Report

Conductance of a Single Conjugated Polymer as a Continuous Function of Its Length

See allHide authors and affiliations

Science  27 Feb 2009:
Vol. 323, Issue 5918, pp. 1193-1197
DOI: 10.1126/science.1168255

This article has a correction. Please see:

Abstract

The development of electronic devices at the single-molecule scale requires detailed understanding of charge transport through individual molecular wires. To characterize the electrical conductance, it is necessary to vary the length of a single molecular wire, contacted to two electrodes, in a controlled way. Such studies usually determine the conductance of a certain molecular species with one specific length. We measure the conductance and mechanical characteristics of a single polyfluorene wire by pulling it up from a Au(111) surface with the tip of a scanning tunneling microscope, thus continuously changing its length up to more than 20 nanometers. The conductance curves show not only an exponential decay but also characteristic oscillations as one molecular unit after another is detached from the surface during stretching.

A key challenge in molecular electronics (1) is the detailed understanding of charge transport through a single molecular wire (24). One of the first conductance measurements of a metal–single molecule–metal junction used a scanning tunneling microscope (STM); the STM tip made electrical contact to a single C60 molecule adsorbed on a metal surface (5). This result was soon followed by break junction experiments (68), in which an ultrathin metallic wire junction is mechanically broken, providing a two-electrode junction where a statistical number of molecules can be located and where conductance histograms are collected to reveal the average conductance of presumably single molecules in this molecular junction (9, 10). These two techniques demonstrated that electrons can easily tunnel through a molecule with a low energy gap between the highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO). This approach can be used to provide statistical information about the conductance of single or very few molecules. However, they do not allow the conductance of a single and long molecular wire to be determined as the function of the distance between the two contacts on the same molecule. We present an experimental procedure to measure the conductance of a single and the same molecular wire with a well-defined and defect-free chemical structure as a function of the distance between the two contact points on the wire while precisely imaging the molecule's conformation before and after the measurement with submolecular resolution.

Recently, it was reported that a single molecule can be picked up from a solid surface with scanning probe techniques by approaching the scanning tip toward the surface and subsequently lifting up a single or several molecules when retracting the tip. This method was applied to molecules (1117) by STM and to oligomers (18) or DNA strands (19) by atomic force microscopy (AFM), although the latter one is not capable of conductance measurements. With such a pulling manipulation, one end of a molecule is bound to the tip of an STM while maintaining surface adsorption at the other end. The pulling was mostly practiced at random on many molecules at the same time, with no characterization of molecular conformations in the junction before and after (1116). A few studies compared the statistically obtained conductances of an oligomer series of different lengths up to 7 nm (1315). However, in the controlled pulling experiment on a single perylene-tetracarboxylic-dianhydride (PTCDA) molecule, the length of the molecule in the junction was fixed to only 1 nm (17).

Our measurements require long π-conjugated oligomers adsorbed on a surface that cannot be deposited intact by conventional sublimation techniques under clean conditions because of their large molecular weight (20). To access long one-dimensional molecular chains, polyfluorene composed of conjugated fluorene repeat units was targeted by our recently developed in situ polymerization (21). For this purpose, we used dibromoterfluorene (DBTF) monomers, consisting of three fluorene units, carrying lateral methyl groups and a Br atom at each end (Fig. 1A). These terminal groups are dissociated from the terfluorene (TF) molecular core in the first activation step of our on-surface synthesis (21). At a surface temperature of 10 K (22), single DBTF molecules on a Au(111) surface appeared in constant-current STM as three intense lobes corresponding to the dimethyl groups (Fig. 1B). DBTF molecules adopt a zig-zag shape on the Au(111) surface, corresponding to the energetically favored alternate conformation of the methyl groups with respect to the molecular board (22). This result is in very good agreement with STM image calculations (Fig. 1C) that allow us to extract the exact position of each dimethyl group along a DBTF unit. Because of the low evaporator temperature during DBTF deposition, the Br atoms should still be attached to the TF core; a comparison between the observed and calculated images indicates that this is the case.

Fig. 1.

On-surface polymerization of DBTF to conjugated molecular chains. (A) Chemical structure of DBTF molecules with experimental (B) and calculated (C) STM images (2.5 by 4.5 nm) of intact molecules, with bright protrusions associated with the lateral dimethyl groups. (D) Overview STM image (80 by 120 nm, 1 V, and 0.1 nA) after on-surface polymerization. The produced long covalently bound molecular chains, i.e., polyfluorene, follow the herringbone reconstruction of the substrate. (E) STM image (5.9 by 3.6 nm) of a single polyfluorene chain end with its chemical structure superimposed (using a different scaling). The arrows indicate three identical (in the STM image and the chemical structure), newly formed covalent bonds between individual building blocks.

After heating the Au(111) surface up to 520 K for 5 min, the DBTF's Br atoms dissociate and covalent bonds are formed between different activated TF monomers, which diffuse randomly on the surface (Fig. 1D). The high mobility of the monomers, being a prerequisite for polymerization, is favored by the presence of lateral methyl groups on each monomer. Hence, this on-surface synthesis allows the formation of long, well-defined chains on the Au(111) surface; oligomer lengths greater than 100 nm were observed.

These polyfluorenes, carrying only methyl side groups, cannot be prepared by conventional polymerization processes in solution because of the absence of sufficiently solubilizing side chains. The homogeneous appearance of each molecular chain demonstrates the extreme regularity of their chemical composition, meaning that the newly formed covalent bonds are equivalent to the existing bonds, connecting the three fluorene units within each DBTF monomer, and that defect-free polymers were synthesized on the surface (Fig. 1E). The resulting polymers, commonly following the herringbone substrate reconstruction (Fig. 1D), are mobile enough on the Au(111) surface to be manipulated laterally with an STM tip for chains as long as 25 nm (22). Furthermore, the manipulation proves their high flexibility, enabling different curvatures of the chain without breaking the chemical bonds between the different TF monomers.

After selecting an isolated oligomer chain on the surface on the basis of a first STM image, this chain was pulled upward by the STM tip apex (Fig. 2A). For this purpose, the STM tip was first positioned at one end of the chain where, because of the Br dissociation during the on-surface synthesis, a chemical radical is presumably located. Subsequently, the tip apex was gently brought into close proximity of the selected chain to establish the electronic contact and then progressively retracted to lift it upward. The precise tip location on the molecular chain end is of great importance in this process, as evidenced by characteristic differences of the pulling success rate (22). The variation of the tunneling current I as a function of vertical distance z during this procedure was recorded, giving detailed insight into the vertical manipulation (Fig. 2B). After a characteristic I(z) exponential increase during the tip approach to the surface (23), a different I(z) dependence was recorded during retraction, when the molecular chain has been bonded to the tip apex. The current was much greater at the same value of z, when the tip is retracted very far from the surface, as compared to the case where no molecular chain was attached. Apparently, the tip-molecule interaction, which is less defined than the contact on the surface (that is imaged before the pulling experiment), is stronger than the chain segment's interaction with the substrate, keeping the chain attached to the tip during the pulling process. Although it is possible to lift entire chains from the surface, the tip-chain connection sometimes ruptures during the retraction (but not at constant tip-surface distance), which manifests as an abrupt drop of the tunneling current.

Fig. 2.

Lifting a single molecular chain with the STM tip. (A) Scheme of the chain pulling procedure: After contacting a molecular chain to the STM tip, it can be lifted from the surface in a ropelike manner upon retraction because of its flexibility and weak interaction with the substrate. (B) Tunneling current as a function of the tip height during a vertical manipulation (approach and retraction with different slopes are marked by arrows). The maximum current is limited in this experiment to 100 nA (by the current pre-amplifier). (C to E) STM images (25.4 by 13.7 nm) of the same surface area during a vertical manipulation series (the cross indicates the position of tip approach and retraction). The manipulated chain, extending beyond the lower image border, changes its shape during the pulling processes. The fixed chain at the upper image border serves as reference.

Complementary to the I(z) curve, STM images recorded before and after a vertical manipulation can aid in the visualization of the changes in the chain conformation and position on the surface. The high mechanical stability of the STM junction (lateral and vertical drift of less than 7 pm/min) renders such a vertical STM manipulation extremely reliable for conductance measurements of long molecular wires. STM images before and after two vertical manipulations are shown in Fig. 2C and Fig. 2, D and E, respectively. In each experiment, the STM tip picked up a single chain and dropped it down (after an arbitrary retraction distance), as monitored by the I(z) curve. The polymer in Fig. 2D is not imaged at the same position as before the pulling (Fig. 2C) but has moved to the left, following the pulled chain end in a ropelike manner. The consequence of the pulling-release sequence is that the initial linear conformation of the chain first adopts a left- and then a right-handed curvature with a large lateral displacement of the contacted chain end (Fig. 2D). Such a modification of the molecular conformation is hardly possible to obtain by lateral manipulation when the full chain remains adsorbed on the surface. Furthermore, we have not observed chain scission during any of the several hundreds of vertical manipulation procedures.

To measure the charge transport through a constructed surface–single molecule–tip junction as a function of the distance, the junction conductance variation G(z) was recorded at a 100 mV bias voltage while pulling a given molecular chain (Fig. 3A). At low temperature, G(z) reflects the ability of the chain to electronically couple the two electrodes of the STM junction through its more or less delocalized electronic structure as a function of the distance between the two electronic contact points on the chain (the surface and the tip apex) (24). Fitting G(z) by the exponential relationship G = Go × e–βz (25), a damping factor β of 0.38 ± 0.09 Å–1 is obtained experimentally. This value, a measure for the conductance of the molecular chain, is in very good agreement with the calculated one of 0.36 Å–1 [calculations were done with the Extended Hückel Molecular Orbital Theory–Elastic Scattering Quantum Chemistry (EHMO-ESQC) method (22)]. The average value of Go, which is determined by the contacts to the electrodes, is 300 nS (with a variation between about 10 and 500 nS). The z distance is shorter than the real length of the molecular chain between the two contact points because of the ropelike bending of the chain. Correcting β by taking into account the relation between z and the effective molecular wire length L, between the tip apex and the part of the molecular wire remaining adsorbed flat on the surface (Fig. 3C, inset), leads to z = L.cos φ, that is, β = 0.3 Å–1. The decay factor β is much less than the 2.4 Å–1 inverse decay length through a vacuum gap (indicated by a dashed line in Fig. 3A) (2), confirming that the electrons are tunneling through a single molecular chain. Other examples of molecular wires with smaller β values have been reported. However, they also display smaller HOMO-LUMO gaps (26) than the polyfluorene molecular wire investigated here [a gap between 3 and 4 eV has been determined for similar poly(dioctylfluorene) molecules (27)].

Fig. 3.

Conductance as a function of the length of the molecular wire. Experimental (A) and calculated (C) G(z) curves (equally scaled), both exhibiting characteristic oscillations with a period of zo (the decay of a vacuum gap is plotted for comparison). The experimental curve is composed of two data sets from measurements below and above about 20 Å, respectively, using different setups and thus ranges for current detection (each about four orders of magnitude). (B) I-V curves (of single wires and thus not averaged) at three tip-surface distances (2, 3, and 4 nm). (D) Schematic views of characteristic conformations during the pulling process, just before the detachment of another molecular unit (z1 = 10.2 Å, z2 = 17.2 Å, and z3 = 25.2 Å). The inset in (C) shows a sketch with the characteristic parameters z, L, and φ.

With our pulling technique, the current through the wire can also be measured as a function of the bias voltage at any fixed wire length, that is, tip-sample distance, by stopping the pulling process. The characteristic asymmetric I(V) curves of such single molecular wires, taken, for instance, at z of 2, 3, and 4 nm in Fig. 3B, confirm the presence of two electronic resonance thresholds. We attribute these to the first reduced and oxidized states of the molecular wire, corresponding in a first approximation to its HOMO and LUMO levels, with the HOMO closer to the Fermi level than the LUMO, which leads to asymmetric I-V characteristics.

The exponential conductance decay of a molecular chain with increasing length should follow a linear curve on a logarithmic scale (4). However, the one presented in Fig. 3A is not straight but shows characteristic oscillations with maxima at retraction distances of 10, 19, 29, and 41 Å (with an error of about ±3Å), that is, a typical oscillation period zo of about 10 Å. Using the EHMO-ESQC calculation method (22), G(z) was calculated as presented in Fig. 3C, taking into account the mechanical deformation of the molecular wire, its sliding over the surface during the pulling process, and the full junction valence electronic structure, including the interaction with the tip apex and the physisorption of the molecular chain on the Au(111) surface at the other end, which defines the electronic contact of the molecular wire with the surface.

At small z values, a steep G(z) decay is observed both in the experimental and the calculated curve, due to the tunneling through vacuum at low tip heights. Aside from its general exponential trend at larger z values, the calculated G(z) (Fig. 3C) shows regular oscillations with a zo between 6 and 8 Å, reproducing the experimental ones. Given the known optimized molecular chain conformation of a calculated pulling, we can attribute each oscillation maximum to a given monomer unit being adsorbed flat on the surface and about to be pulled away from the surface after a short sliding motion (see schemes in Fig. 3D). Hence, superimposed on the exponential decay trend, the contact conductance is oscillating because of the chainlike behavior that involves one monomer unit after the other being removed from contact with the surface by the pulling process. The maxima in the G(z) curve reflect the conformations slightly before the detachment of another chain unit (shown for three different heights in Fig. 3D) that in turn leads to an increased chain length between tip and surface and thus a drop of G(z). Interestingly, the electronic contact between one monomer of the chain and the surface is ensured by its π system and not by its methyl leg, which would cause a G(z) oscillation period of the order of the 2.88 Å Au(111) surface lattice constant (22). Instead, it must be of the order of the monomer unit length (8 to 9 Å), but not exactly equal to it because of the absent registry between the spatial extension of the monomer π system weakly overlapping with the surface and the Au(111) surface corrugation. The difference between the 10 Å experimental and our 7 Å calculated periods originates from the difficulty in theoretically reproducing in detail the full molecular mechanics of the total junction that arises from the large number of atoms involved. For example, upon pulling, the molecule is sliding but also snaking over the surface, a mechanical motion we have found difficult to reproduce in the calculation and which easily shifts the conductance oscillation by a few angstrom in the z direction.

Very often, the chain end to tip apex bond breaks before a complete vertical pulling, but in some cases, it remains attached to the tip until the entire chain is lifted up vertically. We measured the conductance of such extremely long molecular wires up to 20 nm in length. Figure 4A shows such a case, in which a single chain is lifted up at the left end (the position of the tip approach is marked by a cross). The I(z) curve during lifting (Fig. 4B) exhibits an abrupt decrease at the end of the pulling process and shows that the tip to surface electronic interaction through the molecular wire is lost exactly at a z = 20 nm distance. Hence, the chain is in a fully elongated conformation before the break, while being contacted vertically between the tip apex and the surface (Fig. 4C). To record the I(z) curve for large tip–molecular wire–surface junctions in Fig. 4B, it was necessary to increase the junction bias voltage continuously from –100 mV up to –4 eV during the pulling, compensating for the tunneling current exponential decay at low bias voltage. This result confirms that β is energy dependent, with a lower value in the energy gap when approaching the HOMO or LUMO resonance of the molecular wire (28). The increased conductance at higher bias voltages then compensates for the molecular wire length increase in Fig. 4B (a factor of 40 is not sufficient for the achieved current increase). Thus, such a setup allows the determination of the small conductance (8.6 × 10–13 S) of a single and the same molecular wire with 20 nm length (the conductance at small bias voltages cannot be measured over such a large distance, due to the extremely low current—below the detection limit—passing through the polymer in this case). In this regard, it would be interesting to prepare and study conjugated polymers with smaller HOMO-LUMO gaps. Such molecular wires should exhibit higher conductances and allow charge transport to be determined over even larger distances.

Fig. 4.

Complete detachment of a long single chain. STM images (20 by 15 nm) before (A) and after (C) a vertical manipulation process in which the entire chain is lifted from the surface. (B) Current curve as a function of the vertical tip distance from the surface (the corresponding bias voltage scale is shown below). The abrupt decrease at a tip height of about 20 nm reflects the disruption of the tip-surface connection through the chain and hence its complete removal at a vertical distance equal to its length do. (D) Schematic views of the tip-chain configurations before and after this pulling process. At the end, the chain is attached vertically to the tip, which is thus elongated by do, and therefore removed from the surface [see (C)].

Supporting Online Material

www.sciencemag.org/cgi/content/full/323/5918/1193/DC1

Materials and Methods

Figs. S1 to S6

References

References and Notes

View Abstract

Stay Connected to Science

Navigate This Article