Benzene and NO x photocatalytic assisted removal using indoor lighting conditions

Modern life-style is creating an indoor generation: human beings spend approximately 90% of their time indoors, almost 70% of which is at home – this trend is now exacerbated by the lockdowns/restrictions imposed due to the COVID-19 pandemic. That large amount of time spent indoors may have negative consequences on health and well-being. Indeed, poor indoor air quality is linked to a condition known as sick building syndrome. Therefore, breathing the freshest air possible it is of outmost importance. Still, due to reduced ventilation rates, indoor air quality can be considerably worse than outdoor. HVAC, air filtration systems and a well-ventilated space are a partial answer. However, these approaches involve only a physical removal. Photocatalytic mineralisation of pollutants into non-hazardous, or at least less dangerous compounds, is a more viable solution for their removal. Titanium dioxide, the archetype photocatalytic material, needs UVA light to be “activated”. However, modern household light emitting diode lamps irradiate only in the visible region of the solar spectrum. In this short-communication, we show that the surface of titanium dioxide nanoparticles modified with copper oxide(s) and graphene shows promise as a viable way to remove gaseous pollutants (benzene and NOx) by using a common light emitting diode bulb, mimicking real indoor lighting conditions. Titanium dioxide, modified with 1 mol% CuxO and 1 wt% graphene, proved to have a stable photocatalytic degradation rate, three times higher than that of unmodified titania. Materials produced in this research work are thus strong candidates for offering a safer indoor environment.

In brief: sols were synthesised via the controlled hydrolysis and peptisation of titanium (IV) isopropoxide [Tii-pr, Ti(OCH(CH3)2)4, Sigma-Aldrich, ≥ 97.0%] with Milli-Q water (resistivity = 18.2 MΩ.cm; H2O:Ti-i-pr = 4:1) diluted in isopropyl alcohol (IPA). Concentrated HNO3 (Merck, 69%) was used to peptise the sol -Ti 4+ :HNO3 = 2.5:1. Copper (II) nitrate trihydrate (Sigma-Aldrich, ≥ 98.5%) was used to make copper-modified TiO2 sols [1]. To this aim, stoichiometric amounts of the copper salt, having a copper molar amount equal to 1 mol% [2], were added to the TiO2 sol when this had a 1 M concentration. Graphene modifications were made in a similar fashion: 0.5 and 1 wt% of graphene nanoplatelets dispersed in ethanol (supplied by Graphenest) were added to the sols when those had a 1 M concentration. The sols were then dried in an oven at 80 °C for 24 h to remove water. Subsequently, the dried gels were thermally treated following thermal cycles reaching two maximum temperatures: (a) 250 °C with a heating rate of 15 °C/min, before-heating the furnace at 200 °C, with an 8 h dwell time; (b) 450 °C, with a heating rate of 5 °C/min, with 2 h dwell time. Specimens were referred to as: Cu-Gx/Y were x stands for graphene wt% (0.5 or 1.0 wt%), and Y is a number, designating the maximum temperature reached (i.e. 250 or 450). For example, the TiO2 sol in which 1 mol% Cu and 1.0 wt% graphene were added, and thermally treated at 450 °C/2 h, will be labelled as: Cu-G1.0/450.

S1.2 Sample characterisation
The mineralogical composition of the prepared specimens was resolved by means of X-ray powder diffraction (XRPD). The relative fractions of crystalline phases were determined by means of Rietveld refinements of the XRPD data. XRPD patterns were collected at room temperature on a θ/θ diffractometer (PANalytical X'Pert Pro, NL), equipped with a fast RTMS detector (PIXcel 1D, PANalytical), with Cu Kα radiation (45 kV and 40 mA, 20−80 °2θ range, with a virtual step scan of 0.02 °2θ, and virtual time per step of 200 s). The Rietveld data analysis for obtaining semi-quantitative phase analysis (QPA) information was assessed using the GSAS software package and its graphical interface EXPGUI [3,4]. The instrumental broadening was obtained from the refinement of LaB6 standard (NIST SRM 660b) and included in all of the Rietveld refinements. We followed the same refinement strategy as that reported in a previous work by the Authors [5]. XRPD was used also to obtain microstructural information. To this aim, the same instrument and setup as that for the QPA was employed, but XRPD patterns were recorded in the 20−145 °2θ range, with a virtual step-scan of 0.1 °2θ, and virtual time per step of 500 s, so to deal with data with a higher signal-to-noise ratio. The whole powder pattern modelling (WPPM) [6], as implemented in the PM2K software package [7], was used for the microstructural analysis of the diffraction data. We adopted the same modelling approach as that we Supporting Information S−3 described in previous researches, assuming crystalline domains being spherical, and their diameter distributed according to a log-normal curve [5].
Spatially-resolved electron energy-loss spectroscopy (SR-EELS), high-resolution scanning transmission electron microscopy (HR-TEM) and scanning TEM annular dark field (STEM ADF) imaging were performed using a FEI Titan Cubed Themis microscope operated at 80 kV. This microscope is equipped with a double Cs aberration-corrector, a monochromator, a X-FEG gun, an ultrahigh resolution energy filter (Gatan Quantum ERS), which allows for working in dual-EELS mode, and a super X EDS detector. The monochromator was excited during imaging and spectroscopic experiments to minimise chromatic aberrations. Convergence and collection angles during EELS experiments were 19 and 42 mrad, respectively. The energy resolution, measured as the full width at half maximum of the zero loss peak, was 0.9 eV with a dispersion of 0.25 eV.pixel -1 . Background subtraction for the core-loss spectra was performed by using the usual inverse power law function. The EELS spectra were acquired in dual EELS mode allowing for a precise calibration of the spectra, and correction of energetic instabilities by recording simultaneously low-loss and core-loss spectra. EELS datasets were processed by using Digital Micrograph and scientific python packages, such as hyperspy [8], which were used for principal component analysis. Interpretations of the Fourier fast transform (FFT) patterns in HR-TEM images were performed by using JEMS software package [9].
Information about the apparent optical band-gap (Eg) of the specimens was investigated via diffuse reflectance spectroscopy (DRS). Optical spectra were recorded on a Shimadzu UV-3100 spectrometer (JP), equipped with an integrating sphere, and a white reference material made of Spectralon; UV−vis spectral range was explored (250−850 nm), using 0.2 nm in resolution. DR data were converted into pseudoabsorption spectra F(R) by means of the Kubelka−Munk (K-M) transformation: where R is the DR [10]. Apparent optical Eg was then estimated adopting the Tauc procedure [11]. (anhydrous basis)) and 2,2'-azino-bis(3-ethylbenzothiazoline-6-sulfonic acid) salt (ABTS) from Sigma-Aldrich were utilised. The solution of ABTS •+ cation radical was prepared by dissolving 17.2 mg ABTS and 3.3 mg K2S2O8 (Sigma-Aldrich) in 5 mL distilled water, and this mixture was left to stand in the dark for 24 h at room temperature, according to Ref. [12]. A stock solution of ABTS •+ was prepared by mixing 1 mL of this reaction mixture with 60 mL deionised water; the concentration of ABTS •+ was determined by UV-Vis spectroscopy
EPR experiments with titania dispersions in DMSO or water were carried out by means of EMXplus X-band spectrometer (Bruker) with a TE102 EPR EMX cavity (Bruker) in small quartz flat cell (WG 808-Q, Wilmad-LabGlass, optical cell length 0.045 cm). Titania stock suspensions (1 mg.mL −1 ) were homogenised in an ultrasonic bath for 1 min before the addition (DMPO or MV 2+ ) and diluted to a final concentration in the range 0.67-0.80 g.L −1 . The prepared suspensions were then carefully aerated by a gentle airstream, or precisely saturated with argon. Samples were irradiated at 295 K directly in the EPR resonator using a LED source (λmax = 365 nm; irradiance 16 mW.cm −2 ), and the EPR spectra were recorded in situ.

S1.3 Gas-solid phase photocatalytic activity
De-NOx and benzene removal gas-solid phase photocatalytic experiments were carried out in the same reactor, that operated in continuous mode [14]. It is made of a cylindrical chamber (3.8 L internal volume) built from a high grade stainless steel with a stainless steel top with a sealed glass window to allow the light to reach the sample that was placed inside. A LED lamp, as those commonly used in domestic households (Philips warm white, 2700 K LED bulb, 9.5 W), was employed to mimic an indoor environment. The LED was placed 12 cm above the photocatalyst. The irradiance reaching the photocatalyst was estimated to be approximately 5.8 mW.cm −2 in the visible-light range and nil in the UV-A range -values measured by means of a radiometer (Delta OHM, HD2302.0, IT). The emission spectrum of the visible-light LED lamp is displayed in Figure S1. A Petri dish, 6 cm in diameter, was covered with a layer of photocatalyst (0.10 g). The tests were performed at 24±1 °C (temperature inside the reactor) with a relative humidity of 36%. These parameters were controlled by a thermocouple that was placed inside the chamber, and a humidity sensor placed in the inlet pipe, and they were stable throughout the tests.
As per the de-NOx photocatalytic experiments, their initial concentration (prepared using synthetic air and NOx gas) was set at 200 ppb. That value is higher than that commonly reported indoor, e.g. according to the EU INDEX project, NO2 concentrations were declared to be in the range 13-62 μg.m −3 (approximately in the range 7-33 ppb) indoors [15]. However, maximum levels associated with the use of gas appliances (gas cooking and heating) in European households were reported to be in the range 180-2,500 μg.m −3 (around 96-1,330 ppb) [16]. The outlet concentration of NOx gases was measured using a chemiluminescence analyser (AC-31 M, Environment SA, FR), according to a procedure that we described in very detail previously [17].
When benzene was used as model gas, its concentration exiting through the outlet was measured using a VOC-72M gas analyser (Environment SA, FR), that is based on gas chromatography and photoionisation detector. The inlet gas mixture (prepared mixing gas cylinders containing synthetic air and benzene) was allowed to flow into the chamber until it stabilised at a concentration of 260 μg.m -3 (80 ppb). This concentration was chosen because the World Health Organisation recommends an exposure to total VOCs < 300 μg.m -3 during 8 h/day [18]. The mixture of air with that concentration of benzene was guaranteed using two mass flow controllers with a flow rate of 150 mL.min -1 . The photocatalytic experiments were assessed by placing the photocatalyst inside the reactor and covering the glass window. Once the desired NOx/benzene concentration was reached and it attained a stable level into the reactor, the window glass was uncovered,

Supporting Information
S−6 the lamp turned on, and the photocatalytic reaction started. Photocatalytic activity for de-NOx abatement and benzene removal was reported as formal quantum efficiency (FQE) [19], defined as the number of molecules degraded per incident photon. It has to be highlighted that FQE represents the lowest estimate of the true quantum efficiency [20]. Besides, aiming to have a better comparison between the tested specimens, data relative to the first 2 min (de-NOx) or 60 min (benzene removal) of reaction time, were interpreted according to a first-order kinetic law. Additionally, the photo-oxidation of NO might give nitrite (NO2 − ), nitrate (NO3 − ), and nitrogen dioxide (NO2) as intermediate products [21]. NO2 is far more hazardous than NO to human health [22]. Therefore, NO conversion (XNO), and selectivity for the formation of ionic species (S), and NO production, are also reported in the discussion of de-NOx experiments. They were calculated according to equations (1) to (3) [21].
In equations (1) to (3), CNO and CNO2 are the concentrations of NO and NO2, respectively, whilst the superscript symbols in and out refer to the inlet and outlet streams. Eventually, in indoor environment, nitrous acid (HONO) can also be formed via heterogeneous hydrolysis of NO2 [23]. However, with the experimental setup used in this research, we were unable to measure its concentration.
Photocatalytic tests were repeated in triplicate to check recyclability and (photo-)stability of the photocatalysts. Commercial nano-anatase powder KRONOClean 7000 (K7000) was used for comparison in the photocatalytic tests. This consisted of a small fraction of anatase crystalline domains in the size range 11.9-15.5 nm, coexisting with a much larger fraction of domains of size 3.8 to 4.4 nm, with an amorphous phase content of around 7.2 wt%, apparent indirect optical band gap equal to 3.28 eV, and having a specific surface area higher than 225 m 2 .g −1 [5]. An aromatic carbon compound (likely aryl carboxylate species) is at the origin of the visible-light absorption [24].

S2.1 Microstructural and optical information
Microstructural information as obtained by means of XRPD methods are listed in Table S1. Pseudo-absorption spectra of unmodified titanias are shown in Figure S2. The change in the apparent optical Eg of specimens modified with 0.5 wt% graphene and 1 mol% Cu, upon visible-light irradiation, are listed in Table S2 1.  In Figure S3 are reported Cu-G1.0/450 gas-chromatograms as extracted from the gas-analyser at selected reaction times, during the photocatalytic experiments. Repeated photocatalytic tests are depicted in Figure   S4 -Cu-G1.0/450.  Figure S5 reports a de-NOx plot upon white-light LED exposure of selected specimens. The photocatalytic activity of Cu-G1.0/450 over three consecutive tests, proving it to be fully re-usable, is displayed in Figure   S6.