Introduction

Cellulose nanofibers (CNFs) are novel nanomaterials derived from plants. Based on their inherent attributes, which include high strength, high water retention characteristics, and transparency, hydrophilic CNFs have been extensively applied across diverse fields, such as pharmaceutical products, cosmetics, and food thickeners. Hydrophobic CNFs also demonstrate versatility over a spectrum of applications that span automotive engineering, reinforcing tire, household electrical appliances, and house construction materials, and ongoing technological advancements have been made to increase their practical applications (Ministry of the Environment 2021). Market projections conducted by Shatkin et al. (2014) suggested that their utilization as fundamental constituents in various commodities will increase. Although CNFs are expected to contribute to a sustainable society, an increase in the mass production of CNFs could result in increased material disposal into the environment, leading to potential ecological effects on aquatic organisms. Hence, the effects of CNFs on aquatic organisms should be evaluated in parallel with the development of CNFs. There have been multiple efforts in recent years to collect data on the ecotoxicity of nanocellulose materials to aquatic organisms to evaluate their environmental safety (e.g., Kovacs et al. 2007; Pereira et al. 2014; Munk et al. 2015; Tlili et al. 2017).

To date, ecotoxicity testing of CNFs has been based on test methods, such as those based on the Organization for Economic Co-operation and Development (OECD) test guidelines (TGs) for water-soluble chemicals. However, these methods do not consider issues that originate from the unique characteristics of CNFs. For example, the properties of CNFs, such as dispersibility and dispersion stability, are prone to alteration in response to changes in water quality (Bitounis et al. 2019). Therefore, the properties of CNFs may change when they interact with the test medium. Without an understanding of the properties of CNFs in the test medium and the variation in the properties due to variations in test conditions, the ecotoxicity of CNFs cannot be adequately assessed. Changes in the concentration of CNFs in the test media owing to agglomeration and settling may result in lower exposure to the test organisms than the nominal concentrations and underestimate their ecotoxicological risks (Harper et al. 2016; OECD 2021). Thus, these factors should be considered during ecotoxicity testing of CNFs. Recently, the OECD published a guidance document (GD) for conducting aquatic and sediment toxicological testing of nanomaterials (OECD 2021). The GD recommends that ecotoxicity testing be conducted according to the distinct features of nanomaterials. Because the dispersibility and dispersion stability of CNFs are affected by the composition of the test medium, the OECD TGs recommend that ecotoxicity tests for CNFs be conducted while maintaining dispersibility and dispersion stability. To meet this requirement, a test medium must be selected that maintains the dispersibility and dispersion stability of CNFs, and the ecotoxicity test should be conducted accordingly.

CNFs are produced using methods such as chemical modification and mechanical fibrillation, and their properties, such as the fibrillation degree, fiber size, and surface charge, vary depending on the manufacturing method (Isogai et al. 2011). CNFs with different properties can exhibit different ecotoxicity profiles and values to aquatic organisms. Therefore, the ecotoxicity of CNFs should be evaluated individually. The 2,2,6,6-tetramethylpiperidine-1-oxyl (TEMPO) catalytic oxidation can yield uniform CNFs with a fibrillated width of 3–4 nm through chemical modifications (Saito et al. 2007). The CNFs produced using this method (hereafter referred to as TEMPO-CNFs) are expected to be utilized in various fields. However, ecotoxicity data for TEMPO-CNFs is limited. There has been only one study on TEMPO-CNFs, wherein the development and physiological functions of zebrafish embryos and juvenile fish morphology were evaluated (Harper et al. 2016). Acquiring ecotoxicity data for TEMPO-CNFs is necessary to assess its environmental safety.

In this study, acute toxicity tests of TEMPO-CNFs on Daphnia magna and Oryzias latipes were conducted using a test medium in which the dispersibility and dispersion stability of the CNFs were maintained to the best possible extent. First, the test medium suitable for the toxicity tests was selected by comparing the properties of the CNFs in five media with different constituent contents; the results of concentration uniformity tests were also compared. Next, the acute toxicity of TEMPO-CNFs to D. magna and O. latipes in the selected test medium was evaluated according to the OECD TGs. The temporal uniformity of the CNFs at a low concentration decreased in the acute toxicity tests of D. magna. Hence, to evaluate the dispersion and dispersion stability of the CNFs, the spatiotemporal distribution of the CNF concentration was further investigated under conditions identical to those used for the acute toxicity test. Based on these results, we propose a test method to accurately assess the acute toxicity of CNFs.

Materials and methods

Preparation of materials

An aqueous slurry of TEMPO-CNFs without preservatives was obtained from Nippon Paper Industries Co., Ltd. (Tokyo, Japan). The sodium carboxylate content in the TEMPO-CNFs was 1.3–1.6 mmol/g; the typical width and length of the TEMPO-CNFs were 3–4 nm and 0.1–1 μm, respectively, according to the manufacturer's brochure. The obtained CNFs were irradiated with 1 kGy gamma radiation to preclude fungal proliferation. Gamma irradiation is a commonly used method for sterilizing industrial products (Criado et al. 2017). No microorganisms were detected in the CNFs after sterilization. The changes in the fiber length and fiber width distributions of the CNFs due to 1 kGy gamma radiation were slight (Horie et al. 2023).

Selection of test medium

Tested media and preparation of stock solution

Five media with diverse salt concentrations, namely (1) ultra-pure water, (2) dechlorinated tap water, (3) very soft synthetic water (US EPA 2002), (4) moderately hard synthetic water (US EPA 2002), and (5) a mixture of very soft synthetic water and dechlorinated tap water at a ratio of 8:2 (referred to as blended water; Mano et al. 2022), were prepared to examine the properties, dispersibility, and dispersion stability of TEMPO-CNFs in the respective media. The order of ion concentration in the media was as follows: ultra-pure water < very soft synthetic water < blended water < dechlorinated tap water < moderately hard synthetic water. The composition of each medium is listed in Table S1 (Online Resource 1).

The stock solutions used in this study were prepared according to the method described by Fujita et al. (2022) with minor modifications. Stock solutions were prepared by diluting the sterilized TEMPO-CNFs slurry (approximately 10 g/L) to a concentration of 1 g/L using the media. The CNF slurry and medium were mixed for 30 min at 1500 rpm using a planetary centrifugal mixer (ARE 310, Thinky Corp., Japan) to ensure homogeneity of the suspension. The resulting solution was used as the stock solution.

Evaluation of the characteristics of TEMPO-CNFs

The rheology and zeta potential of the stock solutions were measured to examine the effects of the media on the properties of the TEMPO-CNFs. The rheology was analyzed using a modular compact rheometer (MCR 302, Anton Paar, Austria), and the zeta potential was analyzed using a Zetasizer Pro Red Label (Malvern Panalytical Ltd., UK).

Study of aggregation properties

The aggregation of TEMPO-CNFs in 100 mg/L solutions prepared with ultra-pure water, moderately hard synthetic water, and blended water was examined via differential centrifugal sedimentation analysis (CPS, DC24000UHR) and microscopy. A system comprising a laser confocal microscope and a scanning probe microscope (OLS4500, Olympus) was used for the microscopy analyses. Moreover, a drop of each solution was put on a silicon substrate and dried under ambient conditions for microscopic observation.

Study of concentration uniformity

The stock solution of TEMPO-CNFs was added to a 100 mL glass beaker containing 50 mL of medium and adjusted at 30 mg/L. The test solutions were stirred at 550 rpm for 2 min using a magnetic stirrer (KSS-12; AS ONE, Osaka, Japan). The spatiotemporal uniformity of TEMPO-CNFs concentration in each beaker was evaluated using 1 mL samples collected from two depths: 5 mm below the liquid surface and 5 mm above the bottom of the beaker. The 1 mL samples were collected at the start of the test, after 24 h, and after 48 h. Three replicates were used for each medium.

Ecotoxicity tests

Ecotoxicological tests were conducted using D. magna and O. latipes in the selected test medium (i.e., blended water). Acute toxicity tests were conducted according to the OECD TGs. The stock cultures and test conditions of the organisms used in the ecotoxicity tests are detailed below.

Stock culture and test preparation of D. magna and O. latipes

The D. magna used in this study was obtained from the National Institute for Environmental Studies (NIES), Japan. The culture medium for D. magna was changed twice a week. The cultures were maintained at 20 ± 1 °C under a light/dark photoperiod of 16/8 h. Chlorella vulgaris (Recenttec Corp., Japan) was added to each individual at a dose of 0.1 to 0.2 mg C/day. The stock culture of D. magna, which consisted of 20 mature organisms, was maintained in a 1 L beaker containing blended water for a period of 72 h before the start of the test. Neonates born within 24 h of the adaptation period of the stock culture were used for the toxicity tests.

For the O. latipes toxicity test, the NIES-R strain obtained from the NIES was used. Fertilized eggs were procured from organisms raised in a partially submerged aquatic enclosure, hatched, and then nurtured for 9 weeks under controlled conditions, wherein the water temperature was set to 25 ± 1 °C under a light/dark photoperiod of 16/8 h. Following the aforementioned growth period, individuals with an average total length of 1.7 cm and mass of 0.043 g were selected and acclimated to the test medium for 7 days. During the acclimation, the test medium was aerated using an air pump to achieve a dissolved oxygen saturation concentration of 97.0–99.2%, and Artemia was provided as the daily food to the test individuals until 24 h before the start of the toxicity test.

Acute ecotoxicity tests for D. magna

Acute (48 h) ecotoxicity tests for D. magna were performed in accordance with the OECD TG 202 Daphnia sp. Acute Immobilization Test (OECD 2004). The test solutions were prepared by adding a stock solution to the test medium. The CNFs concentrations were 3.1, 6.3, 12.5, 25, 50, and 100 mg/L, and 0 mg/L was used as the control. Four replicates were used for each CNF concentration. Each replicate was prepared in a 100 mL beaker using a 50 mL test solution and five neonates. The toxicity test conditions, including the water temperature, photoperiod, and light intensity, were identical to those used for the stock cultures. At 24 and 48 h after the start of the test, the number of immobilized individuals in each beaker was counted. The pH (D-21, HORIBA, Japan) and dissolved oxygen (DO) (SevenGo pro (SG9), METTLER TOLEDO, Japan) were measured at the beginning and end of the test.

The TEMPO-CNFs concentrations in the test solutions were analyzed by collecting 1 mL of test solutions at the start and end of the test. The test solutions were sampled from the middle portions of the beakers in sterilized 1.5 mL microtubes (1 beaker per concentration). A fresh test medium was prepared for the test solutions. The fresh test medium was filtered through 0.45 μm hydrophilic polytetrafluoroethylene filters and stocked in a sterilized 15 mL volume centrifuge tube and a 100 mL polypropylene bottle. The 100 mL polypropylene bottle was prewashed with filtered fresh test medium. The medium stocked in the centrifuge tube was used for ion concentration analysis, and that in the polypropylene bottle was used for dissolved organized carbon (DOC) analysis.

Acute ecotoxicity tests for O. latipes

Acute (96 h) ecotoxicity tests for O. latipes were conducted in accordance with the OECD TG 203 Fish Acute Toxicity Test (OECD 2019). The test solutions were prepared by adding a stock solution to the test medium. The CNF concentrations were 3.1, 6.3, 12.5, 25, 50, and 100 mg/L, and 0 mg/L was used as the control. A 3 L test solution in a 5 L glass tank with seven fish was used for each CNF concentration. The test solutions were freshly prepared and renewed 48 h after the start of the test. The test was conducted at 25 ± 1℃ with a light period of 16 h with room light and a dark period of 8 h. The number of dead individuals in each tank was counted every 24 h after the start of the test. The pH (D-21, HORIBA, Japan) and DO (Seven Go Pro (SG9), METTLER TOLED, Japan) were measured at the beginning, before renewing the test solution, after renewing the test solution, and at the end of the test.

The TEMPO-CNFs concentrations in the test solutions were analyzed by collecting 1 mL of test solutions at the beginning of the test, before renewing the test solution, after renewing the test solution, and at the end. The test solutions were sampled from the middle portions of the tanks in 1.5 mL microtubes. Each fresh test medium prepared for the initial and renewal test solutions was filtered through a 0.45 μm hydrophilic polytetrafluoroethylene filter and stocked in a sterilized 15 mL volume centrifuge tube (for ion concentration analysis) and a 100 mL polypropylene bottle (for DOC analysis). The 100 mL polypropylene bottle was prewashed with filtered fresh test medium.

Concentration uniformity test of the TEMPO-CNFs in test solutions used for D. magna acute toxicity test

In the acute toxicity tests of D. magna, the temporal uniformity of the CNF concentration decreased at low CNF concentrations. Hence, the concentration uniformity of the CNFs was investigated under conditions identical to the acute toxicity test conditions. The stock solution was added to the test media to prepare test solutions with CNF concentrations of 3 mg/L and 100 mg/L. Three replicates were used for each treatment. Each replicate was prepared in a 100 mL glass beaker using a 50 mL test solution and five neonates born within 24 h. The test was conducted at 21 ± 1 °C under a 16 h light/ 8 h dark photoperiod. The CNF concentrations were measured by sampling 1 mL of the solutions at the start and end of the test; the samples were collected at two depths: 5 mm below the liquid surface and 5 mm above the bottom of the beaker. The pH (AS800, AS ONE Corp., Japan) and DO (HQ30d, HACH, U.S.) were measured at the start and end of the test. A fresh test medium, prepared for the test solutions, was filtered through a 0.45 μm hydrophilic polytetrafluoroethylene filter and stocked in a sterilized 15 mL volume centrifuge tube for ion concentration analysis. In addition, the concentration uniformity of TEMPO-CNFs in the test solutions used in the O. latipes acute toxicity tests was investigated.

Measurement of ion and dissolved organized carbon concentrations

The concentration of ions (Cl−, SO42−, Na+, K+, Mg2+, and Ca2+) was determined using ion chromatography (ICA-7000, DKK-TOA Corporation, Tokyo, Japan). DOC was analyzed using a total organic carbon meter (Multi N/C 3100, Analytik Jena, Germany).

Measurement of CNF concentrations in test solutions

The CNF concentrations of test solutions with nominal CNF concentrations of 25–100 mg/L were determined using the phenol–sulfuric acid method (Hodge and Hofreiter 1962) with some modifications (the use of microplates) (Ogura et al. 2023). The CNF concentrations of test solutions with nominal CNF concentrations of 3.1–12.5 mg/L were measured using the cellulase enzymatic degradation method (Ogura et al. 2020). Outliers were removed using the modified Thompson’s tau test (95% confidence level).

Calculations of ecotoxicity values

The swimming inhibition percentage in the D. magna test and mortality percentage in the O. latipes test were determined by dividing the number of immobilized and dead individuals, respectively, by the number of individuals tested. The median effective concentration (EC50, the concentration at which 50% of the test organisms were affected) of D. magna and the median lethal concentration (LC50, the concentration at which 50% of the test organisms died) of O. latipes were calculated by fitting the data to two-parameter log-logistic models with binomial error distributions.

The significance level was set to 5% for all the statistical analyses. Statistical analysis software R (ver. 4.1.2) was used for all statistical analyses (R Development Core Team 2022).

Results and discussion

Selection of test medium

Evaluation of the characteristics of TEMPO-CNFs

The rheology and zeta potential values of the stock solutions prepared using the five media are shown in Figs. 1 and 2, respectively. The viscosities of stock solutions prepared with very soft synthetic water and blended water were comparable to that of the stock solution prepared with ultra-pure water (Fig. 1). The viscosities of stock solutions prepared with moderately hard synthetic water and dechlorinated tap water were higher than that of the stock solution prepared with ultra-pure water. The absolute zeta potential values of the stock solutions prepared with moderately hard synthetic water and dechlorinated tap water were lower than those of stock solutions prepared with ultra-pure water (Fig. 2).

Fig. 1
figure 1

Results of the rheological analyses of the media

Full size image
Fig. 2
figure 2

Zeta potentials of the test media; the error bars indicate the standard deviation

Full size image

Study of aggregation properties

Figure 3 shows the laser confocal microscope and scanning probe microscope images of the CNFs dispersed in ultra-pure water, moderately hard synthetic water, and blended water. The microscopic observations were conducted just after preparation (0 h) and 48 h after preparation to examine the formation of large aggregates in the dispersion during the test. The laser confocal microscope images of the six samples (Fig. 3a, c, e, g, i, k) did not show significant differences, indicating that CNF aggregations with a size >1 μm were not formed in the tested solutions. The results of differential centrifugal sedimentation analysis (data not shown) also confirmed the absence of aggregation with a size >0.1 μm. The diameter (3–4 nm) and length (~0.5 μm) of the CNFs were consistent with those reported in previous studies (Saito et al. 2007; Isogai et al. 2011).

Fig. 3
figure 3

Laser confocal microscope images of CNFs in ultra-pure water (a, c), moderately hard synthetic water (e, g), and blended water (i, k) at 0 h (a, e, i) and 48 h (c, g, k) after sample preparation. Scanning probe microscope images of CNFs in ultra-pure water (b, d), moderately hard synthetic water (f, h), and blended water (j, l) at 0 h (b f, j) and 48 h (d, h, l) after sample preparation

Full size image

Study of concentration uniformity

The results of the concentration uniformity tests of each medium are shown in Fig. 4. The TEMPO-CNFs concentration in each medium was measured per beaker every 24 h to investigate the spatiotemporal variation in concentration and the percentage deviation from the nominal concentration.

Fig. 4
figure 4

Results of the concentration uniformity test used to determine a test medium for the ecotoxicity tests; the error bars indicate the standard deviation

Full size image

For the ultra-pure water, the averages of the percentage deviation from the nominal concentration in each beaker were 2.9–7.6% in the upper part and 2.8–10% in the lower part. The averages of the temporal percentage change were 4.4–6.8% in the upper part and 4.7–11% in the lower part. The averages of the spatial percentage change were 5.7–6.1% in the upper part and 5.4–6.4% in the lower part. The averages of the spatiotemporal percentage change were 4.1–9.8%.

For the very soft synthetic water, the averages of the percentage deviation from the nominal concentration were 2.7–8.9% in the upper part and 1.8–12% in the lower part. The averages of the temporal percentage change were 1.7–7.7% in the upper part and 7.8–17% in the lower part. The averages of the spatial percentage change were 1.5–15% in the upper part and 1.5–13% in the lower part. The averages of the spatiotemporal percentage change were in the range of 2.1–19%.

For the blended water, the averages of the percentage deviation from the nominal concentration were 1.2–5.9% in the upper part and 3.6–9.4% in the lower part. The averages of the temporal percentage change were 3.0–6.1% in the upper part and 3.0–12% in the lower part. The averages of the spatial percentage change were 4.1–6.3% in the upper part and 3.9–6.9% in the lower part. The averages of the spatiotemporal percentage change were 3.0–9.3%.

The observed percent changes in TEMPO-CNFs concentrations in test solutions prepared with ultra-pure water, very soft synthetic water, and blended water were less than 20%, which was used as a guide for dispersion stability in the OECD guidance for nanomaterials and is considered acceptable (OECD 2021).

For the dechlorinated tap water, the averages of the percentage deviation from the nominal concentration were in the range of 10.1–27% in the upper part and 3.8–25% in the lower part. The averages of the temporal percentage change were 10.5–19% in the upper part and 7.0–21% in the lower part. The averages of the spatial percentage change were 5.7–13% in the upper part and 5.4–12% in the lower part. The averages of the spatiotemporal percentage change were 5.4–24%.

For the moderately hard synthetic water, the averages of the percentage deviation from the nominal concentration were in the range of 16–34% in the upper part and 12–50% in the lower part. The averages of the temporal percentage change were 11–28% in the upper part and 28–47% in the lower part. The averages of the spatial percentage change were 13–55% in the upper part and 11–23% in the lower part. The averages of the spatiotemporal percentage change were 8.5–46%. The TEMPO-CNFs concentration was non-uniform in both dechlorinated tap water and moderately hard synthetic water. The detailed results of the percentage deviations and percentage changes in the measured concentrations are shown in Tables S2–S11 (Online Resource 2).

The salt concentration in the test medium is known to affect the agglomeration of the CNFs. Pereira et al. (2014) conducted an algal growth inhibition study investigating the agglomeration of cotton-derived CNFs in artificially prepared water and water from the Seine River. CNF agglomeration was observed in the Seine River water medium, attributed to its higher salinity than that of the artificial water medium. The Na+ ions in TEMPO-CNFs can be exchanged with other monovalent, divalent, and trivalent metal ions (Saito and Isogai 2005; Sone et al. 2016). Consistent with these findings, changes in the viscosity, zeta potential, and concentration uniformity of CNFs in media increased with increasing salt concentrations.

Table 1 summarizes the ionic levels and states of the TEMPO-CNFs in the five media. The viscosity, zeta potential, and concentration uniformity of the TEMPO-CNFs changed with an increase in the ionic concentration. Blended water was selected as the medium for ecotoxicity testing because of its concentration uniformity and suitability for rearing organisms.

Table 1 Ionic levels and states of TEMPO-CNFs in five media
Full size table

Ecotoxicity of TEMPO-CNFs to test organisms

Acute ecotoxicity tests for D. magna

The water temperature, pH, and DO of the test solutions during the exposure period were in the ranges 19.5–20.1 °C, 7.1–7.7, and 8.9–9.4 mg/L, respectively. The DOC value of the fresh test medium was 0.6 mg/L. These results meet the conditions established by the OECD TG 202 (OECD 2004). The TEMPO-CNFs concentrations at the start and end of the test are shown in Fig. 5. For the treatment with a nominal CNF concentration of 6.3 mg/L, the TEMPO-CNFs concentration decreased by approximately 32%, and the percentage deviation from the nominal concentration decreased by approximately 29%. For all other treatments, the percentage deviation from the nominal concentration and the percentage change in the TEMPO-CNFs concentration between the start and end of the test were within 20%. The measured CNF concentrations are listed in Table S12 (Online Resource 3).

Fig. 5
figure 5

Measured TEMPO-CNFs concentration in D. magna ecotoxicity tests. The cellulase enzymatic degradation method was used for the control to low-concentration treatments, and the microplate-based phenol–sulfuric acid method was used for the high-concentration treatments; the error bars indicate the standard deviation

Full size image

Swimming inhibition was not observed at the end of the toxicity test in any of the tested concentrations. In the high-concentration treatments, the measured concentration was within 20% of the nominal concentration. These results indicate that the half-swimming inhibition concentration is above the highest test concentration, which resulted in an EC50 >100 mg/L.

Acute ecotoxicity tests for O. latipes

The water temperature, pH, and DO of the test solutions were in the range of 24.3–25.8 °C, 7.2–7.7, and 7.1–8.0 mg/L, respectively. The DOC concentrations at the beginning of the test and after renewing the test solution were 0.4 and 0.3 mg/L, respectively. The values of water temperature, DO, pH, and DOC meet the validity criteria of OECD TG 203 (OECD 2019).

The TEMPO-CNFs concentrations measured at the beginning of the test (0 h), before renewing the test solution (48 h), after renewing the test solution (after renewal at 48 h), and at the end of the test (96 h) are displayed in Fig. 6. The percentage deviations from the nominal concentration as well as the TEMPO-CNFs concentrations at the start and end of the test were within 20% for all the treatments. Given the minimal changes in concentration across all CNF treatments, nominal concentrations were used to calculate LC50. The measured CNF concentrations are listed in Table S13 (Online Resource 3).

Fig. 6
figure 6

Measured concentrations of CNFs in O. latipes ecotoxicity tests; The cellulase enzymatic degradation method was used for the control to low-concentration treatments, and the microplate-based phenol–sulfuric acid method was used for the high-concentration treatments; the error bars indicate the standard deviation

Full size image

Dead individuals were not observed for any CNF treatments at the end of the toxicity test. The results indicated that the LC50 was >100 mg/L.

Comparison with previous studies

In this study, acute toxicity of TEMPO-CNFs to the two species was not observed even at the highest exposure concentration. Harper et al. (2016) exposed zebrafish at 6 h post-hatching to varying concentrations (2.0 to 250 mg/L) of TEMPO-CNFs. Their findings indicated that the mortality rate among developing zebrafish embryos remained below 20%, even at the highest concentration of 250 mg/L. However, the species and developmental stages differed from those in this study, and a consistent pattern emerged suggesting lower toxicity associated with TEMPO-CNFs. Previous studies have not investigated the acute toxicity of TEMPO-CNFs to D. magna and O. latipes. Hence, the results of this study were compared with those of previous studies on various CNFs with different fibrillations. Vartiainen et al. (2011) investigated the acute toxicity of microfibrillated cellulose to D. magna and reported that no acute toxicity was observed at the highest concentration (600 mg/L) used. Ong et al. (2017) reported an acute toxicity test wherein D. magna was exposed to 6000 mg/L of lignin-coated fibrils: no reduction was observed in 2-d survival. An acute toxicity test of lignin-coated fibrils using zebrafish as the test organism showed an EC50 > 2000 mg/L (Ong et al. 2017). These studies do not specifically investigate TEMPO-CNFs, and the results therein cannot be directly compared with our results. Nonetheless, the ecotoxicity of TEMPO-CNFs is considered to be as low as that of previously reported CNFs.

Concentration uniformity test of the TEMPO-CNFs in test solutions used for the D. magna acute toxicity test

The D. magna ecotoxicity test with a nominal CNF concentration of 6.3 mg/L indicated that the CNF concentration changed by more than 20%. Hence, the concentration uniformity of the CNFs was investigated under conditions identical to those used in the acute toxicity test. The water temperature, pH, and DO in the test solutions were in the range of 20.9–21.6 °C, 7.5–7.6, and 8.7–8.9 mg/L, respectively. These results meet the validity criterion of OECD TG 202.

Figure 7 shows the measured TEMPO-CNF concentrations before (0 h) and after the test (48 h) when 3 and 100 mg/L TEMPO-CNFs were added. When the nominal concentration was 3 mg/L, the average percentage deviation from the nominal concentration for each beaker was in the range of 11–13% in the upper part and 16–22% in the lower part. However, when the nominal concentration was 100 mg/L, the average percentage deviation from the nominal concentration was in the range of 3.0–5.3% in the upper part and 1.7–3.0 in the lower part. When the nominal concentration was 3 mg/L, the average temporal percentage change was 18% in the upper part and 31% in the lower part. When the nominal concentration was 100 mg/L, the average temporal percentage change was 8.1% in the upper part and 3.3% in the lower part. When the nominal concentration was 3 mg/L, the average spatial percentage change was in the range of 5.1–15% in the upper part and 5.6‒13% in the lower part. When the nominal concentration was 3 mg/L, the average spatiotemporal percentage change was 20% in the upper part and 29% in the lower part. When the nominal concentration was 100 mg/L, the average spatiotemporal percentage change was 4.5% in the upper part and 6.9% in the lower part. The detailed results, including all percentage deviations and percentage changes in the measured concentrations, are shown in Tables S14–S17 (Online Resource 4).

Fig. 7
figure 7

Concentration uniformity tests of TEMPO-CNFs in (a) 3 mg/L CNF and (b) 100 mg/L CNF solutions under conditions identical to those used in the D. magna acute toxicity test (containing 5 neonates); the error bars indicate the standard deviation

Full size image

In the high-concentration treatments, the concentration uniformity was maintained in both the ecotoxicity and concentration uniformity tests. These results imply that the concentration uniformity of TEMPO-CNFs tends to be maintained in high-concentration treatments. In contrast, the item "percent change within 20%" listed in the OECD GD No. 317 (OECD 2021), which is a document on ecotoxicity testing of nanomaterials, was not achieved in the D. magna acute toxicity test with 6.3 mg/L CNFs and the concentration uniformity test with 3 mg/L CNFs. Thus, the concentration of TEMPO-CNFs in the solution varied temporally at low concentrations. The decrease in the CNF concentration when testing at lower concentrations may be due to the adhesion of CNFs to the inner surface of the beaker and D. magna or D. magna feeding on CNFs.

The ion concentrations measured at the beginning and end of the concentration uniformity tests are shown in Fig. 8. In the solution containing 100 mg/L TEMPO-CNFs, the concentration of Na+ ions increased while the concentrations of Mg and Ca ions decreased. This result indicates that the sodium counterions in TEMPO-CNFs were exchanged with Mg and Ca ions in the test medium.

Fig. 8
figure 8

Ion concentration in the medium during the concentration uniformity test

Full size image

The results of the concentration uniformity study with O. latipes are shown in Fig. S4 (Online Resource 5) and Tables S18-S21 (Online Resource 5). The change in concentration was within 20%, indicating that the ecotoxicity tests were conducted under conditions wherein the dispersibility and dispersion stability of TEMPO-CNFs were maintained.

Lessons from this study for future ecotoxicity testing and risk assessment

Our results showed that CNF concentration uniformity can decrease owing to the interaction between the medium components and CNFs. In particular, test media with higher ion concentrations tended to lower the CNF concentration uniformity. Thus, it is important to select a test medium that maintains the dispersibility of the CNFs to the best possible extent. In addition, measuring the concentration of CNFs in the test solution during ecotoxicity testing is crucial. If concentration uniformity cannot be maintained, periodic measurement of the CNF concentration throughout the study is preferable (OECD 2021). Our results also showed that the interaction between the medium components and CNFs could alter the ion concentrations in the test solution. Thus, it is important to check the ion concentrations in the test solution. If adverse effects on organisms are observed during ecotoxicity tests of CNFs, the effect of the change in ion concentration (e.g., depletion of essential ions) due to the addition of CNFs needs to be investigated as well. Our findings and the methods used in this study will be useful for future ecotoxicity testing studies with different species or CNFs.

Several studies have explored the hazards associated with increased CNF emissions into the environment. A comprehensive review conducted by Endes et al. (2016) compiled existing information on the impact of nanocellulose on human and environmental health. The study revealed significant data gaps, particularly in terms of chronic, low-concentration, and repetitive exposures. Despite these insights, toxicity testing with low CNF concentrations remains notably challenging, thus highlighting the need for pragmatic impact assessments based on real environmental conditions. However, few methods of directly quantifying released nanomaterials are available, and reliable data were difficult to collect in this study under low exposure conditions. Although reliable indoor experimental data are still being collected, field experimental methods must be implemented to fill the data gap in the future.

Stoudmann et al. (2019) utilized current data and predicted the volume of nanocellulose discharge into the actual environment of areas in Europe. Through the application of probabilistic species sensitivity distributions, they forecasted the future risk of nanocelluloses and categorized the primary sources of discharge into freshwater, soil, and air. In this study, the authors mentioned that certain applications are likely to release nanocelluloses into wastewater during their use period, and the largest share of nanocelluloses entering the ‘wastewater’ compartment will flow into the ‘sewage treatment plants’ compartment. The findings indicated that none of the discharge sites required risk management interventions. While this conclusion does not serve as an exclusive risk assessment for CNFs, considering the comprehensive evaluation of nanocelluloses, it provides a certain level of confidence in the safety of CNFs. Future assessments of CNF safety require meticulous examinations focusing on the behavior of CNFs upon release into actual environmental matrices as a key safety metric. In the future, the collection of more precise data on CNFs may allow for a more detailed assessment of the safety of CNFs.

Conclusion

In this study, acute toxicity tests of TEMPO-CNFs to D. magna and O. latipes were successfully conducted according to the OECD TGs by maintaining CNF dispersion and concentration uniformity. No acute toxicity to either species was observed, even at the highest concentration tested in this study (100 mg/L). The results obtained herein showed that the interaction between the medium components and CNFs can affect the viscosity, zeta potential, concentration uniformity of the CNFs, and the ion concentrations in the test solution. The concentration uniformity of the CNFs decreased with increasing ionic levels in the test media. Thus, we demonstrated the importance of selecting a suitable test medium and measuring the CNF and ion concentrations in the test solution for ecotoxicity testing of CNFs. The findings and methods used in this study will be useful for future ecotoxicity tests using other species or CNFs.