Nicholas
Black
a,
Tonouchi
Daiki
b,
Michio M.
Matsushita
*b,
J. Derek
Woollins
*a,
Kunio
Awaga
b and
Neil
Robertson
*c
aEaStCHEM School of Chemistry, University of St Andrews, St Andrews, Fife KY16 9ST, UK
bResearch Centre for Materials Sciences & Department of Chemistry, Nagoya University Furo-cho, Chikusa, Nagoya 464-8602, Japan
cEaStCHEM School of Chemistry University of Edinburgh, Edinburgh, EH9 3FJ, UK. E-mail: neil.robertson@ed.ac.uk
First published on 7th December 2017
The magnetic, structural, conductivity and magnetoresistance properties of [Ni(quinoline-8-selenoate)2] ([Ni(qs)2]) have been studied. Despite the insolubility of the material necessitating its study as a powdered sample, a remarkably high conductivity has been measured. The conductivity is an order of magnitude greater than the thin-film processable thiol analogue previously reported and has been interpreted through the same space-charge limited conduction mechanism with charges injected from the electrodes. The introduction of selenium, results in a material with conductivity approaching metallic due to the enhanced interaction between adjacent molecules. Additionally, under an applied magnetic field, the material displays a negative magnetoresistance effect above 35% at 2 K. The effect can still be observed at 200 K and is interpreted in terms of a double-exchange mechanism.
Coinciding with these developments the field of spintronics, the utilisation and manipulation of electronic spin in order to carry information, has come into prominence.4–8 Beginning with the discovery of giant magnetoresistance (GMR) and with development throughout the 1980s it now underpins the magnetic data storage industry, giving rise to magnetic data reading and hybrid logic-storage devices.9,10 Investigation into electronic materials for spintronics and related fields, has yielded devices showing magnetoresistance,11 switching,12 memory effects13 and is aimed towards quantum information processing,14 and can be seen in inorganic, organic and single-molecule materials. It is therefore somewhat natural that the two fields should combine into the field of molecular spintronics.6,15,16
Molecular spintronics is still in the early stages of its development. Exploitation of spin in molecular systems demonstrates remarkable potential for merging the higher functionality of spintronics with the molecular design and processing advantages of molecular materials. The most common method of incorporating molecular materials utilises non-magnetic organic spacers between ferromagnetic materials in a spin valve.11,15,16 This approach is, however, hampered by difficult interface engineering.17–19 Magnetoresistance has also been observed in organic diode arrangements employing a thin film of molecular semiconductor, although the interpretation is still controversial and the sign of the effect changes with experimental conditions.20 A particularly appealing alternative is utilising the intrinsic GMR possible in a film of paramagnetic molecular materials. This approach has previously been used to generate negative MR up to 95% and relies upon a well-understood double-exchange mechanism, but has so far been limited to very few classes of molecule.21–23 ([TPP]Dicyano(phthalocyaninato)iron)2 was the first material of its type to be examined for the relationship between molecular magnetism and charge transport.23,24 A controllable giant negative magnetoresistance (GNMR) of up to 95% was achieved, using magnetic fields up to 15 Tesla. The benzo-TTF based molecule, bearing a nitronyl nitroxide radical group, ETBN, was discovered to demonstrate both conductivity and magnetism.25 The diselena analogue (ESBN) was the first example of the coexistence of both conductivity and magnetism based upon organic spins without the presence of inorganic magnetic ions, and a giant negative magnetoresistance effect of 70% was observed.26
Whereas the previously discussed examples have been studied as brittle single crystals, our observation of giant negative magnetoresistance in vapour processed thin films of [Nickel(quinoline-8-thiolate)2] ([Ni(qt)2]) opened up new avenues in the application and design of molecular materials for spintronic devices. [Ni(qt)2] is paramagnetic, with Ni(II) in a distorted octahedral geometry due to the formation of intermolecular S–N bonds orientated along the a-axis (Fig. 1a). Measurements revealed a drop in electrical resistance of greater than 60% in a magnetic field. This magnetoresistive effect was still discernible up to 200 K, albeit lower than 1%. The magnetic interactions are explained by the well-understood double exchange mechanism.27
In an effort to further explore and improve the intrinsic magnetoresistive effect observed in the [Ni(qt)2] material we have now studied the selenium analogue, Ni(quinolone-8-selenoate)2, [Ni(qs)2] (Fig. 1b). Sulfur and selenium have similar chemical properties however the larger selenium atom has more diffuse orbitals and so it was hypothesised that replacement of S with Se would increase orbital overlap and hence intermolecular interactions enhancing the previously observed effects.
Fig. 2(a) shows the magnetisation against magnetic field curve measured at 2.0 K, likely arising from a chain structure in analogue to [Ni(qt)2]. The metal has a formal oxidation state of +2 with a d8 electron count and would be diamagnetic in square planar geometry due to the crystal-field splitting (Fig. S3, ESI†). As the material is paramagnetic this provides a clear indication that the material synthesized is not in a square-planar geometry, but more likely the desired distorted octahedral geometry, as also seen for the sulfur analogue, since the octahedral crystal-field splitting gives two unpaired electrons for a d8 configuration (Fig. S3, ESI†). Fig. 2(b) shows the temperature dependence of the magnetic susceptibility for Ni(qs)2 measured at 1000 Oe. Fig. 2(b) inset shows the temperature dependence of the product of magnetic susceptibility and temperature (χT) for Ni(qs)2 measured at 1000 Oe. Above 40 K the χT value becomes almost constant after subtracting the temperature-independent components from χ − 1/T plot (Fig. S4, ESI†), and the value (Curie constant, 1.12) is consistent with the anticipated spin S = 1 with g = 2.12. The χT value increases with decreasing temperature below 30 K suggesting dominant ferromagnetic interactions, reaching 3.0 at 12 K. The sudden rise and then drop-off with decreasing temperature below 12 K suggests that antiferromagnetic ordering may exist between chains, or may be due to spin–orbit coupling, however without the single crystal structure reliable fitting of this is not possible. Overall however, paramagnetic S = 1 character of the Ni(II) centres, attributed to the formation of a chain structure, is apparent from the magnetic data.
Fig. 3 Magnetic field dependence of I–V at 2 K for [Ni(qs)2], the inset shows the non-linear I–V behaviour for the same device. |
From measurement of the resistivity temperature-dependence, an activation energy of Ea = 0.11–0.13 eV was calculated (Fig. S6, ESI†). This is very small and comparable to values measured for examples of single-component molecular conductors.30,31 The small activation energy suggests very strong intermolecular interaction, with a narrow band gap between occupied and unoccupied energy levels. This is in agreement with the visible spectrum measured by diffuse reflectance spectroscopy (Fig. S7, ESI†), which shows a broad absorption extending beyond the visible into the near-IR. Although an increase in conductivity was expected due to the enhanced interaction caused by the introduction of selenium, it is remarkable that the conductivity is so high in a powder sample where the effect of grain boundaries will likely be significant. A space-charge-limited conduction (SCLC) mechanism with carriers injected from the electrodes can be used to interpret the non-linear behaviour, as is the case in some other previously reported molecular examples.28,29 It is expected that, as is the case for [Ni(qt)2], the material possesses strong electron-donor property and that the injected carriers are holes and this is consistent with the measured electrochemical behaviour (Fig. S8, ESI†). Above the threshold voltage, the I–V behaviour can be modelled by an exponential law (I ∝ Vm+1) with m ranging from 4 to 18 at 200 and 2 K respectively, confirming that the current is dominated by a trapped-charge-limited conduction (TCLC) regime within the (SCLC) mechanism (Fig. S9, ESI†). In these conditions, the current is due to the bulk properties of the compound rather than the contact effects.
The same devices were also used to measure the magnetoresistance effect. A constant bias was applied and the magnetic field modified between −5 and 5 T. In order to ensure that each measurement of the current was made over a suitable region of the non-linear region an appropriate bias was selected for each temperature at which a measurement was made.
The experiment was setup with the magnetic field parallel to the current direction in order to avoid the appearance of Lorentz force. The observed giant negative magnetoresistance effect of [Ni(qs)2] is illustrated in Fig. 4. The magnetic field dependence of resistance of the [Ni(qs)2] complex reported as a percentage (R − R0T)/R0T, where R and R0T are the resistance with and without applied magnetic field, respectively. At 2 K the resistance decreases by more than 35%. This effect is still apparent (8%) at 50 K and still observable at 200 K although it has decreased below 0.5%. This is lower than that observed for the previous thiol analogue, however this may be partially explained by the measurements being made on a powder sample as opposed to ordered single crystals or evaporated multicrystalline thin-film.
Fig. 5 shows the calculated fully occupied HOMO which possesses predominantly ligand character. The HOMO is based largely upon the selenium atoms of the ligand and the positions on the ring ortho and para to the selenium, with some additional character on the other positions of the rings and the central Ni atoms. This is similar to what was previously calculated for [Ni(qt)2].
The calculated spin density is shown in Fig. 6 which, as anticipated, is predominantly located around the Ni(II) centre but with some character also on the ligand. This electronic structure resembles that of the thiolato analogue and is consistent with what would be expected for a system which follows the double exchange mechanism. As is the case for observations in nitronyl nitroxide-tetrathiafulvalene radicals, and the thiolato analogue, unpaired spins give rise to a Singly-Occupied Molecular Orbital (SOMO) of lower energy than the HOMO as necessitated by the double exchange mechanism. As shown in the density of states plot (Fig. S10, ESI†), contribution of the SOMOs of Ni(II) ions are found in the lower energy level (α − β), and the large energy differences (>0.5 eV) between α and β spins (spin-polarization) at the HOMO band is caused by the low-lying SOMOs. Hole transport via the HOMO orbitals based on the ligands is facilitated by the hole doping of the HOMO band at high applied electric field. The HOMO electrons on the ligand couple with the localized SOMO spins of the molecular unit which are aligned by an applied magnetic field. This results in the alignment of the spins of the HOMO electrons between neighbouring molecular units. As favourable spin alignment is maintained upon hole transfer this instigates hole hopping with reduced scattering. Due to co-location on a small molecular unit, the coupling between the HOMO and the SOMO spins is strong and consistent with the magnetoresistance effect persisting to high temperatures.
Footnote |
† Electronic supplementary information (ESI) available: Fig. S1: Mass spectrum of [Ni(qs)2]; Fig. S2: PXRD obtained for [Ni(qs)2]; Fig. S3: Square-planar and octahedral crystal-field splitting; Fig. S4: Magnetic susceptibility plots; Fig. S5: Powdered sample of [Ni(qs)2] on an interdigitated circuit; Fig. S6: T-Dependence of resistivity; Fig. S7: Diffuse reflectance of [Ni(qs)2]; Fig. S8: Cyclic voltammogram of solid [Ni(qs)2]; Fig. S9: log–log plots of I–V characteristics; Fig. S10: Calculated density of states plot for Ni(qs)2. See DOI: 10.1039/c7cp06273g |
This journal is © the Owner Societies 2018 |