Encapsulation of single-molecule magnets in carbon nanotubes

Abstract

Next-generation electronic, photonic or spintronic devices will be based on nanoscale functional units, such as quantum dots, isolated spin centres or single-molecule magnets. The key challenge is the coupling of the nanoscale units to the macroscopic world, which is essential for read and write purposes. Carbon nanotubes with one macroscopic and two nanoscopic dimensions provide an excellent means to achieve this coupling. Although the dimensions of nanotube internal cavities are suitable for hosting a wide range of different molecules, to our knowledge, no examples of molecular magnets inserted in nanotubes have been reported to date. Here we report the successful encapsulation of single-molecule magnets in carbon nanotubes, yielding a new type of hybrid nanostructure that combines all the key single-molecule magnet properties of the guest molecules with the functional properties of the host nanotube. The findings may pave the way to the construction of spintronic or ultrahigh-density magnetic data storage devices.

ARTICLE

Received 3 Mar 2011 | Accepted 27 Jun 2011 | Published 26 Jul 2011 DOI: 10.1038/ncomms1415

Maria del Carmen Gimnez-Lpez1, Fabrizio Moro1, Alessandro La Torre1, Carlos J. Gmez-Garca2, Paul D. Brown3, Joris van Slageren1, & Andrei N. Khlobystov1

Next-generation electronic, photonic or spintronic devices will be based on nanoscale functional units, such as quantum dots, isolated spin centres or single-molecule magnets. The key challenge is the coupling of the nanoscale units to the macroscopic world, which is essential for read and write purposes. Carbon nanotubes with one macroscopic and two nanoscopic dimensions provide an excellent means to achieve this coupling. Although the dimensions of nanotube internal cavities are suitable for hosting a wide range of different molecules, to our knowledge, no examples of molecular magnets inserted in nanotubes have been reported to date. Here we report the successful encapsulation of single-molecule magnets in carbon nanotubes, yielding a new type of hybrid nanostructure that combines all the key single-molecule magnet properties of the guest molecules with the functional properties of the host nanotube. The ndings may pave the way to the construction of spintronic or ultrahigh-density magnetic data storage devices.

Encapsulation of single-molecule magnets in carbon nanotubes

1 School of Chemistry, University of Nottingham, University Park, Nottingham NG7 2RD, UK. 2 Instituto de Ciencia Molecular, Universidad de Valencia, Parque Cientco, Paterna 46980, Spain. 3 Division of Materials, Mechanics and Structures, Department of Mechanical, Materials and Manufacturing Engineering, Faculty of Engineering, University of Nottingham, University Park, Nottingham NG7 2RD, UK. Present address: Institut fr Physikalische Chemie, Pfaffenwaldring 55, D-70569 Stuttgart, Germany. Correspondence and requests for materials should be addressed to A.N.K. (email: [email protected]) or to J.v.S (email: [email protected]).

NATURE COMMUNICATIONS | 2:407 | DOI: 10.1038/ncomms1415 | www.nature.com/naturecommunications

2011 Macmillan Publishers Limited. All rights reserved.

ARTICLE

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms1415

Single-molecule magnets (SMM) possess a wide range of functional properties unique to this class of materials, including magnetic bistability1, quantum tunnelling of magnetization2,3,

and quantum coherence4,5. These properties are unattainable in other types of magnetic molecules, including endohedral fullerenes, such as N@C60, which, however, exhibits much longer electron spin relaxation times68 than those of SMM. The properties of SMM can be exploited for applications in molecular electronics9, spintronics10, data storage devices11 and quantum information processing12. However, any practical application of SMM requires their interaction with the macroscopic world, to allow read-and-write processes. A possible approach to this end is deposition of SMM on conductive surfaces, which, however, oen leads to SMM decomposition13,14.

Carbon nanotubes (NT) oer an excellent solution to connecting SMM to the outside world, without attendant problems of deterioration of the SMM functionality. The state of the SMM in such a hybrid NTSMM material can be controlled electrically, because NT are excellent electrical conductors14. Such electrical control is essential for local addressability for data storage, spintronics or quantum information processing applications15,16. Furthermore, NT are chemically and thermally very robust, which is highly advantageous for device fabrication17. Finally, the nanotube that encapsulates the SMM also protects the guest molecule from the surroundings that would otherwise induce strong decoherence, which is highly detrimental for quantum information processing applications18. The magnetic bistability of SMM, essential for data storage applications, derives from a combination of two factors11. The intramolecular superexchange interactions give rise to a well-dened ground state with high spin. For example, the ground state of the archetypical SMM, [Mn12O12(O2CCH3)16(H2O)4] or Mn12Ac for short, has a spin of S = 10. Secondly, the zero-eld splitting parameter D is large and negative, leading to a preferential magnetization direction, the easy axis. These two factors combined result in a large energy barrier for magnetization reversal, which gives rise to slow relaxation of the magnetization and concurrent magnetic bistability.

The shape of M12Ac can be described as a discoid with a diameter of 1.6 nm and a height of 1.1 nm (Fig. 1a). Considering the minimum van der Waals gap of 0.3 nm required between the surface of the molecule and the inner surface of NT17, the small

est diameter of NT capable of hosting Mn12Ac is 2.2 nm. Therefore,

insertion of Mn12Ac, as indeed other SMM whose dimensions tend to be even larger, into single-walled carbon nanotubes (SWNT; typical diameter range 12 nm) is not feasible, and thus wider multiwalled carbon nanotubes (MWNT) (internal diameter range 550 nm) should be used as host-structures for SMM. Indeed, all previous attempts of combining SMM and SWNT resulted in composite structures where the molecules, adsorbed on the nanotube surface, were exposed to the detrimental inuences of the environment (atmospheric oxygen and water, or ambient light), oen leading to fast degradation of their magnetic properties1921.

Here we report the preparation and study of a hybrid material in which Mn12Ac SMM are encapsulated inside MWNT (Fig. 1b).

We show that the SMM functional properties are fully retained, and that the molecules undergo a large degree of orientational ordering inside the nanotube. This ordering is advantageous for addressing purposes and for controlling electronic properties of the nanotube.

ResultsEncapsulation of SMM into carbon nanotubes. We chose graphitized MWNT (GMWNT), produced by catalytic chemical vapour deposition at 2,800 C, with a length of 1050 m, and a mean internal diameter of 6.5 1.8 nm (Supplementary Figs S1S4), because these GMWNT are sufficiently wide for insertion of SMM, and are completely free of residual catalyst that complicated the analysis of the magnetic properties of other nanotube-based structures20.

Mn12Ac is by far the most-studied SMM and exhibits a blocking temperature as high as 4 K, which is much higher than that of the vast majority of other SMM11. To allow encapsulation of Mn12Ac

SMM, the GMWNT were pre-treated with concentrated nitric acid followed by heating in air, yielding open GMWNT with an average length of 400 200 nm. We have used supercritical CO2 (scCO2) for the transport of the SMM molecules into the nanotubes. scCO2 is an excellent carrier uid for the encapsulation of molecules, because the small size of CO2, its low viscosity, high diusivity and zero surface tension allow it to penetrate the nanotubes without hindrance, enabling insertion of the desired guest species2224. The mild lling

conditions, with temperatures not exceeding 40 C, allowed for the successful insertion of Mn12Ac, which is far too fragile to be transported into GMWNT through the gas or molten phase. The resulting hybrid material, Mn12Ac@GMWNT (1), was carefully washed with acetonitrile (a good solvent for Mn12Ac) to eliminate any Mn12Ac molecules outside the NT (Supplementary Figs S5 and S6).

The nanotubes in material Mn12Ac@GMWNT (1), as observed by transmission electron microscopy (TEM), comprise discrete, free-standing entities (Fig. 2a). There is no indication of Mn12Ac

aggregates outside the NT in Mn12Ac@GMWNT, whereas in a control sample before washing (Supplementary Fig. S7), these aggregates can be clearly seen. The higher magnication micro-graph (Fig. 2b) shows a series of parallel lines corresponding to the sidewalls of the GMWNT, and dark contrast attributable to clusters of individual Mn12Ac molecules positioned within the

GMWNT cavity (Fig. 2d). Indeed, energy-dispersive X-ray analysis (EDX) spectra taken from lled nanotubes using a 4 nm e-beam probe unequivocally prove the presence of manganese inside the nanotubes (Fig. 2c). It is noted that Mn3O4, an expected decomposition product of Mn12Ac, which would exhibit a characteristic crystal lattice structure in TEM, was not observed in this sample (Supplementary Fig. S8), thus conrming that intact Mn12Ac molecules are present inside the nanotubes, consistent with the magnetic measurements described below. Under the inuence of the e-beam in the TEM, the Mn12Ac was observed to move within the nanotube cavity (Supplementary Fig. S9), indicating that the molecules are held by van der Waals forces only, with no covalent bonds formed between the SMM and the nanotube sidewalls.

Top-down 0.3 nm

1.6 nm

1.6 nm Side-on

1.1 nm

0.3 nm

1.6 nm

Figure 1 | Chemical structure of Mn12Ac SMM and schematic representation of its encapsulation in a carbon nanotube. (a) Balland stick structure of [Mn12O12(O2CCH3)16(H2O)4] (Mn12Ac), with its dimensions indicated by double-headed arrows. Carbon, hydrogen, manganese and oxygen atoms are shown as grey, light-blue, purple and red circles, respectively. (b) Schematic representation of the innermost layer of a GMWNT hosting Mn12Ac molecules.

NATURE COMMUNICATIONS | 2:407 | DOI: 10.1038/ncomms1415 | www.nature.com/naturecommunications

2011 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms1415

ARTICLE

a c

C

Intensity (a.u.)

Mn/O

Cu

Mn Cu

Mn

0

2 4 6 8 10

5.4 nm

Energy (keV)

b d

Figure 2 | Structural and compositional characterization of Mn12Ac@GMWNT. (a,b) Conventional bright eld and phase contrast transmission electron micrographs of GMWNT lled with Mn12Ac (1)

(black scale bar is 100 nm white scale bar is 5 nm). (c) EDX spectrum of the selected area in the GMWNT (inset, circled), conrming the presence of Mn-containing molecules within the nanotube (Cu peaks are due to a copper TEM support grid). (d) Schematic representation of the packing of Mn12Ac molecules in a nanotube.

5

Intensity (a.u.)

110

101

211

002

321

411

200

200

*

10 15 20

2 ()

25 35 40

100

80

60

Weight (%)

40

20

30

0 200 400 600

T (C)

800 1,000

Figure 3 | Preferential interactions of SMM with GMWNT interior revealed through control experiments. Powder X-ray diffractograms(a) of the composite material where Mn12Ac is adsorbed on the surface of closed (untreated) GMWNT before (orange) and after (blue) washing with acetonitrile, conrming the efcient removal of Mn12Ac molecules fromthe nanotube surface. The peak labelled with an asterisk is assigned to the nanotube. (b) TGA of Mn12Ac@GMWNT (1) (red) and closed GMWNT after washing (blue).

Our encapsulation procedure combined with a washing step with acetonitrile ensures that all SMM are inside nanotubes, and that no molecules remain on the nanotube surface. A series of rigorous control experiments involving untreated (sealed) GMWNT were performed to ascertain the efficacy of the washing procedure, using exactly the same conditions as for the preparation of Mn12Ac@

GMWNT (1) (Supplementary Fig. S10). The composite of Mn12Ac

with sealed GMWNT showed a clear signature in the powder X-ray diractogram attributable to the presence of Mn12Ac clusters adsorbed on GMWNT surface, before washing (Fig. 3aorange). Aer washing, the only feature remaining in the diractogram corresponded to the atomic lattice of sealed GMWNT (Fig. 3a blue). Thermogravimetric analysis (TGA) of the sealed GMWNT aer washing showed that no metal-containing material remained

aer heating above 600 C (Fig. 3bblue). Thus, both X-ray diraction and TGA data indicate that all the Mn12Ac molecules have been removed from the nanotube surface in the washing step. In contrast, the hybrid material Mn12Ac@GMWNT (1) subjected to the same washing procedure did not lose its guest Mn12Ac molecules, as conrmed by TGA (Fig. 3bred). These experiments demonstrated that Mn12Ac interacts stronger with the interior of GMWNT than with its exterior because of a greater contact area of the disc-shaped molecule with the concave side as compared with the convex side of the nanotube cylinder, which ensures eective encapsulation of SMM into GMWNT driven by van de Waals force.

The weight loss of the Mn12Ac@GMWNT (1) in TGA proceeded in three distinct steps (Fig. 3bred). The rst step corresponded to decomposition of Mn12Ac with the loss of all coordinated

ligands and, hence, the consequent formation of Mn3O4. The second step corresponded to the catalytic oxidation of the GMWNTs in the presence of the Mn3O4 formed inside the nanotubes, and the third step corresponded to thermal oxidation of the GMWNT themselves (that is, sections of nanotubes that are not in direct contact with Mn3O4)25,26. The residual weight remaining at 1,000 C was attributed to Mn3O4 that does not decompose further. The rst step was also observed in the TGA of pristine, unconned Mn12Ac (Supplementary Fig. S11), thus supporting the fact that the Mn12Ac molecules remain structurally intact upon insertion into the GMWNT (Supplementary Figs. S12S17; Supplementary Tables S1S3).

Dierential scanning calorimetry (DSC) measurements showed essentially the same Mn12Ac thermal behaviour before and aer treatment with scCO2 (Supplementary Figs S12 and S13), again conrming that the molecules are transported and encapsulated in the GMWNT in their intact form. It is interesting to note that Mn12Ac molecules inside the nanotubes decompose at a slightly higher temperature (ca. 240 C) as compared with free molecules (ca. 220 C). This indicates that the connement of SMM within the nanoscopic cavities of GMWNT enhances their thermal stability and resistance to oxidation in air. These ndings demonstrate that Mn12Ac@GMWNT (1) fabricated in supercritical uid is a clean, well-dened nanostructured material, with all of the SMM contained inside the nanotubes.

Magnetic behaviour of SMM in nanotubes. The magnetic properties of SMM inserted in nanotubes have been extensively investigated in our study (Fig. 4). A control sample of free, non-encapsulated Mn12Ac (2), treated with scCO2 under the same conditions as

used for the preparation of Mn12Ac@GMWNT (1), was also studied to ascertain the inuence of nanotubes on SMM properties (Supplementary Figs S16 and S17). Alternating current (AC) magnetic susceptibility measurements recorded for Mn12Ac@GMWNT (1)

(Fig. 4a,b) show an out-of-phase ( ) signal (Fig. 4b) that is a manifest signature of slow relaxation of magnetization. Furthermore, both the in-phase () and out-of-phase ( ) signals are strongly frequency dependent, indicating that the observed slow relaxation is due to the SMM behaviour, that is, the presence of an energy barrier for relaxation of the magnetic moment. It is worth noting that the observed frequency dependence of the AC signals follows an Arrhenius law, = 0 exp(E/kBT), where is the relaxation time of magnetization, T temperature, E activation energy and kB Boltzmann constant, which allows calculation of the energy barrier (E/kB)

for the relaxation process of the magnetization of molecular magnets in nanotubes (Supplementary Fig. S19). Importantly, these AC magnetic susceptibility measurements conrmed that the Mn12Ac

molecules remain intact inside the nanotube, with their functional magnetic properties fully preserved (for Mn3O4, the maxima in

would occur at much higher temperatures, and would not show such a pronounced frequency dependence). To our knowledge, this is the rst example of intact SMM that have been encapsulated

NATURE COMMUNICATIONS | 2:407 | DOI: 10.1038/ncomms1415 | www.nature.com/naturecommunications

2011 Macmillan Publishers Limited. All rights reserved.

ARTICLE

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms1415

0.35

0.300.25

0.0 100

104

[R(H)-R(0)]/R(0)

dM/ dHM(a.u.)

(a.u.)

(a.u.)

[R(H)-R(0)]/R(0)

0.150.100.050.00

8 6 4 2 0

H (T)

2 4 6 8

0.20

1 (0) 1 (0)

1 (90) 1 (90)

3 (0) 3 (0)

3 (90) 3 (90)

5.0

1.0

1.5

103

103

103 102 101 100 H (T)

1Hz

2Hz

5Hz

10Hz

20Hz

50Hz

100Hz

200Hz

500Hz

1000Hz

1500Hz

10

2 3 4 5 6

T (K)

7 8 9 2 1 0

Energy (a.u.)

0H (T)

1 2

Figure 4 | Magnetic characterization of Mn12Ac@GMWNT. In-phase (a) and out-of-phase (b) components of the AC susceptibility of Mn12Ac@

GMWNT (1) measured with an AC eld of 3.5 Oe at different frequencies, showing slow relaxation of the magnetization of the encapsulated Mn12Ac

molecules. (c) Hysteresis curves of magnetization recorded for Mn12Ac@

GMWNT (red) at 1.8 K and for the control sample Mn12Ac (2) (black). (d) Derivatives of the hysteresis curves shown in (c), including the energy level scheme for Mn12Ac plotted as a function of eld (thin crossing lines), calculated using parameters reported previously11.

H ( = 0)

H ( = 90)

I V +V +I

Figure 5 | Magnetoresistance measurements of nanotubes hosting Mn12Ac SMM. (a) Magnetoresistance of Mn12Ac@GMWNT (1) andthe control sample of empty (open and short) GMWNT (3) at 2 K with magnetic eld (H) either perpendicular ( = 0) or parallel ( = 90) to the conducting layers. (b) Low eld region of the magnetoresistance measurements on the logarithmic scale. (c) Schematic representation of the device used for magnetoresistance measurements. Inset: schematic representation of the preferential alignment of Mn12Ac molecules in the internal cavity of nanotube with respect to the magnetic eld.

inside nanotubes. In all previous studies, SMM mixed with SWNT were adsorbed on the external nanotube surfaces1921. Although surface deposition of Mn12Ac is known to cause decomposition of the delicate magnetic molecules13, our procedure for the transport and encapsulation of SMM into GMWNT has no detrimental eects on the structural integrity and magnetism of the Mn12Ac.

Both and display two distinct maxima (labelled with arrows, Fig. 4a,b), attributable to the presence of two slightly different Mn12Ac species, with dierent values of zero-eld splitting parameter, D. The lower-temperature maximum corresponds to a fast-relaxing species in which JahnTeller elongation axis of one of the Mn3 + centres has been tilted by 90 compared with the normal Mn12Ac molecule (slow-relaxing species)27. For Mn12Ac@GMWNT (1), the fraction of fast-relaxing species, which is determined from the intensity of the low-temperature peak, is ~20% compared with 7% for the control sample Mn12Ac (2) (Supplementary Fig. S16 and

Supplementary Table S4). This type of Jahn-Teller isomerization has also been observed previously for Mn12Ac aggregated in small clusters embedded in polymer matrices28,29.

The SMM in nanotubes exhibit magnetic hysteresis at 1.8 K with sizeable coercivity of 0.2 T (Fig. 4c), again demonstrating that their properties remain largely unaected by nanoscale connement. Step-like features in the hysteresis curves, which become more evident in the plots of numerical derivatives dM/dH (Fig. 4d), are attributed to quantum tunnelling of the magnetization. The tunnelling occurs for each longitudinal eld at levels (MS states) on both sides of the magnetization relaxation energy barrier that are in resonance2,3. These steps (dotted lines, Fig. 4d) occur at the same elds for both Mn12Ac@GMWNT (1) and the control sample Mn12Ac (2), showing both samples to have the same D values. The observed variation of the step positions with the strength of magnetic eld correlates well with the crossing energy levels (thin lines, Fig. 4d) calculated for Mn12Ac (ref. 11). These observations are consistent with a signicant fraction of the molecules becoming oriented with their easy axes aligned parallel to the applied magnetic eld. Without such orientation the longitudinal component of the applied eld

for each molecule would be dierent resulting in a smearing-out of the magnetization steps.

The derivative peaks of the magnetization steps near zero eld (labelled with a black arrow, Fig. 4d) are due to relaxation of the fast-relaxing species, the magnetization of which is not blocked at this temperature. The relative step size found for Mn12Ac@GMWNT (1)

is higher than for Mn12Ac (2), indicating more efficient quantum tunnelling and, hence, larger transverse anisotropy of the conned magnetic molecules. The lower coercitivity value in the hysteresis curve for Mn12Ac@GMWNT (1) (Fig. 4c) with respect to the control sample Mn12Ac (2) also conrms the shorter spin relaxation time for the conned Mn12Ac inside GMWNT. This observation is also in agreement with the calculated energy barrier for Mn12Ac@

GMWNT (1) (E/kB = 57 1 K) that appears to be lower than for Mn12Ac (2) (E/kB = 72 4 K).

Magnetoresistance of Mn12Ac@GMWNT. Understanding the response of electric properties of Mn12Ac@GMWNT to applied magnetic eld is crucial for future applications of this material in nano-electronic devices. The inuence of a magnetic eld applied parallel ( = 90) or perpendicular ( = 0) to a layer of Mn12Ac@

GMWNT (1) on the resistance of the nanotubes lled with the magnetic molecules has been evaluated, and compared with a control sample of empty (short and open) GMWNT (3), the same as used for the preparation of Mn12Ac@GMWNT (1). Both samples

Mn12Ac@GMWNT (1) and GMWNT (3) exhibited high room-temperature-specic conductivities of ca. 20 S cm 1, which decrease with temperature (Supplementary Figs S20 and S21) due to inter-tube contact resistance30,31. This indicates that the nanotubes lled with SMM molecules retain their intrinsic electric properties.

The eect of the magnetic eld on conductance of Mn12Ac@

GMWNT became measurable only at temperatures below the blocking temperature of the single-molecule magnet Mn12Ac (ca. 5 K,

Supplementary Fig. S22) residing in nanotubes. For example, at 2 K and the eld above 1 T, all samples exhibited a simple positive magnetoresistance, regardless of the direction of the applied magnetic eld (Fig. 5a). However, at lower elds (below 1 T), the magnetoresistance of empty nanotubes and nanotubes lled with SMM

NATURE COMMUNICATIONS | 2:407 | DOI: 10.1038/ncomms1415 | www.nature.com/naturecommunications

2011 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms1415

ARTICLE

are qualitatively dierent (Fig. 5b). The sample Mn12Ac@GMWNT (1) exhibits a signicant negative magnetoresistance in this range, which is in a sharp contrast to the control sample GWMNT (3) (Fig. 5b; Supplementary Fig. S23). Another important eect of the magnetic guest molecules is that they change the response of the host nanotubes to the direction of the applied eld. For example, magnetoresistance of Mn12Ac@GMWNT (1) becomes more negative and its minimum shis substantially as the eld changes from perpendicular ( = 0) to parallel ( = 90), whereas for empty nanotubes (3) both the depth of the minimum and its position remain unchanged.

The eects of the SMM on nanotubes can be also observed at higher magnetic eld ( > 1 T, Fig. 5a). The gradient of the positive magnetoresistance of Mn12@GMWNT (1) changes by 50%

as the directions of the eld changes, whereas the control sample of GMWNT (3) is signicantly less sensitive showing only 10% change. Encapsulated SMM modify the susceptibility of the resistance of nanotubes to the angle of the magnetic eld, which relates to a preferential orientation of the Mn12Ac molecules in the internal cavities of the nanotubes (Fig. 5c). As the Mn12Ac have a shape of a disc, with the easy magnetic axis perpendicular to the disc, this magnetic anisotropy of the guest molecules translates into the anisotropic magnetoresistance behaviour of the host nanotubes.

Discussion

This study describes a methodology for integration of SMM into the internal cavity of carbon nanotubes. The non-covalent interactions, which are responsible for efficient transport and encapsulation of the guest molecules into nanotubes, allow the molecules to stay mobile within the nanotube and align along the applied magnetic eld. Although the encapsulated Mn12Ac molecules retain their

SMM properties, the host nanotubes provide an alternative pathway for relaxation of SMM magnetization. In turn, the magnetic guest molecules inuence the electrical resistance of the host nanotubes. Therefore, the new class of hybrid structures SMM@GMWNT represents a synergistic material combining the functional properties of nanotubes and magnetic molecules.

Encapsulation of SMM into carbon nanotubes oers a number of exciting opportunities for the applications of these materials in nano-electronics. For example, electrical conductivity of the nanotube that could be precisely controlled by the magnetic states of SMMs can provide a method for generating spin-polarized currents at the nanoscale, which are important for the emerging area of spintronic devices. Insertion of SMMs into carbon nanotubes may enable integration of these molecular materials into ultrasensitive magnetic devices, such as a nanoscale superconducting quantum interference device (nano-SQUID) or a magnetic force microscopy probe, which will open a pathway for harnessing functional properties of SMMs at a single-molecule levela long-standing challenge in molecular nanomagnetism32.

Methods

Preparation of materials. Mn12Ac was synthesized following the procedure described previously33. All reagents and solvents were purchased from Sigma-Aldrich and used without further purication. GMWNT (catalysed chemical vapour deposition) were purchased from CheapTubes. GMWNT were pre-treated to shorten and open them, and to enable the encapsulation of Mn12Ac molecules.

Preparation of short and open GMWNT via oxidative cutting. GMWNT(200 mg) were added to a 16 M solution of nitric acid (50 ml) and the resulting black suspension was treated with ultrasound in a water bath for 30 min at room temperature. The suspension was then heated in air under reux for 5 h. The resulting mixture was then diluted with deionized water (200 ml), ltered using a 0.45 m pore size PTFE membrane lter, washed thoroughly with water until neutral pH and ethanol (50 ml) and dried under vacuum to yield a black solid (192 mg). The product was then placed in an alumina crucible and heated in air at 700 C until weight loss of 80% was achieved.

Encapsulation of Mn12Ac in GMWNT. Freshly annealed GMWNT (10 mg, short and open) were added to a suspension of Mn12Ac (30 mg) in hexane (8 ml), and the

resulting black suspension was transferred into a 10 ml autoclave that was topped up with supercritical carbon dioxide (40 C, 275 bar). The pressure was cycled between 270 bar and 120 bar for 20 h. The resulting black solid was then washed with 250 ml acetonitrile and ltered using 0.2 m pore size PTFE membrane lter to give Mn12Ac@GMWNT (1).

Control measurements. A series of rigorous control experiments involving untreated (sealed) GMWNT were performed to ascertain the efficacy of the washing procedure, and therefore, less eective interactions of Mn12Ac molecules with the outer surface of the nanotubes as compared with the interior of GMWNT (Supplementary Figs S7 and S10).

To conrm that Mn12Ac molecules are transported and encapsulated in the GMWNT in their intact form, a control sample of Mn12Ac (2) treated with scCO2 under the same conditions as used for preparation of 1 was carefully analysed by dierent analytical techniques all supporting the fact that structure and properties of Mn12Ac

remain unchanged (Supplementary Figs S11S17 and Supplementary Tables S1S3).

To prove that the Mn12Ac remains intact inside NT, the thermal propertiesof Mn12Ac@GMWNT (1) were characterized by DSC and TGA. Additionally, a Mn12Ac@GMWNT sample heated at 340 C was analysed by TEM (Supplementary Figs S24 and S25), revealing that Mn3O4 is formed as a product of thermal decomposition of Mn12Ac in GMWNT.

To verify that Mn-peaks in the EDX are indeed due to Mn12Ac guest molecules, EDX spectra were taken for empty sections of nanotubes in Mn12Ac@GMWNT and the supporting carbon lm, none of which showed any indication of Mn (Supplementary Fig. S26).

Analytical techniques. TGA measurements were carried out on a SDT Q-600 TA instrument over the range 251,000 C in air and at a scan rate of 5 C min 1.

DSC measurements were carried out on a Mettler Toledo DSC 821e over the range 25500 C with a scan rate of 5 C min 1 in air or nitrogen. X-ray powder diraction measurements were performed on a PANalytical XPert PRO diractometer equipped with a Cu K radiation source ( = 1.5418 ) in BraggBrentano geometry using a Si zero-background holder. Infrared spectra were measured as KBr discs on a Nicolet Avatar 380 FTIR spectrometer over the range 400 4,000 cm 1.

Elemental analyses were performed by the Elemental Analysis Service of London Metropolitan University. Mass spectrometry was carried out on a Bruker ApexIV with electro spray ionization.

Transmission electron microscopy characterization. High-resolution TEM imaging and energy-disperse X-ray analyses were performed using a Jeol 2100F transmission electron microscope using an accelerating voltage of 100 kV. TEM samples were prepared by casting several drops of hexane suspension of Mn12Ac@

GMWNT onto copper-grid mounted lacey carbon lms before drying under a stream of nitrogen. Statistical analysis of TEM images were performed using Gatan Digital Micrograph soware.

Magnetic measurements. A commercial Quantum Design MPMS-XL5 SQUID magnetometer was used for the magnetic characterization of the samples: Mn12Ac@

GMWNT (1), Mn12Ac (post-scCO2) (2) and pristine Mn12Ac. Pellet samples were pressed and wrapped into Teon lm with negligible diamagnetic contribution. Variable-temperature (1.8300 K) and eld-dependent (05 T) magnetization measurements were carried out using the reciprocating sample option with a sensitivity of 510 9 emu. AC susceptibility measurements were carried out at dierent frequencies (11,500 Hz) of a 3.5 Oe oscillating eld between 1.8 and 10 K.

Methodology used in the magnetic measurements. The investigation of the eects of the encapsulation into nanotube on the magnetic properties of Mn12Ac molecules was carried out by comparing the results obtained for 1 with those for 2 and untreated Mn12Ac taking into account the diamagnetic contribution of GMWNT.

Considerations in the magnetic analysis of 1. The analysis of the magnetic measurements of 1 requires the quantication of both the diamagnetic contribution of the GMWNT and the molar concentration of Mn12Ac in the sample. To disentangle the eects of the GMWNT on the magnetic properties of the Mn12 Ac molecules, we rst measured the magnetic response of the empty GMWNT. The magnetization loop (Supplementary Fig. S27) shows that GMWNT do not exhibit hysteresis. Moreover, GMWNT have diamagnetic response, because the measured magnetization is negative for positive elds and positive for negative elds. The linear response of M versus 0H allows the determination of the susceptibility = ( 10.0 0.1)10 8 emu cm 3

that is simply given by the slope of the curve.

The susceptibility measurements for GMWNT as function of the temperature are presented in Supplementary Figure S28. The signal from the GMWNT is always negative, that is, diamagnetic (of the order of 6.3810 6 emu g 1 at 300 K) and slightly increases below 42 K (inset Supplementary Fig. S28). The mass susceptibility of Mn12Ac is 2.8310 5 emu g 1 at 300 K that is 77% higher than that of the

GMWNT. Below 100 K the dierence in mass susceptibility increases further with the Mn12Ac reaching 9.0310 3 emu g 1 at 1.8 K that is three orders of magnitude higher than that of the GMWNT at the same temperature.

The quantication of the molar concentration has been carefully assessed from the magnetic measurements reported in Supplementary Figure S27. MGMWNT, MMn12Ac

NATURE COMMUNICATIONS | 2:407 | DOI: 10.1038/ncomms1415 | www.nature.com/naturecommunications

2011 Macmillan Publishers Limited. All rights reserved.

ARTICLE

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms1415

and MMn12Ac@GMWNT were dened as the magnetic moments of the GMWNT, Mn12Ac and Mn12Ac@GMWNT samples, respectively. If mGMWNT , mMn12Ac and mMn12Ac@GMWNT are their respective masses, then the magnetic moments per mass of the

GMWNT and Mn12Ac, MGMWNT

* and MMn Ac

12

* , respectively, can be calculated.

As mMn12@GMWNT = mGMWNT + mMn12Ac, the following relation holds

M m M m M

Mn Ac GMWNT GMWNT GMWNT Mn Ac Mn Ac

12 12 12

@ * *

= +

Then the mass of Mn12Ac and GMWNT can be calculated as

m

M m M

M M

Mn Ac

Mn12Ac GMWNT Mn Ac GMWNT GMWNT Mn Ac GMWNT

12

@ @ *

* * ;; @

m m m

GMWNT Mn Ac GMWNT Mn Ac

=

12 12

From the magnetic moment measured at 1.8 K and 2 T (Supplementary Fig. S27), it can be deduced that MGMWNT

* = 0.17 emug1, MMn Ac

12

=

12 12

* = 37.47 0.06 emug1 and MMn12Ac@GMWNT = 0.172 0.005 emu. Because mMn12Ac@GMWNT = 33.8 0.1 mg, then,

mGMWNT = 29.0 and mMn12Ac = 4.8 0.7 mg.

Magnetoresistance measurements. DC conductivity measurements were performed with the four contacts method for two pressed pellets of each sample: Mn12Ac@GMWNT (1) and GMWNT (3). Both pellets of each sample gave very similar results demonstrating reproducibility of the measurements. The magnetoresistance measurements were performed in the temperature range 2300 K with magnetic elds in the range 8 to 8 T applied either parallel or perpendicular to the conduction planes. Platinum electrodes (25 m diameter) were connected to the samples using graphite paste. The samples were measured with cooling and warming rates of 1 K min 1 and with DC intensity currents of 0.11.000 A in a Quantum Design PPMS-9, giving reproducible results in all cases. Further details of the magnetoresistance measurements can be found in the Supplementary

Figs S20S23.

Quantication of the amount of SMM inside nanotubes. The amount of Mn12Ac

in Mn12Ac@GMWNT (1) can be estimated from TGA (11.6% wt). From the TGA measurements, the amount of Mn12Ac has been deducted by comparing the amount of manganese oxide remaining at 1,000 C aer all carbon of Mn12Ac@

GMWNT being burnt out (5.6% of the total weight, Supplementary Fig. S25) with the residual manganese oxide found at the same temperature for pristine Mn12Ac

(47.2% of the total weight, Supplementary Fig. S11).

Magnetic measurements described above gave 14.2% wt for Mn12Ac in

Mn12Ac@GMWNT that is slightly higher than TGA. The accuracy of quantitative SQUID measurements is expected to be greater than that of TGA.

References

1. Sessoli, R., Gatteschi, D., Caneschi, A. & Novak, M. A. Magnetic bistability in a metal-ion cluster. Nature 365, 141143 (1993).

2. Friedman, J. R., Sarachik, M. P., Tejada, J. & Ziolo, R. Macroscopic measurements of resonant magnetization tunneling in high spin molecules. Phys. Rev. Lett. 76, 38303833 (1996).

3. Thomas, L. et al. Macroscopic quantum tunnelling of magnetization in a single crystal of nanomagnets. Nature 383, 145147 (1996).

4. Ardavan, A. et al. Will spin-relaxation times in molecular magnets permit quantum information processing. Phys. Rev. Lett. 98, 057201 (2007).

5. Schlegel, C., van Slageren, J., Manoli, M., Brechin, E. K. & Dressel, M. Direct observation of quantum coherence in single-molecule magnets. Phys. Rev. Lett. 101, 147203 (2008).

6. Weiden, N., Kass, H. & Dinse, K. P. Pulse electron paramagnetic resonande (EPR) and electron-nuclear double resonance (ENDOR) investigation of N@C-60 in polycrystalline C-60. J. Phys. Chem. B 103, 98269830 (1999).

7. Dietel, E. et al. Atomic nitrogen encapsulated in fullerenes: eects of cage variations. J. Am. Chem. Soc. 121, 24322437 (1999).

8. Morton, J. J. L. et al. Electron spin relaxation of N@C60 in CS2. J. Chem. Phys. 124, 014508 (2006).

9. Rocha, A. R. et al. Towards molecular spintronics. Nature Mater. 4, 335339 (2005).

10. Joachim, C., Gimzewski, J. K. & Aviram, A. Electronics using hybrid-molecular and mono-molecular. Nature 408, 541548 (2000).

11. Gatteschi, D. & Sessoli, R. Quantum tunneling of magnetization and related phenomena in molecular materials. Angew. Chem. Int. Ed. 42, 268297 (2003).

12. Leuenberger, M. N. & Loss, D. Quantum computing in molecular magnets. Nature 410, 789793 (2001).

13. Voss, S. et al. Electronic structure of Mn12 derivates on the clean and functionalized Au surface. Phys. Rev. B 75, 045102 (2007).

14. Saywell, A. et al. Self-assembled aggregates formed by single-molecule magnets on a gold surface. Nat. Commun. 1, 75 (2010).

15. Bogani, L. & Wernsdorfer, W. Molecular spintronics using single-molecule magnets. Nature Mater. 7, 179186 (2008).

16. Trif, M., Troiani, F., Stepanenko, D. & Loss, D. Spin-electric coupling in molecular magnets. Phys. Rev. Lett. 101, 217201 (2008).

17. Chamberlain, T. W., Gimnez-Lpez, M. C. & Khlobystov, A. N. Carbon nanotubes as containers: in Carbon Nanotubes and Related Structures: Synthesis, Characterization, Functionalization and Applications (eds Guldi, D.M. & Martn, N.) 349384 (Wiley-VCH, 2010) .

18. Benjamin, S. C. et al. Towards a fullerene-based quantum computer. J. Phys. Condens. Matter 18, S867S883 (2006).19. Bogani, L. et al. Single molecule magnet carbon nanotube hybrids. Angew. Chem. Int. Ed. 48, 746750 (2009).

20. Kyatskaya, S. et al. Anchoring of rare-earth-based single-molecule magnets on single-walled carbon nanotubes. J. Am. Chem. Soc. 131, 1514315151 (2009).

21. Giusti, A. et al. Magnetic bistability of individual single-molecule magnets graed on single-wall carbon nanotubes. Angew. Chem. Int. Ed. 48, 49494952 (2009).

22. Ciric, L. et al. La@C82 as a spin-active lling of SWCNTs: ESR study of magnetic and photophysical properties. Phys. Status Solidi B 245, 20422046 (2008).

23. Khlobystov, A. N. et al. Low temperature assembly of fullerene arrays in single-walled carbon nanotubes using supercritical uids. J. Mater. Chem. 14, 28522857 (2004).

24. Ye, X. R. et al. Supercritical uid synthesis and characterization of catalytic metal nanoparticles on carbon nanotubes. J. Mater. Chem. 14, 908913 (2004).

25. Pan, X. & Bao, X. Reactions over catalysts conned in carbon nanotubes. Chem. Commun. 62716281 (2008).

26. Folch, B., Larionova, J., Guari, Y., Guerin, C., Mehdi, A. & Reye, C. Formation of Mn3O4 nanoparticles from the cluster [Mn12O12(C2H5COO)16(H2O)3]

anchored to hybrid mesoporous silica. J. Mater. Chem. 14, 27032711 (2004).27. Aubin, S. M. et al. Single-molecule magnets: Jahn-Teller isomerism and the origin of two magnetization relaxation processes in Mn12 complexes.

Polyhedron 20, 11391145 (2001).28. Domingo, N. et al. Particle-size dependence of magnetization relaxation in Mn12 crystals. Phys. Rev. B 79, 214404 (2009).

29. van Slageren, J., Dengler, S., Gmez-Segura, J., Ruiz-Molina, D. & Dressel, M. Magnetism and magnetic resonance studies of single-molecule magnets in polymer matrices. Inorg. Chim. Acta 361, 37143717 (2008).

30. de Heer, W. A. et al. Aligned carbon nanotube lms: production and optical and electronic properties. Science 268, 845847 (1995).

31. Kim, G. T. et al. Magnetoresistance of an entangled single-wall carbon-nanotube network. Phys. Rev. B 58, 1606416069 (1998).

32. Cleuziou, J. P., Wernsdorfer, W., Bouchiat, V., Ondarcuhu, T. & Monthioux, M. Carbon nanotube superconducting quantum interference device. Nature

Nanotech. 1, 5359 (2006).33. Lis, T. Preparation, structure, and magnetic-properties of a dodecanuclear mixed-valence manganese carboxylate. Acta Crystallogr. Sect. B 36, 20422046 (1980).

Acknowledgements

This work was supported by the Marie Curie Fellowship within the Seventh European Community Framework Programme (M.C.G. and F.M.), UK Engineering and Physical Sciences Research Council, European Science Foundation (A.N.K., J.v.S., A.L.T.and P.D.B), the Royal Society (A.N.K.), Spanish Ministerio de Ciencia y Tecnologia, Generalitat Valenciana (C.J.G.G.) and the Nottingham Nanoscience and Nanotechnology Centre (access to transmission electron microscope). We would like to thank Jose Maria Martinez-Agudo (ICMOL, University of Valencia) for the DSC measurements.

Author contributions

M.C.G. prepared the materials, developed the experimental methodology, performed the analytical characterization, analysed and interpreted the data. A.L.T. and M.C.G. carried out the scCO2 experiments. A.N.K. conducted the transmission electron microscopy experiments. F.M. and J.v.S. carried out and analysed the magnetic measurements data. C.J.G.G. carried out the magnetoresistance experiments and analysed the data. P.D.B. assisted with the manuscript preparation and analysis of TEM data. M.C.G. and A.N.K. conceived the idea, coordinated the experimental work, discussed the data and wrote the original manuscript. All the authors read and commented on the manuscript.

Additional information

Supplementary Information accompanies this paper at http://www.nature.com/ naturecommunications

Competing nancial interests: The authors declare no competing nancial interests.

Reprints and permission information is available online at http://npg.nature.com/ reprintsandpermissions/

How to cite this article: Gimnez-Lpez, M.C. et al. Encapsulation of single-molecule magnets in carbon nanotubes. Nat. Commun. 2:407 doi: 10.1038/ncomms1415 (2011).

NATURE COMMUNICATIONS | 2:407 | DOI: 10.1038/ncomms1415 | www.nature.com/naturecommunications

2011 Macmillan Publishers Limited. All rights reserved.