Spin polarization measurements and sensor applications in thin films and carbon nanotube-based devices

Spin polarization measurements and sensor applications in thin films and carbon nanotube-based devices

Material Information

Spin polarization measurements and sensor applications in thin films and carbon nanotube-based devices
Sanders, Jeff T
Place of Publication:
[Tampa, Fla]
University of South Florida
Publication Date:


Subjects / Keywords:
Andreev reflection
Spin transport
Dissertations, Academic -- Physics -- Doctoral -- USF
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )


ABSTRACT: The unique properties of carbon nanotubes (CNTs) show a great deal of potential for nanoelectronic devices, spintronic devices, biosensing and chemical sensing applications. Their applicability as interconnects for spintronic devices derives from their one-dimensionality and theoretically predicted preservation of spin current. In this work, we combine an investigation of spin polarization in materials such as half metallic oxides in thin film and bulk form with studies on several aspects of CNTs for sensing and spin transport applications. These two areas of study are intimately related within the umbrella of spin-electronics and nanoscale sensors that are being pursued with great topical interest in recent times. A measurement system has been developed to perform Point-Contact Andreev Reflection (PCAR) in the presence of variable magnetic fields and temperatures. It was designed and built, accepted for patent by the USF, and submitted to the U.S. Patent Office. A study ^of spin polarization in superconductor-magnet junctions has been performed over a wide range in magnetic fields (0 to 3T) and temperature (2 to 300K)on several systems including copper, strontium ruthenate, and chromium dioxide. Spin transport experiments have been extended to single walled carbon nanotube (SWNT) networks inorder to explore spin transport in nanotube networks for potential sensor applications.Carbon nanotube networks have been used as the electronic material for chemical and biological sensing where capacitance and conductance response to the adsorbtion of a chemical or biological analyte are simultaneously measured and a very fast response and recovery is observed. Chemical specificity has been investigated through different means since a goal of the U.S. Navy is to have an array of these sensors, each chemically specific to a unique analyte. Finally, research is ongoing in the analysis of our PCAR spectra in the strontium ruthenate series and the lanthinum strontiu m manganite series to investigate the square root dependence of the background conductance data and the fundamental aspects of the fitting procedure by using a chi-square statistical model to more accurately determine the spin polarization, P.
Dissertation (Ph.D.)--University of South Florida, 2006.
Includes bibliographical references.
System Details:
System requirements: World Wide Web browser and PDF reader.
System Details:
Mode of access: World Wide Web.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains 202 pages.
General Note:
Includes vita.
Statement of Responsibility:
by Jeff T. Sanders.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
001910779 ( ALEPH )
173517096 ( OCLC )
E14-SFE0001711 ( USFLDC DOI )
e14.1711 ( USFLDC Handle )

Postcard Information



This item has the following downloads:

Full Text
xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam Ka
controlfield tag 001 001910779
003 fts
005 20071001125101.0
006 m||||e|||d||||||||
007 cr mnu|||uuuuu
008 071001s2006 flu sbm 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0001711
1 100
Sanders, Jeff T.
0 245
Spin polarization measurements and sensor applications in thin films and carbon nanotube-based devices
h [electronic resource] /
by Jeff T. Sanders.
[Tampa, Fla] :
b University of South Florida,
3 520
ABSTRACT: The unique properties of carbon nanotubes (CNTs) show a great deal of potential for nanoelectronic devices, spintronic devices, biosensing and chemical sensing applications. Their applicability as interconnects for spintronic devices derives from their one-dimensionality and theoretically predicted preservation of spin current. In this work, we combine an investigation of spin polarization in materials such as half metallic oxides in thin film and bulk form with studies on several aspects of CNTs for sensing and spin transport applications. These two areas of study are intimately related within the umbrella of spin-electronics and nanoscale sensors that are being pursued with great topical interest in recent times. A measurement system has been developed to perform Point-Contact Andreev Reflection (PCAR) in the presence of variable magnetic fields and temperatures. It was designed and built, accepted for patent by the USF, and submitted to the U.S. Patent Office. A study ^of spin polarization in superconductor-magnet junctions has been performed over a wide range in magnetic fields (0 to 3T) and temperature (2 to 300K)on several systems including copper, strontium ruthenate, and chromium dioxide. Spin transport experiments have been extended to single walled carbon nanotube (SWNT) networks inorder to explore spin transport in nanotube networks for potential sensor applications.Carbon nanotube networks have been used as the electronic material for chemical and biological sensing where capacitance and conductance response to the adsorbtion of a chemical or biological analyte are simultaneously measured and a very fast response and recovery is observed. Chemical specificity has been investigated through different means since a goal of the U.S. Navy is to have an array of these sensors, each chemically specific to a unique analyte. Finally, research is ongoing in the analysis of our PCAR spectra in the strontium ruthenate series and the lanthinum strontiu m manganite series to investigate the square root dependence of the background conductance data and the fundamental aspects of the fitting procedure by using a chi-square statistical model to more accurately determine the spin polarization, P.
Dissertation (Ph.D.)--University of South Florida, 2006.
Includes bibliographical references.
Text (Electronic dissertation) in PDF format.
System requirements: World Wide Web browser and PDF reader.
Mode of access: World Wide Web.
Title from PDF of title page.
Document formatted into pages; contains 202 pages.
Includes vita.
Adviser: Hariharan Srikanth, Ph.D.
Andreev reflection.
Spin transport.
Dissertations, Academic
x Physics
t USF Electronic Theses and Dissertations.
4 856
u http://digital.lib.usf.edu/?e14.1711


Spin Polarization Measurements and Sensor Applications in Thin Films and Carbon Nanotube-based Devices by Jeff T. Sanders A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy Department of Physics College of Arts and Sciences University of South Florida Major Professor: Hari haran Srikanth, Ph.D. Garrett Matthews, Ph.D. Rudy Schlaf, Ph.D. Lilia Woods, Ph.D. Date of Approval: July 14, 2006 Keywords: Andreev reflection, half-metals, spin transport, spintronics, biosensors Copyright 2006, Jeff Sanders


Dedication This dissertation is dedicated to th e late Randy Ertenberg for his friendship, comaraderie, and motivation throughout the f our years we spent t ogether. His courage during a long battle with Leukemia will be an inspiration forever.


Acknowledgments I would like to thank my advisor Dr Hariharan Srikanth for discussions, encouragement, and his wealth of physics knowledge. Thanks to my committee members Dr. Rudy Schlaf, Dr. Lilia Woods, and Dr. Ga rrett Matthews and to Dr. Carol Williams, the chair of my dissertation defense. I also want to acknowledge my mentors at the U.S. Naval Research Lab in Washi ngton D.C.: Dr. Robert Soulen Dr. Mike Osofsky, Dr. Eric Snow, Dr. Keith Perkins, and Dr. Paul Campbell. I owe a debt of gratitude regarding the technical aspects of this proj ect and would like to thank Sa m Valente for his hard work and extreme skill in the USF Physics mach ine shop, Joe Sanders for his mechanical engineering and CAD expertis e, and Mark Lefevre for hi s electronics knowledge and assistance. I would also like to especi ally thank Dr. Gerald Woods for his mentorship during my first NRL ONR internship and his assist ance, expertise, and friendship throughout the past four years here at USF. Special r ecognition also to Bernar d Batson, whose tireless work ethic with the NSF IGERT fellowship pr ogram here at USF is much appreciated. Thank you to Ranko Heindl for his friendshi p throughout our Ph.D. programs together and his assistance with the LabVIEW data acq uisition programming fo r this work, and I appreciate the intrinsic cu riosity and enthusiasm that James Gass shows for physics discussions and all aspects of life. Krystal McCann gave tirelessly of her time and skill in assisting with ed iting, word processing and genera l support and I canÂ’t thank her enough.


i Table of Contents List of Tables iii List of Figures iv Abstract vii Chapter 1 – Introduction and Motivation 1 1.1 Introduction 1 1.2 Motivation and rese arch plan 5 Chapter 2 – Background and Fundamental Physics 10 2.1 Magnetic Oxides 10 2.1.1 Device applications 10 2.1.2 Future prospects for spintronic devices 13 2.2 Chromium Dioxide CrO2 15 2.2.1 Crystal structure 15 2.2.2 Energy band calculations and density of states (DOS) 18 2.3 Thin film growth and magnetism of CrO2 21 2.3.1 Production and depositio n techniques 21 2.3.2 Magnetic properties 22 2.3.3 Transverse susceptibility data 24 2.4 Superconductivity and Andreev Reflection 27 2.4.1 Superconductivity 27 2.4.2 Measurement of spin polarization using PCAR 36 Chapter 3 – Point-contact experime ntal results and discussion 43 3.1 Point-Contact Andreev Reflection (PCAR) data 43 3.2 Superconductor-Normal Metal (SC-NM) junction 44 3.3 Superconductor-Half Metal (SC-HM) junction 50 3.3.1 Sn-CrO2 junction 51 3.3.2 Temperature dependence 55 3.3.3 Magnetic Field dependence 58 3.4 Blonder-Tinkham-Klapwik (BTK) m odeling of conductance curves 62 3.4.1 BTK theory and modified BTK model 62 3.4.2 Modified BTK fit to conductance curves 65 3.5 Ferromagnetic Metal – Superconductor (FM-SC) junction 66 3.5.1 Ni-MgB2 junction 66 Chapter 4 – PCAR on Correlated Oxides SrRuO3 and La1-x(Ba, Sr)xMnO3 70 4.1 Strontium Ruthenate SrRuO3 70


ii 4.1.1 Properties of SrRuO3 71 4.1.2 PCAR measurements on SrRu0.8Ti0.2O3 72 4.1.3 PCAR measurements on SrRu0.92O3 78 4.1.4 SrRu0.9Mn0.1O3 and SrRu0.9Cr0.1O3 80 4.2 Lanthinum Manganate series: La1-x(Ca, Ba, Sr)xMnO3 87 4.2.1 Properties of La1-x(Ca, Ba, Sr)xMnO3 88 4.2.2 PCAR measurements of La1-x(Ca, Ba, Sr)xMnO3 92 Chapter 5 – Carbon nanotube grow th and applications 99 5.1 Carbon nanotube growth 99 5.2 Carbon nanotube–based chemical sensors 107 5.2.1 Background and motivation for CN T capacitive sensors 107 5.2.2 Sensor fabrication 108 5.2.3 Capacitive response C/C to chemical analytes 112 5.2.4 Chemical specificity for analyte identification 119 5.3 Carbon nanotube biosensors 124 5.3.1 pH testing and DNA functionalization 127 5.3.2 Creatinine sensing 131 Chapter 6 – Spin transport in carbon nanotube devices 133 6.1 Background theory and magne toresistance experiments 133 6.2 Fabrication of SC/CNN/FM samples and measurement 136 6.3 Fabrication of SC/CNN/SC samples and measurement 141 Chapter 7 – Ongoing analyses and future directions 145 7.1 PCAR analysis 145 7.1.1 Investigation of V 1/2 background conductance 145 7.1.2 Statistical 2 analysis of spin polarization P 157 7.2 Instrumentation improvements and future experiments 166 7.2.1 Sample stage improvements 166 7.2.2 Incorporation of lock-in amplifier 167 7.2.3 Future spin transport experiments 167 References 169 Bibliography 178 Appendices 179 Appendix A: CAD of rotating base for T experiments 180 Appendix B: Instumentation of PCAR probe 182 Appendix C: USF carbon nanotube CVD growth furnace 200 Appendix D: Publications & Presentations List 201 About the Author End Page


iiiList of Tables Table 1 Summary of experimental Andreev reflection resu lts to determine 42 spin-polarization P for various materials Table 2 Capacitance response to various chemical vapors 114 Table 3 Ratio of C/ G showing intrinsic specificity for chemical analytes 123 Table 4 Basic material parameters of the SrRuO3 series of samples 156


iv List of Figures Figure 1 Some comm on magnetoelectronic device schematics 12 Figure 2 The Ju liere plot of tunneling magnetoresistance R/R vs. spin 14 polarization P. Figure 3 Illust ration of the rutile structure of CrO2 16 Figure 4 Rutile structure for CrO2 17 Figure 5 Energy band stru cture and energy density of states (DOS) 19 for CrO2 Figure 6 Density of st ates for majority and minority spins in CrO2 20 Figure 7 M-H Hysteresis loops for CrO2 23 Figure 8 T vs. H for CrO2 thin film at 0o and 90o to easy axis 25 Figure 9 Angular dependence of T for CrO2 thin film 26 Figure 10 Density of states diagrams for superconduc tor-normal metal (S C-NM) and a superconductor-half metal (SC-HM) junctions 39 Figure 11 Normalized conductance vs. bias voltage for a series of poi nt-contact experiments.curves for several systems 40 Figure 12 Current vs. bias voltage for a Sn tip and a Cu foil sample 45 Figure 13 Normalized conduc tance vs. bias voltage for Sn-Cu junction 46 Figure 14 Field dependence of G(V)/GN vs. bias voltage for Sn-Cu 48 junction


v Figure 15 Field dependence of G(V)/GN vs. bias voltage for Nb-Cu 49 junction Figure 16 Current vs. bias voltage for Sn-CrO2 junction 52 Figure 17 Normalized conductance vs. bias voltage for SnCrO2 53 junction Figure 18 Temperature dependence data for Sn-CrO2 junction 57 Figure 19 Field dependence data for Sn-CrO2 junction 59 Figure 20 Critical field Hc(T) versus temperature for several 60 superconductors Figure 21 BTK model fit to experimental Sn-CrO2 data 67 Figure 22 Conductance vs. Bias Voltage for Ni-MgB2 junction 69 Figure 23 Resistivity vs. T for SrRuO3, SrRu0.92O3, and SrRu0.8Ti0.2O3 73 Figure 24 PCAR spectra for a Sn-SrRu0.8Ti0.2O3 junction 74 Figure 25 An analyzed PCAR spectrum of Sn-SrRu0.8Ti0.2O3 at zero field 76 Figure 26 PCAR spectra for a SnSrRu0.92O3 junction 79 Figure 27 G(V) vs. V for SrRu0.9Mn0.1O3 as a function of applied field, Ha 82 Figure 28 G(V) vs. V for SrRu0.9Mn0.1O3 as a function of Temperature, T 84 Figure 29 PCAR spectra for SrRu0.9Mn0.1O3 and SrRu0.9Cr0.1O3 86 Figure 30 Resistivity vs. Te mperature and Magnetizati on vs. Temperature 91 Figure 31 PCAR spectra for LBMO (disordered) 93 Figure 32 PCAR spectra for LCMO as a function of temperature 94 Figure 33 Best fits for the G(V) spectra of LBMO and LSMO 95 Figure 34 PECVD-grown carbon nanotubes 100 Figure 35 AFM and SE M images of carbon nanotube networks 102


vi Figure 36 SEM images of SWNTs grown using CVD 104 Figure 37 Water assisted CVD growth of carbon nanotubes 106 Figure 38 Networks of SWNTs for el ectronic materials applications 109 Figure 39 Schematic of capaci tive CNT sensor, circuit schematic 110 Figure 40 Capacitance response C/C for acetone and DMF 113 Figure 41 Capacitance response C/C vs. molecular dipole moment, 116 Figure 42 CAD schematic of CNT devices, Adsorbion on SWNTs 118 Figure 43 Sensor response for se lf-assembled monolayer (SAM) and 120 chemically selective polymer HC, showing selectivity Figure 44 G and C responses and G/ C for eleven different 122 analytes Figure 45 CAD drawing of CNN biosensor showing the active area 125 and electrodes for acquisition of C and G data Figure 46 Microfluicic cell containing CNN biosensor, and syringe 126 pump Figure 47 Biosensor response vs. time for pH testing 128 Figure 48 C(V) curves for DNA functi onalization on CNN biosensor 130 Figure 49 Creatinine molecule 132 Figure 50 Co/MWNT/Co de vice, Magnetoresistance for parallel and antiparallel states of magnetization. 135 Figure 51 HM/CNN/SC concept for measur ing spin transport in CNTs 137 Figure 52 (a) Optical imag e of Co/CNN/Sn device (b ) SEM image of 139 the CNN in the gap between the Sn thin film and Co thin film Figure 53 I-V data and dI /dV data on Co/CNN/Sn samples 140 Figure 54 Sn/CNN/Sn sa mple fabricated with UV litho and e-beam 143


vii Figure 55 (a) (T) in zero magnetic field. (b) in zero B-field 147 Figure 56 PCAR spectra of a Sn-SrRu0.9Mn0.1O3 as a function of T 149 Figure 57 G(V) for a Sn-SrRu0.9Mn0.1O3 junction at T=2.0 K and H=0 151 Figure 58 Sn-SrRu0.9Mn0.1O3 PCAR spectra as a function of Ha 153 Figure 59 Derived normalization curves for Sn-SrRu0.9Mn0.1O3 155 Figure 60 Degeneracy in the f our-parameter PCAR fitting routine 158 Figure 61 2 vs. P and two simulated PCAR spectra for Cu 160 Figure 62 2 vs. P at three different normalization values 162 Figure 63 2 vs. P for experiemental LBMO data 165 Figure A1 CAD orthogona l projections of our rotating base for T 180 Figure A2 Three dimens ional CAD drawing of rotating base 181 Figure A3 Schematic of in itial electronics for PCAR experiments 183 Figure A4 A photograph of the PCAR probe, inserted into the PPMS 184 Figure A5 Point Cont act Andreev Reflecti on cryostat probe 186 Figure A6 Lower copper housing of PCAR probe 189 Figure A7 Exploded view of the lower Cu housing of the PCAR probe 190 Figure A8 Upper aluminum housing 192 Figure A9 Concept of m odifications for coarse/fine translation 194 Figure A10 Mech. drawings of new shaft on upper aluminum housing 195 Figure A11 Mech. drawings of knurled Al knob for coarse translation 196 Figure A14 CAD of modified lower Cu housing and new Cu sliding part 197 Figure A15 Overall, cuta way, and exploded views of modification 198


viii Spin Polarization Measurements and Sensor Applications in Thin Films and Carbon Nanotube-based Devices Jeff T. Sanders ABSTRACT The unique properties of carbon nanot ubes (CNTs) show a great deal of potential for nanoelectronic devices, spin tronic devices, biosensing and chemical sensing applications. Their applicability as interconnects for spintr onic devices derives from their one-dimensionality and theoretically predicted preservation of spin current. In this work, we combine an investigation of spin polarization in materials such as halfmetallic oxides in thin film and bulk form w ith studies on several aspects of CNTs for sensing and spin transport appl ications. These two areas of study are intimately related within the umbrella of spin-electronics a nd nanoscale sensors that are being pursued with great topical intere st in recent times. A measurement system has been devel oped to perform Poin t-Contact Andreev Reflection (PCAR) in the presence of variable magnetic fields and te mperatures. It was designed and built, accepted for patent by th e USF, and submitted to the U.S. Patent Office. A study of spin polarization in superconductor-magnet junctions has been performed over a wide range in magnetic fields (0 to 3T) and temperature (2 to 300K) on several systems including Cu, SrRuO3, LaSrMnO3, and CrO2. Spin transport experiments have been extended to single wa lled carbon nanotube (SWNT) networks in order to explore spin transport in nanotube networks for potential sensor applications.


ix Carbon nanotube networks have been used as the electronic material for chemical and biological sensing where capacitance a nd conductance response to the adsorbtion of a chemical or biological analyte are simultaneously measured and a very fast response and recovery is observed. Chemical specific ity has been investig ated through different means since a goal of the U.S. Navy is to have an array of these sensors, each chemically specific to a unique analyte. Finally, research is ongoing in the analys is of our PCAR spectra in the SrRuO3 series and the La1-x(Ca, Ba, Sr)xMnO3 to investigate the square root dependence of the background conductance data and the fundament al aspects of the fitting procedure by using a 2 statistical model to more accurately determine the spin polarization, P.


1 Chapter 1 Introduction and Motivation 1.1 Introduction Magnetic oxide thin films like CrO2, Fe3O4, etc. are currently used in applications like media for magnetic reco rding. These materials show unique magneto-transport properties that also ha ve potential applicat ions in spintronic devices. There is much interest in using the magnetic properties and spin polarization of materials to engineer recording media and devices with higher memory density, faster switching speeds greater nonvolatility and lower power consumption [1]. There is also intere st in producing more efficient devices by coherent manipulation of the spin of the electron as well as its charge. Proposed devices include spin valves, spin-FET (fie ld effect transistor), spin-LED (lightemitting diode), spin RTD (res onant tunneling device), etc. The unique properties of carbon nanotube s lead to extensive potential [2] for nanoelectronic devices, memory storag e applications, biosensing and chemical sensing for warfare agents, antiterrorism and national security. Their potential as interconnects in spintronic devices has motivated this study of the fusion of spin transport studies with experiments on car bon nanotube-based sensors and devices. Single walled carbon nanotubes (SWNT) are th e most interesting because of their


2 small diameters (~ 1 nm) and lengths up to a few millimeters [3] leading to extremely high aspect ratios, good electr ical conductivity and high mechanical strength. The Department of Defe nse (DOD) and its funding agency, the Defense Advanced Research Projects Ag ency (DARPA) have also earmarked funds for scientific and technological adva nce of so-called spintronic devices. The 1999 SPINS initiative (SPins IN Semiconductors) spearheaded a multi-billion dollar effort toward a 10-year goal of deve loping devices such as spin-transistors, spin-diodes, spin-LEDs in order to revol utionize electronic devi ces, especially for the U.S. Navy and other military branches. Spintronic devices have been predicted to reduce power consumption, in crease switching speeds, and increase memory density, among other qualities [1]. This work has been derived from these two initiatives: studies of thin-f ilm and bulk materials with potential for spintronics applications, and studies of applicatio ns of carbon nanotubes for sensing applications and spintr onic interconnect applications. In this work, we have conduc ted a comprehensive study of the magnetic properties and spin polarization in CrO2 thin films grown by chemical vapor deposition (CVD) technique [4]. We have also done some preliminary studies on Fe3O4 deposited by laser ablation [5]. But we have not included most results in this thesis as further work is needed in that case. A ferromagnetic oxide materi al can be used as a memory storage material as part of a magnetic multila yer to store information in computer hard drives. The


3 magnetizations of domains parallel and antiparallel to an applied external magnetic field are used as dist inct states for binary code. Another interesting aspect of a storage material is to exploit its anisotropy in magnetism. A material that has a fa irly large switching field in one crystal orientation and a smaller switching field in another crystal orientation can be tailored to meet desired specifications. Computer e ngineers, for example, are interested in tailoring a material to ma tch the specific magnetic field needed for the read/write functions of a transducer head. This anisotropic behavior has been seen in CrO2 thin films and has been verified by us using a novel radio frequency (RF) tunnel-diode oscillator (TDO) expe riment [6] developed by our group and then mounted on a home-made rotating base (see appendix A). The data are consistent with large in-plane anisotropy. These features are highlighted in our RF experiments where angul ar anisotropy measurements indicate that our data can be described as a combination of contributions from longitudinal and transverse susceptibility. Previous re search in our laboratory at USF on thin films, nanoparticles [7], and bulk sa mples has included measurements of magnetization vs. applied field, magnetization vs. temperature, magnetoresistance, resisitivity vs. temper ature and other standard experiments. There is current interest in us ing the spin of the electron as well as its charge in new devices and memory stor age applications. This could vastly improve the memory density of hard driv es and efficiency of devices. This control of the spin characteristics of ma terials is the motivation behind so-called


4 “spintronics” [1]. Controlling the spin tr ansport may be readily accessible in materials that are spin polarized. A fully spin-polarized material is called a “halfmetal,” a term first coined by de Groot in 1983 [8]. Half-metals are unique materials that behave as insulators for the transport of one electron spin orientation and as conductors for the other spin orienta tion. Note that the term “half-metal” refers to the material being ‘metallic’ for spin transport in one direction and ‘insulating’ for spin transport in the the opposite direction. CrO2 thin films have been shown to exhibit half-metallic behavior and spin polarization both with theoretical band-structure calcula tions [9-11] and by experimental observations using spinpolarized photoemissi on spectroscopy [12] and Point-Contact Andreev Reflection (PCAR) [13]. The PCAR method is introduced and discussed in detail in a later section (2.3.4). Polarization values for CrO2 have been observed by some groups to be in the 60-90% range [14], giving encouraging evidence of nearly id eal half-metallicity. It is somewhat difficult to measure the spin polarization in Fe3O4. Band structure calculations [15, 16] indicate a half-metallic nature with a gap in the majority density of states, but PCAR cannot verify this due to its property of ferromagnetism at room temperature changing to antiferromagnetic insulator below the so-called Verwey transition around 120K. Since PCAR re quires forming a junction between Fe3O4 and a superconductor, it is impossible to acc ess the ferromagnetic metallic state in the experiment using conventional low Tc superconductors. Efforts are currently


5 underway in our lab to overcome th is limitation by growing granular Fe3O4 films within a conducting matrix. 1.2 Motivation and research plan A major motivation of th is project was to set up PCAR experiments and for the first time combine it with the PPM S, and take advantage of the excellent temperature and field control afforded by it. This allows the user to collect conductance vs. bias voltage data whil e changing any of the following three variables; junction resistance, temperat ure and external magnetic field. A group at Argonne National Laboratory in Illinoi s has developed a somewhat similar apparatus that integrates with a continuous flow cryostat [17], but to our knowledge ours is the first probe of its kind to be combined with a PPMS. This unique home-built probe has been designed and developed to carry out point contact Andreev reflecti on (PCAR) experiments to determine the spin polarization in magnetic materials. To date, this instrument and the research generated by it have resulted in a Mast ers thesis, two published papers [18, 19], two pending manuscripts, and several confer ence presentations [20-25]. A patent has been accepted by the USF patent and licensing office and has been submitted to the U.S. Patent office [26]. This inst rument has very versatile capabilities with respect to studying charge and spin trans port over a large range of temperatures and magnetic fields. One goal of this disse rtation has been to extend PCAR to the measurement of spin polarized current through SWNT patterned structures. And


6 in this capacity as in with standard th in film and bulk samples, the temperature and field capabilities make our PCAR probe unique. Our PCAR method is of interest to other res earch groups currently working in this area. For example, the Naval Research Lab (NRL) group of Bob Soulen and Mike Osofsky have develope d similar PCAR experiments [14]. In their system, the point/sample junction is completely immersed in liquid helium. This affords only limited choice in operati ng temperatures. In fact, there are only two possibilities: 4.2 K or 1.6 K achieved by pumping on Helium. In discussions with our University of South Florid a (USF) materials physics group, the NRL group has shown a great deal of interest in not only using the PPMS to control temperature, but to also use the extern al magnetic field capability to vary the superconducting and magnetic properties of the junction. Since there are very few experien ced groups currently active in this research area, we were motivated to deve lop a versatile, turnkey system that can be applied to a wide range of current and topical problems in magnetic and superconducting junctions. Our first task was to reproduce known results in conventional materials. Data has been acquired for a Sn-Cu junction and a Nb-Cu junction to verify the ability of th e apparatus to repr oduce normal Andreev reflection [27] at a superconductor-normal metal (SC-NM) junction. Sn and Nb are superconducting materials below 3.7 K and 9.5 K, respectively while Cu is a normal non-magnetic metal with zero spin polarization. The conductance is seen to increase by a factor of two at low-bias voltages. Data has also been acquired


7 for a Sn-CrO2 junction to investigate the spin-polarization of the CrO2 sample. The conductance drops at low bias volta ges and the data indicate a spin polarization of 70%. The deta ils of our experiments and results will be discussed in subsequent chapters. The issue of spin polarization is st ill not completely resolved. For example, one never measures 100% spin polar ization as predicted by theory. It is believed that the surface degrad es and a layer of insulating Cr2O3 forms on the CrO2 surface. The NRL group has indicated that others are attempting to circumvent this problem by freshly depos iting films and literally running them to the next lab to perform spin -polarization tests. A que stion may arise as to why there is any interest at all if this material degrades so quickly, but in the process of fabricating devices the CrO2 would be deposited and then covered while still in the chamber. This would prevent exposure to O2. In this way the film would still be of high quality and would not have the Cr2O3 layer. It is only in the spin polarization and other transpor t experiments that the materi al needs to be exposed. Nanotubes can be grown by chemi cal vapor deposition (CVD) [28], plasma enhanced chemical vapor de position (PECVD), carbon arc method, and laser ablation. Nanotubes have been grown with CVD as well as PECVD in this work. Sensing applications are numerous for SWNT devices that use mechanical resonance, resistance, and cap acitance [29, 30]. Ongoing research at USF incl udes growth of carbon nanotubes in two different furnaces. On uses plasma enhanced chemical vapor deposition


8 (PECVD) in collaboration with the laboratory of Dr. Rudy Schlaf. This growth is initiated with Ni thin film as the metallic catalyst. Dropcasting techniques and Langmuir-Blodgett (LB) deposition technique s were extensively investigated for aligning the nanotubes. A chemical vapor deposition (C VD) nanotube growth system has been built through a collaboration between the laboratories of Dr. Hariharan Srikanth and Dr. Garrett Matthews, both of the USF Dept. of Physics. This system typically uses Fe nanoparticle magnetic ca talysts. This has allowed in-house growth of nanotubes, SWNT structures which could be patterned into various device geometries, and in general will be useful for many different applications. With access to CNT growth systems here at USF and also at NRL-Washington, we have achieved our specific goal of developing and testing SWNT network structures for bio and chemical sensi ng. Sensing with SWNT devices has the potential to extend to biological analyt es that are excreted through the human skin. These devices have been shown to be very sensitive to a wide variety of chemical analytes that readily adsorb ont o the nanotube surfaces. Specificity is a key to applying this method to a desired biomolecule. Initial experiments have also been performed in the NRL biosensing system by initializing the microfluidics, calibrating the syringe pumps, performing initial pH testing, and completing several sets of DNA functionalizations. The final portion of our research pl an is to study the modifield BlonderTinkham-Klapwik (BTK) model for fitting our experimental PCAR spectra. We


9 are studying the square root behavi or in the analysis of the SrRuO3 series. Studies are also continuing regarding fundamental aspects of the theoretical fitting in the La1-x(Ca, Ba, Sr)xMnO3 series. We are are using a 2 statistical model to more accurately determine the spin polarization, P.


10 Chapter 2 Background and Fundamental Physics 2.1 Magnetic oxides Magnetic oxides are used in a large number of important applications like read heads for hard drives, magnetic field sensors, galvanic isolators, and magnetoresistive random access memories [1]. The magne tizations of individual domains can be designated as ones and zeros and thus forming the binary logic for writing and reading. Magnetic oxide materials in general have been in use for many decades in storage media such as cassette tapes, VCR tapes, and co mputer floppy disks. The observation of the giant magneto-resistive (GMR) effect in 1988 [ 33] is considered to be the beginning of the new spin-based electronics age. While GMR is seen in magnetic multilayers, a related effect called colossal magnetoresistance (CMR) is seen in a class of magnetic oxides. 2.2.1 Device applications In a GMR device (figure 1), two ferromagnetic layers sandwich a nonmagnetic layer with one of the two magnetic layers be ing “pinned” and the other being “free”. This refers to the magnetiz ation of the pinned layer being relatively


11 insensitive to moderate magnetic fields, while the magnetization of the free layer can be changed by a relatively small magnetic field. As the magnetizations in the two layers change from parallel to antiparallel alignmen t, the resistance of the spin valve rises typically from 5 to 10%. New materials operate at room temperature and exhibit substantial changes in resistivity when subjec ted to relatively small magnetic fields (100 Oe to 1000 Oe). A magnetic tunnel juncti on (MTJ) is a device with a pinned layer and a free layer, separated by a very thin insulating layer, commonly aluminum oxide. The tunneling resistance is controlled by a magnetic field in the same way as a GMR device, but can exhibit a 20 to 40% change in the magnetoresistance, and requires a saturating magnetic field equal to or somewhat less th an that required for spin valves. No commercial sensors using MTJ st ructures are available yet, but are under development. For these devices and other spin-based el ectronic structures, there is a need for ferromagnetic materials with a high degree of spin-polarization. Materials like CrO2 and other oxides like La1-xSrxMnO3 that exhibit CMR effects are excellent candidates for spintronics applications.


12 Figure 1: Some common magnetoelectroni c device schematics. [from Osofsky, NRLWashington, powerpoint presen tation, with permission] Magnetolectronic Devices GMR Spin Valve F N F Lo w High F N F Nonvolatile Memory (RAM) superconducto r injecto r barrier Current Switch Hard disk read head


13 2.1.2 Future prospects for spintronic devices Recently, there has been much effo rt to develop a new class of electronic devices whose prope rties are determined by the spin of the electron. The performance of such devices are enhanced as the spin polarization, P, of the ferromagnetic materials increases. As an example, the tunneling magneto-resistance (TMR) of ferromagnetic/insulator/ferromagnetic (FIF) devices increase dramatically for high values of P. The TMR is expressed as R/R, or the change in resistance of a FIF device when the orientation of the moments switch from parallel to anti-parallel. This is described by Juliere [11] as follows: 2 1 2 11 2 P P P P R R where P1 and P2 are the spin polarizations of the first ferromagnetic layer and the second ferromagnetic layer, respectively. If the same ma terial is used for both layers, as is often the case, this expression becomes: 2 21 2 P P R R A plot of log R/R vs. P reveals in dramatic fashion the slow increase in R/R, and then a sharp increase as P approaches a valu e of one (figure 2). (P = 1 corresponds to 100% spin polarization.) This observation has led to a search for materials with very large P values. Several oxide materials are believed to have P values near one, including CrO2 and some optimally doped manganese oxides. [14]


14 2 21 2 P P R R Figure 2: The Juliere plot of tunneling magneto-resistance R/R vs spin polarization P. From Osofsky et al [14]


15 2.2 Half-metallic system: chromium dioxide (CrO2) Chromium Dioxide shows potential to be a very promising and versatile magnetic oxide for applications in memory st orage and spin-based devices. There is a vast amount of current work dealing with CrO2 due to its classification as a half-metal as previously mentioned, its anisotropic propertie s, and its desirable va lues of saturation magnetization and coercivity. This section includes previous work dealing with its crystal structure and energy band calculat ions, production and deposition techniques, magnetic properties, anisotropy data, and measurement of spin polarization. 2.2.1 Crystal structure Chromium dioxide crystallizes in the rutile structure with a unit cell containing two CrO2 formula units for a total of six atoms (figure 3) It is tetragonal with a c/a ratio of 0.65958 and a lattice constant of a=0.4421 nm [11]. The measur ed values of the structural parameters a, c, and u are 4.419 A, 2.912 A and 0.303 A, respectively [11]. The Cr atoms form a body-centered tetragonal lattice and are su rrounded by a slightly distorted octahedral array of oxygen atoms. Chromium atoms are locat ed at the positions [0,0,0] and [,,] in lattice coordinates, and the four oxygen atoms are located at [u,u,0], [1-u,1-u,0], [+u, -u, ], and [u, +u, ], where u is a dimensionless internal coordinate less than unity. Lewis et al [11] fixed the axes at their experimental lengths and have optimized the internal c oordinate, u. They computed a value of u=0.3043, in excellent agreement with experiments.


16 Figure 3. Illustration of the rutile structure of Cr02. Heavy solid and dashed lines demarcate unit cells of the crystal. Thin lines emphasize the oxygen octahedral surrounding each chromium atom, with solid lines highlighting the equatorial planes and dotted lines linking them to the oxygen atoms. From Lewis et al. [11].


17 Figure 4: Rutile Structure for CrO2. From Kulatov & Mazin [10]


18 2.2.2 Energy band calculations a nd density of states (DOS) The energy band structure of chromium dioxide was first calculated in 1986 by Karlheinz Schwarz of IBM [9]. He used the local-spin-density-approximation (LSDA) method and concluded that the majority-spin electrons are metallic and the minority-spin electrons are semiconducting. These results fo r the energy bands of the majority spins and the minority spins are shown (figure 5). The gap of about 1.5 eV between the ‘O 2p’ and the ‘Cr 3d’ bands of the spin-down elect rons yield a slightly smaller gap than observed experimentally. It is suspected that CrO2 minority electrons are in an insulating structure as opposed to semiconducting. These results were used to ca lculate the energy density of states (DOS) in states per atom per eV for the majority-spin electr ons (spin up) and the minority spin electrons (spin down). The most important results ar e that a metallic character is found for the majority-spin electrons, while the Fermi En ergy falls in a gap for the minority-spin electrons (figure 5). Thus Ch romium dioxide is predicted to be a semiconductor only for minority-spin electrons. Kulatov and Mazin support the half-metallic ferromagnet classification of CrO2 [10], and they calculated simila r band structure and DOS usi ng a so-called spin-polarized scalar relativistic linear muffin-tin orbi tal (LMTO) method. The first plane-wave pseudopotential technique (PWPP) within the local-spin-d ensity-approximation (LSDA) was performed in 1997 [11] and also verifies half-metallicity and ener gy density of states (figure 6).


19 Figure 5: Energy band structure and en ergy density of states (DOS) for CrO2. From Schwarz [9].


20 Figure 6: Densities of states for major ity and minority spins in ferromagnetic CrO2. Majority (minority) states are plotted along the positive (negative) vertical axis. The dotted line denotes the Fermi en ergy. Schwartz et al [9]


21 2.3 Thin film growth and magnetism of CrO2 Chromium dioxide has been shown experi mentally to exhibit ferromagnetism and large in-plane anisotropy. In a later secti on, we will discuss the magnetic properties. We now briefly discuss the growth of film s used in our experimental studies. 2.3.1 Production and deposition techniques Chromium dioxide samples have been prepared through chemical vapor deposition (CVD) techniques [4] by collaborators. CrO2 is a metastable phase, and it can be challenging to obtain this phase using c onventional thin film growth techniques under ambient conditions. The films used in thes e studies were 5 mm x 5 mm squares or 6 mm x 2 mm rectangular samples and typi cally 1000 to 2000 thick. The CrO2 films have been epitaxially grown on TiO2 substrates using CrO3 as a precursor. X-ray diffraction indicate that the films grew co mpletely (100) oriented, in a ccord with the (100) oriented substrates. The CVD reactor consists of a twozone furnace and a quartz tube. The precursor powder is placed in a quartz boat in the source zone and the substrate is placed on a susceptor in the reaction zone. Oxygen is used as a carrier gas for the sublimed CrO3 to be transported to the reaction zone wh ere it decomposes on the substrates to form CrO2 with evolution of O2. The source zone temperatur e and the oxygen flow rate are fixed at 260 oC and 100 cc/min, respectively. The film s were characterized using x-ray diffraction, transport and magnetic measurem ents. Details of film growth and characterization are presente d by DeSisto et al [4].


22 2.3.2 Magnetic properties The band structure and density of stat es calculations have predicted CrO2 to be a half-metallic ferromagnet. Magnetization vs. applied magnetic field data verifies this prediction and shows a hystere sis in the M-H loop. In th is kind of experiment, an external magnetic field is ramped up to a high positive value allowing the sample magnetization to saturate. The field is then reversed and ramped back to zero, which reveals the remnant magnetization. The field is ramped all the way to saturation in the opposite direction and then back to positive saturation once more. Magnetic hysteresis M-H loops for CrO2 are shown (figure 7). The chromium dioxide films are grown on single-crystal (100) TiO2 substrates that also have rutile structure, thus leadi ng to a well-defined magnetic easy and hard axis along the [100] (c axis) and [010] (b axis), respectively. The M-H data in figure 7 is thus shown at room temperature and low temperatur es with the applied magnetic field parallel ( = 0o) and perpendicular ( = 90o) to the easy axis. An important property of CrO2 is the very large in-plane anisotropy that is reflected in the shapes of these hysteresis loops. With field parallel to easy-axis, distinct co ercivity is observed with the curve almost rectangular and the sample saturates at low ma gnetic field. With field perpendicular to easy-axis, magnetic saturation oc curs at fields a few hundred Oe higher than that in the other case.


23 Figure 7: M-H Hyster esis loops for CrO2. From Spinu et al [34]


24 2.3.3 Transverse susceptibility data Transverse susceptibility ( T) is an extremely useful method to probe the dynamic magnetic behavior of materials in general and the effective anisotropy in particular. The basic concept of these experiments is to meas ure the change in transverse component of the magnetization as the applie d static field is varied. Our research lab has a unique home-built probe that can measure these changes at radio frequencies (~10 MHz) with high sensitivity. Thin-film samples are placed in the core of an inductive coil that forms part of a self-resonant LC circuit with a tunnel diode oscillator (TDO) having a resonant frequency around 10 MHz. The change in resonant freque ncy is measured as the static field is ramped from negative to positive saturation a nd vice-versa. This quantity is proportional to the change in inductance th at is in turn governed by the T of the films. The field dependence of T is plotted as a dimensionless ratio, 100 ] ) ( [ X Hsat sat T T T where sat is the transverse susceptibility at a satu ration field of H = 5 kOe. The field dependent T at room temperature for two different orientations [34] are plotted in figure 8. The top panel is the data for the case wher e the field is parallel to the easy axis of magnetization. The data is consistent with switching and anisotropy fields around 60 Oe. For field perpendicular to easy axis (figure 8b) the curve has a characteristic difference in peak positions which are now recorded at 600 Oe. These two curves demonstrate the large in-plane magnetic anisotropy in the CrO2 thin film. Using an electromagnet mounted on a rotating base, we have also explor ed the full angular dependence (figure 9)


25 Figure 8a and 8b: Transverse susceptibility ( T) vs. applied magnetic field (H) for CrO2 thin film at 0o and 90o orientation of applied field with respect to the easy axis of magnetization. From Spinu et al [34]. -6-4-20246 0.000 0.005 0.010 H (kOe) (%) T=300K, =00 H increasing H decreasing- -10-50510 0.0 0.2 0.4 0.6 0.8 1.0 H/HK -6-4-20246 0.000 0.002 0.004 0.006 0.008 0.010 0.012 (%)H (kOe)T=300K, =900 H increasing H decreasing-10-50510 0.0 0.2 0.4 0.6 0.8 1.0 H/HK


26 Figure 9: (a) TDO circuit for measuring tr ansverse susceptibility (b) CAD drawing of TDO probe (c) CAD drawing of rotating base (d) Angular dependence of transverse susceptibility for CrO2 thin film. From J. Sanders L C HRF HDC TDO -20020406080100120140160180200 -2 0 2 4 6 8 10 12 14 16 18 T = 296 K Field = 3030 G Angular Dep. of Transverse Susceptib ility(T/T) %Angle (Degrees)


27 as the field orientation is c ontinuously changed from 0 to 90o with respect to the easy axis. Figure 9 shows the presence of a sma ll angular hysteresis. The origin is not understood as this time but we generally believ e that it may be related to the nucleation and motion of domain walls. 2.4 Superconductivity and Andreev Reflection Superconductivity is a physical phenomenon that is central to the experimental measurement in this work. The use of a s uperconducting experiment al probe facilitates the measurement of spin polarization in ma terials. The superconducting metals of Nb, Pb, and Sn have been used in the experiment s herein due to their expense, availability, ease of machining and the superconducti ng properties of critical field (Hc), critical temperature (Tc) and superconducting energy gap ( ). A brief discussion of these three properties as well as the quantum and clas sical concepts of superconductivity will precede an introduction to Andreev Reflection. 2.4.1 Superconductivity Superconductivity was first discovered by the Dutch physicist H. KamerlinghOnnes in 1911 by measuring a sharp decrease in the electrical resistan ce of Mercury at a temperature just above 4.2 Kelvins. The pr ocess required using liquid Helium to achieve such low temperatures, and it was Onnes himsel f who first liquefied the rare gas in 1908. His work and other early experiments verifi ed a very low resistance (on the order of milliohms) and subsequent experiments have observed electrical currents in


28 superconducting rings persist for over 1.5 years without a decrease in current whatsoever, until the experimentalist finally tired of the obs ervation. This persistent current over long time periods implies a resistance of exactly zero. The experimental temperature below which is observed this sharp decrease in resistance, where the metal becomes superconducting is called the critical temperature and is denoted by Tc Critical temperatures of elemental superconductors ar e typically near abso lute zero (0-10K) and thus need to be cooled with liquid Helium or even pump on the liquid helium to achieve evaporative cooling, or other methods to achieve temperatures below 4.2 Kelvins. Niobium has the highest critical temperat ure of elemental superconductors with a Tc equal to 9.2 Kelvins. Niobium wire is used in this work, along with Pb and Sn which have critical temperatures of 7.2 K and 3.7 K, respectively. In the early years superconduc tivity was thought to be me rely a disappearance of electrical resistance until a more sophisticated phenomenon was observed in 1933. Two German physicists, W. Meissner and R. Ochsenfeld observed a weak but constant magnetic field being forced out of a samp le as it was cooled below its critical temperature. Subsequent experiments also show that if a sample is already cooled below Tc, then a weak external applied field also wi ll not permeate the sample. This is called the Meissner Effect The causality was unknown at the time, but clearly there was an opposition to a magnetic flux in a superconducting material. This led to the discovery of anothe r physical property of superconducting materials. There is an energy expenditure re quired to expel the magnetic field from the sample as in the Meissner effect. If a suffi ciently high external ma gnetic field is applied


29 to a sample, it cannot be expelled in the c ourse of metal cooling, and superconductivity will not occur. Such a magnetic field is called the critical magnetic field and is denoted by Hc. The experiments performed in this work will utilize superconducting metals above and below both their critical fields and critical temperatur es. Conductance vs. voltage curves will be presented as functions of both temperature and magnetic field. We will explore the behavior of a superconductor/sa mple interaction as we scan from zero field to values well above the critical fiel d. We will also observe conductance curves from 2 Kelvins to temperatures well above the critical temperature of our superconducting probes. The decisive role in revealing the corr ect theory of superconductivity was played by the isotope effect. The other experimental facts above, in spite of their uniqueness, gave insufficient material for understanding the nature of supe rconductivity. Both experimentalists and theorists did their best to discover in which direction they should search for the explanation of this mysterious phenomenon. The isotope effect, which had been studied theoretically by Frohlich and discovered in 1950 by Maxwell and Reynolds, revealed this direction. A study of different superconducting isotopes of mercur y established a relationship between the critical temperature and the isotope mass; as the mass number M was varied from 199.5 to 203.4, the critical temperature decreased from 4.185 K to 4.140 K. With sufficient precision, it was established th at for a given element the fo llowing relationship holds: TcM1/2 = constant


30 This implies that the isotope mass is a characteri stic of the crystal lattice and can affect its properties. For example, the frequency of latt ice vibrations is related to the ion mass in the following way: ~M-1/2. Superconductivity, which is a property of the electron system, is shown by the isotope effect to be rela ted to the state of the crystal lattice. It follows that superconductivity is due to the interaction between th e electrons and the lattice. Interestingly, this very interaction is also responsible for electrical resistance in metals. Under certain conditions the el ectron-lattice interaction leads to the disappearance of resistance, namely superconductivity. The vibrations of the lattice can be con ceptualized classically as small spheres on springs in a periodic latti ce, simple cubic (SC) for example. This mass-on-a-spring analog can be extended to the mathematics of the harmonic oscillator. Quantummechanically, the vibrations of the lattice are called phonons, since they have discrete energy values and can be considered packets (o r particles) of vibra tional energy. The fact that electron-phonon scattering is one of the principle mechan isms of resistance was for many years a stumbling block, making it extremely hard to imagine that the same interaction can lead to a vani shing of resistance. The isot ope effect was discovered in mercury and this curiously was the element th at served in both discovering the longmysterious phenomenon of superconductivity and discovering the isotope effect which helped to crack the puzzle. The theory of superconductivity was de veloped soon after the discovery of this effect. A theory of superconductivity, which explained the nature of this phenomenon, was formulated in 1957 by Bardeen, Cooper, a nd Schrieffer (BCS Theory). The mystery


31 of superconductivity was solved, and the forma tion of the theory brought further progress in the field. The BCS theory can be c onceptualized both classically and quantummechanically. The classical description of the conceptual physics is as follow s: the interaction between the electrons and the la ttice leads to a certain eff ective interaction between the electrons. An electrons moving in a metal defo rms (or polarizes) the crystal lattice due to attractive Coulomb forces. The displacement of the ions in the lattice caused by this electrical attraction now affect s the state of the other electr on, since the latter now finds itself in the field of the polarized lattice with a somewhat changed structure. Thus there is an effective attraction between the two electrons. The appearance of the attractive force can be visualized by conceptualizing the deformation of the lattice and the electr on being surrounded by a cloud of positive charge which is attracted to the elec tron. The magnitude of this positive charge can exceed the electron charge. Then this electron, togeth er with the surroundi ng cloud, represents a positively charged system, which will then be attracted to another electron. Hence the formation of pairs of electrons. The quantum-mechanical description is con ceptually as follows: In an electron scattering in a lattice wave (electron-phonon collision) th e energies, velocities, and directions of motion of the particles change but the laws of cons ervation of energy and momentum hold. An electron moving through the lattice will absorb a phonon and take on its quantized values of energy and momentum; or an electron emits a phonon,


32 inducing lattice oscillations by giving away part of its en ergy. All such processes are called electron-phonon interactions. In the normal state (as opposed to th e superconducting state), this interaction gives rise to electrical resistance: a movi ng electron excites lattice oscillations in the course of which it is slightly decelera ted. Electron-phonon inter action provides not only the cause of resistance, but also its di sappearance at low temperatures, namely superconductivity. To understand this phenomenon, think of a phonon excited by an electron. When an electron moves through a lat tice it induces oscillat ion in the ions in the lattice sites. Say we have a metal at a temperature of absolute zero, where its crystal lattice is not at absolute rest, but executes the so-called “zero-poi nt” vibrations, which corresponds to the ground stat e of the harmonic oscillat or from quantum mechanics theory. An electron moves thr ough the crystal and disturbs th e state of these zero-point oscillations and excites the lat tice. When the latti ce returns to the ground state, it radiates energy that is absorbed by another elect ron. This electronphonon interaction, or exchange of phonons in the quantum picture, leads to the additiona l attraction between electrons. This attraction prevails over Coulomb electron repulsion and at low temperatures, thermal vibrations are not enough to ove rcome this phenomenon. As temperature increases however, thermal agitation becomes so great, that this effect disappears. At low temperatures the electron system becomes a bound unit, Cooper pairs. A conceptual model describes cooper pairs as a singlet pair of individual electrons, one with a wave vector + k with spin up intrinsic angular momentum and the other with – k and spin down


33 intrinsic angular momentum. This model ca nnot be taken literall y, though because there are on the order of 106 electrons in the volume occupied by a Cooper pair [36]. These electrons behave collectively and on average as in many concepts of quantum mechanics. Conceptually, the Cooper pairs ac t as Bosons (spin zero partic le) because they are always composed of two electrons each with half -integer spin (Fermions) of opposite sign. Fermions obey Fermi-Dirac statistics and thus obey the Pauli exclusion principle. Thus the electrons in the normal st ate have a continuous energy spectrum. In contrast, Cooper pairs do not follow the Pauli exclusion principl e and are perfectly fine to be all in the same energy level. For this reason, a large am ount of energy must be imparted in order to raise these Cooper pair s to an excited state, hence an energy gap. Another way to conceptua lize these Cooper pairs is that these electron pairs behave as a bound unit, and finite energy must be expended in order to excite it. The energy spectrum of the system in this case wi ll not be continuous. The excited state is separated from the ground state by an energy interval, referred to as the energy gap These Cooper pairs are bound to each other as Bo sons, and since they are in this bound state, a finite energy must be expended in order to excite this state. Electrical resistance of a metal is due to the interaction of th e moving electrons, or current dq/dt, with the vibrations of the crys tal lattice or with im purities. However, if there is a gap in the energy sp ectrum, quantum transitions in the electron fluid will not always be possible. The electrons system will not be excited when it is moving slowly, i.e. at low temperatures below 10 Kelvins. This implies movement without friction, that


34 is, the absence of electrical resi stance. This gives rise to pe rsistent currents without any decrease in magnFitude, namely superconductivity. The size of a Cooper pair was mentioned ear lier and is an interesting exercise. An estimate of the size can be made by conceptu alizing the simplest case where the Fermi energy EF is equal to pF 2/2m. The superconducting state is characterized by the presence of a gap near the Fermi surface. The momentum spread connected with this is determined by the relation pF p/m Therefore p m / pF From the uncertainty relation p r one can estimate the spatial spread as r vF/ where vF is the Fermi velocity. The coherence length is defi ned as the size of the Cooper pair and is defined as: Fv 0 The quantity 0 characterizes the scale of special correlation in a superconductor. Substituting typical values of vF and we obtain 0 ~ 10-4 cm. The frequency of vibration of a crystal lattice is on the order of tens of GHz so it would have a period of about 10-8 cm. This means that electrons form ing a Cooper pair are separated by an immense distance of 104 lattice spacings. Even so, they are the electrons that are most strongly attracted to each other. The final concept within the concept of superconduc tivity needed to understand the phenomenon of Andreev Reflection is the behavior of the supe rconducting gap as a function of temperature. In the preceding di scussion, the thermal mo tion of the electrons was neglected, and it was effectively a discus sion of processes at absolute zero. Now finite nonzero temperatures will be consid ered. Thermal motion excites the electron


35 system and disrupts, or at least reduces, electro n interaction so we must think about excited states of the electron system rather than only the ground states. At finite temperatures in superconductors there appe ar quasiparticles which can change their energies by an arbitrary amount and behave just like ordinary electrons. Their energy is somewhat different than in the normal Ferm i-Dirac distribution a nd BCS theory derives the supercondicting gap as a function of temperature as: c cT T aT T 1 ) ( where a is equal to aBCS = 3.06, a constant from BCS superconductivity theory. Gap values for Sn have been calculated for temperatures from 2.0K to 4.0K, using its critical temperature of 3.69K. Conductan ce curves are fitted using an iterative computational method that holds the parameters of temperature, sample resistance, and superconducting gap fixed while it fits the e xperimental data to two other variable parameters. The theoretical fitting pr ocedure is described in section 3.4.2. The effective interelectr on attraction in superconducting materials leads to the appearance of the energy gap in the energy spectrum. This results in the electronic system being unable to absorb arbitrarity small amounts of energy. Experiments have been able to verify the existence of a superconducting gap and also measure its magnitude. The existence of a superconducting gap allows the fundamental processes in our experimental experiment al measurements. However, before explaining Andreev Reflection, let us consider tunneling. Think of electrons flowing across a thin insulating layer about 10 Angtroms in thickness which separates a normal film from a superconducting material. There is a


36 finite probability that the electrons will tunnel across the barrier in the presence of a bias voltage. This is an example of quantum-mechanical tunneling and will give rise to a tunneling current. The electron cannot tunnel in to a disallowed state, but only into an allowed state in the spectrum of th e superconducting material. Thus when experimentalists observe a curr ent-voltage curve they see a discontinuity in the curve. The levels are filled up to the maximum in the normal metal, namely the Fermi level (EF), and are filled up to EF in the superconductor, where is the superconducting gap. In such a case no tunneling current can appear. Now that we have c onceptualized a clear picture of the origin and hist ory of superconductivity and th e quantized ener gies leading to the superconducting gap we can continue on with the specific behavior of superconducting metals in this work. This phenomenon in the tunneling current spectra is an analog to how the superconducting ga p leads to the observation of Andreev reflection in our experiments and th e extraction of spin polarization. 2.4.2 Measurement of spin polarization using PCAR As mentioned previously, the performance of devices increases dramatically as the spin-polarization P appr oaches 100% (see figure 2). In this section this P value will be more precisely defined as where N(E) is the spin-dependent density of states at the Fermi level, EF. Measuring P requires a spectroscopic technique that can discriminate be tween the spin-up and spin-


37 down electrons near EF. Spin-polarized photoemissi on spectroscopy is capable of providing the most direct measurement of P, but lacks the necessary energy resolution (~1 meV) [14]. An alternative is the tunne l-junction technique pi oneered by Tedrow and Meservey [13]. However, this technique is constrained by the require ment to fabricate a layered device consisting of a thin-film ferromagnet (FM) on top of a uniform oxide layer 10 to 20 thick that is formed on top of the superconducting (SC) base. The need for a thin, uniform oxide layer is a severe limitation of the technique because many interesting materials cannot be deposited in such a way. In a technique developed by Soulen and Os ofsky at the Naval Research Lab [14], a metallic point-contact is formed between the spin polarized magnetic sample and a superconducting tip using a simple mechanical adjustment. The electronic transport properties at the point-contact junction between the tip and sample can then be used to extract the P value of the sample. This technique is called Po int-Contact Andreev Reflection (PCAR). The conversion of normal current to superc urrent at a metallic interface is through a process called Andreev re flection [27] and is a well-understood phenomenon in the field of superconductivity. To understand AR, consider figure 10a showing the energy band schematic for an electron in a metal w ith P = 0 propagating toward the interface. For the electron to enter the superconductor, it must either have an energy above the superconducting gap energy or be a member of a Cooper pair. So, it would appear that for energies below the superconducting gap, a si ngle electron cannot cross the junction. However through the AR process, there is a finite probability of the single electron


38 forming a Cooper pair with a second electr on in the metal and traveling across the junction into the superconducting condensate. For current conversion, the hole left behind in the metal is ba ck reflected. This Andreev reflected hole acts as a nother conducting chan nel to the initial electron current, thus doubling the normal-state conductance Gn (where G = dI/dV and V is the bias Voltage) of the point -contact for applied voltages eV < where is the superconducting gap at the interface. This is illustrated in figure 11 where the uppermost data showing a Nb tip pressed into a Cu fo il sample. Note the doubling of the normalized conductance from 1.0 outside the su perconducting gap to 2.0 within Now consider figure 10b to understand the effect of P on the Andreev reflection process. Since a superconducting pair consis ts of a spin-up and a spin-down electron, and incident spin-up electron in the metal requires a spin-down electron to be removed from the metal as well for conversion to s upercurrent. However, if P=100% as shown then there are no spin-down states in the metal at EF to provide another member of the pair. This is because of the gap in the spin polarized metal at EF for spin-down states. The consequence of this is that AR is suppressed and thus th e doubling of conductance for eV< is also suppressed. This suppression of the normalized conducta nce can be easily seen in figure 11. This figure is a reproduction from the Science ar ticle of Soulen et al [14]. Notice that the conductance vs. bias voltage curve for CrO2 shows a dramatic suppression within the


39 Figure 10a and 10b: Density of states for a superconductor-normal metal (SC-NM) junction and a superconductor-HM (SC-HM) junc tion. Andreev reflection occurs in the former but is suppressed to zero conducta nce at low bias voltages in the latter. From Soulen et al. [14].


40 Figure 11: Normalized conduc tance vs. bias voltage for a series of point-contact experiments. The superconducting tip in this case is Nb. From Soulen et al [14].


41 superconducting gap of niobium (~ 1.5 meV) SoulenÂ’s group reported a spinpolarization of 90% 3.6% for CrO2 and 78% 4.0% for La1-xSrxMnO3 (LSMO), another material calculated to be a half-metal One can see from this figure that PCAR is also a quick and convenient method to look at other materials incl uding the standard ferromagnets of Ni, Co, and Fe as well as NiFe alloy. These P values have been calculated using a model due to Blonder-T inkham-Klapwijk (BTK) for conventional Andreev reflection modified to ta ke spin polarization into account.


42 Table 1. From Soulen et al [14].


43 Chapter 3 Point-contact experimental results and discussion 3.1 Point-Contact Andreev Reflection (PCAR) data The design and constructuction of our home built PCAR probe is described in Appendix C. Initial systematic studies on a number of junctions were done using our PCAR probe and we successfully demonstrated conventional Andreev reflection (AR) in superconductor–normal metal (SC-NM) j unctions and suppression of AR in superconductor-half metal (SC-HM) junctions Immediately after final wiring and vacuum testing of the probe, the first set of experiments was performed using a Nb wire point and a Cu foil sample to esta blish a SC-NM junction. While good AR characteristics were observed, it was difficult to completely eliminate the Niobium oxide layer on the tip surface (assumed to be Nb2O5). So we also decided to conduct AR experiments with a superconducting Sn tip. After modifications to the PCAR probe, a series of experiments was performed using a Sn tip and a Cu foil sample. A similar set of experiments using a Sn tip and a CrO2 sample were done to establish a SC-HM junc tion. In all cases, data was collected at different temperatures below the superconducting Tc and at a number of fixed fields up to the critical field (Hc) of the superconductor.


44 Finally, a series of experiments was performed using a Ni wire tip and a Magnesium Diboride (MgB2) pellet to investigate a ferromagnetic metal – superconductor (FM-SC) junction. MgB2 is a material that is thought to be a two gap superconductor with a fairly high transition temperature near 38 – 40K. This unique material is of current intere st and is “full of surprises for both experimentalists and theorists” [35]. 3.2 Superconductor Normal Metal (SC-NM) junction The first set of data shown is a simple current vs. voltage curve for an Sn-Cu junction (figure 12). This experiment was perf ormed with a stable temperature of 2.00 K, which is below the superconducting transition temperature of Sn (Tc = 3.722K) according to Kittell [36]. At high bias voltage one can see a linear vari ation of I vs. V, inferring a constant normal state conductance dI/dV. At lo w bias voltage the slope is twice as steep, indicating conventional Andreev Reflection (AR) at this volt age. Conventional AR is indicated here by the signature doubling of the conductance with in the superconducting gap of Sn occurring at = 0.58 meV [36]. This effect ma y be more easily seen in the normalized conductance (G(V)/GN) vs. bias voltage curve for Sn-Cu (figure 13). Here we define G = dI/dV where I is current and V is voltage. GN is the normal-state conductance. This doubling in conductance occurs in conventional AR due to the formation of Cooper pairs and the back-re flection of a hole across the junction, as discussed earlier. The data in figure 13 is a classic signature of AR in SC-NM junctions. Note that one has to achieve a good, clean cont act without any impurity or oxide barrier


45 -2.0-1.5-1.0- -4 -3 -2 -1 0 1 2 3 4 5 T = 2.00 K H = 0 Gauss I-V for Sn-Cu JunctionCurrent (mA)Bias Voltage (mV) Figure 12: Current vs. bias voltage for a Sn Tip and a Cu foil sample establishing a Superconductor – Normal Me tal (SC-NM) junction. The superconducting gap features occur at = 0.58 meV. From J. Sanders


46 -2.0-1.5-1.0- 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 T = 2.0 K H = 0 Gauss Normalized Conductance vs. Bias Voltage Sn-Cu JunctionG(V)/GNBias Voltage (mV) Figure 13: Normalized conducta nce (dI/dV) vs. bias voltage for a Sn Tip and a Cu foil sample establishing a SC-NM junction. Th e superconducting gap of Tin occurs at = 0.58 meV. From J. Sanders


47 in the junction area to observe ideal AR. Th is is clearly the case in our Sn-Cu point contact junction. The next two figures display several sets of data taken for the Sn-Cu and Nb-Cu junctions, respectively (figures 14 and 15). Here, the field dependence is shown for several values below and a bove the critical field, Hc of the superconducting tip. Note that Nb and Sn are type I superconductors with only one critical fiel d. Note the field dependence of conductance of Sn-Cu in figure 14 and that the critical field for Sn has a value extrapolated to absolute zero of Hc(0K) = 309 Gauss and Hc(2K) = 203 Gauss at a temperature of 2 K [36]. The uniqueness of the PCAR probe being integrated with the PPMS also shows up in these two sets of field dependence data The superconducting Nb-Ti solenoid in the commercial PPMS can accurately stabilize a magne tic field with precise control. Using this apparatus, one can investigate effects of a magnetic field on conductance for different fields below and above the critical field of the superconducting tip. In this case, data was taken at 0 Gauss, 100 Gauss, 200 Gauss, 300 Gauss, and 400 Gauss. The second field dependence data set f eatures a Nb-Cu junction experiment (figure 15). The experiment was performed in the PPMS at a stable temperature of 5.00 Kelvin, well below the transition temperature for Nb at 9.50 Kelvin [36]. Once again the distinct features associated w ith the Nb superconducting gap of = 1.53 meV [36] show up clearly. The magnetic field was set at 0 Gauss, 500 Gauss, 2000 Gauss, 5000 Gauss, and 1 Tesla. These fields show varying degrees of suppression of the superconductivity in Nb (figure 15) The overall shape of the dI/dV data look similar for


48 -2.0-1.5-1.0- 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 For Sn: Tc=3.7K = 0.55 mV Hc(2.0K)=295 GT = 2.00K Sn-Cu Junction Normalized Conducta nce vs. Bias VoltageG(V)/GNVoltage (mV) 0 Gauss 100 Gauss 200 Gauss 300 Gauss 400 Gauss Figure 14: Conductance vs. Bias Voltage for an Sn tip and a Cu foil sample establishing a Superconductor-Normal Metal (SC-NM ) junction. The peaks are at the superconducting gap of Tin at = 0.58 meV. Curves are rela tively shifted for clarity. From J. Sanders.


49 -6-4-20246 0.65 0.70 0.75 0.80 0.85 0.90 0.95 1.00 1.05 T = 5K dI/dV vs. V for Nb-Cu Point-Contact JunctiondI/dV Voltage (mV) 0G 500G 2000G 5000G 1Tesla Figure 15. Conductance vs. bias voltage for a Nb tip and a Cu foil sample junction. The peaks are at the superconductin gap of Nb at = 1.53 meV. From J. Sanders.


50 the case of Sn and Nb except for the differen ce in the region close to zero bias voltage. For Nb, the conductance exhibits a shallow dip at V=0. This is ascribed to the presence of an oxide barrier at the junction that can be thought of as an effective junction impedance. Since ideal AR characteristics are seen only for barrier less junctions, even a weak barrier term will cause the dip feature. There are also other features in the data like the “wings” just above the gap voltage. Detail ed analysis of these features is not done here but they may be associated with pr oximity effect at th e SC-NM interface. 3.3 Superconductor-Half Metal (SC-HM) junction Once the initial tests were completed a nd successful data was acquired for a SCNM junction to verify the operation of th e PCAR probe, the highly spin polarized samples of interest could now be investigated. CrO2 is thought to be a half-metallic ferromagnet and is the subject of the majority of the work reported here. A colossal magnetoresistance (CMR) oxide La1-xSrxMnO3 (LSMO) is also of interest to our experiments as it is a predicted half-metal and reseachers have not concurred on the value of its spin polarization as of yet. We ha ve investigated several bulk samples of doped LSMO and SrRuO3 and will report on them in the following chapter. Collaborators at the Argonne National Lab in Illinois have provided high quality thin film LSMO samples for this work.


51 3.3.1 Sn-CrO2 junction The Sn-CrO2 data was acquired after several tria ls with different samples of CrO2 and Pb, Nb, and Sn tips of different diameter s and lengths. Chromium Dioxide is known to develop an oxide layer of antiferromagnetic insulating Cr2O3 on its surface [9] which must be penetrated by the point-contact tip in order to achieve a low junction resistance. The Nb tip is very firm, retaining its sharpness for penetrating this ba rrier, but it also has an oxide layer that has been shown to exist on its surface. The lead tip works very well with many metallic and magnetic samples, but does not easily penetrate the Cr2O3 layer due to its softness, whereas Sn has been the most successful for experiments. A mechanically sharpened Sn tip of a large diameter (0.190”) was used to establish a junc tion with a CrO2 sample obtained from collaborators at NIST. This allowed penetration of the Cr2O3 barrier and acquisition of clean conductance curves. The data shows a clear change in the slope of the I-V curve at low bias voltage (figure 16). The normalized dI/dV is plotted in figur e 17. In contrast to the AR curves for conventional metals, here a sharp dip is observed in the region – to + followed by a nearly flat normal state conductance (GN). This type of curve is reminiscent of a tunnel junction characteristic. Howeve r, it should be emphasized that the junction resistance is very low (< 10 ) and one cannot observe tunneling at such low junction resistance. We will later on argue that this is suppression of AR that results from density of states at the Fermi level for CrO2 influenced by a high degree of spin polarization. The effective junction impedance can be expressed as a di mensionless parameter Z. This Z value indicates the type of contact occu rring at the junc tion [14]. A


52 -6-4-2024 -0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 Current vs. Bias Voltage Sn-CrO2 JunctionT = 2.00 KCurrent (mA)Bias Voltage (mV) Figure 16: Current vs. Voltage for Sn-CrO2 junction. From J. Sanders.


53 -6-4-20246 0.65 0.70 0.75 0.80 0.85 0.90 0.95 1.00 1.05 1.10 Normalized Conductance vs. Bias Voltage Sn-CrO2 junctionG(V)/GNVoltage (mV) Figure 17: Conductance vs. Bias Vo ltage for an Sn tip and a CrO2 film junction. The characteristic dip at low bias represents suppression of Andreev reflection in a half-metal. From J. Sanders.


54 barrierless point contact with no scattering has Z = 0, whereas a pe rfect tunnel junction corresponds to a limit Z As Z increases, AR at lo w voltages is suppressed and the characteristic spikes of a t unnel junction develop at eV = These sharp peaks at are due to the square root singularities in the quasiparticle density of states in superconductors. Determining if Z is finite and large is a stra ightforward observation since the conductance peaks that develop at the gap edges are sensitive to the increase in Z at low temperatures T. This study focuses on point contact junctions where Z is small. The sharp decrease in th e conductance near zero bi as indicates the spin polarization of the CrO2 sample. In an ideal half-metal, the conductance curve would dip to zero conductance at ze ro bias voltage due to the complete absence of possible Cooper pairs and thus no Andreev Reflection. The data in figure 17 does not show full suppression of Andreev reflecti on and thus does not verify the theoretically predicted 100% spin polarization of CrO2. The simple formula ) 1 ( 2 ) 0 ( P G GN can be used to calculate spin polarization P in this case sinc e we have a low Z value. The variables G(0) and GN represent conductance at zero bias volta ge and conductance at high bias voltage, respectively and P is the spin polarization of th e material at the Fermi level. At higher Z values, this simple formula cannot be used and a 2-parameter theoretical model allowing for variation in Z and P must be used to calcu late the P value [14]. We have used such a model developed by our collaborato rs at NRL to obtain best fits to our experimental data. We will present a detailed discussion of theory and fitting in a later section (3.4). The simple calculation yields a value of 66% for this CrO2 sample:


55 ) 1 ( 2 ) 0 ( P G GN ) 1 ( 2 00 1 68 0 P 0.68 = 2 – 2P 2P = 1.32 P = 0.66 = 66% This is much lower than 100% as theoretically predicted for this so -called “Half-metal” and considerably lower than the 90% reported by Soulen, et al [14], but is consistent with the Naval Research Lab results for samples whose surfaces have degraded over time. Many groups [15] are finding this degradation of CrO2 with time, and that the quality of the junction resistances, the e xperimental data, and thus the value of P all change with time after deposition. Accordingly, our several-week-old samples show a lower measured spin polarization. It should be pointed out that even 66% spin polarization is very large compared with spin polarization in conve ntional ferromagnetic metals like Fe, Ni and Co. In these metals, P typically ranges from 20% to 30%. Thus, the use of CrO2 in spintronic devices would certainly lead to better control of c oherent spin states and spin transport. 3.3.2 Temperature dependence The next set of data features the temp erature dependence of dI/dV vs. V for the same Sn-CrO2 junction (figure 18). Th ese data as well as the magnetic field dependence shown in figure 19 are essentially new results where our instrument capabilities have led


56 us to investigate these effects systematically To our knowledge, this PCAR probe is the first to have magnetic field and temperature dependence capabilities. One focus of this work is to achieve a reliable, repeatable t unneling and PCAR probe that can scan a wide range of precise temperatures and external magnetic fields. Other groups are limited to temperatures of 4.2K or 1.6 K by using a liquid Helium bath either with or without pumping [14, 17]. Being able to scan magnetic fields is also of interest in order to investigate the spin precession and spin lif etime effects across the junction [14, 15]. The transition temperature for Sn from its normal state to its superconducting state occurs at 3.722 Kelvin [36] which is in the range of temperat ures in the comparison of data in figure 18. At 2.0 Kelvin, suppr ession of AR is seen at low bias voltage indicating the spin pol arization of the CrO2 as before. The dI/dV – V characteristics were systematically followed as the temperature wa s warmed up from 2K to 4K as shown in figure 18. As expected, the AR suppression f eature (dip at zero bias) tends to become shallower and smeared out as T increases and eventually disappears completely above the superconducting Tc of Sn. This set of data represents the first time that PCAR data has been collected as the junction nature change s from SC-NM to NM-NM. The temperature dependence can be understood and modeled by incorporating standard thermal smearing effects of the superconducting gap function. However, there may also be additional thermal effects associated with dissimilar thermal expansion coefficients of the tip and sample that would alter the contact pre ssure and thus the junction resistance. Careful investigation of the normal state c onductance (data at T=4K ) reveals that the


57-4-2024 0.8 1.0 1.2 1.4 1.6 1.8 2.0 T=4.00K T=3.75K T=3.50K T=3.25K T=3.00K T=2.75K T=2.00KNormalized Conductance vs. Bias Voltage Temperature DependenceG(V)/GnVoltage (mV) Figure 18: Temperature de pendence data for Sn-CrO2 junction. Normalized conductance vs. bias voltage, with the superconducting tran sition temperature of Sn at 3.722K. Data are vertically shifted with a constant valu e of 0.2 for clarity. From J. Sanders


58 conductance is not quite flat and has a V-shape on a ll junctions formed on CrO2. This is also observed by other groups in their measurements on CrO2. The origin of this feature is generally ascribed to strong electron correlation e ffects that lead to a correction in the density of states at the Fermi level for CrO2. Further discussion of this feature will be included in chapter 7 as it is a currently being investigated with our NRL collaborators. 3.3.3 Magnetic Field Dependence The data displayed in figur e 19 is the magnetic field de pendence of the dI/dV vs. V characteristics for the Sn-CrO2 junction. These data were collected at a constant temperature of 2.00 Kelvin (well below the transition temperature of Sn) and with a varying magnetic field. The zerofield data is the same as in figure 19 at 2.0 K, but this time an increasing field is used to suppr ess the superconductivity in the tip. According to Kittel [36], a sufficien tly strong magnetic field will destroy superconductivity. The limiting or critical magnitude of the applied magnetic field for the destruction of superconductivity is denoted by Hc(T) and is also a function of the temperature (figure 20). Cr itical fields are shown versus temperature for several superconductors. Since we are using a superconducting Sn tip it is interesti ng to track the Hc vs. T curve for Sn through our experiments, which have all been performed at 2.00 K. According to figure 20, the critical field of Sn at 2.00 K should be approximately 200 Gauss. This value almost exactly corresponds to an experimentally obtained value of Hc = 203 Gauss from our Sn-Cu field dependen ce experiment (figure 14). However,


59-4-2024 0.8 1.0 1.2 1.4 1.6 1.8 H=1 Tesla H=1000 G H=500 G H=200 G H=100 G H=O G Normalized Conductanc e vs. Bias Voltage Field DependenceG(V)/GnBias Voltage (mV) Figure 19: Field dependence data for Sn-CrO2 junction. Normalized conductance vs. bias voltage. Data relatively shif ted for clarity. From J. Sanders.


60 Figure 20: Critical Fields Hc(T) versus temperature for several superconductors. From Kittel [36]


61 in the case of the Sn-CrO2 junction one sees something diff erent. In fact, the signature due to suppression of AR and thus superc onducting behavior is obs erved well above 200 Gauss. There are clearly observable f eatures at even 500 G and 1000 G. These observations may provide a clue about the eff ect of a highly spin polarized material on the process of Cooper pairing and AR in SC-HM junctions. Since formation of Cooper pairs involves spin-up and spin-down pairing to form a zero angular momentum singlet state, the presence of a half-metal could lead to in teresting consequences on the field dependent characteristics. Future experiments will explore further in the possible new physics showing up in the field-dependent AR data. Junctions of Sn-CrO2 can be set up, cooled to 2.0K, and data can be collected at sma ller intervals from zero Oe up to some field where the Andreev signal is seen to disappear completely. A February 2006 Nature article by Gupta et al has recently shown experimental evidence for triplet state superconducting curre nt in the half-metallic ferromagnet CrO2 [37]. GuptaÂ’s group measured I-V characteristics across a NbTiN-CrO2-NbTiN junction and saw a supercurrent through the chromium di oxide at low bias voltages. This is the first experimental evidence of a supercurrent through a ferromagnetic metal. Our future experiments will possibly collaborate with this very group in order to obtain more samples of high-quality CrO2 films with which to perfor m PCAR experiments. As a cautionary note, one must take care regard ing magnetization of the sample which may lead to a remnant field, making quan titative interpretations difficult.


623.4 Blonder-Tinkham-Klapwijk (BTK) modeling of conductance curves A theoretical model is now required in order to understand the G(V)/GN curves in more detail. A framework for such a m odel has been developed for conventional Andreev reflection at a superconductor – norma l metal (P = 0) junction by G. Blonder, M. Tinkham, and T. Klapwijk [38] and is refe rred to as the BTK model. This model was then modified by I. Mazin and the NRL group and used to fit PCAR data [14]. 3.4.1 BTK theory and and modified BTK model. This BTK model represents the barrier at the junction as a 1-D repulsive potential: ) ( x H U This is meant to represent the effect of a typical native oxide layer at the point contact or the intentiona l oxide layer in a high current-d ensity tunnel junction. The BTK model then goes on to define the Z barrier parameter more explicitly: F F Fv H H k Z 2 Rearranging for H, one arrives at: Z v HF And finally, backsubstitution gives the e xpression for the potential: ) ( x Z v UF And thus the 1-D repulsive poten tial is a simple expression in terms of Planck’s constant, the Fermi velocity, the barrier parameter Z and a Dirac delta function. The NRL group [14] has developed a modi fied BTK model for Andreev reflection in the presence of a spin-polarized metal. First the distinction is made between spin polarization P in general and PT for tunneling and Pc for point contact.


63 Recall from sect. 2.3.4 that the spin polarizati on P in general for a material is defined as Where N(E) is the spin-dependent density of st ates. For the case of P = 0 and P = 100%, the definition of P is not critical. However, for intermediate values more careful consideration must be given to the type of experiment being performed. The above expression for spin polarization is nearly impo ssible to obtain in a transport experiment, yet transport is to date the only viable wa y to obtain the needed energy resolution. Experiments of Tedrow and Meservey use t unneling methods to determine P, which is thus more accurately described as tunneling polarization PT: Where T are spin-dependent tunne ling matrix elements. These matrix elements are determined by overlap of wave functions at the interface and should generally differ for spin-up and spin-down bands [10]. For the point-contact experime nts in this study a preferred expression is th e contact polarization, Pc. This is due to the fact that pointcontact experiments have neglig ible interfacial scattering. The junctions have a clean contact and low resistance. The value of PC shall be defined as: where vF is the Fermi velocity of the respective band.


64 The appearance of vF in this expression is expected for a point contact and leads to the expression: because I vF N(EF). These different but necessary definitions of P are useful, and PC is used here in the modified BTK model. In this model for point contact experiments, one assumes small Z values due to low resistan ce contacts. For these purposes consider the current into the point contact as d ecomposed into spin-up and spin-down or unpolarized current a nd polarized current. where the unpolarized current, Iunpol, carries no P and obeys the conventional BTK theory. The remaining current, Ipol, carries all of P and as such is entirely quasiparticle current (since supercurrent can carry no net polarization si nce it is composed of Cooper pairs). This current can be calculated only by allowing nonAR processes at the point contact. Within the BTK model this procedure amounts to setting the Andreev coefficient, A(E) to zero and renormalizing al l of the remaining processes to 1 for current conversion. PC can be extracted from the conducta nce curves by noting that the total conductance is the sum of the polarized conductance and the unpolarized conductance: unpol C pol CI dV d P I dV d P dV dI ) 1 (


65 If the interfacial scattering is minimal (Z 0), then for eV << and kBT << (where kB is BoltzmannÂ’s constant) the term 2 1 unpol nI dV d G and 0 polI dV d substitution yields: ] 2 ][ 1 [ ] 0 [n C CG P P dV dI But since eV<< and is very close to zero, we can call dI/dV as G(0), and thus: n CG P G ) 1 ( 2 ) 0 ( or as previously used in section 4.3. 1 to calculate the spin polarization: ) 1 ( 2 ) 0 (C nP G G a result for the extreme limit when Z=0 and no tunneling peaks appear in the conductance curve. Under these restricti ons, calculating P is straightfo rward using this technique as previously shown. 3.4.2 Modified BTK fit to conductance curves When these strict conditions are relaxed to include all families of curves between Z = 0 and Z and a finite T, a numerical fit ting procedure over the entire voltage range of the experiment is used. This fit was obtained from the NRL modified BTK model [14] and is used as an iterative program. The program requests the user to input static values for contact resistance Rs, temperature T in Kelvin, and superconducting gap Secondly, the user must input


66 columns of data for current, voltage, and conduc tance. Finally, the program requests an initial guess of the barrier parameter Z and the spin polarization P (actually PC), and the program fits a curve to the experimental data. The user simply varies the Z and P values until she/he visually finds the best-fit curve. This exact fitting procedure has been used for the following Sn-CrO2 data (figure 21). The superconducting Tin tip has a value of 0.55 mV, th e temperature of the experiment was 1.75 K, and the contact resistance was 1.8 These three parameters were input into the program and held static during the iterations. The conductance vs. bias voltage data points were then input and can be seen as open circles in figure 21. Finally, a Z parameter of 0.15 and a spin polarization of PC = 0.7 are the final outputs. This indicates a spin polarization of PC = 70% which is consistent with the simplified (Z=0) calculation of PC = 66% in section 3.3.1. 3.5 Ferromagnetic Metal – Superco nductor (FM-SC) junction The next experiment to be f eatured in this study focused on a Nickel wire tip pressed against an MgB2 sample. MgB2 is a unique superconducting material with a high Tc and a unique gap structure. In our experiments, we used a pressed pellet sample of MgB2. The sample was formed into a pellet under high temperature and pressure. 3.5.1 Ni-MgB2 junction Magnesium Diboride is an intermetalli c compound that has been known since


67 -6-5-4-3-2-10123456 0.6 0.8 1.0 1.2 Normalized Conductance vs. Bias VoltageP = 0.7 Z = 0.15 RS = 1.8 = 0.55 mV T = 1.75 KSn-CrO2 experiment BTK fitG (V)/GNBias Voltage (gap units) Figure 21: Normalized c onductance vs. bias voltage for an Sn tip and a CrO2 sample establishing a Superconductor – Half Metal (SC-HM) junction. The modified BTK model was used to fit the experi mental data. From J. Sanders


68 January 2001 to have a superconducting transi tion temperature around 40K, which almost doubled the previous 23K record [39]. Not only has its high Tc created interest, but also its two superconducting gaps, upper (HC2) and lower (HC1) critical magnetic fields, and especially its selec tive coupling between sp ecific electronic stat es and specific phonons [35]. The field dependence of conducta nce vs. bias voltage for Ni-MgB2 is shown in figure 22. Evidence can be seen for the tw o superconducting gaps based on the different peaks in the conductance spectrum. The two gaps have been previously found to be 1 = 3 mV and 2 = 7 mV and conductance spectra of ten interestingly reveal peaks at not only these gap valu es but at locations like and to name just a few. It is quite remarkable that the AR experiments directly exhibit the two-gap nature of this superconductor. The ga p features seen in our AR experiment occur at fairly large bias voltages. This is to be expected as in the PCAR experiment, the energy dispersion at the band stru cture is probed as an averaged quantity. So it is likely that higher multiples of the two-gap combinations are being observed here.


69-80-60-40-20020406080 0.0000 0.0002 0.0004 0.0006 0.0008 0.0010 0.0012 0.0014 0.0016 0.0018 0.0020 0.0022 T = 10K Conductance vs. Bias Voltage Ni-MgB2 JunctiondI/dVVoltage (mV) 0 Gauss 1500 Gauss 9000 Gauss Figure 22: Normalized c onductance vs. bias voltage for a Ni tip and a MgB2 sample establishing a Ferromagnetic Meta l Superconductor (FM-SC) junc tion. From J. Sanders.


70 Chapter 4 PCAR on Correlated Oxides SrRuO3 and La1-x(Ca, Ba, Sr)xMnO3 Strontium Ruthenate (SrRuO3) and La1-x(Ca, Ba, Sr)xMnO3 are interesting materials that show half-metallic behavior and thus show potential for spintronics applications. Samples of these series have been acquired from Argon National Lab and NRLWashington. Experimental results have shown suppression of Andreev Reflection, verifying their half-metallicity. 4.1 Strontium Ruthenate SrRuO3 Point-contact Andreev reflection (PCAR) measurements were made on bulk polycrystalline SrRu0.8Ti0.2O3 and SrRu0.92O3 samples to determine the transport spin polarization. The parent compound SrRuO3 undergoes ferromagnetic ordering at TC ~ 160K with relatively high spin polarization (~60%). For the SrRu0.8Ti0.2O3 and SrRu0.92O3 samples, a reduction in TC occurs. Moreover, our measurements indicate that the SrRu0.8Ti0.2O3 system retains a high degree of spin polarization (P~0.6), similar to the parent compound. However, due to the insulating nature of SrRu0.92O3 the PCAR technique did not yield information on P. Instead, the data displayed nonlinear I-V consistent with the tunneling-like spectra.


71 4.1.1 Properties of SrRuO3 One of the most interesting itinerant ferromagnetic systems studied both theoretically and experimentally is SrRuO3. This ferromagnet transition-metal oxide is very promising for several applications due to its high magnetic moment (m~1.6 B/Ru), chemical stability, and excellent interface qualities with Al2O3. [40, 41]. SrRuO3 has a distorted perovskite structure, which is the same as that of LaMnO3. Measurements of thin films, polycrystalline bulk samples, and single crystals yields a Curie temperature of TC~160K, [40-43], although the TC measured on very thin films (~10nm) was considerably lower. Importantly, the hybridization between the Ru 4d orbitals and the O 2p orbitals gives rise to the ferromagnetism in SrRuO3 [44, 45]. If this hybridization changes substantially, either increasing or decreasing, it would lead to changes in both the magnetic and electrical properties. Site substitution of the Ru in the SrRuO3 system results in diminishing of the ferromagnetic ordering. It was shown that sythesis of this material under high pressure causes a random distribution of Ru vacancies, which forms SrRu1-vO3. [42]. Depending upon the amount of vacancies in the material, TC decreases significantly. Further, it was shown that this decrease is from ~160 to ~90 K for vacancies of v=0-0.08. Structural disorder and an increase in the Ru valence were the key ingredients in such a decrease in the ordering temperature in that study. Also, it has been seen that site substitution over the full range (0

72 ferromagnetic to an insulating paramagnetic phase, which is accompanied by a lowtemperature resistivity several orders of magnitude higher than that for the parent compound [47]. Details of the synthesis of our polycrystalline samples used in this study can be found elsewhere [42]. Both dc magnetic su sceptibility and rf transverse susceptibility measurements (not presented here) show that the SrRu0.92O3 sample has a TC~75K, while that of the SrRu0.8Ti0.2O3 showed a reduced TC of ~65K. The dc resistivity of SrRuO3, SrRu0.8Ti0.2O3, and SrRu0.92O3 are shown in Fig. 23. All three samples show metallic behavior at room temperature. Interestingly, the nonstoichiometric samples have lower resistivity at room temperature. However, for the stoichiometric compound, the metallic behavior persists down to low temperatures. In contrast, as seen in the plot, the resistivity for SrRu0.92O3 increases as the temperature is lowered with the increase being rather steep below 50K. This clearly shows the effect of disorder on the electrical conductivity. Measurements of the magnetoresisitance of these samples from TC down to low temperatures as seen in Ref. 42, signifi es the presence of magnetic field dependent scattering at the grain boundaries as well as a high degree of spin polarization at EF. 4.1.2 PCAR measurements on SrRu0.8Ti0.2O3 We have measured Sn/SrRu0.8Ti0.2O3 junctions using PCAR at T=2.0K. The PCAR data were collected with our homebuilt system. Figure 24 shows a plot of three spectra for a single point contact of this j unction as a function of applied magnetic field, Ha. Several spectra were collected but in this figure we only plot three for clarity. The


73 Figure 23. Resistivity vs. T for SrRuO3 annealed in Ar at 1100oC, SrRu0.8Ti0.2O3 annealed in 600 atm of O2 at 1100oC, and SrRu0.92O3 made from SrRuO3 annealed in 600 atm of O2 at 1100oC. The data show that the sample with Ru defects becomes more insulating at low temperatures compared to the other two samples. From J. Sanders et al [19].


74 Figure 24. PCAR spectra of a single point contact of Sn/SrRu0.8Ti0.2O3 junction. The spectrum was taken with Ha=250 Oe shows normal state conductance. From J. Sanders [19].


75 bottom spectrum was taken at Ha = 0. Several junctions were made in situ to check for reproducibility and also to achieve junctions that yield the best overall data to analyze. This is necessary with PCAR as the point-contact area is undefined and care should be taken to prepare the surface as well as penetrate any “dead” surface layer to realize good quality junctions [18]. We were able to achieve a low barrier, clean spectrum which clearly shows the suppression of Andreev Re flection. Above the critical field of Sn (~210 Oe) at T=2.0K, the superconductivity is suppressed and we obtain the normal-state conductance representing the single-particle spectrum. This approach, measuring at the lowest temperature and using the applied field to achieve the normal state, would allow a careful study of the metal-insulator transition in these materials. The PCAR spectra at Ha=0 were analyzed using the modified BTK model [48]. The fitting parameters used within this model are the barrier strength (Z), and the spin polarization (Pc). All of the other parameters such as the serial resistance (Rs) of the sample, temperature, and the superconducting gap ( ) were not varied in the fitting routine but rather remained fixed throughout the fitting procedure of the spectrum. Displayed in Fig. 25 is an analyzed conductance curve taken at Ha=0. Several of the Ha=0 spectrum were taken and all the fits agreed very well. The results from the fitting procedure yielded a value of Pc=0.61. As indicated, Z is very small, which is indication of a clean junction. Also given is the experimentally determined junction resistance Rj=2.4 which is a reasonable value for a clean junction. The values obtained from our analysis are consistent to that obtained by Na dgorny et al [49] in their investigation of


76 Figure 25. An analyzed PCAR spectrum taken of Sn/SrRu0.8Ti0.2O3 with zero field. The fitting parameters are listed in the figure. From J. Sanders [19].


77 thin-film samples of SrRuO3. The highest value reported in Ref. 49 was Pc =0.61 with a barrier strength of Z=0.17. However, in a recent study by Raychaudhuri et al [50] on epitaxial thin films of SrRuO3, results for Pc were no higher than 0.51. Moreover, the authors in Ref. 50 relied on an empirical extrapolation to Z=0 to acquire their upper bound for Pc. This was the motivation behind performing several trials to obtain a low Z Sn/SrRu0.8Ti0.2O3 junction in this work. Our aim was to collect the conductance with a clean, low barrier junction to achieve an upper bound without justifying the appropriateness of an extrapolation to Z=0. According to several studies, Pc decreases with increasing Z, which is the result of po ssible spin-flip scattering at the contact [18, 50, 51]. The spin polarization obtained in our measurement is related to the density of states and Fermi velocity v of each spin by the following equation as shown previously: Where n=0 calculates P for photoemission, n=1 and 2 calculates the transport spin polarization (our measurement) for ballistic and diffusive transports, respectively. One of the points to make regarding the spin polarization in SrRuO3 is that the value of Pc is mainly attributed to the difference in Fermi velocity between the majority and minority spin carriers, due to the fact that the density of states for both the minority and majority spin states is N(EF) ~ 23 st/Ry [45]. Further, the band structure calculations in Ref. 45 also show that the difference in spin current is negative, though PCAR measures the magnitude of the spin polarization. The important aspect is that the PCAR measurements


78 of the Sn/SrRu0.8Ti0.2O3 junction, with minimal Z, yields a relatively high spin polarization, albeit the magnetic properties have changed significantly. As noted earlier, the hybridization is probably weakened, and while the global ferromagnetic order is affected through the suppression of TC, the magnitude of the difference in spin current is hardly affected, thus accounting for our large Pc values obtained in our experiments. This is an important observation because it indicates that this material can be chemically tuned to realize better junctions for spintronic devices and yet the vital property of spin polarization remained unaffected. 4.1.3 PCAR measurements on SrRu0.92O3 We also measured the PCAR spectra of a Sn/SrRu0.92O3 junction. Plotted in figure 26 is a PCAR spectrum of a Sn/SrRu0.92O3 junction taken at T=2.0K. As was the case for the Sn/SrRu0.8Ti0.2O3 junction, we performed several trials to achieve the best overall quality junction, which is presented in this work. Unfortunately, the data cannot be analyzed with the modified BTK model due to a sharply rising background above 4 mV. This is not surprising due to the high resistivity of the sample at 2K (~11m cm). While the data look reminiscent of tunneling conductance, one has to use caution in interpreting features directly with the tunneling density of states. For example, the contact resistance is not in the kilohm or megohm range but rather around 23 These types of curves generally associated with nonlinear I-V characteristics have been observed in many oxides and are difficult to interpret due to several possible origins such as


79 Figure 26. PCAR spectrum of a Sn/SrRu0.92O3 junction. Plotted is the best overall junction for this sample. Data were taken at T=2.0K. The junction resistance is Rj=23.0 From J. Sanders et al [19].


80 parallel shorts and tunneling channels, surface states, charging in the junction region, etc. We do not attempt to subscribe to any one view of this case. Rather, we can certainly point out an empirical observation from expe rience that if one is unable to achieve junction resistance considerably below 10 conducting PCAR analysis and extraction of spin polarization would lead to ambiguous results [52]. 4.1.4 SrRu0.9Mn0.1O3 and SrRu0.9Cr0.1O3 Point contact Andreev reflection measurements (PCAR) were made on bulk polycrystalline SrRu0.9Mn0.1O3 and SrRu0.9Cr0.1O3 samples to determine the transport spin polarization. For both of these samples, a reduction in TC occurs relative to TC ~ 160 K for the parent compound SrRuO3. For both the SrRu0.9Mn0.1O3 and SrRu0.9Cr0.1O3 samples, suppression of Andreev reflection is present but due to the rising backgrounds for bias voltages outside the superconducting gap voltage, a value of Pt could not be obtained with a high degree of certainty. Work has been done on the subtraction of the V1/2 background which comes from the density of states calculation. This work will be described in chapter seven. In this section, we present the results from our studies of SrRu0.9Mn0.1O3 and SrRu0.9Cr0.1O3 using PCAR, magnetic susceptibility and zero-field resistivity. Herein, we investigate the effects of transition metal (TM) doping and thus the role of both disorder and correlation effects on the transport and magnetic properties of the parent compound.


81 As mentioned earlier, the electronic structure at the Fermi energy of SrRuO3 consists of spatially extended Ru 4d (t2g) orbitals with bandwidth W (W > U), where U is the strength of the on-site Coulomb energy, which is a relevant parameter in band structure calculations of strongly correlated oxides. Importantly, SrRuO3 is highly susceptible to disorder due to the fact that the carrier mean free path is on the order of an inter-atomic spacing. Once more, this disorder can be introduced into the lattice by a TM. Without disorder, the bandwidth W is a maximum. When the TM concentration increases; W will decrease and the on-site Coulomb energy increases, which is due to the inclusion of the more highly localized (less extended) 3d orbitals. The reduction of W will in turn reduce the Ru-Ru magnetic exchange coupling, which is to say that the RuRu separation increases. (Note that the hybridization of the Ru-O bands gives rise to ferromagnetism in this system, and thus the O contributes substantially to the moment.) Our main objective was to investigate how the disorder affects the magnetic and electronic properties, along with the transport spin polarization. Polycrystalline bulk SrRu0.8Mn0.2O3 samples were synthesized by collaborators from stoichiometric mixtures of SrCO3, RuO2, and MnO2. Samples were processed using the solid state reaction method and fired in air several times at various temperatures up to 1300 oC followed by natural furnace cooling. Under these conditions, no single-phase perovskite samples could be obtained. The single-phase samples were synthesized by firing twice with intermittent grinding in 600 atm. of O2 at 1100 oC. To investigate how the disorder effects the spin-dependent transport properties of our samples, we performed PCAR measurements on each using Sn as our


82 -6-4-20246 6.2 6.4 6.6 6.8 7.0 7.2 7.4 7.6 Sn-SrRu0.9Mn0.1O3 Junction Conductance vs. Bias VoltageConductanceBias Voltage (mV) O Oe 25 Oe 50 Oe 75 Oe 100 Oe 125 Oe 150 Oe 175 Oe 200 Oe 225 Oe 300 Oe 500 Oe Figure 27: PCAR spectrum for an SnSrRu0.9Mn0.1O3 junction as a function of applied magnetic field. From J. Sanders.


83 superconducting probe for temperatures ranging 2.0-4.0 K and for applied magnetic fields ranging 0-500 Oe. Fig. 27 displays PCAR spectra for a Sn-SrRu0.9Mn0.1O3 junction for a single point contact (single esta blished junction) as a function of applied magnetic field, Ha. Several junctions were made for each sample to achieve junctions that yield the best overall data to analyze. This is necessary with PCAR as the point contact area is undefined and care should be taken to prepare the surface as well as penetrate any “dead” surface layer to realize good quality junctions [18]. We were able to achieve a low barrier, clean spectrum which clearly shows the suppression of Andreev reflection. At T = 2.0 K, we established a critical field value between 200-250 Oe, in which case the superconductivity is suppressed and we obtain the normal state conductance for the sample. We plot more spectra for low fields for the Mn sample to illustrate that very small changes occur well below 200 Oe. Importantly, our results in Fig. 27 show that the transition to the normal state is not continuous. Further, these junctions are a result of a coherent process, which can essentially be controlled by an applied magnetic field. Note that all samples reported here were zero-field cooled and that the coercive field (Hcoer) obtained from Magnetization vs. Applied field (M-H) data far exceeded Ha. Typically, Hcoer was measured to be approximately 2.0 kOe for all samples. In Figure 28 we plot the relative dependence of the superconducting gap as a function of T for the Mn sample. Our G (V) spectra agree well with expectations on approaching the critical temperature of the superconductor, where the energy gap is disappearing along with suppression of


84 -6-4-20246 6.0 6.2 6.4 6.6 6.8 7.0 7.2 7.4 Temperature Dependence Sn-SrRu0.9Mn0.1O3 Junction Conductance vs. Bias VoltageConductanceBias Voltage (mV) 2.00 K 2.25 K 2.50 K 2.75 K 3.00 K 3.70 K 4.00 K Figure 28: PCAR spectrum dI/dV vs. V as a function of temperature for an SnSrRu0.9Mn0.1O3 junction. From J. Sanders.


85 Andreev reflection. Unlike the field dependence spectra, these curves show a continuous transition to the normal state. Figure 29 plots the G (V) curves for the T = 2.0 K and Ha = 0 for the SrRu0.9Mn0.1O3 sample along with the SrRu0.9Cr0.1O3 sample which has the same experimental parameters and a comparable junction resistance Rj (~ 2 ). We attempted to normalize the data points at 4 mV to compare the amount of Andreev reflection (AR) suppression at zero-bias. We can say that the SrRu0.9Mn0.1O3 sample appears to be more spin-polarized due to the extent in which AR suppression occurs. While recording the data, it was noted that the suppression, or dip in the G (V) curve at zero bias, was much more pronounced for SrRu0.9Mn0.1O3. Both samples however display more AR type features than the SrRu0.92O3 sample shown in Fig. 26. The shape of the G (V) curve for this sample is not to surprising due to the high resistivity of the sample at 2 K (~ 11 m cm). In any event, determining Pt from the data in Fig. 4 using existing models would not yield confident results. Recent results however from this system reveal that the magnetoresistance from TC down to low temperatures signifies the presence of magnetic field dependent scattering at the grain boundaries as well as a high degree of spin polarization at EF. Unfortunately, this cannot be confirmed by PCAR. Referring to Fig. 29 again, the strong broadening of the DOS for V > /e is not unusual in these types of oxides. It is well known that Coulomb interactions increase in the presence of disorder. The spectra, aside from that in Fig. 29, display no evidence of thermal broadening due to local heating at the contact region but display fine spectral resolution,


86 Figure 29 PCAR spectra of SrRu0.9Mn0.1O3 and SrRu0.9Cr0.1O3 junctions at Ha=0. The data were normalized for bias voltages V> to compare. At zero bias, it was noted when taking the data that the AR suppression was larger for SrRu0.9Mn0.1O3. From J. Sanders. -4-3-2-101234 0.82 0.84 0.86 0.88 0.90 0.92 0.94 0.96 0.98 1.00 0.86 0.88 0.90 0.92 0.94 0.96 0.98 1.00 1.02 SRMO/Sn junction SRCO/Sn junction conductancebias voltage (mV)T = 2.0 K Rj ~ 2.0 SRMO SRCONormalized conductancebias voltage (mV)


87 as one normally obtains with low resistance junctions [31]. Though the spectra in Fig. 29 shows behavior characteristic of AR suppression, analyzing it with existing models would lead to ambiguity in Pt, particularly if the fitting procedure involves several fitting parameters. In summary, we presented PCAR data of polycrystalline bulk samples of SRO, SRMO, SRCO. Both SRMO and SRCO displa yed coherent AR suppression but analysis could not be done due to the rising background outside of the gap voltage of the superconductor. Further studies of the electronic structure of SrRuO3, considering more substitutions for Ru should be undertaken in order to understand the robust nature of the spin polarization in this ruthenate series. 4.2 Lanthinum Manganate series: La1-x(Ba, Ca, Sr)xMnO3 Another classic example of an exciting spintronic oxide is the lanthanum manganite system which is known to exhibit “half metallic” behavior and colossal magnetoresistance (CMR). We present point-contact-Andreev-reflection (PCAR) results for polycrystalline samples of the ferromagnetic manganite series La1 xAxMnO3 (A = Ba, Ca, Sr). We analyzed the suppression of Andreev reflection from conductance-vs.voltage data within the superconducting gap of Sn ( [0] = 0:59 meV) to determine an intrinsic value of the spin polarization for each sample. The average value of the spin polarization for these samples was 0.64. The data were analyzed, with careful error propagation, using Mazin et al.'s modifcation of the Blonder-Tinkham-Klapwijk theory.

PAGE 100

88 Our results were compared to theoretical predictions based on calculations on La0.7Sr0.3MnO3 (LSMO) from the literature. 4.2.1 Properties of La1-x(Ca, Ba, Sr)xMnO3 The transition-metal-oxide series La1 xAxMnO3 (A = Ba; Ca; Sr) has been an attractive system due to the strong coupling of magnetic, electrical, and structural properties [53]. These materials possess interesting transport properties, which are responsible for such effects as colossal magnetoresistance (CMR), and are predicted to have a high degree of spin polarization [54]. When substitution is made for the rare-earth element La with elements such as Ca, Ba or Sr, ferromagnetism arises due to the conversion of Mn3+ into Mn4+. The coupling of these ions is described by a doubleexchange mechanism proposed by Zener [55] and de Gennes [54]. Double-exchange alone does not describe the transport properties in these systems [56-59]. Due to a strong electron-phonon coupling, the carriers are polarons [60], which above Tc are magnetic [61] and self-trapped to the lattice. The ferromagnetic state is accompanied by a freeing of the polarons [62]. Importantly, influences such as grain sizes in polycrystalline samples of the manganites modify the coupling mechanism. Magnetism and the subtle coupling with the electronic and structural properties have been investigated extensively using a variety of experimental probes. [63-65]. Nadgorny et al [65], using PCAR, have determined that the value of the transport spin polarization Pc (denoting the spin polarization measured using PCAR) for La0.7Sr0.3MnO3 ranges from roughly 0.6 for low-residual-resistivity films to 0.9 for

PAGE 101

89 higher-residual-resistivity films, where the latte r is more in agreement with theoretical predictions for diffusive transport [54, 65]. The resistivity of their films was varied using ion-beam damaging techniques. They determined that as the disorder increased, along with the resistivity, the minority d-electrons became more localized. Thus, the electrical conductivity maintained its metallic nature (albeit with increasing residual resistivity consistent with elastic scattering introduced by the ion-beam irradiation), but the spin polarization of conduction electrons was enhanced. In this section, we present a systematic study of the manganites La0.6Ba0.4MnO3 (LBMO), La0.58Ba0.42MnO3 (LBMO-ORD), La0.62Ca0.38MnO3 (LCMO), and La0.6Sr0.4MnO3 (LSMO) using point-contact Andreev reflection (PCAR). Our spectra were obtained below the superconducting transition of Sn (Tc ~ 3.7 K), which was used as the superconducting counterelectrode. This allowed us to probe the polarization of the conduction electrons for each sample. Polycrystalline samples were synthesized from stoichiometric mixtures of MnO2, CaCO3, SrCO3, BaCO3, and pre-fired La2O3. Initial calcinations were done in air at 1000oC and were followed by several grindings, pressing into pellets, and firing for 24 hours in air at increasing temperatures up to 1400oC. La1 xCaxMnO3 with 0 < x < 0.5 and La1 ySryMnO3 with 0 < y < 0.5 can be easily synthesized under normal conditions in the form of perovskite materials. Stoichiometric compositions near optimum doping for the highest ferromagnetic transition temperature at x = 0.38 and y = 0.40 were obtained during final firing in air at 1440oC followed by slow cooling to room temperature. Perovskite samples with similar compositions of Ba, which has a larger ionic size, are

PAGE 102

90 much more difficult to stabilize. For this work, La1 zBazMnO3 (z = 0.40 and 0.42) perovskite samples were obtained using a two-step method previously developed for similar metastable compounds. Synthesis of the randomly substituted La1 zBazMnO3 (z = 0.40) system was done at high temperatures (1300-1400oC) using Ar flow to achieve the oxygen-deficient compositions O2.7-2.8 followed by the complete oxygenation (in O2 flow) at a temperature low enough (below 500oC) that the cations do not diffuse. All samples were found to be single-phase from x -ray and neutron powder diffraction data. Figure 30 shows the magnetization and resistivity data for the samples studied in this work. For each sample, the insulator-metal transition coincided approximately with the paramagnetic-to-ferromagnetic phase transition, consistent with other studies of these materials. The compounds investigated here encompass a wide range of ionic sizes at approximately fixed doping level z = 0.40 (fixed ratio of the Mn3+ and Mn4+). The effect of ionic sizes and associated structural f eatures on ferromagnetic transition temperatures was previously investigated experimentally in great detail. Rodriguez-Martinez and Attfield [66] have shown that Tc for manganites at fixed doping level z = 0.30 is a function of the ionic sizes of the A-site cations of the ABO3 perovskite. According to these results, the highest values for Tc appear for the least distorted structures. One of the goals of the present research is to learn if similar relationships can be found between structural features and the spin polarization Pc for a group of optimally doped perovskite manganites.

PAGE 103

91 Figure 30 (a) Resistivity vs. T for the samples in this study. (b) Magnetization vs. T for the samples in this study. The two figure show that the insulator-metal transition coincides with the ferromagnetic-paramagnetic phase transition. From J. Sanders.

PAGE 104

92 4.2.2 PCAR measurements on La1-x(Ca, Ba, Sr)xMnO3 To investigate the spin polarization of the conduction electrons in these materials, we performed PCAR experiments on each sample. In Fig. 31, we show one representative conductance curve which is in this case a Sn-LBMO junction. In Fig. 32, we have shown the full bipolar scans of the conductance spectra in the vicinity of the superconducting gap of Sn ( [T = 0] = 0.59 meV) for one representative point-contact junction. Our PCAR insert integrated with the PPMS allows us to take a series of such spectra at closely spaced fixed temperatures across the superconducting transition into the normal state. Thermal effects and temperature suppression of the order parameter on Pc can be explored in detail with our measurements. However, in this work, for the purpose of conducting a comparative analysis of Pc in different manganite systems and modeling the data, we restrict our discussions to the PCAR data collected at around 2-2.5 K, which is the lowest temperature attainable in our system. We extract transport polarization, again using the modified BTK model of Mazin, Golubov, and Nadgorny [48], figure 33 plots experimental dI/dV data against fits to the model. The best-fit values for polarization were 0.63 (ordered LBMO at 2K), 0.64 (disordered LBMO at 2.3K), 0.65 (LSMO at 2.5K), and 0.62 (LCMO at 2K). Since the value of Pc is related to the density of states at the Fermi energy, we compared our results with some theoretical calculations on a similar system. Bandstructure calculations [67] on La1 xCaxMnO3 for the intermediate region x = 1/3 show that the Ca and La ions contribute valence electrons to the Mn-O states. Further, these calculations show that both Ca and La merely act as spectator ions with no states at or

PAGE 105

93 -4-3-2-101234 0.75 0.80 0.85 0.90 0.95 1.00 1.05 T = 2.35K Sn-LBMO junction (disordered) Normalized C onductance vs. Bias VoltageG(V)/GnVoltage (mV) Figure 31 PCAR spectra for an Sn-LBMO (disordered) junction. From J. Sanders.

PAGE 106

94 Figure 32: PCAR spectra of a Sn-LCMO junc tion as a function of temperature. The spectra are shown from 2.0 K to above the 3.7 K critical temperature of Sn. A vertical shift has been included for clarity. From J. Sanders.

PAGE 107

95 Figure 33: Best fit curves for the G(V) spect ra of LBMO and LSMO. Part (a) plots G(V) vs. V for the disordered LBMO sample at 2.3 K and the best fit to MazinÂ’s model. Part (b) plots G(V) vs. V for the LSMO sample at 2.5 K. From J. Sanders.

PAGE 108

96 below the Fermi energy. Interestingly, the theoretical calculations of LCMO predict a spin polarization of P0 = 0.36 for the bulk (n = 0 from [53]). Spin-polarized photoemission results from Park et al [68] however show that majority spin states occupy the Fermi energy below Tc with no measurable presence of minority states. This technique probes only a few Angstroms of the top surface layer, and thus it is possible that the surface of LSMO is half-metallic contrary to its theoretical bulk value. The equation for spin polarization, P (see pg. 68) can be used to estimate both the ballistic and diffusive values for the transport spin polarization Pc, which is relevant to our experiment. Using the theoretical values [67] for the densities of states Nup(EF) = 0.58 states/(eV Mn ion) and Ndown(EF) = 0.27 states/(eV Mn ion), and Fermi velocities vFup = 7.4 x 105 m/s and vFdown = 2.2 x 105 m/s, we obtain P1 = 0.74 and P2 = 0.92 from [53]. Comparing our results (Pavg = 0.64) with those of band theory, we seem to agree more with the ballistic rather than the diffusive value of polarization [69]. Our values are consistent with the ballistic values obtained from LSMO spectra seen in other works [62, 68]. There are several fundamental reasons that could explain why our values consistently fall below the theoretical prediction. In a PCAR experiment, the (intrinsic) spin polarization is not necessarily always what is measured at the contact. It is possible that mechanisms such as spin-flip scattering and localization can have a signicant effect on the measured value of Pc. For the case of spin-flip scattering, it has been reported [37] in CrO2 that surface effects such as tunnel barriers formed from degradation of CrO2 into

PAGE 109

97 Cr2O3, a well-known insulator and the more stable of the chromium oxides, may reduce Pc and increase Z. We observed no real direct evidence of a tunnel barrier reducing the spin polarization in our measurements. This coul d be attributed to the fact that in pointcontact junctions, we can penetrate any “dead" surface oxide layer and achieve good contact with the ferromagnetic material. Further, since resistivity is high in our samples, we would initially assume that the mean free path L was very small, which would result in strong diffusive scattering at the interface. Nonetheless, the ballistic model fits the data well (Fig. 33). While this appears to be somewhat counterintuitive at first glance, it is not surprising given the electronic nature of many of these perovskite oxides. One real distinction between our PCAR study and that of Nadgorny et al [65] comes from the mechanism responsible for high resistivity in the samples. Those samples studied by Nadgorny et al. that displayed higher resistiv ity did so due to defects produced by iondamaging the films, as confirmed by spectroscopic analysis. In contrast, the high resistivity for all four samples in this work can be attributed to grain boundaries. It is also important to discuss the effects of thermal broadening that could potentially appear in PCAR spectra. Auth et al [52] attribute loss of coherence peaks in PCAR measurements to local heating in the contact area, although, as already mentioned, their several parameter model of thermal broadening makes it difficult to extract definitive values for polarization. Other effects, such as magnetic scattering at the interface [70] could be mistaken for thermal broadening. Our contacts showed no indication of these effects and thus no eff ective temperature or modified superconducting gap value was needed to analyze the PCAR data in this work.

PAGE 110

98 In summary, we have measured the transport spin polarization using the PCAR technique of well characterized polycrystalline samples of the La1 xAxMnO3 (A = Ba, Ca, Sr) series for the intermediate range of x. The average spin polarization obtained from our PCAR data was 0.64. Thus, our results indicate that the value of Pc is not affected by the A-site cation significantly. We successfully analyzed all experimental PCAR spectra with the ballistic model. More experimental work is underway in our laboratory to understand fully the spin-dependent transport properties which directly probe the electronic structure of these fascinating materials.

PAGE 111

99 Chapter 5 Carbon nanotube growth and applications 5.1 Carbon nanotube growth Carbon nanotubes have been grown by plasma enhanced chemical vapor deposition (PECVD) through a collaboration with the lab of Dr. Rudy Schlaf in the USF Electrical Engineering department. The details of this process are described in detail in Z. F. Ren [69], and will be briefly summa rized here. PECVD growth was done using both magnetic nanoparticles and magnetic thin films as a catalyst. The reactive gases used in this growth were acetylene (C2H2) and methane (CH4) with nitrogen as a purge gas. Growth occurred at a sample temperature of 600oC and between two electrodes establishing a 400 Volt plasma. The tubes grown from a Ni thin film catal yst have been imaged and are shown in figure. This SEM image shows vertically a ligned forests of nanotubes, each having a Ni nanoparticle at its top. One i mmediate question is if these magnetic catalyst dots attached to the nanotubes could be used to ma nipulate and align the nanotubes. Ongoing experiments are being performed to system atically remove the CNTÂ’s from their substrates and align and texture them using ma gnetic fields, electric fields, or Langmuir Blodgett (LB) techniques. To date, we have failed to achieve any high degree of alignment using simple procedures based on static magnetic fields.

PAGE 112

100 Figure 34. SEM image of PECVD grown multiwalled carbon nanotubes. From J. Schumacher and J. Sanders.

PAGE 113

101 The next portion of this work was perfor med in the electronic materials division at NRL-Washington in the labor atory of Dr. Eric Snow, sec tion head of Nanotechnology. His group works with carbon nanotubes and thei r sensor applications [70-74]. Figure 35 shows a representative AFM and SEM image of their CVD technique [70]. This process is very reproducible and is th e backbone of all of our sens or applications. The process involves preparing a catalyst, cleaning the substrate, depositing the catalyst on the substrate, loading the sample in the furnace, purging the quartz tube within the furnace, heating the furnace, initiating growth, termin ating the growth, and finally flushing and cooling the quartz tube and removing the samp le. The details of this process can be found in Campbell et al [70]. The AFM image in figure 35 is of the sili con substrate surface after 5 minutes of growth under the following conditions: the sample is loaded into a furnace tube and then purged with a flowing argon gas (600 sccm ) and hydrogen (400 sccm) for at least 30 minutes. The samples were then heated to 800oC at a ramp time of 10 min and then annealed at that temperature for 30 minutes. Ethylene gas can then be admitted at a rate of 5 sccm for a growth time between 5 and 30 minutes. At the end of the growth period the ethylene and hydrogen were turned off, the top of the furnace was opened, and the sample was allowed to cool to ambient te mperature for approximately 60 minutes under an argon purge. The nanotubes in the AFM image are clearly visible and are at a density typical of the catalyst thickness and growth conditions The nanotubes grow only on the bare

PAGE 114

102 Figure 35. (a) TEM image of random networks of SWNT grown with CVD (b) SEM image taken at a grazing angle of almost 90o. In addition to nanotubes lying in the plane, clearly seen in (a), nanotubes that have grown out of the plane are clearly visible. Campbell et al [70].

PAGE 115

103 silicon surface and not on the SiO2, although occasionally a t ube will originate on the silicon surface and cross over onto the SiO2. The tube lengths vary from a fraction of a micron to over 10 m long. The overwhelming majority of the tubes shown have diameters determined from the heights of th eir AFM images to be in the range of 0.8 – 4.0 nm, with a mean diameter of 2.3 nm. The sm all size of the Fe nanoparticle catalyst is critical to the growth of su ch small-diameter tubes. The SEM image of the silicon substrate su rface is taken at a grazing angle (nearly 90o) with the surface. In addition to the na notubes on the surface the image also shows some vertical nanotubes extending out of the plane of the substrate, with an average density of one approximately every few squa re microns. While it is not possible to determine their diameters directly from th e SEM image because of the resolution limits of the instrument (~10 nm), their diameters are smaller than this limit and are thus consistent with those in the planar AFM im age. Attempts to determine whether those vertical tubes, as well as th e planar tubes shown in the AFM image, are single walled or multiwalled using transmission electron mi croscopy (TEM) were unsuccessful. Figure 36 shows four representative sample s that were grown during the author’s visit to NRL. Nanotube networks and forests were grown using both magnetic nanoparticles and magnetic thin f ilms as catalysts. These different catalysts grew vertical forests as well as horizontal random networks. The upper left image is a disperse horizontal random network of nanotubes grow n from Fe nanoparticle catalysts. The remaining images are growths from FeNi alloy thin film catalysts with varying thicknesses and growth conditions.

PAGE 116

104 Figure 36. SEM images of SWNTs grown using CVD. From J. Sanders

PAGE 117

105 Sixteen of these growths we re performed over nine days. The growth times and ethylene flow rates were varied along with the catalyst thickn ess. Obviously this was not enough time to systematically study each variable one at a time. In fact, the effect of water vapor was also experimentally studie d. A paper had just recently appeared in SCIENCE in November 2004 by Hata et al [3 ], which discusses enhanced CVD growth of single walled tubes using water vapor. This created a good deal of enthusiasm and the CVD gas system was redesigned to incorpor ate a water bubbler. The author performed all the plumbing with Teflon tubing and Swagelok connectors to accomplish this integration. The Argon carrier gas was split with a Swagelok union-T splitter and one section was sent to a water bubbler and another flow cont roller was added. This allowed a fourth variable to be added to the study, namely a fl ow rate of water vapor. According to Hata, the water vapor performs th e function of getting rid of some of the amorphous carbon that is thought to cause a term ination of the catalytic behavior In the presense of water vapor, HataÂ’s group was able to grow single walled nanotubes with di ameters in the 1-3 nm range but with lengths up to 2.5 mm. Th is growth was not able to be reproduced during the authorÂ’s stay at NRL, but the publis hed results are very promising. Hata et al report water vapor concentration levels from 20 ppm to 500 ppm and 10 minutes as standard growth time. They also report the ab ility to make precise patterns within which long tubes with uniform small diameters are constituents.

PAGE 118

106 Figure 37. Water assisted CVD growth. Hata et al [3]

PAGE 119

107 5.2 Carbon nanotube-based chemical sensors The random single walled carbon nanotube networks that are ro utinely grown at NRL were studied for electronic materials app lications [73, 74]. Figure 38 shows an AFM image of how a carbon nanotube network (CNN) is grown between two electrodes for a wide variety of electronics applications. At low nanotube densities, these networks are electrically continuous and behave like ptype semiconductors. They have field mobilities on the order of ~ 10 cm2/V s and a transistor on/off ratio ~ 105. 5.2.1 Background and motivation for CNT capacitive sensors The unique structural and el ectrical properties of si ngle-walled carbon nanotubes have inspired researchers to investigate a nd develop SWNT-based chemical [75-82] and biological sensors [83, 84]. Initial work in this area by Kong, et al, has shown that the resisitance of SWNTÂ’s changes in response to the exposure to certain molecules that undergo a charge transfer upon adsorbtion on the SWNT surface [75]. Such SWNTbased chemiresistors have been used to dete ct both toxic industrial chemicals [79-81] and a stimulant for chemical nerve agents [82]. In addition, Snow et al have shown that by using random networks of SWNTÂ’s as th e active material such sensors can be manufactured using conventional microfabrica tion techniques [82, 85] facilitating their use in commercial or defense applications. While these initial result s are promising, several critical problems must be addressed before SWNT-based sensors can successfully transition from impressive laboratory demonstrations to widespread commercial applications. Such problems

PAGE 120

108 include a high level of 1/f noise [86, 87], slow or partial recovery from exposure [78, 80, 82], and a susceptibility to cont act effects that are difficult to control [88]. To circumvent these issues we have explored alternative transduction mechanisms that do not rely on the charge transfer process. In this effort we have found that th e capacitance of SWNTbased structures is highly sensitive to a broad class of chemical vapors. This capacitance response is caused by the polarization of surface adsorbates and does not depend on charge transfer. In comparison to SWNT chemiresistors the capacitance response is faster, more sensitive, and can be measured simultaneously with the resistance response to provide complementary information. 5.2.2 Sensor fabrication Random CNT networks are grown on SiO2/Si++ substrates to build a capacitive sensing device for chemical and bio agents. Th e schematic of a device is shown in figure 39 (a), and experiments are continuing to enhance the response, sensitivity and selectivity. Several processing st eps are involved in the produc tion of this device. First, an active area is patterned w ith photoresist on top of a SiO2/Si++ substrate to prepare for thinning. Then the 250 nm SiO2 layer is thinned to 50-100 nm with standard etching techniques. The photoresist is removed and then a carbon nanotube network is grown via CVD as described earlier. The device is then patterned with photoresist; resist for the tubes, holes where the metal pattern is to be deposited. Then the me tal is deposited using a standard liftoff technique.

PAGE 121

109 Figure 38. Networks of SWNTs for elect ronic materials applications. [73]

PAGE 122

110 Figure 39. (a) Capacitive sensor using SWNT ra ndom networks. (b) Lock in Amplifier technique to measure the capacitive response of the sensor. From J. Sanders Chemiselective polymer layer (or biological linking molecule) Deposited Metal Contact Pads Ti/Au CNT network Dielectric Layer Conductive plate

PAGE 123

111 The area desired to be active with carbon nanotubes is then masked with p hotoresist and the tubes are removed from all other areas. The remaining photoresist is then removed with a solvent and finally an etch is pe rfomed in a convenient spot down to the Si++ layer so a good ohmic contact can be made with a probe to the bottom capacitor layer. The completed device is then wired into a series RC circuit as seen in figur e 39 (b). A lock-in amplifier is used to measure the voltage acr oss a known resistor which gives the overall response. In the presence of a chemical vapor, phys isorption concentrates the analyte on the SWNT surface. The chemical vapors were prepared by mixing saturated vapors of the analyte with dry air at 25oC, and the concentrations are reported as a fraction of the equilibrium vapor pressure, P/Po. Under bias the fringing electr ic field that radiates from the SWNT electrode produces a net polarization of the analyt e that is detected as an increase in capacitance. Because the nanotube s are ~ 1 nm in diameter the electric field drops rapidly within a few nanometers of the surface. This large field gradient coupled with the high density of adsorbates results in a capacitance response that is dominated by the dielectric properties of adsorbates on the SWNTs. Because many molecular adsorbates undergo a weak, but finite, in teraction with the SWNT surface, these capacitors rapidly respond to a broad spectrum of molecular adsorbates and can serve as general-purpose transducers for dete cting dilute chemical vapors.

PAGE 124

112 5.2.3 Capacitive response C/C to chemical analytes The capacitance of single-walled carbon nanotubes (SWNTs) is highly sensitive to a broad class of chemical vapors. This transduction mechanism can form the basis for a fast, low-power chemical sensor. In the presence of a dilute chemical vapor, molecular adsorbates are polarized by th e fringing electric fields radiating from the surface of a SWNT electrode, which causes an increase in it s capacitance. This effect has been used to construct a high-performance chemical sensor by coating the SWNTs with a chemoselective polymer that provides a large, class-specific gain to the capacitance response. As vapor containing analyte is blown onto the surface of the nanotubes, resulting in a change in cap acitance and thus a spike in the voltage across the known resistor shown in figure 39. Such SWNT chem icapacitors are fast, highly sensitive, and completely reversible. This can clearly be seen in figure 40 sh owing the capacitance change C/C, of one CNT sensor in response to repeated 20 second vapor pulses of N,Ndimethylformamide (DMF) at varying values of P/Po. The response that we see is rapid (~ 1 s, currently limited by the performance of our vapor delivery system), proportional to the analyte concentration, and completely reversible. The slow cap acitance delay seen in figure 40b at high-dose exposures is cause d by residual DMF desorbing from the vapor delivery system. We observe a similar, ra pid capacitance response that is completely reversible upon removal of the va por in the analytes that we ha ve tested. In table 2 [89] we list values of C/C for a number of chemical vapors, each measured at P/Po = 1%. We also list literature values of the equilibrium vapor

PAGE 125

113 Figure 40: Capacitance response in terms of C/C x 103 vs. time showing 20-second doses of dimethylformamide (DMF) at different vapor c oncentrations P/Po. From Snow et al [89].

PAGE 126

114 Table 2: Capacitance response to various ch emical vapors. Listed are the measured values of C/C corresponding to P/Po = 1%. Also listed are th e values of the dipole moment, m, the equilibrium vapor pressure, Po, at 25oC, and the vapor concentration, P, in parts per million. From Snow et al [89].

PAGE 127

115 pressure, Po [90], the vapor con centration, P (at P/ Po = 1%), and the molecular dipole moment, [90, 91]. In Fig. 41, we plot the values of C/C reported in table 2 for each of the analytes versus their respective dipole moments. In Fig. 41, we observe that for many an alytes the magnitude of the capacitance response correlates with the va lue of its dipole moment (the solid line corresponds to 2 C C). Nonpolar molecules such as hexane and benzene produce a small response, while relatively polar molecules like di methyl methylphosphonate (DMMP) and DMF produce a large capacitance respons e. DMMP is a stimulant for the nerve agent sarin. This correlation with dipole moment holds under the condition that the vapors are each delivered at a constant value of P/Po, and not for a constant value of P. For example, acetone ( = 2.88D) and DMMP ( =3.62 D) produce a comparable capacitance response when both are delivered at P/Po = 1% even though their vapor concentrations, P, differ by more than two orders of magnitude. In terestingly, several analytes such as chlorobenzene, 1,2-dichlorobenzene, and wate r (represented by the squares in Fig. 41) produce a small capacitance response even t hough they possess a rela tively large dipole moment. The SWNT network forms an array of nanoscale electrodes that serves as one plate of the capacitor with the other electrode formed by the heavily doped Si substrate. We measured the capacitance of the devi ce by applying a 30 kHz, 0.1 V AC voltage between the electrodes and detecting the out-of-phase AC current with a lock-in amplifier. The measured capacitance, 10 nF/cm2, is close to the parallel plate value

PAGE 128

116 Figure 41: Measured cap acitance response to P/Po = 1% doses of various chemical vapors plotted as a function of molecula r dipole moment. Th e capacitance response generally increases with dipole moment; however, large de viations are observed. From Snow et al [89].

PAGE 129

117 corresponding to a 250 nm-thick SiO2 gate dielectric. For SWNT networks with an internanotube spacing smaller than the SiO2 thickness, the capacitance should approximate that of a parallel-plate capacitor because of the overlapping fiel d lines of neighboring SWNTs. Initial work on the physics of th is interaction involved conceptually understanding the capacitance: d A d A Co where the dielectric constant of a polarizable molecule is: N N 3 4 1 4 1 And the polarizability is the sum of the intr insic molecular polarizability and the field induced alignment of the ot herwise randomly oriented mo lecular dipole moment. k T mol 3 2 There is ongoing research as to the physics of the interaction between an analyte molecule and the nanotubes in the capacitive sensor. Researchers are investigating the orientation of analytes with ringed mol ecules and whether they lie parallel or perpendicular to the surface onto which the nanotubes adsorb. Figure 42 shows a CAD drawing (a) of the fabrication of many devices on one chip in and a conceptual drawing of an analyte molecule adsorbing (b) on th e surface of a carbon nanotube within the active network of the sensor.

PAGE 130

118 Figure 42: (a) CAD schematic of 84 devices on one chip. (b) Conceptual drawing of an analyte molecule adsorbing on the surface of a CNT within the active sensor network. From Powerpoint presentation: Eric Snow of NRL-Washington with permission. SWNT network

PAGE 131

119 5.2.4 Chemical specificity for analyte identification The physics of the analyte/SWNT interact ion is still being resear ched and at the same time, sensor applications are advancing rapi dly. This capacitive sensor method is very sensitive, has a quick recovery, and has b een made very selective via two different methods shown in figure 40. The first method that was tried involves depositing a layer of chemically selective polymer over the CNT network. To provide chemical selectivity to the sensors we use the approach of other sorption-based chemical detectors [92] and coat the SWNT networks with thin films of chemoselective polymers. This is shown in the bottom of figure 40. The polymers highly co ncentrate particular classes of chemical vapors in the vicinity of th e SWNTs and produce a large, cl ass-specific gain to the capacitance response. The resulting sensors are fast, sensitive, and reversible. By using arrays of such polymer-functionalized SWNT chemicapacitors, a re sponse “fingerprint” can be obtained in order to detect and identi fy the vapors of toxic industrial chemicals, explosives and chemical warfare agents [93]. Through a collaboration with a chemistry group at NRL, a chemiselective polymer was chosen to match the desired anal yte. Initally, a somewhat thick layer of polymer was deposited. This gave excellent re sults for selectivity compared to the naked nanotube sensor which responded to everything. But the big drawback was response time and recovery time. After many iterations of depositing thinner and thinner polymer layers to decrease these values, and self-a ssembled chemically-selective monolayer was used and is shown in the top portion of figure 43. This solution presented the best of both worlds by presenting a high chemical specifi city and a very fast response time.

PAGE 132

120 Figure 43. Selectivity with (a) a self-assembled monolayer (S AM) of allyltrichlorosilane and (b) a chemically selective polymer, HC, specifically engineered for selectivity to nerve agents. HC is an acidic, str ong-hydrogen-bonding poly carbosilane. From Powerpoint presentation: Eric Snow of NRL-Washington with permission. 010203040506070809010 0 Time (s) 0 1 2 3 4 5 Sensor Sensor S en so r C re sp o Response to 2 nerve-agent simulants, (CH3O)2P(O)H and (CH3O)2P(O)CH3 Self-assembled monolayer Polymer

PAGE 133

121 Chemical specificity may also be achie ved independently of using a chemically selective polymer or a self -assembled monolayer. Several more experiments were performed while simultaneously coll ecting data on change in conductance G as well as change in capacitance C. This allowed us to investig ate both sets of data and notice some interesting properties. The ratio of the change in c onductance to the change in capacitance, G/ C remains constant for each individual chemical analyte. That ratio is an intrinsic property of each molecular adso rbate and can be exploited to determine a “fingerprint” or “signature” of the exact chemical that is present. In figure 44 (a) notice th at the two nerve agent s imulants respond and recover quickly in both capacitance and co nductance, and that the ratio G/ C remains constant. In figure 44 (b) we can see yet another signatu re of specificity wh en investigating the ratio G/ C from the data taken on devices coated with the chemically selective polymer HC. This gives us a fifth variable for achie ving chemical selectivity. The polymer and SAM coatings produces chemically selectiv e gain, altered response times, and allowed the determination of the four independent phys iochemical properties of charge transfer, polarization, solubility and diffusion constant. Combining that with G/ C enhances our results for chemical identification. The ideal sensor for the U. S. Navy would be an array of CNN devices on a chip, each coated selec tively so that any and all nerve agents, chemical warfare agents, and biological warf are molecules could be accurately detected and correctly identified.

PAGE 134

122 Figure 44: (a) G, C for two Nerve Agent Simulants ov er 10x pressure range (b) Analyte-specific effect of the polymer HC on G/ C for eleven different analytes. From Powerpoint presentation: Eric Snow of NRL-Washington with permission. 0100200300400500 -1 0 1 2 3 4 5Time (s)(CH3O)2P(O)CH3(CH3O)2P(O)H -0.2 -0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 MeOHEtOHIPABuOHHeOHH2OTolueneDCPDMMPhexaneacetone

PAGE 135

123 Table 3: Ratio of C/ G showing intrinsic specificity for chemical analytes From Powerpoint presentation: Eric Snow of NRL-Washington with permission. Chemical Vapor G/ C DNT 20.0% Dichloropentane10.0% Nitrobenzene7.7% Water4.5% Hexane4.3% Toluene2.5% Benzene1.3% MeOH0.6% EtOH-1.4% IPA-1.7% BuOH-1.0% HeOH-0.8% Acetone-3.1% THF-10.0% DMMP-12.5%

PAGE 136

124 5.3 Carbon nanotube-based biosensor Carbon nanotube biosensors have been fa bricated using the same CNN material as the basis for the transduction mechanis m. And again the sensor design, while different, uses simultaneous measurement of capacitance and conductance. The CAD drawing in figure 45 shows one of our biosen sors. The SWNT network is again grown using CVD and then selectively removed so the nanotubes remain only in the two areas with the interedigita ted Pd electrodes. There are si x contact pads around the outside, three in the lower left of figure 45 and three in the upper right. Also notice the circles at the top center and at the bottom center. Th ese represent the locations of the input and output capillary tubes for the microfluidic flow system. We then designed and fabricated a polymer gasket in order to surround the input/output locat ions and the active sensor area, but staying inside of the six contact pads for acquisition of capacitance and conductance data. Biological an alytes are delivered in saline solution, thus we needed to design a sealed, air-bubble-free ce ll to contain the analyte so lution that would allow the analyte to adsorb onto the nanot ubes. But at the same time, we needed to isolate the saline solution from the six contact pads and the probes. The entire apparatus was then contained in a plexiglass cell with an aluminum backing that was screwed into the plexiglass (Figure 46 a). A sy ringe pump was them calibrated for several volume flow rates and used to deliver the solution to be investigated (Figure 46b). The initial calibration and testing of th e syringe pump and microfluidic flow system was then performed. As the pump was driven in the forward and reverse modes,

PAGE 137

125 Figure 45: CAD drawing of CNN biosensor showing the active area and electrodes for acquisition of capacitance and conductance data. From J. Sanders

PAGE 138

126 Figure 46: (a) Microfluicic cell containing CNN Biosensor (b) Syringe pump to control microfluidic flow of the biological analyte in saline solution to the active area in the cell. From J. Sanders

PAGE 139

127 the pressure within the capi llary tubes varied greatly, wh ich introduced ai r bubbles into the system. If an air bubble enters the cel l and the active area of the CNN network, the surface tension of the saline solution will cause it to be very difficult to remove. Then the entire cell would have to be di sassembled and the entire proce ss would have to start over. The flow rate threshold that would start to cause the introduction of air bubbles was systematically recorded. 5.3.1 pH testing and DNA functionalization After the syringe pump had been calibrat ed and the microfluidics problems had been solved, we began intial testing of pH solutions. In collaboration with an NRL chemistry group, we prepared several acid and ba se solutions. In a typical experiment, as shown in figure 47, three beakers were prepare d. The first contained deionized water (DI water), the second contained an acidic soluti on (pH 6.00) and the third beaker contained a basic solution (pH 7.98). We started the data acquisition while flowing DI water in order to establish a baseline value for both c onductance and capacitance and allowed that baseline to continue for 500 seconds. We th en quickly reversed the flow on the syringe pump, submerged the input capi llary tube into the acid, then quickly switched the pump to forward mode once again. As can be seen in figure 47, we again have a very fast response as the rise/fall of the capacitance/ conductance curve has an almost vertical slope. The process was then repeated, going first into the DI water where our sensor recovers extremely quickly to the baseline values for C and G. The process was repeated again for base, DI water, acid, DI water, ba se, and finally DI wate r through 2500 sec.

PAGE 140

128 Figure 47: pH testing on microfluic ce ll with CNN biosensor. From J. Sanders 0500100015002000 0.00010 0.00012 0.00014 0.00016 0.00018 0.00020 0.00022 0.00024 0.00026 0.00028 0.00030 0.00032 0.00034 pH 7.98 pH 6.00 pH 6.00 pH 7.98 DI Water DI Water DI Water DI Water DI Water Biosensor Response vs. TimeConductance (Seimens) Capacitance (Arb. Units)Time (sec) Capacitance Conductance

PAGE 141

129 Continued work on our biosenso r included the objective of initializing studies of DNA functionalization as a method of tagging biological analytes The concept is to tag a biological molecule with a strand of DNA, either Adenine, Cytosine, Guanine, or Thymine. Then we would tag the SWNTÂ’s in our sensor with the complementary strand, for example Adenine is complementary to Thymine. When a DNA strand links to its complementary strand, we would then detect its presence by a change in capacitance and/or conductance. Figure 48 shows our initial data in this DNA study. First we ran a voltage sweep of several bare devices on four different chips to collect families of C(V) curves from 1.0 Volt to 1.0 Volt. Next, we functionalized each device with a different strand of DNA and ran the sweep once again. Chip 1 was f unctionalized with Adenine (A-DNA), Chip 2 with Thymine (T-DNA), Chip 3 with Guanine (G-DNA) and Chip 4 with Cytosine (CDNA). Finally, we functionalized one half of each chip with its complementary strand and the other half of the chip with one of the other two non-complimentary strands. For example, to acquire the data shown in Fi gure 48, we covered half of chip 1 (base functionalization was A-DNA) with Thymine (A denineÂ’s complement) and the other half with Guanine (a non-complement ). It is clearly seen th at there is an increase in capacitance from the bare device to the devi ce funtionalized in A-DNA, and then another increase in capacitance with the device f unctionalized in T-DNA on A-DNA. We did not see a second increase in capacitanc e when the non-compliment was added.

PAGE 142

130 Figure 48: DNA functionaliz ation with CNN biosensor. From J. Sanders. -1.0-0.8-0.6-0.4- 550 600 650 700 750 800 Capacitance vs. Voltage DNA-functionalized CNT sensorsCapacitance (F)Voltage (mV) Bare device Functionalized with A-DNA Functionalized with T-DNA on A-DNA

PAGE 143

131 5.3.2 Creatinine sensing Another objective was to study the possibility of sensing biological analytes that are excreted from the human skin as a conn ection to the goals of the USF IGERT SKINS program as sponsored by the NSF. Many molecules are excreted through the human skin, including amino acids, creatinine, narcotics, alcohol, etc. The U.S. Navy is already working on a real-time narcotics monitoring device that would measure drug levels in sailorÂ’s sweat and wirelessly transmit the da ta back to some central laboratory. We initiated an investigation of the creat inine excreted through human skin. (see figure 49). Creatinine is a waste product of the molecule creatine monohydrate which is consumed in the average personÂ’s diet at th e rate of approximately 2-3 grams per day. Soldiers and athletes will supplement with an additional 5 grams per day as creatine monohydrate aids in the ATP process and has be en shown to increase power, speed, and acceleration in plyometric moves. Howeve r, if supplementation exceeds 5 grams per day, excess demand is placed on the kidneys a nd the byproduct creatinine is excreted in the urine and in sweat. The accurate identifi cation and quantification of creatinine by our CNN sensor could be a valuable aid in this pro cess. This could also be a benefit, not only for soldiers and athletes, but to patients of kidney diseases and kidney failure. Excess creatinine in the urine and sweat is symptom of poor ki dney function. A very sensitive carbon-nanotube sensor could aid in the early detection and successful treatment of this disease.

PAGE 144

132 Figure 49: Creatinine molecule. 1-Methylglycocyamidine; 2-Amino-1-methylimidazolidin-4-one 4H-Imidazol-4-one, 2-amino-1,5-dihydro-1-methylC4H7N3O Creatinine

PAGE 145

133 Chapter 6 Spin transport in carbon nanotube devices 6.1 Background theory and m agnetoresistance experiments This section will describe the final expe rimental portion of this work. Carbon nanotube devices were fabricated with ferromagnetic contacts and superconducting contacts for spin transport studies. As di scussed earlier, interest is growing in “spinelectronic” devices whose operation depends both upon the electron spin as well as the charge of the conduction electrons. The unusual properties of carbon nanotubes offer intriguing possibilities for such devices. Thei r elasticand phasescattering lengths are extremely long [94] and car bon nanotubes can behave as one-dimensional conductors [95]. Our research plan is motivated by this goal of finding direct evidence for coherent transport of electrons spins in carbon nanotubes through the use of our alreadyestablished PCAR system. There are several theoretical papers de scribing the boundary states, chiral spin currents, spin transport, and spin filtering in carbon nanotubes [96104]. There are also several experimental papers that report mangetoresistance results on Co/CNT/Co samples, Co/CNT/NiFe samples, or Ni/CNT /Ni samples [105-113]. Spin coherence length in a SWNT can be calculated from th e MR data using MTJ theory [105].

PAGE 146

134 In Figure 50 (a) we have shown the Co/MWN T/Co device fabricated by Tsukagoshi et al [105]. They used shape anisotropy of the Co electrodes to establish four states of saturation which gave them two parallel and two anti-parallel magnetization states. They were then able to measure a high differential resistance (dV/dI) in the antiparallel states and a low dV/dI in the parallel states. The spin-injection model for the nano tube magnetoresistance requires a sufficiently small amount of spin scattering to occur both within the nanotube, and at the interfaces between the nanotube and the contacts. The spin scattering length in the CNT can them be estimated using JulliereÂ’s mode l for the magnetic tunnel junction (MTJ), as shown in figure 2. The difference between the tunnel resistance in the parallel (Rp) and antiparallel (Ra) states is given by: 2 1 2 11 2 ) ( P P P P R R R R Ra p a a Here, P1 and P2 are the percentage of conduction electrons polarized in the majority spin band in the ferromagnetic contacts 1 and 2. Fo r Co, the polarization has been determined [13] to be 34% giving a maximum resistan ce change of 21%. In their best case, R/Ra reaches a maximum value of 9% (Figure 50b) so that ~ 14% of the spin-polarized electrons travel the 250 nm through the na notube without spin-f lipping. The spinscattering length, ls, can then be estimated by assuming that the spin polarization reduces as exp (-l/ ls) within the nanotube. This gives ls = 130 nm. Although fairly long, this is probably an underestimation. The spin polar ization near the FM/CNT interface will depend on the interface quality, and could be appreciably lower than 34%.

PAGE 147

135 Figure 50: (a) Spin valve Co/MWNT/Co device to measure spin transport, (b) Magnetoresistance for para llel antiparallel stat es of magnetization. Figure reproduced from Tsukagoshi et al [105].

PAGE 148

136 Also, spin scattering at the ferromagnet/nanot ube interface was not taken into account. Finally, it should be noted that the MTJ theory cannot be expected to completely describe the CNT device, and a more exact th eoretical description is needed. 6.2 Fabrication of SC/CNN/FM samples and measurement Our initial sample concept for measurement of spin transport in CNNs is as shown in figure 51. We fabricated a set of these samples with one exception: we deposited Co instead of CrO2 so that we could easily do this in-house at the USF NNRC building. Future experiments will involve our collaborators at NIST or the Univ. of Alabama and their expertise in CrO2 growth. We first grew a CNN on a Si/SiO2 substrate with the CVD method described in section 5.1. Th en we masked half to two-thirds of the area and used electron beam evaporation to deposit 200 nm of Co on the complementary area. Next, Ag ink and 30-gauge Cu wire were used to make two contacts to the Co thin film which was electrically connected to th e PCAR apparatus and the data acquisition system. Finally, a Sn superconducting tip was machined and mechanically polished with progressively finer grades of sandpaper in the usual way for PCAR experiments. We then aligned the tip above the nanotube networ k so that it would be driven down into the active area after the entire sample region was cooled to T=2.0K. We ran several sets of experiments on an entire series of samples, and also as a function of distance between the ferromagnetic Co and the superconducting Sn ti p. We did not acquire any clean data showing a sign of a superconducting gap or a ny signature of spin polarized transport through a suppression of an Andreev signal.

PAGE 149

137 Figure 51: HM/CNN/SC concept for measuri ng spin transport in CNTs. From M. Almand and J. Sanders.

PAGE 150

138 We decided to fabricate an entire new se t of samples by depos iting a thin film of superconducting Sn instead of using a Sn ti p. Again, a CNN was grown using CVD on a Si/SiO2 substrate. Half the sample was masked and again approximately 200 nm of Co was deposited using e-beam evaporation. The Co and crucible was allowed to cool, the chamber returned to ambient pressure, and th en the sample was removed from the e-beam evaporation chamber. Next, we masked the Co film and a very thin area of the active CNN. The sample was replaced in the e-beam chamber and now approximately 200 nm of Sn was deposited. This final fabricati on step left us with a Co/CNN/Sn device as shown in figure 52. An optical image of the sample is shown in (a) where Sn film is seen above, then a 40 m gap, and finally a Co film. Part (b) shows an SEM image of the CNN within the gap between the two metals. Silver ink and 30-gua ge Cu wire were again used but this time to make two contac ts to the Co thin film and another two contacts with the Sn thin f ilm. All four leads were then electrically connected to the PCAR apparatus and the data acquisition system. Unfortunately, at this point in time, we ran into several months of troubleshooting w ith establishing a stable 2.0 K temperature in our PPMS. We ran several diagnostic experime nts with the help of the manufacturer, but this severely delayed our e xperimental liquid time. We ran several sets of experiments on an entire series of Co/CNN/Sn samples, and did not acquire any clean data showing superconducting gap or any signature of spin polarized transport through an Andreev signa l (Figure 53). However, we did achieve a low resistance, ohmic contact across the na notube network through the deposited Co and Sn electrodes. We meas ured on the order of 300 across the gap, which is a success in

PAGE 151

139 Figure 52 (a) Optical image of Co/CNN/Sn devi ce (b) SEM image of the CNN in the gap between the Sn thin film and Co thin film. From J. Sanders. 100 nm x25000 25.0kV 150 m

PAGE 152

140 Figure 53 Raw dI/dV data on Co/CNN/Sn samples. From J. Sanders. Sn/CNN/Co Conductance0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 -0.04-0.03-0.02-0.0100. Bias Voltage (V)dI/dV

PAGE 153

141 itself considering the challenges of achieving that level of contact as reported in the literature [105-113]. The I-V data and dI/d V data in figure 53 was collected by the author at NRL-Washington in the laboratory of Dr. Robert Soulen and Dr. Mike Osofsky. 6.2 Fabrication of SC/CNN/SC samples and measurement Our third and final set of samples were fabricated to make a SC/CNN/SC junction. Instead of using ferromagnetic Co, we chose to deposit Sn for both electrodes. This came from a suggestion by Dr. Mike Osofsky at NRL to look for a Josephson junction first with SC/SC device and them go back to the Andreev setup with SC/FM. Also, to see a Josephson signal, we would expe ct to need a much smaller gap, like when Josephson effects are seen in SC/Insulator/S C devices and the oxide barrier is on the order of nm. Thus we tried to do some different masking methods using UV lithography, and in the future another student will ex tend this work by using e-beam lithography and/or using the focused ion beam (FIB). The sample shown in Figure 54 was fabr icated using UV lithography. We grew a carbon nanotube network in the usual way with CVD on a Si/SiO2 substrate, then used a hot-roller lamination method w ith SU8, a negative photor esist, to pattern a 10 m line down the middle of our sample. SU8 is a negative, epoxy-type, n ear-UV photoresist (365 nm). It was originally developed and pa tented by IBM-Watson Research Center and applications include microelectronics, microfluidics, MEMS, and magnetics (when magnetic material is added). Photoresists su ch as SU8, are based on epoxies, specifically a 1,2-epoxy resin which is a bridge consisti ng of one or more groups, each with an

PAGE 154

142 oxygen atom bonded to two carbon atoms. Su ch molecules are capable of being converted to a thermoset form or three-dime nsional network structure by a curing process called cross-linking. Cross-linking de scribes when one or more ki nds of reactants, i.e., an epoxide and a curing agent, are transformed from a low-molecular-weight material to a highly crosslinked network. Photoepoxies, such as the SU8, are normally polymerized by a cationic photopolymerization induced by Lewis acids generated during UV illumination. The polymerization is done by the ring-opening of the 1,2-epoxy. Polymerization was done by UV exposure in the SF-100, manufactured by Intelligent Micr o Patterning of St. Petersburg, Florida. This t ool uses light reflected from a DMD array to control exposure pixel by pixel through a maskless projection. Th e line developed here was a 2 pixel line, producing a linewidth of 10 m. The resist is developed using SU8 developer and rinsed in isopropol alcohol. This 10 m distance is much better than the 40-50 m gap that we had fabricated in the previous set of samples. The line is 10 m wide, approximately 0.7 cm long and 120 m thick. We then used e-beam evapora tion again to deposit 200 nm of Sn across the entire middle third of the sample. Because of the thickness of the resist, we chose to forgo the removal process since the Sn thickness is three orde rs of magnitude thinner than the resist. Thus, there would be no electrical conduction across the resi st at the gap. This was verified with an Ohmmeter after using Ag ink to make four contacts (A, B, C, D from left to right in the fi gure) and verifying continuity everywhere save across the gap.

PAGE 155

143 Figure 54 Sn/CNN/Sn sample fabricated wi th UV lithography and e-beam evaporation. Hot-roller lamination method used to draw a 10 m line with SU8 nega tive photoresist. From J. Sanders.

PAGE 156

144 We then soldered all four Cu wires to the PCAR apparatus and connected to our Keithley meters and LabVIEW data acquisition sy stem. We cooled the sample to 2.0K to bring the Sn electrodes into a superconducting state (Tc = 3.7 K for Sn). We collected conductance data but did not see any Josephson signal. This initi al sample is a step in the right direction, but we would like to build our gap much, much sm aller between the two superconducting electrodes. Again, with el ectron beam lithography and FIB technology available at the USF NNRC building, this s hould be possible in the future for a new graduate student in the Functional Materi als Laboratory. My advisor and our group intend to initiate focused research on the interplay of magnetis m and superconductivity which will likely be a Ph.D. topic for a future gr aduate student. It is anticipated that these experiments will be continued as part of that project.

PAGE 157

145 Chapter 7 Ongoing analyses and future directions 7.1 PCAR analysis There are two ongoing studies related to the analysis and fitting of our PCAR data from the SrRuO3 series and the La1-x(Ca, Ba, Sr)xMnO3 series. In the former, we are investigating the square root dependence of the background conductance data. We have the unique capability of being able to study our junctions abov e the critical temperature, Tc, and above the critical magnetic field, Hc of the superconducting probe. In this way we can subtract the background and fit the remaining data to the modified BTK model. In the latter, we are studying the fundamenta l aspects of the theoretical fitting and are using a 2 statistical model to more accurately determine the spin polarization, P. 7.1.1 Investigation of the V 1/2 background conductance A new analysis is ongoing regarding our experimental PCAR data on the SrRuO3 series from chapter 4 though a collaboration with R. Soulen and M. Osofsky at NRLWashington. The purpose of this study is to characterize the electrical and magnetic properties of SrRuO3 as it is disordered and to in vestigate how this influences the transport spin polarization. This section is a summary of a draft that will be submitted for

PAGE 158

146 publication in the near future. We are inve stigating the results from our studies of SrRu0.94Ti0.06O3, SrRu0.9Mn0.1O3, SrRu0.9Cr0.1O3, and SrRu0.92O3. They were lightly doped (~6-8%) in order to keep the electrica l resistivity reasonably low. Samples with higher doping would be closer to the metal-insulator transition where the resistivity becomes sufficiently high as to make the PCAR data hard to analyze. Thus we did not measure samples with higher TM concentratio ns. The temperature dependencies of the ac susceptibility and the dc magnetization were determined for each sample using our PPMS. The magnetic properties of the samp les displayed the effects of disorder introduced by the TM substitution or Ru deficiency. The highest TC (defined from the position of the p eak in the ac susceptibility) was found in SrRu0.9Cr0.1O3 (see Table 4), which is significantly greater than that of the pure compound SrRuO3. An increase in TC of SRO has been reported previously in chapter 4. We fitted the measured molar ac susceptibility ( m = M/H) curves with the Curie-Weiss Law: m = 0 + ( B NA/3kB) eff 2/(T), where 0 is a temperature-independent background susceptibility, B is the Bohr magneton, NA is the Avogadro’s constant, kB is the Boltzmann constant, is the paramagnetic Curi e-Weiss temperature, and eff is the effective paramagnetic moment. Results fr om the fits are displayed in Table 4. SrRuO3 is a highly correlated metal and t hus its properties should be very sensitive to disorder. This expectation is borne out in the el ectrical conductivity, Figure. 55a shows that the temperat ure dependence of the resistivity =1/ is strongly influenced by disorder. That is: Pure SrRuO3 exhibits classic “metallic” behavior at low

PAGE 159

147 0.000 0.002 0.004 0.006 0.008 0.010 0.012 0100200300400resistivity ( -cm)temperature (K)SrRuO3SrRu.94Ti.06O3SrRu.9Mn.1O3SrRu.9Cr.1O3SrRu.92O3 (a) 0 200 400 600 800 1000 05101520conductivity ( -1 cm-1)T1/2 (K1/2)SrRu.92O3SrRu.94Ti.06O3SrRu.9Cr.1O3SrRu.9Mn.1O3SrRuO3 (b) Figure: 55. (a) Resistivity of the samples as a function of temper ature in zero magnetic field. (b) Conductivity of the samples as a func tion of the square root of temperature in zero magnetic field. From J. Sanders.

PAGE 160

148 temperatures (i.e., (d /dT) is positive), whereas the re sistivity of the others rapidly increases at low temperatures. Indeed, the ma tter may be put more forcefully. When the data are replotted in Fig. 55b as the conductivity versus the s quare root of T, it is seen that the data follow a straight line at low temperatures. This dependence of on the square root of temperature is predicted by theo ries for disordered metals [114]. In this framework, the metal-insulator transition (MIT) occurs when (0) is zero. From Fig 55b it is easy to arrange the samples by their pr oximity to the MIT: the closest is SrRu0.92O3, the next closest is SrRu0.94Ti0.06O3, followed by SrRu0.9Cr0.1O3, followed by SrRu0.9Mn0.1O3. Pure SRO is seen clearly as the most remote from the MIT. Since (0) is greater than zero for all the samples, they all are considered to be metals, albeit endowed with the curious features of dirty metals. Clearly both the magnetic and the resistivity data show that transition metal substitution or the Ru in SRO as well as th e presence of Ru vacancies influences the magnetic and electronic properties. Given this fact, we sought to determine how the disorder and/or the correlation modify the spin polarization in the SRO system using PCAR. Accordingly, we performed PCAR measurements on each sample for temperatures ranging from 2.0-4.0 K and for a pplied magnetic fields ranging from 0-500 Oe. The applied magnetic field and temperat ure controls of the PPMS allow us to investigate the quality of a single junction with ease. Fig. 56a displays the results for the SrRu0.9Mn0.1O3 sample for a single point contact as a function of temperature in zero applied magnetic field ( Ha=0). The curves are typical of many other materials studied, but with one important exception Note that the conductance does not become horizontal

PAGE 161

149 6.0 6.3 6.6 6.9 7.2 7.5 -6-4-20246 4 K 3.7 K 3.0 K 2.8 K 2.75 K 2.5 K 2.25 K 2.0 K # 1 2.0 K # 2 2.0 K # 3 2.0 K # 4G(V)/G(4K, 4mV)V(mV) (a) 0.85 0.90 0.95 1.00 1.05 -6-4-20246 3.7 K 3.0 K 2.8 K 2.75 K 2.5 K 2.25 K 2.0 K # 1 2.0 K # 2 2.0 K # 3 2.0 K # 4G(V,T)/G(V,4K)V(mV) (b) Figure 56: PCAR spectra, take n with a Sn point and a SrRu0.9Mn0.1O3 counter-electrode, as a function of temperature. The magnetic field was zero. (a) The data were normalized at a voltage of 4 mV and (b ) the curves are normalized by dividing the data by the experimentally measured G(V) curve at T= 4 K. From J. Sanders.

PAGE 162

150 at large voltages as is commonly observed in mo st metals. Indeed it continues to increase as a function of V. This result is most likely or iginating from the effect of disorder on the DOS in the SRO alloys. The argument is as follows: The conductance of a PCAR junction is always proportiona l to the product of the DOS of the superconductor and the DOS of the normal metal. Ordered metals ha ve a nearly constant DOS near the Fermi surface and thus the DOS may be considered constant (i.e., independent of energy or voltage). In this case G(V) is constant wh en V is much larger than the superconducting energy gap, Disordered metals, by contrast, have a DOS which varies with energytypically as the square root of V [114]. Cognizant of this possibility, but not wanting yet to presume that it was the explanation for the observation, we just normalized all the curves by the single curve obtained at 4 K-i.e., well above the superconductive transition temperature of Sn. The resu lts are shown in Fig. 56b, where we note that G is now horizontal for large voltages. Once the problem of the V1/2 behavior at large V was eliminated by proper normalization, the PCAR spectra for SrRu0.9Mn0.1O3 could be analyzed using the modified BTK model [14]. The fitting parameters used within this model are the barrier strength ( Z ), and the spin polarization ( Pt). All of the other parameters such as the serial resistance ( Rs) of the sample, temperature, and the superconducting gap () were not varied in the fitting routine but rather remained fixed. Displayed in Fig. 57 is a conductance curve of SrRu0.9Mn0.1O3 taken at Ha = 0 and at T=2.0 K. The fit by the modified BTK model is shown as the so lid line. The fit yielded a value of Pt = 0.64 0.02 and Z = 0.01.

PAGE 163

151 0.75 0.80 0.85 0.90 0.95 1.00 1.05 -6-4-20246G(V)/Gnvoltage (gap units) Figure 57: G(V) for a Sn point and SrRu0.9Mn0.1O3 base at T=2.0 K and H=0. The open circles represent the measured G(V) data whil e the solid line represents the BTK fit to the data. For this curve, the fitted values were: Z=0.01, P=0.64 From J. Sanders.

PAGE 164

152 Also given is the experimentally determined junction resistance Rj = 2.4 The Pt obtained from our analysis agrees well with Pt obtained in the low Z limit from Nadgorny et al [49]in their investig ation of thin film samples of SrRuO3. Curiously, both our results and the results obtained by Nadgorny et al. show that the value the magnitude of Pt agrees with the theoretical diffusive value a nd not the ballistic value. This may indicate that the transport limit is intermediate, namely L ~ a In addition to measuring the PCAR curves for this sample as a function of temperature, we also studied the effect of ma gnetic fields at the very lowest temperature, T = 2.0 K. The data normalized at 4 mV are shown in Fig. 58a. Note first the dependence on magnetic field. Several spectr a in the low field regime ( Ha < 200 Oe) are shown to illustrate that very small changes occur until th e critical field is reached for the junction. Our results in Fig. 58a show that the transiti on from AR suppression to the normal state is not continuous but rather happens abruptly. That is, at the critical field, where the superconductivity of Sn is suppressed, the conduc tance drops dramati cally to the normal state conductance. The empirical value obt ained for Hc(T=2 K) is between 200 and 250 G. This value is in good agreement with the calculated value of 230 G at T=2 K for pure Sn. It should be noted that all samples reported here were zero-field cooled and that the coercive field ( Hcoer) obtained from M ( H ) data far exceeded Ha. (Typically, Hcoer was determined to be of order of several kOe at T = 5 K, see: Table 4). The finer features observed in the field dependence such as the abru pt transition could be due to fact that the induced field in the vicinity of the junction should strictly be taken as B = Ha + 4 M where the magnetization of the sample should also be considered to account for the pair

PAGE 165

153 6.0 6.3 6.6 6.9 7.2 7.5 -6-4-20246 4 K; 0 G 0 G 25 G 75 G 100 G 125 G 150 G 200 G 225 G 250 G 300 G 500 GG(V)/G(4K, 4mV)V(mV) (a) 0.85 0.90 0.95 1.00 1.05 -6-4-20246 0 G 25 G 75G 100 G 125 G 150 G 200 G 225 G 250 G 300 GG(V,H)/G(2K, 500G)V(mV) (b) Figure 58: PCAR spectra, take n with a Sn point and a SrRu0.9Mn0.1O3 counter-electrode, as a function of magnetic field. The temperatur e was fixed at T= 2 K. (a) The data were normalized at a voltage of 4 mV and (b) The curves are normalized by dividing the data by the experimentally measured G(V) curve at T= 2 K and H=500 G. From J. Sanders

PAGE 166

154 breaking in the contact region. A systematic comparative study of the field-dependent junctions in spin polarized materials will be fu rther explored in the future. The second thing to note from Fig. 58a is that the G(V) curves do not re ach horizontal limits at large voltages. Applying the same reasoning as for Fig. 56a, the data in Fig. 58a were normalized by the curve for G taken at T= 2 K and H= 500 G. The subsequent family of curves of normalized conductance as a function of H are shown in Fig. 58 (b). They have the proper behavior now at large V. A quantitative analysis of the data shown in Fig. 59 supports the following interpretation. Both normalization curves appear to fit on a common curve at high voltages (i.e., G~V1/2), that is expected from the DOS of disordered metals. Note, however, that the 4 K curve deviates from the square root behavior at 2 mV, whereas the 2 K data deviate from the square root behavior at 1 mV. We believe that the deviations are caused by thermal broadening. An elemen tary analysis indicat es that the fitted thermal broadening is given by: T k VB4 7 32 1 2 That is, it is about 4 times the thermal broadening. Such factor s are common for tunneling curves. In summary, we have presented PCAR data of polycrystalline bulk samples of SrRu0.9Mn0.1O3, SrRu0.9Cr0.1O3, SrRu0.94Ti0.06O3 and SrRu0.92O3. Lightly doped samples were studied due to the difficulty of PCAR st udies on higher resistivity samples. The Ti doped sample, for which the Ti4+ ion provided no magnetic moment, displayed a high degree of spin polarization with the sa me value as the parent compound, though the magnetic ordering decreased substantially and th e electrical properties also differed from SrRuO3. Both SrRu0.9Mn0.1O3 and SrRu0.9Cr0.1O3 displayed coherent suppression of

PAGE 167

155 6.90 7.00 7.10 7.20 -6-4-20246G(V)/G(4K, 4mV)V(mV) Figure 59: Experimentally derived normaliza tion curves for PCAR spectra, taken with a Sn point and a SrRu0.9Mn0.1O3 counter-electrode. (solid dots): T=4.0 K and H=0. (solid triangles) : T= 2 K and H= 500 G. From J. Sanders.

PAGE 168

156 Sample Tc (K) (K) eff ( B) M70kOe ( B/f.u.) Hcoer (kOe) Pt SrRuO3 163 165 2.73 1.44 2.16 0.6 0.02 SrRu0.9Cr0.1O3 185 190 2.34 1.21 1.89 0.51 0.02 SrRu0.9Mn0.1O3 115 124 2.82 1.34 2.8 0.64 0.02 SrRu0.94Ti0.06O3 92 110 2.35 0.70 5.6 0.61 0.02 SrRu0.92O3 69 136 1.86 0.75 6.1 _______ Table 4: Basic material parameters of our samples determined from magnetic measurements. M70kOe is the magnetization measured in Ha=70 kOe at T=5 K. Hcoer is the coercivity field at T=5 K. eff and are determined from fitting the magnetic susceptibility with the Curie-Weiss Law. Al so presented are the fitted values of Pt

PAGE 169

157 Andreev reflection, though the s quare root behavior was clear ly visible. Values of 0.64 0.02 and 0.51 0.02 for Pt were obtained from the fits respectively. Though these uncertainties of the latter two were estimated from the fitting proce dure of the subtracted spectra, it is difficult to determine the total uncertainty by removing the broadened part of the spectrum (i.e. the broadening of the DOS) for each sample. Future studies of the electronic structure of SrRuO3, considering substitution of a TM could be undertaken in order to understand the correlated nature and its ultimate effect on the electronic structure, which is seen from the definition of spin polarization to influence the transport spin polarization. 7.1.2 Statistical 2 analysis of spin polarization P Another investigation is cu rrently underway in colla boration with Doug Lovelady and Dr. David Rabson of USF. Their theoretical/computational physics group has already verified and improved a modified BT K fitting routine and used it to extract P values from our experimental data (see figure 33a and 33b on pg 87). A recent PRB paper by Bugoslovsky et al [31] suggests a re fined method for analysis of PCAR curves in order to reduce the ambiguity in fitting a multiparameter model [52] (figure 60) and find an accurate value for the spin polarization P. They show that the methods used in the last several years in the analysis of PCAR spectra on the basis of the generalized Blonder-Tinkham-Klapwijk model may lead to de generate solutions. In other words, the straightforward determination of P from the multiparameter fitting is ambiguous.

PAGE 170

158 Figure 60: Example of degeneracy in the fou r-parameter fitting routine in the modified BTK model. Notice that both f its that have excellent correl ation with the experimental measurement (open circles) and shows the ambiguity in finding an accurate value for spin polarization P. From Auth et al [52]

PAGE 171

159 The fitting procedure we are using is based on an optimization algorithm with the target function being the normalized sum of squared de viations between the fitted data and the trial function: Where the analyzed conductance-voltage [G (V)] curve is comprised of N points (Vi, gi), and the fitting function G(Vi) is comprised of either ballis tic or diffusive regime of the generalized BTK. The algorithm allows optim ization in all four pa rameters, or in any subset, while the others are kept fixed. By minimizing this function, we would be able to find the correct value of th e spin polarization P. At T=0K, the theoretical BTK spectra s how very clearly the distinction between the effects of increased P and increased Z [48]. This is, however, no longer the case when the spectral broadening becomes significan t. Indeed, a broadened theoretical curve generated for a set of parameters can be cl osely fitted by another curve, with different values of parameters. An example of a cl ose match between two generated curves is shown in the inset of figure 61. There is a finite difference between the curves, but it is so small ( 2<10-7) that it would be impossible to see in experimental data. The ambiguity is due mainly to the fact that with lim ited spectral resolution, the model fails to distinguish whether it is high P or high Z that causes the depression of the conductance at small bias. The degeneracy of the BTK fitting function is a major problem for the applicability of the PCAR method. This 2 method will show how the effect of the ambiguity can be reduced. 2 2) , ; ( 1 ) , ( i i iP Z V G g N P Z

PAGE 172

160 Figure 61: Inset shows two generated curves th at are closely coincident in spite of the polarizations being 0.4 and 0, respectively. For relatively strongl y broadened spectra, degenerate fits can be produced by simultane ously varying three fitting parameters: Z, and Main Panel: Results of the 2 (Ptrial) analysis for the two curves. Although there are strict minima that correspond to corre ct values of P, the total variation 2 is minute below Ptrial = 0.4 From Bugoslavsky et al [31].

PAGE 173

161 The main panel of figure 61 presents the results of 2 vs. P computed for the two spectra shown in the inset, after performing the three-parameter fitting. The main feature is the very shallow, almost horizontal part in an extended region above P=0. Mathematically, the minimum occurs at P=0, but it is realistical ly impossible to resolv e with any degree of experimental noise. The horizon tal part represents the compensation effect: the model is able to adjust Z in such a way that it co mpensates for P and produces virtually equally good fits over a wide range of P. The comp ensation effect works for higher P as the spectra get even more broadened. However, beyond a certain value of P there is an abrupt threshold and the quality of fit deteriorates dramati cally. This distinct feature leads to an absolute upper limit to th e possible value of spin polarization. Another issue is the fact that the experi mental data is always normalized to some arbitrary high-bias conductance outside the superconducting gap. This arbitrary judgement may be erroneous, especially if th ere are features in the high-bias voltage function. As a dramatic example, Bugosl ovskyÂ’s group has normalized the simulated curves shown in figure 62 by 0.995 and 1.005, corresponding to a 0.5% experimental uncertainty. The implications on the determination of P are very serious. In the case of the 0.995 normalization the optimization has a minimum at P=0.24 instead of the true value of P=0 for Cu. The correctly norma lized curve and the 1.005 normalization both yield minima at P=0. A solu tion for this problem exists. Notice that the correctly normalized data yields a zero slope at P=0 where both of the incorrect normalizations have a non-zero slope. Also notice that the positiv e error (1.005) has a positive slope at P=0 and the negative error (0.995) has a negative slope at P=0. This fact

PAGE 174

162 Figure 62: The effect of spectrum normalization on the convergence of the fitting procedure for a generated spectrum with P= 0. A 0.5% error in normalization produces a false but robust minimum in 2, which would lead to an erroneous inferred polarization of 0.24. From Bugoslavsky et al [31].

PAGE 175

163 can be understood on the basis of a simple mathematical argument. Start with the observation that the slope of the 2(P) curve is equal to zero at P=0 when correctly normalized. This can be proved as follows: Let G1(V) be the experimental spectra of a point-contact with Z>0 and P>0, and let G2(V) be the fitting function with the BT K model. Assume that the correct normalization is unknown, so that the function G1(V)(1+ ) is being fitted, where | | << 1. Consider the 3-parameter fitting with fixed Ptrial so that the best fit function 0 2Gdelivers minimum to the target function: dV V P G V G dV V G V G Ptrial Z trial2 0 2 1 2 2 1 , 2)] ; ( ) 1 )( ( [ )] ( ) 1 )( ( [ min ) ( Due to the degeneracy of the spectra pr oduced by the BTK model at small P, G2(Ptrial = 0) yields a perfect fit (to within the numerical accuracy) to G1. However, with 0 the term G1 is very dissimilar in sh ape to the trial function, as the former has very low amplitude at high bias compared to the latter (the ratio is :1). The slope of the function 2(P) is easily obtaine d by differentiation: dV P G V P G V G P P0 2 0 2 1 2)] ; ( ) 1 )( ( [ 2 ) (

PAGE 176

164 At zero temperature and within the gap voltage, the polarized spectrum G(Z, P) is obtained from the classical zero-P BTK function as G(Z, P) = GBTK(Z)(1-P). Thus we can approximate the derivative under the integral as: BTKG P G 0 2 Due to the (almost) perfect fit between G1 and 0 2G only one non-negligible term remains: dV G V G P PBTK] ) ( [ 2 ) (1 2 The product of the two spectra is always pos itive; hence the sign and the magnitude of the derivative of 2(P) is dermined by The derivative becomes zero in the case of precise normalization; i.e. when = 0. Thus our procedure will be to program our fitting algorithm to look for a zero slope at P=0 via an iterative loop. Once we have established the cu rve with the correct normalization, we can then program our code to look for a horizontal slope at some other point, namely 2(Pmin) which will give us the correct value of P. This will get around both problems of ambiguous fitting with multi-parameter models and arbitrary normalizations leading to inco rrect P values. We have al ready reproduced figures 61 and 62 to verify our fitting routine and our 2 method with BugoslovskyÂ’s results. Initial analysis of our own experimental data is shown in figure 63 featuring 2 (P) vs. P for a sample of LBMO and using Sn as the superconducting probe. Notice the horizontal slope at P=0 and the minimum value of P at 0.60.

PAGE 177

165 0.040 0.045 0.050 0.055 0.060 0.065 0.070 2 vs. P Sn-LaBaMnO3 Junction2P Figure 63. 2 (P) vs. P for a sample of LBMO and using Sn as the superconducting probe. Notice the horizontal slope at P=0 a nd the minimum value of P at 0.60. From D. Rabson, D. Lovelady, and J. Sanders.

PAGE 178

166 7.2 Instrumentation improvements and future experiments The PCAR measurement system developed by us has been shown to be a very unique and versatile apparatus for investigation of spin polarization in materials. The flexibility afforded by integration with the PPMS with its variable magnetic field capability and large range of temperatures, makes the probe us eful for many types of experiments. The probe will be improved with some engi neering modifications and electronics improvements. Our PCAR probe has alrea dy produced successful results for several series of half-metallic oxides. Collaborators have expressed interest in this experiment and several types of materials will be inve stigated in the futu re for studies with conventional thin films and bulk samples as well as carbon nanotube based devices and other nanostructured materials. 7.2.1 Sample stage improvements Another modification will be made to the probe itself by adding another machined part to improve the translation stage. A third adjustment in design and engineering will improve this measurement system even more. The engineering modifications described in Appendix C have al ready given control ov er coarse and fine translation. Another copper pa rt is being engineered to c onnect the upper and lower parts of the assembly, which will recover the differe ntial micrometer capabilities. Thus the desired coarse and differential translation wi ll replace the current coar se/fine translation. Piezoelectric technology and/or stepper motors could also be added in the future to have more precise control over the resistan ce and oxide penetration at the junction.

PAGE 179

167 7.2.2 Incorporation of a lock-in amplifier The dI/dV can directly be measured by a st andard lock-in technique [14]. A lockin amplifier will be integrat ed into the data acquisition system for direct conductance measurement using an AC modulation technique. This technique will improve the quality of the data by increasi ng signal to noise ratio. Th e lock-in amplifier could be used as an output for a small AC signal th at could be mixed with a simple modulation circuit with the DC current sent through the point-contact junction. In this way small changes in I-V would be directly measured a nd dI/dV data could be collected in real-time as well. A new version of the Labview program will be written to integrate the lock-in amplifier with the existing data acquisition instruments. 7.2.2 Future spin transport experiments Of immediate interest is to apply this e xperiment to nanostructured materials. A collaborator at Carnegie Mellon University is interested in exploring spin transport in arrays of magnetic nanoparticles. We e xpect to explore Coulomb charging effects through I-V measurements across these structur es which can be viewed as an array of quantum dots. Our lab also plans to continue spin transport experiments on superconcuctor/nanotube/ferromagnet sample s, superconductor/nanotube/half-metal samples, and superconductor/nanotube/superc onductor samples to investigate Andreev signals and Josephson junctions across the nanotube interconnect. With the availability of FIB and e-beam lithography at the NNRC at USF, much smaller gaps across a carbon

PAGE 180

168 nanotube network would be achievable. In addition, these series of experiments could also be performed on single MWNTs and SWNTs in the future. An interesting subset of experiments would be to investigate the spin transport through a si ngle defect-free carbon nanotube as a function of distance between the superconducting electr ode and half-metal electrode. With better imaging techni ques, lithographic techniques and AFM manipulation capabilities the potential of thes e spin transport measurements will be very promising.

PAGE 181

169 References [1] S.A. Wolf, D. D. Awschalom, R.A. Buhrman, J. M. Daughton, S. von Molnar, M. L. Roukes, A. Y. Chtchelkanova, D. M. Tregar “Spintronics: A Spin-Based Electronics Vision for the Future,” Science 294 1488 (2001). [2] R. Saito, G. Dresselhaus, M. S. Dre sselhaus, “Physical Properties of Carbon Nanotubes”, Imperial College Press, 2003. [3] K. Hata et al. “Water-Assisted Highly Efficient Sythesis of Impurity-Free SingleWalled Carbon Nanotubes”, Science 306 19 November 2004, p. 1362-1364. [4] W. J. DeSisto, P. R. Broussard, T. F. Ambrose, B. E. Nadgorny, and M. S. Osofsky, “Highly spin-polarized chromium dioxide thin films prepared by chemical vapor deposition from chromyl chloride.” Appl. Phys. Lett. 76 3789, (2000). [5] S. Witanachchi, K. Ahmed, P. Sakthivel, and P. Mukherjee. “Dual-Laser ablation for particulate-free film growth,” Appl. Phys. Lett ., 66 1469, (1995). [6] H. Srikanth, J. Wiggins and H. Rees “Radio-frequency impedance measurements using a tunnel-diode os cillator technique.” Rev. Sci. Instrum 70 3097 (1999) [7] H. Srikanth, R. Hajndl, C. Chirinos, J. Sanders. “Magnetic studies of polymer-coated Fe nanoparticles synthesized by microwave plasma polymerization,” Appl. Phys. Lett. 79 3503 (2001) [8] R. A. de Groot and K. H. J. Buschow, “New Class of Materi als: Half-Metallic Ferromagnets.” Phys. Rev. Lett., 50 2024, (1983). [9] K. Schwartz, “CrO2 predicted as a half-metallic ferromagnet.” J. Phys. F: Met. Phys 15 L211, (1986). [10] E. Kulatov and I. I. Mazin, “Extende d Stoner factor calculations for the halfmetallic ferromagnets NiMnSb and CrO2.” J. Phys.: Condens. Matter. 2 343, (1990). [11] S. P. Lewis, P. B. Allen, T. Sasaki “Band structure and tr ansport properties of CrO2.” Phys. Rev. B 55 10253, (1997).

PAGE 182

170 [12] M. A. Korotin and V. I. Anisimov, “CrO2: A Self-Doped Double Exchange Ferromagnet.” Phys. Rev. Lett. 80 4305, (1998). [13] P. M. Tedrow and R. Meserv ey, “Spin-Polarized Electron Tunneling,” Physics Reports 238 173, (1994). [14] R. J. Soulen Jr., J. M. Byers, M. S. Osofsky, B. Nadgorny, T. Ambrose, S. F. Cheng, P. R. Broussard, C. T. Tanaka, J. Nowa k, J. S. Moodera, A. Berry, J. M. D. Coey, “Measuring the Spin Polarization of a Meta l with a Superconducti ng Point Contact”. Science 282 85 (1998). [15] R. A. de Groot and K. H. J. Bu schow, “Recent Developments in Half-Metallic Magnetism,” J. Magn. Magn. Mater 54-7 1377, (1986) [16] M. Penicaud, B. Siberchicot, C. B. So mmers, and J. Kubler, “Calculated electronic band structure and magnetic moment of ferrites,” J. Magn. Magn. Mater 103 212, (1992) [17] L. Ozyuzer, J. F. Zasadzinski, and K. E. Gray, “Point cont act tunneling apparatus with temperature and magnetic field control,” Cryogenics 38 911, (1998). [18] G. T. Woods et al, “A nalysis of point-contact Andreev reflection spect ra in spin polarization measurements.” Phys. Rev. B 70 054416 (2004). [19] J. Sanders et al, “Spi n Polarization measurements on polycrystalline strontium ruthenates using point-cont act Andreev reflection,” J. of App. Phys 97 10, 10C912/1-3 (2005) [20] J. Sanders, R. Hajndl, H. Srikanth, A. Houssam, P. Mukherjee, S. Witanachchi, “Static and dynamic magnetic studies of Fe3O4 thin films.” American Physical Society March Meeting. Oral Presentation, Sessi on W15.007, Indianapolis APS, March 2002. [21] J. Sanders, R. Hajndl, H. Srikanth, “Magneto-transport measurements in magnetic oxide thin films” University of South Flor ida Graduate Research Symposium, February 2002. [22] J. Sanders, P. Poddar, H. Srikant h, “Point contact Andr eev Reflection (PCAR) investigation of half-metal thin films w ith variable temperature and magnetic field control.” American Physical Society March Meeting. Oral Presentation, Session N28.008, Austin, TX, March 2003. [23] J. Sanders, P. Poddar, H. Srikant h, “Point-contact Andreev Reflection (PCAR) investigation of half-metal thin films w ith variable temperature and magnetic field control.” University of South Florida Graduate Res earch Symposium, April 2003

PAGE 183

171 [24] J. Sanders, G. T. Woods, H. Srikan th, “Spin-polarization measurements in La1-x(Ba, Ca, Sr)xMnO3 using point contact Andreev Reflectio n” American Physical Society March Meeting. Oral Presentation, Sessi on A23.011, Montreal, Quebec, March 2004. [25] J. Sanders, G. T. Woods, C. Siyambalap itya, H. Srikanth, P. Poddar, S. Kolensnik, B. Dabrowski, “Spin-polarization measurements on SrRu0.92O3 and SrRu0.8Ti0.2O3 using point contact Andreev Reflection” American Physical Society March Meeting. Oral Presentation, Session X42.011, Los Angeles, CA, March 2005. [26] Patent Pending: USF ref #03A008PR_Ha riharan, USF patents and license office reference #1372.140, intellectual property disclosure. [27] A. F. Andreev, “The Thermal C onductivity of the Intermediate State in Superconductors,” Sov. Phys. JETP 19 1228 (1964). [28] P. M. Campbell et al, “Simple catalys t for the growth of small-diameter carbon nanotubes.” Appl. Phys. Lett. 81 24, 2003, p. 4586-4588. [29] J. P. Novak et al, “Nerve agent dete ction using networks of single-walled carbon nanotubes.” Appl. Phys. Lett. 83 19, 2003, p. 4026-4028 [30] E. S. Snow et al, “Chemical Dete ction with a Single-Walled Carbon Nanotube Capacitor.” Science 307 25 March 2005, p. 1942-1945. [31] Y. Bugoslavsky et al, “Possibilities and limitations of point-contact spectroscopy for measurements of spin polarization.” Phys. Rev. B 71 104523, (2005) [32] R. Soulen, M. Osofsky, G. Woods, J. Sanders “Measurement of Spin Polarization of doped strontium ruthenates using point-contac t Andreev Reflection,” To be submitted. [33] M. Baibich, et al., “Giant Magnet oresistance of (001) Fe/(001) Cr Magnetic Superlattices,” Phys. Rev. Lett 61 2472 (1988). [34] L. Spinu, H. Srikanth, A. Gupta, X. W. Li, Gang Xiao, “Probing magnetic anisotropy effects in epitaxial CrO2 thin films,” Phys. Rev. B 62 8931 (2000) [35] P. C. Canfield and G. W. Crabtree. “Magnesium Diboride: Better Late than Never,” Physics Today 56 34 (2003). [36] Introduction to Solid State Physics, Charles Kittel (John Wile y and Sons, Inc., New York, 1996.)

PAGE 184

172 [37] R. S. Kaizer, S. T. B. Goennenwein, T. M. Klapwijk, G. Miao, G. Xiao, and A. Gupta. “A spin triplet supercurrent through the half-metallic ferromagnet CrO2,” Nature 439 825 (2006). [38] G. Blonder, M. Tinkham, and T. Kl apwijk, “Transition from metallic to tunneling regimes in superconducting microconstrictions: Excess current, charge imbalance, and supercurrent conversion.” Phys. Rev. B, 25 4515 (1982). [39] J. Nagamatsu et al., “Superconduc tivity at 39 K in magnesium diboride,” Nature 410 63 (2001). [40] D. C. Worledge and T. H. Geballe “Negative Spin-Pol arization of SrRuO3,” Phys. Rev. Lett 85 5182 (2000). [41] C. B. Eom et al., Science 258 1766 (1992). [42] B. Dobrowski et al., “Reduced fe rromagnetic transition temperatures in SrRu1– vO3 perovskites from Ru-site vacancies,” Phys. Rev. B 70 014423 (2004). [43] A. Kanbayasi, J. Phys. Soc. Jpn 41 1879 (1996). [44] I. I. Mazin and D. J. Singh, “Elect ronic structure and magnetism in Ru-based perovskites,” Phys. Rev. B 56 2556 (1997). [45] D. J. Singh, J. Appl. Phys. 79 8 (1996). [46] K. Yamaura, D. P. Young, and E. Takayama-Muromachi, eprint cond-mat/0311228. [47] C. Bansal, H. Kawanaka, R. Takaha shi, and Y. Nishihara, “Metal–insulator transition in Fe-substituted SrRuO3 bad metal system,” J. Alloys Compd 360 47 (2003). [48] I. I. Mazin, A. A. Golubov, and B. Nadgorny, “Probing spin polarization with Andreev reflection: A theoretical basis,” J. Appl. Phys 89 7576 (2001). [49] B. Nadgorny, M. S. Osof sky, D. J. Singh, G. T. Woods, R. J. Soulen, Jr., M. K. Lee, S. D. Bu, and C. B. Eom, Appl. Phys. Lett 82 427 (2003). [50] P. Rayshaudhuri, A. P. Mackenzie, J. W. Reiner, and M. R. Beasley, “Transport spin polarization in SrRuO3 measured through point-c ontact Andreev reflection,” Phys. Rev. B 67 020411 (2003). [51] Y. Ji, G. J. Strijkers, F. Y. Yang, C. L. Chien, J. M. Byers, A. Anguelouch, G. Xiao, and A. Bupta, Phys. Rev. Lett 86 5585 (2001).

PAGE 185

173 [52] N. Auth, G. Jakob, T. Block, and C. Fe lser, “Spin polarizati on of magnetoresistive materials by point contact spectroscopy,” Phys. Rev. B 68 024403 (2003). [53] M. B. Salamon and M. Jaime “The phys ics of manganites: Stru cture and transport,” Rev. Mod. Phys 73 583 (2001). [54] P. G. de Gennes, Phys. Rev. 118 141 (1960) [55] A. J. Millis, P. B. Littlewood, and B. I. Shraiman, Phys. Rev. Lett 74 5144 (1995). [56] H. Roder, J. Zang, and A. R. Bishop, Phys. Rev. Lett. 76 1356 (1996), Phys. Rev. Lett. 76 4987 (1996). [57] A. J. Millis, R. Mueller, and B. I. Shraiman, “Fermi-liquid-to-polaron crossover. I. General results,” Phys. Rev. B 54 5389 (1996). [58] A. J. Millis, R. Mueller, and B. I. Shraiman, “Fermi-liquid-to-polaron crossover. II. Double exchange and the physics of colossal magnetoresistance,” Phys. Rev. B 54 5405 (1996). [59] J. Appel, Solid State Physics 21 193 (1968) [60] S. Blundell, Magnetism in Condensed Matter (Oxford, 2001). [61] J. L. Cohn, “Electrical and Therma l Transport in Perovskite Manganites,” J. Supercond.: Incorporating Novel Magn. 13 291 (2000). [62] S. Kolesnik, B. Dabrowski, Z. Bukow ski, and J. Mais, “Enhancement of the lowfield magnetoresistance of La1–x–yCaxSryMnO3 + d at room temperature,” J. Appl. Phys 89 1271 (2001) [63] J. H. Park, C. T. Chen, S. W. Che ong, W. Bao, G. Meigs, V. Chakarian, and Y. U. Idzerda, Phys. Rev. Lett 76 4215 (1996). [64] M. S. Osofsky, B. Nadgorny, R. J. S oulen, Jr., P. Broussard, M. Rubenstein, J. Byers, G. Laprade, Y. M. Mukoskii, D. Shulyatev, and A. Arsenov, J. Appl. Phys 85 5567 (1999). [65] B. Nadgorny, I. I. Mazi n, M. Osofsky, R. J. Soulen, Jr., P. Broussard, R. M. Stroud, D. J. Singh, V. G. Harris, A. Arsenov, and Y. Mukovskii, “Origin of high transport spin polarization in La0.7Sr0.3MnO3: Direct evidence for minority spin states,” Phys. Rev. B 63 184433 (2001).

PAGE 186

174 [66] L. M. Rodriguez-Martin ez and J. P. Attfield, “Cation disorder and size effects in magnetoresistive manganese oxide perovskites,” Phys. Rev. B 54 R15622 (1996). [67] W. E. Pickett and D. J. Singh, “Electronic structure an d half-metallic transport in the La1-xCaxMnO3 system,” Phys. Rev. B 53 1146 (1996). [68] J. H. Park, E. Vescovo, H. J. Kim, C. Kwon, R. Ramesh, and T. Venkatesan, “Direct evidence for a half-metallic ferromagnet,” Nature 392 794 (1998). [69] Z. F. Ren (Boston College). The invent or of the plasma enha nced chemical vapor deposition (PECVD) process for a ligned growth of carbon nanotubes. [70] P. M. Campbell et al, “Simple catalys t for the growth of small-diameter carbon nanotubes.” Applied Physics Letters 81 24, 2003, p. 4586-4588. [71] J. P. Novak et al, “Ner ve agent detection using networ ks of single-w alled carbon nanotubes.” Applied Physics Letters 83 19, 2003, p. 4026-4028 [72] E. S. Snow et al, “Chemical Det ection with a Single-Walled Carbon Nanotube Capacitor.” Science 307 25 March 2005, p. 1942-1945. [73] E. S. Snow et al, “Ra ndom networks of carbon nanotubes as an electronic material.” Applied Physics Letters 82 13, 2003, p. 2145-2147. [74] E. S. Snow et al, “C arbon nanotubes networks: Nano materials for macroelectronic applications.” J. Vac. Sci. Technol B 22 4, Jul/Aug 2004, p. 1990-1994. [75] J. Kong et al., “Quantum Interference and Ballistic Transmission in Nanotube Electron Waveguides,” Science 87 622 (2000) [76] L. Valentini et al., “Sensors for sub-ppm NO2 gas detection based on carbon nanotube thin films,” Appl. Phys. Lett. 82 961 (2003). [77] T. Someya, J. Small, P. Kim, C. Nu ckolls, J. T. Yardley, “Alcohol Vapor Sensors Based on Single-Walled Carbon Nanotube Field Effect Transistors,” Nano Lett. 3 877 (2003). [78] J. Li et al., Carbon Nanotube Sens ors for Gas and Organic Vapor Detection,” Nano Lett. 3 929 (2003). [79] A. Goldoni, R. Larciprete, L. Pet accia, S. Lizzit, “Single-Wall Carbon Nanotube Interaction with Gases: Sample Contaminants and Environmental Monitoring,” J. Am. Chem. Soc 125 11329 (2003).

PAGE 187

175 [80] P. Qi et al., “ Toward Large Arrays of Multiple x Functionalized Carbon Nanotube Sensors for Highly Sensitive and Se lective Molecular Detection,” Nano Lett 3 347 (2003). [81] L. Valentini et al., “Highly sens itive and selective se nsors based on carbon nanotubes thin films for molecular detection,” Diamond and Rel. Mat. 13 1301 (2004). [82] J. P. Novak et al, “Ner ve agent detection using networ ks of single-w alled carbon nanotubes.” Applied Physics Letters 83 19, 2003, p. 4026-4028 [83] R. J. Chen et al., “Noncovalent functionalization of carbon nanotubes for highly specific electronic biosensors,” Proc. Nat. Acad. Sci. U. S. A. 100 4984 (2003). [84] Y. Lin, F. Lu, Y. Tu, Z. Ren, “G lucose Biosensors Based on Carbon Nanotube Nanoelectrode Ensembles, ” Nano Lett. 4 191 (2004). [85] E. S. Snow, J. P. Novak, P. M. Ca mpbell, D. Park, “Random networks of carbon nanotubes as an electronic material,” Appl. Phys. Lett. 82 2145 (2003). [86] P. G. Collins, M. S. Fuhrer, A. Zettl, “1/f noise in carbon nanotubes,” Appl. Phys. Lett 76 894 (2000). [87] E. S. Snow, J. P. N ovak, M. D. Lay, F. K. Perkins, “1/f noise in single-walled carbon nanotube devices,” Appl. Phys. Lett. 85 4172 (2004). [88] R. J. Chen et al., “An Investigation of the Mechan isms of Electronic Sensing of Protein Adsorption on Carbon Nanotube Devices, ” J. Am. Chem. Soc. 126 1563 (2004). [89] E. S. Snow, F. K. Perkins, E. J. Hous er, S. C. Badescu, T. L. Reinecke, “Chemical Detection with a Single-Walle d Carbon Nanotube Capacitor.” Science 307 1942 (2005). [90] D. R. Lide, CRC Handbook of Chemistry and Physics (CRC Press, Boca Raton, ed. 75, 1995), pp. 15 -43 – 15 -49. [91] G. M. Kosolapoff, J. Chem. Soc 3222 (1954). [92] R. A. McGill et al., "The ‘NRL-SAWRHINO’: a nose for toxic gases,” Sens. Actuators B 65 10 (2000). [93] R. E. Shaffer, S. L. Pehrsson, R. A. McGill, “A comparison study of chemical sensor array pattern re cognition algorithms,” Anal. Chim. Acta 384 305 (1999). [94] Bachtold, A. et al., “Aharonov-Bo hm oscillations in carbon nanotubes,” Nature 397 673-675 (1999).

PAGE 188

176 [95] Frank, S. et al., “Car bon nanotube quantum resistors,” Science 280 1744-1746 (1998). [96] M. S. Ferreira and S. Sanvito, “Contact -induced polarization in carbon nanotubes,” Phys. Rev. B 69 035407 (2004). [97] S. Ryu, Y. Hatsugai, “Correlation effects of carbon nanotubes at boundaries: Spin polarization induced by zero -energy boundary states,” Phys. Rev. B 67 165410 (2003). [98] A. Kleiner, “Chiral spin current s and quantum Hall effect in nanotubes,” Phys. Rev. B 67 155311 (2003). [99] L. Balents, R. Egger, “Spin Trans port in Interacting Quantum Wires and Carbon Nanotubes,” Phys. Rev. Lett. 85 3464 (2000). [100] L. Balents, R. Egger, “Spin-de pendent transport in a Luttinger liquid,” Phys. Rev. B 64 035310 (2001). [101] C. S. Peca, L. Balents, “Fabry-P erot interference and spin filtering in carbon nanotubes,” Phys. Rev. B 68 205423 (2003). [102] A. De Martino et al., “Spin-Orbit Coupling and Electron Spin Resonance Theory for Carbon Nanotubes,” Phys. Rev. Lett. 88 206402-1 (2002). [103] S. Kokado, K. Harigaya, “Magnetoresistance effect in spin-polarized junctions of ferromagnetically contacting multiple conductive paths: Applications to atomic wires and carbon nanotubes,” Phys. Rev. B 69 132402 (2004). [104] P. E. Lindelof et al., “Electron Sp in in Single Wall Carbon Nanotubes,” Physica Scripta. T102, 22-29, 2002. [105] K. Tsukagoshi, B. W. Alphenaar, H. A go, “Coherent transport of electron spin in a ferromagnetically contacted carbon nanotube,” Nature 401 572 (1999). [106] K. Tsukagoshi, B. W. Alphenaar, M. Wagner, “Spin transport in nanotubes,” Mat. Sci. & Eng. B B84 26-30 (2001). [107] B. W. Alphenaar, K. Tsukagoshi, M. Wagner, “Spin transport in nanotubes (invited),” Journal of Applied Physics 89 6863 (2001). [108] B. Zhao et al., “Spin-coherent tr ansport in ferromagnetically contacted carbon nanotubes,” Appl. Phys. Lett 80 3144 (2002).

PAGE 189

177 [109] B. Zhao et al., “Spin-dependet n transport in multiw alled carbon nanotubes,” Journal of Applied Physics 91 7026 (2002). [110] D. Orgassa, G. J. Mankey, H. Fujiwar a, “Spin injection into carbon nanotubes and a possible application in spin-resolved scanning tunnelling microscopy,” Nanotechnology 12 281 (2001) [111] S. Sahoo, T. Kontos, C. Schonenberger, an d C. Surgers, “Electrical spin injection in multiwall carbon nanotubes with tran sparent ferromagnetic contacts,” Appl. Phys. Lett. 86 112109 (2005). [112] X. Hoffer et al., “Spin-dependen t magnetoresistance in multiwall carbon nanotubes,” Europhysics Letters 67 103 (2004). [113] Jae-Ryoung Kim et al., “S pin-dependent transport prop erties in a single-walled carbon nanotube with mesoscopic Co contacts,” Phys. Rev. B 66 233401 (2002). [114] A. K. Raychaudhuri, K. P. Rajeev, H. Srikanth, and N. Gayath ri, “Metal-insulator transition in perovskite oxides : Tunneling experiments,” Phys. Rev. B 51 7421 (1995).

PAGE 190

178 Bibliography Ginzburg, V. L. and Andryushm, E. A. Supe rconductivity. World Scie ntific Press, New Jersey, 2004. Ji, Y. et al. “Measurement of spin polarization of singl e crystals of La0.7Sr0.3MnO3 and La0.6Sr0.4MnO3”. Phys. Rev. B, 66 2410, (2002). Kresin, V. Z. and Wolf, Stuart A. Funda mentals of Superconductivity. Plenam Press, New York, 1990. Osofsky, M. S. et al. “Measurement of the transport spin-polariza tion of oxides using point contact Andreev reflec tion (PCAR),” Materials Sc ience and Engineering B, B84 49 (2001). Wolf, E. L. Principles of Electron Tunne ling Spectroscopy, Oxford University Press, New York, 1985.

PAGE 191

179 Appendices

PAGE 192

180 Appendix A: CAD of rotating base for T and L experiments

PAGE 193

181 Appendix A (Continued)

PAGE 194

182 Appendix B Instrumentation for our home-built PCAR probe Point-Contact Andreev Reflection (PCAR) instrumentation A versatile measurement system has been developed that integrates with a commercial PPMS system and can be used for conducting point-contact tunneling as well as PCAR experiments. The system consists of a cryostat probe, Keithley sourcemeter and nanovoltmeter, PPMS system and Labview programs to control all instruments. The sourcemeter and nanovoltm eter supply current and measure voltage, respectively. The nanovoltmeter is advantag eous as the measurement requires resolution of superconducting gaps of the order of a fe w millivolts. These two instruments allow acquisition of I-V data which are then differentiated numeri cally with EXCEL to plot and analyze dI/dV vs. V curves. Obtaini ng accurate dI/dV data is important in calculating the conductance spectra and th e spin polarization of the material. Labview software programming controls the desired current range, acquisition of voltage data, number of data points taken, as well as te mperature and magnetic field control of the PPMS. It also controls data management by graphing the I vs. V curves in real-time on the PC monitor and sends the data to a separate file to be analyzed in EXCEL and ORIGIN. Figure A5 shows a schema tic of the experimental setup and figure A6 shows a photograph of the probe housing mated to the PPMS.

PAGE 195

183 Figure A3: Schematic of initial setup for poi nt contact tunneling experiments. The I-V data is collected with Labview software and a Keithley sourcemeter and nanovoltmeter. The dI/dV conductance data is then acquire d through numerical differentiation in EXCEL and ORIGIN data analysis packages.

PAGE 196

184 Figure A4: A photograph of the PCAR probe, inserted into the PPMS cryostat Dewar. A 10-pin ceramic to glass connector compatible with cryogenic inserts, is used to connect two current leads and two voltage leads fr om the sample stage to the Keithley instruments.

PAGE 197

185 Cryostat PCAR probe: initial design The cryostat probe itself shall now be described in detail. The probe integrates with the aforementioned elec tronics and houses the sample the superconducting tip, and the connecting wires to allow I-V data to be collected. The probe schematics were engineered to match the specification of the PPMS dewar in diameter, length, and vacuum seal requirements. The body of the pr obe was machined from the same material used in other PPMS inserts commerci ally available from Quantum Design. Overall design The probe consists of a copper housing for the sample space and differential micrometer, a garolite tube encasing a stainl ess steel rod for rotation of the micrometer, and a vacuum-sealed aluminum housing with electrical connectors and micrometer knob (figure A7). The copper housing has been m achined with 27 threads per inch (TPI) to hold a matched titanium lug. The bottom half of the titanium lug ha s been machined to 28 TPI to fit into the tip holder. The titanium tip holder has a set screw and a 1/16” bore made to match a gold sheath that holds the tin, lead, or niobium tip. Before soldering to the sheath, the tip is formed into a cone-shaped point with a gri nder. Then several grades of 3M diamondaluminum oxide lapping paper are used to sand the tip to a sharp point. The tip is then thoroughly cleaned with acetone and/or meth anol before being mounted above the sample. Finally, two copper wi res are connected onto the uppe r portion of the tip using

PAGE 198

186 Figure A5: Point Contact Andr eev Reflection cryostat probe. 1. Resistivity puck 2. Copper housing and differential micrometer 3. Garolite hollow rod and spacers with inner stainless steel rod 4. upper vac uum-sealed aluminum housing with 10-pin connector. 1 2 3 4

PAGE 199

187 Indium and/or Sn/Pb alloy solder. With this method, two contacts are made to the tip and two to the sample. Since a considerable amount of time wa s spent in engineering the design and development of the PCAR probe, we now pres ent construction details of each stage. Several innovative modifications were neces sary before a perfect working probe was realized. Our design has generated quite so me interest in the research community pursuing similar experiments. Engineers at Quantum Design have also expressed interest in our probe. This has led to filing a pr ovisional patent on our instrumentation. Lower copper housing The lower portion of the sample housing has a standa rd PPMS resistivity puck screwed directly into the lo wer portion of the copper housi ng (figure A8). The sample itself is mounted on the puck with GE varnis h and/or vacuum greas e, and the two leads are connected with indium solder, Sn/Pb a lloy solder, or Ag conductive ink. Having two leads on the point and two lead s on the sample allows a st andard four-probe currentvoltage measurement and the I-V characteri stics across the tip/sample junction can be measured accurately. The differential micr ometer allows the tip to be translated up and down the shaft (away and toward the sample) with very fine control. This serves the dual purpose of changing the gap distance between the tip a nd sample and thus the junction resistance and also being able to drive th e tip to pierce any native oxide layer that is present. The latter is the case when working with CrO2 since it is well-known to build up an anti-

PAGE 200

188 ferromagnetic insulating layer of Cr2O3 in the presence of oxygen gas. This apparatus has the advantage of being able to use the mi crometer along with a short, sturdy, sharp tip to penetrate this barrier and make contact with the ferromagnetic metallic CrO2. The micrometer was constructe d of titanium due to its low thermal coefficient of expansion and the fact that experimental temperatures range dow n to 2.0 Kelvin. The micrometer is machined with 27 TPI and 28 TP I cut onto the same shaft, hence the name “differential micrometer” (figure A9). The l ug of the differential micrometer translates down relative to the copper housing and the tip holder translates up relative to the lug. Thus the difference between 27 TPI and 28 TPI of the two pieces is the translation of the lug relative to the copper housing. This eas ily works out to be 1/27-1/28 = 0.0013” or 1.3 thousandths of an inch per revolution. Ther e are 12 revolutions avai lable in the lug, so the experimentalist has about 16 thousandths tran slation to work with. This range allows several different gap resistances to be esta blished for a given junction. (Note that by having a large diameter knob at the top, one full revolution can further be fine tuned to achieve smaller translation). The differential micrometer ha s a 3/16” diameter stainle ss steel rod connected to the lug with a set screw. The rod then exte nds upward though a garo lite tube, through the aluminum housing, and into a knob at the very t op. The garolite tube also serves to hold spacers for placement into the PPMS neck and for wrapping coated copper wires from the tip and sample to the upper connectors. Garoli te was the material of choice because of its lighter weight, ease of mach inability and also its low thermal conductivity that is essential for effective cryogenic probes to cu t down heat transfer to the sample region.

PAGE 201

189 Figure A6: Lower copper housing of PCAR probe. The standard resistivity puck (Quantum Design) can be seen in the lo wer left. The copper housing contains the titanium differential micrometer a nd a shank for holding the tip itself.

PAGE 202

190 Figure A7: Exploded view of the lower copper housing of the PCAR probe. The differential micrometer is easily seen here The two sections are machined with 27 threads per inch and 28 threads per inch, resulting in a small translation of 1.3 thousandths per revolution.

PAGE 203

191 Upper aluminum vacuum housing The uppermost component of the probe is a vacuum-sealed aluminum housing with a ten-pin and eight-pin connector, seve ral O-ring seals, and the rotation knob (figure A10). The seal around the rotating rod cont ains a specifically chamfered O-ring for rotation along with a standard O-ring underneath the aluminum disc. This is generally referred to as the Wilson seal arrangement. Each of the four side panels containing the connectors has a counter-sunk groove with a 1 ” outer diameter (OD) and 1.150” ID oring. Each SMA connector itself also has an o-ring seal while the 10-pin and 8-pin connectors have commercial Torr Seal between the mount and the panel. The bottom of the aluminum housing has been precisely machined to an inner diameter of 1 ” to match the commercial f itting from a standard Quantum Design probe. A double o-ring system has been machined at this junction. These were challenging engineering and machining aspects due to the small clearance between the inner diameter of the commercial fitting and the outer diamet er of the garolite tube The technique of machining the inside of the housing with a cut ting tool was also a somewhat complex job. This housing was initially vacuum tested in the PPMS during a liquid helium run. The amount of vacuum grease on each o-ring wa s adjusted and the torque on each of the Allen screws was varied. A successful seal was achieved and s ubsequent experiments consistently achieved a very low pressure inside the sample space. This probe continually holds a pressure of 0.65 Torr at a temperature of 2.0 Kelvin. This remarkable stability at low temperatures is a successf ul outcome and vindicated the design approach for our home-built system.

PAGE 204

192 Figure A8: Upper aluminum housing. It is vacuum-sealed with a double O-ring Wilson seal configuration. The Wilson seal allows translation and rotati on of the center rod without breaking the vacuum.

PAGE 205

193 Cryostat PCAR probe: Engineering modifications A technical aspect of sample preparation nece ssitated an engineering modification of the PCAR probe. The differentia l micrometer gave very fine control, but there was no coarse translation capabilit y. Several experiments yielded no data whatsoever due to either failure to make cont act at all and thus no I-V characteristics or the opposite problem of cracking a sample due to too much compression between the sample and tip. To attack this problem, a new assembly for coarse translation was engineered, machined, tested, and im plemented into experimentation. Coarse translation in upper housing Conceptually, there was a need for a slider that would move the entire micrometer assembly at the bottom using coarse thr eads located in the upper aluminum housing (figure A11). Another knob would be added to the top of the probe to achieve coarse control while the original knob would still control the micrometer. Due to budgetary constraints and time constraints, only three parts could be machined (see figures A12A14). The new shaft that was attached to the upper aluminum housing and a knurled aluminum knob to control the coarse transl ation are both shown in figures A12 and A13, respectively. The new shaft was machined at 20 meaning 1/2” inner diameter and 20 threads per inch. The knob was machined with matching threads over the bottom 1 ” of it’s travel. This gives the e xperimenter 35 revolutions to wo rk with for wiring the tip and sample in the sample space and then moving th e sharpened tip into close proximity prior to the cooling process.

PAGE 206

194 Figure A9: Conceptual drawing of modi fications for coarse/fine translation.

PAGE 207

195 Figure A10: Mechanical drawings of new shaft on upper aluminum housing

PAGE 208

196 Figure A11: Mechanical drawings of knur led aluminum knob for coarse translation.

PAGE 209

197 Figure A12: Mechanical drawings of m odified lower copper housing and new copper sliding part.

PAGE 210

198 Figure A13: Overall, cutaway, and exploded views of new modifications

PAGE 211

199 Lower assembly slider The lower copper housing was previously th readed on its inner diameter to accept the upper portion of the titanium differential micrometer. In the modifications, a sliding part was engineered to move relative to this copper housing according to adjustments made in the upper coarse translation control. Thus the threads were removed and the copper housing was bored out to an inner diamet er of ” in order to mate with the new sliding part (figure A14). Init ially, it was thought that the sliding part would have two rectangular wings on its sides to prevent rotation during tr anslation (figure A11). This initial conceptual idea was chan ged, however, again du e to budgetary and time constraints. Instead, a cheaper and fast er solution was to cut rectangular grooves on the sides of the part and add two more set screws into the copper housing as shown in figure A14. This allows the entire microm eter assembly translate longitudinally along the length of the probe w ithout rotating or torquing at an angle to that axis. The inside of the copper sliding part was threaded with a 28 die to match with the existing titanium micrometer part. Thus the experimenter can move the entire assembly to a point-tip separation distance and then switch to the fine translation knob to establish different contact resistances before taking data. Eventually a third modification will be engineered, machined, and implemented in order to have coarse and differential translation. Note that all these modificat ions were incorporated with focus only on vertical translation i.e. the z-axis translat ion stage. Future plans are to add an x-y translation stage to achieve lateral motion so that point contact junctions can be made in situ at several places on the sample surface, a nd a step motor to drive the entire assembly.

PAGE 212

200 Appendix C: USF carbon nanotube CVD growth furnace

PAGE 213

201 Appendix D Journal Publications and Conference Presentations Publications: J. Sanders, G. T. Woods, D. C. Lovelady, J. Ga ss, H. Srikanth, and D. A. Rabson. “Point-Contact-Andreev-R eflection (PCAR) Investigati ons of the Double-Exchange Ferromagnetic Materials La1-x(Ba, Sr)xMnO3.” To be submitted E. S. Snow, F. K. Perkins, P. M. Campbell, J. Sanders, J. Robinson, “A single-walled carbon nanotube-based biosensor,” To be submitted E. S. Snow, F. K. Perkins, P. M. Campbell, J. Sanders, J. Robinson, “Chemical specificity in carbon nanotube capacitive sensors,” To be submitted J. Sanders, G. T. Woods, P. Poddar, H. Srikanth, B. Dobrowski and S. Kolesnik. “Spin polarization measurements on polycrystalline strontium ruthenates using point-contact Andreev reflection” Journal of Applied Physics 97 10, 10C912/1-3. G. Woods, R. J. Soulen, I. I. Mazin, B. Nadgorny, M. S. Osofsky, J. Sanders H. Srikanth, W. F. Egelhoff and R. Datla. “Ana lysis of Point-contact Andreev Reflection Spectra in Spin Polarization Measurements,” Physical Review B 70, 054416, 2004. R. Hajndl, J. Sanders H. Srikanth, and N. J. Dudney. “Growth and Characterization of BSTO/hexaferrite composite thin films,” Journal of Applied Physics 93 10, May 15, 2003, pp. 7999-8001. H. Srikanth, R. Hajndl, C. Chirinos, J. Sanders “Magnetic studies of polymer-coated Fe nanoparticles synthesized by microwave pl asma polymerization,” Applied Physics Letters, vol. 79, no. 21, November 19, 2001, pp. 3503 – 3505.

PAGE 214

202 Presentations: American Physical Society Meeting, Baltimore, MD, March 2006 Session N32-09 “Sensor applications and Spin Transport Measurements in Carbon Nanotube Nanocomposites” J. Sanders J. H. Gass, H. Srikanth, M. S. Osofsky, R. J. Soulen, F. K. Perkins, P. M. Campbell, E. S. Snow NATO Advance Studies Instit ute, Carbon Nanotubes, Sozopol, Bulgaria,May 2005 Invited Talk: “Growth, Deposition and Characterization of Carbon Nanotubes for Charge and Spin-based Sensor Applications” J. Sanders J. H. Gass, H. Srikanth, E. S. Snow, F. K. Perkins, P. M. Campbell American Physical Society Meeting, Los Angeles, CA, March 2005 Session X42-11 “Spin-Polari zation measurements on SrRu0.92O3 and SrRu0.8Ti0.2O3 using point contact A ndreev Reflection” J. Sanders G. T. Woods, C. Siyambalapitya, H. Srikanth, P. Poddar, S. Kolensnik, B. Dabrowski Office of Naval Research Outbriefing, NSWRC – Panama City, FL August 2004 “Analysis of Conductivity of Seawater a nd Saturated Seabottom for Open-loop and Closed-Loop Magnetic Minesweeping” J. Sanders N. Lassiter, M. Wagstaff, B. Williams American Physical Society Meeting, Montreal, Quebec March 2004 Session A23-11 “Spin-Polari zation measurements in La1-x(Ba, Ca, Sr)xMnO3 using point contact Andreev Reflection” J. Sanders G. T. Woods, H. Srikanth USF Graduate Research Symp. University of South Florid a, Tampa, FL April 2003 Poster Presentation: “Poi nt-contact Andreev Reflection (P CAR) investigation of halfmetal thin films with variable temperature and magnetic field control.” J. Sanders P. Poddar, H. Srikanth American Physical Society Meeting, Austin, TX March 2003 Session N28-8 “Point-contact Andreev Reflec tion (PCAR) investigat ion of half-metal thin films with variable temperat ure and magnetic field control.” J. Sanders P. Poddar, H. Srikanth American Physical Society Meeting, Indianapolis, IN March 2002 Session W15-7 “Static and dynamic magnetic studies of Fe3O4 thin films” J. Sanders R. Hajndl, H. Srikanth, A. Hou ssam, P. Mukherjee, S. Witanachchi

PAGE 215

About the Author Jeff Sanders received a bachelor’s de gree in physics and mathematics from the University of Wisconsin – Platteville in 1995 and an M.Ed. from the University of Wisconsin – LaCrosse in 1999. He was teaching high school math and physics in Mauston, WI while in the UWL Masters progr am, as well as coaching and officiating athletics. He continued teaching and coach ing until he entered graduate school at the University of South Florida in 2000, where he earned the M. S. in physics in 2003. While in the Ph.D. program at the Univ ersity of South Florida, Jeff was very active as a research assistant and laboratory teaching assistant and he also volunteered as a strength & conditioning coach for USF baseball. He taught at Hillsborough Community College as an adjunct faculty of physics and astronomy during six different semesters and has also remained active in th e American Association of Physics Teachers (AAPT) by attending regional and national meeti ngs. Jeff has presented research at five different March meetings of the American Physical Society (APS) and worked at the Naval Research Laboratory in Washington, D. C. through the Office of Naval Research (ONR) internship programs.


Download Options

Choose Size
Choose file type
Cite this item close


Cras ut cursus ante, a fringilla nunc. Mauris lorem nunc, cursus sit amet enim ac, vehicula vestibulum mi. Mauris viverra nisl vel enim faucibus porta. Praesent sit amet ornare diam, non finibus nulla.


Cras efficitur magna et sapien varius, luctus ullamcorper dolor convallis. Orci varius natoque penatibus et magnis dis parturient montes, nascetur ridiculus mus. Fusce sit amet justo ut erat laoreet congue sed a ante.


Phasellus ornare in augue eu imperdiet. Donec malesuada sapien ante, at vehicula orci tempor molestie. Proin vitae urna elit. Pellentesque vitae nisi et diam euismod malesuada aliquet non erat.


Nunc fringilla dolor ut dictum placerat. Proin ac neque rutrum, consectetur ligula id, laoreet ligula. Nulla lorem massa, consectetur vitae consequat in, lobortis at dolor. Nunc sed leo odio.