Effects of Electrodialysis Physical and Operating Parameters on the Performance of an Anaerobic Digestion–Electrodialysis System for Volatile Fatty Acid Production and Recovery

Abstract

Anaerobic digestion (AD) has two key challenges for producing volatile fatty acids (VFAs) from organic-rich wastes: (i) VFA production can become inhibited by VFA/NH₄ ⁺ accumulation, and (ii) VFAs must be selectively recovered. To address both challenges, VFAs and NH₄ ⁺ must be separated from the AD digestate, which has been previously accomplished using electrodialysis (ED) as a separation method. Previous studies have validated the technical feasibility of combining AD and ED to enhance VFA production and recovery but only for cases using one ED configuration. Therefore, the role of ED physical and operating parameters, which govern VFA/NH₄ ⁺ concentrations in the ED-treated digestate that is recycled back to the AD bioreactor, is unknown. We hypothesize that using ED parameters that enable high VFA/NH₄ ⁺ removal and recovery will enhance VFA production in a combined AD-ED system. Accordingly, our objective was to investigate how ED treatment conditions of AD digestate affect VFA production when the ED-treated digestates are recycled back into the AD bioreactors. To do this, we studied the effect of applied voltage, number of ion exchange membrane (IEM) pairs, and ED operation time on VFA/NH₄ ⁺ transport, energy consumption, current efficiency, and VFA production in a sequential batch AD-ED system. We found that VFA/NH₄ ⁺ transport scaled with voltage, increasing the number of IEM pairs improved VFA/NH₄ ⁺ transport and reduced energy consumption, and increasing ED operation time improved VFA/NH₄ ⁺ recovery with minimal impact on energy consumption. Recycling the ED-treated digestate back to the AD bioreactors led to small (7.91–19.8%, p = 0.30–0.66, three ED treatments) to moderate (40.4%, p = 0.06, one ED treatment) increases in VFA production relative to a control without ED-treated digestate recycling. However, we did not observe a strong or significant relationship between VFA production and the VFA/NH₄ ⁺ concentration in the AD bioreactors after ED-treated digestate recycling (R ² = 0.18–0.26, p = 0.49–0.58). Therefore, our results indicate that while recycling ED-treated digestate has the potential to increase VFA production in the AD bioreactor, the concentrations of VFAs and/or NH₄ ⁺ alone do not solely govern the extent of VFA production. Our results provide a foundation for further optimization of VFA production and recovery in an AD-ED system.

1. Introduction

Volatile fatty acids (VFAs) are short-chain carboxylic acids (two to six carbon-chains) that are economically valuable because they have versatile applications in the food, petrochemical, cosmetic, pharmaceutical, and biopolymer industries. , For example, acetic acid (C2 VFA) is the main compound in vinegar and is used as preservative in the food industry. Further, mixed solutions of VFAs can be used as precursors to make polyhydroxyalkanoates which are biodegradable bioplastics. , Therefore, due to the expansive applications of VFAs, the market demand for VFAs is growing where in 2024 the market was valued at 1.2 billion USD and is expected to reach 2.0 billion USD by 2033. The market value of VFAs ranges from 650 USD (acetic acid – C2 VFA) to 3,000 USD (caproic acid – C6 VFA) per ton. , Current VFA production methods rely heavily on nonrenewable petrochemical resources for VFA synthesis. This route of VFA production has high manufacturing costs and negative impacts on the environment and human health due to the excess greenhouse gas emissions associated with use of petroleum products. , As a result, biobased generation of VFAs from organic-rich wastes has gained increasing attention. , Biobased VFA production strategies offer advantages to VFA synthesis because they recycle organic waste streams (that would otherwise be diverted to landfills), reduce stress on nonrenewable resources, and promote a shift toward a circular bioeconomy. ,, Anaerobic digestion (AD) is one such microbial-based waste valorization method that produces VFAs from organic-rich waste streams (e.g., food waste, municipal waste, and sewage). ,

VFA production in AD faces two major challenges. First, VFAs and ammonium (NH₄ ⁺) accumulate in anaerobic digesters as the organic wastes are biodegraded, , and VFA and NH₄ ⁺ accumulation can have inhibitory effects on the microorganisms. ,− Second, VFAs are present together with other solutes in the digestate (e.g., salts including high levels of NH₄ ⁺, neutral solutes, organic acids, and alcohols). Therefore, VFAs must be selectively separated from the digestate for downstream applications. One approach to both minimize inhibitory effects of NH₄ ⁺ and VFAs on AD and selectively recover VFAs is to couple AD with a separation process. − Various membrane-based separation technologies have been used for VFA recovery including nanofiltration, electrodialysis (ED), membrane contactors, and membrane distillationeach offering different advantages and limitations. ,,, However, to minimize inhibitory effects of both NH₄ ⁺ and VFAs, a separation technique capable of both NH₄ ⁺ and VFA separation is required.

ED is a separation technology that can be integrated with AD for both (i) VFA recovery and (ii) VFA production enhancement in AD because ED can separate VFAs and NH₄ ⁺ from digestate. , ED is an ion exchange membrane (IEM)-based technique that uses electricity to selectively drive transport of anions and cations through anion and cation exchange membranes (AEMs/CEMs), respectively. The AEMs and CEMs separate concentrate and diluate compartments in the ED cell. Over time, the concentrate and diluate compartments become concentrated and depleted with ions, respectively. ED has shown to effectively remove and recover VFAs from digestates. ,, However, relevant ED performance metrics and ED process parameters for a combined AD-ED system remain underexplored. First, NH₄ ⁺ removal from the digestate during ED is often not reported, which is important to evaluate given the inhibitory effects of NH₄ ⁺ on AD. Second, while there is evidence in the literature of the economic benefits of producing and separating VFAs through combined biological and electro-driven processes, the energy consumption of ED in AD-ED systems is not well established. ,,− Finally, there is limited understanding of how the ED physical and operating parameters impact ED performance in terms of VFA and NH₄ ⁺ transport (i.e., removal and recovery) and energy consumption when treating digestates. − In ED, the physical and operating parameters (e.g., applied current and voltage, IEM properties, number of IEM cell pairs, flow rate, and ED operation time) affect ion recovery, ion removal efficiency, and the energy consumed to recover a given unit of ions and should be further explored for applications of VFA recovery. ,−

The technical feasibility of an AD-ED system, which operates by first treating digestate by ED and recycling the ED-treated digestate back to the AD bioreactor, has been demonstrated. − ,, AD-ED systems can lead to enhanced H₂ gas and/or VFA production compared with AD alone (i.e., an unamended control). − ,, However, only three studies have evaluated the effect of recycling ED-treated digestate back to an AD bioreactor on VFA production compared with an unamended control, and these studies have only used one ED configuration (i.e., one set of ED process parameters) in the AD-ED system. − Therefore, the effect of ED physical and operating parameters on VFA production in the coupled AD step is unknown. We hypothesize that using ED physical and operating parameters that enable high VFA/NH₄ ⁺ removal and recovery will enhance VFA production in a combined AD-ED system.

Accordingly, the overall goal of this study was to determine the effects of ED physical and operating parameters on the performance of a sequential batch AD-ED system. To do this, we first operated a bench-scale ED cell to study the effects of applied voltage, number of IEM pairs, and operation time on ED performance when treating AD digestate. Next, we operated bench-scale AD bioreactors to determine the effect of subsequently recycling ED-treated digestate to the bioreactors on VFA production. Overall, we found that applied voltage, number of IEM pairs, and operation time substantially impacted ED performance, including VFA recovery, during ED. Further, results showed recycling the ED-treated digestate to the bioreactors led to small (7.91–19.8%, p = 0.30–0.66) to moderate (40.4%, p = 0.06) VFA production enhancement relative to the control (i.e., no recycling). However, we did not observe a strong or significant relationship between VFA production and VFA/NH₄ ⁺ concentration in the bioreactors after ED-treated digestate recycling (R ² = 0.18–0.26, p = 0.49–0.58). Our results indicate that while recycling ED-treated digestate to the AD bioreactors may enhance VFA production, the VFA/NH₄ ⁺ concentration in the AD bioreactor alone does not solely govern the extent of VFA production. This is the first study to comprehensively evaluate the effects of ED physical and operating parameters on ED performance when treating digestate and to evaluate the effect on VFA production of recycling ED-treated digestates with varying VFA/NH₄ ⁺ concentrations to AD bioreactors. Our results provide insight on how ED physical and operating parameters influence VFA recovery and VFA production when ED is coupled with AD bioreactors.

2. Materials and Methods

2.1. Overview of the Sequential Batch AD-ED System

For this bench-scale study, we chose to operate the AD-ED system as a sequential batch system because (i) a previous study showed that in a fully integrated AD-ED system, the electric current from ED harms the microbial community in the digestate, and (ii) operating times for batch AD (days to months) and batch ED (minutes to hours) vary drastically. ,, An overview of the sequential batch AD-ED system is shown in Figure , and a simplified description is provided as follows. First, the stage 1 AD bioreactors were operated for 25 days. Then the digestate in the stage 1 bioreactors was removed from the bioreactors and filtered for ED treatment. Next, the filtered stage 1 digestate was treated by one of seven ED treatments (T1–T7), where each treatment had a different combination of physical and operating parameters. After the stage 1 digestate was treated by the seven ED configurations, four ED-treated digestates (specifically for ED treatments T1, T4, T6, and T7; discussion of selected treatments is in Section ) were used to replace 100 mL of the digestate in each stage 2 bioreactor after 25 days of operation. The stage 2 bioreactors continued to run for an additional 29 days (day 25 to day 54) to study the effect on VFA production of recycling ED-treated digestate back to the stage 2 bioreactors. Detailed descriptions of the equipment, instruments, and experimental methods used for each step of the sequential batch AD-ED system are described in Sections , 2.3, S2, and S3.

1.

Schematic of the sequential batch AD-ED system. T1, T4, T6, and T7 stand for the treated digestates from the corresponding ED treatments added to the stage 2 bioreactors. FW stands for food waste, and TWAS stands for thickened waste activated sludge.

2.2. Setup and Operation of the AD Bioreactors

The organic waste used in the bioreactors was a mixture of food representing food waste (FW). FW was selected as the organic substrate because of its significant production in the United States, and because of the increasing quantity of food waste that is processed by AD. , Thickened waste activated sludge (TWAS) from a full-scale water resource recovery facility was used to inoculate the bioreactors. Properties of the FW and TWAS are in Section S1. TWAS was heat-treated in a water bath under 65 °C for 15 min to reduce methanogenic activity and promote VFA production. Three identical bioreactors (stage 1) were set up in polypropylene bottles (3.8 L) to generate the digestate for ED treatment. The FW (276 g; equivalent to 30 gCOD·L^–1), 1.5 L heat-treated TWAS, and a salt medium were added to give a working volume of 3 L. The salt medium, which was used to maintain a balanced osmotic pressure for cell viability, , was comprised of NaCl (0.9 g·L^–1), MgCl₂·6H₂O (0.2 g·L^–1), and CaCl₂·2H₂O (0.1 g·L^–1) and was adjusted to pH 8.25 with NaOH for a slightly basic starting pH to promote VFA production. This salt medium has been used in previous studies and does not hinder VFA production. , To remove O₂, the liquid and gas phases were purged with N₂ for 15 min. The bioreactors were operated in a dark room kept at a mesophilic condition (37 °C) for 25 days until the soluble chemical oxygen demand (sCOD) concentrations increased by 10 gCOD·L^–1 (indicative of VFA production). The liquid fraction of the three bioreactors was then combined and filtered (0.2 μm polyether sulfone membrane filters). Filtration removed active microbes from the digestate before the digestate was treated by ED. The characteristics of the combined liquid fraction of the digestate after 25 days of operation are in Table . Analytical and experimental details for determining the composition and characteristics of the digestate are in Section S2. The filtered digestate was subsequently treated by ED (see Section for details).

1. Composition and Characteristics of the Digestate after 25 Days of Stage 1 Operation.

VFAs (mM)		inorganic ions (mM)		other
acetic (C2)	57.3	NH4 +	62.6	pH (preadjustment)	5.76
propionic (C3)	4.7	K+	8.8	pH (postadjustment)	7.4
i-butyric (C4)	5.5	Mg2+	3.3	κ (mS·cm–1)	8.1
butyric (C4)	24.3	Ca2+	3.2	TOC (g·L–1)	4.7
i-valeric (C5)	6.5	Cl–	5.3	TOC VFAs (g·L–1)	3.72
valeric (C5)	2.9	Na+	21.7	sCOD (g·L–1)	14.8
i-caproic (C6)	0.1			Ortho-P (mg-P·L–1)	179
caproic (C6)	2.4

^a“C#” represents the number of carbons in each VFA.

^bpH was adjusted using NaOH prior to ED to ensure VFAs were >99.99% deprotonated.

^cConductivity.

^dTotal Organic Carbon.

^eAveraged Na⁺ concentration after NaOH addition.

^fOrthophosphate which includes H₃PO₄, H₂PO₄ ^–, HPO₄ ^2–, and PO₄ ^3–.

A second set of bioreactors (stage 2) was operated to study the influence of recycling ED-treated digestate to the stage 2 bioreactors on VFA production during AD. This set consisted of six groups of bioreactors set up in 250 mL glass media bottles (triplicates, 18 AD bioreactors in total). The bioreactors were set up identically to the stage 1 bioreactors except that the volume was smaller. On day 25, when the sCOD in the bioreactors increased by roughly 10 gCOD·L^–1, 100 mL of the liquid fraction in five of the six bioreactor groups were withdrawn, and 100 mL ED-treated digestate or DI water were added. Of these five bioreactor groups, four received ED-treated digestates from four ED treatment configurations (T1, T4, T6, and T7; ED treatment key and details are in Section and Table ), and one received DI water to provide the highest extent of VFA and NH₄ ⁺ dilution. The sixth and the last bioreactor group served as a control where no liquid was extracted or added. After ED-treated digestate addition, DI water addition, or no addition (i.e., control), the bioreactors continued to operate at 37 °C for an additional 29 days. We monitored the pH, sCOD, VFA, and NH₄ ⁺ levels in the bioreactors (methods in Section S2). Gas compositions were monitored over time to assess CH₄ production (methods in Section S2).

2. Physical and Operating Parameters for Each ED Treatment.

ED treatment	diluate composition	voltage (V)	number of IEM pairs	operation time (min)
T1	digestate	0.5	2	60
T2	digestate	1	2	60
T3	digestate	1	2	120
T4	digestate	0.5	4	60
T5	digestate	1	4	60
T6	digestate	1	4	120
T7	digestate	1	4	240
synthetic 1	VFA mix	1	2	60
synthetic 2	VFA mix	1	4	60

^aThe concentrate composition was 84 mM NaCl for all experiments.

2.3. Electrodialysis

2.3.1. Electrodialysis Setup

The ASA (Selemion, AGC Engineering, Japan) and FKL-30 (Fumasep, fumatech GmbH, Germany) IEMs were used as the AEM and CEM, respectively, for all ED experiments (IEM properties in Section S3). Batch ED experiments were performed using a commercial ED cell (PCCell 64002, PCCell GmbH, Germany). A batch ED configuration was selected to enable time-resolved monitoring of solute concentrations in the C and D compartments and to accommodate the limited available digestate volume (0.26 L). The membrane stack in the ED cell contained either 2 AEMs and 3 CEMs (i.e., two IEM pairs; “N = 2”) or 4 AEMs and 5 CEMs (i.e., four IEM pairs; “N = 4”). Keeping in mind that in commercial systems there are greater electrical resistances and higher capital and maintenance costs associated with a greater number of IEM pairs (Table S3), we chose a relatively low number of IEM pairs to specifically evaluate the role of doubling the number of IEM pairs on ED performance. Additional ED cell details are in Section S3. Figure shows the ED cell setup.

2.

Schematic of the ED cell setup. The example shown is for the two IEM pair case (i.e., 2 AEMs and 3 CEMs). VFAs move through AEMs toward the positively charged (+) anode. NH₄ ⁺ moves through CEMs toward the negatively charged (−) cathode. The symbols κ_C and κ_D represent the conductivity probes in the C and D reservoirs, respectively. The symbols ΔP _C and ΔP _D represent the pressure transducers connected to the inlet and outlet of the C and D compartments, respectively.

An electrode rinse solution (ERS) of 0.1 M Na₂SO₄ was recirculated between the electrolyte compartments of the ED cell from a 1-L reservoir at 150 mL·min^–1. The concentrate (C) reservoir consisted of 84 mM NaCl (to approximate the conductivity of the digestate, 8.2 mS·cm^–1), and the diluate (D) reservoir consisted of digestate or a synthetic VFA mix solution. The synthetic VFA mix solution consisted of 57 mM sodium acetate, 5 mM sodium propionate, 30 mM sodium butyrate, and 9 mM sodium valerate at pH 7.4 to approximate the VFA composition in the digestate (Table ). Prior to loading the digestate in the D reservoir, the pH of the digestate was adjusted to 7.4 using NaOH to ensure the VFAs were >99.99% deprotonated (pK _a ≈ 4.75). Given that electromigration is the main transport mechanism during ED, using pH conditions at which VFAs are predominantly in their anionic (i.e., deprotonated) form ensures that VFA transport is maximized (and therefore greater than under low pH conditions). The total volume of each reservoir plus compartment in the ED cell was approximately 0.26 L, and the solutions in each of the C and D reservoirs were pumped into the ED cell at 41.5 mL·min^–1. Differential pressure transducers (Omega) were connected between the inlet and the outlet of the C and D compartments to measure the pressure drop of each compartment. Conductivity probes (eDAQ, Colorado Springs, CO) were placed in the C and D reservoirs to measure the conductivities. The conductivities of the C and D reservoirs were used to monitor the total ion (i.e., salt) transport during ED.

2.3.2. Electrodialysis Operation

All electrochemical measurements were performed using a potentiostat (VMP-3, Bio-Logic, France) as the power source. Prior to each ED experiment, 100 mM NaCl was recirculated through both the C and D compartments to measure the baseline resistance of the membrane stack to evaluate its integrity given that the constituents in digestate have a high potential to foul IEMs (details in Section S4).

After measuring the baseline resistance, 84 mM NaCl was added to the C reservoir, and the digestate or the synthetic VFA mix solution was added to the D reservoir. Solutions were pumped into the C and D compartments in a single-pass configuration until the open circuit voltage (OCV) became stable (approximately ± 1 mV·min^–1). Once the OCV stabilized, the solutions were recirculated from each C and D reservoir (0.26 L each) in a closed-loop configuration. In one instance with two IEM pairs and one instance with four IEM pairs, the resistance of the membrane stack was measured using a linear sweep voltammetry (LSV) method (sweeping from 0 to 1.0 V) with the C and D reservoirs containing 84 mM NaCl and digestate, respectively. The baseline resistance, initial OCV, and membrane stack resistance with 84 mM NaCl and digestate for conditions with two and four IEM pairs are reported in Section S4. The current-voltage data obtained from the LSV resistance experiments with NaCl (C reservoir) and the digestate (D reservoir) were used to determine the limiting current densities (Section S5).

Once the NaCl and the digestate or VFA mix solutions were recirculating in the ED cell, the ED experiments were performed using a chronoamperometry (CA) method where a voltage of 0.5 or 1 V was applied across the membrane stack for 60 to 240 min. Voltages of 0.5 and 1 V were selected to perform experiments above the OCV and close to the ohmic region based on limiting current density experiments (Section S5). Throughout the ED experiment, ∼ 3 mL samples were collected in 5 mL amber glass vials from the C and D reservoirs at various time intervals for VFA and NH₄ ⁺ analysis. The pressure drop and the conductivity of the C and D compartments were continuously tracked. After the ED experiment was complete, an additional 6 mL sample was taken from the C and D reservoirs for pH, sCOD, and TOC measurements (Section S2). All ED experiments were performed in duplicate. The operating parameters including voltage, number of IEM pairs, and operation time for each of the seven ED treatments plus the ED experiments with the synthetic VFA mix solution are in Table . After each ED experiment, the ED cell underwent an extensive cleaning procedure to remove organic solutes that may have sorbed to (i.e., fouled) the IEMs or the ED cell components during ED (Section S6).

2.3.3. VFA Adsorption Tests

Two adsorption experiments were performed using the ED cell including one with two IEM pairs and one with four IEM pairs. All components of the ED cell were analogous to what was described in Section except no external voltage was applied (i.e., 0 V). The C reservoir contained 84 mM NaCl, and the D reservoir contained the synthetic VFA mix solution. Adsorption experiments were performed for 240 min, and ∼3 mL samples were taken from the C and D reservoirs at 60 min increments for VFA analysis. The OCV was recorded periodically.

2.3.4. Electrodialysis Performance Metrics

For each ED treatment, we evaluated VFA and NH₄ ⁺ recovery and removal from the digestate, VFA loss, electrical and total energy consumption per mol of VFA recovered and removed, and current efficiency for VFA recovery. Metrics of recovery, energy consumption, and current efficiency are relevant ED performance metrics, and metrics of VFA and NH₄ ⁺ removal are relevant for the ED-treated digestate addition to the stage 2 bioreactors. In this work, VFA and NH₄ ⁺ transport was the primary focus; however, AD also contains other salts (Table ). Therefore, for a select treatment (T6), we evaluated the transport of all inorganic ions in the system (discussed in Section ). Given that the relative VFA and NH₄ ⁺ transport trends were consistent across all treatments (e.g., short-chain VFAs transported preferentially over long chain VFAs), we expect the relative transport behavior between ions to be consistent across all ED treatments.

The percent solute (VFAs or NH₄ ⁺) recovery was calculated as

$$\mathrm{recovery}(\%)=\frac{C_{\mathrm{C},\mathrm{f}}{V}_{\mathrm{C}}}{C_{\mathrm{D},0}{V}_{\mathrm{D}}}\times 100$$ 1

where C _C,f (mol·L^–1) is the solute concentration in the concentrate reservoir at the end of the ED experiment, V _C (0.26 L) is the volume of the concentrate reservoir, C _D,0 (mol·L^–1) is the initial solute concentration in the diluate reservoir, and V _D (0.26 L) is the volume of the diluate compartment. The percent solute removal was calculated as

$$\mathrm{removal}(\%)=\frac{C_{\mathrm{D},0}{V}_{\mathrm{D}}-C_{\mathrm{D},\mathrm{f}}{V}_{\mathrm{D}}}{C_{\mathrm{D},0}{V}_{\mathrm{D}}}\times 100$$ 2

where C _D,f (mol·L^–1) is the final solute concentration in the diluate reservoir. The percent VFA loss was calculated to account for VFA lost to sorption to various components of the ED cell apparatus and was calculated as

$$\mathrm{loss}(\%)=\frac{{(C}_{\mathrm{D},0}{V}_{\mathrm{D}}-C_{\mathrm{D},\mathrm{f}}{V}_{\mathrm{D}})-C_{\mathrm{C},\mathrm{f}}{V}_{\mathrm{C}}}{C_{\mathrm{D},0}{V}_{\mathrm{D}}}\times 100=\mathrm{removal}(\%)-\mathrm{recovery}(\%)$$ 3

The VFA loss served as an indicator of the potential for IEM fouling by VFAs.

The electrical energy consumption per mol of VFAs recovered (EC _Elec., kWh·mol_VFA ^–1) was determined as

$${\mathrm{EC}}_{\mathrm{Elec}.}=\frac{{\int }_0^t\mathrm{UI}(t)\mathrm{d}t}{C_{\mathrm{C},\mathrm{f},\mathrm{VFA}}{V}_{\mathrm{C}}}$$ 4

where U (V) is the voltage, I(t) (A) is the current, and C _C,f,VFA (mol·L^–1) is the final VFA concentration in the C reservoir. The electrical energy consumption per mol of VFAs removed is reported in Section S7 and followed similar trends as EC _Elec. (R ² = 0.74, p = 0.01).

Further, we measured the total energy consumption per mol of VFAs recovered (EC _Elec.+Pump, kWh·mol_VFA ^–1) which includes electrical energy and pumping energy. The total energy consumption was calculated as

$${\mathrm{EC}}_{\mathrm{Elec}.+\mathrm{Pump}}=\frac{{\int }_0^t\mathrm{UI}(t)\mathrm{d}t+({\phi }_{\mathrm{C}}\mathrm{\Delta }{P}_{\mathrm{C}}+{\phi }_{\mathrm{D}}\mathrm{\Delta }{P}_{\mathrm{D}})\mathrm{\Delta }{t}_{\mathrm{total}}}{C_{\mathrm{C},\mathrm{f},\mathrm{VFA}}{V}_{\mathrm{C}}}$$ 5

where φ_C and φ_D (41.5 mL·min^–1) are the flow rates of the solutions pumped into the C and D compartments, respectively, ΔP _C and ΔP _D (Pa) are the pressure drops across the C and D compartments, respectively, and Δt _total is the operation time of the ED experiment (60–240 min).

The current efficiency for VFA recovery (CE _VFA, %) was calculated as

$${\mathrm{CE}}_{\mathrm{VFA}}(\%)=\frac{C_{\mathrm{C},\mathrm{f},\mathrm{VFA}}{V}_{\mathrm{C}}\times \mathit{zF}}{N{\int }_0^{t}I(t)\mathrm{d}t}$$ 6

where F (C·mol^–1) is Faraday’s constant, z is the charge of the ion of interest (z = 1 for VFAs), and N is the number of IEM pairs in the ED cell (N = 2 or 4).

2.4. Statistical Analysis

We performed simple linear regression analysis between ED performance metrics and VFA carbon-chain length to assess whether VFA carbon-chain length affected transport in ED. We also performed linear regression analysis between various ED performance metrics to understand how ED performance metrics related to each other. Further, we performed two-sided Welch’s t-tests at a 95% confidence interval to compare ED performance between two groups (e.g., two IEM pairs vs four IEM pairs). The performances of AD bioreactors in stage 2 were compared using a mixed model ANOVA unless otherwise stated and p-values are reported. In the model, bioreactor treatment was the “between subject”, and time was defined as the “within subject”.

3. Results and Discussion

3.1. Effect of ED Physical and Operating Parameters on ED Treatment of Digestate

We evaluated the effects of ED physical and operating parameters including voltage, number of IEM pairs, and operation time on metrics relevant to ED performance including (i) VFA and NH₄ ⁺ recovery, (ii) electrical energy consumption per mol of VFAs recovered, (iii) total energy consumption per mol of VFA recovered, and (iv) current efficiency for VFA transport. Additionally, given that the ED-treated digestates were added to the stage 2 bioreactors, we also evaluated the impact of ED physical and operating parameters on VFA and NH₄ ⁺ removal from the digestate. In some cases, we discuss VFA loss due to sorption. For each ED treatment (T1–T7), the energy consumption per mol of VFAs removed, time-dependent current (density), final pH of the C and D reservoirs, conductivities of the C and D reservoirs, and current efficiency for VFA recovery are reported in Section S7 and referenced where relevant throughout Section .

3.1.1. Effect of VFA Structure on Solute Recovery and Removal during ED

Prior to discussing the effect of each tested ED physical and operating parameter on ED performance, we will first assess trends related to VFA structure that were consistent for all seven ED treatments. We also present results for NH₄ ⁺ to provide a comparison between transport of VFAs and an additional constituent whose accumulation can inhibit further VFA production. Figure shows an example of the VFA and NH₄ ⁺ recovery and loss for a representative ED treatment (T6; analogous figures for all ED treatments are in Figures S6–S12).

3.

Percentage VFA and NH₄ ⁺ recovery and loss from the digestate after ED Treatment 6 (T6). The summation of recovery and loss is the solute removal. T6 was operated at 1 V, with four IEM pairs, and for 120 min. “C#” represents the number of carbons in each VFA. The descriptor “i” indicates that the VFA is the isomer to the linear VFA. ∑VFA represents the total VFA recovery/loss. Each bar is the average value from duplicate experiments, and error bars represent the standard deviation.

For all ED treatments, we observed a strong negative relationship between VFA recovery and VFA carbon-chain length (R ² = 0.79–0.92, p < 0.001; Table S6) as well as between VFA removal and VFA carbon-chain length (R ² = 0.66–0.86, p < 0.005; Table S6). Therefore, greater VFA transport was observed for VFAs with shorter carbon-chain lengths. The inverse relationship between VFA transport though IEMs and VFA size (i.e., carbon-chain length) has been observed previously and primarily stems from increased steric hindrance, drag, and VFA-AEM interaction strength as the size of the VFA increases. ,− Further, we observed that VFA recovery and VFA removal moderately to strongly positively correlated with each other for each ED treatment (R ² = 0.56–0.99, p < 0.001; Table S7), and VFA removal was greater than VFA recovery for all ED treatments (p = 0.02; Table S8). The higher VFA removal than recovery occurred because VFAs sorb to components of the ED cell, namely the IEMs. ,

We also observed that linear VFAs were recovered (p = 0.01) and removed (p < 0.001) to a greater extent than their isomeric VFA counterparts (Table S9), which has been observed previously and can also be attributed to greater steric hindrance and greater drag associated with isomeric VFAs than linear VFAs. Additionally, VFA recovery (p = 0.29) and VFA removal (p = 0.19) were lower than NH₄ ⁺ recovery and removal, respectively (Table S10). The lower permeation of VFAs than NH₄ ⁺ can be attributed to a combination of the larger size and lower mobility of VFAs than NH₄ ⁺, the hydrophobic properties of VFAs, and the concurrent permeation of other inorganic anions (e.g., Cl^–) with VFAs in the digestate. −

3.1.2. Effect of Applied Voltage on ED Performance

We evaluated the effect of applied voltage on ED performance for the cases with (i) two IEM pairs (Figure S13, Section S8) and (ii) four IEM pairs operated for 60 min (Figure A,B). General trends were the same regardless of the number of IEM pairs in the ED cell; therefore, specific trends will only be discussed for the case with four IEM pairs. VFA recovery and removal were 11.1 and 11.7 percentage points higher, respectively, when operating at 1 V compared with 0.5 V. Similarly, NH ₄ ⁺ recovery and removal were 6.5 and 7.7 percentage points higher, respectively, when operating at 1 V compared with 0.5 V (Figure A). The greater solute recovery/removal for cases with 1 V compared with 0.5 V can be attributed to the higher measured current density at 1 V than 0.5 V since the current density is proportional to ion transport during ED (Figure S3, Section S7). Increased solute transport with higher voltages has been observed previously. ,

4.

(A, C, E) Percentage of total VFA (∑VFA) and NH₄ ⁺ recovery (“Rec.”) and loss. The summation of recovery and loss is the solute removal. (B, D, E) Electric (“Elec.”) and pumping (“Pump”) energy consumption (EC) per mol of VFAs recovered. The summation of the electric and pumping energy is the total energy consumption. Data in Panels A and B are from experiments performed with four IEM pairs for 60 min (T4 vs T5). Data in Panels C and D are from experiments performed at 1 V for 60 min (T2 vs T5). Data in Panels E and F are from experiments performed at 1 V with four IEM pairs (T5 vs T6 vs T7). All bars are the average values from duplicate experiments, and error bars represent the standard deviations.

Although the voltage was doubled from 0.5 to 1 V, solute transport did not correspondingly double. The only moderate increase (6.5–11.1%) in VFA and NH₄ ⁺ recovery from 0.5 to 1 V occurred because the ED cell resistance increased faster when operating at 1 V (0.23 Ω·min^–1) compared with 0.5 V (0.14 Ω·min^–1) to thereby hinder ion transport to a greater extent, as described by Ohm’s law: V = IR (Figure S14). The ED cell resistance increased faster when operating at 1 V because the solute flux from the diluate compartment was higher at 1 V than 0.5 V (e.g., J _VFA:0.5V = 6.9 × 10^–5 mol·m^–2·s^–1; J _VFA:1V = 1.0 × 10^–4 mol·m^–2·s^–1), which led to correspondingly lower conductivities (Figure S4), and therefore higher ED cell resistances, in the diluate compartment at 1 V. Results are consistent with a previous study that evaluated the effect of voltage on solute (e.g., NH₄ ⁺ and SO₄ ^2–) flux where doubled voltage only moderately enhanced rates of ion migration.

The electrical energy consumption and total energy consumption per mol of VFA recovered (eqs –) were 104 and 80.7% greater, respectively, when operating at 1 V compared with 0.5 V (Figure B). The observed greater energy consumption with 1 V resulted from the doubling of the applied voltage (numerator of eqs –) yielding an increase in current density (Figure S3) but only a moderate increase in VFA recovery (11.1%, Figure A). We note that for all configurations, the total energy required to recover VFAs from digestate was 0.005–0.026 kWh per mol of VFA which is consistent with electro-driven processes reported in literature (approximately 0.002–0.5 kWh per mole of VFAs recovered). ,, Overall, there was a trade-off between solute removal/recovery and energy consumption as a function of applied voltage, where a higher voltage resulted in higher solute recovery/removal, but at the expense of higher energy consumption per mol of VFAs recovered. Additionally, ED operation at high voltages has shown to degrade the surface of IEMs after long-term operation; therefore, the negative relationship between voltage and membrane structural stability may provide motivation to operate ED at lower voltages for VFA recovery.

3.1.3. Effect of Number of IEM Pairs on ED Performance

We evaluated the effect of the number of IEM pairs (N = 2 vs N= 4) on ED performance for cases operated at (i) 0.5 V for 60 min (Figure S15, Section S9), (ii) 1 V for 60 min (Figure C,D), and (iii) 1 V for 120 min (Figure S16, Section S9). Trends were generally consistent across cases; therefore, specific trends will primarily be discussed for conditions operated at 1 V for 60 min. VFA recovery was 11.3 percentage points lower with two IEM pairs compared with four IEM pairs (Figure C). Conversely, VFA removal was not significantly different with two IEM pairs compared with four IEM pairs. Accordingly, VFA loss, an indicator of IEM fouling, was 12.6 percentage points greater with two IEM pairs compared with four. Assuming all VFA loss was due to sorption to the membranes, the VFA loss was significantly greater with two IEM pairs (169 mmol·m^–2) than with four IEM pairs (35.2 mmol·m^–2), a trend that was also observed for the other N = 2 vs N = 4 comparisons (p < 0.001; Tables S11–S12).

We next compared VFA loss between cases with (i) digestate in the D reservoir operating at 1 V (Figure B), (ii) the synthetic VFA mix solution in the D reservoir operating at 1 V, and (iii) the synthetic VFA mix solution in the D reservoir operating at 0 V under conditions with two and four IEM pairs to isolate the effects of the digestate matrix, electric current, and number of IEM pairs on VFA loss/IEM fouling during ED. Detailed results from the experiments with the synthetic VFA mix solution are in Section S10. Results showed VFA loss was 7.5–20.1% for case i (1 V, real digestate), 8.2–14.3% for case ii (1 V, synthetic VFA solution), and <2% for case iii (0 V, synthetic VFA solution). On average, the VFA loss was greater when operating at 1 V with digestate compared with when operating at 1 V with the synthetic VFA solution, but this difference was not significant (p = 0.56; Table S14). The additional organic solutes and multivalent ions present in the digestate may have had some effect on VFA sorption during ED given that several previous studies have found organic solutes and multivalent ions enhance membrane fouling. − Further, since the VFA loss was significantly greater for the 1 V synthetic VFA solution case than the 0 V synthetic VFA solution case (p = 0.01; Table S15), results indicated that the electric current played a significant role in VFA loss, as has been observed previously. −

The number of IEM pairs had opposing effects on VFA loss depending on the D reservoir matrix when operating at 1 V (there was no effect of number of IEM pairs when operating at 0 V). With the digestate, N = 2 led to greater VFA loss than N = 4 (as previously discussed; Figure C) whereas with the synthetic VFA solution, N = 2 led to lower VFA loss than N = 4 (Figure S19). Accordingly, when operating at 1 V, results indicate that with digestate, the electric current (which was always substantially greater for N = 2 than N = 4 cases; Figures S3 and S17) governed the extent of VFA loss whereas with the synthetic VFA solution, the membrane area available for VFAs to adsorb (i.e., sorption sites) played a greater role than electric current on VFA loss.

Regarding NH₄ ⁺ transport, NH₄ ⁺ recovery and removal were 16.1 and 14.9 percentage points lower, respectively, with two IEM pairs than four IEM pairs (Figure C). Overall, solute recovery (i.e., VFA and NH₄ ⁺) from digestate was greater for cases with four IEM pairs than two IEM pairs. The finding that more IEM pairs results in higher solute recovery has been similarly observed for nutrient recovery using ED. Although the N = 4 case has doubled the number of IEM pairs than the N = 2 case, we did not observe doubled solute transport as would be expected for conditions operating with a constant current. The only moderately greater transport for the N = 4 case stems from operating at a constant applied voltage where the membrane stack resistance affects ion transport (i.e., higher stack resistance with more membrane pairs; R _N=2 = 2.9 Ω; R _N=4 = 5.3 Ω; Table S3). Therefore, the 83% greater resistance associated with the N = 4 case led to a substantially lower current compared with the N = 2 case (I _0:N=2 = 375 mA; I _0:N=4 = 230 mA; Figure S3).

The electrical energy consumption and total energy consumption per mol of VFA recovered were 181 and 183% higher, respectively, with two IEM pairs than with four IEM pairs (Figure D). The relationship between number of IEM pairs and energy consumption can be explained by Ohm’s law (V = IR). Since a constant voltage was applied (1 V), and the case with two IEM pairs had a lower resistance than the case with four IEM pairs (R _N=2 = 2.9 Ω; R _N=4 = 5.3 Ω; Table S3), a higher current was measured with the two IEM pairs than four IEM pairs (Figure S3). Therefore, the case with two IEM pairs used more power (P = I × V) to recover VFAs than the case with four IEM pairs (eqs –). Overall, the case with two IEM pairs had greater power usage and lower VFA recovery than for the case with four IEM pairs; therefore, the energy consumption per mol of VFA recovered was substantially greater for the case with two IEM pairs.

In practice, it is beneficial to use as few IEMs as possible in the ED cell due to the cost and operational burden of a higher number of IEMs, especially when operating with complex feedwaters (e.g., digestate). However, our results showed that using more IEM pairs (i.e., N = 4) resulted in enhanced solute recovery, lower solute loss (which translates to lower sorption/fouling , ), lower energy consumption per mol of VFAs recovered, and higher current efficiency for VFA recovery (Figure S5) than when operating the ED cell with fewer IEM pairs (i.e., N = 2) during treatment of digestate. We attribute the higher current efficiency for VFA recovery observed with four IEM pairs compared with two to the lower VFA loss associated with four IEM pairs. Overall, it is important to balance the trade-off between number of IEM pairs in an ED cell with ED performance, given the challenges in maintaining IEMs and high ED cell resistances associated with large IEM stacks.

3.1.4. Effect of ED Operation Time on ED Performance

We evaluated the effect of ED total operation time on ED performance for cases operating at 1 V using (i) two IEM pairs (Figure S22, Section S11) and (ii) four IEM pairs (Figure E,F). Since the operation time went up to 240 min for only the cases with four IEM pairs, trends observed with four IEM pairs are discussed herein. The trends observed for the case with two IEM pairs were similar to those with four IEM pairs. As expected, VFA recovery and VFA removal had a moderate to strong positive correlation with ED total operation time (Rec: R ² = 0.82, p = 0.01; Rem: R ² = 0.50, p = 0.11; Figure E). Although VFAs were continuously recovered for up to 240 min, VFA recovery only increased linearly with time for the first approximately 60 min of operation (Figure A–C). In the following 60–180 min, the VFA recovery rate decreased (Figure B,C). The VFA recovery rate declined over time because the current declined over time (Figure S3), and the current declined because ED cell resistance increases as the conductivity of the D reservoir decreases (Figure S4).

5.

Total VFA (∑VFA) and NH₄ ⁺ transport to the concentrate compartment as a function of time for ED treatments operated at 1 V with four IEM pairs for (A) 60 min – T5, (B) 120 min – T6, and (C) 240 min – T7. The measured solute concentrations in the concentrate compartment (C_C,t) were normalized to the initial solute concentrations in the diluate compartment (C_D,0). The vertical dashed lines in Panels B and C indicate the times where the ED treatment(s) in the previous panels were terminated. All points are the average values from duplicate experiments, and error bars represent the standard deviation.

Similarly, NH₄ ⁺ recovery and removal had a moderate positive correlation with ED total operation time (Rec: R ² = 0.35, p = 0.21; Rem: R ² = 0.53, p = 0.10; Figure E). The correlation between VFA recovery and total operation time was stronger than that between NH₄ ⁺ recovery and total operation time since VFA recovery occurred for up to 240 min while NH₄ ⁺ recovery rate decreased substantially after 120 min (Figure C). Specifically, in the first 120 min of ED for T7, the concentration-normalized fluxes of VFAs and NH₄ ⁺ were 5.7 × 10^–7 m·s^–1 and 7.6 × 10^–7 m·s^–1, respectively, and decreased to 2.0 × 10^–7 m·s^–1 and 2.8 × 10^–8 m·s^–1 for VFAs and NH₄ ⁺, respectively, in the final 120 min.

The variations in transport rates over time between VFAs and NH₄ ⁺ can be rationalized from (i) higher mobility of NH₄ ⁺ (u = 7.6 × 10^–8 m²·V^–1·s^–1) than VFAs (u = 3.4–4.2 × 10^–8 m²·V^–1·s^–1) and (ii) selectivity between VFAs and other anions (e.g., Cl^–, Ortho-P; Table ) compared with the selectivity between NH₄ ⁺ and other cations (e.g., Ca²⁺, Mg²⁺, K⁺; Table ) in the digestate. First, ion mobility describes the drift velocity of an ion under an electric field and is directly proportional to the concentration-normalized flux of an ion. Accordingly, the higher mobility of NH₄ ⁺ than that of VFAs explains the initially faster transport of NH₄ ⁺. Observations are in agreement with a previous study that evaluated VFA flux over time in ED where (i) VFA flux decreased over time, and (ii) the flux of VFAs with higher mobilities (shorter carbon-chain) decreased at faster rates than those with lower mobilities (longer carbon-chain). Further, in terms of selectivity, Cl^– transport is substantially selective over VFAs (Figure S23); thus, Cl^– selectively permeated AEMs during the earlier portion of the ED operation time resulting in continuous VFA transport until the end of the ED experiment (Figure C). Additionally, the small size of NH₄ ⁺ promoted a slight selectivity toward NH₄ ⁺ compared with other cations, which led to initially faster recovery of NH₄ ⁺ (i.e., earlier depletion of NH₄ ⁺ in the diluate compartment) and continued recovery of other cations (namely Ca²⁺ and Mg²⁺) after approximately 120 min of ED operation time (Figure S24).

There was no correlation between energy consumption per mol of VFAs recovered and total operation time (EC_Elec.: R ² = 0.08, p = 0.59; EC_Elec.+Pump: R ² = 0.01, p = 0.88; Figure F). However, the current (proportional to energy consumption; eq ) in the first 60 min was greater than the following 180 min with the initial 60 min dominating the total energy consumption for all total operation times evaluated (Figure S3). The decline in current with time was consistent with the reduced VFA recovery rates in the final 180 min (Figure B,C). The high energy consumption in the first ∼60 min of ED occurred because a substantial fraction of the current was used to transport non-VFA species (e.g., Cl^–, HPO₄ ^2–, NH₄ ⁺, K⁺ etc.; Figures S23–S24). Specifically, for the case that operated for up to 120 min, in the first 60 min of the experiment, the conductivity in the C reservoir increased by 53.9% (Figure S4F) while only 32.9% of VFAs were recovered (Figure B) which indicated a substantial portion of the current from 0 to 60 min transported non-VFA species. Next, from 60 to 120 min, the C reservoir conductivity increased by 12.3% (Figure S4F) and an additional 14.6% of VFAs were recovered (Figure B) which indicated that the portion of the current that was used to transport VFAs in the 60–120 min period was greater than in the 0–60 min period. Overall, the trade-offs between total operation time, VFA recovery, and current (eq ) led to approximately the same energy consumptions between the operation times evaluated.

Results from the evaluation of the effect of operation time on ED performance showed that VFA recovery increased linearly for ∼60 min and then slowed down thereafter as the ED cell resistance increased. However, VFAs were still continuously recovered for up to 240 min while the ED cell current remained low to thereby not adversely affect the energy consumption. Therefore, one strategy when operating at a constant voltage is to operate the ED cell until the benefits of low energy consumption outweigh those of VFA recovery.

3.2. VFA Production in the Stage Two Bioreactors

We evaluated if recycling the ED-treated digestate to the stage 2 bioreactors would have an impact on VFA production relative to the control bioreactor and the bioreactor that received DI water instead of ED-treated digestate. We selected the following four ED-treated digestates: (1) T1: 0.5 V/N = 2/60 min which had the lowest NH₄ ⁺ removal; (2) T4: 0.5 V/N = 4/60 min which had the lowest VFA removal, (3) T6: 1 V/N = 4/120 min which had the highest VFA and NH₄ ⁺ removal, and (4) T7: 1 V/N = 4/240 min which had the highest VFA recovery. We chose these four treatments because they had the largest variation in VFA and NH₄ ⁺ concentrations across treatments. The VFA concentration, removal, and recovery, the NH₄ ⁺ concentration, removal, and recovery, and sCOD for treatments T1, T4, T6, T7 are tabulated in Section S12.

We first compared the change of sCOD, VFA, and NH₄ ⁺ concentrations in the stage 2 bioreactors from the time the ED-treated digestate, DI water, or nothing (i.e., control) was added (day 25) to the end of the incubation (day 54) (Figure A–C). The concentrations of VFAs, NH₄ ⁺, and sCOD immediately before and after the addition of ED-treated digestate, DI water, or nothing are shown in Table . After addition of the ED-treated digestates, the order of dilution (measured as percent decrease in VFA) was as follows: T7 > T6 > T4 > T1 (Table ). In the control (i.e., no ED-treated digestate or DI water addition), sCOD, NH₄ ⁺, and VFA concentrations increased by 4.2 ± 0.4 g·L^–1, 27.6 ± 0.7 mmol·L^–1, and 24.23 ± 0.30 mmol·L^–1 respectively, over the 29 day period (Figure A–C). This increase in VFA production for the control was the lowest among all bioreactors aside from the bioreactor that received DI water. For the bioreactors where the ED-treated digestates were added, sCOD, VFA, and NH₄ ⁺ increased following a similar trend as the control (Figure A–C). The addition of ED-treated digestates to the bioreactors resulted in greater increases in VFA production compared with the control for all treatments, with the greatest difference after 29 days (9.78 mmol·L^–1) corresponding to the T4 bioreactor (Figure C). The observed increases ranged from moderate (40.4%, p = 0.06 for T4, Table S18) to low (7.91%, p = 0.66, T6, Table S18). In contrast, adding DI water to the bioreactors resulted in a significant decrease (34.1%, p = 0.01) in sCOD and VFAs compared to the control (Figure A,C; Table S18).

6.

Net (A) sCOD, (B) NH₄ ⁺, and (C) VFA increase after the ED-treated digestate addition, DI water addition, or no addition (i.e., control) from day 25 to day 54. The points are the average values from triplicate bioreactor experiments, and error bars represent the standard deviation.

3. Composition of Stage 2 Treatment Bioreactors Immediately before and after the Addition of ED-Treated Digestates or DI Water on Day 25 .

	∑VFA (mM)			NH4 + (mM)		sCOD (g·L–1)
treatment	before	after	% decrease	before	after	before	after
T1	107.5 ± 1.4	93.5 ± 2.2	13.0%	63.5 ± 1.7	54.3 ± 2.0	16.0 ± 0.2	14.1 ± 0.1
T4	99.5 ± 4.4	83.2 ± 5.8	16.4%	63.6 ± 2.6	50.8 ± 2.6	14.9 ± 0.6	13.2 ± 0.2
T6	103.7 ± 3.2	79.6 ± 1.0	23.3%	63.3 ± 3.9	42.3 ± 2.4	15.4 ± 0.4	12.7 ± 0.2
T7	100.7 ± 2.7	72.0 ± 4.2	28.5%	64.0 ± 3.0	36.6 ± 2.3	15.1 ± 0.7	11.6 ± 0.5
DIW	104.8 ± 4.9	51.5 ± 2.5	50.8%	60.4 ± 2.7	29.9 ± 2.9	15.2 ± 0.5	7.3 ± 0.2
control	111.6 ± 2.8		NA	62.5 ± 2.2		16.1 ± 0.1

^aNA: not applicable.

Overall, results indicate that recycling ED-treated digestate to AD increased VFA production compared with an unamended controlalbeit only slightly to moderately. Specifically, only one ED treatment resulted in a moderate (40.4%, p = 0.06) VFA production increase (i.e., the other three resulted in small, not statistically significant increases; 7.91–19.8%, p = 0.30–0.66). Further, we did not observe a strong or significant relationship between VFA production and VFA or NH₄ ⁺ concentration in the bioreactors after ED-treated digestate recycling (R ² = 0.18–0.26, p = 0.49–0.58; Section S13). Therefore, results indicate that while recycling ED-treated digestate has the potential to increase VFA production in the AD bioreactor, the VFA/NH₄ ⁺ concentration in the bioreactor alone, which was dictated by the ED physical and operating parameters, does not govern the extent of VFA production. Findings deviate from our initial hypothesis which stated that using ED parameters that enable high VFA/NH₄ ⁺ removal and recovery would enhance VFA production in a combined AD-ED system.

Recirculating VFA-depleted digestate has been reported to enhance VFA production − ,, which is consistent with our results recognizing that we observed only slight (7.91–19.8%, p = 0.30–0.66) to moderate (40.4%, p = 0.06) increases in VFA production relative to the control. For example, Guo et al. tested the effects of effluent recirculation on a semicontinuous two-stage AD system. Compared to a two-stage AD system without digestate effluent recirculation at an organic loading rate of 2.6 gVS·L^–1·d^–1, larger VFA concentrations and hydrolysis rates were seen in the first-stage acidogenic reactor when the effluent from the second-stage methanogenic reactor (where VFAs were used to produce CH₄) was returned. Similarly, in Jones et al., − where ED-treated digestate was semicontinuously returned to the bioreactors, an increase in total VFA production in the ED-AD systems was observed compared with the control stage when no dilution from ED-treated digestate occurred.

The trends observed including (i) only one ED treatment resulted in a moderate (40.4%, p = 0.06) VFA production increase but the other three resulted in small and not statistically significant increases (7.91–19.8%; p = 0.30–0.66), and (ii) there was not a strong correlation between VFA production and VFA/NH₄ ⁺ concentration (R ² = 0.18–0.26, p = 0.49–0.58) could be explained by three possible scenarios. First, VFA concentrations in our bioreactors may have not reached inhibitory levels. We note that throughout AD operation, the pH in our bioreactors ranged from 5.1 to 6.03 and is not expected to have played a role on VFA production given the relatively small pH change with time after the initial 2–3 days, and the small pH variation across bioreactors (Figure S26). Within that pH range, Babel et al. reported that VFA inhibition can occur when VFA concentrations are ∼7.5 to 10 g·L^–1. Those values are similar to or larger than the VFA concentrations in our treatments (7.1–8.1 g·L^–1), so it is possible that the degree of inhibition in our bioreactors was very low. Second, we operated our bioreactors in fed-batch mode where the incubation time in our bioreactors (∼50 days) was longer than the typical incubation time based on hydraulic residence times in the above-mentioned studies (2–10 days). Longer incubation times can lead to the exhaustion of some reactants (e.g., sugars) for the acidogenesis stage of AD (i.e., where VFAs are produced). Loss of these reactants can slow down VFA production. Hydrolysis of the FW, which was the source of the reactants for acidogenesis, may also have been slowed due to the longer retention time. For example, Guo et al. studied the hydrolysis kinetics in anaerobic digesters of different configurations and found that specific hydrolysis rates decreased in all bioreactors as solid retention time increased. Third, for the cases of DI water addition and ED-treated digestate recycling, dilution of the reactants needed for VFA production likely decreased kinetics and may have affected the microbial dynamics in the bioreactor. For example, propionate and caproate production decreased after dilution with DI water (Figure S27). We speculate that the substrates of these acidogenic pathways decreased when they were (1) removed during the removal of the digestate, and (2) the introduction of DI water or ED-treated digestate further diluted their concentrations. These hypotheses may explain why recycling ED-treated digestates did not significantly increase VFA production in the bioreactors.

Our findings have implications for future studies that combine AD and ED toward enhancing VFA production and VFA recovery. First, our results showed that VFA production generally increased following the order of ED Treatment recycling > Control > DI water addition which indicated that the matrix of the solution added to the AD bioreactors affected VFA production and that adding ED-treated digestate to bioreactors led to generally higher VFA production than the unamended control. Future studies should systematically evaluate the role on VFA production of NH₄ ⁺, VFA, and neutral solute concentrations and ratios in synthetic ED-treated digestates recycled to AD bioreactors. Further, future studies are needed to evaluate the effects of AD operating parameters not studied in this work, such as waste feedstock compositions, pH, and hydraulic residence time, on VFA production and recovery in a combined AD-ED system. Additionally, in this work, we performed only one cycle of the sequential batch AD-ED system. Once the process parameters of a combined AD-ED system are further optimized, future studies should perform multiple cycles of AD and ED to further assess long-term effects of the digestate on membrane fouling during ED and solution conditions in the AD bioreactor on VFA production. While a previous study evaluated VFA recovery from digestate across multiple ED cycles under conditions where the C compartment was progressively enriched across cycles, the stability and performance of multiple AD-ED cycles remains underexplored.

4. Conclusions

In this work, we evaluated the effects of ED physical and operating parameters on the performance of a sequential batch AD-ED system. Specifically, we first aimed to understand how the applied voltage, number of IEM pairs, and ED operation time affected VFA recovery and removal, NH₄ ⁺ recovery and removal, and energy consumption per mol of VFAs recovered during treatment of digestate. We next aimed to understand how recycling ED-treated digestates with variable concentrations of VFAs and NH₄ ⁺ to the AD to bioreactors after 25 days of operation impacted VFA production relative to an unamended control bioreactor. The following points summarize our main findings:

• For all ED configurations, VFA transport negatively correlated with VFA carbon-chain length, and linear VFAs transported through AEMs to a greater extent than their isomeric counterparts. Both findings are attributed to steric hindrance.

• Doubling the voltage from 0.5 to 1 V moderately enhanced solute recovery (5.1–11.1%) but at the expense of substantially increasing energy consumption per mol of VFAs recovered (80.7–106% increase). The only moderately enhanced solute recovery as a function of applied voltage occurred because the ED cell resistance increased more quickly when operating with 1 V compared with 0.5 V.

• ED conditions with double IEM pairs (i.e., four versus two pairs) did not lead to double VFA recovery. Rather, doubling the number of IEM pairs from two to four led to only moderately higher solute recovery (5.2–25.8%), moderately lower VFA loss/sorption (0.6–16.3%), and substantially lower energy consumption per mol of VFAs recovered (61.0–64.8%). The only moderately higher solute recovery stemmed from the higher ED cell resistance for the four IEM pair case than the two IEM pair case.

• Increasing the ED operation time up to 240 min enhanced VFA recovery while the NH₄ ⁺ recovery plateaued after approximately 120 min. This behavior was rationalized by the different selectivities between (i) VFAs and Cl^– and (ii) NH₄ ⁺ and other cations (Mg²⁺, Ca²⁺, K⁺).

• Increasing the ED operation time from 60 to 240 min had no effect on energy consumption per mol of VFAs recovered since VFA recovery continued even after the measured ED cell current substantially declined.

• Recycling the ED-treated digestate to the stage 2 bioreactors on day 25 increased the VFA production relative to the control; however, the increase was moderate (40.4%, p = 0.06) for only one of the ED treatments and small and insignificant for the others (7.91–19.8%, p = 0.30–0.66).

• The increase in VFA production did not strongly or significantly correlate with the VFA or NH₄ ⁺ concentration in the AD bioreactors on day 25 for cases where the ED-treated digestates were recycled back to the bioreactor (R ² = 0.18–0.26, p = 0.49–0.58). The weak and not statistically significant correlation deviates from our initial hypothesis which stated that applying ED parameters that enable high VFA/NH₄ ⁺ removal and recovery would enhance VFA production in a combined AD-ED system.

• The fact that the effect of recycling the ED-treated digestates to the stage 2 bioreactors on VFA production had a moderate effect only for one ED treatment (and small and insignificant for the others) may have been due to not reaching inhibitory levels of VFAs in the bioreactors and/or loss/dilution of reactants required to produce VFAs when a portion of the bioreactor was replaced with ED-treated digestate.

Overall, we showed that ED physical and operating parameters have a substantial effect on ED performance when treating digestate. Further, we provided a comprehensive evaluation of how ED parameters impact performance metrics that are relevant for a combined AD-ED system (e.g., VFA and NH₄ ⁺ recovery/removal and energy consumption). Additionally, we showed that recycling ED-treated digestate to a subsequent AD step may improve VFA production relative to an unamended control, but the VFA/NH₄ ⁺ concentration in the bioreactor alone does not govern the extent of VFA production in the stage 2 bioreactor. Therefore, future work should focus on the effects of ED-treated digestate matrix and number of AD-ED cycles on VFA production along with evaluation of how the microbial populations in AD respond to the addition of ED-treated digestate. This work will inform future studies that aim to couple AD and ED toward enhancing VFA production and recovery.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.5c00937.

• Preparation and properties of substrates and inoculum; analytical methods; IEM and ED cell properties, testing, and cleanup; detailed performance of each ED treatment and synthetic controls including energy consumption, current, pH, conductivities, current efficiency, and solute removal/recovery; characteristics of digestates post ED treatment; VFA profile for stage 2 AD bioreactors (PDF)

§.

H.H. and H.D. contributed equally to this work.

The authors declare no competing financial interest.