Gas-phase ionization potentials ID (eV) and gas-phase electron affinities AA(eV), updated from , except where noted.
Ferrocene is a good electron donor, and as such has been used to test asymmetric conduction (rectification) in molecules that contain ferrocene. Of the five ferrocene-containing molecules that rectify (structures 11, 15, 19, 20, and 22), the last (22) exhibits a record rectification ratio, which should be a dramatic incentive for searching for more high-efficiency rectifiers.
- unimolecular electronics
- rectification ratio
- highest occupied molecular orbital
- Aviram-Ratner proposal of 1974
“Unimolecular electronics” (UME)  was born in 1974 with a theoretical proposal by Arieh Aviram and Mark Ratner (AR) for a one-molecule rectifier (or diode) of electrical current donor-bridge-acceptor (D-σ-A)  (Figure 1, structures 1 and 2): within that molecule D represents a π-electron-rich one-electron donor (D) moiety, σ is a short and saturated bridge of sp3-hybridized C atoms (between two and maybe eight C atoms long), and A is the electron-poor moiety that can act a one-electron acceptor. One small correction, AR had suggested that the first mechanistic step would move electron and hole from metal electrodes to the highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) and the second step would involve the relaxation of the excited-state zwitterion . The experimental direction of rectification for D-σ-A molecules has been shown to be “anti-AR” (Figure 1 structure 3): in step (1), under applied electric field, the neutral ground-state molecule D-σ-A forms an excited-state zwitterion D+-σ-A−; in step (2) the electron and hole are transferred to the metal electrodes .
The first rectifier (4 in Figure 1) was measured in 1990–1993 as a Langmuir-Blodgett (LB) multilayer between dissimilar metal electrodes by J. Roy Sambles (University of Exeter) and Geoffrey J. Ashwell (Cranfield University) [3, 4]. The asymmetric electrical current was confirmed at the University of Alabama (UA) as a LB monolayer of 4 between Al electrodes in 1997  and then between oxide-free Au electrodes in 2001 [6, 7].
As of 2015, 53 unimolecular rectifiers had been measured worldwide , 15 of which at the UA (Figure 1, structures 4–18 [7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20]). Also, 169 molecular wires were measured around the world . Several more rectifiers have been published worldwide since and many review articles on this subject have appeared [21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32].
The present article focuses on the use of one particular powerful one-electron donor in rectifiers: ferrocene.
In the 1980s, UME had hoped to develop useful molecular-scale (~2 nm3) devices for ultrahigh-density and high-speed industrial electronics. To interrogate such molecules (or monolayers of molecules), metal electrodes or nanoelectrodes (Al, Ag, Au, etc.) are used: this is sketched below and explained in detail in many review articles [8, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32]. UME learned how inorganic metals couple (associate with or bond to) single organic molecules and how one can reliably exchange electrons and photons with these molecules.
In the meantime the much wealthier and commercially driven electronic industry has made gigantic strides and has developed high-speed almost-nanoscale electronic circuits using inorganic semiconductors (Si, Ge, and GaAs). Therefore the original hope and promise of UME have been defeated. Nevertheless, UME has learned to interrogate and control individual molecules, and its present challenge is how to combine and exploit electronic, photonic, and spintronic functions in new ways.
The present review looks at how ferrocene-containing molecules have contributed valid and promising and most recently also very dramatic UME rectifiers: in particular molecules 11  and 15  already mentioned in Figure 1 and four other molecules shown in Figure 2, 19 [33, 34], 20 [33, 34], and 21 [33, 34] studied by the Whitesides group at Harvard University and 22  studied by the Nijhuis group at Singapore National University. As discussed below, molecule 21 should not, and does not, rectify.
Electrical measurements of rectifiers. Rectification can be measured with some difficulty at the single-molecule level, but more conveniently as a monolayer between macroscopic metal electrodes as a “metal | molecule | metal” sandwich . All molecules discussed here were studied either as a Langmuir-Blodgett monolayer (11 ) or as self-assembled monolayers (SAMs) with thiol terminations that could be bound covalently to either super-flat “template-stripped” AgTS or ATS electrodes (15 , 19 [33, 34], and 20 ) or PtTS electrodes (22 ). Most (but not all) rectifier measurements have been done with direct current [8, 32].
Candidate unimolecular rectifiers can be (i) electron donor molecules, (ii) electron acceptor molecules, or (iii) D-σ-A molecules [8, 32]. The search for organic rectifiers started in the era of quasi-one-dimensional organic metals with an enduring synthetic emphasis toward strong (easily oxidized) donor moieties D and strong (easily reduced) acceptor moieties A, connected by a covalently saturated and electrically insulating “sigma” (σ) bridge, forming a D-σ-A molecule. For instance, the proposed tetrathiafulvalene (TTF)-σ-tetracyanoquinodimethane molecule (Figure 1, structure 1) would have a presumed low barrier to form the corresponding excited zwitterion D+-σ-A−. Yet in 1974, a (weak donor)-σ-(weak acceptor) molecule (Figure 1, structure 2) was also proposed . Surprisingly, the recently studied (and dimensionally very tiny) molecule 18 (that resembles molecule 2) incorporates a weak electron donor D and a moderate electron acceptor A, yet is an excellent rectifier . We have also seen that a strong, easily oxidized donor like tetramethyl-para-phenylenediamine in molecule 16 blocks the current across the monolayer between −0.5 and +0.5 Volts (Coulomb blockade) .
Table 1 shows some relevant gas-phase ionization potentials ID for electron donors D and gas-phase electron affinities AA for electron acceptors A . It should be noted that ferrocene (Fc or Cp2Fe, where Cp is cyclopentadienyl) is as good an electron donor (i.e., has a relatively small ID value) as tetrathiafulvalene TTF, but not as good as N,N,N′,N′-tetramethyl-para-phenylenediamine (TMPD). Perylenebisimide (PBI) is as good an electron acceptor (i.e., has a similarly large AA value) as 7,7,8,8-tetracyanoquinodimethane (TCNQ).
Asymmetric placement of the electrophore in the electrode gap (“A” rectifier) .
Unimolecular processes depending on molecular energy levels (“U” rectifier) .
A recent fourth mechanism for rectification is asymmetric polarization (“AP” rectifier), when highly polar solvents can induce an asymmetric conductance of a symmetrical molecule between very asymmetric electrodes in a scanning break junction (SBJ) .
Purists would prefer pure-“U” rectifiers, requiring “S” = 0 and “A” = 0. For many molecules, for reasons of assembly, “U” and “A” effects are combined  (e.g., Figure 1 for structures 4, 5, 6, 10, 11, 14, 15, 16). For molecules 19, 20, and 22 in Figure 2, only the “A” effect is operative: the chromophore donor moiety (indicated as “D” and shown in red) yields rectifiers because it is asymmetrically placed within the “metal | monolayer | metal” sandwich. When the D moiety is in the middle of molecule 21, there can be and is no rectification [33, 34], as predicted .
Reversal of rectification: “Janus effect.” The molecules studied routinely at the UA rectify in the “anti-AR” direction, that is, intramolecular electron flow occurs from D to A (e.g., Figure 1 structure 3) [2, 8]. However, D-σ-A rectifiers 13, 14, 15, and 16 also show an additional “Janus effect”: at lower bias they rectify one way, and at higher bias (e.g., at ±2.5 V), they rectify the other way [17, 18]! At lower bias, AR rectification may involve only one energy level (e.g., LUMO); at higher bias, anti-AR rectification may involve both HOMO and LUMO. Such bias-switchable rectifiers may be useful!
Rectification ratio. The asymmetry in electrical current I is quantified by the rectification ratio:
where V is the applied bias or voltage. Typical RR values span several orders of magnitude; for the rectifiers 4–18 studied at the UA, RR(Vmax) is reported in Figure 1. The first rectifier, 4, had RR = 26 ; 16 has a large RR = 3000 . Why is the RR typically seen for unimolecular rectifiers (RR ≤ 103)  so much smaller than the RR for commercial inorganic pn junction devices (RR = 105–106) ? If low RRs were intrinsic to UME rectifiers, then traditional Ge, Si, and Ga As semiconductor physicists could safely look down at UME as a harmless curiosity, not as a competitor. But, as discussed next, a huge increase in RRs was imminent.
We next discuss rectifier 16, in which the electron donor moiety is the powerful electron donor TMPD instead of ferrocene (Figure 5). Figure 6 shows a surprisingly large room temperature Coulomb blockade : too much of a good thing, the TMPD oxidizes too easily and prevents current from flowing for a large bias range ! Beyond where the Coulomb blockade was operative, a relatively impressive RR ≈ 3000 is reached.
The Whitesides group (including Nijhuis) studied the rectification of self-assembled monolayers of thiol-containing molecules 19 and 20 and the non-rectification of the symmetric 22 , in sandwich “Au | SAM | Ga2O3 | GaIn” with a thorough effort to isolate the potential influence of the disordered Ga2O3 oxide that forms at the surface of the GaIn eutectic (without completely covering it) [33, 34, 35] (Figure 7).
Recent huge rectification ratio. A very dramatic result was published recently for the (“A-type”) rectifying monolayer sandwich “PtTS-S-C15H30-Fc-C≡C-Fc | EGaIn” (Figure 2, structure 22) consisting of a diad of ferrocene (Fc) donors (linked by an alkynyl-C ≡ C-), with a pentadecanethiol “tail” . This sandwich was studied between a bottom template-stripped electrode MTS (=PtTS, AuTS, or AgTS) and an EGaIn droplet top electrode. The new record is a very dramatic RR = 6.3 × 105 at ±3 Volts for PtTS (but much less for AuTS or AgTS) (Figure 8) . Also, the conductance “plateaued” around −2 Volts when the AgTS electrode was used .
The key improvements in  were (i) using Pt as the “bottom” electrode, because PtTS tolerates a larger bias range than AuTS or AgTS, (ii) a presumed efficient van der Waals contact between Fc-C ≡ C-Fc and EGaIn, and (iii) a “long enough alkyl tail” to get a very small reverse-bias current .
Also, light emission was measured (with blinking) for 22, with a broad peak at 1.7 eV (λmax = 730 nm), but only at the large negative bias V that corresponds to rectification: this emission was attributed to surface plasma polaritons excited distally within the Pt electrode after tunneling. Thus, the electrical excitation at large negative bias may have accessed the HOMO and HOMO-1 of Fc, but the energy is emitted neither directly (electroluminescence from Fc+ with an expected narrow energy distribution) nor indirectly (as lattice phonons), but indirectly and effectively, as surface plasma polaritons with a wide spectral distribution .
However, the measured RRs for alkanethiols are hundreds of times smaller than expected from careful theoretical simulations : this puzzle must be solved, so that measurements are not victims of unforeseen inefficiencies in the “metal | molecule” interface. The number of measured unimolecular rectifiers has grown dramatically, but their preselection as candidate rectifiers has been somewhat haphazard. Once the “metal | molecule” interface is brought under experimental control, better measurements may provide valid physical organic criteria to guide the design of the better unimolecular rectifiers of tomorrow.
There has also been a recent brief review on this exact topic ; for the sake of brevity, we refer the reader to the papers cited for other significant rectifiers containing the donor ferrocene [43, 44, 45, 46, 47, 48, 49].