An Inkjet‐Printed Field‐Effect Transistor for Label‐Free Biosensing

A flexible, biological field‐effect transistor (BioFET) for use in biosensing is reported. The BioFET is based on an organic thin‐film transistor (OTFT) fabricated mainly by inkjet printing and subsequently functionalized with antibodies for protein recognition. The BioFET is assessed for label‐free detection of a model protein, human immunoglobulin G (HIgG). It is characterized electrically to evaluate the contribution of each step in the functionalization of the OTFT and to detect the presence of the target protein. The fabrication, structure, materials optimization, electrical characteristics, and functionality of the starting OTFT and final BioFET are also discussed. Different materials are evaluated for the top insulator layer, with the aim of protecting the lower layers from the electrolyte and preserving the BioFET electrical performance.


Introduction
Thin-fi lm transistors (TFTs) are powerful devices in semiconductor manufacturing and form the basis of countless electronic devices, such as memory chips, [ 1 ] photovoltaic cells, [ 2 ] logic gates, [ 3 ] and sensors. [ 4 ] Most contemporary TFTs are based on inorganic materials (mainly silicon). An interesting alternative to inorganic TFTs is organic TFTs (OTFTs), which offer the possibility of mass production by using the conventional printing technology and working with low-cost materials, resulting in an especially good approach to obtain functional as it entails rapid, additive manufacturing with small volumes of ink to achieve high pattern precision, resolution, and reproducibility. Furthermore, inkjet printing is better suited for rapid prototyping in research than are technologies that require the use of masks. [ 25 ] It can be used to deposit a wide range of materials (electrical, chemical, optical, biological, etc.) of different compositions (soluble liquids, dispersions of microor nanoparticles, melts, and blends) and viscosities. Lastly, inkjet printing is amenable to deposition of certain compounds (e.g., conductive and/or semiconducting polymers, and large biomolecules) that can only be deposited using a solution-based technique, as they cannot be deposited by conventional vacuum-deposition techniques.
Among the most important applications for biosensors is medical diagnosis. Researchers are endeavoring to create highly effi cient, low-cost, point-of-care biosensors that enable earlier diagnosis (and therefore, faster treatment) of ill patients. [ 26 ] The biosensors for this use are typically based on protein-analysis immunoassays, which can require the use of a label [ 27 ] or be labelfree. [28][29][30] In the former type, labels such as fl uorescent dyes, [ 31 ] radioisotopes, [ 32 ] or quantum dots [ 33 ] are used to indirectly quantify the target protein. In contrast, in label-free immunoassays some inherent property of the primary transducing signal (i.e., the mass, conductivity, or insulator characteristics) is measured.
Herein are reported the design, fabrication, and evaluation of a fl exible BioFET for label-free biosensing. The BioFET is based on an inkjet-printed OTFT whose insulator was biofunctionalized with specifi c antibodies for protein recognition. As proof-of-concept, the resulting BioFET was evaluated for quantifi cation of a model protein, G (HIgG).

Operating Mechanism
The BioFET reported here, made by functionalizing a previously described OFET with anti-HigG antibodies, ensures the electrical detection of biomolecular interactions between the target protein and the antibody. These interactions occur on the BioFET dielectric through capacitive coupling with the organic semiconductor. A schematic of the BioFET is shown in Figure 1 A. The structure is a dual-gate OFET that comprises source, drain, and two gates (Figure 1 A-i,ii,iii). However, in this BioFET, rather than a top gate contact, there is an electrolyte solution (phosphate buffer saline [PBS]), an antibody-protein conjugate and a reference electrode (Pt wire; 0.2 mm diameter). The accumulated charge on the insulator in the BioFET can lead to a variation in the source-drain current. Since the change in the electrical charge is related to the interaction between the target biological molecules and the semiconductor, the BioFET can be used to monitor this interaction effectively by measuring, with high sensitivity, the variations in transistor response ( I -V curves).
The developed BioFET has two main advantages. Firstly, it is easy to miniaturize, making it suitable for future applications involving complex systems. Secondly, and in terms of biosensing, it can be equipped with a bottom OFET for using as test component, to confi rm the reliability of the organic semiconductor. In fact, the dual-gate structure could be used for current control (although in the present work, this was not done). Photographs of the printed OTFTs in the BioFET are shown in Figure 1 B.

Considerations for the BioFET Fabrication
The BioFET was fabricated by inkjet printing. The printing parameters for printed BioFETs can be tailored to defi ne optimal morphologies according to the intrinsic constraints of the materials. To explain the relationship between layer morphology and electrical performance, one must consider all the variables that affect the transfer characteristics in the saturation regime of an OFET, as shown below in Equation 1 : where I DS is the drain-source current (A); µ is the fi eld effect mobility (cm 2 ·V −1 ·s −1 ); C i is the capacitance per unit area of the insulating layer (F·cm −2 ); W is the channel width of the transistor (µm); L is the channel length of the transistor (from source to drain, in the direction of the current fl ow [µm]); V G is the gate voltage; and V τ is the threshold voltage (V), which is the minimum gate voltage required for establishing the conductive channel. Thus, the most important printing parameters are related to the dimensions of the OFET, including the thickness of each layer. Another important factor in printing is the type of printer used. Drop-on-demand (DoD) piezoelectric printers, such as the one employed in the work reported here, use a piezoelectric transducer to eject the ink droplets. The transducer is actuated by a voltage pulse whose frequency ranges from 1 to 20 kHz (in commercial printers).
The relative distribution of deposited material across the printed geometry can be controlled by tuning certain printing parameters related to temperature or to drop-on-demand delays, such as the temperature of the cartridge or platen (in this work, the temperature of cartridge was disabled), [ 34 ] the jetting frequency, the jetting waveform, or the voltage of the print-head nozzles. Other important printer parameters that can be adjusted to achieve better features include drop spacing (DS), the distance between two consecutively printed droplets), which is scaled by the wetting diameter of the droplet on the substrate. The resolution of the fi nal pattern is determined by the drop spacing. The nozzle orifi ce size of the printer used for all layers was 21.5 µm (corresponding to a 10 pL printhead).
One crucial parameter of the ink itself is its viscosity, as the state of the droplet strongly infl uences the morphology of the deposited fi lm. Higher viscosities enable better control over droplet spreading, but lower viscosities (range of cPa) are needed to ensure that the print-head nozzles can eject the ink. Ink viscosity is usually adjusted by modifying the solvent or the concentration of the solute in the solvent.
Before the printer is set up, and the ink and printing parameter values are chosen, the substrate must be selected. Albeit the most commonly used materials are glasses or fl exible plastics, including polyethylene terephthalate (PET) due to its low cost and polyethylene naphthalate (PEN) for its low surface roughness, a wide variety of substrates, including paper and even biological tissue can be used. PEN and PET roughness were analyzed by atomic force microscopy (AFM). In Table S5 and S6 (showing the properties of these two materials in comparison with others such as PI, PEI, and PSS; see the Supporting Information (SI)) indicate the measured values of root-mean-square (Rms) and average (Ra) roughness for each substrate (PEN and PET).
In this work, PEN substrate is used due to its many convenient properties such as strength, heat resistance, and dimensional stability, among others. This material is specially appropriated for printed electronics, since it has a pretreated smooth surface with a roughness around 0.82 nm, being in the range of silicon wafers surface roughness.
Finally, the drying of the ink onto the substrate is also important-namely, the physical and chemical interactions between the ink solution and the substrate during evaporation. This factor is especially important when there is a recirculation fl ow (induced by a Marangoni fl ow effect) opposite to the convective fl ow, [ 35 ] a scenario which is the leading cause of the so-called "coffee ring effect" Both bottom gate and drain and source electrodes show a low coffee ring effect while the cross linker crosslinked poly-4-vinylphenol (cPVP) presents a clear accumulation of the material at the edges. Figure S1 shows the profi le of the Ag bottom gate, bottom cPVP, drain, and source electrodes and FS0027 organic semiconductor respectively. Both bottom gate and drain and source electrodes show a low coffee-ring effect while the cPVP presents a clear accumulation of the material at the edges.

Characterization of the Insulator Layer
As previously explained, the fabricated BioFET comprises a dual-gate OTFT structure. In previous work by Ramon et al., [ 36 ] the structure of the OTFT was optimized and its constituent materials were tested. The conductive layers are made of silver nanoparticles, the insulator layer is made from cPVP, and the semiconductor layer is made of the amorphous p -type polymer FS0027. However, for most biosensing applications that use BioFETs, the top insulator fi lm should also act as a coating layer to protect the device, since it is in contact with an aqueous solution. In the BioFET reported here, silver nanoparticles were used for the source and drain contacts. Since silver nanoparticles are electrocatalytic in aqueous media, [ 37 ] it is very important to insulate the source and drain contacts from the electrolyte. Insulator had to be analyzed for such purpose.
Various materials were explored for the top insulator (i.e., the insulating interface between the semiconductor and the electrolyte). The fi rst studied material was cPVP, which offers good printability and insulation for an OTFT. However, cPVP dissolves in aqueous media, [ 38 ] consequently a fast decomposition reaction was triggered upon application of a sweep voltage in the electrolyte. cPVP is not stable during the conjugation steps due to its poor resistance to the humidity; however, cPVP can support electrical fi eld of up to 0.3 MV cm −1 in dry environments ( Figures S2A,B in the SI). A simple test for the insulator material pre-selection was done, which has to be biocompatible, amenable to use in the desired BioFET, easy to process and suitable for mass-production. A schematic of the metal-insulator-electrolyte (MIE) structure used to characterize the different insulator materials is shown in Figure S3A.
The next three materials separately evaluated for the top insulator layer were Norland Optical Adhesive (NOA) 68, a polyester resin and polymethyl methacrylate (PMMA). Each one was deposited onto the silver contact, and then put into contact with the electrolyte. A voltage sweep (0 to -20 V) was applied, and the surface characteristics of each material were studied by optical microscopy and by current measurements. In all three cases application of the voltage led to soft breakdown produced by many pinholes between the reference electrode and the silver contact, which led to high current density and consequently, to a total loss of insulation (Figures S3B,C, and D). Therefore, all three of these materials were ruled out.
The last material evaluated as possible top insulator was MMAcoMAA (methyl methacrylate copolymer methacrylic acid). Despite never having been reported as an insulator for organic electronics, MMAcoMAA and related compounds form the basis of many positive resists. They are distinguished by their high-resolution features, strong adhesion to metallic or semiconducting supports, and good resistance to solvents and reagents. Thus, MMAcoMAA was subjected to various electrical measurements (capacitance, resistance, and current) with the objective to characterize its electrical behavior and determine its constraints-namely, in terms of achieving the lowest current leakage and the greatest resistance to the aqueous medium possible.
Once MMAcoMAA had been chosen as the material for the top insulator in the BioFET, two additional factors were optimized: the number and thickness of the layers to be used. Current-voltage ( I -V ) measurements were taken by applying a bias voltage to the top of the MIE structure, using aq. KCl (0.1 M ) as electrolyte. The current density plot for different numbers of layers, and different layer thickness values, is shown in Figure 2 . The best results were obtained for one layer of 1.5 µm thickness. Thus, a layer of MMAcoMAA was chosen as the top insulator in the BioFET.

SEM Characterization of Each Layer in the BioFET
The morphology of each layer in the BioFET was characterized by optical microscopy and by scanning electron microscopy (SEM).

Figure 3 A and B show optical microscopy
images of the top view of each layer (top row) and SEM images of the cross-section for each layer (bottom row), the latter of which were used to measure the real thickness of each layer: i) the bottom gate, based on silver nanoparticle ink (thickness: 573 nm); ii) the bottom insulator, made of cPVP (thickness: 650 nm); iii) the source and drain contacts, printed with silver nanoparticle ink (thickness: 328 nm; note: due to the wettability of the cPVP, this layer is thinner than the silver bottom-gate); iv) the ( p -type) semiconductor, FS0027 (thickness: ca. 300 nm); and, fi nally, v) the top MMAcoMAA insulator (thickness: 1.5 µm). Figure 3 C shows a SEM image of a focused ion beam (FIB) cross-section of the BioFET (cutting dimensions: 200 × 10 × 8 µm 3 ) that had been subjected to an extra polishing step to maximize image quality.

Electrical Characterization of the OFET
In any transistor, the control of the electrons or holes from source to drain depends on the size and shape of the channel and on the applied gate voltage. When p -channel devices are in accumulation mode , the applied voltage between gate and source (negative) produces an accumulation region in which the channel is formed by the charge carrier that is produced in the gate insulator/semiconductor interface, thereby enabling the current to fl ow (ON state). In contrast, when the device is in depletion mode , the applied voltage between gate and source (positive) closes the channel, increasing the resistance from source and drain and consequently, stopping the fl ow of current (OFF state). In n -channel devices, the positive voltage attracts electrons toward the gate, forming a conductive channel. The voltage necessary to open the conduction channel is known as the threshold voltage ( V τ ). [ 39 ] Most OTFTs operate in accumulation mode, whereas most inorganic TFTs operate in inversion mode. Conductance occurs in the ON state, due to the layer of charge carriers that are formed in the semiconductor within a few angstroms from the gate insulator/semiconductor interface, and only upon application of a suitable V G . Thereby, OTFTs operate through the creation and elimination of a thin layer of charge carriers at the gate insulator/semiconductor interface.
In order to have an approach of the leakage process of the bottom OFET, we fabricated a simple device consisting of a metal-insulator-metal (MIM) capacitor. All experimental conditions such as the material, printing parameters, and posttreatment of the cPVP are exactly the same as those for the BioFET fabrication. This dielectric withstands an electric fi eld of 0.3 MV cm −1 with a current density of 8 nA cm −2 , moreover, the degradation of the dielectric upon bias stress is not observed.
Some interesting high-performance semiconductors, including 6,13bis (triisopropylsilylethylthynyl) pentacene (TIPSpentacene), amorphous polymer FS0087 provided by Flexink Inc. among others, were tested. In terms of morphology, these mentioned semiconductors lacks of homogenous surface basically due to high crystallization while deposited by inkjet printing technology. [40][41][42] In terms of electrical behavior, high motilities and low threshold voltage were reached thanks to the use of fl uoropolymer [ 43 ] dielectrics but they are very hydrophobic and require conventional techniques for their biofunctionalization. TIPS-pentacene (commercially available) was discarded since (i) it is not appropriate when high-temperatures are required in the materials deposited on top because it suffers destructive structural changes when the temperature is higher than 120 °C; since we bake MMAcoMAA material at 150 °C after the semiconductor deposition, we would be cracking and losing electrical properties of the semiconductor placed on the bottom; (ii) formation of very rough fi lms surfaces due to the crystallization when inkjet printing technique is applied using both 1% wt. in chlorobenzene and 3% wt in anisole as solvent [ 44,45 ] (see Figure S8 and S9 in the SI). Finally, FS0027 was selected for its stability along the time, good printed morphology and low threshold voltage.
Since the electrical properties are essentially determined by the interface between the dielectric and the semiconductor, a metal-insulator-semiconductor (MIS) structure was fabricated using the same materials, printing conditions, and post-treatment (as for BioFET fabrication). MIS devices are very promising because they can provide valuable information about the quality of the dielectric/semiconductor interface. The MIS capacitors where also characterized by using small signal impedance techniques mainly, capacitance-voltage ( C -V ). The electrical characterization shown in Figure S10 reveals that both accumulation and depletion regions can be established. The voltage at which the accumulation region starts is -0.9 V and indicates that there is not a deep trapping phenomenon in the interface dielectric/semiconductor. The electrical characteristics for each OFET as a function of the bending radius were determined with a convex direction radii R of approximately 1.4, 1.6, 2.8, 3.5, and 5.5 mm, corresponding to strains ( S) of approximately 4.5%, 4%, 2.2%, 1.8%, and 1.1%, respectively ( S = d /2 R , where d is the thickness of the substrate); thus, for higher values of S , the bending curvature is lower. Figures 4 C and D show the transfer characteristics and the mobility of the OFET as a function of bending radius. Before bending, good average mobility was observed according to our technology. As the strain induced increases, the mobility on the interface cPVP-semiconductor decreases approximately 90% (from 1.64 × 10 −4 to 1.65 × 10 −5 cm 2 ·V −1 ·s −1 for a strain such as 4.5% corresponding to a radius of 1.4 mm. Nevertheless, for low strains from 1.1% up to 4%, the decrease of the Adv. Funct. Mater. 2014, 24, 6291-6302 www.afm-journal.de www.MaterialsViews.com mobility is not signifi cant. In contrast, the threshold voltage increases approximately 10 V for higher strain values. This result is in accordance with the literature. [ 46,47 ]

Functionalization of the MMAcoMAA Surface
The surface of the MMAcoMAA layer was amino-functionalized by treatment with 1% (3-aminopropyl triethoxysilane) APTES. The fi nal surface density of amino groups in the functionalized layer fi nal was 0.18 µM·cm −2 ( Figure 5 A), as determined by colorimetric quantifi cation with acid orange II dye.
The presence of surface amino groups on the MMAcoMAA layer was confi rmed by various methods, including binding of the fl uorescent label bovine serum albumin (BSA) labeled with the ALEXA 555 fl uorophore to the amino groups and subsequent fl uorescence quantifi cation by epifl uorescence microscopy (Figure 5 B). Figure S5A shows the change in contact angle of the MMA-coMAA. First, the angle decreased from 74° to 13°, due to the appearance of more-hydrophilic functional groups after the plasma treatment. However, the contact angle increased upon amino-functionalization of the surface, reaching values from 46° to 75°, depending on the concentration of APTES used (0.1%, 1%, 2% or 4%). Finally, the bare and amino-functionalized MMAcoMAA surfaces were each incubated with BSA, and then studied by AFM for changes in phase angle. The greatest change in phase angle was observed in the aminofunctionalized MMAcoMAA surface. Although proteins such as BSA can easily attach non-specifi cally to almost any surface, in this application there was a clear difference between the bare MMAcoMAA surface and the amino-functionalized one ( Figure S5B in the SI).

The EIS
Since the electrolyte-insulator-semiconductor (EIS) structure is a precursor of the BioFET, it was used to evaluate detection of an immobilized model protein, BSA (1 µg mL −1 ), as shown in Figure 6 A. The EIS comprised, from bottom to top: a conductor (Ag nanoparticle ink); a semiconductor (FS0027); an insulator (MMAcoMAA); an electrolyte (PBS); and a reference electrode (Pt wire). Two EIS structures were tested by C -V measurements (1 kHz, room temperature): one with an amino-functionalized MMAcoMAA surface and one with a BSA-functionalized MMAcoMAA surface (made by amino-functionalization and subsequent incubation with BSA). This technique enabled quantifi cation of parameters such as surface-state response, surface accumulation, and layer thickness. [ 48 ] The C -V characteristics of the two EISs are shown in Figure 6 B-i (aminofunctionalized EIS) and Figure 6 B-ii (BSA-functionalized EIS). In each case, the capacitance depended on bias voltage and showed two regions of accumulation/depletion, indicating the modulation of the carriers. The capacitance remained constant under forward bias; although above the threshold, it gradually increased to reach the maximum peak value. This phenomenon can be explained by short-circuited parallel capacitance and is based on an RC model circuit. As shown in C -V plots in

Electrochemical Evaluation of the OTFT/BioFET in Three Confi gurations
The OTFT/BioFET was electrochemically assessed in three confi gurations.  a drop of the electrolyte placed onto the MMAcoMAA using the confi guration shown in Figure 7 A-ii. In this case, the electrolyte is assumed to induce pronounced hysteresis, thereby increasing the OFF-current and consequently, decreasing the I ON / I OFF ratio. The transfer characteristic for the BioFET (Figure 7 B-iii) using the confi guration shown in Figure 7 A-iii reveals a decrease in current, which is caused by the differences in insulator thickness and in the permittivity between the bottom insulator (cPVP) and the top insulator (MMAcoMAA). Lastly, Figure 7 C shows the repeatability of the transfer curve for the BioFET confi guration.
Once the BioFET confi guration in which an electrolyte was on the top gate had been evaluated, the steps of functionalizing its MMAcoMAA insulator with specifi c antibodies ( Figure 8 A) were assessed by measuring the change in current after each one. Figure 8 B shows the transfer characteristic for each step: i) amino-functionalization (with 1% APTES); ii) immobilization of the capture antibodies; iii) blocking to prevent non-specifi c absorption; and, iv) analyte conjugation. Figure 8 B-i shows the I -V curve for the amino-functionalized MMAcoMAA surface. Immobilization of the antibodies onto the surface led to a slight decrease in current (Figure 8 B-ii); blocking of the surface with buffer (PBS supplemented with casein and tween) led to another, albeit negligible, decrease (Figure 8 B-iii); and, fi nally, incubation of the surface with the antibody led to a signifi cant decrease in current, due to detection of HIgG at the conjugation concentration (100 ng mL −1 ; Figure 8 B-iv).
The main characteristics of the detection capacity of a BioFET device is its inherent sensitivity to surface charge or surface potential, which can be quantifi ed by the transistor gain or transconductance. Thus, the gain performance of a BioFET in solution can be quantifi ed by replacing the gate voltage in the traditional transistor Gain defi nition Equation ( 2) , with V ref , which yields: Figure 8 C shows the transconductance for each of the aforementioned functionalization steps. The transconductance of the fully functionalized BioFET at HIgG (Figure 8 B-iv) detection was lower than that of the bare BioFET (Figure 8 B-i). Considering the lower capacitance in the BSA-functionalized EIS structure compared to that of the functionalized EIS (as shown in Section 3.1), the reduction in g m can be attributed to the changes in the effective capacitance due to the addition of organic layers, which by extension, applies to the BioFET. [ 49 ] As mentioned previously, the main limitation of the BioFET has been found to be the detection limit level of around micrograms per milliliter. In this work we present a fl exible BioFET capable of detecting 100 ng mL −1 (1.5 n M ) with a relative standard deviation (RSD) of 9% for BSA detection. Other works [ 50,51 ] employ fl uoropolymers such as perfl uor-1-3-dimethylcyclohexam (PFDMCH) and cyclic transparent optical fl uoropolymer (CYTOP) or inorganic copper phthalacyanine (CuPc) as a top insulators. Fluoropolymers are good candidates to withstand wet environment and also do not suffer from hysteresis during bias stress, etc. Despite of higher sensitivity, the main challenge of these materials is the subsequent functionalization, due to the high hydrophobicity of the materials, without using expensive and sophisticate techniques such as plasma-enhanced chemical vapor deposition (PECVD), thermal evaporation, etc.; Table S7 (in the SI) shows a comparison of the mentioned works. Therefore, it remains desirable to develop: i) biosensors based on alternative top insulator materials, such as copolymer MMAcoMAA where these limitations are lessened, Adv. Funct. Mater. 2014, 24, 6291-6302 www.afm-journal.de www.MaterialsViews.com and, ii) biosensors onto fl exible substrate benefi ting from lowtemperature organic semiconductors unlike of the rigid silicon substrates. To the best of our knowledge, this work joins organic materials, fl exible substrates and fi nally low-cost fabrication thanks to inkjet printing technology providing in this way a proof-of-concept of a label-free BioFET easily scalable to mass production and accessible for low-resource places/countries.

Conclusion
A fully integrated organic BioFET was designed, fabricated and tested for label-free protein detection. It was made by inkjet printing of an OFET and subsequent functionalization of the insulator with specifi c antibodies. Several constituent materials and operational confi gurations for the OFET and the fi nal BioFET were evaluated electrically. Moreover, different materials and thicknesses for the insulator layer were assessed by insulator rupture evaluation. Both cPVP (for the bottom OFET) and MMAcoMAA (for the top BioFET) were validated as good insulators in the studied range of voltages. Biofunctionalization of the MMAcoMAA surface, and selective adhesion of target proteins to the BioFET, were studied by colorimetric, fl uorescence measurements and AFM images. For preliminary studies, an EIS structure was used to detect the protein BSA by C -V measurements. The repeatability was studied with different devices. Confocal and interferometry images were taken to assess the roughness of functionalized and immobilized protein structures. Finally, as proof of concept, the BioFET platform was tested for detection of the model protein HIgG (at a fi xed concentration of 100 ng mL −1 ) . The BioFET reported here represents an important starting point for the design and fabrication of fl exible, organic biosensing devices by inkjet printing. Nevertheless, certain issues with such devices remain to be improved, including reproducibility in the fabrication process (in terms of achieving greater sensitivity and therefore, a lower limit of detection), the development of functional inks and extension of this technology to a wider array of substrates. The authors are confi dent that once this technology has matured, it will be amenable to miniaturization for integration into a fully functional device for point-of-care diagnosis.
Instruments : A Dimatix DMP2831 desktop printer (Fujifi lm, USA) was used to print the layers. It employs one user-fi llable, piezobased inkjet print cartridge with 16 nozzles, able to jet droplets (1 or 10 pL) of diverse fl uids with a variable jetting resolution. Drop-ondemand printing was performed in a class 10 000 cleanroom. The printer includes an alignment system to prevent layer misalignment when the sample is moved for curing processes or for changing the cartridge. Electrical measurements were performed using a B1500A semiconductor analyzer and E4980A LCR meter (both: Agilent, USA). Finally, a Süss Microtec EP 6 probe station is used as a tested for device characterization.

Optimization of the Insulator Layer
In order to choose an insulator material that would be biocompatible and would not compromise the performance of the BioFET, different insulators were tested using an MIE structure based on capacitance measurements. The MIE structure comprised (from bottom to top): a conductive layer; a spin-coated insulator layer; a drop of electrolyte (acetate buffer pH 4.0 + PBS pH 7.4; total volume: 2 µL); and fi nally, a reference electrode (Pt wire; diameter: 0.2 mm). Each of the following materials was separately evaluated as insulator: NOA 68 (adhesive), polyester resin, cPVP, and MMAcoMAA. The external reference electrode was used to perform electrical measurements to study the leakage current through the insulator. The measurements included I -V curves, and parallel capacitance ( C p ) and resistance ( R p ) at 1 KHz.
Additionally, each layer was studied for morphology and thickness measurements by scanning electron microscopy (SEM), using a Magellan microscope (FEI, USA).

Fabrication of the BioFET
A BioFET comprising (from bottom to top): a bottom gate (silver nanoparticle ink); a bottom insulator (cPVP); source and drain electrodes (silver nanoparticle ink); a p -type organic polymer semiconductor (FS-0027); a top insulator (MMAcoMAA); and a reference electrode (Pt wire) in contact with the electrolyte. These technical specifi cations of each ink and the fabrication process are described in Tables S1-4 in the SI.
Deposition of the Bottom Gate : Prior to printing, the substrate (PEN foil 125 µm thickness) was cleaned with ethanol for a few seconds, and then dried under nitrogen to remove any remaining particles. The bottom gate was deposited onto the substrate by inkjet printing of a silver nanoparticle solution. The cartridge and platen temperature were set at room temperature and 40 °C respectively. Drop spacing of 20 µm by using a 10 pL cartridge was established, and the jetting frequency and voltage was 10 KHz and around 22-23 V each. The resulting layer (average thickness: ca. 573 nm) was sintered at 130 °C for 30 min in a convection oven ( Figure S6A and Table S1 in the SI).
Deposition of the Bottom Insulator : Cross-linked PVP (cPVP) was prepared by dissolving PVP in PGMEA at room temperature. The resulting solution was stirred for 30 min and, to obtain cPVP with the desired physical properties for jetting, it was subsequently fi ltered. The bottom insulator was deposited by inkjet printing of the previously prepared cPVP solution at room temperature (cartridge and platen), with a drop spacing of 20 µm by using a 10 pL cartridge, being the jetting frequency and voltage 10 KHz and around 23-26 V respectively. The resulting layer was baked at 200 °C for 20 min (fi nal thickness: ca. 650 nm; Figure S6B and Table S2 in the SI).
Deposition of the Source and Drain Electrodes : Interestingly, silver can be printed directly onto cPVP; no surface pre-treatment is required. Thus, the source and drain electrodes were deposited directly onto the bottom insulator using the same silver nanoparticle ink, printing conditions and post-printing procedure as for the bottom gate ( Figure S6C).
Deposition of the Semiconductor : Before deposition of the semiconductor, the source and drain electrodes layer were plasmatreated (O 2 , 60%, 25 W) for 30 s, and then cured on a hot plate for 15 min at 120 °C ( Figure S6D). The semiconductor layer was deposited onto the plasma-treated layer by inkjet printing of the amorphous p -type polymer FS0027, at room temperature (cartridge and platen), with a drop spacing of 25 µm by using a 10 pL cartridge. The jetting frequency and voltage was 10 KHz and around 12-14 V each ( Figure S6D and Table S3 in the SI).
Deposition of the Top Insulator : The top insulator, comprising MMAcoMAA was deposited on top of the semiconductor layer by spincoating it onto the substrate, and then baking the resulting device at 150 °C for 90 s on a hot plate ( Figure S6E). The thickness of the MMAcoMAA insulator layer was controlled by two spin-coating steps, the fi rst one in order to spread the material onto the device at 750 rpm for 10 s, and the second one to obtain the fi nal thickness (1.5 µm) at 1500 rpm for 30 s. The layer was characterized and optimized by SEM. The insulator breakdown and the semiconductor/insulator interface were characterized by electrical measurements. ( Figure S6E and Table S4 in the SI). Finally, regarding the jetting properties, examples of the waveform and the printability of the Sunchemical silver ink, FS0027 organic semiconductor, and cPVP are shown in SI. For the silver ink, the waveform shown in Figure S7A that generates stable and qualitative drop ejection has been used. The maximum jetting voltage was in the range of 22-29 V and the jetting frequency was set to 10 kHz. The perpendicular to the nozzle plate ejected drops exhibit a spherical shape which is an indicator for high quality materials deposition. For the semiconductor FS0027 the used waveform generates stable and good drop ejection. This w-shaped waveform leads to a stable drop formation of the organic semiconductor. The jettability of the organic semiconductors was very stable using the Dimatix 16 nozzles inert printheads. The droplets are ball-shaped and the drop ejection is perpendicular to the nozzle plate. In general, the ink can be used without problems in multi nozzle mode. (Figure S7B). Finally the waveform for the c-PVP ink was developed for stable drop ejection ( Figure S7C).
Addition of Electrolyte : To enable operation of the BioFET in aqueous media for biosensing experiments, a drop of electrolyte solution (0.1 M freshly prepared PBS pH 7.4) was added to the top of the top insulator, for use as both electrolyte and incubation buffer.

Amino-Functionalization of the MMAcoMAA Insulator Surface
The surface of the MMAcoMAA insulator was functionalized to enable binding of the target protein. This entailed two main steps: aminofunctionalization (described here), and antibody attachment (described below). The MMAcoMAA layer was plasma-oxidized, and then incubated with APTES. Different surface concentrations of amino groups were compared and characterized by quantitative colorimetry, using acid orange II dye. This method comprised immersion of the substrate with the amino-functionalized MMAcoMAA in an aqueous solution of acid orange II (the pH of the solution was adjusted to 3, using HCl 1 M). After shaking overnight at room temperature for at least 12 h, the samples were washed twice with the acid solution. In order to dissolve the adsorbed acid orange II, the samples were shaken for 15 min at room temperature in aqueous NaOH (pH 12). The concentration of dye (which was similar to the surface amino concentration of functionalized MMAcoMAA) in the solution was determined by absorbance measurements, using an optical spectrometer at 485 nm. [ 52 ] Functionalization was confi rmed by other methods, including fl uorescence detection by using a fl uorophore, immobilization of polystyrene microparticles, contact-angle studies, and AFM. In the fi rst experiment, the BSA-ALEXA 555, containing 1:1 EDC/NHS, was added dropwise onto two different MMAcoMAA surfaces (either bare or aminofunctionalized) for 2 h. The resulting MMAcoMAA surfaces were then cleaned with PBST, blocked with the blocking buffer (described in the experimental part), washed again with PBST, and, fi nally, subjected to fl uorescence measurements. Another simple test used to confi rm the surface functionalization with amino groups was the immobilization of polystyrene microparticles. The microparticle solution was added dropwise to each of two surfaces (bare or amino-functionalized), which were then incubated for 15 min and, fi nally, washed with PBST. Polystyrene microparticles remained bound to the functionalized surface, but were washed off from the unfunctionalized surface by PBST ( Figure S4). Finally, complementary studies were realized by using AFM as well as angle contact ( Figure S5).

Attachment of Antibodies to the Amino-Functionalized MMAcoMAA Insulator Surface
Once the amino-functionalization of the MMAcoMAA surface had been optimized, specifi c antibodies were attached to the surface amino groups. To this end, 15 µL of goat anti-human IgG (0.5 mg mL −1 ) were covalently bound to the amino groups through peptide coupling, using 135 µL of a solution of 1:1 EDC/NHS in PBS (5 mg mL −1 ) at room temperature for 30 min. The resulting surface was then washed with PBST. The free sites on the antibody-functionalized surface were then blocked, using PBS supplemented with 5% (w/v) casein and 0.005% (v/v) Tween 20 for 30 min. Finally, the blocked, functionalized surface was washed with PBST.

Label-Free Protein Detection
As proof-of-concept, the fi nal BioFET was evaluated for label-free detection of Human IgG, used as model protein. Human IgG was dropcasted onto the blocked, antibody-functionalized surface of the BioFET. The system was incubated for 30 min and fi nally, washed with PBST to remove any excess analyte. The amount of Human IgG bound to the BioFET was quantifi ed by I -V measurements.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.