Building blocks for bioinspired electrets: molecular-level approach to materials for energy and electronics

Preparation of the anthranilamide residues

As polypeptide conjugates of aromatic non-native β-amino acids, we build the Aa oligomers from their C- to their N-termini [46]. Each Aa residue is added to the sequence as a 2-nitrobenzoic acid derivative [45, 46]. A selective reduction of the nitro group to amine prepares the thus added N-terminal residue for the next amide-coupling step [46]. The use of “traditional” synthetic protocols, where each residue is introduced as an N_β-Fmoc or N_β-tBoc anthranilic acid derivative, renders negligible to no yields. The electron-withdrawing nitro group at ortho position ensures the electrophilicity of the carbonyl carbon of the activated carboxylate that is needed for coupling it with the N-terminal amine. The N-terminal anthranilic amines are weak nucleophiles due to the neighboring electron-withdrawing carbonyls.

To attain a diversity of Aa residues, we focus on the preparation of a variety of 4- and 5- derivatives of the 2-nitrobenzoic acid as precursors for non-native Aa residues. Three types of substituents at R₁ and R₂ positions (Fig. 1) allow for producing electron-rich Aa residues with a wide distribution of the energy levels of their frontier orbitals: (1) strong electron-donating groups, dialkylamines; (2) moderately strong electron-donating groups, alkoxyls; and (3) a weak electron-donating group, methyl. Starting with fluoro-derivatives, nucleophilic aromatic substitution allow for introducing amines as substituents at the 4^th and 5^th positions of the 2-nitrobenzoic acid. In the 5-fluoro-2-nitrobenzoic acid, the C–F bond at the para position in relevance to the nitro group is quite polarized, making that carbon susceptible to the attack from an amine nucleophile. Using a cyclic secondary amine, i.e. piperidine, requires relatively short reaction times under conventional heating to produce in quantitative yields the nitrobenzoic precursor for the 5Pip residue (Scheme 1a) [45]. In the 4-fluoro-2-nitrobenzoic acid, the C–F bond is not as polarized due to the meta (rather than para) position of the nitro group. For the precursor for the 4Pip residue, therefore, the same nucleophilic aromatic substitution with the 4-fluoro-2-nitrobenzoic acid requires longer reaction times for attaining similar yields (Scheme 1b).

Scheme 1:

Syntheses of the 2-nitrobenzoic acid precursors for (a) 5Pip, (b) 4Pip, (c) 5Hxm, (d) 4Hxm, and (e) Hox.

A synthetic challenge arises when the nucleophiles are non-cyclic secondary amines. Entropic restrictions decrease the nucleophilic reactivity of dialkylamines as the length of their chains increases [59]. Under conventional heating, the substitution of fluorine in 4-fluoro-2-nitrobenzoic acid with hexylmethylamine leads to negligible yields even when the reaction proceeds for unreasonably long periods of time (Scheme 1c).

Utilizing microwave radiation as a heat source allows for addressing this challenge. Although still debated, microwave heating may enhance reaction rates via the entropy components of their activation energies [60, 61], making it appropriate for overcoming the limitations imposed by amine nucleophiles with long alkyl chains. Employing microwave heating, indeed, allows us to develop procedures that lead to completion of the syntheses of the hexylmethylamino precursors for 4Hxm and 5Hxm (Fig. 2) within reasonable time durations (Scheme 1c,d).

The starting material for the Hox precursor is 5-hydroxy-2-nitrobenzoic acid, the carboxylate and the phenolate of which are indiscriminately strong nucleophiles readily producing the dialkyl derivatives. An extra hydrolysis step leads to the Hox nitrobenzoic precursor (Scheme 1e). Coupling of the 2-nitrobenzoic acids with 1-hexylamine, followed by selective reduction of the nitro group to amine and another amide coupling produces the eight Aa residues [45].

Reduction potentials

Electrochemical studies allow for quantifying the propensity of the Aa residues to serve as electron donors and to potentially mediate hole transfer. The reduction potentials of the oxidation of the Aa residues (i.e. EAa•+/Aa(0) : Aa•+ + e− → Aa) provide a means for quantifying their electron-donating capabilities [62], and for estimating the energy levels of their HOMOs [63].

To elucidate the media effects on the electronic properties, we focus on the dependence of the Aa electrochemical potentials on the solvent polarity [64–68]. Our selection includes five aprotic solvents with different polarity that have electrochemical windows extending over the expected potentials needed for the oxidation of the Aa residues: chloroform (CHCl₃), dichloromethane (DCM), benzonitrile (PhCN), acetonitrile (MeCN), and propylene carbonate (PC).

As a representation of the standard electrode potentials, E⁽⁰⁾, the half-wave potentials, E^(1/2), are readily obtained from cyclic voltammetry (CV). For reversible electrochemical oxidation, E^(1/2) represents the average between the peak potentials of the anodic and the cathodic waves. If the lifetimes of the radical cations, Aa^•+, generated on the surface of the working electrode do not extend over milliseconds and seconds (the time scales of CV), the cathodic wave becomes undetectable. For such electrochemically irreversible oxidation, the potential at the inflection point of the rise of the anodic wave provides an estimate for E^(1/2) (see Supplementary Material). 5Hxm, 5Pip, and Dmx manifest electrochemically reversible oxidation indicating that these residues produce radical cations with pronounced stability.

While electrochemical measurements require media with high electrolyte concentrations, it is the information for E⁽⁰⁾ in neat solvents that is directly relevant to spectroscopy and computational data [65]. From the dependence of the measured EAa•+/Aa(1/2) on the electrolyte concentration, C_el, therefore, we extrapolate the values of the reduction potentials of the residues for zero electrolyte concentration, i.e. the Aa potentials for neat solvents, EAa•+​​/​Aa(Cel=0) (Fig. 3a) [65].

Fig. 3:

Solvent dependence of the electrochemical potentials of the anthranilamide residues. (a) Dependence of the half-wave potentials of the Aa residues on the electrolyte concentration, C_el (for DCM in the presence of (C₄H₉)₄NPF₆ as electrolyte). Extrapolation to zero electrolyte concentration from exponential data fits provides the estimates for the reduction potentials of the residues in neat solvents. (b, c) Dependence of the extrapolated potentials for neat solvents on the media dielectric characteristics obtained from measurements for five different solvents: propylene carbonate, PC (ε^–1 = 0.016); acetonitrile, MeCN (ε^–1 = 0.027); benzonitrile, PhCN (ε^–1 = 0.040); dichloromethane, DCM (ε^–1 = 0.11); and chloroform (ε^–1 = 0.21).

Based on the Born solvation energy [69], a linear correlation between EAa•+​/​Aa(Cel=0) and the inverse dielectric constant of the neat solvents, ε^–1, reveals the effects of the media polarity on the electrochemical properties of each Aa residue (Fig. 3b,c). As expected, an increase in the solvent polarity (i.e. a decrease in ε^–1) shifts the potentials to less positive values, elevating the energy levels of the HOMOs, and improving the capabilities of the residues as electron donors (Fig. 3b,c). The stabilization of the radical cations, Aa^•+, by polar media accounts for this negative shifts of the measured potentials.

For each solvent, an increase in the electron-donating strength of the substituents, from methyl to amines, causes negative shifts in the reduction potentials. The potential of the best electron donor, 5Hxm, is about 1 V more negative than that of Ant (Fig. 3b, Table 1). This finding is consistent with our theoretical predictions that placing dialkylamine at the 5^th position of anthranilamides elevates the energy levels of their HOMOs with about 1 eV [48].

While the amine-derivatized Aa residues are the best electron donors, moving the amine substituents from the 4^th to the 5^th position causes another negative shift (of about 0.2–0.4 V) in the potentials (Fig. 3b,c). The alkylamines in 5Pip and 5Hxm are para-oriented to the electron-donating N-terminal amide and meta-oriented to the electron-withdrawing C-terminal amide. This para-orientation between the two electron-donating groups in 5Pip and 5Hxm can account for the more negative values of their potentials in comparison with 4Pip and 4Hxm.

This “reinforcement” from two electron-donating groups para-positioned to each other can account for stabilizing the radical cations, Aa^•+, and the reversible electrochemical oxidation of 5Pip and 5Hxm. This argument should hold also for the other electron-donating groups at the 5^th position. While Dmx manifests reversible oxidation, however, the cyclic voltammograms of Hox and Met exhibit irreversible behavior. Despite the stabilization that 5-methyl and 5-hexyloxy groups might provide to the racial cations of Met and Hox, respectively, their reduction potentials appear positive enough to irreversibly cleave the amide bonds attached to the aromatic rings [70].

Photophysical properties

While electrochemical analysis provides information about the capabilities of the Aa residues to serve as electron donors and hole transducers, optical spectroscopy reveals complementary features about the energetics of the Aa frontier orbitals. In particular, UV/visible absorption and fluorescence spectroscopy provide a means for estimating the zero-to-zero energies, ℰ₀₀. ℰ₀₀ represents optical HOMO-LUMO gaps, which in molecular photophysics can be viewed as the optical band gaps of these Aa building blocks for organic materials.

The eight Aa residues absorb in the UV spectral region and fluoresce with substantial quantum yields, ranging between about 0.1 and 0.3 (Fig. 4) The wavelength where the intensity-normalized absorption and emission spectra cross provides a means for estimating ℰ₀₀ (Fig. 4c) [71–74]. For the Aa residues, ℰ₀₀ ranged from about 3 to 3.7 eV (Fig. 5c). The capability of the alkyloxy and the dialkylamine substituents to extend the π-conjugation of the aromatic rings leads to a decrease in ℰ₀₀ that is consistent with narrowing the HOMO-LUMO gaps of the residues. This effect, however, was pronounced only for strong electron-donating substituents placed at the para position to the N-terminal amides (5Pip and 5Hxm vs. 4Pip and 4Hxm, Fig. 5).

Fig. 4:

UV/visible absorption and emission spectra of anthranilamide residues for various solvent media: propylene carbonate (PC), acetonitrile (MeCN), benzonitrile (PhCN), dichloromethane (DCM), chloroform (CHCl₃), and cyclohexane (CH). (a) Absorption spectra of the eight residues for MeCN. (b) Fluorescence spectra of the residues recorded for MeCN and DCM (λ_ex = 310 nm; each fluorescence spectrum was normalized by ×(1 – 10^–A(λex))^–1). (c) Absorption and fluorescence spectra for Hox in the different solvents (λ_ex = 310 nm; each fluorescence spectrum was normalized to the height of the red-most band of the corresponding absorption spectrum; except for PC, the baselines of the spectra are elevated from 0 for improved visualization; the arrows point to wavelength, λ₀₀, of crossing point between the two spectra that is used for calculating the zero-to-zero energy, ℰ₀₀ = hc/λ₀₀).

Fig. 5:

Absorption and emission properties of the anthranilamide residues. (a) Wavelengths of the maxima of the most red-shifted bands of the absorption spectra of the eight residues for different solvents: propylene carbonate (PC), acetonitrile (MeCN), benzonitrile (PhCN), dichloromethane (DCM), chloroform (CHCl₃), and cyclohexane (CH). (b) Wavelengths of the maxima of the fluorescence spectra of the eight residues in different solvents (λ_ex = 310 nm). For each residue that, due to aggregation, exhibits two fluorescence bands, the wavelength of the blue-shifted maximum is reported. (c) Zero-to-zero energy values of the anthranilamide residues, extracted from wavelength where the normalized absorption and emission spectra cross (Fig. 4).

Similar comparison for the alkyloxy-derivatized residues reveals that the spectral features of Hox are red-shifted compared to these of Dmx (Fig. 5). This red spectral shifts for Hox vs. Dmx, are most likely due to the methoxy group at the 4^th position of Dmx. An electron-donating group at the 4^th position appears to cause blue spectral shifts. Even strong electron-donating groups, such as amines, placed at the 4^th position, i.e. 4Hxm and 4Pip, result in Aa residues with spectral features similar to those of Met and Ant (Fig. 5).

While ℰ₀₀ depends on the R₁ and R₂ substituents, the solvent polarity has insignificant to no effect on the spectral properties of the Aa residues (Figs. 4 and 5). This lack of substantial solvatochromism is consistent with our previous experimental observations and theoretical findings for Ant and 5Pip derivatives [45, 46]. The anthranilamides are, indeed, polar molecules. The observed lack of significant solvatochromism, therefore, suggests that photoexcitation of the Aa residues does not substantially alter their polarity. That is, the permanent ground-state dipoles have dominating effect on the ground- and excited-state polarity of the Aa conjugates.

Another feature revealed by the emission spectra of the Aa residues is their propensity to aggregate. As expected from our previous studies [46], two of the residues, Ant and Met, exhibit fluorescence bands with two peaks (Fig. 4b), the ratios between which are concentration dependent. We ascribe the red-shifted peak, the intensity of which increases with an increase in concentration, to aggregates that form at the excited and/or the ground state [46, 75–80]. The trends from this assignment of the fluorescence spectra indicate that the chlorinated hydrocarbons, such as DCM, tend to suppress aggregation (Fig. 4b).

In addition, 4Pip and 4Hxm also aggregate at μM concentrations when dissolved in some of the tested organic solvents (Fig. 4b). This finding was somewhat surprising because the residues with identical alkyl chains attached to them, 5Pip and 5Hxm, did not manifest detectable aggregation even at concentrations reaching 1 mM. These findings show that the position of substituents with alkyl chains (i.e. R₁ vs. R₂) pronouncedly affects the aggregation propensity of the Aa derivatives. To confirm this trend, other two residues, Hox and Dmx, that also contain alkyl chains at the 5^th, also show a single fluorescence peak, indicating that they do not manifest detectable aggregation in the tested organic solvents (Fig. 4b).

Comparison between 4Pip and 4Hxm reveals an important trend about the dependence of the aggregation propensity on the structure of the substituents. Both residues aggregate when dissolved in most organic solvents at μM concentrations. In chlorinated solvents, however, while 4Pip exists as a monomer, 4Hxm manifests some propensity for aggregation (Fig. 4b). Both residues contain secondary amines at the 4^th positions of their aromatic rings. Both substituents contain more than five carbons in their alkyl chains (piperidinyl for 4Pip, and hexyl and methyl for 4Hxm, Fig. 2). In fact, 4Hxm has a longer chain than 4Pip. Contrarily to correlating improved solubility with increased length of alkyl substituents, however, 4Hxm has a larger propensity for aggregation than 4Pip. In the piperidinyl substituents more carbons are located closer to the aromatic ring than in hexylmethylamine. This increase in the volume of the solvation cavity immediately next to the aromatic moieties (that may drive aggregation) appears to have dominant effect on improving the residue solubility. This trend indicates that long linear chains, such as hexyls, do not improve the solubility in organic solvents to the extent that branched substituents, such as piperidinyl, do.

Permanent electric dipoles and distribution of the frontier orbitals

Ab initio computational studies provide further key information about the electronic properties of the Aa residues and the dependence of these properties on the media polarity. Ground-state DFT calculations at the B3LYP/6-311 + G(d,p) level [81–83] performed using Gaussian 09 [84] revealed that the HOMOs and LUMOs of all eight Aa residues are predominantly localized on their aromatic rings (Fig. 6). For the residues with no substituents or with relatively weak electron-donating R₁ and R₂ groups, i.e. for Ant, Met, Hox and Dmx, the HOMOs extend over the N-terminal amide bond, while the LUMOs tend to delocalize over both amides (Fig. 6). Placing a strong electron-donating group on the 5^th position (i.e. R₂ = dialkylamine) amplifies these orbital-delocalization trends (see 5Pip on Fig. 6, and 5Hxm in the Supplementary Material). Conversely, placing the same strong electron-donating groups on the 4^th position shifts the delocalization of the HOMOs to the C-terminal amides (see 4Pip Fig. 6, and 4Hxm in the Supplementary Material).

Fig. 6:

HOMOs and LUMOs of Ant, Met, 4Pip and 5Pip for the gas phase (GP) and for acetonitrile (MeCN), obtained from DFT calculations. For the computational studies, the alkyl chains at the C- and N-termini were truncated to C₂H₅. The residues are displayed with their N-termini oriented to the left and the C-termini – to the right. (See the Supplementary Material for the HOMOs and LUMOs of all eight residues.)

To account for the solvent effects on the electronic properties of the Aa residues, we introduced DCM and MeCN to the calculations using a polarizable continuum model [85–87]. Inclusion of the solvents had no visible effect on the distribution of frontier orbitals (Fig. 6). Most importantly, the optimized structures of the eight Aa residues remain practically identical when varying the solvent media (see Supplementary Material). While the anthranilamides are polar molecules, principally because of the amide dipoles, these computational findings suggest that the preferential Aa conformation, with trans amides, is not affected by changes in the solvent polarity.

The permanent electric dipoles are the most important feature of the Aa residues, making them promising building blocks for electrets. The predicted dipole moment of Ant is about 4.7 D (Table 1), which is consistent with the contributions of 1.9 D from each of the two amides and of 0.9 D from the polarization due to the hydrogen bonding (Fig. 1) [48]. Adding methyl to the 5^th position, such as in Met, slightly enhances the dipole (Table 1), which is consistent with the 5-methyl-induced polarization of the aromatic ring co-directionally with the total dipole from the amide and hydrogen bonds. This enhancement effect on the Aa dipole is even more pronounced for Hox, 5Hxm and 5Pip, where R₂ electron-donating groups have mesomeric, rather than inductive, effects on the aromatic ring (Table 1). Conversely, when an electron-donating group is at the 4^th position, such as in 4Hxm and 4Pip, the substituent-induced polarization of the aromatic ring has a diminishing effect on the total residue dipole (Table 1).

Solvent polarity further enhances the magnitude of the Aa dipoles (Table 1). Because the electronic and structural properties of the Aa residues have a negligible dependence on the media, this polarity-driven enhancement can be attributed to the effect of the Aa dipoles on the solvent itself. As we have shown for simple aliphatic amides, such dipole enhancement results from the Onsager fields inside the solvation cavities of the solutes [66, 88]. The Aa dipoles polarize the solvent media in the proximity to the solvation cavities. This polarization increases the displacements between the centers of the positive and negative charges of the solvated Aa molecules, increasing the magnitudes of their total dipole moments.