Migration of Quaternary Ammonium Cations from Exfoliated Clay/Low-Density Polyethylene Nanocomposites into Food Simulants.

Clay/polymer nanocomposites (CPNs) are polymers incorporating refined clay particles that are frequently functionalized with quaternary ammonium cations (QACs) as dispersion aids. There is interest in commercializing CPNs for food contact applications because they have improved strength and barrier properties, but there are few studies on the potential for QACs in CPNs to transfer to foods under conditions of intended use. In this study, we manufactured low-density poly(ethylene) (LDPE)-based CPNs and assessed whether QACs can migrate into several food simulants under accelerated storage conditions. QACs were found to migrate to a fatty food simulant (ethanol) at levels of ∼1.1 μg mg-1 CPN mass after 10 days at 40 °C, constituting about 4% total migration (proportion of the initial QAC content in the CPN that migrated to the simulant). QAC migration into ethanol was ∼16× higher from LDPE containing approximately the same concentration of QACs but no clay, suggesting that most QACs in the CPN are tightly bound to clay particles and are immobile. Negligible QACs were found to migrate into aqueous, alcoholic, or acidic simulants from CPNs, and the amount of migrated QACs was also found to scale with the temperature and the initial clay concentration. The migration data were compared to a theoretical diffusion model, and it was found that the diffusion constant for QACs in the CPN was several orders of magnitude slower than predicted, which we attributed to the potential for QACs to migrate as dimers or other aggregates rather than as individual ions. Nevertheless, the use of the migration model resulted in a conservative estimate of the mass transfer of QAC from the CPN test specimens.

Introduction

Clay/polymer nanocomposites (CPNs) are polymeric materials incorporating refined clay particles as fillers. Because of their high specific surface area, exfoliated clays enhance many of the physical properties of plastics.[1] As such, CPNs have been explored for applications including flame-retardant textile coatings,[2−6] high strength automotive and structural components,[7−11] and food packaging.[12−14] In the latter case, dispersed clays improve toughness and tensile strength of the host polymer, and they create a tortuous path for the diffusion of gasses and other molecules.[12] These improvements bring opportunities for better food quality and safety via shelf-life elongation and cost savings and reduced environmental footprint via packaging material downgauging.

The introduction of CPNs and other nanocomposites into the marketplace is not without its challenges. One hurdle is ensuring that CPNs are safe for humans and the environment throughout product life-cycles, particularly the use phase. In association with food packaging applications, studies have explored both the potential toxicity of clay particles[15−20] and the release (migration) of substances from CPNs into food simulants.[20−30] Most published CPN exposure studies have fixated on the inorganic clay particles, with less attention paid to the fate of organic surfactant modifiers, such as quaternary ammonium cations (QACs), used to enhance the compatibility of hydrophilic clays with the hydrophobic polymer phase. This constitutes an important knowledge gap in the peer-reviewed literature, especially given what is known about the potential health and environmental impact of QACs.[31−34]

To date, the only experimental effort to quantify QAC surfactants migrated from CPNs in a food packaging application and documented in the scientific literature was undertaken by Xia and co-workers.[21,22] These studies provided evidence that QACs migrate from CPNs manufactured from poly(propylene) and Nylon 6 into ethanol. The quantity of migrated QACs was found to be significantly higher than the amount of released inorganic clay, suggesting that QACs may constitute the greater exposure risk to consumers from CPNs. A dependence of QAC release on the host polymer, attributed to the strength of interactions between dispersed clay and the polymer, was also indicated. A broader array of QAC migration data, however, is still needed to formulate a general understanding of how embedded clay particles attenuate the QAC migration process. In particular, it is important to verify that common mathematical migration models, which were developed for neutral organic additives to food packaging polymers, provide a conservative estimate of potential exposure to QACs from CPNs under intended conditions of use.

We incorporated organically modified montmorillonite (OMMT) clay particles into low-density poly(ethylene) (LDPE) cast films at several weight percentages and used them to evaluate potential human exposure to QACs from CPNs in a food contact application. LDPE was selected as a host polymer because it finds broad use as a contact layer in food packaging, is inexpensive, and possesses low barriers to migration, which would make it an attractive industry candidate for improvement with clay incorporation. After fabrication, the CPN films were fully characterized, and migration of the QACs from these materials into a series of food simulants was assessed under conditions recommended by the U.S. Food and Drug Administration to simulate extended room-temperature storage.[35] The food simulants used were 100% ethanol (fatty food), water (aqueous food), 3% aqueous acetic acid (acidic food), and 10% aqueous ethanol (alcoholic food). In addition to the CPNs, LDPE films incorporating the surfactant mixture only (no clays) were fabricated to evaluate how the presence of clay particles changes the migration kinetics of the QAC. Finally, we compared the experimental data to predictive migration models to evaluate the suitability of these models for conservative exposure assessments of QAC additives from CPNs.

Results and Discussion

Characterization of OMMT Clay

The OMMT used in this study was Cloisite 20. Cloisite 20 is a purified montmorillonite, a naturally occurring, 2:1 layered aluminum phyllosilicate mineral in the smectite group, modified with bis(hydrogenated tallow alkyl) dimethylammonium chloride. This organic surfactant modifier, which is available under the trade name Arquad 2HT-75, is a mixture of QACs derived from purified triglycerides in beef tallow. Thermogravimetric analysis (TGA) revealed that Cloisite 20 powder is composed of <0.5 wt % water and 61.32 (0.41) wt % inorganic clay (Supporting Information, Figure S1 and Table S3). X-ray diffraction (XRD) analysis of Cloisite 20 powder (Figure A) yielded an interplatelet layer spacing (d001 value) of 3.3 nm, in good agreement with data supplied by the manufacturer (nominal d001 value = 3.16 nm). In contrast, XRD analysis of an MMT powder without organic functionalization (Cloisite 116, BYK Additives) possessed a d001 value of 1.3 nm, demonstrating that organic surface functionalization of the MMT clay weakens interplatelet attractive forces and facilitates delamination during CPN manufacture.

Figure 1

X-ray diffractograms of (A) Cloisite 20 (black) and Cloisite 116 (red) powder and (B) a 5 wt % Cloisite 20/PE-g-MA/LDPE CPN film (blue) and a neat LDPE film (black). The approximate interplatelet spacings (d001 values) are indicated based on the respective peak maxima.

Fabrication and Characterization of CPNs

Cloisite 20 was incorporated into LDPE via melt processing on a laboratory scale, twin-screw microcompounder using a masterbatch process. To facilitate dispersion of the OMMT into LDPE, a compatibilizer (maleic anhydride-grafted polyethylene, PE-g-MA) was used. CPN films with 1, 3, 5, and 7 wt % OMMT in LDPE were prepared as well as negative (PE-g-MA/LDPE only) and positive (Arquad 2HT-75/PE-g-MA/LDPE) control films. Tabulated nominal compositions for each CPN and control material are provided in the Supporting Information (Table S2). For OMMT-containing films, the OMMT/PE-g-MA mass ratio was kept constant at 1:3. This ratio was chosen to ensure good compatibility of the OMMT with the polymer.[36] In the case of the positive control film, LDPE was charged with 4 wt % Arquad 2HT-75. This amount of Arquad 2HT-75 has a similar QAC content to what is bound to Cloisite 20 platelets in the 7 wt % OMMT/PE-g-MA/LDPE film (Table S2). Note that two separate, nominally identical sets of OMMT-containing LDPE films spanning an OMMT weight range of 1–7 wt % were fabricated to check the reproducibility of results, and these are referred to as series I and series II.

After sufficient mixing, the polymer melt was extruded through a 65 mm film die under constant pressure and rapidly cooled with a stream of nitrogen to obtain cast thin films of approximately 40–60 μm thickness. A photograph of representative 5 wt % OMMT/PE-g-MA/LDPE and neat LDPE films is shown in Figure . Both neat and clay-containing films were slightly hazy, which is typical of semicrystalline LDPE, but the addition of clay does not result in additional haze or coloration. TGA of OMMT-containing LDPE films confirmed that the experimental wt % of OMMT reasonably matched the target wt % in all cases (Supporting Information, Table S3). Scanning transmission electron microscopy (STEM) (Figure ) revealed the presence of OMMT aggregates typically on the order of 20–30 nm thick and up to 150–200 nm long, although smaller clusters of platelets and well-separated platelets were also observed. The lateral dimensions of the clay aggregates and isolated platelets observed in the polymer are roughly on the same order of what has been found for exfoliated sodium montmorillonite suspended in water.[37]

Figure 2

Photograph of representative 5 wt % OMMT/PE-g-MA/LDPE nanocomposite (left) and neat LDPE (right) cast films.

Figure 3

Electron microscopy images of 7 wt % (i, ii) and 1 wt % (iii, iv) OMMT/PE-g-MA/LDPE nanocomposites under low (i, iii: scale bar = 2 μm) and high (ii, iv: scale bar = 0.2 μm) magnifications.

XRD performed on the OMMT/PE-g-MA/LDPE film (5 wt % OMMT, series I) shows that the mean interplatelet distance increases to ∼8.8 nm compared to the ∼3.3 nm that was observed for Cloisite 20 powder (Figure ). This finding confirms that mixing OMMT into the polymer melt using extensional flow screws and a high concentration of PE-g-MA compatibilizer leads to effective intercalation of polymer strands into the interplatelet gallery volume. We did not observe a significant difference in the clay scattering peak position as a function of the clay content (data not shown), but this may be due to the poor performance of the XRD hardware at such low scattering angles and/or the low concentration of platelets in films with OMMT content <3 wt %. For this reason, XRD is a poor method for measuring OMMT exfoliation when the interplatelet distance exceeds 8 nm.[38]

Fourier transform infrared-attenuated total reflection (FTIR-ATR) spectroscopy provides additional information about the dispersion of OMMT in the LDPE-based CPNs. MMTs and other layered silicates possess strong absorption bands in the 1000–1100 cm–1 region that arise from silicon–oxygen stretching modes.[39] The silicon–oxygen stretching region is typically understood to be comprised of four overlapping peaks. Stretches involving Si–O–Si linkages parallel to the clay layer surface have their transition moments lying in the plane of the clay platelet layer (peaks I, III, and IV; in-plane stretches), whereas stretches involving Si–O bonds directed toward the alumina octahedra have transition moments oriented perpendicular to the platelet surface (peak II, out-of-plane stretch).[40] Cole has shown that the peak II (∼1080 cm–1) peak max frequency and the peak III/IV (∼1045/∼1024 cm–1) intensity ratio are both sensitive to the degree of OMMT exfoliation in LDPE-based CPNs, with higher degrees of MMT platelet delamination favoring shifts of peak II to higher frequencies and broader bandwidths and larger peak III/IV intensity ratios.[38]

Figure shows FTIR-ATR spectra in the silicon–oxygen stretching region of the series I OMMT/PE-g-MA/LDPE films, all recorded with force settings <1 (minimal compression of the film to enhance contact of the film with the ATR crystal). The spectra of the Cloisite 20 and Cloisite 116 powders are also provided. Compared to the FTIR-ATR spectrum of Cloisite 116 (an unmodified inorganic clay), the silicon–oxygen stretching peak envelope of Cloisite 20 is narrower, although the out-of-plane stretching peak (II) is still obscured by the in-plane stretching peaks (III and IV), showing that inorganic platelets are closely stacked in the organically modified Cloisite 20 powder. In contrast, the incorporation of Cloisite 20 into LDPE at all concentrations results in the out-of-plane stretching vibration (II) coming into clear prominence and narrowing of the peak III/IV spectral envelope. These changes signify intercalation of polymer strands between OMMT platelets and are consistent with the XRD measurements. Interestingly, as the wt % of Cloisite 20 increases from 1 to 7 wt %, the relative prominence of peak II decreases and shifts to lower frequency (Δν = 4 cm–1), and the peak III/IV intensity ratio decreases. These changes mean that OMMT platelets in films with higher total clay loadings exhibit lower mean separation distances between individual platelets. FTIR-ATR spectral features associated with the OMMT in these CPNs are highly sensitive to the pressure used to enforce good contact between the film samples and the diamond ATR crystal (Supporting Information, Figure S2). Generally, increasing the force gauge pressure results in an intensification of peak II and little change in the peak III/IV intensity ratio. This supports the conclusion that polymer-dispersed clay platelets are at least partially aligned with the film plane, probably during the drawing of the polymer melt through the film die, and they become more aligned during compression of the film for FTIR-ATR analysis. A more detailed explanation of these effects is provided in the Supporting Information.

Figure 4

FTIR-ATR spectra of the silicon–oxygen stretching region of LDPE films containing (a, black) 1 wt %, (b, red) 3 wt %, (c, green) 5 wt %, and (d, blue) 7 wt % Cloisite 20 and of neat (e, teal) Cloisite 20 and (f, orange) Cloisite 116 MMT powders. The spectra have been offset for clarity. In-plane (I, III, and IV) and out-of-plane (II) silicon–oxygen stretching vibrations are indicated. All spectra were recorded with a force gauge setting of <1. The spectrum for 1 wt % Cloisite 20 in LDPE was subjected to a 1% LOESS curve smoothening function to improve clarity.

The crystalline content and melting behavior of the LDPE phase in OMMT/LDPE CPNs were also evaluated. Crystallinity values and melting points are provided in Table . XRD curves in the 10–40° 2θ range and extracted crystalline parameters are plotted in the Supporting Information, Figures S3 and S4. As the OMMT content is increased, the crystallinity of the polymer phase decreases significantly from 47.1% in the neat LDPE film to 33.6% in the film containing 7 wt % OMMT. It is unclear whether it is the clay particles, QAC surfactants, or the PE-g-MA compatibilizer that interferes with LDPE crystallization at higher clay loading. Previous work has shown that exfoliated clay fillers in polymers may decrease the crystallinity of the polymer phase by restricting the movement of polymer chains;[41] concurrently, we note that the crystallinities of the control films containing 21/79 wt % PE-g-MA/LDPE (46.4 ± 1.1%) and 2.8/21/76.2 wt % Arquad 2HT-75/PE-g-MA/LDPE (43.9 ± 2.1%) are only slightly lower than that determined for neat LDPE, suggesting that the exfoliated clay in the CPNs is primarily responsible for the decreased crystallinity of the polymer phase in these materials. Interestingly, the XRD data show that as the clay content increases, the relative peak area of the 110 and 200 Bragg reflections decreases (Table ). A lower ratio has been interpreted to signal an improved alignment of the polymer chains within crystallites with the drawing direction.[42] Since the drawing torque is nominally the same for all of the films, the enhanced alignment of polymer crystallites in CPNs with higher clay content must be primarily related to the change in CPN composition, possibly due to a more strongly preferred orientation of crystals during growth in close proximity to clay particle surfaces. A similar effect has been observed for graphene/PE composites[43] and other exfoliated clay systems.[44,45] This would be consistent with the FTIR-ATR data presented above, which show that clay particles are at least partially aligned with the film plane during the manufacturing process, and this alignment apparently defines the anisotropic growth of polymer crystallites as well.

Table 1

Melting Points and Crystalline Character of CPN Filmsa

film IDb	crystallinity [%]	A110/A200c	melting pointd [°C]
neat LDPE	47.1 (0.3)	7.05 (0.07)	111.11 (0.15)
1% OMMT/LDPE	43.2 (1.1)	5.06 (0.15)	110.45 (0.06)
3% OMMT/LDPE	40.7 (1.3)	4.15 (0.51)	110.44 (0.04)
5% OMMT/LDPE	37.8 (2.4)	3.39 (0.25)	110.23 (0.27)
7% OMMT/LDPE	33.6 (0.7)	2.21 (0.02)	109.83 (0.10)

The values presented in the table are averages, and the standard deviations are provided in parentheses. N = 4 for the XRD results (crystallinity and A110/A200 ratios) and N = 3 for the melting points.

These characterization data are reported for the series I of CPN films. See the Supporting Information, Table S2 for a complete compositional description of these materials.

This value represents the ratio of peak areas for the (110) and (200) LDPE crystallite Bragg reflections. The areas were determined by fits of the XRD data to Voigt line shape profiles.

The melting points were determined by differential scanning calorimetry (DSC) and reflect the melting endotherm peak maxima during the first heat in a heat–cool–heat cycle.

Characterization of QACs in Food Simulants

In an early experiment to ascertain the potential for QACs from Cloisite 20 to migrate from CPNs to food simulants, 42 mm diameter sections were punched from a 7 wt % OMMT/PE-g-MA/LDPE CPN film (and a neat LDPE control film), and the sections were stored in 100% ethanol for 30 days at 75 °C. A temperature of 75 °C was used for this preliminary work to accelerate the migration process and aid in the initial method development. After the experiment, the solutions were cooled, upon which a white precipitate formed in the ethanol contacting the OMMT-containing films that was assumed to be migrated QACs. Representative mass spectra of these ethanolic solutions in positive ionization mode over the m/z range of relevance to the target QACs are shown in Figure . Spectra over the full m/z range are shown in the Supporting Information, Figure S5. Both solutions produced common peaks in the m/z = 325–375 and 600–700 regions, but the solution stored in the presence of 7 wt % OMMT/PE-g-MA/LDPE exhibited additional peaks in the m/z = 450–560 region.

Figure 5

Representative high-resolution mass spectra of a 100% ethanol solution in which a neat 42 mm circular section of (A) neat LDPE or (B) 7 wt % OMMT/PE-g-MA/LDPE has been stored at 75 °C for 30 days. The identified peaks correspond to QACs in Arquad 2HT-75. The four peaks highlighted in red are the target ions (Table ), with the major component ion identified. Other peaks are also QACs but were not analytical targets. The m/z values have been rounded to the nearest hundredth to simplify the presentation.

Table 2

Chemical Structures of Target QACsa in Arquad 2HT-75

Multiple QACs in Arquad 2HT-75 share the same mass. For example, C20–C16, C19–C17, and C18–C18 are all in Arquad 2HT-75 and all have the exact same mass of 550.62 g mol–1. Due to certain fatty acids being significantly more abundant in beef tallow, a single QAC is usually responsible for most of the peak intensity in the mass spectrum at any given m/z value. To simplify the discussion, we refer to the peaks by the most statistically abundant isomer giving rise to that peak, which is indicated in the “major ion” column.

The relative proportion of each QAC in the sample, calculated by dividing the integrated peak area at the corresponding m/z value by the total peak area for all QACs over a range for m/z = 410–590 in the mass spectrum of a 10 ppm infusion of Arquad 2HT-75 in ethanol.[46] Note that because of natural variation in the lipid content of beef tallow, QAC abundances vary slightly for different Arquad 2HT-75 lots. For simplicity, the Cloisite 20 used for PCN manufacture is assumed to have a similar QAC profile to the pure Arquad 2HT-75 lot.

Arquad 2HT-75 is composed of QACs that contain a central nitrogen bonded to two methyl groups and a statistically determined pair of aliphatic carbon chains derived from the lipid profile of beef tallow. In an earlier methodological paper,[46] we provided a description of the distribution of QACs present in an Arquad 2HT-75 lot that could migrate to food simulants from CPN films fabricated with Cloisite 20. We have focused our analysis on four ions that together comprise the majority (>85%) of QACs in Arquad 2HT-75, and their chemical identities and molecular weights are summarized in Table . Note that these ions correspond to four of the major peaks in the m/z = 450–560 region of the mass spectrum of the ethanolic solution that was contacting the 7 wt % OMMT/PE-g-MA/LDPE film (Figure ), and the peak intensities closely match the corresponding weight-based abundances expected from a pure sample of Arquad 2HT-75 (Table ). Based on this preliminary analysis, we concluded that QACs were migrating from Cloisite 20-containing CPNs in significant amounts, and we could use the QAC markers at m/z = 466.5, 494.6, 522.6, and 550.6 to quantify this process. Note that the additional peaks that show up in the mass spectra of the ethanol solutions stored in the presence of both the OMMT/PE-g-MA/LDPE and neat LDPE films (Figure S5) correspond to additives (e.g., slip agents, antioxidants) in the LDPE polymer resin; by cross-referencing the experimental m/z values with tables of common polymer additives,[47] some of them could be identified (see the Supporting Information, Table S4).

Kinetics of QACs Migration from CPNs into Ethanol

The kinetics of QAC migration from 42 mm diameter sections of 7 wt % OMMT/PE-g-MA/LDPE films was measured at 40 °C over a period of several weeks. Figure plots the sum of the masses of the four target QAC ions migrated from this CPN (blue line) into 100% ethanol versus immersion time, as determined by flow-injection mass spectrometry (FI-MS). The migrated QAC mass has been expressed relative to the initial film section mass to account for the varying film thicknesses for each of the three experimental replicates. The amount of QACs migrated from negative (PE-g-MA/LDPE film, black line) and positive (PE-g-MA/LDPE containing a similar amount of QACs from Arquad 2HT-75 but no clay, red line) control materials is also plotted. Tabulated migration data for all four tracked QACs are provided in the Supporting Information, Table S5.

Figure 6

Total mass of QACs migrated from 7 wt % OMMT/PE-g-MA/LDPE (blue), 4 wt % Arquad 2HT-75/PE-g-MA/LDPE (red), and PE-g-MA/LDPE (black) into 100% ethanol after storage at 40 °C, plotted as a function of storage time. The total QAC migration is the sum of QACs having molecular weights of 550.6, 522.6, 494.6, and 466.5 g mol–1, divided by the initial film mass. Error bars represent the standard deviation from the mean (n = 3). The gray bar is meant to highlight migration at 10 days. The dashed blue line shows the 7 wt % OMMT data multiplied by 5 to highlight the trend.

The kinetic data (Table S5) show that at the first time point the three heavier QACs are quantifiable in the simulant stored in the presence of the film containing Cloisite 20, and all four QACs are quantifiable by 1 day of storage at 40 °C. The summed concentrations of all four QACs continually rise as the storage time increases and is still rising even at 29 days (696 h), indicating a kinetically controlled migration process. QACs at the four target masses migrate at approximately the same rate, as shown by normalizing the mean released mass of each QAC at each time point to the average amount released at 10 days (Figure ). This trend was unexpected, because smaller polymer additives tend to migrate faster than larger ones. The release of QACs from the Arquad 2HT-75/PE-g-MA/LDPE film containing a similar concentration of QACs, but no clay, also increases over time, but the migrated concentrations are significantly higher than the amount migrated from the clay-containing film (Figure , red line). Unlike what is observed in the OMMT-containing CPN, no further increase in the amount of migrated QACs is observed in the Arquad 2HT-75 film after 15 days, suggesting that equilibrium may be reached sooner for the positive control film than for the CPN. The longer equilibration time in the CPN film may be in part due to the tortuous path effect created by the exfoliated clay particles, which is known to slow rates of molecular diffusion in CPNs.[12] This is consistent with gas barrier measurements we conducted on an earlier batch of CPNs, which revealed a reduction in the oxygen transmission rate from 10.96 ± 0.40 × 10–10 μmol O2 @STP m–1 s–1 Pa–1 in a neat LDPE film to 7.32 ± 0.61 × 10–10 μmol O2 @STP m–1 s–1 Pa–1 in a 7 wt % OMMT/PE-g-MA/LDPE film.

Figure 7

Mass of QACs with m/z = 550.6 (C18–C18, black square), 522.6 (C18–C16, red circle), 494.6 (C16–C16, red upward triangle), and 466.5 (C16–C14, blue downward triangle) migrated from a 7 wt % OMMT/PE-g-MA/LDPE film into 100 % ethanol as a function of time spent at 40 °C, divided by the corresponding amount of QAC migrated at 10 days. Error bars represent standard deviations. A trend line (black) has been added to guide the eye.

The FDA recommends a standard testing protocol of 10 days at 40 °C to simulate the potential migration during long-term storage at ambient temperature. This recommendation stems from experimental studies showing that migration levels under these accelerated conditions are approximately the same as those observed after extended storage (6–12 months) at 20 °C (68 °F).[35] Therefore, the 10 day time point serves as a suitable benchmark at which to compare QAC migrant levels from different materials and into different food simulants. After 10 days at 40 °C, the total QAC level migrated from the 7 wt % OMMT/PE-g-MA/LDPE film was measured to be 998.43 ± 85.36 ng mg–1 CPN material. The four target ions constitute about 89 wt % of all of the QACs present in Arquad 2HT-75 (Table ). Assuming the remaining QAC ions not quantitated by our method also migrate in approximately proportionate amounts, the total amount of QACs migrated from the CPN material after 10 days is expected to be ∼1120 ng mg–1 CPN. Because the CPN contains 2.837 wt % (28.37 μg mg–1) Arquad 2HT-75 (Table S3C), this suggests that the percent total migration (amount migrated divided by the initial amount of QACs in the film) was about 3.9% after 10 days storage. By comparison, the % total migration of QACs from the 4 wt % Arquad 2HT-75/PE-g-MA/LDPE film (red line, Figure ) was about 66.1%, >16 times higher. No migration was observed from the negative control film (neat LDPE with PE-g-MA only).

In organically modified MMTs, QACs bind to clay particles via strong ionic interactions between the positively charged QAC head groups and negatively charged defects in the silicate surface layer.[48] We conclude that strong binding of QAC surfactants by the high surface area clay platelets restricts their release into the environment from CPNs during prolonged storage. To verify this hypothesis, we performed aggressive extraction tests of the Arquad 2HT-75/PE-g-MA/LDPE and OMMT/PE-g-MA/LDPE films into methylene chloride for several days at 50 °C and again found >90% QAC recovery values from the Arquad 2HT-75 film but recoveries typically in the range of 3–10% for the CPNs. Likewise, the extraction of QACs directly from Cloisite 20 powder into neat ethanol at 40 °C showed recoveries of the different QAC ions ranging from 10 to 25% and remained essentially unchanged over 96 h. These experiments confirm that strong binding of QACs to the clay particles is responsible for the significantly reduced migration of QACs from the CPN compared to migration from the positive control material containing Arquad 2HT-75, in which all of the QACs are free to diffuse out of the polymer matrix.

Modeling Migration of QACs into 100% Ethanol

To better understand QAC migration to a fatty food simulant, we compared the migration data to a diffusion-based theoretical model. Because Arquad 2HT-75 is a mixture of QACs having different sizes, we focused on QACs with a molecular weight of 550.6 g mol–1, which are dominated by a single geometric isomer (C18–C18). The key parameters that determine mass transfer of an additive from a polymer to a food simulant are the diffusion rate constant, DP, and the relative solubility of the additive in the two phases, embodied in the partition coefficient, KPF. The amount of QAC migrating from the CPN film into the surrounding food simulant at time t (MF,) can be determined from these two parameters using Fick’s second law, which has the following analytical solution[49,50]Here, A is the contact area (cm2); cP, and cF, are the concentrations (μg g–1) of QAC in the polymer and simulant at time t; ρP and ρF are the densities (g mL–1) of the polymer and simulant; VP and VF are volumes (mL) of polymer and simulant; and dP is the film thickness (cm). Because migration may happen from either side of the film, the total two-sided surface area is considered, and the effective film thickness through which migration occurs is taken as half of the actual thickness. Note that this model assumes that the food simulant is well-mixed, which prevents locally high concentrations of the migrant at the polymer–simulant interface from reducing the concentration gradient that drives the migration process. Under the assumption that the solubility of QACs in the CPN and the Arquad 2HT-75 positive control film is similar, the ratio of cP,∞ and cF,∞ was determined from the kinetic experiment with the Arquad 2HT-75 film at 29 days (696 h), where no further migration of QACs into the ethanol was observed. This resulted in an estimated KP,F ∼ 205 and α ∼ 5. Although a good estimate of initial QAC concentration in the polymer is known, most of these QACs are unavailable to migrate due to binding through strong ionic interactions to clay surfaces; for this reason, an upper limit effective initial QAC concentration was estimated as 10% of the actual QAC concentration, based on aggressive extraction experiments in which QAC recovery ranged from 3 to 10%. Additional information on how these and other parameters were determined is provided in the Supporting Information.

Figure plots experimental migration data for m/z = 550.6 QACs from the 7 wt % OMMT/PE-g-MA/LDPE film into 100% ethanol at 40 °C on a μg cm–2 basis, overlaid with data simulated using eq . An initial estimate of DP was made using a semiempirical model that was developed specifically for (conservatively) predicting migration from food contact materials[50] and relates the upper-bound diffusion constant, DP*, to the migrant’s molecular weight (Mr), the absolute temperature (T), and a polymer-specific parameter, (AP)Using an AP value of 11.5 for pure LDPE[49] and an Mr value of 586.85 g mol–1 (to account for a chloride counterion), DP* = 1.4 × 10–9 cm2 s–1. It is immediately apparent from Figure that the migration of C18–C18 is slower than what would be predicted from this estimated diffusion coefficient, in which migration would reach equilibrium within just a few hours. Via a least square fit of eq to the experimental data, the real DP appears to be on the order of 1.6 × 10–12 cm2 s–1, almost 3 orders of magnitude smaller than that predicted by the semiempirical model. Although eq is designed to return a conservative estimate of DP to ensure an adequate safety margin for the predicted migration, (DP*/DP)1/2 tends to be in the range of 2–8.[50] Here, (DP*/DP)1/2 ∼ 30. We note that our experimental DP value for the C18–C18 QAC is about three times larger than that determined by Xia et al.[22] in a poly(propylene)-based CPN (DP ∼ 4.8 × 10–13 cm2 s–1); this is consistent with the higher permeability of LDPE in comparison to poly(propylene).

Figure 8

Experimental data (black squares) for migration of QACs with m/z = 550.6 (C18–C18) from a 7 wt % OMMT/PE-g-MA/LDPE film along with data simulated from an analytical solution to Fick’s Second Law (eq ) for two different diffusion coefficients. The larger diffusion constant (DP* = 1.4 × 10–9 cm2 s–1) was determined from a conservative semiempirical model (eq ). The smaller diffusion constant (DP = 1.6 × 10–12 cm2 s–1) was determined by a least square fit to the averaged experimental data. Each experimental data point is the mean of three independent replicates and error bars represent standard deviations. Note that when using DP* (blue line), equilibrium is predicted to be reached within a few hours.

It is possible that some of the difference between the estimated and experimental diffusion constants for QACs in the LDPE-based CPN may be accounted for by the tortuosity effect of dispersed clay particles, but this effect is typically modest. Approximations used in the various input parameters in eqs and 2 (e.g., the AP value, which was determined for pure LDPE) may also account for some of the discrepancy. Nevertheless, the magnitude of the difference led us to question the fundamental validity of the diffusion model, particularly the assumption of the isotropic distribution of QACs in the polymer that are available for migration. The ionic nature of QACs implies strong interactions between the polar head groups when they are dispersed in the relatively hydrophobic polymer interior, such that individual QACs may not diffuse independently of each other. This hypothesis is supported by prior work that has shown that QACs aggregate strongly in nonpolar solvents, typically as dimers when the QAC side chains are long and the QAC concentration is moderate.[51] We note that C18–C18 dimers would have a molecular weight (with two chloride counterions) of 1172.1 g mol–1, and eq predicts DP* = 3.2 × 10–11 cm2 s–1 for these aggregates. This is in more reasonable agreement with the experimental data [(DP*/DP)1/2 ∼ 4.5]. The tendency for QACs to migrate as aggregates is supported by the observation that QACs with different molecular masses (sizes) exhibit almost identical migration kinetics (Figure ): aggregate composition would be determined statistically, and, thus, the variation in their molecular weights would be smaller than the variation in the molecular weights of individual QACs. The aggregation of QACs may also help to rationalize the complex character of the QAC migration curves observed for the Arquad 2HT-75 positive control film. Attempts to compare these experimental data to eq were unsuccessful (Supporting Information, Figure S6), because the functional forms are quite different. In the positive control film, the “free” QAC concentration is substantially higher because of the absence of clay platelets, and it has been noted that as the QAC concentration increases in nonpolar media, the formation of higher-order aggregates may be implicated.[51] A mixture of different sized aggregates, not to mention the possibility of blooming (phase separation) and other interfacial effects that may occur at such high QAC concentrations, would result in complex migration curves.

Impact of Simulant Chemistry on QAC Migration

Due to poor solubility of the tallow-derived QACs in the polar aqueous phase, it was anticipated that migration into aqueous (100% water), alcoholic (10% ethanol in water), and acidic (3% acetic acid in water) food simulants would be negligible, and this is borne out in the experimental data (Table ). No QACs could be quantified in the 10 day, 40 °C samples for any of these food simulants, suggesting that significant QAC exposure is only likely when CPNs are in prolonged contact with fatty/oily foods. Note that quality-control (QC) samples (Arquad 2HT-75 dissolved in simulant) performed alongside migration tests exhibited low recovery rates in simulants containing significant amounts of water. Rinsing the QC sample vials with 100% ethanol and including the rinses in the QC analysis resulted in recovery rates in excess of 90% (Supporting Information, Table S8), suggesting that QACs adhere to the container walls in the presence of aqueous simulants. There was some concern that the lack of measurable QAC concentrations in the experimental samples may have been due a similar phenomenon. Applying similar rinsing steps to the aqueous simulants stored in the presence of OMMT-containing film samples, however, still resulted in no quantifiable QAC migration, reaffirming that the lack of quantifiable QACs in the aqueous simulants is due to slow migration, not an experimental artifact.

Table 3

Migration of QACs from Cloisite 20/PE-g-MA/LDPEa into Four Food Simulants after 10 days at 40 °Cb

simulant	MW = 550.6 [μg g–1]	MW = 522.6 [μg g–1]	MW = 494.6 [μg g–1]	MW = 466.5 [μg g–1]	total [μg g–1]
100% ethanol	149.00 (16.27)	165.38 (18.77)	72.20 (8.24)	15.55 (1.72)	402.13 (26.23)
water	<5.59	<6.21	<5.59	<5.59	<22.98
10% ethanol in water	<4.50	<4.50	<4.50	<4.50	<17.99
3% acetic acid in water	<3.85	<3.85	<3.85	<3.85	<15.39

The nominal mass ratio of Cloisite 20 to PE-g-MA to LDPE in this film prior to the migration test was 7:21:72.

The reported values are the mean masses of the respective QACs transferred to 100 mL of each food simulant per unit film mass, as measured by flow-injection mass spectrometry. The values in parentheses represent standard deviations of three independent replicates. Samples with reported values after a less than “<” symbol had at least one replicate with a concentration below the limit of quantitation (LOQ). The reported value was determined by assuming the concentration in these replicates was equal to the LOQ value and then calculating the mean across the replicates; an upper limit to the migration is implied by this result.

Impact of OMMT Concentration and Temperature on QAC Migration

Experimental migration data in ethanol for the series of CPNs with Cloisite 20 content ranging from 1 to 7 wt % are plotted in Figure , and the data are tabulated in the Supporting Information, Table S6. Although it is generally observed that the amount of QACs released increases as the initial Cloisite 20 concentration in the films increases, the trend is not exactly linear. Initially, it was thought that the deviations from linearity may be due to unavoidable film-to-film variations in the manufacturing processes (e.g., the cooling rate of the melt), which could impact critical parameters like crystallinity that affect polymer permeability. However, a second completely independent series of films (series II, red line) shows almost an identical relationship between Cloisite 20 content and potential exposure to QACs. We note that our extensive characterization data showed that the degree of MMT interplatelet spacing, polymer crystallinity, and both platelet and crystallite orientation relative to the polymer machine direction changes in complex ways as the amount of clay content increases, and each of these factors can play significant and often conflicting roles in determining the macroscopic barrier properties.[52] Additionally, a swell test using thermogravimetric analysis (100 °C hold for 20 min) to monitor absorbed ethanol loss showed that although neat LDPE absorbed a negligible amount of ethanol (<0.1 wt %) during 40 °C storage conditions, the presence of 21 wt % PE-g-MA in LDPE enabled the polymer to absorb up to 1.1 wt % ethanol. Although the degree of swelling in PE-g-MA-containing LDPE is small, it may contribute to non-negligible changes in the permeability of the host polymer to migrating QACs. These various complex relationships are likely responsible for the trends observed in Figure , but more systematic study will be needed to decouple the role of the clay dispersion characteristics and host matrix polarity/permeability in determining QAC migration levels.

Figure 9

Total mass of QACs migrated from Cloisite 20/PE-g-MA/LDPE films, per unit film mass, as a function of the initial Cloisite 20 mass percentage, after storage for 10 days at 40 °C. The mass ratio of Cloisite 20 to PE-g-MA was 1:3. The total QAC migration is the sum of the masses of migrated QACs having molecular weights of 550.6, 522.6, 494.6, and 466.5 g mol–1, divided by the initial film mass. The red and blue lines represent data acquired from experiments using two completely different sets of PNC films, which have been called series I and series II. Error bars along the y-axis and x-axis direction represent, respectively, standard deviations from the mean migrated total QAC mass (n = 3) and from the mean initial Cloisite 20 content in the films (n = 3).

The impact of temperature on the QAC release from two independent 5 wt % OMMT/PE-g-MA/LDPE films was also investigated. For both CPNs investigated (one from series I and one from series II), the amount of QAC migration increased exponentially with temperature, as expected from an Arrhenius-type process (Supporting Information, Figure S8).

Conclusions

This study found that QACs migrated from LDPE-based CPNs into a fatty food simulant (100% ethanol) under simulated long-term storage conditions. For the CPN incorporating 7 wt % organically modified clay, the amount of migrated QACs of varying molecular weights was on the order of 1.12 ± 0.10 μg mg–1 of the polymer into 100 mL of the simulant with a contact area of 27.7 cm2. The migrated QACs represented only a small (<10%) fraction of the total QACs initially added to CPNs as components of organically modified clay filler. Evidently, most of the QACs are irreversibly bound to the embedded clays and are not available for migration, a fact that was verified by aggressive extraction tests into ethanol from the neat OMMT powder. The amount of migrated QACs scaled with both temperature and the initial clay concentration added to the CPNs. These trends were complex and likely influenced by such factors as the crystallinity of the polymer phase, the dispersion characteristics of embedded clay particles, the amount of PE-g-MA co-compatibilizer added, and the processing parameters (e.g., melt temperature and spooling conditions). Additional work will be needed to disentangle the influence of these parameters on the mass transport. Negligible QACs were found to migrate from CPNs into acidic, aqueous, and alcoholic simulants due to the poor solubility of QACs in these solvents.

Migration tests are expensive and technically challenging, so it is often desirable to be able to estimate, rather than experimentally measure, potential exposure to new food contact substances. Mathematical models currently employed for this purpose usually incorporate a safety margin to ensure that the projected migration is likely to be less than the experimental migration. Because CPNs are complex biphasic materials that possess attenuated molecular diffusion rates, it was a prime goal of this work to evaluate whether migration models commonly used to predict exposure to food contact substances from conventional polymers also provide a similarly conservative estimate of exposure to additives incorporated in these novel materials.

In the case of the CPNs studied here, the experimental migration was less than any migration projection that was made. A summary of these results for the C18–C18 QAC is shown in Table . The crudest estimate was achieved by assuming all of the QAC initially present in the film ends in the food simulant (100% migration), but this approach overestimated the exposure by a factor of nearly 25. Refinements to the model, such as accounting for the fact that at least 90% of QACs is bound irreversibly to clay platelets, provides a more accurate, but still conservative, exposure assessment. Interestingly, the use of a diffusion coefficient determined by a common semiempirical method and the default partition coefficient used in migration models for migrants very soluble in the food simulant (KP,F = 1) did not improve the accuracy of the migration model compared to simply assuming 100% migration of the free QAC fraction. This is because the predicted diffusion coefficient DP* is over three orders of magnitude larger than the experimental diffusion coefficient DP, which ensures that equilibrium is more or less reached by the 10 day time point, and at equilibrium, the migration approaches 100% when the simulant volume is large. The discrepancy between the diffusion constant determined by a fit of the experimental data to the Fick’s Second Law solution (DP) and that estimated from the QAC molecular weight (DP*) was speculated to be due to the aggregation of QACs in the hydrophobic polymer phase. This finding implies that QAC migration in these materials may be non-Fickian, particularly when the concentration is very high, and additional factors, such as surface blooming, may play a role in mass transfer dynamics.

Table 4

Predicted and Experimental Migration of C18–C18 QACs from a 7 wt % OMMT/PE-g-MA/LDPE Film Sections into Ethanol after 10 days Storage at 40 °C

model	predicted migration [μg cm–2]
100% migrationa	27.71 ± 2.99
100% migrationa, free QACs onlyb	2.77 ± 0.30
Fickian diffusion modelb,c	2.77 ± 0.30
DP* = 1.9 × 10–9 cm2 s–1d, KP,F = 1e
Fickian diffusion modelb,c	2.31 ± 0.25
DP* = 1.9 × 10–9 cm2 s–1d, KP,F = 205f
Fickian diffusion modelb	1.13 ± 0.01
DP = 1.6 × 10–12 cm2 s–1g, KP,F = 205f
experimental	1.08 ± 0.25

All available QACs in the film migrate to the simulant.

Only 10% of QAC is unbound to clay and available to migrate.

See eq .

Determined from eq using a molecular weight of 586.85 g mol–1.

Recommended value of the partition coefficient to use in the absence of experimental data, when the migrant is expected to be very soluble in the simulant.[49]

Partition coefficient estimated from experimental data.

Determined by a least square fit of eq to the experimental data. In all cases, the predicted migration represents an average across three independent replicates having the same contact area but different thicknesses. The error is a standard deviation.

Despite the complexity of clay/polymer materials and the ionic nature of QACs, migration models are satisfactorily conservative for the CPNs evaluated in this study. Of course, because CPNs differ markedly in their composition, crystallinity, and other properties, it will be necessary to study additional examples to be confident that these findings are broadly applicable to this class of materials. We stress that the experiments described in this study were not intended to provide any conclusion on the potential toxicological or health impact of polymer-embedded clays or associated tallow-derived QACs that may come into contact with foods. All food contact materials in the U.S. market must be reviewed for safety and approved by the FDA under the provisions of the Federal Food, Drug, and Cosmetic Act.

Experimental Methods

Materials

Polymers and Clay

LDPE (density = 0.925 g mL–1 at 25 °C, melt index = 25 g/10 min, product # 428043, batch # MKBX4360V and MKCB9440) was purchased from Millipore-Sigma. This grade has a molecular weight of ∼ 8 × 104 g mol–1. Poly(ethylene) grafted with maleic anhydride (PE-g-MA) (product # 456632, batch # MKBK3735V and MKBV7760V) was also purchased from Sigma-Aldrich. Cloisite 20 (lot # 5102006) was acquired from BYK Additives and Instruments. As reported by the manufacturer, the nominal water content of Cloisite 20 is <3%, the density is 1.77 g cm–3, and the interplatelet spacing (d001 value) is 3.16 nm. Arquad 2HT-75 (lot# BCBT6127 and BCBC0749V) was purchased from Millipore-Sigma. An analysis of a pure solution of the Arquad 2HT-75 revealed that it is ∼79% QACs by weight based on recovery values, the remainder being water and inorganic material (e.g., chloride counterion and other salts).

Solvents and Other Chemicals

Optima grade acetic acid used for the simulant media and Optima LC/MS grade acetonitrile, acetic acid, water, isopropanol, methanol, and ammonium acetate for MS analysis was purchased from ThermoFisher Scientific, Inc. (Fair Lawn, New Jersey). Dimethylsulfoxide (spectranalyzed grade) and glacial acetic acid (ACS grade) were purchased from ThermoFisher Scientific, Inc. Dimethyldioctadecylammonium bromide (96% purity) was purchased from TCI Chemicals (Portland, OR). Ethanol (99.5% purity, ACS reagent grade) and benzyldimethylhexadecylammonium chloride (BDMHD-Cl, assay 97% minimum, 10% water weight maximum) were purchased from Acros Organics (New Jersey). All water was deionized to 18.2 MΩ cm and dispensed from a Millipore-Sigma MilliQ Direct Q3 water purification system. BDMHD stock solutions (1000 ppb) were made by accurately weighing 61.10 mg of neat BDMHD-Cl in a 50 mL class A volumetric flask and diluting to the mark.

Film Manufacture

A DSM Xplore microcompounder with a volume capacity of 15 mL was used to mix the organically modified MMT (OMMT) with the polymer melt and form free-standing cast films. The microcompounder is a scaled-down version of a conical co-rotating twin-screw extruder fitted with a 65 mm cast film extrusion die. A valve controls the flow of the polymer melt to either a recirculation channel within the mixing chamber or an exit channel.

OMMT/PE-g-MA/LDPE films were prepared using a two-stage masterbatch process. The first stage involved the dispersion of OMMT powder into the grafted polymer compatibilizer (PE-g-MA); the mass ratio of OMMT to PE-g-MA was 1:3. The second stage involved dilution of the masterbatch into neat LDPE. The OMMT/PE-g-MA/LDPE melt was extruded to form CPN films with varying concentrations of OMMT in LDPE with identical OMMT/PE-g-MA mass ratios. Films containing Arquad 2HT-75/PE-g-MA/LDPE, as well as negative control films with PE-g-MA and LDPE only, were prepared directly using a single-stage process. Detailed fabrication procedures for all materials are provided in the Supporting Information section, including tabulated information on amounts of components added (Table S1) and nominal compositions (Table S2).

Film Characterization

The CPNs and other materials were fully characterized prior to conducting migration tests. The wt % of clay and other components was determined by thermogravimetric analysis (TGA). Melting points were determined by differential scanning calorimetry (DSC). Crystallinity values were determined by X-ray diffraction (XRD). Clay dispersion characteristics were further evaluated by scanning transmission electron microscopy (STEM) and Fourier-transform infrared attenuated total reflectance (FTIR-ATR) spectroscopy. A full account of the instrumentation and procedures used in these analyses is provided in the Supporting Information.

Migration Tests

Disks (42 mm diameter) were removed from sample film rolls with a steel punch, hammer, and cutting board. Film thickness was measured at 3 points on each disk with a digital micrometer to the nearest 0.1 μm. The disk weight was measured on an analytical balance to the nearest 0.01 mg.

Migration tests were conducted in four food simulants under time/temperature conditions recommended by FDA for the preparation of premarket submission for food contact substances.[35] A large portion of the tests was performed with 100% ethanol as a fatty food simulant. Ethanol was used in lieu of a purified food oil due to the difficulty of extracting the QACs from food oils for chemical analysis. This substitution is supported by the FDA Guidance and numerous experimental evaluations on the use of ethanol in migration testing.[53,54]

For the migration tests on neat LDPE, 7 wt % OMMT/PE-g-MA/LDPE and 4 wt % Arquad/PE-g-MA/LDPE films, the disks were placed in 125 mL glass jars with PTFE-lined screw-caps. The food simulant (100 mL) was added to the jar, and the caps were screwed-on tightly. The jars containing film disks and solvent were placed on a Thermo Max Q4000 shaker set to 100 RPM shaking and a temperature of 40 °C. At time intervals of 2, 4, 8, 12, 24, 48, 72, 96, 168, 240, 485, and 696 h, the jars were removed from the shaker and placed in a cool water bath for approximately 10 min before sampling. This procedure was used for each of the four food simulants (in a preliminary experiment used for the method development, ethanol-based samples were stored in a circulating laboratory oven at 75 °C; all other kinetic experiments were performed at 40 °C, as outlined above.) Equivalent volumes of the simulants spiked with Arquad 2HT-75 and stored under identical conditions served as quality-control (QC) samples.

For the migration tests of 5 wt % OMMT/PE-g-MA/LDPE into 100% ethanol at different temperatures and OMMT/PE-g-MA/LDPE samples with different wt % of OMMT, the circular film sections were stored in 25 mL glass crimp-top vials. The ethanol volume for these samples was 20 mL. Crimp-top vials were used to reduce evaporation of ethanol during storage at 66 °C. These samples were agitated daily by hand. After storage for the appropriate time period, the samples were cooled for approximately 10 min before decrimping and sampling.

Quantitative measurement of the QAC concentrations in food simulants was carried out by flow-injection mass spectrometry (FI-MS; Acquity-TQD, Waters Corporation) using a published method.[46] The limit of quantitation (LOQ) was 5 μg L–1 in the solution. A brief description of sampling methods and instrument run parameters is provided in the Supporting Information. High-resolution mass spectra of QAC mixtures were acquired on a ThermoFisher Exactive Plus Orbitrap in the positive ion mode using the same mobile phase as the FI-MS analysis of migration samples.