Colour-sensitive conjugated polymer inkjet-printed pixelated artificial retina model studied via a bio-hybrid photovoltaic device

In recent years, organic electronic materials have been shown to be a promising tool, even transplanted in vivo, for transducing light stimuli to non-functioning retinas. Here we developed a bio-hybrid optoelectronic device consisting of patterned organic polymer semiconductors interfaced with an electrolyte solution in a closed sandwich architecture in order to study the photo-response of photosensitive semiconducting layers or patterns in an environment imitating biological extracellular fluids. We demonstrate an artificial retina model composed of on an array of 42,100 pixels made of three different conjugated polymers via inkjet printing with 110 pixels/mm2 packing density. Photo-sensing through three-colour pixelation allows to resolve incoming light spectrally and spatially. The compact colour sensitive optoelectronic device represents an easy-to-handle photosensitive platform for the study of the photo response of artificial retina systems.


Results
The results section is organized following our research approach: first we designed and validated a bio-hybrid sandwich structure consisting of a polymer semiconductor and a physiological electrolyte solution; second we developed a simplified concentric model of a pixelated retina via inkjet printing of different conjugated polymers that mimic the three spectra of rods and two cones typical of mammalian retinas.
Bio-hybrid device: design, fabrication and biocompatibility. Here we present a bio-hybrid device composed of both biological (physiological electrolyte solution mimicking extracellular fluid) and non-biological components. The developed device architecture consists of two transparent electrodes separated by a thermoplastic encapsulant with the photosensitive polymer thin film interfaced with an electrolyte mediator. The electrolyte solution was a buffered isotonic solution (Phosphate buffered saline solution, PBS) commonly employed in a biological context to reproduce a physiological extracellular fluid 34,35 . The concept of the device is the result of the amalgamation of different set-ups from different fields, i.e. those from electrophysiology 12,13,15 , organic photovoltaics (OPV) 9,36 and Dye-Sensitized Solar Cells (DSSC) [37][38][39] . Figure 1a,b illustrates a schematic of the device structure. The working electrode (sensitive photo-electrode) is 100 nm-thick polymer semiconductor film, P3HT at first (see "Methods" section), spin-coated on a glass substrate covered with a transparent conducting oxide (fluorine tin oxide-FTO). P3HT is a prototype material for OPV applications 40 , it works as an absorber Figure 1. Bio-hybrid device layout for biological system/organic electronic interfaces and biocompatibility evaluated as cell adhesion, proliferation and viability. (a) Schematic of the device. It consists of two transparent conductive electrodes. The photo-electrode (polymer thin film on FTO) is grounded. The platinum layer deposited on FTO is the counter-electrode. A phosphate buffered saline solution (PBS) was used as electrolyte and it is enclosed in the inner chamber (60 µm thick) formed between the two electrodes (Image created by using Microsoft PowerPoint). (b) Picture of the fabricated device. Dimensions are also shown. (c) Phase contrast micrographs of randomly selected fields of SH-SY5Y cells cultured on standard polystyrene dish (CTRL) and on Glass|FTO|P3HT substrate (P3HT) for 3 days. Scale bar = 50 μm. (d) Cell adhesion evaluated at 16 h after seeding SH-SY5Y cells on polystyrene dish (CTRL) and on Glass|FTO|P3HT (P3HT). Data are expressed as percentage of the CTRL. Statistical difference vs CTRL **p < 0.01 (Data were analysed with GraphPad Prism software V7.0). (e) Cell proliferation evaluated after three days of culture and expressed as cell doubling time for SH-SY5Y cultured on polystyrene dish (CTRL) and on the P3HT layer (Image J software 1.49 V was used. Data were analysed with GraphPad Prism software V7.0). (f) Representative micrographs of SH-SY5Y cells cultured on polystyrene dish (CTRL) and on Glass|FTO|P3HT substrate (P3HT) and analysed via the Click-iT EdU proliferation assay after 3 days of culture (T3). White and red arrowheads indicate non-proliferating and proliferating (EdU positive) cells, respectively. Scale bar = 50 μm. (g) Quantification of cell proliferation measured by Click-iT EdU proliferation assay, expressed as percentage of positive cells on the total number of cells analysed at the beginning (T0) and at the end (T3) of culture (Data were analysed with GraphPad Prism software V7.0). (h) Representative micrographs of SH-SY5Y cells cultured on polystyrene dish (CTRL) and on Glass|FTO|P3HT substrate (P3HT) and analysed via the in-situ Cell Viability Imaging assay after three days of culture (T3). White and red arrowheads indicate live and dead (green) cells, respectively. Scale bar = 50 μm. (i) Quantification of cell death measured by in-situ Cell Viability Imaging assay, expressed as percentage of dead (green) cells on the total number of cells analysed at the beginning (T0) and at the end (T3) of culture (Image J software 1.49 V was used). www.nature.com/scientificreports/ and electron donor material. The counter-electrode is a platinum (Pt) layer screen printed through a precursor on FTO-coated substrate and then fired at 480 °C for 30 min 41 . The latter procedure, borrowed from DSSC technology 42 , converts the used Pt precursor paste into catalytic Pt nanometre-sized islands that permit a facile exchange of electrons/ions at the electrode/electrolyte interface maintaining a high degree of transparency with a resulting transmittance of up to 79% between 400 and 700 nm 43 (Supplementary Fig. S1). Moreover platinum electrodes are widely used in bioelectronic applications because they can safely deliver charge densities ranging from 0.15 to 5.57 mC/cm 244 . PBS is contained in an inner chamber delimited by the thermoplastic sealant gasket (i.e. Surlyn). The presence of the biological electrolyte is important to provide a similar physiological aqueous environment for cells, mimicking eye conditions, and to test whether the polymer layer will retain its optoelectronic features in such biological/aqueous environment. The spacing between the two electrodes can be controlled via the thickness of the thermoplastic spacer. Here 60 µm thick gaskets were used representing a thickness which elicit measurable signals and potentially being able to accommodate smaller biological cells. Thicknesses greater than 60 µm can be used to place larger retinal tissues into the device for future studies. To address the issue of biocompatibility, human neuroblastoma cells SH-SY5Y were seeded on a P3HT layer and grown for up to 3 days, as described in "Methods" section. Cell adhesion, proliferation and viability were evaluated and compared with parallel cultures carried out with SH-SY5Y cells grown on tissue culture tested polystyrene dishes, as positive control (CTRL) (Fig. 1c). Adhesion was analysed 16 h after cell seeding (T0) and the results showed an adhesion rate of 50-60% on P3HT layer in comparison to control (CTRL = 100 ± 9.15%; P3HT = 54.07 ± 8.14%) (Fig. 1d).
For the analysis of cell proliferation, the number of cells seeded on P3HT was adjusted based on the calculated rate of adhesion to have the same number of adhered cells at the beginning of culture (T0) in CTRL and P3HT conditions. When the adhered cells (CTRL = 1.49 ± 0.38 × 10 4 cells/cm 2 ; P3HT = 1.40 ± 0.29 × 10 4 cells/cm 2 ) were allowed to grow for 3 days, no significant differences were observed between the two culture conditions in the proliferation rate expressed as cell doubling time (CTRL = 1.50 ± 0.02 day; P3HT = 1.50 ± 0.10 day) (Fig. 1e). Proliferation was also confirmed by the Click-iT EdU proliferation assay performed on the cells at the beginning (T0) and after 3 days (T3) of culture ( Fig. 1f, Supplementary Fig. S2) was shone on three different device architectures containing: (i) PBS, (ii) polymer, and (iii) the complete structure (polymer+PBS), to gauge the effectiveness of the concept. Without either polymer layer or electrolyte, no signals were elicited. Current densities at short circuit (J ph ) and photo-voltage at open circuit (V ph ) were instead generated upon photoexcitation for the complete device ( Fig. 2a,b). We illuminated the device for 300 ms with a high-speed white LED (5000 K) as falling within the ranges used for in vitro retinas and living cells light stimulation studies 12,14,27,32,45,46 and below the limit allowed for prosthetic application 47,48 . The steady state photo-current after an initial spike is an evidence of a Faradaic process at the electrode/electrolyte interface 49 . Similarly to the concept of a DSSC 38 charges are photogenerated in the semiconductor film, whilst the ions present in the biological liquid permit the exchange of charge at both the polymer/electrolyte and electrolyte/electrode interfaces enabling the presence of a sustained current. The platinum film catalyses charge transfer between the ions and the top transparent electrode. Current spikes visible when the system is perturbed 50,51 , i.e. when light is turned ON or OFF, are the result of a capacitive process due to electrostatic perturbation of accumulated carriers and ions with strong Coulomb interactions localized close to the polymer/electrolyte and electrolyte/metal interfaces (e.g. Helmholtz layers) 49 . Capacitive coupling is one of the most biologically safe mechanism used for cells and tissues electrical stimulation 15,52 . In this device the electrolyte maintains the physiological environment of biological cells and tissues (which would be present in an in vivo investigation or that could be potentially placed inside the device) and enables the movement and transport of charge carriers along the section of the device between photo-electrode and counter-electrode. J ph values, referred to the steady state (at the half of the light pulse width, namely at 150 ms from the light switched ON) current signals are in the order of 0.17 ± 0.10 µA/cm 2 . J ph spike values are in the order of 0.51 ± 0.28 µA/cm 2 and of − 0.29 ± 0.12 µA/cm 2 at the start and the end of the impulse of light respectively (Fig. 2a). Under the convention of positive J values typically used in physiology reporting, the negative V ph signals decrease as a function of time upon light stimulation with V ph values in the order of − 5.4 ± 1.8 mV at a radiant exposure of 53.4 J/m 2 corresponding to 17.8 mW/cm 2 for 300 ms (Fig. 2b). The elicited photo-responses were similar to those reported in patch-clamp electrophysiology 27 , contact pad 14 , and MEA 12 experiments where the samples are immersed in an open electrolyte bath with a reference electrode. This demonstrates the successful operation of our bio-hybrid device architecture concept. The opto-electrical response of the device was tested at different light power levels (Fig. 2c,d) ranging from 17.8 mW/cm 2 (∼10 × 10 3 lx) to 93.8 mW/cm 2 (∼60 × 10 3 lx) representing the intensity range typically experienced in human photopic vision outdoors 53 (from 10 to 100 × 10 3 lx) and the range of irradiance levels generally used for in vitro retinal stimulation experiments 12,14,47,54 . Currents (J ph first peaks, Fig. 2c; J ph second peaks Supplementary Fig. S4b; J ph steady state Supplementary Fig. S4c) increase linearly with irradiance whereas V ph max (measured after 300 ms from light switched ON) as function of the increasing irradiance levels shows a quasilogarithmic trend (Fig. 2d). Considering possible future application of the bio-hybrid device for in vitro retinas stimulation, we also used shorter light pulse stimuli of 10 ms (optical pulse width for retinas stimulation 13,14 ), to opto-electrically characterize the device ( Supplementary Fig. S3). Recorded values of current capacitive peaks are similar with good approximation for 10 and 300 ms light pulse duration ( Supplementary Fig. S4a,b) We also examined the total charge generated by the device upon illumination. The total charge (calculated as the time integral of photocurrent) at different irradiance levels was 2.8 ± 0.2 nC/cm 2 and 10.4 ± 3.3 nC/cm 2 for 10 ms light pulse stimulation (Fig. 3a) and 106.4 ± 23.7 nC/cm 2 and 331.9 ± 42.8 nC/cm 2 for 300 ms light pulse stimulation (Fig. 3b) at 17.8 and 93.8 mW/cm 2 respectively. Thus, the device is able to generate a charge level in the range of few nC/cm 2 which is useful for neural stimulation and in particular for retinas stimulation (bipolar and ganglion cells) 55 . The total photo-generated charge consists of a capacitive and a faradaic component. While the capacitive charge corresponds to the current spikes generated during few ms after the light is switched ON and OFF and includes redistribution of charges at the polymer/electrolyte and electrolyte/metal interfaces, the faradaic component is related to the steady-state photocurrent and involves transfer of electrons across the electrode/electrolyte interface 51 . As suggested by previous works, faradaic reactions implicate reduction/oxidation reactions involving the oxidation of the organic semiconductor interfaced with the electrolyte solutions. Oxygen reduction and hydrogen peroxide production or hydrogen evolution reactions have been implied 56,57 . Transport of mobile ions present in the electrolyte solution across the device cross-section and charge-transfer at the electrodes are also processes that give rise to photo responses of this type 58 .
To gauge the lifetime of the device, we performed an accelerated ageing test. The sandwiched device containing PBS was subjected to continuous white light illumination at 17.8 mW/cm 2 (density power typically  S5a) up to 10 min, i.e. sufficient for testing the response of the devices, whilst the J ph max decreased by 15% ( Supplementary Fig. S5c). At the very high optical powers of 93.8 mW/cm 2 degradation accelerated significantly: V ph max diminished by 12% ( Supplementary Fig. S5b) and J ph max by 40% ( Supplementary Fig. S5d) after 5 min. We continued the accelerated test at these high powers for 48 h noticing a reduction of the current by 64% from the fresh device and visible alteration of both the photo-electrode (formation of blisters and wrinkles) and counter-electrode (residues of the biological medium) as well as evaporation of the electrolyte ( Supplementary Fig. S5e,f). This is not surprising as the test at the high irradiance of 93.8 mW/cm 2 is the equivalent of looking straight at the sun which is known to damage the eye even after few seconds 59,60 . It would be interesting to carry out a long detailed study on lifetimes by varying systematically the irradiance levels and determine the threshold that leads to significant degradation on materials and electrolyte solutions in order to evaluate what changes are needed to improve the lifetimes of artificial retina systems.

Pixelated polymer artificial retina via inkjet printing.
Once the biocompatibility and the optoelectronic operation of the devices presented in Fig. 1 was demonstrated, the next step was to extend the singlepolymer-layer device to a pixelated, colour sensitive, three-polymer-based device towards an artificial retina model. The size of a human adult retina is typically around 22.0 mm in diameter 61 . Photoreceptors are organized in a well-defined mosaic 62 where rod densities are predominant in the peripheral retinal region while cones in the central part (fovea). Out of all cones, about 64% are L type, 32% M type and only 2% S type 63,64 . Relative ratio and position of cones varies greatly among different people even with regular vision 65 . Recently the concepts of pixelated visual implants have been exploited and retinal stimulating pixels (~ 100 µm in diameter) have been obtained by spin-coating, sputtering and photolithography techniques 14 . We designed a simplified model to fabricate the pixelated artificial retina system using the resulting anatomical map of the photoreceptors mosaic and placing locally different polymer semiconductors with specific absorbance spectra via printing techniques (Fig. 4a,b). Conjugated polymers have been previously proposed as promising materials for colour detection 30,31,33 due to band gap and optical tunability. We firstly identified polymers with absorbance spectra similar to that of human photoreceptors (for human photoreceptors absorbance we referred to Bowmaker and Dartnall studies 66 ) (Fig. 4c). Regio-regular P3HT polymer (see "Methods" section) and a blend of P3HT with [6,6]-Phenyl C61 butyric acid methyl ester (PC 61 BM) (P3HT:PCBM) were used, mimicking well the green cone and rods absorbance spectra, respectively. As a third photosensitive polymer we chose Poly(9,9-di-n-octylfluorenyl-2,7-diyl) (PFO), to mimic S-cones. Its peak is shifted significantly from both those of P3HT and the P3HT:PCBM blend and thus they are all spectrally well distinguishable. Then the biocompatibility was evaluated for the P3HT:PCBM and PFO layers ( Supplementary Fig. S6a), as previously described for P3HT (Fig. 1 Fig. S6b-e) were observed on the polymer layers compared to control. Cell death analysed at T0 and T3 was very low, with no differences between CTRL, P3HT:PCBM and PFO ( Supplementary Fig. S6f,g). These results indicated no toxicity of all the polymer layers analysed. Thus, a pixelated retina system based on M-cones (P3HT), S-cones (PFO) and rods (P3HT:PCBM) www.nature.com/scientificreports/ was printed, providing a dichromatic inkjet-printed artificial retina model. The three-polymer bio-hybrid printed device mimics the dichromatic vision of most mammals including rodents often used in biomedical studies 67 such as mice that only possess S-and M-cones 68 . Furthermore, it delineates a starting point for a future trichromatic human artificial retina system. Indeed, the device-concept can be easily extended to other polymers. For instance, a possible candidate to mimic human L-cones (maximal absorption at λ = 560 nm) could be Poly[N-9′-heptadecanyl-2,7-carbazole-alt-5,5-(4′,7′-di-2-thienyl-2′,1′,3′-benzothiadiazole)] (PCDTBT) which has an absorbance spectra maximum at 576 nm. In this study we assembled a simplified model of the anatomical retinal scheme consisting in a concentric ring design where each annulus represents a different retinal photoreceptor region. This simplification made it possible to selectively shadow-mask the different annuli in order to test the spatial and spectral response of our pixelated device. The outermost annulus of 360.5 mm 2 corresponds to rods (P3HT:PCBM), whilst the two central annuli correspond to the M-cones (P3HT) of 12.6 mm 2 and the S-cones (PFO) of 7.1 mm 2 areas respectively. Total diameter was 22.0 mm. Inkjet printing was used for selective deposition of the polymers (Fig. 4d) being a powerful technique that offers excellent design flexibility and precise control in material deposition 69 , enabling contactless, mask-free and digital patterning minimising materials usage and waste produced.
Semiconducting polymer inks formulation (molecular weight, concentration, solvents, viscosity and surface tension) and inkjet deposition represented an important part of creating a complete colour-sensitive artificial retina system. High viscosity, nozzle clogging, agglomeration, precipitation, and uncontrollable drying patterns are among the frequently challenges encountered with inkjet printing 70,71 . PFO, P3HT and P3HT:PCBM were dissolved in a mixture of chlorobenzene (CB) and tetrahydronaphthalene (THN) (1:1) to obtain inkjet-printable polymer inks. The combination of CB:THN-medium/high boiling point solvent mixture-serves two purposes. CB was used to achieve homogeneous film formation owing to excellent solubility of the chosen polymers in CB. THN (b.p. = 270 °C) was used to prevent the nozzles clogging and to provide reliable jetting over time. Inkjet parameters (pulse voltages, cartridge temperature, substrate temperature, drop spacing, waveform) were carefully adjusted in order to obtain stable droplet formation and to achieve rounded-pixels layout (see "Methods" section). The artificial retina system was printed on Glass|FTO substrates. The inkjet-printed device consisted of ~ 42,100 pixels over the total area of 380.1 mm 2 . Optical microscope images show the concentric layout of The printed pixelated device is a starting point for higher pixel density artificial retina models. Indeed, to reduce the pixel size, it is possible to use additional layers to modify the wettability and surface energy of the substrate. An example is shown in Supplementary Fig. S7, where a spin coated P3HT:PCBM layer was used as a layer on which to print P3HT and PFO pixels that enabled us to obtain small diameters of ~ 50 μm. A different underlying layer with negligible optical absorption in the visible range would be suitable for this purpose. In general, inkjet printing equipment with smaller nozzle orifices/small droplet volume and higher viscosity inks along with substrate chemical surface functionalisation, can enable the realisation of pixel sizes smaller than 30 μm 72 . During printing from a single nozzle, individual sub-nanoliter ink droplets are ejected. The resulting drop diameter is related to the size of the nozzle orifice (1 pL drop can create a 20-30 μm spot size) and final, dry, printed area size resulting from a single drop, can be adjusted significantly by the operating conditions 73 and the surface treatment of the substrate. In order to approach the 2-10 μm size, typical for human photoreceptors 74,75 , alternative printing technologies can be adopted such as imprinting 76,77 , flexographic printing 78 , laser patterning 79 or high resolution lithography techniques (optical and electronic).
To the best of our knowledge, no visual prosthetic devices can restore the natural colour vision to date. Our proposed concentric artificial retina model is simplified in order to enable demonstration of spatial resolution of the photo-response. Human cones are distributed radially from the fovea in a geometry that is unique from individual to individual. Even if it was not possible to achieve the ideal case of superimposing pixelated artificial photoreceptors with the same dimension and in the same position as that of the real retina it wants to mimic (e.g. for implantation), there remains possibilities for inducing and possibly restoring some trichromatic vision to the blind. For example, it would be convenient to identify local density variations of cone types across a region of the retina and to place the larger pixels of one type of polymer over the areas where there are larger concentrations of the homologous cones. The avenues for exploration and increasing understanding of the visual system are many.
The artificial retina device was completed by assembling a top transparent platinized counter-electrode with a 60 µm thick thermoplastic spacer and filling the space with PBS. It was then characterised by placing it in a customized black box and illuminating the whole active area (380.1 mm 2 ) via a train of 6 light pulses using a LED lamp, 6 mW/cm 2 at 5 cm distance (standard warm white light spectrum, Supplementary Fig. S8). Chronoamperometric J ph curves (impulses of 2 s light ON, 2 s light OFF) are consistent with those measured for the prototypical single layer device (Fig. 2a,b), showing a capacitive switch and a more steady contribution to the current with J ph ranging from 0.2 to 0.3 μA/cm 2 (Supplementary Fig. S9a). On average responses were 7% greater  Fig. S9b). Photovoltage data were recorded with a train of white light pulses of 5 s duration followed by 60 s dark period. A significantly high V ph of 12.0 mV was recorded after a 5 s long light stimulus ( Supplementary Fig. S9c) (it was ∼0.3 mV and ∼1.1 mV after 10 ms and 300 ms respectively, thus consistent with short light pulse stimuli).The darkness period was extended to 60 s to enable V ph to recover within 10% of its initial value. It would be interesting to evaluate the optoelectrical responses of a single polymer pixel in order to assess whether the one-polymer pixel/one-photoreceptor stimulation is possible. A low-level electrical measurement set-up with a current resolution of less than 1 pA would be required for this analysis. Thus, this will be part of a future investigation. With our current systems we were able to gauge the spectral and spatial responsivity of the different polymer regions of the pixelated device by masking/unmasking larger annuli regions (see "Methods" section). The resulting photocurrent (I ph ) as a function of wavelength of the light source is shown in Fig. 6 with the same measurements carried out on each of the single polymer (or blend) annulus of the printed artificial retina. Customized shadow masks were used to exclusively illuminate the polymer annulus of interest. Notwithstanding the geometrical domination of the P3HT:PCBM component (PFO 2.0%, P3HT 3.3%, P3HT:PCBM 94.7% of total inkjet-printed artificial retina area), each pixelated polymer annulus contributes to the overall I ph signal (Fig. 6, Supplementary Fig. S10). The pronounced red peak in the P3HT:PCBM spectrum can be due to various effects including layer thickness 80 , solvent used 40 , thermal treatment 81 or the interaction with materials such as dopants 82 as well with components of the electrolytic medium.

Discussion
The main focus of this study was the development of a conjugated polymer-based bio-hybrid artificial retina model to study photo-response properties of different band-gap polymer pixels, interfaced with electrolyte medium, aiming to mimic the colour response of mammalian cones and rods, from the single-polymer-layer device concept to a fully pixelated system. The device layout consists of two transparent, planar and parallel electrodes: the P3HT-spin-coated photosensitive working electrode and the Pt-screen printed counter-electrode. Planar polymer layer for retinal opto-electrical stimulation have been already successfully demonstrated in literature, both as polymer thin films spin-coated on conductive glass substrates and as polymer-coated MEAs electrodes [12][13][14][15] . In these works patch-clamp 27 , contact pad 14 , and MEA 12 electrodes are interfaced with electrolyte in an open container configuration with a metal counter-electrode immersed in the electrolyte. Distinctively, the presented bio-hybrid device enables the study of the spectral and spatial responses of the substrate/polymer system interfaced with biological electrolyte within a closed architecture. Consistently with other works on conjugated polymers 12,13,15,29,32,33,46 , the fabricated bio-hybrid photo-electrode shows good biocompatibility. Thus, the device can represent a useful tool for studying light stimulation and recording of bio-electrical signals. The fabricated bio-hybrid device response to incident light elicits transient photo-voltage (mV) and photo-current (µA/cm 2 ) outputs that resemble those seen in patch-clamp and MEA experiments and that would suffice to elicit a response in a retina 83 or in neurons 84 . Even if capacitive coupling (0.51 ± 0.28 µA/cm 2 and of -0.29 ± 0.12 µA/ cm 2 spikes at the beginning and the end of light stimuli, 17.8 mW/cm 2 ) is the dominant process in the device upon illumination, a minor faradaic component (0.17 ± 0.10 µA/cm 2 ) is present. Clearly charge balancing is a vital process to ensure the electrode-tissue interface safety operation. However, we think that self-regulation of ionic www.nature.com/scientificreports/ gradients in living systems could help to prevent tissues and cells damage over time 85 . Thus, we can accept the recorded current density signals and faradaic component values with a good margin of tolerance also in respect to similar work found in literature where faradaic processes were not completely excluded [13][14][15] . The bio-hybrid device presents some advantages compared to electrophysiological investigative systems: easy-to-handle and transportable closed system, controllable size, and a required small amount of electrolyte. It permits the utilization of tools from an electronic engineering/physics/chemistry laboratory rather than the more sophisticated tools found in bio-medical laboratories. In addition, the device architecture opens the possibility of inserting biological materials (cells or tissues) in the chamber of a compact device filled with the biological electrolytic medium. Limitations of this device compared to the open immersions used in typical MEA or electrophysiology experiments, may be represented by the constraints in inserting fairly large tissues, such as retinas, in the small chamber for their study, as well as the large dimensions of the current electrodes used (see Fig. 1a,b). The latter may need to be patterned in a smaller size (i.e. similar to those found in MEAs, i.e. of the order of ≤ 0.1 mm) and be more numerous for better efficiency to record bioelectrical signals from local areas of live retinas. It will also be critical to devise ways to keep living samples functioning in a restricted environment which may require the introduction of inlet and outlet orifices for perfusion. Until these evolutions are implemented and tested, the device can be useful to spatially and spectrally investigate photo-responses of semiconductor films or patterns immersed in physiological fluids that mimic those found in living organisms. The device was also fabricated via inkjet printing technique. Thus, an inkjet-printed colour-sensitive artificial retina device consisting of three different types of semiconducting round polymer pixels, with distinct absorption spectra, mimicking the chromatic sensitivity of photoreceptors in the eye and interfaced with a physiological medium was demonstrated. The number of pixels was 42,100, the density of the artificial photoreceptors was ∼11,000 pixels/cm 2 and the corresponding spatial resolution was 267 dpi (dots per inch), with pixel diameters of 95 ± 5 μm. Printing technologies enable placement of different materials in the locations of choice. In the future, with higher resolution techniques being developed (i.e. ≤ 10 µm), one could first scan/ image an individual retina and then print the pixels spatially where exactly one wants to place them and this can be done for each individual retina (more costly if one uses photolithographic techniques). Secondly, printing techniques enable deposition on uneven or curved surfaces which can be beneficial for artificial retina concepts. The polymer pixelation allows a light-sensing device to spectrally and spatially resolve incoming light providing photo-responses that are both chromatically and spatially sensitive. The optoelectrical characterization of a single polymer pixel awaits further investigation. Future studies can be devoted in replacing glass with flexible substrates such as PET, poly (dimethylsiloxane) (PDMS), and silk fibroin films, the latter of which are implantable in vivo 46 . Moreover, additional studies on the device biocompatibility and its possible integration with healthy or degenerated explanted retinas from animal models are necessary to demonstrate future optoelectronic applications in the field of retinal prostheses.

Methods
Substrates preparation for bio-hybrid interface. Transparent Glass|FTO substrates (2.2 mm thickness) were cut in 2.5 cm by 2.5 cm and cleaned by subsequent rinses in ultrasonic bath using deionized water, pure acetone, isopropyl alcohol (5 min each). Then dried with an air gun.
Polymer formulations for bio-hybrid interface. rr-P3HT (Rieke Metals, Inc.) (regio-regularity ≥ 95%, molecular weight: 51,000 g/mol) was used without any further purification. Polymer solutions in chlorobenzene (30 mg/mL) were prepared by heating polymer and solvent mixture at 100 °C for 5 min and then stirring at 50 °C overnight. rr-P3HT-chlorobenzene solution was spin-coated on FTO glass substrates (2000 r.p.m., rotation duration 60 s) inside the glovebox (Mbraun Inc.). Organic thin films were annealed on a hotplate (120 °C, 2 h). To obtain a square active area of 0.8 cm by 0.8 cm, the excess of polymer was removed using p-Xylene solvent and a polymer protective film mask.
Electrolyte compartment for bio-hybrid interface. Thermo-plastic sealant (Surlyn, 60 µm thickness, DU PONT) masks, creating the milli-volume chamber, were designed to cover the frame of the 0.8 cm by 0.8 cm active area on the device and cut using a cutting machine (Small-one, magicut cutting plotter). Two orifices placed in opposite sides were designed to fill electrolyte inside the device. Masks were pre-attached on the substrates via pneumatic heat press (model Special) (90 °C, 0.5 bar pressure).

Counter-electrode for bio-hybrid interface. A transparent platinum layer was screen-printed on
Glass|FTO substrates (2.5 cm by 2.5 cm). After platinum paste precursor (3D-nano) deposition, the wet platinum layers were processed by a firing process (30 min, 480 °C).

Electrolyte solution.
Phosphate buffered saline solution (Gibco) was used as electrolytic solution. PBS is a balanced salt solution containing: CaCl 2 (0.9 mM), MgCl 2 (0.5 mM), KCl (2.6 mM), KH 2 PO 4 (1.5 mM), NaCl (137.9 mM) and Na 2 HPO 4 (8.0 mM) dissolved in double distilled H 2 O. PBS was filled inside the device using a 1 mL syringe by the orifices present in the Surlyn masks. Devices were then closed using a biphasic glue.
Inkjet-printed artificial retina fabrication. FTO glass substrates (2.5 cm by 2.5 cm) were treated with a standard cleaning process as described before. They were then treated with Argon/Oxygen plasma (100 W, 10 min) and UV treatment (1000 W, 5 min) to improve wettability and hydrophilicity. A model design reproducing the photoreceptors distribution on the human retina was designed and drawn via a standard vector   www.nature.com/scientificreports/ and Bonferroni post-analyses. The level significance was set at P < 0.05 (*, a), P < 0.01 (**, b), P < 0.001 (***, c) and P < 0.0001 (****, d).

Measurement setup.
Three measurements set ups were used. ARKEO apparatus (Cicci Research s.r.l.) was used to opto-electrically characterize the bio-hybrid interface. Transient photovoltage (TPV) and charge extraction (CE) software routines were used. Light stimuli (from the bottom of the device) of 10 and 300 ms duration after 100 ms of dark were presented every 60 s with increasing light intensities (17.8, 26.6, 35.3, 43.9, 52.4, 60.9, 69.3, 77.4, 85.7, and 93.8 mW/cm 2 ). Each measurement was done immediately after the electrolyte injection in the device. Plotted signals are the mean over 3 samples. Each measurement represents an average of twenty consecutive sweeps (10 kHz and 1 kHz sampling rate for 10 ms and 300 ms light stimuli pulse duration).
In TPV recordings a baseline potential was removed to evaluate the potential difference (from dark to light). A 6 W LED lamp (6 mW/cm 2 , at 5 cm distance from the lamp, as measured with ThorLabs PM100D broadband power meter) driven by a microcontroller (schematics and connection in Supplementary Fig. S11) was used for pixelated device characterization. A train of 6 light pulses (alternating periods of 2 s light ON and 2 s OFF for J ph measurements, and 5 s ON and 60 s OFF for V ph measurements) was used. A Keysight precision currentvoltage analyser (Keysight technologies) was used to measure the J ph and V ph signals from the inkjet-printed artificial retina device (5 Hz sampling rate). I ph and V ph across different wavelengths were recorded by using a light source filtered through a monochromator before been focused on the polymer film/inkjet-printed artificial retina device. Data were collected at 2 nm intervals, with scans conducted from 300 to 800 nm. Monochromatic light was narrow-band, with approximately 3 nm FWHM. The light source was chopped by a mechanical chopper at 70 Hz, whose reference signal was fed to a lock-in amplifier (0 db gain). All measurements were taken at room temperature. To examine if the electrical signals drift over time, a device-tester with a P3HT thin film and PBS was stimulated via a 530 nm wavelength light chopped at 70 Hz for ca. 2 min. Monochromator/lock-in set up was used to record electrical signals. No evident dependence on time was observed. J ph , I ph and V ph generated were measured between platinum electrode (counter-electrode) and polymer electrode (photo-electrode) in short circuit (V applied = 0 mV) and open circuit (I applied = 0 mA) respectively. The photoelectrode was grounded.
The power and spectra of ARKEO fast speed LED (5000 K) were measured using a spectrometer (AVASPEC) and an integrating sphere (AVASPHERE-50-IRRAD). The 6 W LED lamp spectra were measured using an Ocean Optics Optics HR2000 + high resolution spectrometer (Ocean Optics, Inc.).
Data were analysed with OriginPro 2016.

Data availability
Authors declare that all relevant data supporting the findings of this study are available in this paper and in its Supplementary Information file. Access to our raw data can be obtained from the corresponding authors upon reasonable request.
Scientific Reports | (2020) 10:21457 | https://doi.org/10.1038/s41598-020-77819-z www.nature.com/scientificreports/ Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creat iveco mmons .org/licen ses/by/4.0/.