1-s2.0-S0022459611005172-main

download 1-s2.0-S0022459611005172-main

of 10

Transcript of 1-s2.0-S0022459611005172-main

  • 7/27/2019 1-s2.0-S0022459611005172-main

    1/10

    Oriented SrFe 12 O19 thin lms prepared by chemical solution deposition Josef Bur sk a ,n , Ivo Drbohlav b , Zden ek Frait b , Karel Kn z ek b , Radomr Kuz el c , Karel Kou ril d

    a Institute of Inorganic Chemistry of the AS CR, v.v.i., 250 68 Husinec- Rez 1001, Czech Republic b Institute of Physics of the AS CR, v.v.i., Na Slovance 2, 182 21 Praha 8, Czech Republic c Charles University in Prague Faculty of Mathematics and Physics, Ke Karlovu 5, 121 16 Praha 2, Czech Republic d Charles University in Prague, Faculty of Mathematics and Physics, V Hole sovic ka ch 2, 180 00 Praha 8, Czech Republic

    a r t i c l e i n f o

    Article history:

    Received 6 April 2011Received in revised form9 September 2011Accepted 14 September 2011Available online 22 September 2011

    Keywords:Hexagonal strontium ferriteChemical solution depositionEpitaxial thin lms

    a b s t r a c t

    Thin lms of SrFe 12 O19 (SrM) were prepared from a solution of iron and strontium alkoxides throughthe chemical solution deposition method on both amorphous (glassy SiO 2 ), and single crystal substrates(Si(100), Si(111), Ag(111), Al 2 O3 (001), MgO(111), MgAl 2 O4 (111), SrTiO 3 (111)) substrates. The process of crystallization was investigated by means of powder diffraction, atomic force microscopy and scanningelectron microscopy. Magnetization measurements, ferromagnetic and nuclear magnetic resonancewere used for evaluation of anisotropy in the lms. Whilst amorphous substrates enabled growth of randomly oriented SrM phase, use of single crystal substrates resulted in samples with different degreeof oriented growth. The most pronounced oriented growth was observed on SrTiO 3 (111). A detailedinspection revealed that growth of SrM phase starts through the breakup of initially continuous lminto isolated grains with expressive shape anisotropy and hexagonal habit. A continuous lm withepitaxial relations to the substrate was produced by repeating recoating and annealing.

    & 2011 Elsevier Inc. All rights reserved.

    1. Introduction

    Strontium hexagonal ferrites SrFe 12 O19 (SrM) have a magne-toplumbite structure with an easy magnetocrystalline anisotropyaxis along the c axis of the hexagonal unit cell. In bulk form theyare commonly used for permanent magnets, in electro-acousticdevices, and are of great interest in thin lm form for high densitymagnetic recording and millimeter-wave devices because of theirlarge chemical resistivity, high permeability at high frequencies,and large uniaxial anisotropy [13]. In majority of these applica-tions highly oriented lms with c -axis perpendicular to thesubstrate are desirable. The lms are in majority prepared bymethods relying on cost expensive apparatus, such as sputtering[4] , pulse laser deposition (PLD) [1,2,5], or liquid phase epitaxy(LPE) [6] . As a convenient alternative, chemical solution deposi-

    tion (CSD) method can be used, thanks to the facts that it issimple, low cost, and can be rapidly developed for the evaluationof a new material system, and is amenable with standardsemiconductor manufacturing methods. Using different types of organometallic precursors (alkoxides [7] , ethyhexanoates [8] , ornitrates [9] ) and depending on the type of substrate (SiO 2 glass[7] , polycrystalline alumina [8] , or single crystal Al 2 O3 (001) [9] )

    either randomly [7 ,8] or oriented [9] ferrite lms have beenprepared through deposition by dipping or spinning. In order toobtain higher crystallinity and more pronounced anisotropy, it isnecessary to do nal crystallization at temperatures above 900 1C,where most of the substrates already react with material of ferritelm giving rise to either secondary phases (Ba xSi yO z [7] ), orinterdiffusion occurs (BaAl xFe12 xO19 [5 ,7]). The aim of this workwas to develop a procedure for preparation of highly orientedmagnetically anisotropic hexagonal ferrite thin lms by means of CSD method.

    2. Materials and methods

    In house synthesized alkoxides Fe(OCHCH 2 (CH3 )2 )3 and Sr(OCH 2

    CH2 OCH3 )2 were used as metal precursors. Calculated amounts of Fe(OCHCH2 (CH3 )2 )3 and Sr(OCH 2 CH2 OCH3 )2 (with 5% surplus in Srto compensate its losses during annealing due to its reactivity withsubstrate) were dissolved in iso-butanol, mixed and heated forseveral hours to accomplish homogenization. Subsequently a sui-table amount of 2,2-diethanolamine (DEA) used as a modier(slowing down the reactivity and sensitivity towards air humidity)was added. All reactions and handling were done under dry nitrogenatmosphere to prevent reaction with air humidity and preliminaryformation of strontium carbonate in solution. The concentration of this stock solution was c 0.6 M (Fe), the modier molar ration wasn(DEA)/n(Fe) 1. For the preparation of the thin seed layer, thestock solution was diluted to 0.10.3 M.

    Contents lists available at SciVerse ScienceDirect

    journal homepage: www.elsevier.com/locate/jssc

    Journal of Solid State Chemistry

    0022-4596/$- see front matter & 2011 Elsevier Inc. All rights reserved.

    doi: 10.1016/j.jssc.2011.09.024

    n Corresponding author. Fax: 420 220941502.E-mail addresses: [email protected] (J. Bur sk), [email protected] (I. Drbohlav) ,

    [email protected] (Z. Frait) , [email protected] (K. Kn z ek) ,[email protected] (R. Kuz el) , [email protected] (K. Kou ril).

    Journal of Solid State Chemistry 184 (2011) 30853094

    http://www.elsevier.com/locate/jsschttp://www.elsevier.com/locate/jsschttp://localhost/var/www/apps/conversion/tmp/scratch_6/dx.doi.org/10.1016/j.jssc.2011.09.024mailto:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]://localhost/var/www/apps/conversion/tmp/scratch_6/dx.doi.org/10.1016/j.jssc.2011.09.024http://localhost/var/www/apps/conversion/tmp/scratch_6/dx.doi.org/10.1016/j.jssc.2011.09.024mailto:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]://localhost/var/www/apps/conversion/tmp/scratch_6/dx.doi.org/10.1016/j.jssc.2011.09.024http://www.elsevier.com/locate/jsschttp://www.elsevier.com/locate/jssc
  • 7/27/2019 1-s2.0-S0022459611005172-main

    2/10

    Polished SiO 2 glass, Si(111), Si(100), Ag(111), Al 2 O3 (001),MgO(111), MgAl 2 O4 (111), SrTiO 3 (100), and SrTiO 3 (111) wereused as substrates. They were cleaned by soaking in KMnO 4 H2 SO4 mixture for 20 min, washing with deionized water andacetone, and nally dried in the ow of puried air. As-receivedAg and MgO substrates were shortly washed with acetone andcleaned under ow of ltrated air just prior to deposition. The solswere ltered through 0.45 mm PTFE membrane lter unit just

    before the deposition. First of all, a thin lm of SrM (2040 nm of effective thickness) serving as a seed layer to promote thecrystallization of bulk layer was deposited by spin-coatingtechnique (RC8 Gyrset, KarlSuss) from diluted stock solutionand crystallized at different temperatures. Then stock solutionwas used for spin-coating of the bulk layer. After drying at110 1C for several minutes and pyrolysis of gel lms at 450 1C for15 min, crystallization annealing was done at temperatures in therange 5001100 1C for 6 h in conventional tube furnace withtemperature ramp 10 1C/min, under open air atmosphere. Thedepositionannealing cycle was repeated eight times to obtainthe desired lm thickness.

    Part of the starting precursor solution was converted intopowder by drying at 110 1C for 24 h in air and used for DTA/TGcoupled with mass spectrometry (Netzsch STA QMS 409), andhigh temperature XRD experiments (Bruker D8 powder diffract-ometer, MRI TC-wide range temperature chamber, Cu K a 1,2 radia-tion). The composition of solutions and thin lms was checked bythe volumetric titration and electron microprobe (Jeol JXA-733)analyses, respectively. Thickness of lms was measured by astylus apparatus (AlphaStep IQ KLA Tencor). The thickness of seed layer was calculated as a mean value gained on separatelyprepared sample composed of 10 crystallized seed layers. Thethickness of nal sample (seed layer bulk layer) typicallyreached values around 600 nm for samples crystallized at1100 1C. The thickness and composition were also evaluated fromthe SEM/EDAX analyses. SEM (Philips XL30 CP, Jeol JSM 6700 FNT) and AFM (Explorer, Thermomicroscopes, USA) analyses wereused to evaluate microstructure of thin lms. The room tempera-ture y 2y diffraction patterns of SrM lms were measured onPANalytical X Pert diffractometer (Cu K a 1,2 radiation, secondarygraphite monochromator, PIXcel detector). Texture characteriza-tion and pole gure measurements were done by PANalyticalXPert MRD diffractometer (Cu K a 1,2 radiation, polycapillary in theprimary beam, the Eulerian cradle, parallel-plate collimator andsecondary monochromator). Magnetization curves were mea-sured at room temperature on SQUID (MPMS-5 S QuantumDesign) apparatus. The microwave properties of SrM lms werestudied by ferrimagnetic resonance (FMR) measurements per-formed at room temperature, in reection mode with externaleld applied normally to the sample plane, at frequencies 49 and69 GHz using eld modulation (ac eld at 94 kHz) and lock-inamplier technique [10] . Solid state 57 Fe NMR spectra weremeasured in a zero static external magnetic eld using pulsecoherent spectrometer Bruker Avance (equipped with homemadetunable probe head at temperature 4.2 K, with modied CarrPurcellMeiboomGill (CPMG) pulse sequence applied). Theresulting frequency swept spectrum was evaluated as amplitudeof the Fourier transform at excitation frequency.

    3. Results

    DTA/TG/MS measurements were used for optimization of drying,pyrolysis, and annealing regime. Results obtained on SrM powderprepared from dried precursor sol are shown in Fig. 1. It wasobserved that after the dehydration at 120 1C thermal degradation

    starts at temperatures $ 250 1C with evolution of CO 2 and N xO y

    gases (detection of CO 2 and NO 2 fragments by MS) generating byburning out of metalloorganic precursor species and amine modier(broad exotherm centered at 330 1C), accompanied by decrease inmass. The pyrolysis is nished at 450 1C, and above this temperatureonly minor mass decrease is observed. Though DTA curve did notshow any endotherm at temperatures around 700 1C, two distin-guishable bumps on DTG and MS curves located at 640 and 700 1C,respectively, can be assigned to decomposition of remaining SrCO 3during the formation of SrM phase (compare with Fig. 2).

    From high temperature XRD measurements ( Fig. 2) i t i sevident that SrCO 3 formed already during drying at 110 1C,survives the pyrolysis and slowly disappears by reaction with

    iron precursor at temperatures slightly above 600 1C. The onset of

    100 200 300 400 500 600 700 800temperature (C)

    Fig. 1. DTA/TG/MS curves on dried SrFe 12 O19 sol measured in air.

    2 theta ()40 60 8020

    25

    1000800

    700

    300400500

    600

    platinum

    650

    SrCO 3

    SrFe 12 O19

    Fig. 2. High temperature powder diffraction patterns measured on powder from

    SrM precursor sol.

    J. Bur sk et al. / Journal of Solid State Chemistry 184 (2011) 308530943086

  • 7/27/2019 1-s2.0-S0022459611005172-main

    3/10

    crystallization of SrM phase was found at 650 1C, where traces of a -Fe 2 O3 , but not SrCO 3 , were observed. At temperatures higherthan 750 1C only pure ferrite phase was observed.

    Fig. 3 shows the powder XRD patterns and Table 1 summarizesvalues of lattice mist and experimental results about phasecomposition and degree of orientation of SrM thin lms depositedon selected substrates. They were selected to include bothamorphous and single crystal materials with both positive and

    negative mist values. According to these results use of amor-phous silica and silicon single crystals led to randomly orientedSrM phase for lower temperatures and to formation of SrSiO x anda -Fe 2 O3 phases due to enhanced reactivity of basic alkali earthwith acidic Si at annealing temperatures above 900 1C. Use of

    substrates with even more convenient values of mist MgO(111) and MgAl 2 O4 (111) induced problems due to highreactivity of Mg 2 with Fe 3 . It was observed in case of MgO(111) substrate that randomly oriented SrM phase is formedbelow 800 1C and oriented MgFe 2 O4 spinel phases were obtainedabove this annealing temperature. Use of MgAl 2 O4 (111) led tomoderately oriented SrM phase (see Fig. 3(e)) for samplesannealed at temperatures below 900 1C. In an effort to promote

    the orientation growth by rising further the annealing tempera-ture weakly oriented MgFe 2 O4 spinel phase started to form. Use of Ag single crystal (not shown here) for orientation growth led toonly randomly oriented SrM phase due to its low melting point(961 1C) and due to its relative softness making difcult for properpolishing of crystal surface. Al 2 O3 (001) single crystal, which isotherwise commonly used for preparation of oriented hexagonalferrites [1 ,2,9], led to only moderate orientation growth of SrMphase (not shown here) probably due to strong tendency of Al 3

    for diffusion into ferrite structure where it substitutes for Fe 3 inferrite structure.

    Diffraction pattern on SrM thin lms deposited on SrTiO 3 (100)substrate showed enhanced intensities of (00l) reections for SrMphase evidencing certain degree of orientation growth. Wellresolved reections with indexes h , k a 0 observed in patternimply presence of signicant portion of nonoriented grains. By farthe best orientation growth was observed with SrM deposited onSrTiO3 (111) single crystal substrate and results gained on thistype of substrate will be subject of following paragraphs.

    Fig. 4 shows the XRD pattern of SrM thin lms deposited onSrTiO3 (111) substrate and annealed at different temperatures. Itcan be seen that samples are amorphous for temperatures lowerthan 550 1C. Annealing at 600 1C induces crystallization of ran-domly oriented SrM phase, which becomes preferentially orientedwhen temperature is increased to 700 1C as it is apparent fromincreasing intensity of 00l reections. With further temperatureincrease up to 1100 1C this character of diffraction patternbecomes even more pronounced. Two of minority reectionscan be ascribed to SrM phase (reections 107 at 32.2 1 and 114at 34.3 1), there are no traces of the frequently noticed impurityphases, namely a -Fe 2 O3 and Sr xFe yO z .

    In order to semi-quantitatively assess the preferential growth,the so called o scans (rocking curves) were measured on 008reection planes of SrM phase and are shown in Fig. 5 togetherwith inset table containing their full width at half maximum(FWHM) values. According to these data the evolution of orientedgrowth starts at temperatures below 600 1C: layers annealed at500 1C show random angular dispersion of [00l] directions.Samples annealed at 600 1C exhibit already sharp rocking curve

    008

    20 30 40 50 602 theta (deg)

    006

    (a)

    (b)

    (c)

    (d)

    (e)

    (f)

    0012

    0014

    MgFe 2O 4 (hhh)

    substrate

    SrFe 12O 19

    Fig. 3. y2 y powder diffraction patterns of SrM thin lms deposited on differentsubstrates (a) Si(100), (b) Si(111), (c) amorphous SiO 2 , (d) MgO(111),(e) MgAl 2 O4 (111), and (f) SrTiO 3 (100) and annealed at 850 1C.

    Table 1Lattice mists, phase composition, and orientation of SrFe 12 O19 thin lms deposited on different substrates.

    Substrate Mist (%) D1 or D2 a Reactivity substrate 2 SrM precursors Phase composition of thin lm

    Si(111) 23.4 ( D1 ) t Z 900 1C: t r 900 1C: random oriented SrMSi(100) Sr - SrSiO x t Z 900 1C: SrM a -Fe 2 O3SiO2 amorph. t Z 900 1C: t r 900 1C: random oriented SrM

    Sr - SrSiO x t Z 900 1C: SrM a -Fe 2 O3MgO(111) 1.35 ( D2 ) t Z 800 1C: t r 800 1C: random oriented SrM

    Fe - MgFe 2 O4 t Z 800 1C: (111) oriented MgFe 2 O4MgAl2 O4 (111) 2.7 ( D2 ) t Z 900 1C: t r 900 1C: (00l) oriented SrM

    Fe - MgFe 2 O4 t Z 900 1C: weakly (111) orient. MgFe 2 O4Ag(111) 1.8 ( D2 ) m.p. (Ag) 961 1C weakly (00l) oriented SrM, problems with polishing

    of single crystal surfaceAl2 O3 (00l) 6.0 ( D1 ) t Z 900 1C: moderately (00l) oriented SrM

    Al- Sr(AlFe) 12 O19SrTiO3 (100) 6.5 ( D2 ) t r 1200 1C:no reaction t Z 900 1C: (00l) oriented SrMSrTiO3 (111) 6.5 ( D2 ) t r 1200 1C:no reaction t Z 700 1C: strongly (00l) oriented SrM

    a

    D1 (a(substr) a(SrM)/a(substr)) 100, D2 (O2a(substr) a(SrM)/ O2a)substr)) 100

    J. Bur sk et al. / Journal of Solid State Chemistry 184 (2011) 30853094 3087

  • 7/27/2019 1-s2.0-S0022459611005172-main

    4/10

  • 7/27/2019 1-s2.0-S0022459611005172-main

    5/10

    hexagonal grains become isolated and reach size of about 600 nmand heights of about 50 nm above the base level ( Fig. 8(c)).Because of the observed hexagonal growth habit of crystallitesand powder diffraction measurements (not shown here) thatconrmed epitaxial relation between islands and substrate, itwas concluded that the seed layer crystallized at 1100 1C pos-sesses the required characteristics for the growth of orientedlms. With further increase of temperature to 1200 1C (Fig. 9(a))the hexagonal grains start to coalescence forming very largegrains but at the same time small droplet are formed due tosome melting process causing degradation of hexagons. At1300 1C the melting led to complete deterioration of hexagons(Fig. 9(b)). Fig. 9(c) exhibits typical microstructure of the surfaceafter the deposition of bulk 600 nm thick layer onto the 40 nmthick seeding layer, both crystallized at 1100 1C.

    The microstructure of SrM thin lms inspected by means of SEM is shown in Fig. 10. Annealing at temperatures slightly higherthan the onset of SrM phase crystallization induces formation of lm microstructure of equiaxed grains with size below 50 nmincluding certain porosity ( Fig. 10(a)). Increasing the annealingtemperature causes the transformation of crystallites fromequiaxial to platelets with length of about 300400 nm andthickness less then 100 nm ( Fig. 10(b)) that are randomly orientedthroughout the cross-section of the lm. By further increase inannealing temperature (cross-section view not shown here) theplatelets start to sinter changing grainy microstructure into morecompact one, losing its original appearance of well resolvedgrains. The process of sintering is illustrated in Fig. 10(c) and(d), displaying plane view of surface of SrM lm crystallized at

    1100 1C before ( Fig. 10(c)), and after ( Fig. 10(d)) the removal of

    the surface layer by means of ion etching. According toFig. 10(c) the surface of 600 nm thick lm is composed of thinplatelets of hexagonal shape oriented mostly parallel to thesurface with basal diameter about 1000 nm, frequently with theirbases glued-up together and forming stacking microstructure.Removal of top surface by ion etching uncovers the interior of thelm ( Fig. 10(d)). The microstructure looks like being composed of continuous mass without any grainy structure and incorporatescertain porosity indicating incomplete sintering.

    Fig. 11 shows the static magnetic properties of SrM thin lmson SrTiO 3 (111) crystallized at 800 1C (Fig. 11(a)) and 1100 1C(Fig. 11 (b)) measured with orientation of external magnetic eldboth parallel, and perpendicular to the sample surface. The datain Fig. 11 (a) give broad hysteresis loops with a coercive eld of 67 kOe, a remanent magnetization of 110130 emu/cm 3 , and asaturation magnetization of approximately 200 emu/cm 3 for bothorientations of magnetic eld. The data collected on samplescrystallized at 1100 1C (Fig. 11 (b)) give for perpendicular orienta-tion square shaped hysteresis loop with a coercive eld 2.8 kOe, aremanent magnetization 220 emu/cm 3 , and a saturation magne-tization of 230 emu/cm 3 . Parallel orientation of magnetic eld hasbeen observed as almost linear response of magnetization versuseld and reduced hysteresis character of the loop. Our lms arec -axis oriented with pronounced magnetic anisotropy and reachmagnetic parameters, which are consistent with values on SrMferrite in literature. The static magnetization data were used todetermine the saturation magnetization for elds used in ferro-magnetic resonance measurements.

    Fig. 12 shows the differential absorption dA/dH vs. external

    eld H for 600 nm thick SrM lm deposited on SrTiO 3 (111) and

    14

    rms = 3.4 nm

    distance (nm)

    7

    21

    h e

    i g h t ( n m

    )

    h e

    i g h t ( n m

    )

    h e

    i g h t ( n m

    )

    0 197 394 591 788 985

    1000 nm0

    2500 nm0

    2500 nm0distance (nm)

    13

    26

    39

    0 202 405 607 810

    rms = 22.0 nm

    1012

    distance (nm)

    34.3

    68.5

    0 417 833 1250 1666 2083

    rms = 26.1nm

    99.8

    Fig. 7. AFM images of seed SrM layer deposited on SrTiO 3 (111) and crystallized at 600 (a), 700 (b) and 800 1C (c), together with height prole scans over selected featuresin surface and calculated r.m.s. values.

    J. Bur sk et al. / Journal of Solid State Chemistry 184 (2011) 30853094 3089

  • 7/27/2019 1-s2.0-S0022459611005172-main

    6/10

    crystallized at 800 1C (Fig. 12 (bottom)), and at 1100 1C (Fig. 12(top)). For elds above saturation magnetization eld and in caseof material exhibiting preferential growth with c axis collinearwith sample normal, resonance frequency is given by the follow-ing equation [11] :

    o res gH res H demag H A g mBh

    H res 4p M sat 2K AM sat

    where o res is the resonance frequency, g is gyromagnetic ratio,H res is resonance external eld, H demag is demagnetization eld, H Ais anisotropy eld, mB is the Bohr magneton, K A is uniaxialanisotropy constant, and g is the g-factor. Using the values of

    saturation magnetization M sat from magnetization SQUID mea-surements and by performing FMR measurements at least at twofrequencies we can obtain the values of parameters g and K A.

    Sample crystallized at 800 1C did not exhibit saturation eldvalue for elds needed for resonance at both 49 and 69 GHz(see Fig. 11(a)), and show broad resonance curve widths atboth frequencies in Fig. 12 (bottom); therefore the g -value1.991 found for sample crystallized at 1100 1C was used for theevaluation of spectra of both samples. Then we have obtained forsample crystallized at 800 1C for both working frequenciesH A 15.9 kG and K A 1.46 erg/cm 3 (Fig. 12 (bottom)). The samplecrystallized at 1100 1C exhibits narrower FMR curves at bothfrequencies ( Fig. 12 (top)), the values of anisotropy eld H Aand anisotropy constant K A were evaluated as 16.5 kG and

    1.88 erg/cm3

    , respectively.

    The 57 Fe NMR spectrum measured on the lm samples crystal-lized at 800 and 1100 1C together with inset plots of tted lineproles are plotted in Fig. 13. The four spectral lines correspond-ing to resonance of 57 Fe nuclei in 12 k, 4 f 1 , 2a and 4 f 2 sites of SrFe12 O19 magnetic structure [12 ,13 ] (the 2 b resonance was notmeasured because of suspected too low signal/noise ratio) areplotted as shifted from resonant frequencies observed in bulksingle crystal [13] . In comparison to the spectrum of a bulk SrMsingle crystal it is evident that the spectral lines of magneticsublattices with magnetization parallel to total magnetization(12 k and 2 a , spin up) are shifted to higher frequencies, while thelines of sublattices with antiparallel magnetization (4 f 1 and 4 f 2 ,spin down) are shifted to lower frequencies. Shift of lines foundfor the sample crystallized at 800 1C are markedly (about0.2 MHz) higher relatively to sample crystallized at 1100 1C.While the line shifts of sample 800 1C are nearly uniform, in caseof sample crystallized at 1100 1C the shift of 2 a line is lower thanthose of the remaining lines. In addition, the lines are signicantlybroadened, having the linewidth of approximately 0.4 MHz (cor-responding to width of the distribution of magnetic eld beingabout 0.3 T), which is more than an order of magnitude higherthan in the single crystal. Spectral shapes were tted to theexperimental data supposing the same line prole for all the fourdetected lines, of course considering the mirror symmetry of the12 k and 2 a lines with respect to those of 4 f 1 and 4 f 2 . The lines incase of the sample crystallized at 1100 1C were asymmetric, whilethe Gaussian prole was sufcient for tting spectrum of the

    sample crystallized at 800 1C. Integral intensities of individual

    2500 nm0

    62

    124

    186

    0 597 1193 1790 2386 2983

    rms = 55.9 nm

    h e

    i g h t ( n m

    )

    h e

    i g h t ( n m

    )

    h e

    i g h t ( n m

    )

    distance (nm)

    2500 nm0 distance (nm)

    31.7

    63.5

    95.2

    0 216 432 648 864 1080

    rms = 26.6 nm

    2500 nm0 distance (nm)

    24.2

    48.5

    72.7

    0 397 795 1192 1590 1987

    rms = 26 nm

    Fig. 8. AFM images of seed SrM layer deposited on SrTiO 3 (111) and crystallized at 900 (a), 1000 (b) and 1100 1C (c), together with height prole scans over selectedfeatures in surface and calculated r.m.s. values.

    J. Bur sk et al. / Journal of Solid State Chemistry 184 (2011) 308530943090

  • 7/27/2019 1-s2.0-S0022459611005172-main

    7/10

    resonance lines obtained from the t well correspond to multi-plicities of underlying sites.

    4. Discussion

    To obtain single phase thin lm, optimization of preparativeconditions must be achieved. In our case, two effects emerged aspotentially hazardous in an attempt to prepare good quality lms:destruction of lm microstructure due to uncontrolled evolutionof combustion products during pyrolysis of gelled lms, and anincomplete decomposition of alkali earth carbonates before theonset of crystallization of ferrite phase. Thus, the heat treatmentregime has been scheduled based on DTA/TG and high tempera-ture XRD measurements, so that the pyrolysis has been conductedat rather higher temperature 450 1C with 15 min delay whichenabled complete removal of organics from the lm. The processof crystallization in SrM thin lms has been studied starting withannealing temperature of 600 1C where most of the SrCO 3 isalready decomposed and crystallization of ferrite phase starts (seeFigs.1 and 2 ).

    In most of the reports on oriented hexagonal thin lms, single-crystal Al 2 O3 (001) (e.g. [1 ,2]) or MgO(111) (e.g. [5 ,8]) substrateswere used in combination with some physical deposition method.Although lms deposited onto substrates with different symme-try or large value of lattice mist, or amorphous substrates wouldnot intuitively be expected to be oriented, results published[4,14 ,15 ] has exhibited certain degree of c -axis oriented growth

    of hexagonal ferrite onto amorphous SiO 2 with ZnO buffer layer

    by PLD [15] , and on Si(100) and YSZ(100) by laser ablationmethod [4,14 ]. Regarding the in-plane orientation of ferrite lmsno direct experimental evidence is given in these papers.

    Our attempt to grow oriented SrM thin lms on selected groupof substrates by means of CSD method illustrate commonlyaccepted fact that for the same type of materialsubstrate arrange-ments different preparation techniques can give thin lm withdifferent properties. Generally, using our chemical method weobserved randomly oriented SrM phase for temperatures at whichthe onset of orientation growth is usually observed using physi-cal deposition methods (typically 800900 1C [1,2,4,5,14 ,15 ]).Increasing the annealing temperature above 900 1C caused fre-quent formation of lms with better c -axis orientation but at thesame time noticeable smearing of lmsubstrate boundary accom-panied by interdiffusion or formation of other phases occurred (seeTable 1 ). This effect can be observed also by comparing the valuesof FMR line widths (see Fig. 12) where higher annealing tempera-ture leads to pronounced narrowing of the line width (the value of the anisotropy eld changes only slightly). The reason for this canlie among others in the difference, in which both types of methodsdeliver thermal energy to atoms needed for them to undergooriented growth due to their increased mobility: continuousdeposition of adatoms on concurrently preheated substrates ine.g. PLD experiments is more convenient than use of post-deposi-tion pyrolysis and annealing typical for CSD methods. Seeing ourndings about phase composition of thin lms prepared ondifferent substrates summarized in Table 1, conclusion can bedrawn that not only structural similarity and low lattice mist

    between deposit and substrate are substantial for reaching of

    2500 nm0

    rms = 29 nm

    distance (nm)

    34.2

    68.5

    102.7

    0 403 805 1208 1610 2013

    h e

    i g h t ( n m

    )

    h e

    i g h t ( n m

    )

    h e

    i g h t ( n m

    )

    5000 nm0

    28.3

    56.5

    84.8

    rms = 29.7 nm

    0 892 1784 2675 3567 4459distance (nm)

    2500 nm00 585 1170 1755 2340 2925

    58

    116

    rms = 29.5 nm

    distance (nm)

    a

    b

    c

    Fig. 9. AFM images of seed SrM layer (a, b) crystallized at 1200 1C (a), 1300 1C (b); and top surface of bulk layer (c) crystallized at 1100 1C (both seeding and bulk layer),all deposited on SrTiO 3 (111), together with height prole scans over selected features in surface and calculated r.m.s. values.

    J. Bur sk et al. / Journal of Solid State Chemistry 184 (2011) 30853094 3091

  • 7/27/2019 1-s2.0-S0022459611005172-main

    8/10

    oriented growth. It is evident by comparison of phase compositiongained on substrate with far the best lattice mist value andappropriate symmetry relation (MgO(111), with results gained on

    substrate with almost ve times higher value of lattice mist

    (SrTiO3 (111)), or phase composition gained on otherwise com-monly used Al 2 O3 (001) substrate, with results on SrTiO 3 (111)

    substrate exhibiting almost the same value of lattice mist, that

    SrM film

    SrM film

    Fig. 10. Cross-section (a, b) and plane view (c, d) SEM images of the SrM thin lmdeposited on SrTiO 3 (111) and crystallized at 700 1C (a), 900 1C (b), and 1100 1C(c,d) before (c) and after (d) the ion etching of the surface.

    -15.0 -10.0 -5.0 5.0 10.0 15.0

    -200

    -150

    -100

    -50

    50

    100

    150

    200

    perpendicular

    parallel

    magnetization (emu)/cm 3

    mg. field strenght (kOe)

    -25 -20 -15 -10 -5 5 10 15 20 25

    -200

    -150

    -100

    -50

    50

    100

    150

    200magnetization (emu)/cm 3

    mg. field strenght (kOe)

    perpendicular

    parallel

    Fig. 11. Magnetization hysteresis curves measured at room temperature on SrMthin lms deposited on SrTiO 3 (111) and crystallized at 800 1C (a), and 1100 1C (b),with both parallel and perpendicular orientations of mg. eld relatively to thinlm surface.

    0 1 2 3 4 5 6 7 8 9 10 11 12 13 14

    800C

    derivative of absorption (dA/dH)

    external mg. field H (kG)

    =49.1 GHz =69.7 GHz

    1100C 1100C

    800C

    Fig. 12. FMR absorption derivative ( dA/dH ) vs. static external eld ( H ) spectra of thin SrM lms deposited on SrTiO 3 (111), crystallized at 800 1C (bottom) and1100 1C (top), and measured at room temperature at 49.1 and 69.7 GHz.

    J. Bur sk et al. / Journal of Solid State Chemistry 184 (2011) 308530943092

  • 7/27/2019 1-s2.0-S0022459611005172-main

    9/10

    also chemical reactivity of constituent elements at elevated tem-peratures is of importance.

    X-ray diffraction data of y2 y scans and rocking curve scans onlms deposited on SrTiO 3 (111) substrate ( Figs. 4 and 5 ) illustrate,that the process of oriented growth starts almost simultaneouslywith the onset of phase crystallization (i.e. at approximately600 1C) and continues further with increase in annealing tem-perature. Rocking curve measurements ( Fig. 5) enable us tospeculate about this development. The at curve of SrM lmsannealed at 500 1C (the contribution from the seeding layer tototal measured signal was indiscernible) indicate almost randomorientation of grains. Rocking curve measured on sampleannealed at 600 1C showing quite narrow distribution of [001]directions and relatively lower intensity indicates that the orien-tation growth commences. We believe that this process proceedsin bottom-up manner through the consumption of ne, randomlyoriented grains by large epitaxial grains from seeding layer thatpossess a low interfacial energy and act as nucleation sites for thegrowth of oriented lms as it is described in literature [16 18 ].The driving force for this transformation should be the reductionin excess grain boundary energy [17] : larger well oriented grainsat the bottom interface consume smaller poorly oriented grainsabove. We consider the bimodal shape of rocking curves observedfor samples annealed at temperatures between 700 and 900 1C(centered at the same o , and with progressively increasingsignicance of the narrow peak and diminishing contribution of broad component to the envelope curve) to be the consequence of incompleteness of this transformation. After annealing at tem-peratures above 1000 1C, fully transformed highly oriented lmshowing monomodal and very narrow rocking curves is obtainedand epitaxial relationship between substrate and thin lm wasproved (see Fig. 6).

    Two types of sample surfaces were studied by means of AFM:

    surface of the 40 nm (effective thickness) seed layer and top

    surface of the 600 nm thick nal sample ( Figs. 79 ). The measure-ments on the rst type of sample proved that under optimumconditions isolated islands of grains having a hexagonal shapewith basal plane oriented parallel to the lm surface are formed.Usage of this phenomenon for getting better oriented lms hasbeen described for variety of oxide systems: Zr(Y)O 2 on Al2 O3(001) [16] , PbTiO3 on SrTiO 3 (111) [17] , PbTiO3 on LaAlO3 (100)[18] or SrTiO 3 , BaTiO3 on LaAlO3 (100) [19] .

    In order to be the trick of using a thin seed layer for theoriented growth effective, the deposited layer must be sufcientlythin to form a discontinuous layer of in nature isolated grainswith an epitaxial relation to the substrate to allow for theelimination of nonoriented grains and simultaneously the cover-age should be sufcient to have sufciently high number of theseseeds [17] . These isolated, single crystal islands with an epitaxialrelation are the conguration that has the lowest free energywhen the lm thickness is less than a certain critical value[16 ,17 ]. By recoating the substrate covered with seed layer, theseeds consume much smaller nonoriented or poorly orientedgrains within the new deposits through grain growth occurringwith increasing annealing temperature as it can be illustrated inFig. 5 by decreasing FWHM values of the broad texture compo-nent for bimodal rocking curves observed at intermediate tem-peratures. Based on XRD texture analysis and AFM measurementsit was determined that optimum conditions for growing of seedlayer is 40 nm effective thickness and crystallization at 1100 1C.

    According to SEM measurements, the microstructure changedfrom nely grained equiaxed grains observed in sample annealedat 700 1C (Fig. 10(a)) to the lms composed of needle-like shapedgrains with entrapped porosity in sample annealed at 900 1C(Fig. 10(b)). Comparing these results with XRD data on lms (seeFig. 4) and with rocking curve measurements ( Fig. 5) we canconclude, that despite the use of epitaxial seed layer, (homo-geneous) nucleation still can occur within the bulk of the lm.Similar ndings were mentioned by other authors [19 21 ]. Theyargue that either the thermal energy barriers for hetero- andhomo-nucleation (which differ in factor governed by contactangle [20] ) are not dramatically different, or the decompositionpathway from amorphous pyrolysed material leading to theformation of relatively stable intermediates within the lmdictates the nal microstructure. The negative inuence of reac-tion pathway on microstructure and orientation developmentduring the annealing has been conrmed for systems where oneof the constituent elements was alkali earth metal giving rise torelatively stable carbonates. It was found in [21] that even underalmost optimal conditions, when BaTiO 3 epitaxial seed layer of high quality was prepared by means of molecular beam epitaxy,the formation of nonoriented crystallites within the bulk of lmdominates microstructure in BaTiO 3 upper lms deposited byCSD. Similarly, in [19] the crystallization of SrTiO 3 and BaTiO 3 thinlms on LaAlO 3 substrates using seeded growth, bulk nucleationwas observed instead of nucleation starting at the seeding layerwith following consumption of an amorphous matrix lyingupwards. It is believed that in all these cases the decompositionpathway takes place via intermediate carbonate which form attemperatures between 300 and 400 1C and undergo decomposi-tion by reaction with other constituent at temperatures around600 1C [19 ,20 ]. Our results for the retarded decomposition of SrCO3 observed in dried sol by means of high temperature XRD(Fig. 2) are apparently in agreement with those studies.

    The removal of material from top surface of lms heat treatedat 1100 1C (Fig. 10(c and d)) demonstrate that beneath the surfaceof densely packed thin platelets of hexagonal shape the micro-structure collapsed into a more compact microstructure losing itsoriginal appearance of well resolved grains. Though, certain

    porosity is still observed, evidencing incomplete sintering.

    Fig. 13. 57 Fe NMR spectra of SrM thin lms deposited on SrTiO 3 (111) andcrystallized at 800 1C (top) and 1100 1C (bottom). Circles experimental data,solid line t. Vertical dashed lines are drawn at resonance frequencies corre-sponding to perfect single crystal. Nested plots: tted shape and shifts of spectrallines from resonant frequencies in a single crystal sample.

    J. Bur sk et al. / Journal of Solid State Chemistry 184 (2011) 30853094 3093

  • 7/27/2019 1-s2.0-S0022459611005172-main

    10/10

    The static magnetic measurements ( Fig. 11 ) essentially con-rmed the ndings of the structural and microstructural studies.The decrease in coercive eld for the external eld appliedperpendicular to sample plane from 6.5 kOe (annealing at800 1C) to 2.8 kOe (annealing at 1100 1C), together with pro-nounced squareness ( M rem /M sat ratio) reaching value 0.97 forsample annealed at 1100 1C, indicate improved magnetic aniso-tropy with increase in annealing temperature. The presence of

    certain hysteresis for data measured with parallel orientation(Fig. 11 (b)) is related to domain considerations and is indicative of presence of various types of defects (point defects, off-stoichio-metric materials, dislocations, stacking faults, etc.) all of whichcan inuence the sample response to external magnetic eldduring magnetization.

    Results of FMR measurements conrmed that our g-factor andanisotropy eld H A values are comparable with values obtainedfor lms prepared by PLD method ( g 1.99 7 0.01 measured in[21] , and H A around 17 kG found in [6 ,22 ,23 ]). However, ingeneral, our CSD lms exhibit broader linewidths compared withPLD lms (see [21 23 ], where some samples yield line widthsaround 30 Oe for only BaFe 12 O19 ).

    One can perceive the common NMR lineshape as proportional(by the 57 Fe gyromagnetic ratio) to distribution of projection of the demagnetizing elds throughout the sample to local easy axis.The maximal possible value of demagnetizing eld induced 57 FeNMR lineshift in SrM is about 0.9 MHz (corresponding to isolatedsingle domain grain with demagnetizing factor N 4/ p ). Theobserved lineshifts are much lower, despite the fact, that as seenby SEM, the grains are very oblate. This is due to compensation of demagnetizing elds via interaction between grains. The line-shifts of sample annealed at 1100 1C are further reduced theoriented grains, have better potential to compensate internal eldwithin the lm, furthermore, since the lateral dimensions of grains are about 1 mm, domain walls may be formed, thuscontributing to the reduction as well. Not perfectly uniformlineshifts in sample crystallized at 1100 1C indicate, that mechan-ism other than macroscopic eld contributes to resulting spec-trum. Additional effects could bring along slight modication of electronic structure, which would alter hyperne eld in eachiron site differently. Strain or structural defects are the mostprobable candidates.

    5. Conclusion

    This study has shown that single phase, epitaxial SrFe 12 O19thin lms can be prepared by chemical solution depositionmethod on SrTiO 3 (111) single crystal substrates using seedinglayer composed of high density of well-oriented, isolated SrMgrains. These well-oriented seeds were formed by the breakup of polycrystalline, nely grained layer prepared under optimum

    deposition and heat treatment conditions. By overcoating thisseeding layer, epitaxial SrFe 12 O19 thin lms was obtained, withpronounced crystallographic and magnetic anisotropy as shownby XRD, SQUID and FMR and NMR measurements. Nevertheless,the problem of homogeneous nucleation of unoriented grains wasnot totally avoided as it was indicated by SEM and XRD measure-ments. Samples crystallized at 1100 1C gave values for in-planeand out-of plane coercivity elds 1.1 and 2.8 kOe, respectively,

    with M rem /M sat ratio reaching value 0.97 for out of plane orienta-tion. Relatively higher values of FMR linewidth prepared can beattributed to strains due to higher crystallization temperaturesand lm-substrate lattice mist used in this work. The strainsmay be as well responsible for nonuniform NMR lineshifts insample crystallized at 1100 C. To improve magnetic characteris-tics of ferrite lms, several approaches will be tested: develop-ment of a more suitable buffer layer, changes in annealing

    conditions to avoid formation of stable alkali earth carbonates.Work on these is currently under progress.

    Acknowledgments

    This work was supported by the Grant Agency of Academy of Sciences of the Czech Republic under Grant nos. IAA100100803and MS0021620834. The authors would like to thank toM. Mary sko (IoP) for SQUID measurements, P. Bezdic ka andA. Pet rina (IIC) for powder XRD y2 y and rocking curve measure-ments, M. Ma rkova (IIC) for DTA measurements, F. M ka (ISSBrno) for SEM measurements and H. St epa nkova (FMP CU) forhelpful discussions on NMR results.

    Appendix A. Supplementary Information

    Supplementary data associated with this article can be foundin the online version at doi:10.1016/j.jssc.2011.09.024 .

    References

    [1] S.R. Shinde, S.E. Loand, C.S. Ganpule, S.B. Ogale, S.M. Bhagat, T. Venkatesan,R. Ramesh, J. Appl. Phys. 85 (1999) 7459.

    [2] Y.-Y. Song, S. Kalarickal, C.E. Patton, J. Appl. Phys. 94 (2003) 5103.[3] G. Albanese, A. Deriu, The LandoltB ornstein Database, Group III Condesed

    Matter, vol. 4b, Part B, Chapter 10, Springer Verlag, Eds. K.-H. Hellwege,

    A.M. Hellwege, 547.[4] B.R. Acharya, S. Prasad, N. Venkataramani, S.N. Shringi, R. Krishnan, J. Appl.Phys. 79 (1996) 478.

    [5] S.A. Oliver, I. Kozulin, M.L. Chen, C. Vittoria, Appl. Phys. Lett. 76 (2000) 3612.[6] S.G. Wang, S.D. Yoon, C. Vittoria, J. Appl. Phys. 92 (2002) 6728.[7] J. Bur sk, Z. Sim sa, M. Cer nansky, J. Sol-Gel Sci. Technol. 8 (1997) 947.[8] D.F. Ryder, J.H. Fenstermacher, J. Am. Ceram. Soc. 73 (1990) 282.[9] H. Tang, W. Zhang, B. Peng, W. Zhang, Thin Solid Films 518 (2010) 3342.

    [10] Z. Frait, Czech. J. Phys. 9 (1959) 403.[11] J.R. Macdonald, Proc. Phys. Soc. 64 (1951) 968.[12] H. St epa nkova, J. Englich, P. Nova k, D. Sedla k, M. Pfeffer, Hyperne Interact.

    50 (1989) 639.[13] H. St epa nkova, K. Kou ril, V. Chlan, P. G ornert, R. M uller, J. St epa nek, J. Magn.

    Magn. Mater. 322 (2010) 1323.[14] V. Papakonstantinou, M.O. O Neill, R. Atkinson, I.W. Salter, R. Gerber,

    J. Magn.Soc. Jpn. 20 (1996) 333.[15] P.C. Dorsey, C. Vittoria, J. Magn. Magn. Mater. 137 (1994) 89.[16] K.T. Miller, F.F. Lange, J. Mater. Res. 6 (1991) 2387.[17] A. Seifert, A. Vojta, J.S. Speck, F.F. Lange, J. Mater. Res. 11 (1996) 1470.

    [18] J.H. Kim, F.F. Lange, J. Mater. Res. 14 (1999) 1626.[19] R.W. Schwartz, P.G. Clem, J.A. Voight, E.R. Byhoff, M. van Stry, J. Headley,

    N.A. Missert, J. Amer. Ceram. Soc. 82 (1999) 2359.[20] R.W. Schwartz, Chem. Mater. 9 (1997) 2325.[21] M.C. Gust, N.D. Evans, L.A. Mamoda, M.L. Mecartney, J. Amer. Ceram. Soc.

    80 (1997) 2828.[22] C.A. Carosella, D. Chrisey, P. Lubitz, J.S. Hotwitz, P. Dorsey, R. Seed,

    C.J. Vittoria, Appl. Phys. 51 (1992) 5107.[23] L.V. Saraf, S.E. Loand, A.V. Cresce, S.M. Bhagat, R. Ramesh, Appl. Phys. Lett.

    79 (2001) 385.

    J. Bur sk et al. / Journal of Solid State Chemistry 184 (2011) 308530943094