Dye decolorization test for the activity assessment of visible light photocatalysts: Realities and limitations

Catalysis Today 224 (2014) 21–28

Contents lists available at ScienceDirect

Catalysis Today journal homepage: www.elsevier.com/locate/cattod

Dye decolorization test for the activity assessment of visible light photocatalysts: Realities and limitations Sugyeong Bae a , Sujeong Kim a , Seockheon Lee b , Wonyong Choi a,∗ a School of Environmental Science and Engineering/Dept. of Chemical Engineering, Pohang University of Science and Technology (POSTECH), Pohang 790-784, Republic of Korea b Center for Water Resource Cycle Research, Korea Institute of Science and Technology (KIST) , Seoul 136-791, Republic of Korea

a r t i c l e

i n f o

Article history: Received 24 September 2013 Received in revised form 29 November 2013 Accepted 9 December 2013 Available online 10 January 2014 Keywords: Photocatalytic activity test Visible light photocatalyst Dye degradation Environmental photocatalyst Dye sensitization

a b s t r a c t The development of photocatalysts with visible light activity has been extensively investigated. Their activities are usually tested by measuring the degradation rate of different organic compounds. Among these organic substrates, dyes are the most widely employed due to their rapid decolorization and simple kinetic analysis using a spectrophotometric method. However, the dye test has much uncertainty in the evaluation of photocatalytic activity. To assess the validity of the dye test, six visible-light photocatalysts (N–TiO2 , C–TiO2 , C60 (OH)x /TiO2 , Pt/WO3 , BaBiO3 , and Bi2 WO3 ) were tested and compared for the degradation of five organic dyes (anionic: acid orange 7, indigo carmine, and new coccine; cationic: methylene blue and rhodamine B) in this study. This study aimed to assess how the measured activities depend on the kind of test dyes and how reliable the dye test is as an activity evaluation method. The activities determined by the dye test were highly specific to the kind of dye and photocatalyst. For example, N–TiO2 is the most active photocatalyst for the degradation of acid orange 7 at pH 3 but is one of the least active at pH 9; Pt/WO3 is the best photocatalyst for the degradation of methylene blue but not much active for the degradation of acid orange 7. This is ascribed to the fact that the dye test is significantly influenced by various factors such as the dye sensitization of catalyst particles, the absorption spectral overlap between dyes and photocatalysts in the visible region, the electrostatic interaction (attractive or repulsive), and the properties of dye degradation intermediates. In general, the dye decolorization efficiency was poorly correlated with the dye mineralization efficiency, which limits the practical value of the dye test. Therefore, the practice of dye test for the activity assessment of visible light photocatalysts should not be recommended and the activity results obtained for a specific combination of a dye and a photocatalyst should not be generalized. © 2013 Elsevier B.V. All rights reserved.

1. Introduction Photocatalysts have been widely investigated for various energy and environmental applications including hydrogen production [1], degradation of organic pollutants [2], bacterial disinfection [3,4], and CO2 reduction [5]. In particular, the photocatalytic remediation of contaminated water and air has been extensively investigated to demonstrate its viability as a useful cleanup process. Various contaminants such as chlorinated aromatics [6], chlorinated hydrocarbons [7], heavy metal ions [8,9], and volatile organic compounds [10,11] can be degraded or transformed by photocatalysis. Among numerous kinds of organic compounds, dyes are the most tested substrates in photocatalytic studies because not only they are common industrial pollutants [12], but also their

∗ Corresponding author. Tel.: +82 54 279 2283; fax: +82 54 279 8299. E-mail address: [email protected] (W. Choi). 0920-5861/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.cattod.2013.12.019

degradation can be simply monitored by a colorimetric method. Dye discoloration test is now widely used as a de facto standard method of photocatalytic activity assessment although the concerns about the dye test problems have been repeatedly raised [13,14]. The research publications on photocatalysis that used the dye decolorization as a test method have rapidly increased for the last decade (see Fig. 1). In particular, the dye test is being widely used as an activity test method for visible-light active photocatalysts. Table 1 shows some recent examples that used the dye test in the studies of visible light photocatalysis: the dye test studies are classified according to the kind of photocatalytic materials such as doped TiO2 [15–27], composite semiconductor [28–31], and binary metal oxide [32–39]. The dye decolorization as an activity test method for visiblelight active photocatalysts suffers from many problems such as: (1) Dye itself absorbs visible light and its degradation can be initiated from the excited dye (e.g., direct photolysis [40], dye sensitization [41,42]), not the excited photocatalyst. The decolorization result

22

S. Bae et al. / Catalysis Today 224 (2014) 21–28

Table 1 Literature study examples of photocatalyst activity test using dye decolorization under visible light. Photocatalyst type Doped semiconductor

N–TiO2 C–TiO2

S–TiO2 Composite semiconductor

CdS/TiO2 WO3 /TiO2

Binary metal oxide semiconductor

Bi2 WO6

Bi2 VO4

obtained via the excited dye does not represent the intrinsic photocatalytic activity of the tested materials. In addition, the visible light absorption by the dyes themselves attenuates the incident light flux available to the photocatalyst (i.e., dyes shield the photocatalyst from irradiation), which may underestimate the intrinsic visible light activity of the catalyst material. (2) The decolorization of dye reflects the selective transformation of chromophoric groups only, and not necessarily the full degradation (or mineralization). It has been frequently demonstrated that there is poor correlation between the color removal and TOC (total organic carbon) removal [43,44]. Therefore, the selective degradation of a specific functional group in a dye molecule should lead to the rapid decolorization, but the overall degradation can be inefficient. (3) The decolorization monitored by absorbance measurement at a single wavelength can be inaccurate because the generation of intermediate products is usually accompanied by the spectral change. Therefore, the Beer’s law that relates the absorbance to the dye concentration cannot be applied to such system. (4) Most commercially available dyes are impure (typically 70–90%). They contain many unknown components that may interfere with the photocatalytic degradation of the parent dye, which makes dyes unsuitable as a test substrate. Keeping the above general problems in mind, this study aims to evaluate the validity and limitation of the dye test in a systematic way. We measured the activities for various combinations of five dyes and six visible-light photocatalysts and assessed how the measured activities depend on the kind of test dyes and how

Number of publications

1000

800

600

Type of dye

Ref.

Acid orange 7 (Azo) Methylene blue (xanthene) Acid orange 7 (Azo) Methylene blue (xanthene) Rhodamine B (phenothiazine) Methylene blue (xanthene) Indigo carmine (indigo) Methylene blue (xanthene) Acid orange 7 (Azo) Methylene blue (xanthene) Acid orange 7 (Azo) Methylene blue (xanthene) Rhodamine B (phenothiazine) Methylene blue (xanthene) Rhodamine B (phenothiazine)

15 16–18 19–21 22–24 25 26 27 28, 29 30 31 32 33 34–36 37 38, 39

reliable the dye test is as an objective evaluation method. Despite the wide popularity of dye tests in the field of environmental photocatalysis, this is the first comprehensive study that compared and assessed the validity of various dye tests. 2. Experimental 2.1. Reagents and materials Six visible light active photocatalysts were selected for this study and pure TiO2 (P25) was compared as a control sample (listed in Table 2). The following visible-light photocatalysts were synthesized according to the literature method: nitrogen-doped TiO2 (N–TiO2 ) [45], carbon-doped TiO2 (C–TiO2 ) [46], fullerolanchored TiO2 (C60 (OH)x /TiO2 ) [47] and Pt/WO3 [43], BaBiO3 [48], and Bi2 WO6 [49]. Five dyes that were selected as the substrate for the activity test are acid orange (AO7, Aldrich), new coccine (NC, Aldrich), indigo carmine (IC, Aldrich), methylene blue (MB, Aldrich), and rhodamine B (RhB, Aldrich). Their structures and properties are listed in Table 3. Methyl orange (MO, Aldrich) was also used as a substrate in a separate activity test. 2.2. Photocatalysts characterization All photocatalysts were characterized, and their properties are listed in Table 2. X-ray powder diffraction analysis using Cu K␣ radiation (Marc Science Co. M18XHF) was carried out to measure the crystalline phase. The BET surface area of powder samples was measured using nitrogen as an adsorptive gas. The zeta potentials of the photocatalyst particles suspended in water were measured using an electrophoretic light scattering spectrophotometer (ELS 8000, Otsuka) equipped with a He–Ne laser and a thermostated flat board cell. Diffuse reflectance UV/visible absorption spectra (DRUVS) were recorded using a spectrophotometer (Shimadzu UV2600) with an integrating sphere attachment and BaSO4 was used as the reference.

400

2.3. Photocatalytic activity measurement 200

0

2000

2002

2004

2006

2008

2010

2012

Year Fig. 1. Annual number of papers published in the subject area of “photocatalytic dye degradation”. The literature search was carried out at the Scoupus website (www.scoupus.com) using the key words “photocatal*” and “dye degradation, removal, decolorization, or decoloration”.

Photocatalyst powder was dispersed in distilled water at the concentration of 0.5 g/L. An aliquot of the dye substrate stock solution was subsequently added to the suspension to make a desired substrate concentration (100 ␮M). The pH of the suspension was adjusted to 3 or 9 with HClO4 or NaOH solution, and then the suspension was stirred for 30 min in the dark to reach the adsorption equilibrium of dyes on the photocatalyst surface. A 300 W Xe arc lamp (Oriel) was used as a light source. The light beam was passed through a 10 cm IR water filter and a cutoff filter ( > 420 nm

S. Bae et al. / Catalysis Today 224 (2014) 21–28

23

Table 2 Properties of visible light photocatalysts. Color

N–TiO2

C60 (OH)x /TiO2

C–TiO2

Pt/WO3

BaBiO3

Bi2 WO6

TiO2 (P25)

BET surface (m2 /g) Crystallite Sizea (nm) pHpzc b Ref.

Yellowish 119 13 5.3 45

Brownish 57 22 3.2 47

Brown 247 5.7 5.0 46

Grey 4.7 45 2.0 43

Blackish brown <0.5 28 –c 48

Yellowish <0.5 74 –c 49

White 56 21 6.5 –

a b c

Calculated from Debye–Scherrer equation. Point of zero charge (measured at point of zero zeta potential). Not determined because BaBiO3 and Bi2 WO6 were not well dispersed in solution.

for visible and  > 300 nm for UV irradiation) and focused onto a cylindrical pyrex reactor (30 mL) with a quartz window. The incident photon flux was measured by using an optical power meter (1830-C, Newport) and determined to be about 155 mW/cm2 in the wavelength range of 420–550 nm. The reactor was open to the ambient air to prevent the depletion of dissolved dioxygen, and agitated magnetically during irradiation. Sample aliquots were extracted from the reactor at a periodic time interval during the illumination and filtered through a 0.45 ␮m PTFE syringe filter (Millipore) to remove photocatalyst particles. Multiple measurements of the photocatalytic activity were carried out under the identical reaction condition to confirm the reproducibility.

dyes (MB and RhB). The activity variation across the table shows that the dye test method is highly specific to the kind of dyes and photocatalysts. The most active photocatalyst is different for each dye. For example, the most active photocatalyst is C–TiO2 for AO7, but C–TiO2 and fullerol-TiO2 for NC and IC, respectively. Similarly, the most reactive (degradable) dye is different for each photocatalyst. Therefore, each photocatalyst is the most active at least for one dye except for BaBiO3 and Bi2 WO6 which are generally much lower in the activity because their surface area is too low compared with other photocatalysts. The visible light activities of BaBiO3 and Bi2 WO6 are even lower than those of pure TiO2 (P25) in most cases. The overall trend in the activity variation in Table 4 is hard to be generalized. This indicates that the visible light activity of a specific

2.4. Analysis

3.1. Dye-specific activity test Six visible light active photocatalysts tested in this study are very different among one another in their working mechanisms. All photocatalysts have a significant absorption in visible region between 400–500 nm (see Fig. 2a). The elevated background (460–800 nm) in Pt/WO3 spectrum is due to the presence of Pt, which is similar to the previously measured spectra of Pt/TiO2 6 and Pt/WO3 [43]. N–TiO2 and C–TiO2 have non-metal dopants that have mid-gap energy levels in the band gap region. The photo-induced electronic transition between the mid-gap levels and the conduction (or valence) band is responsible for the visible light activity [45,46]. The visible light activity of fullerol-TiO2 is ascribed to the photoinduced charge transfer between the ground state of fullerol and the conduction band of TiO2 [47]. Pt/WO3 , BaBiO3 , and Bi2 WO6 have smaller band gaps that can be directly excited by visible light absorption. Table 3 shows that dyes can be classified by their structure, charge, and functional group. Five tested dyes are clearly different in their spectral absorption and intensity in the visible region: the maximal absorption band of AO7, NC, and IC are weaker than that of MB and RhB (see Fig. 2b). Various types of dyes were employed to obtain the comprehensive assessment of the dye test for different kinds of visible active photocatalysts. Table 4 summaries the results of the dye tests under visible light irradiation in a matrix form which has seven photocatalysts (including pure TiO2 as a control sample) in the column and five dyes in the row. Most activity tests were carried out at the initial pH 3. As for AO7, MB, and RhB, the tests were done for both pH 3 and pH 9 to check out their pH dependence. There were no significant pH changes during the photodegradation of dyes in all cases. Anionic dyes (AO7, NC, and IC) are generally more degraded than cationic

A. N-TiO2 B. C60(OH)X/TiO2 C. C-TiO2 D. Pt/WO3 E. BaBiO3 F. Bi2WO6 G. TiO2 (P25)

0.8

Abs (K.M.)

3. Results and discussion

1.0

0.6 E

(a)

0.4

A

0.2

F

D B

G 0.0

400

500

C

600

700

800

Wavelength (nm) 1.6 1.4 1.2 1.0

Abs

The removal of dye color was monitored using a UV/visible spectrophotometer (Shimadzu UV-2401PC). The monitored wavelength for each dye is listed in Table 3. The change of the organic carbon content in the irradiated dye solution was monitored using a TOC analyzer (Shimadzu TOC-VSH ).

(b)

E A. AO7 B. NC C. IC D. MB E. RhB

D

0.8 0.6 0.4

B

0.2 0.0

C

A 400

500

600

700

800

Wavelength (nm) Fig. 2. (a) Diffuse reflectance UV/visible spectra of six visible-light active photocatalysts. (b) UV/visible absorption spectra of five dyes employed as the test substrate ([Dye] = 10 ␮M; pH = 3).

24

Table 3 Properties of tested dyes. Ref.

87

485 (2500)

15, 19–21, 30, 32

Azo-anionic

75

507 (3000)

12

C16 H8 N2 Na2 O8 S2 466.3

Indigo-anionic

93

610 (2100)

27

Methylene blue (MB)

C16 H18 ClN3 S 373.9

Phenothiazine-cationic

89

663 (9100)

16–18, 22–24, 26, 28, 29, 31, 33, 37

Rhodamine B (RhB)

C28 H11 N2 NaO4 S 479.0

Xanthene-zwitter ionic

80

554 (13,500)

25, 34–36, 38, 39

Molecular formula/weight

Classification

Purity (dye content %)

Acid orange 7 (AO7)

C16 H11 N2 NaO4 S 350.3

Azo-anionic

New coccine (NC)

C20 H14 N2 O10 S3 604.5

Indigo carmine (IC)

Structure

S. Bae et al. / Catalysis Today 224 (2014) 21–28

max (nm) (εmax , M-1 cm-1 )

Name of dye

S. Bae et al. / Catalysis Today 224 (2014) 21–28

25

Table 4 Visible light activities of seven photocatalysts measured with various dyesa . Dye

No catalyst (photolysis control)

N–TiO2

C60 (OH)x /TiO2

C–TiO2

Pt/WO3

BaBiO3

Bi2 WO6

TiO2 (P25)

AO7 AO7 (pH 9) NC IC MB MB (pH 9) RhB RhB (pH 9)

3(±2) 4(±6) 2(±3) 2(±1) 3(±3) 6(±3) 6(±1) 2(±2)

84(±2) 18(±4) 29(±3) 46(±2) 4(±1) 5(±2) 19(±0) 19(±1)

80(±2) 47(±4) 32(±4) 99(±1) 13(±2) 23(±5) 47(±1) 31(±5)

97(±6) 94(±4) 54(±6) 89(±3) 21(±0) 32(±1) 39(±1) 36(±2)

49(±3) 26(±2) 18(±3) 87(±3) 29(±1) 32(±4) 33(±5) 35(±1)

21(±1) 4(±0) 4(±1) 36(±5) 6(±2) 27(±4) 5(±3) 4(±1)

21(±3) 5(±4) 1(±1) 22(±2) 2(±0) 25(±7) 6(±3) 4(±1)

60(±1) 20(±4) 10(±1) 64(±6) 4(±1) 11(±1) 25(±2) 8(±0)

a The listed numbers represent percentage (%) of dye removal (100х[D]/[D]0 after 2 h irradiation of visible light ( > 420 nm) at pH 3 (or pH 9 when indicated); [D] represents the concentration of dye). The most active photocatalysts for a specific dye are indicated in bold numbers.

even less active than pure TiO2 (P25) that does not absorb any visible light. 3.3. Dye sensitization effect in photocatalysis It is well known that dyes can be degraded through the sensitization pathway on the surface of semiconductor as long as the excited dye has the energy level higher than the conduction band edge. This pathway does not require the excitation of the band gap and enables the visible light-induced degradation of dyes on

100

80

N-TiO2

C60(OH)X/TiO2 60 Pt/WO3

TiO2 (P25)

40 Bi2WO6 20

BaBiO3 0

50

100

150

200

250

300

2

BET surface area (m /g)

3.2. Correlation between surface area and visible light activity for dye degradation

40

(b) MB 30

[D]/[D] x 100 (%) 0

The specific surface area (SSA) is one of the most critical parameters in determining the catalytic activity. SSA also influences the photocatalytic activities but the relationship between SSA and the photocatalytic activity is weak in many case [50]. Fig. 3 shows the correlation between SSA and the dye degradation activity for AO7 and MB (at pH 3). Although AO7 shows some correlation, MB does no correlation between SSA and the activity. As for the degradation of MB, Pt/WO3 shows the highest activity despite its lowest SSA. This should be ascribed to the negative surface charge of Pt/WO3 at pH 3 (pHpzc ∼2) whereas other photocatalysts have the positive surface charge at pH 3. The cationic MB is attracted electrostatically onto the negatively charged surface of Pt/WO3 , which should accelerate the degradation of MB. As a result, Pt/WO3 is the best photocatalyst for the degradation of MB at both pH 3 and 9. On the other hand, Pt/WO3 is the least active photocatalyst for the degradation of AO7 (at pH 3) because of the electrostatic repulsion between the anionic dye and the negative surface charge. Incidentally, BaBiO3 and Bi2 WO6 that have extremely low SSA show much lower activity than other photocatalysts: in most cases they are

(a) AO7 C-TiO2

[D]/[D] x 100 (%) 0

photocatalyst that is assessed with a specific dye cannot be generalized to other dyes. This also clearly confirms that the single dye test cannot represent the overall activity of a visible active photocatalyst. This general behavior is very similar to what we reported for the substrate-specific photocatalytic activities that were measured for various commercial TiO2 samples under UV irradiation [50]. The effect of pH is important in controlling the photocatalytic activity because pH controls the surface charge and the adsorption of charged substrates. Since most commercial dyes are charged molecules, the pH effect should be important in the dye test as well. The catalyst surface takes positive charge below pHpzc (point of zero charge) and negative charge above pHpzc . Therefore, anionic substrates are more adsorbed and degraded faster at pH < pHpzc whereas the adsorption and degradation of cationic substrates are much favored at pH > pHpzc because of the electrostatic interaction between the substrate and catalyst surface. The pH effects on the degradation of charged dye substrates were investigated by comparing the degradation of AO7 (anionic), MB (cationic), and RhB (zwitter ionic) between pH 3 and pH 9. For most photocatalysts tested in this study, the surface charge is positive at pH 3 and negative at pH 9. As a result of the electrostatic interaction, the degradation of anionic AO7 is clearly favored at pH 3 than at pH 9 for all tested catalysts (see Table 4). On the other hand, cationic MB is degraded faster at pH 9. The zwitter ionic RhB does not show clear preference to acidic or basic condition. Therefore, the most active photocatalyst for a given dye is also dependent on pH. For example, the best photocatalyst for AO7 is N–TiO2 at pH 3 but N–TiO2 is one of the least active photocatalysts at pH 9.

Pt/WO3

C-TiO2

20 C60(OH)X/TiO2 10

0

BaBiO3 Bi2WO6 0

TiO2 (P25)

50

N-TiO2

100

150

200

250

300

2

BET surface area (m /g) Fig. 3. The correlation between the dye degradation activity for (a) AO7 and (b) MB and the BET surface area of photocatalysts under visible light illumination. ([Catalyst] = 0.5 g/L; [Dye]0 = 100 ␮M; pHi = 3;  > 420 nm).

S. Bae et al. / Catalysis Today 224 (2014) 21–28

3.4. Dye decolorization versus mineralization The decolorization of a dye indicates the destruction of the chromophore group, but not necessarily the full degradation (mineralization). To examine how the degree of decolorization is correlated with that of mineralization, the photocatalytic degradation of AO7 was tested with various photocatalysts under both visible light and UV light irradiation and compared for not only the color removal but also the TOC removal. Since the mineralization was much less under visible light, the visible light irradiation was extended to 8 h whereas UV light was irradiated for 2 h. Fig. 4 summarizes the results. The extent of mineralization under visible light is generally much lower than that under UV irradiation despite the longer irradiation time of visible light (Fig. 4a). This is mainly because the visible light photocatalysis generates less energetic electrons and holes and consequently the photocatalytic reaction is thermodynamically and kinetically limited [14]. The band edge positions in semiconductors determine the redox potentials of electrons in CB and holes in VB. Although the decolorization of dyes can be achieved by the selective transformation of chromophoric groups by less energetic electrons/holes, the mineralization of dyes requires the holes with highly positive potential (or OH radicals) which are generated preferably under UV irradiation. As for TiO2 -based visible light photocatalysts (N–TiO2 , C–TiO2 , and fullerol-TiO2 ), the intrinsic bandgap (∼3 eV) cannot be excited under visible light while the visible light activation proceeds without generating the VB holes (or OH radicals). The less energetic holes generated in the sub-bandgap states under visible light can be effective in destructing chromophoric groups but further degradation to CO2 (i.e., mineralization) cannot be achieved. The color removal efficiencies of the titania-based catalysts are all similar between the visible and UV irradiation conditions but the TOC

120

(a) for AO7

[TOC]/[TOC] x 100 (%)

100

TiO2 (P25)

UV light (2 h) 80

N-TiO2

0

Visible light (8 h)

C60(OH)x/TiO2

60

N-TiO2

40

C60(OH)x/TiO2

20 0

BaBiO3 BaBiO3 w/o cat. 0

Pt/WO3

w/o cat.

Bi2WO6 20

TiO2 (P25)

Bi2WO6

40

60

80

100

120

[D]/[D] x 100 (%) 0

60

(b)

50

0

wide bandgap semiconductors like TiO2 of which bandgap cannot be excited under visible light. On the visible light-irradiated photocatalysts, the discoloration of dyes can be contributed by both the dye sensitization and the bandgap excited photocatalysis. Assessing the contribution from each path is not an easy task. As shown in Table 4, the pure TiO2 (P25) was also tested as a control catalyst to estimate the sensitization effect under visible light irradiation. All dyes could be degraded on pure TiO2 under visible light although the degradation efficiency highly varies from dye to dye. AO7, IC, and RhB were significantly degraded while MB was little degraded. This is ascribed to the weak thermodynamic driving force for the electron injection from excited MB to TiO2 CB as well as the electrostatic repulsion between the cationic dye (MB) and the positively charged TiO2 surface under acidic condition [51]. Since MB exhibited particularly low sensitization effect at both acidic and basic conditions, MB should be the most appropriate among the tested dyes in assessing the intrinsic visible light activity of photocatalysts. The presence of the dye sensitization effect should generate an intrinsic error in evaluating the photocatalytic activity. The color removal kinetics, which is related with the destruction of the chromophore group, can be further complicated by the concurrent side processes such as N-deethylation (in RhB degradation) and Ndemethylation (in MB degradation) [52], which is discussed in the later part. A previous study claimed that the degradation of MB by absorption spectrum analysis may cause misunderstanding of genuine visible light activity because of the dye sensitization effect. Moreover, dyes cannot absorb visible light after the destruction of the chromophore structure and the resulting degradation intermediates may not be further degraded due to the limited oxidative power of visible active photocatalyst [53]. On the basis of the above arguments, it is apparent that the activity assessed through measuring the dye decolorization efficiency may not represent the genuine photocatalytic activity.

[TOC]/[TOC] x 100 (%)

26

AO7

for N-TiO2 40

for Pt/WO3

30

RhB

20

MB

10

NC

MB 0 20

NC 40

IC

RhB

60

AO7 80

IC 100

120

[D]/[D] x 100 (%) 0

Fig. 4. (a) The correlation between the decolorization of AO7 (color removal) and the mineralization (TOC removal) with various visible light photocatalysts under UV ( > 300 nm) irradiation for 2 h and visible light ( > 420 nm) irradiation for 8 h. (b) The correlation between the decolorization of test dyes and the mineralization with N–TiO2 and Pt/WO3 under visible light ( > 420 nm) irradiation for 8 h ([Catalyst] = 0.5 g/L; [Dye]0 = 100 ␮M; pHi = 3).

removal efficiencies are much higher under UV irradiation. On the other hand, the TOC removal activities of narrow-bandgap semiconductors (BaBiO3 , and Bi2 WO6 ) are negligibly small under not only visible light but also UV irradiation although, the decolorization efficiencies vary depending on the kind of semiconductors. On UV-irradiated semiconductors with narrow bandgaps, the hot electrons and holes are generated but rapidly thermalized to the band edge energy, which wastes the excess photon energy. Therefore, the activity difference between visible and UV irradiation conditions is not significant for narrow-bandgap semiconductors. In a similar way to Fig. 4a, the photocatalytic degradation of different dyes was tested with N–TiO2 under visible light irradiation to compare the color removal with the TOC removal. The near complete decolorization of RhB and IC under visible light was accompanied by less degree of TOC removal. NC and MB were also hardly mineralized. N–TiO2 exhibited the highest mineralization efficiency for AO7 but it did not exceed 50% even at the point of near complete decolorization. Since N–TiO2 is unable to generate OH radicals under visible light illumination, it has restricted oxidation power [54]. In general, the level of photocatalytic mineralization

S. Bae et al. / Catalysis Today 224 (2014) 21–28

27

Initial Rate (umol / L min)

0.8

AO7 IC

0.6

0.4

0.2

0.0

0

50

100

150

200

250

300

[Dye]0 (uM) Fig. 5. Effect of the initial concentration of dye on its decolorization rate in the visible light-irradiated suspension of N–TiO2 . ([Catalyst] = 0.5 g/L; pHi = 3;  > 420 nm).

of dye is affected by not only the oxidizing power of the photocatalyst, but also the dye properties (e.g., molecular charge, the kind of functional groups, the kind of dye degradation intermediates). Therefore, the photocatalytic activity measured with a specific dye cannot be generalized to different dyes as the data in Table 4 clearly show. 3.5. Dye concentration effect The degradation rates of dyes depend on their concentration and this dye concentration effect can be also complex. The initial degradation rate of dye may increase with increasing the dye concentration because the concentration of dyes adsorbed on the photocatalyst surface is enhanced at higher concentration. However, when both the substrate (dye) and the photocatalyst can absorb the visible light in the same wavelength region, the photocatalytic degradation kinetics may be retarded because excess dye molecules in the solution absorb visible light photons and attenuate the light flux incident onto the catalyst surface [55]. To investigate this effect, AO7 and N–TiO2 , both of which absorb mainly in the wavelength region of 400–500 nm (see Fig. 2), were selected and the dye degradation rate was compared with varying the dye concentration (10–300 ␮M). The degradation of IC with N–TiO2 was also compared as a control case since IC has much lower absorption spectral overlap with N–TiO2 (see Fig. 2). Fig. 5 shows that the dye degradation rates as a function of the dye concentration are very different between AO7 and IC. Although AO7 has a high degree of absorption spectral overlap with N–TiO2 , its degradation rate monotonously increases with increasing [dye] up to 300 ␮M. On the other hand, the degradation rate of IC was retarded at higher concentration although IC has much lower spectral overlap with N–TiO2 than AO7. The results imply that the light shielding effect by dye is not straightforward. The dye-induced light shielding effect might be counterbalanced by the dye sensitization effect that should be higher at higher dye concentrations, which complicates the practice of the dye test. 3.6. Formation of intermediates and absorption peak shift Most tests of photocatalytic decolorization of dyes are carried out by monitoring the absorbance at a specific wavelength. The absorbance is directly converted to the dye concentration on the basis of Beer’s law (A = εbC), which is valid only if dye degradation

Fig. 6. UV/visible absorption spectral change of MO in the presence of (a) C–TiO2 (b) Pt/WO3 ([Catalyst] = 0.5 g/L; [Dye]0 = 100 ␮M; pHi = 3;  > 420 nm).

intermediates do not have any interfering absorbance in the measured wavelength. When a degraded intermediate has a visible absorption band which is different from that of the parent dye, the spectral shift is accompanied in the course of the dye degradation. For example, the hypsochromic shift (blue shift) by Ndemethylation of MB and N-deethylation of RhB has been observed, which indicates that N-demethyl/deethylation concurs with the cleavage of RhB chromophore [56,57]. As an example, Fig. 6 compares the absorption spectral shift during the visible light-induced degradation of methyl orange (MO) with C-TiO2 and Pt/WO3 . MO has both amine (-N(CH3 )2 ) and sulfonate (-SO3 - ) groups and its photocatalytic degradation can be accompanied by the absorption peak shift, which is induced by either the demethylation in the dimethylamino group or the hydroxylation of benzene rings in MO [58]. In particular, the generation of demethylated products induces a marked blue-shift [58]. The photocatalytic degradation of MO on C–TiO2 exhibited a significant spectral shift (Fig. 6a) whereas the spectral shift with Pt/WO3 is insignificant (Fig. 6b). It is interesting to note that the degradation of MO on C–TiO2 slightly increased the absorbance in the initial stage of the degradation, which implies that colored intermediates are momentarily generated from the photocatalytic degradation of MO [56,58]. Such intermediates do not seem to be generated with Pt/WO3 . Therefore, in the case like MO/C–TiO2 system, the absorbance monitoring at a single wavelength cannot provide accurate data about the dye degradation kinetics. This example clearly demonstrates that the dye degradation mechanism can be different depending on the kind of photocatalysts. In such case, the time-profiles of the absorbance decay monitored at a specific wavelength may not properly represent the complex photocatalytic activity behaviors.

28

S. Bae et al. / Catalysis Today 224 (2014) 21–28

4. Conclusions The development of visible light active photocatalysts is being intensively investigated to achieve higher solar conversion efficiencies and their activities can be assessed by various methods. One of the most common methods is the dye decolorization test, which has been systematically evaluated in this study. The dye test has a great merit for its simplicity and rapidity, which is usually done by monitoring the dye absorbance at a specific wavelength as a function of irradiation time. However, despite the strong merit of the facile measurement, it suffers from many problems which make the dye test unsuitable as a standard method to assess the visible light photocatalytic activity. The activities determined by the dye test depend on not only the intrinsic photocatalytic activity but also other parameters/phenomena that are related with various interactions between the dye and the photocatalyst such as the absorption spectral overlap in the visible region, the electrostatic interaction (attractive or repulsive), the dye sensitization of catalyst particles, and the properties of dye degradation intermediates. Therefore, the practice of the dye test for the activity assessment of new photocatalytic materials should not be recommended (whenever possible although it is convenient and useful to some extent) and the activity results obtained for a specific combination of a dye and a photocatalyst should not be generalized. The visible light photocatalytic activities should be measured with test substrates that do not absorb visible light. As demonstrated with the TiO2 /UV system [50], the photocatalytic activities are highly substrate-specific and the test results can be very different depending on the choice of the test substrates. It is recommended that several substrates are selected from different classes of compounds. In our previous study [50], we proposed that the photocatalytic activity assessment be carried out using the following four substrates: phenol, dichloroacetate, tetramethylammonium, and trichloroethylene to take the substrate-specificity into account. They are the aromatic, anionic, cationic, and chlorohydrocarbon compounds, respectively, which are markedly different in their molecular properties and structure. They are also proposed as test substrates for visible light activity assessment. Nevertheless, when the dye test is the only option available for the activity assessment, a test dye that has a minimal absorption spectral overlap with the visible light photocatalyst and/or a minimal dye sensitization effect (e.g., MB) should be selected and the TOC removal should be measured as well. In addition, it is always better to use multiple dyes instead of a single dye. Acknowledgements This work was supported by the Green City Technology Flagship Program funded by KIST (KIST-2012-2E23322), the Global Frontier R&D Program on Center for Multiscale Energy System (2011-0031571), and the KOSEF EPB center (No. 20080061892). References [1] A. Kudo, Y. Miseki, Chem. Soc. Rev. 38 (2009) 253. [2] M.R. Hoffmann, S.T. Martin, W. Choi, D.W. Bahnemann, Chem. Rev. 95 (1995) 69.

[3] A. Fujishima, K. Sunada, Y. Kikuchi, K. Hashimoto, Environ. Sci. Technol. 32 (1998) 726. [4] J. Yoon, M. Cho, H. Chung, W. Choi, Water. Res. 38 (2004) 1069. [5] M. Anpo, H. Yamashita, Y. Ichihashi, S. Ehara, J. Electroanal. Chem. 396 (1995) 21. [6] S. Kim, S.J. Hwang, W. Choi, J. Phys. Chem. B 109 (2005) 24260. [7] W. Choi, M.R. Hoffmann, J. Phys. Chem. 100 (1996) 2161. [8] J. Ryu, W. Choi, Environ. Sci. Technol. 38 (2004) 2928. [9] H. Kyung, J. Lee, W. Choi, Environ. Sci. Technol. 39 (2005) 2376. [10] W. Choi, J.Y. Ko, H. Park, J.S. Chung, Appl. Catal., B 31 (2001) 209. [11] S. Hwang, M.C. Lee, W. Choi, Appl. Catal., B 46 (2003) 49. [12] K. Tanaka, K. Padermpole, T. Hisanaga, Water. Res. 34 (2000) 327. [13] X. Yan, T. Ohno, K. Nishijima, R. Abe, B. Ohtani, Chem. Phys. Lett. 429 (2006) 606. [14] B. Ohtani, J. Photochem, J. Photochem. Photobiol. C 11 (2010) 157. [15] Y. Wu, M.Y. Xing, B. Tian, J. Zhang, F. Chen, Chem. Eng. J. 162 (2010) 710. [16] M. Qiao, Q. Chen, S. Wu, J. Shen, J. Sol–Gel Sci. Technol. 55 (2010) 377. [17] G. Yang, Z. Jiang, H. Shi, T. Xiao, Z. Yan, J. Mater. Chem. 20 (2010) 5301. [18] K.M. Parida, B. Naik, J. Colloid Interface Sci. 333 (2009) 269. [19] J. Zhong, F. Chen, J. Zhang, J. Phys. Chem. C 114 (2010) 933. [20] Y. Wu, M. Xing, J. Zhang, F. Chen, Appl. Catal. B 97 (2010) 182. [21] Y. Wu, L. Xiao, J. Zhang, F. Chen, Appl. Surf. Sci. 256 (2010) 4260. [22] M.-S. Wong, S.-W. Hsu, K.K. Rao, C.P. Kumar, J. Mol. Catal. A: Chem. 279 (2008) 20. [23] Q. Wang, Z. Jiang, Y. Wang, D. Chen, D. Yang, J. Nanopart. Res. 11 (2009) 375. [24] Q. Xiao, J. Zhang, C. Xiao, Z. Si, X. Tan, Sol. Energy 82 (2008) 706. [25] W. Ren, Z. Ai, F. Jia, L. Zhang, X. Fan, Z. Zou, Appl. Catal., B 69 (2007) 138. [26] H. Li, X. Zhang, Y. Huo, J. Zhu, Environ. Sci. Technol. 41 (2007) 4410. [27] Y. Wang, Y. Wang, Y. Meng, H. Ding, Y. Shan, X. Zhao, X. Tang, J. Phys. Chem. C 112 (2008) 6620. [28] J. Zhu, D. Yang, J. Geng, D. Chen, Z. Jiang, J. Nanopart. Res. 10 (2008) 729. [29] G.-S. Li, D.-Q. Zhang, J.C. Yu, Environ. Sci. Technol. 43 (2009) 7079. [30] A.K.L. Sajjad, S. Shamaila, B. Tian, F. Chen, J. Zhang, J. Hazard. Mater. 177 (2010) 781. [31] K. Lv, J. Li, X. Qing, W. Li, Q. Chen, J. Hazard. Mater. 189 (2011) 329. [32] D. He, L. Wang, H. Li, T. Yan, D. Wang, T. Xie, CrystEngComm 13 (2011) 4053. [33] H. Xie, D. Shen, X. Wang, G. Shen, Mater. Chem. Phys. 110 (2008) 332. [34] Z. Zhang, W. Wang, W. Yin, M. Shang, L. Wang, S. Sun, Appl. Catal. B 101 (2010) 68. [35] Z. Cui, D. Zeng, T. Tang, J. Liu, C. Xie, Catal. Commun. 11 (2010) 1054. [36] M. Shang, W. Wang, S. Sun, L. Zhou, L. Zhang, J. Phys. Chem. C 112 (2008) 10407. [37] H. Fan, D. Wang, L. Wang, H. Li, P. Wang, T. Jiang, T. Xie, Appl. Surf. Sci. 257 (2011) 7758. [38] T. Saison, N. Chemin, C. Chaneac, O. Durupthy, V. Ruaux, L. Mariey, F. Mauge, P. Beaunier, J.P. Jolivet, J. Phys. Chem. C 115 (2011) 5657. [39] W. Liu, Y. Yu, L. Cao, G. Su, X. Liu, L. Zhang, Y. Wang, J. Hazard. Mater. 181 (2010) 1102. [40] J. Bandara, J. Kiwi, New J. Chem. 23 (1999) 717. [41] F. Zhang, J. Zhao, L. Zang, T. Shen, H. Hidaka, E. Pelizzetti, N. Serpone, J. Mol. Catal. A: Chem. 120 (1997) 173. [42] F. Zhang, J. Zhao, T. Shen, H. Hidaka, E. Pelizzetti, N. Serpone, Appl. Catal. B: Environ. 15 (1998) 147. [43] J. Kim, C.W. Lee, W. Choi, Environ. Sci. Technol. 44 (2010) 6849. [44] M. Vautier, C. Guillard, J.M. Herrmann, J. Catal. 201 (2001) 46. [45] M. Sathish, B. Viswanathan, R.P. Viswanath, C.S. Gopinath, Chem. Mater. 17 (2005) 6349. [46] Y. Park, W. Kim, H. Park, T. Tachikawa, T. Majima, W. Choi, Appl. Catal. B 91 (2009) 355. [47] Y. Park, N.J. Singh, K.S. Kim, T. Tachikawa, T. Majima, W. Choi, Chem. Eur. J. 15 (2009) 10843. [48] N. Lakshminarasimhan, Y. Park, W. Choi, Chem. Phys. Lett. 452 (2008) 264. [49] J. Tang, Z. Zou, J. Ye, J. Catal. Lett. 92 (2004) 53. [50] J. Ryu, W. Choi, Environ. Sci. Technol. 42 (2008) 294. [51] H. Park, W. Choi, J. Phys. Chem. B 109 (2005) 11667. [52] T. Wu, G. Liu, J. Zhao, H. Hidaka, N. Serpone, J. Phys. Chem. B 102 (1998) 5845. [53] M. Stylidi, D.I. Kondarides, X.E. Verykios, Appl. Catal., B 40 (2003) 271. [54] K. Naito, T. Tachikawa, M. Fujitsuka, T. Majima, J. Phys. Chem. C 112 (2008) 1048. [55] N. Daneshvar, D. Salari, A.R. Khataee, J. Photochem. Photobiol. A 157 (2003) 111. [56] P. Qu, J. Zhao, T. Shen, H. Hidaka, J. Mol. Catal. A: Chem. 129 (1998) 257. [57] T. Zhang, T. Oyama, A. Aoshima, H. Hidaka, J. Zhao, N. Serpone, J. Photochem. Photobiol. A 140 (2001) 163. [58] K. Dai, H. Chen, T. Peng, D. Ke, H. Yi, Chemosphere 69 (2007) 1361.