Role of semiconductivity and ion transport in the electrical conduction of melanin

Abstract

Melanins are pigmentary macromolecules found throughout the biosphere that, in the 1970s, were discovered to conduct electricity and display bistable switching. Since then, it has been widely believed that melanins are naturally occurring amorphous organic semiconductors. Here, we report electrical conductivity, muon spin relaxation, and electron paramagnetic resonance measurements of melanin as the environmental humidity is varied. We show that hydration of melanin shifts the comproportionation equilibrium so as to dope electrons and protons into the system. This equilibrium defines the relative proportions of hydroxyquinone, semiquinone, and quinone species in the macromolecule. As such, the mechanism explains why melanin at neutral pH only conducts when “wet” and suggests that both carriers play a role in the conductivity. Understanding that melanin is an electronic-ionic hybrid conductor rather than an amorphous organic semiconductor opens exciting possibilities for bioelectronic applications such as ion-to-electron transduction given its biocompatibility.

The melanins are responsible for multiple critical functions in humans such as photoprotection and free radical scavenging (1). These molecules are also found in the substantia nigra of the human brain stem where their exact biological role is unknown; however, it has been speculated that neuromelanin may be involved in neural transmission (2). Melanin phototoxicity is also implicated in deadly melanoma skin cancer (3, 4). Despite decades of intense studies across biology, chemistry, and physics, the full details of the structure and functions of the melanins are still not clearly understood. Eumelanin, the major component in human skin pigment is viewed as the archetypal “true” melanin (here we adopt the standard nomenclature in which the terms “eumelanin” and “melanin” are used interchangeably). Eumelanin is composed of aggregated oligomeric and polymeric species based upon the indolic monomers 5,6-dihydroxyindole (DHI) and 5,6-dihydroxyindole-2-carboxylic acid (DHICA) and their various redox forms (1). These monomers are randomly cross-linked to form planar sheets, stacked via aromatic π-interactions and with varying conjugation length and character (5, 6, 7). For many years, melanins were considered as extended linear homo- or heteropolymers with a conjugated backbone (8). However, this view no longer has any credence and the disordered 2D protomolecular sheet model is now widely accepted. This sheet model has quite profound implications for the treatment of melanins within a conventional polymer framework, and many of the standard methods and theories for probing structure property relationships in macromolecules are inadequate or simply do not apply.

Two properties of melanin have particularly intrigued physicists and chemists for decades: (i) Melanins are electrical conductors showing photoconductivity in the solid state; and (ii) melanins are black with broad featureless optical absorption and near unity nonradiative conversion of absorbed photon energy. The latter property is clearly linked to the photoprotective function. However, the biological need for the former is somewhat of a mystery, but may be linked to neurotransmission. More recently, other potentially useful properties of melanins and melanin-like materials have also been reported, such as multifunctional adhesives (9, 10), structural agents in the jaws of marine worms (11) and in cell walls of fungi (12) as well as protection against ionizing radiation (13).

For forty years, the accepted paradigm to explain electrical conductivity of melanins has been that they are amorphous organic semiconductors (14, 15). Important experimental evidence to support this model came from McGinness, et al. who observed electrical bistable switching between two resistive states (15), which, at that time, had only been seen in inorganic amorphous semiconductors. The model of melanin as an amorphous semiconductor also provides a possible explanation for the observed broad optical absorbance (14).

A critical point to note concerning McGinness, et al.’s observations is that switching was only seen when the melanin was hydrated (15). The authors argued that this observation could be rationalized in terms of the modified dielectric theory (16, 17) whereby the absorbed water lowers the activation energy for hopping transport. Following this seminal work, the amorphous semiconductor model became the accepted framework through which all electronic properties of melanin were interpreted (18–24). However, a small collection of electrical observations including electrochemical evolution of hydrogen during calorimetry (17), have cast doubt on this hypothesis. Also, it has been shown (25– 27) that chemical disorder gives a natural explanation of the broad band optical absorption observed in eumelanin, particularly the absence of an optical gap. Finally, it is worthy of note that melanin chemistry is largely defined by its many ionizable moieties—carboxylic acids, amines and catechols in various states of oxidation within the macromolecular structure (1). One may therefore anticipate that melanin (like many other functional biomacromolecules) will exhibit generic polyelectrolyte behavior in which its weakly acidic nature plays an important role. This concept is yet to be explored within the amorphous semiconductor model, but certainly presents an alternative framework through which to interpret melanin structure-property relationships.

Given these inconsistencies and the clear role that hydration plays, we were motivated to undertake a carefully controlled study of the effect of humidity on the electrical properties of melanin and in so doing, clarify the underlying charge transport mechanism.

Results and Discussion

Fig. 1A shows the measured electrical conductivity of a solid pellet of melanin made by compressing melanin powder. One observes a strong hydration dependence of the conductivity (approximately 2 orders of magnitude dry to wet, note the logarithmic ordinate scale) that increases rapidly once the water content is greater than approximately 12 % by weight.

Fig. 1.

(A) Conductivity of melanin as a function of water content. The measured humidity was converted into weight percent gained using a previously reported adsorption isotherm (28). The line shows the best fit to the modified dielectric theory (16, 17) (Eq. 1 and Eq. S1). (B) The paramagnetic and diamagnetic μSR relaxation rates for melanin pellets as a function of water content. The muon hopping rate, ν, was found to be constant at 0.28 ± 0.03 μs^-1 (SE). These results shows that proton mobility does not change in the same way as conductivity, but carrier concentration does.

The amorphous semiconductor hypothesis predicts dramatic changes in conductivity, σ, as the humidity is varied because the local dielectric constant changes as water is absorbed into a semiconductor. The modified dielectric theory (16, 17) predicts that

[1]

where σ₀ is the dry electronic conductivity in the infinite temperature limit, E_D is the band gap of the material in the dry state, R is the molar gas constant, T is the temperature, r is the screening length, κ is the dielectric constant in the dry state and κ^′ is the dielectric constant at the relevant hydration. The dependence of κ^′ on hydration is given by the Lorentz–Lorenz relation (SI Text, Eq. S1).

In Fig. 1A we also plot the predicted variation of the electronic conductivity with water content from the amorphous semiconductor theory Eq. 1. The theory predicts that the conductivity will increase subexponentially, but the experimental data shows a superexponential behavior. Furthermore, the best fit to the data gives unphysical parameters (see SI Text).

We now examine other possible models of charge transport in melanin because this data is qualitatively and quantitatively inconsistent with the modified dielectric-amorphous semiconductor theory. An increase in conductivity as a function of hydration must arise from either an increase in the carrier density or an increase in the mobility. Further, there are two possible charge carriers: electrons (which we will not distinguish from holes at this stage) and ions, specifically protons.

We used muon spin relaxation (μSR) spectroscopy to probe the possible role of proton mobility in the hydration dependent behavior as well as examine bound electronic spins (29, 30). Muons are fundamental spin-half leptons, but from a chemical perspective are better thought of as light radioactive proton isotopes (the mass of the moun is about 1/9 the proton mass). Fully spin polarized muons can be implanted into a material via a beam derived from a proton source. These implanted muons simulate the diffusion and interactions of protons. Moreover, the implanted muon spin relaxes in a manner that is precisely analogous to those of the nuclear and electron spins in nuclear magnetic resonance and electron paramagnetic resonance (EPR) experiments, respectively. Measurement of the relaxation of the spin of an implanted muon therefore probes the local microscopic environment. Experimentally, muon relaxation is monitored via the decay of muons to positrons, which are emitted preferentially in the direction parallel to their spin. The basic measurement parameter is the asymmetry in the number of positrons detected in front of and behind the sample relative to the direction of propagation of the muon beam. The time, t, and magnetic field, B, dependence of this asymmetry yields important information about the behavior of the muons within the sample. The total asymmetry, A_T, is given by:

[2]

where A_D is the asymmetry associated with the fraction of diamagnetic muons (i.e., those not strongly associated with a local unpaired electron spin), A_P is the asymmetry associated with paramagnetic muons (i.e., those that are strongly associated with a local unpaired electron), A_Bg is the background asymmetry, K(Δ,ν,B,t) is the Kubo-Toyabe relaxation function (31), ν is the muon hopping rate, Δ is the relaxation rate for diamagnetic muons dominated by the nuclear dipolar fields of surrounding protons, and λ is the relaxation rate for paramagnetic muons. Importantly, for our purposes, ν provides a proxy for proton mobility, μ_P, (ν ∝ μ_P).

We measured the asymmetry as a function of field and hydration and extracted the parameters Δ, λ, and ν. Fig. 1B shows the humidity variation of λ, which is proportional to the unpaired spin density—the similarity to the hydration dependence of the conductivity is striking. Fig. 1B also shows that Δ, which is representative of the local field experienced by free muons (and hence protons), is sensitive to hydration in the same way as λ and σ, but much more weakly. However, perhaps the most important result from this muon experiment is that the muon hopping rate, ν, was found to be constant across the hydration range [0.28 ± 0.03 μs^-1 (SE)]. Thus, the μSR data shows that proton mobility is unaffected by hydration and that hydration adds new free radicals to the system.

It is instructive to compare these results with the complimentary information available from EPR experiments, which directly probe the behavior of the electronic spins. Indeed, melanin was one of the first biological materials to be studied with this method (32), and EPR has provided important insights into the behavior of free radicals in this system (33, 34). In particular, this large body of work has shown that there are two populations of unpaired spins in melanin: “intrinsic” spins derived from carbon centered radicals remaining from the polymerisation process (8) and the “semiquinone” or “extrinsic” spin regulated by the comproportionation equilibrium (Scheme 1) (35). In solid-state samples, the intrinsic spin is predominantly observed (8, 36), whereas the extrinsic spin is predominantly seen in melanin solutions or colloidal suspensions (35, 37).

Scheme 1.

The comproportionation reaction.

We measured the intrinsic spin population as a function of hydration in a solid state EPR experiment analogous to the μSR and electrical conductivity measurements. The EPR results are shown in Fig. 2. The intrinsic spin density does not display the same hydration dependence as the electrical conductivity and the μSR relaxation rates. In fact, the spin density decreases with water content, showing the intrinsic spin concentration is not linked to the electrical conductivity. Because there are suggestions in the literature to the contrary [directly linked to the semiconductivity interpretation (23)], we further examined the intrinsic spin density in a combined photo-EPR/photoconductivity experiment. The results of this measurement at a hydration level of 7.6% by weight are shown in Fig. 3. No change is observed in the intrinsic spin population even though the sample shows clear photoconductivity. This result is replicated across the hydration range for which there is observable photoconductivity (see Figs. S1 and S2), confirming that the intrinsic spin is not linked to the electrical properties.

Fig. 2.

The EPR peak-to-peak intensity of a melanin pellet at neutral pH as a function of water content. A typical spectrum is shown in the inset (14% hydration; microwave power 0.0502 mW; frequency 9.84917 GHz). The data proves that the intrinsic spin concentration does not vary with hydration in the same way as the conductivity.

Fig. 3.

Simultaneous photo-EPR (red) and photoconductivity (blue) measurements of a melanin pellet (7.6% hydration). Light bulbs indicate when the light was switched on and off. Clearly, there is no change in the EPR signal while the photoconductivity is strongly modulated. This result shows that the intrinsic spin is not connected with electrical conduction in melanin. The background slope of the photoconductivity data is due to resistive and photo-induced heating causing a decrease in hydration. Note also the long RC time constants (approximately seconds) associated with the build-up and decay of photocurrent when the light source is switched on and off. Long RC time constants are typical of any electrical measurement on melanin.

The semiquinone spin population is governed by the comproportionation reaction (Scheme 1), which defines the equilibrium concentrations of the hydroxyquinone, quinone, and semiquinone. As one would expect from Le Châtelier’s principle this spin population increases with pH. The form of this pH dependence can be measured by a solution titration EPR experiment (see Fig. 4 and refs. 35, 37, 39). It is once again striking that the semiquinone population increases with pH in the same way that both the conductivity and muon relaxation rates vary with hydration. This observation provides support for the hypothesis that water also perturbs the comproportionation equilibrium in the solid state. Water titrates the system and increases the semiquinone free radical concentration and likewise the hydronium ion (proton) populations. Thus, the densities of both electrons and ions increase with hydration via water induced self-doping.

Fig. 4.

EPR signal of a colloidal suspension of melanin as a function of pH. We have previously reported this data (38), which shows that the extrinsic spin concentration varies in a similar manner to the electrical conductivity.

Given this hypothesis, the question arises as to whether it is possible to calculate the ion concentration as a function of hydration using a thermodynamic, chemical equilibrium approach. This calculation, however, presents a number of significant problems precluding any sensible estimate to be made from first principles—notably, an inability to determine in the solid state the activities of the reactants and products. Furthermore, such problems clearly indicate the limitation of describing melanin as a polyelectrolyte. One cannot define a “melanin bulk pK_a” and indeed, the individual ionizable moieties possess different pK_as depending upon their position within the macromolecule. A more detailed description of these issues is given in SI Text, but it is clear that the key challenge in estimating the ionic carrier density is to accurately measure the semiquinone, quinone and hydroquinone concentrations. In this regard, certain spin quantitation methods may be of use; however, the issue of an indeterminate solid-state pH remains.

One may now ask whether self-doping via the comproportionation reaction can explain the inconsistencies between observation and the amorphous semiconductor theory. Certainly, the generation of carriers as a result of a local chemical reaction at the catechol site would lead to the electrochemical evolution of hydrogen. Electrical signatures such as nonohmic blocking behavior, long resistance-capacitance (RC) time constants (see, for example, the slow build up of photoconductivity in Fig. 3), and dielectric hopping (17–22) are all characteristics of ionic systems. AC conductivity and impedance measurements are powerful tools in understanding the transport dynamics of polyelectrolyte and ionic systems and have been applied to melanin assuming an a priori standard physical model within the semiconducting framework to yield parameters such as hopping frequency (22). Our DC measurements assume no transport model and merely define the macroscopic electrical response. It would be instructive to reexamine AC and impedance measurements within the comproportionation framework to extract meaningful transport parameters.

The apparent Boltzmann temperature behavior of the conductivity (17, 20, 24) is also consistent with the new model. Finally, we return to why McGinness, et al. (15) observed bistable electrical switching. An examination of the experimental geometry (a melanin powder sandwiched between metal contacts) reveals a simple explanation: The sample was acting as a parallel plate capacitor with unblocking of contacts at high fields. This phenomenon would lead to hysteresis in resistance as a function of applied voltage and is characteristic of ionic systems with electrically blocking contacts. This effect also explains why the switching only occurred in hydrated material—no significant free carrier density exists in dry melanin. Hence, it would appear that a transport model based upon self-doping via the comproportionation equilibrium is consistent with many of the major electrical, optical and electrochemical measurements made on melanin over the past 40 y. This insight stimulates a range of new fundamental questions such as: what is the kinetic mechanism for the proton-transfer reaction—there have been suggestions of efficient excited state proton transfer mediating nonradiative relaxation in melanin (40, 41); can one manipulate the comproportionation equilibrium and hence control the “doping” using metal ion chelation or the addition of other cationic species (38, 39, 42)? These are questions for a new era of melanin bioelectronic materials.

Conclusion

In summary, we have demonstrated that the conventional view of melanin as an amorphous semiconductor is incorrect. Upon the absorption of water, free carriers are produced from the comproportionation reaction, which drives the production of extrinsic free radicals (electrons) and hydronium ions (protons). This chemical self-doping leads to hybrid ionic-electronic behavior and explains the exotic electrical properties of melanin. Given that melanin is a naturally occurring biopolymer, this opens up exciting possibilities to create bioelectronic interfaces and devices.

Methods

Sample Preparation.

Eumelanin was synthesized following a standard method (42) that does not require metal ion catalysis. (thus, metal ion contamination can be excluded from the μSR and EPR studies.). The resulting melanin powder was dried overnight at 80 °C and then approximately 200 mg of powder was pressed into a 13 mm diameter pellet under a pressure of 5 Torr for 5 min. Pellets were contacted by evaporating gold through a custom-made shadow mask. The contact geometry was a van der Pauw configuration, leaving 2 mm of bare melanin between adjacent gold contacts. The van der Pauw configuration ensured that melanin samples achieved equilibrium with the environment (43).

Conductivity Measurements.

The electrical conductivity of melanin pellets was measured in a vacuum-sealed aluminum chamber with optical window and electrical feed-throughs [schematics have been published elsewhere (43)]. Current-voltage (IV) measurements were made using a Keithley 2400 source meter unit (SMU) in two- and four- contact (conductivity measurements) configuration. The pressure in the vacuum line attached to the chamber was monitored with a calibrated BOC-Edwards GK series, 0–50 mbar gauge. The line was also connected in series with a vial containing Milli-Q deionized water (freeze-pump-thawed three times) and an appropriate bleed valve. Conductivity measurements as a function of controlled hydration were performed as follows: The sample chamber was thoroughly evacuated and IV sweeps were performed using the SMU to check the integrity of the contacts. The conductivity measurement was then obtained by applying a DC 15 V potential to the sample while monitoring the current. Once the current reached equilibrium (after approximately 50 s) the value was recorded and the conductivity determined using the van der Pauw conductivity equation (43). The melanin pellet was then hydrated by bleeding water vapor into the vacuum line up to a pressure of 3 mbar and left to equilibrate (1 h); at which point further IV measurements were made. The sample was then hydrated in sequential 3 mbar steps in an identical manner with corresponding sets of IV measurements. The water content (% weight) of the samples was determined from the water vapor pressure in the vacuum line in conjunction with an absorption isotherm previously reported (28).

Muon Spin Resonance Measurements.

Pellets of melanin were glued to a silver plate in order to present the samples in an upright position into the path of the muon beam (EMU spectrometer, ISIS facility). The sample/silver plate was placed within a standard liquid sample holder (4K-Closed Cycle Refrigerator-Cell) with vacuum capability, to which a vacuum line was attached (similar to the conductivity measurements). Initially, a transverse field of 20 Gauss was applied to determine the geometric correction of the setup. The sample was hydrated in a similar procedure for the conductivity, starting at a relative water pressure (RP) of 0.2 (about 6 mbar, approximately 8.9%). After equilibrium was established, a longitudinal repolarization experiment was performed at fields ranging from 0 to 4,500 Gauss. After the repolarization experiment, the system was roughed with a turbo pump for approximately 10 s to recalibrate the pressure gauge. This procedure was then repeated for higher hydration levels in 0.1 RP steps to 0.7 RP. The μSR data was analyzed using Wimda software (44).

EPR Measurements.

Continuous wave EPR spectra were obtained with either a Bruker Elexsys E-500 or E-580 spectrometer consisting of a 10′′ electromagnet with a 12 kW power supply, an X-band SuperX microwave bridge, and an optical super high Q cavity. Furthermore, a Eurotherm 4131VT variable temperature controller was employed in conjunction with a liquid nitrogen flow through system (for photo-EPR/photoconductivity experiments) to ensure a constant sample temperature (296 K). A Bruker Teslameter (ER036M) and an EIP 548B (E500) or Bruker (E580) frequency counter were used to provide calibration of the magnetic field and microwave frequencies respectively. The equipment was controlled by the Bruker Xepr (v. 2.66.45) data acquisition software. A piece of melanin pellet (the largest face was oriented perpendicular to the pole faces) was placed within a standard EPR quartz tube, which was attached to a vacuum line setup similar to that used in the μSR and conductivity experiments. Hydration was controlled in a similar manner as described above. After equilibrium was reached (at a given hydration level), a two-dimensional power saturation (PS) dataset was recorded. It should be emphasized that the diode current (and hence the Q-factor), did not change over the course of an EPR hydration isotherm [varying from 0–21 mbar (0–18% weight gained) over an experimental time of 16 h]. Thus, the intensity changes observed in the PS data as a function of hydration exclude dielectric losses.

Photo-EPR and Photoconductivity Measurements.

A simultaneous photo-EPR/photoconductivity experiment was performed under identical hydration control on the Elexsys E-580 spectrometer described above. In order to perform the photoconductivity experiments, the vacuum assembly required modification. Internal wiring was added, which was attached to a custom-made KF25 blanking flange with an embedded electrical plug. The electrical equipment was the same for the conductivity experiment with the sample configured in the sandwich geometry. Light was coupled into the EPR cavity using a white LED light powered by an Agilent E3640A DC source. A CW EPR spectrum was recorded and a static magnetic field was chosen corresponding to the peak. The spectrometer was then configured to perform a time scan at this magnetic field. To avoid any fluctuations and/or drift over time in the microwave frequency and/or the static field, a field frequency lock involving the frequency counter and teslameter was employed. The photo-EPR/photoconductivity data acquisitions were then initiated simultaneously. After an approximately 50 s delay, the LED was switched on and kept on for about 50 s. The light was then switched off and another approximately 50 s of data collection ensued. The EPR data was filtered using a moving average (over three data points).