Investigation of electrode passivation during electrocoagulation treatment with aluminum electrodes for high silica content produced water Open Access

Electrocoagulation (EC) is an electrochemical water treatment technology that operates by applying an anodic potential to metal electrodes submerged in contaminated water. In this process, the sacrificial anode corrodes to produce metal cations such as aluminum or iron that subsequently hydrolyze to form coagulants in the bulk electrolyte. These coagulants neutralize, sorb, and remove contaminants from the aqueous phases by precipitation, flotation, or filtration. When applying a direct current to an EC reactor (DC-EC) with aluminum electrodes, the anode generates Al³⁺ while the electrolysis of water takes place at the cathode. The fundamental concepts and emerging applications of EC have been widely explored (Garcia-Segura et al. 2017; Tahreen et al. 2020), and economic analyses indicate that for many applications EC is more economical than alternative chemical treatment processes (Palahouane et al. 2015). EC has several other advantages, such as avoiding the use of externally added chemicals, and facilitates automation through the electrochemical control of coagulant dosing (Yasri & Gunasekaran 2017). As environmental concerns about water consumption and the distribution of pollutants in the environment increase, there is increasing interest in the use of EC for treatment and recycling of end-of-pipe effluents in industry (Al-Raad & Hanafiah 2021; Almukdad et al. 2021; Jing et al. 2021).

In the steam-assisted gravity drainage (SAGD) oil production process, which is used in the Canadian oil-sands industry, the produced water (water downstream of de-oiling) and concentrated blowdown water (from steam generators) typically contain high concentrations of inorganics and organic residuals (Liang et al. 2018). The produced water from the SAGD process is recycled to the steam generators and thus requires an effective produced water treatment process (Abdol Hamid et al. 2008). Typically, produced and blowdown waters contain moderate to high concentrations of total dissolved solids (TDS), including inorganic metals, metalloids, non-metals, and have a pH within the range of 7–11 (Liang et al. 2018). Additionally, produced water contains a wide range of water-immiscible organic contaminants (see Table 1). Limited information is available regarding the properties of the inorganic and organic matter in blowdown water, but dissolved silica (SiO₂) has been identified as problematic because it is linked to the failure of equipment due to its co-precipitation with magnesium, calcium, and other organic products (Maiti et al. 2012; Liang et al. 2018). Evidence of silica fouling and clogging of subsurface disposal wells has been reported in a range of steam generation processes including SAGD oil production process (Maiti et al. 2012), and evaporative desalination processes, which produce concentrated blowdown streams that are difficult to dispose of (Den & Wang 2008).

The methods currently used to remove silica from produced water include chemical softening, physical evaporation, and chemical coagulation (Liang et al. 2018; Tahreen et al. 2020). While these methods are effective, they are expensive, require large amounts of chemicals, and generate substantial quantities of sludge. For example, a recent study by Maiti et al. (2012) found that complete silica precipitation can occur at acidic pH below 4, and that the precipitate particles can block the filter pores in the subsequent filtration step.

In this study, the affinity of silica to form stable precipitates with aluminum (Liang et al. 2018) was exploited using EC with aluminum electrodes (Al-EC) to remove silica from produced water. The selection of aluminum electrodes for the treatment of silica-containing effluent over the commonly used iron electrodes in the EC process is due to the formation of more stable aluminosilicates (Al₂Si₂O₅(OH)₄, ΔG = ‒3,759 KJ/mol) compounds as compared with iron silicate (Fe₂SiO₄, ΔG = ‒1,379 KJ/mol) which allow faster coagulation with Al electrodes (Lange & Speight 2005). Moreover, in contrast to EC with iron electrodes, Al³⁺ ions are not expected to undergo reversible electrochemical side reactions (Ingelsson et al. 2020), which help understanding the polarity reversal process without encountering sides reactions.

The complex chemistry of produced and blowdown water can lead to operational and performance problems that are major challenges for the commercialization of EC. Among these challenges, the precipitation of material on the electrodes – referred to as fouling – is particularly problematic (Ingelsson et al. 2020). This effect increases the electron transfer resistivity at the electrode/electrolyte interfaces and impedes the mass transfer of the electrochemically dissolved coagulant to the bulk solution. Electrode fouling has consequently been shown to reduce the faradaic efficiency for metal dissolution in EC processes during long-term operation (Ingelsson et al. 2020). It has also been reported that severe fouling can plug the EC reactor (Kabdaşlı et al. 2012). Electrode fouling is the result of electrochemical and physicochemical mechanisms involving metal coagulant and aqueous-phase species react and deposit at the electrode interfaces. The electrolysis of water contributes to the formation of the acidic and alkaline pH boundaries at the anodic and cathodic interfaces, respectively (Ingelsson et al. 2020). Thus, ionic species in the supporting electrolyte could precipitate on the electrode surfaces if their solubility limits are exceeded. This effect is exacerbated by high ionic concentrations at the electrode/electrolyte interfaces, which may occur due to the slow electrophoretic migration in the electrolyte (Ingelsson et al. 2020). Hence, to create an EC process that sustains effective treatment over time, fouling must be controlled or eliminated. Thus, many engineering strategies have been proposed to mitigate the precipitation of material on the electrodes. For example, researchers have explored intermittently cleaning the electrodes mechanically, installing oscillating electrodes to enhance coagulant mass transfer, or periodically reversing electrode polarity (Ingelsson et al. 2020). Among these methods, EC with polarity reversal (PR-EC) has received the most attention (Eyvaz et al. 2009).

Although PR-EC or DC-EC systems operating with the same current theoretically produce the same amount of coagulant (according to Faraday's Law of electrolysis), it has been reported that they exhibit differences in performance (Mollah et al. 2001). For instance, using aluminum electrodes in an EC process, Eyvaz et al. (2009) found that the removal efficiencies of contaminants and the operating cost of PR-EC were superior to DC-EC during the treatment of effluent containing reactive and disperse dyes. A possible mechanism for the improved performance with PR-EC is the reduction in fouling and passivation due to the inversion of the acidic and alkaline pH layers at the anode and cathode, respectively (Fekete et al. 2016). This may lead to a change in the rate of formation of aluminosilicate particles precipitating on the electrode when the high pH environment at the cathode becomes acidic following polarity reversal. However, contradictory findings have also reported that the coulombic efficiency (and hence EC performance) for aluminum dissolution was diminished when operating PR-EC with a high-frequency of polarity reversal (Fekete et al. 2016).

To address these contradictory findings regarding PR-EC performance, in this study we investigate the effect of the frequency in the polarity reversal under various conditions. In order to provide insight into the electro- and physiochemical processes occurring during PR-EC operation, unlike previous studies which typically focus on performance and faradaic efficiency, a range of additional techniques were used, including voltammetry, and high-resolution mass spectroscopy, for a range of DC-EC and PR-EC operating conditions. The use of PR-EC to mitigate fouling in Al-EC applications with moderate or high silica contents was explored for the first time by investigating the treatment of SAGD boiler blowdown water. By comparing the treatment of organic-free synthetic and field samples of blowdown water, the impact of organics on the silica removal, and the removal of organics from the SAGD boiler blowdown water by EC was studied. In order to investigate the limitations of PR operation at low and high frequencies, the effect of varying the period of polarity reversal on the silica removal rate and the fouling layer morphology in synthetic and field samples of blowdown water was investigated.

The following reagent-grade chemicals were used as received from Sigma-Aldrich to prepare the synthetic blowdown water: sodium metasilicate (Na₂SiO₃), sodium sulfate (Na₂SO₄), and sodium chloride (NaCl). Field samples of blowdown water were obtained from an industrial site, analysed and a simulated synthetic effluent was prepared.

To investigate the performance of EC, two types of blowdown samples were studied: a synthetic model and a field sample of blowdown water. The synthetic blowdown contained 8.13 g L⁻¹ of Na₂SiO₃ (equivalent to 4 g L⁻¹ SiO₂ and 1.87 g L⁻¹ Si), 30 g L⁻¹ Na₂SO₄, and 60 g L⁻¹ NaCl. The initial pH of the prepared solution was 12.8, which was adjusted to 11.8 using concentrated HCl. The use of a synthetic model of the blowdown water allows us to study EC performance in a simplified system, and to eliminate variations in the composition of the experimental samples, which were observed in the field samples. The EC experiments were initially studied using the synthetic blowdown solution, and the most promising operating conditions were then evaluated with field samples of boiler blowdown water. This approach enabled the estimation of the treatment performance and energy consumption of EC, to evaluate its viability for integration into the SAGD water treatment process.

The field samples of blowdown water were collected on-site at a SAGD processing plant in Alberta, Canada. Thus, to obtain blowdown water, produced water was taken from upstream of the softening process and was treated by ion-exchange to remove its hardness. To produce blowdown water, the produced water (now boiler-feed water) was concentrated through steam generator at 80% steam quality, (i.e., 80% of the boiler-feed water mass was converted to steam). The samples were stored in a refrigerator at 4 °C, and some precipitation was observed during storage due to their high silica content, however upon heating to about 85 °C, the precipitate was observed to re-dissolve. Analysis of samples before and after three months of storage did not indicate any changes in their chemical compositions. All experiments were repeated at least three times, and showed reproducible behavior.

The characteristics of these samples, and the samples collected following EC treatment, were studied using a range of analytical techniques. The concentrations of silicon, sodium, calcium, magnesium, and potassium were determined using inductively coupled plasma optical emission spectroscopy (ICP-OES, Thermo Scientific IRIS Intrepid II, USA). To determine the coulombic efficiency of the EC process, the concentration of aluminum was determined by ICP-OES analysis after acidifying the electrolyte with 5.0% nitric acid to dissolve any aluminum precipitates. The concentrations of chloride and sulfate ions were determined using anion-chromatography (Thermo Dionex ICS-2000). The organic content was determined using a total organic carbon (TOC) analyser (OI Analytical Aurora 1030 W). The detailed organic composition was characterized using a Bruker 12 T SolariX ultra-high-resolution Fourier transform ion cyclotron resonance mass spectrometer (FTICR-MS). The samples were analyzed in electrospray ionization negative ion mode (ESI-N) and atmospheric pressure photoionization in positive ion mode (APPI-P) [see (Oldenburg et al. 2014) and supporting information for further details]. Prior to analysis, samples were filtered using 0.45 μm polypropylene syringe-filters. The pH of the treated water was determined using a pH meter (Mettler Toledo S220). The surface morphology of the electrode fouling and the precipitate materials after EC treatments were studied by Scanning Electron Microscopy-Energy Dispersive X-Ray Spectroscopy (SEM-EDS) (Phenom PROX Desktop).

Voltammograms and Tafel plots were obtained using an electrochemical workstation (PGSTAT204/FRA32M, Metrohm Autolab, B.V., The Netherlands). Potentiodynamic experiments were performed at temperatures 40, 60, and 80 °C in a 25 mL three-electrode electrochemical cell using an aluminum working electrode, Ag/AgCl (3 M KCl) reference electrode, and a platinum wire counter electrode. The working electrode was prepared from the same aluminum electrode material used in the electrocoagulation process (grade 6061). The 3.5 cm by 7.0 cm working electrode was polished before each experiment and covered with an adhesive resin tape allowing a circular area of 2.8 cm² to be exposed to the electrolyte.

The physicochemical characteristics and analysis of the field sample of blowdown water are presented in Table 1. The total dissolved silicon was about 481 ppm (equivalent to 1,029 ppm SiO₂). The concentrations of contaminants are in the range expected for SAGD boiler blowdown, considering that these concentrations are a function of the steam quality (i.e., the proportion of water evaporated in the boiler) (Maiti et al. 2012). Analysis carried out on-site indicated that the silica concentration in blowdown water can reach about 4 g L⁻¹. Hence, a higher silicon concentration was used in the synthetic blowdown water.

The nature of the organic content of the boiler blowdown water was analysed using ultra-high resolution FTICR-MS in APPI-P and ESI-N ion modes. About 50 classes of organic compound were detected in APPI-P ion mode, while 32 compound classes were identified in ESI-N ion mode (see Figure S1). The boiler blowdown contains multi-oxygenated species, where classes containing 2–9 oxygen per molecule (O_2–9) have the highest contribution to the relative intensity of all the compound classes present. Most of the dispersed hydrocarbons are separated upstream, therefore the remaining organics are the soluble or emulsified constituents with significant content of functional oxygenated groups. The spectrum also contains a high concentration of sulfur containing species ranging from O_2–8S. The physicochemical characteristics reflect the complex chemistry of the blowdown water, including high concentrations of polar organic molecules originating from heavy oil/bitumen with significant oxygen, sulfur, and inorganic contents. The inorganic/organic content analysis is consistent with the results reported by Maiti et. al. (Maiti et al. 2012).

In the EC process, the amount of aluminum metal dissolved from the anode is expected to be proportional to the charge passed in the EC cell, and the efficiency of the metal dissolution is typically compared with the theoretical dissolution rate calculated from Faraday's law (Yasri & Gunasekaran 2017). Most research studies of EC have been carried out at room temperature. However, in the present study, the treatment was performed at higher temperatures because the water treatment process in SAGD operates at elevated temperatures (Maiti et al. 2012). The operating temperature has a significant impact on the dissolution rate of aluminum in EC, the solubility of the precipitates, the electrolyte conductivity, and the coagulation process (Vepsäläinen et al. 2009). Thus, to determine the effect of temperature, DC-EC experiments were performed with the synthetic blowdown water at different temperatures, using a fixed charge loading of 800 C L⁻¹. Figure 1(b) shows the percentage removal of silica and the measured coulombic efficiency of aluminum dissolution (ϕ) as a function of temperature. Coulombic efficiencies of greater than 100% were observed at all cases studied, and the efficiency of the dissolution increased with increasing temperature. These super-faradaic efficiencies indicate that non-faradaic chemical corrosion occurred at the electrodes (Canizares et al. 2005), which is favored at higher temperatures. Interestingly, if the aluminum electrodes were immersed in the blowdown water in the absence of an externally applied current for more than one day at both room temperature and 80 °C, no dissolved aluminum was detected, and the dissolved silica concentration remained stable. This phenomenon was likely caused by the preservation of the alumina (Al₂O₃) passive film, which may have been reinforced by silica or other species in the sample (Mechelhoff et al. 2013). This suggests that under the conditions studied, de-passivation and pitting corrosion were activated via electrochemical reactions (Mechelhoff et al. 2013).

Coulombic efficiencies greater than 100% are common in EC reactors, especially with aluminum electrodes (Ingelsson et al. 2020). The observation of increasing coulombic efficiencies and contaminant removal with temperature has previously been reported for EC using aluminum electrodes (Yılmaz et al. 2008). For example, Yilmaz et al. studied the removal of boron from wastewater using Al-EC and found that the removal increased significantly with temperature (in the range 10–40 °C) (Yılmaz et al. 2008). They attributed the increase in contaminant removal to the de-passivation of the aluminum oxide film at higher temperatures, as well as the enhancement of the interaction between boron and the coagulant. However, in the case of silica removal from synthetic blowdown water, it was found that an increase in the operating temperature of the DC-EC process led to a slight increase in silica removal (Figure 1(b)). As the temperature was increased from 40 °C to 80 °C, the silica removal increased from 24.4% ± 1.2 to 28.9% ± 1.4, for a charge loading of 800 C L⁻¹. The solubility of precipitating species and the rate of the processes that occur during the coagulation stage (including sorption/desorption, dissolution, precipitation) are all temperature dependent, and interpretation of the impact of temperature on Si removal is complicated. However, it can be concluded that the increase in the rate of aluminum coagulant production was responsible for the increased Si removal at higher temperatures.

Although the increase in the aluminum dissolution rate with temperature has been reported in the literature, the phenomenon has not been widely studied by voltammetry methods, which may shed light on the reaction mechanism (Panikulam et al. 2018). Figure 1(c) shows the voltammograms of an aluminum working electrode in a synthetic blowdown water system at different operating temperatures (40 °C, 60 °C, and 80 °C). The potentiodynamic polarization curves obtained when scanning from anodic to cathodic potentials demonstrate a corrosion potential (E_cor) of about −0.87 V vs. Ag/AgCl. There was a slight variation in the corrosion potential in the experiments carried out at different temperatures. However, a significant increase in the corrosion current (i_cor) was observed with increasing temperatures. The corrosion currents at 40 °C, 60 °C, and 80 °C were 0.14, 1.17, and 3.06 A m⁻², respectively. Similarly, the current densities at the pitting potentials (point ‘a’ in Figure 1(c)) increased from 0.81 A m⁻² [ln (i, A m⁻²) = –0.21] at 40 °C, to 7.24 A m⁻² [ln (i, A m⁻²) = +1.97], and 10.9 A m⁻² [ln (i, A m⁻²) = +2.38] 60 °C, and 80 °C, respectively. However, no significant variation in the pitting potentials was observed at different temperatures. The increase in the pitting current density with temperature indicates faster reaction kinetics of the pitting corrosion process at higher temperatures.

At all temperatures studied, the reverse anodic scan (points 2–3) exhibited higher currents compared to the forward scans, which produced hysteresis loops (point ‘b’ in Figure 1(c)). Higher currents on the reverse scan are likely due to the formation of pits and the lower surface pH (Fuladpanjeh-Hojaghan et al. 2019) which is an established phenomenon in aluminum corrosion studies (Ingelsson et al. 2020). Passivation occurred in the reverse scan at a potential of approximately −0.76 V vs. Ag/AgCl at all temperatures studied (point ‘b’ in Figure 1(c)). Passivation with a sharper drop in the current of the hysteresis loop was observed at a lower temperature due to the higher dissolved oxygen concentration (Cao et al. 2018). This effect was reported by Cao et al. (2018) who investigated the passivation of Al alloy at various temperatures in a marine atmosphere. They concluded that although increasing the temperature results in lower dissolved oxygen in the thin passive film on the Al surface, the formation of the thermodynamically stable protective AlO(OH) species is favored at the higher temperature (Cao et al. 2018). Thus, to ensure the continuous dosing of aluminum, and to prevent re-passivation of the electrode surface, the applied voltage should always be higher than the anodic pitting potential (Panikulam et al. 2018).

The cathodic curves (3–4), show hysteresis loops (point ‘c’) due to water reduction at potentials of about −1.53 V, −1.38 V, and −1.32 V vs. Ag/AgCl at 40 °C, 60 °C, and 80 °C, respectively. Water reduction at a more negative potential corresponds to a higher cell polarization potential, so these results suggest that operating at a higher temperature will reduce the reactor cell potential. Similarly, the cathodic current density was observed to increase with temperature indicating faster electron transfer.

The potentiodynamic polarization data indicates that passivation can negatively impact the dissolution of aluminum. However, the coulombic efficiencies and the silica removal rates both increased with increasing temperature. Thus, the performance of electrocoagulation could be enhanced at high temperatures. This effect is opportune since blowdown water in the SAGD process is already produced at a high temperature. Therefore, Al-EC would be well-suited for this application. Based on these results, EC water treatment experiments were performed at 85 °C.

The performance of PR-EC in terms of silica removal was evaluated for the treatment of both the synthetic and field samples of boiler blowdown water. The silica removal, the organic contaminants in the treated water, the morphology of the electrode surface layer, and the morphology of the precipitates were monitored at different polarity reversal periods, i.e., 10, 20, 40, and 60 s. The results were compared to those obtained from DC-EC experiments.

The white/yellow precipitate observed in the solutions after DC-EC treatment (Figure S2a,b) and the composition of these precipitates confirmed the presence of both aluminum and silica (see SEM-EDS Figure S2c,d). These results provide evidence of the success of DC-EC treatment for removal of silica from the solution.

The silica removal, determined by ICP-OES analysis, of the synthetic and field samples of blowdown during DC-EC treatment are shown in Figure 2. The silica concentration declined rapidly, and the removal was approximately proportional to the charge loading. The initial silica concentration was 485 ± 11 and 3,300 ± 40 mg L⁻¹ in the field sample and synthetic blowdown water, respectively. The charge loading required for >98% silica removal from the field sample was approximately 1,200 C L⁻¹, while more then 4,800 C L⁻¹ was required to remove 90% of silica from the synthetic sample.

These results indicate that EC can be utilized to remove silica from the concentrated blowdown solution in the SAGD process. However, on increasing the operation period in the following successive runs using the same electrodes (without removing the electrode fouling layer between runs) the removal efficiency at 4,800 C L⁻¹ dropped significantly. On the second cycle, the silica removal after a charge loading of 4,800 C L⁻¹ dropped from 88% to about 60%, and on the third cycle the silica removal did not reach 45%. These results indicate that with DC-EC, electrode fouling rapidly affects the silica removal performance, and continuous operation would not be feasible without very frequent cleaning of the electrodes.

Figure 3 shows the percentage removal of silica from synthetic samples using PR-EC with different polarization periods for three successive treatment cycles (again without cleaning the electrodes between cycles). During the first run, the removal rate in all of the PR-EC experiments were slightly higher than those obtained in the DC-EC system (Run 1) with a high silica removal (>98%) observed at 4,800 C L⁻¹ at all reversal periods. However, the removal efficiency at 4,800 C L⁻¹ dropped significantly after three consecutive treatment cycles to about 40% in DC-EC (Figure 2). With polarity reversal, the decrease in removal efficiency over consecutive cycles was less.

For a charge loading of 4,800 C L⁻¹ the removal efficiency fell to over 60% for polarization periods of 10 s, and 60 s after three cycles. However, with intermediate polarisation periods of 20 and 40 s, the silica removal remained higher at about 80% at the end of the third cycle (Figure 3). The results indicate that the presence of fouling on the electrodes decreased the EC performance, whereas the application of polarity reversal mitigated the effect of fouling on silica removal. Further analysis of the fouling layers was performed to investigate this effect and the role of the polarity reversal period on fouling (see section 3.5 below).

The silica removal achieved during three successive PR-EC treatment cycles (without cleaning the electrodes between cycles) of field samples of blowdown water is presented in Figure 4. Similar to DC-EC, >98% silica removal was achieved at a charge loading of around 1,200 C L⁻¹. Although some decrease in silica removal was observed in the second cycle (Run 2), the removal performance in the third cycle (Run 3) was similar to that obtained with a freshly cleaned electrode (Run 1).

An important difference between the synthetic and field sample of produced water is the presence of organic contaminants in the field sample, which may impact the EC treatment process. The amount of TOC removal was found to be relatively low, and the removal of TOC from field samples of blowdown water was similar for DC- and PR-EC. For example, the initial TOC concentration of 9,650 mg L⁻¹ was reduced to about 8,830 and 8,790 mg L⁻¹ after 4,800 C L⁻¹ of DC- and PR-EC treatment, respectively. The samples treated using DC- and PR-EC were also characterized using FTICR-MS in APPI-P and ESI-N ion modes (see Figure S3). The amount of organics removal observed from the FTICR-MS data are consistent with the findings from the TOC analysis. The treated samples show a decrease in the relative intensity of a few compound classes when compared to the original sample. This decrease was observed from the characterization of the organic compounds based on the distribution of compound classes, double bond equivalent (DBE, a measure of the level of unsaturation), and carbon number (see Figures S4 and S5).

Analysis using the ESI-N ion mode (Figure S4), shows a visible decrease after EC treatment in the relative intensity of the classes of organics with -O₂, -O₄, -SO₂ and -SO₄ groups. A closer inspection of the four classes using the carbon number distribution reveals that larger organic molecules containing these groups (with carbon number >15) were removed more effectively, while there was no significant change in the relative intensity of organic molecules containing these groups with a carbon number of less than 15 (e.g. Figure S5 for the -O₂ class). However, most organic compound classes were not significantly removed by EC treatment in either DC- or PR-EC. No significant differences were observed in the composition of the blowdown water samples treated with DC-EC versus PR-EC, in line with the TOC measurements reported for the two samples.

These results suggest that the mechanism of organic removal was by physicochemical adsorption of long chain, functionalized organic molecules. These molecules are expected to have surfactant like properties, enabling them to interact with the coagulated solids during EC treatment. There was no evidence from the high-resolution mass spectrometry data that any organic oxidation occurred during EC treatment.

Visualization of the electrodes after the three successive PR-EC cycles with the synthetic blowdown water samples are shown in Figure 5(a)–5(d). The fouling layers appeared to be made up of spheroidal particles, and the particle size varied significantly for the different polarity reversal periods. The electrode fouling layers after three consecutive PR-EC treatment cycles showed that the diameter of the precipitated particles increased significantly with the polarisation period. This effect is likely due to the kinetics of the deposition of the fouling material and the local pH conditions at the electrode interface. After polarity reversal, aluminum will initially dissolve in the alkaline solution that was formed at the cathode interface during the prior half cycle, leading to rapid coagulation due to the surrounding hydroxide anions and nucleation/precipitation of aluminum hydroxide (and/or aluminosilicate) particles. With increasing the polarity reversal period, the precipitated particles formed on the electrode surface have more time to grow to a larger diameter before the pH conditions are changed again.

The morphology of the anodic fouling layer observed on the DC-EC anode and cathode (Figure 5(e) and 5(f)) was a hard, crystalline deposit. The crystalline deposits formed in DC-EC were difficult to remove from the electrode (requiring physical scraping). The morphology of the fouling layers were different for the anode and the cathode, and also varied between the synthetic and field samples of blowdown water (see Figures 5(e) and 5(f), S2 and S3).

Figure 6(a) shows the mass of fouling formed on both electrodes (the anode and the cathode, green columns) following DC-EC and PR-EC treatment of synthetic blowdown water with a charge loading of 2,000 C L⁻¹. The mass of fouling was determined by the differences in the mass of the dried electrode before and after treatment. The amount of Al in the fouling was also determined by scraping the fouling layer off the electrodes, digesting in acid, and measuring the Al concentration by ICP-OES, and normalizing to the electrode area (Figure 6(a), brown columns). The total fouling-mass and the amount of Al in the fouling are correlated, which indicates that the Al content of the fouling does not change with the EC operating conditions and suggests that similar compounds are formed in the fouling layer under the range of conditions studied. However, the amount of fouling depends on the operating conditions, with higher fouling with DC-EC compared with PR-EC operation. This result shows the benefit of periodically changing the polarity during EC treatment. In addition, the amount of fouling decreased when the polarization reversal period decreased.

The coulombic efficiency for aluminum metal oxidation is shown in Figure 6(b). The total coulombic efficiency of the Al dissolution is a combination of the Al in the solution (dissolved and in dispersed solids), and that in the fouling layer, relative to the theoretical dissolution calculated from Faraday's law. Figure 6(b) shows both the coulombic efficiency calculated from the amount of Al in the digested electrolyte (analysed by ICP-OES), and the Al accumulated on the electrodes (determined by scraping the fouling layer from the electrode, digesting in acid and analysis by ICP-OES).

The results show that the contribution of Al in the electrolyte to the total coulombic efficiency is lowest in DC-EC (44% in solution vs. the total 137%) and PR-EC with 10 s reversal (44.8% vs. 105%). For PR-EC with longer reversal periods (40–120 s), the rate of the Al in the electrolyte significantly higher, i.e., in the range 65–85% (with a total coulombic efficiency of 135–140%). The ratio of the contributions to the coulombic efficiency from the electrolyte to that in fouling was the highest for PR-EC 20 s, i.e., 1.42. This ratio decreased with increasing the polarization period, to 0.95 at 40 s, 0.92 at 60 s, 0.92 at 120 s, and was only 0.47 for DC-EC. This analysis indicates that a polarization period of 20 s was most effective for providing aluminum coagulant to the solution for Si removal, rather than accumulating aluminum containing solids on the electrodes.

The results in Figure 6(b) indicate that the total coulombic efficiency obtained with PR-EC increased significantly as the polarization period increased from 10 s (105%) to 20 s (140%). Further increase in the polarity period from 20 to 60 s led to a slight decrease in the coulombic efficiency to around 130%. At the longest polarization period studied, the coulombic efficiency increased slightly to around 135%. The coulombic efficiency obtained with PR-EC at polarization periods of 20–120 s was similar to that obtained with DC-EC (137%). This may indicate that operating with longer polarity periods than 20 s had a limited impact on the super-faradaic dissolution rate of aluminum. The PR- EC-10 s showed the lowest total coulombic efficiency of around 105%. However, given that the efficiency was greater than 100% in all cases, it is unlikely that the variations observed in coulombic efficiency were associated with side reactions. Reduction in the coulombic efficiency with shorter polarity reversal times may be due to the capacitive charge of the electrochemical double layer (Fekete et al. 2016), however, this effect is expected to be small (Ingelsson et al. 2020). It seems likely that the lower coulombic efficiency at the shortest polarization time (10 s) was associated with a reduction in the rate of non-faradaic dissolution of aluminum, due to differences in the passivation behavior, rapid pH changes, and the kinetics of pitting corrosion. With fast reversal, the range of surface pH is expected to be lower as there is insufficient time for the pH boundary layer to develop. This may explain a reduction in the rate of non-faradaic dissolution, which occurs when the interfacial pH is either acidic (Fuladpanjeh-Hojaghan et al. 2019) or basic (Ingelsson et al. 2020).

The dissolution of more aluminum at higher PR-EC periods (>20 s), coupled with the applied experimental conditions of a high pH media, may contribute to the more pronounced fouling layer observed at the 40 s–120 s PR-EC (see Figure 6(a). This data shows that increasing the polarization period results in an increase in the amount of fouling accumulated on the electrodes. However, the amount of fouling coated on the electrodes after-treatment of the synthetic blowdown water was significantly less with PR-EC than observed on both the cathode and anode using DC-EC treatment (Figure 6(a)). The amount of fouling after PR-EC-20 s was 14.6 mg cm⁻², around 63% less than the amount of fouling obtained with DC-EC.

In principle, the dissolved aluminum ions at the anode tend to form amphoteric hydroxides species that act as acid sites (Equations (4)–(6)) and cause the pH to drop (Lange & Speight 2005; Yasri & Gunasekaran 2017). The thickness of the acidic layer has been shown to be up to 0.5 mm thick and depends on the surrounding environment (Fuladpanjeh-Hojaghan et al. 2019). At the cathode, the electrochemical generation of hydroxide anions, OH̄, leads to the formation of a metal hydroxide base. However, in the DC-EC system, the process of coagulant formation at the anode interface (which is dominantly acidic) will be limited by the available OH̄ anions near the anode surface. Additionally, the process of coagulation will also be limited by the mass transfer (diffusion) of the dissociated aluminum species from the anode to the bulk solution (Yasri & Gunasekaran 2017). However, the (oxyhydr)oxide species are most likely to carry positive charges, which favorably react with the dissolved negatively charged silica to participate in the bridging process between particles (Milne et al. 2014). Hence, aggregation of aluminosilicates (See Equations (7) and (8)) will likely occur at the electrode/solution interfaces. These aggregations form a gel structure, that is in equilibrium with the monomers (Milne et al. 2014). During this process, the particles rearrange over time and slowly nucleate to form denser and harder gel or glassy-type precipitates.

The slow kinetics of the hardening process depends on the pH, temperature, and saturation level (concentration) of the molecules in the gel (Milne et al. 2014). Thus, the unchanging electrochemical condition near the electrode interfaces in the DC-EC process, and the continuous anodic dosing of Al, allow the saturation condition to take place, resulting in the formation of a hard fouling layer. In PR-EC, the periodic change of polarity promotes a faster conversion of the available hydrated metal to consume the hydronium ions near the electrode interface and to form coagulant hydroxide species (Equations (4)–(6)). Thus, the morphology and particle size of the precipitated solids are dependent on the polarity reversal period, the composition of the effluent, and the temperature. Increasing the polarity reversal period may lead to the formation of more crystalline species on the electrode. Thus, careful control of the polarity reversal periods helps the formation of a soft fouling layer that can be easily removed from the electrodes.

Filtration tests were performed to investigate the filterability of the coagulated solids from the treated effluent following EC treatment (see Figure 6(c)). The coagulated solids from PR-EC with polarisation times of 60 and 120 s were more easily filtered than those from DC-EC effluent and PR-EC with shorter polarization times. With the exception of PR-EC with a polarisation time of 10 s, the DC-EC effluent had the lowest average filtration flux of 0.50 L m⁻² s⁻¹ (Run 1), which reduced further for subsequent cycles. The lowest average flux for the PR-EC effluent was for a polarisation time of 10 s, at approximately 0.27 L m⁻² s⁻¹ for the third cycle of treatment. The average filtration flux decreased with the number of cycles of operation, and with decreasing polarity reversal period, likely due to the smaller particle size of the coagulated solids. This observation supports the hypothesis that increasing the polarization periods results in the formation of larger coagulant particles. Although small loosely bound fouling layers are advantageous, it was found that smaller particles formed in the treated water can impede the filtration flux. The reduction in particle size in the fouling layer may be associated with a reduction in the size of the coagulated particles in the precipitate. Based on the silica removal data presented above, there is no indication that smaller or larger-size particles are associated with different contaminant removal performance.

The electrical energy consumed during the EC processes was determined by numerical integration of the cell voltage during treatment (Figure S6). In the case of PR-EC and during the treatment of synthetic blowdown, the lowest energy consumption for a charge loading of 2,000 C L⁻¹ was obtained with a polarity reversal period of 20 s (250 ± 6.9 Wh m⁻³), followed by polarity reversal times of 40 s (266 ± 8.1 Wh m⁻³), 60 s (269 ± 8.0 Wh m⁻³), and 10 s (334 ± 8.9 Wh m⁻³). With the exception of a polarity reversal period of 10 s reversal period, the energy consumption for PR-EC was lower than that obtained with DC-EC (280 ± 8.2 Wh m⁻³). With a polarity reversal period of 10 s, a rapid increase in the absolute cell voltage was observed during treatment (Figure S6a), indicating increasing resistance during operation. However, the mass of fouling found on the electrodes after a charge loading of 2,000 C L⁻¹ with a polarity reversal period of 10 s was among the lowest (see Figure 6b). The increase in the resistance (for PR-EC-10s) may be due to the formation of a dense resistive layer of fine particles (Al(OH)₃/Al₂SiO₅) with a higher ohmic resistance compared to the particles formed when operating at higher polarisation periods. In addition, Al(OH)₃ and Al₂SiO₅ are known to possess dielectric properties that may lead to a high capacitance on the electrode surface (Kim et al. 2015). The increased capacitance is consistent with lower current efficiency for shorter polarity reversal periods.

Based on the energy consumption data, the passivation rate, which is dependent on the nature of the electrode/electrolyte interface (i.e. the pH and mass transfer rate), was the lowest for a 20 s polarity reversal period. For multiple cycles of treatment, the energy consumption for each cycle increased slightly, due to a higher cell voltage (see Figure S7), presumably due to the increased amount of electrode fouling.

The specific electrical energy and aluminum consumption required for 95% silica removal from synthetic and field blowdown water are summarized in Table 2. The electrical energy consumption to achieve 95% silica removal from synthetic and field blowdown water was only 2.1 and 0.52 kWh m⁻³, respectively. A charge loading of 4,800 C L⁻¹ was required to achieve 95% silica removal from the synthetic blowdown with a feed concentration of silica of 4,000 mg L⁻¹. Assuming a coulombic efficiency of around 140% (from Figure 6(a)), this corresponds to an aluminum dose of 624 g m⁻³ or 23.2 mmole L⁻¹. In the case of the field sample, which had a lower silica concentration, a charge loading of only 1,200 C L⁻¹ was required to achieve 95% silica removal, equivalent to an aluminum dose of 156.1 g m⁻³ or 5.7 mmole L⁻¹.

In order to compare the EC treatment of field and synthetic blowdown water, energy and electrode consumption were normalized based on the mass of SiO₂ removed (i.e. kWh g⁻¹ SiO₂ and g Al g⁻¹ SiO₂; see Table 2). The energy and the amount of aluminum required per kg of silica removed was remarkably consistent (<3% difference) between the synthetic and field blowdown samples. These results indicate that the presence of organic contaminants in the field samples did not significantly affect the silica removal. It is notable that the dose of aluminum required is very low, corresponding to the Al to Si molar ratio of only 0.34–0.33, much lower than would be expected if the removal mechanism was the formation of aluminosilicates (typically Al₂SiO₅).

Based on the data in Table 2, the cost of the replacement of the aluminum electrodes for the treatment of synthetic and field blowdown samples would be around 1.7 and 0.42 C$ m⁻³ respectively, based on a market price for aluminum of 2.72 C$ kg⁻¹. The cost of the corresponding electrical energy consumption is estimated to be around 0.32 and 0.08 C$ m⁻³ for synthetic and field blowdown respectively (based on 0.15 C$ kWh⁻¹). Thus, in both cases, the cost of replacement of the aluminum electrodes is more than five times the cost of electrical energy consumption. The total variable cost for electrode replacement and energy (excluding labour costs) would be around 2.0 and 0.5 C$ m⁻³, respectively. The operating cost and the energy consumption for the treatment of field samples of produced water are within the typical range of 0.4–4.0 kWh m⁻³ and 0.15–1.57 C$ m⁻³ reported for EC processes, respectively (Kuokkanen et al. 2013). Table 3 shows the performance of EC processes for a range of effluents, and the typical estimated operating costs (excluding reactor and operating labor costs). For blowdown water treatment, the electrical energy cost is relatively low, due to the high conductivity of the water, and the effectiveness of silica removal by Al-EC. As illustrated in Table 3, the contaminant concentration used in this study was significantly higher than that used in previous similar investigations of the performance of EC.

An electrocoagulation process that utilizes sacrificial aluminum electrodes was found to be effective for the treatment of a highly saline effluent containing silica with organic and inorganic contaminants. The treatment of these effluents is challenging and characteristic of the evaporative concentrated streams generated from boilers in industrials processes, such as SAGD and evaporative desalination (Den & Wang 2008). Although the process was found effective for the removal of concentrated silica from SAGD boiler blowdown, rapid fouling of the electrodes was observed in DC-EC, and the removal performance deteriorated rapidly during multiple batch cycles. A thick, hard layer of crystals was found to be firmly bound to both electrodes. Periodically reversing the polarity of the electrodes was observed to have multiple effects, including the rate of formation of the coagulant (based on coulombic efficiency), the morphology of the sludge (based on the filterability), the amount and morphology of the electrode fouling. However, it is clear from the results obtained that polarity reversal operation with reversal times of 10–60 s mitigated the problems with electrode fouling for Al-EC treatment of high silica content produced water. In contrast to DC-EC, the fouling layer in PR-EC was a loose spherical particulate layer, with a particle size that increases with the interval periods between polarity reversals.

The potentiodynamic polarization investigation indicated that the kinetics of aluminum dissolution increased with temperature. In addition, applying EC conditions for a prolonged polarization period increased the dissolution efficiency, as acidic and alkaline interfacial conditions had longer to develop at the anode and cathode respectively. The super-faradaic efficiencies for Al-EC (i.e. coulombic efficiencies of >100%) are associated with non-faradaic dissolution, which is favored under the acidic conditions that develop at the anode, and in the alkaline pH conditions at the cathode (2Al⁰ + 6H₂O → 2Al³⁺ + 6OH^‒ + 3 H₂) (Fuladpanjeh-Hojaghan et al. 2019). Thus, the pH at the anode and cathode interfaces are repeatedly inverted during the frequent polarity reversal, which decreases the favourability of chemical corrosion of both electrodes. Hence the amount of non-faradaic aluminum dissolution due to water electrolysis decreased for shorter polarity reversal periods, and thus the apparent coulombic efficiency was lower. In addition, coagulant and fouling particles have less time to grow with short polarity reversal periods, leading to a reduction in their size. This was observed to lead to reduced fouling at short polarity reversal periods, but the filterability of the sludge was also reduced. Hence, there is a trade-off between the amount of dissolved aluminum, the rate of fouling, the efficiency of the coagulation process, and the ease of solid-liquid separation that must be carefully considered (Yasri et al. 2020). Operation with polarity reversal periods of 1 min or less may also present challenges for the electrical control system, due to capacitive current surges. Although EC operation at a higher temperature may increase the efficiency by increasing coagulant dosing, it is only suitable for the treatment of high temperature effluents and requires the use of heat-resistant materials of construction, which may affect the reactor cost. Further engineering and techno-economic analysis is required to determine the viability of these operating conditions.

High-resolution mass spectrometry (FTICR-MS) revealed preferential removal of long-chain functionalized organics, probably by adsorption on coagulated solids. Only around 9% of the organics present were removed (based on TOC analysis) from the field samples after a charge loading of 4,800 C L⁻¹. However, the process required a charge loading of only 1,200 C L⁻¹ in order to remove more than 95% of the problematic silica from field samples of the blowdown (with an initial Si concentration of 481 mg L⁻¹). This corresponds to a consumption rate of aluminum of around 156 g m⁻³ of solution treated. The energy consumption for 95% silica removal was estimated to be only around 0.52 kWh m⁻³. The operating costs for energy and aluminum were estimated to be around 0.5 C$ m⁻³, with around 83% of these costs being for the replacement of the aluminum electrodes. To incorporate the finding of this study for industrial applications, further study of the fouling behaviour in large scale and continuous systems is needed.

This research was funded by the Natural Science and Engineering Research Council of Canada (EPG 515024-17, STPGP 506951-2017).

All relevant data are included in the paper or its Supplementary Information.