Elena
Bellido
ab,
Pablo
González-Monje
ab,
Ana
Repollés
c,
Mark
Jenkins
c,
Javier
Sesé
c,
Dietmar
Drung
d,
Thomas
Schurig
d,
Kunio
Awaga
e,
Fernando
Luis
*c and
Daniel
Ruiz-Molina
*ab
aICN2 Institut Catala de Nanociencia i Nanotecnologia, Campus UAB, 08193 Bellaterra (Barcelona), Spain. E-mail: druiz@cin2.es; Fax: +34 935813717
bCSIC - Consejo Superior de Investigaciones Cientificas, ICN2 Building, Campus UAB, 08193 Bellaterra (Barcelona), Spain
cInstituto de Ciencia de Materiales de Aragón, CSIC – Universidad de Zaragoza, 50009 Zaragoza, Spain. E-mail: fluis@unizar.es
dPhysikalisch-Technische Bundesanstalt (PTB), 10587 Berlin, Germany
eResearch Center for Materials Science and Department of Chemistry, Nagoya University, Nagoya 464-8602, Japan
First published on 13th September 2013
Direct measurements of the linear ac susceptibility and magnetic relaxation of a few Mn12 monolayers deposited on a μ-SQUID sensor are reported. In order to integrate the molecules into the device, DPN has been the technique of choice. It enabled the structuration of the molecules on the most sensitive areas of the sensor without the need for any previous functionalization of the molecule or the substrate, while controlling the number of molecular units deposited on each array. The measurements reveal that their characteristic SMM behaviour is lost, a fact that is attributed to molecular distortions originated by the strong surface tensions arising at the molecular interphases.
The archetypical case for all these years has been the dodecamanganese acetate cluster (Mn12-ac) and its derivatives.6 The number of successful techniques employed for their structuration on surfaces (with a few exceptions, under vacuum conditions) is considerably large.7 However, preservation of their SMM behaviour still remains a controversial and challenging issue. Other complexes, such as the Fe4 derivatives, provide promising alternatives to circumvent this difficulty.8 Though, the large number of studies aimed to integrate Mn12 on surfaces represents an excellent scenario to discuss some of the main challenges still lying ahead. Such information is crucial for an efficient design of future hybrid devices based on SMMs.
For instance, direct charge transfer to the substrate has been reported as one of the main mechanisms leading to the loss of SMM behaviour, upon chemical modification of the magnetic core. O'Shea et al. have observed a significant charge transfer between gold and a monolayer of Mn12-ac molecules even after protection of the magnetic core with large benzoate ligands.9 Sessoli et al. also observed a partial reduction to Mn(II) on Au(111) for two different Mn12 thio-derivatives.10 More successful were Pennino et al.11 These authors organized monolayers of Mn12 with longer thiol-terminated ligands on Au(111), and found that a large fraction of molecular cores retain their structural integrity. Such divergence of results between different experiments can be attributed to the role that the isolating organic layer plays in decoupling the core magnetic ions from the surface, as recently shown by Forment-Aliaga et al.12 and Voss et al.13 One of the best alternatives to overcome these chemical restrictions is to use substrates with different charge transfer properties. For this, Ternes et al. recently demonstrated the controlled deposition of fragile Mn12 SMMs on thin-insulating surfaces such as boron nitride (BN) and the feasibility for preserving their quantum magnetism.14 Some studies performed with different substrates, such as HOPG,15 glass16 or silicon,17 have also been reported.
Further studies have shown that preserving the core is a necessary but not sufficient condition to retain the SMM relaxation behaviour of Mn12 clusters. Local symmetry changes at the Mn sites, arising from the interplay of attractive and repulsive molecule–surface interactions, can also take place. This has been directly inferred from the modification of the spectral line shape of X-ray absorption spectra (XAS)18 or by atomic force microscopy (AFM) force-volume imaging experiments.19 Such symmetry changes prompted modification of local Jahn–Teller distortions in a rather uncontrolled manner and, therefore, modification of their magnetic relaxation through variations of the magnetic anisotropy constant D.
In any case, testing the magnetic properties of the Mn12 molecules on the surface clearly represents the best alternative approach to assess their behaviour. So far, most of the techniques applied to this end have been based on indirect probes, such as Scanning Tunneling Microscopy (STM), β-detected Nuclear Magnetic Resonance (β-NMR) or mainly X-Ray Magnetic Circular Dichroism (XMCD).10,11,17–20 However, direct measurement of the magnetic properties with a μ-SQUID sensor has remained so far elusive due to the insufficient sensitivity and the lack of experimental techniques for their proper integration. This represents a key challenge in the field nowadays.
In this work, the ac magnetic susceptibility of Mn12 benzoate (Mn12bz) is directly measured by depositing a few layers of the molecular material on the most sensitive areas of a miniaturized Superconducting Quantum Interference Device (μ-SQUID) susceptometer. The frequency-dependent blocking of the ac linear response is one of the characteristic traits of a SMM. A variation of the frequency and temperature over wide ranges enables the exploration of spin relaxation phenomena and provides information on the magnetic energy level schemes. Therefore, the implementation of these devices and their application to the study of molecular nanostructures provides a relatively simple and powerful method for directly characterizing SMMs deposited on a substrate. Integration of the molecules onto the optimum region of the device has been addressed by means of Dip-Pen Nanolithography (DPN). It has already been shown that DPN is an excellent technique to accurately deposit magnetic nanoparticles on a μ-SQUID without the need for any previous functionalization,21 while controlling the number of molecular units deposited on each array.22
In view of these considerations, our first experiments were devoted to finding a proper solvent for the structuration of Mn12bz on Si/SiO2 substrates. This is the material used in the sensitive μ-SQUID areas. Most of the above described requirements were met by a binary mixture of dimethylformamide (DMF):glycerol. DMF exhibits a high boiling point (b.p. 153 °C) and a relatively high viscosity, both enhanced by adding glycerol to the ink solution (b.p. 290 °C). The glycerol ratio was tuned by measuring for different solutions both the contact angle (CA) of the deposited drops, which depends on the electrical polarity of the mixture, and the ability to transfer the solution from the AFM tip to the substrate (viscosity).
CA of μl droplets deposited on Si/SiO2 substrates significantly increases with the glycerol concentration; values of ∼10°, ∼18°, ∼20° and ∼24° were measured for, respectively, pure DMF, and mixtures of DMF + 5% v/v glycerol, DMF + 15% v/v glycerol and DMF + 20% v/v glycerol (see Fig. S1 of the ESI†). A larger CA upon glycerol concentration increase allows for a better control on the resulting morphologies, because it helps maintaining the pattern shape intact during the drying of droplets. However, an increase in the glycerol concentration also induces a considerable increase of the viscosity that disrupts the ink transfer from the AFM tip. Therefore, the optimum DMF:glycerol ratio found, after different deposition experiments, is a compromise between 5 and 10% v/v. The stability of the molecular system in such a binary mixture was confirmed by spectroscopic experiments, more specifically by attenuated total reflectance-Fourier transform infrared spectroscopy (ATR-FTIR) and X-ray photoelectron spectroscopy (XPS) (see ESI, S2 and S3,† respectively).
A dark brown-colored solution of Mn12bz (10–20 mg mL−1) in a DMF:glycerol (95:5) mixture was freshly prepared and used as ink for coating the tip (for more details see the Experimental section). The coated tip was brought into contact with the surface to deposit drops of controlled size. Mn12 arrays can then be fabricated by traversing the tip over the desired area. Reproducible and uniform in size structures were obtained in this way. As a representative example, Fig. 1a shows a well-defined array of Mn12bz made of 20 × 20 dots spaced by 2 μm. The height profile analysis of the dots reveals a high uniformity in size, with an average diameter of 555 ± 39 nm and a height of 6 ± 1 nm. Considering the dimensions expected for the disk-like Mn12bz molecule, of approximately 2.1 × 1.2 nm (obtained by molecular modelling), and taking into account the presence of some residues remaining after solvent evaporation, the height of these dots is in agreement with the distribution of the molecules forming no more that 2–4 molecular layers. To illustrate the effect of the solvent on the final dimensions and structure of the arrays, Fig. 1b shows results obtained, under the same conditions, with a Mn12bz-free mixture of DMF and 5% v/v of glycerol. AFM analysis of the structures indicates a surface topography of up to 1 nm, attributed to the presence of some solvent residues that can be clearly differentiated from our material.
Fig. 1 Top: arrays fabricated by DPN on Si/SiO2 substrates. 3-D AFM topography image and height profiles recorded along the white dashed lines in the corresponding AFM images are shown. The specifications for each particular case are: (a) ink composition: Mn12bz (10 mg mL−1) in DMF and 5% v/v of glycerol; (b) ink composition: DMF and 5% v/v of glycerol. Bottom: (c) 3-D AFM topography image of the circular pattern specially designed to fit inside the pickup coil of the susceptometer. The external diameter is 25 μm and the distance between dots is 1 μm. The array was fabricated on Si/SiO2 using the same experimental conditions as in (a). (d) Optical image of the two pickup coils that correspond to the sensitive areas of the microSQUID sensor. The sample to be measured can be located on either the left hand side (A), wherein as an example a sample is indicated with a red circle, or the right hand side (B). |
Further control over parameters such as the ink loading of the tip, its movement over the surface and the tip-substrate contact time allowed us to gain control over the deposited structural motifs. As a representative example, Fig. 1c shows a pattern pre-designed to fit inside the internal diameter of the μ-SQUID (27 μm, see Fig. 1d). This pattern consists of concentric circles of dots separated by 1 μm.
Grazing incidence X-ray diffraction (GI-XRD) was used to determine the degree of crystallinity of the deposits. Experiments were performed on two different samples: a thin-film and a structured sample. The thin-film sample was prepared by drop casting Mn12bz ink on a substrate, while the structured sample was fabricated by DPN. For the latter, the size of the droplets was adjusted to ∼1 μm and the deposition process was repeated over extended areas in order to cover up to a few square millimetres, which are required to have enough sensitivity with this experimental technique (see ESI, S4–S5†).
The absence of any clear diffraction peak at low 2θ angles indicates that both samples are amorphous or that they possess a very small degree of crystallinity (see Experimental section). It could also be possible, especially in the case of the structured samples (arrays), that the signal is too small to be detected by the present XRD setup. Further evidence for the amorphous nature of Mn12bz deposits was obtained by high-resolution transmission electron microscopy (HR-TEM) experiments. It should be noted that HR-TEM images were taken with a minimum electron dose, in order to avoid damaging the sample with the beam. Typical HR-TEM images of the arrays are shown in Fig. 2. Even though several dots were inspected over a range of samples, no lattice fringes were observed in any of them (Fig. 2c, inset). These results suggest that Mn12bz droplets deposited on surfaces and exposed to ambient conditions tend to dry as an amorphous material. This is further supported by the presence of a diffuse halo in the selected area electron diffraction (SAED) patterns obtained from regions inside all dots under study (see Fig. 2d).
Fig. 2 HR-TEM images of (a) three different arrays of Mn12bz fabricated on a carbon-coated TEM grid and (b) magnification view of one of the arrays. Ink composition: Mn12bz (10 mg mL−1) in DMF and 5% v/v of glycerol. (c) Details of one of the dots of the array. (The inset shows the magnification view of a region inside the dot). (d) Electron diffraction pattern obtained from a region inside the dot. |
The two Nb coils forming the SQUID are circular-shaped with external and internal diameters of 35 μm and 27 μm, respectively (see Fig. 3). The coil wire has a cross-section of approximately 5 μm × 450 nm (a more detailed description of the sensor is provided in the ESI†). Five different Mn12bz samples were integrated into one of the μ-SQUID coils. Three of these samples were integrated by DPN, as described below, using the experimental conditions already optimized for the controlled deposition of Mn12bz on bare Si/SiO2 (a mixture of DMF and 5% v/v of glycerol). In addition, two “bulk-like” samples were also measured, for comparison purposes, under the same conditions. The first of these consisted of a micron-sized crystal of Mn12bz placed directly onto the coil (B1). The last sample (B2) was microcrystalline powder obtained after evaporation of a DMF solution of the Mn12bz. These two samples were embedded in apiezon-N grease and placed onto the μ-SQUID with the use of a home-made micromanipulator.
Fig. 3 (a) Optical image of the μ-SQUID susceptometer right after the deposition process (b) FE-SEM images taken at 45° tilt angle on one of the susceptometer's pick-up coils after integration of sample DPN1. (c) AFM topography image of a similar array fabricated on Si/SiO2 using the same experimental conditions used for DPN1. (d) Height profile measured by scanning the tip along the black dashed line in image (c). |
The first DPN sample (hereafter referred to as DPN1) consisted of a circular pattern of dots deposited inside the μ-SQUID coil of the SQUID1 sensor (see Fig. 3b and c) after completely removing the excess ink at the tip. In order to do so, several spots were fabricated by the freshly coated tip on an auxiliary Si/SiO2 surface until writing uniform dots in a very controllable manner (see Experimental section).
Afterwards, the tip was placed on top of the coil wire to create a circular pattern while keeping the tip-substrate contact time at 0.1 s. The process was repeated 2.5 times to deposit 1320 identical dots inside the coil. After the deposition, the sensor was characterized by scanning electron microscopy (SEM). The images show the effective and regular deposition of the molecular material inside the circular sensing area of the susceptometer (see Fig. 3b). The topography of the array was investigated by performing AFM on a replica fabricated on a bare Si/SiO2 surface (Fig. 3c).
The purpose of this method is to avoid damaging the sensor in the course of the AFM experiments. The analysis of the height profiles, shown in Fig. 3d, reveals the formation of highly uniform in size dots, with an average diameter of 249 ± 9 nm and a height of 8 ± 1 nm. These data suggest that each dot consists of approximately 3–5 Mn12bz layers. Taking into account the size and number of the dots deposited, it can be estimated that about 9.0 × 107 Mn12bz molecules were integrated inside the sensing area of the susceptometer (for more details on the determination of the number of molecules deposited see the ESI†).
In the subsequent deposition of sample DPN2 on sensor SQUID2, the amount of Mn12bz transferred to the device was increased by controlling the amount of ink removed from the tip in the previous step. For this, the freshly coated tip was brought into contact with the auxiliary Si/SiO2 surface for a shorter time than in the previous experiment. Afterwards, the tip was again positioned on the wire of one of the pickup coils by optical control and traversed over the surface, as before, forming a circular pattern with a tip-substrate contact time of 0.1 s.
In this case, the larger volume of the droplets makes them merge together on the substrate, forming a continuous thin film that covers the entire coil (sample DPN2). An optical image of the pickup coil captured during the deposition process is shown in Fig. 4a. The height profile of a replica fabricated under identical conditions but on a bare Si/SiO2 surface indicates a height of ∼8 nm (Fig. 4b). This profile is in agreement with that expected from a Mn12bz multilayer formed by 3–5 molecular layers. The number of molecules deposited on the sensing area is estimated to be of the order of 7.2 × 108 (see ESI†).
Fig. 4 (a) Optical image of the μ-SQUID susceptometer taken after depositing sample DPN2. (b) AFM topography image of a deposit formed on Si/SiO2 under identical conditions to those used for sample DPN2 and (inset) cross-section recorded along the white dashed line in image (b). |
Finally, when the same circular pattern is repeated without previously removing any excess ink from the tip, a large droplet completely covering the coil area with the Mn12bz ink is obtained (sample DPN3). For this, the tip was positioned on the centre of the coil of SQUID1 by optical control and brought into contact with the surface to deliver a sufficient amount of ink in a single step. An image of this “bulky” sample as it looked right after the DPN deposition process is shown in Fig. 5. The topography of the dry droplet was measured by AFM on a similar sample fabricated under identical conditions but on a bare Si/SiO2 surface. The topographic profile shows an average height of ∼15–20 nm, in agreement with the distribution of the molecules into up to 6–8 molecular layers.
Fig. 5 Deposition of a continuous Mn12bz thin-film covering the whole coil area (sample DPN3). (a) Optical image taken during the integration process. (b) AFM topography image of the replica sample fabricated, under the same conditions as in (a), onto a broken μSQUID susceptometer. (d) Height profile measured along the white dashed line in (b). |
Fig. 6 Output signals measured on empty μ-SQUID susceptometer SQUID1 (background) and on the same device after the deposition of several Mn12bz samples. Top: in-phase components. Bottom: out-of-phase components. |
The responses of two samples deposited by DPN (DPN-1 and DPN-3) are shown in Fig. 6C and D. The dependence of V′ and V′′ on the frequency differs strikingly from that found for the precursor crystalline material B1. In particular, V′ is close to constant above the background signal of the empty susceptometer, while the imaginary component V′′ vanishes. The same behavior is observed for sample DPN2 (see Fig. S7 of the ESI†).
The upturns observed above 104 Hz can be fully ascribed to the background signal of the μ-SQUID susceptometer. These results show that Mn12bz molecular nanomagnets deposited on silicon substrates remain in equilibrium up to very high frequencies, that is, their magnetic relaxation times become much faster than those in the bulk. It is worth noting that the enhanced magnetic relaxation is not associated with conversion of “slow” into “fast” ones. Instead, the susceptibility data show that drastic changes in the magnetic relaxation mechanisms occur for both isomers.
Before trying to discuss their physical significance, it is important to first make sure that these signals come indeed from the molecular samples. In order to check this question, we have computed, using the commercial simulation software COMSOL, the magnetic flux that an ensemble of Mn12bz molecules couples with one of the μ-SQUID pick-up coils. The model used for these calculations is schematically depicted in Fig. 7A that shows also the spatial distribution of the magnetic field derived from these calculations. The sample is a homogeneous disk, with a density equal to that of Mn12bz but with magnetic easy axes oriented at random, located inside one of the pick-up coils and subjected to the same magnetic field hac = 25 mOe as that used in the experiments. The number of molecules (N) is varied by changing the height h of the disk. For low N (<109), simulations performed on different sample geometries, e.g. on samples formed by discrete dots, give results that are in 30% agreement with those obtained for a continuous disk. Therefore, we conclude that approximating the sample with a continuous disk must be valid for the range of N values that we consider here.
Fig. 7 Top: theoretical color plot of the magnetic field intensity generated by a homogeneously magnetized cylinder (height h, diameter Di = 26 μm) located inside one of the μ-SQUID pick-up coils (thin black lines). These numerical calculations are used to determine the magnetic flux ϕ coupled with the device. Bottom: dependence of ϕ as a function of the number of Mn12bz molecules deposited on one of the pick-up coils. The solid line is calculated using the theoretical model described above and varying h between 8 nm (N = 109) and 80 mm (N = 1013). The prediction for N < 109 is a linear extrapolation of these results. |
The dependence of the equilibrium magnetic signal on N is shown in Fig. 7. The data are taken from the low-ω limits of frequency-dependent measurements, whereas the solid line represents the results of the above model calculations. In these calculations, the equilibrium susceptibility of each molecule is calculated as χT = g2μB2S2/3kBT, with g = 2 and S = 10. Considering the experimental uncertainties, calculations reproduce very well the measured values. Then, it can be safely concluded that the output signals we measured correspond to the magnetic response of the molecules deposited by DPN. The fact that the equilibrium susceptibility does not change much with respect to that expected for Mn12bz suggests also that the molecular magnetic moments (thus g and S) are preserved.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c3nr02359a |
This journal is © The Royal Society of Chemistry 2013 |