Long-term wear effects on nanosilver release from commercially available food contact materials

Abstract

Potential consumer exposure to nanoparticles (NPs) from nanoenabled food contact materials (FCMs) has been a driving force for migration studies of NPs from FCMs. Although NP migration from fresh, unused FCMs was not previously observed, conditions that result in significant changes to the surface of FCMs have not been investigated for NP migration into food. Therefore, a quantitative assessment of nanoparticle release from commercially available nanosilver-enabled FCMs was performed using an abrasion protocol to simulate cleaning, cutting, scraping and other stressful use conditions. Laser scanning confocal microscopy (LSCM) analysis showed a general increase in root mean square (RMS) roughness after FCM abrasion, and particle count (for particle sizes from 80 nm to 960 nm) at the surface was 4 orders of magnitude higher for the abraded FCMs. Migration was evaluated using both water and 3% (v/v, volume fraction) acetic acid as food simulants. Low concentrations of total Ag were detected in water simulants with a small portion (<10 ng dm⁻²) in the form of silver nanoparticles (AgNPs). Median particle diameter ranged from 39 nm to 50 nm with particle number concentrations on the order of 10⁶ particles dm^{− 2}. Total Ag migration into 3% (v/v) acetic acid was significantly higher than in water; however, 3% (v/v) acetic acid was not suitable for evaluation of NP release due to dissolution of AgNPs to Ag⁺ under acidic solution chemistries.

Introduction

Silver nanoparticles (AgNPs) have been the nanomaterial of choice for the manufacture of polymer nanocomposites (PNCs) marketed for use in antimicrobial-based food contact applications (Chaudhry et al. 2008; Duncan 2011; Mihindukulasuriya and Lim 2014; Addo Ntim and Noonan 2017). AgNPs reportedly have excellent antimicrobial properties, making them well suited for applications in food preparation and storage to prevent spoilage and reduce food waste. While AgNPs are not currently permitted for food contact applications in the US, products containing AgNPs are available on the worldwide market in products such as food storage containers, cutting boards, and baby bottles (Hatzigrigoriou and Papaspyrides 2011; Lagarón and Busolo 2012; Mihindukulasuriya and Lim 2014). Although there has been increased interest in nanotechnology-based products for food contact applications, there may be concerns about the safety of these products from the viewpoint of potential consumer exposure to NPs that may migrate into food upon contact. This study aims to provide perspective on the concentration of AgNPs released, if any, under various conditions of use to help inform appropriate dose selection in AgNP toxicity studies.

A number of publications already exist in the literature that attempt to address potential exposure and safety concerns by providing migration data for commercially available products as well as for laboratory bench-scale PNCs (Emamifar et al. 2010; Huang et al. 2011; Song et al. 2011; Cushen et al. 2013; Echegoyen and Nerín 2013; von Goetz et al. 2013; Bott et al. 2014; Cushen et al. 2014a, 2014b; Jokar and Abdul Rahman 2014; Addo Ntim et al. 2015; Artiaga et al. 2015). Data reported from these studies have often been contradictory with regards to detection of AgNPs in food simulants. Differences in experimental methodology, analytical instrumentation, and the choice of food simulants used may account for some of these conflicting data. (Addo Ntim et al. 2016). In this study, the experimental approach is consistent across several studies and uses instrumentation with low detection limits to assess migration of silver from FCM products under different use conditions.

Most migration studies on PNCs involve evaluating the amount, if any, of AgNP release from fresh, unused FCMs that are immersed in or filled with food simulant. Data reported from many of these studies show very low concentrations (sub parts per billion) of total silver (Ag) detected in food simulants. Similarly, migration evaluation performed in our laboratory on commercially available AgNP-enabled FCMs showed low total Ag concentrations in simulant using inductively coupled plasma-mass spectrometry (ICP-MS) without any measurable AgNP detection using transmission electron microscopy (TEM) (Addo Ntim, Thomas, Begley and Noonan 2015). Repeat exposure migration assessment of fresh and unused FCMs showed a migration pattern characteristic of surface desorption/dissolution, where the highest total Ag concentration was detected in the first exposure to simulant with total Ag concentration decreasing across three consecutive exposures (Addo Ntim, Thomas, Begley and Noonan 2015). This suggests that any phenomenon that significantly changes the surface of the FCMs may impact the migration profile of silver. Some surface-altering conditions may be commonly encountered in everyday use of FCMs such as knives cutting through cutting boards or abrasive cleaning tools used on food storage containers. Thus, comprehensive studies focused on changes to the FCM surface are important to assess potential release of nanoparticles under practical long-term use conditions.

This project aimed to investigate the potential for NPs to migrate from commercially available AgNP-enabled FCMs into food simulants under use conditions that significantly impact their surface properties. Three food storage containers and a cutting board were first characterised for the presence of NPs. The products were then subjected to rotary abrasion using NIST-developed metallic wheels to simulate wear from long-term and excessive use. Rotary abrasion has been widely used to evaluate wear and rubbing resistance of coatings and paints, as well as nanoparticle release resulting from mechanical forces (Vorbau et al. 2009; Golanski et al. 2011; Schlagenhauf et al. 2012; Sung et al. 2013). After abrasion, a sub-set of FCMs were evaluated for NP release by applying adhesive tape to the surrounding abrasion wheel surfaces. The tape was then digested in nitric acid (details described later) to determine total Ag content using inductively coupled plasma- mass spectrometry (ICP-MS). NP migration from the FCMs was evaluated using water and 3% (v/v volume fraction) acetic acid as food simulants. Migration simulants were analysed by ICP-MS for total Ag and AgNPs in spectra and single-particle modes, respectively.

Materials and methods

Materials

Glacial acetic acid (Optima™), nitric acid (HNO₃, Optima™), hydrochloric acid (HCl, Optima™), and hydrogen peroxide (H₂O₂, Optima™) were purchased from Fisher Scientific (Pittsburgh PA, USA). Water (18 MΩ.cm) was obtained from an Aqua Solutions (Jasper, GA) water purification system. Silver (10 μg mL⁻¹) in 2% (w/w, mass fraction) HNO₃ (NIST traceable standard) was purchased from High-Purity standards (Charleston SC, USA). NIST Standard Reference Material (SRM) 8013, an aqueous suspension of citrate-stabilised gold nanoparticles (nominal 60 nm diameter) were purchased from NIST (Gaithersburg MD, USA). AgNPs used for imaging controls were capped in polyvinylpyrrolidone (PVP) and suspended in water before being dried down onto either a silicon wafer or polyethylene terephthalate (PET). These AgNPs were synthesised at NIST following the procedure in Gorka et al. (2017). Food contact samples advertised as containing nanosilver were purchased through the US-FDA office in China or on the internet from a commercial firm in the US. Four products, including a low-density polyethylene (LDPE) cutting board, three polypropylene-based food storage containers (Container 1, Container 2, and Container 3), and a Ag-free control FCM made of polypropylene (PP) were evaluated. These products are marketed for use in food preparation, consumption and storage under both single and repeat use conditions. No information was provided by the manufacturers on the amount or particle size distribution of nanosilver filler, nor if the silver was applied mainly at the surface or incorporated uniformly throughout the polymer. However, TEM was used to confirm the presence of AgNPs in the bulk FCMs. ICP-MS of acid-digested samples was also used to confirm the presence of total Ag in the bulk of the material (Addo Ntim, Thomas, Begley and Noonan 2015).

Altering the surface properties of the FCMs by Abrasion

Rotary abrasion has been widely used to evaluate wear and rubbing resistance of coatings and paints, as well as nanoparticle release resulting from mechanical forces (Vorbau et al. 2009; Golanski et al. 2011; Schlagenhauf et al. 2012; Sung et al. 2013). The rotary abrader consists of two vertical abrasive wheels rotating about a horizontal axis that abrade the material continuously while the horizontal specimen is rotating about a vertical axis at a fixed speed (Figure S1) (Sung et al. 2013). The abrasion/rubbing action is produced by the friction at the contact line between the material and the sliding rotation of the two wheels, creating a circular track. Certain parameters can be optimised in the abrader to produce measurable levels of stress on an FCM that would result in significant changes to its surface.

A dual specimen table Taber rotary abrader (Model 5155, Taber, North Tonawanda, NY) containing two NIST-developed metallic wheels was used to simulate FCM wear associated with stressful use (Vorbau et al. 2009; Golanski et al. 2011; Schlagenhauf et al. 2012; Sung et al. 2013). In an XPert® Nano Enclosure (1.5 m × 0.725 m, LABCONCO, model 3887561), four FCM samples (3 cm × 3 cm) were securely fitted in a Teflon sample holder (100 mm × 100 mm, Figure S1(a)) and abraded at a speed of 6.28 radians s⁻¹ for 100 or 500 revolutions with a fixed load of approximately 500 g (Figure S1(b)). Abraded FCM surfaces were characterised with laser scanning confocal microscopy (LSCM, described in the next section).

Released materials from a sub-set of the abraded FCMs were collected from the abrasion wheel with double-sided tape onto a carbon substrate and characterised with a scanning electron microscope (SEM, Jeol JSM 6390 LV SEM, Jeol USA, Inc. Peabody MA) equipped with a Thermo 6733A-1NUS-SN energy dispersive X-ray microanalysis system (EDS, Thermo Electron Corporation, Madison WI). Although not all of the released material necessarily ended up on the abrasion wheel, the highest number of released particles were expected at the wheel. Smaller-sized adhesive tape was used to collect released material after abrasion, which was digested with concentrated acid to determine total Ag concentration by ICP-MS. About 5 mg of released material collected with adhesive tape from the LDPE-based cutting board and one of the PP containers (container 3) were placed into perfluoroalkoxyalkane (PFA) vessels and filled with 2 mL of (67–69)% (w/w) HNO₃ and 0.5 mL of 30% (w/w) H₂O₂. The samples were digested in a high pressure microwave reactor (Milestone UltraCLAVE, Milestone Inc., CT USA). The reaction temperature was ramped from room temperature to 200°C over 25 min and held at 200° C for an additional 20 min. After cooling to room temperature, the digestates were diluted with 18 MΩ.cm water (50 mL), stabilising silver ions with 0.5% (w/w) HCl and subsequently analysed by ICP-MS for total Ag. The conditions used for ICP-MS analysis are presented in the supporting information (SI, Table S1).

Laser scanning confocal microscopy (LSCM)

A Zeiss model LSM 510 reflection laser scanning confocal microscope (LSCM) in reflectance mode was used to characterise surface morphology (topographic profile), measure the number and size of surface-exposed particles, and measure the surface roughness of the FCMs. The scanned surface area was varied from 1697 μm × 1697 μm down to 56 μm × 56 μm using different microscope objectives. The incident laser wavelength was 543 nm. The wavelength, numerical aperture (N.A.) of the objective, and the size of the pinhole dictated the resolution in the thickness or axial (z) direction (Kino and Corle 1996). By moving the focal plane, single images (optical slices) could be combined to build up a three-dimensional (3D) stack of images that could be digitally processed. z-Steps were used to obtain overlapping optical slices (a stack of z-scan images). The z-step size described was 1 μm using a 5x objective and 0.1 μm using 50x and/or 150x objectives (Sung et al. 2004; Faucheu et al. 2006).

Root Mean Square (RMS) surface roughness S_q was calculated from 3-D topographic profiles using Zen 2009 software (Jena, Germany). RMS surface roughness was calculated using the maximum height of the image under analysis and a 3 pixel Gauss filter. A surface tilt correlation was used to produce an automatic plane fit. Plane fit is commonly used to remove tilt from images: a first-order polynomial fit was calculated for the entire image and then subtracted from the image. The RMS surface roughness was calculated without a numerical filter according to Equations (1) and (2).

$$S_q=\sqrt{\frac{1}{N_x\cdot N_y}\cdot {\sum }_{i=1}^{N_x}\cdot {\sum }_{j=1}^{N_y}\cdot \left[z\left(x_i,y_j\right)-S_c\right]}$$ (1)

Nx, Ny … number of pixels in X- or Y-direction.

$$S_c=\sqrt{\frac{1}{N_x\cdot N_y}\cdot {\sum }_{i=1}^{N_x}\cdot {\sum }_{j=1}^{N_y}\cdot z\left(x_i,y_j\right)}$$ (2)

where S_c is the mean surface height value of all surface heights. Higher magnification images generally have lower RMS roughness value relative to lower magnification images; this was a result of the increase in resolution and a decrease in the large features observed (Faucheu et al. 2006). Therefore, comparisons were only made within the same magnification and not between magnifications.

Particle analysis was performed on higher magnification LSCM images (150x) since nanoscale particles were the focus of the investigation. It should be noted that the size of one pixel in LSCM is 80 nm based on the collecting optics of the microscope. Actual particle sizes below 308 nm (the diffraction limit of 543 nm light for a 150x objective, numerical aperture of 0.95) cannot be fully resolved due to the diffraction limit of the reflected light. However, particles with a diameter smaller than the diffraction limit of the light used can still be identified and their sizes estimated according to the pixel numbers/area particle occupied. In Figure S2, this is demonstrated with an example image of polystyrene beads below the diffraction limit of incident/reflected light used. Particles below the pixel size of 80 nm can also still be identified but not estimated for size as shown for the approximately 50 nm diameter PVP-capped AgNPs (46 nm ± 14 nm, average and standard deviation of 90 particle diameters measured randomly with ImageJ from three separate scanning electron microscopy (SEM) images (30 particle diameters measured per image, JEOL 7600f, 6.0 kV, Peabody, MA) in Figure S3. These AgNPs were prepared by drop-coating 100 μL of aqueous suspension onto polyethylene terephthalate (PET), which does not reflect light strongly with LSCM, followed by drying off the water and imaging with LSCM to validate that particles of this size could be detected (Figure S3). At a particle size below 80 nm in diameter, the difference between a single nanoparticle or a cluster of nanoparticles within one pixel could not be distinguished. Despite this limitation, individual nanoparticles smaller than 80 nm still appeared as bright pixels but were not distinguishable in terms of size from other nanoparticles or nanoparticle clusters in this same size range (Figure S3).

Particle size analysis was performed using the freeware ImageJ (Wayne Rasband, NIH, USA). The number of pixels associated with particles in the LSCM image (pixel number/area) were converted to particle size in microns to an equivalent area circle and binned by size.

Migration evaluation

Migration studies were performed in accordance with the US-FDA Guidance for Industry for repeat use, on the preparation of premarket submissions for food contact substances (USFDA 2007). The FCMs (3 cm × 3 cm) were weighed and placed into acid-cleaned, clear 40 mL vials with polycarbonate lined caps. Average contact surface area of the FCMs ranged from 0.19 dm² to 0.21 dm². Migration into food simulant (7.5 mL to 10 mL of 3% (v/v) acetic acid or water) was performed by static immersion for 4 h at 100°C. After 4 h, the simulant was removed and replaced by fresh simulant for another 4 h. This was repeated a third time to produce three separate aliquots of simulant. The aliquots were analysed for total Ag and AgNPs by ICP-MS in spectra and single particle mode, respectively. Migration evaluation was performed in three replicates per FCM for each food simulant.

Total Ag concentration in simulants was determined after digesting 1 mL aliquots of simulant with concentrated HNO₃ and diluting (50 x) with deionised water (18 MΩ.cm) and HCl (final concentration - 2% (by mass) HNO₃, 0.5% (by mass) HCl). The digestates were analysed with an Agilent 7900 ICP-MS system (Agilent technologies, Santa Clara, CA, USA) in spectra mode. The limit of detection (LOD) was calculated as the student-t value (99.9% confidence) of the mean of seven blanks multiplied by the standard deviation. The limit of quantitation (LOQ) was calculated as 10 x the standard deviation. The LOD and LOQ were 0.009 ng g⁻¹ and 0.03 ng g⁻¹, respectively.

For AgNP analysis, 1 mL aliquots of simulants were diluted (15x) with deionised water (18 MΩ.cm) and introduced directly into the Agilent 7900 ICP-MS system (Agilent technologies, Santa Clara, CA, USA) using a standard peristaltic pump with Tygon tubing (internal diameter of 1.02 mm), and ASX-520 auto-sampler. The isotope measured was Ag¹⁰⁷. The spICP-MS analyses were performed using an integration time of 0.1 ms for all measurements (Table S1). Blank matrix and ionic solutions were run in single particle mode to determine instrument response to the determined isotope. A rinse solution containing 1% (w/w) nitric acid was used to ensure sample washout between each analysis. Transport efficiency was determined with NIST SRM 8013 (60 nm gold nanoparticles).

Results and discussion

FCM surface abrasion

Physicochemical properties of the FCMs were reported by our team in a previous study (Addo Ntim, Thomas, Begley and Noonan 2015). The polymer used in the FCMs was determined to be PP and LDPE by Fourier transform infrared spectroscopy-attenuated total reflectance (FTIR-ATR). Their densities were also consistent with PP and LDPE. TEM imaging of the FCMs showed inclusions that were characteristic of metal or metal oxide nanoparticles in the size range of approximately 10 nm to 50 nm which were confirmed to be AgNPs by EDS. ICP-MS analysis of the FCMs after microwave-assisted acid digestion showed total Ag concentrations ranging between 7 μg g⁻¹ to 30 μg g⁻¹ (Addo Ntim, Thomas, Begley and Noonan 2015).

The FCMs were abraded to produce measurable wear on the surface of the FCMs and this resulted in visible physical scratching (Figure 1(a)). Released material from a sub-set of the abraded FCMs was collected from the abrasion wheel with double-sided adhesive tape for both SEM analysis (Figure 1(b)) and ICP-MS analysis of total Ag concentration (after acid digestion). Although not all of the released material necessarily ended up on the abrasion wheel, the highest number of released particles were expected at the wheels. SEM imaging of the release material showed polymer particles without any discernable AgNPs. The presence of large volumes of micron-sized polymer particles made it impossible either to identify any AgNPs or to detect any Ag signal with EDS. However, small amounts of total Ag were detected in the release material by ICP-MS after microwave-assisted acid digestion, 0.70 (0.05) μg g⁻¹ and 5.0 (2.0) μg g⁻¹ in the release materials from the cutting board and container 3, respectively. The concentration of total Ag detected in the release material from abrasion was approximately 10% of the concentration of total Ag in the FCMs (7.20 (0.01) μg g⁻¹ and 28.8 (4.6) μg g⁻¹ for the cutting board and container 3, respectively). The percentage of total Ag in the release material was surprising because our sampling of the bulk material suggested that on a micrometre or larger scale, Ag is homogeneously distributed in the FCMs. The low total Ag concentration in the released material may be attributable to the relative ease of removal, by abrasion, of the softer polymeric matrix of the FCM, leaving AgNP aggregates bound to the FCM surface.

Figure 1.

(a) A 100 mm × 100 mm Teflon sample holder with FCM after abrasion with 100 cycles of abrasion with a cross-cut metallic wheel (b) release material from the abraded FCM collected with double sided tape.

Laser scanning confocal microscopy (LSCM)

LSCM images of the FCM surfaces before and after abrasion are displayed as 2-D intensity projections formed by summing the stack of images over the z direction (Figure 2). 2-D intensity projection images are the sum of all the light scattered by different layers of the sample, as far into the sample as light is able to penetrate. The pixel intensity level represents the total amount of back-scattered light; darker areas represent regions scattering less light than brighter areas (Faucheu et al. 2006). Bright spots were visible in the LSCM images of the FCMs before and after abrasion. Bright spots in the image typically represent metallic inclusions (including metal oxides) such as Ag particles or particles from the abrasion wheels, inorganic inclusions, or small polymer fragments containing embedded inorganic particles (Clarke and Eberhardt 2002; Born and Wolf 2013). Particles smaller than or equal to 80 nm, the size of one pixel, could be observed (Figure S3) but not discriminated from other particles within the same pixel (Faucheu et al. 2006). Thus, particle sizes ranging from 80 nm to 960 nm were reported. Elemental composition of the particles could not be confirmed since LSCM does not have elemental analysis capabilities.

Figure 2.

2-D projection images at 150x (56 μm × 56 μm) magnification with average Root Mean Square roughness and standard deviation in four replicate areas of the control polymer and the cutting board before abrasion (before), after abrasion (abraded), after exposure to water (after abrasion + H₂O), and after exposure to acetic acid (after abrasion + acetic acid).

Root Mean Square (RMS) surface roughness S_q was calculated from 3-D topographic profiles using the maximum height of the image under analysis and a 3 pixel Gauss filter. Roughness has a linear log-log scaling relationship to length (L), and the L² is equivalent to the surface area (Faucheu et al. 2006). For all the FCMs evaluated, RMS values were lower at higher magnifications (150x) as expected due to an increase in resolution and a corresponding decrease in large features (Faucheu et al. 2006). For the control polymer and the FCM containers, the RMS roughness values were all comparable prior to abrasion and increased significantly, at both magnifications, after abrasion. The surface roughness then remained statistically unchanged after migration evaluation with water and acetic acid (Table 1). This was not the case for the cutting board sample. The starting roughness for the cutting board was higher than the containers and was comparable to the roughness determined on the containers after abrasion. This is consistent with both visual and tactile characteristics of the cutting board. There were no changes, at either magnification, in the surface roughness of the cutting board after abrasion or after exposure to either of the simulants. It is likely that the intrinsic roughness of the cutting board was large enough that the changes made during abrasion were not detectable by LSCM.

Table 1.

RMS surface roughness Sq of the FCMs before and after abrasion and after migration evaluation using water and 3% (v/v) acetic acid as food simulants. Roughness measurements are the average and standard deviation of roughness measurements from four images taken in different areas of the sample, except for the 5x before abrasion samples, which are the average roughness and standard deviation (in parenthesis) from two images taken in different areas of the sample.

	RMS surface roughness Sq (μm)
	5x Magnification				150x Magnification
Sample	Before Abrasion	After Abrasion	Water Exposed	Acetic acid Exposed	Before Abrasion	After Abrasion	Water Exposed	Acetic acid Exposed
Control FCM	4.1 (1.1)	37.4 (0.4)	34.4 (0.5)	33 (3)	0.04 (0.01)	1.27 (0.05)	1.2 (0.1)	1.1 (0.1)
Cutting Board	42.2 (0.2)	38.7 (0.5)	41.2 (0.7)	41.3 (0.7)	1.7 (0.1)	1.7 (0.2)	4 (3)	2.1 (0.5)
Container 1	4 (1)	36.8 (0.8)	35 (3)	35 (2)	0.05 (0.01)	1.24 (0.03)	1.2 (0.1)	1.1 (0.1)
Container 2	3 (1)	36.5 (0.4)	35 (2)	32.7 (0.6)	0.04 (0.01)	1.22 (0.01)	1.2 (0.2)	0.99 (0.08)
Container 3	2.7 (0.2)	36 (1)	36 (2)	31 (1)	0.03 (0)	1.25 (0.02)	1.4 (0.4)	0.9 (0.2)

Particle analysis was performed on LSCM images at 150x magnification using ImageJ software. Particle counts with sizes ranging from 80 nm to 960 nm increased significantly after abrasion and were at least 4 orders of magnitude higher than the unabraded FCMs (Table 2). It is important to note that the particle counts were determined by the pixel intensities (bright spots) in the LSCM images, which cannot be definitively identified as silver particles. There is the potential for particles from the metallic wheels used in abrasion to be present on the FCM surface or for protruding polymer fragments or inorganic particles to show up as bright spots. The presence of particles on the control FCM surface (PP without AgNP) after abrasion indicates that the majority of the bright spots in the LSCM image used for particle detection were due to particles other than Ag particles: potentially polymer fragments, inorganic particles, and/or particles released from the metallic wheel. Thus, there was not a significant amount of detectable AgNPs at the abraded FCM surfaces, which was consistent with the relatively low total Ag concentration detected in the release material collected on adhesive tape. However, differences in particles counts before and after migration were markedly different for the abraded control (no AgNPs) versus most of the abraded FCMs, with particle counts increasing for abraded FCMs likely due to the removal of loose material on the FCM surface to expose underlying AgNPs.

Table 2.

Particles detected on the FCM surfaces after abrasion and after migration evaluation using water and 3% (v/v) acetic acid as food simulants. Particle counts are represented by the average of four replicate images with coefficient of variation in parenthesis (presented as percentages for ease of comparison).

	Particle count at 150 x (Particle counts from 80 nm – 160 nm)
Sample	Before Abrasion	After Abrasion	Water Exposed	Acetic acid Exposed
Control FCM	6 (67%)	39301 (41%)	33881 (58%)	24713 (79%)
Cutting Board	7892 (47%)	31887 (74%)	35316 (26%)	1116 (68%)
Container 1	2 (22%)	42649 (64%)	67921 (68%)	52934 (74%)
Container 2	3 (38%)	28698 (61%)	80118 (45%)	14190 (53%)
Container 3	6 (67%)	11798 (55%)	24713 (52%)	13313 (92%)

In migration evaluation studies, abraded FCMs were exposed to water or 3% (v/v) acetic acid and then imaged with LSCM. After water exposure, image analysis revealed that particle counts for the abraded containers increased while particle counts for the control FCM surface (PP without AgNP) and cutting board showed a marginal but not statistically significant decrease (Table 2). The increase in particle count observed for the abraded containers may be attributed to removal of loose materials from the container surface such as polymer fragments and particles from the abrasion wheel during water immersion, thereby leading to exposure of underlying Ag particles. In contrast, the control FCM did not lead to an increase in particle count after water immersion because underlying AgNPs were not present when polymer fragments and particles from the abrasion wheel were removed by water. In contrast to water migration studies, particle counts on the abraded FCM surfaces decreased or remained the same after migration evaluation with acetic acid (Table 2). This was particularly prominent for the cutting board, where particle count was reduced to approximately 4 times lower than the unabraded cutting board (Table 2). This was attributed to dissolution of metallic particles by the acidic simulant, consistent with previous observations (Addo Ntim et al. 2016).

Migration evaluation

Migration into water

For all the FCMs evaluated, very low concentrations of total silver (<100 ng dm⁻²) were detected in water simulants (Figure 3). The highest concentration of total Ag was detected in the first of three consecutive aliquots for all the FCMs evaluated and decreased through the second and third aliquots (Figure 3), consistent with previous observations (Addo Ntim, Thomas, Begley and Noonan 2015). The decrease in silver migration with increasing exposure time is characteristic of a surface or near-surface desorption, where the silver available for migration is depleted within the time frame of the experiment, implying that there was negligible contributions from mass transfer of Ag from the bulk polymer (Addo Ntim, Thomas, Begley and Noonan 2015; Duncan and Pillai 2015).

Figure 3.

Total Ag in water simulant after acid digestion as measured by ICP-MS, dashed line represents the highest AgNP mass concentration (not acid digested) detected in water simulant during the first aliquot by sp-ICP-MS. Error bars represent standard deviation of the mean of three replicates.

The low total Ag concentrations detected in water simulant were similar to those previously observed when the unabraded FCMs were evaluated for migration without abrasion (Addo Ntim, Thomas, Begley and Noonan 2015). This shows that the abrasion process did not significantly increase Ag migration into water despite an increase in FCM surface roughness, suggesting that most exposed AgNPs were still partially embedded in polymer. However, in contrast to our previous work with unabraded products, sp-ICP-MS analysis of the water simulant showed that a fraction of the migrating silver from the abraded FCMs was in the form of AgNPs (Figure 3) (Addo Ntim, Thomas, Begley and Noonan 2015). The concentration of AgNPs detected in water simulants ranged from (2–6) ng dm⁻², accounting for roughly 6% to 13% of the total Ag migrating from the abraded FCMs. Particle number concentration was on the order of 10⁶ ([0.5 × 10⁶−3.5 × 10⁶] particles dm⁻²).

A sp-ICP-MS time scan of water simulant showed numerous spikes with a consistent offset of ionic Ag signal from the background (Figure 4(a)) indicating the presence of AgNPs (Degueldre et al. 2006). The size distribution of AgNPs in water simulant was determined using the signal intensity (Pace et al. 2011). Most of the AgNPs released were less than 50 nm in diameter with the median diameter ranging between 39 nm and 50 nm for the different FCMs (Figure 4(b)). These particle sizes are comparable to the sizes determined by TEM, as reported earlier in the results and discussion, during the characterisation of the unabraded FCMs. Particles with diameters greater than 50 nm were also present. The observed nanoparticle release in this study was attributed to degradation of the polymer matrix by mechanical abrasion (Newsome 2014; Noonan et al. 2014; Duncan 2015). Larger particle aggregates protruding from the polymer matrix after abrasion may also have been fully released and then dissolved by the simulant into smaller particles resulting in their release (Duncan 2015; Duncan and Pillai 2015). This is supported by the fact that particle release was observed mostly in the first of three aliquots of simulant collected (Figure 3). Particle detection in the second aliquot was very low, and no particles were detected in the third aliquot for any of the FCMs. The observed decrease in particle release across the three aliquots suggests a surface release phenomenon devoid of actual particle migration from the bulk polymer, consistent with expectations from the physicochemical viewpoint that diffusion-controlled mass transfer of AgNPs in a polyolefin matrix is negligible or exists only minimally for particles with diameters less than 5 nm (Šimon et al. 2008; Bott et al. 2014).

Figure 4.

(a) sp-ICP-MS time scan showing AgNPs in water simulant (b) size distribution of AgNPs released from Container-1 into water simulant (Aliquot 1), (c) Total Ag in 3% (v/v) acetic acid simulant after acid digestion and (d) sp-ICP-MS time scan of 3% (v/v) acetic acid simulant from container 1 (Aliquot 1). Error bars represent standard deviation of the mean of three replicates.

Migration into 3% acetic acid

Migration of Ag into 3% (v/v) acetic acid followed a surface desorption trend similar to what was observed with water simulants, where the highest concentration of total Ag was detected in the first of three consecutive aliquots and decreased with subsequent aliquots (Figure 4(c)). Total Ag concentrations detected in acetic acid simulant for the abraded FCMs were higher than previously reported total Ag concentrations migrating from their unabraded forms (Addo Ntim, Thomas, Begley and Noonan 2015). This increase in total Ag migration was attributed to the increased contact surface area due to abrasion and potential exposure of AgNPs that would have otherwise been buried within bulk polymer and less accessible to the food simulant. Total Ag concentrations detected in 3% (v/v) acetic acid simulant was significantly higher than in water (Table 3), consistent with previously reported data (Addo Ntim, Thomas, Begley and Noonan 2015). The higher concentration of total Ag in 3% (v/v) acetic acid was expected due to the significant difference in the pH of the simulants. Prior work demonstrated that contact with 3% (v/v) acetic acid caused oxidative dissolution of surface AgNPs resulting in higher Ag concentrations (Bott et al. 2014; Addo Ntim et al. 2016). This is consistent with the LSCM data, where particle count on the abraded FCM surface significantly reduced after exposure to 3% (v/v) acetic acid (Table 2).

Table 3.

Ag concentration in first aliquot of simulants after migration tests on abraded FCMs. Ag concentrations are means of three replicates with standard deviation in parenthesis.

	Silver Concentration (ng dm−2)
	Total		Nanoparticles
Sample	Water	3% (v/v) Acetic Acid	Water	3% (v/v) Acetic Acid
Cutting Board	21 (5)	282 (82)	2 (0.4)	< 0.4
Container 1	83 (5)	731 (41)	5 (0.3)	< 0.4
Container 2	46 (7)	369 (45)	6 (1.4)	< 0.4
Container 3	11 (3)	66 (6)	4 (4)	< 0.4

Although total Ag migration was significantly higher in 3% (v/v) acetic acid than water, there were no detectable AgNPs in the acetic acid simulants (Table 3). The sp-ICP-MS time scan of acetic acid simulant produced a stable signal offset from background without any pulses (Figure 4(d)). The absence of detectable AgNPs in 3% (v/v) acetic acid was expected because AgNPs have been shown to have limited stability under acidic conditions (Bott et al. 2014; Addo Ntim et al. 2016). These studies show that in a typical 4 h migration protocol, any AgNPs that may have been released were dissolved to Ag⁺ ions. Migration evaluation with water simulant showed small but detectable AgNPs released from the abraded FCMs; therefore, the potential for AgNP release into 3% (v/v) acetic acid at time periods shorter than 4 h cannot be ruled out. The absence of AgNPs can only be attributed to the relative sensitivity of the Ag/Ag⁺ system to oxidation in the acidic environment. The mechanism of AgNP dissolution is reportedly expedited by the presence of protons (H⁺) in an acidic (low pH) simulant, accounting for the significantly higher total Ag concentration in 3% (v/v) acetic acid (pH 4) relative to water (Peretyazhko et al. 2014).

Conclusions

The findings in the study show that conditions that alter the surface properties of FCMs can potentially influence AgNP release from products when in contact with food. The data shows that rotary abrasion alters the FCM surface in a measurable, reproducible fashion by increasing the surface roughness and the number of particles (particle sizes ranging from ≤ 80 nm to 960 nm) on the FCM surface. The increase in surface roughness and particle count, however, does not translate to an increase in the total Ag concentration migrating into simulant, since total silver concentrations migrating from the abraded FCMs were comparable to those from unused FCMs. The distinguishing factor between unused and abraded FCMs, as evidenced in the data presented here, is that abrasion results in the release of some AgNPs into simulant which was not the case with fresh unused FCMs. This information can help inform dose selection for Ag ions and AgNP in toxicological studies. AgNP release into simulant was attributed to two potential mechanisms: degradation of the polymer matrix by mechanical abrasion resulting in AgNP release; and dissolution of AgNP aggregates protruding from the abraded FCM surface into smaller AgNPs particles that released.