CECILIAANDERSSON ExploringtheMagnetismofUltraThin3dTransitionMetalFilms 175 DigitalComprehensiveSummariesofUppsalaDissertationsfromtheFacultyofScienceandTechnology

Full text

(1)Digital Comprehensive Summaries of Uppsala Dissertations from the Faculty of Science and Technology 175. Exploring the Magnetism of Ultra Thin 3d Transition Metal Films CECILIA ANDERSSON. ACTA UNIVERSITATIS UPSALIENSIS UPPSALA 2006. ISSN 1651-6214 ISBN 91-554-6554-4 urn:nbn:se:uu:diva-6836.

(2)

(3)

(4)

(5)

(6)

(7)

(8)

(9)

(10) !

(11) " #$ #%%& %' ( ) ) ) * +

(12)

(13) ,

(14)

(15) -

(16) * .

(17)

(18) /* #%%&* -

(19) "

(20) ) +

(21) 0 +

(22)

(23) " 1* ..

(24)

(25) *

(26)

(27)

(28)

(29) 2(* 3 $2 * * 4567 8 9(($9&(($9$* 4

(30)

(31)

(32) ) 9

(33) 0

(34)

(35) )

(36)

(37)

(38) 1 7

(39) / )* :9 "

(40) / ;:"/<

(41) )

(42) ; <

(43) ; <

(44) =

(45)

(46) 9* -

(47) )

(48)

(49) ) :9 >

(50)

(51) "

(52) 5

(53) ;:>"5<

(54)

(55)

(56)

(57) -

(58) :9 .

(59) 1

(60) 5 ;-:.15<* 6 )

(61) :"/

(62) 1?.; %%< ,

(63)

(64)

(65) )

(66) 9

(67) 9 )9

(68) 1 @

(69) ,* . , 0%%A

(70) (92

(71) ;"< ) 1* !

(72)

(73)

(74) =

(75) 9 )9

(76)

(77)

(78) , #(B

(79)

(80)

(81) ;(B< )

(82)

(83) * -

(84) :9 .

(85) 1

(86) 5 ;-:.15<

(87)

(88) ) & ".

(89) @ @9@ * 1

(90) 9 )9

(91) ) * +

(92)

(93)

(94) ) .?/ ?. 9

(95) )

(96)

(97) ) /

(98) . @

(99) * .

(100)

(101) )

(102)

(103)

(104)

(105)

(106) )

(107)

(108) * .

(109) 9 ,

(110)

(111)

(112) )

(113)

(114) 9

(115)

(116) 9 )9

(117)

(118) =

(119)

(120) , ) 0%%A #%%A* 4

(121) 9

(122)

(123)

(124)

(125)

(126)

(127) )* + .?/

(128) )

(129)

(130) .

(131)

(132) )

(133)

(134) * 4

(135) 9

(136) 9 ) 1

(137) 7 ,

(138)

(139)

(140)

(141)

(142)

(143) 1

(144) 7 @

(145) * .

(146)

(147)

(148) 9

(149)

(150)

(151)

(152)

(153) ) 0 )

(154)

(155)

(156)

(157) ) ) * +

(158)

(159)

(160) )

(161)

(162)

(163)

(164)

(165)

(166)

(167)

(168) *

(169) ) )

(170)

(171) 1 7 / . /. !

(172) "! # $ %&'! ! ()*%+,+ ! C / .

(173)

(174) #%%& 4557 &( 9&# $ 4567 8 9(($9&(($9$

(175) '

(176)

(177) ''' 9&D0& ; '??

(178) *@*? E

(179) F

(180) '

(181)

(182) ''' 9&D0&<.

(183) Till mojan och far.

(184)

(185) List of papers. This thesis is based on the following papers. Reprints were made with permission from the publishers. I. Structure and magnetism on in-situ ultrathin epitaxial films: XMCD and EXAFS on Fe/Ag(100) A. Hahlin, C. Andersson, J. Hunter Dunn, O. Karis, and D. Arvanitis Surf. Sci. 532-535, 76, 2003. II. Structure and magnetism for ultra-thin epitaxial Fe on Ag(100) A. Hahlin, C. Andersson, J. Hunter Dunn, B. Sanyal, O. Karis, and D. Arvanitis (Accepted in Phys. Rev. B) III. On the temperature driven spin reorientation in Au/Co/Au A. Hahlin, E. Holub-Krappe, H. Maletta, C. Andersson, O. Karis, J. Hunter Dunn, and D. Arvanitis (In manuscript) IV. Spin reorientation in Au/Co: in-situ prepared ultra-thin films C. Andersson, T. Konishi, E. Holub-Krappe, O. Karis, H. Maletta, J. Hunter Dunn and D. Arvanitis (In manuscript) V. Magnetism and electronic structure of in-situ prepared Au/Co/Au C. Andersson, T. Konishi, E. Holub-Krappe, O. Karis, J. Hunter Dunn, H. Maletta and D. Arvanitis (In manuscript) VI. Influence of ligand states on the relationship between orbital moment and magneto-crystalline anisotropy C. Andersson, O. Eriksson, L. Nordström, O. Karis and D. Arvanitis (In manuscript) VII. Vanishing magnetic interactions in ferromagnetic thin films J. Hunter Dunn, O. Karis, C. Andersson, D. Arvanitis, R. Carr, I.A. Abrikosov, B. Sanyal, L. Berqvist and O. Eriksson Phys. Rev. Lett 94, 217202, 2005 iii.

(186) iv VIII. Manipulating the magnetisation axis in Fe and Ni bi- and tri-layer films on Cu(100) C. Andersson, J. Hunter Dunn, R. Carr, O. Karis and D. Arvanitis (In manuscript) IX. Exploring non-collinear magnetic coupling in bi- and tri-layer ultrathin films of Fe and Ni on Cu(100) C. Andersson, J. Hunter Dunn, O. Karis and D. Arvanitis (In manuscript) The following papers are not included in the thesis as they go beyond the scope of the thesis. Interface magnetic properties of La0.7 Sr0.3 MnO3 thin films P. E. Roy, C. Andersson, R. Gunnarsson, O. Karis, Z. Ivanov and P. Svedlindh J. Magn. Magn. Mater. 272-276, 1207, 2004. Final state effects in the X-ray absorption spectra of La0.7 Sr0.3 MnO3 ˚ O. Wessely, P. E. Roy, D. Aberg, C. Andersson, S. Edvardsson, O. Karis, B. Sanyal, P. Svedlindh, M.I. Katsnelson, R. Gunnarsson, O. Bengone and O. Eriksson J. Magn. Magn. Mater. 272-276, 1780, 2004.. Comments on my own participation The research area of experimental physics contains many facets including the service of experimental equipment, instrumentation development, sample preparation, carrying through the measurements, and finally analysing and presenting the data for the community. This work is usually not done by one person only. In the work presented here my role has more been active in the later steps, that is to carry through the experimental work and to extract physical information from the experimental data. My contribution in the published work is to some extent reflected by my position in the author list..

(187) Contents. List of papers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Comments on my own participation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 Populärventenskaplig sammanfattning . . . . . . . . . . . . . . . . . . . . . . . 2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Magnetism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1 Magnetic interactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2 Magnetic domains . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3 Thin film magnetism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.1 Temperature dependence . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.2 Magnetic anisotropy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.3 The Bruno approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Experimental techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1 Synchrotron radiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1.1 MAX-laboratory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2 X-ray Absorption Spectroscopy (XAS) . . . . . . . . . . . . . . . . . . . 4.3 Extended X-ray Absorption Fine Structure (EXAFS) . . . . . . . . . 4.4 X-ray Magnetic Circular Dichroism (XMCD) . . . . . . . . . . . . . . 4.4.1 The XMCD sum rules . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.5 Soft X-ray Resonant Reflectivity . . . . . . . . . . . . . . . . . . . . . . . . 4.6 Photoelectron Spectroscopy (PES) . . . . . . . . . . . . . . . . . . . . . . . 4.7 Sample preparation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.7.1 Crystal preparation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.7.2 Film preparation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 Fe/Ag(100) - Structure and magnetism . . . . . . . . . . . . . . . . . . . . . . 5.1 Magnetic properties of ultrathin Fe films on Ag(100) . . . . . . . . 5.2 Structural characterization . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Spin reorientation in Au/Co/Au system . . . . . . . . . . . . . . . . . . . . . . 6.1 Temperature dependence of the spin moment . . . . . . . . . . . . . . . 6.2 The influence of film and cap thickness . . . . . . . . . . . . . . . . . . . 6.3 The orbital moment upon Spin Reorientation . . . . . . . . . . . . . . . 7 Bi- and tri-layer of Fe and Ni . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.1 Manipulating the easy axis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.2 Decoupling between ferromagnetic Fe and Ni . . . . . . . . . . . . . . Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. v. iii iv 1 5 7 7 8 9 9 9 12 13 13 14 15 17 19 21 21 23 24 25 26 27 27 28 33 33 34 35 39 39 39 43.

(188)

(189) Populärventenskaplig sammanfattning. Det finns en historia om en herde som hette Magnés och bodde p˚a Kreta, och som medan han vaktade sina f˚ar vid berget Ida upptäckte att spikarna i hans stövlar och järnspetsen p˚a hans käpp drogs till marken. Nyfiken som han var ville han hitta orsaken till detta s˚a han började gräva i marken och upptäckte att det som drog ner stövlarna och käppen var n˚agra speciella stenar. Materialet i stenar kallar vi idag för magnetit, vilket a¨ r ett naturligt förekommande magnetisk material (Fe3 O4 ). Kanske s˚a a¨ r den här historien p˚ahittad, men man vet att de tidigaste upptäckterna av magnetit och dess magnetiska egenskaper gjordes antingen i det antika Grekland eller i Kina, n˚agra hundra a˚ r förre Kristus. Det sägs ocks˚a att Djingis Kahn, som var härskare o¨ ver mongolerna p˚a 1200-talet, använde n˚agot magnetisk material som kompass för att korsa den Mongoliska o¨ knen. De första vetenskapliga texterna om magnetism var “De Magnete” och publicerades a˚ r 1600 av William Gilbert. Han beskrev det magnetiska fältet kring en bit magnetit och jämförde det med jordens magnetfält, och sa att jorden i sig var en stor magnet.. Figur 1.1: Ett ferromagnetiskt material p˚a atomär niv˚a, med ett magnetiskt moment p˚a varje atom.[1]. S˚a magnetism upptäcktes för längesedan men hur fungerar det egentligen? Alla som har använt magneter t.ex. som h˚allare p˚a kylsk˚apet vet att det upp1.

(190) 2. ¨ KAPITEL 1. POPULARVENTENSKAPLIG SAMMANFATTNING. st˚ar en attraktion mellan kylsk˚apsmagneten och metallen i dörren. Magneter fungerar inte om vi försöker använda dem p˚a glas eller p˚a metaller som aluminium. Det a¨ r uppenbarligen s˚a att olika material uppvisar olika egenskaper d˚a de utsätts för en magnet. Om man sätter tv˚a stavmagneter med lika a¨ ndar mot varandra s˚a skyr de varandra men om man vänder den ena stavmagneten s˚a dras stavmagneterna till varandra. S˚a hur ser d˚a en stavmagnet ut om vi tittar lite närmare p˚a dem? I Fig. 1.1 ser vi en stavmagnet genom ett väldigt. 6 /. Figur 1.2: Elektronernas spinnmoment (S) och orbitalmoment (L) ger tillsammans det atomära momentet. [1]. bra förstoringsglas. Där avbildas varje atom som en magnet. Denna atomära magnet a¨ r resultatet av en eller flera elektroner som a¨ r hopparade i en riktning. De olika atomernas moment (magnetens riktning) kopplas ihop parallellt med varandra till ett gemensamt nettomoment. Denna koppling mellan atomernas magnetiska moment kallas ferromagnetism. De atomära magnetiska momenten best˚ar av ett spinnmoment (S) och ett orbitalmoment (L). Detta avbildas p˚a ett förenklat sätt i Fig. 1.2 där, utöver varje elektrons spinn, elektronens rörelse runt atomkärnan ger upphov till ett orbitalmoment. Nu när vi har en uppfattning om hur magnetsim fungerar, s˚a infinner sig nästa fr˚aga: Vad man använder magnetism till idag? Ja, det första man tänker p˚a a¨ r kanske kylsk˚apsmagneter eller kanske p˚a en kompass, men det finns m˚anga andra användningsomr˚aden. Ett av de omr˚aden, relevant för min forskning, a¨ r magnetisk datalagring. I v˚ara h˚arddiskar idag s˚a används magnetiska tunna filmer. För att utläsa den lagrade informationen behövs mycket känsliga sensorer. Dessa baseras p˚a ett speciellt magnetiskt fenomen den gigantiska magnetoresistiva effekten [Giant Magneto Resistance (GMR)]. Data lagras som magne-.

(191) 3. 5HODWLYHLQWHQVLW\ DUEXQLWV

(192). . D

(193). . 0/)H$J

(194). . PLQRULW\ ;0&'GLIIHUHQFH. . . . . . PLQRULW\ ;0&'GLIIHUHQFH. N. N. . 0/)H$J

(195). PDMRULW\. PDMRULW\. . E

(196). . 0 . 0. 3KRWRQHQHUJ\ H9

(197). 3KRWRQHQHUJ\ H9

(198). Figur 1.3: XMCD spektra av Fe film växt p˚a en enkel kristall av Ag. tiska”ettor och nollor. Min forskning kan användas för att undersöka magnetiska material för framtida datalagringsenheter eller andra nya avancerade teknologiska system. Jag har undersökt de magnetiska och strukturella egenskaperna hos filmer som a¨ r n˚agra atomlager tjocka. Dessa filmer best˚ar av Fe, Ni och Co och a¨ r växta p˚a ytor av välordnade kristaller. Dessa system kan ses som prototyper för framtida lagringsmedia. De magnetiska egenskaperna hos dessa tunna filmer a¨ r väldigt annorlunda jämfört med stora bitar av samma material. Genom att a¨ ndra t.ex. temperaturen eller tjockleken hos den magnetiska filmen eller tjockleken av ett skyddande lager av n˚agon icke magnetisk film ovanp˚a den magnetiska filmen kan vi a¨ ndra riktningen och storleken p˚a magnetismen i filmen. Metoden som jag använt mest i mina studier benämns magnetisk cirkulär dikroism med röntgenljus (XMCD fr˚an engelskans X-ray Magnetic Circular Dichroism). Ett exempel p˚a spektrum, vilket a¨ r resultatet av mätningen med XMCD, fr˚an ett s˚adant experiment visas i Fig. 1.3. För att kunna använda denna metod m˚aste man ha röntgenljus med variabel energi. Detta f˚ar man endast fr˚an en synkrotron. Den enda synkrotronen i Sverige a¨ r MAX-lab i Lund, och det a¨ r dit jag a˚ ker när jag utför mina experiment. Genom att variera energin hos röntgenljus kan man undersöka de magnetiska egenskaperna p˚a ett a¨ mne i taget i t.ex. en legering eller tv˚a filmer med olika magnetiska material ovan p˚a varandra. Riktningen av det magnetiska momentet kan ocks˚a bestämmas med hjälp av XMCD. När röntgenljuset infaller parallellt eller antiparallellt med riktningen av magnetiseringen hos filmen s˚a uppmäter man en stor sig-.

(199) 4. ¨ KAPITEL 1. POPULARVENTENSKAPLIG SAMMANFATTNING. nal medan XMCD-signalen försvinner helt d˚a ljuset infaller vinkelrätt mot det magnetiska momentet. Slutligen s˚a kan man bestämma storleken av magnetismen genom att analysera arean av skillnaden fr˚an de tv˚a spektra i Fig. 1.3 och p˚a s˚a sätt f˚a fram spinn- och orbitalmoment..

(200) Introduction. The first thing that you think about when it comes to magnetism is probably a compass or the magnets you have on your refrigerators, but there are tremendously many other applications of magnetism, and new ones are found frequently. However magnetism is still a phenomena that keep on fascinating people and intensive research is made in this area. An interesting fact concerning the expansion of use of magnets are that the average family in the 1950’s owned two magnets, one in the bicycle dynamo, the other in the windscreen wiper motor of the family car, a similar average family now owns hundreds of them, in all different kinds of motors, in computers, in domestic microwaves ovens etc. The magnetic storage of data is very important for modern life, and widely used in such applications as audio tape, video cassette recorders, computer hard disk, floppy disks, credit cards any many others. The magnetic tape was originally used for analogue sound storage but is now used to store both analogue and digital information and this is no longer limited to tapes which has slow access times. Magnetic discs offer rapid storage and access of ever increasing amounts of information. The first commercial hard disk was made for the RAMAC-computers released by IBM in 1956, which could hold five Mbytes of information which is about two compressed songs in audio format. Since then an tremendous increase in disk capacity have been made. Especially the discovery of the Giant Magneto Resistance (GMR) effect in the 90’s was extremely important to reach the type of capacities found today [2, 3]. Scientists are now developing newer magnetic materials for computers. The disc drive storage system is mechanical and still quite slow, unlike chip (RAM) storage of data which is about 10000 times faster. But whilst discs retain the information when the power is off, chips forget everything without power. The race is now on to produce MRAMs, or magnetic random access memories, which will retain the information permanently. MRAM chips use magnetic rather than electrical structures to store information, so they do not need to be constantly powered to retain data, as current RAM technologies. They are also much faster and less expensive to make than today’s flash memory. According to data storage companies MRAM will be introduced on the market in a nearby future [4]. 5.

(201) 6. CHAPTER 2. INTRODUCTION. MRAMs are built up of magnetic and non magnetic layers, where the structure and interface of the layers have a significant role. In this work the magnetic and structural properties on ultra-thin Fe, Co and Ni films on single crystal substrates has been investigated by polarized soft x-rays, see the papers included in the end of the thesis referred to with roman letters. These systems can be viewed at as prototypes for e.g. the magnetic films used in storage technology. The magnetic properties of these thin film systems are quite different from the bulk system. The direction and size of the magnetization are depending on, e.g., the temperature and the thickness of the film and the cap that covers the magnetic film. The aim of the work is to get better knowledge of the fundamental magnetic properties, which may ultimately be useful in the designs of new advanced technological devices for the future..

(202) Magnetism. Magnetism is a very large and comprehensive research area. I have no ambition to give a full account of all aspects in this chapter, but only to briefly describe the aspects of magnetism that are relevant for the results presented in this work. For a more complete account I instead refer the reader to one of the many textbooks on the subject, e.g. the classic textbook by Chikazumi [5].. 3.1. Magnetic interactions. There are several contributions to the magnetic energy from different interactions within the solid. The magnetic dipole-dipole interaction will cause a demagnetizing field, generated by the magnetic sample itself [6]. The interaction with an external field and the magnetic moments will give a contribution to the total energy that is described by the Zeeman term (−µ0 ∆m · Hext ) [6]. The exchange interaction has an electrostatic origin, but is really a quantum effect since it is a consequence of the Pauli principle. Due to the exchange interaction the individual spin moments in a system will order themselves in various ways. The direct exchange, when the two nearest neighbors has overlapping charge distributions. The indirect exchange, also known as the RKKY interaction, is the interaction of magnetic impurities is mediated by conduction electrons [7]. In the Heisenberg model of ferromagnetism the exchange field gives an approximate representation of the exchange energy [8]: H=−. . Jex si · sj ,. (3.1). where Jex is the exchange integral between the spins located on sites i and j. Based on the sign of Jex the Bethe-Slater curve predicts ferromagnetic order as a function of the interatomic distance and the orbital radius, see Fig. 3.1. Jex > 0 favors a ferromagnetic order and Jex < 0 favors an anti-ferromagnetic order. Jex > 0 for the ferromagnetic 3d transition metals Fe, Co and Ni, but for Mn Jex < 0 and Mn is thus expected to show anti-ferromagnetic order as observed experimentally. 7.

(203) CHAPTER 3. MAGNETISM. Exchange interaction Jex. 8. Fe. Co Ni. 1.5. 2.0 D/d. Mn. D = interatomic distance (Å) d = diameter of d-shell (Å). Figure 3.1: The Bethe-Slater curve shows the exchange integral, Jex , as a function of D/d where D is the interatomic spacing and d the radius of the unfilled d shell.. 3.2. Magnetic domains. There is a natural tendency for a ferromagnetic body, to exhibit magnetic domains in an attempt to minimize the total energy [7]. In Fig. 3.2.a, a single. Figure 3.2: The figure shows different domain configurations; a) a mono domain. b) a sample with two domains. c) a sample divided into four domains, and finally d) closure domains where the demagnetization energy is minimized. (mono) domain is shown. In this configuration the magnetic energy will be high due to a large demagnetization contribution. For the case in Fig. 3.2.b the magnetic energy will be reduced relative to the the value for the mono domain, this by dividing the sample into two domains which are magnetized in the opposite way. There is however a penalty associated with the formation of these domains and that is the formation of the domain walls between the.

(204) 3.3. THIN FILM MAGNETISM. 9. domains. Depending on the magnitude of the exchange interaction and the magnetic anisotropy (see Section 3.3.2 for explanation), the energy contained in the domain wall might be so high that domain formation is not favorable. By further dividing the crystals into N domains, see Fig. 3.2.c where N=4, the demagnetization energy will be decreased about N times. If the domains are arranged like in Fig. 3.2.d the magnetic flux may close within the sample and the demagnetization energy is minimized [9]. Of course this is a very simplified example, often the domains structures are much more complicated. However, domains always lower the energy from a saturated system with high magnetic energy to a domain configuration with lower energy.. 3.3. Thin film magnetism. In thin films, the magnetic properties can be very different from what is found in the bulk. For example, the reduced dimensionality (a film can be considered as two dimensional) imply that many atoms in the film will experience interactions from less neighbors than they would in a bulk sample.. 3.3.1. Temperature dependence. Below a certain temperature, the Curie temperature (TC ), magnetic ordering appears and the spins line up resulting in a spontaneous magnetization, which is for a parallel arrangement described as the ferromagnetic state. As the temperature, T, increases the magnetization decreases until T = TC , when it becomes zero (see Fig. 3.3). Above TC the system is paramagnetic with no macroscopic magnetic ordering. Here the thermal energy overcomes the exchange energy and the spins fluctuate randomly. For thin films the Curie temperature changes according to the finite-size scaling [10], Tc (∞) − Tc (d) = cd−1/ν (3.2) Tc (∞) where Tc (∞) is the Curie temperature for the bulk, Tc (d) is the Curie temperature of a film with thickness d, c is a crystal structure dependent coefficient. ν is the critical exponent which is 1 in the 2D case and 0.705 for the 3D case according to the Heisenberg model [11]. As an example the Curie temperature for Co bulk is 1388K [9] and for a 1.7 ML Co film on W(110) TC is 300K [12].. 3.3.2. Magnetic anisotropy. The magnetic anisotropy can phenomenologically be divided into magnetocrystalline anisotropy and shape anisotropy. The shape anisotropy depends on.

(205) CHAPTER 3. MAGNETISM. 10 1.0. Ms(T)/Ms(0). 0.8 0.6 0.4 0.2 0.0 0. 0.2. 0.4. 0.6. 0.8. 1.0. T/Tc. Figure 3.3: The figure shows a typically behavior for the saturation magnetization as a function of temperature. The curve was obtained theoretically with the mean field approximation, but experimental values would show a similar curve.. the shape of the sample, hence the name. It originates from the demagnetizing field (i.e. dipolar interaction) also driving the formation of magnetic domains. The demagnetizing field is due to poles at the surface. Inside a ferromagnetic sample it is opposed to the spontaneous magnetization direction -N·Ms , where the demagnetizing tensor N depends on the shape of the sample and Ms is the saturation magnetization. This contribution favors an in-plane magnetization for thin films. Let us now imagine that we have a spherical specimen that is homogeneously magnetized. Even though a sphere looks the same from all directions we will generally find that there is a remaining contribution to the anisotropy energy that most be related to the properties of the material itself. This is the magnetocrystalline anisotropy, Ek . In a solid, the magnetocrystalline anisotropy energy is minimized when the magnetization is along a certain crystallographic orientations called the easy directions. The least preferred direction is the hard directions. The easy directions for the ferromagnetic 3d metals are shown in Fig. 3.4. Iron has a body centered cubic (bcc) structure with the easy direction in its principal axis, e.g. (100). Cobalt has a hexagonal close packed (hcp) structure with the easy axis along the c-axis. Nickel has a face centered cubic (fcc) structure with the easy axis diagonal through the cube, i.e. along the (111) direction. For a uniaxial ferromagnet Ek can empirically be written.

(206) 3.3. THIN FILM MAGNETISM. EFF)H. 11. KFS&R. IFF1L. Figure 3.4: The easy direction, indicated by the arrows, for the ferromagnetic 3d transition metals: bcc Fe, hcp Co and fcc Ni.. in a series of powers of sin2 θ [5]: . Ek = K0 + K1 sin2 θ + K2 sin4 θ + (K3 + K3 cos(6φ))sin6 θ + .... (3.3). where K0 , K1 , K2 ... are empirical ”constants”. They are generally found to be very material dependent, but the anisotropy also depends on e.g. temperature. θ is the angle between the magnetization vector and the easy axis, and φ is the azimuthal angle. The anisotropy constants are often separated into a volume (V) and a surface (S) part when dealing with thin films: Ki = KiV +. 2KiS , t. (3.4). where t is the thickness of the film. A contribution to the magnetocrystalline anisotropy is the magnetoelastic anisotropy which can induced by strain in the film [13, 6]. The magnetoelastic contribution to the magnetic anisotropy is especially important when the lattice mismatch between the film and the substrate is large. Due to competition between different anisotropy terms, a spin reorientation transition (SRT) can occur. Since the anisotropies have a temperature dependence, the SRT can be a function of temperature. It is also dependent on the thickness of the film..

(207) CHAPTER 3. MAGNETISM. 12. 3.3.3. The Bruno approach . The difference in orbital moment, (m⊥ l − ml ), between the perpendicular and in-plane magnetic direction is according to Bruno et al. [14] directly linked to the anisotropy energy primarily arising from the presence of the spin-orbit interaction. G ξ ∆ESO = − (m⊥ − ml ), (3.5) H 4µB l where the factor G/H depends on the details of the band structure and ξ is the spin-orbit interaction. This relation is a “zero K” approximation, that should hold best for TTC . This relation has been used to explain the reorientation of the easy magnetization axis in terms of the competing anisotropy energies for example for the Au/Co/Au system investigated by Weller et al. [15]. However in Paper VI we demonstrate that a direct proportionality of the magnetocrystalline anisotropy and orbital anisotropy, as in the Bruno’s formula, cannot generally be expected, it is only valid for special cases..

(208) Experimental techniques. In this chapter I will briefly describe the techniques I have used for the work presented here. All the experimental techniques I present in this thesis are synchrotron radiation based. For more extensive information about the generation and the application of synchrotron radiation and the related instrumentation, I refer to a textbook by Attwood [16]. The methods used here for sample preparation of the ultra thin metal films on single crystals are also presented in this chapter.. 4.1. Synchrotron radiation. For the experiment described in this thesis, Paper I - IX, a synchrotron light source was used. Synchrotron light is generated when electrons (or positrons) traveling at a speed close to the speed of light are accelerated in a magnetic field that alters the trajectory. The electrons (positrons) are stored in a storage ring and to prevent them from colliding with atoms or molecules, the storage ring has to be kept under ultra-high vacuum. There are three types of radiation sources: bending magnets, wigglers and undulators, located in the storage ring. The bending magnet is a dipole magnet at a vertex in the storage ring, and it produces synchrotron radiation in a large range continuous spectrum [17]. Undulators and wigglers consists of regular arrays of alternating Beamline D1011. exit slit grating 1. (a) View from the side light source. mirror 1. mirror 3 end station. mirror 2. (b) View from above mirror 1 light source. mirror 2. grating 1. mirror 3. exit slit. end station. Figure 4.1: The schematic layout of the modified SX-700 monochromator at beamline D1011 at MAX-lab.. 13.

(209) CHAPTER 4. EXPERIMENTAL TECHNIQUES. 14. magnetic fields. These insertion devices produce higher photon intensities and brilliance1 than what is possible to achieve from a bending magnet. The beamline starts at an insertion device or an bending magnet in the storage ring and continues until the end-station and is kept in ultra high vacuum (UHV). In Fig. 4.1 the schematic layout of the D1011 at MAX-lab is shown. The monochromator at the D1011 beamline is a modified SX-700. Mirror 1, a plane-elliptical mirror, focus the beam onto the exit slit in the horizontal plane. Then the light hits mirror 2, a rotatable plane mirror, continuing to grating 1 the dispersive element, which is a rotatable plane grating. Finally the mirror 3, a plane-elliptical mirror, focus the light vertically onto the exit slit.. 4.1.1. MAX-laboratory. MAX-lab is a synchrotron facility in Lund, Sweden [18]. It consists of three storage rings: MAX I (550MeV), MAX II (1.5GeV) and MAX III (700MeV), see Fig. 4.2.. Figure 4.2: Schematic of MAX-laboratory, showing the storage rings MAXI, MAXII and MAXIII . Also showing are existing and future beamlines.. 1. The brilliance is defined as proportional to the ratio of the intensity and the line-width of the x-rays..

(210) 4.2. X-RAY ABSORPTION SPECTROSCOPY (XAS). 15. Beamline D1011 The main part of this work were performed at Beamline D1011 at MAX-lab. This is a bending magnet beamline located at MAX II, see Fig. 4.2. It is a soft x-ray beamline covering the energy range from 30 eV to 1500 eV. The monochromator, as mentioned before, is a modified SX-700 [17] (see Fig. 4.1). D1011 enables magnetic studies because it delivers elliptically polarized light. At this beamline two UHV endstations exists, one after the other. The front endstation consists of two chambers separated by a valve. The sample is placed on a rod which is mounted on a manipulator sitting on top of the upper chamber. This enables the sample to be moved down through both chambers. This mount also allows rotation of the vertical axis and movements in the horizontal directions. The sample can be cooled down to 80K with liquid nitrogen. To what temperature the sample can be heated varies with different setup, e.g. with electron bombardment a W crystal can be heated to 2400K. The lower chamber is the analysis chamber and is equipped with a Scienta SES-200 electron energy analyzer for X-ray Photoelectron Spectroscopy (XPS) and a channel plate detector for X-ray Absorption Spectroscopy (XAS). The upper chamber, the preparation chamber, offers the possibility for Low Energy Electron Diffraction (LEED) studies. There are also possibilities to add other UHV equipment such as evaporators and magnets. If light is let through the front end station it will end up in the back end station. Also here the sample is placed onto a rod which is mounted on a manipulator attached on top of the chamber. This manipulator offers the same possibility as the one on the front system. In this chamber the sample can be cooled down to 20K by means of liquid helium cooling. Also here the maximum temperature is dependent on the sample mount. The chamber is equipped with a channel plate detector for XAS measurements. Coils for magnetization of the sample are mounted from below. The coils have cooling possibilities which enables measurements in applied field, essential for measuring hysteresis loops by means of of X-ray Resonant Magnetic Scattering (XRMS, see Section 4.5). In one of the coils a photo diode is mounted for reflectivity measurements. Also mounted in this chamber is a LEED and the chamber allows for other UHV equipment to be mounted, e.g. a ion sputter gun for cleaning of the sample.. 4.2. X-ray Absorption Spectroscopy (XAS). X-ray Absorption Spectroscopy (XAS) is a technique used to characterize the unoccupied electronic states of the sample of interest [19]. By exciting core electrons to unoccupied electronic states (i.e. above the Fermi level (EF ) for e.g. a metal) information on the absorption cross section µ is obtained. The principle for the measurements is shown in Fig. 4.3, where monochromatic x-.

(211) CHAPTER 4. EXPERIMENTAL TECHNIQUES. 16. rays impinge upon the sample with intensity I0 . By measuring the transmitted light intensity I, the absorption coefficient can be determined for a sample of thickness x by the following relation I = I0 · e−µx. (4.1). Relative intensity (arb. units). This way of obtaining the absorption coefficient only allows very thin samples to be investigated. However, the absorption can also be monitored via secondary processes such as by detecting the electrons released in the decay of the core excited state created in the absorption event or by photons in the decay process. In different energy regimes the X-ray Absorption (XA) process. Fe L3,2. 4. I0. I. 3 EXAFS * 10 2 1. I=I0.e-mx. 0 700. 800. 900 1000 1100 Photon energy (eV). 1200. Figure 4.3: The most straightforward way to extract the absorption coefficient is to measure the intensity of the transmitted light, I, of X-rays shined on the sample with the intensity I0 . In the figure the background corrected µ versus the photon energy is plotted. yields different kinds of information. The sharp features in the XAS spectrum shown in Fig. 4.3 are the L3,2 absorption edges of Fe, reflecting absorption at the 2p3/2 and 2p1/2 core level thresholds. The process can simplistically be viewed as the promotion of a 2p3/2,1/2 core electron to empty states of predominately d-character above EF . The energy region in the vicinity of these edges is known as the Near Edge X-ray Fine Structure (NEXAFS) region. The measurements in Fig. 4.3 were made with linearly polarized light. If circularly polarized light had been used it would have been possible to extract magnetic information from the NEXAFS spectra [20]. This technique is called X-ray.

(212) 4.3. EXTENDED X-RAY ABSORPTION FINE STRUCTURE (EXAFS) 17 Magnetic Circular Dichorism (XMCD) and is described in Section 4.4. The energy region beyond the NEXAFS region is the Extended X-ray Absorption Fine Structure (EXAFS) region, which contains information on the local structure. In Fig. 4.3 the EXAFS signal is enhanced 10 times. This technique is the focus in next section.. 4.3. Extended X-ray Absorption Fine Structure (EXAFS). The region 40-1000 eV above the absorption edge is usually referred to as the Extended X-ray Absorption Fine Structure (EXAFS), see Fig. 4.3. EXAFS is a final state interference effect due to scattering of the outgoing photoelectron wave on to neighboring atoms, manifested in low frequency oscillations in the measured absorption spectrum [21]. The frequency of each resulting EXAFS oscillation depends on the distance between the absorbing atom and the neighboring atom since the photoelectron wave must travel from the absorber to the scatterer and back. Depending on their relative phase they will interfere hn. constructive inteference destructive inteference. Figure 4.4: EXAFS is a interference effect of the outgoing photoelectron with the nearest neighbors, interfering either constructively or destructively.. constructively or destructively, as shown Fig. 4.4. The characteristic EXAFS function, χ(E), is defined as the modulation in the absorption coefficient [21]: χ(E) =. [µ(E) − µo (E)] , µo (E). (4.2). where E is the photon energy, µ(E) is the measured absorption and µo (E) is the atomic background absorption. It is necessary to convert the energy E into.

(213) CHAPTER 4. EXPERIMENTAL TECHNIQUES. 18. the wavevector k via the relation . k=. 2m (E − Eo ), 2. (4.3). 5. (a). 4 3 2 1 0 700. 0.4. 800. 900. 1000 1100 Photon energy (eV). (b). 0.2 k · c(k). Fe on Ag(100) Normal incidence T = 120K. EXAFS x 10. Modulus FT (k · c(k)). Relative intensity (arb. units). to be able to relate χ(E) to structural parameters. E0 is the energy at the inflection point at the absorption edge. Transforming E into k-space gives the. 0.0 -0.2. 1200. 1300. 3 AL 25 AL 1400. (c). 0.4 0.3 0.2 0.1. -0.4 4. 5. 6. 7. 8 9 k (Å-1). 10 11 12. 0. 1. 2. 3. 4 R (Å). 5. 6. 7. Figure 4.5: The figure is showing the three steps in the EXAFS analysis for the 3 and 25 AL Fe films on Ag(100): (a) the EXAFS signal when the atomic background is subtracted; (b) the EXAFS expressed as k ·χ in the reciprocal space; (c) the EXAFS in real space after Fourier transform of the k · χ. following expression, found in Ref. [21], for χ(k) χ(k) =. j. 2 2. Nj Si (k)Fj (k)e−2σj k e−2rj /λj (k). sin(2krj + φij (k)) , krj2. (4.4). where Nj = number of back-scatterers in the j th shell. rj = the distance between the absorbing atom i and the backscattering j th shell in single scattering. Si (k)= the amplitude reduction factor due to multiple excitations, etc.. Fj (k) = the backscattering amplitude function for each scattering path. 2 2 e−2σj k = the Debye-Waller factor in the harmonic approximation..

(214) 4.4. X-RAY MAGNETIC CIRCULAR DICHROISM (XMCD). 19. σj2 = the Debye-Waller parameter accounting for thermal and static disorder. λj = the photoelectron mean free path. e−2rj /λj (k) = the mean free path factor. (2krj + φij (k))= the total phase. φij (k)= the phase shift due to the Coulomb potential of the central atom i and of the backscattering j th shell. Equation (4.4) holds for K- and L1 -edge EXAFS. The EXAFS formula for L3,2 -edges is more complex since the initial p state can adopt either a d- or a s-symmetry in the final state. However, transitions to the d-type final states are favored by a factor of 20 compared to transitions to s final states [21]. Therefore the L3,2 -edge can be treated in the same way as the K- and L1 -edge to a good approximation. The determination of real space structure via EXAFS by resolving the data into individual contributions corresponding to different types of neighbors of the absorbing atom, can be done by Fourier transforming k · χ, as shown in Fig. 4.5. The Fourier transform of k · χ provides a photoelectron scattering profile as a function of the radial distance from the absorber. In this function, the peaks are related to the distance between the absorber and the neighboring atoms. The size (both intensity and width) of the peaks are related to the numbers and types of neighboring atoms, and also to the temperature through the Debye-Waller factor. Complimentary to the EXAFS measurements, simulations can be done using, e.g., FEFF 8.10 software [22]. The FEFF 8.10 calculations are based on ab initio self-consistent real space Green’s functions approach for clusters of atoms and the outputs are the theoretical EXAFS. Cluster-models, that could, e.g., be built by the program ATOMS 3.0 by Bruce Ravel [23], can be used to simulate the experimental spectra. The EXAFS simulations combined with experiments then gives local crystallographic information e.g. the nearest neighbor distance. This means that EXAFS may be applied even to disordered systems. In contrast to this are the more commonly used structure methods as x-ray diffraction and LEED that only gives information about the long range order.. 4.4. X-ray Magnetic Circular Dichroism (XMCD). X-ray Magnetic Circular Dichroism (XMCD) is a experimental technique that in a element specific way determined both the size and the direction of spin and orbital moments [24]. This technique works especially good for investigating the magnetic properties of the 3d transition metals, which are mainly determined by the d valence electrons [25], and has been used in Papers I - IV. The XMCD technique relies on spin selective photon excitation of the 2p3/2,1/2 electrons to the empty 3d band. Circularly polarized light transfer its angular momentum for right helicity (σ + ) or - for left helicity (σ − ),.

(215) CHAPTER 4. EXPERIMENTAL TECHNIQUES. 20. (a). (b). Energy. 600. 3 AL Fe on Ag(100) T=120 Kelvin. 4s 400. EF. s+. hu 2p3/2. Intensity (arb. units). 3d. 200. 0. DAL3. -200. DAL3 minority majority half sum XMCD diff.. 2p1/2 690. 700. 710 720 730 740 Photon energy (eV). 750. Figure 4.6: (a) The XMCD principle were circular polarized x-rays excites, in a spin selective way, the 2p3/2,1/2 levels into the 3d band. (b) The XMCD difference is obtained by subtracting the majority spectra with the minority spectra. When the double-step function is subtracted from the half-sum it reveals the 3d contribution to the spectra.. to a spin-orbit split core electron, from the 2p3/2 or 2p1/2 levels, see Fig. 4.6. Since the 2p state is split by the spin-orbit interaction it is no longer a pure spin state, and the photon angular moment can be transferred both to the spin and orbital degrees of freedom of the excited electron. As the two different helicities transfer opposite angular momenta, photoelectrons with opposite spins are created. For right handed circularly polarized light the transition probabilities for spin-up and spin-down are 62.7% and 37.5%, respectively for 2p3/2 level. For the 2p1/2 initial state the transition probability for spin-up and spin-down are 25% and 75%, respectively [26]. The opposite would be obtained for light with the opposite helicity. The spin moment (ms ) is given by the difference between the number of spin-up (N↑ ) and the number of spin-down (N↓ ) d-holes [27]: ms ∝ (N↑ − N↓ ). (4.5). In contrast to the free atom, the orbital moment (ml ) is most often very small in the solid phase. This is denoted quenching of the orbital moment. Any remaining moment is mainly due to the interaction between the spin and orbital.

(216) 4.5. SOFT X-RAY RESONANT REFLECTIVITY. 21. moment, the spin-orbit interaction.. 4.4.1. The XMCD sum rules. By applying the magneto optic sum rules [28, 29, 24] the spin and orbital moments, normalized per atom, and the ratio between them can be extracted from the data in Fig. 4.7 and Fig. 4.8 using the following relations [24] 3p − 2q 7 + < Tz > r 2 2q = −C 3r 2 = 9 pq − 6. ms = −C. (4.6). ml. (4.7). ml ms. (4.8). where p and q are values from the integral of the XMCD difference spectrum, see Fig. 4.7, and r is a value from the integral from the halfsum, see Fig. 4.8. The constant C is determined by the number of empty d-states (nd ) taken from theoretical calculations and the degree of circularity of the x-rays . The term < Tz > is the expectation value of the intra-atomic magnetic dipole operator [28, 29]. It can be shown that < Tz > ≡ 0 for a high symmetry cubic lattice, ml while it may be non-zero for samples with lower symmetry. The ratio m is s only dependent on p and q which means that only the XMCD difference is needed to get the ratio. This makes it a quantity which has more accuracy than the spin and orbital moments in themselves.. 4.5. Soft X-ray Resonant Reflectivity. Soft x-ray resonant reflectivity, also denoted x-ray resonant magnetic scattering (XRMS), is a technique where the photons are detected after the light has interacted with the sample [30, 31]. The basic principle of XRMS is that xrays irradiated upon the sample at an angel θ, relative to the surface, and the reflected photons are detected at the specular angel 2θ with e.g. a photo diode, see Fig. 4.9. As for XMCD measurements the XRMS spectra dominates by the 2p3/2,1/2 thresholds → unoccupied 3d transitions, see Fig. 4.10. The magnetic contrast is often larger for the XRMS measurements than for the corresponding XMCD signal. Due to the fact that the electron escape depth is small compared to the ˚ and λphoton ∼ 100 photon escape depth for the 3d transition metals (λe ∼ 17 A ˚ A) the XMCD technique is more surface sensitive than the XRMS technique. This fact enable us to investigate the magnetic properties of the surface with XMCD and compare with the bulk properties from the XRMS measurements, this is done in Paper III. Furthermore, since photons are detected instead of electrons in XRMS it is possible to measure in applied magnetic fields. Un-.

(217) CHAPTER 4. EXPERIMENTAL TECHNIQUES. 22. XMCD difference (arb. units). Fe. 20. q. p. -50. 0 -100 -20 -150. -40 -60. -200. -80. -250. 680. 700. 720. 740. 760. Integrated XMCD difference (arb. units). 0. Photon Energy (eV) Figure 4.7: XMCD difference spectrum of Fe, with the determined quantities p and q from the integration of the area of the XMCD difference spectrum.. Intensity (arb. units). Fe. 2500. 300. 2000. r. 200. 1500 1000. 100 500. 680. 700. 720. 740. Integrated intensity (arb. units). 400. 760. Photon Energy (eV) Figure 4.8: XMCD halfsum spectrum of Fe, with the determined quantity r from the integration of the area of the halfsum spectrum..

(218) 4.6. PHOTOELECTRON SPECTROSCOPY (PES). 23. 2q. k q Diode. M. Figure 4.9: Setup for the soft x-ray resonant reflectivity measurements. X-rays (k) are reflected from the sample and detected by a photo diode.. fortunately there does not exist as straight forward sum-rules for obtaining the spin and orbital moments, in the literature, as it does for XMCD. In Fig. 4.11 the hysteresis curves for Fe and Ni are shown. They were obtained by recording the XRMS-signal at fixed photon energies corresponding to L3 (or in principal L2 ) thresholds of Fe and Ni, respectively. The XRMS signal is then recorded while varying the applied magnetic field. This makes it the possible to measure element specific hysteresis loops, which provides a tool to measure the magnetic response for each element separately in e.g. a bi-layer or an alloy. In Fig. 4.11, taken from Paper VII, this has been done for a tri-layer of Fe and Ni, and it reveals that the films, though in contact to each other, are magnetically decoupled since the have different coersive fields. This is a very good complementary method to earlier mentioned XAS based methods. However only in-plane magnetization hysteresis loops can be measured due to very small reflection coefficients at large angles.. 4.6. Photoelectron Spectroscopy (PES). Photoelectron Spectroscopy (PES) technique will only be shortly described here. For a more detailed description see the textbook by Hüfner [33]. The photo electric effect described by Einstein [34] in 1905 is the basis for PES. When a photon with energy hν irradiate the system, photoelectrons will leave the system with characteristic kinetic energies given by the relation Ek = hν − Eb + φ. (4.9). where Ek is the kinetic energy of the photoelectron released from the system. Eb is the binding energy of the energy level the photoelectron was emitted from and φ is the workfunction. Since the binding energies of the electrons.

(219) CHAPTER 4. EXPERIMENTAL TECHNIQUES. 24. ,QWHQVLW\ DUEXQLWV

(220). 0/&R RQ&X

(221). .

(222) YLFLQDO ²

(223) YLFLQDO. . . 3KRWRQHQHUJ\ H9

(224). . Figure 4.10: XRMS spectra for Co for the different polarities, showing a large magnetic contrast.. in different elements are unique, this method is suited for chemical analysis of different samples. This is the reason why the technique also is known as Electron Spectroscopy for Chemical Analysis (ESCA) [35]. Due to low mean free path for electrons with energies in the range 50 - 500 eV, PES is a very surface sensitive technique. This fact makes it ideally suited to study the surface cleanliness. With the appropriate choice of excitation energies, surface contaminations below 0.01 monolayer (ML) can be detected. PES has been used in several papers in this thesis, e.g. to ensure surface cleanness.. 4.7. Sample preparation. Most of the samples investigated in this work were in the form of ultra-thin metal films grown on single crystals. In this section it is described how the single crystals were cleaned and how the films were prepared on the crystals..

(225) 4.7. SAMPLE PREPARATION. 300. 25. 2ML Ni / 2ML Fe/ 8ML Ni on Cu(100). Intensity (Arb. Units). 200 100 0. Ni Fe. -100 -200 -300 -20. 0 Applied Field (G). 20. Figure 4.11: Element specific XRMS based hysteresis loops of Fe and Ni in a tri-layer film grown on Cu. The photon energy was fixed at an energy corresponding to the L3 threshold for each element. (From Ref. [32]). 4.7.1. Crystal preparation. Single crystals has been used as substrates for the ultra-thin films investigated in-situ in this thesis. For these in-situ experiments the substrate cleaning plays a significant role, since one of the main purposes is to have a well controlled experiment. In my work I have dealt with two kinds of cleaning procedures of single crystals. The Ag(100) and the Cu(100) crystals were cleaned through argon ion bombardment followed by annealing. These cycles were repeated until sharp diffraction spots were observed by means of Low Energy Electron Diffraction (LEED) and the PES spectrum showed a clean surface. This indicated a surface with well defined long range crystallographic order. These cleaning cycles were typically performed with 20 minutes of argon sputtering at 2 keV and 10 µA followed by a one minute anneal to 900 K. When the crystal is clean a last sputtering cycle is done at 1 keV and 2 µA for 5 minutes followed by a one minute anneal to 900 K, to ensure a smooth surface. The argon pressure during sputtering varies for different sputtering guns, but a typical value were ∼10−5 mbar. For removing C from W(110) crystals argon sputtering is not efficient nor sufficient. Here another cleaning procedure had to be used. The W(110) crys-.

(226) 26. CHAPTER 4. EXPERIMENTAL TECHNIQUES. tal was prepared by annealing cycles in oxygen atmosphere. Specifically, the crystal was heated to 1600 K in 1·10−7 mbar O2 atmosphere for eight minutes. The crystal was then rapidly heated to 2100 K which was followed by a five minutes cool-down in O2 atmosphere. Finally, the sample was flashed to 1700 K. The purpose of the rapid flash is to remove any adsorbed oxygen. The above outlined cleaning cycle was repeated until PES showed no traces of C and O contaminations and diffraction spots were detected by means of LEED indicating a well defined long range order.. 4.7.2. Film preparation. The 3d metal film were deposited by means of a custom made evaporator. The evaporator was operated using electron bombardment heating. During evaporation the pressure was typically at <5 · 10−9 mbar with an evaporation-rate of approximately 0.5 layer per minute. A schematic picture of the evaporators is shown in Fig. 4.12. A rod of purified metal of Fe, Co or Ni was put on +1.2 kV. e-. e-. eIf. Ir. If e-. e-. Figure 4.12: Schematic drawing of the custom made electron bombardment evaporators used for evaporation of 3d metal films. The metal rod is put on positive high voltage and is then bombarded by emitted electron from the thin W wires. a positive high voltage U, typically 1.2 kV. Two thin W filament positioned on opposite sides of the metal rod were resistively heated and eventually electrons were emitted and accelerated onto the rod. The induced emission current Ir was measured enabling us to determine the heating power. Since the temperature is dependent of the power, this gives us a possibility to control the evaporation rate. In Papers IV - VI we have worked with Co films that were capped with different amounts of Au and for Au another type of custom made evaporators were used. The purified Au was put into a ceramic cup which was resistively heated. The evaporators was typically operated at a rate of 0.5 ML per minute. The evaporation was controlled by measuring the temperature of the Au by a tungsten iridium thermocouple..

(227) Fe/Ag(100) - Structure and magnetism. The magnetic properties of ultra-thin films are much dependent on the structure of the film. In our investigation of in-situ grown ultra-thin Fe films on Ag(100) single crystals we have done a combined study of the magnetism and the local structure of the system. This is the focus of Paper I and II.. 5.1. Magnetic properties of ultrathin Fe films on Ag(100). XMCD measurements, made in Paper I and II, revealed an out-of-plane magnetization for a 3 ML Fe films on Ag(100) and an in-plane magnetization for a 25 ML Fe film on Ag(100), see Fig. 5.1. For the transition region, 6 - 8 ML Fe,. 5HODWLYHLQWHQVLW\ DUEXQLWV

(228). . D

(229). 0/)H$J

(230). . PLQRULW\ ;0&'GLIIHUHQFH. . . . . . . 0/)H$J

(231). PLQRULW\ PDMRULW\. PDMRULW\. . E

(232). ;0&'GLIIHUHQFH. N. N. . 0 . 0. 3KRWRQHQHUJ\ H9

(233). 3KRWRQHQHUJ\ H9

(234). Figure 5.1: XMCD spectra taken at 120 K for (a) 25 ML and a (b) 3 ML thick Fe films on Ag(100). In the inset the experimental geometry is shown.(From Ref. [36]) no remanent magnetization out-of-plane was found. The 6 ML film only exhibited a very small magnetic contrast as it was given an applied magnetic field 27.

(235) 28. CHAPTER 5. FE/AG(100) - STRUCTURE AND MAGNETISM. pulse in-plane. The 8 ML film on the other hand experienced a fully saturated magnetization in-plane. In Table 5.1 the orbital and spin moment for the four films are shown, obtained by applying the magneto-optical sum rules on the XMCD data, partly shown in Fig. 5.1, for the 3, 6, 8 and 25 ML Fe on Ag(100). There is a 125% Fe/Ag(100). ml (µB ). ms (µB ). ml /ms. easy dir.. 3 ML. 0.45. 2.35. 0.19. ⊥. 6 ML. 0.04. 0.37. 0.10. . 8 ML. 0.23. 2.28. 0.10. . 25 ML. 0.20. 2.25. 0.09. . Table 5.1: In the table the orbital (ml ) and spin (ms ) magnetic moments for 3, 6, 8, and 25 ML Fe films on Ag(100) are shown. The 3 ML Fe film show out-of-plane magnetization, whereas the thicker films (8 and 25 ML Fe) show in-plane magnetization. The 6 ML Fe film only shows a minor in-plane magnetization. increase of the orbital moment when the Fe film thickness is decreased from 25 ML to 3 ML. For the spin moment there is no significant change, only a small increase of about 5%. The 6 ML film shows small moments since it is in the spin reorientation transition region. For the somewhat thicker 8 ML film, the magnetization shows bulk-like values. The orbital moment shows a 15% increase compared to the 25 ML film, but only a tiny increase (of 1.5%) of the spin moment. XMCD studies of Fe grown on Cu(100) [37, 38] show a similar increase in the orbital moments going from 25 ML to 3 ML of Fe. In contrast to Fe on Ag(100), an increase of the spin moment is also observed for Fe on Cu(100) [37, 38] at decreasing thickness.. 5.2. Structural characterization. As a complement to the XMCD study, L-edge EXAFS has been performed to understand the structure of the system. Since the lattice mismatch for Ag(100) and the 45◦ rotated Fe is only 0.8%, Fe is expected to initially adopt to the Ag(100) structure and grow in a bulk Fe bcc structure. Our EXAFS measurement indicates that this may not be the case for all the films here. Fig. 5.2 shows the Fourier transformed k-weighted modulus of χ(k) recorded at 90◦ x-ray incidence angle. In Fig. 5.2(a) 3 ML Fe film is plotted together with the.

(236) 5.2. STRUCTURAL CHARACTERIZATION. D

(237). -LVU (N 5VYTHS PUJPKLUJL. A. 0RGXOXV)7 NÃC N

(238)

(239). . 29. -LVU (N 5VYTHS PUJPKLUJL. E

(240) A . 0/. 0/ 0/ . 0/. . B G. D. B . . . . . 5 c

(241). . . . . . . 5 c

(242). D. . . . Figure 5.2: The k-weighted modulus of the Fourier transform χ(k) of the (a) 3ML, 25 ML and (b) 6 ML, 8 ML of Fe on Ag(100). α, β, γ and δ denotes the peak positions.(From Ref. [36]). 25 ML film. The shape of the EXAFS for the thick film can be recognized as bulk Fe bcc structure [37]. For the 3 ML film, the interpretation is more complicated. In Fig 5.3 simulations of the EXAFS are presented. The simulations were performed using FEFF 8.10 [22] code. In the simulations the Fe cell has been expanded from the pure bcc structure to the fcc structure in three steps. During the expansion the nearest neighbor distance was kept constant. When comparing the simulated peak positions for the pure bcc phase with the different peak positions for the 25 ML film we again see proof for a bulk bcc structure. For the 3 ML case it seems a bit more complicated. None of the simulations for the bcc → fcc transformation seem to coincide with the experimental EXAFS of the 3 ML film. It is evident that also FEFF 8.10 simulations for a pure 3 ML bcc film on Ag(100) fails to reproduce the experimental EXAFS. However Canepa et al. [39] report that Fe on Ag(100) already at the second layer starts to form islands, and cannot be simulated as layers. The authors suggest that even if precautions are taken during film deposition, Fe and Ag atoms has exchange position during deposition, leading to combinations of Fe/Ag, Fe/Fe and Ag/Fe structures in the film. In Fig. 5.4 EXAFS simulations for the 3 ML Fe which has been intermixed with Ag at the following concentrations: 50%, 25% and 12.5% for the interface, middle and top layer respectively are shown. The.

(243) CHAPTER 5. FE/AG(100) - STRUCTURE AND MAGNETISM. 30. a. 1.4. a. a ÷2•a. ÷2•a. a. Modulus FT (k · c(k)). 1.2 1.0 bcc. fct. 0.8 bcc (fct c/a=1.41). 0.6. fct c/a=1.21 fct c/a=1.1 fcc (bct c/a=1.41). 0.4 0.2 0.0 0. 1. 2. 3. 4 R (Å). 5. 6. 7. Figure 5.3: EXAFS simulations using the FEFF 8.10 code [22] k-weighted modulus of Fourier transform for different bcc phases.(From Ref. [36]). solid curve represents the weighted sum from all unique Fe sites in the cluster model, whereas the dashed curve is the EXAFS signal arising from an Fe atom in the center of the film. It is clear that these simulations do not reproduce the experimental EXAFS for the 3 ML Fe film fully. However, the dashed curve exhibits the spectral features we observe in the experimental EXAFS data. In ˚ is only present when backscattering from particular, the feature around 4.1 A one or more Ag atoms is considered. This finding tends to support Canepa et al. [39]. The fact that these structures are not clearly visible in the weighted sum indicates that our simulations underestimate the number of Fe sites where Ag atoms are nearest or next nearest neighbors. It is therefore reasonable to suggest that intermixing occurs to an even larger extent than what is proposed in Ref. [39]. Such intermixing between film and substrate leads to a lower coordination for the Fe, and could then also explain the enhancement of the orbital moment for the 3 ML Fe film..

(244) 5.2. STRUCTURAL CHARACTERIZATION. Fe. 0.5. Ag. 0.4 Modulus FT. 31. 0.3 Weighted sum One atomic site. 0.2 0.1 0.0 0. 1. 2. 3. 4. 5. 6. 7. R (Å). Figure 5.4: EXAFS simulations of 3 ML Fe/Ag(100) including interface intermixing. The solid line represent the sum of all scattering paths from unique Fe sites in the model cluster. The dashed line represents scattering from one particular site in the model cluster, where all scattering events include scattering from one or more Ag nearest neighbor.(From Ref. [36]).

(245)

(246) Spin reorientation in Au/Co/Au system. In this chapter the investigation of the reorientation of the easy magnetization axis for the Au/Co/Au/W(110) system is discussed, and how the spin reorientation (SR) effects the spin and orbital moment. These results are taken from Paper III - VI.. 6.1. Temperature dependence of the spin moment. We have performed a temperature dependent ex-situ XMCD investigation of a Au/Co/Au tri-layer system, see Paper III. As a complement to the XMCD data, XRMS measurements have been performed. The sample configuration was 97 ˚ of W(110 ) grown on a sapphire single crystal, followed by 51 A ˚ Au, 21 A ˚ A ˚ Co and finally 20 A Au. In Fig. 6.1 the Co spin moment, ms , is plotted together with the XRMS dichroic contrast as a function of the temperature, for the inand out-of-plane components. At 300 K the Co film has a large ms in-plane, but no out-of-plane component. As the temperature goes down, the in-plane ms decreases. Between 280-260 K the out-of-plane component increases rapidly and at around 220 K it has reached saturation. The in-plane ms decreases until 200 K where it reaches zero. The reorientation transition takes place under a temperature range of about 100 K under which it is possible to stabilize both an in-plane and out-of-plane component of the magnetization. It is not possible to stabilize both at the same time, excluding the possibility of canted magnetization as reported in other cases [40]. The smooth spin reorientation found in Fig. 6.1 has been explained theoretically within the mean-field approximation by Jensen et al. [41]. The main assumption in their approach is to distinguish between the anisotropy in the second order expansion term of the anisotropy energy K2 of the interfaces (K2,S ) and of the interior layers of the film (K2,V ). The stronger decrease of the surface magnetization would then be the foundation for the strong temperature dependence in the magnetic anisotropies through Kα (T ) = Kα (m(T )). Hence, it is the interfaces that is the driving force for the temperature dependent spin reorientation transition. This we could see in Fig. 6.1 where the more surface sensitive XMCD data decrease faster when the temperature is lowered. It indicates that the reorientation takes place at higher temperatures for the upper Au/Co interface, or the upper part of the Co layer, than for the deeper buried Co layers probed by XRMS. 33.

(247) 34. CHAPTER 6. SPIN REORIENTATION IN AU/CO/AU SYSTEM. Figure 6.1: In the Figure the squares and the circles represent the meff s for in-plane and out-of-plane magnetic moment, respectively. The triangles represent the XRMS dichroic contrast at remanence as obtained from the XRMS hysteresis loops. The inset illustrates schematically the electron and photon detection schemes used for XMCD and XRMS respectively.. 6.2. The influence of film and cap thickness. We have varied the Co film thickness and also the thickness of the Au cap of ˚ Au evaporated on a W(110) single crystal, see in-situ grown samples on 50 A Papers IV and V. Interestingly we observed that the phase diagram, presented in Fig. 6.2 for in-situ grown Au/Co/Au films grown on a W(110) single crystal, differs from the ex-situ Au/Co/Au films grown on c-axis oriented W(110) [42]. The moments extracted from the XMCD data taken at room temperature are shown in Table 6.1. There the spin (ms ) moments and orbital (ml ) moments for three different sample preparations are shown. They indicate that a much lower thickness of Co is needed for the SR to occur for our in-situ grown films on a single crystal substrate than for the ex-situ grown samples [42]. ˚ film is lower, 1.61µB , than the We observe that the spin moment of the 5 A moment obtained upon Au capping, at 1.93µB . It is interesting to note that this value is close to the one observed for the slightly thicker bare Co film of ˚ For the 5A ˚ film the anisotropy of the Co film is expected to be enhanced, 7 A..

(248) 6.3. THE ORBITAL MOMENT UPON SPIN REORIENTATION. 35. $X&R$X:

(249) VLQJOHFU\VWDOOLQH . ([SHULPHQWDOGDWD 5RRPWHPSHUDWXUH SDUDOOHOPDJ SHUSHQGLFXODUPDJ. $XWKLFNQHVV c

(250). . \\. . >. . . . &RWKLFNQHVV c

No results found