Article pubs.acs.org/Langmuir
Odd−Even Effect in Molecular Electronic Transport via an Aromatic Ring Tal Toledano,† Haim Sazan,‡ Sabyasachi Mukhopadhyay,† Hadas Alon,‡ Keti Lerman,‡ Tatyana Bendikov,† Dan T. Major,‡ Chaim N. Sukenik,‡ Ayelet Vilan,*,† and David Cahen*,† †
Department of Materials & Interfaces, Weizmann Institute of Science, Rehovot 76100, Israel Chemistry Department, Institute of Nanotechnology, and the Lise Meitner-Minerva Center of Computational Quantum Chemistry, Bar-Ilan University, Ramat-Gan 52900, Israel
‡
S Supporting Information *
ABSTRACT: A distinct odd−even effect on the electrical properties, induced by monolayers of alkyl-phenyl molecules directly bound to Si(111), is reported. Monomers of H2CCH− (CH2)n−phenyl, with n = 2−5, were adsorbed onto Si−H and formed high-quality monolayers with a binding density of 50−60% Si(111) surface atoms. Molecular dynamics simulations suggest that the binding proximity is close enough to allow efficient π−π interactions and therefore distinctly different packing and ring orientations for monomers with odd or even numbers of methylenes in their alkyl spacers. The odd−even alternation in molecular tilt was experimentally confirmed by contact angle, ellipsometry, FT-IR, and XPS with a close quantitative match to the simulation results. The orientations of both the ring plane and the long axis of the alkyl spacer are more perpendicular to the substrate plane for molecules with an even number of methylenes than for those with an odd number of methylenes. Interestingly, those with an even number conduct better than the effectively thinner monolayers of the molecules with the odd number of methylenes. We attribute this to a change in the orientation of the electron density on the aromatic rings with respect to the shortest tunneling path, which increases the barrier for electron transport through the odd monolayers. The high sensitivity of molecular charge transport to the orientation of an aromatic moiety might be relevant to better control over the electronic properties of interfaces in organic electronics.
■
INTRODUCTION Organic monolayers, down to a few nanometers thick, are of scientific interest as a test bed for molecular electronics as well as for their possible technological potential.1,2 Specifically, thin, dense (and in some cases ordered) self-assembled monolayers (SAM) provide a relatively controlled model system for studying the organic−inorganic solid interface. The overarching aim is to use synthetic chemical modifications of molecules to control and predict the chemical, physical, and mechanical properties of the modified surface. Alkyl chains show a tilt of the terminal group with respect to the chain’s longitudinal axis, which is dictated by the parity of n, where n is the number of carbons in the chain. This is known as an odd−even effect (OE) and is well documented in the alkyl-SAM literature. OE is most prominent in SAMs where binding and ordered packing induce a collective alignment of molecular orientation. The OE effect affects the molecular packing, wettability, tribological behavior, electrochemistry, work function, and electron transfer.2−7 These effects were seen for monolayers on metals and semiconductors (SC) alike and for monolayers made up of molecules with different anchoring groups (to match the substrate affinity).8 Because the quality of alkyl layers decreases drastically for short chains, no OE effect is observed for a monolayer on Si or Au with alkyl chains shorter than 2 nm.9,10 © 2014 American Chemical Society
Here we report an odd−even effect that occurs in much thinner molecular monolayers, 0.9−1.2 nm thick, with a noticeable effect in terms of wetting, the work function, and electrical transport. Besides parity, an OE effect is dictated by the following factors: (1) the geometrical conformation between the anchor group and substrate; (2) the monolayer uniformity, dictated by steric hindrance, the degree of interchain attraction, and the density of substrate binding sites; (3) the anisotropy in the chemical/physical properties of the terminal group of the molecules. The bond geometry differs, depending on the anchoring group and substrates. For example, Au−S and Si−C bonds have an sp3 molecular orbital configuration (angle of 104°) whereas Ag−S has an sp (180°) configuration, thus geometry dictates an opposite OE effect for identical thiol molecules adsorbed on either Au or Ag.11 Observing an OE effect also depends on establishing an averaged monomer orientation based on sufficient intermolecular interactions, such as van der Waals (vdW) interactions between the methylene units or π−π Received: September 4, 2014 Revised: October 21, 2014 Published: October 22, 2014 13596
dx.doi.org/10.1021/la503536f | Langmuir 2014, 30, 13596−13605
Langmuir
Article
stacking if aromatic units exist. vdW interactions between methylene units increase as a function of chain length; therefore, SAMs of short-chain molecules have weak intermolecular interactions, explaining why the OE effect is less readily seen.10 A notable exception is Ag, where a short lattice parameter dictates dense alkane packing for short chains. In this case, the OE effect of short alkyl thiols, as seen by the systematic variation in the contact angle, is for the Ag-alkyl chain monolayer system even more pronounced than for longer ones. Clearly, an OE effect requires a variation in properties with orientation. A well-known orientation-dependent property is the surface energy, which affects wetting and tribology. Electrostatic polarization can also depend on the orientation of the terminal group, as in the case of the terminal CH3 group for alkyl chains, where the net molecule-induced surface dipole and thus the work function of the surface vary with orientation.12 An aromatic end group is a natural candidate for expressing the OE effect because of the distinctive quadrupole moments of the phenyl ring, i.e., the ring electrons cloud (δ-) and ring’s core atom-edge polarities (δ+) (will be elaborated on in the Discussion section). Lee et al.7 could not observe any significant OE effect on the contact angle, packing density, and friction of monolayers made of an aromatic group bound to an alkane SAM bound to Au via a −C−S−Au link. It was argued that the distance between binding sites (4.9 Å)7 and tilted (due to vdW interaction) long alkyl chains forced the distance between rings to be beyond that required for effective π−π interactions (∼3.5 Å).7 Using biphenyls or polyaromatic groups has the advantage of stronger π−π attraction compared to single phenyl rings, but their bulkiness and rigidity could hinder dense self-assembly.13 Still, monolayers of biphenyl with short alkyl thiol spacers (one to six carbons) showed a clear OE effect on the orientation of the biphenyl14 and even on the binding distance and unit cell.15 There is a growing understanding that molecular interactions play a critical role in dictating net electrical transport.16,17 The OE effect induces a structural effect on otherwise identical molecules while introducing minimal perturbations. Reports on the OE effect on electrical transport properties are few. A noticeable example is the clear OE effect on electrical transport across SAMs of S−C10−18 chains on Au, with about 1 order of magnitude higher current for even-numbered chains than for odd ones.6 Because the even-numbered chains are also effectively thicker (more perpendicular), it was impossible to identify whether the OE effect was merely geometrical or fundamental such as a change in molecular energy levels or coupling to electrodes. In this study, we show that monolayers of CnBz (n = 2−5, Figure 1), bound via direct Si−C bonds to Si, show a pronounced OE effect on molecular orientation, contact angle, surface dipole, and charge transport. We note that n for CnBz equals m + 1 in CmS because the binding S atom is part of the spacer. The combination of a short alkyl spacer, π−π interaction, and Si substrate leads to very dense monolayers and a clear alteration in the tilt of the external phenyl ring. Focusing on short alkane chains and resulting layers that are thin (∼1 nm) enables us to study the electrical effect induced by the rings while maintaining a fairly high electronic transport efficiency across the monolayers, which might be useful in future applications.
Figure 1. Four molecules used to form SAMs in this study, sketched in the surface-bound form, resulting from the hydrosilylation of Si−H with the terminal double bond of the original reactant molecules (left to right): styrene, 3-phenyl-1-propene, 4-phenyl-1-butene, and 5phenyl-1-pentene, with acronyms as stated in the figure. C2Bz and C4Bz are the molecules with an even number (referred to as “even”) and C3Bz and C5Bz are those with an odd number (referred to as “odd”) of methylene groups in the chain.
■
EXPERIMENTAL SECTION
Organic Ligands. All reagents and solvents were purchased from Sigma and were of >95% purity unless stated otherwise. Styrene (99% purity) was passed through a basic alumina (Merck, 70-230 mesh) column under positive gas pressure (dry air); C3Bz and C4Bz (97% purity) were purified by distillation under N2 at 156−157 °C (C3Bz) or column chromatography through silica gel (Grace, 230−400 mesh) eluted with n-hexane (C4Bz). C5Bz was synthesized by the following procedure. Allyl magnesium bromide (1 M solution in THF, 21 mmol) was added dropwise at 0 °C to a solution of (2-bromoethyl-benzene (13.7 mmol) in THF and stirred at reflux overnight until being quenched with saturated NH4Cl solution. The product was extracted with Et2O, washed with saturated NaCl, dried over MgSO4, and concentrated in vacuum. The residue was purified by column chromatography on silica gel (eluted by hexane) to afford the product as a colorless oil. Sample Preparation and Adsorption Procedure. Si wafers (Virginia Semiconductors, USA) with ⟨111⟩ orientation, As doping of ∼1019 cm−3 (∼0.001 Ω·cm), polished on one side were used for all cases except for FT-IR experiments conducted on low-doped, n-type (P, 1014 cm−3) double-side-polished Si wafers. The deionized water (Milli-Q ultrapure) had a 1018 MΩ·cm resistivity. Solvents used for rinsing and cleaning the Si substrate were used as received with a purity of at least 95%. Mesitylene (98% purity) was distilled over sodium (160 °C, 1 atm) and degassed before dissolving the precursors in it. Si wafers were cut into 1 × 2 cm2 pieces and scratched with a diamond pen on their back side to maintain the smoothness of the opposite face with etching.18 Substrates were cleaned by 3 min of sonication in ethyl acetate and then sequentially rinsed with ethyl acetate, acetone, ethanol, and DI water, followed by plasma ashing for 3 min under a mixed flow of 3:2 Ar/O2. Plasma ashing is preferred over wet piranha etching19−21 because it is dry, efficient, quick, and not hazardous. The Si oxide was removed by a 15 min etch in a 40% aqueous NH4F solution that had previously been degassed for 1 h by bubbling N2 through it. After being etched, samples were rinsed with DI water and dried in a stream of dry N2. The plasma and wet etching were repeated twice to ensure that the resulting surface will be atomically smooth, contaminant-free Si−H. Immediately after being etched, samples were placed in the reaction cell filled with 10 mM precursor solution in 1 mL of mesitylene, as described earlier.22 Briefly, adsorption was photoactivated, usinga Hg lamp (254 nm band), for 3 h in a glovebox with O2 < 0.5 ppm. Samples were then rinsed with and sonicated in mesitylene and then immersed in boiling DCM, each for 3 min. Samples were dried under a stream of dry nitrogen (99.999%) and transferred to an N2-filled vial for storage until further characterization. Characterization was always done within 3 days but mostly immediately. To minimize surface oxide formation, sample exposure to ambient air was as short as possible. Preparing the monolayers via the photochemical rather than the thermal route leads 13597
dx.doi.org/10.1021/la503536f | Langmuir 2014, 30, 13596−13605
Langmuir
Article
Figure 2. Ball and stick models of conformations of monolayers attached to the ⟨111⟩ Si−H surface with 50% coverage, showing side views of (a) C4Bz and (b) C5Bz as representative configurations for even and odd numbers of methylene units per molecule. The tilt angle, α, is defined between the surface normal and the molecule’s skeleton axis (Table 1). (c) Top view of C2Bz showing that despite the 50% coverage, the nearest-neighbor distance, marked by dotted lines in (c), is 3.9 Å. Side and top views for other derivatives are reported in the Supporting Information. to denser monolayers, less Si oxidization, and more reproducible electrical measurements.23 Electrical Contacts. Top electrical contacts were made by thermally evaporated (BOC, Edwards FL, 400) Pb24 (Kurt J. Lesker, 99.9999%) at a rate of 4 Å/s to a thickness of 300 nm through a 100 μm mask and a base pressure of 3 × 10−6 Torr; a back ohmic contact to the Si was made on the bottom of the Si wafer by rubbing an eutectic mixture of In/Ga and rubbing it into the Si with a diamond pen. InGa was prepared by heating and mixing a 1:3 In/Ga (99.99%, Fluka) ratio until the mixture melted. The product was allowed to cool to create a eutectic mixture for further use. Characterization Tools. Ellipsometry. Data were obtained with a multiple-wavelength ellipsometer (550−1000 nm) at a constant angle of incidence of 70° (M 2000 V from J. A. Woollam Co. Inc.). Data analysis was done by fitting to a Cauchy model for organic monolayers (index of refraction between 1.460 and 1.483) using commercial software (WVASE32). Contact Angles. Water contact angles were measured with a RameHart manual contact angle goniometer. Static contact angles were obtained by averaging both sides of the drops (±1°) without the syringe contacting the droplet during measurements. Reported values and errors reflect the average and standard deviation of measurements on at least five samples. Fourier Transform Infrared (FT-IR). FT-IR spectroscopy was conducted with a Nicolet 6700 spectrometer using an MCT-A detector, averaging over 500 scans, in 4 cm−1 steps and freshly etched Si−H as a reference. Dichroism experiments to extract the molecular tilt25,26 from comparing the methylene’s absorption ratio of the p- and s-polarized beams in transmission mode were conducted with doubleside-polished, low-doped Si(111) wafers mounted on a shuttle that permitted background recording under identical environments and mounting the Si sample at a Brewster angle of incidence of 73.7°. High-sensitivity IR for the detection of benzene modes was done by pressing the Si substrate against a Ge-ATR accessory (HARRICK VariGATR, pressing force 200 N). Atomic Force Microscopy (AFM). A Multimode/Nanoscope-V (Bruker-Nano, Santa Barbara, CA, USA) SPM system was used in combination with Si probes (Olympus Micro Cantilevers, 2 N/m, 70 kHz, UV clean prior to measurements) under constant N2 flow purge (4% relative humility). The topography and phase images of an adsorbed monolayer were obtained simultaneously at a scan rate of 1 Hz with 512 pixel × 512 pixel resolution. The apparent molecular adhesion force of clean C4Bz and C5Bz molecular monolayers was obtained in peak-force tunneling AFM (TUNA) scanning mode with a peak force of 105°) and those of the much less hydrophobic aromatic systems (70−80°), similar to those reported for semihydrophilic surfaces that are covered with aromatic rings.3,38 Contact angle values show a clear OE alteration with a 7° higher contact angle for even than for odd numbers of methylene units. This can be understood because the former has a more vertical orientation with exposed C−H bonds than the latter where the phenyl rings are more inclined with the π orbitals pointing outward. The lower contact angles for odd methylene cannot be due to poorer monolayer quality because the positions of the FTIR CH2 stretch frequencies indicate the opposite, i.e., a slightly denser monolayer with odd than with even numbers of methylene units. The orientation of the different layers was further checked by measuring the local adhesion force with AFM measurements. The scan areas were 256 pixels × 256 pixels and 512 pixels × 512 pixels for different regions (2 × 2 and 1 × 1 μm2), and the apparent adhesion force was obtained from the average of histogram centers (2 pixel binning) over different scan areas. The apparent adhesion force is twice as large for evennumbered alkyl molecules monolayers than for the oddnumbered ones (even, 8 ± 1 nN; odd, 4 ± 0.7 nN). The increase in the force of adhesion can be understood from the less-tilted orientation and lower packing density of even compared to odd alkyl chains. As the tip approaches a nonrigid surface (with an even number of methylenes), it can penetrate between molecules and interact with surrounding molecules more effectively than is the case with the odd methylene layers.39 Monolayer Quality. Si−C binding occurs via a Si-radical reaction,23 with the double bond on the alkyl side reacting with the Si−H surface. Both the contact angle and FTIR data confirm that the layers attach to the Si surface by opening the alkene double bond and not via their aromatic ring. Binding via the phenyl ring would expose a double bond end with a typical contact angle of 90°,40 higher than observed here (Table 1). In addition, monolayers with a free CC−H2 have detectable21 IR absorptions around 2980 and 3080 cm−1, but none were observed here (Figure 3). The high quality of the monolayers is also evident from the residual amount of SiOx. XPS elemental analysis indicates that the oxidized Si level was smaller than 0.5 at. % (SI, Figure S1). Such a small value is less than continuous coverage and was shown to have a negligible effect on electrical transport across the Si-monolayer system.41 The monolayer thickness values, both those extracted from ellipsometer data and those calculated from XPS data (from the C 1s signal; SI, section 1b), are comparable to the theoretical (ChemDraw) values, deduced from the molecules’ structures (length of molecule with aromatic ring and zigzag conformation of methylene groups attached to it). We attribute the slight excess in experimental thickness values compared to theoretical
Table 2. Electrostatic Characterization for CnBz tool potential difference (eV) work function (eV)a band bending (eV)
Kelvin probe (± 0.025) Kelvin probe (± 0.025) XPS (± 0.01)
C2Bz
C3Bz
C4Bz
C5Bz
−0.23
−0.13
−0.21
−0.10
4.0
4.1
4.0
4.1
0.26
0.26
0.24
0.26
a
The work function was derived from the CPD using a work function (WF) of 4.2 eV for the Au probe (Experimental Section).
We find that the work functions of surfaces covered with monolayers of molecules with an even number of methylenes are consistently 0.10 eV higher than of those covered with a monolayer of molecules with an odd number of methylenes. A difference of 0.1 V is about 25% of the net dipole induced by alkyl chains on a Si(111) surface (∼0.4 eV).45 Although the type of highly doped Si that we used for these experiments is often referred to as degenerate, it actually can hold a non-negligible space charge, which, balanced by a surface counter charge, yields an electrical potential gradient that is known as band bending, BB. We extracted BB values from the difference between the XPS Si 2p binding energy values relative to a standard value of 98.74 eV for a flat band. For n-type Si, the BB is46 Eg − (Si 2p + 98.74), where Eg is the energy of the forbidden gap (band gap) and equals 1.12 eV for Si. The CnBz layers all showed BB values of ∼0.25 eV, irrespective of parity. This value is comparable to previously reported band bending 13600
dx.doi.org/10.1021/la503536f | Langmuir 2014, 30, 13596−13605
Langmuir
Article
for an alkane-modified Si surface,34 indicating fair surface passivation. Odd−Even Effect on Charge Transport. Having established the OE effect on surface potential, we moved to the most challenging part of this study, detecting an OE effect on electronic charge transport across the monolayers. The results from the previous sections showed that, notwithstanding their short lengths, CnBz monomers form high-quality monolayers, a condition for reliable transport measurements across macroscopically wide (100 μm) but ultrathin (∼1 nm) molecular monolayers. To minimize transport attenuation by Si, because of its semiconducting nature, highly doped (HD) nSi substrates were used for these measurements. In HD n-Si, the Fermi level, although still in the forbidden gap, is nearly at the conduction band bottom. In addition, the n-type doping assures a good match between the Si work function and that of Pb, the top contact used to complete the circuit for the transport measurements. Such a match minimizes the electrostatic barrier within the Si,24 which leads to minimal depletion of the Si and, thus, a thin space charge layer barrier to transport (∼10 nm instead of ∼1 μm).46,47 Because of the short length and partial conjugation, CnBz was expected and indeed showed high transport efficiency, i.e., high currents passed at a low applied bias voltage. Such high conduction required us to decrease the contact area, A, to increase the net junction resistance relative to the series resistance of the system (Rs = 10 to 100 Ω, independent of the contact area). For Hg and other metallic liquid electrodes, it is challenging to reduce the diameter of the drop to below ∼300 μm. Nevertheless, the contact size is readily reduced by evaporating the top contact using a shadow mask. Unfortunately, evaporating metal top contacts onto such a thin soft matter monolayer will nearly inevitably form shorts. We have recently shown that this is not so for evaporating Pb, which provides reliable electrical contacts to monolayers on Si,24,48 and here we made contacts by vacuum evaporation of Pb through a shadow mask with 100-μm-diameter holes. The current density−voltage (J−V) curves are almost symmetric with bias polarity (SI, Figure S2c), confirming an effective elimination of Si diode (rectifying) behavior due to the high Si doping levels and the band alignment between n-Si and Pb.45,47,48 Figure 4a shows the current as a function of positive bias on Pb for Si-monolayer/Pb junctions, formed with the four molecules. In tunneling-controlled transport, the current should decrease exponentially with molecular length, and for CnBz, n = 4 is more conducting than n = 3. This flip-over is not fully developed at room temperature (inset to Figure 4a), which we attribute to the increased molecular tilt of the odd methylene molecules with cooling. In addition, cooling decreases the contribution of tunneling into interface states of the Si, which, even at low densities, can obscure the direct tunneling transport characteristics.21,46,48,49 Therefore, we will focus on lowtemperature data. Another way to look at the transport data is to consider the exponential decay of the current (at a specific bias) with molecular length, as shown in Figure 4b for near equilibrium (−0.05 V) and at higher bias (+0.5 V). Figure 4b shows that an OE effect evolves with bias. Relative to an imaginary averaged trend, the odd methylene films are slightly more conducting at low bias (red X) but less conducting at high bias (black circles); the latter is just a different expression of the 3−4 switchover in the raw J−V curves of Figure 4a. Another bias effect is the range of change or the scale on the left and right Y axes of Figure 4b,
Figure 4. Transport characteristics across n-Si-CnBz/Pb junctions showing (a) current−voltage, (b) current−length, and (c) conductance−voltage plots. Current is shown as the current density (A/ cm2) and is averaged over all measured sets, and error bars are standard deviations. All plots refer to 6 K except for the inset to (a), which is for measurements at room temperature. Panel (b) shows a plot of current density on a semilog scale at +0.5 V (black circles, left Y axis) and at −0.05 V (red crosses, right Y axis) against the nominal molecular length, from the bonding Si to the most distant H, as computed by ChemDraw for an isolated molecule. Dotted lines are a guide for the eye. Panel (c) shows conductance, computed numerically (G = dJ/dV), from the experimental I−V data and normalized by forcing G(0 V) = 1.
which is much larger at higher bias. In terms of the basic relation of current attenuation with length J ∝ exp(−βL), where L is the molecular length (Å) and β is the tunneling attenuation factor (Å−1), we find β = 0.5 and 1 for V= −0.05 and +0.5 V, respectively. This bias-induced change in β is possibly because of a change in the relative contribution of the Si Schottky barrier and the molecular tunneling barrier. The applied bias flattens the Si barrier and by that decreases its weight and makes the molecular barrier more pronounced.21 In addition to the current magnitude, we examine the variation of the conductance with bias (G−V plot, Figure 4c).46,50 The conductance is computed numerically, G = dJ/dV, rather than measured directly and is normalized to emphasize bias variation rather than net magnitude. There is a prominent OE effect, where the conductivity of the even methylenecontaining films is much more bias-sensitive than that of the odd methylene ones. 13601
dx.doi.org/10.1021/la503536f | Langmuir 2014, 30, 13596−13605
Langmuir
■
Article
surprising. We suggest that it originates from the π−π interactions, which induce order, because such interactions have very specific geometric requirements of parallel-displaced benzene rings. This order results in an odd−even effect that is larger than with regular alkane monolayers. Therefore, we conclude that the combination of short alkyls with a single phenyl ring and a short lattice constant of the binding surface presents an excellent combination that maximizes the π−π stacking enhancement of monolayer quality. The resulting benefits are improved surface coverage and robust, stable monolayers, which are also very thin. This combination is specifically important for applications involving electronic transport, where strong coupling or high injection into the electrode is needed. Ring Conformation Effect on the Work Function. As deduced from the measurement of the IR dichroic ratio and supported by MD simulations, the orientation of the aromatic ring with respect to the surface normal is dictated by the odd or even number of carbons in the alkyl linker of the molecules. When aromatic ring quadrupole characteristics are considered, OE alternates the exposure of the aromatic ring’s electron cloud, which modifies the surface characteristics as shown by the work-function, contact angle and adhesion force measurements. Carbon-based monolayers generally act to reduce the work function of Si by merely blocking the electron density spillover (as a result of electron wave-function extension beyond the substrate).12,47,53 It requires adding a specific electron-withdrawing group to increase the work function (e.g., Br).19,21 The work function of 4.0 eV found for the surface onto which a monolayer with an even number of methylenes is grafted is similar to that of highly doped Si(111) modified by monolayers of 14−18 carbon long alkyl chains.45 The alkyl tether pushes electrons into the ring so that, in principle, the phenyl-alkyl should increase the work function. However, this effect is opposed by the high polarizability of the phenyl ring, compared to simple alkyls, which increases the effective dielectric constant of the monolayer and thus reduces the net change in electrostatic potential for a given dipole moment. The OE effect on the work function is even clearer if we consider that the work functions (from CPD) for the simple alkyl and even phenyl monolayer-covered surfaces are the same and those for the odd phenyl monolayer-covered ones are some 25% higher. The trend in the relative increased work function as the rings are oriented more parallel to the substrate is in qualitative agreement with a UPS-DFT study that found the orientation of thin films of sexithiophene to dictate their ionization potential. The ionization potential was 0.6 eV higher when the photoelectrons were emitted perpendicular to the ring plane (flat-lying thiophenes) than along it (for standing-up monomers).43 The reason is that in an array of flat-lying rings lateral layers of high and low electron density are formed that add up to form a net potential step. This is in contrast to a standing-up aromatic array where the electron density is alternating for each molecule and no net potential profile is formed (Figure 5). Origin of the OE Effect on Charge Transport. Below we discuss several plausible ways to explain the OE effect on charge transport and by elimination reach the conclusion that the main cause for the observed OE effect on transport is variation in the transport barrier induced by the different orientations of the phenyl with respect to the charge path. Considering the use of Si electrodes, the major indirect effect would be a variation in
DISCUSSION Improved Surface Coverage Induced by π−π Interactions. The CnBz monomers form surprisingly high quality monolayers on Si(111), considering the short alkyl chain and rigid phenyl group. We attribute this to efficient π−π stacking of the aromatic rings. The coverage values are ∼50 and ∼60% for even and odd methylene films, respectively, which is higher than the anchoring density of alkyl and alkoxy chain monolayers (42−52%).42 The positions of the antisymmetric methylene IR stretch (2923/2919 cm−1 for C4Bz/C5Bz) support the conclusion of a higher packing density observed for odd than for even methylene-based monolayers. The position of the methylene antisymmetric stretch suggests a monolayer density comparable to that of monolayers made by much longer simple alkyl chains on Si(111), where the methylene antisymmetric stretching vibrations are at 2922, 2920, and 2918 cm−1 for Si-(CH2)9CH3,4 Si-(CH2)11CH3, and Si-(CH2)17CH3 monolayers,21,34 respectively. The dynamic packing density of alkyl-based monolayers on Si was previously demonstrated when alkenyl chain monolayers were found to be denser than those composed of saturated alkyl chains.37 There are a few parameters that influence the adsorption density. First, in densely packed monolayers on silicon, only about half of the Si surface atoms are bound to the alkyl chains as a result of steric constraints (with the exception of C125,51 or alkenyl37 monolayers). Second, the length of the alkyl chain also affects the packing density because the vdW interchain attraction is proportional to the number of methylene units. This may indicate that the π system in the CnBz molecules helps in directing the grafting and facilitates the dense binding. The π−π stacking plays an important role in determining the packing geometries between molecules, which consist of conjugated systems, and this is even more so for aromatic system. This is relevant where the chain-to-chain distance is shorter than 5 Å. This constraint is not satisfied for Au and Ag lattices with 4.9−5.1 Å binding site distances, and in these cases, a herringbone packing arrangement is found.52 In contrast, our MD simulations (Figure 2c) show that even with only 50% coverage of the Si(111) surface atoms the shortest distance (diagonal) between adsorbed molecules is
