Microplastic contaminations in a set of beverages sold in France/ Contaminações por microplásticos em um conjunto de bebidas vendidas na França
Link: https://www.sciencedirect.com/science/article/pii/S0889157525005344?via%3Dihub
Abstract
Microplastics are present in all environments and have even been detected in humans. As a result, more and more worldwide studies are focusing on the contamination of food and beverages by microplastics. To date, none of these studies has examined the contamination levels in beverages sold in France. The current study was set up to address this gap and investigate the level of microplastic contamination in water, soft drinks, beer and wine. This study does not aim to provide an exhaustive overview of all the drinks sold in France. However, efforts were made to study the impact of different containers: plastic, glass, brick, can, cubitainer, on this contamination. Heterogeneous results were obtained with mean contamination levels of 2.9 ± 0.7 MPs/L in waters, 31.4 ± 16 MPs/L in colas, 28.5 ± 13.1 MPs/L in teas, 45.2 ± 21.4 MPs/L in lemonades, 82.9 ± 13.9 MPs/L in beers and 8.2 ± 3.3 MPs/L in wines. It was observed that the most contaminated containers were glass bottles. Caps were suspected to be the main source of contamination, as the majority of particles isolated in beverages were identical to the color of caps and shared the composition of the outer paint.
1. Introduction
Initially considered as a revolutionary material, plastic production has steadily increased over the years. Excluding fibers, it has grown from 1.5 million tons in the 1950s to 400.3 million tons in 2022 (PlasticsEurope, 2023). Because of their high strength, low cost, low weight, and ease of use, plastics are used worldwide. More specifically, packaging made of polyethylene (PE) (high-density HDPE and low-density LDPE), polypropylene (PP), polyethylene terephthalate (PET), and polystyrene (PS) accounts for about 44 % of the volume produced on the global market (PlasticsEurope, 2022).
However, the increased manufacture of single-use plastics has also led to a rise in garbage found in terrestrial and aquatic habitats due to inefficient waste management. The 1970s marked the beginning of research into the existence of plastic debris in marine environments (Carpenter et al., 1972). The accumulation of plastics has been widely documented as a significant environment and human health concern.
The degradation of these plastics represents one threat, as fragmentation releases micro- or nanoplastics (Alimba and Faggio, 2019). Microplastics (MPs), defined in the current study as particles smaller than five millimeters (Arthur et al., 2009a), are present in all ecosystems. Their small size allows humans exposure through inhalation, ingestion, or skin penetration (Sun and Wang, 2023), with ingestion being the main route of human due to their presence in foods and beverages (Prata et al., 2020). Numerous studies have evaluated the degree of contamination of edible organisms, including bivalves, crustaceans, and different types of fish, due to the absorption of MPs by aquatic organisms (Rubio-Armendáriz et al., 2022).
In one hand, contamination of food and beverages can also originate from their packaging, most of which is made of plastic. It has already been shown that contamination of meat can come from its packaging (Kedzierski et al., 2020), and drinks can be contaminated by the screwing and unscrewing of the plastic cap (Winkler et al., 2019).
In other hand, research also examined whether the contamination could result from product processing, like canning or salting fish (Akhbarizadeh et al., 2020, Gündoğdu and Köşker, 2023, Hussien et al., 2021, Kim et al., 2021, Mohsen et al., 2023, Piyawardhana et al., 2022).
Few studies have shown that different beverages available for consumption are contaminated by MPs, with most of these studies focusing on bottled and tap water worldwide (Almaiman et al., 2021, Chu et al., 2022, Feld et al., 2021, Kirstein et al., 2021, Li et al., 2023, Mukotaka et al., 2021, Semmouri et al., 2022, Shen et al., 2021, Shruti et al., 2020, Tong et al., 2020, Volgare et al., 2022). However, more and more studies have been carried out on other beverages such as soft drinks (Altunışık, 2023, Basaran et al., 2024, Crosta et al., 2023, Lam et al., 2024, Pham et al., 2023, Shruti et al., 2020, Wang and Wang, 2024), beer (Kosuth et al., 2018, Liebezeit and Liebezeit, 2014, Pham et al., 2023, Wiesheu et al., 2016) and wine (Prata et al., 2020), but none on those sold in France.
Therefore, the purpose of this study exploratory study is to (i) evaluate the level of contamination present in a set of diverse beverages purchased in France and (ii) assess the potential influence of packaging on contamination. This study does not intend to provide an exhaustive overview of all beverages sold in France. However, it has been meticulously designed to assess the impact of various containers — plastic, glass, brick, can, and cubitainer — on this contamination.
2. Materials and methods
2.1. QA/QC
All manipulations were carried out under a Thermo Scientific Herasafe 2030i laminar flow cabinet (Saint-Herblain, France) in a room specifically dedicated for studying microplastics to lower airbone contaminations as well as the contribution of exogenous MPs by solutions. In compliance with EN14644–1, this laminar flow cabinet has an air purity class of ISO 5 and therefore a minimum filtration efficiency of 99.995 % for particles down to 0.3 µm.
All glassware and filters were heated to 450°C for 6 H in a Nabertherm L 40/11 muffle furnace (Lilienthal, Germany) prior to use in analysis.
All solutions, including water conform to EN-ISO 3696 grade 3, absolute ethanol from Carlo Erba (Val-de-rueil, France) diluted with water to obtain a 70 % (v/v) ethanol solution, and Orapi 2.6 % (v/v) sodium hypochlorite (NaOCl, Saint-Vulbas, France), were filtered at least eight times using Whatman GF/F 90 mm glass fiber filters with a 0.7 µm pore size (Freiburg im Breisgau, German) until no particle was observed, prior to sample analysis. The operators used nitrile gloves and lab coats.
Controls were still employed for each beverage:
- air control consists in placing a decontaminated GF/F filter in an open Petri dish under a laminar flow cabinet during the entire handling phase from degassing stage to filtration.
- negative control was a decontaminated Erlenmeyer flask containing filtered water only. Microplastics observed in these controls were then deduced from the samples, when they were identical in shape, color and polymer.
- positive controls were employed to estimate the recovery rates (Dehaut et al., 2023). Briefly, 25 known microplastics between 45 and 100 μm - made of PEHD, PS, PP, PET, and Polyester (PES) fiber (n = 5 for each polymer) - were added to the hydroxypropyl methylcellulose (HPMC) capsules. Capsules were then placed into decontaminated Erlenmeyer flasks containing filtered water and after filtration, MPs were tallied and recovery rates were determined using the following formula:
2.2. Sample selection
In June 2023, samples were bought from local resellers. They were categorized as follow: water, beer, wine, and soft drinks including colas, teas, and lemonades. The selection of a brand for each category was influenced by several factors, especially for the possibility of finding the drink in a wide variety of container types. Six samples were examined for every reference chosen. The samples were all taken from the same production batch.
2.2.1. Bottled waters
Different brands of bottled water were selected in different types of containers (plastic, glass and brick), from different origins (mineral water and spring water) and type of water (still and sparkling). The volumes of water analyzed ranged from 0.5 to 1.5 L, depending on the sample, and included five brands of mineral water and two brands of spring water (Table 1).
2.2.2. Soft drinks
Two brands of lemon-flavored water were analyzed: a mineral sparkling water in 0.2 L glass bottle and a spring still water in 1.5 L contained in plastic bottle (Table 1).
Three brands of cola, iced tea and lemonade were also studied in glass and plastic bottles as well as in cans with volumes ranging from 0.33 to 1.75 L with and without sweeteners for colas (Table 1).
2.2.3. Beers
Two brands of fruity, blond, amber and alcohol-free beers, sold in glass or cans, were examined. with volumes ranging from 0.25 to 0.75 L (Table 1). The glass bottles were available in two different sizes: small bottles (0.33 L) and large bottles (0.75 L).
2.2.4. Wine
Three types of wine were analyzed: red, white, and rosé wines packaged in bricks (0.25 L), cubitainers (3 L), glass bottles (0.75 L), and plastic bottles (1.5 L) (Table 1).
2.3. Degassing step
Preliminary analyses carried out prior to final sampling showed that the presence of carbon dioxide in a beverage led to degradation of the glass fiber filter during the filtration. To prevent this filter degradation, soft carbonated drinks, i.e. colas and lemonades, had to be degassed before analysis. For sparkling water, this degassing process was not necessary because the bubbles were finer and its pH is less acidic (around 6) than colas and lemonades (< 3). Degassing consisted in the samples opening and placing an aluminum foil to cover the bottle neck for a week in order to perform a first passive degassing process. Afterward, the contents were magnetically stirred for an hour at 100 rpm at room temperature, ca. 20–25°C. The bottles were always covered with aluminum foil. After these two degassing steps, the samples were ready for filtration. To check that this degassing process does not lead to air contamination, a negative control was performed. This control consisted of a bottle filled with filtered water undergoing the same process as the samples.
2.4. Sample filtration
Plastic bottle labels were first removed and all the bottles (plastic, glass, brick, can and cubitainer) cleaned with a pressurized air before being placed under the operating laminar flow hood.
Samples were then filtered with a Whatman GF/A glass fiber filter. To maximize particles recovery from the container walls and filtration equipment, the inside of the analyzed container, the inside of the funnel and the edges of the unclipped funnel in contact with the filter were then rinsed with water, 70 % (v/v) ethanol, and water. Except for the water and lemonades samples, 2 mL of 2.6 % (v/v) NaOCl was poured onto filters right after filtration, with system still under vacuum, to lighten them and ease their observations (Duman et al., 2023). The filter was then put in a 90 mm glass Petri dish and allowed to dry at room temperature, ca. 20–25°C, before observations.
2.5. Observation and identification
To count the particles based on their color and shape (fragment or fiber), the filters were examined using an Olympus SZX16 stereomicroscope (Rungis, France) equipped with a SDF PLAPO 1XPF lens and a UC90 camera. Olyvia software (3.3) and Cellsens (4.2) were then used to process the images. This processing consisted in counting particles, measuring their mean Ferret diameters and classifying them based on shapes, color and size classes. As no particle bigger than 500 µm was observed, and particles smaller than 30 µm were difficult to identify with µFT-IR, the selected size ranges were as follow [30 – 50 µm[, [50 – 100 µm[, [100 – 500 µm].
Once the particles were observed and counted, their nature (plastic, cellulose, mineral, etc.) and if applicable their polymer types (polyethylene (PE), polyester (PES), polypropylene (PP), etc.) were determined by microscopy coupled to Fourier Transform Infrared Spectroscopy (µFT-IR) using a Perkin Elmer Spectrum 3 – Spotlight™ 400 (Villebon-sur-Yvette, France) using the μ-ATR mode. To optimize analysis time, the particles analyzed were selected according to the abundance of particles identical in shape and color (Duman et al., 2023). Once the particles were selected, a screening of the area surrounding the particle was carried out to verify its presence and characteristics. A background was carried out before the particle was analyzed, then 25 acquisitions of the spectrum of the particle from 4000 to 600 cm−1 with a resolution of 4 cm−1 were made. The spectra obtained were then compared with several databases such as Flopp/Flopp-e (De Frond et al., 2021) and PerkinElmer (customized database), supplemented with Openspecy (Cowger et al., 2021). The identification of a particle was valid when the correspondence score between its spectrum and the database was greater than 0.7.
2.6. Caps contamination
Glass bottles corresponding to cola brand 2 (n = 10–0.33 L) were reused after being cleaned and muffled, and the water was filtered through GF/F filters until no particle was observed on the filters. Once ready, they were filled with filtered water and sealed with a new yellow cap using a soft-cover mallet and a hammer capper.
The experiment was carried out following three distinct scenarii: first the capsules were encapsulated directly without pre-treatment; secondly the capsules were blown out using an air bomb; and finally, the capsules were blown out and cleaned using filtered water, 70 % (v/v) ethanol and water. To avoid contamination, the hammer capper was rinsed with water, 70 % ethanol and water between each encapsulation. After capping, the bottles were turned upside down twice to recover any potential MP present under the capsule on the side in contact with the drink.
The bottles were then processed as described in 2.4, and then filters were observed under stereomicroscope to count yellow particles. Additionally, the solutions obtained after capsule rinsing were collected for each capsule and analyzed in order to count the number of yellow particles.
2.7. Statistics analysis
Results for contamination are expressed as a level of microplastic that is a number of microplastics per liter (MPs/L). Except explicit mention, levels are represented as the mean ± 2 error type (2 s.e) of the six analyzed replicates, in order to visualize the 95 % confidence interval of the mean (IC95). In order to study the significance of the observed differences, a Shapiro-Wilks’s test was first performed to determine the normality of the distribution. As data were not normally distributed (p-value < 0.001, Shapiro-Wilks test), a Kruskal-Wallis test (KW) was performed to highlight the significant differences in microplastic levels within the container and brand groups. When a significant difference (p-value < 0.05) was noticed within a group, a pairwise comparison test was performed with a Dunn-Bonferroni post-hoc test (DB). All data processing and statistical tests were performed using R software, version 4.4.1.
3. Results
This study aimed to investigate the level of microplastic contamination in various beverages (water, colas, tea, lemonade, beer and wine) sold in France because no previous studies have been conducted on water or other beverages in this country. In addition, the study examined the impact of different types of packaging (plastic, glass, brick, can and cubitainer), as other studies have shown that the container can influence the level of microplastic contamination.
3.1. QA/QC
All the results presented below are extrapolated data according to the rules explained above (cf 2.5). The MPs observed in the controls (air and negative controls) were subtracted from the results on the beverages according to their form (fiber or fragment), color and polymer. From the 13 batch analyzed, an average of 0.9 ± 1.4 MPs (mean ± s.d) were counted for the negative and air control. As well, a calculated recovery rate of 85.5 ± 7.7 % (mean ± s.d) was found for the positive controls.
3.2. MPs shape
There were only two types of MPs found in all beverages (water, soft drinks, beer, and wine): fibers and fragments. Additionally, 96.9 % of MPs in waters, 93 % in colas, 93.4 % in teas, 96.1 % in lemonades, 92.7 % in beers, and 72.9 % in wines were in fragment form.
3.3. Polymers
Polymers were grouped into polymeric clusters (Fig. 1): the polyester cluster (PET and Alkyd lacquers), the polyolefin cluster (PE, PP, Poly(ethylene-co-vinyl acetate) (PEVA) and Ethylene vinyl acetate (EVA)), the polymethacrylic cluster (poly(methyl methacrylate) (PMMA) and Polyacrylonitrile (PAN)), the polyamide cluster (polyamide (PA) and Nylon), the styrenic cluster (polystyrene (PS), Acrylonitrile butadiene styrene (ABS) and SR) and the polyvinyl cluster (PVC).The polyester cluster was more prevalent in glass containers than in other types of containers for all beverages, but especially for teas (26.3 MPs/L), lemonades (28.5 MPs/L), colas (33.2 MPs/L), and beers (95.9 MPs/L).
Although polyolefin was less common in plastic bottles - beer containing a maximum concentration of 1.3 MPs/L for polyolefins - they remained a prevalent polymer class found in all containers. However, their content was higher in glass with up to 24.7 MPs/L for polyolefins.
3.4. Bottled waters
3.4.1. MPs levels
Across all water samples, the global average concentration was 2.9 ± 0.7 MPs/L. First, measured levels of MPs varied according to the type of packaging, with significant differences in MPs levels between packaging (KW p-value < 0.001). The MP content in glass bottles was significantly higher than plastic containers only (DB p-value < 0.001), with respectively 4.5 ± 1.2 MPs/L and 1.6 ± 1.7 MPs/L (Fig. 2 A). For brick water, the MPs level is neither different from the one in glass containers nor from plastic bottles, but it should be noted that only two brands were analyzed.
Further investigations of the origin waters revealed that mineral waters tended to reach higher level of MPs (3.7 ± 1.0 MPs/L) compared to spring waters (1.6 ± 0.6 MPs/L) with a significant difference (DB p-value < 0.01) (Supplementary material Figure S1 A).
The results observed in two types of water i.e., sparkling and still; also demonstrated variations in MP contamination levels. Sparkling water had, on average, higher MP level with 3.4 ± 1.0 MPs/L, compared to still water with 2.4 ± 0.9 MPs/L, but there are no significant difference (Supplementary material Figure S1 B).
There were variations in the levels of contamination among the brands, caused first by heterogeneous data within the same brand, as data from different containers have been grouped together. However, brand W3 showing the highest concentration of MPs (5.0 ± 2.0 MPs/L), followed by brand W2 (4.1 ± 1.8 MPs/L), brand W1 (2.1 ± 1.6 MPs/L), brand W6 (1.7 ± 1.1 MPs/L), and brand W7 (1.5 ± 0.7 MPs/L). The two brick brands, W4 and W5, had the same MPs contamination level of 3.0 ± 2.5 MPs/L and 3.0 ± 2.3 MPs/L respectively (Fig. 2 B). No significant difference was detected between the different brands.
3.4.2. MPs size
There was minimal variation in the size proportions among the various water containers, origins, and forms, with a range from 18.5 % to 23.9 % of MPs falling between 30 and 50 µm, from 37.1 % to 44.6 % between 50 and 100 µm, and a range from 35.3 % to 41.4 % between 100 and 500 µm (Supplementary material Figure S2 A, B, C).
However, notable variations were observed across brands: the majority of MPs in brand 1 and brand 6 (54.2 % and 58.6 %) measured between 100 and 500 µm, while most MPs in brand 7 (52.8 %) were between 50 and 100 µm (Supplementary material Figure S2 D).
3.5. Lemon flavored water
3.5.1. MPs levels
Two brands of lemon flavored water (LW1 and LW2) were studied, and the MPs level in these waters was 8.6 ± 5.3 MPs/L. With regard to the difference between brands, packaging and origins, the quantity of MP is greater in brand LW1, which is mineral water in glass bottles (15 ± 9.5 MPs/L) than in brand LW2, which is spring water in plastic bottles (2.3 ± 1.4 MPs/L) (Supplementary material Figure S3), with a significant difference (KW p value < 0.01).
3.5.2. MPs size
Regarding the size distribution for the two lemon flavored water brands. The proportions of the three size classes for brand LW1, which is mineral water in glass bottle, are almost identical, with 27.8 % [30–50 µm[, 33.3 % [50–100 µm[ and 38.9 % [100 – 500 µm[. However, for brand LW2, a spring water in plastic bottle, the proportion of large MPs ([100–500 µm[) rises to 75.7 % (Supplementary material Figure S4).
3.6. Colas
3.6.1. MPs levels
The MPs content was evaluated for three cola brands (C1, C2 and C3) with and without sweeteners. These colas were found to have an average MP contamination level of 31.4 ± 16.0 MPs/L.
These results for colas highlighted the effect of containers on MP contamination levels, with a significant difference in the packaging group (KW p-value < 0.001). The most contaminated colas were those from glass bottles with 103.4 ± 44.1 MPs/L, compared with cans (3.4 ± 2.1 MPs/L) or plastic (2.1 ± 0.7 MPs/L) ones. (Fig. 2 C). The level of MPs was significantly different between glass and plastic bottles cola (DB p-value < 0.001) as well as between glass and can (DB p-value < 0.001). Contrarywise, no significant difference was observed between plastic containers and cans.
Furthermore, colas with sweeteners had an average content of 14.3 ± 6.2 MPs/L, while those without sweeteners had an average of 48.5 ± 30.6 MPs/L across all container types (Supplementary material Figure S5).
There was a significant variation (KW p-value < 0.05) in the contamination levels among the brands of colas; brand C2 displayed the highest level of MPs (76.0 ± 49.3 MPs/L), followed by brand C1 (20.6 ± 10.8 MPs/L) and brand C3 (3.0 ± 2.0 MPs/L) with the lowest level of MPs (Fig. 2 D). The level of microplastics differed significantly between brand C2 and C3 (DB p-value < 0.01).
3.6.2. MPs size
Among the MPs from canned colas, the majority (51.4 %) measured between 50 and 100 µm, while smaller MPs (30 and 50 µm) accounted for just 2.9 %. On the one hand in glass bottles, the majority of MPs were also between 50 and 100 µm (51.2 %). Plastic bottles, on the other hands, contained mostly larger MPs, with 52.4 % in the 100–500 µm range (Supplementary material Figure S6 A).
Analysis of MP sizes for the two varieties of colas revealed that the proportion of MPs between 100 and 500 µm tend to be higher with sweetener (41.7 %) than without (15 %) (Supplementary material Figure S6 B).
In terms of brands, the proportions of the different MP size classes for brands C1 and C2 were almost nearly identical, with the exception of minor increase in the [50–100 µm] class for C2. In contrast, smaller MPs (30–50 µm) accounted for just 5.4 % of MPs for brand C3, a decrease compared to 25.2 % for C1 and 33.3 % for C2 (Supplementary material Figure S6 C).
3.7. Cold tea
3.7.1. MPs levels
Three brands (T1, T2, and T3) of cold tea were chosen for the study. According to the findings, cold teas have an average of 28.5 ± 13.1 MPs/L. Cold tea contamination level varied according to the type of container, with a significant difference in the packaging group (KW p-value < 0.001). Tea from cans contained 16.3 ± 3.9 MPs/L, while that from plastic bottles had 2.2 ± 1.0 MPs/L. These teas were less contaminated than those stored in glass bottles, which averaged 86.3 ± 35.3 MPs/L (Fig. 2 E). The level of MPs was significantly different between glass and plastic, plastic and can (DB p-value < 0.001), as well as between can and glass (DB p-value < 0.05).
These results for MPs contamination also showed that there was no significant brand dependency. Brand T2 displayed a higher MP level (51.4 ± 31.1 MPs/L), while brand T1 and brand T3 contained 19.3 ± 9.6 and 8.1 ± 3.9 MPs/L, respectively (Fig. 2 F).
3.7.2. MPs size
The percentage of each size class of MPs present in iced tea were identical for all packaging, with the exception of plastic bottle, where the 50–100 µm size class predominates (54.5 %) (Supplementary material Figure S7 A).
There was little difference between the 1 and 2 brands. Compared to brand 1 and 2, which had respective proportions of 38 % and 39.1 % for the 50–100 µm class, brand 3 had a higher proportion of 51.1 % in this range (Supplementary material Figure S7 B).
3.8. Lemonades
3.8.1. MPs levels
Three distinct brands of lemonades (L1, L2, and L3) were analyzed. For all containers combined, the average MPs content in lemonades was estimated to be 45.2 ± 21.4 MPs/L. Contamination levels varied from one container to another, with a significant difference in the packaging group (KW p-value < 0.001). Lemonades in glass bottles had significantly higher levels of MPs (111.6 ± 41.1 MPs/L) than those in plastic bottles (1.5 ± 0.7 MPs/L) or can (10.9 ± 4.6 MPs/L) (Fig. 2G). The level of MPs was significantly different between glass and plastic bottles (DB p-value < 0.001), between plastic and can (DB p-value < 0.01) and between can and glass (DB p-value < 0.05).
In addition, there was no significant variation in the MP contamination between the different lemonade brands that ranges from 29.3 ± 18.0 (brand L2) to 66.6 ± 44.1 (brand L1) (Fig. 2 H).
3.8.2. MPs size
It appeared that the proportions of the different size classes of MPs from lemonade cans and plastic bottles were comparable, with most MPs between 100 and 500 µm (50 % and 42.4 %, respectively) (Supplementary material Figure S8 A). For glass, the proportion of small MPs (30–50 µm) was higher (36.4 %) than for other containers, while the proportion of large MPs (100–500 µm) falled to 21.1 %. However, the percentage of MPs between 100 and 500 µm varied according to brand, with a proportion of 15.4 % for brand 1, 21.1 % for brand 2, and 29.5 % for brand 3 (Supplementary material Figure S6 B).
3.9. Beer
3.9.1. MPs levels
Two beer brands (B1 and B2) were studied including different containers and types. According to the results, beers had an average MP level of 82.9 ± 13.9 MPs/L.
Analysis of MP contamination levels revealed a significant difference between the containers (KW p-value < 0.001) and that small glass bottles contained a higher level of MP compared to other containers (133.7 ± 15.9 MPs/L). Contamination levels in the other two containers cans and large glass bottles were comparable with 31.8 ± 17.3 MPs/L and 32.8 ± 12.2 MPs/L, respectively (Fig. 2 I). The levels of MPs were significantly different between small glass and large glass bottles (DB p-value < 0.001) and between small glass and can (DB p-value < 0.001).
For the different types of beer, no significant difference in the level of MPs contamination was observed. Amber beers have the highest average level of MPs (117.8 ± 44.3 MPs/L), while fruity beers have the lowest level (69.8 ± 26.1 MPs/L). The levels of contamination for the other three types of beer - Alcohol-free, Alcohol-free fruity, and Blond - are 74.8 ± 29.6 MPs/L, 94.0 ± 37.3 MPs/L, and 78.9 ± 22.1 MPs/L, respectively (Supplementary material Figure S9).
In conclusion, there was no significant variation in contamination levels between the two brands: brand B1 and B2 contained 79.9 ± 20.4 MPs/L and 85.9 ± 19.1 MPs/L respectively (Fig. 2 J).
3.9.2. MPs size
The results showed no difference in MP size distribution for beers from large glass bottles and cans. Conversely, large glass bottles (42.6 %) and cans (44.1 %) had slightly fewer MPs in the 50–100 µm range than small glass bottles (47.9 %) (Supplementary material Figure S10 A).
Equal amounts of each particle class were found in both fruity and non-alcoholic beer varieties, with the majority of MPs (46.2 % and 43.3 %) between 50 and 100 µm. The percentage of small MPs (30–50 µm) reached 27.6 %, 28.2 %, and 28.9 % for the other three types of beer, respectively (Supplementary material Figure S10 B).
Lastly, a difference in particle size distribution was observed between the brands. On the one hand, brand 1 had a higher percentage of MPs in the 100–500 µm range (37.7 %) compared to brand 2 (21.2 %). On the other hand, brand 1 also exhibited a higher proportion of smaller MPs, with 30.8 % in the 30–50 µm range, while brand 2 contained 18.5 % of this size range (Supplementary material Figure S10 C).
3.10. Wine
3.10.1. MPs levels
Analyses were performed to determine the levels of MP contamination in white wines (WW), sparkling white wines (SWW), rosé wines (RoW), and red wines (RW). The impact of packaging was also investigated by selecting wines packaged in different types of containers including plastic, glass, brick, and cubitainer. On average, 8.2 ± 3.3 MPs/L were found in the wines.
Similarly, to other drinks, the MPs content varied depending on the container, with significant difference in the packaging group (KW p-value < 0.001). The level of MPs contamination in large plastic bottles, glass, and cubitainer were 2.1 ± 0.8 MPs/L, 5.3 ± 2.0 MPs/L, and 3.7 ± 0.9 MPs/L, respectively. Small plastic bottles showed a slight but not significant increase in MPs (8.7 ± 3.8 MPs/L). Nonetheless, the wine found in bricks had the highest MPs content (30.0 ± 16.9 MPs/L) (Fig. 2 K). The level of MPs was significantly different between brick and large plastic bottles (DB p-value < 0.001), between brick and glass and between small plastic and large plastic (DB p-value < 0.05), and finally between brick and cubitainer (DB p-value < 0.01).
A significant difference in contamination levels was also observed between the different types of wine (KW p-value < 0.05). The average content of MPs/L was 12.0 ± 8.0 for white wine (WW), 5.8 ± 1.9 for sparkling white (SWW), 8.5 ± 4.0 for red wine (RW), and 2.6 ± 0.9 for rosé wine (RoW) (Fig. 2 L). The level of microplastics differed significantly between rosé wine (RoW) and red wine (RW) (DB p-value < 0.05).
3.10.2. MPs size
Similar size proportions of MPs were found in wine packaged in bricks, glass, small and large plastic bottles, with an average of 39.5 % from 30 to 50 µm, 36.2 % from 50 to 100 µm, and 24.2 % from 100 to 500 µm. A higher proportion of large MPs (100–500 µm) - 45.2 % - were found in cubitainers (Supplementary material Figure S11 A).
The percentages appeared to be the same for all four types of wine (red, rosé, sparkling white, and white): 31.3 % from 30 to 50 µm, 33 % from 50 to 100 µm, and 35.6 % from 100–500 µm. The only exception was for sparkling white wine, which had fewer large MPs (14.3 % of 100–500 µm) (Supplementary material Figure S11 B).
3.11. Caps contamination
The results show that glass containers were more contaminated than other packaging for all beverages except wine, because wine bottles were closed with cork stoppers rather than metal caps.
It was noticed that most of the microplastics isolated from glass bottles had the same color as the paint on the outer layer of the cap. FTIR analysis of the paint on the metal cap revealed that it was mainly composed of polyester, like the particles isolated from glass bottles, which mainly belong to the polyester class. Therefore, it was hypothesized that these particles could originate from the cap.
First, the inside and outside of the capsules were also observed under binocular magnifying glass before use, and scratches were found on the outer surface, as well as pieces of the capsule paint adsorbed to the inside one (Fig. 3).
Then experiments were carried out to confirm the potential origin of colored particles within beverages, as well as to purpose a simple method to lower contaminations by caps. It was shown that when the caps were not pre-cleaned, 287.3 ± 81.4 MPs/L were found (Fig. 4). This MPs levels in the bottles significantly decreased (KW p-value < 0.001 - DB p-value < 0.001) when they were blown prior to encapsulation, with 105.8 ± 32.1 MPs/L, and reached to 86.7 ± 42.3 MPs/L when the caps were blown and rinsed beforehand. Interestingly, the analysis of the rinsing solution, water/ethanol/water mixture, contained an average of 47.8 ± 12.6 yellow particles per rinsed cap.
4. Discussion
4.1. Comparison with previous studies
The number of studies on the MP contamination of various beverages, and the influence of packaging on this contamination has steadily grown in recent years. Soft drinks and water have been the focus of most studies. The levels of MPs in each beverage varies considerably from one study to another (Table 2).Regardless of the type of container, various studies have shown that the MPs content for bottled water ranged from 4 to 87.9 ± 46.4 MPs/L (Table 2). In comparison, the bottled water in this study exhibited a relatively low concentration (2.9 ± 0.7 MPs/L).
Soft drinks are sometimes referred to fruit or carbonated drinks in other studies, but since these terms do not always specify the exact type of beverage, making comparisons becomes even more challenging. The contents of MPs in this study (colas: 31.4 ± 69.7 MPs/L and lemonades: 101.5 ± 123.7 MPs/L) were similar to previous findings, ranging from 1 to 49.3 ± 54.5 MPs/L (Crosta et al., 2023, Shruti et al., 2020). In addition, previous studies in bottled iced teas reported MPs concentrations from 0 to 36.4 ± 36.9 MPs/L (Pham et al., 2023, Shruti et al., 2020), which is comparable to the content of MPs in this study (14.6 ± 43.2 MPs/L).
Regarding alcoholic beverages, several studies have assessed the presence of MPs, particularly in beer. The MPs content in these studies widely varied, ranging from 0 to 95.5 ± 92.0 MPs/L (Kosuth et al., 2018, Liebezeit and Liebezeit, 2014, Pham et al., 2023, Socas-Hernández et al., 2024, Wiesheu et al., 2016), which is quite similar to the content found in this study (84.0 ± 5.7 MPs/L).
Moreover, a previous investigation on the presence of MPs in white wine in glass bottles reported an average level of 183 ± 123 MPs/L (Prata et al., 2020). This is approximately 15 times higher than the MPs found in this study (12.0 ± 18.1 MPs/L). Compared with the present results (8.5 ± 6.1 MPs/L in red wine, 2.6 ± 0.2 MPs/L in rosé wine, and 12.0 ± 18.1 MPs/L in white wine), a more recent study found 32.6 ± 11.2 MPs/L, 4.0 MPs/L and 32.8 ± 37.1 MPs/L respectively (Socas-Hernández et al., 2024).
However, comparing studies on microplastics (MPs) is challenging due to differences in their methodologies, particle size considered, sample types and volume studied, and the countries where the studies were conducted. The absence of standardized methods for MP analysis is a significant issue, leading researchers to adopt different analytical strategies. For instance, some studies dry filters containing MPs at temperatures of 60°C or higher. However, research has shown that certain polymers, such as polyester fibers, can degrade and change morphology at temperatures above 40°C (Karami et al., 2017). This degradation can result in an underestimation of contamination levels. The number of samples analyzed is also an important parameter to consider. To obtain reliable and comprehensive data, it is crucial to analyze as many samples as possible, ideally from various brands or batches. Without global standards, insufficient sample sizes can lead to inaccuracies, either over- or underestimating MP contamination.
5. Conclusion
This study is the first to assess microplastic contamination in various beverages sold in France. It showed that MPs were present in all the samples analyzed, with contamination levels heterogeneous depending on beverages, with a size between 30 and 500 µm. In addition, the comparative study of different containers showed that the container has an impact on this MP contamination. Counterintuitively, drinks sold in glass bottles were more contaminated by MPs. Experiments have shown that these MPs originate from the exterior paint of capsules. A cleaning step before encapsulation can significantly reduce beverages’ contaminations. However, cleaning does not remove all the MPs from the capsule.
