Amorphous matrix effects on silicon nanoparticles

AMORPHOUS MATRIX EFFECTS ON SILICON NANOPARTICLES

by

c

Copyright by Connor P. Pierce, 2019 All Rights Reserved

A thesis submitted to the Faculty and the Board of Trustees of the Colorado School of Mines in partial fulfillment of the requirements for the degree of Master of Science (Applied Physics). Golden, Colorado Date Signed: Connor P. Pierce Signed: Dr. K. Xerxes Steirer Thesis Advisor Golden, Colorado Date Signed: Dr. Uwe Greife Professor and Head Department of Physics

ABSTRACT

The search for new materials is a vital endeavor, as the world demands more from the technologies available, new technologies must be developed. As current semiconductors are pushed to their operating limits, the need for new wider bandgap semiconductor technology has become apparent. In this work, the quantum confinement of silicon quantum dots (SiQDs) in various semiconductor materials was examined. The chosen materials were the wide bandgap semiconductor amorphous silicon carbide (SiC) as well as amorphous silicon synthesized using plasma enhanced chemical vapor deposition (PECVD). Amorphous silicon carbide is an interesting material as it has a similar band structure to silicon while functioning in higher temperature and operating power conditions due to it’s larger bandgap. The measurement of confinement of the dots was attempted using photoluminescence (PL) and spectroscopic ellipsometry (SE). In addition to measuring quantum confinement, since SiC had not yet been synthesized using the PECVD system in this work, a literature review was conducted to both determine the feasibility and capabilities of growing the SiC films. After the review was finished, the PECVD system was modified to allow for the growth of SiC films both with and without SiQDs. Once growth of the films was complete, the films had to be characterized to determine if the growths were successful. This was accomplished using Fourier transform infrared (FTIR) spectroscopy and SE. Growth and characterization of the SiC films was successful. Growth of SiC films containing SiQDs was also accomplished in this work, while the measurement of confinement of the SiQDs within yielded varying levels of success. Quantum confinement was successfully observed with the SiQDs bandgap increasing to between 1.2 and 1.3 eV, however no significant differences in bandgap were observed as the surrounding matrix was changed.

TABLE OF CONTENTS ABSTRACT . . . iii LIST OF FIGURES . . . vi LIST OF TABLES . . . ix LIST OF SYMBOLS . . . x LIST OF ABBREVIATIONS . . . xi ACKNOWLEDGMENTS . . . xii DEDICATION . . . xiii CHAPTER 1 INTRODUCTION . . . 1 1.1 Motivation . . . 1 1.2 Objectives . . . 3 CHAPTER 2 BACKGROUND . . . 4 2.1 Semiconductor Physics . . . 4

2.2 Energy Levels of Silicon Quantum Dots . . . 5

2.3 Quantum Confinement Effect . . . 7

CHAPTER 3 EXPERIMENTAL METHODS . . . 10

3.1 PECVD Synthesis . . . 10

3.1.1 PECVD System . . . 10

3.1.2 Silicon Carbide Synthesis . . . 11

3.2 Optical Characterization Methods . . . 15

3.2.2 Photoluminescence Spectroscopy . . . 17

3.2.3 Photoluminescence System . . . 18

3.2.4 Spectroscopic Ellipsometry . . . 22

CHAPTER 4 MATERIAL CHARACTERIZATIONS . . . 27

4.1 Qualitative Carbon Incorporation in a-Si and a-Si1 – xCx films . . . 27

4.2 Changes in the Amorphous Matrix Chemical Bonding as a Result of Methane Incorporation . . . 30

4.3 Determination of Bandgap for a-Si and a-Si1 – xCx films . . . 32

CHAPTER 5 ATTEMPTING TO MEASURE THE QUANTUM CONFINEMENT EFFECT . . . 36

5.1 Photoluminescence measurements of quantum confinement in SiQDs . . . 36

5.2 Spectroscopic ellipsometry as a measurement method for quantum confinement . . . 40

CHAPTER 6 CONCLUSIONS & FURTHER WORK . . . 48

6.1 Summary . . . 48

6.2 Future Work . . . 49

REFERENCES CITED . . . 50

LIST OF FIGURES

Figure 2.1 Confinement parameter behavior with varying QD size and desired emission wavelength in a 1.1 eV bandgap confining material, as both the QD size and emission increase the confinement parameter must also

increase. . . 9 Figure 3.1 PECVD Reactor Co-Deposition growth chamber, the amorphous film

growth occurs from the laterally flowing precursor gas (orange arrow) while SiQD synthesis occurs below the substrate and flows up to be

deposited (green arrow). . . 12 Figure 3.2 Pressure vs Temperature plot, blue and red dots represent papers that

used a range of temperatures to indicate the low and high temperatures in their range. The green point indicates the average across all sources and the yellow point indicates work done on similar equipment to this work . The blue point indicates the substrate temperature and chamber pressure used in this work. . . 13 Figure 3.3 Gas Ratio vs Temperature plot, the green point indicates the average

across all sources. The yellow point indicates the values used on a system similar to the one from this work. The blue x points indicate the values used for this work. (Note: The pure a-Si parameters are not

shown since the ratio is 1:0 SiH4:CH4) . . . 14

Figure 3.4 A typical excitation scheme for PL in a direct semiconductor system with the horizontal lines indicating the individual energy states available in each band. An incoming photon excites an electron from the valence band into the conduction band leaving a hole in the valence band behind, then before de-exciting back into the valence band, the electron relaxes to the bottom of the conduction band while the hole relaxes to the top of the valence band, both assisted by a phonon to conserve momentum. Then the recombination occurs and a photon is emitted.

This diagram demonstrates . . . 19 Figure 3.5 Temperature dependence on photoluminescence of a-SiC, with the colors

indicating the various temperatures, black is room temperature, red is 270 K, purple is 200 K, green is 150 K, and blue is 100 K. As the temperature decreases the emission shifts and condenses as the

Figure 3.6 Overhead view of the PL system, the green line indicates the path of the excitation laser, while the red line indicates the paths taken by the

emitted light from the sample. . . 21 Figure 3.7 Typical set up of an ellipsometry measurement . . . 24 Figure 3.8 Absorption coefficient plot for a crystalline silicon wafer, the Tauc line

provides the ability to extract the bandgap, approximately 1.1 eV in

this case. . . 26 Figure 4.1 FTIR spectra for the produced films with the gas ratio used indicated.

An increase in the peaks related to carbon are seen as the amount of

methane used increased. . . 29 Figure 4.2 Zoom in on FTIR spectra for the amorphous films, with an emphasis on

the shift of the SiHn peak from SiH to SiH2 as a result of an increase in

methane during film growth. . . 31 Figure 4.3 Raw (Colored) and Modeled (Dashed) ellipsometry data for the a-Si film. . 33 Figure 4.4 Acquired dielectric functions for a-Si, ǫ1 (red) and ǫ2 (green). . . 34 Figure 4.5 Generated absorption curve for a-Si thin film, with an extracted

bandgap of 1.53 eV. . . 34 Figure 4.6 Collection of the absorption curves for the different gas ratio films

allowing for the bandgap of each film to be extracted. . . 35 Figure 5.1 Approximate PL Measurement locations on sample for scan across

sample. . . 38 Figure 5.2 PL spectra acquired while scanning across the a-SiC sample grown with

2:1 gas flow ratio (SiH4:CH4). Locations seen in Figure 5.1. . . 38

Figure 5.3 Captured PL from various samples with little difference besides amplitude, suggesting a different method for measuring the quantum

confinement of the SiQDs is needed. . . 39 Figure 5.4 Approximate SE measurement locations as the sample was scanned

across. . . 44 Figure 5.5 Absorption coefficients for scanned across sample of a-Si1 – xCx

Figure 5.6 Simulated absorption coefficients for an EMA of a-Si and c-Si acting as SiQDs where the percentage is the amount of c-Si present in the mixture and the remaining part being the a-Si matrix. A distinct transition from the a-Si spectrum to the c-Si spectrum is seen as the composition goes

from purely amorphous to purely crystalline . . . 46

Figure 5.7 Absorption plot with extracted bandgap for the SiQDs in the 2:1 SiH4:CH4 a-SiC matrix with bandgap 1.687 eV. The bandgap of the SiQDs has shifted to between 1.2-1.3 eV . . . 47

Figure 5.8 Absorption plot with extracted bandgap for the SiQDs in the a-Si matrix with bandgap 1.528 eV. The bandgap of the SiQDs has shifted to between 1.2-1.3 eV. . . 47

Figure A.1 Dielectric Functions for sample C658 . . . 59

Figure A.2 Dielectric Functions for sample C665 . . . 60

Figure A.3 Dielectric Functions for sample C666 at the center of the dot haze . . . 60

Figure A.4 Dielectric Functions for sample C666 at the edge of the dot haze . . . 61

Figure A.5 Dielectric Functions for sample C666 midway between the dot haze and the edge of the sample . . . 61

Figure A.6 Dielectric Functions for sample C666 at the edge of the sample . . . 62

Figure A.7 Dielectric Functions for sample C669 . . . 62

LIST OF TABLES

Table 3.1 Expected FTIR peaks for a-Si & a-SiC absorption spectra . The different modes indicated in the third column serve to give an idea on the nature

of the vibrational mode in question. . . 17

Table 4.1 Ratio of Peak Heights in the Synthesized Films . . . 28

Table 4.2 Extracted bandgap energy for the different gas ratio films. . . 35

LIST OF SYMBOLS

Absorption Coefficient . . . α Bandgap energy . . . Eg

LIST OF ABBREVIATIONS

Amorphous Silicon . . . a-Si Amorphous Silicon Carbide . . . a-SiC Crystalline Silicon . . . c-Si Effective Medium Approach . . . EMA Fourier Transform Infrared . . . FTIR Mean Square Error . . . MSE Nanocrystalline Silicon . . . nc-Si Plasma Enhanced Chemical Vapor Deposition . . . PECVD Silicon Quantum Dot(s) . . . SiQD(s) Spectroscopic Ellipsometry . . . SE Standard cubic centimeter per minute . . . sccm Wide Bandgap . . . WB

ACKNOWLEDGMENTS

This work would not have been possible without the endless support from my family and friends, thank you for pushing. I’m very grateful for my advisor, Dr. Xerxes Steirer for his encouragement and inspiration especially when progress was slow. Thank you to my committee members, Dr. Reuben Collins and Dr. Jeramy Zimmerman, for their advice as I stumbled through the many problems that arose during this work. Lastly, much thanks to the cohort of physics graduate students for their camaraderie. This journey would not have been nearly as fruitful without all of you. Physics provides the deepest insight into the inner-workings of our universe and I am forever grateful to be afforded the opportunity to investigate such a rewarding discipline at an institution of this caliber.

Remember to look up at the stars and not down at your feet. Try to make sense of what you see and wonder about what makes the universe exist. Be curious. And however difficult

life may seem, there is always something you can do and succeed at. It matters that you don’t just give up.

CHAPTER 1 INTRODUCTION

1.1 Motivation

Silicon is the most common semiconductor in use today, largely in part to the quality of it’s native oxide, a useful band gap, the ease of doping the silicon as well as the prevalence of silicon containing materials in the Earth’s crust. Silicon by itself is not conductive enough to be used in circuits, however through the fabrication process of doping, charge carriers can be introduced to the material allowing for electrical conduction to occur. This doping allowed for the many advances in electronics such as transistors and solar cells. This doping allows for the creation of a p-n junction, a layer of n-type and p-type silicon joined together, which is the basis for all microelectronics.

Silicon comes in a variety of phases: crystalline, amorphous, and nanocrystalline. Crys-talline silicon (c-Si), which has a periodic atomic structure, is used as the basis for nearly all microelectronic fabrication processes [1]. Wafers of silicon can be doped to become con-ductive with either positive (p-type) or negative (n-type) charge carriers. Amorphous silicon (a-Si), unlike c-Si, has no long range order in its atomic structure. Additionally, a-Si is a cheaper alternative to c-Si and has its own benefits in solar cells such as an increase absorp-tion of light [2]. The third phase of silicon that is of interest is quantum confined silicon which has crystals on the order of the Bohr radius for silicon (≈ 8.6 nm), begin to exhibit quantum effects in their optical and electronic properties due to the confining potential they reside in [3, 4].

Quantum confined silicon is a fascinating material on its own due to its size and quantum mechanical nature and becomes more interesting when used in conjunction with other mate-rials [5]. The electronic capabilities of quantum confined silicon embedded in an amorphous silicon matrix are very promising due to a combination of the advantages for each of the

different phases [6]. For example, the combined phases of material could lead to similar cost savings of a-Si with the quantum confined silicon providing the efficiency of c-Si in solar cells [7]. The nanocrystalline phase of a material is also referred to as a quantum dot [8].

One drawback of using silicon in semiconductors is the movement towards the general need for higher power circuits and temperature operating conditions [9]. Silicon loses it ability to be used in electronics around 250 ◦_{C due to an increase in the thermal generation}

of charge carriers [10]. To circumvent this, a different base material must be chosen to take over for pure silicon as society moves towards higher power circuits. A candidate for this replacement is silicon carbide (SiC). SiC is a wide band gap (WB) semiconductor that has begun to see broad use in diverse applications. The larger band gap of SiC compared to more commonly used pure silicon has a much higher potential for use in high power devices [10, 11]. Other uses of SiC include on-board vehicle charging and even radiation detection [12, 13]. As more uses of SiC are proposed, many synthesis techniques will be required to best produce the increasingly many varieties of SiC needed.

For this project, the chosen synthesis method was plasma enhanced chemical vapor depo-sition (PECVD) since amorphous silicon carbide (a-Si_1−xCx) was the necessary phase of SiC.

This method allows for high quality film deposition and fine control of the film properties with growth parameters that are well understood [14]. Due to the extensiveness of research into PECVD, novel materials can be produced by slightly adjusting the growth parameters. In particular, the PECVD system used for this work is capable of growing and depositing both amorphous films & nc-Si simultaneously in a single layer of material.

Combining the two phases of silicon leads to interesting behavior of the quantum confined phase. Due to the crystalline nature of the quantum confined silicon phase, it has a lower band gap ( 1.1 eV) than the surrounding amorphous silicon matrix ( 1.5 eV). This leads to the silicon quantum dots (SiQDs) becoming confined in the material in the manner of the particle in a box model from foundational quantum mechanics [15]. Knowing this allows for the parameters involved in this problem to be varied in an attempt to gain a better

understanding of the physical system and how it would behave when those parameters are changed. This leads to a desire to embed the SiQDs into amorphous silicon carbide (a-SiC), a material which has little experimental study as a matrix for Si nanocrystals, to verify the hypothesis that this is an advantageous system to explore [16]. Thus, a critical goal of this work is to uncover how the properties of nc-Si vary when placed into a-SiC matrices as opposed to a-Si.

1.2 Objectives

The main purpose of this research was to explore differences in how a matrix affects the nc-Si that is embedded within. To begin, Chapter 2, gives a background understanding of semiconductor physics and the quantum confinement effect. This includes discussing the band theory of semiconductors (specifically Si & SiC), and how the combined phases of Si and SiC give rise to a quantum system. The quantum confinement effect gives a working theory that describes how this novel quantum system should behave. Chapter 3 details the experimental methods used to explore this phenomena. The base materials grown for this work are characterized in Chapter 4. Chapter 5 explores how the quantum confinement effect influences the proposed system of a-SiC/nc-Si and how this was measured. Finally, Chapter 6 brings the pieces together to discuss the impact of this work and how it could be further explored by others.

CHAPTER 2 BACKGROUND

In this chapter, the basics of semiconductor physics including band structure will be discussed as well as the quantum mechanical nature of nc-Si. The effects of these on quantum confinement is then explored.

2.1 Semiconductor Physics

Semiconductor materials have provided one of the most important technological advances in the world, pushing society into the age of modern electronics [17, 18]. Semiconductors are called as such because they have a conductivity between that of conductors and insulators. This classification can be determined by looking at the electronic band structure of these materials. The electronic band structure defines the states electrons will occupy when a large collection of atoms come together to form a solid. Quantum mechanics forbids any electrons from sharing the same set of quantum numbers. This causes all of the electrons to have slightly different energies, and when their energies (ǫ) are plotted against their crystal momentum (~k), the bands of electron states emerge. The band gap in a material is the distance between the highest energy in the lower group of filled bands (called the valence bands) and the minimum energy in the upper group of empty bands (called the conduction bands). This separation dictates the range of energy levels where no electron states can exist in the ideal solid. From this band structure, many properties of a material can be garnered including how well the material conducts electricity. Conductive materials have an overlapping conduction and valence band, giving the electrons many free states to occupy allowing the electrons to flow through the material into the different available states. Insulators on the other hand have a large band gap and there are no available states for the electrons to occupy. For a semiconductor, the band gap ranges in size from 0.5-4 eV, which lies between that of a conductor and insulator [19]. Semiconductors tend to have the

electron states in the valence band filled leaving the conduction band states empty at 0 K. The size of the band gap in a semiconductor is surmountable by the electrons in the valence band. When energy from an external source is provided to the electrons, they can be excited into the conduction band of the solid which allows for a flow of electrons and thus electrical conductivity.

The exact band structure for every material is unique [20]. One of the biggest differences between band structures in semiconductors is whether the band gap in the material is direct or indirect. This distinction comes from looking at where the maximum and minimum of the valence and conduction bands occur in momentum space, respectively. If the momentum vector is the exact same for both of these points the band gap is said to be a direct gap. The importance of this distinction is evident when looking at how recombination events in direct and indirect gap materials occur. When an excited electron in the conduction band relaxes back to the valence band to recombine and annihilate the hole left behind in the valence band, the released energy can be emitted as a photon. In indirect gap materials, because of the difference in both energy and crystal momentum, the emitted photon is only able to conserve energy and not momentum. To conserve both quantities, a phonon must be created in the indirect material. This phonon helps to conserve crystal momentum but also ends up taking some of the energy from the excitation used to excite the charge carrier, and the emitted photon is a lower overall energy than the initial excitation energy. In a direct gap material, the emitted photon has the same energy as the excitation required to excite the electron because there is no difference in crystal momentum that needs to be accounted for. Since indirect bandgap materials have to emit two particles which is a lower probability event leading to a lower effieciency for these materials.

2.2 Energy Levels of Silicon Quantum Dots

SiQDs are inherently a quantum mechanical system due to their size [21]. This is ad-vantageous because it allows for certain properties of the dots, such as the energy levels, to be well defined. These energy levels can be approximated using the 3D infinite square well,

which as found using the Schr¨odinger equation,  − ¯h 2 2m∇ 2_{+ V (r)}_{Ψ(r) = EΨ(r) (2.1)}

where Ψ is spherically symmetric in the azimuthal and axial angles and V(r) is the infinite square well. The solution for these boundary conditions is given in [8]. As the form of the wavefunctions are not important for this research, only the obtained eigenvalues will be discussed. The eigenvalues are,

E_nle,h = ¯h

2

2me,h

χ2_nl

R2 (2.2)

where me,h is the mass of the hole/electron, χnl is the nth zero of the lth order spherical

Bessel function, and R is the radius of the dot. The energy of the dot is only dependent on the mass of the hole/electron and the size of the dot. This leaves two variables that can be varied in an attempt to increase the energy levels of the dot. However, in order to get energy levels for the finite well scenario, the situation complicates quickly. The starting point is still 2.1, but as the wavefunction is capable of existing outside the well, the eigenvalues are drastically different, and the general form of the eigenfunction is [22],

El(ξ, η) = ξjl−1(ξ)hl(η) + (l + 1) m∗ me − 1  jl(ξ)hl(η) + m∗ me  ηjl(ξ)hl−1(η) (2.3) where ξ = αR, η = βR, & α = √ 2m∗_(V0_−E) ¯ h , β = √ 2meE ¯

h and m∗ is the effective mass for the

electron. jl(ξ) is the spherical Bessel function of lth order, and hl(η) is the spherical Hankel

function of the first kind. The Hankel function is just a linear combination of the first and second kinds of the Bessel functions. The eigenvalues are then the values of E that cause the eigenfunction to be equal to zero. In the finite potential case, there is now another variable that impacts the overall energy of the electron/hole, the strength of the potential. The energy levels of the SiQDs is then seen to be directly related to the strength of the surrounding potential, giving a new pathway to increase the energy levels of the SiQDs.

2.3 Quantum Confinement Effect

The quantum confinement effect is a phenomena that drastically changes the character-istics of quantum dots [23]. This confinement effect is observed in the widening of the band gap of the confined dots [24]. Effective mass theory defines the size of the band gap, Ebg to

be of the following form

Ebg = Ebulk +

C

Rx, (2.4)

where Ebulk is the band gap energy of the bulk material of the quantum dots, C is the

confinement parameter and defines how much confinement the dot is experiencing, R is the radius of the dot and the exponent x is the confinement power [25, 26]. Since Eqs. 2.2 2.3 and 2.4 are related to energy, the confinement parameter must be the parameter altered by the strength of the potential. This leads to the conclusion that the confinement parameter for the dot has a direct dependence on the surrounding matrix confining it. Based on this, the ”C” should change with respect to the bandgap of the confinement material. If we solve for the confinement parameter given a known bulk bandgap and varying bandgap and dot radius we can determine some trends that can be experimentally tested. Here is an example of the confinement parameter that we would expect from a 5 nm SiQD in an confining potential. Setting the bulk bandgap to 1.1 eV, the dot size to 5 nm and having the bandgap of the dot go up to 1.3 eV we can find the confinement parameter needed with a confinement power of x = 2 to accomplish this increase,

Ebg = Ebulk+ C Rx (2.5) 1.3 = 1.1 + C 52 (2.6) 25(1.3 − 1.1) = C (2.7) C= 5 _{[eV ∗ nm}2] (2.8)

This calculation can be repeated across a variety of dot sizes and dot bandgaps to explore trends in the confinement parameter. These trends can be seen in Figure 2.1, where the size of the dot and emission energy are varied between 2-10 nm & 1.1-1.7 eV respectively and setting the confinement power x = 2. We see that in order to achieve a higher dot emission energy, i.e. a larger dot bandgap, we have two options: increase the amount of confinement the dot experiences or decrease the size of the dot. Changing the material the dot is confined in allows us to readily change the bandgap of the confining material. This leads to the hypothesis that changing the confining material around the SiQDs should allow us to increase the confinement of the dots which is accomplished by increasing the depth of the well (the bandgap of the confining material) in which the dot’s electrons are confined.

Figure 2.1: Confinement parameter behavior with varying QD size and desired emission wavelength in a 1.1 eV bandgap confining material, as both the QD size and emission increase the confinement parameter must also increase.

CHAPTER 3

EXPERIMENTAL METHODS

The methods used to synthesize samples and measure their properties are detailed in this Chapter. The synthesis method chosen was plasma enhanced chemical vapor deposition (PECVD). The characterization methods were FTIR and SE to determine various properties of interest about the grown films such as the composition and bandgap of the films.

3.1 PECVD Synthesis

The basic principle of PECVD is the use of a plasma to dissociate a chosen precursor to be deposited onto a substrate. Specifically, silicon depositions occur by stripping the hydrogen atoms off of silane (SiH4) molecules. This provides elemental silicon that deposits

through a chemical reaction with the surface of the substrate of choice and further layers bond with the first to construct a thin film. Using PECVD offers many advantages over other synthesis methods of silicon through precise growth parameter control allowing the properties of the resulting films to be fine tuned. Due to the chemical reaction nature of this method the growths can be replicated easily. A main advantage of PECVD over that of traditional chemical vapor deposition (CVD) is the use of lower substrate temperatures (>600◦_{), which is useful for semiconductor synthesis. Another advantage of this synthesis}

method is the technique’s ability to produce device quality silicon [27–29]. 3.1.1 PECVD System

The PECVD system used for this project is from MVSystems, model CT8-100-P3. In previous work, a secondary deposition component was added to the growth chamber to allow for co-deposition of both a-Si and SiQDs [30]. This is advantageous for these co-deposition growths because it allows for the optimal growth conditions for both materials to be achieved rather than compromising the conditions to attempt to grow both materials in the same

component. The growth chamber for this model is schematically shown in Figure 3.1. The electrodes for the plasma responsible for growing the a-Si are 10 x 10 cm and an RF power supply at a frequency of 13.56 MHz is used to generate the plasma. The SiQDs are grown in a tube below the deposition chamber and are directed to the substrate. The substrate is placed upside down and transferred to the growth chamber using a motorized transfer arm. Once in the chamber, the substrate is allowed to heat up to the desired temperature for the growth. Once the set temperature is achieved, the system undergoes a purge and pump process using argon gas to remove any possible impurities from the growth chamber. Next, the chosen precursor gases are introduced at fixed flow rates into the chamber and using a throttle valve the desired chamber pressure for growth is achieved. Finally, RF power is applied across the electrodes to generate a plasma of the precursor gas to start the growth. Typical growth parameters for a-Si on the system used in this work were a substrate temperature of 200◦_{C, an RF power of 1 W, silane flow rate of 20 standard cubic centimeters}

per minute (sccm), and the chamber pressure is maintained at 0.500 torr. 3.1.2 Silicon Carbide Synthesis

At the onset of this project, the only precursor gases available were pure Silane (SiH4)

for a-Si growth and SiH4 diluted with argon (99.55%:0.45% :: Ar:SiH4) for SiQD growth.

In order to create a-SiC films, a methane (CH4) gas cylinder was installed into the system.

Similar to silane, this allows for the creation of elemental carbon that can be incorporated into the amorphous film depositions by stripping off the hydrogen atoms of the molecule, which then form hydrogen gas (H2). The PECVD system allows for control over the growth

parameters allowing for fine tuning of film composition. In order to see any noticeable effects from carbon being introduced to the film a high enough concentration of carbon is needed otherwise the silicon effects will dominate. Incorporation of carbon into the film is dependent on the parameters used for the film growth, in particular the ratio of the gases used. Growing a-SiC via PECVD, has been performed by many others [31–54]. In order to choose the parameters that were to be used for the best quality films, a survey of this

Figure 3.1: PECVD Reactor Co-Deposition growth chamber, the amorphous film growth occurs from the laterally flowing precursor gas (orange arrow) while SiQD synthesis occurs below the substrate and flows up to be deposited (green arrow).

literature was completed of 24 papers over the years 1989-2017. The results of this survey are shown in Figure 3.2 and Figure 3.3. Figure 3.2 shows the relationship between the temperature of various growths compared to the pressure used, while Figure 3.3 shows the relationship between temperature of growth and the ratio of the gases used. These plots were used in determining what parameters for the silicon carbide films were to be used. A similar PECVD set up to the one used for this project was used in another study and is marked with a yellow point in the plots [48]. As a result of the literature review, it was determined that the required growth parameters for a-SiC were achievable in our system. This was due to the parameters that were common in high quality a-SiC synthesis studies being possible on the system available in this work.

Figure 3.2: Pressure vs Temperature plot, blue and red dots represent papers that used a range of temperatures to indicate the low and high temperatures in their range. The green point indicates the average across all sources and the yellow point indicates work done on similar equipment to this work [48]. The blue point indicates the substrate temperature and chamber pressure used in this work.

For this work, a variety of samples were grown to investigate the difference in film charac-teristics and how they influence the SiQDs. The sole parameter that was varied for the films

Figure 3.3: Gas Ratio vs Temperature plot, the green point indicates the average across all sources. The yellow point indicates the values used on a system similar to the one from this work. The blue x points indicate the values used for this work. (Note: The pure a-Si parameters are not shown since the ratio is 1:0 SiH4:CH4)

was the ratio of the gases in the plasma to influence carbon incorporation in the film. The flow rate ratios used were 20:0, 20:10, 10:10 & 5:10 sccm (SiH4:CH4) for ratios of 1:0, 2:1, 1:1,

and 1:2. These ratios provided a film of pure a-Si and films of a-Si1 – xCx with (varying (but

unknown) x). Other parameters for the growths were 500 mTorr chamber pressure, 200 ◦_C

substrate temperature and the minimum power to maintain a plasma depending on if only silane was present or if a mixture was used in the growth (1 & 1.95 W respectively). The flow of the dilute silane for SiQDs was 100 sccm, and the power used for the SiQD growth plasma was 50 W.

3.2 Optical Characterization Methods

Various optical methods were used to characterize and measure the properties of the various samples that were produced. This includes Fourier Transform Infrared (FTIR) spec-troscopy to verify the successful synthesis of the a-Si and a-Si1 – xCx films, photoluminescence

(PL) to gain some insight on the emission spectra of the various samples, and spectroscopic ellipsometry to provide further insight into the energy levels of the quantum dots embedded into samples.

3.2.1 Fourier Transform Infrared Spectroscopy

In order to verify what the thin films grown were composed of, Fourier transform infrared spectroscopy (FTIR) was chosen as the experimental technique. However, this technique is unable to identify the phase of material present. This method solely measures the absorption of infrared light by a material. The naming of this technique comes from the use of an IR source that generates a broadband of the IR spectrum. This light is sent into an interfer-ometer to produce an interferogram, this interferogram is then sent incident to the sample which ’filters’ some of this signal by absorbing particular frequencies that excite vibrational modes of the molecules in the sample. The amount of absorption given at any particular wavenumber is defined by the Beer-Lambert Law,

A= log(T0

where A is absorption, and T & T0 are the measured transmission of the light after it has

passed through the sample and the reference substrate, respectively. The measurement itself goes as follows, the light is incident on the sample and is transmitted through it, next, the light is sent into the detector which measures the remaining light as an interferogram and takes a Fourier transform to recover the remaining spectrum of the frequencies of IR light used, giving us a transmission spectrum. The measurement is then repeated with the reference substrate, finally the absorbance is calculated. When the sample absorbs light at a particular wavenumber, the transmission for that wavenumber decreases in the spectrometer which leads to a peak in the absorption spectrum. Every material has different frequencies of light that it will absorb based on its molecular structure. For vibrational modes of atoms in the material, these absorbed frequencies are in the infrared range. For silicon and carbon based materials IR spectroscopy allows vibrational modes associated with Si – H, C – H, and Si – C bonds to be detected. It is the ideal technique to identify materials based on absorption spectra. In addition to the IR region being the region where a-Si and a-SiC have their vibrational absorption peaks, FTIR was chosen due to its several advantages over other IR identification methods such as grating IR spectroscopy [55]. The most significant of these is the increase in the signal to noise ratio for these sorts of measurements due to a special property of multiplexed signals called Fellgett’s advantage [56]. Since a-Si and a-SiC films share many characteristics, their absorption spectra are very similar, the major difference between them being the carbon peaks associated with incorporated carbon are highly prevalent in samples with carbon and absent in samples that were grown only using silane. A collection of expected vibrational peaks from samples containing a-Si and/or a-SiC is shown in Table 3.1.

This spectroscopy was performed on a Nicolet 8700 FT-IR spectrometer. In order to remove sources of error such as water in the air absorbing infrared light, the sample chamber is purged with dry air for at least 12 hours. To acquire absorption spectra, a reference sample must also be measured. For these samples, since the film is grown on a silicon wafer, a piece

Table 3.1: Expected FTIR peaks for a-Si & a-SiC absorption spectra [57, 58]. The different modes indicated in the third column serve to give an idea on the nature of the vibrational mode in question.

Bond Assignment Wavenumber (cm−1_{) Mode}

SiHn 650 Rocking/Wagging Si – CH3 750 Rocking/Wagging Si – C 780 Stretching SiH2 890 Wagging CHn 1000 Bending Si – O 1100 Rocking/Wagging Si – CH3 1250 Stretching C – H 1300 Stretching C – H 1400 Stretching C –– O 1700 Stretching SiHn 2000-2100 Stretching C ≡ C 2200 Stretching C – H 2840 Stretching (Doublet) C – H 2850 Stretching (Doublet) C – H 2900 Stretching C – H 2950 Stretching

of the same wafer the sample was grown on was used as the reference. This allowed for easy identification of the contents of the film.

3.2.2 Photoluminescence Spectroscopy

PL spectroscopy examines the process through which a material absorbs and emits elec-tromagnetic radiation. This process is quantum mechanical in nature. In order to produce PL, an excitation source with energy higher than the band gap of the sample must be used. The excitation source (an Ar-ion laser for this work) produces photons incident with the sam-ple of interest, exciting electrons from the valence band into the conduction band creating an electron-hole pair. The conduction band electron then relaxes to the ground state of the conduction band as the hole relaxes to the top of the valence band before recombining and emitting a photon in the process. PL has the advantage of being non-destructive allowing for additional measurements to be done on a sample. Identification of materials is possible due to PL spectra being unique for a particular material. This is possible due to the different

spectra that are produced by a material that allow for many things to be determined about the sample studied including the general crystal structure and the band gap of the material. Crystal bonding structure can be inferred from the shape of the spectra acquired, amor-phous materials tend to have broad emission bands where crystalline materials have sharp features. The excitation process of PL is diagrammed below in Figure 3.4. The main emis-sion regime for silicon comes from a phonon assisted band to band relaxation with the energy levels of the emission determined by the Schr¨odinger equation for the appropriate boundary conditions [59]. At low temperatures, the emission can also be related to donor or acceptor related recombinations. We want this emission to be the most prevalent in our spectra. To make the majority source of emission the band to band relaxation, other sources must be removed. This is accomplished by cooling the sample down to 10K, which removes the con-tribution of secondary emissions from other emission regimes and in an amorphous sample, the center of the emission peak shifts and sharpens as the other sources of emission become diminished [60]. This phenomena can be observed in Figure 3.5, which shows the effect on a-SiC PL emission as the sample was cooled, the colors represent different temperatures, black is room temperature, red is 270 K, purple is 200 K, green is 140 K, and blue is 100 K. The large spikes in the red and black curves at 1028 nm are due to the 514 nm excitation laser line coming through the spectrometer and reaching the detector. The fluctuations in the 650-750 nm range are most likely small amounts of noise from LEDs throughout the lab as they become less observable as the sample cools and produces a more intense emission relative to the constant emission intensity from the LEDs.

3.2.3 Photoluminescence System

The system used for PL measurements contains both a visible light spectrometer as well as an FTIR spectrometer to measure emission in the infrared. The system was setup and aligned to allow for collection of both visible and IR emissions from the exact same spot on the sample, accomplished through the use of a removable mirror that allows for either spectrometer to be selected by the user to collect the emission from the sample. An

Exciting

Photon

Excitation

Phonon Assisted Relaxation

De-excitation

Conduction Band

Valence Band

Emitted

Photon

Figure 3.4: A typical excitation scheme for PL in a direct semiconductor system with the horizontal lines indicating the individual energy states available in each band. An incoming photon excites an electron from the valence band into the conduction band leaving a hole in the valence band behind, then before de-exciting back into the valence band, the electron relaxes to the bottom of the conduction band while the hole relaxes to the top of the valence band, both assisted by a phonon to conserve momentum. Then the recombination occurs and a photon is emitted. This diagram demonstrates

600 700 800 900 1000 1100 0.0 0.2 0.4 0.6 0.8 1.0 Wavelength (nm) No rmal ized In ten si ty

Temperature Dependence of PL Spectra

Figure 3.5: Temperature dependence on photoluminescence of a-SiC, with the colors indi-cating the various temperatures, black is room temperature, red is 270 K, purple is 200 K, green is 150 K, and blue is 100 K. As the temperature decreases the emission shifts and condenses as the important emission intensity grows rapidly.

overhead view of the system is shown schematically in Figure 3.6. The removable mirror serves the purpose of allowing the emitted light from the sample to be directed to the desired spectrometer. If the mirror is in place, the light will be directed towards the visible spectrometer, and if the mirror is removed the light is sent into the near IR spectrometer. This allows for the emitted light to come from the exact same spot from the sample.

Mirror (Removable)

Mirror Laser Line Filter

Cryostat Sample Lens Ar:Ion Laser Visible Spectrometer Excitation Laser @ 514.5 nm Emitted Light from sample

Near-IR Spectrometer

Figure 3.6: Overhead view of the PL system, the green line indicates the path of the ex-citation laser, while the red line indicates the paths taken by the emitted light from the sample.

Another issue that arose with the PL system during this work was temperature control. As discussed previously, the sample must be cooled down to minimize other non radiation recombination channels. To accomplish this, the sample is attached using thermal paste to

a copper cold finger on the cryostat. Then a heat shield is placed over the cold finger, and a vacuum housing is placed over this set up. The sample is then put under vacuum using a roughing and a turbo pump to get the sample to ≈ 10−6 _{torr. This minimizes thermal}

conduction from the outside ambient into the sample and keeps the cryostat windows from fogging. The cryostat used is a closed cycle He refrigerator. Once the chamber is at this pressure, a helium compressor is turned on, and the sample is allowed to cool. Prior to this work, the readout from the thermocouple used to measure the temperature was trusted to be accurate. During this work, due to both the readout of the thermocouple beginning to malfunction and a test sample of multiple layers of GaAs giving a spectrum that disagreed with the temperature reading, the sample temperature had to be determined manually. This was accomplished by measuring the voltage produced across the thermocouple and finding the table of values that matched. It was determined that the thermocouple type in the PL system is a chromel vs. gold with 0.07 at.% iron. This was determined by measuring the thermocouple voltage using a known temperature source to cool one end of the wire. Liquid nitrogen was used in this case to provide a stable and known temperature. After the thermocouple type was determined, the temperature of the sample was accurately measured. The cooling procedure of the sample was able to achieve a temperature of 10 K.

3.2.4 Spectroscopic Ellipsometry

Spectroscopic ellipsometry is a non-contact, non-destructive optical technique used to characterize thin films [61]. The basic principle of ellipsometry is creating a beam of linearly polarized light incident on a thin film. When the linearly polarized light strikes the film the components of the light that are perpendicular and parallel to the surface undergo different phase shifts. These phase shifts for the perpendicular and parallel light are described by the Fresnel equations,

Rs=     n1cos(θi) − n2cos(θt) n1cos(θi) + n2cos(θt)     2 (3.2) (3.3) Rp =     n1cos(θt) − n2cos(θi) n1cos(θt) + n2cos(θi)     2 (3.4)

where Rs and Rp are the reflection coefficients of the perpendicular and parallel light, n1

and n2 are the indices of refraction for the first and second mediums with the first medium

being the air above the sample for ellipsometry and the second being the sample, θi and θt

are the angles of the incident and transmitted light through the interface. For incident angles between 0 and π

2, Rs 6= Rp, turning the linearly polarized light

into elliptically polarized light. The light then continues through to an analyzer, which is another polarizer, then into a detector which measures the light. A diagram of a typical ellipsometry set up is shown in Figure 3.7. The reflected light has two components: a component parallel to the surface of reflection and one perpendicular. Comparing these two component’s reflection coefficients provides a large amount of information about the sample the light was reflected off of to be extracted, such as the thickness, indices of refraction, and dielectric constants for the sample.

The defining equation for ellipsometry is,

ρ= Rp Rs

= tan(Ψ)ei∆ (3.5)

where ρ is the ratio of the reflection coefficients for the parallel and perpendicular components of the light, Rp and Rs are the reflection coefficients for the parallel and perpendicular

components of the lights respectively. Ψ is the angle that’s tangent is equal to the magnitude of ρ, and δ is the phase difference between the components. To accurately determine the characteristics of a thin film being measured, different angles of incidence and/or wavelengths of incident light must be used. This is because of the cyclic nature of both Ψ & ∆ for the same value of ρ (multiple values of Ψ and ∆ will return the same ρ).

Sample

Source

Detector

Polarizer _Analyzer

θ

Figure 3.7: Typical set up of an ellipsometry measurement

The fitting routine for this work is performed in multiple steps. Immediately following data acquisition, the ellipsometry data is fit using a series of polynomials, known as a B-Spline. This allows for a general idea of the optical constants of the measured film to be determined. Once the B-splines have been fit to the data with a Mean Square Error (MSE) <10, the layer of interest’s dielectric functions can be parameterized. The parameterization is done by fitting oscillators to the B-spline to then apply physical meaning to the data. The particular oscillator used in this study is the Tauc-Lorentz oscillator, which is a model for an oscillator that describes an electromagnetic wave interacting with a dielectric material [62]. The oscillator attempts to model the dielectric function for the material, which allows the bandgap of the film to be extracted. The imaginary portion of the oscillator has 4 parameters that can be fit: the amplitude of the oscillation A, the center energy of the oscillation E0,

the broadening of oscillator Br, and the actual bandgap of the oscillation, Eg. The exact

form of the dielectric function produced by the oscillator is

where ǫ1 and ǫ2 are defined as ǫ1(E) = 2 πP ∞ Z Eg ξǫ2(ξ) ξ2_{− E}2 (3.7) ǫ2(E) = ( 0, E _{≤ E}g AE0Br_(E−Eg)2 (E2 −E02) 2 +Br2_E2 1 E, E > Eg (3.8)

where P represents the principal part of the given integral [63]. To get these dielectric functions from ellipsometry data requires a fit of ǫ2, then taking a Kramers Kronig transform

gives ǫ1.

Samples that contained SiQDs embedded into the film are much harder to model than purely amorphous films. This is due to the layer of material being examined containing multiple materials. Since the wavelengths of light used were still larger than the dots in the film, the layer can be treated as it’s own unique material, and multiple oscillators can be used to account for the different materials. This approach was chosen over a second method for accomplishing this combined layer model, known as an effective medium approach (EMA) for multiple reasons. The main reason is that the EMA requires knowledge of both materials to be mixed, and while data for crystalline silicon existed there was no data for SiQDs. The expected contribution of the quantum confinement effect however made it more reasonable to use multiple oscillators to capture the effect of the surrounding matrix on the SiQDs.

Another benefit of using these oscillators to model the dielectric function of our material, is that it allows for a comparison between the absorption coefficient (α) and energy for the layer of interest to be constructed. The comparison is generated using the dielectric function which is used to define the absorption coefficient as a function of energy. The oscillators used give us the shape the absorption curve and from this conclusions on the properties of the layer and its constituents can be made. The onset of absorption is indicative of the bandgap of the material, and the bandgap can be acquired by extrapolating the linear onset of the absorption [64]. An example of an absorption coefficient plot of a material is shown

below from a measured crystalline silicon wafer in Figure 3.8. From this the bandgap can be extracted by looking at the linear region of the onset of absorption. For crystalline silicon this gives a band gap of approximately 1.1 eV, which is exactly what we would expect for crystalline silicon. Absorption Coefficient 1.0 1.5 2.0 2.5 3.0 3.5 4.0 0 200 400 600 800 1000 1200 1400 Energy (eV) (α ℏω ) 1 /2 (cm -1eV ) 1 /2

Determination of Dot Bandgap in a-SiC (2:1 SiH4:CH4)

Figure 3.8: Absorption coefficient plot for a crystalline silicon wafer, the Tauc line provides the ability to extract the bandgap, approximately 1.1 eV in this case.

CHAPTER 4

MATERIAL CHARACTERIZATIONS

For this work, various samples of a-Si and a-SiC were produced via PECVD. The variable for these samples was the gas ratio of the precursors. This was done in an attempt to produce amorphous films with silicon and a varying amount of carbon incorporated into the a-Si. Increasing carbon should increase the bandgap of the film, which is desirable to be able to test the quantum confinement effect. The methods to test these objectives were FTIR spectroscopy to qualitatively measure carbon incorporation and spectroscopic ellipsometry to measure the bandgaps of the new films. There are methods to quantitatively measure carbon content in these films but determining the exact carbon content was deemed unnecessary for the goals of this project.

4.1 Qualitative Carbon Incorporation in a-Si and a-Si_{1 – x}Cx films

To determine a successful growth of a-Si and a-Si1 – xCx films, FTIR spectroscopy was

performed and the measured peaks were compared to the expected peaks from literature (Table 3.1). Films of a-Si should contain only the peaks related to silicon and any carbon peaks are attributed to surface carbon due to exposure to the atmosphere. Films of a-SiC will have peaks relating to both silicon and carbon bonds, it is the intensity of the peaks relative to each other that show that carbon is incorporated into the film rather than just on the surface of the sample. The absorption for the films was measured over wavenumbers from 600(cm−1_)-4000(cm−1_{). To improve the accuracy of the measurements, the sample chamber}

is purged with dry air to remove water as water has a large absorption in this spectral range. Once this purge process is complete the absorption spectrum is obtained for both the sample of interest and a reference. The reference sample was a piece of silicon wafer cleaved from the same wafer used as the substrate for growing the amorphous films. This was done to remove any background absorption due to the substrate and ensure that the peaks observed

were from the amorphous films. The four films produced used ratios of silane to methane of 1:0, 2:1, 1:1 and 1:2.

The IR absorption spectra for these samples are shown in Figure 4.1. In the samples where methane was present during the growth, additional peaks related to carbon being in the film are seen. Most importantly, the peaks at 780 cm−1 _{and 3000 cm}−1 _{which correspond}

to SiC and CH respectively. The presence of these peaks in samples where methane was present in the growth, and the lack thereof in the pure a-Si sample are strong indications that the measured carbon peaks are from carbon incorporation and not from atmospheric contamination. Another characteristic of the films that can be garnered from these spectra, is the relative carbon incorporation between the films. To do this, the main peaks of interest are the SiHn peak around 600 cm−1 & the SiC peak just below 800 cm−1. Comparing these

two heights across the films allows for a qualitative understanding of the difference in carbon incorporation. The exact ratios are detailed in Table 4.1. From these ratios, it can be seen that an increase in the amount of methane used for the films, leads to an increase in the ratio of the peaks. This leads to the conclusion that the ratio of gases used has a direct impact on the amount of carbon incorporated into the film. The increase in the carbon proportion in the film is nearly linear with respect to the amount of methane used during the PECVD synthesis which agrees with previous study [65]. This result is promising to begin precisely growing a-Si1 – xCx films with any desired ratio. Other methods are necessary

to quantitatively measure the precise carbon content for these films such as absorbance cross sections, secondary ion mass spectroscopy, and x-ray photoelectron spectroscopy, but these methods were outside the scope and intentions of this work.

Table 4.1: Ratio of Peak Heights in the Synthesized Films Gas Ratio Used (SiH4:CH4) Peak Height Ratio (SiC : SiHn)

1:0 0

2:1 0.294

1:1 0.365

1:0 SiH4:CH4

Absorption(Arb. Units)

2:1 SiH4:CH4

Absorption(Arb. Units)

1:1 SiH4:CH4

Absorption(Arb. Units)

1:2 SiH4:CH4

Absorption(Arb. Units)

Figure 4.1: FTIR spectra for the produced films with the gas ratio used indicated. An increase in the peaks related to carbon are seen as the amount of methane used increased.

4.2 Changes in the Amorphous Matrix Chemical Bonding as a Result of Methane Incorporation

In addition to the qualitative determination of carbon incorporation, there are some other interesting points from this data set. If the peaks in the range 1950-2100 cm−1 _{are examined}

further, it becomes apparent that they are different across the samples grown. This can be seen in Figure 4.2. There is a distinct shift of a single peak at 2000 cm−1 _{to a single peak at}

2080 cm−1_{. This was initially thought to represent a distinct shift from SiH bonds to SiH} 2

bonds [66]. However, this shift only occurred in the samples where methane was used in the film growth. This shift is rather interesting and suggests there are distinct differences in a growth process that contains methane, and upon further review of the literature, this shift is likely due to carbon back-bonding in the SiH stretch mode [67]. Considering this shift only occurs when methane is present during growth it makes sense the shift is due to carbon back bonds rather than an increase in the number of SiH2 bonds. This is also supported from

the above discussion where the shift in energy of the vibrational mode occurs because the lighter carbon atom replaces a heavier silicon atom increases the frequency and therefore the energy of the stretching mode. This also signals that the addition of methane has negligible impact on the overall quality of the film as the main hydrogen silicon bond present is still SiH instead of SiH2 which is desirable in high quality a-Si [29]. Monitoring this shift could

1:0 SiH4:CH4

Absorption(Arb. Units)

Raw Data SiH Model Sum

2:1 SiH4:CH4

Absorption(Arb. Units)

Raw Data

SiH w/ Carbon Back Bond SiH Model Sum

1:1 SiH4:CH4

SiH w/ Carbon Back Bond SiH Model Sum

1:2 SiH4:CH4

Absorption(Arb. Units)

Raw Data

SiH w/ Carbon Back Bond SiH

Model Sum

Figure 4.2: Zoom in on FTIR spectra for the amorphous films, with an emphasis on the shift of the SiHn peak from SiH to

4.3 Determination of Bandgap for a-Si and a-Si_{1 – x}Cx films

Another important characteristic that needed to be determined was the bandgap of the films as this would allow for a more precise determination of the existence of the quantum confinement effect. The bandgap was measured using spectroscopic ellipsometry as discussed in Chapter 3. The spectral range for these measurements was 1.24-5.88 eV, chosen due to detector limitations on the ellipsometer system, and this range was used at 3 different incident angles of 50, 60 and 70 degrees. This data was fit with Tauc-Lorentz oscillators. To acquire the bandgap for an amorphous film, the full process requires acquiring the dielectric functions from oscillators, then the absorption coefficient plot can be generated. The process is shown for an a-Si film, the other dielectric functions are shown in the appendix.

First, the raw data is acquired and fit with oscillators, shown in Figure 4.3. The fit itself is not perfect at high energies for Ψ, but the overall MSE for the fit was adequate (10 > MSE) to proceed with the analysis. Not much can be inferred from this data by itself, it only helps us to provide something to fit our oscillators to. The oscillators are used as they provide a way to apply physical meaning to our data. Once the fitting is completed, the dielectric functions based off of the oscillators can be plotted (Figure 4.4). Finally, the absorption coefficient can be plotted as a function of energy based off of this dielectric function as shown in Figure 4.5. The shape of this absorption makes it trivial to determine the bandgap, which for this particular a-Si film was 1.528 eV, perfectly in line with what we expect from a-Si.

As we repeat this process for the other samples and the four different absorption curves are acquired, a change in bandgap becomes evident. The four absorption curves are overlaid in Figure 4.6, to demonstrate the difference in the bandgap energy for the films as the gas ratio changes. The extracted bandgap for each of these films is given in Table 4.2. The bandgap of the a-SiC film increases as the gas ratio contains more methane precursor. This is as expected and is in agreement with the FTIR spectroscopy result suggesting that increased methane ratios leads to more carbon being incorporated into the films.

2 3 4 5 Energy (eV) 10 20 30 40 Log(α) (cm-1)

Ψ

Model

Δ

Model

2

3

4

5

-10

0

10

20

30

Energy (eV)

P

ermi

ti

vi

ty

(F

/m

)

a

_{-Si Dielectric Functions}

Figure 4.4: Acquired dielectric functions for a-Si, ǫ1 (red) and ǫ2 (green).

1 2 3 4 5 6 Energy (eV) -2 0 2 4 6 Log(α) (cm-1)

Figure 4.5: Generated absorption curve for a-Si thin film, with an extracted bandgap of 1.53 eV.

1:2 SiH4:CH4 1:1 SiH4:CH4 2:1 SiH4:CH4 1:0 SiH4:CH4 1.5 2.0 2.5 3.0 3.5 4.0 Energy (eV) -4 -2 0 2 4 6 Log(α) (cm-1)

Figure 4.6: Collection of the absorption curves for the different gas ratio films allowing for the bandgap of each film to be extracted.

Table 4.2: Extracted bandgap energy for the different gas ratio films. Gas Ratio (SiH4:CH4) Estimated Bandgap (eV)

1:0 1.528

2:1 1.687

1:1 1.907

1:2 2.047

Combined, these two results provide the baseline for optimizing the PECVD synthesis of a-Si1 – xCx films on the system used in this work. These results agree to those of others using

CHAPTER 5

ATTEMPTING TO MEASURE THE QUANTUM CONFINEMENT EFFECT

As the surrounding matrix has an increase in its bandgap, the SiQDs embedded in the matrix should also have an increase in bandgap, due to the quantum confinement effect as detailed in section 2.3. To measure this effect, the methods of PL and SE, detailed in sections 3.2.2 & 3.2.4, were used. The PL measurements were done in an effort to observe the light emission from the SiQDs, which if blue-shifted when the dots are in larger bandgap matrices would indicate quantum confinement. The observed PL showed no discernible shift in the emission from the SiQDs as the surrounding matrix was changed and a new method for observing the confinement was needed. The new method was SE, which provided another way of measuring the bandgap of the SiQDs through absorption curves. Quantum confinement was observed in this fashion but again there was little perceptible shift in the amount of confinement. To gain a better understanding of the samples, the system of SiQDs embedded in a-Si was simulated using an EMA, which gives rise to an estimate of the amount of SiQDs in the sample.

5.1 Photoluminescence measurements of quantum confinement in SiQDs Measurement of PL was attempted as a technique to demonstrate quantum confinement of SiQDs in the various matrices fabricated. The spot being excited on the sample is impor-tant as the density of the dots in the sample at any particular spot has a large impact on the acquired spectra [69]. For this work, multiple scans moving across a-SiC containing SiQDs sample (scan locations shown in Figure 5.1) with gas ratio of 2:1 (SiH4:CH4) were acquired

to look for any variance in the acquired PL spectra as location varies as shown in Figure 5.2, shows a large change in the spectrum captured depending on the location of excitation in the sample. As the measurement location moves towards the center of the dot deposition, so called the dot ’haze’, the overall counts drop and the peak associated with the amorphous

film decreases as the amount of crystalline silicon increases in these locations. The reason for this change in spectrum is that the dot density decreases away from the center of the deposition due to the use of a nozzle to direct the dots onto the substrate. This causes the largest number of dots to be deposited directly above the nozzle and with a decreasing concentration farther from the center of deposition. The number of counts in the spectrum are also reduced towards the center of the dot deposition due to an increase in the number of non-passivated SiQDs, which are not as bright as their passivated counterparts. This shift makes it difficult to have reproducible results even when remeasuring a sample.

Next, when changing the surrounding matrix for the samples, PL phenomena was observ-able, although there was very little distinguishable difference between spectra acquired for the different matrices. Spectra for these samples were expected to be similar in shape, with variation in peak location for the different films as the confinement of the SiQDs changed. The acquired spectra, shown in Figure 5.3, do not show any significant differences in their PL spectra with the peaks ranging between ≈ 910−920 nm. The total count differs for these samples, but this is not indicative of any difference in confinement of the SiQDs. The most likely reason for this is the slight change in the band structure of the a-Si when carbon is added to the system. The band tail widens in the a-Si1 – xCx leading to the similar emission

spectra due to the carbon increasing the number of defect states in the film [70]. The peak appearing around 1028 nm in these spectra comes from the laser used for excitation coming through in second order into the spectrometer.

The results from PL measurements seem to trend in the right direction. In this work, due to equipment malfunctions the only samples grown with SiQDs were amorphous films with gas ratios of 1:0 and 2:1 (SiH4:CH4) and they showed no difference in the amount

of confinement experienced by the dots in the different matrices. Based on the results in Table 4.2 the matrices only differ in bandgap by 0.159 eV, which could possibly not be enough of a difference to see a change in the amount of confinement experienced by the SiQDs. The largest bandgap film that was produced in this work was the 1:2 gas ratio film

Film with no noticeable dot haze Beginning of dot 'haze'

Center of dot 'haze'

Figure 5.1: Approximate PL Measurement locations on sample for scan across sample.

Far Below Dot Haze Edge of Dot Haze

Middle of Dot Haze

700 800 900 1000 1100 0 5000 10 000 15 000 20 000 Emission Wavelength (nm) Count s

Figure 5.2: PL spectra acquired while scanning across the a-SiC sample grown with 2:1 gas flow ratio (SiH4:CH4). Locations seen in Figure 5.1.

a-SiC w/QDs (2:1 SiH4:CH4) a-Si w/ SiQDs a-SiC (2:1 SiH4:CH4) 700 800 900 1000 1100 0 5000 10 000 15 000 20 000 Emission Wavelength (nm) Count s

Figure 5.3: Captured PL from various samples with little difference besides amplitude, suggesting a different method for measuring the quantum confinement of the SiQDs is needed.

with a bandgap of 2.047, which has a difference of 0.51 eV to the a-Si matrix. This might be a large enough difference and should be tested in further studies of this phenomena. As a result of the PL measurements, a second method to measure quantum confinement was needed to observe any difference in the amount of confinement and ellipsometry was chosen. 5.2 Spectroscopic ellipsometry as a measurement method for quantum

confine-ment

Quantum confinement, as it was defined in chapter 2, is really due to an increase in the bandgap of the confined material. Thus, spectroscopic ellipsometry can be used to measure the bandgap of confined SiQDs and has been done before [71–75]. The idea for this work is that in the different surrounding matrices the confinement can be determined by comparing the change in bandgap for SiQDs and comparing this to the bandgap of the matrix the dot is embedded in. The process for determining the bandgap for the films containing SiQDs was the same as before in section 4.3 with the stages of raw data → dielectric function → absorption coefficient → bandgap. The layer of interest is now a more complex material (an amorphous film with embedded SiQDs), more Tauc-Lorentz oscillators were used to capture both material’s dielectric functions. The samples containing SiQDs were measured as detailed in Chapter 4. Similar to the work in section 5.1, the sample was measured at multiple points on the sample to look at the effect of concentration of dots on the measurement. The locations measured are shown in Figure 5.4, and the resulting absorption coefficients are detailed in Figure 5.5. From this series of measurements, one can see that as we approach the center of the dot haze, there are two absorption onsets for the layer, which is indicative that the layer contains two different materials.

To compare our results (with quantum confinement present), a simulation of the same system was performed using an EMA. An EMA allows for the approximation of a mixture of materials, in this case a-Si and a secondary material of c-Si (to emulate the SiQDs). De-pending on the method for this approximation different styles of inclusions of the secondary material into the main material. The Bruggeman method was chosen for incorporating the

c-Si into the system, which gives the second material a spherical inclusion into the layer, a shape similar to the shape of the SiQDs [76, 77]. This allowed for a simulation of the material that was being tested which allowed for a better understanding of the results. The percentage of the c-Si was varied from 0 - 100%, ranging from a pure a-Si system to a pure c-Si system. The absorption coefficients for the simulated samples are shown in Figure 5.6. From this we can see that in samples containing between 20-80% c-Si begin to show the two absorption onsets as seen in Figure 5.5, one around 1.1 eV, the band gap of c-Si, and one around 1.53, the bandgap of a-Si, in addition to the higher energy absorption features seen in c-Si. These two peaks are not seen in the samples with SiQDs most likely due to the lack of long range order in the SiQDs so that while they may be c-Si in nature, they are unable to fully reconstruct c-Si behavior. In addition to changing the shape of the curve, from these simulations it becomes possible to some degree to estimate the crystallinity of a sample by comparing the shape of the obtained absorption curve to these simulated ab-sorption coefficients. Once the simulation of possible data to acquire was completed, the samples containing SiQDs were measured as detailed in chapter 4. Similar to the work in section 5.1, the sample was measured at multiple points on the sample to look at the ef-fect of concentration of dots on the measurement. The locations measured are shown in Figure 5.4, and the resulting absorption coefficients are detailed in Figure 5.5. From this series of measurements, one can see that as we approach the center of the dot haze, there is both crystalline silicon and a surrounding amorphous matrix present in the sample, as seen with two absorption onset similarly to the simulation of an a-SiC matrix containing SiQDs. Comparing the shapes of the absorption coefficient plots to those found from our EMA simulation, the sample becomes more crystalline as we approach the center of the dot ’haze’. The estimate for the crystallinity of the sample in the measured sample is somewhere between 0 − 20% for the edge of the dot haze, and between 20 − 40% at the center of the dot haze.