Effects of organic modifiers on the colloidal stability of TiO2 nanoparticles. A methodological approach for NPs categorization by multivariate statistical analysis
Graphical abstract
Introduction
Manufactured nanoparticles (NPs) are being used in a wide variety of industrial applications and consumer products (Piccinno et al., 2012). However, the high heterogeneity of novel nanoforms released on the market has made their safety assessment very demanding in terms of testing. To reduce this high regulatory burden of proof of the nanotechnology industry it has been suggested to employ in silico modelling as well as grouping and read-across approaches to enable safety by design (SbD) strategies that target the early stages of product innovation (Hutchison, 2016). This is challenging as the physicochemical identity of the nanomaterials can be easily affected upon contact with any biological, environmental or industrial dispersion media. One of the most frequently observed phenomena is agglomeration of the NPs in the medium as a result of e.g. its chemical composition, pH, ionic strength, dissolved concentration of oxygen and sulphide, light, suspended particle matter, or content of natural organic matter. Thus, changes in the size distribution, shape, surface area and charge of the agglomerated NPs can be frequently observed, maybe varying their industrial functionality, exposure potential, and/or adverse (eco)toxicological effects. These fast and unpredictable modifications pose challenges not only to the safety assessment of these materials, but also to the reproducibility of product performance, which are major barriers to nanotechnology innovation.
Therefore, understanding how the interactions between NPs and the surrounding medium can alter their colloidal dispersion stability is essential not only to predicting their risks, but also to developing SbD strategies (Sharma et al., 2014) that can prevent these risks early in the R&D process (Ortelli et al., 2017). Specifically, elucidating the NP-medium interaction can help to derive descriptors for in silico and materials modelling of both properties and effects and to design in vitro (eco)toxicological tests as part of Intelligent Testing Strategies that aim at reducing testing costs and the use of animal experiments. It can also help in the better interpretation of the modelling/testing results (Canesi and Corsi, 2016) to derive criteria and guiding principles for grouping and/or read-across and for classification according to regulatory requirements and industrial product quality criteria.
To contribute to the above priorities, the goal of this paper is to investigate the influence of surface modification on the extrinsic properties of the NPs, defined as the “characteristics that are linked to the material's functionality in its environment” (Arts et al., 2016), e.g. agglomeration, surface charge, dispersibility etc. Indeed, the approach employed and the outcomes achieved by this work are not intended to replace the huge efforts already carried out on describing methods as well as standardized and validated protocols for synthesis, purification, and characterization of nanomaterials (Hühn et al., 2017; Rasmussen et al., 2018) but rather to support nanomaterials categorization within relative stability classes by combining easy-to-use analytical and statistical techniques.
Our case study is nanoscale titanium dioxide (TiO2), which was selected due to its widespread use in many consumer products, very low solubility, and surface which can be easily modified (Mitrano et al., 2015; Gonçalves et al., 2010). Specifically, we used different modifying substances: catecholate derivatives (i.e. catechol, 3,4-dihydroxybenzaldehyde, 3,4-dihydroxybenzoic acid, dopamine hydrochloride), salicylic acid (SAL), and polyethylene glycol (PEG), exploiting the optimal geometry of these ligands to get covalently linked to the NPs' surfaces. The catecholate-type ligands were chosen because of their versatile chemistry, which allowed easier attachment of different functional groups, leading to new optically active nanomaterials (Savić et al., 2014) as well as to fundamental building blocks for the synthesis of more complex architectures (Kobayashi and Arai, 2017; Burger et al., 2015; Wei et al., 2014). Salicylic acid was chosen for its similarity to catechols in terms of structure, functional groups, and way of binding to TiO2 surface. The surface modification with PEG was performed because polymeric coatings are considered one of the main approaches to effectively control physicochemical properties such as size, surface charge and solubility, all of which are parameters known to determine the toxicokinetics and toxicity of nanomaterials (Selli and Di Valentin, 2016).
Once the surfaces of the materials were functionalized, the investigation of the stability of colloidal dispersions, which by definition is defined in terms of a change in one or more physical properties over a given time period (ISO. ISO/TR 13097, 2013), was assessed in different dispersion media varying electrolyte concentrations and pH levels, by combining Electrophoretic Light Scattering (ELS), Dynamic Light Scattering (DLS) and Centrifugal Separation Analysis (CSA) techniques. The obtained data were analysed through statistical clustering methods and Principal Component Analysis (PCA) (Bishop, 2006). Clustering have been already employed to assist the development of (Q)SAR models for nanomaterials (Fourches et al., 2010; Epa et al., 2012; Fourches et al., 2016), and as a tool for grouping NPs into different toxicity classes, which were used to predict toxicity of untested materials (Gajewicz et al., 2015). As far as PCA, it was previously applied for nanomaterials classification (Sayes et al., 2013; Wang et al., 2014) as well as for quality assessment of nano-based dispersions (Tantra et al., 2011). In this work, clustering was adopted to subdivide the dataset into categories of samples showing similar stability, while PCA was used to display in a bi-dimensional space the obtained classification into high-, moderate- and low-stability dispersions, and to understand which extrinsic properties affected the most this categorization. This approach is one of the first attempts to in silico modelling the colloidal stability of TiO2 NPs, and it could be a useful starting point for developing SbD strategies.
Section snippets
Case-study nanomaterial and other reagents
The inorganic Aeroxide® P25 titanium dioxide nanopowder was purchased from Evonik Degussa (Germany). P25 powder (declared average particle size: 21 nm) is a mixture of approx. 80% anatase and 20% rutile, with 99.5% purity. According to our previous work (Brunelli et al., 2013), P25 pristine powder showed a size distribution ranging approx. from 10 to 65 nm, with a shape partly irregular and semi-spherical, 50 ± 15 m2/g as surface area, and a bulk density of 3.8 g/cm (Sharma et al., 2014).
Binding of organic ligands to P25 NPs surface
The coating of P25 NPs by chemisorption of the ligands selected was investigated by ATR-FTIR and TGA-DSC analysis.
The ATR-FTIR spectra of catechol free and adsorbed on P25 NPs are displayed in Fig. 2, as a zoom-in image of the wavelength region between 1800 and 1000 cm−1. The main bands of free catechol (Fig. 2a) are the following: stretching vibration of the aromatic ring ν(CC)/ν(CC) at 1618, 1600, 1512, 1467 cm−1 and stretching of phenolic group ν(COH) at 1278, 1254 and 1237 cm−1, while the
Conclusions
The work herein presented is one of the first studies employing multivariate statistical analysis methods to categorize experimental data of NPs dispersions into relative stability classes. The study highlighted that even small modifications of the NPs' surfaces can affect their colloidal stability toward the investigated parameters (i.e. dispersion media composition, pH, and electrolyte concentration). The performed statistical analyses helped to derive conclusions on the relationships of
Conflict of interest statement
We declare that we have no financial and personal relationships with other people or organizations that can inappropriately influence our work, there is no professional or other personal interest of any nature or kind in any product, service or company that could be construed as influencing the position presented in, or the review of, the manuscript entitled.
Acknowledgements
The authors are grateful to the European Commission for funding SUN project (FP7-NMP-2013-LARGE-7, Grant Agreement N° 604305) and University Ca’ Foscari of Venice for a post-doc cofunding (E.B.).
References (50)
Case studies putting the decision-making framework for the grouping and testing of nanomaterials (DF4nanoGrouping) into practice
Regul. Toxicol. Pharmacol.
(2016)DLS and zeta potential - what they are and what they are not?
J. Control. Release
(2016)- et al.
Effects of nanomaterials on marine invertebrates
Sci. Total Environ.
(2016) - et al.
Design and construction of self-assembling supramolecular protein complexes using artificial and fusion proteins as nanoscale building blocks
Curr. Opin. Biotechnol.
(2017) pH-dependent surface charging and points of zero charge. IV. Update and new approach
J. Colloid Interface Sci.
(2009)- et al.
TiO2 nanoparticles aggregation and disaggregation in presence of alginate and Suwannee River humic acids. pH and concentration effects on nanoparticle stability
Water Res.
(2013) - et al.
Breakage of TiO2 agglomerates in electrostatically stabilized aqueous dispersions
Powder Technol.
(2005) - et al.
Review of nanomaterial aging and transformations through the life cycle of nano-enhanced products
Environ. Int.
(2015) Physico-chemical properties of manufactured nanomaterials - characterisation and relevant methods. An outlook based on the OECD Testing Programme
Regul. Toxicol. Pharmacol.
(2018)- et al.
Organic-coated silver nanoparticles in biological and environmental conditions: fate, stability and toxicity
Adv. Colloid Interf. Sci.
(2014)
Stability of commercial metal oxide nanoparticles in water
Water Res.
Evaluation of the stability of concentrated emulsions for lemon beverages using sequential experimental designs
PLoS One
Standard practice for the preparation of substitute ocean water
Pattern Recognition and Machine Learning (Information Science and Statistics)
Agglomeration and sedimentation of titanium dioxide nanoparticles (n-TiO2) in synthetic and real waters
J. Nanopart. Res.
Extrapolated long-term stability of titanium dioxide nanoparticles and multi-walled carbon nanotubes in artificial freshwater
J. Nanopart. Res.
Layer-by-layer assemblies of catechol-functionalized TiO2 nanoparticles and Porphyrins through electrostatic interactions
Chem. - A Eur. J.
Modeling biological activities of nanoparticles
Nano Lett.
Quantitative nanostructure−activity relationship modeling
ACS Nano
Computer-aided design of carbon nanotubes with the desired bioactivity and safety profiles
Nanotoxicology
Novel approach for efficient predictions properties of large pool of nanomaterials based on limited set of species: Nano-read-across
Nanotechnology
Synthesis of TiO2 Nanocrystals with a high affinity for amine organic compounds
Langmuir
Comparing partitions
J. Classif.
Selected standard protocols for the synthesis, phase transfer, and characterization of inorganic colloidal nanoparticles
Chem. Mater.
The road to sustainable nanotechnology: challenges, progress and opportunities
ACS Sustain. Chem. Eng.
Cited by (14)
An overview on dispersion procedures and testing methods for the ecotoxicity testing of nanomaterials in the marine environment
2024, Science of the Total EnvironmentCytotoxicity inhibition of catechol's type molecules by grafting on TiO<inf>2</inf> and Fe<inf>2</inf>O<inf>3</inf> nanoparticles surface
2022, Aquatic ToxicologyCitation Excerpt :In order to investigate the interaction between catecholate type ligands and both TiO2 and Fe2O3 NPs potentially occurring in the aquatic environment, the new nanomaterials formed according to the procedure described above were characterized from the physicochemical and ecotoxicological point of view, and the results obtained were compared with those obtained for both pristine NPs and free organic ligands as follows. The chemisorption of the selected molecules to both NPs was investigated by means of FTIR-ATR and TG-DSC (Fe2O3 NPs spectra are reported in Figures S0-S12, while TiO2 NPs spectra are reported in Brunelli et al., 2018). In the case of FTIR-ATR analysis, the spectra of the modified NPs were compared with those of free organic ligands.
LED light-induced enhanced photocatalytic hydrogen evolution on Au@TiO<inf>2</inf> core–shell modified by nitrogen doping and reduced graphene oxide
2022, International Journal of Hydrogen EnergyCitation Excerpt :The activity of Au loaded photocatalysts depends largely on their stability and morphology and size of Au NPs [21,22]. Cheng et al. reported that the Au/TiO2 photocatalysts with Au NPs of size 3–5 nm produced twice amount of hydrogen than the Au NPs of size larger than 28 nm [23]. The smaller size of Au NPs reduced the height of Schottky barrier, enhancing the charge separation through the Schottky transfer hub to the TiO2.
Colloidal stability classification of TiO<inf>2</inf> nanoparticles in artificial and in natural waters by cluster analysis and a global stability index: Influence of standard and natural colloidal particles
2022, Science of the Total EnvironmentCitation Excerpt :Transmission profiles were recorded every 5 s (41 min of runtime) at 470 nm, T = 25 °C, and rotation per minute (RPM) = 2000, which corresponds to 537 relative centrifugal force (RCF) at 120 mm far from the rotor of the centrifuge. The linear dependency between RCF and V-sed already demonstrated in several works by our research group (Badetti et al., 2021; Brunelli et al., 2018, 2016) allowed to estimate V-sed data at gravity by dividing the sedimentation velocity values calculated by the instrument for the RCF applied. All the measurements were performed in triplicate and the results are expressed as median.
Blueprint for a self-sustained European Centre for service provision in safe and sustainable innovation for nanotechnology
2021, NanoImpactCitation Excerpt :These could be in the form of QSAR models exploring mechanistic information to provide a prediction of the relationship between descriptors (structural, physico-chemical, biological) and the investigated endpoint (environmental fate, transformation, reactivity, toxicity). In this context, various chemoinformatic methods using artificial neural networks, machine learning, multivariate statistical modelling etc. have demonstrated added value e.g. in predicting activity of nanostructured fullerenes (Fjodorova et al., 2020), assessing the environmental fate of TiO2 nanoparticles (Brunelli et al., 2018) or modelling nanotoxicity (Furxhi et al., 2020; Afantitis et al., 2018), or for SbD functionalisation of carbon nanotubes (Varsou et al., 2019). The use of computational models for predicting biokinetics, environmental fate, exposure levels and toxicological effects of manufactured NMs would also bring significant benefits e.g. to prioritise extended experimental studies requiring dedicated infrastructure; to reduce experimental effort and costs regarding the need for experiments; to anticipate 3R principles for reducing animal testing as a major priority for the EC (European Commission, 2019).
- 1
These authors contributed equally.