Synthesis and characterization of Zr- and Hf-doped nano-TiO2 as internal standards for analytical quantification of nanomaterials in complex matrices

The reliable quantification of nanomaterials (NMs) in complex matrices such as food, cosmetics and biological and environmental compartments can be challenging due to interactions with matrix components and analytical equipment (vials and tubing). The resulting losses along the analytical process (sampling, extraction, clean-up, separation and detection) hamper the quantification of the target NMs in these matrices as well as the compatibility of results and meaningful interpretations in safety assessments. These issues can be overcome by the addition of known amounts of internal/recovery standards to the sample prior to analysis. These standards need to replicate the behaviour of target analytes in the analytical process, which is mainly defined by the surface properties. Moreover, they need to carry a tag that can be quantified independently of the target analyte. As inductively coupled plasma mass spectrometry is used for the identification and quantification of NMs, doping with isotopes, target analytes or with chemically related rare elements is a promising approach. We present the synthesis of a library of TiO2 NMs doped with hafnium (Hf) and zirconium (Zr) (both low in environmental abundance). Zirconia NMs doped with Hf were also synthesized to complement the library. NMs were synthesized with morphological and size properties similar to commercially available TiO2. Characterization included: transmission electron microscopy coupled with energy-dispersive X-ray spectroscopy, X-ray diffraction spectroscopy, Brunauer–Emmett–Teller total specific surface area analysis, cryofixation scanning electron microscopy, inductively coupled plasma optical emission spectroscopy and UV–visible spectrometry. The Ti : Hf and Ti : Zr ratios were verified and calculated using Rietveld refinement. The labelled NMs can serve as internal standards to track the extraction efficiency from complex matrices, and increase method robustness and traceability of characterization/quantification.


L-JAE, 0000-0003-2781-4509
The reliable quantification of nanomaterials (NMs) in complex matrices such as food, cosmetics and biological and environmental compartments can be challenging due to interactions with matrix components and analytical equipment (vials and tubing). The resulting losses along the analytical process (sampling, extraction, clean-up, separation and detection) hamper the quantification of the target NMs in these matrices as well as the compatibility of results and meaningful interpretations in safety assessments. These issues can be overcome by the addition of known amounts of internal/recovery standards to the sample prior to analysis. These standards need to replicate the behaviour of target analytes in the analytical process, which is mainly defined by the surface properties. Moreover, they need to carry a tag that can be quantified independently of the target analyte. As inductively coupled plasma mass spectrometry is used for the identification and quantification of NMs, doping with isotopes, target analytes or with chemically related rare elements is a promising approach. We present the synthesis of a library of TiO 2 NMs doped with hafnium (Hf) and zirconium (Zr) (both low in environmental abundance). Zirconia NMs doped with Hf were also synthesized to complement the library. NMs were synthesized with 2018 The Authors. Published by the Royal Society under the terms of the Creative Commons Attribution License http://creativecommons.org/licenses/by/4.0/, which permits unrestricted use, provided the original author and source are credited.

Introduction
Titanium dioxide (TiO 2 , also known as titania) nanomaterials (NMs) below 100 nm have gained increased interest for their incorporation into cosmetic products [1] and other industrial products (e.g. paints and surface coatings) [2], particularly the rutile and anatase crystalline structures. The use of TiO 2 NMs is based on their photoelectronic properties and their ability to photoactivate with UV radiation and act as antibacterial and/or antifouling agents in windows, pavements, walls and roofs [2], or block sunlight when used in cosmetics and help formulate a sun protection factor [1,3]. The type of TiO 2 NM (rutile and anatase) to be used in each case is defined from their crystal structure, as anatase NMs present higher density and lower refractive index and are more photoactive than rutile NMs [4] (although these properties can be modified either through the production process or intentionally through the synthesis protocol used) [4,5]. As a result, anatase NMs are mainly used in infrastructure applications, while rutile NMs are used in sunscreens and cosmetics.
However, nanoscale properties have been linked to potential increased toxicity [6] and, therefore, nanoparticle inclusion in sunscreens has prompted research to identify potential health hazards and/or environmental impacts from their use [7]. This is particularly important because increasing evidence supports that nano-TiO 2 properties differ substantially from those of their larger counterparts [8]. TiO 2 is a photocatalyst and upon excitation can produce hydroxyl radicals and other reactive oxygen species (ROS) that can lead to DNA damage [1], although this same attribute is desirable when TiO 2 NMs are used as antifouling coatings, e.g. in windows [9]. At the same time, the presence of dopants inside the TiO 2 crystal structure can lead to undesirable effects through potential dissolution and the release of Ti 4+ and other metals. Such releases may be potentially hard to identify, especially when Ti 4+ and the potential dopants are naturally occurring elements (e.g. Ag). The International Agency for Research on Cancer (IARC) has classified TiO 2 as a group 2B possible carcinogen to humans for inhalation exposure [10]. Irrespective of risk, European legislation, as well as that of other jurisdictions including the USA, regulates that the maximum concentration of nano-TiO 2 that can be added to sunscreen products is 25% weight [3,11]. In the European Union (EU), since 2013 the use of nanosized ingredients in cosmetic products must be indicated on the product label [12]. Despite restrictions, with recreational uses, rutile TiO 2 NMs will inevitably be dispersed into the environment [13,14]. Additional environmental release will also come from the widespread use of larger-scale pigment grade titania (median size approx. 200 nm) in paints, and anatase-based nanocoatings on external walls of buildings (facades, bricks and concrete), pavements and windows caused by environmental weathering [15,16].
It is essential that the potential exposure routes, kinetics, environmental fate and behaviour of manufactured NMs are fully evaluated and can be monitored if necessary [17]. This knowledge can be used for the risk assessment of nano-TiO 2 and NMs in general to study and potentially decrease hazardous effects on human, animals and the environment in general. This knowledge can then be applied in safer-by-design (SbD) principles and the design and production of more stable and less hazardous TiO 2 nanoproducts. To do so, relevant reference materials are needed that include a variation of sizes and surface functionalization for in vitro, mesocosm and environmental studies, which are currently absent from the literature [18]. TiO 2 is highly abundant in the environment and can be found in forms chemically and structurally indistinguishable from its engineered counterparts [19], while the same is true for potential dopants used to modify their physico-chemical and photoelectronic properties [20]. Thus, there is urgency for manufactured TiO 2 to be produced in a safer form, which also allows it to be identified against an environmental background. Therefore, doped titania particles, with modified photocatalytic properties, can be used as tracers to improve identification method robustness, and increase quantification, reliability and traceability by being distinguishable from their natural counterparts. The EU FP7 project NanoDefine was launched in 2013 and ended in October 2017. The project aimed to develop cost-efficient screening methods and suitable reference materials to support the implementation of the EU Recommendation on the Definition of Nanomaterial [21][22][23]. More information can be found about the project at www.nanodefine.eu. In 2014, the NanoDefine consortium identified the need for more validated information on standard reference materials [21]. To address the absence of suitable reference materials, BAM (in partnership with NanoDefine) now provides an expanding database of 'Nanoscaled Reference Materials', accessible via: http://www.nano-refmat.bam.de/en/ in cooperation with the International Organization for Standardization (ISO)/TC 229 Nanotechnologies.
In partnership with the NanoDefine consortium, we obtained cosmetic materials and samples of powder TiO 2 , known to be used in commercial cosmetic and industrial products, which could be released into the environment, for characterization. The intention was to obtain the morphology, size and physico-chemical properties of the TiO 2 NMs as a foundation to allow production of suitable reference materials with systematic variation of their physico-chemical properties. In addition, the challenges of extracting of TiO 2 NMs from cosmetics for characterization were also explored using different sample extraction methods. Characterization was carried out by transmission electron microscopy (TEM) coupled with energy-dispersive X-ray spectroscopy (EDS), X-ray diffraction (XRD) analysis and Brunauer-Emmett-Teller (BET) surface area analysis.
In this paper, we present results from the development and optimization of two synthesis methods, hydrolysis/oxidation and hydrothermal synthesis, aimed at producing a library of titania NMs with a range of surface treatments (including alumina, acetate and stearate) and doping agents (including hafnium (Hf) and zirconium (Zr)). These methods are reproducible and the final products mimic particles/materials commercially available for use in industrial, sunscreen and other cosmetic products. Moreover, because Zr and Hf are chemically similar to titanium (Ti), the oxides of these elements may potentially be synthesized using the same protocols as TiO 2 [24,25] or incorporated into TiO 2 as dopants. At the same time, Zr and Hf NMs present either similar [26] or lower [27,28] toxicity than TiO 2 NMs during in vitro experiments, respectively. Given their lower abundance in the environment compared with Ti [29], they could be used as tracers and assist the development of relevant risk assessment protocols. We also synthesized ZrO 2 NMs doped with Hf as an additional reference material that could serve as a complement to nano-TiO 2 tracers in environmental samples and for environmental fate studies.

Materials
All chemicals and solvents were purchased from Sigma-Aldrich (Dorset, UK) and were of analytical reagent grade. Ultra-pure water (UPW) with a maximum resistivity of 18.2 MΩ · cm was used throughout the experiments.

Extraction of TiO 2 from sunscreen products and TiO 2 nanopowders
To produce suitable labelled TiO 2 particles, the NanoDefine consortium provided two cosmetic samples termed BAM-13A and BAM-13B, as cited in their Technical Report D2.4 [22] for characterization purposes and to serve as model materials. Sample BAM-13A is a complete formulated sample containing 4% nano-TiO 2 with an aluminium (Al) salt-based surface coating, micro-TiO 2 and iron oxides (for colouring purposes). Sample BAM-13B is a simplified formula containing only the 4% nano-TiO 2 (same particles as BAM-13A), with the Al salt-based surface coating [22]. Prior to characterization, the TiO 2 NMs were removed from the formulated samples using a solvent extraction process. Briefly, 3 g of each sample was mixed with 15 ml of either ethanol, methanol or acetone to remove the cosmetic matrix (such as emulsifiers, paraffin and silicones). The dispersions were vortex mixed; this was followed by centrifugation (5000 r.p.m. for 20 min). The pellet was resuspended with the solvent and the process was repeated three times to ensure all organic components were removed, as reported in the literature [30]. Finally, the solid pellet was removed and left to air-dry for 24 h to obtain the metal oxide powder for characterization. In addition to the samples provided by the NanoDefine consortium and in order to maximize the relevance of the work, a number of samples of TiO 2 powders for commercial use were obtained for characterization from www.ulprospector.com. These included: All obtained samples were characterized to identify size, composition, morphology and surface area, and were subsequently used as a guide for the in-house synthesized TiO 2 and ZrO 2 reference NMs. For reference, these samples will be referred to as the commercial TiO 2 powders for the rest of the paper.

Characterization of TiO 2 NMs
To assess composition, size, crystalline structure and purity, complete and simple formula samples (NanoDefine), commercial TiO 2 powders and in-house synthesized TiO 2 and ZrO 2 reference NMs were characterized by means of TEM, EDS, XRD, BET, scanning electron cryomicroscopy (cryo-SEM) and inductively coupled plasma optical emission spectroscopy (ICP-OES).
TEM analysis was performed using a JEOL 1200EX 80 kV Max system. All sample dispersions were prepared by suspending 0.01 g of the TiO 2 solids into 5 ml UPW. TEM grids were prepared by a dropcasting method, whereby a 20 µl drop of TiO 2 suspension was deposited on a 300 mesh carbon-coated copper TEM grid (Agar Scientific, UK). The TiO 2 suspension drop was left for approximately 30 min to allow the NMs to adhere to the carbon membrane, and grids were rinsed with UPW to remove excess water to avoid aggregation. EDS analysis was carried out using a JEOL 2100 TEM fitted with an Oxford Inca EDS, with an accelerating voltage of 200 kV. Particle diameter measurements were conducted using the Gatan Digital Micrograph software by measuring at least 100 particles.
Cryo-SEM was used to image the NanoDefine samples in the original formulated products to avoid any drying artefacts from solvent extraction methods. Imaging was conducted using a Philips XL-30 FEG ESEM coupled with an HKL Channel 5 electron backscatter diffraction detector and an Oxford Inca X-ray EDS detector. Samples of the complete and simple formula (BAM-13A and BAM-13B) were frozen under liquid nitrogen on the cold stage inside the instrument and were etched to reveal the particles inside the emulsion layers of the formulae.
Crystalline phases were identified for all dispersions using a Bruker D8 autosampler powder diffraction XRD system and the data were analysed using the Eva software. Scans were performed at 2θ between 20°and 60°. The presence of dopants and ratios between Ti and either Hf or Zr and Zr and Hf were verified and calculated through Rietveld refinement of the XRD diffractograms using the GSAS II software [31]. Owing to the low crystallinity of the diffractograms, high-quality phases were acquired from the FIZ Karlsruhe Inorganic Crystal Structure Database for the rutile [32] and anatase [33] TiO 2 and monoclinic ZrO 2 [34]. Those phases were used to calculate the unit cell dimensions, volume and density, and the atomic coordinates of the elements present inside the undoped reference phases. The latter were then used to verify the presence of dopants in the respective phases and calculate the unit cell parameters, atomic coordinates and the percentage of doping of the doped phases (Ti x Hf y O 2 , Ti x Zr y O 2 and Zr x Hf y O 2 ).
Specific surface area analysis using BET was conducted on all samples (except the complete and simple formulated NanoDefine samples due to lack of sufficient material) using a Beckman Coulter SA 3100 instrument. Briefly, the TiO 2 -containing powders were weighed within the ranges of 0.2-0.4 g and were outgassed for 280 min prior to analysis at a furnace temperature of 300°C. Next, using liquid nitrogen, the volume of gas adsorbed to the particle surface was measured to give the total surface area (m 2 g).
A Jenway 6800 double beam UV-visible spectroscopy (UV-vis) was used to measure the absorbance of the TiO 2 NM suspensions. The absorption peak was also used to calculate the energy band gap of the studied NMs using Tauc Plots [35]. This was achieved using the right shoulder of the absorption peak for which linear fitting was performed to calculate the cut-off wavelength (λ, the wavelength where the fitting crosses the horizontal wavelength axis) and the absorbance reaches a minimum. This energy band gap is then calculated using the equation E = h × C/λ, where h is Planck's constant (6.626 × 10 −34 J s), C is the speed of light (3.0 × 10 8 m s −1 ), λ is the cut-off wavelength in nm and 1 eV = 1.6 × 10 −19 J (conversion factor).
The concentration of the in-house synthesized TiO 2 series NMs was measured on a Perkin Elmer Optima 8000 ICP-OES using the Syngistix Software. Samples of 0.01 g were prepared using an aqua regia digest (2%) 24 h prior to analysis and calibrated with TiO 2 standards from 0 to 100 mg l −1 .  Initial TiO 2 synthesis was based on revising the approaches by Cassaignon et al. [36] and Valsami-Jones et al. [7], which involved a traditional hydrolysis and oxidation method. A titanium (III) chloride (TiCl 3 ) solution in 12% hydrochloric acid (HCl) was introduced to UPW under vigorous stirring, to produce a final solution strength of 0.15 M and the final solution pH was adjusted to pH 4. The dispersions were heated at 60°C on a hot plate until the violet solution became a white precipitate. Solids were washed and obtained by centrifuging the precipitate with UPW (to remove excess salts) at 5000 r.p.m. for 20 min. The washing procedure was conducted three times to ensure all excess salt solutions were removed, and the remaining solid was dried at 60°C for 6 h. The TiO 2 particles were then film-coated according to a modified protocol based on Wu et al. [37]. Briefly, a stock solution of 0.4 g of synthesized TiO 2 was suspended in 100 ml of UPW and sonicated for 2 h. Twenty-five millilitres of the stock solution was diluted 50 : 50 under vigorous stirring conditions. A 0.03 mol l −1 solution of aluminium chloride hydrate (AlCl 3 ) was added by titration to the TiO 2 suspension, and the pH was adjusted again to pH 4 using a 0.1 mol l −1 NaOH solution. After vigorous mixing, the dispersions were heated to 60°C for 2 h and centrifuged at 5000 r.p.m. for 30 min to obtain a solid white pellet. The solids were oven-dried at 60°C for 12-24 h [37].

Hafnium-and zirconium-doped TiO 2
It was not possible to synthesize good-quality doped TiO 2 NMs using the hydrolysis and oxidation method as described above. This method did not produce homogeneous size/structures or morphologies of the desired particles in question nor did it incorporate the dopant into the metal oxide core of the NMs. Alternative hydrothermal methods were therefore considered to complement the synthetic metal oxide NM library [38][39][40][41]. After trials, the synthesis of a range of doped NMs was achieved by hydrothermal synthesis using Ti (IV) isopropoxide, as the precursor, and a 0.01% molar ratio of Hf (IV) chloride (HfCl 4 ) or Zr (IV) oxynitrate hydrate (ZrO(NO 3 ) 2 · xH 2 O) as the doping agents in an alcohol aqueous dispersion. All dispersions were subject to hydrothermal treatment, which used Teflon-lined autoclave vessels (Parr Instruments) maintained at 200°C for 24 h. After hydrothermal treatment, the autoclave vessel was cooled at room temperature before the precipitate was removed, washed with UPW and centrifuged at 5000 r.p.m. for 30 min. The remaining precipitate was oven-dried at 60°C to obtain a white powder containing the Hf-or Zr-doped titania particulates. The drying procedure was repeated for all particles synthesized by hydrothermal treatment including NMs described in §2.4.3.-2.4.6.

Hafnium-doped ZrO 2 without surface treatment
In addition to the traditional TiO 2 synthesis, ZrO 2 NMs were also synthesized as pure ZrO 2 and with an Hf-doping agent to produce reference materials similar to the TiO 2 NMs. A starting solution of 0.5 M Zr (IV) oxynitrate hydrate (ZrO(NO 3 ) 2 · xH 2 O) and the doping agent of 0.01 M HfCl 4 were prepared in 10 ml UPW and stirred vigorously to mix. A 5 M solution of NaOH (10 ml) was added to the dispersion at a rate of 1 drop per second to produce a white precipitate. Once all the NaOH was added, the mixture

Results and discussion
3.1. Characterization of the complete formula sunscreen Table 1 displays the characteristics of the simple and complete sunscreen formulae obtained from the NanoDefine consortium (BAM-13A and BAM-13B). Three solvent extraction methods were performed in order to remove the TiO 2 NMs from the surrounding compounds found in the formulated samples [3]. The extracted titania particles were then characterized. The individual particle size distributions (TEM) from the various extraction conditions agreed well for each process for both samples BAM-13A and BAM-13B, producing size ranges within the standard deviation of the measurements. TEM imaging (figure 1a,c,e) identified sample BAM-13A to have mixed morphology containing both rods and semispherical-shaped NMs sized between 66 ± 23 nm on average. The two morphologies (rods and semispheres) may be attributable to at least two types of TiO 2 NMs (and possibly also the iron oxide used for pigmentation) present in the complete formula. Sample BAM-13B only contained rod-shaped NMs (figure 1b,d,f ) that were also individually sized between 66 ± 21 nm. The TEM imaging shows some clustering of the NMs, with definitive outlines of individual particles to confirm loose agglomerates. Despite some ice contamination from the sample preparation, cryo-SEM imaging (figure 2a,b) revealed nanostructured features inside the complete and simple formula sunscreen residues, similar to the nanostructures observed from the TEM imaging.

Characterization of commercially available powder TiO 2 NMs
The characterization data for the commercially available TiO 2 are given in    (table 1). The XRD patterns of the commercial powder preparations containing the TiO 2 NMs are shown in figure 6. The diffraction patterns at a normalized intensity all show strong peaks at 27°, 36°and 55°, agreeing with literature-cited sources for TiO 2 nanostructures in the rutile phase [42]. Although subtle, additional peaks identified at 39°, 44°and 57°are conformational of the Al oxide layer coating the particle surfaces (as highlighted in green, figure 6) [1] and agree with the EDS findings ( figure 5). Note that the iron oxide was not observed in any of the four commercial TiO 2 powder samples analysed, when compared with the samples observed in §3.2 for the complete formulated sunscreen sample BAM-13A. This is due to the iron oxide (non-nano) being incorporated as a colouring agent in the complete formula at a later production stage.
The UV-vis spectra of the commercially available NMs demonstrated the typical pattern produced for TiO 2 NMs for the JTTO MS 7 NM (figure 7a), the EBG of which is 2.89 eV. The rest of the available NMs (TA 100, TTO NJE8, UV Balance powder) presented a flat line, which is probably due to the low concentration of the available NMs that did not allow a proper signal to be acquired. It is interesting to note a small peak observed in all cases at approximately 530 nm, which can be attributed to the presence of impurities inside the structure of the materials. Similar peaks have been observed in the case of Ru    The commercial manufactured powder TiO 2 samples had additional surface treatments; this information was provided in the delivery notes of the samples and not obtained in house. For samples TTO NJE8 and UV Balance Powder 100 NJE8, the surface treatment was made up of jojoba esters, whereas JTTO MS7 was surface-treated with methicone, and TiO 2 TA-100 was surface-treated with silica (exact ratios were not provided by the manufacturers). The alumina coating is designed to prevent catalytic properties, such as the formation of oxidative species or free radicals on the surface of the TiO 2 , while the hydrophobic organic coating improves the dispersion and stability of the NMs in oil/water emulsion of the formula [1,37].  Overall, all commercial TiO 2 powder particles for use in sunscreen products investigated in this study contained NM properties as defined by the European Commission and the ISO, having size distributions that fall between 1 and 100 nm [45]. The powder TiO 2 NMs provided sufficient structure, morphology and size data to support the design of the TiO 2 and ZrO 2 reference NMs.

Characterization of synthesized TiO 2 and ZrO 2 doped for reference materials
Based on the characterization data of commercially available NMs and those in simple and complete sunscreen formulae, we now report the synthesis of a catalogue of crystalline anatase and rutile TiO 2 and ZrO 2 NMs in a range of sizes and surface coatings to mimic those commercially available. The characterization data are summarized in table 2.

Labelled TiO 2 reference materials
The hydrolysis and oxidation method of TiO 2 synthesis produced rod-shaped NMs, as shown by TEM analysis, regardless of the presence or absence of the alumina coating (table 2 and figure 8). The TiO 2 NMs coated with alumina show larger (40 ± 11 nm) and less aggregated particles than the TiO 2 without an alumina coating (figure 8). The size increase observed confirms the successful attachment of the surface coating. The alumina coating process exploits the dissociation of the AlCl 3 , which is introduced in deionized water containing the TiO 2 to form a hexaaquaaluminium complex [Al(H 2 O) 6 ] 3+ which transforms when the pH is adjusted with NaOH to produce Al 2 O 3 . The aluminium species is then adsorbed onto the TiO 2 surface to form a coating around the particle surface, resulting in increased size, as observed [37].
The morphological appearance of the alumina-coated TiO 2 particles matched well the NJE8 and JTTO MS7 commercial samples ( figure 4a,b), and the NMs observed in both samples BAM-13A and BAM-13B of the complete and simple sunscreen formulae (figure 1). EDS analysis in figure 9a,b provides information on the TiO 2 particles before and after the coating process. The presence of Al after the coating process is comparable to the EDS presented in figure 5 for the commercially available powder TiO 2 NMs, further confirming the presence of the alumina surface treatment on the in-house synthesized particles.     In likeness to the commercially available powder NMs, the XRD results (figure 10a) confirm that the TiO 2 particles with the alumina film coating have diffraction peaks at 27°, 36°and 55°, which agree with literature sources for TiO 2 nanostructure in the rutile phase [42]. Additional peaks at 39°, 44°and 55°-57°confirmed the presence of the alumina coating as observed in the XRD spectra for the commercially available powder TiO 2 particles (figure 4) (and confirms the EDS results; figure 9b). The Rietveld refinement of the TiO 2 aluminium oxide-coated NMs is presented in table 3. The unit cell dimensions and properties are in good agreement with those of the reference phase, and the reduction in volume and density can be explained by the lower crystallinity of the synthesized material. Additionally, no Al incorporation into the crystal structure of the TiO 2 was detected, which means that the Al is only present as a surface modification to the TiO 2 NM core.
When analysed by BET in the absence of the Al oxide layer, the surface area was 119.18 m 2 g −1 (table 2). In the presence of the alumina layer, the surface area was reduced slightly to 106.61 m 2 g −1 , which is consistent with increased particle size in the presence of the surface coating modification (table 2).  On the contrary, synthesized TiO 2 NMs with doping agents Hf and Zr produced by hydrothermal methods formed semi-spherical-shaped NMs (table 2 and figure 11a-c), which were similar to those observed for the UV Balance Powder and TiO 2 TA-100 commercial powder samples (figure 4c,d) and those reported in the literature [41]. The addition of acetate and stearate was also used as a surface treatment to produce a hydrophobic layer, similar to the commercially available NMs and complete formula NMs (BAM-13a) (table 1). The adsorption of stearate and acetate forms a strong bidentate bond between the TiO 2 surface and the carboxylic acid COOH group [41,46].
When TiO 2 is doped with other metals, its properties are modified, including the crystalline phase. Diffraction peaks at 25°, 37°and 48°( figure 10a,b) in the literature indicate that the synthesized TiO 2 NMs with doping agents Hf and Zr have an anatase crystalline phase [47] in comparison with the rutile rods produced previously by the hydrolysis and oxidation method. For the Hf-doped TiO 2 , the diffraction peak at 37°had drifted slightly to 38°, which is consistent with changes in the Ti crystalline lattice [48,49]. The ionic radii of Hf (IV) and Zr (IV) are 72 and 80 pm, respectively, and are both larger than that of Ti (IV), which is 56 pm. Therefore, substitutions of Ti positions in the crystal lattice by Hf and Zr ions will increase the size of the resulting NMs and will also cause slight drifts of the XRD diffraction peaks towards lower angles [48]. The XRD patterns also confirmed the addition of the doping agents using the cited literature and the diffraction patterns in the XRD Eva software database. The Hf-doped TiO 2 NMs had additional diffraction peaks at 22°, 23°, 29°, 32°, 34°, 44°, 47°-49°, 51°and 57°-59°(marked in black in figure 10a), which were absent when compared with the commercial TiO 2 samples and the complete formula TiO 2 samples (figures 3 and 6). Furthermore, longer scans between 20°and 80°2θ were required to obtain all corresponding peaks for the TiO 2 -Zr-doped particles at 30°, 53°, 55°, 62°, 70°a nd 75°(figure 10b) [48]. The additional diffraction peaks for the TiO 2 -Zr NMs were not observed in the Hf-doped TiO 2 or the commercial and simple/complete sunscreen formula TiO 2 preparations.
Rietveld refinement of the TiO 2 anatase NM, doped with either Hf or Zr, showed a good agreement and low error between the measured XRD diffractograms and the model produced through GSAS II, despite the low crystallinity of the NMs. The results suggest that both Hf and Zr were incorporated in the TiO 2 structure with respective changes in the unit cell properties and atomic coordinates (table 3), which can be explained by the difference in the ionic radii of those elements (Ti: 147 pm; Hf: 159 pm; and Zr: 160 pm). In the case of the Hf-doped TiO 2 , approximately 5% and 3% of Hf was incorporated inside the TiO 2 structure and their chemical formulae correspond to Ti 0.95 Hf 0.5 O 2 and Ti 0.97 Hf 0.03 O 2 for the acetate-and the stearate-coated NMs, respectively. The higher-and lower-doped TiO 2 led to a less and more crystalline material, respectively, when compared with the unlabelled NMs, as the changes in the α and c dimensions and unit cell volume (table 3) suggest, which can also be observed through the diffractograms, which present either broader or sharper peaks for the acetate-and stearic acid-coated NMs, respectively. Similarly, doping with 9% of Zr (Ti 0.91 Zr 0.09 O 2 ) and no surface treatment led to an NM, which is slightly less crystalline than the unlabelled NMs, which suggests that the crystallinity of the NM is affected by both the dopant and the potential surface treatment, which prevents NMs from growing [37].
When analysed by BET, the Hf-labelled TiO 2 NMs had a total surface area (  Figure 10c shows that the TiO 2 -Zr-doped NMs without surface treatment are larger at 42 (±10) × 12 (±3) nm. Therefore, the type of doping agent influences NM size, as has surface treatment. There were no morphological shape differences between the different surface-treated TiO 2 -Hf-doped NMs and the TiO 2 -Zr-doped NMs without surface treatment. For confirmation of the Hf and Zr doping, figure 12a reports the EDS spectra for TiO 2 NMs with the presence of Hf and figure 11b confirms the presence of the Zr dopant. XRD patterns (figure 10a,b) produced line broadening diffraction peaks, which are due to the nanosized nature of the samples and are consistent with other literature findings [41].
The UV-vis spectra of the doped TiO 2 NMs can be seen in figure 7b. The pure TiO 2 NM demonstrated an absorption peak at approximately 313 nm and had an EBG of 3.14 eV. The difference between this value and that calculated for the commercial JTTO MS 7 NM (2.89 eV) is due to the differences in the anatase and rutile crystal structures. These values are also similar to those reported in the literature of 3.3 and 3.0 eV for anatase and rutile structures [50], respectively, and the differences can be attributed to the different synthesis protocols and potential impurities present during production. Doping the TiO 2 NMs with Zr (TiO 2 Zr) resulted in a shift of the observed spectrum towards the lower wavelengths and the UV part of the spectrum, which signifies decreased photocatalytic activity. This is also supported by the calculated EBG of 3.61 eV. On the other hand, doping with Hf (TiO 2 Hf acetate-coated) resulted in a shift of the observed spectrum towards the visible part of the spectrum and EBG of 3.13 eV. To compare with the commercial NMs, the TiO 2 Hf (stearate-coated) was too dilute to measure. The shift towards the visible part of the wavelength spectrum and the reduction in EBG for the Hf-doped NMs also signify increased photocatalytic activity for the synthesized NMs, which can be explained by the presence of a new energy level in the band gap of the TiO 2 NM, due to the presence of other metal elements [20,51]. These new energy levels are more able to absorb photons and increase the number of excited electrons in the conduction and holes in the valence band [20,51]. The production of potentially damaging radicals can also take place, especially for pure anatase TiO 2 NMs due to their smaller size, when compared with rutile TiO 2 , and the overlapping of its conduction band with the redox potential of biological reactions that can lead to oxidative stress [52,53].
On the other hand, HfO 2 NMs are less likely to produce ROS, and thus oxidative stress, due to the lack of overlap with the redox potential [52], which is further supported by the lack of toxic observations during in vitro experiments [27,28]. As a result, the use of Hf as a doping agent can decrease the toxic potential of TiO 2 NMs, while enhancing its photocatalytic activities, thus making safer the use of such NMs in cosmetics. Doping with Zr, on the other hand, resulted in a shift towards the UV part of the spectrum and increases in the EBG, which signifies a decrease in the photocatalytic activity of TiO 2 , although such NMs could be useful in applications requiring the filtering of UV light, i.e. windows filters. At the same time, the conduction band of ZrO 2 NMs is closer to the redox potential range, which means that any reduction in the toxicity potential of TiO 2 NMs will be lower than that for Hf. This is further  supported by the moderate toxicity observed during in vitro experiments [26]. In any case, further study is needed, including toxicity experiments, to verify this safer-by-design approach that can also act as a tracer for the risk assessment of TiO 2 NMs.

Labelled ZrO 2 reference materials
Labelled ZrO 2 NMs were also synthesized in addition to the TiO 2 NMs (figure 13b). The average size of the ZrO 2 NMs containing Hf are 132 (±30) × 57 (±14) nm based on 100 particle counts (table 2). Morphological observations ( figure 13) show that the majority of the ZrO 2 NMs are much larger than those prepared with the Ti precursor and are morphologically different from each other. Supplementary to the TEM data, EDS measurements (figure 14a) show the ZrO 2 NMs synthesized are absent of the Hf dopant, and with the addition of the Hf dopant ( figure 14b). The measurements show that the presence of Zr accounts for approximately 98.87% for Zr, 0.16% for oxygen and 0.97% for Hf (atomic percentages). Furthermore, the XRD spectra in figure 15 show that the ZrO 2 NMs without a Hf dopant have a monoclinic crystalline structure, corresponding to a standard pattern (PDF: 00-037-1484) and previous literature [29,54]. Although the XRD spectra for the doped ZrO 2 NMs are very similar to those without Hf, subtle differences show that the ZrO 2 NMs containing the Hf dopant have additional diffraction peaks at 29°, 30°and 38°at 2θ. According to the standard data PDF 01-075-3557, the particles are composed of Zr-Hf oxide with a monoclinic crystalline structure, agreeing with the EDS results ( figure 14).
The presence of Hf inside the ZrO 2 structure was again verified through Rietveld refinement. In this case (table 3)  undoped material. This increase is opposite to that observed for the Zr-doped TiO 2 , which also had no surface treatment, and further supports the suggestion that both the ionic radius of the dopant and the surface treatment affect the degree of crystallinity of an NM. The difference in this case possibly lies in the similar ionic radii of Zr and Hf (160 and 159, respectively).

Conclusion
The aim of this study was to test a safer-by-design approach for the production of labelled TiO 2 NMs, which will also lead to increased NM traceability and assist exposure, fate and risk assessments. In recent years, research into the fate and behaviour of such TiO 2 NMs from sunscreen and industrial products has relied on either solvent extraction from commercially available products or nanopowders [30], such as those referenced in this paper. However, these products are often difficult to trace in environmental matrices, and thus, accurate results are difficult to obtain, and concentrations are hard to distinguish from naturally occurring titania and NM titania. We compared the size, morphology, surface area and crystalline structures of several TiO 2 -containing products from complete formulae to simple nanopowders for use in commercial products. We used these inputs to influence the successful synthesis of a library of particles, which can be used as reference materials comparable to the TiO 2 used in sunscreen and industrial formulae and can be distinguished from background concentrations of titania in environmental matrices, cosmetics and food. In summary, this paper reports a simple hydrolysis and oxidation method of TiO 2 synthesis, which successfully produced rod-shaped rutile phase particles, both with and without Al oxide coating and similar to nano-TiO 2 found in commercial sunscreen formulae. The novel hydrothermal method also successfully produced TiO 2 and ZrO 2 NMs, as well as Hf-and Zr-doped variants, each of which were shown to be structurally unique phases. Given that Hf and Zr are both relatively rare in the environment, the NMs presented in this paper can be uniquely identified in complex matrices and therefore serve as tracers in either laboratory or environmental studies.
Data accessibility. There are no additional data to accompany this manuscript. All relevant datasets are within the main body of the manuscript. All protocols and software used are stated fully in the methodology section. There are no coding lists for this research.