**2.1 Equipment and raw materials**

The basic raw material is surface and groundwater samples from the Mid-Ohio River Valley in Parkersburg (West Virginia) and untreated water from the Grand St. Mary's Lake in Celina (Ohio). The illustration in **Figure 2a** represents the experimental design describing the general process of treating contaminated water using electrical energy. **Figure 2b** displays the laboratory-built UV-ozone water purification prototype, consisting of a 100-gallon plastic tank batch reactor fitted with an ionized nitrogen-oxygen (NI-OX) generator. The device is also fitted with a small fractional horsepower delivery compressor and 1-μm electron separation porous cellulose fiber water filter. Using basic engineering ideas, the unit was powered by a 110-V electrical source with the generator fastened to the tank cover and connected to a 1-μm polarized polymer filter suspended 10 inches above the inside base of the tank. The filter was connected to a fine bubble aeration diffuser using a half-inch poly tubing designed to eliminate debris, suspended solids, and microcystins pollutants. The principal component of the generator is a UV radiation lamp (λ = 155 nm) capable of

#### **Figure 2.**

*(a) Schematic of the electrocoagulation purification system; (b) UV-ozone filtration unit; (c) Electrocoagulation unit; and (d) Coupled UV-ozone-electrocoagulation filtration system.*

*Removal of Microcystins from Drinking Water by Electrocoagulation: Upscaling, Challenges… DOI: http://dx.doi.org/10.5772/intechopen.105751*

splitting ambient gases (e.g., O2 and N2) into monoatomic-charged particles using ultraviolet ionizing energy and magnetic emission.

**Figure 2c** is a modified version of the prototype in **Figure 2b** designed to suit field conditions. It consists of a 400-gallon steel tank powered by a high-amperage (250 A), low-voltage generator (40 V) constructed to provide energy via switching polarity from direct current electric discharge. A characteristic component is 34 pairs of submerged anode and cathode crosslinked aluminum electrodes secured over a steel tank (**Figure 2c**). Crosslinked electrodes are the main reason for the (switching) reverse polarity. The coagulator works by establishing an intense electromagnetic field creating simultaneous oxidation-reduction reactions. An attached high-pressure pump was designed to channel polluted water over the metal plate contact areas. Treated water was pumped into a clean glass tank before samples are drawn for testing.

The third equipment (**Figure 2d**) is a coupled UV-EC (UV-ozone-EC) system, designed to mimic an industrial treatment system, with capacity to purify approximately 10,000 gallons of contaminated water within 12 h. **Table 1** shows a summary of technical characteristics of the EC processing system.

#### **2.2 Experimental methods**

Raw water and treated water samples were tested at the Industrial Chemical Laboratories, LLC (ICL) of Denver (CO). ICL is a specialized facility for testing chemical and biological pollutants in drinking water and wastewater including cyanobacteria. The samples were tested every 10 min for 90 min and analyzed using the Agilent 1100 tandem high-performance liquid chromatography-mass spectrometer (HPLC-MS). The glassware was thoroughly washed and rinsed with methanol and distilled water to prevent cross-contamination. The samples were first filtered with Whatman filter paper (1.2 μm) and chilled overnight at 20°C to dilute concentration of the pollutants. The filtrate was dissolved in 400 μL methanol and treated with a 2 mg/L sodium thiosulfate and acidified with trifluoroacetic acid (TFA, 0.1%, v/v),


#### **Table 1.**

*Characteristics of the water treatment techniques.*

concentrated via solid phase cartridges (SuperClean LC-18, 3 mL tube), and eluted with 15 mL of 0.1% TFA in methanol. Aliquots of 20 μL were injected into system's column (150 � 4.60 mm) at a flow rate of 1 mL/min at 30°C column temperature. The mobile phase consisted of H2O plus 0.05% TFA and acetonitrile plus 0.05% TFA with a linear increase from 30 to 70% of the latter between 0 and 30 min. Chromatograms were recorded at 238 nm based on the literature. UV spectra and all chromatographic peaks were examined and compared to spectra standards of MC moieties. Peaks possessing the UV spectrum characteristic for MCs were quantified using a calibration curve. Unidentified peaks possessing the UV spectrum characteristic for MC but not matching the retention time of the standards were determined as MC-LR equivalents with a detection limit of 0.01 μg/L [25]. To understand seasonal variations, MC distribution was measured seven consecutive days in spring, summer, and fall and a regional mean calculated (**Table 2**).

The UV-ozone ionization reaction process was produced following the reactions below. Charged nitrogen particles were activated to release of free electrons (e�) to accelerate oxygen ionization:

$$N\_2 \frac{UV}{\text{MagE}} N^{a+} - N^{a-} \frac{e^-}{O\_2 \text{ionized}} (\text{free electron [e}^-\text{] plus accelerating ionized O}\_2\text{)}. \tag{1}$$

The oxygen radiation produced ozone vapor, ionized ozone, and superoxide ions and dissociated into more singlet oxygen (Eq. 2), resulting in a chain reaction of highenergy ionized oxygen in (Eq. 3):

$$\text{O}\_2 \frac{\text{UV}}{\text{MgE}} \text{O}\_3 \frac{\text{UV}}{\text{MgE}} \to \text{O}\_3^+ \leftrightarrow \text{O}\_3^- \frac{\text{UV}}{\text{MgE}} \to \text{O}\_2^- \text{ (superoxide ion)}\tag{2}$$

$$O-O^- - + xO^- - O^-(\text{singlet}) - \frac{UV + electrons}{MagE} \text{--} O-\text{~}O-\text{~}O(-O\_x^-) - \text{~}O-(chain) \tag{3}$$

The reaction of singlet oxygen (or chained ionized oxygen) with water was generated to produce high concentrations of hydrogen peroxide and/or hydroxide ions as saturated water produces excess peroxyl-reactive (oxidizing, disinfecting, and coagulating) ionized water in the subsequent reactions:

$$\text{xO}\_2^- + \text{H}\_2\text{O} \frac{e^-}{\text{stream}} \rightarrow \text{xH}\_2\text{O}\_2 \leftrightarrow \text{xOH}^- + \text{xOH}\_2^- (\text{κavenger}) \tag{4}$$

$$\text{H}\_2\text{O}\_2^- + \text{H}\_2\text{O} \frac{e^-}{\text{MgE}\_{\Delta T}} \to \text{H}\_2\text{O}\_2 \tag{5}$$

Thermal reaction of hydrogen peroxide and ozone was created to release free electrons and trioxidanes, superoxide ions, and peroxone Eqs. (6)–(8). The charged nitrogen and superoxide ions in aqueous solution were designed to produce additional free electrons, dinitrogen tetraoxide (nitroxyl ions), and hydroxide ions toxic to cyanobacteria:

$$H\_2O\_2 + O\_3 \frac{e^-}{\text{MagE}\_{\Delta T}} \to H\_2O\_3 + O\_2^- \tag{6}$$

*Removal of Microcystins from Drinking Water by Electrocoagulation: Upscaling, Challenges… DOI: http://dx.doi.org/10.5772/intechopen.105751*


#### **Table 2.**

*Relationship between the microcystin and nutrient load.*

$$O\_3 + H\_2O\_2 \frac{e^-}{\Delta T} \to H\_2O\_5 \tag{7}$$

$$\mathrm{M}\_{2}^{+} + \mathrm{3O}\_{2}^{-} + 2\mathrm{H}\_{2}\mathrm{O} \frac{e^{-}}{\mathrm{MgE}\_{\Delta T}} \mathrm{^{-}} \mathrm{O}\_{2} - \mathrm{N}\_{2}^{+} - \mathrm{O}\_{2}^{-} + 4\mathrm{OH}^{-} \tag{8}$$

The electrocoagulation reverse polarity reaction follows an electrolytic procedure [26]. The primary reactions at the anode and cathode are described in Eqs. (9)–(13):

$$2\text{H}\_2\text{O}(l)\rightarrow\text{O}\_2(\text{g}) + 4\text{H}^+(aq) + 4\text{e}^-\text{Anode} \tag{9}$$

$$4\text{H}\_2\text{O}(l) + 4\text{e}^- \rightarrow 2\text{H}\_2(\text{g}) + 4\text{OH}^-(aq)\text{Cathode} \tag{10}$$

$$\text{6H}\_2\text{O}(l) \rightarrow 2\text{H}\_2(\text{g}) + \text{O}\_2(\text{g}) + 4\text{H}^+(aq) + 4\text{OH}^-(aq) \qquad \text{Overall} \tag{11}$$

$$4\text{H}^+ + 4\text{OH}^-(aq) \to 4\text{H}\_2\text{O}(l) \tag{12}$$

$$2\text{H}\_2\text{O}(l)\rightarrow 2\text{H}\_2(\text{g}) + \text{O}\_2(\text{g})\tag{13}$$

While reductants (free electrons) are released from the anode, oxidants and flocculation aggregates (e.g., H2O2, Al(OH)3, Al2O3) are generated at the cathode as shown in Eqs. (14)–(18):

$$\text{Xe}^- + \text{H}\_2\text{O} \xrightarrow{\text{Metal Elextudes}} \text{H}\_2\text{\uparrow} + \text{O}\_2\uparrow \xrightarrow{\text{Electrous}} \text{O}^- - \text{O}^- - \text{O}\_x^-(\text{Oxidation}) \quad \text{(14)}$$

$$\xrightarrow{\text{Electrous}} \text{OH}^- + \text{H}\_2\text{O}\_2\tag{15}$$

$$\xrightarrow{Electrons} Al(OH)\_3\downarrow \tag{16}$$

$$\xrightarrow{Electrons} Al\_2O\_3\,\text{\textquotedblleft}\tag{17}$$

$$\xrightarrow{\text{Electrons}} \text{Electropulliic} + \text{Nucleophilic cleavage of } \text{C}-\text{F} \text{ bonds} \tag{18}$$

The pollutant removal efficiency (% *r*) was calculated using Eq. (19) as follows:

$$\%r = \left(\frac{\text{CO} - \text{Cr}}{\text{CO}}\right) \times 100\% \tag{19}$$

where *C0* is the initial concentration of pollutant and C*<sup>t</sup>* is the concentration of pollutant at time *t*.

To initialize the EC process and augment flocculent formation, about 80 g of potassium aluminum sulfate dodecahydrate (KAl2(SO4)2.12H2O) (potash alum) solution was used both as an electrolyte and coagulant following previous experiments by Johnson [26]. Finally, treated water samples were compared with data from a commercial gravity block ionic adsorption unit fitted with granular activated carbon filters and coated by silver-impregnated ceramic outer shells. The experimental results are discussed in the subsequent section.

### **3. Results and discussion**

#### **3.1 Electrocoagulation and scalability**

The results in **Figure 3** displays microcystin (MC) response to laboratory UV exposure compared with field and municipal electrocoagulation (EC). To determine *Removal of Microcystins from Drinking Water by Electrocoagulation: Upscaling, Challenges… DOI: http://dx.doi.org/10.5772/intechopen.105751*

**Figure 3.** *Comparison of microcystin removal techniques.*

the EC production efficiency, MC filtration data were compared with the WHO's maximum contaminant level of 1.0 μg/L. While the laboratory and field experiments decomposed MC within 10–20 min, the municipal system disrupted MC bonds after 50 min. Multiple reasons can explain this. The first one is a technical challenge. As expected, building and testing the 10,000-gallon reactor (**Table 1**) was more arduous than the 90-gallon reactor. The installation of high-intensity UV lamps to generate optimal radiation density in the larger reactor was particularly challenging. Another difficulty was how to generate maximum turbulence to aerate and circulate radiation. Although this was improved using bubble diffusers, predicting diffuser size was still time consuming, requiring several iterations. The reaction delays were also attributable to differences in surface energy interactions between radiation pulse and pollutant substrates. A recent study by Cavitt et al. [27] has shown that molecular bond disruption in aqueous media is controlled by many thermodynamic factors such as reaction rates, solvent volume, acid-base equilibria, and interfacial alignment of reactants versus products. Given similar temperature doses, reaction rates were better favored in the laboratory (90-gallon) reactor than its municipal (10,000-gallon) counterpart. The general conclusion is that MC bond disruption is easier in smaller reaction tanks than larger ones.

The aforementioned result notwithstanding the results in **Figure 3** shows a slight deviation between UV irradiation (20 min) and EC (electrolytic) treatments (10 min). Notice that UV irradiation is closely related to physical factors such as UV lamp size, pulse intensity, and radiation diffusion (15, 18), while EC is controlled by direct electrical vibration against C–N bonds. The aforementioned, therefore, is a reasonable explanation for the observed discrepancy. Another reason for the reaction delay is ozone deficiency possibly resulting from coupling glitches between the system's ionized nitrogen-oxygen (NI–OX) generator and its compressor (**Figure 2b**). This matter will be further investigated in subsequent studies. It is worth noting, however, that all the prototypes (**Figure 2**) decomposed MC molecules reasonably well. Notice for

instance, that the municipal EC unit destroyed MC by approximately 80% within the first 10 min, while the coupled UV-ozone treatment was even better at 95%. Studies such as Wolfe et al. [28], Langlais et al. [29], and Folcik et al. [14] have reported polar bond disruption from electron bombardment, free radical attack, and ozone and peroxone toxicity. Peroxone is a mixture between ozone and hydrogen peroxide Eqs. (1)-(8). The theoretical basis is that heavy oxidizing agents (e.g., peroxides, trioxidanes, and peroxones) can break down functional C–N bonds of microcystins [30, 31]. Previously, He et al. [32] reported the destruction of cyanobacteria using hydroxyl-free radicals. In addition, studies such as Yoo et al. [33] and Lui et al. [34] have demonstrated how low doses of peroxyl and nitroxyl ions can disrupt chemical bonds in molecular compounds*.* Results from this study quite closely reflect conclusions by previous researchers.

To validate MC decomposition data in **Figure 3**, UV treatments were matched against granular activated-carbon filtration data. The carbon filtration was from a commercial source and readily available. The results are displayed in **Figure 4a**. Two important things are observable from the outcomes. While the UV filtration disrupted C–N bonds within 20 min, the activated carbon produced treatments after 30 min. Still, the UV filtration outperformed the commercial granular carbon filtration by nearly 40%. This was expected knowing that UV radiation is more energetic in splitting C–N bonds. To predict the rate of MC removal, the data were subjected to a crude regression analysis. The curve followed a polynomial decay in the form of *y* = 8.394*x*<sup>2</sup> –88.635*x +* 221.97; *R*<sup>2</sup> = 0.9399 (**Figure 4b**) confirming the robustness of UV radiation in removing MC pollutants within 1 h. This knowledge is relevant and applicable to wastewater treatment and chemical recycling.

#### **3.2 Principal component analysis**

This section discusses electrocoagulation (EC) principal components that control MC decomposition. The parameters below were considered important in the published literature [15]: (a) voltage (as proxy data for radiation density), (b) pH (acid-base equilibria), (c) electrical polarity (reverse polarity), and (d) temperature. The parametric data were derived using Eq. (19). The results are displayed in **Figure 5**. The municipal EC results were evaluated using the published data by Miao et al. [31] and further verified [35]. Two important points are observable here. First, MC removal increased with increasing radiation density. Second, there was a difference in optimal voltage density coincidental with maximum MC decomposition. While the field reactor completely removed MC (100% decomposition) at 24 V, the municipal reactor performed maximally at 32–40 V showing 95% pollutant removal (**Figure 5a**). The difference in energy dosage was attributed to the sheer size of the municipal reactor, which in turn resulted from generator adjustments to solve solvent turbulence and flocculent formation deficiencies.

As expected, no large differences were observed between the field and municipal reactions in terms of acid-base equilibria and thermodynamics (**Figure 5b** and **d**). The most significant conclusions are that (1) pH and temperature elevations are more favorable to MC decomposition; and (2) optimal pH for pollutant removal lies in the alkaline range with pH > 8.00. While the pH data strongly agreed with recent findings by Folcik and Pillai [14], it was strikingly contrary to previous conclusions by Bao et al. [16] whose studies on C–F decomposition was more productive in acidic media. It is possible to explain our findings in two ways. First, the coagulation "seeding agent" (i.e., potash alum (KAl2(SO4)2.12H2O) may have contributed to pH elevation.

*Removal of Microcystins from Drinking Water by Electrocoagulation: Upscaling, Challenges… DOI: http://dx.doi.org/10.5772/intechopen.105751*

#### **Figure 4.**

*(a) Comparison between UV radiation and granular activated-carbon removal of microcystins; and (b) Regression curve of the UV-ozone water treatment.*

Secondly, the production of metal hydroxides such as Al(OH)3 and Fe(OH)3 from sacrificial anodes (aluminum and iron metals electrodes in the electrolytic cell (**Figure 2**)) may have produced more alkaline conditions. Thermodynamic effects on MC decomposition are well researched including published articles by Folcik and Pillai [14], Folcik et al. [36], and Wang et al. [37]. The conclusion here is that MC bonds are significantly disrupted at temperatures beyond 40°C. The results in **Figure 5d** quite closely matched some of the aforementioned findings. In this study, however, 30°C was observed as the starting point of MC decomposition with maximum disruptions encountered between 70 and 90°C. The difference may largely be due to high generator amperage (250 A) employed (Section 2.1).

Another important EC factor is electrical polarity. Previous studies such as Triantis et al. [38] and Gajda et al. [39] have reported limitations of conventional single-anode polarity in EC production. For this reason, we experimented a more robust switching polarity procedure using crosslinked aluminum electrodes with its energy from direct

#### **Figure 5.**

*Factors controlling electrocoagulation efficiency in decomposing microcystin (a) radiation density; (b) bond polarity; (c) pH; and (d) temperature.*

current electrical discharge. Using a trial-and-error optimization approach, the reactor was "trained" to switch electrical current bombardments between the anode and cathode electrodes. The data in **Figure 5c** show that MC removal was maximum (100% removal) at every 5 s. The question is: Why is the switching polarity so important? The answer relates to two important things. First, optimizing the system saved time, power, and ultimately, cost. Second, the switching polarity ensured that the C–N bonds were attacked at both the anode and the cathode, contrary to conventional one-way electrical bond splitting. The dual attack against C–N bonds is a major reason for effective EC production. Notice, however, that the response from the largesize municipal reactor was inferior compared with its field counterpart. The discrepancy is still not easy to explain. However, operational problems such as electrode size determination for maximum flocculent distribution may be responsible for deviations. This is another topic worth investigating in future studies.

How does the new EC equipment compare with conventional community water treatment in terms of cost economics? To answer this question, the municipal prototype was "starved" of ozone and UV radiation, while extending treatments beyond 90 min. The goal was to examine whether the EC system would provide cheaper filtration compared with conventional treatments in the study area. The results are displayed in **Table 3**. The data show that while groundwater MC treatment was unimportant in West Virginia, the importance of surface water treatment was without question. The heavy pollution associated with both the Ohio River and Grand St. Mary's Lake (**Table 2**) is noticed. The data in **Table 3** show that the EC procedure was much cheaper than conventional membrane filtration or chemical disinfection. Specific to MC removal, the EC method was predicted to be nearly 800 times cheaper than conventional treatments at the Celina plant. On the basis of this study,

*Removal of Microcystins from Drinking Water by Electrocoagulation: Upscaling, Challenges… DOI: http://dx.doi.org/10.5772/intechopen.105751*


#### **Table 3.**

*Costs comparison between conventional water treatment and new electrocoagulation.*

incorporating EC methods at conventional treatment plants has potential to both improve water treatment chemistry and save cost.

#### **3.3 Further studies**

This study has confirmed published reports that advanced EC methods are effective in removing MC pollutants from drinking water. However, some key topics remain to be investigated including (a) chemical elucidation of decomposed MC fragments, (b) scalability of UV lamps, ionized nitrogen-oxygen (NI-OX) generators and EC electrodes, (c) technical operability and cost assessments, and (d) applicability of the technique for wastewater treatment and chemical recycling. Some of these topics will be addressed in subsequent studies.

### **4. Conclusion**

The removal of microscopic pollutants in drinking by electrocoagulation (EC) is becoming increasing popular because of the use of radiation energy in decomposing molecules that contain polar bonds including C–F and C–N bonds. The purpose of this study was to discuss three EC techniques for removing microcystins (MC) in contaminated drinking water at the Celina (OH) and Parkersburg (WV) treatment plants and to compare their effectiveness at the laboratory, field, and municipal scales. While the laboratory and field experiments employed UV-ozone and electrolytic cell filtration techniques, respectively, the municipal experiment applied a coupled UV-ozone and EC technique. To validate the effectiveness of the methods, the EC results were evaluated against a commercially available granular activated carbon filtration unit. The EC technique outperformed the activated carbon filtration by more than 40%. When the laboratory treatments were upscaled and tested at a municipal level, effectiveness of the technique declined by nearly 10–20% because of pulse dissipation from UV lamps, vapor-ion plasma underactivity, and limitation of membrane filters. We

confirmed previously published studies that pollutant coagulation and MC decomposition were affected by physical factors such as radiation density, reverse electrical polarity, pH, and temperature. These results have other applications in industrial wastewater treatment and chemical recycling.
