Lecture 14. Molecular Switches - three terminal devices Two-terminal molecular switches have several important advantages, but they have also a serious drawback that needs to be overcome. The drawback is the lack of power gain – a remarkable property that a three-terminal device like a transistor has. For this reason, three terminal molecular switches have been pursued in recent years. Examples include the carbon nanotubes, 1 C60 molecules. 2 We will discuss some of them. The first one is the work by the Cornell group 3 and by the Harvard group. 4 Coulomb blockade and the Kondo effect in single-atom transistors Device Architecture The basic structure of the molecular switch is sketched in the figure below (figure 1). Preparation of the device begins with the thermal growth of a 30-nm SiO2 insulating layer on top of a degenerately doped Si substrate used as a back gate. Continuous gold wires with widths of less than 200 nm, lengths of 200–400 nm and thicknesses of 10–15 nm are fabricated on the SiO2 layer by electron beam lithography. The wires are cleaned with acetone, methylene chloride and oxygen plasma, and placed in a dilute solution of the molecules in acetonitrile for a day or more in order to form a self-assembled monolayer on the Au electrodes. The wires coated with molecules are then broken by electromigration, by ramping to large voltages (typically over 0.5 V) at cryogenic temperatures while monitoring the current until only a tunneling signal is present. This produces a gap about 1–2-nm-wide, across which a molecule is often found (Fig. 2).Fig. 1. A schematic diagram of the device. Fig. 2. A topographic atomic force microscope image of the electrodes with a gap (scale bar, 100 nm) The molecule Cornell’s group used a transition-metal complex designed so that electron transport occurs through well-defined charge states of a single atom. They studied two related molecules containing a Co ion bonded to polypyridyl ligands, attached to insulating tethers of different lengths. Changing the length of the insulating tether alters the coupling of the ion to the electrodes, enabling the fabrication of devices that exhibit either single-electron phenomena, such as Coulomb blockade, or the Kondo effect.The molecules that we have investigated are depicted in Figure 3a. They are coordination complexes in which one Co ion is bonded within an approximately octahedral environment to two terpyridinyl linker molecules with thiol end groups, which confer high adsorbability onto gold surfaces. The two molecules ([Co(tpy-(CH2)5-SH)2]2+ and [Co(tpy-SH)2]2+) differ by a five-carbon alkyl chain within the linker molecules (see Methods for details). These molecules were selected because it is known from electrochemical studies that the charge state of the Co ion can be changed from 2+ to 3+ at low energy. A cyclic voltammogram for [Co(tpy-SH)2]2+ adsorbed on a gold electrode in an acetonitrile/supporting electrolyte solution is shown in Fig. 3b, indicating that a positive voltage Vs + 0.25 V (measured against an Ag/AgCl reference) applied to the solution removes one electron from the ion. Similar results were obtained for [Co(tpy-(CH2)5-SH)2]2+. Fig. 3 The molecules used in this study and their electronic properties. a, Structure of [Co(tpy-(CH2)5-SH)2]2+ (where tpy-(CH2)5-SH is 4'-(5-mercaptopentyl)-2,2':6',2"-terpyridinyl) and [Co(tpy-SH)2]2+ (where tpy-SH is 4'-(mercapto)-2,2':6',2"-terpyridinyl). The scale bars show the lengths of the molecules as calculated by energy minimization. b, Cyclic voltammogram of[Co(tpy-SH)2]2+ in 0.1 M tetra-n-butylammonium hexafluorophosphate/acetonitrile showing the Co2+/Co3+ redox peak. Measurements Electrical characteristics of the molecule are determined by acquiring current versus bias voltage (I–V) curves while changing the gate voltage (Vg). Fig. 4 shows the result for the longer molecule, [Co(tpy-(CH2)5-SH)2]. The measurements were performed in a dilution refrigerator with an electron temperature of less than 100 mK. In about 10% of 400 broken wires they saw I–V curves as shown in Fig. 4. The current is strongly suppressed up to some threshold voltage that depends on Vg, and then it increases in steps. Fig. 4. I–V curves of a [Co(tpy-(CH2)5-SH)2]2+ single-electron transistor at different gate voltages (Vg) from -0.4 V (red) to -1.0 V (black) with Vg -0.15 V.Fig. 5 shows higher-resolution colour-scale plots of the differential conductance ( VI∂∂/ ) at low bias, as a function of V and Vg for three different devices. The darkest areas on the left and right of the plots indicate the regions of no current. The bright lines located outside these regions correspond to a fine structure of current steps visible near the voltage thresholds. The behaviour is the signature of a single-electron transistor (we have discussed in previous lectures). In this case the island is a single Co ion. For most values of Vg, the charge state of the ion is stable at low V (dark regions). An electron does not have sufficient energy to tunnel onto the island and therefore current is blocked (Coulomb blockade). The bright lines that define the boundaries of the Coulomb-blockade regions illustrate the tunnelling thresholds for transitions between charge states. Conductance in the vicinity of V = 0 is allowed at a value of gate voltage Vc where the Fig. 5. Colour-scale plots of differential conductance as a function of the bias voltage (V) and the gate voltage (Vg) for three different [Co(tpy-(CH2)5-SH)2] single-electron transistors at zero magnetic field.charge states are degenerate. We label the charge states as Co2+ and Co3+, in analogy with the electrochemical measurements, and this is supported by a spin analysis presented below. Additional lines in Fig.5 running parallel to the tunnelling thresholds indicate the contributions of excited states to the tunnelling current. Lines that end in the Co3+ (Co2+) blockade region correspond to excited levels of the Co3+ (Co2+) charge state. The pattern of excited states is qualitatively, but not quantitatively, similar from molecule to molecule. Typically, we observe several lines at energies below 6 meV. No additional lines are resolved between about 6 and 30 meV, at which point additional strong peaks are seen. A notable feature of the excited-state spectra is that the pattern of low-lying excitations is the same for both charge states of a given molecule. This, together with the small energy scale,