Correlation between the microstructure and the electromagnetic properties of carbonyl iron filled polymer composites

This article was published in an Elsevier journal. The attached copyis furnished to the author for non-commercial research and

education use, including for instruction at the author’s institution,sharing with colleagues and providing to institution administration.

Other uses, including reproduction and distribution, or selling orlicensing copies, or posting to personal, institutional or third party

websites are prohibited.

In most cases authors are permitted to post their version of thearticle (e.g. in Word or Tex form) to their personal website orinstitutional repository. Authors requiring further information

regarding Elsevier’s archiving and manuscript policies areencouraged to visit:

http://www.elsevier.com/copyright

http://www.elsevier.com/copyright

Author's personal copy

Correlation between the microstructure and the electromagneticproperties of carbonyl iron filled polymer composites

Madina A. Abshinova a, Alexander V. Lopatin a, Natalia E. Kazantseva a,b,*,Jarmila Vilcakova a, Petr Saha a

a Polymer Centre, Faculty of Technology, Tomas Bata University in Zlin, nam. T.G.M. 275, Zlin 762 72, Czech Republicb Institute of Radio Engineering and Electronics, Russian Academy of Sciences, Fryazino, Moscow region 141190, Russia

Received 14 November 2006; received in revised form 31 July 2007; accepted 1 August 2007

Abstract

Electromagnetic properties of polymer composites based on different types of carbonyl iron (CI) are investigated in the frequencyrange from 1 MHz to 10 GHz. A significant difference in the high frequency permeability of composites filled with primary and processedCI powders is revealed, although the chemical composition and particle size distribution of these powders show small difference. Com-posites based on processed CI exhibit two regions of magnetic dispersion and higher absolute values of permeability and permittivity inthe radio-frequency (RF) band. The observed differences are attributed to the microstructure of particles; namely, these differencesdepend on whether or not the particles are characterized by ‘‘onionlike’’ multilayered morphology. Electron microscopy and X-ray dif-fraction analyses show structural changes in the processed CI, which are responsible for the variety of electromagnetic properties of thecomposites.� 2007 Elsevier Ltd. All rights reserved.

Keywords: A. Metal-matrix composites; B. Electric properties; Magnetic properties; Microstructure

1. Introduction

Carbonyl iron (CI) has long been used as a component ofelectronic devices such as plastic encapsulated inductorcores [1–6] and electromagnetic wave absorbers (EWAs)[7–11]. Though, in the case of EWAs, it is quite difficultto predict the frequency dependence of the permittivitye*(f) = e 0 � je00 and permeability l*(f) = l 0 � jl00 of CI com-posites from the characteristics available in the productinformation (such as electrical resistivity (q), initial mag-netic permeability (li), and quality factor (Q factor). More-over, the electromagnetic properties of CI powdersproduced by different companies, for example, by BASF

Corporation, Vogt (Germany), READE, GAF and Ami-don, Inc. (USA), INCO-MOND (UK), ONJA (France),Labdhi Chemical Industries (India), and SINTEZ (Russia),are evaluated using the electromagnetic characteristics ofmagnetic cores produced from these powders; the lattercharacteristics depend not only on the fabrication methodsof test samples but also on the measurement methods. Acomparative analysis of CI powders that we have carriedout using information available in the literature [1–17]and the product information has shown that the CI powdersproduced by BASF and by Russian companies have lowerhysteresis losses compared with the CI powders producedby the firms of UK, India, France, and USA; this fact isattributed to the higher dispersity of the former two typesof powders. However, Russian CI powders have higherRF losses because of less efficient isolation of ferrite parti-cles. The initial magnetic permeability li of different typesof CI powders does not show any significant variation,

1359-835X/$ - see front matter � 2007 Elsevier Ltd. All rights reserved.

doi:10.1016/j.compositesa.2007.08.002

* Corresponding author. Address: Polymer Centre, Faculty of Technol-ogy, Tomas Bata University in Zlin, nam. T.G.M. 275, Zlin 762 72, CzechRepublic.

E-mail address: [email protected] (N.E. Kazantseva).

www.elsevier.com/locate/compositesa

Available online at www.sciencedirect.com

Composites: Part A 38 (2007) 2471–2485


but there are remarkable differences in the magnetic anddielectric spectra of composites with different types of CIin the RF band.

Thus, it is required to investigate the effect of the chem-ical composition and morphology (microstructure, shape,and the dispersity) of CI particles on the correlationbetween the structure and the electromagnetic propertiesof composites. Moreover, due to the nonuniform disper-sion of CI in a polymer matrix, this research is of keyimportance for the design of EWAs. Carbonyl iron parti-cles are characterized by spherical shape and a granularsurface structure [18]; these particles have a tendency toaggregate, thus showing poor dispersion behavior whenintroduced into various polymer matrices. The tendencyof aggregation is associated with the chemical inhomogene-ity and the surface roughness. To improve the dispersion ofCI particles in polymers, it is required to modify the sur-faces of individual particles or to use special processes forcomposite preparation [19]. Commercially available CIpowders are modified in order to isolate the particles andthus to increase the resistivity and the Q factor of the mag-netic cores produced from these CI powders [2–4]. To pro-duce such modified CI particles, industry uses differenttypes of agents in different cases, for example, hydrogensulfide, metal silicates, and a mixture of chromic and phos-phoric acids. As a result, a sulfide, silicate, or phosphatefilm with a thickness of about 0.01 lm is formed on the sur-face of a particle. However, these surface films do not oftenimprove the dispersion properties of particles in a polymermedium; moreover, they show a tendency to interact withthe components of the polymer mixture and hamper theconsolidation of composites. Therefore, while choosing apolymer matrix, we took into account the chemical struc-ture of the CI surface and the parameters of the solidifica-tion process: temperature, time, and the vacuum level.

In our investigations, composites with different types ofCI (primary or processed) have been obtained. These com-posites are characterized by the frequency dispersion e*(f)and l*(f) in the frequency range from 1 MHz to 10 GHz

and are discussed from the viewpoint of correlationbetween the electromagnetic properties of composites andthe morphology of CI.

2. Experimental

2.1. Carbonyl iron powders

In this study we used the following types of carbonyliron powders: SL, ES, and HQ from BASF (Germany);R-20 from SINTEZ (Russia); and MCI (mechanicallymilled Primary CI powder with onionlike structure) fromthe State Research Institute of Chemistry and Technologyof Organic Compounds (GNIICHTEOS, Moscow, Rus-sia). The main characteristics of the CI powders summa-rized in Table 1 were taken from the ProductInformation and come from our investigations. All car-bonyl iron powders were classified according to their man-ufacturing conditions: primary CI (ES, HQ, R-20 – hardgrades); processed CI, such as reduced with hydrogen (SL– soft grade); and mechanically milled CI (MCI – specialgrade). The mechanical treatment of primary carbonyl ironpowders obtained by the thermal decomposition of ironpentacarbonyl is performed through the high-energy ballmilling in the presence of aqueous–alcoholic mixture andsurface-active agent, while by the physico-chemical treat-ment of primary carbonyl iron powders is meant a specialheat treatment in hydrogen steam [2,20,21].

An X-ray diffraction analysis of all CI powder was car-ried out by a DRON – 2 X-ray diffractometer (Russia) withFe Ka radiation at a voltage of 30 kV, a tube current of18 mA, a counter rotation speed of 4 �/min, and the tra-verse speed of a chart recorder equal to 2400 mm/h. Therange of diffraction angles 2h varied from 5� to 155�. Toincrease the sensitivity of the instrument, the Soller slitswere removed. Pure iron was used as a reference material.The particle morphology of CI powders was observed by aJEOL JEM-200CX scanning electron microscope and byZeiss Axio Imager Z1m optical microscope.

Table 1Characteristics of CI powders and pressed composites

Type Hard grades Soft grade Special grade

ES HQ R-20 SL MCI

Chemical composition at% Iron >97.5 97.29 >97.7 >99.5 >97.7Carbon <0.9 0.95 <1.1 <0.05 <0.8Oxygen <0.5 0.9 <0.4 <0.2 <0.8Nitrogen <0.9 0.86 <1.1 <0.01 <0.7

Particle morphology Spherical Spherical Spherical Spherical FlakyMicrostructure Onionlike Onionlike Onionlike Polycrystalline Partly disrupted onionParticle size distribution d10 0.5 lm – 1.4–1.9 lm – –

d50 1.1 lm 2.5 lm 3.0–4.0 lm 9 lm –d90 2.2 lm – 6.0–8.0 lm – 1 lm

True density, g cm�3 �7.8 �7.8 �7.8 �7.8 �7.8Composite density (50 vol.%, g cm�3)* 4.42 4.20 4.43 4.52 4.33Initial permeability* �11 �8.5 �10 �21 �11.5

* Data obtained by our investigations. Initial permeability measured on pressed CI powders with PVA (12 vol.%) at the pressure 1.5 GPa.

2472 M.A. Abshinova et al. / Composites: Part A 38 (2007) 2471–2485


2.2. Polymers

Silicone elastomer (SYLGARD 184, Dow Corning,USA) and polyurethane (AXSON UR 3420, Axson,France) were chosen as a polymer matrix material accord-ing to their mechanical properties and good compatibilitywith CI powder. Silicone elastomer has durometer hard-ness of 40 Shore, tensile strength of 6.2 MPa, and elonga-tion at break of 100%, while polyurethane has hardnessof 50 Shore, tensile strength of 3.5 MPa, and elongationat break of more than 1000%.

2.3. Polymer–carbonyl iron composites

Polymer-filled composites were prepared by two meth-ods: by pressing and by molding.

Bulk samples of SL type of CI and polymer compositesof polyvinyl alcohol (PVA, 12 vol.%) and CI were preparedby pressing with the use of toroidal press forms (compac-tion pressure – 1.5 GPa). Excellent compaction propertiesof the CI powders reduced with hydrogen (SL) make it pos-sible to press them without a binding agent.

CI-containing composites of silicone elastomer or poly-urethane were prepared by the molding process. The poly-

mer mixture was placed between two metallic platesseparated by a 2 mm gap. The samples containing polyure-thane were kept at 80 �C for 4 h in a vacuum chamber,while the samples containing silicone elastomer were keptbetween 65 and 100 �C for 4–1 h, depending on the typeand the concentration of CI.

In this way, composites of CI (10–52 vol.%) were pre-pared. The volume fraction of CI in a composite was calcu-lated using the density of the CI powder (qf � 7.8 g cm�3)and the density of the silicon elastomer and the polyure-thane matrix (qm � 1.02 g cm�3). Depending on the fillerconcentration, the density of composites varied from1.7 g cm�3 (for 10 vol.%) to 4.55 g cm�3 (for 52 vol.%).

2.4. Measurement of the complex permittivity and

permeability

Circular samples of outer diameter of 15 mm and athickness of 1 mm were used for investigating the complexpermittivity e*(f); toroidal samples of outer diameter of8 mm, inner diameter of 3.1 mm, and a thickness of 2.5–3 mm were used for measuring the complex permeabilityl*(f) in the range from 1 MHz to 3 GHz by an RF Imped-ance/Material Analyzer (Agilent E49991A). Cylindrical

Fig. 1. SEM photographs of the primary (a), reduced with hydrogen (b), and mechanically milled (c, d) carbonyl iron powders in different magnifications:(a) R-20, (b) Sl, (c) and (d) MCI.

M.A. Abshinova et al. / Composites: Part A 38 (2007) 2471–2485 2473


Fig. 2. Reflected light microscope image of mechanically milled carbonyl iron MCI.

Fig. 3. SEM (left) and BSEM (right) photographs of the composites at 50% volume fraction of CI: (a) ES and silicone elastomer, (b) SL and siliconeelastomer.



samples of polymer–carbonyl iron composites with anouter diameter of 1 mm and length of 15 cm were usedfor investigating l*(f) and e*(f) by the resonant cavitymethod in the range from 2 GHz to 10 GHz [22,23].

3. Results

3.1. Scanning electron and optical microscopies

3.1.1. CI powders

The CI particles used in this study are reasonably spher-ical and have different diameters (Fig. 1a and b). Moreover,a considerable part of them forms irregularly shapedagglomerates, which are of different size as well. However,after mechanical treatment of CI powders, irregularlyshaped agglomerates were almost all de-aggregated anddeformed into a flake shape, the thickness of each flakebeing below 1 lm (Fig. 1c and d). The polyhedral surfacestructure of flakes is well visible in brightfield reflected lightmicroscopy micrograph (Fig. 2).

3.1.2. Polymer composites

SEM and BSEM microphotographs of composites filledwith different types of CI (Figs. 3 and 4) show a uniformdispersion of filler particles in the polymer matrix on a

micrometer scale. The bright dots seen in the BSEM micro-graphs correspond to CI particles.

3.2. X-ray diffraction

X-ray diffraction study was carried out in two steps:phase analysis and structural analysis.

The phase analysis of CI powders did not show any vis-ible difference in their composition. The phase compositionwas characterized by the pattern location (the angle ofreflection h and the lattice spacing) and the intensity ofthe diffraction patterns of samples. All these parameterswere compared with those of the reference material.

The structural profile analysis of diffraction patternsshowed a manifest difference in the intensity and the half-width of the diffraction peaks of CI powder samples inthe Æ110æ and Æ200æ directions, which correspond to thebasic reflection of the space-centered cubic lattice of a-iron(Fig. 5). Fig. 5a shows that, while there is a small differencein the intensities of the Æ110æ and Æ200æ peaks of the ES,HQ, R-20, and MCI samples, the SL samples show higherintensity of the Æ110æ peak and a twice higher intensity ofthe Æ20 0æ peak. This fact may be attributed to the highdegree of crystallite alignment and to the magnetic textureof the a-iron phase in the Æh00æ direction [24]. On the other

Fig. 4. SEM (left) and BSEM (right) photographs of the composites at 50% volume fraction of CI: (a) R-20 and polyurethane, (b) MCI and siliconeelastomer.



hand, Fig. 5b shows that the Æ110æ and Æ200æ peaks of theSL samples are very narrow, while, in other cases, they are

much broader, and an MCI sample shows a high value ofthe half-width of the diffraction line in the Æ200æ direction.The broader Æ200æ peak in MCI is associated with theenhancement of structural defects and microstresses (latticedistortion of a-iron) in CI structure after its mechanicaltreatment [25].

3.3. Complex permeability of CI and CI composites

The complex magnetic permeability of bulk samples andpolymer composites with high concentration of CI (morethan 40 vol.%) is a function of frequency. The magneticdispersion of bulk samples occupies a frequency intervalfrom 106 to 109 Hz (Fig. 6a), whereas highly filled polymercomposites are characterized by a higher frequency regionof dispersion, from 107 to 3 · 109 Hz (Figs. 6–10). Figs. 6–10 also show that, when the concentration of CI in com-posites decreases, the permeability of the compositesbecomes independent of frequency.

It should be mentioned that the behavior of the mag-netic spectra and the absolute values of the permeabilityjl*j of composites with the same concentration of CIdepend on whether the CI primary or processed.

Pressed composites with primary CI (onionlike struc-ture) are characterized by a broad plateau (106–109 Hz)

Fig. 5. X-ray diffraction patterns of different CI powders: (a) [110] in theangular range 27–31�, (b) [200] in the angular range 42–44�.

a

b

106 107 108 109 1010

0

10

20

30

40

50

μ,,

μ,

Com

plex

per

mea

bilit

y

Frequency, Hz

107 108 109 1010-5

0

5

10

15

20

25

SL

HQR-20ES

MCI

Perm

eabi

lity,

real

par

t

Frequency, Hz107 108 109 1010

0

2

4

6

8

10

12

SL

R-20

ES

MCI

HQ

Perm

eabi

lity,

imag

inar

y pa

rt

Frequency, Hz

Fig. 6. Complex permeability spectra of the pressed samples: (a) SL powder pressed at 1.5 GPa; composites with CI and PVA (12 vol.%) pressed at1.5 GPa.



and the dispersion of the real part of complex permeabilityl 0(f) in a higher frequency range (109–3 · 109 Hz) (seeFig. 6). Composites based on processed CI (those with dis-rupted onion structure) are characterized by a narrow low-frequency plateau (106–108 Hz) but a broader range ofmagnetic dispersion (108–3 · 109 Hz), as well as by highervalues of jl 0j and jl00j (Fig. 6). Moreover, the magneticspectrum of SL filled composites has two regions of disper-sion of the permeability. As it follows from l 0(f), the firstregion lies within 108–109 Hz, while the second one lieswithin 109–3 · 109 Hz. The frequency dependence of theimaginary part of the complex permeability (i.e., l00(f))has a broader resonance (Fig. 6).

Highly filled composites obtained by the moldingmethod exhibit similar behavior of l*(f) (Figs. 7–10). How-ever, the composites based on MCI are characterized bygreater values of jl*j compared with other composites.The magnetic losses ðl00maxÞ of MCI composites attain a

maximal value of 10 for a concentration of 50–52 vol.%of MCI (Fig. 9), whereas the maximal values of the mag-netic losses of composites based on other types of CI donot exceed 5 (Fig. 10).

Composites with a high concentration of CI differ fromeach other in the frequency range of the magnetic disper-sion but do not differ much in the resonance frequencies,except for the composites based on SL (Fig. 10). The mag-netic spectra of SL-filled composites are characterized bytwo resonance frequencies, f1 � 2 · 108 Hz andf2 � 3.5 · 109 Hz, while the resonance frequencies of com-posites filled with other CI powders lie in the microwavefrequency band in the range from 1 GHz to 3 GHz.

Composites based on CI also show difference in thecharacter of the concentration dependence of l 0. Thus,the l 0(c) of CI with onionlike structure exhibits lineardependence, whereas the l 0(c) of CI with disrupted onionstructure is typically nonlinear (Fig. 11). The concentrationdependence of l 0 is approximated by the Lichtenekker

107 108 109 10100

1

2

3

4

5

6

7

8

9

10

10 vol%

20 vol%

30 vol%

40 vol%

50 vol%

52 vol%

Per

mea

bili

ty, r

eal p

art

Frequency, Hz

107 108 109 1010-1

0

1

2

3

4

5

10 vol%

20 vol%

30 vol%

40 vol%

50 vol%

52 vol%

Per

mea

bili

ty, i

mag

inar

y p

art

Frequency, Hz

Fig. 7. Complex permeability spectra of composites based on siliconeelastomer for different contents of SL (reduced with hydrogen CI havingpolycrystalline microstructure).

107 108 109 1010

1

2

3

4

5

6

7

8

10 vol%

20 vol%

30 vol%

40 vol%

50 vol%

52 vol%

Per

mea

bili

ty, r

eal p

art

Frequency, Hz

107 108 109 1010-1

0

1

2

3

4

10 vol%

20 vol%

30 vol%

40 vol%

50 vol%

52 vol%P

erm

eab

ility

, im

agin

ary

par

t

Frequency, Hz

Fig. 8. Complex permeability spectra of composites based on siliconeelastomer for different contents of ES (primary CI) having onionlikestructure. The data scatter is given for composite with 40 vol.% of ES.



equation, which well fits the experimental data (Fig. 11a),while variants of the effective-medium theory (Maxwell-Garnet, Clausius-Mossotti and etc.) are incorrect whenthe volume concentration of CI is over 15% (Fig. 11b). Thisis most probably associated with the fact that the magneticspectra of composites based on CI are mainly determinedby the inhomogeneity of the magnetic filler structure ratherthan by the inhomogeneity of the polymer composite. Infact, the variants of the effective-medium theory take intoconsideration the interaction between magnetic particlesinside the composite, whereas the Lichtenekker equation[26] empirically defines the relation between the effectivevalues of the permeability of a composite and the perme-ability of a pure ferromagnetic: lc ¼ lp

m, where lc is thepermeability of a composite, lm is the permeability of a fer-romagnetic, p is the loading factor (p = Vm/Vc), Vm is thevolume of the magnetic phase, and Vc is the compositevolume.

3.4. Complex permittivity of polymer composites with CI

Figs. 12–15 show the frequency dependence of e* forcomposites with different types of CI and various concen-trations of CI in the polymer matrix. Highly filled compos-ites can be classified into two groups according to thecharacter of the e*(f). The first group of composites, i.e.,those based on SL, R-20, and MCI, are characterized bymanifest frequency dispersion of e* (Fig. 15) and nonlinearbehavior of e 0 and e00 as a function of the CI concentration(Fig. 16). The second group of composites, i.e., those filledwith HQ and ES types of CI (Fig. 15), exhibit no dispersionof e* and linear dependence of e 0 and e00 on the CI concen-tration (Fig. 16). As the concentration decreases, e*

becomes independent of frequency.There is a significant difference between the absolute val-

ues of e 0 and e00 in the frequency range from 106 to 107 Hz.In this frequency range, the values of je 0j and je00j of high-concentration composites with SL and MCI types of CI

107 108 109 1010

0

1

2

3

4

5

6

7

8

9

10

11

10 vol%

20 vol%

30 vol%

40 vol%

50 vol%

52 vol%

Per

mea

bili

ty, r

eal p

art

Frequency, Hz

107

108

109

1010

-1

0

1

2

3

4

5

6

7

8

9

10

10 vol%

20 vol%

30 vol%

40 vol%

50 vol%

52 vol%

Per

mea

bili

ty, i

mag

inar

y p

art

Frequency, Hz

Fig. 9. Complex permeability spectra of composites based on siliconeelastomer for different contents of MCI (machined milled CI withdisrupted onion structure).

107

108

109

1010

0

1

2

3

4

5

6

7

8

9

10

11

HQ

R-20

ES

SL

MCI

Per

mea

bili

ty, r

eal p

art

Frequency, Hz

107 108 109 1010-1

0

1

2

3

4

5

6

7

8

9

10

11

HQ ESSL

R-20

MCIP

erm

eab

ility

, im

agin

ary

par

t

Frequency, Hz

Fig. 10. Comparative analysis of the complex permeability spectra ofcomposites with different types of CI and the same content of CI 50 vol.%.



are almost twofold greater than those for other composites.In the microwave range, the e* of the composites is mostlyindependent of frequency and does not differ too much inthe absolute value for different samples.

4. Discussion

Before discussing the electromagnetic properties of poly-mer composites with CI we will dwell on the characteristicfeatures of the magnetic spectra of the composites.

The magnetization changes of any ferromagnetic in RFband are due to several types of magnetization mecha-nisms: domain-wall motion, magnetization rotation, andgyromagnetic spin rotation (a natural resonance) [24,27,28]. All these magnetization processes affect the l*(f) of aferromagnetic medium. Generally, the domain-wall reso-nance appears in the lower frequency RF band, whereas

the magnetization rotation and the natural resonance areresponsible for the magnetic dispersion in the higher fre-quency microwave band. Frequently, because of the inter-play between the above-mentioned mechanisms, it isimpossible to distinguish between the contributions ofthese processes to the permeability and to determine thedispersion region with a high degree of accuracy. More-over, composites materials in witch magnetic particles areembedded in insulating binders introduce additional disor-der in the form of filler–matrix interfaces. This makes themagnetic relaxation behavior of composite materials evenmore complex [29,30].

The first and the principal feature of the magnetic spec-tra of composite materials is a shift of the dispersion regionof permeability to higher frequencies compared with a bulkferromagnetic (Figs. 6 and 10).

The shift of the dispersion region of l* in composites isassociated with several physical mechanisms, first ofall, with the magnetic polarization of magnetic particles

0.0 0.1 0.2 0.3 0.4 0.51

2

3

4

5

6

7

8

9

10

11

HQ ES R-20 SL MCI

Co

mp

lex

per

mea

bili

ty, r

eal p

art

Loading coefficient

0.0 0.2 0.4 0.6 0.8 1.00

10

20

30

40

50

60

70

experimental points Clausius-Mossotti

Maxwell-Garnet

Lichteneker

Co

mpl

ex p

erm

eab

ility

, rea

l par

t

Loading coefficient

Fig. 11. Concentration dependences of complex permeability (real part) ofcomposites with different types of CI. The experimental results are givenfor 100 MHz and are approximated by the Lichtenekker equation (top)and in the case of SL-filled composites additionally by variants of theeffective-medium theory (bottom).

107

108

109

1010

0

10

20

30

40

50

60

70

Per

mit

tivi

ty, r

eal p

art

10 vol%

20 vol%

30 vol%

40 vol%

50 vol%

52 vol%

Frequency, Hz

107

108

109

1010

0.1

1

10

Per

mit

tivi

ty, i

mag

inar

y p

art

10 vol%

20 vol%

30 vol%

40 vol%

52 vol%

50 vol%

40

Frequency, Hz

Fig. 12. Complex permittivity spectra of composites based on siliconeelastomer for different contents SL.



isolated by a nonmagnetic polymer layer. The interphasepolarization and the formation of effective magneticcharges/dipoles are responsible for the nonuniform distri-bution of magnetization over the bulk of the material. Asa result, the effective magnetic field ð~HiÞ acting on a mag-netic particle decreases by the value of the demagnetizingfield ð~H dÞ, which is proportional to the demagnetizationfactor [27]:

~H i ¼ ~H e þ ~Hd ¼ ~H e � bN M~M ;

where ~H e is the strength of the magnetic component of theapplied electromagnetic field;~H i is the strength of the magnetic component of the

external field that acts on the magnetic composite;bN M is the tensor of demagnetization factors (a second-

rank material tensor that determines the relationshipbetween the magnetization vector and the vector of thedemagnetization field);

~M is magnetization.

Demagnetizing fields in a composite material are associ-ated with the morphology and the microstructural inhomo-geneity of magnetic particles, as well as with the structuralinhomogeneity of the composite material. While the micro-structural inhomogeneity of magnetic particles (distortionof chemical composition, inhomogeneous elastic stresses,pores, cracks, surface roughness, etc.) is ultimately respon-sible for the broadening of the ferromagnetic resonancebandwidth in a bulk ferromagnetic, the structural inhomo-geneity of composite magnetic materials is responsible forthe significant decrease of both components of l* com-pared with those in bulk homogeneous magnetic materialsdue to the violation of the ‘‘magnetic coupling’’ of mag-netic inclusions. The gaps/interphase boundaries (forexample, polymer layers) break the magnetic flux and giverise to local demagnetizing fields on the particle scale in thecase of a low-concentration magnetic component (distrib-uted magnetic charges on the surface of particles) and tothe external demagnetizing field on the sample scale when

107

108

109

1010

0

5

10

15

20

25

30

10 vol%

20 vol%

30 vol%

40 vol%

50 vol%

52 vol%

Per

mit

tivi

ty, r

eal p

art

Frequency, Hz

107

108

109

1010

0,00

0,25

0,50

0,75

10 vol%

20 vol%

30 vol%

40 vol%

50 vol%

52 vol%

Per

mitt

ivit

y, im

agin

ary

par

t

Frequency, Hz

Fig. 13. Complex permittivity spectra of composites based on siliconeelastomer for different contents of ES. The data scatter is given forcomposite with 40 vol.% of ES.

107

108

109

1010

1

10

100

10 vol%

20 vol%

30 vol%

40 vol%

50 vol%

52 vol%

Per

mit

tivi

ty, r

eal p

art

300

Frequency, Hz

107

108

109

1010

0.01

0.1

1

10

100

1000

10 vol%

20 vol%

30 vol%

40 vol%

50 vol%

52 vol%

Per

mit

tivi

ty, i

mag

inar

y p

art

Frequency, Hz

Fig. 14. Complex permittivity spectra of composites based on siliconeelastomer for different contents of MCI.



the concentration of the magnetic component is greaterthan the critical value (magnetic percolation threshold,Cl). For most polymer composites, the value of Cl is closeto 0.45, when the interaction between any two particlesgives rise to a cluster, thus forming a closed magnetic sys-tem. The obstruction of the interparticle interaction is themain reason for the reduction of jl*j in magnetic compos-ites compared with that in bulk magnetic materials. Thisprocess is essential for ferrite filled composites, where theabsolute values of the complex permeability decrease hun-dreds of times compared with those in ferrite ceramics [30],whereas, in the case of CI filled composites, the absolutevalue of the complex permeability decreases approximatelyby a factor of 10 (Fig. 5). This fact is associated with themicrostructure of a CI particle and the nonuniform magne-tization of the particle caused by its microstructure.

The effect of the above-listed factors on the l*(f) of CI

composites is explained as follows:In the present study, we consider three types of primary

CI composites that consist of spherical CI particles with a

small dispersion in size and onionlike internal structure(ES, HQ, and R-20). According to the microstructuralstudies of primary CI given in the literature [2–4,17,31]and obtained in our investigations, the a-iron phase repre-sents an aggregate of anisotropic a-iron crystallites with amean diameter of about 100 A; the amorphous phase ofcementite, which occupies about 30 vol.% of the composite,consists of carbides and nitrides of iron; the mean thicknessof the amorphous layers is about 25 A. A particle of 7 lmin size contains from 7 to 9 layers (Fig. 17a). Therefore, aCI particle can be considered as heterogeneous medium inwhich nanosized ferromagnetic blocks of a-iron are sepa-rated by dielectric nanolayers of nonmagnetic cementite.The nonmagnetic interphase boundaries break the mag-netic coupling of a-iron blocks and thus lead to the nonuni-form distribution of magnetization over the particle. Theabsolute values of the permeability and permittivity incomposites with onion structure CI particles do not differmuch. The magnetic spectra are characterized by constant

107

108

109

1010

20

40

60

80

100

HQ

ES

MCI

R-20

SL

Per

mit

tivi

ty, r

eal p

art

Frequency, Hz

107

108

109

1010

0.1

1

10

100

R-20

MCI

HQ

ES

SL

Per

mit

tivi

ty, i

mag

inar

y p

art

Frequency, Hz

Fig. 15. Comparative analysis of the complex permittivity spectra ofcomposites with different types of CI and the same content of CI 50 vol.%.

0.0 0.1 0.2 0.3 0.4 0.5

0

10

20

30

40

50

60

70 HQ MCI R-20 SL ES

Per

mit

tivi

ty, r

eal p

art

Loading factor

0.0 0.1 0.2 0.3 0.4 0.5

0

5

10

15

20

25

MCI

HQ

SL

R-20

ES

Per

mit

tivi

ty, i

mag

inar

y p

art

Loading factor

Fig. 16. Concentration dependences of the complex permittivity ofcomposites with different types of CI. The real and imaginary parts ofthe complex permittivity are given at 100 MHz.



values of l 0 in the lower frequency RF band and by mag-netic dispersion in the microwave band (Fig. 10). We eval-uated the frequency dependence of l* by the frequencydispersion formula introduced by Tsutaoka [29], whichdescribes a magnetic spectrum as a superposition of tworesonance magnetization mechanisms: a resonance ofdomain wall motion and a natural resonance (gyromag-netic spin rotation). We applied a nonlinear least-squarefitting method to decompose the pilot complex permeabil-ity spectra. We have discovered that the contribution ofgyromagnetic spin rotation is dominant,1 while the contri-bution of domain wall motion is negligible (Fig. 18). Thisresult is associated with the fact that the domain wall dis-placement is suppressed by dielectric layers of cementiteinside the multilayered onionlike structure of primary CIpowders. The frequency dependence of e* is also deter-mined by the onion structure: there is no frequency disper-sion of e* in a highly filled composite; moreover, theabsolute values of permettivity in the composites withonion structure are lower than those in the compositesbased on disrupted CI powders (Fig. 15).

The disrupted microstructure of the particles thatappears after their physico-chemical treatment (reductionwith hydrogen) or mechanical grinding leads to significantchanges in the electromagnetic properties of composites onbased on these types of CI, namely, to the broadening ofthe magnetic dispersion region and the increase of jl*jand je*j.

The magnetic spectra of SL filled composites have twodispersion regions (Figs. 6, 7, and 10), which point to thepresence of two resonance phenomena and two resonancefrequencies, respectively. As already mentioned, the lowerfrequency magnetic dispersion region is attributed to thenonreversible domain wall displacement, while the higherfrequency magnetic dispersion region is attributed to themagnetization rotation and the natural resonance. Forexample, the analysis of l*(f) of SL filled composites bythe frequency dispersion formula of [29] has shown the sig-nificant role of the domain wall motion (Fig. 19). The for-mation of a polycrystalline structure of SL particles, whichis revealed by X-ray diffraction analyses and well visible onthe microsection view of pressed CI reduced with hydrogen(Fig. 17c), increases the contribution of the domain-walldisplacement, which is suppressed in the onionlike struc-ture of primary CI (Fig. 17a).

As to the magnetization rotation and spin rotationcomponents, it is also higher in SL filled composites com-pared with that in composites based on primary CI. Thecrystal texture of the a-iron phase with the Æh00æ direc-tion, which is determined from the intensities of the

Fig. 17. View of particle microstructure of primary and processed CI: (a) onionlike structure of primary CI [Product information BASF, July 2002]; (b)disrupted onion structure of mechanically milled CI [2], enlargement 1:750; (c) polycrystalline structure of reduced type CI at 700 �C and at 1000 �C [2],enlargement 1:750.

1 Since it is difficult to separate the contribution of magnetizationrotation and natural resonance because of the interplay of thesemechanisms in high-frequency region, the frequency dispersion formula[29] containing two components owing to domain wall motion andgyromagnetic spin rotation. Nevertheless, the higher-frequency resonancelinewidth is attributed to superposition of magnetization rotation and thenatural resonance [24,27,28].



Æ110æ and Æ20 0æ peaks on the X-ray diffraction spectrumof SL samples (Fig. 5), plays a key role here. It is knownthat this type of crystal texture in ferromagnetics, calledmagnetic texture, reduces the magnetoelastic energy andthereby enhances the magnetic properties, as, for example,in the case of crystallographic texturized transformer iron[27].

Note that the l00(f) of SL composites exhibits an anom-alous behavior in the region of magnetic dispersion,namely, a broad resonance with a linewidth of 108–3 · 109 Hz (Figs. 6, 7, and 10). Such behavior of l00(f) isassociated with several important factors: (i) the disruptionof cementite layers, which suppresses the growth of a-irondomains, and (ii) the considerable reduction of the concen-tration of carbon and nitrogen in crystalline a-iron lattice,which leads to the suppression of magnetostriction andstructural relaxation phenomena and therefore increasesthe region of magnetic losses. The wide region of magnetic

losses also results from the large size dispersion, shapeanisotropy, and the multidomain structure of a-ironcrystallites.

The large content of a-iron and the polycrystalline struc-ture of a particle is also responsible for the increase of thereal and imaginary parts of the complex permittivity ofcomposites based on SL.

Remarkable changes in the frequency dependence of thecomplex permeability and permittivity of composites arealso observed after mechanical treatment of primary CI(Figs. 10 and 15). The complex permeability l*(f) of com-posites based on R-20 is of relaxation type, while that ofcomposites based on MCI (which is obtained from R-20by mechanical grinding) is of resonance type, with tworegions of dispersion. Compared with R-20-filled compos-ites, MCI-filled composites show a significant increase injl 0j in the wide range of frequencies. Moreover, the maxi-mum of magnetic losses are two times greater than that

106 107 108 109 1010

1

2

3

4

5

6

7

8

domain wall+spin

experimental

spin

domain wall

Co

mp

lex

per

mea

bili

ty, r

eal p

art

Frequency, Hz

106 107 108 10 9 10 10

0.0

0.5

1.0

1.5

2.0

2.5

3.0spin

domain wall+spin

experimental domain wall

Co

mp

lex

per

mea

bili

ty, i

mag

ian

ary

par

t

Frequency, Hz

Fig. 18. Complex permeability spectra of composite filled with 50 vol.% ofES. Solid, dotted, and dashed lines are calculation curves for domain-walland spin components of complex permeability.

106 107 108 109 10100

1

2

3

4

5

6

7

8

9

10

11

Co

mp

lex

per

mea

bili

ty, r

eal p

art

Frequency, Hz

domain wall

spin

domain wall+spin

experimental

106 107 108 109 1010

0

1

2

3

4

5

spin domain wall

experimental

domain wall+spin

Com

ple

x p

erm

eab

ility

, im

agin

ary

par

t

Frequency, Hz

Fig. 19. Complex permeability spectra composite filled with 50 vol.% ofSL. Solid, dotted, and dashed lines are calculation curves for domain-walland spin components of complex permeability.



in composites with other types of CI. Fig. 20 illustrates theanalysis of the complex permeability spectra of MCI-filledcomposites carried out by the Tsutaoka method. It can beseen the dominant role of the domain wall motion. Theconcentration dependence of l 0 in composites based onR-20 is linear, while that for MCI-based composites is non-linear. Just as the absolute value of complex permeability,the absolute value of complex permittivity sharplyincreases (Fig. 15).

The observed changes in l* and e* are also attributed tothe structural changes in CI particles after mechanicaltreatment. Firstly, the mechanical treatment leads to thede-aggregation and breaks spherical particles into frag-ments of flaky shape with well visible polyhedral surfacestructure (Fig. 2). The de-aggregation increases the fillingability of CI and thus provides denser packing of particlesin a polymer matrix (Fig. 4a). The flaky structure of parti-cles reduces the demagnetization factor. Finally, this

increases the permeability of MCI-filled composites(Fig. 10). Secondly, the X-ray analysis shows the increaseof microstresses in MCI particles, which is connected withthe formation of boundaries with high content of defects(Fig. 5b). The photograph of milled CI particles revealspartly disruption cementite layers (Fig. 17b). The X-rayphotoelectron microscopy of the surface of a CI particlebefore and after its mechanical treatment also shows thedisruption of cementite layers in the MCI particle struc-ture. Therefore, before mechanical treatment the surfaceof a particle is rich in iron carbides and iron nitrides, while,after the treatment, it is rich in a-iron. The disruption ofdielectric cementite layers intensifies the domain wall dis-placement and, moreover, increases eddy-current losses,because the resistivity decreases due to the appearance ofconducting paths between a-iron blocks. The latter factoris responsible for the significant increase of the complexpermittivity of composites filled with MCI in the wholeRF band (Fig. 15).

5. Conclusions

We have presented the results of a comparative analysisof the electromagnetic parameters of polymer compositesbased on CI powders. These parameters have been mea-sured in the range of frequencies from 1 MHz to 10 GHzby two different methods, viz., the impedance methodand the resonant cavity method. We have established thatthe changes in the microstructure and the shape of CI par-ticles caused by their physico-chemical or mechanical treat-ment have a crucial effect on the absolute values and thecharacter of the frequency dependence of the complex per-meability and permittivity of CI composites. The mechan-ical treatment de-aggregates and breaks spherical particlesinto the fragments of flaky structure, partly disrupts theamorphous layers of cementite in the onion structure,and thus increases internal microstresses compared withthose in primary CI. As a result, the de-aggregation leadsto a denser packing of particles in a polymer matrix, andthe flaky shape reduces the demagnetization factor andthereby increases the absolute value of the permeability.The disruption of dielectric cementite layers intensifiesthe domain wall displacement and increases eddy-currentlosses, which also leads to the increase of permeability.Owing to the same features, the permittivity of MCI com-posites is an order of magnitude greater than that of pri-mary CI composites. The chemical treatment of primaryCI with hydrogen also disrupts the cementite layers in themicrostructure of particle, decreases the carbon and nitro-gen concentrations, and allows a-iron domains to grow andform a polycrystalline structure with magnetic texture of a-iron phase in the Æh00æ direction. Such a particle micro-structure leads to the intensification of domain-wall motionand increases the magnetization rotation and spin rotationmechanisms. On the whole, this broadens the resonancelinewidth and considerably increases the absolute valuesof l* and e* of composites.

106 107 108 109 1010

0

2

4

6

8

10

12

domain wall

spin

experimental

domain wall+spin

Co

mp

lex

per

mea

bili

ty, r

eal p

art

Frequency, Hz

106 107 108 109 1010

0

2

4

6

8

10

domain wall

domain wall+spin

experimental

spin

Co

mp

lex

per

mea

bili

ty, i

mag

inar

y p

art

Frequency, Hz

Fig. 20. Complex permeability spectra composite filled with 50 vol.% ofMCI. Solid, dotted, and dashed lines are calculation curves for domain-wall and spin components of complex permeability.



From the practical point of view, the variation of thetype of CI in a composite, namely, the microstructureand the shape of particles, makes it possible to controlthe absolute values of its electromagnetic parameters andto monitor the frequency dependence of l* and e*.

Acknowledgements

The financial support of the Ministry of Education,Youth and Sports of the Czech Republic (ME 883 KON-TAKT) and the Russian Foundation for Basic Research(Project No. 06-08-00145) is gratefully acknowledged. Weare grateful to Dr. Gerald Lippert from the BASF Corpo-ration and Dr. Nabanita Saha from Tomas Bata Univer-sity for fruitful discussion.

References

[1] Tailor A, Richards A. Soft magnetic materials for telecommunica-tions. Press. L.; 1953.

[2] Volkov VS, Syrkin VG, Tolmasskii IS. Carbonyl iron. Mos-cow: Metallurgy; 1969 [in Russian].

[3] Syrkin VG, Romanova LN, Maslov NM, Sazonov AI, Romanovs-kaya OR. Powder Metall Met Ceram 1973;12:942.

[4] Syrkin VG, Tolmasskii IS. Powder Metall Met Ceram 1974;13:958.[5] Handbook of Powder Metal Technologies and Applications, vol. 7,

American Society of Metals; 1998.[6] Picos S. Latvian J Phys Tech Sci 1999;2:33.[7] Kazantseva NE, Tchmutin IA, Ryvkina NG. J Commun Technol

Electron 2003;48:173.[8] Johnson RN, Loret C. US Patent 20050139282; 2005.

[9] Zhang B, Feng Y, Xiong J, Yang Y, Lu H. IEEE Trans Magn2006;42:1778.

[10] Nie Y, He H, Zhao Z, Gong R, Yu H. JMMM 2006;306:125.[11] Feng Y-B, Qiu T, Shen C-Y, Li X-Y. IEEE Trans Magn 2006;42:363.[12] Naziripour GA, Patton CE, Kogekar MV. J Appl Phys 1988;63:3771.[13] Qiu RC, Liu SZ, Lin W. Acta Electron Sinica 1994;22:104.[14] Chung KH, Wu CS, Malawer EG. Thermochim Acta 1989;154:195.[15] Wang Y, Afsar MN. Microwave Opt Technol Lett 2004;42:458.[16] Heaney DF, Mueller TJ, Davies PA. Powder Metall 2004;47:1.[17] Lauriat J-P. J Non-Cryst Solids 1983;55:77.[18] Heim L, Farshshi M, Morgeneyer M, Schwedes J, Butt H-J, Kappl

M. J Adhesion Sci Technol 2005;19:199.[19] Wang Y, Xu J, Bechtel SE, Koelling KW. Rheol Acta 2006;45:919.[20] Syrkin VG, Tolmasskii IS. Powder Metall Met Ceram 1965;4:7.[21] Syrkin VG, Tolmasskii IS. Powder Metall Metal Ceram 1966;5:7.[22] YuN Kazantsev. In: Proceedings of the XIV international conference

on gyromagnetic electronics and electrodynamics (Microwave Fer-rites), vol. 2. Moscow, Russia; 1998. p. 205.

[23] Agilent, Basics of measuring the dielectric properties of materials.Application note, Agilent Technologies, Inc., USA, April 28, 2005.5989-2589 EN.

[24] Smith J, Wijn HPJ. Ferrites. Eindhoven: Philips Technical Library;1959.

[25] Novakova AA, Adladze OV, Kiseleva TYu. J Solid States2001;43:1443.

[26] Lichtenekker K. Phys Z S 1926;27:118.[27] Vonsovskii SV. Magnetism. Moscow: Nauka; 1971 [in Russian].[28] Gurevich AG, Melkov GA. Magnetization oscillation and

waves. CRC Press, Inc.; 1996.[29] Tsutaoka T. J Appl Phys 2003;93:2789.[30] Kazantseva NE, Vilcakova J, Kresalek V, Saha VP, Sapurina I,

Stejskal J. J Magn Magn Mater 2004;269:30.[31] Umanskii JS. X-ray diffraction analysis of metals. Moscow: Metal-

lurgy; 1967 [in Russian].


Correlation between the microstructure and the electromagnetic properties of carbonyl iron filled polymer composites

Documents