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 001681114
007 cr mnu|||uuuuu
008 060105s2005 flu sbm s000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0001000
Abou Mourad, Houssam.
Metallic to insulating transition in disordered pulsed laser deposited silicide thin films
h [electronic resource] /
by Houssam Abou Mourad.
[Tampa, Fla.] :
b University of South Florida,
Thesis (Ph.D.)--University of South Florida, 2005.
Includes bibliographical references.
Text (Electronic thesis) 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 128 pages.
ABSTRACT: A metal-to-insulating transition has been observed in iron, iron oxide, iron silicide and cobalt silicide thin films when deposited on Si substrate with a native SiOx layer. This transition produced a change in resistance of 5 orders of magnitude at a temperature of 250 K. To the best of the author's knowledge, this effect has not been reported in the literature prior to this study. This work reports a systematic experimental investigation carried out to understand the fundamental mechanism involved in the manifestation of this metal-to-insulator transition. The films were deposited using the pulsed laser deposition technique (PLD) in a base vacuum of the order of 10-6 torr at 400o C and room temperature. Several experiments were systematically conducted to understand the nature of the transition and the current path.Deposition of films on different substrates and the deposition of different transition metal films were made to narrow down the physical origin of the transition in the sample. Temperature-dependent resistance measurements not only exhibited a transition but also suggested more than one conduction mechanism. This is confirmed by the data collected for the IV curves. Current and voltage have a linear relation at temperatures greater than the transition temperature, and a non-linear relation at lower temperatures. Magnetoresistance (MR) measurements revealed a quadratic dependence of the resistance on the applied magnetic field. This is an indication that the MR observed is due to Lorentz forces acting on the charge carriers. Transmission electron microscopy and x-ray photoelectron spectroscopy have identified different layers that are believed to be responsible for the observed transition.
Adviser: Pritish Mukherjee.
Co-adviser: Sarath Witanachchi
t USF Electronic Theses and Dissertations.
Metallic to Insulating Transition in Disordered Pulsed Laser Deposited Silicide Thin Films by Houssam Abou Mourad 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 Co-Major Professor: Pritish Mukherjee, Ph.D. Co-Major Professor: Sarath Witanachchi, Ph.D. Robert S. F. Chang, Ph.D. Srikanth Hariharan, Ph.D. George S. Nolas, Ph.D. Date of Approval: April 15, 2005 Keywords: iv, fesi, cosi, temperature de pendence, resistance, magnetoresistance. Copyright 2005 Houssam Abou Mourad
Dedication I would like to dedicate this work to my wife, Myrna Y. Padn Abreu, for the unconditional support, and to my daughter Tamara Abou Mourad, for being an understanding 4 year old. I al so would like to dedicate this degree to my mother, Mountaha Abou Mourad, for always believing.
Acknowledgements I would like to thank Prit ish Mukherjee, Ph.D. and Sarath Witanachchi, Ph.D. for allowing me to do this work at their laborator y. I also would like to thank them for their guidance throughout my experience at th e University of South Florida. My thanks goes to my lab-mates, Robert Hyde, John Cuff and James Winslow for the help they have given me and the long heated discussions. My gratitude also goes to Srikanth Hariharan, Ph.D. for allowing me to use the PPMS for the MR measurements. IÂ’m also very grateful for the help provided by Gerald Woods, Ph.D. using the PPMS system. Last but not least, I want to thank David Rabson, Ph.D. for the many conversations I held with him and his advises.
i Table of Contents List of Figures............................................................................................................ii Abstract......................................................................................................................v ii Chapter 1. Introduction..............................................................................................1 Chapter 2. Iron Oxides...............................................................................................4 Chapter 3. Sample Depositi on and Characterization.................................................11 Section 3.1.Deposition...................................................................................11 Section 3.2. Sample Characterization............................................................19 Chapter 4. Experiments and Results of Iron Oxide...................................................25 Chapter 5. Iron Silicide..............................................................................................33 Chapter 6. Results......................................................................................................39 Section 6.1. Resistance De pendence on Temperature...................................39 Section 6.2. IV Characteristics.......................................................................69 Section 6.3. Resistance and Magneto resistance as Measured by PPMS.............................................................................................................84 Chapter 7. Conclusion and Suggestions.....................................................................106 References..................................................................................................................111 Bibliography..............................................................................................................116 About the Author.............................................................................................End Page
ii List of Figures Figure 2.1 Fe3O4 crystalline inverse spinel structure..........................................5 Figure 2.2 Spin polarized electron bands for F3O4..............................................6 Figure2.3 Spin projected one-electron dens ities of states and numbers of states for magnetite........................................................................7 Figure 2.4 Resistance properties of Fe3O4 and its dependence on thickness.............................................................................................8 Figure 2.5 Magnetoresistance for 6600 thick Fe3O4 films deposited on (100) MgO....................................................................................9 Figure 2.6 Resistivity (at 0 magnetic field) and magnetization (measured in 300 Oe) as a function of temperature for Fe3O4 film deposited on MgO (100)..................................................9 Figure 2.7 Heat capacity versus temperature plots of homogeneous single crystal Fe3O4.....................................................10 Figure 3.1.1 Depiction of the PLD system used to deposit the silicide films ................................................................................................12 Figure 3.1.2 Si crystallographic structure..............................................................15 Figure 3.1.3 Resistivity curve of B doped Si with a density of ~ 1015 atoms/cm3...........................................................................................16 Figure 3.2.1 Schematic of the voltage measurement system.................................22
iii Figure 4.1 Resistance measurements of a PLD deposited magnetite film ................................................................................................26 Figure 4.2 A standard 4-point probe meas urement showing a metallicinsulator transition.............................................................................27 Figure 4.3 Resistance dependence on temperature of Si doped iron oxide films..................................................................................28 Figure 4.4 Resistance dependence on temperature for Si doped iron oxide films deposited on Si in different oxygen ambient pressures.............................................................................................29 Figure 4.5 Illustration of spin orientat ion before and after Morin transition for Fe2O3............................................................................30 Figure 4.6 Results of resistance measurement versus temperature for iron oxide films (Fe2O3) deposited in a 2 mtorr O2 atmosphere.........................................................................................31 Figure 5.1 FeSi crystallographic structure..........................................................34 Figure 5.2 FeSi susceptibilities de pendence on temperature..............................34 Figure 5.3 FeSi resistivity dependence on temperature......................................35 Figure 6.1.1 Resistance dependence on temperature of an FeSi sample deposited at 400o C on (100) Si.........................................................40 Figure 6.1.2 Resistance dependence of FeSi on temperature for deposition on different substrates......................................................42 Figure 6.1.3 Dependence of the resistive transition on deposition temperature........................................................................................43
iv Figure 6.1.4 Resistance dependence on temperature of the FeSi film (FeSi/Sapphire), FeSi with in terface interactions (FeSi/Si) and the substrate by itself...................................................................45 Figure 6.1.5 (a) Atomic profile and (b) TEM image of an FeSi/Si sample deposited at 400o C............................................................................47 Figure 6.1.6 TEM and passive circuit model for the sample.................................48 Figure 6.1.7 Four-parameter data fit for resistance versus temperature curve for FeSi deposited on sa pphire to isolat e the filmÂ’s resistance characteristics....................................................................50 Figure 6.1.8 Subtraction solution to equation 5.1.3...............................................51 Figure 6.1.9 Measured voltages and results of modeling for an FeSi/Si sample based on the passive circuit network of figure 6.1.6.............52 Figure 6.1.10 Resistance dependence of FeSi deposited on SiOx/Si........................54 Figure 6.1.11 Effect of substrate resistivity on the transition..................................55 Figure 6.1.12 Resistance dependence on temperature of a discontinuous sample deposited on a Si substrate.....................................................57 Figure 6.1.13 Samples deposited using a contact mask...........................................59 Figure 6.1.14 Resistance versus temperature for four dot-masked sample..............60 Figure 6.1.15 Three samples of FeSi deposited under the same conditions...........................................................................................61 Figure 6.1.16 Resistance versus temperature curve dependence on in-situ sample annealing................................................................................62 Figure 6.1.17 600o C post annealing of an FeSi film...............................................64
v Figure 6.1.18 900o C post anneal study of an FeSi film...........................................66 Figure 6.1.19 Resistance dependence on current for FeSi films..............................67 Figure 6.1.20 The resistance dependence on temperature of different transition metals (TM) and TM silicides deposited on Si..................68 Figure 6.2.1 (a) IV characteristics of an FeSi film deposited on a Si substrate at 400o C and (b) the samples resistance dependence on temperature measured with a current of 100 A ................................................................................................70 Figure 6.2.2 IV curves of a discontinuous FeSi sample.........................................71 Figure 6.2.3 J-FET diagrams..................................................................................74 Figure 6.2.4 IV curves for the J-FET under various conditions.............................76 Figure 6.2.5 IV fit for intermediate temperatures..................................................77 Figure 6.2.6 IV curves for room and high temperature deposited samples............79 Figure 6.2.7 IV characteristic curve for a room temperature deposited sample ................................................................................................80 Figure 6.2.8 IV curve of a commercial pn junction...............................................81 Figure 6.3.1 Resistance versus temperature measured by the PPMS ................................................................................................85 Figure 6.3.2 PPMS current output..........................................................................88 Figure 6.3.3 Resistance and magnetore sistance dependence on temperature........................................................................................89 Figure 6.3.4 Applied magnetic field orientati on with respect to the film..............90 Figure 6.3.5 dR/dT versus temperature for an FeSi/Si sample..............................91
vi Figure 6.3.6 Percentage difference due to a temperature shift...............................92 Figure 6.3.7 Exponential decay fit near the transition region................................94 Figure 6.3.8 Magnetoresistance dependence on an out-of-plane applied magnetic field.....................................................................................95 Figure 6.3.9 Magnetoresistance dependence on in-plane applied magnetic field parallel to current path...............................................96 Figure 6.3.10 MR of CoSi/Si....................................................................................98 Figure 6.3.11 Current output dependence of the PPMS on the applied magnetic field.....................................................................................99 Figure 6.3.12 Resistance dependence on the applied magnetic field for an FeSi sample........................................................................................102 Figure 6.3.13 Resistance dependence on the in-plane applied field fit for an FeSi film........................................................................................103 Figure 6.3.14 Resistance and magnetore sistance dependence on temperature of FeSi/Si.......................................................................104 Figure 6.3.15 Magnetoresistance dependence on magnetic field for a 400o C deposited FeSi sample for 45 minutes............................................105
vii Metallic To Insulating Transition In Diso rdered Pulsed Laser Deposited Silicide Thin Films Houssam Abou Mourad ABSTRACT A metal-to-insulating transition has been observed in iron, iron oxide, iron silicide and cobalt silicide thin films when de posited on Si substrate with a native SiOx layer. This transition produced a change in resistance of 5 orders of magnitude at a temperature of ~ 250 K. To the best of the authorÂ’s knowledg e, this effect has not been reported in the literature prior to this study. This work repor ts a systematic experimental investigation carried out to understand the fundamental m echanism involved in the manifestation of this metal-to-insulator tran sition. The films were deposit ed using the pulsed laser deposition technique (PLD) in a ba se vacuum of the order of 10-6 torr at 400o C and room temperature. Several experiments were syst ematically conducted to understand the nature of the transition and the current path. Depositi on of films on different substrates and the deposition of different transition metal f ilms were made to narrow down the physical origin of the transition in the sample. Te mperature-dependent resistance measurements not only exhibited a transition but also sugge sted more than one conduction mechanism. This is confirmed by the data collected for the IV curves. Current and voltage have a linear relation at temperatures greater than the transition temperature, and a non-linear relation at lower temperatures. Magnetoresistance (MR) measurements revealed a
viii quadratic dependence of the resistance on the applied magnetic field. This is an indication that the MR observed is due to Lorentz forces acting on the charge carriers. Transmission electron microscopy and x-ray photoelectron spec troscopy have identifi ed different layers that are believed to be responsible for th e observed transition. X-ray diffraction (XRD) experiments were conducted to identify any crystalline ordering. F ilms that have been studied using XRD revealed th at the layers were amorphous.
1 Chapter 1 Introduction Fe3O4 (magnetite) is classified as a ha lf-metallic ferromagnet [Z. Zhang, 1991], which makes it quite attractive for tunnel junc tions and spintronic de vices. A material is classified as half metal when it exhibits a metal-like conduction for one spin orientation of electrons and is insulating for the other orientation. The co nduction in this material is due to the hopping of electrons between the iron atoms occupying the octahedral positions in an inverse spinel structure [X. W. Li, 1998; N. Tsuda, 2000]. As the hopping stops, a metal to insulator tran sition is observed. This is kno wn as the Verwey transition. This transition is manifested as a discontinuity in resistivity (an increase of several orders of magnitude) around Tv ~120 K for bulk Fe3O4 [X. W. Li, 1998; G. Q. Gong, 1998; E. J. Verwey, 1941]. This transition and its character istic are modified by the stoichiometry of the film [L. Wang, 1999; E. J. Verwey, 1941], th e film thickness [S. P. Sena, 1997; X. W. Li, (1998); M. Ziese, (2000)], and crystalline quality. This change in the transition characteristics can be used to determine the qualities of a film. The first efforts of the work presented in this dissertation were intended to study the use of pulsed laser deposition (PLD) and dual laser ablation (DLA) techniques for the growth of magnetic junctions and structures rather than other deposition techniques such as evaporation (CVD) and/or sputtering techni ques. It is known that the plume expelled from the target as a result of the energy depos ited by the laser on the target surface in a
2 PLD system is highly energetic [K. L. Saenge r (1993); G. K. Hubler (1994)]. The energy and ionic content of the plume is increased fu rther when DLA is used as it couples more energy to the plume through the incorporation of a CO2 laser in addition to the standard excimer laser in a PLD system [S. Witana chchi, (1995); P. Mukherjee (1998 & 1999)]. This highly energized plume might change the crystalline quality of the film, the grain sizes and/or reactions of th e arriving plume with the subs trate, thereby effectively changing the interactions of th e different layers and interface s of the magnetic structure. Before starting the growth of the magnetic structures it was of interest to study the characteristics of the individual layers to be deposited. In the case of magnetite, a measurement of resistance versus temperatur e should give an insight on the film quality reflected in the characteristics of the Ve rwey transition. Resistance measurements of magnetite films deposited on Si using PLD show ed a broad Verwey transition, that can be attributed to the thickness and/or crysta lline quality of the film. The fact that a transition was observed indicated that the film is most likely magnetite. As a second step to characterizing the individual layer, DLA films were grown and examined in comparison to PLD samples and samples grown by other techniques as published in literature by other groups. An initial magnetite film grown using DLA did not show the Verwey transition, but instead ex hibited a metal to insu lator transition that is greater than 4 orders of magnitude at ~ 230 K. Efforts to replicate this transition in the laboratory by reproducing the experiment were not successful. X-ray photoelectron spectroscopy (XPS) st udies of the sample exhibiting the transition compared to one that does not was made. These experiments showed that the difference was that the sample with the transi tion at 230 K contained an 8 atomic percent
3 Si doped magnetite film. Consequently, Fe2O3 (Hematite) target was intentionally doped with 5.6 atomic percent of Si to try to reproduce the transition. Studies with the doped target using PLD revealed that only sample s deposited in an oxyge n ambient presented a transition similar to what was observed fo r the DLA magnetite sample. Furthermore, hematite samples deposited in oxygen also reve aled the transition negating the apparent need of Si contamination for the transition to take place. To test the role of Si in the transiti on an extreme measure was taken and FeSi films was deposited using PLD and in vacuum. These samples consistently exhibited the transition at a temperature hi gher than 220 K. Samples of Fe and CoSi deposited on Si also exhibited the transition, but Ti and TiSi samples did not. This prompted the study of the magneto-transport properties of these interesting films. Through the chapters following this intr oduction a systematic study of the nature of the transition and its possible causes are discussed. Several possi bilities for explaining this transition have been proposed and ex amined through this work. Scattering, magnetic scattering, charge freeze out and interface inte raction have been suggested as possible mechanisms and experimentally examined.
4 Chapter 2 Iron Oxides The initial intention of this work was to study the effects of PLD and DLA on the fabrication of magnetic structures. This aris es from the high energies involved in the plume as it is expelled from the target, whic h might trigger the depos ition of a different film quality compared to other deposition methods. This might effectively change the properties of a magnetic structure. The magnetic material we investigated w ith intentions of us ing it in a layered structure fabricated w ith PLD and DLA was Fe3O4. Magnetite (Fe3O4) is a well studied ferromagnet and for that reason it was chosen as our starting mate rial. The crystalline structure of magnetite is that of a cubic inverse spinel one [X. W. Li, 1998; L. Wang 1999]. The spinel structure has a generic formula of AB2O4. One cation occupies octahedral sites, the other cation the tetrahedra l sites and both s ites are surrounded by oxygen. The inverse spinel has the same physi cal structure as spinel except that the octahedral sites house two cations. In the case of magnetite the formula would be represented as Fe3+[Fe2+Fe3+]O4, where the Fe3+ occupy both tetrahedral and half the octahedral sites, and the Fe2+ occupy the other half of the oc tahedral sites. A depiction of the structure of inverse spinel is shown in figure 2.1 along with octahedral and tetrahedral sites.
5 Being classified as a half metallic ferromagnet [Z. Zhang, 1991], magnetite is quite attractive for tunnel junc tions. A material is classifi ed as half metal when the material exhibits metal like conduction for one spin orientation of electrons and is insulating for the other orientation. This can be seen in Fe3O4 from the location of the Fermi energy with respect to the spin pol arized electron bands obtained from linear muffin tin orbital (LMTO) cal culations and from the density of states calculations presented in figure 2.2 and 2.3 respectively [Z. Zhang and S. Satpathy, (1991)]. It can be easily seen that the Fermi energy falls in a band gap for electrons with spin while it falls inside a band for spin electrons. These metallic like electrons belong to the Fe atoms in the octahedral sites. This not only predicts the half metallic nature of Fe3O4 but also predicts or confirms the electrical tran sport mechanism discussed below. From this theoretical prediction one can expect Fe3O4 to be highly spin polarized, although only ~ 40% negative spin polarization has b een measured [S. F. Alvarado, 1975]. Octahedral position Tetrahedral p osition Figure 2.1: Fe3O4 crystalline inverse spinel structure. Large open circles O2-, small hatched circles Fe3+, small open circles Fe2+ and Fe3+. N. Tsuda, K. Nasu, A. Fujimori, K. Siratori: Electronic Conduction in Oxides, Springer, 2000, p. 246.
6 Electrical conduction in this ma terial is generally agreed on, and is attributed to the hopping of electrons between the iron at oms occupying the octahedral positions (occupied by both Fe3+and Fe2+ and illustrated as small open circles in figure 2.1.1) [X. W. Li, 1998; N. Tsuda, 2000]. As hopping is fr ozen due to the ordering of Fe ions occupying the octahedral sites a metal to insulator transition known as the Verwey transition is observed [Z. Zhang, 1991; N. Tsuda, 2000; E. J. Verwey, 1941]. This transition is manifested as a discontinuity in resistivity (an increase of several orders of magnitude) around Tv ~120 K for bulk Fe3O4 [X. W. Li, 1998; G. Q. Gong, 1998; E. J. Verwey, 1941]. Characteristics of the Verwey transiti on change with the properties of the material, such as if the material is bulk or film. For example, the transition point Tv ~ 120 Figure 2.2: Spin polarized electron bands for F3O4 obtained from local spin density LMTO calculations [Z. Zhang and S. Satpathy, (1991)].
7 K is the bulk value, but when deposited as a film, the transition temperature is measured to be at 121 K instead [X. W. Li, (1998)] and it is found that Tv decreases with decreasing film thickness [S. P. Sena, 1997; M. Ziese, (2000 )]. Even in film form the characteristics of magnetite vary, such as the observed broadening of the transition with decreasing film thickness [X. W. Li, (1998); M. Ziese, (2000) ]. This is noticeable from figure 2.4 for films ranging in thickness from 150 Â– 6600 . Cr ystalline quality of the film also plays a role, as evident from the absence of the tran sition in quenched polyc rystalline samples [L. Wang, 1999]. The transition could even decrease in magnitude and eventually disappear with an increase in oxygen content [E. J. Verwey, 1941], which is an indication of the transition and transport dependence on the stoichiometry of the sample. All these Figure2.3: Spin projected one-electron densities of states and numbers of states for magnetite [Z. Zhang and S. Satpathy, (1991)].
8 physical properties and their dependence on film quality were used in this work to verify the growth of magnetite as deposited by PLD and DLA. It is also known that magnetite has a small negative magneto resistance (MR) with MR defined as (RH-R0)/R0 where RH and R0 are the in-field and no-field resistance [G. Q. Gong, 1998; V. V. Gridin, 1996; X. W. Li, 1998]. The MR does not vary much with temperature until Tv where it spikes. As plotted in figure 2.5 the sharp increase reaches about ~20 % for thick high quality films at 4T. Note that MR in figure 2.5 is positive but it is due to the formula the authors used to define MR which is (R0-RH)/RH. The MR dependence on temperature is observed both in films and bulk. Other indications of the Verw ey transition are an anomal y in the heat capacity [J. P. Shepherd, 1991] and a sudden drop in magnetization. Both phenomena are shown in figures 2.6, and 2.7 a and b. From the heat capa city versus temperature measurements one can clearly see that there is stoichiometry de pendence on whether the tr ansition is first or second order, and at what temperature it w ill occur. These samples were bulk samples Figure 2.4: Resistance properties of Fe3O4 and its dependence on thickness [X. W. Li et. al. (1998)]. The insert is a plot of Tv for different film thickness.
9 (5 5 0.7 mm3) and the stoichiometry was varied by varying the Fe content in the samples according to the chemical formula Fe3(1)O4, where 0.00018 0.0121. Magnetite has been previously grown ep itaxially on MgO using the pulsed laser deposition (PLD) technique [G. Q. Gong, 1997]. Fe3O4 was also grown epitaxially on Si and GaAs substrates buffered with MgO, but grows with a random in-plane (111) orientation when Si and GaAs are ba re and oxide free [R. J. Kennedy, 1999]. Figure 2.5: Magnetoresistance for 6600 thick Fe3O4 films deposited on (100) MgO [G. Q. Gong et. al. (1997)]. Figure 2.6: Resistivity (at 0 magnetic field) and magnetization (measured in 300 Oe) as a function of temperature for Fe3O4 films deposited on (100) MgO [G. Q. Gong et. al. (1997)].
10 (a) (b) Figure 2.7: Heat capacity versus temperature plots of homogeneous single crystal Fe3O4, (a) exhibiting first order transition, (b) exhibiting higher orde r transition [J. P. Shepherd et. al. (1991)].
11 Chapter 3 Sample Deposition and Characterization Section 3.1 Deposition The deposition process took place at our la boratory [the Laboratory for Advanced Materials Science and Technology (LAMSAT)], Department of Physics, University of South Florida. The technique used was pulsed laser deposition (PLD). Just like any other technique, PLD has advantages and disadva ntages. Some of the advantages are: stoichiometric film deposition, energetic plume, relative safety, ease of operation and the flexibility of using any kind of ambient gas during deposition, includ ing reactive species if necessary [K. L. Saenger (1993); G. K. Hubler (1994)]. The most prominent disadvantages of this process are the deposition of micron and sub-micron size particulates and the narrow angular distribution of the pl ume. These disadvantages are most noticeable when smooth films or larg e uniform coverage areas are required. A variety of techniques can in principle al leviate the particulate problem, such as appropriate adjustment of laser fluence, m echanical shuttering, colliding beam deposition and dual laser ablation. Particularly for re search purposes where the radial sample dimensions are small PLD is an excellent deposition technique and the disadvantages become irrelevant.
12 The PLD system layout is very simple a nd is depicted in fi gure 3.1.1. It consists of an excimer laser acting as the power sour ce, a deposition chamber that contains both the target and the substrate, and optics (mirrors and a lens) that are used to direct and focus the laser beam on the target. The ope ration of the system involves striking the target repetitively by the laser, but intera ctions between parame ters controlling the E X C I M E R L A S E R MIRROR LENS SUBSTRATE TARGET VACUUM CHAMBE R PLUME Figure 3.1.1: Depiction of the PLD system used to deposit the silicide films. The system consists of a laser, acting as a power source, and lenses and mirrors to direct and focus the laser beam on the target which consists of the material to be deposited. In fron t of the target is the substrate where the material is collected and deposited. The substrate could be held at a temperature different than that of room temperature, and the atmosphere of the chamber c ould be vacuum or filled with a reactive or nonreactive gas.
13 process and the dynamics of the ejected mate rial are very complex. For an excellent review of PLD the reader is referred to th e two-part article by K. L. Saenger [K. L. Saenger (1993)] and the book by G. K. Hubler [G. K. Hubler (1994)] where both authors explain the technique in detail They also explain the rela tion of the different deposition parameters with the film quality. The deposition of the energy from the laser to the target occurs in such a short time that thermal relaxation or dissipati on does not spread from the area of laser impingement. This causes an explosive evaporati on from the targetÂ’s surface, resulting in a laser-ablated plume that forms and travels perpendicular to the target surface in an expansive mode. The plume consists of a mi xture of electrons, ions, atoms, molecules and particulate of sizes in the micron and s ub-micron range. The explosive material is collected by a substrate mounted at a fixed distance in front of the target. This process of explosive evaporation, or ablati on, could take place in vacuum or in a certain ambient, which is either reactive or not. Parameters that control the filmÂ’s quality are the laser energy, laser spot size, target to substrat e distance, substrate properties, substrate temperature and deposition atmosphere. Dual laser ablation (DLA) is an offspring of the traditional pulsed laser deposition technique (PLD) used for decades. DLA was developed in the University of South Florida at the Laboratory for Advanced Material Science and Technology [S. Witanachchi, (1995); P. Mukherjee (1998 & 1999)]. DLA incorporates a temporally synchronized CO2 laser in the PLD setup, where the beam of the addi tional laser would spatially overlap that of the excimer lase r and precede it in time by a predetermined amount dependent on the CO2 laser fluence, and the thermal and optical properties of the
14 target. Incorporation of the CO2 laser with proper time delay (~ 50 ns) virtually eliminates particulates and greatly improve s the coverage area bot h of which are common drawbacks of PLD. DLA causes the enhancem ent of plume excita tion by increasing the number of ions present in the plume. This ex citation also induces better reactivity with the ambient gas present in the chamber and in creases the nucleation sites on the substrate. It also enhances the radial expansion of the plume produci ng large uniform area coverage of the substrate, while preser ving the stoichiometry for multiple component targets. Since adjusting the overlap of the two laser beams in time allows the user to eliminate the particulates in a film or cont rol the particulate size [S. W itanachchi, (1995); P. Mukherjee (1998 & 1999)], DLA is useful for the pr oduction of heterostructures and/or nanoparticles. Even though a number of substrates we re used during the experiment, we concentrated our efforts on silicon (Si). The Si substrates we used were commercially acquired p-type (boron-doped), with a (100) or ientation. The resistivity of the substrates varied from very low values to > 3000 ohm-cm This variation in substrate resistivity was purposely used to study its effect on the sampleÂ’s properties. Si crystallizes in the diamond structure with a lattice cons tant of 5.43 . The diamond structure could be viewed as two facecentered cubic Bravais lattices one inside the other but shifted along the diagonal by one f ourth of its length as illustrated in figure 3.1.2. It is known that Si is an indirect se miconductor, with a band gap value of about 1.12 eV when measured at 300K [S. M. Sze, (1981)]. Doping Si causes changes in the conduc tivity depending on the amount and type of doping. Silicon (or any semic onductor) is said to be p-type when the dopant accepts an
15 electron from its neighboring silicon atoms, like in the case of boron or gallium (IIIA elements), creating a hole or an absence of an electron. On the other hand if we dope silicon with an element such as arsenic ( VA element), it donates an electron and changes the silicon to an n-type semiconductor. Figure 3.1.3 shows the conductivity of borondoped Si (p-type) with respect to temper ature. The boron concentration is ~1015 atoms/cm3. Figure 3.1.3 was produced by extracti ng data from a digitized figure containing the results of conductivity depende nce on temperature for Si of different dopant concentration [F. J. Morin (1954)]. At high temperatures the doped Si behaves like a metal, where the resi stance (conductivity) drops (inc reases) with a decrease (increase) in temperature. At ~120 K the resistance increases exponentially exhibiting behavior expected from an intrinsic semiconductor. Figure 3.1.2: Si crystallographic structure. It is the diam ond structure, which is two fcc structures shifted along the diagonal with respect to each other. [C. Kittel, (1996)]
16 Usually chemical cleaning is enough to remove any surface contaminants, but mechanical or heat treatment are also used to clean and reconstruct the surface of the substrate. The cleaning eliminates any possibl e problems due to contamination or lattice imperfections at the interface. In the case of the films deposited for the experiments presented in this study, only chemical cleani ng with acetone and me thanol in ultrasound baths were performed, except for those sample s implicitly specified during specific data discussions in later chapters. The substrate was first put in an acetone ultrasound bath for 10 minutes. Then it was moved to a metha nol ultrasound bath for another 10 minutes. 050100150200250300350400 10-1100101102103104 Resistivity ( cm)Temperature (K)Figure 3.1.3: R esistivity curve of B-doped Si with a density of ~ 1015 atoms/cm3, which is the same density as the substrates used in this report. This fi gure was extracted from F. J. Morin et. al. Phys. Rev. 96, 28 (1954) by digitizing the figure and re-plotting the data here.
17 After the ultrasound baths were done, the subs trate was rinsed with methanol and then blown dry with pure nitrogen. After the substrate was cleaned it was mounted on a heating block facing the target which would be used in the ablation pr ocess. Two methods were used to secure the substrate to the heating block. Each method was implemented depending on the temperature at which the sample would be depos ited. If the sample was to be deposited at room temperature, the substrate would be m echanically pressed against the surface of the heating block by a stainless steel tab. The tab was attached to the block by a screw which when tightened would press against the substrat e and when loosened would release it. If a high temperature deposition was to take plac e the substrate was secured to the heating block by silver paint. This method ensured e qual heating of the substrate and eliminated any temperature gradients that might be present if the substrate was secured mechanically. The target was 1.25Â” in diameter and 0.25Â” in height. It was fixed to a 0.5Â” shaft connected to the target motor by means of a rotational feed through. The motor was used to rotate the target during deposition to mi nimize the damage to the target from the ablation process. The distance between the s ubstrate and the target during deposition was set to 4 cm. After mounting the target, the laser spot area was measured after being directed and focused on the targetÂ’s surface. This measurement was used to calculate the fluence, which is equal to the energy of th e laser pulse divided by the area of the spot. The fluence of the laser dur ing the deposition of the samples for this study was 2.5 5 J/cm2. Prior to mounting the target, it was sande d and polished to remove any grooves or mechanical damage that would be produced by previous ablations. The targets used in
18 this study were usually commercially acquired, unless specified for a particular sample or experiment. More elaboration of the targetÂ’s content will be made during discussion of the specific samples used in the experiments. After securing the target and substrate in place, the chamber was sealed and pumped down to a base pressu re in the range of high 10-7 torr. The samples were deposited in vacuum at base pressure, unl ess specified during the discussion of a particular sample. The substrate was brought up to deposition temperature by heating the block it is mounted on. A 600-Watt light bulb is fixed inside the bloc k and serves as the heating source when a voltage is applied to it. The laser used was a Lambda Physik KrF Compex 102 excimer laser. The wavelength is in the UV range with a value of 248 nm and a pulse width of about 20 ns. The frequency at which the target was ablate d was 4 Hz (four shots per second). At the end of the deposition cycle the sample was cooled down to room temperature and taken out of the chamber for characterization.
19 Section 3.2 Sample Characterization The samples were characterized using several techniques to gain information about the atomic composition, crystalline st ructure, sample layering and transport properties of the deposited film s and their interaction with the underlying substrate. X-ray photoelectron spectroscopy (XPS) measurements were taken using a Physical Electronics 5400 ES CA at AMPAC, University of Central Florida. This characterization technique consis ts of the study of the energies of electrons that have been ejected from the core or valence shel l of an atom. The ejection of the electron occurs when it is hit by an x-ray photon orig inating from a monochr omatic source. The energy difference between the kinetic energy of the electron and the energy of the x-ray photon give the investigator the data needed to identify the atom from which the electron was ejected and what the atom is bonded to. A profile of the compos ition of the film can be studied when this technique is coupled w ith a sputtering process. The profiling can be done by taking an XPS data set, sputtering for some determined time and taking another XPS data set. This procedure would be re peated until the desired sampling depth has been reached generating a film composition versus depth profile. The crystalline structures of the sample s were studied by x-ray diffraction (XRD) theta 2-theta powder technique. A Philips XÂ’Pert system was used for the XRD studies at the Engineering Metrology Lab, University of South Florida (USF). This is done by
20 shining a collimated monochromatic x-ray be am on the sample at different angles and plotting the diffracted patte rn following BraggÂ’s law, sin d 2 n where n is the order of reflection, is the wave length of the x-ray source, d is the distance between the crystallographic planes and is the angle of incidence. XRD theta 2-theta measurements were done on the samp les with 2-theta sweep ing values from 20o to 80o at a step size of 0.005o and 1 second of time per step. The sample dimensions were usually small (about 2 2 cm) and for this reason so me mounting procedures were necessary to be able to take the diffracti on measurements. A glass piece of about 1.5Â” by 1.5Â” was used as the substrate where the sa mple was fixed with double sided tape. The sample had to be physically fixed to a bigge r substrate for the mach ine mounting stage to be able to grab the sample. The glass substr ate also was useful to provide an amorphous background and get rid of unwanted peaks, wh ich would have been observed from the mounting stage. Crystalline structure and thickness measurements were the first two characterizations that were performe d because of the non-destructive and noncontamination nature of the instruments us ed. The thickness of the samples was also studied at the Engineering Metr ology Lab by using a profilomete r. This is a very simple method and it consists of a need le-like probe that m oves across the surface of the sample giving a height profile of the structures present on the surface. The thickness measurements were done by profiling the edge of a shadow formed by a mask covering the substrate and shielding it from deposition. When depositing the sample, care was taken that part of the center of the plume impinged on the mask. This type of alignment
21 assured that the measured thickness would be across a limited portion of the radial profile of the film. The substrate in tercepted a cone angle of 14o of the central profile of the plume. The values later presented for thickne ss are an average calculated from different readings along the shadow of th e film deposited on the substrate. The layering structure of the film was imaged using a Hitachi HD-2000 CFSTEM transmission electron microscope (TEM ); which also allowed the study of the ratios of the elements deposited in the f ilm. A TEM basically operates as an optical microscope does, except that instead of visible light, electrons are used as the illumination source, and the lenses that are used are not made out of glass but of electric and magnetic fields. Transport properties such as DC re sistance and IV characteristics were investigated using a standard four-point probe configura tion at our laboratory. After deposition was complete and the sample remo ved from the deposition chamber, indium contacts were mechanically placed on the sample by pressing a piece of indium wire against the sample. The sample was then pla ced on a Cu finger, which was attached to a compressed He cooling system. Silicon grease was used to fix the sample in place. Cu wires were used as leads for the current de livery and the voltage probing. The leads were placed on the indium pads making sure that contact had been made by observing the reading of the voltage (for the voltmeter leads), and that the voltage applied to deliver the current was not exceeding the constant curr ent power supply maximum voltage settings. Silver paint was placed on the le ads and the indium contact pads to insure that the leads would remain in electrical contact throughout the experiment A schematic representation of the system setup for voltage measurements is presented in figure 3.2.1.
22 The devices controlling the parameters of the measurement (applied current, measured voltage, and temperature control a nd reading) were controlled by a PC through Constant Current Pulse Power Supply Voltage Meter Temperature Controller Cu Finger 50100150200250300 102103104 Resistance ()Temperatue (K)Sample Contact pads and leads Thermocouple Figure 3.2.1: Schematic of the voltage measurement system. A PC through a GPIB connection to the different devices controlled the measurement process, namely, applying the current pulse, measuring the voltage, temperature control, and data storage.
23 a GPIB board. A LabVIEW program was written to control the temperature at which a measurement is to be taken, the temperature steps between measurements, the number of current pulses per measurement, and values of the current to be applied before a measurement is taken. Resistance was calculated using OhmÂ’s law I V R where V is the voltage measured along the sample and I is the applied current. This calculation was made when a current pulse was applied to the sample using a Keithley 224 constant current source and the voltage measurement was done by a Keithley 182 nano-voltmeter. Measurements were taken ten times for every current value, five times in one current direction and five others in the opposite dire ction. An average and standard deviation of the voltage measured for every current value was taken and used to calculate an average resistance. The current pulses varied from 1 to 100 A; which enabled the measurement of IV characteristics at the same time as sa mple resistance. These resistance calculations were plotted as a function of temperature. An APD closed-cycle cooling system and Lake Shore temperature controller adjusted the temp erature. In combination they maintained the temperature within a degree while measurements were being taken. Magnetoresistance (MR) measurements were performed using a Physical Properties Measurement System (PPMS) at USF; which was also used as an independent system to verify the existence of a temperat ure dependent resistance transition. MR is the measurement of change of the resistance when a magnetic field is applied. This effect is calculated using the following equation: 0 0 HR R R MR
24 Here RH is the resistance in a magnetic field and R0 is the resistance out of the magnetic field. It is said that the MR is positive when resistance in creases in a magnetic field and negative when resistance decreases. Magne toresistance dependence on temperature and the dependence on magnetic fi eld at a fixed temperature were obtained. The samples were in an inline four-point probe configuration. They we re fixed to the measurement puck by vacuum grease. The magnetic field was applied parallel and pe rpendicular to the sample surface and current path. This was done to elucidate the cause of the transition. Since the PPMS provided the magnetic field in one orientation (per pendicular to the sample surface) an aluminum stub was used to prop the sample and orient it parallel to the magnetic field. We were able to test up to three different samples at the same time. The configurations of the magnetic field with respect to the sample were parallel to sample surface and current path, parallel to th e sample surface and perpendicular to the current path, and perpendicular to both sample surface and current path. Measurements were taken at field values vary ing from 0 to 7 T. The same method that was used to make the electrical contacts to generate the resi stance curves was also employed in the MR measurements.
25 Chapter 4 Experiments and Results of Iron Oxide Samples were grown at our laboratory on Si (100) substrates using the PLD technique. According to the li terature reviewed in the prev ious section, polycrystalline samples should be expected, which would be reflected by a broad Verwey transition in the resistance versus temperature measurements. Measurements of the resistance dependence on temperature were used as an indication for the quality of the films by just studying the characteristics of the Verwey transition. The first samples made with PLD seemed to agree with the literature (see figure 4.1) for thin polycrystalline films. This was based on the fact that resistance measurements with respect to temperature of PLD grown magnetite films exhibited a broad transition at ~ 120 K. Following these results DLA samples were made in order to compare them to PLD samples. The resistance versus temper ature measurements made on a DLA sample; which was deposited under the same conditions as the PLD samples, is shown in figure 4.2. A difference between the two samples (PLD a nd DLA) is visible. It is obvious that a transition in resistance from a metallic-like behavior to insulator (or semiconductor) is present, which was expected. However, the transition temperature was expected to be about 100 degrees lower than the obse rved value which is ~ 220 K.
26 To understand the nature of the transition and to confirm the type of iron oxide present in the sample, X-ra y photoelectron spectroscopy (X PS) studies were performed on the sample that did show the transition and on a sample that did not show it. The results indicated that the sample with a tr ansition at ~220 K contai ned an eight atomic percent of Si in the film ben eath the surface. In order to inve stigate the role of Si in the film, an Fe2O3 target doped with 5.6 atomic percen t of Si was made and used to grow films on Si using the PLD method. 50100150200250300 102103104105106 Resistance ()Temperature (K)Figure 4.1: Resistance measurements of a PLD deposited magnetite film. It was deposited at ~400o C for 1.5 hrs. A broad transition is an indication of either a polycrystalline or thin film.
27 A film deposited on Si with the silicondoped target in vacuum (M031) did not show the metal to insulator transition (see figure 4.3). However, when the film was annealed in an oven at the deposition temperature (~ 400o C) in air and at atmosphere pressure for half an hour, a sharp transition of approximately 5 orders of magnitude from a conductive, to a highly resi stive state was observed. Onset of the transition was around 230 K. Based on these results it was clear that in the presence of air at high temperatures, the iron oxide silicon doped film undergoes a st ructural and/or compositional change that gives rise to the conduction to insulation transition at a temp erature much higher than the Verwey transition temperature. This hypothesi s was confirmed by the observation of the 50100150200250300 101102103104105106 Resistance ()Temperature (K)Figure 4.2: A standard 4-point probe measurement showing a metallicinsulator transition. This magnetite film was deposited using DLA at ~ 400o C.
28 transition in a film (M032) that was deposit ed from the Si doped target at a 1 mtorr oxygen ambient pressure (figure 4.3). According to results in figure 2.2.3 a study of oxygen pressure in samples doped with silicon was appropriate, since the transition is manifested when both elements are present during the deposition. Se veral films were deposited at different oxygen ambient pressures. The environment pressure was va ried from 0 up to 10 mtorr maintaining all other parameters constant. All samples showed the transition with some slight differences (see figure 4.4). It would be expected that when iron oxide films are deposited at such 50100150200250300 101102103104105106 M031 M031anneal M032Resistance ()Temperature (K)Figure 4.3: Resistance dependence on temperature of Si-doped iron oxide films. The metallic to insulating transition is not quite clear when deposited in vacuum (M031), but is apparent when annealed in air (M031 annealed) or when deposited in an O2 ambient (M032).
29 high oxygen pressures, Fe2O3 is more likely to form instead of Fe3O4, especially when the target is Fe2O3 to start with. Fe2O3, also known as hematite, is considered as a semiconductor [S. K. Pawar, 1983] or a material with high resistivity [S. Bae, 2001; N. Hasegawa, 1996; Y. Gao, 1997]. Hematite exhibits a first order transi tion at about 263 K, which is known as the Morin transition [M. Vasquez-Mansilla, 2001]. This transition changes the material from a weak ferromagnetic state to an anti -ferromagnetic state by flipping spins 90o from a basal plane orientation to an orientation al ong the c axis [M. Vas quez-Mansilla, 2001; F. 050100150200250300 10-1100101102103104105106107108 Resistance ()Temperature (K) 0 mtorr 0.5 mtorr 5 mtorr 10 mtorrFigure 4.4: Resistance dependence on temperature for Si-doped iron oxide films deposited on Si in different oxygen ambient pressures
30 Bodker, 2000; G. Kletetschka, 2002; C. G. S hull, 1951] (see figure 4.5). Because of the nature of the Morin transition, the resistance tr ansition was attributed to a spin dependent scattering mechanism introduced by the spin f lip. However, the resistance measurements made in and out of a magnetic field of up to 4 T did not seem to change the point of transition or the sharpness (i .e. temperature range covered by the transition), creating doubts about the nature of the transition and its relation to magnetic scattering. To try to identify the part played by Si doping in the transition, we made samples of un-doped iron oxide in a 2 mtorr oxygen atmos phere. When tested for their resistance characteristics, the results were as shown in figure 4.6. A transition in resistance at ~230 K was obtained, where the magnitude of the tr ansition is about 5 orders over a short range of temperature. This experi ment suggested that Si in the Fe2O3 film might not play any role in producing the resistive trans ition. Resistance data collected from these samples have shown non-smooth data, espe cially around the transition point. Figure 4.5: Illustration of spin orientation be fore and after Morin transition for Fe2O3. D. J. Craik Magnetic Oxides Part 2, John Wiley & Sons, 1975, p. 665.
31 With this data in mind, it was thought that Fe2O3 is behaving as a metal down to 230K, suggesting that hematite could be c onducting anisotropically along grains that might be a product of columnar growth (per pendicular to the substrate), and the Morin transition is responsible for the sudden incr ease in resistance by magnetic scattering. The current path would be along the grains down to the substrate, then across the substrate towards the second current lead and up the grai ns again. To further study this possibility we also decided to check on the role of th e interface between the substrate and the film. 50100150200250300 100101102103104105106107 Resistance ()Temperature (K)Figure 4.6: Results of resistance measurement versus temperature for un-doped iron oxide (Fe2O3) film deposited in a 2 mtorr O2 atmosphere.
32 The deposition of Fe2O3 and Fe3O4 on substrates other than Si (glass, MgO, sapphire) produced only films as predicte d by the literature already refe renced. This suggested that indeed a certain interface or conducting subs trate is needed and the Morin transition might not be the only factor or a factor at all for the transition to take place. As observed by previous results an O2 atmosphere is needed for a manifestation of the transition. It was also found that the presence of Si was not necessary (to get a transition), or alternatively the right proporti on of Si to Fe was not used, since we only used 5.6 atomic percent doping instead of the 8 % measured in XPS, and the measurements never reached the film-substrat e interface.It was thought that the presence of the transition in iron oxide films, observed earlier, was because of the possibility that the film itself was reforming or restructuring, at least at the interf ace. K. Ruhrnschopf et. al., found that oxygen could diffuse down ( by oxidizing Fe) to the Si substrate by exposing the iron film to an oxygen atmosphere At the right temper atures O also could change partners from iron to silicon and form SiO2, which is distributed all over the FeSi2 interface produced as result of the reacti on. This transformation gives as a product, clusters of SiO2. If this actually is the case for the samples produced by the experiments discussed here, it might explain the noise, or non-smoothness of the data in the iron oxide films that present the transition, which coul d be attributed to scattering across grain boundaries with different stoichiometry and conduction mechanisms. The deposition of Fe films with high Si doping was done to verify if an increase of Si content in the films would change the transition characteristics. Re sults of such experiment are discussed in later chapters. A review of th e knowledge of the characteristics and growth of FeSi is presented in the following chapter.
33 Chapter 5 Iron Silicide FeSi is described as a narrow gap semi conductor that has the B20 simple cubic structure [T. Jarlborg (1995), A. Cahinani et al. (1994), G. Y. Guo, (2001)], and a lattice constant of 4.48 [S. Paschen et. al. ( 1997)]. The classification comes from the Strukturbericht designation, where B stands for a diatom ic structure of equal ratio and the 20 is the historical order of when the latti ce was studied. The atomic arrangement is very simple as can be observed in figure 5.1. In the  direction a la yer of iron atoms is followed by a layer of silicon atoms. This ar rangement is quite interesting for magnetic application in thin film form, since Fe -terminated (111) surfaces could show ferromagnetism as calculated by G. Y. Guo, while Si-terminated (111) surfaces only would show as weak or nonmagnetic. The magnetic moment in Fe-terminated surfaces decreases rapidly as one moves inward. The susceptibility in FeSi, figure 5.2, ha s a strong dependence on temperature; it exhibits a maximum at 500 K [V. Jacarino et. al. (1967)]. Around r oom temperature and down to 100 K the susceptibility can be fitted to an activated behavior [Z. Schlesinger et. al. (1993)]. Below 100 K the susceptibility ri ses again. The whole temperature range of 330 to ~ 4 10-4 K was modeled to fit an equation containing 1 activation term and 2 Curie-Weiss terms as seen in equa tion 5.1 [S. Paschen et. al. (1997)],
34 2 2 1 1 0 T C T C T T exp T c (5.1) where c = 1.8 K, T0 = 670 K, C1 = 8.0 mK, 1 = -870 mK, C2 = 0.26 mK, and 2 = 34 mK. Figure 5.1: FeSi crystallographic structure. It is cla ssified as B20 simple cubic structure. The classification comes from the Strukturbericht Designation, where B stands for a diatomic structure of equal ratio and the 20 is the historical order of when the lattice was studied [G. Y. Guo, (2001)]. Figure 5.2: FeSi susceptibility dependence on te mperature [Jaccarino et. al. (1967)].
35 There are reports of a gap being formed at low temperatures and that it disappears above T~250 K [A. Cahinani et. al. (1994), Z. Schlesinger et. al. (1993)]. This material presents a gradual change in its resistivity below 300 K [A. Cahina ni et. al. (1994), L. Digiorgi (1995)], changing its transport properties from a di rty metal to a semiconductor. The changes in resistivity are gradual and c over a broad range of temperature (see figure 5.3). Early studies explained the magnetic and electric properties of FeSi by interpreting results from el ectrical conductivity, thermal power and Hall coefficient experiments [V. I. Kaidanov et. al. (1968)]. V. I. Kaisdanov et. al. reached the conclusion that mono-silicide materials properties are dictated by th eir d electrons, which do not participate in bonding. In the case of FeSi Kaisdanov concluded that the magnetic and electric properties are governed by the 3d metal electrons which form two narrow subbands, the lower filled with electrons and th e upper one totally empty at absolute zero. More recently iron silicide has been describe d as a Kondo insulator [C Fu et. al. (1994), G. Aeppli et. al. (1999), Z. Schlesinger et. al. (1993), D. Mandrus et. al. (1995)]. Z. Figure 5.3: FeSi resistivity dependence on temperature. The data in this plot is for single crystal [M. Mihalik et al., (1996)].
36 Schlesinger got to this c onclusion after analyz ing data obtained from near normal incident reflectivity and conductiv ity experiments. In a later publication he re-argues that it is not quite correct to classify FeSi as a Kondo insulator but rather as a strongly correlated insulator [Z. Schl esinger et. al. (1997)]. As it stands today, the scientific comm unity has not reached a complete consent on the transport mechanism. It is mostly agreed that the Fermi energy of FeSi falls in the middle of a gap where the value for the dir ect gap is 75 100 meV and 55 Â– 70 meV for the indirect gap [S. Paschen et al. (1997)]. The density of states (DOS) around the gap is high and narrow, with a width of about 50 meV. The values just stated above for the band gaps and the widths of the bands surrounding them were extracted from susceptibility measurements. These values when used to model or verify the resistivity behavior of the same samples do not explain the elect rical transport prop erty as well as they did for susceptibility [D. Mandrus et. al. (1995)], suggesti ng that the gap size relevant for one property is not relevant for the other [S. Pa schen et. al. (1997)]. Chainani implies similar conclusions when he justifies the different values for the band gap he extracted from photoemission spectroscopy e xperiments in comparison with values obtained from electrical resi stivity, magnetic susceptibil ity, and optical conductivity experiments [A. Chainani. et. al. (1994)]. It also has been presented that there ar e different regions, with different physics governing FeSi properties in each region [T Jarlborg (1995), V. V. Glushkov et. al. (2000)]. Glushkov identifies three regions with different mechanisms: region I (T = 75 K 300 K) intrinsic conduction with an indirect band gap of 60 meV, region II (T = 7 K-75 K) extrinsic conduction with a direct band gap of 6 meV and region III (T < 7 K)
37 coherent state formation. K. G. Lisunov et. al. state that conducti on in the temperature range of 100 K Â– 170 K satisfies the Arrheniu s laws while it satisfies variable range hopping as the temperature is lowere d [K. G. Lisunov et. al. (1996)]. The magnetoresistance (MR) measured fo r different samples showed different behavior [H. Ohta et. al. ( 1997), K. G. Lisunov et. al. (1996) ]. These differences in MR could be attributed to differe nces in sample qualities. In general MR decreased with increasing temperature. Lis unov also pointed out that ma gnetic field and temperature dependences of the observed MR suggest both positive and negative magnetoresistance contributions that have different field and temperature dependence. The positive contribution agrees with the model that takes into account both the in tra-site and the spin flip scattering of the hopping scattering, and the ne gative magnetoresistance is explained by Zeeman splitting of the mobility thresholds [K. G. Lisunov et. al. (1996)]. Several groups have studied the inter action of Fe with Si under different conditions and the formation of iron silicide s. Rhrnschopf et. al studied the growth mechanism of iron silicide at 300K and its reactivity at higher temperatures. These studies were carried out by the deposition of iron using a pure iron coil on a clean Si substrate followed by x-ray photoelectr on spectroscopy (XPS) measurements. As deposited, the films did not show iron silicide formation or diffusion of either element. Layers of up to a thickness of 2 monolayers do show some inte raction with the substrate. The iron silicide is formed af ter annealing the sample, and th e conversion of the entire Fe film to iron silicide is film thickness and te mperature dependent [K. Rhrnschopf et. al. (1996)]. These results are different from th e ones obtained by A. V. Zenkevich et. al. where they deposited Fe films on clean Si us ing pulsed laser deposit ion (PLD) instead of
38 evaporation. They found that iron mono silicide is formed an d this is attributed to an explosive crystallization. The interaction of Fe with the Si substrate can also be altered or modified not only by deposition technique, but also by deposition conditions such as substrate surface [C. Chemelli et. al. (1993)] and atmosphere [H. C. Swart (1994)]. C. Chemelli et. al. demonstrated that a thin native oxide layer is enough to stop the formation of an iron silicide film at the Si interface when the film is deposited at room temperature. FeSi begins to form when the sample is annealed at temperatures higher than 450o C as the Fe diffuses throug h the oxide layer forming SiOx/FeSi/Si or SiOx/FeSi2/Si structures. The presence of an oxygen atmosphere during deposition enriches the film with oxygen ma king Fe the diffusion species instead of Si. Annealing of the sample induces FeSi formation with SiOx collecting at the Fe/silicide interface. The presence of oxygen in the film with a concentr ation higher than 2 atom ic percent prevents any silicide formation due to FeO formation [H. C. Swart et. al. (1995)]. The properties of FeSi discussed above are those of bulk samples that exhibit good crystalline quality. To our knowledge, no transport data has been reported on disordered (or even good crystalline quality) FeSi films on Si for temperatures below 290 K. In the range of 290 KÂ– 400 K N. G. Galk in et. al. found that the FeSi (111) films grown on Si (111) display meta l conductivity with holes as majority carriers [N. G. Galkin et. al. (2001)].
39 Chapter 6 Results Section 6.1 Resistance Dependence on Temperature Samples of FeSi were deposited to further investigate the role of Si in the samples compared to oxygen. These samples were e xpected to contain minimum amounts of oxygen since the samples were deposited in a base pressure of approximately 1 10-6 torr. The temperature at which the samples were deposited was about 400o C. After the deposition was completed and samples were cool ed down, they were tested for resistance dependence on temperature. Figure 6.1.1 is a typical result of such measurements. The figure clearly shows a sharp transiti on at about 253 K, and the trend of the resistivity seems to go from metallic-like beha vior above the transition temperature to a semiconductor-like behavior after the transition with high resistance values. The transition height from the lowest resistance value to the highest measured at 20 K was greater than 3 orders of magnitude as measur ed by a 1 A current pu lse. The sharpness of the transition, changing from one behavior to another, only covere d a range of about 50 degrees of temperature. When this temperat ure range is compared to the resistivity behavior of bulk FeSi presen ted in figure 5.3, one can see that even though FeSi goes through a change in resistivity, it is very di fferent from the transition that was witnessed in the present samples. The total magnitude or height of the transition in bulk FeSi was
40 only about 3 orders of magnitude, and the shar pest part of the tran sition covered a range of about 180 degrees of temperature and is only noticeable when the temperature is plotted logarithmically. This experiment de monstrated that a high content of oxygen is not needed and that an FeSi film is enough to show the transition. The transition is very similar to the one observed in the iron oxide samples, but here we see a much cleaner data line or curve. When measurements of voltage taken for FeSi samples in order to calculate the resistance are analyzed they gave standard deviation values of 1 to 3 orders 050100150200250300 102103104105 100 A 1 A Resistance ()Temperature (K)Figure 6.1.1: Resistance dependence on temperature of an FeSi sample deposited at 400o C on (100) Si. The current applied to the sample was 100 A and 1 A. The insert is the resistance versus temperature for a Si doped Fe2O3 sample deposited in an O2 ambient pressure of 1 mtorr. The measurement was also made using 100 A. 050100150200250300 10-1100101102103104 Resistance ()Temperature (K)
41 of magnitude less than the measured average valu e. This is reflected in the error bars of figure 6.1.1. Also in figure 6.1.1 one is able comp are results obtained from FeSi films to results from Fe2O3 films. Note the smaller error bars for FeSi samples below the transition temperature. This indicates a bett er signal to noise ratio, which could be an indication of better conductivity. We have seen the transition in films w ith different concentrations of oxygen and silicon. Even though there are different transi tions associated with the different oxides and FeSi (see chapters 2, 4 and 5), none of them are the same as the one observed in the samples discussed here. It was considered worthw hile to verify the idea that the transition is only related to a film characteristic rather than to a bulk property. This may be caused by the thickness of the thin films making the samples essentially two dimensional, or it may be because of the different interfaces fo rmed when a film is deposited. To control the interface between the film and the substr ate it was chosen to deposit thin films over different substrates. In addition a buffer la yer was also implemented as a means of changing the properties of the in terface. The substrates used were Si, sapphire, glass, and glass buffered with Ti. The findings of these experiments are shown in figure 6.1.2. It is easy to observe from the data displayed in figure 6.1.2 that a transition takes place whenever FeSi is deposited on a Si substrate. Even when Fe3O4 is deposited on top of FeSi a transition is visible when deposited on Si. The fact that no transition is observed for Fe3O4/FeSi/Sapphire means that the transition is due to the FeSi/S i interface and the Fe3O4 layer does not have much influence on the transition. The FeSi/Ti/glass sample demonstrates th at this behavior does not occur when FeSi is encountered with any conductive inte rface. This sample shows a resistance with
42 exponential proportionality with the inverse of the temperature, as expected from a semiconductor. Presenting the data as resis tivity would have made a better comparison between different samples, but that is not po ssible to do with accuracy. This is because it is clear that the interface is important, a nd calculation of resistivity requires the exact dimensions of the interface part of the sample expressing the transition effect. 050100150200250300 102103104105106 Resistance ()Temperature (K) Fe3O4/FeSi/Sapphire Fe3O4/FeSi/Si FeSi/Si FeSi/Glass FeSi/Sapphire FeSi/Ti/GlassFigure 6.1.2: Resistance dependence of FeSi on temperat ure for deposition on different substrates (glass, sapphire and Si), buffer layers (Ti) and as a buffer layer for other materials to be deposited on (Fe3O4).
43 Since this substrate dependence data sugge sts that the interface between FeSi and the Si substrate is an important factor, expe riments controlling the interface between the substrate and the film were conducted. The di ffusion of the film into the substrate is temperature dependent. Therefore, by depositin g at different temper atures it should be possible to change the interface, making it leas t defined or less sharp at high deposition temperatures and sharply defi ned at lower temperature. Figure 6.1.3 shows the resistance dependence on temperature for samples deposited at different deposition temperatures. This data is normalized with respect to the maximum resistance measured. The normalization makes it easier for comparison, Figure 6.1.3: Dependence of the resistive transition on deposition temperature. As the deposition temperature is increased, the magnitude an d sharpness of the transition decreases. 050100150200250300350 10-410-310-210-1100 R. T. 420o C 900o C R/RmaxTemperature (K)
44 especially since the data is not being presented as re sistivity. It is clea r from the data in figure 6.1.3 that as the deposition temperature is increased from room temperature (R.T.) to 900o C the magnitude of the transition drops fro m 3 orders of magnit ude to a factor of 2. And the temperature range over which the transition takes place changes from 40 K to 60 K. The deposition temperature increase lo wered the magnitude of the effect and broadened the transition range This change is attributed to the broadening of the interface as the Fe diffuses through the substrate. The data displayed in the different figures presented so far indicates that the thin film itself is not the main reason for the transi tion. The substrate and the interface that the film forms with it appears to play as big a ro le as the film itself. We characterized the substrateÂ’s (p-type Si (100), 1-10 -cm, 1014/cm3) resistance dependence on temperature to gain more insight on the process of the transition. The data for the substrate along with the data for a sample deposited on Si and sapphire as well are presented in figure 6.1.4. The resistance behavior of th e Si substrate at high temperature seems to agree with literature cited in Chapter 3. Due to the doping of the substrate, the resistance decreases with temperature at high temperatures, exhibi ting a metal-like behavior. This is the exact behavior of the combination film depos ited on Si at high temperatures. As the temperature drops, the FeSi/Si samples seem to emulate the behavior of a sample deposited on an insulating substrate (e.g. Fe Si/Sapphire), even though the substrate still exhibits lower resistance than the film itsel f. The evidence presented by the data up to this point suggests that the cu rrent actually flows through the substrate down to a certain temperature where some kind of charge fr eeze or charge blockade takes place at the
45 interface impeding the flow of th e current. If the current path to the substrate is blocked, then it will have to flow through the film. Transmission Electron Microscope (TEM) im ages revealed multiple layers in the samples. It was expected to find a layer of the material being deposited, a native oxide layer and finally the substrat e, but the TEM images reveal ed the presence of a fourth layer. TEM images of the same sample are shown in figure 6.1.5 (b) and figure 6.1.6 (figure 6.1.6 is displayed at the end of th e section). Atomic profiling of the sample, 50100150200250300 101102103104105 Resistance ()Temperature (K) FeSi/Si FeSi/Sapphire SiFigure 6.1.4: Resistance dependence on temperature of th e FeSi film (FeSi/Sapphire), FeSi with interface interactions (FeSi/Si) and the substrate by itself.
46 shown in figure 6.1.5 (a), indicate s that the bulk of the film contains roughly a one to one ratio of Fe to Si. This was expected because the sample is FeSi. As the profiling moves deeper into the sample we see the continuous decrease of Fe content. Just as the second layer is crossed a slight but sudden increase in Fe content is observed in the atomic profile followed by a continuous decr ease as the substrate is reached. A passive circuit such as the one in fi gure 6.1.6 was used to roughly simulate the interfaceÂ’s resistance dependence on temperature. This circuit assumes a four point probe measurement system with the spacing between the leads being equal thereby resulting in equal resistance between the leads. The three la yers of resistance in the circuit represent the resistances of the sample between the leads (r), the interface linking the deposited material to the substrate (R), and the substrat e (rÂ’) respectively. Analysis of the network simulating the sample (film, interface and s ubstrate) shows that the current going through the sample (I2) between the voltage contacts of th e four point probe is equal to R 2 r r' 2 R r 4 R 2 r r' R 2 r r' I r' R r' R I 2 I2 2 2 2 (6.1.1) where I is the current pulse a pplied to the sample by the constant current power supply. From equation 6.1.1 one finds that I2 is equal to I as R tends to infinity. This is consistent with the reality that if the interface no longer allows current to flow, it will have to pass through the deposited material. I2 will have a dependence on the film and substrate resistance when R is zero, which implies that a parallel flow of current will take place if the interface is not highly resistive. Our system actually measures the voltage (V), so if equation 6.1.1 is multiplied by r, we get that the measured volta ge according to our model to be,
47 R 2 r r' 2 R r 4 R 2 r r' R 2 r r' I R R I 2 r V2 2 2 r r (6.1.2) 05101520253035 0 50 100 150 200 250 CountsDepth (nm) Si Fe (a) (b) Figure 6.1.5: (a) Atomic profile and (b) TEM image of an FeSi/Si sample deposited at 400o C.
48 I3 Â’Â’ rÂ’ rÂ’ r Â’R R R r r r R I I I1 Â’ I1 Â’ I2 Â’ I2 Â’Â’ I3 Â’Â’ I3 Â’ I2 I3 I1 Figure 6.1.6: TEM and passive circuit model for the sample. Each layer of the circuit represents one part of the sample. r is the resistance of the film, R is the interface resistance between the film, and the substrate and rÂ’ is the substrate resistance.r.
49 From equation 6.1.2 one could solve for R (the interface resistance) and get the solution rI V 4 rI V rIr' r r' V r r' 8 rIr' r' r V 16 rIr' r' r V 4 R2 (6.1.3) This supplies two solutions for R, one usi ng the additive sign and the other using the subtraction. It was possible to get numerical values for R by fitting the data of an FeSi film for r, and the data fit of a Si substrate for rÂ’. To get an accurate representation of r, FeSi was deposited on sapphire. The sapphire substrate is olates the film since it is not a conductor. By electrically isolating the FeSi film, any resistance data collected would be representative of the film only. The data fo r the sample and the four parameter fit curve are presented in figure 6.1.7. It is recogni zed that almost anything could be fitted by a four parameter curve, but the interest here is to study the characteristics of the interface (R), and not the film itself. The substrate data at high temperatures was also fitted and the generated curve was substituted in equation 6. 1.3 to replace rÂ’. The equation representing r and rÂ’ have the form, 2 1 4 2T c 3 T c 1e c e c r (6.1.4) T a 12e a r' (6.1.5) where c1, c2, c3, c4, a1 and a2 are fitting parameters. Equation 6.1.4 has two terms, the first represent a normal semiconductor equation, wh ile the second term shows what is known as a variable range hopping conduction mechan ism. Equation 6.1.5 indicates that the resistance of the substrate will continue to decrease with temperature. Even though it is known that this is not true, no corrections have been made to compensate for the fact that
50 the resistance will eventually have to increase at low enough temperatures. This is because according to our model and assumptions no current will flow through the substrate below the transition temperature. A data set of measured V and applied cu rrent were also substituted in equation 6.1.3. This in combination with values of resistance from equation 6.1.4 and 6.1.5 were used to obtain R, which was plotted versus temperature. The substitutions in equation 6.1.3 generate a curve as seen in figure 6. 1.8 for the subtraction solution. The addition solution of R generated negative values, and they were not considered. It is possible to fit the low temperature (less than 200 K) generate d R to one curve with two parameters and Figure 6.1.7: Four-parameter fit for the resistance versus temperature curve for FeSi deposited on sapphire to isolate the filmÂ’s resistance characteristics.
51 the high temperature (greater than 234 K) ge nerated R to the same type of equation but with different parameter values. The equation that fits R in both regimes is T be b R21 (6.1.6) where b1 and b2 have different values for the different temperature regimes. On taking these fits along with the equations used for r and rÂ’ and substituting these in equation 6.1.2 we can obtain new fitted valu es of the voltage. These are plotted along with the measured voltage as a function of te mperature in figure 6.1.9. Obviously the fit for R at low temperatures generates a V that fits the data at temp eratures below 200 K. The curve used for fitting R at high temperatures however simulates V much better than 050100150200250300 10110210310410510610710810910101011101210131014 R ()Temperature (K)Figure 6.1.8: Subtraction solution to equation 5.1.3. Data fitted curves were substituted for r and rÂ’ while actual data were substituted for V and I.
52 the low temperature fit of R and actually follows the trend of the measured data after the transition. At temperatures lower than 63 K, both curves are very close. One of two conclusions could be drawn from these results. One is that the electronic character of the interface changes with temperatur e and one would need at least two different equations to describe this behavior, one at high temperatures and one at low temperatures. The second would be that there might be one solution for R that is capable of describing the complete transition therefore describing the electro nic characteristics of the interface. 050100150200250300 10-410-310-210-1 Data R low T fit R high T fit Voltage (V)Temperature (K)Figure 6.1.9: Measured voltages and results of modeling for an FeSi/Si sample based on the passive circuit network of figure 6.1.6. The solid line is th e result of using the R fit for high temperatures and the dashed line is when the low temperature part of R is fitted and used in equation 5.1.2 instead.
53 Refining this crude model was abandoned in light of results from different experiments that suggest that this model might not be the right one to fully describe the interface, and a different, more complex model, which includes electronic components such as diodes and J-FET transistors, would be more descriptive of the behavior of the interface. These results will be discussed in the following section (section 6.2). One way to cut the current path to the s ubstrate is to create an insulating layer between the Si substrate and th e FeSi film. For that purpose a substrate was placed in an oven at atmospheric pressure and environment. The oven was heated up to 1000o C for an hour. This procedure would allow the growth of a silicon oxide layer; which is electrically insulating. After insulating the substrate an FeSi film was deposited at room temperature to minimize the diffusion of the Fe through the oxid e layer. Resistance characteristics of the film with respect to temperature are plotted in figure 6.1.10. The transition seen in films of FeSi when deposited on Si is not observed. This again suggests that a contact or interaction between the film and the Si substrate is necessary for the observed transition to take place. Direct phys ical contact might not be necessary, since the transition is observed in substrates th at have not been etched. These un-etched substrates have a native oxide layer formed due to the dangling Si bonds on the substrate. Even at room temperature in ambient air th ese native oxide layers have a thickness of up to 50 .
54 The following figure (figure 6.1.11) re-e nforces the idea of the current path actually going through the substrat e and then being confined to the film itself after the transition takes place. As the doping level is increased in the s ubstrate, lowering its resistivity to 0.005 cm, the film acts as if it does not exist allowing current to flow through the substrate showing metallic-like behavior. The transiti on is therefore not observed. As the substrate resistance is increased (lower doping levels) the transition appears again. Note that the substrate resistivity is at least 5-50 times greater than what is used in the default setup for sample de position. When doping levels are too low 50100150200250300350 104 Resistance ()Temperature (K)Figure 6.1.10: Resistance dependence of FeSi deposited on SiOx/Si. The transition seen on films deposited on Si substrates is not observed here, ostensibly because of the current being denied access to the substrate.
55 (resistivity > 3000 cm) then the sample behaves as if it is deposited on an insulating substrate. This indicates that there is a ra nge of substrate doping for which the transition can take place, and/or enhanced. Note that the transition magnitude using a 5-50 -cm substrate is less than the magnit ude of a transition using a 1-10 -cm substrate. This again indicates the importance of the subs trate properties, and specifically electric properties, in the role of the transition. In any of the cases where the transition is present three distin ct regions can be seen. Each region with its distinct behavior relates to a specific conduction mechanism. At high temperature (higher than the transiti on temperature) it is eas y to distinguish the 050100150200250300 10-210-1100101102103104 Resistance ()Temperature (K) < 0.005 -cm > 50 -cm > 3000 -cmFigure 6.1.11: Effect of substrate resistivity on the transition. FeSi films were deposited under the same conditions as for films deposited on 1-10 -cm Si substrates.
56 metallic behavior of the sample. At low temper atures (< 175 K) it changes the behavior to one that is more associated with a semi conductor-like behavior, where the resistance increases with a decrease in temperature. The intermediate regime of temperatures constitutes the transition phase. For diagnostic purposes several samples with specific patterns were deposited. If the current path was actually through the s ubstrate then a continuous sample should not be necessary for the transition to take place until the resistance of the interface becomes higher than that of the film. Therefore, a discontinuous film should show the metallic behavior, and go through a transi tion. At low temperatures two results could occur, either the current would go through the interface anyw ay and we would end up with a transition that would look the same as that for the c ontinuous film, or there would be no conduction in the absence of a continuous film and th e interface path is blocked. In the second scenario, no data could be obtained. To probe these scenarios a sample was pr epared with a mask placed in the middle of the film. This mask produced Â“twoÂ” films (two halves of the same film) separated by the bare Si substrate. The contacts for the f our-point probe were pl aced as two contacts on each half of the film as depicted in figure 6.1.13 (a). Figure 6.1.12 shows the results of the electrical measurements on such a sample. Th e transition is present in this sample as if it was a continuous film. This indicates that the current actu ally flows through the interface, across the s ubstrate, and through the interface again. A second masked sample was prepared where only four FeSi dots were deposited on a Si substrate as seen in figure 6.1.13 (b). When resistance versus temperature measurements were made it was found that at high temperatures the sample followed a
57 metallic behavior, but at 260 K negative resist ances were being measured which might be an indication of bad contacts being made. Th e main difference between the first masked sample and the second one is that in the firs t case one current lead was in contact with one voltage lead through half of the deposited film. The contacts were essentially made in pairs to the sample. Even though the film be tween the leads was highly resistive it was not insulating. As a second step to the expe riment one current lead was shorted to the nearest voltage lead with silver paint (see figure 6.1.13 (c)) forming a three-point probe. The resistance curves obtained with respect to temperature indicated that a transition was 50100150200250300350 101102103104105 Resistance ()Temperature (K)Figure 6.1.12: Resistance dependence on temperature of a discontinuous sample deposited on a Si substrate. This is further evidence that the current path is not necessarily though the film itself, but through the substrate instead.
58 present in one current directi on and not in the other. The remaining two contacts were shorted with silver paint and a similar resi stance curve was obtained for both current directions (see figure 6.1.13 (d)), and they both exhibit the transition. The result of these experiments is depicted in figure 6.1.14. Note that the resistance axis in figure 6.1.14 is presented in a linear scale, in contrast to the form that the data has been pr esented so far. Since ne gative resistance was measured it would not have been possible to present the negative values on a log scale. The voltage leads were not able to detect or sense the transition when they were isolated from the substrate. The voltage drop is somewhat the same in magnitude and the same in sign in either current direction, the negativ e resistance being introduced by dividing the voltage by a negative number ( opposite current direction). Wh en the voltage leads were shorted to the current leads, the instrument wa s able to detect the voltage applied to the film substrate junction to pass a constant current determined by the user. The presence of the transition is reproducib le. We have observed this transition in all samples that we have deposited under th e conditions mentioned in chapter 3. Even though the transition itself is reproducible, th e temperature at which the transition takes place, the magnitude of the transition and th e general dependence of the resistance with respect to temperature is not.
59 Figure 6.1.15 shows the data from th ree samples deposited under the same conditions. The same FeSi target was used fo r all samples. The deposition was made at 4 I1 V1 V2 I2 Si FeSi FeSi Si FeSi FeSi FeSi FeSi Si FeSi FeSi FeSi FeSi Si FeSi FeSi FeSi FeSi I1 V1 V2 I2 (a) (b) (c) (d) Figure 6.1.13: Samples deposited using a contact mask. (a) A discontinuous sample exhibited the transition as a continuous one would. (b) Four dots of FeSi deposited on Si using a contact mask. (c) Two of the dots where electrically shorted by silver paint producing a 3-point probe. (d) The last two contacts were shorted rendering effectively a 2-point probe.
60 Hz for 5 minutes and the subs trate was held at about 420o C. From the figure it can be observed that the samples MS104 and MS115 ha ve a very close tr ansition temperature but differ from sample MS149. MS149 and MS115 exhibit a similar magnitude transition, which is higher than the one seen in MS104. At low temperatures all three samples behave differently. This behavior is also seen in samples th at have been deposited for longer times to make thicker films. It is believed that this is due to the relatively high pressures (low 10-6 50100150200250300350 -1x1040 1x1042x1043x1044x1045x1046x1047x104 Resistance ()Temperature (K) 4 leads + 4 leads 3 leads + 3 leads 2 leads + 2 leads -Figure 6.1.14: Resistance vs. temperature for four dot-masked sample. Four leads, all current leads and voltage leads are separately connected to one FeSi dot. Three leads, one current and one voltage leads are shorted. two leads, the remaining two contacts (one current and one voltage) are also shorted. The plus and minus sign is just a current direction representation.
61 torr) at which the samples are being deposited, in combination with possible contaminants, which might be present at di fferent stages of sample preparation and deposition. The system used for these deposi tions is not an ultra-high vacuum chamber and is used to deposit other films as well. These big differences were not observed for samples that were deposited as a series (i.e. one sample right after the other). Figure 6.1.16 shows the resistance dependence on temperature for samples deposited under the same conditions but post annealed in-situ for different periods of time at the deposition temperature. From this figure one can draw tw o conclusions. Firstly th at in-situ annealing 050100150200250300350 101102103104105 Resistance ()Temperature (K) MS149 MS115 MS104Figure 6.1.15: Three samples of FeSi deposited under the same conditions. All samples show the transition in general, but the details of the transitio n and the behavior in the different regions is not reproducible.
62 at deposition temperatures does not significantly affect the characteristics of the transitions, and secondly, that samples deposited under the same conditions, and consecutively, produce results that are reproducible. Samples that have been annealed in-s itu at the deposition temperature did not change the filmÂ’s characteristics as shown in figure 6.1.16. It was decided to place one of the films back in the chamber and anneal it at higher temperatures. The results of in-situ 50100150200250300 101102103104105 Resistance ()Temperature (K) 0 min 15 min 30 min 60 min 120 minFigure 6.1.16: Resistance versus temperature curve depend ence on in situ sample annealing. All samples were deposited under the same conditions but were annealed for different times at the deposition temperature, which was 400o C.
63 annealing at 400o C and post-deposition annealing at 600o C are shown in figure 6.1.17. It is clear that even though the transition did not vanish, it was certainly modified. There is a dip in resistance of the post-annealed sample that coincides with the transition knee for the un-annealed sample. The dip in resist ance was followed with a sharp rise in resistance. This sudden increase was sharper th an the increase in the as-deposited sample and equal in magnitude. To see if this modification is temperature dependent a second sample was annealed at 900o C after resistance measurements were made on a chip from the same sample as deposited. From figure 6.1.18 one can clearly see that the transition was destroyed. The post-annealed sample shows a dip in resistance at the transition point, similar to the one observed in figure 6.1.17, except that it was not followed by a sharp increase. There is no data after 165 K because the instruments were measuring a negative resistance. The significance of this negativ e resistance is not exactly known. It is interesting to note that a sample that has been deposited at 900o C does exhibit a transition even though it is to a lesser degree (see figure 6.1. 3). One explanation could be that the characteristics depend on the depositi on temperature and are destroyed when they are brought up to higher temperat ures during annealing. However, it is much more likely that the cause is related to the contamina tion of the sample surface when it was exposed to atmosphere to perform the resistance measurements. These contaminants, such as oxygen, might be diffusing through the sample modifying it and modifying the filmÂ’s interaction with the subs trate and the interfaces.
64 One of the characteristic s that was investigated was the dependence of the resistance on the current that was used during the measurement. Figure 6.1.19 shows the resistance dependence on temperature and current. It is clear from this figure that there is a dependence of the transitionÂ’s characteristics on the amount of current used. At low currents the transition takes pl ace at lower temperatures. Ev en though the total increase of resistance for low currents is higher than fo r higher values of curre nt, the transition is broader. All three current values follow the same metallic like behavior before the 50100150200250300350 101102103104105 Resistance ()Temperature (K) As deposited Post annealedFigure 6.1.17: 600o C post annealing of an FeSi film. The sample was deposited at 400o C and was annealed at deposition temperature in situ for 1 hour Resistance measurements were made on one piece of the sample and a second piece was place d in the chamber and annealed at 600 o C.
65 transition with the same slope, but after th e transition the low temperature behavior changes for different current values. An intere st in studying the IV ch aracteristics of these samples emerged from these results. Such IV curves were generated simultaneously as the resistance measurements were take n. The temperature at which resistance measurements were to be made were preprogrammed into the system and at each temperature step the resistance was measured for different currents. Data of the current used and voltage measured were stored at the same time to be analyzed later on. To investigate if this behavior is limited only to FeSi/Si samples, where Si must have a certain doping, some other preliminary experiments were performed by depositing Fe, CoSi and TiSi on Si substrates with resistivity values of 1-10 -cm. Fe and CoSi samples exhibited the transition while samples of TiSi or Ti on Si did not. Results of these experiments are illustrated in figure 6.1. 20. This indicates that not any metal, or transition metal, deposited on Si, would produc e a transition similar to the one observed in FeSi samples. Even though all three metals Fe, Co, and Ti are transition metals, only samples containing Co and Fe deposited on Si exhibited the resistance transition as the temperature was lowered. This is interestin g since Fe and Co are considered magnetic materials (with a net magnetic moment) while Ti is not. This curious result signaled the need for some magnetic measurements, wh ich are discussed in a later chapter.
66 50100150200250300350 100101102103104105 Resistance ()Temperature (K) As deposited Post annealedFigure 6.1.18: 900o C post anneal study of an FeSi film. Th e resistance characteristics of this sample were measured before placing a piece of the same sample in the chamber and annealing it at 900o C. This sample was deposited at the same temperature, as the samples in figure 5.1.16, but was not in situ annealed.
67 50100150200250300350 101102103104105106 Resistance ()Temperature (K) 1 A 50 A 100 AFigure 6.1.19: Resistance dependence on current for FeSi fi lms. This figure shows that the resistance characteristics of the film are not oh mic. This means that the dependence of resistance on current is not linear.
68 050100150200250300 101102103104105106107108 Resistance ()Temperature (K) TiSi Ti Fe CoSiFigure 6.1.20: The resistance dependence on temperature of different transition metals (TM) and TM silicides deposited on Si. All samples were deposited under the same conditions.
69 Section 6.2 IV Characteristics Figure 6.1.19 showed that the resistan ce was not constant when measured using different current values throughout the temper ature range. The non-constant behavior of resistance is an indication that the relation be tween current and voltage is not linear at all temperatures. This ignited the interest in st udying the IV characteristics of the sample under different temperatures. The IV studies would give an indication of the conduction mechanisms that are governing carrier trans port in the samples and causing the sharp increase in resistance as a function of temperature. From the resistance versus temperature curve one can identify three distinct regions and two knees smoothly connecting one region or mechanism to another. Figure 6.2.1 (a) is a set of typical curves of the IV characteristics of an FeSi sample deposited at 400o C. These curves were taken from different points along the resistance temperature curve, which is shown in figure 6.2.1 (b). The data points at which the IV curves are shown in (a) are marked with arrows in (b). The IV plot exhibit a linear relation from room temperature and down to the transition point in the resistance vs. temperatur e curve. This could have been expected by observing figure 6.1.19. The negative linear sl ope down to the transition knee indicates a metal-like conduction mechanism.
70 -8x10-3-6x10-3-4x10-3-2x10-30 2x10-34x10-36x10-38x10-3-100.0 -50.0 0.0 50.0 100.0 -2.0x10-1-1.0x10-10.0 1.0x10-12.0x10-1-100.0 -50.0 0.0 50.0 100.0 -4x100-2x1000 2x100-100.0 -50.0 0.0 50.0 100.0 -1x101-8x100-4x1000 4x1008x1001x101-100.0 -50.0 0.0 50.0 100.0 Current (A) 300 K 255 K Current (A)Voltage (V) 185 K Voltage (V) 20 K050100150200250300 102103104105 20 K 185 K 300 K 255 K Resistance ()Temperature (K) ( a ) ( b ) Figure 6.2.1: (a) IV characteristics of an FeSi fi lm deposited on a Si substrate at 400o C and (b) the samples resistance dependence on temper ature measured with a current of 100 A.
71 When two different materials get in cont act with each other a junction is formed. This junction could exhibit properties as in pn junctions or metal semiconductor junctions where band bending is present and barriers fo r conduction from one side of the junction to the other are formed. This enables the stru cture to act as a rect ifier, allowing current from one side of the junction to flow (forwa rd bias) while impeding the current flow from the other side of the junction (reverse bias). This junction formation exhibits a nonlinear relation (exponential) be tween current and voltage for forw ard bias, and constant (ideally zero current) up to a breakdown voltage. At vol tages higher than the break down voltage current flows for reverse bias. In the absen ce of these barriers, a linear relation between Figure 6.2.2: IV curves of a discontinuous FeSi sample. The sample was deposited on 1-10 -cm Si substrate -6x10-2-4x10-2-2x10-20 2x10-24x10-2-100.0 -50.0 0.0 50.0 100.0 -6x10-3-4x10-3-2x10-30 2x10-34x10-3-100.0 -50.0 0.0 50.0 100.0 -1x101-8x100-4x1000 4x1008x100-100.0 -50.0 0.0 50.0 100.0 -4x101-3x101-2x101-1x1010 1x1012x1013x101-100.0 -50.0 0.0 50.0 100.0 Current (A) 275 K 300 K Current (A)Voltage (V) 220 K Voltage (V) 50 K
72 current and voltage is present and it is said that the materials form an ohmic contact, where the current and voltage obeys OhmÂ’s law and the resist ance is just the slope of a straight line in a V versus I curve. Based on this, and the negative slope for resistance vs. temperature plot, it could be said that down to the tran sition temperature the film substrate junction forms an ohmic contact, a nd the conduction mechanism is that of a metal-like one. As the temperature is lowered and the sa mple goes through the transition, the IV curves begin to gradually show a non-linear relation. This relati on progresses to a maximum effect shown as the IV curve from figure 6.2.1 (a) at 255 K. This non-linearity shows a tendency towards saturation in current. A first reaction to explain this is that the voltage signal that is being measured is due to only part of the current as it travels through the film while the rest of the current is moving th rough the substrate, giving the impression of a current saturation. This anal ysis changes as we take into account the results of the masked film measurements where a discontinuous sample was deposited and its resistance measurements were taken. The discontinuous sample showed the same resistance dependence on temperature as a continuous sample (see figure 6.1.12). Also, the IV curves of the discontinuous sample we re found to have the same results as the continuous sample, as seen in figure 6.2.2. Th ese results re-affirm that the current travels across the film substrate junction and not n ecessarily along the film, and the transition would be due to changes taking place at the fi lm and substrate interface. It is thought that this (the resistance transition and the curre nt saturation) is due to gradual current suppression through the different layers present between the bulk of the sample and the substrate. Based on the evidence of Fe diffusion in Si and SiOx, it could be assumed that
73 Fe intrusions, nanowires or islands, thr oughout the layers between the film and the substrate are present and they act as conduc tion paths. As the temperature is dropped below the linear relation between current and voltages, these conduction sites are modified due to a charge freeze out or loca lization. The freeze out changes the carrier density throughout these conduction paths, creati ng different carrier de nsities between the conduction sites and the host layers. As voltage is applied to drive the current through the conduction path a pinch-off point is created. This is observed in the IV curve as a slopeover towards saturation in current, which mimics IV curves for a field effect transistor (JFET). A J-FET consists of a channel of one t ype of doped semiconductor (n or p type), which is sandwiched between two layers of the opposite type of semiconductor (p or n type) as seen in figure 6.2.3 (a). The two ends of the channel are usually named the source (S) and the drain (D) of the device, sin ce they act as the source and the drain of the charge carrier. The top and bottom layers ar e called the gate (G ) and are usually tied together electrically. If the voltage at the drain (VD), source (VS) and gate (VG) are the same, the device would be in equilibrium and a depletion layer woul d be present at the junction of the channel with th e gate layers as expected from a pn junction. As small values of VD are applied and maintaining VG at zero (figure 6.2.3 (b)), a small current will flow through the channel and one would obser ve a linear relation in the IV curves as seen in figure 6.2.4 (a). So for small values of VD the J-FET acts as a regular ohmic resistor. When voltage at the drain is in creased further the ch annel width decreases considerably as the depletion layer increases. Th is decrease in channel width is due to the increased reverse bias between the gate and the channel; the proce ss is shown in figure
74 6.2.3 (d). A slope-over due to channel narrowing is observed in the IV curve as a result of the increase in VD, see figure 6.2.4 (b). If VD is increased furthe r the depletion region P+N P+ 1 2 3 4 5 V 0 V Widens Pinch-off Current Flow region ID S G G (a) (b) (c) (d) (e) Figure 6.2.3: J-FET diagrams. These diagrams depict the relation between VD and the depletion region and how they effect the current. (a) Equilibrium conditions just s how the depletion layer of a pn junction. (b) At low applied VD the IV the J-FET act as an ohmic resistor. (c) For an applied voltage at the drain, a voltage distribution through the channel will appear, this will cause the depletion regions to be wider at the higher voltage end as in (d). (e) If VD is high enough a pinch-off point is established. R. F. Pierret, Semiconductor Device FundamentalsÂ”, Addison-Wesley Publishing Company, (1996)
75 from both sides of the channel will come in c ontact, as seen in figure 6.2.3 (e), decreasing the channel width further and creatin g a pinch-off point. The value of VD at which the pinch-off point is achieved is known as saturation voltage (VDsat). Any increase in VD above saturation voltage would show as saturation in current (IDsat) as seen in the IV curve of figure 6.2.4 (c). The equation that describes the relation between current and voltage for a J-FET device below saturation voltage is as follows [R. F. Pierret, (1996)]: 2 3 2 33 2P bi G bi P bi G bi D P bi D DV V V V V V V V V V V V I (6.2.1) Here ID and VD are the drain current and voltage respectively, Vbi is the voltage drop across the depletion region under equilibrium conditions, VG is gate voltage and Vp is the gate bias at which the conduction channe l is depleted (pinch-off voltage) and VD is set to zero. The built in voltage is characteristic of the materials involved and the pinch-off voltage is characteristic of both the mate rials involved and the dimensions of the conduction channel. A three parameter, and a two-parameter fit for one of the samples at intermediate temperatures, where slope over due to channe l narrowing is observed, is shown in figure 6.2.5. Even though a physical meaning for the parameter values obtained by the fit have not been extracted, this still illustrates that th is process (current slope over) is apparent. It is not expected for the behavior of the sa mples to have an identical current voltage relation as that of a J-FET. Since it is not pos sible to fix the voltage of what would be the gate in the deposited samples to a certain value, it effectively gives VG the characteristics of a floating voltage. The geometry involved in the samples is somewhat different than
76 VD ID Linear VD ID Slope-over due to channel narrowing VD ID IDsat VDsat Pinch-off point Approximately level for VD > VDsat (a) (b) (c) Figure 6.2.4: IV curves for the J-FET under various conditions, (a) small values for VD, (b) moderate VD values and (c) VD VDsat. R. F. Pierret, Â“ Semiconductor Device Fundamentals Â”, Addison-Wesley Publishing Company, (1996).
77 the one-dimensional analysis invol ved in obtaining equation (6.2.1), so it is expected to find similar behavior to a F ET device, but exact equations simulating our structure must be derived theoretically. Also one must consider the possible complexities of the carrier density along the current path in sample s like the ones discussed here, since no engineering is being employed to control the diffusion or reaction of the different elements involved in the sample, reactions involving elements from the material being deposited, and the substrate w ith itÂ’s dopant and surface prope rties. The parameter that was controlled was the carrier density in the substrate. Results of such an experiment have been illustrated in figure 6.1.11 enforc ing the notion that carrier density along the Figure 6.2.5: IV fit for intermediate temperatures. Two and three parameter fit for an IV curve where current saturation is taking place. 0.0 5.0x10-21.0x10-11.5x10-12.0x10-10.0 50.0 100.0 Current (A)Voltage (V) data a V1/2 + b V + c V3/2 d V1/2 + e V
78 current path is an important parameter, a nd that its variation produces very noticeable effects, preserving the transition or elimina ting it by making the sample metallic like or highly resistive like, throughout the te mperature range of 320 to 20 K. The intermediate temperature IV curve, where current saturati on is apparent, does not behave in exactly the same for the diffe rent samples, such as room temperature deposited or high temperature deposited sa mples. The fitting parameters might be considerably different for different samples. Fi gure 6.2.6 (a) and (b) is an IV curve for an intermediate temperature region of a 400o C and room temperature deposited sample respectively. As compared to figures 6.2.1 a nd 6.2.7 it can be seen that the curves in figure 6.2.6 tend to a slope-over at lower curre nts. Since this change in behavior is apparent in both room temperat ure and high temperature deposite d film it is not attributed to deposition conditions but more toward po ssible contaminants which also could have diffused through the oxide layer and/or modified the electronic structure of Fe formations present throughout the layers be tween the film and substrate. A better reproducibility of the IV curves is expected to be possible if the conduction channe ls were engineered. As the sample was cooled down from 255 K the IV curves gradually began to return to an ohmic-like behavior. After th e resistance goes through the second knee, the IV curves again began to show a non-linear relation. This ap parent ohmic behavior close to the second knee is only due to a transi tion period and is observed because of an intersection of resistan ce vs. temperature curves for different currents (see figure 6.1.19). The sample began to show a slight upward curvature as if tending toward a voltage breakdown. This is most evident in figure 6. 2.7 where IV curves of a room temperature
79 deposited sample are shown. The room temper ature deposited sample exhibits the same IV behavior as the high te mperature deposited sample except at low temperatures. When room temperature deposited samp les make a transition through the second knee, the IV figure begins to curve toward a more evident voltage breakdown than the one shown for high temperature deposited samp les. Figure 6.2.8 (a) and (b) display the IV relation for a commercial pn junction, both forward and reverse bias, and reverse bias Figure 6.2.6: IV curves for room and high temperature de posited samples. These do not show the same rate of saturation as samples shown in figures 5.2.1 and 5.2.2. -1x100-5x10-10 5x10-11x1002x1002x100-100.0 -50.0 0.0 50.0 100.0 -1.0x10-1-5.0x10-20.0 5.0x10-21.0x10-11.5x10-1-100.0 -50.0 0.0 50.0 100.0 Current (A)Voltage (V) Room Temperature Current (A) 400o C
80 in detail respectively. In figur e 6.2.8 (a) one can easily identify the forward and reverse bias regions for the pn junction. The forwar d bias region is the region where current passes through the junction with low applied voltage, which is the relation between the positive voltage and current in the figure. The reverse bias is when ideally no current passes through the junction until a break down through avalanching or Zener effect takes place after a certain voltage. This breakdown volta ge is a characteristic of the device and the materials involved. The reverse bias beha vior is shown as th e relation be tween the negative voltage and current in figure 6.2.8 (a). But real junctions do let some current through the junction in reverse bias mode and this is appreciated in figure 6.2.8 (b). -2x10-3-1x10-30 1x10-32x10-3-100.0 -50.0 0.0 50.0 100.0 -8x10-2-6x10-2-4x10-2-2x10-20 2x10-24x10-26x10-28x10-21x10-1-100.0 -50.0 0.0 50.0 100.0 -1x101-8x100-6x100-4x100-2x1000 2x1004x1006x1008x100-100.0 -50.0 0.0 50.0 100.0 -1x101-8x100-6x100-4x100-2x1000 2x1004x1006x1008x1001x1011x101-100.0 -50.0 0.0 50.0 100.0 Current (A) 300 K 257 K 230 K Current (A)Voltage (V) Voltage (V) 140 KFigure 6.2.7: IV characteristic curve for a room temper ature deposited sample. These figures show similar behavior except at low te mperature the upward curvature is mo re pronounced in these samples.
81 If a comparison between figure 6.2.8 (b) a nd figure 6.2.7 at low temperatures (140 -40-35-30-25-20-15-10-505 -350 -300 -250 -200 -150 -100 -50 0 Current (pA)Voltage (V)(a) ( b ) Figure 6.2.8: IV curve of a commercial pn junction. This image was scanned from Â“ Semiconductor Device FundamentalsÂ” R. F. Pierre page 261-162. (a) This figu re clearly shows the IV characteristics of pn junction, forward and reverse bias. (b) Detailed Current voltage relation for reverse biased pn junction. -40-35-30-25-20-15-10-50 -1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 Current ()Voltage (V)
82 K) is made one can notice the similarities. It has been established that the current passes through the sample substrate junction, and it actually goes through the junction twice. When the current pulse is applied, it travels from the film to the substrate and from the substrate to the film to be collected by the second lead of the power supply. If a conduction barrier is built due to the contact of different materials and the freeze out of charge transport through the in terface layers takes place, as argued above for intermediate temperatures, the current will pass through two p-i-n like junctions, one entering the substrate and one leaving. As it goes through th e first junction it could be in forward bias (reverse bias) mode, depending on which si de of the junction is p or n type semiconductor, but as it exits the substrate it will encounter a reve rse biased (forward biased) junction. The current will go through both junctions as long as the breakdown voltage of the junction is lower than the ma ximum voltage of the constant current power supply and the resistance of the film itself is higher. As one measures the voltage, the measurement would be equivalent to the m easurement of the voltage drop across two resistances in series. The forward bias regi on resistance would be low compared to the resistance of the reverse bias junction, and as a result the measurement would effectively show the characteristics of the reverse bias mode. There is one more junction with the possi bility of interface effects taking place; this is the interface between th e lead contacts with the f ilm itself. To rule out the possibility that the non-linearities observed in the IV data are due the metal contacts with the film rather than the film-substrate interface, figure 6.1.4 and 6.1.10 should be considered. If the transition was due to the leadÂ’s interface with the film, a transition
83 would have been present in these samples wher e the only difference is the isolation of the film from the Si substrate (FeSi/SiOx/Si and FeSi/sapphire).
84 Section 6.3 Resistance and Magnetoresista nce as Measured by PPMS. The resistance dependence on temperature for samples of Fe, FeSi and CoSi deposited on Si substrates exhibited a transi tion where the sampleÂ’s resistance went from metallic like characteristics to a highly re sistive like state. This transition was not observed in Ti or TiSi samples. Acknowledgi ng that Fe and Co are considered magnetic, it was thought that magneto-transport experime nts could shed some light on the nature of the transition, whether the transition is inherent to the deposited material or whether it is characteristic of the sample substrate interaction and structure. Experimental measurements of magnetoresist ance (MR) were performed, where magnetoresistance is defined as the measurement of the relative cha nge in resistance with the application of a magnetic field. Four samples of FeSi were deposited on Si to study the effect of an applied magnetic field on the transport properties of Fe Si films and the nature of the transition. Two samples were deposited at room temper ature, one for 5 minutes and another for 45 minutes, and two other samples were deposited at 420o C, again one for 5 minutes and another for 45 minutes. All the samples were deposited at approximately the same pressure and laser fluence. The measured thickness of the 45 minutes deposited sample at room temperature was 695 This generates a growth rate of 0.06 /laser pulse or 15.4 /minute. This set of samples should give some information on the dependence of the
85 resistance transition on the magnetic field, sa mple thickness and de position temperature. Also, a confirmation of the transition measured by a different system would eliminate any doubt about whether the transition observed at our laboratory is real or just a systematic error. Therefore, the following measurements on samples prepared at our laboratory were performed in the Physical Properties Measurement System (PPMS) at USF. Results of resistance dependence on temperature for the four samples are presented in figure 6.3.1. Several features can be observed from this figure. All samples, independent of the deposition conditions, show the metal-to-insulator transition. The Figure 6.3.1: Resistance versus temperature measured by the PPMS. The only two parameters varied for the deposition of these FeSi samples are the substrate temperature and the deposition time. All samples exhibit the transition. 050100150200250300350 100101102103104105106107108 420o C 45 min 25o C 45 min 420o C 5 min 25o C 5 min Resistance ()Temperature (K)
86 transition takes place at lower temperatures for samples deposited only for 5 minutes. This suggests a dependence of the transi tion temperature on thickness. From the 5 minutes samples one could infer that the transition knee is lower for the sample deposited at 420o C, suggesting another dependence of the transition temperature. From the same figure one can see that for the samples depos ited for 45 minutes the one deposited at room temperature exhibits a slightly lowe r transition temperature, contradicting the conclusion drawn from the 5 minute deposit ed samples. The transition temperature dependence on thickness or minutes of deposition in figure 6.3.1 are systematic variations. As illustrated in chapter 6.1 figure 6.1.15, samples deposited under the same conditions exhibit different transition temperat ures. It is more likely that the transition temperature is strongly dependent on impur ities introduced during sample preparation and/or variations in the subs trate properties. These substrat e variations would be due to doping variations from wafer to wafer. From the results in figure 6.3.1 and othe r experiments, it is consistent that the magnitude of the transition is thickness and temperature dependent. Samples deposited at low temperatures (room temperature for exam ple) exhibit higher magnitude transition then samples deposited at higher temperatur es. Similar behavior is consistent with thickness; films deposited for 5 minutes s how higher magnitudes of the transition compared to films deposited for 45 minutes under the same conditions. One feature that was observed in the data taken by the Physical Properties Measurement System (PPMS) and displayed in figure 6.3.1 that was not observed in the data taken at our laboratory is a change in slope during the sharp increase in resistance and the upward curvature at low temperatures. Careful analysis of the raw data showed
87 that the current applied to the sample during the measurement changed from the set maximum value as temperature dropped. This is due to the operation parameters of the PPMS. For resistance measurements using the PPMS, the user sets maximum values for current, voltage and power. The PPMS operate s without going over these maximums. As the temperature drops the resistance increases fo rcing the current to drop in order to stay under the limits of maximum voltage and power From chapter 6 section 2, we saw that the current and voltage share a nonlinear rela tion causing the resistance measured to have a dependence on applied current. This depe ndence explains the discrepancy in the behavior of measurements made at our la boratory and the PPMS. Figure 6.3.2 illustrates the current output of the PPMS with temperature for one of the samples. A constant current of about 100 A was output by the system down to 265 K, at which temperature the current output dropped sharply down to about 1 A and continued decreasing but at a much slower rate. The current at the lowest temperature ( 20 K) was about 0.03 A. Subsequent to confirmation of th e transition by a second independent experimental setup (PPMS), magnetoresistan ce measurements were performed. Figure 6.3.3 contains data for magnetoresistance m easurements vs. temperature for different applied magnetic field orientations with respec t to the surface of the sample and assumed current path. All in-field resistance measuremen ts were made in a field of 7 Tesla (T) in magnitude. The data in figure 6.3.3 corresponds to FeSi/Si deposited at room temperature for 45 minutes. The orientation of the field w ith respect to the samp le and the electrical contacts is depicted in figure 6.3.4. Also in figure 6.3.3 is the resistance dependence on temperature (thick solid line) as measured with a zero applied magnetic field. The left axis of the figure is magnetoresistance (MR) in units of percent and was calculated as:
88 0 10 R R R MR0 0 H (6.3.1) where RH and R0 are the resistance measured in a nd out of the applied magnetic field respectively. The right axis is the resistance as measured out of field and the data is presented on a log scale. An out of field re sistance was measured before an in-field resistance measurement was taken. These two co nsecutive runs (in and out of field) were used to calculate the magnetore sistance as in equation 6.3.1. 050100150200250300 0 20 40 60 80 100 265 K Current (A)Temperature (K) 7T out of plane 0TFigure 6.3.2: PPMS current output. Below 265 K, the current drops to maintain the limits of voltage and power of the PPMS set by the user.
89 Several features can be observed in figure 6.3.3. A maximum value of about 47 % for magnetoresistance sharply rises as th e transition knee is just passed. A sudden decrease takes place close to but before the decrease in current output of the PPMS. A smooth hump follows the sharp spike in ma gnetoresistance coinciding with the second knee of the resistance curve and settles to neglig ible values relative to the spike seen at higher temperatures. Even though the general behavior is followed by all the magnetic field configurations with respect to the sample surface and current path, the maximum magnetoresistance was measured when the magne tic field was applied parallel to sample surface and current path (inlin e with the four contacts on th e surface of the sample). The Figure 6.3.3: Resistance and magnetoresistance dependence on temperature. The FeSi sample was deposited for 45 minutes at room temperature. The magnitude of the applied magnetic field was 7 T. 20406080100120140160180200220240260280300 -10 0 10 20 30 40 50 101102103104105106107 B Perpendicular to sample B parallel to sample and I B in plane perpendicular to I ((R7T-R0)/R0) 100Temperature (K) 270 K 0 B Resistance Resistance ()
90 current path is defined here as the path that the current would have taken along the sample instead of across the sample as demonstrated in chapter 6 section 2. The fact that the highest spike in magne toresistance coincides with the sharpest change in resistance might indicate some type of phase change, which is affected by the presence of the magnetic field. This phase is not necessarily chemical or crystallographic, but a change in the conduction mechanism governing the sample. Figure 6.3.5 shows a FeSi I1 V1 V2 I2 Si B (a) FeSi I1 V1 V2 I2 Si B (b) FeSi I1 V1 V2 I2 Si B (c) Figure 6.3.4: Applied magnetic field orientation with respect to the film. Applied field (a) in plane and parallel to the assumed current path, (b) in plane and perpendicular to the current path, and (c) out of plane.
91 plot of the peak of the MR, which coincides with the change in the derivative of the resistance with respect to temperatur e signaling the change in the conduction characteristics or mechanisms. The resistan ce curve used in the differentiation was one that was measured at an applied magnetic field of 0 T. One must be careful and make sure that this peak in MR is real and not a systematic error. A systematic error was disc ussed earlier in this ch apter for the resistance measurement with the PPMS, when the slope of the curve of resistance with respect to temperature changed half way through the transi tion. This was attributed to the changing current output of the PPMS and not some thing characteristic of the sample. 0 10 20 30 40 50 210220230240250260270280290300 -3000 -2000 -1000 0 1000 dR/dT (/K)Temperature (K) dR/dT, R at 0T MR (%) MR 7TFigure 6.3.5: dR/dT versus temperature for an FeSi/Si sample. Maximum of MR coincides with the change in dR/dT.
92 One of the systematic errors that could be present is that the resistance reading from one run to another could change due to heating of the sample if the sample does not dissipate heat at a rate that is high enough. Since the temperat ure sensor is not in direct contact with the sample, the temperature re ading always has some error. This would cause the reading of one temperature and reco rding the resistance at another temperature, creating a shift in the data. If one compares tw o different runs and a shift is present, one can expect to see a spike in difference wher e the resistance shows the highest rate of change. Figure 6.3.6 exhibits the equivalent magnetoresistance of a similar error. The Figure 6.3.6: Percentage difference in resistance due to a temperature shift. The resistance measured at a lower temperature was shifted up by one temperature step (RH) and used to calculate the percentage difference with respect to resistance at the actual temperature reading (RL). 050100150200250300 0 20 40 60 80 100 (RH-RL)/RL (percent)Temperature (K)
93 data from the same run was used to generate this figure. To get a high positive magnetoresistance we needed to shift the data to higher temperatures than what it was read at. For this kind of shift in data, the real and spurious magnetoresistance figure have the same distinct peak in magnetoresistance, but there are some differences between the two curves that ultimately show that this type of error, at least in a magnitude to affect the real MR, is not possible. The first discrepancy is that at high temperatures, before the transition, the magnetoresistance is negative in figure 6.3.6 while it is positive in figure 6.3.3. At low temperatures and after a di scontinuity at 200 K, the magnetoresistance calculated from the shifted data begins to increase as temperature drops while the data in figure 6.3.3 is approximately constant below 160 K and no discontinui ties are observed at 200 K. For figure 5.3.6 to be generated the whole data set had to be shifted up one temperature step, which is approximately 2 degrees. The difference in measured temperature for the same data point for 0 T applied field and 7 T a pplied field in plane and parallel to the path is between 0.01 % and 8.7 10-6 %. The difference in temperature for the data point around the peak of the MR curve is about 0.001 %. To be able to estimate resistance at lower temperature steps, the data around the knee was fitted to an exponential decay function. A figure containing the data used in th e fit and the equation generated is shown in figure 6.3.7. For a temperature step of 0.004 K, the magnetoresistance was calculated to be only 0.1 % at 273.5 K. This indicates that the magnetoresistance calculated from two different runs, one in-field and one out of field is reliable as long as the sample has not been changed and/or rewired with different contacts between runs.
94 The dependence of magnetoresistance on field magnitude and direction was also studied using the PPMS. Figure 6.3.8 illustrates such measurements at different temperatures for a magnetic field being app lied perpendicular to the sampleÂ’s surface. The sample is FeSi/Si deposited at room temperature for 45 min. These temperatures were selected as confirmation points from the magnetoresistance versus temperature curve in figure 6.3.3, including the temperature (273 K) at which maximum magnetoresistance was observed. The values of MR at a magnetic field of 7 T for the four 270272274276278280282 20 40 60 80 100 120 140 Data fit y = 33.52 + 8.23 e-x/1.93Resistance ()Temperature (K)Figure 6.3.7: Exponential decay fit near the transition region. In this region a maximum in MR is observed. A shift of 0.004 K in the data would induce only a 0.1% change in resistance. This confirms that the MR measured is not a systematic error.
95 different temperatures in figure 6.3.8 matc h the corresponding values in figure 6.3.3 fairly closely and provide an independen t verification of those measurements. There are three interesting features to poi nt out from figure 6.3.8: the asymmetry for positive and negative applied field at 273 K, the steps or discontinuities observed at 261 K and 257 K, and the non-reproducibility of the MR at 200 K. For the asymmetry for positive and negative applied fields at 273 K, note that the magnetic field was oriented in the positive direction when measurements for figure 6.3.3 were taken. Even though similar results were observed for the magnetoresistance as was observed in figure 6.3.3 for the out of plane curve when the magnetic field is applied in Figure 6.3.8: Magnetoresistance dependence on an out-of-plane applied magnetic field. The sample was deposited at room temperature for 45 minutes. -80000-60000-40000-20000020000400006000080000 0 10 20 30 40 50 -80000-60000-40000-20000020000400006000080000 0 2 4 6 8 10 12 14 -80000-60000-40000-20000020000400006000080000 -2 0 2 4 6 8 10 12 14 -80000-60000-40000-20000020000400006000080000 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 (RH-R0)/R0 (percent)273 K 261 K (RH-R0)/R0 (percent)Applied Field (G) 257 K Applied Field (G) 200 K
96 the positive direction, we observe higher magnitudes for a negative applied field. Different runs of the same sample at different field orientations we re taken as could be seen in figure 6.3.9. The additional runs did not exhibit any asymmetry in our measurement at 273 K, but showed it at 300 K. A CoSi sample deposited under the same conditions was placed along with the FeSi in the same run as for the data in figure 6.3.8. The CoSi sample had also exhibited an as ymmetry but for the oppos ite direction of the applied field and the asymmetry was not as severe. The current was applied in opposite Figure 6.3.9: Magnetoresistance dependence on an in-plane applied magnetic field parallel to current path. The sample is FeSi/Si and was deposited at room temperature for 45 minutes. -80000-60000-40000-20000020000400006000080000 0 5 10 15 20 25 -80000-60000-40000-20000020000400006000080000 -2 0 2 4 6 8 10 12 -80000-60000-40000-20000020000400006000080000 -10 0 10 20 30 40 50 -80000-60000-40000-20000020000400006000080000 0 2 4 6 8 (RH-R0)/R0 (percent)Applied Field (G) 251 K Applied Field (G) 222 K 273 K (RH-R0)/R0 (percent)300 K
97 directions for the CoSi and FeSi samples, wh ich might indicate some relation between the asymmetry and the geometry of measurement setup. The data for the CoSi sample are shown in figure 6.3.10. It is not possible to point out the exact cause of this asymmetry, and there is evidence that sugge sts that this a systematic e ffect, since the asymmetry is present in both samples at about the same temperatures. The fact that they had opposite directions could be related to the fact that the current was in opposite directions. Also the asymmetry disappears for both samples and re -appears at lower temperatures and for different temperatures at different orientati ons suggesting that the effect is actually sample related. Theoretical studies had showed that gradients in the resistivity along the conduction path have an effect on the transver se MR measurement and is more noticeable if the divergence was along the path instead of perpendicular to it. The MR dependence on the orientation of the diverg ence of resistivity is reversed as the length to width ratio of the current path goes to zero. These devi ations form homogeneity introduce a linear MR dependence on the transverse applied ma gnetic field, which would give different magnitude readings as the magnetic field di rection is changed [H. H. Wieder, (1979)]. Clearly more experiments are needed to confirm the nature of this asymmetry. This can be done by repeating the present ex periments on the same samples, performing the same experiments on a different sample deposited under the same conditions and conducting MR measurements using an AC cu rrent instead of DC. The AC measurement will eliminate any thermoelectric effects th at might change the resistanceÂ’s value and influence the MR calculation.
98 The discontinuities observed for both te mperatures of 261 and 257 K in figure 6.3.8 were only observed for negative applied magnetic fields and samples that are room temperature deposited. It is believed that th ese discontinuities are related to the power output restrictions of the PPMS. Figure 6.3.11 is a plot of the current output of the PPMS as a function of the applied field when meas urements where taken for the FeSi/Si sample at 261 K. It is easy to see a nd relate the discontinuities in the current output to the discontinuities in the MR measurements. Th e same figure also explains the deviation from the parabolic dependence in the positive di rection as the current decreases linearly -80000-60000-40000-20000020000400006000080000 -5 0 5 10 15 20 25 30 35 -80000-60000-40000-20000020000400006000080000 0 5 10 15 20 -80000-60000-40000-20000020000400006000080000 0 5 10 15 20 25 30 35 -80000-60000-40000-20000020000400006000080000 0 5 10 15 20 25 30 35 (RH-R0)/R0 (percent)Applied Field (G) 257 K (RH-R0)/R0 (percent) 273 K 261 K Applied Field (G) 251 KFigure 6.3.10: MR of CoSi/Si. Sample was deposited at room temperature for 45 min. The magnetic field was applied normal to the surface of the sample.
99 with the increase in magnetic field beyond 3 T. The same is true for the discontinuities observed at 257 K for the same film. The last feature is that at 200 K we do not reproduce the magnetoresistance observed in figure 6.3.3. Even though we see a negative magnetoresis tance at low applied magnetic fields, it does not maintain the sign or the magnitude as observed in earlier experiments. The current output of the PPMS remains constant for that temperature down to Â–6.3 T and up to 6.5 T, which are magnitudes much higher than the value of the field at which the magnetoresistance changes from negative to positive. A negative -80000-60000-40000-20000020000400006000080000 60.5 61.0 61.5 62.0 62.5 63.0 63.5 64.0 64.5 Current (A)Applied Field (G) 261 KFigure 6.3.11: Current output dependence of the PPMS on the applied magnetic field. The discontinuities in the current output are responsible for the discontinuities observed in the MR measurements and the deviation from its parabolic dependence on the applied field.
100 magnetoresistance is an indication that the chan ge in resistance is due to a spin dependent transport process. This negative magne toresistance is observed in both giant magnetoresistance (GMR) and colossal magneto resistance (CMR) samples or devices, where resistance is decreased when a fiel d is applied to align the spins in the ferromagnetic layers (in the case of the GMR sample) or particles (in the case of the CMR). It is possible that at a certain temperature range and for low enough applied magnetic fields the magnetic impurities in the sample play the dominant role in the MR measurement. The parabolic dependence of the MR on the applied field along with the fact that it is a positive MR indicates th at the effect is due to Lore ntz forces [S. Tumanski, (2001); R. S. Popovic, (2004); S. A. Solin, et. al. ( 2000)]. The force acting on the carriers due to the applied magnetic field distor ts the current paths, which gives rise to a resistance dependence on the field given by 2 2 0CB 1 R R(B) (6.3.2) where R0 is the resistance at zero field, is the carrier mobility and C is a coefficient that is geometry dependent. These distortions can be increased (or modified), and this would increase (or modify) the MR, by changing geom etrical factors of the device or current path [S. Tumanski, (2001); R. S. Popovic, (2004); S. A. Solin, et. al. (2000)]. Geometrical factors such as the length to width ratio of the conduction path, size of the contacts on the samples and the position of the contact leads with respect to the length of the sample introduce an error in the magnetore sistance. For example, a change in length to width ratio changes the magnitude of the magnetoresistance while the dimensions of
101 the electrodes changes the dependence of MR on the applied magnetic field by changing the linearity between the Ha ll voltage and applied fiel d [H. H. Wieder, (1971)]. According to the force a charge experien ces from a magnetic field, a maximum in MR is to be observed when the magnetic fiel d is perpendicular to the current path. The force experienced by the charge is mathematically expressed as B v F qB (6.3.3) where q is the charge of the electron, v is the velocity of the charge carriers and B is the applied field. As seen from previously presented data the maximum MR was observed when the magnetic field was applied parallel to the surface of the sample and parallel to the current path, assuming that the current flows along the film. In chapter 6.1 we showed that this is not the case and that in fact the current flows th rough the film and into the substrate. This path would be effectivel y perpendicular to the applied magnetic field and is consistent with the ma ximum MR observed. Also in ch apter 6.2, it was argued that the conduction mechanism changes as the temp erature is lowered due to freeze out of carriers. Therefore the carri er density changes along th e high temperature conduction channels created by the diffused Fe, or transition metal, from the film to the substrate. After maximum freeze out take s place and a change in conduction mechanism is established by the sample, the effects of the magnetic field on the sample changes and is observed as a decrease in the measured MR. This is consistent with literature which states that the MR due to Lorentz forces is modifi ed due to changes in carrier densities along the current path [C. Herring, (1960)]. Figure 6.3.12 shows the resistance de pendence on the applied magnetic field perpendicular to the sample along with a di fferent data fit for positive and negative
102 applied fields. It was found that two different equations were needed to better fit the data collected, as noted by the equations included in the figure. This difference is due to the asymmetric dependence of the resist ance on the applied field direction. Somewhat different results were obtained for magnetic field being applied in the plane of the sample surface and inline with the four contact leads. The asymmetry was not as severe (see figure 6.3.9), as when th e field was perpendicular and at a temperature Figure 6.3.12: Resistance dependence on the applied magnetic field for an FeSi sample. The increase in resistance and the parabolic dependence on the field is an indication that MR arises from Lorentz forces acting on the charge carriers. Also seen is the fit for positive applied field and the negative applied field. -80000-60000-40000-20000020000400006000080000 65 70 75 80 85 90 95 RB-=66.74 + 5.766*10-9 B2RB+=66.07 + 3.358*10-9 B2 Resistance ()Applied Field (G) R(B) R(-B) fit R(+B) fit
103 of 273 K. The data at which the maximum of MR is observed was fitte d satisfactorily to a single parabolic curve as seen in figure 6.3.13. The magnetoresistance results for FeSi shown so far were collected from one of the samples shown in figure 6.3.1; that sample deposited for 45 minutes at room temperature. The rest of the samples exhibi ted similar characteristics except that the magnitudes of magnetoresistance measured were less for samples deposited at 420o C. Figure 6.3.14 contains the data of magne toresistance dependence on temperature for a sample deposited at room temperature for 5 minutes. The data was collected for a -80000-60000-40000-20000020000400006000080000 90 100 110 120 130 (RH-R0)/R0 (percent)Applied Feild (G)Figure 6.3.13: Resistance dependence on the in-plane ap plied field fit for an FeSi sample. The dependence is clearly quadratic indicating that the magnetoresistance is mostly caused by Lorentz forces acting on the charge carriers. 92.6795+7.8253910-9B2
104 magnetic field that was directed perpendicular to the samples surface. It exhibits the same features as the figures shown for the sa mple deposited for 45 minutes under the same condition. Figure 6.3.15 shows the magnetoresi stance dependence on the magnetic field applied in plane and parallel to the path of the current for a sample deposited at 420o C for 45 minutes. A parabolic dependence is obs erved at high temperat ures (300 K), which vanishes at low temperatures and follows a linear trend with positive slope at 222 K and low values. Figure 6.3.14: Resistance and magnetoresistance dependence on temperature of FeSi/Si. This was a sample that was deposited at room temperature for 5 minutes. The magnetic field was applied out of plane at a magnitude of 5 T. 50100150200250300 -20 -10 0 10 20 100101102103104105106107108 (RH-R0)/R0 (percent)Temperature (K) Resistance ()
105 The difference in MR for samples deposited at room temperature and 420o C could be due to a difference in the struct ure and/or composition of the interface. The higher substrate temperature could encourage gr eater diffusion of Fe to the interface. It also could cause a chemical reaction. This modifies the composition of the interface causing a change in the electronic prop erties of the conduction channels. Figure 6.3.15: Magnetoresistance dependence on magnetic field for a 400o C deposited FeSi sample for 45 minutes. The magnetic field was applied parallel to the current path. -80000-60000-40000-20000020000400006000080000 -1 0 1 2 3 4 5 6 7 8 -80000-60000-40000-20000020000400006000080000 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 -80000-60000-40000-20000020000400006000080000 -0.45 -0.40 -0.35 -0.30 -0.25 -0.20 -0.15 -0.10 -0.05 0.00 0.05 0.10 -80000-60000-40000-20000020000400006000080000 -0.20 -0.15 -0.10 -0.05 0.00 0.05 [(RH-R0)/R0] (pecent)Applied Feild (G) 300 K (RH-R0)/R0 (percent)Applied Feild (G) 273 K (RH-R0)/R0 (percent)Applied Field (G) 251 K (RH-R0)/R0 (percent)Applied Field (G) 222 K
106 Chapter 7 Conclusions and Suggestions This dissertation presents to the best of our knowle dge, for the first time, a temperature dependent transition of up to 5 orders of magnitude in the resistance of selected thin films deposited on Si. Samples of Fe, FeSi and CoSi have been deposited on Si (100) and the resistance transition exhibited by these films was invest igated. Even though the first sample to exhibit the transition was made using the dual laser deposi tion technique it was possible to reproduce the transition using single pulsed la ser deposition. Four point probe resistance measurements have revealed that the transition is present only in two of the three transition metal silicide materials used (FeSi, CoSi), and in Fe films, and not in Ti or TiSi. The tr ansition magnitude and sh arpness was higher in films deposited at room temperature than fo r those deposited at higher temperatures. Samples that are deposited at high temperatures (600 and 900o C) exhibit the transition, but the transition is modified (at 600o C) or destroyed (at 900o C) when annealed at higher temperatures than the deposition te mperature. This might indicate that the diffusion of elements from the sample into the substrate and vice versa are not the key element for the transition to take place. The transition rather depends on a reaction between the arriving species with the subs trate, which would be plume energy and substrate temperature dependent. The transition is consistently present in all the samples
107 deposited but the characteristics of the tran sition, such as the resistance dependence on temperature after the transition and the transi tion temperature itself, changed from sample to sample if they have been deposited at different times, with other experiments involving other materials being made in between. If samples were sequentially deposited on substrates from the same wafer then the tr ansition is completely reproducible. This discrepancy is attributed to di fferences in the resistivity of the different wafers in the same box, since the box of wafers contain wafers with a range of resi stivities (e.g. 1 Â– 10 cm), and to contaminants present in the chamber, since the chamber is used in the deposition of other materials. The fact that the transition is present only for a range of substrate resistivities is a strong indication that contamin ants or dopants play a very important role in the characteristics of the transition. Similarities between the resistance charac teristics of an FeS i/Sapphire and FeSi/Si at temperatures lower than the transition te mperature and similarities between Si and FeSi/Si at high temperatures sparked the anal ysis of a passive thr ee layered resistance circuit. It was possible to extract some in formation on what might be happening at the interface layers observed by the TEM images. Two possible approaches were explored. Either the interface resistance goes through change in its pr operties and two different fitting parameters (for the same type of equation) would be needed to describe the interface resistance, one at temperatures highe r than the transition temperature and one at temperatures that are lower. Or it is possi ble to find one set of fitting parameters to describe the entire trend of th e transition. Since the type of equation is known, a simple iterative process can be used to find the best fitting parameters.
108 Resistance measurements of non-conti nuous FeSi/Si films showed the same characteristics as a continuous sample, suggest ing that a continuous sample might not be necessary. If a continuous sample is not neces sary, then no, or very little current is flowing through the sample itself. This would render the model presented above invalid, since it gives an incomplete pictur e of what is actually taking place. Resistance dependence on temperature showed different values after the transition took place when a different measuring current was used in the determination of the resistance. IV curves obtained at different temperatures show ed three distinct conduction mechanisms taking place throughout the entire temperature range. The relation between applied current and measured voltage show ed a linear relation above the transition temperature. During the sharp increase in th e resistance, a slope over of the current is observed as if tending toward saturation in cu rrent, similar to what is observed in a JFET. At low temperatures, a breakdown similar to what is observed in a pn junction in reverse bias mode is measured. These thr ee different regions smoothly transition into each other. The breakdown voltage is most noticeable in samples deposited at room temperature. Magnetoresistance measurements showed that the resistance (therefore MR) had a parabolic dependence on the a pplied magnetic field. The parabolic dependence is usually a result of Lorentz forces acting on the charge ca rriers. It is believed that this is the case for these samples. The magnitude of the pa rabolic dependence is proven (through out the references presented in this work) to be geometry and carrier density dependent. This relates to this study since the geometry of the conduction paths ta ken by the carriers and the carrier density throughout th e layers of the sample are directly responsible for the
109 transition. MR dependence on te mperature showed a spike in MR coincident with the temperature at which the change in c onduction mechanism takes place, around the transition temperature. The increase in MR doe s not last through out the sharp increase in resistance but actually drops, which might be only a signal that a phase change took place. This is not necessarily a crystallo graphic structure change, but one involving electronic structure. It is believed that a reaction takes place between the material from the plume and the substrate forming conduction paths connectin g the deposited film and the substrate. These paths are plume energy and substrate te mperature dependent upon the arrival of the plume during the deposition. An increase of the temperature after deposition could cause the modification of these conduction paths, therefore modifying or destroying the transition in resistance. As temperature is lowe red a change in electr onic structure of the conduction paths starts to take place, either changing them physically or affecting the carrier density throughout the paths. This is s hown in the IV curves as a slope-over and in the MR as a spike through its dependence on temperature. As the electronic modification of the conduction paths is complete, it is belie ved that a conduction ba rrier is formed as in a pn junction. For conduction to take pl ace the carriers would have to go over the junction in reverse bias, which is seen as a breakdown in voltage. Conduction across the barrier would happen if the breakdown voltage is lower than the voltage needed to drive the current along the sample. Even though the MR results were observed in 8 different samples (four FeSi/Si and four CoSi/Si) it is str ongly recommended that the MR e xperiments be repeated as done in this work, since it was not possible to do so here. Also, it is recommended that
110 similar experiments be done on the same samples using an AC power source instead of a DC power supply. Detailed nano-structure analysis of the in terface is in order for the confirmation or the denial of the existence of such c onduction paths and the carrier concentration dependence on temperature along them. Once the existence of the conduction path s has been confirmed along with their electronic characteristics, an electronic circuit model could be developed which contains electronic elements describing the interface more accurately th an the one presented here in the early stages of the study. A simila r equation for current dependence on applied voltage for J-FET should be derived, k eeping in mind a floating voltage for VG and that the Â“deviceÂ” is three-dimensional instead of two (as in the case of a J-FET) in order to describe the transition region more accurately. Fe and Co are elements that are used in the investigation and development of devices. It is important to understand the in teractions between these elements and Si, since current fabrication technol ogy is Si based. This effect might be unique to PLD, and has been so far to the best of the authorÂ’s knowledge, but we recognize that this transition might be due to a reaction between deposited material and the substrate, and that the reaction might be present for other deposition me thods that are used in the scien tific and industrial communities if th e conditions are favorable.
111 References A. Chainani, T. Yokoya, T. Takahashi, S. Yoshi, M. Kasaya, Phys. Rev. B, 50, 8915, (1994). A. V. Zenkevich, V. N. Nevolin, I. D. Kh abelashvili, O. B. Uvarov, E. V. Chubunova and S. V. Kabanova, Translated fr om Mikroelektronika, 21, 64 (1992). C. Fu, M. P. C. M. Krijn, S. Doniach, Phys. Rev. B (rapid comm.) 49, 2219, (1994). C. G. Shull, W. A. Strauser, E. O. Wollan, Phys. Rev., 83, 333, (1951). C. Herring, J. Appl. Phys., 31, 1939, (1960). C. Kittel, Introduction to Solid Stat e Physics, 7th Ed., Wiley, (1996). D. J. Craik Magnetic Oxides Part 2, John Wiley & Sons, (1975). D. Mandrus, J. L. Sarrao, A. Migliorio, J. D. Thompson, Z. Fisk, Phys. Rev. B, 51, 4763, (1995). E. J. Verwey, P. W. Haayman, Physica VIII, no 9, 979, (1941). F. Bodker, M. Hansen, C. B. Koch, K. Lefman, S. Morup, Phys. Rev. B, 61, 6826, (2000). G. Aeppli, J. F. DiTusa, Mater. Sci. Engin. B, 63, 119, (1999). G. K. Hubler, Â“Pulsed Laser Deposition of thin filmsÂ”, J. Wiley, New York, (1994). G. Kletetschka, P. J. Wasilewski, Phys. Of the Earth and Plan etary Interiores, 129, 173, (2002).
112 G. Q. Gong, A. Gupta, G. Xiao, W. Qian V. P. Dravid, Phys. Rev. B, 56, 5096, (1997). G. Y. Guo, Physica E, 10, 383, (2001). H. C. Swart and G. L. P. Berni ng, Appl. Surf. Sci., 78, 77 (1994). H. H. Wieder, Â“Hall generators and MagnetoresistorsÂ”, Pion Limited, London, (1971). H. H. Wieder, Â“Laboratory Notes on Electri cal and Galvanomagnetic MeasurementsÂ”, Material Science Monograph, 2; Elsevier Scientific Publishing Company, (1979). H. Ohta T. Arioka, Y. Yamamoto, S. Mitsudo, T. Hamamoto, M. Motokawa, Y. Yamaguchi, and E. Kulatov, Physica B, 237-238, 463 (1997). J. P. Shephard, J. W. Koenitzar, R. Aragon, J. Spaxek, J. M. Honig, Phys. Rev. B., 43, 8461, (1991). J. Tang, K. Y. Wang, W. Zhou, J. Appl. Phys., 89, 7690, (2001). K. G. Lisunov, E. K. Arushanov, Ch. Kloc J. Broto, J. Leotin, H. Rokoto, M. Repaud, and E. Bucher, Physica B, 229, 37 (1996). K. L. Saenger, Procs. Adv. Mater., 2, 1, (1993). K. L. Saenger, Procs. Adv. Mater., 3, 63, (1993). K. Ruhrnschopf, D. Borgmann, G. Wedler, Surf. Sci., 374, 269, (1997). K. Rhrnschopf, D. Borgmann, G. Weld er, Thin Solid Films, 280, 171 (1996). L. Digiorgi, M. Hunt, H. R. Ott, Z. Fisk, Physica B, 206, 810, (1995). L. Wang, J. Li, W. Ding, T. Zhou, B. Li u, W. Zhang, J. Wu, Y. Du, J. Magen. Magen Mater. 207, 111, (1999).
113 M. Vasquez-Mansilla, R. D. Zysler, C. Ar ciprete, M. Dimitrijewits, D. RodriguezSierra, C. Saragovi. J. Magen. Magen. Mater. 226-230, 1907, (2001). M. Ziese, H. J. Blythe, J. P hys. Condens. Mater., 12, 13, (2000). N. G. Galkin, D. L. Goroshko, S. Ts. Krivoshchapov, E. S. Zakharova, Appl. Surf. Sci., 175-179, 230 (2001). N. Hasegawa, A. Makino, F. Koike, K. Ikarashi, IEEE Tans. Magens., 32, 4618, (1996). N. Tsuda, K. Nasu, A. Fujimori, K. Sira tori, Electronic c onduction in oxides, Springer, (2000). P. Mukherjee, J. B. Cuff, S. Witahach chi, Appl. Surf. Sci., 5028, 1, (1998). P. Mukherjee, S. Chen, S. Witanacchc hi, Appl. Phys. Lett., 74, 1546, (1999). R. F. Pierret, Semiconductor Device F undementalsÂ”, Addison-Wesley Publishing Company, (1996). R. J. Kennedy, P. A. Stampe, J. Phys. D.: Appl. Phys. 32, 16, (1999). R. S. Popovic, Â“Hall Effect DevicesÂ”, 2nd edition, IOP Publishing Ltd (2004). S. A. Solin, T. Thio, D. R. Hines, J. J. Heremans, Science, 289, 1530, (2000). S. Bae, S. Zurn, W. F. Egelhoff Jr., P. J. Chen, L. Sheppard, J. Torok, J. H. Judy, IEEE Tans. Magens. 37, 1986, (2001). S. F. Alvarado, W. Eib, F. Meier, D. T. Pierce, K. Sattler, H. C. Siegmann, J. P. Remeika, Phys. Rev. Lett., 34, 319, (1975). S. K. Pawar, S. H. Pawar, Mat. Res. Bull., 18, 211, (1983). S. M. Sze, physics of semiconductor devi ces, John Wiley & Sons Inc. New York, (1981).
114 S. P. Sena, R. A. Lindley, H. J. Blyth, Ch. Sauer, M. A. Al-Kofarji, G. A. Gehring, J. Magn. Magn. Mater., 176, 111, (1997). S. Paschen, E. Felder, M. A. Cherinkov, L. Degiorgio, H. Shcwer, H. R. Ott, D. P. Young, J. L. Sarrao, Z. Fisk, Phys. Rev. B, 56, 12916, (1997). S. Tumanski, Â“Thin Film Magnetoresis tive SensorsÂ”, IOP Publishing Ltd (2001). S. Witahachchi, K. Ahmed, P. Sakthivel, P. Mukherjee, Appl. Phys. Lett., 66, 1469, (1995). S. Witanacchchi, P. Mukherjee, J. V ac. Sci. Technol. A 13, 1171, (1995). T. Jalborg, Phys. Rev. B, 59, 15002, (1999). T. Jarlborg, Phys. Rev. B, 51, 11106, (1995). V. I. Kaidanov, V. A. Tselishchev, I. K. Iesalniek, L. D. Dudkin, B. K. Voronov, N. N. Trusova, Sov. Phys. Semicondutors, 2, 382, (1968). V. Jaccarino, G. K. Wertheim, J. H. Wernic k, and L. R. Walker, Sigurds Arajs, Phys. Rev. 160, 476 (1967) V. V. Glushkov, N. E. Sluchanko, S. V. De mishev, M. V. Kondrin, A. A. Pronin, K. M. Petukhov, Y. Brunseraede, V. V. Moshchalkov, A. A. Menovsky, Physica B, 284288, 1179, (2000). V. V. Gridin, G. R. Hearne, J. M. Honig, Phys. Rev. B., 53, 15518, (1996). X. W. Li, A. Gupta, G. Xiao, G. Q. Gong, J. Appl. Phys. 83, 7049, (1998). Y. Gao, Y. J. Kim, S. Thevuthasan, S. A. Chambers, J. Appl. Phys. 81, 3253, (1997). Z. Schlesinger, Z. Fisk, H. T. Zhang, M. B. Maple, J. F. DiTusa, and G. Aeppli, Phys. Rev. Lett., 71, 1748 (1993).
115 Z. Schlesinger, Z. Fisk, H. T. Zhang, M. B. Maple, Physica B, 237-238, 460, (1997). Z. Zhang, S. Sutpathy, Phys. Rev. B., 44, 13319, (1991).
116 Bibliography B. D. Culity, Elements of X-ray Diffraction, Addison-Wesley Publishing Company, Inc. (1978). D. Chescoe, P. J. Goodhew, The Operation of the Transmission Electron Microscope, Oxford University Press (1984). E. M. Slayter, H. S. Slayter, Li ght and Electron Micr oscopy, Cambridge University Press (1992). T. A. Carlson, Photoelectron and Auger Spectroscopy, Plenum Press, New York and London (1975).
About the Author Houssam Abou Mourad received his Bachel orÂ’s degree in phys ics in 1996 at the University of Puerto Rico, Mayagez Campus where he participated in undergraduate research projects in the field of high-energy physics. In 1999 he receiv ed his MasterÂ’s in Science degree in Physics. His masterÂ’s th esis was in the deposition of controlled composition of KTa1-xNbxO (KTN) thin films for ferroel ectric applications under the advice of Dr. Felix Fernandez. In 2000 Houssam Abou Mourad was admitted to the Ph.D. Applied Physics program at the University of South Florida. While in the Ph.D. program he made several presentations and was aw arded the Tharp (2000) and the Duckwall (2001) fellowships.