You are on page 1of 3

IEEE TRANSACTIONS ON MAGNETICS, VOL. 24, NO.

2, MARCH 1988 QUANTITATIVE STUDY OF MAGNETIZATION RIPPLE IN C061B39 AMORPHOUS THIN FILMS In-Seop Jeong' and Rodger M. 1Centerfor Materials Science and Engineering 2Deparhnent o fElectrical and ComputerEngineering 3J.H.Herrinn Centenxial Professor in Engineering The University ofTexas>t Austin, Austin, TX 78712, U S A .

1725

We have performed the first quantitative, two-dimensional, image processing analysis of magnetization ripple. The measured longitudinal and transverse ripple wavelengths of Co61B39 amorphous thin films were in the range 0 . 6 - 0 . 7 ~and 21.9pm respectively and remained constant for heating below the Curie temperature Tc=350"C. The measured value of the longitudinal ripple wavelength agreed well with the value calculated from Hoffmann's ripple theory, but the transverse ripple wavelength is nearly ten times larger than expected. The ripple angle calculated from image processed data is = 0.2-0.3 radian and a constant for temperatures below Tc. Digital image processing techniques have also been used to analyze thermally induced changes in the magnetization ripple of these thin films. The intensity of the magnetization ripple decreased monotonically with increasing temperature and vanished at T,. Intense ripple was observed in films cooled from the =45OoC crystallization temperature, Tx, and ripple width was observed to vary inversely with cutoff in the angle of incidence of the deposition beam. We also discuss the significance of these observations relative to previously observed anisotropy-nanostructuralrelationships in these films. INTRODUCTION Magnetization ripple, defined as the sinusoidal fluctuation of local magnetization, has been widely investigated theoretically and experimentally because of its importance in detekining micromagnetic properties df magnetic materials 11-51. Although several researchers have quantitatively studied magnetization ripple, the results have been obtained from I-D microdensitometry scans[2-4]. Most of these have compared their measured magnetization ripple parameters, mainly the longitudinal ripple wavelength hi, and the ripple angle 8, with Hoffmann's ripple theory [I]. There is, however, a paucity of experimental measurements of the transverse i and ht must be known to calculate the ripple wavelength Xt. Since both X magnetic coupling volume and other magnetic parameters, it is of interest to measure the transverse ripple Wavelength, and to assess the validity of Hoffman's theory for amorphous thin films. In this work, we performed the first quantitative, 2-D analysis of the magnetization ripple. We used 2-D image processing techniques to study ripple in sputter deposited C061B39 amorphous films. The results were compared to the theoretical values from Hoffmann's ripple theory. We also used the image processing technique to analyze the variation of magnetization ripple parameters of films deposited at two deposition beam incident angles, and with the film temperature. The results are of interest for understanding how the nanostructural morphology of these films influences their soft magnetic properties, and for correlating nanostructures and magnetization ripples. Lorentz micrographs of the amorphous Co61B39 films were obtained using a transmission electron microscope (TEM) with the smallest coherent electron probe [5] and constant lens currents. The Lorentz images were digitized and stored. The magnetization ripple could then be quantitatively analysed by manipulation of the digital image intensities. The purposes of this work were to; analyze the magnetization ripple two-dimensionally, examine the effects of incident angle and temperature on the magnetization ripple, and correlate the thermally induced changes in the nanostructural morphology and magnetization ripple of amorphous thin films. EXPERIMENTS Bmple preparation Co61B39 amorphous films were directly deposited by rf sequential cosputtering onto rotating (2 rpm), water-cooled, substrates with substrate bias of -6OV. The substrates were carbon-backed, 3mm diameter, Cu-grids for Lorentz microscopy, and carbon-backed slide glasses for magnetic property measurements. The background pressure and the argon sputter gas pressure were 2xlO-7Torr and lOmTorr respectively. The maximum deposition vapor incidence angle L2 was controlled by placing the substrates inside Pyrex cylinders. The 10' and 20' vapor incidence angles used in this work were achieved by using cylinders with inner diameter 25mm and 33mm respectively, and a 23mm height. The composition and the thickness of the films were controlled by adjusting the target voltages and deposition time, respectively. Details of the sputtering geometq, vapor incidence, and deposition beam pattern have been described previously 161.

Lorentz microscopy For quantifying the ripple structures the Lorentz micrographs were processed with fixed electron microscope operating parameters, i. e., film thickness, specimen position, electron probe size, acceleration voltage, and lens current. The lens circuit voltages in Table 1 were used to obtain constant defocus length (I.lcm and 0.63cm for L2=10 and 20' respectively), magnification (1800x), and exposure time (llsec). For parallel beam operation (nearly zero illumination beam angle) the smallest spot size, with open apertures, was used. We used a JEM-1200EX with a heating stage, operating in the TEM mode at 120KV. The specimen chamber pressure was 7x 10-7 Torr during heating. The Lorentz images were formed by scanning a small, coherent, electron probe with the fixed parameters given above. The film morphologies were obtained from micrographs taken with slightly underfocused phase contrast. W l e 1. V o Condenser lens 1 Condenser lens 2 Objective lens Objective mini

.f .
4.50 5.15 off 7.06

Intermediatelens 1 Intermediatelens 2 Intermediate lens 3 Projector lens

1.31

6.86
3.16

8.02

and Data Processing Source images, with consistent contrast for image processing, were obtained from negative films of Lorentz micrographs which were carefully prepared under nearly identical TEM operating conditions. The negatives were mounted on a light box and digitized from a video camera yielding images with 512x512 pixels which were then placed in disc storage. Image processing was accomplished on a VAX 11/780 computer system that controlled data acquisition, processing, and display devices as shown in Fig.1. A 128x128 pixel portion of the ripple area (4pm x 3.8pm areas outlined in Figs. 2 and 4) was selected for detailed processing. The computer results were viewed and photographed on a GRINNELL 270 color graphic display system and a matrix camera (Fig.3). The ripple intensity was calculated from the peak to peak height of the intensity profile at selected positions on the boundary of the ripple image. The longitudinal and transverse ripple wavelengths were calculated from the most dominant periodic frequencies obtained from the 2-D Fourier transformations of the ripple intensity profiles.

-1
Negative Film Camera Digitizer1 Video Frame Store VAX Ill780

I
MATRIX Camera

MRsArrc
Plotter

m~
plotter

(Fig.]) Schematic diagram of the image and data processing

EXPERIMENTAL RESULTS A N D DISCUSSION In this section we discuss the results of the quantitative, 2-D, image processing analyses of the magnetization ripple in amorphous C061B39 thin films and compare these with calculations from Hoffmann's ripple theory Ouanbtabve Studv of the Magnetization Rioulp The ripple intensities and wavelengths calculated from the digitized data of the ripple image area marked with the squares in Figs. 2 and 4, are shown in Table 2. Longitudinal and transverse ripple wavelengths of 0.6-0.7pm and 21.9pm respectively were obtained. These values were constant below

VI.

OO18-9464/88/03OO- 1725$01.OO@1988 IEEE

1126

the Curie temperature as previously observed by Grundy et. al. [7] for CoFeB foils. From Hoffmann's ripple theory [l], the ripple wavelengths at zero applied field are

hi= 2x: (A/K,)ln (1) ht = Mst lnA'/4 Ku-3/4 (2) for either crystalline films or amorphous films. In (1) and (2) A is the
exchange constant and K, is the effective anisotropy constant associated with the total uniaxial anisotropy including the growth anisotropy and the local anisotropy. K, is assumed constant between room temperature and the Curie temperature as observed by Labrune et. al. [8]. The calculated l.1 from (1) is = 0 . 6 3 p with the parameters A and K, taken at lxlO-'%rgs/cm and Ixl04ergs/cm3 respectively. This value agrees well with the value =0.67pm measured in this work for the films deposited with R=lOo. However, the value of hg(0.18 pn) calculated from (2) with Ms = 435 G is only the order of a tenth of the measured value ( h p l . 9 p m ) for the film of R=lOO. Moreover, the calculated hl is much smaller than hi, which clearly contradicts the experimental evidence (readily apparent in Figs.2 and 4) that hl is much larger than 11.

(Fig.3) One dimensional intensity profiles of digitized ripple images. Polaroid picture from matrix camera. Ripple images and intensities at (a) 25C @) 100C (c) 150C (d) 250C i n CO61B39 N m s deposited with -6OV bias and R=lO" (e) magnified,digitized ripple image and intensity at 25"C, R=lOO. (a-d) 4pm x 3 . 8 (e) ~ 5.lpm x 4.8pm with Iave the average (normalized) intensity, and Z is the defocus length. The angular deflection Qxis
Qx = 4~ethMy (hC)-1 (5) where the electron wavelength h=0.00335nm at the 120KV accelerating voltage, t is the film thickness, and e, C: and h are respectively, the electron charge, the velocity of light, and Planck's constant Then, the ripple angle 8 e = 0, sin(2ny / hi) (6) becomes eo at y = d l (n integer). The ripple angle e calculated from (3)-(6) is of the order of 10-1 radian and assumed to be constant as shown in Table 2.

(Fig.2) Variations of domain structures in Co61B39 films deposited with -6OV bias and R=10" by heating (a-f) and cooling (f-g) (a) 25C @) 100C (c) 150C (d) 250C (e) 350C (f) 450C (g)Z"C after cooling f r o m 450C The peak-to-peak heights of the intensity profiles of the areas shown in Figs. 2 and 4, were calculated at different temperatures (Fig. 3a-d) and plotted in Fig.5. The curves exhibit the functional M(T)-T relation in ferromagnetic films because, assuming a constant ripple angle [3], the intensity is proportional to the magnetization [9]. Since M, at room temperature for these films was known from the hysteresis loop measurements, we used this data to estimate the magnetization values at elevated temperatures. Further assuming that the domain magnetization My = Ms, the ripple angle could be calculated from the measured ripple intensities and wavelengths. From Fuller and Hale [9], and including wave , is given by optical effects, the ripple amplitude 0 eo = I h1( 2 m ~ ~ ) - 1 (3) where the intensity I = (Imax-Imin)/(Imax+Imd = (Imax-Imid(2 L e ) (4)

(Fig.4) Variations of domain structures in COfjlB39 films deposited with -6OV bias and R=20" by heating (a-d) and cooling (d-g) (a) 25C @) 150C (c) 250C (d) 350C (e) 250C (f) 150C (g) 25C after cooling from 350C

1121

-sEn

. .

eo

loo

20

H 25 ,262 HI00 .I98 H150 .I85 H250 '.141 H 25 ' ,203 Hl00 ,176 H15O .156 H200 .I45 H250 .137 H300 .090 C300 .078 C250 .lo9 CloO .I25 C150 .I48 Cl00 .152 C 25 .168

P;Heating, C;Cooling 4aM 0, 0 OLm) W) (cm) (nm) (KG) (IWrad) (rad) 5.47 .796 .601 4.13 .667 1.9 1.10 18 3.86 S62 .318 2.94 .428 5.62 1.590 4.87 1.378 4.32 1.222 4.01 1.135 3.79 1.072 S71 23.8 0.63 35 2.49 .705 .184 2.16 .611 3.02 ,855 3.46 .979 4.10 1.160 4.21 1.191 4.65 1.316
ly

ht

Annealing was observed to change nanostructural features such as the size, shape, number of nanometer scale amorphous columnar clusters and voids, and the surface roughness. All of these can have an important effect on the growth induced anisotropy [6,13]. The improved soft magnetic properties of amorphous films with small anisotropy produced in this study by annealing (with or without magnetic field) between T, and Tx can be explained by these nanostructural changes. In our future research the effects of annealing and/or aging on the domain and ripple structure will be directly related to changes in the nanoscale morphology. We are particularly interested in correlating the nanostructure and magnetization ripple with the high frequency, small signal susceptibility of the type of soft, amorphous thin films used in this study.

C I

Q-10"

a-60"

tures with T e m D e m Domain structures of 18 nm Co61B39 films deposited at an angle of incidence R=lO"were stable to =150C. However, most of domain walls disappeared at =250C (Fig. 2d) and the ripple vanished at T,p35O0C (Fig.2e). Further increase in temperature led to crystallization of the amorphous film at Tx=4500C. The very fine domain structure that appeared at this temperature (Fig. 2 0 was likely due to a randomly dispersed magnetocrystalline anisotropy and a small magnetostrictive anisotropy. After cooling, the ripple intensity in the crystallized film was very strong (Fig. 2g) because of the randomly dispersed magnetocrystalline anisotropy, the shape anisotropy of the large grains, and internal stresses. Fine mesh domains were observed in films cooled from T,, and became larger with decreasing temperature. Qualitatively similar structures were observed in a 35 nm C061B39 film deposited at n=20". This film exhibited a smaller X.1 and a reduced ripple intensity from those of the thinner R=lOo film (Table 2). The smaller X.1 in the films deposited at a more oblique incidence was due to their larger values of KUas previously observed[4, 8,101. Films deposited without a cylinder (C2=6Oo)had a large Ku, small X.1, and barely observable magnetization ripples [6]. Large uniaxial anisotropy can result from a combination of the oblique incidence growth anisotropy [6,11], the local anisotropy from internal stresses of quasi-dislocation dipoles [12], and non-random atomic arrangements [3]. As these films were cooled from T,, the ripple intensity nearly recovered to its original value (Fig.5 and Table 2), but the changes in the domain structure were irreversible (Fig.4 e-g). These films remained in the amorphous state (Fig. 6). Except for a small decrease in the ripple intensity, and small displacements of domain walls, the domain structure of both films was unchanged to =15OoC. Domain, walls were attracted to each other (convergent and divergent walls : A' in Fig.%a),or to defects (see 'B' in Fig. 2a) thereby decreasing the wall and magnetostatic energies. Some imperfections, such as nonmagnetic defects and voids, acted as pinning sites for domain walls. Also, even at low temperature, small domains changed their magnetization direction parallel to that of a neighboring large domain (see 'C' in Fig. 2a). At higher temperature, domain walls disappeared completely. In addition, during cooling to room temperature, intense ripples split into blocking domains as shown in Fig.4 e-g (see arrow).

(Fig.6) Structural relaxation of C06lB39 amorphous films by post annealing at 350C for 1 hr. SUMMARY (1) We have carried out the first quantitative, two-dimensional, image processing analysis of magnetization ripple in amorphous thin films. The longitudinal and transverse ripple wavelengths measured for sputter deposited C06lB39 thin films were 0.6-0.7pm and 21.9pm, respectively. These values were constant as the films were heated and cooled below the Curie temperature =350"C. The ripple angle was 0.2-0.3 radian order and also constant with temperature below the Curie temperature. (2) The ripple wavelength decreased with increasing angle of incidence due to the large induced uniaxial anisotropy associated with growth and local anisotropies. (3) The ripple intensity decreased with temperature due to decreasing internal stress and magnetostriction. Intense ripple was observed in films cooled from the crystallization temperature of =450"C due to the magnetocrystalline anisotropy of the polyclystalline films. (4) The domain structure was stable to =150C. However, domain wall displacement, attraction of domain walls to defects, and magnetization rotation occurred during heating. Splitting of ripples into domains was observed during cooling. (5) Anneal-induced nanostructural changes in the film morphology appear capable of explaining the reduction in their anisotropy and improved soft magnetic character.

ACKNOWLEDGMENTS
We gratefully acknowledge valuable discussions, on Jmentz microscopy with Prof. L.Rabenberg, Center for Materials Science and Engineering, and on image processing with Prof. J.K.Aggarwa1, Advanced Graphics Laboratory, both at the University of Texas at Austin. H.Hoffmann, J. Appl. Phys. 1790(1964) / IEEE Trans. on 32 (1968) / Thin Solid Films 3 223 (1979) Mag. 6 2239 (1983) M.Ali, P.J.Grundy, J. Phys. D; Appl. Phys. 1 P.J.Grundy, J. of Magn. and Magn. Materials. 21 1 (1980) J.Nowak, Thin Solid F i l r n s N 1 (1987) J.N.Chapman, J. Phys. D; Appl. Phys. 1z 623(1984) I.S.Jeong, D.Y.Kim, A.P.Valanju, R.M.Walser, Materials Research Society Fall Symposium, Boston (1986) P.J.Grundy, G.A.Jones, S.F.H.Parker, IEEE Trans. on Mag. 2627(1981) M.Labrune, S.Hamzaoui, I.B.Puchalska, J. of Magn. and Magn. Materials2 323 (1982) H.W.Fuller, M.E.Hale, I. Appl. Phys. 1L 238 (1960) I.B.Puchalska, A.Sukiennicki, T.Tymosz, Phys. Stat. Solidi. 9 575 (1965) D.Kim. R.M.Walser. Materials Research Societv SvmDosium Proc. 3 173(1985) 1218 (1979) H.Kronmuller, IEEE Trans. on Mag. 1.S.Jeone. A.P.Valaniu. R.M.Walser. Submitted to Int. Conf. on Thin Filzs, New Delk,'India, Dec. 1987

0.30

Col6l/B/39 films with -6OV bias R = I O ? heating 0 R =2U? heating R =204 cooling

REFERENCES

IMENSrrY (ARBITRARY UNIT)


0.10..
0.05

..
0
100

200

300

400

TEMPERATURE pC)

(Fig.5) Variation of the ripple intensities with temperatures

1-

Film M 1 m ChanpB with T m The nanostructural morphological changes produced in amorphous thin films by annealing play an important role in the modification of their physical and magnetic properties. The micrographs in Fig.6 show the annea!-induced relaxation of the morphology of the Co61B3g films.

- . .

You might also like