Applied Catalysis B: Environmental 60 (2005) 1–11 www.elsevier.com/locate/apcatb
UV-induced photocatalytic degradation of azo dyes by organic-capped ZnO nanocrystals immobilized onto substrates R. Comparelli a, E. Fanizza a, M.L. Curri b,*, P.D. Cozzoli a, G. Mascolo c, A. Agostiano a,b b
a Dipartimento di Chimica, Universita` di Bari, via Orabona 4, I-70126 Bari, Italy CNR IPCF – Sez Bari c/o Dip. di Chimica, Universita` di Bari, via Orabona 4, I-70126 Bari, Italy c CNR-IRSA – Sez. Bari, Via F. De Blasio 5, I-70123 Bari, Italy
Received 19 November 2004; received in revised form 14 January 2005; accepted 15 February 2005 Available online 14 March 2005
Abstract ZnO nanocrystals (mean particle size: 6 nm) with different surface organic coating and commercial ZnO powder (mean particle size: 200 nm) have been immobilized onto transparent substrates and comparatively examined as photocatalysts for the UV-induced degradation of two azo dyes, Methyl Red and Methyl Orange, in water. The effects of the pH, of the catalyst surface status, and of the dye chemical structure on the course of the photocatalysis are discussed. Reasonable degradation pathways for both target molecules are proposed on the basis of the structural identification of several by-products. The results demonstrate that surfactant-capped ZnO nanocrystals exhibit more versatile performances than those of conventional ZnO-based photocatalysts, because the surface organic coating makes the oxide resistant to photocorrosion and to pH changes. Surface-protected ZnO nanocrystals can be regarded as a valuable alternative to standard TiO2 photocatalysts. # 2005 Elsevier B.V. All rights reserved. Keywords: Nanocrystals; ZnO; Organic capping; Azo dyes; UV-induced photocatalysis
1. Introduction Semiconductor-assisted photocatalysis has attracted considerable attention among advanced oxidation process (AOP) as a promising tool for implementing the large-scale purification of wastewaters at low cost. This methodology exploits the strong reactivity of hydroxyl radicals in driving oxidation processes, ultimately leading to the extensive mineralization of a variety of environmental contaminants [1,2]. Most studies have focused on large bandgap semiconductors oxides, such as TiO2 and ZnO, whose photoexcitation by UV light provides electron–hole pairs able to initiate the production of hydroxyl radicals in water [3–5]. Although * Corresponding author. Tel.: +39 080 5442027; fax: +39 080 5442128. E-mail address:
[email protected] (M.L. Curri). 0926-3373/$ – see front matter # 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.apcatb.2005.02.013
such photocatalysts have been generally applied in the form of aqueous suspensions [7–15], more recently, oxide powders have also been used as thick films deposited onto suitable substrates in order to overcome the technological problems related to the catalyst recovery and recycle. Unfortunately, as a drawback of immobilization, the active surface area of the photocatalyst is dramatically reduced, in turn leading to a relevant decrease in the performances [16,17]. This fundamental disadvantage can be significantly offset by the use of nanostructured semiconductors. When the catalyst grain size is reduced down to a few nanometers, an elevate density of active sites for substrate adsorption and/or catalysis can be guaranteed, as small particles possess a significantly higher surface-to-volume ratio as compared to the bulk material. These characteristics thus offer the potential for gaining elevate performances even with immobilized photocatalysts. Furthermore, when the
2
R. Comparelli et al. / Applied Catalysis B: Environmental 60 (2005) 1–11
nanocrystal size is comparable or smaller than the bulk exciton diameter, the bandgap becomes size-dependent due to quantization effects [3,18]. This opens the opportunity of designing selective photochemical reactions by means of sizemodulating the electron–hole redox potentials, provided that a narrow size distribution is available. However, an efficient photocatalytic process requires highly crystalline semiconductors to minimize electron–hole pairs loss owing to the trapping of either charge carriers at defect states [4,6]. Although TiO2 is universally recognized as the most photoactive catalyst, many reports have highlighted the effectiveness of ZnO in removing organic compounds in water matrices [13,14,19–24]. Recently, the emitting properties of ZnO have been made possible to set up an original catalytic system able to ‘‘sense and shoot’’ environmental contaminants [25], thus motivating further exploration of the properties of this oxide. Nevertheless, the susceptibility of ZnO to facile dissolution at extreme pH values as well as to photocorrosion with subsequent release of metal ions into the environment remain the major obstacle for making this semiconductor competitively useful in photocatalysis [26–29]. In a previous paper, we have demonstrated that the presence of passivating molecules on the oxide surface can affect the photocatalytic activity of organic-capped TiO2 nanocrystals to a remarkable extent [30]. The present work explores further this topic considering the case of zinc oxide. For this purpose, organic-coated ZnO nanocrystals with a narrow size distribution were obtained by two colloidal routes, being basically characterized by the type of reaction, i.e. either hydrolytic or non-hydrolytic, respectively, involved in ZnO formation. The resulting ZnO nanocrystals essentially differed with respect to: (1) the steric hindrance offered by the surface ligand, i.e. either hexadecylamine/tbutylphosphonic acid or acetate moieties; (2) the nature of the surfactant functional group binding to the crystal surface, i.e. either –NH2/–PO3H2 or –CO2; (3) the residual availability of surface terminating –OH groups. The ZnO nanoparticles were deposited onto a transparent support to photocatalyze the degradation of Methyl Red and Methyl Orange chosen as model compounds. The photodegradation course was monitored by UV–Vis absorption and HPLC-MS measurements. The photocatalytic efficiency of nanostructured oxides of a selected mean particle size (6 nm) was compared with that obtained with commercial ZnO purchased by Aldrich (d 200 nm) under the same experimental conditions. A few significant parameters, such as the pH, the catalyst surface status as well as the dye chemical structure, were considered in the evaluation of the nanocatalyst performances. Finally, several by-products were identified by HPLC-MS analysis, which allowed to depict a possible mechanism for the dye degradation. Overall, the results suggest that the use of surfactantcapped ZnO nanocrystals allows to extend the performances of traditional ZnO-based photocatalysts, because the surface organic coating provides the oxide with resistance to
photocorrosion and pH variation, as compared to the unpassivated material. Photocatalysis with nanostructured ZnO can actually become a versatile alternative to that assisted by TiO2, provided that a suitable protection of the catalyst surface is attained.
2. Experimental 2.1. Materials ZnO (mean diameter 200 nm, surface area 5 m2 g1), Methyl Red (Acid Red 2 – C.I.: 13020, or MeRed), Methyl Orange (Acid Orange 52 – C.I.: 13025, or MeOr), anhydrous zinc acetate (C4H6O4Zn or ZnAc2, 99.99%), zinc acetate dihydrate (C4H6O4Zn2H2O or ZnAc22H2O, 99.999%), and t-butylphosphonic acid (C4H9PO3H2 or TBPA, 98%) were purchased from Aldrich. n-Hexadecylamine (C16H33NH2 or HDA, 98%) and lithium hydroxide monohydrate (LiOHH2O, 98%) were purchased from Fluka Biochemika. All solvents were of the highest purity available and purchased from Aldrich. All aqueous solutions were made by using twice distilled water. 2.2. Preparation of the catalysts Two different methods have been used for the synthesis of ZnO nanocrystals. 2.2.1. Non-hydrolytic synthesis of ZnO nanocrystals ZnO nanocrystals were synthesized by thermal decomposition of ZnAc2 in hot TBPA/HDA mixture, by using a previously established method [31]. All manipulations were performed using standard airless techniques. Briefly, a mixture of ZnAc2, HDA and TBPA was degassed under vacuum for 1 h at 110 8C under vigorous stirring. The reaction vessel was then slowly heated up to high temperature (200–300 8C, depending on the TBPA content) under nitrogen flow at a rate of 10 8C/min to induce the decomposition of ZnAc2. The nanocrystal size was modulated by varying the TBPA:ZnAc2 molar ratio. The extraction procedures were subsequently performed in air: the ZnO nanocrystals were readily precipitated upon addition of ethanol to the reaction mixture at 50 8C before solidification took place. The resulting precipitate was isolated by centrifugation and was washed twice with ethanol to remove residual surfactants. The surfactant (i.e. HDA + TBPA)-coated ZnO nanoparticles were then easily re-dissolved in CHCl3 to give optically clear solutions. For the purpose of the present paper, this material will be referred to as nh-ZnO. 2.2.2. Hydrolytic synthesis of ZnO nanocrystals The hydrolytic synthesis of ZnO nanocrystals was performed as reported elsewhere [32]. Briefly, an organometallic Zn precursor (identified as Zn4O(CH3COO)6
R. Comparelli et al. / Applied Catalysis B: Environmental 60 (2005) 1–11
[33–35]) was prepared by distilling 0.5 L of a 0.1-M ZnAc22H2O ethanolic solution at 80 8C for 180 min under ambient moisture protection. At the end of this procedure, 0.3 L of condensate and 0.2 L of hygroscopic Zncontaining reaction mixture were obtained. The latter was placed into a flask and diluted with ethanol to yield a 0.1-M Zn precursor solution. Different dilution ratios provided the means of varying the mean particle size of the final ZnO nanoparticles. Next, LiOH powder was added to form a 0.14 M base solution, while keeping the reaction bath temperature at 0 8C. The resulting suspension was placed under air into an ultrasonic bath to accelerate the release of OH ions from the weakly soluble LiOH, resulting in the hydrolysis of the Zn precursor to form a stable ethanolic solution of ZnO clusters. This process was allowed to continue for 10 min, after which ethanol was gently evaporated by a rotavapor. The resulting powder was washed twice with ethanol/water to remove unreacted ZnAc2 and LiOH residuals, and it was finally redispersed in ethanol. The acetate-stabilized ZnO nanocrystals will be referred to as h-ZnO.
2.3. Characterization of the catalysts UV–Vis absorption spectra of ZnO nanocrystal solutions were recorded with a UV–Vis-near IR Cary 5 (Varian) spectrophotometer. Powder X-ray diffraction (XRD) patterns were collected with a Philips PW1729 diffractometer in a conventional u–2u reflection geometry using filtered Cu Ka radiation ˚ ). For XRD measurements the nanocrystal (l = 1.54056 A powder was placed on an Al sample holder. Transmission electron microscopy (TEM) images were obtained using Philips EM 430 microscope (TEM) operating at 300 kV. The samples for the analysis were prepared by dropping dilute solutions of ZnO nanocrystals onto 400mesh carbon-coated copper grids and leaving the solvent to dry. The samples were stable under the electron beam and did not degrade within the typical observation times. FT-IR spectra of ZnO powders were collected with a Perkin-Elmer Spectrum GX FT-IR Spectrometer with a resolution of 4 cm1. Measurements were performed with pressed pellets which were made using KBr powder as diluent.
3
2.4. Photocatalytic degradations For the photocatalytic experiments, the ZnO nanocrystals were freshly prepared and the obtained mean particle size and size dispersion were checked by UV–Vis absorption, XRD and TEM measurements before use. A concentrated CHCl3 solution of nh-ZnO nanocrystals, or an ethanolic solution of h-ZnO, or an ethanolic suspension of nanocrystalline ZnO powders from Aldrich were cast onto a quartz slide, and the solvent was left to evaporate at room temperature. Typically, 0.1 mmol of ZnO was spread over a support area of 3.6 cm2. After deposition, the resulting film was thermally treated at 150 8C for 20 min to improve catalyst adhesion to the substrate. The transparent support was of a suitable shape in order to fit a 1 cm 1 cm quartz cell and positioned against the cuvette wall which was further with respect to the light beam. The radiating source was a medium pressure 200 W mercury lamp (l > 250 nm). The system was arranged in a suitable geometry in order to monitor in situ the reaction course by UV–Vis spectroscopy. All experiments were performed under ambient atmosphere keeping the system under vigorous stirring. MeRed and MeOr, two azo dyes, were chosen as the target compounds. The initial concentration of the dyes (the respective pKa values are reported in Table 1) was typically 3 105 M. The desired pH was obtained by adding the proper amount of 0.1 M HCl or NaOH. Blank experiments, performed irradiating solution containing only the dye, indicated negligible degradation in the absence of the catalyst. Furthermore, in the presence of catalysts, no changes in the dye concentration were detected if the system was kept in the dark. 2.5. UV–Vis and HPLC-MS analysis Dye decolouration was monitored by using an Ocean Optics UV–Vis diode array spectrophotometer equipped with an optical fibre and a deuterium lamp. The determination of the concentration of the dyes and the identification of their respective by-products were performed by HPLC-UVMS using a Varian 9012 chromatographic system equipped with a Ultimate UV detector (LC-Packing Dionex), set at 425 and 220 nm, interfaced to a QSTAR hybrid Qq-TOF mass spectrometer (Applied Biosystem/MSD Sciex, Canada) equipped with a turbo ion spray interface. Samples,
Table 1 Chemical structures of the dyes employed as the target substrates in the ZnO-assisted photocatalysis Organic dye
Chemical structure
Catalyst-interacting functionality
pKa
MeRed
–COOH
5.3
MeOr
–SO3
3.8
4
R. Comparelli et al. / Applied Catalysis B: Environmental 60 (2005) 1–11
injected by a Gilson 234 autosampler equipped with a 9010 Rheodyne valve and a 40 ml loop, were eluted at 0.4 mL/min through a Luna phenyl-hexyl column (3 mm, 150 mm 3 mm) and pre-column (Phenomenex) with the following gradient: from 5/25/70 (ammonium acetate 50 mM in methanol/methanol/water) to 5/75/20 in 10 min, which was then maintained for 5 min. The interface conditions were, for the positive and the negative ion mode, respectively, as follows: nebulizer gas (air) = 1.2 L/min, curtain gas (nitrogen) = 1 L/min, turboionspray gas (nitrogen at 350 8C) = 6 L/min, needle voltage = 5000 and 4400 V, orifice declustering potential = +40 and 40 V and focusing potential = 150 and 120 V. The flow from the HPLC-UV was split to allow 200 ml/min to enter the turbo ion spray interface. The determination of dissolved Zn2+ concentration was performed by inductively coupled plasma (ICP) analysis using an Optima 3000 instrumentation (Perkin-Elmer).
3. Results and discussion The preparation of the ZnO nanocrystal powders employed in this work was carried out by means of two different routes. The nh-ZnO was obtained by thermal decomposition of ZnAc2 in anhydrous and oxygen-free surfactants (i.e. HDA and TBPA) [31]. As opposed, the lowtemperature hydrolysis reaction of an organometallic Zn precursor [32], provided acetate-capped h-ZnO nanocrystals [36,37]. 3.1. Structural and surface characterization of the catalysts In Fig. 1, the typical powder X-ray diffraction (XRD) pattern of ZnO nanocrystals used in the present photocatalytic experiments is reported. Irrespective of the synthetic method, a wurtzite structure can be assigned to all the samples by comparing the diffraction peak positions
Fig. 1. Typical powder XRD pattern of organic-capped ZnO nanocrystals. The mean particle size was 6 nm as calculated by the Debye–Sherrer formula. In the bar diagram, the standard pattern of bulk ZnO wurtzite is shown.
with those reported in the International Crystallographic Data Table. The characteristic line broadening of the diffraction peaks points to nanosized crystalline domains. The average grain size, estimated to be about 6.0 nm by applying the Debye–Sherrer formula to the (1 1 0) reflex, was in good agreement with the mean particle size measured by transmission electron microscopy as well as with the value extrapolated from the size-dependent absorption onset of quantum-sized ZnO solutions (quantum confinement effects can be observed in nanocrystals with diameter 200 nma polydisperse 6 nmb (size distribution < 10%) 6 nmb (size distribution < 10%)
None Acetate groups TBPA HDA
Higha Highc Negligiblec
Wurtzite Wurtzite Wurtzite
a b c
From Sigma-Aldrich. From TEM and XRD. [31,32].
ZnO Aldrich and for nh-ZnO the abatement remained at the level of 67 and 23%, respectively. This discrepancy can be explained by considering that, whereas the optical absorption comprises the contribution from both the residual parent dye molecules and the by-products that maintain conjugation through the azo bond, the relative decay of the quasimolecular ion reflects the fate of the dye only [30] (see Section 3.3). In the degradation of MeOr (Fig. 5A and B), for the three catalysts the same level of photocatalytic efficiency was inferred from both the absorption and the MS data. At the end of the experiments, the percentage of dye removal was 95% for h-ZnO, 89% for ZnO Aldrich and 38% for nh-ZnO. The general behaviour can be accounted for by considering three factors: (1) the nature of surface coating and surface active area left available for catalysis; (2) the surface-to-volume ratio; (3) the density of surface –OH groups, related to nature of the reaction involved in the synthesis of the material. These characteristics are summarized in Table 2. As discussed above, the presence of –OH groups is directly related to the local production of hydroxyl radicals. These functionalities also provide sites for adsorption of the substrates in addition to unsaturated surface metal atoms. Furthermore, nanosized oxides are characterized by a higher surface-to-volume ratio with respect to ZnO Aldrich, however both nh-ZnO and h-ZnO are surface-coordinated by ligand molecules which occupy the catalyst surface and reduce the active area available for adsorption and catalysis owing to both ligand passivation of metal sites and ligand steric hindrance. As suggested by IR measurements, the phosphonic acids likely bind to the oxide in a bi- and/or tridentate anchoring geometry [38,39], as opposed to the acetate ligands which can be assumed to establish mainly a weak unidentate attachment (see Section 3.1). It follows that a higher number of surface sites can be expected to be passivated by TBPA, as compared to the case of the acetatecapped ZnO. Moreover, the long alkyl chain of HDA and the bulky t-butyl group of TBPA on nh-ZnO can actually oppose a significant barrier against the adsorption of the target molecules. This characteristics can account for the very different photocatalytic efficiency recorded for the two nanocrystalline catalysts, despite their mean particle size (and thus, the surface-to-volume ratio) were comparable. By similar arguments, the higher photocatalytic efficiency of
ZnO Aldrich with respect to nh-ZnO takes benefit from the organic-free surface, which is naturally available for absorption and catalysis. 3.2.3. Molecular structure of dye In Table 3, the percentages of dye removal after a fixed irradiation period for reactions carried out at pH 6 results are summarized. As a general trend, the experiments revealed that, irrespective of the catalyst nature, the efficiency in the removal of the two dyes was comparable. Bauer et al. [64] have demonstrated that the SO3 groups are weakly anchored to the surface of ZnO by means of a single oxygen atom, in a way analogous to that adopted by unidentate carboxylate complexes. Thus, if we reasonably assume that the carboxylic and the sulphonic moieties of MeRed and MeOr, respectively, bind to the ZnO surface with similar strength, the slightly different extent of dye removal could be explained on the basis of an intrinsically different reactivity of the two compounds. The presence of –COOH group in ortho position with respect to the azo bond could increase the susceptibility of MeRed to the attack of hydroxyl radicals due to positive electronic effects. Additionally, it can be suggested [65,66] that loss of planarity of the MeRed molecule, sterically induced by the proximity between the –COOH and the azo bond, may reduce the N N p overlap, in turn benefiting the dye reactivity toward hydroxyl radicals. 3.3. By-products formation The dye by-products were identified by HPLC-MS analysis of the reaction mixtures along the course of irradiation. In Tables 4 and 5, the chemical structures of several by-products originating from the degradation of MeOr and MeRed, respectively, are reported. Similar to Table 3 Percentages of dye removal ([dye]t=0 = 3 105 M) measured at pH 6 after 145 min of illumination Catalyst
MeRed (%)
MeOr (%)
h-ZnO nh-ZnO ZnO Aldrich
100 54 93
95 37 89
The values were estimated from the relative decay of the parent quasimolecular ion peak by MS.
8
R. Comparelli et al. / Applied Catalysis B: Environmental 60 (2005) 1–11
Table 4 Proposed chemical structures of several identified by-products during MeRed degradation By-product
Chemical structure
Molecular weight
h-ZnO
nh-ZnO
ZnO Aldrich
1
255
X
X
X
2
241
X
X
3
271
X
X
X
4
287
X
X
X
5
285
X
X
X
what previously found for dye degradation photocatalyzed by organic-capped TiO2 nanocrystals [30], two different degradative routes can be operative under the present reaction conditions. The first mechanism involves the homolytic rupture of the nitrogen–carbon bond of the aminic group giving rise to the substitution of methyl group
with a hydrogen atom. Such a photolytic route can occur consecutively, thus justifying the detection of by-products 1 and 2 for both dyes. The second mechanism is based on the aromatic ring substitution by one or more hydroxyl groups upon attack of hydroxyl radicals. This is supported by the identification of by-product 4 for MeOr and of by-product 5
Table 5 Proposed chemical structures of several identified by-products during MeOr degradation By-product
Chemical structure
Molecular weight
h-ZnO
nh-ZnO
ZnO Aldrich
1
291
X
X
X
2
277
X
X
X
3
227
X
X
X
4
307
X
X
X
5
321
X
X
X
6
242
X
X
X
7
151
X
X
R. Comparelli et al. / Applied Catalysis B: Environmental 60 (2005) 1–11
for MeRed, whose molecular weights are consistent with the insertion of one oxygen atom. For both dyes, it is likely that the OH substitution initially occurs on the benzene ring carrying the dimethylamino group due to its capability to stabilize the intermediate hydroxy-benzene radical, in contrast to the other benzene ring which, in fact, carries either a –COOH (MeRed) or –SO3H (MeOr). In addition, the identification of by-products 3 and 6 from MeOr demonstrates that the ipso-substitution by a hydroxyl radical can also take place at the carbon position which carries the sulphonic moiety. In addition, the presence of by-products 3 and 4 for MeRed and of by-product 5 for MeOr, deriving from both hydroxyl substitution and photolysis, suggests that the two mechanisms are independently active. In Fig. 6, the formation–decay profiles of MeRed byproduct 1 and MeOr by-product 4 are reported as representative examples of by-product temporal evolution during dye degradation. The trends reported in this figure show that the by-products formation was maximum at intermediate reaction times (30–90 min) during the treatment with h-ZnO and ZnO Aldrich, whereas by-products formation was retarded in the presence of nh-ZnO. Also, these findings are in agreement with the plots of dye decay in Figs. 4 and 5 showing that the removal rate follows the order: h-ZnO > ZnO Aldrich > nh-ZnO and h-ZnO ZnO Aldrich > nh-ZnO for MeRed and MeOr, respectively. It is important to stress that, due to the lack of standards for the identified by-products, the reported trends are just semiquantitative. However, as the structures of the identified by-products are similar to that of the parent dye, it is reasonable to assume that their MS responses are similar too.
9
Fig. 6. Evolution of (A) MeRed by-product 1 and (B) MeOr by-product 4 during photocatalysis with ZnO-based catalysts.
In addition, at longer reaction times during the degradation with h-ZnO and ZnO Aldrich, it was found that the identified by-products represented just a minor fraction (
