1

FERROGELS BASED ON ENTRAPPED METALLIC IRON NANOPARTICLES IN

POLYACRYLAMIDE NETWORK: EXTENDED DERJAGUIN-LANDAU-VERWEY-

OVERBEEK CONSIDERATION, INTERFACIAL INTERACTIONS AND

MAGNETODEFORMATION

Ajay Shankar1, Alexander P. Safronov1,2, Ekaterina A. Mikhnevich1, Igor V. Beketov1,2,

Galina V. Kurlyandskaya1,3

1
Ural Federal University, Yekaterinburg, Russian Federation

2
Institute of Electrophysics, RAS, Ural branch, Yekaterinburg, Russian Federation

3
University of the Basque Country UPV-EHU, Department of Electricity and Electronics, 48940,
Leioa, Spain

Abstract. A new kind of ferrogel with entrapped metallic iron nanoparticles causing unusual

magnetodeformation is presented. The crosslinked polyacrylamide (PAAm) based ferrogels

embedded with iron nanoparticles (MNPs) were synthesized by free radical polymerization in

aqueous medium. Spherical iron MNPs with average diameter 66 nm were synthesized by the

electrical explosion of wire and modified by interfacial adsorption of linear polyacrylamide

(LPAAm). Extended Derjaguin-Landau-Verwey-Overbeek (xDLVO) calculations based on the

superposing of electrostatic, Van-der-Waals, steric, and magnetic contributions showed that

polymeric encapsulation of nanoparticles by LPAAm is one of the most suitable pathways for

preparing stable aqueous dispersions of iron nanoparticles. Microcalorimetry confirmed the

presence of strong interfacial adhesion forces between LPAAm chains and surface of iron

nanoparticles. By keeping the same crosslinking density of polymer network (i.e. 100:1,

monomer to crosslinker ratio) and varying the initial monomer concentration, an influence of the

2
extent of polymer network reticulation on mechanical properties and subsequently, magneto-

elastic properties were demonstrated. It was found that upper limit of shear modulus for

synthesized new kind of polyacrylamide based ferrogels to exhibit any usable

magnetodeformation under application of the uniform external magnetic field of 420 mT is ca. 1

kPa. Magnetodeformation of cylindrical ferrogel samples was observed in the form of an overall

volume contraction accompanied by a homogeneous decrease in all dimensions. The deformation

was found maximum (around 10%) for aspect ratio 1/1 and it was lower and similar for the

samples with 1/2 and 2/1 aspect ratios. Such type of magnetic response is significantly different

from the behavior observed in existing reports on ferroealstomers.

1. Introduction.

Polymeric gels, especially hydrogels are extensively studied as advanced materials for

biomedical and bioengineering applications due to their ability to undergo large mechanical

deformations in response to a variety of external stimuli such as ionic concentration, pH,

temperature, solvent composition, electrical field etc.1 By their nature polymeric gels have very

low sensitivity with respect to an external magnetic field due to either para- or/and diamagnetic

response of the gel components. At the same time, substantial efforts are being given to add

magnetic field stimuli to the list. The obvious way to increase a polymeric gel sensitivity to the

magnetic field up to well detectable level is to introduce magnetic particles in its structure, which

will respond to the application of the external field and mediate its influence to the polymeric

matrix. The presence of ferromagnetic fillers embedded into polymeric network imparts

responsive ability when exposed to external magnetic field. Ferrogels comprise a polymeric

network swollen in a solvent and ferro- or superparamagnetic particles embedded in it. Among

3
stimuli-responsive materials, ferrogels have attracted considerable scientific interest because of

their potential biomedical applications in targeted drug delivery systems2, soft actuators1,

artificial muscles3, regenerative medicine4, and biosensors5. In ferrolastomers and ferrogels, the

kind of magnetoresponse i.e., expansion or contraction, is controlled by the demagnetizing fields

features, shape anisotropy factors and the change in magnetic susceptibility due to magnetic

interaction contributions as a result of change in mutual disposition of magnetic centers6. In

ferrogels, these features responsible for magnetoelasticity can be affected by concentration and

type of magnetic fillers, nature of polymeric network, extent of crosslinking in polymer network

and choice of imbibed solvent.

Typically, the synthesis of a ferrogel involves polymerization and/or cross linking of a

fluidic system with dispersed magnetic nanoparticles (MNPs). Prior to synthesis, it requires a

concentrated and stable dispersion of magnetic nanoparticles in suitable carrier medium. The

stabilization of MNPs against aggregation and sedimentation in fluids requires surface treatment

by suitable stabilizers, modifiers or surfactants. Thus, subsequent stabilization of dispersion

necessarily provides a uniform distribution of magnetic particles after the gel synthesis. Due to

this reason, the reported ferrogels are exclusively based on mixed ferrite particles (i.e., Fe3O47,8,

CoFe2O49,10 and Fe2O311,12) as there are a variety of means for the preparation of their stable

dispersions in various carrier mediums. Meanwhile, there is a demand for the other magnetic

materials with stronger magnetization to be used as fillers in ferrogels. For instance, metallic iron

particles are well known for their superior magnetization compared to previously attempted

ferrites as magnetic fillers for same field strength. In applications for which high magnetic

permeability is desired the magnetic particles should have domain walls, i.e. to be in a

multidomain state. Therefore, the assumption about critical particle diameter for the transition

4
from multidomain into single domain state is important. The critical diameter for above

discussed transition in the case of spherical metallic iron nanoparticle and strong anisotropy is

close to 6 nm and it is close to 60 nm for -Fe2O3 if one assumes two magnetic domains state and

that the domain wall in a particle has the same structure as that in an infinite wall case. For the

weak anisotropy case, the magnetization orientation will tend to follow the surface of the

spherical particle and critical diameter for transition into single domain state appears to be as

high as 50 nm13. Although multidomain iron particles as magnetic filler are expected to enable

enhanced magneto-response of ferrogel or even giant magnetodeformation, the estimation of the

critical diameter for transition into multidomain state requires hard work of fabrication of stable

dispersion of iron MNPs of rather large size. The choice of iron particles as magnetic component

for ferrogel system makes our work different from previous reports7,8,1012, where mostly mixed

ferrite particles are used. In our recent report14, a successful attempt for the preparation of

ferrogel embedded with oleate modified iron particles has been demonstrated. It also revealed the

problem of strong aggregation of iron MNPs in the reaction mixture.

Magnetic nanoparticles are widely needed for biosensor applications. A magnetic

biosensor is a compact analytical device in which an external magnetic field variations are

converted by a magnetic transducer into a change off frequency, current or voltage. Different

types of magnetic effects are capable of creating magnetic biosensors but the search for the

enhanced sensitivity with respect to the applied field is a strong present day demand5. The

magnetic label detection principle is simple: the stray fields induced by the magnetic labels are

employed as biomolecular labels providing a way of the measurement of the concentration of the

biomolecules of interest 15,16. The sensitivity limit is related to the type of MNPs as the magnetic

moment of an individual MNP in the external magnetic field governs the stray field features.

5
However, one of the most requisite for cancer therapy cases, the detection of the MNPs

incorporated into biological tissues, has not been yet properly addressed. The development of

such detector is conditioned by the reliable sampling as the biological tissues present a huge

variety of morphologies. In our earlier published work, we propose to substitute biological tissue

samples at the first stage of the development of biosensor by the synthetic hydrogel with certain

amount of iron oxide MNPs, which is capable to mimic main properties of the living tissues.

Despite the successful demonstration of capability of thin film based magnetoimpedance sensor

element for a detection of -Fe2O3 MNPs in ferrogel synthesized by radical polymerization of

acrylamide in a stable aqueous suspension, the obtained sensitivity was still below desired4.

Testing of magnetoimpedance biosensor with iron particles embedded in ferrogel can be the next

step of the sensitive magnetic biosensor development.

In present study, we used aqueous suspension of multidomain Fe MNPs stabilized by

linear polyacrylamide (LPAAm) chains. At macro and micron scale, the stability aspects of

similar systems, but relatively simpler ones, are modelled using classical Derjaguin-Landau-

Verwey-Overbeek (DLVO) approach17. Therein the additivity of interaction potentials is usually

considered and has been found consistent with experimental results. Meanwhile, this approach

has been extended to nano level in many colloidal systems. In addition, the inclusion of steric

and magnetic forces alongwith classical DLVO forces has been widely reported17 with modified

approach being called as extended DLVO (xDLVO) approach. Although a significant certainty

about existence of nondditive nature of interaction potentials, unlike in xDLVO approach, has

been proposed at nano level due to multiscale collective effects18. Though, in absence of a well-

defined theoretical model to include such interplay of interactions at nanoscale, the xDLVO

approach is indeed a reasonable option. Thus, in view of above, we focused on the stability

6
aspects of prepared suspension which were modeled using xDLVO approach. The steric

stabilization by polymeric chains was shown to be the only route to overcome strong magnetic

interactions between iron MNPs in the reaction mixture for the synthesis of ferrogels. The

synthesized suspension was later used for preparing the crosslinked polyacrylamide (PAAm) gel

network filled with surface modified iron MNPs. The PAAm network is well known for its

nonporous nature where mesh size is too small which allows diffusion of only molecular species

through its polymeric network19. This special feature is one of the important reasons to make

them suitable choice to test for biomedical applications. Further, the optimization of softness of

gel matrix decides the responsive magneto-elastic behavior, i.e. equilibrium of mechanical

strength vs magneto-response. In present work, it has been shown to be affected by taking

different initial monomer concentrations.

2. Experimental section

2.1. Materials

2.1.1. Magnetic particles

Iron magnetic nanoparticles (10-200 nm in diameter) were synthesized using the electric

explosion of wire technique, which was described in detail in our earlier reports20. The method is

based on the evaporation of the iron wire under the high-voltage impulse in the inert working

gas. The following parameters were used. The applied pulsed voltage and length of exploded

wire were 30 kV and 90 mm, respectively. The reaction chamber was maintained under

circulating mixture of 70% Ar and 30% N2 at 0.12 MPa. The production rate of MNPs was as

high as 200 g.hr-1. The produced MNPs were collected in the filter and removed from it after

passivation by in-leakage of oxygen with a flow rate of 0.5 cm3 per second. This leads to surface

passivation due to formation of amorphous metal oxide layer with 3-5 nm of thickness20. The air-

7
dry MNPs were further used in preparation of MNP/LPAAm composites and in the synthesis of

ferrogels.

2.1.2. MNP/LPAAm composites

Linear polyacrylamide was synthesized by heating 10% aqueous solution of acrylamide (AAm,

99%, AppliChem, Darmstadt, Germany) in presence of 0.06% H2O2 as initiator at 80 oC for 1 hr.

The molecular weight of LPAAm was determined by viscometry of its dilute aqueous solutions

(< 1% w/w). Using Mark-Houwink constants, K = 6.31 10-5 dL/g and a = 0.88, the molecular

weight of LPAAm was determined to be 143.6 kDa.

The MNP/PAAm composites for microcalorimetry studies were prepared in the entire range of

weight fraction of MNPs i.e., 0 - 100%. The composite preparation involved grinding the proper

proportions of MNPs and LPAAm 10% water solution in agate mortar till homogenized mixtures

were obtained in each case. The obtained suspension was then cast upon Teflon plate and dried

to the constant weight at 70 oC. The obtained films of MNP/PAAm composites were then used

for the microcalorimetry measurements of the enthalpy of dissolution in distilled water21.

2.1.3 Ferrogels

N,N-Dimethylacrylamide (MDAA, >98%, Merck, Schtuchardt, Germany), acrylamide (AAm,

99%, AppliChem, Darmstadt, Germany), ammonium persulfate (APS, Chemreaktivsnab, 98%,

China) and N,N,N,N-Tetramethylethane-1,2-diamine (TEMED, 99%, Merck, Germany) were

used as received. The gels were synthesized at room temperature by free radical polymerization

of AAm as monomer using MDAA as cross-linker22. APS and TEMED were used as initiator

and catalyst, respectively. All solid reagents except iron nanoparticles were dissolved in water

prior to synthesis. The homogenization of reaction mixture was carried out after addition of each

8
reactant at every step. To provide ferrogels with different strength and subsequent threshold of

magnetodeformation the monomer concentration in the reaction mixture was varied while the

molar ratio monomer to cross-linker was set as 100:1 in all gel preparations (Table. 1). 0.5 g of

bare iron particles and 2.5 g of 10% solution of LPAAm were used as magnetic filler and

modifier in all ferrogel preparations. In a typical synthesis procedure of ferrogels, 3 g of a

suspension of MNPs in the solution of LPAAm were ground in an agate mortar and then were

mixed with 1 mL of 48 mM MDAA solution followed by addition of 0.341 g of crystalline

AAm. After the complete dissolution of AAm in the mixture 0.5 mL of 50 mM APS solution

were added. In the last synthetic step, 20 L of TEMED was added and vigorously mixed for

few seconds before transferring to cylindrical polyethylene mold and subsequently left

undisturbed for ~ 4 hrs. The gels were taken out of mold and washed for at least 14 days by

renewal of distilled water at interval of 2 days. The preparation of blank hydrogels was carried

out in the absence iron particles keeping the rest of the procedure the same. Ferrogels and blank

hydrogels without MNPs were labelled as FG and HG series, respectively. All synthesized gels

retained their cylindrical shape but varied in their physical softness. Compositions, swelling

degree and softness of ferrogels and their respective hydrogels are outlined in Table 1. The

equilibrium swelling degree () of gels was determined gravimetrically and calculated by = (ms

- md)/ md, where ms is the mass of swollen gel and md is the mass of dried gel. The equilibrium

swelling degree values were further used in the approximation of mass fraction of iron particles

() in swollen FGs.

mMNPs 1
' (1)
mMNPs mAAm 1

9
mMNPs is the mass of MNPs in the reaction mixture and mAAm is the total mass of acrylamide units

including both LPAAm and monomer. The latter two parameters were known from the followed

synthesis procedure. By knowing experimental swelling degree of gels, the mass swelling ratio

() solely related to crosslinked polymer network can be calculated. Thus, for hydrogels =

whereas, for ferrogels,

mMNPs mAAm
' (2)
mAAm

Further, the volume fraction of polymer in swollen gels, p,s can be calculated by employing

densities of both polymer (p = 1.349 g.cm-3) and solvent (w = 1.0 g.cm-3) as

1
p (3)

1 ' p
w

Table 1. Description and selected characteristics of ferrogels (FGs) and hydrogels (HGs)

AAm Swelling Swelling degree of Weight Physical

Sample mark conc. degree of PAAm () fraction of softness of gels
(M) gels () MNPs (%)
Ferrogels
FG-01 2.05 9.6 4.3 Soft
17.9
FG-02 1.76 9.7 19.5 4.6 Very soft
FG-03 1.46 14.9 32.7 3.4 Extremely soft
Hydrogels
HG-01 2.05 23.4 - Soft
23.4
HG-02 1.76 29.5 29.5 - Very soft
HG-03 1.46 61.7 61.7 - Extremely soft

2.2. Methods.

10
The powder X-ray diffraction (XRD) patterns were recorded on Bruker D8 Discover with Cu

K1.2 radiation ( = 1.542 ) with graphite monochromator. The Rietveld renement of XRD

patterns were performed using Topas-3 software. The morphology of MNPs was examined using

JEOL JEM2100 Transmission electron microscope (TEM) operating at 200 kV. The specific

surface area of MNPs was measured by the low-temperature adsorption of nitrogen (Brunauer-

Emett-Teller (BET) approach) using Micromeritics TriStar3000 analyzer. The magnetic

measurements were carried out using (Cryogenics Ltd. VSM) vibrating sample magnetometer

(VSM) at room temperature for powder samples placed in a gelatine capsule. The magnetization

values in a field of 1.8 T were designated as the saturation magnetization values (Ms). Dynamic

light scattering (DLS) and electrophoretic light scattering (ELS) measurements were performed

using Brookhaven ZetaPlus particle size analyzer: 5 and 3 runs were recorded for hydrodynamic

size and zeta-potential measurements, respectively. Thermogravimetric analysis (TGA) of

LPAAm coated MNPs was carried out using a NETZSCH STA409 thermal analyzer. TGA

measurements were performed from room temperature to 1000oC at heating rate 10 oC / min.

Microcalorimetry measurements of the enthalpy of dissolution of MNP/PAAm composites in

distilled water were performed at 25 oC using SETARAM C80 microcalorimeter. The glass

ampoule technique was elaborated. The weighted sample of composite film (20 40 mg) in a

thin glass ampoule was placed in a stainless steel cell filled with 10 mL of distilled water. After

thermal equilibration, the ampoule was broken by a special rod and the enthalpy of dissolution

was measured with 2% accuracy. The bright field optical microscopy on thin slices of ferrogels

were performed using Olympus BX51 microscope using 10X and 40X objective lenses. Shear

modulus was measured using a laboratory setup providing compressive loading of gel samples

and their simultaneous optical registration during deformation. The values of shear modulus

11
were taken as the slope of the plot vs (1/2 ), where was the normalized stress and was

the relative compression. Magnetodeformation of ferrogels in magnetic eld was monitored by

an optical system equipped with programmable photocamera in a laboratory setup. A magnetic

system of NdFeB permanent magnets provided a uniform magnetic eld of 420 mT directed

against gravity in the central zone of 10 10 10 (mm3).

3. Results and their discussion.

3.1. Structural and magnetic characterization of MNPs.

Air-dry iron MNPs, before the synthesis of ferrogels, were characterized by XRD, TEM and

VSM. TEM image, SEM image, and particle size distribution (PSD) of MNPs are given in

Figure 1. As shown in Figure 1 the particles are spherical in shape. MNPs do not coalesce as no

junctions between separate particles are noticeable. PSD of the ensemble of MNPs as determined

by the analysis of TEM images was found lognormal (Figure 1a inset) with parameters: mean

particle size 66 nm and dispersion 0.47. The specific surface area of MNPs determined using

BET approach was 9.0 m2/g.

12
Figure 1. TEM (a) and SEM (b) microphotographs of iron MNPs. Inset presents the particle size

distribution with respect to diameter based on the analysis of over 1000 particles. N is the

number of MNPs in the certain range of the diameter. The line in the inset corresponds to the

lognormal fit of the histogram. The mean particle diameter is 66 nm with log-standard deviation

of 0.47.

X-ray diffraction patterns of MNPs show Bragg peaks corresponding to (110), (200) and

(211) reflections which are characteristic of -Fe phase (Figure 2a). The Rietveld refinement of

diffraction patterns yielded average crystallite size of 61 4 nm.

Figure 2. (a) XRD pattern of MNPs. 1 experimental points, 2 Rietveld refinenment, 3

difference between calculated profile and observed data. (b) M-H magnetization curves of
MNPs. Inset show detailed view of loops at low fields.

The hysteresis loop of MNPs measured at T = 20 oC is shown in Figure 2b: M is a

magnetization and H is an applied field. The low field behaviour (inset for M-H loop) reveals

13
the existence of magnetic hysteresis and coercivity Hc 30 mT. As usually magnetic hysteresis

is understood as the dependence of the magnetization on the external magnetic field due to the

complex non-linear response of magnetic system related to the existence of metastable energy

minima separated by energy barriers. It is well known that obtaining the exact coercivity results

by modeling is a very difficult problem whose rigorous solution is possible only for a few simple

cases like coherent rotation model of Stoner-Wolfarrth. Although some progress had been made

in solving the coercivity problem by numerical simulation the volume of the simulated structures

is limited by 106 cubic lattice points. In the case of pure iron, it is equivalent to about 30 nm 30

nm 30 nm cube. It immediately means (Figure 1a) that the behavior of at least two-thirds of the

ensemble of individual MNPs cannot be predicted using simulation techniques. The coercivity of

small particles is often well described by the coherent rotation model but incoherent or curling

modes can appear even in perfectly spherical particles of the larger size.

As to expect for an assay of spherical iron MNPs with average diameter of about 65 nm

the majority of them are in multidomain state. At the same time, one cannot exclude the presence

of a small fraction of single domain MNPs contributing to non-zero of the remnant

magnetization (Mr 110 kA/m) and coercivity. The value of the saturation magnetization in bulk

state for pure iron is well known13: Ms = 1710 kA/m at room temperature. The obtained Ms

1250 kA/m is about 27% below the bulk value for pure iron. Is obtained difference reasonable?

As it was mentioned above due to strongly pyrophoric features of iron MNPs their surface must

be passivated prior to exposure to the atmospheric oxygen. In our previous studies20,23, we have

shown that passivation oxide layer for low oxygen flow rates is usually about very few nm.

Considering the passivation layer of 3 nm for 33 nm radius MNP and keeping in mind the value

of the lattice constant a 0.83 nm for Fe3O4 iron oxide one can very roughly estimate the

14
thickness of four upper layers as 3.3 nm in plausible agreement with observed reduction of the

magnetization.

Another way to understand the reduction of the saturation magnetization value is based

on the well-known concept of laws at nanoscale24. In the case of iron nanoparticle at least three

surface layers are not contributing to the ferromagnetic response due to the insufficient number

of the nearest neighbours. For the lattice constant a 0.286 nm for pure -Fe one can obtain a

corresponding reduction of Ms down to 92% of the Ms value of the bulk iron. Both processes are

contributing to Ms reduction but it is difficult to make more precise analysis first of all due to the

existence of the MNPs size distribution. For example, we have calculated passivation layer of 3

nm contribution for D = 66 4 nm. Considering lower value of 31 nm for the diameter and

making corresponding reduction of Ms we obtain the corresponding Ms reduction down to 74%

of the bulk Ms value. This does not mean that provided analysis is not reliable, on the contrary,

the experimental values of magnetic parameters are in a good agreement with structural data and

established theoretical models but the nanosystem is just very complex and requires careful

analysis of all parameters.

3.2 Analysis of the stability of suspensions of Fe MNPs using xDLVO theory.

The aggregation features of elaborated iron MNPs, which are essential for the synthesis of

ferrogels can be qualitatively modeled by xDLVO approach. In classical DLVO theory, the

attractive and repulsive interactions are modeled for Van der Waals and electrostatic interactions

only. In case of the magnetic colloidal dispersions where both steric and magnetic interactions

are present, they must also be taken into account. The modified approach to consider all these

interactions is known as xDLVO approach17.

15
 The Van der Waals interactions were calculated using classical Hamaker approach.

Instead of using combining relations17 to approximate Hamaker constant, we used the calculated

size dependent Hamaker constant (A) values for iron-water-iron system. The calculation of

Hamaker constant for the ensemble of polydisperse iron MNPs in water (see Electronic

Supplementary Information) gives a third-order exponential decay function.

A r 1.77 1019 1.60 1019 e r /3.05 6.35 1020 e r /10.75 2.05 1020 e r /52.18 [J] (4)

where r is the radius of a particle in nm. Using the A(r) function (Equation(4)) for Hamaker

constant the energy of Van der Waals interactions (VvdW) between iron MNPs depended on their

size was calculated as17

A(r ) 2r 2 2r 2 (4r s)
VvdW ln s 2
(5)
6 s 4r s (2r s) 2
(2r s)

Here, s is surface-to-surface separation of particles. The electrostatic repulsion between particles

for .r < 5 was taken as25

Ve 2 r r o o2 e s (6)

1

k T 2
where, 2 B o r2 is the inverse Debye length. Here r is the relative dielectric
q N A zi ci
constant of water, 0 is the permittivity of free space, 0 is the surface potential, q is the
elementary charge, zi is the charge of simple ions, ci is their molar concentration, NA, kB, and T
have their usual meanings.
The steric repulsion between MNPs was taken into account by a combination of a hard core

potential with a soft tail potential, as modeled in the self-consistent field (SCF) theory26. The

steric repulsion between two identical MNPs with adsorbed polymeric layer can expressed as

follows:

16
 for s < 0

r p k BT
3 3
s 9 s 1 s 1 s
3 6

Vst ln 1 1 1 for 0 < s < 2 (7)

2
2 5 2 3 2 30 2
12 N p l

0 for s > 2

where is the thickness of adsorbed polymeric layer, p is the surface density of adsorbed

polymeric chains, Np is the number of Kuhn segments in the chain, and l is the length of one

segment.

Magnetic interactions between Fe MNPs, which are to be taken into account, depend

strongly on the size of the particles. The superparamagnetic MNPs are well known to have zero

magnetization in the absence of the external magnetic field. However, in the case of multidomain

particles, it is not as simple as for superparamagnetic MNPs: even very slight deviations from the

shape of surface anisotropy features may cause curling mode appearance. On the other hand,

single-domain MNPs with physical size lying in range of magnetic domain size have finite

intrinsic magnetic moments. In case of iron, the threshold of superparamagnetic to single domain

transition lies near 6 nm for particle radius for a short term stability of the order of 1 second27. At

the same time, superparamernetism limit describing particle size for magnetic stability against

ambient thermal demagnetization is smaller than the critical radius (25 nm) for single domain

limit of Fe particles in a small anisotropy consideration model. Even so, according to the PSD of

the ensemble of iron MNPs obtained from TEM images (Figure 1a), the considerable fraction of

MNPs stay in the single domain range and thus they can impart finite magnetization to the

ensemble through magnetic particle-particle interactions even in the absence of external applied

magnetic field. The maximum magnetic attraction energy (VM) between MNPs can be written

as17

17
 8o M r 2 r 3
VM 3
(8)
s
9 2
r

where 0 is the permeability of vacuum and Mr is the remnant magnetization of iron MNPs.

Thus, to model aggregation stability of aqueous suspension of Fe MNPs using xDLVO approach

the total energy of particle to particle interaction comprised two attraction terms: van der Waals

(Equation (5)) and magnetic (Equation (8)) and two repulsion terms: electrostatic (Equation (6))

and steric (Equation (7)).

V Vvdw Ve Vst VM (9)

To perform the calculations, we took the necessary parameters of MNPs obtained in their

structural characterization. We used the value 86 nm for the weight average diameter of MNPs

estimated from the PSD. This value is in very good agreement with the estimation of the mean

diameter based on the specific surface area of the MNPs. The specific surface area (Ssp) of

spherical MNPs is related to its average diameter (D) (nm) according to the simple equation,

which stems from geometry consideration:

6
D (10)
S sp

where, is the density of a particle. The value of the specific surface area Ssp=9.0 m2/g for Fe

MNPs gives 85 nm for their average diameter.

We used the value of zeta-potential of the suspension of Fe MNPs (= -16 mV) for the

estimation of surface potential (0) in the calculation of electrostatic repulsion. The remnant

magnetization value for MNPs was taken as Mr =114 kA/m according to experimentally

measured hysteresis loop (Figure 2 b). In the calculation of the steric repulsion we used, l=2.0

nm for the length of the Kuhn segment of LPAAm. The number of segments in one chain

18
Np=290 was estimated using the molecular weight of LPAAm, the surface density of adsorbed

polymeric chains p=0.14 was assumed based on the weight of the adsorbed polymer per m2 of

MNP surface as determined by TG measurement. The dependence of the attraction and repulsion

terms on the particle to particle separation calculated using these parameters is presented in

Figure 3.

It is evident that two terms are dominant in the total energy of the interaction between Fe

MNPs. These are the magnetic interaction VM (attraction) and steric interaction Vst (repulsion).

The repulsion electrostatic term is very small as compared to VM and van der Waals attraction

became substantial only at very short distances. It is worthwhile to note that the magnitude of the

magnetic term is very large. It reaches the values around several tens of kBT at relatively large

distances between particles. It means that the magnetic forces between single-domain MNPs,

keeping the moment corresponding to the remnant magnetization effectively overwhelm all other

types of interaction possibly except the steric ones.

19
Figure 3. Repulsion and attraction terms in the energy of Fe MNPs interaction in the suspension
in dependence of particle to particle separation (s). Parameters used: r = 43 nm, = -16.0 mV,
Mr = 114 kA/m, l = 2.0 nm, Np = 290, p = 0.14, = 20 nm. 1 Van der Waals attraction, 2 -
electrostatic repulsion, 3 - steric repulsion, 4 magnetic forces related attraction.

Considering the steric repulsion, it should be noted that it becomes very strong if the adsorbed

layers on the surface of interacting MNPs overlap. Equation (7), which is the product of self-

consistent field (SCF) theory28 implies that steric repulsion vanishes to zero if the particle to

particle distance is larger than the double thickness of the adsorbed layer. Therefore, the balance

between attraction magnetic forces and steric repulsion strongly depend on the thickness of this

layer. Figure 4 presents the plots of the total energy of interaction between Fe MNPs calculated

according to Equation (9) with different values of the thickness of the adsorbed layer.

Figure 4. Overall interaction energy curves in dependence of particle-to-particle separation (s)

for aqueous dispersion of iron MNPs coated with variable LPAAm layer thickness. 1 10 nm, 2
- 15 nm, 3 20 nm, 4 30 nm, 5 40nm.

20
 It might be noted that the superposing of the magnetic and steric terms resulted in the

minimum at the curve of the total energy of interaction. The particle to particle separation, which

corresponds to the minimum, correlates with the thickness of the layer on the surface of MNPs.

If the thickness diminishes, the minimum shifts to small particle to particle distances. Taking into

account the dependence of the magnetic forces on separation, it provides the strengthening of

magnetic attraction between particles. It is clear that the depth of the minimum progressively

increases if the thickness of the protective polymeric layer decreases. Obviously, this effect is

crucial for the stabilization of Fe MNPs suspensions. The minimum at the total energy curve

corresponds to the formation of the aggregates of MNPs in suspension. Whether they are stable

or not depends on the relation between the depth of the minimum and the energy of thermal

motion. It is conventionally accepted that the aggregates can be disrupted by the thermal motion

if the corresponding minimum is less than 20 kBT, as statistically, only a few particles will cross

barrier in this case29. Thus, if the protective layers on the surface of Fe MNPs are thin the

aggregation of particles due to magnetic forces is very strong and the stable suspensions of

individual particles cannot be obtained. The only possibility for the deaggregation of Fe MNPs in

suspension is to provide thick polymeric protective layers on their surfaces.

It is worthwhile noting that the quantitative analysis was done using specific structural

and magnetic parameters of Fe MNPs. Some of them are quite definite, but some are not.

Namely, it corresponds to the value of magnetization of the single domain particles which

governs the magnetic interaction (Equation (8)). It was noted that we have evaluated the

maximum attraction, which corresponded to the remnant magnetization of Fe MNPs. Actually;

the magnetization of the particles might be lower. Therefore, the analysis given above should be

21
considered as the upper limit for the aggregation forces. Meanwhile, the qualitative trends, which

were disclosed, remain the same at any certain value of Fe MNPs magnetization. According to

these trends, the successful dispersion of Fe MNPs can be achieved only with the use of

polymeric steric stabilizers.

We can add one simple illustration to how much this conclusion depends on the particle

size distribution in a low size tail part. Above we have discussed such a limit as the critical

radius for superparamagnetic spherical particle transition into a single domain state which

depends on the magnetic anisotropy constant Ku and the value of flipping probability over the

selected time interval. For Ku 105 J/m3, the time difference of 1 year and 1 second represent

difference between 7.3 nm and 6.0 nm radius of superparamagnetic transition threshold. These

energies are governed by either (10kBT)1/3 or (6kBT)1/3 terms accordingly. Coming back to the

observation that aggregates can be disrupted by the thermal motion if the corresponding

minimum is less than 20 kBT, one can see that energies involved into superparamagnetic to

single domain transition are much smaller for practically important time scale of a few months of

colloid stability. Therefore, the conclusions about the dispersion of Fe MNPs will be affected

very little by the change of the low size tail part of the size distribution curve (Fig. 1 a).

3.3. Steric stabilization of iron MNPs by LPAAm

Based on this conclusion of xDLVO consideration we tested linear polyacrylamide

(LPAAm) for this purpose. At the first step, we have evaluated the ability of LPAAm to interact

with the surface of iron MNPs and to provide its efficient coverage. The basic thermodynamic

property, which stands for that ability is the enthalpy of adhesion, which is the enthalpy change if

LPAAm macromolecule adsorbs on the surface and establishes adhesive contacts with it due to

molecular interactions. Such enthalpy change cannot be measured directly in calorimetric

22
experiment as both counterparts iron MNPs and LPAAm are solids. To evaluate the enthalpy

of adhesion one should elaborate an appropriate thermochemical cycle, which includes

measurable processes, which contain the desired process as their combination. In the case of the

enthalpy of mixing of a composite such thermochemical cycle constitutes of the following steps.

1) Wetting of air-dry iron MNPs in water. 2) LPAAm dissolution in the excess of water. 3)

Mixing of aqueous suspension of MNPs and aqueous solution of LPAAm. 4) Dissolution of

LPAAm/MNPs composite with certain MNPs to LPAAm ratio in the excess of water.

Typically, the value of the enthalpy change in Step 3 is much less than the enthalpy

changes in the other steps and usually it is neglected.

Figure 5. Scheme of the thermochemical cycle elaborated for the determination of the enthalpy
of mixing of a filled polymeric composite. is the weight fraction of filler in a composite.

23
 The composites LPAAm/MNPs with MNPs weight fraction () varying in 10 90%

range, were prepared and equilibrated as described in Experimental section. The calculated

enthalpy of the combined process of mixing of composite is given in Figure 6 as a function of the

weight fraction of iron MNPs.

In a wide composition range (weight fraction of iron MNPs up to 70%) the enthalpy of

interaction between LPAAm and MNPs is negative, which means strong interaction at the

interface. The enthalpy of mixing at MNPs weight fraction above 70% is positive. It means that

other contributions to the enthalpy of mixing besides the enthalpy of interfacial adhesion are

present and they become dominant at high MNPs content. To evaluate separately these

contributions to the enthalpy we elaborated thermodynamic model for the enthalpy of mixing of

filled polymer composites taking into account previous works21,30.

Figure 6 Enthalpy of mixing of LPAAm/MNPs composite at 298 K as a function of the weight

fraction of iron MNPs. Line interpolation of experimental points by Equation (11).

24
 The model implies the formation of the polymer adsorption layer on the surface of solid

particles and the glass transition of a polymeric matrix in this layer due to the strong adhesion.

The enthalpy of mixing between polymer and filler is then given by the equation:

K (1 ) Ssp (1 )

H comp H ads cohp exp
Sspp L
(11)
K (1 ) Ssp

Here H ads is the characteristic enthalpy of polymer adsorption at the solid surface per 1 m2, K

is the apparent constant of adsorption, coh is the cohesion enthalpy of polymer matrix per 1 g of

polymer, p is the volume fraction of polymer in composite, p is the density of polymer, L is the

characteristic thickness of the adsorption layer, is the excess fraction of metastable voids in the

glassy structure of polymer at the surface.

The solid line in Figure 6 corresponds to the fitting of the experimental points using

Equation (11) using the set of fitting parameters: H ads = 110 J/m2, K = 0.36 m2/g , L = 0.76

m, = 0.025. Concerning the adhesive interaction of LPAAm to the solid surface of iron

MNPs, the first fitting parameter is of major importance. The characteristic enthalpy of polymer

adsorption at the surface H ads according to the model used is the enthalpy of the formation of

an adsorption layer up to its maximum capacity. Its large negative value means that LPAAm

chains are strongly interacting with the surface of iron MNPs.

The strong adhesion of LPAAm leads to irreversible adsorption of its macromolecular

chains on the surface of MNPs. It was confirmed by TGA of MNPs with the adsorption layer.

Adsorption of LPAAm was performed in the suspension of MNPs in the aqueous solution of

LPAAm (10%). The particles were collected and then re-dispersed in water to wash out free

25
LPAAm. Several washing cycles were performed and thermogravimetry analysis of MNPs was

done then. According to it, the extensively washed MNPs contained around 10% of LPAAm

irrespective of the number of washing cycle. This value was used for the estimation of

parameters and p of the xDLVO model (see section 3.2, Equation (7)). Note that strong

adsorption of a certain portion of LPAAm to the surface of MNPs does not affect the validity of

thermochemical cycle in Figure 5, as the same thermodynamic state of MNPs with strongly

adsorbed LPAAm layer can be achieved either by step 3 or by step 4 of the cycle.

Thus, thermodynamic consideration of the system insures the favored deposition of

LPAAm on the surface of iron MNPs providing their steric stabilization in water suspensions.

Figure 7 shows the features of multimodal distribution obtained by DLS of the suspension of

iron MNPs stabilized by LPAAm.

Figure 7. Particle size distribution (PSD) determined by DLS in diluted aqueous suspension of
iron MNPs sterically stabilized by LPAAm. (a) numerical PSD. (b) weighted PSD. Numbers

26
give the number (a) and weight (b) fractions of individual particles and aggregates. The
concentration of LPAAm coated MNPs was 0.23 g/L.

According to DLS studies, PSD in the suspension of iron MNPs stabilized by LPAAm is

bimodal. It contains a fraction with particle diameter 150 200 nm and a fraction with the

average diameter around 1000 nm. The first mode likely corresponds to individual MNPs. The

average diameter of air-dry MNPs as determined from TEM is around 90 nm, which is less than

the size of MNPs corresponding to the position of the first mode. Meanwhile, these particles are

sterically stabilized by the layer of LPAAm. The thickness of this layer can be estimated using

the residual amount of LPAAm on the surface of MNPs determined by thermal analysis. As

noted above the weight fraction of LPAAm was 10%. Corresponding conversion into volume

fraction gives 40% of polymer. The calculation of the thickness of a layer at the surface of the

spherical particle with the diameter 90 nm gives ca 8 nm for the layer. Meanwhile, this value

corresponds to the dry layer of polymer on the surface. If the MNPs are dispersed in water the

polymeric layer swells and its thickness increases. If we assume that the conformation of

LPAAm macromolecules in the layer is a random Gaussian coil, the volume fraction of a

polymer in a coil is given by the following relation31

63/2
G (12)
8 No

N0 is the number of monomeric units in the chain. The molecular weight of LPAAm (143.6 kDa)

gives N0=2000, which in turn gives G=0.04. It is a reasonable estimation for the volume fraction

of LPAAm in a swollen Gaussian coil. Thus, in water, the volume of LPAAm layer increases by

a factor of 1/0.04=25. Then, the thickness of a layer increases up to ca. 70 nm. Of course, it is

rough estimation based on the assumption that LPAAm coils at the surface are Gaussian. In

27
reality, they are compressed due to the adhesion forces, which bind the chains to the surface.

Then, the volume fraction of polymer in the adsorbed layer is higher. Thus, the calculation done

should be considered as the upper level estimation. Meanwhile, using this calculation one can

estimate the average diameter of MNPs with the steric layer to be as high as 230 nm, which is

very close to the value of the first mode experimentally measured by DLS. The second mode in

the DLS plot (Figure 7) certainly corresponds to the aggregates of MNPs. These aggregates are

large. Although their fraction in numerical PSD (Pn) is not dominant (9.5%), it becomes

dominant (94.5%) in weighted PSD (Pw). It means that there is a large number of individual iron

MNPs in the suspension but the main weight fraction corresponds to aggregates. In general, this

result is consistent with the xDLVO theory for the steric stabilization, which predicts the stability

of iron MNPs coated with polymer in suspension. Meanwhile, calculations were done for the

ensemble of monodisperse particles. Magnetic forces strongly depend on the diameter of

particles and in the particle size range both superparamegnetic-to-single domain and single

domain-to-multidomain state transitions are possible as well as appearance of the curling

magnetization states. Therefore, the presence of large particles in the actual ensemble of MNPs

might certainly be a reason for the processes of the substantial aggregation in suspension.

3.4. Structure and mechanical properties of PAAm ferrogels with embedded iron MNPs

PAAm ferrogels with sterically stabilized iron MNPs were synthesized as described in

the Experimental Section. The typical appearance of a hydrogel and a ferrogel is given in Figure

8 (a). To determine the extent of particle aggregation in synthesized ferrogels, the optical images

were taken from thin slices of gels. Figure 8 (b) shows an example of the bright field image of

FG-03. The aggregation of MNPs in ferrogel is clearly observed. Apparently, there are at least

two levels of aggregation. The statistical size analysis of images were performed using image

28
analysis program ImageJ32 by analyzing ~ 300 aggregates in total. The inset in Figure 8 (b)

shows the size distribution of aggregates in ferrogel with LPAAm modified iron nanoparticles. A

two-parameter Weibull distribution function was used for fitting statistical size distribution of

aggregates (mean size of aggregates is 1.40.7 m).

Figure 8 (a) Typical appearance of a hydrogel (HG series) and a ferrogel (FG series). (b) Bright-
field optical microscopic image of FG-03. Inset: size distribution of aggregates fitted with
Weibull distribution function. Mean size of aggregates is 1.40.7 m.

The image analysis program gives the average diameter of aggregates predominantly in

the range of 1-2 m. This value is close to the hydrodynamic diameter of aggregates in the

suspension of sterically stabilized MNPs (see Figure 7). Therefore, one may conclude that the

aggregates, which existed in suspension, were also kept aggregated in the ferrogel matrix.

Individual particles, which are also present in the suspension, cannot be observed by optical

microscopy due to the resolution limit. These aggregates constitute the low level of aggregation

of MNPs in ferrogel. At the same time, the distribution of micron-sized aggregates is not

uniform in ferrogel. One can easily notice large clusters in images. These clusters of primary

aggregates are incorporated into irregular super-network structure on top of polymeric network

29
of ferrogel. The formation of the super-network of magnetic particles in ferrogel was earlier

reported for strontium ferrite micron-sized particles embedded in gel of poly(acrylamide-co-

potassium acrylate)21. Although there are distinct loose and dense areas in the super-network of

clusters of aggregates, ferrogel is uniform by an eye view (Figure 8 a).

The important structural feature of ferrogel is the relation among the polymeric network,

forming the continuous matrix of the gel, and the superstructure of magnetic particles, aggregates

and clusters, which are imbedded into the matrix. There are two questions to answer with this

respect: Can polymeric matrix of a gel be considered as a continuous medium and are the

MNPs/aggregates fixed in the matrix or they have sufficient degree of freedom to move around?

The answer to the first question needs physicochemical consideration of the mesh size of

polymeric hydrogel. The equilibrium conformation of electrically neutral PAAm subchain in

water is random Gaussian coil with hindered rotation33, which mean that square end-to-end

distance <R2> can be calculated according to the equation:

1 cos
R 2 Nb a 2 (13)
1 cos

where Nb is the number of bonds in the subchain, a is the bond length, is the bond angle.

The end-to-end distance between the cross-links, calculated using Equation (13) was 5.5 nm for a

= 0.154 nm for the ordinary CC bond, 109.5 for the bond angle, and Nb = 100 for the

number of bonds among cross-links. This value is by an order of magnitude smaller than median

diameter of PSD for bare iron MNPs (66 nm) and by two orders of magnitude smaller than the

diameter of sterically stabilized by LPAAm iron MNPs (230 nm). It means that the cross-links of

the polymeric network are closer to each other than the particle diameter. For sterically stabilized

MNPs, aggregates and clusters the mesh size of polymeric network becomes progressively

30
negligible. Therefore, it is possible to conclude that the polymeric matrix can be considered as a

continuous media for any type of magnetic structural elements, which exist in the ferrogels under

consideration.

The answer to the second question is based partly on the space limitations and partly on

the adhesion of polymeric subchains to the surface of embedded particles. As it was shown

above, the mean diameter of MNP is much larger than the mesh size of polymeric matrix. It

means that MNPs are entrapped in the network. As for the adhesion, it was shown that PAAm

chains strongly interact with the surface of iron MNPs (see Section 3.3, Figure 6). It means that

due to both reasons iron MNPs, their aggregates, and clusters are affixed in PAAm gel. From this

assertion, the swollen PAAm network can be supposed to serve as nonporous network for

magnetic fillers which allow only water molecules to move in and out. The scheme of the

structural setup of the PAAm ferrogel with embedded iron MNPs is given in Figure 9.

Figure 9. Scheme of the structural composition of PAAm ferrogel with embedded sterically
stabilized MNPs and their aggregates. Characteristic dimensions of gel mesh size, iron MNPs
and their aggregates are compared.

31
 The mechanical strength of gel networks was measured as stress-strain dependence under

unidirectional compression in absence of external magnetic field. Figure 10 (a) shows the actual

stress strain plots for gels and ferrogels with MNPs. In all cases, the experimental points are

well fitted by a linear dependence, which allowed the calculation of the shear modulus of the

gels. The dependence of the shear modulus on the volume fraction of PAAm in the network is

shown in Figure 10 (b). A monotonous increase of shear modulus with increasing polymer

volume fraction is observed in all studied gel and ferrogel networks. Note that polymer volume

fraction (Figure 10 (b)) represents the polymer content exclusively in swollen gels and excludes

any contribution by MNPs (see subsection 2.1.3).

Figure 10. (a) Stress vs strain curves of ferrogels and hydrogels. (b) Shear modulus vs polymer
volume fraction of all labelled gels. Schematic illustrations are for comparative perspective only.

In general, the parameters which affect mechanical properties of ferrogels are

crosslinking density, polymer network reticulation, swelling degree and the content of filler. In

our study, the cross-linking density of the gel was set constant. The weight fraction of MNPs was

32
close in all swollen ferrogels (see Table 1). It varied by approximately 1% by weight, which

was negligible with respect to the influence on shear modulus. The only value varied

substantially was the content of monomer (AAm) in the reaction mixture, which eventually

resulted in the different values for the swelling degree of polymeric network and hence for the

volume fraction of PAAm chains in its structure (see Table 1). It is worth noting that the swelling

degree decreases if the concentration of monomer increases although the cross-linking density is

kept the same. It is due to the additional entanglements of PAAm subchains, which happen if

their polymerization takes place in the concentrated solutions. The results given in Figure 10

show that the shear modulus increases simultaneously with the volume fraction of PAAm in the

hydrogel matrix.

Now let us compare both gels and ferrogels. Here, one may notice in the corresponding

pairs of gels and ferrogels (HG-01-FG-01 etc.) that the polymer volume fraction is higher in the

case of FGs. It indicates that MNPs grafted with LPAAm chains diminish the swelling degree of

the polymeric network. Most likely, it stems from the difference of the conformation of LPAAm

chains in the gel and ferrogel. In the case of gels, the LPAAm chains are in the Gaussian coil

conformation, whereas in the case of ferrogels they are absorbed on the surface of MNPs. As it

was shown above (see section 3.3.), the LPAAm chains exhibit strong adhesion to the surface of

iron MNPs. It certainly would count for the dense conformations of LPAAm chains attached to

the surface. Consequently, the polymeric matrix of ferrogels is more dense and shows higher

resistance towards the straining.

In addition, it can be seen in Figure 10 that moduli of polymer network are higher for the filled

gels than for the blank gels. So the plot of the shear modulus is shifted to the right and upward.

Hence, the presence of LPAAm modified MNPs affects the stiffness of gel network. In physical

33
sense, the particles themselves are incompressible constituents and impart the same characteristic

to overall gel network.

All FGs were cut in cylindrical shape with aspect ratio ~1, 1/2 and 2/1, and kept under

aqueous medium to test their deformation in the uniform magnetic field of 420 mT. Keeping in

mind that terrestrial magnetic field at the point of measurements and local laboratory fields did

not exceed 0.03 mT value one can easily neglect this kind of interference. The applied field was

parallel to cylindrical axis and Earths gravitational field. To avoid the weight loss of the samples

due to the evaporation of water, the samples were kept in aqueous medium in plastic cuvettes

during magnetodeformation measurements. All FGs except FG-03, either showed a very small

volume decrease (below 1%) or no response to the magnetic field over one-day period. The

significant difference in responsive nature of FGs can be related to the competition between field

induced deformation and the strength of gel network. The applied magnetic field changes the

magnetic state of the MNPs causing the magnetic moments appearance, interaction between

them and creation of an additional force. In a ferrofluid, the result of the application of an

external magnetic field is well known:34 particles form needle-shaped agglomerates constituting

ferrofluid with higher particle concentration. The volume of the agglomerates is larger than that

of the individual MNPs and agglomerates susceptibility become large due to dense packing and

the shape anisotropy (elongated shape of the agglomerates oriented along the field direction)35.

In ferrogel samples, as a consequence of application of external magnetic field one can expect

certain change of the particle configuration and minimization of the free energy of the composite

first of all by minimization of the Zeeman energy of elongated aggregates via magnetic moment

rotation toward the parallel to external field direction. The mobility of the MNPs and their

34
aggregates in ferrogel is, of course, very limited but might be sufficient for macroscopic

deformation in very soft gels.

Although external magnetic field using MNPs as an action instrument induces additional

deformation for polymer network the inherent strength of this crosslinked network resists this

change from occurring. Magneto-elastic behavior of a gel matrix depends on balance between its

mechanical strength and magneto-response. The balance between these contributions is

responsible for macroscopic deformation. The magnetic force induced by the magnetic field is

roughly proportional to the concentration of MNPs in the ferrogel for the case of rather low

concentrations. This concentration is approximately the same for all ferrogels under study. Thus,

the magnetic response for all of them should be the same if the aggregation features are similar.

The origin of the observed difference lies in their inherent mechanical properties. FG-03 is the

softest of the gels and it is responsive to the external field, while FG-02 and FG-01, which are

more rigid do not respond to the field application at the macroscopic level. The non-responsive

FGs have a polymeric network which is strong enough to resist the force created as a

consequence of applied magnetic field. Thus, by considering the modulus values and polymer

volume fraction for corresponding FGs (shown in Figure 10), it can be ascertained that the

threshold value of modulus pertaining to 420 mT field, lies between 0.63 and 2.15 kPa. Note

that FG-03 has a shear modulus under 1 kPa, which indicates its similarity with magnetic field

induced storage modulus of magnetorheological fluids36.

The time dependent changes in volume induced by applied magnetic field for FG-03 with

different aspect ratios (length of the cylindrical sample to its diameter), are plotted in Figure 10

(a), (b) and (c). For the aspect ratio, c/a ~1, FG-03 showed a ~ 10% steady decrease in overall

volume and reaches saturation at around 1200 min. The samples with aspect ratios 1/2 and 2/1

35
show an enhanced volume contraction rate but simultaneously also leads to a decrease in overall

extent of deformation. The total volume contraction is not more than 3%, in both cases.

From thermodynamic viewpoint6, the magnetodeformation observed under homogeneous

field in the form of contraction or elongation depends on the shape of sample and its internal

microstructure from the spatial distribution of particles. First, demagnetizing factor always

stimulates elongation of the sample along the applied magnetic field direction. The

demagnetizing factor is determined only by the shape and the aspect ratio of the sample, not by

the distribution of the particles in it. Second, the change of magnetic susceptibility, as a

consequence of the deformation, can stimulate either, elongation and contraction, depending on

the concentration of the MNPs in the polymeric matrix. In accordance with this, various

simulations on ferroelastomers have predicted either contraction or elongation, of the polymer

matrix3739.

36
Figure 11. Magnetodeformation under 420 mT with time for FG-03 with (a) aspect ratio ~1, (b)
aspect ratio ~1/2, (c) aspect ratio ~ 2/1. The rate of volume decrease for FG-03 with (d) aspect
ratio ~1, (e) aspect ratio ~1/2, (f) aspect ratio ~2/1. Insets show the snapshots of ferrogel samples
with different aspect ratio. Symbols represent the experimentally observed decrease in volume.
All solid curves are fitted exponential function to serve as an eye-guide for the general trend.

In the experimental studies, the elongation of the cylindrical sample of ferroelastomer in

the uniform magnetic is typically observed4044. It is accompanied with simultaneous contraction

along axes orthogonal to the applied field. Meanwhile, the contraction of a flat disk-shaped

ferroelastomer sample in the uniform field was also reported30. The important feature both for

the elongation and contraction of ferroelastomers was that under low deformation cases the

volume change was ignorable. In contrast to these, we observed an overall decrease of both

length and width of cylindrical ferrogel samples leading to comprehensive contraction of volume

under external applied field which was homogenous in the absence of a magnetic sample. Such a

behavior is consistent with all studied aspect ratios (1/1, 1/2 and 2/1) checked in the present

work.

This behavioral dissimilarity may be arising due to the primary difference between

ferrogels and ferroelastomers, i.e presence and absence of the solvent medium in the former and

the latter, respectively. Ferroelasatomers are a two-component system comprising magnetic

particles and polymer matrix, whereas, ferrogels are a three-component system: magnetic

particles, polymer matrix and imbibed solvent. In swollen gels, the solvent molecules may enter

or leave the solvated polymer network. In course of deformation of ferrogels, this flexibility

offered by solvent is expected to greatly affect the whole sample in an isotropic manner.

Obviously, this feature of solvent medium is inaccessible to ferroelastomers during their

magnetodeformation. Typically, the extent of swelling of as synthesized non-solvated

37
crosslinked polymer network is controlled by balance between free energy corresponding to

solvation of polymeric network strands and elastic strength of network. The extent of imbibed

water in course of swelling greatly affects the mechanical strength of gel network and

subsequently its magneto-elastic behavior. So far, this factor of swelling degree is not taken into

account in any of the previous theoretical considerations6,3739,45. In one of the previous reports, it

has been shown that depending on magnetic filler content, both swelling and contraction of

ferrogels can be observed in polyacrylamide ferrogels in uniform magnetic field46.

Further, a careful look at the kinetics of volume contraction also reveal another

interesting feature. Along with a general trend of exponential decrease in volume shown by

colored curves (Figure 11), a weak and superimposed wavy behavior can also be noticed. The

existence of this feature indicates that a consistent volume decrease is resisted by the recovery of

deformed polymer network. This competitive network recovery phenomena is different from the

magneto-mechanical hysteresis reported in previous works22,47,48, where usually a rest time or

cyclic compression-decompression methods were employed to characterize mechanical

hysteresis. In simple words, the overall dynamics of magnetodeformation in the ferrogels studied

in this work is controlled by a complex equilibrium between magnetic forces and polymeric

architecture of the gels.

4. Conclusions.

Ferrogels based on PAAm embedded with linear PAAm modified Fe nanoparticles were

synthesized and studied. Extended Derjaguin-Landau-Verwey-Overbeek calculations showed

that magnetic interaction dominates over classic colloidal interactions (Van der Waals and

electrostatic forces). The only interacting force which can challenge magnetic forces is a steric

one. Even so, the thickness of the steric layer should be of the order of several tens of nanometer.

38
It was shown that linear PAAm was an exemplified steric stabilizer of iron MNPs due to the

strong interfacial adhesion between MNP surface and polymer chains, remaining in the form of a

strongly adsorbed layer. However, a substantial fraction of MNPs still remain aggregated inside

ferrogel matrix. The spatial distribution of MNP aggregates was uniform throughout in all

ferrogels. It has been shown that the mechanical strength of ferrogels can be controlled by the

density of the polymeric network, which is linked to its swelling degree. The upper limiting

value for the observable magnetodeformation is shear modulus ca 1 kPa. A uniform field

induced deformation of ferrogels was observed only in the case of the softest gel. An overall

contraction of volume was accompanied by a decrease in both length and diameter

simultaneously. Such deformation was maximum for the aspect ratio of about 1.

Acknowledgements

This work was supported in part by Russian Science Foundation grant 14-19-00989 and

ELKARTEK grant KK-2016/00030 of the Basque Country Government. Authors appreciate the

special support of Dr. A.I. Medvedev in XRD characterization, Dr. A.M. Murzakayev in TEM

studies and Dr. Andrey Galyas for optical microscopy measurements. AS is thankful to UrFU for

the postdoctoral fellowship.

References

1 J. Thvenot, H. Oliveira, O. Sandre and S. Lecommandoux, Chem. Soc. Rev., 2013, 42, 7099.

2 C.-C. Lin and A. T. Metters, Adv. Drug Deliv. Rev., 2006, 58, 13791408.

39
3 G. Filipcsei, I. Csetneki, A. Szilgyi and M. Zrnyi, in Oligomers - Polymer Composites -

Molecular Imprinting, Springer Berlin Heidelberg, Berlin, Heidelberg, 2007, vol. 206, pp.

137189.

4 N. C. Hunt and L. M. Grover, Biotechnol. Lett., 2010, 32, 733742.

5 G. V. Kurlyandskaya, E. Fernndez, A. P. Safronov, A. V. Svalov, I. Beketov, A. B. Beitia, A.

Garca-Arribas and F. A. Blyakhman, Appl. Phys. Lett., 2015, 106, 193702.

6 A. Y. Zubarev, Soft Matter, 2012, 8, 3174.

7 A. M. AL-Baradi, O. O. Mykhaylyk, H. J. Blythe and M. Geoghegan, J. Chem. Phys., 2011,

134, 094901.

8 R. Muzzalupo, L. Tavano, C. O. Rossi, N. Picci and G. A. Ranieri, Colloids Surf. B

Biointerfaces, 2015, 134, 273278.

9 C. Barrera, V. Florin-Algarin, A. Acevedo and C. Rinaldi, Soft Matter, 2010, 6, 3662.

10 R. Messing, N. Frickel, L. Belkoura, R. Strey, H. Rahn, S. Odenbach and A. M. Schmidt,

Macromolecules, 2011, 44, 29902999.

11 L. Roeder, M. Reckenthaler, L. Belkoura, S. Roitsch, R. Strey and A. M. Schmidt,

Macromolecules, 2014, 47, 72007207.

12 G. V. Kurlyandskaya, D. S. Portnov, I. V. Beketov, A. Larraagad, A. P. Safronov, I. Orue, A.

I. Medvedev, A. A. Chlenova, M. B. Sanchez-Ilarduya, A. Martinez-Amesti and A. V. Svalov,

Biochim. Biophys. Acta BBA - Gen. Subj., 2016.

13 R. C. OHandley, Modern magnetic materials: principles and applications, Wiley, New York,

2000.

14 A. Shankar, A. P. Safronov, E. A. Mikhnevich and I. V. Beketov, J. Magn. Magn. Mater.,

2017, 431, 134137.

40
15 D. R. Baselt, G. U. Lee, M. Natesan, S. W. Metzger, P. E. Sheehan and R. J. Colton, Biosens.

Bioelectron., 1998, 13, 731739.

16 G. V. Kurlyandskaya, J. Magn. Magn. Mater., 2009, 321, 659662.

17 M. Sanchez-Dominguez and C. Rodriguez-Abreu, Nanocolloids: A meeting point for

scientists and technologists, Elsevier, Netherlands, 2016.

18 C. A. Silvera Batista, R. G. Larson and N. A. Kotov, Science, 2015, 350, 1242477.

19 L. M. Lira, K. A. Martins and S. I. C. de Torresi, Eur. Polym. J., 2009, 45, 12321238.

20 A. P. Safronov, G. V. Kurlyandskaya, A. A. Chlenova, M. V. Kuznetsov, D. N. Bazhin, I. V.

Beketov, M. B. Sanchez-Ilarduya and A. Martinez-Amesti, Langmuir, 2014, 30, 32433253.

21 A. P. Safronov, A. S. Istomina, T. V. Terziyan, Y. I. Polyakova and I. V. Beketov, Polym. Sci.

Ser. A, 2012, 54, 214223.

22 S. Rose, A. Dizeux, T. Narita, D. Hourdet and A. Marcellan, Macromolecules, 2013, 46,

40954104.

23 I. V. Beketov, A. P. Safronov, A. I. Medvedev, J. Alonso, G. V. Kurlyandskaya and S. M.

Bhagat, AIP Adv., 2012, 2, 022154.

24 Y. Jun, J. Seo and J. Cheon, Acc. Chem. Res., 2008, 41, 179189.

25 D. A. Walker, B. Kowalczyk, M. O. de la Cruz and B. A. Grzybowski, Nanoscale, 2011, 3,

1316.

26 J. K. Lim, S. A. Majetich and R. D. Tilton, Langmuir, 2009, 25, 1338413393.

27 R. Skomski, J. Phys. Condens. Matter, 2003, 15, R841R896.

28 U. Genz, B. D. Aguanno, J. Mewis and R. Klein, Langmuir, 1994, 10, 22062212.

29 R. E. Rosensweig, Ferrohydrodynamics, Dover books on physics, United States of America,

2014.

41
30 G. V. Kurlyandskaya, A. P. Safronov, T. V. Terzian, N. S. Volodina, I. V. Beketov, L.

Lezama and L. M. Prieto, IEEE Magn. Lett., 2015, 6, 38001044.

31 A. P. Safronov, F. A. Blyakhman, T. F. Shklyar, T. V. Terziyan, M. A. Kostareva, S. A.

Tchikunov and G. H. Pollack, Macromol. Chem. Phys., 2009, 210, 511519.

32 C. A. Schneider, S. R. Wayne and K. W. Eliceiri, Nat. Methods, 2012, 9, 671675.

33 M. Rubinstein and R. H. Colby, Polymer Physics, Oxford University Express, 2003.

34 A. Hubert and R. Schfer, Magnetic domains: the analysis of magnetic microstructures,

Springer, Berlin; New York, 1998.

35 A. Shankar, M. Chand, G. A. Basheed, S. Thakur and R. P. Pant, J. Magn. Magn. Mater.,

2015, 374, 696702.

36 M. Chand, A. Shankar, N. Ali, K. Jain and R. P. Pant, RSC Adv, 2014, 5396053966.

37 O. V. Stolbov, Y. L. Raikher and M. Balasoiu, Soft Matter, 2011, 7, 8484.

38 A. M. Biller, O. V. Stolbov and Y. L. Raikher, J. Appl. Phys., 2014, 116, 114904.

39 D. Romeis, V. Toshchevikov and M. Saphiannikova, Soft Matter, 2016, 12, 93649376.

40 G. Y. Zhou and Z. Y. Jiang, Smart Mater. Struct., 2004, 13, 309.

41 S. Bednarek, J. Magn. Magn. Mater., 2006, 301, 200207.

42 X. Guan, X. Dong and J. Ou, J. Magn. Magn. Mater., 2008, 320, 158163.

43 G. V. Stepanov, D. Y. Borin, Y. L. Raikher, P. V. Melenev and N. S. Perov, J. Phys. Condens.

Matter, 2008, 20, 204121.

44 G. Diguet, E. Beaugnon and J. Y. Cavaill, J. Magn. Magn. Mater., 2010, 322, 33373341.

45 A. Y. Zubarev and D. Y. Borin, J. Magn. Magn. Mater., 2015, 377, 373377.

46 A. P. Safronov, T. V. Terziyan, A. S. Istomina and I. V. Beketov, Polym. Sci. Ser. A, 2012,

54, 2633.

42
47 G. V. Stepanov, S. S. Abramchuk, D. A. Grishin, L. V. Nikitin, E. Y. Kramarenko and A. R.

Khokhlov, Polymer, 2007, 48, 488495.

48 M. P. Algi and O. Okay, Eur. Polym. J., 2014, 59, 113121.

43
ELECTRONIC SUPPLEMENTARY INFORMATION

FERROGELS BASED ON ENTRAPPED METALLIC IRON NANOPARTICLES IN

POLYACRYLAMIDE NETWORK: EXTENDED DERJAGUIN-LANDAU-VERWEY-

OVERBEEK CONSIDERATION, INTERFACIAL INTERACTIONS AND

MAGNETODEFORMATION

Ajay Shankar1, Alexander P. Safronov1,2, Ekaterina A. Mikhnevich1, Igor V. Beketov1,2,

Galina V. Kurlyandskaya1,3

1
Ural Federal University, Yekaterinburg, Russian Federation

2
Institute of Electrophysics, RAS, Ural branch, Yekaterinburg, Russian Federation

3
University of the Basque Country UPV-EHU, Department of Electricity and Electronics, 48940,
Leioa, Spain

Evaluation of Hamaker constant in Fe-water dispersion

We employed the method recently described by Pinchuk et. al.1,2. According to Lifshitz theory,

the Hamaker constant can be written as

3
AH hFe H 2O Fe (S1)
4

where, Fe H2O Fe is Lifshitz constant which can be approximated as

2

2 H 2O (i )
Fe H O Fe 1 d
H O (i ) Fe (i )
2
(S2)
0 2
 H O (i ) and Fe (i ) represent the dielectric permittivity of water and iron as a function of
2

imaginary frequency (i), respectively. Such permittivity of any medium can be written as

2 x Im n ( x)

n (i ) 1 2 dx (S3)
0 x 2

The Lifshitz theory applied here using Equation S1 and S2 includes a first few important terms

for the calculation of nonretarderd Hamaker constants. Theoretically, Equations (S2) and (S3)

are calculated in entire frequency range, i.e. 0 < < . But, experimental data for iron and

water are only available in specific range, which is even narrower for water36. We used recently

measured values of dielectric permittivity of water by Hayashi et. al.4 There are two reasons for

this choice. First, it covers a wider range i.e. 1-100 eV as compared to previously reported data

for water by Hale et. al.3 i.e. 0.006 6 eV, which were used for size dependent Hamaker constant

calculations in previous reports.1,2,7 Second, the measured data for optical constant and

subsequently dielectric permittivity values in 1 - 7.2 eV range differ significantly in these two

literatures.3,4 A significant discrepancy in measured values of optical constants in high

wavelength region for water is widely reported due to limited accuracy in experimental setup.3,8,9

Due to these reasons, we opted for more reliable reported values4 of dielectric permittivity in

wider frequency range for water in calculations. Thus, lower and upper bounds for the integrals

were set as 1 and 100 eV, respectively.

The size induced effect on Fe was calculated using modified Drude model.2 The Drude model is

given as

p2
Drude 1 2 (S4)
i bulk
 ne2
where, p and bulk = 0.01p.
om

p and bulk are plasma frequency and electron damping constant for bulk iron, respectively. n =

17 X 1028 m-3, e = 1.6 X 10-19 C, o = 8.85 X 10-12 F.m-1 and m = 9.1 X 10-31 kg are free electron

density, electronic charge, vaccum permittivity and effective electron mass, respectively.10 Finite

size effect results in the change electron scattering rate which can be taken into account by

writing it as

Fe
( D) bulk 2 A (S5)
D

where, Fe = 1.98 X 106 m.s-1 is the Fermi velocity of electron in iron10, D is the diameter of

nanoparticle, and coefficient A 1. Thus, modified Drude model incorporated with size effect

can be written as

p2
Drude ( D) 1 2 (S6)
i ( D)

The modified permittivity values for nanosized iron ( Fe nano ( D) ) can be calculated using

Equations (S4),(S5) and (S6), and bulk permittivity values of iron ( Fe bulk ( D) ) by1,2

Fe nano ( D) Fe bulk Drude ( D) Drude (S7)

Using above calculated values for iron and data for water from literature4, the values were

incorporated in Equations (S1) and (S2) for variable size of iron nanoparticles. The calculated

values for Hamaker constants are shown in Figure S1. The values can be well represented by a

third order exponential decay which was further used in xDLVO calculations.
Figure S1. Size-dependent Hamaker constant for iron-water-iron system. Blue symbol and red
line represents calculated points and exponential fit, respectively. Blue line represents Hamaker
constant value for bulk iron.

References
S1 A. O. Pinchuk, J. Phys. Chem. C, 2012, 116, 2009920102.

S2 K. Jiang and P. Pinchuk, Nanotechnology, 2016, 27, 345710.

S3 G. M. Hale and M. R. Querry, Appl. Opt., 1973, 12, 555563.

S4 H. Hayashi and N. Hiraoka, J. Phys. Chem. B, 2015, 119, 56095623.

S5 M. A. Ordal, R. J. Bell, R. W. Alexander, L. A. Newquist and M. R. Querry, Appl. Opt.,

1988, 27, 12031209.

S6 S. Adachi, The Handbook on Optical Constants of Metals: In Tables and Figures, World

Scientific, Singapore, 2012.

S7 L. A. Wijenayaka, M. R. Ivanov, C. M. Cheatum and A. J. Haes, J. Phys. Chem. C, 2015,

119, 1006410075.

S8 D. J. Segelstein, Thesis, University of Missouri, 1981.

S9 F. M. Sogandares and E. S. Fry, Appl. Opt., 1997, 36, 86998709.

S10 N. W. Ashcroft and N. D. Mermin, Solid State Physics, Harcourt College Publishers,

1976.