Fang Jianga,
Wenhui Tana,
Huan Chen*a,
Ling Tana and
Jingliang Liub
aKey Laboratory of Jiangsu Province for Chemical Pollution Control and Resources Reuse, School of Environmental and Biological Engineering, Nanjing University of Science and Technology, Nanjing 210094, China. E-mail: hchen404@njust.edu.cn; Fax: +86-25-84315352; Tel: +86-25-84303209
bSchool of Biochemical Environmental and Engineering, Nanjing XiaoZhuang University, Nanjing 211171, China
First published on 1st June 2015
Catalytic hydrodechlorination (HDC) of chlorophenoxyacetic acids was performed over Pd/graphitic carbon nitride (Pd/g-C3N4) catalysts in the present work. A series of Pd/g-C3N4 catalysts were prepared by a deposition-precipitation method, and characterized by X-ray diffraction (XRD), N2 adsorption–desorption, transmission electron microscopy (TEM), CO chemisorption and X-ray photoelectron spectroscopy (XPS). The Pd/g-C3N4 catalysts showed excellent activity to convert dichlorophenoxyacetic acid (2,4-D) into phenoxyacetic acid (PA) and the catalytic activity was correlated with the ratio of Pd2+/(Pd0 + Pd2+) and Pd particle size. In addition, HDC of 2,4-D over Pd/g-C3N4 catalyst followed the Langmuir–Hinshelwood model, indicating an adsorption-controlled mechanism. Other chlorophenoxyacetic acids, such as 2-chlorophenoxyacetic acid (2-CPA), 4-chlorophenoxyacetic acid (4-CPA) and 2,4,5-trichlorophenoxyacetic acid (2,4,5-T) could also be completely dechlorinated to PA over Pd/g-C3N4 catalysts.
An alternative approach for the removal of chlorophenoxyacetic acids is liquid phase catalytic hydrodechlorination (HDC), which is one of the most effective and environmentally friendly methods for the abatement of many organochlorinated pollutions, such as chlorophenols,17–19 chlorobenzenes20,21 and halogenated hydrocarbons,22,23 etc. More recently, Diaz et al.24 explored the HDC process of 2,4-D and found that 2,4-D could be dechlorinated to 2-CPA and PA. Besides of this, there are no far investigations on HDC of chlorophenoxyacetic acids and the reaction mechanism is also scarce.
Previous literatures on HDC of organochlorinated pollutions have demonstrated the important role of Pd active species on catalytic efficiency.25,26 The dependence of HDC activity on Pd particle size was investigated, while no general consensus has been achieved. Gómez-Quero et al.27 investigated the effect of metal particle size on the HDC of 2,4-dichlorophenol over Pd/Al2O3 catalysts and found that a higher activity was achieved over smaller Pd particles. Nevertheless, a volcano-type dependence of catalytic activity on Pd particle size was observed in HDC of chloroacetic acid over Pd/ZrO2 catalysts.28 Apart from Pd particle size, Pd electronic structure is another key factor which markedly affects the catalytic activity. Gómez-Sainero et al.29 studied the catalytic HDC of CCl4 over Pd/AC and found both Pd0 and Pdn+ species had effect on HDC efficiency. Similar conclusions were obtained in HDC of 2,4-dichlorophenol over mesoporous carbon supported Pd catalysts.25 Besides of Pd supported catalysts, bimetallic catalysts, such as Pd–Fe,30 Pd–Rh,31 Pd–Cu,32 were approved to be efficient catalysts to treat many organochlorinated pollutions.
Carbon nitride is a fascinating material which has attracted worldwide attention for its excellent textural properties and extreme chemical and thermal stability.33,34 We have recently shown that the carbon nitride material was an effective support for Pd dispersion due to its nitrogen functionalities, which could serve as active sites to anchor Pd particle,35 and the Pd/graphitic carbon nitride catalysts (Pd/g-C3N4) prepared by photo-deposition method could be used in catalytic hydrodechlorination of 2,4-dichlorophenol.36 Similarly, Feng et al.37 prepared an effective palladium catalyst supported on carbon–nitrogen composites for aqueous-phase hydrogenation of phenol.
In order to get further insight about the effect of Pd properties on HDC of chlorophenoxyacetic acids, we used deposition-precipitation method to prepare a series of Pd/g-C3N4 catalysts, and these catalysts were tested in HDC of 2-CPA, 4-CPA, 2,4-D and 2,4,5-T in this work. Results showed that chlorophenoxyacetic acids could be completely dechlorinated to PA, revealing the potential of the liquid phase catalytic HDC method to remove chlorophenoxyacetic acids.
Serial number | Catalysta | pHb | NaBH4![]() ![]() |
SBETd (m2 g−1) |
---|---|---|---|---|
a The Pd content were determined by ICP.b The pH value during catalyst preparation.c The molar ratio of NaBH4 and Pd during catalyst preparation.d Determined by N2 adsorption–desorption using the Brunauer–Emmett–Teller (BET) method. | ||||
— | g-C3N4 | — | — | 57.35 |
1 | Pd(2.09)/g-C3N4 | 10 | 0.5![]() ![]() |
56.42 |
2 | Pd(1.94)/g-C3N4 | 10 | 1![]() ![]() |
55.87 |
3 | Pd(1.96)/g-C3N4 | 10 | 3![]() ![]() |
49.15 |
4 | Pd(2.07)/g-C3N4 | 10 | 10![]() ![]() |
49.22 |
5 | Pd(1.98)/g-C3N4 | 9 | 3![]() ![]() |
30.65 |
6 | Pd(1.93)/g-C3N4 | 8 | 3![]() ![]() |
35.40 |
7 | Pd(0.46)/g-C3N4 | 10 | 3![]() ![]() |
54.68 |
8 | Pd(0.98)/g-C3N4 | 10 | 3![]() ![]() |
49.51 |
9 | Pd(4.11)/g-C3N4 | 10 | 3![]() ![]() |
40.80 |
X-ray diffraction (XRD) patterns of the catalysts were recorded on a Rigaku D/max-RA powder diffraction-meter. The wavelength of the beam was 0.154 nm. The XRD patterns were scanned in the 2θ range of 10–80° with step width 0.05°. Plots of the 2θ data were generated from the EVA software developed by Bruker.
Brunauer–Emmett–Teller (BET) specific surface areas of the support and catalysts were obtained from the N2 adsorption–desorption isotherms, measured at 77 K using Micromeritics ASAP 2020. Before measurement, the samples were outgassed at 250 °C for 4 h.
The morphologies of the catalysts were observed on a transmission electron microscope (JEM-2100, JEOL, Japan). The average Pd particle size could be quantified using equation:39
![]() | (1) |
X-ray photoelectron spectroscopy (XPS) was conducted on a PHI5000 Versa Probe equipped with a monochromatized Al Kα excitation source (hν = 1486.6 eV) (ULVAC-PHI, Japan).
Pd dispersion of the catalysts was determined by CO chemisorption measurement. 200 mg of reduced catalyst was flushed by Ar flow (30 mL min−1) at 200 °C for 1 h. After the temperature of the catalyst was cooled down, the CO chemisorption was conducted with pulses of 0.4 mL. The Pd dispersion was calculated on the assumption of adsorption stoichiometry of Pd/CO = 1.40
The effect of mass transfer limitation under the reaction condition was evaluated. HDC of 2,4-D over catalyst 3 with different catalyst dosages was compared and the results were presented in Fig. 1S (see ESI†). Increasing catalyst dosage enhanced the HDC efficiency; while the dosage normalized initial activities remained nearly constant, indicating the absence of mass transfer limitation.41 The initial activity was calculated by the removal rate of 2,4-D within initial 3 min.
The XRD patterns of g-C3N4 and Pd/g-C3N4 catalysts are compiled in Fig. 2S.† Clearly, g-C3N4 has two distinct peaks at 13.1° and 27.7°, which are indexed as (100) and (002) diffraction, respectively (JCPDS 87-1526). The intensities of the two peaks slightly decreased for catalyst 1 to 9, attributed to the introduction of Pd. However, there was no diffraction peaks assigned to metallic or oxide Pd in all catalysts, implying of a high Pd dispersion.
![]() | ||
Fig. 1 Liquid phase catalytic HDC of 2,4-D over catalyst 1, 2, 3, 4, 5 and 6. (![]() ![]() ![]() ![]() ![]() |
![]() | ||
Scheme 1 Dechlorination pathways of the liquid phase catalytic HDC of 2-CPA, 4-CPA, 2,4-D and 2,4,5-T. |
Notably, the concentration of 2-CPA was much higher than that of 4-CPA in all HDC processes over these catalysts in Fig. 1, indicating that the para-position Cl atom was more susceptible to be eliminated than ortho-position Cl as a result of steric hindrance. The role of steric hindrance has also been reported in HDC of 2,4-dichlorophenol, as well as electrocatalytic dechlorination of 2,4-D.15,16
Among the Pd/g-C3N4 catalysts prepared with different amount of NaBH4 (catalyst 1 to 4), catalyst 3 possessed the highest catalytic activity. After reaction for 90 min, 0.48 mmol 2,4-D was completely dechlorinated to PA. The catalytic efficiency varied in the order catalyst 3 > catalyst 2 > catalyst 4 > catalyst 1. In comparison with Pd/g-C3N4 prepared in different pH condition (catalyst 3, 5 and 6), the catalytic efficiency followed the order catalyst 3 > catalyst 5 > catalyst 6, suggesting that Pd/g-C3N4 prepared in high pH solution favored the HDC process.
Considering the reason causing the difference in catalytic activity among these Pd/g-C3N4 catalysts, we should know that the catalytic activities are relevant to the properties of Pd, such as Pd particle size and Pd species. To obtain the morphology and dispersion of Pd particles, TEM and CO chemisorptions measurements were investigated. The TEM images of catalyst 1 to 6 are displayed in Fig. 2, with uniform Pd particles clearly identified in all catalysts. The average sizes of the Pd particles calculated by eqn (1) are listed in Table 2. The Pd dispersions measured by CO chemisorption are also listed in Table 2. They were found to be 44.20%, 48.13%, 46.31%, 45.66%, 45.68% and 48.52% for catalyst 1 to 6, respectively. The corresponding Pd particle sizes calculated from CO chemisorption was consistent with the results of TEM. Both TEM and CO chemisorption results indicated that these catalysts exhibited very similar Pd dispersion and Pd particle size, thus the Pd particle size was not a major factor causing the differences in HDC activities among these catalysts. Worth noting is Pd dispersion on these catalysts was relevantly high. The interaction between the pyridine nitrogen groups and the Pd nanoparticles enabled high Pd dispersion on the g-C3N4 support. A previous work has also shown well-dispersed Pd nanoparticles on NH3 functionalized carbon nanofibers, which attributed to the similar interaction between the pyridinic nitrogen and Pd nanoparticles.43
Catalyst | CO uptakea (μmol g−1) | Pd dispersiona (%) | Average Pd diameter (nm) | Binding energy (eV) | Pd2+/(Pd0 + Pd2+) | ||
---|---|---|---|---|---|---|---|
d1a | d2b | Pd0 | Pd2+ | ||||
a Calculated from CO chemisorption.b Calculated from TEM. | |||||||
1 | 86.8 | 44.20 | 2.54 | 2.51 | 335.2 | 337.3 | 0.71 |
2 | 87.7 | 48.13 | 2.33 | 2.40 | 335.2 | 337.1 | 0.64 |
3 | 85.3 | 46.31 | 2.43 | 2.57 | 335.5 | 337.2 | 0.50 |
4 | 88.8 | 45.66 | 2.46 | 2.47 | 335.1 | 337.0 | 0.39 |
5 | 85.0 | 45.68 | 2.46 | 2.49 | 335.2 | 337.2 | 0.42 |
6 | 88.0 | 48.52 | 2.31 | 2.69 | 335.2 | 337.2 | 0.41 |
7 | 25.5 | 59.00 | 1.90 | 2.07 | 335.0 | 337.0 | 0.64 |
8 | 45.1 | 48.98 | 2.29 | 2.36 | 335.2 | 337.0 | 0.55 |
9 | 110.3 | 28.55 | 3.93 | 4.22 | 335.3 | 337.1 | 0.46 |
XPS was performed to understand the surface chemical states of Pd species on Pd/g-C3N4, and the results were illustrated in Fig. 3. The binding energy of Pd 3d5/2 around 335.2 eV and 337.2 eV could be attributed to the presence of Pd0 and Pd2+ species, respectively. Upon NaBH4 reduction treatment, Pd2+ is partially retained due to the stabilization of Pd2+ by interaction of Pd2+ to N-containing groups, such as pyridine nitrogen groups which existed on the support surface.44,45 The intensity of Pd2+ species decreased when the amount of NaBH4 increased during catalyst preparation (catalyst 1 to 4), indicating weakened interaction. The ratios of Pd2+/(Pd0 + Pd2+) in these catalysts are summarized in Table 2. It can be clearly seen that the ratio of Pd2+/(Pd0 + Pd2+) decreased from 0.71 (catalyst 1) to 0.39 (catalyst 4), reflecting of decreased Pd2+ content. For samples prepared in different pH, the ratios of Pd2+/(Pd0 + Pd2+) followed the order catalyst 3 (pH 10) > catalyst 5 (pH 9) > catalyst 6 (pH 8), implying that higher Pd2+ content existed in the catalyst prepared in higher pH condition. As reported, the isoelectric point of g-C3N4 was around 2.0.36 The zeta potential of g-C3N4 decreased as elevating the solution pH, caused strengthening of the interaction between Pd2+ and negative g-C3N4. As a result, more Pd2+ species were retained during the NaBH4 reduction process at high pH solution.
More information was gained by the comparison of turnover frequency (TOF) versus ratios of Pd2+/(Pd0 + Pd2+) as shown in Fig. 4. As shown in Fig. 4, the initial activity and TOF increased with the ratio up to 0.5, and then progressively decreased. During the process, molecular H2 could chemisorb and activated on the surface of Pd0, whereas 2,4-D tended to chemisorb onto Pd2+ by slightly electrostatic attraction for the presence of chloride anion in 2,4-D. Thus, both Pd0 and Pd2+ could affect the catalytic activity of Pd/g-C3N4. Similar result has been depicted by Gómez-Sainero et al.29 on liquid-phase hydrodechlorination of CCl4. It was demonstrated that the HDC reaction rate was related to the amount of Pd0 and Pdn+, and the maximum reaction rate achieved when [Pdn+]/[Pd0] = 1. Here, we observed that a maximum reaction rate could be obtained in the catalyst 3 with Pd2+/(Pd0 + Pd2+) = 0.5, which demonstrated that both Pd0 and Pd2+ species play an important and complementary role on HDC reaction. A dual mechanism was also proposed by Ordóñez et al.46 to explain the HDC of tetrachloroethylene (TTCE) over supported Pd catalysts. They found that Pd2+ could form a complex with TTCE, and then the complex was attacked by hydrogen activated by Pd0. A reaction scheme similar to previous reports can be proposed: 2,4-D can form a complex with the Pd2+, and this complex will be attacked by the hydrogen to accomplish the Cl–H exchange, which is accomplished by the reduction of the Pd2+ to Pd0 (as speculated in HDC of TTCE46). The XPS spectra of used catalyst 3 shown in Fig. 3S† confirmed the reaction scheme. As seen, some Pd2+ species were reduced to Pd0 during the HDC process, causing the increasing amount of Pd0.
Further insight was acquired by investigating the TOF versus Pd particle size and Pd species. Fig. 7a demonstrated a steep enhancement of TOF with Pd particle size increased from 1.90 to 2.54 nm, while further increase of Pd particle size caused a gentle increase in TOF, suggesting that the reaction is structure sensitive. This effect of structure sensitive has been found in other hydrodechlorination reactions. Aramendía et al.50 studied the liquid-phase HDC of chlorobenzene over Pd-supported catalysts; they found the reaction is structure-sensitive and the increase in dispersion leads to a decrease in TOF. Díaz et al.51 studied the Pd-catalyzed aqueous-phase trichloroethylene hydrodechlorination and found the reaction is strongly structure sensitive, being the largest Pd particles the most active. They considered the formation of palladium hydrides in the catalyst was especially important to the catalytic activity.
![]() | ||
Fig. 7 TOF of catalyst 3, 7, 8 and 9 versus (a) Pd particle size and (b) Pd2+/(Pd0 + Pd2+) for liquid phase catalytic HDC of 2,4-D. The dashed line in (b) was the result in Fig. 4. |
Notably, TOF versus the ratio of Pd2+/(Pd0 + Pd2+) exhibited in Fig. 7b differed from the result of Fig. 4 (the dashed line). It indicated that the variation of TOF was related to both Pd particle size and Pd species. As mentioned above, the activation of H2 and C–Cl bond was believed to be two crucial steps during the HDC process of 2,4-D. Small Pd particles with high Pd2+ content facilitated the activation of C–Cl bond, while catalyst with large Pd particles would boost H2 activation due to the formation of β-PdH.50 Hence, although the Pd(2.09)/g-C3N4 had the most appropriate ratio of Pd2+/(Pd0 + Pd2+), the TOF was lower than that of Pd(4.11)/g-C3N4, confirming the important role of Pd particle size on catalytic activity.
![]() | (2) |
![]() | (3) |
![]() | ||
Fig. 8 (a) Initial activity of catalyst 3 as a function of initial 2,4-D concentration and (b) linear plot of 1/r0 versus 1/C. |
The plot of 1/r0 with 1/C is further summarized in Fig. 8b. A good linear relationship (R2 > 0.97) suggested that the HDC of 2,4-D could be well described by Langmuir–Hinshelwood model and the reaction was controlled by 2,4-D adsorption on the catalyst surface. Such phenomena was also be observed by Wu et al.;54 they found the liquid phase hydrodechlorination of diclofenac over Pd/CeO2 could be well described using the Langmuir–Hinshelwood model, indicating the adsorption of reactor was crucial to catalytic efficiency.
![]() | (4) |
Initial activity (mM gcat−1 h−1) | S2,4-D (%) | S2,5-D (%) | S2-CPA (%) | S4-CPA (%) | SPA (%) | ||||||
---|---|---|---|---|---|---|---|---|---|---|---|
t10 | t30 | t10 | t30 | t10 | t30 | t10 | t30 | t10 | t30 | ||
2-CPA | 61.78 | — | — | — | — | — | — | — | — | 100 | 100 |
4-CPA | 80.56 | — | — | — | — | — | — | — | — | 100 | 100 |
2,4-D | 40.08 | — | — | — | — | 49.62 | 31.28 | 1.47 | 0.52 | 54.13 | 70.23 |
2,4,5-T | 9.64 | 23.32 | 12.95 | 12.59 | 6.63 | 14.38 | 13.92 | — | 0.006 | 56.14 | 70.42 |
It was found that the selectivity of 2-CPA was much higher than that of 4-CPA in HDC of 2,4-D at reaction time of 10 min and 30 min. The higher dechlorinated rate of 4-CPA and the higher selectivity of 2-CPA in HDC of 2,4-D could be attributed to steric effect. Cl on the ortho-position of 2,4-D was less susceptible to be attacked than Cl on the para-position. The possible dechlorination pathway of 2,4,5-T is presented in Scheme 1. 2,4-D, 2,5-D, 2-CPA, 4-CPA and PA were detected as the dechlorinated products during the HDC process of 2,4,5-T, while 4,5-D and 5-CPA were not detected. The selectivity of each product in HDC of 2,4,5-T at the reaction time of 10 and 30 min was listed in Table 3. As can be seen, the selectivity of 2,4-D and 2-CPA was much higher than that of 2,5-D and 4-CPA, indicating that 2,4-D and 2-CPA were the main partially dechlorinated products.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ra07913f |
This journal is © The Royal Society of Chemistry 2015 |