Embedded Dipole Self-Assembled Monolayers for Contact Resistance Tuning in p-Type and n-Type Organic Thin Film Transistors and Flexible Electronic Circuits

Based on the powerful concept of embedded dipole self-assembled monolayers (SAMs), highly conductive interfacial layers are designed, which allow tuning the contact resistance of organic thin-film transistors over three orders of magnitude with minimum values well below 1 k Ω cm. This not only permits the realization of highly competitive p-type (pentacene-based) devices on rigid as well as flexible substrates, but also enables the realization of n-type (C 60 -based) transistors with comparable characteristics utilizing the same electrode material (Au). As prototypical examples for the high potential of the presented SAMs in more complex device structures, flexible organic inverters with static gains of 220 V/V and a 5-stage ring-oscillator operated below 4 V with a stage frequency in the range of the theoretically achievable maximum are fabricated. Employing a variety of complementary experimental and modeling techniques, it is shown that contact resistances are reduced by i) eliminating the injection barrier through a suitable dipole orientation, and by ii) boosting the transmission of charge carriers through a deliberate reduction of the SAM thickness. Notably, the embedding of the dipolar group into the backbones of the SAM-forming molecules allows exploiting their beneficial effects without modifying the growth of the active layer. (FHI-aims) [69] and employing the PBE functional [39] combined with the surface version [70] of the Tkatchenko–Scheffler dispersion corrections [71] for the Van der Waals interactions. The system was modeled using periodic boundary conditions within the so called repeated slab approach, inserting a vacuum region of at least 20 Å in the z direction and including a self-consistently calculated dipole correction [72] between the slabs, to electrostatically and spatially decouple them. The metallic substrate was modeled with five Au layers, holding the three bottom ones fixed during all the calculations and turning off the dispersion corrections between the Au atoms. Two molecules were put in a (3 × √ 3) rect unit cell, whose dimensions in the x and y directions were defined accordingly to the calculated Au lattice constant. The docking groups were placed in fcc-hollow sites and both cofacial and herringbone arrangements were tested, with the latter being energetically more favorable. The systems were optimized using the FHI-aims default “tight” setting and a 9 × 5 × 1 k-points grid. The total energy criterion for the self-consistency cycle was set to 10 − 6 eV and the optimizations were performed until the maximum residual force component per atom was below 0.01 eV Å − 1 . The declare no conflict of


Introduction
Contact resistances hamper the development of organic thinfilm transistors (OTFTs) with operation frequencies in the molecular packing of the organic semiconductor, [10,19] they serve as adhesion layer to prevent irregular contact edges and voids, [5] and they remove contaminants from the surface by replacing them with covalently linked molecules.
To maximize the favorable properties of SAM-modified contacts, several conditions ought to be met: i) Independent of whether aliphatic [8,15,[20][21][22][23][24][25][26][27] or aromatic [6,10,17,21,[28][29][30][31][32] backbones are employed in the devices, it must be possible to alter efficiently the chemical structures of the molecules such that the level alignment at the electrode/semiconductor interface can be adjusted in a straightforward manner. Ideally, the chosen structural motifs should be flexible enough to provide the possibility to realize monolayers that allow either electron or hole injection into suitably chosen semiconductors from a single type of electrode. ii) The SAM structures used to induce the desired changes in the substrate work-function should not compromise the growth of the active organic semiconductor layer. [19] Thus, systems in which level alignment and semiconductor growth can be tuned independently are particularly promising. iii) To maximize the injected currents, carrier transport through the monolayers has to be as efficient as possible; i.e., SAMs should by themselves be as transmissive for electrons or holes as possible. [33] iv) For a truly knowledge-based design of SAM-modified electrodes, it is crucial to combine an in-depth understanding of the SAM properties with an extended characterization of the organic layers grown on the SAMs, suitable modeling approaches both at the atomistic and at the device level, and a thorough investigation of the ensuing device structures. v) To allow the realization of more complex, multicomponent electronic circuits, it must be possible to incorporate the SAM deposition onto the contacts into a highly reproducible and simple OTFT fabrication process.
All these aspects are realized in the present manuscript with a versatile SAM platform, which meets the above requirements in an elegant and highly adaptive way. An outstanding feature is that this SAM concept allows the straightforward realization of both p-type and n-type transistors and circuits using the same electrode material and geometry with minimized contact resistances (on the order of 1 kΩ cm for devices operated at ≈3 V). Thus, instead of multistep and error-prone process routes for fabricating combinations of contact materials, we simply need to immerse the gold electrodes in a solution of the SAM-forming molecules with the appropriate dipole orientation. This can be done without the need to adapt the semiconductor growth conditions to the electrode material. Moreover, due to the embedded-dipole approach, detrimental effects of the SAM on the morphology of the semiconductor deposited on top of the SAM are avoided. The employed strategy increases the device-to-device reproducibility (especially on flexible substrates), improves yields, and saves time and costs; these aspects are all inevitable prerequisites for the large area production of complex organic circuits. As prototypical examples of such circuits, we realized flexible unipolar inverters with high static gain and 5-stage ring-oscillators.
Our SAM platform is based on an aromatic backbone to minimize additional tunneling barriers caused by the interfacial layer. [11,[34][35][36] In a first step, we used terphenyl-4-methanethiol (TP1; TP referring to the terphenyl backbone and 1 denoting one methylene spacer). To trigger work-function changes, a polar 2,5-pyrimidine moiety is introduced into the backbone, as shown in Figure 1a (first generation). [37,38] Employing mixed monolayers of these molecules, it is even possible to continuously tune the work function of a Au substrate over a range of 0.9 eV. [37] To further boost the beneficial effect of the SAMs, we deliberately synthesized molecules derived from a biphenyl-4-thiol (BP0) backbone that is shorter and has no methylene spacer in the linker (see Figure 1a, "2nd generation"), thus hoping to minimize the detrimental tunneling barrier due to the SAM.

SAM Characterization
To assess the suitability of the pyrimidine-substituted SAMs for engineering Au surfaces serving as electrodes in OTFT contacts, we first grew SAMs on plasma-cleaned Au layers deposited on glass-based OTFT substrates according to a process described in the Experimental Section and illustrated in Figure S1 in the Supporting Information. The embedded dipoles in the SAMs pointing downward (upward) cause the work function (W f ) to increase (decrease) with respect to the dipole-free reference SAMs. For both generations of SAMs, the covered work-function range amounts to at least ≈0.9 eV, as shown by Kelvin-probe (KP) measurements, W f,KP (Figure 1b), and via the secondary electron cutoffs W f,UPS in the Ultraviolet photoelectron spectroscopy (UPS) spectra (Table 1; Figure S3 in the Supporting Information). To trace the origin of that shift, we calculated the electrostatic energy in the SAM employing dispersion-corrected density-functional theory using the repeated slab approach (with the Perdew, Burke, Ernzerhof (PBE) exchange functional). [39] The results for BP0-up and BP0down, shown in Figure 1c, reveal that the potential drop primarily occurs within the layer in the spatial vicinity of the N atoms. Notably, the energy has become essentially constant at the top of the SAM.
Grazing incidence X-ray diffraction (GIXD) measurements show that pentacene grows on SAM-treated electrodes in crystallites consisting of upright standing molecules. These crystallites belong to two polymorphs, the thin film phase, and the Campbell phase. [40,41] The structures are essentially the same on all surfaces; small deviations are found only for the growth on the TP1 surface (see Figure S4 of the Supporting Information). On the Al 2 O 3 surface used as gate dielectric in the devices (see below), the GIXD measurements show the presence of only the pentacene thin film phase.
To assess the film morphology, we performed atomic force microscopy (AFM) experiments and to relate the morphology to the properties of the underlying SAM-covered electrodes, we conducted contact angle (CA) measurements. Consistent with all SAMs having a surface energy higher than that of pentacene (γ pentacene = 42.1 mJ m −2 ; [42] cf. Table 1), we observe pronounced Stranski-Krastanov-type island growth (Figure 2). The morphologies of the pentacene films on top of all 1st generation SAMs are very similar (see Figure 2a). They are fine-grained with nucleation densities N between 20.3 and 23.8 grains µm −2 . This is consistent with similar values of the surface energy, 46.4 mJ m −2 ≤ γ ≤ 50 mJ m −2 for all three SAMs.
For the shorter and purely aromatic 2nd generation SAMs, we observe the following differences compared to the firstgeneration systems: The RMS surface roughness values of BP0based SAMs are significantly smaller than those of TP1-based SAMs and the surface energies of BP0 and BP0-down SAMs are lower. Both aspects lead to a change of the semiconductor morphology. [42][43][44][45][46] Consequently, on the 2nd generation SAM treated Au surfaces, we observe larger average sizes of pentacene grains (see Figure 2b) along with lower nucleation densities N. On average, the morphologies of the pentacene layers are very similar to the ones observed on blank, plasma-cleaned Aulayers but with higher uniformity (see Table S1 and Figure S5d in the Supporting Information). The decrease in N is more pronounced for BP0-down and BP0 SAMs (N ≈ 7.3-7.9 grains µm −2 ; Table 1). For growth on BP0-up SAMs, the nucleation   W f,KP and W f,UPS refer to work functions determined by Kelvin probe and from secondary electron cut-offs of UPS-spectra. The measurements have been performed on SAMs prepared on substrates equivalent to those used for device fabrication; b) γ is the total surface energy extracted from contact angle measurements with γ D and γ P being its dispersive and polar components; c) RMS is the root mean square of the surface roughness, and N the grain density. The listed numbers correspond to the mean values over four experiments for each monolayer with the respective standard deviations cited in brackets. density (N ≈ 10.9 grains µm −2 ) is somewhat higher, but still far below that one on the TP1-based SAMs. We attribute the higher N-value to the distinctly higher polar component of the surface energy originating from pyrimidine as the terminal ring of the BP0-up molecule.

OTFTs with Embedded Dipole SAM-Treated Au Electrodes
Knowing the key electronic, structural, and morphological properties of the studied layers, the next step is to incorporate them into the bottom gate, bottom contact OTFT architecture shown in Figure 3a. This architecture features an 18 nm ultrathin Al 2 O 3 gate dielectric that allows for operation voltages below 3 V due to its large gate capacitance of 440 nF cm −2 and its remarkable density (gate leakage currents below 150 pA, see Table S3 in the Supporting Information). A thiol-terminated phosphonic acid SAM (PA-SAM) immobilized on the Al 2 O 3 substrate serves as an adhesion layer for the Au electrodes patterned by photolithography. Channel lengths were varied between 7.5 and 51 µm and channel widths between 1 and 4 mm. At each channel length, several devices were produced (e.g., 5-10 transistors with L ≈ 7.5 µm for dipole-down SAM treated electrodes). For the extraction of the contact resistance, several transmission line method (TLM) sets were fabricated (e.g., 3-5 TLM-sets of transistors with dipole-down SAM treatment), each made of 4-6 transistors with different channel lengths. The number of considered devices is specified in detail in the respective table captions. More information on the fabrication process and characterization of the devices can be found in the Experimental Section. Typical I D (V DS )-electrical output characteristics for bottom contact OTFTs with SAM-modified source and drain electrodes for channel lengths of L = 7.5 µm are shown in Figure 3b. The magnitude of the current changes by a factor of ≈20 going from down to up SAM modified OTFTs. This is consistently observed for applying the 1st (upper panels) and the 2nd generation SAMs (lower panels). Since within each SAM generation, the pentacene morphology on the electrodes is very similar irrespective of the direction of the embedded dipole moment, we attribute these differences primarily to dipole-induced changes in the hole-injection barriers (HIB). The latter can be estimated from the (SAM-modified) work-function of the electrode (W f ) and the ionization potential of upright standing pentacene molecules (IP pentacene ) [47,48] as HIB = IP pentacene − W f , considering that no significant interface dipoles [49,50] are expected in the contact region between the SAM and the organic semiconductor. In all cases in which IP pentacene < W f , we set HIB = 0 due to Fermi level pinning (see Table S3 in the Supporting Information). Transistor parameters for the reference samples with H 2 and O 2 -plasma cleaned Au electrodes and no SAM treatment are listed in Table S2   For SAMs containing no or upward-pointing dipoles, the sizable values of HIB between 0.35 eV and ≈0.8 eV give rise to large R c values. These also cause the S-shape in the low voltage region of the measured output curves. [51] By contrast, the SAMs with downward pointing dipole moment (TP1-down, BP0down) and vanishing HIBs (see Table S3, Supporting Information) develop a perfectly linear I D (V DS ) characteristics in the respective region.
Further insight in the influence of the dipole-SAM treatment on the device parameters is provided by the transfer curves (e.g., Figure S6a,b in the Supporting Information). Since the ON current reaches a maximum in devices with dipoledown SAMs on Au (TP1-down and BP0-down), whereas the OFF current is virtually constant (see Table S3, Supporting Information), the ON/OFF current ratio increases continuously from dipole-up over no dipole to dipole-down SAM treated Au electrodes with best values for OTFTs with the shorter BP0down SAMs on Au (see Table S4, Supporting Information). The dipole-down SAMs also have the smallest sub-threshold swing S, and the lowest positive onset voltage V ON , which indicates a smaller contribution from the voltage across the Schottky barrier at the charge injecting interface (see Equation (S4) in the Supporting Information).
To quantify R c , we analyzed all sets of output and transfer curves in two distinct ways. For a comparison with the bulk of literature, we first determine R c by systematically varying the channel length and employing the TLM described in the Supporting Information (see also Figures S7 and S8 in the Supporting Information). [52,53] As an alternative, to explicitly discriminate between ohmic (linear) and nonohmic (nonlinear) resistance contributions [53] (the latter arising from the injection barrier between electrode surface and semiconductor), we determine R c with a newly developed fitting approach (FA) described in detail in the Supporting Information (see also Figure S9, Supporting Information). In short, inspired by the suggestion of Fischer et al., [54] we simultaneously fit the output and transfer curves for different operation conditions to an equivalent circuit model combining an ideal transistor with a gate-controlled Schottky diode. The characteristics reconstructed from the fitted model parameters fully reproduce the measured curves, in particular in terms of their pronounced S-shapes at large HIB (Figure 3b, magenta lines). As shown in Figure 4a,b for the two analysis approaches, the obtained channel-width normalized contact resistances R c ⋅W for different dipole orientations (and HIBs) vary by more than two orders of magnitude within each generation of SAMs. The full set of transistor parameters and extracted contact resistance values are found in Table S4 in the Supporting Information. Notably, for the dipole-down SAMs on Au (TP1-down, BP0-down) the contact resistances are purely ohmic and are clearly below 10 kΩ cm (cf. Table S4, Supporting Information). For BP0-down, they even fall as low as 0.58 kΩ cm (best value, see Table 2). Much larger    R c W values are obtained for dipole-free SAMs (TP1, BP0; 100 kΩ cm ≤ R c W ≤ 1000 kΩ cm) and for the dipole-up systems (TP1up, BP0-up; R c ⋅W ≈ 1000 kΩ cm). The employed fitting routine shows that these high contact resistances are dominated by contributions that are nonlinear in V G and, thus, arise from injection processes across a barrier (cf. Table S4, Supporting Information). This nonlinear contribution is, in fact, also responsible for the pronounced S-shapes in the output curves discussed above. [55] The R c W values for devices without any SAM treatment (see Table S2, Supporting Information) are similar on average to those of TP1-down treated OTFTs but with a much larger spread. When comparing transistors containing SAMs with different backbone structures, the R c -values for 2nd generation, BP0derived SAMs are consistently lower compared to 1st generation, TP1-derived ones. For the dipole-down SAMs, this difference reaches nearly an order of magnitude (see Table 2). This might seem insofar surprising, as based on the measured work functions (see Figure 1b), there should be essentially identical hole-injection barriers into the pentacene layer for a given dipole orientation. The more favorable morphology of the pentacene layers on BP0-based SAMs should also primarily result in a slight improvement of the carrier mobility (see below). Therefore, we attribute the better performance of OTFTs treated with the 2nd generation molecules predominantly to a higher conductance of the BP0-based SAMs due to the shorter length of the conjugated backbone and the absence of the methylene linker. [35] Indeed, conductivity measurements through the SAMs on Au using an EGaIn droplet as counter electrode [56] (Figure 4c) show that for a monolayer junction the currents through 2nd generation up and down molecules (blue) are 1-2 orders of magnitude higher compared to the 1st generation ones (red). A more detailed discussion of the energetic alignment between the states in the SAMs and the Fermi-level of the substrate can be found in Figures S10 and S11 (Supporting Information).
Contact resistances also determine the extent to which the effective charge carrier mobility, µ, extracted from an OTFT characteristics, represents the actual property of the organic semiconductor. [57] Typically, mobility values in the saturation regime, µ sat , and in the linear regime, µ lin , are extracted from the transconductance dI DS /dV GS . [57] To remove the influence of the contact, we determined the equivalent mobility in the transistor channel, µ eqv,FA , for negligible drain-source bias from the abovementioned fitting approach (see Table S4, Supporting  Information).  (Figure 4d) is independent of the channel length only for R c ⋅W ≤ 10 kΩ cm (shaded region, channel-resistance dominated transport). For higher values of R c ⋅W, µ sat undergoes a clear transition to a contact-resistance dominated transport regime with a strongly L-dependent, effective mobility. That is, in that regime, the extracted value of µ sat is primarily determined by the contact resistance and is no longer a direct property of the active material. The linear mobility is even more affected by the contact resistance, with L-independent values for µ lin only for R c ⋅W ≈ 1 kΩ cm, i.e., only for BP0-down SAMs/Au (shaded region in Figure 4e). In stark contrast, the R c -corrected mobility µ eqv,FA (Figure 4f) does not systematically depend on the channel length or on R c ⋅W (respectively, the dipole orientation). Rather, µ eqv,FA of the 2nd generation (filled symbols) typically surpasses the µ eqv,FA of the 1st generation. Values in the shaded region are excluded from this comparison because of the lower quality of the fit in the presence of dominating nonlinear R c contributions. The remnant spread of the values stems from device to device fluctuations between different channel lengths.
Although it does not directly relate to the discussion of the influence of R c on µ, it has to be noted that even our devices with ultralow R c have not reached the best mobility values reported for staggered pentacene OTFTs on high-k dielectrics; our mobilities are smaller by a factor of approximately three for Al 2 O 3 [58] or approximately six for ZrO 2 -PαMS nanocomposites. [59] Despite very low evaporation rates, our pentacene layers presumably contain more grain boundaries and thus a smaller intrinsic mobility due to two reasons: First, the anodized Al 2 O 3 surface, that emerged after the photolithographic process and the plasma cleaning, is not pristine, because we had to omit a passivating and smoothening SAM treatment of the dielectric to avoid any cross-influence on the pyrimidine SAMs. In order to allow for a clear assignment of the contact resistance tuning to the type of embedded dipole SAMs on the contacts, we, thus, deliberately accepted a dielectric surface that is good but not perfected for OSC growth with small grain boundaries. Second, despite the SAM treatment of the contacts, we cannot exclude that our coplanar setup causes a less homogeneous film growth in the contact-channel transition region.

p-Type and n-Type OTFTs with Embedded Dipole SAM-Treated Au Electrodes on Flexible Substrates
The effective lack of injection barriers in OTFTs containing TP1down or BP0-down SAMs leads also to reduced spreads in the OTFT parameters and to the smallest subthreshold swings in the transfer characteristics of all devices studied here (Table S3 in the Supporting Information). This makes such devices particularly promising for applications in more complex electronic circuits. Realizing, for example, inverters and ring oscillators on mechanically flexible substrates would be particularly interesting, as it could significantly impact the area of flexible electronics (Figure 5a). Additionally, a possibility to fabricate n-type devices on flexible substrates without changing the electrode material would be highly intriguing. Consequently, the next logical step is to test the transferability of our concept for contact engineering to devices on flexible substrates and to n-type organic semiconductors (in both instances varying channel lengths between 3.5 and 52 µm).
BP0-down SAM modified pentacene OTFTs (L = 7.5 µm) fabricated on a flexible polycarbonate (PC) substrate yield output ( Figure 5b) and transfer characteristics ( Figure S12a, Supporting Information) that favorably compare to the OTFTs on glass discussed above. Table 2 compares the short channel transistors (L = 5-7.5 µm) with optimum SAM treatment on glass and on PC substrates regarding mobility and contact resistance values. Notably the mobilities (≈0.2 cm 2 V −1 s −1 ) and the ON/OFF ratios www.afm-journal.de www.advancedsciencenews.com (≈10 6 ) are essentially identical (for the full set of parameters, see Table S5 in the Supporting Information). Also the R c ⋅W values amounting to ≈2.5 kΩ cm are essentially the same as for the equivalent glass-based OTFTs when employing the more sophisticated fitting analysis. We believe that this competitive device performance on the flexible substrates is due to the excellent surface quality of our PC substrate, which is exceptional in terms of surface roughness (0.9 nm) and process stability owing to a special hard coating layer. Finally, it has to be noted that mechanical stress tests (bending or strain tests) for pyrimidine-SAM treated p-type OTFTs are ongoing and will be reported in the near future.
For fabricating n-type devices, BP0-up SAMs should be particularly promising considering that they efficiently lower the electron injection barrier (EIB). For example, for C 60 -based OTFTs, a barrier of EIB = W F,KP − EA C60 ≈ 0.15 eV (for EA C60 ≈ 3.92 eV) [60] can be anticipated. The high application potential of the dipole-up SAMs for contact resistance reduction is fully confirmed by the performance of the C 60 OTFTs that we fabricated on flexible PCsubstrates. The corresponding output characteristics for OTFTs containing 2nd generation SAMs with L = 7.5 µm are shown in Figure 5c, the transfer characteristics in Figure S12b (Supporting Information). They reveal a clear n-type behavior for all three SAMs. However, only for the BP0-up modified electrodes, I D (V DS ) curves without S-shape are observed. In that case, we observe an average contact resistance as low as R c,TLM ⋅W = 4.6 kΩ cm and a corrected mobility of µ eqv,FA = 0.09 cm 2 V −1 s −1 (see Table 2), which are both in the same range as for the BP0-down treated, pentacene-based p-type OTFTs on the PC-substrate.

Flexible Inverters and Ring-Oscillators Containing Embedded-Dipole SAMs
The excellent performance and the small device-to-device variation (see Table S5, Supporting Information) observed for the flexible BP0-down modified p-type OTFTs qualify these transistors for integration into flexible organic circuits. Accordingly, as representative examples, we fabricated unipolar inverters and 5-stage organic ring-oscillators on a flexible 1 in. × 1 in. polycarbonate substrate.
Typical voltage transfer characteristics (VTC) of a zero-V GS inverter for supply voltages of V DD = 2, 3, and 4 V are shown in Figure 6a. Up to an operation voltage of V DD = 4 V, the inverter exhibits a high static gain of 220 V/V, a very sharp transition regime (∆V in < 140 mV) and full rail-to-rail swing. Figure 6b also demonstrates the remarkably large noise margin of 0.82 V and the symmetric transition close to V DD /2 for forward and reverse voltage sweeps. [61] A photograph of a flexible 5-stage ring-oscillator made of such inverters and the measured output signal at V DD = 4 V are displayed in Figure 6c. The output signals of two oscillators on the same sample are displayed in Figure S13   consequence of a contact resistance, but is rather the consequence of a parasitic capacitance contribution from the stage interconnections and due to the load transistor being operated just above the threshold. A detailed discussion of the performance of the inverters and ring oscillators and their transfer frequency is found in the Supporting Information.

Conclusion
In summary, in this work we show how the new concept of embedded-dipole SAMs can be adapted to tune the contact resistance in bottom contact, bottom gate organic thin-film transistors by three orders of magnitude. This tuning is achieved by switching the orientation of a pyrimidine ring incorporated in the aromatic backbones of the SAM-forming molecules, which allows modifying the work function of the electrodes over a range of ≈0.9 eV (verified by Kelvin probe and UPS). Quantummechanical simulations show that the underlying shift in the electrostatic energy occurs within the self-assembled monolayer. The concept of embedding the dipolar elements into the molecular backbones ensures that switching the dipole orientation neither impacts the crystal structure of the organic layers (confirmed by X-ray diffraction), nor their morphology (as seen in atomic force microscopy). Varying the length of the aromatic systems and including/omitting a methyl spacer between the backbones and the thiolate docking group, we find that the reduced SAM-tunneling barrier in shorter and fully aromatic SAM-forming molecules helps further boosting device performance. This is consistent with transport measurements employing EGaIn junctions.
The best-performing p-type, pentacene devices operated at 3 V show width-normalized contact resistances as low as R c ⋅W ≈ 0.58 kΩ cm on glass and 1.7 kΩ cm on plastic substrates as determined via the transfer-line method. A particularly intriguing aspect of the SAMs presented here is that by switching the dipole moment within the backbone, they also allow the realization of n-type, C 60 based transistors on flexible substrates using photolithography patterned gold electrodes. Notably the contact resistances and mobilities of the ensuing devices are very similar to those of the best p-type transistors.
To highlight the potential of SAMs with embedded dipoles for more complex device applications, we also fabricated functional 5-stage organic ring-oscillators on flexible substrates that were based on very well performing unipolar inverters (gain of ≈220 V/V at 4 V) and operated below 4 V.

Experimental Section
While the synthesis of the TP1-derived molecule is described in ref. [38], details regarding the synthesis of the BP0-based molecules will be published elsewhere.
Preparation of SAMs: The SAMs were prepared by immersion of the freshly prepared substrates (cleaned "photopatterned" Au layers) into 100-500 × 10 −6 m ethanol (≥99.9%, HPLC grade, VWR) solutions of the pyrimidine-substituted thiols and references molecules for 18 h. After immersion, the samples were carefully rinsed with pure solvent and blown dry with nitrogen.
Tunneling Current Measurements across SAMs with Eutectic Ga-In (EGaIn) Top-Contacts: Electrical conductance measurements were performed with a custom-made two-terminal tunneling junction setup, based on the Keithley 2635A source meter. [63] The gold substrate and a sharp tip of eutectic GaIn (EGaIn) served as the bottom and top electrodes, respectively. [56] Tunneling junctions were formed by contacting grounded SAM/Au samples with the EGaIn tips and applying a potential. The voltage was varied between −0.6 and +0.6 V in 0.05 V steps. At least 10 I-V curves measured at several different places were recorded for each sample; average values were calculated.
Work Function Determination by UPS and KP: UPS experiments were performed in a multiprobe surface analysis ultrahigh vacuum (UHV) system from Omicron Nanotechnology with a base pressure of 1 × 10 −10 mbar. The spectra were recorded using a HIS 13 vacuum ultraviolet source (incident photon energy of 21.22 eV, He I line) and an EA 125 hemispherical analyzer with a pass energy of 1 eV and an energy resolution of 80 meV. The secondary cut-off was recorded while the sample was biased at −5 V (relative to the analyzer). The absolute error in the determination of W f is estimated to be ±100 meV. Complementary  The inverter was fabricated on a polycarbonate plastic film. A circuit scheme of the inverter is also displayed. b) Determination of the noise margin at V DD ≈ 2.5 V using the "maximum equal criterion." The noise margin (NM) of the inverter is determined by finding the maximum square that fits between the inverter transition curve and the mirrored inverter curve. c) Circuit scheme, photograph, and measured output signal of a 5 stage-ring oscillator with an output buffer stage fabricated on a PC substrate. The dimensions of contact pads on the left side of the photograph are 600 × 600 µm.
to the UPS experiments, work function measurements were also carried out using a UHV Kelvin Probe 2001 system (KP technology Ltd., UK). The pressure in the UHV chamber was ≈10 −9 mbar. Freshly sputtered gold and hexadecane SAMs on Au(111) were used as references with the work function values set at 5.2 [64] and 4.32 eV, [65] respectively.
Surface Analysis-H 2 and O 2 Plasma-Cleaned Au and Al 2 O 3 : X-ray photoelectron spectroscopy was used to verify the qualitiy of the H 2 and O 2 -plasma cleaning step prior to SAM coating. The measurements were performed in the UHV system mentioned above using a DAR 400 X-ray source (Al K α1 radiation hν = 1486.7 eV), an XM 500 quartz crystal monochromator (energy width 0.15 eV), and a take-off angle of 90°. The pass energy of the hemispherical analyzer for the detailed element spectra was set to 20 eV resulting in a total energy resolution of 0.5 eV.
Characterization of Pentacene Thin Film Morphology with AFM, and GIXD/XRD: The surface roughness of the SAM-treated and nontreated Au layers was extracted from atomic force microscopy (VEECO Dimension 3100 AFM) topography images. AFM was also used to investigate the morphology of the organic semiconductor (pentacene). AFM images were analyzed using the free WSxM software from Nanotec Electronica. [67] Grazing incidence X-ray diffraction experiments were performed at the XRD1 beamline at the synchrotron ELETTRA, Trieste (Italy). Radiation with a wavelength of 1.4 Å was chosen. An incidence angle of 0.2° was chosen for the primary beam. The diffraction pattern was taken with a Pilatus-2 M detector from DECTRIS (Switzerland) by integrating for 0.2 s. Data acquisition took place at a fixed sample position. Calibration of the detector in terms of detector-sample distance and the associated tilt angles was performed by using a LaB 6 reference material filled in a capillary. Data processing was performed by using the in-house developed software GIDV; data from these images are merged and transferred from pixel space to reciprocal space using standard procedures.
Fabrication Process of p-Type OTFTs on Glass: Organic thin film transistors were fabricated in a bottom-gate, bottom-contact architecture according to the setup illustrated in Figure 3a. In Figure S1a (Supporting Information), the process flow is shown schematically.
The aluminum gate electrode was deposited by thermal evaporation of a 50 nm thick aluminum layer through a shadow mask at a rate of 1 nm s −1 , followed by anodization of the aluminum to create an 18 nm thick aluminum oxide. A detailed description of the anodization process can be found in ref. [68].
Prior to the thermal evaporation of a 50 nm thin Au layer, a SAM with a phosphonic acid anchor group and a thiol tail group (12-mercapto dodecylphosphonic acid, Sigma-Aldrich) was applied as adhesion layer. The source/drain electrodes were formed by photolithography employing the following steps: after deposition of a 50 nm thin Au layer, a positive photoresist (AZ1505, MicroChemicals) was applied by spin-coating, followed by a brief baking step. After UV illumination (λ = 365 nm, E ≈ 50 mJ cm −2 ) through a photomask (Compugraphics), the UV-exposed areas were easily removed by dipping the sample in a developer solution (AZ Developer from MicroChemicals) for 12 s. Then the predefined pattern of the photoresist was transferred to the Au layer by wet chemical etching of the nonphotoresist protected parts (Au etchant recipe: KI:I 2 :H 2 O/4 g:1 g:40 mL from MicroChemicals). Finally, the photoresist was removed from the substrate (mr-REM 500 from MicroResist). The photopatterned source-drain electrodes were briefly plasma (H 2 and O 2 ; applied successively) cleaned and treated with pyrimidine-containing SAMs as described in section "Preparation of SAMs." The H 2 and O 2 -plasma cleaning step prior to the SAM treatment also removed the adhesion SAM in the transistor channel (to obtain defined interface properties at the gate dielectric). Finally, a 40 nm thick layer of pentacene (from Sigma-Aldrich) was deposited by thermal evaporation through a shadow mask at a very low evaporation rate of 0.1 nm min −1 for the first 5 nm and 1.2 nm min −1 for the remaining 35 nm.
After production, all samples were protected from light and stored under argon atmosphere in a glovebox. The channel length of the fabricated OTFTs on the glass slides varied between 7.5 and 51 µm; the channel width was varied between 1 and 4 mm.

Fabrication Process of p-Type and n-Type OTFTs, Unipolar Organic Inverters, and Ring Oscillators on Flexible PC-Substrates:
The fabrication process of the bottom gate bottom contact flexible OTFTs, unipolar organic inverters, and ring oscillators is presented schematically in Figure S1b (Supporting Information).
As substrates, flexible 120 µm thick hard-coated polycarbonate films (1 in. × 1 in.) with a surface roughness as low as 0.9 nm (measured with profilometer from VEECO Dektak 150) were used. The gate electrodes were formed by photolithography employing the following steps: after deposition of a 50 nm thin aluminum layer (thermal evaporation on the precleaned substrates under high vacuum conditions), a positive photoresist (AZ1505, MicroChemicals) was applied by spin-coating, followed by a brief baking step. After UV illumination (λ = 365 nm, E ≈ 50 mJ cm −2 ) through a photomask (Compugraphics), the UV-exposed areas were easily removed by dipping the sample in a developer solution (AZ Developer from MicroChemicals) for 12 s. Then, the predefined pattern of the photoresist was transferred to the aluminum layer by wet chemical etching of the non-photoresist protected parts (Aluminum etchant: ANPE 80/5/5/10 from MicroChemicals). Finally, the photoresist was removed from the substrate (mr-REM 500 from MicroResist). A 18 nm thin Al 2 O 3 gate dielectric was formed by anodization. During the anodization step, all gates have to be electrically connected (realized by additional connection lines) and vias (vertical interconnect access; necessary for the ring-oscillators) have to be protected by a thin photoresist layer. After anodization, the additional gate connection lines were removed by a further photolithographic step. Source/drain electrodes were also fabricated by photolithography. Prior to the thermal evaporation of a 50 nm thin Au layer, a SAM with a phosphonic acid anchor group and a thiol tail group (12-mercapto dodecylphosphonic acid, Sigma-Aldrich) were applied as adhesion layer. For the photolithographic patterning of the Au layer, the same processing steps as for the gate patterning were used, albeit employing a different etching solution for the gold layers (etchant recipe: KI:I 2 :H 2 O/4 g:1 g:40 mL from MicroChemicals). The photopatterned source-drain electrodes were briefly plasma (H 2 and O 2 ; applied successively) cleaned and treated with pyrimidine-containing SAMs as described in sub-section "Preparation of SAMs." Finally, a 40 nm thick layer of pentacene for the p-type transistors, unipolar inverters, and ring-oscillators or a 60 nm thick C 60 (from Sigma-Aldrich) layer for the n-type transistors was deposited by thermal evaporation through a shadow mask. Very low evaporation rates of 0.1 nm min −1 for the first 5 nm and 1.2 nm min −1 for the remaining 35 nm were used for the pentacene films and a constant rate of 0.3 Å sec −1 for the C 60 films. After production, also all devices on PC-substrates were protected from light and stored under argon atmosphere in a glovebox. The channel length of the OTFTs varied between 3.5 and 52 µm and the channel width was fixed at 1 mm. The channel length of the diode-load and zero-V GS ring oscillators was 7.5 µm.
Determination of Surface Energy: Contact angle measurements with ultrapure water and diiodomethane were used to determine the surface energy with a KRÜSS DSA 100 ContactAngle Measuring System. The surface energy was calculated via the Owens-Wendt-Rabel-Kaelble method, using contact angles of different liquids with known disperse and polar fractions of the surface tension. [66] Electrical Characterization: Electrical measurements on OTFTs, inverters, and ring-oscillators were carried out in a glovebox under argon, using a parameter analyzer from MB-Technologies, a manual probe station from Suess, and a digital oscilloscope von Textronix (TDS2024). From the I D (V GS ) plot of the square root of the drain current as a function of the gate bias, the threshold voltage and the onsetcorrected semiconductor charge carrier mobility could be extracted (@V GS −V ON = −3 V). The mobility was obtained from the slope of a linear fit through the data in this plot, while the threshold voltage corresponded to the extrapolation of the line to zero current. The subthreshold swing is the inverse of the maximum slope of the (quasi) linear part of the subthreshold current.
Quantum-Mechanical Simulations: The simulations were performed using the Fritz Haber Institute ab initio molecular simulations package