Spectral Selective Solar Light Enhanced Photocatalysis: TiO₂/TiAlN Bilayer Films

TiAlN films were prepared on 50 × 50 mm aluminum (Al) substrates heated to about 200 °C by dc reactive magnetron sputtering using a Ti_0.5Al_0.5 target (99.99% purity, Plasmaterials, Livermore, CA, USA) and an argon/nitrogen ratio of 6 and a total pressure of 8 mbar, employing a Balzers UTT400 sputter system, as described elsewhere [23, 24]. Prior to deposition the protective oxide layer on the Al substrates was etched away in H₃PO₄/H₂O solution. To verify the removal of the oxide layer, reflectance was measured in the infrared region using FTIR spectroscopy (Tensor 27, Bruker Optics, Ettlingen, Germany). The thickness of the TiAlN film was measured to be 110 ± 3 nm using profilometry (Bruker DektakXT, Karlsruhe, Germany). A 102 ± 3 nm thick TiO₂ layer was subsequently sputtered on top of the TiAlN film using a Ti target (99.99% purity, Plasmaterials, Livermore, CA, USA) using a substrate temperature of ≈ 230 °C, and an oxygen-to-argon ratio of 0.045. For comparisons a TiO₂ layer with the same thickness was also deposited directly on the etched Al substrates.

The film structure was investigated by grazing-incidence X-ray diffraction (GIXRD) using CuK_α1 radiation employing a Siemens D5000 diffractometer equipped with parallel-beam optics and 0.4° Soller-slit collimator (Bruker AXS, Karlsruhe, Germany). The grazing angle was set to 1.0° and the diffractograms were collected with 1.5 s integration time and 0.02° resolution.

Raman spectroscopy was obtained with a confocal Raman microscope (inVia, Renishaw plc, Gloucestershire, UK) equipped with a 514 nm laser and 1800 groves/mm grating. Neutral density filters were used to attenuate the beam to prevent sample damage. Smoothing of the acquired spectra was done using a Savitzky-Golay filter with a three data point window.

Figure 1a shows Raman spectra of TiO₂ and TiO₂/TiAlN films on Al. The Raman spectrum of TiO₂ reveal the typical anatase E_g peak at 144 cm⁻¹ as well as minor E_g, B_1g, A_1g + B_1g, and E_g peaks at ∼ 197, 397, 515, and 637 cm⁻¹, respectively [25]. Similarly, the Raman spectrum of the TiO₂/TiAlN film shows bands due to anatase TiO₂, which is overlaid a broad background with resolved bands at ≈ 255 cm⁻¹ due to the acoustic phonon mode in crystalline TiAlN, and at ≈ 710 cm⁻¹ due to amorphous TiN_x chains, while the optical phonon mode, which overlaps the anatase E_g mode at 637 cm⁻¹, are not resolved [26, 27]. Figure 1b shows the corresponding XRD diffractograms and reveal only the anatase TiO₂ structure with contributions from the underlying TiAlN films [24, 28]. The resulting films have about equal Ti and Al concentration, as expected from their similar sputtering yield. It is known that Ti_(1−x)Al_xN have the same crystal structure as TiN if x < 0.5. The XRD peaks associated with TiN shift with addition of Al, and for an Al concentration of x > 0.7 the structure also changes to that of AlN [29, 30]; in our case only the TiN phase is observed. Minor contributions from broad TiN reflections, in particular at about 43° and 63° can also be discerned [27], which is consistent with the Raman spectroscopy results of a minor TiN_x contribution.

The reflectance of the films was measured from 300 to 20 µm wavelength in order to determine the normal solar absorptance, \({\alpha }_{sol}\), and normal thermal emittance, \(\epsilon\), of the films, viz.

where \({E}_{sol,T}\left(\lambda \right)\) is the incident solar irradiance (“sol”), or thermal radiation spectrum (“T”) for a black-body at temperature, T, at wavelength λ, respectively,\({R}^{0}\left(\lambda \right)\) is the bulk reflectance, and the integration is performed over the solar spectrum (0.3 µm < λ < 4.0 µm), or the measured thermal radiation region (2.5 µm < λ < 20 µm).

As seen in Table 1 the solar absorptance is about \({\alpha }_{sol}\) = 0.7 for the pure TiAlN films, and the thermal emittance is about \(\epsilon\) = 0.06 at 373 K (approximate working temperature of the solar absorber film). Addition of the TiO₂ layer on top of the TiAlN does not change \({\alpha }_{sol}\) and \(\epsilon\) significantly. The absorptance is higher for the TiO₂/Al than bulk TiO₂ due to absorption in the Al substrate and interference effects.

Acetaldehyde photo-degradation on TiO₂/Al and TiO₂/TiAlN samples was performed in a gas reaction cell operated in flow mode [31]. Briefly, it consists of a stainless steel body with two identical 70 mm × 70 mm glass substrates, which were mounted opposite to each other and separated by 5 mm, so that they enclose a volume of 12.5 mL. The samples were placed on the bottom glass in the reactor and were fully exposed to illumination. Acetaldehyde, CH₃CHO (90 ppm in N₂, 99.99%, Air Liquide) was mixed with synthetic air and diluted to 10 ppm, and 5 × 30 min illumination pulses, followed by 30 min purging in synthetic air between the pulses, were performed. The total flow rate was 150 ml min^− 1. The experiments were conducted in dry gas (RH ≲ 0.5%) and at a cell temperature of 301 K. A capacitance sensor (HIH-4000, Honeywell Inc., USA), positioned in the gas stream, was used for on-line monitoring of the relative humidity. The acetaldehyde concentration, \({C}_{\text{CH}_{3}\text{CHO}}\), was measured in the exit gas stream by a semiconductor gas sensor (HS-130AS, Sencera Co., Taiwan) which recorded the voltage, \(V\), across a shunt resistor coupled in series with the sensing element. Data treatment of the transient responses was performed as described elsewhere [31]. The acetaldehyde concentration, \({C}_{a}\left(V\right)\) is proportional to \({V}^{2}\), and taking into account the baseline voltage and small amounts of residual gases remaining after each acetaldehyde pulse,

where \({V}_{b}=0.2293 V\) is the baseline voltage of the shunt resistor, \({V}_{c}\) is the residual voltage in a purged reactor. \({V}_{10}\) is the voltage in a flow of 10 ppm acetaldehyde in synthetic air. The reaction rate per unit area of the photocatalysts was then calculated as

where \(F\) is the molar flow rate; A is the geometrical film area which for the rather dense films is close to the true film area [16]; \({C}_{a,sat}\) and \({C}_{a,ss}\) are the measured acetaldehyde concentration at a set concentration of 10 ppm using the mass flow controllers, and at steady-state at the end of each illumination pulse, respectively. Prior to the measurements the samples were cleaned in a UV-Ozone reactor for 30 min.

In all photocatalytic measurements a Xe arc lamp (Newport, CA, USA) was employed together with a AM1.5 filter (Oriel Instruments) and a set of filters to direct and focus the light onto the samples. The measured photon power was 209 mW cm⁻², as measured with a thermopile detector (Ophir Optronics, Jerusalem, Israel). The total incident photon power above the measured bandgap energy of the TiO₂ films (\({E}_{g}\gtrsim\) 3.3 eV, or \({\lambda }_{g}\lesssim\) 371 nm) was calculated to be about 5 mW cm⁻². The differential quantum yield, \(\varphi\), was calculated from the measured acetaldehyde photodegradation rates, viz.

where \({k}_{r}\) is the rate in units of mole s⁻¹, \({N}_{A}\) is Avogadros constant and \({\mathcal{F}}_{ph}\) is the absorbed photon flux (photons s⁻¹). The absorbed photon flux in the 102 nm TiO₂ films was calculated to be 4.0 × 10¹⁵ photons cm⁻² s⁻¹, respectively, using Eq. (7):

where \({T}_{g}\) = 0.92 is the transmittance through the cover glass of the cell, \(\alpha \left(\lambda \right)\) is the absorption coefficient calculated from the measured reflectance data and film thicknesses, as described elsewhere [32] and \(\left(\lambda \right)\) is the photon flux from the lamp.

Table 2 shows the reaction rates (Eq. 5) and quantum yield (Eq. 6) for photocatalytic acetaldehyde photo-degradation on a double layer TiO₂/TiAlN/Al solar absorber film upon solar light illumination. As a reference the corresponding data for TiO₂/Al films are also shown. The acetaldehyde photo-degradation rate, \({k}_{r}\), on the TiO₂/Al film is similar to the value which can be extracted from the data reported by Papadimitriou et al. [33], albeit detailed data on e.g. photon power and illumination conditions are missing in this study. In contrast, values for the quantum yield should be independent on experimental setup and suitable for comparisons with published data. Most often data on quantum yield is, however, not reported and rather the quantum efficiency, or formal quantum efficiency (FQE), is used, where the rate is normalized to the incident photon flux. By definition \(\phi\) is always higher than FQE, but for bandgap illumination of TiO₂ where the absorption coefficient is high, \(\phi\) is not much higher than FQE. Nevertheless, our value of \(\phi\) for the TiO₂/Al film is about the same as FQE for P25 powders (the most commonly used reference TiO₂ photocatalyst), and > 10 times higher than the Pilkington Activ^TM TiO₂ coating on glass [34], while \(\phi\) for the TiO₂/TiAlN solar absorber film is almost 6 times higher than reported values of FQE for P25 powders and > 100 times higher than Activ^TM TiO₂. The similarity of FQE for P25 TiO₂ and \(\phi\) for the TiO₂/Al film suggest that the sputtered TiO₂ films do not exhibit different intrinsic electronic properties regarding photo-excitation and recombination as crystalline TiO₂ nanoparticles. Moreover, the difference of \(\phi\) for the double layer TiO₂/TiAlN film and a single TiO₂/Al film seen in Table 2 (more than 8 times higher on the TiO₂/TiAlN film) cannot be explained by structural differences of the TiO₂/Al and TiO₂/TiAlN films; both have very similar structure (Sect. 1.2) and surface roughness (of the order ∼ 1 nm [16]). Instead, the collected data suggest that the different film temperatures under solar illumination due different solar absorptance, shown in Table 1, for the single and multilayer films must be considered to explain the \({k}_{r}\) and \(\phi\) data.

The thin TiAlN solar absorber film absorbs > 70% of the solar light and is readily heated under solar illumination. An estimate of the solar light induced temperature increase of the TiO₂/TiAlN film can be made with reasonable assumptions about the film. We assume a free surface with constant \({\alpha }_{sol}\), and thermal emittance, \(\epsilon\), from Table 1. The stagnation temperature in vacuum, \({T}_{s}\), is then given by

where \({T}_{a}\)is the ambient temperature, \({P}_{sun}\) is the incident photon power and \(\sigma\) is Stefan–Boltzmann’s constant. Heat loss by convection will reduce \({T}_{s}\). In the acetaldehyde photocatalysis reactor, the flow rate of 150 mL/min and the reactor dimension (\(L\) ∼70 mm) with a reactor volume of 12.5 cm³, yields a fluid velocity of about 2 cm/s. Thus with Reynolds number, \(Re\) = 10 for air and Prandtl number, \(Pr\) = 0.7 for air at 300 K, and using the Nusselt number,\(Nu = hL/k\), where \(k\) is the heat conductivity of air (\(k\) = 2.6 × 10⁻² W/mK for air at 300 K), we obtain [35], \(Nu\) = 2 + 0.493\({Re}^{0.5}\) ≈ 4, and hence \(h\) ≈ 11 W/m²K, which is in the range for natural convection (5–25 W/m²K). For small temperature increases we can linearize the radiation term in Eq. 8 and obtain

Equation 9 is a rough estimation of the film temperature neglecting e.g. heat conduction through sample support. The right column in Table 1 shows the estimated steady-state film temperatures for well isolated (free-hanging) TiO₂, TiO₂/TiAlN and TiAlN films on Al. It is evident that a temperature increase of more than 100 degrees can be realized on the thin TiAlN solar absorbing film, which under steady-state conditions also will be the temperature of the overlaid TiO₂ film. In contrast, the TiO₂ absorbs only in the UV part of the solar spectrum, with contributions from the Aluminium substrate (\({\alpha }_{sol} ~ 0.1\)), and the concomitant solar heating of the single TiO₂ film is much less. These results suggest that the reaction kinetics on TiO₂ will be significantly modified, as we discuss in more detail in the next section.

The observed enhancement of \(\varphi\) on the TiO₂/TiAlN double layer film compared to the single TiO₂ film can be understood by considering the temperature increase caused by the absorbed heat in the underlying TiAlN film upon solar light illumination. Two distinctly different effects of the temperature induced film heating can be distinguished. First, for thermally activated reactions, where the reaction kinetics is governed by the Arrhenius factor (\({e}^{-{E}_{a}/kT}\)), as a rule of thumb, the reaction rate doubles for a 10 K temperature increase. In the photo-induced degradations studies here, no reactions take place without bandgap illumination, and merely increasing the temperature of the catalysts without illumination about 100 K does not result in any chemical reaction. In contrast, since desorption is normally the rate limiting step in low temperature catalysis, an increased film temperature may indirectly contribute to an overall increased reaction rate due to an enhanced the desorption rate of reaction products, which can free sites for acetaldehyde adsorption.

Another more subtle effect can also be realized by considering the influence of water coverage. The water adsorption–desorption equilibrium on TiO₂ is shifted from multilayer to submonolayer coverages around room temperature. Assuming zeroth-order water adsorption,

where

is the rate of impinging water molecules from the gas-phase (molecules/m²s), p is the water pressure (in Pa), M is the water molar mass, and \({s}_{0}\) is the sticking coefficient which here is assumed to be \({s}_{0} = 1\). In equilibrium we have

We assume that water desorbs according to

where \({E}_{des}\) is the desorption energy of water, \({N}_{H2O}\)is the number of water molecules/m²., R is the gas constant, and \({\upsilon }_{0}\) is the pre-exponential factor, which we set to \({\upsilon }_{0}={10}^{13}\)s⁻¹. Combining Eqs. 10–13, we obtain

Defining one monolayer (ML) of water to be \({N}_{H2O}\left(1 ML\right)\equiv {N}_{ML}=1.15\bullet {10}^{19}\) molecules/m²,[36] the water coverage, \({\theta }_{H2O}\), is

Equation 15 can be rewritten in terms of relative humidity (RH). At 293 K the saturation water pressure is 2338 Pa and Eq. 15 then becomes (with RH in %):

We can distinguish three regimes with different desorption energy for water. At θ > 2 ML, the vaporization energy of water at 298 K can be used (44 kJ/mole [37, 38]). At coverages 1 < θ < 2 ML, it has been reported that water is bonded stronger in a bilayer configuration, with desorption energy of about 48 kJ/mole [39, 40]. Finally, at θ < 1 ML the desorption energy of water from the (101) anatase surface has been estimated from temperature-programmed desorption measurements to be about 63 kJ/mole (and slightly coverage-dependent) [40]; similar values have been given for anatase nanoparticles from calometric measurements (about 62 kJ/mole) [41], while calculations have predicted lower values (52–59 kJ/mole in the low-coverage regime at θ < 0.5 ML) [42].

Figure 2a shows that in dry air (RH ≤ 0.5%) multilayers (> 2 ML) of water build up on the surface below 300 K, where desorption is governed by the heat of vaporization of liquid water. In the temperature interval 300 < T < 380 K water is bonded in a bilayer configuration, and above this temperature from about 380 K to about 440 K submonolayer of water exists on the surface. At even higher temperatures the surface becomes completely dehydrated. Figure 2b shows that at higher relative humidity, up to RH = 10%, the bilayer region is extended to about 390 K, and complete dehydration occurs only above about 600 K (not shown in Fig. 2b). In Fig. 3, the data in Table 2 for \(\varphi\) and the temperature dependence of the water coverage, shown in Fig. 2a, are combined. In Fig. 3 it is clearly seen that the increased surface temperature dramatically changes the equilibrium water coverage on the TiO₂ surfaces. At low temperature multilayers of water exist and a situation arises where diffusion of pollutants and radicals across the water film is rate determining [22]. The large enhancement of \(\varphi\) for the TiO₂/TiAlN double layer film is directly correlated with a dramatic decrease of the equilibrium water coverage on the illuminated TiO₂/TiAlN film, which brings the water from multilayer to submonolayer coverage. Our data thus supports a mechanism where acetaldehyde photodegradation is diffusion limited at low film temperatures (around room temperature), which corresponds to most practical situation for photocatalytic air purification. A dramatic increased photocatalytic activity on the TiO₂/TiAlN double layer film under solar light illumination (about 2 Suns in our experiments) is thus a result of decreased water coverage, which is further enhanced by an enhanced desorption rate of reaction products due to an increased surface temperature.

The mechanism for photo-oxidation of organics on TiO₂ has been attributed to either direct hole oxidation [43], or OH radical reaction [44]. If water is required as hole acceptor to generate OH radicals it is vital that not all water is desorbed from the surface, which would results in a decreased reactivity and gradual deactivation of the photocatalyst [13, 18]. An interesting implication of this model is that for a given film temperature and relative humidity, and hence a given equilibrium water coverage, it is possible to adjust either the relative humidity or the surface temperature to reach an optimum water coverage, which according to Fig. 3 should be in the submonolayer regime; it is apparent that such optimization can increase the photocatalytic activity by approximately an order of magnitude, which is on par with, or better than, the best TiO₂ photocatalyst dopant schemes, or new catalyst materials [45].