- Email: [email protected]
Electrochemical oxidation of phenol at boron-doped diamond electrode in pulse current mode
Electrochimica Acta 56 (2011) 5310–5315
Contents lists available at ScienceDirect
Electrochimica Acta journal homepage: www.elsevier.com/locate/electacta
Electrochemical oxidation of phenol at boron-doped diamond electrode in pulse current mode Junjun Wei ∗ , Xiuping Zhu, Jinren Ni Department of Environmental Engineering, Peking University, The Key Laboratory of Water and Sediment Sciences, Ministry of Education, Beijing 100871, China
a r t i c l e
i n f o
Article history: Received 28 February 2011 Received in revised form 30 March 2011 Accepted 1 April 2011 Available online 12 April 2011 Key words: Boron-doped diamond electrode Phenol Electrochemical oxidation Pulse current Response surface methodology
a b s t r a c t In this study, a pulse current supply was initially used in a BDD anode system (pulse-BDD anode system) for electrochemical oxidation of phenol. The influences of operative parameters (current density, retention time, pulse duty cycle, power frequency) on the system performances were exmamined by response surface methodology (RSM). As for COD degradation efficiency (DCOD ) and specific energy consumption (Es ), the influence of retention time was more important than current density and pulse duty cycle, while power frequency hardly presented significant influence. By the comparison with constant current mode, an obvious specific energy consumption reduction was achieved in the pulse-BDD anode system, though the DCOD was slightly lower. The significant Es decrease might be attributed to the reduction of side reactions and concentration polarization in pulse current mode. The pulse-BDD anode system demonstrated an efficient technology to simultaneously obtain high pollutant degradation efficiency and low energy consumption. © 2011 Elsevier Ltd. All rights reserved.
1. Introduction Electrochemical oxidation is a promising technology to treat bio-refractory wastewaters [1,2]. The degradation effectiveness mainly relies on the nature of the electrodes, as well as reactor construction, electrolysis conditions, etc. Boron-doped diamond (BDD) anodic oxidation demonstrates many attractive electrochemical performances, such as strong oxidation stability, high current efficiency, weak electrode fouling, and long service life [3–6]. However, the BDD anode system is still not applied extensively. It is clear that a technically efficient process must also be economically feasible with regard to its initial capital and operating costs. Energy consumption is a very important economical parameter in electrochemical oxidation process like all other electrolytic processes. Conventional BDD anode oxidation systems usually run in the constant current condition [7–10]. For the sake of satisfied COD degradation efficiency (DCOD ) and no anode passivation, relatively high constant current density is required [9–12]. In such a case, side reactions (e.g., oxygen evolution, H2 O2 and O3 generation) were strengthened and resulted high specific energy consumption (Es ) and operating costs [13]. Enhancement of mass transport was proven to be efficient to optimize the electrochemical oxidation process. For example, ultrasound enhance will signifi-
cantly improve electrochemical oxidation and decrease the specific energy consumption [14]. However, the additional energy consumption of ultrasound system is high and the reactor is not easy to scale up. Improvement of the power supplement seems an alternate reasonable approach to reduce the Es . Panizza et al. [15] applied multiple current steps electrolysis and a semi-continuous current control mode in the BDD electrolysis and achieved the mineralization of organic pollutants with high reaction rate and current efficiency and low energy consumption, but this technology is too complex to be applied extensively. Therefore, we proposed a pulse mode for the electrochemical anode oxidation system. In this study, a pulse current supply was initially used in a BDD anode system (pulse-BDD anode system) for electrochemical oxidation of phenol. The influences of the operative parameters (current density (j), retention time (t), power frequency (f), pulse duty cycle (x)) on COD degradation efficiency (DCOD ) and specific energy consumption (Es ) were investigated and optimized by response surface methodology (RSM). Furthermore, the pulse and constant current modes were compared according to the response functions to evaluate the superiority of pulse-BDD anode system. 2. Experimental 2.1. Electrolytic cell and power supply
∗ Corresponding author. Tel.: +86 10 6275 4290; fax: +86 10 6275 6526. E-mail addresses: [email protected] (J. Wei), [email protected] (J. Ni). 0013-4686/$ – see front matter © 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.electacta.2011.04.006
The electrolytic experiments were performed in a cell with two parallel planar electrodes. A Nb/BDD (boron-doped diamond film
J. Wei et al. / Electrochimica Acta 56 (2011) 5310–5315
on niobium substrate) plate with a working area of 24 cm2 (CONDIAS GmbH, Germany) was used as the anode. The boron-doped diamond film (BDD) was synthesized by the hot filament chemical vapor deposition technique (HFCVD) on niobium substrate. The doping level of boron in the diamond layer expressed as B/C ration was about 3500 ppm and the obtained diamond film thickness was about 2 m with a resistivity of 10–30 m cm. A stainless steel piece with the same size was as the cathode. The interelectrode distance was 1.55 cm. Thus, the working volume of this cell was 37.2 cm3 . Pulse electrolysis was performed using a Pulse Current Supply (DH2002, Beijing Dahua Electron Co. Ltd., China) as power source in a BDD anode system. The output current wave of this power supply was square pulses wave. A pulse cycle T is comprised of Ton and Toff , which is defined as turn-on period and turn-off period, respectively. At the Ton period, current supplied as constant amplitude; while at Toff period, the current supply was intermitted. In this study, the power frequency is equal to 1/T in the range of 10 Hz to 500 Hz, and the pulse duty cycle is determined by Ton /T in the range of 0–100%. 2.2. Electrolysis experiment
5311
Table 1 Experimental range and levels of the process independent variables.
Independent variable Current density (mA cm−2 ) Retention time (min) Power frequency (Hz) Pulse duty cycle (%)
Factor
Range and levels
Xi X1 X2 X3 X4
−2 6 15 10 10
−1 14 30 60 30
0 22 45 110 50
1 30 60 160 70
2 38 75 210 90
where X0 is value of the Xi at the center point, and ıX presents the step change. Each response Y can be represented by a mathematical equation that correlates the response surface. Y = b0 + b1 x1 + b2 x2 + b3 x3 + b4 x4 + b11 x12 + b22 x22 + b33 x32 + b44 x42 + b12 x1 x2 + b13 x1 x3 + b14 x1 x4 + b23 x2 x3 + b24 x2 x4 + b34 x3 x4 (2) where Y is the predicted response, b0 is constant, b1 , b2 , b3 , and b4 are the linear effects, b11 , b22 , b33 , and b44 are quadratic coefficients, and b12 , b13 , b14 , b23 , b24 , and b34 are interaction coefficients. In this context, DCOD and Es were as the responses. The COD (in mg L−1 ) of the solutions were measured by a titrimetric method using dichromate as the oxidant in acidic solution at 150 ◦ C for 2 h (Hachi, USA). For pulse current mode, the Es (in kWh kgCOD−1 ) was calculated using the following equation:
Phenol was chosen as the studied object, because it is a common and typical organic pollutant in many industrial effluents. The initial COD concentration was maintained the value of 500 mg/L. Na2 SO4 was used as the supporting electrolyte and the concentration was 0.05 M. Electrolysis experiments were conducted at different current density (j), retention time (t), power frequency (f), and pulse duty cycle (x). During electrochemical oxidation process, phenol simulated wastewater was continuously pumped through the electrolytic cell at a certain flow rate (determined by retention time).
where U is the cell potential (in V), I is the applied current, (in A), t is the retention time (in h), x is the pulse duty cycle of applied current supply, (in %), CODin and CODout are the input and output COD concentration, respectively, (in mg L−1 ), and V is the cell volume (in mL).
2.3. Experiment design and analysis
3. Results and discussion
Design of experiments (DOE) is a powerful tool to study the effect of variables and their responses with minimum number of experiments. Response surface methodology (RSM) is one of DOEs which can be a useful tool to examine the influence of operative parameters on system performances and to determine the optimum conditions. The most popular class of second order designs called central composite design (CCD) was used for RSM. Because the CCD is ideal for sequential experimentation and allows a reasonable amount of information for testing lack of fit while not involving an unusually large number of design point. In this study, RSM was initially used to describe and optimize the electrochemical oxidation process of Pulse-BDD anode system. The CCD with four factors at five levels was applied for the experimental design. The current density (coded as X1 ), retention time (coded as X2 ), power frequency (coded as X3 ) and pulse duty cycle (coded as X4 ) were chosen as independent variables. Each variable was coded at five levels between −2 and 2 at the range determined by the preliminary experiments and changed in the ranges of 6–38 mA cm−2 , 15–75 min, 10–210 Hz, and 10–90%, respectively. The variables design was shown in Table 1. According to the design of RSM in terms of Minitab 15 software, thirty-one experiments were augmented with seven replications at the design center to evaluate the pure error. For statistical calculations, the variables Xi were coded as xi according to the following relationship:
3.1. Influences of operative parameters
xi =
Xi − X0 ıX
(1)
Es =
1000UItx (CODin − CODout )V
(3)
Thirty-one CCD designed batch runs were conducted to visualize the effects of independent factors (j, t, f and x) on Pulse-BDD anode system responses (DCOD and Es ). The results along with the experimental conditions were presented in Table 2. The DCOD was in the range of 33.72–81.82% and the Es was in 23.2–277 kWh kgCOD−1 . Application of RSM offers, the relationships of DCOD and Es with independent studied variables were obtained as shown in Eqs. (4) and (5), respectively. Y (DCOD ) = 70.7171 + 4.8125x1 +11.3975x2 + 1.3208x3 + 3.9175x4 − 2.2999x12 − 2.764x22 − 0.2574x32 − 3.4499x42 − 0.4788x1 x2 − 0.2825x1 x3 + 2.1875x1 x4 − 0.3088x2 x3 + 0.2088x2 x4 − 0.135x3 x4
(R2 = 0.9407)
(4)
Y (Es ) = 110.224 + 41.8152x1 + 19.0435x2 + 1.07x3 + 51.3014x4 + 0.1347x12 + 2.1642x22 − 1.198x32 + 3.342x42 + 8.8574x1 x2 + 0.2283x1 x3 + 20.5113x1 x4 + 1.2404x2 x3 + 9.24x2 x4 − 0.4015x3 x4
(R2 = 0.9920)
(5)
The correlation factors (R2 ) were respectively equal to 0.9407 and 0.9920 indicating a good fit for the dependent variables. The residuals were the deviations of the experimental data value from the predicted values and were estimates of the error terms in
5312
J. Wei et al. / Electrochimica Acta 56 (2011) 5310–5315
Table 2 Experimental planning in DOE and obtained responses in each experiment. Run
j (mA cm−2 )
t (min)
f (Hz)
x (%)
DCOD (%)
Es (kWh kgCOD−1 )
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31
22 22 30 30 30 22 22 14 38 22 30 30 14 14 30 30 14 22 30 22 22 14 22 22 22 22 6 14 14 22 14
45 45 60 60 60 45 45 60 45 45 30 30 30 60 30 30 60 45 60 45 45 30 75 45 15 45 45 30 30 45 60
110 110 160 60 160 110 10 60 110 110 160 60 60 160 60 110 60 110 60 110 110 160 110 210 110 110 110 160 60 110 160
50 50 70 70 30 50 50 30 50 50 70 70 30 30 30 30 70 50 30 10 50 30 50 50 50 90 50 70 70 50 70
69.9 69.1 81.23 81.82 74.19 72.43 65.69 68.14 74.38 71.85 64.81 57.77 45.64 71.85 49.37 51.32 70.49 72.25 72.04 41.54 69.21 51.72 80.65 68.73 33.72 67.34 43.7 47.61 47.4 70.28 75.56
112.8 112.3 277 265.7 97.2 106.4 97.9 45.7 190.8 108.3 179.5 193.3 34.3 46.4 72.9 72.2 118.4 108.3 97.8 23.2 112.6 32.7 160.4 116.3 80.8 227.4 34.1 88.7 88 110.9 111.3
the model. The error terms were assumed to be random and normally distributed with mean equal to zero and constant standard deviation. As for DCOD , a normal probability plot of the residuals was used to assess the validity of this assumption (Fig. 1A). It was shown that the error terms felled on a straight line on the normal probability plot and followed the normal distribution, which demonstrated that the RSM model (Eq. (4)) was very suitable to describe the COD degradation efficiency of this electrochemical oxidation process. The result was also valid to the Es model (Eq. (5)) (Fig. 1B). Pareto graphic of DCOD demonstrated a visual result which was the main factor to affect the COD degradation efficiency (Fig. 2). The factors of current density, retention time, frequency and pulse duty cycle were denoted as A, B, C and D, respectively. Bar represented the standardized effect of each involved factor. The figure indicated that the influence of retention time was the most significant variable affected DCOD within the four independent variables. The role order of factors was t > j > x > f. Filled bars are a graphical representation of positive-affecting factors, while unfilled bars indicated the contrary. A, B, C and D are all filled bars indicated that by varying the four independent variables DCOD increases. The obvious bar of the combination of j − x (denoted as AD) implied a significant interaction between current and pulse duty cycle which would greatly affect DCOD positively. The effect of each independent variable on DCOD was shown in Fig. 3. It was observed that the current density, retention time and pulse duty cycle affect COD degradation efficiency positively and significantly. Frequency hardly affect DCOD , which was explained that in a sufficiently large frequency, the phenol compounds diffusion were not able to follow the potential change [16,17]. As current density, retention time or pulse duty cycle raised, DCOD increased. However, it was noticeable that when pulse duty cycle was over ca. 50%, not significant DCOD increasing was observed. Therefore, continuous current supply probably hardly achieved higher COD degradation efficiency than a suitable pulse intermission current supply mode.
Fig. 1. Normal probability plots of residuals of COD degradation efficiency (A) and Es (B).
Specific energy consumption was affected significantly by the variables based on Table 2. Response surface graphic presented visual results of the primary variables influence on Es (Fig. 4). It was observed that current density, retention time and pulse duty cycle were the obvious variables to affect the specific energy consumption, while frequency almost shown no influence. These two response surface graphics (Fig. 4A and B) were all quite plain which implied that no interactions between any two variables inside the studied region. Lower specific energy consumption was obtained at smaller current density, shorter retention time and lower pulse duty cycle. To obtain satisfied DCOD and simultaneously reduce the Es , the influences of operative parameters on these two responses
Fig. 2. Pareto graphic: standardized effects on COD degradation efficiency. Current density, retention time, frequency and pulse duty cycle was denoted as A, B, C and D, respectively.
J. Wei et al. / Electrochimica Acta 56 (2011) 5310–5315
5313
Optimal conditions to simultaneously achieve expected DCOD and low Es were obtained from RSM models and calculated by Minitab 15 software. In this study, the initial COD concentration was 500 mg L−1 . In the view of National Discharge Class-1 of China, the discharge COD must lower than 100 mg L−1 , which indicated that 80% DCOD was as the desirability (COD was reduced from 500 mg L−1 to 100 mg L−1 ). Therefore, the optimal conditions were obtained taking into account 80% DCOD and the lowest Es , which was determined from RSM models. The calculation processes were finished by Minitab 15 software. The lowest Es with the value of 110 kWh kgCOD−1 was obtained at the optimized condition (The coded values of x1 , x2 , x3 and x4 were −0.31, 1.6, 2.0 and −0.35, respectively. Then the operative parameters were: j = 19.5 mA cm−2 , t = 69 min, f = 210 Hz, x = 43%). Fig. 3. The main effects on COD degradation efficiency: current density, retention time, frequency and pulse duty cycle.
were discussed in this context. As aforementioned results, the combination influence of current density and pulse duty cycle affected DCOD significantly. Therefore, it showed feasibility that through adjusting current density and pulse duty cycle, satisfied DCOD and low Es would be simultaneously achieved. Response surfaces of DCOD and Es with the current density and the pulse duty cycle were shown in Fig. 5, respectively. It was observed that the higher DCOD was obtained at higher current density and middle pulse duty cycle, while lower Es was achieved at lower current density and pulse duty cycle. The different influences of current density and pulse duty cycle on DCOD and Es demonstrated that coordinated pulse duty cycle and current density could optimize DCOD and Es in pulse-BDD anode system. It could be concluded, a relatively high DCOD and low Es were simultaneously obtained when pulse duty cycle was the value of ca. 50% in this study.
3.2. Comparison between pulse and constant current mode The COD degradation efficiency and specific energy consumption of the two modes, pulse-current mode and constant-current mode, were compared for electrochemical oxidation of phenol simulated wastewater. The values of DCOD and Es in pulse current mode were calculated using the above-obtained response functions. The results of phenol electrochemical degradation in constant current mode were shown in previous work [18]. The influence of current density and retention time on the performance of both systems was studied. Initial COD and conductivity of phenol simulated wastewater was 500 mg L−1 and 9 mS cm−1 , respectively. Frequency and pulse duty cycle of pulse current supply was 110 Hz and 50%, respectively. The trends of DCOD and Es as a function of current density (retention time kept constant of 45 min) were shown in Fig. 6A. COD degradation efficiency in constant-current mode was a little higher than that in pulse-current mode at the same applied current density. The specific energy consumption always increased linearly with rising current density in these two modes. However, the Es
Fig. 4. Response surface of Es in the pulse-BDD system vs. two shown factors: (A) current density (j) and retention time (t), (B) frequency (f) and pulse duty cycle (x).
Fig. 5. Response surfaces of DCOD (A) and Es (B) vs. two shown factors: current density (j) and pulse duty cycle (x).
5314
J. Wei et al. / Electrochimica Acta 56 (2011) 5310–5315
period of Ton oxidized the pollutant, and then the pollutant concentration near the anode surface was decreased gradually. Afterward, at the period of Toff , the organic compounds in vicinity of the anode surface were compensated because of the mass-transport. As a result, at the period of Ton , the generated • HO was combined with abundant reactants, and then side reactions such as • HO oxidation to O2 was weak. Therefore, the specific energy consumption in this mode was lower. On the other hand, in constant current mode, a relatively greater amount of hydroxyl radicals was probably wasted in side reactions due to the absence of abundant pollutants in the vicinity of anode surface, thus yielding high energy consumption. Additionally, concentration polarization was decreased due to the pulsating compensation of supporting electrolyte in pulse current mode. As a result, lower cell potential would be achieved and then followed by low energy consumption. 4. Conclusions
Fig. 6. Trends of DCOD and Es as a function of current density (A) and retention time (B). Solid line represents the DCOD and dash line represents the Es , respectively. Line 1 and line 3 are in constant current mode, line 2 and line 4 are in pulse current mode.
increased more slowly in pulse-BDD anode system, which resulted that pulse-BDD anode system exhibited satisfied COD degradation efficiency with lower specific energy consumption, especially at high current density. Take 22 mA cm−2 as an example, COD degradation efficiencies in pulse and constant current modes were 70.7% and 73.7%, respectively, while the specific energy consumptions were 110 kWh kgCOD−1 and 234 kWh kgCOD−1 , respectively. 53% energy saving was achieved in the pulse-BDD anode system compared to the constant-BDD anode system, though the DCOD was slightly lower. Kept the applied current density as a constant (J = 22 mA cm−2 ), the evolutions of DCOD and Es as a function of retention time were obtained (Fig. 6B). COD degradation efficiencies were similar at same retention time, and increased with rising the retention time in these two modes. The specific energy consumption also increased with rising retention time in these two systems, but the influence of retention time on Es in constant current mode was more apparent than that of pulse current mode. As a result, pulseBDD anode system exhibited lower specific energy consumption, especially at large retention time. The obtained comparison results demonstrated that the pulse-BDD anode system was superior to the constant-BDD anode system in energy saving for electrochemical oxidation of phenol. The main oxidation reaction at BDD electrode during bulk electrolysis was the indirect electrochemical oxidation mediated by hydroxyl radicals (• HO) [19–21]. Hydroxyl radicals with an instant lifetime (shorter than 10−9 s) were generated by the discharge of water and then reacted with organic compounds at the vicinity of anode surface. The reaction also competed with side reactions such as O2 evolution [22]. In pulse current mode, • HO generated at the
Pulse current supply was first conducted to a BDD anode system for electrochemical oxidation of phenol. The response surface methodology was used to evaluate the influence of current density, retention time, power frequency and pulse duty cycle on performances of the pulse-BDD anode system. As for DCOD and Es , the influences of current density, retention time and pulse duty cycle were obvious, while power frequency slightly affected these two responses. In the pulse current mode, a suitable pulse duty cycle would simultaneously obtain a relatively high COD degradation efficiency and low energy consumption. In the view of National Discharge Class-1 of China, 80% DCOD was as the desirability and then the lowest Es with the value of 110 kWh kgCOD−1 was obtained at the optimized condition determined from RSM models. By the comparison with constant current mode, an obvious specific energy consumption reduction was achieved in pulse current mode, though the DCOD was slightly decreased. At a similar level of COD degradation efficiency, pulse-BDD anode system shown lower specific energy consumption than constant-BDD anode system, which were attributed that pulse current resulted obvious reduction of side reactions and concentration polarization. This study demonstrated that pulse-BDD anode system was more efficient for minimizing the operative cost than constant-BDD anode system. Pulse current mode is an economically feasible technology in electrochemical oxidation process. Acknowledgment This research was funded by National Natural Science Foundation of China under grant No. 20877001. References [1] M. Hupert, A. Muck, J. Wang, S. Jason, Diamond Relat. Mater. 12 (2003) 1940. [2] J. Iniesta, P.A. Michaud, M. Panizza, G. Cerisola, A. Aldaz, Ch. Comninellis, Electrochim. Acta 46 (2001) 3573. [3] A. Paolo, D. Alain, B. Rabah, S. Sabine, Electrochem. Commun. 10 (2008) 402. [4] A. Carlos, H. Martínez, S. Ferro, Chem. Soc. Rev. 12 (2006) 1324. [5] C. Comninellis, Electrochim. Acta 39 (1994) 1857. [6] X.P. Zhu, S.Y. Shi, J.J. Wei, F.X. Lv, H.Z. Zhao, J.T. Kong, Q. He, J.R. Ni, Environ. Sci. Technol. 41 (2007) 6541. [7] A. Morao, A. Lopes, M.T. Pessoa de Amorim, I.C. Goncalves, Electrochim. Acta 49 (2004) 1587. [8] O. Scialdone, A. Galia, C. Guariso, S. Randazzo, G. Filardo, Electrochim. Acta 53 (2008) 2095. [9] N. Bensalah, H. Trabelsi, A. Gadri, J. Environ. Econ. Manage. 90 (2009) 523. [10] I. Sires, E. Brillas, G. Cerisola, M. Panizza, J. Electroanal. Chem. 613 (2008) 151. [11] B. Boye, E. Brillas, B. Marselli, P.A. Michaud, C. Comninellis, G. Farnia, G. Sandona, Electrochim. Acta 51 (2006) 2872. [12] P. Canizares, J. Garcia-Gomez, C. Saez, M.A. Rodrigo, J. Appl. Electrochem. 33 (2003) 917. [13] B. Marselli, J. Garcia-Gomez, P.A. Michaud, M.A. Rodrigo, Ch. Comninellis, J. Electrochem. Soc. 150 (2003) D79.
J. Wei et al. / Electrochimica Acta 56 (2011) 5310–5315 [14] [15] [16] [17] [18]
X.P. Zhu, J.R. Ni, H.N. Li, Y. Jiang, X. Xing, Electrochim. Acta 55 (2010) 5569. M. Panizza, A. Kapalka, C. Comninellis, Electrochim. Acta 53 (2008) 2289. C. Amatore, M. Berthou, S. Hebert, J. Electroanal. Chem. 457 (1998) 191. C. Polatides, M. Dortsiou, G. Kyriacou, Electrochim. Acta 50 (2005) 5237. X.P. Zhu, J.R. Ni, J.J. Wei, X. Xing, H.N. Li, Y. Jiang, J. Hazard. Mater. 184 (2010) 493.
5315
[19] X.P. Zhu, M.P. Tong, S.Y. Shi, H.Z. Zhao, J.R. Ni, Environ. Sci. Technol. 42 (2008) 4914. [20] M. Panizza, G. Cerisola, Electrochim. Acta 51 (2005) 191. [21] A. Kapalka, G. Foti, Ch. Comninellis, Electrochim. Acta 53 (2007) 1954. [22] X.M. Chen, F.R. Gao, G.H. Chen, J. Appl. Electrochem. 35 (2005) 185.
Contents lists available at ScienceDirect
Electrochimica Acta journal homepage: www.elsevier.com/locate/electacta
Electrochemical oxidation of phenol at boron-doped diamond electrode in pulse current mode Junjun Wei ∗ , Xiuping Zhu, Jinren Ni Department of Environmental Engineering, Peking University, The Key Laboratory of Water and Sediment Sciences, Ministry of Education, Beijing 100871, China
a r t i c l e
i n f o
Article history: Received 28 February 2011 Received in revised form 30 March 2011 Accepted 1 April 2011 Available online 12 April 2011 Key words: Boron-doped diamond electrode Phenol Electrochemical oxidation Pulse current Response surface methodology
a b s t r a c t In this study, a pulse current supply was initially used in a BDD anode system (pulse-BDD anode system) for electrochemical oxidation of phenol. The influences of operative parameters (current density, retention time, pulse duty cycle, power frequency) on the system performances were exmamined by response surface methodology (RSM). As for COD degradation efficiency (DCOD ) and specific energy consumption (Es ), the influence of retention time was more important than current density and pulse duty cycle, while power frequency hardly presented significant influence. By the comparison with constant current mode, an obvious specific energy consumption reduction was achieved in the pulse-BDD anode system, though the DCOD was slightly lower. The significant Es decrease might be attributed to the reduction of side reactions and concentration polarization in pulse current mode. The pulse-BDD anode system demonstrated an efficient technology to simultaneously obtain high pollutant degradation efficiency and low energy consumption. © 2011 Elsevier Ltd. All rights reserved.
1. Introduction Electrochemical oxidation is a promising technology to treat bio-refractory wastewaters [1,2]. The degradation effectiveness mainly relies on the nature of the electrodes, as well as reactor construction, electrolysis conditions, etc. Boron-doped diamond (BDD) anodic oxidation demonstrates many attractive electrochemical performances, such as strong oxidation stability, high current efficiency, weak electrode fouling, and long service life [3–6]. However, the BDD anode system is still not applied extensively. It is clear that a technically efficient process must also be economically feasible with regard to its initial capital and operating costs. Energy consumption is a very important economical parameter in electrochemical oxidation process like all other electrolytic processes. Conventional BDD anode oxidation systems usually run in the constant current condition [7–10]. For the sake of satisfied COD degradation efficiency (DCOD ) and no anode passivation, relatively high constant current density is required [9–12]. In such a case, side reactions (e.g., oxygen evolution, H2 O2 and O3 generation) were strengthened and resulted high specific energy consumption (Es ) and operating costs [13]. Enhancement of mass transport was proven to be efficient to optimize the electrochemical oxidation process. For example, ultrasound enhance will signifi-
cantly improve electrochemical oxidation and decrease the specific energy consumption [14]. However, the additional energy consumption of ultrasound system is high and the reactor is not easy to scale up. Improvement of the power supplement seems an alternate reasonable approach to reduce the Es . Panizza et al. [15] applied multiple current steps electrolysis and a semi-continuous current control mode in the BDD electrolysis and achieved the mineralization of organic pollutants with high reaction rate and current efficiency and low energy consumption, but this technology is too complex to be applied extensively. Therefore, we proposed a pulse mode for the electrochemical anode oxidation system. In this study, a pulse current supply was initially used in a BDD anode system (pulse-BDD anode system) for electrochemical oxidation of phenol. The influences of the operative parameters (current density (j), retention time (t), power frequency (f), pulse duty cycle (x)) on COD degradation efficiency (DCOD ) and specific energy consumption (Es ) were investigated and optimized by response surface methodology (RSM). Furthermore, the pulse and constant current modes were compared according to the response functions to evaluate the superiority of pulse-BDD anode system. 2. Experimental 2.1. Electrolytic cell and power supply
∗ Corresponding author. Tel.: +86 10 6275 4290; fax: +86 10 6275 6526. E-mail addresses: [email protected] (J. Wei), [email protected] (J. Ni). 0013-4686/$ – see front matter © 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.electacta.2011.04.006
The electrolytic experiments were performed in a cell with two parallel planar electrodes. A Nb/BDD (boron-doped diamond film
J. Wei et al. / Electrochimica Acta 56 (2011) 5310–5315
on niobium substrate) plate with a working area of 24 cm2 (CONDIAS GmbH, Germany) was used as the anode. The boron-doped diamond film (BDD) was synthesized by the hot filament chemical vapor deposition technique (HFCVD) on niobium substrate. The doping level of boron in the diamond layer expressed as B/C ration was about 3500 ppm and the obtained diamond film thickness was about 2 m with a resistivity of 10–30 m cm. A stainless steel piece with the same size was as the cathode. The interelectrode distance was 1.55 cm. Thus, the working volume of this cell was 37.2 cm3 . Pulse electrolysis was performed using a Pulse Current Supply (DH2002, Beijing Dahua Electron Co. Ltd., China) as power source in a BDD anode system. The output current wave of this power supply was square pulses wave. A pulse cycle T is comprised of Ton and Toff , which is defined as turn-on period and turn-off period, respectively. At the Ton period, current supplied as constant amplitude; while at Toff period, the current supply was intermitted. In this study, the power frequency is equal to 1/T in the range of 10 Hz to 500 Hz, and the pulse duty cycle is determined by Ton /T in the range of 0–100%. 2.2. Electrolysis experiment
5311
Table 1 Experimental range and levels of the process independent variables.
Independent variable Current density (mA cm−2 ) Retention time (min) Power frequency (Hz) Pulse duty cycle (%)
Factor
Range and levels
Xi X1 X2 X3 X4
−2 6 15 10 10
−1 14 30 60 30
0 22 45 110 50
1 30 60 160 70
2 38 75 210 90
where X0 is value of the Xi at the center point, and ıX presents the step change. Each response Y can be represented by a mathematical equation that correlates the response surface. Y = b0 + b1 x1 + b2 x2 + b3 x3 + b4 x4 + b11 x12 + b22 x22 + b33 x32 + b44 x42 + b12 x1 x2 + b13 x1 x3 + b14 x1 x4 + b23 x2 x3 + b24 x2 x4 + b34 x3 x4 (2) where Y is the predicted response, b0 is constant, b1 , b2 , b3 , and b4 are the linear effects, b11 , b22 , b33 , and b44 are quadratic coefficients, and b12 , b13 , b14 , b23 , b24 , and b34 are interaction coefficients. In this context, DCOD and Es were as the responses. The COD (in mg L−1 ) of the solutions were measured by a titrimetric method using dichromate as the oxidant in acidic solution at 150 ◦ C for 2 h (Hachi, USA). For pulse current mode, the Es (in kWh kgCOD−1 ) was calculated using the following equation:
Phenol was chosen as the studied object, because it is a common and typical organic pollutant in many industrial effluents. The initial COD concentration was maintained the value of 500 mg/L. Na2 SO4 was used as the supporting electrolyte and the concentration was 0.05 M. Electrolysis experiments were conducted at different current density (j), retention time (t), power frequency (f), and pulse duty cycle (x). During electrochemical oxidation process, phenol simulated wastewater was continuously pumped through the electrolytic cell at a certain flow rate (determined by retention time).
where U is the cell potential (in V), I is the applied current, (in A), t is the retention time (in h), x is the pulse duty cycle of applied current supply, (in %), CODin and CODout are the input and output COD concentration, respectively, (in mg L−1 ), and V is the cell volume (in mL).
2.3. Experiment design and analysis
3. Results and discussion
Design of experiments (DOE) is a powerful tool to study the effect of variables and their responses with minimum number of experiments. Response surface methodology (RSM) is one of DOEs which can be a useful tool to examine the influence of operative parameters on system performances and to determine the optimum conditions. The most popular class of second order designs called central composite design (CCD) was used for RSM. Because the CCD is ideal for sequential experimentation and allows a reasonable amount of information for testing lack of fit while not involving an unusually large number of design point. In this study, RSM was initially used to describe and optimize the electrochemical oxidation process of Pulse-BDD anode system. The CCD with four factors at five levels was applied for the experimental design. The current density (coded as X1 ), retention time (coded as X2 ), power frequency (coded as X3 ) and pulse duty cycle (coded as X4 ) were chosen as independent variables. Each variable was coded at five levels between −2 and 2 at the range determined by the preliminary experiments and changed in the ranges of 6–38 mA cm−2 , 15–75 min, 10–210 Hz, and 10–90%, respectively. The variables design was shown in Table 1. According to the design of RSM in terms of Minitab 15 software, thirty-one experiments were augmented with seven replications at the design center to evaluate the pure error. For statistical calculations, the variables Xi were coded as xi according to the following relationship:
3.1. Influences of operative parameters
xi =
Xi − X0 ıX
(1)
Es =
1000UItx (CODin − CODout )V
(3)
Thirty-one CCD designed batch runs were conducted to visualize the effects of independent factors (j, t, f and x) on Pulse-BDD anode system responses (DCOD and Es ). The results along with the experimental conditions were presented in Table 2. The DCOD was in the range of 33.72–81.82% and the Es was in 23.2–277 kWh kgCOD−1 . Application of RSM offers, the relationships of DCOD and Es with independent studied variables were obtained as shown in Eqs. (4) and (5), respectively. Y (DCOD ) = 70.7171 + 4.8125x1 +11.3975x2 + 1.3208x3 + 3.9175x4 − 2.2999x12 − 2.764x22 − 0.2574x32 − 3.4499x42 − 0.4788x1 x2 − 0.2825x1 x3 + 2.1875x1 x4 − 0.3088x2 x3 + 0.2088x2 x4 − 0.135x3 x4
(R2 = 0.9407)
(4)
Y (Es ) = 110.224 + 41.8152x1 + 19.0435x2 + 1.07x3 + 51.3014x4 + 0.1347x12 + 2.1642x22 − 1.198x32 + 3.342x42 + 8.8574x1 x2 + 0.2283x1 x3 + 20.5113x1 x4 + 1.2404x2 x3 + 9.24x2 x4 − 0.4015x3 x4
(R2 = 0.9920)
(5)
The correlation factors (R2 ) were respectively equal to 0.9407 and 0.9920 indicating a good fit for the dependent variables. The residuals were the deviations of the experimental data value from the predicted values and were estimates of the error terms in
5312
J. Wei et al. / Electrochimica Acta 56 (2011) 5310–5315
Table 2 Experimental planning in DOE and obtained responses in each experiment. Run
j (mA cm−2 )
t (min)
f (Hz)
x (%)
DCOD (%)
Es (kWh kgCOD−1 )
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31
22 22 30 30 30 22 22 14 38 22 30 30 14 14 30 30 14 22 30 22 22 14 22 22 22 22 6 14 14 22 14
45 45 60 60 60 45 45 60 45 45 30 30 30 60 30 30 60 45 60 45 45 30 75 45 15 45 45 30 30 45 60
110 110 160 60 160 110 10 60 110 110 160 60 60 160 60 110 60 110 60 110 110 160 110 210 110 110 110 160 60 110 160
50 50 70 70 30 50 50 30 50 50 70 70 30 30 30 30 70 50 30 10 50 30 50 50 50 90 50 70 70 50 70
69.9 69.1 81.23 81.82 74.19 72.43 65.69 68.14 74.38 71.85 64.81 57.77 45.64 71.85 49.37 51.32 70.49 72.25 72.04 41.54 69.21 51.72 80.65 68.73 33.72 67.34 43.7 47.61 47.4 70.28 75.56
112.8 112.3 277 265.7 97.2 106.4 97.9 45.7 190.8 108.3 179.5 193.3 34.3 46.4 72.9 72.2 118.4 108.3 97.8 23.2 112.6 32.7 160.4 116.3 80.8 227.4 34.1 88.7 88 110.9 111.3
the model. The error terms were assumed to be random and normally distributed with mean equal to zero and constant standard deviation. As for DCOD , a normal probability plot of the residuals was used to assess the validity of this assumption (Fig. 1A). It was shown that the error terms felled on a straight line on the normal probability plot and followed the normal distribution, which demonstrated that the RSM model (Eq. (4)) was very suitable to describe the COD degradation efficiency of this electrochemical oxidation process. The result was also valid to the Es model (Eq. (5)) (Fig. 1B). Pareto graphic of DCOD demonstrated a visual result which was the main factor to affect the COD degradation efficiency (Fig. 2). The factors of current density, retention time, frequency and pulse duty cycle were denoted as A, B, C and D, respectively. Bar represented the standardized effect of each involved factor. The figure indicated that the influence of retention time was the most significant variable affected DCOD within the four independent variables. The role order of factors was t > j > x > f. Filled bars are a graphical representation of positive-affecting factors, while unfilled bars indicated the contrary. A, B, C and D are all filled bars indicated that by varying the four independent variables DCOD increases. The obvious bar of the combination of j − x (denoted as AD) implied a significant interaction between current and pulse duty cycle which would greatly affect DCOD positively. The effect of each independent variable on DCOD was shown in Fig. 3. It was observed that the current density, retention time and pulse duty cycle affect COD degradation efficiency positively and significantly. Frequency hardly affect DCOD , which was explained that in a sufficiently large frequency, the phenol compounds diffusion were not able to follow the potential change [16,17]. As current density, retention time or pulse duty cycle raised, DCOD increased. However, it was noticeable that when pulse duty cycle was over ca. 50%, not significant DCOD increasing was observed. Therefore, continuous current supply probably hardly achieved higher COD degradation efficiency than a suitable pulse intermission current supply mode.
Fig. 1. Normal probability plots of residuals of COD degradation efficiency (A) and Es (B).
Specific energy consumption was affected significantly by the variables based on Table 2. Response surface graphic presented visual results of the primary variables influence on Es (Fig. 4). It was observed that current density, retention time and pulse duty cycle were the obvious variables to affect the specific energy consumption, while frequency almost shown no influence. These two response surface graphics (Fig. 4A and B) were all quite plain which implied that no interactions between any two variables inside the studied region. Lower specific energy consumption was obtained at smaller current density, shorter retention time and lower pulse duty cycle. To obtain satisfied DCOD and simultaneously reduce the Es , the influences of operative parameters on these two responses
Fig. 2. Pareto graphic: standardized effects on COD degradation efficiency. Current density, retention time, frequency and pulse duty cycle was denoted as A, B, C and D, respectively.
J. Wei et al. / Electrochimica Acta 56 (2011) 5310–5315
5313
Optimal conditions to simultaneously achieve expected DCOD and low Es were obtained from RSM models and calculated by Minitab 15 software. In this study, the initial COD concentration was 500 mg L−1 . In the view of National Discharge Class-1 of China, the discharge COD must lower than 100 mg L−1 , which indicated that 80% DCOD was as the desirability (COD was reduced from 500 mg L−1 to 100 mg L−1 ). Therefore, the optimal conditions were obtained taking into account 80% DCOD and the lowest Es , which was determined from RSM models. The calculation processes were finished by Minitab 15 software. The lowest Es with the value of 110 kWh kgCOD−1 was obtained at the optimized condition (The coded values of x1 , x2 , x3 and x4 were −0.31, 1.6, 2.0 and −0.35, respectively. Then the operative parameters were: j = 19.5 mA cm−2 , t = 69 min, f = 210 Hz, x = 43%). Fig. 3. The main effects on COD degradation efficiency: current density, retention time, frequency and pulse duty cycle.
were discussed in this context. As aforementioned results, the combination influence of current density and pulse duty cycle affected DCOD significantly. Therefore, it showed feasibility that through adjusting current density and pulse duty cycle, satisfied DCOD and low Es would be simultaneously achieved. Response surfaces of DCOD and Es with the current density and the pulse duty cycle were shown in Fig. 5, respectively. It was observed that the higher DCOD was obtained at higher current density and middle pulse duty cycle, while lower Es was achieved at lower current density and pulse duty cycle. The different influences of current density and pulse duty cycle on DCOD and Es demonstrated that coordinated pulse duty cycle and current density could optimize DCOD and Es in pulse-BDD anode system. It could be concluded, a relatively high DCOD and low Es were simultaneously obtained when pulse duty cycle was the value of ca. 50% in this study.
3.2. Comparison between pulse and constant current mode The COD degradation efficiency and specific energy consumption of the two modes, pulse-current mode and constant-current mode, were compared for electrochemical oxidation of phenol simulated wastewater. The values of DCOD and Es in pulse current mode were calculated using the above-obtained response functions. The results of phenol electrochemical degradation in constant current mode were shown in previous work [18]. The influence of current density and retention time on the performance of both systems was studied. Initial COD and conductivity of phenol simulated wastewater was 500 mg L−1 and 9 mS cm−1 , respectively. Frequency and pulse duty cycle of pulse current supply was 110 Hz and 50%, respectively. The trends of DCOD and Es as a function of current density (retention time kept constant of 45 min) were shown in Fig. 6A. COD degradation efficiency in constant-current mode was a little higher than that in pulse-current mode at the same applied current density. The specific energy consumption always increased linearly with rising current density in these two modes. However, the Es
Fig. 4. Response surface of Es in the pulse-BDD system vs. two shown factors: (A) current density (j) and retention time (t), (B) frequency (f) and pulse duty cycle (x).
Fig. 5. Response surfaces of DCOD (A) and Es (B) vs. two shown factors: current density (j) and pulse duty cycle (x).
5314
J. Wei et al. / Electrochimica Acta 56 (2011) 5310–5315
period of Ton oxidized the pollutant, and then the pollutant concentration near the anode surface was decreased gradually. Afterward, at the period of Toff , the organic compounds in vicinity of the anode surface were compensated because of the mass-transport. As a result, at the period of Ton , the generated • HO was combined with abundant reactants, and then side reactions such as • HO oxidation to O2 was weak. Therefore, the specific energy consumption in this mode was lower. On the other hand, in constant current mode, a relatively greater amount of hydroxyl radicals was probably wasted in side reactions due to the absence of abundant pollutants in the vicinity of anode surface, thus yielding high energy consumption. Additionally, concentration polarization was decreased due to the pulsating compensation of supporting electrolyte in pulse current mode. As a result, lower cell potential would be achieved and then followed by low energy consumption. 4. Conclusions
Fig. 6. Trends of DCOD and Es as a function of current density (A) and retention time (B). Solid line represents the DCOD and dash line represents the Es , respectively. Line 1 and line 3 are in constant current mode, line 2 and line 4 are in pulse current mode.
increased more slowly in pulse-BDD anode system, which resulted that pulse-BDD anode system exhibited satisfied COD degradation efficiency with lower specific energy consumption, especially at high current density. Take 22 mA cm−2 as an example, COD degradation efficiencies in pulse and constant current modes were 70.7% and 73.7%, respectively, while the specific energy consumptions were 110 kWh kgCOD−1 and 234 kWh kgCOD−1 , respectively. 53% energy saving was achieved in the pulse-BDD anode system compared to the constant-BDD anode system, though the DCOD was slightly lower. Kept the applied current density as a constant (J = 22 mA cm−2 ), the evolutions of DCOD and Es as a function of retention time were obtained (Fig. 6B). COD degradation efficiencies were similar at same retention time, and increased with rising the retention time in these two modes. The specific energy consumption also increased with rising retention time in these two systems, but the influence of retention time on Es in constant current mode was more apparent than that of pulse current mode. As a result, pulseBDD anode system exhibited lower specific energy consumption, especially at large retention time. The obtained comparison results demonstrated that the pulse-BDD anode system was superior to the constant-BDD anode system in energy saving for electrochemical oxidation of phenol. The main oxidation reaction at BDD electrode during bulk electrolysis was the indirect electrochemical oxidation mediated by hydroxyl radicals (• HO) [19–21]. Hydroxyl radicals with an instant lifetime (shorter than 10−9 s) were generated by the discharge of water and then reacted with organic compounds at the vicinity of anode surface. The reaction also competed with side reactions such as O2 evolution [22]. In pulse current mode, • HO generated at the
Pulse current supply was first conducted to a BDD anode system for electrochemical oxidation of phenol. The response surface methodology was used to evaluate the influence of current density, retention time, power frequency and pulse duty cycle on performances of the pulse-BDD anode system. As for DCOD and Es , the influences of current density, retention time and pulse duty cycle were obvious, while power frequency slightly affected these two responses. In the pulse current mode, a suitable pulse duty cycle would simultaneously obtain a relatively high COD degradation efficiency and low energy consumption. In the view of National Discharge Class-1 of China, 80% DCOD was as the desirability and then the lowest Es with the value of 110 kWh kgCOD−1 was obtained at the optimized condition determined from RSM models. By the comparison with constant current mode, an obvious specific energy consumption reduction was achieved in pulse current mode, though the DCOD was slightly decreased. At a similar level of COD degradation efficiency, pulse-BDD anode system shown lower specific energy consumption than constant-BDD anode system, which were attributed that pulse current resulted obvious reduction of side reactions and concentration polarization. This study demonstrated that pulse-BDD anode system was more efficient for minimizing the operative cost than constant-BDD anode system. Pulse current mode is an economically feasible technology in electrochemical oxidation process. Acknowledgment This research was funded by National Natural Science Foundation of China under grant No. 20877001. References [1] M. Hupert, A. Muck, J. Wang, S. Jason, Diamond Relat. Mater. 12 (2003) 1940. [2] J. Iniesta, P.A. Michaud, M. Panizza, G. Cerisola, A. Aldaz, Ch. Comninellis, Electrochim. Acta 46 (2001) 3573. [3] A. Paolo, D. Alain, B. Rabah, S. Sabine, Electrochem. Commun. 10 (2008) 402. [4] A. Carlos, H. Martínez, S. Ferro, Chem. Soc. Rev. 12 (2006) 1324. [5] C. Comninellis, Electrochim. Acta 39 (1994) 1857. [6] X.P. Zhu, S.Y. Shi, J.J. Wei, F.X. Lv, H.Z. Zhao, J.T. Kong, Q. He, J.R. Ni, Environ. Sci. Technol. 41 (2007) 6541. [7] A. Morao, A. Lopes, M.T. Pessoa de Amorim, I.C. Goncalves, Electrochim. Acta 49 (2004) 1587. [8] O. Scialdone, A. Galia, C. Guariso, S. Randazzo, G. Filardo, Electrochim. Acta 53 (2008) 2095. [9] N. Bensalah, H. Trabelsi, A. Gadri, J. Environ. Econ. Manage. 90 (2009) 523. [10] I. Sires, E. Brillas, G. Cerisola, M. Panizza, J. Electroanal. Chem. 613 (2008) 151. [11] B. Boye, E. Brillas, B. Marselli, P.A. Michaud, C. Comninellis, G. Farnia, G. Sandona, Electrochim. Acta 51 (2006) 2872. [12] P. Canizares, J. Garcia-Gomez, C. Saez, M.A. Rodrigo, J. Appl. Electrochem. 33 (2003) 917. [13] B. Marselli, J. Garcia-Gomez, P.A. Michaud, M.A. Rodrigo, Ch. Comninellis, J. Electrochem. Soc. 150 (2003) D79.
J. Wei et al. / Electrochimica Acta 56 (2011) 5310–5315 [14] [15] [16] [17] [18]
X.P. Zhu, J.R. Ni, H.N. Li, Y. Jiang, X. Xing, Electrochim. Acta 55 (2010) 5569. M. Panizza, A. Kapalka, C. Comninellis, Electrochim. Acta 53 (2008) 2289. C. Amatore, M. Berthou, S. Hebert, J. Electroanal. Chem. 457 (1998) 191. C. Polatides, M. Dortsiou, G. Kyriacou, Electrochim. Acta 50 (2005) 5237. X.P. Zhu, J.R. Ni, J.J. Wei, X. Xing, H.N. Li, Y. Jiang, J. Hazard. Mater. 184 (2010) 493.
5315
[19] X.P. Zhu, M.P. Tong, S.Y. Shi, H.Z. Zhao, J.R. Ni, Environ. Sci. Technol. 42 (2008) 4914. [20] M. Panizza, G. Cerisola, Electrochim. Acta 51 (2005) 191. [21] A. Kapalka, G. Foti, Ch. Comninellis, Electrochim. Acta 53 (2007) 1954. [22] X.M. Chen, F.R. Gao, G.H. Chen, J. Appl. Electrochem. 35 (2005) 185.