Abstract

Full-text original study online at
https://www.sciencedirect.com/science/article/pii/S2949824426006208

 

Tea contains the trace element fluorine in the form of fluoride, which can be beneficial or potentially harmful depending on the dose ingested, making accurate measurement in food essential. This study examined various methods for analysing fluorine in tea infusions. Combustion ion chromatography CIC and nICPMS -inductively coupled plasma mass spectrometry (detecting all fluorine containing compounds as a negative ion) provided comparable concentrations for fluorine in tea infusions. ISE results depended on the complexing agent used in the TISAB solution. An EDTA-containing buffer yielded lower fluoride concentrations than buffers containing citrate or CDTA, consistent with incomplete release of complexed fluoride under these conditions. Within the limited set of pooled samples analysed in this study, the organic tea pools contained approximately 40% less fluorine than the conventional tea pools. However, the underlying cause could not be determined from the present study design. The maximum estimated mean and P95 exposure through tea consumption for the general German population was 8% and 50% of the safe level of intake for adults aged > 65 years and over (conventional tea), respectively.

Keywords: prepared tea; occurrence data; Combustion ion chromatography; inductively coupled plasma mass spectrometry; fluoride-selective electrode

1. Introduction

Fluorine is a trace element and contributes to human health particularly regarding dental and skeletal health. It binds as fluoride to hydroxyapatite in teeth, resulting in enamel with higher acid resistance. This reduces caries development and contributes to oral health (EFSA-NDA, 2013). Besides being health beneficial, too high fluoride exposure, especially during tooth development, leads to fluorosis. Fluorosis includes symptoms ranging from opaque white lesions in the teeth, which only have aesthetic consequences (EFSA-NDA, 2013), to damaged enamel in severe cases (EFSA-NDA, 2013, Zohoori and Buzalaf, 2022). Chronic high fluoride intake can also lead to skeletal fluorosis (Ghosh et al., 2015). An acute excessive fluoride intake can lead to gastric and kidney disturbances (EFSA-NDA, 2013).

Taking these implications into account, the European Food Safety Authority (EFSA) established an adequate intake (AI) level of 0.05 mg/kg body weight per day in 2013. In the recently published opinion, a safe level of intake (SLI) was established at 3.3 mg/day for adults and children over 8 years as well as a tolerable upper intake level (UL) at 1.0 to 2.0 mg/day for children up to 8 years (EFSA, 2025). The German Society of Nutrition (DGE) recommends for adequate total fluoride intake during childhood between 0.25 mg/day for infants to 3.0 mg/day and 3.5 mg/day for female respectively male adolescents. The recommendations for women are 3 mg/day (including pregnant and breastfeeding women) and for men 3.5 mg/day (DGE, 2025).

For the determination of fluoride in food stuff a fluoride-selective electrode (ISE) or ion chromatography with conductivity detection (IC) can be used (Martínez-Mier et al., 2011). With both methods, only the amount of fluorine present in form of free fluoride anion can be determined. Combustion ion chromatography (CIC) is a variety of IC which includes on-line sample combustion (Miyake et al., 2007). In this case fluorine present in whatever molecular form in the sample is hydrolyzed to fluoride and detected by a conductivity detector as fluoride. This allows an estimation of the total fluorine content. A more recent development is the direct determination of fluorine by inductively-coupled plasma mass spectrometry (ICPMS) (Guo et al., 2017, Raab et al., 2025, Jamari et al., 2017). With commercial triple-quadrupole instruments fluorine can be determined using additional barium for the formation of BaF+ in the plasma followed by detection as positive ion (pICPMS/MS) (Jamari et al., 2018, Guo et al., 2017). With an ICPMS prototype capable of operating in negative ion mode direct determination of fluorine is possible (nICPMS) (Raab et al., 2025).

Fluoride might be added to salt and dental care products. Natural sources of fluoride in our nutrition are particularly water, tea and marine fish (EFSA-NDA, 2013). Especially tea is known to concentrate high levels of fluoride through bio-accumulation. The fluoride concentration of tea depends on various factors, including growing conditions, leaf maturity, and brewing practices (Ma?yszek et al., 2025). Fluoride concentrations of up to 6 mg/L have been reported in black tea infusions from tea bags. Infusions made from brick tea (black tea leaves) can reach fluoride levels of up to 10 mg/L (Jayasingha et al., 2025). The consumption of tea worldwide was estimated to be 7.3 billion kg in 2023 and is prognosed to increase further (Statista, 2025). As tea can have an important impact on the exposure to fluoride and with its estimated increasing consumption over the next years, the importance of knowing the exact concentration of fluoride in tea is growing.

As part of the BfR MEAL Study (“Mahlzeiten für die Expositionsschätzung und Analytik von Lebensmitteln”; meals for exposure estimation and analysis of foods) conducted at the Federal Institute of Risk Assessment (BfR) in Germany, foods are collected, prepared and pooled in a total diet study (TDS) design. The aim is to deliver an extensive data set of levels of substances in foods for later exposure assessment. These foods are collected and prepared representatively for the German household behavior. Levels of other elements are published elsewhere (Stadion et al., 2024, Fechner et al., 2022, Kolbaum et al., 2023).

To determine the levels of total fluorine and specifically fluoride in tea infusion, samples were prepared in the MEAL study kitchen and measured at the University of Graz with three different methods, including ion selective electrode (ISE) with different TISAB buffers, combustion ion chromatography (CIC) and negative ion inductively coupled plasma mass spectrometer (nICPMS). The analytical objective was to determine whether techniques based on fundamentally different measurement principles provide consistent fluorine and fluoride concentrations in a complex food matrix. Particular emphasis was placed on the application of nICPMS to representative pooled tea infusions and on evaluating the influence of TISAB composition on ISE results. The resulting concentration data were subsequently used to estimate dietary fluoride exposure through tea consumption in the German population, while considering method-specific and sampling-related uncertainties.

2. Material and methods

2.1. Chemicals

Deionized water (Milli-Q Reference A+, Merck Germany) was used throughout the experiments. All chemicals used were at least p.a grade and used without further purification. Ethylenediaminetetraacetic acid (disodium-EDTA dihydrate, > 99%, Carl Roth GmbH, Germany), acetic acid (glacial, NORMAPUR, VWR Chemicals, Austria), sodium acetate (Merck, Germany), sodium chloride (>99%, Sigma-Aldrich, Germany), sodium hydroxide (pellets, Fisher Scientific GmbH, Austria), trans-1,2-Diaminocyclohexane-N,N,N’,N’-tetraacetic acid (CDTA, Titriplex IV, Merck, Germany), tri-sodium citrate dihydrate (Carl Roth GmbH, Germany) and ammonia (25%, Chem-Lab nv, Belgium) were used. A 1000 mg/L fluoride (sodium fluoride) ion chromatography standard from Roth was used throughout. The KOH-solution (Dionex EGC III KOH RFIC Eluent Generator Cartridge) used for CIC was ready prepared from Thermo Scientific (Germany). The gases were at least grade 5.0 from Messer (Austria). The sample vials used were PP-vials from Greiner Bio One (Austria), for IC the PP-vials were from BGB (Austria).

2.2. Preparation of tea infusion

The design of the BfR MEAL Study is described elsewhere (Sarvan et al., 2017). Following international recommendations for TDSs (EFSA/FAO/WHO, 2011), foods are prepared as consumed in Germany and referred to as MEAL foods.

MEAL foods discussed in this publication were purchased in the area of Berlin (Germany) in 2022. Each tea sample analysed consisted of 6 to 15 subsamples, selected according to market share data from a representative household panel and the consumption survey NVS ll (National Nutrition Survey II). Both data sets were used to select brands, purchasing location (e.g., supermarket, discounter, specialty shop), product type (green or black tea) and delivery form (loose or in a tea bag). The NVS ll was conducted in 2006 and covers age groups 14 to < 80 years and includes 19,329 subjects (Heuer et al., 2015). To close knowledge gaps and to update data, market share data were used from a representative panel of 30,000 households over a 12-month period (between 2015 and 2017) (Sarvan et al., 2017).

To mimic consumer behavior, the subsamples were prepared as beverages. Information on the packaging was used to brew the tea ready for consumption. Water was boiled before preparation and used directly or after cooling down to 70 – 90 °C. On average, 9.3 g loose tea or five tea bags were used per 1 L hot water and infused between 3 to 4 minutes. Tea bags contained 1.75 g (green tea) or 1.4 to 2.0 g (black tea) of tea. For the preparation of teas, tap water from the MEAL study kitchen was used. This tap water was also used as blank sample for fluorine determination. For the preparation, typical kitchen utensils were used made of different materials (Hackethal et al., 2023). If stated on the packaging of the teas, the teapot or cup was preheated before brewing. After preparation, 15 subsamples (4 green tea and 11 black tea subsamples) were combined to one food pool, using a Grindomix GM300 blender (Retsch GmbH, Haan, Germany). The pool “Tea bags only” consisted only of tea prepared from tea bags (2 green tea and 4 black tea subsamples). To reflect consumer behavior, all other pools contained 6 subsamples prepared with tea bags and 9 subsamples prepared with loose tea in different ways (tea filter, tea strainer, brewed loose). With this study design the latter pools can be used in a later exposure assessment as they follow the TDS design. The pool “Tea bags only” was sampled for comparison on total fluorine and on fluoride levels.

All preparation methods were conducted for conventionally and organically produced tea. In total 10 pools were analyzed in duplicate. Samples were stored in 100 mL polypropylene (PP) vessels at -20º C before analysis.

2.3. Sample preparation for ISE, CIC and nICPMS

No further sample preparation for CIC and ISE measurements took place. Samples determined by nICPMS were diluted with deionized water 1 + 9 and for standard addition 0.3 and 0.6 mg /L fluoride were spiked. Samples for nICPMS were diluted to reduce the effect of chloride present in the tea extracts prepared with tap water.

2.4. Combustion ion chromatography (CIC)

The CIC system composed of an ASC-270 SL autosampler, HF-210 combustion unit, GA-211 gas absorption unit (a1 Envirosciences GmbH, Germany) and a Dionex Aquion RFIC chromatographic unit (ThermoFisher, Austria) was used with a KOH-gradient. The combustion furnace temperature was 1000 °C (inlet) and 1100 °C (outlet) with Ar (200 mL/min) and O2 (400 mL/min). Details of the combustion program can be found in Table S1.

For the ion chromatography part of the CIC, a Dionex IonPac AS20 (2×250 mm) column was used. KOH buffer was used as mobile phase at a flow rate of 0.3 mL/min using the following gradient: 0 – 4 min: 3 mM KOH, increased linear within 6 min to 30 mM KOH and reset to 3 mM KOH within 1 min. The column was re-equilibrated for a further 14 min. The sample volume was 100 µL.

The calibration for fluoride ranged from 0.2 to 10 mg/L. All tea samples were measured in duplicates without any dilution step. Two samples were spiked with fluoride standard at 5 mg/L to calculate recovery rates.

2.5. Ion selective electrode (ISE)

Fluoride activity (equivalent to free fluoride present in the sample) was determined by ISE. The electrode used was an Orion Ionplus sure-flow /Fluoride electrode from Thermo Scientific (Germany) with integrated reference electrode. The voltameter Orion 5 Star was from Thermo Scientific (Germany). The samples were diluted 1 + 1 with TISAB solution. The solutions were mixed well and then measured. The calibration ranged from 0.1 to 100 mg/L fluoride. The samples were measured in triplicates. Two samples were spiked with fluoride standard at 1 mg/L to calculate recovery rates. To analyze the influence of different TISAB solutions on the fluoride activity in tea infusion, five different recipes of TISAB (Table S2) were prepared according to Forsido et al. (Forsido & Ndibewu, 2021) and the results were compared.

2.6. Inductively coupled plasma mass spectrometer (nICPMS)

The nICPMS used was a modified NexION 2000 system from Perkin Elmer/Canada as described elsewhere (Raab et al., 2025). Briefly the detector and associated electronic components were modified to allow detection of negative ions. Potential fluorine-containing instrumental compounds, like tubing and grease, were replaced to reduce background contamination. The instrument was tuned for highest possible sensitivity and selectivity on m/z 19 (fluorine). All samples were measured using standard addition with two spikes of 0.3 and 0.6 mg /L fluoride.

2.7. Method of fluoride exposure assessment through tea consumption

To simplify reading the text the term fluoride is used in the following text to mean both anionic fluoride and total fluorine. In tea fluoride represents the majority of fluorine present.

2.7.1. Consumption data

The dietary exposure assessment for tea consumption in young children was based on the KiESEL study (Children’s Nutrition Survey to Record Food Consumption) (Golsong et al., 2017, Nowak et al., 2022). KiESEL was conducted between 2014 and 2017 by the BfR and was a module of KiGGS Wave 2 (German Health Interview and Examination Survey for Children and Adolescents) carried out by the Robert Koch Institute (RKI). The KiESEL study included 1,008 children aged between 0.5 and < 6 years. Data on tea infusion consumption was collected via weighed food records over three consecutive days and one separate day. For the present analysis, data from 952 non-breastfed children were included.

Data from the EsKiMo II study (Eating Study as a KiGGS Module; a module in KiGGS Wave 2) were used to estimate tea consumption among children aged 6 to < 12 years. EsKiMo II was conducted between 2015 and 2017. Dietary intake was documented for 1,190 children, for whom weight records were kept for three consecutive days and one separate day. Further details of the EsKiMo II study can be found elsewhere (Brettschneider et al., 2018).

Consumption data for the adult population aged 14 to 80 years were obtained from NVS II. NVS II was carried out between 2005 and 2006. For the present study, 24-hour recalls from 13,926 individuals on two non-consecutive days were included. Further details on the NVS II are provided by Krems et al. (Krems et al., 2006) and MRI (MRI, 2008a, MRI, 2008b).

Body weights were reported at individual level in all these studies.

2.7.2. Data treatment

The fluoride exposure assessment for tea consumption was based on the fluoride concentrations as shown in Table S3 (see Supplementary material). Samples containing only tea bags (A141300825 and A141700725) were excluded from the exposure assessment as they do not reflect the typical tea consumption in the German population. The mean fluoride concentrations for each preparation method were calculated by averaging the results from the three analytical methods (ISE, CIC and nICPMS). This resulted in individual fluoride concentrations for each preparation method. Concentrations per preparation method were then averaged to give an overall mean fluoride concentration for the preparation methods (loose tea in a tea filter, loose tea in a tea strainer and loose tea brewed loose). The mean fluoride concentration was determined separately for conventionally and organically produced tea. The final fluoride levels considered in the exposure assessment were 1.36 mg/L for organic tea and 2.22 mg/L for conventional tea.

2.7.3. Dietary exposure assessment

To calculate long-term fluoride exposure from tea consumption, the mean fluoride concentration of the tea infusion was related to individual long-term tea consumption. Individual long-term tea consumption was determined by averaging the reported consumption over all survey days. Two separate exposure distributions were generated for conventionally and organically produced tea by multiplying long-term consumption by the respective fluoride concentration of the tea infusion.

To describe fluoride exposure from tea consumption for average and high consumers, the mean exposure, the median (50th percentile) and the 95th percentile are presented.

For risk characterization, fluoride exposure from tea consumption was compared with ULs of 1.0, 1.6 and 2.0 mg/day for children under 1 year old, aged 1–3 years, and aged 4–8 years, respectively. For adults and children over 8 years, a SLI of 3.3 mg/day was applied (EFSA, 2025).

All analyses were conducted separately for the general German population and for participants who reported consuming tea on at least one day (tea consumers) during the survey period.

All analyses were conducted using the R software (version 4.3.2; R Foundation for Statistical Computing, Vienna, Austria).

3. Results and discussion

3.1. Quality control

One of the major problems when quantifying fluorine in any kind of sample matrix is the lack of a real reference method and very often the lack of certified reference materials suitable for quality control. One possibility to increase confidence in the determined results is to use as many different independent methods as possible. When the results for all methods agree the likelihood that the results are correct is high.

Tea (leaves of Camellia sinensis) is a matrix quite often studied for its fluoride content (Morés et al., 2011, Horie et al., 1992, Akhdhar et al., 2021, Guo et al., 2020), since the plant is known to accumulate high amount of fluoride in its leaves (Zhang et al., 2023). Despite this there are very few reference materials for fluorine in tea available and none for tea infusions.

3.1.1. Limit of detection

The detection limits of all three techniques were in the same order of magnitude. The limit of detection was estimated via external calibration in all cases. For CIC the limit of detection was 0.3 mg /L tea infusion, for ISE 0.2 mg /L tea infusion and for nICPMS 0.1 mg / L tea infusion.

3.1.2. Spike recovery for CIC and ISE

Tea samples (n=2) were spiked with 5 mg/L fluoride standard for the calculation of spike recovery for CIC and 1 mg/L for ISE. The spike recovery of the CIC experiment was 99 respectively 111% and for the ISE 94 resp. 102%. For both methods together the spike recovery was on average 102 ± 5%. Since nICPMS measurements were done using standard addition no spike recovery experiments were done there. Within-day and between day variation of CIC and ISE were less than 2%, for nICPMS these were less than 10%.

3.2. Fluoride content in different tea infusions

Eight different preparations of conventional and organic tea were measured together with two samples of the drinking water (blank samples TW1 and TW2) used to prepare the tea infusions. The fluoride content in the drinking water samples was below the detection limit for CIC and nICPMS, ISE gave a concentration of 0.21 mg /L, barely above the detection limit.

The fluorine concentrations in the infusions ranged from 1.4 to 2.8 mg /L (average values of all three methods, Table 1, see also Table S3 results of individual methods). The average standard deviation between the results for the replicate measurements and the three methods was between 2.9 and 6.4% for the individual samples. Comparison of the three methods (ISE with TISAB 3, CIC and nICPMS) gave similar results (Table 1, Fig. S2-4). There was no statistically significant difference between the three methods (t-test: p>0.05, Fig. S4).

Table 1. Fluorine content in different tea infusions. Mean and standard deviation of all measurements; mean and standard deviation of duplicate measurement (CIC and nICPMS), triplicate measurement of ISE (TISAB 3)and recovery of ISE compared to CIC and nICPMS in percentage.

Empty Cell mg/L
(average all methods)
ISE
(mg /L)
CIC + nICPMS
(mg F/ L)
Recovery ISE / (CIC+nICPMS) in %
Conventional tea (tea filter; A141000825) 2.3 ± 0.07 2.3 ± 0.05 2.3 ± 0.1 101
Conventional tea (tea strainer; A141100825) 2.2 ± 0.08 2.2 ± 0.04 2.2 ± 0.1 98
Conventional tea (brewed loose; A141200825) 2.2 ± 0.08 2.2 ± 0.05 2.1 ± 0.1 105
Conventional tea (tea bags only; A141300825) 2.8 ± 0.1 2.9 ± 0.06 2.8 ± 0.1 105
Organic tea (tea filter; A141400725) 1.4 ± 0.08 1.5 ± 0.02 1.3 ± 0.1 112
Organic tea (tea strainer; A141500725) 1.4 ± 0.06 1.4 ± 0.03 1.3 ± 0.1 106
Organic tea (brewed loose; A141600725) 1.4 ± 0.07 1.4 ± 0.03 1.3 ± 0.1 104
Organic tea (tea bags only; A141700725) 1.5 ± 0.1 1.7 ± 0.03 1.5 ± 0.1 114

Results for different TISAB preparations (Table S2) for the individual samples are shown in Fig. S5 and show a clear influence of the complexing agent used in the TISAB preparation on the concentration of fluoride determined in tea infusions. Only TISAB buffers containing citrate or CDTA were able to release fluoride completely from its metal-complexes present in tea infusions. There was no statistical difference between the measurement results between TISAB 3 and 4 formulations, indicating both are equally applicable for analysis using ISE with tea extracts (Fig. S4). TISAB 1 and 2 both resulted as did TISAB 5 in lower fluorine activity compared to TISAB 3 and 4, statistically differing from them by 5% and 10% on average, respectively (Fig. S4). These findings underline the importance of selecting the appropriate TISAB for ISE analysis of tea samples.

The three methods are based on different measurement principles and do not necessarily determine exactly the same analytical quantity. Using CIC total fluorine is estimated, after conversion of fluorine containing compounds to fluoride during combustion. nICPMS determines fluorine as element-specific signal at m/z 19 after ionisation in the post-plasma region. ISE determines the activity (= concentration) of free fluoride (present as F–) and the result depends on the extent of fluoride released from metal-fluoride complexes.

All methods use different approaches to detect fluorine and are therefore complementary but also affected by different sources of uncertainty. ISE is sensitive to the presence of aluminum and other metal ions. The results may be influenced by the choice of complexing agent used in TISAB buffer. CIC as all ion exchange methods, has the problem of fluoride being hardly retained on the column and often coeluting together with small organic acids / molecules which give an indistinguishable signal with a conductivity detector. nICPMS is susceptible to chloride-related matrix effects and, as a recently developed prototype technique, has not yet been as extensively validated for routine food analysis as CIC or ISE. Agreement among these analytically independent methods therefore provides orthogonal evidence for the reliability of the concentrations, although it does not replace validation using a certified reference material. In literature, comparable ranges of fluoride concentrations were found in black and green tea infusions (3.10-6.89 and 2.42-4.92 mg/L, respectively) (Biswas & Chinnakoti, 2024). The slightly lower levels in our samples can be caused by the measurement of pooled samples, containing both black and green teas or by differing origin or age of the leaves used. Similar results were found by Kanrar et al. (Kanrar et al., 2024) in infusions of black tea from different regions in India (1.08 – 2.43 mg/L). Besides the fluoride concentration, they also investigated the correlation with infusion numbers and found decreasing fluoride concentrations with increasing infusion numbers. In a representative survey on households in Germany from 2016, more households consume black tea than green tea (data not shown). As black tea is used by the German population typically for one infusion only and is preferred over green tea, decreasing the fluoride content in tea by repeated brewing seems not to be a feasible option.

Ma?yszek et al. found (2025) lower concentrations of fluoride in different types of tea available in Poland they used 2 g tea / 200 mL water. A similar amount of tea (9.3 g / L) was used in this study. Therefore, the lower concentrations found by Ma?yszek et al. (2025) are most likely a result of different origin and or quality of the teas studied. Although differences were reported in the preferred preparation methods for black tea and green tea (Hackethal et al., 2023), no differences were detected in the fluoride concentrations across different preparation methods in this study. In contrast Ma?yszek et al. (2025) reported that brewing time, leaf form, water type and temperature have a strong influence on fluoride concentration in the infusion.

Within the analysed sample set, fluoride concentrations were lower in the four organic tea pools than in the four conventional tea pools (t-test, p = 0.003). However, the limited number of pooled samples and the study design do not allow this difference to be attributed specifically to the production system. Tea origin, leaf maturity, harvest period, product quality and the composition of the individual pools were not controlled independently. The observed difference should therefore be interpreted as an association within the samples analysed rather than as evidence for a general effect of organic or conventional production.

Kanrar et al. (Kanrar et al., 2024) showed varying concentrations of fluoride in tea grown in different areas of India. Zhang et al., showed that levels of extractable fluoride depend among others on tea cultivar (Zhang et al., 2026). The factor mostly contributing to extractable F in the soil is phosphate fertilizer, followed by rainfall and groundwater of the region (Peng et al., 2021). Furthermore, older leaves contain higher F in comparison to younger leaves and leaves picked in spring showed lower fluoride levels than leaves harvested later in the year (Cai et al., 2016). Packages of tea do not normally give information when and where the leaves for this tea were picked, or the estimated age of the leaves. Therefore, the reason for the reduced fluoride content in organically produced tea compared to conventional tea cannot be stated.

3.3. Long-term dietary exposure through the consumption of tea infusion

Here the term fluoride is used, since this is the main form of fluorine in tea. Additionally, it is the form for which recommended daily intake data exists. The mean fluoride exposure through tea consumption for the general German population ranged from < 0.01 to 0.28 mg/day when considering the fluoride content of conventional tea and from < 0.01 to 0.17 mg/day for organic tea (Table 2). Adults aged 18 and older were the most exposed population group. For high consumers (P95), fluoride exposure through the consumption of tea infusion was estimated from 0.01 to 1.67 mg/day for conventional tea and from 0.01 to 1.02 mg/day for organic tea. Consumers of tea infusion can reach a mean fluoride exposure through tea consumption of up to 1.07 mg/day (18 – < 65 years) for conventionally produced tea and up to 0.65 mg/day (18 – < 65 years) for organically produced tea. The highest fluoride exposure among consumers of tea infusion in the 95th percentile was also determined for adults between 18 and 65 years (conventional tea: 2.81 mg/day; organic tea: 1.72 mg/day).

Table 2. Fluoride exposure through the consumption of tea infusion for the German general population and for consumers of tea infusion (in brackets) in mg/day.

Empty Cell Empty Cell Conventional tea Organic tea
Age group (years) n* Mean P50** P95 Mean P50** P95
0.5 – < 1 57 (5; 9%) <0.01 (0.01) 0 (0.01) 0.01 (-***) <0.01 (0.01) 0 (0.01) 0.01 (-***)
1 – < 3 308 (28; 9%) 0.03 (0.32) 0 (0.19) 0.19 (0.81) 0.02 (0.20) 0 (0.12) 0.12 (0.50)
3 – < 6 588 (73; 12%) 0.03 (0.21) 0 (0.20) 0.22 (0.41) 0.02 (0.13) 0 (0.12) 0.13 (0.25)
6 – < 10 789 (62; 8%) 0.02 (0.25) 0 (0.17) 0.13 (0.64) 0.01 (0.15) 0 (0.10) 0.08 (0.39)
10 – < 12 401 (28; 7%) 0.03 (0.37) 0 (0.29) 0.11 (1.11) 0.02 (0.22) 0 (0.18) 0.06 (0.68)
14 – < 18 744 (95; 13%) 0.08 (0.64) 0 (0.44) 0.55 (1.73) 0.05 (0.39) 0 (0.27) 0.34 (1.06)
18 – < 65 10,525 (2,115; 20%) 0.21 (1.07) 0 (0.78) 1.33 (2.81) 0.13 (0.65) 0 (0.48) 0.82 (1.72)
> 65 2,657 (702; 26%) 0.28 (1.04) 0 (0.79) 1.67 (2.66) 0.17 (0.64) 0 (0.49) 1.02 (1.63)

** Median fluoride exposure through the consumption of tea infusion is zero for the general population due to the small number of tea consumers (< 50%).

*** Due to the small number of tea consumers (n=5), no reliable statements can be made regarding the 95th percentile.

*

The number of tea consumers and their proportion in the corresponding age group are shown in brackets.

The mean fluoride exposure through tea consumption for the general German population was estimated at a maximum of 8% (conventional tea; 65 years and over) of the SLI (Table 3). High consumers (P95) can reach 50% of the SLI through tea consumption. Among tea consumers, the consumption of tea infusion contributed up to 32% to the SLI for average consumers and up to 85% for high consumers when the fluoride concentration of conventionally produced tea was taken into account. The results for the consumption of organic tea were lower due to the lower fluoride concentration in organically produced tea.

Table 3. Comparison of fluoride exposure through tea consumption with the UL for children aged < 8 years and the SLI for adults and children > 8 years. Results are shown for the German general population and for consumers of tea infusion (in brackets).

Empty Cell Empty Cell % of UL and SLI*
Empty Cell Conventional tea Organic tea
Age group (years) n** Mean P50*** P95 Mean P50*** P95
0.5 – < 1 57 (5; 9%) <1 (1) 0 (1) 1 (-****) <0 (1) 0 (1) 1 (-****)
1 – < 3 308 (28; 9%) 2 (20) 0 (12) 12 (51) 1 (12) 0 (7) 7 (31)
3 – < 6 588 (73; 12%) 1 (12) 0 (11) 12 (24) 1 (7) 0 (7) 8 (15)
6 – < 10 789 (62; 8%) 1 (12) 0 (7) 6 (32) 1 (7) 0 (5) 4 (20)
10 – < 12 401 (28; 7%) 1 (11) 0 (9) 3 (34) <1 (7) 0 (5) 2 (21)
14 – < 18 744 (95; 13%) 2 (19) 0 (13) 17 (52) 2 (12) 0 (8) 10 (32)
18 – < 65 10,525 (2,115; 20%) 6 (32) 0 (24) 40 (85) 4 (20) 0 (14) 25 (52)
> 65 2,657 (702; 26%) 8 (32) 0 (24) 50 (80) 5 (19) 0 (15) 31 (49)
UL: Tolerable upper intake level; SLI: Safe level of intake
* Applied ULs were 1.0, 1.6 and 2.0 mg/day for children aged < 1 year, 1 – 3 years and 4 – 8 years, respectively. For adults and children > 8 years a SLI of 3.3 mg/day was considered.
** The number of tea consumers and their proportion in the corresponding age group are shown in brackets.
*** The median contribution of fluoride exposure through the consumption of tea infusion to the UL and SLI is zero for the general population due to the small number of tea consumers (< 50%).
**** Due to the small number of tea consumers (n=5), no reliable statements can be made regarding the 95th percentile.

The fluoride exposure of 79 consumers (<1% of the total population) of tea infusion in the age of 18 and older exceeded the SLI when considering the fluoride concentration of conventionally produced tea. Taking the fluoride concentration of organically produced tea into account, the SLI was exceeded by 13 consumers (data not shown). Consequently, the proportion of people exceeding the SLI was less than 1% in the age group of 18 years and older in the general population. For all tea consumers in the younger age groups, exposure from tea consumption was well below the UL or SLI.

Black and green teas are known to account for a significant proportion of fluoride exposure. In the Irish TDS, for example, black tea accounted for 76% of total fluoride exposure in adults (aged > 18 years) (FSAI, 2018). For children aged 5 to 12 years, black tea accounted for 29% of fluoride exposure. Adults in the 95th percentile had a significantly higher fluoride intake from black tea (76 µg/kg bw/day) than in the present study (conventional tea: 19 µg/kg bw/day). This difference can be attributed to a higher average fluoride concentration in the Irish TDS (4.8 mg/L; prepared with deionized water) and a higher tea consumption in the Irish population (FSAI, 2018). Tea has also been identified as a significant source of dietary fluoride intake in New Zealand. With a contribution of 60 – 80%, tea was found to be the main source of fluoride exposure in the adult population of New Zealand (Pearson et al., 2018).

However, a recently published EFSA opinion indicated a lower contribution of tea to total fluoride exposure in European countries. Depending on the country and the age group considered, tea consumption can account for up to 63% of the total fluoride exposure. For Germany, the contribution ranges between 0% and 22% among all considered age groups. Within all age groups of the German population EFSA estimated the highest fluoride exposure through tea consumption for people aged 65 years and older (2.4 µg/kg bw/day) (EFSA, 2025). This is consistent with the findings of the present study. As the fluoride concentration in tea considered by EFSA (1.34 mg/L) is similar to that determined in the present study for organically produced tea (1.36 mg/L), exposure through consumption of organic tea in the present study is also in the same order of magnitude (2.3 µg/kg bw/day) for people aged 65 years and over (data not shown).

3.4. Exposure uncertainties

The percentage of black and green tea per food pool was determined from the consumption study NVS ll from 2006. In this survey 26% of tea consumers drank green tea. In a more recent household survey from 2015 (n = 1008), the percentage of households preparing regularly green tea was given with 40% of all tea consumers. As this was a survey on households, it is not directly comparable with the consumption study but might suggest a shift to more green tea consumption in recent years.

The present exposure assessment for fluoride was based on concentration data from the BfR MEAL Study. The BfR MEAL Study’s TDS approach averaged the fluoride concentration of different tea types by pooling them. Different types of tea, brands or charges can contain varying fluoride concentrations. The variability between tea subsamples cannot be indicated due to the pooling process. This uncertainty is accepted since the BfR MEAL study was designed to collect representative occurrence data. This can result in variations in fluoride exposure, particularly among brand-loyal consumers.

The fluoride concentration of the tea infusion could also be influenced by the drinking water used for preparation. However, the drinking water used in the BfR MEAL Study does not account for regional variations in fluoride concentration. Therefore, tap water in different regions of Germany may have different fluoride concentrations, which can lead to varying fluoride concentrations in tea infusions and, ultimately, different fluoride exposure levels for tea consumers.

The consumption surveys were conducted in different periods, and the number of tea consumers was small in some age groups. Consequently, high-percentile estimates are less robust for these groups, particularly for children aged 0.5 to <1 year, for whom only five tea consumers were available.

Finally, the present assessment considers fluoride exposure from tea infusion only. Fluoride intake from drinking water, other foods, fluoridated salt, supplements and swallowed dental-care products was not included. Comparison with the UL or SLI therefore characterizes the potential contribution of tea to fluoride exposure and should not be interpreted as a complete assessment of total fluoride intake.

4. Conclusion

4.1. Analytical methods

CIC and nICPMS give comparable concentrations for fluorine in tea infusions. nICPMS is sensitive to the presence of high amounts (> 500 mg/L) of chlorine in sample matrix, which can be compensated for by using standard addition for quantification. nICPMS is an untried new method for the determination of fluorine in real samples. When matrix and other influences are better known it may well be suitable as a routine method for total fluorine determination. ISE used with TISAB V buffer gives somewhat (ca. 10%) lower concentration than the other two methods. When ISE is used for quantification of fluoride in tea TISAB buffers containing either citrate or CDTA as complexing agents for aluminum and other ions should be used. When only free (none complexed) fluoride is to be measured by ISE no complexing agent should be added to the TISAB solution.

4.2. Samples

Among the pooled samples analysed in this study, the conventional tea pools contained higher fluoride concentrations than the organic tea pools. However, the underlying cause of this difference could not be determined from the present sample design.

4.3. Exposure assessment

The majority of tea consumers in Germany do not ingest more than 24% of the SLI for fluoride through tea. In the general population, tea consumption contributes to less than 10% of the SLI. People aged 65 and over have the highest average fluoride exposure in the general population.

Uncited reference

(Pavlovi? et al., (2023))

CRediT authorship contribution statement

Tobias Höpfner: Writing – review & editing, Formal analysis, Data curation. Marc Preihs: Investigation. Irmela Sarvan: Writing – review & editing, Writing – original draft, Project administration, Methodology, Investigation. Andrea Raab: Writing – review & editing, Writing – original draft, Validation, Methodology, Investigation. Jörg Feldmann: Writing – review & editing, Resources, Project administration, Funding acquisition.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgement

We gratefully acknowledge the loan of the NexION 2000 by PerkinElmer, Canada and their continuous support with instrument development, especially from Hamid Badiei and Bill Fisher. The authors acknowledge the financial support by the University of Graz.

Appendix A. Supplementary material

Data availability

Data will be made available on request. All analytical data are contained within the publication and its associated supplementary material.

References