Evaluation and Simulation of Silver and Copper Nanoparticle

Article pubs.acs.org/JAFC

Evaluation and Simulation of Silver and Copper Nanoparticle Migration from Polyethylene Nanocomposites to Food and an Associated Exposure Assessment M. Cushen,† J. Kerry,‡ M. Morris,§ M. Cruz-Romero,‡ and E. Cummins*,† †

School of Biosystems Engineering, Agriculture and Food Science Centre, University College Dublin, Belfield, Dublin 4, Ireland School of Food and Nutritional Sciences, Food Packaging Group, and §Department of Chemistry, University College Cork, Cork, Ireland

‡

ABSTRACT: Silver nanoparticles (nanosilver) and copper nanoparticles (nanocopper) exhibit antimicrobial activity and have been incorporated into polymers to create antimicrobial packaging materials. Their use in conjunction with food has caused concerns regarding the potential risk of particle migration, resulting in human exposure to nanoparticles. A migration experiment was carried out to investigate the effect of time and temperature on the migration of nanosilver and nanocopper particles from polyethylene (PE) nanocomposites to boneless chicken breasts. Migration of silver ranged from 0.003 to 0.005 mg/dm2, while migration of copper ranged from 0.024 to 0.049 mg/dm2, for a set of four different scenarios representing typical storage conditions. Effects of time and temperature were not significant (p > 0.1). A migration and exposure model was developed on the basis of mathematical relationships defining migratability and subsequent migratables using the Williams−Landel−Ferry equation for time−temperature superposition. The results of the model accurately predicted the nanosilver levels detected in the laboratory migration tests (R values ranging from 0.43 to 0.99); however, the model was less accurate in predicting nanocopper levels (R values ranging from 0.65 to 0.99), probably because of the highly variable background levels of copper observed in the real food matrix. The 95th percentile of the simulated human exposure to nanosilver based on laboratory experimental results of four scenarios ranged from 5.89 × 10−5 to 8.9 × 10−5 mg kgbw−1 day−1. For the measured migration of copper under the same storage conditions, the exposure ranged from 2.26 × 10−5 to 1.17 × 10−4 mg kgbw−1 day−1. This study highlights the potential migration of nanoparticles from PE composite packaging to a food material and the potential for simulation models to accurately capture this migration potential; however, variable background levels of copper in the food matrix can make prediction more difficult for trace migration of nanocopper. KEYWORDS: migration, nanocomposite, food packaging, simulation modeling, exposure

■

behavior,4 hence, its suitability as a filler in an antimicrobial nanocomposite. Some nanomaterials are not permitted in the European Union (EU) because of limited available toxicity research results. In addition, their behavior following disposal must be considered.5 If migration were to occur from nano-containing food contact materials into food, consumers may be exposed to novel substances, for which the toxicity is not yet fully understood. Because of this, conservative usage limits have been placed on such materials. However, nanomaterials that have been shown not to migrate or to migrate in insignificant amounts are permitted.6,7 Toxicity can be based on a number of parameters, in addition to the traditional factor of chemical composition, including particle size, shape, and surface behavior.8 Permission for the use of a particular food contact material may depend upon its performance in migration tests, and hence, migration testing is very important during the introduction process of a new packaging material. The conditions of migration tests should be as close to their

INTRODUCTION

Many new technological applications have emerged on the basis of novel behavior exhibited by some materials at the nanoscale (particle size between 1 and 100 nm). The collective term used to describe these applications is nanotechnologies, and the food industry, among other industries, is likely to benefit from them. An area within the food industry that nanotechnologies have shown promise is advancements in food packaging materials.1 Fillers are entities that are added to conventional food packaging materials, usually in low percentages, enhancing the performance of the original material. The combination of the conventional packaging material and a filler results in a composite. Antimicrobial compounds that do not enlist nanotechnologies and are designed to be incorporated into food contact polymers2 are commercially available in some countries. Nanosilver shows unique antimicrobial behavior and has been incorporated into commercially available food packaging matrices as a filler. Such nanocomposites, where the filler has a particle size at the nanoscale, are designed to exploit the novel properties of their respective nanocomponent. Copper also has novel attributes at the nano level that have been exploited in various new applications.3 Notably, nanocopper exhibits antimicrobial © 2014 American Chemical Society

Received: Revised: Accepted: Published: 1403

September 11, 2013 January 21, 2014 January 22, 2014 January 22, 2014 dx.doi.org/10.1021/jf404038y | J. Agric. Food Chem. 2014, 62, 1403−1411

Journal of Agricultural and Food Chemistry

Article

intended use as possible, i.e., in contact with the food that they are intended to be used with, while maintaining storage times and temperatures consistent with consumer practices. Alternatively, exaggerated conditions can be used to assess a worstcase scenario, so that measures can be taken to protect even the most careless consumer. Where possible, the food that the intended food packaging material is to be used with should be used in the migration tests because the nexus of interrelationships, such as water availability and viscosity, is not accounted for when using food simulants. Simulation models can be a valuable resource in risk assessment, particularly in migration9 and exposure assessments.10 Simulation models can be used as an initial screening tool to provide a range of values likely to result from a given scenario, reducing the need for expensive, time-consuming migration tests. Simulation models can also provide guidance on expected migration ranges, hence guiding the selection of sensitive analytical techniques. Model design inherently involves the incorporation of multiple factors that affect the overall system and the collation of variability and uncertainty of all known factors that contribute to the models output. For example, a migration model requires inputs; these inputs should account for all of the attributes that affect migration in a packaging−food system. Often existing interrelations between input factors can be found in the literature,11 and these can be useful as a starting framework. By reviewing the literature on existing knowledge regarding mass-transfer phenomena, these systems can be mathematically simulated to calculate potential nanoparticle migration for a particular scenario. When the necessary data are available, exposure assessments are a logical progression from migration testing, particularly when dealing with migration into real foodstuffs. Exposure assessments can improve our understanding of the risks associated with migration. They can make migration results more tangible as the migration figures are combined with likely consumption levels of the particular food in question while also considering the relative body weight of an individual. Hence, the objective of this study was to analytically assess the migration of silver and copper nanoparticles from polyethylene (PE) matrices into a food matrix and to use known interrelationships reported in the literature to create a simulation model to predict likely migration levels. The migration results were also used for an exposure assessment, focusing on the oral route of exposure, following the ingestion of chicken meat packaged with a nanopolymer.

■

Figure 1. SEM images of silver nanoparticles that were incorporated into PE. they were prepared in deionized water. Sodium citrate solution was added dropwise to CuCl solution under stirring at a 2:1 volume ratio, at 60 °C. The mixture was stirred for 30 min, before SFS solution was added dropwise. The mixture was stirred for a further 60 min. A brown precipitate was formed. The supernatant was discarded, and the precipitate was washed. The supernatant was again discarded, and the brown precipitate was dried at 60 °C. This procedure was adapted from Khanna et al.13 Composite Manufacture. PE-based composites were manufactured via extrusion in a micro 27 lab-scale twin screw extruder (Leistritz, Nuernberg, Germany) with a 27 mm screw diameter and a 38:1 length/diameter ratio, using the screw geometry as outlined by Gunning et al.14 PE pellets (melt flow rate of 0.33 g/10 min and density of 921 kg/m3) were supplied by Boxmore Plastics Co., Cavan, Ireland. In the compounding process, the temperature profile was increased from 160 °C at the hopper to 200 °C at the die with a screw speed of 120 rpm. Two PE nanocomposites were designed containing 0.5% (w/w) of the nanocomponent (properties of the PE used can be found in Table 1). The silver nanocomposite incorporated nanosilver with a logistic size distribution with a mean of 1.01 × 10−8 m and a standard deviation of 6.4 × 10−10 m (Figure 1). The copper nanocomposite incorporated nanocopper with a logistic size distribution with a mean of 1.14 × 10−8 and a standard deviation of 6.8 × 10−10 m. Skinless, boneless chicken breasts (samples) were sourced from an Irish supplier, and 32 samples were wrapped in 120 cm2 of one of the prepared nanocomposites (n = 16 for both silver and copper). The packaging did not overlap. Aluminum foil was wrapped around these to eliminate any possible variation that light may impart.15 Oxidation of silver nanoparticles because of ultraviolet (UV) radiation is known to facilitate silver dissolution.16 Each sample (nanocomposite and foiled unit) was then vacuumpacked in standard, heat-sealable, clear, PE vacuum packing bags to ensure maximum, consistent contact between the active nanocomposite and the samples. For such active packaging materials, sharing a common interface or physical contact with the food surface is essential for the desired effect to be observed.17 Each unit was performed in quadruplicate. Also, eight control units were prepared, where samples were in contact with a PE packaging that did not contain added nanosilver or nanocopper. Units were kept in constant temperature rooms for the duration of the experiment (t): either 1.1 or 3.1 days. The time periods chosen for the migration tests were thought to represent likely and unfavorable (however, possible) storage periods for chicken.

MATERIALS AND METHODS

Experimental Procedure. Nanoparticle Synthesis. To make the silver nanoparticles, an AgNO3 solution (0.1 M) was prepared in ethanol at 45 °C to make 10 nm nanoparticles. An equal amount of 0.1 M polyvinylpyrrolidone (PVP) solution in ethanol was then added to this at a rate of 0.667 mL min−1. The solution was stirred for 2 h (800 rpm), until the solution had turned a stable orange−brown color. Silver nanoparticles were separated from solution by the addition of acetone at a volume ratio of approximately 1:4. The solution was sonicated for 10 min and centrifuged at 6000 rpm for 15 min. The supernatant was discarded; the pellet was redispersed in ethanol; and the separation step was repeated. The pellet was dried in an oven overnight at 60 °C and crushed to a powder; this procedure was adapted from Chen et al.12 Scanning electron microscopy (SEM) images were taken of the particles (Figure 1) To make the copper nanoparticles, CuCl (99% minimum), sodium formaldehyde sulfoxylate (SFS), and sodium citrate were purchased from Sigma-Aldrich. The concentration of all solutions was 0.5 M, and 1404

dx.doi.org/10.1021/jf404038y | J. Agric. Food Chem. 2014, 62, 1403−1411


Article

Table 1. Inputs, Constants, and Outputs of the Simulation Model name PE

nanoparticle

unit

symbol

equation

reference

glass transition temperature

Pa s K K

η(p) x Tg

density

921

kg/m3

ρ(p)s

kg/m3 kg/m3 m m

ρ(n)c ρ(n) 2r 2r

30 mv mv

s s K

t t T

mv mv mv

nanosilver

10490 8940 logistic (1.14 × 10−8, 6.8 × 10−10) uniform with a mean of 1.009 × 10−8 and σc of 1.145 × 10−9 95040 267840 lognormal [0.306, 0.385, shift (7.8)] + 273 logistic (21.86, 0.31) + 273 uniform (0.0099, 0.0101) d

K m2 kg/m3

T a c

mv mv mv

nanocopper

d

kg/m3

c

mv

J K−1 K K

π kB C1 C2

11 11 11

Pa s

η(Tg)

1

dη/dT

η(T)

2

m

M

3

m

M

3

mg

Nmg

4

mg/dm2

Q

5

dynamic viscosity

density

time (there were two levels of this factor) temperature (there were two levels of this factor)

viscositya temperature

nanosilver nanocopper nanosilver nanocopper 1.1 days 3.1 days T1 T2

system

value Inputs uniform (200,b 2300) uniform (473, 487.87b) logistic (−120.565, 1.5667)

diameter

storage

subdivision

surface area concentration of nanoparticles in PE

29 29 measured values (mv) product specification

Constants pi Boltzmann constant empirical parameter empirical parameter viscosity at the glass transition temperature temperature dependence of viscosity migratability

nanosilver nanocopper

migratables

22/7 1.3087 × 10−23 17.44 51.6 Outputs distribution with a mean of 4.3 × 109 and σc of 2.13 × 109 distribution with a mean of 1.7 × 104 and σc of 8.4 × 104 specific to the levels of the factors: t and T specific to the levels of the factors: t and T specific to the value of M and c and, thus, the factor ρ(n)

migration level

Negatively correlated with the dynamic viscosity temperature, with a value of −0.8. Calculated from extrapolation of values.29 cStandard deviation of the distribution. dSee Table 3. a

b

Figure 2. Experimental design, where composites containing either silver or copper nanoparticles undergo migration tests (into chicken breast samples) and are held for 1.1 or 3.1 days at an average temperature of 8.13 or 21.8 °C according to their respective scenario profile. Temperature probes were used to log the internal temperatures of the samples at the various temperature levels (Figure 2). The effect of

the temperature was determined by subjecting samples to different temperatures during the migration test time periods. The temperatures 1405



Article

Table 2. Scenarios Showing Experimental Conditions with Corresponding Laboratory, Migration Modeling, and Exposure Results

migrant

scenario

time

temperature

mean migrant (mg/kg) (n = 4)

silver silver silver silver copper copper copper copper

1 2 3 4 1 2 3 4

T1 T1 T2 T2 T1 T1 T2 T2

t1 t2 t1 t2 t1 t2 t1 t2

0.026 0.031 0.042 0.039 0.265 0.351 0.207 0.382

mean migrant (mg/dm2) (n = 4) 0.003 0.004 0.005 0.004 0.032 0.044 0.024 0.049

2

mean predicted level (Q) (mg/dm ) (n = 4) (5th−95th percentile) 0.003 0.003 0.004 0.005 0.038 0.040 0.044 0.047

(0.002−0.004) (0.002−0.005) (0.003−0.007) (0.003−0.009) (0.013−0.038) (0.014−0.040) (0.017−0.044) (0.019−0.047)

95th percentile of exposure (e) (mg kgbw−1 day−1) 5.89 7.06 8.9 7.97 2.26 5.67 4.24 1.17

× × × × × × × ×

10−5 10−5 10−5 10−5 10−5 10−5 10−5 10−4

Figure 3. Study framework showing experimental procedures, modeling tasks, and their interrelationships culminating in the comparison of the two results to assess the performance of the migration model. chosen to represent the conditions of a poorly performing/ overcrowded refrigerator and ambient storage. The two levels of temperatures were set and then logged during the migration tests. The means ± standard deviations of the distributions T1 and T2 were 8.13 ± 0.48 and 21.80 ± 0.64 °C, respectively. Following the migration tests, all packaging was removed from the samples. Each chicken breast sample was homogenized, and 2 g of it was added to a Teflon microwave vessel. A total of 10 mL of concentrated nitric acid (Fisher Scientific) was added to the vessel. The sample was digested using microwave-assisted acid dissolution at 200 °C (microwave unit from Mars Xpress supplied by CEM). The extract produced was diluted to 50 mL with deionized water. Digests were analyzed using inductively coupled plasma−mass spectrometry (ICP−MS, Agilent 7500 ICP−MS) according to the protocol assigned the ISO number DIN EN ISO 17294-2-E29. Standards were derived from stock solutions from Spectrosol. Mathematical Model. A migration model was designed, and its validity was assessed by comparing migration predictions to the results of the laboratory migration tests, given the same set of conditions (time and temperature). Migratability and subsequent migratables were intermediate outputs that were also generated by the mathematical model. The equations to follow are applicable to polyolefins and not to non-amorphous polymers. The Williams−

Landel−Ferry equation for time−temperature superposition was used with the dynamic viscosity of the polymer to obtain the viscosity at the glass transition temperature of the polymer, as used by Šimon et al.11 and defined as η(Tg) =

η(T ) ⎛ ⎛⎜ −1(C1(T− Tg) ⎞⎟⎞ ⎜exp⎝ C2+ T− Tg ⎠⎟ ⎜ ⎟ ⎝ ⎠

(1)

where η is the viscosity, Tg is the glass transition temperature, T is the experimental temperature (as mentioned above; in this case, the dynamic viscosity information of the polymer was used), and C1 and C2 are constants with values of 17.44 and 51.6 K, respectively. As seen from eq 2, the Williams−Landel−Ferry equation is also used for the subsequent calculation of the viscosity at a particular temperature. During validation, experimental temperatures were used as inputs. ⎛ −1(C1(T − Tg)) ⎞ ⎜⎜ ⎟ C 2 + T − Tg ⎟⎠

η(T ) = η(Tg)exp⎝

(2)

Migratability (M) is a value of the likelihood of particles migrating from a given system and is defined as 1406



Article

Table 3. Inputs and Outputs of the Model of Nanosilver and Nanocopper Concentrations in the Nanocomposites value quantity of PE in the nanocomposite quantity of silver in the nanocomposite PE density silver density copper density volume of PE volume of silver volume of 1 kg of nanocomposite weight of 1 m3 of nanocomposite weight of nanosilver in 1 m3 of nanocomposite volume of PE volume of copper volume of 1 kg of nanocomposite weight of 1 m3 of nanocomposite weight of nanocopper in 1 m3 of nanocomposite

M=

⎛ kBTt ⎞ ⎜ ⎟ ⎝ 24(π 2)ηr ⎠

Inputs triangular (99.48, 99.5, 99.52) triangular (0.48, 0.5, 0.52) 921 10490 8940 Outputs for Nanosilver Nanocomposite distribution with a mean of 1.08 × 10−3 and σ of distribution with a mean of 4.76 × 10−7 and σ of distribution with a mean of 1.08 × 10−3 and σ of distribution with a mean of 925 and σ of 0.07 distribution with a mean of 1.53 × 10−5 and σ of Outputs for Nanocopper Nanocomposite distribution with a mean of 1.08 × 10−3 and σ of distribution with a mean of 4.76 × 10−7 and σ of distribution with a mean of 1.08 × 10−3 and σ of distribution with a mean of 925 and σ of 0.07 distribution with a mean of 1.53 × 10−5 and σ of

(3)

(4)

where a is the area of the nanocomposite−food interface and c is the initial concentration of nanoparticles in the nanocomposite assuming homogeneous distribution. The final migration level (Q) must consider background levels of the metal (eq 5). In this study, the background level of each metal was estimated using measurements taken from the control samples. This was particularly important in the case of the samples in contact with the copper nanocomposites because chicken breasts have naturally occurring background levels of copper (BL), which are highly variable relative to likely copper migration levels. A dimensionless uncertainty factor (e) is also included in the migration model to take into account the model uncertainty (eq 5). It was calculated by optimizing the representations of the experimental data of the models. Q = (Nmg + BL)e

1.45 × 10−6 8.86 × 10−8 7.78 × 10−9 8.09 × 10−8 1.45 × 10−6

unit

equation

b d ρ(p) ρ(n)s ρ(n)c

%, w/w %, w/w kg/m3 kg/m3 kg/m3

f g h j c

m3 m3 m3 kg/m3 kg/m3

b/ρ(p) d/ρ(n) f+g 1/h j×d

f g h j c

m3 m3 m3 kg/m3 mg/m3

b/ρ(p) d/ρ(n) f+g 1/h j × d × 106

so, it was assumed that all silver was in its most harmful form, i.e., unagglomerated nanoparticles, because it is not known how readily silver nanoparticles produce silver ions or in what ratios they coexist. In addition, both processes are likely to vary according to pH, temperature, and/or time.10 Any agglomeration of the nanoparticles is likely to reduce the active particle surface area and, hence, reduce toxicity, and therefore, the model assumed no agglomeration throughout the process. A mathematical model were created to generate probability distributions of human exposure to silver and copper nanoparticles as a result of the consumption of chicken after contact with the nanosilver PE nanocomposite and the nanocopper PE nanocomposite used in the migration tests. The results of the migration experiments fed into the exposure assessment as an input variable.10 Monte Carlo simulation (10 000 iterations) was used so that the inherent uncertainty and variability of input parameters were considered. Risk analysis software (@Risk 5.5, Palisade, Middlesex, U.K.) was used to generate a predicted exposure estimate from the series of input distributions using eq 6

where kB is the Boltzmann constant (kB = 1.3807 × 10−23 J K−1), t is the time in seconds, and r is the radius of the particles. Migratables (Nmg) is the quantity of nanoparticles migrating from the polymer matrix11 and is calculated as given in eq 4

Nmg = Mac

8.86 × 10−8 7.78 × 10−9 8.09 × 10−8

symbol

Ex = (yx z)/bw

(6)

where E is exposure in mg kgbw−1 day−1 and x is the particular migrant particle (in this case, nanosilver or nanocopper). The term y is migration in mg/kg into the particular food, chicken breast meat in this case. The term z is the consumption of that food in kg/day in a given population. The term bw refers to the body weight of an individual in kilograms in the given population. Chicken consumption data for Ireland,18 for those who consumed chicken (67%; n = 1500), took account of “... all chicken eaten as meat only, chicken on the bone, chicken skin and chicken meat extracted from composite meals by way of an arbitrary percentage of 66.7%.” The data provided was fit to a log-normal distribution (with a mean of 62.01 g/day and a standard deviation of 43.84 g/day) using @Risk 5.5 software to represent consumption (z). Irish adult body weight data18 (bw) was fit to a normal distribution (with a mean of 77.48 kg and a standard deviation of 16.18 kg) using @Risk 5.5 software.

(5)

A Monte Carlo simulation was used so that inherent variability of the input parameters can be considered. The inputs, constants, and outputs of the model are given in Table 1. The @RISK 5.5 software (Palisade, Middlesex, U.K.), a package specifically used for risk assessment, was used as an add in in Microsoft Excel (Microsoft Office 2007). The model was run for 10 000 iterations, giving a final simulated output probability density distribution for the level of nanosilver and nanocopper likely to migrate into the food matrix. Model−Experimental Synchronization. The conditions of the experiments were used as inputs in the model (Table 2). Scenario 1 used a time (t) of 1.1 days and a temperature (T) of 8.13 ± 0.48 °C. Scenario 2 used a time of 1.1 days and a temperature of 21.80 ± 0.64 °C. Scenario 3 was for a time of 3.1 days and a temperature of 8.13 ± 0.48 °C. Scenario 4 was for a time of 3.1 days and a temperature of 21.80 ± 0.64 °C. This allowed for the direct comparison of the model outputs to the experimental results. This framework is explained diagrammatically in Figure 3. The input variable of the concentration was converted from %, w/w (given in the Experimental Procedure above) to kg/m3 to allow for comparison to model outputs (Table 3). Exposure Model. If migration occurs, likely exposure arising from the migration can be calculated by an exposure assessment. In doing

■

RESULTS AND DISCUSSION Experimental Migration. The results of the experiments are presented in Table 2. All four scenarios were conducted at temperatures above the recommended storage temperature of a perishable food product (>5 °C); this is an important consideration when comparing the migration of nanosilver in this study to other studies or ingestion limits.

1407



Article

Figure 4. Probability density distribution of the final migration level (Q) of silver in chicken and the minimum, maximum, and interquartile range of the corresponding experimental values, following simulations and migration tests of the following conditions: (a) scenario 1, (b) scenario 2, (c) scenario 3, and (d) scenario 4.

It should be noted that the levels of migration observed in this study are from a pilot bench-scale nanocomposite developed in a laboratory for experimental uses only. Commercially, it would be unlikely that nanocomposites containing 0.5% (w/w) fill of nanosilver or nanocopper would be produced because novel effects are observed at much lower levels.19 This is particularly important because the level of nanoparticulate fill was identified as the most significant parameter in a recent study10 influencing migration compared to other parameters of time, temperature, and nanoparticle size. Hence, it is feasible to engineer a functional antimicrobial nanocomposite, with a lower nanofiller percentage and which will potentially result in less migration. Model Validation. It is shown in Figure 4 that the model proved to be reliable in predicting the migration results for nanosilver over a range of storage conditions with coefficients of correlation (R) of 0.99, 0.99, 0.943, and 0.62 and root-meansquare error (RMSE) values of 2.5 × 10−4, 1.1 × 10−4, 1.5 × 10−3, and 2.9 × 10−3 for scenarios 1, 2, 3, and 4, respectively. In general, the model was slightly less accurate in predicting the levels of copper in the chicken samples, with R values of 0.65, 0.97, 0.99, and 0.66 and RMSE values of 9.6 × 10−3, 3.4 × 10−2, 1.4 × 10−2, and 2.1 × 10−2 for scenarios 1, 2, 3, and 4, respectively, as shown in Figure 5. This suggests that the model may be a useful tool in the design and engineering of new nanopackaging materials that are likely to come on stream in the coming years as the fruits of the recent surge of nanotechnological research mature. The main appeal of this model to industry is that it reduces the need for expensive and time-consuming migration tests, thus also reducing costs.

SAS 9.1 was used to establish if there were any effects of time and temperature on the migration of silver or copper from the nanocomposites. No significant effects of time or temperature were observed in the samples analyzed for copper or silver (p > 0.1). In the case of both time and temperature, it is suggested that migration levels may have already plateaued over the time period studied, and hence, no significant effect in the migration results was observed. This is supported by the fact that recent research, which focused on a shorter time period, reported a reduction in the migration rate within the first day of contact.19 It is likely that both of these time periods used in the migration tests of this study are beyond the point of the migration−time relationship, where an increase in time brings about an increase in silver levels. Silver ion release from a polyester matrix into liquid simulant reached a maximum in less than 5 min in another study,20 but corresponding SEM imagery seemed to confirm that this was due to the dissolution of particles on the surface or very near the surface of the polyester matrix in which they were bound. The quantities of silver in samples that had been in contact with the nanosilver PE composite (with a mean of 0.0041 mg/ dm2; n = 16) were significantly different from the control samples (with a mean of 0.001 mg/dm2; n = 8) (p < 0.001). The samples in contact with the nanocopper PE composite had significantly higher copper levels (with a mean of 0.037 mg/ dm2; n = 16) than the control samples (with a mean of 0.0157 mg/dm2; n = 8) (p < 0.05) but to a different level of significance than the silver experiments. This may be explained by the presence of copper in chicken meat as a naturally occurring biological mineral. 1408



Article

Figure 5. Probability density distribution of the final migration level (Q) of copper in chicken and the minimum, maximum, and interquartile range of the corresponding experimental values, following simulations and migration tests of the following conditions: (a) scenario 1, (b) scenario 2, (c) scenario 3, and (d) scenario 4.

to nanoparticles, following the consumption of chicken wrapped in nancomposites based on the laboratory migration results and current population data. Because of the extreme nature of the migration testing conditions, the exposure must also be viewed as extreme. The 95th percentiles of human exposure to silver (as shown in Table 2) were 5.89 × 10−5, 7.06 × 10−5, 8.9 × 10−5, and 7.97 × 10−5 mg kgbw−1 day−1, for scenarios 1, 2, 3, and 4, respectively. The 95th percentiles of human exposure to copper (as shown in Table 2) were 2.26 × 10−5, 5.67 × 10−5, 4.24 × 10−5, and 1.17 × 10−4 mg kgbw−1 day−1 for scenarios 1, 2, 3, and 4, respectively. Putting these values into context in terms of their potential to cause adverse effects is difficult because research on toxicity is sparse. However, a provisional ingestion limit (PIL) for silver nanoparticles of 0.084 mg kgbw−1 day−1 was calculated for silver nanoparticles with a diameter of 10 nm10 based on principles outlined by O’Brien and Cummins,21 which, in turn, were derived from a nanosilver exposure study.22 The exposure to silver nanoparticles calculated in this study ranged from 7.01 × 10−4 to 1.06 × 10−3% of that PIL. In the study by Cushen et al. for similar sized silver nanoparticles migrating from a polyvinyl chloride (PVC) composite into chicken breasts stored at 5 °C for 4 days, 16.02% of the same PIL was reached. A crucial factor in the migration levels seen in the PVC study and the subsequent exposure values was the fill level. The fill level in this current study at 0.5% is 10 times less than that of the comparable study,10 and migration and subsequent exposure levels are far lower because of this.

The model, which took into account background levels (BL) in the food material in mg/dm2, generated from control values, and a model error (e) of 0.32, had mean predicted migration levels (Q) ranging from 2.61 × 10−3 to 5.01 × 10−3 mg/dm2 for silver (Table 2). The model for copper migration, which also included BL and an e of 1.1, had Q values ranging from 3.82 × 10−2 to 4.75 × 10−2 mg/dm2 (Table 2). It can be seen from the R and RMSE values (Figures 4 and 5) that the model more accurately predicted the levels of silver in the samples than the levels of copper in the samples. It is likely that the highly variable BL for copper with a mean of 12.6 × 10−2 mg/kg and a standard deviation of 4.76 × 10−2 mg/kg (n = 8) makes it difficult to capture the true treatment effect. Therefore, the use of a food simulant rather than a real food may be more suited to studying nanoparticles that share a common element with food substances. This study assesses the migration in a nanocomposite−food system not only from nanocomposites with a high nanoparticle fill level but at relatively high temperatures for the storage of perishable goods. An additive to the extreme conditions is the vacuum packaging procedure applied to the individual units. This maximized the contact surface area in the experiments. Hence, the study represents a pessimistic evaluation of potential commercial use and poor storage practices. The predicted migration values increased with time and temperature; this trend corresponds to results where a broader temperature range and longer time periods were studied.11 Exposure Assessment. The exposure model generated probability distributions describing the likely human exposure 1409



Article

composites under varying conditions. It is acknowledged that there is still considerable uncertainty regarding the toxicity of materials at the nanoscale; hence, exposure estimates should be re-evaluated when more toxicity data become available.

No such nanocopper exposure studies exist, and therefore, a PIL cannot be calculated. However, the levels of migration measured in this study are not likely to lead to adverse effects because the simulated exposure levels in this experiment (with approximately half attributed to BL of copper in chicken), with the highest being 1.37 × 10−4 mg kgbw−1 day−1, are very low compared to the average dietary intake of copper. These were found to have averages of 1.42 × 10−2 and 1.9 × 10−2 mg kgbw−1 day−1 for women and men, respectively [calculated using the Irish Universities Nutrition Alliance (IUNA) adult body weight distribution18 and U.S.A. population values expressed in mg/day23,24]. It can be concluded that, if the effect of silver or copper nanoparticles incorporated into PE is optimally effective against the microbial spoilage profile of a particular food (and in line with properties and conditions investigated in this study), then their use may be recommended, because the migration and subsequent exposure (resulting for materials investigated in this study and representing pessimistic scenarios) did not exceed the safety limitation sourced from scientific literature. Having said this, the limits used to assess the safety of silver migration vary hugely,2,10 and safety assessments are highly dependent upon the levels to which they are compared. The limit of 0.01 mg/kg (unauthorized substances) from European legislation that was used to assess the migration from the nanosilver−PE composites and Agion−PE composites to food simulants by Cushen et al.2 is very restrictive. This precautionary approach is due to the lack of appropriate toxicity research available, and it is understandable that the European Commission cannot base permissible migration levels on the results of an insufficient number of toxicity studies. Until more research in this area becomes available, a ban has been imposed on nanomaterials being used in food contact materials,9 unless specifically listed in the Union List of EU 10/2011.9 Neither nanosilver nor nanocopper is mentioned on this list. It is likely that the full antimicrobial potential (by ion release or the particles in nanoform) of nanosilver and nanocopper is excessive for delaying food spoilage at the fill rates tested in this study. The antimicrobial profile (including intensity and longevity) that is required to offset food spoilage should be defined and an antimicrobial system should be designed accordingly, including percentage fill rate, antimicrobial release mechanism, and recommended storage conditions. Such a development could see antimicrobial packaging systems becoming more food-specific, which is justifiable given the effect of food pH observed.2 Because of the association between silver release, antimicrobial potential, and toxicity,25−28 it can be deduced that composites that are excessively antimicrobial are more likely to lead to high exposure levels. Although the effect of both silver nanoparticles and silver ions on humans is largely unknown, recent initial studies into their effects on model organisms suggest that nanoparticles are more toxic than ions.26−28 Considering this, antimicrobial potential and, therefore, fill rate should be minimized within the limits that allow for an acceptable antimicrobial effect. The widespread use of antimicrobials in food packaging depends upon their active behavior, but any commercial success relies on how they perform in migration tests. It can be concluded that this study demonstrates the potential for nanosilver and nanocopper for use in food contact materials, with low migration levels observed in all scenarios tested. The migration and exposure models can be used by industry to study the migration behavior of nanosilver and nanocopper

■

AUTHOR INFORMATION

Corresponding Author

*Telephone: 00353-1-716-7476. E-mail: enda.cummins@ucd. ie. Funding

This work was financially supported by Food Institutional Research Measure (FIRM) as part of the Strategy for Science, Technology and Innovation in the Department of Agriculture, Food and the Marine. Notes

The authors declare no competing financial interest.

■

REFERENCES

(1) Cushen, M.; Kerry, J.; Morris, M.; Cruz-Romero, M.; Cummins, E. Nanotechnologies in the food industryRecent developments, risks and regulation. Trends Food Sci. Technol. 2012, 24 (1), 30−46. (2) Cushen, M.; Kerry, J.; Morris, M.; Cruz-Romero, M.; Cummins, E. Silver migration to food simulants from nanosilver and Agion polyethylene composites. Food Addit. Contam., Part A 2014, manuscript under review. (3) Schrand, A. M.; Rahman, M. F.; Hussain, S. M.; Schlager, J. J.; Smith, D. A.; Syed, A. F. Metal-based nanoparticles and their toxicity assessment. Wiley Interdiscip. Rev.: Nanomed. Nanobiotechnol. 2010, 2 (5), 544−568. (4) Cioffi, N.; Ditaranto, N.; Torsi, L.; Picca, R. A.; Sabbatini, L.; Valentini, A.; Novello, L.; Tantillo, G.; Bleve-Zacheo, T.; Zambonin, P. G. Analytical characterization of bioactive fluoropolymer ultra-thin coatings modified by copper nanoparticles. Anal. Bioanal. Chem. 2005, 381 (3), 607−616. (5) O’Brien, N. J.; Cummins, E. J. A risk assessment framework for assessing metallic nanomaterials of environmental concern: Aquatic exposure and behavior. Risk Anal. 2011, 31 (5), 706−726. (6) Sherman, L. M. Chasing nanocomposites. Plast. Technol. 2004; http://www.ptonline.com/articles/chasing-nanocomposites (accessed Jan 2013). (7) EFSA Panel on Food Contact Materials, Enzymes, Flavourings and Processing Aids (CEF).. Scientific opinion on the safety evaluation of the substance, titanium nitride, nanoparticles, for use in food contact materials. EFSA J. 2012, 10 (3), 2641−2649. (8) Magnuson, B. A.; Jonaitis, T. S.; Card, J. W. A brief review of the occurrence, use, and safety of food-related nanomaterials. J Food Sci. 2011, 76 (6), 126−133. (9) European Union (EU).. Commission Regulation (EU) No. 10/ 2011 of 14 January 2011 on plastic materials intended to come into contact with food. Off. J. Eur. Union 2011, 1−132. (10) Cushen, M.; Kerry, J.; Morris, M.; Cruz-Romero, M.; Cummins, E. Migration and exposure assessment of silver from a PVC nanocomposite. Food Chem. 2013, 139, 389−397. (11) Šimon, P.; Chaudhry, Q.; Bakos, D. Migration of engineered nanoparticles from polymer packaging to foodA physicochemical view. J. Food Nutr. Res. 2008, 47 (3), 105−113. (12) Chen, D.; Qiao, X.; Qiu, X.; Chen, J. Synthesis and electrical properties of uniform silver nanoparticles for electronic applications. J. Mater. Sci. 2009, 44 (4), 1076−1081. (13) Khanna, P. K.; Gaikwad, S.; Adhyapak, P. V.; Singh, N.; Marimuthu, R. Synthesis and characterization of copper nanoparticles. Mater. Lett. 2007, 61 (25), 4711−4714. (14) Gunning, M. A.; Istrate, O. M.; Geever, L. M.; Lyons, J. G.; Blackie, P.; Chen, B.; Higginbotham, C. L. The effect of maleic anhydride grafting efficiency on the flexural properties of polyethylene composites. J. Appl. Polym. Sci. 2012, 124, 4799−4808. 1410



Article

(15) Cruz, J. M.; Sanches Silva, A.; Sendón García, R.; Franz, R.; Paseiro Losada, P. Studies of mass transport of model chemicals from packaging into and within cheeses. J. Food Eng. 2008, 87 (1), 107− 115. (16) Gorham, J.; MacCuspie, R.; Klein, K.; Fairbrother, D. H.; Holbrook, R. D. UV-induced photochemical transformations of citrate-capped silver nanoparticle suspensions. J. Nanopart. Res. 2013, 14 (10), 1−16. (17) Vermeiren, L.; Devlieghere, F.; Debevere, J. Effectiveness of some recent antimicrobial packaging concepts. Food Addit. Contam., Part A 2002, 19 (4, Supplement1), 163−171. (18) Personal communication with Dr. Janette Walton based on information collected for the National Adult Nutrition Survey by Irish Universities Nutrition Alliance (IUNA), Dublin, Ireland, 2011. (19) Song, H.; Li, B.; Lin, Q. B.; Wu, H. J.; Chen, Y. Migration of silver from nanosilver−polyethylene composite packaging into food simulants. Food Addit. Contam., Part A 2011, 28 (12), 1758−1762. (20) Quadros, M. E.; Pierson, R.; Tulve, N. S.; Willis, R.; Rogers, K.; Thomas, T. A.; Marr, L. C. Release of silver from nanotechnologybased consumer products for children. Environ. Sci. Technol. 2013, 47 (15), 8894−8901. (21) O’Brien, N.; Cummins, E. Ranking initial environmental and human health risk resulting from environmentally relevant nanomaterials. J. Environ. Sci. Health, Part A: Toxic/Hazard. Subst. Environ. Eng. 2010, 45 (8), 992−1007. (22) Kim, Y. S.; Kim, J. S.; Cho, H. S.; Rha, D. S.; Kim, J. M.; Park, J. D.; Choi, B. S.; Lim, R.; Chang, H. K.; Chung, Y. H.; Kwon, I. H.; Jeong, J.; Han, B. S.; Yu, I. J. Twenty-eight-day oral toxicity, genotoxicity, and gender-related tissue distribution of silver nanoparticles in Sprague-Dawley rats. Inhalation Toxicol. 2008, 20 (6), 575−583. (23) Tokman, N. The use of slurry sampling for the determination of manganese and copper in various samples by electrothermal atomic absorption spectrometry. J. Hazard. Mater. 2007, 143, 87−94. (24) Shrivas, K.; Jaiswal, N. K. Dispersive liquid−liquid microextraction for the determination of copper in cereals and vegetable food samples using flame atomic absorption spectrometry. Food Chem. 2013, 141 (3), 2263−2268. (25) Xiu, Z.-m.; Zhang, Q.-b.; Puppala, H. L.; Colvin, V. L.; Alvarez, P. J. J. Negligible particle-specific antibacterial activity of silver nanoparticles. Nan. Lett. 2012, 12 (8), 4271−4275. (26) Asharani, P. V.; Wu, Y. L.; Gong, Z.; Valiyaveettil, S. Toxicity of silver nanoparticles in zebrafish models. Nanotechnology 2008, 19 (25), 1−8. (27) Olasagasti, M.; Gatti, A. M.; Capitani, F.; Barranco, A.; Pardo, M. A.; Escuredo, K.; Rainieri, S. Toxic effects of colloidal nanosilver in zebrafish embryos. J. Appl. Toxicol. 2014, DOI: 10.1002/jat.2975. (28) van Aerle, R.; Lange, A.; Moorhouse, A.; Paszkiewicz, K.; Ball, K.; Johnston, B. D.; de-Bastos, E.; Booth, T.; Tyler, C. R.; Santos, E. M. Molecular mechanisms of toxicity of silver nanoparticles in zebrafish embryos. Environ. Sci. Technol. 2013, 47 (14), 8005−8014. (29) Liu, C.; Wang, J.; He, J. Rheological and thermal properties of m-LLDPE blends with m-HDPE and LDPE. Polymer 2002, 43 (13), 3811−3818. (30) Illes, B.; Horvath, B.; Shinohara, T.; Harsanyi, G. Tin whisker growth from Sn−Cu (0−5 wt %) surface finishes. Proceedings of the 17th International Symposium for Design and Technology in Electronic Packaging (SIITME); Timisoara, Romania, Oct 20−23, 2011.

1411


Evaluation and Simulation of Silver and Copper Nanoparticle

Recommend Documents