Sei sulla pagina 1di 10

Modeling the Oxygen Diffusion of Nanocomposite-based Food Packaging Films

Kanishka Bhunia, Sumeet Dhawan, and Shyam S. Sablani


Abstract: Polymer-layered silicate nanocomposites have been shown to improve the gas barrier properties of food packaging polymers. This study developed a computer simulation model using the commercial software, COMSOL Multiphysics to analyze changes in oxygen barrier properties in terms of relative diffusivity, as inuenced by conguration and structural parameters that include volume fraction ( ), aspect ratio ( ), intercalation width (W ), and orientation angle ( ) of nanoparticles. The simulation was performed at different (1%, 3%, 5%, and 7%), (50, 100, 500, and 1000), and W (1, 3, 5, and 7 nm). The value was varied from 0 to 85 . Results show that diffusivity decreases with increasing volume fraction, but beyond = 5% and = 500, diffusivity remained almost constant at W values of 1 and 3 nm. Higher relative diffusivity coincided with increasing W and decreasing value for the same volume fraction of nanoparticles. Diffusivity increased as the rotational angle increased, gradually diminishing the inuence of nanoparticles. Diffusivity increased drastically as changed from 15 to 30 (relative increment in relative diffusivity was almost 3.5 times). Nanoparticles with exfoliation conguration exhibited better oxygen barrier properties compared to intercalation. The nite element model developed in this study provides insight into oxygen barrier properties for nanocomposite with a wide range of structural parameters. This model can be used to design and manufacture an ideal nanocompositebased food packaging lm with improved gas barrier properties for industrial applications. Keywords: gas barrier property, nanocomposites, polymer-layered silicate, relative diffusivity

Practical Applications:

The model will assist in designing nanocomposite polymeric structures of desired gas barrier properties for food packaging applications. In addition, this study will be helpful in formulating a combination of nanoparticle structural parameters for designing nanocomposite membranes with selective permeability for the industrial applications including membrane separation techniques.

Introduction

In recent years, the food industry has been faced with the challenge of meeting consumers demand for minimally processed, nutritious food with a prolonged shelf life. To meet this demand, proper packaging systems that make foods commercially available for an extended shelf life must be developed. Packaging materials, such as glass bottles and aluminum cans, have been popular in the past, but preparation of aluminum packaging requires large amount of energy and glass bottles often break during transport (Frounchi and Dourbash 2009). The food industry has turned their focus to polymer packaging during recent years. However, polymer packaging materials are not without their drawbacks. For example, they are more permeable to gases, water vapor, and aroma compounds than traditional glass, metal cans, and paper packaging (Sorrentino and others 2006). The migration of oxygen and water vapor into polymer packaging can lead to undesirable changes in food, such as deterioration of organoleptic properties (Lange and Wyser 2003; Minelli and others 2009). Recently, organicinorganic nanocomposites have attracted attention from researchers because they

exhibit improved gas barrier properties (Sorrentino and others 2006; Pereira de Abreu and others 2007) which reduces the diffusion of oxygen and water vapor across the matrix and helps in maintaining the quality of the food for several months, thereby increasing shelf life. Nanocomposite materials are basically a 2-phase system of polymer and inorganic ller with certain geometries (bers, akes, spheres, particulates) dispersed in a polymer matrix (Azeredo 2009). The crystalline inorganic ller particles belong to the family of 2:1 layered or phyllosilicates which consists of one octahedral aluminum oxide sheet sandwiched between 2 tetrahedral silicon dioxide sheets (Weiss and others 2006; Paul and Robeson 2008). The layered silicates are usually 1 nm thick, whereas the length of the individual sheet varies from 30 nm to several microns, depending on the type of llers. Dispersion of phyllosilicates depends on the strength of interfacial interaction between polymer and particles, and can result in the formation of 4 types of morphology, as noted by Ray and Okamoto (2003): aggregated, intercalated, occulated, and exfoliated. In intercalation, polymer chains swell into the interlayer region between the silicates, forming a well-ordered, multilayer structure with an alternating polymer/silicate layer (Weiss and MS 20120246 Submitted 2/17/2012, Accepted 4/24/2012. Authors Bhunia, others 2006; Lu and Mai 2007). The exfoliated nanocomposDhawan, and Sablani are with Dept. of Biological Systems Engineering, Washington ite results from the extensive incorporation of polymer, leading State Univ., P.O. Box 646120, Pullman, WA 991646120, U.S.A. Direct inquiries to the complete delamination of the silicates. The state of exto author Sablani (E-mail: ssablani@wsu.edu). foliation is considered to exhibit the best gas barrier properties
C 2012 Institute of Food Technologists R doi: 10.1111/j.1750-3841.2012.02768.x

Vol. 77, Nr. 7, 2012 r Journal of Food Science N29

Further reproduction without permission is prohibited

N: Nanoscale Food Science

Oxygen diffusion of Nanocomposite lms . . .


due to the highest aspect ratio of llers acting as a barrier (Sorrentino and others 2006) and also due to optimal interaction between the llers and the polymer matrix (Alexandre and others 2009). Apart from offering good gas barrier properties, nanoparticles have been shown to improve the mechanical properties of the nanocomposites (Ray and Okamoto 2003; Lotti and others 2008). Barrier property of a packaging material can be described in terms of permeability coefcient which is dependent on diffusion coefcient of gas and solubility coefcient of that gas in the polymer matrix (Kumar and others 2011). However, most studies have estimated the diffusion and permeability coefcient for nanocomposites rather than modeling the solubility coefcient. The diffusion process in the nanocomposite is inuenced by major structural parameters, including the volume fraction of ller, aspect ratio, orientation angle, and most importantly the relative dispersion of llers in the polymer matrix. The diffusion behavior in the nanocomposites was studied by several scientists (Maxwell 1873; Raleigh 1892; Nielsen 1967; Wakeham and Mason 1979; Aris 1986; Cussler and others 1988; Bharadwaj 2001; Swannack and others 2005; Statler and Gupta 2007; Minelli and others 2009). Some of the important work is summarized briey in the Appendix. The simplest model describing the diffusion through a ake-lled system as inuenced by tortuosity ( ) can be written as membrane/lm thickness (d) (Bharadwaj 2001) = d L =1+ d 2W (2)

where L and W is the length, and intercalation /aggregate width of the akes, respectively and is the volume/loading fraction. The intercalation width is equal to the width of the stack of akes assuming that polymer layer between each pair of akes is 1 nm thick. Thus, W value is equal to 1 nm when there is no aggregate formation, and single akes are dispersed in the polymer matrix. The above condition is known as state of full delamination or exfoliation (Figure 1). The expression for the relative diffusivity was given as (Bharadwaj 2001): D 1 1 = = D0 1 + (/2) (3)

where is the aspect ratio of nanoparticles and it can be dened as = L . W (4)

Angular orientation ( ) of the akes has a large inuence on the diffusivity. Tortuosity is considered to be highest when the D0 akes are positioned perpendicular to the direction of diffusion D= (1) (Figure 2). For the = 90 , a new order parameter, S, is incor porated to characterize the diffusion behavior in the ake-lled where D and D0 are the diffusion coefcient of nanocomposite, membrane (Bharadwaj 2001): and pure polymer matrix, respectively. Tortuosity ( ) is dened 1 as the ratio of the maximum path length (d ) that a perme(5) S = (3 cos2 1) 2 ate must travel to diffuse in the presence of nanoparticles and

N: Nanoscale Food Science


Figure 1Illustrations showing intercalated and delaminated state of the nanoparticles dispersed in the polymer matrix (adapted from Bharadwaj 2001; Small dots and long black bars represent the polymer matrix and nanoparticles, respectively).

N30 Journal of Food Science r Vol. 77, Nr. 7, 2012

Oxygen diffusion of Nanocomposite lms . . .


volume fraction which is calculated taking into account a unit cell in the polymer matrix. In contrast, nite element modeling (FEM) is more generalized, robust, and exible for both the idealized and random exfoliated structure, where the unit cell approach fails to predict the solutions. The current nite element analysis is concerned with the effect of aspect ratio, volume fraction, and aggregation on the effective diffusivity of the system. The study of aggregation is more relevant than uniformly distributed structure because llers tend to form aggregates due to their surface charges during the manufacture of D 1 = . (6) nanocomposites. In addition, this is the rst nite element analysis that investigates the inuence of intercalation, exfoliation, and the D0 1 + ( L /2W) (2/3)( S + 0.5) orientation angle of nanoparticles on oxygen barrier properties, Most of the analytical models are applicable only for the ide- making this approach more realistic. alized ordered systems with uniform spacing between akes and In this study, our objectives are (1) to develop a nite element based model to describe the diffusion process in nanocomposites, and (2) to analyze the inuence of the conguration and structural parameters (aspect ratio, volume fraction, intercalation level, orientation angle, and exfoliation) of nanoparticles on the oxygen barrier properties of nanocomposite systems. where is the angle between the direction of diffusion and the average orientation of the akes (Figure 3). The average is taken over all akes, with all possible orientations (Choudalakis and Gotsis 2009). The parameter S becomes 1 when all akes are normal to the diffusion ( = 0; Figure 2). When the akes become parallel to the diffusion ( = /2) the order parameter S is equal to 1/2. For random orientation of akes, S = 0. The relative diffusivity for the nonuniform randomly oriented akes is expressed as (Bharadwaj 2001)

Methodology
Model geometry In this study, a simple rectangular geometry of the lm was constructed to represent the packaging lm. The rectangular, plateletlike structure of nanoparticles was represented in the domain of the polymer matrix, as illustrated in Figure 2. The platelet like shape of nanoparticles was preferred because it is considered to have the highest oxygen barrier property. Previous studies have reported that 10 vol% impermeable spheres could provide a diffusivity reduction of only 14%, whereas the platelet-like structure could yield a diffusivity reduction of 91% (DeRocher and others 2005). Governing equation Diffusion of oxygen across the nanocomposite lm is basically a mass transfer process. It was assumed that Ficks law can predict the general behavior of diffusion for geometry with nanoscale range. Cui (2005) studied molecular diffusion of water conned in narFigure 2Schematic diagram of 2-dimensional model geometry of row cylindrical pores of 13 nm diameters and found that Ficks nanocomposite lm. law provides a good description to self-diffusion in uid conned in cylindrical pores. The Ficks diffusion model that governs the steady state diffusion process of a permeate in a lm can be written as
( D c ) = 0 (7)

where D is the diffusion coefcient of the gas in permeable phase in m2 s1 and c is the concentration of the permeate in mol m3 . The following equation was solved in COMSOL Multiphysics 4.2 and can be used to evaluate the steady state permeate ux across the lm after the solution: N = D c (8)

where N is the steady state permeate ux in mol m2 s1 . Boundary conditions for the diffusion phenomenon across the lm thickness are described in the following section.

Boundary Conditions
Figure 3Picture showing orientation angle ( ) of the nanoparticles.

Relevant boundary conditions were provided for the model geometry to simulate the diffusion process. The basic assumption of
Vol. 77, Nr. 7, 2012 r Journal of Food Science N31

N: Nanoscale Food Science

Oxygen diffusion of Nanocomposite lms . . .


The ner mesh size was adopted to track mass diffusion at the critical points such as slit area, where the diffusion process is affected by small gaps between adjacent akes, and additional resistance faced by permeate while turning the corner of akes to go from the wiggle into the slit (Eitzman and others 1996). The mesh system was rened continuously to get a mesh independent solution. The whole system was then solved in COMSOL Multiphysics 4.2, n ( D c ) = 0 (9) which gives the concentration prole of the oxygen through the polymer matrix and allows calculation of the diffusive ux of gas. In the current study, the volume fraction was dened as follows: where n is the normal vector to the boundary. The concentration at the outside of the wall (AB) of the lm is the oxygen concentration of the lm surroundings, taken as 10 mol/m3 (equivalent ( N) LW = (10) to 21% oxygen concentration in atmosphere). Oxygen concenld tration at the inner surface of the left wall (AB) of the lm is dependent on the solubility of oxygen in the polymer and it is where l is the length of the lm, ( N) and is the mean number of governed by Henrys law. The right surface (CD) represents the nanoparticles that a permeate encounters while diffusing through inner surface of the lm, which is in contact with the food mate- membrane. rial and concentration of oxygen is assumed to be zero. Thus, the Steady state nite element analysis was performed for intercaconcentration gradient across the thickness of the polymeric lm lated and exfoliated orientation of nanoparticles, with values is less than 10 mol/m3 . Solubility of oxygen varies with different varying from 50 to 1000, W in the range of 17 nm, and loading types of polymer, and that may result in different values of diffusive from 1% to 7%. The orientation angle varied from 0 to 85 for ux. However, the resulting relative diffusivity from the simulation exfoliation, and lower percentage loading, ranged from 0.25% to will remain same, irrespective of change in solubility values. 1%. Initial analysis was conducted for the computational domain to identify the reference value of diffusive ux for unloaded homoMesh generation and simulation geneous polymeric structure. Further simulation was carried out A very important step in nite element analysis is mesh gen- for the nanocomposites, and obtained solutions were compared eration. Mesh/grid quality plays an important role to achieve an with the reference value, expressed as relative diffusivity (D/D0 ). accurate and stable computational result. A ne unstructured mesh Typical diffusivity value of oxygen in neat PET polymer is 0.35 with triangular element was created for the computational domain. 1012 m2 s1 at 30 C and it was taken as reference diffusivity (D0 ) this process is that mass diffusion occurs only through the permeable phase of the nanocomposite lm from the left (AB) to the right wall (CD) of the lm geometry (Figure 2). No ux condition was applied to the top (AC), bottom surface (BD) of the lm, and all sides of nanoparticles, and it can be expressed in the following formulation:
Table 1 Relative diffusivity (D/D0 ) values at different volume fraction ( ), aspect ratio ( ), and intercalation width (W ). = 50 (%) 1 3 5 7 (%) 1 3 5 7

= 100 W = 7 nm 0.931 0.868 0.825 0.794 W = 7 nm 0.763 0.581 0.458 0.355 W = 1 nm 0.676 0.445 0.327 0.178 W = 1 nm 0.201 0.070 0.050 0.033 W = 3 nm 0.803 0.639 0.524 0.438 W = 5 nm 0.859 0.742 0.662 0.600 W = 7 nm 0.898 0.815 0.757 0.711 W = 7 nm 0.661 0.404 0.225 0.084 0.907 0.819 0.759 0.718

W = 1 nm 0.799 0.593 0.500 0.283 W = 1 nm 0.302 0.099 0.059 0.039

W = 3 nm 0.868 0.746 0.662 0.601

W = 5 nm

= 500

= 1000

W = 3 nm 0.581 0.180 0.148 0.041

W = 5 nm 0.671 0.419 0.299 0.156

W = 3 nm 0.328 0.159 0.085 0.041

W = 5 nm 0.527 0.213 0.152 0.076

N: Nanoscale Food Science

Reference diffusivity (D0 ) value of oxygen in PET polymer was taken as 0.35 1012 m2 s1 at 30 C for simulation purpose.

Table 2 Relative diffusivity (D/D0 ) values at different orientation angle ( ), volume fraction ( ), and aspect ratio ( ). = 50 (%) 0.25 0.5 1 (%) 0.25 0.5 1 (%) 0.25 0.5 1

=0

= 15 0.914 0.861 0.780

= 30 0.938 0.893 0.830

= 45 0.957 0.928 0.896 = 100 = 45 0.922 0.872 0.810 = 200 = 45 0.848 0.778 0.722

= 60 0.971 0.952 0.936 = 60 0.960 0.930 0.896 = 60 0.924 0.880 0.860

= 75 0.982 0.981 0.976 = 75 0.984 0.972 0.961 = 75 0.979 0.968 0.968

= 85 0.990 0.991 0.990 = 85 0.992 0.990 0.992 = 85 0.994 0.997 0.991

0.904 0.845 0.768 = 0 0.852 0.748 0.619 = 0 0.685 0.511 0.309

= 15 0.862 0.762 0.649 = 15 0.710 0.537 0.383

= 30 0.888 0.810 0.723 = 30 0.763 0.633 0.561

Reference diffusivity (D0 ) value of oxygen in PET polymer was taken as 0.35 1012 m2 s1 at 30 C for simulation purpose.

N32 Journal of Food Science r Vol. 77, Nr. 7, 2012

Oxygen diffusion of Nanocomposite lms . . .


for simulation purpose. Diffusive ux was calculated by integrating the local ux over the ux boundary, and dividing it by the length of the boundary (i.e. AB). The ratio of diffusive ux with and without akes was assumed to be equal to the relative diffusivity (Eitzman and others 1996). Different geometries were created to investigate the effect of ake aspect ratio ( ), volume fraction ( ), intercalation width (W ), exfoliation, and orientation angle ( ) on the diffusion behavior of nanocomposite systems. Simulated results were compared with Eqs. (3) and (6). The practical value of width of akes was 1 nm, which remained constant throughout the computational process.

Results and Discussion


Finite element results for the relative diffusivity values of nanocomposite lm as inuenced by different structural parameters, including varying range of aspect ratio; percent loading, intercalation width, and exfoliation are summarized in Table 1. The effect of orientation angle on relative diffusivity at different volume fraction and aspect ratio is presented in Table 2.

Figure 4Comparison of relative diffusivity (D/D0 ) estimated from analytical model (Bharadwaj 2001; Eq. 3) and current study at different loadings ( ): (A) 3%, and (B) 5% (D and D0 are the diffusivity of oxygen in polymer, with and without nanoparticles, respectively).

Figure 5Plot showing comparison of relative diffusivity (D/D0 ) for different orientation angle ( ) estimated from analytical model (Bharadwaj 2001; Eq. 6) and current study at aspect ratio ( ) of 100 for different volume fraction ( ; D and D0 are the diffusivity of oxygen in polymer, with and without nanoparticles, respectively).

Comparison with previous model prediction Results obtained from this nite element analysis were compared with previously developed model (Eq. (3)) to verify the accuracy of this model for predicting the enhancement of oxygen barrier properties. Relative diffusivity values obtained from this model and predicted by previous model were plotted against the intercalation width for the analyzed range of aspect ratios at different volume fractions (Figure 4). From the obtained plots, it can be observed that predictions of the previous model compared favorably with the present simulation result and showed reasonably good agreement. The comparison of relative diffusivity predicted by Bharadwaj model (Eq. (6)) and the FEM simulation as a function of orientation angle of akes for different loading percentages and aspect ratio is shown in Figure 5. Reasonable agreement was found between the predicted values. However, a deviation can be observed between the predicted values at small values of orientation angles. This could be due to the different computational approaches used. It is important to note that the S parameter in Eq. (5) becomes zero only when = 54.74 . However, the author did not consider a dened angular orientation when S value becomes zero. It is worth to mention that nite element method evaluates the diffusivity on the basis of the relative distribution of the nanoparticles, which was not considered in the analytical model, and therefore some disagreement is expected among the predicted results. Furthermore, the analytical model does not include the effect of relative distribution of akes and slits. The FEM could be a better predictor compared to the analytical model in cases where the unit cell approach in the nanocomposite is not taken into account. Different analytical models based on uniform volume fraction, slit shape, and aspect ratios have been developed by several researchers (Wakeham and Mason 1979; Aris 1986; Swannack and others 2005; Minelli and others 2009). The analysis of the unit cell is usually valid for an idealized ordered structure where the property of each cell inside the heterogeneous system is assumed to be similar for all possible conditions. However, in practical scenarios, it is difcult to maintain such an ordered structure of nanocomposites with denite slit shapes and uniform spacing between 2 adjacent layers of akes throughout the polymer matrix. In addition, the unit cell approach cannot be used to predict diffusion behavior for exfoliation where akes are randomly dispersed in the matrix. Compared to analytical models, nite element analysis formulates the solution according to the variation of the structural parameters of akes and can be used to portray the real phenomenon associated with gas transport. Thus, the developed FEM would be able to provide a comprehensive idea and better understanding compared to analytical model to estimate the gas barrier property of a nanocomposite.
Vol. 77, Nr. 7, 2012 r Journal of Food Science N33

N: Nanoscale Food Science

Oxygen diffusion of Nanocomposite lms . . . Effect of aspect ratio Inuence of aspect ratio on relative diffusivity (D/D0 ) at different volume fractions is shown in Figure 6. The akes were oriented at 0 and W = 1 nm (Figure 6A). The gradual displacement of curves to the lower relative diffusivity values with increasing loading percentage and aspect ratio was noticed. The decrease in relative diffusivity with increasing aspect ratio of the akes is attributed to the increasing tortuous path or wiggle which a diffusing molecule must follow to avoid the barrier. Decreasing relative diffusivity with increasing length-to-width ratio has been reported by many authors (Eitzman and others 1996; Swannack and others 2005; Minelli and others 2009). For example, Eitzman and others (1996) reported the linear increment of D0 /D with increasing square of the aspect ratio. They observed the enhancement of the barrier property, even at low ake concentrations. It is important to note that there was no considerable reduction in relative diffusivity beyond the aspect ratio, and volume fraction of 500 and 0.05, respectively for intercalation width 1 (Figure 6A) and 3 nm (Table 1). Hence, results indicate that = 500 and = 0.05 can be considered as an optimum structural parameter to achieve best barrier property for exfoliated system. However, this is not the case for higher intercalation width as shown in Figure 6B. The obtained results show the strong dependence of diffusivity on aspect ratio, volume fraction, and most importantly the degree of delamination of akes. Effect of volume fraction The effect of the volume fractions on relative diffusivity at different intercalation width is illustrated in Figure 7. For all aspect ratios, improvement of barrier properties were observed as the volume fraction of akes increased (Table 1). For example, D/D0 value decreased from 0.676 to 0.178 as increased from 0.01 to 0.07 at = 100 and W = 1 nm. It was also observed that lower intercalation width had a large impact on diffusivity (Figure 7). The resistance offered by higher amount of tortuosity and the large number of constriction between akes can be attributed to the enhancement of barrier properties at high volume fractions. Eitzman and others (1996) considered an ordered structure and found that wiggles had most signicant effect on the hindrance of diffusivity compared to slits. However, the diffusion behavior depends on the relative dispersion and interlayer spacing of the nanoparticles in the permeable matrix. A similar trend of curves can be found in other works (Aris 1986; Cussler and others 1988; Sorrentino and others 2006). Most of the previous analytical models predict the diffusivity as a function of
structural parameters and volume fraction; however, predictions of these models are limited by the unit cell approach, where the distribution of llers is well described and uniform. Unlike the analytical

Figure 7Effect of loading percent ( ) on relative diffusivity (D/D0 ) at aspect ratio ( ) = 1000 for the different intercalation level (W ): 1 nm ( r), 3 nm ( ), 5 nm ( ), and 7 nm ( ; D and D0 are the diffusivity of oxygen in polymer, with and without nanoparticles, respectively).

N: Nanoscale Food Science

Figure 8Plot showing relative diffusivity (D/D0 ) as a function of aggregate width (W ) at loading ( ) of 5% for the different aspect ratios ( ): 50 ( ), 100 ( ), 500 (), and 1000 ( r D and D0 are the diffusivity of oxygen in polymer, with and without nanoparticles, respectively).

Figure 6Inuence of aspect ratio ( ) on relative diffusivity (D/D0 ) at different aggregate width (W ): (A) 1 nm, and (B) 5 nm for the different volume fraction ( ) values: 1% ( ), 3% ( ), 5% (), and 7% ( r D and D0 are the diffusivity of oxygen in polymer, with and without nanoparticles, respectively).

N34 Journal of Food Science r Vol. 77, Nr. 7, 2012

Oxygen diffusion of Nanocomposite lms . . .


model, FEM can be considered a exible approach because it is not from 7 to 1 nm. This indicates that the exfoliated structure is more restricted to regular geometries and simpler boundary conditions. benecial than intercalation to improve the gas barrier properties of nanocomposite polymeric lms. The possible reason for that can be illustrated with the help of Figure 1. The stacks of nanoparEffect of intercalation width Intercalation or aggregation is a crucial factor that alters the ticles with 2 akes in each give an average aggregate width of 3 diffusivity considerably. Variation of relative diffusivity with inter- nm, assuming that the polymer layer between the akes is 1 nm calation width is shown in Figure 8. A gradual decrease in relative thick. Similarly, aggregates of 3 and 4 nanoparticles give the averdiffusivity was noticed with increasing delamination for all volume age intercalation width of 5 and 7 nm, respectively. It is important fractions of nanoparticles. For example, at = 100 and = 0.05, to note that increasing intercalation increases the effective width relative diffusivity decreased from 0.757 to 0.327 as W dropped of the stack, thus lowering the aspect ratio. Alternatively, single akes scattered in the polymer matrix, result in maximizing the aspect ratio, which dramatically increases tortuosity. The diffusion becomes the Knudsen type when the stacks of nanoparticles cause some holes to form in the matrix, creating low resistance to gas diffusion (Choudalakis and Gotsis 2009).

Effect of orientation angle Orientation angle is another parameter that greatly inuences oxygen transport in nanocomposites. In this study, all akes were oriented with a particular angle (range: 085 ). In practical scenarios, llers may have different orientation angle, and the angle may differ from ake to ake. However, the main purpose of the study was to explore the inuence of orientation angle on the diffusion behavior. The effect of orientation angle on the diffusivity for exfoliated structures at different volume fractions is shown in Figure 9. Figure 9 shows the inuence of orientation angle on relative diffusivity at aspect ratio of 200. As expected, increasing the orientation angle affects the oxygen barrier property by increasing the Figure 9Dependence of diffusivity on the orientation angle ( ): at aspect ratio ( ) of 200 for the different volume fraction ( ): 0.25% ( ), 0.5% ( ), diffusivity of gas in the nanocomposite. Diffusion path of oxygen in the polymer matrix is represented by streamlines (Figure 10). and 1% ( ).

Figure 10Concentration and streamline prole of oxygen in the exfoliated system at volume fraction ( ) = 1%, aspect ratio ( ) = 200 for different orientation angle ( ) of nanoparticles: (A) 0 , (B) 15 , (C) 30 , (D) 45 , (E) 60 , (F) 75 , and (G) 85 .

Vol. 77, Nr. 7, 2012 r Journal of Food Science N35

N: Nanoscale Food Science

Oxygen diffusion of Nanocomposite lms . . .


Streamlines follow a tortuous path around akes, showing maximum tortuosity at = 0 (Figure 10A). Inclined akes reduce tortuosity and allows permeate to pass through the system easily, thereby increasing the diffusivity of the gas. It can be noticed that streamlines become almost straight at = 85 (Figure 10G) indicating almost no hindered-diffusion at this stage. For the other 2 aspect ratios, curves with similar trends were observed. Shifting the angle of akes from 0 to 15 reduces the effect of akes (i.e. increase in diffusivity) by 5% for = 100 and = 0.01. Considering the same aspect ratio and volume fraction, almost 17% increase (relative increment almost 3.5 times) in relative diffusivity was observed when the angle changed from 0 to 30 which indicates the strong dependence of diffusivity on orientation angle. From the above example, for all values of aspect ratio, 15 could be considered as an inection angle beyond which a drastic reduction in barrier property can be expected. Few studies have described the effect of orientation angle of akes on the gas barrier property of nanocomposites (Eitzman and others 1996; Ly and Cheng 1997; Bharadwaj 2001). In addition, this study gives a better understanding of the diffusion behavior of nanocomposite with change in orientation angle for a wide range of aspect ratios. Relative diffusivity value becomes almost 1 as the orientation angle reaches 85 . At this stage, the entire nanocomposite system begins to behave like a pure polymer matrix by losing the effect of nanoparticles. Thus, it is always desirable to incorporate nanoparticles with lesser alignments in the polymer matrix.

Case study A theoretical case study was performed to highlight the implication of improvement in gas barrier property of the lm as result of nanoparticle incorporation on the total oxygen ingress inside the package. The estimated diffusivity after addition of nanoparticle can easily be computed by multiplying relative diffusivity values presented in Table 1 and 2 with diffusion coefcient of oxygen through conventional plastic lms without nanoparticles. Here, we analyze the reduction in oxygen ingress inside a retortable pouch during 1-y storage period. A polyethylene terapthalate (PET) based polymeric lm with 500 cm2 surface area and thickness value of 150 m was considered. These are the typical characteristics of a retort pouch. Diffusivity and solubility values of oxygen in PET were taken as 0.35 1012 m2 s1 and 0.049 m3 gas (m3 solid)1 (atm)1 (Light and Seymour 1983), respectively at 30 C. Total amount of oxygen ingress in a year through a polymer matrix can be calculated from the following equation:
Qt = DS( p ) Vm A 24 3600 365 (11) 22.414 ( x )

Table 3 Case study: Change in total amount of oxygen ingress in PET-based package per year at different volume fraction ( ), aspect ratio ( ), and intercalation width (W ). Predicted values: Q (cc O2 /y) = 50 (%) 1 3 5 7 (%) 1 3 5 7 W = 1 nm 30.2 22.4 18.9 10.7 W = 1 nm 11.4 3.7 2.2 1.4 W = 3 nm 32.8 28.2 25.0 22.7 W = 5 nm 34.3 31.0 28.7 27.1 W = 7 nm 35.2 32.8 31.2 30.0 W = 7 nm 28.8 22.0 17.3 13.4 W = 1 nm 25.5 16.8 12.3 6.7 W = 1 nm 7.6 2.6 1.8 1.2 30.4 24.1 19.8 16.5 = 100 W = 3 nm W = 5 nm 32.5 28.0 25.0 22.7 W = 7 nm 34.0 30.8 28.6 26.8 W = 7 nm 25.0 15.3 8.5 3.1

= 500

= 1000

W = 3 nm 21.9 6.8 5.6 1.5

W = 5 nm 25.4 15.8 11.3 5.8

W = 3 nm 12.4 6.0 3.2 1.5

W = 5 nm 19.9 8.0 5.7 2.8

Q0 (for control lm) = 37.8 cc O2 /y.

N: Nanoscale Food Science

Table 4 Case study: Change in total amount of oxygen ingress in PET-based package per year at different orientation angle ( ), volume fraction ( ), and aspect ratio ( ). Predicted values: Q (cc O2 /y) = 50 (%) 0.25 0.5 1 (%) 0.25 0.5 1 (%) 0.25 0.5 1 =0 34.1 31.9 29.0 = 0 32.2 28.3 23.4 = 0 26.0 19.3 11.6

= 15 34.5 32.5 29.5

= 30 35.4 33.7 31.3

= 45 36.2 35.0 33.8 = 100 = 45 34.8 33.0 30.6 = 200 = 45 32.1 29.4 27.3

= 60 36.7 36.0 35.3 = 60 36.3 35.1 33.9 = 60 34.9 33.2 32.5

= 75 37.1 37.1 36.9 = 75 37.2 36.7 36.4 = 75 37.0 36.6 36.6

= 85 37.4 37.4 37.4 = 85 37.4 37.4 37.5 = 85 37.6 37.7 37.4

= 15 32.5 28.8 24.5 = 15 26.8 20.3 14.4

= 30 33.6 30.6 27.3 = 30 28.8 24.0 21.2

Q0 (for Control lm) = 37.8 cc O2 /y.

N36 Journal of Food Science r Vol. 77, Nr. 7, 2012

Oxygen diffusion of Nanocomposite lms . . .


where Qt is the total amount of oxygen ingress in a year (ccyr1 ); D is the diffusivity in m2 s1 ; S is the solubility of oxygen in m3 gas (m3 solid)1 (atm)1 ; p is the pressure gradient of oxygen across the lm thickness in atm; Vm is the molar volume of oxygen in cc (kgmol)1 , A is the diffusing area of the lm in m2 ; and x is the thickness of the lm in m. The partial pressure of the oxygen at wall AB and CD were taken as 0.21 atm, and zero, respectively. The problem of oxygen ingress inside the package can be time dependent due to interactions between oxygen and packaged food, and requires more complex and rigorous analysis which is beyond the scope of this study. Hence for simplicity the oxygen concentration inside the package was taken as a constant. Improvement of oxygen barrier property in terms of total amount of oxygen ingress per year is presented in Table 3. In Table 3, Q0 and Q represent the mass of oxygen ingress per year through a control (without nanoparticles), and nanocomposite-based PET lm, respectively. Total amount of oxygen ingress at = 1000 and W = 1nm for = 7% was obtained as 1.2 cc/yr whereas the control value was 37.8 cc/yr (Table 3).This is the best possible scenario achieved from the simulation for reduction in the total amount of oxygen ingress (97% reduction). Predicted values of Q at different orientation angle are represented in Table 4. It is noticeable that amount of oxygen ingress inside the package reduces signicantly with the presence of nanoparticles in polymeric structure, thus resulting in a possible improvement in the shelf life of packaged food.
Azeredo HMCd. 2009. Nanocomposites for food packaging applications. Food Res Intl 42:124053. Bharadwaj RK. 2001. Modeling the barrier properties of polymer-layered silicate nanocomposites. Macromolecules 34:918992. Choudalakis G, Gotsis AD. 2009. Permeability of polymer/clay nanocomposites: a review. European Polym J 45:967984. Cui ST. 2005. Molecular self-diffusion in nanoscale cylindrical pores and classical Ficks law predictions. J Chem Phys 123(054706):14. Cussler EL, Hughes SE, Ward WJ, Aris R. 1988. Barrier membranes. J Membr Sci 38: 16174. DeRocher JP, Gettelnger BT, Wang J, Nuxoll EE, Cussler EL. 2005. Barrier membranes with different sizes of aligned akes. J Membr Sci 254:2130. Eitzman DM, Melkote RR, Cussler EL. 1996. Barrier membranes with tipped impermeable akes. Fluid Mech Transp Phenom AIChE J 42(1):29. Falla WR, Mulski M, Cussler EL. 1996. Estimating diffusion through ake-lled membranes. J Membr Sci 119:12938. Frounchi M, Dourbash A. 2009. Oxygen barrier properties of poly (ethylene terephthalate) nanocomposite lms. Macromol Mater Eng 294:6874. Kumar P, Sandeep KP, Alavi S, Truong VD. 2011. A review of experimental and modeling techniques to determine properties of biopolymer-based nanocomposites. J Food Sci 76: E2E14. Lange J, Wyser Y. 2003. Recent innovations in barrier technologies for plastic packaginga review. Packag Technol Sci 16(4):14958. Light R, Seymour R. 1983. Soc Plast Eng 29:417. Lotti C, Isaac CS, Branciforti MC, Alves RMV, Liberman S, Bretas RES. 2008. Rheological, mechanical and transport properties of blown lms of high density polyethylene nancomposites. European Polym J 44:134657. Lu C, Mai YW. 2007. Permeability modeling of polymer-layered silicate nanocomposites. Compos Sci Tech 67:2895902. Ly YP, Cheng YL. 1997. Diffusion in heterogeneous media containing impermeable domains arranged in parallel arrays of variable orientation. J Membr Sci 133(2):20715. Maxwell JC. 1873. A treatise on electricity and magnetism, Vol. 1. Oxford: Clarendon Press. 365 p. Minelli M, Baschetti MG, Doghieri F. 2009. Analysis of modeling results for barrier properties in ordered nanocomposite systems. J Membr Sci 327:20815. Nielsen LE. 1967. Models for the permeability of lled polymer systems. J Macromol Sci Part A 1(5):92942. Paul DR, Robeson LM. 2008. Polymer nanotechnology: nanocomposites. Polymer 49:3187204. Pereira de Abreu DA, Paseiro Losada P, Angulo I, Cruz JM. 2007. Development of new polyolen lms with nanoclays for application in food packaging. European Polym J 43: 222943. Raleigh L. 1892. On the inuence of obstacles arranged in rectangular order upon the properties of the medium. Philos Mag 34:481502. Ray SS, Okamato M. 2003. Polymer/layered silicate nanocomposites: a review from preparation to processing. Prog Polym Sci 28(11):1539641. Sorrentino A, Tortora M, Vittoria V. 2006. Diffusion behavior in polymer-clay nanocomposites. J Polym Sci Part B Polym Phys 44:26574. Swannack C, Cox C, Liakos A, Hirt D. 2005. A three-dimensional simulation of barrier properties of nanocomposite lms. J Membr Sci 263:4756. Statler DL, Gupta RK. 2007. A nite element analysis of the inuence of morphology on barrier properties of polymer-clay nanocomposites. Proceedings of the COMSOL Conference; 2007 October 5; Boston. Wakeham WA, Mason EA. 1979. Diffusion through multiperforate laminae. Ind Chem Eng Fundam 18(4):3015. Weiss J, Takhistov P, McClements DJ. 2006. Functional materials in food nanotechnology. J Food Sci 71(9):R107R116.

Conclusions
This study developed a FEM for diffusion behavior in nanocomposite system to increase the quality and shelf-life of packaged food. The diffusion behavior was investigated for different volume fractions, aspect ratio, intercalation width, exfoliation, and orientation angle. Results show a strong dependence of relative diffusivity on aspect ratio, volume fraction and intercalation width. Oxygen barrier properties showed great improvement with increasing aspect ratios, and volume fractions of nanoparticles. However, no considerable improvement in the barrier property was found beyond = 0.05 and = 500 for exfoliation. Results also show that exfoliation can improve gas barrier properties by several times compared to intercalation. A great reduction in gas barrier properties was noticed as orientation angle increased. The simulated results showed a drastic increase in diffusivity beyond = 15 . Case study showed nearly 97% reduction in oxygen ingress at W = 1nm and = 1000 for = 7% for nanocomposite-based PET lm. FEM developed in this study would be more applicable than analytical models for predicting solutions for systems where the distribution of nanoparticles in the polymer matrix are relatively complex compared to ordered structures. Furthermore, this model is more generalized, and exible enough to incorporate the 3D structure of nanollers in future research. The developed model can be reliably used in food packaging applications to develop packaging lms with improved gas barrier properties.

In this section, a brief review of some important previous analytical models is summarized. The rst fundamental theory of diffusion in composite object was developed by Maxwell (1873), who investigated the transport in periodic layers of neutrally buoyant spheres suspended in a membrane. The expression for relative diffusivity (D/D0 ) was given as: 1 D = D0 1 + /2

(A1)

where D and D0 are the diffusion coefcient of nanocomposite, and pure polymer matrix, respectively, and is the volume fraction This work was funded in part by the USDA NIFA Research of spheres. The relative diffusivity for a membrane consisting of Grant # 201168003-20096. cylindrical nanoparticles oriented parallel to the membrane surface was developed by Raleigh (1892):

Acknowledgments

References

Alexandre B, Langevin D, M ed eric P, Aubry T, Couderc H, Nguyen QT. 2009. Water barrier properties of polyamide 12/montmorillonite nanocomposite membranes: structure and volume fraction effects. J Membr Sci 328(12):186204. Aris R. 1986. On a problem in hindered diffusion. Arch Ration Mech Anal 95(2):8391.

D 1 = D0 1+

(A2)

Vol. 77, Nr. 7, 2012 r Journal of Food Science N37

N: Nanoscale Food Science

Appendix

Oxygen diffusion of Nanocomposite lms . . .


where is the volume fraction of cylindrical nanoparticles. It must be noted that, Eqs. (1) and (2) are limited to very low loadings of ller particles, and that the resulting relative diffusivity depends on the volume fraction of particles, rather than their size. Consequently, several theoretical models (Nielsen 1967; Wakeham and Mason 1979; Aris 1986; Cussler and others 1988) have been proposed to estimate diffusion in nanocomposite. Nielsen model (1967) predicts the permeability of a nanocomposite system with very low clay loading (less than 1%), but fails to predict the values and deviates from experimental data for higher loading levels. Wakeham and Mason (1979) proposed a new model based on perforated laminae. Aris (1986) developed a similar model introducing the effect of resistance due to slit shape ( ) and necking, as follows: 2 D 2 2 4 =1+ + + ln D0 1 (1 ) (1 ) The rst term in Eq. (3) is just unity. The relative diffusivity becomes 1 when the loading of akes equals zero. The second term is attributed to the tortuous path around the akes or wiggle. The third term, involving , represents resistance to diffusion due to constrictions between adjacent akes. The last term corresponds to the resistance offered by necking, which a diffusive molecule faces while turning the corner of akes to go into or out of the slit (Falla and others 1996). The last term on the right hand side of Eq. (3) is dependent on the aspect ratio ( ), comprising the main difference between the Aris model and Wakeham and Mason model. Swannack and others (2005) performed 2D and 3D Monte-Carlo simulation of polymer-clay nanocomposite system with regularly spaced platelets to compute the diffusivity of permeate and found a signicant difference between results obtained from 2-D and 3-D simulations. Statler and Gupta (2007) presented a nite element model for an ordered ake-lled system to analyze diffusion in the nanocomposites. However, they did not describe (A3) the effect of orientation angle on the diffusion behavior of the nanocomposite system.

N: Nanoscale Food Science


N38 Journal of Food Science r Vol. 77, Nr. 7, 2012

Potrebbero piacerti anche