Escolar Documentos
Profissional Documentos
Cultura Documentos
a r t i c l e i n f o a b s t r a c t
Article history: In this work, we analyzed the photocatalytic activity of TiO2:Ln (Ln = La, Ce, Pr, Nd, Sm, Eu and Gd) sys-
Received 7 October 2014 tems in the degradation of two organic dyes by means of a probabilistic approach to screen the factors
Received in revised form 6 May 2015 influencing their photocatalytic activity. The photocatalytic properties were described by six responses
Accepted 15 May 2015
(crystallite size, band gap energy, visible light absorption, specific surface area, zeta potential and photo-
catalytic conversion) as a function of lanthanide type (seven elements), lanthanide concentration (two),
pH level (two) and dye type (methyl orange and crystal violet). We analyzed 140 experiments using the
Keywords:
analysis of variance (ANOVA) method. The results show that band gap energy and specific surface area
Photocatalysis
Analysis of variance
exhibited behavior that depended on the atomic number of the element used. However, visible light
Crystal violet absorption was not modified by the type of lanthanide element or its concentration in the range studied
Methyl orange (0.1 and 0.3 wt.% lanthanides/TiO2 ratio), and the zeta potential depended mainly on pH. Photocatalytic
Lanthanides conversion depended on pH, dye type and doping element. The experimental factorial design of the pho-
TiO2 tocatalytic system allowed us to find the best conditions for enhancing the degradation of these two dyes.
Ó 2015 Elsevier B.V. All rights reserved.
1. Introduction and the conduction band (ht P Eg) [6,7]. This generates an exciton
or electron–hole pair within the semiconductor; these migrate to
The discharge of dye-contaminated wastewater by different the material’s surface and initiate a series of redox reactions. The
types of industry seriously damages microorganisms, aquatic life final outcome of these events is the oxidation of pollutants that
and humans [1]. Due to their complexity, a variety of are present in the reaction medium. The mean lifetime of an elec-
non-biodegradable and highly water-soluble dyes are difficult to tron–hole pair is short (30 ± 15 ns) [7]. The incorporation of lan-
remove from wastewater by conventional methods [2]. thanide ions onto the TiO2 surface increases the lifetime of
TiO2-based photocatalysts are commonly used agents for the charge carriers due to energy transfer phenomena between the
decomposition of organic pollutants present in wastewaters from host and dopant ions in TiO2 [8,9].
various industries, factories, laboratories, etc., which represent a Experimentation has been used to understand and/or improve
serious problem for the environment [3–5]. More recently, hybrid these catalytic systems; however, there may be a large number
materials, such as doped or decorated TiO2 particles, have become of factors, but usually only a handful of them are important or sig-
available to improve this photocatalysis process. nificant. A screening experiment is therefore important to identify
The photocatalytic oxidation process consists of the activation the influential factors. Once these factors are identified, the next
of particles of semiconducting materials, such as the abovemen- stage of the investigation focuses on the elucidation and optimiza-
tioned titanium dioxide, which is excited with light having energy tion of the relationship between the response and the factors. The
greater than or equal to the difference between the valence band experiments to screen the factors are usually performed at the
early stages of a design project when many of the factors initially
considered have very little or no effect on the performance. To con-
⇑ Corresponding author. Tel.: +52 (222) 229 5610x2045; fax: +52 (222) 229 56
11.
duct the experiment, the control factors and noise factors should
E-mail address: omeza@ifuap.buap.mx (O. Meza). be identified [10–12]. The effects of these factors are then investi-
1
Both authors contributed equally to this work. gated at different levels. As the number of levels increases for each
http://dx.doi.org/10.1016/j.commatsci.2015.05.014
0927-0256/Ó 2015 Elsevier B.V. All rights reserved.
E.G. Villabona-Leal et al. / Computational Materials Science 107 (2015) 48–53 49
control and noise factor, the number of combinations becomes kept in the dark under magnetic stirring for 30 min. After 60 min
large and unmanageable. Therefore, running experiments with all under UV–vis light, the suspension was filtered and the dye con-
these combinations becomes uneconomical and time consuming. centration was determined by UV–vis absorption spectroscopy
In this work, we present a screening of factors that may influ- using a Hewlett Packard 8453 spectrophotometer.
ence the photocatalytic properties of TiO2–Ln systems. We ana- X-ray diffraction patterns were collected using a Siemens D500
lyzed six responses: crystalline size, band gap energy, visible diffractometer with a Cu tube with Ka radiation at 1.5405 Å, scan-
light absorption, surface area, zeta potential and photocatalytic ning in the 20–70° 2h range with increments of 0.021 and as wept
conversion. These responses were studied as a function of four fac- time of 1 s. Nitrogen adsorption isotherms were measured at liquid
tors: (A) seven lanthanide elements, (B) two lanthanide concentra- nitrogen temperature (77 K) using a Micromeritics ASAP 2020
tions, (C) two pH levels and (D) two dye types. We used dyes with spectrometer in order to determine the specific surface area
opposite electrical natures, i.e. crystal violet is a cation at both (SSA) of the photocatalysts by BET. Before the measurements, the
reaction pH levels, while methyl orange is an anion under natural samples were outgassed at 350 °C for 2 h [16]. Electrophoretic
conditions. These dyes also possess different electrical characteris- measurements of zeta potential (Zp) at different pH values were
tics than the TiO2–Ln system, as well as chemical stability and performed using a Zeta-Meter 3.0+ unit equipped with a molybde-
minor discoloration when exposed to light for a long period of num anode and a platinum cathode [17,18]. All the experiments
time. As a result we find that photocatalytic activity of all doped (or measurements) are summarized in Table 1.
samples does not depend on the type of Ln, nor on its concentra-
tion, nevertheless if we compare the doped and the undoped sam-
ples, the doped samples have better photocatalytic activity (for any 3. Results and discussions
Ln). This is because the addition of lanthanides inhibits crystal
growth, and this increases SSA. We are confident that our statisti- Analysis of variance (ANOVA) is an effective analysis tool that
cal analysis will help the understanding of the role of each factor allows the simultaneous comparison of factors to determine if they
studied in the photocatalytic activity. are identical or significantly different. ANOVA tell us if the variance
in our data is significant and measures that significance [19,20].
2. Experimental The experimental results are organized in three ANOVA designs
(see Table 1). In the first ANOVA, we analyzed four responses: crys-
TiO2–Ln powders were synthesized and characterized as previ- tallite size, band gap energy, visible light absorption and surface
ously reported [13]. Briefly, these systems were obtained by the area as a function of two factors: (A) seven lanthanide elements
sol–gel method using titanium n-butoxide as the precursor. The and (B) two concentrations, i.e. 0.1 or 0.3 wt.% Ln. The number of
materials were doped with lanthanides (Ln = La, Ce, Pr, Nd, Sm, experiments (or measurements) in this first ANOVA was
Eu and Gd) by the addition of lanthanide nitrate as a precursor salt. 4 7 2 = 56. The average crystallite size d was estimated using
Lanthanide aqueous solutions were prepared by stoichiometric the Scherrer equation and diffraction spectra (see Fig. 1): d = K
addition to obtain 0.1 and 0.3 wt.% Ln on the TiO2. The obtained k/b cos h, where b is the full width at half maximum (FWHM) of
gel was dried in an oven at 120 °C for 12 h and finally annealed the anatase peak (28.17°). The band gap energy (Eg) was estimated
at 500 °C for 4 h using a heating rate of 2 °C/min. from the spectra in percent reflectance mode by means of the
Photocatalytic tests were carried out in a microscale photoreac- Kubelka–Munk function: F(R) = (1R)2/2R, where R is the diffuse
tor equipped with a 5.5 W low-pressure cold-cathode mercury reflectance, through the corresponding Tauc plots: (F(R) E)1/2
lamp (UVP Pen-Ray 90-0012-01). In a typical experiment [13– and E is the energy of absorbed light, by extrapolation of the
15], 10 mg of the photocatalyst was added to 10 mL of a solution straight line to the abscissa [21–23]. The visible light absorption
containing 35 ppm of methyl orange (mo) or crystal violet (cv); (VLA) was calculated from the spectra of absorbance mode by
the pH was adjusted to 5.6 or 3.1 with HCl. The suspension was numerical integration in the range of 400–700 nm [24,25]. All
Table 1
It shows the experimental design with related factors and responses. Four factors: (A) lanthanide type, (B) concentration, with – and + equal to 0.1 and 0.3 wt.%, (C) two pH
conditions, with – and + equal to 3.1 and 5.6 and (D) two dyes, with – and + equal to crystal violet and methyl orange, respectively. Five responses: energy gap, UV–visible
absorption, surface area, zeta potential and photocatalytic conversion.
D
+
C
+ + +
A B Size (nm) Eg (eV) VLA SSA (cm2/g) Pz Conversion%
La 8.4 3.23 33.99 90 9.5 18.7 34.1 34.9 29.5 14
Ce 8.4 3.18 32.06 98 9.5 23.9 28.5 38.5 35.9 13,5
Pr 8.4 3.25 29.45 99 7.8 27.3 20.8 41.3 51.9 8,2
Nd 8.4 3.22 27.99 99 7.8 24.2 22.3 38.5 57.1 11,1
Sm 8.4 3.18 25.59 106 7.8 25.0 27.5 38.4 46.7 12
Eu 8.3 3.18 31.16 102 9.5 20.4 26.2 43.5 45.9 15,4
Gd 9.9 3.17 32.62 111 7.7 16.3 35.2 45.0 39.2 17,6
La + 7.2 3.15 31.70 119 25.8 25.8 30.3 40.9 40.2 13,5
Ce + 9.0 3.13 53.96 109 14.5 18.2 19.7 40.1 50.2 17,3
Pr + 8.4 3.19 38.87 112 8.7 17.0 32.3 41.7 38.2 11,9
Nd + 8.4 3.22 33.96 112 9.4 31.2 25.4 54.0 30.3 6,7
Sm + 7.1 3.18 32.15 110 15.5 22.5 24.0 35.6 52 15,2
Eu + 8.3 3.20 30.55 101 16.4 13.6 21.7 40.9 42.5 12,4
Gd + 8.3 3.22 29.37 94 10.4 20.9 27.6 58.3 43.4 8,5
Undoped 9.9 3.18 33.99 86 6.3 33.1 25.8 35.1 31.5 13.4
Designs (1) (2) (3)
50 E.G. Villabona-Leal et al. / Computational Materials Science 107 (2015) 48–53
TiO2 : Ln (0.3%) als did not depend on the type of Ln, nor on its concentration,
because in all cases the P-values were greater than 0.05. Thereby,
the mean value of crystallite size for all doped samples was
8.4 ± 0.7 nm, while the undoped sample had crystallites 9.9 nm
TiO2 : La (0.1%)
in diameter, i.e. the experimental results show that the size was
reduced (16%) when the material was doped. Table 2 also shows
that the energy band gap depended on the interaction of factors
TiO2 (AB) and not in the individual factors A and B. This is further illus-
trated in Fig. 2, where Eg is plotted against the type of Ln (factor A)
at the two lanthanide concentrations (factor B). In order to obtain a
20 30 40 50 60 70 relationship between Eg and the individual effects of the type Ln
2θ (factor A) and the concentration of lanthanide (factor B), including
the interaction factor (AB) (suggested by P-value, see Table 2), a fit
Fig. 1. Diffraction patterns of TiO2, TiO2:La (0.1 wt.%) and TiO2:La (0.3 wt.%). For of the experimental values of Eg with these factors was done and
studied cases, the crystal structure did not change with the type of dopant or the fit is given by Eq. (1). As result, it is worth noting that the con-
dopant concentration.
tribution of the individual effects of A and B are weak compared to
the combined effect AB.
Eg (eV)
Eg (eV)
Eg (eV)
3.2 3.2 3.2 3.2
0.1 0.2 0.3 0.1 0.2 0.3 0.1 0.2 0.3 0.1 0.2 0.3
B B B B
Eg (eV)
Eg (eV)
3.2 3.2 3.2
Fig. 3. Experimental and ANOVA prediction for the Eg as a function of lanthanide type and concentration.
55
0.3 B = 0.1 wt%
B = 0.3 wt%
0.2 50
0.1
Δ Eg / Δ B (eV/Wt.%)
45
0
VLA (a.u.)
40
−0.1
−0.2 35
−0.3
30
−0.4
−0.5 25
57 58 59 60 61 62 63 64 57 58 59 60 61 62 63 64
Fig. 4. Slope of Eg vs. atomic number from the experimental data. An increase in the Fig. 5. Experimental VLA as a function of atomic number and Ln concentration (B).
atomic number (N) increases the slope, going from a negative to a positive slope. VLA was not affected by the type of lanthanide element or concentration.
Where B is the Ln concentration.
120 B=0.1wt%,pH=3.1
B = 0.1 wt% 30 B=0.3wt%,pH=3.1
B = 0.3 wt%
B=0.1wt%,pH=5.6
115
20 B=0.3wt%,pH=5.6
110
10
pH
SSA (cm 2/g)
105 0
Pz
−10
100
−20
95
−30
90
57 58 59 60 61 62 63 64
−40
Atomic Number 57 58 59 60 61 62 63 64
Atomic Number
Fig. 6. Experimental SSA as a function of atomic number and Ln concentration (B).
Fig. 8. Zeta potential as a function of all the studied factors. No relationship was
observed between the type of lanthanide and its concentration (B).
strength and pH. Fig. 8 shows photocatalytic conversion as a func-
tion of all the factors; see also Table 4. We can see that no relation-
ship was found between the type of Ln and its concentration on
Table 3
photocatalytic conversion. Photocatalytic conversion was modeled Analysis of variance tables for the second experimental design. Zeta potential as a
by the following equation: function of three factors: (A) lanthanide type, (B) concentration and (C) pH.
Statistically significant factors are denoted with ⁄.
Pcon ¼ 44:24 2:99 C þ 36:54 D 9:17 CD ð5Þ
Zeta potential
where C is the pH level (3.1 or 5.6) and D is the dye type (1 for Source Sum of squares Df Mean square F-ratio P-value
crystal violet and +1 for methyl orange). Eq. (5) predicts (with A 41.8951 1 41.8951 1.15 0.2950
R2 = 82%) that the maximum photocatalytic conversion is around B 25.8432 1 25.8432 0.71 0.4084
C 7056.44 1 7056.44 194.33 0.0000⁄
42% for crystal violet (D = 1) at the high pH level (C = +1); see
AB 73.4508 1 73.4508 2.02 0.1696
Fig. 9. The opposite electrical nature of TiO2–Ln (negative) and crys- AC 170.776 1 170.776 4.70 0.0417
tal violet (positive) at pH 5.6 promotes dye adsorption onto the TiO2 BC 6.70321 1 6.70321 0.18 0.6718
surface. However, the maximum photocatalytic conversion (around Total error 762.561 21 36.3124
43%) for methyl orange (D = +1) occurred at the low pH level Total (corr.) 8137.67 27
(C = 1) because the methyl orange has a bipolar nature, and the
ionized groups are around 70% positive and 30% negative at pH
3.1 (calculated using the Henderson–Hasselbalch equation) [26]. Finally, Table 5 shows the average for the doped and undoped
Evidently, this dye is susceptible to preferential adsorption on the materials for crystallite size, specific surface area and photocat-
TiO2 surface by groups Ti–OH+2 (low pH) and Ti–O (high pH). alytic conversion. This table shows that the maximum
2
2
90 90 90 90
0.1 0.2 0.3 0.1 0.2 0.3 0.1 0.2 0.3 0.1 0.2 0.3
B B B B
90 90 90
0.1 0.2 0.3 0.1 0.2 0.3 0.1 0.2 0.3
B B B
Fig. 7. Experimental and ANOVA prediction for the SSA as a function of lanthanide type and concentration.
E.G. Villabona-Leal et al. / Computational Materials Science 107 (2015) 48–53 53
Table 4
8.4 ± 0.7 nm; the undoped sample was 9.9 nm in diameter, so the
Analysis of variance tables for the third experimental design. Photocatalytic conver-
sion as a function of three factors: (A) lanthanide type, (B) concentration, (C) pH and experimental results show that the size was reduced by 15.7%
(D) dye type. Statistically significant factors are denoted with ⁄. when the material was doped. This size reduction increased
21.4% the SSA when the material was doped. The photocatalytic
Photocatalytic conversion
activity depended on the surface interactions between the catalyst
Source Sum of squares Df Mean square F-ratio P-value
and the dye. As expected, an increase in the specific surface area
A 71.8911 1 71.8911 1.90 0.1753 increased the conversion by around 20.4% and 31.5% for crystal
B 0.07875 1 0.07875 0.00 0.9639
violet and methyl orange, respectively.
C 785.252 1 785.252 20.71 0.0000⁄
D 623.779 1 623.779 16.45 0.0002⁄
AB 30.9772 1 30.9772 0.82 0.3709 4. Conclusions
AC 6.14469 1 6.14469 0.16 0.6892
AD 3.83254 1 3.83254 0.10 0.7520
This study screened factors influencing the photocatalytic activ-
BC 41.3145 1 41.3145 1.09 0.3021
BD 20.0402 1 20.0402 0.53 0.4710 ity of TiO2:Ln using the analysis of variance (ANOVA) method.
CD 7353.19 1 7353.19 193.93 0.0000⁄ Surface area and energy gap were found to depend on the interac-
Total error 1706.26 45 37.917 tion between the dopant Ln element and its concentration. In both
Total (corr.) 10642.8 55
cases, a clear dependence on the lanthanide atomic number was
observed. Visible light absorption did not depend on the type of
lanthanide or its concentration. This likely occurred since the
B=0.1wt%,pH=3.1 B=0.3wt%,Ph=3.1 dopants did not change the internal structure of the material.
B=0.1wt%,pH=5.6 B=0.3wt%,pH=5.6 Zeta potential was proportional to the pH level. Ln elements affect
60 the photocatalytic activity because doping inhibits the growth of
(a) cv
Conversion %