1 Introduction

Membrane separation processes are attractive because of their low energy cost as well as their ability to deliver contaminant-free final product [1,2,3]. Recently, interest in using natural materials for membranes has increased because of their biocompatibility and environmentally friendly disposability. The success of synthetic biopolymers can be attributed to their wide range of mechanical properties and the development of transformation processes that allow a variety of different shapes to be easily produced inexpensively as well as their good biocompatibility [4,5,6].

Sodium alginate is a biopolymer that can be possibly used as an environmentally compatible material for medical, food, and industrial applications. It can be sustainably and easily produced from kelp, which is cultivated in oceans worldwide [7,8,9].

The ability of sodium alginate to form gels with the aid of metal ion exchange (ex. Ca2+) has been extensively applied to form gel particles and membranes [10,11,12]. The molecular size of the alginate chain is generally not controlled because it is a natural biological product. To improve biopolymer characterization, using a multiphase composite material from a plural biopolymer showed potential to be a promising approach to prepare a tailor-made membrane.

Cellulose is used to produce potential reinforcing bionanomaterials called cellulose nanofibers (CNF) [13,14,15]. CNF have been researched for their use in biodegradable packaging because they are renewable, narrow diameter, and high mechanical strength [16,17,18]. Additionally, they have the potential to be used as agents in transparent composites, polymer hybrid-gels, and flexible conductive composites [19, 20]. Interest in the fabrication of films using CNF exists because they have the ability to form a cross-linked network in a very dilute aqueous suspension. All these important characteristics have made CNF an attractive candidate in the field of nanomaterial research. Recently, membranes composited with cellulose nanofibers have gained considerable attention because is a naturally occurring material that has intrinsic and structural advantages such as excellent mechanical properties and hydrophilicity. However, to the best of the authors’ knowledge, composited membranes composited with cellulose nanofibers was not previously investigated for selective mass transfer. Using biopolymer-based composite membranes contributes to a more sustainable society. Cellulose nanofibers are compatible with multiphase calcium alginate membrane due to their structural similarity. Thus, their composites are expected to have the physicochemical properties of alginate acid and the mechanical properties of cellulose nanofibers.

In this study, the authors examined the mechanical strength and mass transfer characteristics of a multiphase composite membrane that was prepared using calcium alginate based biopolymer membranes composited with CNF.

2 Experiments

2.1 Materials

Sodium alginate, calcium chloride, urea, glucose, methyl orange, and Bordeaux S were purchased from Wako Pure Chemical Industries, Ltd. (Osaka, Japan). Cellulose nanofibers were purchased from Sugino Machine, Ltd. (Toyama, Japan).

2.2 Methods

2.2.1 Preparation of calcium alginate membrane composted with cellulose nanofibers

An aqueous solution of sodium alginate (10 g/L) was prepared using distilled water, and a desirable amount of CNF was added to this solution. The mixed solution (20 mL) was poured into a glass petri dish (d = 8.3 cm) and then dried in a thermally controlled oven chamber regulated at 333 K for 12 h. The dried sodium alginate with composite CNF from the petri dish was then immersed in a 0.1 mol/L CaCl2 solution (25 mL) to cross link the alginate polymer chain. After 20 min at room temperature (T = 298 K), the membrane was separated from the glass dish. Subsequently, it was washed with distilled water to remove excess metal ions. The additional rate of CNF, RCNF (%), was evaluated by Eq. (1). In this study, calcium alginate membranes were prepared with the following ratios of cellulose nanofibers: 0%, 5%, 10, 15%, and 20%.

$$ R_{CNF} = \frac{Cellulose\,nanofiber\,\left( \text{g} \right)}{{Sodium\,alginate\left( \text{g} \right)}} \times 100 $$
(1)

2.2.2 Scanning electron microscopy

The membranes were snap-frozen in liquid nitrogen then dried in a vacuum-freeze dryer (RLE-103, Kyowa Vacuum Engineering. Co., Ltd., Tokyo, Japan) at 298 K for 24 h. The membranes were then sputter-coated using a thin Pt-membrane on a sputter-coater (E-1010 Ion Sputter, Hitachi, Ltd., Tokyo, Japan). Images of the membrane cross sections were obtained using a scanning electron microscope (SEM, Miniscope TM-1000, Hitachi, Ltd., Tokyo, Japan).

2.2.3 Mechanical strength of the membrane

The mechanical strength of swollen membranes was measured using a rheometer (CR-DX500, Sun Scientific Co., Ltd., Tokyo, Japan). The swollen membrane was cut into sample pieces (1 cm × 4 cm) that were stretched at a speed of 1 mm s−1. These tests were replicated three times. The maximum stress, δ (Pa), and the maximum strain, λ (%), at membrane rupture were calculated using Eqs. (2) and (3).

$$ \delta = \frac{{B_{max} }}{{A_{c} }} $$
(2)
$$ {\uplambda } = \frac{{L - L_{i} }}{{L_{i} }} \times 100 $$
(3)

where Bmax (N) is the load at membrane rupture, Ac (m2) is the initial membrane cross-sectional area, Li (m) is the initial length of the membrane and L (m) is length of the membrane at break.

2.2.4 Volumetric water content

To determine the inner structure of the swollen membrane, analysis was conducted by estimating the volumetric water content from the water content of the membrane. The swollen membrane was cut into squares (4 cm × 4 cm) and contained water in its void spaces. Therefore, volumetric water content was reasonably equivalent to the void fraction in the swollen-state membrane. Excess water on the surface was removed using filter paper, followed by drying in a thermally-controlled oven at 333 K for 24 h. Water loss was gravimetrically measured using an electronic balance. These tests were replicated three times. Volumetric water content was obtained by recalculating the gravimetric change in the membrane. The volumetric water content in the swollen membrane, Hv (–), was estimated by Eq. (4).

$$ H_{V} = \frac{{V_{w} }}{{V_{m} }} $$
(4)

where Vw is the volume of water in the membrane, and Vm is the volume of the swollen membrane.

2.2.5 Water permeation flux

The water permeability of the membrane was determined using an ultra-filtration apparatus (UHP-62K, Advantec Tokyo Kaisha, Ltd., Tokyo, Japan) [21]. The water permeability area of the membrane was estimated by the ultra-filtration apparatus to be 2.21 × 10–3 m2. The initial volume of distilled water was constant at 190 mL. The mass of permeated water with running time was measured by an electric balance, and the volume of permeated water was obtained by recalculation using the density of water. Pressure was applied using nitrogen gas introduced from a cylinder. These tests were replicated three times. The water permeation flux, Jv, was measured at 298 K and was defined by Eq. (5).

$$ J_{V} = \frac{{V_{p} }}{{A_{m} \cdot t}} $$
(5)

where Vp is the volume of permeated water, Am is the membrane surface area, and t is the operating time.

2.2.6 Mass transfer characterization

The mass transfer characterization of the membrane was evaluated from the effective diffusion coefficient of target components. Mass transfer cells were made by sandwiching the examined membranes between two pieces of glass. Aqueous solutions containing each of the target components and stripping water, of the same volume (190 mL), were fed into each cell. Urea (60 Da), d-glucose (180 Da), methyl orange (327 Da), and Bordeaux S (604 Da) were employed as the target components. The aqueous phase was stirred thoroughly (above 850 min−1) to ignore film mass transfer resistance. Under this condition, the overall mass transfer coefficient KOL is the directly the mass transfer coefficient. It was calculated using Eq. (6). Effective diffusion coefficient, Deff (m2 s−1) was evaluated by Eqs. (7) and (8).

$$ \ln \left( {1 - \frac{{2C_{s} }}{{C_{fi} }}} \right) = - 2\frac{{A_{m} }}{V} \cdot K_{OL} \cdot t $$
(6)
$$ K_{OL} = k_{m} $$
(7)
$$ k_{m} = \frac{{D_{eff} }}{{l_{m} }} $$
(8)

For comparison, the diffusion coefficient in bulk solvent, D (m2 s−1) was estimated according to the Wilke–Chang equation.

$$ D = 1.86 \times 10^{ - 18} \frac{{(\varphi M)^{0.5} T}}{{\mu \nu^{0.6} }} $$
(9)

where M is the molar mass of the solvent, ν is the molar volume of the solute at the boiling point, μ is the viscosity of the solvent, and φ is the association factor for the solvent at the required temperature T. The value of φ is 2.6 for water.

In this study, tortuosity was calculated using the following equations as diffusion models.

$$ D_{eff} = \frac{D \cdot \varepsilon }{\tau } $$
(10)
$$ \tau = \frac{{D \cdot H_{V} }}{{D_{eff} }} $$
(11)

where τ is the tortuosity of the membrane, and ε is the void fraction. Because HV was assumed to be the void fraction ε of the swollen membrane, it was substituted in Eq. (11).

3 Results and Discussion

3.1 Morphology of the CNF composite membranes

Figure 1 shows images of the CNF composite membranes. A stable CNF composite membrane (RCNF = 20%) was successfully prepared using this method. The membrane without CNF (RCNF = 0%) was transparent. The CNF appeared to be well dispersed in the composite membrane [22].

Fig. 1
figure 1

Scanning electron microscope images of the CNF composite membrane

Full size image

Scanning electron microscope images of the cross-sectional morphology showed good dispersion and a tightly spaced structure in the calcium alginate membrane. In the case of the CNF composite membrane, uniform morphology was observed with small flakes appearing as dots that had a tendency to accumulate on the membrane. The appearance of the CNF composite membrane became rougher as the amount of CNF was increased. It speculated that the dense structure of the membranes is attributed to their rigid hydrogen-bonded network and high crystallinity.

3.2 Mechanical strength

Figure 2 shows the value of maximum stress and maximum strain at break of the membrane. The maximum stress gradually increased when the amount of CNF was increased. In contrast, the maximum strain at membrane rupture reduced. In general, the maximum stress increases and the maximum strain decreases at break decreases because the amount of cross-linking agent increases. In this study, the stress and strain curve of the nanocomposite membrane indicated the transition from ductile to plastic behavior when the amount of CNF was increased. The lowering of elongation at break is a common trend that is affected by the volume fraction of the added CNF, the dispersion in the matrix, and the interaction between the CNF and the matrix. The maximum stress of an amylopectin membrane increased by incorporation of CNF into the membrane [23]. Likewise, Nanofiber showed the associated structure to form loose, tree-dimensional new works [24].

Fig. 2
figure 2

Change of maximum stress and maximum strain with the CNF composite membrane

Full size image

3.3 Volumetric water content

Figure 3 depicts the relation between the volumetric water content of the membrane and the membrane thickness. HV decayed linearly and the membrane thickness slightly increased when RCNF increased. The value of HV reasonably indicated the void fraction in the swollen membrane because the water in the membrane was contained in the voids of the polymer framework of the swollen membrane. The membrane had a greater tendency to not exhibit cellular structures containing free water in void space when the amount of CNF was increased [25].

Fig. 3
figure 3

Effect of CNF on the volumetric water content of calcium alginate membrane and membrane thickness

Full size image

3.4 Water permeation flux

Figure 4 shows the relation between the water flux and operational pressure acting on the membranes. The water flux of both membranes increased linearly when the operating pressure was increased. The presence of CNF in the calcium alginate membrane noticeably increased the water flux at the same pressure.

Fig. 4
figure 4

Volumetric water flux of CNF composite membrane at different pressures

Full size image

This is a commonly observed result in the performance of CNF [26, 27] and is in accordance with Hagen –Poiseuille flow. It can be expressed via non-equilibrium thermodynamic theory using Eq. (10).

$$ J_{V} = L_{P} \left( {\Delta P - \sigma \Delta \Pi } \right) $$
(12)

where Lp (m3water m−2area Pa−1 s−1) is the water permeability coefficient, \(\Delta P\) is the operational pressure, σ is the reflection coefficient of the solute, and \(\Delta \Pi\) is the osmotic pressure difference. When σ equals zero, Jv is linearly proportional to \(\Delta P\). The value of Lp of RCNF = 0% was 2.99 × 10–8, and that of RCNF = 20% was 8.13 × 10–9. The permeability of the composite CNF membrane was higher. The value of Lp for the CNF composite membrane was less than that of a chitosan membrane.

3.5 Mass transfer characterization

The effective diffusion coefficient versus the molecular weight of the tested component has been shown in Fig. 5. The effective diffusion coefficient was logarithmically decayed with increasing molecular weight. These results suggest formation of intermolecular hydrogen bonds between the hydroxyl groups of the cellulose nanofibers and carboxyl groups of the alginate. It is hypothesized that Ca2+/alginate membranes can form in CNF and decrease the membrane pore size.

Fig. 5
figure 5

Effect of molecular weight on the effective diffusion coefficient (Deff)

Full size image

For the CNF composite membrane, Deff dramatically decreased 2.5 × 103 times when the molecular weight increased 10 times. For the membrane without CNF, Deff decreased 1.7 × 103 times. In contrast, D in the bulk aqueous phase decreased only 3.8 times. This result suggests that the mass transfer channel in the CNF composite membrane was smaller than that in the membrane without CNF. Nomoto and Imai reported a large dependence on molecular size for specific polymer frameworks using a chitosan membrane [28]. Many food ingredients such as amino acids, polyphenols, and saccharides, have a molecular weight within the range tested (60 < MW < 600). Therefore, the CNF composite membrane is expected to be useful for the selective separation of food ingredients.

Figure 6 presents the tortuosity of methyl orange on the amount of CNF in the membrane. The tortuosity of the CNF composite membrane increased when the amount of CNF was increased. This result indicates that the polymeric framework of the membrane was more densely populated when the amount of CNF was increased.

Fig. 6
figure 6

Effect of the tortuosity on the CNF composite membrane

Full size image

4 Conclusion

A multiphase calcium alginate membrane composited with CNF was successfully prepared using a casting method. The maximum stress gradually increased when the amount of CNF was increased. The CNF composite membrane has sufficient mechanical strength to be repeatedly used. For the CNF composite membrane, Deff dramatically decreased 2.5 × 103 times when the molecular weight increased 10 times. Many functional food ingredients such as sugars, amino acids, and polyphenols are found in this special molecular weight range (60–604 Da). Therefore, high potential exists for the selective separation of glucose using a CNF composite membrane for application in downstream processing in food chemical engineering.