Innovative photocatalyst (FeO_x–TiO₂): transients induced by femtosecond laser pulse leading to bacterial inactivation under visible light

Abstract

This study reports the photosensitizing effect/mechanism of FeO_x under visible light irradiation and charge transfer to TiO₂ on FeO_x–TiO₂ cosputtered film. The transients which were photo-induced by femto-second laser pulses at 545 nm (25 fs) were identified, in addition to the concomitant bleaching bands at <440 nm. Electron trapping at 150 fs was observed to compete with electron–hole recombination. A regularized inverse Laplace transform of the transient decay kinetics was worked out to determine the lifetime of the transients. About 50% of the charge carriers recombined within 500 fs, less than 50% of the charges recombined within 25 ps and a small percentage remained trapped. The transient decay within 500 ps did not decay to zero. This is the evidence for the photo-induced charges in FeO_x–TiO₂ leading subsequently to bacterial inactivation in the minute range. The second part of this study addresses the details of the FeO_x–TiO₂–PE photocatalyst performance during E. coli inactivation under solar light irradiation. The bacterial inactivation kinetics was similar under solar or visible light irradiation (>400 nm). This observation unexpectedly confirms the role of FeO_x absorbing light in the visible region. The important role of TiO₂ is reported, leading to bacterial inactivation in the FeO_x–TiO₂ co-sputtered films.

---

Introduction

Bacterial inactivation materials in the form of powders or supported on textiles, polymers, steel and glass have recently used TiO₂, since under sunlight irradiation it produces highly oxidative radicals able to inactivate bacteria and degrade organic compounds,^1–5 and have been reported by many studies out of our laboratory of which we cite only a few.^6–8 But TiO₂ absorbs only 4–5% of the incident solar radiation with a maximum absorption at 392 nm. To enhance the TiO₂ absorption in the visible region, doping with cations such as Cr⁴ and anions such as N⁴ has been tried but they have acted as recombination centres, decreasing considerably the organic compound degradation and the bacterial inactivation yields.^4,5 Another method to increase the visible light absorption of TiO₂ is to couple this wide band-gap semiconductor with a narrow band-gap iron-oxide(s) semiconductor.^9,10 Fe-oxides have been used as TiO₂ dopants due to their small band gap (2.2 eV for α-Fe₂O₃), low cost and non-toxicity as reported in some studies.^11,12 Studies with the FeO_x–TiO₂ bi-functional compound have recently shown an increased visible light absorption compared to bare TiO₂, leading to an acceleration in the degradation of arsenite,¹³ phenol,¹⁴ 4-chlorophenol¹⁵ and humic acid,¹⁶ but no FeO_x–TiO₂ bi-functional photocatalyst has been reported until now to inactivate bacteria under visible light irradiation.

Until now, sol–gel preparations have been used to coat TiO₂ and doped TiO₂ on heat resistant substrates such as glass, ceramics and metal plates, requiring temperatures of a few hundred degrees for adequate adherence to the substrate. But this approach is not possible on low thermal resistance materials like polyethylene (PE), thermally stable up to 98 °C during long-term operation. Also, the thickness of the sol–gel deposited films is not reproducible, they are not mechanically stable, and they exhibit low adhesion and can be wiped off by a cloth or thumb.¹⁷ To deposit adhesive, uniform, tuneable FeO_x–TiO₂ layers, we have chosen in this study to co-sputter FeO_x–TiO₂ films at relatively low temperature for short times on PE. These films will be shown to photo-induce kinetically acceptable bacterial inactivation.

The FeO_x–TiO₂ film was prepared by co-deposition (co-sputtering), to a surface occupied at random by FeO_x and TiO₂ nanoparticles. Physical insight is provided taking into account the random nature of the film microstructure. In this study, we intend to suggest a mechanism for charge injection from the FeO_x absorber into TiO₂ under visible light. Pumping FeO_x–TiO₂ in the visible range generates electron–hole pairs with relaxation dynamics that has been followed by transient absorption (TA).^18–22 Until now, FeO_x–TiO₂ has been investigated using a femto-second pump in the UV-region.²³

In this study we address the following objectives: (a) to report the first uniform, adhesive and reproducible FeO_x–TiO₂ sputtered photocatalyst inducing bacterial inactivation under visible light, (b) to present the first evidence for transients by femto-second laser pulses for the chosen photocatalyst using visible light laser pulses (in the region 545 nm/25 fs) by FeO_x–TiO₂, (c) to suggest a mechanism of charge transfer induced by visible light from FeO_x to TiO₂, based on some new experimental evidence, (d) to report the unambiguous characterization of the oxidative radicals leading to bacterial inactivation, (e) to provide by X-ray photo-electron spectroscopy (XPS) the evidence for the redox catalysis occurring on the FeO_x–TiO₂ surface during the period of bacterial inactivation, and finally (f) to describe the surface properties of the photocatalyst by using several complementary surface science techniques, such as X-ray photoelectron spectroscopy (XPS), X-ray fluorescence (XRF), diffuse reflection spectroscopy (DRS) and inductively coupled-plasma mass-spectrometry (ICP-MS).

Experimental section

Synthesis of the FeO_x–TiO₂–PE co-sputtered films by direct current magnetic sputtering (DCMS), X-ray fluorescence (XRF) and diffuse reflectance spectroscopy (DRS)

The FeO_x was sputtered from a 5 cm diameter target (Kurt Lesker, East Sussex, UK) by reactive direct current magnetron sputtering (DCMS) on PE, applying a current intensity of 200 mA. Since Fe is paramagnetic, the target was modified by creating an artificially eroded area in order to let the magnetic field pass through the target, inducing a hopping trajectory along the target. TiO₂ was sputtered from the 5 cm diameter target (Kurt Lesker, East Sussex, UK) by reactive DCMS on PE, applying a current intensity of 280 mA. The current intensity was adjusted to attain a film composition according to the deposition rate of Fe and Ti in the Ar + O₂ gas atmosphere of the DCMS cavity. The residual pressure P_r in the sputtering chamber was adjusted to 10⁻⁴ Pa. The substrate to target distance was set at 10 cm.

The Fe and Ti content sputtered on the PE film were evaluated by X-ray fluorescence (XRF) in a PANalytical PW2400 spectrometer. The results are shown in Table 1.

Table 1 Fe and Ti-oxide surface loadings sputtered and co-sputtered on FeO_x–TiO₂–PE films detected by X-ray fluorescence (XRF)

	Species	Wt%/wt PE (error% = 0.006)
TiO2–FeOx (co-sputtered for 2 min)	TiO2	0.033
	FeOx	0.045
TiO2–FeOx (co-sputtered for 1 min)	TiO2	0.019
	FeOx	0.025
FeOx (sputtered for 30 s)	FeOx	0.03
FeOx (sputtered for 1 min)	FeOx	0.05
TiO2 (sputtered for 8 min)	TiO2	0.18

---

Diffuse reflectance spectroscopy was carried out on a Perkin Elmer Lambda 900 UV-VIS-NIR spectrometer provided for with a PELA-1000 accessory within the wavelength range of 200–800 nm and a resolution of 1 nm. Table 1 presents Fe and Ti loadings of FeO_x–TiO₂–PE films as detected by X-ray fluorescence.

The polyethylene (PE) film used consisted of highly branched low crystalline semi-transparent film with the formula H(CH₂–CH₂)H. The 0.1 mm thick low-density polyethylene (LDPE) was obtained from Longfellow, UK (ET3112019) and had a density of 0.92 g cm⁻³. PE fabrics were pretreated with RF-plasma as reported in a recent study in our laboratory.¹⁷

X-ray photoelectron spectroscopy (XPS) of the film surface and deconvolution of the Fe and Ti peak signals

The X-ray photoelectron spectroscopy (XPS) of the FeO_x–TiO₂–PE films was determined using an AXIS NOVA photoelectron spectrometer (Kratos Analytical, Manchester, UK) provided with monochromatic AlK_α (hν = 1486.6 eV) anode. The carbon C1s line with a position at 284.6 eV was used as a reference to correct the charging effect. The surface atomic concentration was determined from peak areas using the known sensitivity factors for each element.^24,25 The spectrum background was subtracted according to Shirley.²⁶ The XPS spectral peaks were deconvoluted with a CasaXPS-Vision 2, Kratos Analytical UK.

Femto-second laser spectroscopy

The femtosecond laser spectroscopy setup is shown in ESI (ESI1†). The output of a Ti sapphire oscillator (800 nm, 80 MHz, 80 fs, Tsunami, Spectra-Physics, USA) was amplified by a regenerative amplifier system (Spitfire, Spectra-Physics, USA) at the repetition rate of 1 kHz. The amplified pulses were split into two beams. One of the beams was directed into a non-linear phase-matched optical amplifier with the output centred at 710 nm compressed by a pair of quartz prisms. The Gauss pulse was tuned to 25 fs at 545 nm. The second beam was focused onto a thin quartz cell with H₂O to generate super-continuum probe pulses. The probe pulses were time-delayed with respect to each other. The pulses were then attenuated, recombined, and focused onto the sample cell. The pump and probe light spots had diameters of 300 and 120 μm, respectively. The pump pulse energy was attenuated to 500 nJ, to optimize the light excitation. Experiments were carried out at 278 K. Laser pulse frequency was adjusted by a control amplifier SDG II Spitfire 9132, manufactured by Spectra-physics (USA). The pulse operation frequency was 50 Hz, which is sufficiently low to exclude permanent bleaching of the sample. The set-up allows the change of the amplifier pulse repetition frequency from 0 to 1000 Hz.

The circulation rate in the flow cell was fast enough to avoid multiple excitations in the sample volume. The relative polarizations of pump and probe beams were adjusted to 54.7° (magic angle) in parallel and perpendicular polarizations. The super continuum signal out of the sample was dispersed by a polychromator (Acton SP-300) and detected by CCD camera (Roper Scientific SPEC-10). Transient absorption spectral changes ΔA(t,λ) were recorded within the range of 380–800 nm. Because the super-continuum is chirped, a time correction was applied at each kinetic trace. Control experiments were carried out for non-resonant signals of the coherent spike from the net PE film.²⁷

E. coli inactivation on FeO_x–TiO₂–PE films and irradiation procedures

The samples of Escherichia coli (E. coli K12 ATCC23716) on 2 cm by 2 cm FeO_x–TiO₂–PE were placed into a glass Petri dish and irradiated in the cavity reactor. The irradiation of the samples was carried out in a cavity of a Suntest sunlight simulator with a light dose of 50 mW cm⁻², emitting in the range 310–800 nm. Sample irradiation in the visible region was carried out by addition of a filter in the reactor cavity to block light <400 nm.

The 100 μL culture aliquots with an initial concentration of ∼10⁶ colony forming units (CFU mL⁻¹) in NaCl/KCl (pH 7) were placed on sputtered PE. After preselected irradiation times, the fabric was transferred into a sterile Eppendorf tube containing 900 μL of autoclaved NaCl/KCl saline solution. This solution was subsequently mixed thoroughly using a Vortex. Serial dilutions were made in NaCl/KCl solution. Samples of 100 μL were pipetted onto a nutrient agar plate and then spread over the surface of the plate using a standard plate method. Agar plates were incubated lid down at 37 °C for 24 h before counting. Three independent assays were done for each sputtered sample. The sputtered and unsputtered (control) films were kept in a sterile oven at 60 °C to avoid contamination prior to the bacterial test. The 100 μL bacteria samples were uniformly distributed on the FeO_x–TiO₂ samples. Film irradiation was carried out on Petri dishes provided with a lid to prevent evaporation in the irradiation cavity. The agar was purchased from Merck GmbH, Microbiology division KGaA under the catalogue no. 1.05463.0500. The CFU statistical analysis of the bacteria inactivation data was performed, calculating the standard deviation values.

Ti and Fe detected by inductively coupled-plasma mass-spectrometry (ICP-MS) during the disinfection period

The determination of Ti and Fe was carried out by way of a Finnigan™ ICPS unit equipped with a double focusing reverse geometry mass spectrometer with an extremely low background signal and a high ion-transmission coefficient. The washing solutions from FeO_x–TiO₂–PE samples were digested with 69% nitric acid (1 : 1 HNO₃ + H₂O) to remove the organics in solution and to ensure that there were no remaining ions adhered to the flask wall. The sample droplets were introduced to the ICP-MS through a peristaltic pump to the nebulizer chamber at ∼7700 °C, allowing sample component evaporation and ionization. The Ti and Fe found in the nebulizer droplets were subsequently quantified by mass spectrometry (MS). The results are shown in Table 3.

Results and discussion

Fe and Ti loadings on PE and deconvolution of Fe- and Ti-bands obtained by X-ray-photoelectron microscopy (XPS)

Fe and Ti loadings of FeO_x–TiO₂–PE films are presented in Table 1, as detected by X-ray fluorescence.

Fig. 1a shows the deconvolution of the Fe2p_3/2 peak in the XPS-spectrogram of the co-sputtered FeO_x–TiO₂–PE samples before bacterial inactivation.

Fig. 1 (a) Deconvolution of the Fe2p_3/2 XPS signals in the co-sputtered FeO_x–TiO₂ samples before bacterial inactivation (time zero). (b) Deconvolution of Fe2p_3/2 XPS signals in the co-sputtered FeO_x–TiO₂ samples after bacterial inactivation (time 60 min). (c) Deconvolution of Ti2p_1/2 XPS signals in the co-sputtered FeO_x–TiO₂ samples before bacterial inactivation (time zero). (d) Deconvolution of Ti2p_1/2 XPS signals in the co-sputtered FeO_x–TiO₂ samples after bacterial inactivation.

The peaks in Fig. 1a and b were deconvoluted, showing the Fe₂O₃, FeO, Fe₃O₄ (FeO–Fe₂O₃) and metallic Fe⁰ at the binding energies (B.E.) shown in these figures.

The XPS signals after 60 min bacterial inactivation are shown in Fig. 1b. It is readily seen that the band shifts ≥0.2 eV in Fig. 1b for Fe₂O₃, FeO and Fe₃O₄ with respect to the B.E. shown in Fig. 1a, indicating that redox reactions occur in FeO_x-species during bacterial inactivation. Fig. 1c shows the deconvolution of the Ti2p_1/2 peak in the XPS-spectrogram of co-sputtered FeO_x–TiO₂–PE samples before bacterial inactivation. The Ti³⁺/Ti⁴⁺ shifts as shown in Fig. 1d for the Ti-XPS signals with respect to Fig. 1c, and confirm redox processes taking place between the FeO_x–TiO₂–PE and bacteria.

Table 2 shows the results for the surface atomic percentage concentration on FeO_x–TiO₂–PE before, during and after bacterial inactivation, as detected by XPS.

Table 2 Surface atomic percentage concentration on FeO_x–TiO₂–PE before, during and after bacterial inactivation (60 min) under low intensity solar simulated light as determined by XPS measurements

	Before bacterial inactivation	During bacterial inactivation (30 min)	After bacterial inactivation
Ti2p	27.88	28.01	28.09
Fe2p	7.87	7.54	7.39
O1s	31.11	32.08	35.27
C1s	61.02	61.17	57.34
N2p	0.9	1.26	1.19

---

It is readily seen in Table 2 that the amounts of Ti, Fe, O and N on the 10 upmost layers (2 nm) of the FeO_x–TiO₂–PE film remain unchanged during the 60 min bacterial inactivation period (Fig. 7a, trace 1). This is the evidence for the rapid destruction of the bacteria decomposition residues by the photocatalyst. The C-amount present on the FeO_x–TiO₂–PE film is seen to decrease by ∼4% after bacterial inactivation, due to the oxidation of bacterial carbonaceous residues on the photocatalyst being converted into CO₂.

The signal for the surface peaks of Fe2p decreases modestly during C-oxidation due to the deposition of C-residues left by the bacterial oxidation. The oxidation of the C-residues into more oxidative species is shown by the increase in the O1s surface signals after the bacterial inactivation, shown in Table 2. Finally, the modest increase in the N2p signal observed in the last row in Table 2 is due to the N-containing species left on the photocatalyst film during bacterial inactivation. The data in Table 2 is proof for the rapid destruction of the bacterial residues left by bacterial inactivation. No residues were seen to accumulate on the film surface blocking the catalyst performance.

Diffuse reflection spectroscopy (DRS) and scanning transmission electron microscopy (STEM) of FeO_x–TiO₂ co-sputtered films

Diffuse reflectance spectra (DRS) in Fig. 2 suggest that the hematite is the main component on the FeO_x due to the similarity of both spectra. Fig. 2, trace 2 shows that the diffuse reflectance spectroscopy (DRS) of the PE–TiO₂ sputtered for 8 min, is significantly lower compared to the FeO_x–PE samples sputtered for 2 min (Fig. 2, trace 2) due to the considerably higher molar extinction coefficients of the Fe(III)/(FeO_x)-species, ε = 1000 M⁻¹ cm⁻¹ at 366–375 nm.²⁸

Fig. 2 Diffuse reflectance spectroscopy (DRS) of samples: (1) TiO₂–PE sputtered for 8 min, (2) FeO_x–PE sputtered for 2 min and (3) FeO_x–TiO₂–PE co-sputtered for 2 min. For other details see text.

The spectrum amplitude of Fe₂O₃ between 250 and 500 nm has been reported to be a function of the film thickness and grain size.^1,10,28 In Fig. 2, the addition of traces (1) of TiO₂–PE and (2) of FeO_x–PE does not add up to trace (3) FeO_x–TiO₂–PE. This is the evidence that the new composite formed does not correspond to TiO₂ or to FeO_x. The shift of the band in trace (3) to longer wavelengths upon addition of FeO_x does involve the incorporation of Fe(III) into the TiO₂ through Ti–Fe charge transfer bands. Also, the formation of additional Fe–O transfer bands cannot be discounted.

Furthermore, when the FeO_x–TiO₂ was co-sputtered for 3 minutes the optical absorption in Kubelka–Munk unit was similar to the band obtained by bands co-sputtered for 2 min. This suggests that when co-sputtering for longer times, the FeO_x chromophore did not drastically increase the absorbance, lending further support that a new composite FeO_x–TiO₂ film was deposited on PE. The inset in Fig. 2 shows the second derivative of the FeO_x–TiO₂ spectrum with two bands at 382 nm and 526 nm related to FeO_x–TiO₂, representing the two light absorption maxima.

Fig. 3 presents the scanning transmission electron microscopy in bright field mode (STEM-BF) image of TiO₂–PE films used in the bacterial inactivation (see Fig. 7a, trace 1).

Fig. 3 Scanning transmission electron microscopy in bright field and HAADF modes (STEM-BF) image of PE–TiO₂.

A uniformly dispersed continuous coating was deposited on the PE-surface. The TiO₂ nano-particles represent sizes between 0.5 and 2 nm. The FeO_x particle sizes were reported to be lower or equal to the TiO₂ but could not be unambiguously differentiated.

Femto-second transient absorption spectroscopy of FeO_x–TiO₂ co-sputtered films

Fig. 4 shows the transient absorption spectra of the FeO_x–TiO₂–PE films photo-activated by femto-second laser pulses (25 fs, 545 nm) at different pulse delay times as a function of wavelength. Due to the femto-second pulse excitation in the visible region, only FeO_x absorbs laser pulses, leading to the separation of the conduction band electrons (cbe⁻) and valence band electrons (vbh⁺) or excited d–d states.

Fig. 4 Transient spectra of the FeO_x–TiO₂–PE as a function of wavelength after femtosecond laser pulse of 25 fs (544 nm). Time delays: (1) 150 fs; (2) 500 fs; (3) 1 ps; (4) 3 ps; (5) 10 ps; (6) 500 ps.

The TiO₂ absorption band <400 nm is not excited by the laser pulse, since it is below the absorption range of the 545 nm light pulses. Fig. 4 also shows that the spectral features are about the same for different delays. An increase in the pulse delay up to 500 ps leads to a decrease in absorption bands in Fig. 4. Additional experiments show that the main part of the absorption amplitude at 600 nm decays within 25 ps (t_1/2 ∼ 25 ps). The d–d states are considered local excitons in the FeO_x matrix.

Fig. 4 indicates that the electron trapping process is initiated at times of 150 fs and will compete with electron–hole recombination. Since the spectra cannot be attributed to recombination, the absorption band in Fig. 4 may be assigned to the cbe-spectrum^29–32 of the FeO_x–TiO₂ with a maximum at 600 nm. Until now, no unambiguous identification for the FeO_x or TiO₂ components of the FeO_x–TiO₂ spectrum has been reported.³³ It has been recently reported that hematite electrons relax within 300 fs and then recombine with holes or traps in <5 ps.³¹ Mid-gap Fe d–d states have been suggested as the main trapping sites, showing lifetimes of a few hundred picoseconds (2 ns). In the case of TiO₂, work on dye-sensitized TiO₂ nanoparticles reported electron lifetimes of <100 fs by way of electron injection experiments.³³

Bleaching bands (BL bands) with a delta absorption <0 are shown in Fig. 4 between 390 nm to 440 nm. At time delays of <3 ps, two bleaching bands were observed, the first band between 400 and 440 nm and the second band in the shoulder region of <405 nm. The electron trapping process is initiated at times of 150 fs and will compete with electron–hole recombination. Bleaching bands <400 nm have been observed due to: (a) the depletion of the ground state, (b) the formation of local FeO_x excitons (excited states without charge separation) induced by the 545 nm laser pulses, (c) the electron injection from FeO_x into TiO₂ surface states positioned at ∼1 eV below the TiO₂ cb (−0.1 eV) due to the Kerr effect in the FeO_x–TiO₂ surface,^34–36 a loss of photo-induced electron absorption due to electron trapping and charge recombination significantly decreasing electron injection signals, and finally (d) the depletion of the FeO_x ground state due to the electrons jumping to higher excited energetic states. The bleaching <405 nm in Fig. 4 shifts to a shorter wavelength, due to the Kerr effect.³⁶ The bleaching bands have been reported due to electron trapping/relaxation into localized states of the FeO_x close to the conduction band-edge.³⁸

The TiO₂ cb(e⁻) band overlaps with the FeO_x absorption and contributes to the ESA bands in Fig. 4. Optical absorption from the lowest excited state to the higher lying state is common for various transition metals³⁹, being the excited state absorption (ESA) important in laser spectroscopy.⁴⁰ ESA bands relate to the absorption band of FeO_x cb(e⁻) and FeO_x vb(h⁺). The charge-separation in oxides has been reported to occur within 1 ps.⁴¹ The ESA band shapes have been reported with an optical absorption similar to hematite colloids. The bleaching shown in Fig. 4 was due to FeO_x and with some additional effects due to the TiO₂ low-lying surface states.^1,2

Meaningful transient absorption kinetics within the femto- to sub-nanosecond range is shown in Fig. 5.

Fig. 5 Transient absorption decay profiles of FeO_x–TiO₂–PE films for probe wavelength 688 nm (-■-), 518 nm (-▲-) and 405 nm (-●-) in time windows: (a) 0–2 ps, (b) 0–15 ps and (c) 0–500 ps. Fitting incorporated a non-linear least-squares algorithm to curves: (a) was one-exponential fitting y₀ + A₁ exp(−invTau1t), in (b) and for (c) a double-exponential fit y₀ + A₁ exp(−invTau1t) + A₂ exp(−invTau2t). The negative OD absorbance region is due to bleaching. The parameters of fitting are shown in ESI2.†

The femto-second decay transients present non-exponential transient decay. The non-exponential character of the transient decay is confirmed by the fact that a satisfactory fitting is achieved only in the time ranges: (a) 0–2 ps, (b) 0–15 ps and (c) 0–500 ps. For time window (a), the best χ² fit can be achieved by 1-exp y₀ + A exp(−invTau × t) function, whereas for the time window (b) and (c), the best χ² fit is 2-exp function y₀ + A₁ exp(−invTau1 × t) + A₂ exp(−invTau2 × t), with time constants indicated in ESI (ESI3†).

The analysis of the rate constants in the ESI (ESI4†) shows that the values obtained by fitting of the data in Fig. 4 depend on the selected time window. One or two exponential models fit the transient decay poorly. Non-exponential decays indicate a dispersive charge recombination or exciton decay with a high degree of disorder in their charge distribution.^37,38

The non-exponential decay shown in Fig. 5 in the time window from 100 fs to 500 ps was initially fitted by a Kolrauch’s stretched exponential fitting^7,37,38 (see ESI (ESI3†)). But the Kolrauch’s fitting did not fully match the experimental data in this time window. For this reason, we used the regularized inverse Laplace transform.¹⁵ Four characteristic times for the FeO_x–TiO₂–PE decay up to 500 ps were obtained and the fitting of the data is presented in Fig. 6 (ESI1†).

Fig. 6 The results of the analysis carried out by a regularized inverse Laplace transform. Kinetic transients at wavelengths: (a) 405 nm; (b) 518 nm; and (c) 688 nm.

The results in Fig. 6 show the dependence of peak amplitude vs. time constants τ_i in eqn (1).

The peaks at 0.26 ps, 1 ps, 37 ps and 624 ps in Fig. 6a show the characteristic time constants required to fit the decay transient. The same approach was used to find the characteristic time constants shown in Fig. 6b at 518 nm. Fig. 6a shows a value of 0.26 ps for the faster rate near 200 fs followed by values of 1 ps, 37 ps and finally 624 ps; similar values at 688 nm are reported in Fig. 6c.

Bacterial inactivation kinetics followed by plate counting under sunlight, visible light and in the dark

Fig. 7a, trace 1 presents the E. coli bacterial inactivation kinetics by FeO_x–TiO₂–PE co-sputtered samples for 2 min, leading to the fastest bacterial inactivation (60 min) under simulated solar light (310–800 nm). Fig. 7a, trace 2, shows the inactivation for FeO_x–PE sputtered for 1 min within 120 min. This is a faster bacterial inactivation compared to Fig. 7 trace 3, a film with a thicker PE–TiO₂ coating sputtered for 2 minutes. The latter coating presented: (a) an increase in the layer thickness leading to bulk inward diffusion of the charge carriers^11,12 and (b) FeO_x-clusters with larger size (agglomerates) causing a decrease in the catalytic activity per exposed atom. The TiO₂–PE sample in Fig. 7a, trace 4 shows slower kinetics due to the low fraction of the sunlight absorbed by TiO₂ (<390 nm) compared to FeO_x. Control experiments show that bacterial inactivation does not proceed on PE under sunlight (Fig. 7a, trace 5) or in the dark (Fig. 7a, trace 6).

Fig. 7 (a) E. coli inactivation under low intensity simulated solar light (50 mW cm⁻²) on: (1) FeO_x–TiO₂–PE co-sputtered for 2 min, (2) FeO_x–PE sputtered for 1 min, (3) FeO_x–PE sputtered for 2 min, (4) TiO₂–PE sputtered for 8 min, (5) light only and (6) FeO_x–TiO₂–PE co-sputtered for 2 min in the dark. (b) E. coli inactivation on FeO_x–TiO₂–PE co-sputtered for 2 min under low intensity solar simulated light (50 mW cm⁻²): full sunlight irradiation (310–800 nm) and in the presence of a cut-off filter letting light ≥400 nm pass. (c) E. coli inactivation on FeO_x–TiO₂–PE co-sputtered for 2 min at different solar light intensities: (1) 50 mW cm⁻², (2) 70 mW cm⁻² and (3) 30 mW cm⁻².

Consistent with the results obtained in Fig. 7a, electron transfer from the photo-sensitizer FeO_x to the lower lying TiO₂ trapping states is suggested as noted in Scheme 1. Leytner et al.,³⁹ used time-resolved photo-acoustic spectroscopy (TRPAS) and identified the existence of electron trapping sites within anatase TiO₂ that are 0.8 eV below the anatase cb. Gray et al.,^40,41 used electron paramagnetic resonance (EPR) spectroscopy to study the charge transfer between the mixed TiO₂ (anatase–rutile) interface. They reported that even though the rutile cb is about 0.2 eV lower than the anatase cb, the electrons can transfer from rutile cb to electron trapping sites of anatase, considering that these trapping sites were located ∼0.8 eV lower than the anatase cb. Under visible light irradiation, electron-pairs cannot be produced on TiO₂ but only by FeO_x (mainly Fe₂O₃), since the cb band of Fe₂O₃ is located 0.5–0.8 eV lower than that of TiO₂.^1,2,32,33

Scheme 1 Suggested electron transfer from the photosensitized FeO_x to the low-lying TiO₂ trapped states. For further details see text.

The mechanism of the reaction between FeO_x–TiO₂–PE and bacteria under visible light is suggested below by eqn (2) through to eqn (8).

FeO_x + vis light → FeO_x (e⁻) + FeO_x (h⁺) (2)

FeO_x (e⁻) + TiO₂ → FeO_x + TiO₂ (e⁻_{trapped site}) (3)

FeO_x (e⁻) + O₂ → FeO_x + O₂⁻ (4)

O₂⁻ + H⁺ → HO₂˙ → ROS (reactive oxygen species) (5)

TiO₂ (e⁻_{trapped site}) + O₂ → TiO₂ + O₂⁻ → followed by eqn (5) (6)

TiO₂ (h⁺) + bacteria → CO₂ + H₂O + inorganic/organic residues (7)

An alternative shorthand notation for the visible light induced bacterial inactivation could be suggested by eqn (8):

Bacteria + [FeO_x–TiO₂–PE] + vis light → [bacteria⋯Fe*O_x–TiO₂]PE → [bacteria*⋯FeO_x–TiO₂]PE → [bacteria⁺⋯FeO_x–TiO₂ (e⁻)]PE (8)

The bacteria (RH) interact with the TiO₂ vb(h⁺) within the time of the laser pulse and compete with the charge recombination reaction e⁻ + h⁺ → decay. The decay rate of the cbe⁻ slows down in the presence of bacteria as reported previously by Nadtochenko et al.²⁷

Fig. 7a, trace 1 shows the fast bacterial inactivation kinetics on FeO_x–PE compared to FeO_x–TiO₂–PE due to two effects: (a) the electron transfer from the photo-sensitizer FeO_x to lower lying TiO₂ trapping states and (b) the formation of additional FeO_x interfacial sites active in bacterial oxidation.

The effect of the visible light irradiation on bacterial inactivation compared to the full sunlight irradiation by a Suntest solar simulator (310–800 nm) is shown next in Fig. 7b. The E. coli inactivation kinetics under Suntest simulator full spectral light or under visible light >400 nm was seen to be similar. Therefore, the more energetic photons between 310 and 400 nm do not lead to any significant acceleration of the bacterial inactivation. This unexpected result lends further support to the predominant role of FeO_x as the photo-sensitizer during the bacterial inactivation by FeO_x–TiO₂–PE films. Fig. 7c presents the faster bacterial inactivation kinetics as a function of the applied sunlight dose.

It is readily seen from the data presented in Fig. 7c, that a light dose of 50 or 70 mW cm⁻² led to bacterial inactivation within 60 min. Applying a lower light dose of 30 mW cm⁻², FeO_x–TiO₂–PE shows the semiconductor character of the composite film at lower light intensities.

The stability of the FeO_x–TiO₂–PE inducing repetitive E. coli inactivation is shown in Fig. 8. The bacterial inactivation kinetics slows down marginally only after the 7^th cycle. Table 2 presents the inductive coupled plasma mass spectrometry (ICP-MS) determination of Ti and Fe after the seventh cycle.

Fig. 8 Recycling of the FeO_x–TiO₂–PE photocatalyst film during bacterial inactivation. The samples were irradiated by way of a solar simulator (50 mW cm⁻²).

In Table 3, Ti could not be quantitatively detected since the limit of detection of the ICP-MS unit is 0.1 ppb. In the case of Fe, a value of 4.2 ppb was observed far below the Fe-level cytotoxic limit of 200 ppb set for mammalian cells.⁴² At these ppb low levels of Fe, the interaction between the Fe-ions and the bacteria seems to proceed through an oligodynamic effect. The oligodynamic effect consists of bactericide effect induced by an extremely low concentration of metal.

Table 3 Ion-release after the 7^th recycle from FeO_x–TiO₂–PE as detected by ICP-MS

	Ti (ppb)	Fe (ppb)
TiO2/FeOx (8 min/30 s)	n.d.	3.5
TiO2/FeOx (8 min/60 s)	n.d.	4.1
FeOx (30 s)	—	6.0
FeOx (60 s)	—	9.0
TiO2–FeOx (co-sputtered for 1 min)	n.d.	0.2
TiO2–FeOx (co-sputtered for 2 min)	n.d.	0.5

---

Scavenging of oxidative radicals (ROS) during bacterial inactivation in aerobic media

Photocatalytic inactivation of bacteria in aerobic conditions proceeds by highly oxidative radicals OH˙, HO₂˙/O₂⁻ and cb(h⁺). Fig. 9 presents the scavenging by dimethyl-sulfoxide (DMSO), superoxide dismutase (SOD) and ethylenediaminetetraacetic acid di-sodium salt (EDTA-2Na) to sort out the roles of OH˙, O₂⁻ and TiO₂ vb(h⁺)/FeO_x vb(h⁺) intervening in bacterial oxidation.

Fig. 9 E. coli inactivation kinetics on FeO_x–TiO₂–PE co-sputtered for 2 min by itself or in the presence of ROS scavengers under sunlight simulator irradiation of 50 mW cm⁻². For more details see text.

It is seen that the bacterial inactivation mediated by FeO_x–TiO₂–PE co-sputtered for 2 min proceeds within 60 min as already shown above in Fig. 7a, trace 1. The kinetics slows down in the presence of the three added scavengers as shown in Fig. 9. This means ROS mediate the bacterial inactivation. Dimethyl-sulfoxide (DMSO) and superoxide-dismutase (SOD) have been widely reported to scavenge the OH˙ and HO₂˙⁻ radicals, respectively.^1,2 Ethylenediaminetetraacetic acid di-sodium salt (EDTA-2Na) shows in Fig. 9 that the most effective species leading to bacterial inactivation was TiO₂ vb(h⁺)/FeO_x vb(h⁺). The toxicity effect of the Fe-ions does not affect the bacterial inactivation in anaerobic conditions, since it is below the cytotoxicity levels conforming with the present health standard regulations⁴² (see Table 3).

Conclusions

Femtosecond spectroscopy was employed to report the generation of transients in FeO_x–TiO₂–PE under 545 nm/25 fs femto-second pulses. Transient decay comprised ultrafast charge recombination, charge separation, charge trapping and low-lying trapping states in TiO₂. The transient kinetics revealed that the transients induced by the 545 nm femto-second pulses do not decay to zero within a few hundred ps as reported for hematite by some recent studies. This study shows that ∼10% of the transients generated >500 ps by the femto-pulse laser were available to induce at later stages bacterial inactivation in the minute range. This work addressed the FeO_x photo-sensitizer features in the FeO_x–TiO₂–PE film under light irradiation >400 nm. The ad-atoms and agglomerate diffusion of FeO_x and TiO₂ in FeO_x–TiO₂–PE thin film determine the random distribution of these species at the photocatalyst surface.

The second part of this investigation presented the FeO_x–TiO₂–PE photocatalysis leading to E. coli inactivation. Unexpectedly, the E. coli inactivation kinetics was the same under the full sunlight emission or after insertion of a 400 nm cut-off filter. This observation lends further support to the photosensitizing role of FeO_x, leading to bacterial inactivation by FeO_x–TiO₂–PE.

Acknowledgements

We thank the EPFL, the Swiss National Science Foundation (SNF) Project (200021-143283/1), the EC7th Limpid FP project (Grant No. 3101177) and the Russian Foundation for Basic Research under Grants 14-03-00546, 13-02-12433, 14-03-93181 for financial support. We also thank the COST Action 1106 for interactive discussions during the course of this study.

References

1. A. Fujishima, X. Zhang and D. Tryck, Surf. Sci. Rep., 2008, 63, 515–582.
2. J. Schneider, M. Matsuoka, M. Takeuchi, J. Zhang, Y. Horiuchi, M. Anpo and D. Bahnemann, Chem. Rev., 2014, 118, 9919–9986.
3. H. Foster, I. Ditta, S. Varghese and A. Steele, Appl. Microbiol. Biotechnol., 2011, 90, 1847.
4. M. Pelaez, N. Nolan, S. C. Pillai, M. Seery, P. Falaras, A. Kontos, P. S. M. Dunlop, J. Hamilton, J.-A. Byrne, K. O’Shea, M. Entezari and D. Dionysiou, Appl. Catal., B, 2012, 125, 331.
5. D. Keane, K. Mc Guigan, P. Fernandez, I. Polo-Lopez, J.-A. Byrne, P. S. M. Dunlop, K. O’Shea, D. Dionysiou and S. C. Pillai, Catal. Sci. Technol., 2014, 4, 12111.
6. J. Kiwi and V. Nadtochenko, J. Phys. Chem. B, 2004, 17675.
7. J. Nesic, S. Rtimi, L. Laub, C. Pulgarin, G. M. Roglic and J. Kiwi, Colloids Surf., B, 2014, 123, 593.
8. S. Rtimi, C. Pulgarin, R. Sanjines, V. Nadtochenko, J.-C. Lavanchy and J. Kiwi, ACS Appl. Mater. Interfaces, 2015, 7, 12832.
9. S. Rtimi, C. Pulgarin, A. Houas, R. Sanjines, J.-C. Lavanchy and J. Kiwi, J. Hazard. Mater., 2013, 260, 860.
10. D. Adams, L. Brus, C. Chidsey, S. Creager, C. Creutz, C. Kagan, P. Kamat, M. Lieberman, S. Lindsay, R. Marcus, R. Metzger, M. Michel-Beyerle, J. Mille, M. Newton, D. Rolison, O. Sankey, K. Schanze, Z. Yardley and X. Zhu, J. Phys. Chem. B, 2003, 107, 6668.
11. M. Mishra and D.-M. Chen, Appl. Catal., A, 2015, 498, 126.
12. J. Kiwi and M. Graetzel, J. Chem. Soc., Faraday Trans. 1, 1987, 83, 1101.
13. W. Zhou, H. Fu, K. Pan, C. Tian and Y. Qu, J. Phys. Chem. C, 2008, 112, 19584.
14. Y. Cong, Z. Li, Y. Zhang, Q. Wang and Q. Xu, Chem. Eng. J., 2012, 191, 356.
15. B. Palanisamy, C. Babu, B. Sundarevel, S. Anandan and V. Murugesan, J. Hazard. Mater., 2013, 252, 233.
16. Q. Zhang, G. Rao, J. Rogers, C. Zhao, L. Liu and Y. Li, Chem. Eng. J., 2015, 271, 180.
17. S. Rtimi, R. Sanjines, M. Andrzejczuk, C. Pulgarin, A. Kulik and J. Kiwi, Surf. Coat. Technol., 2014, 254, 333.
18. M. Horng, J. Gardecki, A. Papazyan and M. Moroncelli, J. Phys. Chem., 1995, 99, 17331.
19. T. Pollard and S. Mathies, Annu. Rev. Phys. Chem., 1990, 41, 647.
20. W. Pollard, B. Brito-Cruz, C. Shank and R. Mathies, J. Chem. Phys., 1989, 90, 199.
21. T. Kang, J. Yu and M. Berg, J. Chem. Phys., 1991, 94, 2413.
22. S. Kovalenko, N. Ernsting and J. Ruthman, Chem. Phys. Lett., 1995, 258, 445.
23. L. Fu, Z. Wu, X. Ai, J. Zhang, Y. Nie, S. Xie, G. Yang and B. Zou, J. Chem. Phys., 2004, 120, 3406.
24. Handbook of X-Ray Photoelectron Spectroscopy, ed. C. D. Wagner, M. W. Riggs, E. L. Davis and G. E. Müllenberg, Perkin-Elmer Corp. Phys. Electron Div, 1979, p. 197.
25. J. Nogier, M. Delamar, P. Ruiz, M. Gratzel, R. Thampi and J. Kiwi, Catal. Today, 1994, 20, 109.
26. A. D. Shirley, Phys. Rev. B: Solid State, 1972, 5, 4709.
27. V. Nadtochenko, N. Denisov, V. Gak, E. Gostev, A. Titov, M. Sarkisov and V. Nikandrov, Russ. Chem. Bull., 2002, 51, 457.
28. K. Hardee and A. Bard, J. Electrochem. Soc., 1978, 125, 215.
29. V. Nadtochenko, N. Denisov, O. Sarkisov, D. Gumy, C. Pulgarin and J. Kiwi, J. Photochem. Photobiol., A, 2006, 181, 401.
30. J. Cherepy, P. Moore and A. Mathies, J. Phys. Chem. B, 1997, 101, 3250.
31. A. Joly, J. Williams, A. Chambers, G. Xiong, W. Hess and D. Laman, J. Appl. Phys., 2006, 99, 053521.
32. S. Pendlebury, X. Wang, F. Le Formal, M. Cornuz, A. Kafizas, S. Tilley, M. Gratzel and J. Durrant, J. Am. Chem. Soc., 2014, 136, 9854.
33. B. C. Fitzmorris, J. M. Patete, J. Smith, X. Mascorro, S. Adams, S. S. Wong and J. Z. Zhang, ChemSusChem, 2013, 6, 1907.
34. D. Beydoun, R. Amal, G. Low and S. McEvoy, J. Nanopart. Res., 1999, 1, 439.
35. N. Geoffrey, Introduction to Nonlinear Optics, Cambridge University Press, 2011, ISBN 978-1-139-500760.
36. D. Wheeler, G. Wang, Y. Ling, Y. Li and J. Zhang, Energy Environ. Sci., 2012, 5, 6682.
37. J. L. Andrews and V. S. Hitelman, in Spectroscopy of Solid-State Laser Materials, ed. B. di Bartolo, Plenum Press, N.Y, USA, 1987, p. 513, W. Majorana Int. Sci. Ser.
38. S. Rtimi, C. Pulgarin, R. Sanjines and J. Kiwi, RSC Adv., 2015, 5, 80203.
39. S. Leytner and J. T. Hupp, Chem. Phys. Lett., 2000, 330, 231.
40. D. C. Hurum, A. G. Agrios, K. A. Gray, T. Rajh and M. C. Thurnauer, J. Phys. Chem. B, 2003, 107, 4545.
41. D. C. Hurum, A. G. Agrios, S. E. Crist, K. A. Gray, T. Rajh and M. C. Thurnauer, J. Electron Spectrosc. Relat. Phenom., 2006, 150, 155.
42. S. Hilger, W. Fruhauf and W. Linss, J. Magn. Magn. Mater., 2003, 261, 7.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ra18960h

---