Aluminium oxide

Endpoint summary

Description of key information

Reliable aluminum toxicity studies represent tests with aquatic organisms conducted over a pH range from 6 – 8 as being representative of conditions in most European surface waters as evaluated in a preliminary exposure assessment (EURAS 2007).

Most of the short-term studies included were only those used for purposes of characterization and to evaluate potential effects of water quality on Al toxicity.

 

The long-term chronic studies in this dossier represent all known studies of sufficient reliability according to Klimisch criteria levels “1” (reliable without restriction) and “2” (reliable with restriction; Klimisch et al. 1997). These data are being assessed with a view towards developing a Probable No Effect Concentration (PNEC) principally for use in setting Environmental Quality Standards (EQS) values. It should be noted that aluminum chloride was not classified for the aquatic environment by the EU Classification and Labeling Committee, therefore other less soluble forms of Al such as the oxides, powders, and massive metal would also not classify for the aquatic environment. Therefore, no PNEC is required for REACH purposes.

A justification for using data of soluble Al-compounds for the environmental effects assessment of sparingly soluble Al-compounds is provided in the RAAF document which is attached to Section 13 of the dossier.

Additional information

The long-term goal of the aluminum industry was to develop an aquatic PNEC that covers the pH range of < 6, 6-7, and > 7 for the purpose of future water quality standards. These pH ranges reflect that the form of Al present in water changes significantly as a function of pH. Scientists around the world have worked towards a PNEC for Al for the past 40 years. Over the past twenty years, the aluminum industry has utilized several decades of work to develop and improve a biotic ligand model (BLM) for fish, invertebrates and algae for pH values across the range of 5.5-8.0. The current state of the model is summarized in the Background document which is attached to the dossier (see IUCLID Section 13.2). It is recognized that there is still a need to further demonstrate and validate that this model is applicable to a broader range of aquatic organisms. For the majority of the aquatic toxicity studies, endpoints are expressed as a function of total Al, rather than as dissolved or monomeric Al. This is because for most test solutions with pH from 6 – 8, Al will be largely insoluble, and so dissolved and monomeric concentrations remain relatively constant even with large increases in total or nominal Al. Thus, dose-dependent responses observed by aquatic organisms can only be reliably quantified using total Al across the full pH range from 5.5-8.0.

 

It is important to point out that the substance being registered in this dossier is a sparingly soluble form of aluminum and not a soluble metal salt. Therefore any review of the aquatic toxicity literature where aluminum salts were used for assessing the toxicity of the metal ion, needs to take into account that reported (no)-effect levels may not be relevant for substances with a (very) low solubility. The use of a PNEC, once derived, has to take into consideration the transformation/dissolution of the sparingly soluble aluminum compounds. There is little or no evidence that aluminum metal, metal powders or metal oxides have ever resulted in aquatic toxicity effects.

 

The aluminium (Al) toxicity studies compiled in this section represent tests with aquatic organisms conducted over a pH range from 6 – 8 as being representative of conditions in most European surface waters as evaluated in a preliminary exposure assessment (EURAS 2007). Most of the short-term studies included were only those used for purposes of characterization and to evaluate potential effects of water quality on Al toxicity. The long-term chronic studies represent all known studies of sufficient reliability according to Klimisch criteria levels one (reliable without restriction) and two (reliable with restriction; Klimisch et al. 1997). These data are being assessed with a view towards developing a Probable No Effect Concentration (PNEC) principally for use in setting Environmental Quality Standards (EQS) values. We point out that aluminium chloride was not classified for the aquatic environment by the EU Classification and Labeling Committee, therefore other less soluble forms of Al such as the oxides, powders, and massive metal would also not classify for the aquatic environment (see reference below). Therefore, no PNEC is required for REACH purposes. The long term goal of the industry is to develop an aquatic PNEC that covers the pH range of < 6, 6-7, and > 7. These pH ranges reflect that the form of Al present in water changes significantly as a function of pH. PNEC values calculated in this dossier are preliminary and reflect the state of the science to date. Scientists around the world have worked towards a PNEC for Al for the past 30 years. Over the last decade, the aluminum industry has utilized several decades of work to develop a biotic ligand model (BLM) fish, invertebrates and algae for pH values across the range of 5.5-8.0. The state of the model is summarized in this CSR. It is recognized that there is a need to demonstrate that this model applies to a broader range of aquatic organisms. 

 

 For the majority of the aquatic toxicity studies, endpoints are expressed as a function of total Al, rather than as dissolved or monomeric Al. This is because for most test solutions with pH from 6 – 8, Al will be largely insoluble, and so dissolved and monomeric concentrations remain relatively constant even with large increases in total or nominal Al. Thus, dose-dependent responses observed by aquatic organisms can only be reliably quantified using total Al across the full pH range from 5.5-8.0.

 

It is important to point out that the substances being registered in this dossier are sparingly soluble forms of aluminium and not soluble metal salts. Therefore the review of the aquatic toxicity literature where aluminium salts were used for assessing the toxicity of the metal ion or hydroxide have limited direct application to the substances represented in this CSR. The use of a PNEC, once derived, has to take into consideration the transformation/dissolution of the sparingly soluble aluminium compounds. There is little or no evidence that aluminium metal, metal powders or metal oxides have ever resulted in aquatic toxicity effects.

 

 

Biotic Ligand model for Aluminium

Toxic effects of aluminium have been observed in several types of aquatic organisms under certain exposure conditions. Factors that influence aluminium toxicity are consistent with the factors that influence aluminium speciation. These factors include pH, dissolved organic matter concentration (DOC), and water hardness (Roy and Campbell 1997; Gensemer and Playle 1999). Fluoride has also been shown to influence aluminium toxicity (Hamilton and Haines 1995), though fluoride is not commonly found at elevated levels in the environment. Several studies have demonstrated that some forms of aluminium are only bioavailable and potentially toxic in freshly prepared solutions, and that this toxicity declines or is eliminated after several minutes of aging (e.g. Exely et al. 1996; Witters et al. 1996; Teien et al. 2006). Toxicity in these cases may depend on short-lived transient chemical forms of aluminium hydroxide whose environmental relevance would be restricted to mixing zones where aluminium-rich acidic waters mix with a more alkaline water.

 

A biotic ligand model (BLM) was developed to address the bioavailability and toxicity of dissolved, particulate, and transient forms of aluminium. Application of the BLM framework to understanding aluminium toxicity was reasonable because many of the factors that influence aluminium bioavailability are consistent with the factors that influence aluminium speciation or forms in the environment. As with BLMs for other metals, the Al BLM combines information about chemical speciation and interaction with gill surfaces to explain and predict aluminium bioavailability and toxicity (DiToro et al, 2001; Santore et al 2001; Paquin et al 2002). Factors that affect aluminium bioavailability by altering the chemical speciation of the metal (such as DOC, pH, and fluoride) are directly considered by the speciation model (Tipping 1994; Santore and Driscoll, 1996).  Other factors (such as hardness cations), affect aluminium bioavailability by competing with gill binding sites in a manner similar to what has been observed for other metals (Playle et al 1992; Meyer et al, 1999) and are considered by including interactions for these cations with the BL sites on the gill. The detailed speciation within the aluminium BLM allows the model to predict bioavailability for a number of different aluminium fractions. Depending on available input data, the model can be run with monomeric, dissolved, or total aluminium as the primary input parameter, and the distribution among dissolved species and precipitated forms can be simulated by the model. Comparison of predicted and measured distribution of aluminium fractions in waters where aluminium toxicity has been extensively studied typically shows very good agreement.

 

Factors that are known to affect speciation have also been shown to affect bioavailability and toxicity in both acute and chronic exposures, and the consistency of these affects in different exposure durations allows a common model framework for prediction of both acute and chronic affects. Acute data were useful in model development due in part to the large amount of available data that combined coincident measurement of detailed speciation measurements, measures of Al accumulation in gills, and observation of lethal and sub-lethal effects over wide ranges of water chemistry. Data for development of the Al BLM included Atlantic salmon (Salmo salar) and brown trout (Salmo trutta) from studies performed by NIVA and collaborators from UMB (Kroglund et al. 1997; Kroglund et al. 1998a,b,c,d; Erstad et al. 2002; Teien et al. 2004a,b; Teien et al. 2006; Andren et al. 2006). These studies typically investigated the effects of water chemistry on the accumulation of Al on/in the gills of S. salar and S. trutta, but in some cases, mortality was reported. Many of these studies purposefully investigated the effects of water chemistry on the level of Al accumulation in S. salar and S. trutta gills. The pH conditions varied from approximately pH 5 (Andren et al. 2006) to pH 10 (Erstad et al. 2002). The total organic carbon concentrations (TOC) ranged from approximately 0.5 mg/L to 16 mg/L (Kroglund et al. 1998a,b, and Erstad et al. 2002, respectively). Calcium concentrations ranged from approximately 1 mg/L to 11 mg/L (Kroglund et al. 1998a,b, and Erstad et al. 2002, respectively).

 

From these data it was clear that observed toxicity was strongly related to aluminium accumulation on the gill. The calibrated Al BLM was able to reasonably predict the level of Al accumulation on the gills of S. salar and S. trutta, with one consistent set of BLM parameters (Figure 7.1.1.-3) over a range of approximately 2 orders of magnitude. This wide range in gill accumulation was primarily due to the diverse water chemistry conditions tested, and suggests that the BLM is relatively robust over this wide range of conditions. These data were also used to estimate critical Al accumulation levels for those datasets that reported associated mortality data (i.e. Suldal Fall 1997 – Kroglund et al. 1998a). For example, critical accumulation levels corresponding to the LC10 and LC50 values for mortality could be derived as 2995 and 4225 nmol/g wet weight, respectively.   

 

Although data from acute exposures were extensively used to parameterize the prediction of gill-accumulation over a wide range of conditions, the goal in model development is the evaluation of the ability of the Al BLM to predict effects in chronic exposures. The use of data from both acute and chronic exposures in model development is justified by the consistency of the observed effects of changing water chemistry (such as pH, NOM, and hardness) in both acute and chronic exposures. Adjustment of the Al BLM for different exposure durations (acute versus chronic) and endpoints (lethal or sub-lethal) is primarily accomplished by adjustment of the critical accumulation level for each endpoint and exposure condition. 

 

 

The chronic BLMs that were developed for fish (P. promelas), invertebrates (C. dubia) and algae (P. subcapitata) were first used to normalize all retained individual chronic NOEC values of respectively fish, invertebrates and algae/plant species. Briefly, the bioavailability normalization process normalizes the ecotoxicity data to sets of standard physico-chemical conditions for important abiotic factors (i.e., pH, hardness, and dissolved organic carbon (DOC)). This approach allows for the comparison of intrinsic toxicity among organisms on an equal basis.

 

In analogy to risk assessment exercises for other metals (e.g. Cu, Ni) normalization were carried out towards seven EU ‘eco-region scenario’s’, selected to include a range of typical cases of bioavailability and to encompass the 10th/90th percentile of the DOC, pH and hardness for surface waters. Normalization was done at a fixed temperature of 20°C The normalization of the NOECs with the BLMs allowed to obtain robust and meaningful species-specific NOEC and HC5,50% values.

 

 

More detailed information on the Al-BML, as well as the application of the BLM in deriving ecoregion-specific water quality standards, is provided in the report “Background document – Environmental Effects Assessment of Aluminium” which is attached to Section 13 of the IUCLID Dossier.