Current-Induced Hydrogen Tautomerization and Conductance Switching of Naphthalocyanine Molecules

See allHide authors and affiliations

Science  31 Aug 2007:
Vol. 317, Issue 5842, pp. 1203-1206
DOI: 10.1126/science.1144366


The bistability in the position of the two hydrogen atoms in the inner cavity of single free-base naphthalocyanine molecules constitutes a two-level system that was manipulated and probed by low-temperature scanning tunneling microscopy. When adsorbed on an ultrathin insulating film, the molecules can be switched in a controlled fashion between the two states by excitation induced by the inelastic tunneling current. The tautomerization reaction can be probed by resonant tunneling through the molecule and is expressed as considerable changes in the conductivity of the molecule. We also demonstrated a coupling of the switching process so that the charge injection in one molecule induced tautomerization in an adjacent molecule.

The concept of using single molecules as electronic components is well-established, with many examples on small numbers of or even individual molecules serving as memory elements, diodes, transistors, or switches (15). However, to construct more complex molecular devices requires that components are brought together and electronically coupled in a controlled manner. Most molecular switches are based on drastic conformational changes in the molecule (69); this is not compatible with the aim of controlling the coupling between the molecules. The development of molecular logic devices will also require single-molecule switches that can be coupled without compromising their function and that do not involve changes in the molecular frame.

Here, we present a single-molecule switch based on hydrogen tautomerization that meets these requirements. We operated and characterized the switch by low-temperature scanning tunneling microscopy (STM). The lowest unoccupied molecular orbital (LUMO) of a free-base naphthalocyanine (Fig. 1B) can have two orientations, depending on the position of the two inner hydrogens in the central cavity of the molecule (arrow in Fig. 1B). By increasing the bias voltage between the tip and the sample, a hydrogen tautomerization reaction can be induced by the tunneling electrons in the scanning tunneling microscope junction. This change is formally equivalent to the rotation of the molecule by 90° and causes a substantial change in the tunneling current measured at the scanning tunneling microscope tip positioned over the molecule. Because the switching is well-defined, highly localized, reversible, intrinsic to the molecule, and does not involve changes in the molecular frame, this class of molecules can be used as building blocks for more complex molecular devices such as logic gates.

Fig. 1.

Spectroscopy and orbital images of naphthalocyanine obtained by low-temperature STM operated at T = 5K. (A) Spectroscopy of naphthalocyanine on a NaCl bilayer on Cu(111) where the peaks correspond to tunneling into the LUMO (positive bias) and out of the HOMO (negative bias). (B) STM images at 1 pA, –1.6 V (left) and 1 pA, 0.65 V (right), as well as at low bias (1 pA, 0.05 V) compared with the calculated HOMO and LUMO of the free molecule. The lower center panel shows the structure model to scale where the arrow indicates the central hydrogen atoms that are along the horizontal arms. The STM images were obtained with a molecule-terminated tip. (C) Spectroscopy of naphthalocyanine on a RbI bilayer can resolve both the LUMO and the LUMO+1 (separated by ∼0.23 V). a.u., arbitrary units. (D) Corresponding orbital maps to (C); in this case, the dI/dV signal is in the constant-height mode, as compared with DFT calculations. All images are 30 by 30 Å2.

The molecules were adsorbed on ultrathin insulating films (NaCl, RbI, and Xe) on Cu single crystals and studied by low-temperature STM operated at T = 5 K(1013). Individual naphthalocyanine molecules were adsorbed at a sample temperature of T = 5 K, with the sample located in the scanning tunneling microscope. Bias voltages refer to the sample voltage with respect to the tip.

A tunneling spectrum (with current I and differential conductance dI/dV as a function of the bias voltage V) acquired on an isolated naphthalocyanine molecule on a NaCl(100) bilayer on Cu(111) (Fig. 1A) shows two resonances corresponding to the tunneling through the LUMO and highest occupied molecular orbital (HOMO) at a positive and negative bias, respectively. The molecule is adsorbed along the nonpolar [100] direction of the substrate; this is the most stable configuration on NaCl and RbI insulating films. The corresponding STM images acquired with a molecule-terminated tip at bias voltages corresponding to the resonances and to in-gap conditions are shown in Fig. 1B. As shown previously (12), the unperturbed molecular orbitals can be directly imaged by STM, and they compare very well with the calculated electronic wave-functions of a free molecule. The computed orbitals shown in Fig. 1 are based on density functional theory (DFT) calculations at the B3LYP/TZV level (14). The LUMO image allows for easy determination of the position of the inner hydrogens: They impose D2h symmetry on the molecule, which is reflected in the LUMO over the entire molecule. The “arms” with hydrogens show a single-lobe structure at the end, as opposed to the nodal plane along the other two arms.

The energy resolution in tunneling spectra can be increased by changing the insulating film to RbI (13), and we can resolve the peaks corresponding to the LUMO and LUMO+1 orbitals in the spectrum shown in Fig. 1C. These orbitals are separated by ∼0.23 eV and cannot be resolved separately in the dI/dV spectra on NaCl films. The difference in the energy of the LUMO resonance as compared with the measurements on NaCl is caused by the different work functions of the substrates (12). Constant-height dI/dV images acquired with a metallic tip (Fig. 1D) illustrate how the LUMO and LUMO+1 have the same nodal structure but are rotated by 90° with respect to each other. These experiments are again well corroborated by DFT calculations of the free molecule concerning both the orbital structure and the energy separation between the LUMO and LUMO+1 (0.19 eV). The separation of the LUMO and LUMO+1 images in the experiment is important for the direct assignment of the two different electronic states with the two different tautomers.

The hydrogen tautomerization can be induced by positioning the tip above the molecule and substantially increasing the bias above the LUMO resonance. Because the LUMO images are distinctly different for the two tautomers, the reaction can be directly monitored in the current signal or vertical-tip position in constant-height or constant-current mode, respectively. In these measurements, the current or vertical-tip position switches back and forth between two well-defined levels, as shown in Fig. 2A for a bias voltage of 1.7 V. When we lower the bias and image the LUMO at resonance, the two current levels correspond to a 90° rotation of the orientation of the LUMO. On the basis of the DFT calculations and optical and NMR spectroscopy (15), this observation can be assigned to changes in the position of the imino hydrogens in the central cavity; i.e., hydrogen tautomerization (Fig. 2B). A rotation of the whole molecule can be ruled out as we observe the switching of molecules at step edges and in arrays of molecules, where a rotation of the molecule cannot occur. In addition, we also observe switching on a Xe monolayer, where a rotation of the molecule by 90° would be incompatible with the symmetry of the surface.

Fig. 2.

Switching of a single naphthalocyanine molecule by the tunneling current. (A)(Left) Current-trace obtained at a bias of 1.7 V when the tip was positioned at one end of the molecule (red dot in STM images). (Right) Orbital images showing the orientation of the LUMO corresponding to the high- or low-current state (2 pA, 0.7V). (B) Schematic of the hydrogen tautomerization reaction responsible for the switching. (C) The bias dependence of the switching rate measured with a tunneling current of 1 pA on a naphthalocyanine molecule on a NaCl bilayer on Cu(111), showing an overall exponential trend with a slope of 165 mV/decade. (D) Spatial map of the switching rate for the hydrogen tautomerization reaction shown in (B) for a tunneling current of 1 pA at a bias of 1.825 V. For reference, the structure of the molecule is displayed to scale.

The dependence of the switching rate on the current is linear, and the distribution of residence times in the low- and high-current states is exponential. These observations are consistent with a statistically independent one-electron process. The switching rate increases with increasing bias voltage in a roughly exponential fashion (Fig. 2C, measurements in constant-current mode). Because no saturation is observed in the experimentally accessible range of resolvable switching rates, we think that the resonance responsible for the switching is at an even higher voltage (16). At voltages corresponding to the LUMO resonance (switch readout), switching was not observed. There is a slight but significant deviation from purely exponential dependence that may help to reveal the details of the switching process in conjunction with future theoretical efforts.

There is a strong dependence of the switching rate on the position of the electron injection into the molecule (Fig. 2D). Each pixel in Fig. 2D corresponds to a time trace with 100 switching events on average, giving in total ∼75,000 switching events over the molecule. The spatial map corresponding to the reverse reaction is similar but rotated by 90°. The white pixels in Fig. 2D at the center of the molecule do not indicate a zero switching rate; instead they signify that we cannot observe the tautomerization reaction in the time traces, because both tautomers result in the same current for symmetry reasons. These measurements were carried out in constant-current mode, and thus the switching rate is directly proportional to the quantum yield of the process. Figure 2D also shows that the switching probability is distinctly different on the two inequivalent arms of the molecule. This difference can be exploited to control the direction of the tautomerization reaction: If the current is injected in a position where the probabilities for switching back and forth are very different, the final position of the hydrogen atoms can be selected with a high probability (∼90%). However, the striking feature apparent in these plots is that the largest switching rate is achieved when the tip is above the far periphery of the molecule (i.e., >10 Å from the reaction site). This is in contrast with the typical inelastic electron-tunneling mechanism if no insulating film is present (1721). Because of decoupling provided by the insulating film, the lifetime of an additional electron in a molecular resonance is expected to be relatively long (13, 16, 22). To study this effect in more detail, we spatially mapped the switching rate on one, two, and three monolayers of RbI. On one RbI monolayer (shortest lifetime), the highest switching rate was achieved near the center of the molecule. On two and three monolayers of RbI, we observed similar spatial dependence, as shown in Fig. 2D. When the molecules were directly adsorbed on a Cu(100) substrate, we could not observe tautomerization.

The spatial dependence of the switching probability highlights the role of electron and/or energy transport within the molecule, making this system particularly interesting for related studies. For example, we can probe the coupling between neighboring naphthalocyanine molecules. We have coupled together three naphthalocyanines by simply bringing them close together by STM lateral manipulation (23), as shown in Fig. 3, A to D. In this configuration, we can switch one molecule by current injection through the neighboring ones. The switching yield in this experiment is determined by (i) the electron-transport properties through the molecule into which the current is injected, (ii) the coupling of the two adjacent molecules, and finally, (iii) the sensing molecule. Because energy transport in the absence of electron transport between the molecules seems extremely unlikely (by a dipolar or other mechanism), step (ii) corresponds to electron tunneling between the molecules. In this experiment, the molecules are weakly coupled, implying electron tunneling between the molecules but no substantial energy-level hybridization.

Fig. 3.

Examples of interacting and noninteracting assemblies of molecular switches. (A) A trimer of naphthalocyanine molecules on a NaCl bilayer formed by STM manipulation (image obtained at 2pA, 0.3V). (B to D) Current injection through the top or bottom molecules of the trimer [yellow dots in (B) and (C)] can cause the switching of the molecule in the middle, as shown by the LUMO images (2 pA, 0.8V). Images are 44 by 58 Å2. (E) Arrays of phthalocyanine molecules on a RbI monolayer on Cu(331). The in-gap image shows four phthalocyanine molecules that were isolated from a larger array of molecules (1 pA, –0.1 V). (F) Image at a bias voltage corresponding to the LUMO (1 pA, –0.5 V) (24). Current injection through the point indicated by a yellow dot induces only the switching of the molecule directly under the scanning tunneling microscope tip, as shown in the LUMO images in (G). Images in (E) to (G) are 50 by 50 Å2. (H) The two orientations of the LUMO of an isolated phthalocyanine molecule (2 pA, 0.35 V).

We can decrease the coupling of adjacent molecules, as shown for an array of phthalocyanine molecules with a center-to-center distance of 16 Å (Fig.3, E to H) (24, 25). In this case, the switching of molecules through neighboring ones was not observed, even though the total distance between the injection point (tip) and the reaction site was smaller than in the previous experiment. This finding rules out a field-induced mechanism for the switching.

Switching phenomena within individual molecules could potentially be used as nonvolatile memory with extremely high density, as has been proposed many times (5, 6, 16, 26). In contrast to previously investigated systems, the present molecular switch is planar, does not involve conformational changes at the periphery of the molecule, and is well-suited for use in self-assembled monolayers. Another advantage is that the symmetry inherent to the system implies that both positions of the switch have the same total energy and do not differ in binding to the substrate. Thus, we could observe this switching process, on a variety of insulating films (NaCl, RbI, and Xe), for two related molecules (phthalocyanine and naphthalocyanine) and for different charge states of the molecules. These measurements demonstrate the robustness of the process, and given that no changes occur in the molecular framework, it can be anticipated that the switching will also work with molecules embedded in all solid-state devices and in multicore porphyrin-class molecules acting as more complex devices.

Supporting Online Material

Fig. S1

References and Notes

View Abstract

Navigate This Article