Characterization of photocatalytic paints: a relationship between the photocatalytic properties – release of nanoparticles and volatile organic compounds

D. Truffier-Boutry*a, B. Fiorentinoa, V. Bartolomeiae, R. Soulasa, O. Sicardya, A. Benayada, J.-F. Damlencourta, B. Pépin-Donatbcd, C. Lombardbcd, A. Gandolfoe, H. Worthame, G. Brochardf, A. Audemardf, L. Porcarg, G. Gebela and S. Gligorovski*eh
aUniversité Grenoble Alpes, CEA, Laboratoire en Nanosécurité et Nanocaractérisation, 17 Rue des Martyrs, F-38054 Grenoble Cedex 9, France. E-mail: delphine.boutry@cea.fr; Tel: +33 (0)4 38 78 08 81
bUniversité Grenoble Alpes, INAC-SPRAM, F-38000 Grenoble, France
cCNRS, INAC-SPRAM, F-38000 Grenoble, France
dCEA, INAC-SPRAM, F-38000 Grenoble, France
eAix Marseille Université, CNRS, LCE, UMR 7376, 13331, Marseille, France
fALLIOS, Les Docks Mogador, 105, Chemin de St Menet aux Accates, F-13011 Marseille, France
gInstitut Laue-Langevin, Paul Langevin, BP 156, F-38042 Grenoble Cedex 9, France
hState Key Laboratory of Organic Geochemistry, Guangzhou Institute of Geochemistry, Chinese Academy of Sciences, Guangzhou 510 640, China. E-mail: gligorovski@gig.ac.cn; Tel: +862085291497

Received 23rd May 2017 , Accepted 28th July 2017

First published on 2nd August 2017


Photocatalytic TiO2 appears to be a promising material to eliminate many air pollutants such as nitrogen oxides (NOx) and volatile organic compounds (VOCs). However, a number of questions remain unanswered prior to its full optimization. Some photocatalytic materials are already commercialized but their photocatalytic effects are questionable. In the present study, characterization of two paints for indoor and outdoor applications, one containing micro-sized titanium dioxide (TiO2) particles and the other based on nano-TiO2, is undertaken in order to understand their environmental impact during the use phase. The photocatalytic efficiency of the paints is determined before and after climatic ageing. The degradation of the paints induced by their ageing is characterized in parallel. Powders, dispersions and paints applied on a substrate are investigated to characterize the state of the nanoparticles (NPs) as a function of their surrounding media. The abrasion of the photocatalytic materials indicates that the presence of TiO2 (NPs) enhances the organic matrix degradation of the paints due to a greater photocatalytic effect. The online and continuous measurements by PTR-ToF-MS indicate that the degradation of the organic matrix leads to release of organic compounds (formaldehyde, methanol, acetaldehyde and formic acid) into the air which suggests that monitoring only the removal of VOCs (in this case xylene) is not enough to make a proper evaluation of the effectiveness of photocatalytic paints towards VOC elimination. These VOCs emerge exclusively from the degradation of the organic matrix as much lower VOC emissions were measured in the case of the aged paint which exhibits a lower amount of organic components in the matrix. This study links the morphological observations, chemical determination, structural parameters and photocatalytic properties of the paints for future optimization of safer-by-design photocatalytic paints.



Environmental significance

We investigate the morphological structure, chemical composition, structural parameters and properties of photocatalytic paints based on TiO2 nanoparticles, which is important for future optimization of safer-by-design photocatalytic paints. These paints are able to remove VOCs (e.g. xylene), but emission of harmful carbonyl compounds such as formaldehyde occurs, which is induced by degradation of the organic matrix. The abrasion tests undertaken before and after weathering aging of the photocatalytic paints indicate the emission of TiO2 nanoparticles into the air. The results obtained in this study challenge the usefulness of TiO2-based paints as a remediation technology to improve the air quality. We suggest a few actions based on a “safer-by-design” approach to optimize these paints prior to their launch onto the market.

Introduction

In the construction industry, nanotechnology creates the possibility to produce materials with novel functionalities and improved characteristics. Products containing nanomaterials (NMs) have been developed for the construction sector1 (cement, wet mortar and concrete, paints, coatings, insulation materials, glass, etc.) because of their enhanced properties such as higher durability, fire resistance, thermal stability, self-cleaning and photocatalytic properties. Engineered nanoparticles (ENPs), which by convention mean objects having at least one dimension within the nanometer range (1–100 nm), are more and more used in industrial applications. Different types of NPs are added to commercial paints to improve for instance their mechanical properties (SiO2),2 UV resistance,3 and rheological properties4 or to modify their coloring.5 The industry associated with production of NPs takes into consideration the responsible development of its products.6 This requires, in particular, control of the NPs throughout the life cycle of the product. In this context, many toxicity studies are devoted to properly assessing the impact of different types of NPs and their surrounding matrix on human health.7 Nevertheless, the best way to prevent inhalation and ingestion of NPs and also to avoid dermal contamination is to protect workers and consumers with appropriate equipment or to reduce the release of NPs into the environment. Therefore, some studies are more focused on the risk protection by studying collective8,9 and personal10–13 protection systems while other studies investigate all the ways leading to reduction of the release of nanoparticles while maintaining their initial properties (photocatalytic, antibacterial, reinforcement, etc.).14,15

A paint is generally composed of different components such as binders, fillers, pigments, solvents, and diluents and different additives like dispersing agents, thickeners and antifoaming agents. The polymeric binder induces the adhesion to the substrate and the chemical resistance of the paint, fillers reduce cost and enhance the physical properties of paints and pigments provide the coloring to the paint. Titanium dioxide (TiO2) particles with micro-size are usually used as the white pigment in paints. When TiO2 particles with nano-range size are embedded in the paint, the coloring property is lost but these NPs provide a photocatalytic effect, reducing air contaminants such as NOx and VOCs in the environment, to indoor and outdoor construction materials16–19 (concrete, asphalt, tiles, paints, etc.). Regarding the outdoor applications, TiO2 NPs were used in an attempt to reduce the adverse effects of urban air pollution on human health. Titanium dioxide (TiO2)-based photocatalytic surfaces have been tested, both on a laboratory scale and in the real-life urban environment for the remediation of nitrogen oxides (NOx), volatile organic compounds (VOCs), ozone (O3) and atmospheric particles. The results of these studies are controversial and the reasons for these contradictory results are still under discussion suggesting that further studies are necessary in order to accurately determine the impact of photocatalytic materials on air quality. Few studies have demonstrated that, instead of VOC elimination, the emission of carbonyl compounds occurs during the irradiation of photocatalytic paints. For example, Gunschera et al.20 identified emissions from building materials as formaldehyde, furfural, acetophenone, n-butylbutyrate, n-butyl-i-butyrate, n-butylpropionate, 4-heptanone, acetic acid, i-butyraldehyde and crotonaldehyde, in the case of photocatalytic tiles. The release of harmful aldehydes such as formaldehyde, acetaldehyde, 2-ethylacrylaldehyde, pentanaldehyde, and hexanaldehyde was observed upon irradiation of photocatalytic paints.21 Another controversial issue is the potential formation of by-products such as nitrous acid (HONO),22 which is even more harmful than the primary reactants NO and NO2. Regarding the indoor application of photocatalytic paints, Gandolfo et al.23 have shown that HONO is released into the indoor air as a function of the embedded quantity of TiO2 nanoparticles.

In parallel, other studies are more focused on the possible release of TiO2 NPs into the environment during the use phase of photocatalytic paints. Al-Kattan et al.24,25 studied the release of TiO2 NPs from paints before and after climatic ageing and also their behavior when exposed to different media (pH and different organic matter). Olabarrieta et al.26 immersed photocatalytic paints in different media during UV irradiation. Shandilya et al.27 investigated the emission of TiO2 NPs during ageing in water with the analysis of the run-off and in the air via abrasion experiments. All stated that the photocatalytic effect initiated by UV-light degrades the organic matrix of the paints and increases the release of NPs into the environment.

The present paper links both NP and VOC release into the environment during the ageing of a photocatalytic paint. The aim of the study consists of a better understanding of the main parameters controlling the release of NPs from nanomaterials in order to formulate, in the future, safer-by-design paints, i.e. less NP- and VOC-releasing paints with the same photocatalytic efficiency. For this purpose, two different paints were investigated. Both paints exhibit the same formulation, but they differ by the amount and size (micro and nano) of the TiO2 particles. The paint containing micro-TiO2 is already commercialized, and the one with nano-TiO2 is still under development. The use phase of these two products was investigated by simulating environmental and mechanical ageing.

TiO2 powders, dispersions and paints applied on a substrate were characterized to determine the chemical nature, the particle size distribution, the particle shape and the dispersion state. In a second phase, outdoor ageing of the products was simulated: painted panels were exposed to accelerated weathering. The photocatalytic properties of the TiO2 powders and applied paints were investigated by EPR (electron paramagnetic resonance) and the photocatalytic activity was measured with HR-PTR-ToF-MS (high resolution-proton transfer reaction-time of flight-mass spectrometry) before and after ageing to correlate with the surface characterization. All these techniques allowed the analysis of the surface and the bulk of each sample to be performed, which facilitated the understanding of the weathering effect on paints and of its impact on NP release and on the interaction between the polymeric binder and pigments. Abrasion tests were performed with a Taber Abraser to simulate mechanically induced ageing.

Materials and methods

Preparation of the paints

All the analyzed samples were provided by the same manufacturer ALLIOS. Two types of paints were investigated. Both paints were composed of the same constituents with the same proportion such as an acrylic-based binder, inactive microsized TiO2 for the white color, calcium carbonate particles, aluminosilicate, water and other additives, called “basic paint”. Their formulation differed only by the size of TiO2 particles used. For the formulation of the photocatalytic paint, a slurry was first made. The D2 slurry was composed of TiO2 NPs called P2 in anatase form and the D1 slurry was composed of inactive TiO2 microsized particles called P1. Both slurries D1 and D2 contained about 70% (w/w) P1 and P2 TiO2, respectively, in water. About 5% (w/w) D1 and D2 slurries were mixed with the “basic paint” to reach 2.5% (w/w) P1 and P2 in PM1 paint (for Matte paint 1) and PM2 paint (Matte paint 2). PM1 and PM2 were both investigated to compare a non-reactive paint with a reactive one. The manufacturer provided powders, dispersions, liquid paints and also paints applied on different substrates allowing us to perform various experiments and characterization studies.

TiO2 powders (P1 and P2) were characterized by X-ray diffraction (XRD, Bruker Advance diffractometer), scanning electron microscopy coupled with energy-dispersive X-ray spectroscopy (SEM-EDX, LEO Electron Microscopy Ltd, Quantas EDX, Bruker Nano), transmission electron microscopy coupled with energy-dispersive X-ray spectroscopy (TEM-EDS, FEI Tecnai Osiris) and X-ray photoelectron spectroscopy (XPS, PHI Versa Probe II spectrometer). As the P2 NPs were agglomerated in the powder state, they were characterized in solution by high resolution cryo-transmission electron microscopy (HR cryo-TEM, FEI Tecnai Osiris). Dispersion D1 composed of about 70 wt% P1 microparticles in water and dispersion D2 composed of about 70 wt% P2 nanoparticles in water were also investigated since the manufacturer uses these dispersions to formulate the paints. Small-angle X-ray scattering (SAXS) and small-angle neutron scattering (SANS) were used to determine the real size and the structure of the P2 NPs in the dispersed state. These techniques were chosen because they provide results which are more representative of the samples compared to TEM observations. Paints applied on Leneta plastic substrates (black plastic-vinyl chloride/acetate copolymer with a smooth matte surface, a thickness of 0.25 mm and a size of 165 × 432 mm) were characterized by SEM-EDS and XPS before and after accelerated weathering. The photocatalytic properties were measured by EPR (EMX Bruker spectrometer) on powders and applied paints under solar simulation (Newport Oriel, Sol 3A class AAA solar simulator 94023A). A proton transfer reaction-time of flight-mass spectrometer (PTR-ToF-MS) (Ionicon) was connected to a sealed reactor during the injection of a VOC onto painted surfaces stored over 21 days in an oven ventilated with clean air at 23 °C and 55% relative humidity prior to evaluation of their photocatalytic efficiency with respect to xylene. Paints applied on metallic substrates (Taber Industries aluminum panels of 100 mm2 with a 6.35 mm center hole) were used for abrasion tests (Taber Abraser type 5131, Taber Industries) before and after accelerated weathering in a QUV climatic chamber (Q-Lab, USA).

Photocatalytic efficiency

The photocatalytic effect was measured before and after accelerated weathering of both paints.

For these measurements, paint samples were placed in a flow tube reactor and exposed to VOCs.

For details about the working principle of the flow tube reactor, the readers are referred to Gandolfo et al.19,23 The flow tube reactor is coupled to a PTR-ToF-MS for on-line and continuous measurements of the variations of the concentrations of VOCs. This system is very sensitive for real-time monitoring of VOCs present in the air (acetone, acetaldehyde, methanol, ethanol, benzene, toluene, and xylene among others). Two types of experiments were conducted: (i) following the concentrations of all the released VOCs from the samples upon light irradiation and (ii) following the removal of one selected VOC (xylene) due to the photocatalytic efficiency of the paint. We have chosen to monitor the elimination of xylene by photocatalytic paints because it is considered toxic to humans,28 it is not emitted into the air by paints and it is present in non-negligible concentrations in indoor atmospheres29 because some materials generate this compound.30

Photocatalytic properties measured by EPR

EPR spectra were obtained at 9 GHz at room temperature. Samples were observed either wetted or with pure 5,5-dimethyl-1-pyrroline N-oxide (DMPO) used as a “spin trap”. The procedure consists of trapping radicals that are extremely difficult to be observed because of their short lifetime, using a diamagnetic molecule in order to obtain a stable paramagnetic species. The spin trapping technique involves the addition of an unstable radical to the double bond of a diamagnetic spin trap. This technique allows the identification of the nature of the unstable trapped radical. Such analyses were performed in the dark or under 15 min light irradiation with a solar simulator to characterize the photocatalytic properties of the paints. For irradiation tests, samples were placed in a fixed position located 5 cm below the device producing photons from the solar simulator. In this way, the center of the solar simulator coincides with the center of the sample. The EPR spectra were recorded under an Ar atmosphere after air exchange inside a glovebox in order to avoid broadening of the line widths due to the presence of oxygen.

Ageing: Q-UV exposure conditions

Paint-coated panels (Leneta and aluminum) were exposed to UV lamps in an accelerated weathering chamber for 500 h according to the ISO norm 16474-3:2013 (ISO 2013). The applied lamp is a fluorescent UVB 313 and the UV light was emitted in the wavelength range from 290 nm to 400 nm with a maximum applied irradiance of 0.71 W m−2 at 310 nm. Paints underwent two different cycles: a) 5 h with lamps switched on at a temperature of 50 ± 2 °C and b) 1 h with lamps switched off with a water spray at a temperature of 25 ± 2 °C. Paints exposed to UV lamps were denoted “QUV”. The total radiance exposure was 1.44 MJ m−2 giving a maximum acceleration using short wavelength UV.

X-Ray diffraction (XRD)

The crystallographic structure of the nanoparticles was investigated using a Bruker Advance diffractometer in geometry θ–2θ, equipped with a Cu-Kα source and a LynxEye linear detector. θ–2θ patterns were achieved in order to identify the various crystallographic phases and precisely measure their lattice parameters.

Small-angle scattering

Small-angle neutron scattering (SANS) experiments were performed on the D22 spectrometer of the Institut Laue-Langevin (ILL, Grenoble). Three different configurations (incident wavelength, λ and sample-to-detector distance) were used to cover a momentum transfer (q) range from 3 × 10−3 to 0.5 Å−1 (where q = 4π[thin space (1/6-em)]sin[thin space (1/6-em)]θ/λ and θ is the scattering angle). Small-angle X-ray scattering (SAXS) experiments were performed on the D2AM beamline of the European Synchrotron Radiation Facility (ESRF, Grenoble). Both dispersions D1 and D2 were studied in liquid cells with quartz and Kapton windows for SANS and SAXS, respectively. Usual corrections for intensity normalization and background subtraction were applied.31

Scanning electron microscopy coupled with energy-dispersive X-ray spectroscopy (SEM-EDS)

Images were acquired with an ultra-high resolution SEM (HR-SEM). Samples were covered with a 10 nm layer of platinum to prevent charging during observations. For the acquisition of images, the working distance was less than 4 mm with an accelerating voltage of 5 kV and a diaphragm of 30 μm. EDS analysis was performed to obtain chemical information on the observed samples. The quantification of EDS spectra was conducted using the “ESPRIT” software, allowing an accurate chemical elemental analysis. For this experiment, the working distance and the accelerating voltage were raised to 8 mm and 15 kV, respectively, with a diaphragm of 60 μm.

SEM-EDS allowed us to characterize powders and paint samples applied on Leneta substrates before and after ageing.

Transmission electron microscopy coupled with energy-dispersive X-ray spectroscopy (TEM-EDS)

TEM samples were prepared by simple drying of a drop of the sonicated solution on a carbon film grid. This analytical system has four EDX detectors giving a solid angle close to 1 steradian. Analysis was carried out in STEM mode using a HAADF (high angle annular dark field) detector at a high voltage of 200 kV with a probe current of 0.4 nA. The probe size was approximately 5 angstroms. Cryo-preparation was carried out using a Vitrobot plunger in liquid ethane. After sonication, a drop of the solution was deposited on a C-FLAT carbon grid and frozen at liquid ethane temperature. Observations were made using a single tilt Gatan cryo-holder working at liquid nitrogen temperature. Images were recorded using a low-dose mode at 200 kV (dose reduced to 10 e A−2 s−1) with a Gatan BM-Ultrascan camera equipped with a biological scintillator.

X-Ray photoelectron spectroscopy (XPS)

XPS allowed us to determine the surrounding environment of elements present at the near surface of the samples (first 10 nm). The XPS measurements were performed using a spectrometer with a monochromatized Al Kα X-ray source (1486.6 eV) focalized to a spot of 100 μm and with an electron take-off angle of θ = 45°. Survey spectra were recorded with a pass energy of 117 eV and the high-resolution spectra were registered with a pass energy of 23.5 eV. Known as a non-destructive surface analysis technique, XPS allowed us to probe the chemical bonds and their evolution after climatic ageing.

Abrasion process

All manipulations were performed in a so-called low-noise glove box equipped with an HEPA filter in order to limit the initial presence of particles in the atmosphere.32 The background noise in the sealed glove box was quantified using a condensation particle counter (CPC) to be less than 5 particles per cm3. This particular condition was achieved by creating a vacuum which sucks the air at a rate of 150 l min−1 at the top of the box. Clean air was obtained using a filter placed at the bottom of the glove box. For the abrasion of the surface paints, a Taber Abraser was used. This test simulated the ageing effect induced by mechanical friction. An S42 abrasive sticker paper was set on a CS-0 wheel to simulate the sanding process of a surface. A weight of 500 g was applied on both wheels of the Taber during 100 cycles of rotary abrasion at 60 rpm. These abrasion conditions conform to the standard ISO 7784-2 (ISO 2006).

The characterization of the particles released due to the abrasion process was performed by the determination of the particle size distribution with an electrical low-pressure impactor (ELPI) (DEKATI Ltd, Tampere, Finland). The ELPI allowed collection of aerosolized particles according to their aerodynamic diameter on hydrophilic polycarbonate membranes containing holes of 100 nm. Both membranes located in the two last impactor stages of the ELPI column, corresponding to the smallest particles' size impaction (of about 10 nm), were then analyzed by SEM-EDS. The goal was to observe the morphology and the size and also to determine the elemental composition of the collected nanoparticles.

Results

Characterization of pristine particles P1 and P2

Pristine particles were observed by microscopy (SEM and TEM), chemical identification was conducted by EDS and their crystalline structures were determined by XRD. As seen in Fig. 1, particles P1 had an elongated shape and their size was about 240 nm in length and 150 nm in width. P2 particles were aggregated, forming pellets with size above 200 nm as observed by SEM. TEM-EDS observations show that P1 particles were coated with a thin layer of carbon. To better observe P2 NPs, a HR cryo-TEM was used to observe a solution composed of P2 NPs and water in a frozen state. Isolated NPs with an average diameter of about 5 nm were observed.
image file: c7en00467b-f1.tif
Fig. 1 a) SEM micrograph, b) TEM micrograph, and c) XRD measurement of pristine P1 microparticles. d) SEM micrograph, e) HR cryo-TEM micrograph, and f) XRD measurement of pristine P2 nanoparticles.

EDS measurements identify predominantly titanium and oxygen for both powders. XRD measurements reveal that P1 is mainly composed of particles having a rutile crystalline structure and a minor anatase crystalline structure phase. On the other hand, P2 has only an anatase crystalline structure.

Characterization of P1 and P2 dispersions

The structures of P1 and P2 dispersions were studied by SANS and SAXS. The spectra are presented in Fig. 2. A scaling factor, the same for both P1 and P2, was applied to the SAXS data to superimpose with the SANS results. The scattered intensity varies over 8 orders of magnitude. The spectra obtained with the D1 and D2 dispersions are very different on the overall angular range. The D1 spectra vary as q−4 as expected for large sized particles according to Porod's law. The deviation observed at large angles is significant and cannot be eliminated, modifying the background subtraction. The D2 spectra appear strongly different with a q−3 upturn in intensity at low angles, characteristic of aggregated particles with strong dispersion in the aggregation number and an oscillation around 0.1 Å−1 that should originate from particles with a size around 5 nm. The SANS and SAXS data for both dispersions are exactly superimposable at low angles. The contrast for X-rays is directly related to the very large electron density difference of TiO2 compared to water while Ti presents a negative scattering length enhancing the contribution of a carbon shell. But interestingly, the spectra do not match anymore at large angles, revealing a non-homogeneous structure. Based on the characterization results presented above, the spectra were fitted with a core–shell model assuming spherical particles with a TiO2 core surrounded by a thin carbon shell for calculating the scattering length densities with a Gaussian polydispersity of 0.35. The same particle core diameter and shell thickness were used to adjust the SAXS and SANS spectra. For D1, a value of 150 nm was chosen for the core radius because this value cannot be adjusted due to the large size of the particles. The shell thickness is 0.6 nm. For D2, the core radius is 1.2 nm and the shell was found to be very thin (0.16 nm) which indicates that it cannot be undoubtedly identified by TEM or XPS characterization. The possible reason for this could be the carbon contamination of the TiO2 NPs. The signal scattered at low angles for D2 originates from aggregated particles with a fractal structure, the contribution of which is not encountered in the model. Comparing the level of intensities between the upturn in intensity and form factor of the nanoparticles, it can be estimated that less than 10% of the nanoparticles are aggregated in agreement with cryo-TEM observations in Fig. 1e.
image file: c7en00467b-f2.tif
Fig. 2 SANS and SAXS profiles of D1 and D2 dispersions as a function of the scattering-vector modulus q and their respective theoretical simulations.

Surface characterization of applied paints PM1 and PM2 before and after ageing

PM1 and PM2 were applied on Leneta substrates (polymer substrates). The SEM images of the paints' surface before and after QUV ageing are shown in Fig. 3 on different scales (×1000, ×10[thin space (1/6-em)]000, ×100[thin space (1/6-em)]000). On each scale, differences were observed between aged and non-aged paints. By focusing on the largest scale (×100[thin space (1/6-em)]000) before ageing, it could be seen that P1 and P2 particles were covered by a thin layer of organic matrix. After ageing, the organic layer disappeared, and P1 and P2 were found at the surface of the samples. For PM1, a dense layer of P1 was observed after ageing. In the case of the PM2 sample, not only P2 particles but also the plastic substrate was observed at the surface.
image file: c7en00467b-f3.tif
Fig. 3 SEM images of a) PM1, b) PM2, c) PM1 Q-UV, and d) PM2 Q-UV on a ×100[thin space (1/6-em)]000 scale.

To better understand the effects emerging from the ageing of both paints, XPS measurements were performed before and after ageing (Fig. 4). The XPS C 1s core peak registered at the surface of the PM1 and PM2 samples before ageing (in black in Fig. 4) was the signature of species related to the organic matrix, for instance C–C/C–H at 284.8 eV, C–O at 286.3 eV, COO at 289.0 eV and CO3 bonds related to CaCO3. The O 1s core peak corroborated this effect by the presence of a main peak at 532.4 eV related to the acrylic and vinyl copolymer. The Ti 2p core peak was not detected before ageing in the case of both paints, indicating that the organic layer (observed by SEM in Fig. 1) covering the surface of TiO2 particles was thicker than 5 nm.


image file: c7en00467b-f4.tif
Fig. 4 XPS spectra of PM1 and PM2 before and after ageing. a) C 1s, b) O 1s, and c) Ti 2p core level spectra of PM1 and d) C 1s, e) O 1s, and f) Ti 2p core level spectra of PM2.

After ageing (in red in Fig. 4), the intensity of the C–O and C–C peaks related to the C 1s peak decreased and simultaneously the O 1s spectra showed the appearance of a shoulder at 530 eV assigned to TiO2 particles. The valence state of TiO2 particles was probed by the mean of the Ti 2p peak indicating the presence of the +4 oxidation state of titanium (peak at 458.6 eV).

In the case of the PM2 paint, a similar behavior was observed qualitatively. However, some differences were observed quantitatively. Before ageing, the amount of C–O was lower. The ratio O[O2−]/O[organic] before and after UV exposure was evaluated for both samples, and reported to be 0 and 0.42 in the case of PM1 and 0 and 2.08 for PM2.

In the case of the PM2 paint, the higher intensity of Ti4+ and O2− peaks registered after ageing was related to the presence of mainly P2 TiO2 nanoparticles at the extreme surface of the sample. In parallel, polymer matrix degradation was observed with the decrease in intensity of the C–O-related peak of the C 1s spectrum.

Aerosolized particles' characterization

Particle size distributions analyzed using the ELPI counter during an ongoing abrasion process are depicted in Fig. 5. This type of counter measured the particle diameter as a function of the aerodynamic diameter (ranging from 7 nm to 10 μm). It was observed that all the non-aged paints released fewer particles than the aged paint containing nano-TiO2 (PM2).
image file: c7en00467b-f5.tif
Fig. 5 A) Concentration of particles released as a function of the particle size measured with ELPI; B) SEM-EDS characterization of the released particles.

The aged PM2 sample released mostly NPs with a diameter smaller than 100 nm. NPs impacted on the last stage of the ELPI column were observed by SEM-EDX and were found to be TiO2 NPs.

Testing of the photocatalytic performance

Signals provided by both powders with and without solar simulation have been recorded and are gathered in Fig. 6a. Since the considered photocatalytic paints in this study absorb in the visible range of wavelengths, even in the absence of solar light irradiation (only the presence of ambient light), both TiO2 powders give EPR signals. The light irradiance supplied by the lamps present in the room is sufficient to shift the electron of the valence band towards the conduction band forming an electric dipole (electron/electron hole). The electron (e) and the electronic hole (h+) react with gas, such as molecular oxygen (O2) and water vapor (H2O) adsorbed on the semiconductor surface (TiO2). The formation of these hole–electron pairs creates highly reactive species such as hydroxyl radicals (˙OH), superoxide radicals (˙O2) and hydrogen peroxide radicals (˙OOH) that oxidize organic compounds which are adsorbed on the surface of the catalyst.33 The EPR signal is composed of three lines with respective g-factors of 2.023, 2.009 and 2.004 which can be interpreted in two ways: either to the Ti4+O2−Ti4+OH˙ species34–38 or to oxygenated radicals39 due to the photoreaction of residual organic groups as reported for cysteine-modified TiO2 colloids37,40 ascribed to TiO2 reactivity. The broad signal observed at a higher g value (between 2.08 and 2.06) may be likely ascribed to ˙O2 adsorbed on the nanoparticle surface.41 At a lower g value (g = 1.96), we did not observe the signal corresponding to electrons,42 which could possibly be attributed to the difference between the temperatures of observation, 10 K in our case against 4 K in the study by Kumar et al.42 The signals in the spectra of both nano- and micro-TiO2 particles look very similar, except for the first line (higher g value) for micro-TiO2 which appears at a slightly higher g-factor compared to nano-TiO2 (2.035 against 2.026).42 Under solar simulation, the signals are more intense for both powders. However, the nano-TiO2 powder was the most reactive (in dashed grey line) and showed higher intensity signals.
image file: c7en00467b-f6.tif
Fig. 6 EPR signals of a) micro-TiO2 P1 and nano-TiO2 P2 powders with and without solar irradiation, b) PM2 paint applied on Leneta substrates with DMPO as a function of the duration of the solar irradiation and c) PM1 and PM2 applied on Leneta substrates both with DMPO after 30 min of solar irradiation.

EPR experiments were also conducted on paints applied on Leneta substrates in the presence and absence of solar irradiation (Fig. 6b and c). In order to observe the various radical species involved during the TiO2 catalytic photoreaction, we have applied the so-called “spin trapping method”. The unstable free radical reacts with DMPO and forms a relatively stable nitroxyl radical exhibiting very distinguishable EPR spectra. When the paints which were in contact with pure DMPO were illuminated, new lines appeared which varied with the irradiation time (Fig. 6b). The separation in Gauss between the two extreme lines which appears (*) upon illumination is on the order of 42 Gauss. This fits well with hyperfine coupling expected in the case of trapped ˙OH radicals. Even embedded in a paint formulation, TiO2 reacts and generates ˙OH radicals at the sample surface or within the paint coat for both paints PM1 and PM2 as shown in Fig. 6c. The formation of ˙OH radicals upon illumination of TiO2 and polyacrylonitrile is a well-known mechanism and has already been observed.43,44

The photocatalytic performance of the paints was investigated as well. In this sense, two types of experiments were performed: i) on-line analysis of VOCs released by the paints upon light irradiation and ii) on-line monitoring of the removal of xylene induced by the photocatalytic effect of the paints. All the VOCs released from pristine and aged PM1 and PM2 paints are observed in Fig. 7. Four types of VOCs appeared in the reactor under illumination of the paints: formaldehyde (m/z 31), methanol (m/z 33), acetaldehyde (m/z 45) and formic acid (m/z 47). At first glance, it can be seen that the PM2 paint released a bigger amount of VOCs compared to PM1, especially for non-aged samples. Thus, although these photocatalytic materials are aimed at eliminating a broad range of volatile organic compounds, they release harmful pollutants, such as formaldehyde, for instance. These observations are in agreement with the literature data. For example, the release of formaldehyde was also observed upon light irradiation of photocatalytic paints by Salthammer and Fuhrmann.21


image file: c7en00467b-f7.tif
Fig. 7 Surface emission rates of released VOCs for both PM1 and PM2 paints, pristine and aged ones.

Concerning the removal of xylene induced by the photocatalytic activity of both paints, the first-order rates of xylene disappearance were measured. The PM1 paint was not able to eliminate xylene; hence, it was impossible to measure the corresponding rate constant. On the other hand, a first-order rate constant of (9.5 ± 4.5) × 10−4 s−1 was measured for pristine PM2 and (5.7 ± 0.9) × 10−2 s−1 for aged PM2. Hence, the aged paint containing nano-TiO2 was much more effective (ca. two orders of magnitude) in terms of photocatalytic removal efficiency towards xylene.

Discussion

Nano-TiO2 is known for its photocatalytic properties. Manufacturers add it to their formulation for its VOC removal action. In this study, we reveal that the paint containing TiO2 particles with sizes up to 100 nm and a rutile crystalline structure (PM1) cannot remove xylene from the air, independently of its aging state, as the decay of xylene could not be measured (see Fig. S2 in the ESI). On the other hand, the paint composed of TiO2 NPs with an anatase crystalline structure enables the degradation of xylene as the first-order decay rates of xylene were (9.5 ± 4.5) × 10−4 s−1 and (5.7 ± 0.9) × 10−2 s−1 for non-aged and aged PM2 paints, respectively. While the photocatalytic properties of the paints improve the elimination of VOCs such as xylene, they lead also to the generation of four additional VOCs that are released into the environment at the same time. Among the four identified compounds, formaldehyde, which is a sensory irritant, has been classified as a carcinogenic substance.45 The benefit of the photocatalytic properties to clean up the environment is counterbalanced by the emission of harmful VOCs and also by the possible release of NPs. Indeed, a high amount of NPs, essentially composed of titanium, was observed in the EDX spectrum (Fig. 5B), released by the aged PM2. Despite the fact that the same free radicals are generated during EPR experiments on the P1 and P2 powders, the three lines with respective g-factors of 2.023, 2.009 and 2.004 related to TiO2 reactivity indicate that the P2 nanoparticles with an anatase crystalline structure are more efficient than the rutile microparticles. Although the EPR peak intensities of P1 and P2 are similar when non-irradiated, the P2 EPR spectrum showed more intense peaks compared to that of P1 under UV irradiation. P1 does not act as a VOC removal agent contrary to P2 which exhibits a high reactivity under UV irradiation, in agreement with the literature data.46,47 A possible explanation could be that the anatase crystalline structure exhibits lower recombination rates combined with higher surface adsorptive capacity with organic compounds,48 leading to production of more free radicals acting on VOCs. When P1 and P2 are added to a paint (so-called PM1 and PM2 formulations), the EPR spectra show more peaks attributed to trapped ˙OH radicals (Fig. 6c) generated at the paint surface or within the paint,44 which are responsible for the degradation of VOCs. As observed for irradiated P1 and P2, PM2 shows more intense EPR peaks and the amount of generated ˙OH radicals increases with the irradiation time (see Fig. 6b), confirming that nano-TiO2-based paint PM2 is indeed more reactive than PM1. Concerning the crystalline structures, it is well known that these two principal structures exhibit different photocatalytic performances49 and that a mixture of anatase and rutile showed a more pronounced catalytic effect,50 as observed for PM2 containing 2.5 wt% rutile TiO2 microparticles and 2.5 wt% anatase TiO2 NPs. Obviously, the increase in the amount of produced ˙OH radicals at the paint surface with irradiation time corroborates the increase of the first-order decay rate of xylene after ageing of PM2. In addition to the difference in crystal structure, a carbonaceous coating was well observed by TEM (Fig. 1b) and SAXS (Fig. 2) in the case of the P1 particles which could also reduce the photocatalytic properties. A very thin shell of about 0.16 nm was also found at the surface of the P2 NPs in the D2 dispersion (Fig. 2). This shell might be surface contamination of the NPs rather than a carbon coating as observed for P1 but its presence could contribute to the dispersion of the NPs in an organic matrix, as for the P1 particles which are intentionally coated with a thin carbon layer in order to improve the compatibility between the inorganic particles and the organic matrix of the paint. Indeed, if P2 NPs are agglomerated/aggregated in the solid state (Fig. 1d), they look really well dispersed in liquid media, when observed by cryo-TEM (Fig. 1e). Furthermore, only 10% aggregation was measured by SAXS/SANS when studying the D2 dispersion. As the manufacturer uses this dispersion to formulate the PM2 paint, it is expected that P2 NPs would be well dispersed in the paint. However, the SEM observations of PM2 following the weathering step (Fig. 3d) indicate that clusters of NPs are visible but also the plastic substrate, which suggests that the dispersion state of the NPs was not as good as expected and also that significant degradation of the organic matrix occurred. The SEM observation and XPS analysis of PM1 and PM2 performed prior to and after the weathering revealed the deterioration of the paint surface following the ageing of the paints. Prior to the ageing, the SEM analysis showed a thin layer of organic matrix covering the TiO2 NPs on PM1 and PM2 (Fig. 3a and b). After climatic ageing of the paints, the SEM pictures showed only microparticles and NPs of TiO2 at the surface, indicating the disappearance of the organic matrix for PM1 Q-UV and PM2 Q-UV. The XPS measurements revealed the total disappearance of the peaks related to the organic matrix (C–C and C–H in Fig. 4d) for the PM2 paint after ageing compared to the C 1s spectra of PM1 (Fig. 4a). These results highlight the total degradation of the PM2 organic matrix compared to PM1. The surface degradation of the PM1 paint is probably only due to UV irradiation, whereas for PM2, the degradation can be entirely ascribed to the photocatalytic effect induced by the TiO2 NPs. At the same time, the TiO2 signature significantly increases as observed in the Ti 2p and O 2s XPS spectra in Fig. 4e and f, in agreement with the SEM observations. A significant difference is observed for the O 1s spectra (Fig. 4b and e) since there is a change in the relative intensity of the C–O* and O–C[double bond, length as m-dash]O* peaks related to the organic matrix and the O2− peak related to TiO2 NPs. The disappearance of the organic matrix in the case of PM2 seems to be directly linked to the emission of high levels of VOCs. The emission of four VOCs is reduced after ageing as the organic matrix completely disappears. These results also indicate that more TiO2 NPs are present at the near surface of the aged PM2 samples compared to PM1. This accumulation of TiO2 NPs at the surface allows better removal of xylene and the absence of the organic matrix leads to a high amount of TiO2 NPs released into the environment. After ageing, TiO2 NPs can be released into the environment which could affect the workers during a sanding step or when people rub against a painted wall.

The whole process is illustrated in Scheme 1: a) paint containing TiO2 NPs is applied on a substrate; VOCs are present in the atmosphere, b) UV irradiation activates the photocatalytic effect of TiO2 NPs; ambient VOCs are degraded as well as the organic matrix of the paint leading to the emission of new VOCs into the air, and c) the degradation of the paint leads to accumulation of TiO2 NPs at the surface of the painted walls and a slight friction causes the release of NPs into the air.


image file: c7en00467b-s1.tif
Scheme 1 Simplified illustration of the process leading to emission of new VOCs and release of NPs into the air from the photocatalytic paints.

These results show that the emitted VOCs have a direct link with the organic matrix of the paint. When the content of organic components in the matrix is higher, more VOCs are created during the irradiation of the paints. In the case of non-reactive paint PM1, a small quantity of VOCs was emitted compared to the PM2 one. P1 micron sized particles showed low photocatalytic efficiency while degrading the organic matrix. As the degradation of the matrix was weak, the same amount of emitted VOCs was observed after aging of the PM1 paint. In contrast, the PM2 paint, which contains anatase TiO2 NPs, induces greater degradation of the organic matrix due to the higher photocatalytic efficiency which in turn leads to greater emission of VOCs prior to aging, i.e. before the disappearance of the organic matrix. After aging, a very small amount of organic matrix was present in the PM2 paint and the concentration of emitted VOCs decreased.

Conclusion

Few studies20,21,51 have shown that irradiation of the photocatalytic paints leads to generation of VOCs in the air instead of their removal. The origin of these compounds was mainly attributed to the degradation of the organic binder in agreement with our findings. These few studies speculated that the cause of release of carbonyl compounds could be the photochemical decomposition of paint binders, without going into deeper systematic investigations.

All these previous studies are based on off-line analytical techniques coupled to the sampling methodology. Thus, the choice of the sampling support and the associated analytical techniques play a crucial role during the identification of the VOCs, beyond the sampling artefacts which can be generated by these methods. This indicates that the selected coupling system (sampling method/analytical technique) will determine the class of detected and identified VOCs. Proton transfer reaction-mass spectrometry (PTR-MS) offers a possibility to observe a large selection of VOCs continuously and on-line with considerably reduced temporal evolution compared to the off-line techniques.

In this study, for the first time to the best of our knowledge, by using the state-of-the-art PTR-ToF-MS instrument, we show that the tested photocatalytic paints released organic compounds such as formaldehyde, methanol, acetaldehyde and formic acid upon irradiation.

These results indicate that monitoring only the decaying concentrations of a target compound (in this study xylene) is not sufficient to make an assessment of the effectiveness of photocatalytic paints towards VOC elimination in various indoor environments. Such an approach does not take into account the fact that stable reaction products, mainly the harmful carbonyls, may occur which can profoundly impact the indoor air quality. These VOCs emerge exclusively from the organic matrix decomposition since much lower emissions were measured in the case of the aged paint which exhibits a lower amount of organic components in the matrix. In addition, the SEM pictures and XPS analysis indicate that decomposition of the organic matrix induced the accumulation of TiO2 NPs at the surface of the paint followed by their release into the air. There is no direct relationship between the emission of VOCs and the release of NPs although both phenomena are caused by the photocatalytic degradation of the organic matrix.

Thanks to the large suite of experiments reported in this paper and the results obtained, different solutions were selected to decrease the release of particles and VOCs into the air and therefore to formulate a “safer-by-design” photocatalytic paint. A first approach is to control the degree of the photocatalytic efficiency by modifying the TiO2 NPs with different coatings and also to control the dispersion state of the NPs in the organic matrix to decrease its degradation while maintaining an optimum photocatalytic efficiency of the paints. The selected coatings should have a good affinity with the organic matrix to have good dispersion of the NPs in the paint. Another choice is localization (grafting) of TiO2 NPs on bigger particles to prevent NP release and to improve the affinity between NPs and the organic matrix to reduce the release of NPs. Finally, more resistant binders should be tested to reinforce the organic matrix. All of the suggested solutions are currently under investigation in our laboratory in cooperation with the manufacturer of these paints.

Acknowledgements

This work was carried out in the framework of the LABEX SERENADE (ANR-11-LABX-0064) funded by the French Government program, Investissements d'Avenir, and managed by the French National Agency (ANR). HR cryo-TEM was funded by the EQUIPEX project ANR-10-EQPX-39-01 called NanoID. We are equally grateful to Frédéric Amblard, Cécile Ducros and Manuel Maréchal for their contribution.

References

  1. M. J. Hanus and A. T. Harris, Prog. Mater. Sci., 2013, 58, 1056–1102 CrossRef CAS.
  2. M. Malaki, Y. Hashemzadeh and M. Karevan, Prog. Org. Coat., 2016, 101, 477–485 CrossRef CAS.
  3. H. Ye, L. Zhu, W. Li, G. Jiang, H. Liu and H. Chen, Chem. Eng. J., 2017 DOI:10.1016/j.cej.2017.03.088.
  4. E. Ghafari, S. A. Ghahari, Y. Feng, F. Severgnini and N. Lu, Composites, Part B, 2016, 105, 160–166 CrossRef CAS.
  5. A. Zhang, B. Mu, Z. Luo and A. Wang, Dyes Pigm., 2017, 139, 473–481 CrossRef CAS.
  6. P. van Broekhuizen, F. van Broekhuizen, R. Cornelissen and L. Reijnders, J. Nanopart. Res., 2011 DOI:10.1007/s11051-010-0195-9.
  7. J. Lee, S. Mahendra and P. J. J. Alvarez, ACS Nano, 2010, 4(7), 3580–3590 CrossRef CAS PubMed.
  8. S. J. Tsai, R. F. Huang and M. J. Ellenbecker, Ann. Occup. Hyg., 2010, 54(1), 78–87 CAS.
  9. V. Cesard, E. Belut, C. Prevost, A. Taniere and N. Rimbert, Ann. Occup. Hyg., 2013, 57(3), 345–359 CAS.
  10. C. Brochot, N. Michielsen, S. Chazelet and D. Thomas, Ann. Occup. Hyg., 2012, 56(5), 595–605 CAS.
  11. S. A. Grinshpun, H. Haruta, R. M. Eninger, T. Reponen, R. T. McKay and S. A. Lee, J. Occup. Environ. Hyg., 2009, 6(10), 593–603 CrossRef PubMed.
  12. S. Rengasamy and B. Eimer, Ann. Occup. Hyg., 2011, 55(3), 253–263 CAS.
  13. L. Vinches, N. Testori, P. Dolez, G. Perron, K. J. Wilkinson and S. Hallé, Nanosci. Methods, 2013, 2(1), 1–15 CrossRef CAS.
  14. Y. Lynch, C. Weiss and E. Valsami-Jones, Nano Today, 2014, 9, 266–270 CrossRef.
  15. G. Morose, J. Cleaner Prod., 2010, 18, 285–289 CrossRef CAS.
  16. E. Boonen and A. Beeldens, Coatings, 2014, 4, 553–573 CrossRef CAS.
  17. C. Toro, B. T. Jobson, L. Haselbach, S. Shen and S. H. Chung, Atmos. Environ., 2016, 139, 37–45 CrossRef CAS.
  18. A. L. da Silva, D. N. F. Muche, S. Dey, D. Hotza and R. H. R. Castro, Ceram. Int., 2016, 42, 5113–5122 CrossRef CAS.
  19. A. Gandolfo, L. Rouyer, H. Wortham and S. Gligorovski, Appl. Catal., B, 2017, 209, 429–436 CrossRef CAS.
  20. J. Gunschera, J. R. Andersen, N. Schulz and T. Salthammer, Chemosphere, 2009, 75, 476–482 CrossRef CAS PubMed.
  21. T. Salthammer and F. Fuhrmann, Environ. Sci. Technol., 2007, 41, 6573–6578 CrossRef CAS PubMed.
  22. M. E. Monge, B. D'Anna and C. George, Phys. Chem. Chem. Phys., 2010, 12, 8991–8998 RSC.
  23. A. Gandolfo, V. Bartolomei, E. Gomez Alvarez, S. Tlili, S. Gligorowski, J. Kleffmann and H. Wortham, Appl. Catal., B, 2015, 166-167, 84–90 CrossRef CAS.
  24. A. Al-Kattan, A. Wichser, S. Zuin, Y. Arroyo, L. Golanski, A. Ulrich and B. Nowack, Environ. Sci. Technol., 2014, 48, 6710–6718 CrossRef CAS PubMed.
  25. A. Al-Kattan, A. Wichser, R. Vonbank, S. Brunner, A. Ulrich, S. Zuin and B. Nowack, Environ. Sci.: Processes Impacts, 2013, 15, 2186–2193 CAS.
  26. J. Olabarrieta, S. Zorita, N. Pena, N. Rioja, O. Monzon, P. Benguria and L. Scifo, Appl. Catal., B, 2012, 123-124, 182–192 CrossRef CAS.
  27. N. Shandilya, O. Le Bihan, C. Bressot and M. Morgeneyer, Environ. Sci. Technol., 2015, 49, 2163–2170 CrossRef CAS PubMed.
  28. O. Geiss, C. Cacho, J. Barrero-Moreno and D. Kotzias, Build. Environ., 2012, 48, 107–112 CrossRef.
  29. D. Kotzias, O. Geiss, S. Tirendi, J. Barrero-Moreno, V. Reina, A. Gotti, G. Cimino-Reale, B. Casati, E. Marafante and D. Sarigiannis, Fresenius Environ. Bull., 2009, 18, 670–681 CAS.
  30. A. M. Dafni, E. Demetriou, N. Michael, E. I. Tolis and J. G. Bartzis, Atmos. Environ., 2010, 44, 4388–4395 CrossRef.
  31. Neutron, X-rays and Light. Scattering Methods Applied to Soft Condensed Matter, ed. Th. Zemb and P. Lindner, 1st edn, North Holland, 2002 Search PubMed.
  32. B. Fiorentino, L. Golanski, A. Guiot, J.-F. Damlencourt and D. Boutry, J. Nanopart. Res., 2015, 17, 149 CrossRef.
  33. A. Fujishima, T. N. Rao and D. A. Tryk, J. Photochem. Photobiol., C, 2000, 1, 1–21 CrossRef CAS.
  34. R. F. Howe and M. Grätzel, J. Phys. Chem., 1987, 91, 3906–3909 CrossRef CAS.
  35. M. Anpo, M. Che, B. Fubini, E. Garrone, E. Giamello and M. C. Paganini, Top. Catal., 1999, 8, 189–198 CrossRef CAS.
  36. C. Naccache, P. Meriaudeau, M. Che and A. J. Tench, J. Chem. Soc., Faraday Trans., 1971, 67, 506–512 RSC.
  37. J. M. Coronado, A. J. Maira, J. C. Conesa, K. L. Yeung, V. Augugliaro and J. Soria, Langmuir, 2001, 17(17), 5368–5374 CrossRef CAS.
  38. S.-C. Ke, T.-C. Wang, M.-S. Wong and N. O. Gopal, J. Phys. Chem. B, 2006, 110, 11628–11634 CrossRef CAS PubMed.
  39. M. D. Sevilla, D. Becker and M. Yan, J. Chem. Soc., Faraday Trans., 1990, 86(19), 3279–3286 RSC.
  40. T. Rajh, A. E. Ostafin, O. I. Micic, D. M. Tiede and M. C. Thurnauer, J. Phys. Chem., 1996, 100, 4538–4545 CrossRef CAS.
  41. Y. Nakaoka and Y. Nosaka, J. Photochem. Photobiol., A, 1997, 110, 299–305 CrossRef CAS.
  42. C. P. Kumar, N. O. Gopal, M.-S. Wang and K. S. Ke, J. Phys. Chem. B, 2006, 110, 5223–5229 CrossRef CAS PubMed.
  43. M. V. Motyakin and S. Schlick, Macromolecules, 2001, 34, 2854–2864 CrossRef CAS.
  44. R. Morelli, I. R. Bellobono, C. M. Chiodaroli and S. Alborghetti, J. Photochem. Photobiol., A, 1998, 112, 271–276 CrossRef CAS.
  45. T. Salthammer, Int. J. Hyg. Environ. Health, 2015, 218, 433–436 CrossRef CAS PubMed.
  46. Y. Gao, H. Wang, J. Wu, R. Zhao, Y. Lu and B. Xin, Appl. Surf. Sci., 2014, 294, 36–41 CrossRef CAS.
  47. N. Yuangpho, S. T. T. Le, T. Treerujiraphapong, W. Khanitchaidecha and A. Nakaruk, Phys. E, 2015, 67, 18–22 CrossRef CAS.
  48. D. C. Hurum, A. G. Agrios, K. A. Gray, T. Rajh and M. C. Thurnauer, J. Phys. Chem. B, 2003, 107, 4545–4549 CrossRef CAS.
  49. S. Wang, J. S. Lian, W. T. Zheng and Q. Jiang, Appl. Surf. Sci., 2012, 263, 260–265 CrossRef CAS.
  50. L. Zao, M. D. Han and J. S. Lian, Thin Solid Films, 2008, 516, 3394–3398 CrossRef.
  51. J. Auvinen and L. Wirtanen, Atmos. Environ., 2008, 42(18), 4101–4112 CrossRef CAS.

Footnotes

The authors declare no competing financial interest.
Electronic supplementary information (ESI) available. See DOI: 10.1039/c7en00467b

This journal is © The Royal Society of Chemistry 2017