<%BANNER%>

Atomic Layer Deposition of the Al sub 2 O sub 3-Y sub 2 O sub 3 Pseudo-Binary System

Permanent Link: http://ufdc.ufl.edu/UFE0024884/00001

Material Information

Title: Atomic Layer Deposition of the Al sub 2 O sub 3-Y sub 2 O sub 3 Pseudo-Binary System
Physical Description: 1 online resource (104 p.)
Language: english
Creator: Rowland, Jason
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: ald, alumina, aluminum, atomic, ceria, deposition, garnet, layer, monoclinic, yag, yam, yttria, yttrium
Materials Science and Engineering -- Dissertations, Academic -- UF
Genre: Materials Science and Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: The growth of thin films of selected phases from the pseudo-binary Al2O3-Y2O3 material system was demonstrated using atomic layer deposition (ALD). Specifically ALD growth of Al2O3, Y2O3, Ce2O3, Y2Al4O12 (Yttrium Aluminum Monoclinic - YAM), and Y3Al5O12 (Yttrium Aluminum Garnet - YAG) was accomplished. All films were grown using the same precursors: AlCl3 at 105degreeC and H2O, Y(thd)3 thd = 2,2,6,6-tetramethyl-3,5-heptanedione at 140degreeC and O3, and Ce(acac)3 acac = acetylacetonate at 140degreeC, and O3. The Al2O3 films were grown at substrate temperatures from 295degreeC to 515degreeC. A surface-controlled growth temperature window (ALD window) was found for Al2O3 between 365degreeC to 465degreeC using AlCl3 and H2O. The resultant films grown at all temperatures were amorphous as characterized by X-ray diffraction, and showed a rough surface morphology. The growth rate was determined to be 1 capital a ring/cycle within the ALD window . The thickness of films grown in the ALD window varied linearly with the number of cycles. Films up to 1 um thick were grown (10,000 cycles). The Y2O3 films were grown at substrate temperatures ranging from 200degreeC to 500degreeC. No surface-controlled growth temperature window could be determined using Y(thd)3 and O3. The resultant films were polycrystalline with a cubic structure and a smooth surface morphology. The growth rate was determined to be 3 pm/cycle at 350degreeC. Films up to 30 nm thick were grown (10,000 cycles). Atomic layer deposition of Ce2O3 thin films were also studied because Ce3+ is often used as a luminescent rare earth dopant in YAG and YAM. The Ce2O3 films were grown at substrate temperatures from 200degreeC to 500degreeC. No surface-controlled growth temperature window could be found using Ce(acac)3 and O3. The resultant films had a rough surface morphology. Using X-ray photoelectron spectroscopy (XPS), it was determined that the Ce3+ oxidation state was present in the as-deposited films rather than Ce4+. When included in the growth process for YAG and YAM, Ce2O3 was found to create very rough surfaces. The ternary oxide phases of YAM and YAG were produced by ALD growth of alternating nano-scaled multilayer stacks of Al2O3 and Y2O3 and subsequent calcinations of these thin nano-scale stack structures.. By adjusting the ratio of Al2O3 deposition cycles to Y2O3 deposition cycles, stoichiometric amounts of material were deposited in an alternating fashion. A 300 nm YAG film was deposited using the ALD method at a substrate growth temperature of 350degreeC. X-ray diffraction data showed that these films were amorphous as-deposited, but they were transformed to a polycrystalline cubic garnet structure when calcined at 975degreeC for 12 hrs in air. The surface morphology was uniform and smooth. A 400 nm YAM thin film was also successfully deposited using the ALD method at a growth temperature of 350degreeC. X-ray diffraction of the YAM film showed an amorphous film as-deposited and a polycrystalline monoclinic structure after calcining at 975degreeC for 12 hrs in air. The surface morphology of this YAM film was rough and non-uniform.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Jason Rowland.
Thesis: Thesis (Ph.D.)--University of Florida, 2010.
Local: Adviser: Holloway, Paul H.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0024884:00001

Permanent Link: http://ufdc.ufl.edu/UFE0024884/00001

Material Information

Title: Atomic Layer Deposition of the Al sub 2 O sub 3-Y sub 2 O sub 3 Pseudo-Binary System
Physical Description: 1 online resource (104 p.)
Language: english
Creator: Rowland, Jason
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: ald, alumina, aluminum, atomic, ceria, deposition, garnet, layer, monoclinic, yag, yam, yttria, yttrium
Materials Science and Engineering -- Dissertations, Academic -- UF
Genre: Materials Science and Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: The growth of thin films of selected phases from the pseudo-binary Al2O3-Y2O3 material system was demonstrated using atomic layer deposition (ALD). Specifically ALD growth of Al2O3, Y2O3, Ce2O3, Y2Al4O12 (Yttrium Aluminum Monoclinic - YAM), and Y3Al5O12 (Yttrium Aluminum Garnet - YAG) was accomplished. All films were grown using the same precursors: AlCl3 at 105degreeC and H2O, Y(thd)3 thd = 2,2,6,6-tetramethyl-3,5-heptanedione at 140degreeC and O3, and Ce(acac)3 acac = acetylacetonate at 140degreeC, and O3. The Al2O3 films were grown at substrate temperatures from 295degreeC to 515degreeC. A surface-controlled growth temperature window (ALD window) was found for Al2O3 between 365degreeC to 465degreeC using AlCl3 and H2O. The resultant films grown at all temperatures were amorphous as characterized by X-ray diffraction, and showed a rough surface morphology. The growth rate was determined to be 1 capital a ring/cycle within the ALD window . The thickness of films grown in the ALD window varied linearly with the number of cycles. Films up to 1 um thick were grown (10,000 cycles). The Y2O3 films were grown at substrate temperatures ranging from 200degreeC to 500degreeC. No surface-controlled growth temperature window could be determined using Y(thd)3 and O3. The resultant films were polycrystalline with a cubic structure and a smooth surface morphology. The growth rate was determined to be 3 pm/cycle at 350degreeC. Films up to 30 nm thick were grown (10,000 cycles). Atomic layer deposition of Ce2O3 thin films were also studied because Ce3+ is often used as a luminescent rare earth dopant in YAG and YAM. The Ce2O3 films were grown at substrate temperatures from 200degreeC to 500degreeC. No surface-controlled growth temperature window could be found using Ce(acac)3 and O3. The resultant films had a rough surface morphology. Using X-ray photoelectron spectroscopy (XPS), it was determined that the Ce3+ oxidation state was present in the as-deposited films rather than Ce4+. When included in the growth process for YAG and YAM, Ce2O3 was found to create very rough surfaces. The ternary oxide phases of YAM and YAG were produced by ALD growth of alternating nano-scaled multilayer stacks of Al2O3 and Y2O3 and subsequent calcinations of these thin nano-scale stack structures.. By adjusting the ratio of Al2O3 deposition cycles to Y2O3 deposition cycles, stoichiometric amounts of material were deposited in an alternating fashion. A 300 nm YAG film was deposited using the ALD method at a substrate growth temperature of 350degreeC. X-ray diffraction data showed that these films were amorphous as-deposited, but they were transformed to a polycrystalline cubic garnet structure when calcined at 975degreeC for 12 hrs in air. The surface morphology was uniform and smooth. A 400 nm YAM thin film was also successfully deposited using the ALD method at a growth temperature of 350degreeC. X-ray diffraction of the YAM film showed an amorphous film as-deposited and a polycrystalline monoclinic structure after calcining at 975degreeC for 12 hrs in air. The surface morphology of this YAM film was rough and non-uniform.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Jason Rowland.
Thesis: Thesis (Ph.D.)--University of Florida, 2010.
Local: Adviser: Holloway, Paul H.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0024884:00001


This item has the following downloads:


Full Text

PAGE 1

1 ATOMIC LAYER DEPOSITION OF THE Al2O3Y2O3 PSEUDO -BINARY SYSTEM By JASON CONRAD ROWLAND A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 20 10

PAGE 2

2 20 10 Jason Conrad Rowland

PAGE 3

3 To my loving wife

PAGE 4

4 ACKNOWLEDGMENTS I would like to express my gratitude to my advisor, Dr. Paul Holloway, for his patien t support throughout my thesis project. I am deeply honored to have been a part of Dr. Holloways group. His many achievements, awards, and contributions to the field of materials science and engineering are a true inspiration. In addition, this work would not have been possible without Dr. Mark Davidsons knowledge and expertise in installing, operating, and modifying the ALD reactor. M y appreciation for Marks help cannot be adequately expressed with mere words. I would also like to thank my committee mem bers, Dr. Paul Holloway, Dr. Mark Davidson, Dr. Cammy Abernathy, Dr. David Norton, and Dr. David Tanner, for their participation o n my thesis committee. Thanks to the staff of Microfabritech, especially Dr. Maggie PugaLambers for SIMS data and data reduc tion, and Mr. Chuck Rowland for all the facilities work that was necessary for the ALD reactor becoming fully operational. Thanks to the staff of the Major Analytical Instrumentation Center, Department of Materials Science and Engineering, University of Fl orida, with a special thank you to Eric Lambers for his help with XPS. I would also like to thank Ludie Harmon, program assistant at Microfabritech and secretary to the Holloway research group, for attending to the administrative details, thereby allowing me to devote myself to this project And finally I want to thank my wife, Chris, for all of her patience, love, and support. She is the proof that behind every good man, there is a better woman.

PAGE 5

5 TABLE OF CONTENTS Page ACKNOWLEDGMENTS ...................................................................................................... 4 LIST OF TABLES ................................................................................................................ 7 LIST OF FIGURES .............................................................................................................. 8 LIST OF ABBREVIATIONS .............................................................................................. 10 ABSTRACT ........................................................................................................................ 11 C H APT ER 1 INTRODUCTION ........................................................................................................ 13 2 LITERATURE REVIEW .............................................................................................. 16 2.1 Introduction ........................................................................................................... 16 2.2 Atomic Layer Deposition ...................................................................................... 18 2.2.1 Introduction ................................................................................................. 18 2.2.2 ALD Precursors ........................................................................................... 21 2.2.3 Nucleation and Growth ............................................................................... 23 2.2.3.1 Adsorption ......................................................................................... 23 2.2.3.2 Nucleation ......................................................................................... 28 2.2.3.3 Growth modes ................................................................................... 29 2.3 Atomic Layer Deposition of Bi nary Oxides .......................................................... 33 2.3.1 Thermochemistry ........................................................................................ 33 2.3.2 Al2O3 ............................................................................................................ 34 2.3.3 Y2O3 ............................................................................................................. 36 2.3.4 Ce2O3 .......................................................................................................... 38 2.4 Pseudo-binary Oxides of the Al2O3Y2O3 System ............................................... 40 2.4.1 Introduction ................................................................................................. 40 2.4.2 Y4Al2O9 ........................................................................................................ 44 2.4.3 YAlO3 ........................................................................................................... 44 2.4.4 Y3Al5O12....................................................................................................... 45 2.4.5 Y3Al5O12:Ce ................................................................................................. 45 3 EXPERIMENTAL PROCEDURE ............................................................................... 46 3.1 Precursor Analysis ................................................................................................ 46 3.2 Substrate Preparation .......................................................................................... 46 3.3 Film Deposition ..................................................................................................... 47 3.3.1 Al2O3 Deposition ......................................................................................... 52 3.3.2 Y2O3 Deposition .......................................................................................... 53

PAGE 6

6 3.3.3 Ce2O3 Deposition ........................................................................................ 54 3.3.4 Pseudo -binary Deposition .......................................................................... 55 3.4 Heat Treatment ..................................................................................................... 57 3.5 Characterization .................................................................................................... 58 3.5.1 X -Ray Diffraction (XRD) ............................................................................. 58 3.5.2 X -Ray Photoelectron Spectroscopy (XPS) ................................................ 58 3.5.3 Secondary Ion Mass Spectroscopy (SIMS) ............................................... 58 3.5.4 Thi ckness Measurements ........................................................................... 59 3.5.5 Scanning Electron Microscopy (SEM) ....................................................... 59 4 RESULTS AND DISCUSSION ................................................................................... 61 4.1 Introduction ........................................................................................................... 61 4.2 Al2O3 Deposition ................................................................................................... 61 4.2.1 Al2O3 Precursors ......................................................................................... 61 4.2.2 Al2O3 Growth Parameters ........................................................................... 62 4.2.3 Al2O3 Characterization ................................................................................ 63 4.3 Y2O3 Deposition .................................................................................................... 66 4.3.1 Y2O3 Precursors .......................................................................................... 66 4.3.2 Y2O3 Growth Parameters ............................................................................ 68 4.3.3 Y2O3 Characterization ................................................................................. 69 4.4 Ce2O3 Deposition .................................................................................................. 71 4.4.1 Ce2O3 Precursors ....................................................................................... 71 4.4.2 Ce2O3 Growth Parameters ......................................................................... 72 4.4.3 Ce2O3 Characterization .............................................................................. 72 4.5 Al2O3Y2O3 Pseudo -Bin ary Deposition ................................................................ 75 4.5.1 Al2O3Y2O3 Pseudo-Binary System Growth Parameters ........................... 75 4.5.2 Y3Al5O12 (YAG) Deposition ......................................................................... 76 4.5.2.1 Y3Al5O12 sequence #1....................................................................... 77 4.5.2.2 Y3Al5O12 sequence #2....................................................................... 80 4.5.2.3 Y3Al5O12 sequence #3 ....................................................................... 87 4.5.3 Y4Al2O9 (YAM) Deposition .......................................................................... 89 5 CONLCUSIONS AND FUTURE STUDIES ................................................................ 94 5.1 Conclusions .......................................................................................................... 94 5.2 Future Studies ...................................................................................................... 96 LIST OF REFERENCES ................................................................................................... 99 BIOGRAPHICAL SKETCH .............................................................................................. 104

PAGE 7

7 LIST OF TABLES Table page 2 -1 Thermodynamic properties of precursors and films grown in t his study [49,50]. .................................................................................................................... 33 3 -1 Pulse sequences for the binary oxides deposited in this study. ........................... 50 3 -2 Necessary thickness ratios for Y2O3 to Al2O3 thin film bilayers to achieve stoichiometry for a given phase. ............................................................................ 56 3 -3 ALD sequences for Y -Al -O films. ........................................................................... 56

PAGE 8

8 LIST OF FIGURES Figure page 2 -1 Schematic of an idealized surface reaction with AlCl3 and H2O precursors. ....... 19 2 -2 Schematic of ALD process temperature window for saturation and rate limiting mechanisms at lower and higher temperatures [41]. ............................... 24 2 -3 An ALD sequence illustrating the influence of steric effects on surface roughness. .............................................................................................................. 31 2 -4 Pseudo binary phase diagram of the Al2O3Y2O3 material system reprinted from [15]. ................................................................................................................. 41 3 -1 Schematic of Atomic Layer Deposition vacuum sy stem. ...................................... 51 3 -2 Photograph of the Planar Systems P400 hot wall flow type ALD reactor showing the gas manifold side of the reactor. ....................................................... 52 4 -1 Growth rate of Al2O3 versus substrate temperature. ............................................. 63 4 -2 X -Ray diffraction spectrum from an as deposited 1.0 m thick Al2O3 film deposited at 350C. ................................................................................................ 64 4 -3 SEM micrograph of an as deposited Al2O3 film deposited at 350C. ................... 65 4 -4 SEM micrograph of a fractured cross -section as -deposited Al2O3 film deposited at 350 C. ................................................................................................ 65 4 -5 Differential thermogravimetric mass loss (DTG) of Y(thd)3 as a function of temperature. The broad peak centered at 67 C is due to desorption of methanol used to clean the sample hol der. .......................................................... 67 4 -6 XRD spectrum from a 25 nm thick Y2O3 film as -deposited at 350C. The lines show peaks from JCPDS Card #43-1036 for cubic Y2O3. .................................... 70 4 -7 SEM micrograph of an as deposited Y2O3 film deposited at 350C ..................... 70 4 -8 Plot of mass loss TG of Ce(acac)3 vs. temperature. .......................................... 72 4 -9 XPS spectrum showing that Ce and O were detected on SiO2/Si substrate after 9000 cycles of growth at 350C. ................................................................... 73 4 -10 High resolution XPS spectrum showing the Ce3+ 3d peaks fr om figure 4-9: the intense charge transfer peaks, and the small Ce4+ 3d3/2 peak. The Ce4+ 3d5/2 peak is buried beneath the Ce3+ 3d3/2 peak. ................................................. 74

PAGE 9

9 4 -11 SEM images of a Ce2O3 coated Si/SiO2 surface (left panel 100 nm scale bar) and a bare SiO2/Si surface (right panel 500 nm scale bar). ...................... 75 4 -12 XPS of Y2O3-Al2O3 film grown using YAG sequence #1 (section 4.5.1.1). .......... 79 4 -13 SEM images of YAG films via sequence #1. a.) as deposited and b.) calcined at 975C for 12 hours. ............................................................................................ 80 4 -14 XRD spectrum from an as deposited film grown with YAG sequence #2 (see section 4.5.1). ......................................................................................................... 81 4 -15 XRD spectrum from a film grown with YAG sequence #2 and annealed at 975C for 12 hrs in air. ........................................................................................... 82 4 -16 XPS spectrum from an as -deposited film grown with YAG sequence #2 after a 10 minute sputter with 4 keV Ar+. ....................................................................... 84 4 -17 SEM images of YAG films via sequence #2. a .) as deposited and b.) calcined at 975C for 12 hours. ............................................................................................ 84 4 -18 SEM images of as -deposited (topa, c, e) and calcined (bottom b, d, f) at 975C YAG films with increasing Y:Ce ratios of (a. & b.) 10:1, (c. & d.) 20:1, and (e. & f.) 445:1. .................................................................................................. 86 4 -19 Cross -section SEM image of calcined YAG sequence #2 film. ............................ 86 4 -19 XRD spec trum from a film grown with YAG sequence #3 and annealed at 975C for 12 hrs in air. ........................................................................................... 89 4 -20 XRD spectrum from a film grown with YAM sequence #4 and calcined at 975C for 12 hrs in air. ........................................................................................... 92 4 -21 XPS spectrum from an as -deposited film grown with YAM sequence #4 after a 10 minute sputter with 4 keV Ar+. ....................................................................... 93

PAGE 10

10 LIST OF ABBREVIATION S acac 2,4pentandione, acetylacetonate, C5H7O2 AFM Atomic Force Microscope/Microscopy ALD Atomic Layer Deposition ATO Aluminum Titanium Oxide, Al2O3-TiO2 CVD Chemical Vapor Deposition GGG Gadolinium Gallium Garnet JCPDS Joint Committee on Diffraction Standards LED Light Emittin g Diode MOCVD Metal -Organic Chemical Vapor Deposition MOSFET Metal -Oxide -Semiconductor Field-Effect Transistor PET Positron Emission Tomography PL Photoluminescence SIMS Secondary Ion Mass Spectroscopy SEM Scanning Electron Microscopy TFEL Thin Film Electr oluminescence thd 2,2,6,6-tetramethyl 3,5dionate, C11H20O2 YAG Yttrium Aluminum Garnet, Y3Al5O12 YAM Yttrium Aluminum Monoclinic, Y4Al2O9 YAP Yttrium Aluminum Perovskite, YAlO3 XRD X -ray Diffraction XPS X -ray Photoelectron Spectroscopy

PAGE 11

11 Abstract of Dis sertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy ATOMIC LAYER DEPOSITION OF THE Al2O3Y2O3 PSEUDO -BINARY SYSTEM JASON CONRAD ROWLAND May 2 0 10 Chair: Paul Holloway Major: Materials Science and Engineering The growth of thin films of selected phases from the pseudo-binary Al2O3Y2O3 material system was demonstrated using atomic layer deposition (ALD). Specifically ALD growth of Al2O3, Y2O3, Ce2O3, Y2Al4O12 (Yttrium Aluminum Monoclinic YAM), and Y3Al5O12 (Yttrium Aluminum Garnet YAG) was accomplished All films were grown using the same precursors: AlCl3 at 105C and H2O Y(thd)3 [thd = 2,2,6,6 -tetramethyl 3,5heptanedione] at 140C and O3, and Ce (acac)3 [acac = acetylacetonate] at 140C, and O3. The Al2O3 films were grown at substrate temperatures from 295C to 515C. A surface-controlled growth temperature window (ALD window) was found for Al2O3 between 365C to 465C using AlCl3 and H2O. The resultant films grown at all temperatures were amorphous as characterized by X -ray diffraction, and showed a rough surface morphology. The growth rate was determined to be 1 /cycle within the ALD window. The thickness of films grown in the ALD window varied linearly with the number of cycles. Films up to 1 m thick were grown (10,000 cycles). The Y2O3 films were grown at substrate temperatures ranging from 200C to 500C. No surface-controlled growth temperature window could be determined using Y(thd)3 and O3. The resultant films were polycrystalline with a cubic structure and a

PAGE 12

12 smooth surface morphology. The growth rate was determined to be 3 pm/cycle at 350C. Films up to 30 nm thick were grown (10,000 cycles). Atomic layer deposition of Ce2O3 thin films were also studied because Ce3+ is often used as a luminescent rare earth dopant in YAG and YAM. The Ce2O3 films were grown at substrate temperatures from 200C to 500C. No surface -controlled growth temperature window could be found using Ce(acac)3 and O3. The resultant films had a rough surface morphology. Using X -ray photoelectron spectroscopy (XPS) it was determined that the Ce3+ oxidation state was present in the as -deposited films rather than Ce4+. When included in the growth process for YAG and YAM Ce2O3 was found to create very rough surfaces. The ternary oxide phases of YAM and YAG were produced by ALD growth of alternating nano-scaled multilayer stacks of Al2O3 and Y2O3 and subsequent calcinations of these thin nano -scale stack st ructures .. By adjusting the ratio of Al2O3 deposition cycles to Y2O3 deposition cycles, stoichiometric amounts of material were deposited in an alternating fashion. A 300 nm YAG film was deposited using the ALD method at a substrate growth temperature of 3 50C. X -ray diffraction data showed that these films were amorphous as deposited, but they were transformed to a polycrystalline cubic garnet structure when calcined at 975C for 12 hrs in air. The surface morphology was uniform and smooth. A 400 nm YAM thin film was also successfully deposited using the ALD method at a growth temperature of 350C. X -ray diffraction of the YAM film showed an amorphous film as -deposited and a polycrystalline monoclinic structure after calcining at 975C for 12 hrs in air. Th e surface morphology of this YAM film was rough and nonuniform.

PAGE 13

13 CHAPTER 1 INTRODUCTION Atomic layer deposition (ALD) [ 1] has emerged as a cost effective chemical vapor deposition alternative in the creation of very thin films. The advantage s of ALD compar ed to other deposition methods are its excellent conformity, accurate thickness control at the smallest scales, large area uniformity, sharp interfaces, multilayer processing, and excellent quality films [ 2 -5 ]. The applications for ALD films are numerous. The insulating material aluminum titanium oxide (ATO) is a commercially available thin film deposited via atomic layer deposition Alumina and titania are deposited as alternating layers with thicknesses at the nanoscale. The result is a nanolaminate of the immiscible alumina and titania materials and is used as a dielectric layer in electroluminescent displays [6] Thin film electroluminescent displays were among the first commercial applications for ALD. There is considerable interest in ALD with regards to the semiconductor industry and the creation of very thin, high gate oxides in metal oxide -semiconductor field effect transistors (MOSFETs). As the transistor size continues to shrink the oxide gate material thickness shrinks with it, there by requiring a high[ 7 8 ] ALD is currently being used to deposit very thin, highby companies such as Intel [9 ] Numerous companies such as Applied Materials [10 ], Aixtron [11 ], Beneq [12 ], and Kurt J. Lesker [13 ], manufacture ALD systems for both commercial and research use. The pervasiveness of the semiconductor industry in modern life and the industries pursuit of very thin conformal highcontinue to be a large driving force for the growth i n ALD research. Another area of application for ALD is in luminescent materials such as phosphors and scintillators A scintillators resolution is inversely proportional to the

PAGE 14

14 thickness of the scintillato r [ 14 ] therefore ALD can be used to make thin fi lms of scintillator materials for higher spatial resolution, albeit with a lower sensitivity This research focuses on the atomic layer deposition of the pseudo binary Al2O3Y2O3 material system and two of its three primary phases: Y4Al2O9, and Y3Al5O12 [ 15] In addition, ALD of Ce2O3 was investigated as a possible dopant for ALD deposited Y4Al2O9, and Y3Al5O12. The phases in this system are used as phosphors, scintillators, lasers, dielectrics, and optical waveguides. For example, Y3Al5O12:Ce is a common phosphor material that converts some of the blue light from a GaN light emitting diode (LED ) into yellow light [16 ] that in turn compl e ments the residual blue light to produce a white light The Y3Al5O12:Ce phosphor is typically synthesized from a solid s tate reaction of stoichiometric Y2O3, Al2(OH)3 and Ce2O3 at temperatures on the order of 1400C [16]. In addition to a phosphor, Y3Al5O12:Ce is used as a scintillator for radiation detection such as in Positron Emission Tomography (PET). With ALD, very thi n Y3Al5O12:Ce scintillators can be made for greater resolution. Chemical vapor deposition and physical vapor deposition methods are used for applications where these materials are integrated onto a device, such as a dielectric on a semiconductor. ALD provi des a means for depositing these materials onto devices whereby very thin, uniform, and conformal films are needed. In addition, the deposition temperature for ALD processes are typically less than 700C, approximately half the temperature used in solid st ate reactions. The low temperature process of ALD allows for integrating materials onto other more temperature sensitive materials. This work demonstrates a novel low temperature process for the synthesis of Y4Al2O9 and Y3Al5O12 using atomic layer depositi on. By varying the delivery sequence of

PAGE 15

15 the precursor materials while keeping deposition temperature and pressure constant, the three different solid phases of the pseudobinary Al2O3Y2O3 material system can be produced. These phases have different physic al properties that can be taken advantage of either as a single thin film of one phase or in combinations of thin films using two or three phases. In addition, doping of these films is demonstrated, widening the number of potential applications for this pr ocess and material sy s tem These materials can be formed at the nanoscale which can lead to new developments in optical devices such as quantum confinement of optical states, subwavelength optics, and photonics to name a few. Nanolaminates produced by ALD are shown to result in materials whose properties are different from the constituent materials [ 17,18]. As a material s thickness (or characteristic length) decreases to length scales less than the characteristic length scale of a physical property, novel physical properties emerge. Light emitting q uantum d ots, for example, are materials whose diameters are much less than the wavelength of light emitted. Quantum confinement of energy states in the quantum dot arise as the quantum dots size is reduced to th e nano -scale in all three dimensions. ALD allows for quantum well structures in one dimension that can be further refined to three dimensions by careful patterning. A literature review of the atomic layer deposition process and the atomic layer deposition of Y2O3 and Al2O3 films are presented in Chapter 2 including a review of Y4Al2O9, and Y3Al5O12 structure and properties. In Chapter 3 the experimental and characterization methods used to create the YAlO3, Y4Al2O9, and Y3Al5O12 films are discussed Results and discussion are presented in Chapter 4. C onclusions and f uture studies conclude the dissertation in Chapter 5.

PAGE 16

16 CHAPTER 2 LITERATURE REVIEW 2.1 Introduction The rareearth doped materials of the pseudo-binary Al2O3Y2O3 material system are important in optoelectronic devices such as scintillators (YAlO3:Ce) [ 1 9 ] lasers (Y3Al5O12:Nd) [ 20 ] and phosphors (Y2O3:Eu Y3Al5O12:Ce ) [ 16,20 ] therefore improved fabrication methods have wide -ranging applications Typically this material system is synthesized as b ulk crystals using the Czo ch ralski method [ 2125 ] or as powders using a variety of chemical methods such as sol -gel [ 26 ], precipitation [ 27] rf -magnetron sputtering [ 28 ], and air or vacuum sintering [ 29 ] processes The Czochralski method for the growth o f Y3Al5O12:Ce is conducted at a temperature of ~1700C and produces single crystals about 3 cm in diameter and 10 cm in length [ 30 ]. Czochralski growth of bulk crystals produces bulk stoichiometric materials and in regards to the Al2O3Y2O3 material system can produce bulk crystals of all phases. However, it is not possible to use the Czochralski method to deposit thin films onto a substrate. The sol gel method to synthesize Y4Al2O9:Pr, Y3Al5O12:Eu, YAlO3, has been demonstrated and produces polycrystalline powders after annealing [ 26,31,32] The products of the sol gel method are powders and can be formed into films upon heat treatment. However, the conformity of sol -gel powders to substrates with high aspect ratios is difficult and accurate thickness contr ol down to the nanoscale for thin film formation is an even greater challenge with the sol gel method. Another wet chemistry method, the precipitation method, can produce products such as Y3Al5O12:Ce from the

PAGE 17

17 Al2O3Y2O3 material system The precipitation m ethod produces powders and presents the same limitations for thin film formation as the sol -gel method. Thin film deposition of Y3Al5O12:Ce h as been demonstrated for this material system using rf -magnetron sputtering [ 28 ] known for its advantages of low -t emperature deposition and stoichiometric control, but with the disadvantage of poor surface conformation as it is a line of -sight physical vapor deposition (PVD) process In rf magnetron sputtering, the target used is s toichiometr ic for the desired film, b ut effects such as re -sputtering of the film can lead to films that are non-stoichiometric. Thin films by pulsed laser deposition (PLD) [ 33,34 ] is another PVD process used to grow thin films of Y3Al5O12. PLD possesses many of the same advantages as rf ma gnetron sputtering such as low temperature processing and good stoichiometric control with some of the same disadvantage s such as poor surface conformation and a problem with particulate deposition due to splashing of the ablated target. For example, t he PLD of Y3Al5O12 films produced were nanocrystalline particles embedded in an amorphous Y3Al5O12 matrix [ 33 ] and presented a large degree of splashing onto the resultant thin film. M etal organic chemical vapor deposition (MOCVD) of Y3Al5O12 [ 35 ] has been dem onstrated and possesses many advantages such as good uniformity, crystallinity, and large scale batch capability among others. For example, previous literature of heteroepitaxially grown Y3Al5O12 on single crystal (111 )oriented Gd3Ga5O12 (GGG) resulted in single crystalline Y3Al5O12 films with the (111) orientation [ 35] When the MOCVD technique is modified to operate in the surface-controlled regime, the result is

PAGE 18

18 atomic layer deposition (ALD) also known as atomic layer chemical vapor deposition (ALCVD) Atomic layer deposition (ALD) has received more attention in the semiconductor industry as films and interfaces approach the nanoscale. ALD is well known for its ability to deposit very thin conformal films. The semiconductor industrys interest in ALD fo r thin film gate dielectrics used in MOSFETs has resulted in, improved reactor design, new precursors, and new applications of the technique are being developed. This work expands the state of the art of atomic layer deposition by demonstrating the growth of Y2Al4O12 and Y3Al5O12 by ALD for the first time 2.2 Atomic Layer Deposition 2.2.1 Introduction Atomic Layer Deposition (ALD), also known as Atomic Layer Epitaxy (ALE), was developed by Dr. T. Suntola and coworkers at Instrumentarium Datex in 1974 [ 36 ] and patented in 1977 by Suntola and Antson [ 1 ]. ALD is a chemical vapor deposition process in which film growth occurs in a surface controlled manner, layer by layer, also known as digital growth. Advantages of ALD are excellent conformity, accurate thic kness control at the smallest scales, large area uniformity, sharp interfaces, multilayer processing, and excellent quality films [ 2 5 ] A limitation is the low growth rate (~1/sec) which is significant ly lower when compared to other growth processes (1 1 000nm/sec) However, as demand for ever thinner films increases, this drawback becomes less important Originally, ALD was developed to produce ZnS:Mn and amorphous Al2O3 thin films for alternating current thin film electroluminescent displays ( AC TFEL Ds ) [ 1] Until the mid 1990s, most films grown by ALD were of a binary character such as the

PAGE 19

19 aforementioned ZnS, or oxides such as Al2O3. ALD growth of ternary oxides has also been demonstrated [ 37 -40 ] and continued interest exists for more complex oxide film growth by ALD. Figure 21 Schematic of an idealized surface reaction with AlCl3 and H2O precursors. Figure 21 is a schematic of the atomic layer deposition process that shows Al2O3 growth on a silica substrate. The substrate used in the process show n in figure 21 is a hydrogen passivated silica surface on a silicon wafer. The first precursor, AlCl3 in this case, is introduced to the heated substrate surface (figure 21a). The AlCl3 chemisorbs onto the surface (figure 2-1b) in an exchange reaction. The Clligand is exchanged with a surface hydrogen atom to form a saturated chlorine terminated surface (figure 21b) and HCl by products in the gas phase. The reaction chamber is then purged using an inert gas like N2 (figure 21c). The second precursor, water, is then introduced to the chlorine terminated surface (figure 2 -1d). The water is chemisorbed onto the surface

PAGE 20

20 (figure 21e) in a second exchange reaction. The OH molecule is exchanged with a surface chlorine atom to form a saturated hydroxide term inated surface and HCl by products in the gas phase. The reaction chamber is again purged with N2 (figure 2-1e) completing the growth of one molecular layer ( 0.2 nm) in times of 1 -10 sec. The surface now resembles the original hydroxide passivated surfac e and the cycle can be repeated. With multiple cycles, the reactions result in a film with a thickness determined by the number of cycle s The precursors can also be switched in -situ to create new films. Atomic l ayer d eposition growth proceeds layer -by lay er in which each layer is a monatomic or monomolecular layer. In order to achieve monatomic or monomolecular layers, the dosing of a precursor such as AlCl3, needs to be uniformly delivered onto the substrate surface to achieve saturation of the substrate surface reaction sites Saturation occurs when all available surface bonding sites are filled with new chemical species originating from the reaction between a precursor and a surface bonding site. T he resulting surface does not create new bonding sites f or the precursor being used, instead, bonding sites for the next precursor are formed. After saturation the surface is then either exposed to a different precursor (e.g. water) or treated to re -create new bonding sites [ 4 1 ]. D efining a monatomic layer via the ALD process is difficult. In general terms, Suntola defines a monatomic layer as, a surface configuration formed by adatoms directly bonded to the underlying surface and a full monolayer as, a layer of atoms identical to layers of corresponding atoms inside the lattice in the direction parallel to the surface studied for a crystalline material [ 4 1 ] Often, multiple ALD cycles are necessary

PAGE 21

21 in order to produce a single full monolayer. For example, it has been previous reported that ZnS has a prefe rred growth rate of 1/3 monolayer (ML) per cycle when grown as a polycrystalline film using ZnCl2 and H2S as precursors [ 4 1 ]. This 1/3 ML per cycle rate for ZnS is explained as a steric hindrance from the two chlorine atoms on the surface after exposure to ZnCl2 [ 4 1 ] effectively blocking ZnCl2 from bonding sites on the surface By comparison, CdS shows a 1 monolayer/cycle growth rate when using CdCl2 and H2S as precursors [ 4 1 ] The explanation is that Cd has a larger ionic radius than Zn and CdS has a larg er lattice constant than ZnS such that the chlorine ligands do not block (no steric hindrance) other surface sites as they are further apart than the ligand length [ 4 1 ]. 2.2.2 ALD Precursors The correct choice of precursor materials is important for ALD. T he precursors need to be suitably volatile in order for sufficient vapor to be introduced into the reaction chamber [2 ]. They need to have about 0.1 torr equilibrium vapor pressure at a temperature where they do not decompose. However, the vapor pressure o f the material doesnt have to be very high. Solid precursors such as metal halides need only to provide enough flux for surface saturation. The precursors cannot react with an identical precursor and must be stable enough so as to not decompose on the sur face or in the gas phase so as to achieve self -terminating surface reactions. The precursors can be solids, liquids or gases as long as they meet these requirements [2].The precursors must chemisorb on the surface in question or react aggressively with sur face groups in order to reach surface saturation quickly to ensure a reasonable growth rate [2] at lower ALD temperatures.

PAGE 22

22 Metal halides are the most widely used precursors, especially chlorides [5]. Even though most metal halides are solids, their vapor pressures can be sufficient when heated for surface saturation when dosed into the reaction chamber. Homogeneity of the dose or a constant flux rate is typically unnecessary [2]. Metal alkyl compounds behave very well in making oxide and sulfide films but do not behave well in making other types of films [2]. Metal alkoxides have been used successfully in several oxide processes [2]. For electropositive metals, such as rare earths, it is challenging to find suitable precursors that meet the necessary requi rements of volatility and thermal stability for ALD from commercial sources as these are relatively new to ALD. Most precursors for rare earth metals are expensive and available in limited quantities diketonates are the most studied and show good result s for rare earth sulfides [2]. For reactions of -diketonates to form oxides ozone must be used as an oxygen source rather than water because -diketonate anions are very weak br nsted acids [ 42 ] and a strong oxidizer such as ozone is therefore needed to form the rareearth oxide. H2S, H2O are the most common sulfide and oxide precursors. For reactions with halides, such as AlCl3, H2O is an effective oxygen source. Surface terminating hydroxyl groups from the surface metal halide and water reaction in ALD growth are an important intermediate surface chemical species The hydroxyl groups are formed after the water exposure and act as bonding sites for the subsequent metal halide precursor [ 43,44 ] A high density of surface hydroxyl groups can increase growt h rates when compared to a low density hydroxide surface [ 44 ]. Precursor size plays a role in the size of the surface species formed at each step in an ALD cycle. For example, in growth of Ni films, Ni(acac)2 (acac = acetylacetonate

PAGE 23

23 or pentane 2,4-dionate) precursors have a projected surface area of 0.47 nm2 whereas Ni(thd)2 (thd = 2,2,6,6-tetramethylheptane 3,5-dionate) has a projected surface area of 0.87 nm2 which results in a calculated saturation density of 2.1 Ni/nm2 and 1.1 Ni/nm2, respectively [ 4 1 ]. Because of the large footprint of the Ni precursor (acac or thd), many of the surface bonding sites are covered (sterically hindered) at saturation during a single pulse. The acac or thd ligands are removed by H2O on the following ALD step to form H(acac) or H(thd), respectively, which then desorb from the surface. The new surface then has both Ni -O -H and R -O -H sites available for the next Ni(acac/thd) pulse (where R is the original substrate surface). The next few reaction cycles of Ni precursor are wit h either the Ni -O -H or R -O -H sites until after a few cycles only Ni -O -H sites are available as the substrate surface (R) is covered [ 4 1 ]. 2. 2.3 Nucleation and Growth Atomic layer deposition has many similarities with other chemical vapor deposition (CVD) t echniques. Nucleation and growth in ALD can use the same theoretical treatments as CVD techniques The next few sections will briefly cover adsorption, nucleation, and growth of atomic layer deposition. 2.2.3.1 Adsorption The ALD process window for a sa turated surface (fractional or single monolayer ) as a function of temperature is shown schematically in Figure 2 2 [41] Figure 22 is an arbitrary plot of monolayer saturation (adsorption) versus temperature. The saturation (non-shaded) region shown in fi gure 22 represents a single monolayer or fractional monolayer. A fractional monolayer is simply a monolayer where full saturation has occurred and available bonding sites remain due to steric effects of the adsorbed precursor i.e. a monatomic layer as de scribed in section 2.21 ALD requires a sufficient

PAGE 24

24 precursor flux for monolayer saturation unlike MOCVD which requires careful stoichiometric control of continuous precursor flux in the gas phase Insufficient precursor flux in ALD results in a partially s aturated surfaces. As precursors are alternatively introduced to the surface, growth proceeds only on those areas where the previous precursor saturated the surface and formed a selectively reactive chemical species (bonding site) for the next precursor I nsufficient flux effectively lowers the growth rate of the film as the reaction is starved of precursors. Figure 22 Schematic of ALD process temperature window for saturation and rate limiting mechanisms at lower and higher temperatures [ 4 1 ]. Surface adsorption of a gaseous precursor at a constant temperature is described by the Langmuir Isotherm (equation 21) where is the fractional coverage of the ALD Process 'Window' 200 250 300 350 400 450 500 Temperature Monolayer Saturation Temperature Window for Saturation Condensation Insufficient Activation Energy Desorption of Surface Decomposition

PAGE 25

25 precursor on the surface, P is the partial pressure of the precursor, and K is the equilibrium constant [ 45 ] = 1 + (2 -1) The equilibrium constant K, is given by equation 2 -2, where kads and kdes are the ads orption rate constant and desorption rate constant respectively. The rate constants have an Arrhenius relationship k ~ exp (E/kbT) where Kb is Boltzmanns constant, T is temperature and E is the energy of adsorption or desorption as the case may be. = (2 -2) The Langmuir Isotherm describes surface coverage, or surface saturation. One can relate to the monolayer saturation depicted in figure 2-2 (the y axis). Within the saturation window (figure 22) once surface saturation has occurred no other saturation can take place until a new reactive precursor is introduced. This is the characteristic self -limiting mechanism characteristic of ALD. The shaded regions show deviations from this window. If the temperature is too low, condens ation of the precursor on the surface is a possible mechanism for deviating from monolayer saturation. From the Langmuir Isotherm one can see that if the partial pressure of a precursor is very large, approaches one (full saturation). If the partial pressure of a precursor approaches zero then saturation also approaches zer o However, if the precursor gas has a partial pressure greater than its vapor pressure, supers atura tion can occur resulting in co ndensation. In this case the precursor undergoes physisor ption on the surface and lacks enough energy to overcome both the activation energy barrier to chemisorb and the energy barrier for desorption from the surface Surface saturation increases with

PAGE 26

26 decr easing temperature when condensation occurs This growth regime is not strictly limited to surface saturation and therefore monolayer growth is not achieved [ 4 1 ]. I nsufficient energy for a surface reaction to overcome the activation energy of a reaction i s another possible mechanism for deviating from monolayer surface saturation at low temperatures In this case, t he surface reaction of the precursor with a bonding site lacks sufficient energy to overcome the activation for chemisorption The rate of adso rption, rads, of a precursor to a surface has an Arrhenius relationship given by equation 2-3. (2 -3) Surface saturation decreases with decreasing temperature in this regime as the precursor lacks sufficient activation energy, Ea, for chemisorption For temperatures above the ALD window, two other mechanisms are p ossible : decomposition and desorption. D ecomposition of the precursor into nonvolatile material leads to an excess of available surface bonding sites for the precursor as the decomposed precursor material collects on the substrate [ 4 1 ] Monolayer saturati on may occur as desired but in addition, the non volatile decomposed products on the surface create new bonding site s for additional precursor adsorption Surface saturation increases with increasing temperature as new bonding sites are being formed and an increase in saturation is observed (figure 22). New s urface species formed from chemisorption of the precursor onto the surface may also undergo desorption from the surface at higher temperatures When t he rate of desorption of the new ly formed surface species begins to exceed the rate of adsorption

PAGE 27

27 of the precursor at higher temperatures a decrease in surface saturation is observed (figure 22) [ 4 1 ]. Within the ALD window, saturation occurs such that all available surface bonding sites are filled with new surface species resulting from the reaction of the precursor and the available bonding site s [ 4 1 ] In the window, neither an increase of bonding sites from decomposition or condensation occur, nor does insufficient energy for chemisorption or excess en ergy for desorption occur as a function of temperature. In other words, the available surface bonding sites remain constant as a function of temperature. The temperature range of the saturation window var ies for different precursor and surface combinations For a multiple film process such as an alternating Al2O3Y2O3 process the saturation window s of each film should overlap in order to make the desired film. If the windows do not overlap, one can either change the temperature between reaction processes or use extra energy to improve the saturation of one or more reaction sequences [ 4 1 ]. The density of the bound surface precursor molecules in a surface reaction is controlled by the density of available bonding sites and the size of the surface species for med at a particular reaction step in the ALD cycle [ 4 1 ]. Large ligands from the chemisorbed precursor block bonding sites on the surface from reacting with the precursor. The filling density of the surface species formed is dependent on the available bondi ng sites on the surface and on the size of the surface species formed [ 4 1 ]. Bonding sites can also be blocked or created by reconstruction or unwanted chemical groups, such as chlorine in figure 2-1f.

PAGE 28

28 2.2.3.2 Nucleation Alternating cycles of reactive precursors in ALD leads to the desired film formation. In traditional CVD methods, a gas phase reaction of the precursors occurs and the resultant molecular species adsorbs on the surface. These molecules then coalesce (nucleate) and grow. In ALD, the precursor s react on the surface rather than the gas phase. The resultant molecular species is on the surface just as in traditional CVD. As such, treatment of nucleation and growth is similar. Film growth begins with nucleation. The capillarity theory of heterogeneous nucleation provides a qualitative model of film nucleation [ 45 ]. This theory assumes that atoms or molecules impinge on the substrate surface and coalesce into a cap-shaped nuclei (film) of some mean dimension, d, on the substrate. The free energy chan ge of this nucleus formation is given by equation 2-4 [ 45]. = 33 + 12+ 22 22 (2 -4) The first term is the volume of the nuclei and the free energy change per unit volume and drives the coalescence of the nuclei. In t he second term, fv is the surface energy of the nuclei (film) surface at the nuclei vapor interface. In the third term, fs is the surface energy of the nuclei surface at the nuclei -substrate interface. And in the last term, sv is the surface energy of the substrate at the substrate vapor interface. This last term has the same area as the third term ( a2r2). The last term is negative since the surface energy of this area of substratevapor interface has been lost All the energy terms in equation 2 -4 have units of J/m2. A simple d efinition of s urface energy is related to the change in work on a material when its surface area changes (equation 25). A surface can therefore alter its energy

PAGE 29

29 by engaging in mechanical work to change its surface area. In solids, this involves surface st resses and therefore, surface energy is sometimes referred to as surface tensions. T he surface energies in equation 24 therefore, can be equat ed to interfacial surface tensions between the film and the substrate. These surface tensions are in mechanical equilibrium parallel to the substrate surface and are described by Youngs equation (equation 26 ), where is the wetting angle. = (2 -5) = + (2 -6 ) In this work, we have already looked at the wetting of the precursors (section 2.2.3.1) in ALD. Monolayer saturation occurs in ALD when = 0 (complete wetting of the s urface). It is t he reaction of surface saturated precursors with the underlying surface that produces a film in ALD The resultant film wets the substrate no more than the precursor with the least surface saturation. The relative size of the surface tensions in equation 26, as well as the wetting angle, determine what type of growth occurs after nucleation. 2.2.3.3 Growth modes There are three well known growth modes: Island (Volmer -Weber), layer (Frank Van der Merwe), and a mixture of these two known as S transki Krastanov [ 45 ]. Island growth occurs when the deposited material preferentially attaches to growth material already present on the surface. Layer growth occurs when deposited material preferentially seeks the lowest (normal to the surface) unfilled bonding sites on the surface. The growth is twodimensional and the material grows monolayer by monolayer. Stranski -Krastanov growth occurs when the substrate is first covered by a

PAGE 30

30 few monolayers of growth in a layer growth mode and then proceeds to the i sland growth mode. Finally, a fourth growth mode, random growth, has also been proposed for ALD [ 46 ]. Random growth occurs when the deposited material settles with equal probability to all surface bonding sites in an ALD reaction cycle [ 46 ]. For island growth > 0 and therefore equation 26 becomes equation 27 If we neglect fs, equation 26 suggests that island growth occurs when the surface energy of the film exceeds that of the substrate. In this growth mode, increased lattice mismatch between the film an d substrate increases the influence of island growth. In ALD, the adsorbed precursors preferentially settle on top of material already deposited on the surface. < + (2 -7 ) For layer growth, the deposit wets the substrate ( 0 ) and equation 26 becomes equation 2 8 This suggests that layer growth occurs when the surface energy of the substrate exceeds that of the film. In homoepitaxy, the film is the same material as the substrate and therefore vanishes. In ALD, the ads orbed precursors preferentially settle in the lowest unfilled layer [ 46]. + (2 -8 ) Lastly, f or Stranski -Krastanov growth, initially the film wets the substrate and layer growth occurs. The transition from layer to island growth, occu rring after approximately 5 -6 monolayers is not completely understood [ 45] It is likely that after a few nanometers of layer growth, a lattice mismatch between the film and the substrate accumulates strain energy in the film surface and may trigger the onset of island growth.

PAGE 31

31 In ALD, the adsorbed precursors initially settle in the lowest unfilled layer, but soon preferentially settle on material already deposited on the surface. Another growth mode has been proposed for ALD referred to as random dep osition [ 46 ]. This growth mode bears some similarity with the so called rain model in continuous deposition processes like MOCVD [ 46 ]. In random deposition, the adsorbed precursors do not settle preferentially but instead settle randomly. This growth mo de shows a dependence on the growth rate ( /cycle) in ALD such that large growth rates result in a smoother film whereas a much smaller growth rate results in rougher films [ 46]. Figure 23 An ALD sequence illustrating the influence of steric effects on surface roughness. Steric effects influence nucleation and growth in ALD as briefly discussed in section 2.2.3.1. An adsorbed precursor with large ligands can effectively block other

PAGE 32

32 precursors from adsorption sites. If such a situation occurs, t he surf ace reaction between an absorbed precursor and the next precursor in the cycle can produce an uneven surface. Figure 2-3 shows a simple cartoon of how this might occur. Frame A shows a functionalized substrate and large ligand precursors (open box) about to absorb. Frame B shows those large ligand precursors adsorbed (a fractional monolayer) on the surface and the second precursor (closed box) about to absorb. Frame C shows the absorbed (reacted) second precursors and a new pulse of large ligand precursor about to absorb. Note that in frame C the reaction of the two precursors produces a functionalized substrate similar to frame A, but with slight surface roughness. In Frame D some of the large precursor has absorbed in the low spaces before tho se sites could be blocked. Upon reacting with the second precursor, frame E shows further roughening of the surface. Frame F shows multiple areas of steric hindrance occurring, with some of the original surface from frame A still unsaturated. In fram e G we see the surface after 3 full cycles have occurred. Even if the surface tensions are such that layer growth is expected, large ligand precursors can result in some surface roughening. Typically, growth of a binary film such as Al2O3 or Y2O3, consi sts of multiple ALD cycles such that a monolayer of the binary material is formed. For ternary films using ALD, the common method is to grow alternating monolayers of the binary constituents and then post -calcine the multi layered film to achieve the final structure [2, 3 8 47 and 48]. For instance, it was shown that to grow a film of ternary LaGaO3, alternating monolayers of binary La2O3 and Ga2O3 were grown and calcined to form LaGaO3 [ 3 8 ]. For more complex stoichiometries, the monolayers were deposited such that the overall

PAGE 33

33 stoichiometric ratio was maintained during growth and post growth calcining for inter diffusion. For growth of LaGaO3 films, Nieminen [ 3 8 ] has shown that the pulsing sequence ] ) ( [ 2 ] ) ( [ 53 3 3 3O acac Ga O thd La produces stoichiometric LaGaO3 films at 325C 425C. T he pulsing ratio of La:Ga was 5:2 even though the stoichiometric ratio of the resultant film is 1:1. 2.3 Atomic Layer Deposition of Binary Oxides 2.3.1 Thermochemistry Table 2 1. Thermodynamic properties of precursors and films grown in this study [49,50]. Material f (kJ/mol) S (J/mol K) C p (J/mol K) AlCl 3(s) 704.2 109.3 91.1 AlCl 3(g) 583.2 314.4 Al 2 O 3(s) 1675.7 50.9 79.0 HCl (g) 92.3 186.9 29.1 H 2 O (g) 241.8 188.8 33.6 O 2(g) 0 205.2 29.4 O 3(g) 142.7 238.9 39.2 (thd) (g) 3 46.2 YCl 3(s) 1000 YCl 3( g ) 750.2 75.0 Y(thd) 3(g) 2269.1 Y 2 O 3(s) 1905.3 99.1 102.5 CO 2(g) 393.5 213.8 37.1 Atomic layer deposition of binary oxides is a surface controlled chemical reaction. For ALD, chemical precursors that are highly reactive with each other are desired as discussed in section 2.2. The chemistries of t he highly reactive precursor s used in ALD are exothermic with large enthalpies of reaction. Table 21 shows thermodynamic data for the precursors used in this study and thei r products. In table 2-1, thd stands for 2,2,6,6-tetramethyl 3,5heptanedionate and the enthalpy values for both thd and Y(thd)3

PAGE 34

34 come from Santos et. al. [ 49 ]. The other values are from the Handbook of Chemistry and Physics [ 50 ]. From these values, enthalpies of reaction are calculated for the ALD reactions in this work In table 2-1, Hf is the standard molar enthalpy of formation at 298.15K, S is the standard molar entropy, and Cp is the molar heat capacity at constant pressure at 298.15K. 2.3. 2 Al2O3 The first known process for ALD of Al2O3 us ed AlCl3 and H2O at 250C and was devel oped by Suntola, Pakkala, and Lindfors [ 51]. D eposition temperatures from 100C to 660C have been explored [ 39,51 ]. The deposition of ALD alumina films has primarily focused on using the metal sources of trimethylaluminum (TMA) and aluminum trichloride (A lCl3) [ 3 9 51 and 52 ]. Water is the primary source of oxygen used for alumina growth. In the present study for the ALD of Al2O3 thin films AlCl3 and H2O precursors are used and t he overall chemical reaction is shown in equation 2-8. 2 3 ( )+ 3 2( ) 23 ( )+ 6 ( ) (2 -8) Equation 28 shows an ion exchange reaction where Al is exchanged for H The enthalpy of reaction for the above chemical equation can be determined using Hesss Law (equation 2 9) with the standard molar enthalpies of formation given in table 21. The calculation is given in equation 2-10 (the units have been dropped for clarity) and results in = 337 7 = (2 -9) = ( 6 ) ( 92 3 ) + ( 1 ) ( 1675 7 ) ( 2 ) ( 583 2 ) ( 3 ) ( 241 8 ) (2 -10) The large negative value for H rxn means this reaction is very exothermic. Likewise, using both the standard molar entropies from table 21 and Hesss Law

PAGE 35

35 (equation 2 -9) the standard entropy of reaction for the above chemical reaction is = 22 9 The calculation is given by equation 211 (the units have been dropped for clarity). = ( 6 ) ( 186 9 ) + ( 1 ) ( 50 9 ) ( 3 ) ( 188 8 ) ( 2 ) ( 314 4 ) (2 -11) T herefore, the standard-state Gibbs energy of reaction at constant temperature and pressure is given by equation 212 and results in = 330 9 = (2 -12) A large negative G rxn for this chemical reaction means that the reaction is very favorable to proceed spontaneously when at standard temperatures and pressures and occurs spontaneously. Raising the temperature of this reaction increases the G rxn (less favorable) but at typical ALD temperature s the reaction remains favored with a large negative G rxn. To grow Al2O3 by an ALD reaction between AlCl3 and H2O, the incoming AlCl3 vapor is introduced to the reaction chamber, adsorbs on the substrate surface, and reacts with surface hydroxyls to produce hydrochloric (HCl) gas as a byproduct in an exchange reaction leaving surface chlorides. Once all of the hydroxyl sites have reacted, no more AlCl3 can be adsorbed on the surface. As discussed in section 2.2. 3.1, t his is known as surface saturation and is what makes the ALD process to be self limiting. The remaining AlCl3 and the HCl gas product are then removed from the reaction chamber by the nitrogen pulse Next, H2O is introduced in to the reaction chamber, absorbs on the substrate surface, and react s with surface chlorides to produce HCl gas as a byproduct in an exchange reaction that leav es surface hydroxyl groups. The reactor is again purged with nitrogen. This constitutes one cycle and the

PAGE 36

36 process is repeated several thousand times to grow Al2O3 f ilms 1001000 nm thick The temperature range for the AlCl3 and H2O ALD process is wide, 100C -600C with a growth rate of 0.5/cycle to 0.7/cycle [ 39 53 ]. The AlCl3 source temperature has been reported at 110C [ 53] The formation of Al2O3 from AlCl3 an d H2O is not as simple as indicated by the above idealized reaction sequence. Surface reactions can proceed through a variety of different pathways. In the above idealized sequence, an adsorbed AlCl3 may result in a surface terminated with a single bonded R* -AlCl2, a double bonded R* =AlCl, or triple bonded R* R*is the substrate surface bonding site ). While all of these surface bonding configurations can lead to film growth, the growth rates as well as the density of the resultant films can be different. The reverse of equation 2-8 is also possible. T he gaseous HCl byproduct formed during the AlCl3 pulse can etch the Al2O3 surface being formed (self -etching) The HCl and Al2O3 reaction produces gaseous water and AlCl3. The gaseous water reacts with the AlCl3 (either the by -product or the precursor flux ) in the gas phase to form Al2O3: ie CVD rather than ALD. This reverse reaction forms particulate that deposits on the substrate as well pits where the etched material was removed leading to increased surface roughness in the final film. 2.3. 3 Y2O3 Atom ic layer deposition of Y2O3 using Y(thd)3 [thd = 2,2,6,6-tetramethyl 3,5 heptanedione] and O3 above 425C was shown by Mls, Niinist, and Utrianen [ 54 ]. The deposition of Y2O3 at temperatures of 200C -425C was then reported by Putkonen et. al. [ 55]. Put konen also reported an ALD window of 250C 350C for a Y(thd)3 and

PAGE 37

37 O3 process with a growth rate of 0.23 (cycle)1 [ 55 ]. However, Mls et. al. [ 54 did not show a saturation window and instead report ed a linear increase in growth rate as a function of t emperature. The films were crystalline and showed strong X -ray diffraction (XRD) (400) and (440) peaks [ 55 ] consistent with the cubic phase of Y2O3. The overall chemical reaction is shown in equation 213 [ 56] 2 ( )3 ( )+ 60 3 ( ) 23 ( )+ 57 2( )+ 66 2 ( ) (2 -13) This is a combustion reaction with a very large enthalpy of reaction. Using the standard molar enthalpies of formation from table 21, the enthalpy of reaction for the above chemical equation can be determined using Hesss Law in the same manner as for Al2O3 (section 2.3.2) and results in = 22308 8 The calculation is shown in equation 2-14 (the units have been dropped for clarity). = ( 66 ) ( 393 5 ) + ( 57 ) ( 241 8 ) + ( 1 ) ( 1905 3 ) ( 2 ) ( 226 9 1 ) ( 60 ) ( 142 7 ) (2 -14) This chemical reaction assumes a complete reaction, but previously reported experimental results are not consist e nt with this reactio n and a ligand exchange model is proposed [ 56,57]. This model assumes that only partial combustion occurs during the ozone delivery step. The H2O from partial combustion during the O3 pulse form s surface hydroxyl groups and oxygen bridges can then presumably form from dehydroxylation of these hydroxyl groups. In the following Y(thd)3 delivery pulse a li gand exchange reaction then occurs between Y(thd)3 precursors and the surface hydroxyl groups which remain on the surface during the previous O3 delivery pulse. Partial combustion may result in the formation of CO and hydrocarbons in addition to the H2O an d CO2 products. These may adsorb on the surface as well as provide possible ligand exchange bonding sites, further complicating the reaction mechanism.

PAGE 38

38 The deposition by ALD of yttria films has primarily focused on using the diketonate precursors such as the aforementioned Y(thd)3 as well as Y(thd)3(bipyridyl), and Y(thd)3(1,10phenanthroline) with ozone [ 54,55 ]. The sublimation of Y(thd)3 has been shown to be stable above 130 C [ 54]. The growth of Y2O3 using these precursors show only small differences i n the growth rate s, possibly due to steric hindrance [ 55]. In the present study, ALD was used to grow yttria thin films using Y(thd)3 and O3 as precursors In the idealized Y2O3 ALD reaction between Y(thd)3 and O3, the incoming Y(thd)3 vapor is introduced to the reaction chamber, adsorbs on the substrate surface, and reacts with surface sites to produce a Y(thd)x surface (where x=0,1, or 2) Once surface saturation is complete the nonadsorbed Y(thd)3 and the reaction by products (H2O, CO2, and any thd fr agments) are removed from the reaction chamber during the purge sequence using inert nitrogen gas. Next, O3 is introduced to the reaction chamber, absorbs on the substrate surface, and reacts with the surface Y(thd)x to produce a Y2O3 surface with some hyd roxyl surface groups and combustion byproduct s The reactor is again purged with nitrogen. This constitutes one complete cycle which is repeated several thousand times to grow films 1001000 nm thick 2.3. 4 Ce2O3 Atomic layer deposition of Ce O4 using Ce (t hd)4 [thd = 2,2,6,6tetramethyl -3,5 heptanedione] and O3 with a temperature window of 175250C was shown by Jani Paiva saari, Matti Putkonen and Lauri Niinisto [ 58 ]. The growth rate of Ce O4 was found to be 0.32 /cycle [58] The films were poly crystalline with no preferred orientation. The overall chemical reaction is shown in equation 2-1 5 [ 56 ].

PAGE 39

39 2 ( )4 ( )+ 46 3 ( ) 2 ( )+ 76 2( )+ 88 2 ( ) (2 -1 5 ) This chemical reaction assumes a complete reaction, however, previously reported experimental results were not consistent with this reaction and a ligand exchange reaction with some thermal decomposition was observed [ 58]. A ligand exchange reaction occurs between Ce (thd)4 precursors and the surface hydroxyl groups which remain on the surface during the previous O3 delivery pulse In the idealized Ce O4 ALD reaction bet ween Ce (thd)4 and O3, the incoming Ce (thd)4 vapor is introduced to the reaction chamber, adsorbs on the substrate surface, and reacts with hydroxyl surface sites to produce a Ce (thd)x surface (where x=0,1, 2, or 3 ) and H (thd) as a byproduct. Once surface saturation is complete, the non adsorbed Ce (thd)4, H (thd) and any thd fragments are removed from the reaction chamber during the purge sequence using inert nitrogen gas. Next, O3 is introduced to the reaction chamber, absorbs on the substrate surface, and reacts with the surface Ce (thd)x to produce a Ce O2 surface with some hydroxyl surface groups and combustion byproduct s The reactor is again purged with nitrogen. This constitutes one complete cycle which is repeated several thousand times to grow films Atomic layer deposition of Ce2O3 has not been previously demonstrated. In this dissertation, ALD of Ce2O3 deposition is studied as a potential dopant for Y4Al2O9 and Y3Al5O12 using Ce ( acac )3 [acac = acetylacetonate or 2,4 pentanedionate] and O3 as precurs ors. A possible combustion reaction is shown in equation 216. 2 ( )3 ( )+ 25 3 ( ) 23 ( )+ 24 2( )+ 30 2 ( ) (2 -1 6 ) This proposed chemical reaction assumes a complete combustion. It is suspected that much like the Ce( thd )4 process above and that for Y(thd)3 (section 2.3.2) a ligand

PAGE 40

40 exchange reaction with some combustion is expected. In this manner, a ligand exchange reaction would occur between Ce(thd)3 precursors and the surface hydroxyl groups which remain on the surface during the previous O3 delivery pulse. In the ideali zed Ce2O3 ALD reaction between Ce (thd)3 and O3, the incoming Ce (thd)3 vapor is introduced to the reaction chamber, adsorbs on the substrate surface, and reacts with hydroxyl surface sites to produce a Ce (thd)x surface (where x=0,1, or 2 ) and H (thd) ligands as a byproduct. Once surface saturation is complete, the non adsorbed Ce (thd)4, H (thd) and any thd fragments are removed from the reaction chamber during the purge sequence using inert nitrogen gas. Next, O3 is introduced to the reaction chamber, absorbs on the substrate surface, and reacts with the surface Ce (thd)x to produce a Ce2O3 surface with some hydroxyl surface groups and combustion byproduct s The reactor is again purged with nitrogen. This constitutes one complete cycle which is repeated several thousand times to grow films 2.4 Pseudo -binary Oxides of the Al2O3-Y2O3 System 2.4.1 Introduction The pseudobinary Al2O3Y2O3 material system has several well known solid phases. In order from least aluminum content to the greatest, these are: Y2O3, Y4Al2O9, YAlO3, Y3Al5O12, and Al2O3. A pseudobinary phase diagram is shown in figure 24 The solid phases continue as presented in figure 24 at temperatures below 1600 C Both Al2O3 and Y2O3 are discussed in section 2.3. T he other three phases of this material system are discussed in this section. As can be seen from figure 24, the three solid phases of the Al2O3Y2O3 material system have discrete phase boundaries between them that do not vary in concentration at temperatures below ~ 1 8 00 C [10] Mixture s of finely ground Al2O3 and Y2O3 particles

PAGE 41

41 at high temperatures crystallize into one of the phases The phases formed depend solely on the concentration of the mixture. In these solid-state reactions, repeated grinding of the material is necessary to prov ide intimate contact between Al2O3 and Y2O3 to facilitate their inter -diffusion into the new phases. Even with stoichiometric amounts of Al2O3 and Y2O3, it is difficult to get complete inter diffusion and remnant Al2O3 and Y2O3 phases are often present. Figure 24 Pseudo -binary phase diagram of the Al2O3Y2O3 material system r eprinted from [ 15 ]. At thermodynamic equilibrium, t here are four solid -state regions present in the phase diagram in figure 24. In order of increasing Al2O3 content these are: a mixed

PAGE 42

42 phase of Y2O3 and Y4Al2O9, a mixed phase of Y4Al2O9 and YAlO3, a mixed phase of YAlO3 and Y3Al5O12, and a mixed phase of Y3Al5O12 and Al2O3 Only by the precise mixing of Al2O3 and Y2O3 can pure phases Y4Al2O9, YAlO3, Y3Al5O12 be synthesized. The relative amounts of each present can be determined by using the lever rule [ 59] Tie lines in any given region in this phase diagram always correspond to the same concentrations at temperatures below 1800 C For example, i n the first region of the phase diagram in figure 24 the fraction of Y2O3 in the mixture is given by equation 2-17 (lever rule) where C0 is the initial concentration of Al2O3 in the first region (before thermodynamic equilibrium). The fraction of Y4Al2O9 in the first region would then be 10 0% minus the % Y2O3 calculated by equation 2 17. At any given C0, these relative amounts of Y2O3 and Y4Al2O9 would remain the same for temperatures below 1800 C % 23 = 0 333 00 333 100 % 2 -1 7 In ALD, intimate mixtures can easily be obtained by making alternating stacks of films, each of only a few nanometers or less in thickness. These nanolaminate stacks can then be calcined at high temperatures t o facilitate self -diffusion into the desired solid phase. Stoichiometric amount s of deposited materials are made by noting that the number of moles of a film can be determined from the film thickness and area as shown in equation 218 as both density and m olar mass are easily obtained values. = 2 -18 The stoichiometric ratio of a binary system is shown in equation 219. For an alternating stack of films, area A is equal to area B and therefore Cx (where x is A or B) is a material constant equal to density over molar mass.

PAGE 43

43 = 2 -19 Solving equation 2 19 for tA results in equation 2 -20. For a desired stoichiometric ratio, the thickness of film A must equal to the stoichiometric ratio times a constant times the thickness of film B (equation 2 20). = 2 -20 Inter -diffusion such that complete homogenization is approached can be approximated from Ficks second law [ 5 9 ]. A nanolaminate stack with very thin films approximates a diffusion couple that is approaching complete homogenization. A s olution of Ficks s econd l aw for the complete homogenization case is referred to as the long -time solution [ 59 ] In this ca se, the longtime solution for a diffusion of solute out of a film (slab) leads to a result of 2 where L is the diffusion length D is the constant diffusion coefficient, and t is time [ 5 9 ]. This solution provides an approximation for the extent of a diffusion-controlled process. For a nanolaminate stack, atoms of one layer will need to have diffusion distance s no more than 1/2 the thickness of the other layer to transform the nanolaminate into a solid-solution in thermodynamic equilibrium. In other words the thickness of an individual film in a nanolaminate stack would need to be 4 where h is the film thickness, for thermodynamic equilibrium to be reached for some time, t Rearranging this equation in terms of D results in equation 2 -21. h216 2 -21

PAGE 44

44 Equation 221 can provide estimates of inter diffusion constants b etween two films can be determined if the thickness of the nanolayer is known and the time to reach thermodynamic equilibrium at some temperature is determined. 2.4.2 Y4Al2O9 Atomic layer deposition of the Y4Al2O9 has not been previously demonstrated. The Y4Al2O9 phase has a monoclinic structure [ 60 ] and is also known as yttrium aluminum monoclinic or YAM. This phase is used as a luminescent material [ 61 62] and as a thermal barrier coating [ 63 ]. In YAM, two of the Y ions are sevenfold coordinated and the other two are six fold coordinated, while all of the Al ions are tetrahedrally coordinated [ 64 ]. The Oxygen sites are more complex with four distinct arrangements among the nine sites [ 64]. YAM contains 13.3 atomic % Al and 26.7 atomic % Y. From the phase diagram (section 2.4.1) [ 1 5 ], YAM is 33.3 mol% Al2O3 in Y2O3. A mixture containing m ore than 33.3 mol% and less than 50 mol% Al2O3 in Y2O3 is a combination of YAM plus YAP at thermal equilibrium. Less than 33.3 mol% Al2O3 in Y2O3 results in a combination of YAM plus Y2O3 at thermal equilibrium. 2.4.3 YAlO3 The YAlO3 phase has a perovskite structure of the GdFeO3 type [ 65] and is known as yttrium aluminum perovskite or YAP. This phase is often used as a scintillator material such as YAlO3:Ce [ 63,64 ]. In Y AP, the Y and Al are both octahedrally coordinated in a distorted octahedral formed by the O atoms [ 64 ]. YAP contains 20 atomic % Y and 20 atomic % Al. In terms of mol%, YAP is 50 mol% Al2O3 in Y2O3 according to the phase diagram (section 2.4.2) [ 1 5 ]. A co mbination of more than 50 mol% but less than 62.5 mol% Al2O3 in Y2O3 is a mixture of YAP plus YAG at thermal

PAGE 45

45 equilibrium. When there is a combination of less than 50 mol% but greater than 33.3 mol% Al2O3 in Y2O3, a YAP plus YAM mixture is formed at thermal equilibrium 2.4.4 Y3Al5O12 Atomic layer deposition of Y3Al5O12 has not been previously demonstrated. The Y3Al5O12 phase has a cubic structure of the garnet type and is also known as yttrium aluminum garnet or YAG. The YAG phase is used widely as a phosphor, a scintillator, and as a laser host material. YAG has a complicated structure with 160 atoms in the cubic cell. In YAG, there are two Al sites, an octahedral and a tetrahedral, and Y is eightfold coordinated [ 64 ]. YAG contains 1 5 atomic % Y, and 2 5 ato mic % Al or 62.5 mol% Al2O3 in Y2O3. From the phase diagram (section 2.4.1) [ 1 5 ] more than 62.5 mol% Al2O3 in Y2O3 is a combination of YAG and Al2O3 when at thermal equilibrium A combination with less than 62.5 mol% but greater than 50 mol% Al2O3 in Y2O3 results in a mixture that is YAG plus YAP at thermal equilibrium. 2.4.5 Y3Al5O12:Ce The material Y3Al5O12:Ce (YAG:Ce) is a phosphor utilized in the white light emitting diode (LED) market [ 1 6 ]. White light is produced by absorbing a fraction of the blue li ght from the LED, where the absorbed light stimulates emission of a broad band yellow light from the YAG:Ce crystal The blue light emitted from the LED has a wavelength between 450 nm -480 nm [ 1 6 ] which corresponds well to the broad excitation band of Y AG:Ce which peaks at about 470nm. In this material, Ce3+ is substituted into the Y3+ dodecahedral site of the Y3Al5O12 host [ 66 ] Ce2O3 is used in this work to demonstrate the ability to dope cerium into the YAG or YAM host material. As a dopant, very li ttle material is needed. Concentrations on the order of 1% are typical for light emitting dopants

PAGE 46

46 CHAPTER 3 EXPERIMENTAL PROCEDU RE 3.1 Precursor Analysis Atomic Layer Deposition (ALD) is a surface limited chemical reaction, as discussed in Chapter 2, and typically consists of two sequential half -reactions on a substrate surface. The precursor chemistry determines surface limited growth and therefore is an important factor in ALD A precursor must be stable prior to delivery into the reaction chamber as well as be highly reactive with the other precursor when it is absorbed on the substrate (section 2.2.2) The metal precursors used in this study were commercially available from Fisher Scientific. The precursors and their purity were as follows: AlCl3 (99%) Tris(2,2,6,6 tetramethyl -3,5 heptanedionato)yttrium(III) or Y(thd)3 (98%), and Tris(2,4 pentanedionate)cerium(III) or Ce(acac)3 (99.9%), H2O (10 M conductivity) and ozone. An oxygen source (99.994%) was purchased from Airgas inc. The oxygen passed th rough a corona discharge generated in house using an Enaly HGOZ -1000 ozone generator. The ozone generator has a maximum output of 1% O3 in an O3/O2 mixture. The solid precursors were characterized by thermogravimetric analysis (TGA) to determine volatility from 20 C to 600 C in a Seiko Instruments TG/DTA -320. The TGA results were used to determine the appropriate source temperature of the solid precursors to use during deposition. A small nitrogen flow was present during the analysis and all materials volat ized completely. 3.2 Substrate Preparation All ALD films were deposited onto boron doped Si (100) polished wafers with a 100nm SiO2 surface layer grown at 800 C for 12 hours in air? Atomic layer deposition

PAGE 47

47 results in amorphous or polycrystalline films. Post -growth calcining is needed to inter diffuse the Al2O3 and Y2O3 films deposited in this work. Glass substrates would not be able to handle the post growth calcining temperatures of ~1000 C. SiO2 on silicon can handle the calcining temperatures used (97 5 C). ALD rare earth oxide growth on silicon has been demonstrated [7]. The common chemistry uses a diketonate complex and ozone (section 2.4). In this work, a diketonate and ozone process is also used. The use of ozone would oxidize bare silicon to fo rm an interfacial SiO2 layer, therefore, SiO2 on silicon is used in this work to both segregate the bare silicon from ozone and to prevent inter diffusion between the ALD deposited films and bare silicon during the high temperature calcine. Gadolinium Gall ium Garnet (GGG) could also be used rather than silicon if epitaxial growth of YAG was desired since the lattice parameter and structure of GGG has little mismatch with that of YAG. This work sought to investigate more than just the YAG phase, and using th e same substrate for all films for comparative purposes was desired. The oxidized silicon s ubstrates were cleaned with Alconox (a commercially available powder detergent) and water, rinsed with de ionized water, dried with nitrogen, and finally cleaned in an UVOCS Inc. ultraviolet light ozone cleaner for 10 minutes to remove organic contaminants. 3.3 Film Deposition ALD oxide films were deposited onto the oxidized silicon substrates in a commercial (Planar Systems P400) hot wall flow -type atomic layer depo sition reactor. A schematic is provided in Figure 3.1 and photograph of the system in Figure 3.2. Deposition occurred at ~6 t orr pressure with nitrogen (99.99% purity LN2 source ) as a carrier and purging gas. Precursors were alternately pulsed into the reactor using fast -

PAGE 48

48 -diketonate precursors (thd, acac) were evaporated from open glass crucibles inside the reactor with pulsed nitrogen to carry the precursors to the reaction chamber and su bstrate. The aluminum chloride precursor was heated (105oC) in an external aluminum container and transported into the reactor using fast -switching solenoid valves and heated (105oC) lines. Water was transported into the reactor through fast -switching sole noid valves from an external bottle at 20 C. Ozone was generated externally from oxygen (99.994%) in a corona discharge ozone generator (Enaly HGOZ -1000). The 1% ozone in O2 was pulsed into the reactor using fast -switching valves. The valves are normally c losed valves. A pulse consists of opening the valve for 100ms -600ms. The cleaned substrates were placed within a titanium reactor chamber that was previously coated with Al2O3. This reactor chamber has two openings, one which connects to the plenum on the exhaust side and one that connects to the precursor gas manifold (figures 31, 3-2).The chamber is inserted into another chamber, the hot wall chamber, and connections made to the precursor gas manifold and the exhaust. The hot wall is sealed and with the pumping unit on, both the reactor chamber and the hot wall chamber are pumped down to a base pressure of ~2 t orr The schematic for the ALD reactor is given in Figure 3.1. The rough vacuum pumps are Leybold Trivac D65BCS wet rotary pumps with a Leybold RU VAC WSU251 Roots blower. A paper filter (Planar Systems, inc.) upstream from the pump was used to collect particulate with a size >0.1 m generated in the plenum. The plenum (labeled in Figure 31) is a cold walled chamber where excess precursors from the half -reactions of the deposition process can combine to form less reactive chemical species, such as

PAGE 49

49 oxides. But rather than being formed by surface reactions, these oxides are formed in the gas phase resulting in fine powders. The plenum is connected to t he reactor chamber with the hot wall chamber surrounding both the reaction chamber, and the gas manifold (figure 3 -1). The hot wall chamber has a MKS Baratron pressure gauge (10 t orr ) attached to it and is kept at a slightly lower pressure than the reaction chamber, approximately 8 t orr for the hot wall chamber and 6 Torr for the reaction chamber during deposition. The hot wall chamber is surrounded by heating coils, and the reaction chamber is heated by radiative heating from the hot walls. The hot wall ch amber has a large door through which the reaction chamber can be removed and inserted into the hot -wall chamber (figure 32). The reaction chamber is a rectangular box made of titanium. Two ports are built into the reaction chamber to allow gases to enter at one end from the gas manifold and for excess gasses to exhaust at the opposite end. The reaction chamber is coated with Al2O3 before any films are deposited onto substrates. The reaction chamber is sealed from the hot wall reactor by a metal to metal seal at the two ports. The gas manifold is under a constant purge of N2 from the N2 manifold. The N2 manifold has a pressure gauge attached to it and with a 10 sccm flow reads 170 t orr during deposition. The N2 is constantly flowing through every precursor line purging them of any residual gases from the pulsing sequences. When a pulse occurs at one of the solenoid valves, the total flow of N2 into the reactor remains constant as the N2 flow is diverted through the solenoid valve.

PAGE 50

50 It should be noted that ac cording to the drawing there is a path for pulsed AlCl3 to go directly into the exhaust. However, this drawing does not show the line diameters and the conductance of such lines. There is a N2 pulse in addition to the AlCl3 pulse such that the pressure dif ferential favors the AlCl3 to move towards the reaction chamber rather than directly towards the exhaust. Atomic layer films were deposited using the following pulse sequence: metal precursor pulse, nitrogen purge, water or O3/O2 precursor pulse, and nitrogen purge. This sequence was repeated to grow a metal oxide to the desired thickness. Table 3-1 shows the pulse sequence for the binary oxide materials deposited in this study. The growth conditions for each material was determined separately in individual experiments and then the materials were grown together in order to form a polymorph of yttrium aluminum oxide. Table 3 1 Pulse sequence s for the binary oxides deposited in this study. 1 st Pulse 2 nd Pulse 3 rd Pulse 4 th Pulse Y 2 O 3 Y(thd) N 2 Purge O 3 N 2 Purge Al 2 O 3 AlCl 3 N 2 Purge H 2 O N 2 Purge Ce 2 O 3 Ce(acac) N 2 Purge O 3 N 2 Purge

PAGE 51

51 Figure 31 Schematic of Atomic Layer Deposition vacuum system.

PAGE 52

52 Figure 32 Photograph of the Planar Systems P400 hot wall flow -type ALD reactor showing the gas manifold s ide of the reactor. 3.3.1 Al2O3 Deposition Alumina (Al2O3) was the first material grown in this study. The precursors used were AlCl3 and H2O. The chemical reaction is shown in equation 31. Section 2.3. 2 provides a discussion on this reaction. 2 3 ( )+ 3 2( ) 23 ( )+ 6 ( ) (3 -1) The water was maintained at room temperature, 20C, and the AlCl3 was heated to 105C. The pulse sequence (alumina cycle) consisted of (in chronological order) : 0.2s N2 purge through the AlC l3 line 0. 2s of AlCl3 and N2 through the AlCl3 line 0.2s N2 purge through the AlCl3 line

PAGE 53

53 0.2s pause 0.2s N2 purge through the H2O line 0. 2s of H2O and N2 through the H2O line 0.2s N2 purge through the H2O line 0.2s pause f or a total alumina cycle time of 1.6 seconds. The reactor temperature was varied from 200C to 550C with the number of alumina cycles fixed at 5000. For each run at a given temperature, the thickness was measured using a Tencor P2 Profilometer and the average growth rate determined by d ividing the thickness by the number of cycles. A plot of the growth rate as a function of temperature reveals the so-called ALD window (section 2.2.3.1) [ 67 ] X -Ray p hotoelectron spectroscopy (XPS) was used to verify the films composition. X -ray diffractio n (XRD) was used to determine crystallinity of the film. 3.3.2 Y2O3 Deposition Yttria (Y2O3) was also grown in this study. The precursors used were Tris(2,2,6,6tetramethyl -3,5 heptanedionato)yttrium(III), also known as Y(thd)3, and O3 generated from an O2 source The chemical reaction is shown in equation 3 -2. Section 2.3. 3 provides a discussion on this reaction. 2 ( )3 ( )+ 60 3 ( ) 23 ( )+ 57 2( )+ 66 2 ( ) (3 -2) The O3 was delivered at room temperature, 20C, and the Y (thd)3 source was heated to 140C The pulse sequence (yttria cycle) consisted of (in chronological order) : 0.2s N2 purge through the Y(thd)3 line 0.2s of Y(thd)3 and N2 through the Y(thd)3 line 0.2s N2 purge through the Y(thd)3 line 0.2s pause 0.2s N2 pur ge through the O2/O3 line 0.2s of O2/O3 and N2 through the O2/O3 line 0.2s N2 purge through the O2/O3 line 0.2s pause

PAGE 54

54 for a total yttria cycle time of 1.6 seconds. The reactor temperature was varied from 200C to 550C with the number of yttria cycles fi xed at 10000 For each run at a given temperature, the thickness was measured using a Filmetrics F20 optical thin film measurement tool and the growth rate determined by dividing the thickness by the number of cycles. A plot of the growth rate as a function of temperature reveals the ALD window (section 2.2.3.1 ) [ 67 ] X-ray photoelectron spectroscopy ( X PS ) was used to verify the films composition and X-ray diffraction ( XRD ) was used to determine crystallinity 3.3.3 Ce2O3 Deposition Ceria (Ce2O3) was anothe r material grown in this study. Ce2O3 doped YAG is a well known phosphor material [ 16,19,20 ]. The precursors used were t ris (2,4pentanedionate) cerium(III), also known as cerium(III) acetylacetonate ( Ce(acac)3), and O3. The chemical reaction is shown in eq uation 33. Section 2.3.3 provides a discussion on this reaction. 2 ( )4 ( )+ 46 3 ( ) 2 ( )+ 76 2( )+ 88 2 ( ) (3 -3) The O3 was delivered at room temperature, 20C, and the Ce(acac)3 source was heated to 140C The pulse sequence ( Ce2O3 cycle) consisted of (in chronological order) : 0 .2s N2 purge through the Ce(acac)3 line a 0.2s of Ce(acac)3 and N2 through the Ce(acac)3 line 0.2s N2 purge through the Ce(acac)3 line 0.2s pause 0.2s N2 purge through the O2/O3 line 0.2s of O2/O3 and N2 through the O2/O3 line 0.2s N2 purge through the O2/ O3 line 0.2s pause

PAGE 55

55 for a total Ce2O3 cycle time of 1.6 seconds. The reactor temperature was varied from 200C to 550C with the number of Ce2O3 cycles fixed at 10000. For each run at a given temperature, the thickness was measured using Filmetrics F20 optical thin film measurement tool and the growth rate determined by dividing the thickness by the number of cycles. A plot of the growth rate as a function of temperature reveals the ALD window (section 2.2.3.1 ) [ 67 ]. XPS was used to verify the films compos ition and XRD was used to determine crystallinity. 3.3.4 Pseudo -binary Deposition To grow a pseudo-binary oxide in ALD, the alumina and the yttria were grown sequentially as a multilayered stack of very thin films. The pulse sequences for each material we re introduced sequentially in such a way as to achieve the desired stoichiometry for the bulk film. For example the Al2O3 pulse sequence was followed by the Y2O3 pulse sequence. The necessary thickness of these materials deposited by ALD to achieve a desir ed stoichiometry can be determined using methods described in section 2.4.1. The density and molar mass of yttria are 5.010g/cm3 and 225.81g/mol, respectively. The density and molar mass of alumina are 3.97/cm3 and 101.96g/mol, respectively. The molar dens ity of each is given by dividing density by the molar mass. This results in 2.219x102 mol/cm3 Y2O3 and 3.894x102 mol/cm3 Al2O3. The volume of each material can be expressed as the area of the substrate (A) times the film thickness: At. Let be th e thickness of Y2O3 and be the thickness of Al2O3. Then the molar densities can be expressed as 0.219x102 mol/A Y2O3 and 3.894x102 mol/A Al2O3. If Yttrium Aluminum Garnet (YAG) with the chemical formula Y3Al5O12

PAGE 56

56 is desired, th en 3 moles Y2O3 to 5 moles Al2O3 in a given volume are required. Setting the stoichiometric ratio equal to the ratio of molar densities results in equation 3 -4 (a solved equation 219) and s olving for results in = 1 04 3 235 23 = 2 219 x 10 2mol Y2O3 x A x ty ( 3 89 x 10 2 mol Al2O2) x ( A x tAl) (3 -4) In other words, the thickness of the yttria layers needs to be 105% of the alumina layer thickness. Repeating this pro cedure for YAlO3 and Y4Al2O9 by setting the left side of equation 1 to the appropriate stoichiometric ratio and solving for / results in the desired bi -layer thickness profile for a given phase and is summarized in Table 32 Total film th ickness, when in the ALD window [ 67 ], is determined by the number of cycles. Knowing the growth rate of each material in thickness per cycle, the required number of cycles of each constituent binary oxide needed to achieve stoichiometry can be determined. Table 3 2 Necessary thickness ratio s for Y2O3 to Al2O3 thin film bilayers to achieve stoichiometry for a given phase. ty/tal ratio Y/Al Stoichiometric ratio Y4Al2O9 3.54 2/1 YAlO3 1.77 1/1 Y 3 Al 5 O 12 1. 04 3/5 Table 3 3 ALD sequences for Y Al -O film s. YAG Sequence #1 120 Al2O3 + 200 times (103 Y2O3 + 10 Ce2O3 + 103 Y2O3 + 6 Al2O3) + 120 Al2O3 YAG Sequence #2 120 Al2O3 + 175 times (315 Y2O3 + 32 Ce2O3 + 315 Y2O3 + 6 Al2O3) + 120 Al2O3 YAG Sequence #3 6 Al2O3 + 136 times (445 Y2O3 + 2 Ce2O3 + 445 Y2O3 + 3 Al2O3) + 3 Al2O3 YAM Sequence #4 3 Al2O3 + 100 times (612 Y2O3 + 122 Ce2O3 + 612 Y2O3 + 3 Al2O3) + 3 Al2O3

PAGE 57

57 Multiple run sequences to achieve YAG and YAM were developed (Table 3-3). The pulse sequences show the number of cycles for each binary film and the order of deposition. The parenthetical sequence fragment was repeated a number of times equal to the number shown to the left of the parenthesis. The sequences will be described in greater detail in chapter 4. The reactor temperature for these exp eriments has to be such that the ALD windows for both materials overlap. This was determined to be true at 350C (chapter 4.5.1 ). 3.4 Heat Treatment The films were calcined in a Barnstead/Thermolyne furnace (model F6010) in ambient air to inter diffuse th e Al O, and Y into the desired polycrystalline YAM or YAG. For this study the calcining was performed at 975 C for 12 hours in air The crystallinity of the as -deposited and calcined films was examined using x -ray diffraction. Inter -diffusion such that c omplete homogenization is approached can be determined from Ficks second law (section 2.4.1) [ 59]. In this case, the longtime solution for a diffusion of solute out of a film leads to a result of 2 where L is the thickness of the film, D is the diffusion coefficient, and t is time [ 5 9 ]. An element must diffuse at most a distance equal to one half of the film thickness of the constituent films. Concentration gradients prevent an element diffusing from near an interface through the entire film to another interface. In order to facilitate diffusion during heat treatment, the films are grown in multiple pairs such that the stoichiometric ratio (via bi -layer thickness ratio) is preserved whi le minimizing the thickness of the individual films. For example, rather than grow a 95nm Y2O3 film followed by a 100nm

PAGE 58

58 Al2O3 film for a 195 nm film, fifty pairs of films are grown such that the thickness of each film is 1.9 nm Y2O3 and 2.0 nm Al2O3 for a 195 nm film. 3.5 Characterization 3.5.1 X -Ray Diffraction (XRD) The crystal structure of the as deposited and calcined films were determined using a Philips APD 3720 X = 1.54056. X-Ray Diffraction is a well know n method for determining crystal structure [ 68 ]. Samples were o to 70o with a step size of 0.02o. XRD patterns were compared with powder diffraction files from the JCPDS International Center for Diffraction Data to determine the crys talline phases present. 3.5.2 X -Ray Photoelectron Spectroscopy (XPS) The surface composition of the ALD films was determined by X -Ray Photoelectron Spectroscopy (XPS) [ 69 ]. The XPS measurements were performed on a Perkin -Elmer PHI 5100 XPS/ESCA spectrome ter. Mg K radiation with an energy of 1253 eV was used to irradiate the sample with a spot size of 4 x 10 mm Survey scans were taken from 0-1000 eV binding energy with a step size of 0.5 eV. 3.5.3 Secondary Ion Mass Spectroscopy (SIMS) Secondary Ion Mass Spectros copy (SIMS) [ 70 ] was used to determine the relative concentration of the atomic species as a function of film thickness. SIMS is a characterization technique that is used to analyze the elemental composition of a thin film by sputtering the surface with en ergetic primary ions and analyzing the ejected secondary ions using a mass spectrometer. The SIMS measurements were performed on a Perkin-Elmer/Physical Electronics 6600 Quadrapole Dynamic SIMS system at 6

PAGE 59

59 keV and 40-70 nA with a raster size of 250 m2 wit h electronic gating and Cs+ as the primary ion. 3.5.4 Thickness Measurements The thickness of the Al2O3 films was determined by masking a separate substrate with a permanent marker. Post growth the carbon residue of the mask is removed using methanol an d gentle scrubbing. T he thickness of the film is then measured using a Tencor P2 Profil ometer. Multiple step heights were measured and the average of values was used to determine the thickness. For all other films, the ozone precursor prohibited the use of a permanent marker on a separate substrate. Instead, film thicknesses were measured using a Filmetrics F20 optical thin film measurement tool. This tool requires a materials library containing the indices of refraction for the material for a range of wavelengths. For the mixed, multilayer films of Al2O3 and Y2O3, both the targeted material (YAM or YAG) as well as a double layered Al2O3 Y2O3 were used as models. To verify the Filmetrics results, t he thickness of some ALD films was determined by using Second ary Ion Mass Spectroscopy depth profiling to locate the film -substrate interface and measuring the sputtered crater depth using the Tencor profilometer 3.5.5 Scanning Electron Microscopy (SEM) Surface morphology of films was investigated using a Phillips XL 40 field emission source scanning electron microscope (SEM) [ 71 ]. The entire chamber was vented to load or unload samples as this particular SEM does not possess a load lock. The chamber was evacuated using a turbomolecular high vacuum pump backed by a mechanical vacuum pump. An accelerating voltage of 30 kV was used and the probe

PAGE 60

60 current set to a spot size of 3. The beam was stigmated and focused using a goldoncarbon standard sample.

PAGE 61

61 CHAPTER 4 RESULTS AND DISCUSSI ON 4.1 Introduction The results for the ALD growth of Al2O3, Y2O3, Ce2O3, and the pseudo-binary Al2O3Y2O3 doped with Ce2O3 thin films are now summarized. The precursors used, the growth parameters, and characterizations of each material are discussed in Sections 4.2 to 4.5. In Section 4.5, the intermixing growth parameters and their relationship with the properties of the resultant films are also discussed. 4. 2 Al2O3 Deposition The first material studied in this work is Al2O3. First, the analysis of the AlCl3 precursor is presented in Section 4.2.1. Next, data for the growth parameters are shown in Section 4.2.2 with emphasis on the ALD temperature window. The films are then characterized and the results are discussed in Section 4.2.3. The temperature window for ALD growth is determined and t he films are characterized for crystallinity and surface morphology. 4.2.1 Al2O3 Precursors As discussed in s ection s 2. 2 .2 the precursor chemistry plays an important role in surfacelimited growth. The AlCl3 used for Al2O3 growth was purchased from Fisher Scientific and had a purity of 99% The precursor H2O was deionized water available on site with a conductivity of 10 M cm and kept at ~ 2 0 C The AlCl3 was characterized by thermogravimetric analysis (TGA) to determine the weight loss ( volatility ) from 20 C to 600 C to determine the appropriate deposition source temperature for the solid precursor. A small nitrogen flow w as present during the analysis with a heating rate of 10C/min, and all materials volatized completely The

PAGE 62

62 AlCl3 precursor showed an onset of sublimation at 105 C close to the reported source temperature of 110C [ 53 ] (section 2.3. 2 ) therefore 105C was used for the AlCl3 source temperature for Al2O3 growth. The source holder for AlCl3 can hold approximately 30 grams of AlCl3. 4.2.2 Al2O3 Growth Param e ters The temperature dependence of the growth rate of Al2O3 was studied to determine the ALD window (Sec tion 2.2. 3.1). In this work the number of cycles was fixed at 5000 The AlCl3 source temperature was maintained at 105C. The substrate temperature was varied from 295 C to 515 C at 25 C intervals Growth temperatures lower than 295C resulted in films too thin (< 20 nm) to accurately resolve thickness using the Tencor Profilometer indicating a low growth rate. As discussed in section 2.2.3.1, a decreasing growth rate at decreasing temperatures is due to insufficient activation energy being provided for pr ecursor dissociation and subsequent chemisorption. Growth temperatures higher than 515 C resulted in very thick films. This high growth rate is explained by decomposition of the AlCl3 precursor as previously discussed in section 2.2.3.1. All deposition was conducted at a pressure of ~ 6 t orr The films were deposited onto cleaned (section 3.2) glass substrates. The film thickness was measured for each temperature step and these values were divided by the 5000 cycles to determine a growth rate expressed as a thickness per cycle. As can be seen from figure 41, a substrate temperature window for ALD growth exists from about 365C to 465C with a growth rate of ~ 1/cycle. At this growth rate, films up to 2 m can be grown in ~ 12 hours.

PAGE 63

63 Figure 41 Growth rate of Al2O3 versus substrate t emperature. 4.2.3 Al2O3 Characterization X -Ray d iffraction spectra for the as deposited Al2O3 films show a very broad peak at 17 degrees and no sharp peaks indicating that the Al2O3 film was amorphous Figure 42 shows the XRD spectrum for a 500 nm thick film deposited at 350C on a SiO2/Si substrate and is typical for as deposited films within the temperature window The increased intensity in figure 42 at a 2-theta of 70 degrees is due to the Si substrates (002) diffrac tion peak The SEM micrograph in figure 43 of a 1.0 m thick (step profile) Al2O3 film deposited at 350 C shows a qualitatively rough surface morphology suggesting island growth of the film (section 2.2.3.3) Figure 4 4 shows an SEM micrograph of a fractu red cross -section of that same film. The rough morphology of the film can be attributed to self -etching (section 2.3.2) When self etching occurs, the pits formed by self etching as well as the CVD formed Al2O3 particulate result in a rough surface morphol ogy. If we assume that self etching 0.00 0.50 1.00 1.50 2.00 2.50 295 315 340 365 390 415 430 440 465 515 Growth Rate (/cycle)Temperature ( C)Al2O3Growth Rate v. Temperature Al2O3Growth Temperature 'Window'

PAGE 64

64 creates very shallow pits in a single AlCl3 pulse, perhaps only two A ngstroms in depth (~ the Al -O bond length), 5000 cycles would generate a pit about 1 m in depth. This assumes subsequent depositions do not fill the pit and that etching always occurs in the pit Without these assumptions the effects would be less pronounced. This assumption does indicate that self etching can have a significant impact on surface morphology In addition, Al2O3 particulate would be fallin g on the surface in a random fashion rather than layer by -layer, further disrupting the surface morphology. Figure 42 X-Ray d iffraction spectrum from an as -deposited 1.0 m thick Al2O3 film deposited at 350C. 0 50 100 150 200 250 10.000 20.000 30.000 40.000 50.000 60.000 70.000Intensity (arbitrary units)2 -Theta As -deposited Al2O3XRD

PAGE 65

65 Figure 43 SEM micrograph of an as -d eposited Al2O3 film deposited at 350C Figure 44 SEM micrograph of a fractured cross -section as deposited Al2O3 film deposited at 350C Film Surface Cleaved Film Face Cleaved Substrate Face

PAGE 66

66 4.3 Y2O3 Deposition The TGA analysis of the Y(thd)3 precursor for Y2O3 is presented in Section 4.3.1 while the growth parameters are shown in s ection 4.3.2 with emphasis on the low Y2O3 growth rate. The film characteri stics are discussed in Section 4.3.3. An ideal growth temperature was determined and the films were characterized for crystallinity and surface m orphology. 4. 3 1 Y2O3 Precursor s The Y(thd)3 precursor used for Y2O3 growth was purchased from Fisher Scientific and had a purity of 98%. It was reacted with ozone generated inhouse from 99.994% pure oxygen. The ozone was deliver ed in a n O2/O3 mix ture wit h approximately 1% O3 (section 3.1) The Y(thd)3 precursor w as characterized by thermogravimetric analysis (TGA) to determine volatility from 20 C to 600 C The TGA data were used to determine the appropriate source temperature of the solid precursors. A small nitrogen flow ( 100 sccm) was used with a heating rate of 10C/min The Y(thd)3 precursor shows a differential thermogravimetric ( TG) mass loss of 3 2 .0 m g/ min. at 1 4 0 C (Figure 4-5 ) The large peak at 64C is due to desorption of methanol used to cle an the sample holder. The cycle time for a yttria cycle is 1.6s ( section 3.3.2) therefore 0.8 m g of Y(thd)3 is consumed for every yttria cycle at this source temperature.

PAGE 67

67 Figure 45 Differential thermogravimetric mass loss (D TG) of Y(thd)3 as a function of temperature. The broad peak centered at 67 C is due to desorption of methanol used to clean the sample holder. The following calculation was used to determine if a source temperature for Y(thd)3 of 140C wa s sufficient for Y2O3 deposition (equation 42) The reaction chamber has a total surface area of about 0.5 m2 and a Y2O3 thickness of 350 nm was desired, yielding a total Y2O3 volume of 175 x109 m3. The density of Y2O3 is 5.01x106 g/m3. The molar mass of Y2O3 and Y(thd)3 are 225.81g/mol and 638.72 g/mol respectively Therefore the amount of Y(thd)3 needed for a 350 nm thick film is estimated to be 5 gms assuming the growth rate was the same and constant for all part s of the subst r ate and reaction chamber and all the precursor is absorbed ( 175 10 9 3) 5 01 106 233 1 23225. 81 23 2 1 23 1 ( )31 638. 72 ( )31 ( )3 = 5 ( )3 4 -1 0 10 20 30 40 50 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 TG ( g/minute)Temperature (C) Y(thd)3 TG vs. Temperature

PAGE 68

68 4.3. 2 Y2O3 Growth Parameters The temperature dependence of the Y2O3 growth rate was determined using 10000 cycles at a source temperature of 1 40 C T he substrate deposition temperature was varied from 2 00C to 5 00 C at 25 C intervals All deposition s were cond ucted at a pressure of ~6 t orr onto SiO2/Si substrates with a 100 nm SiO2 layer. The film thickness was measured and divided by the 10000 cycles to determine a growth rate expressed as thickness per cycle. Due to an extremely low growth rate for Y2O3, accu rately measuring the thickness of the resultant films was problematic using the Filmetrics optical thin film measurement The thickest Y2O3 film achieved was 30 nm at 350 C corresponding to a growth rate of only 3.0 pm/cycle Without accurate thickness mea surements at varying temperatures, an ALD temperature window could not be observed. As discussed in the literature review (section 2.3.3), reported growth rates for Y2O3 at this temperature should be 0.23 /cycle, a factor of ten greater than the 3.0 pm/cy cle reported here. The difference may be due to the use of a different ALD reactor i.e. a tooling factor The reactor used in this study relies on careful placement of the solid source configuration and the balance of nitrogen flows around and through t he solidsource crucible The delivery of the Y(thd)3 vapor is from an tube-shaped glass crucible (t he Y(thd)3 source in figure 3-1) with nitrogen used as a carrier gas Reasonable assumptions and estimates were made as to the solid-source crucible design and placement These are two important tooling factors that the flux of Y(thd)3 vapor into the reaction chamber depend upon A low flux c an result in low growth rates if it is insufficient for monolayer saturation (section 2.2.3.1). While the temperature used for the Y(thd)3 source were in agreement with both literature (section 2.3.3) and TGA data

PAGE 69

69 (section 4.3.1 and figure 45) the crucible design influences the position of Y(thd)3 in the heating zones. If the Y(thd)3 was placed in lower temperature zones due to the crucible configuration, a low vapor pressure could result in a low flux as well. At higher temperatures the Y(thd)3 source would have a higher vapor pressure resulting in a high flux. There would be an excess flux of Y(thd)3 after surface satur ation occurred and this excess flux would be removed from the reaction chamber. The rate of Y(thd)3 sublimation could result in the consumption of all the Y(thd)3 precursor before the entire ALD sequence is complete. All attempts were made to design the cr ucible such that it was placed in the correct zone. Due to the cost of the purchased Y(thd)3 at the time of this work, once Y2O3 growth was demonstrated further development and improvement of growth parameters ceased and is a topic for future work (section 5.2) 4.3.3 Y2O3 Characterization The XRD spectrum for as deposited Y2O3 films show several diffraction peaks from a polycrystalline film that matched the diffraction peaks reported on the Joint Committee on Diffraction Standards ( JCPDS) card #43-1036 for the cubic phase of Y2O3. A typical XRD spectrum is shown in f igure 4-6 for a film deposited at 350C. The peak s in figure 4-6 at 2-theta s of 34, 62 and 69 are from the Si substrate Using the Scherrer formula [ 49 ] an average crystallite size of 13.3 nm i s calculated. This film had a smooth, uniform surface morphology indicating a layer by layer growth mode (figure 4 -7 ) [ 45 ].

PAGE 70

70 Figure 46 XRD spectrum from a 25 nm thick Y2O3 film as deposited at 350C. The lines show peaks from JCPDS Card #43-1036 for c ubic Y2O3. Figure 47. SEM micrograph of an as -deposited Y2O3 film deposited at 350C 0 20 40 60 80 100 120 140 160 180 200 10.000 20.000 30.000 40.000 50.000 60.000 70.000Intensity (arbitrary units)2 -Theta XRD of ALD Y2O3Films Silicon (222) (400) (134) (622) Silicon (440)

PAGE 71

71 4.4 Ce2O3 Deposition Ceria ( Ce2O3) was the third binary oxide studied. First, the analysis of the Ce(acac)3 precursor is presented in Section 4. 4 .1. Next, the growth parameters are shown in Section 4. 4 .2. The films are then characterized and the results are discussed in Section 4. 4 .3. The films were characterized for crystallinity and surface morphology. 4.4.1 Ce2O3 Precursors The c e rium (III) a cetylacetonate h ydrate [ Ce (acac)3] used for Ce2O3 growth was commercially available from Fisher Scientific with a purity of 95%. It was reacted with ozone generated inhouse from 99.994% oxygen and deliver ed in a n O2/O3 mix ture with 1% O3 (section 3.1) The Ce(acac)3 was charac terized by thermogravimetric analysis (TGA) from 20 C to 400 C to determine volatility and all materials volatized completely. Figure 48 shows the TG plot for Ce(acac)3 as a function of temperature, and a number of peaks are evident. From low temperature to high temperature, the first and second peaks at 64C and 78C result from volatilization of methanol (boiling point 64.6 C [50] and ethanol (boi ling point 78.3 C [50) used to clean the sample holder prior to analysis. The third peak at 100C is attributed to water, consistent with the fact that the Ce(acac)3 is a hydrated chemical and the boiling point of water is 100 C The fourth and sharpest p eak at 134C is attributed to volatilization of Ce(acac)3 and is in agreement with the literature reported value of 140 C [58]. T he sharp peaks at higher temperatures are attributed to thermal cracking of the (acac) ligands to produce organic fragments. T he source temperature for the Ce(acac)3 was chosen as 140C

PAGE 72

72 Figure 48 Plot of mass loss TG of Ce(acac)3 vs. t emperature. 4.4.2 Ce2O3 Growth Parameters The growth of Ce2O3 was investigated using 10000 cycles and a source temperature of 140C. The s ubstrate temperature was varied from 200C to 500C at 25 C intervals and at a pressure ~6 t orr The growth rate of Ce2O3 was so low that film thickness could not be measured at any substrate temperature (thickness < 20 nm ). 4.4.3 Ce2O3 Characterization D espite the lack of a film thickness measureable by the profilometer X -ray photoelectron spectroscopy ( XPS ) spectra (figure 4-9 ) show that Ce2O3 is present on the SiO2/Si substrate. As to the oxidation state ( Ce+3 or Ce+4), the XPS spectra in figure 4 -9 s uggests Ce3+ is the dominant oxidation state. The Ce3+ 3d3/2 peak (904.5 eV) is larger than the Ce4+ 3d3/2 peak (917.0 eV) indicating a greater concentration of the Ce3+ valence state. Both valence states have a charge transfer peak that appears as shoulde rs at lower binding energies in figure 410. It has been shown by Rack and 0 1 2 3 4 5 6 7 8 9 10 21 47 64 83 103 127 145 165 185 207 228 250 275 358 TG ( g/minute)Temperature (C) TG of Ce(acac)3

PAGE 73

73 Holloway [73] that there is an increase in the Ce3+ charge transfer peak compared with the Ce3+ ground state peak in XPS analysis as oxygen concentration increases. The large charge transfer peaks for both Ce3+ 3d5/2 and 3d3/2 compared with their corresponding ground state peak is another indicator that Ce3+ is the abundant valence state. The 3d3/2 for Ce4+ is buried within the Ce3+ 3d5/2 peak. Typically, diamagnetic Ce4+ show less s hake-up and satellite peaks than does paramagnetic Ce3+ [ 69 ]. The presence of silicon peaks in the XPS spectra (figure 49) suggests either that the Ce2O3 film is very thin and/or the uniformity of the Ce2O3 coverage on the SiO2/Si substrate is non-uniform The sodium (Na) Auger peaks are likely due to contamination ( fingerprints ) on the sample. Figure 49 XPS spectrum showing that C e and O were detected on SiO2/Si substrate after 9000 cycles of g r owth at 350 C Binding Energy (eV) N(E) Min: 2467 Max: 508280 1000 900 800 700 600 500 400 300 200 100 0 Ce 4d Si 2p3 Si 2s Na KVV O 1s Ce MNV O KVV Ce 3d5 Ce 3d3

PAGE 74

74 Figure 410 High resolution XPS spect rum showing the Ce3+ 3d peaks from figure 4-9: the intense charge transfer peaks and the small Ce4+ 3d3 /2 peak. The Ce4+ 3d5/2 peak is buried beneath the Ce3+ 3d3/2 peak. No XRD peaks were detected after the 10000 cycles of Ce2O3 growth However, SEM mic rographs in figure 4 -11 show that island growth dominated as evidenced by the island features ( left ) when compared to a feature-less bare Si/SiO2 substrate (right ). Since these films were to be used as a dopant rather than a bulk material, thick films we re deemed unnecessary for this work. As will be seen (section 4.5), the inclusion of these films into the mixed Al2O3Y2O3 phases resulted in very rough films. 855 875 895 915 935 955N(E) arbitrary unitsBinding Energy (eV)High Resolution XPS of Ce 3d peaks Ce4+ 3d3/2Ce3+ 3d 3/2 Ce3+ 3d5/2Ce3+Charge transfer peaks

PAGE 75

75 Figure 411 SEM images of a Ce2O3 coated Si/SiO2 surface ( left panel 100 nm scale bar ) and a bare SiO2/Si surface (right panel 500 nm scale bar). 4.5 Al2O3-Y2O3 Pseudo Binary Deposition Alternating deposition of the components of the Al2O3Y2O3 pseudo-binary system followed by heating can result in a ternary phase By alternating the deposi tion cycles of Y2O3 and Al2O3, a stoichiometric mixture can be produced for the two phases discussed in section 2. 4 : monoclinic Y4Al2O9 ( YAM ) and garnet Y3Al5O12 ( YAG ). During calcining the layered films inter -diffuse and crystallize into the appropriate phase. The different phases can be formed by changing the ratio of the number of cycles to grow Al2O3 to the number of cycles to grow Y2O3, T he growth parameters used, including the number of cycles for each of the component phases, are reported in Section 4.5.1 T he results and discussion of characterization of the as deposited and calcined films are reported in Section 4.5.2. 4.5.1 Al2O3-Y2O3 Pseudo Binary System Growth Parameters The optimal growth temperature of Y2O3 wa s 350C (section 4.3.3) and nearly overlaps the ALD temperature window for Al2O3 ( 365 C to 465 C -see section 4.2.2). Since Al2O3 exhibited high deposit ion rates, 350C was chosen for the growth

PAGE 76

76 temperature for the mixed deposition so as to optimize the very slow Y2O3 growth. T his temperature is slightly outside the Al2O3 ALD temperature window (figure 4-2). At this temperature the Al2O3 growth rate begins to decline due to insufficient activation of the precursors for adsorption and/or growth (section 2.2.3.1) With the Al2O3 growth rate ap proximately 100 times greater than the Y2O3 at this temperature (sections 4.2.2 and 4.3.2) very few Al2O3 cycles are needed when combined with many Y2O3 cycles to achieve stoichiometry. The pressure was ~6 torr and all deposit s were on cleaned SiO2/Si su bstrates. After deposition, samples were calcined at 975C for 12 hours to aid in crystallization of the thin films These ALD process conditions were used for both YAG and YAM film growth 4.5. 2 Y3Al5O12 (YAG) Deposition To achieve stoichiometr ic Y3Al5O12 (YAG) phase, the thickness of the Y2O3 needs to be 105% of the Al2O3 layer (section 2.4.1) or tY = 1.04 tAl. A single thick film of Al2O3 followed by a single thick film of Y2O3 followed by high temperature calcining for a long time could result in the YAG phase However the inter diffusion distances needed are much too large for the given calcin e temperature (975 C) and time (12 hours) and would probably result in an inhomogeneous sample. As discussed in section 2.4.1, ALD is capable of producing an intimate mixture that is near complete homogeneity for stoichiometric solid solution of that mixture This is achieved by creating a nanolaminate stack of the constituent binary materials I n this case, alternating layers of nanoscale Al2O3 and Y2O3 are made into a nanolaminate stack forming an intimate mixture. A pair of films containing both Al2O3 and Y2O3 is called a bilayer and is a diffusion couple during post growth annealing. As a first approximation for reasonable diffusion distances for an intimate m ixtu r e and therefore bilayer thickness the lattice parameter

PAGE 77

77 of the desired material was considered. For YAG, the lattice parameter is ~ 1.2 nm; therefore a reasonable thickness of the bilayer (Al2O3Y2O3 couple) is 1.2 nm Within this distance both Al2O3 and Y2O3 need to be present or tY + tAl = 1.2 nm. In addition, tY must equal 1.04tAl in order to maintain stoichiometry, as discussed above. The calculated layer thickness es for stoichiometry and intimate mixing are very small at tY = 0.62 nm and tAl = 0. 59 nm These thickness es maintain the 1.04 :1 thickness ratio of Y:Al and sum up to 1.2 nm providing both the necessary stoichiometric ratio and intimate mixing The growth rate for Al2O3 was determined to be 1.0 /cycle at 350C (section 4.2.2) and the gr owth rate for Y2O3 was determined to be 3.0 pm/cycle (section 4.3.2). Therefore, to achieve the above thicknesses of tAl and tY, 6 Al2O3 cycles and 207 Y2O3 cycles are needed and together maintain t he proper mixture of Al2O3 and Y2O3 for the stoichiometr y of YAG This pair of cycles can be repeated many times in order to create thicker films of YAG T o dope with Ce2O3, the Y2O3 cycle s were split in to two equal parts with the Ce2O3 cycle s between them. The cerium atom substitutes for yttrium in Y3Al5O12:Ce. Cerium is a large rare earth element with an ionic radius of 114 pm [ 59 ] Using Paulings rules [ 59 72 ] for coordinat ion with the oxygen anion shows that cerium has a coordination number of 8, and therefore sit s in the dodecahedral site in the Y3Al5O12 lat tice which is the yttrium site Therefore, depositing Ce2O3 within the Y2O3 allows for Ce2O3 to readily occupy the dodecahedral sites. T he number of Ce2O3 cycles was set at 5% of the Y2O3 cycles, so 10 Ce2O3 cycles we re inserted between two sets of 103 Y2O3 cycles. 4.5.2.1 Y3Al5O12 sequence #1 With the theoretical justification from section 4.5.2 in mind, a complete ALD sequence for the deposition of Y3Al5O12 was determined. The total cycle sequence wa s :

PAGE 78

78 120 Al2O3 cycles + 200 times ( 103 Y2O3 cycles + 10 Ce2O3 cycles + 103 Y2O3 cycles + 6 Al2O3 cycles) + 120 Al2O3 ( YAG sequence #1). Sequence #1 includes 2 cladding la y ers of Al2O3. Since YAG is the aluminum rich phase, the boundaries of the film were made Al2O3 to provide a slight excess Ideal atomic concent rations for Y3Al5O12 are 15% yttrium, 25% aluminum, and 60% oxygen. Given the above growth rates for Al2O3 (section 4.2.2) and Y2O3 (section 4.3.2) a film thickness of 270 nm was expected The films deposited with YAG sequence # 1 were measured to be 330 n m thick and XRD data show they are amorphous both as deposited and after heat -treatment at 975 C for 12 hrs in air Quantification of XPS data in figure 4-12 show that growth sequence #1 produced films with Y2O3 concentrations well below stoichiometry i ndicating a faster than expected growth rate for Al2O3 and/or a sl ow er growth rate for Y2O3 when compared to their respective binary phase growth. The XPS data show 37.7% (atomic percent) aluminum, 2.9 % yttrium and 59 .4 % oxygen. Using these atomic percents and the measured film thickness, the relative growth of Al2O3 and Y2O3 in this film can be determined using the equations from section 2.4.1 The accuracy of the atomic percents in XPS can be as high as 90% for high intensity peaks and as low as 50% for l ow intensity peaks [69,74] The oxygen concentration at 59.4% is close to the ideal of 60%. The yttrium concentration, however, could be 1.5% to 6% due to the inaccuracy of XPS at low peak intensities. As a first approximation, the XPS values indicate too much aluminum and too little yttrium. New growth rates of 2.0 /cycle for Al2O3 and 1. 0 pm/cycle for Y2O3 can be calculated and are consistent with the measured thickness (330 nm) and the measured atomic percents of yttrium and aluminum in XPS (figure 4-

PAGE 79

79 12). The XPS plot shows no cerium peaks. This is expected as very little cerium was added and XPS only detects those atoms within ~1 nm from the surface. It is possible that the excess Al2O3 on the surface did not completely diffuse upon calcining resulting in XPS results showing a higher than stoichiometric amount of aluminum. However, the phase diagram (section 2.4.1) shows that a mixed phase of Y3Al5O12 and Al2O3 should occur when there is an excess of Al2O3. The films likely remained amorphous due to the excess Al2O3 and insufficient Y2O3 throughout the deposited film for transformation into Y3Al5O12 rather than the film terminating in Al2O3. The increase in Al2O3 growth rate is attributed to an increase in the wetting of the AlCl3 with the Y2O3 surface (s ection 2.2.3.3). A monolayer of Al2O3 formed on Y2O3 approaches a full monolayer in fewer cycles than a monolayer of Al2O3 formed on SiO2. Therefore, for a given number of cycles, the growth rate will be larger. Binding Energy (eV) N(E) Min: 1793 Max: 207363 1000 900 800 700 600 500 400 300 200 100 0 Al 2p 37.7 % Al 2s Y 3d 2.9 % Y 3p3 C 1s Y 3p1 Y 3s O 1s 59.4 % F 1s C KVV O KVV Figure 412 XPS of Y2O3-Al2O3 film grown using YAG sequence #1 (section 4.5.1.1).

PAGE 80

80 Figure 413 SEM images of YAG films via sequence #1. a.) as deposited and b.) calcined at 975C for 12 hours. Figure 413 shows the SEM images of the film deposited using YAG seq uence #1. The films are very rough as -deposited, in what appears to be peeling sheets of material (the string shaped bright areas are edges of the peeled sheets). Upon calcining at 975C, the materials coalesce into what appear to be rods. 4.5.2.2 Y3Al5O12 sequence #2 These new growth rates of 2.0 /cycle for Al2O3 and 1.0 pm/cycle for Y2O3 were used for mixed film depositions To improve the stoichiometry and produce a YAG film, the number of cycles of Y2O3 was increased in YAG sequence #2: 120 Al2O3 cycl es + 175 times ( 315 Y2O3 cycles + 32 Ce2O3 cycles + 315 Y2O3 + 6 Al2O3) + 120 Al2O3 cycles. This sequence was expected to produce a film thickness of 375 nm. The growth cycles are such that Al2O3 thickness should be twice that of Y2O3. In addition, calcul ations show that the resulting mixture will contain 31.9% Al and 7.7% Y. Since YAG is an aluminum -rich phase (section 2.4.4) it should co exist with Al2O3 if there is more than 25% Al (62.5 mol%, section 2.4.1) At twice the thickness of Y2O3, YAG plus a.) b.)

PAGE 81

81 Al2O3 should be the case since the nece ssary condition for stoichiometry is roughly equally thick films (section 3.3.4, table 32). The as deposited films grown with YAG sequence # 2 were 350 nm thick in reasonable agreement with the predicted thickness of 3 75 nm (section 4.5.1) The XRD spectra from as -deposited films using YAG sequence # 2 d id not contain any peaks from Y2O3 or YAG and had a broad peak from an amorphous phase at a 2-theta of 17o and sharp diffraction peaks at 32o, 61o, and 70o from the Si substrate (figure 4 14). However after calcining at 975 C for 12 hrs in air the XRD pattern consists of peaks from a YAG polycrystalline film (figure 4 15 ). The (420) peak is dominate with a smaller (400) peak showing. The peaks from the calcined film m atch ed well with those reported on JCPDS card #33 -0040. Using Scherrer's formula [ 71 ], an average crystallite size of 25 nm was calculated from XRD peak broadening. Figure 414 XRD spectrum from an as deposited film grown with YAG sequence #2 (see secti on 4.5.1). 0 50 100 150 200 250 300 350 400 450 500 10 20 30 40 50 60 70Intensity2 -theta XRD of YAG Sequence #2 Film As -Deposited

PAGE 82

82 Figure 415 XRD spectrum from a film grown with YAG sequence #2 and annealed at 975C for 12 hrs in air. Using equation 221 the c oefficient for inter -diffusion of the Y2O3 and Al2O3 thin layers can be estimated. Crystalline YAG was obtained after a 12 hour (~43200 sec) calcine. The distance of inter diffusion between Al2O3 and Y2O3 is calculated by multiplying the growth rate for Al2O3 by the number of cycles in a single cycle. In sequence #2 this is 2 0 ( 6 ) = 1 2 Therefore, the diffusion coefficient for Y2O3 inter -diffusion with Al2O3 is 2.1x1024 m2/s. The diffusion coefficient for O in Al2O3 is ~1014 cm2/sec at 1400C [59]. At 975C, the calcine temperature of these films, the diffusion coefficient f or O in Al2O3 can be projected to be as low as 1023 m2/sec [59]. The diffusion coefficient at 975C for O in Y2O3 is ~5x1013 m2/sec [59]. Diffusion coefficients at 975 C for Y in Y2O3 and Al in Al2O3 are 1018 m2/sec and 1022 m2/sec, respectively [59]. Complete homogenization is limited by the element with the slowest diffusion rate to the interface between Al2O3 and 0 10 20 30 40 50 60 70 80 90 100 20 30 40 50 60 70Intensity (arbitrary units)2 -Theta XRD of YAG Sequence #2 Film after calcining at 975 C for 12 hrs in air Film Y3Al5O12 Cubic JCPDS #33 0040 (400) (420)

PAGE 83

83 Y2O3 where YAM, YAP, or YAG is then formed. This suggests that the diffusion mechanism is rate -limited by oxygen diffusion through Al2O3. The XPS data from the as -deposited YAG sequence #2 films continue to show a low amount of yttrium (Figure 416 ). T he Y concentration was higher at 7.5 atomic % from YAG sequence #2 versus 2.9% from YAG sequence #1 and is remarkably close to the expected va lue of 7.7 atomic % (section 4.5.1) There is also a reasonable match of the expected Al concentration at 31.9 atomic % Al with the XPS measured 29.4 atomic %. Note that the Al2O3 cap is still present in YAG sequence #2. The XPS spectra also show peaks from carbon and chlorine, even after a 10 min sputter with 4 keV Ar+ to remove advantageous carbon from the surface. The carbon and chlorine appears to be incorporated throughout the film and is likely due to incorporation of ligands from the Y(thd)3 and chlo rine from the AlCl3 precursors. In addition, there is no detectable XPS peak from Ce in this film (figures 4 -16 ). XPS only detects elements within 1 nm of the surface of a material with atomic concentrations > 1% so a cerium signal is unlikely at dopant concentrations In addition, there was a 120 cycle capping layer of Al2O3 with an expected thickness of 24 nm therefore detection of yttrium by XPS was unexpected However, a non -uniform surface coverage of Al2O3 on Y2O3 would allow detection of yttrium a nd the XPS data not being equal to the expected stoichiometry for YAG. Indeed, the SEM images discussed below suggest that this hypothesis may be true. The surface of this YAG film is very rough as seen by the SEM micrographs in figure 417 The f igures 4 17a ) and 4-17b ) show the as -deposited surface (the scale bar shows 1 m) and the calcined surface, respectively. The surface is rough both before

PAGE 84

84 and after calcining at 975C. As will be discussed below, it is believed that Ce2O3 incorporation into these f ilms leads to the very rough surface morphologies. Binding Energy (eV) N(E) Min: 3450 Max: 316460 1000 900 800 700 600 500 400 300 200 100 0 Y 4p Y 4s Al 2p 29.4 % Al 2s Y 3d 7.5 % C 1s 1.9 % Cl 2p3 0.8 % Ar 2p3 Y 3p3 Y 3p1 Y 3s O 1s 60.4 % O KVV Figure 416 XPS spectru m from an as -deposited film grown with YAG sequence # 2 after a 10 minute sputter with 4 keV Ar+. Figure 417 SEM images of YAG films via sequence #2. a.) as deposited and b.) calcined at 975C for 12 hours. Other depositions, using sequence #2, were made except that the number of Ce2O3 cycles was adjusted in an attempt to achieve dopant levels of cerium in the YAG b.) a.)

PAGE 85

85 film that would result in ph otoluminescence (PL) It was found that the ratio of yttria cycles to Ce2O3 cycles (Y:Ce) influenced the roughness of the films. Figure 418 shows SEM image s of 3 different films with increasing Y:Ce ratios Panels a. and b. are from YAM sequence #4 (see b elow) and have a Y:Ce ratio of 10 :1. Panels c. and d. are from YAG sequence #2 and have a Y:Ce of 20:1 and are the same as in those in figure 417. Panels e. and f. are from YAG sequence #3 (see below) and have a Y:Ce of 445:1. From the sequence of SEM ima ges in figure 4 18, it is apparent that the number of Ce2O3 cycles plays a critical role in surface morphology. The Volmer Weber growth mode (island growth) [ 45 ] occurs when the films surface energy is greater than the substrate (section 2.2.3.3) It is postulated that i n this case the adsorbed Ce(acac)3 on Y2O3 is such that the Ce(acac)3 surface energy is greater than the underlying Y2O3 surface energy leading to island growth. Or p erhaps the adsorbed Y(thd)3 surface energy is greater than the underlying Ce2O3. In addition, crystalline Ce2O3 is hexagonal compared to the cubic structure of Y2O3. A lattice mismatch between the Ce2O3 and Y2O3 films could lead to island growth as well. Figure 419 shows a SEM cross-section of a calcined YAG sequence #2 film. The cross-section shows that the film has a sharp interface with the SiO2 layer of the Si substrate The film shows either some porosity or more likely, inhomogeneity from the film mixture of Al2O3 and Y3Al5O12.

PAGE 86

86 Figure 418 SEM images of as deposited (topa, c, e ) and calcined (bottom b, d, f ) at 975C YAG films with increasing Y:Ce ratios of ( a. & b. ) 10 :1, (c. & d. ) 20 :1, and ( e. & f. ) 445:1. Figure 419. Cross -section SEM image of calcined YAG sequence #2 film. a.) b.) c.) d.) e.) f.) YAG sequence #2 film SiO 2 layer Si

PAGE 87

87 4.5.2.3 Y3Al5O12 sequence #3 The XRD spectrum f rom the smooth, calcined film shown in figure 418f is shown in figure 419 and a strong (400) peak from polycrystalline YAG is observed (YAG Sequence #3) In addition a broad, diffuse peak at 18 could be a nascent (211) peak. The sharp peak at 33 is likely from silicon and not from the YAG (420) due to its very narrow shape. This film was grown with the following growth sequence: 6 Al2O3 cycles + 136 times (445 Y2O3 cycles + 2 Ce2O3 cycles + 445 Y2O3 cycles + 3 Al2O3 cycles) + 3 Al2O3 (YAG sequence #3). This sequence was expected to produce a film thickness of ~ 205 nm. The actual film thickness measured was 300nm. Using Scherrers formula [ 71 ], a crystallite size of 13.3 nm was calculated by XRD peak broadening The deposited film was ~300 nm whereas the predicted value was ~205 nm. The growth rate for Al2O3 and/or Y2O3 must have changed. An increase in the Al2O3 growth rate such that an additional 100 nm of Al2O3 is deposited would imply a Al2O3 growth rate of 4.3 /cycle and the stoichiometry would favor YAG. Likewise, an increase in the growth rate for Y2O3 such that an additional 100 nm of Y2O3 is deposited would imply a growth rate of 1.8 pm/cycle for Y2O3 and th e stoichiometry would favor YAM and YAP (~40 mol%, section 2.4.1) The XRD results in figure 4 -19 show a YAG phase, suggesting an increase in the Al2O3 growth rate. In section 2.3.2 it was noted that the AlCl3 and H2O water chemical reaction produces HCl which can etch the Al2O3 surface. A similar reaction can occur between HCl and Y2O3, etching the Y2O3 surface during the initial cycles of Al2O3 growth. Etching of Y2O3 can continue as long as there is an exposed Y2O3 surface. Once Al2O3 covers the Y2O3 surface, etching of Y2O3 by HCl ceases. This would result in a de crease of Y2O3 thickness. At constant temperature, pressure, AlCl3 flux and surface coverage, the

PAGE 88

88 amount of material removed will be constant. Assuming the bulk growth rate of 3 pm/cycle is reduced by etching such that there is an apparent growth rate of 1.8 pm/cycle, a difference of 1.2 pm/cycle, the amount of material apparently lost is 890 cycles x 1.2 pm/cycles = 1. 1 nm. That is, the films would have been 2.7 nm without etching, but etching reduced them to 1. 6 nm thick. The expected thickness total of Y2O3 was 890 cycles x 1.0 pm/cycles = 0.89 nm. The thicker than expected Y2O3 films resulted in the ~100nm of overall film thickness. This explains the disparity in growth rates from bulk Y2O3 to the nano -laminated Y2O3 (section 4.5.2.1). In section 4.5.2.2, YAG sequence #2 showed a slight decrease from the expected thickness to the measured thickness. In that case, assuming a 3 pm/cycle growth rate for Y2O3 and then subtracting 1.4 nm for etching effects explains the lower than expected film thickness. Us ing equation 221, the inter diffusion coefficient for Y2O3 and Al2O3 can be estimated as in section 4.5.2.2. Crystallinity of YAG was obtained after a 12 hour (~43200 sec) calcine. The diffusion distance of Al2O3 and Y2O3 was determined by multiplying the growth rate for Al2O3 by the number of cycles in a single cycle. In sequence #3 this is 2 0 ( 3 ) = 0 6 Therefore, the diffusion coefficient for Y2O3 and Al2O3 is 2 .1x1024 m2/s. Again, oxygen diffusion in Al2O3 is discusse d in section 4.5.2.2 and has a value of ~1023 m2/s at 975 C. This reinforces the hypothesis that the diffusion mechanism is limited by oxygen diffusion.

PAGE 89

89 Figure 419 XRD spectrum from a film grown with YAG sequence #3 and annealed at 975C for 12 hrs in air. Both YAG sequence #2 and #3 resulted in a polycrystalline YAG material after calcining at 975 C in air. The thicknesses of these films were correctly predicted from the calculated growth rates including etching effects Stoichiometric thickness va lues for the constituent Al2O3 and Y2O3 bilayers resulted in the YAG phase after calcining at 97 5 C for 12 hrs in air. 4.5. 3 Y4Al2O9 (YAM) Deposition For the stoichiometric Y4Al2O9 (YAM) phase, the thickness of the Y2O3 needs to be 354% of the Al2O3 layer (section 3.3.4) or tY = 3.54 tAl. As a first approximation for reasonable diffusion distances, the lattice parameter s of the desired material w ere considered. For YAM, the monoclinic lattice parameters are 0 .74 nm, 1 0 nm, and 1 1 nm; therefore a reasonable thickness of the bilayer (Al2O3Y2O3 couple) is 1. 1 nm. 0 20 40 60 80 100 120 140 10.00 20.00 30.00 40.00 50.00 60.00 70.00Intensity (Arbitrary Units)2 -theta XRD of YAG Sequence #3 film after calcining at 975 C for 12 hrs in air Film at 975 C (420) (400) (211)

PAGE 90

90 Within this distance both Al2O3 and Y2O3 need to be present or, tY + tAl = 1. 1 nm. In addition, tY must equal or be greater than 3.54tAl in order to maintain stoichiometry, as discussed above. Ther efore, the layer thicknesses can be determined to be tY = 0. 86 nm and tAl = 0. 24 nm. These thicknesses maintain the 3.54 :1 thickness ratio of Y:Al and sum up to 1. 1 nm. These thickness values for Al2O3 and Y2O3 are also smaller than twice the calculated di ffusion length (2.6 nm) for O in Al2O3 (section 3.4) therefore calcin ing should produce a homogenous film. The growth rate for Al2O3 was determined to be 2 .0 /cycle at 350C and the growth rate for Y2O3 was initially determined to be 1.0 pm/cycle for the multilayered deposition (section 4.5.2) In section 4.5.2 it was found that a larger number Y2O3 cycles can lead to greater than expected growth rates for Y2O3, namely 1.8 pm/cycle. The growth rate for Y2O3 remains 3.0 pm/cycle for the nano laminate stack as in the bulk. However, in the nano laminate stack HCl from the initial Al2O3 cycles etches the Y2O3 surface. For YAM sequence #4, the 3.0 pm/cycle is assumed for the Y2O3 growth and a 1. 1 nm etch is factored in. In order to achieve the monoclinic phase (YAM) the following sequence was used: 3 Al2O3 cycles + 100 times (612 Y2O3 cycles + 122 Ce2O3 cycles + 612 Y2O3 cycles + 3 Al2O3 cycles) + 3 Al2O3 (YAM sequence #4). The cycles of Y2O3 are such that a slight excess of Y2O3 is present. This sequence shoul d result in a film ~ 340 nm thick and with the stoichiometry of the YAM phase. The number of Ce2O3 cycles was set to 10% of the Y2O3 cycles. The measured thickness of the as -deposited film was ~400 nm ~ 60 nm greater than the predicted thickness The deposited film was ~400 nm whereas the predicted value was ~340 nm, This suggests that there was less than 1.0 nm of etching.

PAGE 91

91 Assuming an etch of 0.4 nm rather than 1.0 nm explains the discrepancy in film thickness. Figure 420 shows the XRD plot for the film gr own using YAM sequence #4 a fter being calcined at 975C for 12 hrs in air Multiple peaks indicate a polycrystalline YAM phase The ( 122) and ( 112) peaks from the YAM phase are the dominant peaks. The crystallite size of the YAM phase was calculated to be 20 nm using Scherrers formula [49]. Using equation 221, the inter diffusion coefficient for Y2O3 and Al2O3 can be calculated as in sections 4.5.2.2 and 4.5.2.3 Crystalline YAM was obtained after a 12 hour (~43200 sec) calcine. The thickness of Al2O3 is calculated by multiplying the growth rate for Al2O3 by the number of cycles. In s equence #2 this is 2 0 ( 3 ) = 0 6 Therefore, the inter diffusion coefficient for Y2O3 and Al2O3 is 2.1x1024 m2/s. Oxygen diffusion in Al2O3 is discussed in section 4.5.2.2 and has a value of ~1023 m2/s at 975 C. This suggests that the diffusion mechanism is limited by oxygen diffusion. The number of growth cycles for Ce2O3 was the greatest in this film (YAM sequence #4) The surface morphology was rough (figure 4 -18a and 4 -18b) for both as deposited and calcined samples respectively. As discussed above (section 4.5.1.1) inclusion of a larger number of Ce2O3 growth cycles results in a rough surface morphology. T he 10:1 ratio of Y2O3 to Ce2O3 cycles in this film is the highest Ce2O3 concent ration attempted. The XPS results are shown in figure 4 -21. This figure shows a Y concentration of 20.7% and no apparent Al. There is a significant Si peak resulting in an 11.5%

PAGE 92

92 concentration. The presence of Si in the XPS spectrum is likely due to the very rough surface as shown in figure 418b. Nonuniform surface coverage of the film could allow the underlying substrate to be detected in XPS. The XRD spectrum for YAM sequence #4 is strong evidence for the formation of polycrystalline YAM. Stoichiometric thickness values for the constituent Al2O3 and Y2O3 bi layers resulted in the YAM phase after calcining at 975C for 12 hrs in air. Figure 420 XRD spectrum from a film grown with YAM sequence #4 an d calcined at 975C for 12 hrs in air. 0 10 20 30 40 50 60 70 80 90 100 10.000 20.000 30.000 40.000 50.000 60.000 70.000Intensity (Arbitrary Units)2 -theta XRD of YAM Sequence #4 film after 975 C Heat Treatment Film 975 C Y4Al2O9 Monoclinic JCPDS #34 0368 (122) (222)

PAGE 93

93 Binding Energy (eV) N(E) Min: 3097 Max: 177830 1100 990 880 770 660 550 440 330 220 110 0 Cl 2p3 3.8 % O 2s Y 4s Si 2p 11.5 % Y 3d 20.7 % Ar 2p3 Y 3p3 Y 3p1 Y 3s O 1s 64.0 % O KLL Figure 421 XPS spectrum from an as -deposited film grown with YAM sequence #4 after a 10 minute sputter with 4 keV Ar+.

PAGE 94

94 CHAPTER 5 CONLCUSIONS AND FUTURE STUDIES 5.1 Conclusions Atomic layer deposition (ALD) of Al2O3, Y2O3, and Ce2O3 films has been demonstrated as well as ALD of pseudo binary Y3Al5O12 (YAG), and Y4Al2O9 (YAM) films YAG and YAM are two of three phases in the pseudobinary Al2O3Y2O3 system B y alternating deposition of very thin films of Al2O3 and Y2O3 to create a nanolaminate stack in stoichiometric amounts, the yttrium aluminum garnet ( YAG Y3Al5O12) and yttrium aluminum monoclinic ( YAM Y4Al2O9) phases can be produced after calcining in air at 975C for 12 hours The proper s toichiometry for a particular phase can be achieved by varying the ratio of the Al2O3 and Y2O3 thicknesses. In this work the ALD precursors for Al2O3, Y2O3, and Ce2O3 were AlCl3/H2O Y(thd)3/ O3 and Ce(acac)3/O3 respectively As is typical for AL D, a layer growth cycle consisted of a metal precursor pulse, a N2 purge pulse, an oxygen precursor pulse, and a final N2 purge pulse. The growth rate at 350 C for Al2O3 was found to be 1.0 /cycle and films up to 1 m thick were grown. Al2O3 was found to be amorphous as -deposited when grown at substrate temperatures of 295C to 515C. The growth rate for Y2O3 was found to be 3.0 pm/cycle and films up to 30 nm thick were grown. Y2O3 deposited at 200C to 500C was found to be a polycrystalline cubic phase a s -deposited. Alternating l ayered deposition of Al2O3 and Y2O3 was found to be amorphous as deposited. Attempts to d op e the samples by inserting Ce2O3 cycles in the middle of Y2O3 cycles resulted in rough films. The growth rate for Ce2O3 was too low to be d etermined when using Ce(acac)3 and 1% O3 in O2/O3 as precursors. The Ce3+ oxidation state was found

PAGE 95

95 to be present in greater amounts than the Ce4+ oxidation state using X -ray photoelectron spectroscopy for Ce2O3 films Films with the YAG phase were produc ed from a multilayer film consisting of sub nano -scale Al2O3 and Y2O3 layers The films were deposited using a pulsing sequence that consisted of YAG sequences # 2 & # 3. These films were amorphous after deposition at 350 C. Post growth calcining at 975C fo r 12 hrs in air transformed the amorphous as deposited multilayer film into a polycrystalline YAG film. It was shown that polycrystalline YAG was detected when using a thickness ratio of Y2O3 to Al2O3 of 1.04 to maintain the proper stoichiometry. Similarly a YAM thin film was produced from an Al2O3 and Y2O3 subnanoscale multilayer film. These films were also deposited at 350C and were calcined at 975C for 12 hrs in air Th e YAM phase required a Y2O3 to Al2O3 thickness ratio of 3.54 to maintain the proper stoichiometry. Both of these phases have been accomplished by growth of sub nanoscale multi layered binary oxides It was found that t he growth rates for the individual binary oxides are different from the growth rates for multi layered structure oxides. T he Al2O3 g r owth rate in mixed deposition with Y2O3 was found to be 2.0 /cycle rather than the 1.0 /cycle for bulk Al2O3 growth The Y2O3 growth rate in mixed deposition with Al2O3 was found to effectively vary from 1. 0 pm/cycle to the 3.0 pm/cycle for bu lk Y2O3 growth due to etching effects from HCl byproducts produced during the initial Al2O3 cycles The layer thickness of Al2O3 and Y2O3 was less than the crystalline lattice parameter (either for YAG or YAM) in order to reduce inter -diffusion distances and therefore the times and temperatures for achieving homogeneous samples.

PAGE 96

96 This approach resulted for the first time the synthesis of Y3Al5O12 and Y4Al2O9 thin films by atomic layer deposition. This w ork allow s for novel and unique developments in optical devices such as phosphors, photonics, quantum confinement of optical states, scintillators, subwavelength optics, and waveguides. Some specific examples of using ALD to deposit Y3Al5O12 include: deposition into high aspect ratio trenches to act as a laser waveguide in a monolithic films, very thin film deposition to increase the resolution of a scintillator, and as organic -free thin film complementary phosphor for light emitting diodes. 5.2 Future Studies This work allows for novel and unique developments in optical devices such as phosphors, photonics, quantum confinement of optical states, scintillators, subwavelength optics, and waveguides. Some specific examples of using ALD to deposit Y3Al5O12 include: deposition into high aspect ratio trenches to act as a laser waveguide in a monolithic films, very thin film deposition to increase the resolution of a scintillator, and as organic -free thin film complementary phosphor for light emitting diodes. Future work on atomic layer deposition of the Al2O3Y2O3 ps eudo binary system is needed for both the formation of yttrium aluminum perovskite ( YAP YAlO3) phase as well as the doping of rare earth materials such as cerium. In addition, continued improvements to the formation of YAG and YAM can be made. With a stabl e YAG process, and improved cerium incorporation, the synthesis of light emitting Y3Al5O12:Ce can be realized. Furthermore, substituting gallium for aluminum and/or gadolinium for ytrrium in Y3Al5O12:Ce can alter the crystal field around cerium. The format ion of pure YAP phase will be difficult. Any deviation from the stoichiometric ratio of unity for Y to Al should result in the formation of either YAM or YAG in addition to YAP based on the

PAGE 97

97 phase diagram (section 2.4.1) Such a balancing act requires very precise thickness control. Thus, process improvements for the formation of Y2O3 and Al2O3 can aid in the development of the YAP phase. Specifically, t he very low growth rate for Y2O3 should be increased. The re are reports in the literature [ 3 9 54 ] of grow th rates on the order of 0.25 /cycle (section 2.3.2) It is believed that the design of the solid source manifold configuration in the ALD reactor used in this could be improved in order to achieve higher Y(thd)3 flux into the reactor Different precursor s could also be used. The highly corrosive nature of AlCl3 may be leading to etching of the Y2O3, yielding a lower than expected growth rate. Other ALD studies of Al2O3 films have use d trimethylaluminum (TMA). R e -tooling of the ALD reactor used in this stu dy for TMA delivery would be more in -line with best practices in the field. It would be desirable to dope YAG and YAM with Ce2O3. The use of Ce(thd)3 as a precursor may be more compatible for Y(thd)3 chemistry and surface saturation of the Y2O3 films. In addition, dependable growth would allow for accurate determin ation of the Ce2O3 growth rate Reliable Ce2O3 growth rate information would aid in depositing the appropriate film thickness needed to achieve the correct Ce3+ concentration for photoluminescence Further development could further explore the concept of multilayered diffusion couples. Using radioactive ions in some layers and examining the position of the radioactive ions in the as -deposited and calcined films could be used to determine the domina te diffusion species and pathways. The use of Ficks second law assumes semi infinite films in a diffusion couple, but clearly the boundary conditions in nano -scaled

PAGE 98

98 films are much different. Indeed, defining a concentration gradient in a sub -nanoscale mul tilayer stack is difficult and would be a very interesting line of research Finally epitaxial deposition of the pseudo -binary Al2O3Y2O3 phases would be an interesting avenue to explore. It may be possible to achieve epitaxial crystalline films without post -growth annealing as the thickness of an individual film approaches bond lengths, literally building a crystal one atomic plane at a time.

PAGE 99

99 LIST OF REFERENCES [1] U.S. Patent 4058430 (1977), to T.S. Suntola and M. J. Antson. [ 2 ] M. Leskela, M. Ritala, Thin Solid Films 409 (2002) 138136. [ 3 ] T. Suntola, Materials Science and Engineering Reports 4 (1989) 261. [ 4 ] T. Suntola in Handbook of Crystal Growth, Vol. 3, by D.T. Hurle, Elsevier, 1994 p. 601. [ 5 ] M. Ritala and M. Leskela in Handbook of Thin Film Materials, Vol. 1, by H.S. Nalwa, Academic Press, San Diego, 2001 p. 103. [6] J.H. Kim, N. Shepard, M.R. Davidson, P.H. Holloway, Applied Physics Letters 83 (2003) 4279-4281. [ 7 ] M. Leskela, M. Ritala, Journal of Solid Sta te Chemistry 171 (2003) 170 174. [8] A. Jones, H. Aspinall, P. Chalker, R. Potter, K. Kukli, A. Rahtu, M. Ritala, M. Leskela, Journal of Materials Chemistry 14 (2004) 3101-3112. [9] Mistry, Kaizad A, International Electron Device Meeting, IEEE Internatio nal, Washington D.C. 2007, 247 250. [10] www.appliedmaterials.com [11] www.aixtron.com [12] www.beneq.com [13] www.lesker.com [ 14 ] M. Kirm, J. Aarik, M. Jurgens, I. Sildos, Nuclear Instruments and Methods in Physics Research A 537 (2005) 251 -255. [ 15 ] T.I Mah, M.D. Perry, Journal of the American Ceramic Society 75 (1992) 2006 2009. [ 1 6] Y. Pan, M. Wu, Q. Su, Journal of Physics and Chemistry of Solids 65 (2004) 845850. [17] J.W. Elam, Z.A. Sechrist, S.M. George, Thin Solid Films 414 (2002), 43-55. [18] J.M. Jensen, A.B. Oelkers, R. Toivola, D.C. Johnson, J.W. Elam, S.M. George, Chemistry of Materials 14 (2002), 22762282.

PAGE 100

100 [ 19 ] V. Baryshevsk yt, M. Korzhikt, B. Minkovt, S. Smimovat, A. Fyodorovt, P. Dorenbost, C. vanEijkt, Journal of Physics: Condensed Matter 5 (1993), 7893 7902. [ 20 ] J. Geusic, H. Marcos, L. Van Uitert, Applied Physics Letters 4 (1964) 182-184. [ 21 ] J. Czochralski, Zeitsch rift fur physikalische Chemie, 92 (1918) 219. [ 22 ] P. Tomaszewski, Journal of Crystal Growth 236 (2002) 1 -4. [ 23 ] P. Yang, P. Deng, J. Xu, Z. Yin, Journal of Crystal Growth 216 (2000) 348351. [ 24 ] G. Zhao, X. Zeng, S. Zhou, J. Xu, Y. Tian, W. Huang, Phy sica Status Solidi (a) 199 (2003) 186191. [ 25 ] G. Neuroth, F. Wallrafen, Journal of Crystal Growth, 198/199 (1999) 435439. [ 26 ] Y. Huang, K. Jang, X. Wang, G. Xian, E. Cho, H. Lee, S -H Kim, Journal of NonCrystalline Solids 353 (2007) 4102-4107. [ 27 ] H Fadlalla, C. Tang, Materials Chemistry and Physics 114 (2009) 99 102. [28] J.W. Kim, Y.J. Kim, Optical Materials 28 (2006) 698 -702. [ 29 ] W. M. Yen and M. J. Weber, Inorganic Phosphors, Compositions, Preparation and Optical Properties p. 1, CRC Press, New York 2004. [30] Ji. Kvapil, Jo. Kvapil, B. Manek, B. Perner, R. Autrata, and P. Schauer, Journal of Crystal Growth 52 (1981), 542 -545. [31] S -K Ruan, J -G Zhou, A -M Zhong,, J -F Duan, X -B Yang,, M -Z Su, Journal of Alloys and Compounds 275 -277 (1998), 7275. [32] P. Tanner, P -T Law, K -L Wong, L. Fu, Journal of Materials Science 38 (2003), 48574861. [ 33 ] L. Diaz -Torres, E. la Rosa, P. Salas, C. Angeles -Chavez, L. Arenas, J.Nieto, Optical Materials 27 (2005) 1217-1220. [ 34 ] H. Kumagai, K. Adachi, M. Ezaki K. Toyoda, M. Obara, Applied Surface Science, 109/110 (1997) 528532. [ 35 ] G. Bai, H. Chang, C. Foster, Applied Physics Letters 64 (1994) 17771779. [ 36 ] Planar Systems, Inc. [http://www.planar.com/Advantages/Innovation/docs/Planar_ALD_Background.pdf] (2004).

PAGE 101

101 [ 37 ] H. Seim, H. Molsa, M. Nieminen, H. FJellveg, L. Niinisto, Journal of Materials Chemistry 7 (1997) 449 -455. [ 38 ] M. Nieminen, S. Lehto, L. Niinist, Journal of Materials Chemistry 11 (2001), 3148 3153. [ 39 ] L. Niinist, M. Ritala, M. Leskela, Materials Science and Engineering B41 (1996) 2329. [ 40 ] L. Niinisto, J. Pivsaari, J. Niinist, M. Putkonen, and M. Nieminen, Physica Status Solidi (a) 201 (2004) 1443-1452. [ 41 ] T. Suntola, Applied Surface Science 100 -101, (1996) 391398. [42] M. Fanciulli G. Scarel G. Lucovsky A. Svane Rare Earth Oxide Thin Films Springer -Verlag, Berlin/Heidelberg 2007. [43] S. Haukka, E L. Lakomaa, A. Root, Journal of Physical Chemistry 97 (1993), 5085 5094. [44] R. Matero, A. Rahtu, M. Ritala, M. Leskela, T. Sajavaara, Thin Solid Films 68 (2000), 17. [45] M. Ohring, Materials Science of Thin Films, Deposition and Structure, 2nd edition Academic Press, San Diego, 2002. [46] R. Puurunen, Chemical Vapor Deposition 10 (2003), 159170. [ 47 ] H. Seim, M. Nieminen, L. Niinisto, H. Fjellvag, and L. -S. Johanson, Applied Surface Science, 112 (1997), 243 -250. [ 48 ] O. Nilsen, M. P eussa, H. Fjellvag, L. Niinisto, and A. Kjeckshus, Journal of Materials Chemistry 9 (1999) 1781-1784. [49] L. Santos Jr., S. Roca, C. Airoldi, Journal of Chemical Thermodynamics 29 (1997), 661-668. [50] D. Lide, Handbook of Chemistry and Physics, 84th Edi tion, CRC Press, New York, 2003. [ 51 ] U.S. Patent 4413022 (1983), to T.S. Suntola, A.J. Pakkala, and S.G. Lindfors. [ 52 ] E L. Lakomaa A. Root, T. Suntola, Applied Surface Science, 107 (1997) 107 115. [ 53 ] L. Hiltunen, H. Kattelus, M. Leskela, M. Makela, L. Niinisto, E. Nykanen, P. Soininen, M. Tiitta, Materials Chemistry and Physics 28 (1991) 379388. [ 54 ] H. Mls, L. Niinist. M. Utrainen, Advanced Materials for Optical Electronics 4 (1994), 389.

PAGE 102

102 [ 55 ] M. Putkonen, T. Sajavaara, L -S. Johansson, L. Niini st, Chemical Vapor Deposition 7 (2001), 44-50. [56] R. Puurunen, Chemical Vapor Deposition 9 (2003), 327-332. [57] R. Puurunen, Chemical Vapor Deposition 9 (2003), 249-257. [58] J. Paivasaari, M. Putkonen, L. Niinisto, Journal of Materials Chemistry 12 ( 2002), 18281832. [59] W. Kingery, H. Bowen, D. Uhlmann, Introduction to Ceramics 2nd edition John Wiley and Sons, New York, 1976. [ 60 ] A. Christensen, R. Hazell, Acta Chemica Scandinavica 45 (1991), 226230. [ 61 ] D. Wang, Y. Wang, Journal of Alloys and Compounds 425 (2006), L5L7. [ 62 ] Y. Yang, K. Jang, X. Wang, G. Xian, E. Cho, H. Lee, S -H. Kim, Journal of NonCrystalline Solids 353 (2007), 41024107. [ 63 ] S. Yang, X. Lan, N. Huang, Journal of Materials Science and Technology 23 (2007), 568 -570. [ 64 ] W. Ching, Y -N, Xu, Physical Review B 59, 12815 -12821. [ 65 ] A.Geller, E. Wood, Acta Crystallographica 9 (1956), 563. [ 6 6] G. Blasse, B.C. Grabmaier, Luminescent Materials Springer -Verlag, Berlin, 1994. [ 67 ] T. Suntola, Applied Surface Science 100/101 (1996), 391398 [ 68 ] B.D. Cullity, S.R. Stock, Elements of X -Ray Diffraction, 3rd edition Prentice Hall, New Jersey 2001. [ 69 ] J.F.Moulder, W.F.Stickle, P.E.Sobol, and K.D.Bomben Handbook of X -ray Photoelectron Spectroscopy Perkin -Elmer Corp., Eden Prairie 1992. [ 70 ] R. Wilson, F. Stevie, C. Magee, Secondary Ion Mass Spectrometry: a practical handbook for depth profiling and bulk impurity analysis, Wiley, New York, 1989. [ 71 ] J. Goldstein, D. Newbury, D. Joy, C. Lyman, P. Echlin, E. Lifshin, L. Sawyer, and J. Michael, Scanning Electron Microscopy and X -ray Microanalysis, 3rd Edition Springer, New York, 2003. [ 72 ] L. Pauling, Journal of the American Chemical Society 51 (1929), 10101026. [73] P. Rack, P. Holloway, S -S Sun, Journal of the Society for Information Display 4/4 (1996), 281 -286.

PAGE 103

103 [74] M. Seah, M. Brown, Applied Surface Science, 144145 (1999), 183187.

PAGE 104

104 BIOGRAPHICAL SKETCH Jason C. Rowland was born in Arcadia, California, on June 11th 1972. He graduated from Eti wanda High School in 1990. He attended the California Polytechnic University, Pomona from 19901991 as an English major. From 1991 -1997 he took time to work and explore the world. During this period he worked as a ski lift operator, a customer service repr esentative, and as a fisherman in Alaska, among other odd jobs. In 1997 Jason re enrolled at California Polytechnic University, Pomona. In 2002 he graduated with a Bachelor of Science degree in Physics with a minor in Chemistry. During his undergraduate st udies, he worked full time at Tecstar Inc., Applied Solar Division, as a Production Technician, a Senior P rocess Control Technician, and eventually promoted to Engineer. At Tecstar, he performed Metal Organic Chemical Vapor Deposition of AlGaAs solar cell devices and their post processing into a working solar cell. Jason was admitted into the University of Florida, Department of Materials Science and Engineering in August 2002 as a University of Florida Alumni Graduate Fellow in Dr. Paul Holloways research group. He earned a Master of Science degree in Materials Science and Engineering in 2004. Jasons work focused on the development of Atomic Layer Deposition and the growth of pseudo-binary Al2O3Y2O3 materials for optoelectronic applications. He received his Ph.D. from the Department of Materials Science and Engineering at the University of Florida under the advisement of Dr. Paul Holloway in May 2010.