Abstract

Epitaxial ferroelectric thin films are potentially enabling materials for microphotonic applications due to their large non-linear optical properties. Subsequently, efforts have been directed toward developing these materials for use as the active layer in integrated, high-bandwidth, electro-optic modulators. This dissertation supplement documents recent progress toward the integration of BaTiO₃ films with silicon substrates. In this study epitaxial BaTiO₃ thin films have been successfully deposited by MOCVD on (001) MgO/ (001) silicon substrates prepared by metal-organic molecular beam epitaxy (MOMBE). To achieve epitaxial MgO a thin, epitaxial b-SiC interlayer was first deposited on the Si using MOMBE. X-ray diffraction and transmission electron microscopy indicate that BaTiO₃ and MgO films are epitaxial, with an orientational relationship of BaTiO₃ (100) // MgO (100) // Si (001) and BaTiO₃ [011] // [011] // Si [110]. Initial polarization and dielectric measurements indicated that the BaTiO₃ / Si films are in a FE state.

Acknowledgements

I would like to begin by thanking my research advisor, Professor Bruce W. Wessels, for his support and guidance through the course of this research. I would also like to thank Professors Prem Kumar, Michael J. Bedzyk, and Vinayak P. Dravid for serving on my thesis defense committee. This work was supported under DARPA through the MURI program entitled "Integrated Devices for Terabit per Second 1.3 and 1.5 Micron Network Applications" at Northwestern University and monitored by the AFOSR under grant F49 620-96-1-0262. Equipment support was provided by BMDO through the DURIP program under grant F49620-96-1-0460. This work also made extensive use of the Central Facilities supported by the MRSEC program of the National Science Foundation (DMR-9632472) at the Materials Research Center of Northwestern University.

Special thanks must be given to several collaborators. Dr. Antonio DiVenere, Dr. Douglas M. Gill, and Seong-Soo Kim (Professor S.T. Ho’s group) provided assistance with thin film device processing and fabrication. Dr. Jinha Hwang was responsible for the high temperature dielectric measurements of the K(Ta,Nb)O₃ films while resident in Prof. T.O. Mason’s group. Dr. Feng Niu (Prof. B.W. Wessels’ group) provided all TEM/SEM work needed for the MgO/Si project, which is not included in this report due to length and time constraints.

I am especially grateful to my fellow group members who provided ferroelectric films for this work: Dr. Bruce Block, Dr. Michael Nystrom, Dr. Gregory M. Ford, David Towner, Andrew R. Teren, and Barbara M. Nichols. I would also like to thank the rest of the Wessels’ group members, visiting scientists, and summer students with whom I had the pleasure of interacting: Roman Korotkov, Joel Gregie, Aaron Blattner, Dr. Carolyn (Duran) Block, Dr. Scott Theiring, Dr. Kuo-Wei (Taylor) Chang, Dr. Gyu-Chul Yi, Dr. Soma Chattopadhyay, Dr. Michael Reshchikov, Dr. Fatemeh (Shadi) Shahedipour, Prof. Michael McInerney, Prof. Ibrahim Abdel-Motaleb, Jerry Majewski, Lisa Kaufmann, and Ellen Siem.

Lastly, I would like to thank all of my friends and family, especially my parents, for their patients and understanding while I was finishing this program. I would like to thank Mark Seniw, Kevin Klug, Rachel Bishop, Eric Seelig, and Dr. Victoria Snyder for befriending me while at Northwestern. Mike McKenna, Randy Smith, and Brian Eschmann, your attempts to revive my ailing social life are greatly appreciated. I would like finish by acknowledging the emotional stability and support provided by favorite performing artist, stress-relief-mechanism, and significant other, Amber E. Dow.
 

Contents

Abstract
Acknowledgements
List of Figures
List of Tables
Chapter

1. Introduction *

List of Figures

List of Tables

1. Introduction

A growing need exists for the integration of silicon microelectronics and advanced materials with specialized electrical and optical properties. The union of ferroelectric oxides with silicon is providing for the inception of a new generation of electronic and photonic devices. Furthermore, products in the developmental stage range from dynamic random access memories (DRAMs), to non-volatile random access memories (NVRAMs), to tunable microwave filters, and spatial light modulators for thin film displays.

Ferroelectric materials promise to fulfill the need for integration of high-speed electro-optic modulators with silicon for optical communication applications. The required performance for such devices includes: low optical loss, large depth of modulation, minimal driving voltage, and operation at microwave frequencies. Optical waveguides of Ti drifted LiNbO₃ form the most practical technology for meeting such performance criteria, but inherent limitations associated with embedded waveguides, and the desire for monolithic integration with silicon have led to a search for more suitable technologies. Increasing speed and efficiency may be achieved by forming a multilayered, thin-film waveguide directly on silicon. Thin film techniques and integration with silicon facilitate the growth of large active areas and leads to tight waveguide confinement and lower operating voltages. The variety of structures afforded by using thin films allow simple optimization of traveling-wave modulator properties, and modeling of device performance.

Actual performance, however, will depend on the intrinsic properties of the layers which can be much different than those of single crystals. Engineering of material properties, through precise control of thin film microstructure, will play the key role in ensuring the best device performance. The active layer must be formed from an epitaxial, ferroelectric thin film exhibiting a large electro-optic effect. This layer must be integrated with silicon using a low index-of-refraction buffer layer to maintain tight waveguide confinement. To achieve high efficiency, both layers must have low optical and microwave losses.

The first step in engineering this structure is to develop optimized methods for forming the ferroelectric and buffer layer films. The growth of KNbO₃ and BaTiO₃ on oxide single crystals using metal-organic chemical vapor deposition (MOCVD) has been particularly successful. Structural characterizations have revealed the high quality of the resulting epitaxial layers. Optical loss mechanisms in the ferroelectric thin films have been studied, and losses below 1 dB/cm are reported for several epitaxial systems. Static electro-optic measurements have been performed, and test devices operating up past 10 MHz have been fabricated.

Currently two main problems are preventing the realization of these high-speed, integrated electro-optic modulators. First, effective integration of epitaxial oxides with silicon has not been achieved. Second, the electro-optic coefficients of thin films are smaller than expected, and there are concerns about the response of the electro-optic effect at microwave frequencies. While the main body of this dissertation has covered investigations of the electro-optic effect in ferroelectric thin films, this supplement will focus on the integration of epitaxial oxides with Si using metal-organic molecular beam epitaxy..

2. Background

2.1 Oxide Buffer Layers

 
Aside from the growth technique, substrate choice is the most critical parameter in ensuring an epitaxial film. Materials are chosen to provide the proper structural match with the film (Table 2–1). Thermal expansion mismatch must be minimized to prevent cracking upon cooling from growth temperatures and to reduce unwanted strain effects in the film. To achieve epitaxial growth the crystal structure of, and lattice match between, the layers must be considered.

Substrates used for waveguide and electro-optic structures have further requirements. They must be optically transparent with a low surface roughness and posses an index of refraction lower than that of the waveguide layer. A low dielectric constant is desirable for use in electro-optic modulator structures to reduce dielectric losses. Several materials meet these criteria, among them, sapphire (Al₂O₃), periclase (MgO), and spinel (MgAl₂O₄) are the most common.

Typically, single crystals of these materials are chosen as substrates for thin film growth because they provide very smooth, defect-free surfaces. However, producing high-purity, single crystals of these materials is often costly, and it is difficult to grow crystals larger than a few centimeters in diameter, leading to difficulties in the scale-up of growth processes on single crystals.

One way to avoid these problems is to grow ferroelectric films directly on semiconductor substrates, which can be produced in large dimensions at low cost. This reasoning has prompted many researchers to explore the growth of ferroelectric materials directly on semiconductor wafers such as silicon, which is a natural selection due to its dominance in electronic applications. In most of these reports Table 2–2, the resulting films were not of high crystalline quality and epitaxy proved difficult, if not impossible, to obtain. Diffusion of atomic species, in particular oxygen, and formation of secondary phases such as SiO₂ at the interface lead to poor electronic and optical properties, and in some cases, to cracking and decohesion.

These problems could be minimized by growing a thin oxide buffer layer epitaxially on the silicon, prior to growth of the ferroelectric film. This intermediate layer would need to meet all of the criteria previously discussed while fulfilling several additional requirements. It should provide a lattice match that relaxes the strain formed at both interfaces. The ideal buffer layer must also be chemically inert, act as a thermally stable nucleation template for the active layer, and be grown thick enough to act as a suitable diffusion barrier.

Several researchers have taken this route to integration for both ferroelectric and superconducting thin film materials. Table 2–4 and Table 2–5 list several studies of the growth and formation of oxide buffer layers on silicon by MOCVD and MBE, respectively. These layers have been used successfully for epitaxial growth of ferroelectric thin films, as illustrated in Table 2–6. Furthermore, Hubbard and Schlom have used thermodynamic data to identify those binary oxides that are stable in contact silicon up to 1000^oC, to identify potential buffer materials.

Table 2–1 compares the structural properties of the most commonly used buffer layers with Si. MgAl₂O₄ has a lattice misfit of 0.8% with Si for the (100)||(100); [001]||[001] with a ratio of 3 Si to 2 spinel unit cells. The lattice mismatch for MgO on Si is 8.8% for the (100)||(100); [001]||[011] orientation. Despite the large disparity, MgO has shown a strong tendency to orient (100) on Si, and many other surfaces, perhaps because of the high stability of the {100} type faces. Experimental evidence (Table 2–4, Table 2–5, and Table 2–6) and thermodynamic calculations suggest these materials are stable in the presence of silicon.

2.2 MOMBEgrowth process

Growth techniques such as sputtering, PLD, MOCVD, and MBE are not well suited to the deposition of epitaxial MgO layers on Si for various reasons. This section will discuss the problems associated with these conventional deposition techniques. Second, the advantages of using the relatively unconventional technique of metal-organic MBE (MOMBE) to deposit the epitaxial MgO buffer layers will be discussed. Lastly, reports of other attempts to deposit oxide buffer layers using MOMBE will be reviewed.

The primary disadvantage of sputtering and pulsed laser deposition (PLD) is the high energy of the depositing species, which promote interface mixing, and damage to surface layers, leading to alloying and micrograin layers. The main advantage of MBE lies in superior compositional and thickness control which provide for greater uniformity of film properties and the command of film thicknesses to less than a monolayer. The low vapor pressures of the metal sources used in MBE require significant heating, achieved by electron-beam heating or laser ablation, which can cause oxidation at the surface leading to unstable fluxes. Source temperatures in MOCVD are much lower due to the higher volatility of the metal-organic precursors. However, the presence of oxygen in the carrier gas flow can lead to the premature formation of an amorphous SiO₂ surface layer, interfering with the formation of an epitaxial film.

Researchers investigating the growth of compound semiconductors have combined MOCVD and molecular beam epitaxy (MBE) into a new growth system referred to as metal-organic MBE (MOMBE). In MOMBE the sources are metal-organic compounds, as with MOCVD. This allows for lower source temperatures, and higher, more stable fluxes than conventional MBE. Unlike MOCVD, MOMBE growth takes place in the molecular flow regime (< 10^-4 Torr). Although the pressures inside a MOMBE chamber (10^-6 to 10^-4 Torr) are higher than those in MBE (<10^-9 Torr), in situ film monitoring techniques are still available.

MOMBE was first used for the production of oxides by investigators attempting to grow superconducting films of YBa₂Cu₃O_7-x on single crystalline MgO (Table 2–7). Epitaxial films were grown using MOMBE at higher source fluxes and deposition rates than with MBE, and at lower substrate temperatures than MOCVD. Background chamber pressures were typically in the range 10^-7 to 10^-5 Torr and oxygen source pressures, ~10^-5 to 10^-4 Torr. Metal-organic precursors were evaporated in low temperature Knudsen cells or introduced into the growth chamber through gas ports. Higher growth rates and lower deposition rates were achieved by using an RF plasma O₂ source. Bade concluded that neither dry nor wet O₂ were sufficient for reasonable growth rates and the presence of an activated oxygen species is essential to MOMBE growth. The most significant problem with MOMBE of superconducting films is the metal-organic precursors which appear to be better suited to MOCVD. A very narrow temperature range is required to maintain a proper flux. The gas phase composition of some precursors were found to be insufficiently stable. Basic data regarding volatility for many precursor materials do not exist.

MOMBE growth of buffer layers on silicon has attracted recent attention (Table 2–7). The growth of binary materials, as compared to the more complex structures of high-T_c oxides, is observed to greatly simplify deposition using MOMBE. Growth pressures and temperatures are similar to those used for superconducting oxide films by MOMBE. Both effusion cells and gas sources were used for the various studies. Ikegawa concluded that growth conditions were similar for both methods. Oxygen sources also varied from NO₂, molecular O₂, and plasma activated O₂, to no separate oxygen source aside from the metal-organic ligand itself. The suitability of each O₂ source depends strongly on the particular ligand chosen for the metal-organic source.

This review of the literature suggests that the MOMBE deposition technique should be well suited to the growth of epitaxial MgO on Si. To date there have been no reports on the deposition of MgO using the MOMBE technique. In Chapter 7, the successful deposition of epitaxial MgO on Si substrates using MOMBE will be presented.

 

Table 2–1: Structural, dielectric, and optical properties of Si, GaAs, and common substrates for ferroelectric thin films.
 

Substrate	Anion Packing - Structure	Lattice Parameter (Å)	Thermal Expansion Coeff. (10-6/K)		Dielectric Constant		Index of Refraction
MgO	FCC - Rocksalt	4.211	10.6	9.65		1.74
a -Al2O3	HCP - Corundum	a = 2.75 c = 4.33	^ c = 8.3 || c = 9.0	e 11 = 9.34 e 33 = 11.54		na = 1.761 nc = 1.769
g -Al2O3	Psuedo-FCC - Hausmannite	a = 7.95 c = 7.79	---	e 11 = 9.34 e 33 = 11.54		na = 1.761 nc = 1.769
MgAl2O4	FCC - Spinel	8.08	6.97	8.6		1.72
Si	FCC - Diamond	5.431	2.6	11.9		--
GaAs	FCC - Zincblende	5.65	6.86	13.1		--

Table 2–2: BaTiO₃ and other ferroelectric films grown directly on Si substrates.
 

Ferroelectric Film	Substrate	Growth Method	Reference
BaTiO3	Si (100)	MOCVD	35,36
BaTiO3	Si (100)	UV-CVD	37
BaTiO3	Si (100) and (111)	MO decomposition	36,38
BaTiO3	Si (100)	Sputtering	39,40
SrTiO3 (100)	Si (100)	MBE	23
SrTiO3	Si (100)	MOCVD	41
LiNbO3	SiO2 / Si (100)	Sputtering	42
PbTiO3	Si(111)	Sputtering	43
PbxLa1-xZryTi1-yO3	Si (100)	PLD	44

Table 2–3: MgO and other oxide buffer layers grown on Si substrates.
 

Oxide Buffer Layer and Orientation	Substrate	Growth Method	Growth Temp (oC)	Ref.
MgO (100) and (111)	Si (100)	MOCVD	650 - 740
MgO (100)	Si (100)	MOCVD	500
MgO (100) and (111)	Si (111)	MOCVD	600 - 900
MgO (100)	Si (100)	MOCVD	600
MgO/BaO (100)	Si (100)	MBE	500
MgO (100)	Si (100)	PLD	650
MgO (100)	Si (100)	PLD	300
MgO (100)	Si (100)	PLD	300 - 600
MgO (100)	Si (100)	Sputtering	700
MgO (100)and (111)	Si (100)	Sputtering	250 – 700
MgO (100)	Si (100)	Sputtering	300
MgO (100) and (111)	Si (100)	Evaporation	500
g -Al2O3 (100)	Si (100)	CVD	950-1050
MgAl2O3 (100)	Si (100)	CVD	850 - 1100
BaO (100)	Si (100)	MBE	20 - 800
CeO2(111)	Si (111)	Evaporation	300
CeO2(110)	Si (100)	MBE	600 - 700
CeO2(110), (100)	Si (100), (110)	PLD	600 - 800
(Y2O3)m(ZrO2)1-m (100)	Si (100)	Evaporation	800

Table 2–4: MgO grown by MOCVD methods.
 

Substrate	Growth Temp (oC)	Oxidant	Mg Precursor	MgO Structure	Ref.
Si (100)	650 - 740	O2	Mg(thd)2	(100)-T
Si (100)	500	O2	Mg(acac)2	(100)-T
Si (111)	600 - 900	O2	Mg(C5H5)2	(100)-T
Si (100)	600	O2	Mg(acac)2	(100)-E
SiO2	400	O2	Mg(thd)2	(100)-T
SiO2	740	O2	Mg(thd)2	(100)-T
SiO2	540	O2	Mg(acac)2	(100)-T
SiO2	680	RF-O2	Mg(acac)2	(100)-E
SrTiO3 (100)	700	O2	Mg(thd)2	(100)-T

E = Epitaxial; T = Textured; A = Amorphous; P = Polycrystalline  

Table 2–5: MgO grown by MBE methods.
 

Substrate	Growth Temp (oC)	Oxidant	Mg Precursor	MgO Structure	Ref.
Si (100)	500	e-beam evap. MgO		(100)-E
BaO/Si (100)	200-600	e-beam evap. MgO		(100)-E
MgO (100)	750	ECR O2	Mg	(100)-E
MgO (100)	650	O2	Mg	(100)-E

E = Epitaxial; T = Textured; A = Amorphous; P = Polycrystalline

 

Table 2–6: Ferroelectric films grown on buffered Si and GaAs substrates.
 

Ferroelectric Film	Buffer Layer	Substrate	Growth Method	Ref.
BaTiO3 (100)	BaO	Si(100)	MBE
BaTiO3 (100)	MgO (100)	Si (100)	MBE
BaTiO3 (100) and (001)	MgO (100)	Si (100)	Sputtering
BaTiO3 (100)	MgO (100)	Si (100)	PLD
BaTiO3 (100)	MgAl2O4 (100)	Si(100)	CVD
BaTiO3 and SrTiO3 (100)	MgAl2O4 (100)	Si (100)	Sputtering
BaTiO3 (100)	TiN (100)	Si(100)	PLD
BaTiO3 (100)	TiO2 (100)	Si(100)	PLD
BaxSr1-xTiO3 (100)	CaF2 (100)	Si (100)	PLD
SrxBa1-xNb2O6	MgO (111)	Si (111)	PLD
BaNaNb5O15 (100)	MgO (100)	Si (100)	PLD
Bi4Ti3O12 (100)	ZrO2 (100)	SiO2 / Si (100)	PLD
PbxLa1-xZryTi1-yO3 (100)	MgAl2O4 (100)	Si (100)	Sputtering
PbZrxTi1-xO3 (100)	MgO (100)	Si (100) and GaAs (100)	PLD
BaTiO3 (100)	MgO (100)	GaAs (100)	PLD

Table 2–7: Oxide films grown by MOMBE.
 

Oxide Layer	Substrate	Growth Temp. (oC)	Ref.
PbTiO3	MgO (100)	300-500
g -Al2O3 (100)	Si (100)	700-850
g -Al2O3 (100)	Si (100)	650-850
g -Al2O3 (100) or (111)	Si (100) or (111)	450-700
CeO2 (110) or (111)	SrTiO3 (001) or (110)	350-800
CuO (111) and (202)	MgO (100)	500-750
YBa2Cu3O7 and CuO	MgO (100) and SrTiO3	650
YBa2Cu3O7	MgO (100)	550
(Y, Dy)Ba2Cu3O7-x	MgO (100)	700

 
 

 3. Experimental

3.1 MOMBE apparatus

Magnesium oxide films were deposited on silicon in an SVT Associates SN35 MBE system. A schematic of the MOMBE chamber is shown Figure 3–1 and photographs from two angles are shown in Figure 3–2. The chamber consists of four main sub-systems: the vacuum system, the effusion sources, the RF oxygen source, and the process monitoring system. Each of these sub-systems will be discussed in detail in the following section.
 

3.1.1 Vacuum system

The main growth chamber of the SN35 system is evacuated using a Pfieffer Vacuum TMU1000 turbomolecular pump backed by a MD4T diaphragm pump. The chamber also contains a system of cryopanels, which are filled with liquid nitrogen during film growth. The background pressure in the chamber before filling the cryopanels is typically 10^-7 to 10^-8 Torr. The pressure after fully cooling the panels with liquid nitrogen is approximately 10^-10 Torr. Prior to introduction of the metal-organic source materials, the best vacuum achieved by the system was 10^-11 Torr.

3.1.2 Effusion sources and metal-organic precursor

The growth chamber is equipped with low temperature effusion cells which are capable of operation from room temperature to 800°C. The temperature of the effusion cells is controlled and monitored by a type K thermocouple in conjunction with a Eurotherm model 2408 temperature controller. The control circuit for the sources is capable of maintaining the temperature within 0.1^oC.

The effusion cells were loaded with the metal-organic precursor magnesium acetylacetonate, Mg(acac)₂, supplied by Alfa Aesar (product #12532) with a purity of 98% (metals basis). This material is a solid white powder at room temperature with a melting point of 260^oC. After loading the metal-organic source into the vacuum chamber, the source material is slowly ramped to 120 to 150^oC for a bake out lasting several hours. This source bake provides a stable metal-organic flux by removing water vapor and other volatile contaminants from the material before using it as a source for film growth.

3.1.3 Oxygen plasma source

The oxygen plasma for MgO film growth is an SVTA model RF4.5 RF plasma source. The RF plasma source supplied activated oxygen to the substrate for oxide growth. The RF power supply is an Advanced Energy model RFX600 with an excitation frequency of 13.56 MHz and a variable power up to 600 W. Ultrahigh purity molecular oxygen (Air Gas, 99.90% purity) is supplied to the plasma source through a variable leak valve (Part #315010, MDC Vacuum Products Corp.). This valve was used to control the pressure of oxygen entering the vacuum chamber.

3.1.4 Process monitoring system

Source fluxes were monitored using an ion gauge with a thorium coated iridium filament. This ion gauge can be manipulated between the source and the substrate to read the flux directly at the surface of the substrate. Substrate temperatures were measured and controlled by a C type thermocouple in conjunction with a Eurotherm model 2408 temperature controller. The temperature of the substrate can be maintained within ~ 5^oC. The thermocouple was calibrated with an optical pyrometer to ensure accurate temperature measurements. All chamber temperatures and pressures were monitored and recorded on a PC using the Microsoft Quick Basic 4.5 program listed in Appendix A. Reflection high energy electron diffraction (RHEED) patterns were recorded during the growth using an SVT model RH-10KV RHEED gun and a k-Space Associates KSA-100 system. An SRS model RGA300 residual gas analyzer was used to measure the atomic mass spectra of the vacuum atmosphere during film growth.

3.2 Substrate selection and preparation

Substrates for the MgO film deposition were 3 inch (001) cut Si wafers (International Wafer Service) doped with nitrogen to a resistivity of 1-10 W -cm. The substrates are etched in a solution of HNO₃:HF:H₂O (5:3:92 by volume) for several seconds to remove the native oxide. The wafers are then rinsed in methanol and acetone for 15 minutes each before being placed in the system load-lock. The wafer is then transferred to the main growth chamber, which was previously pumped to a base pressure of £  10^-9 Torr. The Si is degassed at 200^oC for 30 minutes before being ramped to 850^oC for desorption of any remaining SiO₂. After 30 minutes of desorption, the Si surface showed a clear and streaky RHEED pattern indicating an atomically smooth surface is achieved.

3.3 Growth of b-SiC template layer

The b-SiC template layer is grown using the Mg(acac)₂ precursor as a carbon source. The MO source is heated to a temperature (~160^oC) resulting in a flux of ~1x10^-7 Torr as the substrate is ramped to 900^oC at a rate of 40^oC/min. After equilibration of the MO source and Si substrate temperatures, the shutters over the Si substrate and the MO source are opened and growth of the b-SiC proceeds through a simple carbonization reaction. Table 3–1 lists typical deposition parameters for the b-SiC films; details of the growth will be discussed in Chapter 4.

3.4 Growth of MgO overlayer

After a b-SiC film of the desired thickness has been deposited, the growth of the MgO over-layer begins. For growth of the MgO over-layer the source shutters are closed, the temperature of the substrate is reduced to ~800^oC, and the MO temperature is raised to obtain a metal-organic flux of ~1x10^-6 Torr. The shutters are then opened and the oxygen plasma source is ignited at a pressure of ~3x10^-5 Torr of O₂ and an RF power of ~400 W. After the desired layer thickness is achieved, the sample is cooled in vacuum and removed from the chamber for subsequent analysis. Table 3–1 lists typical deposition parameters for the MgO films; details of the MgO growth will be discussed in Chapter 4.

3.4 Structural characterization of thin films

In addition to in-situ RHEED analysis, the deposited films were compositionally and structurally analyzed using a variety of ex-situ techniques. Thickness of the films was determined by using a Tencor P-10 surface profiler to measure the height of a step created by masking a portion of the substrate surface with a small chip of Si. Surface morphology and roughness were examined with a Nanoscope III atomic force microscope (AFM) using a Si₃ N₄ tip. The film composition was determined by a Physical Electronics scanning Auger microscope (SAM). Fourier transform infrared (FTIR) spectroscopy ( Bio-Rad FTS-60)) was used to identify the SiC layer. Film microstructure was investigated with a Matrix x-ray diffractometer (XRD) using LiF filtered CuK_a radiation, an Hitachi HF-8100 conventional transmission electron microscope (TEM), and an Hitachi 2000 cold field emission, high resolution TEM operated at 200 keV. TEM samples were prepared by mechanical grinding and polishing followed by further mechanical dimpling to a thickness of less than 10 mm and final thinning in a Gatan Model 691 precision ion polishing system.  

 

Parameter	SiC Interlayer	MgO Overlayer
Growth Temperature (° C)	800 to 950	700 to 900
O2 Plasma Pressure (Torr)	--	1.0 ´ 10-5
O2 Plasma Power (W)	--	300 to 400
Metal-organic Pressure (Torr)	1 ´ 10-9 to 1 ´ 10-7	1.0 ´ 10-6
Growth Time (hours)	< 1	3 to 5
Film Thickness (nm)	1 to 200	20 to 100
Growth Rate (nm/hour)	10 to 300	10 to 30

 

4. Integration of epitaxial oxide thin films with Si

In the preceding two chapters, the EO properties of FE thin films have been investigated. Integration of these materials with the electrical properties of semiconductors would increase the utility of FE films as the active layer in EO device structures. As silicon is the most widely accepted platform for electrical device production, and is widely available in large diameter wafers, it is a natural choice as a potential integration platform for FE thin films.

The majority of FE oxides have been observed to chemically react with silicon forming silicon dioxide or metal silicates precluding their epitaxial growth on Si. One approach to obtaining epitaxial layers of these reactive oxides on silicon is to first deposit a stable epitaxial oxide buffer, or template layer prior to their deposition. A buffer layer is also necessary to facilitate the production of waveguide EO modulators on the integrated FE thin films. This buffer layer has four main requirements. First, it is necessary to optically isolate the waveguides in the FE film from the semiconductor. This would require a material with a lower index of refraction than the FE film and sufficient thickness to prevent the leakage of light from the waveguide into the Si. Second, the buffer layer must be thick enough to electrically isolate the FE material from the Si substrate. This would require a thickness on the order of the separation of the surface electrodes of the EO modulator. Third the buffer layer must be deposited epitaxially on the Si substrate to promote the overgrowth of high-crystalline-quality FE films. Fourth, the buffer layer must have a low surface roughness to prevent the scattering of light from the waveguide.

The production of a stable, highly uniform, epitaxial layer of (100)MgO on a (100)Si substrate meets these requirements and would facilitate the integration of FE films. MgO is the material of choice as a buffer layer as epitaxial thin films of BaTiO₃ have already been deposited on (100)MgO single crystals using MOCVD. Additionally the substrate has a lower index of refraction than the BaTiO₃ films, facilitating thin film waveguide fabrication.

The immediate goals of this work are to produce high-structural-quality, epitaxial MgO thin films on Si substrates using the MOMBE deposition technique. The following section will discuss the key issues involved in producing epitaxial buffer layers of MgO on Si using MOMBE. Later sections will discuss the growth of a b-SiC nucleation layer on Si, the deposition of the MgO buffer layer, and the initial attempt to produce integrated, epitaxial BaTiO₃ active layers using the MOMBE grown MgO layers.

4.1 Key Issues

In this section the key issues surrounding the deposition of epitaxial MgO films on Si will be discussed. First, the difficulties surrounding the direct integration of MgO with Si will be reviewed. As will be explained, the simplest way to avoid these problems is to deposit a thin cubic SiC (b-SiC) nucleation layer on the Si prior to MgO growth. As such, the deposition of b-SiC using MOMBE will be discussed next, followed by an overview of MgO deposition using MOMBE.

While MgO has been predicted to be thermodynamically stable on the silicon surface, its growth on Si using many different techniques has been hampered by the unintentional formation of an amorphous interlayer of SiO₂. This amorphous layer leads to the growth of polycrystalline films. Even if formation of the SiO₂ layer can be avoided, polycrystalline MgO films often result from the large lattice mismatch between Si and MgO (8.8 to 29%).

To overcome these problems, previous work has focused on the growth of thin, epitaxial buffer layers on Si that are stable in the presence of oxygen and relax the lattice match between MgO and Si. McKee and Walker have focused on the growth of cubic silicides, such as BaSi₂ and SrSi₂, using molecular beam epitaxy (MBE) with electron-beam heated, elemental sources. Despite the inherent technical difficulties associated with using electron-beam heating in an oxygen ambient, they have successfully used these silicide layers to form epitaxial oxide layers on silicon substrates. However, the silicide layer only remains commensurate with the Si surface up to one monolayer of thickness. A silicide film thicker than one monolayer results in an orthorhombic structure, an unsuitable template for MgO formation. Therefore this technique is only feasible with sub-monolayer control of the interlayer deposition.

A possible alternative template layer is cubic SiC (b -SiC or 3C-SiC). Cubic (001)SiC is better lattice matched to (001)MgO (3.3%) for the <100>||<100> orientation. Figure 4-1 shows the unit cells of the three layers in the proposed heterostructure along three different axes. The increased lattice match between the b-SiC and MgO layers is evident from the scaled drawing of the three materials. The b-SiC surface also resists oxidation at the temperatures, < 1000°C, used to grow MgO. This should prevent the formation of amorphous SiO₂ which has been shown to preclude epitaxial growth of MgO on Si.

While the thermal expansion (~8%) and the lattice (20%) mismatches between (001)Si and (001)b-SiC are high, growth of epitaxial b-SiC films on single crystal Si wafers has been widely reported. The simplest method for the growth of a thin b-SiC nucleation layer is through a Si carburization process using either a solid carbon or hydrocarbon gas source such as CH₄, C₂H₂, C₂H₄, C₃H₈. Deposition techniques have ranged from PLD and sputtering to CVD and gas source MBE. Temperatures as low as 600- 800°C have been used for successful growth of epitaxial b-SiC through such a carburization process

The carburization of the Si surface could be accomplished using the Mg(acac)₂ precursor used for the growth of MgO thin films by the MOMBE technique described in Section 2.4. A ball-and-stick model of the Mg(acac)₂ molecule is shown in Figure 4-2. The molecule consists of a central Mg ion surrounded by two acetylacetonate ligands which contain five carbon atoms each. The acetylacetonate ligand is know to fractionate into a variety of C, H, and O containing molecules and radicals:



These hydrocarbon intermediate products, especially the CH_x radicals, could act as a precursor for the carburization of the Si surface thereby forming b-SiC. The overall reaction between Si and the Mg(acac)₂ would proceed as:

Mg(acac)₂ (gas) + Si (solid) à b-SiC (solid) + RH (gas)

where the RH represents the remainder of the hydrocarbon products. Residual O, H, or Mg impurities may be incorporated into the deposited films producing effects on optical and electrical properties. As the b-SiC layer is only used as a structural integration layer only effects on the crystal structure are of practical importance. The incorporation of Mg and H in the film structure is expected to have a negligible effect on the b-SiC structure. The incorporation of oxygen in large amounts may be expected to form a SiC_xO_y structure; however, the Si-C-O phase diagram suggests that SiC can incorporate large amounts of oxygen without formation of a new crystal structure.

The growth mechanism of b-SiC on Si has been widely investigated. As opposed to the simple carburization process observed in metals such as iron, the Si(001) surface does not carburize through a simple diffusion reaction. The process takes place entirely at the Si surface because of the low diffusivity of C in the Si lattice, the high self-diffusion energy of Si (~5 eV), and the low activation energy of Si(001) along the dimer rows (0.6 eV at 750-900°C). The carbon atoms impinging on the Si surface form b-SiC in isolated islands. Three dimensional growth is favored over 2-dimensional growth because of the large lattice and thermal expansion mismatch between the materials and the high surface energy of b-SiC. The islands grow by the diffusion of Si around the b-SiC nuclei. Eventually after the islands coalesce and seal off the Si surface, the Si migrates to the growing b-SiC surface by diffusion of Si through extended defects in the b-SiC lattice. The evidence of this process is the production of large voids beneath the b-Si/Si interface. The pyramid-shaped voids form with edges parallel to Si(110) and faces along Si(111) as Si vacancies congregate underneath the b-SiC surface.

After formation of the b-SiC/Si(001) nucleation layer, the MgO buffer layer must be deposited. A variety of deposition techniques have been used to grow MgO thin films on Si including MOCVD, MBE, PLD, sputtering, and sol-gel (Table 2–4 and Table 2–5). The most successful technique for depositing epitaxial MgO/Si to date has required using electron-beam evaporation of MgO onto Si in an oxygen ambient. The inherent mechanical difficulties associated with using electron-beam sources in an oxygen atmosphere has limited the success of this approach, especially for large area coverage. The technique also requires atomistic control of the amount and type of atomic species deposited to enable epitaxy. This complication has further limited the success of this approach.

While a variety of oxides such as PbTiO₃, g -Al₂O₃, CeO₂, CuO, and YBa₂Cu₃O₇ have been deposited using the MOMBE growth technique (Table 2–7) there have been no reports of MgO deposition by MOMBE. As described in Section 2.4, this technique combines the advantages of both MOCVD (high-flux, metal-organic sources) and conventional MBE (low background O2 pressure and in-situ film characterization). The technique is also potentially scalable to large wafers.

If the deposition of epitaxial MgO on Si(100) is possible by MOMBE, the conditions should be similar to those of both MBE and MOCVD (Table 2–4 and Table 2–5). Mg(acac)₂ has been used in MOCVD to successfully deposit MgO/Si, and should act as a suitable precursor for the MOMBE technique. The molecular structure of Mg(acac)₂ is shown in Figure 4-2. It contains a central Mg ion bonded directly to the four O ions of the acetylacentonate ligands.

MBE deposition of MgO using molecular O₂ at 10^-6 Torr typically suffers from low incorporation of O₂, ~0.1%. Similar results were obtained for MgO growth using PLD with oxygen pressures from 5 x 10^-6 Torr to 10^-4 Torr. Higher incorporation of oxygen can be achieved by using an activated oxygen source which possesses a higher reactivity than molecular O₂. Such an increase in reactivity was observed in the homoepitaxy of MgO by MBE using an ECR plasma at 10^-5 to 10^-6 Torr of O₂. Similarly an increase in deposition rate of MgO was observed using an RF O₂ plasma at powers ranging from 200 to 400 W as the oxidant in MOCVD. Typical growth temperatures used for the MOCVD growth of MgO range from 400 to 900°C (Table 2–4). For MBE the temperature ranged from 650 to 750°C, excluding those reports using e-beam evaporation sources (Table 2–5).

Using the deposition conditions of MgO by MBE and MOCVD as references, the growth of MgO/b-SiC/Si by MOMBE should occur between 400 and 900°C with activated O pressures ranging from 10^-6 to 10^-4 Torr. The MO precursor Mg(acac)₂ should react with the RF-plasma activated O to form MgO through a simple oxidation reaction:

Mg(acac)₂ (gas) + O^* (gas) à MgO (solid) + H₂O (gas) +CO_x (gas)

Carbon and hydrogen impurities are removed by the activated oxygen, O*, to form water, carbon dioxide, and carbon monoxide. The b-SiC layer is expected to remain stable at 400 to 900°C in an atmosphere of O₂ up to 10^-4 Torr as a result of the strength of the Si-C bond.

In the remainder of this chapter the deposition of epitaxial (100)MgO on (100)Si using the MOMBE deposition technique will be reported. These investigations will focus on the production of high-crystalline-quality, epitaxial MgO films with a high growth rate and low surface roughness. First, deposition of the b-SiC nucleation layer will be discussed, followed by the deposition of the MgO overlayer. Lastly, the integration of MOCVD grown BaTiO₃ with Si using the MOMBE grown MgO buffer layers will be presented.

4.3 Growth of b-SiC template layer using MOMBE

This section will discuss the deposition of the b-SiC interlayer using the MOMBE growth technique. The first sub-section will detail the growth procedure and its development. The second sub-section will discuss the results of chemical and structural analysis of the films. The last section will examine the influence of growth conditions on the structural perfection of the layers.

4.2.1 Development of the MOMBE growth procedure

Development of the growth procedure for b-SiC/Si(001) followed several steps: obtaining a clean Si(001) surface, finding an epitaxial growth regime, and improving the growth procedure. The details of these steps will be discussed in the remainder of this section.

Removal of the native oxide from the Si surface was accomplished using a brief chemical etch and a simple desorbtion in UHV. Three inch Si wafers were cleaved into 2 cm square substrates before cleaning. The native oxide was etched for 5 s in a solution of HNO₃:HF:H₂O (5:3:92 by volume). After removal from the etch solution the Si was rinsed in acetone and ultrasonically agitated for 15 min. After removal from the acetone, the sample was rinsed in methanol and ultrasonically agitated for 15 min. The sample was blown dry with a nitrogen jet and placed in the MBE sample holder. The mounted sample was placed in the MBE through the system load-lock and transferred to the MBE stage. The chamber was cooled with liquid nitrogen after which the vacuum typically reached 1 x 10^-9 Torr.

After the vacuum had reached an equilibrium pressure, the stage temperature was raised to 200°C for 20 to 30 minutes to degas water from the stage. A schematic of the heating conditions used for substrate cleaning is illustrated in Figure 4-3. A plot of the stage temperature and chamber pressure recorded during the growth of a b-SiC/Si(001) film (BHH087) is shown in Figure 4-4. As the stage is heated, and water vapor is removed, the vacuum pressure increases to 2 x 10^-8 Torr. After the desorption is complete the vacuum recovers to ~ 10^-9 Torr.

After the 250°C degas, the stage is ramped at 30°C/min. to 850°C for desorption of any remaining SiO₂. A schematic of the heating conditions used for desorption is also illustrated in Figure 4-3. Figure 4-4 shows the increase in the chamber pressure recorded as the stage is ramped to 850°C. Initial RGA analysis suggests that the majority of this flux can be attributed to H₂. As the Si is held at 850°C, the vacuum recovers to ~10^-7 to 10^-8 Torr. RHEED analysis shows that after ~20 min. at 850°C the Si(001) surface becomes clear of any SiO₂. The RHEED pattern becomes clear and streaky with Kikuchi lines, indicating the Si surface was atomically clean. Figure 4-5 is a RHEED image of the cleaned Si(001) surface obtained by this simple procedure.

Initial attempts to grow b-SiC focused on finding an epitaxial growth regime with a Mg(acac)₂ flux between 10^-7 and 10^-5 Torr and a Si temperature from 800°C to 900°C. The growth of b-SiC through a simple carbonization reaction was reported using gas-source MBE under similar temperatures, 750 to 900°C, and pressures, 10^-7to 10^-5 Torr, of C₂H₂. The initial growth procedure consisted of heating the MO source until the proper flux was obtained. During this time the Si was held at 850°C. The MO source temperatures were typically ~ 160°C, but the source did not always produce the same flux at a given temperature. Over time the temperature needed to reach a specific flux had to be increased, typically by a few degrees from one deposition to the next. This instability is thought to stem from impurities in the MO source material. To correct for this, the MO flux had to be checked before the growth was initiated. The flux was checked by heating the source to a given temperature and measuring the resulting pressure. The flux was measured by lowering the flux monitor between the MO source and the stage (time t₁ in Figure 4-4). The MO flux was measured as the change in flux monitor pressure between readings taken with the source shutter in the closed and open positions (time t₂, t₃, and t₄ in Figure 4-4). With the stage at ~900°C the lowest change in pressure detectable was ~10^-7 Torr. After the required MO flux was attained, the flux monitor was raised, and the stage was ramped to the growth temperature. After equilibration of the stage temperature, the stage and source shutters were opened simultaneously to start the growth. The RHEED diffraction pattern was monitored during the growth to probe the film structure.

Initial experiments did not find an epitaxial growth regime in the initial parameter space. Growths at all combinations of MO flux and Si temperatures resulted in amorphous RHEED patterns. The first observation of an epitaxial b-SiC film occurred while incorporating a "pre-deposition" period of 30 min. The pre-deposition step was added after stabilizing the MO flux at 10^-7 Torr and ramping the stage to a growth temperature of 900°C. During the pre-deposition step the stage shutter was opened while the MO shutter remained closed. During this 30 min. period the RHEED pattern showed a slow transformation from the Si(001) diffraction pattern to that of b-SiC, as shown in Figure 4-6. The diffraction patterns observed are similar to those reported for other b-SiC films. After 30 min. the MO shutter was opened and growth of b-SiC continued without further changes to the RHEED pattern. Without the pre-deposition step, an amorphous film resulted from growth under the these conditions. Epitaxial b-SiC films resulting from this initial growth procedure are listed in Table 4–2. The dependence of the growth rate on deposition conditions for these films will be discussed in Section 4.2.3.

While this initial growth procedure resulted in epitaxial b-SiC films, several modifications to the procedure were made to improve reproducibility. First, a cooling ramp to 250°C was added after the desorbtion of the SiO₂ at 850°C. This step is shown in the heating schematic of Figure 4-3 and the data recorded in Figure 4-4. The addition of this step had two benefits. The background pressure in the chamber was lowered to ~10^-9 Torr, thus improving the lowest MO flux measurable to ~10^-9 Torr. Additionally, the lower Si temperature ensures that growth of b-SiC will not occur during the next step, ramping the MO source to the proper temperature and measuring the resulting flux. A cooling step is consistently used in the CVD growth of b-SiC. The second modification to the procedure was removal of the pre-deposition step. Instead of using a closed source shutter to reduce the MO flux during nucleation of the b-SiC layer, a lower (< 10^-7) MO flux was used. A lower MO flux could only be used after addition of the cooling step reduced the background pressure in the chamber, facilitating the measurement of lower fluxes.

Following the SiO₂ desorbtion, there are seven steps in the growth of b-SiC/Si(100) through this improved method:
 

1. The Si is cooled from 850°C to 250°C at 30°C/min.
2. The MO source is ramped at 10°C/min. to the approximate temperature needed to obtain the desired MO flux.
3. The flux monitor is lower between the MO source and stage, and the MO flux is measured.
4. If the flux is insufficient/excessive, the MO source temperature is raised/lowered, then step (3) is repeated.
5. When the desired MO flux is obtained, the stage is ramped to the growth temperature at 40°C/min.
6. When the growth temperature is reached and the temperature is stable, both the stage and MO source shutters are opened.
7. After the desired growth time is complete, all of the shutters are closed, and the stage and MO sources are ramped to 20°C at 40°C/min.
This procedure was used to grow a series of b-SiC films for examination of the conditions under which epitaxial growth is produced. Films of b-SiC grown using this procedure are listed in Table 4–1. Their chemical and structural properties will be discussed in the following section. The dependence of film growth rate and structural perfection on deposition conditions will be examined in Section 4.2.3.

4.2.2 Structural and chemical analysis

In addition to in-situ RHEED analysis, the deposited films were compositionally and structurally analyzed using a variety of ex-situ techniques. The composition of the films was examined by Auger electron spectroscopy (AES). Fourier transform infrared (FTIR) spectroscopy was used to identify the SiC layer. Surface morphology and roughness were examined with atomic force microscopy (AFM). Film microstructure was investigated using transmission electron microscopy (TEM). The results of in-situ and ex-situ chemical and structural characterizations will be presented in this chapter.

Under favorable growth conditions, sharp characteristic single crystal RHEED patterns were observed as shown in Figure 4-6 along the Si <100> azimuth and for the Si <110> azimuth. The diffraction patterns observed are similar to those reported for epitaxial b-SiC films. Thus the RHEED patterns suggest the films are epitaxial.

During the initial stages of film growth the RHEED pattern showed a slow transformation from the Si(001) diffraction pattern to that of b-SiC, as shown in Figure 4-7 for a growth at 950°C (BHH087) and in Figure 4-8 for a film grown at 850°C (BHH079). At 950°C the transformation was complete after ~4 min. At 850°C the transformation was complete after ~ 50 min. For times earlier than these the RHEED patterns show spots characteristic of both the Si(001) and the b-SiC surfaces. The slow appearance of the b-SiC diffraction spots is an indication of the 3-dimentional growth mechanism discussed in Section 4.1. The first b-SiC spots appear as the first nuclei form. The Si pattern disappears once the nuclei have completely coalesced and covered the surface. The faster disappearance of the Si diffraction spots at higher temperatures suggests a faster growth rate. After the Si(001) spots had completely faded, no changes in the RHEED patterns were observed.

Fourier-transform infrared (FTIR) measurements of the carbonized Si surface layer confirmed the presence of b-SiC. Figure 4-9 shows an FTIR transmittance spectra from a 200 nm thick interlayer grown on Si(001). A well resolved absorption peak at 798.6 cm^-1 was observed. This is in good agreement with the b-SiC transverse optical (TO) mode at 797 cm^-1. No peaks were observed in the spectrum at either 1060 or 460 cm^-1 corresponding to the stretching and bending modes of the Si-O bond.

AES results reveal the presence of Si, C, O, and Mg in the b-SiC films. Figure 4-10 is a series of AES energy spectra taken before and after Ar ion sputtering for a 15 nm thick b-SiC film (BHH063). Peaks at 263, 503, and 1184 eV correspond to the C, O, and Mg K_LL transitions respectively. The peaks at 89 and 1640 eV are a result of the L_VV and K_LL transitions of the Si atom. The positions and relative intensities of the Si and C AES peaks are similar to those observed for other b-SiC films. Typically the unsputtered surface showed only trace amounts of Si and Mg, and consists mainly of C and O. After 20 minutes of sputtering with Ar ions, the AES spectrum contains the same peaks, with the addition of a small L_VV Ar peak at 211 eV. The Si peak becomes more pronounced, but the O peak remained intense. This suggests either the presence of large quantities of O in the interior of the b-SiC film, or the presence of H₂O inside the vacuum chamber in which the AES was preformed. This would lead to the continual recontamination of the b-SiC surface with O. If O is present in the interior of the film it results from the cracking of the MO precursor and most likely leads to the formation of SiO₂ in the films. The presence of SiO₂ phase was not detected in TEM analysis, and FTIR results did not detect the presence of the Si-O bond. Phase diagrams of the Si-C-O system suggest that SiO₂ and SiC can exist in equilibrium up to large concentrations of O (at high pressures, ³  0.01 atm, with temperatures ~1700 K) although no information is available for very low partial pressures of oxygen. It is also conceivable that the films contain large amounts of O incorporated into the b-SiC lattice. The growth of SiC_xO_1-x, x = 0.56, films on Si by RF sputtering has been reported, although the resulting films were amorphous.

The surface morphology of the SiC films was determined by AFM. The films consisted of a dense rectangular morphology as shown in Figure 4-11. The film shown in Figure 4-11 is a 60 nm thick b-SiC film (BHH079) grown on Si(001). Shown in the figure is a 10 mm square section of the film showing the elevation data (left) and the corresponding deflection data (right). The block-like features in the image are columnar subgrains with diameters ranging from 10 to 100 nm. The RMS roughness of the film in Figure 4-11 was 11 nm. The typical RMS surface roughness of the SiC films in this study was ~1 to 20 nm depending on growth conditions. Similar roughnesses and surface morphologies were measured for CVD and MBE grown b-SiC. The specific dependence of the RMS roughness and morphology of these films is reported in Section 4.2.3.

Plan view TEM results from the b-SiC film along the <001> zone axis show two sets of diffraction spots, as in the inset of Figure 4-12. Diffraction spots with the larger lattice constant are due to the Si(001) substrate while the smaller are attributed to the b-SiC. Using the Si as an internal calibration standard, the measured lattice spacing, a, of the b-SiC was determined. The value for our films, a = 4.328 Å, agrees with that of bulk b-SiC, a = 4.359 Å. The TEM results also reveal an in-plane relationship of b-SiC{220}//Si{220} and b-SiC{440}//Si{440}. This supports the RHEED observations which also indicate a cube-on-cube, epitaxial relationship of (001)||(001) and <100>||<100>. The TEM micrograph in Figure 4-12 reveals a similar morphology to that of the AFM analysis. Light colored rectangular shapes with dimensions ranging from 10 to 200 nm appear in the TEM image. These features appear to be under the surface of the b-SiC film which is consistent with the presence of pyramid-shaped voids observed in the TEM cross-section. Imaging of the void defects under the surface of the film has also been observed for CVD and MBE grown b-SiC. The base dimensions of the voids in one case were as large as 1 mm.

Cross-sectional TEM (XTEM) micrographs of the MgO/b-SiC/Si composite film, as in Figure 4-13, also confirm the epitaxial relationship. XTEM revealed a columnar growth structure similar to the AFM images. Lamellar features, which are internal boundaries, possibly {111} twin-bands can be seen in the b-SiC layer of Figure 4-13. These defects have also been reported to cause the oblique streaks and additional spots observed in the RHEED patterns (Figure 4-6 through 6-8). At several locations near the b-SiC/Si interface, pits shaped like inverted rectangular pyramids were found penetrating into the Si substrate. This type of defect has been previously observed during the carburization of Si and results from the migration of Si atoms to the b-SiC growth surface, as discussed previously.

4.2.3 Effect of growth conditions on film structure

Several experiments were performed to investigate the growth mechanism of the b-SiC. The goal of the investigations was to find growth conditions which produced the best structural quality b-SiC for use as an integration layer between the Si and MgO overlayer. The measures of the b-SiC structural quality were: (1) its stability on the surface (no delamination or cracking), (2) epitaxy with the substrate as determined by RHEED, and (3) the roughness of the surface as determined by AFM. The first investigation examined the growth rate of the films and the maximum thickness attainable using the simple carburization procedure. The second series of experiments determined the temperatures and pressures at which epitaxial films are produced. The third experiment examined the growth rates produced at different temperatures and pressures as well as the dependence of surface roughness on film thickness.

The first series of films grown for these experiments are those listed in Table 4–2. All films were grown at 900°C with a MO flux ~ 10^-7 Torr. The films were grown after 30 min. of pre-deposition, growth with the stage shutter open and the MO shutter closed. Thickness of the film is plotted as a function of elapsed time after opening the MO shutter in Figure 4-14. Growth of films under the pre-deposition period results in thicknesses ranging from 10 to 30 nm. Assuming a linear growth rate, deposition proceeds at 0.3 to 1.0 nm/min during the 30 minutes of predeposition. After opening the MO shutter the films grow at ~3 nm/min. up to a thickness of ~200 nm. These growth rates are similar to those observed during CVD and MBE of b-SiC/Si. Films grown past 200 nm were observed to blister and delaminate from the Si substrate. A thickness limit to films grown using only a carbon source has been observed by others, in some cases leading to delamination. The cracking and delamination of the thicker films is most likely a result of the large tensile strain built up in the film due to the poor lattice and thermal expansion match between b-SiC and Si. It is also probable that at this thickness the pit defects begin to coalesce leaving the film unsupported, resulting in delamination. Collapse of the b-SiC film into these voids has been observed in TEM samples. Similar fractured surfaces have been observed in SEM micrographs of CVD grown b-SiC.

The pre-deposition step enables epitaxial growth of the b-SiC films because of the lower growth rate. The growth rate with the MO shutter closed is one order of magnitude lower than that with the shutter open. The order of magnitude growth rate reduction suggests the flux is one order of magnitude lower with the shutter closed, 10^-8 Torr, than with it open, 10^-7 Torr. This would suggest a critical flux between 10^-8  and 10^-7 Torr above which an epitaxial film cannot be grown at 900°C.

A second set of b-SiC depositions were performed to test this reasoning and the stability zone of epitaxial growth. The deposition conditions for this series of films is listed in Table 4–1. These films were all grown with a cooling step after the Si cleaning and without a pre-deposition period. The crystal structure of the films was examined using RHEED as the MO flux was varied between 10^-10 to 10^-7 Torr and the growth temperature was varied from 800 to 950°C.

The results of the study are shown in Figure 4-15. As is clear from the plot, epitaxial films are grown only at low MO pressures and high Si temperatures. Outside the shaded epitaxial "stability zone", only amorphous films resulted. The RHEED pattern of one such film is shown in Figure 4-16. The diffraction spots from the Si(001) surface can still be seen, but there is a diffuse background suggesting the Si is covered by a thin amorphous film. The films did not seem to increase in thickness with time, or had a very low growth rate after their initial nucleation. The low sticking coefficient of C on a C-terminated Si surface has been observed by Ferre, et. al. and suggests the formation of a stable graphitic layer at the higher C concentrations. This graphitic layer, once formed, prevents the outdiffusion of Si atoms to the surface and inhibits the growth of b-SiC.

The boundaries of the epitaxial stability zone agree with growths using the initial procedure with the pre-deposition step. At a growth temperature of 900°C the MO flux must be £  10^-8 Torr to promote epitaxial growth. This critical flux increases at higher Si temperatures. At 950°C the critical flux is ~10^-7 Torr. At 800°C an epitaxial film could not be grown at the lowest flux measurable, 10^-10 Torr. Similarly, gas source MBE was not able to produce epitaxial b-SiC below 700°C. Below 700°C the dissociation of the hydrocarbon precursor and sticking of C to the surface was sufficient to deteriorate the RHEED pattern of the Si, but b-SiC was not formed. At lower temperatures b-SiC growth is not favored due to the limited thermal energy necessary to diffuse Si atoms along the substrate surface.

The third experiment examined the linearity of the deposition rates produced at two different temperatures and MO fluxes. Reproducibility in the growth rate is essential for making comparisons between b-SiC films grown for structural investigations and those used as interlayers for MgO film overgrowth. The first set of growth conditions used was 950°C and 10^-7 Torr of MO flux. Four films were grown for 1, 5, 10, and 15 minutes. Film thickness as a function of growth time is plotted in Figure 4-17. The growth rate is linear in this time range at 11 nm/min. The second set of growth conditions was 850°C and 10^-9 Torr of MO flux. Three films were grown for 30, 60, and 90 minutes. Film thickness versus growth time for this series is also plotted in Figure 4-17. The growth rate is linear at 1 nm/min. For both series of films the growth rates are much more consistent than the series plotted in Figure 4-14. This is a result of cooling the Si, removing the possibility of b-SiC growth while checking the flux of the MO source and allowing for more accurate measurement of that flux. Increasing the MO flux by two orders of magnitude did not result in two orders of magnitude faster growth rate. This suggests that at 950°C the growth rate is becoming reaction limited, not MO flux limited.

These two series of films were also used to measure the dependence of surface roughness on film thickness and growth rate. AFM images of four films deposited at the low growth rate (850°C and 10^-9 Torr of MO) are shown in Figure 4-18. For each film the elevation and corresponding deflection data are shown for 4 mm square regions. RMS roughness, s_RMS, and thickness, h, of each film is also shown above each image. As the thickness increases from 1 nm to 30 nm, the RMS roughness increases from 0.6 nm to 5.7 nm. For the thinnest film the morphology is smooth and almost featureless. In the 8 nm thick film, pit-like defects can be found extending into the film. The two thickest films display a columnar morphology which becomes larger and faceted for the 30 nm thick film.

The roughness of the b-SiC films is plotted as a function of thickness in Figure 4-19. Also shown in the figure, marked with a triangle, is the roughness of the Si substrate measured after a typical cleaning procedure. For the low growth rate conditions, a linear dependence is observed. The RMS roughness is ~19% of the film thickness. For the high growth rate conditions, 950°C and 10^-7 Torr of MO flux, the dependence was nearly linear with an RMS roughness, ~2% of the film thickness. A 5 to 10% dependence of roughness on film thickness was also reported for CVD growth at 1300°C. In general, smoother b-SiC films are grown at higher temperatures, in agreement with reports on MBE deposition and in contradiction with the results of CVD grown films.

These experiments suggest that b-SiC interlayers grown to only a few nanometers thickness have the highest structural quality. The low temperature series is also preferred as the low growth rate makes the thickness of the b-SiC easier to control. The thin films should also minimize the size and density of the Si pit defects. Since less b-SiC is grown, less Si will need to be removed from the substrate. In the Section 4.3.4 the different thickness b-SiC films will be used as nucleation layers for MgO overgrowth.

4.3 Growth of MgO overlayer using MOMBE

This section will discuss the deposition of the MgO overlayer on b-SiC/Si using the MOMBE growth technique. The first sub-section will detail the growth procedure. The second sub-section will present the results of chemical and structural analysis performed on the films. Next, the influence of growth parameters on the structural perfection of the MgO layer will be examined. The last section will examine the influence of the b-SiC interlayer on the structural quality of the MgO overlayer.

4.3.1 The MOMBE growth procedure

The MOMBE technique was successful in depositing an epitaxial MgO overlayer on the b-SiC covered Si(001). The procedure used to deposit these films will be described in this sub- section. After the growth of the b-SiC template layer the MgO overlayer growth proceeded through 6 steps: ramping the stage and MO sources, equilibration of the MO source flux, ignition and stabilization of the RF O₂ plasma, deposition and in-situ monitoring, and post-growth cooling. These six steps are shown in the schematic of Figure 4-20, and an actual record of chamber pressure, MO source temperature, and stage temperature versus time is plotted in Figure 4-21.

After the deposition of a b-SiC interlayer on the Si(100) surface the MO source is ramped to 20°C at 40°C/min. The stage is then ramped to the MgO growth temperature, from 800°C to 950°C, at 20°C/min. A second MO source is then heated at 10°C/min. to a temperature necessary to achieve the required MO flux. A second MO source is used so only one of the two MO sources is exposed to the O₂ background present in the chamber while it is hot. Heating the MO sources in the presence of O₂ is believed to decrease the stability of the source and its usable lifetime. This dedication of one MO source to each process increased the reproducibility of both growth processes.

The MO source temperatures were typically ~ 160°C, but since the source does not always produce the same flux at a given temperature, the MO flux must be directly measured prior to growth. The flux monitor was used to measure the MO flux using the same procedure explained in Section 4.2.1. MO fluxes used for MgO growth ranged from 10^-9 to 10^-5 Torr.

After the substrate temperature and MO flux have been set and stabilized, the RF O₂ plasma source is started. First the RF generator is ramped up to the required RF power level, typically 300 to 500 W. Next, the RF matching network on the RF gun is adjusted to minimize the reflected RF power. The variable leak valve on the O₂ supply line is opened until the flux monitor reads the desired pressure of O₂ for the MgO growth. The increase in pressure following the opening of the leak valve is shown in Figure 4-21. Oxygen pressures used for MgO growth ranged from 10^-6 Torr, the lowest pressure which would maintain a stable plasma, to 10^-4 Torr, the highest pressure attainable. The matching network is then re-adjusted to minimize the reflected RF power, and the plasma is ignited with an initial spark from the gun.

The deposition of MgO is initiated by opening the O₂, MO, and stage shutters simultaneously. RHEED diffraction patterns were recorded during the initial moments of growth to determine the crystal structure of the MgO film. RHEED was not used throughout the remainder of the deposition as operation of the electron source in the O₂ atmosphere has been observed to shorten the life of its tungsten filament.

After the required amount of deposition time has elapsed, the stage, O₂, and MO shutters are closed to terminate the MgO growth. The stage and O source are both ramped to 20°C at 20°C/min. as shown in Figure 4-20 and Figure 4-21. RHEED images were recorded after the completion of the deposition to examine the final crystal structure of the MgO surface.

This procedure detailed above was used to grow a series of MgO films for examination of the conditions under which epitaxial growth is maintained. MgO films produced using this procedure are listed in Table 4–3 and Table 4–4. The b-SiC layers in Table 4–3 were grown using the pre-deposition technique described in Section 4.2.1 while those in Table 4–4 were grown using the improved technique. The chemical and structural properties of the MgO films will be discussed in the following section. The dependence of the structural perfection of the MgO layer on the deposition conditions and thickness of the b-SiC interlayer will be examined in Sections 4.3.3 and 4.3.4.

4.3.2 Structural and chemical analysis

In addition to in-situ RHEED analysis, the deposited films were compositionally and structurally analyzed using a variety of ex-situ techniques. The composition of the films was examined by Auger electron spectroscopy (AES). Surface morphology and roughness were examined with atomic force microscopy (AFM). Film microstructure was investigated using transmission electron microscopy (TEM) and x-ray diffraction (XRD). The results of in-situ and ex-situ chemical and structural characterizations will be presented in this chapter.

Under favorable growth conditions, sharp characteristic single crystal RHEED patterns were observed as shown in Figure 4-22 along the Si <100> (a) and Si <110> (b) azimuths. The diffraction patterns observed are similar to those reported for single crystal MgO. Thus the RHEED patterns suggest the deposited films are epitaxial. Occasionally the RHEED patterns consisted of a series of concentric rings as in Figure 4-22(c) which did not change as the azimuthal angle was rotated. This ring pattern is characteristic of polycrystalline MgO films. The growth conditions leading to both types of crystal structures will be discussed in Sections 4.3.3 and 4.3.4.

AES was used to measure the atomic species present in the MgO films. A typical AES spectrum is shown in Figure 4-23 for an epitaxial MgO film grown on b-SiC/Si(001) (FN051000). The spectra shown are similar to those reported for MgO grown by others. Only peaks corresponding to the K_LL transitions of O and Mg at 503 and 1184 eV are observed in the film spectra. The spectrum shown can be directly compared to the AES spectrum of a single crystal MgO substrate measured under similar conditions. The single crystal MgO substrate exhibits Mg and O peaks in the same positions as the film and in the same ratio. Thus, it can be concluded that the films are stoichiometric. Additionally the MgO substrate surface shows some contamination with carbon.

AES was also used to measure the composition of the MgO/b-SiC/Si(001) heterostructure as a function of depth. The film was sputtered with Ar ions and the AES spectrum was measured as the upper layers of the film were removed. In Figure 4-24 the AES spectra are shown as a function of sputtering time. Prior to the sputter the spectrum is similar to the MgO substrate shown for comparison. After 10 minutes of Ar sputtering the spectra are similar to that of the b-SiC interlayer shown in Figure 4-10.

The surface morphology of the MgO films was measured using AFM. A typical AFM micrograph is shown in Figure 4-25 for a 65 nm thick MgO film (FN033000). The morphology of the MgO surface is similar to that of the b-SiC films, consisting of a dense network of columnar sub-grains. The sub-grains often exhibited facets parallel to the <100> planes of the Si substrate. Widths of the columnar features ranged from 10 to 500 nm. Sputtered and PLD grown MgO/Si(100) films showed similar morphology with grain sizes ranging from 10 nm to 1 mm. The RMS roughness of the MgO surface in the figure was 3.3 nm. The RMS roughness of the MgO films in this study ranged from 1 to 5 nm depending mostly on the growth conditions of the b-SiC layer. These values are an order of magnitude larger than the RMS roughness of freshly cleaved MgO single crystals, 4 Å, but similar to reports of other thin films. The dependence on the b-SiC growth conditions will be examined in Section 4.3.4.

XRD q-2q scans of the composite MgO/b-SiC/Si structure (FN072399) show a strong peak at 2q = 42.9^o, as shown in Figure 4-26. This peak is consistent with diffraction from the MgO(002) plane. Also observed in the scan are the Si(002) and Si(004) peaks. The Si(002) reflection is forbidden by symmetry, but weak peaks at 33° were consistently observed in the XRD scans. Similar features were observed in the XRD spectra of MgO films deposited by sputtering, PLD, and MOCVD.. Peaks from the b-SiC layer are not observed due to its limited thickness and the lack of a strong (00h) reflection from bulk b-SiC. The results suggests a (001)||(001) and <100>||<100> epitaxial relationship between the MgO and the Si substrate.

Rocking, w, curves were performed on the MgO(200) peak of the integrated thin films (FN071300), as shown in Figure 4-27. The data in the figure are fit by a Gaussian distribution with a FWHM of 2.8 degrees. Typically the rocking curve FWHM ranged from 2.4 to 4 degrees. These values are larger than the w-curve FWHM of single crystal MgO substrates, 0.1°, but similar to those of sputtered and PLD grown MgO/Si(001), ³  3°. Reports of FWHM < 0.3° have been made for some MgO thin films. While the FWHM provides some measure of the crystalline perfection of the MgO films, the reduced thickness of the films could also cause broadening of the peak.

TEM was also used to probe the MgO structure. A plan view TEM micrograph taken along the <001> zone axis of a composite MgO/SiC/Si film is shown in Figure 4-28. A morphology similar to that observed in the AFM images is observed. The surface consisted of sub-grains with diameters on the order of 10 to 100 nm. The TEM diffraction pattern revealed three sets of spots, as shown in the inset of Figure 4-28. These three contributions are from the Si substrate, the b-SiC interlayer, and the MgO overlayer. Thus the TEM results also suggest a (001)||(001) and <100>||<100> epitaxial relationship is maintained throughout the entire structure from the Si substrate to the MgO over-layer. The calculated lattice of the MgO layer was a = 4.202 Å for the sample in Figure 4-28. This value agrees with that of bulk MgO, a = 4.213 Å.

A cross-sectional TEM (XTEM) micrograph of the entire MgO/b-SiC/Si structure is shown in Figure 4-29. Both the MgO/b-SiC and b-SiC/Si interfaces are abrupt with no evidence of secondary phase formation. A pit defect formed in the Si substrate is also evident in the lower left corner of the micrograph. As explained in Section 4.2.2, these defects are a result of Si migration to the b-SiC surface during the b-SiC formation. A high resolution TEM image of the MgO/b-SiC interface is shown in Figure 4-30. From the image it is clear that an epitaxial relationship exists between the b-SiC interlayer and the MgO overlayer as the columns of atoms proceed continuously from one layer to the next.

4.3.3 Effect of MgO growth conditions on film structure

Several experiments were performed to investigate the growth mechanism of the MgO films. The goal of the investigations was to find growth conditions which produced the best structural quality MgO and the highest growth rate. The measures of the MgO structural quality were: (1) epitaxy with the substrate as determined by RHEED, (2) roughness of the surface as determined by AFM, and (3) the FWHM of the MgO(200) rocking curve. These experiments examined the growth rate and the crystal structure of the films as function of the MgO growth conditions.

The first series of experiments examined the effects of growth temperature on the growth rate and the crystal structure of the films. A series of MgO films were grown at 600, 700, 800, and 900°C while holding the MO flux, O₂ pressure, and RF power constant at 10^-6 Torr, 3x10^-5 Torr, and 400 W respectively. The films were also deposited on b-SiC interlayers grown under the same conditions, 10^-7 Torr of MO and a substrate temperature of 800°C. The results of the study are shown in Figure 4-31 where the growth rate of the films is plotted as a function of temperature.

Two distinct regions are observed in the plot. In region I, at temperatures less than 750°C, the growth rate decreases with increasing temperature and RHEED analysis indicated the growth of polycrystalline films. The gradual loss of epitaxy below 500 -700°C was also reported for sputtered MgO/Si(100) films. Contrary to these results, reports of epitaxial MgO films at temperatures down to 200°C have also been made for MBE. This observation is most likely a result of the e-beam evaporation procedure used. A similar decrease in deposition rate with increasing temperature was observed for MOCVD and PLD growth of MgO on Si. In region II, at temperatures greater than 750°C, the growth rate increases with increasing temperature. In region II, RHEED analysis indicated the growth of epitaxial MgO films.

In the low temperature region I, the growth rate increases with decreasing temperature due to the increased sticking coefficient of the Mg and O atoms at lower substrate temperatures. Others have suggested that the sticking coefficients of both Mg and O₂ decrease with increasing temperature. One report predicted that at 750°C only half of the reactant atoms remain on the surface long enough to be incorporated into the growing film, while at 25°C, a 100% incorporation was observed. However, as the mobility of the reactant atoms on the surface is low at lower temperatures, polycrystalline films result. In the high temperature region II, the growth rate increases with increasing temperature because of the increased cracking efficiency of the Mg(acac)₂ precursor. As the mobility of the reactant atoms on the surface is high, epitaxial films can be formed. The results of the study indicate that the highest growth rate for epitaxial MgO films occurs at high substrate temperatures.

The second series of experiments examined the effects of MO precursor flux on the growth rate and the crystal structure of the films. A series of MgO films were grown from 10^-8 to 4.5 x 10^-6 Torr of MO while holding the substrate temperature, O₂ pressure, and RF power constant at 800°C, 10^-5 Torr, and 400 W respectively. The films were also deposited on b-SiC interlayers grown under the same conditions, 10^-7 Torr of MO and a substrate temperature of 800°C. The results of the study are shown in Figure 4-32 where the growth rate of the films is plotted as a function MO flux. A sharp increase in the growth rate is observed between 10^-8 and 2 x 10^-7 Torr of MO flux. Above 2 x 10^-7 Torr the growth rate is saturated at 30 nm/hr. This behavior indicates that MgO growth below 2 x 10^-7 Torr is occurring under Mg deficient conditions. Above 2 x 10^-7 Torr the growth proceeds under a Mg excess. The MgO films grown under Mg deficient conditions were amorphous or textured while those under excess Mg were epitaxial. These results suggest that increasing the Mg precursor pressure will increase the growth rate up to the point that the films become limited by the amount of activated O, and that epitaxial films appear to be favored only under Mg rich conditions. A similar claim was made for MOCVD prepared MgO/SiO₂.

The third series of experiments examined the effects of RF power on the growth rate and the crystal structure of the films. A series of MgO films were grown with RF power varied between 0 and 500 W while holding the substrate temperature, O₂ pressure, and MO flux constant at 800°C, 10^-5 Torr, 10^-6 Torr respectively. The results of the study are shown in Figure 4-32 where the growth rate of the films is plotted as a function RF power. At 0 W, MgO films were not observed to grow and the b-SiC layer remained unaffected by the O₂ atmosphere. A linear increase in the growth rate between 0 and 25 nm/h is observed. The slope of the line is 56 nm/h/kW. In MOCVD studies of MgO/SiO₂ the RF power was not observed to change the growth rate. The authors suggest that the amount of activated O produced by their plasma does not depend on RF power. The results of the current study suggest that increasing the RF power of the O₂ plasma does increase the percentage of activated oxygen delivered to the film, resulting in a higher growth rate. All of the films in the study were epitaxial which suggests growth is still proceeding under Mg rich conditions even at the highest RF plasma power, 500 W.

Experiments were also conducted on the epitaxial MgO films to determine the dependence of RMS roughness, as determined from AFM, and XRD rocking curve FWHM on thickness and deposition conditions. Rocking curve widths ranged from 2.4° to 3.7° and did not show dependence on either deposition conditions or MgO thickness. Similarly, the RMS roughness of the films in the previous studies was consistently between 2 and 3 nm. The results of these studies suggest that the structural quality and roughness of the films remain constant over the range of deposition conditions and thicknesses used in this study. This is contradictory to reports of (100) textured MgO/Si prepared by sputtering which suggest an improvement in the roughness from 12 to 2.6 nm at the deposition temperature was reduced.

The studies reported in this section suggest that epitaxial MgO films can be deposited under Mg rich conditions in the presence of activated O. The highest growth rates are achieved under high temperatures and RF oxygen plasma powers. The highest growth rates achieved, ~ 30 nm/h, occurred at 800°C, 500 W, and 10^-6 Torr of MO. For comparison, a deposition rate of ~ 1 mm/h was reported for homoepitaxial growth of MgO by MBE and MgO/SiO₂ by MOCVD. Rates of 200 nm/h were reported for sputtered and PLD grown MgO/Si. The results of the current study suggest that higher growth rates should be obtained if a growth temperature of 900°C were used. Improvement in the structural quality of the films was not achieved through variation of the MgO deposition conditions. In the following section the impact of the b-SiC deposition conditions on the quality of the MgO layer will be investigated.

4.3.4 Effect of the b-SiC interlayer on MgO film structure

Experiments were performed to determine the impact of the b-SiC interlayer structure on the MgO overlayer films. The goal of the investigations was to determine whether changing the thickness and/or growth conditions of the b-SiC films could be used to improve the structural quality of the MgO. The measures of the MgO structural quality for these experiments were: (1) epitaxy with the substrate as determined by RHEED, and (2) roughness of the surface as determined by AFM. These experiments examined the crystal structure of the films as function of the MgO growth conditions.

For this study a series of MgO films grown under identical conditions were deposited on b-SiC layers of varying thickness and growth conditions. The MgO films were grown at 800°C, 10^-5 Torr of O₂ at 400 W, and 10^-6 Torr of MO flux. The MgO growth time was 180 minutes, corresponding to a thickness of ~70 nm. The b-SiC layers were deposited under one of two sets of conditions. The first corresponded to the low growth rate described in Section 4.2.3, 1 nm/min. at 850°C and 10^-9 Torr of MO flux. The second set of conditions were those producing the high b-SiC growth rate, 11 nm/min. at 950°C and 10^-7 Torr of MO flux. The data from Figure 4-19 suggest that the best b-SiC films are those grown to thicknesses less than 100 nm. Consequently, MgO layers were deposited on b-SiC films ranging from 1 to ~50 nm thick. The thickness of the b-SiC films was determined by the deposition conditions and the growth time using the rates calculated in Figure 4-17.

RHEED images taken after completion of the MgO deposition were used to determine the crystal structure of the film. While very thin b-SiC films had the lowest RMS roughness, MgO films deposited on interlayers thinner than ~4 nm resulted in a polycrystalline MgO structure. Those deposited on interlayers greater than ~4 nm were epitaxial. This observation was made for MgO films deposited on interlayers grown at both the high and low rates.

The cause for this change form epitaxial to polycrystalline films is the incomplete coverage of the Si(001) surface with b-SiC for thicknesses below ~4 nm. Incomplete coverage of the b-SiC layer would suggest the 3-dimensional growth of b-SiC nuclei on the Si surface. This carburization mechanism has been suggested to occur in atmospheres containing a large ratio of [H]/[C]. This 3-dimentional growth mode leaves some exposed Si surface between the b-SiC nuclei before they coalesce. The growth of MgO on the free Si(001) surface by MOMBE is known to result in polycrystalline films. This explanation also agrees with the slow evolution of the RHEED patterns observed during the b-SiC growth (Figure 4-8). The RHEED patterns show a combination of Si(001) and b-SiC(001) spots at times prior to complete coverage of the Si surface. The time necessary to completely cover the Si surface was typically ~10 minutes at 850°C, corresponding to a b-SiC thickness of ~10 nm. Thus the Si(001) surface needs to be completely covered with b-SiC to promote epitaxial growth.

AFM scans of the MgO surface were used to determine the roughness of the MgO films. AFM images of the series of MgO films deposited on b-SiC formed under the low growth rate conditions are shown in Figure 4-34. Shown in the figure are the height and deflection data from 10 mm square regions of four MgO films deposited different thicknesses of b-SiC, h_SiC. All images have the same columnar morphology. No distinct evolution of the feature sizes is noted, however the RMS roughness, s_RMS, can be used to determine two distinct trends with increasing thickness

The RMS roughness of the MgO surface is plotted versus the b-SiC interlayer thickness in Figure 4-35. Shown in the plot are two sets of data corresponding to the b-SiC films deposited at both the high (squares) and low (circles) growth rates. The region in which polycrystalline films are grown is shaded. The remainder of the films were epitaxial MgO. In both sets of data a decrease in the RMS roughness is observed with an increase in the interlayer thickness up to ~8 nm. Above 8 nm the RMS roughness increased with b-SiC thickness. The transition between the two regions occurs at the same b-SiC thickness as the transition from polycrystalline to epitaxial growth. Thus the increased roughness for interlayers less than ~8 nm is most likely an effect of the polycrystalline structure. If the increase in the roughness for interlayers ³ 8 nm is assumed to be linear with interlayer thickness, slopes of 0.03 and 0.35 are measured for the high and low growth rate b-SiC respectively. These values are similar to those measured in Figure 4-19 for the interlayer surface roughness. This suggests the roughness of the epitaxial MgO is determined by the initial roughness of the b-SiC interlayer.

This study was successful in determining the interaction between the b-SiC interlayer and MgO overlayer. Overall, the structure of the b-SiC interlayer has a large influence on the structure of the MgO in comparison to the effects of the MgO growth parameters. In specific, epitaxial MgO can only be grown on interlayers ³ 8 nm, as b-SiC completely covers the Si surface. The lowest RMS roughness MgO layers, ~1.2 nm, were grown on the thinnest b-SiC interlayers which met the condition of complete surface coverage. This implies that the best b-SiC template layers for the overgrowth of epitaxial MgO are ~ 8 nm.

4.4 Growth of ferroelectric BaTiO₃ films on integrated MgO using MOCVD

The epitaxial MgO/b-SiC/Si(001) heterostructures described in the preceding sections were used as integration layers for the overgrowth of BaTiO₃ using MOCVD. Details of the MOCVD technique and the actual conditions used for the deposition can be found elsewhere. Initial experiments were performed on the BaTiO₃ films to examine their crystal structure and electrical properties. XRD and TEM were used to determine whether the BaTiO₃ films were epitaxially deposited on the thin film MgO. Electrical techniques such as capacitance and polarization vs. voltage and capacitance versus temperature measurements were used to determine whether the films produced are in a FE state. These techniques have been discussed in the main body of the dissertation.

4.4.1 Structural analysis

To make effective EO devices using the integrated BaTiO₃/MgO/biC/Si(001) heterostructures the active BaTiO₃ layer must be deposited epitaxially. The crystal structure of the integrated BaTiO₃ thin films was examined using XRD and TEM. The results of these studies will be described in the following section.

Thin film XRD was used to determine the structure of the BaTiO₃ deposited on integrated MgO. A typical XRD q -2q scan is shown in Figure 4-36 for a ~400 nm thick BaTiO₃ film (AT406) deposited on MgO(70 nm)/b-SiC(10°nm)/Si heterostructure (FN100899). Similar to the q -2q scan of the bare MgO/b-SiC/Si(001) structure (Figure 4-26), peaks are observed at 69° and 43° corresponding to the (004) plane of the Si substrate and the (002) plane of the MgO interlayer respectively. Additional peaks are observed near 2q angles of 22° and 45° corresponding to diffraction from the (001) and (002) planes of BaTiO₃. Thus, the XRD data suggests that the BaTiO₃ films are epitaxial. Additionally, XRD phi scans about the BaTiO₃(022) and Si(022) peaks revealed the four-fold symmetry indicative of cube-on-cube epitaxy. This is the same epitaxial relationship observed for BaTiO₃ thin films deposited on MgO single crystal substrates. The lattice parameter calculated from the (200) diffraction peak is 3.986 Å. This value is slightly smaller than the bulk a, 3.991 Å, and c, 4.035 Å, lattice parameters. Such compression has been attributed to residual strain in the thin film heterostructures.

XRD rocking, w, curves were used to determine the crystalline perfection of the integrated BaTiO₃ thin films. A typical w-curve about the BaTiO₃ (002) reflection is shown in Figure 4-37 for a ~400 nm thick film (AT406). The data are well fit by a Gaussian distribution with a FWHM of 1.7°. While this is the smallest FWHM measured for the integrated BaTiO₃, it is still an order of magnitude larger than the typical FWHM of BaTiO₃ deposited on single crystal MgO under similar conditions. The larger width of the rocking curve is attributable to the increased FWHM of the thin MgO, 2 to 4°.

TEM was also used to analyze the integrated BaTiO₃ microstructure. Figure 4-38 shows a low magnification XTEM image of the BaTiO₃/MgO/b-SiC/Si(001) structure (AT346/FN100899). The thick BaTiO₃ layer is clearly observed at the top of the image. Sub-grain features, which are related to the columnar growth of the BaTiO₃ film, are observed in the film. The columns had faceted boundaries parallel to the BaTiO₃(100) planes with dimensions ranging from 50 to 300 nm. The interfaces between the BaTiO₃/MgO and MgO/SiC layers are not as distinct as expected and may indicate some diffusion between the layers occurs during the BaTiO₃ deposition. Pit defects in the Si substrate, which are voids formed during the b-SiC deposition, are also observed in the micrograph.

A selected area diffraction pattern from an integrated BaTiO₃, plan-view TEM sample (AT346/FN100899) is shown in Figure 4-39. The diffraction patterns were taken along the Si[110] zone axis and consist of two sets of spots attributable to the Si substrate and the BaTiO₃ film. Using the Si as an internal calibration standard, the lattice constant of the BaTiO₃ was measured. The experimentally determined value of a = 3.98°Å is slightly smaller than the bulk a and c spacings and is in agreement with the value determined from XRD. The diffraction pattern also suggests the orientational relationship between the BaTiO₃ film and Si substrate is BaTiO₃ {001} || Si {002} and BaTiO₃ {011} || Si {022}. Thus the TEM analysis suggests the same cube-on-cube epitaxial relationship determined by XRD.

The studies reported in this section suggest that epitaxial BaTiO₃ films can be grown by MOCVD on the MgO thin films deposited using MOMBE. However, the structural quality of the films is not as high as those films deposited on single crystal MgO. Thus it cannot be directly assumed that the deposited films are in a FE state. In the following section dielectric measurements will determine whether the integrated BaTiO₃ films exhibit FE properties.

4.4.2 Electrical analysis

Dielectric measurements were used to determine whether the BaTiO₃ films deposited on the MgO/b-SiC/Si substrates exhibit FE properties. In this section the fabrication of the test devices will be briefly discussed followed by the results of capacitance versus voltage, polarization versus voltage, and capacitance versus temperature measurements.

Since the Si substrate is conducting, a metal-oxide semiconductor (MOS) structure with a parallel plate geometry was used to measure the dielectric properties of the films. A schematic of the structure is shown in Figure 4-40. Aluminum was thermally evaporated through a shadow mask onto the BaTiO₃ surface to form the upper electrode for dielectric measurements. The aluminum electrodes are circular disks 475 mm in diameter and 200 nm in thickness. The aluminum was covered with 100 nm of gold to prevent oxidation of the Al during measurements at elevated temperature. The n-type Si substrate was used as the bottom electrode. Ohmic contact was made to the Si using indium. The capacitor structures exhibited very low leakage currents (~10 mA at 10 VDC), indicating the BaTiO₃ and MgO films are highly resistive.

The results of a typical capacitance versus voltage, C(V), measurement for an integrated BaTiO₃/MgO/b-SiC/Si(001) capacitor structure are shown in Figure 4-41. The BaTiO₃ film (AT346) was grown by MOCVD to a thickness of ~400 nm. The MgO (17 nm) and b-SiC (30 nm) were grown by MOMBE under typical deposition conditions (FN100899). Similar C(V) curves were measured for several different surface electrodes on two integrated BaTiO₃ films (AT346, AT325). The tunability and shape of the C(V) curve appears to be a combination of two different behaviors. First, the hysteresis and "inverted V" shape are characteristic of FE materials. Second, the asymmetry between positive and negative voltages is characteristic of the MOS structure used to measure the dielectric properties. MgO films from this study were used to make MOS structures without the BaTiO₃ layer. While these structures did show the asymmetric tunability expected of a MOS structure, they did not exhibit the inverted V shape, nor did they exhibit a large degree of hysteresis. This component of the C(V) behavior is, therefore, attributed to the BaTiO₃ layer and is a direct indication that the BaTiO₃ is in a FE state.

A typical polarization versus voltage, P(V), measurement made on the integrated BaTiO₃ is shown in Figure 4-42. The measurements was made on a 400 nm thick BaTiO₃ film (AT346) grown on MgO/b-SiC/Si (FN100899). The hysteresis observed in the P(V) plot is typical of FE thin film materials, and also suggests the epitaxial BaTiO₃ in this study is in a FE state. Saturation of the hysteresis loop is not observed up to the ±20 V maximum of the RT6000S. The remnant polarization measured was ~1 mC/cm². Similar values were measured for many different surface electrodes on two integrated BaTiO₃ films (AT346, AT325). This spontaneous polarization is substantially smaller than that of bulk BaTiO₃, 27 mC/cm², but is similar to values obtained for epitaxial BaTiO₃/MgO/GaAs. The reduced value could be the result of several factors discussed in the main body of the dissertation. For these measurements one additional cause must be included since the measuring field is applied normal to the plane of the film. If we assume that the BaTiO₃ films grown on the thin film MgO have the same a-axis orientation as those grown on single crystal MgO, the applied electric field is normal to the polarization direction of all the domains in the film. Thus, the polarization changes observed must come from switching the polarization out of the plane of the film. Because such 90° rotations of the polarization also change the strain state of the film, these changes are expected to be minimal in thin films.

The results of a typical capacitance versus temperature measurement for a 400 nm thick, integrated BaTiO₃ film (AT346) is shown in Figure 4-43. The peak in the capacitance at 130°C is an indication of the FE to paraelectric phase transition. This is further evidence that the film is in a FE state at room temperature. The peak is also very broad, indicative of a diffuse phase transition. The diffuseness of the transition for the integrated BaTiO₃ films is comparable to the that of the thinnest films grown on single crystal MgO. This could be attributed to increased epitaxial strain as a result of the additional MgO/SiC epitaxial layers. An increased amount of structural disorder, as measured by the FWHM of the XRD rocking curve, in the integrated films could also be responsible for the increased diffuseness of the FE phase transition in these films.

The studies reported in this section suggest that the integrated BaTiO₃ films are in a FE state at room temperature despite the structural disorder discussed in Section 4.4.1. Moreover, the FE properties measured for the integrated BaTiO₃ films are similar to those exhibited by BaTiO₃ films deposited on single crystal MgO.

Figure 4-1: Schematic diagram of the MgO/b-SiC/Si (001) heterostructure. The growth proceeds in the (001) direction. Shown for each material is one unit cell along the (110), (010), and (001) projections. Lattice spacing and atomic radii are not drawn on the same scale.

Figure 4-2: Ball and stick model of the Mg(acac)₂ structure.

Figure 4-3: Schematic diagram of the temperature schedule for growth of a b-SiC/Si (001) heterostructure for a time, t_g. Shown in the figure are: the degass (t ~ 50 min.), the desorbtion (t ~ 100 min.), the MO ramp (t ~ 160 min.), and the growth (t ~ 230 min.).

Figure 4-4: Temperature and pressure data recorded during growth of a b-SiC/Si (001) heterostructure (BHH087). The dashed line at time t₁ corresponds to lowering the flux monitor between the MO source and the stage. The shaded regions at t₂, t₃, and t₄ correspond to opening of the MO source shutter to measure the flux. The dashed line at time t₄ corresponds to raising the flux monitor. The shaded region at time t₅ corresponds to opening the MO source and stage shutters during film deposition.

Figure 4-5: RHEED image taken after desorbtion of Si (001) at 850°C and cooling to room temperature. Images are along the (100) Si (a) and (110) Si (b) at an acceleration voltage of 9 keV.

Figure 4-6: RHEED image taken after the growth of b-SiC on Si (001) at 850°C and cooling to room temperature (BHH060). Images are along the <100> Si (a) and <110> Si (b) azimuths at an acceleration voltage of 9 keV.

Figure 4-7: Series of RHEED images as a function of time during growth of b-SiC at 950°C (BHH087). Images are taken at 9 keV along the <100> Si azimuth.

Figure 4-8: Series of RHEED images as a function of time during growth of b-SiC at 850°C (BHH079). Images are taken at 9 keV along the <100> Si azimuth.

Figure 4-9: FTIR spectrum of a b-SiC film shows an absorbtion peak characteristic of the b-SiC TO mode at 798.3 cm^-1.

Figure 4-10: Auger electron energy spectrum of a b-SiC film (BHH063) grown on Si (001) after a brief sputter cleaning. Peaks corresponding to Mg, C, Si, O, and Ar are observed in the spectrum.

Figure 4-11: AFM image (10 mm x 10 mm) of a 60 nm thick b-SiC film (BHH079) grown on Si(001). Shown is the elevation data (left) and the corresponding deflection data (right). The RMS roughness was 11 nm.

Figure 4-12: Bright field TEM micrograph showing a plan view of a b-SiC film grown on Si(001). The inset shows the spot diffraction pattern of the heterostructure along the Si <100> zone axis indicating the cube-on-cube epitaxy of the film with the substrate.

Figure 4-13: High resolution TEM image showing a cross-sectional view of the interface between a b-SiC film and the underlying Si(001) substrate. The epitaxy of the film with the substrate can be clearly seen along with the twin band structures in the SiC.

Figure 4-14: Growth rate determination for b-SiC grown with a 30 min. pre-deposition period during which the stage shutter is open and the MO shutter is closed. Deposition conditions were 900°C and 10^-7 Torr of MO (with the MO shutter open). Film delamination was observed for films grown past 1 hr. Thickness was determined using profilometry, and error bars denote the standard deviation of measurements taken at several locations on the film.

Figure 4-15: The effect of stage temperature and Mg(acac)₂ (MO) source pressure on the crystal structure of b-SiC films as measured by RHEED. A stability region at high temperatures and low MO fluxes is observed (shaded). Films were grown without a pre-deposition period.

Figure 4-16: RHEED images recorded after desorbtion of Si (001) at 850°C (a) and after failed attempt to grow epitaxial b-SiC (b) outside of the stability region in Figure 4-15. Growth conditions were 850°C and 10^-8 Torr of MO for 60 minutes (BHH077). Images were taken after cooling to room temperature along the (100) Si at an acceleration voltage of 9 keV.

Figure 4-17: Growth rate determination for b-SiC deposited at two sets of conditions inside the epitaxial stability zone. Deposition conditions for the high growth rate were 950°C and 10^-7 Torr of MO. For the low growth rate, deposition conditions were 850°C and 10^-9 Torr of MO. The b-SiC films were grown without a pre-deposition period. Thickness was determined using profilometry, and error bars denote the standard deviation of measurements taken at several locations on the film.

Figure 4-18: AFM images (4 mm x 4 mm) of b-SiC films grown on Si(001) as a function of film thickness for films deposited at the low growth rate. Shown for each film is the elevation data (left) and the corresponding deflection data (right). RMS roughness, s_RMS, and thickness, h, of each film is shown above the image. Deposition conditions were 850°C and 10^-9 Torr of MO. The films shown are: 1 nm = BHH095, 8 nm = BHH094, 17 nm = BHH093, 30 nm = BHH092

Figure 4-19: RMS roughness of b-SiC as a function of film thickness for films deposited at two sets of conditions inside the epitaxial stability zone. Deposition conditions for the high growth rate (n ) were 950°C and 10^-7 Torr of MO. For the low growth rate (l ), deposition conditions were 850°C and 10^-9 Torr of MO. The roughness of the Si substrate after desorbtion at 850°C is also plotted (s ). Roughness was measured using AFM.

Figure 4-20: Schematic diagram of the temperature schedule for growth of an MgO/ b-SiC/Si (001) heterostructure for a time, t_g. Shown in the figure are: the ramp from the b-SiC growth temperature (t ~ 10 min.), the MO ramp (t ~ 20 min.), the growth (t ~ 60 min.), and the ramp to room temperature (t ~ 120 min.).

Figure 4-22: RHEED image taken after the growth of an epitaxial (a, b) MgO film on b-SiC/Si (001) at 850°C and cooling to room temperature (FN051000). For comparison a polycrystalline MgO film (FN032300) is shown in (c). Images are taken along the (100) Si (a, c) and (110) Si (b) azimuths at an acceleration voltage of 9 keV.

Figure 4-23: Auger electron energy spectrum of an MgO film grown on b-SiC/Si (001) after a brief sputter cleaning (FN101400). For comparison a spectrum collected from an MgO single crystal is shown. Peaks corresponding to Mg and O are observed in both spectra. Some trace carbon signal is observed in the MgO single crystal sample.

Figure 4-24: Auger electron energy spectrum of an MgO film (FN101400) grown on b-SiC/Si (001) with sputter cleaning time as a parameter. For comparison a spectrum collected from an MgO single crystal is shown. Peaks corresponding to Mg, O, Si, and C are labeled.

Figure 4-25: AFM images (10 mm x 10 mm) of a 18 nm thick MgO film (FN033000) grown on 15 nm of b-SiC. Shown is the elevation data (left) and the corresponding deflection data (right). The RMS roughness was 3.3 nm.

Figure 4-26: XRD q -2q scan (CuK_a) of an MgO film (FN072399) grown on b-SiC /Si(001). The MgO(002) peak at 43° is shown with the Si(002) and Si(004) peaks indicating the cube on cube epitaxy of the film with the substrate.

Figure 4-27: XRD (CuK_a) rocking curve about the MgO (002) peak for an 18 nm thick film (FN071300) grown on b-SiC /Si(001). The line is a Gaussian curve fit to the data with a full width at half maximum (FWHM) of 2.8°.

Figure 4-28: TEM plan view image of an MgO film (FN101399) grown on b-SiC/Si(001). The inset shows the spot diffraction pattern of the heterostructure indicating the cube on cube epitaxy of the film with the b-SiC interlayer and the Si substrate.

Figure 4-29: TEM image showing a cross-sectional view of the interfaces between an MgO overlayer, a b-SiC interlayer, and the underlying Si(001) substrate (FN101399). In the lower left corner of the image a pit-defect formed in the Si can be seen. The defect is the result of Si out-diffusion during growth of the b-SiC layer.

Figure 4-30: High resolution TEM image showing a cross-sectional view of the interface between an MgO film and the underlying b-SiC/Si(001) interlayer (FN101399). The epitaxy of the MgO film with the interlayer can be clearly seen along with the twin band structures in the b-SiC.

Figure 4-31: Growth rate of MgO on b-SiC/Si(001) as a function of MgO growth temperature. Two distinct regimes are observed: (I) polycrystalline films, growth rate decreases with increasing temperature; (II) epitaxial films, growth rate increases with increasing growth temperature. The line is a guide to the eye. Films shown are, in order of increasing temperature: FN101599, FN101499, FN100899, FN101399.

Figure 4-32: Growth rate of MgO on b-SiC/Si(001) as a function of metal-organic (MO) precursor flux. The crystal structure of films grown above 0.5 x 10^-6 Torr of MO were epitaxial; below this pressure the films were textured or amorphous. The remainder of the growth parameters are listed in the figure. The line is a guide to the eye. Films shown are, in order of increasing MO flux: FN060800, FN053100, FN120899, FN051000, FN121699, FN120999.

Figure 4-33: Growth rate of MgO on b-SiC/Si(001) as a function of RF oxygen plasma excitation power. All of the films produced were epitaxial and growth rate increases linearly with increasing RF power. The rest of the growth parameters are listed in the figure. Films shown are, in order of increasing RF power: FN111899, FN120899, FN050900, FN051000, FN121566, FN071900, FN080200.

Figure 4-34: Series of AFM images (10 mm x 10 mm) for 70 nm thick MgO films grown on various thicknesses of b-SiC/Si(001). Shown for each film is the elevation data (left) and the corresponding deflection data (right). RMS roughness of the MgO overlayer, s_RMS, and thickness of the SiC interlayer, h_SiC, of each film is shown above the image. Deposition conditions for the b-SiC layers were 850°C and 10^-9 Torr of MO. The films shown are, in order of decreasing b-SiC thickness: FN033000, FN040600, FN041300, FN040700.

Figure 4-35: RMS roughness of 70 nm thick MgO films grown on various thicknesses of b-SiC/Si(001) at high (n ) and low (l ) growth rates. The curves drawn are guides to the eye. The roughness of similar b-SiC films is shown in Figure 4-19. The shaded area corresponds to films that were polycrystalline as determined by RHEED. Films outside this region were epitaxial. The films for the low growth rate are, in order of decreasing b-SiC thickness: FN033000, FN032900 , FN040600, FN041300, FN040700. For the high growth rate the films are, in order of decreasing b-SiC thickness: FN030900, FN031600, FN031700, FN032300.

Figure 4-36: XRD q -2q scan (CuK_a) of a BaTiO₃ film grown on MgO/b-SiC /Si(001). The BaTiO₃(001) and (002) peaks are shown with the Si(004) and MgO(002) peaks indicating the cube on cube epitaxy of the film with the substrate. The BaTiO₃ film (AT406) was grown by MOCVD to a thickness of ~400 nm. The MgO (33 nm) and b-SiC (10 nm) were grown by MOMBE (FN100899).

Figure 4-37: XRD rocking (w) curve about the BaTiO₃ (002) reflection for a film grown on MgO/b-SiC /Si(001). The line drawn in a Gaussian curve fit with a full width at half maximum (FWHM) of 1.7°. The BaTiO₃ film (AT406) was grown by MOCVD to a thickness of ~400 nm. The MgO (70 nm) and b-SiC (10 nm) were grown by MOMBE (FN051000).

Figure 4-38: TEM image showing a cross-sectional view of the interfaces between a BaTiO₃, the MgO and b-SiC interlayers, and the underlying Si(001) substrate. In the lower right corner of the image a pit-defect formed in the Si can be seen. The defect is the result of Si out-diffusion during growth of the b-SiC layer. The BaTiO₃ film (AT346) was grown by MOCVD, and the MgO/b-SiC layers were grown by MOMBE (FN100899).

Figure 4-39: TEM spot diffraction pattern of a BaTiO₃ film grown on MgO/b-SiC /Si(001). The diffraction spots indicate the cube on cube epitaxy of the film with the MgO/b-SiC interlayer and the Si substrate. The BaTiO₃ film (AT346) was grown by MOCVD, and the MgO/b-SiC layers were grown by MOMBE (FN100899).

Figure 4-40: Schematic of the MOS structure used for dielectric characterization of the integrated BaTiO₃ thin films.

Figure 4-41: Capacitance at 100 kHz of an integrated BaTiO₃/MgO/b-SiC/Si(001) capacitor structure (AT346) as a function of DC bias voltage. The hysteresis is an indication the BaTiO₃ is in a FE state and the asymmetry is characteristic of the metal oxide semiconductor (MOS) structure. The BaTiO₃ film was grown by MOCVD to a thickness of ~400 nm. The MgO (33 nm) and b-SiC (10 nm) were grown by MOMBE (FN100899).

Figure 4-42: Polarization of an integrated BaTiO₃/MgO/b-SiC/Si(001) capacitor structure (AT346) as a function of DC bias voltage. The hysteresis is an indication the BaTiO₃ is in a FE state. The BaTiO₃ film was grown by MOCVD to a thickness of ~400 nm. The MgO (33 nm) and b-SiC (10 nm) were grown by MOMBE (FN100899).

Figure 4-43: Capacitance at 100 kHz of an integrated BaTiO₃/MgO/b-SiC/Si(001) capacitor structure (AT346) as a function of temperature. The maximum at 127°C is an indication of the FE to PE transition. The BaTiO₃ film was grown by MOCVD to a thickness of ~400 nm. The MgO (33 nm) and b-SiC (10 nm) were grown by MOMBE (FN100899).
 

Table 4–1: Tabulation of b-SiC films grown on n-type Si(001). All films were grown with Mg(acac)₂ as the metal-organic precursor after desorbtion of SiO₂ at 850°C for ½ hr and cooling to room temperature. Oxygen was not used during these depositions.
 

Table 4–1 (continued): Tabulation of b-SiC films grown on n-type Si(001). All films were grown with Mg(acac)₂ as the metal-organic precursor after desorbtion of SiO₂ at 850°C for ½ hr. and cooling to room temperature. Oxygen was not used during these depositions.
 

Table 4–2: Tabulation of b-SiC films grown on n-type Si(001) using a 30 minute pre-deposition period. All films were grown with Mg(acac)₂ as the metal-organic precursor after desorbtion of SiO₂ at 850°C for ½ hr. Oxygen was not used during these depositions.
 

Table 4–3: Tabulation of MgO films grown on 30 nm of b-SiC deposited using a 30 min. pre-deposition at 900°C and 10^-7 Torr of Mg(acac)₂ with 15 min. of actual growth time. All films were grown with Mg(acac)₂ as the metal-organic precursor after desorbtion of SiO₂ at 850°C for ½ hr and cooling to room temperature.
 

Table 4–4: Tabulation of MgO films grown on b-SiC using the standard deposition technique. All films were grown with Mg(acac)₂ as the metal-organic precursor after desorbtion of SiO₂ at 850°C for ½ hr and cooling to room temperature.
 

Table 4–4 (continued): Tabulation of MgO films grown on b-SiC/Si(001). All films were grown with Mg(acac)₂ as the metal-organic precursor after desorbtion of SiO₂ at 850°C for ½ hr and cooling to room temperature.
 

5. Summary and Conclusions

This work has shown that epitaxial MgO can be deposited on Si(100) substrates using the MOMBE growth technique in conjunction with a thin b-SiC interlayer. Moreover, these MgO films are suitable for use as the interlayer in ferroelectric heterostructures, as demonstrated by the successful deposition of FE BaTiO₃/MgO/b-SiC/Si(100).
 

6. References

1 C. Basceri, S. K. Streiffer, A. I. Kingon, S. Bilodeau, R. Carl, P. C. van Buskirk, S. R. Summerfelt, P. McIntyre and R. Waser, M.R.S. Symp. Proc. 433, 285 (1996).

2 O. Auciello, J. F. Scott and R. Ramesh, in Physics Today; Vol. July (1998).

3 W. D. Kingery, H. K. Bowen and D. R. Uhlmann, Introduction to Ceramics, Second ed. (Wiley, New York, 1976).

4 G. H. Haertling, Ferroelectrics 75, 25 (1987).

5 M. Bass, Handbook of optics, Vol. 1, 2 ed. (McGraw-Hill, New York, 1995).

6 F. J. Walker and R. A. McKee, Nanostructured Materials 7, 1 (1996).

7 D. Fork, F. Leplingard and J. Kingston, MRS Bulletin 102, 53 (1996).

8 M. Yoshimoto, H. Nagata, T. Tsukahara and H. Koinuma, Japaneese journal applied physics 29, 1199 (1990).

9 R. L. Holman, L. M. Althouse Johnson and D. P. Skinner, Opt. Eng. 26, 134 (1987).

10 E. J. Denlinger, IEEE Trans. Microwave Theory and Techniques MTT-28, 513 (1980).

11 G. M. Ford, Radiative and non-radiative processes in heteroepitaxial rare earth doped barium titanate thin films for photonic applications. PhD Thesis, Northwestern University, (1998).

12 B. A. Block, Optical properties of rare earth doped barium titanate thin films prepared by metalorganic vapor deposition. PhD Thesis, Northwestern University, (1997).

13 M. J. Nystrom, Structural and optical properties of stontium barium niobate and potasium niobate thin films prepared by metalorganic chemical vapor deposition. PhD Thesis, Northwestern University, (1996).

14 F. J. Walker, R. A. McKee, Y. Huan-wun and D. E. Zelmon, Appl. Phys. Lett. 65, 1495 (1994).

15 A. Yamada, H. Tamada and M. Saitoh, Appl. Phys. Lett. 61, 2848 (1992).

16 D. M. Gill, B. A. Block, C. W. Conrad, B. W. Wessels and S. T. Ho, Appl. Phys. Lett. 69, 2968 (1996).

17 D. M. Gill, C. W. Conrad, G. Ford, B. W. Wessels and S. T. Ho, Appl. Phys. Lett. 71,  1783 (1997).

18 M. Ohring, The materials science of thin films (Academic Press Limited, San Diego, CA, 1992).

19 K. J. Hubbard and D. G. Schlom, J. Mater. Res. 11, 2757 (1996).

20 M. Tonouchi, Y. Sakaguchi and T. Kobayashi, J. Appl. Phys. 62, 961 (1987).

21 F. J. Walker, R. A. McKee, S. J. Pennycook and T. G. Thundat, M.R.S. Symp. Proc. 401, 13 (1996).

22 L. L. H. King, K. Y. Hsieh, D. J. Lichtenwalner and A. I. Kingon, Appl. Phys. Lett. 59, 3045 (1991).

23 R. A. McKee, F. J. Walker and M. F. Chisholm, Phys. Rev. Lett. 81, 3014 (1998).

24 S. Yadavalli, M. H. Yang and C. P. Flynn, Phys. Rev. B 41, 7961 (1990).

25 W. T. Tsang, J. Crystal Growth 95, 1 (1989).

26 W. T. Tsang, J. Crystal Growth 107, 1 (1991).

27 P. R. Willmott, P. Felder, M. Lingenauer, J. R. Huber, E. Fritsch, W. Bidell, H. Berke, E. Machler, E. J. Williams, J. P. Locquet and J. G. Bednorz, J. Vac. Sci. Tech. A 13, 248 (1995).

28 K. Endo, Y. Moriyasu, S. Misawa and S. Yoshlda, Physica C 1991, 1995 (1991).

29 J. P. Bade, E. A. Baker, A. I. Kingon, R. F. Davis and K. J. Bachmann, J. Vac. Sci. Tech. B 8, 327 (1990).

30 E. M. Machler, F. Arrouy, E. Fritsch, J. G. Bednorz, H. Berke, J. R. Huber and J. P. Locquet, Appl. Phys. Lett. 71, 710 (1997).

31 S. Ikegawa and Y. Motoi, Thin Solid Films 281, 1 (1996).

32 K. Hayama, T. Togun and M. Ishida, J. Crystal Growth 179, 3 (1997).

33 H. Iizuka, K. Yokoo and S. Ono, Appl. Phys. Lett. 61, 2978 (1992).

34 K. Sawada, M. Ishida, T. Nakamura and T. Suzaki, J. Crystal Growth 95, 494 (1989).

35 L. A. Wills, W. A. Feil, B. W. Wessels, L. M. Tonge and T. J. Marks, J. Crystal Growth 107, 712 (1991).

36 S. Kim and S. Hishita, J. Mater. Res. 12, 1152 (1997).

37 J. Zhang, C. P. J. Beetz and S. B. Krupanidhi , Appl. Phys. Lett. 65, 2410 (1994).

38 J. R. Hanrahan, E. Sanchez, J. J. Santiago, D. H. Berry and Q. Jiang, Thin Solid Films 202, 235 (1991).

39 Q. X. Jia, Z. Q. Shi and W. A. Anderson, Thin Solid Films 209, 230 (1992).

40 L. H. Chang and W. A. Anderson, M.R.S. Symp. Proc. 361, 489 (1995).

41 W. A. Feil, B. W. Wessels, L. M. Tonge and T. J. Marks, J. Appl. Phys. 67, 3858 (1990).

42 C. H. J. Huang, H. Lin and T. A. Rabson, Ferroelectrics 157, 365 (1994).

43 E. Fujii, A. Tomozawa, S. Fujii, H. Torii, M. Hattori and R. Takayama, Jpn. J. Appl. Phys 32, L1448 (1993).

44 D. H. Reitze, E. Haton, R. Ramesh, S. Etemad, D. E. Leaird, T. Sands, Z. Karim and A. R. T. Jr., Appl. Phys. Lett. 63, 596 (1993).

45 B. S. Kwak, E. P. Boyd, K. Zhang, A. Erbil and B. Wilkins, Appl. Phys. Lett. 54, 2542 (1989).

46 C. Jiangong, M. Xiangjian, Y. Pingxong and C. Junhao, J. Crystal Growth 194, 89 (1998).

47 R. Huang and A. H. Kitai, Appl. Phys. Lett. 61, 1450 (1992).

48 X. Fu, G. Wu, S. Song, Z. Song, X. Duo and C. Lin, App. Surf. Sci. 148, 223 (1999).

49 P. Tiwari, S. Sharan and J. Narayan, J. Appl. Phys. 69, 8358 (1991).

50 T. Ishiguro, Y. Hiroshima and T. Inoue, Jpn. J. Appl. Phys. 35, 3537 (1996).

51 D. K. Fork, F. A. Ponce, J. C. Tramontana and T. H. Geballe, Appl. Phys. Lett. 58, 2294 (1991).

52 S. Kim and S. Hishita, Thin Solid Films 281, 449 (1996).

53 Y. Kaneko, N. Mikoshiba and T. Yamashita, Japn. J. Appl. Phys. 30, 1091 (1991).

54 H. Z. Durusoy, Journal of Materials Science Letters 10, 1023 (1991).

55 M. Ishida, I. Katakabe, T. Nakamura and N. Ohtake, Appl. Phys. Lett. 52, 1326 (1988).

56 M. Ihara, Y. Arimoto, M. Jifuku, T. Kimura, S. Kodama, H. Yamawaki and T. Yamaoka, J. Electrochem. Soc. 129, 2569
(1982).

57 R. A. McKee, F. J. Walker, J. R. Conner, E. D. Specht and D. E. Zelmon, Appl. Phys. Lett. 59, 782 (1991).

58 J. Hwang, Y. Fang, Y. Song and D. Yaung, Japanese Journal of Applied Physics Part 1-Regular Papers Short Notes & Review Papers 34 (1995).

59 H. Nagata, M. Yoshimoto and H. Koinuma, J. Crystal Growth 118, 299 (1992).

60 H. Fukumoto, T. Imura, Y. Osaka and F. Nishiyama, J. Appl. Phys. 66, 616 (1989).

61 Y. Zhao and H. Suhr, Applied Physics A 54, 451 (1992).

62 R. Hiskes, S. A. DiCarolis, R. D. Jacowitz, Z. Lu, R. S. Feigelson, R. K. Route and J. L. Young, J. Crystal Growth 128, 781 (1993).

63 S. Fujii, A. Tomozawa, E. Fujii, H. Torii, R. Takayawa and T. Hirao, Appl. Phys. Lett 65, 1463 (1994).

64 J. M. Zeng, H. Wang, S. X. Shang, Z. Wang and M. Wang, J. Cryst. Growth 169, 474 (1996).

65 S. A. Chambers, T. T. Tran and T. A. Hileman, J. Mater. Res. 9, 2944 (1994).

66 F. J. Walker and R. A. McKee, MRS Symp. Proc. 597 (1999).

67 D. M. Hwang, R. Ramesh, C. Y. Chen, X. D. Wu, A. Inam, M. S. Hegde, B. Wilkens, C. C. Chang, L. Nazar and T. Venkatesan, J. Appl. Phys. 68, 1772 (1990).

68 T. Martin, C. R. Whitehouse and P. A. Lane, J. Crystal Growth 107, 1 (1991).

69 M.-B. Lee, M. Kawasaki, M. Yoshimoto, B.-K. Moon and Ishiwara, Ferroelectric Thin Films IV Materials Research Society Symposium Proceedings 361, 521 (1995).

70 M.-B. Lee, M. Kawasaki, M. Yoshimoto and H. Koinuma, Appl. Phys. Lett. 66, 1331 (1995).

71 M. B. Lee, Jpn. J. Appl. Phys. 34, 808 (1995).

72 S. Summerfelt, M.R.S. Symp. Proc. 221, 29 (1991).

73 X. L. Guo, Z. G. Liu, S. N. Zhu, S. B. Xiong, Y. Y. Zhu and J. M. Liu, J. Crystal Growth 167, 1 (1996).

74 X. L. Guo, Z. G. Liu, X. Y. Chen, S. N. Zhu, S. B. Xiong, W. S. Hu and C. Y. Lin, Journal of Physics D Applied Physics 29,
1632 (1996).

75 W. Jo, H. J. Cho, T. W. Noh, B. I. Kim, D. Y. Kim, Z. G. Khim and S. I. Kwun, Appl. Phys. Lett. 63, 2198 (1993).

76 A. Masuda, Y. Yamanaka, M. Tazoe, Y. Yonezawa, A. Morimoto and T. Shimizu, Jpn. J. Appl. Phys. 34, 5154 (1995).

77 K. Nashimoto, D. K. Fork and T. H. Geballe, Appl. Phys. Lett. 60, 1199 (1992).

78 K. Nashimoto, D. K. Fork, F. A. Ponce and J. C. Tramontana, Japanees Journal of Applied Physics 32, 4099 (1993).

79 A. Kingon, K. Hsieh, L. King, S. Rou, K. Bachmann and R. Davis, MRS res soc  symp proc 200 (1990).

80 F. Niu, B. H. Hoerman and B. W. Wessels, Appl. Surf. Sci. 161, 74 (2000).

81 A. R. Teren, F. Niu, B. H. Hoerman and B. W. Wessels, To be submitted (2000).

82 A. R. Teren, PhD Thesis, Northwestern University, (2001).

83 K. D. Preston and G. H. Haertling, Appl. Phys. Lett. 60, 2831 (1992).

84 D. K. Fork, Superconducting thin films on epitaxial magnesium oxide grown on silicon.  Patent US5418216, 1995.

85 N. Aoki, N. Fujimura, T. Yoshimura and T. Ito, J. Crystal Growth 174, 1 (1997).

86 W. A. Krivsky and R. Schuhmann, Trans. Metal. Soc. 221, 898 (1961).

87 F. Danes, E. Saint-Aman and L. Coudurier, J. Mater. Sci. 28, 489 (1993).

88 F. Niu, B. H. Hoerman and B. W. Wessels, M.R.S. Symp. Proc. 606 (1999).

89 M. Kitabatake, M. Deguchi and T. Hirao, J. Appl. Phys. 74, 4438 (1993).

90 Q. Wahab, M. Sardela, L. Hultman, A. Henry, M. Willander, E. Janzen and J. Sundgren, Appl. Phys. Lett. 65, 725 (1994).

91 K. Zekentes, R. Callec, K. Tsagaraki, B. Sagnes, G. Arnaud, J. Pascual and J. Camassel, Mater. Sci. Eng. B 29, 138 (1995).

92 A. Addamiano and J. A. Sprague, Appl. Phys. Lett. 44, 525 (1984).

93 G. Ferro, Y. Monteil, H. Vincent, V. Thevenot, M. D. Tran, F. Cauwet and J. Bouix, J. Appl. Phys. 80, 4691 (1996).

94 A. J. Steckl and J. P. Li, Ieee Transactions On Electron Devices 39 (1992).

95 K. Zekentes, N. Becourt, M. Androulidaki, K. Tsagaraki, J. Stoemenos, J. M. Bluet, J. Camassel and J. Pascual, Appl. Surf. Sci. 102, 22 (1996).

96 L. Rimai, R. Ager, W. Weber, J. Hangas and B. Poindexter, Appl. Phys. Lett. 65 (1994).

97 M. Shigeta, A. Suzuki, K. Furukawa, Y. Fujii, A. Hatano, A. Uemoto and K. Nakanishi, Sharp Kabushiki Kaisha, Heteroepitaxial growth of SiC on Si.  Patent 1989.

98 K. Zekentes, V. Papaioannou, B. Pecz and J. Stoemenos, J. Crystal Growth 157 (1995).

99 T. Yoshinobu, M. Nakayama, H. Shiomi, T. Fuyuki and H. Matsunami, J. Crystal Growth 99 (1990).

100 V. Cimalla, T. Stauden, G. Eichhorn and J. Pezoldt, Mat. Sci. Eng. B 61, 553 (1999).

101 S. Sasaki, Y. Itagaki, T. Kurokawa and K. Nakanishi, Bull. Chem. Soc. Jpn. 40, 76 (1967).

102 C. G. Macdonald and J. S. Shannon, Aust. J. Chem. 19, 1545 (1966).

103 E. Ech-chamikh, E. L. Ameziane, A. Bennouna, M. Azizan, T. A. Tan and T. Lopez-Rios, Thin Solid Films 259, 18 (1995).

104 R. A. McKee and F. J. Walker, Martin Marietta Energy Systems, Inc., Process for depositing epitaxial alkaline earth oxide onto a substrate and structures prepared with the process.  Patent US5482003, 1996.

105 R. A. McKee and F. J. Walker, Martin Marietta Energy Systems, Inc, Process for depositing an oxide epitaxially onto asilicon
substrate and structures prepared with the process.  Patent US005482003A, 1993.

106 K. Kim, S. Choi and K. Wang, Journal of Vacuum Science & Technology B 10 (1992).

107 U. Kaiser, S. B. Newcomb, W. M. Stobbs, M. Adamik, A. Fissel and W. Richter, J. Mater. Res. 13 (1998).

108 S. Nishino, H. Suhara, H. Ono and H. Matsunami, J. Appl. Phys. 61 (1987).

109 M. Diani, A. Mesli, L. Kubler, A. Claverie, J. Balladore, D. Aubel, S. Peyre, T. Heiser and J. Bischoff, Mater. Sci. Eng. B 29 (1995).

110 F. Niu, B. H. Hoerman and B. W. Wessels, (Unpublished) (1999).

111 Y. Li, G. C. Xiong, G. J. Lian, J. Li and Z. Gan, Thin Solid Films 223, 11 (1993).

112 S. Chattopadhyay, A. Teren and B. W. Wessels, Mat. Res. Soc. Symp. Proc.  541,  489 (1999).

113 F. Niu, B. H. Hoerman and B. W. Wessels, M.R.S. Symp. Proc. (In Press) (2000).
 

7. Appendicies

7.1 MOMBE process monitoring program

Not Included. Sorry.

7.2 Miscellaneous MOMBE pictures



















7.3 Documented MOMBE procedures

7.3.1 Chamber Bakeout

• Ensure wafer holder is removed from growth head and shutter is closed.
• Shut off supply and return water for system (growth head and RF matching unit).
• Shut off the process gas for the shutter actuators.
• Remove all unbakeable items from MBE.

• Three waterlines (drain) and Swagelock O-ring fittings (2)
• Sample rotation motor and belt
• Pauls actuator lever (two screws)
• Power cord and thermocouple for sample head
• Rotation magnet for sample head
• Shutter actuators (return screws to shutter posts)
• Power to fan on RF matching unit
• RF power cable to matching unit
• N₂ vent hoses to load-lock
• High pressure gauge on load-lock
• Five RHEED gun cables
• Liquid N₂ supply, routing, and drain tubing.
• RGA electronics

• Reconnect power cord and thermocouple for sample head.
• Fold out heater bank under growth chamber.
• Shutter all ports and cover exterior with Al foil.
• Place largest shroud over growth head and main turbo-pump area.
• Insert support rods.
• Slide bottom shroud under and around K-cells and around load-lock chamber.
• Place final shroud between K-cells.
• Fasten all connectors, snaps, ties, and clips.
• Stop up remaining holes with Al foil.
• Ramp-up K-cell and stage heaters.
• Set bake time and temperature on controllers at cabinet.
• Throw both start switches and monitor the chamber temperature and pressure.
• Adjust room thermostat if necessary.

7.3.2 Dismounting the Wafer Holder

• Ensure all source shutters are closed, to protect the K-cells, and the sample shutter is open.
• Ensure the gate valve between the load-lock and growth chambers is completely open.
• Slowly slide the transfer arm through the load-lock in the horizontal position until the cassette area has been cleared.
• Before entry into the growth chamber, tilt the platen until the stop is reached.
• Center the growth manipulator in the x-y plane.
• Rotate the pauls to ensure they are in the correct orientation. [ The stage can only be rotated while the pauls are engaged. ]
• Raise the growth manipulator head with the z-axis control to avoid collision with the platen.
• Carefully slide the platen under the growth head and wafer holder. Maintain the tilt of the platen and avoid contact with the pauls or the head. Check the position through monitor ports.
• Position the platen directly below the wafer holder and center both in the x-y plane. Check the position through monitor ports.
• Slowly move the growth manipulator head down with the z-axis control.
• Continue until there is minimal separation between wafer holder and platen. The wafer holder should just fit into the cut-out and buttons on the platen Check the position through monitor ports.
• Slowly begin to disengage the pauls (pauls actuator lever -- down). Ensure the wafer holder is seated in the platen.
• Fully disengage the pauls.
• Raise the growth manipulator head slowly until the wafer holder is cleared.
• Slide the platen and wafer holder back, past the growth manipulator head.
• Return platen to the horizontal position and continue to withdraw until clear of the gate valve.

 7.3.3 Film Growth Procedure

• Ensure the vent N₂, process N₂, and source O₂ tanks are on and there is sufficient pressure.
• Ensure the pump and system water supplies are on and the flow rates are correct.
• Ensure that all shutters are closed.
• Fill LN₂ dewar (takes 1/2 to 1 hr) and hook up to system. Half of a tank should last 8 hours.
• Clean and etch a Si wafer.
• While the wafer is etching, vent the load-lock according to documentation.
• When load-lock is up to atmospheric pressure: dry wafer with N₂, set in holder, place in the load-lock, and pump down load-lock.
• Start LN₂ flow. Maintain a steady rate. Continual flushing during the growth may be necessary.
• Begin degassing the RHEED gun filament according to documentation.
• Start RoboMBE or other necessary software.
• Transfer the wafer holder to the stage. Take care to avoid collision with the shutter, stage, or flux monitor. Shut the gate valve.
• When LN₂ flow is steady, begin degassing the stage using program number one.
• The RHEED gun can be brought to operating voltage and current.
• When pressure recovers from degassing, begin the stage ramp to desorbtion temperature using program number two.
• Monitor the RHEED pattern for reconstruction and the pyrometer for accurate temperature measurement. Ensure the sample shutter remains closed after the pyrometer is used.
• After pressure has balanced at the desorbtion temperature and the RHEED pattern has cleared, begin to ramp the stage to growth temperature using program number three.
• Stage rotation may begin.
• After the stage is at growth temperature, begin warming the MO source.
• Monitor the MO source flux with the flux monitor at the 100 mm position. Be sure to avoid the shutter and stage. The source shutter can be momentarily opened to check the flux.

 7.3.4 Loading a Sample

 7.3.5 Mounting the Wafer Holder

7.3.6 Oxygen Gun Operation

• Open system supply water and return valves and ensure the water flow rate is at or above 0.25 GPM.
• Ensure source and sample shutters are in proper positions.
• Depress "power" button on RF power supply to energize.
• The "intlk" light on the supply should remain illuminated. If "intlk" flashes, the water flow is low; correct the flow and turn the RF power supply off then on.
• The power level should be turned to minimum. Depress the "RF on" button.
• Slowly bring up the RF power level until the forward power is ~ 50 W.
• Adjust the capacitor on the matching network to minimize the reflected power. The minimum will be very sharp.
• Repeat steps 7 and 8 at ~50 W increments until the forward power is 300 W.
• Let the system settle for at least 15 min. With no O₂ flow the reflected power should read ~30 W.
• Open the O₂ supply tank and set pressure to ~10 psi.
• Adjust needle valve on the O₂ gun to increase flow (counter-clockwise). Monitor growth chamber pressure and flux monitor.
• When chamber pressure begins to increase, the matching capacitor must be readjusted to minimize the reflected power (~2 W).
• At ~3 x 10^-5 Torr, a plasma discharge should be visible through the monitor port at the back of the gun.
• If the plasma is not visible, snap the red ignition button on the matching unit until the plasma is lit.
• The RF power (100 to 500 W) and flow rate (10^-5 to 10^-4 Torr) can now be adjusted to growth parameters.

• Slowly bring the RF power level down ~ 50 W.
• Adjust the capacitor on the matching network to minimize the reflected power (< 10 % of FWD power).
• Repeat steps 1 and 2 until the power is minimized.
• Depress "RF OFF" button, depress "power" button to de-energize.
• Close the needle valve on the gun (clock-wise) and shut off the O₂ supply tank.

7.3.7 RHEED Operation

• Turn on RHEED power supply.
• Ensure that all voltages and currents read as close to zero as possible. The filament current will not go below » 0.45 A.
• Slowly bring the filament current to 1.2 A. Let the filament out-gas for 1 to 1.5 hours.
• Slowly ramp the filament current to 2.25 A and then bring the beam voltage to 1.1 kV.
• By adjusting the x- and y-deflection voltages toward their centers (4-6 on gauges) the primary beam should be visible a blue spot on the RHEED screen when its shutter is open.
• Focusing of the beam is accomplished by raising the objective voltage. Do not set the objective higher than the beam voltage as arcing may occur.
• The emission current should remain below 1 mA. Keep the screen shutter closed whenever possible to avoid burn-in.

• Set the growth manipulator near the center of the x, y, and z ranges.
• Adjust the x-deflection until the beam spot becomes blocked by the substrate / growth head.
• Carefully ramp the beam voltage to the operating level. By constantly re-adjusting the x-deflection, keep the primary beam blocked by the substrate.
• By adjusting the x-deflection and z-height controls the reflected beam can be aimed at the screen.
• Rotation of the growth head / substrate will enable imaging of the correct RHEED pattern.

• Slowly ramp the beam voltage to 1.1 kV. Using the x- and y-deflection plates, center the primary beam on the screen.
• Close the screen shutter.
• Slowly ramp down the beam voltage to zero. Ensure the objective voltage remains lower than the beam.
• Slowly ramp down the beam current to its baseline level.
• Turn off the RHEED power supply.

7.3.8 Venting the Main Chamber

• Ensure the N₂ (Air) vent gas supply to the MBE is available and open.
• Open the large gate valve between the main chamber and the load-lock.
• Turn off ion gauges in load-lock and growth chamber.
• Turn off flux monitor and ensure the CG high pressure gauge is operational.
• Set the N₂ (Air) vent gas pressure to 3 psi at the gauge below the small turbo pump.
• Close the butterfly valve between small turbo pump and load-lock.
• Shut off small turbo pump.
• Shut off main turbo pump.
• When the main turbo pump begins to vent (loud "pop"), shut off roughing pump at the electrical cabinet.
• Bleed vent gas separately to load-lock chamber with the green Swagelock valve. (optional)

7.3.9 Pumping Down the Main Chamber

• Ensure the green Swagelock valve on the load-lock chamber is closed.
• Close off the N₂ (Air) vent gas pressure at the gauge below the small turbo pump.
• Ensure all ports are closed and the load-lock door is shut and screwed down.
• Open the butterfly valve between small turbo pump and load-lock.
• Switch on the small turbo pump.
• Turn the roughing pump on at the electrical cabinet.
• Switch on the main turbo pump.
• After pressure falls below 10^-4 Torr, turn on the ion gauges in load-lock and growth chamber. Turn on flux monitor.