In a straightforward application of molecular nanospintronics to quantum computing, single-
molecule spin transistors can be used to measure and control nuclear spin qubits. A jump in the
conductance occurs when the electronic spin inverts its polarization, and this happens at a so-
called anticrossing between energy levels, which in turn only takes place at a specific magnetic
field determined by the nuclear spin state. So far, this procedure has only been implemented for
the terbium(III) bis(phthalocyaninato) complex. Here we explore theoretically whether a similar
behavior is expected for a highly stable molecular spin qubit, the vanadium tris-dithiolate
complex [VIV(α-C3S5)3]2-. We consider such molecule sandwiched into a two-terminal device and
determine the spin-dependent conductance. We verify that the transport channel at minimal bias
1 voltage does not overlap with the occupied spin orbitals, indicating that the spin states may
survive in the conduction regime. We estimate some physical parameters to guide the
experiments, and verify the robustness of the theoretical methodology by applying it to two
chemically related vanadium complexes.
TOC GRAPHICS
KEYWORDS: molecular spintronics, spin qubits, electron transport
Molecular spintronics is an emerging branch of nanotechnology that builds on previous
knowledge acquired in its inorganic counterpart, inorganic spintronics, and adds the infinite
possibilities of tailoring and tuning the electronic properties of molecules.1,2,3 The field proposes
new materials to the data storage and data processing industries but also aims at offering novel
possibilities to quantum technologies. In a seminal experiment carried out in Wernsdorfer’s
group,4,5 the read-out of a two-qubit quantum state was performed by the measurement of the
conductance of a single terbium bisphthalocyaninato ([TbPc2]-) molecule, which acted as a
hyperfine-controlled single-molecule spin transistor. In this experiment, the two qubits reside in 159 the I = 3/2 nuclear spin quadruplet of a single Tb3+ ion, J = 6 with a ground doublet MJ = ±6.
The nuclear spin states are read out by a spin-dependent current: the conductance exhibits a jump
2 at an anticrossing between electronuclear spin levels, at a magnetic field determined by the
nuclear spin projection MI (see Figure 1, bottom panel). This effect is only measurable by
working at low temperature, where the molecule is in a fully polarized electronic spin state.
Recently, coherent quantum manipulation has been performed by the hyperfine Stark effect to
implement a minimal four-level Grover algorithm in a single molecule using this approach.6
While these experiments so far have been implemented only with [TbPc2]-, chemistry could offer
a palette of molecular structures matching or improving the possibilities of manipulation. There
are two necessary requirements: firstly, anticrossings between electronuclear spin states should
exist and be accessible, and secondly, the transport channel needs to be orthogonal to spin
orbitals to allow the survival of the electronuclear spin states in the transport regime. In addition,
further characteristics are desirable, such as a slow spontaneous relaxation of the electronic spin
to allow longer operation times, a higher nuclear spin to increase the number of qubits (n qubits
require 2n states), and a large quantum tunneling to improve the Landau-Zener transition
probability at the anticrossings and thus the operational speed.
A theoretical study is useful to explore the wide variety of chemical systems that can in principle
fulfill these requisites and thus be candidates for such kind of experiments. Among lanthanide-
based single ion magnets (SIMs) but beyond the phthalocyaninato sandwiches used in the
original experiments, other obvious candidates are polyoxometalate complexes7,8,9 because of
their electronic versatility, 10,11 and in particular DyW30, which has been experimentally proven to
function as a single-molecule rectifier.12 We did not choose this kind of systems for the present
study because our electronic structure method of choice, local and semi-local density functional
theory, does not provide a good description of ions containing 4f electrons (see SI for a broader
discussion on the process of selection of the molecule to study). A second chemical family of
3 potential interest is the recently investigated and smaller family of transition-metal based
molecular spin qubits.13,14,15,16 Among them, we decided to start our exploratory search with the
vanadium complex [V(α-C3S5)3]2- (figure 1): 13 This presents a high nuclear spin octuplet I=7/2,
meaning its 2x2x2 states give rise to 3 qubits for each single molecule.9 This complex presents a
tunneling splitting that is at least three orders of magnitude above the 1 µK found in [TbPc2]-,
improving the operational speed (see Supplementary Information). Furthermore, [V(α-C3S5)3]2-
has the potential to be attached to gold through its sulfur terminal atoms, and it presents a long
relaxation time that increases by cooling (the rapidly increasing times T1 = 0.05, 1, 20 ms are
achieved just by lowering the temperature from 40 K to 20 K and 10 K).13 Moreover, because of
its small size it is computationally inexpensive to deal with. Although its vanadyl analogue,
[VO(α-C3S5)2]2-, is expected to present a much more convenient magnetic field separation
between the anticrossings, we chose [V(α-C3S5)3]2- since preliminary calculations indicated that
it is more favorable from the point of view of spin-dependent conductance (see Supplementary
Information).
4 Figure 1: (Top panel) Calculated structure of [V(α-C3S5)3]2- after the structural relaxation (see
Supplementary Information) while attached to two Au(111) electrodes. (Color code: Au = dark
yellow, C = gray, S = yellow, V = green). (Bottom panel) Pictorial scheme of the hyperfine-
controlled single-molecule spin transistor for a I = 7/2 system, showing the anticrossing in the
nuclear splitting, where (at B1) an abrupt jump in the current intensity would be measured. Using
a slowly relaxing electron spin minimizes the direct transitions and thus improves the
signal/noise ratio.
In order to evaluate whether the apparent advantages of [V(α-C3S5)3]2- can be used in practice,
one needs to verify two remaining requirements (a) the conductance needs to be spin-dependent
at a reasonable gate voltage and, (b) for a chosen transmission peak, there must exist a
conduction channel that is compatible with the survival of the electronuclear spin state. This
second point is critical. In fact it is easy to imagine an organic radical or a metal complex where
the electron spin resides at the Fermi level, EF, and in that case the electron spin state would not
be well defined in the conduction regime.
In this work, we have evaluated these two conditions by theoretically studying the transport
properties of a two-terminal junction made of Au(111)-oriented electrodes sandwiching [V(α-
C3S5)3]2-. The theoretical transport calculations are based on the standard NEGF-DFT framework
using the SMEAGOL17 code, where atomic self-interaction correction (ASIC) is included in
5 order to avoid the pinning of the HOMO/LUMO to EF (details about the procedure are shown in
Supplementary Information).18,19
The first questions one has to answer are: does [V(α-C3S5)3]2- conduct current at all and, if so,
is the current spin-dependent; furthermore is a gate voltage required for driving a spin-dependent
current? Since the experiments are usually conducted for an infinitesimal source-drain voltage
and at a low enough temperature to ensure that the electronic spin is fully blocked, we have
calculated the zero-bias transmission spectra, and distinguished between spin-up and spin-down
transmissions. We have limited our calculations to an energy range corresponding to the gate
voltage range explored in the original experiments, namely up to ±2 V. In order to obtain an
initial, rough estimate of the energy overlap between the conductance pathway and the magnetic
orbitals, we have also computed the projected density of states (PDOS) on the vanadium ion.
These two molecular fingerprints are shown superimposed (see Figure 2, upper panel).
The transmission plotted in Figure 2 manifests two overlapping conductance peaks with very
strong spin asymmetry at and around EF, as well as four sharp peaks at accessible gate voltages,
which also present very strong spin asymmetry. At any of these peaks and in the presence of a
magnetic field, the expected conductance behavior of the spin-up configuration of the molecule
is very different from the expected behavior for the spin-down configuration. Therefore, we
could expect to see the kind of jump in the conductance accompanying a flip in the
electronuclear spin state, as experimentally observed at each anticrossing in the hyperfine-
controlled single-molecule spin transistor experiment.5 There is also a broad conductance peak
1eV below EF, but for that peak the spin dependence is more complex and the spin asymmetry is
weaker. Among all conductance peaks calculated for [V(α-C3S5)3]2-, the vanadium-projected
density of states is minimal at the Fermi level, thus we will be assuming no gate voltage is
6 applied throughout the remainder of this work.
In order to check the properties of the junction at finite bias, we have calculated the current-
voltage, I-V, characteristics at zero gate voltage, by varying the bias voltage both in the
conventional range [-1.0:1.0] V and zooming in at small bias [-1.0:1.0] mV as done in the
original experiments. In both cases, we have worked in the same low-temperature regime that
was employed experimentally to ensure full polarization of the electronic spin state of the
molecule: our calculations are therefore performed at an electronic temperature of T = 0.1 K. The
calculated I-V curve (see Fig. 2, lower panels) shows that the spin current is almost entirely
polarized for voltages between -1 and +1 V. The spin up component of the spin current is at least
three orders of magnitude more intense than the spin down one, meaning that the transmission
path is influenced by the molecular spin. The fact that the molecule conducts at zero gate voltage
and infinitesimal bias voltage is due to a lucky arrangement of the hybridized molecular orbitals
and the electrode work function; note that other analogous vanadium complexes where the same
calculations were performed did not display this feature, (see SI and Ref. 20). One should notice
that the large value found for the total molecular conductance is unsurprising, since we have
looked here at a situation of strong chemical bonding between the molecule and the electrodes.
13 Zadrozny, J. M.; Niklas, J.; Poluektov, O. G.; Freedman, D. E. ACS Central Science 2015, 9,
488. 14 Yu, C.-J.; Graham, M. J.; Zadrozny, J. M.; Niklas, J.; Krzyaniak, M. D.; Wasielewski, M. R.;
Poluektov, O. G.; Freedman, D. E. Journal of the American Chemical Society 2016, 138 (44),
14678.
13 15 Bader, K.; Winkler, M.; Van Slageren, J. Chem. Commun. 2016, 52, 3623. 16 Atzori, M.; Tesi, L.; Morra, E.; Chiesa, M.; Sorace, L.; Sessoli, R. Journal of the American
Chemical Society 2016, 138 (7), 2154.
17 A. R. Rocha, V. Garcia-Suarez, S. W. Bailey, C. J. Lambert, J. Ferrer, S. Sanvito, Nat.
Mater. 2005, 4, 335.
18 C. Toher and S. Sanvito, Phys. Rev. Lett., 2007, 99, 056801.
19 C. D. Pemmaraju, T. Archer, D. Sánchez-Portal and S. Sanvito, Phys. Rev. B, 2007, 75,
045101– 045116.
20 Cardona-Serra, S. et al. Work in preparation. 21 Akiba, K.; Matsubayashi, G. E.; Tanaka, T. Inorganica Chimica Acta 1989, 165 (2), 245.
