Long-Term Evaluation of CNT-Clad Stainless-Steel Cathodes in Multi-Channel Microbial Electrolysis Cells Under Variable Conditions

Abstract

Microbial electrolysis cells (MECs) present a viable platform for sustainable hydrogen generation from organic waste, but their scalability is limited by cathode performance, cost, and durability. This study evaluates three hybrid carbon nanotube (CNT) cathodes—acid-washed CNT (AW-CNT), thin layer non-acid-washed CNT (TN-NAW-CNT), and thick layer non-acid-washed CNT (TK-NAW-CNT)—each composed of stainless-steel-supported CNTs coated with molybdenum phosphide (MoP). These were benchmarked against woven carbon cloth (WCC) under varied operational conditions. A custom multi-channel reactor operated for 341 days, testing cathode performance across applied voltages (0.7–1.2 V), buffer types (phosphate vs. bicarbonate), pH (7.0 and 8.5), buffer concentrations (10–200 mM), and substrates including acetate, lactate, and treated acid whey. CNT-based cathodes consistently showed higher current densities than WCC across most conditions with significant difference found at higher applied voltages. TK-NAW-CNT achieved peak current densities of 259 A m⁻² at 1.2 V and maintained >41 A m⁻² in real-waste conditions with no added buffer. Long-term performance losses were minimal: 4.5% (TN-NAW-CNT), 0.1% (TK-NAW-CNT), 10.8% (AW-CNT), and 6.8% (WCC). CNT cathodes showed improved performance from reduced resistance and greater electrochemical stability, while proton transfer improvements benefited all materials due to buffer type and pH conditions. These results highlight CNT-based cathodes as promising, scalable alternatives to WCC for energy-positive wastewater treatment.

1. Introduction

Transitioning away from fossil fuel-based energy sources is critical for the coming decades to mitigate climate change and reduce ongoing environmental degradation. The development and deployment of alternative energy technologies that support carbon neutrality, including wind, solar, hydroelectric, and bioenergy, are becoming increasingly important. A major challenge shared across renewable energy systems is the need for reliable energy storage. Current solutions often rely on costly batteries that require rare elements typically extracted through environmentally damaging mining practices. Hydrogen, by contrast, represents a promising alternative as an energy carrier: it is abundant, clean-burning, and yields only water as a byproduct at point of use. This makes it especially attractive in applications where air quality and emissions are key concerns [1].

Microbial electrolysis cell (MEC) technology is an emerging bioenergy system capable of directly producing hydrogen gas from organic substrates. As a type of bio-electrochemical system (BES), MECs have shown strong potential not only to generate hydrogen from organic-rich waste streams [2], but also to simultaneously reduce their chemical and biological oxygen demand (COD and BOD). This dual functionality enables MECs to enhance the energy efficiency of wastewater treatment by offsetting the energy normally required for conventional treatment. This is achieved by converting embedded energy into a usable form as hydrogen. MECs leverage a specialized group of electrochemically active microorganisms (EAMs), such as Geobacter spp., that oxidize substrates like volatile fatty acids (VFAs) at the anode [3]. Electrons released from this oxidation process are transferred to the cathode via a low applied voltage, where they reduce protons to hydrogen gas [4]. Consequently, MECs offer the opportunity to recover energy from waste streams that would otherwise demand costly aerobic treatment processes. Although MECs require some external energy input, their net energy efficiency is often more favorable than that of conventional hydrogen production methods, such as water electrolysis.

Despite these advantages, scaling MECs to industrial levels remains a major challenge. Three primary obstacles are associated with cathode base-material selection: material resistance, longevity, and cost. To ensure economic viability, cathodes must be composed of earth-abundant elements and manufacturable with low energy inputs. They must also maintain performance over extended operational periods to minimize replacement and maintenance costs. At the same time, minimizing electrode resistance is essential for achieving high current densities and maximizing hydrogen production.

Cell resistance (R), which directly influences current (I) output according to Ohm’s law (I = V/R) at a given applied voltage (V), arises from multiple components: the intrinsic resistivity of electrode materials and current collection systems, limitations in proton transfer across the electrolyte, and catalytic activity at both the anode (biotic) and cathode (abiotic). As reactor dimensions increase, resistance scales with the length and surface area of the electrode material, often reducing current densities unless material conductivity is optimized. This is particularly important because the cathodic hydrogen evolution reaction (HER) is not spontaneous. For the oxidation of acetate to hydrogen under biological standard conditions, a theoretical minimum voltage of E⁰′ = 0.13 V is required. However, slightly lower thresholds are possible under operational conditions with reduced hydrogen partial pressure. While hydrogen production has been observed near this thermodynamic minimum [5], it typically occurs at low rates. Substantially higher external voltages (0.6–1.4 V) are generally required to reach the current densities and hydrogen production rates (HPRs) necessary for industrial applications [6,7,8,9]. This means that lowering MEC resistance through cathode material selection is necessary for increasing current at low applied voltages, thus increasing efficiency. While electrode material resistance can cause a decrease in performance at small scales, this effect is inevitably exacerbated at larger scales. Many studies report reduced performance in upscaled reactors [10], or employ costly current collection infrastructure, such as those made of titanium, to mitigate this decrease [9,11].

MEC cathodes typically fall into two categories: stand-alone and hybrid materials. Stand-alone materials, such as nickel or stainless steel (SS), serve both as the structural and catalytic component. Hybrid cathodes, by contrast, combine materials to optimize both structural and catalytic functions. Ideal supporting materials should possess high conductivity, structural integrity, corrosion resistance, a high specific surface area (SSA), material abundance, and low manufacturing costs. Optimal catalyst materials should offer high catalytic activity, long-term stability, and be composed of commonly available elements.

Among hybrid cathodes, platinum (Pt) has long been recognized as a benchmark catalyst for HER [12,13]. However, the high cost and scarcity of Pt have prompted exploration of alternatives, including nickel [14], stainless steel [15], titanium [16], mixed-metal [17,18], nanocomposites [19], and biocathodes [20,21]. Molybdenum phosphide (MoP), in particular, has emerged as a strong candidate, demonstrating current densities of up to 157 A m⁻², comparable to Pt, while meeting the material and cost criteria required for scale-up [9,22].

For cathode base materials, carbon-based substrates such as carbon fiber, carbon felt, carbon paper, and carbon brush are frequently used due to their corrosion resistance and long-term stability in MEC environments [23,24,25]. Other carbon base material such as activated carbon have also been explored to enhance physical properties such as wettability [26]. However, many suffer from drawbacks such as high electrical resistance, low surface-to-volume ratios, and elevated costs. Carbon nanotubes (CNTs) offer a promising alternative that may overcome these limitations while remaining cost-effective. New CNT manufacturing technologies are driving down manufacturing costs of CNT from conventional prices of about $100 kg⁻¹ to under $10 kg⁻¹, with future projection even lower [27], making these materials cost competitive with carbon cloth-type electrodes. Despite their potential, CNTs remain underexplored as cathode base materials in MECs. CNT has been successfully tested as a cathode catalyst when applied to carbon cloth base materials, producing current densities of 192 Am⁻³ [28], 205 Am⁻³ [29], and 163 Am⁻³ [30]. While CNT base material has been tested previously as anode material [31], CNT as cathode base material has not yet been tested to the best of our knowledge. Using CNT as a base material, as opposed to CNT as a catalyst, draws on the high current conducting and surface area properties of the material. Combining the robust qualities of CNT as base material with higher-activity catalysts, improvements to cathode kinetics can be achieved.

CNTs are essentially rolled graphene sheets with exceptional electrical conductivity and high specific surface area relative to other carbonaceous materials. Pure CNTs can achieve conductivities in the range of 10⁶ to 10⁷ S m⁻¹, approaching those of top conductive metals like silver (6.2 × 10⁷ S m⁻¹) [32]. In bulk form, however, their performance is often limited by lattice imperfections and grain boundary effects [33]. To mitigate these limitations, CNTs can be grown directly on a more conductive material backbone. This approach not only improves electrical performance but also provides structural stability at the macro scale. It also has the potential to reduce reactor fabrication complexity by eliminating the need for additional electrode support structures, such as cassettes used with flexible materials like carbon cloth.

In this study, we evaluated the performance of hybrid cathodes composed of novel dual-composite CNT base materials combined with an MoP catalyst under realistic operational conditions. A multi-walled CNT layer was grown directly on stainless-steel mesh to enhance surface area and reduce cathode resistance. A custom-designed multi-channel test cassette, implemented in a single-chamber reactor, enabled simultaneous electrode comparisons. The performance and long-term stability of these hybrid cathodes were benchmarked against MoP-coated carbon cloth. Additionally, we tested cathode material resilience under real-world reactor conditions, including variable buffer types and concentrations, pH shifts, and use of pre-treated acid whey—an abundant waste stream from cheese manufacturing.

2. Materials and Methods

2.1. Design and Fabrication of a Multi-Channel Electrode Test Cassette

A multi-channel electrode test cassette was constructed for all cathode evaluations (Figure 1A). The test cassette comprises 16 individual channel locations. Each location accommodates a 1.27 cm diameter test electrode and provides a working surface area of 0.71 ${\mathrm{c}\mathrm{m}}^2$. The cassette assembly was 3D printed using 1.75 mm PRO Series Ryno™ filament (Matter Hackers, Lake Forest, CA, USA) at a 245 °C extrusion temperature and a 65 °C print bed temperature. The wire for all electrode current collection was 0.025” Ultra-Corrosion-Resistant Grade 2 Titanium Wire (McMaster-Carr, Elmhurst, IL, USA). The current collection wires at the electrode contact point were formed into a zig-zag pattern and were appressed onto the entire circumference of the test electrode during assembly to maximize wire–electrode contact area. Each electrode sub-assembly was held into place by an O-ring (0.067 in × 0.5 in dia.) (McMaster-Carr, Elmhurst, IL, USA) and secured with a 3D-printed retaining bracket (Figure 1A). The vessel housing was a 2 L Wheaton, double-neck vessel equipped with butyl septa for gas sampling (VWR International, LLC, Radnor, PA, USA) (Figure 1B).

A triple layer of carbon fiber cloth (Type-B, fuelcellearth.com) was used as the anode, with a projected surface area of 85 ${\mathrm{c}\mathrm{m}}^2$ and a total anode surface area of 255 ${\mathrm{c}\mathrm{m}}^2$. The distance between anode and cathode was held constant at 1.2 cm with additional 3D-printed frames. The cassette was assembled with 6-32 Nylon hardware (McMaster-Carr, Elmhurst, IL, USA).

2.2. MEC Operation and Cathode Evaluation

The MEC anode was inoculated with effluent from a mature MEC previously operated with acetate as substrate. The original culture was enriched from sewage sludge [34]. All electrode testing was carried out at 32 °C in an air temperature-controlled incubation cabinet. The electric voltage applied to the MEC circuit varied with treatment (0.7 V, 1.0 V, or 1.2 V) and was supplied via a DC power supply (CSI1802X, Circuit Specialists, Mesa, AZ, USA). The reactor was fed in batch mode with exchange intervals determined by the first indication of current density decline due to substrate limitation. To facilitate solution proton transfer and maintain pH levels, both phosphate buffer system (PBS) and bicarbonate buffer system (BBS) were used comprising NaH₂PO₄ and Na₂HPO₄, and NaHCO₃, respectively. For both buffer systems, testing concentrations of 10 mM, 50 mM, 100 mM, and 200 mM were used to evaluate the effect of buffer concentration on cathode performance. The carbon substrates used for MEC operation consisted of sodium acetate (40 mM), sodium lactate (40 mM), or diluted pre-treated acid whey effluent from cheese manufacturing (OSU Beaver Classic Creamery, Corvallis, OR, USA). The acid whey effluent was diluted to match the lactate concentration of the synthetic medium (40 mM lactate) which rendered an acetate concentration of 20.3 mM. The medium vitamins are as follows: (Biotin (2 mg L⁻¹), folic acid (2 mg L⁻¹), pyridoxine HCl (10 mg L⁻¹), riboflavin (5 mg L⁻¹), thiamin (5 mg L⁻¹), nicotinic acid (5 mg L⁻¹), pantothenic acid (5 mg L⁻¹), B-12 (0.1 mg L⁻¹), p-aminobenzoic acid (5 mg L⁻¹), and thioctic acid (5 mg L⁻¹). Minerals used are as follows: Nitrilotriacetic acid (NTA) (1.5 g L⁻¹), MgSO₄ (3 g L⁻¹), MnSO₄·4H₂O (0.5 g L⁻¹), NaCl (1 g L⁻¹), FeSO₄·7H₂O (0.1 g L⁻¹), CaCl₂·2H₂O (0.1 g L⁻¹), CoCl₂·6H₂O (0.1 g L⁻¹), ZnCl₂ (0.13 g L⁻¹), CuSO₄·5H₂O (0.01 g L⁻¹), AlK(SO₄)₂·12H₂O (0.01 g L⁻¹), H₃BO₃ (0.01 g L⁻¹), Na₂MoO₄ (0.025 g L⁻¹), NiCl₂·6H₂O (0.024 g L⁻¹), and Na₂WO₄·2H₂O (0.025 g L⁻¹). With each medium batch exchange, the headspace of the reactor was purged with nitrogen gas (Airgas, Radnor Township, PA, USA) for 15 min. The pH was adjusted with sodium hydroxide (NaOH) and hydrochloric acid (HCl) for each pH treatment (pH 7.00, 8.00, and 8.50). The MEC was operated continuously for a total of 341 days. During baseline operation, all test cathodes were connected simultaneously, resulting in a lower anode-to-cathode ratio (7:1). To reduce the influence of anode limitations and better assess cathode performance, operational condition tests were conducted using a higher anode-to-cathode ratio (60:1). This was achieved by sequentially testing randomized pairs of cathodes for one hour each before switching to the successive pairs. The (60:1) ratio is used to supply ample anodic surface area, thereby reducing any imbalance in mass transfer between channels, a benefit not inferred to test cells at the (7:1) ratio.

2.3. Fabrication and Preparation of CNT and Carbon Cloth Cathode Material

The woven carbon cloth (WCC) cathode base material was identical to that used for the anode material as described above and was tested in duplicates. The CNT cathode base material tested was acid-treated CNT (AW-CNT, 0.956–1.24 mm thickness), tested in quadruplicate, and two non-acid-treated material types grown to different thicknesses: thin non-acid-washed CNT (TN-NAW-CNT, 1.034–1.101 mm thickness), tested in triplicate, and thick non-acid-washed CNT (TK-NAW-CNT, 1.503–1.551), tested in triplicate. All CNT material was grown on 316 stainless-steel mesh with wire diameter of 0.165 mm and a pore size of 0.680 mm, which provided a conductive, mechanically robust scaffold.

CNT synthesis was conducted via chemical vapor deposition (CVD) on a nickel-plated stainless-steel mesh, as described in detail in our previous work [31]. The plating solution contained NiSO₄·6H₂O (300 g L⁻¹), NiCl₂·6H₂O (35 g L⁻¹), and boric acid (30 g L⁻¹). Electroplating was performed at a current density of 2 mA cm⁻² for 2 min. After thorough rinsing and drying with deionized water, the nickel-coated mesh was punched into 12 mm circular discs and placed in a three-zone quartz tube furnace. Ferrocene served as the catalyst precursor in zone 1, while the mesh resided in zone 3. The reactor was purged with argon (Ar), then zones 1 and 3 were heated to 120 °C and 650 °C, respectively, under a 400 sccm flow of H₂. Growth was initiated by introducing a gas mixture of C₂H₄ (120 sccm), Ar (120 sccm), and H₂ (400 sccm) for 5–15 min. After the reaction, the furnace was cooled down to room temperature, and NAW-CNT samples were collected. The details of the CNT growth parameters are given in

Table 1. Summary of CNT growth parameters on Ni-plated stainless-steel mesh.

Process	Step	Variables	Setting
Ni Plating	Bath	Composition	NiSO4·6H2O (300 g L−1), NiCl2·6H2O (35 g L−1), H3BO3 (30 g L−1)
	Plating	Current density/time	2 mA cm−2/2 min
CVD	Purge	Gas (flow rate)/time	Ar (600 sccm)/15 min
	Ramp	Temperature (Zone 1/Zone 3)	120 °C/650 °C under H2 400 sccm
	Growth	Gas (flow rates in sccm)	C2H4 (120 sccm); Ar (120 sccm); H2 (400 sccm)
		Duration	15 min
		Catalyst	200 mg ferrocene (solid in zone 1)
	Cooling	Gas (flow rates)	Ar (20 sccm)

.

For AW-CNT samples preparation, an oxidative treatment adapted from a prior report [35] was used. The process involves applying a solution of KMnO₄ (0.5 g L⁻¹) dissolved in concentrated sulfuric acid (95–98%) dropwise onto the CNT-coated mesh samples in a vacuum filtration setup, maintaining controlled vacuum pressure to ensure uniform oxidation. This treatment partially unzipped the graphitic carbon walls, facilitating the attachment of carboxyl and hydroxyl groups to defect sites on the CNT surface, improving hydrophilicity and proton accessibility. After treatment, samples were rinsed thoroughly with deionized water and ethanol to remove residual oxidants and acids and then dried at 80 °C overnight.

Each test cathode piece was loaded with 2.5 mg cm⁻² molybdenum phosphide (MoP) catalyst using a Nafion™ binder (D520, Fuel Cell Store, Bryan, TX, USA) at a rate of 7$\mathsf{\mu }\mathrm{L}$ Nafion/1 mg MoP catalyst as previously described [9]. The catalyst mixture was applied via airbrush using nitrogen propellant. Cathodes were dried in open-air for 48 h before installation into the MEC cassette.

2.4. Acid Whey Feedstock Characterization

The treated acid whey effluent used as MEC influent was created using raw acid whey provided by “OSU Beaver Classic Creamery” (Corvallis, OR, USA) as a feedstock. The sludge used as inoculum for acid whey treatment was obtained from an anaerobic digester of a municipal wastewater treatment plant (Tigard, OR, USA). The raw acid whey was stored at 4 °C until use. The characteristics of the feedstock (raw acid whey) and fermentation effluent (treated acid whey) are shown in Table S44.

The sludge was first stored at 35 °C for 24 h for degasification prior to inoculation. The fermentation was run in semi-batch mode with a working volume of 3 L. The ratio of the inoculum to feedstock was 1:2 based on the pre-batch results. Feeding events took place every 15 h, at which time all lactose was consumed, and 2 L of fresh acid whey influent was exchanged. The bioreactor was continuously mixed at 100 rpm. pH was measured and monitored by a pH probe mounted through the bioreactor head plate. Automated pH control of the bioreactor-mixed liquor was maintained at pH 5.5 ± 0.1 using a controlled dosing pump. The temperature inside the reactor was maintained at 35 °C by heat jacket. The produced gases were collected in a gasbag in the upper part of the reactor. VFA concentrations were determined using a high-performance liquid chromatograph (Agilent Technologies 1200 Series, Agilent Technologies, Santa Clara, CA, USA) equipped with an Aminex HPX-87H column as previously described [36].

2.5. Data Analysis and Calculations

The MEC current (I) was determined using Ohm’s law ($I= {\displaystyle \frac{V}{R}}$) and was measured as voltage across a 1 $\mathsf{\Omega }$ resistor. Where V is the measured voltage across the resistor and R is the resistance value of known magnitude. All voltage measurements were acquired using a multi-channel data acquisition system (2700, Keithley, Cleveland, OH, USA) and recorded in seven-minute intervals. Cathode current values were normalized and reported as (A m⁻²). The area-specific resistance (ASR) of each material was calculated using ($ASR= {\displaystyle \frac{∆V}{∆J}}$), where $∆V$ is change in voltage and $∆J$ is change in current density.

All gas samples were taken using a 50 $\mathsf{\mu }\mathrm{L}$ syringe (1705N Hamilton) and analyzed using a gas chromatograph (6890N Network GC System, Agilent Technologies, Santa Clara, CA, USA). Hydrogen recovery ${(R}_{cathode})$ was calculated using $R_{cathode}={\displaystyle \frac{V_t}{V_E}}$, where V_t is the total volume of hydrogen produced and V_E is the expected volume based on the total calculated Coulombs from the circuit.

To analyze statistical differences between treatments, whether between materials, substrate type, applied voltage, or buffer type and concentration, a linear mixed effects (LME) model was used. This statistical approach simultaneously models fixed and random effects which allows for dependent samples within a replicate; hence, it is suitable for our time series data. We fit LME models using the R package lme4 (version 1.1-37) [37], including the treatment as the fixed effect and replicates as a random effect. For example, when analyzing if current densities differ between cathode materials we use ‘cathode material’ as the fixed effect and we use ‘replicates’ as a random effect. We then obtained the estimated marginal means (EMMs) from R package emmeans (version 1.11.2-8) [38] from the fitted LME for the expected mean current density across replicates for each treatment. Next, we tested all pairwise comparisons between the treatments’ EMMs using a Tukey’s test to correct for multiple comparisons to obtain p-values. Similar statistical analyses can be found in prior work [39,40,41]. Differences were considered significant at p < 0.05. Performance results are reported as mean ± standard deviation. All statistical analyses and figures were generated using Python 3.8.18 and Rstudio 4.1.3.

3. Results and Discussion

3.1. Characterization of CNT Materials

The structural and morphological properties of the CNT-coated stainless-steel mesh cathodes were evaluated using photographic, optical, and electron microscopy techniques. Figure 2a shows the digital image of the CNT-coated stainless-steel mesh sample, revealing a uniform, continuous coating across the substrate surface. High-magnification optical microscopy further confirms the dense coverage of the CNTs, forming an interconnected network on the SS mesh wires (Figure 2b). This continuous network morphology is critical, as it facilitates efficient electron transport pathways by bridging the conductive SS scaffold and the electrochemically active catalyst sites.

Figure 2c,d show the SEM analysis at low and high magnifications, demonstrating that the CNTs form a three-dimensional, entangled network across the SS wires. This interconnected CNT network significantly contrasts with the planar fiber structure of conventional WCC, providing enhanced surface area and porosity. The nanotubes exhibit diameters of ~150 nm and display excellent mechanical adhesion to the SS substrate. Notably, the direct growth of CNTs on the conductive SS backbone ensures a low-resistance electrical interface, minimizing contact resistance losses that typically arise from binder-based coatings on carbon cloth. The porosity of the CNT electrodes was also quantified using a simple mass/volume approach: the total CNT layer volume was calculated from the electrode area and average CNT thickness (excluding the SS mesh), and the solid volume from the CNT mass divided by the bulk carbon density (2 g cm⁻³). Porosity was then calculated as $\phi =1-{\displaystyle \frac{V_{solid}}{V_{total}}}$, giving a porosity of ~92%, consistent with the SEM morphology.

Electrochemical characterization using Tafel plots (Figure 2e) further highlights the protective effect of the CNT network. The bare SS316 exhibits an initial corrosion potential of −398 mV, which shifts negatively to −720 mV after 50 cyclic voltammetry (CV) cycles, indicating substantial passive film degradation. In contrast, the CNT-coated SS mesh shows an initial corrosion potential of −217 mV with only a minor 16 mV negative shift after 50 CV cycles, demonstrating enhanced corrosion resistance provided by the CNT coating. This stability, combined with the interconnected morphology, highlights the advantage of directly growing CNTs on SS mesh for long-term electrochemical performance. Collectively, these distinctions in CNT layer thickness, surface chemistry, and conductivity are expected to influence cathode performance, durability, and scalability in microbial electrolysis cells.

3.2. Effect of Applied Voltage on MEC Cathode Performance

Increasing applied voltage increases current density which in turn increases hydrogen production. However, energy efficiency during MEC operation is known to decrease with increased applied voltage due to resistive losses [2]. Therefore, when selecting cathode material for optimal performance, it is important to evaluate the maximal current density across both lower and upper operational voltages, to determine the optimal cathode material. Current density increases significantly (p < 0.001, Tables S12–S14) within each material type across the three tested voltages (0.7 V, 1.0 V, and 1.2 V) (Figure 3). At the lower applied voltage (0.7 V), woven carbon cloth (WCC) produced a non-significant higher current density than all three CNT materials (Figure 3). While at the middle-applied voltage (1.0 V) and, to a greater extent, at higher applied voltage (1.2 V), each CNT material produced higher current densities than WCC. At the 1.0 V applied voltage level, no significant differences were found among material types. At the 1.2 V applied voltage level, TK-NAW-CNT current density was found to be significantly higher than WCC (p = 0.014, Table S11). AW-CNT and TN-CNT exhibited nearly significantly higher current densities than WCC (p = 0.058 and p = 0.054, Table S11). No significant differences were found among CNT variants at any voltage level. The elevated performance of WCC at 0.7 V was unexpected and is attributed to its behavior during the earliest stages of the testing period. Specifically, WCC showed transiently higher current densities at the beginning of the experiment before stabilizing at a lower performance level over time. Despite efforts to increase hydrophilicity of CNT through the addition of functional groups (hydroxy and carboxyl), CNT remains fundamentally hydrophobic [42]. This attribute of CNT can cause relative performance instability in nascent testing periods, when material is first introduced to an aqueous environment, leading to higher relative performance of WCC. The highest current density was observed for TK-NAW-CNT, exceeding 259 A m⁻² at 1.2 V with 200 mM phosphate buffer. This value was over 50 A m⁻² higher than that achieved by WCC (p < 0.014, Table S11) and higher than TN-NAW-CNT and AW-CNT by 12 A m⁻² (p < 0.71, Table S11) and 15 A m⁻² (p < 0.53, Table S11), respectively (Figure 3). All material in this applied voltage condition were high, relative to the field with current densities of 245.18 ± 14.32 A m⁻², 259.90 ± 7.24 A m⁻², 247.74 ± 10.20 A m⁻², and 208.07 ± 24.55 A m⁻² for AW-CNT, TK-NAW-CNT, TN-NAW-CNT, and WCC, respectively. To the best of our knowledge, the highest reported current densities are 90 A m⁻² [43], 63 ± 11 A m⁻² [31], 43.1 A m⁻² [44], and 44.4 A m⁻² [45]. Our high reported values in the 1.2 V applied voltage test are substantially higher than these reported values although a direct performance comparison is not practical as each of these studies are using more realistic reactor conditions such as lower anode-to-cathode ratios and for longer sustained periods. These high values do however demonstrate the high potential of these materials when unconstrained by other MEC system limitations.

The elevated current density achievable with CNT cathode material, when compared to WCC, is likely a result of decreased material resistance, a benefit of using a stainless-steel skeleton bolstering the nanotube arrays. The conductive stainless-steel mesh serves as a highly efficient electron highway, directly supporting the interconnected CNT network and minimizing contact resistance losses commonly observed in binder-based electrodes. In our previous study, the active surface area of the CNT on the SS mesh electrodes was estimated to be ~740 m² g⁻¹ based on the time-dependent current response using the Cottrell equation [46]. In contrast, typical untreated carbon cloth exhibits a much lower specific surface area of around 2.39 m² g⁻¹ [47]. The higher active surface area for the CNT on the SS mesh significantly increases the density of accessible catalytic sites and facilitates improved electrolyte infiltration, both critical for enhanced catalytic activity and hydrogen evolution kinetics. This substantial increase in accessible surface area explains the enhanced electrochemical performance observed in our study and demonstrates the clear advantage of the CNT-coated electrodes at higher applied voltage. The polarization curve slopes yielded area-specific resistance (ASR) values of 3.4 mΩ m², 2.9 mΩ m², 3.1 mΩ m², and 4.6 mΩ m², for AW-CNT, TK-NAW-CNT, TN-NAW-CNT, and WCC cathodes, respectively. As we scale this system, these performance enhancements are expected to be amplified as the resistance of a material is proportional to the length of electron travel, making the low-resistance SS-CNT architecture particularly advantageous for thick, high-loading cathodes. Among the CNT materials tested, the thick non-acid-washed CNT (TK-NAW-CNT) exhibits the highest electrochemical performance. The dense and highly interconnected CNT network in the thick layer provides continuous pathways for electron and proton transport, increasing the density of accessible catalytic sites and minimizing resistive losses. In comparison, thinner CNT layers show less uniform coverage and reduced network connectivity, which can limit the accessibility of catalytic sites despite shorter electron transport distances. Acid-washed CNTs, irrespective of thickness, introduce structural defects that compromise electrical conductivity and long-term durability. These results emphasize that optimizing both CNT layer thickness and surface chemistry is crucial for balancing conductivity, catalytic activity, and stability in scalable microbial electrolysis cell cathodes.

3.3. Effect of Nutritional Substrate on MEC Cathode Performance

Cathodic performance in MECs can be directly impaired by substrate-dependent fouling mechanisms, such as suspended solids accumulation, non-electrogenic biofilm formation, and mineral precipitation, when exposed to complex feedstocks. These processes reduce the effective surface area and disrupt local electrochemical conditions, ultimately hindering hydrogen evolution. However, distinguishing the effects of substrate type on cathode performance in MECs is inherently complex due to the interconnected nature of anodic and cathodic processes. Because cathodic performance in an operating MEC cannot be fully decoupled from upstream anodic activity, especially in single-chamber designs, we interpret substrate effects at the system level, recognizing that observed differences in current density may result from interactions across both electrodes. To better isolate cathodic effects, these experiments were designed with a high anode-to-cathode projected surface area ratio (60:1), helping to minimize performance limitations arising from anodic inhibition. Our focus is on evaluating the performance of MECs with these cathode materials under different substrate conditions, including synthetic media and a real-waste stream, both with and without buffer amendment.

As expected, the synthetic substrate containing 40 mM acetate and 200 mM bicarbonate buffer consistently produced the highest current densities of 187.24 ± 4.55 A m⁻², 188.28 ± 7.91A m⁻², 189.78 ± 7.65 A m⁻², and 184.25 ± 8.87A m⁻², for AW-CNT, TK-NAW-CNT, TN-NAW-CNT, and WCC, respectively (Figure 4), confirming acetate’s status as an ideal electron donor for anodic bacteria in MECs. The superiority of acetate for Geobacter species is well documented and is attributed to its direct entry into the TCA cycle via acetyl-CoA, enabling efficient electron transfer [48].

In contrast, lactate (40 mM with 200 mM buffer), despite being thermodynamically more favorable than acetate as substrate, resulted in lower current production: 159.35 ± 9.69 A m⁻², 154.28 ± 7.21 A m⁻², 162.28 ± 8.60 A m⁻², and 143.85 ± 2.00A m⁻², for AW-CNT, TK-NAW-CNT, TN-NAW-CNT, and WCC, respectively. This reduction in performance is likely due to decreased anodic oxidation efficiency arising from metabolic complexity: lactate must first be converted to pyruvate and then to acetyl-CoA before it can enter the TCA cycle. While this explains a potential decrease arising from anodic processes, non-acetate substrates such as those present in the synthetic lactate treatment and the treated acid whey (buffered or unbuffered) may promote the development of non-electrogenic microbial communities as lactate is easily converted to other VFAs such as propionate and butyrate [49]. These conditions can lead to the formation of biofilms on both the cathode and anode surface that are not involved in extracellular electron transfer. Such biofilms may obstruct electron transfer pathways or impede hydrogen evolution by creating diffusional limitations or altering the local electrochemical environment. In contrast, acetate tends to support microbial populations more directly engaged in electrogenic activity, contributing to its superior performance.

Further reduction in current density was observed with treated acid whey (40 mM lactate + 20 mM acetate) even when the buffer was maintained at 200 mM. This suggests that the complexity of the real feedstock containing particulates, biofilm-promoting constituents, or inhibitory metabolic byproducts such as butyrate and propionate may contribute to cathodic fouling or subtle anodic inhibition. When the buffer was removed entirely, cathodic performance deteriorated sharply across all materials, highlighting the essential role of buffering in maintaining local pH, conductivity, and electrochemical stability. While the performance decreased dramatically from the optimal conditions, and bearing in mind the extreme anode-to-cathode ratio, we report a high current density relative to the field for real feedstock without added buffer to yield 41.91 ± 2.16 A m⁻², 41.81 ± 3.59 A m⁻², 41.86 ± 1.20A m⁻², and 38.80 ± 2.67 A m⁻² for AW-CNT, TK-NAW-CNT, TN-NAW-CNT, and WCC, respectively. However, it should be noted that during the testing period, the pH of the electrolyte declined from approximately 8.5 to 6.5. This pH shift may have affected the relative current density values. Statistical analysis confirmed that these substrate-driven differences were robust. For all four cathode materials, changes in substrate type led to statistically significant differences in current density (p < 0.001 for all pairwise comparisons, Tables S5–S8). Within substrate type, CNT materials exhibited consistently higher current densities, although no significant differences in current density were found among material types. These results suggest that while advanced CNT-based electrodes often exhibit higher current densities under varying substrate conditions, their relative benefits are not statistically significant across all treatments. Instead, MEC performance appears to be more strongly influenced by substrate quality and buffering capacity. Cathodic inhibition becomes increasingly pronounced as substrate complexity rises and buffer support is withdrawn.

3.4. Effect of Buffer Type and Concentration on MEC Performance

The addition of buffer in MECs serves two primary functions: buffering the pH drop caused by the oxidation of volatile fatty acids (VFAs) at the anode, and facilitating charge transfer [50]. However, for practical implementation of MEC technology in wastewater treatment, adding significant quantities of buffering reagents could be costly and impractical. Phosphate buffers have commonly been used in MEC research, as this buffering system has a higher buffering capacity at a neutral pH, which is within the suitable range for the anodic microbial community. However, due to the increasing scarcity, high cost, and environmental impacts associated with the mining of phosphorous [51], it is important to evaluate these cathode materials using cheaper, alternative buffering systems. Bicarbonate is one such buffer system that has been shown to facilitate high current density in MECs [52,53], while remaining relatively cost-effective and with low environmental impact. Additionally, the bicarbonate buffer system has the potential to be produced in situ by conversion from carbon dioxide produced during VFA oxidation provided the solution pH is maintained above pKa1 of the carbonate system. This process does, however, require the addition of caustic soda to maintain reactor pH at the desired level, though caustic soda is already a commonly added reagent in wastewater treatment [54]. Three buffering conditions were tested here to evaluate these materials under conventional and alternative buffer conditions: bicarbonate at pH 8.5, bicarbonate at pH 7.0, and phosphate at pH 7.0.

The highest performances were found at the 200 mM buffer concentration for each buffer type and pH level (Figure 5). At this concentration, cathodic performance varied by both buffer composition and pH, with differences observed across the four tested materials. In bicarbonate buffer at pH 8.5, all four materials exhibited the highest current densities, with TK-NAW-CNT (188.28 ± 7.91 A m⁻²), TN-NAW-CNT (189.78 ± 7.65 A m⁻²), and AW-CNT (187.24 ± 4.55 A m⁻²) slightly outperforming WCC (184.25 ± 8.87 A m⁻²). CNT-based materials showed slightly higher current densities than WCC with no significant differences found, indicating all materials function well under alkaline bicarbonate conditions.

When the pH was lowered to 7.0 in 200 mM bicarbonate buffer condition, current densities significantly declined for all materials (p < 0.001, Tables S28–S31). TK-NAW-CNT, TN-NAW-CNT, and AW-CNT maintained relatively high performance (150.53 ± 7.76 A m⁻², 150.91 ± 11.13 A m⁻², and 148.19 ± 6.88 A m⁻², respectively), while WCC experienced a marked decline to 124.70 ± 5.74 A m⁻², indicating greater sensitivity of WCC performance to pH in this buffer system. While WCC exhibited comparable performance to CNT-based cathodes under bicarbonate buffer at pH 8.5, its relative performance dropped sharply at pH 7. TK-NAW-CNT, TN-NAW-CNT, and AW-CNT significantly outperformed WCC in this condition (p = 0.036, p = 0.034, and p = 0.044, respectively). This may be attributed to the pH-dependent buffering capacity of bicarbonate, which is significantly reduced at pH 7. Unlike CNTs, WCC relies more heavily on favorable local conditions due to its lower catalytic activity. CNT materials, with their high surface area and higher active sites, are better equipped to maintain hydrogen evolution rates under proton-limited or poorly buffered environments. This trend continued in the 100 mM bicarbonate buffer at pH = 7.0 condition where TK-NAW-CNT and TN-NAW-CNT significantly outperformed WCC (p = 0.041 and p = 0.047, respectively).

Switching to 200 mM phosphate buffer at pH 7.0, TK-NAW-CNT exhibited a significant recovery in performance (167.34 ± 1.24 A m⁻²) compared to bicarbonate buffer at pH 7.0 (p < 0.001, Table S29), though still below its performance in bicarbonate buffer at pH 8.5 (p < 0.001, Table S29). TN-NAW-CNT (159.24 ± 6.78 A m⁻²) and AW-CNT (155.81 ± 11.73 A m⁻²) also rebounded relative to their bicarbonate pH 7.0 values (p < 0.001 for both, Tables S28 and S30), while still underperforming compared to bicarbonate at pH 8.5 (p < 0.001 for both, Tables S28 and S30). WCC performance significantly improved in phosphate buffer at pH 7.0 (150.48 ± 2.85 A m⁻²) recovering from the sharp decline seen in bicarbonate at pH 7.0 (p < 0.001, Table S31) but again underperforming the performance in bicarbonate buffer at pH 8.5 (p < 0.001, Table S31).

All CNT materials demonstrated high stability and performance across all buffer systems and pH ranges, indicating robustness to changes in both pH and buffer type. In contrast, WCC showed greater variability, with performance more strongly influenced by both factors. This is likely due to catalytic active site limitations arising from the lower effective surface area of the material, making it more susceptible to local environmental changes. This vulnerability was observed most prominently in the bicarbonate buffer condition at pH 7.0, where diminished proton transport capabilities further exacerbated performance losses. Among the buffer and pH treatments tested, bicarbonate buffer at pH 7.0 resulted in the lowest current densities. This diminished performance is likely attributable to a reduction in effective buffering capacity at this pH, caused by the chemical equilibria between the bicarbonate ion (HCO₃⁻), carbonic acid (H₂CO₃), and dissolved carbon dioxide (CO₂). The dissociation of carbonic acid proceeds via two well-characterized acid dissociation constants: pKa₁ ≈ 6.3 for the H₂CO₃ ⇌ H⁺ + HCO₃⁻ equilibrium, and pKa₂ ≈ 10.3 for the HCO₃⁻ ⇌ H⁺ + CO₃²⁻ equilibrium. At pH 7.0, α-speciation calculations using these constants indicate that approximately 82% of the total dissolved inorganic carbon exists as bicarbonate (HCO₃⁻), while about 18% is present as carbonic acid (H₂CO₃), with negligible contribution from carbonate ion (CO₃²⁻). Importantly, carbonic acid is in rapid equilibrium with dissolved CO₂, which in turn is in dynamic exchange with CO₂ gas in the headspace. In this experimental system, the headspace was continuously released to reduce thermodynamic barriers associated with hydrogen partial pressure buildup. This exchange likely resulted in the loss of CO₂ gas, driving the equilibrium toward further conversion of bicarbonate to carbonic acid and subsequently to CO₂ gas. Over time, this process effectively removes carbonate species from the solution, leading to an appreciable decrease in buffer capacity.

Such carbonate loss not only weakens the buffering action, leading to more dramatic local pH changes near each electrode, but potentially disrupts proton transfer mechanisms essential for cathodic and anodic reactions. These findings emphasize the importance of buffer selection and pH control in bio-electrochemical systems, particularly under conditions where gas-phase CO₂ exchange cannot be eliminated or tightly controlled. In contrast, the phosphate buffer system is not subject to volatilization or gas-phase loss, as none of its acid-base species (H₃PO₄, H₂PO₄⁻, HPO₄²⁻, and PO₄³⁻) exist in equilibrium with a gaseous component. As a result, phosphate buffer concentration remains stable across pH levels, and is not diminished by degassing or atmospheric exchange.

Furthermore, the superior performance of all materials in bicarbonate buffer at pH 8.5 compared to phosphate buffer at pH 7.0 may be attributed to two factors: (1) higher pH creates more favorable conditions for the biological anode reactions, and (2) increased net proton flux resulting from pH-dependent buffer speciation and the corresponding diffusion coefficients of those species. For increases in anode performance, single-chamber MEC experiments have shown that anode biofilm development at slightly alkaline pH (8.0) exceeded development at neutral pH (7.0) in phosphate buffer system, indicating that pH 8.5 allows for more efficient cell synthesis and acetate oxidation than at pH 7.0, and that increases in performance could be observed with increasing pH in phosphate buffer [55]. However, a small increase in anode performance is expected to have a minimal effect on cathode performance given the extreme anode-to-cathode ratio used during testing periods.

Among the primary mechanisms of net proton flux, diffusion, electromagnetic migration, and convection, diffusion is considered dominant due to the low potential between electrodes, which minimizes the influence of the electric field on charged particles [56]. Although convective forces may arise from hydrogen gas bubbling at the cathode, proton flux due to convection is expected to be minimal, as no stirring or agitation was applied during the experiment. Given that diffusion of protonated buffering species (i.e., bicarbonate and phosphate) is the primary mechanism for proton flux, it is important to consider the unequal diffusion rates among the buffer species present. The α-speciation calculations indicate that at pH 8.5, bicarbonate exists primarily as HCO₃⁻ (~ 98%), with only residual amounts of carbonate (CO₃²⁻) and carbonic acid (H₂CO₃), while at pH 7.0, phosphate exists as approximately 62% dihydrogen phosphate (H₂PO₄⁻) and 38% hydrogen phosphate (HPO₄²⁻). HCO₃⁻ has a significantly higher diffusion coefficient (1.185 × 10⁻⁹ m² s⁻¹) compared to H₂PO₄⁻(0.879 × 10⁻⁹ m² s⁻¹) and, to an even greater extent, HPO₄²⁻(0.439 × 10⁻⁹ m² s⁻¹), due to its lower net ionic charge and consequently smaller hydrodynamic radius [57]. This means that the proton shuttling from anode-to-cathode can be faster for HCO₃⁻ than for either phosphate species, leading to a higher cathodic current. Additional benefits of reactor pH maintained at pH 8.5 versus pH 7.0, regardless of buffer type, is that at pH 8.5, carbon dioxide from the oxidation of acetate at the anode is driven into the system as bicarbonate buffer, thus increasing the effective buffer concentration.

In addition to buffer type, buffer concentration was also varied to assess the extent of performance attenuation at concentrations of 10 mM, 50 mM, 100 mM, and 200 mM. As potential waste streams for MEC operation have variable solution conductivity, finding cathode material that performs well at low buffer concentration is highly valuable. As buffer concentration decreased from 200 mM to 10 mM, all cathode materials exhibited significant declines in current density across all buffer and pH conditions (Figure 5), consistent with the expected reduction in ionic conductivity and buffering capacity. For each buffer type/pH group, performance was found to decrease linearly with buffer concentration (R² > 0.975) for all materials. Current density values for all materials under each buffer condition and concentration can be found in Supplementary Materials (Tables S16–S27).

Cathodic performance in the MEC was strongly influenced by both buffer type and pH, with the highest current densities observed under alkaline conditions (pH 8.5) in 200 mM bicarbonate buffer. CNT-based cathodes consistently showed higher current densities than WCC cathodes, with TK-NAW-CNT and TN-NAW-CNT exhibiting the highest and most stable performance across conditions. These results indicate that decreasing buffer concentration leads to a direct loss in performance for all materials regardless of buffer type or pH, providing further evidence that maintaining a highly buffered MEC environment is crucial for maximizing performance. These results also show that high current densities (>50 A m⁻²) are achievable at low buffer concentrations (10 mM) at the high anode-to-cathode ratios tested here. The use of a high anode-to-cathode ratio minimized anode-driven limitations on proton and electron fluxes, allowing cathodic current densities to be governed primarily by the electrochemical properties of the cathode materials. In practice, MEC performance is evaluated in terms of both electrodes and such disproportionate anode-to-cathode ratios cannot maximize volumetric current densities. However, these results indicate that cathodic current densities using these combined base materials and catalysts are not limiting factors, pointing to limitations in anodic performance or flux capacity of the medium. Further improvement to anode performance and waste stream pre-treatment methods which maximize buffer concentrations can help realize the potential of this cathode material.

3.5. Long-Term Stability of Cathode Material

Long-term stability is a critical factor in evaluating the viability of cathode materials for MECs, particularly in applications where extended operation, minimal maintenance, and material longevity directly influence system economics and feasibility. Materials that can sustain high current densities over time without significant performance losses reduce operational costs by lowering the frequency of replacement and maintenance interventions.

To assess long-term performance, MECs were operated continuously for 341 days under a constant condition of 1.0 V applied voltage and 200 mM phosphate buffer concentration. Current density values encompassing the structured testing periods with high anode-to-cathode ratio (60:1) were removed to assess the background operational condition with a lower anode-to-cathode ratio (7:1) (Figure 6). The higher current density values during testing periods were not evenly distributed over the 341 days of operation which would otherwise work to skew the performance trend over the testing duration. This lower anode-to-cathode ratio yielded correspondingly lower current densities and is the dominant condition over the entire length of the MEC operation.

Across the full operational period, the CNT-based cathodes consistently outperformed WCC in terms of average current density. AW-CNT, TN-NAW-CNT, and TK-NAW-CNT achieved mean current densities of 67.20 ± 16.21, 69.79 ± 16.06, and 67.69 ± 15.93 A m⁻², respectively, compared to 61.56 ± 15.69 A m⁻² for WCC. Furthermore, linear trend analysis revealed statistically significant declines in performance over time for WCC (slope = −0.0128, p < 0.0001), TN-NAW-CNT (slope = −0.0093, p < 0.0001), and AW-CNT (slope = −0.0225, p < 0.0001), while TK-NAW-CNT showed no significant trend (slope = −0.0003, p = 0.6906). The stability of TK-NAW-CNT, in particular, suggests strong long-term potential. This sustained performance, even under less favorable operating conditions, highlights the value of CNT materials for long-term MEC operation. In addition to improved electrochemical output, their durability offers potential cost savings through reduced material replacement intervals and lower system maintenance needs. Previous studies outline the importance of lowering electrode cost for moving MEC technology to real-world scales [58]. Estimates of more than half of MEC startup costs have been attributed to electrodes and catalyst cost [59]. These findings not only confirm the stability of these cathode base materials over time, but the combination of the base material with the Nafion binder and MoP catalyst. Unlike the higher anode-to-cathode ratio used for the structured testing period, the lower ratio (7:1), when all test cells are connected simultaneously, could impart an imbalance in mass transport to cells based on test cassette location. This is a potential limitation of this study.

3.6. Hydrogen Production Performance

Gas sample analysis was conducted under synthetic substrate condition containing 40 mM acetate and 200 mM bicarbonate buffer at pH 8.50. The influent medium was treated with 0.02% chloroform as described by Wang et al. (2020) [60]. All test cathodes were connected simultaneously, rendering the lower anode-to-cathode ratio and thus lower current density. Current densities were 70.56 ± 1.35 A m⁻², 67.07 ± 3.92 A m⁻², 65.55 ± 2.73 A m⁻², and 64.33 ± 5.01 A m⁻² for TN-NAW-CNT, AW-CNT, TK-NAW-CNT, and WCC, respectively. The first gas sample was taken approximately one hour after medium exchange, producing a very high cathodic hydrogen recovery (CHR) of 99.94%, indicating that each material can maintain high current density under hydrogen producing condition. This shows that without the development of planktonic hydrogen scavengers, a high CHR can be achieved. As the anodic biofilm microbes reduce the anode, these electrons are moved through the circuit to the cathode to reduce protons to hydrogen gas. Specialized methanogenic species have been shown to directly utilize these electron and electron potentials to reduce carbon dioxide to methane, thereby inflating the current density measurements. These results indicate that electrophy at the cathode surface by the electrothrophic methanogens is negligible, owing the high current density to hydrogenic processes. The CHR does, however, decrease over the next sampling period, which is likely due to incomplete reactor sealing or homoacetogenesis, as no methane was detected. An average CHR of 92.44% was found over the complete 5 h sampling period used for gas analysis.

4. Conclusions

This study demonstrates that carbon nanotube (CNT)-based hybrid cathodes supported on stainless steel and coated with molybdenum phosphide (MoP) offer a durable and cost-effective alternative to traditional carbon cloth cathodes in microbial electrolysis cells (MECs), across a broad range of operational conditions including variations in applied voltage, buffer type and concentration, substrate complexity, and pH. CNT cathodes consistently showed higher current densities than WCC, though only some differences were found to be significant. It is important to note that non-significant findings do not necessarily indicate the absence of a difference but may instead reflect limited sample sizes and reduced statistical power to detect an effect. Performance advantages were particularly pronounced under high applied voltages. The CNT materials sustained high performance over 341 days of operation, with minimal deterioration, underscoring their long-term stability and reducing the need for frequent maintenance or replacement. Additionally, high cathodic hydrogen recoveries were achieved under real-feedstock conditions, demonstrating the viability of these materials in practical applications without added buffer, minerals, vitamins, or synthetic substrates. These results collectively indicate that CNT-clad stainless-steel electrodes not only enhance MEC performance but also address key economic and operational barriers to scaling up this technology. By combining conductivity, structural integrity, and long-term electrochemical performance, CNT-based cathodes represent a promising path forward for cost-competitive, energy-positive wastewater treatment.