Published: August 01, 2011

r 2011 American Chemical Society 3751 dx.doi.org/10.1021/nl201784z |Nano Lett. 2011, 11, 3751–3754

LETTER

pubs.acs.org/NanoLett

Scanning Transmission Electron Microscopy Analysis of GrainStructure in Perpendicular Magnetic Recording MediaFaraz Hossein-Babaei,*,† Robert Sinclair,† Kumar Srinivasan,‡ and Gerardo A. Bertero‡

†Department of Materials Science and Engineering, Stanford University, Stanford, California 94305, United States‡Western Digital Corporation, 1710 Automation Parkway, San Jose, California 95131, United States

The data storage density of magnetic recording media con-tinues to increase steadily.1�6 Further progress requires the

controlled arrangement of smaller magnetic grains in the record-ing medium. Obtaining a desirable nanostructure in the magneticlayer (ML), the topmost active layer in a hard disk layer stack(Figure 1), is of critical importance to the performance of thedevice.7,8 The ML is a continuous nanocomposite thin film ofmagnetic cobalt-alloy grains isolated from one another by anideally nonmagnetic intergranular matrix.1,3,9 To enhance themagnetic properties and maintain the required signal-to-noiseratio in the read and write processes, downscaling of the aver-age ML grain size should be accompanied by tightening their sizedistribution.10 The average grain size in these recording mediahas reached nanometer scales, about 7 nm at present.3,11�14

Cobalt-based perpendicular magnetic recording (PMR) med-ia comprising Co-rich grains of about 7 nm average diameter withstandard deviation of ∼1 nm are manufactured by sputterdeposition of the different device layers on both sides of asubstrate at precisely controlled rates and atmospheric condi-tions. Different underlayers and seed layers are employed toobtain such grain structures in the ML. Figure 1 shows the layerstacking in a typical device schematically. ML is deposited on aRu layer which, like ML, has a hexagonal close-packed (hcp)crystal structure. The Ru seed layer is thought to control the MLnanostructure.4,8,15 The grains in both ML and Ru layer exhibitstrong textures with their crystallographic c axes aligned perpen-dicular to the thin film surface.9,16 Understanding the structuralcorrelation between the ML and Ru layer is important for bettercontrol of the PMR media nanostructures.

Owing to the nanometric feature scales involved and theintended simultaneous observation of the two layers, the trans-mission electron microscope (TEM) is the most suitable tool.

However, discriminating the ML and Ru layer from one anotherusing phase or diffraction contrast, as used in the broad beamTEM imaging mode, is impeded by the overlapping of thestructural features of both layers in the images. Moreover,cross-sectional examination requires specialized and time-con-suming specimen preparation, yielding limited thin area for theanalysis. Accordingly, here we report a novel use of scanningtransmission electron microscope energy dispersive spectrome-try (STEM-EDS) to simultaneously observe the nanostructuresof the ML and Ru layer. The results show a strong grain-to-grainagreement between the two layer structures. The method usedcan be utilized for the simultaneous observation and the deter-mination of compositional and orientational correlations amongconsecutive nanometric layers in general.

The disks studied comprise a 1.5 mm thick Al�Mg substratewith the functional layers shown in Figure 1 sputter deposited onboth sides. A 20 μm thick layer of Ni�P is electrodeposited priorto the sputtering. All the other layers are then grown in a serialarrangement of 20 sputtering chambers. The ML deposited is ofCoCrPt�TiO2 composition.17,18 Plan-view TEM specimenswere prepared using a conventional method:17,19 Each samplewas ground and polished from the back side of the sample. Afterbeing punched into standard 3 mm diameter TEM specimendisks, the samples were further ground using a Gatan-656 dimplegrinder from the back side to a thickness of less than 15 μm at thecenter. The specimens were then ion milled with Ar+ ions at 4�5keV at incident angles of 4�5� using a Gatan-691 precision ionpolishing system. Electron transparent regions near the holes

Received: May 25, 2011Revised: July 28, 2011

ABSTRACT: The key component of a hard disk medium is aCo-based magnetic layer (ML) grown on a Ru seed layer. TheML nanostructure, composed of less than 10 nm grains, isbelieved to be controlled by this seed layer.We successfully usedscanning transmission electron microscopy energy dispersivespectrometry simultaneous composition-based imaging andMoir�e pattern analysis for determining the mutual structuraland orientation relationship between the two layers revealing agrain-to-grain agreement. The method presented here can beutilized for observing structural correlations between consecu-tive polycrystalline thin film layers in general.

KEYWORDS: PMR medium, Co nanoparticle, seeded growth, multilayer nanostructure, STEM-EDS, Moir�e pattern, Runanoparticle

3752 dx.doi.org/10.1021/nl201784z |Nano Lett. 2011, 11, 3751–3754

Nano Letters LETTER

perforated after ion milling in the 10�30 nm specimen thicknessrange were used for TEM studies. These regions included theML�Ru layer interface. The TEM instrument used for theanalysis was FEI G2 F20 Tecnai operated at 200 kV in bothTEM and STEM modes.

Figure 2 shows a bright field (BF) TEM image along with theselected area electron diffraction (SAED) pattern of the observedregion. The hcp crystal structure is confirmed for both ML andRu layer from the diffraction pattern. Owing to the verticalorientation of the c axes in all grains (i.e., the c axes are parallel tothe electron beam), only the plane systems in the [0001] zone,i.e., {hki0}, contribute diffraction rings. Table 1 presents theinterplanar spacings of the primary plane systems determinedfrom the SAED pattern along with those calculated based on

alloy data.20,21 For calculated values, pure Ru for Ru layer and thenominal ML composition (excluding Ti and O which are largelysegregated to the intergranular phase)17 for ML grains wereassumed. The lattice constant, a, of the CoCrPt alloy in theML issmaller than that of Ru, and the slightly larger radius rings in theSAED patterns are, therefore, due to the diffraction from theML.The lattice mismatch between the ML and Ru layer depends onthe ML composition.

For the specified sample, the lattice constants were calculatedfrom the SAED pattern to be 0.271( 0.003 nm for the Ru layergrains (compared with 0.270 nm of bulk pure Ru20) and 0.261(0.003 nm for the ML grains (compared with 0.257 nm linearlycalculated for the nominal Co�Pt�Cr composition ofML grainsfrom the available Co�Pt and Co�Cr data21) after calibrationwith a standard Au nanoparticles specimen. The small differencesbetween the measured and calculated lattice constants areattributed to the incorporation of oxygen and Ti species in MLand Ru layer grains during their sputter deposition, which wasnot considered in the calculations.

The BF TEM image in Figure 2a shows the nanostructure ofthe specimen, where both the ML and Ru layer are present,comprising nanosized crystalline grains embedded in a lightershaded amorphous intergranular matrix.22 The large periodstriations within the grains are due to the Moir�e effect whicharises from the difference in d-spacings of the respective planes inthe Ru layer and ML grains. The Moir�e fringes are consistentwith the in-plane polycrystalline orientations and show that boththin films are contained in the viewing area. By averaging the dataobtained from 60 different grains observed in BF TEM images,the Moir�e pattern period, λM, was measured to be 3.1( 0.3 nm.

Figure 1. A schematic diagram of layers in a CoPtCr�TiO2 alloy-basedPMR medium. All layers with the exception of Ni�P (electrodepositedon the substrate for rigidity) are sputter deposited at precisely controlledconditions onto a 1.5 mm thick Al�Mg substrate.8

Figure 2. A BF diffraction contrast TEM image of the ML (a) in aCoPtCr�TiO2 PMRmedium plan-view specimen showing the Co alloygrain structure and Moir�e fringes in each grain. Both the Ru layer andML are present in this region producing the Moir�e fringes. Theassociated SAED pattern (b) is produced from the contribution of planesystems in the [0001] zones of both the ML and the Ru layer.

Table 1. Measured and Calculated Interplanar Spacings forML and Ru Layer with Uncertainty of about 1% for thePrimary Plane Systems Observed in the SAED Pattern Shownin Figure 2

measured calculated

hkil dRu (nm) dML (nm) dRu (nm) dML (nm)

1010 0.235(3) 0.226(3) 0.234 0.223

1120 0.135(2) 0.130(2) 0.135 0.129

2020 0.117(2) 0.113(2) 0.117 0.111

2130 0.089(1) 0.085(1) 0.0884 0.0842

Figure 3. A high-resolution phase contrast TEM image of the ML in aCoCrPt�TiO2 PMR medium plan-view specimen showing the {1010}lattice fringes and the Moir�e pattern resulting from interference of{1120} planes.

3753 dx.doi.org/10.1021/nl201784z |Nano Lett. 2011, 11, 3751–3754

Nano Letters LETTER

As observed in Figure 2a, this Moir�e fringe periodicity is quitecommon among the grains of this sample. Only a few percent ofgrains examined demonstrate Moir�e patterns of significantlysmaller periods (not seen in Figure 2a), which presumably arisedue to strong diffraction from higher order planes in thesecrystals as the c axis crystallographic texture is not exact.9 Figure 3shows the high-resolution phase contrast TEM image of a grainexhibiting the typical Moir�e pattern period. The dominant latticefringes show {1010} planes. In such phase contrast images, thelattice fringes from the crystalline grains extend a short distancebeyond the grains and overlap in the intergranular regions,making the matrix phase appear crystalline. This artifact arisesowing to the delocalization effect caused by the sphericalaberration of the objective lens of the TEM.23

The orientational relationship between the Moir�e and latticefringes reveals that theMoir�e pattern arises from the interferenceof {1120} planes. Denoting the {1120} interplanar spacings ofRu layer and ML grains by d1 and d2, respectively, the resultingMoir�e interference pattern period, λM, for a parallel plane systemsuch as that assumed for the ML grains and their respective seedgrains at the Ru layer, is24,25

λM ¼ 1g1 � g2

¼ d1d2d2 � d1

ð1Þ

The reciprocal lattice vectors g1 and g2 have magnitudesequivalent to inverses of the respective interplanar spacings. Usingthe {1120} d spacings resulted from the SAED patterns ofML andtheRu layer (seeTable 1), theMoir�e pattern period is calculated as3.5( 0.3 nm similar to the value directly measured from the TEMimages. Presumably Moir�e fringes from {1010} planes are toowidely separated (6.1 nm) to be seen clearly within 7 nm grains.

Energy dispersive spectrometry (EDS) spectra were obtainedfrom regions of specimens thick enough to contain both the MLand Ru layer such as that shown in Figure 2. In these experimentsthe specimen was tilted +15� toward the EDS detector toenhance the detection efficiency of the X-rays emitted. Figure 4shows an EDS spectrum obtained from the specimen over a largearea (800 nm across) and X-ray collection time of 30 s at a countrate of∼3000 s�1. All elements in the sputtered layers present inthe region analyzed show X-ray peaks in the spectrum.

Maps were produced by rastering the electron beam over thespecimen and collecting X-ray signals over the dwelling time foreach pixel point of the map. The parameters of the STEM-EDSanalysis software (gun lens current, beam spot size, pixel size,analysis area, and pixel dwelling time) were selected to obtain EDSelemental maps with a 1 nm resolution. Short dwelling times of 2 sper pixel point, imposed by the nanoscale sample drifts, and thelow brightness of the 1 nm diameter probe limited the number ofpixels in each map to 20 � 30. The collected EDS spectra of thespecimen were used for constructing two-dimensional elemental

X-ray maps based on each element’s KR peak filtered from thespectra. An energy window was manually selected around acharacteristic peak of an element after its identification in thereference spectrum collected over a long period of 30 s; forexample, the 6.8�7.1 keV energy window was used to identifyCo. The maps were produced by measuring the relative con-centration of an element at a pixel point by integrating thespectrum intensity curve over the energy window specified. Thebrightness of each pixel on the map represents the relativeconcentration of the element within that map. The obtainedSTEM-EDS data were analyzed using the software TEM Imagingand Analysis 2.

The elemental maps extracted from a 21� 13 nm2 area on thesample are given in Figure 5. Nanometer-scale features arenoticeable. In the ML, Co and Pt are more concentrated in thegrains while Ti andO are segregated into the intergranular matrixas described by others.17 In the Ru layer map, likewise, the grainsappear brighter than the grain boundaries owing to the formationof voids at grain boundaries due to the high ambient pressureused during the sputter deposition of this layer.26 The agreementin the nanostructures of the two layers is Ru grain to Co alloygrain. This was confirmed in three different regions examined.

Figure 4. Long irradiation time (30 s) and wide area (800 nm diameterirradiating beam diameter) EDS spectrum obtained from the specimencomprising the carbon overcoat, ML, Ru seed layer, and theNi�Wlayer.

Figure 5. The BF TEM image (a) and the elemental maps obtainedsimultaneously for Co (b), Pt (c), Ti (d), Ru (e), andO (f) from a 21 nmwide region of a specimen comprising the ML and Ru layers. Note therelative displacement of the Co grains to the left relative to the Ru seedgrains in the Co and Ru maps shown in (g) and (h) at highermagnification.

3754 dx.doi.org/10.1021/nl201784z |Nano Lett. 2011, 11, 3751–3754

Nano Letters LETTER

Close inspection shows that there is a small displacementbetween the corresponding features in the Co and Ru elementalmaps, which was measured to be about 2.5 nm for all the grainsexamined in this case. This is not due to a displacement of the Ruand Co alloy grains but rather the TEM specimen tilt. Theobserved lateral displacement is consistent with the specimen tiltangle, and the layer thicknesses where examined. Denoting thedistance between centers of the two layers as t, the displacementbetween their features in the elemental maps would be t sin(θ),where θ is the actual tilt angle of the specimen examined. Thepredicted displacements of�1.4, +0.7, and +3.9 nmmatched themeasured values (see Figure 6) at actual tilt angles of �9�, +6�,and +26�, respectively, with t being 9 nm. The total specimenthickness based on this value is 23 nm equivalent to thethicknesses of the carbon overcoat, the ML, and part of the Rulayer needed to examine the structures of the two layers simul-taneously.

We reported observing a grain-to-grain structural agreementbetween the ML and the Ru seed layer in a PMR medium andconcluded that, at the utilized fabrication conditions, each grainin the Ru layer acts as seed for the growth of a single Co grain inthe ML. The grain structural correlation was observed by thesimultaneous mapping of the two layers using a novel STEM-EDS technique to distinguish the two structures based on theirdifferent chemical compositions. SAED and Moir�e pattern an-alyses revealed a close crystallographic relationship between thecorresponding grains inML and Ru layer. Themethod presentedis applicable for the simultaneous nanostructural analyses ofmultilayer thin film stacks.

’AUTHOR INFORMATION

Corresponding Author*E-mail: farazhb@stanford.edu, farazhb@gmail.com.

’REFERENCES

(1) Piramanayagam, S. N.; Srinivasan, K. J. Magn. Magn. Mater.2009, 321, 485–494.(2) Judy, J. H. J. Magn. Magn. Mater. 2001, 235, 235–240.(3) Piramanayagam, S. N. J. Appl. Phys. 2007, 102, 011301.(4) Weller, D.; McDaniel, T. Media for Extremely High Density

Recording. In AdvancedMagnetic Nanostructures; Sellmyer, D. J., Skomski,R., Eds.; Springer: New York, 2006; pp 295�324.(5) Piramanayagam, S. N.; Tan, H. K.; Ranjbar, M.; Wong, S. K.;

Sbiaa, R.; Chong, T. C. Appl. Phys. Lett. 2011, 98, 152504.

(6) Li, W. M.; Chen, Y. J.; Huang, T. L.; Xue, J. M.; Ding, J. J. Appl.Phys. 2011, 109, 07B758.

(7) Bertram, H. N.; Zhou, H.; Gustafson, R. IEEE Trans. Magn.1998, 34, 1845–1847.

(8) Judy, J. H. J. Magn. Magn. Mater. 2005, 287, 16–26.(9) Kwon, U.; Sinclair, R.; Velu, E. M. T.; Malhotra, S.; Bertero, G.

IEEE Trans. Magn. 2005, 41, 3193–3195.(10) Weller, D.; Moser, A. IEEE Trans. Magn. 1999, 35, 4423–4439.(11) Srinivasan, K.; Piramanayagam, S. N.; Chantrell, R. W.; Kay,

Y. S. J. Magn. Magn. Mater. 2008, 320, 3036–3040.(12) Lister, S. J.; Thomson, T.; Kohlbrecher, J.; Takano, K.;

Venkataramana, V.; Ray, S. J.; Wismayer, M. P.; de Vries, M. A.; Do, H.;Ikeda, Y.; Lee, S. L. Appl. Phys. Lett. 2010, 97, 112503.

(13) Jung, H. S.; Kwon, U.; Kuo, M.; Velu, E. M. T.; Malhotra, S. S.;Jiang, W.; Bertero, G. IEEE Trans. Magn. 2007, 43, 615–620.

(14) Wang, J.-P.; Shen, W.; Hong, S. IEEE Trans. Magn. 2007,43, 682–686.

(15) Bertero, G. A.;Wachenschwanz, D.;Malhotra, S.; Velu, S.; Bian,B.; Stafford, D.; Wu, Y.; Yamashita, T.; Wang, S. X. IEEE Trans. Magn.2002, 38, 1627–1631.

(16) Hirayama, Y.; Honda, Y.; Kikukawa, A.; Futamoto, M. J. Appl.Phys. 2000, 87, 6890–6892.

(17) Risner, J. D.; Nolan, T. P.; Bentley, J.; Girt, E.; Harkness, S. D.,IV; Sinclair, R. Microsc. Microanal. 2007, 13, 70–79.

(18) Chen, C.; Sakurai, R.; Hashimoto, M.; Shi, J.; Nakamura, Y.Thin Solid Films 2004, 459, 200–202.

(19) Risner, J. D.; Sinclair, R.; Bentley, J. J. Appl. Phys. 2006, 99,033905.

(20) Pearson’s Handbook of Crystallographic Data for IntermetallicPhases, 2nd ed.; Villars, P., Calvert, L. D., Pearson, W. B., Eds.; ASMInternational: Materials Park, OH, 1991.

(21) Buschow, K. H. J.; van Engen, P. G.; Jongebreur, R. J. Magn.Magn. Mater. 1983, 38, 1–22.

(22) Nolan, T. P.; Risner, J. D.; Harkness, S. D., IV; Girt, E.; Wu,S. Z.; Ju, G.; Sinclair, R. IEEE Trans. Magn. 2007, 43, 639–644.

(23) Erni, R. Aberration-Corrected Imaging in Transmission ElectronMicroscopy: An Introduction; Imperial College Press: London, 2010; pp73�77.

(24) Williams, D. B.; Carter, C. B. Transmission Electron Microscopy:A Textbook for Materials Science; Williams, D. B., Carter, C. B., Eds.;Springer: New York, 1996; Vol. 3, pp 445�448.

(25) Creath, K.; Schmit, J.; Wyant, J. C. Optical Metrology of DiffuseSurfaces. In Optical Shop Testing, 3rd ed.; Malacara, D., Ed.; Wiley:Hoboken, NJ, 2007; pp 757�762.

(26) Thornton, J. A. J. Vac. Sci. Technol., A 1986, 4, 3059–3065.

Figure 6. Co and Ru maps obtained at different sample tilt angles fromthe same sample region demonstrate different lateral displacementswhich are indicated with the relative positions of the bright vertical lines.The actual tilt angles were determined frommatching the measured andcalculated displacements.