A Dissertation Entitled Production of Biofuels and Value-Added

A Dissertation

entitled

Production of Biofuels and Value-Added Chemicals from Oleaginous Biomass

Yaser Shirazi

Submitted to the Graduate Faculty as partial fulfillment of the requirements for the

Doctor of Philosophy Degree in Chemical and Environmental Engineering

______Dr. Sridhar Viamajala, Committee Chair

______Dr. Sasidhar Varanasi, Committee Member

______Dr. Ana C. Alba-Rubio, Committee Member

______Dr. G. Glenn Lipscomb, Committee Member

______Dr. Maria R. Coleman, Committee Member

______Dr. Patricia Ann Relue, Committee Member

______Dr. Amanda Bryant-Friedrich, Dean College of Graduate Studies

The University of Toledo March 2018

This document is copyrighted material. Under copyright law, no parts of this document may be reproduced without the expressed permission of the author. An Abstract of

Production of Biofuels and Value-Added Chemicals from Oleaginous Biomass by

Yaser Shirazi

Submitted to the Graduate Faculty as partial fulfillment of the requirements for the Doctor of Philosophy Degree in Chemical and Environmental Engineering

The University of Toledo

March 2018

Oleaginous biomass such as oilseeds and microalgae are attractive feedstocks for biofuels and chemicals production due to the presence of energy-dense triglycerides.

Pyrolysis is a promising technique that can thermally degrade whole biomass and/or triglycerides to drop-in fuels or fuel precursors. However, low yield of liquid products has remained the major obstacle for commercialization of triglyceride pyrolysis since feedstock price is the most significant component of the cost for fuel production from triglycerides. The current practice of feeding bulk liquids into hot pyrolysis reactors results in excessive decomposition and polymerization due to slow volatilization and long residence time. As such, we designed and built a novel continuous pyrolysis system equipped with an atomizer to introduce micron-sized droplets of oil into hot pyrolysis reactors. This approach facilitates rapid vaporization of the feed such that the subsequent vapor phase degradation is efficient and results in high yields of bio-oil. Further, when the volatilized feedstocks are passed through a heterogeneous catalyst bed, this method enables vapor-phase catalytic reactions with high-selectivity and high-yield (with low coke formation). Besides triglycerides, the pyrolysis system has the ability to process feedstocks containing free fatty acids (e.g. tall oil from wood pulping, waste cooking oils or oils from iii microalgae) that are generally unsuitable for biodiesel production. Finally, the reactor can be used for synthesis of fatty acid derivatives (e.g. fatty nitriles, fatty esters or alcohols) by introducing additional reactants (e.g. ammonia, methanol or hydrogen) and suitable catalysts.

Pyrolysis of soybean oil in the absence of catalyst resulted in bio-oil yields as high as 88% (theoretical yield is approximately 92%) at optimal conditions (Trxn =500 °C and τ

= 60s) with products consisting of 38% hydrocarbons (22% C5-C12 and 16% >C12), 33% long-chain fatty acids (C16-C18, but primarily oleic acid) and 15% short-chain fatty acids

(C6-C12). To promote triglyceride deoxygenation and produce high value aromatics, pyrolysis of soybean oil as well as non-edible oils (such as pennycress, camelina and waste cooking oil) in the presence of zeolite resulted in a yield of nearly 60% liquid products, of which more than 70% was benzene, toluene and xylene (BTX). Regeneration of catalyst allowed prolonger reuse - 12 reaction-regeneration cycles were performed without any measurable loss in catalyst performance. To assess the direct production of fatty acid derivatives, triglyceride feed was allowed to react with ammonia that was co-fed into the reactor. In the presence of V2O5 near-theoretical fatty nitrile yields (84 wt.% relative to the feed mass) were obtained. Energy balance calculations indicate that the one-pot reaction vapor phase reaction requires significantly lower energy than the conventional process that relies on the energy-intense triglyceride hydrolysis.

To allow (1) separate recovery of energy-dense lipid pyrolysis products and the lower calorific value bio-oils from the degradation of starch and protein and (2) tailored vapor phase upgrading of the resulting fractions, we implemented a two-stage fractional pyrolysis integrated with vapor phase upgrading on whole microalgae biomass. Chlorella

iv sp. was first pyrolyzed at 320 °C to volatilize and degrade the biomass starch and a majority of the protein. Then, the residual biomass was pyrolyzed again at 450 °C to recover products from lipid decomposition. The volatiles from each fraction were passed through an ex-situ zeolite catalyst which resulted in high yield of BTX and catalyst-free biochar.

Dedicated to

My Parents

Acknowledgements

I would like to thank Dr. Sridhar Viamajala (my advisor) for his help, advice and mentorship throughout my PhD studies. I also thank him for assisting me to write this dissertation.

I thank Dr. Sasidhar Varanasi for his guidance, advice and support during my PhD research.

I thank my dissertation committee - Drs. G. Glenn Lipscomb, Sasidhar Varanasi,

Ana C. Alba-Rubio, Maria R. Coleman and Patricia Ann Relue for their help to improve my dissertation.

I also acknowledge the funding agencies for supporting this thesis. This work was supported by National Science Foundation (CHE-1230609) and the U.S. Department of

Energy (DOE) Office of Energy Efficiency and Renewable Energy (EERE) Bioenergy

Technologies Office (BETO) (contract No. DE-EE0005993).

vii

Table of Contents

1 Background and significance ……………………………………………………..1

1.1 Production of biofuels and value-added chemicals from biomass…..……………..1

1.2 Oleaginous biomass...……………………………………………………………..2

1.3 Triglyceride conversion into biofuels ……………………………………………..5

1.4 Research outline…….……………………………………………………………..8

2 High-Yield Production of Fuel- and Oleochemical- Precursors from

Triacylglycerols in a Novel Continuous-Flow Pyrolysis Reactor…...……………………12

2.1 Abstract……..……………………………………………………………………12

2.2 Introduction………………………………………………………………………13

2.3 Experimental..……………………………………………………………………16

2.3.1 Materials…………………………………………………………………………16

2.3.2 Experimental set-up...……………………………………………………………16

2.3.3 Fractionation of pyrolysis liquid products……………………………………….18

2.3.4 Experimental design……………………………………………………………...19

2.3.5 Feedstock analysis………………………………………………………………..20

2.3.6 Gas chromatography analysis…………………………………………………….20

2.4 Results and Discussion…………………………………………………………...23

2.4.1 Design of the experimental pyrolysis system…………………………………….23

2.4.2 Feedstock characterization……………………………………………………….26

2.4.3 Pyrolysis yield ……………………………………………………………………27

2.4.4 Products analysis…………………………………………………………………30

2.4.5 Product distillation and detailed hydrocarbon analysis…………………………...36 viii

2.4.6 Feasibility of isolating oleic acid from soy oil via pyrolysis……………………..40

2.5 Conclusions………………………………………………………………………41

3 High Yield Production of Hydrocarbons from Non-edible Oils Through Reactive

Pyrolysis System…………………………………………………………………………43

3.1 Abstract…………………………………………………………………………..43

3.2 Introduction………………………………………………………………………44

3.3 Experimental……………………………………………………………………..48

3.3.1 Materials…………………………………………………………………………48

3.3.2 Experimental set-up……………………………………………………………...49

3.3.3 Experimental design……………………………………………………………...51

3.3.4 Products analysis…………………………………………………………………51

3.3.5 Fractionation of pyrolysis liquid products………………………………………..53

3.4 Results and Discussion…………………………………………………………...54

3.4.1 Catalytic pyrolysis of triglyceride…………………………………………………..54

3.4.1.1 Effects of temperature and residence time on products yield……………………..54

3.4.1.2 Effects of temperature and residence time on products composition……………..58

3.4.1.3 Feasibility of complete deoxygenation of liquid product…………………………62

3.4.1.4 Effect of reaction time on stream…………………………………………………64

3.4.2 Catalytic pyrolysis of non-edible oils feedstock……………………………………67

3.4.2.1 Products yields and compositions………………………………………………...67

3.4.2.2 Catalyst long-term reusability …………………………………………………...69

3.4.2.2.1 Design of reaction-regeneration cycle………………………………………….69

3.4.2.2.2 Products yields and compositions from reaction-regeneration cycle…………...70

3.4.2.2.3 Fractionation of liquid products………………………………………………...71

3.5 Conclusion……………………………………………………………………….73

4 High-Yield Production of Fatty Nitriles by One-Step Vapor Phase Thermo-

Catalysis of Triglycerides………………………………………………………………...74

4.1 Abstract…………………………………………………………………………..74

4.2 Introduction………………………………………………………………………75

4.3 Experimental……………………………………………………………………..80

4.3.1 Materials…………………………………………………………………………80

4.3.2 Experimental set-up……………………………………………………………...80

4.3.3 Experimental conditions………………………………………………………….82

4.3.4 Catalyst characterization…………………………………………………………83

4.3.5 Feedstock and products analysis………………………………………………….84

4.4 Results and Discussion…………………………………………………………...85

4.4.1 Feedstock characterization…………………………………………………………85

4.4.2 Catalyst characterization…………………………………………………………...85

4.4.3 Products yields and compositions…………………………………………………..86

4.4.4 Effects of catalyst acidity on fatty nitrile yield……………………………………...89

4.4.5 Effect of NH3/triglyceride molar ratio……………………………………………...90

4.4.6 Proposed reaction mechanism of fatty nitrile production…………………………...92

4.4.7 Comparison between conventional and one-step vapor phase nitrile process……...93

4.5 Conclusions………………………………………………………………………95

5 Thermochemical Conversion of Microalgae to Biofuels and Chemicals: A Study on

Integration of Microalgae Fractionation Pyrolysis with Ex-situ Catalytic Upgrading…..96

5.1 Abstract…………………………………………………………………………96

5.2 Introduction……………………………………………………………………..97

5.3 Experimental……………………………………………………………………100

5.3.1 Materials………………………………………………………………………..100

5.3.2 Pyroprobe-GC-MS pyrolysis………………………………………………...…101

5.3.3 Experimental procedure………………………………………………………...102

5.3.4 Analytical method………………………………………………………………103

5.3.4.1 Feedstock characterization……………………………………………………...103

5.3.4.2 Analysis of the pyrolysis products………………………………………………104

5.3.4.2.1 Gas chromatography (GC) analysis …………………………………………..104

5.3.4.2.2 Calorific value………………………………………………………………...105

5.4 Results and Discussion………………………………………………………….106

5.4.1 Feedstock characterization ...... 100

5.4.2 Single step pyrolysis ...... 100

5.4.2.1 Products yields ...... 115

5.4.2.2 Bio-oil compositions...... 115

5.4.2.3 Bio-char elemental analysis ...... 115

5.4.3 Fractional pyrolysis of microalgae ...... 100

5.4.3.1 Products yields ...... 115

5.4.3.2 Products compositions ...... 116

5.5 Conclusion ...... 119

References…………………………………………………………………………...... 120 xi

Appendix A: High-Yield Production of Fuel- and Oleochemical- Precursors from

Triacylglycerols in a Novel Continuous-Flow Pyrolysis Reactor……………………….139

Appendix B: High Yield Production of Hydrocarbons from Non-edible Oils Through

Novel Reactive Pyrolysis System……………………………………………………….151

Appendix C: High-Yield Production of Fatty Nitriles by One-Step Vapor Phase Thermo-

Catalysis of Triglycerides……………………………………………………………….161

Appendix D: Thermochemical Conversion of Algae to Biofuels and Chemicals: A Study on Integration of Algae Fractionation and Ex-situ Catalytic Pyrolysis………………169

xii

List of Tables

Table 2.1: Relative mass of different classes of compounds in the distilled fraction

(DF)...... 38

Table 3.1: Analytical techniques used to characterize feedstock and liquid/gas products.

...... 53

Table 3.2: Products yields and composition from catalytic pyrolysis of non-edible oils.

...... 68

Table 3.3: Distillation of OLP from 12 reaction-regeneration of WCO pyrolysis. .... 72

Table 4.1: Operating conditions for the one-step vapor phase nitrile reaction of triglyceride...... 82

Table 4.2: Texture-properties and acidity of tested catalysts...... 86

Table 4.3: Comparison of energy requirements for conventional nitrile process and the proposed one-step vapor phase nitrile production. Detailed energy calculations and assumptions are given in Appendix C...... 94

Table 5.1: Compositional, proximate and ultimate analysis of feedstocks.

Carbohydrates include cellulose and hemicellulose in the woodchips and starch in soybean flake and Chlorella sp. microalgae. Woodchips consisted of lignin instead of lipid. Mass fraction of oxygen was calculated by difference...... 107

Table 5.2: Bio-oil composition from pyrolysis of microalgae at different reaction temperatures. The values are average of two experiments and based on weight percentage relative to dry and ash-free biomass...... 111

Table 5.3: Elemental analyses of bio-char obtained from microalgae pyrolysis in presence of ex-situ catalyst. ‘‘Dry-basis’’ values were obtained by CHN analyzer. ‘‘Dry- xiii ash free basis’’ values were calculated by using ‘‘dry-basis’’ values, and ash content.

Calorific values (HHV) were calculated using Equations 5.4 and 5.5. All values are reported as mass fractions (%). Mass fraction of oxygen was calculated by difference. 114

Table 5.4: Bio-oil composition from fractional pyrolysis. The values are average of two experiments and based on weight percentage relative to dry and ash-free biomass...... 117

Table 5.5: Elemental analyses of bio-char obtained from fractional pyrolysis of algae.

‘‘Dry-basis’’ values were obtained by CHN analyzer. ‘‘Dry-ash free basis’’ values were calculated by using ‘‘dry-basis’’ values, and ash content. Calorific values (HHV) were calculated using Equations 5.4 and 5.5. All values are reported as mass fractions (%). Mass fraction of oxygen was calculated by difference...... 118

xiv

List of Figures

Figure 1.1: Schematic diagram of the three major steps in energy and products production from oleaginous biomass...... 3

Figure 1.2: Triglyceride structure...... 5

Figure 1.3: (a) Transesterification reaction of triglyceride to produce FAMEs and (b) schematic diagram of biodiesel plant. Adopted from UOP report [42]...... 7

Figure 1.4: Summary of production of biofuel and value-added chemicals from oleaginous biomass...... 9

Figure 2.1: Schematic diagram of pyrolysis system...... 17

Figure 2.2: (a) Conversion of feed to non-glyceride products (see Equation 2.1), (b) yield of liquid recovered (see Equation 2.2), and (c) yield of non-glyceride liquid products

(see Equation 2.3) versus vapor residence time at the pyrolysis temperatures. Dashed line represents the theoretical yield of non-glyceride liquid products...... 28

Figure 2.3: A general thermal cracking mechanism of soybean oil...... 32

Figure 2.4: Content of long-chain fatty acids (a), short-chain fatty acids (b), heavy hydrocarbons (c), and light hydrocarbons (d) in the recovered pyrolysis liquids at the tested reaction conditions...... 33

Schematic diagram of catalytic pyrolysis of triglyceride. 1: catalyst (HZSM-

5); 2: quartz wool (as support for catalyst); Evap: stands for evaporation of feedstock. . 51

Yields of (a) organic phase (b) non-condensable gases (c) aqueous phase and (d) coke versus HZSM-5 loading at tested reaction temperatures...... 56 xv

A general thermo catalytic cracking mechanism of soybean oil ...... 59

Yields of (a) total aromatics (b) L-FAs (c) C5-C13 olefins (d) C5-C13 paraffin and (e) C14- C18 aliphatic in the organic phase...... 61

(a) Products yields and (b) composition of organic phase from catalytic pyrolysis of triglyceride at WHSV of 3-12 h-1...... 63

Effect of reaction time on stream on (a) products yields and (b) OLP compositions...... 66

(a) Products yields and (b) compositions of WCO catalytic pyrolysis at 12 reaction cycles...... 71

Figure 4.1: Pathways to produce fatty amines...... 76

Figure 4.2: Schematic diagram of one step vapor phase nitrile reaction system...... 81

Figure 4.3: (a) Products composition and (b) fatty nitrile composition from the one-step vapor phase nitrile reaction over tested catalysts. The sum of the individual fatty nitrile weight fractions in the product represents the total fatty nitrile selectivity. C8-C18 represents the carbon numbers in fatty nitriles...... 87

Figure 4.4: Correlation of catalyst acidity and fatty nitrile yields from vapor phase nitrile reaction of coconut oil...... 90

Figure 4.5: Effects of NH3/triglyceride molar ratio on products yields from one step vapor phase nitrile reaction...... 91

Figure 4.6: Possible vapor phase nitrile reaction mechanism over metal oxide catalysts.

(a) catalysts activation step (b) fatty amide formation in presence of active catalyst (c) fatty amide dehydration (d) active catalyst regeneration. MOx: metal oxide; ∆: high temperature.

...... 93

xvi

Figure 5.1: Pyroprobe-GC/MS set up...... 101

Figure 5.2: (a) Bio-oil and (b) bio-char yields from single step pyrolysis of microalgae in absence and presence of HZSM-5 catalyst at tested temperatures. The error bars denote the standard deviation from two experiments...... 110

Figure 5.3: Products yields from fractional (a) non-catalytic and (b) catalytic pyrolysis.

...... 116

xvii

Chapter 1:

Background and significance

1.1 Production of biofuels and value-added chemicals from biomass

Steady growth of world population, finite fossil fuel resources, energy security and environmental concerns necessitate the need for developing viable renewable energy resources [1] such as solar, wind, hydropower, geothermal and biomass. Of these, biomass is the only carbon-based alternative energy resource and the only sustainable source of liquid fuels. Moreover, biomass is the only renewable source for production of ethylene, benzene, toluene and xylene that are widely used in polymer, surfactant and other industries. Biomass broadly includes any organic materials or wastes such as forest resources, agricultural residue, micro- and macro-algae, plant seeds and organic industrial waste. Potential biomass feedstock for biofuel production can be classified into three categories according to the sources: cellulosic (or lignocellulosic) biomass, starch-rich derived biomass and triglyceride-rich biomass [2]. The Energy Independence and Security

Act of 2007 established a mandatory Renewable Fuel Standard (RFS) that requires transportation fuel sold in the US to contain a minimum of 36 billion gallons of renewable fuels by 2022 [3]. Of the 36 billion gallons, corn ethanol is capped at 15 billion gallons,

with the remaining 21 billion gallons coming from advanced biofuels from various sources

(e.g. lignocellulosic and triglyceride-based biomass) [4].

Lignocellulosic biomass is derived from terrestrial plants such as forest resources and agricultural residues. While lignocellulosic biomass is abundant and relatively inexpensive, it can only partially provide the US energy needs. In fact, to supply the US energy demands, it requires a plantation area of one million square miles, which is roughly one-third of the area of the 48 contiguous states. It is impossible that plantation could be implemented at this scale, not to mention that soil exhaustion would eventually occur.

Moreover, the use of lignocellulosic biomass for energy supply in large scale should be carefully controlled to prevent any negative environmental impacts due to deforestation.

Therefore, other biomass resources (e.g. algae) should be considered to complement the lignocellulosic biomass [5, 6].

1.2 Oleaginous biomass

Oleaginous biomass such as oilseeds and algae are attractive biomass resources for biofuel production due to presence of energy dense triglycerides and fatty acids [7]. As shown in Figure 1.1, cultivation, harvesting and conversion are three main steps for producing fuels and chemicals from oleaginous biomass. Algae are photosynthetic organisms capable of harvesting solar energy and converting CO2 and water to O2 and organic macromolecules such as carbohydrates and lipids. Under stress conditions such as high light or nutrient starvation, some microalgae accumulate lipids such as triacylglycerols (commonly known as triglycerides) as their main carbon storage

compounds [8]. Studies have shown that the lipid fraction of certain algae strains is comparable and sometimes greater than the oil-content of lipid producing seeds such as soybeans [7]. Better yet, the land required to produce microalgae is significantly less than that of oilseeds due to their faster growth rates and higher lipid productivities [9]. Finally, algae can be cultivated using low quality sea water or wastewater, desert lands and waste nutrients, and thus would not compete with the irrigation water, fertile land and chemical fertilizer resources required for food production [10, 11].

Energy Products

CO2 Recycle Utilization

Biofuels and Value-added Biochemical

Step 1 Separation, Blending, Modifications CO2 Sunlight

H2O Nutrients Biomass

Cultivation Pyrolysis

Fermentation

Transesterification Animal food

Hydrodeoxygenation Chemical Chemical conversion Sugars Triglycerides Protein acid Amino production Upgrading

Step 2 Energy Harvesting & H2O

Drying

Hydrolysis

Residue Oilextraction

Pyrolysis productsIntermediate Fractionation Whole Biomass Biomass Liquefaction Conversion

Gasification

Step 3 Energy

Figure 1.1: Schematic diagram of the three major steps in energy and products production from oleaginous biomass.

After cultivation, microalgae cultures are harvested and dewatered before being used in conversion step (step 3). While harvesting methods such as centrifugation, filtration and flocculation followed by settling are commercially practiced, hydrogel technology developed in our lab has shown promising results in effectively dewatering algae with relatively low energy requirement. Hydrogels are superabsorbent cross-linked polymers that can absorb large quantities of water and “swell” in aqueous solutions. Stimuli-sensitive hydrogels can also “de-swell” (i.e. release the absorbed water) based on an external stimulus. The water then can be recycled and reused for algae cultivation [12].

Oleaginous biomass is mainly comprised of carbohydrate, protein and lipid. One approach to produce fuels and chemicals from oleaginous biomass, especially microalgae, is to first recover biomass constituents (“fractionation”), and then separately process each of the recovered fractions [5]. The biomass fractionation would allow the downstream processing of each fraction to be tailored to the specific chemical and physical properties of the fractions (protein, carbohydrate and lipid). In this context, the carbohydrate fraction in algae can be recovered as monomeric sugar through enzymatic hydrolysis [13]. The lipid fraction can be then extracted from the algae by different strategies such as oil press/expeller, solvent extraction and supercritical CO2 fluid extraction [14, 15]. Finally, the protein-rich residue can be used as animal feed and/or converted to amino acids [13,

16].

Alternatively, the whole biomass can be converted to fuels and chemicals through thermochemical conversion such as pyrolysis, liquefaction and gasification [17-19].

Typical reaction conditions for hydrothermal liquefaction are pressures between 50-200 atm and temperatures between 250-450 °C [18, 20, 21] and as a result, hydrothermal 4

liquefaction can incur high capital cost. Pyrolysis operates at temperature range of 400-600

°C, near-atmospheric pressure and in absence of oxygen. While a relatively dry (<10 wt.% water) feedstock used for pyrolysis, using low-value heat sources in the algae drying step would significantly decrease the processing cost. During pyrolysis process, microalgae breakdown due to thermochemical reactions, in which gases, liquids (bio-oil), and solids

(bio-char) can be produced [22, 23]. The bio-char from microalgae pyrolysis is N-rich and can be used as soil amendment or fertilizer. Moreover, the bio-oil can be upgraded to produce biofuels and value-added chemicals.

CH2O C R1 O Glycerol Fatty acid CHO C R2 backbone chains O

CH2O C R3 Figure 1.2: Triglyceride structure.

1.3 Triglyceride conversion into biofuels

Energy-dense triglycerides from oilseeds or microalgae have the potential to, at least partially, displace fossil-based fuels and chemicals [24-28]. Triglyceride is comprised of three fatty acids and one glycerol backbone (Figure 1.2). The fatty acid chains in triglyceride are composed of straight chain hydrocarbons (R1, R2 and R3) that are similar to petroleum hydrocarbons. Triglyceride’s oxygenated functional groups can also be

utilized for value-added chemicals, which are traditionally produced from functionalization and oxidation of petroleum-derived hydrocarbons [29, 30].

Vegetable oil, the current main source of triglycerides, is extracted from oilseeds such as soybean, rapeseed, and palm kernel. According to the US Department of

Agriculture (USDA) estimates, over 190 million metric tons of the vegetable oil was produced worldwide in 2016, of which the US produced more than 11 million metric tons, mainly from soybean [31]. Vegetable oils cannot be used directly in current combustion engines due to their low volatility and high viscosity [25, 32-36]. However, they can be converted to usable fuels through transesterification [37, 38]. Figure 1.3 shows the transesterification process, where triglycerides react with methanol, usually in the presence of a homogenous catalyst (e.g. sodium hydroxide or methoxide), to produce a mixture of fatty acid methyl esters (FAMEs; commonly known as biodiesel) [39]. While biodiesel has similar fuel properties as petrodiesel, it contains nearly zero sulfur content and is biodegradable [40].

However, biodiesel fuel and the production process have some significant drawbacks. Transesterification process requires a high quality vegetable oil feedstock with low water and free fatty acid content. The presence of water hydrolyzes triglycerides and leads to free fatty acid formation that can cause soap formation in the presence of alkaline catalysts. This results in a loss of catalyst and a decrease in biodiesel yield [41]. Extra pretreatment steps would be inevitable for biodiesel production from algal lipid and low quality vegetable oils that are rich in free fatty acids. Finally, since the transesterification reaction is reversible, excess methanol is required to shift the reaction towards ester

production; thus, unreacted methanol along with catalyst and glycerin need to be separated

/recovered at the end of reaction.

(a)

(b)

Figure 1.3: (a) Transesterification reaction of triglyceride to produce FAMEs and (b) schematic diagram of biodiesel plant. Adopted from UOP report [42].

The use of biodiesel as a fuel is somewhat restricted since it is not considered as a

“drop-in” alternative to hydrocarbon fuels and does not meet the ASTM D975 specification in the US and EN 590 in Europe. Most diesel engine manufacturers recommend a maximum of 5% biodiesel in petroleum diesel blends. Biodiesel has corrosive properties and appreciable amount of oxygen (~11%), which inhibits full compatibility with existing fuel infrastructure such as pipelines and storage tanks. Moreover, biodiesel has lower oxidative stability compared to petrodiesel. It also exhibits a relatively high melting point which limits its usage in cold climate regions [27, 33, 43, 44].

As an alternative, pyrolysis (or thermal cracking in the absence of O2) of vegetable oil can produce hydrocarbons that are compatible with a variety of petro-fuels such as gasoline, jet and diesel [45-48]. Studies on pyrolysis of seed oils, such as canola, palm, and soybean reveal that the primary compounds in the product are paraffins, olefins, carboxylic acids, and small amounts of aromatics [49-53]. Although a mixture is produced, the pyrolysis products can be separated (e.g. via distillation) and used either directly or processed by common refinery methods such as hydrogenation, hydrotreatment or alkylation to obtain gasoline and/or diesel like fuels. Further, unlike transesterification, conversion of triglycerides by pyrolysis avoids the use of methanol and unrecoverable homogenous catalysts.

1.4 Research outline

As discussed above, pyrolysis is a promising alternate to traditional transesterification for processing triglycerides (e.g. vegetable and algal oil) to produce biofuels and value-added chemicals. However, low yield of liquid products has remained the major obstacle for commercialization of triglyceride pyrolysis since feedstock price is the most significant component of the cost for fuel production from triglycerides. The current practice of feeding bulk liquids into hot pyrolysis reactors results in excessive decomposition and polymerization in the liquid phase due to slow volatilization. In principle, pyrolysis yields are highest when liquid phase reactions are minimized through rapid volatilization. As such, we designed and constructed a novel continuous pyrolysis system, where feed was injected through an atomizer into the reactor in the form of micron-

sized droplets. This approach allows the oils to vaporize rapidly and mitigates the excessive unfavorable reactions that occur in the liquid phase. Further, this approach also facilitates high-selectivity and high-yield vapor-phase catalytic reactions when the volatilized feedstocks pass through a heterogeneous catalyst bed. Besides triglycerides, the pyrolysis system has the ability to process feedstocks containing free fatty acids (e.g. tall oil from wood pulping, waste cooking oils or oils from microalgae) that are generally unsuitable for biodiesel production. Finally, the reactor can be used for synthesis of fatty acid derivatives

(e.g. fatty nitriles, fatty esters or alcohols) by introducing additional reactants (e.g. ammonia, methanol or hydrogen) and suitable catalysts. The dissertation is comprised of the following specific objectives related to pyrolysis-based triglyceride conversion. Figure

1.4 shows a summarized scheme of each chapter of the dissertation.

Oleaginous Biomass

Oil extraction Whole biomass

Catalytic Pyrolysis (Ch5) Triglyceride BTEX Pyrolysis Catalytic Reactive (Ch2) Pyrolysis (Ch3) Pyrolysis (Ch4) Polystyrene Polyeurethane Fatty nitriles Oleochemicals, Hydrocarbons BTEX Poly ethylene therphethalic acid

Polystyrene Surfactant, Gasoline, jet fuel, diesel Polyeurethane detergent Polymer, surfactant, detergent Poly ethylene therphethalic acid

Figure 1.4: Summary of production of biofuel and value-added chemicals from oleaginous biomass.

(Chapter 2): Non-catalytic pyrolysis of triglycerides

Pyrolysis of triglyceride was performed in a continuous reactor equipped with an atomized feeding system to produce biofuels and valuable chemicals. The pyrolysis reactions were carried out over a range of temperatures (450-500 °C) and vapor residence times (1-300 s) to assess the impact of operating conditions on triglyceride conversion, liquid products yields and compositions. Product composition and overall liquid product yields were quantified. Thereafter, the liquid products were distilled to separate hydrocarbons from other chemical compounds.

(Chapter 3): Catalytic pyrolysis of triglycerides

Non-catalytic triglyceride pyrolysis liquids typically contain oxygenated compounds, in addition to hydrocarbons. Greater deoxygenation can be achieved using catalysts that contain acidic sites such as zeolites. Moreover, catalytic pyrolysis of vegetable oil over a zeolite catalyst can increase selectivity of aromatics that are important to improve octane number of gasoline and jet fuel. These aromatics (C6-C8) are also platform molecules for production of commodity chemicals such as solvents, detergents and polymers. We hypothesized that the rapid vaporization of triglycerides due to feed injection through an atomizer would allow vapor phase catalytic reactions and result in high hydrocarbon yields. The impact of reaction temperature, catalyst loading (or weight hour space velocity; WHSV) and reaction time-on-stream on the conversion, products yields and compositions were investigated. We focused on optimizing the operating conditions to maximize benzene, toluene and xylene yields in the products. Non-edible oils such as pennycress, camelina, and waste cooking oil were also used as feedstock to demonstrate high yields from diverse feedstocks. To assess the long-term reusability of the catalyst, several reaction-regeneration cycles were performed and the product yields and 10

compositions from each cycle were analyzed. Thereafter, the liquid products collected from all cycles were combined and distilled to separate benzene, toluene and xylene from other chemicals in the products.

(Chapter 4): Reactive pyrolysis of triglycerides to produce fatty nitriles

ZrO2, V2O5, Fe2O3, CuO and Al2O3 was investigated to assess the impact of catalyst on conversion and fatty nitrile yields.

(Chapter 5): Catalytic and non-catalytic pyrolysis of algae

A two-step fractional pyrolysis of microalgae was performed to collect volatiles from protein and carbohydrate deconstruction separately from lipid degradation. Algae was first pyrolyzed at 320 °C to volatilize and degrade the biomass carbohydrates and a majority of the protein. The residual biomass was pyrolyzed again at 450 °C to recover products from lipid decomposition. The volatiles from each fraction were passed through a zeolite bed to assess hydrocarbon yields. Thermal fractionation of algae followed by catalytic upgrading of pyrolysis vapors allows the upgrading process to be tailored to biopolymer-specific bio-oil characteristics in an integrated reactor system. 11

Chapter 2

High-Yield Production of Fuel- and Oleochemical- Precursors from Triacylglycerols in a Novel Continuous-Flow Pyrolysis Reactor1

2.1 Abstract

Conversion of soybean oil was carried out in a continuous pyrolysis system with feed injected through an atomizer. This allowed introduction of micron-sized droplets of oil that could be rapidly vaporized inside the reactor. With this novel design, we were able to achieve feed vapor residence times (τ) of 6-300s without use of carrier gas, which would significantly reduce the overall cost of pyrolysis. Effects of reaction temperature

(450

At the optimum experimental conditions of Trxn =500 °C and τ = 60s, the yield of pyrolysis liquids was as high as 88% (relative to feed mass). Under these conditions, the identified products consisted of 38% hydrocarbons (22% C5-C12 and 16% >C12), 33% long-chain fatty acids (C16-C18, but primarily oleic acid) and 15% short-chain fatty acids (C6-C12).

1 Shirazi et al., Applied Energy, 179 (2016) 755-764. 12

Upon distillation of the liquid products, the long-chain fatty acids were cleanly separated from the hydrocarbons. Overall, our results demonstrate the feasibility of producing liquid products at high yields, including a wide range of fuels (gasoline, jet and diesel) and enriched oleic acid (for oleochemicals production), using our reactor design for pyrolytic conversion of vegetable oil.

Key words: Biofuel, pyrolysis, vegetable oil, soybean oil, atomization

2.2 Introduction

Pyrolysis of triglyceride-based feedstock has typically been carried out in batch reactors at a temperature range of 300-500 °C and atmospheric pressure [34, 54]. However, batch processes are not appropriate for commodity scale industrial operations due to low throughput. In addition, most batch studies report low yield of liquid products, likely due to the high residence time in these closed systems which allows secondary reactions of primary products to low molecular weight (C1-C4) noncondensable gases. For example,

Chang and Wan [55] performed batch pyrolysis of tung oil at 450 °C and residence time of 100 min and reported only 55% yield of liquid products. More recently, Kubatova et al.

[56] conducted batch pyrolysis of canola and soybean oil at 440 °C but at a lower residence time of 10 min and also with high hydrogen pressure (2200 kPa). This approach resulted in higher organic liquid product (OLP) yields (67% OLP for soybean oil and 61-69% for canola oil) likely due to the shorter reaction time and in situ hydrogenation of unsaturated primary products that lowered the production of secondary noncondensable gases.

In an effort to develop a continuous process for vegetable oil pyrolysis, Idem et al.

[53] performed thermal cracking of canola oil. Their reactor consisted of a fixed-bed of inert materials (ceramic and quartz glass chips) that was kept at 500 °C. Due to the high reactor temperature and surface area created by inert particles, feed likely vaporized and cracked simultaneously. However, the residence time was still high (18 min) and only 15% of feed was recovered as OLP in their experiments. Nearly 75% of feed was lost as uncondensed gases such as small chain hydrocarbons (C1-C4) and H2.

Akin to the concepts of fast pyrolysis of solid substrates (e.g. biomass and coal), which are carried out at high temperature and short residence time to maximize liquid products [57-59], OLP yields from vegetable oils are expected to be better in continuous reactors with low residence time, since secondary reactions would be minimized due to continuous product removal and condensation. Theoretical reaction mechanisms proposed by Maher et al. [34] also suggest that liquid formation would be more favorable at low residence time. In recent times, Meier and co-workers have explored this approach in a series of studies and have reported improved liquid product yields [60-62]. For instance, continuous pyrolysis of waste fish oil during these studies resulted in the relatively high yield of 72% liquid products at reaction temperature of 525 °C and low vapor residence time of 24s [60, 61]. In addition to short residence time, a high free fatty acid content in the feed also possibly contributed to the high yields of OLP since fatty acids are more amendable to thermal cracking than triglycerides [49]. While yields were improved, nearly

30% of feed material was still lost to uncondensed gases. An additional drawback of the reaction system proposed by Meier and co-workers is the requirement of preheating the feed. Since the objective was to quickly vaporize the feed in the reactor (for short residence 14

time), the feed was preheated to 475 °C. Thus, while the pyrolysis residence time was short, the overall time period for which feed was exposed to high temperature was still large and could have possibly resulted in some oil degradation during preheating. Including a preheater would also increase the capital cost of the reaction system.

From these studies reported in the literature, it can be noted that liquid product yields were typically in the range of 20-70% depending on reaction conditions, feedstock and reactor design. Liquid products, rather than gas are more desirable, since they have higher heating values on a volumetric basis and are easy to store and transport. Although pyrolysis of vegetable oil is more desirable than transesterification since it allows the direct production of hydrocarbons fuels, pyrolysis technology has not yet attracted commercial interest due to the key bottleneck of the inability to achieve high liquid yields. The goal of this study was to improve the yield of liquid products during pyrolysis. As such, conversion of soybean oil was carried out in a novel continuous pyrolysis system equipped with an atomizer. This allowed introduction of micron-sized droplets of oil that could be rapidly vaporized inside the pyrolysis tube. With this design, we were able to achieve low vapor residence times without preheating the feed. Since atomization is frequently utilized in car engines, painting/coating industries, and pharmaceutical manufacturing processes, it is expected that this common approach could be easily implemented on industrial scale. The influence of temperature and vapor residence time on conversion, product yields and composition of soybean oil pyrolysis was investigated.

2.3 Experimental

2.3.1 Materials

Soybean oil was obtained from Zoyeoil (Zeeland, MI, USA). Coconut oil was obtained from Zoyeoil (Zeeland, MI, USA). Hexane, chloroform, methanol, N-methyl-2- pyrolidone (NMP), sulfuric acid and enriched oleic acid were purchased from Fisher

Scientific (Pittsburgh, PA, USA). Analytical standards for fatty acids (stearic acid, oleic acid, linoleic acid, and palmitic acid), glycerides (triolein, diolein, and monolein), FAMEs

(mixtures of C8-C22 FAMEs), alkanes (C5, C6, C7, C8 and mixtures of C7-C30), olefins-

(Alphagaz PIANO), aromatics-(Alphagaz PIANO), mixtures of benzene, toluene, ethylbenzene and xylene (BTEX), gasoline, jet fuel and naphthalene were purchased from

Sigma-Aldrich (St. Louis, MO, USA).

2.3.2 Experimental set-up

All experiments were performed in a continuous pyrolysis system that is schematically shown in Figure 2.1. A stainless steel tube with an inner diameter (ID) of

2.18 cm and length of 23 cm served as the pyrolysis reactor. The tube was placed in a clam- shell furnace (Applied Test Systems, Butler, PA, USA) with 10 cm ID and 38 cm length and equipped with a temperature controller to maintain the set-point temperature. A high temperature ultrasonic atomizer (Sonazop, Farmingdale, NY, USA; model: HTNS40) was attached to one end of the reactor tube to introduce soybean oil into the pyrolysis reactor.

The atomizer consists of an ultrasonic probe with a 4 mm diameter, which was operated at constant frequency of 40 kHz. 16

Figure 2.1: Schematic diagram of pyrolysis system.

For precise control of feed flowrate, an HPLC pump (Waters, Milford, MA, USA; model: 515) was used. After pyrolysis, product vapors were passed through a condenser system which consisted of one cold-trap and two Allihn condensers connected in series.

Commercial antifreeze was used as coolant for the Allihn condensers. The antifreeze was cooled to approximately -20 °C using dry ice and pumped through the Allihn condensers via a MasterflexTM L/S peristaltic pump (Cole-Parmer, Vernon Hills, IL, USA).

Prior to each experiment, the reactor was placed in the furnace and heated while being purged with pure N2 to remove any traces of O2 from the system. After the reactor reached set point temperature, the N2 purge was stopped, atomizer was turned on, feed pump was started at a set flow rate to provide a pre-specified vapor residence time (1-300 s) in the reactor. The system was operated in continuous mode for a duration of 10-170 min, depending upon the set feed flow rate. Over this period, approximately 10 g of soy oil was processed. The exact feed mass introduced into the reactor was calculated from the difference in the weight of feed tank before and after each experiment.

At the end of the experiment, the reactor furnace was turned off and allowed to cool to room temperature. Thereafter, the liquids from all condensers were collected and weighed on an analytical balance (Mettler Toledo, USA) with ± 0.1 mg accuracy.

Conversion of soybean oil to non-glyceride products was calculated as

W −(W ) Conversion (%) = O TG out × 100 (2.1) WO where, wO is the mass of oil introduced into the reactor and (WTG)out is the mass of glycerides in collected liquid.

The yield of pyrolysis liquids was calculated as

w Yield (%) = L × 100 (2.2) wO where, wL is the weight of pyrolysis liquids collected.

The yield of non-glyceride liquid products was calculated as

W −(W ) Yield of non-glyceride liquid (%) = L TG out × 100 (2.3) WO

The feed was assumed to be all triglycerides.

2.3.3 Fractionation of pyrolysis liquid products

A simple lab-scale vacuum distillation unit was used to separate the components of pyrolysis liquid products. The procedure used is as follows: the liquid products collected after pyrolysis were transferred to a round-bottom boiling flask that was placed in a silicone oil bath. The bath was kept on a hot stir plate and heated to distillation temperature. The boiling flask was connected to a short Vigreux-type condenser followed by a Liebig condenser, and finally a collecting flask. A vacuum pump (Fisher Scientific; model:

Maxima C plus) was connected to the collecting flask to maintain the distillation unit under 18

a vacuum of 60 mm Hg. Commercial antifreeze cooled to -25 °C was used as a coolant for the Liebig condensers and recirculated via a Masterflex L/S pump. The collecting flask was kept immersed in liquid N2 during distillation.

2.3.4 Experimental design

According to the literature, temperature and vapor residence time are the two variables that can impact the yield and characteristics of pyrolysis products [63]. In our experiments, vapor residence times of 1, 6, 60 and 300 s were tested to compare flash, fast and slow pyrolysis regimes. Feed mass flowrates were adjusted to obtain the desired vapor residence time (detailed calculations are described in Section 2.4.1). Reaction temperatures of 450, 475, and 500 °C were selected to investigate effects of temperature. The lowest reactor temperature in our experiments (450 °C) is above the expected volatilization temperature of triglycerides (> 410 °C) [64]. A full factorial design with twelve runs was employed. In addition, one experiment at severe operational conditions (T=550 °C and

τ=300s) was performed to analyze the liquid products composition and the productivity of system at harsh reaction conditions. In this paper all experiments are represented by a T-τ code system, where T represent the reaction temperature in °C and τ indicates the vapor residence time in seconds. For instance, the code 475-6 describes an experiment at reaction temperature of 475 °C and a vapor residence time of 6s.

2.3.5 Feedstock analysis

To identify and quantify fatty acid constituents in the feedstock, soybean oil was transesterified to FAMEs by reacting 5 mg soybean oil with 2 mL of a methanol-sulfuric acid mixture (95/5 vol. %). The reactants were placed in a 5 mL sealable vial and immersed in a hot water bath at 90 °C for 90 min. Afterward, the reaction mixture was cooled and the

FAMEs were extracted into 2 mL hexane (extraction conditions: 90 °C, 15 min). The extract was analyzed by gas chromatography.

2.3.6 Gas chromatography analysis

To identify and quantify chemical compounds in the pyrolysis products and feedstock, a gas chromatograph (Shimadzu 2012 plus) equipped with a flame ionization detector (FID) and an auto sampler (AOC-20i) was used. For analysis of triglycerides, diglycerides, long-chain fatty acids and feedstock FAMEs (obtained from feedstock transesterification, Section 2.3.5), an RTX-biodiesel (Restek, Bellefonte, PA, USA) column was employed (15 m length, 0.32 mm ID, and 0.1 µm film thickness). The injector and FID detector temperatures were 370 °C; 1µL sample was injected. H2 was used as the carrier gas with a flowrate of 6.02 mL min -1and a split ratio of 1:10. The column temperature was initially set at 60 °C for 1 min, and was subsequently heated at a temperature ramp rate of 10 °C min-1 to 370 °C; this final temperature was maintained for

5 min at the end of the run. FID detector response was first calibrated using appropriately diluted (in hexane or chloroform) FAME standards, long-chain fatty acid standards, monoolein, diolein and triolein (0.05 – 5 mg mL-1). For sample analysis, the samples were

diluted to concentrations of 2-5 mg-sample mL-1 in chloroform (for pyrolysis biooil) or hexane (for FAME analysis of transesterified feed). Monoglyceride, diglyceride, triglyceride, FAME and long-chain fatty acid concentrations in the samples were estimated from calibration curves as previously described [14, 65].

For preliminary analysis of hydrocarbons in the pyrolysis products, an RTX-5

(Restek, Bellefonte, PA, USA) column (15 m length, 0.25 mm ID, and 0.25 µm film thickness) was used. The injector and FID detector temperatures were kept at 350 °C; 1µL

- sample was injected. H2 was used as the carrier gas with a constant flowrate of 1 mL min

1 and a split ratio of 1:10. The column temperature was initially at 30 °C for 3 min, and was subsequently heated at a temperature ramp rate of 10 °C min-1 to 350 °C; this final temperature was maintained for 3 min at the end of the run. Since the hydrocarbons were not well separated on this column, only relatively few peaks were observed, likely due to overlap of similar chemical compounds. In these analyses, calibration curves were developed using external analytical alkane standards and the concentration of each hydrocarbon group (e.g. C8 hydrocarbons) in the sample was estimated based on the calibration response of a representative alkane (e.g. C8 alkane). Pyrolysis liquid product samples were diluted to concentrations of 2-5 mg-sample mL-1 in chloroform for analysis.

For additional detailed analysis of hydrocarbons in the distilled fraction of the pyrolysis products (see Section 2.3.3), a DB-petro (Agilent, Santa Clara, CA, USA) column

(50 m length, 0.25 mm ID, and 0.5 µm film thickness) was employed. The injector and detector temperatures were kept at 300 °C; 1µL sample was injected. H2 was used as the carrier gas with a constant flowrate of 1 mL min-1 and a split ratio of 1:10. The column temperature was initially maintained at 35 °C for 15 min. Thereafter, the temperature was 21

increased at a ramp rate of 1 °C min-1 to 60 °C. The column was then maintained at this temperature for 20 min. Finally, the column temperature was increased to 300 °C at a ramp rate of 2 °C min-1 and held at the final temperature for 2 min at the end of the run. All samples were analyzed three times and the product compositions were calculated based on relative peak area. Mean values of the estimated composition are reported. Distilled fraction samples were diluted to concentrations of 2-5 mg-sample mL-1 in chloroform for analysis.

GC-MS (Bruker, 450-GC equipped with 300-MS) analysis was performed to identify the chemical compounds in the pyrolysis products and also for quantification of derivatized short-chain fatty acids (see more detailed description in the next paragraph).

An Agilent DB-5MS fused silica capillary column (length: 30 m, ID: 0.25 mm, and film thickness: 0.25 µm; Agilent Technologies, Santa Clara, CA) was used. The injector temperature was held constant at 300 °C and a split ratio of 1:100 was maintained during each sample analysis. 1µL sample was injected. Helium was used as carrier gas with constant column flow of 1.0 mL min-1. The column was programed as follows: constant temperature of 30 °C for 10 min, followed by a temperature ramp 10 °C min-1 to 300 °C and a final hold for 10 min. The transfer line, ion source, and manifold were maintained at

300, 150, and 40 °C, respectively. Chemical compounds corresponding to chromatogram peaks were identified using NIST2008 mass spectral database. A minimum 70% confidence level was used as a threshold for positive identification of IDs provided by the spectral analysis software.

For direct analysis via GC-MS, liquid samples obtained after pyrolysis or after subsequent distillation, were diluted in chloroform (2-5 mg-sample mL-chloroform-1). In 22

addition, these samples were also treated with acidified methanol to derivatize fatty acids within the sample to fatty acid methyl esters. Kubatova et al. [56] have suggested that this derivatization procedure improves the sensitivity of fatty acid detection, especially for the detection of short chain fatty acids. The derivatization procedure was similar to the transesterification method described in Section 2.3.5. In brief, 5 mg of biooil samples were reacted with 2 mL of a methanol-sulfuric acid mixture (95/5 vol. %) at 90 °C for 90 min.

Afterward, the samples (now containing FAMEs) were extracted into 2 mL hexane and analyzed by GC-MS as described in the previous paragraph.

2.4 Results and Discussion

2.4.1 Design of the experimental pyrolysis system

To prevent secondary reactions and achieve a high yield of liquid products, we focused on designing a reactor that would allow rapid vaporization of feed. It was anticipated that such a design would provide the means to maintain a short residence time of feed due to its rapid transition to vapor phase which quickly flows out of the reactor. In addition, liquid build up would be avoided and thereby polymerization/coke formation would be prevented. One way to improve vaporization/volatilization rates of viscous vegetable oil is by injecting it into a hot reactor in the form of small droplets. This would create a very large surface area to promote rapid heat transfer and vaporization of the oil.

In this study, an atomizer was used to create microscopic oil droplets. In addition to rapid vaporization, this approach allows uninterrupted addition of feed and hence continuous operation of the reactor. Stagnation of fluid was also avoided due to rapid formation of

vapors. Finally, by this approach the vapors and volatile products were pushed out without the need to apply additional carrier gas. Thus, the cost associated with supplying and heating the carrier gas could be eliminated. Elimination of carrier gas would also minimize the condenser size.

Disintegration of a liquid film into fine droplets in the surrounding environment is known as atomization [66]. In the ultrasonic atomizer used in our reactor, the mechanical vibration generated from a piezoceramic element is transferred to the in-flowing liquid creating capillary waves, which disintegrate into fine droplets and form a dense fog that exits the atomizer nozzle [67]. The droplet size generated by atomizers can be estimated from the characteristics of the atomization device as well as fluid properties and flow rates.

Based on the correlation proposed by Rajin et al. [68], we estimated that the average droplet size of oil ejected from our atomizer would be between approximately 20 to 60 µm

(depending on feed flowrates) under ambient temperature conditions (soybean oil density

910 kg m-3 [69], surface tension 32.9 mN m-1 [70], and viscosity 38 mPa-s [71]). At high temperature pyrolysis conditions, droplet size would be lower; however, we were unable to estimate the droplet size within the hot reactor due to unavailability of temperature dependent vegetable oil properties.

Assuming that vaporization is nearly instantaneous, Equation 2.4 derived from the

Ideal Gas Law was used to calculate the mass flowrates of the generated vapors at different reaction temperatures to obtain a target vapor residence time.

LAPM τ = w (2.4) RTṁ where, L is reactor length (m), A is reactor cross sectional area (m2), P is pressure (atm),

-1 Mw is soybean oil molecular weight (875 g gmol ), R is universal gas constant (0.000082 24

m3.atm mol-1.K-1), T is reactor temperature (K), and τ is vapor residence time (s). Mass flowrate was divided by soybean oil density to estimate the feed volumetric flowrate (see more detailed derivation of Equation 2.4 in Appendix A, Figure A1). In Equation 2.4, the residence time is a function of temperature and feed mass flow rates and can be adjusted by varying these parameters.

The residence time can also be calculated based on product composition. Since the average molecular weight of the product is much lower than the feed, the residence time calculated based on product composition will be lower since residence time is directly proportional to the molecular weight. The detailed calculations for product molecular weights and product-based residence time estimates are shown in pages 1-4 of Appendix

A. In addition, Table A1 (Appendix A) shows the correlation between feed- and product- residence times. Since product composition changes with reaction conditions, Table A1 provides an estimate of the residence time based on output product (τo) under the various experimental conditions reported in this manuscript. From this table, it can be seen that residence time values are drastically different when reaction temperatures are increased for the same feed flow rate. In addition, it is important to note that the estimates of residence time based on product composition are likely less accurate than the residence time estimates from feed mass (τ) due to the uncertainty (especially under more severe reaction conditions) associated with unidentified liquid and uncondensed gas products. Finally, since the reactor is of a column design (“plug-flow” rather than a “complete back mix”), the product compositions likely vary along the length of the reactor.

Since the primary purpose of residence time calculation (for the gas phase reaction described here) is to allow empirical reactor design, rather than to assess fundamental 25

reaction kinetics, the choice of feed-based residence time estimates allow for correlation of product yields with feed flow rates – a parameter that could be easily controlled in commercial practice. As such, we have used a feed-based residence time (τ) to discuss the results of this manuscript.

Reports from literature have revealed that thermal decomposition and volatilization of vegetable oil is slow below 420 °C [72]. In addition, thermodynamic simulations of vegetable oil cracking reactions suggest that the scission of C=C starts at 400 °C [73]. To stay sufficiently above the minimum reaction temperature threshold, the lowest temperature used in our experiments was 450 °C. Very high temperatures are also reported to decrease the liquid products yield (due to further cracking) [62], in addition to increasing the operating costs. Consequently, the highest temperature in our studies was kept at 550

°C.

2.4.2 Feedstock characterization

Soybean oil was derivatized to FAMEs to quantify the fatty acid composition of feed by GC-FID. Consistent with previous observations [56], our results show that the soybean oil used in this study was comprised of linoleic and linolenic (53.4 wt. %), oleic

(30.6 wt. %), palmitic (9.8 wt. %), and stearic (6.2 wt. %) acids. As expected, the majority of fatty acids in soybean oil (84%) were unsaturated.

2.4.3 Pyrolysis yield

Figure 2.2a shows the increase in pyrolytic conversion (Equation 2.1) of soybean oil with increasing τ and reaction temperature. Thermal cracking of vegetable oil is an endothermic reaction, and is thus favored at high temperature [53]. At high reaction temperature (475-500 °C), nearly 100% conversion was observed even at low τ (~6 s).

However, at low temperature (450 °C) complete conversion was achieved at much higher vapor residence time (60-300 s). Interestingly, at very short vapor residence time (τ = 1 s) conversion of soybean oil was nearly 40-60 % at all reaction temperatures, which suggests that reaction is initially fast and possibly first-order.

To assess loss of material to non-condensable gases, Figure 2.2b shows the mass fraction of liquid product recovered relative to feed mass (Equation 2.2) with change in vapor residence time at the pyrolysis temperatures of our experiments. These values include unconverted feed, which is especially significant at low vapor residence time (τ <

6 s) (refer to Figure 2.2a). It is interesting to note that at even very short reaction residence time (τ = 1 s), nearly 7-10 % of feed mass is lost. Previous studies have demonstrated that first step in triglyceride pyrolysis is the release of the glycerol backbone and its conversion to non-condensable gases [34]. From our results, this first step appears to be rapid and seems to occur within 1 s of reaction. After 1 s, loss of products as gases remains nearly constant until 60 s. Thereafter, at high vapor residence time (τ > 60 s) the yield of recovered liquid decreases, likely due to further decomposition of products.

Figure 2.2: (a) Conversion of feed to non-glyceride products (see Equation 2.1), (b) yield of liquid recovered (see Equation 2.2), and (c) yield of non-glyceride liquid products (see Equation 2.3) versus vapor residence time at the pyrolysis temperatures. Dashed line represents the theoretical yield of non-glyceride liquid products.

Figure 2.2c shows the effects of τ and temperature on the yield of non-glyceride liquid products (equal to mass of total liquid recovered minus the mass of glyceride in liquid product; see Equation 2.3). For instance, at the experimental condition of 500-6, the 28

total mass fraction of recovered liquids is 87% (relative to feed mass, Figure 2.2b) but contains 3% of unreacted triglyceride (Figure 2.2a), therefore the yield of non-glyceride liquid product for this experiment would be 84% (Figure 2.2c). As observed from Figure

2.2c, the yield of non-glyceride liquid product increases most significantly between 0-6 s, but decreases at high vapor residence time (τva p> 60 s). At τ=1 s, although yield of liquid recovered is high (Figure 2.2b), conversion of feed is low (Figure 2.2a) and thus yield of non-glyceride liquid product is low (Figure 2.2c). At τ between 6-60 s, due to high yield of liquid recovered (Figure 2.2b) and conversion (Figure 2.2a), high yield (>85%) of non- glyceride liquid product was achieved.

When the glycerol backbone is released during triglycerides pyrolysis (the first step

-1 in triglyceride decomposition), acrolein (C3H4O; MW=56 g gmol ) is produced (one mole per mole of triglyceride) which then decomposes to non-condensable C2H4 and CO [53,

55, 74]. Given this reaction mechanism, the theoretical yield of liquid from soybean oil pyrolysis was calculated to be approximately 93% (w/w) by discounting the estimated mass of acrolein produced from the feed mass. An average soybean oil molecular weight of 875 g gmol-1 was used for these calculations. In Figure 2.2c, the dashed line represents this theoretical maximum liquid product yield. When pyrolysis was performed under the reaction conditions of 6 s < τ < 60 s and 475

(when τ < 60 s) suggest that secondary reactions were sufficiently slow such that pyrolysis products remained as condensable liquids but did not significantly decompose to non- condensable gaseous products. 29

At τ > 60 s and T ≥ 475 °C, yield of non-glyceride liquid product decreased with temperature (except at 450 °C) likely due to further breakdown of non-glycerides to non- condensable gases at higher reaction temperature and residence time. Interestingly, at reaction temperature of 450 °C, yield of non-glyceride liquid product continued to increase over τ between 1-300 s suggesting that secondary reactions (which cause further cracking of liquid product and gas formation) are even slower (or absent) at low reaction temperature.

In our pyrolysis system, liquid yield from pyrolysis of lipids were significantly improved (as high as 88%) relative to previous studies (20-70%) [34]. The high yields are likely due to the rapid volatilization of the feed and the short vapor residence time, which decreases the chance for pyrolysis products to undergo secondary reactions. Our experiments show that in order to achieve high liquid product yield, the optimal vapor residence time should be between 6-60 s. To reach such a short vapor residence time, rapid vaporization of feed is required. In our pyrolysis system, due to atomization of vegetable oil, small droplets of oil could vaporize and reach the reactor temperature rapidly, which allowed the pyrolysis to be carried out at a short vapor residence time.

2.4.4 Products analysis

Figure 2.3 shows a general schematic of the principal steps involved in thermal cracking of vegetable oil as described previously in the literature [49, 50, 53, 55], and also is evident from the present work. As discussed in the previous section, the first step in thermal cracking of triglycerides is the disintegration of the glycerol backbone.

Subsequently, the long-chain fatty acids (L-FAs) can further degrade via two parallel 30

pathways. In one case, they might decarboxylate and/or decarbonylate and produce heavy hydrocarbons (Hy-HCs) (reaction 3). Alternatively, the L-FAs can degrade at the sites of the C=C bonds and form short-chain fatty acids (S-FAs) and light hydrocarbons (Lt-HCs)

(reaction 4). The Hy-HCs produced from reaction 3 can also subsequently crack at the C=C sites and form Lt-HCs (reaction 5). In parallel, the S-FAs from reaction 4 can release CO2 and/or CO due to decarboxylation and/or decarbonylation and produce Lt-HCs (reaction

6). Aromatics and cyclo-compounds can also form from the Diels-Alder reaction of dienes and alkenes [53] (reaction 7). Finally, unsaturated Lt-HCs also may break down further

(due to C=C bond) and produce C1-C4 uncondensable hydrocarbons (reaction 8). In Figure

A2 (Appendix A), we have illustrated the triglyceride pyrolysis reaction pathways for the thermal cracking of trioleate and show the C-C bond cleavage, decarboxylation/decarbonylation and Diels-Alder reaction pathways for this model triglyceride. Overall, the mechanisms shown in Figures 2.3 and A2 suggest that due to the complexity of the cracking reactions, a diverse distribution of chemicals constituents (such as L-FAs, S-FAs, Hy-HCs, Lt-HCs, G, cyclic and aromatics) can be produced during pyrolysis. The mechanisms also suggest that lighter products (secondary reactions) are progressively formed with an increase in residence time. If the residence time is excessive, it is expected that a large proportion of the feed would form gases. The mechanism further suggests that with optimization of reaction temperature and residence time, the formation of non-condensable gases can be minimized. It might also be possible to partially control product compositions (low MW versus high MW) through changes in reactor temperature and residence time.

Figure 2.3: A general thermal cracking mechanism of soybean oil: (1) initial cracking of triglyceride that generates free radical RCOO· or/and RCO· (2) gases produced as the result of loss of the glycerol backbone (3) decarboxylation and/or decarbonylation of long chain fatty acids, followed by hydrogenation and/or dehydrogenation that produces heavy hydrocarbons (4) C-C double bond cleavage of long chain fatty acids to produce short chain fatty acids and light hydrocarbons (5) long chain hydrocarbons breakdown due to cleavage of unsaturated C-C bonds to produce light hydrocarbons (6) decarboxylation of short chain fatty acids to produce light hydrocarbons (7) Diels-Alder reaction (reaction between a diene and alkene) to produce cyclic compounds and/or aromatics (due to dehydrogenation of cyclic compounds) (8) breakdown of light hydrocarbons to non-condensable gases (e.g. ethylene elimination). TG: triglyceride, L-FAs: long-chain fatty acids, G: gases, Hy-HC: heavy hydrocarbons, S-FAs: short-chain fatty acids, Lt-HC: light hydrocarbons, Ar: aromatic, and Cyc: cyclic compounds.

Product composition from pyrolysis reactions were estimated by GC analysis of the recovered liquid. Figure 2.4 shows the content of L-FAs (C16-C18; Figure 2.4a), S-FAs

(C6-C11; Figure 2.4b), Hy-HCs (>C12; Figure 2.4c), and Lt-HCs (C5-C12; Figure 2.4d) in the recovered liquids. Identification and quantification of hydrocarbons (heavy and light), and L-FAs were relatively straightforward. Concentrations of these components were estimated from GC-FID peak areas by comparison with calibration curves of known standards. However, direct identification and quantification of S-FAs was more difficult because these compounds showed low sensitivity on both GC-FID and GC-MS. Therefore,

derivatization of fatty acids to fatty acid methyl esters (FAMEs) was performed to improve sensitivity as suggested by Kubatova et al. [56]. Even after derivatization, identification and quantification of S-FA methyl esters was difficult using GC-FID, because of close peak proximity with Lt-HCs. GC-MS, however allowed a positive identification of the S-

FAs methyl esters. Therefore, S-FAs were quantified by GC-MS by comparison of peak areas with calibration curves of S-FA methyl ester standards.

From Figure 2.4, one can observe that the concentrations of L-FAs (Figure 2.4a) in the products first increased (1 s < τ <6 s), and then decreased (τ > 6 s). At τ < 6 s, the conversion of triglycerides is incomplete (refer to Figure 2.2a) and L-FAs (that formed

from triglycerides degradation) likely further decompose to only a small extent due to the short residence time. In other words, the rate of formation of L-FAs is higher than the rates of their decomposition when τ < 6s. For instance, at 450 ° C and τ =1 s, although only

~28% of non-glyceride liquid products were formed, the L-FA fraction of the non- glyceride liquid products was nearly 70%. However, after 6 s, when nearly complete triglyceride conversion was achieved (Figure 2.2a), the produced L-FAs would be expected to decompose via reactions 3 and/or 4 (Figure 2.3), when allowed to stay in the reactor longer. Thus, the L-FAs concentrations decreased when τ > 6 s (Figure 2.4a). Interestingly, oleic acid remained the major fatty acid component of the L-FA fraction, while linoleic acid was only 5% of the total recovered liquid (Table A2; Appendix A). In contrast, the feed contained ~50% linoleic and linolenic acids (refer to Section 2.4.2). The enrichment of oleic acid in the pyrolysis products indicates that linoleic acid (C18:2) and linolenic acid

(C18:3) break down more readily than oleic acid (C18:1).

From Figure 2.4b, it can be observed that S-FA concentrations increased consistently with vapor residence time. As discussed in Section 2.4.2 and in the previous paragraph, fatty acid chains in soybean oil are largely polyunsaturated (linoleic and linolenic acid) and prone to C=C cleavage [34]. These decompositions are expected to result in formation of S-FAs and Lt-HCs (Figure 2.2; reaction 4). Table A3 (Appendix A) shows the composition of S-FAs in the recovered liquid. From these data, it can be observed that the majority of the produced S-FAs were saturated (such as hexanoic, heptanoic, octanoic, nonanoic, and decanoic acid). In addition, the yields of nearly all S-

FAs increased at higher vapor residence time and reaction temperature most likely because the cleavage of C=C bond is an endothermic reaction. Since fatty acid chains in the parent 34

triglyceride are predominantly unsaturated, it is reasonable to presume that these S-FAs were produced by C=C cleavage. For instance, oleic acid contains one double bond at C9 and linoleic acid has an additional double bond at C12. Cleavage of the C-C double bond

(from L-FAs) at the α, β position relative to the C9 unsaturation would produce heptanoic acid, and similar cleavage relative to C12 would produce decenoic acid. Thereafter, hydrogenation of decenoic acid (due to hydrogen generation from glycerol decomposition) could form decanoic acid. Heptanoic, decanoic and decenoic acid were the major S-FAs compounds observed in our products.

As observed in Figure 2.4c, the concentrations of Hy-HCs in liquid products also increased at 1 s <τ <60 s, but decreased at higher reaction times. These C12 and higher hydrocarbons were likely produced from decarboxylation and/or decarbonylation of L-FAs along with partial cracking of the fatty acid chain. The Hy-HCs that have unsaturated C-C bonds can break down and produce Lt-HCs. Like S-FAs, the content of Lt-HCs increased with τ and Trxn. Formation of Lt-HCs is thermodynamically more favored at higher Trxn due to the endothermicity of cracking (C=C cleavage) and decarboxylation reactions.

Table A4 (Appendix A) shows the mass of the identified and unidentified chemical compounds in the recovered liquid as a fraction of feed mass. Production of non- condensable gases (such as C1-C4 hydrocarbons, H2, CO2 and CO [53, 75]) was calculated by subtracting liquid product mass from feed mass. The mass of unidentified components was then calculated by subtracting the identified liquid products and uncondensed gases from feed mass. From Table A4, it can be seen that at τ = 1 s, nearly all liquid products were identified. However, for experiments at higher τ identification of all products was

difficult, due to the vast number of small peaks that were present on the GC-FID chromatogram.

Between 10-25% of the products remained unidentified in the recovered liquid with higher τ. Other studies have also reported similar mass closure of identifiable products

(typically 30-80%) [28, 53, 56]. At higher τ, secondary reactions are enhanced and the concentrations of Lt-HCs increased. These Lt-HCs consisted of several components. For instance, the detailed results of hydrocarbons identification (as described in Section 2.4.5) showed 360 distinct peaks on the GC-FID chromatogram. Although each unidentified compound was likely produced at a low concentration, the cumulative effect is significant.

A majority of the unidentified products were in the range of Lt-HCs (GC column elution temperature < 90 °C).

To assess the products at an extreme reaction condition, soybean oil was pyrolyzed at 550 °C and τ =300s. Under this severe reaction condition, liquid products likely underwent extensive cracking to small molecules which resulted in a high (50%) production of uncondensed gases (Table A4, Appendix A). The high content of Lt-HCs

(23.1% of feed) under these conditions is likely due to break down of unsaturated Hy-HCs, decarboxylation of S-FAs, and degradation of L-FAs. As a result, low mass of fatty acids

(long- and short-chain) as well as Hy-HCs was observed.

2.4.5 Product distillation and detailed hydrocarbon analysis

Products of soybean oil pyrolysis at Trxn=500 °C and τ = 60 s were distilled to separate free fatty acids from hydrocarbons, since a high yield of non-glyceride liquid products (~85%; Figure 2.2c) was achieved at these experimental conditions. Distillation 36

was performed under vacuum (60 mm Hg) to minimize thermal degradation (if any) of liquids during the separation. The goal of distillation was to separate L-FAs from other components. To estimate the appropriate distillation temperature, boiling point (b.p.) of hydrocarbons and fatty acids were estimated at the vacuum conditions employed. The b.p. of heptadecane (highest carbon-number hydrocarbon in our liquid products) at 60 mm Hg was estimate to be 200 °C from extrapolation of its normal b.p. using the correlation provided by Maxwell and Bonnell [76]. For fatty acids, the b.p. values at 10 mm Hg from

Cermak et al. [77] were extrapolated to a pressure of 60 mm Hg [76]. From these correlations, the b.p. of oleic and palmitic acid were estimated to be 271 and 261 °C, respectively. Since there was sufficient difference in b.p.s of hydrocarbons and L-FAs at

60 mm Hg, vacuum distillation was performed at 200 °C.

Vacuum distillation was carried out for 4.5 h (including 1.5 h for heating up to distillation temperature). After 4.5 h, no further boiling of the liquid phase was visually apparent. Thereafter, the distilled fraction (DF) and the residue fraction (RF) were analyzed using GC-FID. Figure A3 (Appendix A) shows the GC-FID chromatogram of the DF.

Identification of compounds was performed by matching retention time of analytes with standard mixtures (listed in the Section 2.3.1) and was confirmed by GC-MS through a comparison of MS spectra of analytes with corresponding entries from the NIST 2008 database. The GC-FID analysis detected 360 peaks with relative concentration above

0.01% in the distilled fraction. Since quantification of all individual compounds using their corresponding standards is difficult, chemicals were quantified using relative peak areas.

Table A5 (Appendix A) shows the more abundant compounds (with concentrations approximately above 0.1%) that represent nearly 85% of the DF. As observed, different 37

classes of hydrocarbons were present in the products. Light hydrocarbons in the range of gasoline (C5-C12) comprise almost 65% of the DF. The major compounds identified are C5-

C17 alkanes and alkenes, aromatics including benzene, toluene, ethylbenzene, propylbenzene, buthylbenzene and xylene, and cyclic compounds such as cyclooctane and cyclopentane methylene. Table 1 shows the relative mass of aromatics, olefins, paraffin and cyclic compounds in the DF. While aromatic compounds improve the octane number of a fuel, their use in gasoline is strictly regulated by EPA air quality standards [33]. Our

DF contains 13% aromatic, which is below the regulated value (maximum 20%).

Formation of aromatics can occur due to Diels-Alder reaction and/or intramolecular radical cyclization. In Diels-Alder reaction, a diene and alkene reacts and form polysubstituted cyclohexenes and then via hydrogenation polysubstituted cyclohexanes and via dehydrogenation polysubstituted aromatic produce [78]. It can also be noted that majority of olefins and paraffins are unbranched. Dienes (e.g. undecadiene, tridecadiene, and tetradecadiene) were also present in the DF.

Table 2.1: Relative mass of different classes of compounds in the distilled fraction (DF).

Classes Wt. (%) Paraffin (C5-C17) 22.9 Olefin (C5-C17) 32.1 Aromatic 13.0 Cyclic compounds 8.9 Dienes 8.5 Unidentified 14.6 Total 100

Figure A4 (Appendix A) shows the GC-FID chromatogram and Table A6

(Appendix A) shows the composition of residual fraction (RF) of the distillation. It can be

clearly seen that C18 fatty acids were the major (59%) components of the RF. In addition, much smaller amounts of C16 fatty acids, and C13-C17 hydrocarbons were also observed in the RF. In contrast with GC-FID results of the DF (Figure A3, Appendix A), the analysis of the RF (Figure A4, Appendix A) showed no significant peak prior to a retention time of

100 min. This indicates that the simple distillation (that was used in our experiment) could separate most hydrocarbons from L-FAs. Although the RF contained small amounts of C13-

C17 hydrocarbons (~11%), with a better control of distillation conditions, more purified L-

FAs could possibly be recovered.

S-FAs in DF and RF were identified and quantified separately since derivatization of S-FAs was needed to improve the sensitivity of detection by GC-FID (as described in

Section 2.4.4). The procedure for derivatization was similar to transesterification of soybean oil, which was explained in Section 2.3.5. Derivatized S-FAs were quantified by comparison with FID response of known methyl ester standards. Figure A5 (Appendix A) shows the GC-FID chromatogram of the derivatized DF and RF. The composition of S-

FAs in DF and RF is given in Table A7 (Appendix A). From these data, S-FAs comprised

~7% of the DF with heptanoic acid (C7-FAs) as the major S-FAs. Also, S-FAs (mainly

C10-FAs) comprised ~9% of the RF. It is important to mention that the GC-FID analysis of the DF (Figure A5a, Appendix A) did not show any peak correspond to L-FAs. This indicates that all L-FAs (of the liquid products at experiment 500-60) remained in RF.

Our distillation results show that pyrolysis products could be relatively easily separated into hydrocarbon-rich and L-FAs-rich fractions. The hydrocarbon containing fraction could then be converted to fuels or further distilled to recover aromatics.

2.4.6 Feasibility of isolating oleic acid from soy oil via pyrolysis

Analysis of products from vegetable oil pyrolysis suggested that oleic acid was relatively thermally-stable (Table A2, Appendix A). To better understand the thermal degradation of the mono-unsaturated oleic acid under more controlled conditions (in the absence of other poly-unsaturated fatty acids), pyrolysis of enriched oleic acid was performed. The experimental set-up was kept similar to pyrolysis of soybean oil in order to compare the results of oleic acid and soybean oil pyrolysis. The feed contained small amounts of saturated fatty acids- C16:0 (~4%), C14:0 (~3%) and C17:0 (~2%); oleic acid was ~85% of the feed. Pyrolysis was carried out at 450 °C and τ of 60 s and 300 s

(conditions that showed low degradation of oleic acid during soybean oil pyrolysis).

Identification of feed and pyrolysis products was performed using GC-MS, and quantification of components was carried out via GC-FID. Figure A6 shows the GC-MS chromatogram of feed (Figure A6a, Appendix A), and pyrolysis products at 450 °C and τ of 60 s (Figure A6b, Appendix A) and 300 s (Figure A6c, Appendix A). Product composition at 450 °C and τ=60 s is nearly same as the feed (Table A8, Appendix A), which indicates that the extent of degradation was small under these conditions. The heptadecene (2.9%) was likely produced due to decarboxylation of oleic acid. Even at τ

=300 s, a large portion of the pyrolysis products was still oleic acid (~63%), and only

~4.3% decanoic acid (due to C-C cleavage) was observed in addition to heptadecene

(5.7%).

Overall, pyrolysis results of oleic acid demonstrate that oleic acid is thermally stable and can be largely recovered in its native fatty acid form after pyrolysis. In contrast,

poly-unsaturated fatty acids are much more thermally labile and undergo extensive degradation, as evidenced by the low concentration of these in the products of soybean oil pyrolysis (Table A2, Appendix A). From Table A8 and Figure A6 (Appendix A), approximately 25% of oleic acid degraded into smaller molecules at 450 °C and τ of 300s.

However, when soybean oil was pyrolyzed under similar conditions, a slightly greater amount (~33%) of oleic acid (from triglyceride) was measured to be degraded (Table A2,

Appendix A). The higher degradation of oleic acid in the triglyceride feed could be attributed to more free radical formation due to extensive degradation of the thermally labile linoleic/linolenic acid, which may have contributed to more extensive cracking of oleic acid. Moreover, in pyrolysis of triglycerides, the produced long chain fatty acids (as a result of glycerol backbone release) are in the form of free radicals (RCOO.) and perhaps these highly reactive fatty acid radicals are more prone to crack than oleic acid in a whole free fatty acid form. Oleic acid is an important precursor in oleochemicals [79-81] and our results suggest that pyrolysis could be used for high yield recovery of this fatty acid from oleic acid-rich feedstocks such as sunflower oil.

2.5 Conclusions

Pyrolysis of soybean oil was performed to convert it into fuels and chemicals. An atomizer was used to introduce the oil into the reactor in the form of micron-sized droplets to rapidly volatilize oil and achieve short vapor residence time. As a result, near-theoretical yields of liquid products (up to 88% relative to feed mass) were achieved. The liquid products were comprised of light- (C5-C12) and heavy- (>C12) hydrocarbons in addition to

short- (C6-C12) and long- (C16-C18) chain fatty acids. Distillation of liquid products resulted in distinct hydrocarbon-rich and long-chain fatty acids-rich fractions. The hydrocarbon- rich fraction contained paraffin (23%), olefin (32%), aromatic (13%), cyclic compounds

(9%) and dienes (5%).

In contrast with transesterification that produces biodiesel but does not produce drop-in hydrocarbon fuels, pyrolysis can generate multiple fuels/fuel precursors (e.g. gasoline, jet fuel and diesel) and chemicals (e.g. olefin and aromatics as primary petrochemicals; fatty acids for oleochemicals). In addition, low quality oils with high free fatty acid content are also suitable as pyrolysis feedstocks, but would not be usable for biodiesel production. Thus far, low yield of liquid products has remained the major obstacle for commercialization of vegetable oil pyrolysis since feedstock price is the dominant cost for fuel production from triglycerides. In this chapter, we have demonstrated that our simple reactor design allows for near-quantitative yield of products. Further, since the design is based on the widely used industrial practice of atomization, the reactor system is easily scalable for commercial use and would allow the implementation of pyrolysis as a viable alternative to transesterification for production of fuels and chemicals from vegetable oils.

Chapter 3

High Yield Production of Hydrocarbons from Non-

edible Oils Through Reactive Pyrolysis System

3.1 Abstract

We report the high yield production of hydrocarbons and aromatics from non-edible oils in a novel reactive pyrolysis system. In this system, the oil feed is atomized and injected into a hot catalytic reactor. The micron-size droplets rapidly volatilize in the hot reactor to facilitate high-efficiency vapor phase reactions over a homogeneous catalyst rather than reactions in the liquid phase that are prone to polymerization and coke/char formation. In this chapter, conversion of triglyceride-based feedstock such as pennycress, camelina and waste cooking oil was performed on a reactive pyrolysis system and in presence of zeolite catalyst to produce high yields of aromatics. As such, at reaction temperature of 500 °C and WHSV of 12 h-1 an organic liquid products (OLP) yield of 71 wt. % was obtained, where the OLP contained 53.5% aromatics (mainly benzene, toluene and xylene; BTX),

17.6% C16-C18 fatty acids and 7.8% aliphatic. Impacts of reaction temperature, residence time, reaction time on stream and catalyst regeneration on products yields and composition were investigated. Finally, the BTX were separated from other hydrocarbons in the liquid

products. This approach is promising to achieve high liquid yields when feedstock is hardly volatile and thermally labile such as triglyceride, bio-oil from lignocellulose biomass or crude-oil from microalgae.

Key words: Non-edible oil, Pyrolysis, Triglyceride, Zeolite and BTX

3.2 Introduction

Energy-dense triglycerides from oilseeds or microalgae are a good source of hydrocarbon precursors for fuels, chemicals and polymers [24-28]. However, resource availability and negative societal impacts (i.e. food versus fuel or chemical considerations) have to be considered. Among natural oils, non-edible oils do not compete with food supplies due to presence toxic hydroxyl fatty acids, ricin and glucosinolates [82, 83]. In the

US, some varieties of non-edible oil seed producing plants (e.g. pennycress and camelina) are being actively developed by the USDA as “off season rotation crops” that can be grown in rotation with traditional food crops [84-87]. Waste cooking oil (plant oil used in food preparation and is no longer suitable for human consumption) is another attractive source of non-edible oil. It is estimated that the approximately 10 million metric tons of waste cooking oil (WCO) is generated in the US per year [88]. Improper disposal of WCO can lead to clogging of sewer lines and wastewater treatment plant upsets. In many urban areas in the US, regulatory requirements and/or government incentives have allowed “grease collection” by third party vendors [89]. In addition to restaurants and cafeterias, other large sources of WCO are commercial fried food operations (e.g. potato chips manufacturing).

After collection, the oils are taken by biodiesel manufacturers [89], or disposed. WCO has

different physical and chemical properties than virgin vegetable oil due to the thermolytic, oxidative, and hydrolytic reactions that occur during frying of virgin oil. As a result, WCO contains impurities such as free fatty acids, water and hydrocarbons that decrease the efficiency of biodiesel production through transesterification reaction and necessitate chemical and/or energy intense pretreatment [41, 90].

Catalytic pyrolysis is an attractive pathway for conversion of triglyceride-based feedstock to the products that can be fully compatible with the existing petroleum infrastructure. Typically, the catalysts promote triglyceride degradation and deoxygenation to hydrocarbons [45-48, 91]. The product yields and compositions from triglyceride catalytic pyrolysis are dependent upon reaction conditions (e.g. temperature and residence time) and catalyst properties (e.g. acidity, pore size and shape selectivity). Idem et al. [92] performed cracking of canola oil at atmospheric pressure and temperatures of 400-500 °C over a slew of catalysts - HZSM-5, silicalite, silica, silica-alumina, γ-alumina, calcium oxide and magnesium oxide. They concluded that non-shape selective catalysts (e.g. silica, silica-alumina and γ-alumina) promote secondary reactions and produce substantial amount of gaseous products. In contrast, catalysts with high shape-selectivity (e.g. HZSM-

5 and silicalite) produced higher organic liquid product (OLP) as well as aromatics.

However, basic sites in the catalyst (e.g. calcium oxide and magnesium oxide) strongly inhibited cracking of oil and resulted in significant production of high molecular weight residual oils.

Zeolites have generally been regarded as appropriate catalysts for cracking of vegetable oils due to their crystallinity, well-defined pore structures, large surface area and strong acidity [93, 94]. In the literature, the use of several zeolite catalysts such as HZSM- 45

5, zeolite β, USY zeolites and hybrid of HZSM-5-zeolite β and HZSM-5-USY has been reported for triglyceride pyrolysis [95, 96]. Of these, HZSM-5 has shown better oil conversion due to high Brønsted acid sites; however, the OLP yields have remained low.

For instance, Katikaneni et al. [97] studied the conversion of canola oil over HZSM-5 and showed that quantitative conversion was obtained at 500 °C and weight hourly space velocity (WHSV) of 1.8 h-1. However, only 41% of feed was collected as OLP while nearly

50% of the feed mass was converted to non-condensable gases and 5% of the feed was converted to coke. Although Katikaneni et al. attempted to crack the canola oil in presence of steam to eliminate coke formation, the OLP yield decreased to 23%. Srinivas et al. [98] used HZSM-5, silica-alumina and a mixed catalyst containing HZSM-5 and silica-alumina to upgrade the bio-oil derived from fast pyrolysis of maple wood. A maximum OLP yield of 27% was achieved over HZSM-5 at reaction temperature of 370 °C and WHSV of 7.2 h-1. Botas et al. [99] reported a 46% OLP yield from cracking of rapeseed oil in a fixed bed reactor at 550 °C and WHSV of 7.7 h-1 using HZSM-5.

In an attempt to improve the OLP yield from catalytic cracking of vegetable oil, mesoporous catalyst MCM-41 was used [100, 101]. Although gas yield decreased due to larger pore size (relative to HZSM-5), low conversion of feedstock was achieved likely due to low acidity and shape selectivity of MCM-41. Ooi et al. [102] also reported similar results over mesoporous catalyst SBA-15, which has similar pore structure as of MCM-41.

In order to marry the advantages of both microporous zeolite and mesoporous molecular sieve, Twaiq et al. [103] performed palm oil cracking over composite of HZSM-5/MCM-

41. The HZSM-5 composite of 10 and 20 wt.% MCM-41 resulted in nearly same conversion, OLP and coke yield as those obtained using pristine HZSM-5. However, the 46

conversion as well as gasoline yield decreased when more than 20% of the composite catalyst was a mesoporous phase, likely due to the low activity of amorphous MCM-41.

While pyrolysis is a promising approach to produce fuels and chemicals from triglyceride feedstock, the key bottleneck of the inability to achieve high liquid yields has remained the major obstacle for commercialization. Current pyrolysis practices result in excessive decomposition and polymerization in the liquid/vapor phase due to long residence time and slow volatilization. Akin to the concepts of fast pyrolysis of solid substrates (e.g. biomass and coal) [57, 104], pyrolysis yields are highest when liquid phase reactions are minimized and reaction residence time is shortened through rapid volatilization. As such, we designed and built a novel continuous reactive pyrolysis system, where feed was injected through an atomizer into the reactor in the form of micron-sized droplets. This approach allows the oils to vaporize rapidly and mitigates the excessive unfavorable reactions that occur in the liquid phase. Further, this approach facilitates vapor-phase catalytic reactions when the volatilized feedstocks pass through a heterogeneous catalyst bed. Besides, the reactive pyrolysis system has the ability to process diverse liquid feedstocks (e.g. triglycerides, free fatty acids, bio-oil from lignocellulose biomass or crude-oil from microalgae) and to carry our diverse reaction chemistries (e.g.

ammonization in presence of NH3 and hydrocracking in presence of H2) by selecting suitable catalysts and reactants. In this report, diverse triglyceride-based feedstocks (edible, nonedible and waste oil) were used to produce high-yield and high-selectivity of renewable aromatics via a vapor phase reactive pyrolysis system. The impacts of reaction conditions such as temperature, residence time and “reaction time on stream” were assessed to maximize target aromatics (benzene, toluene and xylene; BTX) that are essential in 47

petrochemical industries. Moreover, several reaction-regeneration cycles were carried out and the products yields and compositions were analyzed to evaluate the long-term reusability of catalyst.

3.3 Experimental

3.3.1 Materials

Soybean and camelina oil was obtained from Zoyeoil (Zeeland, MI, USA) and Bulk

Apothecary (Aurora, Ohio, USA), respectively. Waste cooking oil (WCO) was provided by The University of Toledo cafeteria. Pennycress oil was recovered from ground seeds by extraction with hexane for 24 h in a Soxhlet apparatus followed by evaporation of hexane in a rotary evaporator under reduced pressure.

Hexane, chloroform, methanol, sulfuric acid and enriched oleic acid were purchased from Fisher Scientific (Pittsburgh, PA, USA). Analytical standards for fatty acids (haxanoic, octanoic, palmitic, stearic, oleic, linoleic and linolenic acid), glycerides

(triolein, diolein, and monolein), FAMEs (mixtures of C8-C22 FAMEs), alkanes (C5, C6,

C7, C8 and mixtures of C7-C30), olefins-(Alphagaz PIANO), aromatics-(Alphagaz PIANO), mixtures of benzene, toluene, ethylbenzene and xylene (BTEX), gasoline, jet fuel and naphthalene were purchased from Sigma-Aldrich (St. Louis, MO, USA).

NH4-ZSM-5 powder with a SiO2/Al2O3 molar ratio of 23 was purchased from

Zeolyst International, USA. The NH4-ZSM-5 sample was calcinated in a muffle furnace for 5.5 h at 550 ºC to obtain HZSM-5. The texture and acid properties of the HZSM-5 are

listed in Table B1 (Appendix B) and a detailed description of HZSM-5 characterization is provided elsewhere [105].

3.3.2 Experimental set-up

Catalytic pyrolysis of vegetable oil was carried out in a continuous reactor system schematically shown in Figure 3.1. The reactor tube was packed with desired amount of catalyst (1-5 g) supported by stainless steel disc (10 µm pores) and quartz wool. Prior to each experiment, the reactor was placed in the furnace and heated while being purged with pure N2 to remove any traces of H2O/O2 from the system. After the reactor reached set point temperature, the N2 purge was stopped, and triglyceride-rich feedstock was introduced into the reactor in the form of micron-size droplets using an atomizer (Sonazop,

Farmingdale, NY, USA; model: HTNS40). The atomizer consists of an ultrasonic probe with a 4 mm diameter, which was operated at constant frequency of 40 kHz. The atomizer used in our reactor system, creates extremely large surface area for heat transfer and allows rapid evaporation of feedstock. A detailed description of the atomizer is given elsewhere

[106]. The system was operated in continuous mode for 60 min and the products from the reaction were recovered in a condenser. A gas bag (made of Tedlar® PVDF fluoropolymer film; Sigma-Aldrich, MO, USA) was placed at the end of the condenser to collect the non- condensable gases. At the end of the experiment, the reactor furnace was turned off and allowed to cool to room temperature. Thereafter, the liquids from condenser were collected and weighed on an analytical balance (Mettler Toledo, USA) with ± 0.1 mg accuracy. The liquid product was separated in a separatory funnel into aqueous and organic phases. The

catalyst was also recovered from the reactor to and the mass of coke formed on the spent catalyst was estimated by thermogravimetry as described in Section 3.3.4.

Non-condensable GC-TCD gas

Gas

Coolant Temperature out T1 Triglyceride Controller

Furnace

Condenser Atom e Evap 2 1 2 Products Coolant Pump in Furnace iquid GC-FID Products GC/MS Schematic diagram of catalytic pyrolysis of triglyceride. 1: catalyst (HZSM- 5); 2: quartz wool (as support for catalyst); Evap: stands for evaporation of feedstock.

Conversion of triglyceride to non-glyceride products was calculated as

(W ) −(W ) Conversion (%) = TG 푖푛 TG out × 100 (3.1) (WTG)푖푛 where, (WTG)𝑖푛 is the mass of oil introduced into the reactor and (WTG)out is the mass of glycerides in collected liquid.

The yield of organic liquid products (OLP), aqueous phase and coke was calculated as

w Yield (%) = i × 100 (3.2) (WTG)푖푛 where wi is the mass of OLP, aqueous phase and coke.

3.3.3 Experimental design

The impact of reaction temperature and WHSV on triglyceride conversion, products yield and composition from catalytic pyrolysis of vegetable oil were investigated.

In our experiments, reaction temperatures of 450, 475, and 500 °C were selected to investigate temperature effects. WHSV can be adjusted by changing feed mass flow rate or catalyst loading inside the reactor. However, changing feed mass will affect the vapor residence time, as described in our previous report [106]. Moreover, the previous experiments on pyrolysis of soy oil indicated that vapor residence time (τ) of 6-60 s is the optimum range to achieve quantitative triglyceride conversion and high liquid yields [106].

To eliminate the confounding effects of varying residence time, τ was kept constant (10 s) by injecting a constant feed mass flowrate of 1 g min-1. The WHSV of 60, 30, 20 and 12 h-

1 was achieved by loading the reactor with 1, 2, 3 and 5 g of HZSM-5, respectively. A full factorial design with fifteen runs was employed.

3.3.4 Products analysis

Feedstock (pennycress, camelina, soy and waste cooking oil) was transesterified to

FAMEs and analyzed by gas-chromatography using mass spectrometry (GC-MS) and flame ionization detection (GC-FID) to identify and quantify the fatty acid constituents in the feedstock, as explained elsewhere [106].

GC-MS and GC-FID were employed to identify and quantify liquid products from catalytic and non-catalytic pyrolysis of triglyceride. Table 3.1 shows a summary of analytical techniques used to analyze feedstocks and liquid/gas products. All chemical

concentrations were estimated based on calibration curves developed using corresponding external analytical standards. All samples were analyzed two times and the mean value is reported. Chemical compounds corresponding to the GC-MS chromatogram peaks were identified using NIST2008 mass spectral database. A minimum 70% confidence level was used as a threshold for positive identification of IDs provided by the spectral analysis software.

To identify and quantify non-condensable gases produced from catalytic pyrolysis of triglycerides, a GC (Shimadzu 2012 plus) equipped with a thermal conductivity detector

(TCD) was used. A Carboxen-1010 PLOT (Supelco Inc, PA, USA) column was employed

(30 m length, 0.32 mm ID) to separate gases. The injector and detector temperatures were

200 and 230 °C, respectively. Helium was used as the carrier gas with a flowrate of 3 mL min-1. The column temperature was initially set at 35 °C for 8.5 min, and was subsequently heated at a temperature ramp rate of 24 °C min-1 to 250 °C; this final temperature was maintained for 5 min at the end of the run.

A thermo-gravimetric analyzer (TGA; Model: SDT Q600 series analyzer, TA

Instruments, Schaumburg, IL) used to measure the amount of coke deposited on the catalyst. Catalyst was heated from room temperature to 550 °C (temperature ramp rate of

-1 10 °C min ) and held for 4 h under O2 atmosphere until no further weight loss were observed. The weight loss during the TGA experiment corresponds to coke combustion.

Table 3.1: Analytical techniques used to characterize feedstock and liquid/gas products.

Analytical Column type Application Quantification/identification method method Analysis of triglycerides, RTX- diglycerides, monoglyceride and FAME standards, fatty acid biodiesel1 fatty acids in the pyrolysis liquid standards, monoolein, diolein and products and feedstock triolein

GC-FID derivatized FAMEs External analytical standards: Quantification of hydrocarbons in alkane (C , C , C , C and DB-Petro2 5 6 7 8 the liquid products mixtures of C7-C30), olefins, aromatics, naphthalenes Identification of the chemical compounds in the pyrolysis liquid NIST2008 mass spectral GC-MS DB-5MS3 products; identification and database; short-chain FAMEs quantification of short-chain fatty standards acids in the pyrolysis products Identification: retention time Carboxen- Identification/quantification of match with gas standards GC-TCD 1010 PLOT4 non-condensable gases Quantification: calibration curve 1 Column properties: 15 m length, 0.32 mm ID, and 0.1 µm film thickness. Oven temperature program: 60 °C (1 min), 10 °C min-1 to 370 °C (5 min). Injector temperature: 370 °C; -1 Carrier gas: H2 with a flowrate of 6.02 mL min and a split ratio of 1:10. Detector: FID, 370 °C 2 Column properties: 50 m length, 0.25 mm ID, and 0.5 µm film thickness. Oven temperature program: 35 °C (15 min), 1 °C min-1 to 60 °C (20 min), 2 °C min-1 to 300 °C (2 min). Injector temperature: 300 °C; -1 Carrier gas: H2 with a flowrate of 1 mL min and a split ratio of 1:10. Detector: FID, 300 °C 3 Column properties: 30 m length, 0.25 mm ID, and 0.25 µm film thickness. Oven temperature program: 30 °C (10 min), 10 °C min-1 to 300 °C (10 min). Injector temperature: 300 °C; Carrier gas: He with a flowrate of 1 mL min -1 and a split ratio of 1:100. Detector: MS; the transfer line, ion source, and manifold were maintained at 300, 150, and 40 °C, respectively. 4 Column properties: 30 m length, 0.32 mm ID Oven temperature program: 35 °C (8.5 min), 24 °C min-1 to 250 °C (5 min). Injector temperature: 200 °C; Carrier gas: He with a flowrate of 3.0 mL min -1 and a split ratio of 1:10. Detector: TCD, 230 °C

3.3.5 Fractionation of pyrolysis liquid products

Figure B1 (Appendix B) shows a lab-scale distillation apparatus used to separate the components of catalytic pyrolysis liquid products. We focused on separating benzene, toluene and xylene (BTX) form other aromatics. These aromatics (BTX) contribute to the

large world market for commodity chemicals and have a diverse uses across several industries. For instance, benzene is used as precursor for styrene, phenol, nylon and aniline production; toluene is blended into unleaded gasoline to improve the octane number; and xylene is used to produce polyethylene terephthalate (PET) and resins [107].

The liquid products collected form the catalytic pyrolysis was first distilled to separate benzene and toluene from other fractions. The collected distillates (fraction 1) were weighed and stored for chemical analysis. Thereafter, the residue was redistilled to separate xylene (and/or chemicals that have boiling point close to xylene) from other heavy products. Afterwards, the distillates and the residue from the second distillation were weighed and stored for analysis. A detailed description of the distillation apparatus and method used in present study is provided in Appendix B (Figure B1).

3.4 Results and Discussion

3.4.1 Catalytic pyrolysis of triglyceride

Soybean oil is a well-defined source of triglyceride that has been used widely in previous studies. Soybean oil was used as feedstock to optimize the operating conditions such as reaction temperature, residence time and reaction time on stream. Our results with soybean oil were then compared with the literature.

3.4.1.1 Effects of temperature and residence time on products yield

Figure 3.2 shows the yields of organic liquid products (OLP) (Figure 3.2a), non- condensable gases (Figure 3.2b), aqueous phase (Figure 3.2c) and coke (Figure 3.2d) at the 54

pyrolysis temperatures of 450-500 °C and catalyst loading of 0-5 g. From Figure 3.2a, it can be observed that increase of catalyst loading decreases the OLP yield. For instance, the liquid yield at 450 °C using 1g of HZSM-5 was 87.9% and those using 5g catalyst was

78.9%. HZSM-5 is a highly shape selective catalyst with approximately 5Å pore size and contains strong Brønsted acid sites (see Table B1, Appendix B). These acidic sites enhance the decomposition of large molecules that are produced from initial thermal cracking of triglycerides. As a result of the cracking facilitated by the catalyst, small molecules that fall in non-condensable gases (e.g. C1-C4) will increase, and consequently the liquid yield decreases. Moreover, pyrolysis in presence of catalyst improved the conversion of triglyceride to non-glyceride products (see Table B2, Appendix B), likely due to increase of cracking reactions rate in presence of HZSM-5. Figure B2 (Appendix B) shows the TGA results from decomposition of soy oil in the absence and presence of HZSM-5. For degradation in presence of catalyst, HZSM-5 was mixed with soy oil (approximately 1:1 mass ratio). One can observe that both samples (with and without catalyst) showed one major derivative weight loss peak (dw/dT, where w is the sample weight and T is the temperature). However, in the presence of the catalyst, the temperature corresponding to the derivative weight loss peak shifted from 430 °C (in absence of catalyst) to 360 °C. This indicates that using HZSM-5 enhances the rate of triglyceride degradation.

(a) (c) 450 ºC 475 ºC 500 ºC 5 100 4

80 %)

60 3 %)

40 2 Organic phase yield phase (wt. yield Organic

20 1 Aqueous phase yield phase (wt. yield Aqueous 0 0

1.0 (d) 50 (b) 40

gases yield yield gases 30 0.5

20 (wt. %) (wt.

10 %) (wt. yield Coke

condensable condensable - 0 0.0 Non 0 1 2 3 4 5 6 0 1 2 3 4 5 6 HZSM-5 (g) HZSM-5 (g) 60 30 20 15 12 60 30 20 15 12

WHSV (h-1) WHSV (h-1) Yields of (a) organic phase (b) non-condensable gases (c) aqueous phase and (d) coke versus HZSM-5 loading at tested reaction temperatures.

Non-condensable gases (such as C1-C4 hydrocarbons, CO, CO2 and H2) yield was calculated by subtracting liquid products (OLP and aqueous phase) and coke mass from feed mass. As expected, non-condensable gases yields increased with temperature and catalyst loading (Figure 3.2b). Cracking of vegetable oil is an endothermic reaction, thus higher temperature breaks down large molecules more readily, and forms more gaseous product [53]. In addition, increasing catalyst loading increases the contact between product vapors and catalyst and consequently increases generation of non-condensable gases increases.

Figure 3.2c shows the mass fraction of aqueous phase (relative to feed) from cracking of soy oil. As observed, there was no evidence of water in liquid products from pyrolysis without catalyst. This implies that decarboxylation and/or decarbonylation are likely the main reactions to deoxygenate the heavy oxygenated compounds (e.g. oleic acid;

RCOOH). However, pyrolysis in the presence of HZSM-5 resulted in water formation due to dehydration of the heavy oxygenated compounds. Dehydration of heavy oxygenated compounds followed by decarbonylation results in production of hydrocarbons.

From Figure 3.2d, the highest measured coke production was 0.6 wt. % (relative to feed mass). The low coke yields can be attributed to small (if any) extent of polymerization/polycondensation of soy oil due to rapid vaporization of feed. Additionally, due to sufficiently low contact time between vapors and catalyst the polyaromatization/dehydrogenation that cause coke formation is low. It is important to note that coke deactivates catalyst and negatively affects its performance. Therefore, low coke yield achieved from our catalytic pyrolysis will decrease catalyst deactivation rate.

Overall, the OLP yields from pyrolysis of triglyceride over HZSM-5 were significantly improved (71-87%) relative to previous studies (20-50%). The high yields are likely due to the rapid volatilization of the feed and the short residence time, which decreases the chance for pyrolysis products to undergo extensive secondary reactions. To reach such a short residence time, rapid vaporization of feed is required. In our reactive pyrolysis system, due to atomization of triglycerides, small droplets of oil could vaporize and reach the reactor temperature rapidly, which allowed the pyrolysis to be carried out at a short time.

3.4.1.2 Effects of temperature and residence time on products composition

Figure 3.3 shows a general schematic of the principal steps involved in thermo- catalytic pyrolysis of triglyceride as described previously in the literature [34, 106, 108], and also is evident from the present work. The first step in cracking of triglycerides is the disintegration of the glycerol backbone [50, 53]. Subsequently, the long-chain fatty acids

(L-FAs) produces from first step can further degrade via three parallel pathways. First, they might decarboxylate and/or decarbonylate and produce heavy hydrocarbons (Hy-HCs),

CO2 and/or CO (reaction 3). Alternatively, L-FAs may deoxygenate due to dehydration followed by decarbonylation reactions and produce Hy-HCs, H2O and CO (reaction 4).

Acid sites of HZSM-5 facilitate the dehydration reaction. Additionally, the L-FAs can degrade at the sites of the C=C bonds and form short-chain fatty acids (S-FAs) and light hydrocarbons (Lt-HCs) (reaction 5). The S-FAs can also subsequently deoxygenate due to decarboxylation and/or decarbonylation (reaction 6) or dehydration (reaction 7) and produce Lt-HCs, CO2, CO and H2O. In parallel, the Hy-HCs produced from reactions 3 and 4 can break down at the C=C sites and form Lt-HCs (reaction 8), this can be facilitated due to acid sites of HZSM-5. Afterward, the unsaturated Lt-HCs (olefins) undergo oligomerization reactions (reaction 9) to produce C2-C10 olefins due to shape selectivity of

HZSM-5. The restructure of light olefins (oligomerization) to the size and shapes permitted by the pores of catalyst occurs within the HZSM-5 pores. C2-C10 olefins also undergo cyclization, dehydrogenation and aromatization reactions that lead to aromatic and naphthalene formation (reaction 12). The produced naphthalene may undergo further aromatization and dehydrogenation reactions and form polyaromatics (reaction 13) and

eventually coke as the results of dehydrogenation (reaction 14). Triglyceride may also undergo polymerization/condensation that results in coke formation (reaction 15).

A general thermo catalytic cracking mechanism of soybean oil: (1) initial · · cracking of triglyceride that generates free radical RCOO or/and RCO (2) gases produced as the result of glycerol backbone loss (3) decarboxylation and/or decarbonylation of long chain fatty acids (L-FAs), followed by hydrogenation and/or dehydrogenation that produces heavy hydrocarbons (Hy-HCs) (4) dehydration of L-FAs followed by decarbonylation (5) C-C double bond cleavage of L-FAs to produce short chain fatty acids (S-FAs) and light hydrocarbons (Lt-HCs) (6) decarboxylation of S-FAs to produce Lt-HCs (7) dehydration and decarbonylation of S-FAs (8) Hy-HCs breakdown due to cleavage of unsaturated C-C bonds to produce Lt-HCs (9) oligomerization of Lt-HCs to produce C2-C10 olefins due to shape selectivity of HZSM-5 (10) decomposition of olefins to gases (e.g. propene and ethene) (11) oligomerization of olefins (12) cyclization/aromatization of olefins followed by dehydrogenation to produce aromatics (13) polymerization of aromatics to produce polyaromatics (14) coking from polyaromatics (15) coking from polymerization/polycondensation of triglycerides. TG: triglyceride, L-FAs: long-chain fatty acids, G: gases, Hy-HC: heavy hydrocarbons, S- FAs: short-chain fatty acids, Lt-HC: light hydrocarbons, Olef: olefins, Ar: aromatic, PAH: polyaromatics and C: coke. *: dominant pathway in presence of HZSM-5

Figure 3.4 shows the yields of total aromatics (Figure 3.4a), C16-C18 fatty acids

(Figure 3.4b), C5-C13 olefins (Figure 3.4c), C5-C13 paraffin (Figure 3.4d) and C14- C18 aliphatic (Figure 3.4e) in the OLP. As observed, the OLP is mainly consisted of aromatics, and clearly the aromatics content increased with catalyst loading and temperature. These aromatics are C6-C12, however, the majority of aromatics are benzene, toluene and xylene

(BTX). For instance, at 500 °C and using 5g HZSM-5, the OLP was comprised of 43.1%

BTX and 10.4 % C9-C12 aromatics (Table B3, Appendix B). This selective production of

BTX is due to molecular sieve characteristic of HZSM-5, which means the molecules that are structurally larger than the catalyst pore size will decompose in order to accommodate within the pores.

Figure 3.4b shows the content of fatty acids (C16-C18) in the OLP. It can be seen that fatty acids content decreases with temperature and catalyst loading, likely due to dehydration, decarbonylation and/or decarboxylation. Pyrolysis without catalyst under all examined temperatures resulted in higher fatty acids compared to those with catalyst indicating that HZSM-5 facilitates deoxygenation reactions. The produced fatty acids (C16-

C18) can be separated/purified from other hydrocarbons through a downstream fractional distillation due to significant difference in boiling point (b.p.) of ethyl naphthalene (as largest identified hydrocarbons in the produced OLP; b.p. of 251 °C) and palmitic acid (as smallest identified fatty acid in produced OLP; b.p. of 390 °C).

The OLP also contained of C5-C13 olefins (Figure 3.4c), C5-C13 paraffin (Figure

3.4d), and C14-C18 aliphatic (Figure 3.4e), and their content decreased using higher amount

of catalyst. This can be due to more cracking at higher catalyst loading, which probably converts the aliphatic hydrocarbons into aromatics or non-condensable hydrocarbons such as C1-C4.

(a) (c) 450 ºC 475 ºC 500 ºC 20 60

40 %) (wt.

C -

5 10

Olefins C Olefins Total aromatics (wt. %) (wt. aromatics Total 0 0

60 20 (b) (d) 15

(wt. %) (wt.

13 C

- 10

5 fatty (wt. %) acid fatty

20 18 18

C 5

Paraffin C Paraffin C 0 0 0 1 2 3 4 5 6 0 1 2 3 4 5 6 HZSM-5 (g) HZSM-5 (g) 60 30 20 15 12 60 30 20 15 12

WHSV (h-1) WHSV (h-1) 20 (e)

(wt. %) (wt. 15

14 10

Aliphatics C Aliphatics 0 0 1 2 3 4 5 6 HZSM-5 (g) 60 30 20 15 12

WHSV (h-1)

Yields of (a) total aromatics (b) L-FAs (c) C5-C13 olefins (d) C5-C13 paraffin and (e) C14- C18 aliphatic in the organic phase.

3.4.1.3 Feasibility of complete deoxygenation of liquid product

Catalytic pyrolysis of triglycerides at WHSV of 60-12 h-1 resulted in partial deoxygenation of fatty acids. In order to improve fatty acids deoxygenation rate, vegetable oil was pyrolyzed at 500 °C where WHSV decreases from 12 to 3 h-1. The desired WHSV was achieved by adjusting the feed mass flowrates. The total mass of oil that was introduced into the reactor was kept constant to eliminate effect of oil mass that has been pyrolyzed by the catalyst on products yield and composition. Thus, to maintain WHSV of

12, 6 and 3 h-1, the feed mass flowrate was set at 1, 0.5 and 0.25 g min-1, and reaction was carried out for 30, 60 and 120 min, respectively.

Figure 3.5 shows products yields (Figure 3.5a) and the OLP composition (Figure

3.5b). From Figure 3.5a, it can be seen that OLP yields decreased from 71% to 65% where

WHSV reduced from 12 to 3 h-1. At lower WHSV, the contact time between vapors and catalyst increases, thus cracking activity enhances. Consistently, aqueous phase yield is higher at lower WHSV due to more dehydration of fatty acids. Dehydration of fatty acids mainly occurs on acid sites of HZSM-5. At lower WHSV, there are more accessible sites on catalyst surface, therefore the dehydration rates increase. Interestingly, coke yields are small (0.6-1%) even at low WHSV. Coke formation could result from polymerization/condensation of triglyceride or/and polyaromatization/dehydrogenation of the formed aromatics (see Figure 3.3). The former reaction mainly occurs in the liquid phase, however in our reactive pyrolysis system, due to rapid vaporization of triglyceride, there is a small chance (if any) of liquid build up. Moreover, low coke yield achieved in our pyrolysis system indicates small extent of polyaromatization/dehydrogenation.

(a)

100 Oil-phase Noncondensable gases 10 Aqueous phase Coke

80 8 . %) .

(wt. %) (wt. 60 6

40 4 Products yields yields Products 20 2 yields (wt Products

0 0 0 5 10 15 WHSV (h-1) (b) 100 Unidentified

(wt. %) (wt. 80 Fatty acids Aliphatic>C13 60 Paraffins C5-C13

omposition Olefins C5-C13 40 Naphthalenes C8+ aromatics 20 BTEX

Liquid c Liquid products 0 12 6 3 WHSV (h-1)

(a) Products yields and (b) composition of organic phase from catalytic pyrolysis of triglyceride at WHSV of 3-12 h-1.

Figure 3.5b and Table B4 (Appendix B) show the composition of OLP at WHSV of 3-12 h-1 and. As observed, fatty acids content decreases with reduced WHSV. For instance, at WHSV of 3 h-1 the OLP was comprised of only 4% fatty acids and as high as 63

57% BTX. This can be attributed to more dehydration of L-FAs followed by aromatization of olefins at lower WHSV. This is consistent with higher aqueous phase yield at lower

WHSV (Figure 3.5a). In other words, at lower WHSV the contact time between fatty acids

(as intermediate products) and acid sites of HZSM-5 is higher, thus dehydration reactions are promoted. Moreover, the BTX content increased when WHSV decreased, likely due to more deoxygenation and C=C break down of fatty acids, which results in more short chain olefin production. The produced olefins then undergo aromatization and dehydrogenation to produce aromatics. The selective production of BTX (C6-C8 aromatics) is due to molecular sieving characteristics of HZSM-5. The size of aromatic molecules that can diffuse through the pores of HZSM-5 is known to be less than C10 [92, 109], thus larger molecules (>C10) undergo further cracking in order to be accommodated within the HZSM-

5 pores.

3.4.1.4 Effect of reaction time on stream

Figure 3.6 and Table B5 (Appendix B) shows the products yields (Figure 3.6a) and

OLP composition (Figure 3.6b) from catalytic pyrolysis of triglyceride over reaction time on stream of 1-5 h. As observed, the yield of organic phase increased and non-condensable gases yield decreased at higher time on stream. At higher reaction time on stream, more coke deposits on the catalysts surface and inner pores. Therefore, catalyst activity decreases, and consequently the cracking rates slow down. From Figure 3.6b, it can be seen that at reaction time on stream of 1 h, the liquid product is mainly comprised of BTEX and naphthalene and no trace of oxygenated compounds such as fatty acids were observed.

However, at higher time on stream, the fatty acids yield gradually increased and reached 64

28% (relative to the liquid products) at a reaction time on stream of 5 h. Interestingly, the aqueous phase yield (Figure 3.6a) decreased with the time on stream, which indicates that dehydration of the fatty acids is reduced under these conditions, likely due to fewer available acidic sites of the HZSM-5 catalyst. The BTEX yields decreased from 70% to

23% when the time on stream increased from 1 to 5 h, which contributes to the less dehydration and aromatization reactions resulted from the catalyst deactivation.

(a)

100 Organic phase Noncondensable gases 10 Aqueous phase Coke 80 8

60 6

yields %) (wt. yields %) (wt. yields

40 4 Products Products 20 2 Products

0 0 0 1 2 3 4 5 6 Reaction time on stream (h)

(b) 100 Unidentified

80 Fatty acids

.%) (wt 60 Aliphatic

Naphthalenez 40 composition C8+ aromatics

OLP OLP 20 BTEX 0 1 2 3 5 Reaction time on stream (h)

Effect of reaction time on stream on (a) products yields and (b) OLP compositions.

3.4.2 Catalytic pyrolysis of non-edible oils feedstock

3.4.2.1 Products yields and compositions

Catalytic pyrolysis of pennycress, camelina and waste cooking oil performed at 500

°C, WHSV of 3 h-1 and reaction time on stream of 1 h and the products yields and compositions is shown in Table 3.2. As observed, 53-58% OLP yields were achieved from catalytic pyrolysis of the tested non-edible oils. The OLP was mainly comprised of benzene, toluene and xylene. However, the products from camelina oil contained higher benzene, probably due to higher amount of unsaturated C-C chain in original fatty acid (see

Table B6, Appendix B).

The non-condensable gas (NCGs) is mainly comprised of light hydrocarbons (C1-

C3), CO and CO2. Additionally, small amount of H2 (0.2-0.3 mole %) was observed in the non-condensable gases due to dehydration of the cyclohydrocarbons such as cyclohexane.

Interestingly, the CO content in NCGs was much higher that CO2 demonstrating that the oxygenated compounds produced during the pyrolysis undergoes more decarbonylation reactions rather than decarboxylation. The oxygen content of the tested non-edible oil

(pennycress, camelina and waste cooking oil) was estimated to be ~10 w.t % relative to the feed mass. From aqueous phase and CO yields (see Table 3.2), it can be concluded that the deoxygenation occurred mostly through dehydration (~50%) and decarbonylation (~30%) pathways. The NCGs contained more than 70 mole % hydrocarbons, in particular, propylene, ethylene and methane.

Table 3.2: Products yields and composition from catalytic pyrolysis of non-edible oils.

Properties Feedstock Products yields (wt. %) Pennycress Camelina WCO

OLP 56.5 53.2 58.6 Aqueous phase 5.3 6.5 5.1 Coke 1.1 1.8 0.9 Non-condensable gas 37.0 38.5 35.3

Products compositions

OLP (wt. %) Benzene 21.0 26.6 15.6 Toluene 34.5 32.2 37.5 Ethylbenzene 1.8 2.0 1.7 Xylene 15.2 16.9 18.6 C8+ Aromatics 10.3 4.1 9.7 Naphthalene 9.0 11.3 9.3 Total identified 91.8 93.1 92.3

* NCG (mole % ) H2 0.3 0.2 0.3 CO 15.6 14.6 13.5 CO2 4.9 6.1 4.3 CH4 15.9 10.4 22.5 C2H4 14.7 20.8 11.3 C2H6 3.3 5.0 2.2 C3H6 32.6 25.5 30.8 C3H8 3.7 4.8 5.9 Total identified 91.1 87.4 90.8 * NCG: Non-condensable gas

The amount of energy that can be produced from combusting these light hydrocarbons was calculated to be 35-42 kJ g-1. Considering the non-condensable gases yields, the NCGs contains an approximately 11-13 kJ g-1 feedstock. Of these, 1.3-2.6 kJ g-

1 feedstock was generated from methane. Moreover, the pyrolysis process requires nearly

0.8 kJ g-1 feedstock, which includes raising the oil temperature from room temperature to

500 °C, evaporation and enthalpy of reaction. Therefore, part of the methane in NCGs can be used to supply heat while most of the propylene and ethylene can be separated to be used in petrochemical industries. Moreover, the propylene and ethylene can be returned back to the reactor to improve the yield of BTX.

3.4.2.2 Catalyst long-term reusability

3.4.2.2.1 Design of reaction-regeneration cycle

To assess the reusability of the HZSM-5 catalyst, the reaction-regeneration cycle was performed for several times. It is desirable that catalyst performance remains same after regeneration, this allows reusing the catalyst for long time and results in significantly lower catalyst replacement. Regeneration of the zeolite catalyst carried out by introducing

O2 into the reactor at high temperature to combust the coke deposited on the catalyst. TGA instrument was employed to find the combustion temperature and duration. After catalytic pyrolysis of WCO for 1 h at 500 °C and WHSV of 3 h-1, HZSM-5 catalyst was recovered from the reactor and used for regeneration study on the TGA instrument, where coke on the catalyst combusted at 500, 550 and 600 °C. The TGA data indicates that at 600 °C and after 10 min there was no mass loss, which implies a complete coke combustion.

Figure B3 (Appendix B) shows the reaction-regeneration procedure of WCO catalytic pyrolysis. Prior to each experiment, the reactor was packed with 5 g of HZSM-5 catalyst and placed in the furnace and heated while being purged with pure N2 to remove any traces of H2O and O2 from the system. After the reactor reached 500 °C, the N2 purge was stopped and WCO fed into the reactor at mass flow rate of 0.25 g min-1 for 60 min.

Afterwards, the pump was stopped and the condenser trap was disconnected from the reactor and the liquid products were collected and stored. Then, reactor temperature increased to 600 °C and held for 10 min under O2 atmosphere to regenerate the catalyst by combusting the coke deposited on the catalyst. At the end, the O2 stopped and reactor cooled down to 500 °C under N2 atmosphere to remove any O2 and H2O from the reactor.

To assess the long-term reusability of the HZSM-5 catalyst and performance of the regenerated catalyst, 12 reaction-regeneration cycles was performed and the products yields and compositions from each cycle were analyzed.

3.4.2.2.2 Products yields and compositions from reaction-regeneration cycle

Figure 3.7 shows the products yields (Figure 3.7a) and compositions (Figure 3.7b) from 12 reaction cycles of WCO catalytic pyrolysis. As observed, the organic phase yield from 12 reaction cycles is relatively constant and near 57%, which demonstrates that catalytic performance remained same after regenerations. From Figure 3.7b, it can be seen that more than 90 wt. % of the liquid products is aromatics (mainly BTX) through all 12 reaction cycles. The yields of C8+ alkylbenzene and xylene increased and then stayed constant after three reaction cycle. In contrast, the naphthalene yields decreased under the similar reaction condition. From 12 reaction cycles, average yields primary products were benzene (12.5%), toluene (31.8%), xylene (24%), ethyl benzene (4.3%), C8+ alkylbenzene

(11.9%) and naphthalene (5.9%). The products yields and compositions from 12 reaction cycles suggest that catalysts can be reactivated and reused for a long time.

(a) 100 Organic phase Aqueous phase 80 Non-condensable gas

40 Products yields ) yields (wt.% Products 20

0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 Reaction cycle

40 (b) 100 Benzene

80 Toluene 30

60 Ethylbenzene 20 Xylene 40 Napthalene

Total atomatics (wt.%) atomatics Total Aromatic yields (wt.%) Aromatic yields 20 C8+ aromatics

0 0 Total aromatics 0 1 2 3 4 5 6 7 8 9 10 11 12 13 Reaction cycle (a) Products yields and (b) compositions of WCO catalytic pyrolysis at 12 reaction cycles.

3.4.2.2.3 Fractionation of liquid products

Table 3.3 shows the mass and composition of each fraction from distillation of 67 g liquid products that were produced from 12 reaction-regeneration cycle of WCO catalytic pyrolysis. Moreover, Figure B4 (Appendix B) depicts the GC-FID chromatogram of each distillates and residue fraction. The goal of distillation was to separate aromatics such as

benzene, toluene, xylene and alkylbenzene from other fractions. As observed from Table

3.3 and Figure B4 (Appendix B), the first fraction was predominantly comprised of

benzene and toluene. The second fraction was mainly contained xylene, ethylbenzene and

C8+ alkylbenzenes (e.g. methyl ethyl benzene, trimethyl benzene). Finally, the residue

fraction from distillation experiment was comprised of naphthalene (e.g. naphthalene,

methyl naphthalene, methyl ethyl naphthalene) and small amount of C8+ alkylbenzene.

The residue fraction also contained some other chemicals that we were not able to

identify/quantify using GC-MS and GC-FID. Those chemicals might be heavy

polyaromatics or aliphatic that formed during the long duration of distillation process.

Table 3.3: Distillation of OLP from 12 reaction-regeneration of WCO pyrolysis.

Properties Initial 1st fraction 2nd fraction Residue Material lost

W (g) Wt. % W (g) Wt. % W (g) Wt. % W (g) Wt. % W (g) Wt. % Benzene 8.4 12.5 7.1 27.5 0 0 0 0 1.3 15.6 Toluene 21.3 31.8 17.7 68.7 0 0 0 0 3.6 17 Ethylbenzene 2.9 4.3 0.1 0.2 1.6 7.6 0 0 1.2 42.2 Xylene 16.1 24 0.1 0.6 14.5 69.1 0 0 1.5 9.1 Naphthalene 3.9 5.9 0 0 0 0 3.2 38.3 0.8 19.4 C8+ aromatics 8 11.9 0 0 4.4 20.9 0.6 7.5 3 37.5 Total aromatics 60.6 90.5 25 97 20.5 97.5 3.8 45.8 11.4 18.8

Total mass (g) 67 25.7 21 8.3 12

The first and second distillates fraction has clear and transparent appearance. These

fractions can be further distilled through a downstream fractional distillation to produce

high purity aromatics that can be directly used in polymer industries. For instance, benzene

is used as precursor for styrene, phenol, nylon and aniline production; toluene is blended

into unleaded gasoline to improve the octane number; and xylene is used to produce polyethylene terephthalate (PET) and resins.

3.5 Conclusion

Catalytic pyrolysis of triglyceride-based feedstock performed on a reactive pyrolysis system that equipped with an atomizer to introduce liquid feedstock into the reactor in form of micron-size droplets to rapidly volatilize the oil and achieve short residence time. As such, organic liquid products (OLP) yield of 84-71% was obtained from triglyceride conversion in presence of HZSM-5 zeolite catalyst. At reaction temperature of

500 °C and WHSV of 12 h-1, the OLP contained 53.5% aromatics (mainly benzene, toluene and xylene; BTX), 17.6% C16-C18 fatty acids and 7.8% aliphatic. To improve the BTX yields, pennycress, camelina and waste cooking oil were converted at lower WHSV of 3 h-

1. An OLP yield of 53-59% was produced from these non-edible oils and the OLP was predominantly comprised of BTX. Moreover, the non-condensable gases produced from catalytic pyrolysis contained more than 70% C1-C3 hydrocarbons, mostly propylene, ethylene and methane. Additionally, catalyst reusability was assessed by performing 12 reaction-regeneration cycles of WCO pyrolysis. The products yields and composition remained constant after 12 reaction cycles, which demonstrates the catalyst can be used for long time.

Chapter 4

High-Yield Production of Fatty Nitriles by One-Step

Vapor Phase Thermo-Catalysis of Triglycerides2

4.1 Abstract

Fatty nitriles are widely used as intermediate molecules in the pharmaceutical and polymer industries. In addition, hydrogenation of fatty nitriles produces fatty amines that are common surfactants. In the conventional fatty nitrile process, triglycerides are first hydrolyzed and the resulting fatty acids are catalytically reacted with NH3 in a liquid phase reaction. We report a simpler one-step fatty nitrile production method that involves a direct vapor phase reaction of triglycerides with NH3 in the presence of heterogeneous solid acid catalysts. The reactions were performed in a tubular reactor maintained at 400 °C into which triglycerides were injected through an atomizer to allow rapid volatilization and reaction; NH3 was fed as a gas. Several metal oxide catalysts were tested and reactions in the presence of V2O5 resulted in near-theoretical fatty nitrile yields (84 wt.% relative to the feed mass). In general, catalysts with higher acidity such as V2O5, Fe2O3 and ZnO showed higher fatty nitrile yields compared to low acidity catalysts such as ZrO, Al2O3 and CuO.

2 Shirazi et al., ACS Omega, 2 (2017) 9013-9020.

Energy balance calculations indicate that the one-step reaction described here would require significantly lower energy than the conventional process primarily due to elimination of the energy-intense triglyceride hydrolysis.

Key words: Triglyceride, Hydrolysis, Fatty nitrile, Metal oxide, NH3-TP.

4.2 Introduction

Triacylglycerols (commonly known as triglycerides) from oilseeds or algae have the potential to, at least partially, displace petroleum-derived fuels and chemicals [24, 106,

110-112]. Their built-in oxygenated functional groups can be utilized for value-added chemicals, which are traditionally produced from functionalization and oxidation of petroleum-derived hydrocarbons [29, 30]. For instance, fatty nitriles (platform molecules in pharmaceutical and polymer industries) [113] and fatty amines (common cationic surfactant precursor) [29], which are nitrogen derivatives of fatty acids, olefins or alcohols can be produced from fats and oils, instead of petrochemical raw materials. Due to their high affinity to natural surfaces, fatty amines have been used in surface modification applications such as fabric softening, hair conditioning, corrosion inhibition, mineral flotation and bactericides [114, 115]. Fatty amines also have good oil solubility and lubricity and are thereby particularly useful in friction modification. In addition, the emulsification and dispersion properties of fatty amines are exploited in numerous cleaning and agricultural formulations [114, 115]. A recent global market study on fatty amines projected that global industrial consumption of fatty amines is expected to reach >650,000 metric tons with a revenue of more than 2 billion US dollars by 2020 [116].

Figure 4.1 shows conventional routes for fatty amine production from petroleum as well as oleaginous biomass feedstocks. While petroleum is still the main resource for amine production, the bio-based route is potentially more environmentally sustainable. In the conventional bio-based process for producing fatty amines (C8-C22), the first step is the production of fatty nitriles through the “nitrile process” [117] wherein triglycerides

(extracted from oilseeds) are converted to fatty acids through hydrolysis and then reacted with NH3 in the presence of metal oxide catalysts such as alumina or zinc oxide (see purple box in Figure 4.1). The nitrile reaction is performed at 280-360 ºC in liquid state for several hours. Excess ammonia (2-4 times of the required stoichiometric value) is added and water is continually removed from the reactor to move the equilibrium reaction towards nitrile production. Downstream, the fatty nitriles are hydrogenated typically in the presence of a nickel catalyst (e.g. RANEY® Ni) to produce fatty amines [118].

Figure 4.1: Pathways to produce fatty amines.

In batch implementations of the conventional nitrile process, fatty acids stay at high temperature and liquid state for extended periods. Consequently, several side-reactions such as isomerization, polymerization, Piria, Diels–Alder or peroxidation reactions can 76 occur, particularly in the presence of the unsaturated fatty acids [113, 119]. Moreover, in batch reactor systems, the produced fatty nitriles could also be hydrated to fatty amides.

Thus, at the end of the reaction, the product contains unreacted fatty acids, fatty nitriles, fatty amides and other undesired byproducts thereby lowering fatty nitrile yields and complicating product recovery. To improve the fatty nitriles yields, some studies have investigated nitrile reactions in continuous flow reactors packed with catalysts with countercurrent flow of fatty acids and NH3 [119-121]. The continuous nitrile process has the potential to decrease the undesired side products (compared to batch reactor) due to more efficient contact between NH3 and fatty acids, lower reaction time (10-60 min) and continuous products removal [121]. 70-90 % of theoretical fatty nitrile yields are reported from the continuous nitrile process and the product stream is reported to contain unreacted fatty acids, fatty amides as well as dimers of fatty nitrile. Higher yields are possible with additional processing such as dehydration of the intermediate fatty amide [119].

Purification/distillation is used as a final step to produce high purity fatty nitriles required for surfactant applications [119].

An additional variation of the nitrile process involves gas phase reactions of fatty acids with NH3 at a temperature range of 300-600 C and in the presence of a solid catalyst bed. Gas phase reactions can potentially prevent undesired products due to short reaction time (few seconds) [122, 123] and better contact between NH3 and fatty acids [124]. Also, gas phase nitrile reactions significantly decrease the tendency for dimerization/polymerization due to greater distance between molecules (reactant and intermediate products), particularly when unsaturated fatty acids are used as feedstock [34,

53, 106]. Additionally, the process does not require catalyst recovery since a heterogeneous catalyst bed is used [113].

In contrast to several published studies on the conventional liquid phase nitrile process, there are relatively few reports on gas phase nitrile reaction of fatty acid (or its derivatives, e.g. fatty esters). From the published literature, the important reaction parameters that govern fatty nitrile yields in gas phase reactions are temperature, NH3/fatty acid molar ratio and catalyst type. For instance, Wortz et al. [125] used silica gel as catalyst for nitrile reaction of several fatty acids (e.g. lauric, palmitic and stearic acid) at reaction temperature range of 425-450 °C. They reported high yields of long chain fatty nitriles (up to 98% of the theoretical yield) and only ~1% short chain fatty nitriles likely due to degradation of the produced long chain nitriles and/or feedstock. Ralston et al. [126] also carried out gas phase nitrile reaction of long chain fatty esters in presence of Al2O3 supported by activated charcoal. In their process, the reaction was performed at much higher temperature (above 500 °C) in order to eliminate production of any heavy byproducts such as polymers and resins. However, higher reaction temperature favors short chain fatty nitriles due to C-C cleavage of original fatty acid (feedstock). Thus, in their study, the products were mainly comprised of short chain fatty nitriles and hydrocarbon gases.

Various metal oxide catalysts such as ZnO, Zr2O and Al2O3 have been used to convert fatty acids (or its derivatives, e.g. fatty ester) into fatty nitrile. It has been reported that oxides of metals from Groups III and IV of the periodic table are especially effective for fatty acid ammonization reaction [122, 126]. Mekki-Berrada et al. [122] studied several basic (e.g. MgO), amphoteric (e.g. ZnO, Zr2O and Al2O3) and acidic (WO3) solid catalysts

78 for gas phase nitrile reaction of lauric acid methyl ester at reaction temperature of 300 °C.

They showed that catalysts with amphoteric character provide higher fatty nitrile yields compared to basic or strong acidic catalysts. Although, acidic catalysts such as Nb2O5 yielded as high as 97% fatty nitriles, stronger acidic catalysts (e.g. Bentonite and WO3) promoted side products (e.g. methyl lauramide) and decreased the fatty nitriles yields.

Overall, the gas phase nitrile reaction has the potential to produce high yield of fatty nitrile from fatty acids. Previous studies (both gas and liquid phase reaction) have used fatty acid, which are typically produced from hydrolysis of triglycerides, as reactant for nitrile reactions. However, hydrolysis is an energy intense process due to high temperature

(250-350 °C) and pressure (45-60 bar) and requires long reaction time [127, 128].

Moreover, fatty acid purification/distillation and catalyst recovery is required after hydrolysis. Furthermore, polymerization and degradation can occur during the hydrolysis step, particularly in presence of unsaturated fatty acids. If nitrile reaction can be performed in the vapor phase by directly reacting triglycerides with ammonia in the presence of a catalyst bed, the hydrolysis step as well as catalyst recovery can be avoided. In present work, we investigated a one-step vapor phase nitrile reaction to directly convert triglycerides into fatty nitrile using diverse solid acid catalysts. Product yields were correlated to catalyst properties. Effects of triglyceride/NH3 ratios on product yield were assessed for reactions in the presence of V2O5 catalyst.

4.3 Experimental

4.3.1 Materials

Coconut oil was obtained from Zoyeoil (Zeeland, MI, USA). The following catalysts were purchased from Strem Chemicals, Inc. (Newburyport, MA, USA) - V2O5,

Fe2O3, ZrO2, ZnO, Al2O3, CuO. NH4-ZSM-5 powder with a SiO2/Al2O3 molar ratio of 23 was purchased from Zeolyst International (Conshohocken, PA, USA). The purchased NH4-

ZSM-5 was calcined in a muffle furnace for 5.5 h at 550 ºC to obtain HZSM-5. Hexane, chloroform, methanol and sulfuric acid were purchased from Fisher Scientific (Pittsburgh,

PA, USA). Analytical standards of lauric nitrile, lauramide, lauric acid, glycerides (triolein, diolein, and monolein), fatty acid methyl ester (FAMEs; mixtures of C8-C22) were purchased from Sigma-Aldrich (St. Louis, MO, USA).

4.3.2 Experimental set-up

All experiments were performed in a continuous vapor phase reactor system that is schematically shown in Figure 4.2, as described previously [106]. Prior to each experiment, the reactor was packed with catalysts, placed in the furnace and heated while being purged with pure N2 to remove any traces of H2O and O2 from the system. After the reactor reached set point temperature, the N2 purge was stopped. Then, NH3 gas was introduced into the reactor at a controlled flow rate measured by a mass flow meter (Alicat scientific, USA).

Thereafter, feed coconut oil was introduced into the reactor in the form of micron-size droplets through an atomizer (Sonazop, Farmingdale, NY, USA; model: HTNS40). The atomizer consists of an ultrasonic probe with a 4 mm diameter, which was operated at

80 constant frequency of 40 kHz. The atomizer used in our reactor system, creates extremely large surface area for heat transfer and allows rapid evaporation of feedstock. The system was operated in continuous mode for 3h and the products were condensed in a liquid N2 trap. At the end of the experiment, feed flow (NH3 and coconut oil) was stopped and the reactor furnace was turned off and allowed to cool to room temperature. Thereafter, the collected liquid products were weighed on an analytical balance (Mettler Toledo, USA) with ± 0.1 mg accuracy. The liquid product was separated in a separatory funnel into aqueous and organic phases.

Figure 4.2: Schematic diagram of one step vapor phase nitrile reaction system.

Conversion of coconut oil to non-glyceride products was calculated as

(W ) −(W ) Conversion (%) = TG 푖푛 TG out × 100 (4.1) (WTG)푖푛 where, (WTG)𝑖푛 is the mass of oil introduced into the reactor and (WTG)out is the mass of glycerides in collected liquid.

The yield of fatty acid, amide and nitrile in the liquid products was calculated as

w Yield (%) = i × 100 (4.2) (WTG)푖푛 where wi is the mass of fatty acids, amides or nitriles in the product.

The reaction residence time (휏) was calculated using Equation 4.3 as reported in previous studies [122, 123].

푉 휏 = (4.3) 푉̇ where 푉 is the catalyst bed volume and 푉̇ is the NH3 volumetric flow rate at reaction conditions (400 °C and 1 atm pressure).

4.3.3 Experimental conditions

Table 4.1 shows the experimental conditions for the reaction. Reactor temperature was set at 400 °C to allow vaporization of feedstock and simultaneously prevent extensive cracking reactions [126]. Previous studies on ammonization of fatty acids reported a residence time range of 3-13 s as sufficient contact time for fatty nitrile production [122].

As such, the residence time in our experiments was set to 10 s by introducing 8.9 mL min-

1 3 NH3 (at room temperature, 20 °C) into the reactor that was packed with 3.4 cm of catalyst. An NH3/triglyceride molar ratio of 12 (4×stoichiometric amount) was used in order to allow the reversible reactions to proceed in the forward direction.

Table 4.1: Operating conditions for the one-step vapor phase nitrile reaction of triglyceride. Feedstock Coconut oil Reactor temperature (°C) 400 Catalyst vol. (cm3) 3.4 NH flowrate (mL min-1) 8.9# 3 NH molar rate (mmole min-1) 0.369 3 Triglyceride molar rate (mmole min-1) 0.092 NH /triglyceride molar ratio 4 3 Residence time$ (s) 10 -1 NH3 linear velocity (mm s ) 5

# NH3 flowrate at room temperature and atmospheric pressure (STP) conditions. $ Calculated from Equation 3.

4.3.4 Catalyst characterization

Texture properties of catalysts were measured using N2 adsorption-desorption isotherms at 77 K by a Micromeritics ASAP 2020 instrument. Prior to the analysis, the samples were pretreated at 350 C for 3 h under vacuum to remove any adsorbed compounds. Multipoint adsorption isotherms were obtained on the samples using N2 as adsorbate. The BET equation was used to calculate surface area and the micropore volume was determined by the t-plot method [129, 130].

Ammonia temperature programmed desorption (NH3-TPD) technique is a useful method to measure catalyst acidity and strength of acid sites [131]. In this study, a custom dynamically-pumped TPD system was used as shown schematically in Figure C1

(Appendix C). Prior to each TPD measurement, ~130 mg catalyst was degassed at 350 °C under high vacuum (<10-6 Torr) for at least 5 h. Then, 100 mg of degassed catalyst was weighed and transferred into a borosilicate glass tube with an inner diameter of 7 mm and placed inside the TPD furnace. Thereafter, the system was evacuated for 30 min to remove trace air and H2O. Subsequently, the valve connected to the main chamber (gas detector) was closed, and NH3 gas was introduced into the sample tube that was placed inside the furnace at 80 °C. During this step, NH3 was allowed to adsorb on catalyst surface for 1 h.

Thereafter, the NH3 injection was stopped, the main chamber valve (see Figure C1,

Appendix C) was opened, and the sample tube was again pumped down to high vacuum

-6 for ~ 1 h to reduce NH3 partial pressure to the background level (<10 Torr). In other words, all “free NH3” in the system was pumped out with the exception of the NH3

83 adsorbed on catalyst surface. Afterwards, the catalyst was heated to 600 °C at a constant

-1 ramp rate of 5 °C min . The NH3 gas desorbing from the catalysts was monitored by a mass spectrometer (RGA 300, Stanford Research System) connected to the dynamically- pumped high-vacuum TPD chamber (see Figure C1). Mass spectra in the range 1-50 atomic mass units (AMU) was recorded in 3 s intervals. The concentration of the desorbed NH3 was measured by the integrated-signal approach as reported previously [132].

4.3.5 Feedstock and products analysis

Coconut oil was transesterified to FAMEs and analyzed by gas-chromatography using mass spectrometry (GC-MS) and flame ionization detection (GC-FID) to identify and quantify the fatty acid constituents in the feedstock, as explained elsewhere [106].

GC-MS (Bruker, 450-GC equipped with 300-MS) analysis was also performed to identify the chemical constituents in the liquid products from the reaction. An Agilent DB-

5MS fused silica capillary column (length: 30 m, ID: 0.25 mm, and film thickness: 0.25

µm; Agilent Technologies, Santa Clara, CA) was employed, where the column was initially set at 30 °C for 10 min, then heated at 10 °C min-1 to 300 °C and finally held at 300 °C for

10 min. The injector, transfer line, ion source, and manifold were maintained at 300, 300,

150, and 40 °C, respectively.

To quantify chemical compounds in the products, a GC-FID (Shimadzu 2012 plus) with an RTX-biodiesel (Restek, Bellefonte, PA, USA) column (15 m length, 0.32 mm ID, and 0.1 µm film thickness) was used. The column temperature was programmed as follows: initially set at 60 °C for 1 min, followed by a temperature ramp rate of 10 °C min-1 to 370

°C and a final hold for 5 min. The injector and FID temperatures were maintained at 370

°C. The FID detector response was first calibrated using standards of known concentrations of FAME, fatty acid, fatty amide, fatty nitrile, monoolein, diolein and triolein. The chemical compound concentrations in the samples (liquid products and feedstock) were estimated from calibration curves as described previously [14, 65, 104, 106].

4.4 Results and Discussion

4.4.1 Feedstock characterization

The coconut oil used in this study was comprised of 5% caprylic (C8:0), 5% capric

(C10:0), 46% lauric (C12:0), 20% myristic (C14:0), 12% palmitic (C16:0) and 12% stearic and oleic (C18:0 and C18:1) acids (Figure C2, Appendix C). As expected, the coconut oil was mainly comprised of fatty acids in the C12-C18 range, which are important in soaps, surfactants for cosmetics, pharmaceuticals and foodstuffs since they carry a biocompatible lipophilic group [133].

4.4.2 Catalyst characterization

Table 4.2 shows the physiochemical properties of the tested catalysts. The texture properties were determined from N2 adsorption isotherms at 77 K and acid properties were measured by the NH3-TPD method. As seen from Table 4.2, most of the metal oxide catalysts have similar BET surface areas (~2-6 m2 g-1), with the exception of HZSM-5 which has a much higher surface area (337 m2 g-1) and pore volume (0.2 cm3 g-1). Figure

C3 (Appendix C) shows the NH3 desorption profile obtained during catalyst heating. As observed, most metal oxide catalysts (with the exception of V2O5) showed two distinct

85 peaks at different temperatures. The first peak corresponds to weak/medium acidity and the second peak is from strong acid sites on the catalyst. A summary of catalyst acid properties is also shown in Table 2. ZrO2, V2O5, Fe2O3 and ZnO contained more acid sites

-1 than Al2O3 and CuO. HZSM-5 showed a significantly higher acidity of 459 mol g due to strong Brønsted acid sites.

Table 4.2: Texture-properties and acidity of tested catalysts.

Properties Texture† Acidity‡ Surface Pore Pore Total 1st acidity$ 2nd acidity* T T Catalyst max.1 max.2 area¥ size# volume@ acidity (µmol g-1) (µmol g-1) (K) (K) (m2 g-1) (nm) (cm3 g-1) (µmol g-1)

V2O5 6 18.7 0.028 14.1 8.6 5.5 378.7 441.3 Fe2O3 4.5 16.2 0.014 12.2 3.7 8.6 485.7 703.1 ZrO2 5.5 4.7 0.007 23.8 23.3 0.4 468.5 748.7 ZnO 4.6 4.4 0.006 8.5 5.9 2.7 490.6 586.8 Al2O3 2.6 12.8 0.009 3.1 2.6 0.5 448.4 615.6 CuO 1.7 20.3 0.009 1.0 1.0 ND 432 ND HZSM-5 337.7 0.5 0.204 459.1 171.9 287.2 397.1 503

† Texture properties were measured by ASAP2020 instrument. ‡ Acidity was measured by NH3-TPD method. ¥ BET surface area; # BJH Adsorption average pore diameter; @ Cumulative pore volume $ Measured from area of first peak; corresponds to medium/weak acid sites. * Measured from area of second peak; corresponds to strong acid sites. Tmax.1: Temperature where maximum NH3 desorption was observed in first peak. Tmax.2: Temperature where maximum NH3 desorption was observed in second peak. ND: Not detected. 4.4.3 Products yields and compositions

Figure 4.3a shows the liquid product yields (relative to feed mass) from the one- step vapor phase nitrile reaction of triglycerides in the presence of tested catalysts. For the reaction in the absence of catalysts, glass beads (1 mm OD) were packed (5 cm length) inside the reactor to maintain a similar residence time (10 s) as the reactions in the presence of catalysts. The dashed horizontal line represents the theoretical maximum fatty nitrile yield. For quantitative coconut oil conversion and 100% selectivity towards fatty nitriles

(see the generalized reaction pathways in Figure C4, Appendix C), the mass of fatty nitriles

86 produced would be 598 g per mol of coconut oil fed into the reactor. Using an average coconut oil molecular weight of 693 g mol-1 (based on measured fatty acid composition of coconut oil; Figure C2), the weight-basis theoretical yield of fatty nitriles equals 86 wt. % relative to coconut oil.

(a) 100 Theoretical fatty nitrile yields (86 wt.%) 80

Fatty acid 60 Fatty amide 40 Fatty nitrile

20 Products compositions (g/g of (g/g compositions feed) Products 0

No catalyst V2O5 Fe2O3 ZrO2 ZnO Al2O3 CuO HZSM-5

100 (b)

C18 80 C16

(g/g (g/g products) of 60 C14

C12 40 C10

20 C8

Fatty nitrile selectivity selectivity Fatty nitrile 0

No catalyst V2O5 Fe2O3 ZrO2 ZnO Al2O3 CuO HZSM-5

From Figure 4.3a, it can be seen that liquid product from the reaction performed in the absence of catalyst is comprised of 28 wt.% fatty acids, 26 wt.% fatty amides and only

22 wt.% fatty nitriles (all values are relative to feed mass). The product also contained 6 wt. % unreacted triglyceride feed (not shown in Figure 4.3a). The fatty nitrile yields significantly increased in presence of catalysts (Figure 4.3a) due to greater ammonization of the triglycerides and/or produced fatty acids and dehydration of the fatty amides (second and third reactions in Figure C4, Appendix C). It must be noted that while the reaction scheme in Figure C4 shows reactions of fatty acids with ammonia, it is also possible that the triglycerides react directly with NH3. Our results show that the composition of liquid products was catalyst-dependent. For example, a near-theoretical yield of 84 wt. % fatty nitrile (relative to feed mass) was achieved in the presence of V2O5. Yields were also high with HZSM-5 and Fe2O3 – nearly 81 wt.% fatty nitriles. Moderate yields of 77% and 73% were obtained with ZrO2 and ZnO, respectively. Reactions in the presence of Al2O3 yielded only 50% fatty nitriles and the fatty nitrile yields were even lower (34 wt.%) when CuO was used as catalyst. The GC-FID chromatograms of products from reactions in the absence of catalyst, with Al2O3 and with V2O5 (to compare products from low- and high- yield catalysts) are shown in Figure C5.

Figure 4.3b shows the composition of the fatty nitriles in the product from the catalytic reactions (= mass of nitrile/total product mass). The sum of the individual fatty nitrile weight fractions in the product represents the total fatty nitrile selectivity. We observed a mixture of C8-C18 fatty nitriles that corresponded to the C8-C18 fatty acids esters of coconut oil (see Figure C2). In the presence of V2O5, the selectivity was as high as 97% such that the fatty nitrile composition was nearly the same as coconut oil fatty acid

88 composition (see Figure C2). Table C1 (Appendix C) shows the detailed composition of liquid products from all the reactions. Fatty nitrile products smaller than C8 or larger than

C18 were not observed indicating that excessive cracking or polymerization reactions were prevented due to the combination of low reaction temperature and short residence time.

Although the liquid product from V2O5 contained a mixture of fatty nitriles with different carbon chain length, sufficient boiling point differences among the products would allow for purification of specific fatty nitriles via distillation, if desired.

4.4.4 Effects of catalyst acidity on fatty nitrile yield

Vapor phase nitrile reaction in presence of metal oxide catalysts improved the fatty nitrile yields, however some catalysts such as V2O5 and Fe2O3 showed much better fatty nitrile yields and selectivity compared to others such as Al2O3 and CuO. In order to understand the difference in performance of the tested catalysts, we further investigated the differences in catalyst properties. Figure 4.4 shows that catalyst acidity and fatty nitrile yields were positively correlated. For instance, V2O5 showed desorption of 14 μmol NH3

-1 per gram of catalyst and resulted in ~84% fatty nitriles, however, Al2O3 has 8 μmol g acidity and showed only 50% fatty nitrile yields (acidity values are given in Table 4.2).

Moreover, the products from the one-step vapor phase nitrile reaction in presence of V2O5 did not contain any measurable fatty amide content likely due to fast dehydration of fatty amide in presence of Lewis acid sites on the catalyst (see the reaction steps in Figure C4,

Appendix C). On the other hand, liquid products in presence of Al2O3 contained up to 12% fatty amide (see Figure 4.3a) possibly due to lower acidity of Al2O3 compared to V2O5.

Furthermore, it is likely that the catalysts acid sites improved the ammonization reaction

89 of triglyceride and/or produced fatty acid in addition to increasing the rate of the fatty amide dehydration. As a result, there was no evidence of triglyceride and/or fatty acid in the liquid products when V2O5 was used. However, the products in presence of Al2O3 and

CuO contained fatty acid (see Figure 4.3a) likely due to less ammonization of the produced fatty acids as a result of lower catalyst acidity.

Fe2O3 V2O5 ZrO2 ZnO

Al2O3

CuO

Glass beads

Figure 4.4: Correlation of catalyst acidity and fatty nitrile yields from vapor phase nitrile reaction of coconut oil.

4.4.5 Effect of NH3/triglyceride molar ratio

Figure 4.5 shows the fatty acids and fatty nitriles yields from one-step vapor phase nitrile reactions at various NH3/triglyceride molar ratios and Figure C6 (Appendix C) shows the GC-FID chromatograms of the liquid products at the various reactant molar ratios. The reactions were performed at 400 °C using V2O5. The molar ratios were adjusted by changing mass flow rates of coconut oil into the reactor while maintaining a constant

NH3 flowrate. As shown in Figure 4.5, in the absence of NH3 the liquid products were

90 mainly comprised of fatty acids (76 wt.% of C8-C18 fatty acids) along with some hydrocarbons (mainly C11-C14). These results are consistent with our previous observations of triglyceride pyrolysis [106]. In absence of NH3, triglycerides degrade into fatty acids and release glycerol backbone; then, the produced fatty acids can undergo deoxygenation and produce hydrocarbons. When stoichiometric amount of NH3 was introduced into the reactor, the liquid products contained a mixture of fatty nitriles and acids. The fatty acid yields decreased from 76% in absence of NH3 to 45% at NH3/triglyceride molar ratio of 3

(stoichiometric value). Moreover, under these conditions, 41% fatty nitriles yield was obtained. The fatty nitriles yields consistently increased when the NH3/triglyceride molar ratio increased up to 7.5 (2.5 times of stoichiometric value) and reached the maximum nitrile yields of ~97%. Interestingly, there was no evidence of fatty amide productions even when stoichiometric amount of NH3/triglyceride molar ratio was used. This indicates that dehydration of fatty amides in the presence of V2O5 acid sites is a fast reaction step.

Figure 4.5: Effects of NH3/triglyceride molar ratio on products yields from one step vapor phase nitrile reaction.

4.4.6 Proposed reaction mechanism of fatty nitrile production

Figure 4.6 shows the mechanism for production of fatty nitrile from triglyceride over Al2O3. Although yields were lower with Al2O3 than other catalysts, the reaction mechanism is described with this catalyst since its structure is well-known. Partially dehydroxylated metal oxides provide acidic sites and the density of the Brønsted/Lewis acidic sites on the surface highly depends on the dehydroxylation temperature and gas flow.

In our reactor system, high temperature and a flow of inert gas (N2) was applied prior to the exposure of the metal oxide to ammonia and triglyceride to generate active sites. We propose that adsorption of ammonia over the surface acidic sites and its activation to form a bridging amide (–NH2) is a plausible initiation step in this process (see Figure 4.6a).

Then, we hypothesize that a series of acid-base reactions occur as shown in Figure 4.6b.

The nucleophile O on carboxylic group coordinates with the electrophile Al, and the electrophile C on the carboxylic group undergoes nucleophilic attack by strongly basic –

NH2 to generate the active complex i. Transfer of alkoxy group from the intermediate ii to the Lewis acidic site yields a bridging alkoxide and the desired fatty amide product. The produced fatty amide undergoes dehydration (Figure 4.6c) and yields fatty nitrile and water. While this reaction can occur at high temperature even in the absence of a catalyst, the acidic catalyst sites facilitate the dehydration reaction. The alkoxide-bridged surface species is active and can act as an active site and regenerate the catalyst (Figure 4.6d).

(a)

(b)

i ii

(c)

(d)

Figure 4.6: Possible vapor phase nitrile reaction mechanism over metal oxide catalysts. (a) catalysts activation step (b) fatty amide formation in presence of active catalyst (c) fatty amide dehydration (d) active catalyst regeneration. MOx: metal oxide; ∆: high temperature.

4.4.7 Comparison between conventional and one-step vapor phase nitrile process

Table 4.3 shows the energy consumption during the conventional and one-step vapor phase nitrile process. A detailed energy calculation and assumptions are given in the

Appendix C (see Figure C7 and accompanying description). In the conventional nitrile

93 process, the first step is hydrolysis of triglycerides to produce fatty acids. A triglyceride/water ratio of 1/5, temperature of 250 °C, pressure of 45 bars and 100% triglyceride conversion was used to calculate the energy required in hydrolysis step [127,

128]. As shown in Table 4.3, hydrolysis of 1 kg triglyceride requires a minimum energy of

5652 kJ. Moreover, conversion of the fatty acid into fatty nitrile through conventional nitrile process requires a minimum energy of 387 kJ kg-1 triglyceride. Assuming 75% of the consumed energy can be recovered via appropriate heat exchange, the conventional method requires more than 1500 kJ to convert 1 kg triglyceride into fatty nitriles.

Energy components Process methods Conventional Proposed one-step vapor phase (kJ kg-1 triglyceride) (kJ kg-1 triglyceride)

Hydrolysis 5652a n/a Nitriles production 387b 925c Total energy consumed 6039 925 Energy that can be recoveredd 4529 694 Total net energy required 1510 231 n/a: not applicable a: T= 250 °C, P= 45 bar, water/triglyceride mass ratio of 1/5 b: T= 300 °C, P= 1 bar c: T= 400 °C, P= 1 bar d: Assuming 75% of the consumed energy assumed that can be recovered.

The one-step vapor phase nitrile reaction requires energy for increasing the feedstock temperature from ambient to the boiling point (~400 °C) and also vaporization.

From Table 4.3, one can see that the one-step vapor phase nitrile process requires only 925 kJ kg-1 triglyceride for this step. This energy requirement would even decrease to 230 kJ kg-1 triglyceride if 75% energy recovery could be achieved through heat exchange from the hot product stream. The lower energy consumption in our system is due to elimination of

94 the hydrolysis step. Furthermore, the one step vapor phase nitrile process eliminates (or at least significantly decreases) the complicated distillation/purification steps required in hydrolysis and fatty nitrile production steps through the conventional process. Overall, the one step vapor phase nitrile reaction is able to directly transform triglyceride into fatty nitrile at nearly theoretical yields with lower energy consumption and potentially much simpler reactor and separation units.

4.5 Conclusions

One-step vapor phase fatty nitrile production was performed in a single reactor, in which triglyceride reacted with ammonia. In absence of catalyst, the products contained

28% fatty acid, 26% fatty amide and only 21% fatty nitrile (relative to the feed). However, the fatty nitriles yields significantly increased when catalyst was used. A near-theoretical fatty nitriles yield of 84% (relative to feed mass) was achieved from nitrile process in presence of V2O5. Catalysts such as Fe2O3, ZrO2, ZnO and HZSM-5 also showed high fatty nitriles yields. However, the fatty nitriles yields were low in presence of Al2O3 and CuO, likely due to low acidity of these catalysts. Overall, nitrile yields positively correlated with catalyst acidity. The results showed that a minimum NH3/triglyceride molar ratio of 7.5 is required to achieve near-theoretical fatty nitrile yields. Energy assessments suggest that the one-step vapor phase nitrile process would require lower energy inputs than the conventional nitrile process. Furthermore, due to the high purity of the nitrile product, purification/distillation steps are expected to be simple. Finally catalyst recovery from the hydrolysis reactions and nitrile production steps in the conventional nitrile process, are eliminated due to the vapor phase reaction of triglyceride.

Chapter 5

Thermochemical Conversion of Microalgae to Biofuels and Chemicals: A Study on Integration of Microalgae Fractionation Pyrolysis with Ex-situ Catalytic Upgrading

5.1 Abstract

Microalgae are attractive feedstocks for biofuel production and are especially suitable for thermochemical conversion due to the presence of thermally labile biomass constituents - lipids, starch and protein. However the presence of starch and proteins also poses challenges and microalgae pyrolysis produces water as well as other O- and N- compounds that are mixed-in with energy-dense lipid pyrolysis products. In This chapter, we describe a two-step fractional pyrolysis approach integrated with vapor phase catalytic upgrading. This approach allows (1) the separate recovery of energy-dense lipid pyrolysis products and the lower calorific value bio-oils from the degradation of starch protein and

(2) tailored vapor phase upgrading of the resulting fractions. To evaluate the validity of this approach, Chlorella sp. was first pyrolyzed at 320 °C to volatilize and degrade the biomass starch and a majority of the protein. Then, the residual biomass was pyrolyzed again at 450 °C to recover products from lipid decomposition. The volatiles from each fraction were passed through an ex-situ zeolite catalyst bed to assess hydrocarbon yields.

In presence of ex-situ catalyst, high yield of aromatics, in particular benzene, toluene and

96 xylene was achieved. Besides, the bio-char was N-rich and contamination-free and can be used as soil amendment or fertilizer.

Key words: Microalgae, Fractional pyrolysis, Zeolite catalyst, Bio-oil and Bio-char

5.2 Introduction

Microalgae are primarily comprised of starch, proteins and lipids and are especially attractive biomass resources for biofuel production due to presence of energy dense triglycerides and fatty acids [7]. Traditional approaches of lipid extraction followed by transesterification (or in situ transesterification) can produce fuel from only the biomass lipids [134, 135]. Thermochemical methods, however, can convert all organic components of microalgae into fuel precursors/molecules. In pyrolysis, biomass is thermally degraded in the absence of oxygen to produce gases, liquids (bio-oil), and solids (bio-char) [22, 23].

Since starch, proteins and lipids are all highly thermally labile, pyrolysis is a promising pathway to produce biofuels from microalgae [72, 136]. However, bio-oil from microalgae biomass cannot be directly used as liquid fuel due to its undesirable characteristics such as high heteroatom (O and/or N), water, acidity and thermal/chemical instability [137-139].

These undesirable properties of bio-oil are directly related to microalgae constituents [140].

For instance, pyrolysis of the carbohydrate fraction from microalgae produces oxygenated compounds such as aldehydes, ketones, carboxylic acids, alcohol and water [22].

Furthermore, protein constituents from microalgae pyrolysis are converted into N- compounds, and consequently the bio-oil may contain high nitrogen depending upon the

N content of microalgae feedstock [22, 136, 141]. These oxygen and nitrogen containing compounds that produced from carbohydrate and protein constituents negatively impacts

97 the bio-oil quality. One the other hand, pyrolysis of lipids produces bio-oil that is rich in hydrocarbons [142]. Thus, if pyrolysis volatiles from protein, carbohydrate and lipid constituents could be collected separately, this would allow the downstream processing of each fraction to be tailored to take advantage of the different chemical and physical properties of these fractions (protein, carbohydrate and lipid).

In recent work, we have observed that the biopolymer components (protein, starch and lipids) volatilize over narrow and distinct temperature regions [143]. This allows recovery of N-containing compounds and carbohydrate products from the biomass by first heating to ~320 °C and holding temperature steady until all protein and starch are thermally degraded or stabilized via polymerization to bio-char. The biomass remaining after thermal removal of protein and starch can be further pyrolyzed at higher temperature to recover bio-oil from lipids that contains much less N- and O-containing compounds [5, 144].

The bio-oil produced from thermal deconstruction of microalgae is typically upgraded through hydrotreating or catalytic cracking to drop-in fuel molecules [145, 146].

The hydrotreating process is typically conducted at elevated temperature (300-450 °C) and pressure (up to 20 MPa) and requires H2. The bio-oil quality improved during the hydrotreating due to decrease in O and N and increase in H/C ratio. However, due to high pressure H2 requirements, hydrotreating can incur high capital cost. Furthermore, catalyst deactivation and low catalyst lifetime (<200h) is a challenge for commercialization of the biomass hydroprocessing [147]. Alternatively, catalytic cracking can deoxygenate the bio- oil to produce hydrocarbons that are compatible with petro-fuels. The reaction is carried out at elevated temperature (400-600 °C) and near-atmospheric pressure over solid acid catalysts [148]. Zeolites such as ZSM-5, SAPO-34, zeolite Y and β have been used for

98 cracking of biomass due to their crystallinity, well-defined pore structures, large surface area, strong acidity and high thermal resistance [93, 149, 150].

Catalytic cracking is often combined with pyrolysis in an in situ reaction system commonly referred to as “catalytic cracking”. While this approach provides intimate contact with pyrolysis vapors as soon as they are produced and results in high yields of desired product, a big challenge in this process is recovery of catalyst after reaction is complete since the catalyst becomes co-mingled with char. Char combustion allows catalyst recovery and is a viable option for lignocellulose pyrolysis, but biochar from microalgae is N-rich and combustion produces significant NOx. Biochar from microalgae is also a valuable source of nutrients and is valuable as a fertilizer.

In this study, we have investigated pyrolytic fractionation coupled with ex-situ catalytic upgrading for producing drop in fuels from microalgae pyrolysis that allows for easy catalyst recovery while also preserving the biochar for applications other than combustion. This approach also provides the opportunity to develop biopolymer-specific upgrading strategies such that catalysts and/or operating conditions for fatty acid-rich bio- oils (from lipid degradation) may be separately optimized from N- and O-rich bio-oils

(from starch and protein degradation). Experiments were performed with Chlorella sorokiniana str SLA-04 in both in-situ and ex-situ catalyst configurations with HZSM-5.

Product yields and compositions from the novel pyrolytic fractionation coupled with ex situ catalytic upgrading were compared with the more traditional in situ catalytic pyrolysis approach.

5.3 Experimental

5.3.1 Materials

Chlorella sorokiniana str. SLA-04, a natural isolate [151], was mixotrophically cultivated in 750L outdoor raceway ponds with the 750 L volume [12]. Stationary phase cultures were centrifuged (2500×g), washed with de-ionized water and freeze-dried

(Labconco Freezone 2.5 L bench-top freeze drying system, Kansas city, MO) to obtain the feedstock used in this study.

NH4-ZSM-5 powder with a SiO2/Al2O3 molar ratio of 23 was purchased from

Zeolyst International, USA. The NH4-ZSM-5 was calcined in a muffle furnace for 5.5 h at

550 ºC to obtain HZSM-5. The texture and acid properties of the HZSM-5 are listed in

Table B1 (Appendix B) and a detailed description of HZSM-5 characterization is provided elsewhere [105].

Hexane, chloroform, methanol and sulfuric acid were purchased from Fisher

Scientific (Pittsburgh, PA, USA). Analytical standards for fatty acids (oleic acid and palmitic acid), glycerides (triolein, diolein, and monolein), fatty acid methyl esters

(FAMEs; mixtures of C8-C22), indole, pyrrole, lauramide, acetic acid, hexanoic acid, furfural, levoglucosan, pentanone, phenol, alkanes (C5, C6, C7, C8 and mixtures of C7-C30), olefins-(Alphagaz PIANO), aromatics-(Alphagaz PIANO), mixtures of benzene, toluene, ethylbenzene and xylene (BTEX) and naphthalene were purchased from Sigma-Aldrich

(St. Louis, MO, USA).

100

5.3.2 Pyroprobe-GC-MS pyrolysis

Pyrolysis GC-MS (using PyroprobeTM or similar instruments) is widely used to assess fast pyrolysis behavior of biomass feedstocks [152-156]. High heat transfer, precise control of temperature and short vapor residence time that are required for fast pyrolysis can all be achieved in the PyroprobeTM instrument [23]. In this study, pyrolysis experiments were performed on a CDS PyroprobeTM 5200 unit (CDS Analytical, Oxford, PA) connected to a Bruker 450 gas chromatograph (GC) equipped with a 300 series mass spectrometer

(MS) (Billerica, MA) and flame ionization detector (FID). A schematic diagram of the experimental set-up is shown in Figure 5.1.

Figure 5.1: Pyroprobe-GC/MS set up.

The sample was loaded in a quartz tube that serves as a micro-pyrolysis reactor.

The quartz tube is placed in a platinum heating element to rapidly heat up the sample.

Moreover, the heating element is connected to a temperature controller to maintain the micro-pyrolysis reactor temperature at the desired set point. The volatiles produced from the pyrolysis of biomass, immediately removed from the reactor (via He as carrier gas) and

101 routed to a trap packed with Tenax® adsorbent material to adsorb the volatiles from pyrolysis reaction. In pyrolysis in the presence of ex-situ catalyst, the volatiles from the biomass pyrolysis first passed through a catalyst bed (maintained at the same temperature as the micro-pyrolysis reactor) and then the upgraded vapors trapped on a Tenax® adsorbent bed. After completion of pyrolysis, the volatiles adsorbed on the trap were desorbed (by increasing the trap temperature) and routed to the GC-MS for analysis. A heated transfer line connected the trap to the GC injector to prevent any condensation of the volatiles.

5.3.3 Experimental procedure

For each pyrolysis experiment 5-7 mg of accurately weighed microalgae biomass was used. For in-situ catalytic pyrolysis, microalgae samples were mixed with HZSM-5 at a biomass/catalyst weight ratio of 1/5 and the mixture was loaded into the PyroprobeTM quartz tube. A same microalgae/catalyst weight ratio was used for ex-situ catalytic pyrolysis, in which the catalyst was packed in the ex-situ reactor connected to the end of the Pyrolyzer section. Each experiment was performed two times and the average values are reported.

In the fractional pyrolysis (in the absence and presence of ex-situ catalyst) experiments, the microalgae was pyrolyzed first at 320 °C for 10 min. For experiments in the presence of ex-situ catalyst, the volatiles from the first fraction passed through a catalyst bed that maintained at 320 °C. The bio-oil from the first fraction then transferred to the

GC-MS for analysis. After completion of first pyrolysis fraction, the reactor cooled down and the quartz tube that contains the solid residue was removed from the PyroprobeTM and

102 weighed to measure the bio-char. Afterward, the solid residue was pyrolyzed again at 450

°C and the bio-oil from the second fraction was analyzed by GC-MS. At the end, the residue from the second fraction was removed from the PyroprobeTM and weighed. The solid residue fractions were weighed by an analytical balance (XP6; Mettler Toledo, USA) with ± 0.01 mg accuracy.

The yield of pyrolysis products were calculated as

WBio−oil YBio−oil = × 100 (5.1) WBiomass

WBio−char YBio−char = × 100 (5.2) WBiomass where WBio−oil were calculated by adding the concentration of different chemical compounds in the bio-oil that measured by GC-MS (see section 5.3.4.2.1), WBio−char is ash-free bio-char weight that measured gravimetrically and WBiomass is dry and ash-free biomass weight. To avoid absorption of room moisture by the biomass (and/or bio-char) during the gravimetrical measurement, all weight analysis performed immediately after the biomass (and/or bio-char) exposure to the room atmosphere.

5.3.4 Analytical method

5.3.4.1 Feedstock characterization

Proximate analysis was performed using dried biomass to measure volatile matter, fixed carbon and ash content of biomass. Volatile matter content was determined using a thermo-gravimetric analyzer (SDT Q600 series analyzer, TA Instruments, Schaumburg,

IL) by measuring weight loss after heating biomass samples under N2 atmosphere from room temperature to 575 °C at a temperature ramp rate of 10 °C min-1 followed by holding

103 the biomass temperature constant at 575 °C for 7 min. Ash content (푓푎푠ℎ) was measured by heating the oven dried biomass at 575 °C for 24 h in a muffle furnace. The fixed carbon fraction was calculated by subtracting the volatile matter and ash (in percentage) from 100.

Elemental analysis was performed using a Thermo Scientific Flash 2000 Organic

Elemental Analyzer equipped with autosampler to measure C, H and N. Biomass and bio- char samples were analyzed for C, H, and N content.

Microalgae lipids (푓푙𝑖푝𝑖푑) were quantified as fatty acid methyl esters (FAMEs) using an in situ transesterification method [151]. Protein content (푓푝푟표푡푒𝑖푛) was calculated from multiplying elemental nitrogen content by a factor of 6.25 [157]. The carbohydrates mass fraction (푓푐푎푟푏) was obtained by Equation 5.3.

푓푐푎푟푏 = 100 ̶ 푓푙𝑖푝𝑖푑 ̶ 푓푝푟표푡푒𝑖푛 ̶ 푓푎푠ℎ (5.3)

5.3.4.2 Analysis of the pyrolysis products

5.3.4.2.1 Gas chromatography (GC) analysis

GC-MS (Bruker, 450-GC equipped with 300-MS) analysis was performed to identify and quantify the chemical compounds in the pyrolysis products. An Agilent DB-

5MS fused silica capillary column (length: 30 m, ID: 0.25 mm, and film thickness: 0.25

µm; Agilent Technologies, Santa Clara, CA) was employed. The injector temperature was held at 300 °C and a split ratio of 1:100 was maintained during the analysis. Helium was used as carrier gas and column flow was constant at 1.0 mL min-1. The column temperature was held constant at 30 °C for 10 min, and then heated at a temperature ramp of 10 °C min-

1 to 300 °C and hold for 10 min. The transfer line, ion source, and manifold were maintained at 300, 150, and 40 °C, respectively. Chemical compounds corresponding to chromatogram

104 peaks were identified using the NIST2008 mass spectral database. A minimum 70% confidence level was used as a threshold for positive identification of IDs provided by the spectral analysis software. Concentration of chemical compounds was estimated based on calibration curves developed using corresponding external analytical standards. The chemicals in bio-oil that had less than 70% confidence level defined as “unidentified” and their concentration was estimated using an average slope of the calibration curves developed for identified chemicals.

5.3.4.2.2 Calorific value

The higher heating values (HHV) of the biomass and bio-char were calculated using empirical Equations 5.4 and 5.5, that were previously reported from experimentally measured calorific values of several biomass [158]. In this study, HHV of biomass and bio- char presented from average of the OLS and PLS values.

HHV (OSL) = 1.87C2 ̶ 144C ̶ 2820H + 63.8C×H + 129N + 20147 (5.4)

HHV (PLS) = 5.22C2 ̶ 319C ̶ 1647H + 38.6C×H + 133N +21028 (5.5)

Where, C, H and N denote the mass fractions (range of mass fraction is 0-100%) of carbon, hydrogen and nitrogen within the sample, measured from the elemental analysis. The HHV retrieved from the Equations 5.4 and 5.5 is in kJ kg-1.

105

5.4 Results and Discussion

5.4.1 Feedstock characterization

Table 5.1 shows the results from composition, proximate and ultimate analysis of microalgae. For later comparison of bio-oil and bio-char yields and compositions, data for other biomass feedstocks – a terrestrial biomass feedstock (woodchips) and soybean flake, are also shown in Table 1. These data were taken from our previous studies [72, 104]. As observed, the microalgae used in this study, contains nearly 38% lipid, 27% carbohydrates and 16% proteins. Soybean flake has similar carbohydrate, but less lipid and more protein than microalgae [104]. Woodchips has a much higher carbohydrate content than microalgae or soybean flakes and also contain lignin, rather than lipid, but C and H content of woodchips and microalgae were similar. The biomass ultimate analysis also shows that microalgae contains higher N content due to protein and less O due to lower carbohydrates than woodchips. The proximate analysis of the tested biomass indicate nearly similar volatile matter, however, microalgae has lower fixed carbon and higher ash. High ash content in microalgae can catalyze bio-char formation during pyrolysis [159].

106

Table 5.1: Compositional, proximate and ultimate analysis of feedstocks. Carbohydrates include cellulose and hemicellulose in the woodchips and starch in soybean flake and Chlorella sp. microalgae. Woodchips consisted of lignin instead of lipid. Mass fraction of oxygen was calculated by difference.

*: Dry basis #: Lignin equal to 23.7% b: Data taken from reference [72].

5.4.2 Single step pyrolysis

From previous studies, pyrolysis of microalgae mostly performed in the presence of in-situ catalyst, where catalyst and biomass were mixed. From this approach, the bio- char cannot be recovered and has to be combusted to regenerate the catalyst and reuse it.

While in lignocellulosic biomass, combustion of bio-char can provide heat, combustion of bio-char from microalgae produces significant NOx due to the presence of N in the microalgae. However, pyrolysis of microalgae in the presence of ex-situ catalyst can produce catalyst-free bio-char that can be readily recovered. As such, single step pyrolysis of microalgae in the presence of ex-situ catalyst was performed and the products yields and compositions were compared with the in-situ catalyst and catalysts free pyrolysis.

107

5.4.2.1 Products yields

Figure 5.2 shows the bio-oil (Figure 5.2a) and bio-char (Figure 5.2b) yields from pyrolysis of microalgae in absence, and presence of in-situ and ex-situ catalyst. It should be noted that in this report all the bio-oil yields are water free since the trap on the

PyroprobeTM instrument only absorbs the organic compounds. Bio-oil yield from catalyst free pyrolysis and at 550 °C was as high as 62%. Compared to previous studies with

Lyngbya sp., Cladophora sp. [72] and soybean flakes [104] the non-catalytic pyrolysis yields were higher likely due to the higher lipid content of the Chlorella sp. feedstock (see

Table 5.1) used here. In the presence of HZSM-5 catalyst, the bio-oil yields were lower, but expected due to higher cracking activities that results in production of small molecules

(e.g. C1-C4) that form non-condensable gases. It is interesting to note that the bio-oil yields from ex-situ and in-situ catalytic processes were relatively similar. These results suggest that an ex-situ catalysis approach would not compromise bio-oil production while simultaneously preserving biochar and facilitating easy catalyst reuse.

The bio-char yields decreased by increasing reaction temperature, which is similar to our previous study on pyrolysis of soybean flakes in a fluidized-bed flash pyrolysis

[104]. The low bio-char (15-20%) and high bio-oil (50-62%) yields confirm that Pyroprobe can be a reliable instrument to simulate the fast pyrolysis conditions, where high heat transfer between reactor and biomass and short vapor residence time result in low biochar/coke and high bio-oil yields, respectively. Pyrolysis reaction with in-situ catalyst showed higher bio-char compared with the one in absence of catalyst, probably due to strong acidity of the catalysts that promotes bio-char/coke formation. However, in pyrolysis with ex-situ catalyst, biomass is not in contact with catalysts, in contrast to in-

108 situ catalyst where catalyst and biomass are mixed, thus same bio-char were obtained compared with the one from catalyst free pyrolysis. The catalysts used on the pyrolysis in presence of ex-situ catalyst were also recovered and combusted on the TGA instrument at

600 °C under air atmosphere to measure the coke deposited on the catalyst. The TGA confirmed that only small amount of coke (2-4 wt. % relative to biomass) were deposited on the catalyst surface and/or within the pores. Low coke formation on the HZSM-5, would increase the catalyst lifetime and prevent frequent regeneration that would be required when in-situ catalyst was used.

109

( (a) 80 a) Bio-oil

free dry dry free - 450 °C 40 500 °C

microalgae) 550 °C

20 Yield (wt. % of of % ash (wt. Yield

0 No catalyst In-situ catalyst Ex-situ catalyst

(b) 30 Bio-char

25 free dry dry free

- 20 450 °C 15 500 °C

microalgae) 10 550 °C

5 Yield (wt. % of of % ash (wt. Yield

0 No catalyst In-situ catalyst Ex-situ catalyst Figure 5.2: (a) Bio-oil and (b) bio-char yields from single step pyrolysis of microalgae in absence and presence of HZSM-5 catalyst at tested temperatures. The error bars denote the standard deviation from two experiments.

5.4.2.2 Bio-oil compositions

The bio-oil compositions for the pyrolysis reactions without- and with- catalyst is shown in Table 5.2 and their GC-MS chromatograms displayed in Figure D1. In the absence of catalyst, the bio-oil was comprised of oxygenated compounds such as acetic

110 acid, ketone, aldehyde and furan that likely produced from decomposition of carbohydrate constituents; N-compounds such as pyrroles, indole, pyrazoles, oleoamides and fatty nitriles produced from protein decomposition; as well as glyceride, fatty ester, aliphatic and fatty acids (mainly octadecanoic, hexadecanoic, heptanoic and octanoic acid) that are mostly produced from volatilization and/or decomposition of lipid constituents in microalgae. Moreover, by increasing the pyrolysis temperature, glyceride yields decreased, but carboxylic acid and aliphatic content increased, due to more decomposition of glyceride at higher temperatures [142].

Compound Catalyst-free pyrolysis Pyrolysis in presence of catalyst

In-situ Ex-situ

T (°C) → 450 500 550 450 500 550 450 500 550 Benzene 0.8 1.5 1.7 2.8 4.9 5.2 3.8 6.8 7.6 Toluene 0.5 0.5 0.8 7.3 9.8 11.8 9.2 13.0 14.2 Ethylbenzene - - - 1.6 1.8 1.9 0.2 0.4 0.5 Xylene 0.2 0.1 0.2 5.3 7.0 7.8 5.2 6.1 7.4 C8+ aromatic - - - 4.2 4.8 4.8 1.3 0.5 0.5 Naphthalene - - - 4.5 3.9 4.0 2.3 4.0 5.1 Aliphatic 9.3 14.8 24.1 5.8 4.3 3.8 4.8 3.1 2.0 Indane - - - 0.4 0.5 0.4 - - - Indene - - - 0.4 0.6 0.6 - - - Acetic acid 2.4 1.6 1.1 - - - - - Carboxylic acid 7.9 13.9 8.1 1.3 0.6 - 0.9 - - Ketone/Aldehyde 4.5 4.2 3.6 0.6 0.6 0.3 0.5 0.1 - Furan 0.7 1.4 3.7 0.1 0.5 1.3 0.4 0.6 - N-compound 1.1 2.1 3.5 ------Alcohol 1.9 1.7 2.3 ------Fatty ester 1.5 1.3 2.9 ------Glyceride 15.4 10.5 4.3 ------BTX# 1.5 2.1 2.7 15.3 21.7 24.7 18.2 25.9 29.2 Total aromatics 1.5 2.1 2.7 26.4 33.3 36.4 22.1 30.7 35.3 Unidentified 4.2 4.5 5.4 0 0 0 0 0 0 #: Benzene, toluene and xylene

In the presence of HZSM-5 (in-situ or ex-situ), the bio-oil was mainly comprised of aromatics such as benzene, toluene and xylene (BTX), C9-C10 alkylaromatics and

111 naphthalene, aliphatic (C5-C18) and small amount of oxygenated compounds such as phenol, aldehyde, furan and fatty acids. N-compounds were not observed in the bio-oil.

Formation of aromatics occurs due to Diels-Alder reaction and/or intramolecular radical cyclization. In Diels-Alder reaction, a diene and alkene reacts and form polysubstituted cyclohexenes and then via dehydrogenation polysubstituted aromatic produce [78]. The strong Brønsted acid sites in the HZSM-5 enables the oligomerization of light olefins

(produced for dehydration, decarbonylation, decarboxylation and decomposition of microalgae volatiles) to form C4-C10 olefins which then dehydrogenate to form dienes.

Thereafter, dienes and olefins undergo cyclization and dehydrogenation to form aromatics.

The aromatics’ yields were higher at higher reaction temperature. In particular, the

BTX yields increased from 15 to 25% in in-situ catalyst and up to 29% in ex-situ catalytic process when temperature increased from 450 °C to 550 °C. Interestingly, the BTX yields are higher when ex-situ catalyst was used and there was no evidence of oxygenated compounds under these conditions at 550 °C. In ex-situ catalyst configuration, the HZSM-

5 is expected to remain active for much longer reaction time since the bio-char and catalysts are in separate reactors. Under these reaction conditions, the catalyst acid surface and inner pores are more accessible and active for breaking down the volatiles (that produced from thermal decomposition of the biomass) followed by aromatization/dehydrogenation to produce aromatics. Figure D2 shows the selectivity of benzene, toluene and xylene from microalgae pyrolysis in presence of in-situ and ex-situ catalyst at temperature range of 450-

550 °C. As observed, selectivity of benzene and toluene is higher when an ex-situ catalyst configuration was employed. This can be attributed to the less catalyst deactivation during the ex-situ catalytic process. The BTX contribute to the large world market for commodity

112 chemicals and have a diverse uses across several industries. For instance, benzene is used as precursor for styrene, phenol, nylon and aniline production; toluene is blended into unleaded gasoline to improve the octane number; and xylene is used to produce polyethylene terephthalate (PET) and resins.

5.4.2.3 Bio-char elemental analysis

Bio-char from catalyst free process is similar to the one from ex-situ catalyst process (as expected) and under in-situ catalyst process, it is difficult to separate bio-char from catalyst since they are mixed. Thus, bio-char from ex-situ catalyst process is shown in Table 5.3. As observed, the N content in biochar decreased at higher temperature possibly due to cracking of the C-N bonds and release of ammonia. The N mass balance indicates that 16-37% (16% at 550 °C and 37% at 450 °C) of the N content in microalgae was fixed in bio-char in forms of ammonium, nitrate or high molecular weight N- heterocyclic compounds condensed on the bio-char surface or inside the bio-char pores

[22]. Moreover, there was no evidence of N-compounds in the bio-oil, thus the remaining

N likely ended up in non-condensable gases in the forms of ammonia. The C balance shows that more than 70% of C in microalgae recovered in bio-oil and bio-char. The bio-char from ex-situ catalytic process is catalyst-free and contains high amounts of macro- and micro-nutrients (e.g. N, P, K, N, S and Ca) and also has C/N ratio from 17-27%, this ratio is proper for bio-char to serve as soil amendment or fertilizer [160, 161].

113

Table 5.3: Elemental analyses of bio-char obtained from microalgae pyrolysis in presence of ex-situ catalyst. ‘‘Dry-basis’’ values were obtained by CHN analyzer. ‘‘Dry-ash free basis’’ values were calculated by using ‘‘dry-basis’’ values, and ash content. Calorific values (HHV) were calculated using Equations 5.4 and 5.5. All values are reported as mass fractions (%). Mass fraction of oxygen was calculated by difference.

5.4.3 Fractional pyrolysis of microalgae

Starch and protein volatilize/deconstruct at temperature ranges of 160-340 °C, however, microalgae lipid volatilize/degrade at higher temperature (see Figure D3a). This distinct volatilization temperature of microalgae’s constituents allows collecting the products from starch/protein degradation separately from lipid. Moreover, each fraction can be upgraded separately via ex-situ catalytic process. To demonstrate the fractionation of microalgae, a two-step fractional pyrolysis of Chlorella sorokiniana str. SLA-04 microalgae was performed in absence and presence of ex-situ HZSM-5 catalyst. For experiments in the presence of ex-situ catalyst, the microalgae volatiles produced from each fraction were passed through a catalyst bed that was also maintained at the same temperature as the pyrolysis reactor. A microalgae/HZSM-5 weight ratio of 1/5 was also used for these experiments.

114

5.4.3.1 Products yields

Figure 5.3 shows the products yields from fractional pyrolysis (Figure 5.3a) and fractional pyrolysis with ex-situ conversion of bio-oil vapors (Figure 5.3b). From catalyst free pyrolysis, nearly 56% bio-char and 15% bio-oil was produced from the first fraction

(320 °C). A similar residue weight fraction (~60%) was observed when the microalgae samples were pyrolyzed on a TGA instrument under similar conditions (see Figure D3b).

However, when temperature increased to 450 °C (2nd fraction), the bio-char yield decreased to 23% and the bio-oil increased to ~32%. From first and second fractions, a total 47% bio- oil was produced, which is close to the amount of bio-oil (~50%) achieved from single step microalgae pyrolysis at 450 °C (see Figure 5.2a). Fractional pyrolysis with ex-situ catalyst achieved a similar bio-char yield as the one without catalyst. Moreover, 11% and 20% bio- oil was obtained from the first and second fractions, respectively. Overall, ~ 31% bio-oil was obtained from microalgae fractional pyrolysis in presence of ex-situ catalyst. This total bio-oil is relatively similar to the single step pyrolysis with ex-situ catalyst (see Figure

5.2a). Besides, the fractional pyrolysis in presence of ex-situ catalyst would provide a flexible downstream upgrading process, where each fraction can be upgraded separately.

115

80 ( (a) b)

60 Bio-char free dry dry free - Bio-oil

40 microalgae)

20 Yield (wt. % of of ash % (wt. Yield

0 st nd 1 stfraction fraction 2nd2 fraction fraction

80 (b)

free dry dry free Bio-char - Bio-oil

40 microalgae)

20 Yield (wt. % of of % ash (wt. Yield

0 11st stfraction fraction 22ndnd fraction fraction Figure 5.3: Products yields from fractional (a) non-catalytic and (b) catalytic pyrolysis.

5.4.3.2 Products compositions

The bio-oil composition from each pyrolysis fraction is shown in Table 5.4 and their GC-MS chromatograms depicted in Figures D4 (catalyst free) and D5 (ex-situ catalyst). Bio-oil from first fraction (320 °C) of catalyst free pyrolysis was mainly comprised oxygenated compounds (e.g. acetic acid, ketone, aldehyde and furan), N- compounds (e.g. indole and pyrrole) and small amount of hydrocarbons. However, bio-oil

116 from the second fraction contained mostly chemical compounds that derived from lipid volatilization/degradation such as carboxylic acid (C16-C18 fatty acids), glyceride and hydrocarbons. This indicates majority of the lipid fraction in microalgae collected in the second fraction.

Table 5.4: Bio-oil composition from fractional pyrolysis. The values are average of two experiments and based on weight percentage relative to dry and ash-free biomass.

Bio-oil from fractional pyrolysis in presence of ex-situ catalyst was mainly contained C6-C12 aromatics, in particular benzene, toluene, xylene and naphthalene. The chemical compounds in first and second fractions were relatively similar, but the aromatic yields increased in the second fraction due to higher reaction temperature and more volatilization/degradation of microalgae lipid. Moreover, the BTX selectivity in the bio-oil from the second fraction was higher than the first fraction (see Figure D6). From first and second fraction, more than 17% BTX (relative to ash-free dry microalgae) was produced.

117

A similar BTX yield was also achieved from the single-step pyrolysis with ex-situ catalyst.

This demonstrates that integration of fractional pyrolysis with downstream ex-situ catalytic upgrading allows carrying out the desired reaction chemistries on the vapor produced from each fraction.

Table 5.5: Elemental analyses of bio-char obtained from fractional pyrolysis of algae. ‘‘Dry-basis’’ values were obtained by CHN analyzer. ‘‘Dry-ash free basis’’ values were calculated by using ‘‘dry-basis’’ values, and ash content. Calorific values (HHV) were calculated using Equations 5.4 and 5.5. All values are reported as mass fractions (%). Mass fraction of oxygen was calculated by difference.

Table 5.5 shows the ultimate analysis of the bio-char from fractional pyrolysis of microalgae in presence of ex-situ catalyst. As observed, the bio-char from the first fraction has higher HHV than the second fraction due to higher C content. The bio-char from the first fraction contained most of the C and N in the microalgae. However, only 18% of the original microalgae C ended up in the bio-char from the second fraction. Moreover, the O content of bio-char decreased in second fraction, while the N remained relatively constant between fractions.

118

5.5 Conclusion

A two-step fractional pyrolysis of microalgae integrated with ex-situ catalyst was performed to upgrade the volatiles from microalgae’s constituents separately. With no catalyst, the bio-oil from the first fraction was mainly comprised of O-containing compounds, but the bio-oil from the second fraction was mainly glycerides, fatty acids and aliphatic, that likely were produced from volatilization and/or degradation of microalgae lipid. When the volatiles from each fraction were upgraded with an ex-situ catalyst, a total

32% water-free bio-oil yield was achieved and the majority of the bio-oil was aromatics

(C6-C12) such as benzene, toluene, xylene and naphthalene. Moreover, 22% bio-char was produced that was contamination-free and N-rich that can be used as soil amendment or fertilizer. The fractional pyrolysis of microalgae integrated with the ex-situ catalytic process shows promise to produce high value liquid products and bio-char from microalgae.

119

References

[1] J. Cheng, Biomass to Renewable Energy Processes, in, CRC press, Taylor& Francis

Group, 2010.

[2] U.S. DOE, U.S. Billion-Ton Update: Biomass Supply for a Bioenergy and Bioproducts

Industry, in, R.D. Perlack and B.D. Stokes, Oak Ridge National Laboratory, Oak Ridge,

TN, 2011, pp. 227.

[3] U. Doe, National algal biofuels technology roadmap, US Department of Energy, Office of Energy Efficiency and Renewable Energy, Biomass Program, (2010).

[4] N.R. Council, Renewable Fuel Standard: Potential Economic and Environmental

Effects of U.S. Biofuel Policy, The National Academies Press, Washington, DC, 2011.

[5] B. Maddi, Pyrolysis Strategies for Effective Utilization of Lignocellulosic and Algal

Biomass, in, University of Toledo, 2014.

[6] Alternative Energy: Biomass Energy, in, 2017.

[7] D.R. Georgianna, S.P. Mayfield, Exploiting diversity and synthetic biology for the production of algal biofuels, Nature, 488 (2012) 329-335.

[8] J. Ferrell, V. Sarisky-Reed, National algal biofuels technology roadmap, in, EERE

Publication and Product Library, 2010.

[9] P. Duan, B. Jin, Y. Xu, Y. Yang, X. Bai, F. Wang, L. Zhang, J. Miao, Thermo-chemical conversion of Chlorella pyrenoidosa to liquid biofuels, Bioresource Technology, 133

(2013) 197-205.

[10] J.B.K. Park, R.J. Craggs, A.N. Shilton, Wastewater treatment high rate algal ponds for biofuel production, Bioresource Technology, 102 (2011) 35-42.

120

[11] A. Parmar, N.K. Singh, A. Pandey, E. Gnansounou, D. Madamwar, Cyanobacteria and microalgae: A positive prospect for biofuels, Bioresource Technology, 102 (2011)

10163-10172.

[12] A. Vadlamani, Enhanced biomass and lipid productivities of outdoor alkaliphilic microalgae cultures through increased media alkalinity, in, University of Toledo, 2016.

[13] A.K. Vadlamani, P. Relue, S. Viamajala, H. Shao, S. Varanasi, Enzymatic Digestion of Microalgal Biomass for Lipid, Sugar, and Protein Recovery, in, Google Patents, 2016.

[14] P.M. Bollin, S. Viamajala, Reactive Extraction of Triglycerides as Fatty Acid Methyl

Esters using Lewis Acidic Chloroaluminate Ionic Liquids, Energy & Fuels, 26 (2012)

6411-6418.

[15] M. Hannon, J. Gimpel, M. Tran, B. Rasala, S. Mayfield, Biofuels from algae: challenges and potential, Biofuels, 1 (2010) 763-784.

[16] J. Stefels, Physiological aspects of the production and conversion of DMSP in marine algae and higher plants, Journal of Sea Research, 43 (2000) 183-197.

[17] T.M. Brown, P. Duan, P.E. Savage, Hydrothermal Liquefaction and Gasification of

Nannochloropsis sp, Energy & Fuels, 24 (2010) 3639-3646.

[18] D.C. Elliott, P. Biller, A.B. Ross, A.J. Schmidt, S.B. Jones, Hydrothermal liquefaction of biomass: Developments from batch to continuous process, Bioresource Technology, 178

(2015) 147-156.

[19] K. Wang, R.C. Brown, Catalytic pyrolysis of microalgae for production of aromatics and ammonia, Green Chemistry, 15 (2013) 675-681.

[20] G.W. Huber, S. Iborra, A. Corma, Synthesis of transportation fuels from biomass: chemistry, catalysts, and engineering, Chemical reviews, 106 (2006) 4044-4098.

121

[21] D.C. Elliott, Review of recent reports on process technology for thermochemical conversion of whole algae to liquid fuels, Algal Research, 13 (2016) 255-263.

[22] K. Wang, R.C. Brown, S. Homsy, L. Martinez, S.S. Sidhu, Fast pyrolysis of microalgae remnants in a fluidized bed reactor for bio-oil and biochar production,

Bioresource Technology, 127 (2013) 494-499.

[23] B.V. Babu, Biomass pyrolysis: A state-of-the-art review, Biofuels, Bioproducts and

Biorefining, 2 (2008) 393-414.

[24] A. Demirbas, Political, economic and environmental impacts of biofuels: A review,

Applied Energy, 86, Supplement 1 (2009) S108-S117.

[25] L. Labecki, A. Cairns, J. Xia, A. Megaritis, H. Zhao, L.C. Ganippa, Combustion and emission of rapeseed oil blends in diesel engine, Applied Energy, 95 (2012) 139-146.

[26] E. Buzetzki, K. Sidorová, Z. Cvengrošová, J. Cvengroš, Effects of oil type on products obtained by cracking of oils and fats, Fuel Processing Technology, 92 (2011) 2041-2047.

[27] S. Yan, C. DiMaggio, H. Wang, S. Mohan, M. Kim, L. Yang, S. O. Salley, K. Y.

Simon Ng, Catalytic Conversion of Triglycerides to Liquid Biofuels Through

Transesterification, Cracking, and Hydrotreatment Processes, Current Catalysis, 1 (2012)

41-51.

[28] V.R. Wiggers, G.R. Zonta, A.P. Franca, D.R. Scharf, E.L. Simionatto, L. Ender, H.F.

Meier, Challenges associated with choosing operational conditions for triglyceride thermal cracking aiming to improve biofuel quality, Fuel, 107 (2013) 601-608.

[29] A. Corma, S. Iborra, A. Velty, Chemical routes for the transformation of biomass into chemicals, Chemical Reviews, 107 (2007) 2411-2502.

122

[30] A.Y. Mudiyanselage, S. Viamajala, S. Varanasi, K. Yamamoto, Simple Ring-Closing

Metathesis Approach for Synthesis of PA11, 12, and 13 Precursors from Oleic Acid, Acs

Sustainable Chemistry & Engineering, 2 (2014) 2831-2836.

[31] Oilseeds: World Markets and Trade, in, 2017.

[32] D. Chiaramonti, A.M. Rizzo, A. Spadi, M. Prussi, G. Riccio, F. Martelli, Exhaust emissions from liquid fuel micro gas turbine fed with diesel oil, biodiesel and vegetable oil, Applied Energy, 101 (2013) 349-356.

[33] S. estari , P. Mäki-Arvela, J. Beltramini, G.Q.M. Lu, D.Y. Murzin, Transforming

Triglycerides and Fatty Acids into Biofuels, ChemSusChem, 2 (2009) 1109-1119.

[34] K.D. Maher, D.C. Bressler, Pyrolysis of triglyceride materials for the production of renewable fuels and chemicals, Bioresource Technology, 98 (2007) 2351-2368.

[35] A. Demirbas, Biodiesel production from vegetable oils via catalytic and non-catalytic supercritical methanol transesterification methods, Progress in Energy and Combustion

Science, 31 (2005) 466-487.

[36] A.W. Schwab, M.O. Bagby, B. Freedman, Preparation and properties of diesel fuels from vegetable oils, Fuel, 66 (1987) 1372-1378.

[37] A.-M. Williamson, O. Badr, Assessing the viability of using rape methyl ester (RME) as an alternative to mineral diesel fuel for powering road vehicles in the UK, Applied

Energy, 59 (1998) 187-214.

[38] R. Gautam, N. Kumar, H.S. Pali, P. Kumar, Experimental studies on the use of methyl and ethyl esters as an extender in a small capacity diesel engine, Biofuels, (2016) 1-10.

[39] U. Schuchardt, R. Sercheli, R.M. Vargas, Transesterification of vegetable oils: a review, Journal of the Brazilian Chemical Society, 9 (1998) 199-210.

123

[40] A. Demirbas, Importance of biodiesel as transportation fuel, Energy Policy, 35 (2007)

4661-4670.

[41] E. Lotero, Y. Liu, D.E. Lopez, K. Suwannakarn, D.A. Bruce, J.G. Goodwin, Synthesis of Biodiesel via Acid Catalysis, Industrial & Engineering Chemistry Research, 44 (2005)

5353-5363.

[42] S.J. Frey, G.W. Fichtl, UOP–A Honeywell Company, Des Plaines, IL, USA, (2014).

[43] A. Demirbaş, Biodiesel fuels from vegetable oils via catalytic and non-catalytic supercritical alcohol transesterifications and other methods: a survey, Energy Conversion and Management, 44 (2003) 2093-2109.

[44] R. Wang, B. Song, W. Zhou, Y. Zhang, D. Hu, P.S. Bhadury, S. Yang, A facile and feasible method to evaluate and control the quality of Jatropha curcus L. seed oil for biodiesel feedstock: Gas chromatographic fingerprint, Applied Energy, 88 (2011) 2064-

2070.

[45] M. Stumborg, A. Wong, E. Hogan, Hydroprocessed vegetable oils for diesel fuel improvement, Bioresource Technology, 56 (1996) 13-18.

[46] D.G. Lima, V.C.D. Soares, E.B. Ribeiro, D.A. Carvalho, É.C.V. Cardoso, F.C. Rassi,

K.C. Mundim, J.C. Rubim, P.A.Z. Suarez, Diesel-like fuel obtained by pyrolysis of vegetable oils, Journal of Analytical and Applied Pyrolysis, 71 (2004) 987-996.

[47] P.D. Patil, S. Deng, Optimization of biodiesel production from edible and non-edible vegetable oils, Fuel, 88 (2009) 1302-1306.

[48] F. Pinto, S. Martins, M. Gonçalves, P. Costa, I. Gulyurtlu, A. Alves, B. Mendes,

Hydrogenation of rapeseed oil for production of liquid bio-chemicals, Applied Energy, 102

(2013) 272-282.

124

[49] A.W. Schwab, G.J. Dykstra, E. Selke, S.C. Sorenson, E.H. Pryde, Diesel fuel from thermal decomposition of soybean oil, J Am Oil Chem Soc, 65 (1988) 1781-1786.

[50] J.W. Alencar, P.B. Alves, A.A. Craveiro, Pyrolysis of tropical vegetable oils, Journal of Agricultural and Food Chemistry, 31 (1983) 1268-1270.

[51] I.C.P. Fortes, P.J. Baugh, Study of analytical on-line pyrolysis of oils from Macauba fruit (Acrocomia sclerocarpa M) via GC/MS, Journal of the Brazilian Chemical Society,

10 (1999) 469-477.

[52] I.C.P. Fortes, P.J. Baugh, Study of calcium soap pyrolysates derived from Macauba fruit (Acrocomia sclerocarpa M.). Derivatization and analysis by GC/MS and CI-MS,

Journal of Analytical and Applied Pyrolysis, 29 (1994) 153-167.

[53] R.O. Idem, S.P.R. Katikaneni, N.N. Bakhshi, Thermal Cracking of Canola Oil:

Reaction Products in the Presence and Absence of Steam, Energy & Fuels, 10 (1996) 1150-

1162.

[54] H.D. Almeida, O.A. Correa, J.G. Eid, H.J. Ribeiro, D.R. de Castro, M.S. Pereira, L.M.

Pereira, A.D. Mancio, M.C. Santos, J.A.D. Souza, L.E.P. Borges, N.M. Mendonca, N.T.

Machado, Production of biofuels by thermal catalytic cracking of scum from grease traps in pilot scale, Journal of Analytical and Applied Pyrolysis, 118 (2016) 20-33.

[55] C.-C. Chang, S.-W. Wan, China's Motor Fuels from Tung Oil, Industrial &

Engineering Chemistry, 39 (1947) 1543-1548.

[56] A. Kubatova, Y. Luo, J. Stavova, S.M. Sadrameli, T. Aulich, E. Kozliak, W. Seames,

New path in the thermal cracking of triacylglycerols (canola and soybean oil), Fuel, 90

(2011) 2598-2608.

125

[57] A. Bridgwater, G. Peacocke, Fast pyrolysis processes for biomass, Renewable and

Sustainable Energy Reviews, 4 (2000) 1-73.

[58] A. Trubetskaya, P.A. Jensen, A.D. Jensen, A.D.G. Llamas, K. Umeki, D. Gardini, J.

Kling, R.B. Bates, P. Glarborg, Effects of several types of biomass fuels on the yield, nanostructure and reactivity of soot from fast pyrolysis at high temperatures, Applied

Energy, 171 (2016) 468-482.

[59] X.S. Zhang, H.W. Lei, L. Zhu, M. Qian, X.L. Zhu, J. Wu, S.L. Chen, Enhancement of jet fuel range alkanes from co-feeding of lignocellulosic biomass with plastics via tandem catalytic conversions, Applied Energy, 173 (2016) 418-430.

[60] V.R. Wiggers, A. Wisniewski Jr, L.A.S. Madureira, A.A.C. Barros, H.F. Meier,

Biofuels from waste fish oil pyrolysis: Continuous production in a pilot plant, Fuel, 88

(2009) 2135-2141.

[61] A. Wisniewski Jr, V.R. Wiggers, E.L. Simionatto, H.F. Meier, A.A.C. Barros, L.A.S.

Madureira, Biofuels from waste fish oil pyrolysis: Chemical composition, Fuel, 89 (2010)

563-568.

[62] V.R. Wiggers, H.F. Meier, A. Wisniewski Jr, A.A. Chivanga Barros, M.R. Wolf

Maciel, Biofuels from continuous fast pyrolysis of soybean oil: A pilot plant study,

Bioresource Technology, 100 (2009) 6570-6577.

[63] Y. uo, I. Ahmed, A. Kubátová, J. Šťávová, T. Aulich, S.M. Sadrameli, W.S. Seames,

The thermal cracking of soybean/canola oils and their methyl esters, Fuel Processing

Technology, 91 (2010) 613-617.

126

[64] J.W. Goodrum, D.P. Geller, Rapid thermogravimetric measurements of boiling points and vapor pressure of saturated medium- and long-chain triglycerides, Bioresource

Technology, 84 (2002) 75-80.

[65] D.R. Nelson, S. Viamajala, One-pot synthesis and recovery of fatty acid methyl esters

(FAMEs) from microalgae biomass, Catalysis Today, 269 (2016) 29-39.

[66] A. Dalmoro, A.A. Barba, M. Amore, Analysis of Size Correlations for Microdroplets

Produced by Ultrasonic Atomization, The Scientific World Journal, 2013 (2013) 7.

[67] K. Omer, N. Ashgriz, Spray Nozzles, in: N. Ashgriz (Ed.) Handbook of Atomization and Sprays, Springer US, 2011, pp. 497-579.

[68] R. Rajan, A.B. Pandit, Correlations to predict droplet size in ultrasonic atomisation,

Ultrasonics, 39 (2001) 235-255.

[69] H. Noureddini, B.C. Teoh, L. Davis Clements, Densities of vegetable oils and fatty acids, J. Am. Oil Chem. Soc., 69 (1992) 1184-1188.

[70] B. Esteban, J.-R. Riba, G. Baquero, R. Puig, A. Rius, Characterization of the surface tension of vegetable oils to be used as fuel in diesel engines, Fuel, 102 (2012) 231-238.

[71] O.O. Fasina, Z. Colley, Viscosity and Specific Heat of Vegetable Oils as a Function of Temperature: 35°C to 180°C, International Journal of Food Properties, 11 (2008) 738-

746.

[72] B. Maddi, S. Viamajala, S. Varanasi, Comparative study of pyrolysis of algal biomass from natural lake blooms with lignocellulosic biomass, Bioresource Technology, 102

(2011) 11018-11026.

[73] C. Zhenyi, J. Xing, L. Shuyuan, L. Li, Thermodynamics calculation of the pyrolysis of vegetable oils, Energy Sources, 26 (2004) 849-856.

127

[74] C. Castro, F. Rust, Thermal decomposition of acrolein. The attack of methyl and t- butoxy free radicals on acrolein, Journal of the American Chemical Society, 83 (1961)

4928-4932.

[75] H.F. Meier, V.R. Wiggers, G.R. Zonta, D.R. Scharf, E.L. Simionatto, L. Ender, A kinetic model for thermal cracking of waste cooking oil based on chemical lumps, Fuel,

144 (2015) 50-59.

[76] J.B. Maxwell, L.S. Bonnell, Derivation and Precision of a New Vapor Pressure

Correlation for Petroleum Hydrocarbons, Industrial & Engineering Chemistry, 49 (1957)

1187-1196.

[77] R.L.E.a.J.A.K. Steven C. Cermak, Distillation of Natural Fatty Acids and Their

Chemical Derivatives, in: S. Zereshki (Ed.) Distillation - Advances from Modeling to

Applications, InTech, 2012.

[78] A. Kubátová, J. Š t’ávová, W.S. Seames, Y. Luo, S.M. Sadrameli, M.J. Linnen, G.V.

Baglayeva, I.P. Smoliakova, E.I. Kozliak, Triacylglyceride thermal cracking: pathways to cyclic hydrocarbons, Energy & Fuels, 26 (2011) 672-685.

[79] U. Biermann, U. Bornscheuer, M.A.R. Meier, J.O. Metzger, H.J. Schafer, Oils and

Fats as Renewable Raw Materials in Chemistry, Angew. Chem.-Int. Edit., 50 (2011) 3854-

3871.

[80] B.R. Moser, B.K. Sharma, K.M. Doll, S.Z. Erhan, Diesters from oleic acid: Synthesis, low temperature properties, and oxidation stability, J. Am. Oil Chem. Soc., 84 (2007) 675-

680.

128

[81] G. Ameh Abel, K. Oliver Nguyen, S. Viamajala, S. Varanasi, K. Yamamoto, Cross- metathesis approach to produce precursors of nylon 12 and nylon 13 from microalgae, RSC

Advances, 4 (2014) 55622-55628.

[82] A.E. Atabani, A.S. Silitonga, H.C. Ong, T.M.I. Mahlia, H.H. Masjuki, I.A. Badruddin,

H. Fayaz, Non-edible vegetable oils: A critical evaluation of oil extraction, fatty acid compositions, biodiesel production, characteristics, engine performance and emissions production, Renewable and Sustainable Energy Reviews, 18 (2013) 211-245.

[83] M. Tripathi, A. Mishra, Glucosinolates in animal nutrition: A review, Animal Feed

Science and Technology, 132 (2007) 1-27.

[84] A. Boateng, C. Mullen, N. Goldberg, Producing stable pyrolysis liquids from the oil- seed presscakes of mustard family plants: Pennycress (Thlaspi arvense L.) and camelina

(Camelina sativa), Energy & Fuels, 24 (2010) 6624-6632.

[85] H.L. Dose, C.A. Eberle, F. Forcella, R.W. Gesch, Early planting dates maximize winter annual field pennycress (Thlaspi arvense L.) yield and oil content, Industrial Crops and Products, 97 (2017) 477-483.

[86] G.W. Hergert, J.F. Margheim, A.D. Pavlista, D.L. Martin, T.A. Isbell, R.J. Supalla,

Irrigation response and water productivity of deficit to fully irrigated spring camelina,

Agricultural Water Management, 177 (2016) 46-53.

[87] T.A. Isbell, US effort in the development of new crops (Lesquerella, Pennycress

Coriander and Cuphea), OCL, 16 (2009) 205-210.

[88] M.M. Gui, K.T. Lee, S. Bhatia, Feasibility of edible oil vs. non-edible oil vs. waste edible oil as biodiesel feedstock, Energy, 33 (2008) 1646-1653.

129

[89] H. Zhang, U. Aytun Ozturk, Q. Wang, Z. Zhao, Biodiesel produced by waste cooking oil: Review of recycling modes in China, the US and Japan, Renewable and Sustainable

Energy Reviews, 38 (2014) 677-685.

[90] Z. Yaakob, M. Mohammad, M. Alherbawi, Z. Alam, K. Sopian, Overview of the production of biodiesel from Waste cooking oil, Renewable and Sustainable Energy

Reviews, 18 (2013) 184-193.

[91] Y.-S. Ooi, R. Zakaria, A.R. Mohamed, S. Bhatia, Catalytic conversion of palm oil- based fatty acid mixture to liquid fuel, Biomass and Bioenergy, 27 (2004) 477-484.

[92] R.O. Idem, S.P.R. Katikaneni, N.N. Bakhshi, Catalytic conversion of canola oil to fuels and chemicals: roles of catalyst acidity, basicity and shape selectivity on product distribution, Fuel Processing Technology, 51 (1997) 101-125.

[93] G.W. Huber, A. Corma, Synergies between Bio‐and Oil Refineries for the Production of Fuels from Biomass, Angewandte Chemie International Edition, 46 (2007) 7184-7201.

[94] A. Corma, G.W. Huber, L. Sauvanaud, P. O'Connor, Processing biomass-derived oxygenates in the oil refinery: Catalytic cracking (FCC) reaction pathways and role of catalyst, Journal of Catalysis, 247 (2007) 307-327.

[95] F.A. Twaiq, N.A. Zabidi, S. Bhatia, Catalytic conversion of palm oil to hydrocarbons: performance of various zeolite catalysts, Industrial & engineering chemistry research, 38

(1999) 3230-3237.

[96] V. Abbasov, T. Mammadova, N. Andrushenko, N. Hasankhanova, Y. Lvov, E.

Abdullayev, Halloysite–Y-zeolite blends as novel mesoporous catalysts for the cracking of waste vegetable oils with vacuum gasoil, Fuel, 117 (2014) 552-555.

130

[97] S.P.R. Katikaneni, J.D. Adjaye, N.N. Bakhshi, Studies on the Catalytic Conversion of

Canola Oil to Hydrocarbons: Influence of Hybrid Catalysts and Steam, Energy & Fuels, 9

(1995) 599-609.

[98] S.T. Srinivas, A.K. Dalai, N.N. Bakhshi, Thermal and catalytic upgrading of a biomass‐derived oil in a dual reaction system, The Canadian Journal of Chemical

Engineering, 78 (2000) 343-354.

[99] J.A. Botas, D.P. Serrano, A. García, J. de Vicente, R. Ramos, Catalytic conversion of rapeseed oil into raw chemicals and fuels over Ni- and Mo-modified nanocrystalline ZSM-

5 zeolite, Catalysis Today, 195 (2012) 59-70.

[100] F.A. Twaiq, A.R. Mohamed, S. Bhatia, Liquid hydrocarbon fuels from palm oil by catalytic cracking over aluminosilicate mesoporous catalysts with various Si/Al ratios,

Microporous and Mesoporous Materials, 64 (2003) 95-107.

[101] F.A. Twaiq, N.A.M. Zabidi, A.R. Mohamed, S. Bhatia, Catalytic conversion of palm oil over mesoporous aluminosilicate MCM-41 for the production of liquid hydrocarbon fuels, Fuel Processing Technology, 84 (2003) 105-120.

[102] Y.-S. Ooi, R. Zakaria, A.R. Mohamed, S. Bhatia, Catalytic Conversion of Fatty Acids

Mixture to Liquid Fuel and Chemicals over Composite Microporous/Mesoporous

Catalysts, Energy & Fuels, 19 (2005) 736-743.

[103] F.A.A. Twaiq, A.R. Mohamad, S. Bhatia, Performance of composite catalysts in palm oil cracking for the production of liquid fuels and chemicals, Fuel Processing

Technology, 85 (2004) 1283-1300.

[104] B. Urban, Y. Shirazi, B. Maddi, S. Viamajala, S. Varanasi, Flash Pyrolysis of

Oleaginous Biomass in a Fluidized-Bed Reactor, Energy & Fuels, 31 (2017) 8326-8334.

131

[105] Y. Shirazi, H. Tafazolian, S. Viamajala, S. Varanasi, Z. Song, M.J. Heben, High-

Yield Production of Fatty Nitriles by One-Step Vapor-Phase Thermocatalysis of

Triglycerides, ACS Omega, 2 (2017) 9013-9020.

[106] Y. Shirazi, S. Viamajala, S. Varanasi, High-yield production of fuel- and oleochemical-precursors from triacylglycerols in a novel continuous-flow pyrolysis reactor, Applied Energy, 179 (2016) 755-764.

[107] W.A. Sweeney, P.F. Bryan, BTX Processing, in: Kirk-Othmer Encyclopedia of

Chemical Technology, John Wiley & Sons, Inc., 2000.

[108] K. Giannakopoulou, M. Lukas, A. Vasiliev, C. Brunner, H. Schnitzer, Low pressure catalytic co-conversion of biogenic waste (rapeseed cake) and vegetable oil, Bioresource

Technology, 101 (2010) 3209-3219.

[109] S. Bhatia, Zeolite catalysts: principles and applications, CRC press, 1989.

[110] C.H. Zandonai, P.H. Yassue-Cordeiro, S.B. Castellã-Pergher, M.H.N.O. Scaliante,

N.R.C. Fernandes-Machado, Production of petroleum-like synthetic fuel by hydrocracking of crude soybean oil over ZSM5 zeolite–Improvement of catalyst lifetime by ion exchange,

Fuel, 172 (2016) 228-237.

[111] L. Raghunanan, S.S. Narine, Engineering Green Lubricants I: Optimizing Thermal and Flow Properties of Linear Diesters Derived from Vegetable Oils, ACS Sustainable

Chemistry & Engineering, 4 (2016) 686-692.

[112] G.A. Abel, S. Viamajala, S. Varanasi, K. Yamamoto, Toward Sustainable Synthesis of PA12 (Nylon-12) Precursor from Oleic Acid Using Ring-Closing Metathesis, ACS

Sustain. Chem. Eng., 4 (2016) 5703-5710.

132

[113] A. Mekki‐Berrada, S. Bennici, J.P. Gillet, J.L. Couturier, J.L. Dubois, A. Auroux,

Ammoniation–Dehydration of Fatty Acids into Nitriles: Heterogeneous or Homogeneous

Catalysis?, ChemSusChem, 6 (2013) 1478-1489.

[114] F.D. Gunstone, R.J. Hamilton, Oleochemical Manufacture and Applications,

Sheffield Academic Press, 2001.

[115] M. Kjellin, I. Johansson, Surfactants from renewable resources, John Wiley & Sons,

2010.

[116] Global Market Study on Fatty Amines, in, Persistence Market Research, 2016, pp.

152.

[117] P. Foley, E.S. Beach, J.B. Zimmerman, Derivation and synthesis of renewable surfactants, Chemical Society Reviews, 41 (2012) 1499-1518.

[118] J.O. Strüven, D. Meier, Hydrocracking of Organosolv Lignin in Subcritical Water to

Useful Phenols Employing Various Raney Nickel Catalysts, ACS Sustain. Chem. Eng., 4

(2016) 3712-3721.

[119] A.E. Rheineck, R.J. Stenberg, Method of preparing nitriles, in, Google Patents, 1961.

[120] R.H. Potts, Manufacture of nitriles, in, Google Patents, 1950.

[121] E. Hammerberg, G. Lichtenwalter, Production of nitriles, in, Google Patents, 1974.

[122] A. Mekki-Berrada, S. Bennici, J. Gillet, J. Couturier, J. Dubois, A. Auroux, Fatty acid methyl esters into nitriles: acid–base properties for enhanced catalysts, Journal of catalysis, 306 (2013) 30-37.

[123] J.-L. Dubois, A. Mekki-berrada, A. Auroux, S. Bennici, Gas-phase and liquid-gas- phase nitrilation process, in, Google Patents, 2014.

133

[124] N.B. Vargaftik, Handbook of physical properties of liquids and gases-pure substances and mixtures, (1975).

[125] C.G. Wortz, Preparation of long-chain aliphatic mononitriles, in, Google Patents,

1940.

[126] A.W. Ralston, Process of pyrolytically treating higher fatty acid substances, in,

Google Patents, 1935.

[127] N. Thirunavukarasu, R.C. Panda, Modeling, identification, and control for the production of glycerol by the hydrolysis of tallow, Rev. Chem. Eng., 31 (2015) 345-359.

[128] P.T. Vasudevan, M. Briggs, Biodiesel production—current state of the art and challenges, Journal of Industrial Microbiology & Biotechnology, 35 (2008) 421-430.

[129] P.L. Llewellyn, E. Bloch, S. Bourrelly, Surface area/porosity, adsorption, diffusion,

Characterization of Solid Materials and Heterogeneous Catalysts: From Structure to

Surface Reactivity, Volume 1&2, (2012) 853-879.

[130] M. Massaro, C.G. Colletti, G. Lazzara, S. Guernelli, R. Noto, S. Riela, Synthesis and

Characterization of Halloysite-Cyclodextrin Nanosponges for Enhanced Dyes Adsorption,

Acs Sustainable Chemistry & Engineering, 5 (2017) 3346-3352.

[131] G. Busca, The Surface Acidity and Basicity of Solid Oxides and Zeolites, in: Metal

Oxides, CRC Press, 2005, pp. 247-318.

[132] K.E. Hurst, M.J. Heben, J.L. Blackburn, T. Gennett, A.C. Dillon, P.A. Parilla, A dynamic calibration technique for temperature programmed desorption spectroscopy,

Review of Scientific Instruments, 84 (2013) 025103.

[133] J.L. Salager, Surfactants types and uses. Laboratory of formulation, interfaces, rheology and processes, in, Merida, Venezuela: Universidad De Los Andes, 2002.

134

[134] X. Yu, J. Yang, H. Lu, S.-T. Tu, J. Yan, Energy-efficient extraction of fuel from

Chlorella vulgaris by ionic liquid combined with CO2 capture, Applied Energy, 160 (2015)

648-655.

[135] V. Skorupskaite, V. Makareviciene, M. Gumbyte, Opportunities for simultaneous oil extraction and transesterification during biodiesel fuel production from microalgae: A review, Fuel Processing Technology, 150 (2016) 78-87.

[136] A.M. Rizzo, M. Prussi, L. Bettucci, I.M. Libelli, D. Chiaramonti, Characterization of microalga Chlorella as a fuel and its thermogravimetric behavior, Applied Energy, 102

(2013) 24-31.

[137] Z. Sebestyén, E. Barta-Rajnai, Z. Czégény, T. Bhaskar, B.B. Krishna, Z. May, J.

Bozi, Z. Barta, R. Singh, E. Jakab, Thermoanalytical Characterization and Catalytic

Conversion of Deoiled Micro Algae and Jatropha Seed Cake, Energy & Fuels, 30 (2016)

7982-7993.

[138] Y. Elkasabi, B.M.E. Chagas, C.A. Mullen, A.A. Boateng, Hydrocarbons from

Spirulina Pyrolysis Bio-oil Using One-Step Hydrotreating and Aqueous Extraction of

Heteroatom Compounds, Energy & Fuels, 30 (2016) 4925-4932.

[139] D. Chiaramonti, M. Prussi, M. Buffi, A.M. Rizzo, L. Pari, Review and experimental study on pyrolysis and hydrothermal liquefaction of microalgae for biofuel production,

Applied Energy, 185 (2017) 963-972.

[140] X. Gong, B.A. Zhang, Y. Zhang, Y.F. Huang, M.H. Xu, Investigation on Pyrolysis of Low Lipid Microalgae Chlorella vulgaris and Dunaliella salina, Energy & Fuels, 28

(2014) 95-103.

135

[141] X. Miao, Q. Wu, C. Yang, Fast pyrolysis of microalgae to produce renewable fuels,

Journal of Analytical and Applied Pyrolysis, 71 (2004) 855-863.

[142] Y. Shirazi, S. Viamajala, S. Varanasi, High-yield production of fuel-and oleochemical-precursors from triacylglycerols in a novel continuous-flow pyrolysis reactor, Applied Energy, 179 (2016) 755-764.

[143] B. Maddi, A. Vadlamani, S. Viamajala, S. Varanasi, Quantification of triglyceride content in oleaginous materials using thermo-gravimetry, Journal of Analytical and

Applied Pyrolysis, 128 (2017) 232-237.

[144] B. Maddi, S. Viamajala, S. Varanasi, Pyrolytic Fractionation: A Promising

Thermochemical Technique for Processing Oleaginous (Algal) Biomass, ACS Sustainable

Chemistry & Engineering, 6 (2018) 237-247.

[145] E. Furimsky, Hydroprocessing challenges in biofuels production, Catalysis Today,

217 (2013) 13-56.

[146] J.M. Jarvis, N.M. Sudasinghe, K.O. Albrecht, A.J. Schmidt, R.T. Hallen, D.B.

Anderson, J.M. Billing, T.M. Schaub, Impact of iron porphyrin complexes when hydroprocessing algal HTL biocrude, Fuel, 182 (2016) 411-418.

[147] P.M. Mortensen, J.D. Grunwaldt, P.A. Jensen, K.G. Knudsen, A.D. Jensen, A review of catalytic upgrading of bio-oil to engine fuels, Applied Catalysis A: General, 407 (2011)

1-19.

[148] S. Thangalazhy-Gopakumar, S. Adhikari, S.A. Chattanathan, R.B. Gupta, Catalytic pyrolysis of green algae for hydrocarbon production using H+ ZSM-5 catalyst, Bioresource

Technology, 118 (2012) 150-157.

136

[149] P.S. Rezaei, H. Shafaghat, W.M.A.W. Daud, Production of green aromatics and olefins by catalytic cracking of oxygenate compounds derived from biomass pyrolysis: A review, Applied Catalysis A: General, 469 (2014) 490-511.

[150] J. Jae, G.A. Tompsett, A.J. Foster, K.D. Hammond, S.M. Auerbach, R.F. Lobo, G.W.

Huber, Investigation into the shape selectivity of zeolite catalysts for biomass conversion,

Journal of Catalysis, 279 (2011) 257-268.

[151] A. Vadlamani, S. Viamajala, B. Pendyala, S. Varanasi, Cultivation of Microalgae at

Extreme Alkaline pH Conditions: A Novel Approach for Biofuel Production, ACS

Sustainable Chemistry & Engineering, 5 (2017) 7284-7294.

[152] K. Murugappan, C. Mukarakate, S. Budhi, M. Shetty, M.R. Nimlos, Y. Roman-

Leshkov, Supported molybdenum oxides as effective catalysts for the catalytic fast pyrolysis of lignocellulosic biomass, Green Chemistry, 18 (2016) 5548-5557.

[153] C. Mukarakate, J.D. McBrayer, T.J. Evans, S. Budhi, D.J. Robichaud, K. Iisa, J. ten

Dam, M.J. Watson, R.M. Baldwin, M.R. Nimlos, Catalytic fast pyrolysis of biomass: the reactions of water and aromatic intermediates produces phenols, Green Chemistry, 17

(2015) 4217-4227.

[154] C.A. Mullen, P.C. Tarves, A.A. Boateng, Role of Potassium Exchange in Catalytic

Pyrolysis of Biomass over ZSM-5: Formation of Alkyl Phenols and Furans, ACS

Sustainable Chemistry & Engineering, 5 (2017) 2154-2162.

[155] S. Wang, G. Dai, H. Yang, Z. Luo, Lignocellulosic biomass pyrolysis mechanism:

A state-of-the-art review, Progress in Energy and Combustion Science, 62 (2017) 33-86.

[156] K. Wang, K.H. Kim, R.C. Brown, Catalytic pyrolysis of individual components of lignocellulosic biomass, Green Chemistry, 16 (2014) 727-735.

137

[157] F.M. Dong, R.W. Hardy, N.F. Haard, F.T. Barrows, B.A. Rasco, W.T. Fairgrieve,

I.P. Forster, Chemical composition and protein digestibility of poultry by-product meals for salmonid diets, Aquaculture, 116 (1993) 149-158.

[158] A. Friedl, E. Padouvas, H. Rotter, K. Varmuza, Prediction of heating values of biomass fuel from elemental composition, Analytica Chimica Acta, 544 (2005) 191-198.

[159] F.A. Agblevor, S. Besler, Inorganic Compounds in Biomass Feedstocks. 1. Effect on the Quality of Fast Pyrolysis Oils, Energy & Fuels, 10 (1996) 293-298.

[160] T.N. Trinh, P.A. Jensen, K. Dam-Johansen, N.O. Knudsen, H.R. Sørensen, S.

Hvilsted, Comparison of Lignin, Macroalgae, Wood, and Straw Fast Pyrolysis, Energy &

Fuels, 27 (2013) 1399-1409.

[161] F. Verheijen, S. Jeffery, A. Bastos, M. Van der Velde, I. Diafas, Biochar application to soils, A critical scientific review of effects on soil properties, processes, and functions.

EUR, 24099 (2010) 162.

138

Appendix A: High-Yield Production of Fuel- and Oleochemical- Precursors from Triacylglycerols in a Novel Continuous-Flow Pyrolysis Reactor

Mass Balance and residence time calculations Mass balance boundary Mass balance boundary

푞̇푙 휌푙 푞̇푙 휌푙 푞̇푣1 Reactor 푞̇푣2 Input Atomizer Output 휌푣1 휌푣2

Figure A1: Mass balance calculations for residence time estimates

Residence time calculation based on feed properties (the red dashed box)

From equation of continuity 푞̇푙. 휌푙 = 푞̇푣1. 휌푣1 (A1)

푃 푀 From Ideal Gas Law 휌 = 푤 (A2) 푣1 푅푇

Further, 푚̇ = 푞̇푙. 휌푙 (A3)

퐿퐴 and, 푞̇ = (A4) 푣1 휏

퐿퐴 푃 푀 Therefore, 푚̇ = . 푤 (A5) 휏 푅푇

퐿퐴 푃 푀 Thus, 휏 = . 푤 (A6) 푚̇ 푅푇

139

Residence time calculation based on product properties (the green dashed box)

From equation of continuity 푞̇푙. 휌푙 ̇ = 푞푣1. 휌푣1 = 푞̇푣2. 휌푣2 (A7)

푃 (푀푊) From Ideal Gas Law 휌 = 푎푣푒 (A8) 푣2 푅푇

퐿퐴 And; 푚̇ = 푞̇푙. 휌푙 ; 푞̇푣2 = (A9) 휏0

퐿퐴 푃 (푀푊) Therefore, 푚̇ = . 푎푣푒 (A10) 휏 푅푇

퐿퐴 푃(푀푊) and, 휏 = . 푎푣푒 (A11) 0 푚̇ 푅푇

Thus,

휏 푀 = 푤 (A12) 휏0 (푀푊)푎푣푒

Nomenclature

3 -1 푞̇푙 = Volumetric flow rate of liquid (m s ) -3 휌푙 = Density of liquid (g m ) 3 -1 푞̇푣1 = Volumetric flow rate of triglyceride vapor (m s ) -3 휌푣1 = Density of triglyceride vapor (g m ) 3 -1 푞̇푣2 = Volumetric flow rate of products vapor (m s ) -3 휌푣2 = Density of products vapor (g m ) 푃 = Reactor pressure (atm) -1 푀푤= Molecular weight of triglyceride (g gmol ) -1 (푀푊)푎푣푒 = Average molecular weight of products (g gmol ) 푅 = Universal gas constant (m3 atm mol-1 K-1) 푇 = Reactor temperature (K) 푚̇ = Feed mass flow rate (g/s) 퐿 = Reactor length (m) 퐴 = Reactor cross-sectional area (m2) τ = Vapor residence time (s) based on input boundary condition τ0 = Vapor residence time (s) based on output boundary condition

140

Table A1: Residence time estimates at reaction temperatures based on input (τ) and output (τ0) boundary condition.

τ T 450 475 500 (MW)ave τ0 (MW)ave τ0 (MW)ave τ0 1 254.4 0.3 209.2 0.2 173.3 0.2 6 148.5 1.0 128.1 0.9 117.9 0.8 60 127.6 8.7 119.1 8.1 115.7 7.9 300 111.8 38.2 94.4 32.2 87.6 29.9 Note: Average molecular weight ((MW)ave) is calculated using Equation 27 (below). The mole fraction of each compound in liquid products was estimated by GC-FID. For unidentified compounds, molecular weight of hexane (86 g gmol-1) was used as approximation since a majority of unidentified compounds were low molecular weight hydrocarbons. The average molecular weight of non-condensable gases was assumed to be 44 g gmol-1.

Average molecular weight calculations:

For a reaction that produces 푛 chemically-distinct products, we can first calculate the total number of moles of products (condensed and uncondensed) generated after pyrolysis as follows:

푁푇 = 푁1 + 푁2 + ⋯ + 푁푛 (A13) where, 푁1, 푁2 … 푁푛 are number of moles of compounds 1, 2 … 푛 and 푁푇 is the total moles of products.

The moles of each chemical product can be further estimated from the mass fraction of the chemical compound relative to the total mass of products. Thus, for each chemical, the number of moles would be

푀1 푁1 = ; 푀1 = 푋1 × 푀푇 (A14) 푀푊,1

푀2 푁2 = ; 푀2 = 푋2 × 푀푇 (A15) 푀푊,2

141

⋮

푀푛 푁푛 = ; 푀푛 = 푋푛 × 푀푇 (A16) 푀푊,푛 where, 푀1, 푀2 … 푀푛 are mass of compounds 1, 2 … n; 푀푊,1, 푀푊,2 … 푀푊,푛 are molecular weights of compounds 1, 2 … n; 푋1, 푋2 … 푋푛 are mass fractions of compounds 1, 2 … n;

푀푇 is the total mass of vapor products. This would also be equal to the mass fed to the reactor since char formation was negligible in our experiments;

Thus,

푋1 푋2 푋푛 푁푇 = 푀푇 ( + + ⋯ + ) (A17) 푀푊,1 푀푊,2 푀푊,푛

Based on the total number of moles, the mole fraction of each compound in the product can be estimated as:

푁1 푥1 = (A18) 푁푇

푁2 푥2 = (A19) 푁푇

⋮

푁푛 푥푛 = (A20) 푁푇

Thus,

푥푇 = 푥1 + 푥2 + ⋯ + 푥푛 = 1 (A21)

Average molecular weight

(푀푊)푎푣푒 = 푥1. 푀푊1 + 푥2. 푀푊2 + ⋯ + 푥푛. 푀푊푛 (A22)

142

푁1 푁2 푁푛 (푀푊)푎푣푒 = . 푀푊1 + . 푀푊2 + ⋯ + . 푀푊푛 (A23) 푁푇 푁푇 푁푇

푀1 푀푊1 푀2 푀푊2 푀푛 푀푊푛 (푀푊)푎푣푒 = . + . + ⋯ + . (A24) 푀푊1 푁푇 푀푊2 푁푇 푀푊푛 푁푇

푀1 푀2 푀푛 푀1+ 푀2+ ⋯+ 푀푛 푀푇 (푀푊)푎푣푒 = + + ⋯ + = = (A25) 푁푇 푁푇 푁푇 푁푇 푁푇

푀푇 (푀푊)푎푣푒 = 푋1 푋2 푋푛 (A26) 푀푇 ( + + ⋯+ ) 푀푊1 푀푊2 푀푊푛

1 (푀푊)푎푣푒 = 푋 푋 푋 (A27) ( 1 + 2 + ⋯+ 푛 ) 푀푊1 푀푊2 푀푊푛

Reaction mechanism for trioleate pyrolysis:

O O R 1 1 O 3 OH 3 4 4 O OH 6 5 7 Diels-Alder 6 7 5 7

Figure A2: Triglyceride pyrolysis reaction pathways illustrated for thermal cracking of trioleate. Reaction numbering is according to Figure7 3. Red arrows show C-C cleavage. Green arrows show decarboxylation/decarbonylation. Blue arrows show Diels-Alder reactions. Dashed red lines indicate the C-C cleavage locations. The reactions shown here are based on the mechanisms discussed by Maher et al. [34].

143

Table A2: Composition (wt. %) of L-FAs present in the recovered liquid (refer to Figure 2.4(a)). Compound Experimental condition 450-60 450-300 475-60 475-300 500-60 500-300 Palmitic acid 4.0 4.3 4.5 5.8 5.5 5.2 Linoleic acid 7.3 6.2 5.1 3.5 5.3 4.9 Oleic acid 22.3 20.3 21.5 17.7 18.1 15.5 Stearic acid 4.8 4.8 4.1 4.1 4.5 3.9 Total 38.4 35.7 35.2 31.2 33.2 29.5

Table A3: Composition (wt. %) of S-FAs present in the recovered liquid (refer to Figure 2.4(b)). Compound Experimental condition 450-60 450-300 475-60 475-300 500-60 500-300 Hexanoic acid 0.7 1.0 1.4 1.9 1.1 1.1 Heptanoic acid 1.9 2.9 3.4 3.2 3.7 4.7 Octanoic acid 1.0 1.8 1.5 2.2 1.7 1.8 Nonanoic acid 1.1 1.5 1.4 2.0 1.3 1.3 Decenoic acid 1.6 1.9 2.0 1.7 2.7 2.2 Decanoic acid 1.4 2.4 2.1 2.7 2.6 2.9 Undecenoic acid 0.6 0.8 0.9 0.7 1.1 1.1 Decanedioic acid 1.0 0.5 0.5 0.5 0.8 0.8 Total 9.2 12.7 13.2 14.8 15.0 16.0

144

Table A4: Major classes of chemical compounds in recovered liquid.

Experiment Composition of recovered liquid (wt. % of feed) Gas al condition Unreacted L-FAs S-FAs Lt-HCs Hy-HCs Unidentified feed (C18-C16) (C6-C12) (C5-C12) (>C12) liquid products 450-1 55.9 26.3 2.2 2.9 3.6 2.3 6.8 450-6 17.0 40.1 4.1 5.5 5.0 20.4 8.0 450-60 5.7 34.6 8.3 11.3 9.0 21.3 9.7 450-300 0.0 31.0 11.1 16.3 6.8 21.7 13.1 475-1 45.0 28.6 4.0 4.9 4.9 4.4 8.2 475-6 7.2 36.9 6.3 8.0 7.6 24.2 9.8 475-60 0.0 30.9 11.6 15.6 13.3 16.3 12.3 475-300 0.0 23.6 11.2 19.3 8.1 13.4 24.4 500-1 36.8 29.4 5.8 5.5 6.1 5.0 11.5 500-6 2.8 36.3 7.3 9.6 8.7 22.6 12.8 500-60 0.0 28.3 12.8 19.0 13.8 11.3 14.7 500-300 0.0 20.6 11.2 19.8 8.9 9.5 30.1 550-300 0.0 5.2 6.2 23.1 4.3 10.5 50.8 Notes: (1) The “unreacted feed” in Table A4 indicates glyceride mass in the liquid products (relative to the feed). For instance, at experimental condition of 500-6, the unreacted feed is 2.8% wt., that means 2.8% wt. of feedstock did not convert into non- glyceride products. In other words, the conversion of triacylglycerol to non-glyceride compounds is 97.2%. The unreacted feed (glyceride) mass were quantified using GC- FID. (2) The “unidentified liquid products” are the components that we could not positively identify by GC-MS due to low confidence number, possibly due to low concentration or low sensitivity on GC-MS. In Table A4, the liquid products were quantified by GC-FID and the non-condensable gases were calculated by subtracting liquid products mass from feed mass.

145

uV(x10,000) 5.0 uV(x100,000) 1.25

1.00

0.75

4.0)

6 0.50

10 0.25

(

3.0 0.00 0.0 5.0 10.0 15.0 20.0 min

2.0 Intensity

1.0

0.0

0.0 25.0 50.0 75.0 100.0 125.0 150.0 175.0 min Retention Figure A3: GC-FID analysis of distilled fraction. time (min)

Table A5: Major chemical composition of distilled fraction (DF). Peak Relative GC-FID No. RT (min) Compound (%) 1 3.11 1-pentene 0.26 2 3.81 pentane 2.37 3 4.13 2-pentene 1.10 4 4.71 cyclopentene 1.02 5 5.44 methylpentene 1.40 6 6.59 hexene 4.62 7 6.83 hexane 2.14 8 9.68 benzene 0.72 cyclopentane 9 11.53 methylene 2.38 10 12.92 1-heptene 4.18 11 14.11 heptane 2.45 12 15.60 2-heptene 0.44 13 20.61 toluene 1.43 14 24.34 octadiene 0.97 15 26.15 1-octene 2.13 16 27.98 octane 2.67 17 29.81 2-octene 0.74 18 30.85 octadiene 0.90 19 35.13 ethylbenzene 1.22 20 36.65 m-xylene 0.78 21 36.97 p-xylene 0.47 22 40.61 o-xylene 0.59

146

23 42.16 cyclooctene 4.21 24 42.80 1-nonene 1.73 25 44.20 3-nonene 0.26 26 44.42 3-nonene 0.16 27 45.09 nonane 1.87 28 46.98 isopropylbenzene 0.09 29 53.33 n-propylbenzene 0.88 1-methyl-2- 30 55.46 ethylbenzene 0.33 1-methyl-4- 31 56.26 ethylbenzene 0.18 32 58.35 trimethylbenzene 0.17 1-methyl-2- 33 60.15 ethylbenzene 0.13 34 64.14 tert-buthylbenzene 1.50 35 67.32 decene 2.11 36 69.21 decane 1.60 methyl- 37 70.34 Isopropylbenzene 0.43 38 75.50 n-butylbenzene 1.35 39 82.18 undecene 1.68 40 83.28 decane 1.15 41 84.64 2-undecene 0.71 42 85.34 undecadiene 1.73 43 87.35 benzene pentyl 2.19 44 89.63 buthylcycloheptadiene 1.29 45 91.75 dodecene 2.01 46 93.22 dodecane 1.63 47 94.42 dodecadiene 1.06 48 96.92 benzenehexyl 0.48 49 99.92 tridecene 1.81 50 101.25 tridecane 1.24 51 104.69 tridecadiene 0.96 52 107.62 tetradecene 1.76 53 108.36 tetradecane 1.15 54 112.06 tetradecadiene 2.92 55 114.35 pentadecene 2.99 56 114.61 pentadecane 3.44 57 120.35 hexadecene 0.26 58 125.22 heptadecene 1.70 59 126.68 heptadecane 1.20 Total 85.4

147

5.0uV(x10,000)

4.0

)

6 3.0

(

2.0

Intensity Intensity 1.0

0.0

0.0 25.0 50.0 75.0 100.0 125.0 150.0 175.0 min Retention time (min)

Figure A4: GC-FID chromatogram of residue fraction.

Table A6: Chemical composition of residue fraction. Compounds Relative GC-FID (%) C13-HCs 2.4 C15-HCs 5.6 C16-HCs 0.7 C17-HCs 3.1 C16-FAs 9.2 C18-FAs 59.2 S-FAs 8.8 Unidentified 11 Total 100

148

10.0 uV(x10,000)

(a) (

) 7.5

6 a)

(

5.0

Intensity Intensity 2.5

0.0

0.0 25.0 50.0 75.0 100.0 125.0 150.0 175.0 min

Retention

time (min)

) 6

(

Intensity Intensity

Retention

Figure A5: GC-FID analysis of derivatizedtime (min) (a) DF and (b) RF.

Table A7: Composition of S-FAs in DF and RF. RT (min) S-FAs (wt. %) DFa (wt. %) RFb (wt. %) 46 C6 0.8 0.4 70 C7 3.0 1.2 84 C8 1.3 0.7 93.5 C9 0.9 0.5 101.5 C10 0.7 5.4 108 C11 - 0.6 Total 6.7 8.8 a: Values are relative to DF. b: Values are relative to RF.

149

)

( ( (10 Abundance c) (( c) (

bb)) a) 10 15 20 25 30 35

Retention time (min) Figure A6: GC-MS analysis of oleic acid pyrolysis: (a) feed, and pyrolysis products at 450 ºC and τ of (b) 60s, and (C) 300s.

Table A8: Chemical composition of oleic acid pyrolysis (feed and products). Peak No. Compound Feed 450-60 450-300 1 Decanoic acid 0.0 1.1 4.3 2 Heptadecene 0.0 2.9 5.7 3 Tetradecanoic acid 2.7 2.5 2.3 4 Hexadecenoic acid 3.3 3.1 2.9 5 Hexadecanoic acid 4.4 4.2 4.3 6 Heptadecanoic acid 1.9 1.5 1.3 7 Oleic acid 85.5 72.4 63.3 Total 97.7 87.7 84.1

150

Appendix B: High Yield Production of Hydrocarbons from Non-edible Oils Through Novel Reactive Pyrolysis System

Table B1. Texture properties and acidity of HZSM-5 catalysts.

Properties Texture† Acidity‡ st $ nd * Surface Pore Pore Total 1 acidity 2 acidity Tmax.1 Tmax.2 Catalyst area¥ size# volume@ acidity (µmol g-1) (µmol g-1) (K) (K) (m2 g-1) (nm) (cm3 g-1) (µmol g-1)%

HZSM-5 337.7 0.5 0.204 459.1 171.9 287.2 397.1 503

† Texture properties were measured by ASAP2020 instrument. ‡ Acidity was measured by NH3-TPD method. ¥ BET surface area; # BJH Adsorption average pore diameter; @ Cumulative pore volume % The values for acidity are in µmole of NH3 desorption per gram of catalyst. $ Measured from area of first peak; corresponds to medium/weak acid sites. * Measured from area of second peak; corresponds to strong acid sites. Tmax.1: Temperature where maximum NH3 desorption was observed in first peak from the NH3 desorption versus temperature plot. Tmax.2: Temperature where maximum NH3 desorption was observed in second peak from the NH3 desorption versus temperature plot.

151

Figure B1. A laboratory scale distillation apparatus employed to fractionate the liquid products.

Fractionation of pyrolysis liquid products

The liquid products were transferred to a round-bottom boiling flask that was placed in a silicone oil bath. The bath was kept on a hot stir plate and heated to a desired temperature. Temperature of liquid sample inside the boiling flask was monitored using a thermocouple. During the distillation experiments, components start to boil and evaporate, then pass through the distillation column. To reduce the heat losses, distillation column

(from boiling flask up to condenser located at the top of column) was insulated using glass wool wrapped with aluminum foil. A thermocouple also was placed at the column head to monitor the vapors temperature leaving the column (right before the condenser). Tap water

152 served as coolant fluid for the condenser and the condensate were collected in a 100 mL beaker. Before conducting distillation of the liquid products from catalytic pyrolysis, preliminary distillation experiments were performed using a mixture of benzene, toluene, xylene, ethylbenzene and naphthalene. These chemicals are the main compounds observed in the liquid products from WCO catalytic pyrolysis. The preliminary distillation experiments aid to understand temperature profile at oil bath, flask and distillation head.

The liquid products collected form 12 cycles of catalytic pyrolysis was first distilled to separate benzene and toluene from other fractions in the liquid products. 67 g of liquid products transferred into the boiling flask and the oil bath heated up to 190 °C. Under these conditions, the temperature at the column head gradually increased up to 115 °C. The distillation experiment conducted for about 8h until no vapors were observed at the column head. The collected distillates (fraction 1) were weighed and stored for analytical analysis.

Thereafter, distillation of the residue from the first distillation experiment was carried out to separate xylene (and/or chemicals that have b.p. close to xylene) from other chemicals in the catalytic pyrolysis liquid products. However, to maintain a temperature at the column head near to the b.p. of xylene (145 °C), the oil bath estimated to be heated to temperature above 230 °C. To avoid that, a vacuum pump employed and the distillation experiment were performed under 20 mmHg vacuum pressure. The vacuum pump was also equipped with a pressure controller to maintain this pressure for the entire experiment. Under this condition, b.p. of xylene estimated to be ~50 °C. Thus, a column head temperature of 60

°C was selected to make sure most of the xylene would be separated from the other fraction.

The oil bath heated up to 125 °C and the column head temperature increased gradually up

153

to 60 °C. The distillation performed for more than 10 h. Afterwards, the distillates and the

residue weighed and stored for analysis.

Table B2. Overall mass balance for conversion of triglyceride at examined operating conditions.

*: Organic Liquid Products. Run No. Experimental conditions Products yields (wt. %) Conversion (%) Temperature HZSM-5 WHSV OLP* Aq. Coke Non-condensable (°C) (g) (h-1) phase gases 1 450 0 - 93.5 0.0 0.0 6.5 92.2 2 450 1 60 86.6 1.3 0.1 12.0 100 3 450 2 30 82.5 2.1 0.2 15.2 100 4 450 3 20 79.8 2.5 0.3 17.3 100 5 450 5 12 75.9 3.0 0.5 20.6 100 6 475 0 - 91.8 0.0 0.0 8.2 96.6 7 475 1 60 84.5 1.6 0.1 13.8 100 8 475 2 30 80.7 2.3 0.2 16.8 100 9 475 3 20 77.1 2.7 0.4 19.7 100 10 475 5 12 73.7 3.3 0.5 22.5 100 11 500 0 - 90.8 0.0 0.0 9.2 98.3 12 500 1 60 83.3 1.8 0.1 14.9 100 13 500 2 30 78.7 2.6 0.3 18.4 100 14 500 3 20 74.7 3.2 0.4 21.7 100 15 500 5 12 70.7 3.8 0.6 24.9 100

154

100 2

80 1.6

60 In presence of HZSM-5 1.2 In absence of HZSM-5 40 0.8

Residual weight (Wt.%) weight Residual 20 0.4 Derivative weight loss(%/ºC) weight Derivative

0 0 0 100 200 300 400 500 600 Temperature (ºC)

Figure B2. Residual weight (solid line) and its temperature derivative (dash line) of triglyceride degradation in the presence (blue color) and absence (red color) of HZSM-5.

155

Table B3. Composition of organic liquid products (OLP) at the examined experimental conditions (coded as X-Y, where X is the temperature in °C and Y is the HZSM-5 loading in g).

Compound Experimental condition

450-0 450-1 450-2 450-3 450-5 475-0 475-1 475-2 475-3 475-5 500-0 500-1 500-2 500-3 500-5

Aromatic Benzene 0.9 1.8 3.3 5.6 7.3 1.3 2.8 4.7 7.2 9.3 2.1 4.4 6.8 9.8 11.6 Toluene 0.1 3.6 7.8 11.5 15.8 0.2 5.3 9.7 13.9 19.3 0.5 8.2 12.9 17.3 20.0 Ethylbenzene 0.2 0.6 1.5 1.7 1.8 0.1 0.8 1.3 1.9 1.9 0.3 1.1 1.3 1.4 2.3 Xylene 0.5 2.6 6.3 7.5 9.8 0.8 3.7 6.3 9.1 11.0 1.1 5.3 7.8 9.2 11.5 C8+ Aromatics 0.6 1.1 1.4 2.4 2.5 1.1 1.1 1.8 2.5 2.3 1.9 1.0 1.8 1.5 2.8 Naphthalenes 1.0 2.7 1.9 4.5 1.5 3.6 3.9 4.3 2.8 4.8 4.3 5.2 BTEX 1.7 8.8 18.9 26.3 34.7 2.4 12.6 22.0 32.2 41.5 4.0 19.0 28.8 37.7 45.4 Total aromatic 2.3 10.9 23.0 30.7 41.7 3.5 15.2 27.4 38.6 48.1 5.9 22.8 35.4 43.5 53.5 Aliphatic Olefin C5-C13 4.7 15.9 12.2 10.6 7.4 7.7 13.6 11.1 9.3 6.1 9.5 12.8 10.2 8.6 5.1 Paraffin C5-C13 6.9 9.6 7.6 4.3 2.6 8.4 8.3 4.6 3.3 1.5 10.7 7.1 3.7 2.9 0.9 Aliphatic >C13 8.7 12.5 9.4 5.9 3.9 10.3 10.1 7.8 5.1 2.5 11.2 8.4 6.9 2.2 1.8 Total aliphatic 20.3 37.9 29.3 20.8 13.8 26.4 32.0 23.5 17.7 10.1 31.4 28.3 20.8 13.7 7.8 Fatty acids C16-FAs 5.4 5.1 5.0 5.0 3.7 4.1 4.5 3.9 3.4 3.5 3.1 3.4 2.9 3.5 2.5 C18-FAs 38.0 32.3 27.5 21.3 18.2 34.3 29.1 25.1 19.5 16.3 32.4 25.8 22.2 17.1 15.1 S-FAs 7.0 0.0 0.0 0.0 0.0 8.3 0.0 0.0 0.0 0.0 10.1 0.0 0.0 0.0 0.0 Total fatty acid 50.4 37.4 32.5 26.3 21.8 46.8 33.6 29.0 22.9 19.8 45.6 29.2 25.1 20.6 17.6

Glycerides 8.3 0.0 0.0 0.0 0.0 3.7 0.0 0.0 0.0 0.0 1.9 0.0 0.0 0.0 0.0

Unidentified 18.7 13.8 15.2 22.2 22.7 19.6 19.3 20.1 20.8 22.0 15.2 19.7 18.7 22.2 21.2 Total 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100

156

Table B4. Composition of the OLP from conversion of triglyceride at 500 °C and WHSV of 3-12 h-1. Compound WHSV (h-1)

12 6 3 Aromatic Benzene 11.6 11.8 13.3 Toluene 20.0 24.6 27.8 Ethylbenzene 2.3 2.2 2.6 Xylene 11.5 13.9 16.2 C8+ Aromatics 2.8 3.5 3.2 Naphthalenes 5.2 7.9 10.4 BTEX 45.4 52.5 59.9 Total Aromatics 53.5 63.9 73.5

Aliphatic

Olefin C5-C13 5.1 4.1 3.6 Paraffin C5-C13 0.9 0.5 0.4 Aliphatic >C13 1.8 2.1 1.2 Total aliphatic 7.8 6.7 5.2 Fatty acids C16 2.5 2.3 1.5 C18 15.1 7.5 3.1 Total fatty acids 17.6 9.8 4.6

Unidentified 21.2 19.6 16.7 Total 100 100 100

157

Table B5. Products yields and composition from catalytic pyrolysis of triglyceride at 500 °C, WHSV of 3 h-1 and reaction time on stream of 1-5 h.

Properties Reaction time on stream (h) Products yield (wt.% of feed) 1 2 3 5 OLP 59.0 64.9 67.0 72.0 Aqueous phase 6.3 5.8 4.8 4.0 Coke 0.9 1.1 1.6 2.0 Non-condensable gas 33.8 28.3 26.7 22.0

OLP composition (wt.% of OLP) Aromatic Benzene 19.1 13.3 7.6 5.8 Toluene 30.2 27.8 14.2 9.2 Ethylbenzene 2.7 2.6 2.3 1.7 Xylene 17.6 16.2 9.9 6.0 C8+ Aromatics 2.8 3.2 4.7 2.0 Naphthalenes 15.9 10.4 7.2 3.9 Total 88.3 73.5 45.9 28.6

Aliphatic 14.8 27.6 17.2 Olefins C5-C13 2.1 3.6 6.6 8.3 Paraffin C5-C13 0.4 0.4 3.1 5.4 Aliphatic >C13 0.0 1.2 5.8 12.9 Total 2.5 5.2 15.5 26.6

Fatty acids C16 - C18 ND 4.6 15.8 21.5 C6 - C12 ND ND 4.5 7.3 Unidentified 9.2 16.7 18.3 16.0 Total 100.0 100.0 100.0 100.0

158

Table B6. Fatty acid compositions of the tested non-edible oil. Fatty acid composition Pennycress Camelina WCO 16:0* 6.8 6.4 27.4 16:1 -% 2.6 - 18:0 1.9 4.0 19.9 18:1 3.9 7.4 7.5 18:2&3 10.0 46.6 26.0 20:0 3.5 5.8 10.7 20:1 11.5 14.0 - 20:2 & 3 - 3.2 - 22:0 3.5 1.5 - 22:1 25.1 5.5 - 24:0 5.9 - - 24:1 22.9 - - Unidentified 5.2 3.0 8.5 Polyunsaturated 10.0 49.8 26.0 Monounsaturated 63.3 29.5 7.5 Saturated 21.6 17.7 58.0 Free fatty acids$ - - 20.7 * The first and second number indicates the length of the fatty acid chain and number of double bonds, respectively. $ Estimated from GC-FID analysis of WCO. % Not detected.

Reaction Reg Reaction Reg 700

O2 O2

600

C) ° 500 ……

400 . 300 N2

Reactor temperature ( temperature Reactor 200

100

0 0 50 100 150 200 250 Time (min)

Figure B3. Reaction-regeneration procedure for WCO catalytic pyrolysis.

159

uV (x1,000,000) Chromatogram 1.50 1 Peak No. Compound (a) 1.25 1 Benzene 1.00 2 2 Toluene

0.75 3 Ethyl benzene 4 Xylene 0.50 5 C8+ alkylbenzene 0.25 6 Naphthalene 0.00

25.0 50.0 75.0 100.0 125.0 150.0 175.0 min uV (x100,000) Chromatogram 7.5 (b) 4 5.0

2.5 3 5

Intensity Intensity Intensity

0.0

25.0 50.0 75.0 100.0 125.0 150.0 175.0 min uV (x100,000) 2.0 Chromatogram 6 (c) 1.5

1.0

0.5 5

0.0

25.0 50.0 75.0 100.0 125.0 150.0 175.0 min Retention time (min)

Figure B4. GC-FID chromatogram of the (a) first and (b) second distillates and (c) residue fraction from distillation of liquid products from WCO catalytic pyrolysis.

160

Appendix C: High-Yield Production of Fatty Nitriles by One-Step Vapor Phase Thermo-Catalysis of Triglycerides

Main chamber valve

Vacuum chamber valve

Figure C1: Schematic diagram of the temperature programmed desorption (TPD) system.

5% C8:0 5% 12% C10:0

12% C12:0 C14:0 C16:0 20% 46% C18:0; C18:1

161

Figure C2: Fatty acid composition (wt.%) of coconut oil.

Figure C3: NH3-TPD analysis of tested catalysts.

O O O

CH2O-C R1 HO-C R1 H2N-C R1 N≡C R1 O O O pyrolysis N≡C R2 + 6H O CH O-C R2 HO-C R2 + 3NH3 H2N-C R2 2 O O O N≡C R3 CH2O-C R3 HO-C R3 H2N-C R3

Triglyceride Fatty acid Fatty amide Fatty nitrile

162

Figure C4: Generalized reaction steps of nitrile production from triglyceride.

Fatty nitriles

uV(x100,000) uV(x100,000) 8.0Chromatogram 8.0Chromatogram Fatty acids 7.0 7.0

6.0 6.0 Fatty amides

5.0 5.0

) 5 4.0 4.0 (c)

3.0 3.0 (b) 2.0 2.0

Intensity Intensity (10 Triglyceride 1.0 1.0 (a) 0.0 0.0

0.0 5.0 0.0 10.0 5.0 15.0 10.0 20.0 15.0 25.0 20.0 30.0 25.0 min30.0 min Retention time (min) Figure C5: GC-FID analysis of liquid products from one-step vapor phase nitrile reaction in (a) absence of catalyst and (b) presence of Al2O3 and (c) V2O5

Table C1: Chemical composition of one-step vapor phase nitrile reaction over tested catalysts. Values are wt. % relative to total product mass.

163

ND: Not detected and/or less than 0.05% of products.

164

uVChromatogram(x1,000,000)

1.50

1.25 N

N: Fatty nitrile F: Fatty acid 1.00 N

N N

) 0.75 6 N 7.5

0.50

Intensity (10 Intensity 6

0.25 4.5

3 0.00 F F F F F F 0

0.0 5.0 10.0 15.0 20.0 25.0 30.0 min Retention time (min)

Figure C6: GC-FID analysis of liquid products from vapor phase nitrile reaction at different NH3/triglyceride molar ratio (0-7.5).

165

NH3 Acid -H2O (a) Conventional nitrile production Triglyceride Fatty acid Fatty amide Fatty nitrile hydrolysis ZnO

Hydrolysis Fatty nitriles production Water Q1 Q2 NH3 >100°C Triglyceride 20 °C 20°C Fatty acids & water Glycerol Fatty nitriles 1bar (1:5) 1bar 250°C, 45 bar & water Fatty 300°C, ZnO water acid Hydrolysis reactor Phase separator Distillation Nitrile reactor Phase separator

Q1 = energy input for hydrolysis per kg triglyceride Q2 = energy input for nitrile reaction = mtriglyceride Cp, triglyceride ∆T + mwater (Hwater, 250 °C – Hwater, 20 °C) = mfatty acid Cp, fatty acid ∆T = (1 kg × 1.9 kJ kg-1 °C-1 ×225 °C) + 5 kg (1128.8 kJ kg-1 – 83.9 kJ kg-1) = (0.9 kg × 2.15 kJ kg-1 °C-1 ×200 = 5652 kJ = 387 kJ

NH ; catalyst (b) One step vapor phase nitrile production Triglyceride 3 Fatty nitrile

Q' Water

NH3 20 °C 400 °C 100°C Triglyceride 1bar Vapor-phase reactor Fatty nitriles Distillation

Q' = energy input for one-step vapor phase nitrile reaction

= mtriglyceride Cp, triglyceride ∆T + mtriglyceride ∆ Hvap = (1 kg × 1.9 kJ kg-1 °C-1 ×380 °C) + (1 kg × 203.9 kJ kg-1) = 925.9 kJ

Figure C7: Comparison between conventional (a) and one step vapor phase nitrile production (b) from triglyceride.

Detailed description for energy calculations and assumptions

Figure C7 compares the conventional nitrile process with one-step vapor phase nitrile reaction. Figure C7 shows the simplified reaction steps, schematic of process flow diagram and energy consumption calculations. In the conventional nitrile process (Figure

166

C7a) the first step is hydrolysis of triglyceride to produce fatty acids. A typical triglyceride/water ratio of 1/5 was used to calculate the energy required for hydrolysis step.

From previous studies, reaction temperature range of 250-350 °C and pressure range of 45-

60 bar were reported and triglyceride conversion from 40 to 99% was obtained depending on reaction temperature, pressure and time [127, 128]. Although, higher reaction temperature, pressure and time improve the triglyceride hydrolysis, we assumed reaction temperature of 250 °C, pressure of 45 bar and 100% triglyceride conversion to calculate the energy consumed in hydrolysis of 1 kg triglyceride (Figure C7a). Considering the complete hydrolysis of triglyceride, 0.9 kg fatty acids would be produced from 1 kg triglyceride. The energy calculated under these reaction conditions would be the minimum energy required for triglyceride hydrolysis because to achieve nearly complete triglyceride hydrolysis, higher temperatures and pressures are usually required than those assumed in our calculation. Moreover, the produced fatty acid from hydrolysis of triglyceride contains some water and requires distillation/purification to remove the soluble water in fatty acids

[127], however, the energy required by those extra purification steps are not considered in our calculations. As observed in Figure C7a, hydrolysis of 1 kg triglyceride requires a minimum energy of 5652 kJ. Conversion of fatty acid into fatty nitrile through conventional nitrile process was assumed to occur in a batch reactor and it requires a minimum energy of 387 kJ kg-1 triglyceride. It should be noted that the enthalpies of reaction of conventional (hydrolysis and nitrile reactions) and one-step vapor phase nitrile processes are assumed to be similar. Figure C7b gives the detailed calculations of the energy required in the one-step vapor phase nitrile process. Boiling point of the

167 triglycerides was assumed to be 400 °C. The net energy for was estimated to be 9259 kJ kg-1 triglyceride.

168

Appendix D: Thermochemical Conversion of Algae to Biofuels and Chemicals: A Study on Integration of Algae Fractionation and Ex-situ Catalytic Pyrolysis

169

(a) ( ( a) ( b) c)

(b)

Intensity Intensity

(c)

Retention time (min) Figure D1. GC-MS chromatogram of bio-oil from pyrolysis in (a) absence of catalyst and in presence of (b) in-situ and (c) ex-situ catalyst.

170

20 (a)

450 °C 10 500 °C 550 °C

5 Benzene selectivity Benzene

0 In-situ catalyst Ex-situ catalyst

40 (b) 30 450 °C 20 500 °C

10 550 °C Toluene selectivity Toluene

0 In-situ catalyst Ex-situ catalyst

25 (c)

15 450 °C 500 °C 10 550 °C

Xylene Xylene selectivity 5

0 In-situ catalyst Ex-situ catalyst

Figure D2. Selectivity of (a) benzene, (b) toluene and (c) xylene from in-situ and ex-situ catalytic pyrolysis of microalgae at tested temperatures.

171

(a) Region III: Lipid Region IV: Region I: Region II: Large protein volatilization Dehydration Carbohydrate/protein molecules

volatilization volatilization 100 0.8

80 0.6

60 C) ° 0.4

40 (wt.%/

Residual weight % weight Residual loss

0.2 weight Derivative 20

0 0 0 100 200 300 400 500 600 Temperature (°C)

(b) 100

40 Residual weight % weight Residual 20

0 0 100 200 300 400 500 Temperature (°C) 500

(c)

400

300

200 Temperature ( Temperature 100

0 0 50 100 150 200 250 300 Time (min)

Figure D3. Thermogravimetric analysis of microalgae (a) pyrolysis of microalgae (b) fractional pyrolysis at 320 °C and 450 °C (c) temperature versus time profile in fractional pyrolysis.

172

Notes: (1) Figure D3a shows the TGA analysis of microalgae, where it was heated from -1 room temperature to 600 °C at temperature ramp rate of 10 °C min and under N2 atmosphere. (2) Figure D3b demonstrates a simulated fractional pyrolysis on TGA -1 instrument. For this analysis, microalgae heated under N2 to 320 °C at 10 °C min temperature ramp and then held at 320 °C for 10 min. Then, the residue from first fraction cooled down to room temperature and then heated again to 450 °C at 10 °C min-1 and held for 10 min. (3) Figure D3b shows the temperature vs time profile during the simulated fractional pyrolysis experiments on the TGA.

Fractional pyrolysis simulation on TGA

Figure D3a shows the TGA analysis of microalgae, where it was heated from room

-1 temperature to 600 °C at 10 °C min ramp and under N2 atmosphere. Clearly, four regions of derivative weight loss can be observed from Figure D3a. Region I shows slight weight loss due to the dehydration of the water that likely trapped in microalgae cells and did not removed by freeze drying. Region II shows a weight loss from 160 to 340 °C contributed to decomposition of carbohydrate and protein. The weight loss in Region III is due to lipid volatilization/degradation that occurs in temperature range of 345-430 °C. Interestingly, the derivative weight loss profile of microalgae showed another peak at temperature range of 430-530 °C (Region IV). Our previous TGA experiments of pure protein showed some protein (e.g. lysozyme) degradation occurs at higher temperature range (up to 500 °C) [72].

While we did not analyzed the protein type in microalgae in this study, the peak at Region

IV can be due to degradation of larger protein molecules in microalgae. It is also possible that bio-char decompose at such a high temperature.

Figure D3b demonstrates a simulated fractional pyrolysis on TGA. For this

-1 analysis, microalgae heated under N2 to 320 °C with 10 °C min temperature ramp and then held at 320 °C for 10 min. Then, the residue from first fraction cooled down to room

173 temperature and then heated again to 450 °C and held for 10 min. As observed, the weight loss from the first fraction is nearly 40%. The microalgae that is used in this study, is comprised of nearly 40% carbohydrate and protein (Table 5.1). Thus, the weight loss from the first fraction is mainly contributed to volatilization/decomposition of carbohydrate and protein constituents in microalgae. Interestingly, the residue from the first fraction did not show any weight loss when the temperature increased from room temperature up to 320

°C, which indicates that holding the microalgae at 320 °C for 10 min was sufficient to volatilize most of the carbohydrate and protein. Nearly 30% weight loss was achieved from the second fraction, which is close to the lipid content of the microalgae. The TGA analysis shows that by fractional pyrolysis of microalgae first at 320 °C and then 450 °C, the products from carbohydrate and protein volatilization/decomposition can be recovered from the first fraction and the products from lipid constituents can be obtained from the second fraction.

174

(a) Intensity

(b)

Retention time (min)

Figure D4. GC-MS chromatogram of bio-oil from (a) first and (b) second fraction of catalyst free fractional pyrolysis. (

175

( (a)

a) Intensity (

(b)b)

Retention time (min)

Figure D5. GC-MS chromatogram of bio-oil from (a) first and (b) second fraction of fractional pyrolysis in presence of ex-situ catalyst.

176

oil) -

Benzene 20 Toluene

Xylene Aromatic selectivity of % bio (wt. Aromatic selectivity

0 1 stst fraction fraction 2nd2nd fraction fraction

Figure D6. Selectivity of aromatics produced from microalgae fractional pyrolysis in presence of ex-situ catalyst.

177