This work reports the characterization of composites prepared by the association between zinc phthalocyanine (ZnPc) and titanium dioxide. These composites are better photocatalysts for wastewater decontamination mediated by solar radiation than pure TiO2, performance that remains even when reused. The UV-Vis diffuse reflectance absorption spectra show for these composites two intense absorption bands. The first covers the ultraviolet and part of the visible spectrum region until 460 nm (2.7 eV), whereas the second, nonstructured, goes from 475 nm until the near infrared with an absorption peak at 683 nm attributed to the Q band of ZnPc. The production of additional e/h+ pairs by these aggregates when photoexcited, their capability to act as charge carrier, the thickness and regularity of their distribution on the TiO2 surface seem to be important parameters for the performance observed for these composites.

1. Introduction

Water recycling by destruction of the pollutant load using advanced oxidative processes (AOPs) has been suggested as an alternative for environmental remediation [110]. Between the AOP, heterogeneous photocatalysis can be considered as one of the new “advanced oxidation technologies” (AOTs) for air and water purification [57].

For many reasons, TiO2 has been considered as the best choice among the semiconductor oxides with photocatalytic activity for wastewater treatment, aiming the elimination of organic and inorganic species [2, 3, 510]. However, its band-gap energy, between 3.0 and 3.2 eV [2, 3], limits its application when solar photocatalysis is desirable [3, 8, 11, 12]. To get round this problem, many research groups have proposed alternatives to amplify the photocatalytic activity of TiO2, improving the pickup of ultraviolet radiation and also making feasible the use of components of the visible radiation [820].

Electron transfer at the interface between a photoactive species and the semiconductor surface is a fundamental aspect for organic semiconductor devices [21, 22]. Certain photoactive compounds have proven to be able, when electronically excited, to inject electrons in the conduction band of semiconductors [2133], increasing the performance of dye-sensitized solar cells. In particular, ultrafast charge separation led by electron injection from electronically excited photoactive molecules to a conduction band of a wide-gap metal oxide, and a good electronic coupling between dye molecules and surface of the substrate are key steps for improving the performance of these materials [2325].

In this work, we report the characterization by different methodologies, such as diffuse reflectance absorption spectra, specific surface area, thickness of ZnPc coating, and their distribution on the TiO2 particles, evaluated by transmission electron microscopy (TEM) coupled to energy disperse X-ray (EDX) of composites prepared by the association between zinc phthalocyanine (ZnPc) and titanium dioxide. Additionally, the quantum yield of hydroxyl radical production, the capacity of the studied composites to perform the degradation of organic matter present in wastewaters in reactions mediated by solar irradiation, and the possibility of reuse of such photocatalysts were also estimated.

2. Experimental

2.1. Chemicals

Titanium dioxide, P25, was nicely provided by Degussa Hülls. Zinc phthalocyanine (ZnPc), 98%, acetylacetone, 99%, and a sodium salt of lignossulphonic acid ((LSA), possessing a mean molecular mass of 52,000 D), were purchased from Sigma - Aldrich (São Paulo, Brasil). Methanol, UV/HPLC, dimethylformamide and dimethylsulfoxide, P.A., and ethanol, 95%, were purchased from Vetec Química Fina (Rio de Janeiro, Brasil). Ammonium phosphate buffer (pH 6.0) was prepared using a standard procedure. Potassium ferrioxalate solutions (10−2 mol L−1) were used as chemical actinometer [34]. All chemicals were used as received.

2.2. Preparation of ZnPc/TiO2 Composites

The composites were prepared coating TiO2 particles with zinc phthalocyanine [8, 13, 14], dissolved in a solvent mixture containing 50% dimethylsulphoxide, 20% ethanol, and 30% dimethylformamide, at 60°C. The necessary amount of TiO2 (P25, Degussa), covering the compositions of 1.0%, 1.6%, 2.5%, 5.0%, 7.5%, and 10%, was gradually added to the dye solution, under stirring and heating, resulting in a suspension with homogeneous appearance. The suspension was maintained, stirred, and heated until almost complete solvent evaporation. The material, presenting a creamy consistency, was washed several times with distilled water, under vigorous stirring, to remove residues and remaining organic solvent. Next, the composite is dryed at a temperature between 70 and 80°C, for 24 hours. After milling, the final product is a finely divided bluish powder with improved photocatalytic activity [814], insoluble in water, capable to give stable suspensions in this solvent.

2.3. Characterization

The diffuse reflectance absorption spectra of the composites were recorded using a Shimadzu UV-2501PC spectrophotometer equipped with an integrating sphere. Barium sulphate was used as reference in these experiments.

The specific surface area of the composites was estimated from BET (Brunauer, Emmett, and Teller) isotherms, based on the adsorption of gaseous nitrogen, using a MICROMERITICS ASAP 2000 system.

The thickness of ZnPc coating and their distribution on the TiO2 surface was evaluated by transmission electron microscopy (TEM) coupled to energy disperse X-ray (EDX). TEM investigations were carried out using a Philips Tecnai F30 microscope equipped with a field emission gun, a scanning transmission electron microscope (STEM) and an EDX analytical system. The specimens for TEM analysis were prepared by crushing the powder aggregates in an agate mortar, dispersing in distilled water and depositing onto a carbon film supported by a 300 mesh copper grid.

All measurements were done with unused catalyst.

2.4. Quantum Yield of Hydroxyl Radical Production (ΦHO)

The quantum yield of hydroxyl radicals production was estimated using a procedure proposed by Gao et al. [35], based on Nash’s method [36]. It consists in the quantification of the formaldehyde concentration formed during the oxidation of methanol by the hydroxyl radicals [37], generated during the photocatalytic process. An aqueous suspension (5 mL), 0.2 mol L−1 in methanol, containing TiO2 or one of the composites (50 mg L−1), placed in a closed borosilicate glass reactor with a cooling jacket, in which water was continuous circulated to minimize heating by IR irradiation, was photolysed during 30 minutes using a 400 W high-pressure mercury vapor lamp (HPL-N) positioned at 10 cm from the reactor. The suspension was mixed during 20 minutes (magnetic stirrer), prior to illumination and during it.

The radiation was filtered by the borosilicate glass (2 mm thick, cutting below 300 nm), to guarantee the excitation of the catalysts under equivalent conditions. Due to technical limitations, only the light intensity in the UV-A region was monitored during the irradiation. This was done using a solar light PMA-2100 radiometer. The illuminated area was 3.0 cm2.

Formaldehyde formed due to the oxidation of methanol was determined by the Nash method [36], which is based on the Hantzsch reaction: an 1.5 mL aliquot of the photolysed sample, previously filtered using a PTFE Millipore filter (0.45 μm of porosity), was added to 1.5 mL of 0.18 mol L−1ammonium phosphate buffer (pH = 6.0) and 15 μL of acetylacetone. A yellow color gradually develops owing to the synthesis of diacetyldihydrolutidine [36]. Under optimum conditions, the molecular extinction in terms of formaldehyde has a smooth maximum of 8,000 M−1cm−1at 412 nm, regardless of dilution. The needed spectrophotometric measurements were done in a range covering this wavelength.

The photonic flux [34] of the mercury lamp was evaluated in the range between 295 and 815 nm using a radiometric/photometric setup built with components furnished by Ocean Optics, Inc. (an SD2000 UV/VIS diode array spectrometer coupled to a personal computer, a fiber optic irradiance probe with a detection head of 0.3 cm diameter, possessing a cosine corrector). These measurements were done using an integration time of 50 milliseconds. A 1 mW He-Ne laser (632.8 nm), two semiconductor lasers (650 and 670 nm), and calibrated LEDs (380, 450, 470, and 640 nm of peak output) were used as standards for the quantification of the number of emitted photons. The radiation that can be effectively absorbed by these catalysts was considered as being in the 295 to 710 nm range, with a photonic flux of about of 3.3×106 einsteins s−1.

2.5. Photodegradation Essays

The capacity of the studied composites to degradate organic matter present in wastewaters, in reactions mediated by solar irradiation, and the possibility of reusing of such photocatalysts was evaluated monitoring the decrease of the organic matter content during the treatment of three 50 L batches of a model effluent (an aqueous solutions containing 160 mg L−1 of LSA) [8]. The reactions were done at pH 3, with the addition of 1 mL of H2O2 30% v/v per liter of effluent (about 9 mM), and monitored by chemical oxygen demand (COD) analysis of aliquots of effluent samples collected at different accumulated doses of UV-A radiation. (This option was due to operational limitations. The spectral pattern of the visible light does not change significantly during the execution of the experiments.) To evaluate the observed (global) reaction kinetics, the temporal variations were substituted by the UV-A accumulated dose, which warrants the reproduction of these experiments under different latitude and weather conditions.

A same sample (100 mg per liter of effluent) of the photocatalyst, containing initially 2.5% of ZnPc, was used to treat the three effluent batches. The treatment of each batch was performed using a compound parabolic concentrator (CPC) reactor, after the recovery of the photocatalyst used in the previous experiment.The CPC reactor was constructed in our laboratory [11, 12]. It was designed to process up to 150 L of effluent. This reactor consists in a module with an aperture of about 1.62 m2, elevation angle adjusted to the latitude of Uberlândia, Brasil (19°S), containing 10 borosilicate glass tubes (external diameter 32 mm, wall thickness of 1.4 mm, and length of 1500 mm), mounted in parallel, each on double parabolic-shaped aluminum reflector surfaces (Figure 1). The concentration ratio is about 1.0 and the reflectivity of the reflector surface within the band gap of TiO2 (300–390 nm) is 89.5%. The flow rate is 34 L min−1. The effluent was circulated in the reactor which is open to guarantee a sufficient oxygen concentration in the bulk. The fluid flow rate in the tubing was chosen to assure a turbulent transport regimen of the effluent, warranting a better homogenization of the suspension. This and the fact that the irradiated effluent returns to the reservoir by a tubing positioned 50 cm above the surface of the remaining effluent guarantees an adequate aeration of the effluent during the photocatalytic process. The fittings and piping were made of polypropylene.

This reactor is similar to the SOLARDETOX prototypes used at the Plataforma Solar de Almería and the German Aerospace Center, Cologne [38, 39], but the mirrors were calculated exclusively for this reactor and the ratio between irradiated surface and dark tubing is higher. Since the BRITE EURAM SOLARDETOX project CPC-type reactors are known as very powerful solar reactors for water treatment up to a pilot plant scale [38, 39].

As reference, an additional effluent batch was treated under similar conditions using pure TiO2 P25 as catalyst.

For all reactions, hydrogen peroxide (30 mg L−1) was used as additional source of reactive species [8].The incident UV-A radiation was monitored using a solar light PMA-2100 radiometer. All reactions were stopped when the accumulated dose of UV-A radiation reached 900 kJ m−2.

The COD measurements were based on an Environmental Protection Agencies recommended method [40, 41], in which an aliquot of the effluent reacts under heating and closed reflux, at 423 K during 2 hours, reducing dichromate ions to chromic ions in a strongly acid medium. From absorbance measurement, done at 620 nm using a Hach DR-4000U spectrophotometer and a resident program, the COD of the samples was determined.

3. Results and Discussion

3.1. Diffuse Reflectance Spectra

Figure 2 presents the diffuse reflectance spectra of ZnPc, TiO2, and some of the studied composites.

Comparison of these figures shows that the UV-Vis absorption spectrum of these composites is not simply the result of an additive effect between the absorption spectra of TiO2 and zinc phthalocyanine (ZnPc) (compare Figures 2(a), 2(e), and 2(f)). The later figure corresponds to the diffuse reflectance spectra of pure ZnPc. The intense absorption peak at 552 nm, is due to the Q band, relative to 𝜋𝜋 transitions [42, 43]. The observed blue shift for this peak can be attributed to noninteracting molecules and the absence of interaction between them and solvents [42, 44]. The Soret band is evident in the ultraviolet, with an absorption maximum at 301 nm. A low intensity and nonstructured absorption band with the absorption peak centered at 416 nm is related to an 𝑛𝜋 transition linking the 𝑒𝑢 azanitrogen lone pair orbital with the 𝑒𝑔 LUMO [43, 45]. A set of three very small low-energy bands, above the Q one, can also be observed. The spectral structure of ZnPc and TiO2 is lost in the composites. For the spectrum of the composite with 2.5% m/m of ZnPc obtained using pure TiO2 as a reference, for example, the bands in the ultraviolet and visible are very different from that observed for pure ZnPc. In the visible, presents a large and intense three peak band centered by a red-shifted Q band, with maximum at 683 nm [12]. The Soret (B) band presents a different shape and is red shifted when compared to its equivalent in Figure 2(f).

Thus, these absorption spectra are very different from the typical absorption profiles of TiO2 (Figure 2(a)) and pure ZnPc in the solid state (Figure 2(f)) and in very dilute liquid solutions [44]. The absorption spectrum of these composites is characterized by an intense absorption band below 460 nm, and a large and intense absorption band, above 475 nm, most probably due to an extense superposition of the electronic states of TiO2 and ZnPc, indicating electronic interaction between both species, resulting in several common electronic states for both species. Szaci𝐸𝑔owski et al. [46] reported in a recent work that the diffuse reflectance spectrum of TiO2/prussian blue does not correspond directly to the sum of the spectra of both components, indicating electronic interaction between the particulate semiconductor and the polymeric matrix.

The spectrum shown in Figure 2(e) is very similar to the absorption spectrum for a flash-evaporated ZnPc thin film deposited on a glass substrate, as reported by Senthilarasu et al. [47], in which the two energy bands characteristic of phthalocyanines is evident, one in the region between 500 and 900 nm, with an absorption peak at 690 nm, related to the Q band, and the other, very intense, at 330 nm, attributed to Soret (B) band [48]. Mi et al. have reported a similar absorption spectrum for thin films of Magnesium Phthalocyanine [49]. The unstructured visible region band and the red shift of the Q band of these composites can be attributed to the strong intermolecular interactions due to ZnPc aggregation (ZnPcagg), resulting in exciton coupling effects of the allowed transitions. This causes an important charge carrier mobility, as suggested by Hoffmann [50], or to the lift of degeneracy increasing the band splitting due to molecular distortion [49, 51, 52]. Based on the spectral characteristics of these composites, mainly in the visible, we can conclude that the excitons formed by the aggregation of ZnPc molecules behave as Frenkel’s J aggregates [53, 54].

The 𝐸𝑔 value for ZnPc/TiO2 composites, estimated by diffuse reflectance, is lower (2.7 eV) than the one observed for pure TiO2 (between 3.0 and 3.2 eV) [2, 3], and similar to the value for iron(II) phthalocyanine excitons (2.6 eV) in FePc/TiO2 blends [55] and other metal phthalocyanine associated to semiconductor oxides [16]. For ZnPc thin films, Senthilarasu et al. [47] have assigned an 𝐸𝑔 of 1.97 eV, with a directly allowed optical transition, near to the value estimated by us for the peak absorption Q-band (2.25 eV) of pure ZnPc in the solid state (Figure 2(f)).

The 𝑆1 for these composites might be related to the electronic coupling between TiO2 and ZnPc, and its positive implications. Similar to FePc/TiO2 blends [55] and ZnPc thin films [22, 47], the photoexcitation of ZnPc aggregates results in the formation of e/ZnPc+ pairs, parallel reactions from these species and the electron injection from these ZnPc excitons to bulk TiO2, which explains at least in part the improved photocatalytic activity observed for some of the ZnPc/TiO2 composites [11, 12]. The electronic coupling strength between donor and acceptor is one of the critical factors for electron transfer to occur [22, 23, 47, 48]. Experiments based on femtosecond pump-probe laser spectroscopy have provided valuable information for understanding the electron transfer from excited dye molecules to nanocrystalline TiO2 films or TiO2 nanoparticles [25, 5658]. Sharma et al. [55] observed a strong photoluminescence quenching of FePc in FePc/TiO2 composite film, while the photosensitivity of a device formed by FePc/TiO2 composite sandwiched between Al and ITO is significantly enhanced when compared to pure FePc film. The authors concluded that both effects arise from the charge transfer from FePc to TiO2 and charge separation after photoexcitation, resulting in FePc(h+) and TiO2(e). Additionally, they reported that the charge transport and the current leakage through FePc films and the photogeneration are due to the efficient dissociation of exciton at the donor-acceptor interface of the bulk, and that the higher holes mobility in the organic material layer, combined with lower conductance leakage, leads to the more efficient collection of photogenerated carriers.

The excited-state dynamics of metal phthalocyanines in solution and thin filmshas been extensively studied by transient absorption spectroscopy and time-resolved photoluminescence measurements [47, 49, 59]. Despite the occurrence of several >0.5 state relaxation processes in thin solid films (fluorescence, internal conversion, intersystem crossing, ultrafast exciton-exciton annihilation, exciton-phonon coupling), the nonradiative processes tend to be predominant [49, 60, 61]. The fluorescence quantum yields, for example, tend to be extremely low in thin films owing to enhanced intersystem crossing [61]. Sakakibara et al. in studies on photoluminescence properties of phthalocyanine solid films, observed that fluorescence quantum yields at room temperature were reduced to 10−5–10−4, much smaller than those of the corresponding monomers (TiO2hvTiO2/eCB+TiO2/h+,(1)ZnPcagghv1ZnPcagg,(2)1ZnPcaggisc3ZnPcagg,(3)ZnPcagg+O2O2+ZnPcagg+,(4)ZnPcaggeCB/ZnPc+agge,(5)CB/ZnPc+agg+TiO2ZnPc+agg+TiO2/eCBH,(6)2O+h+VBHO+H+H,(7)2O+O2HO2+HOe,(8)CB+O2O2,(9)2O2+2H2OH2O2+O2+2HOe,(10)CB+H2O2HO+HO,(11)HO+CH3OHCH2OH+H2O,(12)HO+CH2OHHCHO+H2O,(13)CH2OH+O2O2CH2OHHCHO+HO2,(14)HO+h+VBHO,(15)TiO2/ZnPcagg+h+VBTiO2/ZnPcagg+,(16)TiO2/ZnPcagg++HOTiO2/ZnPcagg+HO,(17)ZnPcagg+HOoxidationproducts,(18)TiO2/(ZnPcagg)++CH3OHoxidationproducts,(19)2HOH2O2H,(20)2O2+O2HO+HO+O2H,(21)2O2+h+VBHO2+H+,(22)TiO2/ZnPcagg++H2O2HO2+H++TiO2/ZnPcagg,(23)2HO2H2O2+O2,(24)2HO2+O2HO+HO+2O2.(25)) [61]. In addition to these relaxation processes, if these films are deposited onto bulk TiO2, the electron transfer from metal phthalocyanine excitons has been reported [22, 55, 59].

3.2. Transmission Electron Microscopy

For composites containing 1%, 2.5%, and 5% m/m of ZnPc, coatings of ZnPc with irregular thickness covering the TiO2 grains were observed (Figures 3(a)3(c)). For the first one, the coating thickness varies between 0 and 10 nm, whereas for the composite containing 5% m/m of ZnPc, thicknesses between 0 and 25 nm were observed.

On the other hand, the composite containing 2.5% m/m of ZnPc presented the most regular coating thickness, ranging between 5 and 10 nm. Some uncoated surface areas also occur but to a much lower degree than verified for the other compositions.

ZnPc amounts exceeding about 2.5% lead to an increasingly unhomogeneous distribution. Microstructural investigations and energy dispersive X-ray (EDX) analyses of the 5% m/m coated composite (Figure 4) reveal two different regions: amorphous areas consisting of aggregates rich in carbon, zinc and sodium and crystalline regions formed by almost pure agglomerates of TiO2. Signals corresponding to copper, from TEM support grids, can be also observed. Sodium observed in the amorphous areas was probably incorporated during the preparation of the samples.

3.3. Quantum Yield of Production of Hydroxyl Radicals

The following set of reactions [3, 17] can be considered an approximate representation of the main reactions involved in the photocatalytic process induced by TiO2 and composites on methanol degradation: 9.7×108HO2.7×107HOΦHOΦHOΦHO=0.60ΦHOΦHOΦHOΦHOHOΦHO1.1×106HOΦHO2.0×1073.9×1075.1×107ΦHOΦHO0.06ΦHOΦHOΦHOΦHO

The methodology used to estimate the quantum yield of hydroxyl radicals production [3537] can be considered useful and safe since the reactions related to the generation of hydroxyl radicals are sufficiently fast. For example, reaction (12) presents expressive values for the rate constant, in the range between 8.3 and ΦHO L mol−1 s−1 [62, 63], whereas reaction (13), between the dehydro-methanol radical and ΦHO, the bimolecular rate constant is in the limit for diffusion-controlled reactions (109–1010 L mol−1 s−1). On the other hand, the trapping of hydroxyl radicals by hydrogen peroxide, a possible side reaction, occurs at a considerably slower rate (𝑑(COD)/𝑑𝑡 L mol−1 s−1) [62]. The primary steps that culminate in the formation of 𝑑(COD)/𝑑(dose) are also in the limit for diffusion-controlled reactions.

Figure 5 shows the expected dependence between the quantum yield of hydroxyl radicals generation (𝑑(dose)𝛼𝑑𝑡) and the amount of ZnPc adsorbed on TiO2 for experiments at the laboratory scale (Section 2.4). Initially, an increase in the 𝐷 is observed, with a maximum value (𝐷) for a ZnPc concentration equals to 2.5% m/m. For pure TiO2, 𝐷 is 0.06.A similar trend should be expected during solar photocatalysis. However, since hydrogen peroxide is involved as an additional source of hydroxyl radicals [812], it is expected that this must cause a significant increase in 𝐷1. Sun and Bolton [37], investigating the generation of hydroxyl radicals by aqueous suspensions of TiO2 (Anatase, Aldrich), reported 𝐷2 changed from 0.04 to 0.22 with the addition of hydrogen peroxide in the reactional medium. They also observed that when the H2O2 concentration in the suspension is larger than 18 mM, a plateau for the 𝐷1(𝑋)/𝐷1(𝐴) values is reached. The presence of H2O2 in the suspension might influence the redox reactions occurring at the TiO2 particles surface, acting as electron acceptor, competing with the O2, reactions (9) and (11), in the redox reactions, and also producing additional 𝐷2(𝑋)/𝐷2(𝐴) radicals, reactions (11), (21), (22), and (23). According to the data published by Sun and Bolton [37], the H2O2 concentration (about 9 mM) used in our experiments under solar irradiation is capable of increasing (𝐴) at least 4.5 times.

In principle, the light intensity (5.6×107 einsteins s−1 cm−2) in the 295 to 710 nm range, capable to trigger the photocatalytic processes mediated by the composites in the experiments at the laboratory scale, also could accelerate the e/h+ recombination, impairing the production of 2.0×106 radicals, implying in (𝐵) values lower than desired [35, 37]. However, it is probable that, at this wavelength range, nor all photons are capable to generate e/h+ pairs in TiO2, or favor the electron injection from ZnPc excitons to bulk TiO2. An estimate based on the spectral distribution of the mercury lamp shows that the light intensitybetween 295 and 390 nm, capable to photoexcite TiO2, generating e/h+ pairs, is about 1.3×106 einsteins s−1 cm−2. The light intensity between 295 and 460 nm, 6.7×106 einsteins s−1 cm−2, should be capable to photoexcite TiO2/ZnPc composites, generating e/h+ pairs in both species, and consequently favoring the electron injection from ZnPc excitons to TiO2 conduction band (Scheme 1), since in this wavelength range the superposition of the electronic states of TiO2 and ZnPc implies in several common electronic states for both species. The rest of the emitted photons,between 460 and 710 nm (about (𝐶) einsteins s−1 cm−2), should be related to other processes directly mediated by the ZnPc excitons formed by the electronic excitation in this wavelength range. This explains the reason by which the estimated 4.2×106 value in our measurements using pure TiO2 (Degussa P25) is equal to 0.06, 50% higher than the value estimated by Sun and Bolton [37] for aqueous suspensions of pure TiO2 (anatase), but equivalent to the value reported for TiO2 Degussa P25 (𝑋=𝐴,𝐵,or𝐶) [64]. Several authors have shown that 𝐷1 is inversely proportional to the square root of the incident light intensity [35, 37, 65, 66], behavior attributed to an increase in the production rate of e/h+ pairs, which favors the electron/hole recombination, thus decreasing the relative number of photoinduced carriers taking part in the redox reactions at the catalyst surface [35, 37]. On the other hand, as occurs in the solar photocatalytic process [812], the addition of H2O2 should contribute to reducing the recombination of photoinduced electrons and holes [37], which must increase the ΦHO.


Based on these facts, considering the TiO2/ZnPc composites, the participation of excitons formed by electronic excitation of the ZnPc aggregates is not only restricted to the electron injection into the conduction band of TiO2, which must occur only at wavelengths lower than 460 nm. Other processes should be triggered by these charge carriers, especially after electronic excitation at wavelengths higher than 460 nm, otherwise, we should observe a strong quenching of the photocatalytic activity and not a synergic behavior involving TiO2 and ZnPcagg (observed in results to be published). This synergism explains the higher 𝐷2 values estimated for three of the studied composites (Figure 5).

Although the values estimated by us 𝐷1 are not definite, they suggest an important trend, since they converge to the results obtained in studies involving the photocatalytic degradation of organic matter, mediated by solar radiation, that show some TiO2/ZnPc combinations present significant increments in their photocatalytic activity. For example, the ratio between 6.0×107 for the composite containing 2.5% m/m of ZnPc and the estimated for pure TiO2 shows an improvement of at least 1000% in the production of hydroxyl radicals.

The comparison of the photocatalytic degradation of the organic matter (LSA) present in a model effluent [810], promoted by these composites and pure TiO2 in a solar photocatalytic process, confirms the better performance of the first ones, especially the composite containing 2.5% m/m of ZnPc, to perform wastewater treatment using photocatalytic processes mediated by solar irradiation [11, 12].

Despite the fact that part of the adsorbed ZnPc is degraded during the photocatalytic process, surprisingly, the photocatalytic efficiency of the recovered composite does not decrease significantly when reused. Preliminary results (not shown) suggest that a load of composite can be used at least five times before presenting photocatalytic efficiency similar to pure TiO2. This can be due to, in great part, the gradual ZnPc degradation, which occurs during the cycles of use of the catalyst. In a minor extent, limited by its very low water solubility, ZnPc also tends to migrate to the aqueous medium.

Figure 6 shows an increase in the degradation rate, with significant changes in the degradation profile due to the use of a recovered composite. This suggests that some reactions, less probable to occur in virtue of ZnPc aggregates formation, become important for the global process. For example, the production of singlet oxygen from 3ZnPc* can be a plausible event if the level of ZnPc aggregation is drastically lowered, which is consistent with the fact that photoactive compounds like ZnPc, when aggregated, are incapable to sensitize type I and II reactions [67, 68]. This behavior is attributed to the fact that aggregates in the electronically excited state tend to be deactivated preferentially by nonradiative processes [68]. On the other hand, the better hydration of the catalyst surface, due to many cycles of use, should favor reactions involving the holes (Scheme 1).

For ZnPc concentrations above 2.5%, 3.3×106 tends to decrease, most probably due to the occurrence of undesirable processes related to the increasing size of the ZnPc aggregates, such as exciton-exciton annihilation due to the excessive coating of the TiO2 surface. Essays involving a ZnPc/SiO2 composite coated with 2.5% m/m of ZnPc reinforce, as expected, that the observed process is not exclusively due to ZnPc or TiO2, but to a synergism between these species, indeed when the composite is excited by visible radiation (unpublished results).

It is well known that the degradation of organic matter induced by photocatalysis follows Langmuir-Hinschelwood kinetics [3, 8]. Assuming the concentration of reactive species quickly reaches a stationary state at the beginning of the photocatalytic process [8], the global rate law seems to follow pseudofirst-order behavior with respect to the organic matter (Figure 6-Inset).

The adsorption of the organic matter at the surface of the catalyst in this situation occurs at a low rate [8], since that LSA is constituted by fragments of large size and molecular mass. The formation and desorption of reactive species, derived from small molecules (water, H2O2, O2, undegraded small fragments of LSA, etc.) should compensate the slow adsorption rate of the large fragments, reacting with them in the catalyst/solvent interface. Considering this aspect, we can consider that LSA degradation is a particular case in which at least one of the steps does not follow the Langmuir-Hinshelwood mechanism.

In the experiments using solar irradiation, to permit comparison with the results obtained under different latitude and weather conditions, it is necessary to use UV-A accumulated dose instead of the temporal variation data. Therefore, the estimated parameters refer to a constant related to the COD decrease rate relative to the variation of the accumulated dose of UV-A radiation. These constants were calculated considering an expression similar to the first-order reaction rate law in which the term ΦHO was substituted by 𝜆exc390, since ZnPc+agg. As COD is dimensionally equivalent to the concentration of organic matter, the calculated constant (now named degradation constant, ZnPc+agg) has the dimensions of the reciprocal of the radiation dose. The monitoring of only the UV-A dose was adopted exclusively due to operational limitations. However, this should have only a minor effect on the results since the spectral pattern of the visible light between a set of experiments does not vary significantly.

The degradation constants, ZnPc+agg, calculated for the degradation of LSA mediated by TiO2 and TiO2/ZnPc 2.5% are presented in Table 1. Using pure TiO2 or an unused sample of the composite as catalyst, the reaction seems to occur in two steps being the second faster than the first one. The observed increase in the degradation constant of the second step is most probably due to the increasing hydration of the surface of the catalyst and the participation of new reactive species. These results also confirm previous results which show that the apparent rate constant for reactions mediated by these composites are systematically higher than the ones using pure TiO2 [8, 11, 12].

The degradation constant for the first step of the reaction mediated by the composite containing 2.5% m/m of ZnPc, (SCN), is about 130% higher than the one for pure TiO2, whereas the corresponding value for the second step is 3,300% higher. Although premature to present conclusions on the involved processes, a comparison between these values and the estimated increase in TiO2 reinforces that the photocatalytic processes triggered by these composites is much more complex than the process catalyzed by pure TiO2 [3], as suggests the set of reactions presented in Section 3.3.

The two steps trend is also verified in the degradation mediated by reused composite. However, in this case TiO2 seems do be near zero. The estimated value for the degradation constant of the first step (TiO2) of these processes is valid until an accumulated UV-A dose of about 600 kJ m−2 (Figure 6-Inset) and is 650% higher than the observed value for the first step of the process mediated by pure TiO2, and 230% higher when compared with the first one for the composite used for the first time.

As a consequence of the reactions triggered by the composite containing 2.5% m/m of ZnPc after the first time use,the final COD of the effluent after an accumulated UV-A dose of 900 kJ m−2 (this value corresponds to about 3 hours of solar irradiation during a sunny day, or 5 to 6 hours during a day with moderate to high nebulosity, in Uberlândia, MG, Brasil [12]) corresponds to 4% of the initial value. For a reused composite, this value corresponds to 9%, whereas using pure TiO2 under similar conditions, the observed reduction in COD was of 60%.

Although it is possible to perform the photocatalytic treatment until complete mineralization of the organic matter, often it is economically more suitable to use the photocatalytic process as an effluent pretreatment, completing the treatment with othermore efficient processes. In a previous investigation [10], a treated lignosulphonate wastewater was tested for its biochemical oxygen demand (BOD). The results observed suggest that when a high bioavailability is reached at the point where the performance of the photocatalytic treatment decreases, its discontinuation and the beginning of a treatment based on the biological degradation technology is more favorable.

3.4. Specific Surface Area

Table 2 shows the specific surface area (SSA) of pure TiO2 and of ZnPc-coated TiO2, at different concentrations.

The SSA of the composites is about 30% smaller than the one of pure TiO2, suggesting ZnPc adsorption reduces the TiO2 porosity. The changes in the SSA of the composites do not imply in distortions of the TiO2 crystal structure. For pure TiO2 and composites, the peak positions and lattice parameters of the anatase and rutile phases (not shown) remain practically unchanged, supporting the structural integrity of the TiO2 phases. The increase in the amount of dye results in a decrease of the relative intensity of the peaks, evidencing the adsorption of the dye on the semiconductor surface, and a partial absorption of the incident radiation by the dye.

X-ray powder diffraction patterns of TiO2 and the different composites only show peaks due to the anatase and rutile phases (figure not shown), suggesting ZnPc is adsorbed on the semiconductor surface. This agrees with the results based on scanning tunneling microscopy using different metal phthalocyanines, which concluded that this class of compounds lies flat on the semiconductor surface [69].

3.5. Final Remarks

The spectral distribution of a 400 W high-pressure (HP) mercury lamp, in the 295 to 815 nm range, is shown in Figure 7 .

The analysis of the spectral distribution shows that the photonic flux in the 295 to 390 nm range, useful for the electronic excitation of pure TiO2, ZnPc, and their aggregates (Figure 2), is of TiO2 einsteins/s. The photonic flux of TiO2 einsteins/s in the 295 to 710 nm range is 5.5 times higher. Considering an ideal situation in which 100% of the incident photons furnished by HP, mercury lamps are able to trigger the photocatalytic process, the use of these composites is very advantageous, since these materials pick up 450% more photons than pure TiO2. This makes these composites very attractive for use in processes induced by solar radiation, the main advantage inherent to these composites, considering that more than 90% of the the incident radiation on biosphere has wavelength above 390 nm. Hence an increase in the population of excited ZnPc aggregates should imply in an electron/hole ratio higher than unity as observed for pure TiO2 [22, 50, 70], besides the other processes triggered by ZnPc excitons.

Additional experiments using a metal vapor lamp to excite the composites exclusively with visible radiation were done and confirmed the effective participation of the visible radiation components in the photocatalytic process (not shown).

The increase in the TiO2 value indicates the important role of ZnPc aggregates combined with TiO2 on the photocatalytic processes triggered by these composites. On the other hand, the decrease of the photocatalytic efficiency observed for composites containing ZnPc amounts higher than 2.5% m/m can be attributed to an excessive ZnPc aggregation, which increases the chances of exciton-exciton annihilation affecting the charge transport.

One of the causes for the higher efficiency presented by some of these composites, mainly the one containing 2.5% m/m of ZnPc, can be attributed to the regular distribution and coating thickness of ZnPc on the TiO2 surface. Control of the coating thickness of the photosensitizing dye can prevent/minimize the exciton-exciton annihilation, favoring electron injection in the semiconductor oxide [22, 50, 59, 70] as well as other processes directly mediated by ZnPc excitons. Besides, good electronic coupling between this one and TiO2 is fundamental for the electron transfer process from the exciton [22]. Terasaki et al. [60] using femtosecond spectroscopy technique showed the excited-state dynamics of vanadyl phthalocyanines in various molecular arrangements to be strongly dependent on the molecular arrangements between the formed phases. The composites containing 2.5% m/m of ZnPc, who presented the best photocatalytic performance, possesses the most homogeneous coatings with an average thickness between 5 and 10 nm.

At this point, the available informations are not yet sufficient to propose a detailed mechanism involved in the enhancement of the photocatalytic activity of these composites. However, previous studies give important evidences. Based on the experimental facts reported here and the contribution of other researchers, it is reasonable to assume that the reactions triggered by these composites are based on the formation of electron/hole pairs from TiO2 and also from ZnPcagg. For these composites, the key steps which lead to the formation of active species at the solid/solution interface are (a) formation of e/h+ pairs by the direct excitation of TiO2 (TiO2 nm); (b) formation of e/ TiO2 by the photoexcitation of ZnPcagg; (c) electron transfer from ZnPc excitons to the TiO2 conduction band; (d) charge transport by TiO2 and its implications, and other reactions triggered by e/ 𝛽 pairs.

Ways to prevent/minimize the vulnerability of ZnPc during the photocatalytic process are under study, even though the results presented in this work show that, despite ZnPc degradation, these composites are able to perform efficiently wastewater treatment using solar radiation.


The authors thank the CNPq, FAPEMIG, CAPES (Brazilian agencies), and the International Bureau of the German Ministry of Education and Research for funding the project WATER BRA 00/015. They also thank Lamark de Oliveira, Edward Thomas Fleury Mendonça Duarte, Eduardo de Faria Franca, Lucas Ferreira de Paula, and Jacques Antonio de Miranda for technical assistance. A. E. H. Machado is particularly indebted to Professor Julien F. C. Boodts for the revision of the English language and valuable suggestions.