UNIVERSITY of CALIFORNIA, IRVINE Synthesis of Gallium Zinc

Home , Zinc nitrate

UNIVERSITY OF CALIFORNIA,

IRVINE

Synthesis of Gallium Zinc Oxynitride Solid Solution by

Flame Spray Pyrolysis

THESIS

submitted in partial satisfaction of the requirements

for the degree of

MASTER OF SCIENCE

Chemical and Biomolecular Engineering

Ewa Richard Chukwu

Thesis Committee: Assistant Professor Erdem Sasmaz, Chair Assistant Professor Iryna Zenyuk Associate Professor Ali Mohraz

2020

DEDICATION

The Almighty God who only him knows the end of a journey right from the beginning.

Do not fear, for I am with you;

Do not be dismayed for I am your God;

I will strength you and help you;

I will uphold you with my right hand of righteousness.

Isaiah 49:10

TABLE OF CONTENTS

LIST OF FIGURES……………………………………………………………. .iv

LIST OF TABLES……….………………………………………………………vi

ACKNOWLEDGEMENT………………………………….…………………...vii

ABSTRACT……………………………..……………………………………....viii

CHAPTER 1: Introduction………………………………………………..…...... 1

CHAPTER 2: Literature Review…………………………..…………...... …...... 4

CHAPTER 3: Experimentation...... 27

CHAPTER 4: Results and Discussions……………...…………………...... 34

CHAPTER 5: Conclusion and Recommendation….….…..……………...... 50

CHAPTER 6: Future Work …………………..………………….…...... 53

REFERENCES………..…...……………………………………….……...... 69

APPENDIX A...... ………………………………………………..…...... 83

APPENDIX B...... ……………………………………………..…...... 86

LIST OF FIGURES

Fig.1: Flame synthesis versus wet chemistry……………………………………....5

Fig.2: Different particle formation routes and ...... 8

mechanisms for flame aerosol synthesis

Fig.3: Different H2 production routes……………………………………….……12

Fig 4: Diagram showing the Z-system for photocatalytic water splitting…….…..13

Fig.5: (a) customized FSP set-up (b) set-up with the premixed flamelets…….….28

Fig.6: Schematic diagram of the FSP system………...………………………...... 29

Fig.7: XRD pattern of GZN2……....…………………………………………....37

Fig.8: Diffraction pattern of as-prepared zinc oxide………………………...... 38

Fig.9. Fig. 9. XRD patterns of samples ...... 40

made from urea and zinc nitrate solution only

Fig.10. XRD patterns of gallium zinc oxynitride solid solutions made with….....43

methanol + butanol as solvents compared with pristine ZnO

(GZN 3 (top) and GZN4 (middle) to Z3 (bottom))

Fig.11. XRD patterns of gallium zinc Oxynitride solid solutions made…...... 46

with DI H2O + butanol compared with ZnO

(GZN5 (middle), GZN6 (top) and Z3 (bottom))

iii

Fig.12. Tauc’s plot showing direct bandgaps GZN 3 (3.16 eV) ………………48

and (b) GZN4 (3.07 eV).

Fig.13. Tauc’s plot showing the direct bandgaps GZN 5 (3.12 eV) …………...49

and (b) GZN6 (3.08 eV).

Fig.14. Variation of flame height with (a) precursor flowrate (left)……...... ….63

(b) equivalence ratio (right) at constant dispersion gas rate

and pressure drop.

Fig.15. Flame height versus equivalence ratio and precursor feed rate ……...…63

Fig.16. The influence of precursor rate on the particle size of FSP-made…...….64

ceria nanoparticles.

Fig.17. Variation of flame height with oxygen flowrate at ...... 65

constant pressure drop and precursor rate.

Fig. 18. Visualization of pressure drop effect on the flame at....…………....…..66

constant precursor and oxygen rates. (a) & (b) are at

the same precursor rates and (c) & (d) are at the same oxygen rates

Fig. B1: Heat formation as a function of the concentration of ZrO2 in ceria...... 79

Fig. B2: Modelled heat of mixing as a function of the concentration of ……...... 79

ZrO2 in ceria compared with experimental work.

Fig. C1: Variation of flame height with precursor...... 80

flowrate at constant oxygen flowrate.

Fig.C2: Variation of flame height with oxygen flowrate at ...…………...... 80

constant precursor flowrate and pressure drop.

LIST OF TABLES

Table1: Sample compositions and solvents...... 30

Table 2: Samples and their synthesis parameters...... 33

Table 3: Lattice parameters of the synthesized ZnO...... 38

Table 4: Estimated crystallite sizes of ZnO samples ...... 39

Table 5: Lattice parameters of ZN samples...... 40

Table 6: Estimated crystallite sizes of ZN samples ...... 40

Table 7: Lattice parameters of GZN samples...... 43

Table 8: Estimated crystallite sizes of GZN samples...... 43

Table 9: Bandgaps of GZN samples...... 49

Table A1: Solution compositions and their synthesis conditions...... 77

GZN, Z1 -Z3 and ZN1 to ZN5

ACKNOWLEDGEMENT

I remain indebted to God, the giver and sustainer of life, for his continual grace and blessings.

My appreciation goes to my parents, siblings and well-wishers for their encouragement, support, motivations and prayers.

Special thanks to my Advisor and Chair of the thesis committee, Dr. Erdem Sasmaz, whose guidance and support saw to the successful completion of this work. Also, my sincere gratitude to my thesis committee members, Profs. Ali Mohraz and Iryna Zenyuk for their commitment and efforts.

I am also grateful to the past and present Lab mates whose critiques, support and cooperation contributed in shaping my research thought process.

My eternal gratitude goes to the United States’ Government for their Opportunity Fund

Award through the LagosEducationUSA, Nigeria-that facilitated my graduate school dream. Worthy of mention are my EducationUSA advisers, Mrs. Uwadileke and Mrs.

Adejumobi for their guidance and continual moral support.

I would like to thank Dr. Ben Meekins of the University of South Carolina who carried out the UV-Vis spectroscopy and for his insights and suggestions. Additionally, I would like to acknowledge the SAGE, U of SC whose XRD machine was used for the characterization.

Finally, I seize this opportunity to thank the CBE program of UCI for the enabling environment throughout my stay.

God bless you all.

ABSTRACT

Synthesis of Gallium Zinc Oxynitride Solid Solution by

Flame Spray Pyrolysis

Ewa Richard Chukwu

Master of Science

Chemical and Biomolecular Engineering

University of California, Irvine, 2020

Assistant Professor Erdem Sasmaz, Chair

Photocatalytic water splitting is seen as a viable alternative to steam reforming for hydrogen fuel generation. One of the major keys to exploiting photocatalytic technologies as a commercial energy source lies in the development of catalytic materials that can drive photocatalytic reactions using the largest portion of the solar energy spectrum – the visible light region. Gallium zinc oxynitride solid solution has been recommended as a promising candidate for photocatalytic water splitting, however, the current synthesis methods are largely multi- steps and batch wet chemistry techniques that are not easily amenable to scale up. Flame spray pyrolysis (FSP) is a continuous, single step, fast, relatively cheaper and easily scalable nanomaterial synthesis method. This work reports the deployment of flame spray pyrolysis to the synthesis of gallium zinc oxynitride solid solutions. The solid solutions were prepared using two different precursor compositions expressed as urea to total metal ratio, solvent mixtures and synthesis conditions, and X-ray diffraction and UV-Vis spectra were employed for preliminary characterizations. Based on solvent screening results, no single solvent satisfied the solubility, boiling point and combustion enthalpy requirements. The XRD data shows that solvents of weak combustion enthalpy resulted in either amorphous or very low crystalline,

vii bimodal and inhomogeneous phase solid solution. However, solvent mixtures of adequate combustion enthalpy yielded single phase and highly crystalline solid solutions with all peaks indexed to the hexagonal wurtzite ZnO structure irrespective of synthesis conditions. The absence of extra peaks coupled with the slight lattice contraction relative to pristine ZnO satisfied the necessary conditions to qualify the as-prepared samples as solid solutions of GaN and ZnO. Lattice contraction was favored as urea to total metal ratio decreased while peak broadening occurred as the urea to total metal ratio in the precursor solution increased while the crystallinity lowered. This could be possibly due to increased dopant concentration in the

ZnO lattice. Samples made with DI H2O + butanol solvent mixture were more crystalline and of larger crystallite size than those made from methanol + butanol mixture which is attributed to the fact that the solvent mixture supplied higher rate of energy (1.17 kJ/s) than the methanol- butanol (1.12 kJ/s) to the flame.

Tauc’s plots were used to estimate the direct bandgaps of the samples. With DI H2O + butanol mixture, the bandgaps are 3.08 and 3.12 eV with smaller bandgap belonging to the sample with higher urea to total metal ratio in the precursor solution. Comparing with the XRD data, the smaller bandgap corresponds with lattice contraction. But for the methanol-butanol mixture, the bandgaps are 3.07 eV and 3.16 eV with the smaller bandgap belonging to the sample with lower urea to total metal ratio. This perhaps, support the claims as reported in Literatures that bandgap is not solely defined by dopant content. Overall, based on bandgaps, the synthesized samples can drive photocatalytic activity in the UV region. Hence, optimization of the synthesis technique is necessary to further narrow the bandgap to the visible light range.

viii

CHAPTER ONE: INTRODUCTION

Given the rising imbalance between global energy demand and supply coupled with the already well known environmental and allied challenges such as climate effects, limited supply, cost, inefficiency etc associated with the conventional energy mix currently in vogue, a search for not only renewable, eco-compatible but also affordable alternative energy sources has been the hub of modern researches[1], [2]. Among the explored alternative sustainable fuels, hydrogen presents itself as the most viable replacement for the present fossil-based fuels; besides its eco-friendliness and higher energy yield relative to the popular gasoline, hydrogen can be readily found in common energy sources and offers the advantage of multiple storage options[2]. Unfortunately, the current commercial supply of hydrogen comes mainly from the steam reforming of methane and fossil allied fuels - leading us back to the undesirable environmental challenges inherent in the utilization of fossil fuels[3]. On this backdrop, Liao et al.[2] argues that unless its production is untied from fossil fuels, it is perhaps, misleading, to describe hydrogen as an eco-friendly energy carrier. Hence, major efforts are being directed towards the development and scale up of renewable sources for hydrogen fuel production.

Of all alternative sustainable energy sources, solar energy stands out as one of the most preferred choices. It is universally abundant, unreckoned with direct carbon footprints, renewable and most economically viable[4]. But the intensity and geographical distribution of solar energy supply are unsteady and ununiform. For optimal utilization and compensation for the above set-backs, the development of cost-effective techniques for its conversion into a more convenient usable forms and storage media for energy derived from the sun have become increasingly important[2]. Thermochemical water splitting[5], photobiological water splitting[6] and photocatalytic water splitting[7] are the major solar-energy-based technologies for the generation of hydrogen. Among these, Liao et al. [2], states that photocatalytic water

1 splitting is the most viable and scalable in terms of cost and efficiency; hence, it has attracted the most attention.

From the foregoing, harnessing solar energy for water splitting is the main key towards unlocking the sustainable energy of the future and this has engendered the ongoing global search for best techno-economic strategies for its full-scale development, optimization and applications. The focal point of this research is the development of efficient photocatalysts. To maximize photocatalytic energy production, attention is given to photocatalysts that allow for the utilization of the visible light spectrum which accounts for half of the solar energy. Gallium zinc oxynitride has gained acceptance as one of the high-performing visible-light-responsive photocatalysts. So far, the research in this direction is largely driven by wet chemistry synthetic methods[8]–[12]. Being batchwise processes, among other challenges, these synthetic methods are not amenable to scale up. Therefore, the development of synthesis methods that would meet such requirement in addition to other synthesis conditions have become necessary.

Liquid-feed flame spray pyrolysis, a modification of the relatively well-known traditional flame synthesis method is a continuous single step technique for the production of nanoparticles of high purity and stability. It involves the dissolution of precursors in a suitable solvent, injection of the solution into the flame, followed by the evaporation of solvent, decomposition, nucleation, aggregation and/or agglomeration of particles and subsequent collection on a relatively cold surface. For about two decades now, this elegant approach has been adopted for the synthesis of many catalytic materials[13]–[17].

The purpose of this work is to deploy LF-FSP for the synthesis of gallium zinc oxynitride with potentials for photocatalytic water splitting. It not only seeks to contribute to the search for flexible and scalable methods for the development of photocatalysts towards the realization of sustainable energy but also to the projection of (LF-FSP) as an attractive technique for the synthesis of catalysts and viable alternative to the traditional synthesis routes. In furtherance of

2 the work, flame spray pyrolysis is proposed for the synthesis of ceria-based solutions for energy and environmental applications. Preliminary data on flame stability tests and characterization are reported herein.

CHAPTER TWO: LITERATURE REVIEW

2.1 Flame Aerosol Technologies and Classifications

In general, flame aerosol synthesis is grouped into liquid-fed aerosol synthesis (LAFS) and vapor-fed aerosol synthesis (VAFS). In LAFS, the precursors are usually in the liquid or aqueous states and subsequently nebulized or atomized while in VAFS, precursors are gaseous reactants from the onset. Although the states of the precursors differ, each of this involves self- propagating reactions in gas phase during the synthesis and therefore are lumped as gas-phase combustion technologies [18], [19].

2.1.1 Vapor-fed aerosol synthesis (VAFS)

The volatile gaseous precursor system when injected into a pilot flame, are heated, combusted and decomposed. The decomposed species through a series of interactions, collisions, nucleation, agglomeration and/or aggregation give rise to particles via the gas-to-particle conversion route. This sequence of actions generates energy enough to drive the process to completion such that VAFS does not require external energy input. This process gives birth to well crystalline, homogeneous products of high purity. Although, this technology has been successfully employed even on industrial scale, the limited availability of gaseous precursors and the high cost of available ones are discouraging factors [18].

2.1.2 Liquid-fed aerosol synthesis (LAFS)

Following the identified challenges of VAFs, Sokolowaski et al.[20] proposed and applied an elegant strategy for the utilization of cheap and readily available aqueous precursors as alternatives to the high cost gaseous precursors as used in VAFS. This approach was adopted and extended by the Laine’s group at the University of Michigan for the synthesis of various catalytic materials and became known as flame spray pyrolysis (FSP)[21]–[23]. In this strategy, low volatile aqueous or solid phase precursors are dissolved in a suitable solvent or mixture of solvents, injected and simultaneously atomized and dispersed in flame using combustible gases

4 such as oxygen, air or hydrogen etc. The atomization or dispersion of dissolved precursor is achieved either by using ultrasonic nozzle[20] or applying pressure as currently seen[24]–[30].

Unlike the VAFS process, the particle formation route in LAFS is a competition between (or combination of) gas-to-particle and droplet-to-particle conversion routes[18]. Depending on the enthalpy content of the precursor solution, the LAFS process could be majorly driven by the energy produced by the combustion of the precursor solution or augmented or solely driven by an external energy source as the flame or external heating source etc. When the combustion of the precursor generates at least half of the total energy of the LAFS process, it is referred to as flame spray pyrolysis (FSP), otherwise, it is called flame-assisted flame spray pyrolysis

(FASP). Consequently, FSP is more favorable for precursors of high volatility and high enthalpy of combustion while FASP can be adapted for precursors of low volatility and low enthalpy of combustion[18].

Unlike wet chemistry synthetic processes, flame aerosol technologies are cheap, scalable, one- step, continuous synthesis tools for manufacturing nanomaterials of high purity and crystallinity with minimal waste[31]. As shown in Fig. 1, flame synthesis reduces the steps involved in classical wet chemistry techniques to essentially two.

Fig. 1. Flame synthesis versus wet chemistry[32]

2.2 Particle Formation Mechanism in LAFS

The possible particle formation routes in flame synthesis are summarized in Fig. 2. In VAFS, the precursors are already in gas phases and the product formation occurs only by gas-to- particle via nucleation, coagulation and agglomeration/aggregation. When the vapor precursor is introduced into the high temperature environment, evaporation and decomposition occurs simultaneously. The species from the high temperature decomposition move upwards the flame corresponding with decrease in temperature. Due to this drop in temperature as the species move away from the combustion zone, the flame become supersaturated which induces nucleation that leads to formation of small particles. The nucleation step decreases the supersaturation and the nucleation rate continuously fall as the particles move to the lower temperature region of the flame. Then, coagulation sets in. The particles coalesce through interparticle collisions and result in increase in particle sizes. As the temperature progressively drops and particle size increases, coalescence by coagulation is overtaken by agglomeration and/or aggregation. The particles collide and bind either chemically or physically with one another to yield aggregates or agglomerates, respectively. The gas-to-particle route usually yield homogeneous particle properties.[31], [33]. For FSP, the particle formation route can proceed by either gas-to-particle or droplet-to-particle. Regardless of the particle formation routes, evaporation of solvent is seen as the underlying step for all LAFS [18], [34]. Solvent properties such as boiling point, specific combustion enthalpy and precursor’s decomposition temperature are critical factors that drastically contribute to defining particle formation routes in LAFS. When the solvent has enough energy (combustion enthalpy) to ensure uniform and complete vaporization and decomposition of the precursor, the particle formation route follows the gas-to-particle as discussed in VAFS. In a case where the energy of the solvent is insufficient for complete, uniform, vaporization and decomposition of the precursor, droplet- to-particle route or a competition between (or combination of) droplet-to-particle and gas-to-

6 particle formation dynamics occurs with each giving rise to inhomogeneous particle distribution and morphology. A very quick criterion supported by a couple of experimental evidences as discussed by Li and co-workers correlates the solvent boiling point(푇푏푝) with precursor decomposition or melting (푇푑⁄푚푝) temperature to give at least, a qualitative snapshot of the highly possible particle formation routes. This criterion suggests that when

푇 푏푝 > 1.05, the precursor solution are uniformly and completely vaporized resulting in dense 푇푑⁄푚푝 and homogeneous nanoparticles formation via gas-to-particle route. For a precursor solution

푇 system with 푏푝 < 1.05, heterogeneities in both morphology and distribution due to droplet- 푇푑⁄푚푝 to-particle or combination of both routes. The criterion suggests that in the case where the melting point or decomposition temperature exceeds the boiling point of the solvent, the solvent vaporizes faster and earlier before the complete decomposition of precursors and leave with some precursor vapors to form particles via gas-to-particle route while leaving behind some droplets of the precursor in the flame which then undergo mere drying and solid densification to form particles via droplet-to-particle route. Usually, the non-uniformity during the evaporation step causes partial evaporation of solvent and precipitation of precursor droplets which subsequently go through final vaporization by the remaining solvent and slow drying process to yield hollow nanoparticles [18], [34], [35] or very fast drying process leading to explosion of particles to form smaller fragments[36]. To avoid this scenario, external energy input may be supplied to drive the system into complete vaporization and decomposition as seen in the case of FASP[36]. FSP and FASP processes are essentially the same but are distinguished based on the source of combustion energy. While FSP is driven by the energy of the precursor solution, FASP relies on external energy[34]. Although, Fig. 2 suggests that

FASP product route is mainly by droplet-to-particle route but since the energy of combustion

7 does not depend on the precursor solution, FASP can either go through gas-to-particle or droplet-to-particle depending on the amount of energy supplied by the external source[34].

Overall, particle formation mechanism depends on solvent-precursor system, thermodynamics and overall flame condition, it is therefore very important to point out that the above criterion is just a conceptual necessary condition and not sufficient to predict particle formation routes[31].

Fig. 2. Different particle formation routes and mechanisms for flame aerosol synthesis[36]

2.3 Factors that affect FSP-made nanomaterials

2.3.1 Effects of precursor type

One of the key decisions towards bottom-up rational design of nanomaterials via FSP is the choice of precursors. In addition to the specific enthalpy of combustion, solubility in the chosen solvent, the decomposition or melting temperature of precursors are considerable factors in ensuring that the desired quality of the final products is achieved. Although nitrate and acetylacetonate-based precursors were previously common on the basis of their low cost and commercial availability, their intrinsic properties such as high decomposition temperature, low volatility and low specific combustion enthalpy do not allow for the production of

8 homogeneous products as they do not favor gas-to-particle conversion route [31], [32], [37].

However, based on the criterion discussed in section 2.3, some of the identified drawbacks can be compensated utilizing appropriate solvent and external energy source as seen in the case of

FASP[31]. The above argument suggests that in the case of multiple precursors, the highlighted properties should be very similar to favor homogeneous particle formation.

To overcome the challenges posed by the nitrates and acetylacetonates precursors, organometallic precursors are preferred given their low decomposition temperature, low cost and high enthalpy of combustion which show high probability for homogeneity of products[31].

2.3.2 Solvent selection

Since the energy of combustion in FSP comes from solvent, proper choice of solvent becomes very critical. Besides the solubility and miscibility of precursor in the desired solvent, the physical properties of the solvent such as boiling point/volatility, specific enthalpy of combustion and viscosity are determining factors for choosing a solvent[31]. Although, the boiling point should be high enough compared to the decomposition temperature to satisfy the aforementioned conceptual criterion, it is also desired that given the short time residence in

FSP, the solvent volatility should be within the range that allows for complete its vaporization within the milliseconds of FSP operation. More so, the solvent should have a very high energy of combustion to drive the FSP process through the gas-to-particle conversion route to ensure that the desired properties such as homogeneity and morphology of particles are achieved[29],

[35]. Solvent properties are often exploited to compensate for any drawback in precursor properties such as low enthalpy of combustion and high decomposition temperature[31].

A common practice is the combination of at least two miscible solvents when it is difficult or impossible to meet the criteria of solvent-precursor system using a single solvent. For example, the high enthalpy of combustion, high boiling point, low cost, availability and solubility make

2-ethylhexanoic acid (EHA) a very good solvent for most organometallic precursors. However,

EHA is a highly viscous solvent which affects the efficiency of its dispersion by commonly used gases such oxygen and air. To beat this drawback, EHA is used along with one or two miscible solvents of very low viscosity such as toluene and xylene [31], [38].

2.3.3 Precursor and dispersion gas flowrates

Apart from the solvent and precursor effects, the overall flame condition has predominant effects on the particle’s characteristics. Increase in precursor flowrate with other conditions fixed, results in increase in temperature of the combustion zone and flame height due to increase in energy density caused by the increased fuel amount delivered per unit time for combustion. This results in longer residence time of particles in the flame and high temperature exposures with overall effect on the physico-chemical characteristics of the nanoparticles[32],

[39].

The precursor solution is usually dispersed or atomized into the flame through a nozzle system using combustible gases with air and oxygen being the commonly used. The spray sizes and uniformity are majorly defined by the dispersion gas flowrate. Dispersion is achieved by the applied pressure drop between the precursor solution entry point and the spray outlet at the nozzle tip. The pressure drop is determined by the dispersion gas flowrate as well as the area of the nozzle tip. At constant nozzle tip area, the pressure drop has a direct correlation with the dispersion gas flowrate. Invariably, the spray characteristics such as sizes, angle, length and velocity are defined by pressure drop. The higher the pressure drop (high flowrate), the higher the velocity and the finer the spray (smaller particles of higher specific surface area). This leads to fast, uniform and complete combustion with shorter residence for the particles yielding homogeneous fine particles. It has been reported that lower pressure drops could result in low spray velocity, longer residence time, bigger spray (particle) sizes with smaller specific surface area [28], [39]. These observations show that precursor-to-dispersion gas ratio is a very critical

10 factor in not just defining the flame condition but also the overall property of the synthesized nanomaterials. It is also agreed that the final product characteristics is highly impacted by the quenching rate[40]. One of the parameters for influencing the quenching is pressure drop. The higher the pressure drop, the higher the quenching rate. At high quenching rate, the particles have very limited residence time in the flame and quickly leave the flame as either metastable[41] or amorphous materials[42].

For simplicity, the effect of the precursor rate and dispersion gas rates is unified in a single parameter called the equivalence ratio, Φ. In the above definition, the oxidant is the dispersion gas while the precursor solution[28]

푚표푙푒푠 표푓 표푥𝑖푑푎푛푡 ( ) 푚표푙푒푠 표푓 푓푢푒푙 푠푡표𝑖푐ℎ Φ = 푚표푙푒푠 표푓 표푥𝑖푑푎푛푡 ……………………………………………………………. (1) ( ) 푚표푙푒푠 표푓 푓푢푒푙 푎푐푡푢푎푙

2.4 Solar Energy for Hydrogen Production: Photocatalysis

Photocatalysis is a direct mimicry of the well-known natural phenomena of photosynthesis in green plants. Photocatalytic water splitting involves the utilization of solar energy for splitting of water into hydrogen and oxygen in the presence of a catalyst-called photocatalyst. In 1972, using TiO2, Fujishima and Honda successfully achieved water splitting using mainly energy derived from the sun. Since then, there has been an increasing interest in photocatalytic water splitting for hydrogen production[7]. The beauty of photocatalytic water splitting is that it seems to satisfy almost all the criteria of what are considered to be the ideal energy source for the future. It makes use of inexhaustible , readily available and free energy source (sun) to produce the desired safe, comparatively high efficient, eco-friendly fuel (hydrogen) using universally available material(water)[1], [43]. There are several sources of hydrogen. A concise summary of these sources and production routes are presented in Fig.3.

Fig. 3. Different H2 production routes (extracted from https://hydrogeneurope.eu/hydrogen-production-0)

2.4.1 Principle of photocatalytic water splitting

The key reaction sequence for photocatalytic water splitting is photoirradiation of the photocatalyst with light of higher energy, valence-band-to-conduction-band electron transfer,

+ + water oxidation to give O2 and H and reduction of the H to H2 [1],[7]. Hydrogen production via photocatalytic water splitting can be achieved by either photochemical or photoelectrochemical cell reactions. While photochemical reaction is a light-only-induced chemical reaction, photoelectrochemical reaction makes use of light and catalyst[2].

Photocatalytic water splitting is achieved by two complementary reactions occur: reduction of the generated protons into H2 (forward reaction) by the conduction band (CB) and oxidation of the electron donor to give electron acceptor (backward reaction) by the valence band (VB) except in case where recombination occurs-leading to no net reaction. It is also not uncommon to see this reaction facilitated using two photocatalysts [2], [44]. This is called the Z-scheme as displayed in Fig.4 [44].

2H2푂 → 2퐻2 + 푂2 ……………………………….………………………….2

The complementary half-cell reactions of the above overall water splitting reaction can be written as:

+ − Reduction: 4퐻 + 4 푒 → 2퐻2…………………………………………………………... 3

+ + Oxidation: 2퐻2푂 + 4 ℎ → 푂2 + 4 퐻 ……………………………..…………….…….4

From the above, it clear that the electrons drive the reduction reaction while the holes are responsible for the oxidation reaction. Therefore, suppression of recombination of the electrons and holes is a critical parameter to ensure efficient photocatalytic water splitting.

Fig 4: Diagram showing the Z-system for photocatalytic water splitting[44] N.B: CB = conduction band, VB = valence band, D = electron donor, A = electron acceptor.

Although there is yet to be a generally accepted technique for the measurement of the actual performance of a photocatalyst, Quantum Efficiency (QE) and Applied Bias photo-to-Current

Efficiency (ABPE) are among the popular expressions for representing the efficiency of solar energy conversion. None of these methods accounts for any loss of incident photons in a photocatalytic system and hence could be regarded as theoretical efficiencies. However, QE could be more accurate as it does not involve external voltage unlike ABPE. Quantum

Efficiency is the percentage ratio of the number of electrons generated to the incident photon of known wavelength applied on the photocatalyst film[45],[46].

푁푒푓푓 ἠ푄퐸 = ∗ 100………………………………………………………..……………………5 푁푡표푡푎푙

푁푒푓푓 = number of electrons-holes pair per given light irradiation, 푁푡표푡푎푙= total number of incident photons.

To accurately evaluate the photocatalytic efficiency of solar-energy-driven water splitting, energy loss due to solar irradiance and chemical conversion efficiency are important.

Unfortunately, ἠ푄퐸does not include these terms[2]. A more encompassing formula for

13 estimating the true efficiency of solar-energy-based water splitting called solar-to-hydrogen conversion efficiency (STH) and simply defined as energy output (chemical) per energy input

(solar) [46] is given below:

퐻 푘퐽 푚푚표푙 2 × 237 푠 푚표푙 푆푇퐻 = [ 푚푊 ] AM1:5 …………………………………………………………………………………………6 푃 ( )×퐴((푐푚2) 푡표푡푎푙 푐푚2

G (237 kJ/mol) is the change in standard Gibbs free energy per mole of H2, produced from water splitting, 푃푡표푡푎푙 is the input solar power density equivalent to the ASTM’s Air mass 1.5

Global (AM 1.5 G) G173 standard. A is irradiated area. The STH formula above is useful and valid only when the stoichiometric H2 and O2 produced without a sacrificial electrons or acceptor are known. This could be achieved using gas chromatography or mass spectrometry or any other reliable quantitative analytical technique. STH is considered the common and most important efficiency parameter for water splitting[46].

In some occasions, such as during the diagnostic measurements of photoelectrochemical cell, external bias is supplied between the working and a reference electrode. This applied bias

(voltage) is usually less than the required thermodynamic potential for water splitting (1.23

V)[46]. Therefore, the total photocurrent generated is not attributed just to water splitting as the current due to the applied voltage is accounted for and therefore not regarded as a true or direct measurement of the efficiency of the water splitting process. The quantum efficiency

(QE) obtained via this approach is called applied bias photon-to-current efficiency

(ABPE)[46], [47].

퐽 ( 푉 −푉 퐴퐵푃퐸 = 푝ℎ 푟푒푑표푥 푏) ...... 7 푃푡표푡푎푙 푚퐴 Where 퐽 ( ) is the photocurrent density generated alongside with the applied bias, 푉 푝ℎ 푐푚2 푟푒푑표푥 is the thermodynamic redox potential for water splitting = 1.23 V, 푉푏is the applied voltage bias

푚푊 and 푃 is the light intensity in ( ). 푡표푡푎푙 푐푚2

Another way of measuring the efficiency of PEC is by quantifying the number of the photogenerated currents per incident photon as a function of the wavelength of the incident light[46], [47]. Again, this measure of efficiency is used majorly as a diagnostic tool for evaluating the performance of the photoelectrode for water splitting. This measure of efficiency is called incident-photon-to-current efficiency (IPCE). The highest possible solar-to-hydrogen conversion (STH) achievable in a PEC can be estimated by integrating the IPCE measurements over the solar spectrum. However, the measured STH from IPCE is only true if done without applied bias. IPCE can be evaluated with the following formula[47]:

푇표푡푎푙 푒푛푒푟푔푦 표푓 푒푙푒푐푡푟표푛 푐표푛푣푒푟푡푒푑 푒푙푒푐푡푟표푛푠 IPCE (λ) = ………………………………...……..8 푇표푡푎푙 푒푛푒푟푔푦 표푓 푡ℎ푒 𝑖푛푐𝑖푑푒푛푡 푝ℎ표푡표푛푠

퐽푝ℎ표푡표 (λ) ℎ×푐 × = 푒 λ ………………………………………………………………….8.1 푃(λ)

푚퐴 Where 퐽 is the photocurrent density ( ) , λ (nm) is the wavelength of the incident 푝ℎ표푡표 (λ) 푐푚2 photon, h is the Planck’s constant (6.63 × 10-34 J s), c = the speed of light = 3 ×1017 nm/s and

푚푊 P(λ) is the incident light power intensity at λ ( ). 푐푚2

In IPCE, the losses due to transmission and reflection are neglected which affects the accuracy of IPCE. To account for these loses, quantum efficiency based on the amount of photon absorbed is utilized. This is called the absorbed photon-to-current conversion efficiency and shortened as APCE[47].

퐼푃퐶퐸 (λ) 퐼푃퐶퐸 (λ) IPCE (λ) = = × 100 퐴(λ) 1−푅−푇

Where A is the optical absorption, R is the fraction reflected and T is the fraction transmitted.

Of all the various methods, STH is the most important, accurate and is used as benchmark for estimating the efficiency of water splitting[46].

2.4.2 Requirements for photocatalytic overall water splitting

Following the demonstration of photo-electrochemical cell water splitting by Fujishima and

Honda in 1972 using TiO2, several photocatalysts have been developed and tested for the same

15 purpose with the goal of enhancing the process for scalability. However, majority of these photocatalysts just like TiO2 drive overall water splitting only in the UV region of the solar energy spectrum. But the UV light is less than 5% of the solar energy spectrum while the visible light accounts for about 50%. Therefore, photocatalytic water splitting reaction utilizing the

UV light is significantly low and considered a major setback towards the successful commercialization of this technology. Consequently, key research efforts are directed towards the realization of visible-light-responsive photocatalyst for overall water splitting[2]. In general, common requirements for an ideal photocatalyst for solar-based overall water splitting have been identified:

(a). Visible light region has wavelength between approximately 400 nm to 700 nm which corresponds to a bandgap of at most 3.0 eV. Bandgap engineering is therefore, the first step towards the design and development of visible-light-driven photocatalyst.

(b). The photocatalyst must have its band edge located at a position that can drive effectively photocatalytic water splitting.

(c). The catalyst should possess suitable morphological properties such as high crystallinity and particle size/distribution to suppress the potential recombination tendencies of the photogenerated VB holes and CB electrons[48], [49]. Particle sizes of catalyst have been shown to have effects on photocatalytic performance. The rate of transport of these charge carriers, adsorption of reactants and recombination of the charge carriers have been reported to depend on the sizes and morphology of the catalyst[50]–[52]. It is important to understand the compromise between the desired small particle size and excessively small particle size. In the study of photocatalytic reduction of CO2 using pristine TiO2, it is shown that higher yields were obtained as the particle sizes decreased with the optimum yield achieved with 14 nm nanoparticles. Below 14 nm size, the yield decreased. In addition to increase in surface area as particle size dropped, it was reported that particle size affected the electronic properties of the

16 catalyst as seen in the variation of bandgap with particle sizes[52]. However, in another study involving photocatalytic mineralization of acetic acid using TiO2, photocatalytic activity increased as SSA increased with crystallite size. The catalyst with the smallest SSA showed the highest activity. This observation was attributed to the crystallinity. The catalysts with higher crystallinity showed higher activity[53]. It is also suggested that either case, improved activity could be a combined effect of morphological and optoelectronic properties[52].

Furthermore, Amano, F. et al.[54] evaluated the effect of particle sizes of WO3 for photocatalytic water splitting. Their work stated that for water splitting, optimum activity was reached with 200 nm nanoparticles. Above 200 nm, the activity plateaued while below it, activity decreased. The drop in activity was claimed to be due to fast recombination of electrons and holes which competed against oxygen and hydrogen molecules release.

Therefore, from the above, it could be said that the effect of particle size on photocatalytic reactions will vary from one reaction to another reaction. More so, an optimum particle size is one that is not too big to promote efficient transport of charges by diffusion of the photogenerated electrons to reaction sites and not too small to absorb enough sunlight necessary to achieve photocatalytic water splitting[52], [55].

2.5 Visible-Light-Driven Overall Photocatalytic Water splitting

Photocatalytic water splitting has been realized using pristine metal oxides such as TiO2[7],

ZnO, WO3 , Fe2O3[1] etc. However, due to the high band gaps of metal oxides (> 3 eV), these materials can only drive photocatalytic reactions in the UV region of the sunlight spectrum

(wavelength < 400 nm). The UV region makes up only about 5 % of the solar energy while the visible region accounts for almost 50 %. To optimize photocatalysis, it is necessary that the catalysts must be able to drive the reactions using the visible light region which falls in the wavelength range of 400 – 700 nm. For water splitting to occur, the photocatalysts must have a band structure with enough potential relative to the normal hydrogen electrode (NHE). To

17 meet this thermodynamic requirement for stoichiometric water splitting, the valence band maximum must be located at +1.23V relative to NHE while the conduction band minimum must be at negative potential (zero relative NHE). This difference in potential shows that a minimum bandgap of 1.23 eV is required[43], [56]. Metal oxide structure satisfies the band structure described above but its wide bandgap limits their choice for photocatalytic water splitting since they cannot utilize the major component of the solar spectrum. Therefore, photocatalysts that meet the minimum bandgap and utilize the visible region are the desired candidates for water splitting. A common strategy is to develop catalytic materials that have similar band structure as the metal oxide but with band less than 3 eV. Therefore, bandgap engineering is employed to photocatalysts with bandgap less than 3.0 eV which can drive overall water splitting in the visible light region (wavelength > 400 nm) [57].

Doping the primitive metal oxide with some other metallic or non-metallic oxides have been useful for tuning of bandgaps. Most explored photocatalysts to meet the requirements discussed above are based on oxides or nitrides of the d0 or d10 transition elements(semi-conductors).

Some of the reported visible light photocatalyst derived from d0 transition metals include

TaON[58], Ta3N5[58], LaTiO2N[59],CaTa2ON[14], Ln2Ti2S2O5 (Ln = Pr, Ho, Nd, Sm, Gd, Tb,

10 Dy, Er)[60] while some d -based photocatalysts are (Zn1+xGe)(N2Ox)[61], MSnO3 (M = Sr,

Ca, Br)[62], (Ga1-xZnx)(N1-xOx)[63] etc .The strategy relies on partial or complete substitution of the oxygen atoms in the metal oxides with elements of higher negative electrode potential than the oxygen atom. This principle aims at introducing an orbital above the O2p orbital to realize a negative shift of the valence band maximum without affecting the conduction band minimum. While the metal oxides have unsuitable bandgaps and prone to recombination effects, the bandgaps of the photocatalysts from the oxygen atoms substitution with anionic materials were successfully tuned to the desired visible light range; unfortunately, the later class of photocatalysts suffers photodissociation. Zhu and Zach[56] synthesized CdS with

18 bandgap of 2.6 eV for photocatalytic water splitting. Despite its photocatalytic activity in the visible light region, it is reported that the sulfur ion in the photocatalysts got oxidized by the photogenerated holes with the Cd2+ washed into the solution. Similar effect occurred with

ZnS[56].

However, with oxynitride or purely metallic nitrides obtained by partial or complete replacement of the oxygen atoms with nitrogen, photodissociation-resistant photocatalysts with favorable bandgaps have been reported [1], [49], [64], [65].

2.6 Gallium Zinc Oxynitride (GaN:ZnO) as Photocatalyst for Overall Water Splitting

Overall water splitting has been achieved utilizing oxynitrides-based photocatalysts but in the

UV-region. Since the d10- (oxy)nitrides electronic structure show enhanced mobility of the photogenerated in the CB resulting in higher in photocatalytic efficiency than d0 (oxy)nitrides, the former has become subject of research[66]. GaN is well-known d10 semi-conductor that has been explored for various electronic and electrical applications[67]. Even though it shows promise for photocatalytic water splitting, its high bandgap (3.4 eV) limits it to the UV-region application-undesirable. However, by bandgap engineering, its bandgap is tunable to the visible light range. By doping with ZnO (3.2-3.4 eV), it was observed that it is possible to tune the bandgap of primitive GaN without affecting the conduction band minimum. Interestingly, the obtained photocatalyst has a bandgap less than either of the combined two semi-conductors.

This indicates that the as-synthesized photocatalyst was a solid solution and not just a mere mixture of two oxides and the formation is enhanced by the wurzite structure of both ZnO and

GaN [68].

2.6.1 Synthesis methods for GaN:ZnO

Maeda et al.[68] reported for the very first time, the synthesis and evaluation of the photocatalytic overall water splitting performance of a solid solution of GaN:ZnO. Although the photocatalytic efficiency of the prepared photocatalyst was substantially low, a significant

19 improvement was observed in the presence of suitable co-catalyst (RuO2). The reported photocatalytic efficiency with the co-catalyst was claimed to be the highest as available in open literatures. This stimulated interest in this new material as a candidate for overall photocatalytic water splitting with special reference to the development of cost-effective methods for its synthesis. Adeli and Taghipour [48] classified the available methods for the synthesis of gallium zinc oxynitride into three on the basis of temperature requirements: High temperature solid state reaction, medium temperature and micro-wave methods.

(a). High temperature synthesis (HTS)

This is the traditional method for the synthesis of very high crystalline solid/powder materials by heating a system of other solid materials in the presence of reducing agents usually gases such as ammonia at temperatures greater than 1000 K. Maeda et al. [68] utilized the technique for the synthesis of high crystalline gallium zinc oxynitride at 1123 K by heating a mixture of

ZnO and Ga2O3 in the ratio 2:1 using ammonia as the nitridating (reducing) agent. The synthesized yellow solid solution had a bandgap favorable for the visible light range photocatalytic activity. Because of the high temperature reducing condition, a significant loss of Zn atoms through volatilization to metallic Zn and high density of crystal defects, a negligible quantum efficiency was noted and even a very low quantum efficiency was observed using RuO2 as the co-catalyst. To improve the photocatalytic quality of the gallium zinc oxynitride produced via HTSS, several strategies have been employed such as modification of co-catalysts, introduction of excess Zn and oxygen to compensate for the drastic Zn loss, reduction of nitridating time, use of alternative nitridating agents and use of special types of precursors and incorporation of additional synthesis step(s) etc.[4], [69]–[72]. Despite all current efforts, there is yet to be a significant improvement in the photocatalytic efficiency of gallium zinc oxynitride derived from this technique as well as the efficiency of the method.

(b). Medium temperature synthesis (MT) (823 K ≤ T ≤ 1000 K)

Since the band gap and to an extent, the photocatalytic performance of gallium zinc oxynitride are determined by the amount of Zn present in it, a strategy to suppress the huge loss of zinc and reducing the operating temperature in the HTS (>1000 K) technique, is seen as one of the effective ways for achieving zinc-rich gallium oxynitride. More so, this method lay claim to reduced synthesis time. Yan et al. [73] was able to synthesize visible-light-range gallium zinc oxynitride via ammonia-based nitridation of ZnGa2O4 at 953 K.. Even though this method achieves the desired band gap and more Zn-rich as-synthesized materials, there is an obvious decrease in the crystallinity of the product. The photocatalytic performance of the gallium zinc oxynitride by this technique was not reported.

(c). Microwave synthesis technique

Yang et. al [74] is the first to synthesize gallium zinc oxynitride by the indirect heating of a mixture of gallium oxide, zinc oxide, zinc powder and urea as the nitrogen source in a microwave oven for 10 mins. The estimated bandgap of the prepared material falls within the visible- light region (2.47-2.64 eV) as the Zn volatilization was greatly suppressed. Although, the prepared material has not been tested for photocatalytic water splitting, it showed high photocatalytic activity for the decomposition of isopropyl alcohol. This high activity was attributed to the high porosity induced by the foam-structure template generated by the release of release of volatile gases enhanced by urea presence. From the XRD pattern, it is crystal clear that the as-synthesized photocatalyst has the lowest crystallinity compared to the traditional methods. Furthermore, with the low temperature and very short synthesis, there is high probability that other phases resulting from poor nitridation and incomplete combustion of urea could form.

(d). Combustion synthesis

This technique was recently reported for the synthesis of gallium zinc oxynitride and was therefore not included in the review cited above. Kennedy and Meekins [75] prepared gallium

21 zinc oxynitride via combustion synthesis method. This involved the direct heating of a mixture of gallium nitrate, zinc nitrate and urea in water for about 30 minutes at 773 K. The comparatively low temperature and reduced synthesis time compared to the above methods could have been responsible for the reportedly reduced loss of zinc atoms via volatilization and therefore, presents this method as a relatively promising technique. However, the X-ray diffraction pattern shows a reduced crystallinity when compared with the other techniques previously discussed. The synthesized photocatalyst, with bandgaps 2.22-2.8ev demonstrated photocatalytic activity under visible light region but showed no improvement even in the presence of co-catalyst. In most cases, photocatalytic surface reactions are slow, accumulation of charges on the surface of the catalyst will lead to recombination and decreased in activity.

To avoid this, secondary catalysts are added to facilitate consumption of any deposited electrons. These secondary catalysts are called co-catalyst. These co-catalysts can help to provide more suitable active sites for reactions to occur and hinder the recombination of electrons, thereby favoring the formation of molecular hydrogen. Therefore, co-catalysts can be used to improve the efficiency of photocatalysts[55]. This synthesis technique is prone to the inadequacies inherent in a typical batch process. Overall, more work is required for the optimization of this technique.

2.7 Flame Spray Pyrolysis for Synthesis of Photocatalysts

Despite the efforts discussed so far, the current synthesis techniques for gallium zinc oxynitride with visible light photocatalytic activities are still laden with inadequacies. Although the traditional method of nitridation (HT) produces very high crystalline stable photocatalyst, high volatilization of zinc, safety concerns of ammonia and synthesis duration are some of its drawbacks. While other techniques suffer less from Zn loss, their as-prepared photocatalysts are less crystalline and prone to additional phases and impurities[48]. More so, majority of the synthesis methods currently explored are wet-chemistry-based and mostly multi-step batch

22 processes. Furthermore, as noted earlier, the morphological properties such as particle structures and sizes of photocatalysts are of critical importance to their efficiency and performance. None of these discussed methods has demonstrated the capacity to control or systematically control the morphological properties of their prepared materials. Therefore, to harness the potentials of gallium zinc oxynitride as the recommended candidate for large scale visible-light-driven photocatalytic overall water splitting, a best techno-economic synthesis strategy must be devised. FSP as a very high temperature (above 2000 K) technique is expected to give highly pure, homogeneous single phase crystalline and stable nanocatalysts. FSP occurs at a temperature almost twice that of the HTS technique. However, because FSP reaction happens in milliseconds[31], it is expected that zinc losses could be significantly minimized.

More so, unlike the wet chemistry methods, FSP enjoys the reputation of being a single-step technique. This will not only reduce the reaction time but also minimize the potential of introducing impurities as often witnessed in a multi-step process. Furthermore, except for the

HTS (with capacity for continuous operation), current approaches for the development of gallium zinc oxynitrides are basically batch processes, hence, not ideal for large scale production. But FSP enjoys the beauty of having several approaches utilized in the tuning and control of the physico-chemical properties of its products such as solvent-precursor system, precursor and dispersion gas loadings, quenching rates and air entrainments etc. [25].

Therefore, FSP is a continuous, single-step, multi-parametric tuning high temperature synthesis technique that has been utilized for the design and development of various catalysts

Besides the use of FSP for the synthesis of various traditional catalytic materials, it is also being deployed for the development of photocatalysts. Ranging from simple to mixed metal oxides,

FSP-made photocatalysts have been reported [37], [76]–[78]. Teoh [57] presents a very concise but excellent review on FSP-made photocatalysts, highlighting the progress, challenges and prospects. Since TiO2 is usually set as a benchmark for the characterization of photocatalysts,

23 this section presents an overview of some of the FSP-made TiO2 for photocatalytic applications.

The work by Teoh et al.[79] compared the photocatalytic efficiencies of FSP-made TiO2 and commercial flame-made Degussa P25 of comparable structural and morphological features for the degradation of various organic compounds. The FSP-made photocatalyst showed superiority in the mineralization of saccharides while the conventional P25 for phenol and methanol. This therefore implies that the photocatalytic activity of titania for organic compounds degradation is a function of the type of compound and other experimental conditions.

In an enclosed FSP configuration, Kho and his group shows that the polymorphic composition of TiO2 nanocatalyst can be tuned by controlling the flame condition through the manipulating the equivalence ratio. They observe that at stoichiometric flame condition, the anatase phase was more favored while at slightly fuel rich condition, the rutile phase dominated. The works also suggests that besides the basic parameters such as precursor and dispersion feed rates, the geometry of the enclosure for the flame impacts on the structural properties of the as-prepared materials. This is not surprising as the enclosure would determine air entrainment rate which in turn impact on the quenching rate. Overall, the enclosure has a direct consequence on the temperature profile of the flame. This work further portrays FSP as a tool for rational design of nanomaterials. In general, FSP-made TiO2 is often reckoned with being compositionally homogenized, exhibition of efficient charge transfer and surface area that contributes to improved photocatalytic activity[25], [80].

In furtherance of the characterization of FSP-made TiO2 for photocatalytic and photochemical applications, Tsekouras and co evaluated the charge mobility of FSP-made TiO2 for dye- sensitized solar cells. The charge transport properties of the prepared materials agreed with that commercially available Degussa P25. Utilizing the parameters such as feed rates variations and

24 air entrainment (quenching rates), the worked shows that FSP could be used to precisely control the physico-chemical and charge transport behaviors of TiO2 unlike available wet chemistry synthesis methods[78].

Using FSP, Amal et al.[81] doped TiO2 with 2% Cu and 50% F and compared their performance against pristine TiO2 (P25) for photocatalytic oxidation of acetaldehyde in the UV region. While the non-metal doped TiO2 demonstrated slightly improved photocatalytic activity than the P25, the Cu-doped sample showed declined activity-which is attributed to the increased generation of the terminal hydroxyl functional group. A quick inspection of the XRD pattern showed that both doped samples showed high crystallinity and contained both the rutile and anaphase of TiO2 like the P25.

By doping titania with ceria, Chaisuk et al.[24] was able to engineer the bandgap of TiO2 into the visible light region. The work reported that the absorbance of pristine TiO2 shifted from

388 to 467 nm as the dopant concentration increased from 10 to 30 % ceria. It also shows that the surface area and anatase: rutile varies with the amount of ceria incorporated into the lattice of TiO2. Although the photocatalytic activity of the catalyst synthesized in this work is not known as there was no actual photocatalytic application besides the UV-vis measurements, but, relying on the UV-vis spectra, it is expected that the photocatalyst should be active in the visible light region unlike pristine TiO2 and CeO2.

2.7.1 Mixed oxides for photocatalysis and challenges

Formation of mixed oxides and solid solution catalysts is seen as one of the viable strategies for extending the stability and activity of catalyst. This has made it possible to access catalytic properties that are usually difficult, if not impossible to achieve with single oxide materials[82]–[88]. Although this has been deployed for many conventional catalysts, as already seen above, it is also being utilized for Flame-made photocatalysts. Despite the beauties of FSP for the synthesis of conventional catalysts, its utilization for photocatalysts development

25 is limited by some inherent setbacks. Although the high temperature environment during FSP often results in the formation highly crystalline and low defects concentration catalysts[26]–

[30], the very short residence and high quenching rates of particles in the flame give rise to metastable[41] and even amorphous nanomaterials[42]. Poorly crystallized materials are reckoned with unwanted charge trap defects that act as recombination centers for electron and hole pairs which reduces the efficiency of the photocatalytic process[57]. To compensate for the short residence enclosed flame configuration is adopted during FSP but this has drastic effects on the surface area and compositions of the final products[42]. More so, lack of suitable precursors for some class of metals and uncertainty or difficulty in controlling the desired oxidation states and compositions of dopants in the host remain a challenge in the preparation of mixed oxides or solid solution photocatalysts[57].

CHAPTER 3: EXPERIMENTATION

A customized flame spray pyrolysis system was designed and built. The system consists of essentially the flame reactor, feed injection system and particulate collection unit.

3.1 Flame reactor system

The flame reactor is a standard water-cooled 316 stainless steel Mckenna burner with a center tube purchased from Holthius and Associates. To reduce non-uniformity and variation of temperature in the radial direction of the premixed flat flame (flamelets), the burner contains a bronze-made porous sintered plug (6 cm in diameter) equipped with a spiral circuit for the passage of cooling water. The sintered plug is the mixing point for the methane and burner air for the generation of the flamelets. More so, the flat flame burner is fitted with a coaxial sintered shroud ring made of bronze with a 0.25inch compression fitting for the delivery of air. The shroud air protects the premixed flamelets (pilot flame) from outside air interferences and therefore helps to maintain flame stability. The center tube is for the insertion of the atomization nozzle.

With this, the flat burner offers a steady temperature profile across the flame which is assumed to be 1-D especially within the lean to stoichiometric regimes[89]. This is the primary reason for the choice of this burner system so as to help in maintaining uniform temperature and stability of the pilot flame.

The figure below shows the inhouse fabricated FSP system (a) and (b) shows the burner with premixed flamelets. As part of safety measures, a flame arrestor is installed on the methane line.

Fig. 5: (a) Customized FSP set-up (b) set-up with the premixed flamelets

3.2 Nozzle system

This is a Schlick full cone, two-fluid pneumatic atomizer obtained through Orthos Liquid

System. The two-fluid system has a pipe of circular cross-section serving as the route for the precursor and an air cap (dispersion gas – oxygen- route) ending with a conical geometry. The end of the liquid route is fitted with nozzle tip of 0.5 mm while the gas route ends with a diameter of 2 mm. The nozzle tip edges out with the air cap with an adjustable annular gap area between the tip and the air cap. With spacers, the annular gap is adjusted by changing the height of the nozzle tip from the end of the air cap. Although, it is difficult to quantify annular gap area, the pressure drop across the nozzle tip is utilized to have a sense of it. The nozzle is inserted through the center tube of the burner and suspended using a retort stand.

The inlet to the precursor and the dispersion gas routes are connected to 0.125 mm and 0.25 mm tubing respectively. The precursor is delivered to the nozzle by NE-300 infusion syringe pump acquired from New Era Pump Systems Inc. Methane, burner air and oxygen gas flowrates are metered and controlled using Brooks mass flow controllers while the shroud air is controlled by rotameter obtained from Cole Parmer.

3.3 Particles collection unit

Particles collection in flame spray pyrolysis is based on thermophoretic effect. The synthesized nanoparticles are collected on a 15 cm Whatman Glass fiber filter paper (grade B) attached to

28 a stainless-steel filter holder. To keep the filter holder cold, cooling water flows through a 0.25 mm channel in the filter holder. The water cooling is to ensure that the collected particles are not exposed to temperature high enough to cause further reactions of the particles via agglomeration and aggregation. The filter holder is suspended at 15.5 inches above the burner surface and particle collection was facilitated using Varian SD-200 mechanical pump supplied by Duniway stockroom corporation.

A schematic representation of the FSP set-up is presented in fig. 6. In the figure, Rot = rotameter; MFC = mass flow controllers; FBA = flashback arrester; PG = pressure gauge.

Fig. 6: Schematic diagram of the FSP system

3.4. Solvent screening

Due to multiple precursors involved, it was necessary to figure out a common solvent for them.

For solubility tests, the three precursors were dissolved in methanol, ethanol, methanol-ethanol, methanol-acetonitrile, ethanol-water, methanol-acetic acid, butanol-water, butanol-methanol respectively. For each case, the precursors were dissolved, stirred and further sonicated to improve solubilities of the precursors.

Using methanol, a solution containing the three precursors (Ga: Zn: U) in the molar ratio of

1:1:10 was prepared, stirred and sonicated for 1 hr. Similar solutions were prepared (unless otherwise stated) using ethanol, ethanol-deionized water (10 ml of DI H2O), methanol-ethanol, methanol-acetonitrile (40 ml). For the Nitrogen-doped samples (ZN), 75ml of total concentration of 0.4 M was used while 75 ml of 0.2 M was used for the synthesis of Zinc oxide

(Z). Subsequently, Ga, Zn and U are represented with G, Z and N (urea as the source of nitrogen). For example, gallium zinc oxynitride samples are represented as GZN1 with subscript denoting sample’s identification number, Z1 would mean Zinc oxide sample 1 while

ZN1 means nitrogen-doped Zinc oxide with Zn: N = 1 etc.

Table 1: Sample compositions and solvents Sample Ga: Zn: N Solvent

GZN1 1:1:10 Methanol

GZN2 1:1:10 10 ml DI H2O + Ethanol GZN3 1:3:10 40ml Methanol + Butanol GZN4 1:3:20 40ml Methanol + Butanol GZN5 1:3:25 20 ml DI H2O + Butanol GZN6 1:6:25 20 ml DI H2O + Butanol ZN1 0:1:1 40ml Methanol + acetonitrile ZN2 0:1:1 40ml Methanol + acetonitrile ZN5 0:1:5 40ml Methanol + acetonitrile Z1 0:1:0 Ethanol Z2 0:1:0 40ml Methanol + acetonitrile Z3 0:1:0 40ml Methanol + acetonitrile

The choice of suitable solvents and precursors is the first step towards achieving homogeneously distributed ultrafine products in flame spray pyrolysis. Since gas-to-particle formation route leads to those desirable product’s qualities, the simple strategy is to employ precursors and solvents with desirable properties (as discussed in sections 2.4.1 and 2.4.2) that favor gas-to-particle conversion over droplet-to-particle mechanism[90]. Therefore, the initial step was geared towards choosing a common solvent for the ternary system. Nitrates are usually

30 soluble in readily available alcohols[90] and with urea showing some solubility in lower molecular weight alcohols too, the initial trials started with methanol and ethanol.

(a). Methanol (GZN1)

0.01 moles of Gallium nitrate and zinc nitrate with 0.1 mole of urea dissolved in 100 ml of solution to yield GZN1 (1:1:10). Although, the precursors seemed well dissolved, urea is known to exhibit limited solubility in alcohols[91] and with nitrate having fair solubility methanol, the prepared solution was stirred and further sonicated to aid homogenization.

(b). DI H2O + Ethanol and DI + Butanol systems (GZN2 to GZN4)

Urea is much less soluble in ethanol than methanol[91]. Also, it was observed that gallium nitrate solubility decreased across the alcohol series leading to the formation of precipitate. The formation of precipitation could also be due to the upset of the fair solubility of urea in the alcohol. However, ethanol and butanol have slightly higher boiling point and enthalpy of combustion than methanol while butanol has higher combustion enthalpy and boiling point than ethanol. Water is a very excellent solvent for all the precursors. Therefore, 10 ml of DI was used to dissolve the three precursors and made up to 100 ml with ethanol while 20 ml of

DI was used in the case of butanol since solubility of the precursors was observed to decrease in butanol. Both of which gave colorless solutions after stirring and sonicating for 1 hr.

With the low enthalpy of combustion of ethanol and boiling point relative to the poor volatility and decomposition temperature of the precursors, introduction of water into the system is likely to drastically discourage gas-to-particle conversion mechanism by further lowering the flame energy [37], [90]. Therefore, there was a need to eliminate water regardless of the amount.

Based on the findings with methanol, GZN5 and GZN6 samples were prepared using 40 ml of methanol and made up with butanol. Methanol primarily served as the base solvent while butanol that has very boiling point and heat of combustion is employed to favor gas-to-particle

[31], [90]. The solution was unclear due to precipitation. After sonication, no precipitate was observed, and the solution appeared clear. The challenge with the methanol-butanol system, is that the slight solubilities of the nitrates and urea especially the Gallium nitrate in methanol only allowed for the complete dissolution of a very small amount of the precursors.

(d) Methanol, Ethanol and Methanol-Acetonitrile systems

Zinc nitrate is comparatively highly soluble in ethanol and methanol respectively and so did not present any solubility issues. Urea are relatively soluble in methanol and acetonitrile respectively and the energy of combustion of acetonitrile is almost twice that of methanol. Just as discussed above, acetonitrile was therefore added to boost the enthalpy of combustion of the system in addition to its higher boiling point. Therefore, 40 ml of methanol was initially used to dissolve the precursor(s) and made up to the desired volume with acetonitrile for the preparation of Z1 to Z3, ZN1, ZN2 and ZN5.

3.5 Nanomaterial synthesis

Samples of solid solutions of gallium zinc oxynitride, zinc oxide and nitrogen-doped zinc oxide were prepared via liquid-fed flame spray pyrolysis at various compositions and synthesis conditions. Stable premixed methane-air flamelets was achieved at stoichiometric equivalence ratio with a total gas flowrate of 12 L/min (methane = 1 L/min, air = 10 L/min and shroud air

= 1 L/min) and used for all nanomaterials synthesis. In the case of precursor flame, the equivalence ratio (molar basis) was computed by taking the solvents and urea (where present) as fuel and oxygen as oxidant and was used together with pressure drop as the basis for the result analysis. Prior to the actual materials synthesis, the performance of the FSP system in terms of particle collection efficiency and flame stability with respect to possible synthesis conditions such as feed rates and dispersion gas pressure drop was carried by synthesizing zirconia using 0.1 M zirconium (iv) oxynitrate. The precursors for gallium zinc oxynitride are

Gallium (III) Nitrate, Zinc Nitrate and urea purchased from Alfar Aesar. Urea served as the

32 nitrogen source for solid solution. The choice for these precursors was based on commercial availability and cost. The gases used in the experiment were obtained from Airgas.

As-prepared nanomaterials scrapped from glass fiber filter paper (grade B). The table 2 below gives the summary of the samples’ identification including synthesis conditions for the samples while the nominal flowrates are presented in Appendix A. However, due to the various synthesis conditions involved, for clarity purposes, samples would be discussed alongside their respective synthesis conditions. Note that the equivalence ratios in table 2 refers to that of the precursor flame as that of the flamelets was held constant.

Table 2: Samples and their synthesis parameters Sample Ga: Zn: N Pressure Equivalence drop(bar) ratio GZN1 1:1:10 1 0.51 GZN2 1:1:10 1 0.95 GZN3 1:3:10 1.5 2.34 GZN4 1:3:20 1.5 3.16 GZN5 1:3:25 1 1.84 GZN6 1:6:25 1.5 2.04 ZN1 0:1:1 1.5 0.81 ZN2 0:1:1 1 0.81 ZN5 0:1:5 1 1.55 Z1 0:1:0 1 0.65 Z2 0:1:0 1 0.55 Z3 0:1:0 1 0.65

3.6 Characterization of synthesized nanomaterials

Preliminary physical characterizations of the as-prepared nanocatalysts were carried out using powder x-ray diffraction, Ultraviolet-visible spectroscopy. X-ray diffraction patterns of the samples were collected using a Rigaku II desktop diffractometer with

CuK훼 (훼 = 1.5406) source radiation in the 2휃 range of 10-80° with a step rate of 2°/min.

UV-vis spectra were collected using Shimadzu UV-2600 in the diffuse reflectance mode and absorbance data conversions were computed with the aid of UV probe software.

Tauc’s plots were constructed and were used to estimate the direct bandgaps.

CHAPTER 4: RESULTS AND DISCUSSION

4.1 Physical characterization

X-ray diffraction patterns were collected to examine the crystal structure of the as- synthesized samples. GZN1 (1:1:10) synthesized at 1 bar and equivalence ratio of 0.51 using methanol was a colorless amorphous material which did not occur as powder but rather as waxy chunks and was not subjected to x-ray diffraction. As we shall see later in this section, the inability to prepare crystalline nanoparticle with this composition is highly likely not due to the synthesis conditions (equivalence ratio and pressure drop) but could be explained in terms of precursor- solvent effect. The boiling point of methanol and combustion enthalpy are 63℃ and 763 kJ/mol respectively (https://webbook.nist.gov). Urea decomposition temperature[92], [93] ranges from 130℃ to over 250℃, gallium nitrate decomposes at 110℃

(https://pubchem.ncbi.nlm.nih.gov) while zinc nitrate[94] decomposition occurs from 300 to

550℃. Without overlooking the impact of the wide differences in the decomposition temperatures of the precursor themselves, but, comparing the boiling point of methanol to even the smallest decomposition temperature (gallium nitrate), it is far less than unity and by the experimentally supported conceptual criterion previously discussed, it is reasonable to infer that the solvent evaporated long before the precursors decomposed. Therefore, precursors droplets did not get enough energy to be properly combusted. Compositionally similar solution prepared and left to evaporate overnight yielded similar material-which supports the above claim. In addition to the short time residence of FSP, the poor volatility coupled with low combustion enthalpy of nitrates strongly contributed to the creation of the amorphous material

[37], [90]. Despite the poor characteristics of the precursors, a solvent of high boiling point or combustion enthalpy would have yielded at least inhomogeneous crystalline materials via droplet-to-particle route[31].

Fig. 7 shows the diffractogram for GZN2 (1:1:10) synthesized at 1 bar and equivalence ratio of 0.95 using the mixture of DI H2O + ethanol. The XRD spectra shows very low crystalline peaks assigned to ZnO and ZnGa2O4 respectively which indicates the formation of mixed oxides since a single-phase gallium zinc oxynitride shows only ZnO structure[75], [95]. There are possible explanations for the mixed oxides formation in GZN2. In terms of compositional effect, it could mean that either nitrogen is not incorporated, or the amount incorporated is small to lead to formation of a single-phase gallium zinc oxynitride. Using the same precursors ratio (1:1:10), Kennedy and Meekins[75] reported to have achieved a single phase of gallium zinc oxynitride via combustion synthesis technique. However, regardless of the Zn/Ga ratios tested in their work, deviation from urea to total metal molar ratio of 5 yielded mixed oxide phases as obtained in GNZ2. The authors, using uv-vis absorption spectra claims that nitrogen was present in the mixed oxide but more likely to exist in the form of ZnOxN1-x. This result suggests that there could be a favorable precursor ratio necessary for the formation of a single- phase gallium zinc oxynitride. On the other hand, it is possible that nitrogen was incorporated, but the gallium was either not incorporated into the ZnO lattice or not all gallium was incorporated. In doping ZnO with only gallium via modified Pechini synthesis method (wet chemistry), Goncalves et al.[96] stated that while Zn/Ga ratio less than or equal to 2% yielded a single phase of ZnO structure, Zn/Ga molar ratio greater than 2% showed mixed ZnO and

ZnGa2O4 phases. The authors states that there could be solubility limit for gallium in the ZnO lattice. This work[96] seems to support the argument by Kennedy and Meekins that there could be an optimum precursor ratio[75].

Without overruling, the above possibility, even in the absence of the first possibility, the non- uniform decomposition rates of the precursors due to wide difference in their decomposition temperatures (urea = 130 - 250°C, gallium nitrate = 110°C, zinc nitrate = 300 - 550°C) is prone to generation of volatile materials or segregation or poor interaction of the materials in

35 flame[29]. More so, in addition to dampening the flame temperature and inhibiting adequate gas-phase mixing, the water in the system can hydrolyze the precursors to yield products of different decomposition temperatures [35]. The occurrence of one or both of these possibilities is less likely to yield a homogeneous or single-phase product in FSP[29].

A quick look at the pronounced peak of the ZnGa2O4 phase (311) of the spectrum at 2휃 =

36°, the peak shows a sharp tip but broadens towards the base. This is an indication of bimodal particle size distribution. This heterogeneity is a consequence of the possibilities already discussed and is very common with nitrates[37], [90]. In addition to the aforementioned factors, another probable contribution to the heterogeneity of the synthesized nanomaterial is poor mixing of the precursors due to their limited solubilities in the screened solvent. Even though there was no observable precipitation as in the case of GZN2, but it is also possible that the varying degrees of solubilities in the solvent mixture could have inhibited perfect mixing. By this, the materials might have undergone different evaporation rates and particle formation routes, the precursors that adequately dissolved will likely follow gas-to-particle route while the precursor with less solubility will go through the droplet-to-particles route due to non- uniform evaporation thereby resulting in heterogeneous particle size distribution[35]. Overall, insufficient combustion energy might have led to inhomogeneous and low crystalline nanomaterials.

Comparing GZN1 with GZN2, unlike the case of GZN1 (with methanol as solvent), using a mixture of H2O + ethanol (GZN2), a solution of similar precursor composition dispersed at the same pressure drop as GZN1, produced relatively crystalline nanomaterial. While we might not neglect the possible effect of the equivalence ratio, but we shall subsequently see that various crystalline materials were synthesized at comparable equivalence ratio of GZN1 utilizing a different solvent as well. Therefore, it is arguable that solvent effect could have contributed more to the synthesis of crystalline material of GZN2. Despite the presence of H2O,

36 it is assumed that in addition to the somewhat higher boiling point (78°C), the enthalpy of combustion of ethanol (1360 kJ/mol) provided higher energy than methanol (B.pt = 63°C and combustion enthalpy = 763 kJ/mol).

Furthermore, the poor volatility and low combustion enthalpy of the majority of the precursors

푇 coupled with the ratio 푏푝 << 1 might have outweighed any anticipated positive effect due 푇푑⁄푚푝 to enthalpy of combustion of ethanol[31].

Fig 7: XRD pattern of GZN2 Fig. 8 shows that it is possible to suppress the unfavorable effects of the precursors by choice of solvent with better characteristics such as high boiling point and/or high enthalpy of combustion[31]. Exploiting the high combustion enthalpy of ethanol (1360 kJ/mol) compared to methanol (726.56 kJ/mol), highly crystalline ZnO nanoparticles (Z1) was synthesized at pressure drop of 1 bar and equivalence ratio of 0.65. Like claimed earlier in this section, the failure of GZN1 to yield crystalline material is likely not attributable to the synthesis condition.

A mixture of methanol-acetonitrile was used to synthesize highly crystalline and pure ZnO

37 nanoparticles (Z2 and Z3) at pressured drop of 1 bar and equivalence ratios of 0.65 and 0.55, respectively. Since acetonitrile has a higher combustion energy (1256 kJ/mol) than methanol

(763 kJ/mol), it was added to boost the energy of combustion of the solution. The fact that Z3 was synthesized at conditions comparatively similar with those of GZN1 suggests that solvent properties and not necessarily synthesis conditions define particles characteristics[35]. Also, by augmenting methanol with a second solvent of better enthalpy of combustion, crystalline materials were made unlike the case where methanol only was used. The use of solvents of even low boiling points but high enthalpy of combustion to override unfavorable properties of precursors is a common strategy in FSP [31], [90]. The lattice parameters and crystallite sizes of the synthesized ZnO nanoparticles are presented in table 3 and table 4, respectively.

Fig. 8: Diffraction pattern of as-prepared Zinc Oxide.

Table 3: Lattice parameters of the synthesized ZnO Sample a (nm) c (nm) Z1 0.3226 0.5220 Z2 0.3224 0.5215 Z3 0.3221 0.5203 Table 3 above contains the lattice parameters of the ZnO oxide samples which were calculated using peaks (102) and (110). The average lattice parameter of the synthesized ZnO is: a (nm)

= 0.3224 ± 0.00017, c (nm) = 0.5213 ± 0.00052. Both the individual and average lattice

38 parameters are comparable to that of a typical hexagonal wurtzite ZnO (PDF No. 36-1451) which is 0.32498 nm and 0.52066 nm for the a-axis and c-axis respectively[10].

Table 4: Estimated crystallite sizes of ZnO samples Sample crystallite size (nm) Z1 11 ± 0.109 Z2 9 ± 0.533 Z3 9 ± 0.392

퐾 × 휆 The Scherrer’s equation, 퐷 (푛푚) = , where K is the Scherrer constant = 0.9, 휆 is the 훽 ×cos 휃 wavelength of the x-ray source = 0.15406 nm, 훽 is the FWHM in radian nm and 휃 is the peak position (radian) was used to calculate the crystal sizes using peaks (100), (102) and (110).

From the XRD spectra in Fig. 8, There are no observable differences in the peaks of Z2 and Z3 which were synthesized using the same solvent (methanol and acetonitrile) despite a slight variation in the synthesis condition, specifically the equivalence ratio (0.65 and 0.55 respectively) which agrees with the equal crystallites sizes as shown in table 4. Comparing the

Z1 and Z2, Z3, the peaks of Z1 are sharper than those of Z2 and Z3 which is indicative of higher crystallinity which corresponds with Z1 having the largest crystallite size out of the three samples. More so, comparing Z1 and Z3 which were synthesized at the same condition, the peaks of Z3 are broader than those of Z1 which reflected in the crystallite size of Z1 being bigger. The difference in the crystallinity and crystallite sizes could be attributed to differences in the enthalpy of combustion of the solvents since the boiling point of ethanol (78℃) is almost the same with that acetonitrile (81℃). The enthalpy of combustion of ethanol (1360 kJ/mol) is higher than those of methanol (763 kJ/mol) and acetonitrile (1256 kJ/mol) respectively. By implication, the nanoparticles of Z1 were exposed to higher temperature regime than those Z2 and Z3 which favored sintering and yielded larger crystallite size. This observation lends support to the assertion that precursor-solvent effect dominates over synthesis condition in defining particles morphology[35].

From the solubility tests, out of the three precursors, gallium nitrate was visually observed to be the least soluble and easily precipitated in all the tested solvents. As earlier discussed, inhomogeneity of precursor solution could result in different evaporation and decomposition rates which affect product formation. To evaluate possible effect of the poor solubility of gallium nitrate, ZN samples were synthesized under the conditions stated in table 2. Fig. 9 shows the XRD spectra of ZN1, ZN2 and ZN5.

Fig. 9. XRD patterns of samples made from urea and zinc nitrate solution only. The numbers after the letters stand for U/Zn molar ratio

Table 5: Lattice parameters of ZN samples Sample a (nm) c (nm)

ZN1 0.3219 0.5219 ZN2 0.3217 0.5203 ZN5 0.3220 0.5200

Table 6: Estimated crystallite sizes of ZN samples Sample Crystallite size (nm) ZN1 11 ± 0.109

ZN2 9 ± 0.533 ZN5 9 ± 0.392

From Fig. 9, all the samples synthesized using zinc nitrate and urea only are highly crystalline materials with all the peaks indexed to the hexagonal wurtzite structure of ZnO. The absence of any other peaks except those indexed to ZnO shows that the synthesized samples are pure and single-phase[97]. From table 5, the lattice parameters of the synthesized Zn samples are presented. Comparing these unit cells dimensions with the average unit cell dimensions of the pristine ZnO shown in table 3 ( a (nm) = 0.3224 ± 0.0017, c (nm) = 0.5213 ± 0.0052), ZN1 gives a lattice contraction of 0.16 % (a-axis) and while the c-axis slightly increased by 0.11%

, ZN2 gives 0.21 % (a-axis), 0.19% c-axis) while ZN5 gives 0.12 % (a-axis) and 0.25 % (c- axis). Regardless of the minor differences among the ZN samples, all their unit cell dimensions lie within the error margin of that of ZnO which indicates that there are essentially the same with the lattice parameters of ZnO. This could mean that no nitrogen was incorporated or that if incorporated, it is not in the lattice of ZnO or even if present, it is not in substantial amount to cause any significant change in the lattice parameters of the ZnO. Gopinath and Mapa[98] using combustion synthesis claimed to have prepared nitrogen-doped ZnO with urea and zinc nitrate precursors. The authors claimed that with urea/zinc molar ratio =1, the sample showed

ZnO lattice contraction from 0.3250 to 0.3232 nm (0.54 %) in the a-axis and 0.5205 to 0.51984 nm (0.5 % ) in the c-axis while sample made with urea/zinc molar ratio = 5 yielded 0.24 % decrease in both axes. The authors ascribed the lattice contraction to 15 % bulk nitrogen concentration incorporated in the sample with urea/zinc molar ratio = 1 and 9 % in the urea/zinc

= 5. Their results show that the amount of nitrogen incorporated decreased as the urea content increased. According to the authors, the increase in carbonate species due to urea decomposition as the urea amount increased led to segregation which possibly hindered successful incorporation of nitrogen in the lattice of ZnO.

However, to validate combustion synthesis as a simple technique for doping ZnO with nitrogen,

Söllradl et.al[98] repeated the work of Gopinath and Mapa[98] using urea/zinc molar ratio =

1 which gave the highest dopant concentration as claimed by Gopinath and Mapa. The authors reported that although nitrogen was present in the sample (urea/zinc = 1) but only a maximum of 0.48 % bulk nitrogen concentration was achievable and also reported that the lattice parameter of the sample was essentially the same with that of the pristine ZnO. From this result, it could be possible that nitrogen was incorporated in the ZN samples of this present work but in small amount.

Also, inspecting the peaks shows that the FWHM increased as the ratio of urea to zinc increased which reflected in the decrease in the crystallite sizes as shown in table 6. All samples were made with methanol-acetonitrile solvent and synthesis conditions are presented in tables

1 and 2. Since the same solvent was used, any observed characteristic differences can only be described in terms of other parameters such as synthesis conditions and ratio of precursors. A quick observation of the peaks implies a slight decrease in crystallinity as the ratio of urea to zinc amount increased. ZN1 and ZN2 were prepared at the same equivalence ratio of 0.81 but pressure drops of 1.5 bar and 1 bar, respectively. While the ZN1 has an estimated crystallite size of 11 ± 0.109 nm, ZN2 and ZN5 have equal crystallite sizes of 9 ± 0.533 nm and 9 ± 0.392 nm, respectively. This suggests that the presence of impurities might inhibit crystal growth and decrease crystallinity. Even though higher dispersion pressure could mean finer spray[39] but in this case, it cannot be said that the effect of pressure was observed or perhaps, obvious.

Perhaps, similarity in equivalence ratio and precursor-solvent effects had more influence.

From the preceding discussions, in addition to the need for solvent of high combustion enthalpy, it is also very likely that the poor solubility of gallium nitrate could inhibit the formation of single-phase, homogeneous solid solution. Based on this, the amount of gallium precursor was significantly reduced by dropping the molarity to 0.03 M in the subsequent synthesis of gallium zinc oxynitride discussed below. Due to the solubility issues, DI H2O +

42 butanol and methanol + butanol systems were adopted for synthesis and at limited gallium content to suppress precipitation.

Fig. 10. XRD patterns of gallium zinc oxynitride solid solutions made with methanol +

butanol as solvents compared with pristine ZnO (GZN 3 (top) and GZN4 (middle) to Z3

Sample a (nm) c (nm) (bottom) GZN3 0.3197 0.5189 GZN4 0.3218 0.5219 Table 7: Lattice parameters of GZN samples GZN5 0.3212 0.5213 GZN6 0.3199 0.5191

Sample Crystallite sizes (nm) GZN3 7 ± 1.243 GZN4 7 ± 0.495 Table 8: Estimated crystallite sizes of GZN samples GZN5 11 ± 1.08 GZN6 9 ± 0.798

Fig. 10 presents the diffraction pattern of GZN3 and GZN4 compared with that of pristine ZnO

(bottom). The lattice parameters and the crystallite sizes are shown in table 6 and table 7, respectively. From the XRD spectra, all the peaks are indexed to hexagonal wurtzite structure of ZnO. The non-existence of any other extra peak on the XRD suggests that the as-prepared samples are pure and single-phase metal oxides[75]. This observation could be attributed to

43 two major possibilities. It is either the gallium is highly dispersed such that it is below the detectable limit of the XRD or are incorporated into the crystal lattice of the host as dopants[8].

However, as shown by the insert, the peaks of the GZN samples slightly shifted to higher angles of 2휃. From table 6, the shifts correspond with the contraction of the lattice of the synthesized

ZnO (a (nm) = 0.3224 ± 0.0017, c (nm) = 0.5213 ± 0.0052). For GZN3, the a-axis contracted to 0.3197 nm (0.84 %) and c-axis to 0.5189 (0.46 %) while GZN4, the a-axis contracted to

0.3218 nm (0.19 %) and the c-axis slightly increased to (0.12 %). The result shows that the a- axis is more contracted while the c-axis is less affected. The c-axis has been reported to show less affected by changes in the ZnO crystal lattice[99], [100]. The lattice parameters are still within the calculated error margin of the lattice parameters of the pristine ZnO and therefore means, that there is no significant change in the unit cell dimensions. However, it is unlikely that all the gallium atoms completely vaporized from the flame given the very short residence time in FSP (milliseconds) and again, that peaks related to gallium were observed in GNZ2.

More so, the contractions observed in GZN3 and GZN4 are also somewhat higher than those of ZN samples. The contraction of the ZnO is due to the differences in the ionic radii of Ga3+

(0.047 nm )) and Zn2+ (0.074 nm)[95], [97], [101]. This result suggests that the gallium and nitrogen are possibly in the crystal lattice of ZnO as dopants and that the metal oxide are not mere mixtures of GaN and ZnO[75], [101]. Gallium zinc oxynitride solid solutions prepared via nitridation at 850℃ showed lattice contraction by 1.67 % in the a-axis and 0.26 % in c-axis and was reported to contain 91 % GaN and 9 % ZnO while another sample that showed lattice contraction of 1.5 % (a-axis) and 0.26 % in the c-axis contained 77% GaN and 23 % ZnO. The reported bandgaps are 2.7 and 2.64 eV respectively – which are substantially less than 3 eV

[102]. This result implies that lattice contraction is strongly dependent on the amount of GaN incorporated and that bandgaps were strongly tied to the zinc content as discussed below. As previously discussed, all lattice contractions for the GZN samples are less than 1% and being

44 slight contractions, suggests that dopants if present in the lattice, would be of small concentrations and that GZN3 possibly has more dopant concentration than GZN4 as indicated by the extent of contractions. Also, another observation is that the sample (GZN3) with the smaller urea to total metal ratio (2.5) in the precursor solution had a relatively more contraction than GZN4 which urea to total metal ratio of 5. Similar observation was made in the case of

ZN samples where contraction decreased as the urea to metal ratio increased. The same trend has been reported in the synthesis of nitrogen-doped ZnO and was attributed to the increased in the formation of carbonate species which promoted segregation and inhibited doping[103] or the increase in temperature as due to exothermicity of urea combustion possibly led to less incorporation of nitrogen[104] and fast decomposition of urea relative to other precursors[105].

However, the work by Kennedy and Meekins[75] claims that it is Zn/Ga ratio not the urea to total metal content that had higher role in the contraction of the lattice. Currently, it is not clear whether it is nitrogen or gallium content that is playing the significant role in the lattice contraction or a combination of them but, it might be the presence of the dopants that caused the minor lattice contractions.

A look at the peaks shows that the intensities of the peaks of undoped ZnO was slightly higher than those samples of GZN with comparable urea to total metal ratio ( ZN1 versus GZN3 and

ZN5 vs GZN4) which suggests that crystallinity might decrease due to the presence of foreign materials in GZN sample[97]. Table 8 shows the crystallite sizes of the GZN samples. GZN3 and GZN4 have crystallite sizes of 7 ± 1.243 nm and 7 ± 0.496 nm, respectively and are smaller than the average crystallite size of the ZnO. The similarity in crystallite size of GZN3 and

GZN4 might be due to the fact both were made using the same solvent. [75], [97], [106] .

Fig. 11. XRD patterns of gallium zinc oxynitride solid solutions made with DI

H2O + butanol compared with ZnO (GZN5 (middle), GZN6 (top) and Z3 (bottom).

From Fig. 11, the XRD spectra peaks of GZN5 and GZN6 are indexed to the hexagonal wurtzite structure of zinc oxide. The prepared samples are single-phase oxides of high crystallinity and purity as implied by the absence of extra peaks[107]. As shown in table 6, sample GZN6 with smaller urea to metal ratio (3.57) in the precursor solution showed more lattice contraction than

GZN5 (6.25). For GZN6, when compared with the pristine ZnO, the a-axis contracted from

0.3224 nm to 0.3212 nm (0.78 %) and the c-axis from0.5224 to 0.5191 nm (0.42 %) while that of GZN5 contracted to 0.3212 nm (0.37 %) with no change in the c-axis. Also earlier discussed, the more contraction shown by GZN6 suggests than it might contain more dopants than GZN5.

The more contraction obtained with smaller urea to total metal content was also observed in the case of GZN3 versus GZN4 as well regardless of solvent and synthesis conditions. As earlier explained in the case of ZN samples above, it is possible that as urea increased, more decomposition products such as carbonate increased which promoted segregation and perhaps, inhibited doping[103].

The peaks broadened as the urea to total metal content increased with the crystallite sizes GZN5 and GZN6 as 11 ± 1.077 nm and 9 ± 0.798 nm respectively as shown in table 8. Although, urea

46 is very combustible fuel that can increase the temperature of the flame and lead to higher crystallinity of samples[75]. But, generally, from the standpoint of equivalence ratio, it could mean that there is excess fuel in the flame caused by the urea which contributes to the quenching of the flame or dampening the temperature of the flame, hence, the reduction in crystallinity. Another interesting observation is that samples involving water (GZN5and

GZN6) are more crystalline and of larger crystallite sizes than those made without water

(GZN3 and GZN4) and might be because the DI H2O + butanol mixture supplied higher rate of energy (1.17 kJ/s) to the flame than methanol-butanol system (1.12 kJ/s).

In general, comparing all the GZN samples, there is no distinct trend in their physical properties as reflected in the crystallite. At this point, it cannot be said with certainty, the factors that are responsible for the observed particles characteristics inferred from the XRD since the possible effect of the interplay of dopant, solvent effect as well as synthesis conditions are not yet known

4.3 Determination of Bandgaps

Bandgap tuning is the first strategy in the design and development of photocatalysts[49]. The direct bandgaps of the synthesized gallium zinc oxynitride were estimated from the Tauc’s plots shown in Figs. 12 and 13 and summarized in table 3 below.

Fig. 12. Tauc’s plots showing direct bandgaps (a). GZN3 (3.16 eV) and (b). GZN4 (3.07 eV)

Fig. 13. Tauc’s plot showing the direct bandgaps (a). GZN 5 (3.12 eV) and (b). GZN6 (3.08 eV)

Table 9: Bandgaps of GZN samples Sample Bandgap (eV) GZN3 3.16 GZN4 3.07 GZN5 3.12 GZN6 3.08

All estimated bandgaps for the samples as shown in table 3 range from 3.07 to 3.16 eV. The bandgaps of pristine ZnO and GaN are 3.2 eV and 3.4 eV respectively[95]. This shows that all bandgaps are significantly lower than the bandgap of GaN with GZN4 and GZN6 obviously lower than that of ZnO while those of GZN5 and GZN3 are slightly lower than the bandgap of

ZnO. GZN4 with higher urea to total metal ratio than and less lattice contraction (0.19 %) than

GZN3 (0.84 %) has lower bandgap than GZN3 that showed more lattice contraction with smaller urea to total metal ratio while GZN5 with higher urea to total metal ratio and less lattice contraction (0.37 %) shows a higher bandgap than GZN6 (contracted by 0.78 %). While the trend of lattice contraction with higher urea content is the same i.e. more contraction at smaller urea to total metal ratio, it is observed there was no correlation between bandgap and lattice

49 contraction. Also, despite that the GZN6 has Zn/Ga amount twice that of GZN4, their bandgaps are essentially the same - which might be supporting the claims that there are other factors contributing to reduction of bandgaps. The lack of correlation between lattice contraction and total concentration of dopants are reported in literatures. It has been reported that bandgap reduction in gallium zinc oxynitrides do not depend on the total dopant concentration (lattice contraction) but rather depends on the zinc content in the final composition. Yang et al.[102] synthesized gallium zinc oxynitride solutions and their result shows that 0.22 % lattice contraction gave a bandgap of 2.57 eV while the sample with a higher lattice contraction of

0.45 % gave a bandgap of 2.72 eV. The authors stated that although the former sample has less contraction, it has more zinc content than the later and therefore attributed the smaller bandgap to increased repulsion between the N 2p atomic orbital, Zn 3d and O 2p of the ZnO atomic orbital. Maeda et. al[95] earlier observed that the bandgaps of gallium zinc oxynitrides did not depend on the overall amount of dopants incorporated but rather on the Zn content of the final products, with the samples having higher zinc content showing narrower bandgaps. The authors stated that the repulsion between the N 2p and Zn 3d and O 2p orbitals raises the valence band positions towards the conduction band (which are filled by the Ga 3d orbitals) without affecting the conduction band position and that the strength of the repulsion determines the final position of the valence band maximum. The stronger the repulsion, the more the valence band is raised towards the conduction which translates to smaller bandgap. The authors further observed that the p-d repulsion increased as the Zn content increased and claimed that the finding agrees with their DFT result. Similar finding was reported in the use of combustion synthesis for the development of gallium zinc oxynitrides[75]. Using moisture assisted nitridation, Yang et. al[97] prepared a solid solutions of gallium zinc oxynitride with ZnO/GaN = 3 and GaN/ZnO

= 3 in the precursor mixture respectively via nitridation at 850°C. The as-prepared materials were reported to have a band gaps of 2.31 eV and 2.48 eV, respectively. The reduction in

50 bandgaps was attributed to the Zn 3d and N 2p orbitals repulsion which the authors claimed to have increased as the zinc concentration increased in the final product increased. From this result and other preceding discussions and also considering that ZnGa2O4 has a bandgap of 4.4-

5 eV[108], it could be inferred that nitrogen might be playing greater role in bandgap reduction than the gallium content. Based on the presence of peaks associated with Ga in GZN2, it is also likely that Ga could be incorporated in GZN3 to GZN6, but the amount of nitrogen incorporated might be low to achieve a significant bandgap reduction. Correlating bandgaps with the only possible concentration of dopants incorporated into the lattice of the host as commonly done could be misleading. In tracing the origin of bandgap narrowing, the computational works, specifically DFT show that in addition to the strength of repulsion between the N 2p and Zn

3d orbitals, the zinc and oxygen atoms positions in the band structure could also affect bandgap

[95], [109]. From the foregoing, it means that to fully characterize and understand the origin of bandgap differences, experimental works could be complemented by computational studies.

Computational works could assist to provide insights on the formation and positions of donor and acceptor levels in the band structure which affects the Zn 3d and N 2p interaction[109],

[110].

Zn/Ga ≤ 2 is a common ratio that has been reportedly used to achieve bandgap of less than 3 eV. In this present work, it was not possible to achieve single phase solid solution with Zn/Ga

= 1 in the condition tested. Attempts to increase the gallium amount in the precursor ratio was limited by solubility issues already discussed while decreasing the zinc amount in the precursor solution impacted on the particles collection efficiency that it was not reasonable to run at this condition. However, even with similar ratio, different bandgaps are obtained for different precursors and synthesis routes and conditions [4], [63], [75], [111]. While it was possible to synthesize single-phase solid solutions of gallium zinc oxynitride with potentials for visible light region activity using Ga2O3 and ZnO precursors, it was not possible to achieve a single-

51 phase and visible-light responsive bandgap with GaN and metallic Zn powder with the similar ratio as the former regardless of synthesis conditions tested. These observed differences have been attributed to atomic arrangement and interaction. Lack of knowledge of the detailed formation mechanism limits the of understanding the exact origin of bandgap in gallium zinc oxynitride solid solution[49].

Based on the estimated bandgap, the synthesized gallium zinc oxynitrides have potentials for photocatalytic applications in the UV region. In general, the use FSP for rational design of photocatalysts remain a very challenging tasks as it has been difficult to understand how the interplay of its multiple parameters influence not just the physico-structural composition but also the electronic properties of the final products[57].

CHAPTER FIVE: CONCLUSION AND RECOMMENDATION

5.1 Conclusion

• Gallium zinc oxynitrides of two different precursor compositions and solvent mixtures

have been prepared via liquid-fed flame spray pyrolysis.

• Precursors and solvent of weak combustion enthalpy content resulted in the formation of

amorphous and multiphase oxides with bimodal particle sizes.

• Poor mixing of precursor could inhibit the formation of single-phase solid solutions.

• Solvent and precursors properties outweighs synthesis conditions. Solvents of high

enthalpy was used to override the poor combustion enthalpy of nitrates precursors and

facilitated the formation of homogeneous phase solid solutions.

• There was no correlation between bandgap and lattice contraction.

• Based on bandgap measurements, the as-synthesized gallium zinc oxynitrides have

potentials for photocatalytic application in the UV region.

• Although, narrower bandgap agreed with more lattice contraction in the case of DI H2O

+ butanol solvent, such correlation was not observed in the case of methanol-butanol

solvent. This probably supports other computational and experimental works that suggest

that bandgap is not just determined by dopant content.

• To fully characterize flame-made photocatalysts, the precursor-solvent effects, synthesis

conditions and the interplay of both must be accounted for. This requires very detailed

experimentations and characterization.

5.2 Recommendation

Organometallic based precursors or liquid-based precursors and solvents with good combustion enthalpies should be used while paying attention to compatibility and solubility. More solvents screening could be done. To narrow down the bandgap into the visible region, it is critical to increase the dopant content in the crystal lattice of the ZnO.

CHAPTER SIX: FUTURE WORK

The next phase of this research is to extend the knowledge gained from this present work to the synthesis of ceria-based solid solutions catalysts for various energy and environmental applications. For this work, similar FSP set-up has been designed and built with slight modifications in the nozzle design. The adjustment of the annular gap of the nozzle will be solely done using an in-built needle system.

In the light of the rising climate change and health effects, there is a concerted global effort to significantly cut down greenhouse emissions gases (GHG) being released to the environment[112]. The 2018 report on California Climate’s policies shows that of all

GHG sources, automobiles are the largest contributor[113]. To halt this trend, strict environmental regulations have been put in place and more stringent regulations are anticipated in the nearest future. For instance, California environmental stipulation aims to drop GHG emissions to 40% below 1990 level by 2030[113] while working towards raising the number of zero emissions vehicles to 1.5 million by 2025

(https://ww2.arb.ca.gov/). Independently, the European Union (EU) member states have set similar targets as California to cut GHG emissions by 40% below the 1990 level by

2030 (http://data.europa.eu/eli/reg/2018/842/oj).

To meet this goal while transitioning to sustainable transport systems, calls for invention of new technologies or modification of existing ones. This would not just involve improving engine designs but also a search for high performance automobile catalytic converter technologies (three-way catalysts). During operations, the automobile exhaust engines switches between fuel rich and lean regimes. To minimize the release of emissions without introducing extra cost in terms of incorporating technologies for fuel- air ratio control during this cycle requires that the three-way converter have the capability to adjust its oxygen content accordingly[112]. Ceria and ceria-based catalysts that have

55 been noted to show the above desired property coupled with its excellent activity in the degradation of automobile emissions and currently serves as automobile catalytic converter[112], [114], [115]. However, during these fuel regimes, the operating temperatures ranges from 950-1100℃. At this high temperature conditions, pure ceria and ceria-supported materials undergo sintering leading to drastic reduction in their catalytic performance. This limits their candidatures as self-regulating oxygen storage capacity catalysts for three-way catalysis (emissions degradation) in automobiles in order to meet current and emerging environmental regulations[112][115].

Doping pristine ceria with some other metal oxides to form homogeneous stable solid solutions is seen as one of the best techno-economic strategies towards improving the activity and thermal/hydrothermal stability of ceria-based catalysts-so as to be used as

TWCs[112], [114]–[120].

A significant part of the efforts to achieve the above objectives has been largely driven by the classical wet chemistry techniques. These synthetic methods are usually multi- step batch processes, less amenable to automation and waste prone. More so, these methods have little or no control over their products’ properties and their as-prepared solid solutions are thermally unstable and segregate when exposed to high temperatures like those obtainable in the engines of vehicles during operations[118]. These challenges call for the ongoing search for more viable and efficient synthesis approaches.

6.1 Why Flame Spray Pyrolysis for the Synthesis Ceria-based Solid Solution

Catalysts

In addition to the Hume-Rothery rules, thermodynamic considerations suggest high temperature requirement for the synthesis of homogeneously phase-stable solid solution.

The thermodynamic-based phase-stability prediction is done by utilizing the Gibbs free energy of mixing-which provides a correlation between the heat of mixing and entropy

56 of mixing. Out of the three main entropic contributions (electronic, vibrational and configurational) to solid solution, vibrational and configurational entropies are usually the most important[121]. However, configurational entropy dominates over vibrational contribution[122], [123]. Although the definition for configurational entropy (entropy due to mixing) is valid for only ideal or regular solid solutions (random atomic configurations). In ideal and regular solutions, the atoms are randomly distributed and the regular configurational entropy models are applied to them[123]. Although still a subject of ongoing debate, Yeh J. et l.[121] claims that atoms in solid solutions are randomly distributed so that the ideal configurational entropy model applies. Lee, T. et al.[124] reports that mixing in the solid solution of ceria zirconia was not ideal but also agrees that the ideal or regular configurational entropy model of the can be used the entropic analysis. Therefore, the Boltzmann’s equation is used to estimate the total entropic contribution for this analysis. Conventionally, positive heat of mixing indicates unstable solid i.e. tendency to demix while negative heat of mixing suggests phase stability. The Gibbs free energy has similar connotations[121], [125].

The first stage of this work is to dope ceria with zirconia and build upon the knowledge to extrapolate it to other oxides. To demonstrate the ideal temperature requirement for the synthesis of a typical homogenous phase-stable solid solution, the compositionally maximized configurational entropy (equimolar) is chosen for a binary solid solution of ceria zirconia. Due to dearth of calorimetric data at desired composition, estimates of the heats of mixing and formation were obtained by comparing information from theoretical[126] and experimental[124] works. Utilizing the approximated free Gibbs energy of mixing equation, it is shown that more than 1600 K is needed to obtain at least, a thermodynamically stable ceria zirconia solid solution (negative Gibbs free energy).

Furthermore, since alloy is a typical solid solution, some of the design criteria could give

57 more insights into the temperature requirement for the synthesis of the catalytic solid solution. To ensure a single-phase formation in alloy, the minimum annealing temperature (sometimes called critical temperature) is often conventionally fixed as a fraction of the average of the melting points (Tm) of the constituents. Following experimental instances cited by Wu et al, a typical minimum critical temperature is ca. half the Tm[127]. By extension, based on their oxides, the minimum critical temperature for ceria zirconia solid solution is at least 1400 K. This is comparable to the value obtained from the thermodynamic calculations as presented in Appendix C.

More so, it has been reported that high temperature could be employed to increase the allowable amount of dopant into the lattice of ceria from 15% to 40% by increasing the temperature from 1273 K to 1773 K[128]. This evidence further supports that flame spray pyrolysis could be more viable for the synthesis of stable ceria-based solid solution. In fact, at lower entropic contributions (compositions away from equimolar), higher temperature is needed. And entropic contributions could be maximized by increasing the number of constituent oxides[125].

From the above analyses, it is difficult for a traditional wet chemistry synthetic method to provide such a high temperature required for thermodynamically phase-stable solid solutions in one step[129]–[135]. Interestingly, LF-FSP as a one-step technique conveniently provides even more than such a high temperature requirement. It is expected that catalysts made from such harsh temperature will withstand the high temperature regimes of automobile engines (900 -1100°C). It has been reported that LF-

FSP-made ceria-zirconia solid solutions withstand thermal sintering compared to co- precipitated samples[27], [41]. Also, Sutorik and Baliat[136] reported that FSP-made ceria zirconia did not show any structural change when calcined up to 1100°C but increased in particle size due to sintering.

6.2 Literature Review on LF-FSP-made Ceria Zirconia Solid Solution

A range of compositionally homogeneous high surface area LF-FSP-made ceria zirconia exhibit higher degree of resistance to sintering but lower oxygen storage capacity (OSC) in the presence of CO and H2 compared to co-precipitated samples. The low OSC of the as-prepared flame-made materials is attributed to the high degree of crystallinity and reduced defects concentration due to the high temperature synthesis environment[27]. Pratsinis group did not only highlight the OSC or catalytic ability of the LF-FSP-made ceria zirconia but also reported that it shows well-defined crystalline property and conserved ca. 60 % of its surface area after calcination[29]. Similarly, Stark et al. shows that the choice of solvent is critical to obtaining a single-phase ceria zirconia solid solution but did not provide any information on the reducibility or catalytic performance of the as-prepared materials[45]. Sutorik and Baliat[136] and Kim et.al[88] respectively were able to synthesize homogeneous (over certain compositions), sintering-resistant but low surface area ceria zirconia solid solutions.

Additionally, none of the two works tested for the OSC or any catalytic applications.

Furthermore, the Laine group[137] utilizing high throughput experimentation, presents LFSP- made ceria zirconia as candidates for both C3 Hydrocarbons oxidation and NOx reduction.

Unfortunately, the performance of the as-prepared catalysts at temperatures typical of the automobile exhaust was not reported. Recently, Zhang et.al[138] synthesized a series of metastable ceria zirconia solid solutions that show better performance both in terms of activity

(CO oxidation) due to increased oxygen vacancies and sintering resistance than the co- precipitated samples. While this recent work agrees with other earlier works on the superior sintering-resistant ability of LF-FSP-made ceria zirconia, its claim of better catalytic activity than co-precipitated samples is in opposite direction. This interesting discrepancy suggests at least two things: (1) there is still an opportunity for optimization of the LF-FSP for pristine

59 ceria zirconia solution (2) a more systematic investigation of the LFSP with integration of the synthesis conditions for better characterization and comparisons.

More so, the catalytic tests of the potentials of the LF-FSP-made ceria zirconia for TWC applications have not been properly reported-or at least, in explicit catalyst performance terms such as conversion or turnover frequency. This has made it difficult to compare the works of the different groups. Another point is the absence of reports on the performance of the ceria zirconia in the presence of the three pollutants. Most importantly, oxidation of hydrocarbons usually give rise to water vapor. Therefore, a TWC must not just be effective in degrading the pollutants but also maintain its high activity in the presence of water vapor in addition to high temperature[139],[112]. Hence, the use of resistance to thermal sintering only as a measure of phase stability of LF-FSP-made ceria zirconia as seen in the previous works is not a true representation of the real conditions. Hydrothermal stability of LF-FSP-made ceria zirconia is a crucial aspect of this research. This has become more important as ceria zirconia finds increasing applications involving different harsh environments.

Furthermore, despite the available works on the deployment of LF-FSP to the synthesis of ceria zirconia solid solution, no attempt has been made to correlate the effects of particle or crystallite sizes to the phase stability of the as-prepared catalysts. However, particle or crystallite sizes play important roles in defining phase boundary, phase stabilization and catalytic properties of ceria zirconia solid solution and zirconia[140]. Similar experimental evidences regarding such roles of nanoscale particle sizes in solid solutions have been confirmed by other researchers[140]–[142]. This work will also investigate the influence of particles sizes on phase formation, phase stability and catalytic abilities of LF-FSP-made solid solution.

6.3 Preliminary Data on the Effects of Synthesis Parameters on Flame Characteristics

In addition to precursor-solvent effect, the flame conditions contribute to determining the overall properties of FSP-made materials. Flame characteristics such as geometry and temperature are majorly defined by precursor flowrates, type of dispersion gas and flowrates and pressure drop[143]. Usually, the precursor and dispersion gas feed rates are lumped into the single parameter -equivalence ratio.

The effects of precursor rate, dispersion gas (oxygen) rate and pressure drops on flame geometry were investigated and discussed herein.

For this proposed work, the precursors to be used are Cerium(III) 2-ethylhexanoate, 49% in 2- ethylhexanoic acid, Ce 12%, zirconium(IV) oxide 2-ethylhexanoate, in mineral spirits (≈6%

Zr), Manganese(II) 2-ethylhexanoate, 40% w/w in mineral spirits, 6% Mn. Ethyl hexanoic acid

(EHA) is a common solvent for all the precursors and the best solvent candidate based on its high boiling point and enthalpy of combustion. However, due to its high viscosity, a compatible secondary solvent is used to added to lower the viscosity of EHA. Therefore, the solution will be prepared such that the total ratio of EHA: toluene in the precursor solution is 1:1. Hence, the flame tests were carried out using EHA: toluene (1:1) mixture.

It is also important to note that the equivalence ratio (stoichiometric) of the premixed flame and the shroud air flowrate (1 L/min) were held constant. Therefore, the equivalence ratio under investigation is that of the precursor or center flame where the solvent is the fuel and dispersion oxygen gas is the oxidant.

6.4 Effect of Equivalence Ratio Flowrate on the Flame

Since equivalence ratio depends on precursor and dispersion gas flowrates, to probe the effect of the equivalence ratio, one of the flowrates and pressure drop were fixed while the flowrate of interest was varied.

Prior to evaluation of other parameters, attempt was made to ascertain the dispersion gas pressure drop to operate at. Pressure drop can be tuned either by changing the dispersion gas flowrate at constant annular gap area of the nozzle or changing the annular gas (by adjusting the needle) while keeping the dispersion gas constant. In this case, the annular gap area was kept constant while flowrates were varied. Based on visual observation of the flame behavior and stability, the operating pressure drop was fixed 1.5 bar.

(a). Precursor flowrate versus flame properties

Fig.14 (a and b) displayed below show the variation of the flame height with precursor flowrate at constant dispersion gas (oxygen) flowrate and equivalence ratio. The flame heights were generated from pictures of the precursor flames beginning from the tip of the nozzle to the end of the combustion luminous region (appendix C, Fig. C1). The precursor flowrates (ml/min) used are 2, 3, 4 and 5 respectively at constant 6 L/min Oxygen flowrate. The corresponding equivalence ratios were computed and correlated with the flame heights as shown in fig.14 (b).

From the plots, the flame height varied linearly with precursor flame and the equivalence ratio.

As the precursor flowrate increases, the total concentration of fuel droplets[144] and combustion enthalpy in the flame are increased. This in turn increases both the temperature and the length of time it takes to complete burn off the fuel which translates to higher flame height and vice versa[37], [143], [144].

Fig. 14. Variation of flame height with (a) precursor flowrate (right) (b) equivalence ratio (left) at constant dispersion gas rate and pressure drop

Fig. 15. Flame height versus equivalence ratio and precursor feed rate [29]. This trend was reported by the authors in the use of FSP to synthesize ZnO and agrees with other literatures including the findings in this present work. As the height of the flame increases, the residence time spent by the nanoparticles increases as it travels a longer distance to escape the combustion zone. By this, the particles are exposed to high temperature for long duration (however, in milliseconds) which affects the physical characteristics such morphology, increase in particle size, surface area and homogeneity of the final product[28], [144].

Fig. 16. The influence of precursor rate on the particle size of FSP-made Ceria nanoparticles. This shows that as precursor flowrate increases, the average particle size increases due to sintering and coagulation effects consequent of long exposure to high

temperature. The open symbols are the bimodal particles size[28].

More so, the effect of precursor rate is very important as it is the major parameter for controlling the rate of production during the synthesis. It has been reported that BET diameter of particles by FSP (which is correlated with surface area) varies linearly with production rate[143].

(b). Dispersion gas (pure Oxygen) rate versus flame Properties

The effect of the dispersion gas rate on the flame characteristics and possible qualitative inferences on products characteristics was studied at constant pressure drop of 1.5 bar and precursor rates. The O2 flowrates (L/min) used were 5, 6 and 7 respectively at constant precursor rate of 2 mL/min. Similar experiment was repeated at constant precursor rate of 3 mL/min. The heights of flame were measured using the same approach as earlier mentioned above. The pictures of the flame are found in fig. C2 of appendix C. In all cases, it was found that the flame height decreased as O2 increased as shown by fig. 16. As O2 flowrate increases, the rate of mixing within the flame[145] increases so that the evaporation and combustion of

64 fuel droplets happen very fast leading to decrease in flame height[39], [143]. The reduction in flame height implies that the particles have shorter residence time in the high temperature combustion zone which suppresses sintering and coagulation and leads to reduction in particle sizes as oxygen flowrate increases[28].

Fig. 17. Variation of flame height with oxygen flowrate at constant pressure drop and precursor rate. Furthermore, in the synthesis of silica using FSP, Madler et al. noted that as oxidant flowrate increased, the temperature of the flame increased and then continuously decreased as the oxidant flowrate was increased. Similar trend was also observed for the specific surface area of the Silica. The work claims that for the first phase, the initial increase in the oxidant gas flowrate resulted in increased velocity and dispersion of the droplets such that their residence time in the high temperature combustion environment was reduced. More so, as more and more oxygen were introduced into the flame to be heated by the same amount of fuel, the energy and temperature of the flame dropped. The combined effect of these factors inhibited sintering and coagulation [39]. This drop in energy could be likened to quenching of the flame. In the second phase characterized by the drop in specific surface area of the silica, as oxygen rate was increased, the fuel combustion was highly accelerated resulting in high temperature-although for a shorter residence. The high temperature occasioned by the accelerated combustion gave birth to the reduced SSA[39].

Generally, considering that oxygen and precursor flowrates are coupled into the equivalence ratio, the effect of one can be contradicted or complemented by the other[143].

(c). Pressure drop effect versus flame properties

The effect of pressure drop on the flame characteristics was probed by keeping both the precursor and oxygen flowrate constant while varying the pressure drop. The variation of pressure drop was achieved by adjusting the annular gap area of the nozzle using the in-built needle system.

Fig. 18. Visualization of pressure drop effect on the flame at constant precursor and oxygen rates. (a) & (b) are at the same precursor rates and (c) & (d) are at the same oxygen rates. The tested pressure drops were 1.5 and 2 bar and two different precursor and oxygen rates respectively as shown above in Fig. 17.

As seen in the Fig. 17, the pairs of (a) & (b) and (c) & (d) are kept at the same equivalence ratio defined by similarity in precursor and oxygen rates. For each comparable pair, it could be visualized that as the pressure drop was increased from 1.5 to 2 bar, the flame height decreased while an opposite trend occurred for the width of the flame. Increase in pressure drop leads to increase in the spray velocity, fuel droplets dispersion efficiency and finer spray (larger SSA).

The formation of finer spray or droplets coupled with increased velocity translates to faster

66 combustion of the droplets and accelerated removal from the combustion zone - thereby starving the flame of fuel droplets and hence, the reduced flame height and vice versa[39],

[143], [146], [147]. It could also be seen that for each, the dense core representing the high temperature combustion zone is enveloped by white-yellow parts which are more visible when closer to the burner. This envelope is an indication of incomplete combustion of droplets[143].

It is more visible towards the end of the flame due to the drastic reduction in flame energy and temperature along the flame height[39]. It also appears that this incomplete combustion is more pronounced at higher pressure. It could be interpreted that due to the high droplet velocity at higher pressure, the droplets are less combusted than at lower pressure. A higher dispersion pressure drop facilitates the quenching rate of the particles and creates metastable structures[41].

Therefore, from the preceding discussions, FSP has various parameters that could be explored for rational tuning of catalysts. In the proposed work, ceria-based solutions will be synthesized at various synthesis parameters, the chemical and physical properties would be evaluated to provide structure-property correlation of the as-synthesized materials.

REFERENCES

[1] H. Ahmad, S. K. Kamarudin, L. J. Minggu, and M. Kassim, “Hydrogen from photocatalytic water splitting process : A review,” Renew. Sustain. Energy Rev., vol. 43, pp. 599–610, 2015. [2] C. Liao, C. Huang, and J. C. S. Wu, “Hydrogen Production from Semiconductor-based Photocatalysis via Water Splitting,” pp. 490–516, 2012. [3] L. S. Al-Mazroai et al., “The photocatalytic reforming of methanol,” Catal. Today, vol. 122, no. 1–2, pp. 46–50, 2007. [4] J. Yang, Y. Pei, S. Kim, G. Choi, A. Vinu, and J. Choy, “Highly Enhanced Photocatalytic Water-Splitting Activity of Gallium Zinc Oxynitride Derived from Flux-Assisted Zn / Ga Layered Double Hydroxides,” Ind. Eng. Chem. Res., vol. 57, pp. 16264–16271, 2018. [5] R. Strobel and S. E. Pratsinis, “Effect of solvent composition on oxide morphology during flame spray pyrolysis of metal nitrates,” pp. 9246–9252, 2011. [6] I. Akkerman, M. Janssen, J. Rocha, and H. Wij, “Photobiological hydrogen production : photochemical e ciency and bioreactor design,” vol. 27, pp. 1195–1208, 2002. [7] A. FUJISHIMA and K. HONDA, “Electrochemical Photolysis of Water at a Semiconductor Electrode,” Nature, vol. 238, no. 5358, pp. 37–38, 1972. [8] K. Maeda et al., “Overall water splitting on (Ga1-xZnx)(N 1-xOx) solid solution photocatalyst: Relationship between physical properties and photocatalytic activity,” J. Phys. Chem. B, vol. 109, no. 43, pp. 20504–20510, 2005. [9] Z. Wang et al., “Moisture-Assisted Preparation of Compact GaN:ZnO Photoanode Toward Efficient Photoelectrochemical Water Oxidation,” Adv. Energy Mater., vol. 6, no. 20, pp. 1–8, 2016. [10] J. H. Yang, Y. R. Pei, S. J. Kim, G. Choi, A. Vinu, and J. H. Choy, “Highly Enhanced Photocatalytic Water-Splitting Activity of Gallium Zinc Oxynitride Derived from Flux-Assisted Zn/Ga Layered Double Hydroxides,” Ind. Eng. Chem. Res., vol. 57, no. 48, pp. 16264–16271, 2018. [11] T. Ohno, L. Bai, T. Hisatomi, K. Maeda, and K. Domen, “Photocatalytic water splitting using modified GaN:ZnO solid solution under visible light: Long-time operation and regeneration of activity,” J. Am. Chem. Soc., vol. 134, no. 19, pp. 8254– 8259, 2012. [12] H. N. Kim, T. W. Kim, I. Y. Kim, and S. J. Hwang, “Cocatalyst-free photocatalysts for efficient visible-light-induced H 2 production: Porous assemblies of CdS quantum dots and layered titanate nanosheets,” Adv. Funct. Mater., vol. 21, no. 16, pp. 3111–3118, 2011. [13] B. Thiébaut, “Flame spray pyrolysis: A unique facility for the production of nanopowders,” Platin. Met. Rev., vol. 55, no. 2, pp. 149–151, 2011. [14] G. L. Chiarello et al., “Flame-synthesized LaCoO3-supported Pd. 1. Structure, thermal stability and reducibility,” J. Catal., vol. 252, no. 2, pp. 127–136, 2007. [15] C. Siriwong and S. Phanichphant, “Flame-made single phase Zn2TiO4 nanoparticles,” Mater. Lett., vol. 65, no. 12, pp. 2007–2009, 2011. [16] S. E. Pratsinis, “Flame aerosol synthesis of ceramic powders,” Prog. Energy Combust. Sci., vol. 24, no. 3, pp. 197–219, 1998. [17] A. K. Bhanwala, A. Kumar, D. P. Mishra, and J. Kumar, “Flame aerosol synthesis and

characterization of pure and carbon coated titania nano powder,” J. Aerosol Sci., vol. 40, no. 8, pp. 720–730, 2009. [18] S. Li, Y. Ren, P. Biswas, and S. D. Tse, “Flame aerosol synthesis of nanostructured materials and functional devices: Processing, modeling, and diagnostics,” Prog. Energy Combust. Sci., vol. 55, pp. 1–59, 2016. [19] K. Brezinsky, “Gas-phase combustion synthesis of materials,” Symp. Combust., vol. 26, no. 2, pp. 1805–1816, 1996. [20] A. Sokolowska, “1977, Vol. 8, pp. 219 to 230. Pergamon Press. Printed in Great Britain.,” vol. 8, no. 1946, 1977. [21] R. Baranwal, M. P. Villar, R. Garcia, and R. M. Laine, “Flame Spray Pyrolysis of Precursors as a Route to Nano-mullite Powder: Powder Characterization and Sintering Behavior,” J. Am. Ceram. Soc., vol. 84, no. 5, pp. 951–961, 2001. [22] C. R. Bickmore, K. F. Waldner, D. R. Treadwell, and R. M. Laine, “Ultrafine spinel powders by flame spray pyrolysis of a magnesium aluminum double alkoxide,” Journal of the American Ceramic Society, vol. 79, no. 5. pp. 1419–1423, 1996. [23] A. C. Sutorik, S. S. Neo, D. R. Treadwell, and R. M. Laine, “Synthesis of Ultrafine β″- Alumina Powders via Flame Spray Pyrolysis of Polymeric Precursors,” J. Am. Ceram. Soc., vol. 81, no. 6, pp. 1477–1486, 2005. [24] C. Chaisuk, A. Wehatoranawee, S. Preampiyawat, S. Netiphat, A. Shotipruk, and O. Mekasuwandumrong, “Preparation and characterization of CeO2/TiO2 nanoparticles by flame spray pyrolysis,” Ceram. Int., vol. 37, no. 5, pp. 1459–1463, 2011. [25] Y. K. Kho, W. Y. Teoh, L. Mädler, and R. Amal, “Dopant-free, polymorphic design of TiO2 nanocrystals by flame aerosol synthesis,” Chem. Eng. Sci., vol. 66, no. 11, pp. 2409–2416, 2011. [26] M. T. Matter, L. A. Furer, F. H. L. Starsich, G. Fortunato, S. E. Pratsinis, and I. K. Herrmann, “Engineering the Bioactivity of Flame-Made Ceria and Ceria/Bioglass Hybrid Nanoparticles,” ACS Appl. Mater. Interfaces, vol. 11, no. 3, pp. 2830–2839, 2019. [27] W. J. Stark, M. Maciejewski, L. Mädler, S. E. Pratsinis, and A. Baiker, “Flame-made nanocrystalline ceria/zirconia: Structural properties and dynamic oxygen exchange capacity,” J. Catal., vol. 220, no. 1, pp. 35–43, 2003. [28] L. Mädler, W. J. Stark, and S. E. Pratsinis, “Flame-made ceria nanoparticles,” J. Mater. Res., vol. 17, no. 6, pp. 1356–1362, 2002. [29] W. J. Stark, L. Mädler, M. Maciejewski, S. E. Pratsinis, and A. Baiker, “Flame synthesis of nanocrystalline ceria-zirconia: Effect of carrier liquid,” Chem. Commun., vol. 3, no. 5, pp. 588–589, 2003. [30] W. J. Stark, J. Grunwaldt, M. Maciejewski, S. E. Pratsinis, A. Baiker, and C.- Zu, “Flame-Made Pt / Ceria / Zirconia for Low-Temperature Oxygen Exchange,” Chem. Mater., no. 5, pp. 3352–3358, 2005. [31] W. Y. Teoh, R. Amal, and L. Mädler, “Flame spray pyrolysis: An enabling technology for nanoparticles design and fabrication,” Nanoscale, vol. 2, no. 8, pp. 1324–1347, 2010. [32] R. Koirala, S. E. Pratsinis, and A. Baiker, “Synthesis of catalytic materials in flames: Opportunities and challenges,” Chem. Soc. Rev., vol. 45, no. 11, pp. 3053–3068, 2016. [33] G. D. Ulrich, “Theory of Particle Formation and Growth in Oxide Synthesis Flames,” Combust. Sci. Technol., vol. 4, no. 1, pp. 47–57, 1971. [34] L. Mädler, “Liquid-fed aerosol reactors for one-step synthesis of nano-structured particles,” KONA Powder Part. J., vol. 22, no. March, pp. 107–120, 2004. [35] R. Jossen, R. Mueller, S. E. Pratsinis, M. Watson, and M. K. Akhtar, “Morphology and

composition of spray-flame-made yttria-stabilized zirconia nanoparticles,” Nanotechnology, vol. 16, no. 7, 2005. [36] R. Strobel and S. E. Pratsinis, “Flame aerosol synthesis of smart nanostructured materials,” J. Mater. Chem., vol. 17, no. 45, pp. 4743–4756, 2007. [37] L. Mädler and S. E. Pratsinis, “Bismuth Oxide Nanoparticles by Flame Spray Pyrolysis,” J. Am. Ceram. Soc., vol. 85, no. 7, pp. 1713–1718, 2004. [38] N. Cordonnier, “Scholar Commons Flame Spray Pyrolysis of Ce-Mn Solid Solutions for Catalytic Applications,” 2019. [39] L. Mädler, H. K. Kammler, R. Mueller, and S. E. Pratsinis, “Controlled synthesis of nanostructured particles by flame spray pyrolysis,” J. Aerosol Sci., vol. 33, no. 2, pp. 369–389, 2002. [40] N. C. Castillo, A. Heel, T. Graule, and C. Pulgarin, “Flame-assisted synthesis of nanoscale, amorphous and crystalline, spherical BiVO4 with visible-light photocatalytic activity,” Appl. Catal. B Environ., vol. 95, no. 3–4, pp. 335–347, 2010. [41] Z. Zhang et al., “Tailored metastable Ce-Zr oxides with highly distorted lattice oxygen for accelerating redox cycles,” Chem. Sci., vol. 9, no. 13, pp. 3386–3394, 2018. [42] N. C. Castillo, L. Ding, A. Heel, T. Graule, and C. Pulgarin, “On the photocatalytic degradation of phenol and dichloroacetate by BiVO 4: The need of a sacrificial electron acceptor,” J. Photochem. Photobiol. A Chem., vol. 216, no. 2–3, pp. 221–227, 2010. [43] J. Suk, H. Gyu, and J. Sung, “Heterojunction semiconductors : A strategy to develop efficient photocatalytic materials for visible light water splitting,” Catal. Today, vol. 185, no. 1, pp. 270–277, 2012. [44] K. Maeda, R. Abe, and K. Domen, “Role and function of ruthenium species as promoters with TaON-based photocatalysts for oxygen evolution in two-step water splitting under visible light,” J. Phys. Chem. C, vol. 115, no. 7, pp. 3057–3064, 2011. [45] K. Maeda, “Journal of Photochemistry and Photobiology C : Photochemistry Reviews Photocatalytic water splitting using semiconductor particles : History and recent developments,” "Journal Photochem. Photobiol. C Photochem. Rev., vol. 12, no. 4, pp. 237–268, 2011. [46] Z. Chen et al., “Accelerating materials development for photoelectrochemical hydrogen production: Standards for methods, definitions, and reporting protocols,” J. Mater. Res., vol. 25, no. 1, pp. 3–16, 2010. [47] C. Jiang, S. J. A. Moniz, A. Wang, T. Zhang, and J. Tang, “Photoelectrochemical devices for solar water splitting-materials and challenges,” Chem. Soc. Rev., vol. 46, no. 15, pp. 4645–4660, 2017. [48] B. Adeli and F. Taghipour, “A Review of Synthesis Techniques for Gallium-Zinc Oxynitride Solar-Activated Photocatalyst for Water Splitting,” vol. 2, no. 7, pp. 118– 126, 2013. [49] A. Kudo, “Recent progress in the development of visible light-driven powdered photocatalysts for water splitting,” vol. 32, pp. 2673–2678, 2007. [50] H. Kominami, S. Y. Murakami, J. I. Kato, Y. Kera, and B. Ohtani, “Correlation between some physical properties of titanium dioxide particles and their photocatalytic activity for some probe reactions in aqueous systems,” J. Phys. Chem. B, vol. 106, no. 40, pp. 10501–10507, 2002. [51] F. Amano, K. Nogami, M. Tanaka, and B. Ohtani, “Correlation between surface area and photocatalytic activity for acetaldehyde decomposition over bismuth tungstate particles with a hierarchical structure,” Langmuir, vol. 26, no. 10, pp. 7174–7180, 2010.

[52] K. Kočí et al., “Effect of TiO2 particle size on the photocatalytic reduction of CO2,” Appl. Catal. B Environ., vol. 89, no. 3–4, pp. 494–502, 2009. [53] H. Kominami, J. I. Kato, M. Kohno, Y. Kera, and B. Ohtani, “Photocatalytic mineralization of acetic acid in aerated aqueous suspension of ultra-highly active titanium(IV) oxide prepared by hydrothermal crystallization in toluene,” Chemistry Letters, no. 12. pp. 1051–1052, 1996. [54] F. Amano, E. Ishinaga, and A. Yamakata, “Effect of particle size on the photocatalytic activity of WO3 particles for water oxidation,” J. Phys. Chem. C, vol. 117, no. 44, pp. 22584–22590, 2013. [55] J. Yang, D. Wang, H. Han, and C. Li, “Roles of cocatalysts in photocatalysis and photoelectrocatalysis,” Acc. Chem. Res., vol. 46, no. 8, pp. 1900–1909, 2013. [56] J. Zhu and M. Zäch, “Nanostructured materials for photocatalytic hydrogen production,” Curr. Opin. Colloid Interface Sci., vol. 14, no. 4, pp. 260–269, 2009. [57] W. Y. Teoh, “A perspective on the flame spray synthesis of photocatalyst nanoparticles,” Materials (Basel)., vol. 6, no. 8, pp. 3194–3212, 2013. [58] T. Takata, G. Hitoki, J. N. Kondo, M. Hara, H. Kobayashi, and K. Domen, “Visible- light-driven photocatalytic behavior of tantalum-oxynitride and nitride,” Res. Chem. Intermed., vol. 33, no. 1–2, pp. 13–25, 2007. [59] A. Kasahara et al., “Photoreactions on LaTiO2N under visible light irradiation,” J. Phys. Chem. A, vol. 106, no. 29, pp. 6750–6753, 2002. [60] R. Sasaki, K. Maeda, Y. Kako, and K. Domen, “Preparation of calcium tantalum oxynitride from layered oxide precursors to improve photocatalytic activity for hydrogen evolution under visible light,” Appl. Catal. B Environ., vol. 128, pp. 72–76, 2012. [61] T. Kente and S. Dalton, “Gallium nitride nanostructures : Synthesis , characterization and applications,” J. Cryst. Growth, vol. 444, pp. 55–72, 2016. [62] M. Litter, P. Boule, D. Bahnemann, and P. Robertson, Environmental Photochemistry Part II, vol. 2M, no. September. 2005. [63] K. Maeda et al., “Overall Water Splitting on ( Ga 1 - x Zn x )( N 1 - x O x ) Solid Solution Photocatalyst : Relationship between Physical Properties and Photocatalytic Activity,” pp. 20504–20510, 2005. [64] A. Kudo and Y. Miseki, “Heterogeneous photocatalyst materials for water splitting,” Chem. Soc. Rev., vol. 38, no. 1, pp. 253–278, 2009. [65] H. Kato, H. Kobayashi, and A. Kudo, “Role of Ag+ in the band structures and photocatalytic properties of AgMO3 (M: Ta and Nb) with the perovskite structure,” J. Phys. Chem. B, vol. 106, no. 48, pp. 12441–12447, 2002. [66] “New photocatalyst group for water decomposition of RuO2-loaded p-block metal (In, Sn, and Sb) oxides with d10 configuration,” J. Phys. Chem. B, vol. 105, no. 26, pp. 6061–6063, 2001. [67] N. Kobayashi, J. Li, T. Watanabe, and T. Ishigaki, “Synthesis of gallium nitride powder and improved UV emission of gallium nitride in pulse-modulated ICP,” Ispc_18, pp. 4–7, 2007. [68] K. Maeda et al., “GaN:ZnO Solid Solution as a Photocatalyst for Visible-Light-Driven Overall Water Splitting,” J. Am. Chem. Soc., vol. 127, no. 23, pp. 8286–8287, 2005. [69] B. Adeli and F. Taghipour, “Applied Catalysis A : General Facile synthesis of highly efficient nano-structured gallium zinc oxynitride solid solution photocatalyst for visible-light overall water splitting,” "Applied Catal. A, Gen., vol. 521, pp. 250–258, 2016. [70] Z. Wang, J. Han, Z. Li, M. Li, H. Wang, and X. Zong, “Moisture-Assisted Preparation

of Compact GaN : ZnO Photoanode Toward Effi cient Photoelectrochemical Water Oxidation,” pp. 1–8, 2016. [71] N. Nagata, Y. Uchimoto, A. Miyoshi, and K. Maeda, “Mechanistic Insight on the Formation of GaN:ZnO Solid Solution from Zn-Ga Layered Double Hydroxide Using Urea as the Nitriding Agent,” 2018. [72] K. Maeda, K. Teramura, N. Saito, Y. Inoue, and K. Domen, “Improvement of photocatalytic activity of ( Ga 1 − x Zn x )( N 1 − x O x ) solid solution for overall water splitting by co-loading Cr and another transition metal,” vol. 243, pp. 303–308, 2006. [73] S. Yan, Z. Wang, Z. Li, and Z. Zou, “Two-step reactive template route to a mesoporous ZnGaNO solid solution for improved photocatalytic performance,” J. Mater. Chem., vol. 21, no. 15, pp. 5682–5686, 2011. [74] M. Yang, Q. Huang, and X. Jin, “Microwave synthesis of porous ZnGaNO solid solution for improved visible light photocatalytic performance,” Solid State Sci., vol. 14, no. 4, pp. 465–470, 2012. [75] A. E. Kennedy and B. H. Meekins, “Combustion synthesis and photoelectrochemical characterization of gallium zinc oxynitrides,” pp. 9–11, 2018. [76] B. Tian, C. Li, F. Gu, H. Jiang, Y. Hu, and J. Zhang, “Flame sprayed V-doped TiO2 nanoparticles with enhanced photocatalytic activity under visible light irradiation,” Chem. Eng. J., vol. 151, no. 1–3, pp. 220–227, 2009. [77] M. J. Height, S. E. Pratsinis, O. Mekasuwandumrong, and P. Praserthdam, “Ag-ZnO catalysts for UV-photodegradation of methylene blue,” Appl. Catal. B Environ., vol. 63, no. 3–4, pp. 305–312, 2006. [78] G. Tsekouras et al., “Charge transport in dye-sensitized solar cells based on flame- made TiO 2 nanoparticles,” IEEE J. Sel. Top. Quantum Electron., vol. 16, no. 6, pp. 1641–1648, 2010. [79] W. Y. Teoh, F. Denny, R. Amal, D. Friedmann, L. Mädler, and S. E. Pratsinis, “Photocatalytic mineralisation of organic compounds: A comparison of flame-made TiO2 catalysts,” Top. Catal., vol. 44, no. 4, pp. 489–497, 2007. [80] Y. K. Kho, A. Iwase, W. Y. Teoh, L. Mädler, A. Kudo, and R. Amal, “Photocatalytic H2 evolution over TiO2 nanoparticles. The synergistic effect of anatase and rutile,” J. Phys. Chem. C, vol. 114, no. 6, pp. 2821–2829, 2010. [81] Y. Jiang, J. Scott, and R. Amal, “Exploring the relationship between surface structure and photocatalytic activity of flame-made TiO 2-based catalysts,” Appl. Catal. B Environ., vol. 126, pp. 290–297, 2012. [82] J. Marchal, T. John, R. Baranwal, T. Hinklin, and R. M. Laine, “Yttrium Aluminum Garnet Nanopowders Produced by Liquid-Feed Flame Spray Pyrolysis (LF-FSP) of Metalloorganic Precursors,” Chem. Mater., vol. 16, no. 5, pp. 822–831, 2004. [83] E. C. Lovell, J. Horlyck, J. Scott, and R. Amal, “Flame spray pyrolysis-designed silica/ceria-zirconia supports for the carbon dioxide reforming of methane,” Appl. Catal. A Gen., vol. 546, no. March, pp. 47–57, 2017. [84] T. J. Patey, R. Büchel, S. H. Ng, F. Krumeich, S. E. Pratsinis, and P. Novák, “Flame co-synthesis of LiMn2O4 and carbon nanocomposites for high power batteries,” J. Power Sources, vol. 189, no. 1, pp. 149–154, 2009. [85] J. A. Kemmler et al., “Quenched, nanocrystalline In 4Sn 3O 12 high temperature phase for gas sensing applications,” Sensors Actuators, B Chem., vol. 161, no. 1, pp. 740– 747, 2012. [86] L. Dan, Y. T. Wey, C. Selomulya, R. C. Woodward, R. Amal, and B. Rosche, “Flame- sprayed superparamagnetic bare and silica-coated maghemite nanoparticles: Synthesis,

characterization, and protein adsorption-desorption,” Chem. Mater., vol. 18, no. 26, pp. 6403–6413, 2006. [87] K. K. Akurati et al., “Flame-made WO3/TiO2 nanoparticles: Relation between surface acidity, structure and photocatalytic activity,” Appl. Catal. B Environ., vol. 79, no. 1, pp. 53–62, 2008. [88] M. Kim, “Mixed-Metal Oxide Nanopowders By Liquid-Feed Flame Spray Pyrolysis ( Lf-Fsp ): Synthesis and Processing of Core-Shell Nanoparticles,” 2008. [89] F. Migliorini, S. De Iuliis, F. Cignoli, and G. Zizak, “How ‘flat’ is the rich premixed flame produced by your McKenna burner?,” Combust. Flame, vol. 153, no. 3, pp. 384– 393, 2008. [90] R. Strobel and S. E. Pratsinis, “Effect of solvent composition on oxide morphology during flame spray pyrolysis of metal nitrates,” Phys. Chem. Chem. Phys., vol. 13, no. 20, pp. 9246–9252, 2011. [91] F. M. Lee and L. E. Lahti, “Solubility of Urea in Water-Alcohol Mixtures,” J. Chem. Eng. Data, vol. 17, no. 3, pp. 304–306, 1972. [92] S. Tischer, M. Börnhorst, J. Amsler, G. Schoch, and O. Deutschmann, “Thermodynamics and reaction mechanism of urea decomposition,” Phys. Chem. Chem. Phys., vol. 21, no. 30, pp. 16785–16797, 2019. [93] S. D. Yim et al., “Decomposition of urea into NH3 for the SCR process,” Ind. Eng. Chem. Res., vol. 43, no. 16, pp. 4856–4863, 2004. [94] V. Berbenni and C. Milanese, “Thermal decomposition of gallium nitrate hydrate Ga (NO 3) 3× xH 2 O,” J. Therm. …, vol. 82, no. 3, pp. 401–407, 2005. [95] K. Maeda et al., “Overall Water Splitting on (Ga 1 - x Zn x )(N 1 - x O x ) Solid Solution Photocatalyst: Relationship between Physical Properties a,” J. Phys. Chem. B, vol. 109, no. 43, pp. 20504–20510, 2005. [96] A. de Souza Gonçalves, S. Antonio Marques de Lima, M. Rosaly Davolos, S. Gutierrez Antônio, and C. de Oliveira Paiva-Santos, “The effects of ZnGa2O4 formation on structural and optical properties of ZnO:Ga powders,” J. Solid State Chem., vol. 179, no. 5, pp. 1330–1334, 2006. [97] H. Jansen, J. Menze, and M. Muhler, “Optimizing the Synthesis of Zinc-rich Gallium Zinc Oxynitrides by Combining Co-Precipitation and Moisture-Assisted Nitridation,” Zeitschrift fur Anorg. und Allg. Chemie, vol. 644, no. 24, pp. 1686–1690, 2018. [98] S. Söllradl et al., “Nitrogen-Doping in ZnO via Combustion Synthesis?,” Chem. Mater., vol. 27, no. 12, pp. 4188–4195, 2015. [99] M. Yang, L. juan Wan, and X. qi Jin, “Synthesis of ZnGaNO solid solution–carbon nitride intercalation compound composite for improved visible light photocatalytic activity,” J. Cent. South Univ., vol. 24, no. 2, pp. 276–283, 2017. [100] J. F. D. Figueiredo et al., “Zinc-gallium oxynitride powders: Effect of the oxide precursor synthesis route,” Ceramica, vol. 59, no. 350, pp. 269–276, 2013. [101] K. Maeda and K. Domen, “Solid solution of GaN and ZnO as a stable photocatalyst for overall water splitting under visible light,” Chem. Mater., vol. 22, no. 3, pp. 612–623, 2010. [102] J. H. Yang, Y. R. Pei, S. J. Kim, G. Choi, A. Vinu, and J. H. Choy, “Highly Enhanced Photocatalytic Water-Splitting Activity of Gallium Zinc Oxynitride Derived from Flux-Assisted Zn/Ga Layered Double Hydroxides,” Ind. Eng. Chem. Res., vol. 57, no. 48, pp. 16264–16271, 2018. [103] M. Mapa and C. S. Gopinath, “Combustion synthesis of triangular and multifunctional ZnO 1-xNx (x = 0.15) materials,” Chem. Mater., vol. 21, no. 2, pp. 351–359, 2009. [104] G. Kale, S. Arbuj, U. Kawade, S. Rane, J. Ambekar, and B. Kale, “Synthesis of porous

nitrogen doped zinc oxide nanostructures using a novel paper mediated template method and their photocatalytic study for dye degradation under natural sunlight,” Mater. Chem. Front., vol. 2, no. 1, pp. 163–170, 2018. [105] B. Adeli and F. Taghipour, “Facile synthesis of highly efficient nano-structured gallium zinc oxynitride solid solution photocatalyst for visible-light overall water splitting,” Appl. Catal. A Gen., vol. 521, pp. 250–258, 2016. [106] P. M. Shah, J. W. H. Burnett, D. J. Morgan, T. E. Davies, and S. H. Taylor, “Ceria– zirconia mixed metal oxides prepared via mechanochemical grinding of carbonates for the total oxidation of propane and naphthalene,” Catalysts, vol. 9, no. 5, 2019. [107] Y. L. Hu et al., “ZnGaNO Photocatalyst Particles Prepared from Methane-Based Nitridation Using Zn/Ga/CO 3 LDH as Precursor,” Inorg. Chem., vol. 57, no. 15, pp. 9412–9424, 2018. [108] H. Dixit et al., “Electronic structure and band gap of zinc spinel oxides beyond LDA: ZnAl2O4, ZnGa2O4 and ZnIn 2O4,” New J. Phys., vol. 13, 2011. [109] T. Hirai et al., “ Origin of Visible Light Absorption in GaN-Rich (Ga 1 - x Zn x )(N 1 - x O x ) Photocatalysts ,” J. Phys. Chem. C, vol. 111, no. 51, pp. 18853–18855, 2007. [110] A. Kudo, “Recent progress in the development of visible light-driven powdered photocatalysts for water splitting,” Int. J. Hydrogen Energy, vol. 32, no. 14, pp. 2673– 2678, 2007. [111] T. Hisatomi, K. Maeda, D. Lu, and K. Domen, “The effects of starting materials in the synthesis of (Ga 1-x Zn x )(N 1-x O x ) solid solution on its photocatalytic activity for overall water splitting under visible light,” ChemSusChem, vol. 2, no. 4, pp. 336–343, 2009. [112] R. J. Farrauto, M. Deeba, and S. Alerasool, “Gasoline automobile catalysis and its historical journey to cleaner air,” Nat. Catal., vol. 2, no. 7, pp. 603–613, 2019. [113] “Assessing California ’ s Climate Policies — An Overview,” no. December, 2018. [114] T. Montini, M. Melchionna, M. Monai, and P. Fornasiero, “Fundamentals and Catalytic Applications of CeO 2 ‑ Based Materials,” 2016. [115] M. Pudukudy and Z. Yaakob, “Catalytic aspects of ceria-zirconia solid solution: Part-I An update inthe synthesis, propertiesand chemical reactions of ceria zirconia solid solution,” Der Pharma Chem., vol. 6, no. 1, pp. 188–216, 2014. [116] H. Schulz, W. J. Stark, M. Maciejewski, S. E. Pratsinis, and A. Baiker, “Flame-made nanocrystalline ceria/zirconia doped with alumina or silica: Structural properties and enhanced oxygen exchange capacity,” J. Mater. Chem., vol. 13, no. 12, pp. 2979– 2984, 2003. [117] N. J. Taylor, “Oxide Nanopowders for the Processing of Ceramic,” 2015. [118] D. Devaiah et al., “Ceria – zirconia mixed oxides : Synthetic methods and applications Ceria – zirconia mixed oxides : Synthetic methods and,” Catal. Rev., vol. 60, no. 2, pp. 177–277, 2018. [119] X. Wu, Q. Liang, and D. Weng, “Effect of manganese doping on oxygen storage capacity of ceria-zirconia mixed oxides,” J. Rare Earths, vol. 24, no. 5, pp. 549–553, 2006. [120] T. Baidya, A. Marimuthu, M. S. Hegde, N. Ravishankar, and G. Madras, “Higher catalytic activity of nano-Ce1-x-yTixPd yO2-δ compared to nano-Ce1-xPd xO2-δ for CO oxidation and N2O and NO reduction by CO: Role of oxide ion vacancy,” J. Phys. Chem. C, vol. 111, no. 2, pp. 830–839, 2007. [121] A. Manzoor, S. Pandey, D. Chakraborty, S. R. Phillpot, and D. S. Aidhy, “Entropy contributions to phase stability in binary random solid solutions,” npj Comput. Mater.,

vol. 4, no. 1, pp. 1–10, 2018. [122] B. S. Murty, High-Entropy Alloys. . [123] D. B. Miracle and O. N. Senkov, “A critical review of high entropy alloys and related concepts,” Acta Mater., vol. 122, pp. 448–511, 2017. [124] T. A. Lee, C. R. Stanek, K. J. McClellan, J. N. Mitchell, and A. Navrotsky, “Enthalpy of formation of the cubic fluorite phase in the ceria-zirconia system,” J. Mater. Res., vol. 23, no. 4, pp. 1105–1112, 2008. [125] A. Sarkar et al., “High-Entropy Oxides: Fundamental Aspects and Electrochemical Properties,” Adv. Mater., vol. 31, no. 26, 2019. [126] R. Grau-Crespo, N. H. De Leeuw, S. Hamad, and U. V. Waghmare, “Phase separation and surface segregation in ceria-zirconia solid solutions,” Proc. R. Soc. A Math. Phys. Eng. Sci., vol. 467, no. 2131, pp. 1925–1938, 2011. [127] Z. Wu, M. C. Troparevsky, Y. F. Gao, J. R. Morris, G. M. Stocks, and H. Bei, “Phase stability, physical properties and strengthening mechanisms of concentrated solid solution alloys,” Curr. Opin. Solid State Mater. Sci., vol. 21, no. 5, pp. 267–284, 2017. [128] T. H. Etsell and S. N. Flengas, “The electrical properties of solid oxide electrolytes,” Chem. Rev., vol. 70, no. 3, pp. 339–376, 1970. [129] P. Esteves, Y. Wu, C. Dujardin, M. K. Dongare, and P. Granger, “Ceria-zirconia mixed oxides as thermal resistant catalysts for the decomposition of nitrous oxide at high temperature,” Catal. Today, vol. 176, no. 1, pp. 453–457, 2011. [130] E. C. Vagia and A. A. Lemonidou, “Investigations on the properties of ceria-zirconia- supported Ni and Rh catalysts and their performance in acetic acid steam reforming,” J. Catal., vol. 269, no. 2, pp. 388–396, 2010. [131] C. Wang et al., “The dehydrogenation of ethylbenzene with CO2 over V 2O5/CexZr1- xO2 prepared with different methods,” J. Mol. Catal. A Chem., vol. 329, no. 1–2, pp. 64–70, 2010. [132] W. Wang, S. Wang, X. Ma, and J. Gong, “Crystal structures, acid-base properties, and reactivities of CexZr1-xO2 catalysts,” Catal. Today, vol. 148, no. 3–4, pp. 323–328, 2009. [133] B. M. Reddy et al., “Structural characterization and oxidative dehydrogenation Activity of v2O5/CexZr1-xO2/SiO 2 catalysts,” J. Phys. Chem. B, vol. 110, no. 18, pp. 9140–9147, 2006. [134] B. M. Reddy, P. Lakshmanan, P. Bharali, and P. Saikia, “Dehydration of 4- methylpentan-2-ol over CexZr1-xO2/SiO2 nano-composite catalyst,” J. Mol. Catal. A Chem., vol. 258, no. 1–2, pp. 355–360, 2006. [135] G. Xiao, S. Li, H. Li, and L. Chen, “Synthesis of doped ceria with mesoporous flowerlike morphology and its catalytic performance for CO oxidation,” Microporous Mesoporous Mater., vol. 120, no. 3, pp. 426–431, 2009. [136] A. C. Sutorik and M. S. Baliat, “Solid Solution Behavior of CexZr1-xO2 Nanopowders Prepared by Flame Spray Pyrolysis of Solvent-Borne Precursors,” J. Metastable Nanocrystalline Mater., vol. 13, pp. 371–376, 2002. [137] B. Weidenhof et al., “High-throughput screening of nanoparticle catalysts made by flame spray pyrolysis as hydrocarbon/NO oxidation catalysts,” J. Am. Chem. Soc., vol. 131, no. 26, pp. 9207–9219, 2009. [138] Z. Zhang et al., “Tailored metastable Ce-Zr oxides with highly distorted lattice oxygen for accelerating redox cycles,” Chem. Sci., vol. 9, no. 13, pp. 3386–3394, 2018. [139] L. Nie et al., “Activation of surface lattice oxygen in single-atom Pt/CeO2 for low- temperature CO oxidation,” Science (80-. )., vol. 358, no. 6369, pp. 1419–1423, 2017. [140] F. Zhang, C. H. Chen, J. C. Hanson, R. D. Robinson, I. P. Herman, and S. W. Chan,

“Phases in ceria-zirconia binary oxide (1-x)CeO 2-xZrO 2 Nanoparticles: The effect of particle size,” J. Am. Ceram. Soc., vol. 89, no. 3, pp. 1028–1036, 2006. [141] K. D. Malviya, C. Srivastava, and K. Chattopadhyay, “Phase formation and stability of alloy phases in free nanoparticles: Some insights,” RSC Adv., vol. 5, no. 45, pp. 35541–35550, 2015. [142] L. J. M. Davis, I. Heinmaa, B. L. Ellis, L. F. Nazar, and G. R. Goward, “Influence of particle size on solid solution formation and phase interfaces in Li0.5FePO4 revealed by 31P and 7Li solid state NMR spectroscopy,” Phys. Chem. Chem. Phys., vol. 13, no. 11, pp. 5171–5177, 2011. [143] R. Mueller, L. Mädler, and S. E. Pratsinis, “Nanoparticle synthesis at high production rates by flame spray pyrolysis,” Chem. Eng. Sci., vol. 58, no. 10, pp. 1969–1976, 2003. [144] T. Tani, L. Mädler, and S. E. Pratsinis, “Homogeneous ZnO nanoparticles by flame spray pyrolysis,” J. Nanoparticle Res., vol. 4, no. 4, pp. 337–343, 2002. بببببب ببببب ببببب ببببببب بببNo Title ,م .، .شكري [145] ببببببببب ببببب بببببب ببببببب بببب ببببببب . .ببببببب [146] K. Wegner, B. Schimmoeller, B. Thiebaut, C. Fernandez, and T. N. Rao, “Pilot plants for industrial nanoparticle production by flame spray pyrolysis,” KONA Powder Part. J., vol. 29, no. 29, pp. 251–265, 2011. [147] C. Kuang, Y. Li, S. Zhu, and J. Li, “Influence of different low air pressure on combustion characteristics of ethanol pool fires,” Procedia Eng., vol. 62, pp. 226–233, 2013. [148] http://data.europa.eu/eli/reg/2018/842/oj. Retrieved 04/10/2020 [149] https://hydrogeneurope.eu/hydrogen-production-0. Retrieved on 04/10/2020. [150] https://pubchem.ncbi.nlm.nih.gov. Retrieved on 04/10/2020. [151] https://webbook.nist.gov. Retrieved on 04/10/2020. [152] https://ww2.arb.ca.gov. Retrieved on 04/10/2020

APPENDIX A: Synthesis Conditions for Gallium Zinc Oxynitrides

Table A1: Solution compositions and their synthesis conditions GZN, Z1 -Z3 and ZN1 to ZN5

Sample Ga: Zn: N Solvent Synthesis Conditions GZN1 1:1:10 Methanol (1, 1.7, 5)

GZN2 1:1:10 10 ml DI H O + Ethanol (1, 1.7, 5) 2 GZN3 1:3:25 40ml Methanol + Butanol (1.5, 3, 3.5)

GZN4 1:6:25 40ml Methanol + Butanol (1.5, 3, 3.5)

GZN5 1:3:10 20 ml DI H2O + Butanol (1.5, 3, 3.5)

GZN6 1:3:20 20 ml DI H2O + Butanol (1.5, 3, 3.5)

ZN1 0:1:1 40ml Methanol + acetonitrile (1.5, 2, 3)

ZN2 0:1:1 40ml Methanol + acetonitrile (1.5, 2, 3)

ZN5 0:1:5 40ml Methanol + acetonitrile (1.5, 3, 3.5)

Z1 0:1:0 Ethanol (1, 1.7, 3)

Z2 0:1:0 40ml Methanol + acetonitrile (1, 1.7, 3)

Z3 0:1:0 40ml Methanol + acetonitrile (1, 2, 3)

N.B: The synthesis conditions are (pressure (in bar), Precursor rate (mL/min), O2 rate (L/min)).

APPENDIX B: Thermodynamic calculations for ceria zirconia solid solution

Fig. B1: Heat formation as a function of the concentration of ZrO2 in Ceria[124]

Fig. B2: Modelled heat of mixing as a function of the concentration of ZrO2 in Ceria Compared with experimental work [126]

The goal here is to estimate the temperature required to achieve a thermodynamically stable solid solution of Ceria Zirconia with reference to equimolar composition, Ce0.5Zr0.5O2. The main equation used is the simplified Gibb’s free energy equation. The assumptions and discussions are provided in Chapter 6 of this work.

∆퐺푚𝑖푥 = ∆퐻푚𝑖푥 - T ∗ ∆ 푆푐표푛푓𝑖푔 ……………………………………………. b1

− ∆ 푆 푐표푛푓𝑖푔 = 푅(푥푐퐼푛 푥푐 + 푥푧퐼푛 푥푧) ………………………………………. b2

With equimolar composition,

∆푆푐표푛푓𝑖푔 = 0.6931 ∗R ………………………………………………………... b3

∆퐺푚𝑖푥 = ∆퐻푚𝑖푥 - T ∗ 0.6931 ∗ R ……………....……………………………...b4

To estimated ∆퐻푚𝑖푥 , figs. B1 and B2 were compared at x = 0.5,

∆퐻푓 = 푥 ∗ ∆퐻푚 ↔ 푐((푍푟푂2) + ∆퐻푚𝑖푥 ………………………………………. b5

∆퐻푚 ↔ 푐 = 8.8 ± 2.2 kJ/mol

By this, the minimum ∆퐻푚𝑖푥 = 9.5 kJ/mol ………………………………. b6

Putting (b6) into (b4), and setting the equation to zero (equilibrium)

푇 (퐾) = 1600.

Therefore, a more than 1600 K is required to drive the Gibb’s energy to negative

(thermodynamically stable regime for solid solution).

APPENDIX C: Pictures showing effects of synthesis conditions on flame

Fig. C1: Variation of flame height with precursor flowrate at constant oxygen flowrate

Fig. C2: Variation of flame height with oxygen flowrate at constant precursor flowrate and pressure drop.