Abstract
The redox potential (ERedox) of a ferrocene (Fc) derivative differs, depending on its functional group. In this study, the various Fc derivatives are considered as mediators of anodic catalysts to promote glucose oxidation reaction (GOR) in glucose/oxygen enzymatic biofuel cells (EBFCs). Initially, their lowest unoccupied molecular orbital (LUMO) and highest occupied molecular orbital (HOMO) energies are calculated using density functional theory to predict their ERedox pattern. According to the calculations, the LUMO and HOMO energies of Fc derivatives combined with electron-donating groups (EDGs) are higher than those of Fc derivatives combined with electron-withdrawing groups (EWGs), including the results that Fc(NH2) has the highest molecular orbital (MO), while Fc(CHO) has the lowest. To verify the prediction for ERedox pattern, electrochemical evaluations are conducted. When glucose is provided, the onset potential (EOnset) of GOR is measured, while the ERedox of Fc derivatives and EOnset of GOR are linearly proportional to each other () and DFT calculations. As the energies increase, the above two potentials are shifted more negatively. More specifically, the ERedox of Fc(NH2) and EOnset of GOR show the most negative values at (−0.112 and −0.17) V, respectively, while the ERedox of Fc(CHO) and EOnset of GOR show the most positive values at (0.496 and 0.40) V, respectively. Then, when this correlation is adopted for EBFCs, EBFCs using Fc(NH2) combined with EDG show 3.7 times higher maximum power density than those using Fc(COOH), as representatives combined with EWG. Based on that, it is well established how DFT and electrochemical evaluations should be used to design anodic catalysts including Fc derivatives as mediators for GOR.
1. Introduction
The enzymatic biofuel cell (EBFC) is a device that converts energy using enzyme as catalyst [1, 2]. By using enzymes, the EBFC can be operated under milder conditions, such as room temperature (RT) and natural pH, than other conventional fuel cells. Also, metabolites, such as ascorbate, creatine, and lactate, [3–5] produced in the body, or saccharides, such as glucose, fructose, and sucrose that are biocompatible materials, can be used as fuel [6, 7]. In addition, since enzymes have good fuel selectivity, EBFC can be operated even in conditions of mixed fuel without membrane, and thus, the design flexibility of EBFC is very liberal [8–12]. Due to these characteristics, EBFC is mainly used as a power source for portable, wearable, and implantable devices [13–15].
However, despite these benefits, there are some limitations to enzyme-based catalysts, because the cofactors included in the enzyme body play an important role in transferring electrons and reacting with fuel, but they are placed deep inside the protein shell of the enzyme [16, 17]. To overcome this issue, the enzymes should use electron transfer mediators that are electrically wired with electrode [18–20]. When the mediators are used for transferring electrons, such kind of electron transfer route is denoted as mediated electron transfer (MET), and MET is known as the best way to transfer electrons to electrode [21]. However, this MET induces severe cell voltage loss of the EBFC, because the reaction potential is determined by the inherent redox potential (ERedox) of the mediator [22]. Therefore, mediators should satisfy the following three characteristics: (i) their reversible redox reaction should occur at a proper potential range, (ii) they should be electrically wired to enzyme and electrode, and (iii) they should not be reactive with the substrate. For example, for anode, the occurrence of the more negative potential by MET may affect the increase in open circuit voltage (OCV) for the performance enhancement of the EBFC.
The representative precious metal-based mediators, such as osmium and ruthenium, have satisfied these factors to date [23, 24]. In the early stage, they were considered as mediator for anode with glucose oxidase (GOx) in the EBFC and showed good performance of the EBFC because their ERedox was observed in the negative potential range. For example, Heller’s group fabricated redox polymer by the crosslinking of osmium and polymer. Here, anode including the redox polymer had ERedox of −0.2 V (vs. Ag/AgCl) [25]. The Cosnier group used ruthenium complex for the anodic catalyst with GOx, and the ERedox of the catalyst was −0.1 V (vs. Ag/AgCl) [26]. In addition, the mediator that did not contain precious metals, such as ferrocene, tetrathiafulvalene, or naphthoquinone, which are less expensive than precious metals, were also used as mediators [27–29]. However, these mediators have their ERedox in relatively positive potential ranges. Kwon’s group fabricated the anodic catalyst using tetrathiafulvalene as a mediator, and the ERedox of the anode was observed at 0.05 V vs. Ag/AgCl [30, 31]. Urban’s group conjugated a redox polymer with ferrocene (Fc) and adopted the polymer as anodic catalyst, and its ERedox was observed at 0.1 V (vs. Ag/AgCl) [32].
Among the various mediators, ferrocene (Fc), which is an organometallic compound consisting of two cyclopentadienyl rings having a center core of Fe, has excellent electrochemical properties and is used in both the EBFC and in many commercial energy devices [33–36]. For example, Zhengjin’s group calculated the lowest unoccupied molecular orbital (LUMO) energy of synthesized Fc derivatives by density functional theory (DFT) and established a correlation between the simulation-driven and experimentally measured ERedox. According to their calculations, as the LUMO energy of ferrocenium (Fc+) increased, its ERedox was shifted to a more negative potential range [37]. In addition, Curtiss’s group analyzed the ERedox of 4,178 molecules including Fc derivatives through a high-throughput screening method. As a result, it was explained that the ERedox was shifted differently by the types of functional groups that were bound to the cyclopentadienyl ring within Fc [38].
This suggests that when Fc derivatives are considered as mediators for GOx-based anodic catalysts, the performance of the glucose/oxygen EBFC will be predicted through DFT calculation of the LUMO and the highest occupied molecular orbital (HOMO) energies of the Fc derivatives. Appropriate Fc derivatives predicted by DFT calculation may induce more negative potential values of the (i) ERedox of anodic catalysts, and (ii) onset potential (EOnset) of the glucose oxidation reaction (GOR). Furthermore, with the benefits of ERedox and EOnset, both the OCV and power density of glucose/oxygen EBFC may increase considerably.
In this study, DFT was used to calculate the LUMO and HOMO energies of eight different Fc derivatives combined with different functional groups. Their ERedox of Fc derivatives and EOnset of GOR were then measured in cyclic voltammetry (CV) graphs when Fc derivatives were used as anodic catalyst with GOx, to establish the correlation between the calculation data and experimental results. Although Fc derivatives have been widely used as mediator for the glucose/oxygen EBFC system, the prediction of ERedox and EOnset of anodic catalysts including Fc derivatives by DFT calculation of the LUMO and HOMO energies has not previously been reported. Finally, the performance of EBFCs was investigated in polarization curves to confirm the increment of performance according to the ERedox of Fc derivatives and the EOnset pattern of GOR.
2. Experimental
2.1. Materials
Multiwall carbon nanotube (MWCNT, of ~95% purity), was purchased from Nano Lab (Brighton, MA). Glucose oxidase (GOx, Type X-S, (100,000−250,000 unit/g solid), Nafion 117-containing solution (~5% in a mixture of alcohols and water), ferrocene (Fc), ferrocene carboxylic acid (Fc(COOH)), ferrocene carboxaldehyde (Fc(CHO)), vinylferrocene (Fc(CHCH2)), ferrocenemethanol (Fc(CH2OH)), and acetyl ferrocene (Fc(COCH3)) were obtained from Sigma–Aldrich. Anhydrous D-(+)-glucose was purchased from Alfa Aesar. Phosphate buffer saline (1X PBS, pH 7.4) was acquired from Thermal Fisher. Aminoferrocene (Fc(NH2)) and 1,1-dimethyl ferrocene (Fc2∙(CH3)) were purchased from Tokyo Chemical Industry. Bilirubin oxidase (BOD, from Myrothecium sp., 2.28 unit/mg) was obtained from Amano Enzyme (Nagoya, Japan). 2-Propanol (IPA, 99.9%) and ethyl alcohol (ethanol, 99.5%) were purchased from Samchun Chemicals.
2.2. Preparation of Electrodes
To fabricate anodic catalysts consisting of CNT/Fc derivatives/GOx, 10 μL of 5 mg/mL CNT dispersion in ethanol, 10 μL of 10 mM Fc derivatives, 10 μL of 40 mg/mL GOx solution (dissolved in PBS), and 4 μL of 0.5 wt.% Nafion were sequentially drop cast on glassy carbon electrode (GCE) and dried at RT. To fabricate cathodic catalyst consisting of CNT/BOD, 10 μL of 5 mg/mL CNT dispersion in ethanol, 10 μL of 10 mg/mL BOD solution dissolved in PBS, and 4 μL of 0.5 wt.% Nafion were sequentially drop cast on GCE.
2.3. Electrochemical Evaluations
Cyclic voltammetry (CV) tests were carried out in a three-electrode system using a Bio-Logic SP-240 electrochemical workstation (Bio-Logic, USA) connected to a personal computer. GCE (of diameter 5.5 mm) was utilized as working electrode, while platinum wire was considered as counter electrode, and Ag/AgCl (sat. in 3.0 M NaCl) was considered as reference electrode. All the tests were implemented under PBS solution, and the scan rate of potential was 20 mV/s [39, 40].
To measure the electrochemical performance of EBFCs, an in-house enzymatic biofuel-cell kit was used, and the polarization curve of EBFCs was measured. To measure their polarization curve, catalyst-loaded GCE was used as electrode. For fuel for the EBFC, PBS containing 0.1 M glucose was rotated from an external reservoir to the EBFC kit at a flow rate of 4 mL/min under the O2 state.
2.4. DFT Calculations
All theoretical calculations were conducted with BLYP, B3LYP, and BHandHLYP functionals applied with no symmetry restriction in the gas phase. For main group elements, the 6-31G(d,p) basis set was utilized, while the LanL2DZ ECP (effective core potential) was employed for the analysis of Fe element [27, 28, 41]. The energies of Fc and seven Fc derivatives were calculated to obtain their energy-minimized structures, while the energy-minimized structures of the oxidized form of Fc that were named ferrocenium (Fc+) and its derivatives were also calculated. Vibrational frequency computations showed that the energy-minimized geometries of Fc and Fc+ derivatives have no imaginary frequency, indicating that these geometries correspond to local minima of energy. The solvent effect (water, ) was taken into account by using the Cossi and Barrone CPCM (conductor-like polarizable continuum model) in all of the calculations [42]. Gaussian 16 suite was used to carry out all the theoretical calculations [43].
3. Results and Discussion
3.1. DFT Calculations of Fc Derivatives
The electron transfer mechanism of anodic catalysts containing mediator and GOx is presented by the following reactions: [44, 45]
Figure 1 shows that electrons produced by GOR with the help of flavin adenine dinucleotide (FAD) cofactors inside the enzymes, such as GOx and glucose dehydrogenase, are transferred to the mediator through a reversible redox reaction of FAD cofactor (Equations (2) and (3)). Here, the mediator plays a critical role in promoting electron transfer between enzyme and electrode (Figure 1(a)). More specifically, electrons are initially produced by GOR (Equation (1)), and the electrons are transferred to the FAD cofactor of the enzyme, reducing the FAD into FADH2 (Equation (2)). Then, such reduced FADH2 transfers electrons to the mediator, reducing the mediator (Equation (3)), and such reduced mediator transfers electrons to the electrode, oxidizing the reduced mediator (Equation (4)). This sequence is then repeated. For the circulation, HOMO energy of reduced mediator and LUMO energy of oxidized mediator are needed (see Figure 1(b)), meaning that the HOMO energy of the reduced mediator (Fc derivative) and the LUMO energy of the oxidized mediator (Fc+ derivative) are calculated by DFT.

In previous DFT studies on Fc derivatives [46], as the Hartree–Fock (HF) exchange fraction () in hybrid functionals increased, more errors occurred in predicting the ERedox of Fc derivatives. Based on that, three DFT methods for BLYP series (BLYP, B3LYP, and BHandHLYP) are utilized to calculate the MO energies for each Fc derivative, and such calculated energies are experimentally verified. For the calculation of MO energies using the DFT method, HF exchange fraction () increased in the order (Figure 2), and as increased, the energy gap between the LUMO energy of Fc+ derivatives and the HOMO energy of Fc derivatives increased. Here, Fc derivatives with higher LUMO energy contained more electron donating groups (EDGs), such as Fc(NH2) and Fc·(CH3)2, while those with electron-withdrawing groups (EWGs), such as Fc(COCH3) and Fc(CHO), showed a relatively lower LUMO energy than other Fc derivatives. Therefore, it is expected that as the LUMO energy of Fc+ derivatives increases, their ERedox will be located in a more negative potential region.

3.2. Measurements of Redox Potential and Onset Potential for GOR of Fc Derivatives
In onset potential (EOnset), the oxidation or reduction reaction starts to produce current, while EOnset can be determined by the applied catalyst. Open circuit voltage (OCV) is defined as the difference between the EOnset of the anodic and cathodic currents, which directly affects the power density of the EBFC. Theoretically, as the difference between the EOnset values increases, OCV increases (Figure 3) [47]. Therefore, to improve the power density of EBFCs, the EOnset that is related to the anodic catalyst should be placed in a more negative potential range, while the EOnset that is related to the cathodic catalyst should be placed in a more positive potential range.

To measure the ERedox of Fc derivatives and EOnset of GOR when Fc derivatives are used as mediator of anodic catalyst with GOx, their CV curves were measured. More specifically, anodic catalysts consisting of CNT/Fc derivatives/GOx were prepared to observe the ERedox of Fc derivatives. After that, glucose was injected into the electrolyte to measure the EOnset of GOR. Methodologically, the EOnset of GOR was calculated by the tangent method. This method measures the tangent between the baseline before GOR (non-Faradaic current) and the line where GOR occurs (Faradaic current), while the intersection of the two straight lines is defined as EOnset (Figure 4) [48].

(a)

(b)

(c)

(d)

(e)

(f)

(g)

(h)

(i)
Actually, considering Fc(NH2), Fc2·(CH3), and Fc(CH2OH) as representatives of EDG, their ERedox was observed at −0.112, 0.1, and 0.202 V (vs. Ag/AgCl), while their EOnset for GOR was measured at −0.17, 0.03, and 0.07 V (vs. Ag/AgCl) (Figures 4(a)–4(c)). This means that the EOnset of the three Fc derivatives is placed in a more negative potential range than that of Fc (when Fc is used, its ERedox of Fc derivatives and EOnset of GOR are 0.203 and 0.08 V (vs. Ag/AgCl) (see Figure 4(d)). Even in previous studies using MET, the mediator that does not contain precious metals, such as Fc, tetrathiafulvalene, and quinone derivative, showed relatively more positive EOnset than its metal competitors, such as osmium- and ruthenium-based derivative. However, Fc(NH2), although it was based on organometallic Fc, showed excellent EOnset of GOR.
In contrast, the ERedox of Fc derivatives and EOnset of GOR of the four Fc derivatives including EWG were placed in a more positive potential range than that of Fc. Quantitatively, the ERedox of Fc(CHCH2), Fc(COOH), Fc(COCH3), and Fc(CHO) was 0.231, 0.289, 0.486, and 0.496 V (vs. Ag/AgCl), while their EOnset of GOR was 0.14, 0.22, 0.37, and 0.40 V (vs. Ag/AgCl), respectively (Figures 4(e)–4(h)). These results show that the ERedox of Fc derivatives and EOnset of GOR are strongly affected by the functional group combined with Fc, and the Fc derivatives combined with the amine group or dimethyl functional group were the most suitable mediator candidates for anodic catalyst to improve the maximum power density (MPD) of the EBFC. Figure 4(i) corroborates that the ERedox of Fc derivatives is linearly proportional to their EOnset for GOR. ERedox of Fc derivatives and EOnset for GOR we measured in CV curves are summarized in Table 1.
The ERedox of Fc derivatives and EOnset for GOR of the Fc derivatives were also compared with the MO energies of each Fc derivative (Figure 5). More specifically, as the HF fraction (%) increased in the order BLYP < B3LYP < BH and HLYP, the degree of linear correlation between MO energies and ERedox of Fc derivatives was reduced with low , while a pure functional BLYP method used for measuring the HOMO energy of Fc derivatives showed the best linear correlation with of 0.883 (Figure 5(b)). Even when the EOnset of GOR was compared with the LUMO energy of Fc+ derivatives, the BLYP method showed the best linear correlation with of 0.925 (Figure 5(c)). Since the linearity between MO energy and ERedox of Fc derivatives is the highest when the BLYP method is used, it is speculated that BLYP is the most suitable method to evaluate the electrochemical performance of Fc derivatives. In addition, the above simulations are evidence that the EOnset of GOR using Fc derivatives as mediator can be predicted by DFT calculation.

(a)

(b)

(c)

(d)
3.3. Performance Evaluations of Glucose/Oxygen EBFCs Using Fc Derivatives
The polarization curves of glucose/oxygen EBFCs using anodic catalysts including Fc derivatives are measured to investigate the effects of Fc derivatives on the performance of EBFCs (Figure 6). As parameters showing the performance of EBFCs, their OCV and MPD were measured. For the performance measurements of EBFCs, CNT/Fc derivatives/GOx catalyst was loaded on GCE as anode, while CNT/BOD was loaded on GCE as cathode. Here, bilirubin oxidase (BOD) containing Cu-based cofactor that is operated by DET mechanism was included as cathodic catalyst to promote the oxygen reduction reaction (ORR) [49]. The selected Fc derivatives were Fc, Fc(NH2) including EDG, and Fc(COOH) including EWG. As a control, EBFC using CNT/GOx catalyst as anode without mediator was prepared. To confirm the reproducibility in the performance of EBFCs, the OCV and MPD of EBFCs were measured three times.

According to Figure 6, the OCV and MPD of control EBFCs operated without mediator (CNT/GOx) were lowest at 0.34 V and 4.2 μW/cm2, while those of EBFCs using the EWG-based CNT/Fc(COOH)/GOx were 0.36 V and 5.9 μW/cm2, while those of EBFCs using CNT/Fc/GOx were 0.46 V and 8.6 μW/cm2. Then, those of the EBFCs using the EDG-based CNT/Fc(NH2)/GOx were highest at 0.63 V and 22 μW/cm2. The above polarization curve measurements confirmed two important findings. First, as the EOnset of GOR was more negatively shifted, the OCV of EBFC increased; and second, the MPD of EBFC was proportional to its OCV. Conclusively, the LUMO and HOMO energies of the Fc derivatives calculated using DFT were approximately linearly proportional to the patterns in ERedox of the Fc derivatives and EOnset of GOR, and they were well matched with the performance of EBFCs. Based on this study, the performance of EBFCs using Fc derivatives as mediator can be predicted by calculating their LUMO and HOMO energies by DFT.
4. Conclusion
In this study, when Fc derivatives were used as mediator for the anodic catalyst of the EBFC, their ERedox of Fc derivatives and EOnset of GOR were initially predicted by using DFT, and the prediction was experimentally verified by two modes: half-cell tests of the anodic catalyst and the performance measurements of the EBFC. As a result, a correlation was explained between the performance of EBFCs using Fc derivatives and the two potentials linked to Fc derivatives. The LUMO and HOMO energies of Fc derivatives were calculated by DFT to predict how their ERedox was affected by functional groups, such as EDGs and EWGs. According to the calculations, Fc derivatives combined with EDG showed higher LUMO and HOMO energies than did Fc. The ERedox of Fc derivatives and EOnset for GOR were evaluated by their CV curve measurements to determine the electrochemical behavior of anodic catalyst containing Fc derivatives and identify correlation patterns with their LUMO and HOMO energies. It was found that the ERedox and EOnset for GOR of the EDG-attached Fc(NH2) were lowest at −0.041 and −0.17 V, respectively, while the EWG-attached Fc derivatives showed more positive potentials than did Fc. In addition, a linearly proportional relationship between the ERedox of Fc derivatives and the EOnset of GOR and a correlation between the MO energies and potentials were observed. Regarding the performance of the EBFCs, the polarization curves of EBFCs using Fc derivatives as mediator for anodic catalyst were measured. According to the measurements, the pattern in the OCV and MPD of EBFCs was well matched with that acquired from the CV curve measurements and DFT calculations. Based on that, it is substantiated that when Fc derivatives showing high LUMO energy are prepared, the Fc derivatives can be used as the best mediator to improve the performance of EBFCs.
Data Availability
The underlying data supporting the results of your study can be found, including, where applicable, hyperlinks to publicly archived datasets analyzed or generated during the study.
Conflicts of Interest
The authors declare that they have no conflicts of interest.
Authors’ Contributions
Joonyoung Lee, Jungyeon Ji, and Jae Jun Lee contributed equally to this work.
Acknowledgments
This research was supported by the Basic Science Research Program through the National Research Foundation of Korea (NRF), funded by the Ministry of Education (2021R1A6A1A03039981), and the National Research Foundation of Korea (NRF) and the Ministry of Education of the Republic of Korea (MOE) (no. 2023R1A2C200244411).