*Article* **Effect of Liquid Absorbent Pads and Packaging Parameters on Drip Loss and Quality of Chicken Breast Fillets**

**Marit Kvalvåg Pettersen 1,\*, Julie Nilsen-Nygaard <sup>1</sup> , Anlaug Ådland Hansen <sup>1</sup> , Mats Carlehög <sup>1</sup> and Kristian Hovde Liland <sup>2</sup>**


**Abstract:** Visible liquid inside food packages is perceived as unattractive to consumers, and may result in food waste—a significant factor that can compromise sustainability in food value chains. However, an absorber with overdimensioned capacity may cause alterations in texture and a dryer product, which in turn may affect consumers' satisfaction and repurchase. In this study we compared the effect of a number of liquid absorbent pads in combination with headspace gas composition (60% CO2/40% N<sup>2</sup> and 75% O2/25% CO<sup>2</sup> ) and gas-to-product volume ratio (g/p) on drip loss and quality of fresh chicken breast fillets. A significant increase in drip loss with an increasing number of liquid absorbent pads was documented. The increase was more pronounced in 60% CO2/40% N<sup>2</sup> compared to 75% O2/25% CO<sup>2</sup> . By comparing packaging variants with a different number of liquid absorbent pads, a higher drip loss for all tested was found at g/p 1.8 compared to g/p 2.9. Total viable counts (TVC) were independent of whether there was free liquid in contact with the product, and TVC was independent of gas composition. Differentiation between the gas compositions was seen for specific bacterial analyses. While significant changes were observed using texture analysis, sensory evaluation of the chicken breast fillets did not show any negative effect in texture related attributes. This study demonstrates the importance of optimized control of meat drip loss, as product-adjusted liquid absorption may affect economy, food quality, and consumer satisfaction, as well as food waste.

**Keywords:** food packaging; drip loss; liquid absorbent pad; chicken breast fillet; texture; sensory evaluation

#### **1. Introduction**

In light of the past several years' focus on sustainability, and the global targets of the United Nations Sustainable Development goals [1] related to responsible production and consumption, the drive towards development of food systems for reducing food loss and food waste has been pronounced. Packaging in general, and especially plastics, have for the last years seen increasing public awareness of the related environmental challenges, specifically related to littering and marine debris [2]. However, one of the main functions of packaging is to protect and preserve the food in the total value chain—from the producer to the consumer. Food packaging is recognized to contribute to food waste reduction and more sustainable food value chains [1,3–5].

Through evaluation of the environmental impact of meat products, measured as greenhouse gas (GHG) emissions [6,7], it has been found that the packaging is only responsible for a small part of the GHG emissions [8]. Considering the small environmental footprint of the packaging compared to that of the meat products, it is clear that optimal packaging systems to avoid food loss in the value chain and food waste at consumers should be high priority, and will contribute to more sustainable food systems.

**Citation:** Pettersen, M.K.; Nilsen-Nygaard, J.; Hansen, A.Å.; Carlehög, M.; Liland, K.H. Effect of Liquid Absorbent Pads and Packaging Parameters on Drip Loss and Quality of Chicken Breast Fillets. *Foods* **2021**, *10*, 1340. https:// doi.org/10.3390/foods10061340

Academic Editors: Valeria Rizzo and Muratore Giuseppe

Received: 3 May 2021 Accepted: 8 June 2021 Published: 10 June 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

Fresh meat products contain high amounts of water. The physiological water content in muscle foods, such as chicken, beef, and pork, is approximately 75% [9]. The waterholding capacity (WHC) of the meat refers to the ability of the meat to retain its natural or added water content during postmortem processing and storage [10,11]. The high water content of meat makes these products particularly prone to microbial spoilage, making them highly perishable. Another important factor tightly linked to the high water content, and thereby also to the perceived quality of these products, is their drip loss [12]. Unavoidably, these products will exude some liquid during storage. Scientific literature in the field mainly focuses on how different aspects of prehandling can be related to excessive WHC and drip loss for red meat (pork, beef, and lamb) [13,14]. Factors like genotype, feeding, slaughtering, and chilling have been evaluated [14], as well as the effect of cutting of muscle fibers. The only packaging-related aspects studied in relation to drip loss were the effect of shrinking or non-shrinking films [13], i.e., there are relatively few studies focusing on the effect of packaging variables [12,15,16]. However, in all cases the comparison has been between MAP (80% O2/20% CO<sup>2</sup> or 30% CO2/70% N2) and vacuum [12,16]. Payne et al. (1998) compared vacuum with CO<sup>2</sup> flushing and/or the use of active packaging as an oxygen scavenger [15]. The studied storage temperatures have also varied a great deal, but no studies included storage at traditional/recommended storage temperatures at retailers in Europe [12,15,16]. Thus, to our knowledge, the effects of different MA gas compositions or gas-to-product volume ratios in relation to drip loss have not been reported for chicken meat.

A liquid absorber is often used in packages with fresh meat and fish to improve the appearance of the product. The capacity of these liquid absorbers is typically chosen by the food producers to ensure absorption of all drip loss, and may not be specifically designed for each product. An absorber with an overdimensioned capacity may cause an unnecessarily high drip loss. This can cause a dryer product and an alteration in texture. Furthermore, visible liquid in the packages can be perceived as unattractive by the consumer [17]. Sensory quality attributes such as juiciness and tenderness of the meat may be reduced, and these are important in terms of how the product is perceived by the consumer at the time of consumption [17].

There is an economic aspect to striving to limit the drip loss of muscle foods. Firstly, free liquid inside packages may be perceived as unattractive by the consumer, and may result in reduced sales [17,18]. In addition to this, a common perception seems to be that excess liquid inside packages can give rise to increased microbial growth and reduced meat quality. However, this has been disproven in a previous publication, where no such relation could be documented [18]. On the contrary, the study showed that the most attractive growth medium for bacteria is the product itself, not its exudate. Finally, the liquid lost implies a reduction in product weight and a reduced product yield for food producers.

Another aspect in the context of drip loss is the fact that the EU regulates the use of absorbers containing superabsorbent polymers (SAP), which are considered to be active packaging devices. For non-sealed absorbers it is mandatory for food producers to use absorbers of adequate capacity in order to ensure that the absorber can absorb all liquid lost from the product. This is to ensure that there is no leakage of SAP that can come into contact with the product (Commission Regulation (EC) No 450/2009 on active and intelligent materials and articles intended to come into contact with food) [19]. Due to this, food producers may be prone to choose an absorber that has an overdimensioned capacity for the product, in order to make sure that there is no free liquid inside the packages.

A high amount of CO<sup>2</sup> inside modified atmosphere packaging (MAP) is often associated with increased drip loss of the product, as CO<sup>2</sup> dissolving into the product causes the WHC to decrease [20–22]. One of the assumed mechanisms at play is the reduction of pH in the presence of CO<sup>2</sup> [23]. Others have reported negative correlation between CO<sup>2</sup> content and drip loss of meat [24]. However, often overlooked is the more pronounced effect of underpressure formation inside the packages at high CO<sup>2</sup> levels. CO<sup>2</sup> absorbed

by the product may cause package deformation and a physical squeeze on the product, resulting in increased drip loss [25,26].

In this study we wanted to systematically investigate which packaging parameters are the most determinant for influencing the drip loss—and consequently, the physicochemical, microbiological, and sensory quality—of chicken breast fillets. The initial experimental setup was designed to address the following research questions: How does the number of liquid absorbent pads affect the drip loss and quality of the meat? How is the drip loss affected by different gas atmosphere compositions in MAP? Chicken breast fillets were chosen as the model product due to their relatively high drip loss and well-known challenges related to rapid microbiological spoilage. The number of liquid absorbent pads and different gas compositions were the variables included in the main experiment. Based on the results from the first experimental setup, a follow-up experiment was designed, aimed at addressing the new emerging research question: How is the drip loss affected in MAP with CO<sup>2</sup> at different gas-to-product volume (g/p) ratios? In this experiment the number of liquid absorbent pads and the gas-to-product (g/p) volume ratio were the main variables.

#### **2. Materials and Methods**

#### *2.1. Sample Preparation and Storage Conditions*

Chicken breast fillets (breast fillet tenderloin; pectoralis major) were obtained from a local producer using a fast and highly automated process with low temperature during the process. The slaughtering was performed in the morning with continuous cooling and cutting. The approximate time for slaughtering and cooling was 3 h, followed by cutting and packaging within less than 30 min. The fillets were wrapped in plastic bags and transported chilled in distribution boxes containing approximately 10 kg each. The packaging was performed at the research institute shortly after reception and within 48 h after slaughtering. The fillets were randomly selected from the distribution boxes. Two fillets with a total average weight of 339 g (339.1 ± 4.4 g) (329.9–350.9 g) were packaged in each tray. The samples were stored in dark conditions at 4 ◦C.

#### *2.2. Packaging Materials*

The chicken breast fillets were packaged in thermoformed trays with a base web consisting of amorphous polyethylene terephthalate/polyethylene (APET/PE) (Multipet 550 µm, Wipak, Nastola, Finland). Biaxer 65 XX HFP AFM consisting of polyethylene terephthalate/polyethylene/ethylene vinyl alcohol/polyethylene (PET/PE/EVOH/PE) (Wipak (Nastola, Finland)) was applied as the top web.

The oxygen transmission rates (OTR) of the materials were, according to the producer: 10 cm3/(m<sup>2</sup> d) at 23 ◦C, 50% RH for the base web, and 5 cm3/(m<sup>2</sup> d) at 23 ◦C, 50% RH for the top web.

The trays were thermoformed using a Multivac R145 thermoforming machine (Multivac, Wolfertschwenden, Germany).

In both experiments, different numbers of liquid absorbent pads were used—0, 1, and 2 (Absorber type 109642, MP-2501 75 × 115 mm black, Færch, Denmark)—and thereby 3 different possibilities of liquid absorption for each gas composition.

#### *2.3. Packaging Methods and Experimental Design*

The studies encompass two experiments with chicken breast fillets. In the first and main experiment (all analyses included) (hereafter referred to as Experiment 1), the gas composition and the number for liquid absorbent pads were the experimental design factors. In this experiment, the chicken breast fillets were stored in a modified atmosphere of 60% CO2/40% N<sup>2</sup> or 25% CO2/75% O2, and a gas-to-product volume (g/p) ratio of 1.8 was applied for all samples. Four replicates of each sample type were prepared for each sampling time, performed after 0, 6, 14, and 20 days of storage.

In the second experiment (Experiment 2), the effect of the g/p ratio and the number of liquid absorbent pads was investigated. The chicken breast fillets were stored in 60% CO2/40% N2, and two different tray sizes were used. Trays with a volume of 860 mL resulted in an initial gas-to product-volume ratio (g/p ratio) of approximately 1.8, while trays with a volume of 1390 mL gave an initial g/p ratio of approximately 2.9. Four replicates of each sample type were prepared for each sampling time, performed after 0, 7, 14, and 20 days of storage.

#### *2.4. Analyses*

#### 2.4.1. Headspace Gas Analyses and Drip Loss

The headspace atmosphere of the MA packages was analyzed for CO<sup>2</sup> and O<sup>2</sup> levels (%) immediately after packaging and at each sampling time using a CheckMate 9900 O2/CO<sup>2</sup> analyzer (PBI Dansensor, Ringsted, Denmark). Gas was removed from the packaging for analysis using a needle through self-sealing patches on the packages.

Drip loss was determined by initially weighing the meat, the package, and the absorbent pad(s), and calculating the increase in weight of the packages (including the absorbent pads) at each sampling. Results are given as the percentage (%) of initial muscle weight, and refer to the corresponding drip loss from the meat. These analyses were performed in all experiments.

#### 2.4.2. Texture Analyses and Dry Matter Content

Warner–Bratzler shear force (WBSF) and dry matter content measurements were performed (in Experiment 1) for chicken breast fillets cooked after storage in different packaging conditions after 0, 6, 14, and 21 days of storage. The fillets were vacuum packed in PA/PE (70 µm) (Maskegruppen, Norway) bags and heat treated in a water bath at 70 ◦C for 50 min before being cooled in ice water for 50 min. The samples were stored in the vacuum bags at 4 ◦C until the next day. Prior to WBSF measurements, the temperature of the samples was equilibrated at 20 ◦C for 1 hour. The fillets were cut into rectangular pieces of 1 × 1 × 2 cm along the fiber direction. The samples were sheared perpendicularly to the fiber direction with a triangular device attached to an Instron Materials Testing Machine (model 4202, Instron Engineering Co., High Wycombe, UK). The average maximum force (given as N/cm<sup>2</sup> ) was obtained from measurement of 6 replicates.

For determination of dry matter content, the samples were macerated/homogenized, and approximately 6 ± 0.5 g of the mass was accurately weighed into Petri dishes and oven dried at 105 ◦C for 18 h. The samples were weighed after drying, and the weight loss during drying was equal to the water content of the samples. The dry matter content was calculated as the percentage of the initial weight minus the water content (Dry matter content (%) = wsample (%) − wwater (%)). Two replicates per sample variant were measured at each sampling time.

#### 2.4.3. Microbiological Analyses

The selected microbiological analyses for chicken breast fillets (Experiments 1 and 2) were total viable count (TVC), *Enterobacteriaceae,* lactic acid bacteria (LAB), and *Brochothrix thermosphacta*, performed at the time of packaging and after the selected sampling time.

Samples of 3 <sup>×</sup> 3 cm<sup>2</sup> and 1 cm depth were cut with a sterile scalpel from the surface of the meat, weighed, macerated, and diluted 1:10 with peptone water and spread using a Whitley Automated Spiral Plater (WASP) (Don Whitley Scientific Ltd., West Yorkshire, UK). In addition, in Experiment 1, after 14 days of storage, samples from the bottom surfaces of the fillets were cut and included for analyses. Appropriate 10-fold dilutions were spread in duplicate on PCA (plate count agar; Difco, Difco Laboratories, Detroit, MI, USA) for total viable counts (TVC) (incubation temperature 30 ◦C, 72 h, anaerobic incubation), and on MRS agar (Man, Sharpe and Rogosa agar, MRS; Oxoid, Unipath Ltd., Basingstoke, Hampshire, UK) for lactic acid bacteria (incubated at 20 ◦C, 48 h, anaerobic incubation). *Enterobacteriaceae* were analyzed by use of VRBGA (Violet Red Bile Glucose Agar, Oxoid,

Hampshire, UK) (37 ◦C, 24 h, semi-aerobic conditions, cells embedded in agar with sterile overlay). *Brochothrix thermosphacta* was detected by use of STAA agar (streptomycin thallous acetate actidione) and an agar base (CM 0881 with selective supplement SR 0151E, Oxoid, Hampshire, UK) (25 ◦C for 48 h, aerobic incubation). Microbial counts are expressed as colony-forming units (cfu) per g.

#### 2.4.4. Sensory Analysis

Sensory analysis was performed on both raw and heat-treated chicken breast fillets after 14 days of storage in both modified atmospheres—25% CO2/75% O2, and 60% CO2/40% O2—with 0, 1, and 2 liquid absorbent pads (g/p 1.8) (Experiment 1). A highly trained panel of 10 assessors (10 women; aged, 37–64 years) at Nofima (Ås, Norway) performed a sensory descriptive analysis (DA) according to the "generic descriptive analysis" [27] and ISO standard 13299 [28]. The assessors are regularly tested and trained according to ISO standard 8586, and the sensory laboratory follows the practice of ISO standard 8589 [29,30].

In a pretest session before the main test, the assessors were calibrated on samples that were considered the most different on the selected attributes typical for raw and heattreated chicken fillets. The results from the pretest were evaluated and discussed by the panel leader and the assessors. This calibration procedure was performed in order to arrive at a common understanding and agreement of the selected attributes. This is common practice, with the purpose being to ensure that the assessors have a common understanding of how to evaluate and rank the different sensory attributes, and to obtain consensus for each attribute among the assessors. For raw evaluation the assessors agreed upon six sensory attributes describing odor: sourness odor, metallic odor, cloying odor, sulfurous odor, fermented odor, and chlorine odor. For heat-treated evaluation, the assessors agreed upon eight sensory attributes describing flavor and texture: sourness flavor, metallic flavor, cloying flavor, sulfurous flavor, fermented flavor, hardness, juiciness, and tenderness.

The sensory evaluation of raw samples was performed on chicken fillets stored in the original packaging. The samples were first heated to room temperature. Immediately before evaluation, an opening (4 × 4 cm) was cut into the top web of the packages, and the assessors smelled the sample trays.

For heat-treated sensory evaluation, each assessor was served one piece of chicken in a triangular shape and served from the same position of the fillet throughout the whole test. Heat treatment of chicken fillets was performed in a combi oven (Electrolux Air-o-steam, Model AOS061EANQ) at +100 ◦C with 100% heat for 20 min (core temperature of 72 ◦C). Samples were served in preheated porcelain bowls with warm metal lids and placed on a heating plate in the sensory booth. All samples were served to the panel coded with a three-digit number in duplicate following a randomized block design.

All attributes were evaluated on an unstructured 15 cm line scale with labeled end points ranging from "no intensity" (1) to "high intensity" (9). Each assessor evaluated all samples at individual speed on a computer system for direct recording of data (EyeQuestion, Software Logic8 BV, Utrecht, The Netherlands). Tap water and unsalted crackers were available for palate cleansing.

#### 2.4.5. Statistical Analyses

The experiments were prepared using balanced experimental designs for easy analysis of the packaging choices under investigation. Subsequent data analysis was performed using analysis of variance (ANOVA) with type II sums of squares and proportions of explained variance. Tukey's pairwise comparisons were used to generate compact letter displays (CLDs), indicating which factor levels—e.g., different MAPs or number of absorbent pads—were not significantly different. All tests were performed using a level of significance of 0.05. The software used in the analyses was R version 4.0.4 [31] and the R package "mixlm" version 1.2.4 [32].

For sensory performance, ANOVA was conducted on the descriptive sensory data in order to identify the sensory attributes that discriminated among samples. A two-way mixed model was fitted for each of the sensory attributes, with the assessor and interaction effects considered to be random and the samples as a fixed effect. Least significance differences were calculated using Tukey's test (*p* < 0.05). The statistical software used for the sensory analysis was EyeOpenR (Logic8 BV). mixed model was fitted for each of the sensory attributes, with the assessor and interaction effects considered to be random and the samples as a fixed effect. Least significance differences were calculated using Tukey's test (*p* < 0.05). The statistical software used for the sensory analysis was EyeOpenR (Logic8 BV).

displays (CLDs), indicating which factor levels—e.g., different MAPs or number of absorbent pads—were not significantly different. All tests were performed using a level of significance of 0.05. The software used in the analyses was R version 4.0.4 [31] and the R

For sensory performance, ANOVA was conducted on the descriptive sensory data in order to identify the sensory attributes that discriminated among samples. A two-way

#### **3. Results and Discussion 3. Results and Discussion**

this.

package "mixlm" version 1.2.4 [32].

#### *3.1. Effect of Liquid Absorbent Pads and Gas Composition 3.1. Effect of Liquid Absorbent Pads and Gas Composition*

*Foods* **2021**, *10*, x FOR PEER REVIEW 6 of 18

The percentage drip loss was measured for chicken breast fillets packaged in modified atmospheres of 60% CO2/40% N<sup>2</sup> and 75% O2/25% CO<sup>2</sup> with 0, 1, and 2 liquid absorbent pads (Experiment 1) at selected sampling times during storage: 6 days, 14 days, and 20 days. The results presented in Figure 1 show that there is a clear positive relation regarding the number of liquid absorbent pads (i.e., increased absorbing surface) present in the packages. This implies that when increasing the number of the absorbent pads, the liquid lost from the product will increase as the absorbers draw excess liquid from the product. The tendency is evident for both packaging atmospheres (Figure 1). The measurements show a small increase in drip loss over the time of storage that is documented through these sampling times (6, 14, and 20 days) for all six sample variants. However, the most pronounced differences among the variants are already present at the first sampling time (day 6), revealing that the main part of the drip loss actually occurs during the initial days of storage. The relationship between the sample variants does not change after this. The percentage drip loss was measured for chicken breast fillets packaged in modified atmospheres of 60% CO2/40% N2 and 75% O2/25% CO2 with 0, 1, and 2 liquid absorbent pads (Experiment 1) at selected sampling times during storage: 6 days, 14 days, and 20 days. The results presented in Figure 1 show that there is a clear positive relation regarding the number of liquid absorbent pads (i.e., increased absorbing surface) present in the packages. This implies that when increasing the number of the absorbent pads, the liquid lost from the product will increase as the absorbers draw excess liquid from the product. The tendency is evident for both packaging atmospheres (Figure 1). The measurements show a small increase in drip loss over the time of storage that is documented through these sampling times (6, 14, and 20 days) for all six sample variants. However, the most pronounced differences among the variants are already present at the first sampling time (day 6), revealing that the main part of the drip loss actually occurs during the initial days of storage. The relationship between the sample variants does not change after

**Figure 1.** Drip loss for chicken breast fillets expressed as a percentage (%) of initial product weight as a function of storage time for chicken breast fillets (C) packaged in modified atmospheres of 60% CO2/40% N2 (C-CO2) and 75% O2/25% CO2 (C-O2) with 0, 1, or 2 liquid absorbent pads (0A, 1A, 2A). g/p ratio for all samples was 1.8 (Experiment 1). Sampling times were after 6, 14, and 20 days of cold storage. For each bar, the error bars indicate +/− one standard error. **Figure 1.** Drip loss for chicken breast fillets expressed as a percentage (%) of initial product weight as a function of storage time for chicken breast fillets (C) packaged in modified atmospheres of 60% CO2/40% N<sup>2</sup> (C-CO<sup>2</sup> ) and 75% O2/25% CO<sup>2</sup> (C-O<sup>2</sup> ) with 0, 1, or 2 liquid absorbent pads (0A, 1A, 2A). g/p ratio for all samples was 1.8 (Experiment 1). Sampling times were after 6, 14, and 20 days of cold storage. For each bar, the error bars indicate +/− one standard error.

The gas composition and storage time had significant effects on drip loss (1.8% and 3%, respectively); however, the number of absorbent pads had the most effect (88.7%) (Appendix A Table A1). Furthermore, the measured drip loss was lower for the samples packaged in 75% O2/25% CO2 than for those packaged in 60% CO2/40% N2, though not The gas composition and storage time had significant effects on drip loss (1.8% and 3%, respectively); however, the number of absorbent pads had the most effect (88.7%) (Appendix A Table A1). Furthermore, the measured drip loss was lower for the samples packaged in 75% O2/25% CO<sup>2</sup> than for those packaged in 60% CO2/40% N2, though not different enough to be significant. These effects were present at all sampling times (though more pronounced towards the end of storage) and for different numbers of absorbent pads. This can be explained by the solubility of CO<sup>2</sup> into the product, causing underpressure formation and a physical pressure on the product, resulting in increased drip loss [25,26]. The magnitude of the underpressure formed is proportional to the amount (percentage) of CO<sup>2</sup> present in the package, while being disproportional to the g/p ratio of the packaging

concept (the available volume of the gas in relation to the volume of the product). On the other hand, increasing the CO<sup>2</sup> level may be beneficial in terms of improved microbiological quality—an aspect that will be considered in the following section.

In accordance with the observed increase in drip loss during storage, the dry matter content of the chicken breast fillets also increased during the initial part of storage (from day 0 to day 6), to varying extents, for all sample variants (Figure 2A). For fillets stored in a CO2-rich atmosphere, the dry matter content of the chicken meat increased until 14 days of storage, followed by a decrease in dry matter content. For samples stored in an O2-rich atmosphere the effect of storage time on the dry matter content was more ambiguous, and no clear correlation could be deducted from the results. *Foods* **2021**, *10*, x FOR PEER REVIEW 8 of 18

**Figure 2.** Percentage dry matter content measured for chicken breast fillets (**A**) (two replicates per sample variant) and Warner–Bratzler maximum shear force (N/cm2) (WBSF) required to cut cooked samples of chicken breast fillets (**B**) (presented as an average of 6 replicates per sample variant). The chicken breast fillets (C) were stored in modified atmospheres of 60% CO2/40% N2 (C-CO2) and 75% O2/25% CO2 (C-O2) (g/p 1.8) with 0, 1, or 2 liquid absorbent pads (0A, 1A, 2A) (Experiment 1) and sampling times of 0, 6, 14, and 20 days. **Figure 2.** Percentage dry matter content measured for chicken breast fillets (**A**) (two replicates per sample variant) and Warner–Bratzler maximum shear force (N/cm<sup>2</sup> ) (WBSF) required to cut cooked samples of chicken breast fillets (**B**) (presented as an average of 6 replicates per sample variant). The chicken breast fillets (C) were stored in modified atmospheres of 60% CO2/40% N<sup>2</sup> (C-CO<sup>2</sup> ) and 75% O2/25% CO<sup>2</sup> (C-O<sup>2</sup> ) (g/p 1.8) with 0, 1, or 2 liquid absorbent pads (0A, 1A, 2A) (Experiment 1) and sampling times of 0, 6, 14, and 20 days.

> For storage in high O2-atmospheres, an initial reduction in maximum force was measured for all three numbers of absorbent pads (0, 1, or 2 absorbent pads) added. Perhaps most interesting in these results is the increase in measured maximum force between day 6 and day 14 of storage, indicating a decrease in tenderness of the fillets cooked after this time of storage, seen only for high-O2 samples. The measured shear force dropped at the last sampling day (day 20) of storage regardless of the number of absorbent pads. For fillets stored zero (0A) or with one absorbent pad (1A), the texture evaluated by WBSF was measured to be higher (17.2 N/cm2 and 17.5 N/cm2) at the end of storage compared to the initial level (15.6 N/cm2), indicating a decrease in meat tenderness towards the end of In general, for all samples there was a tendency for the dry matter content to increase from the time of packaging to the end of storage, but the dry matter content varied among the sample variants and by the time of storage. Regarding the number of liquid absorbent pads, no clear correlation between an increased number of pads and dry matter content could be documented—neither for the samples in CO2-rich atmospheres, nor for those in O2-rich atmospheres. Correlation analyses showed that the correlation between dry matter and drip loss was −0.166 in CO2-rich atmospheres, while no correlation (−0.001) was observed in O2-rich atmospheres. Furthermore, the number of pads had no significant effect as a main factor (Appendix A Table A1).

> storage. For chicken stored with two absorbent pads (2A), the measured WBSF was reduced at the end of storage (14.5 N/cm2) compared to the start (15.6 N/cm2). However, according to one-way ANOVA, no significant differences between the samples stored in high-O2 atmospheres were detected at any sampling time. The only significant differences were observed between samples stored in different atmospheres after 20 days of storage (Appendix A Table A2). The physicochemical origin of the observed differences between samples stored in high-CO2 and high-O2 atmospheres is unknown. For differences in tenderness between chicken fillets to be of importance for the perceived quality, the numeric differences in measured shear force would assumedly need to be a great deal larger. Again, even though the fillet selection was randomized at packaging, some of the measured variations may be a result of individual variations. The results from sensory evaluation of the textural traits hardness, tenderness, and The effects of the number of liquid absorbent pads and of gas composition were also evaluated through texture analysis using the Warner–Bratzler method. As displayed in Figure 2B, for fillets stored in CO2-rich atmospheres the measured WBSF shows a net decrease from the time of packaging (15.6 N/cm<sup>2</sup> ) to the end of storage for samples stored without absorbent pads (0A) (13.4 N/cm<sup>2</sup> ) and for those stored with two absorbent pads (2A) (14.5 N/cm<sup>2</sup> ). For fillets stored with one absorbent pad (1A) the measured WBSF was practically unchanged throughout storage. In general, the measured differences between 0, 1, and 2 absorbent pads for this gas composition were very small. In addition, according to one-way ANOVA—computed by defining a six-level factor with the combinations of gas composition and number of liquid absorbent pads—there were no significant differences between these samples at any storage time (Appendix A Table A2). Although all main factors (gas composition, number of absorbent pads, and storage time) were

> pads (flavor evaluation of heat-treated chicken breast fillets and odor evaluation of raw chicken are presented in Appendix A Table A3, and will be discussed in relation to micro-

bial growth).

significant (Appendix A Table A1), the variance explained by these factors was relatively low (7.2%, 6.6%, and 13.2%, respectively), leaving 59% as unexplained and a corresponding R <sup>2</sup> of 41%, i.e., other sources of variation dominated the WBSF.

For storage in high O2-atmospheres, an initial reduction in maximum force was measured for all three numbers of absorbent pads (0, 1, or 2 absorbent pads) added. Perhaps most interesting in these results is the increase in measured maximum force between day 6 and day 14 of storage, indicating a decrease in tenderness of the fillets cooked after this time of storage, seen only for high-O<sup>2</sup> samples. The measured shear force dropped at the last sampling day (day 20) of storage regardless of the number of absorbent pads. For fillets stored zero (0A) or with one absorbent pad (1A), the texture evaluated by WBSF was measured to be higher (17.2 N/cm<sup>2</sup> and 17.5 N/cm<sup>2</sup> ) at the end of storage compared to the initial level (15.6 N/cm<sup>2</sup> ), indicating a decrease in meat tenderness towards the end of storage. For chicken stored with two absorbent pads (2A), the measured WBSF was reduced at the end of storage (14.5 N/cm<sup>2</sup> ) compared to the start (15.6 N/cm<sup>2</sup> ). However, according to one-way ANOVA, no significant differences between the samples stored in high-O<sup>2</sup> atmospheres were detected at any sampling time. The only significant differences were observed between samples stored in different atmospheres after 20 days of storage (Appendix A Table A2). The physicochemical origin of the observed differences between samples stored in high-CO<sup>2</sup> and high-O<sup>2</sup> atmospheres is unknown. For differences in tenderness between chicken fillets to be of importance for the perceived quality, the numeric differences in measured shear force would assumedly need to be a great deal larger. Again, even though the fillet selection was randomized at packaging, some of the measured variations may be a result of individual variations.

The results from sensory evaluation of the textural traits hardness, tenderness, and juiciness of heat-treated chicken breast fillets (Figure 3) reveal that there are significant differences in the perceived texture of the meat as affected by the number of absorbent pads (flavor evaluation of heat-treated chicken breast fillets and odor evaluation of raw chicken are presented in Appendix A Table A3, and will be discussed in relation to microbial growth). *Foods* **2021**, *10*, x FOR PEER REVIEW 9 of 18

**Figure 3.** Sensory intensity scores (scale 1–9) for the textural attributes hardness, tenderness, and juiciness of heat-treated chicken breast fillets (C) stored for 14 days in modified atmospheres of 60% CO2/40% N2 (C-CO2) and 75% O2/25% CO2 (C-O2) with 0, 1, or 2 liquid absorbent pads (0A, 1A, 2A). g/p ratio for all samples was 1.8 (Experiment 1). **Figure 3.** Sensory intensity scores (scale 1–9) for the textural attributes hardness, tenderness, and juiciness of heat-treated chicken breast fillets (C) stored for 14 days in modified atmospheres of 60% CO2/40% N<sup>2</sup> (C-CO<sup>2</sup> ) and 75% O2/25% CO<sup>2</sup> (C-O<sup>2</sup> ) with 0, 1, or 2 liquid absorbent pads (0A, 1A, 2A). g/p ratio for all samples was 1.8 (Experiment 1).

However, the trend is not systematic across gas compositions, as it does not increase/decrease with the increasing number of pads independent of the gas composition, i.e., the only significant difference is found in the interaction between the number of ab-However, the trend is not systematic across gas compositions, as it does not increase/decrease with the increasing number of pads independent of the gas composition, i.e., the only significant difference is found in the interaction between the number of ab-

stored in high-O2 atmospheres, while regarding the juiciness of chicken stored in high-CO2 atmospheres, a lower score for samples stored without absorbent pads (0A) compared to one absorbent pad (1A) was observed. In this experiment, tenderness and juiciness both correlated positively with dry matter content (0.231 and 0.210, respectively) and negatively with WBSF (−0.517 and −0.197, respectively), while hardness did not correlate with dry matter (−0.040) but correlated positively with WBSF (0.397). However, although there were significant differences in sensory scores among different numbers of absorbent pads, the differences were less than 2 in score (hardness 3.75 and 4.45; juiciness 4.37 and 5.40), i.e., less than what is presumed observable by consumers. In addition, no significant differences in dry content or texture measured as maximum force (WBSF) (neither number of absorbent pads nor gas composition) were observed for the samples stored for 14 days. This implies that for this product consumers will not be able to identify differences in the textural quality of fillets packaged with liquid absorbent pads overdimensioned for the

With regard to the effect of modified atmosphere composition on the textural attributes, the samples stored in high-O2 atmospheres were evaluated with slightly higher sensory intensity scores for hardness than the ones stored in high-CO2 atmospheres. This is supported by the fact that the sample variants stored in high-O2 atmospheres were evaluated to have somewhat lower intensity scores on tenderness than the sample variants stored in high-CO2 atmospheres. For juiciness—an important quality trait for chicken breast fillets—no clear tendency can be observed when comparing the sample variants stored in high-O2 compared to high-CO2 modified atmospheres. However, the gas composition significantly affected the texture in samples with one absorbent pad (1A), as shown by lower scores of tenderness and juiciness and higher scores of hardness in the samples stored in high-O2 atmospheres compared to high-CO2 atmospheres. This finding is in accordance with the results of the Warner–Bratzler shear force measurements, which, as discussed, displayed the largest relative differences between the two gas compositions at measurement after 14 days of storage—the same storage time at which sensory

product's drip loss.

sorbent pads and the packaging gas. A significantly higher score of hardness with one absorbent pad (1A) compared to two absorbent pads (2A) was observed for samples stored in high-O<sup>2</sup> atmospheres, while regarding the juiciness of chicken stored in high-CO<sup>2</sup> atmospheres, a lower score for samples stored without absorbent pads (0A) compared to one absorbent pad (1A) was observed. In this experiment, tenderness and juiciness both correlated positively with dry matter content (0.231 and 0.210, respectively) and negatively with WBSF (−0.517 and −0.197, respectively), while hardness did not correlate with dry matter (−0.040) but correlated positively with WBSF (0.397). However, although there were significant differences in sensory scores among different numbers of absorbent pads, the differences were less than 2 in score (hardness 3.75 and 4.45; juiciness 4.37 and 5.40), i.e., less than what is presumed observable by consumers. In addition, no significant differences in dry content or texture measured as maximum force (WBSF) (neither number of absorbent pads nor gas composition) were observed for the samples stored for 14 days. This implies that for this product consumers will not be able to identify differences in the textural quality of fillets packaged with liquid absorbent pads overdimensioned for the product's drip loss.

With regard to the effect of modified atmosphere composition on the textural attributes, the samples stored in high-O<sup>2</sup> atmospheres were evaluated with slightly higher sensory intensity scores for hardness than the ones stored in high-CO<sup>2</sup> atmospheres. This is supported by the fact that the sample variants stored in high-O<sup>2</sup> atmospheres were evaluated to have somewhat lower intensity scores on tenderness than the sample variants stored in high-CO<sup>2</sup> atmospheres. For juiciness—an important quality trait for chicken breast fillets—no clear tendency can be observed when comparing the sample variants stored in high-O<sup>2</sup> compared to high-CO<sup>2</sup> modified atmospheres. However, the gas composition significantly affected the texture in samples with one absorbent pad (1A), as shown by lower scores of tenderness and juiciness and higher scores of hardness in the samples stored in high-O<sup>2</sup> atmospheres compared to high-CO<sup>2</sup> atmospheres. This finding is in accordance with the results of the Warner–Bratzler shear force measurements, which, as discussed, displayed the largest relative differences between the two gas compositions at measurement after 14 days of storage—the same storage time at which sensory evaluation was performed. Still, the differences in sensory scores are small and will most likely not be detectable by the average consumer. Geesink et al. (2015) stated that the effect of high oxygen on tenderness of meat has been reported in a number of studies [33]. To the best of our knowledge, reported effects of packaging atmosphere on textural attributes of chicken/poultry meat are rather scarce. Rossaint et al. (2015) reported no significant difference in texture (measured as sensory attribute) for poultry stored in a high-oxygen atmosphere (70% O2/30% CO2) compared to 70% N2/30% CO<sup>2</sup> nitrogen [34].

Regarding texture measured as maximum force (WBSF), a majority of publications have been on beef and pork meat, and often conducted comparisons of high-oxygen atmospheres with vacuum (oxygen- and CO2-free atmosphere) [35–38]. Lagersted et al. (2011) reported that storage in MAP with high oxygen resulted in higher shear force and negatively affected the juiciness and tenderness of beef steaks compared to vacuum [37]. Moczkowska et al. (2017) reported a decrease in shear force (WBSF) for beef stored in an oxygen-free atmosphere (vacuum), while increased WBSF was observed when stored in a high-oxygen atmosphere [36]. Similarly, Zakrys-Waliwander et al. (2012) detected significantly lower WBSF in beef steaks stored in vacuum compared to a high-O<sup>2</sup> atmosphere after 8 and 14 days, but not after 4 days of storage [38]. For pork meat (porcine longissimus dorsi) a decrease in tenderness was detected already after 4 days of storage in a high-oxygen atmosphere, with further decrease until 14 days of storage, compared to an increase in tenderness when stored in an oxygen-free atmosphere (vacuum) [39]. In oxygen-rich atmospheres the potential for increased oxidation is present, and oxidation of protein can influence properties such as water-holding capacity, and lead to changes in texture, such as tenderness [36,39].

The TVC at the start of storage (day 0) was approximately 2.6 log cfu/g for all sample variants (Appendix A Table A4). In general, the increase in TVC developed in a parallel manner for sample variants in high-O<sup>2</sup> and high-CO<sup>2</sup> atmospheres, with a slightly lower bacterial growth on the samples in high-CO<sup>2</sup> atmospheres, reaching 6.7–7.1 log cfu/g after 20 days of storage for the high-CO<sup>2</sup> variants, and 7.3–7.5 log cfu/g for the high-O<sup>2</sup> atmosphere samples. Overall, there was a small but significant effect of gas composition on TVC (Appendix A Table A4). Previous results show similar total viable count numbers for products packaged with and without high levels of oxygen (though with low CO2-levels; 30% CO2/70% N<sup>2</sup> and 30% CO2/70% O2) [40], whereas in the present study high levels of CO<sup>2</sup> resulted in lower TVCs compared to high-O<sup>2</sup> atmospheres. Statistical analysis also confirmed that gas composition had a significant effect on the TVC (although only 1.7% explained variance), while there was no significant effect of different numbers of absorbent pads on TVC. Hence, the microbiological growth is independent of whether or not there is free liquid in the packages.

This finding was further supported by results from additional analysis performed at day 14 of storage by sampling from the bottom surfaces of the fillets that were in contact with available visible liquid and comparing the TVCs measured in these samples to the TVCs from the top surfaces of the fillets at the same sampling day (Appendix A Table A4). The TVC data for the top and bottom samples for all six packaging variants are summarized in Appendix A Table A4. For samples stored in high-CO<sup>2</sup> atmospheres, a significantly lower bacterial level was measured for samples taken from the bottom surface (and in contact with visible liquid if present) compared to samples from the top surface of the fillets (with two absorbent pads (2A)). However, no such differences in bacterial growth for samples stored in high-O<sup>2</sup> atmospheres with different numbers of absorbent pads could be detected. Hence, free liquid inside packages does not give rise to increased microbiological activity. Even in the packaging variants without an absorbent pad present (0A), the microbiological growth at the bottom and top of the fillets was similar—not significantly different—for the same packaging conditions, as well as similar to the measured TVC levels for packaging with the highest number of absorbent pads (no free liquid inside the packages). This is a significant finding, as it disproves the common perception that packaged meat with excess liquid inside the packages has a poorer microbial quality [18]. Dissemination of these results to the food industry, retailers, and consumers could contribute to reducing food waste in this product category.

Regardless of the number of absorbent pads added, growth of *Brochothrix thermosphacta* reaches a level of approximately 7 log cfu/g for fillets in high-O<sup>2</sup> atmospheres and 3–4 log cfu/g in high-CO<sup>2</sup> atmospheres after 20 days of storage (Appendix A Table A4). This is as expected for this bacterium; *Brochothrix* grows fast in the presence of O2, and it is also able to adapt to an anaerobic environment, though then at a much slower growth rate, which is in accordance with previous studies [34,41].

Lactic acid bacteria increased from 2.45 log cfu/g to approximately 7 log cfu/g (7.12–7.31 log cfu/g and 6.67–7.22 log cfu/g in high-CO<sup>2</sup> and high-O<sup>2</sup> atmospheres, respectively) over a 20-day storage period. Levels of *Enterobacteriaceae* were measured to be about 2 log cfu/g until 15 days of storage, followed by an increase to approximately 3 log cfu/g in high-O<sup>2</sup> atmospheres (2.23–3.17 log cfu/g), and slightly higher approximately 4 log cfu/g—in high-CO<sup>2</sup> atmospheres (4.02–4.27 log cfu/g) after 20 days of storage. According to analyses of variance, the number of liquid absorbent pads had no significant effect on the levels of lactic acid bacteria and *Enterobacteriaceae.* No significant differences were observed at any sampling time within each packaging gas (Appendix A Table A4), and the variance was mainly explained by the storage time (Appendix A Table A1). Moreover, no significant effect of the number of absorbent pads in the packages on the growth of *B. thermosphacta* could be seen in these results, but in this case most of the variance was explained by gas composition and storage time (45.4% and 42%, respectively).

Sensory evaluation of the odor of the raw chicken showed similar trends as the bacterial growth, with no significant effect of different numbers of liquid absorbent pads but clear effect of the gas composition (Appendix A Table A3). In all evaluated attributes, significant differences were detected, with slightly higher scores for sourness and metallic odors and slightly lower scores for chlorine odor in chicken stored in high-CO<sup>2</sup> atmospheres compared to high-O<sup>2</sup> atmospheres. However, for the other attributes, the differences were more pronounced. Chicken stored in high-O<sup>2</sup> atmospheres had high scores (6.88–7.28) for fermented and cloying (6.48–6.89) odors compared to storage in high-CO<sup>2</sup> atmospheres (2.14–2.87 and 3.37–3.94, respectively). The opposite was the case regarding the sulfurous odor: 1.97–2.07 in high-O<sup>2</sup> atmospheres and 5.38–6.23 in high-CO<sup>2</sup> atmospheres. For heattreated chicken the only significant flavor attributes were sourness and fermented flavors, though the scores were relatively low (below 3.37) and the difference was less than 2 units.

#### *3.2. Effect of Liquid Absorbent Pads and G/P Ratio*

The g/p ratio applied in the first experiment in the study (g/p 1.8) is comparable to what is commonly used for modified atmosphere packaging on the market in Norway. Based on the findings of this first and main part of the study, it was relevant to look into the effect of varying the g/p ratio, i.e., the amount of carbon dioxide gas available to the product, and how it affects the drip loss and microbiological quality of the chicken breast fillets when packaged with 0, 1, or 2 liquid absorbent pads. Therefore, the second part of the study includes both g/p 1.8 (as for Experiment 1) and a higher g/p ratio of 2.9.

Carbon dioxide has a well-documented antimicrobial effect [13,42–44]. According to Renerre et al. (1990), Dalgaard et al. (1993), and Randell et al. (1999), the drip loss increases with increased CO<sup>2</sup> content in the packages [20–22]. However, Holck et al. (2014) showed that drip loss is not solely dependent on the amount of CO2, with higher drip loss in chicken packed in 100% CO<sup>2</sup> compared to 100% CO<sup>2</sup> with the addition of a CO<sup>2</sup> emitter [25]. Thus, underpressure formation with high amounts of CO<sup>2</sup> in rigid packages may result in an excessive drip loss from the product, due to physical pressure as the package is compressed [26]. Figure 4 displays the drip loss results from the second experiment on chicken breast fillets. Firstly, this follow-up experiment confirms the results obtained in the initial experiment: the drip loss increases when increasing the number of liquid absorbent pads. After 20 days of storage, two absorbent pads (2A) resulted in significantly higher drip loss compared to without absorbent pads (0A). However, after 7 and 14 days of storage, significantly higher drip loss was detected in packages with one absorbent pad (1A) compared to those without absorbent pads (0A) at g/p 1.8 (Appendix A Table A2). Secondly, the figure shows that there is a general tendency towards higher drip loss for samples in modified atmospheres at g/p 1.8 compared to g/p 2.9. By statistical analysis, this effect was found to be significant, with higher drip loss at g/p 1.8 compared to g/p 2.9 with one absorbent pad (1A) after both 7 and 14 days of storage. Analysis of variance shows that all main factors (g/p ratio, number of absorbent pads and storage time) were significant (Appendix A Table A1), with explained variances of 16.7%, 26.1%, and 31.6%, respectively. At a higher g/p ratio (and identical gas composition) there is a larger amount of CO<sup>2</sup> present in the package and, thus, a higher amount of CO<sup>2</sup> that can be dissolved into the product. This implies that when CO<sup>2</sup> is absorbed by the product the underpressure formed will be relatively small in the packages with g/p 2.9 compared to in the packages with g/p 1.8. The samples with g/p 2.9 consequently have a lower drip loss due to a less pronounced physical pressure on the chicken breast fillets. The variance in the measurements reflects the natural individual variations between the products.

higher drip loss compared to without absorbent pads (0A). However, after 7 and 14 days of storage, significantly higher drip loss was detected in packages with one absorbent pad (1A) compared to those without absorbent pads (0A) at g/p 1.8 (Appendix A Table A2). Secondly, the figure shows that there is a general tendency towards higher drip loss for samples in modified atmospheres at g/p 1.8 compared to g/p 2.9. By statistical analysis, this effect was found to be significant, with higher drip loss at g/p 1.8 compared to g/p 2.9 with one absorbent pad (1A) after both 7 and 14 days of storage. Analysis of variance shows that all main factors (g/p ratio, number of absorbent pads and storage time) were significant (Appendix A Table A1), with explained variances of 16.7%, 26.1%, and 31.6%, respectively. At a higher g/p ratio (and identical gas composition) there is a larger amount of CO2 present in the package and, thus, a higher amount of CO2 that can be dissolved into the product. This implies that when CO2 is absorbed by the product the underpressure formed will be relatively small in the packages with g/p 2.9 compared to in the packages with g/p 1.8. The samples with g/p 2.9 consequently have a lower drip loss due to a less pronounced physical pressure on the chicken breast fillets. The variance in the measurements reflects the natural individual variations between the products.

**Figure 4.** Drip loss for chicken breast fillets expressed as a percentage (%) of initial product weight as a function of storage time for chicken breast fillets (C) packaged in modified atmospheres of 60% CO2/40% N2 (C-CO2) with 0, 1, or 2 liquid absorbent pads (0A, 1A, 2A) and two different g/p ratios (1.8 and 2.9) (Experiment 2). Sampling times were after 7, 14, and 20 days of cold storage. **Figure 4.** Drip loss for chicken breast fillets expressed as a percentage (%) of initial product weight as a function of storage time for chicken breast fillets (C) packaged in modified atmospheres of 60% CO2/40% N<sup>2</sup> (C-CO<sup>2</sup> ) with 0, 1, or 2 liquid absorbent pads (0A, 1A, 2A) and two different g/p ratios (1.8 and 2.9) (Experiment 2). Sampling times were after 7, 14, and 20 days of cold storage.

Statistical analysis of variance showed that the main differences in TVC were explained by storage time (98.4%) (Appendix A Table A1). Some differences in TVC were detected after 14 days of storage, but there were no significant differences for samples stored for 20 days, showing similar bacterial counts as for the previous TVC results (Appendix A Table A4). This indicates that the microbial growth on the chicken breast fillets is not affected by changing the g/p ratio—and thereby, the amount of available CO2—for this packaging concept. Due to the antimicrobial effects of CO2, increasing the g/p ratio may have been expected to produce an increased antimicrobial effect of the packaging gas. However, the percentage of CO2 in the headspace was the same, and the results demonstrate that the amount of CO2 available at a g/p of 1.8 is adequate to produce a comparable effect of CO2 in terms of microbiological quality as measured for a g/p of 2.9. These results also support the finding that microbiological growth on the chicken fillets is independent of the presence of or the number of liquid absorbent pads in the packages, i.e., whether or not there is free liquid inside the packages. Statistical analysis of variance showed that the main differences in TVC were explained by storage time (98.4%) (Appendix A Table A1). Some differences in TVC were detected after 14 days of storage, but there were no significant differences for samples stored for 20 days, showing similar bacterial counts as for the previous TVC results (Appendix A Table A4). This indicates that the microbial growth on the chicken breast fillets is not affected by changing the g/p ratio—and thereby, the amount of available CO2—for this packaging concept. Due to the antimicrobial effects of CO2, increasing the g/p ratio may have been expected to produce an increased antimicrobial effect of the packaging gas. However, the percentage of CO<sup>2</sup> in the headspace was the same, and the results demonstrate that the amount of CO<sup>2</sup> available at a g/p of 1.8 is adequate to produce a comparable effect of CO<sup>2</sup> in terms of microbiological quality as measured for a g/p of 2.9. These results also support the finding that microbiological growth on the chicken fillets is independent of the presence of or the number of liquid absorbent pads in the packages, i.e., whether or not there is free liquid inside the packages.

#### **4. Conclusions**

A significant increase in drip loss with an increasing number of liquid absorbent pads was documented for chicken breast fillets. The percentage of dry matter content increased during storage; however, only minor effects of gas composition and the number of absorbent pads could be detected. Sensory properties related to the texture were not significantly affected by the number of pads for the fillets. Microbiological analyses showed that TVC was independent of the number of absorbent pads, i.e., whether or not there was free liquid in the packages. This is an interesting finding as it disproves the common belief that excess liquid inside packages results in a higher bacterial load on the product. By comparing two different packaging gases in MAP, it was confirmed that for both packaging gases increased numbers of liquid absorbent pads in the packages led to increased drip loss. Still, there was no pronounced effect on sensory quality (texture), though regardless of liquid absorption, higher intensity scores of negatively associated odor attributes were detected for the high-O<sup>2</sup> atmospheres compared to the high-CO<sup>2</sup> atmospheres.

This study demonstrates the importance of product-adjusted capacity of liquid absorbers in order to maintain product yield, which can have economic benefits for food

producers and retailers. Visible free liquid inside the packages did not affect the bacterial load, but might result in rejection by the consumer due to unattractive packages, which might have an effect in terms of increased food waste.

**Author Contributions:** Conceptualization, M.K.P.; methodology, M.K.P. and M.C.; validation, M.K.P., J.N.-N., A.Å.H., M.C. and K.H.L.; formal analysis, K.H.L. and M.K.P.; investigation, M.K.P., A.Å.H., and J.N.-N.; writing—original draft preparation, M.K.P. and J.N.-N.; writing—review and editing, M.K.P., J.N.-N., A.Å.H., M.C. and K.H.L.; visualization, M.K.P. and J.N.-N. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was funded by the Norwegian Levy on Agricultural Products through the Strategic Programs FoodMicro-Pack (Project no 262306) and FutureFoodControl (Project no 314743), and The Research Council of Norway in the "FoodPilot Plant (Project no 296083).

**Informed Consent Statement:** Informed consent was obtained from all subjects involved in the study.

**Data Availability Statement:** The data presented in this study are available on request from the corresponding author.

**Acknowledgments:** We would like to thank Aud Espedal, Janina Berg, Karen Wahlstrøm Sanden, and Jens Petter Wold for their skillful technical assistance.

**Conflicts of Interest:** The authors declare no conflict of interest.

## **Appendix A**

**Table A1.** Summaries of analyses of variance for measurements on chicken breast fillets (Experiments 1 and 2). Numbers show percentage of variation per factor/interaction.




*p*-values are indicated as 0 \*\*\* 0.001 \*\* 0.01 \* 0.05 ◦ 0.1 e.g., one star for *p*-values between 0.01 and 0.05.

**Table A2.** One-way analyses of variance for drip loss, dry matter content and Warner–Bratzler shear force (N/cm<sup>2</sup> ) (maximum load) for each recorded day of storage of chicken breast fillets (Experiment 1). The one-way ANOVAs were computed by defining a six-level factor with the combinations of gas composition and liquid absorbent pads. Tukey's letters signify which of the six combinations of gas composition and liquid absorbent pad numbers are not significantly different (*p*-value > 0.05), i.e., levels sharing a letter are not significantly different. In the lower part of the table (Experiment 2), gas volume/product volume replaces gas composition.


**Table A3.** Sensory intensity scores (scale 1–9) for flavor attributes for heat-treated samples and odor attributes for raw chicken breast fillets stored for 14 days in modified atmospheres of 60% CO2/40% N<sup>2</sup> and 75% O2/25% CO<sup>2</sup> with 0, 1, or 2 liquid absorbent pads (Experiment 1). Levels sharing a letter are not significantly different (*p* < 0.05) between the chicken breasts samples, as measured by two-way ANOVA and Tukey's multiple comparison test.



**Table A3.** *Cont.*

**Table A4.** Microbiological content measured as total viable count of bacteria (TVC), lactic acid bacteria (LAB), *Brochothrix thermosphacta,* and *Enterobacteriaceae* given as log cfu/g for chicken breast fillets (Experiment 1 and 2) as a function of storage time, packaging variables, gas composition (60% CO2/40% N<sup>2</sup> and 25% CO2/75% O<sup>2</sup> ), or g/p ratio (1.8, 2.9, 2.0 and 2.1), with 0, 1, or 2 liquid absorbent pads. Sampling times after 6 (7), 14, and 20 days of cold storage. Values are mean ± standard deviation (*n* = 3).



**Table A4.** *Cont.*

(Bo) = sampling from the bottom surfaces of the chicken breast fillets. (\*) value given as > (too low dilutions applied—at least two replicates over growth and not possible to count); (\*\*) value given as < (too high dilutions applied–at least two replicates below detection limit at this dilution). Means sharing letters (Tukey) within the same column and storage time are not significantly different (*p* > 0.05) (one-way ANOVA with six levels per day).

#### **References**


**Teresa Fernández-Menéndez <sup>1</sup> , David García-López <sup>2</sup> , Antonio Argüelles <sup>3</sup> , Ana Fernández <sup>4</sup> and Jaime Viña 1,\***


**Abstract:** New PET and nanosepiolite materials are produced for its application in innovative packaging with better performance. In our previous work, we demonstrate that the use of different percentages of sepiolite modified with different organosilanes improved mechanical and barrier properties of PET. Nanocomposites permeability can decrease up to 30% compared to that of pure PET and the mechanical analyses show that, although PET nanocomposites are more brittle than virgin PET, they are also harder. In the present work, we are going to study the properties of this innovative packaging with real food analyzing mechanical properties related to the product transport together with permeability and microbiological characteristics. At the same time, it has been seen that it is possible to lighten trays, which is very important both industrially and environmentally. On the other hand, a good quality packaging for food needs to ensure that organoleptic and physicochemical characteristics of the product inside are not modified due to migration of any of the packaging material to the food itself. Results obtained in this work also show lower count of aerobic mesophilic bacteria and *Enterobacteriaceae* (EB), reducing the incidence of food contaminations by microorganisms.

**Keywords:** PET; sepiolite; nanocomposites; MAP; microbiological quality; chicken

#### **1. Introduction**

Food packaging has many useful functions, such as food containment, marketing, protection, and preservation during the shelf life of a product. In order to accomplish all this, a food packaging material must have enough strength to overcome its filling process, transport, and customer handling. At the same time, it needs to have the appropriate barrier properties for certain applications, such as in modified atmosphere packaging (MAP), and, of course, it needs to keep migration of packaging components to food to a minimum, complying with all regulation regarding Food Contact Materials (FCM), such as Food and Drug Administration (FDA) in USA, or European Regulation (EU) No 10/2011 of 14 January 2011.

Due to its low weight and versatility, among other things, polymers have been one of the most important materials used in food packaging. Plastic packaging plays a key role in protecting food from exterior virus and microorganisms as well as helping extend shelf life of packed food. However, it is also very important to keep migration of plastic components into food to a minimum, since that could ruin the product packed. Migration tests are regulated by legislation; thus, the amount of migrants allowed in the food is specified by the global migration limits. At the same time, the legislation records the lists of permitted substances to be used in food contact materials, this is the "positive list". Each of this substance has its migration limit specified, in order to avoid food toxicity. The analysis

**Citation:** Fernández-Menéndez, T.; García-López, D.; Argüelles, A.; Fernández, A.; Viña, J. Application of PET/Sepiolite Nanocomposite Trays to Improve Food Quality. *Foods* **2021**, *10*, 1188. https://doi.org/10.3390/ foods10061188

Academic Editors: Valeria Rizzo, Muratore Giuseppe and Benu P. Adhikari

Received: 15 April 2021 Accepted: 21 May 2021 Published: 25 May 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

done for controlling the amount of each substance in the packaging are called specific migration tests.

Fresh products (poultry, fruits, vegetables, etc.) are a vehicle for the transmission of bacterial, parasitic, and viral pathogens capable of causing human illness [1]. Within food packaging applications, improving shelf life of packed poultry is a huge challenge for the industry. Fresh poultry meat is highly popular among consumers and, at the same time, it is highly perishable (rapid microbial growth) leading to high economic losses [2,3]. Its shelf life depends mainly on poultry handling and processing (in the initial number of microorganisms) [4] or on its storage conditions through all the food chain.

One of the most important food packaging systems is MAP, where the food is packed, together with a certain mixture of gases that will keep freshness of the food, enhancing preservation, and extending shelf life. MPA requires a tray or base container and a lid to seal the packaged content. There are many plastic materials that could be used for this application, being poly(ethylene terephthalate) (PET) one of the most widely used. Depending on the amount of barrier required for the application, sometimes it is needed to apply a multilayer film with a high barrier polymer in it. In the present work, the base material used for the tray will be PET as well as for the lid. Adding the nanoclay to the PET it is expected to obtain trays with improve mechanical and barrier properties. One of the most used clays when talking about nanocomposites is montmorillonite (MMT). However, the clay chosen for the work has been sepiolite, which is a magnesium silicate with the following formula: Mg8Si12O30(OH)2(H2O)<sup>4</sup> 8H2O. It is a fibrous clay with nanometric dimensions that vary between 0.2 and 3 µm in length, 10–30 nm in width, and 5–10 nm in thickness, which gives the sepiolite a high aspect ratio of about 27. In addition, sepiolite has a surface area of about 300 m<sup>2</sup> g −1 , and an outer layer of silanol groups. All these characteristics make the sepiolite perfect for surface modification with organosilanes and other reagents on its surface. It has also shown better mechanical properties than MMT in previous works [5].

The migration analysis of the trays, as well as microbiological tests are done to prove the possibility of using this material for food packaging, complying with actual legislation.

#### **2. Materials**

PET pellets, kindly supplied by LINPAC Packaging (Pravia, Spain), were from Novapet S.A. (Zaragoza, Spain). The pellets' intrinsic viscosity (IV) was 0.81 dL/g. PET-EVOH-PE laminated sheet is a PET sheet laminated with an EVOH-PE flexible film (its structure being EVA/PA/EVOH/PA/PE). The sealing top film of the tray is a film coated with aluminum oxide (AlOx) (BOPET), Mylar® 850 from DuPont Teijin Films UK Limited, Middlesbrough, England. Mylar® 850 is a co-extruded, one side amorphous, heat sealable polyester film, suitable for use in contact with food. Its oxygen transmission rate is 56 cm3/m2/day, at 23 ◦C and 60/70% RH, for a 30 micron film. Two types of sepiolite were supplied by Tolsa S.A. (Madrid, Spain), one modified with 2% of 3-metracyloxypropil trimetoxysilane (MEMO, CAS 2530-85-2) and the other one with 2% of 3-aminopropyltriethoxysilane (AMEO, CAS 919-30-2). Both organomodifiers are suitable for food packaging with restrictions regarding the amount of absorbed substance by kg of product packed, as stated in Regulation CE 975/2009 for MEMO (0.05 mg by kg of packed product) and in Directive 2007/19/CE for AMEO (between 0.05 mg to 3 mg by kg of product in the package). The nanocomposite masters were produced at Repol S.L. (Almazora Castellón, Spain) in an industrial polyamide low shear extruder.

#### **3. Experimental Part**

In order to produce nanocomposites at industrial level, it was necessary to do the first steps at laboratory scale, as shown in previous works [6]. The materials used in this paper, are those found to be the best ones in terms of processability and mechanical properties.

The first step of nanocomposite fabrication at industrial level was to produce the masterbatches.

The PET/nanosepiolite masters were produced at Repol S.L. facilities. Conditions in the production plant were optimized to minimize PET matrix degradation, reducing humidity, and decreasing extrusion shear on the nanocomposites. Two different masters with 10% sepiolite each were produced. In one master, the sepiolite was previously modified with MEMO, and the other one with AMEO in Tolsa S.A. Then, these masters were diluted to the final percentage of sepiolite into the extruder in order to obtain the corresponding sheets for characterization. The use of these modifiers reduces the sepiolite– sepiolite interactions, which favors a better dispersion, and alignment of nanofibers that translate in effectiveness in mechanical properties [7].

Drying conditions for the master were 80 ◦C for 7 h, and 120 ◦C for 7 h for the pristine PET. The drier used was a CRAMER-TROCKNER model PK 100/300F. With the aim of simplification, from now on virgin PET will be referred to as PET.

The industrial extruder used was a Luigi Bandera SpA twin screw extruder, from LINPAC Packaging. In this extruder it was obtained the nanocomposite's sheet that then is taken to a KIEFEL GmbH thermoforming machine to obtain the desired final trays. The trays chosen for this project are MAP trays, with the following dimensions: 18 cm width, 25 cm length, and 45 mm depth. In this work it will be referred to as B1825-45 tray. The nanocomposite trays are sent to a poultry packer (Sada, Nutreco. Spain). There, 2 kg of breast chicken are packed in a modified atmosphere containing 70% CO2, 20% O2, and 10% N<sup>2</sup> in each tray. Control trays are packed in the same way, in a PET tray. Then, samples are taken for microbiological analysis of mesophilic aerobes and *Enterobacteriaceae* for 14 days. In this work, the microbiological quality of chicken fillets was assessed by determining the number of mesophilic aerobic bacteria, and *Enterobacteriaceae*. These analyses will help us determine if microbial load of those species in chicken, packed in nanocomposite trays, is lower than that packed in regular PET, in order to have an idea of food sanitary quality.

#### *3.1. TGA*

Thermogravimetric Analysis (TGA) was used to determine nanosepiolite percentage within the nanocomposite sheets. The analyses were performed in a Mettler Toledo 851e equipment, using a procedure in two steps:

First step: from 50 to 600 ◦C at 20 ◦C/min under nitrogen atmosphere.

Second step: from 600 to 900 ◦C at 20 ◦C/min under air atmosphere.

#### *3.2. Permeability*

The permeability analyses were done on sheet samples; specimens were taken from the extruded sheets before going to thermoforming into trays.

Oxygen transmission rate (OTR) was measured in an OXTRAN with a volumetric sensor (Oxtran SS 2/20, MOCON. Barcelona, Spain). Previously to the analysis the samples were conditioned, 48 h under an atmosphere with 0% RH. Oxygen transmission rate was measured at 23 ◦C and 0% RH following Standard ASTM D3985-17 and the effective area exposed to permeation was 50 cm<sup>2</sup> .

#### *3.3. Puncture Test*

Plastic products are more prone to fail when submitted to a multiaxial impact, rather than to a slow-motion load. In many applications, packaging materials are exposed to penetrating damages, which lead to barrier properties and package integrity deterioration. Thus, it is very important to obtain packaging materials with good impact strength properties that help preserve the food until its use.

These impact tests were done in an MTS-831 equipment, following ISO 6603-2:2000 methodology [8]. The speed used was 4.4 m/s and tests were at room temperature (23 ◦C).

The specimen, with an effective diameter of 40 mm, is hold with two anchor rings, then the impactor (ϕimpactor = 20 mm) hits on the specimen center from below. The curve strength versus strain is registered for each sample, together with absorbed energy (E). However, it is very important to describe the failure mode in order to know if the material is

going to break in a fragile way, a ductile, or in any of the intermediate modes in between [8]. In a ductile break (D), the specimen breaks slowly deforming the material with the absorbed E, while the additional, non-absorbed E, is used to extend the crease (Dc). On the contrary, in a fragile break (F) the crease is spread quickly, suddenly, and totally, causing the break of the sample.

#### *3.4. Compression Test*

Lateral compression tests and stiffness were tested on a Hounsfield H1KS Benchtop equipment following LINPAC Packaging internal procedures for trays. For each trial, between 65 and 70 trays were tested.

#### *3.5. Microbiological Tests*

Twelve specimens of the nanocomposite trays were analyzed for each material, and sixteen for control.

For total viable count (TVC) determination, 25 g of superficial meat are taken aseptically. Samples are mixed with 225 mL of buffered peptone water and is then homogenized in a Stomacher® (dilution 1:10). After that, 1 mL sample is taken from the main dilution and then the dilutions needed to obtain an appropriate number of microorganisms are done. The incubation time and temperature for mesophilic aerobes are 72 h at 30 ◦C and a Petrifilm Aerobic Count Plate is used (ISO 4833-1, 2003 [9]), and 24 h at 37 ◦C for Enterobateriacae using a 3M Petrilm to help counting (ISO 21528-2, 2004 [10]). Bacterial count results are expressed in log<sup>10</sup> of colony-forming units per gram of meat (log cfu/g).

Microbiological analyses were done, on 3 samples per day and treatment, on the following days post packaging: 2, 7, 10, and 14. At the same time, head space gases were measured to see the evolution during the microbial study, using an OXIBABY-M O2/CO2 (WITT-GASETECHNIK. Witten, Germany). The specimens were kept at 5 ◦C during all that period.

#### **4. Results and Discussion**

It has been analyzed before [6], the effect caused by different nanosepiolite masterbatches concentration (one with 20%, and the other one with 10%) on the final nanocomposite properties. It was concluded that those nanocomposites coming from a less concentrated master had better homogeneity, as well as viscosity and mechanical properties of the final material. Thus, for this study, a 10% nanosepiolite master was aimed. This percentage of sepiolite is theoretical because after dosing the sepiolite in an industrial extruder, the final amount of nanoclay in the nanocomposite changes. This is due to the difficulty of adding a powder to an industrial extruder at 600 kg/h. In this way, a PET masterbatch with 7.66% sepiolite modified with MEMO (MB1), and another one with 8.56% of sepiolite modified with AMEO (MB2) were used in this study determined by TGA.

Table 1 shows the percentages of sepiolite in the nanocomposites, after dilution of the masters into the PET industrial extruder.


**Table 1.** Samples and sepiolite content on the nanocomposites.

#### *4.1. Permeability*

The permeability was analyzed on industrially extruded sheet samples with different thickness. Permeability to oxygen, calculated as OTR, improved in all nanocomposite's samples compared to that of pure PET. The improvement can be up to 30% with 1.37% of nanosepiolite. It is shown the OTR/Sheet Thickness, which is calculated dividing the permeability, given in OTR units, by the sample thickness. It was observed that increasing the clay concentration did not change the permeability of the samples substantially (Figure 1a). In addition, as stated by Ke and Yongping [11] and tested in this work, the processability of the nanocomposite is much more difficult when increasing the amount of nanosepiolite over 3% [6,12]. It is possible to decrease the amount of nanoclay in the samples, as long as the nanoparticles are well dispersed and oriented within the matrix [13–16]. However, if the sepiolite content is too low it will not do the job and if it is too much, PET matrix viscosity will decrease and it will not disperse properly, opening the path for gas and vapor molecules.

Comparing the nanocomposite samples to a PET-EVOH-PE laminated sheet, which is the sheet normally used when high barrier is required, we can see 20% improvement in the sample containing 1.37% of nanosepiolite modified with MEMO (sample M2). There are no mayor differences in terms of permeability performance between the two silanes used in this study. *Foods* **2021**, *10*, x FOR PEER REVIEW 6 of 14

**Figure 1.** Oxigen transmission rate (OTR)/Sheet Thickness as a function of sepiolite content on the nanocomposite samples (**a**); OTR for nanocomposites as a function of sheet thickness (**b**). **Figure 1.** Oxigen transmission rate (OTR)/Sheet Thickness as a function of sepiolite content on the nanocomposite samples (**a**); OTR for nanocomposites as a function of sheet thickness (**b**).

These tests are done on the industrially extruded sheet before the thermoforming process. Registered curves for impact tests show a maximum which is related to the initial damage on the sheet, corresponding to the starting point of the fissure that will develop in a fracture. Analyzing the curves obtained in this test, maximum load and its associated deformation can be known, as well as perforation energy. On the other hand, this test shows the way the sample breaks allowing us to define the failure mode of each specimen. Thickness is measure on extruded sheet, and the results shown in Table 2 are the average obtained in all the samples width (795 mm). Since this measure is of great importance in impact results, we will compare samples with the same thickness. Thus, for a sheet with 465 µm, a sample M5 with 1.65% nS shows an 8% improvement in impact strength [14]; nevertheless, the impact energy decreases 18% in the nanocomposite compared to that of pure PET (with 467 µm) (Table 2). This means there is less energy for deformation, but it takes a little higher force to break the nanocomposite sheet. The same behavior is seen in nanocomposites with MEMO, but impact and energy values are a little

tests that can simulate its treatment once at the market.

*4.2. Mechanical Properties of Nanocomposite Trays* 

lower than in those modified with AMEO.

Looking at Figure 1b it can be seen that, in order to achieve the same permeability of a 450 µm PET sample, a nanoPET composites could decreased its thickness in about 100–150 µm. These results are really important in terms of reducing PET consumption, which is good for both company savings in materials and transport, and for the environment.

#### *4.2. Mechanical Properties of Nanocomposite Trays*

In order to overcome its transport from factory to houses, including customer's handling, trays must be tough enough. That is why it is so important to test the packages with tests that can simulate its treatment once at the market.

These tests are done on the industrially extruded sheet before the thermoforming process. Registered curves for impact tests show a maximum which is related to the initial damage on the sheet, corresponding to the starting point of the fissure that will develop in a fracture. Analyzing the curves obtained in this test, maximum load and its associated deformation can be known, as well as perforation energy. On the other hand, this test shows the way the sample breaks allowing us to define the failure mode of each specimen.

Thickness is measure on extruded sheet, and the results shown in Table 2 are the average obtained in all the samples width (795 mm). Since this measure is of great importance in impact results, we will compare samples with the same thickness. Thus, for a sheet with 465 µm, a sample M5 with 1.65% nS shows an 8% improvement in impact strength [14]; nevertheless, the impact energy decreases 18% in the nanocomposite compared to that of pure PET (with 467 µm) (Table 2). This means there is less energy for deformation, but it takes a little higher force to break the nanocomposite sheet. The same behavior is seen in nanocomposites with MEMO, but impact and energy values are a little lower than in those modified with AMEO. *Foods* **2021**, *10*, x FOR PEER REVIEW 7 of 14 A PET sheet (100% virgin) shows a ductile behavior, whilst nanocomposite materials evidence a fragile component when using MEMO modified sepiolite (see Table 2). However, samples modified with AMEO do not show fragile behavior but a ductile with a forming crease one. Otherwise, there are no notable differences in nanocomposites impact behavior depending on the organomodifier used. *Foods* **2021**, *10*, x FOR PEER REVIEW 7 of 14 A PET sheet (100% virgin) shows a ductile behavior, whilst nanocomposite materials evidence a fragile component when using MEMO modified sepiolite (see Table 2). However, samples modified with AMEO do not show fragile behavior but a ductile with a forming crease one. Otherwise, there are no notable differences in nanocomposites impact behavior depending on the organomodifier used. *Foods* **2021**, *10*, x FOR PEER REVIEW 7 of 14 A PET sheet (100% virgin) shows a ductile behavior, whilst nanocomposite materials evidence a fragile component when using MEMO modified sepiolite (see Table 2). However, samples modified with AMEO do not show fragile behavior but a ductile with a forming crease one. Otherwise, there are no notable differences in nanocomposites impact behavior depending on the organomodifier used. *Foods* **2021**, *10*, x FOR PEER REVIEW 7 of 14 A PET sheet (100% virgin) shows a ductile behavior, whilst nanocomposite materials evidence a fragile component when using MEMO modified sepiolite (see Table 2). However, samples modified with AMEO do not show fragile behavior but a ductile with a forming crease one. Otherwise, there are no notable differences in nanocomposites impact behavior depending on the organomodifier used.

**Table 2.** Impact puncture results for nanocomposites containing nanosepiolite modified with MEMO (3-metracyloxypropil trimetoxysilane) and with AMEO (3-aminopropyltriethoxysilane). **Table 2.** Impact puncture results for nanocomposites containing nanosepiolite modified with MEMO (3-metracyloxypropil trimetoxysilane) and with AMEO (3-aminopropyltriethoxysilane). **Table 2.** Impact puncture results for nanocomposites containing nanosepiolite modified with MEMO (3-metracyloxypropil trimetoxysilane) and with AMEO (3-aminopropyltriethoxysilane). **Table 2.** Impact puncture results for nanocomposites containing nanosepiolite modified with MEMO (3-metracyloxypropil trimetoxysilane) and with AMEO (3-aminopropyltriethoxysilane). **Table 2.** Impact puncture results for nanocomposites containing nanosepiolite modified with MEMO (3-metracyloxypropil trimetoxysilane) and with AMEO (3-aminopropyltriethoxysilane).


**AMEO** 

**AMEO** 

**AMEO** 

**AMEO** 

M4 1.59 320 ± 6 520 ± 15 3.1 ± 0.7 FD-Dc

M4 1.59 320 ± 6 520 ± 15 3.1 ± 0.7 FD-Dc

M4 1.59 320 ± 6 520 ± 15 3.1 ± 0.7 FD-Dc

M4 1.59 320 ± 6 520 ± 15 3.1 ± 0.7 FD-Dc

M5 1.65 465 ± 9 780 ± 34 5.4 ± 0.7 Dc

M5 1.65 465 ± 9 780 ± 34 5.4 ± 0.7 Dc

M5 1.65 465 ± 9 780 ± 34 5.4 ± 0.7 Dc

M5 1.65 465 ± 9 780 ± 34 5.4 ± 0.7 Dc

M6 1.88 601 ± 12 1100 ± 40 8.5 ± 0.4 D

M6 1.88 601 ± 12 1100 ± 40 8.5 ± 0.4 D

M6 1.88 601 ± 12 1100 ± 40 8.5 ± 0.4 D

M6 1.88 601 ± 12 1100 ± 40 8.5 ± 0.4 D


*Foods* **2021**, *10*, x FOR PEER REVIEW 7 of 14

*Foods* **2021**, *10*, x FOR PEER REVIEW 7 of 14

*Foods* **2021**, *10*, x FOR PEER REVIEW 7 of 14

*Foods* **2021**, *10*, x FOR PEER REVIEW 7 of 14

**Thickness (µm)** 

**Thickness (µm)** 

**Thickness (µm)** 

**Thickness (µm)** 

behavior depending on the organomodifier used.

behavior depending on the organomodifier used.

behavior depending on the organomodifier used.

behavior depending on the organomodifier used.

**Sample TGA** 

**Sample TGA** 

**Sample TGA** 

**Sample TGA** 

**(% nS)** 

**(% nS)** 

**(% nS)** 

**(% nS)** 

A PET sheet (100% virgin) shows a ductile behavior, whilst nanocomposite materials evidence a fragile component when using MEMO modified sepiolite (see Table 2). However, samples modified with AMEO do not show fragile behavior but a ductile with a forming crease one. Otherwise, there are no notable differences in nanocomposites impact

A PET sheet (100% virgin) shows a ductile behavior, whilst nanocomposite materials evidence a fragile component when using MEMO modified sepiolite (see Table 2). However, samples modified with AMEO do not show fragile behavior but a ductile with a forming crease one. Otherwise, there are no notable differences in nanocomposites impact

A PET sheet (100% virgin) shows a ductile behavior, whilst nanocomposite materials evidence a fragile component when using MEMO modified sepiolite (see Table 2). However, samples modified with AMEO do not show fragile behavior but a ductile with a forming crease one. Otherwise, there are no notable differences in nanocomposites impact

A PET sheet (100% virgin) shows a ductile behavior, whilst nanocomposite materials evidence a fragile component when using MEMO modified sepiolite (see Table 2). However, samples modified with AMEO do not show fragile behavior but a ductile with a forming crease one. Otherwise, there are no notable differences in nanocomposites impact

**Table 2.** Impact puncture results for nanocomposites containing nanosepiolite modified with MEMO (3-metracyloxypropil trimetoxysilane) and with AMEO (3-aminopropyltriethoxysilane).

**Table 2.** Impact puncture results for nanocomposites containing nanosepiolite modified with MEMO (3-metracyloxypropil trimetoxysilane) and with AMEO (3-aminopropyltriethoxysilane).

**Table 2.** Impact puncture results for nanocomposites containing nanosepiolite modified with MEMO (3-metracyloxypropil trimetoxysilane) and with AMEO (3-aminopropyltriethoxysilane).

**Table 2.** Impact puncture results for nanocomposites containing nanosepiolite modified with MEMO (3-metracyloxypropil trimetoxysilane) and with AMEO (3-aminopropyltriethoxysilane).

> **Max Load (N)**

> **Max Load (N)**

**Max Load (N)** 

**Max Load (N)** 

**MEMO** 

**MEMO** 

**MEMO** 

**MEMO** 

M0 0 467 ± 2 720 ± 16 6.6 ± 0.8 D

M0 0 467 ± 2 720 ± 16 6.6 ± 0.8 D

M0 0 467 ± 2 720 ± 16 6.6 ± 0.8 D

M0 0 467 ± 2 720 ± 16 6.6 ± 0.8 D

M2 1.37 351 ± 12 560 ± 42 2.6 ± 0.6 FD

M2 1.37 351 ± 12 560 ± 42 2.6 ± 0.6 FD

M2 1.37 351 ± 12 560 ± 42 2.6 ± 0.6 FD

M2 1.37 351 ± 12 560 ± 42 2.6 ± 0.6 FD

M3 1.78 470 ± 11 730 ± 13 5.0 ± 2.4 FD-D

M3 1.78 470 ± 11 730 ± 13 5.0 ± 2.4 FD-D

M3 1.78 470 ± 11 730 ± 13 5.0 ± 2.4 FD-D

M3 1.78 470 ± 11 730 ± 13 5.0 ± 2.4 FD-D

**Puncture Energy (J)** 

**Puncture Energy (J)** 

**Puncture Energy (J)** 

**Puncture Energy (J)**  **Failure** 

**Failure** 

**Failure** 

**Failure** 

**Mode \* Picture** 

**Mode \* Picture** 

**Mode \* Picture** 

**Mode \* Picture** 

\* Failure Modes: D—Ductile, FD—transition Fragile/Ductile, Dc—Ductile with crease.

A PET sheet (100% virgin) shows a ductile behavior, whilst nanocomposite materials evidence a fragile component when using MEMO modified sepiolite (see Table 2). However, samples modified with AMEO do not show fragile behavior but a ductile with a forming crease one. Otherwise, there are no notable differences in nanocomposites impact behavior depending on the organomodifier used.

Regarding sample's thickness, when increasing this value, maximum load, impact energy, and ductility at break increases in all the samples no matter the sepiolite content or its organic modifier.

In order to normalize result to thickness, data in Newtons were divided by the thickness of each sample. The results show that samples modified with AMEO improve its maximum load at impact when increasing nanosepiolite content, whereas samples with MEMO have poorer results. Impact load increases in all samples with AMEO over that of pure PET, with a 19% increase in sample containing 1.88% nanosepiolite (M6). Meanwhile, the maximum load improvement in MEMO samples is of 9% over that of virgin PET.

#### *4.3. Compression Tests*

Compression tests results are done with the tray in vertical position since is, generally, the most critical force MAP trays are going to be subjected to. Results indicate that the nanocomposite trays have more resistance to compression forces compare to the pure PET trays.

Nanocomposites are more resistant to lateral compression than pure PET. The improvement can be 66.7% for the sample M6 containing 1.88% nS.

In the following graph (Figure 2) it is shown the stiffness per gram. The results show that for the same tray weight the stiffness increase can be up to 85% with 1.88% nanosepiolite AMEO modified. Thus, the addition of nanosepiolite to a PET matrix results in increased stiffness and decreased ductility, as seen in these results [6,15]. This could be

due to a good interaction between the PET matrix and the sepiolite. The high aspect ratio of this clay is key in this interaction, together with the organomodifiers used [16,17]. It is also seen here an increase in stiffness with the amount of sepiolite when using AMEO as organomodifier, an increase of 0.23 g nanosepiolite produces an improvement of 40% in stiffness. Whereas in a sample with nearly double sepiolite content modified with MEMO (1.09 g more), there is an increase of just 4%. creased stiffness and decreased ductility, as seen in these results [6,15]. This could be due to a good interaction between the PET matrix and the sepiolite. The high aspect ratio of this clay is key in this interaction, together with the organomodifiers used [16,17]. It is also seen here an increase in stiffness with the amount of sepiolite when using AMEO as organomodifier, an increase of 0.23 g nanosepiolite produces an improvement of 40% in stiffness. Whereas in a sample with nearly double sepiolite content modified with MEMO (1.09 g more), there is an increase of just 4%.

*Foods* **2021**, *10*, x FOR PEER REVIEW 8 of 14

or its organic modifier.

*4.3. Compression Tests* 

PET trays.

\* Failure Modes: D—Ductile, FD—transition Fragile/Ductile, Dc—Ductile with crease.

Regarding sample's thickness, when increasing this value, maximum load, impact energy, and ductility at break increases in all the samples no matter the sepiolite content

In order to normalize result to thickness, data in Newtons were divided by the thickness of each sample. The results show that samples modified with AMEO improve its maximum load at impact when increasing nanosepiolite content, whereas samples with MEMO have poorer results. Impact load increases in all samples with AMEO over that of pure PET, with a 19% increase in sample containing 1.88% nanosepiolite (M6). Meanwhile, the maximum load improvement in MEMO samples is of 9% over that of virgin PET.

Compression tests results are done with the tray in vertical position since is, generally, the most critical force MAP trays are going to be subjected to. Results indicate that the nanocomposite trays have more resistance to compression forces compare to the pure

Nanocomposites are more resistant to lateral compression than pure PET. The im-

In the following graph (Figure 2) it is shown the stiffness per gram. The results show that for the same tray weight the stiffness increase can be up to 85% with 1.88% nanosepiolite AMEO modified. Thus, the addition of nanosepiolite to a PET matrix results in in-

provement can be 66.7% for the sample M6 containing 1.88% nS.

It is seen that nanocomposites sheets with around 1.6% nanosepiolite need 150 µm less thickness in order to generate the same stiffness as a pure PET samples. Thus, a sample with 1.65% nanosepiolite modified with AMEO (M5) has the same stiffness than a sample that weights 8 g more in pure PET. It is seen that nanocomposites sheets with around 1.6% nanosepiolite need 150 µm less thickness in order to generate the same stiffness as a pure PET samples. Thus, a sample with 1.65% nanosepiolite modified with AMEO (M5) has the same stiffness than a sample that weights 8 g more in pure PET.

**Figure 2.** Nanocomposites stiffness per gram compared to that of pure PET. **Figure 2.** Nanocomposites stiffness per gram compared to that of pure PET.

As it is shown in this paper, results with both AMEO and MEMO as organic modifiers for the sepiolite have quite similar performance. However, those nanocomposites with AMEO are more resistant to lateral compression forces, a little stronger and less brittle. Moreover, between the two nanocomposite's types there are some other differences which are important for good industrial extrusion and thermoforming processes. These differences are mainly that AMEO nanosepiolite seems to disperse better into the PET matrix [6], and its processability in an industrial extruder was a little better, showing a more stable behavior.

As a function of the results obtained, the sheet chosen for the next step is that containing between 1.5 and 1.9% of nanosepiolite modified with 2% of AMEO. At the same time, it will be used the trays with intermediate thickness. Thus, the sheet sample chosen to thermoformed trays to be used in the next step is M5. These trays were thermoformed in order to do microbiological analysis and migration test studies.

#### *4.4. Microbial Analysis in Packed Chicken Breast*

Microbiological analysis is done on breast chicken packed in the same trays as discussed previously, a B1825-45 tray made with M5 sheet (Tray A1, 465 m) and using a virgin PET tray in 601 µm as control (Tray A0). This thickness is the regular one that would be used for this type of product. The tray thickness refers to the average thickness of the sheet before thermoforming (the average value of all sheet width).

Mesophilic aerobes and *Enterobacteriaceae* family include not only pathogenic species, but environmental species as well, which often appear in the food manufacturing environment without posing any health hazard. In fresh food, high number counts are not

recommended, although an elevated count does not imply the presence of pathogenic flora. However, the total count reflects sanitary quality of the analyzed products. If the total count is high, testing of specific pathogens can be done. At the same time, a low count does not mean the sample is pathogen free, it depends on the composition of the microbiota [18]. This control should be done in accordance with Commission Regulation (EC) No 2073/2005.

In the following graphs (Figure 3) it is shown microbiology parameter's growth, together with atmosphere gases evolution with packing time for both samples chosen; which is to say, 2 kg of breast chicken packed in trays A1, and also the control samples which are packed in pure PET (A0). In both cases, the top film is BOPET coated with AlOx. Results shown in the graphics correspond to the average count obtain for the three samples examined, for each period (2, 7, 10, and 14 days). *Foods* **2021**, *10*, x FOR PEER REVIEW 10 of 14

**Figure 3.** Picture of the packed chicken (**a**), aerobic mesophilic and *Enterobacteriaceae* count (**b**,**d**) and gases concentration evolution in 2 kg chicken breast (**c**,**e**), packed in A1 and A0 nanocomposite tray, respectively. **Figure 3.** Picture of the packed chicken (**a**), aerobic mesophilic and *Enterobacteriaceae* count (**b**,**d**) and gases concentration evolution in 2 kg chicken breast (**c**,**e**), packed in A1 and A0 nanocomposite tray, respectively.

Regarding total count of *Enterobacteriaceae* (EB), the average value of PET trays ana-87

lyzed is 3.83 log (cfu/g) (8.25 × 103 ufc/g) whilst this number is 2.98 log (cfu/g) (9.67x102 ufc/g) for the nanocomposite tray. Thus, nanosepiolite trays have lower microbiological charge. EB remains practically constant and very low from day 2 to 14 for the nanocomposite trays, starting at an average log (cfu/g) of 2.83 and ending, at 14 days, at 2.98, whereas in PET trays, although the starting count of *Enterobacteriaceae* is lower, log (cfu/g) 2.43, at day 14 it is of 3.80. The explanation for the lower growth of bacterial counts could be due not only to CO2 [24] but also to N2 capability of inhibiting bacterial growth [25,26]. Figure 3b,d shows head space gases evolution for the breast chicken packed in nanocomposite's trays and in PET tray. CO2 content decrease is mainly due to product absorption into the food [21–28], but also to loss of the gas through permeation across the plastic. Major CO2 absorption takes place during the first two days in the two samples, resulting in an equilibrium concentration, lower than the initial [29]. The absorption of CO2 depends

Generally, in hygienically handled fresh food, with a limited shelf life, if the count of microorganism is high it will decrease the serviceable time of the products. The chicken microbiological quality was initially good in both trays, being at day 2 (first count analyses) lower than 4 log (cfu/g) [19,20]. At day fourteenth, none of the samples had arrived at 7 log (cfu/g) which is the maximum limit stablished for fresh chicken [21–23]. Looking at mesophilic aerobes results (Figure 3a,d), it can be seen the values of the two trays (A0, A1) are quite similar up until the fourteenth day, where results on nanocomposite's tray show lower numbers of colony forming units of mesophilic aerobes in those trays made with nanocomposite, up to one order of magnitude lower (from log (cfu/g) 6.57 to 5.25).

Regarding total count of *Enterobacteriaceae* (EB), the average value of PET trays analyzed is 3.83 log (cfu/g) (8.25 <sup>×</sup> <sup>10</sup><sup>3</sup> ufc/g) whilst this number is 2.98 log (cfu/g) (9.67 <sup>×</sup> <sup>10</sup><sup>2</sup> ufc/g) for the nanocomposite tray. Thus, nanosepiolite trays have lower microbiological charge. EB remains practically constant and very low from day 2 to 14 for the nanocomposite trays, starting at an average log (cfu/g) of 2.83 and ending, at 14 days, at 2.98, whereas in PET trays, although the starting count of *Enterobacteriaceae* is lower, log (cfu/g) 2.43, at day 14 it is of 3.80. The explanation for the lower growth of bacterial counts could be due not only to CO<sup>2</sup> [24] but also to N<sup>2</sup> capability of inhibiting bacterial growth [25,26].

Figure 3b,d shows head space gases evolution for the breast chicken packed in nanocomposite's trays and in PET tray. CO<sup>2</sup> content decrease is mainly due to product absorption into the food [21–28], but also to loss of the gas through permeation across the plastic. Major CO<sup>2</sup> absorption takes place during the first two days in the two samples, resulting in an equilibrium concentration, lower than the initial [29]. The absorption of CO<sup>2</sup> depends on partial pressure in the head space of the packed tray [30] and also on the product itself (for example, chicken with skin or without it). In this study, these factors are the same for the two trays mentioned. However, the final CO<sup>2</sup> content is approximately 20% higher in nanocomposite's trays. Thus, in spite of the lower thickness of the new trays, the decrease in permeability helps in keeping CO<sup>2</sup> inside the package.

Regarding O2, this was maintained quite constant throughout the 14 days due to initial amount being very similar to atmosphere O<sup>2</sup> concentration (approximately 20.9%). O<sup>2</sup> percentage increased during the first two days, due to the high decrease of CO<sup>2</sup> in that period, which caused loss of volume, and to the lack of diffusion of O<sup>2</sup> to the exterior since its partial pressure is nearly the same in and outside the package. After those days, O<sup>2</sup> remained quite constant for the reason already mention. It can be also seen in the gas's evolution figures that N<sup>2</sup> expands at the same time that CO<sup>2</sup> is absorbed into the chicken.

The improved barrier performance of the PET/sepiolite nanocomposites has, therefore, produced a decreased on microbial counts in the poultry trays due to their ability to keep modified atmosphere gases longer [12,31,32].

#### *4.5. Migration*

Global migration test for food contact plastics are done in a variety of food simulants, depending on the food type it is aimed to contain (fatty food, vegetables, etc.) and the storage conditions. In this work, global migration tests have been done following specifications under UNE-EN 1186-14 using iso-octane as a fatty food simulant.

For these analyses, the samples chosen were M3 and M6 (PET + 1.78% nS\_MEMO and PET + PET + 1.88% nS\_AMEO, respectively), since these are the ones with the highest sepiolite content. Global migration limits within European Regulation (UE) No 10/2011 and RD 866/2008 are 10 mg/dm<sup>2</sup> or 60 mg/kg. Results on both trays are lower than 1.0 mg/dm<sup>2</sup> ; thus, both trays (M3 and M6) comply with global migration limits for the chosen simulant and under the test conditions.

Regarding specific migration limits for MEMO and AMEO (0.05 mg/dm<sup>2</sup> ), being the results 0.04 mg/dm<sup>2</sup> for both MEMO and AMEO, it can be concluded that both trays also comply with the limits stablished in Directive 2002/72/EC and its subsequent amendments (Regulation EC 975/2009) for this product types. Actual legislation for food contact materials does not contain specific migration limits for natural silicates (except asbestos) such as sepiolite. However, since sepiolite is a magnesium silicate, the global migration of silicon and magnesium in iso-octane has been analyzed in order to know if there is any sepiolite migration from tray material to the packaged food. Results for silicon migration are lower than 0.05 mg/kg in both trays (M3 and M6), and results for magnesium are lower than 1.0 mg/kg. For this reason, it is concluded that migration of these two elements to the packaged food is well under the legislation limit.

#### **5. Conclusions**

The results obtained in the study have showed that the global properties of the trays have been improved.

It is possible to improve PET permeability by 30% with a nanocomposite containing just 1.37% of sepiolite in its matrix. It would be possible to decrease tray thickness in about 100–150 µm in order to obtain the same permeability of a 450 µm PET sample. These results would imply company savings in terms of materials and transport, and also it would be great for the environment, using less raw materials.

Mechanical properties have been also measured. In the impact test, it has been seen an improvement in the maximum load at impact in all the samples but in M3, which value is equal to that of the PET. In those samples modified with AMEO, impact energy is the same as to a PET sample with 1.88% sepiolite, whilst for the MEMO modified samples the best one, nearly equal to that of PET in impact energy is that one with 1.2% clay.

Failure behavior is ductile for a 100% virgin PET and has a fragile component in the nanocomposite sheet with MEMO modified sepiolite. However, samples modified with AMEO does not show fragile behavior but a ductile with a forming crease one. Otherwise, MEMO and AMEO samples show similar mechanical performances.

Nanocomposites trays are stiffer than PET ones and show better performance to lateral compression forces. This type of compression is most critical for sealed trays, since the top film tends to produce tension and, if the tray is not stiff enough, it will bend or even collapse easily. This would give a poor packaging impression to customers. It has been seen that it is possible to lighten trays with the use of nanosepiolite clay, which is very important both industrially and environmentally.

The final nanocomposite trays show lower numbers of colony forming units of mesophilic aerobes in those trays made with nanocomposite, up to one order of magnitude lower (from log(cfu/g) 6.57 to 5.25). EB count remains practically constant and very stable from day 2 to 14 for the nanocomposite trays, being the final count one order of magnitude lower than that obtained in a virgin PET tray (from log(cfu/g) 3.83 to 2.98).

It is important to note that nanocomposite trays are 136 µm less thick than that of the PET serving as control. Thus, lighter nanosepiolite trays with lower microbiological charge can be produced at industrial machines. Moreover, these nanocomposite trays comply with European Regulation (UE) No 10/2011 and RD 866/2008 for materials intended to be in contact with food products.

**Author Contributions:** Conceptualization, D.G.-L. and A.F.; methodology, A.A.; validation, J.V.; investigation, T.F.-M.; resources, A.F.; writing-review and editing, T.F.-M.; supervision, J.V. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Data Availability Statement:** The data presented in this study are available on request from the corresponding author. The data are not publicly available due to privacy restrictions of the companies involved.

**Acknowledgments:** Authors of this paper would like to thank Klöckner Pentaplast in Pravia (former LINPAC Packaging S.A.U.), Cidaut, Ainia, Grupo Repol S.A. and Tolsa S.A. for their different collaborations in this project.

**Conflicts of Interest:** The authors declare no conflict of interest.

## **Nomenclature**


#### **References**

