Nanoscale Mapping of the Conductivity and Interfacial Capacitance of an Electrolyte‐Gated Organic Field‐Effect Transistor under Operation

Probing nanoscale electrical properties of organic semiconducting materials at the interface with an electrolyte solution under externally applied voltages is key in the field of organic bioelectronics. It is demonstrated that the conductivity and interfacial capacitance of the active channel of an electrolyte‐gated organic field‐effect transistor (EGOFET) under operation can be probed at the nanoscale using scanning dielectric microscopy in force detection mode in liquid environment. Local electrostatic force versus gate voltage transfer characteristics are obtained on the device and correlated with the global current–voltage transfer characteristics of the EGOFET. Nanoscale maps of the conductivity of the semiconducting channel show the dependence of the channel conductivity on the gate voltage and its variation along the channel due to the space charge limited conduction. The maps reveal very small electrical heterogeneities, which correspond to local interfacial capacitance variations due to an ultrathin non‐uniform insulating layer resulting from a phase separation in the organic semiconducting blend. Present results offer insights into the transduction mechanism at the organic semiconductor/electrolyte interfaces at scales down to ≈100 nm, which can bring substantial optimization of organic electronic devices for bioelectronic applications such as electrical recording on excitable cells or label‐free biosensing.


Introduction
In recent years the field of organic bioelectronics has experienced a fast evolution and development, with relevant emerging bioelectronic applications, among which the realization of bioelectronic recordings in excitable cells [1][2][3] and the www.afm-journal.de www.advancedsciencenews.com

(2 of 8)
© 2020 Wiley-VCH GmbH organic transistors operated in air or vacuum, [12,13,16,17] still has not been demonstrated for transistors operated in electrolyte solutions. On the other hand, electrochemical strain microscopy (ESM) has shown the possibility to probe local variations in ion transport in organic electrochemical transistors (OECTs) from the measurement of sub-nanometer volumetric expansions due to ion uptake. [18] However, this technique does not provide relevant information when applied to organic semiconductors in which ion uptake does not take place, as in most EGOFETs. Finally, microscale local potential measurements by scanning electrochemical microscopy have been demonstrated also on OECTs, [19] but they are based on the good conducting properties of OECTs, which are not generally met in EGOFETs.
Therefore, techniques that can directly probe the nanoscale electrical properties of the organic semiconducting channel in contact with the electrolyte solution in an EGOFET during device operation are still lacking. Establishing them would offer new insights into the physicochemical electrical transduction mechanisms at the organic semiconductor/electrolyte interfaces and would pave the way for the optimization of these, and other, organic devices for bioelectronic applications.
In the present study, we focus on mapping the local electrical conductivity of EGOFETs in operando. The conductivity is given by (see also Section S1, Supporting Information) where e is the electron charge, μ is the carrier mobility along the in-plane direction, and n(x) the free carrier concentration in the organic semiconductor film averaged along the transversal direction. The carrier concentration, n(x), can be determined by assuming the semiconductor forms the plate of a capacitor [20] so that (see also Section S1, Supporting Information) where c int,s (x) is the capacitance per unit area of the semiconductor/electrolyte interface, h the film thickness, and V(x) (res. V sol (x)) the electric potential in the semiconductor (res. electrolyte) with respect to the source at position x along the semiconducting channel of the EGOFET.
The mapping of the conductivity then can reveal spatial variations of the free carrier concentration due to variations of the voltage along the channel resulting from space charge limited transport and to local variations in the semiconductor/electrolyte interfacial capacitance. These variations are not accessible from the analysis of the transfer I DS -V GS or output I DS -V DS characteristics that normally provide averaged information (see Section S2, Supporting Information).
To map the local conductivity of an EGOFET channel under operation in an electrolyte solution, we employed in-liquid scanning dielectric microscopy (in-liquid SDM). SDM is a scanning probe microscopy technique that can be applied in either current [21,22] or force sensing [23][24][25] to obtain high spatial resolution electric and dielectric functional maps of materials in air [22,23] and in-liquid environment [24][25][26] (see Section S3, Supporting Information).
As an example of application we consider EGOFETs based on a blend of the organic semiconducting material 2,8-difluoro-5,11-bis(triethylsilylethynyl)anthradithiophene (diF-TES-ADT) with the insulating polymer polystyrene (PS), which we recently used as a bioelectronic recording platform for excitable cells. [3,27] EGOFETs based on organic blends have attracted the attention during the past few years due to their enhanced electrical performance and remarkable stability. [28,29] The impressive performance of these devices is believed to be correlated to the nanoscale structure of the organic semiconductor thin film, which shows an enhanced crystallinity and, in some cases, the presence of a nanostructured top insulating protective layer derived from a vertical phase separation phenomenon. [30][31][32]

Results and Discussion
The setup implemented to carry out the in operando nanoscale electrical measurements is shown schematically in Figure 1.
In it, the metallic (platinum coated) cantilever probe is used as both gate electrode for the EGOFET transistor and as a Figure 1. In-liquid SDM setup for the nanoscale electrical characterization of a functional EGOFET. An amplitude-modulated AC potential with frequency ω el > 1 MHz and modulation frequency ω mod of about 2 KHz is applied between the source of the transistor and the conductive probe in the electrolyte solution. An additional DC voltage offset is applied as a gate voltage V GS . A potential V DS is applied between the source and the drain, and the I DS current is recorded by a current-voltage source meter (not shown in the figure). The tip is scanned in the force volume mode. The SDM probe is used as both gate electrode and force sensor.
is applied to the probe (gate) with respect to the source electrode. The AC part of the voltage is used to generate and detect a cantilever oscillation harmonic at ω mod that is sensitive to only the local AC polarization properties of the sample at the frequency ω el . [24] The angular frequency ω el is in the MHz range, beyond both the dielectric relaxation frequency of the electrolyte and the mechanical resonance frequency of the cantilever. The free carrier concentration in the semiconductor channel of the EGOFET is practically not modulated by the high-frequency voltages used in SDM due to the poor coupling of the electrolyte with the channel at these frequencies (see Section S4, Supporting Information). As a result, SDM measurements are not sensitive to the semiconductor channel capacitance as in C-V measurements [27] but to the (local) conductivity of the organic semiconductor thin film and the (local) interfacial capacitance of the semiconductor/electrolyte interface (see Section S5, Supporting Information). We note that SDM is sensitive to the conductivities along and transversal to the channel. The DC voltage applied to the probe is used as a gate voltage V GS to polarize the transistor. The electric current flowing in the device, in response to the source-drain voltage V DS , is measured in real time with a current-voltage source meter connected to the device while in situ in-liquid SDM measurements are performed on the EGOFET. The EGOFETs based on the organic semiconducting blend composed of diF-TES-ADT and PS were fabricated by using the bar-assisted meniscus shearing (BAMS) technique as described earlier. [33] The active organic semiconducting layer is separated from the gate electrode by an aqueous electrolyte, which in our case was Milli-Q water. Current-voltage transfer characteristics of the EGOFET using the probe as gate electrode have been measured and show the classical response of an EGOFET that works in accumulation mode in Milli-Q water (see Sections S6 and S7, Supporting Information).
In-liquid SDM measurements on the EGOFET in operando were performed by operating the technique in the so-called force volume mode (see Section S3, Supporting Information), recently introduced in air environment. [34] We analyzed an area of the transistor which comprises of the source-channel-drain, as shown by the AFM topographic image in Figure 2a. The channel length is around 30 µm and presents some clearly visible nanoscale morphological heterogeneities, including some holes in the semiconductor layer (from where the thickness of the organic thin film layer was estimated to be 20-30 nm). The electrodes are about 50 nm thick (see cross-section topographic profile in Figure 2b taken along the dashed line in Figure 2a). Representative AC electric force-distance curves taken at a pixel located at the center of the channel at different gate voltages V GS are shown in Figure 2c. The force is represented through the capacitance gradient dC/dz [aF nm −1 ], which is related to the force by F ωmod = 1/8 dC/dz v ac 2 . [35] The dC/dz curves show a dependence of the electric force on the tip-sample distance, which is a sign of the locality of the measurement. The most relevant feature shown by the curves is that they shift up and change their "concavity" when the gate voltage, V GS , is swept from 0 to −0.6 V. Variations of the electric force with distance are more pronounced when the gate voltage surpasses the voltage V GS ≈−0.2 V, which coincides with the threshold voltage of the transistor (as measured from the current-voltage transfer characteristics in Section S6, Supporting Information). The changes observed in the AC electric forcedistance curves as a function of the gate voltage can be directly attributed to a change in the effective (local) conductivity of the semiconducting channel due to the variation of the gate voltage, see Section S5, Supporting Information. Note that the active material is a p-type semiconductor that works in accumulation mode, and an increase in the hole charge carrier density is expected with more negative gate voltages. The dependence of the electric force on conductivity can be qualitatively explained as follows. When the transistor is in the off-state the carrier concentration density in the channel is very low, and the AC electric field inside it is little screened out (the channel behaves like an insulator with losses). As a result, most of the AC voltage applied to the tip drops in the semiconductor layer (which has a relatively small relative permittivity ε sem ≈ 4), and on the SiO 2 substrate (which also has a low relative permittivity ε SiO2 ≈ 4), and very little in the electrolyte solution (which has a large relative permittivity, ε sol ≈ 78), resulting in a low AC electric force acting on the tip (black, red, and blue curves of Figure 2c). When the transistor passes to the on-state, the carrier concentration in the channel increases rapidly. Consequently, the AC electric field inside the semiconductor layer gets strongly screened and very little AC potential drops in it. Therefore, most of the AC potential drops in the electrolyte solution between the tip and the semiconductor layer, resulting in a large AC force acting on the tip, which shows a sharp increase at the closest distances. These results show that the force-distance curves obtained by in-liquid SDM are sensitive to (local) changes in the conductivity of the organic semiconductor film in contact with the electrolyte solution under in operando conditions (see Section S5, Supporting Information, for further details). We have quantified the local conductivity of the semiconductor channel as a function of the gate voltage by using the methods of SDM [23,24] (dashed lines in Figure 2c, see Section S3, Supporting Information). The extracted conductivity as a function of V GS is shown by the green curve in Figure 2d. When the transistor undergoes the transition from the off to the on-state, the local conductivity increases sharply with the gate voltage V GS , followed by an almost linear dependence on it. The qualitative behavior of the local conductivity as a function of the gate voltage is similar to the global conductivity obtained from the analysis of the I-V characteristics (see Section S2, Supporting Information). Quantitatively the values can be somewhat higher due to different effects (e.g., field dependent mobility, contact resistance effects, or sensitivity to the interfacial capacitance) as discussed in Section S8, Supporting Information.
From the electric force-distance curves acquired at different gate voltages, one can derive local electric force-voltage device transfer characteristics and correlate them with the conventional current-voltage transfer characteristics of the EGOFET. In Figure 2d we plot on the left axis, in log-linear scale, the source-drain current I SD versus gate-source voltage V GS measured from the transistor with the current-source meter during the in-liquid SDM measurements. On the right axis, in linearlinear scale, we plot the capacitance gradient (force) measured at the center of the channel by the in-liquid SDM setup as a function of the gate voltage V GS (bottom x-axis) at a given tip-sample distance (in this case Z = 180 nm with respect to the semiconductor surface, see Section S9, Supporting Information, for electric force transfer characteristics at different heights). In this graphical representation, there is a direct overlap between the global current-gate voltage transfer characteristics and the local electrostatic force-gate voltage transfer curve, demonstrating that both types of transfer curves provide similar information. The above results fit well with what one would expect in an OFET and thus confirms that in-liquid SDM measurements probe the change in the local conductivity of the channel during the EGOFET operation.
The previous analysis, which has been performed at a single point in the channel, can be performed in all points of the device (source, drain, and channel). We show the corresponding set of force-gate voltage curves at a distance Z = 180 nm with respect to the surface of the semiconductor in the inset of Figure 2d. We have plotted in different colors the local force-voltage transfer characteristics in the channel (blue), source (black), and drain (red). In absolute values, the forces on the source and drain are higher than on the channel, since the source and drain parts of the transistor are closer to the tip at the chosen distance at which the transfer curves have been extracted, and also because the substrate underneath the semiconductor is an electrode instead of an insulator (see Section S5, Supporting Information). Some variability exists in the electric force transfer characteristics obtained at the different points in the channel (and on the source and drain regions), which reflect slightly different local electrical properties. This spatial variability can be better illustrated by building up constant height electric force images for the different gate voltages. These images are obtained from the local electric force-voltage transfer curves by selecting the values corresponding to a given gate voltage at the selected tip-sample distance. [34] Figure 3a shows examples of electric images obtained for a tip-channel distance Z = 180 nm with respect to the surface of the semiconductor (electric images for different Z can be found in Section S10, Supporting Information) and for gate voltages V GS from 0 to −0.6 V. Concerning the semiconductor channel, when the gate voltage is below the threshold voltage (off-state) it shows an almost uniform and symmetric distribution of the electric forces (except for edge effects), which is independent from the gate voltage (see black, red, and blue central average crosssection profiles in Figure 3b). When the gate voltage passes the threshold voltage (V GS < −0.2 V), we observe a change in the contrast of the electric images, which reflects an increase in the force acting on the tip in all parts of the device. We note that the electric forces are larger on the source than on the drain (see Figure 3c). This fact is due to the presence of a nonzero source-drain voltage V DS that makes that the conductivity in the source and drain are different (for V DS = 0 V the forces are similar in both the source and drain, see Section S11, Supporting Information). In the channel, we also see the gradual decrease of the force from the source to the drain for the same reason. Repeating at each pixel the analysis done at the center of the channel in Figure 2c, we have calculated conductivity maps for the central part of the channel for different gate voltages (see Figure 3c). Figure 3d shows averaged conductivity www.afm-journal.de www.advancedsciencenews.com

(5 of 8)
© 2020 Wiley-VCH GmbH profiles along the channel obtained from each image. The conductivity in the channel changes both globally, because of the gate voltage, and along the channel on a large scale (micrometers) due to the source-drain voltage. This result implies that the conductivity of the channel can be mapped from in-liquid SDM measurements.
Finally, spatial heterogeneities at small scales (sub-micrometric) are also observed in the force and conductivity images of the channel for gate voltages above the threshold voltage. Some heterogeneities can be simply correlated to the sample morphology. For instance, in the upper right corner of the channel there appears an electrical heterogeneity in the force image (Figure 3a) due to a hole in the semiconductor layer (see Figure 2a). Some other heterogeneities, however, are not so simply correlated with the semiconductor morphology, for example, those observed in the central part of the channel (see Figure 3a,c). A possible source of these small scale conductivity heterogeneities could be attributed to variations in the interfacial capacitance. Indeed, the conductivity images have been built by assuming a uniform interfacial capacitance for the semiconductor/electrolyte interface, whose value was determined from measurements performed on the source (see Section S5, Supporting Information). If the actual interfacial capacitance is not uniform, its variations from point to point will be translated into variations into the extracted conductivity values (see Section S12, Supporting Information). This suggests that the surface of the transistor channel shows a nanoscale heterogeneous composition. The origin of such a heterogeneous composition could be the same as that reported for another semiconductor blend containing 2,7-dioctyl [1]benzothieno [3,2b] [1]benzothiophene and polystyrene (C8-BTBT:PS), [30] for which it was reported that a vertical phase separation occurred in which the organic semiconductor is sandwiched between a bottom PS layer and another top non-uniform ultrathin PS layers ≈0.5-2 nm thick. In electrical terms, the presence of a local ultrathin insulating layer would introduce a local interfacial capacitance variation, c int,s (x), to which the in-liquid SDM measurements are also very sensitive, as shown in [25] and also shown here in the Section S5, Supporting Information and as discussed elsewhere for the case of nanopatterned selfassembled-monolayers at metal/electrolyte interfaces. [35] The presence of an insulating layer of low dielectric constant (like PS, ε PS ≈ 2) on top of the semiconducting material induces a voltage drop on the layer which is larger than on the surrounding aqueous solution (which has a much higher dielectric constant, ε sol ≈ 78). Therefore, the force acting on the tip when on top of the ultrathin insulating layer is expected to be smaller than when on the bare semiconductor. Consequently, the regions showing smaller forces are expected to correspond to those with the presence of the ultrathin PS layers. In terms of the extracted conductivities (Figure 3c), if the model assumes a uniform interfacial capacitance, force reductions due to a reduction of the interfacial capacitance will be translated into a decrease of the extracted conductivity (see Section S12, The source-drain voltage was fixed at V DS = −0.5 V. The dark spots represent parts of the sample whose height is above the tip height of 180 nm and whose values are not reported (we arbitrarily assigned a black color to them). b) Average of the 13 AC electric force cross-section profiles along the source/channel/drain in the different images in (a). c) Conductivity maps of the central part of the channel of (a) which is obtained by a theoretical fit of each electrical curve at each pixel as in Figure 2c. d) Average of the conductivity profiles depicting the spatial variation of the conductivity along the channel obtained from the maps in (c), by assuming a uniform interfacial capacitance as determined from measurements on the source (see Section S5, Supporting Information).

(6 of 8)
© 2020 Wiley-VCH GmbH Supporting Information). To further confirm this interpretation we have acquired additional electrical and AFM images (topography, adhesion, and mechanical phase). The results of some representative regions of the channel are shown in Figure 4. In (a)-(c) we show topography, adhesion, and electric force images, respectively, obtained from a single set of SDM measurements on the selected region. In (d) and (e) we show intermittent contact topography and phase images of another region, respectively, and the corresponding in-liquid SDM constant height electric force image in (f). Both electrical images (c) and (f) correspond to Z = 30 nm from the surface of the semiconductor with the transistor in the on-state (V GS = −0.5 V for the case of Figure 4c and V GS = −0.3 V for Figure 4f). V DS is always kept at −0.5 V. We observe that when abrupt decreases in the force are observed in the electric images on top of the smooth variation along the channel, either the adhesion or the mechanical phase in the AFM images also present some abrupt variations (in the topographic images sometimes a small variation below ≈1 nm can be also observed). The fact that the electric force decreases on the heterogeneity indicates that it corresponds to a region covered by the (ultrathin) insulating layer. These results confirm that the heterogeneities observed in the force and conductivity images correspond to variations of the local interfacial capacitance of the semiconductor/electrolyte interface caused by the presence of a heterogeneous distribution of an ultrathin PS layer on top of the organic semiconductor. We have estimated a variation of interfacial capacitance in this case from ≈1.3 µF cm −2 to ≈0.3 µF cm −2 when passing from the semiconductor to the PS coated part (see Section S13, Supporting Information). It should be noted that these heterogeneities resulting from the blend do not hinder the overall performance of the EGOFET. The addition of PS enhances the material processability resulting in more crystalline films and stable devices as compared to the ones based on only the semiconductor. [27] The stability of these EGOFETs could be partly related to the observed ultrathin layer of PS that could act as a capping layer to ion diffusion into the semiconductor. [36] Having access to this information can be specifically useful for the case of biosensors based on organic transistors and to better understand the coupling of excitable cells with these devices for electrical cell recordings. [3] The application of the present approach to other electrolyte gated transistors, including OECTs, requires further study. OECTs do not have a field effect transport but rather an ionic modulation of a high bulk conductivity. SDM can be sensitive to high conductivity variations, but higher frequencies, probably in GHz range would be required. Similarly, measurements in electrolytes at higher ionic concentrations (beyond 100 m) would also require the use of frequencies in the GHz range. [24] Detection of GHz signals with force detection methods has been shown in the analysis of microwave integrated circuits. [37] Properly adapted, it could also be done for the in-liquid SDM measurements reported here. Concerning the spatial resolution and acquisition times, we expect that sub-100 nm and sub-30 s images could be obtained on EGOFETs with the use of high resonance frequency probes. [34,35]

Conclusion
To conclude, in this work we applied in-liquid SDM to study the nanoscale electrical properties (conductivity and interfacial capacitance) of a fully functional EGOFET. We have first shown that the AC electric force versus distance approach curves measured at single points in the channel are sensitive to the local conductivity, that is, to the free hole carriers concentration. Thanks to this sensitivity, local electric force-voltage transfer characteristics can be obtained at all points in the channel, which nicely correlate with the global device current-voltage transfer characteristics. Conductivity maps of the transistor www.afm-journal.de www.advancedsciencenews.com

(7 of 8)
© 2020 Wiley-VCH GmbH channel reveal both global conductivity variations due to changes in the gate voltage and variations along the channel due to the space charge conduction in an accumulation mode field effect transistor. The conductivity maps have also revealed minute heterogeneities at small scales (micron and submicron) that we have correlated to variations of the interfacial capacitance of the semiconductor/electrolyte interface. For the semiconductor blend diF-TES-ADT:PS used in the EGOFETs analyzed here, these results show that a vertical phase separation takes place with the PS forming heterogeneous ultrathin layers on top of the semiconductor material. The electrical images enable to directly identify which parts of the surface are covered by PS and which parts are not and to quantify the local variations in the interfacial capacitance. Such findings would offer the possibility to optimize the device performance such as increasing the coverage of the ultrathin PS layer to enhance stability or eliminate this layer in order to have a better coupling at the semiconductor/electrolyte interface. The results presented open fascinating possibilities to study the transduction mechanisms at the organic semiconductor/electrolyte interface relevant for electrolyte gated transistors and to optimize device fabrication and performance for bioelectronic applications.

Experimental Section
Device Fabrication: A heavily doped n-type silicon wafer featuring thermal SiO x (200 nm thick) was used as substrate for our devices into which source and drain (S/D) electrodes were defined by photolithography as described earlier. [3,27] The channel width (W) and length (L) were 19680 and 30 µm respectively. Prior to the deposition of the organic semiconductor, the substrates were cleaned in ultrasonic bath with acetone and isopropanol for 15 min, respectively, and afterward ozone treated for 25 min. S/D electrodes were subsequently modified by immersing the device in a 15 mm pentafluorothiophenol solution in isopropanol for 15 min. A blend composed of diF-TES ADT and polystyrene (PS) was mixed in a 4:1 ratio and then dissolved in chlorobenzene reaching a final concentration of 2 wt%. The blend solution was kept on a hot plate at 105 °C for 1 h to ensure the complete dissolution of the starting materials. Thin film deposition was realized through BAMS technique by means of a home-adapted bar coater working at fixed speed of 1 cm s −1 and at a fixed plate temperature of 105 °C as reported earlier. [27,31] All the abovementioned processes were realized under ambient conditions.
In-liquid Scanning Dielectric Microscopy on Functional EGOFET: SDM measurements had been performed in liquid following approaches presented earlier. [24,25] In a nutshell, an amplitude-modulated AC voltage (Equation (1)) of amplitude 1.5 V PP , electrical frequency in the range 1-30 MHz and modulating frequency of 2kHz had been applied by means of a function generator (33220A Function Waveform Generator, Keysight) between the conductive probe of an AFM system and the source electrode of the EGOFET. In order to polarize the transistor, a V DC offset (equivalent to the gate-source voltage V GS ) was applied in addition to the AC voltage driving the tip's movement. A constant voltage V DS was applied externally using the Source Measuring Unit Agilent 2912B. The ground of all the equipment had been short circuited in order not to have floating potentials. The so-called force volume mode had been used to acquire the data, as described in the text and previously applied in air. [34] Normal deflection and A ωmod oscillation amplitude approach curves were acquired using the JPK Nanowizard 4 AFM system (JPK) connected to an external lock in (eLockin 204/2, Anfatec). The acquisition was performed by using the Advanced Quantitative Imaging (JPK) mode. The set of SDM data of Figure 3a consisted of 64 × 13 deflection and ω mod -oscillation amplitude approach curves, each one with 300 data points and spanning a length of about 2.5 µm. The SDM data of Figures 4c consisted of 128 × 26 pixels and that of figure 4f consisted of 128 × 23 pixels. The acquisition time per pixel was 300 ms, and the lock-in integration time was set to 2 ms, with a gain G lock-in = 50. The measurements had been carried out by using PPP-CONTPt tips (NANOSENSORS), with equivalent spring constant k = 1.2 N m −1 and resonance frequency in liquid of around 11 KHz. To acquire an image of 65 × 13 pixels it took around 4 min per set of SDM data for each gate voltage. Electrical oscillation amplitudes were converted to electric forces as detailed elsewhere. [23,35] Quantitative Analysis: Numerical calculations were used to relate the electric force acting on the tip to the local conductivity and interfacial capacitance of the organic semiconductor. The numerical calculations had been implemented in COMSOL Multiphysics 5.4 linked to a custom made Matlab interface using the modeling developed earlier in refs. [23,24,38,39] adapted here to include tip interfacial capacitance effects, as detailed elsewhere [35] and to deal with conductive samples. Further details are provided in the Sections S3 and S5, Supporting Information.

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