Hydrogenation of selected carbohydrate molecules in the presence of solid foam catalysts in a stirred tank reactor

UNIVERSITY OF VALLADOLID SCHOOL OF INDUSTRIAL ENGINEERINGS Master’s in Chemical Engineering MASTER’S THESIS Hydrogenation of selected carbohydrate molecules in the presence of solid foam catalysts in a stirred tank reactor Author: Araujo Barahona, German Rodrigo Supervisors: García Serna, Juan University of Valladolid Salmi, Tapio Laboratory of Industrial Chemistry and Reaction Engineering Åbo Akademi University Valladolid, May 2021 Hydrogenation of selected carbohydrate molecules in the presence of solid foam catalysts in a stirred tank reactor Master’s Thesis German Rodrigo Araujo Barahona Johan Gadolin Process Chemistry Centre Laboratory of Industrial Chemistry and Reaction Engineering Faculty of Science and Engineering/Chemical Engineering Åbo Akademi University Turku/Åbo, Finland, 2021 i TFM REALIZADO EN PROGRAMA DE INTERCAMBIO TÍTULO: Hydrogenation of selected carbohydrate molecules in the presence of solid foam catalysts in a stirred tank reactor ALUMNO: German Rodrigo Araujo Barahona FECHA: 17 de mayo de 2021 CENTRO: Faculty of Science and Engineering UNIVERSIDAD: Åbo Akademi University TUTOR: Tapio Salmi ii PREFACE “While you live, shine. Have no grief at all; life exists only for a short while and time demands his due.” - Seikilos epitaph. These two years in the master's program have been a life-changing experience. I still remember that September 11, 2019, at the International Airport of El Salvador with my luggage full of dreams, my family saying goodbye, and a world of experiences waiting for me. None of this would have been possible without the support of many people. First, I would like to express my gratitude to my master's coordinator, Professor Juan García Serna, for all the support from the very beginning, which made it possible for me to pursue my master's degree at the University of Valladolid, his enthusiasm and commitment to students are qualities that I have always admired about him. I would like to thank my supervisor at Åbo Akademi University, Professor Tapio Salmi, for the opportunity to work with him. Tapio is not only a great scientist but also a great person who shares his vast knowledge, kindness, and sense of humor with everyone around him. I feel privileged to have been able to learn from him in all these aspects. The experimental work of this thesis would not have been possible without the help of Docent Kari Eränen, I am grateful not only for his exceptional ability to solve any problem in the lab, but also for all that I have learned about experimental work thanks to his patience and ingenuity. I would like to thank Professor Dmitry Yu. Murzin and Docent Narendra Kumar for their wise advice on catalyst preparation and characterization, a visit to their offices always brings new ideas and perspectives. I would like to express my gratitude to Dr. Zuzana Vajglová for her guidance in the development of the incipient wetness impregnation method, that was definitely a turning point in this research. Thanks to Dr. Atte Aho for his help with HPLC and TPR measurements, thanks to Jay Pee for his help in the anodic oxidation experiments. Thanks to all the good friends that I made in Finland for all the adventures and for making me feel at home: Christoph, Ole, Tom, Pontus, Adriana, Mark, Matias, Bernadette, Pasi, Mouad, Federica, and Luca. Thank you, Marcela, por estar a pesar de la distancia; thank you, Andrea and Hamza for being incredible friends; thank you, Olga for your friendship and good wishes; and thank you, Karen, for your friendship and advice. I would like to thank the Latin America + Asia / University of Valladolid-Santander Bank Scholarship Program for providing the funds that allowed me to pursue my master’s studies, and the Erasmus Plus program for the financial support during my exchange in Finland. Y, por último, pero no por eso menos importante, gracias a mi familia. Ustedes son mi todo y les debo todo lo que tengo; a mi madre Haydeé, a mi padre German, a mi hermana Lilian, a mis tíos Ruth y Carlos Humberto, muchas gracias por el apoyo incondicional y por siempre creer en mí. Mil gracias, tack så mycket, kiitos paljon! -German. iii ABSTRACT Keywords: hydrogenation, heterogeneous catalysis, structured catalysts, ruthenium, L-arabinose, Dgalactose, binary sugar mixtures, sugar alcohols. This research work was carried out at the Laboratory of Industrial Chemistry and Reaction Engineering (TKR) at Åbo Akademi University (Turku/Åbo, Finland) in collaboration with the University of Valladolid (Valladolid, Spain) under the supervision of Tapio Salmi (Professor of the Academy of Finland) and Juan García Serna (Full Professor at the University of Valladolid) as part of the Erasmus Plus exchange program. The growing concern about the short and long-term consequences caused by climate change is driving humankind towards a more sustainable development, which requires adequate diversification of feedstock and industrial production processes. In this context, the use of lignocellulosic biomass as raw material for chemical industry is a promising option on which a lot of research effort has been focused in recent years, such as the production of sugar alcohols. These compounds can be obtained by catalytic hydrogenation of mono and disaccharides present in the cellulose and hemicelluloses fractions of biomass. Sugar alcohols have a wide range of applications e.g., in the alimentary industry as healthier sweeteners or in the pharmaceutical industry as excipients and anti-caries agents. The research effort of this thesis was focused on the development of a novel solid foam catalyst based on ruthenium supported on carbon. This heterogeneous catalyst was used to perform kinetic experiments on the hydrogenation of L-arabinose and D-galactose at different temperatures (90˚C, 100˚C, and 120˚C) and hydrogen pressures (20 and 40 bar) to investigate the effect of these parameters on the hydrogenation rate. Furthermore, kinetic experiments were carried out with binary sugar mixtures at different D-galactose to L-arabinose molar ratios to study the interactions of these sugars during the reaction in the presence of the prepared catalyst. The solid foam catalyst preparation comprised the following steps: cutting of the open-cell foam aluminum pieces, anodic oxidation pretreatment, carbon coating, acid pretreatment, ruthenium incorporation, and ex-situ reduction. The carbon coating method comprised the polymerization of furfuryl alcohol, followed by a pyrolysis process and activation with oxygen. The degree of crosslinking of polyfurfuryl alcohol was identified as a relevant parameter to obtain a carbon coating iv with appropriate properties to act as catalyst support; thus, the polymerization conditions were optimized to obtain the desired catalyst properties. Incorporation of ruthenium on the carbon-coated foam was done by two different methods, homogeneous deposition precipitation (HDP) and incipient wetness impregnation (IWI), using in both cases ruthenium(III) nitrosyl nitrate as the precursor solution. In the HDP method, the carbon content of the foams and the molar ratio of urea-to-ruthenium were the most relevant parameters to obtain an active catalyst. On the other hand, for the IWI method, the carbon content and the concentration of the precursor solution were identified as the most relevant parameters. Using IWI, it was possible to prepare an active catalyst with a ruthenium load of 1.1 wt. % for the conversion of the sugars to the corresponding sugar alcohol. This catalyst was used in the systematic kinetic experiments for both the individual sugars and sugar mixtures. Several catalyst characterization techniques such as Scanning Electron Microscopy (SEM), Transmission Electron Microscopy (TEM), Temperature-Programmed Reduction (TPR), and Inductively Coupled Plasma Atomic Optical Emission Spectroscopy (ICP-OES) were used to interpret the behavior of the catalyst in terms of activity, durability and critical parameters for the catalyst preparation. Extensive kinetic experiments were carried out in an isothermal laboratory-scale semibatch reactor to which gaseous hydrogen was constantly added. Two pieces of solid foam catalysts were placed at the endpoint of an agitating shaft and rotated at a constant speed during the experiments. From the individual kinetic experiments, high selectivities towards sugar alcohols, exceeding 98% were obtained for both sugars, in fact, the conversions were within the range of 60-98%, depending on the temperature. The temperature effect on the reaction rate was very strong, while the effect of the hydrogen pressure was rather minor. Regarding the sugar mixtures, in general, the L-arabinose presented a higher reaction rate, and an acceleration of the hydrogenation process was observed for both sugars as the ratio of D-galactose to L-arabinose increased, evidently as a result of competitive interaction on the catalyst surface. A kinetic model based on a non-competitive adsorption mechanism between sugar molecules and hydrogen was tested with extensive experimental data by applying non-linear regression. A good description of the concentration profiles and the temperature effect on the reaction kinetics was achieved with the mathematical model. Furthermore, a detailed sensitivity analysis revealed that v the estimated parameters were very well defined and all of them had an important contribution to the model. The obtained results demonstrate the feasibility of converting primary sugars from biomass such as L-arabinose and D-galactose and their mixtures into the corresponding sugar alcohols using ruthenium as the active metal on an active carbon support implemented in an open foam structure. A possible next step would be the use of this catalyst in continuous three-phase reactors that allow taking the advantage of the properties associated with structured catalysts, such as high flow rates, high external heat and mass transfer rates and low diffusion resistance in the active catalyst layer. vi REFERAT Nyckelord: hydrering, heterogen katalys, strukturerade katalysatorer, L-arabinos, D-galaktos, binära sockerblandningar, sockeralkoholer. Detta arbete genomfördes vid Laboratoriet för teknisk kemi och reaktionsteknik vid Åbo Akademi, Finland i samarbete med Valladolid universitet, Spanien. Arbetet handleddes av akademiprofessor Tapio Salmi (Finlands Akademi och Åbo Akademi) och professor Juan García Serna (Valladolid universitet) inom ramen för Erasmus Plus –utbytesprogrammet. Den växande medvetenheten om de kortoch långvariga konsekvenser som klimatförändringen orsakar driver mänsligheten mot en mer hållbar utveckling, vilket kräver en adekvat diversifiering av råvaror och industriella processer. I detta sammanhang är användningen av lignocellulosabaserad biomassa som råmaterial för kemisk industri ett lovande alternativ. Extensiv forskningsverksamhet har idkats kring detta tema under de senaste åren, t.ex. produktion av sockeralkoholer, som kan framställas via katalytisk hydrering av monooch disackarider vilka finns tillgängliga i biomassans cellulosaoch hemicellulosafraktioner. Sockeralkoholer har många tillämpningar, t.ex. i livsmedelindustrin som hälsosamma sötningsmedel och i den farmaceutiska indusrin som fyllnadsmedel och anti-carieskomponenter. Detta arbete fokuserades på utveckling av en fast skumkatalysator som baserar sig på ruteniumnanopartiklar på aktivt kol. Katalysatorn användes i kinetiska hydreringsexperiment av Larabinos och D-galaktos vid olika temperaturer (90˚C, 100˚C och 120˚C) och vätetryck (20 och 40 bar). Experiment utfördes för att få fram dessa parametrars inverkan på hydreringshastigheten. Dessutom genomfördes kinetiska experiment med binära sockerblandningar med olika molära förhållanden av D-galaktos och L-arabinos för att studera växelverkan mellan sockerarterna under hydreringsprocessens gång. Prepareringen av den fasta skumkatalysatorn bestod av följande steg: skärning av aluminiumstycken av fast skum, förbehandling av materialet med anodisk oxidation, beläggning av skummet med aktivt kol, förbehandling av skummet med syra, impregnering av materialet med rutenium samt ex situ –reduktion av katalysator-materialet. Beläggningsmetoden baserade sig på polymerisering av furfurylalkohol, pyrolys av polymeren samt aktivering av kolskiktet med syre. Graden av tvärbindning av polyfurfurylalkohol konstaterades vara en relevant parameter, då det gäller att åstadkomma en kolbeläggning med önskade egenskaper så att beläggningen kan fungera som vii katalysatorbärare; därför optimerades polymerisationsbetingelserna för att erhålla de eftersträvade katalytiska egenskaperna. Rutenium inkorporerades på kolbaserat skum med två olika metoder, homogen deponeringsfällning (HDP) samt porimpregnering (IWI). I båda fallen användes rutenium (III)nitrosylnitrat som prekursorlösning. I HDP-metoden var kolinnehållet i skummet och molförhållandet urea:rutenium de mest relevanta parametrarna då det gällde att åstadkomma en aktiv katalysator. För IWImetoden konstaterades kolinnehållet och koncentrationen av prekursorlösningen vara de mest relevanta parametrarna. Genom användning av porimpregnering blev det möjligt att preparera en aktiv katalysator med en ruteniumhalt på 1.1 vikt-% för omvandling av sockerarter till motsvarande sockeralkoholer. Därför användes denna katalysatorvariant för kinetiska studier för både individuella sockerarter och blandningar av dem. Flera katalysatorkarakteriseringsmetoder såsom svepelektronmikroskopi (SEM), transmissionselektronmikro-skopi (TEM), temperaturprogrammerad reduktion (TPR) och induktivt kopplad plasmaspektroskopi (ICP_OES) användes för att tolka och utreda katalysatorns aktivitet, hållbarhet och kritiska parametrar i själva katalysatorprepareringsprocessen. Omfattande kinetiska experiment genomfördes i en isotermisk halvkontinuerlig reaktor i laboratorieskala. Reaktorn tillfördes en kontinuerlig ström av vätgas. Två stycken av fasta skumkatalysatorer placerades i ändan av en omrörare och katalysatorerna roterades med en konstant hastighet under experimentets gång. Höga selektiviteter av sockeralkoholer som överskred 98%, upptäcktes för båda sockerarterna, medan omsättnings-graden av sockerarten varierade mellan 60% och 98% i experimenten, beroende på den aktuella reaktions-temperaturen. Temperaturens inverkan på reaktionshastigheten var stark, medan vätetryckets inverkan på omsättningsgraden var relativt svag. I allmänhet hade L-arabinos en högre reaktionshastighet, men en acceleration av hydreringsreaktionen kunde observeras för båda sockerarterna då förhållandet D-galaktos:L-arabinos ökade, troligen p.g.a. konkurrerande växelverkan på katalysatorytan. En kinetisk modell baserad på icke-konkurrerande adsorptionsmekanism mellan sockermolekyler och väte anpassades till de framtagna experimentella data med hjälp av icke-linjär regressionsanalys. Modellen gav en utmärkt beskrivning av de experimentella koncentrationsprofilerna och temperaturens inverkan på reaktionshastigheten. Utförliga känslighetsberäkningar visade att de estimerade kinetiska och adsorptions-parametrarna var statistiskt sätt väldefinierade och de hade ett väsentligt bidrag till den matematiska modellen. xiv 4.2.3. Parameter Estimation ............................................................................................ 69 4.2.4. Double Logarithmic Plots ....................................................................................... 72 5. CONCLUSIONS AND FUTURE PERSPECTIVES ......................................................................... 75 6. REFERENCES ....................................................................................................................... 78 APPENDIX I: HPLC CALIBRATION DATA .................................................................................... 88 APPENDIX II: HPLC CURVE OF SUGAR MIXTURES (EXAMPLE) .................................................... 89 APPENDIX IV: MODELING SINGLE SUGAR HYDROGENATION (PYTHON CODE) ........................... 94 APPENDIX V: MODELING SUGAR MIXTURES HYDROGENATION (PYTHON CODE)....................... 96 xv INDEX OF FIGURES Figure 1.1. Sugar alcohols from lignocellulosic biomass. .................................................................. 22 Figure 1.2. Structure of arabinogalactan. ......................................................................................... 24 Figure 1.3. Illustration of the semi-competitive sugar/hydrogen adsorption concept (hydrogen is adsorbed in dissociated form)........................................................................................................... 26 Figure 1.4. Optical microscope photograph of an open-cell metallic foam. .................................... 27 Figure 1.5. Linear Polyfurfuryl alcohol structure (PFA). .................................................................... 28 Figure 2.1. Overview of the solid foam catalyst preparation process. ............................................. 31 Figure 2.2. Cutting of aluminum foams. ........................................................................................... 33 Figure 2.3. Experimental arrangement of the anodic oxidation process. ........................................ 34 Figure 2.4. Experimental setup for PFA coating................................................................................ 35 Figure 2.5. Experimental setup for HDP............................................................................................ 36 Figure 2.6. Overview of the setup for sugar hydrogenation experiments. ...................................... 39 Figure 2.7. SEM analysis of foam catalysts. ...................................................................................... 40 Figure 3.1.Changes in the open-cell foam catalyst through its different preparation stages: (a) Al untreated foam, (b) anodized Al foam, (c) foam coated with PFA, (d) pyrolyzed/oxygen treated carbon-coated foam, (e) carbon-coated, Ru impregnated and reduced catalyst............................. 42 Figure 3.2. Potential variation during aluminum foam anodic oxidation at constant current. ........ 42 Figure 3.3. Visual change on the surface of the Al foams. Untreated foam (left) and anodized foam (right). ................................................................................................................................................ 43 Figure 3.4. SEM micrographs of the oxide texture generated in catalyst C8-AO-IWI3-R300. (a) Untreated foam (30X), (b) untreated foam (50 kX), (c) anodized foam (30X), (d) anodized foam (50 kX), (e) anodized and calcined foam (30K X), (f) anodized and calcined foam (50 kX). .................... 44 Figure 3.5.Obtained PFA coating: (a) foamy dark polymer obtained when sudden rising of temperature takes place, (b) golden polymer obtained when not sudden rising of the temperature takes place. ........................................................................................................................................ 46 Figure 3.6. Temperature pattern during PFA coating when a foamy dark PFA coverage is obtained (catalyst C8-IWI3). ............................................................................................................................. 46 Figure 3.7. Surface structure of a carbon-coated foam substrate: (a) C2-HDP2 (12 wt.% carbon), obtained from golden-colored PFA, (b) C7-IWI2 (50 wt.% carbon), obtained from foamy dark PFA, (c) C10-AO-IWI2-R300 (50 wt.% carbon), preanodized and obtained from foamy dark PFA. .......... 47 xvi Figure 3.8. Evolution of pH during HDP at different conditions of Ru nominal load based on carbon (NL) and urea-to-Ru molar ratio (r). .................................................................................................. 48 Figure 3.9. SEM image and elemental analysis (by EDX) of catalyst C3-HDP2 displaying cracks on the aluminum support and incorporation of Ru on the aluminum phase. ............................................. 49 Figure 3.10. TEM images of catalyst C5-HDP4 and particle size distribution. .................................. 50 Figure 3.11. TEM images of catalyst C10-OA-IWI3-R300 and particle size distribution. .................. 51 Figure 3.12. Induction behavior of Catalyst C8-IW3 during hydrogenation of 1:1 D-galactose to Larabinose 0.13 M solution at 120˚C and 20 bar. (a) L-arabinose conversion vs time, (b) D-galactose conversion vs time. ........................................................................................................................... 52 Figure 3.13. Hydrogen-TPR profiles of catalyst C10-IWI-R300 (before ex-situ reduction). .............. 53 Figure 3.14. Deactivation of catalyst C8-IW3 during hydrogenation of (a) L-arabinose, (b) D-galactose at 120˚C and 20 bar. .......................................................................................................................... 53 Figure 3.15.TEM images of catalyst C8-IWI3 after 100 h of use and particle size distribution. ....... 54 Figure 3.16. Effect of temperature on the hydrogenation rates at 20 bar for (a) L-arabinose, (b) Dgalactose. .......................................................................................................................................... 56 Figure 3.17. Effect of pressure on the hydrogenation rates at 120˚C for (a) L-arabinose, (b) Dgalactose. .......................................................................................................................................... 56 Figure 3.18. Effect of the D-galactose to L-arabinose molar ratio on the hydrogenation rates at 120˚ of (a) L-arabinose, (b) D-galactose. ................................................................................................... 57 Figure 4.1. Reaction mechanism for hydrogenation of individual sugars. ....................................... 59 Figure 4.2. Arrhenius plot for the estimated 𝑘′𝑠 parameters: (a) L-arabinose, (b) D-galactose ...... 61 Figure 4.3. Modeling results for L-arabinose hydrogenation at 20 bar and different temperatures: (a) 90˚C, (b) 100˚C and (c) 120˚C. ...................................................................................................... 62 Figure 4.4.Modeling results for D-galactose hydrogenation at 20 bar and different temperatures: (a) 90˚C, (b) 100˚C and (c) 120˚C. ........................................................................................................... 63 Figure 4.5. Sensitivity analysis of fitted parameters: (a) EA, (b) EG, (c) AA', (d) AG', (e)KA, (f) KG. ....... 65 Figure 4.6 Reaction mechanism for the sugar mixtures hydrogenation. ......................................... 67 Figure 4.7. Modeling sugar mixtures hydrogenation results at 120˚C and 20 bar. Ratio=0.5: (a) Larabinose, and (b) D-galactose. ........................................................................................................ 70 Figure 4.8. Modeling sugar mixtures hydrogenation results at 120˚C and 20 bar. Ratio=1: (a) Larabinose, and (b) D-galactose. ........................................................................................................ 70 xvii Figure 4.9.Modeling sugar mixtures hydrogenation results at 120˚C and 20 bar. Ratio=5: (a) Larabinose, and (b) D-galactose. ........................................................................................................ 71 Figure 4.10. Sensitivity analysis of fitted parameters: (a) 𝛋𝐀, (b) 𝛋G, (c) KA. ..................................... 71 Figure 4.11. Double logarithmic plots of sugar mixtures at 120˚C and 20 bar: (a) D-galactose: Larabinose=0.5, (b) D-galactose: L-arabinose=1, (a) D-galactose: L-arabinose=5. ............................. 73 xviii INDEX OF TABLES Table 2.1. Batch codes and general information of the prepared catalysts. .................................... 32 Table 2.2. Ruthenium incorporation conditions. .............................................................................. 37 Table 3.1. Elemental Analysis (EDX) of aluminum foam during the different anodic oxidation stages (sample: C8-AO-IWI2-R300). ............................................................................................................. 44 Table 3.2.Carbon-coating conditions of the prepared catalyst supports. ........................................ 45 Table 3.3. Comparison of the physical properties and reactivity of the prepared catalyst ............. 51 Table 3.4 Selectivity and conversion of L-arabinose and D-Galactose after 6 h of reaction at 20 bar and different temperature (individual sugar hydrogenation). ......................................................... 55 Table 4.1. Arrhenius parameters determined by linear regression. ................................................. 61 Table 4.2. Kinetic parameters estimated for L-arabinose and D-galactose in individual hydrogenation experiments. ............................................................................................................. 64 Table 4.3. Kinetic parameters estimated for L-arabinose and D-galactose in mixtures experiments. ........................................................................................................................................................... 69 Table 4.4. Relative reactivities at different initial mole ratios of D-galactose to L-arabinose.......... 74 xix NOTATION Nomenclature A's Lumped pre-exponential factor. a Slope in HPLC calibration curves AHPLC Area in HPLC graphs C*’H Hydrogen concentration in the active sites *’. C*s Sugar concentration in the active sites *. C'0 Total concentration in the active sites *’. C0 Total concentration in the active sites *. CA Concentration of L-arabinose in the liquid phase. CAOH Concentration of L-arabitol in the liquid phase. Ccacli Estimated concentration of the component i. Cexpi Experimental concentration of the component i. CG Concentration of D-galactose in the liquid phase. CGOH Concentration of D-galactitol in the liquid phase. CH Atomic hydrogen concentration in the liquid phase. Cmean Mean concentration. CS Sugar concentration in the liquid phase. Es Activation energy of the sugar s. KA Adsorption constant for L-arabinose. KG Adsorption constant for D-galactose. KH Adsorption constant for hydrogen. ks Reaction rate constant of the surface reaction between hydrogen and sugars. k's Merged reaction parameter. k''s Merged reaction parameter at constant hydrogen concentration for sugar s. ksi Adsorption constant for component i. mcat Mass of catalyst. Q Objective function. R2 Coefficient of determination. ri Reaction rate of component i. xx rs Reaction rate of sugar s. T Temperature. VL Reaction volume. α Relative reactivity. κ'A Merged parameter of L-arabinose in sugar mixture modeling at constant hydrogen concentration. κ'G Merged parameter of D-galactose in sugar mixture modeling at constant hydrogen concentration. κA Merged parameter of L-arabinose in sugar mixture modeling. κG Merged parameter of D-galactose in sugar mixture modeling. ρB Bulk density. Abbreviations A L-Arabinose. AOH L-Arabitol. EDX Energy dispersive X-ray analysis. FA Furfuryl alcohol. G D-Galactose. GOH D-Galactitol. H Hydrogen. HDP Homogenous deposition precipitation. ICP-OES Inductively coupled plasma atomic emission spectroscopy. IWI Incipient wetness impregnation. PFA Polyfurfuryl alcohol. S Sugar. SEM Scanning electron microscope. SOH Sugar alcohol. SSR Sum of the square residues. TEM Transmission electron microscopy. TPR Temperature-programed reduction. 1. Introduction 21 1. INTRODUCTION 1.1. The Biorefinery Concept The growing concern about the short and long-term environmental, social, and economic consequences caused by climate change is driving humankind towards a more sustainable development [1, 2]. In this context, at the time, 189 countries have signed The United Nations Paris Agreement, which aims to limit global warming from 2.0°C to 1.5 °C [3].The actions required to accomplish the Paris Agreement goals represent enormous scientific, logistic, and political challenges. Despite the advances in terms of technology and the implementation of renewable energy sources such as solar, wind, and biomass, a successful and smooth transition to a carbonneutral economy requires a proper diversification of the feedstock and production processes [4, 5]. Therefore, the use of biomass appears to be a highly promising alternative to tackle the upcoming challenges of the chemical industry through the adoption of a new and sustainable biorefinery concept [6]. The biorefinery is a novel platform based on the conversion of biomass from different sources into high-value products, with the aim of increasing the economic potential by integrating various technologies that allow the use of the by-products generated in different transformation processes [7, 8]. The barriers to implementing this concept are mainly related to factors such as the geographic biomass availability, the chemical diversity of biomass, the economic viability compared to traditional refinery products, and the cultivable land usage, such in case of first-generation biofuels [6, 9]. The so-called second-generation biorefinery is oriented to the utilization of lignocellulosic biomass created from agriculture, forestry, and the alimentary industry, generating chemical compounds from residues [9]. This approach has outstanding advantages such as the wide availability of lignocellulosic materials, which represent 75% of the renewable biomass [10] and the absence of competition for cultivable soil. However, the transformation processes are complex due to the chemically diversified nature of its constituents: cellulose, hemicellulose, lignin, and extractives [5, 11]. Lignocellulosic biomass is composed of 40-50 wt. % of cellulose (glucose-based polymer linked by β1, 4-glycosidic bonds), 16-33 wt. % of hemicelluloses (heteropolymers containing various monomers 1. Introduction 22 of different sugars, such as arabinose, galactose, glucose, mannose, and xylose), and 15-30 wt. % of lignin (complex cross-linked polymer with coniferyl, coumaryl, and sinapyl alcohols as monomeric units). Thus, the elaboration of fuels and chemicals from these materials requires applying thermal, chemical, catalytic, or biological methods to obtain its constituents [12]. From extraction processes combined with chemical treatments, for example, acid hydrolysis, simpler carbohydrates are obtained, such as mono and disaccharides. In this sense, several conversion routes have been proposed to use these compounds as platforms for chemical production. A prime example is glucose as a building block from cellulose and starch, which after a reduction process can be transformed into its respective sugar alcohol, sorbitol, and afterwards, in polyesters, polyamides, and polyurethanes [5]. Figure 1.1. Sugar alcohols from lignocellulosic biomass. 1. Introduction 23 Sugar alcohols are versatile substances with a wide range of applications that have aroused interest in recent years due to their potential to act as precursors for complex molecules and to produce renewable hydrogen or alkanes via aqueous-phase reforming [13]. 1.2. Sugar Alcohols 1.2.1. Applications and Sources Sugar alcohols are polyols with the general formula H2(CH2O)n+1 , which are formed by the reduction of the carbonyl group present in the sugar molecules employing either chemical reagents (e.g. sodium borohydride) or molecular hydrogen in contact with a homogenous or heterogeneous catalyst [4, 5, 14]. The route based on the use of heterogenous catalysts is preferred from an environmental point of view since it avoids the formation of stoichiometric co-products and facilitates the separation processes [5, 11]. Sugar alcohols find their applications in the alimentary, pharmaceutical, and cosmetics industries. The global market size of the sugar alcohols was 3.61 billion USD in 2019 and is projected to reach 6.79 billion USD by 2027, exhibiting an increasing rate of 7.75% within 2020-2027 [15]. Their main applications rely on the alimentary industry as healthier alternatives for sucrose due to their sweet taste and low caloric content, especially in the case of xylitol [16]. Sugar alcohols are also widely used in the production of hand sanitizers, which have had a remarkable demand increase since 2020 [15]. It is noteworthy that some studies have shown that sugar alcohols exhibit significant healthpromoter effects, such as anti-carries and antioxidant activity [17, 18]. Currently, the production of the most important sugar alcohols on the market, sorbitol and mannitol, depends on agricultural resources, such as cassava, corn, and wheat [11]; alternatively, hydrolysis processes have been sought to produce them using cellulose [11, 19–21]. On the other hand, the production of sugars from hemicelluloses has gained much attention because the extraction is an easier task compared to the extraction of cellulose and it requires milder operation conditions; the use of acids, water, steam, or combinations at a moderate temperature range (150190˚C) generally yields to a selective solubilization of hemicelluloses totally or partially hydrolyzed to oligomeric and monomeric sugars [16, 22, 23]. 1. Introduction 30 it. The uptake of the support occurs by capillary pressure differences. In hydrophobic supports such as carbon, the capillary pressure difference becomes negative; hence, the presence of oxygen surface groups (as in the case of HDP) is relevant to enhance the precursor adsorption. In structured catalysts, obtaining a good metal particle dispersion through this method is challenging because of the geometric complexity of these materials [45, 49]. 1.6. Research Strategy The goal of the present research work was to develop a novel open-cell solid foam Ru/C catalyst to study the catalyst activity and the reaction kinetics of the hydrogenation of L-arabinose and Dgalactose and their mixtures to sugar alcohols. To achieve this objective, the following tasks were carried out: ▪ Development of an effective and reproducible carbon-coating method of aluminum foams based on the polymerization of furfuryl alcohol. ▪ Incorporation of a suitable amount of ruthenium on the carbon-coated foams and evaluation of the optimal condition of the incorporation method. ▪ Performance of kinetic hydrogenation experiments with L-arabinose and D-galactose to explore the product selectivity, reactant conversion, and influence of the pressure and temperature on the reaction rate and product distribution. ▪ Conduction of kinetic experiments with sugar mixtures to investigate the interaction of the sugars during the hydrogenation reaction. ▪ Fitting the experimental data to a plausible kinetic model. ▪ Application of catalyst characterization techniques that contribute to the clarification of the results. 2. Experimental 31 2. EXPERIMENTAL 2.1 Catalyst Preparation The catalyst preparation consisted of six general steps: Cutting of the open-cell aluminum foam pieces, anodic oxidation pretreatment, carbon coating, acid pretreatment, ruthenium incorporation (through homogenous deposition precipitation (HDP) or incipient wetness impregnation (IWI)), and ex-situ reduction. An overview of the catalyst preparation process is shown in Figure 2.1. Figure 2.1. Overview of the solid foam catalyst preparation process. Ten catalyst batches were elaborated, in which different preparation parameters were tested and several characterization techniques were applied to obtain an efficient catalyst for the kinetic study of the hydrogenation of sugars. Table 2.1 shows the general information of the catalysts. 2. Experimental 32 Table 2.1. Batch codes and general information of the prepared catalysts. Batch Code Foam Code Initial Mass (Al foam) [g] Anodic Oxidation Pretreatment Ruthenium Incorporation Method C1-HDP1 C1-HDP1-F1 0.4679 No HDP C1-HDP1-F1 0.4934 C2-HDP1 C1-HDP2-F1 0.7970 No HDP C1-HDP2-F2 0.8679 C3-HDP2 C3-HDP2-F1 0.5420 No HDP C3-HDP2-F2 0.4963 C3-HDP2-F3 0.4753 C4-HDP3 C4-HDP3-F1 0.4640 No HDP C4-HDP3-F2 0.4620 C5-HDP4 C5-HDP4-F1 0.5951 No HDP C5-HDP4-F2 0.5863 C5-HDP4-F3 0.5872 C6-IWI1 C6-IWI1-F1 0.5901 No IWI C6-IWI1-F2 0.5773 C6-IWI1-F3 0.6000 C7-IWI1 C7-IWI2-F1 0.4935 No IWI C7-IWI2-F2 0.5495 C8-IWI2 C8-IWI3-F1 0.5544 No IWI C8-IWI3-F2 0.5481 C9-AO-IWI2 C9-IWI3-F1 0.3674 Yes IWI C9-IWI3-F2 0.3763 C10-AO-IWI2R300 C10-IWI3-F1 0.4799 Yes IWI C10-IWI3-F2 0.4464 C10-IWI3-F3 0.5251 2.1.1. Cutting Cylindrical pieces with dimensions of 33 mm length and 11 mm diameter were cut from a pure aluminum foam sheet with a pore density of 40 PPI (Goodfellow Cambridge Ltd) using a diamond hole saw bit (Figure 2.2). The cut foams were sonicated for 15 min in deionized water and for 15 min in acetone, then oven-dried for 2 hours at 70˚C, and overnight at room temperature. 2. Experimental 33 Figure 2.2. Cutting of aluminum foams. 2.1.2. Anodic Oxidation Pretreatment In order to enhance the carbon adhesion to the foams, the surface of some Al supports was pretreated as follows: One cleaned foam with the above-mentioned dimensions was attached to a thin platinum flat strip using PTFE tape, then connected to the anode (working anode) of a power supply (Autolab PGSTAT100N) with a rectangular 4cmx9cm aluminum plate (the immersed area was of 18 cm2) connected to the cathode (counter electrode). Both were immersed in the electrolyte solution keeping 2.5 cm distance. The electrolyte solution consisted of 100 mL of 1.6 M sulfuric acid (Sigma-Aldrich; 96%), 60 g/L of aluminum sulfate hexadecahydrate (Fluka; 98%) were also added to control the dissolution of aluminum during the anodization process [60, 61]. The temperature was set to 40˚C using a thermostat (Grant GR150 GP200) by circulating oil in the jacketed vessel containing the solution. A magnetic stirrer at the bottom of the vessel was utilized to homogenize the temperature. A constant electrical current of 2 A was circulated through the system during 1 h and the voltage was monitored with the General-Purpose Electrochemical System (GPES) version 4.1 software. Thereafter, the foam was taken out from the acid and washed by dipping in deionized water. The same solution was used to anodize three different foam pieces. The obtained foams were ovendried at 70˚C for 30 minutes and afterwards, calcined at 600 °C for 4h. Figure 2.3 shows the setup of the anodic oxidation procedure. 2. Experimental 34 Figure 2.3. Experimental arrangement of the anodic oxidation process. The required electrical current (2 A) was estimated using the geometrical surface area information and the optimal current density reported by Lali et al. (2015) [61]. On the other hand, the time and the electrolyte concentration were chosen by carrying out experiments and evaluating qualitatively the physical stability and homogeneity of the obtained oxide layers. 2.1.3. Carbon Coating The carbon-coated foam batches consisted of two or three pieces, which were attached to a crossed blade stirrer shaft using thin stainless-steel wires and introduced in a 300 mL metallic vessel provided with an electric band heater (Ogden Mighty-Tuff MT-03015-0424). Thereafter, 136.2 g of furfuryl alcohol (Sigma Aldrich; 98 wt.%), 0.42 g of oxalic acid dihydrate (Sigma Aldrich; 99.5 wt.%), and 16.7 g of distilled water were poured into the vessel. The heating rate of the electrical band was adjusted at 2 K∙min-1 from room temperature (about 20˚C) to 120˚C using a temperature process controller (The CAL 9500P). A Heidolph RZR 2021 mechanical stirrer was utilized to rotate the foams during the polymerization process; two different stirring speeds were tested (200 rpm and 700 rpm). The mixture under the above-described conditions kept between 20˚C and 110˚C within 55-60 min. When it reached 110˚C, the water evaporation began, the viscosity and temperature increased sharply due to the reaction heat, therefore, the automatic heating was turned off and the temperature was adjusted manually to reach 120˚C within 45-60 min in such a way that the water 2. Experimental 35 slowly vaporized. Once the polymerization process was finished, the excess of PFA was removed by centrifuging the foams at 1000 rpm for 5 min. The experimental setup of the PFA coating procedure is shown in Figure 2.4. Figure 2.4. Experimental setup for PFA coating. The PFA coated foams were pyrolyzed in a furnace (Carbolite CTF 12/100/900) heating at 5 K∙min-1 up to 550 °C and held for 5 h in a nitrogen stream with a flow rate of 1 L∙h-1. Subsequently, the carbon coating was activated in an oxygen stream of 2 L∙h-1, heating from room temperature at 5 K∙min-1 up to 380°C and held for 2h. The experimental conditions and the obtained carbon loads are presented in Table 3.2. 2.1.4. Ruthenium Incorporation The carbon-coated foams were pretreated in a 3 wt.% nitric acid (Sigma-Alrich;70 wt.%) solution for 2 h. The acid-pretreated foams were then washed in deionized water, oven-dried at 70˚C for 2h and overnight at room temperature. The ruthenium incorporation was carried out testing two different methods: homogeneous deposition precipitation (HDP) and incipient wetness impregnation (IWI). 2. Experimental 36 Homogenous Deposition Precipitation (HDPE) The used homogeneous deposition precipitation procedure was modified from the method proposed by Lali et al.[32] and Najarnezhadmashhadi et al. [34] by suitably tuning the experimental conditions to obtain a proper pH evolution and a catalyst with a feasible Ru content. A volume of 500 mL of Ru precursor solution with a nominal load as specified in Table 2.3 for each catalyst was prepared by diluting in deionized water an adequate amount of Ru(III) nitrosyl nitrate (Sigma-Alrich; 1,4 wt.% Ru, diluted in nitric acid). Two carbon-coated and acid-treated foam pieces were placed inside a vessel containing the Ru precursor solution; a pH meter (InoLab 7310) was also introduced to monitor the pH during the entire process. The vessel was immersed in an oil bath under a hot plate (Velp Scientific). Two magnetic stirrers (one in the solution and one in the oil bath) were utilized to keep the concentration and the temperature homogeneous, and a stream of nitrogen was bubbled through the precursor solution to avoid the accumulation of carbon dioxide [30]. The system was heated until a constant temperature of 80˚C was reached, after which urea (SigmaAlrich; 99 wt.%) was added to get a molar ratio of urea-to-Ru as is specified in Table 2.2. for each catalyst batch. The experimental arrangement is shown in Figure 2.5. Figure 2.5. Experimental setup for HDP. 2. Experimental 37 Table 2.2. Ruthenium incorporation conditions. Batch Code Foam Code Final Mass of Carbon [g] Ru Deposition Method Ru Nominal Load Based on Carbon [%] Urea-to-Ru Molar Ratio (for HDP) Ex-situ Reduction Conditions C1-HDP1 C1-HDP1-F1 0.0635 HDP 170% 5 450˚C for 2h C1-HDP1-F1 0.0618 C2-HDP1 C1-HDP2-F1 0.1212 HDP 100% 5 450˚C for 2h C1-HDP2-F2 0.0935 C3-HDP2 C3-HDP2-F1 0.3472 HDP 10% 20 450˚C for 2h C3-HDP2-F2 0.3259 C3-HDP2-F3 0.3075 C4-HDP3 C3-HDP3-F1 0.0705 HDP 5% 1:20 at t=0min and 1:20 at t=400min 450˚C for 2h C3-HDP3-F2 0.0799 C5-HDP4 C5-HDP4-F1 0.3342 HDP 30% 40 450˚C for 2h C5-HDP4-F2 0.3746 C5-HDP4-F3 0.3576 C6-IWI1 C6-IWI1-F1 0.0043 IWI 24% - 450˚C for 2h C6-IWI1-F2 0.0040 C6-IWI1-F3 0.0024 C7-IWI1 C7-IWI2-F1 0.0384 IWI 24% - 450˚C for 2h C7-IWI2-F2 0.0311 C8-IWI2 C8-IWI3-F1 0.2989 IWI 4% - 450˚C for 2h C8-IWI3-F2 0.3504 C9-AO-IWI2 C9-IWI3-F1 0.3042 IWI 6% - 450˚C for 2h C9-IWI3-F2 0.3063 C10-AO-IWI2R300 C10-AO-IWI2R300-F1 0.5128 IWI 4% - 300˚C for 5h C10-AO-IWI2R300-F2 0.5371 C10-AO-IWI2R300-F3 0.5879 Incipient Wetness Impregnation (IWI) Two concentrations of Ru(III) nitrosyl nitrate (diluted in nitric acid solution; Sigma-Aldrich) were tested for the incorporation of ruthenium in the foam catalyst: a 1.4 wt.% Ru solution for catalysts C6-IWI1 and C7-IWI1, and a 0.6 wt.% Ru solution for catalysts C8-IWI2, C9-AO-IWI2 and C10-AOIWI2-R300. The precursor solution was dripped to distribute it as homogeneously as possible on the surface of the carbon-coated foams using an adequate number of impregnation steps (avoiding overflowing) 2. Experimental 38 until reaching the nominal load of each batch as reported in Table 2.2. For catalysts labeled as IWI1 the amount of precursor solution per step was approximately 0.10 g and 0.25 g for catalysts labeled as IWI2. After each impregnation step, the foams were dried in an oven at 110 ° C for 24 h. 2.1.5. Ex-Situ Catalyst Reduction The ex-situ reduction of the catalysts was carried out in a furnace (Carbolite CTF 12/100/900), using 1 L∙h-1 of hydrogen stream under the conditions of time and temperature described in Table 2.2. The reduction temperature of 450˚C is based on previous works that reported Ru/C catalysts prepared through the HDP method, in which ruthenium hydroxide and ruthenium oxide species are expected to be formed [30, 58]. On the other hand, the temperature of 300 ˚C is based of TPR measurements gauged in this work with catalyst C10-AO-IWI3-R300. 2.2. Kinetic Experiments The kinetic experiments were carried out in a 0.3 L laboratory-scale semi-batch reactor (Parr 4561) provided with baffles, a sampling line with a sintered filter (7 µm), a heating jacket, a temperature and stirring rate controller (Parr 4843), a cooling coil, a pressure display module (Parr 4843) and a bubbling chamber. Two foam catalyst pieces were mounted at the endpoint of the mechanical agitating shaft to work as the stirrer during the experiments (Figure 2.6). Preliminary experiments were conducted in a 1:1 molar ratio solution of L-arabinose (Sigma-Aldrich; 99 wt. %) to D-galactose (Across Organics; 99 wt. %) with a concentration of 0.13 M at 20 bar and 120˚C to test the prepared catalysts. A set of individual kinetic experiments were performed with L-arabinose and D-galactose at three different temperatures (90°C, 100°C, and 120°C) and two hydrogen pressures (20 and 40 bar) in the presence of the catalyst C8-IWI3, using an initial sugar concentration of 0.13 M. To study the interaction of the sugars during the hydrogenation reaction, a series of experiments were conducted using binary mixtures of D-galactose and L-arabinose in the presence of the catalyst C10-AO-IWI3-R300. The reaction conditions were 120°C and 20 bar, varying the molar ratio of Dgalactose to L-arabinose (ratios: 0.5, 1, and 5). 2. Experimental 39 Figure 2.6. Overview of the setup for sugar hydrogenation experiments. Prior to the kinetic experiments, the reactor was purged with argon and hydrogen, the foam catalyst was in-situ reduced for 2 hours at 5 bar pressure of hydrogen and 120˚C. Once the catalyst was reduced, 130 mL of sugar solution were pumped to the bubbling chamber and purged with argon, afterwards, with hydrogen during 15 min each, then the temperature was set to the required one, the hydrogen pressure adjusted, and the hydrogen-saturated solution injected to the reactor; hence the experiments started under the desired conditions of temperature and hydrogen pressure. A stirring rate of 600 rpm was utilized for all the experiments. Samples were withdrawn from the reactor to measure the concentration of reactants and products. The concentration analysis of the sugars and sugar alcohols was conducted using a HighPerformance Liquid Chromatograph (HITACHI Chromaster HPLC) equipped with a refractive index (RI) detector (HITACHI 5450 RI Detector). A Biorad HPX-87C carbohydrate column was used with 1.2 mM CaSO4 solution (0.5 mL∙min-1 flow rate) as the mobile phase, the temperature of the oven was 70˚C and an injection volume of 10 μL was utilized. The calibration data are shown in Appendix I. 3. Results and discussion 46 In general, the carbon made from the dark foamy polymer exhibited better properties to be used as catalyst support: higher surface area, more homogeneous coverage, and better resistance to acids. Figure 3.5.Obtained PFA coating: (a) foamy dark polymer obtained when sudden rising of temperature takes place, (b) golden polymer obtained when not sudden rising of the temperature takes place. Despite that the exact reaction mechanism and products obtained in the polymerization of furfuryl alcohol remain uncertain [12, 13, 14], it is widely accepted that under acid conditions, the main product is a linear aliphatic structure of repeating units of polyfurfuryl alcohol linked by methylene bridges, produced by the condensation of the OH groups [14, 15]. Then, as the branching and crosslinking of the linear PFA take place, the mixture becomes darker and more viscous, and the water vaporizes due to the exothermic character of these phase reactions, creating cavities on the polymer, which enables it to become a good active carbon precursor [68]. Figure 3.6. Temperature pattern during PFA coating when a foamy dark PFA coverage is obtained (catalyst C8-IWI3). 0 20 40 60 80 100 120 140 014 28 43 57 72 86 100 115 Temperature[°C] Time [min] Temperature under automatic control. Sudden temperature rise/water evaporation. . Temperature under manual control (110˚C-120˚C). 3. Results and discussion 47 As Figure 3.7 shows, the carbon coating obtained from the less crosslinked PFA looks inhomogeneous and has a considerable amount of uncovered areas compared to the carbon from the foamy PFA. Additionally, the anodized foams presented a carbon coating with fewer cracks and improved cohesion due to the roughness of their surface. (a) (b) (c) Figure 3.7. Surface structure of a carbon-coated foam substrate: (a) C2-HDP2 (12 wt.% carbon), obtained from golden-colored PFA, (b) C7-IWI2 (50 wt.% carbon), obtained from foamy dark PFA, (c) C10-AO-IWI2R300 (50 wt.% carbon), preanodized and obtained from foamy dark PFA. Another significant parameter identified was the rotation speed. The ruthenium incorporation experiments indicated that a carbon content above 40 wt.% was required to get enough active metal on the support. Thus, a rotation speed of 200 rpm was used, resulting in higher carbon loads at similar polymerization conditions (see Table 3.2). 3. Results and discussion 48 3.1.3. Ruthenium Incorporation Homogenous Deposition Precipitation Results The homogeneous deposition precipitation technique has been reported in the literature as a suitable method to incorporate ruthenium on carbonaceous supports [3, 4, 17] using Ru(III) nitrosyl nitrate as precursor solution due to the feasibility of this reagent to generate small-sized nanoparticles [57]. In the HDP, a precipitant agent, such as urea, in this case, decomposes and releases OHions that interact with the precursor solution, precipitating Ru(OH)3 as the pH smoothy increases to a constant value of 7. Figure 3.8. Evolution of pH during HDP at different conditions of Ru nominal load based on carbon (NL) and urea-to-Ru molar ratio (r). The presence of aluminum in the support and nitric acid in the precursor solution represented additional difficulties for the incorporation of an adequate amount of active metal through this method. Different Ru nominal loads, urea-to-Ru molar ratios, and carbon loads on the support were investigated. 0 1 2 3 4 5 6 7 8 0200 400 600 800 1000 1200 1400 pH Time [min] C5-HDP4, r=40, NL=30% C4-HDP3, r1=20 and r2=20, NL=5% C3-HDP2, r=20, NL=10% C2-HDP1, r=5, NL=100% C1-HDP1, r=5, NL=170% 3. Results and discussion 49 It can be seen in Figure 3.8 that the solution goes through a first plateau of pH attributable to the neutralization of nitric acid, since this behavior is not observed in HDP using only Ru(NO)(NO3)3 [3,17]. If the amount of added urea is not enough to neutralize the acid, the deposition process is not completed in 24 h, hence the prolonged exposure to acidic conditions causes a damage to the Al support, and in some cases, the precipitation of the ruthenium on the aluminum substrate (Figure 3.9). Point Ru [wt. %] C [wt. %] Al [wt. %] O [wt. %] 1 - 59.58 1.53 37.90 2 - 58.31 1.85 32.92 3 12.69 4.70 28.28 49.51 4 12.54 5.83 29.58 29.58 Figure 3.9. SEM image and elemental analysis (by EDX) of catalyst C3-HDP2 displaying cracks on the aluminum support and incorporation of Ru on the aluminum phase. Considering the above-mentioned results, the catalyst C5-HDP4 was elaborated utilizing the found optimal conditions: a urea-to-Ru molar ratio of 40, coated foams with a high carbon content (~ 40 wt. %), and a nominal load of 30 wt.% Ru based on carbon. The result was a catalyst considerably more active, with a Ru load of 0.5 wt. % and a small particle size distribution (~80% of the particles smaller than 4 nm), yielding an average size of 3.3 nm (Figure 3.10). Incipient Wetness Impregnation Incipient wetness impregnation was used to incorporate Ru on the surface of the carbon-coated foams as an alternative method to HDP aiming to increase the active metal content of the catalyst. Two concentrations of Ru(III) nitrosyl nitrate were tested: 1.4 wt.% Ru. and 0.6 wt.% Ru, the amount of precursor solution per step was established as the maximum liquid volume that could be uptaken by the support without overflowing. Two fundamental aspects for the preparation of this type of catalyst were confirmed as a result of the IWI tests; the carbon obtained from the foamy PFA (see Section 3.1.2) exhibited superior adsorption of the precursor solution compared to the carbon from the golden-colored PFA, and the 1 2 C V 3 4 C V 3. Results and discussion 50 presence of nitric acid in the precursor solution represents a risk for the aluminum structure if the carbon load is insufficent. Therefore, the most active catalysts (C8-IWI3 and C10-IWI3) obtained in this work were elaborated using supports with a high carbon content (~50%), a precursor solution with a concentration of 0.6 wt.% Ru, and a nominal load of 4 wt. % Ru based on carbon, yielding a 1.1 wt.% of Ru content with an average nanoparticle size of 3.7 nm, and 70% of the particles smaller than 4 nm (Figure 3.11). 3.1.4. Preliminar Catalyst Tests The catalysts prepared by the optimal conditions of both HDP and IWI were tested for the hydrogenation of a 1:1 molar ratio of D-galactose to L-arabionse solution with a concentration of 0.13 M under the conditions discribed in Section 2.2. Table 3.3 shows a comparison between the prepared catalysts. Figure 3.10. TEM images of catalyst C5-HDP4 and particle size distribution. Average Size=3.3 nm Standard Deviation=1.5 nm 3. Results and discussion 51 Figure 3.11. TEM images of catalyst C10-OA-IWI3-R300 and particle size distribution. Considering the high activity of catalyst C8-IWI3, this was selected to perform the kinetic study of the individual sugars, and catalyst C10-OA-IWI3-R300, made utilizing the same carbon coating and Ru incorporation conditions, was used in the kinetic study of the sugar mixtures. Table 3.3. Comparison of the physical properties and reactivity of the prepared catalyst Catalyst Total Carbon Mass (two foams) [g] Ru load [wt. %] Arabinose conversion in 4h [%] Galactose conversion in 4h [%] Average Ru particle size [nm] C4-HDP3 0.1504 0.23 2% 4% 3.3 C5-HDP4 0.7322 0.52 54% 27.4% 3.3 C8-IWI3 1.2668 1.12 98% 97% 3.7 Average Size=3.6 nm Standard Deviation=2.1 nm 3. Results and discussion 52 3.1.5. Effect of the Reduction Conditions and Deactivation As mentioned in Section 2.1.5., catalyst C8-IWI3, used for the hydrogenation of individual sugars, was ex-situ reduced in a hydrogen stream (1 L∙h-1) at 450˚C for 2h. An induction period (~ 1 h) was observed when using this catalyst in the first two experiments (Figure 3.12), and it displayed an increase of activity in the subsequent experiments until reaching a considerably stable behavior in the fourth experiment; after approximately 32 h of use (including in-situ reduction time). (a) (b) Figure 3.12. Induction behavior of Catalyst C8-IW3 during hydrogenation of 1:1 D-galactose to L-arabinose 0.13 M solution at 120˚C and 20 bar. (a) L-arabinose conversion vs time, (b) D-galactose conversion vs time. Induction periods are relatively frequent in ruthenium catalysts, especially for liquid phase reactions [69–72], and are generally explained in terms of surface oxides, which are reduced during the time of reaction, forming more active metallic Ru0. Temperature-programmed reduction measurements displayed in Figure 3.13 were conducted with catalyst C10-AO-IW3-R300 to establish more adequate reduction conditions, thus avoiding induction periods. A single hydrogen consumption peak appeared at 245˚C, attributable to the reduction of ruthenium oxides [73, 74]. Therefore, the new reduction temperature was set at 300˚C and the reduction time was increased to 5 h (temperature ramp = 3˚C∙min-1) for catalyst C10-AOIWI3-R300, in which no induction periods were observed. 0% 20% 40% 60% 80% 100% 050 100 150 200 250 300 350 400 L-arabinose Conversion [%] Time (min) First Experiment Second Experiment Third Experiment Fourth Experiment 0% 20% 40% 60% 80% 100% 050 100 150 200 250 300 350 400 D-galactfose Conversion [%] Time [min] First Experiment Second Experiment Third Experiment Fourth Experiment 3. Results and discussion 53 Figure 3.13. Hydrogen-TPR profiles of catalyst C10-IWI-R300 (before ex-situ reduction). On the other hand, after 96 h of use, catalyst C8-IWI3 presented a considerable deactivation for the hydrogenation of both L-arabinose and D-galactose (Figure 3.14). TPR and TEM measurements were performed to investigate the possible causes of deactivation. (a) (b) Figure 3.14. Deactivation of catalyst C8-IW3 during hydrogenation of (a) L-arabinose, (b) D-galactose at 120˚C and 20 bar. Figure 3.15 shows the TEM micrograph of the spent catalyst. A substantial agglomeration of particles took place, resulting in the increase of the average size, from 3.6 nm to 5.2 nm. Some authors have reported agglomeration Ru nanoparticles after hydrogenation reactions of sugars [75, 76] and other chemicals [56]. 0200 400 600 800 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 Temperature [˚C] TCD Signal [a.u] 0% 20% 40% 60% 80% 100% 0100 200 300 400 Conversion [%] Time [min] Active Catalyst Spent Catalyst 0% 20% 40% 60% 80% 100% 0100 200 300 400 Conversion [%] Time [min] Active Catalyst Spent Catalyst 3. Results and discussion 54 Figure 3.15.TEM images of catalyst C8-IWI3 after 100 h of use and particle size distribution. Other studies have suggested the formation of Ru(OH)x species during liquid phase reactions in the presence of water [30, 58]. Nevertheless, TPR measurements carried out with the spent catalyst did not shown any significant hydrogen consumption peak within the temperature range of 400-500˚C associated with the reduction of these species [32, 56, 58]. Furthermore, Simakova et al. [29] found that the rates of hydrogenation of L-arabinose and Dgalactose on Ru/C catalysts are highly influenced by the metal cluster size, with a maximum turnover frequency at 3 nm (approximately the particle size of the fresh catalyst), and that decays rapidly as the size increases, indicating that the increased particle size in our catalyst is the main cause of deactivation. Although more research is needed to establish certainly the causes of the deactivation of this kind of catalyst, which also would allow developing suitable regeneration methods. In general, the prepared catalyst exhibited a good selectivity, activity, and stability similar [77] and even superior to other Ru/C catalysts described in the literature [75]. Average Size=5.2 nm Standard Deviation=2.9 nm 3. Results and discussion 55 3.2. Kinetics Results 3.2.1. Individual Sugar Results Selectivity and Conversion Individual sugar hydrogenation experiments were performed at 20 bar and different temperatures (90, 100, 20˚C) on the prepared Ru/C foam catalyst. The overall selectivity towards sugar alcohols was higher than 98% in all the cases, whilst the conversion ranged 60-98% depending on the temperature. L-arabinose presented higher reactivity, as can be seen in Table 3.4. Table 3.4 Selectivity and conversion of L-arabinose and D-Galactose after 6 h of reaction at 20 bar and different temperature (individual sugar hydrogenation). Temperature D-galactose L-arabinose Conversion (%) Selectivity (%) Conversion (%) Selectivity (%) 90 60 100 85 99 100 73 99 97 99 120 97 98 98 98 The yield of by-products was negligible (1-5%) in all the experiments and dependent on the operation conditions; higher pressures and higher temperatures resulted in the formation more byproducts, which could be detected by inspecting the chromatograms. Temperature and Pressure Influence Temperature showed a significant influence on the reaction rate for both sugars, as can be seen in Figure 3.16. The effect of temperature was successfully described by Arrhenius's law, the estimated values for the apparent activation energy of L-arabinose and D-galactose were 44 kJ∙mol-1 and 51 kJ∙mol-1, respectively (see section 4.1.2). 4. Kinetic modeling 62 The parameters obtained from the lineal regression analysis were used as initial values for the simultaneous optimization of the experimental data at different temperatures. The coefficient of correlation for the model was computed from equation 4.18, varying from 0-100%, considering that a good degree of explanation is achieved for values of R2 in the range of 95-99%. R2=(1- ∑(CExp,i-CCalc.i)2 n i=1 ∑(CExp,i-Cmean)2 n i=1 )∙100 (4.18) The model-fitting results are displayed in Figure 4.3 and Figure 4.4, demonstrating how the proposed model very successfully described the experimental concentration profiles during the hydrogenation reaction on the prepared Ru/C foam catalyst at different temperatures. (a) (b) (c) Figure 4.3. Modeling results for L-arabinose hydrogenation at 20 bar and different temperatures: (a) 90˚C, (b) 100˚C and (c) 120˚C. 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 050 100 150 200 250 300 350 400 Concentration [mol∙L-1] Time [min] L-arabinose (Real) L-arabitol (Real) L-arabinose (Model) L-arabitol (Model) 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 050 100 150 200 250 300 350 400 Concentration [mol∙L-1] Time [min] L-arabinose (Real) L-arabitol (Real) L-arabinose (Model) L-arabitol (Model) 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 050 100 150 200 250 300 350 400 Concentration [mol∙L-1] Time [min] L-arabinose (Real) L-Arabitol (Real) L-arabinose (Model) L-arabitol (Model) 4. Kinetic modeling 63 (a) (b) (c) Figure 4.4.Modeling results for D-galactose hydrogenation at 20 bar and different temperatures: (a) 90˚C, (b) 100˚C and (c) 120˚C. 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 050 100 150 200 250 300 350 400 Concentration [mol∙L-1] Time [min] D-galactose (Real) D-galactitol (Real) D-galactose (Model) D-galactitol (Model) 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 050 100 150 200 250 300 350 400 Concentration [mol∙L-1] Time [min] D-galactose (Real) D-galactitol (Real) D-galactose (Model) D-galactitol (Model) 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 050 100 150 200 250 300 350 400 Concentration [mol∙L-1] Time [min] D-galactose (Real) D-galactitol (Real) D-galactose (Model) D-galactitol (Model) 4. Kinetic modeling 64 The parameters for the L-arabinose and D-galactose hydrogenation yielded by regression using the proposed model are listed in Table 4.2. High values of R2 and small values of the objective function were obtained in both cases, which indicates the good performance of the model. The apparent activation energies obtained in this work are in the order of magnitude of values reported in previous research [35, 36]. The adsorption parameter was higher for L-arabinose than for D-galactose, which was expected due to the smaller molecular size of L-arabinose compared to D-galactose. Table 4.2. Kinetic parameters estimated for L-arabinose and D-galactose in individual hydrogenation experiments. Sugar As' [L∙gRu-1∙min-1 mol-1] Es [J∙mol-1] Ks [L mol-1] R2 [%] SRS1 [mol2∙L-2] L-arabinose 181393.97 44220.30 8.78 99.84 0.0008 D-galactose 663007.51 51738.46 1.56 99.89 0.0006 1Sum of the residual squares. 4.1.3. Sensitivity Analysis The sensitivities of the estimated parameters were evaluated by plotting the parameter values against the corresponding objective function equation 4.17 while keeping the other values constant and equal to best-fitted from the model. The sensitivity plots of the objective function for the parameters AS', ES, and KS are shown in Figure 4.5. The presence of sharp valleys in these graphs confirms that the parameters are well-defined, and all of them have an important contribution to the overall model. 4. Kinetic modeling 65 (a) (b) (c) (d) (e) (f) Figure 4.5. Sensitivity analysis of fitted parameters: (a) EA, (b) EG, (c) AA', (d) AG', (e)KA, (f) KG. -0.01 0.04 0.09 0.14 0.19 0.24 0.29 0.34 0.39 30000 40000 50000 60000 Objetive Function [mol2∙L-2] EA[J∙mol-1] -0.01 0.04 0.09 0.14 0.19 0.24 0.29 45000 50000 55000 60000 Objetive Function [mol2∙L-2] EG[J∙mol-1] 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 80000 180000 280000 380000 480000 Objetive Function [mol2∙L-2] A'A[L∙gRu-1∙min-1 mol-1] 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 300000 600000 900000 1200000 1500000 Objetive Function [mol2∙L-2] A'G[L∙gRu-1∙min-1 mol-1 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0 5 10 15 20 Objetive Function [mol2∙L-2] KA [L∙mol-1] 0 0.0005 0.001 0.0015 0.002 0123 Objetive Function [mol2∙L-2] KG [L∙mol-1] 4. Kinetic modeling 66 4.2. Modeling of Sugar Mixtures 4.2.1. Model Hypotheses The hypotheses formulated in Section 4.1.1 regarding the reactor operation conditions are also valid to model the experimental data obtained from the sugar mixtures experiments: isothermal batch reactor, constant concentration of hydrogen during the reaction time, no mass transfer limitations, a constant volume of the reaction medium, and fixed catalyst mass are the main hypotheses for the quantitative modelling. No by-products yields were considered, given the high selectivity observed in the experimental data (see Section 3.3.2.). Thus, the reactions were assumed to proceed towards the exclusive formation of sugar alcohols. As in the case of individual sugar experiments, a non-competitive sugar-hydrogen adsorption mechanism was assumed, so the sugar molecules and hydrogen were adsorbed in different active sites. Moreover, the catalyst surface was supposed as ideal, so the Langmuir isotherm is sufficient to describe the adsorption processes. Hydrogen was presumed to be adsorbed in the dissociated form, and the adsorption of the products was discarded. Additionally, it was assumed that the adsorption of L-arabinose was stronger than D-galactose, so the adsorption of the latter was neglected. 4.2.2. Derivation of Rate Expressions for the Kinetic Model The liquid-phase mass balances for the batch reactor introducing the above-presented assumptions are given by equation 4.19-22, where A=L-arabinose, G=D-galactose, AOH=L-arabitol, GOH=Dgalactitol, and the bulk density is defined as ρB=mCat VL. dCA dt =-rA∙ρB (4.19) dCAOH dt =rA∙ρB (4.20) dCG dt =-rG∙ρB (4.21) dCGOH dt =rG∙ρB (4.22) 4. Kinetic modeling 67 The reaction mechanism based on the hypotheses presented in the previous section is shown in Figure 4.6, where * denotes an active site devoted to sugar adsorption (L-arabinose or D-galactose), whilst *’ is a site for hydrogen adsorption. A+* K𝐴 ↔A* G+* K𝐺 ↔G* H2+2*′K𝐻 ↔ 2H*′ A* +2H*′kA → AOH+*+*′ G* +2H*′kG → GOH+*+*′ Figure 4.6 Reaction mechanism for the sugar mixtures hydrogenation. Consequently, the adsorption quasi-equilibria for L-arabinose, D-galactose and hydrogen are defined by equations 4.23-25. C*A=KA∙CA∙C* (4.23) C*G=KG∙CG∙C* (4.24) C*'H=√KH∙CH∙C*' (4.25) The site balances for sugar and hydrogen adsorption can be written as C*A+C*G+C*=C0 (4.26) C*'H+C*'=C′0 (4.27) Where 𝐶0 and 𝐶′0 denote the total concentration of the adsorption sites available for sugar and hydrogen, respectively. Substituting the quasi-equilibria expressions (equations 4.23-4.25) in the site balances (equations 4.26 and 4.27), the concentrations of vacant sites are C*=C0 1+K𝐴∙C𝐴+K𝐺∙C𝐺 (4.28) C*'=C'0 1+√KH∙CH (4.29) Considering that the surface reactions between the adsorbed sugar molecules and hydrogen are the rate-determining steps, the rate equations for L-arabinose, rA and D-galactose, rG becomes rA=kA∙C*A∙C*'H2 (4.30) rG=kG∙C*G∙C*'H2 (4.31) 4. Kinetic modeling 68 The expressions for C* and C*' are then inserted in the rate equations, which yields rA=kA∙KA∙KH∙C0∙C′02∙CA∙CH (1+KA∙CA+KG∙CG)∙(1+√KH∙CH)2 (4.32) rG=kA∙KA∙KH∙C0∙C′02∙CG∙CH (1+KA∙CA+KG∙CG)∙(1+√KH∙CH)2 (4.33) The merged parameters are defined as κA=kA∙KA∙KH∙C0∙C′02 (4.34) κG=kG∙KG∙KH∙C0∙C′02 (4.35) The reaction rate equations obtain the following forms rA=κA∙CA∙CH (1+KA∙CA+KG∙CG)∙(1+√KH∙CH)2 (4.36) rG=κG∙CG∙CH (1+KA∙CA+KG∙CG)∙(1+√KH∙CH)2 (4.37) Furthermore, since the hydrogen pressure was kept constant during all the experiments, the term C𝐻 (1+√KH∙CH)2 is constant, so equations 4.34 and 4.35 can be simplified to rA=κ′A∙CA (1+KA∙CA+KG∙CG) (4.38) rG=κ´G∙CG (1+KA∙CA+KG∙CG) (4.39) Finally, including the hypothesis about the stronger adsorption of L-arabinose ( K𝐴>>K𝐺) rA=κ′A∙CA (1+KA∙CA) (4.40) 4. Kinetic modeling 69 rG=κ′𝐺∙CG (1+KA∙CA) (4.41) 4.2.3. Parameter Estimation The parameters included in equations 4.38 and 4.39 were obtained by regression analysis, minimizing the sum of the residual squares (equation 4.17) using the Nelder-Mead optimization method, while the ordinary differential equations (ODEs) were solved through the LSODA solver in Python. The coefficient of regression was computed as defined by equation 4.18. The estimated parameters are listed in Table 4.3. As can be observed from the table, a higher value was obtained for the objective function compared with the estimations conducted with individual sugars. Table 4.3. Kinetic parameters estimated for L-arabinose and D-galactose in mixtures experiments. Parameter Value R2 [%] SSE [mol2∙L-2] 𝛋′𝐀 0.119 99.45 0.0012 𝛋′𝐆 0.090 KA 12.75 Figure 4.7-9 display the concentration profiles of the reagents and products obtained from the model compared with the experimental data for the hydrogenation of mixtures of L-arabinose and D-galactose at 120˚C and 20 bar and different D-galactose to L-arabinose molar ratios. The results revealed that the model correctly follows the tendency of the experimental data. Nevertheless, some deviations are noticeable especially in case of L-arabinose and L-arabitol, which can be ascribed to systematic errors when determining the concentrations due to the overlapping of peaks in the chromatograms (see Appendix IV). 4. Kinetic modeling 70 (a) (b) Figure 4.7. Modeling sugar mixtures hydrogenation results at 120˚C and 20 bar. Ratio=0.5: (a) L-arabinose, and (b) D-galactose. (a) (b) Figure 4.8. Modeling sugar mixtures hydrogenation results at 120˚C and 20 bar. Ratio=1: (a) L-arabinose, and (b) D-galactose. 0.00 0.02 0.04 0.06 0.08 050 100 150 200 250 300 350 400 450 Concentration [mol∙L-1] Time [min] L-arabinose (Real) L-arabitol (Real) L-arabinose (Model) L-arabitol (Model) 0.00 0.01 0.02 0.03 0.04 050 100 150 200 250 300 350 400 450 Concentration [mol∙L-1] Time [min] D-galactose (Real) D-galactitol (Real) D-galactose (Model) D-galactitol (Model) 0.00 0.02 0.04 0.06 050 100 150 200 250 300 350 400 450 Concentration [mol∙L-1] Time [min] L-arabinose (Real) L-arabitol (Real) L-arabinose (Model) L-arabitol (Model) 0.00 0.02 0.04 0.06 050 100 150 200 250 300 350 400 450 Concentration [mol∙L-1] Time [min] D-galactose (Real) D-galactitol (Real) D-galactose (Model) D-galactitol (Model) 4. Kinetic modeling 71 (a) (b) Figure 4.9.Modeling sugar mixtures hydrogenation results at 120˚C and 20 bar. Ratio=5: (a) L-arabinose, and (b) D-galactose. The sensitivity plots of the objective function for the parameters 𝛋’𝐀, 𝛋’G, and KA are shown in Figure 4.10. The presence of sharp valleys in these graphs suggests that the parameters are well-defined, and all of them have an important contribution to the overall model. Figure 4.10. Sensitivity analysis of fitted parameters: (a) 𝛋𝐀, (b) 𝛋G, (c) KA. 0.000 0.005 0.010 0.015 0.020 050 100 150 200 250 300 350 400 450 Concentration [mol/L] Time [min] L-arabinose (Real) L-arabitol (Real) L-arabinose (Model) L-arabitol (Model) 0.00 0.03 0.06 0.09 050 100 150 200 250 300 350 400 450 Concentration [mol/L] Time [min] D-galactose (Real) D-galactitol (Real) D-galactose (Model) D-galactitol (Model) 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0 0.1 0.2 0.3 0.4 Objetive Function [mol2∙L-2] 𝛋'G[J∙mol-1] 0 0.005 0.01 0.015 0.02 0.025 0 0.1 0.2 0.3 0.4 Objetive Function [mol2∙L-2] 𝛋'𝐀[J∙mol-1] 0 0.001 0.002 0.003 0.004 0.005 0.006 0.007 010 20 30 40 Objetive Function [mol2∙L-2] KA [L∙mol-1] 5. References 78 6. REFERENCES 1. Masson-Delmotte, V., P. Zhai, H.-O. Pörtner, D. Roberts, J. Skea, P.R. Shukla, A. Pirani, W. Moufouma-Okia, C. Péan, R. Pidcock, S. Connors, J.B.R. Matthews, Y. Chen, X. Zhou, M.I. Gomis, E. Lonnoy, T. Maycock, M. Tignor, & T. W. (eds. )]. (2019). IPCC, 2018: Global Warming of 1.5°C. An IPCC special report on the impacts of global warming of 1.5 °C above pre-industrial levels and related global greenhouse gas emission pathways, in the context of strengthening the global response to the threat of climate change, sustainable development, and efforts to eradicate poverty. Intergovernmental Panel on Climate Change. https://doi.org/10.1038/291285a0 2. Fritsche, U., Brunori, G., Chiaramonti, D., Galanakis, C.M., Hellweg, S, Matthews, R. & Panoutsou, C. (2020). Future transitions for the Bioeconomy towards Sustainable Development and a Climate-Neutral Economy. Luxembourg: © European Union. https://doi.org/10.2760/667966 3. United Nations. (2021). The Paris Agreement. Retrieved from https://www.un.org/en/climatechange/paris-agreement 4. Ruppert, A. M., Weinberg, K., & Palkovits, R. (2012). Hydrogenolysis goes bio: from carbohydrates and sugar alcohols to platform chemicals. Angewandte Chemie International Edition, 51(11), 2564-2601. https://doi.org/10.1002/anie.201105125 5. Murzin, D. Y., Duque, A., Arve, K., Sifontes, V., Aho, A., Eränen, K., & Salmi, T. (2015). Catalytic hydrogenation of sugars. In Biomass Sugars for Non-Fuel Applications (pp. 89-133). Royal Society of Chemistry. https://doi.org/10.1039/9781782622079-00089 6. Ubando, A. T., Felix, C. B., & Chen, W. H. (2020). Biorefineries in circular bioeconomy: A comprehensive review. Bioresource technology, 299, 122585. https://doi.org/10.1016/j.biortech.2019.122585 7. Ubando, A. T., Del Rosario, A. J. R., Chen, W. H., & Culaba, A. B. (2020). A state-of-the-art review of biowaste biorefinery. Environmental Pollution, 116149. https://doi.org/10.1016/j.envpol.2020.116149 5. References 79 8. AliAkbari, R., Ghasemi, M. H., Neekzad, N., Kowsari, E., Ramakrishna, S., Mehrali, M., & Marfavi, Y. (2021). High value add bio-based low-carbon materials: Conversion processes and circular economy. Journal of Cleaner Production, 293, 126101. https://doi.org/10.1016/j.jclepro.2021.126101 9. Pinales-Márquez, C. D., Rodríguez-Jasso, R. M., Araújo, R. G., Loredo-Treviño, A., Nabarlatz, D., Gullón, B., & Ruiz, H. A. (2021). Circular bioeconomy and integrated biorefinery in the production of xylooligosaccharides from lignocellulosic biomass: A review. Industrial Crops and Products, 162, 113274. https://doi.org/10.1016/j.indcrop.2021.113274 10. Werpy, T., & Petersen, G. (2004). Top value added chemicals from biomass: volume I--results of screening for potential candidates from sugars and synthesis gas (No. DOE/GO-1020041992). National Renewable Energy Lab., Golden, CO (US). https://doi.org/10.2172/15008859 11. Zada, B., Chen, M., Chen, C., Yan, L., Xu, Q., Li, W., & Fu, Y. (2017). Recent advances in catalytic production of sugar alcohols and their applications. Science China Chemistry, 60(7), 853-869. https://doi.org/10.1007/s11426-017-9067-1 12. McKendry, P. (2002). Energy production from biomass (part 1): overview of biomass. Bioresource technology, 83(1), 37-46. https://doi.org/10.1016/S0960-8524(01)00118-3 13. Ruppert, A. M., Weinberg, K., & Palkovits, R. (2012). Hydrogenolysis goes bio: from carbohydrates and sugar alcohols to platform chemicals. Angewandte Chemie International Edition, 51(11), 2564-2601.https://doi.org/10.1002/anie.201105125 14. Murzin, D. Y., Daigue, E., Slotte, R., Sladkovskiy, D. A., & Salmi, T. (2020). Techno-economic analysis for production of L-arabitol from L-arabinose. Chemical Engineering & Technology, 43(7), 1260-1267. https://doi.org/10.1002/ceat.202000125 15. Fortune Bussiness Insights. (2020). Sugar Alchol Market Size, Share & COVID-19 Impact Analysis, By Type (Sorbito, Xyliyol, Maltitol, Erythritil and Isomalt), Application (Food and Beberages, Pharmaceuticals, and Cosmetics and Personal Care), and Reginal Forecast 20202027. 5. References 80 16. Ji, X. J., Huang, H., Nie, Z. K., Qu, L., Xu, Q., & Tsao, G. T. (2012). Fuels and chemicals from hemicellulose sugars. Advances in Biochemical Engineering/Biotechnology, 128. https://doi.org/10.1007/10_2011_124 17. Mokhena, T. C., Mochane, M. J., Motaung, T. E., Linganiso, L. Z., Thekisoe, O. M., & Songca, S. P. (2018). Sugarcane Bagasse and Cellulose Polymer Composites. In Sugarcane - Technology and Research. https://doi.org/10.5772/intechopen.71497 18. Grembecka, M. (2015). Sugar alcohols—their role in the modern world of sweeteners: a review. European Food Research and Technology, 241(1), 1-14. https://doi.org/10.1007/s00217-015-2437-7 19. Liao, Y., Liu, Q., Wang, T., Long, J., Zhang, Q., Ma, L., & Li, Y. (2014). Promoting hydrolytic hydrogenation of cellulose to sugar alcohols by mixed ball milling of cellulose and solid acid catalyst. Energy & fuels, 28(9), 5778-5784. https://doi.org/10.1021/ef500717p 20. Liu, Q., Tan, J., Cai, C., Ma, L., & Wang, T. (2016). Enhanced Sugar Alcohol Production from Cellulose by Pretreatment with Mixed Ball-Milling and Solid Acids. BioResources, 11(1). https://doi.org/10.15376/biores.11.1.1843-1854 21. Kobayashi, H., Ito, Y., Komanoya, T., Hosaka, Y., Dhepe, P. L., Kasai, K., Fukuoka, A. (2011). Synthesis of sugar alcohols by hydrolytic hydrogenation of cellulose over supported metal catalysts. Green Chemistry, 13(2). https://doi.org/10.1039/c0gc00666a 22. Maki-Arvela, P., Salmi, T., Holmbom, B., Willfor, S., & Murzin, D. Y. (2011). Synthesis of sugars by hydrolysis of hemicelluloses-a review. Chemical reviews, 111(9), 5638-5666. https://doi.org/10.1021/cr2000042 23. Ramos-Andrés, M., Aguilera-Torre, B., & García-Serna, J. (2021). Hydrothermal production of high-molecular weight hemicellulose-pectin, free sugars and residual cellulose pulp from discarded carrots. Journal of Cleaner Production, 290, 125179. https://doi.org/10.1016/j.jclepro.2020.125179 5. References 81 24. Murzin, D. Y., Kusema, B., Murzina, E. V., Aho, A., Tokarev, A., Boymirzaev, A. S., ... & Salmi, T. (2015). Hemicellulose arabinogalactan hydrolytic hydrogenation over Ru-modified H-USY zeolites. Journal of Catalysis, 330, 93-105. https://doi.org/10.1016/j.jcat.2015.06.022 25. Willför, S., Sjöholm, R., Laine, C., & Holmbom, B. (2002). Structural features of water-soluble arabinogalactans from Norway spruce and Scots pine heartwood. Wood Science and Technology, 36(2). https://doi.org/10.1007/s00226-001-0137-x 26. Van Gorp, K., Boerman, E., Cavenaghi, C. V., & Berben, P. H. (1999). Catalytic hydrogenation of fine chemicals: sorbitol production. Catalysis today, 52(2-3), 349-361. https://doi.org/10.1016/S0920-5861(99)00087-5 27. Chen, C. C., Chen, J. H., & Chao, C. G. (2005). Post-treatment method of producing ordered array of anodic aluminum oxide using general purity commercial (99.7%) aluminum. Japanese Journal of Applied Physics, Part 1: Regular Papers and Short Notes and Review Papers, 44(3), 1529–1533. https://doi.org/10.1143/JJAP.44.1529 28. Kuusisto, J., Mikkola, J. P., Sparv, M., Wärnå, J., Karhu, H., & Salmi, T. (2008). Kinetics of the catalytic hydrogenation of d-lactose on a carbon supported ruthenium catalyst. Chemical Engineering Journal, 139(1), 69–77. https://doi.org/10.1016/j.cej.2007.07.084 29. Simakova, I. L., Demidova, Y. S., Murzina, E. V., Aho, A., & Murzin, D. Y. (2016). Structure Sensitivity in Catalytic Hydrogenation of Galactose and Arabinose over Ru/C Catalysts. Catalysis Letters, 146(7), 1291–1299. https://doi.org/10.1007/s10562-016-1752-3 30. Thakur, D. B., Tiggelaar, R. M., Hoang, T. M. C., Gardeniers, J. G. E., Lefferts, L., & Seshan, K. (2011). Ruthenium catalyst on carbon nanofiber support layers for use in silicon-based structured microreactors, Part I: Preparation and characterization. Applied Catalysis B: Environmental, 102(1–2), 232–242. https://doi.org/10.1016/j.apcatb.2010.12.003 31. Wenmakers, P. W. A. M. (2009). Hairy Foam: Carbon nanofibers on solid foam as catalyst support. Synthesis, mass transfer and reactor. PhD Thesis. Eindhoven University of Technology. 5. References 82 32. Lali, F., Gärtner, S., Haase, S., & Lange, R. (2015). Preparation Method for Ruthenium Catalysts Supported by Carbon-Coated Aluminum Foams. Chemical Engineering and Technology, 38(8), 1353–1361. https://doi.org/10.1002/ceat.201400676 33. Schimpf, S., Bron, M., & Claus, P. (2004). Carbon-coated microstructured reactors for heterogeneously catalyzed gas phase reactions: influence of coating procedure on catalytic activity and selectivity. Chemical Engineering Journal, 101(1-3), 11-16. https://doi.org/10.1016/j.cej.2003.11.009 34. Najarnezhadmashhadi, A., Eränen, K., Engblom, S., Aho, A., Murzin, D., & Salmi, T. (2020). Continuous Hydrogenation of Monomeric Sugars and Binary Sugar Mixtures on a Ruthenium Catalyst Supported by Carbon-Coated Open-Cell Aluminum Foam. Industrial and Engineering Chemistry Research, 59(30), 13450–13459. https://doi.org/10.1021/acs.iecr.0c01565 35. Sifontes Herrera, V. A., Oladele, O., Kordás, K., Eränen, K., Mikkola, J. P., Murzin, D. Y., & Salmi, T. (2011). Sugar hydrogenation over a Ru/C catalyst. Journal of Chemical Technology and Biotechnology, 86(5), 658–668. https://doi.org/10.1002/jctb.2565 36. Sifontes Herrera, V. A., Saleem, F., Kusema, B., Eränen, K., & Salmi, T. (2012). Hydrogenation of L-Arabinose and D-Galactose Mixtures Over a Heterogeneous Ru/C Catalyst. Topics in Catalysis, 55(7–10), 550–555. https://doi.org/10.1007/s11244-012-9833-z 37. Crezee, E., Hoffer, B. W., Berger, R. J., Makkee, M., Kapteijn, F., & Moulijn, J. A. (2003). Threephase hydrogenation of D-glucose over a carbon supported ruthenium catalyst - Mass transfer and kinetics. Applied Catalysis A: General, 251(1), 1–17. https://doi.org/10.1016/S0926-860X(03)00587-8 38. Salmi, T., Murzin, D. Y., Mikkola, J. P., Wärnå, J., Mäki-Arvela, P., Toukoniitty, E., & Toppinen, S. (2004). Advanced kinetic concepts and experimental methods for catalytic three-phase processes. Industrial & engineering chemistry research, 43(16), 4540-4550. https://doi.org/10.1021/ie0307481 39. Najarnezhadmashhadi, A., Wärnå, J., Eränen, K., Trajano, H. L., Murzin, D., & Salmi, T. (2021). Modelling of kinetics, mass transfer and flow pattern on open foam structures in tubular reactors: Hydrogenation of arabinose and galactose on ruthenium catalyst. Chemical Engineering Science, 233, 116385–116395. https://doi.org/10.1016/j.ces.2020.116385 5. References 83 40. Mikkola, J.-P., Salmi, T., & Sjöholm, R. (1999). Modelling of kinetics and mass transfer in the hydrogenation of xylose over Raney nickel catalyst. Journal of Chemical Technology & Biotechnology: International Research in Process, 74(7), 655–662. https://doi.org/10.1002/(SICI)1097-4660(199907)74:7%3C655::AID-JCTB96%3E3.0.CO;2-G 41. Ho, P. H., Ambrosetti, M., Groppi, G., Tronconi, E., Palkovits, R., Fornasari, G., Benito, P. (2019). Structured Catalysts-Based on Open-Cell Metallic Foams for Energy and Environmental Applications. In Studies in Surface Science and Catalysis (pp. 303–327). https://doi.org/10.1016/B978-0-444-64127-4.00015-X 42. Cybulski, A., & Moulijn, J. A. (2005). The present and the future of structured catalysts: An overview. In Structured Catalysts and Reactors. CRC Press. https://doi.org/10.1201/9781420028003 43. Montebelli, A., Visconti, C. G., Groppi, G., Tronconi, E., Cristiani, C., Ferreira, C., & Kohler, S. (2014). Methods for the catalytic activation of metallic structured substrates. Catalysis Science & Technology, 4(9), 2846-2870. https://doi.org/10.1039/c4cy00179f 44. Tronconi, E., Groppi, G., & Visconti, C. G. (2014). Structured catalysts for non-adiabatic applications. Current Opinion in Chemical Engineering, 5, 55-67. https://doi.org/10.1016/j.coche.2014.04.003 45. Pangarkar, K., Schildhauer, T. J., van Ommen, J. R., Nijenhuis, J., Kapteijn, F., & Moulijn, J. A. (2008). Structured packings for multiphase catalytic reactors. Industrial & engineering chemistry research, 47(10), 3720-3751. https://doi.org/10.1021/ie800067r 46. Gancarczyk, A., Sindera, K., Iwaniszyn, M., Piątek, M., Macek, W., Jodłowski, P. J., & Kołodziej, A. (2019). Metal foams as novel catalyst support in environmental processes. Catalysts, 9(7), 587. https://doi.org/10.3390/catal9070587 47. Ho, P. H., Ambrosetti, M., Groppi, G., Tronconi, E., Jaroszewicz, J., Ospitali, F., … Benito, P. (2018). One-step electrodeposition of Pd-CeO2 on high pore density foams for environmental catalytic processes. Catalysis Science and Technology, 8(18). https://doi.org/10.1039/c8cy01388h 5. References 84 48. Wenmakers, P. W., van der Schaaf, J., Kuster, B. F., & Schouten, J. C. (2010). Comparative modeling study on the performance of solid foam as a structured catalyst support in multiphase reactors. Industrial & Engineering Chemistry Research, 49(11), 5353-5366. https://doi.org/10.1021/ie900644e 49. Nijhuis, T. A., Beers, A. E., Vergunst, T., Hoek, I., Kapteijn, F., & Moulijn, J. A. (2001). Preparation of monolithic catalysts. Catalysis Reviews, 43(4), 345-380. https://doi.org/10.1081/CR-120001807 50. Moreno-Castilla, C., Mahajan, O. P., Walker Jr, P. L., Jung, H. J., & Vannice, M. A. (1980). Carbon as a support for catalysts—III glassy carbon as a support for iron. Carbon, 18(4), 271276. https://doi.org/10.1016/0008-6223(80)90050-0 51. Vergunst, T., Kapteijn, F., & Moulijn, J. A. (2002). Preparation of carbon-coated monolithic supports. Carbon, 40(11), 1891-1902. https://doi.org/10.1016/S0008-6223(02)00034-9 52. Iroegbu, A. O., & Hlangothi, S. P. (2019). Furfuryl alcohol a versatile, eco-sustainable compound in perspective. Chemistry Africa, 2(2), 223-239. https://doi.org/10.1007/s42250-018-00036-9 53. Hucke, E. E. (1975). U.S. Patent No. 3,859,421. Washington, DC: U.S. Patent and Trademark Office. 54. Tondi, G., Cefarin, N., Sepperer, T., D’Amico, F., Berger, R. J. F., Musso, M., Vaccari, L. (2019). Understanding the polymerization of polyfurfuryl alcohol: Ring opening and diels-alder reactions. Polymers, 11(12), 2126–2141. https://doi.org/10.3390/polym11122126 55. Munnik, P., de Jongh, P. E., & de Jong, K. P. (2015). Recent developments in the synthesis of supported catalysts. Chemical reviews, 115(14), 6687-6718. https://doi.org/10.1021/cr500486u 56. Thakur, D. B., Tiggelaar, R. M., Weber, Y., Gardeniers, J. G. E., Lefferts, L., & Seshan, K. (2011). Ruthenium catalyst on carbon nanofiber support layers for use in silicon-based structured microreactors. Part II: Catalytic reduction of bromate contaminants in aqueous phase. Applied Catalysis B: Environmental, 101(1–2), 243–250. https://doi.org/10.1016/j.apcatb.2010.12.004 5. References 85 57. Toebes, M. L., Van Der Lee, M. K., Tang, L. M., Huis In ’t Veld, M. H., Bitter, J. H., Jos Van Dillen, A., & De Jong, K. P. (2004). Preparation of Carbon Nanofiber Supported Platinum and Ruthenium Catalysts: Comparison of Ion Adsorption and Homogeneous Deposition Precipitation. The Journal of Physical Chemistry B, 108(31), 11611–11619. https://doi.org/10.1021/jp0313472 58. Manyar, H. G., Weber, D., Daly, H., Thompson, J. M., Rooney, D. W., Gladden, L. F., … Hardacre, C. (2009). Deactivation and regeneration of ruthenium on silica in the liquid-phase hydrogenation of butan-2-one. Journal of Catalysis, 265(1), 80–88. https://doi.org/10.1016/j.jcat.2009.04.013 59. Rodríguez-Reinoso, F., & Sepúlveda-Escribano, A. (2001). Porous carbons in adsorption and catalysis. In Handbook of surfaces and interfaces of materials (pp. 309-355). Academic Press. https://doi.org/10.1016/b978-012513910-6/50066-9 60. Mason, R. B. (1956). Effect of Aluminum Sulfate in the Sulfuric Acid Electrolyte on Anodic Polarization. Journal of The Electrochemical Society, 103(8), 425–429. https://doi.org/10.1149/1.2430373 61. Lali, F., Böttcher, G., Schöneich, P. M., Haase, S., Hempel, S., Lange, R. (2015). Preparation and characterization of Pd/Al2O3 catalysts on aluminum foam supports for multiphase hydrogenation reactions in rotating foam reactors. Chemical Engineering Research and Design, 94, 365–374. https://doi.org/10.1016/j.cherd.2014.08.012 62. Sanz, O., Almeida, L. C., Zamaro, J. M., Ulla, M. A., Miró, E. E., Montes, M. (2008). Washcoating of Pt-ZSM5 onto aluminium foams. Applied Catalysis B: Environmental, 78(1– 2), 166–175. https://doi.org/10.1016/j.apcatb.2007.09.024 63. Burgos, N., Paulis, M., Montes, M. (2003). Preparation of Al2O3/Al monoliths by anodisation of aluminium as structured catalytic supports. Journal of Materials Chemistry, 13(6). https://doi.org/10.1039/b212242a 64. Masuda, H., Hasegwa, F., Ono, S. (1997). Self-Ordering of Cell Arrangement of Anodic Porous Alumina Formed in Sulfuric Acid Solution. Journal of The Electrochemical Society, 144(5), L127–L130. https://doi.org/10.1149/1.1837634 5. References 86 65. Kozhukhova, A. E., du Preez, S. P., Bessarabov, D. G. (2020). The effects of pore widening and calcination on anodized aluminum oxide prepared from Al6082. Surface and Coatings Technology, 383, 125234–125246. https://doi.org/10.1016/j.surfcoat.2019.125234 66. Choura, M., Belgacem, N. M., Gandini, A. (1996). Acid-catalyzed polycondensation of furfuryl alcohol: Mechanisms of chromophore formation and cross-linking. Macromolecules, 29(11), 3839-3850. https://doi.org/10.1021/ma951522f 67. Constant, K. P., Lee, J. R., Chiang, Y. M. (1996). Microstructure development in furfuryl resinderived microporous glassy carbons. Journal of Materials Research, 11(9), 2338–2345. https://doi.org/10.1557/JMR.1996.0297 68. Cepollaro, E. M., Caputo, D., Cimino, S., Gargiulo, N., Lisi, L. (2020). Synthesis and Characterization of Activated Carbon Foam from Polymerization of Furfuryl Alcohol Activated by Zinc and Copper Chlorides. C — Journal of Carbon Research, 6(3), 45–60. https://doi.org/10.3390/c6030045 69. Koopman, P. O. J., Kieboom, A. P. G., Van Bekkum, H. (1981). Induction effects in liquid phase hydrogenation catalyzed by ruthenium on carbon. Colloids and Surfaces, 3(1), 1–12. https://doi.org/10.1016/0166-6622(81)80030-3 70. Koopman, P. G. J., Buurmans, H. M. A., Kieboom, A. P. G., van Bekkum, H. (1981). SolventReactant-Support interactions in liquid phase hydrogenation. Recueil des Travaux Chimiques des Pays‐Bas, 100(4), 129–136. https://doi.org/10.1002/recl.19811000407 71. Maximov, A. L., Zolotukhina, A. V., Mamedli, A. A., Kulikov, L. A., Karakhanov, E. A. (2018). Selective Levulinic Acid Hydrogenation in the Presence of Hybrid Dendrimer-Based Catalysts. Part I: Monometallic. ChemCatChem, 10(1), 222–233. https://doi.org/10.1002/cctc.201700691 72. Wang, X., Lan, G., Liu, H., Zhu, Y., Li, Y. (2018). Effect of acidity and ruthenium species on catalytic performance of ruthenium catalysts for acetylene hydrochlorination. Catalysis Science and Technology, 8(23), 6143–6149. https://doi.org/10.1039/c8cy01677a 73. Koopman, P. G. J., Kieboom, A. P. G., van Bekkum, H. (1981). Characterization of ruthenium catalysts as studied by temperature programmed reduction. Journal of Catalysis, 69(1), 172– 179. https://doi.org/10.1016/0021-9517(81)90139-1 5. References 87 74. Rio, S., Peru, G., Léger, B., Kerdi, F., Besson, M., Pinel, C., … Ponchel, A. (2020). Supported ruthenium nanoparticles on ordered mesoporous carbons using a cyclodextrin-assisted hardtemplate approach and their applications as hydrogenation catalysts. Journal of Catalysis, 383, 343–356. https://doi.org/10.1016/j.jcat.2019.10.021 75. Tourvieille, J.-N. (2010). Ultrasound effect on hydrogenation of sugars : catalyst deactivation and modelling. Master's Thesis. Åbo Akademi and Université de Lyon. 76. Arena, B. J. (1992). Deactivation of ruthenium catalysts in continuous glucose hydrogenation. Applied Catalysis A, General, 87(2), 219–229. https://doi.org/10.1016/0926-860X(92)80057-J 77. Aho, A., Roggan, S., Eränen, K., Salmi, T., Murzin, D. Y. (2015). Continuous hydrogenation of glucose with ruthenium on carbon nanotube catalysts. Catalysis Science and Technology, 5(2), 953–959. https://doi.org/10.1039/c4cy01088d 78. Pyatnitsky, Y. I. (1994). Some new approaches to the competitive catalytic reaction kinetics. Applied Catalysis A, General, 113(1). https://doi.org/10.1016/0926-860X(94)80238-6 79. Kusema, B. T., Hilpmann, G., Mäki-Arvela, P., Willför, S., Holmbom, B., Salmi, T., Murzin, D. Y. (2011). Selective hydrolysis of arabinogalactan into arabinose and galactose over heterogeneous catalysts. Catalysis Letters, 141(3), 408–412. https://doi.org/10.1007/s10562-010-0530-x 80. Hussey, A. S., Baker, R. H., Keulks, G. W. (1968). Competitive hydrogenations on platinum. Journal of Catalysis, 10(3). https://doi.org/10.1016/S0021-9517(68)80007-7 81. Cerveny, L., Ruzicka, V. (1982). Competitive Catalytic Hydrogenation in the Liquid Phase on Solid Catalysts. Catalysis Reviews, 24(4). https://doi.org/10.1080/03602458208079662 82. Rader, C. P., Smith, H. A. (1962). Competitive Catalytic Hydrogenation of Benzene, Toluene and the Polymethylbenzenes on Platinum. Journal of the American Chemical Society, 84(8). https://doi.org/10.1021/ja00867a021 Appendix 94 APPENDIX IV: MODELING SINGLE SUGAR HYDROGENATION (PYTHON CODE) import os import numpy as np import pandas as pd from scipy.integrate import solve_ivp import matplotlib.pyplot as plt from scipy.optimize import minimize print("\n---Sugar Hydrogenation---\n") #CONSTANTS ncomp = 2 #Number of components-- Arabinose, Arabitol or Galactose, Galactitol nre =3 #Number of reactions mC = 0.01418816 #Mass of Ru [g] VL = 130/1000 #Reaction Volume [L] rho = mC/VL #Bulk density [g/L] T= [90+273.15, 100+273.15,120+273.15] #Temperatures [K] R= 8.314 #Gas constant [J/mol] #2. READING DATA dirpath = os.getcwd() + "\\" trainname = "KINETIC_DATA .xlsx" data=np.array(pd.read_excel (dirpath + trainname,"Comb_Galactose")) #Choose "Comb_Galactose" or "Comb_Arabinose" nserie=int(data.shape[1]/2) ndata=data.shape[0] nser=ncomp*nre t_span=np.zeros((nserie,ndata-1)) t_train=np.zeros((ndata)) #Experimental Time Vector C_train=np.zeros((ndata,nser)) #Experimental Concentration Matrix t_train=data[:,0] C_train[:,0]=data[:,1] C_train[:,1]=data[:,2] C_train[:,2]=data[:,4] C_train[:,3]=data[:,5] C_train[:,4]=data[:,7] C_train[:,5]=data[:,8] # KINETIC MODEL dC=np.zeros(ncomp*3) #Initializing Derivatives Vector r=np.zeros(3) #Initializing Reaction Rates Vector k=[814434, 55956, 0.7] #Initial Values for the Parameters A=k[0], E=k[1] and kS=k[2] Appendix 95 def model (t, C, k): #Differential equations function. for i in range (0,3): r[i]=((k[0]*np.exp(-k[1]/(T[i]*R)))*C[2*i])/((1+k[2]*C[2*i])) #Reaction rates dC[2*i]=-r[i]*rho #Mole balances for the sugar dC[2*i+1]=r[i]*rho #Mole balances for the sugar alcohol return dC C0 = C_train[0,:] #Stablishing initial conditions at t=0 min tspan= [0, t_train[-1]] #Defining integration range C_int=np.zeros((len(t_train),ncomp*3)) # interpolation t_train=t_train.flatten() #OPTIMIZATION def of(k): #Objetive Function sol = solve_ivp(lambda t,C: model(t,C,k), t_span=tspan, y0=C0, method='LSODA',t_eval=t_train) C_int = sol.y.T error=abs(C_int-C_train)**2 #Objective function errorval=sum(sum(error)) return errorval def R2(k): #Determination Coefficient sol = solve_ivp(lambda t,C: model(t,C,k), t_span=tspan, y0=C0, method='LSODA',t_eval=t_train) C_int = sol.y.T Cmean=sum(C_train)/ndata R2=(1-((sum((C_train-C_int)**2))/(sum((C_train-Cmean)**2))))*100 R2=sum(R2)/nser return R2 #SOLUTION #Calling optimization solver res = minimize(of, k, method='nelder-mead', options={'xatol': 1e-8, 'disp': True}) k=res.x #Obtained parameters R2v=R2(k) #Computing coefficient of correlation for the model t = np.linspace(0,tspan[1], num=200) # time points #Solving with optimal parameters sol = solve_ivp(lambda t,C: model(t,C,k), t_span=tspan, y0=C0, method='LSODA', t_eval=t.flatten()) C = sol.y.T t=sol.t #Showing Estimated Paramaters print("A=" + str(k[0])+" L/gRu*min*mol") print("EA=" + str(k[1])+" J/mol") print("k'=" + str(k[2])+" L/mol") print("R2="+ str(R2v)+ " %") Appendix 96 APPENDIX V: MODELING SUGAR MIXTURES HYDROGENATION (PYTHON CODE) import os import numpy as np import pandas as pd from scipy.integrate import solve_ivp import matplotlib.pyplot as plt from scipy.optimize import minimize print("\n---Sugar Mixtures Hydrogenation---\n") #CONSTANTS ncomp = 2 #Number of components-- Arabinose, Arabitol nre =1 #Number of reactions mC = 0.01179148318 #Mass of Ru [g] VL = 130/1000 #Reaction Volume [L] rho = mC/VL #Bulk density [g/L] T= [90+273.15, 100+273.15, 120+273.15] #Temperatures [K] R= 8.314 #Gas constant [J/mol] #2. READING DATA dirpath = os.getcwd() + "\\" trainname = "MIXTURES.xlsx" data=np.array(pd.read_excel (dirpath + trainname,"Hoja1")) nserie=int(data.shape[1]/2) ndata=data.shape[0] nser=ncomp*nre t_span=np.zeros((ndata)) ##t_train=np.zeros((ndata)) #Experimental Time Vector C_train=np.zeros((ndata,nser)) #Experimental Concentration Matrix ## t_train=data[:,0] for i in range (0, nser): C_train[:,i]=data[:,i+1] ### KINETIC MODEL dC=np.zeros(nser) #Initilazing Derivatives Vector r=np.zeros(6) #Initilazing Reaction Rates Vector k=[0.2, 0.1, 7.99400900e+00] #Arabinose initial constants #Initial Values for the Parameters A=k[0], E=k[1] and kS=k[2] def model (t, C, k): r[0]=k[0]*C[0]/(1+k[2]*C[0]) r[1]=k[1]*C[1]/(1+k[2]*C[0]) Appendix 97 r[2]=k[0]*C[4]/(1+k[2]*C[4]) r[3]=k[1]*C[5]/(1+k[2]*C[4]) r[4]=k[0]*C[8]/(1+k[2]*C[8]) r[5]=k[1]*C[9]/(1+k[2]*C[8]) dC[0]=-r[0]*rho dC[1]=-r[1]*rho dC[2]=r[0]*rho dC[3]=r[1]*rho dC[4]=-r[2]*rho dC[5]=-r[3]*rho dC[6]=r[2]*rho dC[7]=r[3]*rho dC[8]=-r[4]*rho dC[9]=-r[5]*rho dC[10]=r[4]*rho dC[11]=r[5]*rho return dC C0 = C_train[0,:] tspan= [0, t_train[-1]] #Defining integration range C_int=np.zeros((len(t_train),ncomp*3)) # interpolation t_train=t_train.flatten() #OPTIMIZATION def of(k): #Objetive Function sol = solve_ivp(lambda t,C: model(t,C,k), t_span=tspan, y0=C0, method='Radau',t_eval=t_train) C_int = sol.y.T error=abs(C_int-C_train)**2 #Objective function errorval=sum(sum(error)) return errorval def error (k): sol = solve_ivp(lambda t,C: model(t,C,k), t_span=tspan, y0=C0, method='LSODA',t_eval=t_train) C_int = sol.y.T error=abs(C_int-C_train)**2 #Objective function SE=((sum(sum(error))/(ndata-1))**0.5) return SE def R2(k): #Determination Coefficient sol = solve_ivp(lambda t,C: model(t,C,k), t_span=tspan, y0=C0, method='LSODA',t_eval=t_train) Appendix 98 C_int = sol.y.T Cmean=sum(sum(C_train))/(ndata*6) R2=(1-((sum(sum((C_train-C_int)**2)))/(sum(sum((C_train-Cmean)**2)))))*100 return R2 ###SOLUTION #Calling optimization solver res = minimize(of, k, method='nelder-mead', options={'xatol': 1e-8, 'disp': True}) k=res.x #Obtained parameters R2v=R2(k) #Computing coefficient of determination for the model t = np.linspace(0,tspan[1], num=200) # time points ###Solving with optimal parameters sol = solve_ivp(lambda t,C: model(t,C,k), t_span=tspan, y0=C0, method='Radau', t_eval=t.flatten()) C = sol.y.T t=sol.t ###Showing Estimated Paramaters print("k'1=" + str(k[0])+" L/gRu*min*mol") print("k'2=" + str(k[1])+" J/mol") print("kA=" + str(k[2])+" L/mol") print("R2="+ str(R2v)+ " %")