David
Klar
*a,
Andrea
Candini
b,
Loïc
Joly
c,
Svetlana
Klyatskaya
d,
Bernhard
Krumme
a,
Philippe
Ohresser
e,
Jean-Paul
Kappler
ce,
Mario
Ruben
cd and
Heiko
Wende
a
aFaculty of Physics and Center for Nanointegration Duisburg-Essen (CENIDE), University of Duisburg-Essen, Lotharstraße 1, D-47048 Duisburg, Germany. E-mail: david.klar@uni-due.de; Fax: +49 (0)203 379 3601; Tel: +49 (0)203 379 1662
bCentro S3, Istituto Nanoscienze – CNR, via Campi 213/a, 41125 Modena, Italy
cInstitut de Physique et Chimie des Matériaux de Strasbourg, UMR 7504 UdS-CNRS, 67034 Strasbourg Cedex 2, France
dInstitute of Nanotechnology, Karlsruhe Institute of Technology (KIT), D-76344 Eggenstein-Leopoldshafen, Germany
eSynchrotron SOLEIL, L'Orme des Merisiers, Saint-Aubin - BP 48, 91192 Gif-sur-Yvette, France
First published on 13th May 2014
With element-specific X-ray absorption spectroscopy and X-ray magnetic circular dichroism we have investigated submonolayer coverages of TbPc2 and DyPc2 molecules sublimated on highly ordered pyrolytic graphite. We have studied the field dependence of the magnetization of the central lanthanide ion at very low temperatures. Even in zero applied magnetic field we still observe a remanence in the magnetization. Since there are neither intermolecular coupling nor magnetic interactions with the substrate, this remanent behaviour results just from single-ion anisotropy. On the very inert surface of graphite at temperatures between 0.5 K and 2 K the spin relaxation is slow enough to observe a memory effect in the timescale of the experimental measurements.
DyPc2 and TbPc2 were synthesized according to an earlier established procedure13 and analytical data confirm their intact structure (see ESI†). A submonolayer (0.5 ± 0.2 ML) coverage of LnPc2 was prepared in situ by thermal evaporation of the molecules onto the freshly cleaved HOPG crystal held at room temperature (for characterization details, see ESI†). The measurements relative to the TbPc2 molecule were performed at the DEIMOS29 beamline at the synchrotron SOLEIL, while DyPc2 molecules were studied at the SIM beamline at the SLS synchrotron, Paul Scherrer Institute, Villigen, Switzerland with the unique possibility to perform XMCD experiments at sub-Kelvin temperatures.30 The temperature of 0.5 K is crucial for the detection of the open hysteresis curve of DyPc2 because of its very low blocking temperature. Field-dependent measurements were performed to determine the SMM properties of the molecules. To study the spin relaxation after turning off the magnetic field, we carried out time-dependent XMCD measurements. With angle-dependent XMCD we studied the molecule orientation and the magnetic anisotropy.
Fig. 1 shows the angle-dependent spectra at the Tb and the Dy M4,5 edges for different X-ray incidence angles, normal (0°) and grazing (60° or 45°) at T = 2 K or 0.5 K and external magnetic field B = 6 T or 1 T. The black line represents the circular right polarization μ+, the red line the circular left polarization μ− and the green line the difference of the two spectra, the XMCD signal. The integrated white line (μ+ + μ−)/2 intensity at the M5 edge is normalized to 1. This normalization enables a direct comparison of the magnitude of the XMCD signals corresponding to the different angles. In the recorded spectra there are clear structural differences visible, especially at the M5 edge. However, the shape of the XMCD does not change with variation of the angle. Only the magnitude of the XMCD changes exactly with the cosine of the incidence angle. This behaviour of the magnetic anisotropy was observed for the TbPc2 molecules also on other substrates.14 The large anisotropy prevents the magnetic moments of Tb from turning into the plane of the Pc ring. The magnetization of the LnPc2 is still perpendicular, even in an external magnetic field of 6 T applied in the Pc-plane direction. Since we detect the magnetization just along the incident photons, the spectra only give the projection of the magnetization that is scaled with cos(θ).
The field dependence of the Tb magnetization is given in Fig. 2(a). The curves show the maximum of the M5 edge (1229.8 eV) normalized to the pre-edge (1220 eV). At a temperature of 2 K the magnetization of the Tb ion reaches the saturation value at about 1 T. Between the positive and negative saturation the magnetization curve is clearly open. The inset in this figure demonstrates that the curves do not cross at zero field. There is a remanent signal with a small coercive field of about 50 mT. The measurement of one complete curve takes more than one hour, but the relaxation at this temperature is slow enough to observe the remanent magnetization. The spin quantum tunneling that causes the typical butterfly shape is strongly reduced, but apparently the hysteresis curve is a bit more narrow at B = 0 T, indicating that spin tunneling is still present. The detailed origin of this behaviour is still unclear and falls beyond the scope of the present work. We note that a similar behaviour has been recently reported by another group for a crystallite sample of the same molecule.15 The spin relaxation is very sensitive to the temperature, leading to a vanishing remanence above 4 K.
Each curve in Fig. 2(b), showing the field dependence of the Dy magnetization, consists of only 26 data points. Each point belongs to an XMCD maximum at 1295.1 eV, taken from completely recorded μ+ and μ− spectra. In this way, each data point took approximately 30 minutes. Nevertheless, the opening of the magnetization curve is clearly visible for the black magnetization curve, demonstrating the long relaxation times at this low temperature. The remanent magnetization is also verified by the XMCD signal (Fig. 2b, inset) at B = 0 T. However, at T = 2 K (red line), no remanent magnetization is visible, confirming the lower blocking temperature of DyPc2 compared to TbPc2.
To make a quantitative statement about the relaxation time of the DyPc2 magnetization, we measured the time-dependent magnetization without an applied magnetic field (Fig. 3). Each triangle in this figure represents the normalized difference μ+ − μ− at the maximum XMCD intensity of 1295.1 eV. Approximately one data point was recorded every 90 seconds. The functional correlation between the XMCD intensity and time was fitted with a simple exponential function, while the maximum at the beginning was normalized to 1. The red curve fits the data points very well, demonstrating that the relaxation follows an exponential behaviour with a half-life of τ = 48 minutes. This time is in the range of typical XMCD measurements, giving us the opportunity to observe the magnetic remanence.
In conclusion, we have observed the remanent magnetization of both TbPc2 and DyPc2 molecules. Submonolayers of these molecules on the very inert HOPG substrate were deposited under UHV conditions to ensure that the interaction with the molecules’ environment is kept as low as possible. For the first time we could show the open hysteresis of vacuum deposited and well oriented LnPc2 molecules in an X-ray absorption study due to the very low temperature at which the measurements have been performed. This is a fundamental step for the possibility to use molecular memories in future devices. Because of the strongly suppressed spin relaxation at this temperature we were also able to determine a half life of the magnetization of 48 minutes.
Experiments were performed on the DEIMOS beamline at SOLEIL Synchrotron, France (proposal number 20110125) and on the SIM BEAMLINE (proposal number 20110200) at the Swiss Light Source, Paul Scherrer Institute, Villigen, Switzerland. We are grateful to the SOLEIL staff for smoothly running the facility. We thank Armin Kleibert at Swiss Light Source whose outstanding efforts have made these experiments possible.
Footnote |
† Electronic supplementary information (ESI) available: Bulk reference measurements and characterization of the molecules and the submonolayer. See DOI: 10.1039/c4dt01005a |
This journal is © The Royal Society of Chemistry 2014 |