Static and dynamic X-ray resonant magnetic scattering studies on magnetic domains - 5 AN XRMS STUDY OF ION-BEAM-PATTERNED -GDTBFE THIN FILMS

UvA-DARE is a service provided by the library of the University of Amsterdam (https://dare.uva.nl)

UvA-DARE (Digital Academic Repository)

Static and dynamic X-ray resonant magnetic scattering studies on magnetic

domains

Soriano, J.M.

Publication date

2005

Link to publication

Citation for published version (APA):

Soriano, J. M. (2005). Static and dynamic X-ray resonant magnetic scattering studies on

magnetic domains.

General rights

It is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), other than for strictly personal, individual use, unless the work is under an open content license (like Creative Commons).

Disclaimer/Complaints regulations

If you believe that digital publication of certain material infringes any of your rights or (privacy) interests, please let the Library know, stating your reasons. In case of a legitimate complaint, the Library will make the material inaccessible and/or remove it from the website. Please Ask the Library: https://uba.uva.nl/en/contact, or a letter to: Library of the University of Amsterdam, Secretariat, Singel 425, 1012 WP Amsterdam, The Netherlands. You will be contacted as soon as possible.

5 5

A NN

XRMS

STUDY OF

ION-BEAM-PATTERNED D

0 - G D T B F EE

THIN FILMS

InIn an attempt to produce well-defined domain nucleation centers, square arrays of artifi-cialcial defects were produced by focused-ion-beam irradiation of amorphous Gdn^Tb^jFegs thinthin films. Using X-ray resonant magnetic scattering we followed the domain structure overover the magnetization loop. Rather than affecting the domain nucleation mechanism, thethe irradiated dots are found to hinder domain wall motion and thus strongly affect the alignmentalignment and size of the domain pattern.

5.1.. Introduction

InIn the previous two chapters we studied the static and dynamic evolu-tionn of magnetic domains in nearly defect-free thin films with perpendicular anisotropyy using X-ray resonant magnetic scattering. As is clearly illustrated inn these chapters, the Fourier transform involved in scattering dictates that the amountt of information that can be obtained from a scattering experiment in-creasess with the degree of order of the scattering object.

Inn the present chapter, we impose order on the pinning landscape of sim-ilarr films using a focused ion beam (FIB) to reduce the out-of-plane anisotropy inn a square grid of nanodots with a diameter comparable to the domain wall thickness.. The underlying assumption was that these dots would act as

pref-88 8 C H A P T E RR 5

erentiall nucleation sites, allowing us to film the growing domains using the time-resolvedd technique described in the previous chapter. By reducing the ion fluence,, one would in this way be able to extrapolate to the intrinsic nucleation behavior.. As will become clear in the chapter, even the lowest ion fluence used heree causes the anisotropy to lie in the film plane. As a result, the nucleation fieldd is not affected by the dots. However, the domain patterns are strongly af-fected.. We find that at remanence the dots form strong domain pinning centers, whilee in applied fields they are the preferred sites for the down domains.

Ionn irradiation has been extensively used to modify the magnetic proper-tiess of crystalline [144,145,146,147, 148] and multilayered [149,150] magnetic materials.. In these systems, ion irradiation decreases the crystalline order or leadss to interface mixing. In films with an out-of-plane easy axis, this results in aa reduction of the perpendicular magnetic anisotropy [151,152].

Inn amorphous metallic films, the energy of the incident ion causes a par-tiall annealing of the film, leading to the formation of nanocrystallites [153]. As farr as we can tell, no studies on the effect of ion irradiation on the magnetism of amorphouss films have been performed so far. Since the perpendicular magnetic anisotropyy in rare earth-transition metal films is due to anisotropic pair correla-tionss frozen in during the deposition process, the expectation is that the partial annealingg reduces the anisotropy.

Inn the course of this project, rare earth-transition metal films of different compositionn were irradiated with square lattice patterns, with a range of inter-dott spacings a and ion fluences (p. The film described here differs from the GdFe sampless studied in the previous two chapters in that it contains a small fraction off Tb. Owing to its single-ion anisotropy, small fractions of Tb content are suffi-cientt to cause a marked increase in Ku. This, together with the concomitant slow domain-walll propagation [124], results in a very disordered domain pattern for thee demagnetized pristine sample.

Heree we use XRMS to monitor the evolution of the domain pattern over thee magnetization curve, obtaining the averaged microscopic properties of the systemm such as domain orientation and domain size. We show that the FIB ir-radiationn indeed locally reduces the perpendicular anisotropy in RT films on thee nanometer scale without significantly changing the film topography. This

Ann XRMS study of ion-beam-patterned a-GdlbFe thin films 89 -30000 -2000 -1000 0 1000 2000 3000 300 0 200 0 __ 100 E E

^^

Figuree 5.1: Hysteresis loops of pristine Gdn.3Tb3 jFe^. Out-of-plane loop (bottom axis) measuredd with polar MOKE and in-plane loop (top axis) measured with longitudinal SQUID. .

causess the irregular native domains to have a preferential orientation in the di-rectionn of the dot array.

Inn Sect. 5.2 we will describe the properties of the as-grown GdTbFe sam-pless and the effect of FIB irradiation on their magnetic properties. The results fromm the XRMS data will be presented in Sect. 5.3 and discussed in Sect. 5.4. Finally,, the main conclusions will be outlined in Sect. 5.5.

5.2.. Experimental

5.2.1.. Samples

Thee amorphous 50 nm thin Gdn.3Tb3.7Fe85 films was deposited by molec-ularr electron beam evaporation as described in Sect. 3.2.1. Superconducting quantumm interference device (SQUID) magnetometry and magneto-optical Kerr effectt (MOKE) were used to characterize the sample. Fig. 5.1 shows the perpen-dicularr and in-plane hysteresis loops of the pristine sample.

Thee uniaxial anisotropy constant Ku can be estimated from the in-plane nucleationn field Bcr and the saturation magnetization Ms as 1/2Ms(B(;r — }ioMs). Ann interpolation of literature values [154] for GbjsFess and TbisFess was used too estimate the exchange stiffness constant A. The resulting value yields in turn

90 0 CHAPTERR 5

Tablee 5.1: Properties of a pristine Gdn.^Tb^jFegs film.

t t (nm) (nm) 50 0 Ms s (kA/m) (kA/m) 320 0 Bcr r (mT) ) 1256 6 Bc c (mT) ) 22.6 6 (105//m3) ) 1.4 4

aa domain wall width 6 which is ~50 times smaller than the average domain sizee at remanence TQ ~ 400 nm. The obtained magnetic parameters are listed in Tablee 5.1.

5.2.2.. Focused-ion-beam irradiation

Initiall investigations used the wide-beam ion implanter at AMOLF to ir-radiatee GdFe films with low-fluence doses of 125 keV Ar+ ions. Fig. 5.2 shows thee polar MOKE hysteresis loops of the pristine and irradiated films with in-creasingg fluence (for sake of clarity, only the downwards branch of the hystere-siss loop is shown). In the inset, the increasing perpendicular saturation field andd the decreasing slope at remanence prove that the perpendicular magnetic anisotropyy gradually decreases with increasing fluence, in accordance with the literaturee [145,151]. These results demonstrate that even these low doses strongly decreasee the out-of-plane anisotropy without changing the surface morphology:

B(mT) )

Figuree 5.2: Polar MOKE hysteresis loops of a-GdFes film homogeneously irradiated withh 125 keV Ar+ ions. Inset: saturation field (left axis) and slope at remanence (right axis). .

Ann XRMS study of ion-beam-patterned fl-GdTbFe thin films 91 1 (105//m3) ) 0.62 2 Tablee 5.1: Continued Q Q 2.25 5 A A (10-1 2//m) ) 1.02 2 Ö Ö (nm) (nm) 8.5 5 1m 1m ( 1 0 -4/ / m2) ) 15.1 1

evenn the smallest dose is sufficient to destroy the typical stripe signature of the hysteresiss loop and to cause the magnetization to rotate in-plane.

Ass a next step, the focused-ion-beam facility at MESA+ (University of

Twente)) was employed to pattern the specimens. Using a 30 keV Ga+ ion beam

thatt was focused to nominally 30 nm, nine square lattices of resolution-limited dotss were written on both samples. Both the nominal interdot spacing a (200, 3000 and 400 nm) and ion fluence <p (1,5,10-1014 ions/cm2) were varied (see lay-outt in Fig. 5.3-a). Unfortunately, irradiation of large areas was not feasible with thee employed FIB system, so that the exact dependence of the magnetization andd anisotropy on the ion fluence is not known.

Simulationss of the ion-atom collisions were carried out with the software packagee Stopping and Range of Ions in Matter (SRIM) [155]. For the employed ion, energyy and target, they showed a G a+ depth range of 14 nm and indicated colli-sionss as deep as ca. 70% of the layer thickness. In the following, i will designate thee writing direction and y the normal direction. Due to the scattering of the collisionn process, the dot radius is increased by ~15 nm.

Fig.. 5.3-b shows the topography of the patterned GdTbFe film as imaged withh atomic force microscopy (AFM). The white dots in the AFM images are probablyy AI2O3 grains, coming from the the oxidation of the Al capping layer, thatt dominate the sample roughness. Although the change in height due to the ion-beamm irradiation is small, we can observe the dots as small irregular inden-tationss due to the local sputtering of atoms. The indentations have a diameter off ~60 nm and a depth of 2 nm for high fluence (501014 ions/cm2). This agrees welll with the predicted sputtering yield of 5.3 atoms /ion and the dot broaden-ingg as obtained from SRIM simulations. The indentation depth at high fluence iss comparable to the 3 nm roughness of the sample, as can be seen from the sec-tionn in Fig. 5.3-b. We found that the dot shape is sometimes ellipsoidal with the majorr axis along the writing direction. This means that anisotropy patterning

92 2 C H A P T E RR 5 \ c c 1 1 5 5 10 0 2000 300 400 I I IV V VII I II I V V VIII I III I VI I IX X &X&X 200 m

Figuree 5.3: a) Layout of the FIB-modified sample with nine irradiated areas for nominal interdott spacing a = 200, 300, 400 nm and ion fluence <p = 1,5,10-1014 ions/cm2, b) Top: AFMM image of field IX; bottom: height profile of the region in between dotted lines (2

pmpm scan), c) MFM image measured under in-situ field (15 /<m scan, field IX, B = 60 mT).

withh FIB at these low fluences is almost non-destructive leaving the continuous filmm intact.

5.3.. Results

Thee experimental setup used for the X-ray scattering measurements is describedd in Sect. 3.2 and Ref. [18]. In these measurements we used a 100 }im diameterr beam tuned to the Gd M5 absorption edge. The sample was posi-tionedd so that the writing direction was horizontal and a vertical knife edge wass used as beam stop. For each field, the hysteresis loop was measured using thee XMCD signal. Contrary to our expectations, the results were identical to thatt of the pristine sample for all irradiated areas.

Fig.. 5.4 groups the most representative examples of the X-ray scattering patternss in the vast dataset. Firstly, for the pristine sample (first five images in thee top row), the scattering patterns are typical for very disordered magnetic domainn lattices. At the onset of nucleation (image #1), a circularly-symmetric intensityy disc appears, without any sign of higher orders. As the applied field increases,, the disc evolves to a broad ring with increasing maximum-intensity wavevectorr transfer for higher fields. The largest radius corresponds to a do-mainn size T ~ 400 nm. As the sample magnetization approaches saturation, the ringg intensity fades out and ultimately its radius decreases.

A nn XRMS s t u d y of ion-beam-patterned a-GdTbFe thin films 93 3

qyy (unr1)

Figuree 5.4: Field-dependent evolution of the X-ray resonant magnetic scattering pat-terns.. The top row first gives five images of the pristine sample under increasing mag-neticc fields followed by the diffraction pattern of field IX in the saturated state. The rest off the figure shows four representative examples of the domain scattering patterns for eachh of the nine irradiated regions of the sample in the order as shown in Fig. 5.3. In eachh panel, B increases in the indicated order. In all cases, the transmitted beam (black spot,, size not to scale) indicates the q=(0 0) point. Long tick marks are separated by 10

94 4 C H A P T E RR 5

Secondly,, the last image of the top row shows an example of the diffrac-tionn pattern of an irradiated area (field IX) under conditions where the pris-tinee sample is saturated and does not diffract. The FIB-induced changes in the structuree produce a square diffraction pattern. The diffraction peaks that can be labelledd with Miller indices (/' ƒ), where i and j range enumerate the spots per-pendicularr and parallel to the beam stop. In this example, the (0 5) spot is still visible.. Such patterns could be observed in all nine areas, and the integrated intensityy was found to scale with the ion fluence. The patterns disappear when thee photon energy is moved away from the Gd M5 resonance, which proves thatt they result from a structure in the perpendicular magnetization and not

fromm charge structure due to the implanted Ga+ ions. Since the pristine parts

aree saturated perpendicularly to the film, it implies that the moments in the ir-radiatedd areas must have obtained an in-plane anisotropy component. The total integratedd intensity of this pattern was found to reach a maximum at -30 mT, indicatingg that the spins in the dots rotate in the sum of the applied and demag-netizingg fields.

Thee dot spacing derived from these dot diffraction patterns were deter-minedd to be a = 225, 350, 420 nm with an error of 2 nm. These values are all substantiallyy higher than the nominal values 200, 300 and 400 nm.

Thee remainder of Fig. 5.4 shows four typical domain scattering patterns forr each of the nine areas shown in Fig. 5.3. Each of the four images was taken att fields where the magnetization is approximately -30, 0, 30 and 60% of the saturationn value, increasing with the pattern's ordinal number. The position off the direct beam is indicated by a black spot on the left side of each pattern. Backgroundd subtraction was carried out in all the images, either with a constant valuee or by using an image measured at saturation.

a = 225 nm (left column, array I, II, III): At nucleation (images #1), two

fea-turess different from the pristine case appear: firstly, the (0 1) diffraction peak of thee irradiated lattice appears as a bright, slightly elongated spot. Secondly, the domainn scattering has a wedge-like shape for low fluences and a rod-like shape ass the fluence increases. When the magnetic field is increased (images #2), the domainn scattering shifts to higher q values, evolving towards the circular shape off the pristine case. There is a clear intensity concentration on the qx axis, nearly att the point corresponding to Miller indices (0 1/2)/ which is more pronounced

Ann XRMS study of ion-beam-patterned a-GdTbFe thin films 95 5 forr the high fluences. For fields where the magnetization reaches 30% of the saturationn value (images #3), both the ring and the (0 1/2) peak intensities de-crease,, although the latter remains the most intense feature. Close to saturation (imagess #4), the ring intensity fades out and only the (0 1) peak remains.

a = 340 nm (middle column, array IV, V, VI): Although the overall behaviour

off the scattering patterns is the same as for the areas I, II and III, the rod-like shapee at nucleation is more pronounced in this series and its intensity again in-creasess with fluence. The (0 1) diffraction peak lies now appreciably closer to thee domain scattering circle and is nearly incorporated in it. Again the intensity maximaa in the domain scattering is highest on the qx axis. For higher magnetic fields,, the evolution towards a ring-like pattern with an accumulation of inten-sityy at (0 1) direction is repeated, more pronounced now than for a = 225 nm.

a - 420 nm (right column, array VII, VIII, IX): The initial rod-like intensity

att nucleation and the (0 1) peak are more pronounced than ever. In this case alsoo the (0 2) and (11) diffraction peaks are visible. For higher magnetic fields, thee scattering ring expands (images #2) and engulfs (images #3) the (0 1) peak, clearlyy because the average domain size coincides with the interdot spacing.

Althoughh the (1 0) reflections are hidden by the beamstop, it seems that thee diffraction patterns do not have a true rectangular structure, as indicated byy the rod like structures such as in the pattern IX-1. This asymmetry can be ascribedd to the FIB writing process.

Imagess like the ones discussed here were taken with over the whole mag-netizationn loop with a small field increment. In order to compare the different areass more precisely, the scattering patterns were angularly integrated, even thoughh such an integration is strictly applicable only to the isotropic pristine areaa but not to the patterned areas, due to their rectangular symmetry. The resultingg curves I(qr) were collected in contour plots, shown in Fig. 5.5. The pristinee area shows the quasi-parabolic dispersion of the scattering maxima as observedd in samples with disordered domain patterns. The patterned areas havee an additional ridge of intensity at fixed qr that is caused by the (0 1) re-flection.. For each I(qr) curve, the domain scattering feature was fitted with a Lorentziann function to obtain the evolution of the intensity maximum ^ M ( B ) withh the magnetic field, indicated in the contour plots by white dots.

96 6 _{CHAPTERR 5}

500 100 150 50 100 150 50 100 150

B(mT) )

Figuree 5.5: Contour plots of the scattered intensity as a function of field B and momen-tumm transfer qr. The panels correspond to the pristine sample (top left) and the nine

ir-radiatedd areas with interdot spacing and ion fluence as specified in the previous figure. Thee scattered intensities are normalized to the maximum domain scattering intensity. Thee positions of the maxima of the domain-scattered intensity IJM(B) are indicated with whitee circles.

Ann XRMS study of ion-beam-patterned a-GdTbFe thin films 97

Thee contour plots clearly show that for the smallest dot spacing the do-mainn scattering does not interact with the dot scattering in q space. Although thee spacing is much smaller than the average domain size, the evolution of the latterr with field is not affected, which seems to indicate that the domains can adaptt to the dots. The fact that the (0 1) reflections are so clear indicates that adjacentt dots are relatively often in domains with opposite magnetization, an interpretationn that is corroborated by the MFM image taken in a perpendicular magneticc field (Fig. 5.3-c).

Thee intermediate dot spacing is clearly closer to the average domain spacing,, and the dot scattering is much stronger in this case, a sign that the localizationn of the domains on the dots is stronger. This is even more so the case forr the 420 nm spacing, which matches the intrinsic domain size at high fields. Forr the highest doses, there is a clear change in the position of the maximum intensityy curve.

Too bring this out more clearly, Fig. 5.6 compares the field dependence of thee domain size obtained from the intensity maxima of the pristine area with thatt of the patterned areas for each of the three lattice spacings. For the 225 nm andd 340 nm spacing (panel a and b), the data match up well with the pristine behaviour.. For the largest spacing of 420 nm (c) we clearly see that for the lowestt dose the field-dependence of the domain size is still identical to that off the pristine sample, while the two higher fluences show a clear lock-in of thee domain size with the interdot spacing at the field where the two become comparable. .

5.4.. Discussion

Thee fact that all irradiated areas show the same hysteresis curve as was measuredd with XMCD, nucleating at the same magnetic field, means that the dotss do not act as low-field nucleation centers. This is clearly different from whatt is seen in another study where ion irradiation is used to mix Pt/Co multi-layerss [151]. However, the dots do have an effect on the position and orientation off the domain wall.

98 8 C H A P T E RR 5

OO 50 100 1500 50 100 1500 50 100 150 B(mT) )

Figuree 5.6: Average magnetic correlation length for the pristine sample (full line) and forr the nine irradiated areas (symbols). The interdot spacing is 225 (a), 340 (b), 420 (c) nm,, indicated in (c) by a dotted line.

off the domains. The positional lock-in can be inferred from the large inten-sityy of the (0 1) Bragg peak which strongly varies with the applied field, while higher-indexx peaks have no or very little intensity. This effect can be interpreted withh the help of the MFM image taken under applied field (Fig. 5.3-c): after nu-cleation,, the reversed domains find it energetically more favorable to include thee irradiated areas, independent of their size or orientation. This causes adja-centt spots to have opposite magnetization which brings out the (0 1) and (1 0) diffractionn spots. Because of the slight anisotropy in the FIB writing process, the (00 1) correlations seem to be more strongly present stronger than the (1 0) ones. Thee orientation of the domains is affected by the asymmetry in the FIB writing process,, and should be removable in future. Finally, he size lock-in is observed whenn the interdot spacing matches the domain size and if the fluence is high enough.. This only happens for the largest spacing and the two highest doses.

5.5.. Conclusions

Thiss study illustrates the strength of focused ion beam patterning in tai-loringg the arrangement of magnetic domains by locally changing the magnetic

anisotropy.. The domain structure of FIB-patterned Gdn.3Tb3jVegs thin films

hass been followed with XRMS and MFM. The MFM data show that in applied fieldss the dots are hosts to the down domains. The high sensitivity of X-ray res-onantt magnetic scattering allowed u s to study the effect of these lattices on the

Ann XRMS study of ion-beam-patterned a-GdTbFe thin films 99 magneticc domain structure over the complete field range. We find that fluences

ass small as 1 ion/nm2 of 30 keV G a+ ions are enough to destroy the perpendic-ularr magnetic anisotropy of the material without changing the film topography.

Wee conclude that the magnetic anisotropy patterning has a strong effect onn the position of the domains, which favour to include at least one irradiated dot.. The strength of the effect scales with the ion fluence. When the typical domainn size approaches a multiple of the interdot spacing, the domain lattice accommodatess to the dot array and is locked to that size over a large field inter-val. .

Iff the original aim of creating controlled domain nucleation centers is to bee reached, even lower doses will have to be used. However, the present system formss a very interesting artificial defect system that could serve as a test bed for thee study domain wall propagation in inhomogeneous samples. Indeed, this workk will be followed up with transmission X-ray microscopy experiments.