<%BANNER%>

Magneto-Structural and Magneto-Optical Studies of Prussian Blue Analogs


PAGE 1

MAGNETO-STRUCTURAL AND MAGNETO-OPTICAL STUDIES OF PRUSSIAN BLUE ANALOGS By JU-HYUN PARK A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2006

PAGE 2

Copyright 2006 by Ju-Hyun Park

PAGE 3

For Asako

PAGE 4

iv ACKNOWLEDGMENTS In the course of experimental studies I benefited greatly from the numerous discussions and assistances fr om fellow colleagues. First and foremost, I would like to thank Professor Daniel Talham, Professo r Young-Duk Huh, Dr. Jeff Culp, and Franz Frye for designing and fabricati ng the materials studied in this dissertation. They were enthusiastic chemists, from whom I had the privilege of learning magneto-chemistry and supramolecular structures. I also would lik e to thank Professor Yoonseok Lee for his great advices in experimental techniques and physics. He guided me through the many projects, especially the entir e process of dilution refriger ator operation, from which I learned many experimental techniqu es and the view of physicist. Every member of the Department of Physics instrument shop has been extremely helpful. Especially, I would like to tha nk Marc Link, Ed Storch, and Bill Malphurs for their great craftsmanship. Their expertise and casual discussions always helped throughout the designing process. I also would like to thank members in the Cryogenic Services, Greg Labbe and John Graham for th eir great support in cryogenic matters. As members in the same lab, I would like thank to James Maloney, Sara Gamble, James Davis, and Norman Anderson for their support and useful discussions in physics. With them, I truly enjoyed working in the lab a nd experienced great American customs. Finally, I would like to thank my Ph.D. advise r, Professor Mark Meisel for his guidance and encouragement. From him I learned not only the physics but also the example of being a professional scientist. Furthermore, in all aspect s, he always supported and

PAGE 5

v inspired me when I was dealing with everyda y life in a foreign country, and for that I truly thank him.

PAGE 6

vi TABLE OF CONTENTS page ACKNOWLEDGMENTS.................................................................................................iv LIST OF TABLES.............................................................................................................ix LIST OF FIGURES.............................................................................................................x ABSTRACT.....................................................................................................................xiii CHAPTER 1. INTRODUCTION........................................................................................................1 1.1 Ni-Fe(CN)6 Films...............................................................................................2 1.2 Co-Fe(CN)6 Films..............................................................................................4 1.3 Co-Fe(CN)6 Powders.........................................................................................6 1.4 Other Magnetic Systems....................................................................................7 2. EXPERIMENTAL TECHNIQUES..............................................................................8 2.1 Configuration of MPMS Hardware...................................................................9 2.1.1 Sample Rod..............................................................................................9 2.1.2 Probe......................................................................................................11 2.1.3 Console and Computer...........................................................................14 2.1.4 Principle of DC Magnetization Measurement.......................................14 2.2 DC Magnetization Measurement Procedure....................................................16 2.3 DC Magnetization Data Analysis....................................................................19 2.3.1 Curie-Weiss Law and Dimer Model......................................................19 2.3.2 History Dependent Magnets..................................................................25 2.5 Sample Packing and Background Consideration.............................................30 2.5.1 Sample Packing......................................................................................32 2.5.2 Background Consideration.....................................................................35 2.6 Summary and Future Direction........................................................................37 3. PHYSICAL PHENOMENON AND THEORY.........................................................39 3.1 Photoinduced Magnetism in Prussian Blue Analogs.......................................39 3.1.1 Overview................................................................................................39 3.1.2 Initial Observation and Description.......................................................39

PAGE 7

vii 3.2 Charge Transfer Induced Spin Transition........................................................49 3.3 Summary..........................................................................................................57 4. MAGNETIC STUDY OF EVOLVING STURUCTURE (MONO, BI, AND MULTILAYER OF FILMS)......................................................................................58 4.1 Synthesis of Ni-Fe(CN)6 Films........................................................................58 4.2 DC Low Field Magnetization Measurements..................................................60 4.3 AC Field Magnetization Measurements..........................................................72 4.4 Magnetic Evolution versus Structural Evolution.............................................81 5. ANISOTROPIC PHOTOINDUCED MAGNETISM OF PRUSSIAN BLUE ANALOG FILMS.......................................................................................................87 5.1 Synthesis of Co-Fe(CN)6 Films.......................................................................88 5.2 Photoinduced State Magnetism........................................................................90 5.3 Dipolar Field Model.........................................................................................98 6. CHARGE TRANSFER INDUCED SPIN TRANSITION IN PRUSSIAN BLUE ANALOG.................................................................................................................104 6.1 Experimental Details......................................................................................105 6.2 Experimental Results.....................................................................................106 6.3 Discussion and Future Directions..................................................................108 7. HETEROSTRURED LAYERED MAGNET...........................................................113 7.1 Heterostructured Layered Films....................................................................114 7.2 Conceptual 2D Magnetic Nanometer Island..................................................121 7.3 Summary and Future Directions....................................................................127 8. SUMMARY AND FUTURE DIRECTIONS...........................................................129 8.1 Ni-Fe(CN)6 Films...........................................................................................129 8.1.1 Summary..............................................................................................129 8.1.2 Future Directions.................................................................................130 8.2 Co-Fe(CN)6 Films and Powders....................................................................132 8.2.1 Summary..............................................................................................132 8.2.2 Future Directions.................................................................................133 APPENDICES A. CAPACITANCE MEASUREMENT.......................................................................136 B. LOW TEMPERATURE TRANSPORT MEASUREMENT...................................140 B.1 Configuration of Dilution Refrigerator and Sample Mount..........................140 B.2 Operation of Dilution Refrigerator................................................................145

PAGE 8

viii B.3 Summary and Future Directions....................................................................149 C. ORIGIN SCRIPT FOR MAGNETIC QUANTITIES..............................................151 LIST OF REFERENCES.................................................................................................153 BIOGRAPHICAL SKETCH...........................................................................................159

PAGE 9

ix LIST OF TABLES Table page 2-1. Zero-field-cooled magnetiz ation measurement procedure......................................182-2. Field-cooled magnetiza tion measurement procedure..............................................182-3. Magnetization vs. magnetic field measurement procedure.....................................183-1. Chemical compostion of NaxCoy[Fe(CN)6] z H2O samples.....................................50 3-2. Valence states of NaxCoy[Fe(CN)6] z H2O samples at 290 K..................................50 4-1. Characteristic values from DC magnetization measurements.................................824-2. Characteristic values from AC magnetization measurements.................................82

PAGE 10

x LIST OF FIGURES Figure page 2-1. Schematics of MPMS (Magnetic Property Measurement System)......................... 10 2-2. Schematics of the impurity filter and liquid helium transfer...................................13 2-3. Second-derivative pickup coil and centering magnetic signal.................................15 2-4. Magnetization signals of the holders........................................................................17 2-5. Magnetic plots of various types of magnet..............................................................20 2-6. Susceptibility times temperature simulations of dimers..........................................23 2-7. Field dependent magnetization simulations of dimer model...................................24 2-8. Field-cooled (fc) and zero-field -cooled (zfc) magnetization processes...................26 2-9. Thermal remnant magnetization of RbxNiy[Cr(CN)6]z m H2O film..........................29 2-10. A homemade MPMS optic insert rod.......................................................................31 2-11. Various MPMS sample packing methods................................................................33 2-12. Simulation of weak paramagnetic sample in gelcap holder.....................................36 3-1. Temperature dependent photoi nduced and dark magnetizations.............................41 3-2. Time dependent photoinduced magnetization..........................................................42 3-3. Processes of photoinduced magnetization and demagnetization.............................44 3-4. Complete cycle of photoinduced magnetization and demagnetization....................47 3-5. Magnetizations of a series of NaxCoy[Fe(CN)6] z H2O compounds............................53 3-6. T vs. T data of K0.6Co1.2[Fe(CN)6]4H2O powder upon cooling and warming......55 3-7. M vs. H data of K0.6Co1.2[Fe(CN)6]4H20 powder...................................................56 4-1. Amphiphilic pentacyanoferrate buildi ng unit and 2D two-dimensional grid..........59

PAGE 11

xi 4-2. Sketches of the monolayer, bilayer, and multibilayer..............................................61 4-3. Schematic of Fe-CN-Ni square grid (top view).......................................................62 4-4. Temperature dependent magneti zations of 150-multibilayer film...........................64 4-5. The field dependent magne tizations of 150-multibilayer film at T = 2 K................65 4-6. Temperature dependent magne tizations of bilayer film...........................................67 4-7. The field dependent magnetiz ations of bilayer film at T = 2 K...............................68 4-8. Temperature dependent magne tizations of monolayer film.....................................70 4-9. The field dependent magnetiza tions of monolayer film at T = 2 K.........................71 4-10. Temperature dependent AC sus ceptibilities of 150-multibilayer film.....................73 4-11. Result of Arrhenius law fittings to monolayer, bilayer, and multibilayer................74 4-12. Result of Vogel-Fulcher law fitting on multibilayer film........................................76 4-13. Temperature dependent AC sus ceptibilities of monolayer film...............................77 4-14. Temperature dependent AC su sceptibilities of bilayer film.....................................79 4-15. Comparison of AC data at 17 Hz (mono, bilayer, and multibilayer).......................80 4-16. Dipole field produced by a magnetized disk (R~30 )............................................84 5-1. SEM images of film 1 (a) and film 2 (b)..................................................................89 5-2. Photoinduced magnetization of film 1.....................................................................91 5-3. Photoinduced magnetization of film 2.....................................................................93 5-4. Photoinduced magnetization of film 2.....................................................................94 5-5. Frequency dependent AC susceptibilities of Co-Fe(CN)6 films..............................96 5-6. Vogel-Fulcher law fitting to Co-Fe(CN)6 films.......................................................97 5-7. Schematic description of the spin conf igurations of the domains in the film..........99 5-8. Photoinduced magnetization of film 2 when HE > HD...........................................100 5-9. Field dependent photoinduced and da rk state magnetization of film 2..................101 5-10. Photoinduced magnetization at T ~ 4.5 K and HE = 20 T......................................102

PAGE 12

xii 6-1. Susceptibility time temperatur e as function of temperature...................................107 6-2. Low temperature magnetizations measur ed at three differe nt cooling states........109 6-3. Magnetizations of three different magnetic states..................................................110 7-1. An alternating planar struct ure of Prussian blue analog........................................117 7-2. Schematics of synthesizing Ni -Cr and Ni-Cr / Co-Fe films...................................119 7-3. Temperature dependent magnetizations of Ni-Cr and Ni-Cr / Co-Fe films...........120 7-4. Schematic of LB film generation...........................................................................123 7-6. Deposition of nanometer LB islands on substrate..................................................125 7-7. Deposition of nanometer pa rticles to substrates.....................................................126 A-1. The schematic view of capacitance measurement setup........................................137 A-2. The change in capacitance a nd magnetic property of DMACuCl3........................138 B-1. The configuration of low temper ature transport measurement setup.....................143 B-2. Sample mount for transport measuremnt...............................................................144

PAGE 13

xiii Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy MAGNETO-STRUCTURAL AND MAGNETO-OPTICAL STUDIES OF PRUSSIAN BLUE ANALOGS By Ju-Hyun Park May 2006 Chair: Mark W. Meisel Major Department: Physics The magnetic properties of molecule-based magnets, especially in confined or altered geometries, are of great interest in modern magnetic devices. Two different but related molecule-based magnets, namely Ni-Fe(CN)6 and Co-Fe(CN)6 Prussian blue analogs, were generated. Langmuir-Blodgett and/ or sequential deposition fabrication of thin films provided opportunities to st udy dimensional effects on fundamental magnetism. More specifical ly, a series of monolayer, bilayer, and multilayer of ferromagnetic Ni-Fe(CN)6 films were generated, and the magnetic evolution accompanied by the structural evolution is reported herein. In addition, when Co-Fe(CN)6 was generated as a thin film, unusual photoinduced magnetism was observed. In our investigati on, the photoinduced magnetization of the film increased or decreased due to the orientation of the film with respect to the external magnetic field. This interesting effect provides a new ma gnetic switching mechanism and is reported herein. For some of the Co-Fe(CN)6 Prussian blue analogs, the phenomenon of charge

PAGE 14

xiv transfer induced spin trans ition have been reported. Our study of the unusual spin transition process that depends on the cooling rate is also presented he rein. Finally, in the pursuit of better molecule-based functional magnets, the fabr ication method and magnetic results of heterostructured layered films of Ni-Cr / Co-Fe (CN)6 are presented.

PAGE 15

1 CHAPTER 1 INTRODUCTION The main topic of this dissertation fo cuses on the characterization and physical aspects of the molecular magnetic systems base d on Prussian blue an alogs. Historically, Prussian blue is known as the first artificial pigment originating from the use in German army uniforms. It was accidentally found in 1704, when beef blood was boiled in a strong basic solution. Despite of its long history, the chemical structure of Prussian blue (FeIII 4[FeII(CN)6]3H2O) was not identified until the earl y 1970s [1]. In general, the Prussian blue and its analogs have face-cente red-cubic (fcc) struct ures, and the generic formula can be written as AjMk[M (CN)6]l n H2O,1 where M and M are bridged by CN, and respectively located in the center and at th e vertices of the octahe dral structure. The interstitial alkali metal A occupies some of the tetrahedral sites. Hereafter, the generic form of the Prussian blue anal og will be referred as MM (CN)6. Recently, due to its superb magnetic characte ristics, the family of Prussian blue compounds has received attenti on in molecule-based magnets [1 79]. For example, the compound V[Cr (CN)6]0.86.8H2O was the one of the first molecular magnets whose TC exceeded room temperature [6]. Other compound, K0.2Co1.4[Fe(CN)6].9H2O, exhibited a photoinduced magnetism, in which the increases or decreases of the magnetization were controlled by light [38]. These interes ting results came, however, mainly from experiments performed on bulk ma terials and, thus, may not re flect the properties of the 1 This formula contains a interstitial al kali metal A, transition metals (M and M ) of different valence states, and the waters that replace [M (CN)6] to make overall charge neutral.

PAGE 16

2 material confined to reduced dimensions In addition, the technological demand on modern magnetic devices requires physica lly smaller systems, so the constituent materials must be studied in restricted phys ical dimensions. Consequently, studying the dimensional and structural effects of the magnetic systems seems a natural interest, and this interest motivated the research presented in this dissertation. Within the Prussian blue family, Ni-Fe(CN)6 and Co-Fe(CN)6 systems were selected as research compounds Briefly, the bulk Ni-Fe(CN)6 is a ferromagnetic system with the TC ~ 24 K [15], and the bulk Co-Fe(CN)6 is a photoinduced magnet as mentioned earlier [38]. One of the main ch allenges was to reduce the dimensions of the compounds, and this task requires technique s to manipulate the structures at the nanometer scale. For this challenge, the Langmuir-Blodgett technique [9 13] and the sequential deposition method [13, 14] were utilized for Ni-Fe(CN)6 and Co-Ni(CN)6 respectively. Both the Langmuir-Blodge tt and sequential deposition methods are chemical approaches of building thin solid films from layer-by-layer deposition. General reviews and specific uses of these methods ha ve been presented el sewhere [9 14], and the possibility of using these methods to fabric ate heterostructured la yered films, as well as magnetic nanometer islands, are discussed in Chapter 7. 1.1 Ni-Fe(CN)6 Films Using Langmuir-Blodgett techniques, Ni-Fe(CN)6 films were fabricated in monolayer, bilayer, and multibil ayers motifs. This series of systems was designed to model two-dimensional (2D), quasi-2D, and th ree-dimensional (3D) magnetic systems [9, 11]. The resultant monolayer film contains a 2D network of face centered squares of Ni-Fe(CN)6 on a substrate. In additio n, at the interface to the air, Fe ions are terminated by amphiphilic tails. The bilayer film is composed of two monolayers of Ni-Fe(CN)6

PAGE 17

3 networks placed on the substrate. In a side view and starting from the bottom, the bilayer film contains a base substrate, amphi philic tails, two la yers of Ni-Fe(CN)6 networks, and the terminating amphiphilic tails. The mu ltilayer film was generated by repeating the bilayer fabrication process, and therefore, contains multiple stacks of bilayers, in which each bilayer is separated by others via two amphiphilic tails (~ 35 ). In a local magnetic point of view, the bridging CN provides the ferromagnetic superexchange interaction between the NiII ( S = 1) and FeIII ( S = 1/2) ions, whose separation distance is ~ 5 The TC of the 3D bulk material was reported as ~ 24 K with an exchange constant J ~ 14 K [15]. However, the unit of FeIII-CN-NiII studied in different structural environments yielded different experimental J values. Furthermore, recent computational results suggest the J values of FeIII-NC-NiII units vary according to their bonding distances and angles, and in certain situations, the J value can be even negative [16]. In other words, a subtle difference in the local structures can significantly modify the magnetic properties of material. In addition to the J values, the TC values of the system are also predicted to be changed according to the local environment such as the number of the nearest magnetic neighbors. For example in the previously studied ferromagnetic Ni-Cr(CN)6 compounds, the TC values changed from 53 K to 90 K as the number of the magnetic neighbor increased from 4 to 6 [1]. Thus, the magneto-structural effects must be elucidated to understand the fundamental magnetism and the potential device applications. Chapter 4 presents the magnetic properties of monolayer, bilayer, and multibilayers of Ni-Fe(CN)6 LB films that were investigated using DC and AC magnetic property measurement system (MPMS) in the temp eratures between 2 K and 300 K and in

PAGE 18

4 magnetic fields up to 5 T. In addition, the high field dependent magnetization was studied up to 30 T using the vibration samp le magnetometer (VSM) at the National High Magnetic Field Laboratory (NHMFL). In th e low field measurements, the films were studied in both parallel and perpendicular orientations with respect to the external magnetic fields ( HE) in the frequency ranges from DC to 1 kHz. The main discussion in Chapter 4 is focused on the correlation between the magnetic and structural properties of the films as they evolve from the 2D to 3D system. 1.2 Co-Fe(CN)6 Films As mentioned earlier, Co-Fe(CN)6 Prussian blue anal ogs showed striking photoinduced magnetism.2 In 1996, Sato and coworkers discovered that the magnetization of the bulk compound K0.2Co1.4[Fe(CN)6].9H2O increased under irradiation with red light [38] This enhanced magnetization remained nearly constant at 5 K, even when the light was off. To make things more in teresting, after the photoinduced magnetic state has been formed, the magnetization of the system returned to nearly the initial state by irradiation of bl ue light or by thermally cycling the material to ~ 150 K. In the dark state, the compound K0.2Co1.4[Fe(CN)6].9H2O is a ferrimagnetic material with a TC of ~ 16 K. As a result of red light irradiation, the TC of the compound increased to 19 K, and the magnetization at T = 5 K and HE = 5 T increased by 10% compared to its dark state value. For this photoinduced magnetic system, we utilized a sequential deposition method to generate a variety of Co-Fe(CN)6 films. The original motivation was to construct 2 The phrase photoinduced magnetism is often used in the field of semiconductors, and this terminology has an association with a sp ecific process. In this dissertation, the phrase photoinduced magnetism is used in a more general sense to refer any change in magnetic behavior during, or as a result of, irradiation.

PAGE 19

5 magnetic films that exhibit a fast res ponse to the light by minimizing the photon attenuation. In a conve ntional bulk Co-Fe(CN)6, the light tends to at tenuate, and thus it takes a considerable amount of time to sw itch photoinduced magnetization. However in the course of the investigation, we have discovered an unusual anisotropic photoinduced effect. Briefly, we have obs erved an increase or decrea se of photoinduced magnetization controlled by the orientation of the film with respect to HE [50, 51]. This unique effect has never been previously observed and the pot ential extensions to optically controlled magnetic devices are significant. Furtherm ore, the area of photoinduced magnetic films in Prussian blue analogs had not been activ ely developed prior to our investigations. Thus, our investigation on Co-Fe(CN)6 films gave us the opportuni ty to be at one of the frontiers in the research. In Chapter 5, anisotropic photoinduced magnetism in Co-Fe(CN)6 films is presented. For a comparison study, using se quential deposition methods two similar but different films of RbjCok[Fe(CN)6]l n H2O were generated [51]. The difference in the films lies in the arrangement of the domains within the samples. For example in film 1, the particles that cons titute the powder are randomly deposited, givi ng rise to a bulk-like texture. On the other hand in film 2, the pa rticles are more uniformly arranged parallel to the films, resulting in a quasi-2D texture wi thin continuous patches of approximately 600 nm, as verified by scanning electron mi croscopy (SEM) images. The photoinduced magnetic properties in the bulklike film were observed to be different, especially with respect to the anisotropic photoinduced magnetization, from the 2D-like film. More specifically, in the 2D-like film, the magnetizat ion increased under irradiation of visible light when the film was placed parallel to the HE. However, when the film was placed

PAGE 20

6 perpendicular to the HE, the photoinduced magnetization decreased The main discussion of Chapter 5 is devoted to the mechanis m of this unique anisotropic photoinduced magnetism. 1.3 Co-Fe(CN)6 Powders In addition to the photoinduced magnetism, the compound Co-Fe(CN)6 was reported to exhibit an interesting charge-tra nsfer-induced spin transition (CTIST) [46, 57, 58]. In 2002, Shimamoto and coworkers di scovered that the magnetization of a microcrystalline NaxCoy[Fe(CN)6] n H2O powder sample decreased sharply at temperature ~ 180 K upon cooling and increased sh arply at ~ 220 K upon warming when x = 0.37, y = 1.37, and n = 4.8 [57]. This hysteric, first-order, spin transition resembles the properties of spin crossover (SC) materials and can be applied to the magnetic memory devices due to its spin control characteri stics [80]. As in the case of photoinduced magnetism, the CTIST process occurs in a cooperative ma nner and shows sharp spin transitions. However, the phenomenon is strongly influe nced by the subtle differences in the chemical compositions and structures. For example, when the Co/Fe ratio changed from 1.37 to 1.52, the CTIST disappeared and the spin state of Co was locked in the high spin state at all temperatures. Furt hermore instead of Na, when ot her interstitial alkali metals were used, the phenomenon and the condition of CTIST seemed to be restricted [57]. Therefore, studying the magneto -structural effect is esse ntial to understand the CTIST phenomenon. During our investigation of the photoindu ced magnetism, an interesting CTIST phenomenon was observed in KxCoy[Fe(CN)6] n H2O with x ~ 0.6, y ~ 1.2, and n ~ 4. In usual CTIST and SC phenomena, a rapidly cooled sample bypasses the transition from the high temperature (HT) phase to the low te mperature (LT) phase and will be trapped in

PAGE 21

7 the HT phase even at low temperature [46]. Upon slow warming, the sample eventually releases all the HT phase en ergies at certain temperatur es and follows the LT phase warming curve. However in our warming investigation, the magnetization of rapidly cooled K0.6Co1.2[Fe(CN)6]4H2O sample became smaller than that of the regular LT phase. This new LT phase can be achie ved only by rapid cooling and has never been heretofore observed. Therefore in Chapter 6, we re port the magnetization measurements of K0.6Co1.2[Fe(CN)6]4H2O powder in its three different stat es (i.e. slowly cooled state, rapidly cooled state, and rapidly cooled/warme d state). The main discussion in Chapter 6 is an attempt to describe this new LT phase quantitatively and introdu ce recent theoretical developments on phase tr ansitions in Co-Fe(CN)6 Prussian blue analog systems [78, 79]. 1.4 Other Magnetic Systems Along with Prussian blue analog material s, other magnetic systems were also investigated. Although the expe rimental techniques and the results are interesting, the physical nature of the studied materials doe s not represen t the main theme of this dissertation. Therefore, some of the appe ndices are dedicated these techniques and results. In Appendix A, a capacitance measuremen t [81] and magnetization of the S = 1/2 magnetic chain material (CH3)2NH2CuCl3 [82 90], otherwise known as DMACuCl3 are presented. This unique magnetic chain co mpound seems to possess two similar but different alternating magnetic ch ains parallel to each other, and shows interesting field dependent magnetic information. In Appendix B, an investigation of the low temperature transport properties of Ca2xSrxRuO4 are presented. The main focus on this material was to explore Fermi-liquid behavior at the critical concentration x = 0.5 [91].

PAGE 22

8 CHAPTER 2 EXPERIMENTAL TECHNIQUES This chapter focuses on the methods and an alysis techniques used to probe the magnetic properties of materials. For the macroscopic magnetic measurement, a Quantum Design MPMS (Magnetic Property Measurement System) was used. The MPMS is a computer controlled magnetomete r, whose detecting ability is enhanced by SQUID (Superconducting Quantum Interference Device) electronics. Two customized models (MPMS-5S and MPMS-XL) of co mmercial SQUID magnetometers from Quantum Design, Inc., San Diego, CA were used for the DC and AC magnetic measurements. The model MPMS-5S is an ol der system that is equipped with a 5 T longitudinal superconducting magnet and with options including AC measurement, magnet reset, environmental magnetic shield, and degauss shield. The model MPMS-XL is provided with a 7 T longitudinal supe rconducting magnet but possesses none of the options of the MPMS-5S. However, the MPMS-XL has an additional continuous impedance tube, which allows the system to hold a temperature continuously even below 4.2 K. The detailed specifications and rele vant technical notes of the MPMS can be found in the MPMS manuals [94, 95] and on the companys web site.3 This chapter first introduces the confi guration of the MPMS hardware and the principle of the DC magnetic measurement in Section 2.1. Second, a typical procedure for the DC magnetic measurements will be pr esented in Section 2.2, and the simple data analysis strategy will be given in Section 2.3. Third, the construc tion and properties of 3 http://www.qdusa.com

PAGE 23

9 the homemade fiber optic sample holder (FOSH) will be described in Section 2.4. Before summarizing, Section 2.5 review s the various ways of packing and holding sample and outlines the treatment of the background magnetic signals arising from the holders and sample rods. Finally, Section 2.6 summarizes and introduces the fu ture projects of constructing multi-purpose sample rod for the MPMS. 2.1 Configuration of MPMS Hardware In general, our MPMS-XL consists of six main parts: a sample rod, a probe, a longitudinal superconducting magnet, a liquid helium dewar, a console cabinet, and a computer, as schematically shown in Figure 2-1. 2.1.1 Sample Rod A typical sample rod provides a place to m ount a sample and can be inserted into the probe at standard operati ng temperatures (1.9 K ~ 400 K). A long thin stainless steel tube (39 inch) is connected to a short quantal loy (silicon copper all oy) tube (8 inch) to make the body of sample rod.4 The sample rod goes through a slide seal assembly, which engages to the probe head and allows the samp le rod to move in the longitudinal direction while maintaining an air leak tight envir onment on the sample side. The sample is usually placed inside a transparent beverage straw, and the straw is fixed to the end of quantalloy side of the sample rod. Alt hough the beverage straw provides a sufficient housing for most samples, the auto tracki ng function of the MPMS, which compensates the thermal contraction of sample rod, was or iginally calibrated for quartz sample tube. 4 The long stainless steel tube extends from ro om temperature down to the sample space. In the sample space near the sample detecti on area, the quantalloy tube was used. The stainless steel was chosen to maximize the mechanical rigidity and minimize the thermal heat leak from room temperature to the samp le space. The quantalloy tube was used to minimize the extrinsic magnetic contribution to the detector. The both tube have outer diameters of ~ 0.12 inch and the inner diameters of ~ 0.10 inch.

PAGE 24

10 Figure 2-1. Schematics of MPMS (Magnetic Prope rty Measurement System). In general, our MPMS-XL consists of six main parts: a sample rod, a probe, a longitudinal superconducting magnet, a liquid helium dewar, a console cabinet, and a computer connected to the internet. liquid helium dewar sample space cooling annulus quantalloy tube pumping cables electric cables sample rod longitudina l superconducting magnet slide seal assembl y pickup coil stainless steel tube vacuum sleeve probe impedance tube copper jacket and heater thermometer and heater straw sample thermometer liquid helium console pump magnet power control PID thermometry gas control magnetic measurement computer GBIB Multi-View INTERNET GPIB cable

PAGE 25

11 As a consequence, there is a small drift in sample position with respect to the magnet center, when the straw was used as a sa mple housing, even when the auto tracking function is enabled. When the straw was us ed, a typical sample center shift between 310 K and 2 K is ~ 0.1 cm, which is not negligible when studying the absolute magnetization of a spatially small sample. 2.1.2 Probe The probe contains many parts, such as the probe head, the sample space, a cooling annulus, the liquid helium re servoir, primary and secondary impedance tubes, thermometers and heaters, a detection pic kup coil, and the SQUID electronics. Briefly, the main function of the probe is to provide the computer controlled temperature (within 0.5%) environment for the sample and to measure the SQUID enhanced magnetic signals from the sample. The probe is sitting in the liquid helium dewar and regulates the temperature of the sample space by withdraw ing the liquid helium through the impedance tubes to the cooling annulus, which surrounds the sample space concentrically as shown in Figure 2-1. The liquid helium in the cooling annulus is further manipulated by the PID (Proportional, Integral, and Derivative) te mperature controller and by adjusting the pumping power to deliver the desired temperatur e to the sample space. The heat transfer between the cooling annulus and the samp le space is meditated by the conduction through a copper jacket that shields the sample space. The sample space is filled with low pressure helium gas (~ 0.1 mbar at room temperature) to pr ovide heat exchange between the sample and the wall of the sample space. Since the temperature control mainly re lies on the helium flowing through the impedances, the regulation of the amount wit hdrawn from the dewar is critical. The worst case will be realized if no helium is av ailable to pass from the dewar to the cooling

PAGE 26

12 annulus. Occasionally, this unexpected inci dence happens due to the plugging in the impedance tubes. The exact causes of the pl ugging are not certain but can be related to impurities such as ices (e.g. water ice and ni trogen ice) that nucleated to block the impedance tubes. For this reason, great care is needed when transferring the helium from the transport dewar to the system dewar. In order to filter the impurities from the transport helium dewar during the transfer, a homemade filter, which attaches to the end of the transfer line, was built. Figure 2-2 show s a schematic view of the homemade filter. The filter was assembled by placing charcoal particles in an existing stainless steel porous filter. Wool-like stainless mesh was also inserted to prevent the charcoal particles from moving. Before transferring liquid he lium, the filter was warmed to ~ 330 K or higher, while blowing dry helium gas through it The process was repeated two or three times to remove water contaminants in the charcoal filter. If the impedance tubes become completely plugged, temperature control is lost. In this unfortunate case, the w hole system has to be warmed by boiling the liquid helium, and the impedance tubes must be cleaned. Most of time, if ice blocks the impedance, it will melt and unplug the impedance when the system is warmed to room temperature. However in some cases, impurity, such as resi dual oil, blocks the impedance tubes and does not melt upon warming. In this case, the impedance tubes must be cleaned by blowing helium gas through the cooling annulus pumping line, or the impedance tube itself has to be replaced with a clean one. On the other hand when the impedance is partially plugged, the control of the temperature is partially enabled. The quick solution for this partial plugging is to set the pr obe temperature to 300 ~ 310 K for an extended period, usually overnight, expect ing the impurity ice to melt and to be pumped. However,

PAGE 27

13 when this quick solution fails, the whole syst em has to be warmed as in the case of complete plugging. Figure 2-2. Schematics of the impurity filter an d liquid helium transfer. In order to filter the impurities from the transport helium dewar, a homemade filter, which attaches to the end of transfer line, was built. The filter was assembled by putting charcoal particles in an existi ng stainless steel porous filter. The stainless mesh wool was also inserted into the short tube connecting the filter to prevent the charcoal particles from moving. Be fore transferring liquid helium, the filter was warmed to ~ 330 K or higher, while blowing dry helium gas. The process was repeated tw o or three times to remove water contaminants in the charcoal filter. transport liquid helium dewar MPMS liquid helium dewar impurity filter liquid helium transfer line to helium recovery line pressurized external helium gas ( P ~ 2 PSI) stainless steel mesh charcoal particles porous stainless steel filter stainless steel liquid helium transfer line Clean LHe Dirty LHe

PAGE 28

14 2.1.3 Console and Computer The pumping tubes and electric cable s from the probe and the magnet are connected to the console cabinet. The consol e cabinet consists of a pump, gas handling system, a temperature controller, a magnetic measurement unit, and a power source for the magnet. The main role of the console cabin et is to provide electronically controlled pumping power to the probe for temperature regul ation as well as electric controls to the probe and the magnet. The computer is c onnected to the console cabinet via a GPIB (General Purpose Interface Bus) cable and operates all activities of the console using Multi-View, a LabVIEW based computer prog ram that can be controlled locally and remotely via the internet. Once the sample is placed, all other standard activities of MPMS, such as sample positioning, rampi ng magnetic field, changing temperature, magnetic measurement, and reco rding data, can be controlle d using Multi-View program. 2.1.4 Principle of DC Magnetization Measurement To measure the DC magnetization of the samp le, the sample is first placed near the bottom of the pickup coil, which is a single superconducting coil wound by one clockwise turn at the top pos ition, and then by two counter-clo ckwise turns at the middle position, and finally by one clockwise turn at the bottom position, as shown in Figure 2-3 (a). This type of coil is called a second-derivative coil and the main purpose of winding in different directions is to cancel the uniform external magnetic field ( HE) from superconducting magnet of MPMS and other stray fields in the lab. In principle, as a sample moves along the axis of the dete ction pickup coil to the top position, an electromotive force (EMF) is induced in the pickup coil. This induced EMF is proportional to the sample magnetization, and th e MPMS electronics detect the amplified EMF signals using SQUID electronics as th e sample moves along the pickup coil. The

PAGE 29

15 detected signal is then fit into the calculated model curve [96] to give the actual magnetic moments of the sample. Figure 2-3 (b) shows a typical detected signal and the fit as a Figure 2-3. Second-derivative pickup coil and centering magnetic signal. The Secondderivative pickup coil (a) is a sing le superconducting coil wound by one clockwise turn at the top position then by two counterclockwise turns at the middle position and finally by one clockwis e turn at the bottom position. In principle, as a sample moves along the ax is of the detection pickup coil to the top position, the electromotive force (E MF) is induced to the pickup coil. This induced EMF is proportional to the sample magnetization and the MPMS electronics detect the amplified EMF signals using SQUID electronics as the sample moves along the pickup coil as show n in (b). The sample, in this case moved 4 cm along the pickup coil from the near bottom of the coil (0.0 cm position in (a)) at T = 298 K with HE = 100 G. For this detection 16.04 mg of K0.6Co1.2[Fe(CN)6]4H2O powder was used. ~ 3 cm 01234 -2 0 2 4 VoltagePosition (cm) 0.0 cm 0.5 cm 2.0 cm 3.5 cm 4.0 cm sample sample rod superconducting second-derivative pickup coil measured fit HE = 100 G, T = 298 K M = 5.94 10-5 emu G (a) (b)

PAGE 30

16 function of the sample position. In this cas e the sample, moved 4 cm along the pickup coil from the near bottom of the coil (0.0 cm position in Figure 2-3(a)) at T = 298 K with HE = 100 G. For this de tection, 16.04 mg of K0.6Co1.2[Fe(CN)6]4H2O powder was used. 2.2 DC Magnetization Measurement Procedure A typical DC magnetization measurement invol ves two steps. The first step is to measure magnetization of the sample holder on ly and the next is a measurement of the sample in the holder. The sample in powder or microcrystalline form is usually packed into the holder such as gelatin capsule (gelcap ) or plastic can (cappe d polyethylene vial). The diamagnetic background signals from these holders are measured and then subtracted from the total signal of the sample and the holder. The diamagnetic signals from different holders are similar but not iden tical maybe due to the unequal numbers of electronic centers in polymers [97]. Therefore, the background signal must be measured individually and independently before the sa mple measurement, especially for a weak magnetic sample, whose magnetic moment is comp arable to that of the holder. Figure 24 shows typical temperature and field depende nt magnetic measurements of the plastic can and the gelcap with their pictures. Since the magnetic pic kup coil and the sample moving distance have finite lengths of ~ 3 cm and ~ 4 cm (to 12 cm) respectively, a short sample and a symmetric (along the horizontal plane) sample packing are desired. The sample size dependent magnetization was studi ed previously by Jackson and coworkers [98] and the result shows that in order to minimize the sample size dependency, the optimal value of the sample volume is between 10 mm3 and 15 mm3. Other useful sample mounting consideration can be found in the Reference [99].

PAGE 31

17 Figure 2-4. Magnetization signals of th e holders. The sample, in powder or microcrystalline form, is usually packed into holder such as gelatin capsule (gelcap) or plastic can (capped pol yethylene vial). The diamagnetic background signals from these holders ar e measured and then subtracted from the total signal of the sample in the holder. The figure shows the typical temperature and field dependent magnetic measurements of the plastic can and the gelcap with their pictures. 050100150200250300 -15 -10 -5 0 02468 -10 -8 -6 -4 -2 0 gel-cap canM (10-5 emu G) T ( K ) HE = 1 kGT = 2 K M (10-3 emu G)H (T)

PAGE 32

18 Table 2-1. Zero-field-cooled magnetization measurement procedure. Step Procedure 1 Center the sample and measure the magnetization at 300 K. 2 Cool the sample without the magnetic field to the lowest temperature. 3 Apply a suitable measuring magnetic field. 4 Center the sample by measuring an d adjusting the positional dependent magnetic signal as shown in Figure 2-2 (b). 5 Measure the magnetization upon warming the sample. Table 2-2. Field-cooled magneti zation measurement procedure. Step Procedure 1 Center the sample and measure the magnetization at 300 K. 2 Cool the sample with the magnetic field to the lowest temperature. 3 Center the sample by measuring an d adjusting the positional dependent magnetic signal as shown in Figure 2-2 (b). 4 Measure the magnetization upon warming the sample. Table 2-3. Magnetization vs. magnetic field ( M vs. H ) measurement procedure. Step Procedure 1 Center the sample and measure the magnetization at 300 K. 2 Cool the sample without the magnetic field to the desired temperature. 3 Apply a small magnetic field for centering purpose. 4 Center the sample by measuring an d adjusting the positional dependent magnetic signal as shown in Figure 2-2 (b). 5 Measure the magnetization upon sweeping the field. For an unknown molecule-based magnetic material, typical magnetization measurement sequence involves a zero-fieldcooled (zfc), field-cooled (fc), and magnetization vs. field ( M vs. H ) measurements. These zfc, fc, and M vs. H sequences are usually the standard measur ements to characterize the overall magnetic properties of unknown materials. Variations of these standard measur ements are also performed frequently in order to probe specific aspects of a magnetic property of a sample. Typical procedures to perform the zfc, fc, and M vs. H measurements are listed in Table 2-1 to 2-3.

PAGE 33

19 2.3 DC Magnetization Data Analysis After the magnetization measurement, the MPMS records values of raw magnetization and its associated uncertainty. From these va lues and the information of the sample, all relevant quantities, such as DC susceptibility ( ), effective moment ( eff), susceptibility multiplied by temperature ( T ), and inverse susceptibility (1/ ), can be calculated. To generate these values from MP MS data, a script for ORIGIN, a scientific plotting program was written and listed in Appendix C. The DC susceptibility ( = M / H ) is close to a true differential susceptibility ( diff = dM / dH ) only when the M is a linear function of magnetic field ( H ), which is the case for the high temperature or low field magnetization of paramagnetic samples. 2.3.1 Curie-Weiss Law and Dimer Model When the magnetic data are plotted accordi ng to final quantities, a simple magnetic analysis is possible from the tre nd of the graphs. Generally, the T vs. T graph is useful to extract the temperature dependent magnetic information of the sample. According to the Curie-Weiss law, can be described as = C / (T ), (2.1) for ferromagnet when > 0 and T > TC, and for antiferromagnet when < 0 and T > TN, where the C is a Curie constant, is a Weiss temperature, TC is a Curie temperature, and TN is Nel temperature. Therefore if the material is paramagnetic, i.e. = zero, then T becomes equal to the Curie constant C at all temperature. When the material is ferromagnetic, the T graph will increases quickly around TC, and the 1/ graph will have positive x -intercept at TC. The antiferromagnetic behavior is characterized by decreasing T values or a negative x -intercept value extrapolated in a 1/ vs. T graph. A special, but often encountered case of an antiferromagnet is ferrimagnetic material. This ferrimagnet

PAGE 34

20 Figure 2-5. Magnetic plots of various types of magnet. (a) Inverse susceptibility of a paramagnet. (b) Inverse susceptibilit y and spontaneous magnetization of a ferromagnet. (c) Inverse susceptibility of an antiferromagnet. (d) Inverse susceptibility and spontaneous magn etization of a ferrimagnet. (e) susceptibility times temperature of ferromagnet, ferrimagnet, paramagnet, and antiferromagnet. usually contains antiferromagnetically interac ting spins with unequal spin values so that when the magnet is ordered its magnetization graph resembles the superposition of TN 0 1 / 1 / 0 T TC 1 / M 0 T T TC 1 / M 0 T (a) (b) (c) (d) ferromagnet ferrimagnet antiferromagnet paramagnet T T ferrimagnet ferromagnet antiferromagnet paramagnet (e) 0

PAGE 35

21 ferromagnet and antiferromagnet. The characteristic trend of T vs. T graph of ferrimagnet follows that of the antiferromagnet first (decreasing T ) and then becomes ferromagnetic (increasing T abruptly) upon lowering temperat ures. In addition to the ferrimagnetism, some antiferromagnet shows w eak ferromagnetism due to canting of the spins [100]. In this special class of an tiferromagnet, spins ar e antiferromagnetically coupled but in slightly canted way. Therefore, even when the spin values of coupling spins are same, the canting intr oduces a weak ferromagnetism. Figure 2-5 shows schematically drawn magnetic plots for various types of magnetic ma terials and Figure 2--6 shows specific theoretical simulations of T vs. T plots for one mole of noninteracting isotropic spin 1/2 dimers with intra-dimer magnetic exchange constant J in Kelvin. In this theoretical dimer model, th e magnetization of one mole of non-interacting dimer described as T J T k H g T k H g T k H g T k H g Ng T H MB B B B B B B B B dimerexp exp exp 1 exp exp 2 ) ( (2.2) is derived from the fundamental magnetic equation and Heisenberg exchange Hamiltonian, respectively as ) / exp( )] / exp( ) / [( T k E T k E H E N MB n n B n n n mol (2.3) and H S S g S S JB A B A) ( ) ( H, (2.4) where N is Avogadros number (6.022 1023 / mol), En is energy eigenvalues of S = 1/2 dimer applied to Hamiltonian (Eq. 2.3), kB is the Boltzmann constant (1.38062 10-16

PAGE 36

22 erg / K), J is the Heisenberg exchange constant in Kelvin, SA and SB are the spin operators for spin A and B, g is a g -value, and H is an applied magnetic field. The En values of each dimer are given by E0 = J /4 + gHB, E1 = J /4, E2 = J /4 gHB, and E3 = 3 J /4. The simulations in Figure 2-6 are consistent with abov e Curie-Weiss law in Figure 2-5 and shows that the T value is increasing when J > 0 (ferromagnetic), decreasing when J < 0 (antiferromagnetic), and constant when J = 0 (paramagnetic). For typical molecule-based magnets, the TC values are generally lower than room temperature. Thus, the magnetic behavior above TC can fit into Curie-Weiss law to give a g -value provided that the mass and formula unit of the sample are known. The same type of information can be also extracted from the M vs. H measurement for paramagnet and saturated ferromagnet, by fitting the data to the fundamental magnetization function,5 ) / 10 359 3 tanh( 1 2 1 ) / ) 1 2 ( 10 359 3 tanh( 2 / 1 1 5585 ) (5 5T gH S T S gH S gS H T M (2.5) for one mole of paramagnet case and by comparing the saturated magnetization ( Msat) to ideal saturation value, NgSB. Of course, electron paramagnet resonance (EPR) can be used to determine the g -value directly at a specific te mperature. For a known amount of material, comparing M vs. H of the material with the paramagnetic magnetization function (Eq. 2.5) can also give a hint of magnetic interactions. For example, if the material is ferromagnetic then, the slope of the initial increase in a M vs. H plot is steeper than that of paramagnet and if antiferromagnetic, the slope will be less steep than the paramagnet. Figure 2-7 shows M vs. H plots of theoretically simulated S = 1/2 dimer with 5 This equation is identical to Eq. 2-3. For practical computing reason, all physical constants are already substituted and the final M value is in the unit of emu G / mol.

PAGE 37

23 Figure 2-6. Susceptibility times temperatur e simulations of dimers. Theoretical simulations of T vs. T plots were performed for one mole of non-interacting isotropic spin 1/2 dimers with intr a-dimer magnetic exchange constant J in Kelvin. The simulation used Eq. 2-2 with three different J values at HE = 100 G. The T value is increasing when J > 0 (ferromagnetic), decreasing when J < 0 (antiferromagnetic), and constant when J = 0 (paramagnetic). ferromagnetic and antiferromagnetic intra-dimer interactions compared to the paramagnetic material at two different temperatures (2 K and 5 K) assuming g = 2 in all cases. The upper limit of HE in our MPMS is 7 T. Howe ver, when higher fields are necessary, the resistive magnet at th e NHMFL can be used up to 33 T. 050100150200250300350 0 2 4 6 8 10 T (10-1 emu K / mol)T (K) HE = 100 G, g = 2.00 J > 0 J < 0 J = 0 ferromagnetic dimer antiferromagnetic dimer paramagnetic dimer ferromagnetic ( J = 20 K) dimer simulation antiferromagnetic ( J = -20 K) dimer simulation paramagnetic ( J = 0) dimer simulation

PAGE 38

24 Figure 2-7. Field dependent magnetization simu lations of dimer model. Theoretical simulations of M vs. H plots were performed for one mole of non-interacting isotropic spin 1/2 dimers with intr a-dimer magnetic exchange constant J in the unit of Kelvin. The simulation used dimer model with three different J values with g = 2 in two different temperature environments ( T = 2 K and 5 K). The upper limit of HE in our MPMS is 7 T but highe r field can be reached at the NHMFL up to 33 T. 051015202530 0 5 10 15 M (103 emu G / mol)H (T) J = 0, T = 2 K J = 20 K, T = 2 K J = -20 K, T = 2 K J = -20 K, T = 5 K J = 0, T = 5K J = 20 K, T = 5 K J > 0 J < 0 J = 0 ferromagnetic dimer antiferromagneic dimer paramagnetic dimer

PAGE 39

25 2.3.2 History Dependent Magnets For history dependent magnetic materials su ch as ferromagnets, ferrimagnets, spinglasses, and superparamagnets, the zfc and fc magnetizations are different under certain circumstances. This difference between zfc and fc magnetiz ations is caused by irreversibility of the zfc magnetization and ha s different origins according to classes of the magnets, but the process of zfc in the sample s is similar. For example, in the cluster spin-glass-like material such as K0.6Co1.2[Fe(CN)6] 4H2O powder, zfc magnetization (Figure 2-8) shows characteristic increase in magnetization up to the freezing temperature ( TF ~ 9 K) then decrease in magnetization upon further warming. The zfc magnetization process of this sample can be modeled a nd schematically depicted in Figure 2-8. Briefly, when the sample was cooled below TC value in zero field, the magnetic moments begin to align ferrima gnetically and forms cluster. Upon further cooling below TF, each cluster starts to freeze and the direc tion of each cluster (determined by net spin direction within the cluster) tends to point in random directions to minimize the magnetocrystalline energy. As a result, the net magne tization of the sample achieves its minimum value due to the random distribut ion of spin cluster directions At this point, a small amount of applied HE (100 G in this case) would not be sufficient to align the direction of the cluster or its constituent and other indivi dual spins. Upon increasing temperature, the thermal energy, however, will disturb the fr ozen clusters and individual spins at sufficiently high temperature and will then enable the net cluster direc tion to be realigned parallel to HE. As a result, the net zfc magneti zation will increase until it reaches a maximum at TF, where the magnetic spins in the cluster mostly aligned with the HE to

PAGE 40

26 minimize Zeeman energy in balance with fe rrimagnetic interaction and temperatures. Upon further increasing temperat ure, the frozen cluster will disappear since the spin Figure 2-8. Field-cooled (fc) and zero-field-cooled (zfc) magnetization processes. K0.6Co1.2 [Fe(CN)6] 4H2O powder was used for this measurement. The zfc plot shows a characteristic increas e in magnetization up to the freezing temperature ( TF ~ 9 K) then the decrease in magnetization upon further warming. 051015202530 -1 0 1 2 3 4 M (10-2 emu G)T (K) = HE = 100 G fc zfc

PAGE 41

27 interactions are significantly weaker than the thermal energy fluctuations. As a consequence, the net zfc magnetization will decrease and eventually becomes paramagnetic at T >> TC. The trend of a zfc magnetiz ation plot contains useful information such as the magneto-cryst alline energy, the freezing energy given by temperature, the strength of the magnetic inte ractions between the spins, and the softness of magnet as judged by the strength of HE. All of the information are interrelated and sometimes shows universal scaling be havior as indicated by the linear TF dependence on HE 2/3 in some spin-glass materials [91]. The fc magnetization, on the other hand, cooled with the HE and thus, the direction of clusters will be aligned to the direction of HE as temperature decreases. The zfc magnetization is irreversible in the sense that if the zfc magnetiza tion is measured up to anywhere below TF and cooled again, then the measured magnetization will not be the same upon lowering temperature. Instead, th e magnetization will follow a path that is similar to the fc measurement curve [91]. In an actual magnetization measurement, because of the limitation on the quantity of sample and the lowest temperature limit, occasionally the HE as a measuring field has to be large compared to TF in the energy consideration. In this case, the characteri stic zfc curve (incre ase and decrease below TC) might not be apparent. When this situation o ccurs, a plot of the difference between the fc and the zfc magnetizations helps to determin e the relevant history dependent magnetic information. For a direct measure of TC of history dependent magnets, a thermal remnant magnetization (TRM) measurement can be c onsidered. In a TRM measurement, the sample is cooled to the lowest temperature to below TC in the presence of a finite HE, and

PAGE 42

28 the temperature dependent data are taken upon warming without any HE. Since, there is no magnetic field applied, the paramagnetic and diamagnetic contributions from the sample and holders are expected to be zero. The TRM along with zfc and fc magnetizations measured on sequentially deposited RbxNiy[Cr(CN)6]zm H2O film is shown in Figure 2-9. The sample was cooled to T = 5 K while HE = 100 G, and then the field was lowered to zero for TRM measurement. From the TRM, the TC is determined as ~ 74 K. In comparisons between zfc, fc, and TRM measurements of spin-glass-like materials or superparamagnets, the cooling and warming rates have to be considered because the magnetizations in these materials are not in equilibrium and exhibits notable time dependences. In fact, the zfc and fc m easurements in Figure 2-9 were performed at the same rate, but the sample was warmed slower than the others for the TRM measurement. For some spin-glasses such as Ag(Mn), the sum of the TRM and zfc yielded the same quantity as fc magnetization when all the measurements were performed at the same time rates [103]. An attempt was made to perform this analysis with a RbxNiy[Cr(CN)6]zm H2O film (Figure 2-9), even th ough the TRM was acquired with a different time rate compared to the zfc and fc measurements. The result does not agree with that of Mathieu and coworkers [103]. Another feature of history dependent magnets is a hysteresis loop in M vs. H plot below a certain temperature. For example in the cases of superparamagnetism below the blocking temperature TB, the M vs. H plot possesses a S shape loop. The detailed physics of the hysteresis loop differs according to the different classes of magnets and is beyond the scope of this dissertation and will not be discussed further.

PAGE 43

29 Figure 2-9. Thermal remnant magnetization (TRM) of RbxNiy[Cr(CN)6]zm H2O film. The TRM along with zfc and fc magnetiz ations measured on a sequentially deposited RbxNiy[Cr(CN)6]zm H2O film. For TRM measurement, the sample was cooled to 5 K with HE = 100 G, and the temperature dependent measurement was performed while warming without any HE. 020406080100120 0.0 0.5 1.0 1.5 2.0 2.5 M (10-3 emu G)T (K)TRM after field-cooled with HE = 100 G TRM + zero-field-cooled magnetization zero-field-cooled magnetization, HE = 100 G field-cooled magnetization, HE = 100 G

PAGE 44

30 2.4 Homemade Fiber Optic Sample Holder For photoinduced magnetization studies pr esented in Chapter 5, a homemade fiber optic sample holder (FOSH) for MPMS was used. This inexpensive homemade probe was made to fit the existing parts of MPMS sample rod. Figure 2-10 shows schematics of our homemade FOSH in the experimental se tup. In the long stai nless tube (~ 54 inch I.D. ~ 0.10 inch, and OD ~ 0.12 inch), ten strands of optical fibers (Ocean Optics) were fed through. The room temperature side of th e probe was bent in U-shape and filled with epoxy (2850 FT). The counter balancing optical fibers are optional to give magnetically symmetric sample environment. However, most of time, the background of the holder (gelcap for example) was measured using opt ic sample rod. Therefore, when this background is subtracted from total signal, the contribution from optical fibers is automatically subtracted. Briefly, the choi ce of optical fiber (Ocean Optics, Model 200 UV/VIS, O.D ~ 270 m) was optimized for visible light and the typical power that was used for a photoinduced magnetization measur ement was about 1 ~ 2 mW measured at the sample side when halogen lamp was us ed. Although the current design is sufficient to see rough power dependence of photoindu ced magnetization, the mechanically rigid and optically efficient design is needed for in situ measurement of spectroscopic and magnetization properties. To do so, one so lid optical fiber or bunched fibers using optical fiber collimator can be used in order to fit into the standard spectroscopic sources and detectors. 2.5 Sample Packing and Background Consideration As mentioned earlier in this chapter, pr oper packing of the sample contributes to the accurate measurement of the magnetiza tion. Basically, the MPMS measures the

PAGE 45

31 magnetization by moving the sample for a fin ite distance (usually ~ 4 cm) through the pickup coil, which has a finite dimension (~ 3 cm long and ~ 1.9 cm in diameter). Figure 2-10. A homemade MPMS optic insert ro d. In the long stainless tube (~ 54 inch I.D. ~ 0.10 inch, and OD ~ 0.12 inch), ten strands of optical fibers (Ocean Optics) were fed through the inner space. The room temperature side of the probe was bent in U-shape and filled with epoxy (2850 FT). The balancing optical fibers are optional and are pres ent to provide a magnetically symmetric sample environment. sample counter optical fibers optical fibers carrying light from the source straw ~ 137 cm ~ 11.2 cm Stainless steel tube epoxy optical fibers optical fibers in Teflon tube to light source stainless steel tube, O.D. ~ 0.3 cm. slide seal assembly

PAGE 46

32 Therefore, given the limitation of machine sensitivity, the magnetically symmetric packing of the sample is important to compensate the finite detecting dimension. In addition, the direction of the sample with respect to HE and the intrinsic core diamagnetism of the sample have to be considered along with the magnetic nature of holder to give a pure magnetic response fr om the sample. In the following two subsections, various ways of sample p acking and the consideration of background magnetic signals will be discussed. 2.5.1 Sample Packing Most magnetometers that use superconduc ting magnet as external magnetic field sources measure the longitudinal component of magnetization. In other words, if the axis of superconducting magnet points in the z -direction (longitudinal) as in our MPMS, the magnetometer measures the z -component of magnetization ( Mz). The transverse components of magnetization ( Mx and My) are also of great interest in some cases, but are not applicable to our MPMS and therefore will not be discussed further. Figure 2-11 (a) shows an example of symmetric packing when the homemade FOSH was used. In this case, the optical fibers that are diamagnetic are positioned symmetrically with respect to the center x y plane of the sample assuming that the sample itself is packed symmetrically. The total magnetization of this configuration will be the magnetization of the optical fibers without the sample and the sample without the optical fibers. Since the optical fibers are diamagnetic and longer than th e scan length, the vacant volume created by removing the sample will give a positive magne tic signal, which is proportional to the diamagnetic strength of the optic al fibers. If the counter bala ncing optical fibers were not in place, the detected magnetic signal would not be symmetric, as in Figure 2-3 (b), and an inaccurate measurement value would result. Of course, if the magnetization of the

PAGE 47

33 Figure 2-11. Various MPMS sample packing methods. A symmetric sample packing is shown schematically in (a) and (b). In (b), the bad example of packing sample is also shown (marked as X) along w ith good examples. For powder or small samples, gelcap (c, d), adhesive transp arent tape (e), and grease or epoxy can be used as holders (f). For thin film samples, Mylar shell (g) or gelcap (h) were used for photoinduced magnetic measurements. holder (X) (O) (O) two gelcap tops sample ~ 0.9 cm ~ 0.7 cm ~ 1.4 cm sample in gelcap sample counter optical fibers Optical fibers carrying light from the light source straw (a) (c) (d) wrap transparent tape sample straw sample embedded in grease or epoxy + = films Mylar shell ~ 0.4 cm gelcap Mylar spacer straw spacer films venting hole (b) (e) (f) (g) (h) ~ 0.5 cm sample

PAGE 48

34 sample is significantly stronger than the backgr ound (from optical fibers in this case), the effect of asymmetric packing can be ignored. The contributio n from the straw, which the detection coil sees as a c ontinuous medium, can be also ignored in most MPMS measurements. For a collection of small samples such as a powder or collection of microcrystals, tight symmetric packing is also desired. Figure 2-11 (b) shows schematic representation of good and bad packing of a microcrystalline sa mple. As shown first in figure, when the sample is loosely packed in the holder, an accurate centering is difficult due to the unequal distribution of magnetic weights with respect to the center of the sample. In addition, when the sample is loosely packed a nd if the particles have a magnetic easy axis (for example uniaxial anisotropy), the particle will tend to rotate with its easy axis along HE and will give a value that is different than that of randomly oriented particles. For a large amount (~ 200 mg) or small amount ( < 30 mg) of samples, a gelcap provide a good holder as shown in Figure 2-11 (c) and (d ) respectively. The plastic can, however should be used for a wet or water sensitive sample instead of a gelcap. For some of the photoinduced magnetic measurements of a powde r sample, transparent tape was used as shown in Figure 2-11 (e). The powder sample was first sprinkled on top of the adhesive side of a small section of tape, and then attached to the adhesive side of a larger piece that can be wrapped around the straw. Occasionally, for the air sensitive sample, grease or transparent epoxy can be used as a matrix that holds the samples, see Figure 2-11 (f). In this case, a care must be made to make su re that the thermal contraction and chemical properties of the grease or epoxy do not affect the magnetic properties of the samples.

PAGE 49

35 For the photoinduced magnetic measurements of Prussian blue based thin films, thin Mylar (thickness ~ 100 um) was used as a substrate. Although a silicon wafer is a good choice as a substrate for spectroscopic a nd even mounting purposes (properly cut Si wafer can be wedged inside the straw), it give s a larger magnetic background than that of thin Mylar films. When the directional dependence of the magnetization measurement was needed, the Mylar film sample was cut in squares and stacked together inside a Mylar cubic shell as shown in Figure 2-11 (g). The final assembly of the sample was then wedged inside the straw to any desired direction with respect to HE. A gelcap can also be used as a holder for a film sa mple for a directional dependent magnetic measurement as shown in Figure 2-11 (h). In this case, a piece of straw and the blank Mylar film were used as spacers. The final asse mbly is again pressure fit into the straw. When the pressure fit is impossible due to sm aller size of gelcap, two slots or microholes were made in the straw to hold the gelcap inside. 2.5.2 Background Consideration Although direct measurement of holder back ground and subsequence subtraction is easiest way of removing the background e ffect, sometime, the measurement of the background of an individual holder is impossible due to several reasons. In that case, an indirect estimation of the background is also possible. For example, consider 1 mg of paramagnetic powder ( S = 1/2), which has a molecular we ight, of 1000 g / mol is packed in a typical gelcap (40 mg). The simulated ex perimental magnetic data for this sample at T = 2 K is shown in Figure 2-12 (a) and is labe led as a total. This total signal is a superposition of the magnetization of samp le itself and the background. Assuming that the background is a linear function of field a nd is temperature independent, a linear fit to the total plot at higher field will give the sl ope that comes from the background. This

PAGE 50

36 Figure 2-12. Simulation of weak paramagnetic sample in gelcap holder. A field dependent magnetization of a theore tical paramagnetic sample (1 mg, S = 1/2, M.W. = 1000 g/mol) and gelcap holder (40 mg) is simulated at T (a) = 2 K, T (b) = 10 K, and T (c) = 400 K. estimation is possible because of the satura tion of the paramagnetic magnetization at a low temperature. When the temperature is higher, for instance at 10 K as shown in Figure 2-12 (b), a similar type of background estimation will be difficult. Instead, the 051015202530 -6 -4 -2 0 2 4 6 M (10-3 emu G)H (T) 051015202530 -6 -4 -2 0 2 4 6 M (10-3 emu G)H (T) 051015202530 -6 -4 -2 0 2 4 6 M (10-3 emu G)H (T)(c) (b) (a) sample: mass: 40 mg M.W.: 1000 g / mol S = 1/2, g = 2 holder: gelcap mass: 40 mg The background of the holder was assumed to be temperature independent. sample holder total sample holder total total holder sample T = 400 K T = 2 K T = 10 K

PAGE 51

37 field dependent magnetization can be meas ured at higher temperature where the magnetization from the sample is a minimum, as shown in Figure 2-12 (c) to estimate the background of holder. The origin of the back ground of the holder system, however, is illdefined in the case of Figure 2-12 (a) becau se the estimated bac kground from the linear fit of the magnetization curve actually cont ains the core-diamagnetic contribution from the sample as well as the background from the holder. 2.6 Summary and Future Direction In this section, the conf iguration of the MPMS and the typical magnetic measurement processes, as well as data analys is strategies were reviewed. Briefly, the MPMS is a wide-temperature ranged (1.9 K ~ 400 K), computer-controlled magnetometer using SQUID electronics to enhance magnetic detection. The typical sample measurement consists of zfc, fc, and M vs. H sequences. From the trends of the data and Curie-Weiss fitting processes, th e magnetic properties of the samples can be analyzed. More specifically, by analyzing the behavior of T vs. T and M vs. H plots, the sample can be categorized roughly into, paramagnets, ferromagnets, antiferromagnets, and ferrimagnets. For special magnetic studies, such as photoinduced magnetism, a homemade FOSH was constructed to measur e the magnetizations under irradiation of light. This homemade FOSH consists of 10 stra nds of optical fibers in stainless tube, and the typical light power used for photoindu ced magnetization experiment was about 1 ~ 2 mW / cm2. Although it is limited by the size of sample space and operational temperature, the great advantages of using MPMS are: (1) the accuracy in magnetic measurement, (2) the convenience of the sample changing environm ent using the sample rod, and (3) a fully automated computer-controlled temperature a nd measuring sequence. These advantages

PAGE 52

38 bring systematic and reproducible efficiency to these experimental ma gnetic studies. As long as the dimension of the experimental space is compatible, the MPMS can be used not only as a magnetometer but also as a general purpose low or high temperature measurement device by revising the existing sa mple rod. Some of the modified sample rods for transport, optical measurement, and even pressure study are already commercially available, but at non-negligible co sts. In order to maximize the efficiency of MPMS and to minimize the cost, a cost effective multi-purpose sample rod can considered. In this new sample rod, the top of the room side stainless tube is attached to the quick connector and provides vacuum s ealed connection between the sample space and outside. In principal, any thing that fits into the quick connector and stainless tube can be inserted into the sample space. For ex ample, optical fibers and thin coaxial cables can be introduced at the same time to m easure a magneto-optical-t ransport property under the irradiation of the light. Eventually the cooling power of the system and the size of probe space will limit the use of this multi-purpose sample rod, but otherwise the uses will increase the efficiency in many areas of experimental physics.

PAGE 53

39 CHAPTER 3 PHYSICAL PHENOMENON AND THEORY 3.1 Photoinduced Magnetism in Prussian Blue Analogs 3.1.1 Overview The phenomenon of photoinduced magnetization in a Co-Fe(CN)6 Prussian blue analog was first observed by Sato and coworkers in 1996 [38]. Accordin g to their studies, a ferrimagnetic compound K0.2Co1.4[Fe(CN)6].9H2O showed increases of magnetization upon irradiation of red light. Furthermor e, the resultant photoinduced magnetization disappeared under irradiation of blue light or by increasing te mperature to ~150 K. This unique control of magnetization by light and temperature attracted numerous researchers and opened a new field of functional molecule -based magnets [38 79], which could be useful for future magnetic devices. This section introduces th e phenomena and proposed theoretical descriptions of phot oinduced magnetism in Co-Fe(CN)6 Prussian blue analogs to provide a theoretical background to the an isotropic photoinduced magnetism described in Chapter 5. 3.1.2 Initial Observation and Description Based on the result of Sato and cowork ers, the three-dimensional (3D) bulk compound K0.2Co1.4[Fe(CN)6].9H2O undergoes a short range ferrimagnetic ordering below TC ~ 16 K in the dark state [38]. The majo rity of sample in this dark state was proposed to be a mixture of diamagnetic Co-Fe pairs and ferrimagnetically coupled Co-Fe pairs distinguished by the oxidation st ates of ions and the consequent spin configurations. More speci fically, some portion of the sample is composed of

PAGE 54

40 diamagnetic pairs of low-spin CoIII (6 2 gt, S = 0) ions and low-spin FeII (6 2 gt, S = 0) ions bridged by CN in the form of CoIII-NC-FeII. The other portion on the other hand, consists of magnetic pairs of high-spin CoII (2 5 2 g ge t, S = 3/2) ions and low-spin FeIII (5 2 gt, S = 1/2) ions forming CoII-NC-FeIII units. These magnetic spins of CoII and FeIII in the original dark state, in contrast to the diamagnetic spins of CoIII and FeII, will be referred as primordial spins to be differentiated later by the photoinduced spins that are created as a result of light irradiation. Upon reducing temperature below TC, some primordial spins interact antiferromagnetically via superexcha nge and yield a net ferrimagnetic moment with effective S = 1 per CoII-NC-FeIII pair. The first microscopic evidence of ferrimagnetic interactions between CoII and FeIII ions in a similar compound Rb1.8Co4[Fe(CN)6]3.3H2O, was investigated by Campion and coworkers in their K-edge X-ray magnetic circular dichroism studies [45]. When Sato and coworkers illuminated the K0.2Co1.4[Fe(CN)6].9H2O compound at T = 5 K with red light, the magnetization in creased, and this photoinduced magnetization lasted for several days. As a result of the photoinduced enhancement of the magnetization, the TC increased from approximately16 K to 19 K and the coercive field at 5 K was enhanced from 800 G to 1500 G. The results of our sequentially deposited RbjCok[Fe(CN)6]l n H2O films, which have a similar face-centered cubic structure and photoinduced enhancement of the magnetization, are shown in Figure 3-1 and 3-2. A full discussion of the particular film shown in Fi gure 3-1 is presented in Chapter 5. Briefly, the zero-field-cooled (zfc) and field-cooled (f c) magnetizations are pl otted in Figure. 3-1 as a function of temperature before and afte r white light irradiation, and Figure 3-2 shows time dependent magnetization under white light irradiation. The bumps in the data in

PAGE 55

41 102030 0 5 10 15 20 TC photoinduced Mfc photoinduced Mzfc dark Mfc dark MzfcHE = 200 G || film M (10-5 emu G / cm2)T (K) h TP Figure 3-1. Temperature dependent phot oinduced and dark magnetizations of sequentially deposited RbjCok[Fe(CN)6]l n H2O film. The zero-field-cooled (zfc) and field-cooled (fc) magnetiza tions are plotted as a function of temperature before and after irradiati on with white light. As a result of irradiation, the overall magnetizat ions increased. In addition, TC increased by ~ 2 K and TP, the temperature where zfc magnetization shows a maximum value increased by ~ 2 K.

PAGE 56

42 012345 0.0 0.2 0.4 0.6 0.8 1.0 1.2 202530 3.6 3.8 4.0 4.2 off red LED light on blue LED light on offM (10-4 emu G)Time ( hours ) white light on T = 5 K, HE = 1 kG off M (10-5 emu G)Time (hrs) Figure 3-2. Time dependent photoinduced magn etization. Under irradiation of white light at 5 K, the magnetiza tion increase to ~ 500% of its original value over 4 hours of period. The inset shows ma gnetization responses under blue and red light irradiation. In both cas es, the magnetization increased under irradiation.

PAGE 57

43 Figure 3-2, when light is toggled on and off, are due to a thermal effect associated with the light being a local heating source. The trends (i.e. overall increase in the magnetization and enhancement of TC at the photoinduced state) of our data are consistent with those observed in K0.2Co1.4[Fe(CN)6].9H2O [38]. However in our case, there was no decreasing in the magnetization upon irradiat ion with blue light as shown in the inset of Figure 3-2, and the absence of this demagne tization effect might come from the use of different alkali metals (e.g. Rb in our case) and has been reported by other group [43]. The Sato and coworkers were the firs t to suggest that the mechanism of photoinduced magnetization was due to an inte rnal photochemical redox reaction and this picture was further supported by Verdague r [39]. According to Verdaguer, a K0.2Co1.4[Fe(CN)6].9H2O compound (a typical Co-Fe(CN)6 Prussian blue analog) has a face-centered cubic structure with some complex chemical constitutes: primordial magnetic units of CoIICN-FeIII, water filled Fe vacancies as shown in Figure. 3-3 (a) (sites A), isolated diamagnetic units of FeII(CN)6 (Figure. 3-3 (a) B sites), and diamagnetic pairs of CoIII-NC-FeII (Figure. 3-3 (a), C sites). Below TC when the sample is exposed to the red-light, the charge transf er occurs between the Fe and Co ions within the diamagnetic pair (Figure. 3-3 (a), C sites). The electron from the FeII ion transfers to CoIII ion and the diamagnetic pair becomes a photoinduced magnetic FeIII-CN-CoII pair, which is similar to the primordial magnetic pair. This photoinduced magnetic state of spins may subsequently return to its in itial diamagnetic state upon the blue light irradiation via the process of electron transfer from CoII to FeIII. These charge transfer processes are schematically shown in Figure. 3-3 (b) for photoinduced magnetization (red light) and demagnetization (blue light).

PAGE 58

44 Figure 3-3. Processes of photoinduced magne tization and demagnetization described by Verdaguer [39]. The model compound c ontains primordial magnetic units of FeIII-CN-CoII, water filled Fe vacancies (sites A), isolated diamagnetic units of FeII-(CN)6 (B sites), and diamagnetic pairs of CoIII-NC-FeII (C sites). When the sample is exposed to the red light, the charge transfer occurs between the Fe and Co ions within the diamagnetic pair (C sites, top) The electron from the FeII ion transfers to CoIII ion and the diamagnetic pair becomes a photoinduced magnetic FeIII-CN-CoII pair, which is similar to the primordial magnetic pair. This photoinduced magnetic state of spins returned to the initial diamagnetic state upon the blue light irradiation via the process of electron transfer from CoII to FeIII. These charge transfer processes are schematically shown in (b). C-N H2O K Co Fe (a) (b) A B C CoIII FeII NC CoIIIFeII NC CoII FeIII NC S = 0 S = 0 S = 3/2 S = 1/2 Charge Transfer e Red CoIII FeII NC CoIIFeIII NC CoII FeIII NC S = 0 S = 0 S = 3/2 S = 1/2 Charge Transfer e Blue

PAGE 59

45 As a consequence of photoinduced magne tization, the net number of magnetic neighbors increases, and this increment contributes to the enhancement of TC in the photoinduced magnetic state. Verdaguer also indicated that thes e charge transfer processes occur at the local leve l (for example, at site C of Figure 3-3 (a) under red-light) but activate the cooperative ma gnetic interactions through out the sample. In fact, the cooperative behavior was observed by the Verdaguer group when studying Rb0.52Co[Fe(CN)6]0.84.3H2O [48]. The reversible photoi nduced magnetization scenario of Verdaguer is to be consistent with the subsequent theoretical descriptions based on ab initio quantum chemical cluster calculations by Kawamoto and coworkers [55]. The calculation was performed on the experimentally studied compound K0.4Co1.3[Fe(CN)6]5H2O [66], which showed little or no evidence of primordial spin pairs of FeIII-CN-CoII when compared to the original K0.2Co1.4[Fe(CN)6].9H2O system. As a consequence of the absence of primordi al spins, the model compound did not show ferrimagnetic ordering in dark state. The significance of the concentrations of each chemical constituent and the choice of the alka li ions will be discussed later in this chapter. Briefly, the experimental results on K0.4Co1.3[Fe(CN)6]5H2O [66] indicated weak paramagnetic behavior between 2 K and 340 K in the dark state, and upon irradiation of light (500 ~ 700 nm) at 5 K, the material became ferrimagnetic with TC ~ 26 K. Upon further irradiation by infrared (IR) light (1319 nm), the magnetization returned to its initial weak paramagnetic state. In order to explain revers ible photoinduced magnetism, Kawamoto and coworkers [55] proposed four relevant spin states of the Co-Fe pairs as schematically shown in Figure. 3-4 (a). This theoreti cal description will be referred as Co cluster model. By

PAGE 60

46 going through these four different spin st ates, a complete photoinduced magnetization and demagnetization cycle can be described. The first state (LS0) begins at the ground state of the diamagnetic pair of CoIII (6 2 gt, S = 0)-NC-FeII(6 2 gt, S = 0), and this ground state can be excited to the second state (L S1) by visible light (500-700 nm). The LS1 state, which is an intermediate state, is composed of a CoII (1 6 2 g ge t, S = 1/2)-NC-FeIII (5 2 gt, S = 1/2) pair. Without a ny external stimuli such as light irradiation, the CoII (1 6 2 g ge t, S = 1/2) ions in the LS1 stat e quickly relaxes into a high-spin configuration, CoII (2 5 2 g ge t, S = 3/2), via intersystem crossing. In this HS0 sate, the phot oinduced enhancement of the magnetization is complete, and if the sample is below TC, then the spins in HS0 state will be coupled antiferromagnetically, generating an effective ferrimagnetic moment of S = 1 (3/2 1/2) per Co-Fe pair. On the other hand, if T >> TC, a paramagnetic moment of S = 2 (3/2 + 1/2) per Co-Fe pair will result. U pon further irradiation with IR light, the HS0 state can be promoted to an interm ediate excited state (HS1) in the CoIII (1 5 2 g ge t, S = 1)NC-FeII(6 2 gt, S = 0) configuration. This intermedia te HS1 state will ag ain be transferred to the stable diamagnetic ground state (LS0) via intersystem crossing, thereby completes one cycle of the photoinduced magnetiz ation and demagnetization process. Kawamoto and coworkers also realized that, in the K0.4Co1.3[Fe(CN)6]5H2O model compound, if the water-filled Fe-vacancies ar e randomly distributed inside the sample, then 37% of the Co ions are in a (CN)5-Co-(OH2) environment (number of water per Co is 1, NW = 1) and 21% of the Co ions are in a (CN)6-Co arrangement (no water per Co, NW = 0), as shown in the insets Figure. 3-4 (b) and (c) respectively. Using an ab initio

PAGE 61

47 calculation program, the potential ener gies of these Co clusters with NW = 1 and 0 were plotted for different states (LS0, LS1, HS0, a nd HS1) as a function of Co-N distance [55]. Figure 3-4. Complete cycle of photoindu ced magnetization and demagnetization proposed by Kawamoto and coworkers [ 55]. The complete process can be obtained by going thorough LS0 visible light LS1 HS0 IR light HS1 LS0 cycle (a). Co clusters with one (b) and no water (c) molecules, respectively. (a) e g t2 g CoIII ( S = 0) FeII ( S = 0) LS0 e g t2 g CoII ( S = 1/2) FeIII ( S = 1/2) LS1 e g t2 g CoIII ( S = 1) FeII ( S = 0) HS1 e g t2 g CoII ( S = 3/2) FeIII ( S = 1/2) HS0 (b) Nw = 0 (c) Nw = 1 CN H2O Co Light 1 Light 2 Inter System Crossing Inter System Crossin g

PAGE 62

48 From these calculated potential energy diagrams, Kawamoto and coworkers were able to propose a complete photoinduced magnetization and demagnetization scenario, which is consistent with Verdaguers rationalization in troduced earlier in this section. The basic scenario can be summarized as a cyclic process of LS0 visible light LS1 HS0 IR light HS1 LS0 cycle. In addition, they prop osed that the photoinduced enhancement of the magnetization originates at the (CN)5-Co-(OH2), NW = 1 site in the compound, because the calculation shows that the poten tial energy difference between LS0 and LS1 in Nw = 1 site is close to the experimentally observed value (500-700 nm). On the other hand the photoinduced demagnetization process originates from the (CN)6-Co, NW = 0 site. The potential diagram shows that th e energy difference between the HS0 to HS1 transition in the Nw = 0 site is close to that of expe rimentally observed value (1350 nm). As Kawamoto and coworkers indicated, the impor tance of their results is to verify that the photoinduced magnetization and demagneti zation processes are initiated from the spatially distinct local sites of Co with different ligand environments (mainly at NW = 1 and NW = 0). In general this local effect, for example the photoinduced magnetization arising at local (CN)5-Co-(OH2), induces an overall phase transi tion throughout the crystal to be in stable photoinduced magnetic state. More sp ecifically, when visible light (500-700 nm) is used, a charge transfer in the Co-Fe pa irs occurs if the local Co ion is in (CN)5-Co(OH2) environment. However, in the presence of visible light the charge transfer result in the promotion of the of the LS0 state to the LS1 state should not occur at the (CN)6-Co site because, according to the potential energy diagram, the visible light lacks sufficient energy. Therefore, if the cooperative behavior is not considered, upon irradiation with a

PAGE 63

49 visible light, the (CN)5-Co-(OH2) sites will become magnetic, but the (CN)6-Co sites in the crystal will remain diamagnetic. However, the experimental evidence shows that the material undergoes a photoinduced phase tran sition as a whole and supports cooperative behavior [48]. The same argument is app lied for photoinduced demagnetization process: starting from the (CN)6-Co sites and extending through the crystal upon IR light irradiation to be in LS0 ground state. The origin of the cooperative behavior can be related to the lattice elongati on of Co-Fe pairs when they undergo spin transition such as photoinduced magnetic transition. Briefly, th e local elongation of the lattice in the photoinduced Co-Fe pair will affect neighbor ing diamagnetic Co-Fe pair and in the process of minimizing latti ce distortion energy, the elongation of the lattice and associated magnetic properties will propagate throughout the cr ystals. If the material contained only (CN)5-Co-(OH2) sites without (CN)6-Co sites, the photoinduced magnetization would be possible but phot oinduced demagnetization would not be possible. Hence, in order to have reve rsible control of th e photoinduced magnet, Kawamoto and coworkers emphasized that the ma terial has to be composed of a mixture of (CN)6-Co sites and (CN)5-Co-(OH2) sites, that is to say that the Co-Fe(CN)6 Prussian blue structure should contain water-filled Fe -vacancies. Before discussing additional experimental results it is important to stre ss an obvious point, na mely basic conditions necessary for the existence of photoinduced magnetization is the presence of diamagnetic pairs of CoIII-NC-FeII in the sample. 3.2 Charge Transfer Induced Spin Transition Although the Co cluster mode l can explain the reversib le process of photoinduced magnetization in K0.4Co1.3[Fe(CN)6]H2O, other experimental evidence on a similar material did not show the photoinduced magne tization even if the material contained

PAGE 64

50 mostly (CN)6-Co clusters. This absence of phot oinduced magnetization, and the ability to stimulate photoinduced demagnetization in si milarly generated materials, suggest the effects are linked to subtle differences in the chemical concentrations and to the choice of different alkali metal ions. Considering electroneutrality, general formula for CoFe(CN)6 based Prussian blue so lid can be written as AxCo4[Fe(CN)6]y4y n H2O [44], alternative to AjCok[Fe(CN)6]l n H2O, where A is an interstitial alkali metal cation and is a Fe-vacancy. During the search for the optimal conditions for producing photoinduced magnets, some Co-Fe(CN)6 Prussian blue structures were discovered to undergo charge-transfer-induced spin transition (CTIST) at relatively higher temperature compared to the TC of the sample. The systematic st udies [47, 57] show that the choice of alkali metals and the relative concentrations given by x and y in the formula affect the conditions for the CTIST to occur. More over, the condition fo r being a photoinduced magnet is related to the occurrence of CTIS T. Shimamoto and coworkers studied the magnetic properties of a series of NaxCoy[Fe(CN)6] z H2O compounds [57] by systematically varying the Co/Fe ratio, which in turn modifies the concentration of alkali metal ions and water filled Fe-vacancies. Tables 3-1 and 3-2 show the chemical composition and the valence states at 290 K of the five samples from Shimamoto and coworkers. Table 3-1. Chemical compositions of NaxCoy[Fe(CN)6] z H2O [57]. Sample Na Co Fe(CN)6 Vacancy H2O Co/Fe 1 0.07 1.50 1 0.50 6.3 1.50 2 0.37 1.37 1 0.37 4.8 1.37 3 0.53 1.32 1 0.32 4.4 1.32 4 0.60 1.26 1 0.26 3.9 1.26 5 0.94 1.15 1 0.15 3.0 1.15

PAGE 65

51 Table 3-2. Valence states of NaxCoy[Fe(CN)6] z H2O at 290 K [57]. Valence State at 290 K 1 Na0.07CoII 1.50 [FeIII(CN)6]0.93[FeII(CN)6]0.07.3H2O 2 Na0.37CoII 1.37 [FeIII(CN)6]0.89[FeII(CN)6]0.11.8H2O 3 Na0.53CoII 1.32 [FeIII(CN)6]0.83[FeII(CN)6]0.17.4H2O 4 Na0.60CoII 1.08CoIII 0.18[FeIII(CN)6]0.70[FeII(CN)6]0.30.9H2O 5 Na0.94CoII 0.39CoIII 0.76 [FeII(CN)6]1.00.0H2O The results of dark state magnetic measurements as T vs. T and photoinduced magnetization studies presented as M vs. T are schematically s hown in Figure 3-5 (a) and (b) respectively. These plots are not the actual data plots shown in Reference [57] but the illustrations showing characteristics of the ma in results. Sample 1, which contains the most Fe-vacancies and the least alkali metals shows a constant T value between 50 K and 350 K, whereas Sample 5, which contains the least Fe-vacancies and most alkali metal ions shows constant but lower T value compared to that of sample 1. In other words, the nearly temperature independent T values indicate that the two materials are in the paramagnetic regime, and IR data sugge st that the paramagnetic part of sample 1 and 5 are mainly composed of primordial spins of CoII (2 5 2 g ge t, S = 3/2)-NC-FeIII (5 2 gt, S = 1/2) and CoII (2 5 2 g ge t, S = 3/2)-NC-FeII (6 2 gt, S = 0), respectively, even though Sample 5 contains a majority of diamagnetic pairs, see Table 3-2. Furthermore, Samples 1 and 5 did not exhibit any photoinduced magnetization as shown in the Figure. 3-5 (b). Sample 1 failed to meet the basic condition (exi stence of Co-Fe diamagnetic pairs) for photoinduced magnetism but the sample 5 contai ns plenty of diamagnetic pairs, and moreover the majority the of Co ions are in a (CN)6-Co environment, which according to Kawamoto and coworkers is a necessary c ondition for photoinduced magnetization. The resolution of this apparent discrepancy betw een theory and experiment might come from

PAGE 66

52 the consideration of the streng th of different ligand fields holding the crystal together. When the material undergoes a transition to the photoinduced magnetic state, the crystal experiences a change of volume as indicated by the experiments and theories [55, 66]. In similar Prussian blue structures, the flexibil ity of volume change is then related to the ligand field strength that holds the crys tal. For example Co with six CN ( NW = 0) will hold crystal tighter than the Co i on with five CN and one water ( NW = 1) because the ligand field of the CN is stronger than that of H2O. As a consequence even though Co with six bridging CN links is a good condition for photoinduced magnet, the strong ligand field from all CN do not let the crysta l relax to be in th e photoinduced state. Consequently, in order to have photoinduced magnetization, there needs to be a fine balance between the number of (CN)6-Co configuration and latti ce rigidity, which can be weaken by substituting the (CN)5-Co-(OH2) motif to the (CN)6-Co site. This substituting is equivalent to creating more water filled Fe-vacancies. In term s of Fe-vacancies, if there are too many or too few Fe-vacancies the sample will not undergo a photoinduced transition. Too many vacancies will put the sample in already lattice-relaxed magnetic primordial spin states plus lack of diamagne tic pairs, while too little Fe vacancy leave the lattice too stiff to respond for photoi nduced state of lattice relaxation. On the other hand, the samples in the intermed iate range of Fe vacancies such as in samples 2, 3, and 4 show photoinduced magnetiza tion as indicated in Figure 3-5 (b). Moreover upon cooling, roughly constant T values (close to the value of sample 1) fall quickly at around 200 K, 240 K, and 260 K, and then become nearly constant again at 190 K, 230 K, and 250 K for sample 2,3 and 4 respectively. These changes in T values are characterized as CTIST and show hyste resis upon warming. For these samples (2, 3,

PAGE 67

53 and 4), the concentrations of Fe-vacancies are flexible enough to be photoinduced and CTIST magnets as shown in Figure 3-5. It is noteworthy that the temperatures where the T values fall in sample 2, 3, and 4 systemati cally move to higher temperatures as the Figure 3-5. Magnetizations of a series of NaxCoy[Fe(CN)6] z H2O compounds studied by Shimamoto and coworkers [57]. (a) T vs. T graphs upon cooling and warming. (b) Magnetization befo re and after irradiation. light light light (b) 0 10 20 30 40 T (K) 0 1 0 1 0 1 0 1 0 1 Sample 2 Sample 1 Sample 3 Sample 4 Sample 5 (a) 2 4 0 2 4 0 2 4 0 2 4 0 2 4 0 100 200 300 T (K) cooling warming light dark

PAGE 68

54 numbers of Fe-vacancies of the samples decreas e. This trend can be rationalized as a result of competition between the ligand fiel d and temperature: the stronger the ligand field (less Fe vacancies) the sample has, it ma intains the tight crystal structure to higher temperatures. The CTIST can be suppressed by rapid cool ing (quenching) the samples [46, 56]. Hanawa and coworkers [56] argued that the photoinduced state of the sample is essentially the same as the high temperature phase of the sample such as sample 2,3, and 4 in the states before thermally induced ch arge transfer occur and furthermore, they investigated microscopic do main structures of photoi nduced state of sample in comparison to the rapidly cooled sample usi ng high angle-resolved synchrotron-radiation X-ray powder diffraction techni que. From this study, Hanawa and coworkers concluded that the structural domain si ze of the photoinduced state is larger than the quenched hightemperature phase, and the lattice constant of photoinduced state is larger (10.287(1) for Co to Co distance) than that of th e quenched phase (10.187(3) for Co to Co distance). The conditions for photoinduced ma gnets or CTIST magnets seem to be the same in samples 2, 3, 4, and Hanawas sample but generally these conditions are not necessary the same because some compounds with Rb or K as alkali metal ions do not show the thermally induced charge transf ers but do show photoinduced magnetization [57]. Interestingly, one of the powder compounds that we have studied showed both CTIST and photoinduced magnetization even if the compound contains K as an alkali metal. In addition, the same sample also exhibited temperature i ndependent primordial spins indicated by ferrimagnetic ordering in T plot after the thermally induced charge transfers as shown in Figure 3-6. The e xpected formula for our powder sample is

PAGE 69

55 K0.6Co1.2[Fe(CN)6]4H20 based on the chemical components used in synthesis procedure. The T plot of the sample also shows the m acroscopic fingerprint of ferrimagnetic interactions between the Co and Fe spins indicated by the decrease of T down to ~ 50 K and abrupt increase below ~25 K. The magneti zation measured at 5 K and up to 7 T are 050100150200250300 4 6 8 10 050100150200 6 7 8 9 10 cooling warmingT (10-4 emu K)T (K) HE = 5 kGT = 5 K, HE = 1 kGlight on light offM (10-3 emu G)Time (min) Figure 3-6. T vs. T data of K0.6Co1.2[Fe(CN)6]4H2O powder upon cooling and warming. The inset shows the time dependent ma gnetization under irradiation of white light.

PAGE 70

56 02468 -1 0 1 2 3 4 -0.10.00.1 -1 0 1 M (10-2 emu G)H (T)HcMr photoinduced M dark MM (10-2 emu G)H (T)T = 5 K Ms Figure 3-7. M vs. H data of K0.6Co1.2[Fe(CN)6]4H20 powder measured at 5 K upon sweeping field from 7 T to -0.1 T before and after white light irradiation. The inset shows zoomed view of magnetic da ta near the origin of the graph. plotted in Figure 3-7 for the cases of both phot oinduced and dark stat es. As indicated by the arrows in the Figure, some characteristi c values such as sa turation magnetization ( MS), remnant magnetization ( Mr), and coercive field ( HC) are increased in the photoinduced state.

PAGE 71

57 3.3 Summary From the all experimental and theoretical studies presented in this section, the conditions necessary to support photoinduced enhancement of the magnetization can be summarized as: 1. existence of diamagnetic Co -Fe pairs, 2. suitable local environment of Co such as (CN)6-Co, and 3. structurally flexible la ttice. Similarly, the condition for photoinduced demagnetization can be summari zed as 1. existence of magnetic Co-Fe pairs, 2. suitable local envi ronment of Co such as (CN)5-Co-(OH2), and 3. flexible lattice. In general these three conditions are interrela ted via chemical formula, i.e. choice of alkali ions and relative concentr ation of chemical constituents. As external conditions, of course, the choice of suitable range of light is most important but other variables such as pressure, and orientations of the sample as shown in Chapter 5, can be considered. Since the proper chemical concentration is the cr ucial variable in p hotoinduced magnetism and CTIST, systematic studies varying the choi ce of the alkali metal and the concentration between the constituents of th e Co-Fe Prussian blue structure are needed to optimize the protocol for generating a photoinduced magnet. Furthermore, understanding the physical and chemical conditions for photoinduced ma gnetism and the theoretical predictions of relevant potential energy diagram can contribut e to generate various types of functional magnets required for modern magnetic devices. To this end, Chapte r 5 introduces unique Co-Fe(CN)6 Prussian blue films that shows an isotropic photoinduced magnetic properties and Chapter 6 reviews unusual CTIST phenomenon in K0.6Co1.2[Fe(CN)6]H20 powder when the specimen was cooled rapidly below 100 K.

PAGE 72

58 CHAPTER 4 MAGNETIC STUDY OF EVOLVING STURUCTURE (MONO, BI, AND MULTILAYER OF FILMS) 4.1 Synthesis of Ni-Fe(CN)6 Films The generation of Langmui r-Blodgett (LB) Ni-Fe(CN)6 films [9 13] involves the preparation of a chemical so lution (LB solution) containing bis(tetramethylammonium)pentacyano-(4 -(didodecylamino)-pyr idine)ferrate(III)6H20, here after written as amphiphilic pentacy anoferrate and nickel nitrate, Ni(H2O)6(NO3)2. As shown schematically in Figure 4-1, one un it of amphiphilic pentacy anoferrate consists of a central Fe ion with five cyanide (CN) arms and one amphiphilic tail group in the locations of octahedral vertices. This build ing unit floats on the water-based solution due to the hydrophobic nature of the amphiphilic ta il. When amphiphilic pentacyanoferrate is introduced to the Ni(H2O)6(NO3)2 solution, its four CN arms in the equatorial plane replace the H2O from the Ni(H2O)6(NO3)2 and bond to the Ni ions. As a consequence, a face centered square grid netw ork of Fe-CN-Ni forms at the air-water interface and the hydrophobic amphiphilic tails of the Fe ions poi nt away from the water as shown in Figure 4-1. The Fe-CN-Ni network is then transferred onto solid substrates, and depending on the deposition cycles, monol ayer, bilayer, and multiple bilayers (multibilayer) specimens are generated. For the magnetic studies, a commercial Mylar film was used as a substrate, and for the sp ectroscopic studies, singl e-crystal (100) silicon wafers were used.

PAGE 73

59 Figure 4-1. Amphiphilic pentacyanoferrate bu ilding unit and 2D two-dimensional grid network. One unit of amphiphilic pentacyanoferrate consists of a central Fe ion with five cyanide (CN) arms and one bis(tetramethylammonium), a tail group. This building unit floats on the water-based solution due to the hydrophobic nature of amphiphilic tail. When amphiphilic pentacyanoferrate is introduced to the Ni(H2O)6(NO3)2 solution, its four CN arms in equatorial plane replace the H2O from the Ni(H2O)6(NO3)2 and bond to the Ni ions. As a consequence, a face centered square grid network of Fe-CN-Ni forms at the air-water interface and the hydrophobic am phiphilic tails point away from the water surface. Bilayer and multibilayer films were prep ared using the Y type Langmuir-Blodgett technique [13]. Briefly, one bilayer of the Fe-CN-Ni network was formed through a cycle of immersing and withdrawing a hydrophobic substrate into/out of the aforementioned LB solution. The multibilaye r was produced by repeating the deposition cycles for multiple times. In a slightly different way, the monolayer of Fe-CN-Ni was generated by starting with th e hydrophilic substrate immers ed in the LB solution and pulling it out of the solution. Th e structures of resultant film s are shown schematically in Fe3+ C CN NC C N N C N N N R R Ni Fe Ni Fe Ni Fe Ni Fe Ni N N R R N N R R N N R R N N R R two-dimensional grid network unit of amphiphilic pentacyanoferrate R = (CH2)11CH3

PAGE 74

60 Figure 4-2. The results of X -ray diffraction measurements indicate that the distance between each bilayer (inter-bilayer distance) in the multibilayer film is about 35. The structural coherence within the two-dimensi onal network was also probed using Grazing Incidence X-ray Diffraction (GIXD), and th e result yields an average of 3600 2 coherent area coverage in the multibilayer sample. In other words, within the two-dimensional network, the crystallinity is preser ved in the average area of 3600 2 The GIXD result also indicates that the unit cell edge distan ce (i.e. the Fe Fe spacing in the Fe-CN-NiNC-Fe configuration) is a bout 10.2 as shown in Figure 4-3. The distance and the structure between the monolayers in each bi layer (intra-bilayer region) are not well characterized at this time. The detailed synt hesis procedures and characterizations can be found in Reference [9 13]. 4.2 DC Low Field Magnetization Measurements The DC magnetization measurements of each film were performed using the QUANTUM DESIGN Magnetic Property Measurement System (MPMS), a superconducting quantum interference devi ce magnetometer whose properties are described in Chapter 2. About 10 cm2 of each film were cut into small squares and packed into a commercial plastic can or gelatin capsule (g elcap) for the MPMS measurement. For each sample, the magnetic measurements were performed in two different film orientations with resp ect to the external magnetic field ( HE). In one orientation, the film was placed wi th its surface parallel (||) to HE, and in the other orientation the film was placed with its surface perpendicular ( ) to HE. The background magnetic signal arising from each commercial can or gelcap was measured beforehand

PAGE 75

61 multibilayer Fe Ni Fe Ni Fe Ni Fe Ni Fe Ni Fe Ni Fe Ni Fe Ni Fe Ni Fe Ni monolayer Fe Ni Fe Ni Fe Ni Fe Ni bilayer 36 multibilayer Fe Ni Fe Ni Fe Ni Fe Ni Fe Ni Fe Ni Fe Ni Fe Ni multibilayer multibilayer Fe Ni Fe Ni Fe Ni Fe Ni Fe Fe Ni Ni Fe Fe Ni Ni Fe Ni Fe Ni Fe Ni Fe Ni Fe Fe Ni Ni Fe Fe Ni Ni Fe Ni Fe Ni Fe Ni Fe Ni Fe Fe Ni Ni Fe Fe Ni Ni Fe Ni Fe Ni Fe Ni Fe Ni Fe Fe Ni Ni Fe Fe Ni Ni Fe Ni Fe Ni monolayer Fe Ni Fe Ni Fe Ni Fe Ni Fe Fe Ni Ni Fe Fe Ni Ni monolayer monolayer Fe Ni Fe Ni Fe Ni Fe Ni bilayer Fe Ni Fe Ni Fe Ni Fe Ni Fe Fe Ni Ni Fe Fe Ni Ni Fe Ni Fe Ni Fe Ni Fe Ni Fe Fe Ni Ni Fe Fe Ni Ni bilayer bilayer 36 Figure 4-2. Sketches of the monolayer, bilayer, and multibilayer structures viewed from the side. Bilayer and multibilayer films were prepared using Y type Langmuir-Blodgett technique. Briefly, one bilayer of Fe-CN-Ni network was formed through a cycle of immersing and withdrawing a hydrophobic substrate into/out of the LB solu tion. The multibilayer was produced by repeating the immersing and withdrawing cycles for multiple times. The multibilayer in the figure is a result after 2 cycles. In a slightly different way, the monolayer of Fe-CN-Ni was gene rated by starting with the hydrophilic substrate immersed in the LB solution and pulling it out of the solution. and subtracted from the final result. Th e average magnetic background of the substrate (Mylar) was also measured separately and subt racted from the final result. For each film at each orientation, the temperature ( T ) dependencies of the zero -field-cooled (zfc) and field-cooled (fc) magnetization was measured in HE = 100 G for the multibilayer sample and in HE = 20 G for the monolayer and bilayer sa mples over the temperature range of 2 K to 15 K. The zfc measurements were pe rformed, first by cooli ng the sample from

PAGE 76

62 Figure 4-3. Schematic of Fe-CN-Ni square grid (top view). The Grazing Incidence X-ray Diffraction (GIXD) result indicates that the unit cell edge distance (distance between Fe ion and Fe ion in Fe-CNNi-NC-Fe configuration is about 10.2 For clarity CN bonding is expressed as a line. 300 K to 2 K without any field ( HE = zero), and then at T = 2 K, the external field was applied and the magnetization of the sample was measured upon increasing temperature. The fc measurement was performed in a similar way except HE was present during the initial cooling of the sample from 15 K to 2 K. For field dependent magnetization measur ements, the samples were zero-fieldcooled to 2 K and then the magnetization was measured while sweeping HE from 50 kG to -50 kG and back to 50 kG. In all of these measurements, the strong diamagnetic background signals from the holder and Mylar dominated the relatively weaker magnetic signals arising from the deposited films, esp ecially at high temperatures or in the large Fe Ni Ni Fe Fe Ni Ni Fe Fe Ni Ni Fe Fe Ni Ni Fe 10.2 10.2

PAGE 77

63 external fields. This strong background comp licated the data analys is at high field and temperatures. As a consequence, the absolu te saturation value of the magnetization was not well defined in each film. Figure 4-4 shows temperature dependent zfc and fc magnetizations of multibilayer film at two orientations (film || HE and film HE). A striking feature of the data is M ( HE) < M (|| HE). This anisotropy arises from the demagnetizing field effect from the twodimensional (2D) domain arrangements.6 Overall, the low temperature magnetizations rise rapidly below T ~ 10 K and deviations between th e zfc and fc magnetizations are observable below 5 K. In addition, below 5 K th e zfc values decrease while the fc values slowly increase with decreasing temperature. The rapid increase in magnetization below TC ~ 10 K is an indication of ferroma gnetic interactions between the FeIII ( S = 1/2) ions and NiII ( S = 1) ions. Although the TC value is lower in the film case by considerable amount, the magnetic interaction is consiste nt with the three-dimensional bulk NiFe(CN6) ( TC ~ 24 K) studied by others[15]. Furthe rmore, the difference between the zfc and fc values below T ~ 5 K and the monotonous increase of the fc values are evidence of spin-glass or cluster spin-gla ss-like behaviors. The deta iled results of this glassy behavior are presented in S ection 4.3 when AC magnetic meas urements are presented. It should be mentioned that the difference betw een the fc and zfc data between 5 K and 10 K in Figure 4-4 seems to be an artifact, which was not reproduced in another similar measurement. The field dependent magnetizations of the multibilayer film at T = 2 K are plotted in Figure 4-5. In both orientations of the film there exist hysteresis loops with coercive 6 A different view of this anisotropy concer ns anisotropic environm ents of single metal ions and the full discussion can be f ound in Reference [13], page 97 and 98.

PAGE 78

64 051015 0 5 10 15 fc (film || HE) zfc (film || HE) fc (film HE) zfc (film HE) HE = 100 GM (10-3emu G)T (K) 150multibilayer Figure 4-4. Temperature dependent magnetizati ons of 150-multibilayer film. The plot shows temperature dependent zfc and fc magnetizations of 150-multibilayer film at two orientations (film || HE and film HE) measured at HE = 100 G. A striking feature of the data is M ( HE) < M (|| HE). This anisotropy arises from the demagnetizing field effect from the two-dimensional (2D) domain arrangements. Overall, the low temperature magnetizations rise rapidly below T ~ 10 K and deviations between the zf c and fc magnetizations are observable below 5 K. In addition, below 5 K the zf c values decrease while the fc values slowly increase with decreasing te mperature. The rapid increase in magnetization below TC ~ 10 K is an indication of ferromagnetic interactions between the FeIII ( S = 1/2) ions and NiII ( S = 1) ions.

PAGE 79

65 -600-400-2000200400600 -3 -2 -1 0 1 2 3 film || HE film HEM (10-2emu G)H (G) 150multibilayer T = 2 K Figure 4-5. The field dependent ma gnetizations of 150-multibilayer film at T = 2 K. In both orientations of the film, there exis t hysteresis loops with coercive fields of ~135 G (when film || HE) and ~110 G (when film HE). fields of ~135 G (when film || HE) and ~110 G (when film HE). This hysteresis loop is another indication of magnetic glassy behavior of the film. Consistent with the case of

PAGE 80

66 temperature dependent magnetizations (Figur e 4-4), the magnetization values in the parallel orientation (film || HE) are higher than the values of perpendicular orientation. This difference show demagnetizing effect caused by 2D arrangement of magnetic domains. As a consequence, the ratio of measured magnetization values in the two orientations will be different from the unity, and thus this ratio w ill provide an idea about shape of domain arrangement. In Figure 44 and Figure 4-5, the magnetizations of the multibilayer in the parallel orientation are higher than the magnetizations of the film at perpendicular orientation. In particul ar, the value of fc magnetization at T = 2 K when film || HE is ~ 3.5 times higher than the value when film HE. This particular ratio of parallel magnetization over perpendicular magnetization at 2 K will be referred as shape ratio as it gives width to he ight information of the magnetic domains. For example if the shape ratio is close to infi nity, the height/width of domain is expected to be zero, or equivalently the shape of domain is like 2D sheet. The temperature dependent fc and zfc magne tizations of the bilayer film at two orientations (film || HE and film HE) are plotted in Figure 4-6. As it was the case of multibilayer, that the resulting values, when film HE, are smaller than the values when film || HE, the shape ratio at 2 K in this case is ~4.5, which is sli ghtly higher than the multibilayer case. The shapes of fc and zf c also look similar to those of multibilayer result: increase below 10 K and difference betw een the zfc and fc values below 5 K. The field dependent magnetizations of bilayer film at T = 2 K are plotted in Figure 4-7. In both orientations of the film, hysteresis loops exist with coercive fields of 75 G (when film || HE) and 55 G (when film HE). These hysteresis loop as was in the multibilayer case, is an indication of magnetic glassy behavior of the film.

PAGE 81

67 051015 0 2 4 6 HE = 20 G bilayer fc (film || HE) zfc (film || HE) fc (film HE) zfc (film HE)M (10-5 emu G)T (K) Figure 4-6. Temperature depende nt magnetizations of bilayer film. The plot shows temperature dependent zfc and fc magne tizations of bilayer film at two orientations (film || HE and film HE) measured. The magnetization values when film HE are smaller than the values when film || HE. This anisotropic effect is an indication of demagnetizi ng field effect arising from 2D like magnetic domain arrangements.

PAGE 82

68 -600-400-2000200400600 -3 -2 -1 0 1 2 3 bilayer T = 2 K film || HE film HEM (10-4 emu G)H (G) Figure 4-7. The field dependent magne tizations of bilayer film at T = 2 K. In both orientations, there exist hysteresis loops with coercive fields of ~75 G (when film || HE) and ~55 G (when film HE). This hysteresis loop is an indication of magnetic glassy behavior of the film. Consistent with the case of temperature dependent magnetizations (F igure 4-6), the magnetization values of parallel orientation (film || HE) are higher than the values of perpendicular orientation.

PAGE 83

69 As described previously, the magnetic properties of the bilayer film are qualitatively similar to ones exhibited by the multibilayer film. Moreover, the quantitatively normalized fc value in 150-multibil ayer at 2 K in para llel orientation (film || HE) is about 46 times higher than that of bilaye r at 2 K in the same orientation (film || HE). The fact that the synthesis of mu ltibilayer film went through 150 cycles of processing make the number 46 somewhat sm aller than the expected 150; however, considering non-linear behavior of the magnetization with respect to HE, the value 46 is reasonable. Therefore, from the result of DC magnetizations, two films (multibilayer and bilayer, see Figure 4-4 to Figur e 4-7) are both qualitatively and quantitatively similar. This result is not surprising when consideri ng the synthesis protocols of two films. According to the synthesis processes the multib ilayer film is equivalent to 150 stacks of bilayer and each bilayer is separated by ~ 35 Magnetically, this separation is sufficiently long so that the strong long-range magnetic ordering interactions between the bilayers might not be expected. Thus, it can be expected that multibilayer behaves similar to 150 stacks of isolated bilayer films. In contrast, the overall DC magnetiza tion properties of the monolayer are qualitatively different than the results from two other films, see Figure 4-8 and 4-9. Unlike multibilayer and bilayer f ilms, the magnetizations of monolayer films do not show the difference between zfc and fc below 5 K but show a hint of difference at T ~ 2 K. In addition, there is relatively lower TC ~ 6 K, where the magnetizat ion rises rapidly. Also, the difference between magnetizations in the parallel (film || HE) and perpendicular orientations is much larger (shape ratio is about 22.5) than the cases of the two other

PAGE 84

70 films, namely the multibilayer and bilayer films. The field dependent magnetizations of monolayer film at T = 2 K are plotted in Figure 4-9. 051015 0 1 2 M (10-5 emu G)T (K) HE = 20 G monolayer fc (film || HE) zfc (film || HE) fc (film HE) zfc (film HE) Figure 4-8. Temperature depende nt magnetizations of monola yer film. The plot shows temperature dependent zfc and fc magne tizations of monolayer film at two orientations (film || HE and film HE). The magnetization values when film HE is much smaller than the values when film || HE. This anisotropic effect is an indication of demagnetizing field effect arising from 2D like domain arrangements.

PAGE 85

71 -100-50050100 -3 -2 -1 0 1 2 3 film || HET = 2 K monolayerM (10-5 emu G)H (G) Figure 4-9. The field dependent magne tizations of monolayer film at T = 2 K. The coercive field of monolayer at 2 K is about 10 G, which is close to the resolution of the instrument.

PAGE 86

72 The coercive field of monolayer at 2 K is about 10 G, which is close to th e resolution of the instrument. 4.3 AC Field Magnetization Measurements The DC magnetizations of the bilayer and the multibilayer in the previous section showed some indications of spin-glass-lik e behavior such as different temperature dependent behaviors of zfc and fc magneti zations below a certain temperature and hysteresis loops. The spin-gla ss-like behavior can be expl ored more explicitly by AC magnetization studies. Briefly, the magnetization in the glassy state is dependent upon time, hence the time varying AC field probe s the temporal nature of the magnetization using periodic AC field at different freque ncies. For the AC measurement, the same QUANTUM DESIGN MPMS was used. All the samples were cooled to 2 K without any DC or AC field and then 4 G of AC field was applied to measure the AC magnetizations upon increasing temperatures. Figure 4-10 shows the real and imaginar y parts of AC susceptibilities of multibilayer film. The film was oriented parall el to the direction to the AC field (film || HAC) and the different frequencies (17 Hz, 170 Hz, and 1 kHz) of AC field were used for each measurement. Overall, the real parts of the AC susceptibili ties increase rapidly below T ~ 8 K, and reach the maximum values at a temperature TF and the decrease at lower temperatures. The magnit ude of AC magnetizations be low ~ 8 K is the highest at lowest frequency (17 Hz) and the lowest at the highest frequency (1 kHz). Also the temperature where the maximum AC magnetization occurs ( TF) depends on the frequencies, such that TF is smaller for lower frequencies. The qualitative result of the multibilayer film in the perpendicular orientation (not shown) is very similar to the result

PAGE 87

73 051015 0 1 2 3 4 5 f = 17 Hz f = 170 Hz f = 997 HzHAC = 4 G"AC (104emu)T (K) '150multilayer ( film || HAC) Figure 4-10. Temperature dependent AC su sceptibilities of 150multibilayer film. Overall, the real parts of the AC susceptibilities increase rapidly below T ~ 8 K, and reach the maximum values at a temperature TF and decrease at lower temperatures. The magnitude of AC ma gnetizations below ~ 8 K is highest at lowest frequency (17 Hz) and the lowest at the highest frequency (1 kHz). Also the temperature where the ma ximum AC magnetization occurs ( TF) depends on the frequencies, such that TF is smaller for lower frequency.

PAGE 88

74 0.20.30.4 2 4 6 8 10 monolayer0 ~ 5x1021 rad/s Ea/kB ~ 110 K bilayer0 ~ 5x1021 rad/s Ea/kB ~ 174 K 150multibilayer0 ~ 3x1029 rad/s Ea/kB ~ 347 KLn = (Ea/kB)/ TF+Ln0 = 0exp[Ea/kBTF]Ln (rad / s)1/ TF (1/K)Arrhenius law fitting Figure 4-11. Result of Arrhenius law fittings to monolayer, bilayer, and multibilayer.

PAGE 89

75 of the film at parallel orient ation. As was the case in the DC measurement of the same film, the magnitude of AC ma gnetization is higher in para llel orientation than in perpendicular orientation. The frequency dependent TF values of the multibilayer are a unique indication of spin-glass-like behavior. An empirical constant can be defined as a relative change of TF with respect to the change of corresponding frequencies, f such that = ( TF/ TF) / (log f ), (4.1) where f is a frequency [102]. Appl ying Eq. 4-1 to the real part of the AC measurement result of multibilayer (film || HE) yields ~ 0.04. The experimental values of the empirical constant differ in the limiting cases of s uperparamagnets and spin-glasses. The value ~ 0.04 falls into the regime of insu lating spin-glasses [102]. Two other empirical laws are often used to characteri ze spin-glasses and superparamagnets, namely, Arrhenius (Eq. 4.2) and Voge l-Fulcher law (Eq. 4.3), = 0 exp [Ea/kBTF] (4.2) and = 0 exp [Ea/kB( TF T0)]. (4.3) The Arrhenius law is an empirical law that is used to characterize the thermal activation energy ( Ea) of a superparamagnet. As shown in Figure 4-11, when fitted to the Arrhenius law (Eq. 4.2), the AC result of multibilayer film yields Ea/kB ~ 347 K and 0 ~ 3 x 1029 rad/s. These values are unphysic al considering the fact that 0 ~ 3 x 1029 corresponds to

PAGE 90

76 TeV in energy. Alternatively, the second law, Vogel-Fulcher law, describing interacting superparamagnets or magnetic clusters, yields Ea/kB ~52 K and T0 ~ 3 K, which is a measure of interaction between the clusters, when 0 was fixed at 2 x 1013 rad/s [53], see Figure 4-12. The values of Vogel-Fulc her fit make more physical sense and are consistent with the multibilayer film being id entified as an insulating spin-glass [102]. 5.45.55.65.75.8 -0.044 -0.042 -0.040 -0.038 -0.036 150multilayer0 = 2 x1013 rad/s (fixed) Ea/kB ~ 57 K T0 = ~ 3.3 K1 / Ln(/0)Tf (K) = 0exp[Ea/kB( TFT0)] Vogel-Fulcher law fitting Figure 4-12. Result of Vogel-Fulcher law fitting on multibilayer film.

PAGE 91

77 On the other hand, given that the fc magne tization increases monotonically at lower temperatures, the multibilayer may be classified as cluster-spin-glass [53]. 051015 0.0 1.0 2.0 3.0 4.0 5.0 ''' AC (10-7emu)T (K) HAC = 4 G monolayer ( film || HAC) f = 1 Hz f = 17 Hz f = 170 Hz f = 997 Hz Figure 4-13. Temperature depende nt AC susceptibilities of monolayer film. The film was oriented in parallel direction to the AC field (film || HAC) and the different frequencies (1 Hz, 17 Hz, 170Hz, and 1 kH z) of AC field were used for each measurement. Overall, real parts of AC susceptibili ties increase rapidly below T = ~ 7 K, then reach the peaks a nd returns The magnitude of AC magnetizations below ~ 7 K is highest at lowest frequency (1 Hz) and the lowest at the highest frequency (1 kHz).

PAGE 92

78 As shown earlier in the Section 4-2, th e DC magnetic properties of monolayer sample were qualitatively different than the behavior exhibited by the other two films. The same observation was made for the resu lts of the AC magnetization measurements. Figure 4-13 shows AC magnetization measur ement of monolayer in the parallel orientation (film || HAC). The AC signals of the monol ayer film in the perpendicular orientation (not shown here) were too small to be resolved in the magnetometer. The small or undetectable AC signals in perpendi cular orientation are consistent with a high shape ratio, the result of DC magnetization of the monolayer film in Section 4.2. It is striking to see that the absolu te peak AC magnitude of the monolayer is almost 500 times smaller than those of multibilayer in parallel orientation. Moreover, the TF values of the monolayer are relatively smaller than t hose of the multibilayer. The overall lower TF values in monolayer film can be interpreted as a lack of a cluster-glass like state. In other words, if we correlate the TF as a measure of interactions between the clusters and the size of the clusters, then the size and interactions of cluste rs in the monolayer are less pronounced than those of the multibilayer sin ce the magnetic pathways in the monolayer are confined to the plane. The empirical constant (Eq. 4.1), for the monolayer case is about 0.05, which is slightly higher than = 0.04 of the multibilayer, and closer to superparamagnet limit [102]. Due to the sca tter of the data, only two data points were used to fit the monolayer data to Arrhenius law as shown in Fig 4-11. The fit yields Ea ~ 110 K, which is smaller than the case of multila yer and reflects the qualitative difference between two films. Interestingly the shape of AC magnetizati on of bilayer sample resembles the data of the both monolayer and mutibilayer. Figur e 4-14 shows real and imaginary parts of

PAGE 93

79 AC susceptibilities of bilayer film. The film was oriented in parallel direction to the AC field (film || HAC) and the different frequencies (1 Hz, 17 Hz, 170Hz, and 1 kHz) of AC 051015 0 2 4 6 8 10 12 f = 1 Hz f = 17 Hz f = 170 Hz f = 997 Hz''' AC (10-7 emu)T (K) HAC = 4 G bilayer ( film || HAC) Figure 4-14. Temperature depende nt AC susceptibilities of bi layer film. The film was oriented in parallel directi on to the AC field (film || HAC) and the different frequencies (1 Hz, 17 Hz, 170Hz, and 1 kH z) of AC field were used for each measurement. Overall, real parts of AC susceptibili ties increase rapidly below T = ~ 8 K, then before they reach the T ~ 4 K peaks they slow down the increasing process around T = ~ 6 K. It seems that bilayer shows peak of monolayer at ~ 4 K and peak of 150-mu tibilayer at ~ 6 K as compared in Figure 4-15.

PAGE 94

80 246810 0 2 4 6 (x 1) (x 300) HAC = 4 G f = 17 Hz bilayer monolayer 150-multibilayerAC (104emu)T (K)film || HAC(x 150) Figure 4-15. Comparison of AC data at 17 Hz (mono, bilayer, and multibilayer). field were used for each measurement. Over all, the real part of AC susceptibility increases rapidly below T ~ 8 K and peaks at T ~ 4 K, with a shoulder T ~ 6 K. The

PAGE 95

81 shoulder corresponds to the peak of the 150-multibilayer film at ~ 6 K as compared in Figure 4-15. Using the peaks at around 4 K, Arrhenius law and empirical constant can be evaluated. The results are similar to the case of monolayer and yielded ~ 0.05 and Ea/kB ~ 174 K as shown in the Figure 4-11. 4.4 Magnetic Evolution versus Structural Evolution In the previous sections, the experimental results indicated that the magnetic behavior of the films differ from each other in both quantitative and qualitative aspects. These differences are expected to be related to the systematic differences of the film structures. Consequently, th is section focuses on the correlations between the magnetic and structural evolution of th e films. The experimental re sults of the characteristic DC magnetization values ( TC, HC, and shape ratio) are summarized in Table 4-1. Considering only parallel (film || HE) cases, the TC values increase as films evolve from monolayer (5.5 K) to multibilayer (9.9 K). The increment of TC value between mono layer to bilayer (5.5 K to 8.9 K) is larger than the increments of TC between bilayer and multilayer (8.9 K to 9.9 K). According to a mean field appr oximation first described by Langevin, Weiss, and Nel, the TC values in dinuclear st ructures (such as our Fe -CN-Ni systems) can be expressed as ) 1 ( ) 1 ( 3 M M M M B M M M M cS S S S k J Z Z T, (4.5) where, ZM is the number of the nearest magnetic atoms that interact with a central magnetic atom M, ZM is the number of the nearest magne tic atoms that interact with a central magnetic atom M JMM is the exchange coupling constant between the two nearest metal M and M ions, SM and SM are the spin values of atoms M and M

PAGE 96

82 respectively, and kB is the Boltzmann constant. Defining Z as an effective number of nearest magnetic neighbors such that ZM ZM = Z2, and assuming that Z = 5, the estimated JMM value using TC from the experimental result is JMM ~ 5 K in the multibilayer case. The value of JMM is expected to be approximately the same in all three films since the immediate local environments for all magnetic io ns in all three films are the same, i.e. the (NC)4 M CN M' (NC)4 environment dominates the structures. Using the value of JMM ~ 5 K and TC from Table 4-1, the Z value can be estimated as Z = 4.4 and Z = 2.7 for bilayer and monolayer samples respec tively. A noticeabl e increase in the Z value from Table 4-1. Characteristic values fr om DC magnetization measurements.7 samples orientation TC (K) HC (G) shape ratio || 9.9 135 multibilayer 10.5 110 22.5 || 8.9 75 bilayer 8.4 55 4.5 || 5.5 10 monolayer 6.1 N/A 3.5 Table 4-2. Characteristic values fr om AC magnetization measurements. samples frequency (Hz) TF (K) 0 (rad/s) Ea (K) 17 5.4 170 5.7 multibilayer 997 5.8 0.04 3 x 1029 347 1 3.7 17 3.9 170 N/A bilayer 997 N/A 0.05 5 x 1021 174 1 2.3 17 2.4 170 N/A monolayer 997 N/A 0.05 5 x 1021 110 7 The shape ratio is defined as M (||) / M ( ) at 2 K.

PAGE 97

83 the monolayer to the bilayer suggests that there are more magnetic inte ractions in bilayer, and this can arise from the intra-bilayer ma gnetic interactions. As mentioned earlier in Section 4.1, this intra-bilayer region was not chemically well characterized; however from the fact that bilayer has more magneti c interactions than monolayer, we conclude that the structure of intra-bi layer region is such that it provides an additional magnetic interaction between monolayers in bi layer (intra-bilayer interaction). Structurally, the multibilayer is a stack of bilayers separated by ~ 35 of long amphiphilic tails. As a consequence, the DC behavior of the bilayer is similar to the multilayer. However the magnetic properties of the two samples are not the same, and this fact is especially evident in the AC measurements, see Table 4-2. Microscopically, the bilayer sample, whose top surface is exposed to air is expected to possess different behavior than a bilayer sandwiched between ot her bilayers (as in multibilayer). This possible surface effect of the bi layer might be structurally an d magnetically significant to differentiate the bilayer film from the multibilayer ones. In addition, dipole interactions between the bilayers (inter-bilayer interactio n) might play a role to differentiate the multibilayer from the bilayer film. To investigate the possible role of the dipole interaction between the bilayers, we consider the z -component of the dipole field strength of a magnetically ordered disk (radius, R ~30 ) as a function of height at the center of the disk along the z -axis, see Figure 4-16. This s ituation mimics the dipole field generated by a bilayer in a multilayer sample, assuming that the disk represents coherent portion of the bilayer (for simplicity bilayer wa s treated as a single disk). For the result of the calculation plotted in Figure 4-16, the effective surface magnetization (total

PAGE 98

84 moments per cm2) of the disk was estimated by cons idering the structure of the face centered Fe-CN-Ni square network. The estimated value was ~ 1.4 x 10-5 emu/cm2. 05101520 0 50 100 150 200 dipole fieldheightH (G)Height (nm) R z HE || z Figure 4-16. Dipole field produced by a magneti zed disk (R~30 ) evaluated at different height along the z centered axis away from the disk. The situation mimics the dipole field generated by a bilayer in the multilayer sample assuming that the disk represents a coherent portion of b ilayer (for simplicity bilayer was treated as a single disk). The effective surface magnetization (moments per cm2) of the disk was estimated by considering th e structure of face centered Fe-CN-Ni square network. The estimated value was ~ 1.4 x 10-5 emu/cm2.

PAGE 99

85 An effective volume magnetization (total moments per cm3) was also estimated to be ~ 140 emu/cm3. To get a practical sense of th is value, the element Iron (Fe) ( TC ~ 1000 K) has a magnetization of ~ 1700 emu/cm3, which is an order of magnitude larger. From the result of calculation, the z component of dipole field pr oduced by a coherent portion of the bilayer at the distance of 36 perpendi cular from the center of the disk, is about 55 G. Generally speaking, this field strength produces a magnetic interaction energy of 10 mK, base on the approximation of eH ~ kBT but for ferromagnetically ordered molecules (or crystals) having la rger magnetic moments, the e ffect is stronger and can be an order of ~ 10 K in a multibilayer sample. The detailed dipole field distribution is not calculated here because it requires the know ledge of the exact distribution of the magnetic domains and detailed information on structure. Based on the discussion on this section and re ferring to the experimental results in the previous sections, a descrip tion of the magnetic processes of all samples can be made. When the temperature is well above TC of the multibilayer (9.9 K) all the films behave as paramagnets that follow a Curie law. When temperature equals TC of multibilayer or bilayer, the magnetic interactions within the bilayer exceed the thermal fluctuations of the systems, and the films orders ferromagnetically. In the mean time, the dipole field between the bilayers in the multilayer sample develops and provides an additional magnetic pathway between the bilayers. In addition, the ferromagnetic clusters start to interact each other, and these in teractions give a rise to clus ter-spin glass behavior that exhibits frequency dependent AC susceptibilities. Still, th e magnetic interactions in the monolayer sample are not well developed due to the lack of magnetic pathways. Upon further cooling, the monolayer starts to develop ferromagnetic clusters around 5 K and

PAGE 100

86 the interaction between clusters become su fficiently strong enough to show cluster-spin glass behavior around 2 K.

PAGE 101

87 CHAPTER 5 ANISOTROPIC PHOTOINDUCED MAGNETISM OF PRUSSIAN BLUE ANALOG FILMS The phenomenon of photoinduced magnetism in Co-Fe(CN)6 Prussian blue analogs was first discovered by Sato and coworkers in 1996 [38]. In the experimental studies, they observed that the Co-Fe(CN)6 material undergoes a photoinduced magnetic transition from a diamagnetic state to a ferrimagne tic state. More specifically, in the dark state, the bulk compound Co-Fe(CN)6 develops mainly two different magnetic phases: photoinducible diamagnetic phase and primordial magnetic phase8. A photoinducible diamagnetic phase contains regions of low spin (LS) FeII ( S = 0) and CoIII (LS, S = 0) pairs, and primordial magnetic state contains regions of FeIII (LS, S = 1/2) and high spin (HS) CoII ( S = 3/2) pairs, whose magnetic inte ractions become ferrimagnetic below TC. Upon irradiation, a charge transfer occurs from FeII (LS, S = 0) to CoIII (LS, S = 0) and as a result the photoinducible diamagnetic re gion becomes a ferrimagnetic region with FeIII (LS, S = 1/2) and CoII(HS, S = 3/2). Consequently, the Co-Fe(CN)6 material reduces its diamagnetic phase and becomes more ma gnetic under irradiation [39, 54, 55]. This chapter presents the anisotr opic photoinduced magnetic behavior in sequentially deposited RbjCok[Fe(CN)6]l n H2O films. The main goal of the chapter is to explore the unique photoinduced magnetic behavi or in the quasi-two-dimensional (quasi2D) system compared to the bulk-like th ree-dimensional (3D) system through the experimental evidence and propos ed theoretical desc ription. The chapter is divided into 8 Herein, primordial spins is used to desc ribe spins whose states do not change with temperature.

PAGE 102

88 three sections, and the synthesis and char acterization of two c ontrasting films of RbjCok[Fe(CN)6]l n H2O (film 1 and film 2) are presented in Section 5.1. The experimental study of photoinduced magnetic be havior of these two films are presented in Section 5.2. In Section 5.3, the theoretical model to de scribe the observed anisotropy in photoinduced magnetism is presented, as we ll as the experimental evidence supporting the proposed model. 5.1 Synthesis of Co-Fe(CN)6 Films In order to generate two films with different textures, a sequential deposition method was utilized. For sample film 1, a thin Mylar sheet (thickness ~ 100 m) was immersed sequentially into a 5 10-3 mol of cobalt(II) nitrate a queous solution and into a mixed solution of 2 10-2 mol potassium ferricyanide and 1.25 10-2 mol rubidium nitrate. The deposition cycle was rep eated for 20 times and each immersion was performed by dipping the Mylar sheet in and out of the solution 3 to 4 times. Sample film 2 was prepared similarly, however, for f ilm 2, the film was rinsed with water and methanol after each cycle of immersion. Nitr ogen gas was used to dry the films and the resultant films were cut into small squares (~ 10.5 mm2) and about 22 pieces were packed into a gelcap with Mylar and plastic straw spacers for magnetic measurement.9 One single piece of the film square wa s used for standard SEM images. In the synthesis process, film 2 went through an additional water and methanol rinsing procedure to remove excessive pow der-like surface residuals, whereas in the fabrication of film 1 the additional rinsing process was omitted. As a result, the SEM image in Figure 5-1 (a) indicates that film 1 is mainly composed of randomly deposited particles of Co-Fe(CN)6 and this deposition condition gives rise to a bulk-like texture. 9 See Figure 2-11 (h) for packing details.

PAGE 103

89 (a) (b) Figure 5-1. SEM images of film 1 (a) and f ilm 2 (b). For film 1, a thin Mylar sheet (thickness ~ 100 m) was immersed seque ntially in one solution (cobalt(II) nitrate) after the other (potassium fe rricyanide and rubidium nitrate) for 20 cycles, and each immersion was perf ormed by dipping the Mylar sheet in and out of the solution 3 to 4 times. Film 2 was prepared in a similar manner, except after each cycle of imme rsion, the film was rinsed by water and methanol. For film 2, the Co-Fe(CN)6 particles are continuously and smoothly arranged along the surface of the film, and this uniform arrang ement results in a quasi-2D texture within

PAGE 104

90 continuous patches separated by black lines10 in the SEM image, Figure 5-1 (b). The cracks between the patches might be defectiv e to the system. However, the network connected by these cracks shows a structure similar to the planar network nanostructure studied by Spiecker and coworkers [104], a nd thus suggests a future possibility of generating nanometer networks using this chemical method. 5.2 Photoinduced State Magnetism For photoinduced magnetic study, the samp les were placed in a homemade optic sample holder (Figure 2-10) and a Halogen light source was used to supply white light to the sample. The typical light power received at the sample was measured using a photodiode at the sample side and yiel ded a power between 1 and 2 mW / cm2 depending on the power level of the light source. The DC and AC magnetizations were measured using the Quantum Design SQUID magnetometers (see Chapter 2). For each sample film in the DC measurement, magnetization was meas ured as a function of time at 5 K with the external magnetic field HE = 200 G ( films) upon white light irradiation. In addition, the zero-field-cooled (zfc) and field-cooled (fc) magnetizations under dark and photoinduced states were measured as a functio ns of temperature from 5 K to 30 K with HE = 200 G ( films). For the case when the HE || film, similar measurements were performed only for film 2. The estimate d background magnetic si gnals arising from Mylar sheet and other component s were subtracted from the total magnetic values. As shown in Figure 5-2 (a) the time dependent magnetization of film 1 increased by about 50% of the dark state value under irradiation. For this measurement, film 1 was placed 10 These black lines (cracks) s eem to separate continuous cove rage of the material into small structural islands.

PAGE 105

91 Figure 5-2. Photoinduced magnetization of film 1. (a) The magnetization of film 1 is increasing upon irradiation of li ght. (b) Temperature dependent magnetizations of film 1 for HE film is shown. with HE (200 G) film at 5 K. This photoinduced enhancement in the magnetization is consistent with the results of bulk materials reported by ot hers [38, 42, 53] and confirms that film 1 possesses the powde r-like nature as shown in Fi gure 5-2 (a). Figure 5-2 (b) -50050100 1.8 2.0 2.2 2.4 2.6 2.8 Light off M (x 10-5 emu G / cycle)Time (Min) Light on HE = 200 G film T = 5 K(a)102030 0 1 2 3 HE = 200 G film photoinduced Mfc photoinduced Mzfc dark Mfc dark Mzfc M (x 10-5 emu G / cycle)T (K) (b)

PAGE 106

92 shows temperature dependent magnetizations of film 1 in both the dark and photoinduced states when HE (200 G) is perpendicular to the film surface. The monotonic increase in the fc values at low temperature and hist ory dependent zfc magnetization, where the magnetization shows peak value can be attribut ed to the behavior of cluster-spin glass behavior [53, 102]. Briefly, the cl uster-spin glass state of Co-Fe(CN)6 films can be described in terms of magnetically interacting clusters. Each cluster is mainly composed of a ferrimagnetically ordered FeIII (LS, S = 1/2) and CoII (HS, S = 3/2) region. The dark state TC ~ 15 K was enhanced to TC ~ 18 K as a result of light irradiation. Moreover, the corresponding temperature ( TF) of maximum zfc magnetization has also been enhanced from ~ 7 K in dark state to ~ 11.5 K in the photoindu ced state. Adapting the cluster-spin glass desc ription, these changes in TC and TF can be interpreted as the following. When the system is irradiate d, the photoinducible diamagnetic regions become ferrimagnetic regions and these additio nal ferrimagnetic regions result in the formation of new magnetic cluster as well as increase in size and concentration of primordial spin clusters. As a result of larger cluster forming, the magnetic correlation lengths are less restricted by th e structural disorder and responsible for the increase of TC in the photoinduced state. In the mean time, the increase in cluster concentration and size contribute to the enhancement of the interactio ns between the clusters and results in an increased TF for the photoinduced states. Similar behavior was reported by another group when the bulk material was studied [53]. Ther efore, from the direct SEM images and the comparisons to other powder magnetic studies by others, it can be concluded that film 1 is powder-like film in its structure and magnetism.

PAGE 107

93 Figure 5-3. Photoinduced magnetization of film 2. (a) The magnetization of film 2 is decreasing under irradiation of light. Temperature dependent magnetizations of film 2 when HE film is shown in (b). In contrast to film 1, when qua si-2D-like film 2 was placed in HE (200 G) film, the photoinduced magnetization decreased upon irradiation as shown in Figure 5-3 (a). In addition, when film 2 was oriented with HE (200 G) || film, the magnetic value of the film increased under light irradiation as shown in Figure 5-4 (a). This striking 102030 0.0 0.6 1.2 1.8 HE = 200 G film dark Mfc dark Mzfc photoinduced Mfc photoinduced Mzfc(b) M (x 10-5 emu G / cycle)T (K)050100 1.3 1.4 1.5 (a) HE = 200 G film T = 5 K Light off M (10-5 emu G / cycle)Time (Min) Light on

PAGE 108

94 anisotropic behavior of film 2, where the increase and d ecrease of the photoinduced magnetization occur according to the direction of the system with respect to the external magnetic field will be discussed in detail in the next section within the dipolar field description [50]. Figure 5-4. Photoinduced magnetization of film 2. (a) The magnetization of film 2 is increasing under irradiation of light. Temperature dependent magnetizations of film 2 when HE || film is shown in (b). -2002040 1.2 1.4 1.6 1.8 2.0 2.2 2.4 (a)HE = 200 G || film T = 5 K Light off Light on M (x 10-5 emu G / cycle)Time (Min) 102030 0.0 0.6 1.2 1.8 2.4 (b) photoinduced Mfc photoinduced Mzfc dark Mfc dark MzfcHE = 200 G || film M (x 10-5 emu G / cycle)T (K)

PAGE 109

95 Overall, the behavior of the temperature depe ndent magnetization data of film 2 is similar to the cluster-spin glass descri ption of film 1, Figure 5-2. As shown in Figure 5-3 (b) for HE film and Figure 5-4 (b) for HE || film, film 2 exhibits bifurcations between the zfc and fc magnetizations. In additi on, as a result of irradiation the TC and TF increased in film 2 and this change is similar to that of film 1. It is notewo rthy that the photoinduced increase of TC in film 2 ( HE film) is relatively smaller than the increases observed in film 1 ( HE film) and film 2 ( HE || film). Since the cluster-spin glass description involves dynamic aspects of magnetization, frequency dependent AC magnetic measuremen ts were performed to similar Co-Fe(CN)6 film sample but from a different batch. The AC measurement was performed only for dark state magnetizations. Nevertheless, th e existing primordial spin states provide noticeable glassy behavior. The frequency dependent AC susceptibilities in both film orientations (|| and with respect to the HE) as a function of temp erature are shown in Figure 5-5 for both real (a) and imaginary (b ) components. The frequency was varied from 1 Hz to 1 kHz in a decade interval and the strength of the AC magnetic field was 4 G in all measurement. For these measurements, the sample was cooled in zero magnetic field to the lowest temperature a nd the magnetic measurement was taken while increasing temperatures. In both orientations, the temperature ( TF) where the AC susceptibilities reach the maximum values shift towards higher temperatur es as frequency increases. These shifts in TF indicate dynamic aspects of magnetism in the material and the behavior can be described by the Vogel-Fulcher law = 0exp[-Ea/kB ( TFT0)] discussed earlier in Section 4-3. Briefly, the Voge l-Fulcher law describes a system of interacting magnetic

PAGE 110

96 5101520 0 1 2 3 4 (10-6 emu)T (K) 1 Hz 10 Hz 100 Hz 1 kHzHAC(4 G) || films HAC (4 G) films(a)5101520 0 1 2 3 4 '' (10-7 emu)T (K) HAC(4 G) || films HAC (4 G) films 1 Hz 10 Hz 100 Hz 1 kHz(b)5101520 0 1 2 3 4 (10-6 emu)T (K) 1 Hz 10 Hz 100 Hz 1 kHzHAC(4 G) || films HAC (4 G) films(a)5101520 0 1 2 3 4 '' (10-7 emu)T (K) HAC(4 G) || films HAC (4 G) films 1 Hz 10 Hz 100 Hz 1 kHz(b) Figure 5-5. Frequency dependent AC susceptibilities of Co-Fe(CN)6 films as a function of temperature in both || and film orientations. The real part of susceptibilities (a) and imaginary part of susceptibilities (b) show frequency dependent cluster-spin glass behavior. clusters with an interaction energy of T0. When the 0 was fixed to 2 x 1013 Hz [53], the Vogel-Fulcher law fitting yields Ea/kB ~ 45 K ( HE || film) and ~ 50 K ( HE film). In

PAGE 111

97 both orientations, the interaction energy T0 was approximately 9 K, see Figure 5-6. The quantity ( TF T0) / TF is proposed to classify the magnetic status of material [53]. For example, in the extreme case of non-interacting magnetic particles T0 = 0, and consequently ( TF T0) / TF = 1. Using T0 ~ 9 K from the fit, we obtained ( TF T0) / TF to be approximately 0.15, which is a similar va lue describing cluster-s pin glass behavior obtained from bulk Co-Fe(CN)6 Prussian blue analogs by others [53]. Figure 5-6. Vogel-Fulcher law fitting to Co-Fe(CN)6 films. When the 0 was fixed to 2 1013 [53], the Vogel-Fulcher law fitting yields Ea/kB ~ 45 K ( HE || film) and ~ 50 K ( HE film). In both orientat ions, the interaction energy T0 was approximately 9 K. 10.610.811.011.2 -0.045 -0.040 -0.035 -0.030 1/Ln(/0)TFEa/kB ~ 45 K T0 ~ 9 K HAC (4 G) || films Ea/kB ~ 50 K T0 ~ 9 KHAC (4 G) films

PAGE 112

98 5.3 Dipolar Field Model The anisotropic photoinduced magnetism of fi lm 2 can be described using a dipolar field model where the low-dimensional nature of the system and the dipolar magnetic fields generated by the primor dial domains play important roles. The interactive relationship between the dipolar field and th e dimensionality is shown schematically in Figure 5-7 for the case where HE film. When the sample is cooled below TC in HE, the ferrimagnetic primordial region and photoinducib le diamagnetic region can be situated next to each other as shown in Figure 5-7 (a). This arrangement is possible due to the random distribution of the local chemical composition such as Fe vacancies and interstitial alkali metal ions in film. In this situati on, the primordial region possesses a net magnetic moment, which has been aligned parallel to HE and produces its own dipolar field ( HD),. Consequently, the net magnetic fiel d seen by the photoinducible diamagnetic site is the vector sum of HE and HD, where HD is anti-parallel to HE. When illuminated, the diamagnetic site will become ferrimagne tic, and the direction of the photoinduced magnetization will follow that of the net field as shown in Figure5-7 (b) (where the HD > HE), and Figure 5-7 (c) (where the HD < HE). As a result, the net magnetization of the film in the photoinduced state decreases when HD > HE, and this expectation is consistent with our result for the photoinduced decr ease of the magnetization in film 2 as shown in Figure 5-3. When HD < HE as shown in Figure 5-7 (c), th e net magnetization is expected to increase, and this effect wa s observed experimentally in Fi gure5-8. In order to achieve the condition of HD < HE, two variables were adjusted independently. The value of HE was increased from 200 G (Figure 5-3) to 5 kG (Figure 5-8 (a)) and the observing

PAGE 113

99 Figure 5-7. Schematic description of the spin configurations of the domains in the film when HE surface. In (a), the primordial ferrimagnetic domains coexist with diamagnetic regions when cooled to T < TC before irradiation. The net magnetic field on the diamagnetic site is a vector sum of the external magnetic field ( HE) and the dipolar field ( HD) produced by the ferrimagnetic domains. In (b), at T < TC and HD > HE, the diamagnetic regions are magneto-optically converted to ferrimagnetic ones, and the newly formed ferrimagnetic net moments point in the z direction, resulting in a decrease of the net magnetization. In (c ), the photoinduced ferrimagnetic net moments point +z direction since HD < HE, and in this case, the net magnetization increases in the photoinduced state. HE HD mylar S = 0 Dark, Below TC, HE Film Photoinduced M Primordial M (a) (b) (c) light HD > HE HD < HE

PAGE 114

100 Figure 5-8. Photoinduced magnetization of film 2 when HE > HD. (a) HE = 5 kG. (b) T = 20 K. 0204060 6.5 7.0 7.5 8.0 (a) Light on HE = 5 kG film T = 5 K M (x 10-5 emu G / cycle)Time (Min) 02040 3 4 5 6 7 (b) HE = 200 G film T = 20 K Light on M (x 10-7 emu G / cycle)Time (Min)

PAGE 115

101 temperature was increased from 5 K (Figure 53) to 20 K (Figure 5-8 (b)). In both cases, the photoinduced magnetizations increased in contrast to Figure 5-3 and support the dipolar field model. Since the increase and decrease of photoinduced magnetization is related to the relative strength between the HE and HD in the dipolar field description, the field dependent magnetization was measured at 5 K, see Figure 5-9. As shown in the Figure 5-9. Field dependent photoinduced and dark stat e magnetization of film 2 when the HE film. Inset shows crossover field where HE ~ HD. 02468 0 20 40 60 80 100 0.00.10.20.3 0 20 40 M (10-5 emu G / cm2)H (T) M (10-5 emu G / cm2)H (T) Light Dark H E fil m s

PAGE 116

102 02468101214 0.04 0.06 0.08 HE = 20 TM (A.U.)Time (min) T ~ 4.5 K Light On Figure 5-10. Photoinduced magnetization at T ~ 4.5 K and HE = 20 T. inset, when the HE is around 1.5 kG, the photoinduced magnetization starts to increase more than the dark state magnetization. This crossover field can be interpreted as the field where HE ~ HD, and thus it gives rise to no ch ange in net magnetization of the sample even at the photoinduced state. For the extreme case of HD < HE, the

PAGE 117

103 photoinduced magnetization was measured us ing a vibrating sample magnetometer (VSM) at 20 T at the National High Magnetic Fields Lab (NHMFL). Although signal to noise ratio is low due to the relatively poor sensitivity of VSM, the photoinduced magnetization increased when the film was placed with its surface HE as shown in Figure 5-10.

PAGE 118

104 CHAPTER 6 CHARGE TRANSFER INDUCED SPIN TRANSITION IN PRUSSIAN BLUE ANALOG Recently, the phenomenon of charge-transfe r-induced spin transition (CTIST) has been reported for some Co-Fe(CN)6 Prussian blue analogs [46, 57]. As described earlier in Chapter 3, the basic phenomenon is hysteric, reversible changes in spin states of Co-Fe pairs from magnetic to diamagnetic upon cha nging temperatures. This phenomenon is similar to that of spin cross over (SC) transition [101] and ge nerates great interest due to the potential use for magnetic memory devices [80]. However, unlik e the SC materials, the CTIST materials undergo thermally induced charge transfers between two constituent metals. The Co-Fe(CN)6 Prussian blue analogs ar e also photoinduced magnetic compounds as presented in Chapter 5, and pres ently some theoretical descriptions or models are being developed to describe both thermally induced and photoinduced spin transitions [78, 79]. While investigating the photoinduced magnetism, we have observed unusual CTIST phenomenon when K0.6Co1.2[Fe(CN)6]4H2O powder was cooled rapidly to 100 K. In conventional CTIST magnets [46, 58], the spin states of rapidly cooled specimens bypass the spin transition and remain in the spin state of the high temperature (HT) phase, which is highly magnetic. Upon increasing temperature, the specimen suddenly releases all the en ergy from rapid cooling and follows the less magnetic low temperature (LT) phase that can be achieved by slow cooling. However in our CTIST investigation, the magnetizati on of the rapidly cooled K0.6Co1.2[Fe(CN)6]4H2O powder

PAGE 119

105 became lower than that of the LT phase and be came trapped in a new state, which we call the new low temperature (NLT) phase, w hose magnetic state is less pronounced than that of the LT phase. This effect has never been observed previously and is reported in this chapter. 6.1 Experimental Details The brown K0.6Co1.2[Fe(CN)6]4H2O powder was generated by first preparing three different solutions. Solution 1 contains 5.0 x 10-3 M of Co(NO3)2H2O and 0.291 g of Co(NO3)2H2O in 200 mL H2O. Solution 2 consists of 2.0 x 10-2 M of K3Fe(CN)6 and 0.3293 g of K3Fe(CN)6 in 50mL H2O. Solution 3 contains 1.0 x 10-1 M of KNO3 and 0.5055 g of KNO3 in 50mL H2O. By mixing solution 2 and 3, solution 4 was formed, and using a separatory funnel solution 1 was added to solution 4 over a period of 2 hours. After a powder formed, a centrifuge was used to collect the powder. The resultant powders were then dried in a vacuum oven at 373 K. Magnetic measurements were taken using a commercial MPMS (Magnetic Property Measurement System). A sample of 16.04 mg was packed between two gelcap tops (27.95 mg) and the background signal aris ing from the gelcap was estimated and subtracted in the data analys is procedure. Rapid cooling of the sample was obtained, by first setting the MPMS to a low temperatur e (100 K or 40 K) and then inserting the sample quickly from room temperature in a vacuum environment. In order to have uniform cooling and warming rates, the ma gnetization was measured using the sweeping mode of temperature control. In other wo rds, during the measurement temperature was continuously changed according to the set ra te. Each measurement was performed using a unique cooling and warming sequence, and thus the details of the sequence will be described in the results section.

PAGE 120

106 6.2 Experimental Results The magnetizations of K0.6Co1.2[Fe(CN)6]4H2O powder are presented in the format of T as a function of temperat ure in Figure 6-1. Upon slow cooling at 0.5 K / min between 310 K and 100 K, the T of the specimen starts to decrease rapidly (arrow a). However, while increasing the temperature at the same rate, T was nearly constant until it suddenly increases at ~ 210 K and becomes id entical to the value of the cooling curve above room temperature, see arrow b. These changes in T values are hysteric and the mechanism was proposed to be CTIST between CoII (S = 3/2) FeIII (S = 1/2) pair and CoIII ( S = 0) FeII ( S = 0) pairs [46, 57]. The previous studies indicate that these spin transitions also accompany a structural ch ange evidenced by the lengthening of the distance between the Co ion and N atom in a Fe-CN-Co unit. For the following discussion, the high magnetic state of the samp le near room temperature will be referred to as the high temperature (HT) phase, and the low magnetic state at lower temperature when slowly cooled will be referred to as the low temperature (LT) phase. When the sample was rapidly cooled (que nched) to 100 K, the warming data (arrow 1) shows that the specimen is still in th e HT phase determined by its high magnetic values. Upon warming at 0.5 K / min, however, the specimen quickly falls into a state much lower than the LT phase at a temperature around 160 K, and upon further increasing temperature, the sample returns to the HT phase again (arrow 5). The change in phases (from HT to LT, and HT) upon warming of a quenched sample is normally observed by other groups [58]. However, in our case, the quenched sample showed significantly less magnetization than the LT phase upon warming and reaches its minimum at ~ 210 K and this minimum NLT phase maintained upon cooling back to low temperatures (arrow 3). This new lowest state will be referred as in a new low

PAGE 121

107 Figure 6-1. Susceptibility time temperature as function of temperature. Arrow a and b indicate a slow cooling and warming at 0.5 K / min. At arrow 1, the sample was rapidly cooled from room temp erature to 100 K and upon increasing temperature, the magnetic value rapidl y decreased below the LT phase (arrow 2) and reached the minimum at ~ 210 K. The sample was then cooled again at the rate of 0.5 K / min (arrow 3) below 100 K. While warming the measurement was made (arrow 4 and 5). 100150200250300 1 2 3 4 T (emu K / mol of F.U.)T (K)HE = 100 G Rapidly cooled from room temperature HT HT LT NLT a b 1 2 3 4 5

PAGE 122

108 temperature (NLT) phase and can only be accessible by quenching and warming the sample. From experimental results, the magnetic properties of the specimen can be categorized into three different states, namely rapidly cooled state (effectively in a HT phase), slowly cooled state (in a LT phase), and rapidly cooled/warmed state (in a NLT phase). Figure 6-2 shows the low temperatur e magnetic properties of the sample in three different states. In temperature dependent magnetization data, see Figure 6-2, all of the states showed differences between zero-field -cooled (zfc) and field-cooled (fc) magnetic properties. These differences indicate that th e specimen is still in a magnetic glassy state even though the magnitudes of magnetization in each state are different. For example, the magnetization of the HT phase is highest a nd that of the NLT is lowest. Furthermore, the TC and the temperature where the zfc magnetization reaches its maximum ( TF) shifted toward higher temperatures as the system we nt from the NLT to the LT, and HT phase. This situation is somewhat similar to the photoinduced magnetization presented in Chapter 5, and by using the same argument of increasing the number of magnetic neighbors, the TC and TF shifts can be understood. The field dependent magnetizations in the three states are also shown in Figure 6-3. The magnetization value of the HT phase at HE = 7 T is about twice that of the NLT phase and indicates that the HT phase contains the absolute number of spins, about twice that of NLT phase, which is consistent with the result of the T vs. T analysis. 6.3 Discussion and Future Directions Considering the formula of K0.6Co1.2[Fe(CN)6]4H2O, at high temperature the specimen should contain a net spin value of 2.3 per formula unit. This value came from the sum of Co spins (1.2 x S = 3/2) and Fe spin ( S = 1/2). Taking the g value for Co as

PAGE 123

109 Figure 6-2. Low temperature magnetizations meas ured at three differe nt cooling states. all of the states showed differences be tween zero-field-cooled (zfc) and fieldcooled (fc) magnetic properties. These differences indicate that the specimen is still in a magnetic glassy state ev en though the magnitudes of magnetization in each state are different. 51015202530 0.0 0.5 1.0 1.5 2.0 M (103 emu G / mol of F.U.)T (K)fc, rapidly cooled (HT) zfc, rapidly cooled (HT) fc, slowly cooled (LT) zfc, slowly cooled (LT) fc, rapidly cool/warm (NLT) zfc, rapidly cool/warm (NLT) HE = 100 G

PAGE 124

110 Figure 6-3. Magnetizations of thr ee different magnetic states as a function of the external magnetic field. The magnetization value of the HT phase at HE = 7 T is about twice that of the NLT phase and indicates the HT phase contains the absolute number of spins, about twice that of NLT phase, which is consistent with the result of the T vs. T analysis. 02468 0 5 10 15 M (103emu G / mol of F.U.)H (T)Rapidly cooled (HT) Rapidly cool/warm (NLT) Slowly cooled (LT) T = 5 K

PAGE 125

111 2.6 and that of Fe as 2, the T value at the paramagnetic limit, which is the same as the Curie constant is ~ 4.2 emu K / mole of formul a unit, which is close to the experimental T value of 4.3 emu K / mol, see Figure 6-1. Therefore, as a firs t approximation, it can be assumed that all magnetic states of the sample are in the HT phase. However, exact details of the spin states at room temperat ure can only be determined using microscopic probes. In addition, the CTIST can only occur within a single of Co-Fe pair, and thus the unpaired Co ions (17% of total Co spin) is assumed to be in the HT phase at all temperatures. When this assumption is made, the lowest T value for the specimen at room temperature can be estimated to be the T value of unpaired Co ions and yields T ~ 0.6 emu K / mol. When slowly cooled, the T value of the sample decreased rapidly in the narrow temperature window and became ~ 2.4 emu K / mol at 100 K. Since the charge transfer can only occur as a pair, the loss of magneti zation at 100 K in the LT phase should come from the CTIST. From the previous a ssumption on the residual HT phase, this T value at 100 K indicates that about 51% of spin s in the HT phase at room temperature transferred into diamagnetic states. Furtherm ore, in the NLT phase sample, about 73% of HT phase spins transferred to diamagnetic states. The question such as, why the LT phase does not undergo a complete CTIST to the minimum magnetic state has not been answered yet. In addition, the unusual beha vior of the NLT phase that completed 22% more CTIST than that of LT phase is an interesting theo retical question. Recent study, by Bleuzen and coworkers, have focused on a similar CTIST compound Cs0.7Co4[Fe(CN)6]2.9H2O [77], in which only the kind of alkali metal and metal concentrations are different than our sample, K0.6Co1.2[Fe(CN)6]4H2O. In their

PAGE 126

112 study, they have observed a significant change in the position of the alkali metal between the HT and LT phase. Also, recent theoretic al investigation suggests that the spin transition phenomenon is influenced by the magnetic interactions [78], which is a surprise considering that the CT IST happens much higher than TC (~ 20 K). The unusual behavior of our NLT phase might have b een associated with the aforementioned movement of alkali metal and consequent coope rative structural beha vior. However, the exact qualitative description is not fully understood yet. Since 1991, the discovery of the new ma gnetism in Prussian blue analogs blossomed. From room temperature magne ts to photoinduced and thermally induced magnets, the surprises come each year in this seemingly simple structure. On the other hand, compared to many experimental discoverie s, less the theoretical investigations have been made until the very recent developm ents [78, 79]. The unusual CTIST of K0.6Co1.2[Fe(CN)6]4H2O was never been observed before, and thus no theoretical work exists to this date. Nevert heless, the phenomenon is unique in the area of fundamental phase transition and device applications. T hus, until the theory becomes available, the experimental probing should be made using microscopic investigating methods, with which the correlation between the structural transition and magnetic transition can be studied.

PAGE 127

113 CHAPTER 7 HETEROSTRURED LAYERED MAGNET One of the most fascinating aspects of mol ecule-based magnets is related to the fact that they can be rationally designed to st udy fundamental physical phenomena such as interactions in 1D spin chains [105, 106], ef fects in 2D frustrated spin planes [107], glassy behavior in the spin glasses, Bose -Einstein condensation as in the condensed triplons realized in gapped sp in systems [108], and quantum tunneling of spins in singlemolecule magnets [109]. Furthermore, th e molecule-based magnets not only provide suitable models for fundamental studies but garner great attention for potential future magnetic applications. For example, the phenomena of photoinduced magnetism in Prussian blue derivatives [38] can be used to generate a proto-type of an optically controlled magnetic switch, which might even tually provide a connection to quantum information storage devices. However in ge neral, synthesizing the desired moleculebased magnets faces two challenges. The firs t challenge is to theo retically predict the magnetic interactions in the target molecule and the second difficulty is related to the actual chemical synthesis techniques require to produce the system. As an effort of overcoming these theore tical and experimental difficulties in designing new molecule-based magnets, we c onsidered the variations from the existing chemical synthesis methods and the uses of pr eviously studied materi als. This chapter introduces two new possible methods of desi gning molecule-based ma gnets utilizing the existing expertise of the sequential deposit ion (SD) method and the Langmuir-Blodgett (LB) technique [9 14]. Furthermore, us ing these new methods, the possibility of

PAGE 128

114 constructing conceptual magnets has been c onsidered. For example in Section 7.1, after the introduction of a new SD method, a cons truction of heterostructured layered high TC photoinduced magnet is discussed and a prelimin ary study of its proper ties is presented. In Section 7.2, a new application of LB technique to construct 2D magnetic nanometer islands is proposed. Finally, Section 7.3 su mmarizes the chapter and suggests future directions. 7.1 Heterostructured Layered Films The beauty of the sequential deposition (SD) method is in its simplicity of synthesizing a thin solid film. For example, a typical Pr ussian blue derivative having the AiM j[M(CN)6] k n (H2O) structure, where A is an alkali metal and M and M are transition metals, can be generated by dippi ng a substrate sequentially into a solution containing M metal ions and then in to the other solution containing M metal ions. The fact that each cycle of this SD method creates a quasi-2D monolayer makes it possible to build a solid sample from layer by layer. The interesting sample can be made if we deposit foreign or impurity layer between each host layer of AiM j[M(CN)6] k n (H2O) is made. We can do this deposition by repeating a cycle of a monolayer of AiM j[M(CN)6] k n (H2O) and then a cycle of a foreign layer, provided that the foreign layer chemically bonds to the host layers. Th e thickness and the ver tical arrangement of the host and foreign layers can be controlled easily by varying th e number and order of the deposition cycles. More spec ifically, to construct a thin film of 3 layers of host material, 5 layers of foreign material, and 2 layers of host material we can go through SD process of 3 cycles for host material solutions 5 cycles for foreign material solutions, and 2 cycles for host material solutions. In princi ple, we can use variety of foreign materials to create different structures of thin bulk specimens raging from homogeneous layers of

PAGE 129

115 AiM j[Mk(CN)6] n (H2O) to one giant supramolecule of many different foreign layers. Therefore, applying th e SD method to different material s in a controlled manner provides unique constitutes of different la yers in one solid as if we build a block of bricks using various colors and shapes of LEGO sheets. The Prussian blue analogs of the AiM j[M(CN)6] k n (H2O) variety are excellent materials for SD layer by layer construction because of the well defined octahedral building unit of M(CN)6 and the variety of transition metals available for M and M In addition the magnetic and structural properties, a wide variety of bulk (i.e. powdered) Prussian blue analogs have been reported. In the previous paragraph, we used AiM j[M (CN)6] k n (H2O) as a host layer, here forth called the MM layer, and if we use AiX j[X(CN)6] k n (H2O), hereafter referred to as the XX layer as a foreign layer the chemical bonding between the host and foreign layers is made because both have similar CN bridged face-centered-cubic structures in bulk state. However, despite of the similarity of structures, the magnetic properties of AiM j[M(CN)6] k n (H2O) and AiX j[X (CN)6] k n (H2O) could be different. For instance, one could be a diamagnet whereas the other co uld be a room temperature ordered magnet. Figure 7-1 shows a conceptual schematic of th e newly designed heterostructured layered Prussian blue analogs of different metal combinations of MM and XX This magnet is synthesized by going through the se quential deposition in M, M X, and X solutions. In the resultant film, the MM layer of AiM j[M (CN)6] k n (H2O), and XX layer of AiX j[X(CN)6] k n (H2O) are alternatively stacked together making a one big molecule. Since each layer is homogeneous in its chemical composition, an in-plane magnetic interaction arises and the interaction can be expressed as effective exchange integrals

PAGE 130

116 JMM in MM layer and JXX in XX layer. The values of in-plane interactions ( JMM and JXX ) are expected to be compatible to the pr edictions of Nishino and coworker and with the experimental results of other in 3D materials. However, the expected TC of a single monolayer will be less than the TC value of the 3D bulk materi al [1] of the same kind. This difference in TC values between the monolayer (2D in nature) and bulk materials can be understood by considering a decrease of nearest magnetic neighbors in 2D structure when compared to the 3D structure. Furtherm ore in this alternating structure (Figure 7-1), we expect interlayer magnetic interaction JMM -XX through CN bridges between two different planes of MM and XX The next-nearest interactions are also expected between the layers of the same kind ( JMM 1-MM 2 for between MM 1-MM 2 planes and JXX 1-XX 2 for XX 1-XX 2 planes). With these five basic interactions ( JMM JXX JMM -XX JMM 1-MM 2, and JXX 1-XX 2) as the magnetic ingredients, it is possible to cr eate many interesting magnetic systems with the layer-by-layer solid deposition technique. One of them is a higher TC photoinduced magnet. The recipe for this syst em involves two alternating Prussian blue magnetic planes (MM and XX planes). If we construct the MM plane using a photoinduced magnet such as RbjCok[Fe(CN)6]l n H2O ( TC ~ 20 K presently) and the XX plane with a relatively higher TC Prussian blue analog, an alternating photoinduced and high TC heterostructured layers wi ll result. In the dark stat e, the photoinduced magnetic layers (MM 1 and MM 2) will be diamagnetic and the XX planes will exhibit lower TC values due to the absence of a magnetic pathway in the z -direction. At this point, the

PAGE 131

117 z JMM'-XX' JMM'1-MM'2 JXX'1-XX'2Substrate x yX X 1X X 2JMM'JMM'JXX'JXX' M M 2 M M 1 z JMM'-XX' JMM'1-MM'2 JXX'1-XX'2Substrate x yX X 1X X 2JMM'JMM'JXX'JXX' M M 2 M M 1 Figure 7-1. An alternating plan ar structure of Prussian bl ue analog. The monolayers of AiM j[M (CN)6]k n (H2O) (MM plane) and AiX j[X (CN)6]k n (H2O) (XX plane) are alternatively stacked toge ther using sequential deposition method. inter-layer interaction JMM -XX is weak since the MM planes are diamagnetic, but when we irradiate the system with light the diamagnetic state of the MM layer will become magnetic and provides additional magnetic pathways to the XX layers in the z -direction via the JMM -XX interaction. These enhanced magne tic interactions might alter the TC and magnetizations of the system to higher or lo wer values. Regardless of the modification of TC a highTC molecule-based magnet, that can be controlled by irradiation, will have been generated. Potential variations of th e above mentioned structure are obvious. For example, the thickness of diamagnetic layer can be tuned by repeating more deposition cycles and studied for the interaction betw een the 2D magnetic layers as function of separation distance in the z -direction. Also, if we incr ease the separation between the

PAGE 132

118 magnetic layers, we can possibly study the phys ics of isolated 2D magnetic systems such as correlation length in vari ous situations (ferromagnetic, antiferromagnetic, frustrated, and other situations) using differe nt combinations of metal ions in Prussian blue analogs. As a preliminary attempt, we have gene rated two different but related magnetic films (film 1 and film 2) us ing the sequential deposition method. Briefly, film 1 was designed to be 20 layers of RbxNiy[Cr(CN)6]zm H2O (Ni-Cr) film, and film 2 was to be a total 20 layers of alternating RbxNiy[Cr(CN)6]zm H2O and RbjCok[Fe(CN)6]ln H2O (CoFe) layers. For the syntheses, four diffe rent solutions were prepared, Ni (Ni(NO3)26H2O), Cr (K3Cr(CN)6 + RbNO3), Co (Co(NO3)26H2O), and Fe (K3Fe(CN)6 + RbNO3). The first monolayer of film 1 was generated by a cycle of immersing a substrate (Mylar film) sequentially into Ni-sol ution and Cr-solution, and then the immersing cycles were repeated for total 20 times to fabricate th e final Ni-Cr film, see Figure 7-2(a). An additional immersion of the Ni -Cr layered system into the Co-solution and Fe-solution completed one cycle of film 2, and the proce ss was repeated for total 10 cycles to make alternating Ni-Cr / Co-Fe film 2, see Figure 7-2(b). The ingredients of film 2 consist of Ni-Cr, which is a relatively high TC (~ 90 K) molecule-based magnet in its bulk form [1], and Co-Fe, which shows photoinduced magnetism ( TC ~ 20 K) as described earlier in Chapter 5. The preliminary magnetic measurement of film 2 was performed mainly to observe the magnetic effect of the Co-Fe layers in the Ni-Cr host. The zero-field-cooled (zfc) and fiel d-cooled (fc) magnetizations of film 1 are shown in Figure 7-3. Looki ng at the magnetic response as a function of decreasing temperature, both zfc and fc magnetizations rise ra pidly around 84 K, but

PAGE 133

119 start to deviate from each other below ~ 40 K, and eventually the zfc magnetization falls below ~ 20 K. Figure 7-2. Schematics of synthesizing Ni-Cr an d Ni-Cr / Co-Fe films. Four different solutions were prepared: Ni (Ni(NO3)26H2O), Cr (K3Cr(CN)6 + RbNO3), Co (Co(NO3)26H2O), and Fe (K3Fe(CN)6 + RbNO3). The first monolayer of film 1 was generated by a cycle of immersing a substrate (Mylar film) sequentially into the Ni solution and the Cr solution (a). An additional immersing of Ni-Cr monolayer, to Co solution and Fe solution completed one cycle of film 2 (b). The right side of figures shows schematics of resultant films after 2 cycles. Fe Ni Cr Ni Cr Co substrate Ni-Cr Co-Fe Ni-Cr Co-Fe Ni-Cr substrate Ni-Cr after 2 cycles after 2 cycles (a) Film 1 : Ni-Cr (b) Film 2 : Ni-Cr / Co-Fe

PAGE 134

120 Figure 7-3. Temperature depende nt magnetizations of film 1 (Ni-Cr) and film 2 (Ni-Cr / Co-Fe) at H = 100 G. Both zfc( ) and fc( ) magnetizations of film 1 rise rapidly around 80 K but start to deviate from each other below ~ 40 K, and eventually zfc magnetization falls down below ~ 20 K. On the other hand, fc ( ) magnetization of film 2 is smaller (~ 20 times smaller at T = 5 K and at H = 100 G) than film 1 and the shape of the da ta is different than that of film 1. The inset shows an attempt of photoi nduced magnetization measurement of film 2. The magnetization was measured as a function of ti me while light was illuminated to sample at T = 5 K with H = 1 kG. The result indicates no significant photoinduced ma gnetization but thermal heating effect when the light was on and off. On the other hand, as shown at the bottom of Figure 7-3 the fc magnetization of film 2 is significantly smaller (~ 20 times smaller at T = 5 K and at H = 100 G) than film 1, and the trend of the data is different th an that of film 1. A direct magnitude 020406080100120140 0.0 0.5 1.0 1.5 03060 3.25 3.30 M (10-3emu G)T(K)M (x 10-4 emu G)Time (min)light on light off film 1: Ni-Cr film 2: Ni-Cr / Co-Fe fc film 1 zfc film 1 fc film 2

PAGE 135

121 comparison between the two films is possible because both films have about the same area and thickness. If film 2 was simply th e physical mixture of Ni-Cr and Co-Fe layers, the total magnetization of film 2 would be expe cted to be at least the half the magnitude of film 1, since film 2 contains a total 10 cy cles of Ni-Cr, whereas film 1 contains 20 cycles of Ni-Cr. In addition, there was no significant photoinduced ma gnetization in film 2 as shown in the inset of Figure 7-3, where the magnetization was measured as a function of time while light illuminated the sample at T = 5 K and H = 1 kG. In a typical pure Co-Fe film, photoinduced magnetization is e xpected at this temperature as described in Chapter 5. Therefore, these facts, namely that there are quantitative and qualitative differences between two sample films a nd that no significant photoinduced magnetism was observed in film 2, confirm that the pr eliminary synthesis pr otocol successfully generated a new film, in which the layers of Ni-Cr and Co-Fe ar e not just physically layered together but also chemi cally bonded at the molecular leve l to give rise to different magnetic behavior. 7.2 Conceptual 2D Magnetic Nanometer Island In order to construct a 2D nanometer island, LB film maki ng technique can be used. A strategy of generating a LB monolayer of Prussian blue analog at the air-water interface is described in detail in Reference [9 13]. In general, to make a LB film a solution containing LB units is introduced to a Teflon LB trough, which contains LB subphase solution, see Figure 7-4 (a). The LB trough, also equipped with barriers that can be moved to squeeze the water surface as means of increasing surface pressure. The LB unit is composed of a head (solid circle in figure) and hydrophobic tail (wavy line). In the beginning stage the LB units are not bonded to each other and float at the air-water interface individually because of the hydrophobic tail. As two barriers move toward the

PAGE 136

122 center of the trough, as shown in Figure 7-4 (b ), some of the LB units squeeze together and aggregate to form a cluster, while some units still remain unbounded. When the LB units form a cluster, they bond to each other with help of the subphase and the cluster becomes a 2D solid. Depending on the propertie s of the LB head, this solid can be an atomic thick 2D LB island or thicker quasi 2D structure. When this cluster formation occurs, the system is the mixture of liquid a nd solid phase of LB units. As we increase the surface pressure by squeezing the barriers, a ll the clusters and individual LB units are bonded together with the subphase in between the units at the air-water interface, see Figure 7-4 (c). At this point, the surface of the water inside the barriers is covered by monolayer of bonded LB unit network. With a further reduction of the available surface area, the monolayer of LB units breaks and the cl usters start to stack together as shown in Figure 7-4 (d). When this stacking happens, the system is analogous to an amorphous 3D solid. At each stage, we can insert a hydrophobic substrate into the LB solution and when the substrate passes the air-water interface, th e hydrophobic part of the LB unit (tail) will be attached to the substrate. The resultant film of each stage is shown on the right hand side of Figure 7-4. When the LB units are in liquid phase Figure 7-4 (a), only a small fraction of the LB units will be deposite d to the substrate because of the small concentration of LB units pe r surface area and a lack of bonding force between the tail and substrate. The situation is similar even when some of the LB clusters form, see Figure 7-4 (b). However, when the LB units are in a solid phase at the air-water surface, most of the network will be transferred into th e substrate as we insert the substrate slowly and at the same time as we apply surface pressure, see Figure 7-4 (c). After the substrate

PAGE 137

123 Figure 7-4. Schematic of LB film generation. In general, to make a LB film a solution containing LB units is introduced to a Teflon LB trough, which contains LB subphase solution (a). Upon squeezing th e barriers toward the center of the trough the LB units form clusters (b), we ll defined network (c), and stacks of clusters (d) at the air-wa ter interface. The right si de show the deposited LB unit at each stage. is inserted, if we withdraw the substrate wh ile maintaining a high su rface pressure via the barriers, a Y-type LB film which is bilayer system as described in Reference [13], will be d) a) c) b) barrier LB trough A-type LB unit substrate subphase subphase subphase subphase

PAGE 138

124 obtained. However, if we released the barrie r and withdraw the substrate, we will have a 2D LB monolayer deposited onto the substrate as shown in Figure 74(c). The state of mixed liquid and solid phases of LB units as shown in Figure 7-4(b) and again in Figure 7-5 (b), contains some clusters of LB un its and some individual LB units. Depending on the extrinsic and intrinsic factors such as temperature, the concentration of the LB solution per surface area, and the bonding force between the LB units through the subphase, the size of cluster can be varied, and in some cases, we can expect this size to be of the order of a nanometer. Therefore if we insert the substrate at this point, we can obtain nanometer size LB island deposited on to the substrate. However, since the barriers do not provide the surface pressure of the water, the nanometer island deposition will not be efficient. On the other hand, if we apply excessive surface pressure, the system will be in a 2D monolayer whose size is limited by the size of substrate, which is much larger than the nanometers. In orde r to overcome this difficulty, a new method introduces dummy LB units such as Phospho lipid as shown in Figure 7-5 (c). For convenience, the existing LB unit of interest is labeled as A-type and the dummy LB unit is labeled as B-type. With this introd uction of B-type LB units, upon increasing the surface pressure, the size of the cluster is we ll preserved because the Btype LB units are filling the space between the clusters as shown in Figure 7-5 (d). At each stage of LB layer formation (Figure 7-5), the substrate can be inserted and the final resultant film of each stage is shown schematically in Figure 7-6. As shown in Figure 7-5(d), introducing the dummy LB units helps the or iginal LB units to form as clusters and be deposited onto substrate. The final product is a 2D A-t ype LB island in the sea of 2D B-type LB network. The application of this modified LB technique is useful to create various

PAGE 139

125 Figure 7-6. Deposition of nanometer LB islands on substrate corresponding each stage of Figure 7-5. front top a) b) c) d) substrate A-type LB unit B-type LB unit

PAGE 140

126 Figure 7-7. Deposition of nanometer particles to substrates. Nanometer particles can be deposited by using LB islands as a su bstrate of sequential deposition method (a). Nanometer particles can be also deposited to the substrate using general LB technique where the particle s act as LB units (b-d). nanometer sized magnetic islands. For example, if A-type LB unit is magnetic and Btype is not, we have a magnetic 2D nanometer island in the nonmagnetic solid substrate. b) a) c) d) substrate LB nano particle barrier LB trough substrate (top view)

PAGE 141

127 Conversely, if the B-type is magnetic and Atype is not, we create nanometer size vacancies in the 2D magnetic network. 7.3 Summary and Future Directions In the previous two sections, modifi ed methods of generating new magnetic materials were discussed. In Section 7.1, a modified seque ntial deposition was used to generate a heterostructured layered magnetic film. The ma in point of the modification was in a use of different chemical compositions for the different layers to construct one thin solid. Using this technique, we have attempted to generate high TC photoinduced magnet. In our attempt we constructed a thin film consisting mixed layers of RbxNiy[Cr(CN)6]zm H2O and RbjCok[Fe(CN)6]ln H2O. The preliminary magneto-optic study of this heterostructured layered film sh owed that the film was different than the pure RbxNiy[Cr(CN)6]zm H2O or RbjCok[Fe(CN)6]ln H2O film and indicated the presence of magnetic interactions be tween the mixed layers. In Section 7.2, a modified LB technique was introduced to produce a 2D magnetic nanometer islands. The introduction of an insert LB unit played a key role to isolate each magnetic cluster as nanometer island and to deposit the system onto the substrate. The modified LB method in Section 7.2 can al so provide good template for sequential deposition method. For example consider the 2D LB nanomete r island in Figure 7-6 (d), where A-type is RbjCok[Fe(CN)6]ln H2O, a Prussian blue based photoinduced magnet and the B-type is just insert material whose st ructure is not similar with Prussian blue derivatives (face-centered-cubic structure). Using a substrate of this 2D nanometer magnetic island, we can apply a sequential deposition method to deposit more of RbjCok[Fe(CN)6]ln H2O. Since the B-type part is not similar to the Prussian blue

PAGE 142

128 structure, the sequentially deposited material only will remain on top of the A-type island as shown in Figure 7-7(a). As another futu re direction, nanomete r particles surrounded by hydrophobic tails can be also considered to be deposited on the substrate as shown in Figure 7-7(b d). For this nanometer part icle deposition, first we introduce the the particles with hydrophobic tails into the LB s ubphase. After introducing the particles to the trough, the surface pressure can be increased to insert the substrate into the trough. This insertion of substrate will result in deposition of magnetic array, in which the nanometer particles are closely packed but separated by hydrophobic tails or dummy LB units. All the efforts in this chapter ar e basically focused on generating new magnetic materials using atomic scale molecular mani pulations based on a bottom up approach, such as the sequential deposition method or the LB technique. In the future, the combination of chemical methods and the conventional top down approach will be important for the creation of systems of interest.

PAGE 143

129 CHAPTER 8 SUMMARY AND FUTURE DIRECTIONS This chapter summaries the main experiment al results from the principal topics in this dissertation. The following subsections are divided into two sections according to the investigated materials, namely Ni-Fe(CN)6 and Co-Fe(CN)6 systems. In addition to the summaries, potential improvements and exte nsions are discussed, and possible future directions are suggested. 8.1 Ni-Fe(CN)6 Films 8.1.1 Summary A dimensional evolution of the magnetic properties was realized in Ni-Fe(CN)6 Langmuir-Blodgett (LB) films as the film structure evolves from monolayer (2D) to bilayer (quasi-2D), and to multibilayer (3D). The structural studies confirmed that the size of the face-centered unit square is about 10.2 (edge distance) in each layer and the distance between the bilayers is ~ 35 whic h is about the length of two of amphiphilic tails that link between the bilayers. Mo reover, the structural coherence length was determined to be ~ 60 according to the grazing incidence X-ray diffraction (GIXD) result. The exact structure between the monol ayers within the bila yer is not yet known. The main magnetic interactions of the sa mple originates from the superexchange interactions between the FeIII ( S = 1/2) and NiII ( S = 1) ions through the CN bridges. Overall, the magnetic properties of the samp le resemble that of ferromagnetic clusterspin-glasses with TC ranging from ~ 5.5 K (monolayer) to ~ 10.5 K (multibilayer). These TC values are much lower than that of th e bulk material (24 K) and successfully

PAGE 144

130 demonstrate the characteristic of the samples in reduced geometries. In both the DC and AC magnetic studies, the anisot ropic effect due to domain arrangement was observed in all samples. More specifical ly, the signal of the film was stronger when the film surface was placed parallel to the external magnetic field ( HE). These anisotropic effects suggest that all films possess a 2D nature in their magnetic domain arrangements. In addition, the degree of the anisotropic effect was most pronounced in the monolayer (2D) sample, and this fact is consistent with the evolving TC and TF as the sample evolves to the multilayer (3D) film. From the above experimental results, we concluded that the magnetic interaction in monolayer film is confined to the plane, while the bilayer sample possesses an additional magnetic pathway out of the plane and links two constituent monolayers magnetically. In the case of multibilayer, th e dipolar interaction between the bilayers enhanced the magnetic quantities of the multilayer film. 8.1.2 Future Directions As mentioned in Chapter 4, the exact st ructure between the monolayers in the bilayer is not known. However compared to the monolayer sample, the fact that the bilayer specimens showed less anisotropic effect and higher TC indicates the existence of an intra-bilayer magnetic interaction. The nature of this interaction could be superexchange if two monolayers had chemi cally bonded through CN bridges or could be dipolar if there were no magnetic bridging lig ands between the monolayers. For this unknown region, a careful structural study is needed. When this characterization is done, the magnetic characteristics of multibilayer can be analyzed rigorously by varying the length of amphiphilic tails that bonds two bilaye rs. In Chapter 4, it was assumed that the enhancement of the magnetic quantities in the multilayer compared to the bilayer was due to the dipolar interaction between two bilayers Additional studies, such as those that

PAGE 145

131 vary the strength (via length) and directi on of these interactions are desirable to investigate the detailed behavior of the dipolar interactions. Although the experimental results of the Ni-Fe(CN)6 materials successfully demonstrated the dimensional effect of the magnetic properties, the system itself might have been too complicated for systematic studi es of structural and dimensional effects of magnetism. For example, some of Ni-Fe(CN)6 systems contain H2O molecules that replace some of the Fe vacancies. These Fe vacancies seem to create geometrical frustration that prevents overa ll long range ordering and leaves the system in the glassy state. Of course, the magneto-structural st udy of glassy magnets is interesting, but a simpler physical situation should be considered first. The amphiphilic tails are necessary for the LB technique, and they can play a great role to isolate each bilayer. However, they add complexity to the system. Therefor e, a natural, next step towards the study of the dimensional magnetic effects would be to generate a series of thin solids linked by known pathways. Within the Prussian blue family, ANiII[CrIII(CN)6], where A is an alkali metal, contains no H2O and no vacancies. Therefore this idea l fcc structure has an equal ratio of NiII ( S = 1) to CrIII ( S = 3/2). The previously studie d ferromagnetic bulk material of CsNi[Cr(CN)6] 2H2O showed TC ~ 90 K [1], which is easily accessible. If this ANi[Cr(CN)6] is synthesized into films with well controlled thicknesses, the systematic study of the dimensional effect to the re lationship between the ferromagnetic magnetic correlation length and TC can be conducted. Theoretically, the susceptibilities and coherence lengths of quantum magnets are expected, near TC, to possess | T TC|and | T TC|dependencies, respectively. The exponents and v are predicted to differ for

PAGE 146

132 various classes of materials and also to be affected by the dimensionality of the system. Therefore by investigating the magnetic beha vior of films with va rious thicknesses, the systematic studies of the dimensional effect s on the critical behavior of these magnetic phase transitions can be accomplished. To produce a series of films with a system atic variation of the thicknesses, the combination of the LB technique and sequentia l deposition method can be adapted [9-14]. For example in this combined te chnique, a monolayer of ANi[Cr(CN)6] can be produced by using LB technique (one down stroke). Th is monolayer film then can be used as a templated substrate for the subsequent sequen tial deposition of thicker layered samples. Some of the useful designing aspects are di scussed in Chapter 7 as future directions. 8.2 Co-Fe(CN)6 Films and Powders 8.2.1 Summary The photoinduced magnetic material RbjCok[Fe(CN)6]l n H2O showed an interesting photoinduced anisotropic effect when the mate rial was synthesized in a film form, and this property had not been heretofore reported. The sequential deposition method enabled us to generate the thin films (~ 200 nm) of RbjCok[Fe(CN)6]l n H2O. The scanning electron microscopy (SEM) images s howed that the film is structurally continuous over ~ 600 nm. Before irradia tion, the magnetic beha vior of the film resembled that of the bulk material. More specifically, the film exhibited ferrimagnetic interactions between Co and Fe, indicated by rapid increases in magnetization around 15 K. Furthermore, the film also exhibited cluster-spin glass like behavior as evidenced from the frequency dependent susceptibility a nd history dependent ma gnetization. In our studies, the original sp in state that caused a finite magnetic state under dark conditions was referred to as the pr imordial spin state.

PAGE 147

133 The anisotropic effect was generated when the film was illuminated by visible light. The magnetization of the film at 5 K increa sed under the illumination when the film surface was placed parallel to the HE, but the magnetization decreased when the film surface was placed perp endicular to the HE. In both orientations, the mechanism of photoinduced magnetism originates from the photoi nduced charge transfer that results in diamagnetic Fe-Co pairs converting into magne tic Fe-Co pairs. The anisotropic effect, however, was explained in te rms of a dipolar field ( HD) description and specific experimental results, i.e. the va riation of the strength of the HE or the temperature of the system, support this description. One of th e main assumptions of the dipolar field description was the 2D-like ar rangement of the primordial spin states. A control experiment was performed with a similar RbjCok[Fe(CN)6]l n H2O film but with particle like (3D-like) texture, and the photoinduced anis otropic magnetic effect was not observed. This result is consistent with the pred iction of the dipolar field description. 8.2.2 Future Directions Although many groups have studied phot oinduced magnetic phenomena in the Co-Fe based Prussian blue analogs, the exact mechanism is still not known to this date. In other words, the exact conditions n ecessary for producing a photoinduced magnet seem to be dependent upon various pa rameters such as the amount of H2O and the choice of interstitial alkali metal. The difficulty of controlling these parameters is even greater in the film sample due to the chemical nature of the sequent ial deposition method. However, in some aspects, this difficulty might have created a random distribution of primordial spin regions that are sufficiently large to produce the dipolar field. In other words, the arrangement of the primordial spin states is not known exactly.

PAGE 148

134 To resolve this matter, direct magnetic imaging, either via a magnetic force microscope or through an indirect method su ch as neutron reflectometry might be a useful future probe of the surface magnetism of the film. Alternatively, since the distance between Co ion and Nitrogen atom is longer in primordial spin regions than that of photoinducible diamagnetic region, primordial spin regions will have larger lattice structure than that of diamagnetic regi ons. Therefore, non-magnetic surface imaging such as low temperature TEM can probe surf ace structure, which can be mapped onto the magnetic information of the surf ace. In addition, in order to investigate the dipolar field model, the construction of a simpler model system with know n geometry could prove to be useful. For example, strong magnetic nanom eter-sized particles can be distributed in the sea of the photoinduced magnetic material, in order to investigat e the effect of the arranged pattern on the photoindu ced anisotropic effect. The s ynthesis of this film can be guided by the discussion in Chapter 7. Currently, sequentially deposited fi lms of copper octacyanomolybdate (CuII 2[MoIV(CN)8] 8H2O) are under investigation for photoinduced magnetism. In contrast to many Co-Fe(CN)6 systems, bulk CuII 2[MoIV(CN)8] 8H2O is reported to be a photoinduced magnet,11 which does not have primordial spin states. In other words, in the absence of irradi ation, the sample behaves as a paramagnet and, therefore, is not expected to produce dipolar fields Consequently a film of CuII 2[MoIV(CN)8] 8H2O is predicted to exhibit no anisotropic photoi nduced magnetism due to the absence of the dipolar field. 11 Under irradiation of blue laser, th e material becomes ferromagnetic with TC ~ 25 K.

PAGE 149

135 A direct approach can be attempted to i nvestigate the effect of dipolar field by using a rotating, optical sample probe. Since th e direction of the dipolar field is strongly dependent upon the orientation of the film with respect to the HE, the angular dependence of the photoinduced magnetization measurement will provide insight into the behavior of the dipolar field. For this approach, a new MPMS probe equipped with both optical fiber and rotating sample stage (al ong horizontal axis) can be c onstructed. A packing of the films at an angle w ith respect to the HE is always an alternative option if using the existing optical probe. Alternatively, when the HE is perpendicular to film, monitoring photoinduced magnetization as a function of HE at fixed temperature can quantify the strength of dipolar field. As mentioned in Chapter 3, the exac t mechanism of photoinduced magnetism, especially its cooperative behavior, is not yet understood. This cooperative behavior also seems to be connected to the unusual char ge-transfer-induced spin transition (CTIST) phenomenon presented in Chapter 6. Obvi ously, the understanding of this cooperative behavior will contribute to the generati on of highly efficient photoinduced magnets and/or CTIST magnets. Both magnets are maybe useful for future magnetic devices because of their unique spin control charac teristics. However, the current photoinduced magnets are limited to low temperatures (< 77 K) and thus have not made practical applications yet. As an atte mpt to generate a future high TC photoinduced magnet, a heterostructured layered film was introduced in Chapter 7.

PAGE 150

136 APPENDIX A CAPACITANCE MEASUREMENT Simple capacitance measurement of (CH3)2NH2CuCl3, otherwise known as DMACuCl3 was performed and the details ar e introduced in this appendix. The capacitance measurement is a macroscopic probe targeted to detect a structural transition in a material. During the study of a possible alternating magnetic chain material (CH3)2NH2CuCl3, otherwise known as DMACuCl3 [81], a hint of a structural transition was detected. A more rigorous magnetic measurement of the powder of DMACuCl3 clearly showed the jump in da ta around 288 K, and the capacitance measurement was performed to clarify that th e origin of the jump in the magnetic data arises from the structural change indicat ed by a hysteric kinks in the change of capacitance. A low temperature capacitance measurement cell was built by Watson [87] and the Figure A-1 shows schematic repr esentation of the cell and th e experimental setup. The cell is composed of a copper top, a copper ba se, a base capacitance plate, and a spring loaded top capacitance plate. The powder sa mple was spread on top of the base plate, which is sitting on the copper base but is elec trically isolated by a phenolic spacer. The spring loaded top plate then pushes the sample against the bottom plate. A heater and thermometer were attached to the cell and pl aced in the cryostat. The top and bottom plates are electrically connected to two coax ial cables that connect to the capacitance

PAGE 151

137 bridge (GRC, Type 1615-A). The lock-in amplifier (EG&G Model 5302)12 was used to provide the excitation to the capacitance bridge as well as a null detection from the bridge. Figure A-1. The schematic view of capacitan ce measurement setup. A sample is placed between the capacitance plates and the two coaxial cables are connected from the capacitance bridge to the cell. The lock-in amplifier provides the excitation voltage to the capacitance bridge and detects the capacitance imbalance between the sample and the capacitance bridge. 12 http://www.signalrecovery.com LN2 Power source and voltmeter EG&G 5305 Lock-in Amplifier Vin A Vou t detector L H Generator GRC Capacitance Bridge Copper base Copper top Capacitance plates sample Phenolic insulator Probe Cryostat Thermometer Heater Spring

PAGE 152

138 Figure A-2. The change in capacitance and magnetic property of DMACuCl3. (a) The change in the capacitance was measured as a function of temperature upon cooling and warming. The hysteric ki nks at ~ 286.2 K upon cooling and at ~ 288.6 K upon warming indicate the pres ence of structural transition. (b) The susceptibility times temperature is s hown as a function of the temperature. The measurement was performed while warming with the external field of 100 G. The jump in the data around 288 K is consistent with the capacitance measurement. 180200220240260280300 -0.5 -0.4 -0.3 -0.2 -0.1 0.0 0.1 280285290295 -0.05 0.00 cooling warming Amplitude (mV)T ( K ) 200220240260280300 3.44 3.46 3.48 3.50 3.52 T (10-1 emu K / mol)T (K)(a) (b) HE = 100 G warming

PAGE 153

139 When the system is ready, 5 V (with fre quency ~ 1350 Hz) excitation voltage was given to the capacitance bridge, and the bridge wa s adjusted to give minimum detecting value on the lock-in amplifier. The computer then monitors the change in the minimum value, which corresponds to the changes of the samp le capacitance as temperature varies. The typical capacitance of the sample at room temperature was about 2.9 pF. Figure A-2 (a) shows the resultant capacitance change of the ~ 32 mg (recovered mass) of DMACuCl3. For this measurement, the time dependent lin ear drift of the signal was subtracted. The kinks in the detected voltage show hysteric behavior. Upon cooling, the kink appeared at 286.2 K, and it reappeared at 288.6 K upon warming. This change in capacitance is consistent with the results of the magnetic measurements shown in Figure A-2 (b) and indicates that the structural transition from monoclinic to triclinic accompanies the change in magnetic property. Although the capacitance m easurement of DMACuCl3 was performed in a liquid nitrogen environment, it can be easily reprodu ced in the liquid helium environment. This type of technique will be also useful to dete ct cooperative structural changes of thermally induced spin-crossover or photoinduced spin-crossover materials such as K0.6Co1.2[Fe(CN)6] 4H2O, a Prussian blue based photoinduced magnet. As a future project, the capacitance cell can be revised to incorporate th e optical source or ultrasound source.

PAGE 154

140 APPENDIX B LOW TEMPERATURE TRANSPORT MEASUREMENT Low temperature transport measurement was made to investigate Fermi-Liquid behavior of Sr1.5Ca0.5RuO4. The details of experimental setup are presented in this appendix. The low temperature four probe transport measurement was performed in the homemade dilution refrigerator system. The original dilution units (still, mixing chamber, heat exchangers) were purchased from Oxford Instruments13 (Kelvinox 25 dilution unit) but other components were made in the machin e shop in University of Florida. This appendix first introduces overall configura tion of dilution refrigerator system, and a detailed discussion of sample mounting follows A typical four wire magneto-resistance measurement process will then be reviewed along with an overview of the experimental procedure of cooling the diluti on system. Finally, this secti on closes with a summary and an outline of future possible improvements of the transport measurement system. B.1 Configuration of Dilution Refrigerator and Sample Mount As schematically shown in Figure B-1, th e dilution refrigerator system mainly consists of six parts: a probe, a 8 T superconducting magnet producing axial magnetic field, a gas handing panel, a liquid helium dewa r with liquid nitrogen jacket, a computer, and electronic equipments parts. More specif ically, the probe part consists of an inner vacuum can, 4.2 K flange, 1 K shield, 1 K pot, 1 K flange,14 adjustable 1 K pot impedance, still, still flange, still line, return line, mixing chamber, mixing chamber 13 URL: http://www.oxinst.com 14 Although referred to as K items, they actually operate closer to 1.7 K.

PAGE 155

141 flange, cold finger, sample stage, and thermo meters and heaters as shown in Figure B-1. The gas panel system consist of a hermetica lly sealed vacuum pump, nitrogen cold trap, two gas tanks for 3He-4He mixture15 (mash), copper tubes, pressure gauges, and gas valves. For a standard magneto-resistance measurement, a sample was first mounted on a sapphire disk (0.25 inch in diameter and ~ 0.003 inch thick, from Insaco16) using a tiny amount of GE varnish. Four thin gold wires17 (~ 0.005 inch in diameter) are then attached to the sample using silver epoxy (H20E from Epotek18) and typically cured at T ~ 430 K on a hot plate for about 5 minutes. The sample and wires on sapphire disk are tested for continuity and low contact resistances, and then the ensemble is mounted on the copper sample stage at the bottom of the cold finger. The cold finger is directly attached to the bottom of the mixing chamber flange where the thermometers and heater are also attached. For a four wire measurem ent, two current wires and two voltage wires are soldered to the gold wires from the sample as shown in Figure B-2 (a). In order to reduce the noise due to inductiv e coupling between th e wires, two wires of a current pair are twisted together for 4~5 tim es per inch as shown in Figure B-2 (a). The voltage wires are also twisted in the same manner. The wi res are then heat sunk to the mixing chamber flange, still flange, 1 K pot flange, and the 4.2 K flange by winding wires around copper cylinders (bobbin) that are bolted to the each flange. GE varnish and dental floss were used to provide bonding between the wires and the bobbin as shown in the Figure B-2 (b). 15 The typical total volume of operational mi xture was ~ 3.32 liters at 1 atm, where 3He / (3He + 4He) ratio was ~ 0.16. 16 URL: http://www.insaco.com 17 Two wires are for current supply and the other two are for voltage measurement. 18 URL: http://www.epotek.com

PAGE 156

142 Occasionally, heat shrink was used to wrap the bobbin and wires to provide an extra bonding. The sample wires connecting from mi xing chamber flange to the 1 K pot were made out of superconducting material in order to increase electric conductivity and reduce the thermal conductivity. From the 4.2 K flange, the wires travels to a room temperature wire connector through stainless steel cylindrical tubes. The LR-700 AC Resistance Bridge (Linear Research19) is then connected to measure th e real and imaginary components of resistance. The wires from thermometers and h eaters are also heat sunk and connected to the room temperature thermometer wire conne ctor box. For the control of thermometry, a Lakeshore Model 370 AC Resistance Bridge20 was used. The thermometers and heaters are placed at mixing chamber flange, still, and 1 K pot flange. The additional heater (film burner) was also placed at the t ube connecting to the t op of still to prevent superfluid 4He film moving out of the still. Each heater can be controlled using an individual power source. The magnet current leads from room temperature to the top of 4.2 K flange are made out of solder coated br ass tubes to reduce elec trical resistance and thermal conductivity. The magnet current leads and heat switch wires are then connected to the magnet power supplier and programmer. The magnet has a field to current ratio of 0.990 kG/A, a 4.0 inch bore, an 8.0 inch height, an inductance of 22 H and a homogeneity of 0.5% in a region within a 1 cm diameter spherical volume around the center point. The temperatures, resistance m easurements, and the magnet status are then read and controlled by computer using La bView program via GPIB (General Purpose Interface Bus). 19 URL: http://www.linearresearch.com The company no longer in service. 20 URL: http://www.lakeshore.com/index.html

PAGE 157

143 Figure B-1. The configuration of low temp erature transport measurement setup. The system mainly consists of six parts: a probe, a 8 T superconducting magnet, a gas handing panel, a liquid helium de war with liquid nitrogen jacket, a computer, and electronic equipments part s. More specifically, the probe part consists of an inner vacuum can, 4.2 K flange, 1 K shield, 1 K pot, 1 K flange, adjustable 1 K pot impedance, stil l, still flange, still line, return line, mixing chamber, mixing chamber flange, cold finger, sample stage, and thermometers and heaters. The gas pa nel system consist of hermetically sealed vacuum pump, nitrogen cold trap, two gas tanks for 3He-4He mixture (mash), copper tubes, pressure gauges, and gas valves. The electronics such as resistance bridges, thermometry, a nd magnet controllers are connected to the probe and the computer. pump 1 2 still line return line LN2 trap valve (closed) valve (open) mash tanks dewar IVC 1 K pot still mixing chamber superconducting magnet heat exchanger bath cold finger 1 K shield sample stage probe magnetcontrol thermometrycontrol resistance bridge pot pump computer (LabView) GPIB to probe adjustable impedance heat sink safety valve LN2 jacket to IVC pump P P P P P pressure gauge P P LHe flange (4.2K) flange (1 K) flange (still) flange (M.C.) return

PAGE 158

144 Figure B-2. Sample mount (a), wire heat sink (b ), and local heating e ffect of the sample (c). (a) Four gold wires are attached to the sample using silver epoxy and the twisted pairs of voltage and current tabs are connected to the gold wires for four probe transport measurement. (b) The wires are wrapped around the copper bobbin for thermal heat sink. Th e GE varnish and the dental floss were used to provide tight bonding betw een the wires and the bobbin. (c) The resistance of the sample as function of the excitation current was measured at the constant mixing chamber temperature (~ 100 mK). The decrease of the resistance with the lower excitation curre nt indicates the local heating of the sample. The inset shows the uncertainty of the measurement. As the excitation current is lowered the uncertainty increases. V+V-I+ ISample stage Sapphire disk Sample Silver epoxy Gold wires Twisted superconducting wires ~ 1 inch To voltage tabs To current tabs (a) Bobbin Wires Dental floss 0246810 1.44 1.46 1.48 1.50 1.52 0.00.51.01.5 1.444 1.446 1.448 1.450 R ()Iex (mA) (b) (c)

PAGE 159

145 When the LR-700 is not used, various lock-i n amplifiers were used to measure the sample resistance. A large resistor (~ 47 k ) was used to regulate the excitation current to a 0.1 mA to 10 mA in order to make a constant current measurement. A small current is required (in either LR-700 or lock-in am plifier measurement) to minimize the heating of the sample, but a small current also reduces the signal to noise ratios. Figure B-2 (c) shows the measured resistance of Sr1.5Ca0.5RuO4 as a function different excitation currents from the LR 700 resi stance bridge. The temper ature read from the mixing chamber thermometer is the same in all cases but clearly the local heating effects can be observed, as indicated by the decreasing resist ance of the sample. Mo re noise effect due to low excitation can be also observed in th e inset of Figure B-2 (c). This type of systematic approach of probing local heating wa s necessary in order to obtain an actual resistance of the sample without the heating effect by extrapolating the plot to the zero excitation. This aspect of transport meas urements at milikelvin temperatures is sometimes forgotten and not checked. B.2 Operation of Dilution Refrigerator Before cooling the system, all the electric connections and vacuum status of the 1 K pot, inner vacuum chamber (IVC), dilution units (still, still line, return line, mixing chamber), and gas panel have to be checked. For the electrical connections, the resistance of each wire was measured to check for an electric short or open. For a leak test of the IVC with respect to the 1 K pot and the bath of the liquid helium dewar, the probe was placed in the dewar and the IVC was pumped using a turbo pump. The bath and the 1 K pot were also pumped by a mechanical pump. The impedance of the 1 K pot was in the closed position in order to separate the source of leak, if one was detected. After achieving minimum pressure in the IVC (typically ~ 10 mTorr in the room

PAGE 160

146 temperature), the leak det ector (Spectron 600T, Edwards21) is connected to the back of the turbo pump. The bath is relatively difficult to pump to low pressure because of the large volume and the small copper tubes conne cted to it, but typically over night pumping provided ~ 50 mTorr. For a leak test between the bath and IVC, helium gas was introduced to the bath. The amount of helium gas is such that it does not exceed atmospheric pressure because if the pressure is higher than the atmosphere, th e helium gas might leak out of the bath and go into the possible leaking area on top of the probe such as electric wire connector box, which is connected to IVC. The typical background leak rate was ~ 1x10-9 mbar l / sec and if there is a leak between the bath and IVC, this rate goes up to ~ 10-7 mbar l / sec or higher depending on the source of the leak. To check the leak between the IVC and the 1 K pot, helium gas was introduced to the pumpe d pot and monitoring leak rate change in IVC completes the test. When there is no leak from the bath and pot to the IVC, helium gas was sprayed on top of the probe to check a possible leak from the electric connectors and vacuum components such as valves and t ubes. The leak check between dilution part and the IVC can be performed in the simila r way: pump dilution unit and IVC, and test the leak rate of the IVC while introducing heli um gas to dilution units. When all the leak checks are finished, the dilution units are pum ped as low as possible and the bath is pumped and flushed with helium gas for two or three times. The system is now ready to be cooled. 21 URL: http://www.bocedwards.com

PAGE 161

147 As a first step of cooling, a small amo unt of mash (typically ~ 150 mmHg in return pressure) begins to circulate the sy stem through the nitrogen cold trap.22 The circulation path is indicated by the arrows in Figure B1. As an exchange gas, about ~ 800 mTorr of helium gas is introduced to the IVC, and the bath is pressuri zed with helium gas slightly above the atmosphere to be ready for liqui d helium transfer. The 1 K pot was either pressurized with helium gas to ~ 1 psi ove r atmosphere or pumped continuously while transferring. For practical reasons, the liquid nitrogen pre-cooling of the bath was skipped and the direct transfer of liquid helium was made. The transferring rate of the initial cool down was kept slow (~ 5 liters / hour) to maximize cost effective gas cooling and not to quench the sample and the system When the liquid helium was transferred, the liquid nitrogen was also transferred into the nitrogen jacket of the dewar to maximize thermal efficiency and to prevent the air ge tting into the jacket due to cryo-pumping created by the cold surface of the nitrogen jacket.23 With cooling the system at about liquid ni trogen temperature, the electric and leak checks are repeated because of the physical a nd chemical changes of the materials at low temperature. The final electric and leak checks are also recommended above liquid helium temperature, and after this check, th e residual helium gas in IVC is pumped to around 8 mTorr using turbo pump. When the IVC pumping is complete, the valve is closed, and the 1 K pot should be started using a mechanical pump whose outlet is 22 The cold trap has to be cleaned previously by heating (~ 330 K) and pumping. The mash is also desired to be clean previously by circulating within the gas panel through the cold trap. 23 While transferring liquid helium the jacket te nds to be colder th an the liquid nitrogen temperature because the neck pa rt of the jacket is connected to the helium space. This colder jacket will reduce the boiling pressure of the liquid nitrogen and eventually acts like a pump.

PAGE 162

148 connected to the helium recovery system. Th e adjustable impedance of the 1 K pot also should be tuned to give ~ 1300 mmHg of 1 K po t pressure. Now the temperature of 1 K pot is ~ 1.6 K and the mixing chamber is ~ 800 mK or lower. The return pressure of the gas panel is also reduced to < 50 mmHg, and the condensing of the mash can begin by introducing the mash from the tank to the in let side of the pump. The condensing rate was kept < 150 mmHg as measured on the retu rn side so as not to overheat the dilution system and not to oversaturated liquid nitr ogen trap. When all the volume of mash (~ 3.32 liters at 1 atm) is condensed, the system is placed in closed circulation mode. In this mode, the typical pressure of the return line is < 50 mmHg, and th e still pressure is between 200 mTorr and 300mTorr. Within about 3 hours after the condensing is done, the system will reach around mi nimum temperature. For lower temperature, a heat (~ 0.6 mW) can be given to the sti ll to increase the pumping rate of 3He and thus provide lower temperature to the mixing chamber. Th e detailed principle of dilution cooling is beyond the scope of this dissertation. However, briefly, when the mash is condensed to the still, pumping of the stil l provides the cooling of the mash down to ~ 870 mK where the phase separation between 4He (dilute phase)24 and 3He (rich phase)25 occurs. This phase separation and continuous pumping from the still side eventually pumps the 3He from the rich phase to the dilute phase a nd provides an extra cooling to the lowest temperature. The minimum temperature of our dilution system with typical experimental set up was ~ 41 mK. 24 The dilute phase contains mostly 4He with temperature dependent small amount of 3He. The concentration of the 3He in dilute phase is consta nt (6.5%) even at the lowest temperature (~ 0 K). 25 The rich phase becomes pure 3He as temperature reaches the absolute zero.

PAGE 163

149 The measurement sequence of the magneto-r esistance experiment was performed at either a constant temperature or at a constant magnetic field. The temperature dependence of magneto-resistance was measur ed while sweeping the temperature at the various constant magnetic fields. The field dependent magneto-resistance was measured at the various constant temperat ures while sweeping the field. B.3 Summary and Future Directions In this appendix, the experimental met hods used for low temperature transport measurements were discussed. For the magne to-resistance measurement of the strongly correlated system Sr1.5Ca0.5RuO4, a four probe measurement technique was adopted in a dilution refrigerator environment. More spec ifically, the sample was glued to a bolted sapphire disk and mounted on the sample stag e at the bottom of a cold finger that is bolted to the mixing chamber flange. The positio n of the sample was at the center of an 8 T superconducting magnet, which provides homogeneity to with 0.5% in the 1 cm spherical diameter centered at the central point The four wires connected to the sample were heat sunk at the mixing chamber, the still, 1 K pot, and the 4.2 K flange before reaching the room temperature electronics. Before the operation of the dilution refrigerator system, all the electrical and vacuum components were checked at room temperature. While cooling, the mash was circulated throughout the system, and the nitrogen cold trap was used to filter im purities. The circulation of the mash, which condenses from the 1 K pot to the still, provid ed dilution cooling power to the sample and the minimum temperature of the system was ~ 41 mK. Although the present conf iguration of experimental se tup performs sufficiently for the magneto-resistance measurement, an improve ment can be made for future work. One of the suggested improvements can be made to the sample stage. Presently, the sample

PAGE 164

150 space is fixed at the bottom of the cold finge r and cannot be rotated. As in the case of Sr1.5Ca0.5RuO4, many samples have anisotropic propert ies with respect to the relative orientation between the current and the magne tic field direction. Rotating the sample each time before cool down is also possible, bu t it is too time consuming and furthermore, might change the experimental conditions of the measurement. The solution around this difficulty is to construct a rotating sample st age. When a rotating sample stage is used, the angular dependence of magneto-resistance can be resolved, and the result will contribute to a better understanding of the properties of the material.

PAGE 165

151 APPENDIX C ORIGIN SCRIPT FOR MAGNETIC QUANTITIES This ORIGIN script generates various magnetic quantities based on the magnetic data from MPMS (Magnetic Property Measurement System). // MvsT Generator // The raw data column should be named as TK, BG, rM, rM, drM, Back, dBack. // TK: temp, BG: Field, rM: Raw M, drM: Error of rM, Back: M of sample holder // Enter the Following information //Molecular Weight (g/mol) MW=378.18958 //Error in MW EMW=.0001 //Mass of the sample(g) Not mg MS=.02735 //Error in MS EMS=.0001 //300K error in sample (emu G) not per mol ES3=0 //300K error in the Back ground (emu G)* not per mol EC3=0 // Enter the core diamagnetic correction (emu / mol) CD=0 //Enter the error in corediam agnetic correction (emu/mol) ECD=0 // core magnetic moment wks.addCol(Mdia) wks.addCol(dMdia) // net magnetic moment per sample wks.addCol(M) wks.addCol(dM)

PAGE 166

152 // net magnetic moment per mol wks.addCol(Mmol) wks.addCol(dMmol) // Susceptibility per mol wks.addCol(Chi) wks.addCol(dChi) // Chi T per mole wks.addCol(ChiT) wks.addCol(dChiT) // u effective per mol in the unit of Bohr magneton wks.addCol(uef) wks.addCol(duef) // 1 over Chi time mol wks.addCol(InvChi) wks.addCol(dInvChi) Col(Mdia)=CD*Col(BG)*MS/MW Col(dMdia)=ECD*Col(BG)*MS/MW Col(M)=Col(rM)-Col(Back)-Col(Mdia) Col(dM)=sqrt((Col(drM))^2+(Col(dB ack))^2+(Col(dMdia))^2+EC3^2+ES3^2) Col(Mmol)=Col(M)*MW/MS Col(dMmol)=Col(Mmol)*sqrt((Col(dM) /Col(M))^2+(EMW/MW)^2+(EMS/MS)^2) Col(Chi)=Col(Mmol)/Col(BG) Col(dChi)=Col(dMmol)/Col(BG) Col(ChiT)=Col(Chi)*Col(TK) Col(dChiT)=Col(ChiT)*s qrt((Col(dChi)/Col(Chi ))^2+(0.005/Col(TK))^2) Col(uef)=sqrt(7.997*Col(ChiT)) Col(duef)=Col(uef)*sqrt((0.5*Col(d ChiT)/Col(ChiT))^2+(0.0005/7.997)^2) Col(InvChi)=1/Col(Chi) Col(dInvChi)=Col(InvChi)*Col(dChi)/Col(Chi) //end

PAGE 167

153 LIST OF REFERENCES 1. M. Verdaguer, A. Bleuzen, V. Marva ud, J. Vaissermann, M. Seuleiman, C. Desplanches, A. Scuiller, C. Train, R. Garde, G. Gelly, C. Lomenech, I. Rosenman, P. Veillet, C. Cartier, and F. V illain, Coord. Chem. Rev. 192 (1999) 1023. 2. H. J. Buser, D. Schwarzenbach, W. Pe ter, and A. Ludi, Inorg. Chem. 16 (1977) 2705. 3. D. Gatteschi, O. Kahn, J. S. Miller, F. Pa lacio, Eds.(Molecular Magnetic Materials) NATO AS1 Series, Series E, Vol. 198; Kluwer: Dordrecht, (1991). 4. V. Gadet, T. Mallah, I. Castro, M. Verdaguer, and P. Veillet, J. Am. Chem. Soc. 114 (1992) 9213. 5. T. Mallah, S. Thiebaut, M. Verdaguer, and P. Veillet, Science 262 (1993) 1554. 6. S. Ferlay, T. Mallah, R. Ouahes, P. Veillet, and M. Verdaguer, Nature 378 (1995) 701. 7. E. Dujardin, S. Ferlay, X. Phan, C. De splanches, C. Cartier dit Moulin, P. Sainctavit, F. Baudelet, E. Dartyge, P. Veillet, and M. Verdaguer, J. Am. Chem. Soc. 120 (1998) 11347. 8. J.-H. Park, J. T. Culp, D. R. Talham, and M. W. Meisel, Physica B 329-333 (2003) 1152. 9. J. T. Culp, J.-H. Park, D. Stratakis, M. W. Meisel, D. R. Talham, J. Am. Chem. Soc 124 (2002) 10083. 10. J. T. Culp, J.-H. Park, M. W. Meisel, and D. R. Talham, Inorg. Chem. 42 (2003) 2842. 11. J. T. Culp, J.-H. Park, I. O. Benitez, M. W. Meisel, and D. R. Talham, Polyhedron 22 (2003) 2125. 12. J. T. Culp, J.-H. Park, M. W. Meisel and D. R. Talham, Polyhedron 22 (2003) 3059. 13. J. T. Culp, Ph.D. Thesis, University of Florida (2002). 14. J. T. Culp, J.-H. Park, I. O. Benitez, Y. D. Huh, M. W. Meisel, and D. R. Talham, Chem. Mater. 15 (2003) 3431.

PAGE 168

154 15. S. Juszczyk, C. Johansson, M. Hanson, A. Ratuszna, and G. Malecki, J. Phys. Condens. Matter 6 (1994) 5697. 16. Y. Zhang, C. Luoa and Z. Yu, New J. Chem. 29 (2005) 1285. 17. S. Ohkoshi, T. Iyoda, A. Fujishima, and K. Hashimoto, Phys. Rev. B 56 (1997) 11642. 18. A. Bobk, F. O. Abubrig, and T. Balcerzak, Phys. Rev. B 68 (2003) 224405. 19. S. Ohkoshi, Y. Abe, A. Fujishima, and K. Hashimoto, Phys. Rev. Lett. 82 (1999) 1285. 20. S. Ohkoshi and K. Hashimoto, Phys. Rev. B 60 (1999) 12820. 21. S. Ohkoshi, O. Sato, T. Iyoda, A. Fujishima, and Kazuhito Hashimoto, Inorg. Chem. 36 (1997) 268. 22. S. Ohkoshi, and K. Hashimoto, J. Am. Chem. Soc. 121 (1999) 10591. 23. S. Ohkoshi, Y. Einaga, A. Fujishima, a nd K. Hashimoto, J. Electroanal. Chem. 473 (1999) 245. 24. M. Ohba, H. Ohkawa, N. Fukita, and Y. Hashimoto, J. Am. Chem. Soc. 119 (1997) 1011. 25. N. M. Harrison, B.G. Searle, E. A. Seddon, Chem. Phys. Lett. 266 (1997) 507. 26. G. R. Torres and B. Agricole, Chem. Mater. 14 (2002) 4012. 27. W. E. Buschmann and J. S. Miller, Inorg. Chem. 39 (2000) 2411. 28. R. C. Millward, C. E. Madden, I. Sutherla nd, R. J. Mortimer, S. Fletcher, and F. Marken, Chem. Commun. (2001) 1994. 29. A. Kumar, and S. M. Yusuf, Physica B 362 (2005) 278. 30. M. Ohba and H. Okawa, Coord. Chem. Rev. 198 (2000) 313. 31. S. Ohkoshi and K. Hashimoto, J. Am. Chem. Soc. 121 (1999) 10591. 32. M. Verdaguer, N. Galvez, R. Garde, and C. Desplanches, Interface (Fall 2002) 28. 33. E. Ruiz, A. Rodrguez-Fortea, S. Alvar ez, and M. Verdaguer, Chem. Eur. J. 11 (2005) 2135. 34. R. Garde, F. Villain, and M. Verdaguer, J. Am. Chem. Soc. 124 (2002) 10531. 35. O. Sato, T. Iyoda, A. Fujishima, and K. Hashimoto, Science 271 (1996) 49.

PAGE 169

155 36. M. Ohba, N. Fukita, and H. Ok awa, Dalton Trans. (1997) 1733. 37. S. Takeda, Y. Umehara, H. Hara, and G. Maruta, Mol. Cryst. Liq. Cryst. 379 (2002) 253. 38. O. Sato, T. Iyoda, A. Fujishima, and K. Hashimoto, Science 272 (1996) 704. 39. M. Verdaguer, Science 272 (1996) 698. 40. O. Sato, Y. Einaga, A. Fujishima, and K. Hashimoto, Inorg. Chem. 38 (1999) 4405. 41. A. Bleuzen, C. Lomenech, V. Escax, F. Villa in, F. Varret, C. Cartier dit Moulin, K. Yoshizawa, F. Mohri, G. Nuspl, T. Yamabe, J. Phys. Chem. B 102 (1998) 5432. 42. A. Bleuzen, C. Lomenech, V. Escax, F. V illain, F. Varret, C. Cartier dit Moulin, and M. Verdaguer, J. Am. Chem. Soc. 122 (2000) 6648. 43. C. Cartier dit Moulin, F. Villain, A. Bleuzen, M.-A. Arrio, P. Sainctavit, C. Lomenech, V. Escax, F. Baudelet, E. Dartyge J. J. Gallet, and M. Verdaguer, J. Am. Chem. Soc. 122 (2000) 6653. 44. V. Escax, A. Bleuzen, C. Cartier dit Mou lin, F. Villain, A. Goujon, F. Varret, and M. Verdaguer, J. Am. Chem. Soc. 123 (2001) 12536. 45. G. Champion, V. Escax, C. Cartier dit Moulin, A. Bleuzen, F. Villain, F. Baudelet, E. Dartyge, and M. Verdaguer, J. Am. Chem. Soc. 123 (2001) 12544. 46. A. Goujon, F. Varret, V. Escax, A. Bl euzen, and M. Verdaguer, Polyhedron 20 (2001) 1339. 47. A. Goujon, F. Varret, V. Escax, A. Bl euzen, and M. Verdaguer, Polyhedron 20 (2001) 1347. 48. A. Goujon, O. Roubeau, F. Varret, A. Dolbecq, A. Bleuzen, and M. Verdaguer, Eur. Phys. J. B 14 (2000) 115. 49. J.-H. Park, Y. D. Huh, E. imr, S. J. Gamble, D. R. Talham, and M. W. Meisel, J. Magn. Magn. Mater. 272-276 (2004) 1116. 50. J.-H. Park, E. imr, M. W. Meisel, Y. D. Hu h, F. Frye, S. Lane, and D. R. Talham, Appl. Phys. Lett. 85 (2004) 3797. 51. J.-H. Park, F. Frye, S. Lane, E. imr, Y. D. Huh, D. R. Talham, and M. W. Meisel, Polyhedron 24 (2005) 2355. 52. O. Sato, S. Hayami, Y. Einaga, and Z. Gu, Bull. Chem. Soc. Jpn., 76 (2003) 443. 53. D. A. Pejakovi J. L. Manson, J. S. Miller, and A. J. Epstein, Phys. Rev. Lett. 85 (2000) 1994.

PAGE 170

156 54. K. Yoshizawa, F. Mohri, G. Nuspl and T. Yamabe, J. Phys. Chem. B 102 (1998) 5432. 55. T. Kawamoto, Y. Asai, and S. Abe, Phys. Rev. Lett. 86 (2001) 348. 56. M. Hanawa, Y. Moritomo, A. Kuriki, J. Tateishi, K. Kato, M. Takata, and M. Sakata, J. Phys. Soc. Jpn. 72 (2003) 987. 57. N. Shimamoto, S. Ohkoshi, O Sato, and K. Hashimoto, Inorg. Chem. 41 (2002) 678. 58. S. Gawali-Salunke, F. Varret, I. Maur in, C. Enachescu, M. Malarova, K. Boukheddaden, E. Codjovi, H. Tokoro, S. Ohkoshi, and K. Hashimoto, J. Phys. Chem. B 109 (2005) 8251. 59. C. W. Ng, J. Ding, and L. M. Ga n, J. Solid State Chem. 156 (2001) 400. 60. S. Ohkoshi and K. Hashimoto, Photochem. Photobiol. C: Photochemistry Reviews 2 (2001) 71. 61. O. Sato, Y. Einaga, T. Iyoda, A. Fujishi ma, and K. Hashimoto, J. Phys. Chem. B 101 (1997) 3903. 62. S. Ohkoshi, H. Tokoro, M. Utsunomiya, M. Mizuno, M. Abe, and K. Hashimoto, J. Phys. Chem. B106 (2002) 2423. 63. Y. Moritomo, K. Kato, A. Kuriki, M. Ta kata, M. Sakata, H. Tokoro, S. Ohkoshi, and K. Hashimoto, J. Phys. Soc. Jpn. 71 (2002) 2078. 64. H. Tokoro, S. Ohkoshi, and K. Hashimoto, Appl. Phys. Lett. 82 (2004) 1245. 65. Y. Sato, S. Ohkoshi, and K. Hashimoto, J. Appl. Phys. 92 (2002) 4834. 66. O. Sato, V. Einaga, T. Iyoda, A. Fujishima, and K. Hashimoto, J. Electrochem. Soc. 144, (1997) L11. 67. Y. Moritomo, M. Hanawa, Y. Ohishi, K. Ka to, M. Takata, A. Kuriki, E. Nishibori and M. Sakata, Phys. Rev. B 68 (2003) 144106. 68. T. Yokoyama, M. Kiguchi, T. Ohta, O. Sa to, Y. Einaga, and K. Hashimoto, Phys. Rev. B 60 (1999) 9340. 69. S. Ohkoshi, T. Hozumi, and K. Hashimoto, Phys. Rev. B 64 (2001) 132404 70. H. Osawa, T. Iwazumi, H. Tokoro, S. Ohkos hi, K. Hashimoto, H. Shojid, E. Hiraid, T. akamurad, S. Nanaod, and Y. Isoz umi, Solid State Commun. 125 (2003) 237. 71. M. Kamiya, M. Hanawa, Y. Moritomo, Y. Isobe, J. Tateishi, K. Kato, and A. Nakamura1, Phys. Rev. B 69 (2004) 052102.

PAGE 171

157 72. H. W. Liu, K. Matsuda, Z. Z. Gu, K. Ta kahashi, A. L. Cui, R. Nakajima, A. Fujishima, and O. Sato, Phys. Rev. Lett. 90 (2003) 167403 73. M. Pyrasch, B. and Tieke, Langmuir 117 (2001) 7706. 74. T. Yamamoto, Y. Umemura, O. Sato, and Y. Einaga, Chem. Lett. 33 (2004) 500. 75. T. Yamamoto, Y. Umemura, O. Sato, and Y. Einaga, Chem. Mater. 16 (2004) 1195. 76. T. Yamamoto, Y. Umemura, O. Sato, and Y. Einaga, J. Am. Chem. Soc. 127 (2005) 16065. 77. A. Bleuzen, V. Escax, A. Ferrier, F. Villa in, M. Verdaguer, P. Munsch, and J.-P. Itie, Angew. Chem. Int. Ed., 43 (2004) 3728. 78. K. Boukheddaden, M. Nishino, S. Miyashit a, and F. Varret, Phys. Rev. B 72 (2005) 014467. 79. S. Ohkoshi, H. Tokoro, and K. Hashimoto, Coord. Chem. Rev. 249 (2005) 1830. 80. O. Kahn and C. J. Martinez, Science 279 (1998) 44. 81. J. W. Hall, W. E. Marsh, R. R. Weller, W. E. Hatfield, Inorg. Chem. 20 (1981) 1033. 82. R. D. Willett, J. Chem. Phys. 44 (1966) 39. 83. S. O'Brien, R. M. Gaura, C. P. Landee, B. L. Ramakhrishna, R. D. Willett, Inorg. Chim. Acta 141 (1988) 83. 84. S. Lahiry, and R. Kakkar, Chem. Phys. Lett. 78 (1981) 379. 85. M. Hurley and B. C. Gerstein, J. Chem. Phys. 59 (1973) 6667. 86. B. C. Gerstein, F. D. Gehring, and R. D. Willett, J. Appl. Phys. 43 (1972) 1932. 87. B. C. Watson, Ph.D. Thesis Univer sity of Florida: Gainesville, 2000. 88. Y. Ajiro, K. Takeo, Y. Inagaki, T. Asano, A. Shimogai, M. Mito, T. Kawae, K. Takeda, T. Sakon, H. Nojiri, M. Motokawa, Physica B 329 (2003) 1008. 89. Y. Inagaki, A. Kobayashi, T. Asano, T. Sakon, H. Kitagawa, M. Motokawa, Y. Ajiro, J. Phys. Soc. Jpn. 74 (2005) 2683. 90. Y. Yoshida, O. Wada, Y. Inagaki, T. Asano, K. Takeo, T. Kawae, K. Takeda, Y. Ajiro, J. Phys. Soc. Jpn. 74 (2005) 2917.

PAGE 172

158 91. S. Nakatsuji, D. Hall, L. Balicas, Z. Fisk, K. Sugahara, M. Yoshioka, and Y. Maeno, Phys. Rev. Lett. 90 (2003) 137202. 92. S. Nakatsuji and Y. Maeno, Phys. Rev. Lett. 84 (2000) 2666. 93. S. Nakatsuji and Y. Maeno, Phys. Rev. B 62, (2000) 6458. 94. MPMS Users Manual, Quantum Design, San Diego, CA. 95. MPMS Hardware Manual, Quantum Design, San Diego, CA. 96. MPMS Application Note 1014-201, Quantum Design, San Diego, CA. 97. A. Schuhl, S. Maegawa, M. W. Meisel, a nd M. Chapellier, Phys. Rev. B 36 (1987). 98. M. Jackson, P. Slheid, and J. Marvin, The IRM Quarterly Vol. 11 No. 4 (2001) pg. 1 and 9-11. 99. MPMS Application Note 1014-213, Quantum Design, San Diego, CA. 100. G. E. Fanucci, J. Krzystek, M. W. Meisel L.-C. Brunel, and D. R. Talham, J. Am. Chem. Soc. 120 (1998) 5469. 101. O. Kahn, Molecular Magnetism (VCH, New York, 1993). 102. J. A. Mydosh, Spin Glasses (Taylo r & Francis, Washington, DC, 1993). 103. R. Mathieu, P. Jonsson, D. N. H. Nam, and P. Nordblad, Phys. Rev. B 63 (2001) 092401. 104. E. Spiecker, A. Schmid, A. Minor, U. Dahmen, S. Hollensteiner, and W. Jger, Phys. Rev. Lett. 96 (2006) 086401. 105. H. Bethe, Z. Phys. 71 (1931) 205. 106. M. Yamashita T. Ishii, and H. Ma tsuzaka, Coor. Chem. Rev. 198 (2000) 347. 107. X. Obradors, A. Labarta, A. Isalgue, J. Te jada, J. Rodriguez, and M. Pernet, Solid State Commun. 65 (1988) 189. 108. M. Jaime, V. F. Correa, N. Harrison, C. D. Batista, N. Kawashima, Y. Kazuma, G. A. Jorge, R. Stern, I. Heinmaa, S. A. Zvyagin, Y. Sasago, and K. Uchinokura, Phys. Rev. Lett. 93 (2004) 087203. 109. D. Gatteschi and R. Sessoli, Angew. Chem. 42 (2003) 268.

PAGE 173

159 BIOGRAPHICAL SKETCH Ju-Hyun Park was born in Pusan, the second la rgest city in South Korea. At age six in a toy store, he happened to encounter a model airplane. Ever since, obsessed by the idea of flying, he spent most of his chil dhood designing and building model airplanes and radios. Soon he learned that the area of his fondness was called physics, and as he entered Dong-Rae High School in Pusan, he set his life goal to be a physicist. After he served in the Korean military for more than two years, he came to the United States to continue his education. He was enrolled in the department of physics at Southern Connecticut State University in 1996. Du ring undergraduate studies, his interest in physics was broad, ranging from condensed ma tter to astrophysics, both experimental and theoretical. However, he decided to specialize in condensed matter experimental physics, especially in magnetism when he jo ined Professor Mark W. Meisels group at the University of Florida. While working in the Meisel group, he al so married his lovely wife, Asako Osanai, and shares hi s life with her as a team.


Permanent Link: http://ufdc.ufl.edu/UFE0013792/00001

Material Information

Title: Magneto-Structural and Magneto-Optical Studies of Prussian Blue Analogs
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0013792:00001

Permanent Link: http://ufdc.ufl.edu/UFE0013792/00001

Material Information

Title: Magneto-Structural and Magneto-Optical Studies of Prussian Blue Analogs
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0013792:00001


This item has the following downloads:


Full Text







MAGNETO-STRUCTURAL AND MAGNETO-OPTICAL
STUDIES OF PRUSSIAN BLUE ANALOGS















By

JU-HYUN PARK


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA


2006






























Copyright 2006

by

Ju-Hyun Park
































For Asako








ACKNOWLEDGMENTS

In the course of experimental studies, I benefited greatly from the numerous

discussions and assistance from fellow colleagues. First and foremost, I would like to

thank Professor Daniel Talham, Professor Young-Duk Huh, Dr. Jeff Culp, and Franz

Frye for designing and fabricating the materials studied in this dissertation. They were

enthusiastic chemists, from whom I had the privilege of learning magneto-chemistry and

supramolecular structures. I also would like to thank Professor Yoonseok Lee for his

great advices in experimental techniques and physics. He guided me through the many

projects, especially the entire process of dilution refrigerator operation, from which I

learned many experimental techniques and the view of physicist.

Every member of the Department of Physics instrument shop has been extremely

helpful. Especially, I would like to thank Marc Link, Ed Storch, and Bill Malphurs for

their great craftsmanship. Their expertise and casual discussions always helped

throughout the designing process. I also would like to thank members in the Cryogenic

Services, Greg Labbe and John Graham for their great support in cryogenic matters. As

members in the same lab, I would like thank to James Maloney, Sara Gamble, James

Davis, and Norman Anderson for their support and useful discussions in physics. With

them, I truly enjoyed working in the lab and experienced great American customs.

Finally, I would like to thank my Ph.D. adviser, Professor Mark Meisel for his guidance

and encouragement. From him I learned not only the physics but also the example of

being a professional scientist. Furthermore, in all aspects, he always supported and




inspired me when I was dealing with everyday life in a foreign country, and for that I

truly thank him.








TABLE OF CONTENTS

page

ACKNO W LEDGM EN TS ............ .. ...... ....................................................... iv

LIST O F TA BLES ....................................................................... .............................. ix

LIST O F FIG U R E S ....................................................................................................... x

A B S T R A C T ..................................................................................................................... x iii

CHAPTER

1. IN T R O D U C T IO N ...................................................................................................... 1

1.1 N i-Fe(C N )6 Film s......................................................................................... 2
1.2 C o-Fe(C N )6 Film s............................................................................................4
1.3 Co-Fe(CN )6 Pow ders ................................................................................. 6
1.4 O their M agnetic System s .......................................................... ..........................7

2. EXPERIMENTAL TECHNIQUES........................................................................... 8

2.1 Configuration of MPMS Hardware ...............................................................9
2 .1.1 S am ple R od ...................................................... ....................................9
2.1.2 Probe ............................................................................................. 11
2.1.3 Console and Com puter............................... ............................................... 14
2.1.4 Principle of DC Magnetization Measurement ......................................14
2.2 DC Magnetization Measurement Procedure..................................................16
2.3 DC Magnetization Data Analysis ..................................................................19
2.3.1 Curie-W eiss Law and Dimer M odel..................................................... 19
2.3.2 History Dependent Magnets .......................... ..............................25
2.5 Sample Packing and Background Consideration......................................... 30
2.5.1 Sam ple Packing................................................................................32
2.5.2 Background Consideration............................................................ 35
2.6 Summary and Future Direction...................................................................... 37

3. PHYSICAL PHENOMENON AND THEORY.......................................................39

3.1 Photoinduced Magnetism in Prussian Blue Analogs .......................................39
3.1.1 O verview ........................................................................ ........................39
3.1.2 Initial Observation and Description.....................................................39





3.2 Charge Transfer Induced Spin Transition.....................................................49
3.3 Sum m ary .................................................. ................. .. ...............................57

4. MAGNETIC STUDY OF EVOLVING STRUCTURE (MONO, BI, AND
MULTILAYER OF FILMS)................................................................................58

4.1 Synthesis of Ni-Fe(CN)6 Film s........................................................................58
4.2 DC Low Field Magnetization Measurements................................................60
4.3 AC Field Magnetization Measurements .....................................................72
4.4 Magnetic Evolution versus Structural Evolution...........................................81

5. ANISOTROPIC PHOTOINDUCED MAGNETISM OF PRUSSIAN BLUE
A N A L O G FIL M S........................................................ ........................................ 87

5.1 Synthesis of Co-Fe(CN)6 Films ......................................................................88
5.2 Photoinduced State M agnetism ........................................ ............................90
5.3 D ipolar Field M odel.................................... ................. ............................98

6. CHARGE TRANSFER INDUCED SPIN TRANSITION IN PRUSSIAN BLUE
A N A L O G ..................................... ............ .......................................... ................... 104

6.1 Experim ental Details.................................. ............. ............................105
6.2 Experimental Results ........................... ..............................................106
6.3 Discussion and Future Directions ...............................................................108

7. HETEROSTRURED LAYERED MAGNET...................................................... 113

7.1 Heterostructured Layered Film s ................................................................114
7.2 Conceptual 2D Magnetic Nanometer Island..................................................121
7.3 Sum m ary and Future Directions.................................................................. 127

8. SUMMARY AND FUTURE DIRECTIONS........................................................129

8.1 N i-F e(C N )6 Film s....................................................... ...............................129
8.1.1 Sum m ary ................................................................ ......................... 129
8.1.2 Future D directions ................................... ..........................................130
8.2 Co-Fe(CN)6 Films and Powders ............................................................132
8.2.1 Sum m ary .................................................... ...................................... 132
8.2.2 Future D directions ................................... ..........................................133

APPENDICES

A. CAPACITANCE MEASUREMENT..................................................................136

B. LOW TEMPERATURE TRANSPORT MEASUREMENT ................................... 140

B. 1 Configuration of Dilution Refrigerator and Sample Mount..........................140
B.2 Operation of Dilution Refrigerator.............................................................. 145




B.3 Summary and Future Directions....................................................................149

C. ORIGIN SCRIPT FOR MAGNETIC QUANTITIES .........................................151

LIST O F R EFER EN C ES ............................................................................................... 153

BIOGRAPHICAL SKETCH ........................................................................................ 159
















































viii








LIST OF TABLES


Table page

2-1. Zero-field-cooled magnetization measurement procedure................................... 18

2-2. Field-cooled magnetization measurement procedure............................................18

2-3. Magnetization vs. magnetic field measurement procedure.................................. 18

3-1. Chemical composition of NaCo,[Fe(CN)6]'zH20 samples .....................................50

3-2. Valence states of NaxCoy[Fe(CN)6]-zH20 samples at 290 K ..................................50

4-1. Characteristic values from DC magnetization measurements................................. 82

4-2. Characteristic values from AC magnetization measurements................................. 82








LIST OF FIGURES


Figure pge

2-1. Schematics of MPMS (Magnetic Property Measurement System)....................... 10

2-2. Schematics of the impurity filter and liquid helium transfer .................................. 13

2-3. Second-derivative pickup coil and centering magnetic signal................................. 15

2-4. M agnetization signals of the holders.....................................................................17

2-5. M agnetic plots of various types of magnet ..............................................................20

2-6. Susceptibility times temperature simulations of dimers ..........................................23

2-7. Field dependent magnetization simulations of dimer model ...................................24

2-8. Field-cooled (fc) and zero-field-cooled (zfc) magnetization processes .................26

2-9. Thermal remnant magnetization of RbxNiy[Cr(CN)6]z-mH20 film..........................29

2-10. A home ade M PM S optic insert rod....................................................................31

2-11. Various MPMS sample packing methods...........................................................33

2-12. Simulation of weak paramagnetic sample in gelcap holder....................................36

3-1. Temperature dependent photoinduced and dark magnetizations ...........................41

3-2. Time dependent photoinduced magnetization.................................................. 42

3-3. Processes of photoinduced magnetization and demagnetization ...........................44

3-4. Complete cycle of photoinduced magnetization and demagnetization ..................47

3-5. Magnetizations of a series of NaxCoy[Fe(CN)6]-zH20 compounds............................53

3-6. xTvs. T data of Ko.6CO1.2[Fe(CN)6]-4H20 powder upon cooling and warming ......55

3-7. M vs. H data of Ko.6Co1.2[Fe(CN)6]-4H20 powder .................................................56

4-1. Amphiphilic pentacyanoferrate building unit and 2D two-dimensional grid ..........59





4-2. Sketches of the monolayer, bilayer, and multibilayer......................................61

4-3. Schematic of Fe-CN-Ni square grid (top view) .......................................................62

4-4. Temperature dependent magnetizations of 150-multibilayer film...........................64

4-5. The field dependent magnetizations of 150-multibilayer film at T = 2 K................65

4-6. Temperature dependent magnetizations of bilayer film.........................................67

4-7. The field dependent magnetizations ofbilayer film at T = 2 K .............................68

4-8. Temperature dependent magnetizations of monolayer film ...................................70

4-9. The field dependent magnetizations ofmonolayer film at T = 2 K .......................71

4-10. Temperature dependent AC susceptibilities of 150-multibilayer film.....................73

4-11. Result of Arrhenius law fittings to monolayer, bilayer, and multibilayer...............74

4-12. Result of Vogel-Fulcher law fitting on multibilayer film ....................................76

4-13. Temperature dependent AC susceptibilities of monolayer film.............................77

4-14. Temperature dependent AC susceptibilities of bilayer film.....................................79

4-15. Comparison of AC data at 17 Hz (mono, bilayer, and multibilayer).....................80

4-16. Dipole field produced by a magnetized disk (R-30 A)..........................................84

5-1. SEM images of film 1 (a) and film 2 (b)................................................................89

5-2. Photoinduced magnetization of film 1 ........................................................91

5-3. Photoinduced magnetization of film 2 ................................................................... 93

5-4. Photoinduced magnetization of film 2 .............................. .......................... 94

5-5. Frequency dependent AC susceptibilities of Co-Fe(CN)6 films ..............................96

5-6. Vogel-Fulcher law fitting to Co-Fe(CN)6 films ...............................................97

5-7. Schematic description of the spin configurations of the domains in the film..........99

5-8. Photoinduced magnetization of film 2 when HE > HD.......................................100

5-9. Field dependent photoinduced and dark state magnetization of film 2................101

5-10. Photoinduced magnetization at T- 4.5 K and HE = 20 T.................................... 102




6-1. Susceptibility time temperature as function of temperature.................................107

6-2. Low temperature magnetizations measured at three different cooling states ........109

6-3. Magnetizations of three different magnetic states................................................110

7-1. An alternating planar structure of Prussian blue analog ........................................117

7-2. Schematics of synthesizing Ni-Cr and Ni-Cr / Co-Fe films................................19

7-3. Temperature dependent magnetizations of Ni-Cr and Ni-Cr/ Co-Fe films...........20

7-4. Schem atic of LB film generation ..........................................................................123

7-6. Deposition of nanometer LB islands on substrate................................................125

7-7. Deposition of nanometer particles to substrates. ...........................................126

A-1. The schematic view of capacitance measurement setup ........................................137

A-2. The change in capacitance and magnetic property of DMACuC3 ........................138

B-1. The configuration of low temperature transport measurement setup................... 143

B-2. Sample mount for transport measurement ...........................................................144








Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

MAGNETO-STRUCTURAL AND MAGNETO-OPTICAL STUDIES OF
PRUSSIAN BLUE ANALOGS

By

Ju-Hyun Park

May 2006

Chair: Mark W. Meisel
Major Department: Physics

The magnetic properties of molecule-based magnets, especially in confined or

altered geometries, are of great interest in modem magnetic devices. Two different but

related molecule-based magnets, namely Ni-Fe(CN)6 and Co-Fe(CN)6 Prussian blue

analogs, were generated. Langmuir-Blodgett and/or sequential deposition fabrication of

thin films provided opportunities to study dimensional effects on fundamental

magnetism. More specifically, a series ofmonolayer, bilayer, and multilayer of

ferromagnetic Ni-Fe(CN)6 films were generated, and the magnetic evolution

accompanied by the structural evolution is reported herein. In addition, when

Co-Fe(CN)6 was generated as a thin film, unusual photoinduced magnetism was

observed. In our investigation, the photoinduced magnetization of the film increased or

decreased due to the orientation of the film with respect to the external magnetic field.

This interesting effect provides a new magnetic switching mechanism and is reported

herein. For some of the Co-Fe(CN)6 Prussian blue analogs, the phenomenon of charge




transfer induced spin transition have been reported. Our study of the unusual spin

transition process that depends on the cooling rate is also presented herein. Finally, in the

pursuit of better molecule-based functional magnets, the fabrication method and magnetic

results of heterostructured layered films of Ni-Cr / Co-Fe (CN)6 are presented.







CHAPTER 1
INTRODUCTION

The main topic of this dissertation focuses on the characterization and physical

aspects of the molecular magnetic systems based on Prussian blue analogs. Historically,

Prussian blue is known as the first artificial pigment originating from the use in German

army uniforms. It was accidentally found in 1704, when beef blood was boiled in a

strong basic solution. Despite of its long history, the chemical structure of Prussian blue

(Fe"I4[Fe"(CN)6]3*14H20) was not identified until the early 1970s [1]. In general, the

Prussian blue and its analogs have face-centered-cubic (fcc) structures, and the generic

formula can be written as AjMk[M'(CN)6]rnH20,1 where M and M' are bridged by CN,

and respectively located in the center and at the vertices of the octahedral structure. The

interstitial alkali metal A occupies some of the tetrahedral sites. Hereafter, the generic

form of the Prussian blue analog will be referred as M- M'(CN)6.

Recently, due to its superb magnetic characteristics, the family of Prussian blue

compounds has received attention in molecule-based magnets [1 79]. For example, the

compound V[Cr (CN)6]0.86"2.8H20 was the one of the first molecular magnets whose Tc

exceeded room temperature [6]. Other compound, Ko.2Co1.4[Fe(CN)6]'6.9H20, exhibited

a photoinduced magnetism, in which the increases or decreases of the magnetization were

controlled by light [38]. These interesting results came, however, mainly from

experiments performed on bulk materials and, thus, may not reflect the properties of the

1This formula contains a interstitial alkali metal A, transition metals (M and M') of
different valence states, and the waters that replace [M'(CN)6] to make overall charge
neutral.







material confined to reduced dimensions. In addition, the technological demand on

modem magnetic devices requires physically smaller systems, so the constituent

materials must be studied in restricted physical dimensions. Consequently, studying the

dimensional and structural effects of the magnetic systems seems a natural interest, and

this interest motivated the research presented in this dissertation.

Within the Prussian blue family, Ni-Fe(CN)6 and Co-Fe(CN)6 systems were

selected as research compounds. Briefly, the bulk Ni-Fe(CN)6 is a ferromagnetic system

with the Tc 24 K [15], and the bulk Co-Fe(CN)6 is a photoinduced magnet as

mentioned earlier [38]. One of the main challenges was to reduce the dimensions of the

compounds, and this task requires techniques to manipulate the structures at the

nanometer scale. For this challenge, the Langmuir-Blodgett technique [9 13] and the

sequential deposition method [13, 14] were utilized for Ni-Fe(CN)6 and Co-Ni(CN)6

respectively. Both the Langmuir-Blodgett and sequential deposition methods are

chemical approaches of building thin solid films from layer-by-layer deposition. General

reviews and specific uses of these methods have been presented elsewhere [9 14], and

the possibility of using these methods to fabricate heterostructured layered films, as well

as magnetic nanometer islands, are discussed in Chapter 7.

1.1 Ni-Fe(CN)6 Films

Using Langmuir-Blodgett techniques, Ni-Fe(CN)6 films were fabricated in

monolayer, bilayer, and multibilayers motifs. This series of systems was designed to

model two-dimensional (2D), quasi-2D, and three-dimensional (3D) magnetic systems [9,

11]. The resultant monolayer film contains a 2D network of face centered squares of

Ni-Fe(CN)6 on a substrate. In addition, at the interface to the air, Fe ions are terminated

by amphiphilic tails. The bilayer film is composed of two monolayers of Ni-Fe(CN)6








networks placed on the substrate. In a side view and starting from the bottom, the bilayer

film contains a base substrate, amphiphilic tails, two layers ofNi-Fe(CN)6 networks, and

the terminating amphiphilic tails. The multilayer film was generated by repeating the

bilayer fabrication process, and therefore, contains multiple stacks of bilayers, in which

each bilayer is separated by others via two amphiphilic tails (- 35 A).

In a local magnetic point of view, the bridging CN provides the ferromagnetic

superexchange interaction between the Ni" (S = 1) and Fe"I (S = 1/2) ions, whose

separation distance is 5 A. The Tc of the 3D bulk material was reported as ~ 24 K with

an exchange constant J- 14 K [15]. However, the unit of Fe"I-CN-Ni" studied in

different structural environments yielded different experimental J values. Furthermore,

recent computational results suggest the J values of FeII-NC-Ni units vary according to

their bonding distances and angles, and in certain situations, the J value can be even

negative [16]. In other words, a subtle difference in the local structures can significantly

modify the magnetic properties of material. In addition to the J values, the Tc values of

the system are also predicted to be changed according to the local environment such as

the number of the nearest magnetic neighbors. For example in the previously studied

ferromagnetic Ni-Cr(CN)6 compounds, the Tc values changed from 53 K to 90 K as the

number of the magnetic neighbor increased from 4 to 6 [1]. Thus, the magneto-structural

effects must be elucidated to understand the fundamental magnetism and the potential

device applications.

Chapter 4 presents the magnetic properties ofmonolayer, bilayer, and multibilayers

ofNi-Fe(CN)6 LB films that were investigated using DC and AC magnetic property

measurement system (MPMS) in the temperatures between 2 K and 300 K and in







magnetic fields up to 5 T. In addition, the high field dependent magnetization was

studied up to 30 T using the vibration sample magnetometer (VSM) at the National High

Magnetic Field Laboratory (NHMFL). In the low field measurements, the films were

studied in both parallel and perpendicular orientations with respect to the external

magnetic fields (HE) in the frequency ranges from DC to 1 kHz. The main discussion in

Chapter 4 is focused on the correlation between the magnetic and structural properties of

the films as they evolve from the 2D to 3D system.

1.2 Co-Fe(CN)6 Films

As mentioned earlier, Co-Fe(CN)6 Prussian blue analogs showed striking

photoinduced magnetism.2 In 1996, Sato and coworkers discovered that the

magnetization of the bulk compound K0.2Co1.4[Fe(CN)6]'6.9H20 increased under

irradiation with red light [38]. This enhanced magnetization remained nearly constant at

5 K, even when the light was off. To make things more interesting, after the

photoinduced magnetic state has been formed, the magnetization of the system returned

to nearly the initial state by irradiation of blue light or by thermally cycling the material

to 150 K. In the dark state, the compound Ko.2Coi.4[Fe(CN)6]-6.9H20 is a

ferrimagnetic material with a Tc of- 16 K. As a result of red light irradiation, the Tc of

the compound increased to 19 K, and the magnetization at T = 5 K and HE = 5 T

increased by 10% compared to its dark state value.

For this photoinduced magnetic system, we utilized a sequential deposition method

to generate a variety of Co-Fe(CN)6 films. The original motivation was to construct


2 The phrase "photoinduced magnetism" is often used in the field of semiconductors, and
this terminology has an association with a specific process. In this dissertation, the
phrase "photoinduced magnetism" is used in a more general sense to refer any change in
magnetic behavior during, or as a result of, irradiation.








magnetic films that exhibit a fast response to the light by minimizing the photon

attenuation. In a conventional bulk Co-Fe(CN)6, the light tends to attenuate, and thus it

takes a considerable amount of time to switch photoinduced magnetization. However in

the course of the investigation, we have discovered an unusual anisotropic photoinduced

effect. Briefly, we have observed an increase or decrease of photoinduced magnetization

controlled by the orientation of the film with respect to HE [50, 51]. This unique effect

has never been previously observed and the potential extensions to optically controlled

magnetic devices are significant. Furthermore, the area of photoinduced magnetic films

in Prussian blue analogs had not been actively developed prior to our investigations.

Thus, our investigation on Co-Fe(CN)6 films gave us the opportunity to be at one of the

frontiers in the research.

In Chapter 5, anisotropic photoinduced magnetism in Co-Fe(CN)6 films is

presented. For a comparison study, using sequential deposition methods two similar but

different films of RbjCok[Fe(CN)6]rnH20 were generated [51]. The difference in the

films lies in the arrangement of the domains within the samples. For example in film 1,

the particles that constitute the powder are randomly deposited, giving rise to a bulk-like

texture. On the other hand in film 2, the particles are more uniformly arranged parallel to

the films, resulting in a quasi-2D texture within continuous patches of approximately

600 nm, as verified by scanning electron microscopy (SEM) images. The photoinduced

magnetic properties in the bulk-like film were observed to be different, especially with

respect to the anisotropic photoinduced magnetization, from the 2D-like film. More

specifically, in the 2D-like film, the magnetization increased under irradiation of visible

light when the film was placed parallel to the HE. However, when the film was placed







perpendicular to the HE, the photoinduced magnetization decreased. The main discussion

of Chapter 5 is devoted to the mechanism of this unique anisotropic photoinduced

magnetism.

1.3 Co-Fe(CN)6 Powders

In addition to the photoinduced magnetism, the compound Co-Fe(CN)6 was

reported to exhibit an interesting charge-transfer-induced spin transition (CTIST) [46, 57,

58]. In 2002, Shimamoto and coworkers discovered that the magnetization of a

microcrystalline NaxCoy[Fe(CN)6]-nH20 powder sample decreased sharply at temperature

~ 180 K upon cooling and increased sharply at 220 K upon warming when x = 0.37, y =

1.37, and n = 4.8 [57]. This hysteric, first-order, spin transition resembles the properties

of spin crossover (SC) materials and can be applied to the magnetic memory devices due

to its spin control characteristics [80]. As in the case of photoinduced magnetism, the

CTIST process occurs in a cooperative manner and shows sharp spin transitions.

However, the phenomenon is strongly influenced by the subtle differences in the

chemical compositions and structures. For example, when the Co/Fe ratio changed from

1.37 to 1.52, the CTIST disappeared and the spin state of Co was locked in the high spin

state at all temperatures. Furthermore instead of Na, when other interstitial alkali metals

were used, the phenomenon and the condition of CTIST seemed to be restricted [57].

Therefore, studying the magneto-structural effect is essential to understand the CTIST

phenomenon.

During our investigation of the photoinduced magnetism, an interesting CTIST

phenomenon was observed in KxCoy[Fe(CN)6]-nH20 with x 0.6, y 1.2, and n 4. In

usual CTIST and SC phenomena, a rapidly cooled sample bypasses the transition from

the high temperature (HT) phase to the low temperature (LT) phase and will be trapped in









the HT phase even at low temperature [46]. Upon slow warming, the sample eventually

releases all the HT phase energies at certain temperatures and follows the LT phase

warming curve. However in our warming investigation, the magnetization of rapidly

cooled Ko.6Coi.2[Fe(CN)6]'4H20 sample became smaller than that of the regular LT phase.

This new LT phase can be achieved only by rapid cooling and has never been heretofore

observed. Therefore in Chapter 6, we report the magnetization measurements of

Ko.6Col.2[Fe(CN)6]'4H20 powder in its three different states (i.e. slowly cooled state,

rapidly cooled state, and rapidly cooled/warmed state). The main discussion in Chapter 6

is an attempt to describe this new LT phase quantitatively and introduce recent theoretical

developments on phase transitions in Co-Fe(CN)6 Prussian blue analog systems [78, 79].

1.4 Other Magnetic Systems

Along with Prussian blue analog materials, other magnetic systems were also

investigated. Although the experimental techniques and the results are interesting, the

physical nature of the studied materials does not represent the main theme of this

dissertation. Therefore, some of the appendices are dedicated these techniques and

results. In Appendix A, a capacitance measurement [81] and magnetization of the S = 1/2

magnetic chain material (CH3)2NH2CuCl3 [82 90], otherwise known as DMACuCI3 are

presented. This unique magnetic chain compound seems to possess two similar but

different alternating magnetic chains parallel to each other, and shows interesting field

dependent magnetic information. In Appendix B, an investigation of the low temperature

transport properties of Ca2-SrxRuO4 are presented. The main focus on this material was

to explore Fermi-liquid behavior at the critical concentration x = 0.5 [91].







CHAPTER 2
EXPERIMENTAL TECHNIQUES

This chapter focuses on the methods and analysis techniques used to probe the

magnetic properties of materials. For the macroscopic magnetic measurement, a

Quantum Design MPMS (Magnetic Property Measurement System) was used. The

MPMS is a computer controlled magnetometer, whose detecting ability is enhanced by

SQUID (Superconducting Quantum Interference Device) electronics. Two customized

models (MPMS-5S and MPMS-XL) of commercial SQUID magnetometers from

Quantum Design, Inc., San Diego, CA were used for the DC and AC magnetic

measurements. The model MPMS-5S is an older system that is equipped with a 5 T

longitudinal superconducting magnet and with options including AC measurement,

magnet reset, environmental magnetic shield, and degauss shield. The model MPMS-XL

is provided with a 7 T longitudinal superconducting magnet but possesses none of the

options of the MPMS-5S. However, the MPMS-XL has an additional continuous

impedance tube, which allows the system to hold a temperature continuously even below

4.2 K. The detailed specifications and relevant technical notes of the MPMS can be

found in the MPMS manuals [94, 95] and on the company's web site.3

This chapter first introduces the configuration of the MPMS hardware and the

principle of the DC magnetic measurement in Section 2.1. Second, a typical procedure

for the DC magnetic measurements will be presented in Section 2.2, and the simple data

analysis strategy will be given in Section 2.3. Third, the construction and properties of


3 http://www.qdusa.com







the homemade fiber optic sample holder (FOSH) will be described in Section 2.4. Before

summarizing, Section 2.5 reviews the various ways of packing and holding sample and

outlines the treatment of the background magnetic signals arising from the holders and

sample rods. Finally, Section 2.6 summarizes and introduces the future projects of

constructing multi-purpose sample rod for the MPMS.

2.1 Configuration of MPMS Hardware

In general, our MPMS-XL consists of six main parts: a sample rod, a probe, a

longitudinal superconducting magnet, a liquid helium dewar, a console cabinet, and a

computer, as schematically shown in Figure 2-1.

2.1.1 Sample Rod

A typical sample rod provides a place to mount a sample and can be inserted into

the probe at standard operating temperatures (1.9 K 400 K). A long thin stainless steel

tube (39 inch) is connected to a short quantalloy (silicon copper alloy) tube (8 inch) to

make the body of sample rod.4 The sample rod goes through a slide seal assembly, which

engages to the probe head and allows the sample rod to move in the longitudinal direction

while maintaining an air leak tight environment on the sample side. The sample is

usually placed inside a transparent beverage straw, and the straw is fixed to the end of

quantalloy side of the sample rod. Although the beverage straw provides a sufficient

housing for most samples, the auto tracking function of the MPMS, which compensates

the thermal contraction of sample rod, was originally calibrated for quartz sample tube.


4 The long stainless steel tube extends from room temperature down to the sample space.
In the sample space near the sample detection area, the quantalloy tube was used. The
stainless steel was chosen to maximize the mechanical rigidity and minimize the thermal
heat leak from room temperature to the sample space. The quantalloy tube was used to
minimize the extrinsic magnetic contribution to the detector. The both tube have outer
diameters of ~ 0.12 inch and the inner diameters of ~ 0.10 inch.
















































-liquid helium dewar


Figure 2-1. Schematics of MPMS (Magnetic Property Measurement System). In general,
our MPMS-XL consists of six main parts: a sample rod, a probe, a
longitudinal superconducting magnet, a liquid helium dewar, a console
cabinet, and a computer connected to the internet.







As a consequence, there is a small drift in sample position with respect to the magnet

center, when the straw was used as a sample housing, even when the auto tracking

function is enabled. When the straw was used, a typical sample center shift between

310 K and 2 K is 0.1 cm, which is not negligible when studying the absolute

magnetization of a spatially small sample.

2.1.2 Probe

The probe contains many parts, such as the probe head, the sample space, a cooling

annulus, the liquid helium reservoir, primary and secondary impedance tubes,

thermometers and heaters, a detection pickup coil, and the SQUID electronics. Briefly,

the main function of the probe is to provide the computer controlled temperature (within

0.5%) environment for the sample and to measure the SQUID enhanced magnetic signals

from the sample. The probe is sitting in the liquid helium dewar and regulates the

temperature of the sample space by withdrawing the liquid helium through the impedance

tubes to the cooling annulus, which surrounds the sample space concentrically as shown

in Figure 2-1. The liquid helium in the cooling annulus is further manipulated by the PID

(Proportional, Integral, and Derivative) temperature controller and by adjusting the

pumping power to deliver the desired temperature to the sample space. The heat transfer

between the cooling annulus and the sample space is meditated by the conduction

through a copper jacket that shields the sample space. The sample space is filled with

low pressure helium gas (- 0.1 mbar at room temperature) to provide heat exchange

between the sample and the wall of the sample space.

Since the temperature control mainly relies on the helium flowing through the

impedances, the regulation of the amount withdrawn from the dewar is critical. The

worst case will be realized if no helium is available to pass from the dewar to the cooling








annulus. Occasionally, this unexpected incidence happens due to the plugging in the

impedance tubes. The exact causes of the plugging are not certain but can be related to

impurities such as ices (e.g. water ice and nitrogen ice) that nucleated to block the

impedance tubes. For this reason, great care is needed when transferring the helium from

the transport dewar to the system dewar. In order to filter the impurities from the

transport helium dewar during the transfer, a homemade filter, which attaches to the end

of the transfer line, was built. Figure 2-2 shows a schematic view of the homemade filter.

The filter was assembled by placing charcoal particles in an existing stainless steel

porous filter. Wool-like stainless mesh was also inserted to prevent the charcoal particles

from moving. Before transferring liquid helium, the filter was warmed to 330 K or

higher, while blowing dry helium gas through it. The process was repeated two or three

times to remove water contaminants in the charcoal filter.

If the impedance tubes become completely plugged, temperature control is lost. In

this unfortunate case, the whole system has to be warmed by boiling the liquid helium,

and the impedance tubes must be cleaned. Most of time, if ice blocks the impedance, it

will melt and unplug the impedance when the system is warmed to room temperature.

However in some cases, impurity, such as residual oil, blocks the impedance tubes and

does not melt upon warming. In this case, the impedance tubes must be cleaned by

blowing helium gas through the cooling annulus pumping line, or the impedance tube

itself has to be replaced with a clean one. On the other hand when the impedance is

partially plugged, the control of the temperature is partially enabled. The quick solution

for this partial plugging is to set the probe temperature to 300 310 K for an extended

period, usually overnight, expecting the impurity ice to melt and to be pumped. However,








when this quick solution fails, the whole system has to be warmed as in the case of

complete plugging.


stainless steel liquid helium transfer line


helium recovery line


liquid helium transfer line

I


MPMS liquid
helium dewar


transport liquid
helium dewar


Figure 2-2. Schematics of the impurity filter and liquid helium transfer. In order to filter
the impurities from the transport helium dewar, a homemade filter, which
attaches to the end of transfer line, was built. The filter was assembled by
putting charcoal particles in an existing stainless steel porous filter. The
stainless mesh wool was also inserted into the short tube connecting the filter
to prevent the charcoal particles from moving. Before transferring liquid
helium, the filter was warmed to 330 K or higher, while blowing dry helium
gas. The process was repeated two or three times to remove water
contaminants in the charcoal filter.


impurity
filter







2.1.3 Console and Computer

The pumping tubes and electric cables from the probe and the magnet are

connected to the console cabinet. The console cabinet consists of a pump, gas handling

system, a temperature controller, a magnetic measurement unit, and a power source for

the magnet. The main role of the console cabinet is to provide electronically controlled

pumping power to the probe for temperature regulation as well as electric controls to the

probe and the magnet. The computer is connected to the console cabinet via a GPIB

(General Purpose Interface Bus) cable and operates all activities of the console using

Multi-View, a LabVIEW based computer program that can be controlled locally and

remotely via the internet. Once the sample is placed, all other standard activities of

MPMS, such as sample positioning, ramping magnetic field, changing temperature,

magnetic measurement, and recording data, can be controlled using Multi-View program.

2.1.4 Principle of DC Magnetization Measurement

To measure the DC magnetization of the sample, the sample is first placed near the

bottom of the pickup coil, which is a single superconducting coil wound by one

clockwise turn at the top position, and then by two counter-clockwise turns at the middle

position, and finally by one clockwise turn at the bottom position, as shown in Figure 2-3

(a). This type of coil is called a second-derivative coil and the main purpose of winding

in different directions is to cancel the uniform external magnetic field (HE) from

superconducting magnet of MPMS and other stray fields in the lab. In principle, as a

sample moves along the axis of the detection pickup coil to the top position, an

electromotive force (EMF) is induced in the pickup coil. This induced EMF is

proportional to the sample magnetization, and the MPMS electronics detect the amplified

EMF signals using SQUID electronics as the sample moves along the pickup coil. The







detected signal is then fit into the calculated model curve [96] to give the actual magnetic

moments of the sample. Figure 2-3 (b) shows a typical detected signal and the fit as a


(a)



4.0 cm

3.5 cm





2.0 cm-




0.5 cm-
0.0 cm-


;ample rod
,II


-3 cm


sample/
superconducting
second-derivative
pickup coil


0 1 2 3 4
Position (cm)


O measured
- fit


HE =100 G, T 298 K
M =5.94 x 10-5 emu G


Figure 2-3. Second-derivative pickup coil and centering magnetic signal. The Second-
derivative pickup coil (a) is a single superconducting coil wound by one
clockwise turn at the top position then by two counterclockwise turns at the
middle position and finally by one clockwise turn at the bottom position. In
principle, as a sample moves along the axis of the detection pickup coil to the
top position, the electromotive force (EMF) is induced to the pickup coil.
This induced EMF is proportional to the sample magnetization and the MPMS
electronics detect the amplified EMF signals using SQUID electronics as the
sample moves along the pickup coil as shown in (b). The sample, in this case
moved 4 cm along the pickup coil from the near bottom of the coil (0.0 cm
position in (a)) at T = 298 K with HE = 100 G. For this detection 16.04 mg of
Ko.6Col.2[Fe(CN)6]'4H20 powder was used.









function of the sample position. In this case the sample, moved 4 cm along the pickup

coil from the near bottom of the coil (0.0 cm position in Figure 2-3(a)) at T = 298 K with

HE = 100 G. For this detection, 16.04 mg of Ko.6Co1.2[Fe(CN)6]-4H20 powder was used.

2.2 DC Magnetization Measurement Procedure

A typical DC magnetization measurement involves two steps. The first step is to

measure magnetization of the sample holder only and the next is a measurement of the

sample in the holder. The sample in powder or microcrystalline form is usually packed

into the holder such as gelatin capsule (gelcap) or plastic can (capped polyethylene vial).

The diamagnetic background signals from these holders are measured and then subtracted

from the total signal of the sample and the holder. The diamagnetic signals from

different holders are similar but not identical maybe due to the unequal numbers of

electronic centers in polymers [97]. Therefore, the background signal must be measured

individually and independently before the sample measurement, especially for a weak

magnetic sample, whose magnetic moment is comparable to that of the holder. Figure 2-

4 shows typical temperature and field dependent magnetic measurements of the plastic

can and the gelcap with their pictures. Since the magnetic pickup coil and the sample

moving distance have finite lengths of- 3 cm and 4 cm (to 12 cm) respectively, a short

sample and a symmetric (along the horizontal plane) sample packing are desired. The

sample size dependent magnetization was studied previously by Jackson and coworkers

[98] and the result shows that in order to minimize the sample size dependency, the

optimal value of the sample volume is between 10 mm3 and 15 mm3. Other useful

sample mounting consideration can be found in the Reference [99].







17











-2 -
-5
-4
S-4 gel-cap

o-6 l o can

0 -8 e
C-10 T=2K
-10

0 2 4 6 8
H (T)

-15-

I I I I I I
0 50 100 150 200 250 300
T (K)



Figure 2-4. Magnetization signals of the holders. The sample, in powder or
microcrystalline form, is usually packed into holder such as gelatin capsule
(gelcap) or plastic can (capped polyethylene vial). The diamagnetic
background signals from these holders are measured and then subtracted from
the total signal of the sample in the holder. The figure shows the typical
temperature and field dependent magnetic measurements of the plastic can
and the gelcap with their pictures.









Table 2-1. Zero-field-cooled magnetization measurement procedure.
Step Procedure
1 Center the sample and measure the magnetization at 300 K.
2 Cool the sample without the magnetic field to the lowest temperature.
3 Apply a suitable measuring magnetic field.
4 Center the sample by measuring and adjusting the positional dependent
__ magnetic signal as shown in Figure 2-2 (b).
5 Measure the magnetization upon warming the sample.


Table 2-2. Field-cooled magnetization measurement procedure.
Step Procedure
1 Center the sample and measure the magnetization at 300 K.
2 Cool the sample with the magnetic field to the lowest temperature.
3 Center the sample by measuring and adjusting the positional dependent
magnetic signal as shown in Figure 2-2 (b).
4 Measure the magnetization upon warming the sample.


Table 2-3. Magnetization vs. magnetic field (M vs. H) measurement procedure.
Step Procedure
1 Center the sample and measure the magnetization at 300 K.
2 Cool the sample without the magnetic field to the desired temperature.
3 Apply a small magnetic field for centering purpose.
4 Center the sample by measuring and adjusting the positional dependent
magnetic signal as shown in Figure 2-2 (b).
5 Measure the magnetization upon sweeping the field.


For an unknown molecule-based magnetic material, typical magnetization

measurement sequence involves a zero-field-cooled (zfc), field-cooled (fc), and

magnetization vs. field (M vs. H) measurements. These zfc, fc, and M vs. H sequences

are usually the standard measurements to characterize the overall magnetic properties of

unknown materials. Variations of these standard measurements are also performed

frequently in order to probe specific aspects of a magnetic property of a sample. Typical

procedures to perform the zfc, fc, and M vs. H measurements are listed in Table 2-1 to







2.3 DC Magnetization Data Analysis

After the magnetization measurement, the MPMS records values of raw

magnetization and its associated uncertainty. From these values and the information of

the sample, all relevant quantities, such as DC susceptibility (X), effective moment (Ceff),

susceptibility multiplied by temperature (XT), and inverse susceptibility (1/ ), can be

calculated. To generate these values from MPMS data, a script for ORIGIN, a scientific

plotting program was written and listed in Appendix C. The DC susceptibility (X =

M / H) is close to a true differential susceptibility (xif = dM/dH) only when the M is a

linear function of magnetic field (H), which is the case for the high temperature or low

field magnetization of paramagnetic samples.

2.3.1 Curie-Weiss Law and Dimer Model

When the magnetic data are plotted according to final quantities, a simple magnetic

analysis is possible from the trend of the graphs. Generally, the yT vs. T graph is useful

to extract the temperature dependent magnetic information of the sample. According to

the Curie-Weiss law, X can be described as

= C / (T- ), (2.1)

for ferromagnet when 0 > 0 and T > Tc, and for antiferromagnet when 0 < 0 and T > TN,

where the C is a Curie constant, 0 is a Weiss temperature, Tc is a Curie temperature, and

TN is N6el temperature. Therefore if the material is paramagnetic, i.e. 0 = zero, thenXT

becomes equal to the Curie constant C at all temperature. When the material is

ferromagnetic, the XT graph will increases quickly around Tc, and the 1/X graph will have

positive x-intercept at Tc. The antiferromagnetic behavior is characterized by decreasing

XT values or a negative x-intercept value extrapolated in a 1/x vs. T graph. A special, but

often encountered case of an antiferromagnet is ferrimagnetic material. This ferrimagnet












paramagnet
paramagnet


0 T


T IlT
Immt


0 Tc T


Ittt'l

ferromagnet






ferrimagnet


Figure 2-5. Magnetic plots of various types of magnet. (a) Inverse susceptibility of a
paramagnet. (b) Inverse susceptibility and spontaneous magnetization of a
ferromagnet. (c) Inverse susceptibility of an antiferromagnet. (d) Inverse
susceptibility and spontaneous magnetization of a ferrimagnet. (e)
susceptibility times temperature of ferromagnet, ferrimagnet, paramagnet, and
antiferromagnet.


usually contains antiferromagnetically interacting spins with unequal spin values so that

when the magnet is ordered its magnetization graph resembles the superposition of


0 TN








ferromagnet and antiferromagnet. The characteristic trend of T vs. T graph of

ferrimagnet follows that of the antiferromagnet first (decreasing T ) and then becomes

ferromagnetic (increasing XT abruptly) upon lowering temperatures. In addition to the

ferrimagnetism, some antiferromagnet shows weak ferromagnetism due to canting of the

spins [100]. In this special class of antiferromagnet, spins are antiferromagnetically

coupled but in slightly canted way. Therefore, even when the spin values of coupling

spins are same, the canting introduces a weak ferromagnetism. Figure 2-5 shows

schematically drawn magnetic plots for various types of magnetic materials and Figure

2--6 shows specific theoretical simulations ofXT vs. T plots for one mole of non-

interacting isotropic spin 1/2 dimers with intra-dimer magnetic exchange constant J in

Kelvin. In this theoretical dimer model, the magnetization of one mole of non-interacting

dimer described as


exp gH -exp ggBH
M(H,T)dm,,er = Ng, kT g kBT (2.2)
21+exp -- +exp --B+exp -
k,T ) kT T

is derived from the fundamental magnetic equation and Heisenberg exchange

Hamiltonian, respectively as

NJ [(-OE, /8H)exp(-E, /kT)]
Mo exp(-E, /k,T)


and

S= -J(SA S,)+ g(SA + S,)H, (2.4)

where N is Avogadro's number (6.022 x 1023 / mol), En is energy eigenvalues ofS = 1/2

dimer applied to Hamiltonian (Eq. 2.3), kB is the Boltzmann constant (1.38062 x 10-16







erg / K), J is the Heisenberg exchange constant in Kelvin, SA and SB are the spin operators

for spin A and B, g is a g-value, and H is an applied magnetic field. The En values of

each dimer are given by Eo = -J/4 +gHpB,E = -J/4, E2 = -J/4 gHuB, and E3 = 3J/4. The

simulations in Figure 2-6 are consistent with above Curie-Weiss law in Figure 2-5 and

shows that the XT value is increasing when J > 0 (ferromagnetic), decreasing when J < 0

(antiferromagnetic), and constant when J = 0 paramagneticc).

For typical molecule-based magnets, the Tc values are generally lower than room

temperature. Thus, the magnetic behavior above Tc can fit into Curie-Weiss law to give

a g-value provided that the mass and formula unit of the sample are known. The same

type of information can be also extracted from the M vs. H measurement for paramagnet

and saturated ferromagnet, by fitting the data to the fundamental magnetization function,5

1+1/2S
Stanh(3.359 x 105gH(2S+ 1)/T))


2S 28tanh(3.359 x 10~5 gH /T)j

for one mole of paramagnet case and by comparing the saturated magnetization (Msat) to

ideal saturation value, NgSpB. Of course, electron paramagnet resonance (EPR) can be

used to determine the g-value directly at a specific temperature. For a known amount of

material, comparing M vs. H of the material with the paramagnetic magnetization

function (Eq. 2.5) can also give a hint of magnetic interactions. For example, if the

material is ferromagnetic then, the slope of the initial increase in aM vs. H plot is steeper

than that of paramagnet and if antiferromagnetic, the slope will be less steep than the

paramagnet. Figure 2-7 shows M vs. H plots of theoretically simulated S = 1/2 dimer with

5 This equation is identical to Eq. 2-3. For practical computing reason, all physical
constants are already substituted and the final M value is in the unit of emu G / mol.

















C
S


S
a)



0
o
o


0 50


100 150 200
T(K)


250 300 350


Figure 2-6. Susceptibility times temperature simulations of dimers. Theoretical
simulations of XT vs. T plots were performed for one mole of non-interacting
isotropic spin 1/2 dimers with intra-dimer magnetic exchange constant J in
Kelvin. The simulation used Eq. 2-2 with three different J values at HE =
100 G. The XT value is increasing when J > 0 (ferromagnetic), decreasing
when J < 0 (antiferromagnetic), and constant when J = 0 paramagneticc).



ferromagnetic and antiferromagnetic intra-dimer interactions compared to the

paramagnetic material at two different temperatures (2 K and 5 K) assuming g = 2 in all

cases. The upper limit of HE in our MPMS is 7 T. However, when higher fields are


necessary, the resistive magnet at the NHMFL can be used up to 33 T.
















15
-J=0,T=2K
-JJ=20 K, T= 2 K

-

10 -

SJ= 0,
; 3 // T=5K J>0

J = 20 K, ferromagnetic dimer
2 T=5K J<0
SJ= -20 K -
antiferromagneic dimer
T=5K
J=0
paramagnetic dimer

0 ---- J--= =-20 K, T= 2 K
I 0 I I I I I I1
0 5 10 15 20 25 30
H (T)



Figure 2-7. Field dependent magnetization simulations of dimer model. Theoretical
simulations of M vs. H plots were performed for one mole of non-interacting
isotropic spin 1/2 dimers with intra-dimer magnetic exchange constant J in the
unit of Kelvin. The simulation used dimer model with three different J values
with g = 2 in two different temperature environments (T = 2 K and 5 K). The
upper limit of HE in our MPMS is 7 T but higher field can be reached at the
NHMFL up to 33 T.









2.3.2 History Dependent Magnets

For history dependent magnetic materials such as ferromagnets, ferrimagnets, spin-

glasses, and superparamagnets, the zfc and fc magnetizations are different under certain

circumstances. This difference between zfc and fc magnetizations is caused by

irreversibility of the zfc magnetization and has different origins according to classes of

the magnets, but the process of zfc in the samples is similar. For example, in the cluster

spin-glass-like material such as Ko.6Co1.2[Fe(CN)6]-4H20 powder, zfc magnetization

(Figure 2-8) shows characteristic increase in magnetization up to the freezing temperature

(TF ~ 9 K) then decrease in magnetization upon further warming. The zfc magnetization

process of this sample can be modeled and schematically depicted in Figure 2-8.

Briefly, when the sample was cooled below Tc value in zero field, the magnetic

moments begin to align ferrimagnetically and forms cluster. Upon further cooling below

TF, each cluster starts to freeze and the direction of each cluster (determined by net spin

direction within the cluster) tends to point in random directions to minimize the magneto-

crystalline energy. As a result, the net magnetization of the sample achieves its minimum

value due to the random distribution of spin cluster directions. At this point, a small

amount of applied HE (100 G in this case) would not be sufficient to align the direction of

the cluster or its constituent and other individual spins. Upon increasing temperature, the

thermal energy, however, will disturb the frozen clusters and individual spins at

sufficiently high temperature and will then enable the net cluster direction to be realigned

parallel to HE. As a result, the net zfc magnetization will increase until it reaches a

maximum at TF, where the magnetic spins in the cluster mostly aligned with the HE to








minimize Zeeman energy in balance with ferrimagnetic interaction and temperatures.

Upon further increasing temperature, the frozen cluster will disappear since the spin


4



3



02


0




0


5 10 15 20 25 30
T(K)


Figure 2-8. Field-cooled (fc) and zero-field-cooled (zfc) magnetization processes.
K0.6Co1.2 [Fe(CN)6] 4H20 powder was used for this measurement. The zfc
plot shows a characteristic increase in magnetization up to the freezing
temperature (TF 9 K) then the decrease in magnetization upon further
warming.








interactions are significantly weaker than the thermal energy fluctuations. As a

consequence, the net zfc magnetization will decrease and eventually becomes

paramagnetic at T >> Tc. The trend of a zfc magnetization plot contains useful

information such as the magneto-crystalline energy, the freezing energy given by

temperature, the strength of the magnetic interactions between the spins, and the softness

of magnet as judged by the strength of HE. All of the information are interrelated and

sometimes shows universal scaling behavior as indicated by the linear TF dependence on

HE2/3 in some spin-glass materials [91].

The fc magnetization, on the other hand, cooled with the HE and thus, the direction

of clusters will be aligned to the direction of HE as temperature decreases. The zfc

magnetization is irreversible in the sense that if the zfc magnetization is measured up to

anywhere below TF and cooled again, then the measured magnetization will not be the

same upon lowering temperature. Instead, the magnetization will follow a path that is

similar to the fc measurement curve [91]. In an actual magnetization measurement,

because of the limitation on the quantity of sample and the lowest temperature limit,

occasionally the HE as a measuring field has to be large compared to TF in the energy

consideration. In this case, the characteristic zfc curve (increase and decrease below Tc)

might not be apparent. When this situation occurs, a plot of the difference between the fc

and the zfc magnetizations helps to determine the relevant history dependent magnetic

information.

For a direct measure of Tc of history dependent magnets, a thermal remnant

magnetization (TRM) measurement can be considered. In a TRM measurement, the

sample is cooled to the lowest temperature to below Tc in the presence of a finite HE, and








the temperature dependent data are taken upon warming without any HE. Since, there is

no magnetic field applied, the paramagnetic and diamagnetic contributions from the

sample and holders are expected to be zero. The TRM along with zfc and fc

magnetizations measured on sequentially deposited RbxNiy[Cr(CN)6]z-mH20 film is

shown in Figure 2-9. The sample was cooled to T = 5 K while HE = 100 G, and then the

field was lowered to zero for TRM measurement. From the TRM, the Tc is determined as

~ 74 K. In comparisons between zfc, fc, and TRM measurements of spin-glass-like

materials or superparamagnets, the cooling and warming rates have to be considered

because the magnetizations in these materials are not in equilibrium and exhibits notable

time dependence. In fact, the zfc and fc measurements in Figure 2-9 were performed at

the same rate, but the sample was warmed slower than the others for the TRM

measurement.

For some spin-glasses such as Ag(Mn), the sum of the TRM and zfc yielded the

same quantity as fc magnetization when all the measurements were performed at the

same time rates [103]. An attempt was made to perform this analysis with a

RbxNiy[Cr(CN)6]z'mH20 film (Figure 2-9), even though the TRM was acquired with a

different time rate compared to the zfc and fc measurements. The result does not agree

with that of Mathieu and coworkers [103]. Another feature of history dependent magnets

is a hysteresis loop inM vs. H. plot below a certain temperature. For example in the

cases of superparamagnetism below the blocking temperature TB, the M vs. H plot

possesses a S shape loop. The detailed physics of the hysteresis loop differs according to

the different classes of magnets and is beyond the scope of this dissertation and will not

be discussed further.



































0 20


40 60 80
T (K)


100 120


Figure 2-9. Thermal remnant magnetization (TRM) of RbxNi[Cr(CN)6]z-mH20 film.
The TRM along with zfc and fc magnetizations measured on a sequentially
deposited RbxNiy[Cr(CN)6]z-mH20 film. For TRM measurement, the sample
was cooled to 5 K with HE = 100 G, and the temperature dependent
measurement was performed while warming without any HE.


2.5



2.0


3
0r
0
C-


1.5



1.0



0.5



0.0









2.4 Homemade Fiber Optic Sample Holder

For photoinduced magnetization studies presented in Chapter 5, a homemade fiber

optic sample holder (FOSH) for MPMS was used. This inexpensive homemade probe

was made to fit the existing parts of MPMS sample rod. Figure 2-10 shows schematics

of our homemade FOSH in the experimental setup. In the long stainless tube (- 54 inch

I.D. 0.10 inch, and OD 0.12 inch), ten strands of optical fibers (Ocean Optics) were

fed through. The room temperature side of the probe was bent in U-shape and filled with

epoxy (2850 FT). The counter balancing optical fibers are optional to give magnetically

symmetric sample environment. However, most of time, the background of the holder

(gelcap for example) was measured using optic sample rod. Therefore, when this

background is subtracted from total signal, the contribution from optical fibers is

automatically subtracted. Briefly, the choice of optical fiber (Ocean Optics, Model 200

UV/VIS, O.D 270 jtm) was optimized for visible light, and the typical power that was

used for a photoinduced magnetization measurement was about 1 ~ 2 mW measured at

the sample side when halogen lamp was used. Although the current design is sufficient

to see rough power dependence of photoinduced magnetization, the mechanically rigid

and optically efficient design is needed for in situ measurement of spectroscopic and

magnetization properties. To do so, one solid optical fiber or bunched fibers using

optical fiber collimator can be used in order to fit into the standard spectroscopic sources

and detectors.

2.5 Sample Packing and Background Consideration

As mentioned earlier in this chapter, proper packing of the sample contributes to

the accurate measurement of the magnetization. Basically, the MPMS measures the










magnetization by moving the sample for a finite distance (usually 4 cm) through the

pickup coil, which has a finite dimension (- 3 cm long and 1.9 cm in diameter).


optical


optical


4-
to light
source
S137 cm









stainless steel tube,
O.D. 0.3 cm.


assembly

-11.2 cm


Stainless steel
tube



optical fibers
carrying light
from the source



-sample


counter
optical fibers


---straw


Figure 2-10. A homemade MPMS optic insert rod. In the long stainless tube (- 54 inch
I.D. 0.10 inch, and OD 0.12 inch), ten strands of optical fibers (Ocean
Optics) were fed through the inner space. The room temperature side of the
probe was bent in U-shape and filled with epoxy (2850 FT). The balancing
optical fibers are optional and are present to provide a magnetically symmetric
sample environment.







Therefore, given the limitation of machine sensitivity, the magnetically symmetric

packing of the sample is important to compensate the finite detecting dimension. In

addition, the direction of the sample with respect to HE and the intrinsic core

diamagnetism of the sample have to be considered along with the magnetic nature of

holder to give a pure magnetic response from the sample. In the following two

subsections, various ways of sample packing and the consideration of background

magnetic signals will be discussed.

2.5.1 Sample Packing

Most magnetometers that use superconducting magnet as external magnetic field

sources measure the longitudinal component of magnetization. In other words, if the axis

of superconducting magnet points in the z-direction (longitudinal) as in our MPMS, the

magnetometer measures the z-component of magnetization (Mz). The transverse

components of magnetization (Mx and My) are also of great interest in some cases, but are

not applicable to our MPMS and therefore will not be discussed further. Figure 2-11 (a)

shows an example of symmetric packing when the homemade FOSH was used. In this

case, the optical fibers that are diamagnetic are positioned symmetrically with respect to

the center x-y plane of the sample assuming that the sample itself is packed symmetrically.

The total magnetization of this configuration will be the magnetization of the optical

fibers without the sample and the sample without the optical fibers. Since the optical

fibers are diamagnetic and longer than the scan length, the vacant volume created by

removing the sample will give a positive magnetic signal, which is proportional to the

diamagnetic strength of the optical fibers. If the counter balancing optical fibers were not

in place, the detected magnetic signal would not be symmetric, as in Figure 2-3 (b), and

an inaccurate measurement value would result. Of course, if the magnetization of the















*--straw
Optical fibers
carrying light
from the light
source


.-- /sample


counter
optical fibers


(b)
holder

sample

(X) (0) (0)


(c)



S1.4 cm


sample
n gelcap


(d) two gelcap tops
/


cm


films Mylar shell

IT 0.4 cm


Mylar spacer- venting hole
gelcap
films
straw space"


Figure 2-11. Various MPMS sample packing methods. A symmetric sample packing is
shown schematically in (a) and (b). In (b), the bad example of packing sample
is also shown (marked as X) along with good examples. For powder or small
samples, gelcap (c, d), adhesive transparent tape (e), and grease or epoxy can
be used as holders (f). For thin film samples, Mylar shell (g) or gelcap (h)
were used for photoinduced magnetic measurements.









sample is significantly stronger than the background (from optical fibers in this case), the

effect of asymmetric packing can be ignored. The contribution from the straw, which the

detection coil sees as a continuous medium, can be also ignored in most MPMS

measurements.

For a collection of small samples such as a powder or collection of microcrystals,

tight symmetric packing is also desired. Figure 2-11 (b) shows schematic representation

of good and bad packing of a microcrystalline sample. As shown first in figure, when the

sample is loosely packed in the holder, an accurate centering is difficult due to the

unequal distribution of magnetic weights with respect to the center of the sample. In

addition, when the sample is loosely packed and if the particles have a magnetic easy axis

(for example uniaxial anisotropy), the particle will tend to rotate with its easy axis along

HE and will give a value that is different than that of randomly oriented particles. For a

large amount (- 200 mg) or small amount ( < 30 mg) of samples, a gelcap provide a

good holder as shown in Figure 2-11 (c) and (d) respectively. The plastic can, however

should be used for a wet or water sensitive sample instead of a gelcap. For some of the

photoinduced magnetic measurements of a powder sample, transparent tape was used as

shown in Figure 2-11 (e). The powder sample was first sprinkled on top of the adhesive

side of a small section of tape, and then attached to the adhesive side of a larger piece that

can be wrapped around the straw. Occasionally, for the air sensitive sample, grease or

transparent epoxy can be used as a matrix that holds the samples, see Figure 2-11 (f). In

this case, a care must be made to make sure that the thermal contraction and chemical

properties of the grease or epoxy do not affect the magnetic properties of the samples.







For the photoinduced magnetic measurements of Prussian blue based thin films,

thin Mylar (thickness 100 um) was used as a substrate. Although a silicon wafer is a

good choice as a substrate for spectroscopic and even mounting purposes (properly cut Si

wafer can be wedged inside the straw), it gives a larger magnetic background than that of

thin Mylar films. When the directional dependence of the magnetization measurement

was needed, the Mylar film sample was cut in squares and stacked together inside a

Mylar cubic shell as shown in Figure 2-11 (g). The final assembly of the sample was

then wedged inside the straw to any desired direction with respect to HE. A gelcap can

also be used as a holder for a film sample for a directional dependent magnetic

measurement as shown in Figure 2-11 (h). In this case, a piece of straw and the blank

Mylar film were used as spacers. The final assembly is again pressure fit into the straw.

When the pressure fit is impossible due to smaller size of gelcap, two slots or microholes

were made in the straw to hold the gelcap inside.

2.5.2 Background Consideration

Although direct measurement of holder background and subsequence subtraction is

easiest way of removing the background effect, sometime, the measurement of the

background of an individual holder is impossible due to several reasons. In that case, an

indirect estimation of the background is also possible. For example, consider 1 mg of

paramagnetic powder (S = 1/2), which has a molecular weight, of 1000 g / mol is packed

in a typical gelcap (40 mg). The simulated experimental magnetic data for this sample at

T = 2 K is shown in Figure 2-12 (a) and is labeled as a "total". This total signal is a

superposition of the magnetization of sample itself and the background. Assuming that

the background is a linear function of field and is temperature independent, a linear fit to

the total plot at higher field will give the slope that comes from the background. This
























H(T)


0 5 10 15 20 25 30
H (T)


H(T)
sample:
mass: 40 mg
M.W.: 1000 g / mol
S = 1/2, g = 2


Figure 2-12. Simulation of weak paramagnetic sample in gelcap holder. A field
dependent magnetization of a theoretical paramagnetic sample (1 mg, S= 1/2,
M.W. = 1000 g/mol) and gelcap holder (40 mg) is simulated at T (a) = 2 K,
T (b)= 10 K, and T (c) = 400 K.



estimation is possible because of the saturation of the paramagnetic magnetization at a

low temperature. When the temperature is higher, for instance at 10 K as shown in

Figure 2-12 (b), a similar type of background estimation will be difficult. Instead, the


holder: gelcap
mass: 40 nig
The background of the
holder %\as assumed to be
temperature independent.







field dependent magnetization can be measured at higher temperature where the

magnetization from the sample is a minimum, as shown in Figure 2-12 (c) to estimate the

background of holder. The origin of the background of the holder system, however, is ill-

defined in the case of Figure 2-12 (a) because the estimated background from the linear

fit of the magnetization curve actually contains the core-diamagnetic contribution from

the sample as well as the background from the holder.

2.6 Summary and Future Direction

In this section, the configuration of the MPMS and the typical magnetic

measurement processes, as well as data analysis strategies were reviewed. Briefly, the

MPMS is a wide-temperature ranged (1.9 K 400 K), computer-controlled

magnetometer using SQUID electronics to enhance magnetic detection. The typical

sample measurement consists of zfc, fc, and M vs. H sequences. From the trends of the

data and Curie-Weiss fitting processes, the magnetic properties of the samples can be

analyzed. More specifically, by analyzing the behavior ofXT vs. T and M vs. H plots, the

sample can be categorized roughly into, paramagnets, ferromagnets, antiferromagnets,

and ferrimagnets. For special magnetic studies, such as photoinduced magnetism, a

homemade FOSH was constructed to measure the magnetizations under irradiation of

light. This homemade FOSH consists of 10 strands of optical fibers in stainless tube, and

the typical light power used for photoinduced magnetization experiment was about

1 2 mW / cm.

Although it is limited by the size of sample space and operational temperature, the

great advantages of using MPMS are: (1)the accuracy in magnetic measurement, (2) the

convenience of the sample changing environment using the sample rod, and (3) a fully

automated computer-controlled temperature and measuring sequence. These advantages

















bring systematic and reproducible efficiency to these experimental magnetic studies. As

long as the dimension of the experimental space is compatible, the MPMS can be used

not only as a magnetometer but also as a general purpose low or high temperature

measurement device by revising the existing sample rod. Some of the modified sample

rods for transport, optical measurement, and even pressure study are already

commercially available, but at non-negligible costs. In order to maximize the efficiency

of MPMS and to minimize the cost, a cost effective multi-purpose sample rod can

considered. In this new sample rod, the top of the room side stainless tube is attached to

the quick connector and provides vacuum sealed connection between the sample space

and outside. In principal, any thing that fits into the quick connector and stainless tube

can be inserted into the sample space. For example, optical fibers and thin coaxial cables

can be introduced at the same time to measure a magneto-optical-transport property under

the irradiation of the light. Eventually the cooling power of the system and the size of

probe space will limit the use of this multi-purpose sample rod, but otherwise the uses

will increase the efficiency in many areas of experimental physics.







CHAPTER 3
PHYSICAL PHENOMENON AND THEORY

3.1 Photoinduced Magnetism in Prussian Blue Analogs

3.1.1 Overview

The phenomenon of photoinduced magnetization in a Co-Fe(CN)6 Prussian blue

analog was first observed by Sato and coworkers in 1996 [38]. According to their studies,

a ferrimagnetic compound Ko.2Co1.4[Fe(CN)6]-6.9H20 showed increases of magnetization

upon irradiation of red light. Furthermore, the resultant photoinduced magnetization

disappeared under irradiation of blue light or by increasing temperature to -150 K. This

unique control of magnetization by light and temperature attracted numerous researchers

and opened a new field of functional molecule-based magnets [38 79], which could be

useful for future magnetic devices. This section introduces the phenomena and proposed

theoretical descriptions of photoinduced magnetism in Co-Fe(CN)6 Prussian blue analogs

to provide a theoretical background to the anisotropic photoinduced magnetism described

in Chapter 5.

3.1.2 Initial Observation and Description

Based on the result of Sato and coworkers, the three-dimensional (3D) bulk

compound Ko.2Co1.4[Fe(CN)6]'6.9H20 undergoes a short range ferrimagnetic ordering

below Tc 16 K in the dark state [38]. The majority of sample in this dark state was

proposed to be a mixture of diamagnetic Co-Fe pairs and ferrimagnetically coupled

Co-Fe pairs distinguished by the oxidation states of ions and the consequent spin

configurations. More specifically, some portion of the sample is composed of








diamagnetic pairs of low-spin Co"' (tg S= 0) ions and low-spin Fe1 (t', S= 0) ions

bridged by CN in the form of CoIINC-FeII. The other portion on the other hand, consists

of magnetic pairs of high-spin Co" (t2ge S = 3/2) ions and low-spin FeI" (tg, S = 1/2)

ions forming CoI-NC-Fe.' units. These magnetic spins of Co" and Fe"' in the original

dark state, in contrast to the diamagnetic spins of CoIII and FeI, will be referred as

primordial spins to be differentiated later by the photoinduced spins that are created as a

result of light irradiation. Upon reducing temperature below Tc, some primordial spins

interact antiferromagnetically via superexchange and yield a net ferrimagnetic moment

with effective S = 1 per CoI-NC-FeIII pair. The first microscopic evidence of

ferrimagnetic interactions between Co"I and Fel" ions in a similar compound

Rbi.sCo4[Fe(CN)6]3.3'13H20, was investigated by Campion and coworkers in their

K-edge X-ray magnetic circular dichroism studies [45].

When Sato and coworkers illuminated the Ko.2Coi.4[Fe(CN)6]-6.9H20 compound at

T = 5 K with red light, the magnetization increased, and this photoinduced magnetization

lasted for several days. As a result of the photoinduced enhancement of the

magnetization, the Tc increased from approximatelyl6 K to 19 K and the coercive field at

5 K was enhanced from 800 G to 1500 G. The results of our sequentially deposited

RbjCok[Fe(CN)6]rnH20 films, which have a similar face-centered cubic structure and

photoinduced enhancement of the magnetization, are shown in Figure 3-1 and 3-2. A full

discussion of the particular film shown in Figure 3-1 is presented in Chapter 5. Briefly,

the zero-field-cooled (zfc) and field-cooled (fc) magnetizations are plotted in Figure. 3-1

as a function of temperature before and after white light irradiation, and Figure 3-2 shows

time dependent magnetization under white light irradiation. The bumps in the data in













20 l o pnotoinauceaC lf,
photoinducedMff
0 a dark Mf
^15 hv *<*0 darkMf,
HE = 200 G film




T (K)
10 -



5 ~


TP

0-

10 20 30
T (K)



Figure 3-1. Temperature dependent photoinduced and dark magnetizations of
sequentially deposited RbjCok[Fe(CN)6]rnH20 film. The zero-field-cooled
(zfc) and field-cooled (fc) magnetizations are plotted as a function of
temperature before and after irradiation with white light. As a result of
irradiation, the overall magnetizations increased. In addition, Tc increased by
~ 2 K and Tp, the temperature where zfc magnetization shows a maximum
value increased by 2 K.












1.2 i 1 1 1 1

T= 5 K, HE= 1 kG

1.0
white off
light on
4.2
0.8
0 8 blue LED

C-)
1 light on
0.6 f `4.0 -

SO n
0.4- 3.8 off red LED
0 light on

0.2 -g -
30 25 30
Time (hrs)
0.0 I ***
0 1 2 3 4 5
Time (hours)



Figure 3-2. Time dependent photoinduced magnetization. Under irradiation of white
light at 5 K, the magnetization increase to 500% of its original value over
4 hours of period. The inset shows magnetization responses under blue and
red light irradiation. In both cases, the magnetization increased under
irradiation.








Figure 3-2, when light is toggled on and off, are due to a thermal effect associated with

the light being a local heating source. The trends (i.e. overall increase in the

magnetization and enhancement of Tc at the photoinduced state) of our data are consistent

with those observed in Ko.2Coi.4[Fe(CN)6]'6.9H20 [38]. However in our case, there was

no decreasing in the magnetization upon irradiation with blue light as shown in the inset

of Figure 3-2, and the absence of this demagnetization effect might come from the use of

different alkali metals (e.g. Rb in our case) and has been reported by other group [43].

The Sato and coworkers were the first to suggest that the mechanism of

photoinduced magnetization was due to an internal photochemical redox reaction and this

picture was further supported by Verdaguer [39]. According to Verdaguer, a

Ko.2Col.4[Fe(CN)6]-6.9H20 compound (atypical Co-Fe(CN)6 Prussian blue analog) has a

face-centered cubic structure with some complex chemical constitutes: primordial

magnetic units of Co"- CN-FeII, water filled Fe vacancies as shown in Figure. 3-3 (a)

(sites A), isolated diamagnetic units of FeI(CN)6 (Figure. 3-3 (a) B sites), and

diamagnetic pairs of Co"I-NC-Fe" (Figure. 3-3 (a), C sites). Below Tc when the sample

is exposed to the red-light, the charge transfer occurs between the Fe and Co ions within

the diamagnetic pair (Figure. 3-3 (a), C sites). The electron from the Fe"I ion transfers to

CoIII ion and the diamagnetic pair becomes a photoinduced magnetic FeII-CN-CoI pair,

which is similar to the primordial magnetic pair. This photoinduced magnetic state of

spins may subsequently return to its initial diamagnetic state upon the blue light

irradiation via the process of electron transfer from Con to Fe11'. These charge transfer

processes are schematically shown in Figure. 3-3 (b) for photoinduced magnetization (red

light) and demagnetization (blue light).











0*

A


O 0 0 B
0^$'


0


Co
Fe


H20


* C-N 0


S=ORed
S=0


( NC
S=O


S=0-
S=Q


SNC (Fe --
Charge Transfer



Charge Trafer
Charge Transfer


Figure 3-3. Processes of photoinduced magnetization and demagnetization described by
Verdaguer [39]. The model compound contains primordial magnetic units of
FeI'-CN-Co", water filled Fe vacancies (sites A), isolated diamagnetic units of
Fen-(CN)6 (B sites), and diamagnetic pairs ofCoI"-NC-FeI (C sites). When
the sample is exposed to the red light, the charge transfer occurs between the
Fe and Co ions within the diamagnetic pair (C sites, top). The electron from
the Fe" ion transfers to Cor" ion and the diamagnetic pair becomes a
photoinduced magnetic Fe"-CN-Co" pair, which is similar to the primordial
magnetic pair. This photoinduced magnetic state of spins returned to the
initial diamagnetic state upon the blue light irradiation via the process of
electron transfer from Con to Fe"'. These charge transfer processes are
schematically shown in (b).


S=O


Co" NC
S = 3/2


3/2
S = 3/2


S= 1/2


NC e
S= 1/2








As a consequence ofphotoinduced magnetization, the net number of magnetic

neighbors increases, and this increment contributes to the enhancement of Tc in the

photoinduced magnetic state. Verdaguer also indicated that these charge transfer

processes occur at the local level (for example, at site C of Figure 3-3 (a) under red-light)

but activate the cooperative magnetic interactions through out the sample. In fact, the

cooperative behavior was observed by the Verdaguer group when studying

Rb0.52Co[Fe(CN)6]0.84-2.3H20 [48]. The reversible photoinduced magnetization scenario

of Verdaguer is to be consistent with the subsequent theoretical descriptions based on ab

initio, quantum chemical cluster calculations by Kawamoto and coworkers [55]. The

calculation was performed on the experimentally studied compound

Ko.4Co1.3[Fe(CN)6]-5H20 [66], which showed little or no evidence of primordial spin

pairs of Fe"I-CN-Co" when compared to the original Ko.2Col.4[Fe(CN)6]'6.9H20 system.

As a consequence of the absence of primordial spins, the model compound did not show

ferrimagnetic ordering in dark state. The significance of the concentrations of each

chemical constituent and the choice of the alkali ions will be discussed later in this

chapter. Briefly, the experimental results on KO.4Co1.3[Fe(CN)6]-5H20 [66] indicated

weak paramagnetic behavior between 2 K and 340 K in the dark state, and upon

irradiation of light (500 700 nm) at 5 K, the material became ferrimagnetic with Tc ~

26 K. Upon further irradiation by infra-red (IR) light (1319 nm), the magnetization

returned to its initial weak paramagnetic state.

In order to explain reversible photoinduced magnetism, Kawamoto and coworkers

[55] proposed four relevant spin states of the Co-Fe pairs as schematically shown in

Figure. 3-4 (a). This theoretical description will be referred as Co cluster model. By









going through these four different spin states, a complete photoinduced magnetization

and demagnetization cycle can be described. The first state (LSO) begins at the ground

state of the diamagnetic pair of Co"I (t6 S = 0)-NC-Fe"(tg,, S = 0), and this ground

state can be excited to the second state (LS1) by visible light (500-700 nm). The LS 1

state, which is an intermediate state, is composed of a CoI ( t e S= 1/2)-NC-Fe1 (t',

S= 1/2) pair. Without any external stimuli such as light irradiation, the CoII (t6e, S=

1/2) ions in the LS1 state quickly relaxes into a high-spin configuration, Co" (t5e S=

3/2), via intersystem crossing. In this HSO sate, the photoinduced enhancement of the

magnetization is complete, and if the sample is below Tc, then the spins in HSO state will

be coupled antiferromagnetically, generating an effective ferrimagnetic moment of S= 1

(3/2 1/2) per Co-Fe pair. On the other hand, if T >> Tc, a paramagnetic moment of S=

2 (3/2 + 1/2) per Co-Fe pair will result. Upon further irradiation with IR light, the HSO

state can be promoted to an intermediate excited state (HS 1) in the Co'I ( tte S = 1)-

NC-Fe"I( t6, S = 0) configuration. This intermediate HS 1 state will again be transferred

to the stable diamagnetic ground state (LSO) via intersystem crossing, thereby completes

one cycle of the photoinduced magnetization and demagnetization process.

Kawamoto and coworkers also realized that, in the K.o4Co1.3[Fe(CN)6]'5H20 model

compound, if the water-filled Fe-vacancies are randomly distributed inside the sample,

then 37% of the Co ions are in a (CN)5-Co-(OH2) environment (number of water per Co

is 1, Nw = 1) and 21% of the Co ions are in a (CN)6-Co arrangement (no water per Co,

Nw = 0), as shown in the insets Figure. 3-4 (b) and (c) respectively. Using an ab initio








calculation program, the potential energies of these Co clusters with Nw = 1 and 0 were

plotted for different states (LSO, LSI, HSO, and HS1) as a function of Co-N distance [55].


LSO
eg
t2g
Coll (S = 0)


S HS1

eg
or t( t=1 t2g t4t t4
Co" (S= 1) FeI (S= 0)
S HSO


Light 2


Light 1
Fe (S= 0)
LSI

Co ( =12) FegI ( 2)


Col (S= 1/2) Fe"' (S= 1/2)
-%i


- ( t=3 t2g t t. 1/
Co" (S = 3/2) Fell (S =1/2


Inter System
Crossing


N= 1


CO CN

%1 H20


(c)


N= 0


Figure 3-4. Complete cycle of photoinduced magnetization and demagnetization
proposed by Kawamoto and coworkers [55]. The complete process can be
obtained by going thorough LSO-*visible light-LSl S1HSO-IR
light--HSl--+ LSO cycle (a). Co clusters with one (b) and no water (c)
molecules, respectively.


Inter System
Crossing








From these calculated potential energy diagrams, Kawamoto and coworkers were able to

propose a complete photoinduced magnetization and demagnetization scenario, which is

consistent with Verdaguer's rationalization introduced earlier in this section. The basic

scenario can be summarized as a cyclic process of LSO- visible light--LS1--HSO--IR

light--HS1-* LSO cycle. In addition, they proposed that the photoinduced enhancement

of the magnetization originates at the (CN)5-Co-(OH2), Nw = 1 site in the compound,

because the calculation shows that the potential energy difference between LSO and LS 1

in Nw = 1 site is close to the experimentally observed value (500-700 nm). On the other

hand the photoinduced demagnetization process originates from the (CN)6-Co, Nw = 0

site. The potential diagram shows that the energy difference between the HSO to HS1

transition in the Nw = 0 site is close to that of experimentally observed value (1350 nm).

As Kawamoto and coworkers indicated, the importance of their results is to verify that

the photoinduced magnetization and demagnetization processes are initiated from the

spatially distinct local sites of Co with different ligand environments (mainly at Nw = 1

and Nw = 0).

In general this local effect, for example the photoinduced magnetization arising at

local (CN)5-Co-(OH2), induces an overall phase transition throughout the crystal to be in

stable photoinduced magnetic state. More specifically, when visible light (500-700 nm)

is used, a charge transfer in the Co-Fe pairs occurs if the local Co ion is in (CN)5-Co-

(OH2) environment. However, in the presence of visible light the charge transfer result in

the promotion of the of the LSO state to the LS1 state should not occur at the (CN)6-Co

site because, according to the potential energy diagram, the visible light lacks sufficient

energy. Therefore, if the cooperative behavior is not considered, upon irradiation with a







visible light, the (CN)5-Co-(OH2) sites will become magnetic, but the (CN)6-Co sites in

the crystal will remain diamagnetic. However, the experimental evidence shows that the

material undergoes a photoinduced phase transition as a whole and supports cooperative

behavior [48]. The same argument is applied for photoinduced demagnetization process:

starting from the (CN)6-Co sites and extending through the crystal upon IR light

irradiation to be in LSO ground state. The origin of the cooperative behavior can be

related to the lattice elongation of Co-Fe pairs when they undergo spin transition such as

photoinduced magnetic transition. Briefly, the local elongation of the lattice in the

photoinduced Co-Fe pair will affect neighboring diamagnetic Co-Fe pair and in the

process of minimizing lattice distortion energy, the elongation of the lattice and

associated magnetic properties will propagate throughout the crystals. If the material

contained only (CN)5-Co-(OH2) sites without (CN)6-Co sites, the photoinduced

magnetization would be possible but photoinduced demagnetization would not be

possible. Hence, in order to have reversible control of the photoinduced magnet,

Kawamoto and coworkers emphasized that the material has to be composed of a mixture

of(CN)6-Co sites and (CN)5-Co-(OH2) sites, that is to say that the Co-Fe(CN)6 Prussian

blue structure should contain water-filled Fe-vacancies. Before discussing additional

experimental results it is important to stress an obvious point, namely basic conditions

necessary for the existence of photoinduced magnetization is the presence of diamagnetic

pairs of Co"I-NC-Fe" in the sample.

3.2 Charge Transfer Induced Spin Transition

Although the Co cluster model can explain the reversible process of photoinduced

magnetization in Ko.4Co1.3[Fe(CN)6]-5H20, other experimental evidence on a similar

material did not show the photoinduced magnetization even if the material contained








mostly (CN)6-Co clusters. This absence of photoinduced magnetization, and the ability

to stimulate photoinduced demagnetization in similarly generated materials, suggest the

effects are linked to subtle differences in the chemical concentrations and to the choice of

different alkali metal ions. Considering electroneutrality, general formula for Co-

Fe(CN)6 based Prussian blue solid can be written as AxCo4[Fe(CN)6]y,04-ynH20 [44],

alternative to AjCok[Fe(CN)6]rnH20, where A is an interstitial alkali metal cation and H

is a Fe-vacancy. During the search for the optimal conditions for producing

photoinduced magnets, some Co-Fe(CN)6 Prussian blue structures were discovered to

undergo charge-transfer-induced spin transition (CTIST) at relatively higher temperature

compared to the Tc of the sample. The systematic studies [47, 57] show that the choice

of alkali metals and the relative concentrations given by x and y in the formula affect the

conditions for the CTIST to occur. Moreover, the condition for being a photoinduced

magnet is related to the occurrence of CTIST. Shimamoto and coworkers studied the

magnetic properties of a series of NagCoy[Fe(CN)6]-zH20 compounds [57] by

systematically varying the Co/Fe ratio, which in turn modifies the concentration of alkali

metal ions and water filled Fe-vacancies. Tables 3-1 and 3-2 show the chemical

composition and the valence states at 290 K of the five samples from Shimamoto and

coworkers.



Table 3-1. Chemical compositions of NaxCo[Fe(CN)6 ]zH20 [57].J
Sample Na Co Fe(CN)6 Vacancy H20 Co/Fe
1 0.07 1.50 1 0.50 6.3 1.50
2 0.37 1.37 1 0.37 4.8 1.37
3 0.53 1.32 1 0.32 4.4 1.32
4 0.60 1.26 1 0.26 3.9 1.26
5 0.94 1.15 1 0.15 3.0 1.15








Table 3-2. Valence states of NaxCoy[Fe(CN)6]-zH20 at 290 K [57].
Valence State at 290 K
1 Nao.07Co"1.50 [Fe"(CN)6]0.93[Fe"(CN)6]0o.076.3H20
2 Nao.37C111.37 [Fe "(CN)6].89 [Fe"(CN)6]o.114. 8H20
3 Nao.53C111.32 [Fe"(CN)6]0.83 [Fe (CN)6]0o.174.4H20
4 Nao.60Co 1.8oCo '.18s[Fe "(CN)6]o.70[Fe"(CN)6]o.o303.9H20
5 Na0.94Co 0.39Co0 0.76 [Fe"(CN)6]1.o003.0H20


The results of dark state magnetic measurements as XT vs. T and photoinduced

magnetization studies presented as M vs. T are schematically shown in Figure 3-5 (a) and

(b) respectively. These plots are not the actual data plots shown in Reference [57] but the

illustrations showing characteristics of the main results. Sample 1, which contains the

most Fe-vacancies and the least alkali metals shows a constantXT value between 50 K

and 350 K, whereas Sample 5, which contains the least Fe-vacancies and most alkali

metal ions shows constant but lower XT value compared to that of sample 1. In other

words, the nearly temperature independent XT values indicate that the two materials are

in the paramagnetic regime, and IR data suggest that the paramagnetic part of sample 1

and 5 are mainly composed of primordial spins of Co1 (t ge S= 3/2)-NC-Fe" (t ,, S

= 1/2) and Co" (t5g eg, S = 3/2)-NC-FeII (t6 S= 0), respectively, even though Sample 5

contains a majority of diamagnetic pairs, see Table 3-2. Furthermore, Samples 1 and 5

did not exhibit any photoinduced magnetization as shown in the Figure. 3-5 (b). Sample

1 failed to meet the basic condition (existence of Co-Fe diamagnetic pairs) for

photoinduced magnetism but the sample 5 contains plenty of diamagnetic pairs, and

moreover the majority the of Co ions are in a (CN)6-Co environment, which according to

Kawamoto and coworkers is a necessary condition for photoinduced magnetization. The

resolution of this apparent discrepancy between theory and experiment might come from








the consideration of the strength of different ligand fields holding the crystal together.

When the material undergoes a transition to the photoinduced magnetic state, the crystal

experiences a change of volume as indicated by the experiments and theories [55, 66]. In

similar Prussian blue structures, the flexibility of volume change is then related to the

ligand field strength that holds the crystal. For example Co with six CN (Nw = 0) will

hold crystal tighter than the Co ion with five CN and one water (Nw = 1) because the

ligand field of the CN is stronger than that of H20. As a consequence even though Co

with six bridging CN links is a good condition for photoinduced magnet, the strong

ligand field from all CN do not let the crystal relax to be in the photoinduced state.

Consequently, in order to have photoinduced magnetization, there needs to be a fine

balance between the number of(CN)6-Co configuration and lattice rigidity, which can be

weaken by substituting the (CN)5-Co-(OH2) motif to the (CN)6-Co site. This substituting

is equivalent to creating more water filled Fe-vacancies. In terms of Fe-vacancies, if

there are too many or too few Fe-vacancies, the sample will not undergo a photoinduced

transition. Too many vacancies will put the sample in already lattice-relaxed magnetic

primordial spin states plus lack of diamagnetic pairs, while too little Fe vacancy leave the

lattice too stiff to respond for photoinduced state of lattice relaxation.

On the other hand, the samples in the intermediate range of Fe vacancies such as in

samples 2, 3, and 4 show photoinduced magnetization as indicated in Figure 3-5 (b).

Moreover upon cooling, roughly constant XT values (close to the value of sample 1) fall

quickly at around 200 K, 240 K, and 260 K, and then become nearly constant again at

190 K, 230 K, and 250 K for sample 2,3 and 4 respectively. These changes in XT values

are characterized as CTIST and show hysteresis upon warming. For these samples (2, 3,






























































































































n*i



n*ij



II I*1



( I*1



I*I


10~ TRi





~1~.(-111 nl 1117--11~-111n*11I








numbers of Fe-vacancies of the samples decrease. This trend can be rationalized as a

result of competition between the ligand field and temperature: the stronger the ligand

field (less Fe vacancies) the sample has, it maintains the tight crystal structure to higher

temperatures.

The CTIST can be suppressed by rapid cooling (quenching) the samples [46, 56].

Hanawa and coworkers [56] argued that the photoinduced state of the sample is

essentially the same as the high temperature phase of the sample such as sample 2,3, and

4 in the states before thermally induced charge transfer occur and furthermore, they

investigated microscopic domain structures of photoinduced state of sample in

comparison to the rapidly cooled sample using high angle-resolved synchrotron-radiation

X-ray powder diffraction technique. From this study, Hanawa and coworkers concluded

that the structural domain size of the photoinduced state is larger than the quenched high-

temperature phase, and the lattice constant of photoinduced state is larger (10.287(1) A

for Co to Co distance) than that of the quenched phase (10.187(3) A for Co to Co

distance). The conditions for photoinduced magnets or CTIST magnets seem to be the

same in samples 2, 3, 4, and Hanawa's sample, but generally these conditions are not

necessary the same because some compounds with Rb or K as alkali metal ions do not

show the thermally induced charge transfers but do show photoinduced magnetization

[57]. Interestingly, one of the powder compounds that we have studied showed both

CTIST and photoinduced magnetization even if the compound contains K as an alkali

metal. In addition, the same sample also exhibited temperature independent primordial

spins indicated by ferrimagnetic ordering in xT plot after the thermally induced charge

transfers as shown in Figure 3-6. The expected formula for our powder sample is







Ko.6Coi.2[Fe(CN)6]-4H20 based on the chemical components used in synthesis procedure.
The XT plot of the sample also shows the macroscopic fingerprint of ferrimagnetic

interactions between the Co and Fe spins indicated by the decrease of T down to 50 K

and abrupt increase below -25 K. The magnetization measured at 5 K and up to 7 T are


10





8




O
06





4


0 50 100 150
T(K)


200 250 300


Figure 3-6. T vs. T data of Ko.6Co1.2[Fe(CN)6]-4H20 powder upon cooling and warming.
The inset shows the time dependent magnetization under irradiation of white
light.





56



4 o photoinducedM
darkM 0 O
o I
T=5K *
3 O *
00
1



01 M,-,
0 2


Sr Hj
0
-1 p
-0.1 0.0 0.1
H (T)
-1 I I I
0 2 4 6 8
H (T)



Figure 3-7. M vs. H data of Ko.6Coi.2[Fe(CN)6]'4H20 powder measured at 5 K upon
sweeping field from 7 T to -0.1 T before and after white light irradiation. The
inset shows zoomed view of magnetic data near the origin of the graph.


plotted in Figure 3-7 for the cases of both photoinduced and dark states. As indicated by
the arrows in the Figure, some characteristic values such as saturation magnetization (Ms),
remnant magnetization (Mr), and coercive field (Hc) are increased in the photoinduced


state.











3.3 Summary

From the all experimental and theoretical studies presented in this section, the

conditions necessary to support photoinduced enhancement of the magnetization can be

summarized as: 1. existence of diamagnetic Co-Fe pairs, 2. suitable local environment of

Co such as (CN)6-Co, and 3. structurally flexible lattice. Similarly, the condition for

photoinduced demagnetization can be summarized as 1. existence of magnetic Co-Fe

pairs, 2. suitable local environment of Co such as (CN)5-Co-(OH2), and 3. flexible lattice.

In general these three conditions are interrelated via chemical formula, i.e. choice of

alkali ions and relative concentration of chemical constituents. As external conditions, of

course, the choice of suitable range of light is most important but other variables such as

pressure, and orientations of the sample as shown in Chapter 5, can be considered. Since

the proper chemical concentration is the crucial variable in photoinduced magnetism and

CTIST, systematic studies varying the choice of the alkali metal and the concentration

between the constituents of the Co-Fe Prussian blue structure are needed to optimize the

protocol for generating a photoinduced magnet. Furthermore, understanding the physical

and chemical conditions for photoinduced magnetism and the theoretical predictions of

relevant potential energy diagram can contribute to generate various types of functional

magnets required for modem magnetic devices. To this end, Chapter 5 introduces unique

Co-Fe(CN)6 Prussian blue films that shows anisotropic photoinduced magnetic properties

and Chapter 6 reviews unusual CTIST phenomenon in Ko.6Co1.2[Fe(CN)6]'4H20 powder

when the specimen was cooled rapidly below 100 K.







CHAPTER 4
MAGNETIC STUDY OF EVOLVING STRUCTURE
(MONO, BI, AND MULTILAYER OF FILMS)

4.1 Synthesis of Ni-Fe(CN)6 Films

The generation of Langmuir-Blodgett (LB) Ni-Fe(CN)6 films [9 13] involves the

preparation of a chemical solution (LB solution) containing

bis(tetramethylammonium)pentacyano-(4-(didodecylamino)-pyridine)ferrate(III)-6H20,

here after written as amphiphilic pentacyanoferrate and nickel nitrate, Ni(H20)6(NO3)2.

As shown schematically in Figure 4-1, one unit of amphiphilic pentacyanoferrate consists

of a central Fe ion with five cyanide (CN) arms and one amphiphilic tail group in the

locations of octahedral vertices. This building unit floats on the water-based solution due

to the hydrophobic nature of the amphiphilic tail. When amphiphilic pentacyanoferrate is

introduced to the Ni(H20)6(NO3)2 solution, its four CN arms in the equatorial plane

replace the H20 from the Ni(H20)6(N03)2 and bond to the Ni ions. As a consequence, a

face centered square grid network of Fe-CN-Ni forms at the air-water interface and the

hydrophobic amphiphilic tails of the Fe ions point away from the water as shown in

Figure 4-1. The Fe-CN-Ni network is then transferred onto solid substrates, and

depending on the deposition cycles, monolayer, bilayer, and multiple bilayers

(multibilayer) specimens are generated. For the magnetic studies, a commercial Mylar

film was used as a substrate, and for the spectroscopic studies, single-crystal (100) silicon

wafers were used.









R = (CH2)11CH3


R R



N
I /C

NC- Fe -CN
c/ I
N c
N
unit of amphiphilic
pentacyanoferrate


\/R R /
R /R R R



i F Fe-
-Ni/ e/ .i/ I
I I I
two-dimensional
grid network


Figure 4-1. Amphiphilic pentacyanoferrate building unit and 2D two-dimensional grid
network. One unit of amphiphilic pentacyanoferrate consists of a central Fe
ion with five cyanide (CN) arms and one bis(tetramethylammonium), a tail
group. This building unit floats on the water-based solution due to the
hydrophobic nature of amphiphilic tail. When amphiphilic pentacyanoferrate
is introduced to the Ni(HzO)6(NO3)2 solution, its four CN arms in equatorial
plane replace the H20 from the Ni(H20)6(NO3)2 and bond to the Ni ions. As a
consequence, a face centered square grid network of Fe-CN-Ni forms at the
air-water interface and the hydrophobic amphiphilic tails point away from the
water surface.


Bilayer and multibilayer films were prepared using the Y type Langmuir-Blodgett

technique [13]. Briefly, one bilayer of the Fe-CN-Ni network was formed through a

cycle of immersing and withdrawing a hydrophobic substrate into/out of the

aforementioned LB solution. The multibilayer was produced by repeating the deposition

cycles for multiple times. In a slightly different way, the monolayer of Fe-CN-Ni was

generated by starting with the hydrophilic substrate immersed in the LB solution and

pulling it out of the solution. The structures of resultant films are shown schematically in









Figure 4-2. The results of X-ray diffraction measurements indicate that the distance

between each bilayer (inter-bilayer distance) in the multibilayer film is about 35A. The

structural coherence within the two-dimensional network was also probed using Grazing

Incidence X-ray Diffraction (GIXD), and the result yields an average of 3600 A2 coherent

area coverage in the multibilayer sample. In other words, within the two-dimensional

network, the crystallinity is preserved in the average area of 3600 A2. The GIXD result

also indicates that the unit cell edge distance (i.e. the Fe Fe spacing in the Fe-CN-Ni-

NC-Fe configuration) is about 10.2 A as shown in Figure 4-3. The distance and the

structure between the monolayers in each bilayer (intra-bilayer region) are not well

characterized at this time. The detailed synthesis procedures and characterizations can be

found in Reference [9 13].

4.2 DC Low Field Magnetization Measurements

The DC magnetization measurements of each film were performed using the

QUANTUM DESIGN Magnetic Property Measurement System (MPMS), a

superconducting quantum interference device magnetometer whose properties are

described in Chapter 2. About 10 cm2 of each film were cut into small squares and

packed into a commercial plastic can or gelatin capsule (gelcap) for the MPMS

measurement. For each sample, the magnetic measurements were performed in two

different film orientations with respect to the external magnetic field (HE). In one

orientation, the film was placed with its surface parallel (l|) to HE, and in the other

orientation the film was placed with its surface perpendicular (1) to HE. The background

magnetic signal arising from each commercial can or gelcap was measured beforehand










Ni --Fe- Ni Fe
monolayer





Ni Fe- Ni Fe

Ni FlNia Fe


bilayer


T
36 A


4,


Ni Fe

Ni Fe




Ni Fe


Ni F-- Ni l


multibilayer


Figure 4-2. Sketches of the monolayer, bilayer, and multibilayer structures viewed from
the side. Bilayer and multibilayer films were prepared using Y type
Langmuir-Blodgett technique. Briefly, one bilayer of Fe-CN-Ni network was
formed through a cycle of immersing and withdrawing a hydrophobic
substrate into/out of the LB solution. The multibilayer was produced by
repeating the immersing and withdrawing cycles for multiple times. The
multibilayer in the figure is a result after 2 cycles. In a slightly different way,
the monolayer of Fe-CN-Ni was generated by starting with the hydrophilic
substrate immersed in the LB solution and pulling it out of the solution.


and subtracted from the final result. The average magnetic background of the substrate

(Mylar) was also measured separately and subtracted from the final result. For each film

at each orientation, the temperature (T) dependencies of the zero-field-cooled (zfc) and

field-cooled (fc) magnetization was measured in HE = 100 G for the multibilayer sample

and in HE = 20 G for the monolayer and bilayer samples over the temperature range of

2 K to 15 K. The zfc measurements were performed, first by cooling the sample from








Ni Fe Ni Fe



A Fe Ni Fe Ni



10.2 A Ni Fe Ni Fe



I Fe Ni Ni


10.2 A



Figure 4-3. Schematic of Fe-CN-Ni square grid (top view). The Grazing Incidence X-ray
Diffraction (GIXD) result indicates that the unit cell edge distance (distance
between Fe ion and Fe ion in Fe-CN-Ni-NC-Fe configuration is about 10.2 A.
For clarity CN bonding is expressed as a line.



300 K to 2 K without any field (HE = zero), and then at T = 2 K, the external field was

applied and the magnetization of the sample was measured upon increasing temperature.

The fc measurement was performed in a similar way except HE was present during the

initial cooling of the sample from 15 K to 2 K.

For field dependent magnetization measurements, the samples were zero-field-

cooled to 2 K and then the magnetization was measured while sweeping HE from 50 kG

to -50 kG and back to 50 kG. In all of these measurements, the strong diamagnetic

background signals from the holder and Mylar dominated the relatively weaker magnetic

signals arising from the deposited films, especially at high temperatures or in the large







external fields. This strong background complicated the data analysis at high field and

temperatures. As a consequence, the absolute saturation value of the magnetization was

not well defined in each film.

Figure 4-4 shows temperature dependent zfc and fc magnetizations of multibilayer

film at two orientations (film II HE and film HE). A striking feature of the data is M (I

HE)
dimensional (2D) domain arrangements.6 Overall, the low temperature magnetizations

rise rapidly below T- 10 K and deviations between the zfc and fc magnetizations are

observable below 5 K. In addition, below 5 K the zfc values decrease while the fc values

slowly increase with decreasing temperature. The rapid increase in magnetization below

Tc 10 K is an indication of ferromagnetic interactions between the Fe11' (S = 1/2) ions

and Nin (S= 1) ions. Although the Tc value is lower in the film case by considerable

amount, the magnetic interaction is consistent with the three-dimensional bulk Ni-

Fe(CN6) (Tc ~ 24 K) studied by others[15]. Furthermore, the difference between the zfc

and fc values below T 5 K and the monotonous increase of the fc values are evidence of

spin-glass or cluster spin-glass-like behaviors. The detailed results of this glassy

behavior are presented in Section 4.3 when AC magnetic measurements are presented. It

should be mentioned that the difference between the fc and zfc data between 5 K and

10 K in Figure 4-4 seems to be an artifact, which was not reproduced in another similar

measurement.

The field dependent magnetizations of the multibilayer film at T = 2 K are plotted

in Figure 4-5. In both orientations of the film, there exist hysteresis loops with coercive


6 A different view of this anisotropy concerns anisotropic environments of single metal
ions and the full discussion can be found in Reference [13], page 97 and 98.












15 I v v v v
150-multibilayer

HE = 100G
O1 E
10 o fc (film IIH)
0 zfc (film II H)
;3 o fe (filmI ) H)
[ zfc (film I HE)
0 O1
5 -







0 5 10 15
T(K)

Figure 4-4. Temperature dependent magnetizations of 150-multibilayer film. The plot
shows temperature dependent zfc and fc magnetizations of 150-multibilayer
film at two orientations (film |I HE and film I HE) measured at HE = 100 G. A
striking feature of the data is M (I HE) from the demagnetizing field effect from the two-dimensional (2D) domain
arrangements. Overall, the low temperature magnetizations rise rapidly below
T 10 K and deviations between the zfc and fc magnetizations are observable
below 5 K. In addition, below 5 K the zfc values decrease while the fc values
slowly increase with decreasing temperature. The rapid increase in
magnetization below Tc 10 K is an indication of ferromagnetic interactions
between the FeIII (S = 1/2) ions and Ni"I (S = 1) ions.






















S
Q)
N
O


-2 [ v V 111111 11E



-3
-600 -400 -200 0 200 400 600

H(G)



Figure 4-5. The field dependent magnetizations of 150-multibilayer film at T = 2 K. In
both orientations of the film, there exist hysteresis loops with coercive fields
of 135 G (when film || HE) and -110 G (when film HE).


fields of-135 G (when film || HE) and -110 G (when film I HE). This hysteresis loop is

another indication of magnetic glassy behavior of the film. Consistent with the case of








temperature dependent magnetizations (Figure 4-4), the magnetization values in the

parallel orientation (film 1| HE) are higher than the values of perpendicular orientation.

This difference show demagnetizing effect caused by 2D arrangement of magnetic

domains. As a consequence, the ratio of measured magnetization values in the two

orientations will be different from the unity, and thus this ratio will provide an idea about

shape of domain arrangement. In Figure 4-4 and Figure 4-5, the magnetizations of the

multibilayer in the parallel orientation are higher than the magnetizations of the film at

perpendicular orientation. In particular, the value offc magnetization at T = 2 K when

film II HE is 3.5 times higher than the value when film I HE. This particular ratio of

parallel magnetization over perpendicular magnetization at 2 K will be referred as "shape

ratio" as it gives width to height information of the magnetic domains. For example if

the shape ratio is close to infinity, the height/width of domain is expected to be zero, or

equivalently the shape of domain is like 2D sheet.

The temperature dependent fc and zfc magnetizations of the bilayer film at two

orientations (film II HE and film L HE) are plotted in Figure 4-6. As it was the case of

multibilayer, that the resulting values, when film I HE, are smaller than the values when

film II HE, the "shape ratio" at 2 K in this case is -4.5, which is slightly higher than the

multibilayer case. The shapes of fc and zfc also look similar to those of multibilayer

result: increase below 10 K and difference between the zfc and fc values below 5 K. The

field dependent magnetizations ofbilayer film at T = 2 K are plotted in Figure 4-7. In

both orientations of the film, hysteresis loops exist with coercive fields of 75 G (when

film II HE) and 55 G (when film -HE). These hysteresis loop as was in the multibilayer

case, is an indication of magnetic glassy behavior of the film.
















bilayer


HE=20G


0,


o fc (film II HE)
* zfc(film | HE)
o fc (film I HE)

* zfc (film I HE)


~~mow


T (K)


Figure 4-6. Temperature dependent magnetizations ofbilayer film. The plot shows
temperature dependent zfc and fc magnetizations of bilayer film at two
orientations (film 11 HE and film I HE) measured. The magnetization values
when film I HE are smaller than the values when film |I HE. This anisotropic
effect is an indication of demagnetizing field effect arising from 2D like
magnetic domain arrangements.


a
3
0
s
In
0
'I


S a


______________


--1111111----






68








T=2K bilayer
2 -



1








-o- film || H
E
-2 -v- film I HE



-3
-600 -400 -200 0 200 400 600
H(G)

Figure 4-7. The field dependent magnetizations of bilayer film at T = 2 K. In both
orientations, there exist hysteresis loops with coercive fields of -75 G (when
film || HE) and -55 G (when film -HE). This hysteresis loop is an indication
of magnetic glassy behavior of the film. Consistent with the case of
temperature dependent magnetizations (Figure 4-6), the magnetization values
of parallel orientation (film || HE) are higher than the values of perpendicular
orientation.









As described previously, the magnetic properties of the bilayer film are

qualitatively similar to ones exhibited by the multibilayer film. Moreover, the

quantitatively normalized fc value in 150-multibilayer at 2 K in parallel orientation (film

|| HE) is about 46 times higher than that of bilayer at 2 K in the same orientation (film ||

HE). The fact that the synthesis of multibilayer film went through 150 cycles of

processing make the number 46 somewhat smaller than the expected 150; however,

considering non-linear behavior of the magnetization with respect to HE, the value 46 is

reasonable. Therefore, from the result of DC magnetizations, two films (multibilayer and

bilayer, see Figure 4-4 to Figure 4-7) are both qualitatively and quantitatively similar.

This result is not surprising when considering the synthesis protocols of two films.

According to the synthesis processes the multibilayer film is equivalent to 150 stacks of

bilayer, and each bilayer is separated by 35 A. Magnetically, this separation is

sufficiently long so that the strong long-range magnetic ordering interactions between the

bilayers might not be expected. Thus, it can be expected that multibilayer behaves

similar to 150 stacks of isolated bilayer films.

In contrast, the overall DC magnetization properties of the monolayer are

qualitatively different than the results from two other films, see Figure 4-8 and 4-9.

Unlike multibilayer and bilayer films, the magnetizations ofmonolayer films do not show

the difference between zfc and fc below 5 K but show a hint of difference at T 2 K. In

addition, there is relatively lower Tc 6 K, where the magnetization rises rapidly. Also,

the difference between magnetizations in the parallel (film || HE) and perpendicular

orientations is much larger ("shape ratio" is about 22.5) than the cases of the two other









films, namely the multibilayer and bilayer films. The field dependent magnetizations of


monolayer film at T


'Sn
0


2 K are plotted in Figure 4-9.


T(K)


Figure 4-8. Temperature dependent magnetizations of monolayer film. The plot shows
temperature dependent zfc and fc magnetizations of monolayer film at two
orientations (film || HE and film I HE). The magnetization values when film I
HE is much smaller than the values when film I| HE. This anisotropic effect is
an indication of demagnetizing field effect arising from 2D like domain
arrangements.


monolayer



HE= 20 G


o fc (film HE)

zfc (film|| HE)

In fc (film HE)

zfc(film HE)






l* I ( a I H
-~-~ o ~e49 0eE








71



3 -I i ---- i --- i ---- i I
monolayer
T=2K
2 -




-1 -^






-v- film IHE

-2


-3 ..
-100 -50 0 50 100
H (G)



Figure 4-9. The field dependent magnetizations of monolayer film at T = 2 K. The
coercive field of monolayer at 2 K is about 10 G, which is close to the
resolution of the instrument.








The coercive field ofmonolayer at 2 K is about 10 G, which is close to the resolution of

the instrument.

4.3 AC Field Magnetization Measurements

The DC magnetizations of the bilayer and the multibilayer in the previous section

showed some indications of spin-glass-like behavior such as different temperature

dependent behaviors of zfc and fc magnetizations below a certain temperature and

hysteresis loops. The spin-glass-like behavior can be explored more explicitly by AC

magnetization studies. Briefly, the magnetization in the glassy state is dependent upon

time, hence the time varying AC field probes the temporal nature of the magnetization

using periodic AC field at different frequencies. For the AC measurement, the same

QUANTUM DESIGN MPMS was used. All the samples were cooled to 2 K without any

DC or AC field and then 4 G of AC field was applied to measure the AC magnetizations

upon increasing temperatures.

Figure 4-10 shows the real and imaginary parts of AC susceptibilities of

multibilayer film. The film was oriented parallel to the direction to the AC field (film ||

HAC) and the different frequencies (17 Hz, 170 Hz, and 1 kHz) of AC field were used for

each measurement. Overall, the real parts of the AC susceptibilities increase rapidly

below T 8 K, and reach the maximum values at a temperature TF and the decrease at

lower temperatures. The magnitude of AC magnetizations below 8 K is the highest at

lowest frequency (17 Hz) and the lowest at the highest frequency (1 kHz). Also the

temperature where the maximum AC magnetization occurs (TF) depends on the

frequencies, such that TF is smaller for lower frequencies. The qualitative result of the

multibilayer film in the perpendicular orientation (not shown) is very similar to the result







73




5 1 1 1 .

%, 150-multilayer ( film |H )A

4 '
4 H =4G
A
_3 0~ o f= 17 Hz
SAD f= 170 Hz
0 A 8 f= 997 Hz


o 0
1 A
2 A -


0 .





0 5 10 15
T(K)



Figure 4-10. Temperature dependent AC susceptibilities of 150-multibilayer film.
Overall, the real parts of the AC susceptibilities increase rapidly below T-
8 K, and reach the maximum values at a temperature TF and decrease at lower
temperatures. The magnitude of AC magnetizations below ~ 8 K is highest at
lowest frequency (17 Hz) and the lowest at the highest frequency (1 kHz).
Also the temperature where the maximum AC magnetization occurs (TF)
depends on the frequencies, such that TF is smaller for lower frequency.
















I I I s l
Arrhenius law fitting


50-multibilayer
S- 3x1029 rad/s
?a/ 347 K


k


)o = o-exp[-E/k T ]

Lnw = (-Ea )/lT+Lnw



monolayer
c ~ 5x1021 rad/s -
E /kB 110K
bilayer L


-
C
I


- 5x102 rad/s
/kB- 174 K


0.3
1/TF (1/K)


Figure 4-11. Result of Arrhenius law fittings to monolayer, bilayer, and multibilayer.


E


0/3
rb
c^
-


0.2


0.4








of the film at parallel orientation. As was the case in the DC measurement of the same

film, the magnitude of AC magnetization is higher in parallel orientation than in

perpendicular orientation. The frequency dependent TF values of the multibilayer are a

unique indication of spin-glass-like behavior. An empirical constant (p can be defined as

a relative change of TF with respect to the change of corresponding frequencies,f such

that

p = (ATF/TF) / A(logj), (4.1)



wherefis a frequency [102]. Applying Eq. 4-1 to the real part of the AC measurement

result of multibilayer (film II HE) yields (p 0.04. The experimental values of the

empirical constant (p differ in the limiting cases of superparamagnets and spin-glasses.

The value p 0.04 falls into the regime of insulating spin-glasses [102]. Two other

empirical laws are often used to characterize spin-glasses and superparamagnets, namely,

Arrhenius (Eq. 4.2) and Vogel-Fulcher law (Eq. 4.3),



o = wo exp [-Ea/kBTF] (4.2)

and

w = coo exp [-Ea/kB(TF -T)]. (4.3)



The Arrhenius law is an empirical law that is used to characterize the thermal activation

energy (Ea) of a superparamagnet. As shown in Figure 4-11, when fitted to the Arrhenius

law (Eq. 4.2), the AC result ofmultibilayer film yields Ea/kB 347 K and oo 3 x 1029

rad/s. These values are unphysical considering the fact that oo ~ 3 x 1029 corresponds to









TeV in energy. Alternatively, the second law, Vogel-Fulcher law, describing interacting

superparamagnets or magnetic clusters, yields Ea/kB -52 K and To 3 K, which is a

measure of interaction between the clusters, when coo was fixed at 2n x 1013 rad/s [53],

see Figure 4-12. The values of Vogel-Fulcher fit make more physical sense and are

consistent with the multibilayer film being identified as an insulating spin-glass [102].


-0.036


-0.038


-0.040




-0.042


-0.044
5.


4


5.5


5.6
T (K)


5.7


5.8


Figure 4-12. Result of Vogel-Fulcher law fitting on multibilayer film.


Vogel-Fulcher law fitting

0 )= moexp[-Ea/kB(TF-To)]







150-multilayer
co = 27x x1013 rad/s (fixed)
E /k- 57 K
o= 3.3 K

- I l I l *








On the other hand, given that the fc magnetization increases monotonically at lower

temperatures, the multibilayer may be classified as cluster-spin-glass [53].


5.0


4.0 1


3.0


S
a>
"W
(9


2.0


1.0



0.0


T(K)


Figure 4-13. Temperature dependent AC susceptibilities of monolayer film. The film
was oriented in parallel direction to the AC field (film 11 HAC) and the different
frequencies (1 Hz, 17 Hz, 170Hz, and 1 kHz) of AC field were used for each
measurement. Overall, real parts of AC susceptibilities increase rapidly below
T = 7 K, then reach the peaks and returns The magnitude of AC
magnetizations below 7 K is highest at lowest frequency (1 Hz) and the
lowest at the highest frequency (1 kHz).


" I I I I I I S I


monolayer ( film HA)

Ha=4G
0
AC


A filHz
0


A f= 1 Hz
A O
So f = 17 Hz
a f = 170 Hz
Sr [ f = 997 Hz

[A
A
I0


DA A A
O








As shown earlier in the Section 4-2, the DC magnetic properties of monolayer

sample were qualitatively different than the behavior exhibited by the other two films.

The same observation was made for the results of the AC magnetization measurements.

Figure 4-13 shows AC magnetization measurement of monolayer in the parallel

orientation (film II HAC). The AC signals of the monolayer film in the perpendicular

orientation (not shown here) were too small to be resolved in the magnetometer. The

small or undetectable AC signals in perpendicular orientation are consistent with a high

"shape ratio", the result of DC magnetization of the monolayer film in Section 4.2. It is

striking to see that the absolute peak AC magnitude of the monolayer is almost 500 times

smaller than those of multibilayer in parallel orientation. Moreover, the TF values of the

monolayer are relatively smaller than those of the multibilayer. The overall lower TF

values in monolayer film can be interpreted as a lack of a cluster-glass like state. In other

words, if we correlate the TF as a measure of interactions between the clusters and the

size of the clusters, then the size and interactions of clusters in the monolayer are less

pronounced than those of the multibilayer since the magnetic pathways in the monolayer

are confined to the plane. The empirical constant p, (Eq. 4.1), for the monolayer case is

about 0.05, which is slightly higher than (p = 0.04 of the multibilayer, and closer to

superparamagnet limit [102]. Due to the scatter of the data, only two data points were

used to fit the monolayer data to Arrhenius law as shown in Fig 4-11. The fit yields Ea~

110 K, which is smaller than the case of multilayer and reflects the qualitative difference

between two films.

Interestingly the shape of AC magnetization of bilayer sample resembles the data

of the both monolayer and mutibilayer. Figure 4-14 shows real and imaginary parts of








AC susceptibilities ofbilayer film. The film was oriented in parallel direction to the AC

field (film I| HAC) and the different frequencies (1 Hz, 17 Hz, 170Hz, and 1 kHz) of AC


1--
0
4



2



0


OAL


o


a
0


*4t.


bilayer ( film | HA) .
HA=4G
AC


1 Hz
17Hz
170 Hz
997 Hz


2 2 *n' t*1


A M A


T (K)


Figure 4-14. Temperature dependent AC susceptibilities ofbilayer film. The film was
oriented in parallel direction to the AC field (film II HAC) and the different
frequencies (1 Hz, 17 Hz, 170Hz, and 1 kHz) of AC field were used for each
measurement. Overall, real parts of AC susceptibilities increase rapidly below
T = 8 K, then before they reach the T- 4 K peaks they slow down the
increasing process around T = ~ 6 K. It seems that bilayer shows peak of
monolayer at ~ 4 K and peak of 150-mutibilayer at ~ 6 K as compared in
Figure 4-15.


tm


Sf=

o f=
f=
o f=


-1--1-1-1-1-1








80






6 -i- -- i-
x 3 ) (x 150)
(x 300) ^

film |HAc
SAC 4
4 -
A O f= 17 Hz
0 O
O 0 bilayer
S 0 A monolayer
S o 150-multibilayer
2 O0


(x 1)....,,
0-




2 4 6 8 10
T(K)
Figure 4-15. Comparison of AC data at 17 Hz (mono, bilayer, and multibilayer).




field were used for each measurement. Overall, the real part of AC susceptibility
increases rapidly below T 8 K and peaks at T 4 K, with a shoulder T 6 K. The








shoulder corresponds to the peak of the 150-multibilayer film at 6 K as compared in

Figure 4-15. Using the peaks at around 4 K, Arrhenius law and empirical constant (p can

be evaluated. The results are similar to the case ofmonolayer and yielded (p 0.05 and

Ea/kB ~ 174 K as shown in the Figure 4-11.

4.4 Magnetic Evolution versus Structural Evolution

In the previous sections, the experimental results indicated that the magnetic

behavior of the films differ from each other in both quantitative and qualitative aspects.

These differences are expected to be related to the systematic differences of the film

structures. Consequently, this section focuses on the correlations between the magnetic

and structural evolution of the films. The experimental results of the characteristic DC

magnetization values (Tc, He, and shape ratio) are summarized in Table 4-1. Considering

only parallel (film I HE) cases, the Tc values increase as films evolve from monolayer

(5.5 K) to multibilayer (9.9 K). The increment of Tc value between monolayer to bilayer

(5.5 K to 8.9 K) is larger than the increments of Tc between bilayer and multilayer (8.9 K

to 9.9 K). According to a mean field approximation first described by Langevin, Weiss,

and N6el, the Tc values in dinuclear structures (such as our Fe-CN-Ni systems) can be

expressed as


Tc = T7 MM' I (SM + 1) SM (S + 1)
3k (4.5)

where, ZM is the number of the nearest magnetic atoms that interact with a central

magnetic atom M, ZM' is the number of the nearest magnetic atoms that interact with a

central magnetic atom M', JMM' is the exchange coupling constant between the two

nearest metal M and M' ions, SM and SM' are the spin values of atoms M and M',







respectively, and kB is the Boltzmann constant. Defining Z as an effective number of

nearest magnetic neighbors such that ZM' ZM = Z2, and assuming that Z = 5, the estimated

JMM' value using Tc from the experimental result is JMM ~ 5 K in the multibilayer case.

The value of J~m' is expected to be approximately the same in all three films since the

immediate local environments for all magnetic ions in all three films are the same, i.e. the

(NC)4-M-CN-M'-(NC)4 environment dominates the structures. Using the value of JMM'

~ 5 K and Tc from Table 4-1, the Z value can be estimated as Z = 4.4 and Z = 2.7 for

bilayer and monolayer samples respectively. A noticeable increase in the Z value from


Table 4-1. Characteristic values from DC magnetization measurements.7
samples orientation Tc (K) He (G) shape ratio
|| 9.9 135
multibilayer 10.5 110 22.5
1 10.5 110
|| 8.9 75
bilayer 8.5 4.5
1 8.4 55
| __5.5 10
monolayer 0/A 3.5
S 1 6.1 N/A


Table 4-2. Characteristic values from AC magnetization measurements.
samples frequency (Hz) TF (K)

17 5.4
multibilayer 170 5.7 0.04 3 x 1029 347
997 5.8
1 3.7
17 3.9
bilayer 170 N/A 0.05 5 x 1021 174
170 N/A
997 N/A
1 2.3
17 2.4 21
monolayer 17 2.4 0.05 5 x 102 110
170 N/A
997 N/A


7 The "shape ratio" is defined as M (I ) /M (1) at 2 K.









the monolayer to the bilayer suggests that there are more magnetic interactions in bilayer,

and this can arise from the intra-bilayer magnetic interactions. As mentioned earlier in

Section 4.1, this intra-bilayer region was not chemically well characterized; however

from the fact that bilayer has more magnetic interactions than monolayer, we conclude

that the structure of intra-bilayer region is such that it provides an additional magnetic

interaction between monolayers in bilayer (intra-bilayer interaction).

Structurally, the multibilayer is a stack of bilayers separated by 35 A of long

amphiphilic tails. As a consequence, the DC behavior of the bilayer is similar to the

multilayer. However the magnetic properties of the two samples are not the same, and

this fact is especially evident in the AC measurements, see Table 4-2. Microscopically,

the bilayer sample, whose top surface is exposed to air is expected to possess different

behavior than a bilayer sandwiched between other bilayers (as in multibilayer). This

possible surface effect of the bilayer might be structurally and magnetically significant to

differentiate the bilayer film from the multibilayer ones. In addition, dipole interactions

between the bilayers (inter-bilayer interaction) might play a role to differentiate the

multibilayer from the bilayer film. To investigate the possible role of the dipole

interaction between the bilayers, we consider the z-component of the dipole field strength

of a magnetically ordered disk (radius, R -30 A) as a function of height at the center of

the disk along the z-axis, see Figure 4-16. This situation mimics the dipole field

generated by a bilayer in a multilayer sample, assuming that the disk represents coherent

portion of the bilayer (for simplicity bilayer was treated as a single disk). For the result

of the calculation plotted in Figure 4-16, the effective surface magnetization (total







moments per cm2) of the disk was estimated by considering the structure of the face

centered Fe-CN-Ni square network. The estimated value was 1.4 x 10-5 emu/cm2.


200



150



100



50



0


5 10 15


20


Height (nm)


Figure 4-16. Dipole field produced by a magnetized disk (R-30 A) evaluated at different
height along the z centered axis away from the disk. The situation mimics the
dipole field generated by a bilayer in the multilayer sample assuming that the
disk represents a coherent portion of bilayer (for simplicity bilayer was treated
as a single disk). The effective surface magnetization (moments per cm2) of
the disk was estimated by considering the structure of face centered Fe-CN-Ni
square network. The estimated value was 1.4 x 10-5 emu/cm2.








An effective volume magnetization (total moments per cm3) was also estimated to be

~ 140 emu/cm3. To get a practical sense of this value, the element Iron (Fe) (Tc 1000

K) has a magnetization of- 1700 emu/cm3, which is an order of magnitude larger. From

the result of calculation, the z component of dipole field produced by a coherent portion

of the bilayer at the distance of 36 A, perpendicular from the center of the disk, is about

55 G. Generally speaking, this field strength produces a magnetic interaction energy of

10 mK, base on the approximation of peH kBT but for ferromagnetically ordered

molecules (or crystals) having larger magnetic moments, the effect is stronger and can be

an order of 10 K in a multibilayer sample. The detailed dipole field distribution is not

calculated here because it requires the knowledge of the exact distribution of the

magnetic domains and detailed information on structure.

Based on the discussion on this section and referring to the experimental results in

the previous sections, a description of the magnetic processes of all samples can be made.

When the temperature is well above Tc of the multibilayer (9.9 K) all the films behave as

paramagnets that follow a Curie law. When temperature equals Tc of multibilayer or

bilayer, the magnetic interactions within the bilayer exceed the thermal fluctuations of the

systems, and the films orders ferromagnetically. In the mean time, the dipole field

between the bilayers in the multilayer sample develops and provides an additional

magnetic pathway between the bilayers. In addition, the ferromagnetic clusters start to

interact each other, and these interactions give a rise to cluster-spin glass behavior that

exhibits frequency dependent AC susceptibilities. Still, the magnetic interactions in the

monolayer sample are not well developed due to the lack of magnetic pathways. Upon

further cooling, the monolayer starts to develop ferromagnetic clusters around 5 K and





























86


the interaction between clusters become sufficiently strong enough to show cluster-spin

glass behavior around 2 K.