Introduction

For more than a decade, the poly-indolequinone eumelanin (named “melanin” in the following) has been of interest to electrochemists and material scientists alike due to their properties and functions within the human body1. In addition, photoprotection, radical scavenging, and metal ion chelation are properties of the diffuse polymer that have been used for sensing and energy storage2,3.

To optimize the polymer for these applications, the main focus has been mainly to increase the solubility of melanin in polar solvents to increase film processibility4,5,6. This has led to functionalization like galactosyl-thio or poly(vinyl alcohol) substitution by Corani et al. for water-soluble melanin analogues, or the sulfonated melanin soluble in DMSO by Bronze-Uhle et al.7,8,9. There exists no systematic study backed up by density functional theory (DFT) calculations aiming to understand the mechanisms on how functional groups of different polarity and steric requirements would affect the electrochemistry and charge transport of melanin materials10. This work aims to fill that gap by looking at how two fundamentally different functional groups substituted at different molecular positions affect the morphology and particle size of the resulting polymer and the energetic differences between the highest occupied molecular orbital (HOMO) and lowest occupied molecular orbital (LUMO) on a molecular level. The derivatives were chosen to serve as model molecules for structures with functional groups of similar size and polarity while considering the ease of synthesis. As a model group for reducing metal-ion chelation and increasing sterical bulkiness, tert-butyloxycarbonyl (Boc)-groups were introduced to the previously conjugated nitrogen atoms, giving “Mel-Boc”11. As a model electron-withdrawing group (EWG), nitro-groups were introduced, giving “Mel-NO2”. A figure with Lewis formulas of both structures can be found in Fig. S1. The electrochemical performance of the derivatized melanins was then analyzed in order to correlate it to the spectroscopic/microscopic and computational findings. This enables predictions on other functionalization of melanins for future research. Secondary Zn coin cells were evaluated using the derivatized melanins as the working electrode (cathode material) and Zn sheets as anode in both aqueous and ionic liquid electrolytes depending on the solubility of the synthesized material.

Organic Zn-ion batteries have been of growing interests as alternatives to lithium-ion batteries due to the safety, small ionic size, and low cost of Zn metal. They have also been shown to be well-fitted for testing melanins in electrochemical applications12. To keep the devices safe with either non-toxic or low vapor-pressure electrolytes, the water-insoluble Mel-Boc was tested in aqueous solutions (2 M ZnSO4 and 0.1 M ZnSO4 in 0.9 M Na2SO4 as supporting electrolyte (SE)), while the water-soluble Mel-NO2 was tested in room temperature ionic liquid (RTIL) solutions. Specifically, the interest in such aqueous electrolyte systems and their ion storage mechanisms has been increasing due to their sustainability benefits, as seen in recent literature covering both zinc-organic batteries and more high capacity energy storage systems13,14. To compare the results using RTILs to previous findings, the established RTIL mixture of zinc di-[bis(trifluoromethylsulfonyl)imide] (Zn(TFSI)2) in 1-ethyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide (EMIM TFSI) was used15. This combination has been used due to the high conductivity of sulfonium- and imidazolium containing RTILs mixing well with heterocyclic compounds16. Additionally, pyrrolidinium-/piperidinium-based RTILs are known for being stable and having large electrochemical windows17,18.

The insights obtained from analyzing the two different melanin derivatives show the importance of metal-ion chelation, band gap modification, and redox center availability, when optimizing melanin and melanin-like materials for energy storage. To better understand the rationale behind selecting the specific model molecules, melanin and the derivatives will be firstly introduced in more detail.

Results and discussion

Melanin and the derivatives

The material class of melanins contains substances such as eumelanin, pheomelanin, neuromelanin, and allomelanin1. The macromolecular and disordered polymer eumelanin that is interesting for electrochemical applications, consists of two different monomer units. These are 5,6-dihydroxyindole-2-carboxylic acid (DHICA) and 5,6-dihydroxyindole (DHI)19. The structural formulas of the different melanin redox states (hydroquinone (HQ), semiquinone (SQ), indole quinone (IQ), and quinone imine (QI)) are depicted in Fig. 1.

Fig. 1: The different possible redox states of eumelanin monomer units (top) and the comproportionation reaction (bottom).
figure 1

For DHI species, R = H, and for DHICA species, R = COOH.

Figure 1 also depicts the comproportionation reaction that has been found to be the central reaction for the hydration-dependent electron transport/conductivity and energy storage of melanin polymers5,20. A central discussion in melanin research concerns the general structure of the poly(indolequinone) in bulk materials. One postulated structure is a disordered model, where oligomers of different sizes are present21. One of the main structure motifs of these oligomers was found to be a tetrameric porphyrin-like structure, especially in melanins from natural sources22,23. This work will re-evaluate the tetrameric structure that is still being discussed in very recent work24 by using semi-empirical approaches and DFT calculations to justify using the tetramer as a base for band gap calculations for melanin derivatives with different degrees of functionalization.

The black powder Mel-Boc is obtained by modifying melanin post-polymerization with di-tert-butyl dicarbonate by using a catalytic amount of iodine. As the bulky Boc-groups are used for protecting secondary amines in organic synthesis25, the formation of Mel-Boc was expected to be very selective to the indole nitrogen atoms of the melanin monomer units. This positions the bulky Boc-groups in the center of the proposed porphyrin-like tetrameric oligomers that have been theorized to be important for metal ion chelation in melanin26, Mel-Boc was selected as a derivative to investigate the role of metal ion chelation for the electrochemical properties of melanin in energy storage devices. The structure of Mel-Boc can be seen in Fig. 2a.

Fig. 2: XPS and FTIR results.
figure 2

Spectroscopic analysis of (ad) Mel-Boc films and (eh) Mel-NO2 films. a and e show ATR FTIR-spectra; (bd) and (fh) XP (X-Ray Photoelectron) spectra. Important bands in (a and e) are marked with wave numbers directly next to the plotted graph. For both substances, the C 1 s (b, f), O 1 s (c, g), and N 1 s (d, h) spectra are shown.

Adding nitric acid to melanin post-polymerization, neutralizing the pH level, and removing solvents yields the orange-brown powder Mel-NO2. This derivatization was chosen, as it has been shown in the literature that adding strong EWGs to organic electrode materials can modify the electronic band gap27. This can result in a more easily occurring redox reaction and subsequent charge transfer. Changes in interactions with composite materials can also potentially increase the energy storage capabilities and conductivity in built devices27. The introduction of nitro groups was also expected to lead to higher film processibility and film consistency when using the strongly polar N-methyl-2-pyrrolidone (NMP) as a slurry solvent28. The structure of Mel-NO2 can be seen in Fig. 2e. The synthesis of both derivatives was confirmed using NMR spectroscopy (Fig. S2), FTIR (Fig. 2a, e), and X-ray photoelectron spectroscopy (XPS) (Fig. 2b–d, f–h). The common FTIR-bands (shown in black) for both samples are the following: IR (melanin derivatives): \({\tilde{\nu }}_{max}\) /cm−1 = 3400 (br, indole NH stretching), 3200 (br, OH stretching), 1650 (m, conjugated ketone CO stretching), 1580 (m, indole ring vibration), 1390 (m, CN in-plane vibration), 1240 (m, CN ring stretching). They match literature findings expected for melanins29,30,31. Prominent additional bands (shown in red) are for Mel-Boc: 2900 (m, aliphatic CH stretching) and 1210 (m, ester CO stretching) and for Mel-NO2: 1340/1550 (m, nitro compound NO stretching (symmetric/asymmetric)). These bands match literature values for comparable derivatized compounds and qualitatively confirm that the desired species are present in both Mel-Boc and Mel-NO232,33.

For quantitative confirmation of the degree of derivatization, XPS was performed on casted films on evaporated Au thin films on glass slides (named Au/glass) as a substrate. The survey spectra can be seen in Fig. S3 and show the presence of Au, C, O, and N on the samples, which confirms the presence of the two types of films. The high-resolution spectra are shown in Fig. 2b–d for Mel-Boc and (f–h) for Mel-NO2. The spectral components of the high-resolution data are shown in Table 1.

Table 1 The assignment of components of XP spectra of Fig. 2 to the structures of melanin films: For better comparison, the literature binding energies of Mel-Syn are included36

The observed spectra could be fitted with a maximum of four spectral components with constant full width half maximum (FWHM), to which a structural element of Mel-Boc or Mel-NO2 could be assigned. The C 1s-data of both Mel-Boc and Mel-NO2 show the presence of species to be expected in the poly(indolequinone) structure of melanin34. The peaks for the sp2 and sp3 carbon species, as well as the C-N/C-O and C = O species, are all present. For Mel-NO2, the relative intensities are similar to synthetic melanin (Mel-Syn) as expected, as no other carbon species were introduced into the system. The C-N/C-O peak for Mel-NO2 (EB = 285.9 eV) is shifted to higher binding energies when compared to Mel-Boc.

The Mel-NO2 spectrum exhibits three different nitrogen species. The highest intensities belong to the indole nitrogen species (EB = 399.3 eV) and newly introduced nitro group species (EB = 406.3 eV) with typical shifts for these species35. The low intensity peak at EB = 402.4 eV can be assigned to protonated indole nitrogen generated by the acid treatment. The comparison of peak area ratios of the components in the N 1 s spectra yields a mole fraction of the nitro to indole N-atoms of 0.54. This indicates a degree of nitration of slightly over 50% that can be considered when modelling the system using DFT calculations. As Mel-NO2 does not modify the indole nitrogen, the corresponding signal component is also not shifted. The Mel-Boc N 1s-data however, shows that the indole peak (EB = 399.0 eV) is shifted to lower binding energies compared to Mel-Syn36, corresponding to an increase of binding energy observed for the C 1s-data.

Lastly, the O 1s-data for Mel-Boc shows three peaks. The shift in binding energy of O = C (EB = 530.1 eV) and O-C/C-O-C (EB = 531.1 eV) decreased compared to Mel-Syn, observed in previous work36. This is caused by the presence of additional ester moieties, shifting the electron density due to the higher oxygen electronegativity. The peak at EB = 534.6 eV corresponds to Na KLL Auger signals stemming from trace leftovers from the derivative synthesis route37. The shifts observed in the Mel-NO2 O 1s-data also correspond to the observed shifts in C 1 s binding energy discussed above. From these spectroscopic investigations, the significant presence of the introduced functional groups was confirmed, and the information can be used further in computational studies.

Modified material properties of the derivatives

SEM provides vital information on morphology and particle size after derivatization that can affect film processibility and thus electrochemical performance38. SEM-images of aqueous Mel-Boc and Mel-NO2 dispersions casted on Si wafer substrates at different magnifications can be seen in Fig. 3. Particle sizes were obtained from 100 sample spots across three different sample areas. The stated ranges indicate the size distribution of the particles and the uncertainty of its determination and was taken as the standard deviation of the particle sizes obtained from the software ImageJ. Both materials show morphologies different from synthetic melanin39. For Mel-Boc, rigid and sharp non-uniform chunks, and particles with an average size of (520 ± 23) nm can be determined, with sampling across homogenous areas of the material. Agglomerations of the particles can have sizes of up to roughly 1.5 μm. These particularly large chunks could be the result of the sterically demanding Boc-groups increasing the average spacing between melanin oligomer units and thus increasing the free volume inside the bulk material, which has been reported to result in material changes in organic systems40. Not only should the Boc-groups affect the metal ion chelation of the melanin oligomer units due to interfering with the assumed porphyrin-like structure, but this observed morphological change may also affect the performance of energy storage devices using Mel-Boc as the working electrode material.

Fig. 3: Field emission scanning electron microscope (FE-SEM) images.
figure 3

Depicted are Mel-Boc (a, c) and Mel-NO2 (b, d) on a Si wafer substrate. The images were obtained with secondary electron imaging while using an acceleration voltage of 5.0 kV.

For Mel-NO2, particles with a non-uniform, flakey morphology similar to Mel-Syn are observed. They have an average particle size of (240 ± 13) nm, with sampling across homogenous areas of the material. While these values for Mel-Boc and Mel-NO2 represent statistically averaged estimates, the SEM images in Fig. 3 clearly reveal that both materials display broad size distributions and irregular particle morphologies. As it is expected that only free carbon atoms of the poly(indolequinone) were modified by post-polymerization with nitro groups being relatively small compared to the bulky Boc-groups of Mel-Boc, this could explain the morphological similarities of Mel-NO2 to Mel-Syn. Furthermore, Mel-NO2 was observed to be surprisingly water-soluble during solubility tests in polar solvents, even though nitro-compounds are known to not explicitly increase water solubility41. The change could be a result of the nitro groups significantly affecting the conjugated electronic structure of the indole tetramers. This indicates that Mel-NO2 is a promising material in terms of film processability, which might positively affect the material performance in energy storage applications, besides the changes in electronic structure introduced by the nitro group.

As the Mel-NO2 films seemed promising, ultraviolet photoelectron spectroscopy (UPS) was performed on the Mel-NO2 films on Au/glass. The results are shown in Fig. S4 as a plot of intensity versus the binding energy EB referenced to the Fermi energy EF. The same set-up was used in previous tests where EHOMO of identically prepared synthetic melanin films was determined as (1.3 ± 0.2) eV36. The samples were referenced versus EF of a clean gold sample. From the intersection of abscissa and a tangent to the shoulder of the signal, EHOMO of Mel-NO2 could be determined as (1.9 ± 0.2) eV. This falls in the range of typical melanin HOMO energy values found in literature for poly(indolequinone) systems42. As the HOMO energy of the Mel-Syn and Mel-NO2 films produced in these works are in agreement with previous literature findings, the HOMO-LUMO gaps in these previous works can be used for benchmarking the accuracy of new DFT calculations on the proposed tetrameric structure for Mel-Syn oligomers42. After these DFT calculations are proven to be accurate, they can be used as a basis for predicting the band gap of nitrated melanin tetramers, using the previously discussed XPS data as a reference for the estimated degree of nitration. Mel-Boc could not be analyzed this way, as the unstable film was visibly removed from the substrate while being probed under photons incidence in ultra-high vacuum (UHV) conditions.

Computational results

DFT is valuable for predicting the influence of chemical modifications on electronic properties like the band gap size and thus performance in electronic and opto-electronic devices43. For melanin, it has been mainly used for estimating the structures of potentially stable oligomers. However, the theoretical foundation for these calculations from Kaxiras et al.21,44 is by now over a decade old, and the accuracy of computational approaches in chemistry and materials science has been improving vastly43. Several recent works have shown that further investigation of the proposed tetramer structure as an energetic minimum-structure of melanin is still of high interest for the ongoing structural discourse for melanin24,26,45. Additionally, analyzing free tetramers by calculation has proven to be a good approximation for the changes in the electronic structure of the bulk material after chemical modification. Thus, before using post-DFT methods for predicting the influence of nitro groups on the redox properties of melanin, the stability of the tetrameric structure had to be verified using the recent Conformer-Rotamer Ensemble Sampling Tool (CREST)46. The most stable modeled Mel-Syn tetramer could then be used to select the Mel-NO2 models to study further with different degrees of functionalization.

Utilizing the “tautomerize” function from CREST, all possible permutations of HQ/SQ/IQ/QI monomers inside the tetramer formation were studied at semi-empirical level instead of just hand-picking an arbitrary number of tetramers. In the next step, the most stable base-structure was identified by ranking formation energies after structural optimization of pre-selected tetramers using the meta-GGA density functional r2SCAN-3c implemented in ORCA47. This structure can be used for studying the effect of adding different amounts of nitro groups to the tetramer. The obtained optimized tetramer structures can be found in Fig. S5. They were labelled after the amount of exposed hydrogen atoms in either the HQ/SQ moieties or attached to the nitrogen atoms of the porphyrin-like central structure, ranging from “4H” to “12H”. Additionally, the structure found to be the most stable according to Kaxiras et al.21, labelled “HMIM” in their work (also displayed in Fig. S5), was analyzed with the above mentioned approach and tested for stability against the other tetramer tautomers. Their tetramer includes two SQ, one HQ, and one IQ moiety. The results of the energetic comparison can be seen in Fig. 4a, b.

Fig. 4: Computational determination of stable melanin structures.
figure 4

Structure of the most stable Mel-Syn tetramer determined with CREST after re-optimization with the r2SCAN-3c functional (a). Shown in orange are the distinctive features of this tetramer. (i) shows that all monomers are in the SQ form, (ii) shows the formal double bond between monomer units coupled with a protonated nitrogen atom in the center. b, c show the formation energies of different Mel-Syn tetramers and the band gap energies of Mel-NO2 tetramers with different degrees of nitration compared to the most stable Mel-Syn tetramer, the 6H structure.

The most stable structure, labelled 6H shown in Fig. 4a has a formation energy that is 0.81 eV lower than the HMIM structure and is thus most likely to be present at room temperature (RT). The orange circles show the key properties of the new structure; all monomer units are in SQ state (i) and two of the monomer units are predicted to be connected with a double bond (ii). Formally, the structure is equivalent to HMIM and shares the same sum formula, where the sum of one HQ, one IQ, and two SQ units leads to the same total ratio of hydroxy to keto moieties as in the four SQ structure of 6H.

To further justify the accuracy of these stability calculations when comparing them to bulk melanin films, they need to be referenced against experimental values. Here, we use the electronic structure as a measure to correlate tetramers and films of melanin. Our prior experiments using Mel-Syn films, being explored with UPS36, showed a HOMO energy range that matches other literature investigations of melanin building blocks42, and is in a similar range as the HOMO energy found for Mel-NO2 in this work. The work by Niyonkuru et al. reported the HOMO for DHI and DHICA films at (1.12 ± 0.05) eV and (1.16 ± 0.05) eV, respectively. The same literature gave experimental HOMO-LUMO band gap (ΔE) values that can be compared to those calculated using MOLGW, a Gaussian-orbital-based post-DFT code for molecular calculations48. Different band gap values based on the building block type and solvent-processing were reported (1.7 ± 0.3) eV for DHI, (2.5 ± 0.5) eV for DHICA (both solution-processed), and (2.1 ± 0.5) eV/(2.8 ± 0.5) eV for ammonia-induced solid-state polymerized DHI/DHICA42. In our theoretical studies, we calculated quasiparticle energies of the HOMO and LUMO with eigenvalue-self-consistent GW theory (ev-GnW0). The results are displayed in Table 2. Considering the uncertainty ranges, the calculated HOMO-LUMO gap energy of the 6H structure (2.78 eV) matches the range of the experimental findings of thin-films, confirming it can be used as a base for studies of the nitration of the structure.

Table 2 Calculated HOMO-LUMO gaps (ΔE) of different possible Mel-Syn tetramers compared to experimental band gap data of DHICA from the literature (ΔE(39))

Next, different degrees of nitration for the four available C1-positions in the 6H structure (ranging from being nitrated once, “1 N”, to four nitro groups, “4 N”) were analyzed in stability and band gap size, once again using the approach of CREST tautomerization and ORCA/r2SCAN-3c re-optimization. The band gap energies obtained from MOLGW/evGnW0 can be seen in Fig. 4c and the formation energies are listed in Table 3.

Table 3 Calculated formation energies Ef of Mel-NO2 with different degrees of nitration compared to the formation energy of the 6H Mel-Syn tetramer: The formation energies were calculated via a simple condensation reaction, the substitutions were then considered post-polymerization

The formation energies indicate that “2Ntrans” (−0.050 eV) and 1 N (−0.059 eV) are the most stable structures, although the energy differences are small. The “trans” indicates that the two nitrated monomer units are located diagonally from each other with one unit in between, compared to “2Ncis”, where they are located directly next to each other. The XPS data for Mel-NO2 films indicates that 54% of all monomers are nitrated. This matches with the calculated high stability of the 2Ntrans structure, which is also more likely to occur than the cis equivalent, as the functional groups prefer to be further away from each other due to steric and energetic benefits, resulting in a lower formation energy for the 2Ntrans structure. Due to the unclearly defined molar mass of melanin induced by its structural inhomogeneity and structural disorder, conventional yield calculations for the functionalization were not applicable6. XPS was used to estimate the functionalization for Mel-NO2 as discussed above through investigating the N 1 s signal, while Mel-Boc could not be quantified in such a manner due to spectral overlap, though its presence was confirmed by XPS, NMR, and FTIR.

Looking at the calculated band gap bar graphs for Mel-NO2 in Fig. 4c, only the 4 N and 2Ntrans structures result in significant band gap reductions ranging from 50 meV to 100 meV when compared to the 6H structure. This indicates another reason for potentially favorable electrochemical performance of Mel-NO2. The reduced band gap decreases the constrain of electron movement, resulting in increased conductivity and redox center availability49. If the experimental conditions were modified to achieve a higher degree of nitration than 54%, the even larger band gap reduction observed for the 4 N structure indicates the possibility for even higher ease of redox reactions. However, having a higher formation energy indicates it would be more difficult to access this structure experimentally. Additionally, the improved solubility of Mel-NO2 compared to Mel-Syn leads to better processibility in polar solvents and thus more stable films.

As a summary of the computational results for the 2Ntrans structure, the electronic structure of melanin was clearly affected by the nitro-group functionalization. This must be considered closely when continuing the investigation of the application in electrochemical coin cells. Additionally, in the process of obtaining information about Mel-NO2, a new most stable Mel-Syn tetramer structure, labelled as 6H, was identified.

Electrochemical applications

To test the electrochemical performance of Mel-Boc in devices, aqueous electrolyte Zn coin cells were tested with a polished Zn counter electrode. Mel-Boc mixed with carbon black and polyvinylidene fluoride (PVDF) (6:3:1 ratio) is used as the working electrode. For the active mass, only the melanin was considered and not the redox-inactive composites. The separator was made of glass fiber filter paper soaked in the electrolyte. As a comparison, identical devices using Mel-Syn as the active working electrode material were tested as well. The results using an aqueous 2 M ZnSO4 electrolyte can be seen in Fig. 5.

Fig. 5: Aqueous electrolyte systems electrochemical data.
figure 5

CV (a, e), EIS (b, f), GCD (c, g), specific capacity calculation (d), and long GCD cycling (h) data of Zn coin cells using melanin (derivative) electrodes and an aqueous (2 M ZnSO4) electrolyte. The top row shows the data of Mel-Syn and the bottom row contains the Mel-Boc data.

2 M ZnSO4 has proven to be an ideal electrolyte for aqueous zinc systems50. Selecting a low Zn2+ concentration or removing Zn ions while using a Zn anode has led to low capacity and poor rate performance in previous works50. The Zn plating/stripping ability is greatly improved with more Zn ions inside the solution, as the energy barrier for Zn desolvation is lowered by reducing the availability of water molecules for completing the Zn solvation structure. This leads to the improved performance and reduced chance of dendrite formation when using ZnSO4 in high concentration50.

The CV results shown in Fig. 5a, e depict the selected potential window of 0.3 V to 1.7 V. This window was optimized to avoid decomposition reactions. The overall non-polarizing potential window is increased by 0.3 V when compared to the system using 0.1 M ZnSO4 and 0.9 M Na2SO4 in deionized water (MQ) as electrolyte. This increase can be attributed to the change in hydration of the system due to the high amount of present Zn species, shifting the water splitting reaction to higher/lower oxidation/reduction potentials51. Both CV curves show the melanin-typical shape found in literature4. This shape shows a broad oxidation peak from 1 V to 1.5 V, and a broad reduction wave-peak from 1.4 V to 0.8 V. This is the linear combination of the oxidization/reduction peaks of all possible monomer variations (HQ, SQ, IQ, QI, each with/without carboxylic acid units) located very close to each other4. The presence of Boc-groups did not change the general shape of this curve, confirming the redox inactivity of the Boc-groups. As observed in previous work, the increase in scan rate leads to a shift of the oxidation peak to more positive potential, indicating a slow charge transport from the melanin moieties36.

For Fig. 5b, f, using 2 M ZnSO4 in a two electrode set-up, both Mel-Syn and Mel-Boc show diffusion-controlled impedance data. A magnification of the high frequency region for both materials can be seen in Fig. S6, which confirms only the presence of an onset electrochemical semi-circle corresponding to ohmic losses (e.g., from the separator) for both types of electrodes. Diffusion plays an important role in the systems due to the size and mobility of the ions present during the charge transfer processes.

Finally, Fig. 5c, d, g, h show the GCD results and the resulting specific discharge capacity CD at different current densities, as well as the Coulombic efficiency η and capacity retention over 5000 cycles. The absolute CD of Mel-Boc is lower than for Mel-Syn by a factor of 10, while the cycling stability and rate capability follow a similar trend for both materials. As commonly observed for organic electrodes in aqueous electrolytes52, there is an activation of the electrode surface through initial surface hydration until a stable η and maximum CD is achieved for both Mel-Syn and Mel-Boc.

The differences in mixed conductivity and electrochemical performance can be explained by the reduced metal ion chelation capabilities of Mel-Boc, as this has been found to be important for melanin energy storage systems analyzed using DFT26. The sterically bulky Boc-groups block the centers of the melanin porphyrin-like structures and reduce the amount of overall available chelating moieties. Thus, the likelihood of metal ions taking part in electrolyte decomposition reactions is increased, and the size of the stable potential window is decreased. The Boc-groups could also increase the spacing between proposed π-π stacked tetrameric units, resulting in larger structural gaps through anisotropic particle growth/agglomeration and possibly explaining the observed clustered surface structure and larger average particle size observed for Mel-Boc using SEM53. This results in a decreased surface area and thus less accessible redox centers. Together with the overall mass increase of Mel-Boc due to the high molar mass of the Boc-groups, this could possibly explain the differences in electrochemical performances of Mel-Syn and Mel-Boc in aqueous electrolytes.

While selecting an electrolyte to investigate the Mel-Boc system, the SE composition of 0.1 M ZnSO4 with 0.9 M Na2SO4 was considered as well and tested in a three-electrode set-up. The results can be seen in Fig. S7. The benefit of this SE has been shown in literature to be the high mobility of the sodium ions, leading to improved conductivity54. Melanin systems using composites with electronic conductivity like carbon black and the melanin itself with mostly proton/ionic conductivity, can be classified with the term “mixed conductivity”3. While using the SE, qualitatively, the electrochemical impedance spectroscopy (EIS) data shows to be not strongly dominated by Warburg diffusion for both Mel-Syn and Mel-Boc (Fig. S7b, e). However, the CV (Fig. S7a, d) shows a small potential window due to water decomposition around −0.3 V51. This, together with the low specific capacity values at higher current densities (3.0 mA h g−1 (Syn) and 0.94 mA h g−1 (Boc) at 0.4 A g−1 compared to 17.3 mA h g−1 (Syn) and 0.97 mA h g−1 (Boc) when using 2 M ZnSO4), which is essential for organic energy storage systems, led to 2 M ZnSO4 being investigated in more detail instead.

The importance of metal ion chelation for melanin-based energy storage systems was underlined by the examination of these systems. With this in mind, the next derivative that introduces an EWG was selected as to not negatively affect the chelation capabilities or intermolecular spacing, while also acting as an interesting modification to study the influence of highly polar groups on the electrochemical performance of melanin in energy storage applications.

Ionic liquid electrolyte systems

The empirical studies using SEM and UPS followed up by the computational investigations have indicated promising results for using Mel-NO2 as a melanin modification in energy storage applications. Thus, as Mel-NO2 was found to be water-soluble, it was tested electrochemically using a RTIL electrolyte (Zn(TFSI)2 in EMIM TFSI) that was used in previous work comparing Mel-Syn to melanin from the black soldier fly (Hermetia illucens)36. Mel-Boc and Mel-NO2 were tested in different electrolytes due to their distinct solubility. The water-insoluble Mel-Boc was compatible with aqueous zinc-based electrolytes, while the water-soluble Mel-NO2 required a RTIL to ensure stable electrochemical measurements. While this approach enables the reliable evaluation of each material, comparisons across systems should be interpreted within the context of their respective electrolyte environments. The built devices were of identical structure to the aqueous electrolyte devices and can be seen schematically in Fig. 6f. This figure also displays the results of the CV (a), EIS (b), GCD (c, d), and long cycling GCD (e) investigations for Mel-NO2.

Fig. 6: RTIL electrolyte systems electrochemical data.
figure 6

CV (a), EIS (b), GCD (c), specific capacity calculation (d), and long cycling (e) data of Zn coin cells using Mel-NO2 as working electrode and RTIL (Zn(TFSI)2 in EMIM TFSI) electrolyte. In the bottom right (f), a schematic of the Zn coin cell assembly can be seen.

The CV of Mel-NO2 once again shows the same melanin-like curve shape and shift to more positive potentials of the oxidation peak at higher scan rate, just like Mel-Boc in aqueous and Mel-Syn in aqueous and RTIL systems. The absence of additional signals confirms that the added nitro groups are redox inactive in the chosen potential window.

The potential window size was optimized for the use of the stable low vapor-pressure RTIL electrolyte and is larger than that shown for the aqueous systems, as there is no direct issue of water splitting.

Comparing the EIS data to previous findings using the same electrolyte36, it can be seen that qualitatively, the electrochemical semi-circle is more pronounced for Mel-NO2, and the Warburg diffusion seems to have less influence on the overall system. Literature findings have shown that EWGs can cause a decrease in band gap size for conjugated aromatic systems, thus making the conducting band easier accessible and increasing the overall system conductivity and strengthening the influence of the electrochemical semicircle on the impedance spectrum27. This would correspond to the nitro groups with a negative mesomeric (-M) and negative inductive (-I) effect on the melanin IQ system. The evGnW0 results have shown that such a shift in band gap size would correspond to structures with a degree of nitration similar to that observed in our XP spectra.

Lastly, the GCD results show a similar specific discharge capacity at 0.1 A g−1 for Mel-NO2 (22.0 mA h g−1) when compared to Mel-Syn (31.5 mA h g−1). At higher currents, Mel-NO2 has a higher capacity than Mel-Syn in RTIL electrolyte systems. This starts at 0.2 A g−1 (19.5 mA h g−1 for Mel-NO2 vs. 19.2 mA h g−1 for Mel-Syn36) and the specific discharge capacity trend for Mel-NO2 continues to be more stable around 16.0 mA h g−1 even for higher currents, as can be seen in Fig. 6d. The Coulombic efficiency stability and rate capability for Mel-NO2 are also shown to be excellent in Zn coin cell applications over 5000 cycles (e). The longer charge times observed for both Mel-Boc and Mel-NO2, particularly at low current densities, can be attributed to slower ion diffusion and structural accommodation during ion insertion. In contrast, the discharge process proceeds more rapidly due to easier ion release. The presence of the different functional groups further influences these kinetics, contributing to the observed asymmetry.

In contrast to the aqueous systems, which exhibit irregular points during the initial 0–1000 cycles due to an activation phase involving surface hydration and structural adjustment, the RTIL system displays stable and consistent electrochemical behavior from the beginning. The absence of such fluctuations highlights the inherent stability of the RTIL environment and underscores a key difference in cycling behavior between the two electrolyte systems. To summarize, the clearly improved electrochemical performance of Mel-NO2 in RTIL energy storage systems can be attributed to multiple reasons. First, as discussed, theoretical calculations showed a decrease in band gap size correlating with good mixed conductivity in the electrochemical investigations.

The small particle size measured with SEM ((240 ± 13) nm) provides high surface area. At the same time, the solubility in polar solvents like NMP was increased, resulting in improved film processibility and thus redox center accessibility on the film surface. The high cycling stability of Mel-NO2 without the requirement of an activation process that occurs for Mel-Boc and Mel-Syn in aqueous electrolyte systems is the result of using a low vapor-pressure RTIL electrolyte mixture with high electrochemical stability15, showcasing the excellent energy storage capabilities of the system.

The maximum specific discharge capacity of 22.0 mA h g−1 (~ 38 F g−1) gives the energy density ε = 22000 mW h kg−1. Composite hydrogel materials using synthetic melanin, cellulose nanofibrils, and polypyrrole were shown to have an energy density of 5387 mW h kg−1 3. Solid state devices using soluble melanin derivatives showed power densities of around 1 W kg−1, but lower energy densities of around 2 × 10−3 mW h kg−1 55. Finally, self-synthesized melanin on a carbon paper substrate using aqueous electrolytes showed an energy density of around ε = 12.2 mW h kg−1 at high power densities of around 3800 W kg−1 56. As melanin-based energy storage devices are still improving rapidly, the range of discussed literature values is wide. Still, Mel-NO2 was successfully found to be on the upper end of these devices (including Mel-Boc at 0.97 mA h g−1; ~3 F g−1) regarding energy density, proving further optimizations of devices using this material to be promising. Comparing the redox behavior of Mel-Boc and Mel-NO2 to other doped carbon devices in the literature57, it can be seen that the performance of Mel-NO2 fits into the lower end of oxidized graphene porous materials-based supercapacitors (~ 30 F g−1). N-doped bio-based carbon materials show energy storage capabilities of up to 32 F g−1 due to improved electronic properties of nearby carbon atoms and impacted adsorption behaviors of intermediary substances58. Looking at the redox behavior and device optimization methods in these works, strategies like electrode surface functional group tuning through thermal modification and porosity optimization could be fruitful approaches to improve the melanin device performance in future work. Furthermore, the sustainability aspect of the melanin materials could be explored further and the synthesis approaches could be modified accordingly, as one of the key advantages of the melanin derivative electrodes over highly modified graphene electrodes remains the possibility to obtain the melanin precursor sustainably36.

Conclusions

In this work, chemically modified organic melanin working electrodes were used to tailor their properties in energy storage applications using aqueous and RTIL electrolytes. NMR, FTIR, and XPS confirm the syntheses of two different melanin derivatives. The first derivative, Mel-Boc, contains Boc-groups that are connected to the indole nitrogen atoms of the melanin monomer units, resulting in bulky spacer-groups and reduced metal ion chelation capabilities in the porphyrin-like tetramer centers. The second derivative, Mel-NO2, showed a nitration of 54% of the available monomer units, indicating two monomers of each tetramer being nitrated. The strong polar EWGs improve the electrochemical properties of melanin systems.

Using different modern computational approaches, the previously proposed tetramer structure of unmodified melanin was re-evaluated and a new most stable structure (Ef = −1.77 eV) with four SQ units per tetramer, known as “6H”, was discovered, with calculated HOMO-LUMO gap values (2.78 eV) matching UPS results of melanin and melanin-like films. This was used as a base to confirm the stability of the roughly 50% nitrated Mel-NO2 tautomer and a decrease in band gap of 50 meV to 100 meV depending on the degree of nitration.

Coupled with smaller particle size and consistent film processibility, Mel-NO2 showed improved redox center accessibility. Using aqueous electrolyte solutions, Mel-Boc was compared to Mel-Syn, with similar cycling stability but reduced specific discharge capacity (0.97 mA h g−1 at 0.4 A g−1; 0.2 µA h cm−2 at 0.20 mg cm−2 active mass loading per area). This may be due to the decreased redox-active surface area through larger particle size and reduced metal ion chelation capabilities through the presence of bulky Boc-groups. For all melanin-based systems, the reported areal capacity is lower than the active mass specific capacity, as the electrode area includes not only the redox-active material, but also inactive components like binders and conductive additives. While the areal capacity is provided for ease of comparison with other systems, it needs to be noted that it is less representative about the melanin electrochemistry than the active mass specific capacity. Using RTIL electrolyte solutions, Mel-NO2 was shown to be a promising material with high Coulombic efficiency and rate capability over 5000 cycles, and increased specific discharge capacity values for current densities starting at 0.2 A g−1 (19.5 mA h g−1; 3.3 µA h cm−2 at 0.17 mg cm−2 active mass loading per area) when compared to Mel-Syn.

Previous assumptions of metal ion chelation being important for melanin system charge transport could be confirmed with Mel-Boc, while Mel-NO2 showcased that looking into other EWGs for melanin optimization may be of interest for future work covering organic electronic materials in energy storage or sensing. The newly determined most stable 6H tetramer structure can be used as a basis to determine the feasibility of potential synthesis endeavors.

Methods

Materials and reagents

Analytical grade was given for all used chemicals, and they were used without further purification. Synthetic eumelanin (M8631), acetonitrile (anhydrous, 99.8%), dimethylsulfoxide (DMSO, ≥99.9%), PVDF, NMP, zinc sulfate (ZnSO4, anhydrous, ≥99.0%), di-tert-butyl dicarbonate (Boc2O, 99%), iodine (I2, ≥99.8%, solid), sodium thiosulfate (Na2S2O3, 99%), sodium bicarbonate (NaHCO3, ≥99.7%), sodium sulfate (Na2SO4, anhydrous, ≥99.0%), nitric acid (HNO3, 70%), potassium carbonate (K2CO3, ≥99.0%), and EMIM TFSI (≥ 98% (HPLC grade)) were purchased from Sigma-Aldrich, USA. Zn(TFSI)2 (> 98.0%) was obtained from Tokyo Chemical Industry, Japan. Dimethylsulfoxide-d6 (DMSO-d6, 99.9%) was purchased from Cambridge Isotope Laboratories, Inc., USA. MQ water was filtered with a Milli-Q A10 Biocel water purification system with Millipak Express 20 filters provided by Merck Millipore, USA. Carbon Black (Vulcan XC 72 R) was obtained from Vulcan Materials, USA. Conductive carbon paper was obtained from Toray, Japan. The glass fiber filter papers (GF-3, 0.6 microns) for coin cell assembly were bought from Whatman, UK. The remaining components for coin cell assembly were purchased from Smart Memories Pte. Ltd., Singapore, and the dimensions of each part were discussed in our previous work36. The area of the circular coin cell electrodes was determined as ~1.5 cm2. For XPS/UPS sample preparation, microscopy glass slides (VWR International BVBA, Belgium) were first washed with water and ethanol (Fisher Chemicals, Germany). Then, they were dried with a heat gun and 0.5 nm of chromium as adhesion promoter was deposited on the cut glass slides by electron beam evaporation in vacuum. Finally, 100 nm of Au were deposited while being monitored with a quartz crystal microbalance (MiniCoater, Tectra GmbH, Germany) to prepare the Au/glass substrates. SEM was performed on Si wafer substrates from Sigma-Aldrich, using double-sided carbon tape from Fisher Scientific, USA, to secure the samples on the SEM-stub.

Synthesis of derivatives

Mel-Boc: Synthetic eumelanin (0.20 g) was added to 5 mL of an acetonitrile/DMSO (2:1) mixture and stirred overnight at RT in a glass vial. Then, Boc2O (130 mg) and a catalytic amount of iodine (90 mg) were added to the solution, stirring it for 24 h at 50 °C. After cooling the glass vial to 0 °C for 1 h using an ice bath and letting the reaction mixture sit for another 24 h at RT, a black powder precipitate was formed. The powder was vacuum filtrated and washed multiple times with acetonitrile, 5% aqueous sodium thiosulfate (10 mL), and with 5% aqueous sodium bicarbonate (10 mL). Finally, the powder was dried under reduced pressure overnight, and Mel-Boc was obtained as an insoluble black powder.

Mel-NO2: Synthetic eumelanin (0.20 g) was added to 5 mL of an analytic acetonitrile/DMSO (2:1) mixture and stirred overnight at RT in a glass vial. This mixture was added dropwise at 0 °C to 3 mL of nitric acid (68%–70%) under constant stirring. After complete addition, the reaction container was sealed, and the reaction mixture was stirred for 24 h at 50 °C. Then, the reaction mixture was neutralized to pH = 7 by addition of an aqueous 0.5 M potassium carbonate solution at 0 °C. The organic phase was dried over Na2SO4 and the solvent was removed under reduced pressure, yielding Mel-NO2 as an orange-brown powder.

Spectroscopy and microscopy

To confirm the derivative syntheses, liquid-phase nuclear magnetic resonance (NMR) spectroscopy was performed on solutions of Mel-Boc and Mel-NO2 powders dissolved in deuterated DMSO (DMSO-d6) for reducing the DMSO signal in the subsequent NMR measurements. To ensure the samples dissolving, the solutions were stirred at 650 rpm for 1 h at 65 °C before filling them inside the test tubes. It was ensured that no solid phase remained inside the NMR tubes. Unless stated otherwise, all 1H NMR spectra (400 MHz) of the samples were recorded on a Bruker AVANCE NEO 400 MHz (9.4 T) NMR spectrometer from Bruker, USA. The following parameters were used during measurements: Time domain of 65536 data points, 148 scans, 8 dummy scans, 6250 Hz spectral width, 5.24 s acquisition time, and 80.00 s dwell time. The data was analyzed using the Bruker TopSpin 4.1.3 software, including phase- and baseline-correction, peak picking, and further processing.

Mel-Boc and Mel-NO2 powders were analyzed with FTIR spectroscopy (FTIR Perkin Elmer Frontier, USA) in the attenuated total reflection (ATR) mode. The background was subtracted, and graphs were averaged from 16 scans.

XPS and UPS were performed using an ESCALAB 250 Xi system from Thermo Fisher, UK, using a 0.02 eV step size, 50 ms dwell time, 10 eV pass energy, and monochromatized Al Kα radiation ( = 1486.6 eV). The scan numbers were as follows: 10 for survey spectra (100 eV pass energy), 5 for O 1s, 10 for S 2p, 5 for C 1s, and 10 for N 1s. Using the fitting software Avantage (v.5.9931, Thermo Fisher), a smart background fit (combination of polynomial and Shirley background) and a Gauss/Lorentz function convolution was applied for each spectral component. The UP spectra were taken with a He(I) source (21.2 eV) with 1 eV pass energy, 10 scans, 0.02 eV step size, 50 ms dwell time, and a range from 13.5 eV to −5 eV. Used samples were Mel-Boc and Mel-NO2 films drop-cast on Au/glass. Powders of the materials were first dissolved in NMP. Then, they were sonicated and stirred over night before drop-casting. Finally, the samples were dried in the oven at 1 atm at 60 °C for 24 h.

SEM was performed on a FE-SEM 7600 F from JEOL, USA (5.0 kV beam energy, magnifications between 5 k and 30 k). Aqueous dispersions of the sample powders were sonicated and drop-cast on Si wafers. After oven-drying for a minimum of 12 h, the wafers were grounded to the SEM-stub with double-sided conductive carbon tapes and covered with a few nm of Pt particles to decrease charging.

Computational methods

To analyze the stability of different Mel-Syn and Mel-NO2 tetramers and calculate the HOMO-LUMO gap size of these structures, various computational methods were used. For finding the most stable tautomers, CREST version 2.11 (using ×TB version 6.4.0) was used. CREST is a semi-empirical CREST46. With this, the tautomerize function was applied with cinp (usage of constraints and oxygen atom positions as hold fixes)59. This was necessary for preventing the generation of fragmented structures, proving to be instable after post-optimization. DFT calculations for structure optimization were performed using ORCA version 5.0.2, specifically using geometrical counterpoise correction (gCP), the D4 dispersion correction, def2-mTZVPP basis sets, and def2-mTZVPP/J auxiliary basis sets60,61,62,63. Used parameters were opt for geometry optimization, TightSCF (threshold for SCF converging), and the used functional was r2SCAN-3c63. Finally, using MolGW version 2.F, the HOMO-LUMO energy gaps were calculated for comparison to UPS data48. The following parameters were used: BHLYP functional for the wavefunction generation, followed by an eigenvalue-self-consistent GnW0 calculation (with n being the number of iteration steps; usually n = 6 was sufficient to obtain self-consistency) for the quasiparticle energy calculation, employing cc-pVDZ basis sets, cc-pVDZ-RI auxiliary basis sets, and the frozen core approximation. For displaying atom coordinates in graphics, Avogadro was used64.

Device fabrication

The eumelanin derivatives were tested as working electrode materials in Zn coin cells and compared to synthetic eumelanin. The structure of these devices can be seen in Fig. 6. The devices were assembled in the lab air environment under atmospheric pressure. Slurries of the components were manufactured by using 12 mg of the respective melanin, 6 mg carbon black, and 2 mg PVDF (6:3:1 ratio) in 1 mL NMP. The remaining preparation of the coin cell devices was identical to the process described in, with only the Mel-NO2 devices also using an ionic liquid electrolyte consisting of 0.31 g Zn(TFSI)2 in 1 mL EMIM TFSI. Mel-Boc and Mel-Syn coin cells were instead tested with an aqueous 2 M ZnSO4 electrolyte. For testing Mel-Boc and Mel-Syn in a three electrode-system with Zn as a counter electrode and Ag/AgCl as a reference electrode, a 0.1 M ZnSO4/0.9 M Na2SO4 in MQ SE was used and 30 μL of the slurry was drop-cast in a 1 cm × 1 cm square onto a 1 cm × 2 cm piece of conductive carbon paper, which was oven-dried at 60 °C for 24 h and clamped directly on the non-slurry covered part with an alligator clip before electrochemical testing. It was ensured that during testing, only the slurry-covered part of the samples was submerged in the electrolyte.

Electrochemical measurements

Each electrochemical test was performed at RT for a minimum of three devices, and the results shown are typical data of individual devices.

CV data was measured using a PGSTAT-128N Autolab from Metrohm, Netherlands and a 760E electrochemical workstation from CH Instruments, USA. Each measurement was started/stopped at the system starting OCP. The minimum number of cycles was 5 at scan rates between 10 mV s−1 and 50 mV s−1 and a step size of 0.001 V. To confirm complete activation of the samples, each CV curve shown is the final cycle of each measurement. Mel-NO2 devices were cycled between 0 V and 2 V vs. Zn/Zn2+ in a two-electrode-system, while Mel-Boc and Mel-Syn devices were cycled between 0.3 V and 1.7 V. In the three electrode-system, Mel-Boc and Mel-Syn samples were cycled between −0.5 V and 0.8 V vs. Ag/AgCl. All ranges were optimized to either prevent or visualize the limits caused by electrolyte decomposition reactions. The CV data was visually processed using the OriginPro 2022 software.

EIS (potentiostatic) was also performed with the PGSTAT-128N Autolab with FRA-32M module from Metrohm. After confirming the linearity of the CV curve around that potential for every sample, 0 V was chosen as a parameter for the measurements. The frequency range varied between measurements depending on the samples at a constant 10 frequencies per decade, internal input connection, sine wave type, and VRMS = 0.01. Using the Nova 2.1.5 Software, impedance date was qualitatively fit using a [R([RctW]Q)] equivalent circuit as used for similar systems using melanin as redox-active material together with conductive, redox inactive additives, as described in refs. 3, 36. While R represents the electrolyte layer and Q the constant phase element double layer, the RctW-unit stands in for coupled diffusion and charge transfer resistance.

GCD tests for the coin cell devices were done with the 760E electrochemical workstation. GCD measurements for the three-electrode systems were performed using the PGSTAT-128N Autolab. The step size for all measurements was 1 ms for at least 5 cycles of each sample, while the last cycle was reported similarly to the CV experiments. Mel-NO2 coin cells were measured in the potential window determined by CV, while the aqueous Mel-Boc and Mel-Syn coin cells had to be measured at a reduced potential window of 0.3 V to 1.4 V, and the three-electrode systems from −0.2 V to 0.8 V due to aqueous electrolyte decomposition at high potentials when using a low current. GCD measurements were normalized by the active material mass and thus only use the melanin (derivative) mass and not that of the electrochemically inactive additives, carbon black and PVDF. Current densities between 0.1 A g−1 and 2.0 A g−1 were used during the tests.

The long cycling GCD of coin cells over 5000 cycles was performed using a NEWARE battery testing system from Neware Global, Hong Kong, together with the control software BTSDAV7.1.19. Mel-NO2 was cycled using the ionic liquid electrolyte (Zn(TFSI)2 in EMIM TFSI) at a current density of 0.6 A g−1 between 0 V and 2 V. Mel-Boc was cycled using the aqueous 2 M ZnSO4 electrolyte at a current density of 0.6 A g−1 between 0.3 V and 1.4 V.