Registration Dossier

Ecotoxicological information

Endpoint summary

Administrative data

Description of key information

The outcome of the environmental fate assessment demonstrated that different valence states simultaneously occur under normal environmental conditions, with the pentavalent and trivalent valence being the dominant form under oxic and reducing conditions, respectively.

The environmental effects assessment of Sb and Sb compounds, however, is based on measured concentration of dissolved Sb without taking the valence state into account. Adequate and reliable ecotoxicity data in this dossier are predominantly generated with either antimony trichloride or diantimony trioxide as test substance. Both are trivalent species of antimony. Upon release, a significant fraction of dissolved antimony will oxidize to the pentavalent state during the testing period. This transformation process is governed by several physicochemical properties of the test medium (pH, O2-concentration, presence of oxidizing substances in the medium, biological activity) and currently inufficiently understood to allow a reliable modeling and calculation of Sb-speciation during the testing period. Due to this dynamic equilibrium between trivalent and pentavalent antimony in the environment, it is in practice not possible to determine the acute or chronic ecotoxicity of a specific valence state as the final test solution will always contain both forms. Under normal conditions that are considered relevant for environmental hazard assessmtent (oxic conditions), is is expected that Sb(V) is the predominant form that will be found in the environment.

Skeaff et al (2013) demonstrated in a series of 28d-Transformation/Dissolution tests with several Sb-compounds that – under oxic conditions – Sb released as Sb(V) remained in this valence state, whereas Sb that was released as Sb(III) underwent a significant oxidation over the 28d-exposure period. This transformation was not completed after 28 days. Based on these findings can be concluded that:

  • Acute/chronic toxicity tests with an Sb(V)-substance as testing material reflect the toxicity of the pentavalent form;
  • Acute/chronic toxicity tests with an Sb(III)-substance as testing material reflect the toxicity of a mixture of both the trivalent and pentavalent form; given the duration of most ecotoxicity tests, equilibrium between both valence states may not have been reached.

The simultaneous presence of two different valence states is not critical for a hazard assessment if both are equally toxic; under those circumstances all Sb-related data (expressed as dissolved Sb) can be used for the risk/hazard assessment of Sb-compounds. If Sb(III) would be more ecotoxic than Sb(V), then an assessment that is based on Sb(III)-generated data would still be protective for both trivalent and pentavalent compounds. Only if Sb(V) would be more ecotoxic than Sb(III), it would not be possible to use ecotoxicological information that is generated with trivalent Sb-compounds (unless there is evidence that equilibrium between both valence states is reached at the start of the exposure period).

A review by Filella et al (2009) reported that there is a vast amount of publications stating that Sb(III) is more toxic than Sb(V). However, these publications did not provide no conclusive evidence for this assumption as such, or referred to Venugopal and Luckey (1978) who discussed the toxicity of both valence states to mammals. It should be noted that the latter paper does not irrefutably support the assumption that Sb(III) is more toxic than Sb(V). A recent study on the effect of Sb(III) and Sb(V) on the relative root elongation of rice seedlings after 48h of exposure resulted in EC50s of 24.3 and 67.9 µM for Sb(III) and Sb(V), respectively (Cui et al, 2015). This observation suggests that the trivalent valence state is about a factor of two more toxic than the pentavalent form. It should be noted, however, that inhibition of the root elongation at 20 µM was the same for both valence states (14%), and that an effect exceeding 50% was only noted at the highest test concentrations. 

No evidence has been identified in scientific literature for Sb(V) being more toxic than Sb(III) for the environment.

It can thus be concluded that the use of test data that were generated with trivalent Sb-compounds will be sufficiently protective for all Sb-compounds within a read-across approach: if equilibrium between Sb(III) and Sb(V) in the test medium is not reached at the time of exposure – and assuming that Sb(V) is less (or equally) ecotoxic than Sb(III) – then the outcome of such tests will not underestimate the toxicity of dissolved antimony at equilibrium. This effects assessment – based on tests with trivalent Sb-compounds - will thus refer to dissolved Sb without making reference to the valence state.

Additional information

There is scientific evidence that the toxicity of metals in the aquatic environment is determined by the dissolved fraction, and not by the total fraction. A correct interpretation of ecotoxicity tests therefore requires that effects data are based on measured Sbdissolved. Nominal concentrations are considered to be very misleading guides to the actual amount of dissolved antimony. At equal nominal concentrations, for instance, readily soluble compounds like SbCl3 result in more dissolved antimony during the exposure periods used in toxicity experiments than less soluble compounds like Sb2O3. Moreover, even the readily soluble compound SbCl3 may at high nominal concentrations result in lower amounts of soluble antimony. Brooke et al.(1986), for instance, reported that the use of high concentrations of SbCl3 dissolved in water resulted in precipitation of antimony. What is most likely being observed is an initial formation of chloroantimonate (III) species, which in aqueous solutions are weak, and which then hydrolyse to oxychloride (SbOCl), which has low solubility in water, and is further hydrolysed to Sb2O3 (Filella et al.,2002b; Filella and May, 2003). The percentage of Sb lost from solution by precipitation in the study by Brooke et al.(1986) during a 96-h test ranged from a low 6% for the lowest exposure concentration (nominally 25 mg Sb/L) to 76 % for the highest concentration (nominally 50 mg Sb/L). The apparent trend was to lose a greater percentage of antimony from solution with increasing nominal concentration. The authors improved the methods by introducing mixing and filtering to remove the precipitate from the solution. The maximum concentration maintained in a solution without organisms for 96 h was 35.0 mg Sb/L, using a nominal concentration of 250 mg Sb/L. Nominal bioassay concentrations at 50 mg Sb/L and 100 mg Sb/L using Sb2O3 resulted in measured soluble concentrations of 3.4 and 5.0 mg Sb/L, respectively. Solutions of Sb2O3 in laboratory water at nominal concentrations of 28, 58, and 110 mg Sb/L had measured dissolved concentrations during the 96 h experiment period of 1.9, 2.6, and 3.3 mg Sb/L, respectively.

Similar observations of the removal of some antimony species from solution over time have been observed in transformation/dissolution tests on antimony compounds in some cases (CanMET, 2010).

Filella and May (2003) conducted a critical review of all available thermodynamic data on antimony and developed a computer speciation model of antimony in multi-component solutions, representative of different environmental conditions. Based on the limited data set, their subsequent speciation calculations confirmed that antimony is exclusively present as the pentavalent Sb(OH)6 - in oxic freshwater systems and as the trivalent Sb(OH)3 in anoxic conditions, at all pH values of environmental relevance for aquatic systems. The formation of chloride-antimony species does not appear to be of importance under environmentally relevant conditions, as no Sb(III)-chloride was observed under seawater conditions, and the concentration of possible Sb(V)-chloride could not be calculated due to a lack of data. The very few studies available on Sb(V)-chloride binding had been performed under extremely acidic conditions to prevent hydrolysis and could thus not be used, as it was difficult to establish the strength of such interactions under dilute conditions relative to other antimony species. Therefore no thermodynamic relationship of this kind has been published.

Based on the available information on antimony there is nothing that indicates that the difference observed in toxicity in aquatic systems between different inorganic antimony compounds of the same valence, such as for instance SbCl3 and Sb2O3, would be due to different antimony species exerting different degrees/kinds of toxicity. Instead, an observed difference in toxicity at equal nominal doses of antimony is most probably a reflection of differences in solubility, which means that a more soluble antimony compound will result in more dissolved antimony capable of exerting toxicity. However, it may be that higher concentrations of Sb compounds also reflect an increased presence of counter ions and/or protons.

Toxicity studies in the aquatic compartment in which only nominal antimony concentrations are reported will therefore not be considered to be reliable in this assessment report. However, studies will not be rejected based solely on which antimony compound was used in the test, or whether or not a tri- or a pentavalent compound was used, as long as the results are considered reliable and relevant. All effect and no-effect concentrations are reported as a concentration of the antimony ion.

Six studies provide valid EC50s and NOECs for fish, invertebrates and algae: Brooke et al. (1986), Kimbal (1978), TAI (1990), Heijerick and Vangheluwe (2003), Heijerick and Vangheluwe (2004), and LISEC (2001).

Toxicity results for marine species are scarce. Only the results from Takayanagi (2001) are considered reliable.

None of the NOECs used to derive the PNECs in the aquatic compartment are considered to be confounded by the additions of counter ions (i. e. chloride) and/or protons resulting from the use of SbCl3.