crystallinity index was determined according to Segal’s method (Segal et al. 1959) where the height of the highest diffraction peak corresponded to the amount of crystalline material and the height of the minimum intensity between the major peaks (2θ approximately 18°) corresponded to the amorphous material (French and Santiago Cintrón 2013) as shown in Eq. 2.

$$ Crystallinity\;index \left( \% \right) = \frac{{\left( {I_{200} - I_{am} } \right)}}{{I_{200} }} \times 100 $$
(2)

where I200 is the maximum intensity (in arbitrary units) of the 200-lattice diffraction (representing the crystalline material) and Iam is the minimum intensity (between the major peaks) in the same units at 2θ approximately 18 degrees representing the amorphous material.

Mechanical properties

The tensile properties of the films (and OCC substrates used for lap shear testing) were evaluated using an Instron 5942 (Norwood, MA, USA) testing machine equipped with a 500 N load cell (Fig. 1). Three to four dog-bone specimens were cut from each film (total of 6–9 tested specimens per sample) and conditioned in a conditioning chamber at a relative humidity of 53 ± 2% and temperature of 23 ± 2 °C for 24 h (adequate to reach a constant mass). The specimens were tested at a loading rate of 2 mm/min and an initial grip span of 10 mm.

Fig. 1
figure 1

From left to right: the dimensions of the tensile testing specimen, a typical stress–strain curve, and Instron 5942 testing machine

Full size image

One-way analysis of variance (ANOVA) was used to determine the statistical difference between the means of the independent variables. Duncan’s Multiple Range Test (DMRT) was conducted as a post hoc test to measure the specific differences between pairs of means. All analyses were performed using IBM SPSS Statistics 25 software at 95% confidence level.

Lap shear testing

To investigate the bondability of the CNF, FLC, and FLC-CNF binders to OCC substrates, lap-shear tests were carried out in accordance with ASTM D4896-01 (2016) with modification using an Instron 5942 (Norwood, MA, USA) testing machine equipped with a 500 N capacity load cell. Lap-shear specimens were prepared by cutting rectangular substrates of 50 mm by 20 mm out of 200-gsm OCC and then overlapping two substrates bonded together using the binder (at a spread rate of 0.012 g dry mass per glue-line) in-between. The lap area was 10 mm long and 20 mm wide. The samples were then pressed at 0.17 MPa and 180 °C for 1 min using a hydraulic hot press (Carver, Inc., Wabash, IN, USA). Six replicates of each sample were produced and conditioned in a conditioning chamber at a relative humidity of 53 ± 2% and temperature of 23 ± 2 °C for 24 h. The lap-shear specimens were tested at a loading rate of 1 mm/min and an initial gauge length of 35 mm.

Thermogravimetric analysis (TGA)

The thermogravimetric analysis was conducted using a TGA Q500 (TA Instruments, New Castle, DE, USA). For each test, a 6–8 mg specimen was cut from the film and placed onto a tared platinum pan. The test was conducted at a heating rate of 10 °C/min and a final temperature of 600 °C in a nitrogen gas atmosphere at a flow rate of 40 mL/min. The data were analyzed using a TA Universal Analysis software.

Filtration (dewatering) analysis

The filtration analysis was conducted in an OFITE (Houston, TX, USA) filter press. The analysis was conducted on the three samples; CNFs, FLC, and FLC-CNFs. One hundred grams of 3 wt% solid content of the sample was placed in the cylinder of the filter press. A filter paper of pore size 2.7 µm was used as a filtration medium. A continuous air supply (172 kPa) was applied and the filtrate mass was recorded during 30-min of filtration.

Retention

The total retention of fiber/mineral (i.e. nano/microfibrillated cellulose/calcium carbonate) after film formation and dewatering experiment was determined according to Eq. 3, where the initial dry masses for film formation and dewatering experiment were 1 g and 3 g, respectively.

$$ Total\;retention \left( \% \right) = \frac{Dry\,mass\,of\,retained\,solid}{Initial\,dry\,mass } \times \, 100 $$
(3)

The retention of calcium carbonate was determined by measuring the ash content according to a modified procedure from ASTM D1102 using thermogravimetric analysis (TGA). To this end, a TGA Leco 701 (St. Joseph, MI, USA) was used to ignite the samples (1 g for film samples and 3 g for dewatering samples) at a temperature of 560–580 °C under air atmosphere. At this temperature, cellulose and moisture were be removed leaving calcium carbonate as ash in the crucibles.

Results and discussion

Surface and structural evaluation

Surface zeta potential (ζ) showed that the mean negative potential was − 22.2 mV for CNFs and − 18.2 mV for FLC particles (Table 2). This increase in zeta potential is possibly due to the decrease in the exposed surfaces of microfibrillated cellulose due to calcium carbonate particles. A relative assessment of particle size was achieved by laser diffraction technique. The median for volume distribution (D50) was 65 um and 141 µm for CNFs and FLC, respectively (Table 2). It is important to note that particle size obtained by the laser diffraction technique is equivalent to a sphere diameter having the same diffraction pattern (Hrncirova et al. 2013).

Table 2 Laser diffraction and zeta potential results summary
Full size table

Figure 2 shows SEM images of the surface and cross-section of the films. The surface of CNF films (Fig. 2a, d) showed a high degree of fibrillation which resulted in an entangled and randomly-oriented network of fibrils. In the FLC sample (at × 10,000 magnification) (Fig. 2b), the calcium carbonate particles can be easily seen as agglomerates of particles on the surface rendering the observation of cellulose microfibrils slightly difficult. At higher magnification of × 25,000 (Fig. 2e), calcium carbonate particles can be seen entrapped in the network of microfibrillated cellulose. The same image also showed that the particles of calcium carbonate were not uniform in size. They vary greatly in size, ranging from 50–70 nm to 1–2 µm. The diameter of individual fibrils of CNFs and MFCs (the microfibrillated cellulose in FLC sample) was very comparable, they both ranged from 40 to 70 nm which suggests that the increased size of FLC particles obtained by laser diffraction was potentially due to an increase in the length of fibrils.

Fig. 2
figure 2

Scanning electron microscopy (SEM) images of film surface and cross-section at various magnifications at the same order from left to right, CNFs, FLC, and FLC-CNFs. The first row a, b, and c is the film surface at a magnification of × 10,000. The second row d, e, and f is the film surface at a magnification of × 25,000. The third row g, h, and i is the cross-section at a magnification of × 250. The fourth row j and l is the cross-section at a magnification of × 10,000 (except k at a magnification of × 5000)

Full size image

The degree of entanglement around the calcium carbonate particles increased with the introduction of CNFs to the FLC (i.e. in the FLC-CNF sample) (Fig. 2c, f). It is worth reminding that in the mix of CNFs and FLC, the amount of calcium carbonate was 25 wt% compared to 50 wt% in the FLC sample only. It appears that the CNFs and MFC act together as a web-like network that keeps the calcium carbonate particles in place. Furthermore, one can notice that the calcium carbonate particles are distributed within the film following an interesting pattern. The SEM images show that large particles of calcium carbonate were retained on the top surfaces of the film whereas smaller particles are seen within the inner layers of the film.

The cross-section of the samples that contain CNFs looked smoother than the FLC sample (Fig. 2g–i). In the FLC cross-section sample, a line was observed along the cross-section of the film. This line, at a higher magnification of × 5000 (Fig. 2k), appeared as a gap between the top and bottom surfaces of the film. It seems possible that this phenomenon is due to an intense disruption of the fiber–fiber bond resulting from the high load of mineral in this particular formulation (50 wt% in this case).

Furthermore, EDS mapping was conducted to gain more insight into the distribution of calcium carbonate on the surface and in the z-direction of the film (Fig. 3) inferred from the elemental distribution of calcium. In the surface distribution images (Fig. 3a), the intensity of the yellow dots (first row from the top in Fig. 3a) in FLC sample was noticeably higher than that of FLC-CNF sample. It is also evident that calcium carbonate is uniformly distributed on the film surface in both FLC and FLC-CNF samples. Minute traces are noticed in CNF sample which are likely attributed to calcium-containing contaminants (e.g. from tap-water used to produce CNF at the pilot-plant facility). Expectedly, the distribution of carbon and oxygen remained constant throughout the three samples. The relatively brighter corners observed in the carbon distribution of the surface (and absent in the other images) are likely due to the conductive carbon tape (placed around the corners of the surface samples) which are made of pure carbon, hence showing an increased intensity in the carbon distribution images and absent in the non-carbon images. Turning now to the cross-section images, calcium carbonate was fairly well-distributed across the thickness of the FLC-CNF sample. In the FLC sample, however, agglomerations of the mineral were observed in different spots of the cross-section. Also, carbon and oxygen distributions showed a relatively darker area across the center of cross-section which likely corresponds to the gap previously observed in Fig. 2h, k.

Fig. 3
figure 3

EDS mapping of calcium, carbon, and oxygen of a film surface (scale bar = 1 mm) and b cross-section (scale bar = 20 µm)

Full size image

In order to assess the crystalline structures within the films, X-ray diffraction (XRD) was used (Fig. 4). The three samples exhibited a broad peak between 15° and 16.5° which corresponds to the Miller indices (1–10) and (110) (French 2014). The 22.9° peak is attributed to the (200) lattice diffraction. The 35° represents the (004) crystallographic plane for cellulose I. The intensities of these three peaks are lower in FLC-CNF and FLC films. The peak arising at 28.6° in CNFs and FLC-CNFs is likely attributed to impurities in the CNFs. The films containing calcium carbonate showed characteristic peaks at 29.4°, 31.5°, 36.0°, and 39.4° (Harris et al. 2015) and the planes are indicated in Fig. 3. The crystallinity indices (Segal et al. 1959) of the three films were comparable. In the CNF film, the crystallinity index was 88.2% whereas the crystallinity indices of FLC-CNFs and FLC were 91.4% and 87.9%, respectively. It is important to note that the major reason for the change in the crystallinity indices is generally the variation in the crystallite size (French and Santiago Cintrón 2013) of the cellulosic portion and not the crystallinity. The Segal’s method for calculating the crystallinity index of the cellulosic materials is a relative tool for comparing the differences between samples, however, the method should not be used for estimating the amount of crystalline and amorphous domains in the samples for reasons discussed further by Park et al. (2010) and French and Santiago Cintrón (2013).

Fig. 4
figure 4

XRD patterns of a CNF, b FLC-CNF, and c FLC films. Annotations within parentheses correspond to the cellulose portion and annotations between brackets correspond to the calcium carbonate portion

Full size image

Mechanical properties

The films were tested for their tensile properties and the results are shown in Fig. 5. As Fig. 5a shows, the films prepared from pure CNFs exhibited an average tensile modulus of 4.44 GPa whereas FLC-CNF and FLC films had average tensile moduli of 2.63 and 2.76 GPa, respectively. The statistical analyses showed that this decrease in the tensile modulus in the films that contain calcium carbonate was significant at 95% confidence level. However, there was not a significant difference between the FLC and FLC-CNF samples at the same confidence level.

Fig. 5
figure 5

Tensile properties of films prepared from CNFs (average density: 1.24 g/cm3), FLC-CNFs (average density: 1.15 g/cm3), and FLC (average density: 1.04 g/cm3). a tensile modulus, b tensile strength, and c strain at break. Columns with common letters are not statistically different at 95% confidence level

Full size image

The tensile strength (Fig. 5b) of films prepared from the mixture of CNFs and FLC was not significantly different from the pure CNF films. However, the tensile strength of FLC films (without the additional CNF fraction) was significantly lower compared to the other two samples. The strain at maximum stress, expectedly, decreased when the fraction of the calcium carbonate increased in the film (Fig. 5c). It was also observed that the densities (at 24 °C) of the CNF, FLC, and FLC-CNF films were 1.24 g/cm3, 1.04 g/cm3, and 1.15 g/cm3, respectively. One could argue that the decreased modulus and strength of the FLC samples was attributed to the decrease in the density of the films. On one hand, this could be a reasonable argument, however, it should also be noted that in the FLC-CNF sample, the decrease in the density did not have the same negative impact (on the tensile strength). Put differently, the negative effect in such a case is likely to be mostly attributed to the high load of the mineral than the density of the films.

Focusing now on the tensile strength, it is apparent that when calcium carbonate co-existed with CNFs, a similar (to CNFs) tensile strength was maintained. Previous studies highlighted the negative effect of mineral filler on the tensile strength of paper (Mörseburg and Chinga-Carrasco 2009; Chauhan and Bhardwaj 2014; Hubbe and Gill 2016). Typically, in such cases, minerals are often accompanied by other strength-enhancement additives such as starch or starch derivatives. The study herein, where no starch is added, shows that at a load of 25 wt% calcium carbonate, the tensile strength was not negatively affected. At a higher load of calcium carbonate (50 wt%) in the film, and in the absence of additional CNFs, the tensile strength decreased significantly. This result may be explained by the fact that the CNFs acts as ‘bridging network’ enabling the encapsulation of calcium carbonate particles within the CNFs leading to favorable tensile strength even in the presence of relatively high mineral load (e.g. 25 wt%). He et al. (2016) observed an increase in the tensile strength of the paper when precipitated calcium carbonate (PCC), CNFs, and starch composite filler were added to the paper. It is important to note that He’s system included other additives as well such as poly-aluminum chloride (PAC), anionic polyacrylamide, and micro-polymer. The authors, however, believed that the improved bondability between CNFs and PCC was due to cationic starch and that the network of CNFs also offered better strength properties and denser sheets. We believe that this promising result could pave the way for further advancement in many applications. Previous studies by Amini et al. (2017) and Leng et al. (2016, 2017) showed that particleboard produced from CNF binder (10–20 wt%) met or exceeded the American National Standard Institution (ANSI) 208.1 minimum specifications for low and medium density panels. Therefore, to assess the bonding characteristics of CNF and calcium carbonate binder, the lap shear strength was evaluated.

The subsequent set of mechanical analysis is an assessment of adhesion properties of the composite (Fig. 6a) using OCC paper as substrates via a lap shear test. Since adhesion properties are significantly affected by the properties of the used substrate, Table 3 summarizes the properties of OCC paper used for the test. During lap shear test, the applied tension load was parallel to the machine direction of OCC substrate to preclude any detachment along the substrate as the tensile strength of OCC in machine direction is three times that in the transverse direction.

Fig. 6
figure 6

a Lap shear strength of film specimens, b paper substrate and glue-line assembly, and c SEM images of bond failure. Columns with common letters are not statistically different at 95% confidence level

Full size image
Table 3 Summary of OCC substrate properties
Full size table

The lap shear strength for the CNF, FLC, and FLC-CNF samples ranged between 0.54 and 0.62 MPa. The commercial wood glue and softwood pulp exhibited lap shear strengths of 1.17 and 0.08 MPa, respectively. According to one-way ANOVA and Duncan’s post hoc statistical tests, the tested samples could be grouped into three categories that are statistically different at 95% confidence level: (1) softwood pulp, (2) CNF, FLC, FLC-CNF, and (3) commercial wood glue.

By comparing the strengths obtained from softwood pulp and CNF one can clearly observe the considerable effect of the size reduction (i.e. an increase in the surface area) in promoting a higher bonding strength. The ability of CNF to bond lignocellulosic materials was previously demonstrated when used as a sole binder for particleboard (Amini et al. 2017) and the results showed that the produced panels met or exceeded the minimum ANSI standard for low-density particleboard. However, the effect of the presence of calcium carbonate with CNF when used as binder was not known. This lap shear testing set up is a simple test that offers information on the behavior of CNF-containing calcium carbonate as a binder for lignocellulosic materials.

It was found that despite the presence of the calcium carbonate (regardless of its fraction), a favorable lap shear strength was obtained. This finding is almost consistent with the previously discussed tensile strength results. It could conceivably be hypothesized that despite the decrease in fiber–fiber interaction due to the presence of mineral particles, this decrease in the interaction was compensated by the web-like (Fig. 6b) structure of CNFs and MFC which held the calcium carbonate particles within. Figure 6c shows the SEM of the surface of the glue-line after bond failure. The bond failure could be described as a hybrid between adhesive failure and substrate failure. Although the majority of the failure could be attributed to adhesive failure, fibers from the substrate were also noticed on the adhesive films. In the case of commercial wood glue, all the failure occurred within the substrate whereas an adhesive failure was observed in all the samples bonded with softwood pulp.

Thermal analysis

Figure 7 shows the thermogravimetric analysis (TGA) curves and their corresponding first derivative curves (DTGA) of film samples. It is seen in Fig. 7a that the degradation of CNFs exhibited two regions. The first region (room temperature to ~ 100 °C) was attributed to the loss of moisture in the films. Following this, there was a steep degradation (278–340 °C) attributable to the thermal decomposition of the glycosidic bonds of cellulose (Poletto et al. 2014). At the end of the tested temperature range (600 °C), 15% of char residue remained from the CNF sample. In the case of FLC-CNF and FLC samples, the onset of the steep degradation was shifted towards a higher temperature, 283 °C and 292 °C, respectively, relative to CNF film (278 °C). The typical onset of degradation for calcium carbonate is around 600 °C (Li et al. 2017). This shift in the onset temperature might be related to the presence of calcium carbonate particles which protected the surrounding fibrils from the degradation at a relatively lower temperature. It is worth reminding that in FLC, the cellulosic portion was microfibrillated cellulose and not cellulose nanofibrils which could lead to slight differences in the thermal profile. In this regard, it could also be suggested that the relatively larger size of microfibrillated cellulose could contribute to the shift in the onset of degradation. Caution must be applied, however, with such interpretation as the CNFs used in this study could also contain portions of microfibrillated cellulose.

Fig. 7
figure 7

a TGA and b DTGA curves of film samples

Full size image

Furthermore, in the calcium carbonate-containing films, a distinctive feature in the thermal decomposition profile was noticed between 340 and 550 °C. This feature was more recognizable in the DTGA curves (Fig. 7b). The Figure shows that this additional degradation peak was broader in the case of FLC-CNFs than in FLC. These additional peaks could be explained by the delay in the degradation of the carbohydrate portion caused by the calcium carbonate particles. Stated differently, the calcium carbonate particles act as protective elements for the fibrils which retard the thermal degradation of the fibers to a higher temperature. The broadness of this additional peak in the FLC-CNFs compared to the FLC samples may be explained by the fact that the FLC-CNF samples contain more fibers as opposed to the FLC. In other words, the calcium carbonate particles could still protect the fibers even at a higher fiber to mineral ratio.

Filtration (dewatering) analysis

The assessment of the dewatering is of great importance as it reflects an economical aspect for many applications. What is interesting in the result (Fig. 8a) is the large variability in the dewatering pattern of FLC sample compared to the other two samples (i.e. the samples that contain CNFs). The variability ranged from rapid dewatering (leveling off after 10 min) to slow dewatering (continued increase in the rate during the entire filtration time).

Fig. 8
figure 8

a Filtrate mass as a function of time and b the rate of filtration over time

Full size image

This inconsistency may be due to the ratio of the calcium carbonate to the fiber. In the FLC sample, there was 50 wt% calcium carbonate. In the FLC-CNFs and CNFs, there were 25 wt% and 0 wt% calcium carbonate, respectively. In other words, there was less fiber entities in the FLC sample, which means that more free (unbound to the fiber) water was available. During the filtration, and at this higher portion of free water, there are two possible events. First, when air pressure was applied (172 kPa), the small particles of calcium carbonate rush to the pores of the filter paper (2.7 µm) with the water current leading to a drastic increase in the rate of filtration. The second scenario is when the pores of the filter paper are blocked by an agglomeration of calcium carbonate particles. The extent of blockage could vary depending on the size of the agglomerated particles. This was also supported by the fact of the increased turbidity of the filtrate of the sample that was dewatered at the highest rate. The turbidity decreased drastically when the rate was very slow.

This suggested effect may not be obvious when there was 25 wt% of calcium carbonate in the sample (FLC-CNF sample). The consistency in the CNF-containing samples could be attributed to the web-like network of CNFs and MFC (discussed in surface and structural evaluation section, Fig. 2) which entraps the calcium carbonate particles inside, hence reducing the free motion of the calcium carbonate particles and regulating the dewatering process.

Turning now to the average rate (between every two points) of the FLC-CNF and CNF samples (Fig. 8b), the rate of dewatering of the FLC-CNF sample was higher in the first 15 s. The dewatering rate of the FLC-CNF sample remained slightly higher than that of CNFs until 12–13 min. Following this, the two samples were dewatered at the same rate. Even though the presence of calcium carbonate showed, expectedly, an increase in the rate of dewatering, the unanticipated finding from this result was the effect of the fiber/mineral ratio on the dewatering consistency at this specific pressure.

Table 4 summarizes the retention of fibers/mineral after film formation and dewatering experiment and also shows the retention of individual components in each system. The results agree with the findings from previous studies, in which the positive effect of CNF in increasing the retention of added fillers such as calcium carbonate was demonstrated (Ämmälä et al. 2013; He et al. 2016). The role of CNF in increasing the retention of calcium carbonate in FLC-CNF samples was evident during film formation. In FLC films, there was a loss of 10% during the formation of the film, this loss was attributed mostly to calcium carbonate rather than the cellulosic fibers (9% vs 1%). During the dewatering experiment, the retention of fibers/mineral for FLC and FLC-CNF samples was comparable, indicating a less obvious effect of CNF in such a case. Between the two systems (film formation vs dewatering analysis), the possible interference of filter paper pore size, starting dry mass, and the type/amount of pressure applied cannot be ruled out.

Table 4 Fibers and mineral retention after film formation and dewatering experiment
Full size table

The total dry mass for the formed films was 1 g whereas the dewatering experiment used 3 g dry mass of suspension. The film formation process used filter paper of pore size 2.5 µm and vacuum pressure of 86 kPa. The dewatering experiment, however, used filter paper of pore size 2.7 µm and air pressure of 172 kPa. The increased pressure and pore size in the dewatering could be the reason for the decreased overall retention of fibers/mineral in the FLC-CNF sample. The dewatering of FLC sample showed, however, an opposite trend compared to that of film formation. The retention increased from 90 to 94% with a combination of loss from both fibers and mineral possibly due to the increased variability induced by the relatively higher pressure applied during the dewatering experiment.

Conclusions

This study demonstrated the feasibility of producing composite films from nano/microfibrillated cellulose and calcium carbonate without sacrificing their mechanical properties. The web-like structure of CNFs and MFC was key to maintain similar (to pure CNF film) mechanical properties in the composite films. The three samples showed no significant difference in their lap shear strength indicating that the presence of calcium carbonate particles did not inhibit the binding properties of CNFs. The presence of the mineral also increased the thermal stability of the films and retarded the thermal degradation of the cellulosic portion to higher temperatures. The presence of CNFs in FLC-CNF samples has maximized the retention of solids during film formation and the fibers to mineral had a substantial effect on the consistency of the dewatering behavior. Taken together, these results suggest that a hybrid CNF-based binder could be formulated at a lower cost, leading to better prospects for the commercialization of cellulose nanomaterials and an increase in the economic viability and commercialization of bio-based binders. The study also expands the traditional range of applications especially those that require a high mineral load without sacrificing the mechanical properties.