- Email: [email protected]
Plasma-based water treatment: Conception and application of a new general principle for reactor design
Chemical Engineering Journal 273 (2015) 543–550
Contents lists available at ScienceDirect
Chemical Engineering Journal journal homepage: www.elsevier.com/locate/cej
Plasma-based water treatment: Conception and application of a new general principle for reactor design Gunnar R. Stratton a, Christopher L. Bellona b, Fei Dai b, Thomas M. Holsen b, Selma Mededovic Thagard a,⇑ a b
Clarkson University, Plasma Research Laboratory, Department of Chemical and Biomolecular Engineering, 8 Clarkson Avenue, Potsdam, NY 13699, USA Clarkson University, Department of Civil and Environmental Engineering, 8 Clarkson Avenue, Potsdam, NY 13699, USA
h i g h l i g h t s Basic guidelines for plasma reactor design and optimization have been developed. Rhodamine B degradation depends on the area of the plasma–liquid interface. Reactor performance can be improved by generating foam on the liquid surface.
a r t i c l e
i n f o
Article history: Received 5 February 2015 Received in revised form 13 March 2015 Accepted 15 March 2015 Available online 20 March 2015 Keywords: AOP Plasma Pseudo-first-order Reactor design Rhodamine B Water treatment
a b s t r a c t To improve the feasibility of plasma-based water treatment technology and develop basic guidelines for reactor design and optimization, a study was conducted to identify and characterize design parameters and physical phenomena that influence treatment efficiency. The first phase of the study established that the chemical reactions responsible for the degradation of organic solutes can be more accurately represented, mathematically, as heterogeneous reactions, rather than the common representation as homogeneous reactions. Using Rhodamine B as the model solute, the observed removal rate constant was found to be proportional to the area of the plasma–liquid interface. This observation supported the validity of the proposed heterogeneous rate equation and inspired the conception of a general design principle, which prescribes maximizing contact between the plasma and the treated solution. The second phase of the study involved the application of this design principle to create seven different ‘‘contact-oriented’’ reactors. The design parameters employed to increase contact included feeding liquid streams directly through the discharge region and generating a layer of foam on the liquid surface. The contact-oriented reactors validated their founding principle by achieving removal efficiencies up to 145 times that for the reference case (point-plate with discharge in liquid). The removal efficiencies attained in this work compare quite favorably with those achieved by other advanced oxidation processes for the degradation of Rhodamine B. Ó 2015 Elsevier B.V. All rights reserved.
1. Introduction Advanced oxidation processes (AOPs) have increasingly been under investigation for their applications in the treatment of chemically and biologically contaminated water [1,2]. Some of these processes have reached a high level of development and have been commercialized, such as ultraviolet (UV)/hydrogen peroxide (H2O2) treatment and ozonation (O3) [3,4]. Pulsed electrical discharge plasma formed directly in or above water constitutes another, less-developed, AOP, but is receiving increased attention from researchers due to its unique treatment capabilities. Like ⇑ Corresponding author. Tel.: +1 315 268 4423; fax: +1 315 268 6654. E-mail address: [email protected] (S.M. Thagard). http://dx.doi.org/10.1016/j.cej.2015.03.059 1385-8947/Ó 2015 Elsevier B.V. All rights reserved.
other AOPs, plasma-based water treatment (PWT) makes use of the highly oxidative hydroxyl (OH) radical to oxidize chemical contaminants, and thus does not yield the hazardous byproducts that may be produced by conventional chlorination processes [1,5]. PWT involves the generation of OH radicals in situ, and from the water itself; therefore, unlike most AOPs and conventional processes, little to no chemical additives are required. Additionally, plasma offers a broader range of chemical and physical treatment mechanisms. Besides producing several reactive chemical species (OH, O, H, HO2, O 2 , O3, H2O2, H2), the plasma channels formed directly in water can reach temperatures of up to several thousands of Kelvin and thermally degrade molecules, emit UV and visible light, and generate shockwaves capable of inducing cavitation [6,7]. While these features make plasma attractive as a
544
G.R. Stratton et al. / Chemical Engineering Journal 273 (2015) 543–550
stand-alone technology, they also open up the possibility of combining plasma with other AOPs. For example, the generation of UV light allows plasma to be effectively integrated with photocatalysis [8–10]. While the potential advantages of PWT are evident, the technology has not yet reached a high enough level of development to be used in practice. Some of the obstacles that have hindered progress in developing PWT systems are associated with the absence of general principles to guide the design of new reactors with higher efficiency. The most expansive set of design criteria were offered by Malik in his review, where several different plasma reactors were compared in terms of treatment efficiency [11]. This review provided insight into which of the existing PWT systems performed best, and offered explanations for which process parameters contributed to higher treatment efficiencies. While there is much value in knowing which of the existing process parameters are most effective, it does not provide assistance to the development of new reactor designs. The present study investigates how selected reactor design parameters affect removal efficiency of Rhodamine B (RhB) dye, and uses the resulting relationships to originate a more universal guideline for developing new reactors. In addition to providing insights into the fundamental nature of the overall mechanism responsible for organic solute degradation, this guideline is employed, to considerable benefit, for the development of several novel reactor types.
2. Experimental The high voltage (HV) pulsed power supply used to generate the plasma was a custom built unit (Applied Physical Electronics LC, Spicewood, TX), with a general circuit diagram as shown in Fig. 1. The operational parameters used for this study were 20 kV discharges at 43 Hz, using a 0.94 nF load capacitor. The voltage and current in the plasma reactor were measured using a Tektronix P6015A high voltage probe and a Tektronix P6021 current probe connected to a Tektronix TDS 3032C oscilloscope. Examples of the voltage and current waveforms for discharges in both liquid and gas are shown in Fig. 2. All reactors (Fig. 3) consisted of a 17.3 cm diameter glass vessel fitted with an airtight polymer cap, which was adapted to allow for sample extraction, solution recirculation, and integration of the electrodes. The reactors were operated in semi-batch mode, with liquid recirculating at 1.4 L/min. The liquid recirculation loop ensured thorough mixing and included a heat exchanger to keep the solution at 15 °C. The headspace was purged with argon at 2.1 L/min. All reactors featured the point-plate electrode configuration, and differed in the size, shape, type and position of each electrode, as well as the mode by which the liquid feed stream entered the reactor. The ‘‘liquid discharge’’ reactor (Fig. 3(a)) contained a HV nickel chromium (NiCr) point electrode and a grounded stainless steel (S.S.) plate electrode both in the liquid phase with 1.5 cm
Rotary Spark Gap
Charge Resistor
HVDC Power Supply
spacing. The ‘‘gas discharge’’ reactor (Fig. 3(b)) featured the grounded S.S. plate in the liquid and the HV NiCr point in the gas with 2.7 cm spacing (1.5 cm in liquid, 1.2 cm in gas). Four different sized S.S. plates were used: 0.64, 3.15, 4.75 and 7.60 cm. For all of the ‘‘contact-oriented’’ reactors (Fig. 3(c)–(i)), the electrode positioning was similar to the gas discharge reactor, except a grounded aluminum ring (9.8 cm outer diameter, 6.0 cm inner diameter) was used in place of the grounded S.S. plate. In the turbulent jet reactor (Fig. 3(c)), the liquid feed was sprayed over the HV NiCr point with a 90° spray angle. The laminar jet reactor (Fig. 3(d)) was configured to introduce the liquid feed through a S.S. tube (0.32 cm inner diameter), which also served as the HV electrode. In the RVC HV reactor (Fig. 3(e)), the liquid feed passed through a reticulated vitreous carbon (RVC) plate, which served as the HV electrode. All of the ‘‘bubbling’’ reactors (Fig. 3(f)–(i)) were the same as the corresponding non-bubbling reactors, but involved bubbling argon (2.1 L/min) through a gas diffuser (9.5 cm diameter) centered beneath the hole in the ring electrode. The design characteristics of all the reactors are summarized in Table 1. The treated solution was deionized water containing 7.5 mg/L (0.0157 mM) RhB with sodium chloride added to adjust the conductivity to 300 lS/cm. The pH was 5.4 for the untreated solution, and averaged 5.0 for the treated solution. The conductivity of the treated solution was 300–330 lS/cm. Due to variations in characteristics of the reactors, the solution volume was not the same for all experiments, ranging from 850 to 1350 mL; however, when appropriate, the reported results have been linearly scaled to represent a solution volume of 600 mL (chosen arbitrarily). The concentration of the dye was determined spectrophotometrically (Shimadzu UV-1800) by measuring its absorbance at 554 nm. The concentration of H2O2 was determined spectrophotometrically, using the reaction between H2O2 and titanium sulfate and measuring the absorbance of the resulting yellow complex at 410 nm [12].
3. Results and discussion 3.1. Liquid and gas discharge reactors The study began by using two of the most common reactor types, the point-plate reactor with discharges directly in liquid (Fig. 3(a)) and discharges in gas (Fig. 3(b)). As shown in Fig. 4, the performance of the liquid discharge reactor is poor, though the removal efficiency (discussed in Section 3.6) is of the same order of magnitude as those reported for this reactor type in other plasma studies [11]. The performance of the gas discharge reactor (Fig. 4) was significantly better, with an observed rate constant (discussed in the following section) approximately 16 times higher than that for the liquid discharge. 3.2. Solute degradation kinetics To determine the reason behind the superior performance of the gas discharge compared to that of the liquid discharge, the nature of the overall degradation mechanism was evaluated. The shape of the concentration curves shown in Fig. 4 suggests first order kinetics, which is described by the following rate equation:
Load Capacitor
Fig. 1. Circuit diagram for pulsed power supply.
Plasma Reactor
dC S ¼ kobs C S dt
ð1Þ
where CS (mg L1) is the solute concentration (RhB in this case), t (min) is the treatment time, and kobs (min1) is the observed firstorder rate constant. By integrating Eq. (1) and rearranging, an equation is obtained to calculate the observed rate constant from the concentration data:
G.R. Stratton et al. / Chemical Engineering Journal 273 (2015) 543–550
545
Fig. 2. (a) Voltage waveform for liquid discharge, (b) current waveform for liquid discharge, (c) voltage waveform for gas discharge and (d) current waveform for gas discharge.
ln
C S;0 ¼ kobs t CS
ð2Þ
where CS,0 (mg L1) is the initial solute concentration. Eq. (1) is often interpreted as directly representing the pseudo-first-order reaction between the organic solute and OH radicals [13–15]:
Solute þ OH ! Products
ð3Þ
with kobs treated as the pseudo-first-order reaction rate constant (kobs ¼ k C OH , where k is the second-order reaction rate constant between the solute and OH radicals). However, this interpretation requires the assumption that the reaction is homogeneous, which is doubtful since OH radicals exist only in the immediate vicinity of the plasma channel (the interfacial zone). Thus, the solute degradation will likely only occur in the interfacial zone rather than within the bulk liquid. While RhB is photosensitive, it is not expected that photodegradation occurs to any significant degree, due to the relative lack of emissions in the range of wavelengths in which RhB absorbs (450–600 nm) [7]. Considering the non-homogeneous nature of this system, a more suitable kinetic model would be that for a heterogeneous irreversible pseudo-first-order reaction [16]:
dC S A ¼ 0 C S ¼ kobs C S dt V 1=km þ 1=k
ð4Þ
where A (m2) is the interfacial area, V (m3) is the solution volume, k0 (m s1) is the pseudo-first-order reaction rate constant 0 (k ¼ k C OH ), and km (m s1) is the mass transfer coefficient of the solute (km = D/L, where D (m2 s1) is the solute diffusivity and L (m) is the thickness of the interfacial zone). Note that the fractional term in Eq. (4) is equal to the measured kobs.
One way to support the validity of the proposed rate equation is to show that kobs is proportional to A, which requires that the interfacial area be varied while keeping km and k0 constant. Recalling that km = D/L, it is reasonable to assume that D is independent of A, as it is primarily determined by the characteristics of the particular solute. L has been defined as the thickness of the interfacial zone, which is effectively determined by the diffusion distance of the active species (i.e., OH radicals). The diffusion distance (estimated to be 6–20 nm) is primarily influenced by the concentration of OH radicals (see discussion in Section 3.4) and the density of the liquid, which is not likely to change dramatically with variations in A [6,17]. k0 varies with the OH radical concentration, which could also depend on A (see Section 3.4).
3.3. Plasma area Interfacial area was varied by using the gas discharge reactor with grounded plate electrodes of different diameters, as shown in Fig. 5. The leaders (plasma channels) generated in the gas discharge reactor propagate radially out over the liquid surface. As the leaders propagate across the liquid surface, they branch in such a way that the distance between leaders remains relatively constant, due to electrostatic repulsion. The fully propagated plasma occupies a roughly circular region on the liquid surface with visually uniform leader distribution [18–20]. The length of the leaders and therefore the area of this circular region, which will be referred to as the plasma area, increase with the diameter of the grounded plate. Because it is very difficult to define and measure the actual area of the plasma–liquid interface (A in Eq. (4)), plasma area was investigated as a potential surrogate. The plasma area was calculated for each plate size by recording images of the discharges and measuring the average length of the leaders.
546
G.R. Stratton et al. / Chemical Engineering Journal 273 (2015) 543–550
HV
HV
(a)
(b)
HV
HV
(c)
HV
HV
Argon
(d)
(e)
HV
Argon
(f)
HV
HV
Argon
Argon
(g)
(h)
(i)
Fig. 3. Reactor diagrams: (a) point-plate with discharge in liquid, (b) point-plate with discharge in gas, (c) turbulent jet, (d) laminar jet, (e) RVC HV, (f) point-plate with discharge in gas with bubbling, (g) turbulent jet with bubbling, (h) laminar jet with bubbling, and (i) RVC HV with bubbling.
Leaders are characterized by a very low internal electrical resistance, which results in a negligible voltage drop, thus the voltage gradient across the plasma–liquid interface is independent of the position within the plasma area. This uniform voltage gradient coupled with the uniform conductivity of the water produces a uniform current density, which means that the discharge current is directly proportional to the area of the plasma–liquid interface. It is clear from Fig. 6 that the relationship between the peak discharge current (Imax,meas) and plasma area is not perfectly linear. This non-linearity is because a higher discharge current corresponds to a higher discharge rate and therefore a higher rate of reduction in the discharge voltage. Consequently, the discharge voltage at the time that corresponds to Imax,meas decreases with increasing plate size, causing the Imax,meas vs. plasma area curve to level off for larger plate sizes. The effect of this varying voltage was accounted for by comparing Imax,meas to the theoretical maximum current (Imax,thry) calculated from Ohm’s law (Table 2):
Imax;thry ¼
U r Ap d
ð5Þ
where r (300 lS cm1) is the solution conductivity, d (1.5 cm) is the thickness of the liquid layer separating plasma from the grounded electrode, Ap (cm2) is the plasma area and U (V) is the measured discharge voltage at the time that corresponds to Imax,meas. Because the plasma area is greater than the actual interfacial area, Imax,thry is greater than Imax,meas. However, the ratio of Imax,meas to Imax,thry varies by less than five percent (95% confidence interval) for all plate sizes, which confirms that the variations in U are primarily responsible for the non-linearity seen in Fig. 6. Additionally, the uniformity in the ratio of Imax,meas to Imax,thry indicates that the ratio of Ap to A must also be consistent between plate sizes, which confirms that Ap is a suitable surrogate for A. As shown in Fig. 7, kobs exhibits a strong linear relationship with the plasma area, and thus kobs must be proportional to A, which supports the validity of Eq. (4).
547
G.R. Stratton et al. / Chemical Engineering Journal 273 (2015) 543–550 Table 1 Summary of reactor design characteristics. Reactor type
Corresponding figure
Liquid discharge Gas discharge Gas discharge Turbulent jet Laminar jet RVC HV Gas discharge with bubbling Turbulent jet with bubbling Laminar jet with bubbling RVC HV with bubbling * ** ***
Fig. Fig. Fig. Fig. Fig. Fig. Fig. Fig. Fig. Fig.
HV electrode *
3(a) 3(b) 3(b) 3(c) 3(d) 3(e) 3(f) 3(g) 3(h) 3(i)
NiCr wire NiCr wire NiCr wire NiCr wire S.S. tube RVC plate NiCr wire NiCr wire S.S tube RVC plate
HV diameter (mm)
Ground electrode
Ground diameter (cm)
Solution volume (mL)
0.81 0.81 0.81 0.81 2.2** 25 0.81 0.81 2.2 25
S.S plate S.S plate Al ring Al ring Al ring Al ring Al ring Al ring Al ring Al ring
3.15 0.64, 3.15, 4.75 and 7.60 9.8/6*** 9.8/6 9.8/6 9.8/6 9.8/6 9.8/6 9.8/6 9.8/6
1000 900, 880, 870 and 850 1350 1350 1350 1350 1350 1350 1350 1350
Wire was sharpened. 2.2 mm is the tube’s inner diameter. 9.8/6 refers to the ring’s outer diameter/inner diameter.
adapted to account for the fraction of the volume in which the reactions actually occur. The strong linear relationship between kH2 O2 ;obs and the plasma area indicates that kH2 O2 ;obs is proportional to A (Fig. 8). The lack of deviation from this linear relationship suggests negligible variation 0 in kH2 O2 with changes in A. Due to the quadratic relationship 0
between kH2 O2 and C OH , the degree of variation in C OH must be even 0
0
lower than in kH2 O2 , and thus is also negligible. Because k only var0
ies with C OH , k must be independent of the plate size. Therefore, interfacial area is confirmed as the dominant factor influencing the solute removal rate in this system. 3.5. General design principle
Fig. 4. Normalized RhB concentration profiles for discharges in the liquid and in the gas (ground plate diameter is 3.15 cm in both cases).
The relationship between the removal rate and the interfacial area is likely to explain why gas discharges are so much more effective than liquid discharges, as the plasma channels in the
3.4. The production of H2O2 To investigate whether variations in k0 (i.e., OH radical concentration) with A might account for some of the variation in kobs, the rate of H2O2 production was measured for gas discharge using different plate sizes. The rate of H2O2 production serves as a relative measure of the rate of OH radical production because H2O2 is formed by the recombination of OH radicals [21]:
OH þ OH ! H2 O2
ð6Þ
Similar to the reactions between organic solutes and OH radicals, this recombination reaction can only occur in the interfacial zone, where OH radicals are present. Therefore, like Eq. (4), the overall rate of reaction must be proportional to the area of the plasma–liquid interface. However, unlike Eq. (4), the reaction is pseudo-zero-order and all reactants originate in the plasma interior; therefore, diffusion is negligible. These features are captured in the following kinetic model:
dC H2 O2 A L 0 ¼ kH2 O2 ¼ kH2 O2 ;obs V dt
ð7Þ
where C H2 O2 (mM) is the concentration of H2O2, t (min) is treatment time, kH2 O2 ;obs (mM min1) is the observed rate constant calculated 0
from the C H2 O2 measurements, and kH2 O2 is the pseudo-zeroth-order 0 (kH2 O2
C 2OH ,
rate constant ¼ kH2 O2 where kH2 O2 is the second-order reaction rate constant). Note that Eq. (7) is not technically a heterogeneous rate equation, but rather a homogeneous rate equation
Fig. 5. Gas discharge with four different plate diameters: (a) 0.64 cm, (b) 3.15 cm, (c) 4.75 cm, and (d) 7.6 cm.
548
G.R. Stratton et al. / Chemical Engineering Journal 273 (2015) 543–550
Fig. 6. The relationship between the measured peak discharge current and the measured plasma area. Ground plate diameters: (j) 0.64 cm, (d) 3.15 cm, (N) 4.75 cm, and (.) 7.6 cm.
gas are visibly longer, broader and more numerous than those in the liquid, and thus would have more surface area. Because interfacial area, or in other words, plasma–liquid contact, was found to have such strong influence over reactor performance, increasing the contact became the primary design objective for the reactors described in Section 3.7. 3.6. Reactor performance Reactor performance is compared in terms of the calculated energy yield G50 (g kWh1):
G50 ¼
C0 V kobs 2 P lnð1=2Þ
ð8Þ
where C0 (0.0075 g L1) is the initial RhB concentration, V (L) is the solution volume, P (=1.35 10–4 kWh min1) is the average discharge power and kobs (min1) is the observed rate constant, calculated from Eq. (2). Because kobs is inversely proportional to V (Eq. (4)) and G50 is proportional to V and kobs, the variation in solution volume between reactors (Table 1) has no influence on the G50. 3.7. Contact-oriented reactors
propagated along the surface of this liquid stream, which increased the contact and thus the G50 in comparison to the gas discharge reactor. The turbulent and laminar jet reactors both were more efficient than the RVC HV reactor. These reactors had higher velocity liquid streams that churned the bulk liquid and caused a layer of foam to form on the surface. The foam adds complexity to the liquid surface and influences the leader propagation pattern such that the leader must follow a less direct path across the surface, which effectively increases the length of the leaders and thus the contact area. Much previous work has been done to determine how streamers propagate within gas bubbles immersed in liquid; however, the influence of bubbles adjacent to the bubble in which a streamer propagates is unknown, but may factor into why the presence of foam enhances the removal rate so dramatically [22– 24]. One possible mechanism causing the change in removal rates may involve the electric field enhancements resulting from the positive space charge produced at the head of the streamer as it propagates along the liquid film between two bubbles. This enhanced electric field may trigger small discharges over the surface of the opposite side of the film within the adjacent bubble, which would effectively increase the area exposed to plasma. Another possible mechanism involves the electrostatic interactions between neighboring leaders being dampened by the liquid films present between them, due to the films having a higher dielectric constant than the argon. This might reduce the distance between leaders and thus increase the number of leaders that emerge. Regardless of the particular mechanism, it is clear that the high surface density of the foam enhances the contact and thus accounts for the substantial increase in the G50 compared to the RVC HV reactor. The lower efficiency achieved by the turbulent jet, in comparison to the laminar jet, is the result of the streamers following the liquid streams of the turbulent jet to the surface of the bulk liquid. Because the turbulent jet has a wide spray angle, the streamers reach larger radii before reaching the surface of the bulk liquid, and therefore the plasma area is lower than that of the laminar jet. In an effort to increase the rate of generation of foam in the reactors, argon was bubbled through a diffuser that is positioned directly beneath the hole in the grounded ring electrode (Fig. 3(f)–(i)). This resulted in more foam on the surface, and significantly increased the removal efficiency for all of the reactors except the turbulent jet reactor (Fig. 10). While bubbling argon has a comparable effect on the RVC HV, laminar jet, and gas discharge reactors, it has no significant effect on the turbulent jet
The contact principle was first applied to develop three new reactor types: turbulent jet (Fig. 3(c)), laminar jet (Fig. 3(d)) and RVC HV (Fig. 3(e)). The general idea behind these contact-oriented reactors was to pass the liquid feed directly through the discharge region above the liquid surface to facilitate additional contact between the plasma and liquid. Even though the basic configurations of the three reactors were quite similar, their performances differed significantly (Fig. 9). The RVC HV reactor featured a low velocity liquid stream falling through the RVC plate, which had minimal effect on the characteristics of the liquid surface. However some of the leaders Table 2 Measured and theoretical discharge characteristics for different plate diameters. Ground diameter (cm)
Ap (cm2)
Imax,meas (A)
U (V)
Imax,thry (A)
Imax,meas/ Imax,thry
0.64 3.15 4.75 7.60
34.1 52.9 67.0 90.2
52.2 83.3 99.4 115.3
15,500 15,038 13,637 11,607
106 159 183 209
0.49 0.52 0.54 0.55
Fig. 7. Relationship between the observed RhB removal rate constant and the measured plasma area. Ground plate diameters: (j) 0.64 cm, (d) 3.15 cm, (N) 4.75 cm, and (.) 7.6 cm.
549
G.R. Stratton et al. / Chemical Engineering Journal 273 (2015) 543–550
Fig. 8. Relationship between the observed H2O2 production rate constant and the measured plasma area. Ground plate diameters: (j) 0.64 cm, (d) 3.15 cm, (N) 4.75 cm, and (.) 7.6 cm.
reactor. The lack of improvement for the turbulent jet reactor is likely because of the wide spray angle, as discussed previously, which causes the streamers to pass over the central region where the argon is bubbled without making contact, thus rendering a negligible benefit from the additional foam. 3.8. Comparison with other AOPs In this study, the best performing reactor was the laminar jet with bubbling, achieving a G50 of 11.4 g/kWh, which is nearly 150 times that of the liquid discharge. It is also significantly greater that the RhB removal efficiency of other AOPs (Table 3). The G50 values presented in Table 3 were calculated using the input power and concentration–time data provided in the corresponding articles. The initial RhB concentrations were often different between studies; therefore, when possible, the rate constants calculated from the extracted data were adjusted, via interpolation or extrapolation, to represent an initial RhB concentration of 7.5 mg/L. However, when there was insufficient information to adjust the rate constant, the calculated G50 values were linearly scaled to represent 7.5 mg/L, which necessitated assuming that the rate constant was independent of initial concentration. Therefore, the calculated G50 values should only be considered as an order-ofmagnitude estimate.
Fig. 10. Removal efficiencies for all reactors.
Table 3 G50 values for RhB degradation with different AOPs. AOP
G50 (g/kWh)
References
UV/H2O2 Ultrasound Photocatalysis Hydrodynamic cavitation Ozonation* Electrical discharge plasma: liquid discharge (worst performing reactor) Electrical discharge plasma: laminar jet with bubbling (best performing reactor)
0.1 0.2 0.2 0.01 0.3 0.078
[25] [26] [27] [28] [29] Present work
11.4
Present work
* G50 was calculated from removal rate and corresponding ozone dosage achieved in [29] and ozone generation efficiency achieved by commercial ozone generators [30].
4. Conclusions This study found that the overall rates of solute degradation and H2O2 production are better represented by rate equations that account for the area of the plasma–liquid interface, rather than the standard homogeneous-type rate equations. When all other factors (i.e., mass-transfer coefficient and OH radical concentration) remain constant, the removal rate of RhB and the production rate of H2O2 vary linearly with interfacial area. The strong influence of interfacial area likely explains why gas discharges are more effective than liquid discharges for solute degradation. Additionally, observing the influence of interfacial area prompted the origination of a general principle for reactor design, where maximizing plasma–liquid contact is the leading objective. This principle proved valuable when applied for the development of several new reactor types, all of which perform far better than the standard liquid and gas discharge reactors with which this study began. An effective means of increasing contact and improving reactor performance is to generate a layer of foam on the liquid surface by bubbling gas beneath the surface or by agitating the surface with a liquid jet. Acknowledgement
Fig. 9. Removal efficiencies for the contact-oriented reactors and the liquid and gas discharge reactors.
The authors would like to acknowledge the support by the Environmental Protection Agency (R835332).
550
G.R. Stratton et al. / Chemical Engineering Journal 273 (2015) 543–550
References [1] C. Comninellis, A. Kapalka, S. Malato, S.A. Parsons, I. Poulios, D. Mantzavinos, Advanced oxidation processes for water treatment: advances and trends for R&D, J. Chem. Technol. Biotechnol. 83 (2008) 769–776. [2] M. Klavarioti, D. Mantzavinos, D. Kassinos, Removal of residual pharmaceuticals from aqueous systems by advanced oxidation processes, Environ. Int. 35 (2009) 402–417. [3] W.H. Glaze, Drinking-water treatment with ozone, Environ. Sci. Technol. 21 (1987) 224–230. [4] J.C. Kruithof, P.C. Kamp, B.J. Martijn, UV/H2O2 treatment: a practical solution for organic contaminant control and primary disinfection, Ozone Sci. Eng. 29 (2007) 273–280. [5] Environmental safety of halogenated by-products from use of active chlorine, Science Dossiers, Euro Chlor., 2010. [6] R.P. Joshi, S.M. Thagard, Streamer-like electrical discharges in water: Part II. Environmental applications, Plasma Chem. Plasma Process. 33 (2013) 17–49. [7] B.R. Locke, M. Sato, P. Sunka, M.R. Hoffmann, J.S. Chang, Electrohydraulic discharge and nonthermal plasma for water treatment, Ind. Eng. Chem. Res. 45 (2005) 882–905. [8] D. He, Y. Sun, L. Xin, J. Feng, Aqueous tetracycline degradation by non-thermal plasma combined with nano-TiO2, Chem. Eng. J. 258 (2014) 18–25. [9] Z.L. Liu, Y.F. Zeng, J.H. Sun, K. Di Yang, The effects of different methods of catalyst preparation on the hydro-electric plasma TiO2-catalyzed degradation of 2,4-dinitrophenol, Environ. Chem. Lett. 7 (2009) 149–153. [10] Y. Zhang, Q. Xin, Y. Cong, Q. Wang, B. Jiang, Application of TiO2 nanotubes with pulsed plasma for phenol degradation, Chem. Eng. J. 215 (2013) 261–268. [11] M. Malik, Water purification by plasmas: which reactors are most energy efficient?, Plasma Chem Plasma Process. 30 (2010) 21–31. [12] G. Eisenberg, Colorimetric determination of hydrogen peroxide, Ind. Eng. Chem. Anal. Ed. 15 (1943) 327–328. [13] C. Du, T. Shi, Y. Sun, X. Zhuang, Decolorization of Acid Orange 7 solution by gas–liquid gliding arc discharge plasma, J. Hazard. Mater. 154 (2008) 1192– 1197. [14] J. Gao, Y. Liu, W. Yang, L. Pu, J. Yu, Q. Lu, Oxidative degradation of phenol in aqueous electrolyte induced by plasma from a direct glow discharge, Plasma Sources Sci. Technol. 12 (2003) 533–538. [15] J. Gao, X. Wang, Z. Hu, H. Deng, J. Hou, X. Lu, J. Kang, Plasma degradation of dyes in water with contact glow discharge electrolysis, Water Res. 37 (2003) 267–272.
[16] E.L. Cussler, Diffusion: Mass Transfer in Fluid Systems, Cambridge University Press, 2009. [17] S.B. Gupta, H. Bluhm, The potential of pulsed underwater streamer discharges as a disinfection technique, IEEE Trans. Plasma Sci. 36 (2008) 1621–1632. [18] V. Belosheev, Self-consistent development and fractal structure of leader discharges along a water surface, Tech. Phys. 44 (1999) 381–386. [19] V. Belosheev, Discharge leader self-organization on the water surface, Tech. Phys. 45 (2000) 922–927. [20] V. Belosheev, Leader discharge over a water surface in a Lichtenberg figure geometry, Tech. Phys. 43 (1998) 1329–1332. [21] B.R. Locke, K.-Y. Shih, Review of the methods to form hydrogen peroxide in electrical discharge plasma with liquid water, Plasma Sources Sci. Technol. 20 (2011) 034006–034021. [22] N.Y. Babaeva, M.J. Kushner, Structure of positive streamers inside gaseous bubbles immersed in liquids, J. Phys. D: Appl. Phys. 42 (2009) 132003–132008. [23] N.Y. Babaeva, M.J. Kushner, Effect of inhomogeneities on streamer propagation: II. Streamer dynamics in high pressure humid air with bubbles, Plasma Sources Sci. Technol. 18 (2009) 035010–035023. [24] B. Sommers, J. Foster, N.Y. Babaeva, M.J. Kushner, Observations of electric discharge streamer propagation and capillary oscillations on the surface of air bubbles in water, J. Phys. D: Appl. Phys. 44 (2011) 082001–082007. [25] F.H. AlHamedi, M.A. Rauf, S.S. Ashraf, Degradation studies of Rhodamine B in the presence of UV/H2O2, Desalination 239 (2009) 159–166. [26] M.A. Behnajady, N. Modirshahla, S.B. Tabrizi, S. Molanee, Ultrasonic degradation of Rhodamine B in aqueous solution: influence of operational parameters, J. Hazard. Mater. 152 (2008) 381–386. [27] P. Worajittiphon, K. Pingmuang, B. Inceesungvorn, N. Wetchakun, S. Phanichphant, Enhancing the photocatalytic activity of ZnO nanoparticles for efficient Rhodamine B degradation by functionalised graphene nanoplatelets, Ceram. Int. 41 (2015) 1885–1889. [28] K.P. Mishra, P.R. Gogate, Intensification of degradation of Rhodamine B using hydrodynamic cavitation in the presence of additives, Sep. Purif. Technol. 75 (2010) 385–391. [29] B. Cuiping, X. Xianfeng, G. Wenqi, F. Dexin, X. Mo, G. Zhongxue, X. Nian, Removal of Rhodamine B by ozone-based advanced oxidation process, Desalination 278 (2011) 84–90. [30] Primozone GM-series, Primozone., 2015.
Contents lists available at ScienceDirect
Chemical Engineering Journal journal homepage: www.elsevier.com/locate/cej
Plasma-based water treatment: Conception and application of a new general principle for reactor design Gunnar R. Stratton a, Christopher L. Bellona b, Fei Dai b, Thomas M. Holsen b, Selma Mededovic Thagard a,⇑ a b
Clarkson University, Plasma Research Laboratory, Department of Chemical and Biomolecular Engineering, 8 Clarkson Avenue, Potsdam, NY 13699, USA Clarkson University, Department of Civil and Environmental Engineering, 8 Clarkson Avenue, Potsdam, NY 13699, USA
h i g h l i g h t s Basic guidelines for plasma reactor design and optimization have been developed. Rhodamine B degradation depends on the area of the plasma–liquid interface. Reactor performance can be improved by generating foam on the liquid surface.
a r t i c l e
i n f o
Article history: Received 5 February 2015 Received in revised form 13 March 2015 Accepted 15 March 2015 Available online 20 March 2015 Keywords: AOP Plasma Pseudo-first-order Reactor design Rhodamine B Water treatment
a b s t r a c t To improve the feasibility of plasma-based water treatment technology and develop basic guidelines for reactor design and optimization, a study was conducted to identify and characterize design parameters and physical phenomena that influence treatment efficiency. The first phase of the study established that the chemical reactions responsible for the degradation of organic solutes can be more accurately represented, mathematically, as heterogeneous reactions, rather than the common representation as homogeneous reactions. Using Rhodamine B as the model solute, the observed removal rate constant was found to be proportional to the area of the plasma–liquid interface. This observation supported the validity of the proposed heterogeneous rate equation and inspired the conception of a general design principle, which prescribes maximizing contact between the plasma and the treated solution. The second phase of the study involved the application of this design principle to create seven different ‘‘contact-oriented’’ reactors. The design parameters employed to increase contact included feeding liquid streams directly through the discharge region and generating a layer of foam on the liquid surface. The contact-oriented reactors validated their founding principle by achieving removal efficiencies up to 145 times that for the reference case (point-plate with discharge in liquid). The removal efficiencies attained in this work compare quite favorably with those achieved by other advanced oxidation processes for the degradation of Rhodamine B. Ó 2015 Elsevier B.V. All rights reserved.
1. Introduction Advanced oxidation processes (AOPs) have increasingly been under investigation for their applications in the treatment of chemically and biologically contaminated water [1,2]. Some of these processes have reached a high level of development and have been commercialized, such as ultraviolet (UV)/hydrogen peroxide (H2O2) treatment and ozonation (O3) [3,4]. Pulsed electrical discharge plasma formed directly in or above water constitutes another, less-developed, AOP, but is receiving increased attention from researchers due to its unique treatment capabilities. Like ⇑ Corresponding author. Tel.: +1 315 268 4423; fax: +1 315 268 6654. E-mail address: [email protected] (S.M. Thagard). http://dx.doi.org/10.1016/j.cej.2015.03.059 1385-8947/Ó 2015 Elsevier B.V. All rights reserved.
other AOPs, plasma-based water treatment (PWT) makes use of the highly oxidative hydroxyl (OH) radical to oxidize chemical contaminants, and thus does not yield the hazardous byproducts that may be produced by conventional chlorination processes [1,5]. PWT involves the generation of OH radicals in situ, and from the water itself; therefore, unlike most AOPs and conventional processes, little to no chemical additives are required. Additionally, plasma offers a broader range of chemical and physical treatment mechanisms. Besides producing several reactive chemical species (OH, O, H, HO2, O 2 , O3, H2O2, H2), the plasma channels formed directly in water can reach temperatures of up to several thousands of Kelvin and thermally degrade molecules, emit UV and visible light, and generate shockwaves capable of inducing cavitation [6,7]. While these features make plasma attractive as a
544
G.R. Stratton et al. / Chemical Engineering Journal 273 (2015) 543–550
stand-alone technology, they also open up the possibility of combining plasma with other AOPs. For example, the generation of UV light allows plasma to be effectively integrated with photocatalysis [8–10]. While the potential advantages of PWT are evident, the technology has not yet reached a high enough level of development to be used in practice. Some of the obstacles that have hindered progress in developing PWT systems are associated with the absence of general principles to guide the design of new reactors with higher efficiency. The most expansive set of design criteria were offered by Malik in his review, where several different plasma reactors were compared in terms of treatment efficiency [11]. This review provided insight into which of the existing PWT systems performed best, and offered explanations for which process parameters contributed to higher treatment efficiencies. While there is much value in knowing which of the existing process parameters are most effective, it does not provide assistance to the development of new reactor designs. The present study investigates how selected reactor design parameters affect removal efficiency of Rhodamine B (RhB) dye, and uses the resulting relationships to originate a more universal guideline for developing new reactors. In addition to providing insights into the fundamental nature of the overall mechanism responsible for organic solute degradation, this guideline is employed, to considerable benefit, for the development of several novel reactor types.
2. Experimental The high voltage (HV) pulsed power supply used to generate the plasma was a custom built unit (Applied Physical Electronics LC, Spicewood, TX), with a general circuit diagram as shown in Fig. 1. The operational parameters used for this study were 20 kV discharges at 43 Hz, using a 0.94 nF load capacitor. The voltage and current in the plasma reactor were measured using a Tektronix P6015A high voltage probe and a Tektronix P6021 current probe connected to a Tektronix TDS 3032C oscilloscope. Examples of the voltage and current waveforms for discharges in both liquid and gas are shown in Fig. 2. All reactors (Fig. 3) consisted of a 17.3 cm diameter glass vessel fitted with an airtight polymer cap, which was adapted to allow for sample extraction, solution recirculation, and integration of the electrodes. The reactors were operated in semi-batch mode, with liquid recirculating at 1.4 L/min. The liquid recirculation loop ensured thorough mixing and included a heat exchanger to keep the solution at 15 °C. The headspace was purged with argon at 2.1 L/min. All reactors featured the point-plate electrode configuration, and differed in the size, shape, type and position of each electrode, as well as the mode by which the liquid feed stream entered the reactor. The ‘‘liquid discharge’’ reactor (Fig. 3(a)) contained a HV nickel chromium (NiCr) point electrode and a grounded stainless steel (S.S.) plate electrode both in the liquid phase with 1.5 cm
Rotary Spark Gap
Charge Resistor
HVDC Power Supply
spacing. The ‘‘gas discharge’’ reactor (Fig. 3(b)) featured the grounded S.S. plate in the liquid and the HV NiCr point in the gas with 2.7 cm spacing (1.5 cm in liquid, 1.2 cm in gas). Four different sized S.S. plates were used: 0.64, 3.15, 4.75 and 7.60 cm. For all of the ‘‘contact-oriented’’ reactors (Fig. 3(c)–(i)), the electrode positioning was similar to the gas discharge reactor, except a grounded aluminum ring (9.8 cm outer diameter, 6.0 cm inner diameter) was used in place of the grounded S.S. plate. In the turbulent jet reactor (Fig. 3(c)), the liquid feed was sprayed over the HV NiCr point with a 90° spray angle. The laminar jet reactor (Fig. 3(d)) was configured to introduce the liquid feed through a S.S. tube (0.32 cm inner diameter), which also served as the HV electrode. In the RVC HV reactor (Fig. 3(e)), the liquid feed passed through a reticulated vitreous carbon (RVC) plate, which served as the HV electrode. All of the ‘‘bubbling’’ reactors (Fig. 3(f)–(i)) were the same as the corresponding non-bubbling reactors, but involved bubbling argon (2.1 L/min) through a gas diffuser (9.5 cm diameter) centered beneath the hole in the ring electrode. The design characteristics of all the reactors are summarized in Table 1. The treated solution was deionized water containing 7.5 mg/L (0.0157 mM) RhB with sodium chloride added to adjust the conductivity to 300 lS/cm. The pH was 5.4 for the untreated solution, and averaged 5.0 for the treated solution. The conductivity of the treated solution was 300–330 lS/cm. Due to variations in characteristics of the reactors, the solution volume was not the same for all experiments, ranging from 850 to 1350 mL; however, when appropriate, the reported results have been linearly scaled to represent a solution volume of 600 mL (chosen arbitrarily). The concentration of the dye was determined spectrophotometrically (Shimadzu UV-1800) by measuring its absorbance at 554 nm. The concentration of H2O2 was determined spectrophotometrically, using the reaction between H2O2 and titanium sulfate and measuring the absorbance of the resulting yellow complex at 410 nm [12].
3. Results and discussion 3.1. Liquid and gas discharge reactors The study began by using two of the most common reactor types, the point-plate reactor with discharges directly in liquid (Fig. 3(a)) and discharges in gas (Fig. 3(b)). As shown in Fig. 4, the performance of the liquid discharge reactor is poor, though the removal efficiency (discussed in Section 3.6) is of the same order of magnitude as those reported for this reactor type in other plasma studies [11]. The performance of the gas discharge reactor (Fig. 4) was significantly better, with an observed rate constant (discussed in the following section) approximately 16 times higher than that for the liquid discharge. 3.2. Solute degradation kinetics To determine the reason behind the superior performance of the gas discharge compared to that of the liquid discharge, the nature of the overall degradation mechanism was evaluated. The shape of the concentration curves shown in Fig. 4 suggests first order kinetics, which is described by the following rate equation:
Load Capacitor
Fig. 1. Circuit diagram for pulsed power supply.
Plasma Reactor
dC S ¼ kobs C S dt
ð1Þ
where CS (mg L1) is the solute concentration (RhB in this case), t (min) is the treatment time, and kobs (min1) is the observed firstorder rate constant. By integrating Eq. (1) and rearranging, an equation is obtained to calculate the observed rate constant from the concentration data:
G.R. Stratton et al. / Chemical Engineering Journal 273 (2015) 543–550
545
Fig. 2. (a) Voltage waveform for liquid discharge, (b) current waveform for liquid discharge, (c) voltage waveform for gas discharge and (d) current waveform for gas discharge.
ln
C S;0 ¼ kobs t CS
ð2Þ
where CS,0 (mg L1) is the initial solute concentration. Eq. (1) is often interpreted as directly representing the pseudo-first-order reaction between the organic solute and OH radicals [13–15]:
Solute þ OH ! Products
ð3Þ
with kobs treated as the pseudo-first-order reaction rate constant (kobs ¼ k C OH , where k is the second-order reaction rate constant between the solute and OH radicals). However, this interpretation requires the assumption that the reaction is homogeneous, which is doubtful since OH radicals exist only in the immediate vicinity of the plasma channel (the interfacial zone). Thus, the solute degradation will likely only occur in the interfacial zone rather than within the bulk liquid. While RhB is photosensitive, it is not expected that photodegradation occurs to any significant degree, due to the relative lack of emissions in the range of wavelengths in which RhB absorbs (450–600 nm) [7]. Considering the non-homogeneous nature of this system, a more suitable kinetic model would be that for a heterogeneous irreversible pseudo-first-order reaction [16]:
dC S A ¼ 0 C S ¼ kobs C S dt V 1=km þ 1=k
ð4Þ
where A (m2) is the interfacial area, V (m3) is the solution volume, k0 (m s1) is the pseudo-first-order reaction rate constant 0 (k ¼ k C OH ), and km (m s1) is the mass transfer coefficient of the solute (km = D/L, where D (m2 s1) is the solute diffusivity and L (m) is the thickness of the interfacial zone). Note that the fractional term in Eq. (4) is equal to the measured kobs.
One way to support the validity of the proposed rate equation is to show that kobs is proportional to A, which requires that the interfacial area be varied while keeping km and k0 constant. Recalling that km = D/L, it is reasonable to assume that D is independent of A, as it is primarily determined by the characteristics of the particular solute. L has been defined as the thickness of the interfacial zone, which is effectively determined by the diffusion distance of the active species (i.e., OH radicals). The diffusion distance (estimated to be 6–20 nm) is primarily influenced by the concentration of OH radicals (see discussion in Section 3.4) and the density of the liquid, which is not likely to change dramatically with variations in A [6,17]. k0 varies with the OH radical concentration, which could also depend on A (see Section 3.4).
3.3. Plasma area Interfacial area was varied by using the gas discharge reactor with grounded plate electrodes of different diameters, as shown in Fig. 5. The leaders (plasma channels) generated in the gas discharge reactor propagate radially out over the liquid surface. As the leaders propagate across the liquid surface, they branch in such a way that the distance between leaders remains relatively constant, due to electrostatic repulsion. The fully propagated plasma occupies a roughly circular region on the liquid surface with visually uniform leader distribution [18–20]. The length of the leaders and therefore the area of this circular region, which will be referred to as the plasma area, increase with the diameter of the grounded plate. Because it is very difficult to define and measure the actual area of the plasma–liquid interface (A in Eq. (4)), plasma area was investigated as a potential surrogate. The plasma area was calculated for each plate size by recording images of the discharges and measuring the average length of the leaders.
546
G.R. Stratton et al. / Chemical Engineering Journal 273 (2015) 543–550
HV
HV
(a)
(b)
HV
HV
(c)
HV
HV
Argon
(d)
(e)
HV
Argon
(f)
HV
HV
Argon
Argon
(g)
(h)
(i)
Fig. 3. Reactor diagrams: (a) point-plate with discharge in liquid, (b) point-plate with discharge in gas, (c) turbulent jet, (d) laminar jet, (e) RVC HV, (f) point-plate with discharge in gas with bubbling, (g) turbulent jet with bubbling, (h) laminar jet with bubbling, and (i) RVC HV with bubbling.
Leaders are characterized by a very low internal electrical resistance, which results in a negligible voltage drop, thus the voltage gradient across the plasma–liquid interface is independent of the position within the plasma area. This uniform voltage gradient coupled with the uniform conductivity of the water produces a uniform current density, which means that the discharge current is directly proportional to the area of the plasma–liquid interface. It is clear from Fig. 6 that the relationship between the peak discharge current (Imax,meas) and plasma area is not perfectly linear. This non-linearity is because a higher discharge current corresponds to a higher discharge rate and therefore a higher rate of reduction in the discharge voltage. Consequently, the discharge voltage at the time that corresponds to Imax,meas decreases with increasing plate size, causing the Imax,meas vs. plasma area curve to level off for larger plate sizes. The effect of this varying voltage was accounted for by comparing Imax,meas to the theoretical maximum current (Imax,thry) calculated from Ohm’s law (Table 2):
Imax;thry ¼
U r Ap d
ð5Þ
where r (300 lS cm1) is the solution conductivity, d (1.5 cm) is the thickness of the liquid layer separating plasma from the grounded electrode, Ap (cm2) is the plasma area and U (V) is the measured discharge voltage at the time that corresponds to Imax,meas. Because the plasma area is greater than the actual interfacial area, Imax,thry is greater than Imax,meas. However, the ratio of Imax,meas to Imax,thry varies by less than five percent (95% confidence interval) for all plate sizes, which confirms that the variations in U are primarily responsible for the non-linearity seen in Fig. 6. Additionally, the uniformity in the ratio of Imax,meas to Imax,thry indicates that the ratio of Ap to A must also be consistent between plate sizes, which confirms that Ap is a suitable surrogate for A. As shown in Fig. 7, kobs exhibits a strong linear relationship with the plasma area, and thus kobs must be proportional to A, which supports the validity of Eq. (4).
547
G.R. Stratton et al. / Chemical Engineering Journal 273 (2015) 543–550 Table 1 Summary of reactor design characteristics. Reactor type
Corresponding figure
Liquid discharge Gas discharge Gas discharge Turbulent jet Laminar jet RVC HV Gas discharge with bubbling Turbulent jet with bubbling Laminar jet with bubbling RVC HV with bubbling * ** ***
Fig. Fig. Fig. Fig. Fig. Fig. Fig. Fig. Fig. Fig.
HV electrode *
3(a) 3(b) 3(b) 3(c) 3(d) 3(e) 3(f) 3(g) 3(h) 3(i)
NiCr wire NiCr wire NiCr wire NiCr wire S.S. tube RVC plate NiCr wire NiCr wire S.S tube RVC plate
HV diameter (mm)
Ground electrode
Ground diameter (cm)
Solution volume (mL)
0.81 0.81 0.81 0.81 2.2** 25 0.81 0.81 2.2 25
S.S plate S.S plate Al ring Al ring Al ring Al ring Al ring Al ring Al ring Al ring
3.15 0.64, 3.15, 4.75 and 7.60 9.8/6*** 9.8/6 9.8/6 9.8/6 9.8/6 9.8/6 9.8/6 9.8/6
1000 900, 880, 870 and 850 1350 1350 1350 1350 1350 1350 1350 1350
Wire was sharpened. 2.2 mm is the tube’s inner diameter. 9.8/6 refers to the ring’s outer diameter/inner diameter.
adapted to account for the fraction of the volume in which the reactions actually occur. The strong linear relationship between kH2 O2 ;obs and the plasma area indicates that kH2 O2 ;obs is proportional to A (Fig. 8). The lack of deviation from this linear relationship suggests negligible variation 0 in kH2 O2 with changes in A. Due to the quadratic relationship 0
between kH2 O2 and C OH , the degree of variation in C OH must be even 0
0
lower than in kH2 O2 , and thus is also negligible. Because k only var0
ies with C OH , k must be independent of the plate size. Therefore, interfacial area is confirmed as the dominant factor influencing the solute removal rate in this system. 3.5. General design principle
Fig. 4. Normalized RhB concentration profiles for discharges in the liquid and in the gas (ground plate diameter is 3.15 cm in both cases).
The relationship between the removal rate and the interfacial area is likely to explain why gas discharges are so much more effective than liquid discharges, as the plasma channels in the
3.4. The production of H2O2 To investigate whether variations in k0 (i.e., OH radical concentration) with A might account for some of the variation in kobs, the rate of H2O2 production was measured for gas discharge using different plate sizes. The rate of H2O2 production serves as a relative measure of the rate of OH radical production because H2O2 is formed by the recombination of OH radicals [21]:
OH þ OH ! H2 O2
ð6Þ
Similar to the reactions between organic solutes and OH radicals, this recombination reaction can only occur in the interfacial zone, where OH radicals are present. Therefore, like Eq. (4), the overall rate of reaction must be proportional to the area of the plasma–liquid interface. However, unlike Eq. (4), the reaction is pseudo-zero-order and all reactants originate in the plasma interior; therefore, diffusion is negligible. These features are captured in the following kinetic model:
dC H2 O2 A L 0 ¼ kH2 O2 ¼ kH2 O2 ;obs V dt
ð7Þ
where C H2 O2 (mM) is the concentration of H2O2, t (min) is treatment time, kH2 O2 ;obs (mM min1) is the observed rate constant calculated 0
from the C H2 O2 measurements, and kH2 O2 is the pseudo-zeroth-order 0 (kH2 O2
C 2OH ,
rate constant ¼ kH2 O2 where kH2 O2 is the second-order reaction rate constant). Note that Eq. (7) is not technically a heterogeneous rate equation, but rather a homogeneous rate equation
Fig. 5. Gas discharge with four different plate diameters: (a) 0.64 cm, (b) 3.15 cm, (c) 4.75 cm, and (d) 7.6 cm.
548
G.R. Stratton et al. / Chemical Engineering Journal 273 (2015) 543–550
Fig. 6. The relationship between the measured peak discharge current and the measured plasma area. Ground plate diameters: (j) 0.64 cm, (d) 3.15 cm, (N) 4.75 cm, and (.) 7.6 cm.
gas are visibly longer, broader and more numerous than those in the liquid, and thus would have more surface area. Because interfacial area, or in other words, plasma–liquid contact, was found to have such strong influence over reactor performance, increasing the contact became the primary design objective for the reactors described in Section 3.7. 3.6. Reactor performance Reactor performance is compared in terms of the calculated energy yield G50 (g kWh1):
G50 ¼
C0 V kobs 2 P lnð1=2Þ
ð8Þ
where C0 (0.0075 g L1) is the initial RhB concentration, V (L) is the solution volume, P (=1.35 10–4 kWh min1) is the average discharge power and kobs (min1) is the observed rate constant, calculated from Eq. (2). Because kobs is inversely proportional to V (Eq. (4)) and G50 is proportional to V and kobs, the variation in solution volume between reactors (Table 1) has no influence on the G50. 3.7. Contact-oriented reactors
propagated along the surface of this liquid stream, which increased the contact and thus the G50 in comparison to the gas discharge reactor. The turbulent and laminar jet reactors both were more efficient than the RVC HV reactor. These reactors had higher velocity liquid streams that churned the bulk liquid and caused a layer of foam to form on the surface. The foam adds complexity to the liquid surface and influences the leader propagation pattern such that the leader must follow a less direct path across the surface, which effectively increases the length of the leaders and thus the contact area. Much previous work has been done to determine how streamers propagate within gas bubbles immersed in liquid; however, the influence of bubbles adjacent to the bubble in which a streamer propagates is unknown, but may factor into why the presence of foam enhances the removal rate so dramatically [22– 24]. One possible mechanism causing the change in removal rates may involve the electric field enhancements resulting from the positive space charge produced at the head of the streamer as it propagates along the liquid film between two bubbles. This enhanced electric field may trigger small discharges over the surface of the opposite side of the film within the adjacent bubble, which would effectively increase the area exposed to plasma. Another possible mechanism involves the electrostatic interactions between neighboring leaders being dampened by the liquid films present between them, due to the films having a higher dielectric constant than the argon. This might reduce the distance between leaders and thus increase the number of leaders that emerge. Regardless of the particular mechanism, it is clear that the high surface density of the foam enhances the contact and thus accounts for the substantial increase in the G50 compared to the RVC HV reactor. The lower efficiency achieved by the turbulent jet, in comparison to the laminar jet, is the result of the streamers following the liquid streams of the turbulent jet to the surface of the bulk liquid. Because the turbulent jet has a wide spray angle, the streamers reach larger radii before reaching the surface of the bulk liquid, and therefore the plasma area is lower than that of the laminar jet. In an effort to increase the rate of generation of foam in the reactors, argon was bubbled through a diffuser that is positioned directly beneath the hole in the grounded ring electrode (Fig. 3(f)–(i)). This resulted in more foam on the surface, and significantly increased the removal efficiency for all of the reactors except the turbulent jet reactor (Fig. 10). While bubbling argon has a comparable effect on the RVC HV, laminar jet, and gas discharge reactors, it has no significant effect on the turbulent jet
The contact principle was first applied to develop three new reactor types: turbulent jet (Fig. 3(c)), laminar jet (Fig. 3(d)) and RVC HV (Fig. 3(e)). The general idea behind these contact-oriented reactors was to pass the liquid feed directly through the discharge region above the liquid surface to facilitate additional contact between the plasma and liquid. Even though the basic configurations of the three reactors were quite similar, their performances differed significantly (Fig. 9). The RVC HV reactor featured a low velocity liquid stream falling through the RVC plate, which had minimal effect on the characteristics of the liquid surface. However some of the leaders Table 2 Measured and theoretical discharge characteristics for different plate diameters. Ground diameter (cm)
Ap (cm2)
Imax,meas (A)
U (V)
Imax,thry (A)
Imax,meas/ Imax,thry
0.64 3.15 4.75 7.60
34.1 52.9 67.0 90.2
52.2 83.3 99.4 115.3
15,500 15,038 13,637 11,607
106 159 183 209
0.49 0.52 0.54 0.55
Fig. 7. Relationship between the observed RhB removal rate constant and the measured plasma area. Ground plate diameters: (j) 0.64 cm, (d) 3.15 cm, (N) 4.75 cm, and (.) 7.6 cm.
549
G.R. Stratton et al. / Chemical Engineering Journal 273 (2015) 543–550
Fig. 8. Relationship between the observed H2O2 production rate constant and the measured plasma area. Ground plate diameters: (j) 0.64 cm, (d) 3.15 cm, (N) 4.75 cm, and (.) 7.6 cm.
reactor. The lack of improvement for the turbulent jet reactor is likely because of the wide spray angle, as discussed previously, which causes the streamers to pass over the central region where the argon is bubbled without making contact, thus rendering a negligible benefit from the additional foam. 3.8. Comparison with other AOPs In this study, the best performing reactor was the laminar jet with bubbling, achieving a G50 of 11.4 g/kWh, which is nearly 150 times that of the liquid discharge. It is also significantly greater that the RhB removal efficiency of other AOPs (Table 3). The G50 values presented in Table 3 were calculated using the input power and concentration–time data provided in the corresponding articles. The initial RhB concentrations were often different between studies; therefore, when possible, the rate constants calculated from the extracted data were adjusted, via interpolation or extrapolation, to represent an initial RhB concentration of 7.5 mg/L. However, when there was insufficient information to adjust the rate constant, the calculated G50 values were linearly scaled to represent 7.5 mg/L, which necessitated assuming that the rate constant was independent of initial concentration. Therefore, the calculated G50 values should only be considered as an order-ofmagnitude estimate.
Fig. 10. Removal efficiencies for all reactors.
Table 3 G50 values for RhB degradation with different AOPs. AOP
G50 (g/kWh)
References
UV/H2O2 Ultrasound Photocatalysis Hydrodynamic cavitation Ozonation* Electrical discharge plasma: liquid discharge (worst performing reactor) Electrical discharge plasma: laminar jet with bubbling (best performing reactor)
0.1 0.2 0.2 0.01 0.3 0.078
[25] [26] [27] [28] [29] Present work
11.4
Present work
* G50 was calculated from removal rate and corresponding ozone dosage achieved in [29] and ozone generation efficiency achieved by commercial ozone generators [30].
4. Conclusions This study found that the overall rates of solute degradation and H2O2 production are better represented by rate equations that account for the area of the plasma–liquid interface, rather than the standard homogeneous-type rate equations. When all other factors (i.e., mass-transfer coefficient and OH radical concentration) remain constant, the removal rate of RhB and the production rate of H2O2 vary linearly with interfacial area. The strong influence of interfacial area likely explains why gas discharges are more effective than liquid discharges for solute degradation. Additionally, observing the influence of interfacial area prompted the origination of a general principle for reactor design, where maximizing plasma–liquid contact is the leading objective. This principle proved valuable when applied for the development of several new reactor types, all of which perform far better than the standard liquid and gas discharge reactors with which this study began. An effective means of increasing contact and improving reactor performance is to generate a layer of foam on the liquid surface by bubbling gas beneath the surface or by agitating the surface with a liquid jet. Acknowledgement
Fig. 9. Removal efficiencies for the contact-oriented reactors and the liquid and gas discharge reactors.
The authors would like to acknowledge the support by the Environmental Protection Agency (R835332).
550
G.R. Stratton et al. / Chemical Engineering Journal 273 (2015) 543–550
References [1] C. Comninellis, A. Kapalka, S. Malato, S.A. Parsons, I. Poulios, D. Mantzavinos, Advanced oxidation processes for water treatment: advances and trends for R&D, J. Chem. Technol. Biotechnol. 83 (2008) 769–776. [2] M. Klavarioti, D. Mantzavinos, D. Kassinos, Removal of residual pharmaceuticals from aqueous systems by advanced oxidation processes, Environ. Int. 35 (2009) 402–417. [3] W.H. Glaze, Drinking-water treatment with ozone, Environ. Sci. Technol. 21 (1987) 224–230. [4] J.C. Kruithof, P.C. Kamp, B.J. Martijn, UV/H2O2 treatment: a practical solution for organic contaminant control and primary disinfection, Ozone Sci. Eng. 29 (2007) 273–280. [5] Environmental safety of halogenated by-products from use of active chlorine, Science Dossiers, Euro Chlor.
[16] E.L. Cussler, Diffusion: Mass Transfer in Fluid Systems, Cambridge University Press, 2009. [17] S.B. Gupta, H. Bluhm, The potential of pulsed underwater streamer discharges as a disinfection technique, IEEE Trans. Plasma Sci. 36 (2008) 1621–1632. [18] V. Belosheev, Self-consistent development and fractal structure of leader discharges along a water surface, Tech. Phys. 44 (1999) 381–386. [19] V. Belosheev, Discharge leader self-organization on the water surface, Tech. Phys. 45 (2000) 922–927. [20] V. Belosheev, Leader discharge over a water surface in a Lichtenberg figure geometry, Tech. Phys. 43 (1998) 1329–1332. [21] B.R. Locke, K.-Y. Shih, Review of the methods to form hydrogen peroxide in electrical discharge plasma with liquid water, Plasma Sources Sci. Technol. 20 (2011) 034006–034021. [22] N.Y. Babaeva, M.J. Kushner, Structure of positive streamers inside gaseous bubbles immersed in liquids, J. Phys. D: Appl. Phys. 42 (2009) 132003–132008. [23] N.Y. Babaeva, M.J. Kushner, Effect of inhomogeneities on streamer propagation: II. Streamer dynamics in high pressure humid air with bubbles, Plasma Sources Sci. Technol. 18 (2009) 035010–035023. [24] B. Sommers, J. Foster, N.Y. Babaeva, M.J. Kushner, Observations of electric discharge streamer propagation and capillary oscillations on the surface of air bubbles in water, J. Phys. D: Appl. Phys. 44 (2011) 082001–082007. [25] F.H. AlHamedi, M.A. Rauf, S.S. Ashraf, Degradation studies of Rhodamine B in the presence of UV/H2O2, Desalination 239 (2009) 159–166. [26] M.A. Behnajady, N. Modirshahla, S.B. Tabrizi, S. Molanee, Ultrasonic degradation of Rhodamine B in aqueous solution: influence of operational parameters, J. Hazard. Mater. 152 (2008) 381–386. [27] P. Worajittiphon, K. Pingmuang, B. Inceesungvorn, N. Wetchakun, S. Phanichphant, Enhancing the photocatalytic activity of ZnO nanoparticles for efficient Rhodamine B degradation by functionalised graphene nanoplatelets, Ceram. Int. 41 (2015) 1885–1889. [28] K.P. Mishra, P.R. Gogate, Intensification of degradation of Rhodamine B using hydrodynamic cavitation in the presence of additives, Sep. Purif. Technol. 75 (2010) 385–391. [29] B. Cuiping, X. Xianfeng, G. Wenqi, F. Dexin, X. Mo, G. Zhongxue, X. Nian, Removal of Rhodamine B by ozone-based advanced oxidation process, Desalination 278 (2011) 84–90. [30] Primozone GM-series, Primozone.