Hatami Tahmasb D.Pdf

UNIVERSIDADE ESTADUAL DE CAMPINAS

FACULDADE DE ENGENHARIA DE ALIMENTOS

TAHMASB HATAMI

Effects of grinding time, grinding load, and cold pressing on the aromatic compounds content of extract from fennel obtained by supercritical fluid extraction: experimental and mathematical modeling

Efeitos do tempo de moagem, carga de moagem, e prensagem a frio no conteúdo de compostos aromáticos do extrato de funcho obtido por extração com fluido supercrítico: experimental e modelagem matemática

CAMPINAS 2018 TAHMASB HATAMI

EFFECTS OF GRINDING TIME, GRINDING LOAD, AND COLD PRESSING ON THE AROMATIC COMPOUNDS CONTENT OF EXTRACT FROM FENNEL OBTAINED BY SUPERCRITICAL FLUID EXTRACTION: EXPERIMENTAL AND MATHEMATICAL MODELING

EFEITOS DO TEMPO DE MOAGEM, CARGA DE MOAGEM, E PRENSAGEM A FRIO NO CONTEÚDO DE COMPOSTOS AROMÁTICOS DO EXTRATO DE FUNCHO OBTIDO POR EXTRAÇÃO COM FLUIDO SUPERCRÍTICO: EXPERIMENTAL E MODELAGEM MATEMÁTICA

PhD thesis presented at the School of Food Engineering, Estate University of Campinas, as part of the requirements for obtaining the title of Doctor in Food Engineering.

Tese de doutorado apresentada à Faculdade de Engenharia de Alimentos, da Universidade Estadual de Campinas, como parte dos requisitos exigidos para obtenção do título de Doutor em Engenharia de Alimentos.

Supervisor: Prof. MARIA ANGELA DE ALMEIDA MEIRELES PETENATE

ESTE EXEMPLAR CORRESPONDE À VERSÃO FINAL DA TESE DEFENDIDA PELO ALUNO: TAHMASB HATAMI E ORIENTADA PELA PROF.ª DR.ª MARIA ANGELA DE ALMEIDA MEIRELES PETENATE

CAMPINAS 2018 Agência(s) de fomento e nº(s) de processo(s): Não se aplica.

Ficha catalográfica Universidade Estadual de Campinas Biblioteca da Faculdade de Engenharia de Alimentos Claudia Aparecida Romano - CRB 8/5816

Hatami, Tahmasb, 1985- H28e HatEffects of grinding time, grinding load, and cold pressing on the aromatic compounds content of extract from fennel obtained by supercritical fluid extraction : experimental and mathematical modeling / Tahmasb Hatami. – Campinas, SP : [s.n.], 2018.

HatOrientador: Maria Angela de Almeida Meireles Petenate. HatTese (doutorado) – Universidade Estadual de Campinas, Faculdade de Engenharia de Alimentos.

Hat1. Extraçâo supercrítica. 2. Prensagem. 3. Moagem. 4. Modelagem matemática. 5. Método dos elementos finitos. I. Petenate, Maria Angela de Almeida Meireles. II. Universidade Estadual de Campinas. Faculdade de Engenharia de Alimentos. III. Título.

Informações para Biblioteca Digital

Título em outro idioma: Efeitos do tempo de moagem, carga de moagem, e prensagem a frio no o conteúdo de compostos aromáticos do extrato de funcho obtido por extração com fluido supercrítico : experimental e modelagem matemática Palavras-chave em inglês: Supercritical extraction Pressing Grinding Mathematical modeling Finite element method Área de concentração: Engenharia de Alimentos Titulação: Doutor em Engenharia de Alimentos Banca examinadora: Maria Angela de Almeida Meireles Petenate [Orientador] Priscilla Carvalho Veggi Paulo de Tarso Vieira e Rosa Pedro Esteves Duarte Augusto Reginaldo Guirardello Data de defesa: 30-05-2018 Programa de Pós-Graduação: Engenharia de Alimentos

______Prof.ª Dr.ª Maria Angela de Almeida Meireles Petenate ORIENTADORA – DEA/FEA/UNICAMP

______Prof.ª Dr.ª Priscilla Carvalho Veggi MEMBRO TITULAR - USP

______Prof. Dr. Paulo de Tarso Vieira e Rosa MEMBRO TITULAR – IQ/UNICAMP

______Prof. Dr. Pedro Esteves Duarte Augusto MEMBRO TITULAR - USP

______Prof. Dr. Reginaldo Guirardello MEMBRO TITULAR – FEQ/UNICAMP

A ata da Defesa, assinada pelos membros da Comissão examinadora, consta no processo de vida acadêmica do aluno. DEDICATED TO

my mother and my father for their patient, kindness, and endless support. I have passed many years in abroad without thanking them worthy, but they have never passed even a moment without loving and supporting me continuously. They are the best parents ever.

ACKNOWLEDGEMENTS

I would first like to express my sincere gratitude to my supervisor, Prof. M. Angela A. Meireles, for her perfect supervision and valuable guidance. She is the best supervisor I have ever had in my academic life. My sincere thanks also go to Prof. Ozan N. Ciftci, Department of Food Science and Technology, University of Nebraska-Lincoln, USA who provided me an excellent opportunity to join his lively team as a visiting researcher. I am using this opportunity to thank all professors in School of Food Engineering, UNICAMP, for their patience, motivation, and immense knowledge. In particular, I would like to thanks Prof. Julian Martínez and Prof. Douglas Fernandes Barbin, both from UNICAMP, and Prof. Juliana Martin do Prado, from UFSCAR, for their great comment on the first draft of my thesis. My honest thanks also go to my colleagues in LASEFI, Ariovaldo Astini, Gislaine Chrystina Nogueira de Faria, Renata Vardanega, Grazielle Náthia Neves, Juliana Queiroz Albarelli, Pedro Ivo Nunes, Ricardo Abel Del Castillo Torres, Diego Tresinari dos Santos, Eric Keven Silva, and Juan Felipe Osorio Tobón. In particular, I am deeply appreciated my friend Júlio Cezar Flores Johner for helping and advising me in every stage of this thesis. I am also appreciated my colleagues in Ciftci lab, Lisbeth Vallecilla Yepez, Junsi Yang, and Ali Ubeyitogullari during my stay in UNL for all the assistance provided in performing experiments and for the fellowship and friendship. I would like to thank the examination board of my defense meeting for their participation and suggestions that enhanced the quality of thesis. Last but not the least, I am really grateful my friends, parents, brothers, and sisters for supporting me spiritually throughout not only my PhD program, but also my life.

ABSTRACT This thesis investigated effects of grinding procedures and cold pressing on the supercritical fluid extraction (SFE) from fennel, and modeled the process mathematically. In the first part of this thesis, the impacts of grinding time (GT) and mass of raw material in mill (ms) were considered on both global SFE yield from fennel and its main volatile oil content namely anethole and fenchone. For this purpose, the extractor was filled with milled fennel obtained at various values of ms, from 15 g to 35 g, and GT, from 15 s to 20 min. Extractor was o subjected to pressure of 200 bar, temperature of 40 C, and supercritical CO2 flow rate of 1.67×10-4 kg/s for 10 minutes. Then, extract composition was evaluated by Gas

Chromatography (GC) analysis. For better understanding how GT and ms affect SFE yield and extract composition, their effects were also investigated on temperature raising and diameter of fennel seeds during grinding process. It was found that ms and GT have considerable effects on anethole and fenchone content of fennel extract. In the second part of this thesis, the effect of GT, from 15 s to 20 min, was investigated on the dynamic yield of SFE from fennel over 80 min extraction. The experimental data was then modeled based on mass conservation law for both fluid and solid phases. Partial differential equations (PDEs) of the model were solved using Galerkin´s method on finite element method (FEM). The main feature of this part of thesis is that it reports a first complete solution strategy to solve the PDEs of SFE model using FEM. In the last part of this thesis, supercritical fluid extraction assisted by pressing (SFEAP) and SFE were compared in terms of fennel extraction kinetics as well as extract fractionation. Extractor was subjected to the pressure of 200 bar, temperature of 40 oC, and solvent to feed ratio of 100 using torques of 40 and 70 N.m. Fennel oil extracted with SFE and SFEAP were then successfully fractionated into the volatile oil and other fractions using two series of equal-sized separators. It was found that despite the superiority of SFEAP over SFE in terms of extraction performance, SFE gave better performance for obtaining volatile extract in the second separator.

Keywords: Supercritical extraction, pressing, grinding, mathematical modeling, finite element method

RESUMO

Esta tese investigou os efeitos dos procedimentos de trituração e prensagem a frio na extração com fluido supercrítico (SFE) de funcho e modelou o processo matematicamente. Na primeira parte desta tese, os impactos do tempo de moagem (GT) e da massa de matéria-prima (ms) foram considerados tanto no rendimento global de SFE a partir de funcho e conteúdo de seu principal óleo volátil, ou seja, anetol e fenchona. Para este propósito, o extrator foi preenchido com funcho moído obtido em vários valores de ms, de 15 g a 35 g, e GT, de 15 a 20 min. O o extrator foi submetido à pressão de 200 bar, temperatura de 40 C e vazão de CO2 supercrítico de 1,67 × 10-4 kg / s por 10 minutos. Em seguida, a composição do extrato foi avaliada por análise de cromatográfica gasosa (GC). Além disso, para uma melhor compreensão de como o

GT e o ms afetam o rendimento de SFE e a composição do extrato, seus efeitos foram investigados quanto ao aumento de temperatura e diâmetro das sementes de funcho durante o processo de moagem. Verificou-se que ms e GT têm efeitos consideráveis sobre o teor de anetol e fenchona do extrato de funcho. Na segunda parte desta tese, o efeito do GT, de 15 s a 20 min, foi investigado sobre o rendimento dinâmico de SFE a partir da extração de funcho durante 80 min. Os dados experimentais foram então modelados com base na lei de conservação de massa para as fases fluida e sólida. Equações diferenciais parciais (PDEs) do modelo foram resolvidas usando o método de Galerkin no método dos elementos finitos (MEF). A principal característica desta parte da tese é que ela relata uma primeira estratégia de solução completa para resolver os PDEs do modelo SFE usando FEM. Na última parte desta tese, a extração com fluido supercrítico assistida por prensagem (SFEAP) e SFE foram comparadas em termos de cinética de extração bem como o fracionamento do extrato de funcho. O extrator foi submetido à pressão de 200 bar, temperatura de 40 oC e uma relação de massa de solvente por massa de matéria-prima de 100 utilizando os torques de 40 e 70 N.m. O óleo de funcho extraído com SFE e SFEAP foi então fracionado com sucesso no óleo volátil e outras frações utilizando duas séries de separadores de tamanho igual. Verificou-se que, apesar da superioridade da SFEAP em relação ao SFE em termos de desempenho de extração, o SFE apresentou melhor desempenho na obtenção de extrato volátil no segundo separador.

Palavras-chave: Extração supercrítica, prensagem, moagem, modelagem matemática, método dos elementos finitos.

LIST OF FIGURES Figure 1.1 - Flowchart of thesis structure...... 17 Figure 3.1 - Schematic diagram of the mill...... 33 Figure 3.2 - Schematic diagram of the SFE unit from (a) top view and (b) front view (Johner and Meireles, 2016)…………………………….……………………………………………..34 Figure 3.3 - The average size of Fennel particles at various values of GT and ms ...... 35 Figure 3.4 - The raising temperature of Fennel at the end of grinding process ...... 35 Figure 3.5 - Fennel distributions inside the crushing chamber of mill ...... 36

Figure 3.6 - Effect of GT and ms on the overall extraction yield obtained by SFE from Fennel at 200 bar, 313 K, and 1.67×10-4 kg/s during 10 min...... 36

Figure 3.7 - Effect of GT and ms on the anethol extraction yield obtained by SFE from Fennel at 200 bar, 313 K, and 1.67×10-4 kg/s, during 10 min………………...... 36

Figure 3.8 - Effect of GT and ms on the fenchone extraction yield obtained by SFE from Fennel at 200 bar, 313 K, and 1.67×10-4 kg/s, during 10 min...... 37 Figure 4.1 - The effect of GT on the temperature increase (ΔT) and the average size of fennel particles (dp) (Hatami et al., 2017)...... 44 Figure 4.2 - SFE unit used in this study...... 46 Figure 4.3 - Finite element mesh of the extractor ...... 49 Figure 4.4 - Comparison between the model results and the experimental data for SFE from fennel ...……………………………………………………………………………………….59 Figure 4.5 - Effect of the number of mesh elements (n) on the FEM and FDM results for Exp. 3……………………………………………………………………………………………….60 Figure 4.6 - The extractor with one element……...... 61 Figure 5.1 - Schematic diagram of the SFEAP unit: (A1) Pressing system, (A2) Extractor, (B) First separator, (C) Second separator, (D) Outlet of separators, (E) Leadscrew, (F) Socket sliding, (G) Torque meter, (H) Shut-off valve, (I) Micro-metering valve, and (J) Gas flow meter.……………………………………………………………………………...…………. 72 Figure 5.2 - Extraction yield of milled fennel (GT = 6 min) using SFE and SFEAP at 200 bar and 40 °C. …...……………………………………………………….……………………….74 Figure 5.3 – A schematic of (A) raw material after extraction using SFE, (B) raw material after extraction using SFEAP, (C) SFE extract in Separator 1 (D) SFE extract in Separator 2, (E) SFEAP extract in Separator 1, (F) SFEAP extract in Separator 2……...……………..….77

LIST OF TABLES Table 4.1 - Process parameters for SFE from 1×10-2 kg fennel at 313 K and 200 bar in the extractor, with 2×10-2 m internal diameter…………...... 47 Table 4.2 - The parameters of the SFE model...... 57 Table 5.1 - The significance of temperature, pressure, S/F, particle diameter, and torque on the SFE yield from fennel...... 76

Table 5.2 - Comparison between SFE and SFEAP in terms of extraction and fractionation...78

TABLE OF CONTENT ABSTRACT...... 7 LIST OF FIGURES...... 9 LIST OF TABLES...... 10 CHAPTER 1 - GENERAL INTRODUCTION, OBJECTIVES AND STRUCTURE OF THE THESIS……………………………………………………………………………………….13 1.1. INTRODUCTION...... 14 1.2. JUSTIFICATION...... 15 1.3. OBJETIVES...... 15 1.4. THESIS STRUCTURE...... 16 REFERENCES...... 17 CHAPTER 2 - LITERATURE REVIEW...... 20 1. SFE FROM FENNEL...... 21 2. MATHEMATICAL MODELING OF SFE...... 23 3. SFEAP FROM FENNEL...... 25 REFERENCES...... 26 CHAPTER 3 - EFFECTS OF GRINDING TIME ON SUPERCRITICAL FLUID EXTRACTION……………………………………………………………………………….30 1. INTRODUCTION...... 32 2. MATERIALS AND METHODS...... 33 3. RESULTS AND DISCUSSION...... 35 4. CONCLUSIONS...... 37 REFERENCES...... 37 CHAPTER 4 - MATHEMATICAL MODELING OF SUPERCRITICAL FLUID EXTRACTION……………………………………………………………………………….39 1. INTRODUCTION...... 42 2. MATERIAL AND METHODS...... 45 3. RESULTS AND DISCUSSION...... 57 4. CONCLUSIONS...... 61 REFERENCES...... 62 CHAPTER 5 - SUPERCRITICAL FLUID EXTRACTION ASSISTED BY COLD PRESSING…………………………………………………………………………………....66 1. INTRODUCTION...... 68

2. MATERIAL AND METHODS...... 70 3. RESULTS AND DISCUSSION...... 73 4. CONCLUSION...... 79 REFERENCES...... 79 CHAPTER 6 GENERAL DISCUSSIONS...... 82 CHAPTER 7 GENERAL CONCLUSIONS...... 85 MEMORANDUM OF THE PhD PERIOD...... 87 REFERENCES...... 91 APPENDIX...... 98

CHAPTER 1

GENERAL INTRODUCTION, OBJECTIVES AND STRUCTURE OF THE THESIS ______

CHAPTER 1

General Introduction, Objectives and Structure of the Thesis

1.1. INTRODUCTION

Fennel (Foeniculum vulgare) is a plant belonging to Apiaceae family, which is cultivated in several countries like Brazil, England, Germany, and China (Brender et al., 1997). Fennel extracts can be employed as antispasmodic, anti-inflammatory, expectorant, diuretic, and laxative (Jahromi et al., 2003). It also has applications in treatment of nervous disturbances, carminative, analgesic, and stimulant of gastrointestinal mobility (Jahromi et al., 2003). One of the promising methods for extraction of oil from fennel is supercritical fluid extraction (SFE) (Yamini et al., 2002; Piras et al., 2014; Díaz-Maroto et al., 2005). Supercritical fluids (SCF) have proved to be effective solvents for extraction from solids, and SFE are widely used by many researchers (Mackėla et al., 2017; Conde-Hernández et al., 2017) and industries due to its high mass transfer rate as well as simple separation of solvent from extract at the end of process (Gallo et al., 2017; Antunes-Ricardo et al., 2017). Two of the attractive research areas in the field of SFE from fennel are either maximizing overall extraction yield or maximizing anethole and fenchone content of the volatile oil. The first necessary step for these goals is grinding fennel seeds to increase their specific area. Grinding procedure has not been sufficiently addressed so far in SFE field. Two main key factors in grinding are grinding time and grinding load. Although the effect of grinding time (GT) was already investigated on the SFE yield of other raw materials, researchers limited themselves to small GT (Wilkinson et al., 2014; Yahya et al., 2010). After investigating the effects of grinding time and load on the SFE yield experimentally, it is very important to model the SFE process mathematically (Hatami et al., 2014) in order to evaluate the interaction effects of factors and optimize the process accordingly. To this end, two main SFE models can be found in literature, empirical models, and mass transfer-based models. Although the empirical models have the advantages of simplicity, they are not suitable for scaling-up (del Valle and Fuente, 2006). Mass transfer-based models generally result in two partial differential equations (PDEs), one PDE for fluid phase and the other for solid phase (del Valle and Fuente, 2006; Oliveira et al., 2011). These kinds of models can be solved using numerical techniques such as finite difference method (FDM) or finite element method (FEM).

In addition to grinding procedure, another important preprocessing factor in SFE is cold pressing. Accordingly, a novel extraction technique has been recently developed by Johner et al. (2018), which is called SFEAP. SFEAP stands for supercritical fluid extraction assisted by cold pressing, which performance was successfully evaluated for extraction from pequi (Caryocar brasiliense) (Johner et al., 2018). However, more experiments with various raw materials, like fennel, are still required to evaluate this new technique in comparison with SFE.

1.2. JUSTIFICATION

As effects of milling process on yield and quality of SFE were not fully discussed in literature and researchers limited themselves to small grinding time (GT), the first aim of the current study was to focus on this area for a broad range of GT and various grinding loads. The experimental SFE data was then modeled based on mass conservation law for both fluid and solid phases. The main novelty of modeling part is presenting a systematic FEM technique for solving SFE model, which has not been reported yet. Additionally, most publications of SFE from fennel focused on the effects of temperature and pressures on the SFE performance, but there was no publication about SFEAP from fennel yet. Therefore, the third target of this thesis was to employ SFEAP for the extraction from fennel at constant temperature and pressure, but at different torques. Moreover, this thesis aimed to highlight the fractionation of extracts from SFE and SFEAP into volatile fraction and lipid fraction rich products and compare their composition to literature data.

1.3. OBJECTIVES

1.3.1. General objective

Evaluating the effects of grinding process and cold pressing on the aromatic compounds content of extract from fennel obtained by supercritical fluid extraction.

1.3.2. Specific objectives

 Studying impacts of grinding time (GT) and mass of raw material in mill (ms) on both global SFE yield from fennel and its main volatile oil content, namely anethole and fenchone.

 Mathematical modeling of SFE from fennel, and providing a step by step solution strategy of the model using finite element method.

 Employing supercritical fluid extraction assisted by cold pressing (SFEAP) for extraction and fractionation from fennel.

 Evaluating the effects of torque in SFEAP on the overall extraction yield.

1.4. THESIS STRUCTURE

The text is presented in 7 chapters. In chapter 1 - General Introduction, Objectives and Structure of the Thesis - the most relevant aspects for the formulation of the work are presented. This chapter presents a brief introduction of the scenario that propitiated the development of this thesis, the justification, the objectives outlined and the planned structure for the development of the works. In Chapter 2 - Literature Review - a review on the topics covered in the thesis is presented. Literature on the extraction and fractionation from fennel, mathematical modeling, and integrating of cold pressing with SFE are discussed. In Chapter 3 - Effects of grinding procedure on supercritical fluid extraction - the article published in Industrial Crops & Products is presented. In Chapter 4 - Mathematical modeling of supercritical fluid extraction - the article submitted to the Journal of Food Engineering is presented. In Chapter 5 - Supercritical fluid extraction assisted by cold pressing - the article submitted to Industrial Crops & Products is presented. Chapter 6 - General Discussion - brings an integrated discussion of the results obtained in chapters 3, 4 and 5. Finally, chapter 7 - General Conclusions presents the general understandings and outlines obtained in this thesis. Figure 1.1 shows a flowchart of thesis structure. As depicted in this figure, the thesis contains three main parts, namely the effect of grinding procedure on SFE, mathematical modeling of SFE, and integration of cold pressing with SFE.

Figure 1.1 – Flowchart of thesis structure.

REFERENCES

Antunes-Ricardo, M., Gutiérrez-Uribe, J.A., Guajardo-Flores, D., 2017. Extraction of isorhamnetin conjugates from Opuntia ficus-indica (L.) Mill using supercritical fluids. J. Supercrit. Fluids, 119, 58-63.

Brender, T., Gruenwald, J., Jaenicke, C., 1997. Herbal Remedies, Phytopharm Consulting

Institute for Phytopharmaceuticals (2. ed.). Schaper & Brümmer GmbH & Co., Salzgitter,

Berlin, Germany.

Conde-Hernández, L.A., Espinosa-Victoria, J.R., Trejo, A., Guerrero-Beltrán, J.Á., 2017. CO2-supercritical extraction, hydrodistillation and steam distillation of essential oil of rosemary (Rosmarinus officinalis). J. Food Eng. 200, 81-86.

del Valle, J.M., Fuente, J.C.D.L., 2006. Supercritical CO2 extraction of oilseeds: review of kinetic and equilibrium models. Crit. Rev. Food Sci. Nutr. 46, 131-160.

Díaz-Maroto, M.C., Díaz-Maroto Hidalgo, I.J., Sánchez-Palomo, E., Pérez-Coello, M.S., 2005. Volatile components and key odorants of fennel (Foeniculum vulgare Mill.) and thyme (Thymus vulgaris L.) oil extracts obtained by simultaneous distillation− extraction and supercritical fluid extraction. J. Agric. Food Chem. 53(13), 5385-5389.

Gallo, M., Formato, A., Ianniello, D., Andolfi, A., Conte, E., Ciaravolo, M., Varchetta, V., Naviglio, D., 2017. Supercritical fluid extraction of pyrethrins from pyrethrum flowers (Chrysanthemum cinerariifolium) compared to traditional maceration and cyclic pressurization extraction. J. Supercrit. Fluids, 119, 104-112.

Hatami, T., Moura, L.S., Khamforoush, M., Meireles, M.A.A., 2014. Supercritical fluid extraction from Priprioca: Extraction yield and mathematical modeling based on phase equilibria between solid and supercritical phases. J. Supercrit. Fluids, 85, 62-67.

Jahromi, B.N., Tartifizadeh, A., Khabnadideh, S., 2003. Comparison of fennel and mefenamic acid for the treatment of primary dysmenorrhea. Int. J. Gynaecol. Obstet. 80, 153-157.

Johner J.C.F., Hatami, T., Meireles, M.A.A., 2018. Developing a supercritical fluid extraction method assisted by cold pressing for extraction of pequi (Caryocar brasiliense). J. Supercrit. Fluids, 137, 34-39.

Mackėla, I., Andriekus, T., & Venskutonis, P.R., 2017. Biorefining of buckwheat (Fagopyrum esculentum) hulls by using supercritical fluid, Soxhlet, pressurized liquid and enzyme-assisted extraction methods. J. Food Eng. 213, 38-46.

Oliveira, E.L., Silvestre, A.J., Silva, C.M., 2011. Review of kinetic models for supercritical fluid extraction. Chem. Eng. Res. Des. 89, 1104-1117.

Piras, A., Falconieri, D., Porcedda, S., Marongiu, B., Gonçalves, M.J., Cavaleiro, C., Salgueiro, L., 2014. Supercritical CO2 extraction of volatile oils from Sardinian Foeniculum vulgare ssp. vulgare (Apiaceae): chemical composition and biological activity. Nat. Prod. Res. 28(21), 1819-1825.

Wilkinson, N., Hilton, R., Hendry, D., Venkitasamy, C., Jacoby, W., 2014. Study of process variables in supercritical carbon dioxide extraction of soybeans. Food Sci. Technol. Int. 20(1), 63-70.

Yahya, F., Lu, T., Santos, R.C.D., Fryer, P.J., Bakalis, S., 2010. Supercritical carbon dioxide and solvent extraction of 2-acetyl-1-pyrroline from Pandan leaf: The effect of pre-treatment. J. Supercrit. Fluids 55, 200–207.

Yamini, Y., Sefidkon, F., Pourmortazavi, S.M., 2002. Comparison of essential oil composition of Iranian fennel (Foeniculum vulgare) obtained by supercritical carbon dioxide extraction and hydrodistillation methods. Flavour Fragr. J. 17(5), 345-348.

CHAPTER 2 LITERATURE REVIEW ______21

CHAPTER 2

1. SFE FROM FENNEL

Fennel (Foeniculum vulgare) is a plant belonging to Apiaceae family, which is cultivated in several countries like Brazil, England, Germany, and China among others (Brender et al., 1997). Fennel extracts can be employed as antispasmodic, anti-inflammatory, expectorant, diuretic and laxative (Jahromi et al., 2003). It also has applications in treatment of nervous disturbances, carminative, analgesic, and stimulant of gastrointestinal mobility (Jahromi et al., 2003). Anethole and fenchone are two of the main compounds in fennel volatile oil that compose, respectively, 40 to 70% and 1 to 20% of it (Bernath et al., 1996; Coşge et al., 2008; Raghavan, 2006). Extraction of volatile oil from fennel is generally carried out by several techniques such as supercritical fluid extraction (SFE), liquid CO2 extraction, soxhlet extraction, accelerated solvent extraction, and steam distillation (Baby and Ranganathan, 2016; Rodríguez-Solana et al., 2014a; Bodsgard et al., 2016; Johner and Meireles, 2016). Among all extraction techniques, SFE is usually a preferable technique for extraction from various raw materials due to its advantages such as high extraction performance, simple separation of solvent from extract, and selectivity capability (Durante et al., 2014; Ding et al., 2017). In a typical SFE process, solvent (CO2) flows slowly through a bed containing raw material. Extract is then separated from solvent after passing the solvent- extract mixture through a reduction pressure system. SFE from fennel has been reported in various research works. Reverchon et al. (1999) carried out successfully extraction of fennel seeds in two steps including SFE at 90 bar and 50 °C, and SFE at 200 bar and 40 °C using three CO2 mass flow rates of 0.5, 1.0, and 1.5 kg/h. They also modeled the process based on mass conservation law and demonstrated the reliability of the model by comparing it to experimental data. Yamini et al. (2002) obtained essential oil from Iranian fennel by hydro- distillation and SFE techniques. They investigated the effects of pressure (202.65 and 354.64 bar), temperature (45 and 55 oC), extraction time (30 and 45 min), and modifier (methanol) volume (80 and 400 µl) on the composition of extract at constant supercritical CO2 flow rate of approximately 0.3–0.4 ml/min. It was found that the optimum operating condition for maximizing percentage of anethole in extract using SFE was the pressure of 354.64 bar, temperature of 55 °C, extraction time of 45 min, and modifier percentage of 5%. Under this 22

condition, the percentage of anethole in the extract was 90.14%, while the corresponding value using hydro-distillation method was 69.41%. These numbers together with the shorter extraction time of SFE compared to hydro-distillation (4 h), and selectivity of SFE confirmed its superiority over hydro-distillation. They reported that other main components in the extract were fenchone, 6.25% by SFE versus 11.00 % by hydro-distillation, and Limonene, 2.82 % by SFE versus 10.00 % by hydro-distillation. Moura et al. (2005) studied effect of harvesting season, degree of maturation, bed geometry, operating temperature and pressure on SFE yield from fennel. They also studied extraction of fennel by hydro-distillation and low-pressure solvent extraction, and compared their results to SFE. They found that not only SFE process produced larger relative proportion of anethole and fenchone, but also its overall extraction yield was 5.2 larger than that of hydro-distillation (Moura et al., 2005). Díaz-Maroto et al. (2005) compared SFE from fennel to simultaneous distillation-extraction (SDE) at 40 oC and 120 bar during 25 min. Analyzing the extracts by gas chromatography-mass spectrometry (GC-MS) confirmed that trans-anethole had the highest proportion of extract, 63.80% by SFE versus 49.71% by SDE, followed by estragole, 20.33 % by SFE versus 25.84 % by SDE, and fenchone, 12.71 % by SFE versus 19.33 % by SDE. In another relevant paper, Pereira and Meireles (2007) investigated economic analysis for obtaining extracts and essential oils from fennel using SFE and steam distillation. They concluded that in spite of high cost of SFE equipment, it was cheaper than steam distillation due to higher quality of its product, higher extraction yield, and lower energy consumption (Pereira and Meireles, 2007). Another paper in this topic was written by Hammouda et al. (2014), who studied extraction yield and extract quality from fennel using SFE, microwave-assisted extraction (MAE), and hydro-distillation. Their study showed that MAE gave higher overall yield and higher percentage of fenchone than others. SFE, on the other side, gave the maximum percentage of anethole (Hammouda et al., 2014). Piras et al. (2014) employed SFE and hydro-distillation on Sardinian wild fennel and measured composition of volatile extracts using gas chromatography –mass spectrometry. They found that percentage of the main components in SFE extract were 7.1% (fenchone), 34.9% (estragole), and 24.6% ((E)-anethole), while corresponding number for hydro- distillation extract were 8.8%, 42.6%, and 43.4%, respectively. Rodríguez-Solana et al. (2014b) performed oleoresin extraction from fennel seeds using SFE, and optimized the process for the aim of attaining maximum estragole per kg of dry plant. They found the optimal operating condition at the pressure of 240 bar, temperature of 60 oC, extraction duration of 3.41 h, and methanol percentage of 3%. In this condition, they obtained 1320 ± 260 mg of estragole per kg dry plant. Ivanovic et al. (2014) increased the yield of SFE for 23

several raw materials by employing different mechanical pretreatment such as flaking, impact plus shearing, and cutting plus grinding. They found flaking as the most convenient method for increasing the SFE yield from fennel seeds. According to the above literature review, two of the attractive research areas in the field of SFE from fennel are either maximizing overall yield or maximizing anethole and fenchone content of the volatile oil. The first necessary step for attending these goals is, however, grinding the fennel seeds to increase their specific area. Although the effect of grinding time (GT) was already investigated on the SFE yield of other raw materials, they limited themselves to small GT. For example, Wilkinson et al. (2014) investigated the effect of GT on the SFE yield from soybeans at 483 bar and 80 oC. They placed 60 g of soybeans in a mechanical grinder for different GT from 10 to 60 s. They concluded that increasing GT enhanced the extraction yield due to the reduction of particle sizes. The SFE yield was increased from around 6% to 22% by decreasing the particle diameters from 0.20 mm to 0.07 mm. In another study by Yahya et al. (2010), SFE yield from Pandan leaf increased up to 50% after a pretreatment grinding for 30 s.

2. MATHEMATICAL MODELING OF SFE

After determining the SFE yield experimentally, it is very important to model the process mathematically (Hatami et al., 2014) not only to minimize the required number of experiments for sensitivity analysis, but also to optimize the process. To this end, two main SFE models can be found in literature: 1) empirical models 2) mass transfer-based models. Although the empirical models have the advantages of simplicity, they are not suitable for scaling-up (del Valle and Fuente, 2006). Mass transfer-based models generally result in two partial differential equations (PDEs), one PDE for fluid phase and the other for solid phase (del Valle and Fuente, 2006; Oliveira et al., 2011). Several models were proposed for SFE process in literature. Among all, Madras et al. (1994) proposed one of the general model based on the following assumptions: spherical particles with constant porosity, constant temperature and pressure, axial dispersion flow, unchanged bed porosity, and plug flow pattern for fluid velocity. Accordingly, mass balance equation for bulk fluid phase is: ∂C ∂C ∂2C 3(1 − ε) (1) f + v f − D f = J ∂t ∂z l ∂z2 εR

With the following boundary and initial conditions: 24

At 푡 = 0 ⟹ 퐶푓 = 0 (2) ∂C 퐴푡 푧 = 0 ⟹ 퐷 f = 푣C 푙 ∂z f

∂C 푎푡 푧 = 퐿 ⟹ f = 0 ∂z Mass balance equation for solid phase is: ∂C ∂C (3) p 1 ∂Cs De ∂ 2 p + = 2 (r ) ∂t εp ∂t r ∂r ∂r

With the following boundary and initial conditions:

At 푡 = 0 ⟹ Cp = Cp0 & Cs = Cs0 (4) ∂Cp ∂C 퐴푡 푟 = 0 ⟹ = 0 & s = 0 ∂r ∂r

∂Cp 푎푡 푟 = 푅 ⟹ ε D = −J p e ∂r

3 3 3 Where, Cf (kg/m ), Cp (kg/m ), and Cs (kg/m ) are solute concentrations in the bulk fluid, pore of particles, and solid phase, respectively. Relationship between Cp and Cs must be obtained based on thermodynamic equilibrium between pore and solid of particles. z (m) is axial 2 2 coordinate along extractor, t (s) is time, Dl (m /s) is axial dispersion coefficient, De(m /s) is effective diffusion coefficient in pore of particles, v (m/s) is interstitial velocity, R (m) is particle radius, L (m) is height of the bed, ε is bed porosity, εp is particle porosity, and J is mass transfer rate, which is calculated as follow:

J = k (C | − C ) f p r=R f (5)

Where, kf (m/s) is external mass transfer coefficient. To reduce computational cost in this model, further simplifications may be employed depending on the geometrical and operating factors of the process under consideration. Some of the well-known simplifications are external mass transfer control, internal mass transfer control, no solute dispersion, linear driving force, and steady-state approximations. Each of these assumption changes one or both aforementioned mass balances PDEs of fluid and solid phases. A detail explanation in this issue can be found in a publication by del Valle and Fuente (2006). These kinds of models can be solved using either analytical or numerical techniques. Analytical techniques have been frequently used, and they were explained in literature in detail (Lucas et al., 2007; Mongkholkhajornsilp et al., 2005). As extra simplification is employed for analytical 25

techniques, such as ignoring axial dispersion effect in the papers by Lucas et al. (2007) and Mongkholkhajornsilp et al. (2005), they always have an inherent error. About numerical techniques, SFE models can be solved by either finite difference method (FDM) or finite element method (FEM). The fundamentals of FDM and FEM can be found in a number of sources (LeVeque, 2007; Kwon and Bang, 2000; Reddy, 1993). FDM partitions the PDE domain, and approximates the partial derivation terms at each point from neighboring values (Causon et al., 2010). FDM has been broadly used in SFE, and it was explained in detail by Meireles et al. (2009). FEM is another powerful numerical technique that had already employed for SFE modeling (Madras et al., 1994; Valderrama and Alarcón, 2009). However, there is no publication in literature yet to explain FEM on SFE model in detail. Such a publication is important for conceptual understanding of the FEM solution of the SFE process, and decreasing the computation cost for future relevant researches in this area.

3. SFEAP FROM FENNEL

Cold pressing is another preprocessing factor that can improve the SFE yield. This improvement of extraction yield is possibly due to the breaking of matrix raw material by applying pressing that consequently more oil liberates from particles, and so more oil exposes to supercritical CO2. SFE assisted by pressing (SFEAP) is a novel technique of extraction that has been recently developed by Johner et al. (2018). SFEAP is an integration of SFE and cold pressing methods, whose performance was evaluated for extraction from pequi (Caryocar brasiliense). Prior to study the impact of pressing, they studied SFE yield from pequi at two isotherms, 313 and 333 K, and pressure from 20 to 40 MPa. It was found that a temperature of 333 K and pressure of 40 MPa, among others, gave the maximum SFE yield, 48 g extract/100 g raw material. So, they investigated the impact of pressing with torque of 40 N.m under the optimum condition of temperature and pressure. Comparing the SFEAP yield at 400 bar, 40 oC to that of SFE revealed that the SFEAP yield was eight times greater than that of SFE during the first minute of extraction (Johner et al., 2018). According to the analysis of pequi oil, SFEAP did not change the fatty acid composition compared to SFE. As research in SFEAP area is still in the initial phase, there is no publication in literature regarding SFEAP from fennel yet. SFEAP can be also evaluated in terms of fractionation of fennel extract to produce volatile fraction and lipidic fraction rich products. Previous publications about fractionation of fennel extract using SFE revealed that temperature and pressure of extractor and first separator are the most critical factors affecting the fractionation performance. Simandi et al. (1999) fractionated SFE extract from fennel into volatile fraction rich and 26

lipidic fraction rich products using two subsequent separators. They recommended 80-84 bar and 31-35 °C as the best operating conditions of the first separator to minimize the presence of undesired components. Johner and Meireles (2016) performed extraction and fractionation −4 from fennel using two separators with a flow rate of 2.00×10 kg/s of CO2 and solvent per feed mass ratio (S/F) of 10. Temperature and pressure in the extractor, first separator, and second separator were, respectively, 40 oC and 200 bar, 35 oC and 80 bar, and 8 oC and 20 bar. Analysis of extract showed that from the overall extraction yield of 2.8%, around 97.5% was accumulated in the first separator (waxy phase), and the remaining 2.5% accumulated in the second separator (oily phase containing the volatile compounds).

REFERENCES

Baby, K.C., Ranganathan, T.V., 2016. Effect of enzyme pre-treatment on extraction yield and quality of fennel (Foeniculum vulgare) volatile oil. Biocatal. Agric. Biotechnol. 8, 248-256.

Bernath, J., Nemeth, E., Kattaa, A., Hethelyi, E., 1996. Morphological and chemical evaluation of fennel (Foeniculum vulgare Mill.) populations of different origin. J. Essent. Oil Res. 8, 247-253.

Bodsgard, B.R., Lien, N.R., Waulters, Q.T., 2016. Liquid CO2 Extraction and NMR Characterization of Anethole from Fennel Seed: A General Chemistry Laboratory. J. Chem. Educ. 93, 397-400.

Brender, T., Gruenwald, J., Jaenicke, C., 1997. Herbal Remedies, Phytopharm Consulting Institute for Phytopharmaceuticals. Second ed., Schaper & Brümmer GmbH & Co., Salzgitter, Berlin, Germany.

Causon, D.M., Mingham, C.G., 2010. Introductory finite difference methods for PDEs. Bookboon.

Coşge, B., Kiralan, M., Gürbüz, B., 2008. Characteristics of fatty acids and essential oil from sweet fennel (Foeniculum vulgare Mill. var. dulce) and bitter fennel fruits (F. vulgare Mill. var. vulgare) growing in Turkey. Nat. Prod. Res. 22, 1011-1016.

del Valle, J.M., Fuente, J.C.D.L., 2006. Supercritical CO2 extraction of oilseeds: review of kinetic and equilibrium models. Crit. Rev. Food Sci. Nutr. 46, 131-160. 27

Ding, X., Liu, Q., Hou, X., Fang, T., 2017. Supercritical Fluid Extraction of Metal Chelate: A Review. Crit. Rev. Anal. Chem. 47(2), 99-118.

Durante, M., Lenucci, M.S., Mita, G., 2014. Supercritical carbon dioxide extraction of carotenoids from pumpkin (cucurbita spp.): A review. Int. J. Mol. Sci. 15, 6725-6740.

Hammouda, F.M., Saleh, M.A., Abdel-Azim, N.S., Shams, K.A., Ismail, S.I., Shahat, A.A., Saleh, I.A., 2014. Evaluation of the essential oil of foeniculum vulgare mill (fennel) fruits extracted by three different extraction methods by Gc/Ms. Afr. J. Tradit. Complement Altern. Med. 11, 277-279.

Ivanovic, J., Meyer, F., Stamenic, M., Jaeger, P., Zizovic, I., Eggers, R., 2014. Pretreatment of natural materials used for supercritical fluid extraction of commercial phytopharmaceuticals. Chem. Eng. Technol. 37, 1606-1611.

Jahromi, B.N., Tartifizadeh, A., Khabnadideh, S., 2003. Comparison of fennel and mefenamic acid for the treatment of primary dysmenorrhea. Int. J. Gynaecol. Obstet. 80, 153-157.

Johner, J.C.F., Meireles, M.A.A., 2016. Construction of a supercritical fluid extraction (SFE) equipment: validation using annatto and fennel and extract analysis by thin layer chromatography coupled to image. Food Sci. Technol. (Campinas) 36(2), 210-247.

Kwon, Y.W., & Bang, H., 2000. The Finite Element Method Using MATLAB, CRC Press. 28

LeVeque, R., 2007. Finite Difference Methods for Ordinary and Partial Differential Equations: Steady-State and Time-Dependent Problems (Classics in Applied Mathematics Classics in Applied Mathemat), Society for Industrial and Applied Mathematics.

Lucas, S., Calvo, M., Garcia-Serna, J., Palencia, C., Cocero, M., 2007. Two-parameter model for mass transfer processes between solid matrixes and supercritical fluids: Analytical solution. J. Supercrit. Fluids, 41, 257-266.

Madras, G., Thibaud, C., Erkey, C., Akgerman, A., 1994. Modeling of supercritical extraction of organics from solid matrices. AIChE J. 40(5), 777-785.

Meireles, M.A.A., Zahedi, G., Hatami, T., 2009. Mathematical modeling of supercritical fluid extraction for obtaining extracts from vetiver root. J. Supercrit. Fluids, 49, 23-31.

Mongkholkhajornsilp, D., Douglas, S. Douglas, P.L., Elkamel, A., Teppaitoon, W.,

Pongamphai, S., 2005. Supercritical CO2 extraction of nimbin from neem seeds––a modelling study. J. Food Eng. 71, 331-340.

Moura, L.S., Carvalho Jr, R.N., Stefanini, M.B., Ming, L.C., Meireles, M.A.A., 2005. Supercritical fluid extraction from fennel (Foeniculum vulgare): global yield, composition and kinetic data. J. Supercrit. Fluids 35, 212-219.

Oliveira, E.L., Silvestre, A.J., Silva, C.M., 2011. Review of kinetic models for supercritical fluid extraction. Chem. Eng. Res. Des. 89, 1104-1117.

Pereira, C.G., Meireles, M.A.A., 2007. Economic analysis of rosemary, fennel and anise essential oils obtained by supercritical fluid extraction. Flavour Frag. J. 22, 407-413.

Piras, A., Falconieri, D., Porcedda, S., Marongiu, B., Gonçalves, M.J., Cavaleiro,C., Salgueiro, L., 2014. Supercritical CO2 extraction of volatile oils from Sardinian Foeniculum vulgare ssp. vulgare (Apiaceae): chemical composition and biological activity. Nat. Prod. Res. 28, 1819-1825.

Raghavan, S., 2006. Handbook of spices, seasonings, and flavorings. CRC Press.

Reddy, J.N., 1993. An introduction to the finite element method, McGraw-Hill. 29

Reverchon, E., Daghero, J., Marrone, C., Mattea, M., Poletto, M., 1999. Supercritical fractional extraction of fennel seed oil and essential oil: experiments and mathematical modeling. Ind. Eng. Chem. Res. 38, 3069-3075.

Rodríguez-Solana, R., Salgado, J.M., Domínguez, J.M., Cortés-Diéguez, S., 2014a. Characterization of fennel extracts and quantification of estragole: Optimization and comparison of accelerated solvent extraction and Soxhlet techniques. Ind. Crops Prod. 52, 528-536.

Rodríguez-Solana, R., Salgado, J.M., Domínguez, J.M., Cortés-Diéguez, S., 2014b. Estragole quantity optimization from fennel seeds by supercritical fluid extraction (carbon dioxide– methanol) using a Box–Behnken design. Characterization of fennel extracts. Ind. Crops Prod. 60, 186-192.

Simandi B., Deak A., Ronyai E., 1999. Supercritical Carbon Dioxide Extraction and Fractionation of fennel Oil. Journal of Agricultural and Food Chemistry, 47, 1635−1640.

Valderrama, J.O. and Alarcón, R.F., 2009. A novel hybrid numerical technique to determine mass transport properties in supercritical fluid extraction processes. IJNMBE 25(2), 173-184.

Wilkinson, N., Hilton, R., Hendry, D., Venkitasamy, C., Jacoby, W., 2014. Study of process variables in supercritical carbon dioxide extraction of soybeans. Food Sci. Technol. Int. 20(1), 63-70.

Yamini, Y., Sefidkon, F., Pourmortazavi, S. M., 2002. Comparison of essential oil composition of Iranian fennel (Foeniculum vulgare) obtained by supercritical carbon dioxide extraction and hydrodistillation methods. Flavour Fragr. J. 17(5), 345-348.

CHAPTER 3

EFFECTS OF GRINDING TIME ON SUPERCRITICAL FLUID EXTRACTION ______

INVESTIGATING THE EFFECTS OF GRINDING TIME AND GRINDING LOAD ON CONTENT OF TERPENES IN EXTRACT FROM FENNEL OBTAINED BY SUPERCRITICAL FLUID EXTRACTION

Tahmasb Hatami, Júlio Cezar Flores Johner, Maria Angela de Almeida Meireles

LASEFI/DEA/FEA (School of Food Engineering), UNICAMP (University of Campinas), Campinas, SP, Brazil

Paper published in Industrial Crops & Products, 109: 85–91, Apr.-Aug. 2017.

https://www.sciencedirect.com/science/article/pii/S0926669017305216

Industrial Crops & Products 109 (2017) 85–91

Contents lists available at ScienceDirect

Industrial Crops & Products

journal homepage: www.elsevier.com/locate/indcrop

Investigating the eﬀects of grinding time and grinding load on content of MARK terpenes in extract from fennel obtained by supercritical ﬂuid extraction ⁎ Tahmasb Hatamia,b, , Júlio Cezar Flores Johnera, M. Angela A. Meirelesa a LASEFI/DEA/FEA (School of Food Engineering), UNICAMP (University of Campinas), Campinas, Brazil b Department of Chemical Engineering, Faculty of Engineering, University of Kurdistan, 66177 Sanandaj, Iran

ARTICLE INFO ABSTRACT

Keywords: This paper investigated effects of grinding time (GT) and mass of raw material in mill (ms) on both global SFE Anethole yield from Fennel and its main volatile oil content namely Anethol and Fenchone. For this purpose, extractor of Fenchone SFE equipment was filled with milled Fennel obtained at various values of ms, from 15 g to 35 g, and GT, from Fennel seeds 15 s to 20 min. Extractor was subjected to pressure of 200 bar, temperature of 313 K, and supercritical CO2 flow Grinding effect − rate of 1.67 × 10 4 kg/s for 10 min. Then, extract composition was evaluated by Gas Chromatography (GC) Supercritical fluid extraction analysis. Moreover, for better understanding how GT and ms affect SFE yield and extract composition, their effects were initially investigated on temperature raising and diameter of Fennel seeds during grinding process.

It was found that ms and GT have considerable eﬀects on Anethole and Fenchone content of Fennel extract.

1. Introduction efficiency was 5.2 larger than that of hydrodistillation (Moura et al., 2005). In another relevant paper, Pereira and Meireles (2007) in- Fennel (Foeniculum vulgare) is a plant belonging to Apiaceae fa- vestigated economic analysis for obtaining extracts and essential oils mily, which is cultivated in several countries like Brazil, England, from Rosemary, Fennel, and Anise using SFE and steam distillation. Germany, China and so on (Brender et al., 1997). Fennel extracts can be They concluded that in spite of high cost of SFE equipment, it was more employed as antispasmodic, anti-inflammatory, expectorant, diuretic, economical than steam distillation due to higher quality of its product, and laxative (Jahromi et al., 2003). It also has applications in treatment higher extraction yield, and lower energy consumption (Pereira and of nervous disturbances, carminative, analgesic, and stimulant of gas- Meireles, 2007). Another paper in this topic was written by Hammouda trointestinal mobility (Jahromi et al., 2003). Anethole and Fenchone et al. (2014), whom studied extraction yield and extract quality from are two of the main ingredients in Fennel volatile oil that compose, Fennel using SFE, microwave-assisted extraction (MAE), and hydro- respectively, 40–70% and 1–20% of it (Bernath et al., 1996; Coşge distillation. Their study showed that MAE gave higher overall yield and et al., 2008; Raghavan, 2006). Extraction of volatile oil from Fennel is higher percentage of Fenchone than others. SFE, on the other side, gave generally carried out by several techniques such as SFE, liquid CO2 the maximum percentage of Anethol (Hammouda et al., 2014). In an- extraction, Soxhlet extraction, accelerated solvent extraction, and other interesting paper, Johner and Meireles (2016) constructed a la- steam distillation (Baby and Ranganathan, 2016; Rodríguez-Solana boratory SFE unit that contained one extractor and two separators. et al., 2014a,b; Bodsgard et al., 2016; Johner and Meireles, 2016). From They validated the equipment using SFE from Annatto and Fennel. them, SFE is usually a preferable technique due to its advantages such Fennel extract was obtained by employing 200 bar pressure and 313 K as high extraction performance, simple separation of solvent from ex- temperature in extractor, 80 bar and 312 K in first separator, 20 bar and tract, and its selectivity capability (Durante et al., 2014; Ding et al., 279 K in second separator, mass flow rate of 12 g/min, and solvent per 2017). SFE from Fennel have been reported in various research works. feed ratio of 10. Their results indicated an overall SFE yield of 2.8 g Moura et al. (2005) studied effect of harvesting season, degree of ma- extract/100 g of ground seeds. Moreover, they observed that 97.5% of turation, bed geometry, operating temperature and pressure on SFE the whole extract was collected in the first separator (Johner and yield from fennel. They also studied extraction of Fennel by hydro- Meireles, 2016). Reverchon et al. (1999) carried out successfully ex- distillation and low-pressure solvent extraction, and compared their traction of fennel seeds in two steps including SFE at 90 bar and 50 °C, results with SFE. They found that not only SFE process produced larger and SFE at 200 bar and 40 °C using three CO2 mass flow rates of 0.5, relative proportion of anethole and fenchone, but also its overall 1.0, and 1.5 kg/h. They also modeled the process based on mass

⁎ Corresponding author at: LASEFI/DEA/FEA (School of Food Engineering), UNICAMP (University of Campinas), Campinas, Brazil. E-mail address: [email protected] (T. Hatami). http://dx.doi.org/10.1016/j.indcrop.2017.08.010 Received 16 April 2017; Received in revised form 21 July 2017; Accepted 7 August 2017 Available online 18 August 2017 0926-6690/ © 2017 Elsevier B.V. All rights reserved. 33

T. Hatami et al. Industrial Crops & Products 109 (2017) 85–91 conservation law and demonstrated the reliability of the model by comparing it with experimental data. Piras et al. (2014) employed SFE and hydrodistillation (HD) on Sardinian wild fennel and measured composition of volatile extracts using gas chromatography −mass spectrometry. They found that percentage of the main components in SFE extract were 7.1% (fenchone), 34.9% (estragole), and 24.6% ((E)- anethole), while corresponding number for HD extract were 8.8%, 42.6%, and 43.4%, respectively. Rodríguez-Solana et al. (2014b) performed Oleoresin extraction from fennel seeds using SFE and optimized the process for the aim of attaining maximum estragole per kg of dry plant. They found the optimal operating condition as the pressure of 24 MPa, temperature of 333.15 K, extraction duration of 3.41 h, and methanol percentage of 3%. In this condition, they obtained 1320 ± 260 mg of estragole per kg dry plant. Ivanovic et al. (2014) increased the yield of SFE (increase by up to 1350%) for several plant materials by employing different mechanical pretreatment such as flaking, impact plus shearing, and cutting plus grinding. They found Flaking as the most convenient method for increasing the yield of SFE from fennel seeds. According to the above literature review, two of the attractive research areas in the field of SFE from Fennel are either maximizing overall yield or maximizing Anethole and Fenchone content of the volatile oil. The first necessary step for these goals is, however, grinding the Fennel seeds to increase their specific area. Although the effect of GT was already investigated on the SFE yield of other raw materials, Fig. 1. Schematic diagram of the Mill. they limited themselves to small GT. For example, Wilkinson et al. ff (2014) investigated the e ect of GT on the SFE yield from soybeans at 2.2. Then, in subsection 2.3, the SFE equipment as well as the experi- 48.3 MPa and 80 °C. They placed 60 g of soybeans in a mechanical ments procedure is discussed. Finally, a short description about GC ff grinder for di erent GT from 10 to 60 s. They concluded that increasing analysis is given in subsection 2.4. the GT enhances the extraction yield due to the reduction of particle sizes. The SFE yield was increased from around 6% to 22% by decreasing the particle diameters from 0.20 mm to 0.07 mm. In another 2.1. Sample preparation study by Yahya et al. (2010), SFE yield from Pandan leaf increased up to 50% after a pretreatment grinding for 30 s. Grosso et al. (2008) Dried Fennel seeds were supplied from a municipal market, called “ ” evaluated the effects of mean particle size (0.4, 0.6 and 0.8 mm) to- Temperos Brasil , in Campinas, São Paulo, Brazil, and they were kept in a domestic freezer at 255 K. Prior to each grinding experiment, the gether with the effects of pressure, temperature, and CO2 flow rate on the yield and composition of SFE extract from Italian coriander seeds. raw material were taken out from the freezer to stabilize at the la- They found that a decrease in particle size didn’t have a considerable boratory temperature, 297 K. effect on the SFE volatiles composition, but it increased the SFE yield as more ducts were destroyed with longer milling time. Taking into ac- 2.2. Grinding raw material count the yield and composition of the extract, they found the best operating conditions to be the pressure of 90 bar, temperature of 40 °C, The Fennel seeds were ground in a mill (Marconi, model: MA 340,

CO2 flow rate of 1.10 kg/h, and mean particle size of 0.6 mm. Shrigod São Paulo, Brazil), which is shown in Fig. 1. Its main parts are a rotor, a et al. (2017) carried out SFE of mint leaves, and they investigated the crushing chamber, a collector container, and stainless steel knives effects of temperature (35–55 8C), pressure (100–300 bar), extraction (cutting edge). It has weight of 58 kg, dimensions of 27 × 48 × 50 cm, time (20–90 min), and particle size (0.2–1.0 mm) on both SFE yield and electrical power of 1600 W, and a fixed speed rotor of 1750 rpm. carvone content in volatile oil. They found that SFE yield was mainly An initial study was performed by using the Mill to determine the influenced by particle size followed by pressure, temperature, and ex- influence of ms and GT on physical properties of Fennel. For this pur- traction time, however, carvone content in extract was mostly affected pose, five levels of GT and three levels of ms were considered. These by pressure followed by particle size and extraction time. The effect of levels are 15 s, 2 min, 6 min, 12 min, and 20 min for GT, and 15 g, 25 g, particle size in SFE process was also studied by Sodeifian et al. (2016) and 35 g for ms. Each grinding process was performed two times to and Chougle et al. (2016). Nevertheless, as the effect of milling process ensure that the obtained results are reliable. At the end of each run, on yield and quality of SFE from Fennel were not discussed in literature, Fennel temperatures were measured from the top of the crushing the current study aimed to focus on this area for a board range of GT chamber by using an infrared thermometer with an uncertainty of and various ms. Using higher GT not only affects the particle sizes, but it 0.1 K. The idea of measuring the Fennel temperature in the central parts also heats up the raw materials and probably affects volatile compo- of the crushing chamber was not applicable as opening this chamber, nents. for this purpose, took several seconds and, during this period, temperature decreased quickly. In the next step of the experiment, the 2. Material and methods milled Fennel was subjected to a vibratory agitator and sieves (Bertel, model MAGNETICO, Sao Paulo, Brazil) to determine the particle size This section is organized as follows. First in subsection 2.1, the raw distribution. In this step, six sieves with mesh sizes of 18, 25, 35, 50, 80, material and its pretreatment process are presented. In order to con- and 100 were employed, and they were vibrated for 15 min to ensure ceptual understanding how ms and GT affects the SFE yield and extract that the material in each sieve didn’t change with time. After that, composition, their effects should be firstly investigated on the tem- materials on each sieve and the bottom pan were accurately weighed perature raising and diameter of Fennel seeds during grinding process. and recorded. For calculating the mean particle diameter, the Standard The method for doing this investigation is fully explained in subsection ASAE method was used (Standard, 2003).

86 34

T. Hatami et al. Industrial Crops & Products 109 (2017) 85–91

Fig. 2. Schematic diagram of the SFE unit from (a) top view and (b) front view (Johner and Meireles, 2016).

2.3. SFE from fennel partials of volatile oil evaporated during agitation. In fact, the Fennel seeds for SFE were first ground in the mill, and then directly put in the A schematic diagram of the employed SFE unit is shown in Fig. 2, extractor as we had already measured their size distribution separately. which has total dimensions of 57 × 77 × 115 cm. It has a 100 mL ex- This amount of milled Fennel occupies small part of the extractor, and tractor with an internal height per diameter of 19, and two separators the remaining space is filled with small glass beads. Next, the pump is each with 90 mL volume. Since extraction from Fennel provides low turned on to pressurize the system up to the desired pressure, 200 bar. amount of essential oil (Johner and Meireles, 2016), the mass of oil that It is required to keep the system in this situation for 10 min before remains in the separators, even after recovering the extract oil in the opening the exit valve of the extractor. This 10 min is called static time bed, separators, and lines, is significant in comparison with the total and it is employed to ensure that the supercritical fluid reach equili- mass of extract, and it consequently affects the extraction performance. brium with solid phase. The overall SFE yield is evaluated by collecting To overcome this problem, the separators of the SFE unit in the current and weighing the extract samples at the end of each run. To determine study were substituted by two glass flasks at atmospheric pressure. The the effect of GT and ms on the quality and efficiency of SFE, it is ne- first flask was a separator, but the second flask with cotton inside was cessary to perform the experiment with milled Fennel obtained at used for cleaning the CO2 flow before entering the totalizer. Char- various values of GT and ms. Each SFE run is performed two times to acteristics of other parts of the SFE equipment such as pump, cooling ensure that the obtained results are reliable. and heating bath, flow totalize, temperature and pressure indicators, and valves can be found in our previous paper (Johner and Meireles, 2.4. Chromatographic analysis 2016). In a typical SFE experiment, the thermostatic bath is turned on one The compositions of Anethol and Fenchone in the extract were de- hour before doing the extraction in order to ensure that the system termined by using a GC with flame ionization (GC-FID) (SHIMADZU, reaches the desired temperature, 313 K. The extractor is then charged CG 17-A, Kyoto, Japan) equipped with a fused-silica capillary column with 10 g of milled Fennel. Noticeably, the subjected milled Fennel to ZB-5 (length of 30 m, inner diameter of 0.25 mm, and film thickness of the vibratory agitator was not used for SFE as it was possible that 0.25 μm, Zebron, USA). For this purpose, the analyzable solution was

87 35

T. Hatami et al. Industrial Crops & Products 109 (2017) 85–91

m = 15g s ms = 15g 0.48 m = 25g 30 s ms = 25g 0.46 m = 35g s ms = 35g 25 0.44

0.42 20

0.4 15 0.38 10 0.36

0.34 5

0.32 0 5 10 15 20 0 0 5 10 15 20

Fig. 3. The average size of Fennel particles at various values of GT and ms. Fig. 4. The raising temperature of Fennel at the end of grinding process. obtained by dissolving 10 mg of extract in 10 mL of ethyl acetate. The carrier gas in GC was helium (99.9% purity, White Martins, Campinas, 15 s and 20 min and ms of 15 g and 35 g. These pictures clearly indicate Brazil) at a total flow of 26 mL/min and column flow rate of 1.11 mL/ that as grinding continues, the raw material are moved from the round min. Split injection conducted with an injection volume of 1 μL and body of the mill to the cutting edges. It also illustrates that higher split ratio of 20. The temperature of injector was 493 K, the tempera- percentage of material for the case of ms = 15 g and GT = 20 min lo- ture of detector was 513 K, and the initial temperature of column was cated close to the cutting edges when it compare with that of ms =35g 333 K. Based on a defined program, the column was heated from 333 K and GT = 20 min. This concludes that the reported data in Fig. 4 are to 513 K at 3 K/min, and then held for 2 min. more reliable for ms = 35 g and GT = 20 min than ms = 15 g and the In order to measure the amount of Anethol and Fenchone in the same GT, and the real temperature raising of Fennel for ms = 15 g was extract, it was first required to obtain the calibration curve for both of much higher than that of Fig. 4. Probably, the average temperature them. The analyzable solutions, for this purpose, were obtained by raising of material for the case of ms = 15 g and GT = 20 min was dissolving various masses of Anethol (purity: 99%, SIGMA-ALDRICH) higher than that of ms = 35 g and GT = 20 min. and Fenchone (purity ≥98%, SIGMA-ALDRICH) in 10 mL of ethyl After grinding the raw material, SFE from Fennel was performed acetate, and analyze them with GC. using the home-made SFE equipment (Johner and Meireles, 2016). From the whole ms-GT sets in the grinding experiments, three levels of GT (15 s, 6 min, and 20 min) and two levels of m (15 g and 35 g) were 3. Results and discussion s chosen for SFE experiments. The reason for this selection was that, as shown in Figs. 3 and 4, employing these sets of m -GT produced milled The average size of Fennel particles with respect to GT and m is s s Fennel with clearly different particles size and different temperature presented in Fig. 3. In this figure, three distinct regions can be identi- raising, and thus it would be possible that they gave extract with dif- fied. In region l, which is from 15 s to 2 min, the size of particles ferent volatile oil compositions. Each SFE experiment was performed dropped dramatically with increasing GT, while for region 2, which is two times to guarantee the reliability of the results. from 2 min to 12 min, the size of particles decreased moderately. The Overall extraction yields for m = 15 g and 35 g are presented in size of particles for the last part of the curves, region 3, is approximately s Fig. 6 for GT from 15 s to 20 min. According to this figure, increasing constant. As depicted in Fig. 3, this trend is rather similar for three GT, for m = 35 g, leaded to increasing the overall extraction yield. By material load, 15 g, 25 g, and 35 g, examined in this study. For the case s contrast, increasing the GT, for m =15g,firstly, raised and then re- of 15 g Fennel, the average diameter of particles decreased from s duced the extraction yield. The reason for this phenomenon was that 0.43 mm at 15 s to 0.35 mm at 20 min. For the same GT period, the increasing GT affected the extraction yield in two opposite ways. On average particles diameter of 35 g Fennel was dropped by 0.11 mm and one hand, the particles diameter decreased according to Fig. 3, which it, it leveled off at 0.37 mm. The curve of 25 g Fennel located between the in turn, had a positive effect on extraction yield due to improving the two aforementioned curves. specific area of particles in the SFE bed. On the other hand, the material Fig. 4 shows the raising temperature of Fennel at the end of grinding temperature in the mill increased according to Fig. 4, and this ac- process for various values of GT and m . It is clear from this figure that, s celerated the evaporating of volatile compound that consequently for each value of m , the temperature increased significantly as the s tended to reduce the SFE yield. Temperature raising, at small to mod- grinding process proceeds. What’s more, the temperature raising is also erate values of GT, was not too much, and the particle diameter was the affected considerably by the value of m , and it changed from around s key factor in SFE yield. In this range of GT, as particles diameter for 288 K for m = 15 g to just under 302 K for m = 35 g. In fact, higher s s m = 15 g was smaller than that of m = 35 g, the SFE yield for amount of material inside the mill causes higher friction among them s s m = 15 g was higher. At higher values of GT, the situation was com- or, in the other words, higher conversion of kinematic energy of par- s pletely different. Despite the small particles size produced in higher GT, ticles to internal energy that it consequently leads to more temperature the extraction yield did not necessarily increase due to the significant raising. It is worthwhile emphasizing once more that the reported data temperature raising and evaporation of volatile compound. For the case in Fig. 4 are not the average temperature raising of Fennel inside the of m = 15 g in higher GT, the temperature raising was the prominent mill, but in fact, they are the temperature raising of material located s factor due to the fact that higher percentage of material located close to near the top of the crushing chamber. Evidently, there was a gradient the cutting edges (Fig. 5), and it, as mentioned, increased the Fennel temperature inside the mill so that the material close to the cutting temperature even much higher than that of shown in Fig. 4. For the case edges was often hotter than the other parts. Thus, the real temperatures of m = 35 g in higher GT, the particle diameter was still the prominent raising of material are higher than that of reported in Fig. 4. s factor, and so the extraction yield continued its ascending trend, Material distribution inside the mill is depicted in Fig. 5 for GT of

88 36

T. Hatami et al. Industrial Crops & Products 109 (2017) 85–91

Fig. 5. Fennel distributions inside the crushing chamber of mill.

6 50

5 45 m = 15g s ms = 15g 4 m = 35g s ms = 35g 40 3

35 2

30 1 0 5 10 15 20 0 5 10 15 20

Fig. 6. Effect of GT and m on the overall extraction yield obtained by SFE from Fennel at s Fig. 7. Effect of GT and m on the Anethol extraction yield obtained by SFE from Fennel −4 s 200 bar, 313 K, and 1.67 × 10 kg/s during 10 min. −4 at 200 bar, 313 K, and 1.67 × 10 kg/s, during 10 min. however, with lower slope due to the evaporation of volatile com- Fennel. It can be also interpreted from this figure that the effect of in- pound. The highest extraction yield of 51.14 and 49.34 mg of extract/g creasing ms from 15 g to 35 g on the Anethol extraction yield at small, of Fennel were achieved by grinding, respectively, 15 g of raw material moderate, and high values of GT is, respectively, negative, insignificant, for 6 min and 35 g of raw material for 20 min. and positive. Based on Fig. 8, the effect of ms in Fenchone yield is al- Based on GC analysis of extract and using the calibration curves, the most negligible at small and high values of GT, while its effect for the ff e ects of GT and ms on the Anethol and Fenchone extraction yield are moderate value of GT (6 min) is considerable, which gives the yield of depicted, respectively, in Figs. 7 and 8. It should be noted that there are 0.29 mg Fenchone/g Fennel for ms = 15 g and 0.2 mg Fenchone/g error bars on these two figures, but they are so small that are not seen. Fennel for ms = 35 g. The difference between trends of Anethol and Referring to Fig. 7, the Anethol extraction yield increased with in- Fenchone extraction yield is mainly due to the differences between creasing GT, initially, and then decreased with further increasing GT their Enthalpy of vaporization, which is 61.90 kJ/mol for Anethol after reaching the maximum yield of around 6.19 mg of Anethol/g of (https://www.chemeo.com/cid/44-213-6/Anethole, 2016) and

89 37

T. Hatami et al. Industrial Crops & Products 109 (2017) 85–91

ﬀ m = 15g e ect on the overall extraction yield. For the case of ms = 15 g, how- 0.3 s ever, the extraction yield increased initially and then reduced with GT. m = 35g s In fact, the signiﬁcant temperature raising at higher GT, especially for

ms = 15 g, accelerated the evaporation of volatile compound in the mill, and consequently decreased the SFE performance. Although ex- 0.25 traction yield for both Anethol and Fenchone increased with increasing GT, at ﬁrst, and then decreases with further increasing GT, their curves of extraction yield were not similar due to the diﬀerences between their

0.2 Enthalpy of vaporization. This work evidenced that the eﬀect of ms on Fenchone yield was almost negligible at small and high values of GT, while its eﬀect for the moderate value of GT (6 min) is considerable.

Anethol extraction yield changed signiﬁcantly with ms at low and 0.15 especially high values of GT, but its yield at GT = 6 min was the same

for both ms = 15 g and 35 g. It was also found that among the whole

0 5 10 15 20 values of GT and ms employed in this study, GT = 6 min and ms =15g gave the highest content of Anethole and Fenchone, which are 6.16 mg of Anethol/g of Fennel and 0.29 mg of Fenchone/g of Fennel, respec- Fig. 8. Effect of GT and ms on the Fenchone extraction yield obtained by SFE from Fennel − at 200 bar, 313 K, and 1.67 × 10 4 kg/s, during 10 min. tively. It is worthwhile mention that this work just aimed to investigate the effect of ms and GT, but it didn’t aim to determine the optimum values of these variables. Thus, the aforementioned optimum values are 39.49 kJ/mol for Fenchone (https://www.chemeo.com/cid/70-160-6/ just the best values among the studied values of GT and ms, and more Fenchone, 2016) at standard conditions. As a result, Fenchone is more experiments are still required to find the real optimum point. sensitive to temperature raising than Anethol, and it is the reason why the small difference between Fennel temperature of m = 15 g and s Acknowledgment ms = 35 g at GT = 6 min (Fig. 4) leads to the big difference between their corresponding values of Fenchone yield in Fig. 8. By contrast, this The authors are grateful to the Ministry of Science Research and temperature difference at GT = 6 min doesn’t have any effect on the Technology, Iran for financial support. Anethol extraction yield in Fig. 7. Increasing GT from 6 min to 20 min leads to a dramatic drop for m = 15 g in both figures, while it has just s References a little negative effect for the case of ms = 35 g. This substantial drop for the case of m = 15 g is probably due to the accumulating high s Baby, K.C., Ranganathan, T.V., 2016. Effect of enzyme pre-treatment on extraction yield percentage of material near to the cutting edges (Fig. 5), and it, as and quality of fennel (Foeniculum vulgare) volatile oil. Biocatal. Agric. Biotechnol. 8, mentioned, increases the Fennel temperature even much higher than 248–256. that of shown in Fig. 4, and it consequently accelerate the evaporating Bernath, J., Nemeth, E., Kattaa, A., Hethelyi, E., 1996. Morphological and chemical evaluation of fennel (Foeniculum vulgare Mill.) populations of different origin. J. of volatile compound. These results underline the importance of em- Essent. Oil Res. 8, 247–253. ploying an appropriate GT and ms in order to avoid missing volatile Bodsgard, B.R., Lien, N.R., Waulters, Q.T., 2016. Liquid CO2 extraction and NMR char- compounds of extract. Comparing Fig. 6 with Figs. 7 and 8 shows that acterization of anethole from fennel seed: a general chemistry laboratory. J. Chem. – fl Educ. 93, 397 400. the overall extraction yield is mostly in uenced by particle size, while Brender, T., Gruenwald, J., Jaenicke, C., 1997. Herbal Remedies, Phytopharm Consulting volatile oil composition is depend on both particle size and the tem- Institute for Phytopharmaceuticals, second ed. Schaper & Brümmer GmbH & Co., perature raising in the miller. Salzgitter, Berlin, Germany. Chougle, J.A., Bankar, S.B., Chavan, P.V., Patravale, V.B., Singhal, R.S., 2016. To show the impact of the milling conditions on extraction yield Supercritical carbon dioxide extraction of astaxanthin from Paracoccus NBRC with respect to the influence of other typical SFE parameters like 101723: Mathematical modelling study. Sep. Sci. Technol. 51 (13), 2164–2173. temperature and pressure, it is interesting to compare the obtained Coşge, B., Kiralan, M., Gürbüz, B., 2008. Characteristics of fatty acids and essential oil results in this study with those obtained by Moura et al. (2005). Thy from sweet fennel (Foeniculum vulgare Mill. var. dulce) and bitter fennel fruits (F. vulgare Mill. var. vulgare) growing in Turkey. Nat. Prod. Res. 22, 1011–1016. found that for SFE from a specific type of dried Fennel seeds (Fv_3), Ding, X., Liu, Q., Hou, X., Fang, T., 2017. Supercritical fluid extraction of metal chelate: a which was planted on April 2001 and harvested in September 2001, review. Crit. Rev. Anal. Chem. 47 (2), 99–118. increasing pressure from 100 to 250 bar at 30 °C leaded to increasing Durante, M., Lenucci, M.S., Mita, G., 2014. Supercritical carbon dioxide extraction of carotenoids from pumpkin (cucurbita spp.): A review. Int. J. Mol. Sci. 15, 6725–6740. Fenchone and Anethole content of extract by respectively, 85% and Grosso, C., Ferraro, V., Figueiredo, A.C., Barroso, J.G., Coelho, J.A., Palavra, A.M., 2008. 50%. Moreover, increasing temperature from 30 to 40 °C at constant Supercritical carbon dioxide extraction of volatile oil from Italian coriander seeds. – pressure of 100 bar raised the Fenchone content by 120%, but dropped Food Chem. 111 (1), 197 203. Hammouda, F.M., Saleh, M.A., Abdel-Azim, N.S., Shams, K.A., Ismail, S.I., Shahat, A.A., the Anethole content by 17.5%. These numbers are very comparable Saleh, I.A., 2014. Evaluation of the essential oil of foeniculum vulgare mill (fennel) with the impact of GT in the current study. According to Figs. 7 and 8, fruits extracted by three different extraction methods by Gc/Ms. Afr. J. Tradit. Complement. Altern. Med. 11, 277–279. https://www.chemeo.com/cid/44-213-6/ increasing GT from 15 s to 6 min for the case of ms = 15 g increases the Anethole (Accessed 15 November, 2016). https://www.chemeo.com/cid/70-160-6/ Anethole and Fenchone yield by approximately 40% and 110%, re- Fenchone (Accessed 15 November, 2016). spectively. Although dried Fennel was used in both this paper and a Ivanovic, J., Meyer, F., Stamenic, M., Jaeger, P., Zizovic, I., Eggers, R., 2014. part of paper by Moura et al. (2005), smaller Anethol and Fenchone Pretreatment of natural materials used for supercritical fluid extraction of commercial phytopharmaceuticals. Chem. Eng. Technol. 37, 1606–1611. yield was obtained in the current study mainly due to the smaller SFE Jahromi, B.N., Tartifizadeh, A., Khabnadideh, S., 2003. Comparison of fennel and me- time, 10 min compared to 3 and 4 h, as well as different planting lo- fenamic acid for the treatment of primary dysmenorrhea. Int. J. Gynaecol. Obstet. 80, cation and harvesting time of the employed Fennel. 153–157. Johner, J.C.F., Meireles, M.A.A., 2016. Construction of a supercritical fluid extraction (SFE) equipment: validation using annatto and fennel and extract analysis by thin 4. Conclusion layer chromatography coupled to image. Food Sci. Technol. (Campinas) 36 (2), 210–247. Moura, L.S., Carvalho Jr, R.N., Stefanini, M.B., Ming, L.C., Meireles, M.A.A., 2005. This work studied the compositions of SFE extract from Fennel in Supercritical fluid extraction from fennel (Foeniculum vulgare): global yield, com- terms of the main volatile oil components, namely Anethol and position and kinetic data. J. Supercrit. Fluids 35, 212–219. Pereira, C.G., Meireles, M.A.A., 2007. Economic analysis of rosemary, fennel and anise Fenchone, by taking into account GT and ms as independent variables. essential oils obtained by supercritical fluid extraction. Flavour Frag. J. 22, 407–413. This study proves that increasing GT, for ms = 35 g, has a positive

90 38

T. Hatami et al. Industrial Crops & Products 109 (2017) 85–91

Piras, A., Falconieri, D., Porcedda, S., Marongiu, B., Gonçalves, M.J., Cavaleiro, C., Shrigod, N.M., Swami Hulle, N.R., Prasad, R.V., 2017. Supercritical ﬂuid extraction of

Salgueiro, L., 2014. Supercritical CO2 extraction of volatile oils from Sardinian essential oil from mint leaves (mentha spicata): Process optimization and its quality Foeniculum vulgare ssp. vulgare (Apiaceae): chemical composition and biological evaluation. J. Food Process Eng. 40 (3), 1–9. activity. Nat. Prod. Res. 28, 1819–1825. Sodeifian, G., Ardestani, N.S., Sajadian, S.A., Ghorbandoost, S., 2016. Application of Raghavan, S., 2006. Handbook of Spices, Seasonings, and Flavorings. CRC Press. supercritical carbon dioxide to extract essential oil from Cleome coluteoides Boiss: Reverchon, E., Daghero, J., Marrone, C., Mattea, M., Poletto, M., 1999. Supercritical experimental, response surface and grey wolf optimization methodology. J. Supercrit. fractional extraction of fennel seed oil and essential oil: experiments and mathema- Fluids 114, 55–63. tical modeling. Ind. Eng. Chem. Res. 38, 3069–3075. Standard, A.S.A.E., 2003. Method of Determining and Expressing Particle Size of Chopped Rodríguez-Solana, R., Salgado, J.M., Domínguez, J.M., Cortés-Diéguez, S., 2014a. Forage Materials by Screening. Standard. Characterization of fennel extracts and quantification of estragole: optimization and Wilkinson, N., Hilton, R., Hendry, D., Venkitasamy, C., Jacoby, W., 2014. Study of process comparison of accelerated solvent extraction and soxhlet techniques. Ind. Crops Prod. variables in supercritical carbon dioxide extraction of soybeans. Food Sci. Technol. 52, 528–536. Int. 20 (1), 63–70. Rodríguez-Solana, R., Salgado, J.M., Domínguez, J.M., Cortés-Diéguez, S., 2014b. Yahya, F., Lu, T., Santos, R.C.D., Fryer, P.J., Bakalis, S., 2010. Supercritical carbon di- Estragole quantity optimization from fennel seeds by supercritical fluid extraction oxide and solvent extraction of 2-acetyl-1-pyrroline from Pandan leaf: the effect of (carbon dioxide–methanol) using a Box–Behnken design. Characterization of fennel pre-treatment. J. Supercrit. Fluids 55, 200–207. extracts. Ind. Crops Prod. 60, 186–192.

91 39

CHAPTER 4

MATHEMATICAL MODELING OF SUPERCRITICAL FLUID EXTRACTION ______40

A STEP-BY-STEP FINITE ELEMENT METHOD FOR SOLVING THE

EXTERNAL MASS TRANSFER CONTROL MODEL OF THE SUPERCRITICAL

FLUID EXTRACTION PROCESS FROM FENNEL SEEDS

Tahmasb Hatami, Júlio Cezar Flores Johner, Maria Angela de Almeida Meireles

LASEFI/DEA/FEA (School of Food Engineering), UNICAMP (University of Campinas), Campinas, SP, Brazil

This manuscript will be submitted to Journal of CO2 Utilization, 2018.

A step-by-step finite element method for solving the external mass transfer control model

of the supercritical fluid extraction process from fennel seeds

Tahmasb Hatami*, Júlio Cezar Flores Johner, M. Angela A. Meireles

LASEFI/DEA/FEA (School of Food Engineering), UNICAMP (University of Campinas),

Campinas, Brazil

Abstract

The effect of grinding time (GT) from 15 s to 20 min on the dynamic yield of supercritical fluid extraction (SFE( from fennel was investigated. The extractor was subjected to a pressure of 200 −4 bar, a temperature of 313 K, and a supercritical CO2 flow rate of 2.0×10 kg/s over 80 min. It was found from experimental data that the overall SFE yield was increased significantly from 6.4 g/100 g fennel when GT = 15 s to 9.8 g/100 g fennel when GT = 6 min, while it changed by only 1.5 g/100 g fennel after increasing GT from 6 min to 20 min. The experimental data was then successfully modeled based on the mass conservation law for both fluid and solid phases. Partial differential equations (PDEs) of the model were solved using Galerkin’s method on finite element method (FEM). The main feature of the modeling is that it reports the first complete FEM solution strategy to solve the PDEs of SFE.

Keywords: SFE from fennel; Grinding time; Mathematical modeling; Finite element method

Nomenclature A (m2) Cross section area of extractor C (kg/m3) Solute concentrations in the supercritical fluid phase 3 Cs (kg/m ) Solute concentration in fluid layer sticking to the surface of particles

Dint (m) Internal diameter of extractor 2 Dl (m /s) Axial dispersion coefficient of solute in CO2 2 Dm(m /s) Binary diffusion coefficient of solute in CO2 F (kg) Total mass of raw material inside the bed K Partition coefficient of solute between solid and supercritical fluid

* Corresponding author. Tel.: C55 19 3521 0100; fax: C55 19 3521 4027. E-mail address: [email protected] (Tahmasb Hatami). 42

kf (m/s) External mass transfer coefficient L (m) Height of the particle bed n Number of mesh elements P (Pa) Pressure

Peb Bed Péclet number (Lv/Dl) q (kg/m3) Solute concentrations in the solid phase

Re Reynolds number (ρvdp/μ)

Rp (m) Particle radius

Sc Schmitt number (μ/ρDm)

Sh Sherwood number (kfdp/Dm) T (K) Temperature v (m/s) Interstitial fluid velocity inside the bed x0 (g/g) Initial mass fraction of the extractable solute in the solid phase z Dimensionless axial coordinate along extractor Z (m) Axial coordinate along extractor Greek letter τ Dimensionless time ε Porosity of the bed ω Weighting function μ (kg/m. s) Fluid viscosity ρ (kg/m3) Fluid density

1. Introduction

Fennel (Foeniculum vulgare) is a plant belonging to the Apiaceae family and is cultivated in several countries, including Brazil, England, Germany, and China (Brender et al., 1997). Fennel extracts can be employed in antispasmodic, anti-inflammatory, expectorant, diuretic, and laxative applications (Jahromi et al., 2003). Fennel also is used in the treatment of nervous disturbances, is used as a gastrointestinal mobility stimulant, and is used for carminative and analgesic purposes (Jahromi et al., 2003). One of the most promising methods for the extraction of oil from fennel is SFE. Supercritical fluids (SCF) have proved to be effective solvents for extraction from 43

solids and are widely used by many researchers (Mackėla et al., 2017; Conde-Hernández et al., 2017) due to their high mass transfer rate as well as the simple separation of solvent from extract at the end of the extraction process (Gallo et al., 2017; Antunes-Ricardo et al., 2017). SFE from fennel has been reported in various research works (Johner and Meireles, 2016; Rodríguez- Solana et al., 2014; Hammouda et al., 2014). Most publications on SFE from fennel focused on the effects of temperature and pressure on the SFE performance but rarely placed emphasis on the effect of grinding time (GT). To address this limitation, in our recent publication, we investigated the effects of GT (15 s to 20 min) and grinding load (15 g to 35 g) on global SFE yield from fennel as well as its main volatile oil content, namely, anethole and fenchone (Hatami et al., 2017). The extractor was subjected to a pressure of 200 bar, a temperature of 313 K, and a -4 supercritical CO2 flow rate of 1.67×10 kg/s, with a static time of 10 min and an extraction time of 10 min. It was found that a GT of 6 min and a grinding load of 40.7 g of fennel/lit of mill capacity gave the highest anethole and fenchone contents of 6.16 mg of anethol/g of fennel and 0.29 mg of fenchone/g of fennel, respectively. The short extraction time of 10 min in the paper by Hatami et al. (2017) was chosen to quickly investigate the effect of GT on the SFE yield, though they did not give any insight into the effect of GT on the dynamic extraction yield. Therefore, the first aim of the current study is to fill this gap by investigating the effect of GT on the dynamic extraction yield of SFE from fennel over 80 min. From the whole runs in the previous paper, those with grinding load of 40.7 g of fennel/lit of mill capacity was used in this study as it was proven to have higher SFE yield and better extract quality compared to other grinding loads. It should be noted that the impact of GT on the SFE performance is not solely due to the effect of GT on particle size because GT also changes the temperature of raw materials inside the mill, which consequently affects both the quantity and quality of the extract due to the change of evaporation rate of volatile components (Hatami et al., 2017). For clarification, Fig. 1 presents the impact of GT, from 15 s to 20 min, on the temperature increase and the average size of the fennel particles (Hatami et al., 2017). Although using a higher GT results in a smaller particle diameter that in turn improves the SFE yield, it also enhances the temperature increase of the raw material in the mill, which subsequently increases the evaporation rate of the volatile material and decreases the SFE yield (Hatami et al., 2017). 44

Fig. 1. The effect of GT on the temperature increase (ΔT) and the average size

of fennel particles (dp) (Hatami et al., 2017).

After experimental determination of the SFE yield, it is very important to model the process with the aim of determining the optimum operation conditions (Hatami et al., 2014). Two main SFE models can be found in the literature: 1) empirical models and 2) mass transfer-based models. Although the empirical models have the advantages of simplicity, they are not suitable for scaling-up (del Valle and Fuente, 2006). Mass transfer-based models generally result in two PDEs, one PDE for the fluid phase and the other for the solid phase (del Valle and Fuente, 2006; Oliveira et al., 2011). These kinds of models can be solved using either analytical or numerical techniques. Analytical techniques have been previously explained in the literature in detail (Lucas et al., 2007; Mongkholkhajornsilp et al., 2005), where the authors used further simplification, such as ignoring the axial dispersion effect (Lucas et al., 2007, Mongkholkhajornsilp et al., 2005). Although ignoring the axial dispersion coefficient often does not have any substantial effect on the results (Reverchon and Marrone, 2001), Goto et al. (1996) revealed that in the case of an extraction column with a sharp distribution, the impact of axial 45

dispersion could become considerable, even for a Peclet number of 100. In regards to numerical techniques, SFE models can be solved by either the finite difference method (FDM) or the finite element method (FEM). The fundamentals of the FDM and FEM can be found in a number of sources (LeVeque, 2007; Kwon and Bang, 2000; Reddy, 1993). The FDM has been broadly used in SFE and was explained in detail by Meireles et al. (2009). The FEM is also used in SFE (Madras et al., 1994; Valderrama and Alarcón, 2009), but to the best of our knowledge, no work has been published explaining this method as applied to the SFE model in detail. Such a publication is important not only due to the high volume of calculations but also for potentially decreasing the computation cost in comparison with the FDM. Therefore, the second aim of the present study is to employ the FEM to solve the external mass transfer control model of the SFE process and to explain the solution strategy in detail. The external mass transfer control model was intentionally selected in this study, as it is one of the most applicable models in the SFE process that has been widely used in the literature (Song et al., 2017; Hatami et al., 2012; Oliveira et al., 2011; del Valle and Fuente, 2006). 2. Material and methods

2.1. Sample preparation

The fennel seeds, with a humidity of 8 % (mass), were supplied from the “Temperos Brasil”, in the municipal market of Campinas, São Paulo, Brazil, and they were kept in a domestic freezer at 255 K. Prior to each grinding experiment, the raw material was removed from the freezer and was allowed to stabilize at laboratory temperature, 297 K. The fennel seeds were then ground in a knife mill (Marconi, model: MA 340, São Paulo, Brazil) for 15 s, 6 min, and 20 min to determine the influence of GT on the dynamic extraction yield. The standard ASAE method was then used (Standard, 2003) for calculating the mean particle diameter. In this method, the milled fennel was processed with a vibratory agitator (Bertel, model MAGNETICO, Sao Paulo, Brazil) with six sieves (mesh sizes of 18, 25, 35, 50, 80, and 100). The sieves were vibrated for 15 min to ensure that the mass of material in each sieve did not change over time. The material remaining in each sieve and the bottom pan were then accurately weighed, recorded, and used for calculating the mean particle diameter.

2.2. SFE from fennel

A picture of the employed SFE unit is shown in Fig. 2. The unit includes a 100 mL extractor (internal diameter of 0.02 m, and height of 0.32 m), a 100 ml glass flask as the separator, a filter flask, a flow totalizer, a pump, cooling and heating baths, temperature and pressure indicators, and valves. One of the heating baths is used to maintain a constant extractor temperature, while another heating bath is used to control the temperatures of the remaining equipment.

Fig. 2. SFE unit used in this study.

In a typical SFE experiment, the cooling and heating baths are turned on one hour prior to extraction to ensure that the system reaches the desired temperatures, which are 316 K for the heating baths and -5 °C for the cooling bath. The heating bath temperature was 3 K higher than the extractor temperature due to the heat lost in isolated pipelines. The extractor is then charged with 10 g of milled fennel. This milled fennel occupies a small part of the extractor, and the 47

remaining space is filled with small glass beads, 3.8 mm in diameter. Next, the pump is turned on to pressurize the system up to 200 bar. The system must equilibrize under these conditions for a fixed period of time (20 min in this work) before opening the exit valve of the extractor. This 20 min period is called static time, and its value was selected to be two times the static time of a relevant paper (Moura et al., 2005) to ensure that the supercritical fluid phase reaches equilibrium with the solid phase. After that, the shut-off valve and the micro-metering valve were opened, and the SFE started. The dynamic SFE yield was calculated by collecting and weighing the extract samples at the specified extraction times of 3.5, 10, 15, 22, 30, 55, and 80 min. This experiment was performed two times for each of the GTs of 15 s, 6 min, and 20 min. Table 1 provides the process parameters for these experiments. In this table, GT (s) is grinding time, Q (kg/s) is solvent mass flow rate, dp (m) is fennel particle diameter, L (m) is the height of the particle bed, and ε is bed porosity.

Table 1 Process parameters for SFE from 1×10-2 kg fennel at 313 K and 200 bar in the extractor, with 2×10-2 m internal diameter. Parameter Exp. 1 Exp. 2 Exp. 3 GT (s) 15 360 1200 Q (kg/s) × 104 2.05 1.88 2.03 4 dp (m) × 10 4.30 3.70 3.50 L (m) × 102 4.90 4.40 4.40 ε 0.48 0.43 0.43

3. Theory

3.1. Mathematical model

The concentration profiles in a supercritical extractor can be mathematically described by a set of PDE. From all SFE models in the literature (del Valle and Fuente, 2006; Oliveira et al., 2011), this study aimed to focus on the external mass transfer control model, as it is one of the most applicable models in the SFE process and has been widely used in the literature. The main assumptions for this model are mono-sized spherical particles, constant temperature and pressure, neglected internal mass transfer resistance, flow with axial dispersion, linear equilibrium adsorption isotherm, and unchanged bed porosity (Hatami et al., 2012). Pilavtepe and Yesil-Celiktas (2013) reported that neglecting internal mass transfer resistance is applicable 48

for SFE from particles with smaller diameters. The particle size in their experiments for SFE from P. oceanica was from 0.3 mm to 0.5 mm. Additionally, this assumption was successfully employed for modeling of SFE from St. John’s wort, with an average size of 0.4 mm (Hatami et al., 2012). Therefore, the average particle size of 0.35 mm to 0.43 mm in the current paper (Table 1) justifies the employment of this challenging assumption. According to these hypotheses, the following PDEs can be derived for the fluid and solid phases in the extractor (Reis-Vasco et al., 2000): ∂C 1 ∂2C ∂C 3Lk (1 − ε) (1) f ( ) = 2 − + Cs − C ∂τ Peb ∂z ∂z εvRp

∂q 3Lkf (2) = − (Cs − C) ∂τ vRp where C (kg/m3) and q (kg/m3) are solute concentrations in the supercritical fluid phase and the 3 solid phase, respectively. Cs (kg/m ) is the solute concentration in the fluid layer, sticking to the surface of particles, and is equal to q/K (K is partition coefficient between solid and supercritical fluid). Dimensionless terms Peb, z, and τ are bed Péclet number (Lv/Dl), axial coordinate along 2 extractor (Z/L), and dimensionless time (tv/L), respectively, where Dl (m /s), v (m/s), and Z (m) are the axial dispersion coefficient, the interstitial velocity of fluid inside the extractor, and the axial coordinate, respectively. Rp (m) is the particle radius, ε is the porosity of the particle bed, and kf (m/s) is the external mass transfer coefficient. Boundary and initial conditions for the above PDEs are as follows:

At 휏 = 0 푎푛푑 0 < 푧 < 1 ⟹ 퐶 = 0 , 푞 = 푞 = 푥0퐹 0 퐴퐿(1−휀)

휕퐶 (3) 퐴푡 휏 > 0 푎푛푑 푧 = 0 ⟹ = 푃푒 (퐶 − 0) 휕푧 푏

휕퐶 푎푡 휏 > 0 푎푛푑 푧 = 1 ⟹ = 0 휕푧 where x0 (g/g) is the initial mass fraction of the extractable solute in the solid phase, F (kg) is the total mass of raw material inside the bed, and A (m2) is the cross section area of the extractor 49

휋 2 ( 퐷 ). Dint (m) is the internal diameter of extractor. The boundary conditions in the equation 4 푖푛푡 (3) are the natural boundary conditions.

3.2. Solution strategy using FEM

Weighted residual is a good technique to obtain approximate solutions to the PDEs of the SFE process (Kwon and Bang, 2000). This method for the unsteady state problem can be applied to the spatial domain but not to the temporal domain (Kwon and Bang, 2000). The solution strategy involves two steps. 1) Applying the FEM to the spatial derivatives of PDEs to convert each PDE into a set of ordinary differential equations (ODEs). 2) Approximating time derivatives, often using the FDM, to convert ODEs into a set of algebraic equations. For this purpose, discretization of the problem domain (0

Fig. 3. Finite element mesh of the extractor.

Then, the following functions are considered to approximate the PDE solution: 50

C(푧, τ) = 퐻1푖 (푧)퐶푖(τ) + 퐻2푖 (푧)퐶푖+1(τ) (4) q(푧, τ) = 퐻1푖(푧)푞푖(τ) + 퐻2푖(푧)푞푖+1(τ) (5) where

푧푖+1 − 푧 (6) 퐻1푖 = ℎ푖

푧 − 푧푖 (7) 퐻2푖 = ℎ푖

ℎ푖 = 푧푖+1 − 푧푖 (8)

The main advantage of these trial functions is that their unknown constants (퐶푖, 퐶푖+1, 푞푖, 푞푖+1) are the solute concentrations at the mesh element boundaries. Subsequently, equations (4) and (5) should be substituted into equations (9) and (10) for calculating the residual functions:

∂C 1 ∂2C ∂C 3Lk (1 − ε) (9) f ( ) Residual1 = − 2 + − Cs − C ∂τ Peb ∂z ∂z εvRp

∂q 3Lkf (10) Residual2 = + (Cs − C) ∂τ vRp

As the numerical solutions are not necessarily equal to the exact solution, the above residual functions are not necessarily equal to zero for the whole domain. Thus, a weighting function ω is selected, and the weighted averages of the residuals are set to zero. Thus:

1 (11) ∂C 1 ∂2C ∂C 3Lk (1 − ε) f ( ) ∫ ω ( − 2 + − Cs − C ) dz = 0 ∂τ Peb ∂z ∂z εvRp 0

1 (12) ∂q 3Lkf ∫ ω ( + (Cs − C)) dz = 0 ∂τ vRp 0

The formulation of equation (11), which is called the strong formulation, requires the evaluation of the second derivative of the trial function. However, this term vanishes, as the trial function of equation (4) is a linear function of z. This means that the axial dispersion effect is ignored by the 51

strong formulation, which consequently leads to an inaccurate approximate solution of the PDEs. This problem can be handled by applying integration by parts to the strong formulation, which gives the following equation (weak formulation) (Kwon and Bang, 2000):

1 1 (13) ∂C 1 ∂ω ∂C ∂C 3Lkf(1 − ε) ω ∂C ∫ (ω + + ω − ω (Cs − C)) dz − ( )| = 0 ∂τ Peb ∂z ∂z ∂z εvRp Peb ∂z 0 0

It is clear from equation (13) that the order of differentiability of the trial function reduces to one. The weighting functions can be calculated according to the Galerkin’s method as follows (Kwon and Bang, 2000):

∂C ∂q (14) ω1 = = = 퐻1푖 ∂퐶푖 ∂푞푖

∂C ∂q (15) ω2 = = = 퐻2푖 ∂퐶푖+1 ∂푞푖+1

Equation (12) and (13) can be written in the following manner: n zi+1 (16) ∂C 1 ∂ω ∂C ∂C 3Lkf(1 − ε) ∑ ∫ (ω + + ω − ω (Cs − C)) dz ∂τ Peb ∂z ∂z ∂z εvRp i=1 zi

ω ∂C 1 − ( )| = 0 Peb ∂z 0 n zi+1 (17) ∂q 3Lkf ∑ ∫ (ω + ω (Cs − C)) dz = 0 ∂τ vRp i=1 zi

By combining equations (4), (5), (14), (15), and (16), the terms of the integrals for the ith element can be calculated as follows:

First term of equation (16): 52

zi+1 zi+1 (18) ∂C H 퐶̇ ∫ ω dz = [∫ [ 1i] [H H ] dz] [ 푖 ] = H 1i 2i ∂τ 2i 퐶푖+̇ 1 zi zi zi+1 2 ̇ ̇ H1i H1iH2i 퐶푖 hi 2 1 퐶푖 [∫ [ 2 ] dz] [ ] = [ ] [ ] H2iH1i H2i 퐶푖+̇ 1 6 1 2 퐶푖+̇ 1 zi where 퐶푖̇ is equal to dCi/dτ.

Second term of equation (16):

zi+1 zi+1 dH1i (19) 1 ∂ω ∂C 1 dH dH 퐶 ∫ dz = [∫ [ dz ] [ 1i 2i] dz] [ 푖 ] = Peb ∂z ∂z Peb dH2i dz dz 퐶푖+1 z z i i dz

1 1 zi+1 2 − 2 1 hi hi 퐶푖 1 1 −1 퐶푖 ∫ dz [ ] = [ ] [ ] 1 1 퐶 퐶 Peb 푖+1 Pebhi −1 1 푖+1 zi − 2 2 [ [ hi hi ] ]

Third term of equation (16): zi+1 zi+1 (20) ∂C H dH dH 퐶 ∫ ω dz = [∫ [ 1i] [ 1i 2i] dz] [ 푖 ] = ∂z H2i dz dz 퐶푖+1 zi zi

H1i H1i zi+1 − h h 퐶 1 −1 1 퐶 ∫ i i dz [ 푖 ] = [ ] [ 푖 ] H2i H2i 퐶푖+1 2 −1 1 퐶푖+1 zi − [ [ hi hi ] ]

Fourth term of equation (16): 53

zi+1 (21) 3Lkf(1 − ε) ∫ −ω (Cs − C) dz = εvRp zi

zi+1 3Lkf(1 − ε) H1i 1 푞푖 퐶푖 [− ∫ [ ] [H1i H2i] dz] ( [푞 ] − [ ]) = εvRp H2i K 푖+1 퐶푖+1 zi

zi+1 3Lk (1 − ε) H2 H H 1 푞 퐶 [− f ∫ [ 1i 1i 2i] dz] ( [ 푖 ] − [ 푖 ]) = 2 푞 퐶 εvRp H2iH1i H2i K 푖+1 푖+1 zi

hiLkf(1 − ε) 2 1 푞푖 hiLkf(1 − ε) 2 1 퐶푖 − [ ] [푞 ] + [ ] [ ] 2εvRpK 1 2 푖+1 2εvRp 1 2 퐶푖+1

Fifth term of equation (16):

ω ∂C 1 ω ∂C ω ∂C (22) − ( )| = − ( )| + ( )| = (퐶 − 0) Peb ∂z 0 Peb ∂z z=1 Peb ∂z z=0

By substituting equations (18) to (22) into equation (16), the integral for the ith element can be obtained.

In the other side, the terms of integrals for the ith element of equation (17) can be calculated by combining equations (4), (5), (14), (15), and (17) as follows:

First term of equation (17): zi+1 ri+1 (23) ∂q H1i 푞푖̇ ∫ ω dz = [∫ [ ] [H1i H2i] dz] [ ] = ∂τ H2i 푞푖+̇ 1 zi ri zi+1 2 H1i H1iH2i 푞푖̇ hi 2 1 푞푖̇ [∫ [ 2 ] dz] [ ] = [ ] [ ] H2iH1i H2i 푞푖+̇ 1 6 1 2 푞푖+̇ 1 zi where 푞푖̇ is equal to dqi/dτ.

Second term of equation (17): 54

zi+1 (24) 3Lkf ∫ ω (Cs − C) dz = vRp zi

zi+1 3Lkf H1i 1 푞푖 퐶푖 [ ∫ [ ] [H1i H2i] dz] ( [푞 ] − [ ]) = vRp H2i K 푖+1 퐶푖+1 zi

zi+1 3Lk H2 H H 1 푞 퐶 [ f ∫ [ 1i 1i 2i] dz] ( [ 푖 ] − [ 푖 ]) = 2 푞 퐶 vRp H2iH1i H2i K 푖+1 푖+1 zi hiLkf 2 1 푞푖 hiLkf 2 1 퐶푖 [ ] [푞 ] − [ ] [ ] 2vRpK 1 2 푖+1 2vRp 1 2 퐶푖+1

By substituting equations (23) and (24) into equation (17), the integral for the ith element can be obtained in a matrix form.

After substituting the terms into equation (16) and (17), these equations can be solved numerically using ODE toolbox of MATLAB software. Subsequently, the SFE yield can be obtained by the integration (over dimensionless time) of solute concentration at the extractor outlet as shown below:

휏 퐴퐿휀 퐴퐿휀 (25) 푌𝑖푒푙푑 = ∫ 퐶 (휏′)푑휏′ × 100 = × 100 ∑ Δ휏 퐶 (휏 ) 퐹 푛+1 퐹 푖 푛+1 푖 0 푖 where 퐶푛+1(휏푖) is the solute concentration at the extractor outlet at dimensionless time 휏푖, and

Δ휏푖 is the dimensionless time interval.

3.3. Model parameters

The density and viscosity of SC-CO2 were obtained as functions of temperature and pressure from the NIST Chemistry WebBook (2016). The binary diffusion coefficient of solute in CO2 was obtained using López-Padilla correlation (López-Padilla et al., 2016):

−4 −3 −4 −6 퐷푚 = 2.3239 × 10 + 3.9332 × 10 휇 − 5.6372 × 10 휌 + 1.20 × 10 푇 − 2.71 (26) × 10−7푃 55

2 3 where 퐷푚is binary diffusion coefficient (cm /s), 휇 is viscosity (cP), 휌 is density (g/cm ), T is temperature (oC), and P is pressure (bar). It is worth noting that for proposing the binary diffusion coefficient, López-Padilla et al. (2016) analyzed the literature experimental data for diffusivity of various solutes in supercritical CO2. These authors found that diffusion coefficients of solutes with similar chemical structure, volatility, and molecular weight are approximately equal. Accordingly, they derived three algebraic correlations for calculating the diffusion coefficients of volatile oils, fatty acids and their esters, as well as fixed oils (triglycerides) in supercritical CO2, as a function of temperature and pressure. They revealed that the proposed correlations are in good agreement with the literature data. equation (53) is one of the three equations proposed by López-Padilla et al. (2016), which is applicable for calculating the diffusion coefficient of fatty acids and fatty acid esters in supercritical carbon dioxide. A correlation adapted to fatty acids and fatty acid esters was applied in this case because fennel extract contains less than 15% of combined anethole and fenchone, as determined by GC analysis in our previous publication (Hatami et al., 2017), and the vast majority of the extract was fatty acid.

The axial dispersion coefficient (퐷푙) was calculated using the following correlation (Funazukuri et al., 1998).

휀퐷 (27) 푙 = 1.317(휀푅푒푆푐)1.392 푓표푟 휀푅푒푆푐 > 0.3 퐷푚 where Re is Reynolds number (휌푣푑푝/휇), and Sc is Schmitt number (휇/휌퐷푚). Furthermore, the external mass transfer coefficient was calculated as shown below:

푆ℎ퐷푚 (28) 푘푓 = 푑푝 where dp is the particle diameter, and Sh is the Sherwood number that was calculated using the Wakao and Funazkri correlation (1978):

Sh = 2+1.1(휀푅푒)0.6Sc0.33 (29) 56

This correlation is valid for Re values between 3 and 10000. As Wakao and Funazkri developed their equation based on the superficial Reynolds number, the term porosity was added in equation (56) to update it for the interstitial Reynolds number.

The parameter x0 in the model was calculated using the method proposed by Hatami et al. (2011), who demonstrated that the slope of the SFE yield versus time in the diffusion-controlled extraction region obeys a geometric progression.

The bed porosity was calculated as follows:

휌 휀 = 1 − 푎 (30) 휌푟

where 휌푎 and 휌푟 are apparent bed density and real density of fennel, respectively. The apparent bed density was determined by dividing the mass of fennel in the bed per bed volume. The real density of fennel was measured using a pycnometer.

As there is no empirical correlation for calculating the partition coefficient (K) of fennel extract, it is determined by a trial and error procedure such that the SFE yield obtained by the model best approximates that obtained from the data of Exp. 2 (the obtained value of K was then used for

Exp. 1 and Exp. 3). To this end, K was changed systematically, starting from a small positive quantity, and for each value of K, the mean absolute percentage deviation (MAPD) of the model results from the experimental data was calculated using the following equation:

1 |푌𝑖푒푙푑 − 푌𝑖푒푙푑 | (31) MAPD = ∑ 푒푥푝 푚표푑푒푙 × 100 푁 푌𝑖푒푙푑푒푥푝 where N is the total number of experimental data, and the subscript “exp” and “model” refer to the experimental data and model results, respectively. This procedure was repeated until the minimum value of “MAPD” was obtained. MATLAB software was used to write the required FEM codes for solving the mathematical model.

4. Results and discussion

To compare the model results with the experimental data, the first step is the calculation of the models’ parameters and the only fitted parameter, K. These parameters for SFE from fennel were calculated using the correlations presented in the previous subsection. The numerical values of -9 2 -5 Dm, μ, ρ, ρr, Sc, and K are constant for all experiments and are, 6.06×10 m /s, 7.83×10 Pa.s, 8.40×102 kg/m3, 1.26×103 kg/m3, 1.54×101, and 4.24×101, respectively. Table 2 presents the numerical values of other parameters for each experiment. In addition, the analysis showed that the dimensionless time interval (Δ휏) is one of the most important factors that affects the simulation results. As the forward difference technique is conditionally stable, the numerical value of Δ휏 should be carefully determined. In this study, it was found that a dimensionless time interval of equal to or smaller than 0.03 leads to stable and accurate results.

Table 2 The parameters of the SFE model. Parameter Exp. 1 Exp. 2 Exp. 3 2 6 Dl (m /s) × 10 4.36 3.58 3.70 Re 7.39 6.64 6.80 Sh 7.82 7.06 7.13

Peb 17.99 20.58 21.56 4 kf (m/s) × 10 1.10 1.16 1.23 x0 0.069 0.106 0.116

Fig. 4 compares the model results with the experimental data of SFE from fennel over a solvent per feed mass ratio (S/F) of 100. The MAPD of the model from experimental data of Exp. 1 and Exp. 2 are 21.6% and 15.3%, respectively, so that the SFE yield of the model ends too early. In contrast, the agreement between the model and the experimental data of Exp. 3 is better, with a corresponding MAPD value of 9.0%. These values of MAPD show that better agreement between model and experiments is reached with decreasing particle sizes. This figure also clearly shows, especially for Exp. 2 and Exp. 3, that the model results in the constant rate period are in acceptable agreements with the experimental data, while the deviations in the falling rate period and diffusion rate period are considerable. This outcome can be attributed to the fact that the lumped system assumption in SFE from fennel is more applicable for the constant rate period, as the concentration profile of oil inside the particles remains approximately flat in this period. In 58

contrast, the oil concentration in the falling and diffusion rate periods is strongly correlated with the distance from the center of the particle. Comparison of these findings with those reported by Pilavtepe and Yesil-Celiktas (2013) and other relevant papers shows that the reliability of lump system assumption for the solid phase not only depends on the particle diameter but also depends on the fluid hydrodynamics as well as the type of raw material used. According to the experimental data in this figure, the overall SFE yield for Exp. 1, Exp. 2, and Exp. 3 are 6.4, 9.8, and 11.3 g/100 g fennel, respectively. This result clearly demonstrates that by increasing the GT from 15 s to 6 min, the SFE yield increases sharply due to the reduction in particle diameter. A much smaller increase in the SFE yield was found when the GT was increased from 6 min to 20 min. It is worth noting that in our previous publication, the GT of 6 min gave the highest amount of volatile oil in the extract compared to the GTs of 15 s and 20 min (Hatami et al., 2017). This result is because the small GT of 15 s produces particles with larger average diameters that drop the extraction of volatile oil yield due to the reduction of specific surface area of mass transfer, and the large GT of 20 min leads to a greater temperature increase in the mill that consequently leads to evaporation of the volatile oil. In other words, the overall extraction yield is principally influenced by particle size, while volatile oil composition depends on both particle size and the temperature increase in the mill (Hatami et al., 2017). 59

Model 1 4 Exp. 1 Model 2

Yield (g Yield extract(g / g of 100 fennel) Exp. 2 2 Model 3 Exp. 3

0 0 20 40 60 80 100 120 S/F (kg CO / kg fennel) 2 Fig. 4. Comparison between the model results and the experimental data for SFE from fennel.

In conclusion, it is interesting to compare the FEM results, as outlined in this study, with the results obtained by the FDM. Fig. 5 shows the results of both methods for Exp. 3, with multiple numbers of mesh elements (n). The SFE yield obtained by the FEM in this figure is approximately independent of the number of mesh elements. Actually, increasing the number of meshes in the FEM increases the accuracy for determining the concentration profile along the bed due to employing more nodes but does not have a considerable effect on the concentration of the outlet of the extractor (Cn+1 in Fig. 3), which is used for calculating the SFE yield. However, the FDM results for calculating the SFE yield are strongly influenced by the number of mesh elements because the numerical derivation of the concentration at each node in the FDM solution strategy is obtained using the numerical values of concentration of the neighbor nodes, and this numerical derivation becomes closer to the real derivation if the distances between nodes tends towards zero. As shown in Fig. 5, the FDM results converge to the accurate solution if and only if the appropriate numbers of mesh elements (n=200 in this study) are selected. This outcome means that from all the FDM curves in Fig. 5, only the curve obtained at n=200 is reliable and 60

that the others are inaccurate. Accordingly, the main advantage of the FEM over the FDM in this study is that fewer numbers of mesh elements are required, as the FDM results with values of n greater than or equal to 200 are very close to the FEM results with value of n equal to one element. Employing fewer numbers of mesh elements decreases the computation cost.

6 FDM (n=5) FDM (n=10) FDM (n=20) 4 FDM (n=200) FEM (n=1)

Yield (g Yield extract(g / g of 100 fennel) FEM (n=10) FEM (n=20) 2

0 0 10 20 30 40 50 60 S/F (kg CO / kg fennel) 2 Fig. 5. Effect of the number of mesh elements (n) on the FEM and FDM results for Exp. 3.

As the SFE yield obtained by the FEM in Fig. 5 is approximately independent of the number of mesh elements, it is interesting to further simplify the calculation by considering only one element. Fig. 6 shows the whole bed as one single element together with the corresponding concentrations for the fluid and solid phases. 61

Fig. 6. The extractor with one element.

Considering Fig. 3 and using Eqs. (18) to (24), the Eqs. (16) and (17) can be simplified as follow:

1 2 1 퐶̇ 1 1 −1 1 −1 1 Lkf(1 − ε) 2 1 퐶 (32) [ ] [ 1] + ( [ ] + [ ] + [ ]) [ 1] 퐶 6 1 2 퐶2̇ Peb −1 1 2 −1 1 2εvRp 1 2 2

( ) Lkf 1 − ε 2 1 푞1 −퐶1 − [ ] [푞 ] = [ ] 2εvRpK 1 2 2 0

1 2 1 푞1̇ Lkf 2 1 푞1 Lkf 2 1 퐶1 0 (33) [ ] [ ] + [ ] [푞 ] − [ ] [ ] = [ ] 6 1 2 푞2̇ 2vRpK 1 2 2 2vRp 1 2 퐶2 0

Accordingly, two PDEs of Eqs. (1) and (2) are converted to four ODE of Eqs. (32) and (33). The main advantage of this conversion is that Eqs. (32) and (33) can be solved much easier, and it reduces the computation cost significantly.

5. Conclusion

This paper describes the effect of GT (15 s, 6 min, and 20 min) on the dynamic SFE yield from fennel at 313 K and 200 bar. The process was also modeled mathematically, and a comprehensive step-by-step solution strategy using the FEM was provided. Using this method, the PDEs were converted into a set of algebraic equations so that they could be easily solved 62

using simple computer codes. The results confirm that the FEM is very effective for solving the external mass transfer control model for the SFE process, and the MADP for the model results compared to the experimental data was 21.6% for Exp. 1, 15.3% for Exp. 2, and 9.0% for Exp. 3. Comparing the SFE yield obtained by the FEM and FDM solutions shows that the FEM accuracy is approximately independent of the mesh number; thus, using one single element is adequate to accurately solve the model using FEM. However, the FDM accuracy is strongly dependent on the mesh number; thus, the mesh number value chosen should be high enough to ensure accurate results (the number of mesh elements must be greater than or equal to 200). This study revealed that FEM could simplify the solution strategy of the external mass transfer control model of SFE by converting two PDEs of the model into four ODEs.

Acknowledgements

T. Hatami is grateful to the Ministry of Science Research and Technology, Iran for financial support of PhD program. J. C. Johner F. thanks CNPq (140287/2013-2) for the PhD assistantship and M. A. A. Meireles thanks CNPq for the productivity grant (302423/2015-0).

References

Antunes-Ricardo, M., Gutiérrez-Uribe, J.A., & Guajardo-Flores, D. (2017). Extraction of isorhamnetin conjugates from Opuntia ficus-indica (L.) Mill using supercritical fluids. J. Supercrit. Fluids, 119, 58-63.

Brender, T., Gruenwald, J., & Jaenicke, C. (1997). Herbal Remedies, Phytopharm Consulting Institute for Phytopharmaceuticals. Second ed., Schaper & Brümmer GmbH & Co., Salzgitter, Berlin, Germany.

Conde-Hernández, L.A., Espinosa-Victoria, J.R., Trejo, A., & Guerrero-Beltrán, J.Á. (2017).

CO2-supercritical extraction, hydrodistillation and steam distillation of essential oil of rosemary (Rosmarinus officinalis). J. Food Eng. 200, 81-86.

del Valle, J.M., & Fuente, J.C.D.L., (2006). Supercritical CO2 extraction of oilseeds: review of kinetic and equilibrium models. Crit. Rev. Food Sci. Nutr. 46, 131-160. 63

Funazukuri, T., Kong, C., & Kagei, S. (1998). Effective axial dispersion coefficients in packed beds under supercritical conditions. J. Supercrit. Fluids, 13, 169-175.

Gallo, M., Formato, A., Ianniello, D., Andolfi, A., Conte, E., Ciaravolo, M., Varchetta, V., & Naviglio, D. (2017). Supercritical fluid extraction of pyrethrins from pyrethrum flowers (Chrysanthemum cinerariifolium) compared to traditional maceration and cyclic pressurization extraction. J. Supercrit. Fluids, 119, 104-112.

Goto, M., Roy, B.C., & Hirose, T. (1996). Shrinking-core leaching model for supercritical-fluid extraction. J. Supercrit. Fluids, 9(2), 128-133.

Hammouda, F.M., Saleh, M.A., Abdel-Azim, N.S., Shams, K.A., Ismail, S.I., Shahat, A.A., & Saleh, I.A. (2014). Evaluation of the essential oil of foeniculum vulgare mill (fennel) fruits extracted by three different extraction methods by Gc/Ms. Afr. J. Tradit. Complement Altern. Med. 11, 277-279.

Hatami, T., Glisic, S.B., & Orlovic, A.M. (2012). Modelling and optimization of supercritical

CO2 extraction of St. John's Wort (Hypericum perforatum L.) using genetic algorithm. J. Supercrit. Fluids, 62, 102-108.

Hatami, T., Johner, J.C.F., & Meireles, M.A.A. (2017). Investigating the effects of grinding time and grinding load on content of terpenes in extract from fennel obtained by supercritical fluid extraction, Ind. Crops Prod. 109, 85-91.

Hatami, T., Moura, L.S., Khamforoush, M., & Meireles, M.A.A. (2014). Supercritical fluid extraction from Priprioca: Extraction yield and mathematical modeling based on phase equilibria between solid and supercritical phases. J. Supercrit. Fluids, 85, 62-67.

Hatami, T., Rahimi, M., Veggi, P.C., Portillo-Prieto, R., & Meireles, M.A.A. (2011). Near- critical carbon dioxide extraction of khoa (Satureja boliviana Benth Briq) using ethanol as a co- solvent: experiment and modeling. J. Supercrit. Fluids, 55(3), 929-936.

Jahromi, B.N., Tartifizadeh, A., & Khabnadideh, S. (2003). Comparison of fennel and mefenamic acid for the treatment of primary dysmenorrhea. Int. J. Gynaecol. Obstet. 80, 153- 157. 64

Johner, J.C.F., & Meireles, M.A.A. (2016). Construction of a supercritical fluid extraction (SFE) equipment: validation using annatto and fennel and extract analysis by thin layer chromatography coupled to image. Food Sci. Technol. (Campinas) 36(2), 210-247.

Kwon, Y.W., & Bang, H. (2000). The Finite Element Method Using MATLAB, CRC Press.

LeVeque, R., (2007). Finite Difference Methods for Ordinary and Partial Differential Equations: Steady-State and Time-Dependent Problems (Classics in Applied Mathematics Classics in Applied Mathemat), Society for Industrial and Applied Mathematics.

López-Padilla, A., Ruiz-Rodriguez, A., Reglero, G., & Fornari, T. (2016). Study of the diffusion coefficient of solute-type extracts in supercritical carbon dioxide: Volatile oils, fatty acids and fixed oils. J. Supercrit. Fluids, 109, 148-156.

Lucas, S., Calvo, M., Garcia-Serna, J., Palencia, C., & Cocero, M. (2007). Two-parameter model for mass transfer processes between solid matrixes and supercritical fluids: Analytical solution. J. Supercrit. Fluids, 41, 257-266.

Mackėla, I., Andriekus, T., & Venskutonis, P.R. (2017). Biorefining of buckwheat (Fagopyrum esculentum) hulls by using supercritical fluid, Soxhlet, pressurized liquid and enzyme-assisted extraction methods. J. Food Eng. 213, 38-46.

Madras, G., Thibaud, C., Erkey, C., & Akgerman, A., (1994). Modeling of supercritical extraction of organics from solid matrices. AIChE J. 40(5), 777-785.

Meireles, M.A.A., Zahedi, G., & Hatami, T. (2009). Mathematical modeling of supercritical fluid extraction for obtaining extracts from vetiver root. J. Supercrit. Fluids, 49, 23-31.

Mongkholkhajornsilp, D., Douglas, S. Douglas, P.L., Elkamel, A., Teppaitoon, W., &

Pongamphai, S. (2005). Supercritical CO2 extraction of nimbin from neem seeds––a modelling study. J. Food Eng. 71, 331-340.

Moura, L.S., Carvalho Jr, R.N., Stefanini, M.B., Ming, L.C., & Meireles, M.A.A. (2005). Supercritical fluid extraction from fennel (Foeniculum vulgare): global yield, composition and kinetic data. J. Supercrit. Fluids, 35(3), 212-219. 65

National Institute of Standards and Technology Chemistry Web Book, (2003).

Oliveira, E.L., Silvestre, A.J., & Silva, C.M. (2011). Review of kinetic models for supercritical fluid extraction. Chem. Eng. Res. Des. 89, 1104-1117.

Pilavtepe, M., & Yesil-Celiktas, O. (2013). Mathematical modeling and mass transfer considerations in supercritical fluid extraction of Posidonia oceanica residues. J. Supercrit. Fluids, 82, 244-250.

Reddy, J.N. (1993). An introduction to the finite element method, McGraw-Hill.

Reis-Vasco, E.M.C., Coelho, J.A.P., Palavra, A.M.F., Marrone, C., & Reverchon, E. (2000). Mathematical modelling and simulation of pennyroyal essential oil supercritical extraction. Chem. Eng. Sci. 55(15), 2917-2922.

Reverchon, E., & Marrone, C. (2001). Modeling and simulation of the supercritical CO2 extraction of vegetable oils. J. Supercrit. Fluids, 19(2),161-175.

Rodríguez-Solana, R., Salgado, J.M., Domínguez, J.M., & Cortés-Diéguez, S. (2014). Estragole quantity optimization from fennel seeds by supercritical fluid extraction (carbon dioxide– methanol) using a Box–Behnken design. Characterization of fennel extracts. Ind. Crops Prod. 60, 186-192.

Song, Y., Zheng, L. & Zhang, X., (2017). Kinetics model for supercritical fluid extraction with variable mass transport. Int. J. Heat Mass Transfer 112, 876-881.

Standard, A.S.A.E., (1998). Method of determining and expressing particle size of chopped forage material by screening. St. Joseph. MI: ASAE.

Valderrama, J.O. & Alarcón, R.F., (2009). A novel hybrid numerical technique to determine mass transport properties in supercritical fluid extraction processes. IJNMBE 25(2), 173-184.

Wakao, N., & Funazkri, T., (1978). Effect of fluid dispersion coefficients on particle-to-fluid mass transfer coefficients in packed beds: correlation of Sherwood numbers. Chem. Eng. Sci. 33, 1375-1384. 66

CHAPTER 5

SUPERCRITICAL FLUID EXTRACTION ASSISTED BY COLD PRESSING ______67

EXTRACTION AND FRACTIONATION OF FENNEL USING SUPERCRITICAL FLUID EXTRACTION ASSISTED BY COLD PRESSING

Tahmasb Hatami, Júlio Cezar Flores Johner, Maria Angela de Almeida Meireles

LASEFI/DEA/FEA (School of Food Engineering), UNICAMP (University of Campinas), Campinas, SP, Brazil

This manuscript has been submitted to Industrial Crops & Products, 2018.

Extraction and fractionation of fennel using supercritical fluid extraction assisted by cold pressing

Tahmasb Hatami*, Júlio Cezar Flores Johner, M. Angela A. Meireles LASEFI/DEA/FEA (School of Food Engineering), UNICAMP (University of Campinas), Campinas, Brazil

Abstract

Supercritical fluid extraction assisted by pressing (SFEAP) and supercritical fluid extraction (SFE) were compared in terms of extraction kinetics and extract fractionation using fennel. Extraction was performed at a pressure of 200 bar, a temperature of 40 °C, and a solvent-to-feed ratio of 95, using torques of 0, 40, and 70 N.m. The overall extraction yield increased by 24.5%, from 9.8 g extract/100 g fennel for SFE to 12.2 g extract/100 g fennel for SFEAP with a torque meter reading of 40 N.m, while overall extraction yield increased by only 2.5% after increasing the torque from 40 N.m to 70 N.m. In the second part of this study, fennel oils extracted with SFE and SFEAP were successfully fractionated into fatty oil rich product and essential oil rich product using two serial, equal-sized separators.

Keywords: Supercritical extraction; Cold pressing; Fractionation from fennel

1. Introduction

Supercritical fluid extraction assisted by pressing (SFEAP) is a novel technique for extraction that has been recently developed by Johner et al. (2018). SFEAP is an integration of SFE and cold pressed methods, and its performance was already evaluated by extraction from pulp of pequi (Caryocar brasiliense). Comparing the SFEAP yield at 400 bar, 40 °C, and 40 N.m torque with that of SFE revealed that the SFEAP yield was eight times greater than that of SFE during the first minute of extraction (Johner et al., 2018). The current paper extends on our previous work (Johner et al., 2018) and employs SFEAP for the extraction of fennel (Foeniculum vulgare). This is a plant belonging to the Apiaceae family, and it is cultivated in several

* Corresponding author. Tel.: +55 19 3521.0100; fax: +55 19 3521.4027. E-mail address: [email protected] (Tahmasb Hatami).

countries, including Brazil, England, Germany, China, and others (Brender et al., 1997). The extract oil of Fennel exhibited antibacterial activity (Ruberto et al., 2000), antiviral activity (Shukla et al., 1988), anticancer activity (Anand et al., 2008), antitumor activity (Al Harbi et al., 1995), and antioxidant properties (Oktay et al., 2003). Badgujar et al. (2014) reviewed ethnomedicinal applications, morphology, phytochemistry, and pharmacology of fennel.

SFE from fennel have been reported in various studies. In a recent publication, Hatami et al. (2017) investigated the effects of grinding time (GT), 15 s to 20 min, and the mass of raw material in a mill, 15 g to 35 g, on both global SFE yield from fennel and fennel’s main volatile oil compounds, namely, anethole and fenchone. The extractor was subjected to a pressure of 200 −4 bar, a temperature of 40 °C, a supercritical CO2 flow rate of 1.67×10 kg/s for 10 min, and a mass of solvent-to-feed ratio (S/F) of 10. It was also found that a grinding time of 6 min and a grinding load of 40.7 g of fennel/lit of mill capacity gave the highest content of anethole (6.16 mg of anethole/g of fennel) and fenchone (0.29 mg of fenchone/g of fennel). In these conditions, the whole SFE yield was 51 mg of extract per g of raw material. Most publications on SFE from fennel focus on the effects of temperature, pressure, particle diameter, and CO2 flow rate on SFE performance (Coelho et al., 2003; Damjanovic et al., 2005; Hatami et al., 2017), but there is no publication about SFEAP from fennel to date. Therefore, the first aim of this paper was to employ both SFE and SFEAP for extraction from fennel at a constant temperature (40 °C) and pressure (200 bar), but at different torques (0, 40, and 70 N.m). The reason for selecting this torques level was firstly to prevent the possible damage to the piston holder at higher torques as evidenced by Johner et al. (2018) for a torque of 120 Nm, and secondly to avoid non-uniformly flowing of CO2 through the bed due to excessive compaction of the raw material. Moreover, this paper highlights the fractionation of extracts from SFE and SFEAP into volatile fractions and lipidic-fraction-rich products and compares their composition with data from the literature. Previous publications revealed that the temperature and pressure of both the extractor and the first separator are the most critical factors affecting fractionation performance. Simandi et al. (1999) fractionated SFE extract from fennel into volatile fractions, rich and lipidic fractions, and rich products using two subsequent separators. They recommended 80-84 bar and 31-35 °C as the best operating conditions of the first separator to minimize the presence of undesired components. Johner and Meireles (2016) performed extraction and fractionation from fennel −4 using two separators with a flow rate of 2.00×10 kg/s of CO2, and S/F of 10. Temperature and

pressure in the extractor, first separator, and second separator were, 40 °C and 200 bar, 35 °C and 80 bar, and 8 °C and 20 bar, respectively. Analysis of the extract showed that from the overall extraction yield of 2.8%, approximately 97.5% of that accumulated in the first separator was a waxy phase, and the remaining 2.5% that accumulated in the second separator was an oily phase containing volatile compounds.

2. Material and methods

2.1. Sample preparation

Fennel used in this study was originally harvested in Campinas, São Paulo, Brazil, and were purchased from the “Temperos Brasil”, in the municipal market of Campinas. Fennel seeds, with a humidity of 8 % (mass), were then kept in a domestic freezer at -18 °C. Prior to grinding, the raw material was taken out of the freezer to stabilize at laboratory temperature, 24 °C. The fennel seeds were then ground in a knife mill (Marconi, model: MA 340, São Paulo, Brazil) for 6 min for the extraction process and for 3 min for extraction and fractionation process. The reason for using different GT for extraction and fractionation compared to that used for pure extraction can be explained as follows. To obtain a noticeable mass of extract in both separators during extraction and fractionation, the extractor was charged with a higher mass of milled fennel, 40 g, compared to the corresponding mass for the pure extraction, 10 g. Pressing compacted the whole 40 g of raw material excessively so that using very fine milled fennel may block the extractor and prevent CO2 passage due to the height of the fennel in the extractor. Thus, fennel with a bigger particle diameter (GT of 3 min) was intentionally selected for extraction and fractionation to prevent the possible blocking. Noticeably, the blocking problem may not occur for the extractor with 10 g of milled fennel with GT of 6 min, due to its small height in the extractor and bigger particle sizes. The standard ASAE method was then used (Standard, 2003) for calculating the average particle diameter. For this purpose, the milled fennel was processed with a vibratory agitator (Bertel, model MAGNETICO, Sao Paulo, Brazil) with six sieves (mesh sizes of 18, 25, 35, 50, 80, and 100) together with a pan at the bottom. The sieves were vibrated for 15 min to ensure that the mass of material in each sieve did not change over time. The material remaining in each sieve and the bottom pan were then accurately weighed, recorded, and used for calculating the average particle diameter.

2.2. SFE from fennel

A schematic diagram of the employed SFE unit is shown in Figure 1. In a typical SFE experiment, the cooling and heating baths are turned on one hour before the extraction to ensure that the system reaches the desired temperatures, 43 °C for the heating baths and -5 °C for the cooling bath. This temperature of heating bath maintained the extractor at 40 oC, and the micro- metering valve at 39 oC. The extractor is then charged with 10 g of milled fennel with a GT of 6 min. This amount of milled fennel occupies a small part of the extractor (13.5 mL), and the remaining space (86.5 mL) is filled with small glass beads. Next, the pump is turned on to pressurize the system up to 200 bar, and the system is kept at this pressure for 20 min. After that, the shut-off valve (indicated by the letter H in Figure 1) and the micro-metering valve (indicated by letter I in Figure 1) were opened. The mixture of CO2 and extract is then passed through the micro- metering valve and its pressure drops to the environmental pressure. Therefore, the extract separates from CO2 and accumulates in a glass vial. The dynamic SFE yield is calculated by collecting and weighing the extract samples at specified extraction time intervals. The CO2 flow rate was adjusted by a gas flow meter located after the glass vial (indicated by letter J in Figure 1). The measurable range of flow rate by this flow meter is from 3.25×10-5 to 4.30×10-4 kg CO2/s at the standard condition. Prior to venting CO2 to the ambient, it is passed through a totalimeter at the end of the line to measure the total CO2 consumed. Due to the fluctuation of flow rate shown by flow meter, the average flow rate of CO2 was calculated using totalimeter by dividing the total CO2 consumed per total dynamic time of extraction. Each SFE run is performed two times to ensure that the obtained results are reliable.

The bed porosity was calculated as follows:

휌 휀 = 1 − 푎 (1) 휌푟 3 3 where 휌푎 and 휌푟 are apparent bed density (kg/m ) and real density (kg/m ) of fennel, respectively. The real density of fennel was measured using a pycnometer, while the apparent bed density was determined by dividing the mass of fennel in the bed per occupied bed volume by fennel. Noticeably, the volume of milled fennel in the extractor was calculated using the internal diameter of extractor (2 cm) along with the height of the empty part of the extractor.

2.3. Supercritical fluid extraction assisted by pressing

A schematic of the SFEAP system is shown in Figure 1. The pressing process of the SFEAP method was carried out inside the extractor. In a typical experiment, 10 g of raw material with a GT of 6 min is inserted into the extractor, and then the pressing piston is attached to the extractor to apply two different torques, 40 and 70 N.m. A torque meter (Sata, ST96303SC, Sorocaba, Brazil) is used to control the force that the piston exerts on the bed. After contacting the piston with the raw material, torque is applied to the thread, and the force on the raw material increases until the equipment clicks, indicating that the selected torque has been reached. After the operator hears the clicking sound, the piston is immediately detached, and glass beads are inserted to fill the remaining part of the extraction column. Subsequently, SFE is performed as described in subsection 2.2.

Figure 1 Schematic diagram of the SFEAP unit: (A1) Pressing system, (A2) Extractor, (B) First separator, (C) Second separator, (D) Outlet of separators, (E) Leadscrew, (F) Socket sliding, (G) Torque meter, (H) Shut-off valve, (I) Micro-metering valve, and (J) Gas flow meter.

2.4. Extraction and fractionation

In a typical SFE experiment, the cooling and heating baths are turned on one hour before the extraction to ensure that the system reaches the desired temperatures, 43 °C for the heating baths and -5 °C for the cooling bath. The extractor is then charged with 40 g of milled fennel with a GT of 3 min. After performing the extraction process by either SFE or SFEAP, the mixture of

CO2 and extract flows out the extractor, passes through the micro-metering valve, and subsequently enters the first separator at 80 bar and 35 °C. The first fraction, which has low solubility in CO2 at this condition, recovers and accumulates at the bottom of separator. Then, the mixture of CO2 and remaining extract leaves the first separator and flows into the second separator at 20 bar and 8 °C. The essential oil rich product settles in this separator, and CO2 exits from the top of separator and enters to the ambient atmosphere after passing through a totalimeter. At the end of the experiment, the bottom valves of separators are opened gradually, and the extract of each separator is accumulated in a vial for weighing and analysis.

3. Results and discussion

Particle size analysis using standard ASAE method (Standard, 2003) showed that GT had a negligible effect on the average particle diameter. Changing GT from 3 to 6 min decreased the average particle diameter very slightly from 0.41 to 0.39 mm. This was mainly due to the fact that grinding moved the raw material from the round body of the mill to the cutting edges (Hatami et al. 2017) so that increasing GT only milled the raw material available in the round body of the mill. Accordingly, the raw material divided into two distinct parts in terms of average particle size: a bigger portion of the raw material, which stayed stationary close to the cutting edges (Hatami et al. 2017), had a bigger average size regardless of the value of GT, and a smaller portion of the raw material, which had mobility in the round body of the mill, had a smaller average size. Although the reason for using GT of 3 min for fractionation instead of GT of 6 min was to prevent the possibility of bed blocking, the aforementioned particle size analysis revealed that both GTs gave practically the same average size. Nonetheless, successful performing the experiments evidenced that GT of 3 to 6 min can still be considered as a desirable range for prevent blocking problem.

Figure 2 represents the kinetic extraction yield of fennel using the SFE and SFEAP techniques. This figure clearly shows the negligible impact of pressing at lower S/F (S/F<5) and a substantial

impact of pressing at higher S/F values. The is principally because carbon dioxide is saturated with extract at the early time of extraction due to its long contact with raw material during static time, and so pressing does not initially have any effect on extraction yield. However, extraction yields for S/F>5 increased considerably with employing SFEAP (40 N.m) compared to SFE. This improvement of extraction yield by pressing is because of liberating more oil from particles and exposing more oil to the supercritical CO2.

Figure 2 Extraction yield of milled fennel (GT = 6 min) using SFE and SFEAP at 200 bar and 40 °C.

This section aimed to show the impact of pressing on extraction yield in the current study with respect to the influence of other typical SFE parameters such as temperature, pressure, S/F ratio, and particle diameter in the literature. Table 1 compares the impact of pressing in the current study with the impact of temperature and pressure reported in two papers by Damjanovic´ et al. (2005) and Moura et al. (2005), as well as the impact of particle diameter reported in a paper by Hatami et al. (2017). Damjanovic´ et al. (2005) highlighted the effect of temperature (40 to 57

°C), and pressure (80 to 150 bar) on SFE yield from 80 g fennel with 0.9 mm diameter under a constant CO2 flow rate of 0.2 kg/h and S/F ratio of 10. They found the optimal conditions for both maximizing the percentage of trans-anethole and fenchone and minimizing methyl chavicol and co-extracted cuticular waxes to be at a pressure of 100 bar, a temperature of 40 °C, and an extraction time of 120 min. A summary of that paper is presented in the first to fifth rows of Table 1. According to this table, increasing the temperature from 40 to 57 °C increases SFE yield from fennel by 9.4%, while increasing the pressure from 100 to 150 bar improves the SFE yield by 38.9%. Moura et al. (2005) extracted oil from fennel using SFE at a constant CO2 flow rate (8.33×10−5 kg/s), extraction time (3 h), particle size (0.61 mm), and S/F ratio of 90, but varied the temperature (30 and 40 °C) and pressure (100, 150, 200, 250, and 300 bar). The best operating conditions for maximizing SFE performance were a temperature of 40 °C and a pressure of 250 bar. Under these conditions, 12.5 g extract/100 g fennel was recovered. A short sensitivity analysis of those results is available in the sixth to ninth rows of Table 1. Although increasing the temperature from 30 to 40 °C did not have a considerable effect on the overall extraction yield, increasing the pressure from 150 to 300 bar enhanced the SFE yield at 30 °C by 52.2% and by 44.3% at 40 °C. Hatami et al. (2017) investigated the effects of the particle diameter of fennel (0.37 to 0.48 mm) on the global SFE yield subjected to a pressure of 200 bar, -4 a temperature of 40 °C, a supercritical CO2 flow rate of 1.67×10 kg/s, and an S/F ratio of 10 for 10 minutes. They reported that decreasing the particle diameter from 0.48 to 0.39 mm improved the yield by 31.9%, while further decreasing the particle diameter from 0.39 to 0.37 mm improved yield by only 11.2 %. A general analysis of Table 1 reveals that pressure and temperature are by far the most and least important factors in SFE from fennel, respectively. The impacts of particle diameter and temperature are very comparable with the impact of pressing in the current study, as employing a torque of 40 N.m in SFE increased the yield by 24.5%.

Table 1 The significance of temperature, pressure, S/F, particle diameter, and torque on the SFE yield from fennel.

Torque Change of T (oC) P (bar) S/F d (mm) (N.m) SFE yield Reference 40 100 → 150 10 0.90 0 38.9 % Damjanovic´ et al. (2005) 47 100 → 150 10 0.90 0 36.8 % Damjanovic´ et al. (2005) 57 100 → 150 10 0.90 0 31.5 % Damjanovic´ et al. (2005) 40 → 57 100 10 0.90 0 9.4 % Damjanovic´ et al. (2005) 40 → 57 150 10 0.90 0 3.6 % Damjanovic´ et al. (2005) 30 150 → 300 90 0.61 0 52.2 % Moura et al. (2005) 40 150 → 300 90 0.61 0 44.3 % Moura et al. (2005) 30 → 40 150 90 0.61 0 6.1 % Moura et al. (2005) 30 → 40 300 90 0.61 0 0.6 % Moura et al. (2005) 40 200 10 0.48 → 0.39 0 31.9 % Hatami et al. (2017) 40 200 10 0.39 → 0.37 0 11.2 % Hatami et al. (2017) 40 200 95 0.39 0 → 40 24.5 % Current study 40 200 95 0.39 40 → 70 2.5 % Current study

After proving the superiority of SFEAP over SFE in terms of dynamic extraction yield, it was worthwhile to compare them in terms of fractionation. Figure 3 represents a schematic of raw material after extraction using SFE (A) and SFEAP (B), SFE extract in Separator 1 (C) and Separator 2 (D), and SFEAP extract in Separator 1 (E) and Separator 2 (F). As seen in this figure, pressing seems to have had a considerable effect on the amount of extract in the first separator but had negligible effects on the amount of extract in the second separator (This conclusion is confirmed later based on the data in Table 2). Added to this, SFE and SFEAP produced extract with distinct color strength that indicates different compositions of extract. The extract in the first separator was in the solid phase, while that of the second separator was in the liquid phase.

Figure 3 A schematic of (A) raw material after extraction using SFE, (B) raw material after extraction using SFEAP, (C) SFE extract in Separator 1 (D) SFE extract in Separator 2, (E) SFEAP extract in Separator 1, (F) SFEAP extract in Separator 2.

Table 2 compares SFE to SFEAP in terms of extraction and fractionation. It can be deduced from this table that the bed porosity drops dramatically from 0.41 for SFE to 0.13 for SFEAP due to the significant impact of the pressing stage. Numbers in this table also show that pressing had a positive effect on the amount of extract in the first separator, which increased by 2.4 g/100 g. However, pressing decreased the amount of extract in the second separator from 0.6 without pressing to 0.3 with a torque of 40 N.m. This meant that although SFEAP brought the benefit of

increasing the extract performance in the first separator, it also decreased the amount of extract in the second separator. This can be explained by the fact that employing pressing has two different effects. On the one hand, pressing is not selective for oils, and increases extracting from the whole cellular content including proteins, gums, enzymes, and volatile oil. Accordingly, it causes more oil to be exposed to the supercritical fluid, which consequently increases the amount of extract in the first separator, which mostly contains fatty oil rich product (Simandi et al., 1999). On the other hand, pressing increases the internal energy of raw material inside the bed (according to the first law of thermodynamic) that it consequently increases the temperature of raw material. This temperature increase in the extractor leads to the evaporation of some part of oil components during the detachment of the torque meter and a consequent decrease in the amount of extract in the second separator, which mostly contains the essential oil rich product (Simandi et al., 1999). Therefore, it can be concluded that despite the superiority of SFEAP over SFE in terms of extraction performance, SFE gave better performance for obtaining essential oil rich product in the second separator of fractionation. Table 2 also concludes that pressing improved the total extraction yield accumulated in both separators by 50%, from 4.2 g /100 g of fennel by SFE to 6.3 g /100 g of fennel, which can be considered as the main advantage of SFEAP over SFE.

Table 2 Comparison between SFE and SFEAP in terms of extraction and fractionation.

T (oC) P (bar) Yield (g / 100 g) Method Porosity of Extractor Sep. 1 Sep. 2 Sep. 1 Sep. 2 Sep. 1 Sep. 2 SFE 0.41±0.00 35±1 8±1 80±1 20±1 3.6±1.2 0.6±0.1 SFEAP 0.13±0.04 35±1 8±1 80±1 20±1 6.0±1.5 0.3±0.1

It is interesting to compare the results of the current study with those obtained by Johner and Meireles (2016). Temperature and pressure in their extractor, first separator, and second separator were exactly the same of the current study. As S/F ratio used by Johner and Meireles (2016), S/F = 10, was by far lower than that used in the current study, S/F = 95, it is fairly to make the comparison in terms of the mass fraction of the extract in each separator instead of the mass of extract. The mass fraction of the extract in the first separator in their paper was 97.5%, while in the current study it was 85.7% for SFE and 95.2% for SFEAP. Moreover, the mass

fraction of the extract in the second separator in the paper by Johner and Meireles (2016) was 2.5%, while in the current study it was 14.3% for SFE and 4.8% for SFEAP.

Although SFEAP is still in its initial phase, it is simple to handle and has high performance based on Figure 2. However, its major drawbacks are that it is not a continuous process, and the torque meter should be detached first before starting SFE; thus, it can be time-consuming in terms of technology. Finding a solution for this limitation is highly desirable before scaling up. One possible solution would be employing an electrical pressing piston. Another potential solution is installing an additional outlet in the middle of the bed for CO2 flow-through immediately after pressing.

4. Conclusion

SFEAP is a novel technique currently under development in LASEFI/UNICAMP/Brazil. The major achievement of this study is employing the SFEAP method to improve the extraction yield from fennel. The results confirm that the SFEAP, using torque of 40 N.m, is effective compared to SFE and increased the overall yield by 24.5 %. Increasing torque from 40 to 70 N.m in the SFEAP method does not have a significant effect on the extraction yield. Moreover, extraction and fractionation from fennel showed that pressing has a significant effect on the amount of extract in both separators. While SFEAP with 40 N.m increased the extraction yield in the first separator by 67% (from 3.6 to 6.0 g extract/100 g fennel) in comparison with SFE, it decreased the extraction yield in the second separator by 50% (from 0.6 to 0.3 g extract/100 g fennel). Despite significant improvements in the extraction performance, more research is still required on the economic feasibility of SFEAP to prove its superiority over SFE.

Acknowledgements T. Hatami is grateful to the Ministry of Science Research and Technology, Iran for financial support. J. C. Johner F. thanks CNPq (140287/2013-2) for the PhD assistantship and M. A. A. Meireles thanks CNPq for the productivity grant (302423/2015-0). References

Al Harbi M.M., Qureshi S., Raza M., Ahmed M.M., Giangreco A.B., Shah A.H., 1995. Influence of anethole treatment on the tumor induced by Ehrlich ascites carcinoma cells in paw of Swiss albino mice. Eur. J. Cancer Prev. 4, 307–318.

Anand P., Kunnumakara A.B., Sundaram C., Harikumar K.V., Tharakan S.T., Lai O.S., Sung B., Aggarwal B.B., 2008. Cancer is a preventable disease that requires major lifestyle changes. Pharmaceut. Res. 25, 2097–2116.

Badgujar S.B., Patel V.V., Bandivdekar A.H., 2014. Foeniculum vulgare Mill: a review of its botany, phytochemistry, pharmacology, contemporary application, and toxicology. Biomed. Res. Int. 2014, 1-32.

Brender, T., Gruenwald, J., Jaenicke, C., 1997. Herbal Remedies, Phytopharm Consulting Institute for Phytopharmaceuticals (2. ed.). Schaper & Brümmer GmbH & Co., Salzgitter, Berlin, Germany.

Coelho, J.A.P., Pereira, A.P., Mendes, R.L., Palavra, A.M.F., 2003. Supercritical carbon dioxide extraction of Foeniculum vulgare volatile oil. Flavour Fragr. J. 18(4), 316-319.

Damjanović, B., Lepojević, Ž., Živković, V., Tolić, A., 2005. Extraction of fennel (Foeniculum vulgare Mill.) seeds with supercritical CO2: Comparison with hydrodistillation. Food Chem. 92(1), 143-149.

Hatami T., Johner J.C.F., Meireles M.A.A., 2017. Investigating the effects of grinding time and grinding load on content of terpenes in extract from fennel obtained by supercritical fluid extraction. Ind. Crops Prod. 109, 85–91.

Johner J.C.F., Hatami T., Meireles M.A.A., 2018. Developing a supercritical fluid extraction method assisted by cold pressing for extraction of pequi (Caryocar brasiliense). J. Supercrit. Fluids 137, 34-39.

Johner J.C.F., Meireles M.A.A., 2016. Construction of a supercritical fluid extraction (SFE) equipment: validation using annatto and fennel and extract analysis by thin layer chromatography coupled to image. Food Sci. Technol. (Campinas) 36(2), 210-247.

Moura, L.S., Carvalho Jr.R.N., Stefanini, M.B., Ming, L.C., Meireles, M.A.A, 2005. Supercritical fluid extraction from fennel (Foeniculum vulgare): global yield, composition and kinetic data. J. Supercrit. Fluids 35(3), 212-219.

Oktay M., Gulcin I., Kufrevioglu I., 2003. Determination of in vitro antioxidant activity of fennel (Foeniculum vulgare) seed extracts. Lebensm. Wiss. Technol. 36, 263–271.

Ruberto G., Baratta M.T., Deans S.G. Dorman H.D., 2000. Antioxidant and antimicrobial activity of Foeniculum vulgare and Crithmum maritimum essential oils. Planta Med. 66(08), 687-693.

Shukla H.S., Dubey P., Chatuvedi R.V., 1988. Antiviral properties of essential oils of Foeniculum vulgare and Pimpinella anisum L. Agronomia 9, 277–279.

Simandi B., Deak A., Ronyai E., 1999. Supercritical Carbon Dioxide Extraction and Fractionation of fennel Oil. J. Agric. Food Chem. 47, 1635−1640.

Standard, A.S.A.E., 1998. Method of determining and expressing particle size of chopped forage material by screening. St. Joseph. MI: ASAE.

CHAPTER 6 GENERAL DISCUSSIONS 83

CHAPTER 6 GENERAL DISCUSSIONS

This thesis studied the effects of grinding procedure and cold pressing on the quality and quantity of extract from fennel using SFE experimentally and theoretically. Grinding procedure and cold pressing are both actually pre processing steps, and this thesis aimed to highlight the impact of these preprocessing steps on the SFE performance. Beside experimental evidences for this purpose, mathematical modeling can help better understanding the impacts and interactions of independent variables.

Referring to grinding procedure, impacts of GT and ms were evaluated on the compositions of anethole and fenchone in the extract from fennel. Increasing GT, for ms=35 g, enhanced the overall extraction yield due to reducing of particle sizes. However, the extraction yield, for the case of ms=15 g, increased with increasing GT from 15 s to 6 min and then reduced with further increasing of GT from 6 min to 20 min. The reason is that higher temperature rising at higher GT, especially for ms=15 g, accelerated the evaporation of volatile compound in the mill, and consequently decreased the SFE performance. Although extraction yield for both anethole and fenchone increased with increasing GT, at first, and then decreased with further increasing GT, their curves of extraction yield were not similar due to the differences between their enthalpy of vaporization. This part of the thesis evidenced that the effect of ms on fenchone yield was almost negligible at small and high values of GT, while its effect for the moderate value of GT (6 min) was considerable. Anethole extraction yield changed significantly with ms at low and especially high values of GT, but its yield at

GT= 6 min was the same for both ms= 15 g and 35 g. It was also found that among the whole values of GT and ms employed in this study, GT=6 min and ms=15 g gave the highest content of anethole and fenchone, which were 6.16 mg of anethole / g of fennel and 0.29 mg of fenchone / g of fennel, respectively. Next, the impact of grinding process was evaluated on the dynamic extraction yield of SFE from fennel over 80 min. The grinding time changed from 15 s to 20 min, and its impact was investigated on the dynamic SFE yield from fennel at 40 oC and 200 bar. The process was also modeled mathematically based on mass conservation law, and a comprehensive solution strategy was provided step by step using FEM. This is the first complete solution of SFE model using FEM that has been reported in such a detail ever. By this method, the PDEs were converted into a set of algebraic equations so that they can be easily solved using simple 84

computer codes. The results confirm that FEM is very effective for solving the external mass transfer control model of SFE process, and the model results has acceptable agreements with the experimental results. The SFE model was also solved using FDM, and its results were compared to FEM. It was found that FEM results for calculating the SFE yield were independent to the number of mesh, while FDM results for calculating SFE yield were strongly influenced by the numbers of mesh elements. Accordingly, considering even two elements was sufficient for FEM solution, while around 200 elements should be considered for FDM to converge the solution. As more number of mesh elements means more computation cost, FEM can be considered as a good substitute instead of FDM for solving SFE model. After considering the effects of grinding procedure both experimentally and mathematically, the impact of another important preprocessing step, namely cold pressing, was evaluated in this thesis for extraction from fennel. The integration of SFE with cold pressing was named SFEAP (Supercritical fluid extraction assisted by pressing), which is a novel technique currently under development in LASEFI/UNICAMP/Brazil. The results confirmed that SFEAP method is effective compared to SFE and increased the overall yield by 27% by using a torque of 40 N.m in the cold pressing step. Nevertheless, increasing torque from 40 to 70 N.m in the SFEAP method did not have a considerable effect on the extraction yield. Moreover, extraction and fractionation from fennel showed that pressing has a significant effect on the amount of extract in both separators. SFEAP with 40 N.m increased the extraction yield in the first separator by 67% (from 3.6 to 6.0 g extract/100 g fennel) in comparison with SFE, while it decreased the extraction yield in the second separator by 50% (from 0.6 to 0.3 g extract/100 g fennel). As the first separator contained fatty acid and the second separator contain volatile oil, these experimental data evidenced that pressing had a negative impact on the fractionation process efficiency. This is possibly due to the fact that pressing the bed leaded to a temperature rising in the bed. Actually based on the first law of thermodynamic, the whole work that performed for the pressing step was converted to the internal energy of the bed, which it in turn increased the bed temperature. This temperature rising accelerated the evaporation of volatile oil component during detaching torquemeter and consequently decreases the amount of volatile extract in the second separator.

CHAPTER 7 GENERAL CONCLUSIONS 86

CHAPTER 7 GENERAL CONCLUSIONS

Grinding time and grinding load affect the composition and amount of extract using SFE methods significantly. These two factors affect the SFE in two ways: On one side, they influence the diameter of raw material, so that using higher GT with lower ms results in smaller particles diameter. On the other side, grinding time and grinding load influence the temperature rising of raw material in the mill, which subsequently changes the evaporation rate of the volatile material. Employing lower GT with lower ms results in lower temperature rising and so lower lost of volatile components. With these two opposite effects, it is necessary to choose suitable values for GT and ms to enhance both SFE yield and quality of extract. Cooling the mill during grinding is a strategy that can be considered in future works to prevent temperature rising. If cooling of the mill is not a suitable option for a specific type of mill, then periodically grinding of the raw material may be work. It means that instead of continuous grinding of raw material for 6 min, it is better to grind them three times (2 min grinding for each) with four min stop in between them for naturally cooling. Mathematical modeling of SFE process helps to evaluate the impacts of various operating factors (such as temperature, pressure, and solvent flow rate) together with their interactions on the SFE performance without doing additional experiments. FEM is a very efficient method to solve the PDEs of SFE model. FEM converts PDEs into a set of algebraic equations so that they can be easily solved using simple computer codes. It is recommended for the future works to employ FEM method on the general SFE model, the model which is provided in the chapter 2 of this thesis, and compare it to FDM method. FEM on the general model will be a two dimensional problem, and it makes the solution strategy more complicated, but more accurate than the current study. SFEAP method increases the extraction yield compared to SFE. Despite this superiority, SFE gave better performance for obtaining volatile extract in fractionation as employing pressing causes a temperature rising in the extractor that leads the evaporation of volatile oil component during detaching torquemeter, which consequently decreases the amount of volatile extract. As SFEAP is in very early stage of development, more efforts are still required in terms of economic feasibility to check its superiority over SFE. It is strongly desired in the future researches to employ SFEAP for extraction from other raw materials such as clove. Proposing a mathematical model for SFEAP process will be also interesting. 87

MEMORANDUM OF THE PhD PERIOD

Tahmasb Hatami joined UNICAMP as a PhD student in 2015 through the selective process of the Department of Food Engineering. With financial assistance from MSRT (Ministry of Science, Research and Technology of Iran), he studied 4 compulsory subjects including TP-320 Thermodynamics; TP-322 Transport Phenomena I; TP-323 Transport Phenomena II; TP-199 Seminar, as well as three optional disciplines offered in the respective department including BI-001 Fundamentals of Biomass Production; TP-333 Experimental Design and Processes Optimization; TP-121 Topics in Food Engineering (totally 22 credits). Moreover, Tahmasb participated as Training Assistant of Thermodynamic for undergraduate students in 2016. He has also worked with various relevant software for mathematical modeling of process involved supercritical fluid. For validation of the employed models, he performed the required SFE and SFEAP experiments, and analyzed the extract oil with Gas Chromatography. As a visiting researcher, Tahmasb also worked under supervision of Prof. Ozan Ciftci in the Department of Food, Science, and Technology, University of Nebraska-Lincoln, the USA for six months. He had a very good collaboration with researchers in UNL, and together they did three projects as follow. 1) Formation of hollow solid micro- and nanoparticles from fully hydrogenated soybean oil via supercritical carbon dioxide was best evaluated experimentally and theoretically at varying pressures (100-300 bar), temperatures (60-80 ºC), and FHSO volumes (10-50 mL). 2) SFE of Lycopene from tomato by-product was studied experimentally, and the process was modeled and optimized in terms of peel to seeds ratio. 3) Curcumine loading into nanoporous starch aerogels was performed experimentally, and the process was modeled based on mass conservation law.

ARTICLES PUBLISHED IN JOURNALS

Hatami, T., Johner, J.C.F., Meireles, M.A.A., 2017. Investigating the effects of grinding time and grinding load on content of terpenes in extract from fennel obtained by supercritical fluid extraction, Ind. Crops Prod. 109, 85-91.

Vardanega, R., Dalmolin, I.A., Nogueira, G.C., Hatami, T., Meireles, M.A.A., 2017. Phase behaviour and mathematical modelling for the system annatto seed oil in compressed carbon dioxide+ ethanol as co-solvent. J. Supercrit. Fluids, 130, 56-62. 89

Rahimpur, N., Hatami, T., Meireles, M.A.A., 2016. Modeling and Optimization of Supercritical Fluid Extraction of Compounds from Campomanesia xanthocarpa Fruits: Comparison between Artificial and Diffusion Based Models, Food Public Health, 6(1) 1-7.

ARTICLES PREPARED AND / OR SUBMITTED FOR PUBLICATION:

Hatami, T., Meireles, M.A.A., Ciftci, O.N., 2018. Supercritical carbon dioxide extraction of lycopene from tomato processing by-products: Experiments, mathematical modeling, and optimization. Submitted to J. Food Eng.

Hatami, T., Johner, J.C.F., Meireles, M.A.A., 2018. Extraction and fractionation from fennel using supercritical fluid extraction assisted by cold pressing. Submitted to Ind. Crops Prod.

Johner, J.C.F., Hatami, T., Carvalho, P.I.N., Meireles, M.A.A., 2018. Impact of grinding procedure on the yield and quality of extract from Clove buds using supercritical fluid extraction. Submitted to Open Food Sci. J.

Johner, J.C.F., Hatami, T., Zabot, G.L., Meireles, M.A.A., 2018. Kinetic behavior and economic evaluation of supercritical fluid extraction of oil from pequi (Caryocar brasiliense) for various grinding times and solvent flow rates. Submitted to J. Supercrit. Fluids.

Johner, J.C.F., Hatami, T., Zabot, G.L., Meireles, M.A.A., 2018. Extraction performance, volatile oil composition, and economic feasibility of supercritical fluid extraction assisted by pressing: a case study with clove buds. Under final revision for submition.

Yang, J., Hatami, T., Meireles, M.A.A., Ciftci, O.N., 2018. Mathematical modeling of formation of hollow solid lipid micro- and nanoparticles using supercritical carbon dioxide. Under final revision for submition. 90

Hatami, T., Johner, J.C.F., Meireles, M.A.A., 2018. A step by step finite element method for solving the external mass transfer control model of supercritical fluid extraction. Under final revision for submition.

Hatami, T., Ubeyitogullari, A., Meireles, M.A.A., Ciftci, O.N., 2018. Curcumin loading into nanoporous starch aerogels using ethanol as solvent: Experiment, mathematical modeling, and sensitivity analysis. Under final revision for submition.

REFERENCES

Al Harbi, M.M., Qureshi, S., Raza, M., Ahmed, M.M., Giangreco, A.B., Shah, A.H., 1995. Influence of anethole treatment on the tumor induced by Ehrlich ascites carcinoma cells in paw of Swiss albino mice. Eur. J. Cancer Prev. 4, 307–318.

Anand, P., Kunnumakara, A.B., Sundaram, C., Harikumar, K.V., Tharakan, S.T., Lai, O.S., Sung, B., Aggarwal, B.B., 2008. Cancer is a preventable disease that requires major lifestyle changes. Pharmaceut. Res. 25, 2097–2116.

Baby, K.C., Ranganathan, T.V., 2016. Effect of enzyme pre-treatment on extraction yield and quality of fennel (Foeniculum vulgare) volatile oil. Biocatal. Agric. Biotechnol. 8, 248-256.

Badgujar, S.B., Patel, V.V., Bandivdekar, A.H., 2014. Foeniculum vulgare Mill: a review of its botany, phytochemistry, pharmacology, contemporary application, and toxicology. Biomed. Res. Int. 2014, 1-32.

Bernath, J., Nemeth, E., Kattaa, A., Hethelyi, E., 1996. Morphological and chemical evaluation of fennel (Foeniculum vulgare Mill.) populations of different origin. J. Essent. Oil Res. 8, 247-253.

Bodsgard, B.R., Lien, N.R., Waulters, Q.T., 2016. Liquid CO2 Extraction and NMR Characterization of Anethole from Fennel Seed: A General Chemistry Laboratory. J. Chem. Educ. 93, 397-400.

Brender, T., Gruenwald, J., Jaenicke, C., 1997. Herbal Remedies, Phytopharm Consulting Institute for Phytopharmaceuticals. Second ed., Schaper & Brümmer GmbH & Co., Salzgitter, Berlin, Germany.

Causon, D.M., Mingham, C.G., 2010. Introductory finite difference methods for PDEs. Bookboon.

Chougle, J.A., Bankar, S.B., Chavan, P.V., Patravale, V.B., Singhal, R.S., 2016. Supercritical carbon dioxide extraction of astaxanthin from Paracoccus NBRC 101723: Mathematical modelling study. Sep. Sci. Technol. 51(13), 2164-2173. 92

Coelho, J.A.P., Pereira, A.P., Mendes, R.L., Palavra, A.M.F., 2003. Supercritical carbon dioxide extraction of Foeniculum vulgare volatile oil. Flavour Fragr. J. 18(4), 316-319.

del Valle, J.M., & Fuente, J.C.D.L., 2006. Supercritical CO2 extraction of oilseeds: review of kinetic and equilibrium models. Crit. Rev. Food Sci. Nutr. 46, 131-160.

Ding, X., Liu, Q., Hou, X., Fang, T., 2017. Supercritical Fluid Extraction of Metal Chelate: A Review. Crit. Rev. Anal. Chem. 47(2), 99-118.

Durante, M., Lenucci, M.S., Mita, G., 2014. Supercritical carbon dioxide extraction of carotenoids from pumpkin (cucurbita spp.): A review. Int. J. Mol. Sci. 15, 6725-6740.

Funazukuri, T., Kong, C., Kagei, S. 1998. Effective axial dispersion coefficients in packed beds under supercritical conditions. J. Supercrit. Fluids 13, 169-175.

Gallo, M., Formato, A., Ianniello, D., Andolfi, A., Conte, E., Ciaravolo, M., Varchetta, V., Naviglio, D. 2017. Supercritical fluid extraction of pyrethrins from pyrethrum flowers (Chrysanthemum cinerariifolium) compared to traditional maceration and cyclic pressurization extraction. J. Supercrit. Fluids 119, 104-112. 93

Goto, M., Roy, B.C., Hirose, T., 1996. Shrinking-core leaching model for supercritical-fluid extraction. J. Supercrit. Fluids, 9(2), 128-133.

Grosso, C., Ferraro, V., Figueiredo, A.C., Barroso, J.G., Coelho, J.A., Palavra, A.M., 2008. Supercritical carbon dioxide extraction of volatile oil from Italian coriander seeds. Food Chem. 111(1), 197-203.

Hatami, T., Glisic, S.B., Orlovic, A.M., 2012. Modelling and optimization of supercritical CO2 extraction of St. John's Wort (Hypericum perforatum L.) using genetic algorithm. J. Supercrit. Fluids 62, 102-108.

Hatami, T., Rahimi, M., Veggi, P.C., Portillo-Prieto, R., Meireles, M.A.A., 2011. Near- critical carbon dioxide extraction of khoa (Satureja boliviana Benth Briq) using ethanol as a co-solvent: experiment and modeling. J. Supercrit. Fluids, 55(3), 929-936.

Https://www.chemeo.com/cid/44-213-6/Anethole (Accessed 15 November, 2016).

Https://www.chemeo.com/cid/70-160-6/Fenchone (Accessed 15 November, 2016).

Jahromi, B.N., Tartifizadeh, A., Khabnadideh, S., 2003. Comparison of fennel and mefenamic acid for the treatment of primary dysmenorrhea. Int. J. Gynaecol. Obstet. 80, 153-157. 94

Kwon, Y.W., Bang, H., 2000. The Finite Element Method Using MATLAB, CRC Press.

López-Padilla, A., Ruiz-Rodriguez, A., Reglero, G., Fornari, T., 2016. Study of the diffusion coefficient of solute-type extracts in supercritical carbon dioxide: Volatile oils, fatty acids and fixed oils. J. Supercrit. Fluids 109, 148-156.

Mackėla, I., Andriekus, T., Venskutonis, P.R., 2017. Biorefining of buckwheat (Fagopyrum esculentum) hulls by using supercritical fluid, Soxhlet, pressurized liquid and enzyme-assisted extraction methods. J. Food Eng. 213, 38-46.

Madras, G., Thibaud, C., Erkey, C., & Akgerman, A., 1994. Modeling of supercritical extraction of organics from solid matrices. AIChE J. 40(5), 777-785.

Meireles, M.A.A., Zahedi, G., Hatami, T., 2009. Mathematical modeling of supercritical fluid extraction for obtaining extracts from vetiver root. J. Supercrit. Fluids 49, 23-31.

Mongkholkhajornsilp, D., Douglas, S. Douglas, P.L., Elkamel, A., Teppaitoon, W.,

Pongamphai, S., 2005. Supercritical CO2 extraction of nimbin from neem seeds––a modelling study. J. Food Eng. 71, 331-340.

National Institute of Standards and Technology Chemistry Web Book, 2003.

Oktay, M., Gulcin, I., Kufrevioglu, I., 2003. Determination of in vitro antioxidant activity of fennel (Foeniculum vulgare) seed extracts. Lebensm. Wiss. Technol. 36, 263–271.

Oliveira, E.L., Silvestre, A.J., Silva, C.M., 2011. Review of kinetic models for supercritical fluid extraction. Chem. Eng. Res. Des. 89, 1104-1117.

Pereira, C.G., Meireles, M.A.A., 2007. Economic analysis of rosemary, fennel and anise essential oils obtained by supercritical fluid extraction. Flavour Frag. J. 22, 407-413.

Pilavtepe, M., Yesil-Celiktas, O., 2013. Mathematical modeling and mass transfer considerations in supercritical fluid extraction of Posidonia oceanica residues. J. Supercrit. Fluids, 82, 244-250.

Piras, A., Falconieri, D., Porcedda, S., Marongiu, B., Gonçalves, M.J., Cavaleiro, C.,

Salgueiro, L., 2014. Supercritical CO2 extraction of volatile oils from Sardinian Foeniculum vulgare ssp. vulgare (Apiaceae): chemical composition and biological activity. Nat. Prod. Res. 28, 1819-1825.

Raghavan, S., 2006. Handbook of spices, seasonings, and flavorings. CRC Press.

Reddy, J.N. 1993. An introduction to the finite element method, McGraw-Hill.

Reis-Vasco, E.M.C., Coelho, J.A.P., Palavra, A.M.F., Marrone, C., Reverchon, E. 2000. Mathematical modelling and simulation of pennyroyal essential oil supercritical extraction. Chem. Eng. Sci. 55(15), 2917-2922.

Rodríguez-Solana, R., Salgado, J.M., Domínguez, J.M., Cortés-Diéguez, S., 2014b. Estragole quantity optimization from fennel seeds by supercritical fluid extraction (carbon dioxide– 96

methanol) using a Box–Behnken design. Characterization of fennel extracts. Ind. Crops Prod. 60, 186-192.

Ruberto, G., Baratta, M.T., Deans, S.G. Dorman, H.D., 2000. Antioxidant and antimicrobial activity of Foeniculum vulgare and Crithmum maritimum essential oils. Planta Med. 66(08), 687-693.

Shrigod, N.M., Swami Hulle, N.R., Prasad, R.V., 2017. Supercritical fluid extraction of essential oil from mint leaves (mentha spicata): Process optimization and its quality evaluation. J. Food Process. Eng. 40(3), 1-9.

Shukla, H.S., Dubey, P., Chatuvedi, R.V., 1988. Antiviral properties of essential oils of Foeniculum vulgare and Pimpinella anisum L. Agronomia 9, 277–279.

Simandi B., Deak A., Ronyai E., 1999. Supercritical Carbon Dioxide Extraction and Fractionation of fennel Oil. J. Agric. Food Chem. 47, 1635−1640.

Sodeifian, G., Ardestani, N.S., Sajadian, S.A., Ghorbandoost, S., 2016. Application of supercritical carbon dioxide to extract essential oil from Cleome coluteoides Boiss: experimental, response surface and grey wolf optimization methodology. J. Supercrit. Fluids 114, 55-63.

Song, Y., Zheng, L. Zhang, X., 2017. Kinetics model for supercritical fluid extraction with variable mass transport. Int. J. Heat Mass Transfer 112, 876-881.

Standard, A.S.A.E., 1998. Method of determining and expressing particle size of chopped forage material by screening. St. Joseph. MI: ASAE.

Valderrama, J.O. Alarcón, R.F., 2009. A novel hybrid numerical technique to determine mass transport properties in supercritical fluid extraction processes. IJNMBE 25(2), 173-184.

Wakao, N., Funazkri, T., 1978. Effect of fluid dispersion coefficients on particle-to-fluid mass transfer coefficients in packed beds: correlation of Sherwood numbers. Chem. Eng. Sci. 33, 1375-1384.

Wilkinson, N., Hilton, R., Hendry, D., Venkitasamy, C., Jacoby, W., 2014. Study of process variables in supercritical carbon dioxide extraction of soybeans. Food Sci. Technol. Int. 20(1), 63-70. 97

APPENDIX

Title: Investigating the effects of grinding time and grinding load on content of terpenes in extract from fennel obtained by supercritical fluid extraction Author: Tahmasb Hatami,Júlio Cezar

Please note that, as the author of this Elsevier article, you retain the right to include it in a thesis or dissertation, provided it is not published commercially. Permission is not required, but please ensure that you reference the journal as the original source. For more information on this and on your other retained rights, please visit: https://www.elsevier.com/about/our-business/policies/copyright#Author-rights