Journal of Materials Science

, Volume 47, Issue 18, pp 6724–6732

Arabinoxylan/nanofibrillated cellulose composite films


    • INNVENTIA AB, Fibre and Material Science
    • Wallenberg Wood Science CenterThe Royal Institute of Technology
  • Elina Mabasa Bergström
    • INNVENTIA AB, Fibre and Material Science
    • Wallenberg Wood Science Center, Department of Chemical and Biological EngineeringChalmers University of Technology
  • Paul Gatenholm
    • Wallenberg Wood Science Center, Department of Chemical and Biological EngineeringChalmers University of Technology
  • Lars Berglund
    • Wallenberg Wood Science CenterThe Royal Institute of Technology
  • Lennart Salmén
    • INNVENTIA AB, Fibre and Material Science
    • Wallenberg Wood Science CenterThe Royal Institute of Technology

DOI: 10.1007/s10853-012-6615-8

Cite this article as:
Stevanic, J.S., Bergström, E.M., Gatenholm, P. et al. J Mater Sci (2012) 47: 6724. doi:10.1007/s10853-012-6615-8


There is an increasing interest in substituting petroleum based polymer films, for food packaging applications, with films based on renewable resources. In many of these applications, low oxygen permeability and low moisture uptake of films are required, as well as high enough strength and flexibility. For this purpose, rye arabinoxylan films reinforced with nanofibrillated cellulose was prepared and evaluated. A thorough mixing of the components resulted in uniform films. Mechanical, thermal, structural, moisture sorption and oxygen barrier characteristics of such films are reported here. Reinforcement of arabinoxylan with nanofibrillated cellulose affected the properties of the films positively. A decrease in moisture sorption of the films, as well as an increase in stiffness, strength and flexibility of the films were shown. From these results and dynamic FTIR spectra, a strong coupling between reinforcing cellulose and arabinoxylan matrix was concluded. Oxygen barrier properties were equal or better as compared to the neat rye arabinoxylan film. In general, the high nanofibrillated cellulose containing composite film, i.e. 75 % NFC, showed the best properties.


Production of environmentally friendly packaging materials, based on polymers from renewable resources, is increasingly demanded. Films made of plant-derived non-food polysaccharides, such as cellulose and hemicelluloses (i.e. xylans and glucomannans), excluding storage polysaccharides, such as starch, may provide a potential alternative for today’s petroleum based packaging materials. These polysaccharide polymers are hydrophilic, and have good barrier properties against oils and fats, but are by its nature less efficient as moisture and water vapour barriers [1]. Also, oxygen barrier properties are good at low or moderate relative humidities, particularly for xylans [24]. The mechanical strength of pure hemicellulose-based films does however vary considerably and are generally considered to be too low without additives. One possibility to enhance the performance of such hemicellulose films is by reinforcing them with cellulose, thus initiating this study.

Arabinoxylans are a group of polysaccharides that can be found in almost all annual plants, and many woody plants, such as softwoods [5, 6]. The function of the arabinoxylans is to make up part of the matrix material in the plant cell wall. The chemical structure of arabinoxylans varies significantly depending on their source. The backbone of arabinoxylans is built up of (1 → 4)-linked β-d-xylopyranosyl units, which are further substituted with varying degrees by α-l-arabinofuranosyl groups [7, 8]. Other substituents found in arabinoxylans are α-d-glucopyranosyl uronic acid (or its 4-O-methyl ether), d-xylopyranosyl and acetyl groups [7, 8]. Cereal cell walls, such as from barley, husk, oat and rye, are rich in arabinoxylans and therefore an attractive source since they represent non-food fractions and a potential agricultural waste. In rye (Secale cereale) endosperm arabinoxylan, the β-d-xylopyranosyl units are both mono- and di-substituted by (1 → 2)- and (1 → 3)-linked α-l-arabinofuranosyl units (Fig. 1). The ratio of mono and disubstituted xylopyranosyl residues is 2:1 and the average arabinose to xylose (Ara/Xyl) ratio is between 0.50 and 1.00 [3].
Fig. 1

Chemical structure of rye arabinoxylan

Nanofibrillated cellulose (NFC) is a specifically prepared cellulosic material composed of liberated semicrystalline nanosized cellulose microfibrils with a high aspect ratio (i.e. length to with ratio). The lateral dimension of the NFC nanofibrils and nanofibril aggregates is in the order of 5–20 nm, respectively [9] and, if further aggregated, up to 40 nm [10] (cf. for kraft pulp fibres, the lateral dimension of microfibrils and microfibril aggregates is about 4 and 20 nm, respectively [11]), a value highly dependent upon the preparation technique used. Most frequently, various chemical and/or enzymatic pre-treatments before an intensive mechanical fibrillation are used for preparation of the NFC. The pre-treatments are introduced to purify the origin material from other cell wall components, such as hemicelluloses and lignin, but also to significantly reduce the energy consumption during the microfibril liberation process [12]. The NFC, at the time termed microfibrillated cellulose (MFC), was first introduced by Turbak et al. [13] and Herrick et al. [14]. An improved MFC was later prepared [9, 15] using a combination of mechanical and enzymatic pre-treatments, followed by a high pressure homogenization. The NFC used in the present study was prepared according to such a procedure outlined by Henriksson et al. [15]. This NFC forms a stable hydrocolloidal dispersion and forms a gel already at fairly low concentration. The gel consists of a strongly entangled and disordered network of cellulose nanofibrils and nanofibril aggregates. Filtration and drying can be used to form a nanopaper structure [16] in the form of a NFC fibril network of high tensile strength (>200 MPa), high stiffness (~15 GPa) and relatively high strain-to-failure (6–10 %). In addition, NFC from wood pulp is characterised by biodegradability, it is from renewable resources, and the existing infrastructure for commercial wood pulping also makes NFC attractive as an easily available resource to be used as reinforcement component for biocomposites.

The goal of the present study was to investigate the morphology, physical and mechanical properties of biocomposite films, based on rye arabinoxylan (rAX) and NFC, and to examine effects of addition of NFC.

Materials and methods

Preparation of rAX solution

A high molecular weight rAX (lot 20601a) was used (Megazyme International Ireland Ltd); molar mass (Mw) 289,300 g/mol [17]. The arabinose/xylose (Ara/Xyl) ratio of the rAX was 0.50. An aqueous arabinoxylan solution was prepared by dissolving the rAX flour, 2 g, (pre-wetted with 95 % C2H5OH) in deionised water, heating it at 100 °C under magnetic stirring for 10 min. The solution was allowed to cool down under magnetic stirring to room temperature and its concentration was adjusted to 2 % (w/v).

Preparation of nanofibrillated cellulose

NFC was prepared from a never-dried bleached sulphite pulp based on Norwegian spruce (Picea abies) (kindly provided by Nordic Paper Seffle AB, Sweden), containing 13.8 % hemicelluloses and 0.7 % lignin. A combined pre-treatment, i.e. an enzymatic degradation and a mechanical beating, was used, followed by a disintegration into NFC by a homogenisation, i.e. passing several times through a microfluidizer, as described by Henriksson et al. [15, 16] The solid content of the NFC (deionised water dispersion) was adjusted to 1.4 % (w/v). The charge density of the NFC was 50 μequiv/g, and it contained 0.8 % arabinoglucuronoxylan and 2.5 % galactoglucomannan (most likely a low Gal substituted fraction of galactoclucomannan). This lower amount of hemicelluloses as compared to the original pulp can be explained by the fact that much of the galactoglucomannan has been released upon the liberation of the cellulose microfibrils during the NFC preparation process. The crystallinity of the NFC may be estimated to around 8–12 % using NMR, based on previous studies [9], where the NMR crystallinity value corresponds to an X-ray crystallinity of ca. 50 %, due to the contribution of the paracrystalline material.

Preparation of films

The films were prepared by mixing the NFC dispersion with the rAX solution in different proportions, with the rAX/NFC ratios of 100/0, 75/25, 50/50, 25/75 and 0/100 (see Table 1). No external plasticizer was added.
Table 1

Codes and descriptions of rAX, NFC and composite films thereof; % is percent by weight




100 % Arabinoxylan


75 % Arabinoxylan + 25 % nanofibrillated cellulose


50 % Arabinoxylan + 50 % nanofibrillated cellulose


25 % Arabinoxylan + 75 % nanofibrillated cellulose


100 % Nanofibrillated cellulose

In order to produce homogenous films, a thorough mixing is necessary obtained by a homogenization and further dispersion of the rAX/NFC mixtures in three steps: (a) using an Ultra-Turrax® 9500 rpm for 5 min, (b) exposing the mixture to magnetic stirring for 3 days, and (c) once again using the Ultra-Turrax® 9500 rpm for 10 min; adopting method developed by Svagan et al. [10]. The resulting dispersions were exposed to vacuum under magnetic stirring for 24 h to remove air bubbles. The homogenized dispersions were then cast into Teflon (PTFE) coated polystyrene petri dishes (Ø 8.4 cm). These were left to evaporate at 23 °C and 52 % RH (relative humidity) for 48 h. The thickness of the resulting films was approximately 20 μm (measured using a micrometre screw gauge).

Scanning electron microscopy (SEM)

SEM images of the films, brittle fractured in liquid N2, were collected with a LEO ULTRA 55 FEG SEM (LEO electron microscopy LTD, Cambridge, UK) equipped with a field emission gun and an Inlens detector. The acceleration voltage was set to 5 kV. Prior to the SEM analysis, the films were glued to aluminium specimens, mounted with a colloidal silver liquid and sputter coated with a thin layer of gold.

Atomic force microscopy (AFM)

Surface roughness of the films was measured using a Digital Instrument Nanoscope IIIa AFM (Digital Instrument Inc., Santa Barbara, CA, USA), equipped with a type G scanner and a μMasch ultrasharp nanocontact silicon cantilever NSC15. Areas of 25 × 25 μm of the films were analysed in tapping mode. The surface roughness was expressed as mean roughness, Ra,  and calculated according to Eq. (1):
$$ {\text{Ra}} = \frac{1}{{L_{x} L_{y} }}\int\limits_{0}^{{L_{y} }} {\int\limits_{0}^{{L_{x} }} {\left| {f\left( {x,y} \right)} \right|{\text{d}}x{\text{d}}y} } , $$
where Lx and Ly are the dimensions of the surface and \( f\left( {x,y} \right) \) is the surface relative to the centre plane. The Ra is the arithmetic average of the absolute values of the measured profile height deviation taken within the sampling length and measured from the graphical centre line. It is the mean value of the surface relative to the centre plane.

Dynamic vapour sorption (DVS)

Moisture sorption isotherms were obtained with a Surface Measurement System Plus II Oven DVS (Surface Measurement System Ltd, Rosemont Road, UK). 4 mg material of each film was measured in adsorption, from 0 to 90 % RH. At each step of 0, 20, 40 and 60 % RH the humidity was kept constant for 300 min, at 80 % RH it was kept for 400 min, and at 90 % RH it was kept for 500 min. The weight increase, registered by a microbalance, was recorded as a function of the increase in humidity of the surrounding air. This device generates a moist air by mixing dry and water-saturated air streams at 30 °C. The results were based on a single measurement and generated using a DVS 32 Novideo software. The moisture uptake was expressed according to Eq. (2):
$$ {\text{Moisture}}\,{\text{uptake}} = 100\frac{{W_{\text{moist}} - W_{\text{dry}} }}{{W_{\text{dry}} }}, $$
where Wmoist is the sample weight equilibrated at the chosen relative humidity and Wdry is the weight of the dry sample.

Tensile testing

Mechanical testing (stress–strain scanning) was carried out on a Single Column Table Top Instron 5944 (Instron Ltd., Coronation Road, UK), with a load cell of 2 kN, operating in tension mode. Film stripes cut to the size of 5 × 50 mm (width × length), of 20 μm thickness, were used for testing at 50 % RH and 23 °C. The applied loading rate was 10 % of the span length/min, where the initial span length of films was 30 mm. Ten replicates from each film type were tested. The Young’s modulus (E) was taken as the initial slope of the linear part of the stress–strain curve. The stress at break (σb) and strain at break (εb) were evaluated based on the initial sample dimensions. Data were presented as average values from ten measurements.

Dynamic mechanical analysis (DMA)

Dynamic mechanical testing (humidity scanning) was carried out using a Perkin-Elmer DMA 7e (PerkinElmer Corp., Norwalk, CT, USA), operating in tension mode. Film stripes cut to the dimensions of 3 × 15 mm (width × length), of 20 μm thickness, were first conditioned at 5 % RH for 3 h; (the hysteresis was negligible due to the low moisture content of ca. 3 at 5 % RH). The samples were then scanned in the range of 5–95 % RH at a speed of 1 % RH/10 min at a temperature of 30 °C, a speed low enough to maintain equilibrium conditions during scanning [18]. The desired relative humidities were achieved by mixing dry air and water-saturated air using a Wetsys humidity generator (Setaram Instrumentation, Caluire, France). The static load was adjusted to be equal to 140 % of the dynamic load, the amplitude was set to be constant at 5 μm corresponding to a deformation of 0.033 %, at a frequency of 1 Hz. Three replicates from each film type were tested. The storage modulus (E’) and the loss tangent (tan δ) were recorded using a Pyris DMA 28 software (PerkinElmer Corp., Norwalk, CT, USA).

Dynamic mechanical testing (temperature scanning) was carried out using a Perkin-Elmer DMA 7e (PerkinElmer Corp., Norwalk, CT, USA), operating in tension mode. Film stripes cut to the dimensions of 5 × 20 mm (width × length), of 20 μm thickness, were first conditioned at a relative humidity of 0 % RH in a He atmosphere, a temperature of 30 °C for 30 min and then scanned in a range of 30–300 °C at a speed of 2 °C/min. The static load was adjusted to be equal to 120 % of the dynamic load, the amplitude was set to be constant at 3 μm, at a frequency of 1 Hz. Three replicates from each film type were tested. The storage modulus (E′) and the loss tangent (tan δ) were recorded using a Pyris DMA 28 software (PerkinElmer Corp., Norwalk, CT, USA).

Dynamic FTIR spectroscopy

Dynamic FTIR (Fourier Transform InfraRed) spectra were recorded on a Varian 680-IR spectrometer (Varian Inc., Santa Clara, CA, USA) in transmission mode. A liquid nitrogen cooled MCT (Mercury Cadmium Telluride) detector was used and the IR radiation was polarized by a KRS5 wire grid polarizer at 0° in relation to the stretching direction. An optical filter was added after the polarizer to reduce the spectral range (3950–700 cm−1). The interferometer was run in step-scan mode at a phase modulation frequency of 400 Hz (the moving mirror oscillates around each step point with a frequency of 400 Hz). Thin sheets with a dimension of 10 × 20 mm (width × length) and of 20 μm thickness were mounted between two parallel jaws in a specially constructed polymer modulator (PM-100). The PM-100 was placed in a temperature control system (TC-100) (MAT-Manning Applied Technology Inc., Troy, ID, USA). A modular humidity generator (MHG-32) (Projekt Messtechnik, Ulm, Germany) was connected to the TC-100. Spectra were recorded with a resolution of 8 cm−1; one scan per measurement. To ensure the linear deformation during the collection of the dynamic FTIR spectra, the amplitude of the applied sinusoidal strain was less than 0.3 % of the sample length at a frequency of 16 Hz. The dynamic changes in the IR spectrum were divided into two orthogonal spectra: the in-phase spectrum indicating immediate changes or elastic responses and the out-of-phase spectrum representing the time-delayed changes or viscous responses using Varian Resolutiom Pro 5.1 software (Varian Inc., Santa Clara, CA, USA). The resulting interferograms were Fourier transformed and a Norton–Beer medium apodization function was used. The in-phase and out-of-phase spectra obtained were base-line corrected at 1800 cm−1. They were then converted into a phase spectrum and a magnitude spectrum, where the converted magnitude spectrum was normalised to 1.0 at 1160 cm−1. New in-phase and out-of-phase spectra were calculated from these. Two parallel measurements were made at 50 % RH and 30 °C. The average in-phase and out-of-phase spectra were then calculated.

Oxygen permeability

Oxygen transmission rate (OTR) was measured according to an ASTM F—1927-07 standard method, using a Mocon Ox-Tran® 2/21, ML Master Base Control Module, (Mocon® 7500 Boone Avenue North, Minneapolis, Minnesota, USA) operating with a Mocon’s WinPerm™ permeability software. The measurements were done at controlled conditions of 50 % RH and 23 °C, using a coulometric detector. Duplicates from each film type of 20 μm thickness with an area of 5 cm2 were tested. Oxygen permeability (OP) was calculated from the film thicknesses and expressed as cm3 μm/m2 day kPa.

Results and discussion

Film formation and morphology

Upon casting from aqueous dispersions, both the rAX and the NFC (i.e. 100rAX and 100NFC, respectively) were able to form cohesive films without the use of any plasticiser added. In the case of the composite films (i.e. 25NFC, 50NFC and 75NFC), cohesive films were also produced. With the higher amount of NFC added, the films became successively less transparent and more opaque, while the 100NFC film seems as slightly less opaque than the 75NFC film, as illustrated in Fig. 2. All the films were homogenous without any visible particles present, as inspected by eyes.
Fig. 2

Optical images of films, from left to right: pure rye arabinoxylan film (100rAX), composite films: 25NFC, 50NFC, 75NFC, and pure nanofibrillated cellulose film (100NFC)

A similar trend was observed regarding surface roughness (AFM measurements, Table 2), where an increase was noted with the higher amount of NFC added. This increased surface roughness may contribute to the successive decrease in transparency of the films. The 100NFC film displayed a lower surface roughness than that of the 75NFC film, in line with the slightly higher transparency of the pure NFC film.
Table 2

Mean roughness (Ra) of rAX, NFC and composite films thereof







Surface roughness, Ra (nm)






The morphology of the films was studied by SEM. In Fig. 3, SEM micrographs of the fracture surfaces, obtained by breaking nitrogen frozen films, are shown. The 100rAX film displayed a continuous and homogenous structure. The composite films all showed somewhat more aggregated structures, which became less homogeneous and more layered at higher NFC content. The 25NFC film remained quite homogeneous in structure with visible and dispersed bundles of NFC embedded in the matrix of rAX. The 50NFC film displayed a more heterogeneous structure. The 75NFC film showed a more layered structure with thick NFC lamellae coated by thinner layers of rAX. The 100NFC film showed a clear lamellar structure with the lamellae composed of a random in-plane network of cellulose NFC. This lamellar structure of NFC films, composed of a fibrous network, has previously been reported by several authors [10, 16, 19].
Fig. 3

SEM micrographs of fracture surfaces (obtained by breaking nitrogen frozen films) of films; pure rye arabinoxylan film (100rAX), composite films: 25NFC, 50NFC, 75NFC, and pure nanofibrillated cellulose film (100NFC); scale bar is 2 μm

Moisture sorption isotherms

The moisture sorption behaviour of films is an important characteristic of hemicellulose films. The sorption isotherms of the films, the 100rAX, 100NFC and composite films, all displayed a sigmoidal shape (Fig. 4), normally reported for hydrophilic materials. This shape of the curves is characteristic for systems with strong polymer–polymer and polymer–solvent interactions, i.e. a type II isotherm [20]. At the higher relative humidities, the molecular interactions between cellulose and/or hemicellulose molecules become partly replaced by interactions with water molecules. In general, the water molecules are adsorbed around polar groups (i.e. –OH groups) in the amorphous regions of the molecules.
Fig. 4

Water vapour sorption isotherms of rye arabinoxylan film, nanofibrillated cellulose film and composite films made thereof; recorded at 30 °C

The NFC film displayed the lowest moisture adsorption, reflecting the crystalline nature of the microfibrillar cellulose. However, the present NFC contains a substantial amount of disordered cellulose as well as some hemicelluloses. The 100rAX film showed the highest moisture content, reflecting the highest amount of water sorption sites available. The composite films showed intermediate moisture sorption isotherms, where the moisture sorption isotherm successively decreased with increasing amount of NFC present.

If the moisture sorption data of the composite films are compared with the expected moisture uptake based on the rule of mixture (i.e. calculated on the basis of the moisture uptake contribution of each component) (Fig. 5), it is apparent that the composite films adsorb less moisture than expected. This indicates that strong permanent H-bonds and other interactions were created between the xylan and the cellulose NFC, decreasing the number of available sites for moisture sorption. A larger difference was seen for the 50NFC film, as compared with the other two composite films (i.e. 75NFC and 25NFC). This may be explained by a better mixing of the two components (i.e. xylan and cellulose) in this film (cf. Fig. 3).
Fig. 5

Difference between calculated moisture uptake and measured moisture uptake of the composite films versus RH

Tensile properties

Figure 6 shows average stress–strain curves of the different films tested at 50 % RH and 23 °C (see Table 3). The 100rAX film showed the lowest stiffness, strength and ductility of all films. The Young’s modulus was 3.1 GPa, a higher value than previously reported (i.e. 2.5 GPa at 30 °C [21] and 1.8 GPa at 23 °C [3]), which can be explained by the use of an improved homogenisation procedure in the present study. The composite films showed gradual increase in stiffness, strength and strain-to-failure with increased NFC reinforcement content. The 100NFC film, itself, had lower strength, stiffness and strain-to-failure than the 75NFC film.
Fig. 6

Stress–strain curves of rye arabinoxylan film (100rAX) and composite films of rye arabinoxylan reinforced with nanofibrillated cellulose: 25NFC, 50NFC, 75NFC; recorded at 50 % RH and 23 °C

Table 3

Average values of Young’s modulus (E), stress at break (σb) and strain at break (εb)


Young’s modulus, E (GPa)

Stress at break, σb (MPa)

Strain at break, εb (%)


3.1 ± 0.2

62 ± 3

4.0 ± 0.9


4.8 ± 0.7

108 ± 17

6.0 ± 1.1


6.3 ± 0.4

135 ± 31

6.4 ± 2.4


7.3 ± 0.5

143 ± 15

7.2 ± 1.2


6.6 ± 0.5

107 ± 14

5.7 ± 1.9

The low mechanical properties for the 100NFC film were most likely due to high porosity, since NFC films are fibrillar network structures. One could also observe lamellar separation in the SEM micrographs, see Fig. 3. When smaller denser regions were sampled for DMA measurements, much higher Young’s modulus was measured at 50 % RH (i.e. 11.5 GPa). This was more in line with previously reported data on homogenous films with 19 % porosity, showing mechanical strength properties of approximately 15 GPa Young’s modulus, 200 MPa tension strength and 7 % strain-to-failure [16]. With rAX, as an amorphous polymer [3, 21] present in the structure, a more uniform stress distribution results in the better strength properties.

Moisture sensitivity

The effect of relative humidity (humidity scans–absolute moduli, E′ and loss tangent, tan δ) on moduli and mechanical damping of the films are shown in Fig. 7. With increasing relative humidity, the stiffness of the films decreased, as expected for these hydrophilic materials, while the mechanical damping (tan δ) increased (above 85 % RH the increase was strong). The 100NFC film showed the highest storage modulus and the lowest values for tan delta. The 100rAX film showed the lowest storage modulus and the highest tan delta values. With increasing amounts of the NFC, the composite films showed increased storage modulus and data for the mechanical damping successively decreased. The 100NFC film showed a drop in the storage modulus from 13–7.7 GPa (5–90 % RH), which may be due to both fibril softening and decreased interfibril interactions. The change of the tan δ value with RH was negligible. The pure 100rAX film was strongly sensitive to moisture with the storage moduli dropping from 6.8 to 0.9 GPa (5–90 % RH), reflecting the amorphous and hygroscopic nature of rAX.
Fig. 7

Humidity scans of rye arabinoxylan film, nanofibrillated cellulose film and composite films made thereof; recorded at 1 Hz and 30 °C

Two transition points could be noted, a secondary transition around 20 % RH and a main transition [22] at around 80 % RH. The secondary transition was most apparent for the 100rAX film and may be attributed to the motion of the arabinose substituent [23]. The main transition at 80 % RH was attributed to the moisture-induced glass transition of the xylan [2426]. It is notable that already with an addition of 25 % NFC (25NFC film), the indication of the glass transition almost disappeared in the data. This may be due to possible interactions between NFC and arabinoxylan, limiting the moisture-induced mobility of the xylan chains [7].

In Fig. 8, the relative softening behaviour (humidity scans–relative moduli, E′) is illustrated for the 100rAX, 100NFC and composite films. Even the reinforcement with the lowest amount of the less hygroscopic NFC (25NFC film) showed considerably reduced moisture sensitivity. The low moisture sensitivity of the NFC results in that the reinforcing effect being substantial above the Tg of the arabinoxylan, i.e. above 80 % RH, is similarity with effects on composites above the matrix Tg [27, 28].
Fig. 8

Relative storage modulus of rye arabinoxylan film, nanofibrillated cellulose film and composite films made thereof, as a function of relative humidity; recorded at 1 Hz and 30 °C

Thermal properties

The thermomechanical properties (temperature scans–absolute moduli, E′ and loss tangent, tan δ) are shown in Fig. 9. With increasing temperature the stiffness decreased, and the mechanical damping increased for all films. The pure 100rAX film showed a distinct glass transition (Tg) at 200 °C. It was previously reported that dry hemicelluloses from wood typically have a Tg between 150 and 220 °C (the span depends on differences in chemical composition, configuration, side groups and molecular weight), while dry cellulose may show a Tg of amorphous parts between 200 and 250 °C (the span depends on the degree of crystallinity) [29]. For the pure NFC studied here, as well as for the 75NFC film, no transitions were noted in this temperature range. For the composite films the Tg shifted to higher temperatures with increasing amount of reinforcing NFC. This, again, points to the strong interactions between the cellulose and xylan.
Fig. 9

Temperature scans of rye arabinoxylan film, nanofibrillated cellulose film and composite films made thereof; recorded at a frequency of 1 Hz

Molecular interactions

Figure 10 illustrates dynamic FTIR spectra of the 75NFC film in the wavenumber region of 3950–700 cm−1, showing both the in-phase spectrum representing the elastic response of the components and the out-of-phase spectrum representing the viscous response of the components. The signals occurring in the in-phase spectrum are much more intense than in the out-of-phase spectrum indicating that, at 50 % RH, the film responded mostly elastically. Three absorption vibrations are highlighted in the in-phase spectrum, two cellulose signals at 1160 cm−1 (asymmetric C–O–C bridge stretching vibration) and at 1425 cm−1 (C–OH bending vibration of the CH2–OH group attached at Glc unit) [30, 31], and an arabinoxylan signal at 1460 cm−1 (CH2 symmetric bending on the Xyl ring) [3234]. The cellulose signal at 1160 cm−1 was used for normalization of the spectra. The signal at 1425 cm−1 is selective for cellulose, while the signal at 1460 cm−1 is selective for arabinoxylan, due to the fact that the functional groups absorbing IR radiation are unique for respective components. The elastic response from both components indicates that the two components are strained simultaneously with the applied sinusoidal perturbation. This shows that strong interactions between the arabinoxylan and the cellulose exist in the composite films.
Fig. 10

Dynamic FTIR spectra of the 75NFC film recorded at 0° polarization, 50 % RH and 30 °C; the in-phase spectrum (thin line) indicates elastic responses and the out-of-phase spectrum (thick line) indicates viscous responses

Oxygen barrier properties

The oxygen permeability of the neat rAX and its composite films showed very low values at 50 % RH (Fig. 11). The composite films with higher amount of the NFC showed even lower values than that of the pure rAX. This decreased permeation of oxygen of the composite films, 50NFC and 75NFC, are most likely due to the low permeability of NFC itself. The oxygen permeability of the 100rAX film in this study was 1.0 cm3 μm/m2 day kPa, which was lower than values earlier reported by Höije et al. (i.e. 2.0 cm3 μm/m2 day kPa for the same type of the material measured under similar conditions of 50 % RH and 23 °C [3]). This improvement may be explained by the enhanced film preparation procedure adopted here, where the dispersions were homogenised before casting. The oxygen permeability of the films was in the range of poly ethylene vinyl alcohol (EVOH), which is commercially used as a barrier plastic, having an oxygen permeability of 0.1 cm3 μm/m2 day kPa at 0 % RH which increases to 12 cm3 μm/m2 day kPa at 95 % RH [35, 36].
Fig. 11

Oxygen permeability of rye arabinoxylan film and its composite films; literature values for MFC (carboxymethylated) [19], rAX [3], EVOH (poly ethylene vinyl alcohol) [36], PVDC (polyvinylidene chloride) [37], polyester [36] and PET (poly(ethylene-terephthalate)) [37]; all tests done at 23 °C and 50 % RH, if not specified differently


This study demonstrates that, with the addition of NFC to rAX, it is possible to produce composite films with greatly improved mechanical, thermal, moisture sorption and excellent oxygen barrier properties. Thorough mixing of NFC with rAX resulted in homogenous composite films.

In general, stiffer, stronger and more ductile films were produced when the NFC was added to the arabinoxylan; with a Young’s modulus of up to 7.2 GPa, a tensile strength of up to 143 MPa and a strain-to-failure of up to 7.3 %. The addition of the NFC to the arabinoxylan films caused a shift of the Tg to higher temperatures than the 200 °C of the pure xylan. This indicates a substantial xylan interphase region with reduced molecular mobility. A strong coupling between cellulose and arabinoxylan was also indicated. The moisture sorption and the associated moisture-induced softening were both reduced with increased NFC reinforcement content. With the reinforcement, the oxygen barrier properties of the arabinoxylan films were also improved, showing an oxygen permeability of 0.8 cm3 μm/m2 day kPa at 50 % RH in a film composed of 50–75 % NFC by weight.


The Knut and Alice Wallenberg Foundation are gratefully acknowledged for funding through the Wallenberg Wood Science Center. The authors thank Dr Aihua Pei for assistance in procuring the NFC and for fruitful discussions on the film casting techniques, Anders Mårtensson for providing the SEM micrographs and AFM measurements, Anne-Mari Olsson for performing the DVS and Therese Johansson for performing O2 permeability measurements.

Copyright information

© Springer Science+Business Media, LLC 2012