www.advmat.dewww.MaterialsViews.com

CO

MM

UN

ICATI

Johannes M. Haberl , Antoni Sánchez-Ferrer , Adriana M. Mihut , Hervé Dietsch , Ann M. Hirt , and Raffaele Mezzenga *

Liquid-Crystalline Elastomer-Nanoparticle Hybrids with Reversible Switch of Magnetic Memory

ON

Magnetic shape-memory materials are responsive materials with a strong infl uence of shape changes on the magnetic prop-erties, and vice versa. [ 1 ] They offer access to unique applications, including wireless actuators, switches and magnetic valves. To achieve the required mechanical-magnetic coupling, both inor-ganic materials and polymer-based hybrid systems have been proposed. [ 1–7 ] In inorganic materials, the magneto-crystalline anisotropy and the martensitic transformation are typically exploited to induce magnetic reorganization at the phase transi-tion, but the associated elongation and device sizes, are limited to changes of about 10% and to the millimeter range, respec-tively, [ 2 ] which severely limits their use in the mentioned appli-cations. Polymer-based materials, on the other hand, are more effi ciently sizeable and recover easily strains beyond 50% by magnetic stimuli. [ 3,4 ] However, to achieve magnetic properties of serviceable strength, they require the incorporation of inor-ganic magnetic particles. [ 3–8 ]

To solve this issue and to induce magnetic anisotropy in easy-to-process systems, we propose a new strategy exploiting shape anisotropy of ferrimagnetic nanoparticles in organic-inorganic hybrid materials. The effect of shape-anisotropy is found in soft magnets, such as maghemite, where magneto-crystalline anisotropy is of minor importance and it is explained in terms of the Stoner–Wohlfarth theory. [ 9,10 ] In order to achieve full control of the resultant magnetic anisotropy in response to external stimuli, we use a new liquid-crystalline elastomer (LCE) matrix that allows fully reversible changes in shape and physical properties at the smectic-isotropic phase transition, [ 11 ] enables shape-memory [ 12 ] and can convert external stimuli into mechanical work. [ 3 , 5 , 13–17 ]

© 2013 WILEY-VCH Verlag Gm

DOI: 10.1002/adma.201204406

J. M. Haberl, Dr. A. Sánchez-Ferrer, Prof. R. MezzengaETH ZürichDepartment of Health Science and Technology8092 Zürich, Switzerland E-mail: raffaele.mezzenga@hest.ethz.ch Dr. A. M. Mihut, [+] Dr. H. Dietsch [++]

Adolphe Merkle Institute and Fribourg Center for NanomaterialsUniversity of Fribourg1723 Marly, Switzerland Prof. A. M. HirtETH Zürich, Department of Earth Science8092 Zürich, Switzerland [+] Present address: Physical Chemistry, Lund University, SE-22210 Lund, Sweden [++] Present address: BASF SE, Formulation Platform, 67056 Ludwigshafen am Rhein, Germany

Adv. Mater. 2013, 25, 1787–1791

While in LCE alone output signals beyond shape changes are limited to optical and electrical effects [ 18–21 ] and do not include useable magnetic features, [ 22 ] the integration of ellip-soidal magnetic nanoparticles (MNs) into LCEs, as proposed in this work, opens new scenarios. The LCE composites allow for accurate control over the global orientation of the MNs and the macroscopic magnetic susceptibility. The resulting material is not only a ductile and inexpensive alternative to magnetic shape memory alloys, but it also greatly expands the scope of materials with magnetic memory and actuation capabilities, bypassing classical limits found in these systems. We show that by stretching the hybrid systems at deformation beyond 200%, with stresses as low as 5 MPa, these new nanocomposites can be used to reversibly store magnetic information that can then be erased at temperatures as low as 80 ° C, which are typical conditions of organic materials. We conclude by demonstrating how the anisotropic information is readable in magnetic torque measurements and we point to implications that these mate-rials may have in temperature or strain sensors with magnetic read-out, actuators, magnetic switches and valves.

One major issue in the preparation of hybrid nanocomposites is the suppression of segregation of the nanoparticles from the polymer matrix. Creeping of the polymer matrix and aggregation of the nanoparticles at high temperatures can be avoided when crosslinkable polymers and particles are used; a concept that has already been applied for isotropic networks. [ 23,24 ] Therefore we have selected chemically compatible end-functional groups for the two components: hydroxyl groups for the liquid-crystalline polymer and amino groups for the inorganic nanoparticles, which can be mutually crosslinked into a covalent network with a tri-functional isocyanate (see Supporting Information).

For the purpose of generating a strong anisotropic magnetic susceptibility, silica-coated hematite MNs have been synthesized with a narrow and monomodal size distribution [ 25 ] and were then redox-processed into silica-coated maghemite. [ 26 ] In this study we use new core-shell MNs with a 7:3 maghemite:hematite ratio in the core and an amine-functionalized silica shell (see Supporting Information).

A representative hybrid fi lm with 10 wt% nanoparticles is shown in Figure 1 a. Due to the presence of iron-oxide-based MNs the sample appears dark brown. The homogeneous distribution and fi ne dispersion of MNs over the fi lm can be seen in the transmission electron microscopy (TEM) micrograph (Figure 1 b; Supporting Information, Figure S1). From statistical analysis on high-resolution TEM images (Figure 1 c; Supporting Informa-tion, Figure S2,S3) of 100 MNs, the mean core-shell structure is calculated, with a major core axis found to be 310 ± 7 nm and a minor core axis 55 ± 7 nm, which correspond to an aspect ratio of

bH & Co. KGaA, Weinheim 1787wileyonlinelibrary.com

1788

www.advmat.dewww.MaterialsViews.com

CO

MM

UN

ICATI

ON

Figure 1 . a) Photograph of the liquid-crystalline elastomer-magnetic nanocomposite fi lm. b) TEM micrograph of a stretched sample (80 nm-section microtomed parallel to stretching direction) and c) TEM micrograph of core-shell ellipsoidal nanoparticles. d) Characteristic fea-tures for: i) the calamitic repeating unit in the smectic layering of the liquid-crystalline elas-tomer, ii) the molecular distance in the layers with the crosslinker molecules (green spheres) and iii) the magnetic nanoparticles.

5.6 and the demagnetizing factors N parallel = 0.048 and N perpendicular = 0.476. [ 27 ] The average thicknesses of the silica shell are 22 ± 4 nm along the major axis and 19 ± 2 nm along the minor axis. In Figure 1 d a schematic of the microstructure of the hybrid nanocomposite and the length-scales involved are presented.

Differential scanning calorimetry experiments have been performed in order to analyze the phase behavior of the nano-composites. A glass-transition temperature of T g = 15 ± 5 ° C and a smectic-isotropic transition at T SI = 75 ± 2 ° C have been found, respectively.

We then have analyzed the structural and physical properties of the nanocomposite upon stretching, representatively shown for two states. The fi rst state, referred to as “ON”, is when the fi lm has been stretched at high temperature to an elastic deformation of λ = 3.2, which is fi xed on cooling down, due to the smectic liquid-crystalline phase. The second state, referred

wileyonlinelibrary.com © 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Wei

to as “OFF”, is when the sample has relaxed back to its original length ( λ = 1.0) due to entropy elasticity, after heating above the clearing temperature T SI . Compared to glassy systems with kinetically driven processes, liquid-crystalline elastomers show all the advantages associated with fi rst order tran-sitions, and thus immediate changes in the mesophase structure at the phase transition together with spontaneous changes in shape.

Small and wide-angle X-ray scattering (SWAXS) experiments for both states ON and OFF have been performed to study the microstructure of the LCE and the MNs ori-entation distribution ( Figure 2 ). 2D-scattering patterns have been taken from the perpen-dicular directions normal to the fi lm surface ( y -direction in Figure 2 a,c) and normal to the fi lm thickness (x-direction Figure 2 b,d). From the evaluation of the WAXS patterns a smectic layering distance of d l = 1.9 nm with a correlation length of ξ = 24 nm and a mesogen-mesogen distance of d m = 0.44 nm have been measured. From the azimuthal intensity distribution the orientation distri-bution of scattering elements and the order parameter:

S = 1/2 · ⟨3 cos2 θ − 1

⟩ (1)

have been calculated. [ 28,29 ] This defi nition of the order parameter relies on cylindrical sym-metry. In cases where cylindrical symmetry is lost with respect to any axis, we still use the defi nition, but identify the order parameter with an additional asterisk, S ∗ .

The 2D-WAXS pattern for the OFF state shows a LC polydomain structure without scattering intensity anisotropy (Figure 2 a and blue and green curves in Figure 2 b). How-ever, the SAXS pattern in the y -direction (Figure 2 b, red curve) shows two maxima in the azimuthal intensity distribution with

the order parameter S = − 0.28 (calculated around a director in y -direction). As the scattering of the reference polymer systems presents no features in this q -region (see Supporting Informa-tion, Figure S7), these SAXS profi les can be univocally attrib-uted to a planar orientation distribution of the particles parallel to the fi lm surface due to in-plane confi nement during the synthesis of the fi lm.

In Figure 2 c and Figure 2 d the SWAXS patterns and azimuthal scans for the ON states are presented. In the WAXS pattern in y -direction (Figure 2 c) anisotropic scattering is observed with two maxima in the mesogen distance. These peaks in the azi-muthal intensity distribution (blue curve) indicate preferential orientation of the mesogens long axes parallel to the stretching direction with an order parameter S = 0.70 (calculated around a director in z -direction). Two additional maxima at the lay-ering distance appear perpendicular to the former and indicate

nheim Adv. Mater. 2013, 25, 1787–1791

www.advmat.dewww.MaterialsViews.com

CO

MM

UN

ICATIO

N

Figure 2 . Two-dimensional X-ray scattering patterns in the wide angle (light blue) and small-angle (dark blue) regions for the OFF state after relaxation (a, y -direction; b, x -direction) and ON state after stretching (c, y -direction; d, x -direction), with the azimuthal scans at the corresponding correlation lengths of the magnetic nanoparticles short axes (red curve at q = 0.15−0.25 nm − 1 ), the smectic layering (green curve at q = 3.3 nm − 1 ) and the mesogens (blue curve at q = 14 nm − 1 ). Note: the red arrow indicates the X-ray beam direction.

a smectic A phase (green curve). In the SAXS pattern the azi-muthal intensity distribution at the characteristic nano particle length scale (red curve) shows two intensity maxima that indicate an orientation of the mean particle long axes parallel to the stretching direction with an order parameter S ∗ = 0.56 (calculated around a director in z -direction). This is in agree-ment with an earlier, similar study. [ 23 ]

The 2D-WAXS pattern in x -direction is shown in Figure 2 d. From the azimuthal intensity distribution at the mesogen dis-tance, a degree of order S = 0.68 is calculated and the two maxima for the smectic A layering are found consistently with the evalu-ation in Figure 2 c. From the 2D-SAXS pattern, the azimuthal intensity distribution at the MNs characteristic distance (red curve) is evaluated with a corresponding anisotropy quantifi ed by S ∗ = 0.69. This increase in value, compared to Figure 2 c, is a direct consequence of the in-plane confi nement of the MNs.

From the SWAXS analysis above, we can summarize the restructuring upon stretching for both the liquid-crystalline elastomer and the nanoparticles, which both align the mesogens and the nanoparticles long axes parallel to the strain direction (ON state), while an in-plane isotropic orientation distribution

© 2013 WILEY-VCH Verlag GAdv. Mater. 2013, 25, 1787–1791

of both liquid crystal domains and nanoparticles is found when the stored deformation is released ( λ = 1, OFF state) ( Figure 3 ).

Low-fi eld magnetic susceptibility measurements of the nano-composite have been performed in order to determine the anisotropic magnetic susceptibility of the sample in the OFF and ON states. In order to remove potential effects from fi eld inhomogeneity the ON sample (with λ = 3.2) was folded twice, to yield dimensions directly comparable with the OFF sample and allowing measurements under identical geometrical set-up. Both the OFF and ON samples were rotated stepwise around the three axes x , y and z , and the 2nd-rank susceptibility tensors have been calculated. The normalized eigenvalues K ij (OFF) = {1.04, 1.04, 0.91} and K ’ ij (ON) = {1.19, 0.95, 0.86} are obtained with residuals of 0.5%. For the OFF state, K 1 and K 2 were iden-tical as expected from the isotropic 2D-SAXS pattern (Figure 2 a). K 3 was about 14% lower, which is the result of the in-plane con-fi nement of the nanoparticles (Figure 2 b). For the ON state, K’ 1 is 25% higher than K’ 2 as a consequence of the reorientation process of the MNs along the stretching direction, which leads to a deformation of the bulk susceptibility tensor with expan-sion in z -direction.

1789wileyonlinelibrary.commbH & Co. KGaA, Weinheim

1790

www.advmat.dewww.MaterialsViews.com

wileyonlinelibrary.com © 2013 WILEY-VCH Verlag GmbH & Co. KGaA, We

CO

MM

UN

ICATI

ON

Figure 3 . Control over low-fi eld magnetic susceptibility with the two states, OFF and ON. On the left, the random distribution of magnetic nanoparticles in the liquid-crystalline elastomer nanocomposite in the OFF state after relaxation, characterized by an order parameter S ∗ = 0 and the corresponding low-fi eld anisotropic magnetic susceptibility K 1 / K 2 = 1. The colored arrows indicate triggering (green) and release (red) of the anisotropic magnetic susceptibility. On the right, the liquid-crystalline elastomer nanocomposite and the oriented magnetic nano-particles in the ON state after stretching, characterized by an order parameter S ∗ = 0.56 and the corresponding low-fi eld anisotropic magnetic susceptibility K ’ 1 / K ’ 2 = 1.25.

Figure 4 . a) The magnetic torque experiment set-up [ 30 ] and data of the nanocomposite at the magnetic fi eld B = 1.4 T in x , y and z direction for the ON (green dots) and the OFF (black dots) states, together with the corresponding theoretical curves (solid lines). b) The stereogram of the ferromagnetic tensor for the OFF and ON states with the eigenvalues difference of the major and the semi-major axes.

In Figure 3 , the xz -plane of the low-fi eld magnetic susceptibility measurements for the OFF and ON states are represented with red ellipses. In order to compare the measure-ments of the susceptibilities with theoretical expectations, we apply a Stoner–Wohlfarth model, [ 10 ] which supports the presence of an anisotropic magnetic susceptibility K ’ 1 / K ’ 2 ≈ 2.1 (see Supporting Information), consistently with the measurements. The deviation of the theoretical anisotropic magnetic susceptibility compared to the measured one ( K ’ 1 / K ’ 2 = 1.25) is expected and, as discussed below, in agreement with the results from high-fi eld magnetic measurements and attributed to the effects that arise from the 30 wt-% hema-tite content in the MNs. The model-assump-tion of a particle magnetization obeying the global energy minima (Supporting Informa-tion, Figure S4), multigrain structures and surface effects in the particles, and inter-particle magnetic dipole coupling may further contribute to this overestimation.

To get further details on the magnetic structure we performed complementary high-fi eld magnetic torque measurements ( Figure 4 ). During these experiments the magnetization of the MNs is saturated and aligned with respect to the external fi eld. Any tilt angle θ between the long axis of a MN and the magnetic fi eld generates a torque that is measured and predicted (solid line) for the entire nanocomposite following:

τ = μ0/2 · V · ∣∣Nparallel − Nperpendicular

∣∣ ·M2

s · S∗ · sin(2�) (2)

with the volume of iron-oxide V = 3.95 × 10 − 11 m 3 , the magnetic constant μ 0 = 4 π × 10 − 7 NA − 2 , the saturation Magnetization M s = 120 kA m − 1 , N parallel − N perpendicular and S ∗ defi ned as above, and Θ the angle between the sample and the magnetic fi eld (Figure 4 a). [ 31,32 ]

The torque in the x -direction for the OFF state shows a 2 Θ -signal periodicity which fi ts well with the theoretical curve, using S = − 0.28. Moreover, the signal strength in the ON state maintains the same 2 Θ -periodicity, but increases in magnitude, in agreement with the corresponding theoretical expectation for S ∗ = 0.69. The overestimation compared to the measurement is due to the value S ∗ , which overestimates the order after stretching due to the in-plane confi nement of the MNs in y -direction before stretching, an effect contributing to S ∗ but not to S , that is signifi cant in both “OFF” and “ON”.

For the measurement in y -direction a very weak signal is found for the OFF state. Since

inheim Adv. Mater. 2013, 25, 1787–1791

www.advmat.dewww.MaterialsViews.com

CO

MM

UN

ICATIO

N

theoretically for S = 0 no torque is expected, this suggests that there is a contribution from the 30 wt% of hematite in the MNs. Hematite has a strong magnetocystalline anisotropy that would be oriented in this direction. [ 25 ] The measurement for the ON state shows a strong periodical 2 Θ -signal, which corresponds perfectly to the theoretical calculation, using S ∗ = 0.56.

In z -direction the OFF state also has a 2 Θ -periodicity and the intensity agrees with the prediction for the measured S = − 0.28. For the ON state the torque decreases, because stretching has produced a decrease in the population of MNs peaked around the equatorial direction, the only component which contributes to torque in this case. The theoretical curve is computed with an order parameter S ∗ = −0.04 that has been obtained from the X-ray scattering experiment in the z -direction (Supporting Information, Figure S5).

The resulting tensors of the torque measurements are shown in Figure 4 b. The stereogram from OFF constitutes an oblate-like tensor around z with a small eigenvalue difference of K 1 − K 2 = 3.7 × 10 − 7 that is not oriented with respect to the pre-vious stretching direction, whereas the ON ensemble is repre-sented by a tensor with the eigenvalue difference K’ 1 − K’ 2 = 10.1 · 10 − 7 and K 1 lies in the stretching direction. Interestingly, the OFF state is not completely isotropic in the K 1 − K 2 plane, which is the origin of the non-zero torque that is observed in the y -direction. The minor axis corresponds to the in-plane orientation distribution that has also been confi rmed by SWAXS experiments. Thus the deformation of the material is correctly measurable with both low and high magnetic fi elds.

We have presented a new class of hybrid nanocomposites, where the unique combination of liquid-crystalline elastomers and anisotropic magnetic nanoparticles is exploited to generate a material with magnetic memory that can be reversibly stored by mechanical deformation and erased by heating to moderate temperatures. The resulting shape-memory materials allow manipulation of magnetic properties of inorganic materials via the control of parameters that are typical of soft materials, and thus greatly expand the scope of magnetic materials. As stored magnetic information is reachable by simple measurements, such as magnetic torque, and can be reversibly tuned, the route presented here increases possibilities in the design of actuators beyond presently achievable imprinted magnetic geometries, [ 33 ] and shows great promise in strain sensing devices and magnetic information storage applications.

Supporting Information Supporting Information is available from the Wiley Online Library or from the author.

Acknowledgements The research was fi nanced by the NRP 62 of the Swiss SNF. We gratefully acknowledge I. Usov for assistance with the calculations.

Received: October 23, 2012 Revised: December 11, 2012

Published online: January 29, 2013

© 2013 WILEY-VCH Verlag GAdv. Mater. 2013, 25, 1787–1791

[ 1 ] J. Ma , I. Karaman , Science 2010 , 327 , 1468 – 1469 . [ 2 ] A. Sozinov , A. A. Likhachev , N. Lanska , K. Ullako , Appl. Phys. Lett.

2002 , 80 , 1746 – 1748 . [ 3 ] A. Kaiser , M. Winkler , S. Krause , H. Finkelmann , A. M. Schmidt ,

J. Mater. Chem. 2009 , 19 , 538 – 543 . [ 4 ] R. Mohr , K. Kratz , T. Weigel , M. Licka-Gabor , M. Moneke ,

A. Lendlein , Proc. Natl. Acad. Sci. USA 2006 , 103 , 3540 – 3545 [ 5 ] I. Levine , R. B. Zvi , M. Winkler , A. M. Schmidt , M. Gottlieb ,

Macromol. Symp. 2010 , 291–292 , 278 – 286 . [ 6 ] A. M. Schmidt , Macromol. Rapid Commun. 2006 , 27 , 1168 –

1172 . [ 7 ] U. N. Kumar , K. Kratz , M. Behl , A. Lendlein , Polym. Exp. Lett. 2012 ,

No. 1 , 26 – 40 . [ 8 ] Y. Zhou , N. Sharma , P. Deshmukh , R. K. Lakhman , M. Jain ,

R. M. Kasi , J. Am. Chem. Soc. 2012 , 134 , 1630 – 1641 . [ 9 ] R. C. O’Handley , in Modern Magnetic Materials: Principles and

Applications , Wiley-Interscience , New York 1999 . [ 10 ] E. C. Stoner , E. P. Wohlfarth , Phil. Trans. R. Soc. A 1948 , 240 ,

599 – 642 . [ 11 ] M. Warner , E. M. Terentjev , in Liquid Crystal Elastomers, Vol. 2 ,

Oxford University Press , Oxford, UK 2007 . [ 12 ] I. A. Rousseau , P. T. Mather , J. Am. Chem. Soc. 2003 , 125 ,

15300 – 15301 . [ 13 ] J. Küpfer , H. Finkelmann , Makromol. Chem., Rapid Commun. 1991 ,

12 , 717 – 726 . [ 14 ] H. Finkelmann , E. Nishikawa , G. G. Pareira , M. Warner , Phys. Rev.

Lett. 2001 , 87 , 011501 . [ 15 ] V. Ahir , E. M. Terentjev , Nat. Mater. 2005 , 4 , 491 – 495 . [ 16 ] W. Lehmann , H. Skupin , C. Tolksdorf , E. Gebhard , R. Zentel ,

P. Krüger , M. Lösche , F. Kremer , Nature 2001 , 410 , 447 – 449 .

[ 17 ] A. Sanchez-Ferrer , A. Merekalov , H. Finkelmann , Macromol. Rapid. Commun. 2011 , 32 , 671 – 678 .

[ 18 ] W. Kossak , P. Papadopoulus , P. Heinze , H. Finkelmann , F. Kremer , Macromolecules 2010 , 43 , 7532 – 7539 .

[ 19 ] H. Finkelmann , S. T. Kim , A. Muñoz , P. Palffy-Muhoray , B. Taheri , Adv. Mater. 2001 , 13 , 1069 .

[ 20 ] J. Harden , M. Chambers , R. Verduzco , P. Luchette , J. T. Gleeson , S. Sprunt , A. Jáckli , Appl. Phys. Lett. 2010 , 96 , 102907 .

[ 21 ] P. Heinze , H. Finkelmann , Macromolecules 2010 , 43 , 6655 – 6665 .

[ 22 ] A. Buka , W. H. de Jeu , Physique 1982 , 43 , 361 – 367 . [ 23 ] A. Sánchez-Ferrer , R. Mezzenga , H. Dietsch , Macromol. Chem. Phys.

2011 , 212 , 627 – 634 . [ 24 ] A. Sánchez-Ferrer , M. Reufer , R. Mezzenga , P. Schurtenberger ,

H. Dietsch , Nanotechnology 2010 , 21 , 185603 . [ 25 ] M. Reufer , H. Dietsch , U. Gasser , B. Groberty , A. M. Hirt ,

V. K. Malik , P. Schurtenberger , J. Phys.: Condens. Matter. 2011 , 23 , 065102 .

[ 26 ] V. Städele , U. Gasser , H. Dietsch , Soft Matter 2012 , 8 , 4427 – 4431 .

[ 27 ] J. A. Osborn , Phys. Rev. 1945 , 67 , 351 – 357 . [ 28 ] G. R. Mitchell , A. Windle , in Orientation in Liquid Crystal Polymers ,

Elsevier Applied Science , London, New York 1988 . [ 29 ] R. Lovell , G. R. Mitchell , Acta. Crystallogr. Sect. A 1981 , 37 , 135 –

137 . [ 30 ] F. Bergmüller , C. Bärlocher , B. Geyer , M. Grieder , F. Heller ,

P. Zweifel , Meas. Sci. Technol. 1994 , 5 , 1466 – 1470 . [ 31 ] Y. Watanabe , J. Compos. Mater. 2002 , 36 , 915 – 923 . [ 32 ] J. J. Abbot , O. Ergeneman , M. O. Kummer , A. M. Hirt , B. J. Nelson ,

IEEE Trans. Robotics 2007 , 6 , 1247 – 1252 . [ 33 ] J. Kim , S. E. Chung , S.-E. Choi , H. Lee , J. Kim , S. Kwon , Nat. Mater.

2011 , 10 , 747 – 752 .

1791wileyonlinelibrary.commbH & Co. KGaA, Weinheim

Copyright WILEY-VCH Verlag GmbH & Co. KGaA, 69469 Weinheim, Germany, 2013.

Supporting Information for Adv. Mater., DOI: 10.1002/adma.201204406 Liquid-Crystalline Elastomer-Nanoparticle Hybrids with Reversible Switch of Magnetic Memory Johannes M. Haberl, Antoni Sánchez-Ferrer, Adriana M. Mihut, Hervé Dietsch, Ann M. Hirt, and Raffaele Mezzenga*

Submitted to

1

Supporting Information for: Liquid-Crystalline Elastomer-Nanoparticle Hybrids with Reversible Switch of Magnetic Memory By Johannes M. Haberl, and Antoni Sánchez-Ferrer, Adriana M. Mihut, Hervé Dietsch, Ann M. Hirt, Raffaele Mezzenga* J. M. Haberl, Dr. A. Sánchez. Ferrer, Prof. R. Mezzenga [*], ETH Zürich, Department of Health Science and Technology 8092 Zürich, Switzerland E-mail: raffaele.mezzenga@hest.ethz.ch

Dr. A. M. Mihut†, Dr. H. Dietsch†† Adolphe Merkle Institute and Fribourg Center for Nanomaterials, University of Fribourg 1723 Marly, Switzerland

Prof. A. M. Hirt ETH Zürich, Department of Earth Science 8092 Zürich, Switzerland In order to obtain homogeneous liquid-crystalline elastomer nanocomposites with well-dispersed nanoparticles into the network, special care was taken from the early functionalization of the nanoparticles up to their integration in the organic matrix, devoted to prevent sedimentation at any stage of the process. This included sonication and periodic vortexing during the process. More details of the individual steps taken to the synthesis of the hybrid nanocomposites can be summarized as follow: Synthesis of the liquid-crystalline polymer Dimethyl biphenyl-4,4-dicarboxylate (12.1 g, 44.8 mmol, 1 eq) was mixed with triethylene glycol (7.06 g, 47.0 mmol, 1.05 eq), and titanium (IV) isopropoxide (5 mg). The mixture was heated to 200 °C in a nitrogen atmosphere for 6 h, and methanol was distilled off. The volatile components were removed and the temperature was increased to 230 °C for 1 h to obtain the liquid-crystalline polymer.[1] In two extra steps, an excess of triethylene glycol (0.025 eq) was added to the mixture repeating the same procedure. The final linear polymer was dissolved in dichloromethane and purified by three times precipitating from methanol to obtain the product as a slightly yellowish glassy material (12.8 g, 70%). 1H NMR (400 MHz, CDCl3) δ = 8.10-7.95 (m, 4H, Ar-H), 7.65-7.50 (m, 4H, Ar-H), 4.50-4.37 (m, 4H, -CO2CH2-), 3.90-3.50 (m, 8.8H, -OCH2-) ppm; MS (MALDI-TOF): Mn = 2920 g⋅mol-1, Mw = 3510 g⋅mol-1, DP = 7.8, PDI = 1.2 (Fig. S1). Synthesis of the spindle type maghemite nanoparticles (SCH NPs). In a first step, the bare spindle hematite, α-Fe2O3, nanoparticles (BH NPs) were synthesized based on the method described by Ocaña et al.[2] The particles were coated with a layer of silica using the approach of Graf et al.[3] based on an initial adsorption of polyvinylpyrrolidone (PVP) on the particles to improve their colloidal stability and the subsequent addition of tetraethylorthosilicate (TEOS) as a precursor for the growth of the silica shell. The silica-coated spindle hematite nanoparticles (SCH NPs) dispersion was dried in an air oven at 90 °C for 24 hours. The dried powder is then annealed in a furnace at 360 °C under a continuous hydrogen gas flow. After 2 hours, the hydrogen flow is turned off and the

Submitted to

2

powder exposed to air.[4] The furnace temperature is decreased to 240 °C during 2 hours. The obtained maghemite particles (SCM NPs) have a hybrid composition consisting of 70% maghemite and 30% hematite as determined from XRD data with Rietveld method. Surface functionalization of maghemite nanoparticles Surface functionalized silica coated maghemite nanoparticles (SCM NPs) were obtained following the previously reported method.[5] The surface modification was ensured using 3-aminopropyltriethoxysilane (APTES) coupling agent (ABCR, Germany) used without previous purification. In a typical example, 1 g of SCM NPs was transferred in a mixture of water (330 mL), absolute ethanol (1 L) and tetramethylammonium (12 mL, 25% solution in methanol). 23.6 g of APTES were added to the mechanically stirred suspension under sonication for 2 h at 20 °C. After stirring the suspension overnight, the obtained amino-functionalized SCM NPs were centrifuged at 10000 rpm for 15 min and redispersed in ethanol, repeating this process five times. Thereafter, the mixture was dispersed in 30 mL of dichloromethane. Synthesis of the organic-inorganic nanocomposite A dispersion of MNs in dichloromethane (3.17 g, 2.365 wt-%) was ultrasonicated during 30min, and then added to a triisocyanate crosslinker solution (67.4 mg, 0.130 mmol, 1.0 eq, Basonat HI100-BASF) in freshly distilled dichloromethane (1 mL). The crosslinker and the MNs were kept for 2 h, and the liquid-crystalline polymer (571 mg, 0.391 mmol, 3 eq) was dissolved in dichloromethane (absolute, 1 mL), together with dibutyltin dilaurate (1.55 g, 6 wt-% in dichloromethane). The two mixtures were merged, vortexed, poured in a Petri dish and kept for 18 h. The resulting crosslinked film was heated to 85 °C for 2 days. In order to remove all soluble content swelling in chloroform was used to obtain the final fully crosslinked elastomer (80%). Methods: Differential Scanning Calorimetry (DSC) experiments were performed on a DSC 1 calorimeter from Mettler Toledo equipped with a Huber TC100 cooling system, where the nanocomposite was encapsulated in a 40 µL aluminum oxide crucible under nitrogen atmosphere. The sample was analyzed in a temperature range from -20 °C to 200 °C with heating and cooling rates of 10, 15, 20 and 25 K⋅min−1. The first heating curves were used for removing all thermal history from the sample, and the obtained transition temperatures were extrapolated to the combined 0-heating rate transitions. Bright-field images were taken with a Canon 550D digital camera. Transmission Electron Microscopy (TEM) micrographs were obtained on a Philips CM100-Biotwin microscope operating at 80 kV. The nanocomposite sample was ultramicrotomed using a Diatome diamond knife on a Reichert-Jung UltraCut E Microtome to give 80 nm thick sections. Sections were transferred onto 600-mesh copper grids. For the MNs, the sample was prepared by placing some drops of the 0.1 wt-% silica coated MNs suspension onto a carbon-coated copper grid. Small and Wide Angle X-ray Scattering (SWAXS) experiments were performed using a Rigaku MicroMax-002+ microfocused beam (4 kW, 45 kV, 0.88 mA) with the λCu Kα = 0.15418 nm radiation in order to obtain direct information on the scattering patterns. The scattering intensities were collected by a Fujifilm BAS-MS 2025 imaging plate system (15.2 x 15.2 cm2, 50 µm resolved) and a 2D Triton-200 X-ray detector (20 cm diameter). An effective scattering vector range of 0.05 nm-1 < q < 25 nm-1 was obtained, where q is the scattering wave vector defined as q = 4π·sinθ/λCu Kα with a scattering angle of 2θ. For the order parameter of the nanoparticles, the SAXS pattern was evaluated at low scattering angles (q = 0.15-0.25 nm-1) that have been related to the nanoparticles orientation previously[6] and where scattering is weak in a neat liquid-crystalline elastomer reference sample (Fig. S7). Low-field magnetic susceptibility

Submitted to

3

experiments were performed on a Kappabridge MFK1-FA from Agico. High-field magnetic torque experiments were performed with a homebuilt torque magnetometer with an accuracy of 2⋅10-3 J⋅kg-1. Detailed information on the instrument and the experimental method were previously published.[7] The ferromagnetic tensors OFF and ON were calculated from 15 measurements, using 6 different magnetic field values from B = 1000 mT to 1500 mT (100 mT steps). Uniaxial deformation of the sample at a rate of 1 mm s-1 (0.05 s-1) up to λ = 3.2 lead to a shrinkage in the other two directions according to Poisson’s ratio for elastomers. The stretched sample was cut into a piece of length l’z1 = 60 mm, with a width of l’x1 = 5.1 mm and a thickness of l’y1 = 0.080 mm, it was folded two times to l’z = 20 mm and l’y = 0.240 mm (ON). The thin film in the OFF state had a length lz =18.5 mm width of lx = 9.1 mm and thickness of ly = 0.140 mm. At the operating temperature of 80 °C the sample relaxed fast to its original dimensions due to entropic elasticity, as confirmed by X-ray analysis.

Submitted to

4

Calculation of anisotropic magnetic susceptibility The calculation of the anisotropy of magnetic susceptibility was done on the basis of a Stoner-Wohlfarth model:[8,9] As the magnetization M of a sample can be calculated from the volume susceptibility Kv with

HKM v ⋅= (1) at the applied magnetic field H, the anisotropy of magnetic susceptibility

MMKK 2121 = . (2) In order to calculate M in the nanocomposite, we assume that the magnetic properties are the result of an ensemble of non-interacting single-domain particles following the assumptions made by Stoner and Wohlfarth.[8,9] Magnetic coupling of grains or grains and the liquid-crystal[10] is not assessed in this calculation, because of separation and isolation by the silica shell. So the sample magnetization in the measuring direction is proportional to the mean particle magnetization

( ) ( )

( )∫ ⋅

∫ ⋅

=∝ 2

0

2

0,,

sin

sin

π

π

θθθ

θθθθ

dP

dMPMM

i

isampleisample (3)

with the particle population P(θ), where θ is the angle between the major particle axis and the applied field H, and i is the measuring direction (parallel to H), z or x. In order to get the orientation distribution function P(θ) X-ray scattering patterns were evaluated[11,12] in azimuthal steps am+1/2

( ) ( ) ( )[ ]1sin

1

21221 +−⋅=+

+ mSmSaN

aPm

mπ

, (4)

where 0 ≤ m ≤ (N - 1),

( ) ( ) ( ) ( ) ⎥⎦

⎤⎢⎣

⎡−−−⋅∑= +

−

=−−

2122211

22121 coscoscoscos nmnm

N

mnnN aaaaaImS (5)

and

2π⋅⎟⎠

⎞⎜⎝

⎛=Nnan . (6)

In order to calculate M(θ) of individual particles, the magnetic energy U has to be minimized[8]

Submitted to

5

0=ΦddU , (7)

where φ is the angle between M(θ) and the applied field H. The magnetic energy per volume V can be assumed to be the sum of the magnetostatic energy density ums, due to the demagnetizing field in the shape anisotropic particles, and the Zeeman energy density uH due to the orientation of these in the magnetic field, when we assume that no magneto-crystalline anisotropy is present.

Hms uuVU

+= (8)

with

( ) ( )θφ −⋅⋅−= 22cos

2MNNu s

parallellarperpendicums . (9)

If we assume a constant aspect ratio R, which is a reasonable assumption due to the narrow and mono-modal shape distribution that was found for the particles, we can calculate the demagnetizing factors [13]

12 =⋅+ larperpendicuparallel NN (10)

and

⎟⎟

⎠

⎞

⎜⎜

⎝

⎛

−−

−+

−⋅⋅

−=

1

1ln121

12

2

22 RR

RR

R

RR

Nparallel . (11)

Further, the Zeeman energy density is calculated

φcos⋅⋅= HMu sH . (12) When eq. 9 and eq. 12 are inserted in eq. 8 the differential eq. 7 becomes ( ) ( ) ( ) 0sincossin2 =⋅⋅+−−⋅⋅−− φθφθφ HMMNN ssparallellarperpendicu (13)

or

( )[ ] 0sin22sin =⋅⋅+−⋅ φθφ h (14) with

( )[ ]parallellarperpendicus NNMHh −= . (15)

This equation can be solved numerically following [8]

Submitted to

6

( ) ( ) ( )( ) ( ) 012212sin2cos12 21222

12 =−±−+− mhmmm θθ (16) with m = M / Ms = cos φ, yielding discrete results for the magnetization

( ) mMM si ⋅=θ . (17) For the solution, we chose steps of π / 180 in the limits of [0, π/2]. By entering Mi(θ) as defined by eq. 17, eq. 3 yields iparticleM , . From the calculated four energy minima of the

magnetization, we chose the maximal magnetization curve (Fig. S5), which should correspond to the global minima of the energy curve; this scenario, which excludes potential intermediate metastable magnetization states, might be one reason for the overestimation of the anisotropic magnetization based on this theoretical calculation. Thus, eq. 4 had to be solved for the sample when it was measured in z-direction (K1) and when it was measured in x-direction (K2). For the OFF state, obviously, the isotropic scattering intensity of the X-ray experiment gave an aspect ratio of K1/K2 = 1. However, for the ON state, the distribution functions P1 and P2 were calculated once peaked around 0° and once around 90°, which were inserted into eq. 2 along with eq. 3. When the measuring field of H = 200 A m-1, the aspect ratio of 5.6 and the Ms = 120 kA m-1 were taken into account K1’/K2’ = 2.1 was obtained. In figure 3 in the main manuscript, the aspect ratios of the measured susceptibilities are shown as red ellipses for OFF and ON. Inserted, blue ellipses with the aspect ratio of the particles are presented, placed at an angle θ, corresponding to the definition of the 3D order parameter[11,

12]

( ) ( )

( )∫

∫ −⋅⋅

=−= 2

0

2

0

2

2

sin

sin1cos321

1cos321

π

π

θθθ

θθθθθ

dP

dPS m , (18)

but in 2D projection. Therefore, the order parameter S = 0 corresponds to θm = 54.7° in 3D, which is represented with the ellipse at θ = 45° in the corresponding 2D projection. In the same manner the order parameter of S = 0.56 corresponds to θm = 32.8° in the 3D sample and is shown in its 2D projection at θ = 24.5°.

Submitted to

7

Supporting figures

Figure S1 TEM micrographs of the liquid-crystalline elastomer nanocomposite: different regions in the sample all proof the good dispersion with occasional small aggregates of several particles.

Figure S2 TEM micrographs of the coated maghemite nanoparticles in high resolution: the core-shell-type of the particles is visible. The structure in the darker core region might indicate porosity.

Submitted to

8

Figure S3 TEM micrographs of the coated maghemite nanoparticles dispersed in water: the aspect ratio of the particles is 5.6 and constant in all particles to a good approximation.

Figure S4 The solutions of the derivative eq. 7 calculated from eq. 16 correspond to the stable (1) and metastable magnetizations (2-4) of the relative magnetization in the measuring direction (proportional to cosφ) of a particle at the azimuthal angle θ. To proceed further on the calculation, we assumed that all particles would be magnetized at the maximum, e.g. reached the lowest energy configuration, which is indicated with the red curve.

Submitted to

9

Figure S5 Azimuthal scattering intensity distribution of SAXS experiment performed with the sample ON in z-direction. The peaks correspond to the in-layer confined particles after stretching with the degree of order of approximately S = -0.04.

Submitted to

10

Figure S6 MALDI-TOF mass spectroscopy of the liquid-crystalline polymer in DCTB / Na Mix 1:10:1. 72.2% of the total intensity is assigned to P173 which is the di-hydroxyl terminated chain, 16.5% of the intensity corresponds to P55, the mono-hydroxyl and mono-methyl ester terminated chain that will insert terminal groups in the network, and 9.0% of the intensity corresponds to P23, the macro-cyclic byproduct of the trans-esterification reaction.[14]

Figure S7 Scattering intensity of the SAXS pattern for the liquid-crystalline elastomer nanocomposite and a neat reference sample.

Submitted to

11

References _[1] A. Martínez-Gómez, E. Pérez, C. Álvarez, Polymer 2009, 50, 1447-1455.

_[2] M. Ocaña, M. P. Morales, C. J. Serna, J. Colloid Interf. Sci. 1999, 212, 317.

_[3] C. Graf, D. L. J. Vossen, A. Imhof, A. van Blaaderen, Langmuir 2003, 19, 6693-6700.

_[4] T. Chappuis, I. Bobowska, S. Hengsberger, E. Vanoli, H. Dietsch, Chimia 2011, 65,

979-981.

_[5] A. Sánchez-Ferrer, M. Reufer, R. Mezzenga, P. Schurtenberger, H. Dietsch,

Nanotechnology 2010, 21, 185603.

_[6] M. Reufer, H. Dietsch, U. Gasser, B. Grobety, A. M. Hirt, V. K. Malik, P.

Schurtenberger, J. Phys.: Condens. Mater. 2011, 23, 065102.

_[7] F. Bergmüller, C. Bärlocher, B. Geyer, M. Grieder, F. Heller, P. Zweifel, Meas. Sci.

Technol. 1994, 5, 1466-1470.

_[8] R. C. O`Handley, in Modern Magnetic Materials: Principles and Applications, Wiley-

Interscience publication, New-York, 1999.

_[9] E. C. Stoner, A. Wohlfarth, E. P. Phil. Trans. R. Soc. A 1948, 240, 599-642.

_[10] D. I. Santiago-Quiñone, A. Acevedo, C. Rinaldi, J. of Appl. Phys. 2009, 105, 07B5121-

3.

_[11] G. R. Mitchell, A. Windle, in Orientation in Liquid Crystal Polymers, Elsevier

Applied Science, London, New York, 1988.

_[12] R. Lovell, G. R. Mitchell, Acta. Cryst. A 1981, 37, 135-137.

_[13] J. A. Osborn, Physical Review 1945, 67, 351-357.

_[14] H. R. Kricheldorf, M. Rabenstein, M. Maskos, M. Schmidt, Macromolecules 2001, 34,

713-722.