Proton conduction in inkjet-printed reflectin films

Biomolecular proton conducting materials have been touted as promising for seamlessly and directly interfacing natural biological systems with traditional artificial electronics.^1–7 Within this context, protein-based proton conductors feature several advantages, including well-defined sequences, modular structural characteristics, controllable self-assembly properties, ease of production, customizable physical properties, and intrinsic biocompatibility.^4–16 Consequently, proton conduction has been explored in a number of different protein- and polypeptide-based architectures, such as films from the cephalopod protein reflectin,⁸ mats from bovine serum albumin (BSA),⁹ membranes from squid ring teeth proteins,¹⁰ fibers from amyloid-β peptides,¹¹ and composites from MnO_x/tyrosine-rich peptides.¹² Among these materials, different reflectin isoforms have distinguished themselves as efficacious naturally occurring proton conductors with relatively high bulk conductivities of ∼3 × 10⁻³ S cm⁻¹ and carrier mobilities of ∼0.01 cm² V⁻¹ s⁻¹,^8,13 thus enabling applications as diverse as protonic transistors,¹⁴ photochemically dopable devices,¹⁵ and protochromic color-changing platforms.¹⁶ Moreover, the inherent biocompatibility of reflectins has been exemplified by their ability to support the growth and differentiation of human and murine neural stem and progenitor cells^17,18 and by their utility for the optical engineering of human cells.¹⁹ However, reflectins’ technological potential has been hampered by typically low-throughput material-intensive processing strategies, such as drop-casting [as shown in Fig. 1(a)], and by the frequent need for organic solvents, such as hexafluoroisopropanol (HFIP).^20–26 Thus, there exists an impetus for the development and validation of new approaches to the fabrication of reflectin-based films and structures for electrical device and other applications.

Over the past several decades, inkjet printing has emerged as one prominent and valuable strategy for the processing of protein- and polypeptide-based materials. Indeed, inkjet printers are widely available, relatively inexpensive, user-friendly, straightforward to automate, and material efficient.^27–30 These systems can rapidly produce protein-based microarrays or architectures with nearly arbitrary dimensions from low volume ink solutions with excellent reliability and in high throughput.^29–32 Due to such advantages, inkjet printing has been used not only for the spotting of proteins in arrays but also for the processing of various structural proteins, such as collagen and silk (typically from highly optimized multi-component formulations).^33,34 However, the inkjet printing of conductive proteins such as reflectin from water in the absence of any other additives [as shown in Fig. 1(b)] has not been reported to date, and the inkjet printing of intrinsically conductive proteins for applications in electrical devices remains rare in general.

Herein, we report the preparation of devices from inkjet-printed reflectin-based active layers and the systematic evaluation of their electrical characteristics. First, we fabricate and physically characterize single devices wherein films printed from different recombinant reflectins bridge two gold electrodes. Next, we comparatively interrogate the electrical behavior of such devices by means of electrochemical impedance spectroscopy (EIS) in the presence of both water (H₂O) and deuterium oxide (D₂O) vapor, benchmarking our results against prior observations. In turn, we fabricate and physically characterize arrayed devices wherein printed reflectin fibers bridge multiple palladium electrodes. Subsequently, we interrogate the electrical behavior of these devices when contacted by proton-blocking palladium and proton-injecting palladium hydride electrodes at different relative humidities, again benchmarking our results against prior observations. Taken together, our findings show that inkjet-printed reflectin-based architectures feature proton-conducting properties similar to those of dropcast reflectin films and establish a robust, effective, and general strategy for the processing of reflectin-based functional materials.

We began our experiments by fabricating independent two-terminal devices in which reflectin-based films served as active layers, as illustrated in Fig. 2(a). To this end, we adapted and modified previously reported literature procedures for analogous device configurations.^8,13 We first expressed, characterized, and solubilized both the wild type Doryteuthis (Loligo) pealeii reflectin A1 (WT RfA1) protein and a variant of this protein with an additional six histidine residues incorporated at its N-terminus (histidine-tagged RfA1) (see the supplementary material and Figs. S1–S3). We then electron beam evaporated paired gold metal electrodes (over an adhesion layer) through a shadow mask onto glass substrates [Fig. S4(a)]. We next inkjet-printed films from either WT RfA1 or histidine-tagged RfA1 solubilized in water such that these films bridged the paired gold electrodes [Figs. 2(b) and 2(c)]. We in turn characterized both types of printed films with optical microscopy, finding raised edges around the perimeter presumably due to solvent evaporation, as well as with atomic force microscopy (AFM), finding thicknesses of <∼3 µm and local root mean square (rms) roughnesses of <∼10 nm [Figs. 2(b) and 2(c)]. The overall protocol allowed for the rapid, reliable, and robust preparation of reflectin-based devices, which were amenable to systematic electrical interrogation.

With devices from inkjet-printed WT RfA1 and histidine-tagged RfA1 films in hand, we proceeded to comparatively investigate their electrical properties (and to thus assess the influence of the ionizable histidine residues) under different environmental conditions, as illustrated in Fig. 3(a). First, we performed EIS measurements for devices from WT RfA1 films when switching between H₂O and D₂O vapor in situ at a relative humidity of 90%. We noted that the resulting Nyquist plots exhibited a semi-circle and an inclined spur, as anticipated for a proton-conducting material in this device architecture [Fig. 3(b)].^6,8,13,35 In the presence of H₂O, we found that the films featured an effective conductivity of 0.052 ± 0.008 mS/cm, and in the presence of D₂O, we found that the same films featured an effective conductivity of 0.035 ± 0.006 mS/cm [Fig. 3(b)]. The observed kinetic isotope effect, i.e., a ∼33% ± 4% decrease in conductivity upon moving from H₂O to D₂O, confirmed that our printed material was a proton conductor [Fig. 3(b)]. Similarly, we performed identical EIS measurements for devices from histidine-tagged RfA1 films in the presence of H₂O and D₂O at relative humidities of 90%. We again noted that the Nyquist plots featured a semicircle and an inclined spur, as anticipated for a proton-conducting material in this device architecture.^6,8,13,35 In the presence of H₂O, we found that the films featured an effective conductivity of 0.059 ± 0.019 mS/cm, and in the presence of D₂O, we found that the same films featured an effective conductivity of 0.041 ± 0.017 mS/cm [Fig. 3(c)]. The observed kinetic isotope effect, i.e., a ∼32% ± 13% decrease in conductivity upon moving from H₂O to D₂O, confirmed that our printed material was a proton conductor. Here, the magnitudes of the kinetic isotope effects measured for our two types of inkjet-printed RfA1 films were consistent with models developed for a Grotthuss-type proton transport mechanism³⁶ and were in general agreement with experimental findings for dropcast reflectin films,^8,13 self-assembled peptide fibril networks,³⁷ and electrospun BSA mats.³⁸ Interestingly, the electrical properties of inkjet-printed films from WT RfA1 and histidine-tagged RfA1 were not only similar to one another (i.e., not substantively influenced by the incorporation of a small number of readily protonated histidine residues within the sequence) but were also comparable to those reported for dropcast films from the histidine-tagged reflectin A1 and histidine-tagged reflectin A2 isoforms.^6,8,13 Taken together, our findings indicated that reflectin-based materials maintained their proton-conducting functionality when processed into films from aqueous solutions via inkjet printing.

Having validated inkjet printing as an effective strategy for the processing of conductive films from our two RfA1 variants, we proceeded to extend our approach to the fabrication of arrayed two-terminal devices in which reflectin-based fibers served as the active layers, as illustrated in Fig. 4(a). To this end, we maintained procedures previously reported for similar device configurations and specifically focused our efforts on histidine-tagged RfA1 to facilitate comparisons with prior observations for dropcast reflectin films.^8,13–15 We first electron beam evaporated arrayed palladium metal electrodes (over an adhesion layer) through a shadow mask onto silicon dioxide/silicon (SiO₂/Si) substrates [Fig. S4(b)]. We then inkjet printed aligned fibers from histidine-tagged RfA1 solubilized in water such that the fibers spanned multiple arrayed palladium electrodes [Fig. 4(b)]. We in turn characterized the printed fibers with optical microscopy, finding raised edges along their length, as well as with AFM, finding thicknesses of <∼5 µm and local rms roughnesses of <∼10 nm [Fig. 4(c)]. The overall protocol allowed for the high throughput preparation of arrayed reflectin-based devices with more sophisticated layouts and form factors, facilitating subsequent electrical interrogation.

After producing the more advanced devices from histidine-tagged RfA1 fibers, we sought to study their electrical properties when contacted with different types of electrodes, as illustrated in Fig. 5(a). First, we recorded the current (I) as a function of the voltage (V) for devices from our fibers when contacted with both proton-blocking palladium contacts and proton-injecting palladium hydride contacts [Fig. 5(a)]. For both types of contacts, we noted that the I–V characteristics deviated from linearity and featured hysteresis between the forward and reverse scans presumably due to charge accumulation and/or depletion at the contacts, as reported for reflectins and various other biomolecular proton-conducting materials in comparable device architectures [Fig. 5(b)].^{6,8,13–15,37,39–42} For the palladium electrodes, we found that the fiber-based devices featured an estimated current density of ∼0.2 (±0.1) × 10⁻² A/cm² at 1 V, and for the palladium hydride electrodes, we found that the fiber-based devices featured an estimated current density of ∼1.6 (±0.4) × 10⁻² A/cm² at 1 V (with both measurements at a relative humidity of 90%) [Fig. 5(b)]. The observed approximately eightfold increase in the current levels for the proton-injecting contacts relative to the proton-blocking contacts indicated that our printed fibers were effective proton conductors.^{6,8,13–15,37,40–42} In analogous fashion, we recorded the current (I) as a function of time (t) for devices from our histidine-tagged RfA1 fibers when contacted with both proton-blocking palladium contacts and proton-injecting palladium hydride contacts [Fig. 5(a)]. For both types of contacts, we noted that the I–t characteristics exhibited a steep and rapid initial decay followed by an eventual plateau, as reported for reflectins and other biomolecular proton-conducting materials in comparable device architectures [Fig. 5(c)].^{6,16,39,40,43} For the palladium electrodes, we found that the fiber-based devices featured estimated current densities of ∼0.4 (±0.2) × 10⁻² A/cm² and ∼0.03 (±0.02) × 10⁻² A/cm² at 0 s and 60 s, respectively; and for the palladium hydride electrodes, we found that the fiber-based devices featured estimated current densities of ∼1.7 (±0.8) × 10⁻² A/cm² and ∼0.4 (±0.2) × 10⁻² A/cm² at 0 s and 60 s, respectively (with both measurements at a relative humidity of 90%) [Fig. 5(c)]. The observed approximately 11-fold increase in the current levels for the proton-injecting contacts relative to the proton-blocking contacts (at longer times and presumably at steady state) confirmed that our printed fibers were highly effective proton conductors.^{6,16,39,40,43} Notably, for both the I–V and I–t measurements, the current magnitudes were directly dependent on the relative humidity and significantly increased when the RH was raised from 60% to 90%, in agreement with expectations for a proton-conducting material (Fig. S5).^{6,8,37,40–42} In general, the electrical properties of inkjet-printed fibers from histidine-tagged RfA1 were quite similar to those of dropcast films from the histidine-tagged reflectin A1 and histidine-tagged reflectin A2 isoforms,^8,13–16 with nuanced differences resulting from the distinct processing conditions, device geometries, and/or measurement parameters. Overall, our findings demonstrated that reflectin-based materials retained their known proton-conducting functionality when processed not only into films but also into structures with variable form factors, e.g., fibers, from aqueous solutions via inkjet printing.

In summary, we have prepared devices from inkjet-printed reflectin-based active layers and systematically investigated their electrical characteristics, with our findings holding significance for several reasons. First, our report constitutes the first example of the inkjet printing of reflectin-based materials and one of the rare examples of the inkjet printing of conductive proteinaceous materials of any kind. Additionally, the described printing approach requires minimal amounts of reflectin, water as the only solvent, and low solution volumes, thus improving upon the spincasting, drop-casting, or doctor blading methods (and associated conditions) commonly used for the processing of reflectins.^6,26 Also, our rapid, reliable, and high throughput strategy for preparing reflectin-based structures maintains the constituent proteins’ known advantageous proton-conducting functionality while allowing for diverse and programmable form factors. Moreover, devices from our inkjet-printed reflectin active layers feature electrical characteristics on par with those of analogous devices from other biomolecules, underscoring the general utility of reflectins as conductive materials. Here, we note that there remains future work associated with precisely controlling the long-term stabilities of high-concentration reflectin solutions to prevent precipitation, optimizing the conditions needed for improved uniformity of the edges of printed reflectin films and fibers, extending the methodology to the processing of various reflectin mutants and isoforms, and developing a better understanding of structure–function relationships for reflectin-based active layers. Even when accounting for these challenges, our methodology constitutes a promising step forward for the manufacturing and continued improvement of advanced bioelectronic platforms from reflectins.

See the supplementary material for the detailed experimental procedures, amino acid sequences of wild type reflectin A1 and histidine-tagged reflectin A1 (Fig. S1), HPLC chromatograms and sequence coverage maps obtained from tryptic digestion in tandem with mass spectrometry analysis for wild type reflectin A1 and histidine-tagged reflectin A1 (Figs. S2 and S3), digital camera images of the fabricated devices with gold and palladium electrodes (Fig. S4), and electrical characterization data for two-terminal palladium hydride devices from histidine-tagged reflectin A1 at different relative humidities (Fig. S5).

Y.L. and P.P. contributed equally to this work.

The data required to evaluate the conclusions in this work are included in the main text and/or the supplementary material. The data that support the findings of this study are available from the corresponding author upon reasonable request.

The authors are grateful to the Air Force Office of Scientific Research (Grant Nos. FA9550-17-1-0024, FA9550-16-1-0296, and FA9550-14-1-0144 to A.A.G.) and the Defense Advanced Research Projects Agency (Grant No. D16AP00034) for financial support. F.G.O. acknowledges support from the Office of Naval Research (Grant No. N00014-19-1-2399). The authors also thank the Mass Spectrometry Facility at the University of California, Irvine.