Rights / License: Research Collection In Copyright - Non ......The best wishes for Thomas Pielhop in...

Research Collection

Doctoral Thesis

A process for the complete valorization of lignin into aromaticchemicals based on acidic oxidation

Author(s): Werhan, Holger

Publication Date: 2013

Permanent Link: https://doi.org/10.3929/ethz-a-009790818

Rights / License: In Copyright - Non-Commercial Use Permitted

This page was generated automatically upon download from the ETH Zurich Research Collection. For moreinformation please consult the Terms of use.

ETH Library

https://doi.org/10.3929/ethz-a-009790818

http://rightsstatements.org/page/InC-NC/1.0/

https://www.research-collection.ethz.ch

https://www.research-collection.ethz.ch/terms-of-use

Diss. ETH No. 21132

A Process for the Complete Valorization

of Lignin into Aromatic Chemicals based

on Acidic Oxidation

A dissertation submitted to

ETH Zurich

for the degree of

Doctor of Sciences

presented by

Holger Werhan

Dipl.-Ing. Universität Stuttgart

born on January 16, 1982in Stuttgart (Germany)

citizen of Germany

Accepted on the recommendation of

Prof. Dr. Dr. h.c. Philipp Rudolf von Rohr (ETH Zurich), examinerProf. Dr. Sven Panke (ETH Zurich), co-examiner

Dr. Irene Schober (Sika Technology AG), co-examiner

Zurich, 2013

c© Holger Werhan, 2013

"You can make anything out of lignin... except money"

Old industry saying

Preface I

Preface

The present doctoral thesis was conducted at the Institute of Process En-gineering at ETH Zurich between September 2009 and April 2013. Theproject was facilitated by the financial support of Sika Technology AGand the Commission of Technology and Invention (CTI) in Switzerland.The close and fruitful scientific collaboration with Sika Technology AGand the Functional Materials Laboratory (FML) at ETH contributed tothe successful completion of the present work. Herewith I would like tothank all the people who were involved in the success of the project andthe present doctoral thesis.

First of all, I want to express my gratitude to my doctoral adviser Prof.Dr. Philipp Rudolf von Rohr for the opportunity to conduct the doctoralthesis in his laboratory and for his continuous support throughout thethesis. I appreciate his advice, his unrestricted confidence in me, thenumerous scientific discussions which sometimes helped to get a newperspective on the project, and the necessary freedom he gave me toconduct the doctoral investigations. In addition, I would like to thankProf. Dr. Sven Panke and Dr. Irene Schober for being my co-examiners,regardless of a busy schedule.

Special thanks go as well to Dr. Irene Schober from Sika TechnologyAG for the organization and supervision of the CTI project and to Mar-tin Zeltner from FML for the cooperation within the project. I mustalso acknowledge Martin Zeltner and Karl Fluri (Sika Technology AG)for their analytical support, for performing chemical analyses relatedto the thesis (NMR, Titration / APCI-MS) as well as Martin for thechemical synthesis. Furthermore, I would like to thank Michael Studerand Simone Brethauer-Studer for their help in biomass pretreatment,enzymatic hydrolysis and biomass analysis.

Appreciation also goes out to my predecessor Tobias Voitl for intro-ducing me into the world of lignin and for sharing his scientific knowledgewith me. Very special thanks to Nora Assmann, Thomas Pielhop, Bruno

II Preface

Tidona and Panagiotis Stathopoulos for the fruitful scientific discussionsrelated to this work. The best wishes for Thomas Pielhop in the con-tinuation of the biomass valorization research at the Institute of ProcessEngineering.

I sincerely want to thank all the students who supported me in mydoctoral studies in the form of master theses, semester theses or bachelorstheses. The motivated work of Akbar Farshori, Joan Mora Mir, AndreasLamprecht, Martin Wiese, Silvio Tödtli, and Simon Ambühl essentiallycontributed to the present thesis. It was a pleasure to work with you, Ihope you profited a lot from your theses and wish you all the best foryour professional and personal future.

The successful completion of the doctoral studies would not have beenpossible without the support of Bruno Kramer, Peter Feusi and MartinMeuli from the mechanical and electrical workshop, the administrativework of Silke Stubbe as well as the chemical and analytical advice ofMarkus Huber. I herewith gratefully acknowledge their work.

I would like to thank all my current and former colleagues at theLTR, namely Bruno, Gina, Christian, Martin, Nora, Thomas, Panagi-otis, Adrian, Tobias R., Tobias V., Patrick, Cédric, Lutz, Serge, Axel,Denis, Richard, Yannick, Dragana, Thierry, Agnieszka, Vito and Helenafor the nice atmosphere at work and the great time we spent outside theETH. In this context, I especially appreciated the cooperative and con-structive atmosphere with Bruno, Serge and Gina in the office. Thankyou Patrick, Christian and Martin for climbing, hiking and skiing tripsto the Swiss Alps together. I also acknowledge the efforts of Thomasand Tobias R. in the organization of the soccer and the unihockey team,respectively.

Finally, I would like to cordially thank my parents and my sister forthe support they provided me through my entire life and for giving methe opportunity to pursue all the goals in my private and professionalcareer. Very special thanks are dedicated to my fiancée Lea for her love,patience and support throughout all the years.

Zurich, April 2013 - Holger Werhan

Abstract III

Abstract

Woody biomass represents a very important and promising feedstock interms of the renewable production of fuels and chemicals in the future.A main constituent of wood is lignin, which is the second most abun-dant resource in nature after cellulose and accounts for about 25% ofthe world’s biomass. While designated applications for cellulose alreadyexist in form of the current pulp and paper production as well as itsprospective hydrolysis and fermentation into biofuels, sustainable waysto valorize the lignin fraction of wood are yet to come. Thus, virtually allof the estimated 70 million tons of lignin per year that nowadays accruein the pulp and paper industry are internally burnt for heat and powergeneration. However, the aromatic structure of the lignin molecules andthe large availability as a renewable carbon source open up promisingopportunities to convert lignin into aromatic chemicals.

In this context, especially the valorization of the predominant kraftlignin from the kraft pulping industry is highly attractive. Althoughdifferent chemical conversion techniques which target specific kinds ofmonomeric aromatic products have been investigated throughout the lastdecades and have been complemented by biotechnological approachesduring the last years, none of them proved suitable in yielding significantquantities of high-value aromatic chemicals from kraft lignin withoutleaving large amounts of waste material behind. In contrast, approachesthat completely employ the biopolymeric lignin mixture in applicationsof low value (e.g. dispersants, binders, emulsifiers) usually lack of eithera beneficial effect of the kraft lignin or the economical feasibility. Hence,combinations of the aforementioned approaches can have a superior po-tential in enhancing the generation of valuable chemicals and therebythe valorization of lignin.

In the present thesis, a combined approach for the complete valoriza-tion of kraft lignin into high-value monomeric aromatic chemicals as wellas an oligomeric fraction of high molecular weight lignin oxidation prod-

IV Abstract

ucts is presented. Moreover, a conceptual process for the commercialrealization of the combined approach is proposed at the end of the the-sis and a cost estimate is performed. The process comprises, amongstothers, the major process steps of reaction, extraction and membraneseparation which were all investigated in detail and validated in the con-text of this thesis. As the oxidation of lignin has been evaluated to behighly viable to produce monomeric chemicals and acidic conditions inthe reaction solvent have as well been shown favorable, the acidic oxida-tion was selected as reaction type of choice for the current approach.

The acidic oxidation reaction and the influences on it are described indetail in the beginning of the thesis. Among the different transition metalsalt catalysts that were screened as homogeneous redox catalysts in theoxidation, cobalt chloride gave the highest amount of vanillin and methylvanillate with a yield of 3.19wt% and 3.09wt%, respectively. However,iron chloride and copper chloride featured a fast depolymerization of thelignin molecule from 3200 g mol−1 down to 500 g mol−1 and additionallyrevealed considerable amounts of other monomeric products which lateron could be identified as derivatives of vanillin and methyl vanillate andthus degradation products of them. Furthermore, a high methanol frac-tion in the solvent and a low pH value proved beneficial for the productgeneration while other types of lignin were also advantageous in termsof monomer yield compared to kraft lignin.

The influence of temperature, pressure and lignin concentration wasafterwards investigated in a two-phase flow microreactor that enabledcontinuous operation and short reaction times without having influencesof heating and cooling on the reaction. Thereby, numerous experimentswith varying reaction parameters could be performed in a short timewithout stopping the microreactor plant. Moreover, reaction conditionsof up to 250 C and 96bar of oxygen pressure could be realized in spiteof the fact that the gas phase composition in the reactor with methanoland oxygen is critical. While oxygen pressure and lignin concentrationaffected the product generation to a minor extent, higher maxima inthe product concentrations were obtained at even shorter reaction timesfor higher temperatures. At 250 C, a monomer yield of 5.02wt% wasobtained for a sulfonated kraft lignin after only 0.65min of residencetime. Assuming that the ratio of maximum product yield for differenttypes of lignin is the same as in the batch reactor, a maximum yield of

Abstract V

11.8wt% for an unmodified kraft lignin and 19.0wt% for a lignin fromenzymatic hydrolysis can be expected.

In order to recover the reaction products from the reaction solventfor further processing and thereby allow for the recycling of the reactionsolvent in a potential process, liquid-liquid extraction with an organicsolvent was studied. Ethyl acetate emerged as most promising extract-ing agent in this context. Organic solvent nanofiltration was investigatedin a next step for the separation of the monomeric products from thefraction of oligomeric lignin oxidation products. The polymer membranePuraMemTM S380 which possesses a MWCO of 600Da proved the fea-sibility of the chosen separation method and was found to be the bestof several solvent-stable membranes that were tested for this purpose.With rejection values of 38.4% for the monomeric products and 96.6%for the oligomers, an efficient separation of the monomeric products canbe achieved. While the monomeric products require further separationand purification steps before they can be obtained as products from theprocess, the oligomeric product fraction can either be directly employedin a desired application or further be subjected to a chemical modifica-tion before use.

At the end of the thesis, a conceptual process design as well as a costestimate for different possible cases are presented. Besides the majorprocess steps, all additionally required unit operations were discussedand the recycling of the internal process streams was realized. For thefinal cost estimate, three cases were evaluated: based on the experimen-tal results from the batch reactor, on experimental results from the mi-croreactor or assuming ideal process conditions. Thereby, the low ligninconcentrations as well as the losses of methanol and ethyl acetate werefound to be the main drawbacks of the current concept with regard tothe process economics. In combination with an efficient process heat in-tegration and a valuable application for the fraction of oligomeric ligninoxidation products, the improvement of the mentioned aspects can con-tribute to bring the presented concept closer to commercial realization.

Zusammenfassung VII

Zusammenfassung

Holzartige Biomasse stellt einen wichtigen und vielversprechenden Roh-stoff für die Herstellung von Kraftstoffen und Chemikalien aus erneu-erbaren Quellen in der Zukunft dar. Ein Hauptbestandteil von Holz istLignin, welches der zweithäufigste Rohstoff in der Natur nach Cellulo-se ist und 25% der weltweiten Biomasse ausmacht. Während Cellulosebereits für die Zellstoff- und Papierherstellung sowie für die zukünftigeHerstellung von Biotreibstoffen durch Hydrolyse und Fermentation vor-gesehen ist, sind nachhaltige Verwertungsmethoden des Ligninanteils imHolz noch nicht vorhanden. Aus diesem Grund werden die 70 Millio-nen Tonnen an Lignin, die jährlich in der Papier- und Zellstoffindustrieanfallen, quasi komplett zur Energiegewinnung und Chemikalienrückge-winnung intern verbrannt. Die aromatische Struktur des Lignins sowieseine hohe Verfügbarkeit als erneuerbare Kohlenstoffquelle machen Li-gnin jedoch zu einem aussichtsreichen Kandidaten für die Umwandlungin aromatische Chemikalien.

In diesem Zusammenhang ist speziell die Verwertung des vorherrschen-den Kraft-Lignins aus dem Kraftaufschluss höchst attraktiv. Obwohlverschiedene chemische Umsetzungsstrategien, die auf unterschiedlichemonomere, aromatische Produkte abzielen, in den letzten Jahrzehntenuntersucht und in den vergangenen Jahren zusätzlich durch biotechnolo-gische Ansätze ergänzt wurden, konnte keine davon geeignete Mengen anhochwertigen aromatischen Chemikalien aus Kraft-Lignin erzielen, ohnegrössere Mengen an Ausschuss zu hinterlassen. Auf der anderen Seitefehlt es jenen Ansätzen, die das komplette, biopolymerische Ligninge-misch in Anwendungen mit geringerem Wert einsetzen (z.B. Dispergier-mittel, Bindemittel, Emulgatoren), gewöhnlich an einem vorteilhaftenEffekt des Kraft-Lignins oder an der wirtschaftlichen Durchführbarkeit.Aus diesem Grund können Kombinationen aus den oben genannten An-sätzen bei der Herstellung von wertvollen Chemikalien und damit derder Verwertung von Lignin überlegen sein.

VIII Zusammenfassung

In der vorliegenden Arbeit wird ein kombinierter Ansatz zur komplet-ten Verwertung von Kraft-Lignin in hochwertige, monomere Chemikalienund in eine oligomerische Fraktion von hochmolekularen Ligninoxidati-onsprodukten präsentiert. Darüber hinaus wird am Ende der Arbeit einkonzeptioneller Prozess zur gewerblichen Umsetzung des kombiniertenAnsatzes vorgeschlagen und eine Kostenabschätzung durchgeführt. DerProzess besteht, unter anderem, aus den Hauptprozessschritten Reakti-on, Extraktion und Membrantrennung, die alle im Rahmen dieser Arbeitdetailliert untersucht und validiert wurden. Da die Oxidation von Ligninals äußerst praktikabel unter den verschiedenen Reaktionen zur Herstel-lung von monomeren Chemikalien eingestuft ist und saure Bedingungenim Reaktionslösungsmittel als vorteilhaft nachgewiesen wurden, wurdedie Oxidation in saurer, wässriger Lösung als Reaktion für den vorlie-genden Ansatz ausgewählt.

Die Oxidation in saurem Medium und die Einflüsse auf die Reak-tion werden am Anfang der Arbeit detailliert aufgezeigt. Von den ver-schiedenen Übergangsmetallsalzkatalysatoren, die als mögliche, homoge-ne Redoxkatalysatoren überprüft wurden, ergab Cobaltchlorid die höchs-te Menge an Vanillin und Methylvanillat mit Ausbeuten von 3.19 Gew.-%bzw. 3.09 Gew.-%. Eisenchlorid und Kupferchlorid führten hingegen zueiner schnellen Depolymerisierung des Ligninmoleküls von durchschnitt-lich 3200 g mol−1 hinunter auf 500 g mol−1 und erzeugten zusätzlich be-trächtliche Mengen an anderen monomeren Produkten, die später alsDerivate von Vanillin und Methylvanillat sowie als deren Folgeprodukteidentifiziert wurden. Ausserdem erwiesen sich ein hoher Methanolanteilsowie ein niedriger pH-Wert als zweckmässig für die Produktbildung,während andere Lignine sich ebenfalls als vorteilhaft gegenüber Kraft-Lignin bezüglich der Monomerausbeute herausstellten.

Der Einfluss von Temperatur, Druck und Ligninkonzentration wurdeanschliessend in einer Zweiphasenströmung im Mikroreaktor, der einenkontinuierlichen Betrieb und kurze Reaktionszeiten ohne die Einflüssevon Aufheiz- und Abkühlzeit ermöglichte, untersucht. Dadurch konntendie Reaktionsbedingungen einfach geändert werden, ohne die Reakti-on zu stoppen, wodurch eine Vielzahl an Experimenten in kurzer Zeitdurchgeführt werden konnte. Darüber hinaus waren Reaktionsbedingun-gen von bis zu 250 C und 96bar an Sauerstoffdruck realisierbar, obwohldie Gasphasenzusammensetzung aus Methanol und Sauerstoff im Re-

Zusammenfassung IX

aktor als kritisch einzustufen ist. Während Sauerstoffdruck und Lignin-konzentration die Produktbildung nur in geringem Masse beeinflussten,wurden höhere Maxima der Produktkonzentration bei noch kürzeren Re-aktionszeiten und noch höheren Temperaturen erhalten. Unter der An-nahme, dass das Verhältnis der maximalen Produktausbeuten zwischenden verschiedenen Ligninen dem aus dem Batch-Reaktor entspricht, kanneine maximale Ausbeute von 11.8 Gew.-% für ein unmodifiziertes Kraft-Lignin sowie 19.0 Gew.-% für ein Lignin aus der enzymatischen Hydro-lyse von Biomasse erwartet werden.

Um die Produkte für die weitere Bearbeitung aus dem Reaktionslö-sungsmittel rückzugewinnen und dadurch dessen Rückführung in einemmöglichen Prozess zu erlauben, wurde die Flüssig-Flüssig-Extraktion miteinem organischen Lösungsmittel betrachtet. Ethylacetat stellte sich indiesem Zusammenhang als vielversprechendstes Lösungsmittel heraus.Organophile Nanofiltration wurde anschliessend zur Abtrennung der mo-nomeren Produkte von der Fraktion der oligomerischen Ligninoxidati-onsprodukte untersucht. Die Polymermembran PuraMemTM S380, dieeine Molekulargewichtsgrenze von 600Da besitzt, bewies die Durchführ-barkeit der gewählten Trennmethode und stellte sich als beste von mehre-ren lösungsmittelstabilen Membranen heraus, die in diesem Zusammen-hang getestet wurden. Durch Rückhaltwerte von 38.4% für die monome-ren Produkte und 96.6% für die Oligomere kann eine effektive Abtren-nung der Monomere erreicht werden. Während die monomeren Produktenoch weitere Trenn- und Aufreinigungsschritte benötigen, kann die oli-gomerische Fraktion entweder sofort in einer bestimmten Anwendungverwendet oder vorher noch einer chemischen Modifizierung unterzogenwerden.

Am Ende der Arbeit werden sowohl ein konzeptioneller Verfahrensent-wurf als auch eine Kostenabschätzung für verschiedene Fälle gezeigt. Ne-ben den Hauptprozessschritten werden alle weiteren, benötigten Grund-operationen diskutiert und die Rückführung der internen Prozessströmerealisiert. Für die abschließende Kostenabschätzung wurden drei Fällebewertet: Diese basieren auf den experimentellen Ergebnissen im Batch-Reaktor, auf experimentellen Resultaten im Mikroreaktor oder unter An-nahme idealer Bedingungen. Dabei stellten sich die niedrigen Ligninkon-zentrationen sowie der Zersetzung des Methanols und des Ethylacetatsals hauptsächliche Nachteile des aktuellen Konzepts bezüglich der Pro-

X Zusammenfassung

zessökonomie heraus. Durch Kombination mit einer effizienten Wärme-integration im Prozess sowie einer wertvollen Anwendung für die oli-gomerischen Produkte können Verbesserungen an den zuvor genanntenAspekten dazu beitragen, den gezeigten Prozess näher an die industrielleUmsetzung heranzuführen.

Contents XI

Contents

Preface I

Abstract III

Zusammenfassung VII

Nomenclature XV

1 Introduction 11.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . 11.2 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . 31.3 Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

2 Basics of lignin and state of the art 72.1 The structure of lignin and its occurrence in wood . . . . 72.2 Industrial lignin production . . . . . . . . . . . . . . . . . 11

2.2.1 Kraft pulping . . . . . . . . . . . . . . . . . . . . . 122.2.2 Kraft lignin . . . . . . . . . . . . . . . . . . . . . . 142.2.3 Current availability of industrial lignins . . . . . . 152.2.4 Biorefinery concepts . . . . . . . . . . . . . . . . . 16

2.3 Utilization of lignin . . . . . . . . . . . . . . . . . . . . . . 192.3.1 Depolymerization into aromatic chemicals . . . . . 212.3.2 Direct use of lignin as a polymer . . . . . . . . . . 25

3 Experimental methods 273.1 Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

3.1.1 Lignin sources . . . . . . . . . . . . . . . . . . . . 273.1.2 Laboratory chemicals . . . . . . . . . . . . . . . . 28

3.2 Experimental setup and procedure . . . . . . . . . . . . . 303.2.1 Batch reactors for lignin oxidation . . . . . . . . . 30

XII Contents

3.2.2 High pressure microreactor setup . . . . . . . . . . 323.2.3 Membrane plant for separation experiments . . . . 37

3.3 Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . 403.3.1 Identification and quantification of low molecular

weight reaction products . . . . . . . . . . . . . . . 413.3.2 Analysis of molar mass distribution . . . . . . . . 433.3.3 Functional group analysis . . . . . . . . . . . . . . 45

4 Catalytic oxidation of lignin in acidic media 494.1 Investigation of different transition metal salt catalysts . . 50

4.1.1 Main monomeric reaction products . . . . . . . . . 514.1.2 Performance of the catalysts . . . . . . . . . . . . 544.1.3 The catalyst’s mode of action . . . . . . . . . . . . 634.1.4 Variation of catalyst concentration . . . . . . . . . 67

4.2 Composition of the reaction solvent . . . . . . . . . . . . . 694.2.1 Fraction and type of the co-solvent . . . . . . . . . 694.2.2 Degradation of the co-solvent . . . . . . . . . . . . 724.2.3 Acidity of the solvent . . . . . . . . . . . . . . . . 72

4.3 Dependency on lignin feedstock . . . . . . . . . . . . . . . 744.3.1 Acidic oxidation of other commercial lignins . . . . 744.3.2 Conversion of lignins from enzymatic hydrolysis . . 77

4.4 Continuous oxidation in a two-phase flow microreactor . . 814.4.1 Product concentrations in the microreactor . . . . 824.4.2 Influence of temperature . . . . . . . . . . . . . . . 854.4.3 Influence of pressure . . . . . . . . . . . . . . . . . 874.4.4 Influence of lignin concentration . . . . . . . . . . 894.4.5 Parameter estimation of the vanillin kinetics . . . 894.4.6 Simulation of the vanillin concentration . . . . . . 92

4.5 Summary of the lignin oxidation results . . . . . . . . . . 94

5 Extractive recovery of the reaction products 955.1 Liquid-liquid extraction with organic solvents . . . . . . . 95

5.1.1 Potential extraction solvents . . . . . . . . . . . . 955.1.2 Miscibility of the extraction solvents . . . . . . . . 965.1.3 Determination of distribution coefficients . . . . . 985.1.4 Capacity for lignin oxidation products . . . . . . . 101

5.2 Extraction with supercritical carbon dioxide . . . . . . . . 102

Contents XIII

5.3 Conclusions for the extraction step . . . . . . . . . . . . . 103

6 Membrane separation of lignin oxidation products 1056.1 Organic solvent nanofiltration . . . . . . . . . . . . . . . . 1066.2 Characterization of OSN membranes . . . . . . . . . . . . 107

6.2.1 Pure solvent flux . . . . . . . . . . . . . . . . . . . 1076.2.2 Rejection of model compounds . . . . . . . . . . . 111

6.3 Separation of the product mixture . . . . . . . . . . . . . 1146.3.1 Permeate flux . . . . . . . . . . . . . . . . . . . . . 1146.3.2 Monomer rejection . . . . . . . . . . . . . . . . . . 1146.3.3 Separation Performance . . . . . . . . . . . . . . . 117

6.4 Summary of the membrane separation . . . . . . . . . . . 121

7 Downstream processing 1237.1 Separation and purification of monomeric products . . . . 123

7.1.1 Review on existing approaches . . . . . . . . . . . 1247.1.2 Conclusions for the present process . . . . . . . . . 127

7.2 Processing of high Mw products . . . . . . . . . . . . . . 1297.2.1 Characterization of the fraction . . . . . . . . . . . 1297.2.2 Potential applications . . . . . . . . . . . . . . . . 132

8 Conceptual process design and cost estimate 1358.1 Process design and description of the process . . . . . . . 1358.2 Cost estimate . . . . . . . . . . . . . . . . . . . . . . . . . 140

8.2.1 Evaluated cases . . . . . . . . . . . . . . . . . . . . 1408.2.2 Chemicals cost and revenue . . . . . . . . . . . . . 1448.2.3 Cost-effectiveness of the cases . . . . . . . . . . . . 145

8.3 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . 149

9 Conclusions 151

10 Outlook 15510.1 Realization of the current approach . . . . . . . . . . . . . 15510.2 Enhanced monomer yield by a membrane reactor concept 15710.3 Sequential lignin treatment to enhance the monomer yield 16010.4 Study on applications for the oligomeric product fraction 161

XIV Contents

A Appendix 163A.1 Synthesis of components for GC/MS calibration . . . . . . 163A.2 High pressure microreactor setup . . . . . . . . . . . . . . 164A.3 Ternary phase diagrams . . . . . . . . . . . . . . . . . . . 168A.4 Aspen Plus R© process model . . . . . . . . . . . . . . . . . 172

Bibliography 173

List of publications 189

Curriculum Vitae 191

Nomenclature XV

Nomenclature

Abbreviations

APCI atmospheric pressure chemical ionizationBCR benefit-cost ratioBM ball millingBP benzylphenolBPE benzyl phenyl etherBTX benzene, toluene, xyleneCBA cost benefit analysisCoV coefficient of variationDM DuraMemTM

DME dimethyl etherDMSO dimethyl sulfoxideEA ethyl acetateFTIR Fourier transform infrared spectroscopyGC gas chromatographyHPLC high-performance liquid chromatographyHWP hot water pretreatmentIS internal standardLC liquid chromatographyMDHA methyl dehydroabietateMeOH methanolMS mass spectrometryMV methyl vanillateMw molecular weightMWCO molecular weight cut-offM5CV methyl 5-carbomethoxy-vanillateNIST National Institute of Standards and TechnologyNMR nuclear magnetic resonance

XVI Nomenclature

NRTL non-random two-liquidOSN organic solvent nanofiltrationPAN polyacrylonitrilePDI polydispersity indexPDMS polydimethylsiloxanePET polyethylene terephtalatePF phenol formaldehydePI polyimidePM PuraMemTM

POM polyoxometalatePSS poly(styrene sulfonate) sodium saltPU polyurethaneROI return on investmentRT retention timeSEC size-exclusion chromatographySIL siliconeTCD thermal conductivity detectorUV ultravioletV vanillinVis visualVRF volume reduction factor5CV 5-cabomethoxy-vanillin

Roman symbols

a - correlation parameterA m2 areaAi - peak area in GC/MSAs L m−2 h−1 bar−1 solvent permeability constantAU - absorbance unitb - correlation parameterBj L m mol−1 h−1 solute permeability constantci mol L−1, g L−1 concentrationCV % coefficient of variationDi m2 s−1 diffusion coefficient

Nomenclature XVII

Ei J mol−1 activation energyHi,j bar Henry’s law constanthi - peak height in SECJi L m−2 h−1 permeate flux through membraneki s−1 rate constantki,0 s−1 pre-exponential factorKi - distribution coefficientl m membrane thicknessLi m2 mol J−1 h−1 proportionality parameterm g, t massMi g mol−1 molecular weightMw g mol−1 weight average molecular weightMn g mol−1 number average molecular weightn - batch numberp bar pressurepri $/t pricer mol L−1 s−1 reaction rateR J mol−1 K−1 ideal gas constantRi % rejectionSi - sorption coefficientt s, min, h timeT K, C temperatureTB

C boiling pointv m3 mol−1 molar volumeV L, m3 volumexi - fractionx m axial variable

Greek symbols

γi - activity coefficientµ J mol−1 chemical potentialµ(0)i J mol−1 chemical potential in pure state

XVIII Nomenclature

Subscripts

B boilingBP benzylphenolBPE benzyl phenyl etherCat catalystcool coolingDi dimersend end of membrane plant batchfeed feedstockheat heatingi componentiS internal standardj solutek group of products (e.g. monomers, dimers,...)Lig ligninmax maximumM5CV methyl 5-carbomethoxy-vanillateMeOH methanolMon monomersMV methyl vanillaten number-basedOlig oligomersp permeateProd productsr retentates solventSub substrateTri trimersV vanillinw weight-based0 start of membrane plant batch5CV 5-cabomethoxy-vanillin

Nomenclature XIX

Superscripts

E extract phasen batch numberR raffinate phase0 initial runI phase III phase II

1

Chapter 1

Introduction

1.1 Motivation

In order to face the challenges of becoming independent of crude oil andswitching to a more sustainable and carbon neutral society, a paradigmshift in the production of chemicals and fuels towards renewable rawmaterials is imperative. In contrast to energy which can be obtained byseveral "green" technologies including solar power, hydro power and windpower, particularly the production of chemicals relies on carbon sources.In this context, biomass has emerged as most prospective raw material inthe forthcoming future. While first approaches to use biomass for biofuelproduction relied on edible crops like sugar cane, wheat or corn (referredto as 1st generation biofuels), it is generally accepted that prospectiveconversion strategies for biomass have to rely on non-edible feedstocksof large availability. These factors have led to to increasing interest inlignocellulosic biomass and especially wood as promising feedstocks forfuture biorefineries.

The growing interest in wood as renewable resource has provoked ex-tensive research on the valorization of its components in the productionof chemicals and fuels during the last decades [1–3]. Most of this re-search was dedicated to the conversion of cellulose and hemicelluloseinto fermentable sugars for biofuel production (2nd generation biofuels).The sustainable use of lignin as the other major constituent in wood hasgained less attention so far. This is mostly due to the fact, that the ran-dom structure and the various types of linkages in the lignin moleculecomplicate its controlled depolymerization into single chemicals. Thepotential of lignin, however, is huge. Due to its aromatic structure and

2 1. Introduction

the large availability as a renewable carbon source (lignin constitutes toabout 25% of the world’s biomass by weight), the conversion of lignininto aromatic bulk and fine chemicals appears highly attractive.

Currently, lignin accrues in enormous amounts as a by-product of thepulping process in the pulp and paper industry. An estimated 70 milliontons of lignin are processed in the predominant kraft pulping industryalone. The total amount of processed lignin is even expected to furtherincrease as soon as the lignocellulosic ethanol process gets commercial-ized. In contrast to cellulose which is industrially used for the productionof pulp and paper and discussed as potential feedstock for the produc-tion of biofuels, no relevant industrial application for lignin exists sofar. In spite of its attractive chemical composition it is still insufficientlyexploited and mainly regarded as biowaste. Only less than 2% of thelignin in the pulping liquors worldwide is separated from the process asa readily available raw material. Instead, almost all of the lignin is cur-rently burned in the recovery boilers of the pulping industry for internalheat and power generation and the recovery of the pulping chemicals.

The potential of lignin as a renewable resource for the productionof chemicals is well-established in research [3, 4]. Scientific approacheseither focus on the direct utilization of lignin with optional modifica-tion steps to find application in low value products including disper-sants, binders, emulsifiers or resins, or target the depolymerization ofthe lignin macromolecule into high-value aromatic bulk and fine chem-icals. As these approaches either generate products of low commercialvalue or aromatic products at low yield, which require costly separationand purification steps, virtually none of them has reached an industriallevel so far.

Several different transformations for lignin to yield high value aro-matic products can be found in literature [5]. Among those, especiallythe oxidation of lignin into vanillin was evaluated to be highly viable[6]. As a matter of fact, vanillin is the only aromatic chemical thatwas and still remains commercially produced. The alkaline oxidation oflignosulfonates dissolved in the black liquor of sulfite pulp mills yieldsabout 15wt% of vanillin and accounted for 60% of the world supply ofvanillin in the 1980’s [7]. However, environmental and economic reasonsled to the closure of virtually all of the production plants till today [8].In contrast to lignosulfonates, the alkaline oxidation of the by far more

1.2 Objectives 3

abundant kraft lignin does not yield vanillin in reasonable amounts (yield<4wt%) [9]. Hence, novel methods are required to depolymerize kraftlignin into vanillin and/or other functionalized phenols.

In a recent work of Voitl, the acidic oxidation of kraft lignin in amethanol/water solvent with polyoxometalates as catalysts was proposedas a novel depolymerization concept [10]. It was proven superior to com-mon alkaline oxidation as higher yields of vanillin can be obtained andmethyl vanillate results as an additional monomeric reaction product.Nevertheless, more than 90% in the mixture of reaction products re-mained unused. A complete valorization of lignin as a renewable rawmaterial, however, demands its complete utilization from both, an eco-nomic and an ecological point of view. Thus, a concept for a processthat can completely and competitively convert (kraft) lignin into differ-ent value-added chemicals is required to launch its commercial successon industrial scale.

1.2 Objectives

Within the scope of the present thesis, the main goal is to develop aconceptual design for a process that is able to fully convert lignin intodifferent valuable chemicals. Based on the acidic oxidation reaction de-scribed above, the complete valorization of lignin is intended by usinga twofold approach with two different routes of products. On the onehand, aromatic monomers like vanillin and methyl vanillate that canbe separated and purified and used as high value chemicals or chemicalbuilding blocks thereof. On the other hand, a mixture of non-monomericproducts that can optionally be modified by functionalization and lateron find application in products like dispersants, binders, emulsifiers orresins. In this context, research towards utilization of the latter reactionproducts was conducted in collaboration with Sika Technology AG as in-dustrial partner. The goal of this collaboration was to chemically modifythe fraction of non-monomeric products for the application as water re-ducers in concrete applications. The general concept which depicts thetwofold approach is shown in Figure 1.1.

Before a potential process can be presented at the end of this thesis,its basic unit operations have to be investigated. First of all, the acidic

4 1. Introduction

LigninLigninReaction Reaction

productsproductsAcidic

oxidation

Fractionation

MonomersMonomers

NonNon--monomericmonomeric

products*products*

ModifiedModified

products*products*Chemical

modification

Separation /

purification

Monomeric Monomeric

chemicalschemicals

*for applications like dispersants, binders, emulsifiers or resins.

Figure 1.1: Basic concept for the complete valorization of lignin.

oxidation needs to be studied in detail. A reaction environment (catalyst,pH, solvent composition) that is able to efficiently depolymerize kraftlignin and other prospective lignins into aromatic chemicals is required.As the value generation is primarily based on the aromatic monomers,the reaction has afterwards to be optimized concerning temperature,pressure and lignin concentration to produce a maximum of monomericproducts. The formerly used batch reactor is, however, not suited fora broad and laborious study on these reaction conditions. Instead, acontinuous high pressure microreactor setup is supposed to overcomeany drawbacks of the batch reactor.

For the following processing of the reaction products, process stepsto recover and to separate the reaction products need to be found andinvestigated. In this context, a complete product recovery from the reac-tion solvent and a sharp separating cut of monomers and non-monomericproducts during fractionation is targeted. The downstream processing,especially of the mixture of monomeric products, to obtain marketableproducts has to be addressed at the end of thesis as well as the prospec-tive cost and revenues. For the characterization of the feedstock and thelignin products before and after any investigated process step, a com-plete set of analytical tools is mandatory and needs to be established inthe beginning of the thesis.

1.3 Outline 5

1.3 Outline

The present thesis is structured in the same order as the main unit op-erations of the process which is investigated within it. A comprehensivebackground on lignin origin, lignin production and current approachesfor lignin utilization is provided in Chapter 2. Chapter 3 describes allexperimental setups and analytical methods used within this work. Thefollowing chapters address the oxidation reaction (Chapter 4), the recov-ery of the reaction products by extraction (Chapter 5), the fractionationof the products (Chapter 6) and the downstream processing of the prod-ucts (Chapter 7). As each of the chapters on the unit operations coversrelated background information and relevant literature, any of them canbe read individually. The last part of the thesis comprises a detaileddiscussion on the process and estimated costs in Chapter 8, followed bya conclusion and an outlook in Chapters 9 and 10, respectively.

7

Chapter 2

Basics of lignin and state of the

art

Lignin has been subject to extensive research for more than one centurywith results documented in a virtually unlimited number of publicationsand reviews. To not exceed the scope of the present thesis, the emphasisof this chapter is on the most relevant fundamentals of lignin and itsutilization. In the beginning, the nature of lignin and its structuralproperties are described. Afterwards, the industrial production of ligninduring pulping is shown and the current and prospective utilization byeither depolymerization or direct use including the relevant literature isdiscussed.

2.1 The structure of lignin and its

occurrence in wood

Lignin is, besides cellulose and hemicellulose, one of the three majorconstituents of vascular plants. It fills the spaces between cellulose andhemicellulose in the secondary cell walls and acts like a resin holding thelignocellulosic matrix together. The covalent linking with the carbohy-drate polymers confers on the plant strength and stiffness and therebyrenders possible its growth [11]. Depending on the wood species, thelignin content in a tree can vary between 15% and 40% by weightwhereas softwoods are known to contain higher values (typically 27%–31%) than hardwoods (19%–25%) [12]. With an annual production onearth in the range of 20 billion tons, lignin is the second most abundant

8 2. Basics of lignin and state of the art

raw material in nature after cellulose [13]. Figure 2.1 depicts the locationof lignin in wood.

Cellulose

Lignin

HemicellulosePlant cell

Plant cell wall

Lignin molecule

Figure 2.1: Occurrence of lignin in the plant cell walls of a tree [14, 15].

Unlike other natural biopolymers lignin is a rather randomly cross-linked macromolecule whose native structure has not been completelyelucidated yet. Today, it is widely accepted though, that the biosyn-thesis of the lignin molecule results from the random polymerization ofthe three aromatic monolignol units depicted in Figure 2.2, namely p-coumaryl alcohol (hydroxyphenyl unit), coniferyl alcohol (guaiacyl unit)and sinapyl alcohol (sinapyl unit). The radical polymerization process ofthese phenylpropane units is initiated by an enzyme-catalyzed oxidationof their phenolic hydroxyl group, resulting in a random cross-linking andbuild-up of the amorphous three-dimensional molecule [16]. While soft-wood lignin almost exclusively stems from the polymerization of coniferylalcohol (> 95%; < 5% of coumaryl alcohol), hardwood lignin is mainlycomposed of sinapyl units (45% to 75%) with lower ratios of coniferylalcohol (25% to 50%) and coumaryl alcohol (0% to 8%) [12]. Untiltoday, the lignin structure that results from the polymerization of themonolignols is not fully understood. The isolation of lignin from woodfor analytical purposes causes undesirable changes in its structure andthereby complicates the determination of the native structure in wood.

The research on the structure of lignin and its biosynthesis dates backto the 1950’s and 1960’s. Freudenberg and co-workers conducted compre-hensive and fundamental investigations and summarized their findingsin a textbook [17]. Based on their analytical data, they came up with

2.1 The structure of lignin and its occurrence in wood 9

OH

OH

OH

OH

O

OH

OH

OO

12

34

5

6

αβ

γ

p-coumaryl alcohol coniferyl alcohol sinapyl alcohol

Figure 2.2: Monolignol units as precursors for lignin biosynthesis andnomenclature of carbon atoms.

an early version of the structure of softwood lignin. Adler reviewed theexperimental data on lignin available in scientific literature in the endof the 1970’s and presented a structure which is widely accepted untiltoday [18]. This structure with its main functional groups highlighted isschematically depicted in Figure 2.3.

O O

O

OH

O

OH

OH

O

OH

O

O

O

OH

O

O

O

O

O

OH

O

O

O OH

O

O

OH

O

O

OHO

OHO

OH

O

O

O

O

OHO

OH

OO

OH

OO

O

Aliphatic hydroxyl groups

Phenolic hydroxyl groups

Methoxyl groups

Carbonyl groups

Figure 2.3: Schematic structure of a softwood lignin molecule withmain functional groups highlighted.

As a result from the random polymerization during lignin synthesis,the type of linkages between the aromatic units in a lignin molecule


varies between softwood and hardwood in the same manner as molecu-lar weight and functional group content. However, aryl ether linkages,especially those of the β-O-4 type, are the most dominant ones in thestructures of both lignins. The main interunit linkages are illustrated inFigure 2.4 with their abundance for spruce (softwood) and birch (hard-wood) listed in Table 2.1 [18]. In addition to these linkages, the aromaticlignin molecule can also contain branched and crosslinked structures.The more condensed the structure of lignin is, the more difficult it is todissolve lignin during chemical pulping and to degrade it later on for theproduction of aromatic monomers.

OOH

R

R

O

R

OH R

β-O-4

R

O

R

OH

OHO

R

R

α-O-4

O

OR

O

OH

4-O-5

R

O

R

OH

O

O

β-5

R

O

R

OO

O

R R

β−β

OH

O

OH

O

OH

R

O

R

OH

O

O

R

5-5 β-1

Figure 2.4: Most frequent linkages (bold) between aromatic units innative lignin molecules.

The structure of the lignins that are liberated from the lignocellulosicmatrix during pulping and available on the commercial market (often re-ferred to as "technical lignins") are, however, distinct from the structureof natural lignin. It is almost equally determined by the botanical originand the method of isolation [19]. Type and abundance of both, interunitlinkages and functional groups are subject to change when lignin is sep-arated from the other biomass components. Depending on the isolationtechnology used, different groups can be introduced (e.g. thiol groups

2.2 Industrial lignin production 11

Table 2.1: Major interunit linkages in lignin and their proportions insoftwood and hardwood.

Linkage type Softwood (spruce) [%] Hardwood (birch) [%]

β-O-4 48 60α-O-4 6–8 6–84-O-5 3.5–4 6.5β-5 9–12 65-5 9.5–11 4.5β-1 7 7β-β 2 3Others 13 5

in kraft pulping) or withdrawn from the molecules [20]. As the contentof functional groups determines the properties (especially the solubility)and the performance of lignin in specific applications, the pulping pro-cesses already predetermine potential applications of the isolated lignins.The isolation of lignin by pulping with focus on kraft pulping as well asthe resulting changes in lignin structure are discussed in the following.

2.2 Industrial lignin production

In contrast to cellulose which is commercially used for pulp and paperproduction and discussed for the production of bioethanol in the nearfuture, lignin has very limited applications as a chemical and is not in-tentionally produced in industry. It rather originates in huge amountsas a side product when cellulose is isolated from the lignocellulose ma-trix during the pulping process. Various isolation technologies based onchemical or mechanical treatment exist for the separation of celluloseand hemicellulose from lignin. While mechanical treatments are usedto mechanically separate the constituents from each other, the predom-inant chemical processes usually employ harsh process conditions andpulping chemicals that uniquely alter the structure of lignin [3]. They


either aim at the removal of cellulose and hemicellulose by solubilizationleaving an insoluble lignin fraction behind or vice versa. Lignin frompotential biorefineries that produce ethanol by fermentation of cellulosicsugars is derived from processes of the first category. The pulp and pa-per industry on the other hand uses chemical pulping processes resultingin an insoluble fibrous pulp and lignin-rich black liquor. Kraft pulpingand to a minor extent sulfite pulping are the predominant processes forwood in industry. Other pulping processes are so far mainly used in pilotscale plants or in research. As the scope of the present thesis is on thevalorization of the abundant kraft lignin from the kraft process which isalmost exclusively used in the experiments, the other pulping processesand the resulting lignins are not discussed in detail. Reference is madeto the literature [19, 21, 22].

2.2.1 Kraft pulping

The kraft process is the by far most prominent pulping process in indus-try and accounts for 89% of the chemical pulp production [21]. The highquality of the pulp including its high mechanical strength, the simplicityand rapidity of the process as well as technological advances have ledto its predominant position [20]. Among the latter, the development ofrecovery furnaces by Tomlinson that enable the recovery of the pulpingchemicals and the continuous multi-stage bleaching process have to bementioned. A scheme of the kraft process in the paper industry includingthe main operation units is depicted in Figure 2.5 [23].

In the kraft pulping process, wood chips are heated in a highly al-kaline solution of sodium hydroxide and sodium sulfide from approxi-mately 70 C to 170 C and cooked at this temperature for 1 h to 2h[20]. During this treatment, the lignin macromolecule is broken downby action of the hydroxide and hydrosulfide ions in the pulping liquor("white liquor") and results in smaller water/alkaline-soluble fragments[16]. Consequently, the kraft lignin is dissolved in the pulping liquorswhereas cellulose can be filtered off in the form of pulp. The resultingso-called kraft "black liquor" from the kraft treatment is in the nextstages concentrated by evaporation and used as fuel in the recovery boil-ers. Kraft pulp mills therefore operate as highly integrated facilities thatrely on lignin as an energy source. Since the present infrastructure of the


Lime kiln

Power generation

Recovery

boiler

Bleaching

DelignificationCooking

Wood yard

Recausticizing

Oxygen plant

Paper mill

Black

liquor

White

liquor

Bleaching

chemicals

Evaporators

Figure 2.5: Overview of a kraft process in the paper industry [23].

process was optimized throughout the more than 130 years of operation,the recovery of lignin has not been practiced broadly until recently.

While the recovery boilers are essential to the environmental and eco-nomic performance of the kraft pulp mill, energetic improvements haveled to the fact that current pulping processes are often limited by theenergy load that can be handled therein [24]. In this case, a capacityincrease in pulp production is only possible when the thermal load to therecovery boiler is decreased. As modern kraft mills operate with an en-ergy excess, a debottlenecking is possible in most of the present mills bypartly removing the energy surplus from the black liquor in form of solidlignin [25]. Although this precipitated kraft lignin is often discussed asadditional solid fuel in other areas of the pulp mill, it represents a mar-ketable product and an important raw material for chemical valorization[19].

Different separation techniques to remove lignin from the kraft blackliquor have been demonstrated in the literature. The most common oneis the precipitation of kraft lignin by acidification with carbon dioxide toa pH of 10 to 11 and subsequent filtration [24]. Another approach is its


concentration and isolation by ultrafiltration [26]. Recently, the so-calledLignoboost process, which removes lignin by the former method, hasgained much attention and a pilot plant was successfully implementedin a Swedish pulp mill in 2007 [27]. It is capable of separating 25%of the lignin in the black liquor (corresponding to 24 t d−1), therebyenabling 20% to 25% more pulp production [28]. The first commercialLignoboost plant will start its operation in the beginning of 2013. Ifthis process continues succeeding, the amount of commercially availablekraft lignin can be expected to increase and kraft lignin will get moreand more important as a renewable raw material and feedstock.

2.2.2 Kraft lignin

As previously mentioned, lignin available from the kraft pulping industryis very distinct from the original lignin found in plants. By action of thewhite liquor, the fragmentation of the lignin molecule in combinationwith the introduction of thiol groups from hydrosulfide anions renderkraft lignin soluble. The fragmentation of lignin during kraft treatmentmainly proceeds by cleavage of the α-O-4 and β-O-4 linkages betweenthe aromatic units. 80% to 85% thereof were found to be cleaved dur-ing kraft pulping of softwood resulting in an increase in phenolic hy-droxyl groups [29]. All other interunit linkages in lignin including 4-O-5aryl ether structures basically survive the harsh conditions in the kraftprocess [20]. Apart from the fragmentation of the lignin molecule, con-densation reactions which recombine lignin fragments also occur duringkraft pulping. Thereby alkali-stable carbon-carbon linkages of the diarylmethane type are formed. To illustrate the typical chemical reactionsduring kraft pulping, Figure 2.6 depicts the cleavage of β-O-4 linkages aswell as the primary condensation during kraft cooking [16]. Due to thehigh amount of stable carbon-carbon bonds which either remain unaf-fected or are even formed during kraft cooking, kraft lignin precipitatedfrom black liquor represents a highly refractory material which is hardto depolymerize into aromatic units by chemical or biological processes.

Analytical results of the molecular mass distribution usually report anaverage molar mass Mw of 2000 g mol−1 to 3000 g mol−1 with polydis-persities (PDI) from 2 to 3 for kraft lignin [30]. The functional groupcontent and other characteristics (e.g. elemental composition, impuri-


CH

O-

CH

CH2OH

O

H3CO

H3CO

OH CH

O

CH

CH2OH

O

H3CO

H3CO

CH

O-

CH

CH2OH

O

S-

H3CO

H3CO

CH

O-

CH

CH2OH

S

H3CO

O-

H3CO

CH

O-

CH

CH2OH

H3CO

Sulfidolytic cleavage of β-aryl ether bonds:

-OH +HS, -H+

-S

CH

O

CH2

CH3

H3COO

C-

OCH3

H CH

O-

CH2

CH3

H

O

H3CO

OCH3

-H+

CH

O-

CH2

CH3

O-

H3CO

OCH3

Primary condensation:

Figure 2.6: Cleavage and recombination of linkages in lignin duringkraft pulping.

ties) of kraft lignin are also well-characterized in the literature [31]. Inpart, these results will be listed and discussed in the context of the ownanalytical results in Chapter 7. A schematic structure of a kraft pinelignin molecule was presented by Gargulak and Lebo and is depicted inFigure 2.7.

2.2.3 Current availability of industrial lignins

Out of the approximately 20 billion tons of lignin that are annuallysynthesized in nature, an estimated 70 million tons arise each year inkraft pulp mills during the production of pulp and regenerated fibers[33]. However, as kraft pulp mills have evolved as highly energeticallyintegrated facilities, more than 99% of the so-called kraft lignin fromthe process is not recovered for industrial application but burned in therecovery boilers for the recovery of pulping chemicals and the provisionof energy. Only about 60 000 tons of kraft lignin per year are commer-


SH

O

OH

O

O

OH

O

OH

OHO

OH

O

O

OH O

O O

O

OH

O

OH

OH OH

OH

O

OH

OH

Figure 2.7: Schematic structure of a pine kraft lignin molecule as pre-sented by Gargulak and Lebo [32].

cialized, virtually all of it by MeadwestVaco in the United States [19].In combination with one million tons of lignosulfonates from the sulfitepulping industry and about 10 000 tons from the soda pulping industry,the total marketed amount of lignin is a little lower than 1.1 million tonsannually. This accounts for less than 2% of the total amount of processedlignin [22]. Production of other lignins which also employ different pulp-ing/isolation technologies is so far only conducted on small scale in pilotplants or in research and technical development. An overview of themain current lignin producers (production of ≥ 0.5kt/y) is depicted inTable 2.2.

2.2.4 Biorefinery concepts

The reduction of greenhouse gas emissions and the need for becom-ing independent of limited fossil resources triggered intensive researchon the possibility to replace current crude oil based feedstocks in re-fineries by renewable lignocellulosic biomass [2, 35]. Just as the current


Table

2.2

:M

ajor

indu

stri

allig

nin

sour

ces

(Dat

aof

2011

)[3

4].

Lig

nin

type

Sour

ceSc

ale

ofop

erat

ion

Vol

ume

[kt/

y]Su

pplie

rs

Lig

nosu

lfona

tes

Soft

/har

dwoo

dC

omm

erci

al∼

1000

Bor

rega

ard

(NO

),T

EM

BE

C(F

R,U

S),

Dom

sjö

Fabr

iker

(SE

),La

Roc

hett

eVen

izel

(FR

),N

ippo

nPap

er(J

P)

Kra

ftlig

nin

Softw

ood

Com

mer

cial

60M

eadw

estV

aco

(US)

Soda

ligni

nN

on-w

ood

Com

mer

cial

5–10

Gre

enVal

ue(C

H,IN

)

Kra

ftlig

nin

Softw

ood

Pilo

t8

Lig

nobo

ost/

Met

so(S

E)

Org

anos

olv

Har

dwoo

dP

ilot

0.5–

3Lig

nol(C

A),

Frau

nhof

er(D

E),

Ded

ini(B

R)

Org

anos

olv

Stra

wP

ilot

0.5

CIM

V(F

R)

Hyd

roly

sis

ligni

nN

on-w

ood/

woo

dP

ilot

0.5

SEK

AB

(SE

)

Hyd

roly

sis

ligni

nC

rop

resi

dues

Pilo

t0.

5In

bico

n(D

K,U

S),

Che

mte

x(I

T,U

S,C

N)


petroleum refineries, a so-called biorefinery is intended to produce fuelsand chemicals at the same time. Extensive knowledge on the productionof bioethanol from cellulose by (enzymatic) saccharification and subse-quent fermentation has already been established in research. On theother hand, technologies that are able to at least partly convert lignininto value-added chemicals instead of just burning it are still lacking [3].The production of fuels and chemicals from biomass, however, is an eco-nomical necessity for the realization of biorefineries [36]. Moreover, thevalorization of all biomass components including lignin is essential andtherefore new processes to convert sulfur-free lignin from biorefineriesinto high-value chemicals are imperative. A scheme of possible processstreams in a lignocellulosic biorefinery is depicted in Figure 2.8 [2].

Current concepts for lignocellulosic feedstock biorefineries mostly tar-get bioethanol production from cellulose and hemicellulose by hydrolysisand fermentation. An enzymatic process is generally regarded as themost attractive approach for the hydrolysis step. The rather ambientprocess conditions during the enzymatic treatment are economically ad-vantageous since less cellulose is degraded and higher sugar yields areobtainable [37]. The action of the cellulolytic enzymes causes a disso-lution of the derived sugars and thereby a separation of the biomasscomponents. The recalcitrance of the lignocellulosic matrix, however,inhibits an enzymatic access and makes a pretreatment of the wood nec-essary.

Various pretreatments exist that can be categorized into physical,physico-chemical, chemical and biological pretreatments. They are sum-marized and evaluated in several reviews, e.g. [38, 39]. Due to theabsence of any chemicals except water, steam pretreatment and hot wa-ter pretreatment have emerged as simple and low-cost options for fu-ture biorefineries. Lignins arising from steam pretreatment/hot waterpretreatment and following enzymatic hydrolysis of the lignocellulosicbiomass are supposed to perform better than current lignins in most ap-plications as they possess a lower molecular weight (Mw=1000 g mol−1–2000 g mol−1), are less polydisperse, and sulfur-free. Although the wholebiorefinery development is still in its infancy, these lignins need to be con-sidered as prospective feedstocks for a valorization process of lignin intoaromatic chemicals.

2.3 Utilization of lignin 19

Figure 2.8: Possible process scheme of a lignocellulosic biorefinery [2].

2.3 Utilization of lignin

As stated above, the present applications of precipitated and marketedlignin that make use of its chemical value are limited. The existingmarkets are based on lignosulfonates and cover mostly either low valueapplications or niche segments [22]. Due to the energetically integratedkraft pulp mills that commercialize only low amounts of kraft lignin, ithas so far received little attention with regard to its valorization.

Currently, lignosulfonates from the sulfite pulping industry representthe major fraction of commercialized lignin with an annual production ofabout one million tons. As of 2008, 67.5% of lignosulfonates were used


for dispersant applications and 32.5% for binder and adhesive applica-tions [40]. Concrete admixtures were the leading market, accountingfor 38% of the world consumption. As concrete water reducers, thedispersion capacity of lignosulfonates allows for an increase in worka-bility and better strength properties of concrete with a lower amountof water required at the same time [19, 41]. Other selected dispersantapplications are in gypsum wallboards, agrochemicals, dyestuffs and pig-ments. As binder, lignosulfonates have the most significant market (13%of world consumption in 2001) within animal feed where they supportpellet manufacturing [42]. Selected additional binder applications are inroad construction, in dust control, in fertilizers, in mining and in carbonblack.

The present applications of commercialized kraft lignin cover roughlythe same markets as lignosulfonates with dominant applications as dis-persants and emulsifiers. In order to compete with water-soluble ligno-sulfonates in some of the markets, the hydrophobic kraft lignin has to bechemically modified prior to utilization. Sulfonation and sulfomethyla-tion are the predominant modifications which result in increased watersolubility of kraft lignin. Due to the higher costs, modified kraft ligninhas not entered the major lignosulfonate markets yet, but only nicheareas with low volumes [43]. Mainly dyes and agrochemicals currentlybenefit from the dispersant properties of sulfonated kraft lignin. Unmod-ified kraft lignin is a good stabilizer for oil-water emulsions in asphaltapplications but can also be used as antioxidant, rubber reinforcer andcarbon black. In search for a reasonable industrial utilization of kraftlignin, the use in thermoplastics (e.g. Arboform R©) emerged as a promis-ing additional application during the last decade [44].

Apart from the utilization of precipitated and commercialized ligninfrom the pulping process, two chemicals are directly produced on site onsmall scale from the pulping black liquors. Vanillin is still made fromsulfite black liquor as the only aromatic chemical from lignin and willbe discussed in detail below. Kraft black liquor, on the other hand, isused in small amounts for the industrial production of dimethylsulfoxide(DMSO). The reaction requires the addition of sulfur and high tem-peratures to form dimethylsulfide which is subsequently oxidized withnitrogen oxide to DMSO [19].

In addition to the existent applications of lignin, numerous poten-


tial utilizations are investigated and discussed in research. They eitheraddress the depolymerization of the lignin molecule into valuable aro-matic monomers or the direct use as polymer. Figure 2.9 illustratespotential routes for lignin valorization from the literature. The differ-ent approaches including their products are discussed in detail in thefollowing.

Lignin

Energy

Vanillin

Vanillic acid

Aromatic acids

Aromatic aldehydes

Aliphatic acids

DMSO

Combustion

Depolymerization into aromatics

Polymer

mixture

Dispersants

Binders

Adhesives

Emusli!ers

Resins

Plastics

Carbon !ber

Antioxidants

Oxidation

Pyrolysis

Synthesis gas

Gaseous

hydrocarbons

Alcohols

Ketones

Phenols

Hydrogenolysis

Hydrolysis

Benzene

Toluene

Xylene

Cresol

Catechols

Syringols

Guaiacols

Vanillin

SH

O

OH

O

OOH

O

OHOH

O

OH

OO

OH

OO

O

O

OH

O

OH

OH

OH

OH O

OH

OH

Figure 2.9: Potential routes for lignin utilization and correspondingproducts.

2.3.1 Depolymerization into aromatic chemicals

The knowledge about the aromatic nature of lignin and the possibilityto chemically degrade it into aromatic products was already present toFreudenberg in the end of the 1920’s [45]. In fact, Singer was the first onein 1882 who found vanillin as a product when cooking spruce wood forseveral hours in hot water [46]. During these times the academical scopewas to clarify the structure of lignin by degradation reactions. Sincethen, scientists more and more recognized the huge potential of lignin asa renewable carbon source for the transformation into aromatic fine andbulk chemicals.

Various reviews were published over the decades discussing potential


routes for the transformation of lignin into useful chemicals, prominentones of them are [3, 47–50]. The huge variety of thermochemical treat-ments for the fragmentation of lignin is principally divided into fourcategories: pyrolysis, hydrolysis, hydrogenolysis and oxidation. Delim-itations in terminology of the categories are, however, not always ob-vious and additional terms for lignin degradation treatments are com-mon. Recent progress in lignin degradation with ligninolytic enzymesalso promoted biotechnological processes as an alternative way for lignindepolymerization. As the scope within the present thesis is on chemicaltreatments, this approach will not be discussed but further informationcan be found in the literature [51]. The four predominant thermochemi-cal treatments are briefly summarized in the following. For more detailedinformation on the treatments, reference is made to the previously men-tioned literature reviews.

Pyrolysis

In pyrolysis, the lignin molecules are broken down by applying temper-atures above 200 C in the absence of air. As pyrolysis temperaturescover a wide range of temperatures up to 1000 C, the products whichare obtained generally depend thereon. The major products include butare not limited to gaseous hydrocarbons with carbon monoxide and car-bon dioxide, volatile liquids (e.g. acetone, methanol, water) as well assubstituted aromatic monomers (e.g. phenol, guaiacol, catechol) [52]. Inaddition, char and high-boiling complex phenols are found among theproducts. The total yield of monophenols is generally below 15%. Attemperatures above 700 C, lignin can be converted to synthesis gas andused as feedstock in Fischer-Tropsch synthesis.

Hydrolysis

Hydrolysis (often also referred to as solvolysis) aims at the fragmenta-tion of lignin by aqueous solvents at moderate temperatures mostly incombination with the action of an acid or a base. Thereby, the interunitlinkages in lignin are cleaved resulting in products of lower molecularweight. As condensation of carbonium ions causes repolymerization re-actions in neutral and acidic media, char is a predominant product [49].


Alkaline hydrolysis is more applicable and yields a broad mixture of,amongst others, catechol, syringol, guaiacol and vanillin [48]. The re-ported yields of phenolic products in the mixture, however, do usuallynot exceed 10%.

Hydrolysis in supercritical water is reported as an alternative approachfor lignin fragmentation [53]. Supercritical water offers several advan-tages, e.g. the occurrence of oxidation and hydrolysis reactions withouta catalyst, its thermal stability as well as the miscibility with gases, hy-drocarbons and aromatic substances. Reactions with model compounds(guaiacol and catechol) have shown the ability to produce phenol underthe supercritical conditions. The gasification of lignin into carbon diox-ide and hydrogen with suppression of char formation was also reportedunder supercritical conditions.

Hydrogenolysis

As the hydrolysis of lignin results in product mixtures with low monomeryields, reductive (hydrogenolyis) and oxidative (oxidation) conditionsduring the reaction have become more common. Thereby, phenolicfunctionalities can either be removed towards bulk chemical productsor added to target fine chemicals. In hydrogenolysis, hydrogen-donatingsolvents (e.g. tetralin) or a hydrogen atmosphere are used to obtainsimple monomeric components like BTX, phenol and cresol. At temper-atures from 250 C to 600 C and H2 pressures from usually 30bar to460bar, even more than 30% monophenols are reported in the literature[52]. Some of these results are doubtful though, as they could not bereproduced or part of the solvent reacted to phenolic products [3]. Morereliable results report high oil yields whereas monophenol compoundssum up to 15% and the maximum yield of single phenolics (phenol andcresol) is less than 4% [54]. Due to the low selectivities of phenolic prod-ucts in the lignin-derived oils and the low prices of those products in thecompeting petrochemical industry, valorization of lignin by hydrogenol-ysis has not reached an industrial level so far.


Oxidation

The oxidation of lignin adds functionality to the aromatic rings duringfragmentation and aromatic aldehydes and acids (hydroxybenzaldehyde,vanillin, syringaldehyde and corresponding acids) related to the ligninprecursor molecules (compare Figure 2.2) are obtained. Among them,especially vanillin is a prominent product of commercial interest whichis predominantly obtained by the oxidation of softwood lignin being richin guaiacol units from coniferyl alcohol. The vanillin demand and itseasy formation from lignin compared to sophisticated petrochemical pro-duction routes have made oxidation the only chemical depolymerizationtreatment of lignin that brought it to a commercial scale till today.

Since 1936, vanillin is produced as high-value aromatic product byalkaline oxidation of lignosulfonates at a reasonable yield of about 15%[7, 55]. With an annual consumption of about 12 000 t, vanillin representsone of the most widely used flavoring in food and perfumes [56]. From thewaste sulfite liquors containing the lignosulfonates, vanillin is producedby cooking lignin for 4min at 225 C with 105bar of air as oxidant. Whilein the 1980’s one Canadian plant produced 60% of the world supply ofvanillin, from 1993 Borregaard Lignotech remained the only commercialproducer of vanillin from waste sulfite liquors with an annual productionof 1500 t (≡ 12.5% of world supply) [57]. Severe environmental problemsand economic reasons forced the closure of most of the production plantsaround 1990 [8]. Moreover, sulfite pulping is a declining process due tothe inferior strength properties of sulfite pulp compared to kraft pulp.The decrease in feedstock availability in combination with the problemsin economy and ecology therefore demand for alternative routes in non-petrochemical vanillin production.

Its huge abundance and still unexploited utilization towards valuablechemicals would promote kraft lignin as potential feedstock in this con-text. However, owing to the condensed structure of kraft lignin, itsalkaline oxidation is in contrast to lignosulfonates not suited to yieldvanillin in reasonable amounts. Villar et al. reported a yield of lessthan 5% aldehydes (syringaldehyde + vanillin) for the conversion of ahardwood kraft lignin in sodium hydroxide solution with copper or cobaltsalts as catalyst [9]. Nevertheless, the oxidation of vanillin was evaluatedto be highly viable and represents the most promising approach among


the treatments discussed above [6]. Therefore, improved processes forvanillin production from kraft lignin can introduce a new era of vanillinproduction from renewable resources and valorize the huge amounts ofkraft lignin in a sustainable way.

2.3.2 Direct use of lignin as a polymer

The thermochemical conversion technologies of lignin as discussed abovegenerally result in polydisperse product mixtures with a broad molecu-lar weight distribution. Unless the costs for the following separation andpurification steps are kept on a low level, none of the approaches can beeconomically competitive. A straighter way to chemically use lignin isthus by employing it as a macromolecular mixture in products of lowerpurity requirements and lower value. Since some of the products requirespecial functionalities of lignin and/or its solubility in water, the follow-ing utilizations of lignin may require a modification/functionalizationstep before use. Although neither available in uniform quality nor insignificant amounts yet, sulfur-free lignins from biorefinery-related treat-ments are often preferably discussed in the literature because they offer agreater versatility and can be heat-processed without the irritating odorof sulfur-derived compounds as observed for kraft lignin [4].

Plenty of current and prospective utilizations for undegraded ligninare discussed in a multitude of literature articles and reviews (e.g. [4,33, 43, 58]). As already mentioned in this chapter, lignin presently findsapplication as dispersant, binder, and emulsifier in several industries.While these current uses of lignin mainly include products of lower value,some of the potential uses also target materials with higher revenues.

The majority of discussed applications for future lignin utilization inthe literature suggest the use in different kinds of resins or plastics [33].The similarity of lignin to phenol-formaldehyde (PF) resins make it anobvious candidate as a phenol substitute therein. Lignin can partlyreplace phenol from petrochemical sources in PF resins with additivelevels of up to 35%, depending on the type of lignin. PF resins have abroad range of industrial applications and are among the most widelyused plastics in the world. In this context, lignin is also discussed asadhesive in panel boards and fiberboards where PF resins are commonlyemployed. A similar function as in PF resins can be fulfilled by lignin in


epoxy resins where it can substitute the activator (e.g. bisphenol A ortriethylenetetramine). An IBM research group successfully incorporatedlignin into epoxy resins for the fabrication of printed circuit boards [58].Alkoxylated lignin has been tested as polyol in polyurethane foams. Par-tial and complete substitution of the commercial polyol by alkoxylatedlignin resulted in similar foam properties [4]. In addition to the above-mentioned Arboform R© thermoplastic, several literature studies deal withthe utilization of lignin as additive in polyolefins. Properties that can bebeneficially affected by lignin addition include improved strength prop-erties, good surface properties, recalcitrance to biodegradation and costreduction. Additional lignin applications are in the field of carbon fibers,activated carbon and antioxidants. For more detailed information andfurther niche applications, reference is made to the above-mentionedcomprehensive literature.

27

Chapter 3

Experimental methods

In this chapter, all experimental methods that were used within thepresent thesis are introduced. First of all, the different types of ligninsthat were utilized in the experiments including their natural sourcesand commercial distributors are presented. The laboratory setups forthe various experiments and the related experimental procedures aredescribed in detail in the second section of the chapter. In the end, allanalytical methods that were internally or externally conducted in thecontext of this thesis are described.

3.1 Materials

3.1.1 Lignin sources

As the primary goal of the present thesis was on the valorization of theabundant kraft lignin, a commercial unmodified kraft lignin, namely In-dulin AT, from MeadwestVaco (Charleston, USA) was used throughoutthe majority of experiments. Indulin AT is obtained from the kraft pulp-ing of pine and sold at a purity of 97% (dry) with a moisture content of5%. It is well-characterized and described in the literature [59, 60]. Thesolubility of unmodified kraft lignin is limited to alkaline pH and it is inparticular not soluble in the acidic reaction solvent. A sulfonated kraftlignin (lignin, alkali, low sulfonate content) from spruce (Sigma-Aldrich,Buchs, Switzerland) was therefore additionally used for comparison pur-poses and for the microreactor experiments in which prevention of clog-ging is crucial. Unlike Indulin AT, the latter is also completely soluble ina neutral aqueous solution which was used as feedstock in the microre-

28 3. Experimental methods

actor experiments.Other types of commercially available lignins (compare Table 2.2) were

exclusively used to study the feasibility of using them as alternativelignin feedstock in the final reaction system with regard to the type andamount of reaction products. A commercial lignosulfonate from sprucenamed Ultrazine NA (Borregaard Lignotech, Norway), a soda lignin fromannual crops known as Granit Bioplast (Granit SA, Switzerland) and anorganosolv lignin from poplar (Fraunhofer, Germany) were used in theseexperiments.

Moreover, three lignins were produced in the laboratory by enzymatichydrolysis of lignocellulosic biomass in combination with different pre-treatment methods. Due to the mild isolation conditions, these ligninsresemble potential biorefinery lignins that are expected to be availablein the future. Sawdust of Swiss spruce was sieved to a particle size ofless than 400µm and either subjected to hot water pretreatment, ballmilling at 400 rpm or ball milling at 200 rpm. In hot water pretreat-ment, 20 g of sieved spruce sawdust in 1L of water were heated undernitrogen atmosphere to 200 C and kept at temperature for 30min. Ballmilling was conducted with 10 g of sieved spruce sawdust and 200 g ofceramic balls once at 400 rpm for 5h and once at 200 rpm for 3hour.Subsequently, enzymatic hydrolysis of the cellulose and hemicellulosefraction in the three pretreated biomass samples was performed accord-ing to the standard US National Renewable Energy Laboratory protocol(NREL/TP-510-42629). The three resulting lignin samples which stillcontained some amounts of carbohydrate impurities were then filteredoff and washed with deionized water. The obtained lignins were ana-lyzed for their carbohydrate, lignin, and ash content using the standardUS National Renewable Energy Laboratory protocol (NREL/TP-510-42630). Table 3.1 lists all lignins that were studied in the context of thepresent thesis.

3.1.2 Laboratory chemicals

All chemicals were used in the laboratory as received and were of atleast 98% purity unless otherwise specified. The employed chemicals in-clude the following solvents and acids: methanol (Fluka 65543), ethanol(Scharlau ET0010), propanol (Fluka 59300), butanol (Merck 101990),

3.1 Materials 29

Table 3.1: Commercial and own lignins studied within this work.

Name Origin Pulping Supplier

Indulin AT Pine Kraft MeadwestVacoLignin, alkali Spruce Kraft Sigma-AldrichUltrazine Na Spruce Sulfite BorregaardGranit Bioplast Crops Soda Granit SAOrganosolv lignin Poplar Organosolv Fraunhofer

HWP Spruce Enzymatic (hot water)BM200 Spruce Enzymatic (ball milling@200 rpm)BM400 Spruce Enzymatic (ball milling@400 rpm)

acetone (Sigma-Aldrich 34850), ethyl acetate (Sigma-Aldrich 33211),toluene (Sigma-Aldrich 179418), hexane (Fluka 52770), diethyl ether(Fluka 31700), chloroform (Sigma-Aldrich 132950), 1,4-dioxane (Fluka42500), sulfuric acid (95% to 97%, Fluka 84720), hydrochloric acid(37%, Sigma-Aldrich 258148), nitric acid (65%, Merck 100456), formicacid (Merck 100264), acetic acid (Sigma-Aldrich A6283) and phospho-molybdic acid (Fluka 79563).

The used salts and substrates are: copper chloride CuCl2 · 2H2O(Fluka 61174), iron chloride FeCl3 · 6 H2O (Fluka 44944), cobalt chlorideCoCl2 · 6 H2O (Fluka 60820), copper sulfate CuSO4 (Fluka 61230), potas-sium bromide KBr (Sigma-Aldrich 221864), syringaldehyde (Sigma-Aldrich S760), vanillin (Sigma-Aldrich V1104), methyl vanillate (Sigma-Aldrich 138126), 2-benzylphenol (Sigma-Aldrich 13761), benzyl phenylether (Sigma-Aldrich 404284), bibenzyl (Sigma-Aldrich B33706),4-phenyl phenol (97%, Sigma-Aldrich 134341) and 4-phenoxy phenol(Sigma-Aldrich 230669). As methyl dehydroabietate and 5-carbo-methoxy-vanillin were not commercially available, these two compoundswere synthesized in the laboratory according to the experimental proce-dures described in Appendix A.1.

The gases oxygen, nitrogen and helium were supplied with a purityof at least 99.995% by PanGas (Dagmersellen, Switzerland). Deionized


water was provided in the laboratories by an in-house supply. For thepreparation of HPLC eluents, chemically pure water with a specific re-sistance of 18.2MΩ cm from a Milli-Q Advantage A10 plant (Millipore,Zug, Switzerland) was exclusively used.

3.2 Experimental setup and procedure

3.2.1 Batch reactors for lignin oxidation

Three different high pressure autoclaves, which mainly varied in reactorvolume, were used to study the batch oxidation of lignin. The majority ofexperiments was performed in the batch reactor that offers a "medium"volume of 400mL. When available or required quantities demanded fora different volume, the two additional batch reactors with higher andlower volume were used. The oxidation experiments with the above-mentioned lignins from enzymatic hydrolysis were conducted in the batchreactor with the smaller volume due to the low amounts of the ligninscreated in the laboratory. For the membrane separation experimentswhich required large amounts of oxidized lignin substrate, the largerbatch autoclave was used. The characteristics of all three batch reactorsused within the thesis is depicted in Table 3.2.

Table 3.2: Batch reactors used for the lignin oxidation experiments.

Relative size Small Medium Large

Manufacturer Parr* Premex** PremexVolume [mL] 50 400 3000pmax [bar] 200 325 700Tmax [C] 350 500 300Reactor material SS316 1.4980 1.4571Liner material Glass Ti Gr2 –Sampling system No Yes Yes* Parr Instruments Company, Moline, IL, USA** Premex Reactor AG, Lengnau, Switzerland

3.2 Experimental setup and procedure 31

Standard batch oxidation experiments were performed in the 400mLhigh pressure autoclave. The reactor is equipped with a gas entrain-ment impeller to provide a good dispersion of oxygen in the liquid andthus a high gas-liquid mass transfer. All internal reactor parts are madeof titanium to avoid their corrosion upon contact with the hot acidicsolvent (compare Table 3.2). The reaction solvent was prepared bymixing 160mL methanol and 40mL ml deionized water. Afterwards,the required amount of the catalyst to give a nominal concentration of0.01mol L−1 was added and completely dissolved in the reaction solvent.The pH of the solvent was finally adjusted by adding some droplets ofconcentrated sulfuric acid to pH 1.0 as measured by a Polylite HT120sensor (Hamilton Bonaduz AG, Bonaduz, Switzerland). This mixturewas then transferred into the reactor and 2 g of lignin were added, re-sulting in a lignin concentration of 10 g L−1. As kraft lignin is not solublein acidic media, virtually all of the lignin was present as a solid/liquidsuspension before the reaction. The reactor was sealed and purged threetimes by loading and releasing oxygen at a pressure of 10bar. A workingpressure of 10bar oxygen was finally loaded to the reactor and the im-peller was started at 1000 rpm. The mixture was heated to the reactiontemperature of 170 C at 6K min−1. The time when the reactor reachedthe reaction temperature was defined as t = 0min in all batch experi-ments. After the final reaction time had been reached, the reactor wascooled down to room temperature within 60min.

The medium-sized Premex batch reactor of 400mL could optionally beequipped with an internal dip tube, which was connected to a samplingvalve at the top of the reactor. In this case, a heat exchanger exter-nally attached to that valve allowed the quenching of the hot sample.Just before sampling, a liquid volume of about 17mL (correspondingto a volume slightly higher than the dead volume of the sampling de-vice) was drawn and discarded. The volume of the liquid sample wasalways chosen to be approximately 8mL, which is sufficient for GC/MSand SEC analysis. Thus, up to seven samples could be drawn duringan experiment (e.g. every 20min during 2 h). As the analytical infor-mation is significantly increased, the opportunity was taken upon mostexperiments done in this reactor.

During the batch oxidation experiments conducted within the presentthesis, various process conditions with different parameter values were


investigated. Most of the parameters were kept at a constant standardvalue throughout all batch experiments, except if they were varied tostudy their influence on the reaction. These parameters (standard valuesin brackets) include: temperature (T = 170 C), methanol fraction inthe solvent (xMeOH = 80 vol%), acidity (pH 1), catalyst concentration(cCat = 0.01mol L−1) and, in case of model substances as substrates,the model compound concentration (cSub = 1 g L−1). Parameters thatwere specifically chosen according to the reactor or are intrinsic to it arelisted with their values in Table 3.3.

Table 3.3: Parameter values that were chosen according to the batchreactor or are intrinsic to it.

Value Small Medium Large

Solvent volume V s [mL] 10 200 1000Lignin concentration cLig [g L−1] 10 10 20Oxygen pressure pO

2[bar] 10 10 6*

Reaction time t [min] 20/120** 120 20Heating time theat [min] 14 24 ∼80Cooling time tcool [min] ∼40 ∼60 ∼120* Lower pressure to not overoxidize products during long heating/cooling** Both values studied

3.2.2 High pressure microreactor setup

The experiments to study the influence of temperature, pressure andlignin concentration on the lignin oxidation were performed in a con-tinuous two-phase flow microreactor which is part of a microscale highpressure setup. The setup allowed a fast and safe screening of experimen-tal conditions and is depicted in Figure 3.1. Pictures of the laboratorysetup can be found in Appendix A.2. Two liquid streams and oxygenare used as feedstock and mixed on a silicon/glass mixing element whichgenerates a well-distributed slug flow of gas and liquid. The two-phaseflow then immediately enters the heated capillary microreactor in which


Syrin

ge

pu

mp

Oxyg

en

Ma

ss flo

w

co

ntr

olle

r

Ca

pill

ary

re

acto

r

(he

ate

d in

oil

ba

th)

Ba

ck-p

ressu

re

reg

ula

tor

PIC

Sa

mp

le

Wa

ste

Sa

mp

le

Nitro

ge

n

T =

15

0 -

25

0 °

C

p =

32

- 9

6 b

ar

O2

t =

0.7

- 2

0.7

min

xH

2O =

20

V-%

; x

Me

OH=

80

V-%

cLig

= 0

g L

-1

cH

2S

O4=

0.2

mo

l L

-1

Mix

ing

ele

me

nt

(slu

g flo

w)

Fe

ed

Sa

mp

ling

Syrin

ge

pu

mp

PI

PI

PI

TIC

xH

2O =

20

V-%

; x

Me

OH=

80

V-%

cLig

= 5

g L

-1

cH

2S

O4=

0 m

ol L

-1

Fig

ure

3.1

:H

igh

pres

sure

setu

pfo

rth

eco

ntin

uous

ligni

nox

idat

ion

exper

imen

tsin

atw

oph

ase-

flow

mic

rore

acto

r.


the oxidation of lignin takes place. From the outlet of the microreactorthe flow is directed to the high pressure sampling system which comprisestwo sampling containers and one waste container made of stainless steel.The main parts of the system are explained in detail in the following.

Feedstock

In order to have the lignin completely dissolved in the feedstock andnot to expose it to strongly acidic conditions prior to the reactor, theliquid feed was split into two streams: one lignin feed stream containingtwice the desired concentration of lignin in a neutral solution of 80 vol%methanol and 20 vol% water, and one acidic solvent feed stream con-taining no lignin but 0.2mol L−1 of sulfuric acid in a solution of 80 vol%methanol and 20 vol% water. Both streams were fed at identical flowrates so that the same conditions as the lignin slurry in the previousbatch reactor studies were established upon mixing.

The liquid feed flow rates varied between 25µL min−1 and 200µL min−1

during the experiments and were delivered by a 260D high pressure sy-ringe pump (Teledyne Isco, Lincoln, NE, USA) and a PHD 4400 syringepump with stainless steel syringe (Harvard Apparatus, Holliston, MS,USA) for the lignin feed and the solvent feed, respectively. The oxygenflow rate was chosen to be 1 to 1.3 times the flow rate of the liquid feedsto ensure enough oxygen in the reactor. It was supplied from an exter-nal gas bottle by means of a mass flow controller (F-230M, Bronkhorst,Ruurlo, The Netherlands) at system pressure. As the lignin oxidation iscarried out in liquid phase, the minimum system pressure is dictated bythe dew point curve of the feed mixture. The software package AspenPlus R© was used to calculate the dew point curve of a mixture of 80 vol%methanol and 20 vol% water in order to find the experimentally availabletwo-phase region. It is illustrated in gray together with the calculateddew point curve and the experimentally studied data points in Figure3.2.

Silicon/glass mixing element

The gas and liquid feed streams were mixed on a silicon/glass mixingelement which was fabricated by standard photolithography, dry etching


Temperature [°C]

160 180 200 220 240

Pre

ssure

[b

ar]

0

20

40

60

80

100

Dew point curve of the feed mixture

Two-phase region

Experimental

data points

Gas region

Figure 3.2: Thermodynamically available two-phase region for thelignin oxidation experiments as calculated for a mixture of80 vol-% methanol and 20 vol-% water. The crosses displaythe investigated conditions in the experiments.

and anodic bonding techniques [61]. It allows the monitoring of the slugflow generation which was achieved by a special inlet design guarantee-ing a segmented gas/liquid flow [62]. The main channel of the mixingelement had a length of 349mm and a quadratic cross section of 300µmwidth. The external connections were realized with fused silica capillar-ies (inner diameter 180µm, outer diameter 360µm, Polymicro, Phoenix,AZ, USA) which were coplanarly connected to the inlets/outlet of themixing element and fixed with Duralco 4703 high temperature adhe-sive (Polytec PT, Waldbronn, Germany). The meandering region in theoxygen inlet increases the pressure drop and thereby stabilizes slug flowformation. Detailed information and images of the mixing element canbe found in the literature [63].

In the experiments, the solvent feed stream was first mixed with theoxygen stream in order to saturate the reaction solvent with oxygen.After a meandering saturation region, the lignin feed stream was added


and the mixture immediately exited the mixing element towards themicroreactor.

Microreactor

A Hastelloy C-276 capillary (Vici AG, Schenkon, Switzerland) with alength of 2.4m, an inner diameter of 1000µm and a wall thickness of294µm was used as microreactor. In contrast to many other materialsused for microreactors, Hastelloy C-276 is resistant to both the harshchemical conditions with highly acidic pH at elevated temperature incombination with an oxidant and the high pressures that are required toachieve these temperatures. The capillary was wound and placed in athermostated silicone oil bath which was used to establish the reactiontemperatures in the range of 150 C to 250 C. For the calculation of theresidence time, the volume of the reactor was divided by the volumetricflow rate of the two-phase flow. The latter was determined based onthe ternary mixture methanol/water/oxygen using the process simula-tion software Aspen Plus R© to account for the different thermodynamicphenomena in the multicomponent flow at reaction conditions.

High pressure sampling

After leaving the reactor, the product mixture entered the high pressuresampling part of the setup. Two independent stainless steel containers(Sitec, Maur, Switzerland) were used to collect samples. Thereby, a newexperiment could already be started using one container while samplingand cleaning the other container from the previous run. In betweentwo experiments, the flow was directed to a third stainless steel wastecontainer until the process conditions for a new run were reached andanother one and a half residence times had elapsed. This time has beenproven sufficient to establish steady state conditions in the reactor. Thesampling setup including the valves as depicted in Figure 3.1 was adaptedfrom Assmann et al. and enabled a continuous operation of the plant aswell as an efficient screening of experimental conditions without stoppingthe feed or releasing the pressure in the system [63]. The applied pressurein the range of 32bar to 96bar was established and controlled by anexternal nitrogen bottle and an automated back pressure regulator (BP-


1580-81, JASCO, Tokyo, Japan).

3.2.3 Membrane plant for separation experiments

Membrane plant

The membrane separation experiments to investigate the separation ofmonomeric products from the reaction mixture were carried out in a P-28 membrane laboratory plant (Folex AG, Seewen, Switzerland). Theplant comprises of a stainless steel feed tank with a volume of 500mLand two cross flow cells connected in series with an active membranearea of A = 28 cm2 each. The solvent is pumped through the membranecells by a gear pump and is afterwards recycled to the feed tank. Thetemperature of the feed tank was controlled to 30 C by a thermostatedwater bath. The plant was pressurized to the desired system pressurewith nitrogen from an external gas bottle.

Membranes

The five organic solvent nanofiltration membranes selected for the mem-brane separation study are commercially available. DuraMemTM 500,DuraMemTM 900, and PuraMemTM 280 are integral asymmetric mem-branes made of crosslinked polyimide (PI) and were supplied in a "dry"form by Evonik MET Ltd. (Wembley, UK, www.membrane-extraction-technology.com). PuraMemTM S380 is a thin-film composite membranemade of silicone (SIL)-coated polyimide and was also provided by EvonikMET Ltd. in a "dry" form. The molecular weight cut-off (MWCO) ofthe membranes was specified by the manufacturer to be 500Da, 900Da,280Da and 600Da, respectively. SelRo R© MPF-44 is a composite mem-brane which is composed of a separating layer of polydimethylsiloxane(PDMS) on a polyacrylonitrile (PAN) support and was obtained in a"wet" form from Koch Membrane Systems (Aachen, Germany) with anominal MWCO of 250Da [64]. The membranes employed within thisthesis and their characteristics are summarized in Table 3.4.

As the operation of the two DuraMemTM membranes is limited to20bar, this pressure was chosen for the membrane experiments. Allmembranes were obtained as sheets and membrane disks of 75mm di-ameter were cut using scissors. Before insertion into the membrane plant,


Table 3.4: Studied organic solvent nanofiltration membranes and theircharacteristics.

Membrane Supplier Material MWCO

DuraMemTM 500 Evonik PI (int. asym.) 500 DaDuraMemTM 900 Evonik PI (int. asym.) 900 DaPuraMemTM 280 Evonik PI (int. asym.) 280 DaPuraMemTM S380 Evonik SIL on PI (comp.) 600 DaSelRo R© MPF-44 Koch PDMS on PAN (comp.) 250 Da

the dry membranes were soaked for at least 2h in ethyl acetate. Thewet membrane was directly inserted into the plant.

Membrane characterization

After cutting and optionally soaking the membrane disks, two disks ofthe same membrane were placed in parallel on a sintered support plateand inserted into the cross flow cells of the plant. In order to washthe membrane disks and to address irreversible compaction phenomenawhich are known to occur with OSN membranes, pure solvent was per-meated through the membranes at a pressure of 20bar until a stableflux was obtained [65]. Flux measurements were performed for all OSNmembranes at pressures from 5bar to 20bar. During flux measurements,the permeate volume ∆V was collected in a covered volumetric flask andthe duration ∆t to reach the nominal volume of the flask was recordedby means of a stopwatch.

Rejection measurements were conducted with the same membranedisks as the flux measurements. The solutes vanillin, methyl vanillate,2-benzylphenol and benzyl phenyl ether were dissolved at a concentra-tion of 0.25 g L−1 in either 250mL or 500mL of solvent. The solutionwas filled into the plant and half of its volume was allowed to perme-ate through the two disks of the same membrane at a pressure of 20barbefore measuring rejection. Then, aliquots of the retentate and each per-meate were sampled simultaneously and analyzed concerning the solute


concentrations by GC/MS. This procedure was repeated three times forall membranes.

Fractionation of lignin oxidation products

Before the lignin oxidation products (as produced by the method de-scribed previously in this section) could be used in the membrane plant,they had to be extracted from the reaction mixture. Due to the largevolume of the latter, this was done in two batches. 500mL of the mix-ture were diluted with 900mL of water to ensure immiscibility with theorganic solvent and extracted three times by addition of 400mL ethylacetate. The organic phase was separated after each extraction step andcombined to one organic phase in the end. After partially evaporatingthe solvent in a rotary evaporator (Büchi AG, Uster, Switzerland) to avolume less than 500mL, the mixture was adjusted to a final concen-tration of 5 g L−1 of lignin oxidation products by addition of pure ethylacetate. The ethyl acetate concentration in the solvent was > 99% as cal-culated based on an Aspen Plus R© process simulation. The solution wasthen filtered using 0.3µm quartz filter paper (Fisher Scientific, Wohlen,Switzerland) and afterwards used as feedstock in the membrane plant.The average molecular weight of the product mixture in the feedstocksolution was analyzed by SEC to 1150 g mol−1.

In the beginning of each fractionation experiment two membrane disksof each membrane were run with ethyl acetate until stable flux wasachieved. Then, the solution of products in ethyl acetate was filled intothe feed tank of the membrane plant to the maximum fill level of 500mL.The plant was then run at 20bar of nitrogen pressure until 400mL ofpermeate had been obtained. Afterwards, the feed tank was refilled forthe next batch with pure ethyl acetate to the maximum fill level andmixed by running the plant for approximately thirty seconds at atmo-spheric pressure. Subsequently, a sample of the retentate was taken andanalyzed by GC/MS and SEC. Flux measurements were performed sev-eral times during plant operation. The batchwise operation procedurewas repeated eight times within an experiment with two membrane disksof the same type.


3.3 Analysis

The analytical methods that were applied in the context of this thesiscan be categorized into three groups. The first group covers methodsto identify and quantify the low molecular weight reaction products inthe reaction solvent or the gas phase. The determination of the mo-lar mass distributions of lignin and the products, which represents thesecond group, was done by size-exclusion chromatography. The thirdgroup contains methods to analyze the content of the different func-tional groups present in lignin and the products. While methods of thefirst group can detect individual reaction products from the oxidation oflignin, methods of the latter groups are used to analyze sum parametersof the feedstock mixture or the product mixture. As not all analyticaldevices were present at the Institute of Process Engineering, some of theanalysis were done externally at the Institute for Chemical and Bioengi-neering of ETH Zurich or in cooperation with the industrial partner. Anoverview of all analytical methods is illustrated in Figure 3.3.

Figure 3.3: Overview of the analytical methods used within the presentthesis.

3.3 Analysis 41

3.3.1 Identification and quantification of low

molecular weight reaction products

Aromatic products by gas chromatography/massspectrometry (GC/MS)

All samples from the oxidation experiments were analyzed concerning thenature and amount of monomeric aromatic products by gas chromatogra-phy/mass spectrometry (GC/MS). A defined volume (commonly 5mL)of each sample was diluted with half the amount of water to inhibit mis-cibility with chloroform, which was the solvent used for extraction of thereaction products from the product mixture. Extraction was performedthrice with CHCl3 (usually 3mL) and the organic phase was recovered.The three extracts were afterwards combined and a fixed amount of sy-ringaldehyde was added as internal standard (IS) for quantification. Thesamples arising from the membrane experiments, which were present inethyl acetate as solvent, did not need to be extracted. Aliquots weredirectly injected after addition of a certain amount of internal standard.

Split injections of 1µL were done by an AI 3000 autosampler (ThermoScientific, Waltham, MA, USA) into the GC/MS system (Thermo Sci-entific - Trace GC Ultra/Polaris ITQ ion trap, EI mode). The GC sys-tem was equipped with a Restek RTX-5ms capillary column (30m ×

0.25mm × 0.25µm) and helium was used as carrier gas at a flow rate of1mL min−1 with a split ratio of 25:1. The GC oven was initially kept at80 C for 5min and heated up at 10K min−1 to a temperature of 280 Cwhich was finally held for another 5min. Detection of the compoundswas done in the mass spectrometer of the GC/MS system using electronionization. Two to three injections were made and the average value wascalculated. The mean relative error between the GC/MS measurementswas less than 5%.

The monomeric products were quantified by calibration with the inter-nal standard syringaldehyde. Standard solutions with varying amountsof targeted products and a fixed amount of syringaldehyde that covera wide range of mass ratios were produced in the laboratory and usedfor calibration. When plotting the mass ratios of each monomer withrespect to syringaldehyde versus the corresponding GC area ratios, thecalibration points can be linearized according to Eq. 3.1. The linear re-


gressions showed coefficients of determination (R2) of always above 0.99for each component and were used as calibration curves.

mi

mis= ai ·

Ai

Ais+ bi (3.1)

Due to the missing availability, methyl 5-carbomethoxy-vanillate wasthe only compound which could not be calibrated in this manner. Toanyway quantify the generated amounts in the reaction, its calibrationwas estimated relative to the ones of the other vanillin derivatives. InGC, the specific peak area is inverse proportional to the slope of thecalibration curve, if it is approximated as line through the origin in theregion of low concentrations (Ai/mi ∝ 1/ai). As the specific peak areaon the other hand results from the fragmentation of the molecule inthe MS, the slope of the calibration curve of methyl 5-carbomethoxy-vanillate can be calculated by considering the chemical similarities to

aM5CV =

(

1

aMV+

1

a5CV−

1

aV

)−1

(3.2)

This calculation enabled the approximate calibration for methyl 5-carbomethoxy-vanillate and thereby to estimate its concentrations.

Gas phase analysis by gas chromatography (GC/TCD)

Gas samples which were drawn from the batch reactor after cooling wereanalyzed concerning gas phase products on a HP5890 gas chromatograph(Agilent AG, Basel, Switzerland). It was equipped with a 20-foot Supelcocolumn containing a packed bed of 100/120-mesh Hayesep D. The ovenwas initially kept at 100 C for 8min, heated to 220 C at 20K min−1

and finally kept at this temperature for 10min. Helium was used ascarrier gas and a TCD detector for detection of the eluating gas peaks.The products were quantified by external calibration.

Identification of products by liquid chromatography/massspectrometry (LC/MS)

High pressure liquid chromatography/mass spectrometry (LC/MS) wasused as an supplemental method for getting structural information to

3.3 Analysis 43

identify unknown reaction products. A high resolution LTQ-OrbitrapXL (Thermo Scientific, Waltham, MA, USA) mass spectrometer withatmospheric pressure chemical ionization (APCI-MS) allowed the deter-mination of the products’ molar masses accurate to five decimal places.Thereby the correct sum formula and the degree of unsaturation could becalculated which finally enabled to find likely structures of the unknownmolecules. As the analysis was performed in the analytical laboratoriesof the industrial partner, the analytical method is confidential and nofurther details on it are provided in this thesis.

3.3.2 Analysis of molar mass distribution

Size-exclusion chromatography (SEC)

Size-exclusion chromatography (SEC) was used to determine the molec-ular weight distribution of the lignins and the reaction samples. 2mLof each sample were adjusted to pH 12 at a solute concentration of2 g L−1 by adding sodium hydroxide solution. A solution of IndulinAT dissolved in a pH 12 sodium hydroxide solution at 2 g L−1 was pre-pared as reference. To remove possible traces of solid, all SEC sam-ples were filtered using nylon syringe filters with a pore size of 0.45µm.5µL of the samples were injected into the SEC system [Waters Alliance2695 Separations Module; column cascade MCX 8mm × 300mm (10µm)1000Å & 100 000Å + precolumn (Polymer Standard Service GmbH,Germany); photodiode array detector (UV 210–400nm)] and analyzedusing 0.01mol L−1 sodium hydroxide solution as eluent. It was preparedby using FIXANAL R© cartridges and stored under N2 sparging to preventCO2 uptake. The flow rate of the eluent was set to 0.5mL min−1 and thecolumns were operated at 35 C. A detector wavelength of 320nm wasgenerally chosen for the evaluation to avoid overlying signals of the cat-alyst in the POM experiment. Wavelengths of 254nm or 280nm couldhave been used alternatively for all other experiments as well.

Calibration of the SEC system

The reliable and accurate calibration of SEC systems is a challengingtask since no calibration standards are available that resemble ligninsufficiently. Moreover, different SEC systems that vary in column type,


eluent, and calibration method are used in different lignin-analyzing lab-oratories. For this reason, analytical results on the average molecularweight for the same lignin can vary by several thousand grams per molein the literature [66]. Although the scientific community is aware of thisissue and collects data from different laboratories working on lignin anal-ysis in a round robin test, satisfying solutions are still lacking. Therefore,an own tailor-made solution was found within this project.

For the region between 697 g mol−1 and 148 500g mol−1, calibrationof the molecular weight was done with common poly(styrene sulfonate)(PSS) sodium salt standards using a third-order polynomial. In order tocalibrate the region below 697 g mol−1, the four peaks with the longestelution times of a nanofiltration permeate sample were assigned to besignals from products having one to four aromatic rings. This is espe-cially true for vanillin which eluates at the elution time of the last peakin the chromatogram. A molar mass of 160 g mol−1 (close to vanillin)was assumed per aromatic ring in the products, resulting in 160 g mol−1

to 640 g mol−1 for the four peaks. The molecular weight for these peaksversus elution time was also fitted using a third-order polynomial, whichleads to a very good transition between the two calibration ranges. Fig-ure 3.4 depicts the calibration with the two third-order polynomials.

To validate this calibration, the calculated average molecular weightsof the own Indulin AT samples were compared to results from absoluteanalytical methods in the literature. Measured values in this work of3300±200g mol−1 match literature values of 2990 g mol−1 to 3400 g mol−1

very well and prove the strength of the chosen calibration method [30].The weight average molecular weight Mw, the number average molec-

ular weight Mn, and the polydispersity index PDI were calculated basedon the SEC chromatograms with the absorbance units AUi according toEqs. (3.3) - (3.5) from the literature [67].

Mw =

∑

iAUi ·Mi∑

iAUi(3.3)

Mn =

∑

iAUi∑

iAUi/Mi(3.4)

PDI =Mw

Mn

(3.5)

3.3 Analysis 45

Elution time [min]

26 28 30 32 34 36 38

log M

w /

[lo

g (

gm

ol-1

)]

1

2

3

4

5

6

7PSS standards

Low Mw "standards"

Combined 3rd

order

polynom (calibration)

PSS standards

Elution time [min]

22 24 26 28 30 32 34

AU

[ -

]

0.00

0.01

0.02

0.03

0.04

0.05

Low Mw "standards"

Elution time [min]

24 26 28 30 32 34 36

AU

[ -

]

0.00

0.02

0.04

0.06

0.08

0.10

0.12

0.14

0.16n=1n=2n=3n=4

Mw = n*160 g/mol

Figure 3.4: Calibration of the SEC system with PSS standards andown ultrafiltrated samples.

3.3.3 Functional group analysis

Phenolic hydroxyl groups by UV/Vis spectroscopy

Quantitative and qualitative analysis of phenolic hydroxyl groups inlignin and its reaction products was done according to a modified UV-method published in the literature [68]. The method is state of the artand widely employed in lignin analysis. It is based on the difference inabsorbance of phenolic hydroxyl groups between the protonated stateat pH 6 and the deprotonated state at pH 13.37 (0.2mol L−1 sodiumhydroxide solution). In addition to the quantitative data, informationon the chemical structure surrounding the phenolic hydroxyl can be ob-tained by evaluation of different wavelengths.

The UV/Vis spectrum of the samples was recorded between 200nmand 400nm on a Lambda 35 device (PerkinElmer AG, Schwerzenbach,Switzerland). The absorbance values of the two apparent maxima atapproximately 300nm and 350nm were used for calculation. Validationof the method was done by comparing Indulin AT results with litera-


ture results. The measured value of 4.21wt% phenolic hydroxyl contentmatches literature data of 4.0wt% to 4.5wt% very well [59, 69].

Aliphatic hydroxyl groups by 31P-NMR spectroscopy

Phosphorous nuclear magnetic resonance (31P-NMR) spectrograms wererecorded on a Bruker DRX-500 NMR spectrometer (Bruker, Billerica,MA, USA) to quantify aliphatic hydroxyl groups and carboxyl groupsin lignin and in the reaction products. For this reason, the hydroxylfunctionalities were derivatized with the phosphitylating agent 2-chloro-4,4,5,5-tetramethyl-1,3,2-dioxaphospholane in order to generate phos-phite groups which are suitable for 31P-NMR. The whole experimentwas performed in a mixture of pyridine and deuterated chloroform (1.6:1,v/v). Cyclohexanol and chromium acetylacetonate were used as internalstandard and relaxation agent, respectively. The 31P-NMR measurementwas performed with a 25 s delay between 90 pulses and an inverse gateddecoupling sequence was used in order to obtain quantitative spectra.The analytical method is described in detail in the literature [70].

Carboxyl groups by 31P-NMR spectroscopy and aqueoustitration

The amount of carboxyl groups was analyzed by 31P-NMR spectroscopyand by aqueous titration. The NMR spectrograms were recorded as de-scribed above and allowed the quantification of the free carboxyl groupsthat had reacted with the phosphitylating agent. For the aqueous titra-tion, which was performed on a 862 Compact Titrosampler (Metrohm,Herisau, Switzerland), 250mg of the sample were dissolved in a sodiumhydroxide solution at pH 12 and titrated with a 0.1N hydrochloric acidsolution. As lignin derivatives tend to precipitate at low pH, it is neces-sary to titrate very slowly (flow rate of 4mL min−1). Detailed informa-tion on this analytical method were published by Zakis [71]. Owing tothe saponification reaction at alkaline pH, the titration quantifies bothfree and esterified carboxyl groups. Thus, the numerical difference ofboth methods allows the quantification of esterified carboxyl groups.

3.3 Analysis 47

Carbonyl groups by Fourier-transform infrared (FTIR)spectroscopy

FTIR spectra of selected samples were recorded on a Spectrum BX in-strument (PerkinElmer AG, Schwerzenbach, Switzerland) to quantifythe carbonyl content of the sample and to analyze it concerning the rel-ative abundance of functional groups. For each spectrum, 64 scans inthe range of 600 cm−1 to 4000 cm−1 (resolution of 2 cm−1) were aver-aged. Band assignment and sample preparation was done according tothe literature [72]. The peaks were normalized to the aromatic skeletalvibrations at a wavelength of 1514 cm−1, assuming no significant changein the number of aromatic rings during depolymerization. Quantificationof the carbonyl content was done as in Faix et al. [73].

Methoxyl groups by 13C-NMR spectroscopy

Quantification of methoxyl groups in lignin and the product sampleswas done by 13C nuclear magnetic resonance spectroscopy in combina-tion with either 31P-NMR or UV/Vis analysis. For distinguishing differ-ent types of hydroxyl groups, the lignins and the product samples wereacetylated with acetic anhydride and pyridine. As internal standard, awell-known alcoholic function (either from 31P-NMR or UV/Vis) waschosen and an inverse gated decoupling mode was used for quantitativeanalysis of the methoxyl groups. The spectrograms were as well recordedon a Bruker DRX-500 NMR spectrometer (Bruker, Billerica, MA, USA).More details on the analytical methods can be found in the literature[67, 70].

Reactive 5-positions on the aromatic ring by methylolation

The reactive positions on the aromatic ring, which are assumed to equalthe free position on the ring, were quantified by methylolation based onthe method of Vázquez et al. [74]. 18 g of lignin were dissolved in 60mLof water with 2.34 g sodium hydroxide at 40 C and afterwards methy-lolated with 16 g of formaldehyde for 20h to 24h using a reflux con-denser. For quantification, the amount of remaining formaldehyde wasthen titrated according to the hydroxylamine hydrochloride method. 5 gof the solution in 100mL of water were adjusted to pH 3.85 by addition of


0.1N HCl, and mixed with 25mL of a 2N hydroxylamine-hydrochloridesolution dissolved in 50mL of water at the same pH. After 10min, themixture was titrated with 1N NaOH to pH 3.85. The amount of reactivepositions equals the amount of used formaldehyde.

49

Chapter 4

Catalytic oxidation of lignin in

acidic media

In this chapter, the detailed results on the depolymerization of ligninby catalytic oxidation in acidic aqueous solvent are presented. The gen-eral reaction concept is based on a previous work of Voitl who studied anovel approach for the oxidation of lignin into aromatic chemicals [10].In contrast to most other research approaches and to the industrial pro-cess that was operating in the 1980’s, the oxidative depolymerizationof lignin is carried out in acidic solvent. This was deemed unfeasibleby other scientists as it is known that counterproductive condensationreactions occur in acidic solvent. However, it could be demonstratedthat simple alcohols can prevent the condensation of lignin fragmentsby competitive coupling and even higher yields of aromatic products areobtained [75]. This is in accordance with literature studies showing thatacidic conditions are generally superior for depolymerization of lignin,since cleavage of interunit bonds is not limited to phenolic units [76].

In the above-mentioned work, the oxidation of lignin was carried outin an acidic mixture of 80% methanol and 20% water at 170 C withan oxygen pressure of 10bar. Polyoxometalates (POMs) which are well-known for their potential in lignin degradation during paper bleachingwere used as homogeneous redox catalysts. Possessing a redox poten-tial in between that of lignin and oxygen, POMs can selectively oxidizelignin by donating several electrons and subsequently be reoxidized inthe presence of molecular oxygen [77]. Moreover, POMs have been showncapable of cleaving kraft lignin model compounds of the diarylmethanetype, making them promising redox catalysts especially for kraft ligninvalorization [78]. The main monomeric reaction products from the acidic

50 4. Catalytic oxidation of lignin in acidic media

oxidation of kraft lignin in the presence of a POM catalyst are vanillinand methyl vanillate. In addition to the monomeric products, significantamounts of di-, tri- and oligomeric reaction products were found. Basedon this reaction concept, the acidic oxidation of lignin and the influenceof the main reaction parameters are studied in detail in the following.

4.1 Investigation of different transitionmetal salt catalysts

A crucial step for the successful performance of any reaction, even be-fore studying different reaction conditions, is to find an efficient catalyst.Although POMs have already been found to be an appropriate catalyst,the aim was to find a cheap catalyst that performs as least as good inthe depolymerization of kraft lignin towards monomeric products as thebenchmark phosphomolybdic acid (H3PMo12O40), which had been stud-ied as POM catalyst before [75]. The catalyst screening was restrictedto homogeneous catalysts, as with a heterogeneous catalyst both ligninand the catalyst were dispersed in the solvent, thereby worsening thechance of contact and the performance of the catalyst.

Different catalysts were studied to degrade lignin into aromatic prod-ucts in the literature. Wu et al. obtained up to 14.6% of aromaticaldehydes (4.7% vanillin) from steam-explosion hardwood lignin withCuSO4 and FeCl3 catalysts in a 13.5% NaOH solution [79]. In the workof Villar et al., copper and cobalt salts served as homogeneous catalystin NaOH solution [9]. By oxidation of a hardwood organosolv lignin inacetic acid/water with a Co/Mn/Zr/Br catalyst, Partheimer found upto 10.9% of aromatic aldehydes and aromatic acids [80].

In the present thesis, the performance of different soluble transitionmetal salts (CuSO4, FeCl3, CuCl2, CoCl2) as catalysts for the acidic oxi-dation of kraft lignin at 170 C with 10bar of oxygen pressure was inves-tigated. Since the action of the catalyst is assumed to be based on its re-dox potential as reported for delignification studies in the literature, thetransition metals were chosen to have a broad spectrum of cation redoxpotential (Co/Co2+=−0.28V, Cu/Cu2+=0.15V, Fe/Fe3+=0.77V) [81].All of these values are below the redox potential of oxygen (O2/H2O=1.21V), which is essential for the reoxidation of the catalyst. The results

4.1 Investigation of different transition metal salt catalysts 51

of the catalyst experiments are compared to the reaction carried out inpure solvent acidified by sulfuric acid as well as to results of previouslyperformed experiments with phosphomolybdic acid.

4.1.1 Main monomeric reaction products

Before studying the acidic oxidation of lignin in the presence of tran-sition metal catalysts, the main monomeric reaction products as wellas their formation and decomposition pathways were analyzed. Themost important and most familiar aromatic products from the oxidationof lignin with POM catalyst in H2O/MeOH solvent were vanillin andmethyl vanillate [75]. Unlike vanillin which is generally obtained in theoxidation of lignin, methyl vanillate is formed by acid-catalyzed ester-ification of vanillic acid in the presence of methanol and is not foundin reactions carried out in pure water. The experiments with transi-tion metal salt catalysts, however, revealed several additional productsin considerable amounts. This is illustrated in the exemplary GC/MSchromatogram in Figure 4.1 with retention times given for each peak.

Besides the known peaks of vanillin, methyl vanillate, and the in-ternal standard (syringaldehyde) at retention times (RTs) of 12.66min,14.23min, and 15.94min, respectively, five additional peaks are visiblein considerable amounts. By comparison of its MS spectra with theNIST database, the compound with a retention time of 22.75min wasidentified as methyl dehydroabietate (MDHA). The compounds withRTs of 17.01min, 18.32min, 18.51min, 22.09min and 23.20min couldnot be identified by their GC/MS spectra. The structures of thesecompounds were, however, revealed by means of high resolution MSdata from LC/MS analysis. In combination with various experimentswith model compounds as substrate, it turned out that all of them arederivatives of vanillin and methyl vanillate and formed by their oxida-tion at longer reaction times. The products at RTs of 17.01min and18.32min/18.51min are 5-carbomethoxy-vanillin (5CV) and methyl 5-carbomethoxy-vanillate (M5CV) whereas the latter is present in twoisomeric forms with two different retention times. The two remainingcompounds at RTs of 22.09min and 23.20min are not common in lit-erature. Their chemical structures were postulated based on LC/MSanalysis, but isomeric structures thereof are also possible.


Retention time [min]

10 12 14 16 18 20 22 24

Rel

ativ

e ab

unda

nce

[%]

0

20

40

60

80

100

12.66

14.23

15.9417.01

18.32

18.51

22.75

22.0923.20

Figure 4.1: GC/MS chromatogram of the products from the acidic ox-idation of kraft lignin after 2 h of reaction with CuCl2 cat-alyst (T = 170 C, pO2

= 10bar, cLig = 10 g L−1). Thepeaks, with numbers indicating GC retention time, are as-signed in Table 4.1. The peak at 15.94min is the internalstandard.

Based on these results, a reaction network with chemical structuresand formation pathways of the main products from the acidic oxidationof softwood kraft lignin in the presence of transition metal catalysts wascreated and is summarized in Figure 4.2. The formation of vanillin andmethyl vanillate (reactions A and B) from lignin in acidic media viadifferent intermediate steps is well-understood and described by Giererand Nilvebrant [76] and Voitl et al. [82]. For longer reaction times,these products are derivatized by carbomethoxylation as shown in re-actions D and F. The reactions most probably proceed via the carboxyderivatives as 5-carboxy-vanillin was also reported to be a product fromthe alkaline oxidation of lignosulfonates at similar reaction conditions by


OH

O

O

O

O

OH

O

O

O

O

O

O

OH

O

OO

O

O

O

O

OH

O

O

OH

O

OO

OH

O

OO

O

O

Methyl dehydroabietateC H O

314 g mol21 30 2

-1

VanillinC H O

152 g mol8 8 3

-1

Methyl vanillateC H O

182 g mol9 10 4

-1

O O

Methyl

C H O

240 g mol

vanillate

11 12 6

-1

5-carbomethoxy-C H O

322 g mol15 14 8

-1

C H O

292 g mol14 12 7

-1

5-Carbomethoxy-vanillin C H O

210 g mol10 10 5

-1

Carbon dioxide

Ring-opening products

Further derivatization

A

H

G

C

B

FD E

H

OH O

OH O

-H+

+CH OH3

Gierer and Nilvebrant [76]Voitl et al. [82]

OH

O

O

O

OH

OH

O

OO

O

OH

+CH OH3+CH OH3

Kraft lignin from softwood

Mixture of dimers / trimers Mixture of oligomersLignin biopolymer

SH

O

OH

O

O

OH

O

OH

OHO

OH

O

O

OH O

O O

O

OH

O

OH

OH OH

OH

O

OH

OH

O

O

OH

O

OH

O

O

O

OHO

O

O

OH

O

O

O

OH

O

OH

O

OH

O

O

O

O

OH

O

O

O

OH

O

Exem

pla

ry s

tru

ctu

res b

ased o

n

analy

tical re

sults a

nd lite

ratu

re [3

2]

Main

mo

nom

eric r

eaction

pro

ducts

as iden

tified b

y G

C/M

S a

nd L

C/M

S

Figure 4.2: Reaction network of the main products from the acidic ox-idation of softwood kraft lignin with transition metal saltcatalysts.


Pearl [83], followed by acidic esterification with methanol as describedby Profft and Smirnow [84]. The products from reactions E and G,which were postulated based on LC/MS analysis, are obtained in loweramounts. Independent from the previous components, methyl dehydroa-bietate (reaction C) turned out to be another product from kraft ligninoxidation. It is formed from abietic acid by dehydrogenation and esterifi-cation with methanol as described by Hjulström et al. [85]. Abietic acidis a major constituent of pine resin and bound to the lignin molecule viaester bonds [86]. For even longer reaction times, the concentrations of allcomponents start to decrease. Degradation products including mainlycarbon dioxide but also ring-opening products like maleic acid dimethylester are formed.

4.1.2 Performance of the catalysts

In accordance to the general goals of the oxidative lignin depolymeriza-tion, the performance of the catalyst was evaluated based on two criteria.On the one hand, it should promote the formation of monomeric prod-ucts and result in high yields of favorably single components. On theother hand, it is supposed to cause a rapid breakup of the interunitlinkages in lignin and thus a fast depolymerization. These two criteriaas well as the ability of the catalyst to incorporate the oxygen into theproducts during the oxidation are discussed in the following.

Concentration of low molecular weight products

To evaluate the formation of monomeric products in the presence of dif-ferent catalysts, the concentration of the two main reaction products (viz.vanillin and methyl vanillate) was first monitored. Figure 4.3 depictstheir concentrations versus reaction time up to 2h whereas t = 0minis the time when the reactor had reached the reaction temperature of170 C. The accuracy of the measured concentrations was evaluatedbased on three experiments conducted in the absence of a catalyst. Inthese experiments, the maximum deviation from the average was alwaysless than 7.1% for vanillin and less than 8.6% for methyl vanillate. Toevaluate the progress of the reaction after 2 h, experiments without cat-alyst and with POM were conducted for 6 h. While the vanillin and


0 20 40 60 80 100 120 140

cV [

g L

-1]

0.00

0.05

0.10

0.15

0.20

0.25

0.30

Reaction time [min]

0 20 40 60 80 100 120 140

cM

V [

g L

-1]

0.00

0.05

0.10

0.15

0.20

0.25

0.30

H2SO

4

H3PMo

12O

40

CuSO4

FeCl3

CuCl2

CoCl2

H2SO

4

H3PMo

12O

40

CuSO4

FeCl3

CuCl2

CoCl2

Figure 4.3: Concentrations of vanillin and methyl vanillate during theacidic oxidation with different transition metal salt cat-alysts at 170 C, 10bar of oxygen pressure and 10 g L−1

lignin concentration.


methyl vanillate concentrations remained approximately constant in theexperiment with sulfuric acid, they began to decrease after 2h in thePOM experiment.

According to the figure, the transition metal salt catalysts only slightlyenhance the formation of the monomeric products vanillin and methylvanillate. For all of the studied salts, the concentration maximum ofvanillin is higher than without catalyst (sulfuric acid) or in presence ofH3PMo12O40. In contrast, the concentration of methyl vanillate reachedonly in experiments with CoCl2 and CuSO4 the same level as with thestudied POM but the yield was with any catalyst at least as high aswithout catalyst. The chloride salts FeCl3 and CuCl2, on the other hand,show a fast kinetics in the formation of the two products. The highestvanillin yield at time t = 0min and a concentration maximum after20min reaction time was obtained with the latter mentioned catalysts,while with the other catalysts a maximum is not observed before 60min.The maximum yields of vanillin and methyl vanillate, defined as productconcentration related to the concentration of pure lignin in the feedstock,are summarized in Table 4.1.

The formation of the other monomeric products versus reaction timeis always roughly linear and the maximum amount was therefore usu-ally found at the end of the reaction time. Although the final amountof MDHA does not vary significantly between the experiments, mostMDHA is obtained in the absence of any catalyst. This is not the casefor the four derivatives of vanillin and methyl vanillate. While theirformation without catalyst is rather negligible, the yields of these sub-stances increased significantly in presence of FeCl3 and CuCl2. As puresubstances of these products were not available for GC calibration dur-ing this study, the relative amount of each product i with respect to theinternal standard peak area (Ai/Ais) at a reaction time t = 120min islisted in Table 4.1.

Weight average molecular weight of the product mixture

The performance of the catalysts in breaking interunit linkages in ligninwas evaluated by the molecular weight distribution of the product mix-ture. It was monitored during the reaction by SEC as depicted for theiron chloride experiment in Figure 4.4. From the graph it becomes ap-


Table

4.1

:M

axim

umyi

elds

ofva

nilli

n(V

)an

dm

ethy

lva

nilla

te(M

V)

asw

ell

asre

lative

amou

ntof

othe

rpr

oduc

tsob

tain

edin

the

oxid

atio

nex

peri

men

tsw

ith

the

diffe

rent

inve

stig

ated

tran

sition

met

alsa

ltca

taly

sts

(T=

170C

,pO

2=

10ba

r,c L

ig=

10g

L−1).

Com

poun

dV

MV

5CV

M5C

Vn.

n.M

DH

An.

n.RT

inG

C/M

S[m

in]

12.6

614

.23

17.0

118

.32

18.5

122

.09

22.7

523

.20

Cat

alys

tM

axim

umyi

eld

[wt%

]R

elat

ive

amou

ntin

GC

/MS

att=

120m

in[-]

none

(H2SO

4)

2.41

2.33

0.05

0.05

0.12

0.00

1.36

0.01

H3P

Mo 1

2O

40

2.60

3.15

0.08

0.20

0.28

0.02

1.10

0.06

CuS

O4

3.12

3.15

0.30

0.54

0.52

0.14

0.95

0.18

CoC

l 23.

193.

090.

370.

660.

700.

171.

020.

23Fe

Cl 3

2.93

2.61

0.82

1.37

0.81

0.30

1.02

0.37

CuC

l 22.

932.

331.

031.

570.

850.

551.

070.

55


parent how the whole molecular weight distribution is shifted to lowermasses on the right side with increasing reaction time. The amountsof high molecular weight products decreases in favor of low molecularweight products, especially monomers and dimers which are the twopeaks furthest to the right.

Elution time [min]

28 30 32 34 36

No

rma

lize

d A

U [

-]

0.0

0.2

0.4

0.6

0.8

1.0

1.2

Molar mass [g mol-1

]

39439 14892 3215 934 446 292 219 156 89 34

Indulin AT

t = 0 min

t = 20 min

t = 120 min

Figure 4.4: SEC chromatograms of the product mixtures during thedepolymerization of kraft lignin with iron chloride catalystat 170 C, 10bar of oxygen pressure and 10 g L−1 ligninconcentration.

Based on the SEC chromatograms, the weight average molecularweight can be calculated for each sample as described in Section 3.3. Us-ing these values as calculated for the samples drawn during the catalystexperiments, the depolymerization of kraft lignin can be visualized ver-sus reaction time. Figure 4.5 illustrates the decrease in average molecularweight (starting from 3200g mol−1 of Indulin AT) for the experimentswith the different catalysts.

Obviously, when working without catalyst (only H2SO4), the depoly-


Reaction time [min]

0 20 40 60 80 100 120

Mw

[g m

ol-1

]

0

500

1000

1500

2000

2500

3000

3500Indulin AT ~3200 g mol

-1 H2SO

4

H3PMo

12O

40

Cu2SO

4

FeCl3

CuCl2

CoCl2

Figure 4.5: Decrease in weight average molecular weight of the productmixture versus reaction time for different catalysts (T =170 C, pO2

= 10bar, cLig = 10 g L−1).

merization proceeds very slowly. Even with POM as catalyst, the de-crease in the molecular weight is only slightly enhanced. In contrast, afar better fragmentation is achieved with the studied transition metalsalts by which the molecular weight of the product mixture is decreasedto considerably lower values. This is even emphasized by the fact, thatthe remaining reaction products were completely dissolved in the reac-tion solvent after the reaction time of 2h, except for experiments withsulfuric acid and POM, where small amounts of finely dispersed solidswere present.

The best results were obtained with FeCl3 and CuCl2 which frag-mented the lignin molecules into a product mixture with an average mo-lar mass of almost 500 g mol−1 within two hours of reaction time. Thisis in accordance to the concentration of monomeric products describedabove, where those two catalysts already showed the fastest kinetics andthe highest amounts of monomeric products.


Oxygen consumption

For evaluating the action of the different catalysts concerning the oxida-tion, the consumption of oxygen was analyzed by comparing both FTIRspectra of the reaction products and the loss of pressure in the reactorduring the reaction. FTIR spectra of the products after 2h of reactiontime from experiments without catalyst and with CuCl2 as one of thetwo most promising catalysts were recorded after extraction with CHCl3and drying of the extracts. In Figure 4.6, the spectra are compared toIndulin AT. The labeled peaks show the region of IR absorbance of: (A)aromatic skeletal vibrations (1514 cm−1), (B) aromatic skeletal vibra-tions plus C=O stretch (1593 cm−1), (C) C=O stretch of conjungatedaldehydes and carboxylic acids (1680 cm−1), (D) C=O stretch in uncon-jugated ketones, carbonyls and ester groups (1732 cm−1) and (E) O-Hstretch (3412–3460 cm−1).

The most important difference between the spectra is in the rangeof 1600 cm−1 to 1800 cm−1 (region of C=O bonds). After treatmentwith sulfuric acid, the band areas in this region increased, indicating ahigher amount of C=O compared to Indulin AT. The analysis of prod-ucts from experiments with CuCl2 show further increased band areasand therefore an even higher oxygen consumption. An estimation ofthe C=O content of the samples according to [73] resulted in 3.6wt%(1.27mmol g−1) for Indulin, 5.8wt% (2.06mmol g−1) for sulfuric acid,and 7.2wt% (2.57mmol g−1) for CuCl2. This increase in C=O con-tent can mainly be attributed to the formation of carbonyl and car-bomethoxyl groups in the products as found for vanillin and methylvanillate before. As no significant change in the amount of hydroxylgroups (3412 cm−1 to 3460 cm−1) is observed, their formation seems notto arise from direct oxidation of aliphatic hydroxyl groups. More likely,C=O formation is due to the cleavage of linkages in acidic media in thepresence of oxygen as described by Gierer and Nilvebrant [76].

A comparison of reactor pressures after cooling at the end of the re-action reveals similar results. Table 4.2 lists the pressures at the endof the experiment for the different catalysts. Although these values donot correspond to the absolute amount of oxygen present after the reac-tion (liquid samples are withdrawn during the reaction changing the gasvolume; CO2 and dimethyl ether are formed), a clear trend is observed.


Wavenumber [cm-1

]

9001100130015001700190032003400

No

rma

lize

d A

U [

-]

0.0

0.5

1.0

1.5

2.0

ABC

D

E

Indulin AT

No catalyst

CuCl2

Figure 4.6: FTIR spectra of samples from experiments with CuCl2 cat-alyst and without catalyst after 2h of reaction time, incomparison to Indulin AT. Peak assignment is explained inthe text above. The spectra are normalized to 1514 cm−1.

Only small amounts of O2 are consumed in the experiments without cat-alyst and with POM. In contrast, the far lower pressures at the end ofthe transition metal salt experiments indicate considerably higher oxy-gen consumption. Moreover, the remaining pressure at the end of thereaction correlates with the molecular weight at a reaction time of 120min (compare Table 4.2). The experiment without catalyst consumedthe least amount of oxygen and the molecular weight of the productswas the highest and vice versa for the experiments with CuCl2 or FeCl3.A linear regression of remaining pressure and molecular weight resultsin a good correlation (R2 = 0.95). By calculating the initial molecular


weight of Indulin AT based on this correlation (extrapolation to 10barof initial oxygen pressure) a value of 3946.5g mol−1 is obtained.

Table 4.2: Final pressures in the reactor after 2h of reaction timeduring the experiments with the different transition metalsalt catalysts and average molecular weight of the reactionproducts.

Catalyst Pressure [bar] Mw at t = 120min [g mol−1]

none (H2SO4) 5.6 1636H3PMo12O40 5.4 1250CuSO4 4.2 775CoCl2 4.4 753FeCl3 3.7 529CuCl2 3.8 531

Although the final pressures in the reactor were affected by the sam-pling during the experiments, they allow for an approximate estimationof the oxygen consumption during the reaction. When calculating witha Henry’s law constant of HO

2,Solv = 5000bar as interpolated from the

literature and a 60% oxygen fraction in the final gas phase as determinedby GC/TCD in subsequent batch experiments in the Parr reactor (seeSection 4.3), the oxygen consumption can be estimated to 0.93 g/gLig

for the CuCl2 experiment and 0.76 g/gLig for the reference H2SO4 ex-periment [87]. While these values are valid for a reaction time of twohours in the batch reactor, the same calculation for a 20min experimentwith copper chloride as catalyst returns a consumed oxygen amount of0.72 g/gLig.

Summarizing evaluation of the different catalysts

In general, the selectivity of the studied transition metal salt catalyststowards the monomeric products vanillin and methyl vanillate is onlyslightly enhanced. The maximum yield of vanillin obtained in the exper-iments is always higher for the transition metal salts than for the POM or


without a catalyst. However, the oxidation of vanillin and methyl vanil-late leads to decreasing concentrations with proceeding reaction time.

Among the investigated catalysts, CoCl2 yielded the highest amountof vanillin and methyl vanillate with a maximum of 6.3wt% after 60minof reaction time. It is, however, less efficient in the depolymerization ofthe lignin molecule compared to FeCl3 or CuCl2. This might be due tothe lower redox potential of cobalt. On the other hand, since iron andcopper lead to comparable results concerning product yield and averagemolecular weight of the product mixture, the fragmentation seems not tobe limited by the redox potential of the catalyst when working above acertain value. But the fact that copper chloride and copper sulfate showdifferent results proves that the catalytic action can rather be ascribedto both cation and anion than to the cation alone. This will be addressedin more detail in the next subsection.

The use of CuCl2 or FeCl3 as catalyst resulted in the fastest reac-tion kinetics. Concentrations of vanillin and methyl vanillate indicate amaximum at the very beginning of the reaction while the decreasing con-centrations show their quick reaction in favor of their oxidation products.The fast kinetics is also pointed out by the rapid decrease of the meanmolecular weight down to 500 g mol−1 and the enhanced oxygen con-sumption as shown by FTIR. Moreover, the use of FeCl3 and especiallyCuCl2 as catalyst, resulted in the highest amount of monomeric productsapart from vanillin and methyl vanillate. Based on these findings, thosetwo can be considered the best of the investigated homogeneous oxida-tion catalysts. CuCl2 was therefore chosen as catalyst for all followingexperiments within this thesis.

4.1.3 The catalyst’s mode of action

Although the transition metal salts have proven to be effective in acidicoxidation of lignin, their mode of action has not completely been clarifiedyet. As already stated above, a redox mechanism according to Eqs. (4.1)and (4.2) was assumed as the role of the catalyst during the oxidation oflignin as it was proposed in POM lignification studies in the literaturebefore [81]. The oxidized lignin from Eq. (4.1) can subsequently returnto neutral state either by liberation of a proton, by acquisition of ananion present in solution or by undergoing a further redox reaction.


Lignin + Cu2+ −−→ Cu+ + Lignin+ (4.1)

Cu+ + 14 O2 +H+ −−→ Cu2+ + 1

2 H2O (4.2)

The results from the transition metal salt catalysis experiments re-vealed that the redox potential of the cation seems to contribute to thecatalytic action. However, it could not explain differences for differentsalts of the same metal. In order to shed more light into the effect of thetransition metal salts in the fragmentation of the lignin molecule, reac-tions with different model compounds were performed with and withoutcatalyst. As the fragmentation is linked to the breakage of interunitbonds in lignin, five dimeric model compounds with different types oftypical kraft lignin linkages were studied. Compounds that resemblethe aromatic rings in lignin with respect to their functionality are notcommercially available though. Therefore, dimeric compounds of thebenzene and phenol type were selected as depicted in Figure 4.7. In ad-dition to these compounds, vanillin and methyl vanillate were also usedin these studies.

OH

O

O

OH

OH

Bibenzyl 2-Benzylphenol Benzyl phenyl ether

4-Phenyl phenol4-Phenoxy phenol

Figure 4.7: Selected dimeric model compounds to study the action ofthe transition metal catalysts.

In the presence of the CuCl2 catalyst, vanillin and methyl vanillatewere rapidly converted into their derivatives as described in Section 4.1.Quite surprisingly, all dimeric model compounds except bibenzyl were


also converted rapidly in the reactor as shown in Figure 4.8. If the con-centration decays were caused by the breakage of the interunit bonds,the catalyst would prove able to break the recalcitrant carbon bonds.As usually only ether bonds are known to be easily cleaved in acidic andalkaline oxidation, this could explain the better performance in frag-mentation with the catalyst. However, products analysis turned outthat indeed the ether bond is the only bond that is cleaved. All otherbonds remained intact as the compounds were only derivatized with car-bomethoxyl groups in the same way as vanillin and methyl vanillate.

Reaction time [min]

-20 0 20 40 60 80 100 120 140

Re

lative

am

ount

cj/c

j0 [-

]

0.0

0.2

0.4

0.6

0.8

1.02-Benzyl phenol

Benzyl phenyl ether

4-Phenoxy phenol

4-Phenyl phenol

2-Benzyl phenol (no cat.)

Figure 4.8: Conversion of the model compounds during the acidic oxi-dation at 170 C and 10bar of oxygen pressure with copperchloride catalyst. The slower conversion in the absence ofthe catalyst is illustrated by the additional plot of 2-benzylphenol.

Thus, the only linkages that are disrupted during both experimentswith and without catalyst are ether linkages. Bibenzyl is not affected at


all due to the absence of a phenolic OH group which turns out crucialfor the derivatization. All phenolic dimers on the other hand are subjectto carbomethoxylation. Here, however, the kinetics heavily depend onthe presence of the catalyst which is illustrated in Figure 4.8 by 2-benzylphenol. While it is virtually completely converted with copper chlorideafter 30min of reaction time, incomplete conversion is still observed after120min in the absence of a catalyst.

The reason why this derivatization is significantly faster with catalystcan be elucidated by analysis of products from the reaction of vanillin.5-chlorovanillin is detected in small amounts from the very beginningof the reaction as derivative of vanillin. The reaction is also known inthe literature to occur under acidic conditions in the presence of chlo-ride ions [88]. Moreover, it appears to be an intermediate product inthe formation of 5-carbomethoxy-vanillin based on the concentrations ofthese products during the model reactions. The reaction of vanillin to5-carbomethoxy-vanillin can thus be assumed to take place as depictedin Figure 4.9 and exclusively occurs in the presence of the catalyst. Theorigin of the carboxyl group that substitutes chloride at the aromaticring is, however, not fully understood. Although formic acid that origi-nates from the methanol fraction in the solvent via formaldehyde wouldrepresent a potential candidate, neither favorable reaction conditions forthis oxidation are present nor was any formic acid detected in the solventby HPLC. As methanol is the only carbon source besides the substrate,the fact that it originates in methanol is very likely.

O

OH

O

O

OH

O Cl

O

OH

O

O

OH

CuCl2

R

OH

O

+

-RCl

O

OH

O

O

O

+CH3OH

-H2O

Figure 4.9: Supposed reaction of vanillin to 5-carbomethoxy-vanillinvia chlorovanillin intermediate.

Based on the model reactions, the catalytic action of copper chlo-ride can finally also in part be explained by the promotion of the car-bomethoxylation reaction by the chloride anion. These chloride inter-mediates were also found in experiments with the other phenolic model


substances, showing that the reaction is not limited to vanillin. Thus,the action of the chloride ions in combination with the high redox po-tential of copper and iron explains why these two chlorides yielded thebest results among the different transition metal salt catalysts.

4.1.4 Variation of catalyst concentration

The influence of the copper chloride concentration on the lignin oxi-dation was studied in experiments with catalyst concentrations from0.005mol L−1 to 0.05mol L−1. Figure 4.10 depicts the concentrations ofvanillin and methyl vanillate as a function of reaction time for differentamounts of CuCl2. At a concentration of 0.05mol L−1, the experimentwas stopped after 20 min reaction time as the SS316 stirrer bearings suf-fered from corrosion in the presence of the high chloride concentrations.

While the vanillin concentration is comparable between experimentswith 0.005mol L−1 and 0.01mol L−1 of catalyst, a higher CuCl2 amountresults in a concentration decrease of vanillin after the start of the reac-tion time (t = 0min) due to its discussed conversion. Obviously, a max-imum concentration occurs already during the heating time. This effectis not observed for methyl vanillate. Although a higher catalyst concen-tration decreases the product concentration as well, a rather constantlevel is reached without a pronounced maximum. For the carbomethoxyderivatives of vanillin and methyl vanillate (not shown in the figure), anincrease in product formation at higher amounts of CuCl2 is found. Thisis consistent with both the catalytic action of CuCl2 in their formationand the lower amount of vanillin and methyl vanillate. As expected,their conversion is apparently enhanced for higher amounts of catalyst.

In the same manner, the average molecular weight of the product mix-ture is decreased faster for higher catalyst concentrations. The changein molecular weight as a function of CuCl2 concentration is presented inFigure 4.11. The fastest fragmentation kinetics is observed for a cata-lyst concentration of 0.05mol L−1, which results in an average molecularweight of 386 g mol−1 after only 20min of reaction time. Decreasing theamount of CuCl2 results in a higher final molecular weight, whereas nosignificant difference is observable once again for the two lowest concen-trations of 0.01mol L−1 and 0.005mol L−1.


0 20 40 60 80 100 120 140

cV [

g l-1

]

0.00

0.05

0.10

0.15

0.20

0.25

0.30

0.005 mol L-1

0.010 mol L-1

0.025 mol L-1

0.050 mol L-1

Reaction time [min]

0 20 40 60 80 100 120 140

cM

V [

g l-1

]

0.00

0.05

0.10

0.15

0.20

0.25

0.30

0.005 mol L-1

0.010 mol L-1

0.025 mol L-1

0.050 mol L-1

Figure 4.10: Concentrations of vanillin and methyl vanillate during theacidic oxidation of kraft lignin at 170 C, 10bar oxygenpressure and 10 g L−1 lignin concentration for differentconcentrations of the CuCl2 catalyst.

4.2 Composition of the reaction solvent 69

Reaction time [min]

0 20 40 60 80 100 120

Mw [

g m

ol-1

]

0

500

1000

1500

3500

0.005 mol L-1

0.010 mol L-1

0.025 mol L-1

0.050 mol L-1

Indulin AT ~ 3200 g/mol

Figure 4.11: Weight average molecular weight of the product mix-ture versus reaction time for different concentrations ofcopper chloride catalyst (T = 170 C, pO2

= 10bar,cLig = 10 g L−1).

4.2 Composition of the reaction solvent

In addition to the catalyst, the influence of the solvent composition onan efficient lignin oxidation into monomeric products was investigated.As previously known, acidic conditions in combination with methanolas co-solvent proved to be effective in order to depolymerize lignin andto prevent counterproductive condensation reactions of lignin fragments.The quantitative influence of these parameters is demonstrated in thefollowing.

4.2.1 Fraction and type of the co-solvent

The effect of methanol in preventing repolymerization reactions by com-petitive coupling with lignin fragments has already been proven [10].While this implies the use of a high fraction in the solvent, methanol


is also known to undergo counterproductive degradation into dimethylether during the reaction. Therefore, the influence of methanol wasstudied for volumetric fractions of 0%, 20%, 40%, 60% and 80% todetermine the necessary amount of methanol in the solvent. As wateris required to create hydronium ions and therefore causes the acidic pH,the reaction was not performed in pure methanol.

Analysis of the monomeric products in the reaction mixture showedthat the recovered amount of all products increases continuously withthe methanol content. In case of pure water as solvent, however, noteven traces of monomeric products were found, underlining the necessityof methanol as co-solvent. This is illustrated in Figure 4.12 where theamount of monomeric products after a reaction time of t = 2h is plottedversus the volumetric methanol fraction. Accordingly, the higher themethanol content is, the higher the yields of the monomeric products are.Furthermore, less carbonaceous residues of high molar mass resultingfrom repolymerization are formed when methanol is present in excess.With 80 vol% of methanol in the solvent, the reactor is free of any solidprecipitates after two hours of reaction time and only little amountsare found in experiments with 60 vol% and 40vol%. The amount ofprecipitates, however, increases to 41.8% for 20 vol% methanol and upto 62.1% for the reaction carried out in pure water.

Besides methanol, other co-solvents of the homologous series of alco-hols (ethanol, propanol and butanol) as well as the organic acids formicacid and acetic acids were studied as potential co-solvents. The alcoholsare expected to be effective by competitive coupling with the carboniumions in the same way as it was demonstrated for methanol [10]. Simpleorganic acids, on the other hand, are as well expected to couple with thecarbonium ions due to the negative charge of the deprotonated carboxylgroup. Moreover, they can be advantageous as specific mixing ratioswith water already set the pH to 1 and thereby make the additionalacidification by sulfuric acid redundant.

The results of the investigation revealed that ethanol is the only can-didate out of the chosen co-solvents that can compete with methanol interms of performance. In experiments with the other alcohols and acids,virtually no vanillin was found. In fact, a totally different chemistrywith different reactions and reaction products was observed whereas noknown valuable monomeric products were identified. Moreover, SEC


Methanol fraction in the solvent [vol%]

0 20 40 60 80

Co

nce

ntr

atio

n [

g L

-1]

0.00

0.05

0.10

0.15

0.20

0.25

0.30

Ai /

Ais [

- ]

0.0

0.5

1.0

1.5

2.0

2.5

3.0Vanillin

Methyl vanillate

5-Carbomethoxy-vanillin

Metyhl 5-carbomethoxy-vanillate

Methyl dehydroabietate

Figure 4.12: Vanillin and methyl vanillate concentrations (connectedby solid lines) as well as the relative amounts (dashedlines) of the three other products after a reaction timeof t = 2h (T = 170 C, pO2

= 10bar, cLig = 10 g L−1)for different volumetric methanol fractions in the reactionsolvent (pH 1).

chromatograms showed only a limited effect of these co-solvents in pro-moting the depolymerization of lignin. Ethanol, on the other hand, gavesimilarly good results as methanol. Vanillin was found in comparableamounts in the product mixture and also the other products were de-tected in form of ethanol esters instead of methanol esters. The amountof ring-opening products is higher for ethanol which could indicate astronger degradation of the aromaticity of lignin and the products. Nev-ertheless, the results of ethanol are quite promising, especially as ethanolis a product of potential biorefineries and thus could enhance the inte-gration of the process.


4.2.2 Degradation of the co-solvent

Although a high methanol fraction in the solvent proved essential for thegeneration of monomeric products, its decomposition according to Eq.(4.3) (which likewise happens to ethanol) represents a relevant drawback.

2CH3OHH+

−−→ CH3OCH3 +H2O (4.3)

To study the extent of methanol degradation under the current reac-tion conditions, experiments with pure solvent (80 vol% methanol, pH1) were performed at 170 C with 10bar oxygen in the batch reactor for20min as well as for 120min. Afterwards, the change in liquid volumewas measured and the methanol fraction in the solvent was quantified byHPLC using an Aminex HPX-87H column and external standardization.

The measurements showed a continuous decrease in liquid volume from194.5mL (difference to 200mL due to excess volume of the mixture) be-fore the reaction to 189.2mL after 20min and 173.1mL after 120min.In the same manner, the methanol fraction dropped from 80 vol% to73.3vol% and 64.4 vol%, respectively. Consequently, the methanol con-sumption is 13.3% after 20min of reaction time and 29.5% after 120min.This enhanced demand in process chemicals negatively affects the over-all process performance (as discussed in Chapter 8) and emphasizes theneed for short reaction times.

4.2.3 Acidity of the solvent

As for the methanol content, the presence of hydronium ions in highamounts in the solvent and therefore a highly acidic solution is presumedto be beneficial for lignin fragmentation. On the other hand, increasingcorrosion problems (as already observed for high catalyst amounts) andthe high amount of sulfuric acid required for acidification suggest the useof slightly higher pH values. Four different acidities with pH values of1, 1.5, 2 and 4.35 (corresponding to the intrinsic acidity of the catalystwithout external addition of sulfuric acid) were used in the experiments,keeping all other conditions constant. The amounts of the five mainmonomeric products after a reaction time of t = 2h for the differentstudied pH values is depicted in Figure 4.13.


pH of the reaction solvent

1 2 3 4 5

Co

nce

ntr

atio

n [

g L

-1]

0.00

0.05

0.10

0.15

0.20

0.25

0.30

Ai /

Ais [

- ]

0.0

0.5

1.0

1.5

2.0

2.5

3.0VanillinMethyl vanillate5-Carbomethoxy-vanillinMethyl 5-carbomethoxy-vanillateMethyl dehydroabietate

Figure 4.13: Vanillin and methyl vanillate concentrations (connectedby solid lines) as well as the relative amounts of the threeother significant monomeric products (dashed lines) af-ter a reaction time of t = 2 h (T = 170 C, pO2

=10bar, cLig = 10 g L−1) for different pH values (80 vol%methanol).

The results of these experiments turned up in comparable manneras the previous experiments with different methanol ratios. The totalamount of monomeric products increases with higher concentrations ofhydronium ions (lower pH values). While the final concentrations ofvanillin, methyl vanillate and methyl dehydroabietate decrease contin-uously with increasing pH value, 5-carbomethoxy-vanillin and methyl5-carbomethoxy-vanillate have their maximum value at a pH of 1.5. Al-though this maximum appears reasonable considering for instance thesteep decrease of vanillin concentration between a pH of 1 and 1.5, theexistence of an optimum in this region cannot be explained.

Apart from that, it can be concluded that the formation of products is


better the lower the pH is. For the sake of a high production of aromaticmonomers, the composition of the reaction solvent is not changed to theprevious experiments. A methanol fraction of 80 vol% and a pH of 1 ismaintained throughout all following experiments of this thesis.

4.3 Dependency on lignin feedstock

After the reaction environment in the acidic oxidation of lignin was de-fined within the previous experiments, different types of lignin were usedas feedstock in the next step. The experiments are motivated, on the onehand, by the question whether the products qualitatively (type of prod-ucts) and quantitatively (amount of products) change when other typesof lignin are used. Linked to this, it is also relevant to know, on the otherhand, if the results for the chosen reaction environment are independentof the feedstock which would facilitate the operation of a potential pro-cess. Experiments with commercial lignins and self-made lignins fromenzymatic hydrolysis of lignocellulosic biomass were performed and arediscussed in the following.

4.3.1 Acidic oxidation of other commercial lignins

The four commercial lignins apart from Indulin AT were exposed tothe same acidic oxidation conditions (T = 170 C, pO2

= 10bar, cLig =10 g L−1, pH 1, 80 vol% MeOH, 0.01mol L−1 CuCl2 catalyst) in the batchreactor to compare the results to the previous ones from Indulin AT. TheGC/MS chromatograms of the reaction products from all five commerciallignins after a reaction time of 100min are depicted in Figure 4.14. Peakassignment is the same as in Figure 4.1 (compare Table 4.1) with theinternal standard at a RT of 15.94min.

The composition of the product mixture (as illustrated by the chro-matograms) is on a first glance comparable among the different lignins.The main monomeric products from the oxidation of Indulin AT arein general found in all experiments, especially the peaks of vanillin (RT12.66min) and methyl vanillate (RT 14.23min). Methyl dehydroabietate(RT 22.68min and RT 22.74min, respectively) is the only product thatis not found in all product mixtures as no peaks are found in the chro-

4.3 Dependency on lignin feedstock 75

Re

laitve

ab

un

da

nce

[%

]

Indulin AT (MeadwestVaco)

Lignin, alkali (Sigma aldrich)

Granit Bioplast (Granit SA)

Ultrazine Na (Borregaard)

Organosolv lignin (Fraunhofer)

10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

0

20

40

60

80

100

0

20

40

60

80

100

0

20

40

60

80

100

0

20

40

60

80

100

0

20

40

60

80

10014.23

18.3212.66

22.7415.94 17.0118.52

23.2022.0916.70 24.9919.0015.31 20.5317.9313.8510.78 11.95

18.29

15.9422.6816.9914.22

18.48 23.1312.66 16.50 22.03 24.9023.2220.4718.9717.8615.4312.8010.78 11.43

18.80 19.46

15.9520.72

14.2318.29

24.1320.4816.99 22.4921.3212.66 23.1414.57 17.9113.2710.82 11.52

15.94

14.22

18.48

17.0012.66 22.6823.1321.0120.4718.72 24.9115.14 17.8714.0910.11 10.97

15.95

14.23

18.2918.49 19.8412.66 17.00 22.9720.75 22.49 24.1219.0112.79 14.3610.78 11.74

Retention time [min]

Figure 4.14: GC/MS chromatograms of the reaction products from theacidic oxidation of the five commercial lignins at 170 Cwith 10bar of oxygen pressure and 10 g L−1 lignin concen-tration after a reaction time of t = 120min.


matograms of products from soda lignin (Granit Bioplast) and organo-solv lignin. This can most likely be attributed to the fact that abieticacid as the source of MDHA is already removed from lignin during thepulping process. No significant additional aromatic products were foundin any product sample except in the sample of soda lignin, in which somelarger peaks of non-identifiable products appeared in the chromatogram.In contrast, especially the two kraft lignin experiments shown in the topof the figure give qualitatively comparable results.

Although the oxidation of the different lignin yields basically the sameproducts, the amounts thereof which are obtained in the product mixturediffer clearly as one can assume from the chromatograms. Peak ratios ofthe products related to the internal standard peak at a RT of 15.94minare considerably higher for Indulin AT than for the other types of lignin.This is emphasized by the quantitative results for vanillin and methylvanillate in Table 4.3. The table also lists the average molecular masses ofthe fed lignin and of the resulting product mixture after 100min reactiontime.

Table 4.3: Maximum yield of vanillin and methyl vanillate from theacidic oxidation of different types of lignin as well as the av-erage molecular weight of the lignin and its product mixtureat the end of the reaction.

Maximum yield [wt%] Mw [g mol−1]*

Lignin Vanillin Methyl vanillate Lignin Products

Indulin AT 2.93 2.33 3200±200 536Lignin, alkali 1.17 0.97 6630±780 463Granit Bioplast 1.66 1.07 3972** 589Ultrazine Na 1.66 1.16 19706 456Organosolv 1.59 0.98 1522 634* Product values refer to a reaction time of 120min for kraft lignins and 100min

for other types of lignins, respectively. Except for the two kraft lignins, the listedaverage molecular weights result from single SEC analyses.

** Lignin not completely soluble in eluent.


From these values it becomes apparent, that not only the componentsin the product mixture are roughly the same, but also their averagemolecular weight even though the one of the feedstock differed signifi-cantly. Thus, apart from the fact of lower yields of the main monomericproducts, any of the other types of lignin could as well be used in a po-tential process, making it virtually independent of the feedstock lignin.

4.3.2 Conversion of lignins from enzymatic

hydrolysis

In addition to the commercial lignins, three lignins resulting from theenzymatic hydrolysis of pretreated spruce were converted by acidic oxida-tion. In contrast to the lignins from chemical treatments, these ligninswere not exposed to harsh chemical conditions and are therefore ex-pected to have a more native structure which yields to higher portionsof monomeric products. The three produced lignins differ in the type ofpretreatment (hot water pretreatment - HWP, ball milling at 200 rpm -BM200 and ball milling at 400 rpm - BM400) that was used to break therecalcitrance of the lignocellulosic network (compare Section 3.1).

Analysis of the lignins during production

Analyses on the sugar and lignin content to trace the purity of lignin andthe influence of the pretreatment were performed for the spruce sawdustas well as for the three biomass fractions before the enzymatic hydrolysis(pretreated state) and after the hydrolysis (final lignin). The analyticalresults are presented in Table 4.4. The glucan and xylan percentagesrepresent the total content of all C6 sugars including cellulose and allC5 sugars including hemicellulose, respectively. Although some of themeasured ash contents are negative due to the possible presence of re-maining moisture in the filter paper during initial weighing, the minordeviations of the total percentage from 100% underline a good accuracyof the measurements.

The analysis shows that the original composition of the spruce with43.3% glucose, 20.0% hemicellulose and 35.4% lignin is not significantlychanged during the pretreatment by ball milling. Contrary to this, theelevated temperatures and the presence of the solvent in hot water pre-


Table 4.4: Composition of the three different biomass fractions duringthe production of lignins from the enzymatic hydrolysis ofpretreated biomass. The analyses were conducted accordingto the state of the art NREL protocol.

Weight-based percentage ofFraction Glucan Xylan Ash Lignin Total

Spruce 43.3 20.0 -1.1 35.4 97.6

HWP, pretreated 61.5 3.0 -2.1 42.8 105.2BM200, pretreated 43.6 20.2 -1.8 35.7 97.8BM400, pretreated 39.8 19.3 2.2 34.7 96.0

HWP lignin 39.5 3.4 -2.2 58.1 98.7BM200 lignin 39.9 20.0 -2.2 40.9 98.7BM400 lignin 7.0 10.2 6.0 69.2 92.3

treatment lead to the degradation and dissolution of hemicellulose sothat the resulting biomass almost exclusively consists of glucose andlignin after the pretreatment.

During the following enzymatic hydrolysis, cellulose is hydrolyzed re-sulting in a lower glucan content in all three cases. Due to the dissolutionof the cellulosic sugars as well as the remaining hemicellulosic sugars inthe samples from ball-milling, the lignin percentage increases at the sametime. The enzymatic activity is, however, not the same among the threepretreated fractions. Only 7% of glucan remained in BM400 lignin, in-dicating a very good enzyme accessibility and an efficient pretreatment.In contrast, almost no decrease in glucan content was found for BM200,revealing that the entanglement of the biomass remained mostly present.The different pretreatment efficiencies finally result in the different ligninpurities of 58.1% for HWP, 40.9% for BM200 and 69.2% for BM400.


Acidic oxidation of the lignins

The three lignins from enzymatic hydrolysis were converted by acidic ox-idation in the 50mL batch autoclave for a reaction time of t = 120minat 170 C with 10bar of oxygen pressure and a lignin concentration of10 g L−1. In addition to the quantification of monomeric products byGC/MS, the gas phase products were analyzed by GC/TCD. Whenopening the reactor after the reaction had finished, a significant dropin liquid volume was noticeable in all cases. Analysis of the gas phaserevealed a dimethyl ether content of 29% to 33% (in addition to about5% CO2 and 60% O2), which indicates a relevant degradation of themethanol fraction in the solvent as described before. As this degrada-tion is not beneficial in terms of process economics and the reaction hasturned out rapid regarding monomer production, the experiments wererepeated with a reaction time of t = 20min.

In the experiments with a reduced reaction time, methanol degrada-tion was indeed lowered with dimethyl ether fractions in the gas phase of11% to 16% (plus about 3% CO2 and 80% O2). As expected, the yieldsof the main monomeric products were also higher in the short experi-ments. Figure 4.15 illustrates the calculated yields of vanillin, methylvanillate and 5-carbomethoxy-vanillin. For each lignin, the left bars de-pict results from the 20min oxidation, the patterned bars on the righthand side the lower yields during the 120min oxidation.

All of the lignins that were produced by enzymatic hydrolysis yieldedmore aromatic monomers than the commercial Indulin AT. As expected,the rather native lignins seem to have a less condensed chemical structurewith more original ether bonds and are therefore easier to depolymerizeinto aromatic products. While the lignin from hot water pretreatmenthad already been exposed to higher temperatures during pretreatmentand yielded only slightly more aromatic products than the kraft lignin,the ball-milled lignins and especially the BM200 lignin had been softlypretreated and yielded up to 9.04% of the observed monomers withrespect to the lignin content. Even more products were obtained whenthe original non-pretreated spruce sawdust was converted for 20min inthe batch reactor. A combined yield of 9.83% was measured for thethree monomeric products alone, so that the total yield including theadditional monomeric products, which was not quantified yet during


Indulin HWP BM200 BM400 Spruce

Mo

no

me

r yie

ld [

wt%

]

0

2

4

6

8

10

12


Methyl vanillate

Vanillin

Figure 4.15: Yields of vanillin, methyl vanillate and carbomethoxy-vanillin after acidic oxidation of the lignins from enzy-matic hydrolysis at 170 C, 10bar oxygen with 10 g L−1

lignin for 20min (left bars) and 120min (patterned, rightbars).

these experiments, even exceeds 10%.Surprisingly, it also turned out that the lower the lignin content of

the feedstock is, the more aromatic products are in general obtained.As the oxygen pressure was the same in all experiments, the oxygen tolignin ratio was higher in experiments with high yields and could also besuspected to cause an increased monomer production. This could, how-ever, partly be disproved within the following microreactor experimentsin which different oxygen pressures and thus different oxygen to ligninratios were investigated. Therefore, the improved monomer productionindeed seems to originate from a more native lignin structure with lesscarbon-carbon type bonds.

4.4 Continuous oxidation in a two-phase flow microreactor 81

4.4 Continuous oxidation in a two-phaseflow microreactor

The previously described experiments on the acidic oxidation of ligninwere all performed in high pressure batch reactors of different size. Whilethese autoclaves are commonly used for reactions that presume highpressures and temperatures, several drawbacks were encountered duringtheir use in the lignin oxidation experiments. Especially when it comesto a more detailed and profound investigation on the lignin conversionas the remaining studies on the main process parameters temperature,pressure and lignin concentration, the use of high pressure batch reactorsis limited due to the following disadvantages:

• Since the reactor wall dimensions are designed for high pressureapplications, heating and cooling the steel body of the reactor takestime and may generally not be neglected when studying a reaction.During this time a totally different reaction behavior can be presentin the reactor which may falsify the results and lead to wrongconclusions. This is especially true when the time to reach thereaction temperature and to cool down from it even exceeds thereaction time at constant temperature as in the experiments withtransition metal catalysts. Maximum concentrations were observedwithin less than 20min of reaction time while heating and coolingaccounted for an additional 24min and 60min, respectively.

• Due to the heating and cooling phases of a batch reactor, runninga single experimental data point within a parameter study can bevery costly in terms of time. Therefore, a detailed investigation onreaction parameters can be tedious and may contain unavoidableerrors (e.g. different heating and cooling times at different processtemperatures).

• Although batch reactors may be designed for elevated pressures,safety reasons can enforce a maximum experimental pressure inthe laboratory. This can be due to both, constructional reasonsand process conditions. During the acidic oxidation of lignin, themaximum applicable oxygen pressure was both limited by the risk


of explosion due to the ignitable gas mixtures in the reactor andthe risk of titanium autoignition at high oxygen pressures.

These drawbacks can be overcome by performing the reaction con-tinuously in a microreactor. Due to the high heat transfer within themicroreactor the reaction mixture is rapidly heated to and cooled downfrom the reaction temperature, thereby avoiding transient phases wherereactivity and selectivity can differ significantly [89]. The small scale ofthe reactor assures safe operation even when working at high pressures[90]. Furthermore, the volumes processed in a microreactor are suffi-cient for chemical analysis when studying a reaction but minimize wastegeneration within an experiment.

For this reason, the acidic oxidation of lignin was transferred from thebatch autoclave into a continuous two-phase flow capillary microreactorin order to study the influence of the main process parameters. Theinfluence of temperature (150 C to 250 C), pressure (32 bar to 96bar)and lignin concentration (2.5 g L−1 to 10 g L−1) on the acidic oxidationof lignin is described below.

4.4.1 Product concentrations in the microreactor

The concentrations of the main products for a typical experiment in themicroreactor at 190 C and 48bar with a lignin concentration of 2.5 g L−1

are depicted in Figure 4.16. The statistical error of the concentrationscan generally be estimated from one experiment which was done in trip-licate. The 99% confidence interval of the product concentrations wasbetween 8.6% (vanillin) and 14.7% (methyl dehydroabietate) with anaverage of 11.7%. As slug flow prevails in the microreactor, residencetime at steady state conditions was chosen on the x-axis in all diagramsbut can directly be transferred with the total volumetric flow rate to anaxial position in the microreactor.

The graph visualizes the formation and decomposition of the productswith time. Vanillin and methyl vanillate experience a maximum concen-tration and are later consumed in favor of their carbomethoxy deriva-tives. The latter continuously increase in concentration in the same wayas methyl dehydroabietate until the end of the studied residence times.The observed concentration maxima for vanillin and methyl vanillate are


Residence time [min]

0 2 4 6 8 10 12 14

Co

nce

ntr

atio

n [

mg L

-1]

0

20

40

60

Vanillin

Methyl vanillate


Methyl 5-carbomethoxy-vanillate


Figure 4.16: Concentrations of the main lignin oxidation products inthe microreactor at 190 C and 48bar with a lignin con-centration of 2.5 g L−1.

obtained within less than twelve minutes of residence time. This under-lines the rapid conversion of lignin in the presence of the copper chloridecatalyst at these reaction conditions and the basic need for studyingshort reaction times in a continuous microreactor.

The results from the microreactor were as well compared to the for-mer batch experiments in order to validate the continuous microreactorsetup but also to detect differences in kinetics and selectivity due to thedifferent nature of the reactors. For this purpose, an experiment at thepreviously used batch conditions (170 C and 32bar) with 5 g L−1 ligninconcentration was performed in the microreactor. The concentrations ofthe main products vanillin and methyl vanillate versus reaction time inthe batch reactor and versus residence time at steady state conditions inthe microreactor, respectively, are depicted in Figure 4.17. The negative


reaction time in the batch experiments results from the roughly 20minto heat up the batch reactor to 170 C.

Reaction/residence time [min]

-20 -10 0 10 20 30 40

Co

nce

ntr

atio

n [

mg L

-1]

0

10

20

30

40

50 Vanillin (batch)

Vanillin (micro)

Methyl vanillate (batch)

Methyl vanillate (micro)

Figure 4.17: Vanillin and methyl vanillate concentrations in the batchreactor and the microreactor during a lignin oxidation ex-periment at 170 C and 32bar with a lignin feed concen-tration of 5 g L−1.

In case of vanillin, the concentrations in the microreactor and the batchreactor virtually match. The formation of vanillin in the microreactor israpid and reaches quite high concentrations already within less than twominutes of residence time. The methyl vanillate concentration, on theother hand, is always lower in the microreactor than in the batch reactor.The same is true for methyl dehydroabietate and the two carbomethoxyderivatives which are not shown in the graph. The lower concentrationscan most likely be attributed to the relatively long additional heatingperiod in the batch reactor with even different product formation anddegradation kinetics.


4.4.2 Influence of temperature

The influence of the reaction temperature on product formation wasstudied in the microreactor between 150 C and 250 C with 2.5 g L−1

lignin concentration at 48bar or a pressure just above the dew pointline, if necessary. The experiments revealed a major effect of the reactiontemperature on the concentrations of the monomeric reaction productsduring the acidic oxidation of kraft lignin. Their concentrations versusreaction time for different temperatures are depicted in Figure 4.18.

For 150 C and 170 C the vanillin concentration (graph a)) contin-uously increases with residence time and approaches a constant level.Starting at 190 C, a maximum in vanillin concentration becomes visi-ble. With increasing temperature the peak concentration reaches highervalues at shorter reaction times which is in accordance to results frombatch experiments in the hour range at lower temperatures [91]. At210 C a vanillin concentration of 74.2mg L−1 is found, exceeding themaximum value of 9.9mg L−1 at 170 C by a factor of more than seven.In addition, for temperatures above 210 C the vanillin peak seems tooccur rapidly within considerably less than one minute of residence timein the microreactor and only the concentration decline can be observed.Based on these results, even higher vanillin concentrations than the max-imum value at 210 C can be expected for even shorter reaction timesand higher temperatures.

The same phenomenon of earlier and higher concentration maximawith increasing temperature is generally also found in the concentrationprofiles of the other products (graphs b)-d)). For methyl vanillate, a simi-lar behavior as the one of vanillin can be observed although the maximumis by far not that pronounced and the concentration is lower by a factorof approximately four compared to vanillin. 5-carbomethoxy-vanillin,as a subsequent product of vanillin, experiences a steady increase inconcentration until temperatures of 210 C. For higher temperatures, adecomposition of the product becomes as well visible but with a con-centration maximum at a later residence time than vanillin and methylvanillate. Similar concentrations as for 5-carbomethoxy-vanillin were ob-tained for methyl dehydroabietate. Interestingly, methyl vanillate whichused to be the major product in the batch experiments besides vanillindoes not profit as much from higher temperatures as the other prod-


0 5 10 15 20

Co

nce

ntr

atio

n [

mg

L-1

]

0

20

40

60

80

150°C 48bar

170°C 48bar

190°C 48bar

210°C 48bar

230°C 72bar

250°C 96bar

0 5 10 15 20

0

20

40

60

80

150°C 48bar

170°C 48bar

190°C 48bar

210°C 48bar

230°C 72bar

250°C 96bar


0 5 10 15 20

Co

nce

ntr

atio

n [

mg

L-1

]

0

20

40

60

80

150°C 48bar

170°C 48bar

190°C 48bar

210°C 48bar

230°C 72bar

250°C 96bar


0 5 10 15 20

0

20

40

60

80

150°C 48bar

170°C 48bar

190°C 48bar

210°C 48bar

230°C 72bar

250°C 96bar

a) Vanillin b) Methyl vanillate

c) 5-Carbomethoxy-vanillin d) Methyl dehydroabietate

Figure 4.18: Influence of temperature on the concentrations of a)vanillin, b) methyl vanillate, c) 5-carbomethoxy-vanillinand d) methyl dehydroabietate in experiments with alignin feed concentration of 2.5 g L−1.


ucts do. In the experiments with high reaction temperatures, methyldehydroabietate emerges as the second most abundant monomeric prod-uct after vanillin and even 5-carbomethoxy-vanillin is found in higherconcentrations than methyl vanillate.

In general, the positive effect of temperature on a high and fast productformation is present for all products and temperature turns out as themost influential reaction parameter. High reaction temperatures appearcrucial for both a high yield of any aromatic product and short residencetimes in the reactor. At 250 C, the experiments reveal a maximumtotal yield (as sum of the five single product yields) of 5.02wt% afteronly 0.65min of residence time. This represents a significant increasecompared to the total yield of vanillin and methyl vanillate (2.14wt%)which was obtained from the same type of lignin in the batch experiment(see Table 4.3). When assuming the same yield ratio between batchreactor and microreactor results for other types of lignin, a total yieldof 11.8wt% for Indulin AT and even 19.0wt% for the enzymatic ligninBM200 could be obtained at the same conditions in the microreactor ifthese lignins were sufficiently soluble.

4.4.3 Influence of pressure

To investigate the influence of the oxygen pressure on the formation ofmonomeric products, experiments at three different pressures of 32bar,48bar and 64bar were performed for a temperature of 170 C with2.5 g L−1 lignin concentration. The relatively high pressure values cho-sen are required to keep the reaction solvent in liquid state. In contrastto temperature, the pressure turned out to have a minor impact on theyield of monomeric reaction products formed. To illustrate this fact, thetop graphs a) and b) in Figure 4.19 show the concentrations of vanillinand methyl vanillate versus residence time for different pressures.

While pressure seems to somewhat affect vanillin formation, the con-centrations of all other products like methyl vanillate are virtually notinfluenced by pressure. For vanillin, especially the concentrations at48bar is for some unknown reason lower than the ones at 32bar and64bar which are rather similar. However, the differences do not showany reasonable trend and could not be reproduced in a different exper-imental series. Thus, no significant pressure influence on the vanillin


0 5 10 15 20

Con

cen

tration

[m

g L

-1]

0

20

40

60

80

32 bar 2.5 g L-1

48 bar 2.5 g L-1

64 bar 2.5 g L-1

0 5 10 15 20

0

20

40

60

80

32 bar 2.5 g L-1

48 bar 2.5 g L-1

64 bar 2.5 g L-1


0 5 10 15 20

Yie

ld %

[g

gL

ig-1]

0.0

0.5

1.0

1.5

2.0

2.5

3.0

32 bar 2.5 g L-1

32 bar 5.0 g L-1

32 bar 10.0 g L-1


0 5 10 15 20

0.0

0.5

1.0

1.5

2.0

2.5

3.0

32 bar 2.5 g L-1

32 bar 5.0 g L-1

32 bar 10.0 g L-1

a) Vanillin b) Methyl vanillate

d) Methyl vanillatec) Vanillin

Figure 4.19: Influence of pressure (a) and b)) and lignin feed concen-tration (c) and d)) on the concentrations of vanillin (left)and methyl vanillate (right). The experiments were con-ducted at 170 C with 2.5 g L−1 lignin feed concentrationand at 32bar, respectively.


concentration can be concluded either.The negligible influence of oxygen pressure can generally be attributed

to the high oxygen concentration in the liquid phase at high pressure.This is especially enhanced in the present case by the methanol fractionin the solvent which promotes oxygen solubility [92]. On the other hand,the amount of oxygen consumed in the reaction is low compared to theamount dissolved in the liquid (less than 1% at 170 C and 48bar with2.5 g L−1 lignin concentration as estimated based on literature [93]) andtherefore hardly affects the product concentrations. Nevertheless, highpressures are required to avoid complete evaporation of the solvent whenrunning the reaction at high temperatures as discussed previously.

4.4.4 Influence of lignin concentration

The influence of lignin feed concentration on the product formation wasstudied at 170 C and 32bar. The concentrations of vanillin and methylvanillate are depicted in the bottom graphs c) and d) of Figure 4.19.Since product concentrations are not meaningful on the y-axis of thegraphs when working with different lignin concentrations, their yielddefined as product concentration related to lignin concentration is shown.From both graphs it becomes apparent that approximately the sameyield of products is obtained when increasing the lignin concentrationfrom 2.5 g L−1 to 10 g L−1. The same is true for all other products whichare not shown in the figure.

Although no influence of pressure and lignin feedstock concentrationwas found, these parameters were only studied within a certain range forone specific temperature. For other reaction conditions they may verywell become important. Especially concentrations higher than 10 g L−1

become relevant when it comes to upscaling for commercial application.These higher concentrations could, however, not be realized in the mi-croreactor because of insufficient lignin solubility.

4.4.5 Parameter estimation of the vanillin kinetics

Among the different reaction parameters that were varied during themicroreactor experiments, temperature emerged as the parameter thatessentially influences the product formation. However, the previous mi-


croreactor experiments covered just a definite range of residence times inthe minute range and high vanillin yields were already found within lessthan one minute at high temperatures. For this reason, the concentra-tions of vanillin as the most important product were simulated to find themaximum concentrations. To determine the required vanillin formationand degradation kinetics, a parameter estimation based on the experi-mental concentrations at different temperatures was performed using thesoftware package MATLAB R©.

The model for the parameter estimation included two reactions: theoxidation of lignin to vanillin and the subsequent degradation of vanillinto form products like 5-carbomethoxy-vanillin. While the stoichiometryfor the second reaction is obvious, it is not trivial to determine the stoi-chiometry for the formation of vanillin from lignin. The following dodgeallowed to find out a stoichiometric factor for this reaction anyway. Lit-erature reports a maximum possible vanillin yield from Indulin AT of10.6% based on nitrobenzene oxidation experiments [94]. Although In-dulin AT and kraft lignin from Sigma-Aldrich yield to different amountsof vanillin and methyl vanillate on a mass basis, the yields on a molarbasis are comparable when calculating with the measured average molec-ular weights. The molar yields are 0.57mol mol−1 and 0.51mol mol−1

for Indulin AT as well as 0.49mol mol−1 and 0.42mol mol−1 for the otherkraft lignin. From these data, a stoichiometric coefficient for the forma-tion of vanillin from lignin of 2.23molV/molLig can be calculated.

Due to the excess of oxygen and methanol, the reaction rates of bothreactions were assumed to be only dependent on the concentration of themain reactants (viz. lignin and vanillin). Based on the stoichiometriccoefficients of the main reactants, both reactions were implemented intothe reaction model with a first order Arrhenius approach. The kineticapproach results as follows

r1 = k1 · cLig = k1,0 · e(E1/RT ) · cLig (4.4)

r2 = k2 · cV = k2,0 · e(E2/RT ) · cV (4.5)

dcVdt

= r1 − r2 (4.6)

For each temperature, the rate constants k1 and k2 were fitted to the


corresponding experimental data set in MATLAB R© using the method ofleast squares. The logarithms of their resulting values were plotted in anArrhenius plot versus the inverse temperature to find the values for thepre-exponential factors k1,0 and k2,0 as well as for the activation ener-gies E1 and E2. The Arrhenius plot for the formation and degradationreaction of vanillin is depicted in Figure 4.20.

1/T [K-1

]

0.0019 0.0020 0.0021 0.0022 0.0023 0.0024

ln (

k 1,k

2)

[ln m

in-1

]

-6

-4

-2

0

2

4

r1 (Formation)

r2 (Degradation)

Linear fit

( )

Figure 4.20: Arrhenius plot of the formation (r1) and the degradation(r2) of vanillin resulting from the fit of the model to theexperimental microreactor data sets.

The plotted values show a linear behavior for both reactions andthereby confirm the approach which was used in the model. Both re-gression lines result in a coefficient of determination (R2) of above 0.99.The only value that was disregarded in the regression is the formationrate constant at a temperature of 250 C (≡ 0.0019K−1). However,this appears legitimate as the formation of vanillin can hardly be de-termined from the experimental data which solely show a decreasingconcentration at this temperature (see Figure 4.18 a)). The values for


the pre-exponential factors and the activation energies that result fromthe Arrhenius plot are listed in Table 4.5.

Table 4.5: Pre-exponential factors and activation energies for the for-mation and degradation of vanillin resulting from the pa-rameter estimation.

k1,0 [s−1] E1 [kJ mol−1] k2,0 [s−1] E2 [kJ mol−1]

1.41× 1017 170.8 1.77× 104 56.3

4.4.6 Simulation of the vanillin concentration

Based on the kinetic model and the kinetic constants which were deter-mined in the parameter estimation from the experimental microreactordata, the vanillin concentrations for different temperatures were simu-lated in MATLAB R©. The corresponding profiles are illustrated togetherwith the experimental data in Figure 4.21.

For all temperatures, the simulated data fits the experimental valuesquite well, especially when keeping in mind that the simplified reactionmodel hardly matches the real chemical reactions involved in lignin ox-idation. The simulated vanillin concentrations show exactly the trendthat was already predicted from the experimental microreactor results.The higher the reaction temperature is, the faster lignin is oxidized andthe higher and earlier the concentration maxima for vanillin appear.While at 150 C the vanillin concentration does not exceed 0.05mmol L−1

and the complete oxidation of lignin takes more than 2h, a general maxi-mum vanillin concentration of 0.75mmol L−1 is simulated for 250 C at aresidence time of less than 3 s. This maximum concentration is 15 timeshigher than the peak value at 150 C. Moreover, it equals 89.3% of thetheoretical maximum which was reported in the nitrobenzene oxidationexperiments in literature. For Indulin AT, this corresponds to a vanillinyield of 9.5wt%.

The appearance of a vanillin peak concentration at such short reactiontimes is, however, only valid for low lignin concentrations. For higher,industrially relevant concentrations of lignin, the vanillin maximum can


0 50 100 1500

0.1

0.2

0.3

0.4150°C

Concentration [m

mol L

-1]

Vanillin (exp)

Vanillin (sim)

Lignin

0 50 1000

0.1

0.2

0.3

0.4170°C

Vanillin (exp)

Vanillin (sim)

Lignin

0 10 20 300

0.1

0.2

0.3

0.4190°C

Concentration [m

mol L

-1]

Vanillin (exp)

Vanillin (sim)

Lignin

0 5 10 15 200

0.1

0.2

0.3

0.4

0.5

210°C

Vanillin (exp)

Vanillin (sim)

Lignin

0 5 100

0.1

0.2

0.3

0.4

0.5

0.6

230°C

Concentration [m

mol L

-1]


Vanillin (exp)

Vanillin (sim)

Lignin

0 5 100

0.2

0.4

0.6

0.8250°C


Vanillin (exp)

Vanillin (sim)

Lignin

Figure 4.21: Simulated vanillin concentrations in the microreactor atdifferent temperatures in comparison with the experimen-tal data.


be shifted towards longer reaction times due to a limited lignin solubilityand lower specific contents of methanol, acid, and oxygen, respectively.Nevertheless, the key to high vanillin yields from lignin are high reactiontemperatures and short residence times. Although the realization ofthese parameters values on larger scale seems unfeasible, it should bestriven for values which come as close as possible.

4.5 Summary of the lignin oxidation results

The major aspects of the acidic oxidation of lignin were experimentallyinvestigated in this chapter. While copper chloride turned out to be avery efficient transition metal salt catalysts, a methanol fraction of 80%and a pH value of 1 in the solvent have been proven beneficial for theformation of aromatic monomers. The reaction system is roughly inde-pendent on the lignin feedstock in terms of the type of products that areformed. However, Indulin AT yielded the highest quantities of productsamong the commercial lignins and was only excelled by the more nativenon-commercial lignins from enzymatic hydrolysis of pretreated biomass(up to 10wt% of monomeric products).

In contrast to pressure and lignin feedstock concentration, tempera-ture emerged as most important reaction parameter. High temperaturesand short residence times turned out decisive for high product yields.At 250 C, a maximum total yield of 5.02wt% after only 0.65min ofresidence time was obtained, representing a significant increase for thechosen alkali lignin compared to the batch experiments which yielded atotal of 2.14wt% of vanillin and methyl vanillate. Assuming the sameyield ratios between the batch and the microreactor experiment for otherlignins, a total yield of 11.8wt% would result for Indulin AT in the mi-croreactor and even 19.0wt% for the enzymatic lignin BM200. Based onthe kinetic simulations, nearly 10wt% of vanillin alone can be presumedfrom Indulin AT in the microreactor at 250 C for residence times in thesecond range.

95

Chapter 5

Extractive recovery of the

reaction products

In terms of a continuous processing for the valorization of lignin, thereaction products from the acidic oxidation need to be recovered fromthe reaction solvent for further processing. Thereby, the reaction solventincluding the catalyst, which is virtually insoluble in organic solventsaccording to the literature, can be recycled back to the reactor [95].Liquid-liquid extraction with non-polar or polar aprotic organic solventsis known to be a potential approach to recover aromatic components likevanillin from aqueous solutions [7, 96]. Furthermore, the extraction withsupercritical carbon dioxide is patented as an alternative approach in thiscontext [97]. In this chapter, the recovery of the reaction products fromthe aqueous reaction solvent by both types of extraction is discussed.

5.1 Liquid-liquid extraction with organicsolvents

5.1.1 Potential extraction solvents

The choice of the extraction solvent is decisive in order to achieve ahigh recovery of products at a minimum operational effort and with aminimum loss of reaction solvent in the extraction step. From the com-prehensive pool of organic solvents, potential solvents have to fulfill thecriteria listed below. Out of these criteria, high distribution coefficientsand thus a high capacity of the aromatic products in the extraction sol-vent represents the key criterion for the selection of the solvent. Unlike

96 5. Extractive recovery of the reaction products

in other extraction processes, selectivity is of minor importance in theinvestigated extraction step since the recovery of all products is intended.

The requirements for a potential extraction solvent are:

• High distribution coefficients of the aromatic products between theselected solvent and the reaction solvent

• Large miscibility gap in the ternary system extraction solvent-water-methanol

• Sufficient difference in density to the reaction solvent for phaseseparation

• Easy recoverability and low volatility

• Thermal and chemical stability

• Low environmental impact and no toxicity to living organisms

• High availability at low cost

Five potential organic solvents were investigated in the present the-sis and evaluated according to the mentioned criteria. The non-polarsolvents chloroform, toluene, hexane, diethyl ether as well as the polaraprotic solvent ethyl acetate were first theoretically studied with AspenPlus R© concerning miscibility gaps, and afterwards experimentally stud-ied in the extraction of the reaction products from the reaction solvent.

5.1.2 Miscibility of the extraction solvents

Ternary phase diagrams of each of the selected extraction solvents withwater and methanol were generated in Aspen Plus R© to evaluate themiscibility and the resulting formation of two separate liquid phases. Asthe choice of the property model turned out to be of no matter for thephase diagram, the non-random two-liquid (NRTL) model was used forthe calculation. Figure 5.1 illustrates the phase behavior of the systemchloroform-water-methanol at 20 C and 1bar. The phase diagrams forthe other extraction solvents can be found in Appendix A.3.

5.1 Liquid-liquid extraction with organic solvents 97

xWater

[mol mol-1]

x Chlo

rofo

rm [

mol

mol

-1 ]x

Meth

anol [m

ol mol -1]

0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55 0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

0.45

0.5

0.55

0.6

0.65

0.7

0.75

0.8

0.85

0.9

0.95

0.9

50.9

0.8

50.8

0.7

50.7

0.6

50.6

0.5

50.5

0.4

50.4

0.3

50.3

0.2

50.2

0.1

50.1

0.0

5

Figure 5.1: Ternary phase diagram of the system chloroform-water-methanol at 20 C and 1bar showing the miscibility gapand four selected tie lines.

All of the studied extraction solvents exhibit a region of immiscibilityin the ternary phase diagram. As with chloroform, methanol is com-pletely miscible with all potential extraction solvents and thus promotesthe miscibility of the aqueous methanol/water phase with the extractionsolvent. Hence, the high methanol fraction in the reaction solvent turnsout disadvantageous for the extraction step. On the other hand, thedirection of the tie lines that point towards the pure extraction solventis advantageous since only low amounts of methanol are extracted intothe organic phase.


In comparison with the illustrated phase diagram for chloroform, theuse of toluene and hexane results in a larger miscibility gap whereas di-ethyl ether and ethyl acetate cause a smaller one (see Appendix A.3). Al-though the mixing line of a binary 80 vol% methanol/20vol% water mix-ture (corresponding to 0.64mol mol−1 methanol) with chloroform crossesthe two-phase area in the ternary phase diagram, no phase separationoccurs upon addition of chloroform to the real reaction mixture in thelaboratory. Water had therefore to be added to inhibit miscibility ofthe mixture when chloroform was used as extraction solvent. The sameis true for diethyl ether and ethyl acetate that exhibit an even smallermiscibility gap than chloroform. No addition of water was required fortoluene and hexane which turn out to be the best extraction solvents inthis regard.

In addition to the thermodynamical phase behavior, the distributioncoefficients of vanillin, which was the only product with thermodynam-ical data available, were calculated between the different solvents andthe reaction solvent with Aspen Plus R©. However, the data turned outpoor and neither did different property models result in the same valuesnor did they fit experimental values from the literature [96]. Hence, thedistribution coefficients were determined experimentally.

5.1.3 Determination of distribution coefficients

To compare the performance of the selected extraction solvents regard-ing product capacity, the distribution coefficients of the commerciallyavailable monomeric products vanillin and methyl vanillate were deter-mined experimentally for each solvent. For this purpose, model solutionsthat mimic the reaction solvent with varying fractions of methanol anddefined amounts of vanillin and methyl vanillin were produced in thelaboratory. A defined volume of the model solution was then mixedwith the same amount of each extraction solvent in a separating funnelwhich was afterwards vigorously shaken for several minutes. After com-plete phase separation, the volume of the phases was measured with ameasuring cylinder and the extract phase was analyzed concerning thesolute concentrations by GC/MS.

The distribution of a solute i between two immiscible phases I and II


is governed by its phase equilibrium condition [98].

µIi = µII

i (5.1)

With the chemical potential of component i in each phase

µi = µ(0)i +RT ln(xiγi) (5.2)

Eq. (5.1) can be rearranged for adequate dilution of the solutes to ageneral form of Nernst’s distribution law

K∗i =

xIi

xIIi

= exp1

RT(µ

II,(0)i − µ

I,(0)i ) (5.3)

The distribution law states that the ratio between the concentrationsof a solute i in two different liquid phases at a certain temperature isconstant. A more applicable form of Nernst’s law with regard to liquid-liquid-extraction relates the distribution coefficient Ki with the molarconcentrations of i in the extract phase E and the raffinate phase R

Ki =cEicRi

(5.4)

Accordingly, high values of Ki correspond to a good extraction of com-ponent i by the extraction solvent and vice versa. The distribution co-efficients which are presented in the following were all calculated basedon Eq. (5.4).

Figure 5.2 depicts the vanillin distribution coefficients between the dif-ferent extraction solvents and the reaction solvent versus the methanolfraction in the model solution. As hexane gave the by far worst re-sults with virtually no solutes in the extract phase, it was not furtherconsidered as potential extraction solvent. The error bars represent anaverage uncertainty in GC measurements of 1.5% for vanillin and 3.5%for methyl vanillate. However, as the denominator in Eq. (5.4) is calcu-lated by mR

i = m0i −mE

i , these small deviations can cause huge errorswhen almost the complete solute is extracted (mE

i → m0i ). Therefore,

the high Ki values at 0 vol% methanol can rather be considered approxi-mate distribution coefficients. No values were obtained for ethyl acetate


0

5

10

15

20

25

30

35

40

45

0 vol% 20 vol% 40 vol% 60 vol%

Methanol fraction in the model solution

KV [

- ]

Chloroform

Toluene

Diethyl ether

Ethyl acetate

Figure 5.2: Vanillin distribution coefficients between the potential ex-traction solvents and the model solution for different frac-tions of methanol in the model solution.

and diethyl ether when the model solution contained 60 vol% methanolas the resulting mixture only formed one single phase.

According to the figure, the methanol fraction in the solvent once againturns out disadvantageous for the extraction. The higher the methanolcontent is, the lower are the vanillin distribution coefficients. Espe-cially chloroform, ethyl acetate and diethyl ether show high Ki valuesand almost complete extraction in one single step when extracting outof pure water, whereas toluene seems less suited as extraction solvent.For methyl vanillate, the measured results were comparable with evenslightly higher distribution coefficients than for vanillin. Chloroform,ethyl acetate and diethyl ether were as well the solvents that extractedthe highest amounts of methyl vanillate.


5.1.4 Capacity for lignin oxidation products

As not only the extraction of vanillin and methyl vanillate are of in-terest but the extraction of all reaction products, the overall recoverywas evaluated by analyzing the mass that can be extracted from an ac-tual product mixture. 15mL of a mixture that contained approximately150mg of lignin oxidation products was diluted with the same volumeof water and then extracted once at a 1:1 ratio with 15mL of the ex-traction solvent. The extracted mass was gravimetrically measured aftercomplete evaporation of the solvent and drying of the sample in a vacuumoven overnight. Table 5.1 lists the results for the different solvents. Inaddition, the average molecular weight of products in the extract phaseis specified.

Table 5.1: Extracted mass of lignin oxidation products from 15mL ofa product mixture (cLig = 10 g L−1) as well as the averagemolecular weight of products in the extract phase for thedifferent extraction solvents.

Solvent Extracted mass [mg] ME

w [g mol−1]

Chloroform 111.8 529Toluene 52.5 231Diethyl ether 70.5 340Ethyl acetate 130.5 600

As the table shows, ethyl acetate and chloroform are the solvents thatextracted the highest amounts of products from the reaction mixture (∼75% to 87%). Diethyl ether and toluene, on the other hand, did not onlyextract a lower mass, it also appears that these solvents rather extractproducts of lower molecular weight and show a certain selectivity to-wards these compounds. Since ethyl acetate and chloroform also recoversignificant amounts of high molecular weight products, they emerge asthe best solvents when it comes to the extraction of all reaction productsas intended in the discussed process step.


5.2 Extraction with supercritical carbondioxide

The extraction with supercritical carbon dioxide represents an alterna-tive approach for the recovery of the lignin oxidation products. Carbondioxide is an abundant and environmentally benign solvent that caneasily be removed from a supercritical extraction step by lowering thepressure. Literature reports a large miscibility gap in the ternary systemcarbon dioxide-water-methanol [99]. Moreover, the ability of supercriti-cal carbon dioxide to extract lignin-derived aromatic monomers was ingeneral already proven in several patents, e.g. [100]. Therefore, the ex-traction of the lignin-derived products from the acidic aqueous reactionmixture with supercritical carbon dioxide was investigated in our labo-ratory [101]. Supercritical carbon dioxide and the reaction mixture wereseparately fed to a silicon/glass two-phase flow microreactor and the in-fluence of the pressure on the extracted amount of monomeric productswas studied. The extracted quantities versus pressure are depicted inFigure 5.3. The experimental procedure and the results of the study arereported in detail in the mentioned literature.

Compared to the organic solvents that were used in liquid-liquid-extraction, the amount of extracted products with supercritical carbondioxide turned out to be rather low. While methyl dehydroabietate isalmost completely recovered in the whole pressure range, virtually allof the vanillin and also large amounts of the the other products remainin the raffinate. Correspondingly, the resulting distribution coefficientsare all lower than unity and even lower than 0.1 for vanillin and methylvanillate. Thus, despite it offers several advantages, the extraction withsupercritical carbon dioxide is not the method of choice for the com-plete recovery of reaction products. Nevertheless, the high selectivity formethyl dehydroabietate (and partly for the two carbomethoxy products)potentially qualify the supercritical extraction within the separation andpurification of the monomeric products as discussed in Chapter 7.

5.3 Conclusions for the extraction step 103

Pressure [bar]

60 70 80 90 100 110 120

Ext

racte

d a

mo

unt

[%]

0

20

40

60

80

100

Vanillin

Methyl Vanillate




Figure 5.3: Extracted amount of monomeric products by extractionwith supercritical carbon dioxide at different pressures[101].

5.3 Conclusions for the extraction step

Based on the criteria for a good extraction solvent in the beginning ofthis chapter, ethyl acetate emerges as the best solvent for the desired re-covery of all reaction products from the lignin oxidation product mixture.On the one hand, it offers a high capacity for the oxidation products,which represented the main criterion in the evaluation. On the otherhand, the other solvents encounter different drawbacks which disqualifythem as extraction solvents. Toluene and diethyl ether, for instance,extract comparably low amounts of products from the product mixture.In addition, diethyl ether is highly volatile and tends to form unstableperoxides over time, making its use risky. While all of the studied extrac-tion solvents are harmful or irritant to humans and animals, chloroform,which also performed very well in this study, is not desired as extractionsolvent as it is additionally suspected to cause cancer [102].


For these reasons, ethyl acetate was finally chosen as extraction sol-vent within this thesis. Moreover, all of the selection criteria for theextraction solvent are in principle fulfilled by it. The small miscibilitygap in the ternary system with the reaction solvent, however, demandsfor high amounts of water addition or methanol evaporation before twodifferent liquid phases are formed during extraction. This has to beconsidered in the process design and represents the main disadvantageof ethyl acetate as extraction solvent. In addition, the degradation ofethyl acetate into acetic acid and ethanol, which occurs in acidic condi-tions under the influence of heat, has to be suppressed. Vigneault et al.,who also reported ethyl acetate as the best extraction solvent to recoverlignin-derived aromatic monomers in a different study, therefore proposevinyl acetate and n-butyl acetate as possible alternatives to ethyl acetate[103]. Nevertheless, ethyl acetate is used as extraction solvent in the fol-lowing chapters as well as in the final process considerations, owing toits desirable performance in liquid-liquid extraction.

105

Chapter 6

Membrane separation of lignin

oxidation products

To separate the valuable monomeric lignin oxidation products from theproducts of higher molecular weight, a fractionation step, which followsthe previously discussed extraction of the reaction products, is required.As the separation of the products is based on the number of their aro-matic rings, respectively, on their molecular weight, membrane separa-tion represents a simple but effective approach for the fractionation.

In the context of lignin, membrane processes have mainly been dis-cussed for the recovery of inorganic chemicals or the isolation of ligninfrom black liquor [26, 104]. Within the valorization of lignin, however,membrane processes have played a minor role up to now. Research hasso far focused on chemical transformations towards specific products.Further processing of the product mixture has mostly either not beendiscussed or the product mixture has been employed without prior sepa-ration or fractionation. Nevertheless, some aqueous applications of mem-brane separation to improve the benefit of lignin from biomass can befound in the literature. For example, Toledano et al. studied the poten-tial of lignin pre-fractionation by ultrafiltration to produce fractions forspecific applications [105]. Ultrafiltration with ceramic membranes wasused by Zabkova et al. to separate vanillin from aqueous vanillin/ligninmixtures mimicking lignin oxidation products [106]. Recently, Konc-sag and Kirwan published results on the concentration of dilignols andtrilignols from enzymatic conversion of lignin using solvent stable ultra-and nanofiltration membranes [107]. However, no application of mem-brane separation in the processing of product mixtures from chemicaltransformations of lignin exists.

106 6. Membrane separation of lignin oxidation products

The objective of this chapter is to find a membrane that allows for aneffective separation of monomeric lignin oxidation products from reac-tion products of higher molecular weight. As the complete separationof the different monomers from the complex multi-component mixtureseems not feasible in one step, the investigated membrane separationis regarded as a first fractionation step in the processing of the prod-ucts. Therefore, the main goal is to obtain two fractions considerablyenriched in monomers and high molecular weight products, respectively,that can further be processed and purified. As the lignin oxidation prod-ucts are dissolved in ethyl acetate from the previous extraction step andthe desired separation cut-off is in the range of hundreds of Daltons,organic solvent nanofiltration appears as a tailor-made approach for thefractionation and is studied within this thesis.

6.1 Organic solvent nanofiltration

Organic solvent nanofiltration (OSN) is an emerging field in membraneseparation that opened up novel opportunities in low-energy fractiona-tion and purification of polydisperse mixtures during the last decade. Es-pecially in the recent years, it has gained much attention as a promisingseparation technique in the chemical and pharmaceutical industry [108].The development of membranes that are stable in organic solvents trig-gered intensified research in membrane fabrication, membrane character-ization and application of OSN. Various polymeric OSN membranes witha cut-off ranging from 150 g mol−1 to 1000 g mol−1 which can be used innumerous organic solvents have emerged and been improved throughoutthe last decade. Most of these membranes have an asymmetric structurewith a thin top layer on a porous support, either of the integral asym-metric type or the composite type [109]. Several potential applicationsof OSN have been proposed in the literature, including applications inthe petrochemical industry, homogeneous catalysis and in the food in-dustry [64, 110–112]. Among those, only few processes have already beenrealized on industrial scale, out of which ExxonMobils’s MAX-DEWAXprocess for the dewaxing of lube oil is the largest and best-known one[113]. Within the present thesis, five different commercial OSN mem-branes were studied for the separation of monomeric reaction products

6.2 Characterization of OSN membranes 107

from the mixture of lignin oxidation products: three integral asymmetricmembranes (DuraMemTM 500, DuraMemTM 900 and PuraMemTM 280)made of crosslinked polyimide (PI), one thin-film composite membranemade of silicone (SIL)-coated polyimide (PuraMemTM S380) and onecomposite membrane which is composed of a separating layer of poly-dimethylsiloxane (PDMS) on a polyacrylonitrile (PAN) support (SelRo R©

MPF-44) [64].In a first step, the membranes were characterized concerning their

performance parameters permeability and rejection using pure solventand model compounds dissolved therein. Ethyl acetate was studied asorganic solvent since it represents the most promising extracting agentfor lignin products before the OSN separation step. The rejection wasmeasured for the known monomeric products vanillin and methyl vanil-late as well as for the dimers benzyl phenyl ether and 2-benzylphenol(compare Figure 4.7), which are commonly used as dimeric lignin modelcompounds [3]. In a second step, the fractionation of lignin oxidationproducts was investigated. The performance of the membranes was mon-itored by flux measurements, size-exclusion chromatography (SEC) andgas chromatography/mass spectrometry (GC/MS). Based thereupon, re-jection values were calculated and compared along with the flux mea-surements to the membrane characterization experiments. These datawas finally used for the selection of an appropriate membrane for thedesired fractionation of the lignin oxidation products.

6.2 Characterization of OSN membranes

6.2.1 Pure solvent flux

In the beginning of the membrane investigations, the permeability ofeach membrane for ethyl acetate was experimentally studied. Two mem-brane disks of each membrane were run in the membrane laboratoryplant with pure ethyl acetate as solvent until stable permeate flux wasobtained. A nitrogen pressure of 20bar was applied as the operation ofthe DuraMemTM membranes was limited to this value. The solvent fluxJs was calculated with the membrane area A

Js =∆V

A ·∆t(6.1)


The flux Ji of a component i through a membrane is driven by itschemical potential gradient dµi/dx and can generally be described withthe proportionality parameter Li

Ji = −Lidµi

dx(6.2)

For OSN membranes the solution-diffusion model can be applied to de-scribe the gradient in chemical potential in the membrane [114]. Thisleads to the general transport term for the flux of a component i [115]

Ji =DiSivi

l

[

ci,r − ci,p exp

(

−vi(pr − pp)

RT

)]

(6.3)

Eq. (6.3) can be simplified for the permeation of pure solvent as derivedstep by step in the aforementioned article to

Js =DsSscs,rv

2s

lRT∆p = As∆p (6.4)

The solvent permeability constant As for the ethyl acetate flux acrossthe different membranes is determined from the experimental results inthe following.

Among the studied membranes, an initial flux decline resulting frommembrane compaction (as discussed in Chapter 3, see [65]) was observedfor DuraMemTM 500, DuraMemTM 900 and SelRo R© MPF-44. For thesemembranes it took 3h, 10h and 15h, respectively, until stable fluxesand thus a final compaction of the membranes were achieved. A similarflux behavior was observed in literature for a different membrane of theDuraMemTM type by Sereewatthanawut et al. and for SelRo R© MPF-44by Yang et al. who both also reported an initial flux decline attributedto membrane compaction [116, 117]. In contrast to those membranes,the ones of the PuraMemTM series did not show a decline but virtu-ally constant fluxes from the very beginning of their use. The differentflux behaviors of the two investigated DuraMemTM membranes and ofPuraMemTM 280 are depicted in Figure 6.1.

While the initial flux decline of the two disks of the same mem-brane was quite similar for DuraMemTM 500 and DuraMemTM 900(35.3%/38.8% and 29.4%/33.0%), the two SelRo R© MPF-44 disks be-haved quite differently with flux declines of 16.6% and 59.5%. A similar


Run time [h]

0 2 4 6 8 10 12 14 16

Pe

rme

ate

flu

x [

L m

-2 h

-1]

0

500

2000

2500

3000

3500DuraMem

TM 500 (1)

DuraMemTM

500 (2)

DuraMemTM

900 (1)

DuraMemTM

900 (2)

PuraMemTM

280 (1)

PuraMemTM

280 (2)

Figure 6.1: Fluxes of ethyl acetate at 20bar through different mem-branes with and without initial flux decline addressed tomembrane compaction.

trend was found for the difference in final flux values between the twotested membrane disks. For the DuraMemTM and PuraMemTM mem-branes the flux differences with respect to the absolute values were quitesmall with coefficients of variation below 10%. For SelRo R© MPF-44, onthe other hand, it reached 33.8%. This confirms the results of Yang etal. who also found large discrepancies for ethyl acetate fluxes throughMPF-44 with values in the same range as in our study [117].

The dependency of flux on pressure was measured at four differentpressures at the end of the characterization experiment. All membranesshowed linear pressure dependency corresponding to Eq. (6.4) with coef-ficients of determination (R2) of above 0.97. The permeability constantsAs for constant flux are listed together with the coefficients of variation(CoV) and the constant permeate fluxes in Table 6.1. The constant per-


meate flux Js represents the average value of the two membrane disksafter constant flux behavior was observed including the maximum devi-ation above and below the average.

Table 6.1: Experimental parameters for the permeation of ethyl acetateacross two membrane disks of each type at 20bar pressure(DM=DuraMemTM, PM=PuraMemTM).

Membrane Disk Permeate flux Js Permeability As CoV# [L m−2 h−1]+max

−max [L m−2 h−1 bar−1] [%]

DM 500 1 268.4 +7.0−7.6 14.24

1.75(500Da) 2 271.4 +13.7

−9.0 14.81

DM 900 1 2017.3 +56.5−39.3 96.22

6.38(900Da) 2 2208.0 +47.6

−65.1 105.27

PM 280 1 131.4 +8.3−5.9 6.93

3.16(280Da) 2 137.4 +4.8

−6.8 7.15

PM S380 1 69.2 +4.2−4.6 3.42

6.96(600Da) 2 62.7 +5.0

−5.6 3.19

MPF-44 1 2.24 +0.16−0.09 0.11

33.83(250Da) 2 3.66 +0.32

−0.15 0.18

The by far highest fluxes among the investigated membranes were ob-tained with the DuraMemTM membranes. Especially DuraMemTM 900outperforms all other membranes by a factor of about seven. This isin accordance with the information given by the manufacturer who rec-ommends DuraMemTM membranes for polar aprotic solvents like ethylacetate as well as polar protic solvents. PuraMemTM membranes, on theother hand, are recommended for apolar solvents. In addition, the per-meability of ethyl acetate across the DuraMemTM membranes seems tobe very good compared to other solvents discussed in the literature. Pe-shev et al. found ethanol permeabilities of 1.07 and 0.78L m−2 h−1 bar−1

for DuraMemTM 500 compared to 14.24 and 14.81L m−2 h−1 bar−1 forethyl acetate within this work [118].


For the PuraMemTM membranes the flux across the 280 membraneis higher than for the S380 although its MWCO (280Da vs. 600Da) islower. This is in contrast to the manufacturer’s data for toluene in whichfluxes of 80L m−2 h−1 and 160L m−2 h−1 were reported, respectively.The permeability values of ethyl acetate and toluene are, however, inthe same range. Among all studied membranes, the SelRo R© MPF-44had by far the lowest flux, which was lower by a factor of 40 comparedto PuraMemTM 280, although their cut-off is comparable. This mightbe due to the fact that these membranes were originally designed for theapplication of water/solvent mixtures and are therefore stable in organicsolvents but not intended for the use of pure solvents. Hence, thesemembranes rather promote the permeation of polar protic solvents likewater or methanol than that of ethyl acetate, which is in accordancewith the experimental data of Yang et al. [117].

6.2.2 Rejection of model compounds

To study the rejection of the membranes against aromatic monomers anddimers in lignin, a solution containing four model compounds was usedin the membrane plant. The rejection was determined for the monomericproducts vanillin and methyl vanillate as well as for the dimeric modelcompounds 2-benzylphenol and benzyl phenyl ether based on the generaldefinition of the rejection

Rj =

(

1−cj,pcj,r

)

× 100% (6.5)

By using the average GC/MS area ratios of the four solutes j with theinternal standard, the rejection values can be calculated

Rj =

(

1−(Aj/AiS)p

(Aj/AiS)r

)

× 100% (6.6)

The description of the solute transport across the membranes is ingeneral also derived from Eq. (6.3). It can be simplified for the solutesby neglecting the minor term −vi(pr − pp) which leads to

Jj =DjSjvj

l(cj,r − cj,p) = Bj∆cj (6.7)


The constant Bj represents the permeability of solute j and correspondsto the permeability constant As for the solvents. Since, according toEqs. (6.4) and (6.7), the transmembrane pressure ∆p drives the fluxof solvent but its influence on the solute flux is negligible, the soluterejection depends on solvent flux and thus on pressure. The rejectionvalues determined in the following are measured at a transmembranepressure of 20bar and only valid for this pressure.

Based on three samples which were drawn for each disk during therejection experiments and three GC/MS injections per sample, the valuespresented in Table 6.2 were found. The given error values represent the68% confidence interval of the nine injections. The rejection values ofthe two membrane disks of the same membrane are comparable in allcases and within the calculated error.

Table 6.2: Rejection values of the model compounds vanillin (V),methyl vanillate (MV), 2-Benzylphenol (BP) and benzylphenyl ether (BPE) for the two membrane disks of each typeduring ethyl acetate flux at 20bar pressure.

Membrane Disk # RV [%] RMV [%] RBP [%] RBPE [%]

DM 500 1 13.8±4.6 16.8±5.0 13.5±5.5 15.5±3.9

(500Da) 2 17.9±3.2 18.2±4.7 10.4±6.0 14.6±4.1

DM 900 1 -0.1±1.5 0.0±2.1 1.2±2.7 0.4±1.7

(900Da) 2 -1.1±1.9 -1.0±2.0 -0.8±2.0 -0.8±1.5

PM 280 1 24.7±4.1 42.3±4.6 16.2±3.3 46.7±4.3

(280Da) 2 25.8±4.5 41.5±4.5 14.3±3.0 45.6±4.4

PM S380 1 26.2±2.8 26.5±3.0 10.7±3.4 25.5±2.6

(600Da) 2 31.7±3.2 31.1±3.3 13.8±3.2 28.9±3.1

MPF-44 1 2.5±3.1 2.2±2.8 1.4±3.0 1.3±1.9

(250Da) 2 -1.9±4.1 -0.8±3.7 0.6±3.3 1.1±2.8

Among the studied membranes, DuraMemTM 900 and SelRo R© MPF-44 show virtually no rejection against the four investigated model sub-


stances. While this is in agreement with the MWCO of DuraMemTM

900, one would expect higher rejection values for SelRo R© MPF-44 asits MWCO of 250Da is close to the molecular weight of the modelsubstances. The other three membranes show a considerable rejectionwhereas the highest rejection values were obtained with PuraMemTM

280 which also has the lowest MWCO among the membranes suppliedby Evonik MET. However, when comparing the values of the four modelsubstances for any of the membranes, it becomes apparent that the re-jection against all model substances is roughly the same. No selectivepermeation of monomeric compounds indicated by significant differencesbetween monomeric and dimeric compounds can be found. In fact, 2-benzylphenol passes most of the membranes with the least rejection,although it possesses the highest molecular mass and is a dimeric ligninmodel compound composed of two aromatic rings.

The observations described above underline the fact that flux and sep-aration properties of organic solvent nanofiltration membranes heavilydepend on solvent-membrane interaction as described by White [114].Mulder et al. also found that the MWCO should not be considered asan absolute value in nanofiltration since additional effects such as solu-bility become predominant [119]. For this reason, molecules of similarmolecular mass can experience totally different rejections by a mem-brane. Especially swelling of the membrane in the solvent has a largeimpact on the separation performance and thus on the MWCO as foundin our study.

Based on the results from the characterization experiments no ob-vious preselection of a suitable membrane for the desired separationof monomeric lignin products is possible. The monomeric model com-pounds permeate successfully across all of the studied membranes withrejection values (as in Table 6.2) of below 50% but no significant differ-ences in the permeation of monomeric and dimeric model substances wasfound for any membrane. The permeability values, on the other hand,differ significantly among the studied membranes. As a good perme-ability of ethyl acetate is also required for the separation step, SelRo R©

MPF-44 was not further considered in the following lignin separationstudy.


6.3 Separation of the product mixture

6.3.1 Permeate flux

The most obvious difference between experiments with pure solvent andthe lignin product solution was the difference in permeate flux. Just af-ter starting the separation experiment, the fluxes through all the studiedmembranes reached constant values which were considerably lower com-pared to the characterization experiments (compare Table 6.3). Espe-cially for DuraMemTM 500 and DuraMemTM 900 the difference was im-mense with factors of 49 and 233 between the experiments, respectively.In fact, it turned out that in contrast to the previously measured perme-abilities As, the ethyl acetate permeation across the membranes of theDuraMemTM type is even worse during separation than for PuraMemTM

S380 and PuraMemTM 280. For the latter ones, the flux difference wasless pronounced with factors of 1.6 and 5.7, respectively.

The reason for the significantly lower permeate flux during the sep-aration of lignin oxidation products is mainly membrane fouling. Es-pecially the DuraMemTM membranes, which are almost in the ultra-filtration range, have a rather porous structure that is to some extentclogged by the variety of molecular species in the product mixture. ThePuraMemTM membranes, on the other hand, are less porous and prod-uct separation is generally based on the solution-diffusion mechanism.Therefore, the performance of those membranes is less affected by mem-brane fouling.

6.3.2 Monomer rejection

The permeation of the monomeric reaction products was monitored dur-ing the separation experiments by GC/MS. A similar analytical proce-dure as for the rejection measurements of the model components wasused for the determination of monomer rejection. In order to calculatethe rejection values, the relative amount of vanillin and methyl vanillateafter each membrane separation batch n related to the initial amountmust be known. Based on the GC/MS analysis, the relative amountcnj /c

0j can be determined by

6.3 Separation of the product mixture 115

cnjc0j

=(Aj/AiS)

nr

(Aj/AiS)0r(6.8)

When knowing the relative amount of one component j versus time, therejection Rj can be calculated based on a mass balance of the plant.Within one batch, the change in concentration can be derived as

lncnj,endcnj,0

= Rj · lnV0

Vend(6.9)

Considering the refill of the plant after each batch

cn+1j,0

cnj,end=

Vend

V0(6.10)

the decrease in concentration of one component j with increasing batchnumber n can be calculated for a constant volume reduction factorV RF = V0/Vend by

cnjc0j

= exp(Rj − 1) · ln (V RF ) · n (6.11)

The batch number can be replaced by the total volume of permeateVp,total for a constant permeate volume per batch Vp,batch, so that thedecrease in concentration with permeate volume can be described by

cjc0j

= exp(Rj − 1) · ln (V RF ) ·Vp,total

Vp,batch(6.12)

The rejection values Rj were determined by numerically fitting thisexponential equation to the relative amount of the monomeric productsj using a Levenberg-Marquardt algorithm. The obtained rejections forvanillin and methyl vanillate are compared to the values from the previ-ous characterization experiments for all membranes in Table 6.3.

As for the permeability before, the rejection values obtained duringcharacterization and separation differ significantly. The only membranewhich gave similar permeability and rejection values during both experi-ments was PuraMemTM S380. For all other membranes large differences


Table 6.3: Permeability and rejection of the membranes during separa-tion and characterization.

Membrane Experiment As[L m−2 h−1 bar−1] RV [%] RMV [%]

DM 500 (Sep.) 0.29 37.7 61.4(500Da) (Char.) 14.53 15.9 17.5

DM 900 (Sep.) 0.43 -4.8 26.2(900Da) (Char.) 100.75 -0.6 -0.5

PM 280 (Sep.) 1.24 -5.3 19.6(280Da) (Char.) 7.04 25.2 41.9

PM S380 (Sep.) 2.06 31.4 32.8(600Da) (Char.) 3.30 29.0 28.8

were observed without any clear trend. Solely the rejection of methylvanillate is always remarkably higher than the rejection of vanillin whilethey were rather similar during characterization. Since rejection is di-rectly linked to flux (see Eqs. (6.4) and (6.7)), the differences within therejection values in general seem to be caused by the huge differences inpermeability attributed to membrane fouling as discussed above.

The general rejection against the different monomeric products dur-ing the separation experiment is illustrated in Figure 6.2. The plotdepicts the relative amount of each monomer versus permeated volumefor PuraMemTM 280, for which the most significant differences in thepermeation of the monomeric products was observed.

The rejection values of the different monomers cover a wide range.While vanillin can pass the membrane unhindered, methyl dehydroa-bietate is largely rejected by the membrane. As a matter of fact, therejection values are generally in the same order as the molecular weightand the size of the molecule, respectively. Vanillin has both the low-est molecular weight and the lowest rejection, for methyl vanillate and5-carbomethoxy-vanillin they are both in the same range, and methyldehydroabietate as the by far largest molecule experiences also the high-est rejection. Due to these differences in rejection against the different


Volume permeated [L]

0.0 0.5 1.0 1.5 2.0 2.5 3.0

Re

lative

am

ount

cj/c

j0 [

-]

0.0

0.2

0.4

0.6

0.8

1.0Vanillin

Methyl vanillate




Numerical fit

RMV = 19.6%RV = -5.3%

RM5CV= 48.4%

R5CV = 24.4%

RMDHA = 85.1%

Figure 6.2: Separation of monomeric products from the reaction mix-ture with PuraMemTM 280 and numerical fit.

monomers, a membrane like PuraMemTM 280 is also capable of separat-ing certain monomers from the mixture within a multi-stage membraneseparation unit. Hence, the membrane could optionally be used in thedownstream processing of the monomer fraction in a lignin process.

6.3.3 Separation Performance

In order to evaluate the overall separation performance of the mem-branes, retentate samples were drawn before each batch run and analyzedby SEC. Rejection values for different groups of products (monomers,dimers,...) were calculated by their relative amount in SEC after eachbatch with respect to the initial amount as it was done for the monomericproducts by GC/MS before. Sample preparation was done by evaporat-ing the solvent of 1mL of the sample and redissolving it in 2mL ofeluent (0.01N NaOH solution). Due to the overlapping of product peaksin the SEC chromatograms, the peak height hk was used for the calcu-


lation of the relative amount cnk/c0k of monomers (k = 1, ∼160 g mol−1),

dimers (k = 2, ∼320 g mol−1), trimers (k = 3, ∼480 g mol−1) and of theoligomeric peak (k = 4, ∼750 g mol−1) for all batches n

cnkc0k

=hnk,r

h0k,r

(6.13)

The relative amount of each group of products can finally be used todetermine their rejection values as already derived in Eq. (6.12) by

ckc0k

= exp(Rk − 1) · ln (V RF ) ·Vp,total

Vp,batch(6.14)

By SEC, the separation performance can be monitored for the wholerange of molecular weight as depicted in Figure 6.3. The graph shows

Elution time [min]

29 31 33 35 3730 32 34 36

AU

[-]

0.00

0.02

0.04

0.06

0.08

0.10

0.12

Molar mass [g mol-1

]

14892 3215 934 446 292 219 156 89 34

Vp

Vp

Vp

MonomersDimers

Trimers

Figure 6.3: SEC chromatograms showing the decline of lignin oxidationproducts in the retentate with increasing permeate volumefor PuraMemTM S380.


the decline of products in the retentate with increasing permeate vol-ume. From the SEC chromatograms it becomes apparent that not onlylow molecular weight products are able to pass the membrane but prod-ucts of all molecular masses. This was observed for all membranes, nomembrane was able to retain the high molecular weight products com-pletely.

The permeation of the different groups of products can even betterbe illustrated by plotting the relative peak height of the four apparentmaxima versus the permeated volume as in Figure 6.4. The depicteddata points result from a single SEC analysis as the error of absolutepeak height in SEC turned out to be negligible (below 2.2%). In or-der to determine the rejection values for one group, the data points ofeach curve were numerically fitted according to Equation (6.14) using aLevenberg-Marquardt algorithm as described above.

Volume permeated [L]

0.0 0.5 1.0 1.5 2.0 2.5 3.0

Re

lative

am

ount

ck/c

k0 [

-]

0.0

0.2

0.4

0.6

0.8

1.0Monomers

Dimers

Trimers

Oligomers

Numerical Fit

RMon= 38.4%

RDi= 83.2%

RTri= 93.2%

ROlig= 94.6%

Figure 6.4: Decline of different product groups with increasing perme-ate volume and numerical fit for PuraMemTM S380.

As expected, the monomeric molecules permeate the most easily, fol-lowed by dimers and trimers so that the obtained rejection values in-


crease with the size of the molecules. After permeation of approximately2L of solvent, the monomeric lignin oxidation products have passedthrough the membrane leaving only products of higher molecular weightin the retentate. The performance of the other studied membranes isin general very similar, the obtained rejection values are summarized inTable 6.4.

Table 6.4: Rejection values for the different groups of products ob-tained by numerical fit to SEC data.

RMon RDi RTri ROlig

Membrane [%] [%] [%] [%]

DuraMemTM 500 15.7 77.5 88.8 90.2DuraMemTM 900 23.2 80.9 91.8 93.1PuraMemTM 280 8.0 77.3 86.6 87.4PuraMemTM S380 38.4 83.2 93.2 94.6

All four studied membranes turn out to be very effective in the sepa-ration of monomeric reaction products from the reaction mixture. Whilethe monomer rejection values are below 39% for all membranes, the re-jection of the other groups of products is always above 77%, therebyallowing for an efficient separation of monomers. However, when com-paring the rejection values among the different membranes, it turns outthat the obtained values are quite similar. No membrane yields the low-est overall rejection for monomers and the highest overall rejection forthe other lignin oxidation products at the same time. Therefore, it re-mains unobvious which of the membranes performs the best based onthe rejection values alone. Nevertheless, the permeate flux as includedin Table 6.3 is also an important criterion when it comes to the separa-tion performance. Out of both, rejection values and flux, the membranewith the best separation performance can be found. For this purpose,the time to separate 98% of the monomers and the amount of remainingother products at this time are compared in Table 6.5.

According to the table, especially the two PuraMemTM membranes of-fer a relatively fast separation of monomeric products from the rest of the

6.4 Summary of the membrane separation 121

Table 6.5: Time for permeation of 98% of monomeric products in thelaboratory plant and remaining amount of other products.

t Dimers Trimers OligomersMembrane [h] [%] [%] [%]

DuraMemTM 500 35.5 35.3 59.3 63.6DuraMemTM 900 26.3 37.8 65.7 70.4PuraMemTM 280 7.6 38.1 56.6 58.5PuraMemTM S380 6.8 34.3 65.0 71.1

product mixture. However, it takes a little longer for PuraMemTM 280to separate the monomers with a higher permeation of other products atthe same time. Based on these criteria, PuraMemTM S380 finally turnedout to be the best membrane for the separation of monomeric lignin ox-idation products from the remaining product mixture concerning bothmembrane flux and rejection of other components.

6.4 Summary of the membrane separation

Five solvent stable nanofiltration membranes were evaluated within thischapter concerning their performance in separating the monomeric ligninoxidation products from the remaining product mixture. In a first mem-brane characterization step, a high permeability of ethyl acetate wasfound for the membranes of the DuraMemTM series after an initial mem-brane compaction period. The two PuraMemTM membranes did notshow any compaction but far lower permeability values. Because of thevery low permeate fluxes obtained for SelRo R© MPF-44, this membranewas not further considered for the following selection of an appropriatemembrane for lignin product separation. Measurement of rejection val-ues revealed a very good permeation of the studied monomeric productsfor all membranes. As the difference in rejection between monomericand dimeric model compounds was not significant for any membrane, nofurther membrane preselection was possible at this point.


In the subsequently performed lignin separation study, the permeationof different groups of oxidation products was monitored. Rejection valuesagainst these product groups were fitted to the experimental results.All four studied membranes showed a very efficient separation with lowrejection against monomers and high rejection against the products ofhigher molecular weight including dimers and trimers. Interestingly, thesolvent fluxes within the experiments differed significantly among themembranes and were in contrast to the permeation of pure solvent before.In the same manner, rejection values of vanillin and methyl vanillatediffered between the separation and the characterization experiments. Incase of a sufficient difference in rejection against the different monomericproducts (as found for PuraMemTM 280), the membrane can as wellbe employed in the separation of a certain monomer from the mixture,qualifying it as a potential method in the purification of the monomericfraction.

In general, the desired separation of monomers from the product mix-ture can effectively be achieved with all of the four membranes of theDuraMemTM and PuraMemTM type. However, in search of the bestmembrane, evaluation of the time for separating monomeric productsbased on permeate flux as well as the amount of high molecular weightproducts remaining in the retentate tipped the balance for PuraMemTM

S380 which gave the most satisfying results in this regard.

123

Chapter 7

Downstream processing

After separation of the product mixture by organic solvent nanofiltra-tion, the two resulting process streams need further processing to yieldboth single aromatic products in sufficient purity and a high molecularweight fraction that can be used in dispersants, resins, adhesives or an-tioxidants. The mixture of aromatic monomers requires separation ofthe compounds and purification steps to remove impurities. In contrast,the other fraction does not necessarily demand for a separation but mayrequire a chemical modification to improve its performance in the desiredproducts. These aspects are discussed in the following.

7.1 Separation and purification ofmonomeric products

The isolation of the monomeric compounds and their purification repre-sent a crucial part of the process in terms of its economy. Based on thefirst existing lignin-related processes which were realized on an indus-trial scale, literature in the 1950s rated the recovery of pure vanillin ingood yield as "extremely difficult" with "great expenditure in appara-tus, chemicals, manpower, etc. without attaining the quality of naturalvanillin or that produced by fully synthetic means" [120]. However,due to improvements in vanillin separation and purification during thelast decades, lignin-related processes have become economically morecompetitive. Various approaches for the recovery of vanillin and otheraromatic products, which were partly realized in commercial processes,are discussed in the literature and described below.

124 7. Downstream processing

7.1.1 Review on existing approaches

The invention of the process which is able to convert waste sulfite liquorsinto vanillin by alkaline oxidation in the 1930s initiated intensified re-search on separation and purification methods to obtain pure vanillinfrom the product stream. Several different approaches that are capa-ble of separating vanillin and purifying it to food grade were patentedthroughout the decades. As this task involves several process steps andtherefore affects the overall economy of the process, most inventionsaimed at modifying existing techniques or introducing novel methods inthe way that they require less process steps or consume lower amountsof chemicals and energy.

The original version of the process for the production of purifiedvanillin as patented by Sandborn et al. applied a chemical derivati-zation to separate vanillin from all other reaction products after thereaction step [121]. The ability of aldehydes to form water soluble bisul-fite adducts opens up the opportunity to derivatize vanillin with sulfurdioxide in alkaline solution and subsequently precipitate all products ex-cept the vanillin bisulfite complex by acidification. The bisulfite complexcan be transformed back to vanillin by further addition of a mineral acidand liberates the sulfur dioxide for recycling. After extraction of vanillinwith benzene, several fractions of vanillin with different purities are ob-tained by washing the benzene solution with dilute sodium hydroxide inmultiple stages of scrubbing units.

The isolation and purification of vanillin from the crude liquors inthe original version was modified by the patents of Sandborn and Servis[122, 123]. The extraction of sodium vanillate with butanol and re-extraction into an alkaline solution leaves large amounts of unconvertedlignosulfonates behind and thereby eliminates the drawback of high acidrequirements for neutralization as well as extraction problems that wereencountered in the benzene extraction step.

Inventions during the following decades tried to lower the complex-ity of the above-mentioned processing in terms of chemical need andthe number of process steps. Vacuum distillation, as disclosed in thepatent of Bryan, came up as one of them [124]. While vanillin normallyrequires high vacuum as it suffers from decomposition at temperaturesof above 165 C, the addition of an oil as distillation aid renders vac-

7.1 Separation and purification of monomeric products 125

uum distillation at higher pressures and lower temperatures (10mbarand <160 C) feasible. A similar approach was pursued by Töppel whorecommended fractional steam distillation of vanillin from solution in anadequate solvent. Apart from that, multi-stage crystallization is knownto be essential for obtaining food grade vanillin as disclosed in the patentof Schöffel [125].

The processes that were commercially operating on industrial scaleand produced large amounts of the worldwide vanillin demand, appliedtechniques that are equal to or at least related to the aforementioned in-ventions. A process scheme of the formerly existing process, as describedby Faith in the middle of the 1960s, is illustrated in Figure 7.1 [7].

Accordingly, this embodiment of the process applies extraction andre-extraction as a first step after flash cooling of the reactor effluent.Vanillin and the ligneous side products in the aqueous phase after re-extraction are then subjected to sulfur dioxide treatment in order totransform vanillin in its water-soluble bisulfite complex. Thereby, theside products can be precipitated by acidification leaving only vanillinand some impurities behind. After decomposition of the vanillin bisulfitecomplex with sulfuric acid, vanillin is purified by vacuum crystallization,centrifugation and vacuum tray-drying. As mentioned in the same lit-erature, other commercially existing embodiments of the process at thistime also employed vacuum distillation to separate vanillin from the con-taminating by-products.

After the heyday of commercial plants for vanillin production fromwaste sulfite liquors, more contemporary process steps were introducedto facilitate the known methods of vanillin separation and purification.Several patents were filed that describe the benefit of supercritical ex-traction with carbon dioxide either of the extracted crude vanillin in solidform or direct extraction from the reactor effluent [97, 100]. Moreover,the use of cationic ion exchange resins for vanillin recovery emerged inthe literature and is still today discussed as a potential process step forvanillin production from lignin-related sources [126, 127].

While the previous approaches focused on the recovery of vanillin as asingle product, one article in literature also deals with the separation of amonomeric product mixture. Vigneault et al. propose a strategy on howto separate a mixture of twelve monomers that were obtained by base-catalyzed depolymerization of steam exploded aspen lignin [103]. Not


Figure 7.1: Process scheme of a formerly existing commercial processfor vanillin production from waste sulfite liquors includingseparation and purification stages for vanillin [7].

7.1 Separation and purification of monomeric products 127

surprisingly, they found that the separation of the phenolic compoundsin a single step by conventional methods is impractical. Instead, theyrecommend a combination of vacuum distillation, flash liquid chromatog-raphy and crystallization. Vacuum distillation is implemented with fourdistillation columns (p = 0.13mbar, T → 160 C) as pre-fractionationinto five streams. While two components can already be isolated by thismethod, the other ones require further separation by crystallization withor without a preceding liquid chromatography step. Among the proposedprocess steps, solely vacuum distillation and liquid flash chromatographywere demonstrated on a laboratory scale.

7.1.2 Conclusions for the present process

Some of the discussed processes or at least combinations thereof havebeen proven to be capable of yielding vanillin in food grade purity. Inprinciple, a sequence of the same process steps can be used in the con-text of the present thesis to produce highly pure vanillin. However, thedownstream processing of the reaction products becomes even more so-phisticated when intending to produce more than one of the aromaticmonomers, as seen in the work of Vigneault et al. Whether it makessense to separate not only vanillin but also other monomeric reactionproducts solely depends on economical considerations. As no signifi-cant commercial markets exist for the products except vanillin, furtherinvestigations on their market potentials or promising chemical transfor-mations have to be done before this question can be answered. This isdiscussed in more detail in the next chapter.

To separate the five main monomeric products present in the lowmolecular weight fraction, a combination of methods either reported inthis thesis or described above is necessary. As shown in the previouschapters, both supercritical extraction and membrane separation can beused to isolate certain products when conducted in multiple stages owingto the sufficiently different solubility and rejection, respectively. Sepa-ration of the two aldehydes from the mixture is possible by the bisulfitemethod as described above, assuming that 5-carbomethoxy-vanillin isalso susceptible of forming bisulfite complexes like vanillin. Evaluatingthe potential qualification of distillation or crystallization to separatethe products is even more vague since solubilities and boiling points for


all five components are neither reported in vacuum nor are they exactlyknown at atmospheric pressure. While the boiling points at atmosphericpressure are comparable between vanillin (285 C) and methyl vanillate(285 C to 287 C), methyl dehydroabietate (390 C) evaporates at highertemperatures. In case of 5-carbomethoxy-vanillin, the boiling point isnot reported in literature but at least its melting point (135 C) differssignificantly from vanillin (82 C) and methyl vanillate (64 C to 67 C),rendering e.g. a crystallization step possible.

By employing selected of these methods in an appropriate sequence,complete separation of the monomeric products appears feasible. A po-tential separation strategy is presented in Figure 7.2.

Vacuum

Distillation

CrystallizationVanillin,Methyl

Vanillate

Monomers

(Ethyl Acetate)

Monomers

Ethyl Acetate

Solvent

Recovery scCO2

Extraction

Methyl Dehydroabietate

Vanillin

Methyl Vanillate

Crystallization

5-Carbo-

methoxy-Vanillin

Methyl 5-Carbo-

methoxy-Vanillate

Other

Monomers

Ethyl Acetate

Recycling

Figure 7.2: Possible separation strategy to isolate the five mainmonomeric products from the permeate stream obtainedduring organic solvent nanofiltration

After evaporation and recycling of the organic solvent, fractional vac-uum distillation has been demonstrated for the separation of vanillin inthe literature. As methyl vanillate has a comparable boiling point atatmospheric pressure, it is assumed to evaporate in the same fraction asvanillin under vacuum. All other products are expected to possess highervacuum boiling points and to remain in the bottom fraction due to theirboiling and melting points at atmospheric pressure. Vanillin and methyl

7.2 Processing of high Mw products 129

vanillate can then either be separated by crystallization owing to a suf-ficient difference in melting points or by flash column chromatography.The other products first undergo a supercritical extraction step withcarbon dioxide to selectively remove methyl dehydroabietate from thisfraction. By placing this process step behind distillation, lower amountsof impurities are co-extracted. The separation of the two remainingcompounds should as well be feasible by a thermal separation step likecrystallization or by liquid chromatography. Nevertheless, several othersequences and replacement of unit operations by other ones as discussedabove appear also reasonable.

Finally, purification efforts for each of the compounds, which are notincluded in the scheme, are not to be underestimated. Hence, a detailedevaluation of the cost-effectiveness for the isolation and purification ofeach monomeric product is mandatory. Apart from a complete isola-tion of each component, the combined chemical transformation of theremaining products into other products (e.g. benzene and phenol bydefunctionalization) could as well be of commercial interest.

7.2 Processing of high Mw products

As previously discussed, several potential applications are considered forthe fraction of high molecular weight lignin oxidation products. Besidesthe desired use as water reducer in concrete, the same applications asdiscussed for lignin itself (e.g. resins, adhesives or antioxidants) arepossible, either with or without prior functionalization or modificationstep. To evaluate the potential applications for this fraction at the endof this section, it is first analyzed concerning its chemical properties inthe following.

7.2.1 Characterization of the fraction

The high molecular weight oligomeric product fraction that results fromthe organic solvent nanofiltration step was analyzed along with IndulinAT according to the analytical procedures described in Chapter 3. Allanalytical methods were validated with literature results for Indulin AT.The analyzed properties of both fractions are listed in Table 7.1.


Table 7.1: Chemical properties of Indulin AT and the high molecularweight fraction which is obtained as retentate from organicsolvent nanofiltration.

Kraft lignin High Mw(Indulin AT) fraction

Mw

(SEC)3200±300 1125±125 g mol−1

16-19 5-7 phenolic C9 units

Phenolic OH(UV)

2.39 1.10 mmol g−1

0.44 0.20 per phenolic C9 unit7.64 1.24 per chain

Aliphatic OH(31P-NMR)

2.08 1.49 mmol g−1


Free carboxyl(31P-NMR)

0.44 0.50 mmol g−1


Total carboxyl(Titration)

0.42 2.30 mmol g−1


Carbonyl(FTIR)

1.27 2.57* mmol g−1


Methoxyl(13C-NMR)

3.85 0.32 mmol g−1


Reactive 5-pos.(Methylolation)

3.80 1.35 mmol g−1


* determined for the complete product mixture before nanofiltration


As known from results of the oxidation and the membrane separa-tion step, the obtained oligomeric product fraction possesses a far lowerweight average molecular weight than the original lignin in the feed-stock. While values of 3200±300 g mol−1 are generally analyzed by SECfor Indulin AT, the average molecular weight is decreased during the re-action to values of 650±100 g mol−1 and increases during liquid-liquid-extraction and organic solvent nanofiltration to values in the range of1125±125 g mol−1, depending on the process parameters.

Regarding the functional group content, a significant increase in es-terified carboxyl groups (difference of total and free carboxyl groups) ismeasured. As expected and also found in the monomeric products before,the oxidation introduces carboxyl groups which are readily esterified inthe presence of methanol. The amount of free carboxyl groups is com-parable between Indulin AT and the fraction of high molecular weightreaction products. In addition to the carboxyl groups, the content ofthe carbonyl groups is also considerably increased. Although the ana-lyzed quantities are questionable since quantitative FTIR spectroscopywas so far only used for lignin itself, the increase in carbonyl groupsduring the oxidation appears reasonable when regarding the monomericreaction products. Another functional group whose content decisivelychanges is the methoxyl group. Its quantity in the product fraction isfar lower than in Indulin AT. This is less evident when considering theunchanged methoxyl group in the monomeric products, despite the factthat they have been separated before. The content of both, aliphaticand phenolic hydroxyl groups, is lowered as well. The reactive aliphatichydroxyl groups can be assumed to undergo oxidation into carbonyl andcarboxyl groups. The content of unreactive phenolic hydroxyl groups,on the other hand, has been found to not change during the reactionbut upon separation and fractionation. Analysis revealed a higher phe-nolic hydroxyl content in a sample of the permeate fraction than in thereaction mixture, thus explaining the low phenolic hydroxyl content inthe oligomeric retentate fraction. The analysis of reactive 5-positions onthe aromatic ring as well reveals a lower number for the fraction of highmolecular weight products.


7.2.2 Potential applications

Potential applications for the fraction of oligomeric lignin oxidation prod-ucts should take advantage of their chemical properties in comparisonwith Indulin AT. More precisely, this fraction can be expected to besuperior to kraft lignin when the desired application relies on chemicalfunctionalities that are present in higher amounts. In this case, it is ofno matter whether the application benefits from the functionality itselfor from the possibility to chemically modify it. However, as possiblechemical modifications would exceed the scope of the thesis, only directapplications without further modification are discussed in the following.

Out of the pool of potential applications which were discussed in Chap-ter 2 and in the cited literature, the fraction of oxidized high molecularweight lignin products appears promising for the use in dispersants dueto the increased amount of carboxyl groups. Although they are onlypartly present as free carboxyl groups, the esterified ones can easily betransformed by a saponification reaction. The action of dispersants is ingeneral based on steric stabilization, electrostatic repulsion or a combina-tion thereof. The carboxyl groups in the lignin products can in this casecontribute to the electrostatic repulsion of the molecule when present inthe deprotonated form. Steric stabilization can as well be caused by themacromolecules in the high molecular weight fraction but is assumed tobe less pronounced due to the lower molar mass compared to kraft ligninitself.

Another potential application that could benefit from the chemicalproperties of the high molecular weight fraction is the use in PF resins.The lower amount of analyzed functional groups ideally opens up morepossibilities for formaldehyde attack. However, the analysis of reactive5-positions in the aromatic ring which relies on formaldehyde polymer-ization resulted in a lower reactivity with formaldehyde and therebyproved this expectation wrong. Apart from PF resins, the abundance ofcarboxyl groups qualifies the fraction for its use in other polymer resins.

For the other applications that are discussed for lignin, no apparentadvantage of the oligomeric product fraction compared to the originallignin can be found based on the chemical properties. The substratesof epoxy resins and PU foams are polymerized by involving hydroxylgroups which have been found to be lower. Emulsifiers require interfacial


properties based on specific polarities that the high molecular weightfraction cannot provide in a better way than the original kraft lignin. Theuse as antioxidants or substrates for the fabrication of carbon fibers canneither be assumed to result in a better performance. Instead, chemicalmodifications on the abundant functional groups can make the fractionof high molecular weight products interesting for these applications andmay result in a superior performance compared to conventional kraftlignin. However, no enhanced performance can a priori be assured andappropriate tests in the laboratory are required for verification.

135

Chapter 8

Conceptual process design and

cost estimate

After all main process steps have been validated and investigated in theprevious chapters, a conceptual process as well as a cost estimate arefinally presented in the current chapter. The process including the mainunit operations is described and different aspects of the process designare discussed. Subsequently, a basic cost estimate is performed revealingpotentials and drawbacks of the current design and process conditions.Thereby, possibilities to bring the valorization of lignin into aromaticchemicals closer to commercial scale can be found and discussed.

8.1 Process design and description of theprocess

Based on the main unit operations that were presented in the previouschapters, a conceptual process design for the catalytic oxidation of lignininto monomeric aromatic chemicals as well as into a fraction of oligomericlignin products of higher molecular weight was done. The process com-prises the main process steps of reaction, fractionation and separation aswell as the recovery of the internal process chemicals and the purificationof the monomeric products. The conceptual flow diagram is depicted inFigure 8.1 and described in the following.

136 8. Conceptual process design and cost estimate

Me

thano

l/Wate

r

+C

ata

lyst

O2 /a

ir

Lig

nin

Rea

ctio

n P

rod

ucts,

Meth

anol/W

ate

r

+C

ata

lyst

Extra

ctio

n

(Ca

taly

st/S

olv

ent

Re

co

ve

ry)

Fra

ctio

natio

n(O

rga

nic

So

lve

nt

Nan

ofiltra

tion)

Lig

nin

Oxid

atio

n

So

lve

nt

Re

cove

ryV

acu

um

Dis

tillatio

n

So

lve

nt

Re

co

ve

ry

(Mo

dific

atio

n)

Me

thano

l/Wate

r

+C

ata

lyst

Reactio

n P

rod

ucts

,

Eth

yl A

ce

tate

Eth

yl A

ceta

te

Eth

yl A

ce

tate

Eth

yl A

ce

tate

Hig

h M

w

Pro

du

cts

,

Eth

yl A

ce

tate

Hig

h M

w

Pro

du

cts

Mo

no

me

rs,

Eth

yl A

ceta

te

Mon

om

ers

O2 /a

ir,

(DM

E,C

O2 )

Eth

yl A

ce

tate

Meth

anol,

(Wate

r),

(Cata

lyst)

Rea

ctio

n P

rod

ucts

,

Wa

ter/(M

eth

ano

l)

+C

ata

lyst

Me

tha

no

lR

em

ova

l

Wa

ter/(M

eth

ano

l)

+C

ata

lyst

Filtra

tion

Cry

sta

llizatio

nV

an

illin,

Meth

yl

Va

nilla

tescC

O2

Extra

ctio

n

Me

thyl D

eh

yd

roa

bie

tate

Va

nillin

Me

thyl V

an

illate

Cry

sta

lliza

tion

5-C

arb

o-

meth

oxy-V

an

illin

Me

thyl 5

-Ca

rbo-

me

tho

xy-V

an

illate

Oth

er

Mo

no

mers

So

lids

Hig

h M

w P

rod

ucts

for d

esire

d A

pp

lica

tion

Meth

anol,

(Wate

r)

Op

tion

al

mo

dific

atio

n

ag

ent R

esid

ue

s

Fig

ure

8.1

:C

onceptualflow

diagramof

aprocess

toconvert

kraftlignin

intoarom

aticm

onomers

anda

fractionof

highm

olecularw

eightlignin

oxidationproducts

comprising

them

ainprocess

stepsof

reaction,fractionation

andseparation

asw

ellas

therecovery

ofthe

internalprocess

chemicals.

8.1 Process design and description of the process 137

Lignin oxidation

In the first step of the process, lignin is converted by acidic oxidation inthe methanol/water solution that contains the copper chloride catalyst.Lignin, the reaction solvent, and oxygen are fed to the reactor whereasthe solvent feed results from a process-internal recycling. Oxygen can beprovided in pure form or potentially be replaced by air which only needsto be compressed to the pressure in the reactor. Although several typesof reactors and modes of operation can be realized, the investigations inthe microreactor have shown that a continuous reactor that features fastheating and cooling is highly beneficial for the production of monomericproducts. The gas phase leaving the reactor contains, besides oxygenor air, carbon dioxide and dimethyl ether in low amounts. While theunused oxygen should be recycled to the reactor, dimethyl ether caneither be combusted to gain heat or energy for the process, or separatedand sold as product of the process.

Methanol removal

As the liquid reactor effluent cannot directly be subjected to liquid-liquid extraction with ethyl acetate due to its complete miscibility, it ismandatory to lower the methanol fraction in the solvent of the effluent.This can either be accomplished by addition of water or by evaporation ofa methanol-rich vapor. To avoid unnecessary dilution, the latter optionis preferred and the withdrawn methanol can be recycled to the solventbefore re-entering the reactor. The extent of methanol removal is atrade-off between the separation effort and ethyl acetate losses in theraffinate of the following extraction step. To avoid extensive losses ofthe extraction solvent in the following step, nearly complete removal ofmethanol is favorable.

Liquid-liquid extraction

In the next main process step, the reaction mixture is extracted withethyl acetate to recover the reaction products as described in Chapter 5.Thereby, the reaction solvent which accrues as raffinate stream can berecycled to the reactor. An Aspen Plus R© process simulation, which isbased on the process model in Appendix A.4 and will be explained in the


following cost estimate section, reveals a loss of 8.0% ethyl acetate in theraffinate when the same volumetric flow rates are fed to the extractioncolumn. These losses need to be recovered from the raffinate to avoidadditional costs within the process. Moreover, the raffinate phase needsfiltration to remove solid residues that are neither soluble in the reactionsolvent nor in the extraction solvent. Depending on the process condi-tions, this fraction may account for up to 25% of the fed lignin. Thesolid residues can either be chemically used for applications as previouslydiscussed for the main stream of high molecular weight reaction productsor ultimately act as a fuel for the process as well. Any reaction productswhich remain dissolved in the raffinate either get degraded in the follow-ing reaction or equilibrate during the following cycles as demonstratedin the literature [93].

Membrane separation

The extract phase enters the fractionation step which is realized as or-ganic solvent nanofiltration operated in discontinuous diafiltration mode.Ethyl acetate from process-internal recycling is added until the desiredpurity of either the retentate or the permeate fraction is achieved. Sub-sequently, the ethyl acetate can be recovered from both fractions bythermal processing and recycled back as the products do not require thesolvent for further processing or even need to be free of it. Althoughthe solvent composition was not determined experimentally in the re-tentate and permeate, it is expected to be comparable to the one of theextract phase before the membrane separation step in both cases. Asderived from the above-mentioned Aspen Plus R© process simulation, theextract phase contains impurities of water and methanol in the orderof 13.5mol % and 24ppm, respectively. The composition is depicted inthe ternary diagram methanol/water/ethyl acetate in Figure 8.2 with an"X". Owing to the distillation boundary between the two low boilingazeotropes "A" and "B" in the binary systems methanol/ethyl acetateand water/ethyl acetate, the composition of the solvent remains in thedistillation region in the lower left corner of the diagram during evap-oration, even when the methanol impurities in the extract are higher.Thus, the complete evaporation of the solvent at atmospheric pressurecan be realized at temperatures that do not exceed the boiling point of

8.1 Process design and description of the process 139

X

A (62.2°C)

B (71.6°C)

x Eth

yl a

ceta

te [m

ol m

ol-1 ] x

Meth

anol [m

ol mol -1]

xWater

[mol mol-1]

0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55 0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95

0.0

5

0.1

0.1

5

0.2

0.2

5

0.3

0.3

5

0.4

0.4

5

0.5

0.5

5

0.6

0.6

5

0.7

0.7

5

0.8

0.8

5

0.9

0.9

5

0.9

50.9

0.8

50.8

0.7

50.7

0.6

50.6

0.5

50.5

0.4

50.4

0.3

50.3

0.2

50.2

0.1

50.1

0.0

5

EA

(77.1°C)

Figure 8.2: Ternary diagram methanol/water/ethyl acetate at 1barshowing residue curves and the distillation boundary. "X"denotes the approximate composition of the solvent in thepermeate and retentate, "A" and "B" are the low boilingazeotropes of the binary systems methanol/ethyl acetateand water/ethyl acetate, respectively.


ethyl acetate (TB = 77.1 C) and will not cause thermal degradation ofthe products.

Downstream processing

Depending on the aromatic monomers that need to be isolated from themonomeric product fraction, different embodiments for the processingare possible. Options that yield only vanillin as product as well as op-tions that recover all monomeric products can be realized as alreadydiscussed in the previous chapter. In any case, vacuum distillation rep-resents a promising first step in the processing of the products. Theresidues from the vacuum distillation can either be further processed(e.g. defunctionalization to phenolic compounds) or, as shown in theflow diagram, be added to the fraction of oligomeric lignin oxidationproducts if beneficial.

As discussed before, there are two options for the high molecularweight fraction after the complete evaporation of the extraction solvent.It can either be directly utilized in a desired application or be subjectedto a chemical modification reaction. As the chemical modification mayrequire costly reagents, the recovery of the unused modification agentcan be mandatory.

Alternative process designs

Although this conceptual design of the process for the oxidation of lignininto aromatic chemicals represents a promising approach from a technicalpoint of view, numerous alternative process configurations are realizable.The optimum one for a commercial realization of the process has tobe found based on more profound experimental and numerical data incombination with economical considerations.

8.2 Cost estimate

8.2.1 Evaluated cases

To eventually evaluate the presented process with regard to its econom-ical potential, a cost estimate was performed. As only a conceptual

8.2 Cost estimate 141

design was done so far and no detailed data for all process steps andconditions is available, three different cases were analyzed concerningtheir cost-effectiveness and are explained below.

The first case is the "batch case" and based on the results from thelaboratory experiments in the batch reactor. The second case is partlybased on the experimental results in the continuous microreactor butalso contains, in contrast to the first case, assumptions on the best prod-uct yield, chemical losses in the process as well as feasible improvements.The last case is based on a best case calculation, assuming a combinationof the best results obtained within this thesis together with the processimprovements considered as feasible. In any of these cases, the cost esti-mate reveals potentials and drawbacks of the current conceptual processdesign. The basic assumptions for the three cases are summarized inTable 8.1.

Table 8.1: Relevant assumptions for the cost evaluation of the threecases.

Case 1 Case 2 Case 3Experimental basis Batch Continuous Ideal

Lignin Indulin AT Indulin AT Enz. HydrolysiscLig [g L−1] 10 20 100O2 source Pure Air AirMeOH consum. [%] 12.3 5 5Acid consum. [t/tLig] 0.21 0.21 0.21O2 consum. [t/tLig] 0.72 0.72 (Air) 0.72 (Air)DME use Energy Product ProductMonomer yield [wt%] 5.26 11.8 19.0

First Case

As mentioned above, the data for the first case originates in the resultsfrom the batch experiments for a reaction time of 20min. The consump-


tion of process chemicals was as well calculated from the batch experi-ments. While oxygen and acid are assumed to be exclusively consumedby lignin and therefore calculated based on the amount of convertedlignin, the methanol degradation is independent thereof and listed in apercentage loss. The product yield in the first case just includes vanillinand methyl vanillate which were the only quantified products in thiscase. Dimethyl ether is incinerated for energy production and oxygencompletely recycled except the consumption during the reaction.

Second Case

In the second case, the reaction is continuously conducted at the con-ditions that resulted in the maximum yield of 5.02wt% within the mi-croreactor experiments. However, as alkali lignin from Sigma-Aldrichwas used instead of Indulin AT to lower the chance of precipitation, thetotal yield including all monomeric products was multiplied by the fac-tor of 2.35 that represents the yield ratio of these lignins in the batchautoclave. As lower consumption of methanol can be expected duringthe short residence times, even though conducted at higher reaction tem-peratures, its use is supposed to be lower by a factor of about 2.5. Air isused as oxidant as it was done in the former vanillin processes whereasdimethyl ether is now assumed to be separated from the gas phase asadditional product. Based on the cited literature on lignin oxidation inthis thesis, an increase in lignin concentration by a factor of two shouldvery well be feasible on a non-laboratory scale.

Third Case

Finally, the third and best case includes mainly the same assumptionsas the second one. However, a lignin concentration of 100 g L−1 whichequals a 10% loading is assumed. Although this value appears highon a first glance, it is, on the one hand, still in the upper range ofconcentrations used in literature and, on the other hand, required tomake lignin the main reactant with higher consumptions than methanolas will be seen later on. In addition, a more native lignin from ball-milling followed by enzymatic hydrolysis, as evaluated in Chapter 4, wasassumed as feedstock. Hence, the yield from Indulin AT was additionally


multiplied by a factor of 1.61 which was found as monomer yield ratiobetween the lignins in those experiments.

Assumptions

As a designated application for the oxidized high molecular weight frac-tion is still pending, potential applications were assumed to be simi-lar to those of lignin itself. Thus, the value of this fraction was pre-sumed to be identical to lignin in all cases. The presented cost estimatetherefore rather assesses the benefit of separating and isolating valuablemonomeric products if no additional revenue can be generated by theoligomeric product fraction. Also, the cost estimate does not considerutility and equipment cost in the first stage, but these aspects will bediscussed later on. The assumptions that are valid for all cases are listedbelow

• Only material streams that enter or exit the main unit operationssignificantly contribute to the process cost.

• The fraction of oligomeric lignin oxidation products can be usedin the same applications as lignin and therefore creates the samerevenue.

• The recovery of the monomeric products during separation andpurification is 80% based on industrial experiences provided inthe literature [7].

• Energy and utility cost are not considered in the calculation.

• Oxygen and acid are solely consumed by lignin and do not changein between the cases while methanol conversion decreases withlower reaction severity.

• The consumption of ethyl acetate equals the raffinate losses of 8.0%as derived from the Aspen Plus R© process simulation.

Cost benefit analysis

The cost estimate of the process for each of the three cases is performedbased on a cost benefit analysis (CBA), a tool which is commonly applied


in business administration. Therein, the cost-effectiveness is evaluatedin terms of the benefit-cost ratio (BCR) to summarize the overall valuefor money. Applied to the present cost estimate, it relates the monetarygain as total revenue of chemicals sale to the acquisition cost of feedstockchemicals. With the component mass flows mi and the component pricespri it calculates to

BCR =Total revenue

Total cost=

∑

Prod mi · pri∑

Feed mi · pri(8.1)

Obviously, the higher the BCR the better the economical success thatcan be expected from the process. Moreover, the equation must resultin a value above unity to be further considered as a realistic option.To be economically competitive even when all additional costs includingutility and equipment cost are considered, a BCR of at least 2 to 3 ismandatory. In the following, the BCR is calculated for all three casesand the influence of additional costs is discussed.

8.2.2 Chemicals cost and revenue

The current market prices (as of November 2012) of all chemicals thatenter or exit the process are summarized in Table 8.2. The values forthe commodity chemicals were collected from an online database [128].In contrast, the forecast price for kraft lignin in commercial quantitiesas found in literature and discussed in science, covers a wide range ofmore than one order of magnitude owing to the fact that no reliablemarket exists yet. The lignin price was finally chosen according to theaforementioned database for a kraft lignin that will become availablefrom the Lignoboost plant which is currently built in the United States[28]. Price information for oxygen and vanillin were as well obtainedfrom the literature [56, 129]. However, no values for the revenues ofany monomeric product except vanillin could be found as no commercialmarkets for these products exist either. Nevertheless, methyl vanillatecan be used as a flavoring (taste like "vanilla aroma obtained from oak")and perfuming agent just as vanillin. Methyl dehydroabietate, on theother hand, can function as an emollient, skin conditioner or viscositycontroller in cosmetics. Although no applications for the carbomethoxylderivatives of vanillin and methyl vanillate exist, potential uses are in


the same commercial markets as vanillin where in general many othervanillin derivatives find application as well. Therefore, the sale prices ofall monomeric products were assumed to equal the one of vanillin.

Table 8.2: Current or prospected (indicated by *) market prices of allchemicals that affect the cost balance of the process in thecost estimate.

Chemical Price pri [$/t] Reference

Kraft lignin 720* [128]Oxygen 210 [129]Methanol 435 [128]Sulfuric acid 230 [128]Ethyl acetate 1200 [128]Dimethyl ether 750 [128]Vanillin 12000 [56]Other monomers 12000* as vanillin

The energy that is set free during dimethyl ether combustion wasassumed to have a value of 0.07 $/kWh. This price was chosen accordingto the current industry price of energy from natural gas in the EuropeanUnion (as illustrated on their website www.energy.eu).

8.2.3 Cost-effectiveness of the cases

The values presented in the previous table were used to calculate thecost and revenue of the material streams in the process for each casebased on 1 t of lignin as feedstock. The cost balance, which includes theassumptions as stated above, is presented for all three cases in Table 8.3.

First Case

According to the table, the cost estimate based on the batch experiments(Case 1) results in a BCR of 0.71 and in a negative balance of −1724.3$per converted ton of lignin. Surprisingly, in terms of both consumption

www.energy.eu


Table

8.3

:C

ostestim

atefor

thethree

discussedcases

basedon

theconversion

ofone

tonof

lignin.

Case

1C

ase2

Case

3C

hemical

Cost/R

evenue[$]

Cost/R

evenue[$]

Cost/R

evenue[$]

Lignin

(1t)

-720.0(1t)

-720.0(1t)

-720.0O

xygen/air(0.72

t)-125.7

(0.72

t)0

(0.72

t)0

Methanol

(7.77

t)-3381.5

(1.58

t)-687.3

(0.32

t)-137.5

Sulfuricacid

(0.21

t)-47.4

(0.21

t)-47.4

(0.21

t)-47.4

Ethyl

acetate(1.44

t)-1722.2

(0.72

t)-861.1

(0.14

t)-172.2

Monom

ericproducts

(0.05

t)505.0

(0.12

t)1132.8

(0.19

t)1824.0

High

Mw

products(0.95

t)682.1

(0.88

t)635.0

(0.81

t)583.2

Dim

ethylether

(5.59

t)3085.4

(1.14

t)851.7

(0.23

t)170.3

Balance

-1724.3303.8

1500.5B

enefit-costratio

(BC

R)

0.711.13

2.39


and cost/revenue, the calculation in this column is governed by methanolconversion and dimethyl ether production and not by lignin and its reac-tion products. While oxygen and sulfuric acid consumption are tolerable,ethyl acetate losses in the raffinate represent another significant entry.These costs can, however, be reduced in the case of complete ethyl ac-etate recovery from the raffinate which would result in a BCR of aboutone. Furthermore, an increase in lignin concentration will also lower thesolvent-related costs which represent the main economic drawback of thecurrent case.

Second Case

The calculation for the second case already results in a positive balanceof 303.8 $ per converted ton of lignin and a BCR > 1. Owing to the lowerspecific methanol and ethyl acetate consumption for a lignin concentra-tion of 20 g L−1, these entries are less significant than before. Neverthe-less, methanol and dimethyl ether still contribute in higher masses to theoverall process balance than lignin itself. Eventually, the total balancereveals a positive value due to the increased yield of monomers, as de-rived from the continuous microreactor experiments. This total revenueis, however, still too low to account for additional costs for utility usage,apparatus construction, staff costs, etc.

Third Case

The cost estimate for the ideal case results in a reasonable balance of1500.5$ per converted ton of lignin and a BCR of 2.39. Due to its highconcentration, lignin and the derived products finally represent the mainprocess chemicals in the balance in terms of conversion and production.Moreover, the lower consumption of methanol and ethyl acetate does notsignificantly affect the balance anymore. Although the assumption of alignin from enzymatic hydrolysis as feedstock with an ideal monomeryield of 19% holds, even a kraft lignin as feedstock with the yield asassumed in case 2 would return a BCR of 1.80. Hence, the third of thethree evaluated cases is the only one that justifies a more profound costanalysis and more detailed investigations under the assumptions statedabove.


Utility cost

In order to estimate the additional utility cost for the third case, a pro-cess simulation, as already mentioned before, was performed in AspenPlus R©. The underlying model that was built for this purpose comprisesthe sequential processing units air compressor, solvent pump, mixer,heater, cooler, phase separator, distillation column, and extraction col-umn in the reaction solvent cycle, as well as an additional evaporatorand condenser in the extraction solvent cycle. A flowchart of the modelas well as properties of selected process streams from the simulation canbe found in Appendix A.4. The simulation, which will not be discussedin detail in the context of the thesis, calculates the energy requirementto heat, cool, pump and compress the solvents and gases for one possibleembodiment of the process. Electricity for heating/electrical power aswell as cooling water which can be heated from 15 C to 25 C are usedas process utilities in the model. Assuming a lignin concentration of100 g L−1 and neglecting the exothermic reaction heat of the oxidation,the simulation reveals a total process duty per ton of converted lignin of37.0GJ for heating/electrical power and 34.7GJ for cooling. This resultsin a total utility cost of 1109.9$ per ton of converted lignin when calcu-lating with general utility prices provided in the literature (0.10 $/kWhof electricity and 0.10 $/t of cooling water) [130]. Taking these costs intoaccount, the BCR decreases from 2.39 to 1.18. Energy integration in theprocess of 25%, 50%, 75% or 100% would result in an even better BCRof 1.34, 1.55, 1.84 and 2.26, respectively.

Conclusion

The calculated BCR values show, that an improved version of the processis promising to run economically even though not all costs are includedand some assumptions still have to be proven valid. Besides the afore-mentioned points, a relevant value creation can be expected when thefraction of high molecular weight reaction products can efficiently beused and marketed. To generate a higher revenue than lignin itself (asrequired based on the assumptions in the cost balance), the fraction mustprofit from its chemical properties and be incorporated in applicationsas discussed in Chapter 7. As this fraction can account for up to 90%

8.3 Summary 149

of the mass of lignin fed to the process, an incremental increase in valuecan significantly improve the cost balance.

8.3 Summary

A conceptual process to convert lignin into both aromatic monomericchemicals and a fraction of oxidized high molecular weight products wassuggested. It is based on the unit operations that were discussed inthe previous chapters, namely acidic oxidation of lignin, recovery of thereaction products from the reaction mixture by liquid-liquid extraction,fractionation of the products by organic solvent nanofiltration, as wellas isolation and purification steps for the products. In addition, thermalprocessing is required to recover the internal process chemicals and tocontribute to an ecological and economic process.

In order to evaluate the economic potential of the designed process,three potential cases were evaluated. They were based on different exper-imental data sets and assumed an application and a value of the oxidizedhigh molecular weight lignin fraction equal to lignin. Thus, all cases infact evaluate the potential of producing and separating the monomericproducts. The benefit-cost ratio revealed the high solvent consumptionand the low yields in the batch reactor as main economic drawbacksin the first case. For higher lignin loadings and higher monomer yieldswhich are expected for lignins from mild isolation processes, a BCR of2.39 was calculated without including additional costs like utility costsand investment costs. When considering the utility costs for one poten-tial embodiment of the process, the BCR decreases, depending on thepercentage of process heat integration, to values in the range of 1.18(0%) to 2.26 (100%). For a more detailed evaluation of the process eco-nomics, the return on investment (ROI) will also play a decisive role. Tocalculate its value, the annual production has as well still to be defined.

Based on the economical considerations and the cost estimates for thethree cases that were performed in the current chapter, the followingpoints have to be addressed to render the valorization of lignin econom-ically successful:

• The concentration of lignin in the reaction solvent has to be signif-icantly increased compared to the 10 g L−1 which were convention-


ally used in the batch reactor. Otherwise the methanol consump-tion and dimethyl ether production govern the cost balance of theprocess in a negative manner.

• A valuable application for the fraction of oxidized high molecularweight products has to be found. The application must benefitfrom the chemical properties of the fraction to create a value higherthan that of lignin which is as well mandatory to contribute to apositive cost balance.

• A maximum of monomeric products has to be produced from lignin.For this purpose, commercial scale reactors must feature continu-ous processing with fast heating/cooling and short reaction times.

• The virtually complete recovery of ethyl acetate from the raffinateis required to positively influence the cost estimate and contributeto a low solvent use/degradation. Alternatively, the replacementof ethyl acetate by cheaper and less miscible derivatives for theextraction is also an option in this context.

• Possibilities for heat integration within the process have to be as-sessed and can significantly alter the BCR towards an economicallyfeasible value.

As soon as these points are fulfilled or at least addressed, a moredetailed process design and cost calculation can be performed towards atechnically and economically feasible valorization of lignin into aromaticchemicals.

151

Chapter 9

Conclusions

A novel approach for the complete valorization of lignin into differ-ent high value monomeric aromatic chemicals as well as a fraction ofoligomeric lignin oxidation products is presented in this thesis. Themonomers can be purified and commercialized as fine chemicals or inter-mediates thereof while the high molecular weight fraction can find appli-cation in e.g. dispersants, binders, adhesives or resins. The approach isbased on the acidic oxidation of lignin with subsequent recovery, fraction-ation and processing of the reaction products. In contrast to previousapproaches in the literature that either focus on the transformation intomonomers or directly use lignin as polydisperse mixture in various appli-cations, the presented concept exploits the complete potential of ligninby using a combination thereof. Consequently, lignin is sustainably con-verted into chemical products, maintaining its aromatic structure andvalorizing it without leaving ligneous waste material behind.

Several process steps are required within a potential process to producethe above-mentioned products from lignin. Among those, the depoly-merization of lignin by the acidic oxidation reaction, the recovery of thereaction products from the reaction solvent by liquid-liquid extractionand the fractionation of the product mixture into the two desired frac-tions by organic solvent nanofiltration represent the most crucial unitoperations. These process steps were studied in detail in the presentthesis and validated in the laboratory. The influence of catalyst, solventcomposition, solvent acidity and lignin type during the reaction wereclarified by experiments in the batch reactor. Moreover, the impact oftemperature, pressure and lignin concentration during the continuousprocessing were investigated in a microreactor. For the liquid-liquid ex-traction, different organic solvents were evaluated based on their physic-

152 9. Conclusions

ochemical properties as well as on economic considerations and safetyaspects. Furthermore, several solvent stable polymeric membranes werestudied with ethyl acetate concerning their transport properties in puresolvent and afterwards investigated for the fractionation of lignin productmixtures. Besides these main process steps, additional unit operationsare required in a potential process that cover distillation, evaporation,filtration as well as purification steps for the monomeric units. Whilethe former steps were also validated in the laboratory, various patents onthe separation and purification of the monomeric product mixture existin the literature and are discussed within this work. Based on the men-tioned process steps, a conceptual design for the process was performedand discussed at the end of the thesis. Moreover, a cost estimate wasfinally conducted for this conceptual process design. Three cases wereeconomically evaluated and revealed the main potentials and drawbacksof the process. Thereby, the cost estimate opened up possibilities on howto improve the valorization of lignin towards its industrial commercial-ization.

Hence, the investigations presented in this thesis address many aspectsrelated to the valorization of lignin as well as to the required unit oper-ations to realize it as a process on industrial scale. The major scientificfindings of the present thesis, which concern reaction and separation, aresummarized in the following.

Acidic oxidation of lignin in the presence of transition metalsalt catalysts

In the beginning of the thesis, the acidic oxidation of lignin was in-vestigated in detail. As the oxidation was assumed to be catalyzed bythe redox potential of a homogeneous catalyst, different transition metalsalts were experimentally tested. Copper chloride emerged as the bestof the transition metal salts, besides iron chloride, which both even out-performed the former polyoxometalate catalyst. Although the use ofcopper chloride does not yield significantly improved amounts of vanillinor methyl vanillate, early peak concentrations of these products and arapid depolymerization of the lignin molecule from 3200 g mol−1 downto 500 g mol−1 proved the enhancement of the depolymerization kinet-ics. Moreover, the oxygen incorporation into the products was proven

153

by both FTIR and the amount of remaining oxygen in the reactor.In addition to the known monomeric products vanillin and methyl

vanillate, five other aromatic monomers were detected in the reactionwith copper chloride catalyst. Four of them could be identified as prod-ucts from the oxidation of vanillin or methyl vanillate, revealing thedestiny of the two main products for longer reaction times. The fast ox-idation kinetics in the reactions with the new catalyst and the possiblederivatization and degradation of the monomeric products demandedfor other reaction systems that enable short residence times with fastheating and cooling in a preferably continuous system.

Continuous oxidation of lignin in a two-phase flowmicroreactor

A continuous high pressure microreactor setup as presented within thisthesis was effectively employed to overcome the drawbacks from conven-tional batch reactors during lignin valorization. The heating and coolingphases that exceed the intentional reaction times in the batch reactor areavoided on a microscale. In addition, high oxygen pressures can safelybe realized and experimental conditions can rapidly be screened with aminimum of reactant. Thereby, a detailed study on the influence of re-action temperature, oxygen pressure, lignin concentration and residencetime on a minute scale was possible.

Concentration profiles of the five main monomeric products versus res-idence time in the microreactor confirmed the proposed reaction path-ways and allowed for the construction of a reaction network for themonomeric reaction products. Vanillin and methyl vanillate are obtainedas main aromatic monomers but undergo further derivatization to theircarbomethoxy derivatives with increasing reaction time. Methyl dehy-droabietate is formed from abietic acid in lignin via dehydroabietic acidbut is later on decomposed as all the other aromatic products.

The influence of temperature, pressure and lignin feedstock concen-tration on the monomeric products was studied and temperature turnedout to be the most influential parameter. An increase in temperatureresulted in a very fast reaction with early concentration maxima of theproducts, especially for vanillin. Its maximum concentration at 210 Cwas more than seven times higher than at 170 C and obtained within

154 9. Conclusions

less than one minute of reaction time. The phenomenon of higher andearlier concentration peaks was generally observed for the other prod-ucts as well. A maximum yield of 5.02wt% monomeric products wasobtained at 250 C for the alkali lignin from Sigma-Aldrich. Assumingthe same ratio of product yields as in the batch reactor, these conditionswould result in a yield of 11.8wt% for Indulin AT and even 19.0wt% fora mildly treated lignin as the ball-milled and enzymatically hydrolyzedlignin produced within this work. Based on the data generated in themicroreactor, the concentration of any main aromatic product arisingfrom the acidic oxidation of lignin can be optimized.

Fractionation of the lignin oxidation products by organicsolvent nanofiltration

In the studied approach of valorizing lignin by conversion into monomericchemicals and oxidized products of higher molecular weight, the efficientfractionation represents a decisive process step. As ethyl acetate emergedas a good extraction solvent for the recovery of the products from thereaction solvent, organic solvent nanofiltration with solvent-stable flatsheet membranes was investigated for the fractionation. Among the fivemembranes that were studied in this context, SelRo R© MPF-44 had alow permeability for the solvent and was therefore not suitable for thedesired separation. Although the DuraMemTM membranes underwentirreversible membrane compaction, characterization experiments withpure ethyl acetate and model compounds revealed reasonable permeatefluxes and model compound rejections for the four other membranes.

In the subsequently studied separation of the real product mixture,the two DuraMemTM membranes, which have molecular weight cut-offvalues almost in the ultrafiltration range and are somewhat porous, weresubject to intensive fouling. This was not the case for the more densePuraMemTM membranes. While all of the four remaining membranesshowed rejection values that enable a fractionation as intended for thevalorization of lignin, PuraMemTM S380 slightly outperformed the othermembranes with a permeability of 2.06L m−2 h−1 bar−1 and rejectionvalues of 38.4% for the monomers, 83.2% for the dimers, 93.2% for thetrimers and 96.6% for the oligomers. Hence, this membrane was chosenfor the desired fractionation step in lignin valorization.

155

Chapter 10

Outlook

The valorization of lignin into chemicals represents a promising approachin the efforts to replace the fossil-based feedstocks of the chemical in-dustry by renewable resources. The present thesis contributes to theseefforts by introducing a twofold approach for the complete exploitationof lignin without leaving ligneous waste material behind, by validatingthe main unit operations to accomplish this concept and by presenting aconceptual process to realize it on a commercial scale. However, the eco-nomical breakthrough is linked to several improvements of the processwhich have to be implemented as discussed before. These necessary im-provements in combination with required laboratory validations for thecurrent approach are discussed in the first section of this chapter. Al-ternative reaction concepts to improve the monomer yield are discussedin the middle part. The last section describes ways to find a valuableapplication for the oligomeric lignin oxidation product mixture.

10.1 Realization of the current approach

As resulting from the cost estimate in the present thesis, several changesin the process and the related process conditions are mandatory to renderthe current concept more viable. These changes are recommended basedon assumptions and experiences collected within this work but still haveto be realized and validated in the laboratory to prove them correct.

First of all, the concentration of lignin has to be significantly increased,as already stated in the conclusions of the cost estimate section. Forthis purpose, preliminary experiments with Indulin AT concentrationsof 25 g L−1 and 50 g L−1 were performed in the batch reactor. In com-

156 10. Outlook

parison with a lignin concentration of 10 g L−1, the results with 25 g L−1

showed even higher yields of 5.2wt% vanillin and 3.1% methyl vanillateafter 75min of reaction time. The experiment at 50 g L−1, on the otherhand, returned lower yields of 2.8wt% and 1.3wt%, respectively. Like-wise, the high lignin concentration did not result in significant depoly-merization. While the product mixture from the 25 g L−1 experimenthad an average molecular weight of 1785 g mol−1 after a reaction timeof 75min, the one from the 50 g L−1 experiment resulted in an averageof 2932 g L−1, which reveals a very slow depolymerization. Nevertheless,the experiments show, that an increase in lignin feedstock concentrationis in general possible. Adaption of the reaction conditions (e.g. longerreaction times) is, however, required to achieve comparable results fordifferent lignin loadings.

The feasibility of fast heating and cooling in a continuously operatingreactor was as well listed as a prerequisite in obtaining higher productyields and to improve the process’ economical viability. Based on theresults of the present thesis, a monomeric product yield of 11.8wt% and19.0% is presumed when Indulin AT or a lignin from enzymatic hydrol-ysis of lignocellulosic biomass had been used in the microreactor exper-iments instead of the post-sulfonated alkali lignin from Sigma-Aldrich.As a matter of fact, these assumptions have to be backed up by exper-iments in the laboratory. However, an appropriate reactor is missing,since the latter types of lignin are mostly insoluble in the solvent andwould clog the microreactor. As a limited heat transfer in tubular re-actors of larger diameter would result in more moderate temperaturegradients during heating and cooling, a compromise in size has to befound for an ideal continuous reactor. So-called tubular millireactorswith diameters on the milliscale represent promising reactors to realizethe required reaction conditions and to validate the maximum possibleyield under the suggested optimum conditions.

Moreover, several other requirements concerning the complete processare essential as discussed in Chapter 8. They address an efficient opera-tion of a potential plant including but not limited to the heat integrationand the recycling of chemicals in the process. In this context, processsimulations at a higher project definition level which also include a pinchanalysis are obligatory. While most of the unit operations have alreadybeen validated in the laboratory, the separation and purification of the

10.2 Enhanced monomer yield by a membrane reactor concept 157

monomeric products was proposed based on patented literature but stillrequires confirmation by a proof-of-concept study. Finally, the valida-tion of a more detailed process design on a laboratory and pilot scale areinevitable before a commercial realization of the process.

10.2 Enhanced monomer yield by a

membrane reactor concept

When studying the reaction pathways of the main monomeric productsas illustrated in Figure 4.2 on page 53, the derivatization and degradationof the aromatic products represent a main drawback of the reaction froma chemical point of view. Thereby, the maximum yield of monomericreaction products that can be obtained in a conventional batch or con-tinuous reaction step is limited. This is either due to the fact that ligninunits are not yet fragmented to monomeric units at the end of the re-action or that the the targeted products are further oxidized. Thus, theremoval of the products after formation to protect them against degra-dation implicitly represents a superior approach. One possible conceptto realize the product removal would be an extractive reaction, in whichthe products are extracted from the reaction solvent by an organic sol-vent as second liquid phase in the reactor. However, as the productswould still be exposed to high temperatures and pressures, a protectionfrom degradation is not assured. Instead, an involvement of the organicsolvent in the reaction as well as the degradation of the solvent at thereaction conditions could occur.

A more promising concept to realize the product removal during thereaction is a membrane reactor approach. The selective and simulta-neous removal of the products from a continuous or batch reactor cancontribute to a higher yield of aromatic monomers. The technical chal-lenge, however, is to find a reactor material that both selectively allowsthe monomeric units to permeate and to withstand the thermal, me-chanical and chemical conditions that are required for the acidic oxida-tion. Membrane materials that are also studied for membrane bioreac-tors like metal, ceramics or cermet potentially qualify for this application[131, 132]. On the other hand, the permeation of oxygen gas should besuppressed at the best which, for instance, could be realized in a reactor

158 10. Outlook

similar to the high temperature/high pressure zeolite membrane reactorwith low gas permeability used by Gallucci et al. for the hydrogenationof carbon dioxide [133]. Alternatively, low oxygen permeation can berealized by phase separation in a reactor in which only the bottom partwould be made of permeable material. In any case, the reactor wouldneed to be specifically customized to all present requirements as thereare no generic solutions at hand [134]. If no technical concept for therealization were found, the sequential reaction and separation in cyclewould be an alternative approach to realize the discussed concept.

To study the feasibility of the membrane reactor concept from a chem-ical point of view, the question whether the mixture of high molecularweight lignin oxidation products can further be converted into monomericproducts was addressed in preliminary experiments in the laboratory.Instead of running the reaction in a membrane reactor, the followingsteps were consecutively repeated three times: acidic reaction in thebatch reactor, extraction with ethyl acetate, separation by organic sol-vent nanofiltration and evaporation of the organic solvent. The exper-iments were performed in the 400mL Premex autoclave with nominalreaction times of 0min (i.e. the reactor was cooled down after reachingthe reaction temperature). The average molecular weights of the mixturebefore and after the reaction steps as analyzed by SEC and the yieldsof vanillin and methyl vanillate related to the substrate amount in thecurrent run as analyzed by GC/MS are depicted in Table 10.1.

Table 10.1: Average molecular weight and yields of vanillin and methylvanillate during three consecutive oxidation steps withseparation of the monomeric products by organic solventnanofiltration in between each step.

Mw [g mol−1] Product yield [wt%]Run Before After Vanillin Methyl Vanillate Sum

I 3241 686 3.19 2.04 5.23II 1565 460 0.31 0.40 0.71III 849 458 0.00 0.77 0.77

10.2 Enhanced monomer yield by a membrane reactor concept 159

The results of the first run are in very good agreement with the resultsshown in Chapter 4. After extraction and separation of the monomericproducts, the average molecular weight of the feedstock mixture for thesecond run is more than doubled in comparison to the product mixturefrom the previous run and about half the one of the original lignin. Asanticipated, the high molecular weight oligomeric products of the firstrun can further be depolymerized. By just heating and cooling, the av-erage molecular weight is decreased in the second run from 1565 g mol−1

to 460 g mol−1. The yields are low compared to the first run though,and methyl vanillate seems to predominantly form in this run. Thesame is true for the third run, in which no vanillin but still noticeableamounts of methyl vanillate are created. The decrease in average molec-ular weight reveals that the depolymerization is carried on during thethird run. However, the comparison of the substrates of each run sug-gests that the continued depolymerization will cease during the followingruns. Although the total monomer yield related to the original lignin canbe calculated to have just increased by 12.8wt% after two and 26.7wt%after three runs, the membrane reactor concept deems feasible and spacefor improvements still remains.

To mimic the membrane reactor concept more accurately, a slightlyimproved concept was as well validated in the laboratory. The organicsolvent nanofiltration step was replaced by a conventional aqueous mem-brane separation, thereby making the extraction with a second solventand its evaporation expendable. Hence, the reaction mixture was justrepeatedly subjected to reaction and separation. For the membrane sep-aration step, a CMS-KX-036 membrane (Celfa AG, Seewen, Switzerland)with a molecular weight cut-off of 1000Da was employed.

According to the previous experiment with an organic solvent nanofil-tration step, the yield related to the original lignin was again increasedby 10.4wt% in a second run. The continued depolymerization of theoxidized fraction of oligomeric products could once again be proven by adecrease of the average molecular weight from 690 g mol−1 to 464 g mol−1

in the second run. In this experiment, the feedstock of the second runalready possessed a far lower average molecular weight compared to theprevious experiment which can be attributed to the different membranecharacteristics and the different MWCO of the membrane for the aque-ous separation. This emphasizes the fact, that the separation properties

160 10. Outlook

of the membrane as well play a decisive role in the efficiency of the mem-brane reactor concept. In any case, the preliminary experiments revealthe potential of the membrane reactor concept for the enhancement ofthe monomeric yield from lignin.

10.3 Sequential lignin treatment to enhance

the monomer yield

As already indicated within this thesis, another way to enhance themonomeric yield and thus the efficient conversion of lignin into chemicalproducts is a sequential lignin treatment, i.e. a two-step approach. Inthis concept, the remaining non-monomeric product mixture after the ox-idation reaction is subjected to another reaction step, which alternativelycould also be placed prior to the oxidation. Several approaches are foundin literature that pursue similar objectives in this context. For instance,Yan et al. used a two-step approach for an improved lignin hydrogenol-ysis by applying two steps with two different types of carbon supportednoble metal catalysts under hydrogen atmosphere [135]. Yoshikawa etal. combined a depolymerization step using a silica-alumina catalyst in abutanol/water mixture with a catalytic cracking over an iron oxide cat-alyst to produce phenols from lignin [136]. A two-step catalytic reactionconsisting of a base-catalyzed depolymerization in supercritical alcoholand a hydrodeoxygenation using a Co-Mo/Al2O3 catalyst into a refor-mulated hydrocarbon gasoline product was even patented by Shabtai etal. [137].

In the same manner, a two-step approach can potentially increase themonomeric yield of a single oxidation step as conducted within this the-sis. Although a hydrogenolysis as second step could be promising toyield phenolic bulk chemicals from the reaction products as suggestedin the literature, it appears not reasonable to run a reduction reactionjust after an oxidation which is in fact opposing to it [10]. Instead, anadditional oxidation step could be one of the options to continue thedepolymerization of the product mixture. In contrast to the previouslydiscussed membrane reactor concept, different reaction conditions, a dif-ferent oxidant or a different catalyst could be employed. To not degradeor derivatize the products that have already formed, a membrane separa-

10.4 Study on applications for the oligomeric product fraction 161

tion step as employed herein before is advantageous and thus preferableto the approaches in literature that do not include an intermediate sep-aration step.

In particular, the enzyme-catalyzed oxidation could be a potentialsecond reaction step to complement the chemical oxidation. Several en-zymes are known to be capable of oxidizing lignin, including lignin per-oxidase, manganese peroxidase and laccases. Thus, an increase in theyield of monomeric aromatic chemicals could be achieved by the sequen-tial combination of enzyme-catalyzed oxidation and catalytic chemicaloxidation with intermediate separation of the products [138]. A projectfor the investigation of this sequential treatment was recently launchedand is currently in progress.

10.4 Study on applications for the

oligomeric product fraction

Apart from efforts to convert the non-monomeric reaction products intofurther aromatic monomers by an additional reaction, the fraction canalso be valorized as a mixture in the applications which are also con-sidered for lignin like dispersant or resins as discussed in Chapter 7.Among them, the most promising applications in the competition withlignin are the ones that directly benefit from the chemical properties ofthe mixture.

In this context, prospective effects in possible applications need to beforecast based on the chemical properties of the fraction. Afterwards,laboratory experiments have to be performed with the high molecularweight lignin oxidation products in the desired application to prove theintegration as an additive beneficial. In case of resins, additional me-chanical tests are in general compulsory and have to be performed after-wards as well. First approaches to use a fraction of oxidized lignin withinpolyurethane formulations including experimental verifications and me-chanical stress tests of the foam can already be found in literature [127].

If the functionality of the non-monomeric fraction does not turn outadvantageous for a desired application, the chemical modification stillrepresents a possible intermediate step. Besides chemical reactions, mod-ifying the functionality by enzymes evolved as an alternative technique

162 10. Outlook

during the last years [139]. Several functional groups in the productsincluding carboxyl groups, phenolic hydroxyl groups and even free posi-tions on the aromatic ring can be used to perform modifying chemistrywith the product mixture. Based on its chemical functionality, adequatechemical modification reactions have to be chosen to get a desired effectof the fraction in the application. Possible modification reactions havethen to be successfully realized in the laboratory and afterwards as wellto be tested regarding their effect. With this possibility at hand, anappropriate tailor-made derivatization reaction can be used to make thefraction of high molecular weight lignin oxidation products a potentialcandidate in any of the various possible applications.

163

A Appendix

A.1 Synthesis of components for GC/MS

calibration


As methyl dehydroabietate was found to arise from abietic acid via de-hydroabietic acid under the reaction conditions in the batch reactor,the same conditions were applied for its synthesis. 300mg abietic acidwere mixed with 200mL of the acidic reaction solvent (80 vol% MeOH,20 vol% water, pH 1, 0.01mol L−1 CuCl2) in the 400mL batch reactor.After adding 10bar of oxygen, the mixture was heated to 170 C, followedby a reaction time of 90min. The reaction mixture was cooled down,transferred to a searating funnel and extracted in two steps with 50mLof chloroform. The solvent of the combined extracts from the two stepswas removed by disitllation in a rotary evaporator (Büchi LabortechnikAG, Uster, Switzerland) and the brownish solid residues were dryed ina vacuum oven at 60 C for one week. GC/MS analysis of the dryedsolids revealed methyl dehydroabietate with 85Area% which was finallyassumed as purity for the GC/MS calibration.


5-Carbomethoxy-vanillin was produced by esterification of 5-carboxy-vanillin with methanol. The latter was synthesized at the Institutefor Chemical and Bioengineering of ETH Zurich according to a pro-cedure described in the literature [140]. Ortho-vanillic acid was formy-lated by hexamethylenetetramine in glacial acetic acid at 100 C for 7h.After addition of 6N hydrochloric acid at 90 C and heating for an-other 10min, the reaction mixture was cooled and crystals of 5-carboxy-vanillin were obtained. The crystals were filtered, washed and recrystal-

164 A. Appendix

lized from an ethanol-water solution. The esterification with methanolto yield 5-carbomethoxy-vanillin was finally performed at 60 C to 70 Cin methanol with hydrochloric acid as catalyst.

A.2 High pressure microreactor setup

Solvent Inlet

Gas Inlet

Two-Phase

Outlet

Lignin Inlet

Silicon/Glass Mixing Element

Figure A.1: Establishment of gas-liquid two-phase flow in the mixingunit before entering the heated reaction zone.

A.2 High pressure microreactor setup 165

Syringe Pump

(Solvent)

Syringe Pump

(Lignin)

Oxygen Supply

Oxygen Mass

Flow ControllerFeed Valves

Figure A.2: Feedstock part of the high pressure microreactor setup

166 A. Appendix

Capillary

ReactorHeated

Oil Bath

Mixing

Element

Back-Pressure

Regulator

Figure A.3: Mixing unit and heated reactor of the high pressure mi-croreactor setup

A.2 High pressure microreactor setup 167

Measuring

Station

Sampling

Container

Waste

Container

Sampling

Valves

Figure A.4: Sampling part of the high pressure microreactor setup withmeasuring station

168 A. Appendix

A.3 Ternary phase diagrams

Ethyl acetate-water-methanol

0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55 0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95

0.0

5

0.1

0.1

5

0.2

0.2

5

0.3

0.3

5

0.4

0.4

5

0.5

0.5

5

0.6

0.6

5

0.7

0.7

5

0.8

0.8

5

0.9

0.9

5

0.9

50.9

0.8

50.8

0.7

50.7

0.6

50.6

0.5

50.5

0.4

50.4

0.3

50.3

0.2

50.2

0.1

50.1

0.0

5

xWater

[mol mol-1]

x Eth

yl a

ceta

te [

mol

mol

-1 ]x

Meth

anol [m

ol mol -1]

Figure A.5: Ternary phase diagram of the system ethyl acetate-water-methanol at 20 C and 1 bar.

A.3 Ternary phase diagrams 169

Toluene-water-methanol

0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55 0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95

0.0

5

0.1

0.1

5

0.2

0.2

5

0.3

0.3

5

0.4

0.4

5

0.5

0.5

5

0.6

0.6

5

0.7

0.7

5

0.8

0.8

5

0.9

0.9

5

0.9

50.9

0.8

50.8

0.7

50.7

0.6

50.6

0.5

50.5

0.4

50.4

0.3

50.3

0.2

50.2

0.1

50.1

0.0

5

xWater

[mol mol-1]

x Tol

uen

e [

mol

mol

-1 ]x

Meth

anol [m

ol mol -1]

Figure A.6: Ternary phase diagram of the system toluene-water-methanol at 20 C and 1 bar.

170 A. Appendix

Hexane-water-methanol

0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55 0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95

0.0

5

0.1

0.1

5

0.2

0.2

5

0.3

0.3

5

0.4

0.4

5

0.5

0.5

5

0.6

0.6

5

0.7

0.7

5

0.8

0.8

5

0.9

0.9

5

0.9

50.9

0.8

50.8

0.7

50.7

0.6

50.6

0.5

50.5

0.4

50.4

0.3

50.3

0.2

50.2

0.1

50.1

0.0

5

xWater

[mol mol-1]

x Hex

ane [

mol

mol

-1 ]x

Meth

anol [m

ol mol -1]

Figure A.7: Ternary phase diagram of the system hexane-water-methanol at 20 C and 1 bar.

A.3 Ternary phase diagrams 171

Diethyl ether-water-methanol

0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55 0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95

0.0

5

0.1

0.1

5

0.2

0.2

5

0.3

0.3

5

0.4

0.4

5

0.5

0.5

5

0.6

0.6

5

0.7

0.7

5

0.8

0.8

5

0.9

0.9

5

0.9

50.9

0.8

50.8

0.7

50.7

0.6

50.6

0.5

50.5

0.4

50.4

0.3

50.3

0.2

50.2

0.1

50.1

0.0

5

xWater

[mol mol-1]

x Die

thyl

eth

er [

mol

mol

-1 ]x

Meth

anol [m

ol mol -1]

Figure A.8: Ternary phase diagram of the system diethyl ether-water-methanol at 20 C and 1 bar.

172 A. Appendix

A.4 Aspen Plus R© process model

COMP-AIR

SOLVENT

REACTION EFFLUENT EFFLCOLD

MEOH-RICH

OFFGAS

SLV-BACK

REC-SOLV

EA

RAFFINATE

EXTRACT

MEOH

EA-REC

EA-CLOSE

EA-ADD

MIXER

HEATER COOLERPHASESEP

PUMP

COMP

BACKPRES

EXTRACTION

RECYCLE

EVAP

COND

MEOHREM

COOL

PREHEATER

EA-MIX

SLV-CLOSE

FEED

EXT-IN

MIXER

MIXER

Heat and Mat erial Balance T able

Stream SOLVENT COMP-AIR FEED REACTION MEOH-RICH EXT-IN EXTRACT RAFFINATE EA-ADD SLV-CLOSE

Temperature C 20.0 954.0 78.6 250.0 10.0 20.0 21.5 20.8 20.0 20.0

Pressure bar 100.000 100.000 100.000 100.000 1.000 1.000 1.000 1.000 1.000 100.000

Vapor Frac 0.000 1.000 0.087 0.854 0.000 0.000 0.000 0.000 0.000 0.000

Mole Flow kmol h-1 3084.814 625.023 3709.837 3709.837 3054.060 916.218 206.571 934.563 18.339 3072.405

Mass Flow kg h -1 83335.168 20000.000 103335.168 103335.168 82398.321 16506.396 16252.367 17940.584 1433.825 83832.508

Volume Flow m3 h -1 100.000 637.688 218.698 1405.288 97.536 16.527 17.857 18.081 1.571 100.368

Enthalpy GJ h-1 -791.216 19.202 -772.014 -637.522 -785.364 -262.063 -93.705 -270.413 -8.298 -790.817

Mole Fraction

METHANOL 0.642 0.533 0.533 0.638 37 PPM 24 PPM 36 PPM 0.634

WATER 0.358 0.298 0.298 0.361 1.000 0.135 0.983 0.142 0.360

ETHYL ACE 0.865 0.017 0.858 0.005

OXYGEN 1.000 0.168 0.168 0.002 trace 0.002

Figure A.9: Aspen Plus R© flowchart for the calculation of the utilitydemand and stream impurities as well as properties of se-lected streams. The calculation is based on Case 3 of theevaluated cases in the cost estimate.

Bibliography 173

Bibliography

[1] I. S. Goldstein. Organic chemicals from biomass. CRC Press,Inc.,Boca Raton, FL, United States, 1981.

[2] B. Kamm, P. R. Gruber, and M. Kamm. Biorefineries - IndustrialProcesses and Products. Wiley-VCH Verlag GmbH & Co. KGaA,2006.

[3] J. Zakzeski, P. C. A. Bruijnincx, A. L. Jongerius, and B. M. Weck-huysen. The catalytic valorization of lignin for the production ofrenewable chemicals. Chemical Reviews, 110(6):3552–3599, 2010.

[4] J. H. Lora and W. G. Glasser. Recent industrial applications oflignin: A sustainable alternative to nonrenewable materials. Jour-nal of Polymers and the Environment, 10(1):39–48, 2002.

[5] M. P. Pandey and C. S. Kim. Lignin depolymerization and conver-sion: A review of thermochemical methods. Chemical Engineering& Technology, 34(1):29–41, 2010.

[6] I. Hermans, E. Spier, U. Neuenschwander, N. Turrà, and A. Baiker.Selective oxidation catalysis: Opportunities and challenges. Topicsin Catalysis, 52(9):1162–1174, 2009.

[7] W. L. Faith, D. B. Keyes, and R. L. Clark. Vanillin. In IndustrialChemicals, pages 796–799. John Wiley & Sons.

[8] M. B. Hocking. Vanillin: Synthetic flavoring from spent sulfiteliquor. Journal of Chemical Education, 74(9):1055–1059, 1997.

[9] J. C. Villar, A. Caperos, and F. García-Ochoa. Oxidation of hard-wood kraft-lignin to phenolic derivatives with oxygen as oxidant.Wood Science and Technology, 35(3):245–255, 2001.

174 Bibliography

[10] T. Voitl. Oxidation of Lignin for the Production of Chemicals.PhD Thesis ETH No. 18780, ETH Zürich, 2009.

[11] S. K. Ritter. Lignocellulose: A complex biomaterial. Chemical &Engineering News Archive, 86(49):15, 2008.

[12] G. Gellerstedt and G. Henriksson. Lignins: Major sources, struc-ture and properties. In M. N. Belgacem and A. Gandini, editors,Monomers, Polymers and Composites from Renewable Resources,pages 201–224. Elsevier, Amsterdam, 2008.

[13] D. S. Argyropoulos and S. B. Menachem. In D. L. Kaplan, edi-tor, Biopolymers from Renewable Resources., page 292. Springer,Berlin,1998.

[14] K. Milner. Structure & function of the plant cell. http://www.

ehow.com/facts_5894544_structure-function-plant-cell.

html, accessed on September 11, 2012.

[15] S. Scott, M. Dagher, E. Myles, and A. Tillman. Identifying en-ergy needs & biomass resources within the state of Mississippi.Technical report, Alcorn State University, 2002.

[16] F. S. Chakar and A. J. Ragauskas. Review of current and fu-ture softwood kraft lignin process chemistry. Industrial Crops andProducts, 20(2):131–141, 2004.

[17] K. Freudenberg and A. C. Neish. Constitution and biosynthesis oflignin. Springer-Verlag, Berlin, Heidelberg, New York, 1968.

[18] E. Adler. Lignin chemistry - past, present and future. Wood Scienceand Technology, 11(3):169–218, 1977.

[19] J. Lora. Industrial commercial lignins: Sources, properties and ap-plications. In M. N. Belgacem and A. Gandini, editors, Monomers,Polymers and Composites from Renewable Resources, pages 225–241. Elsevier, Amsterdam, 2008.

[20] J. Gierer. Chemical aspects of kraft pulping. Wood Science andTechnology, 14(4):241–66, 1980.

http://www.ehow.com/facts_5894544_structure-function-plant-cell.html



Bibliography 175

[21] H. Sixta. Handbook of pulp. Wiley-VCH Verlag GmbH & Co.KGaA, Weinheim, 2006.

[22] R. J. A. Gosselink, E. de Jong, B. Guran, and A. Abächerli. Co-ordination network for lignin–standardisation, production and ap-plications adapted to market requirements (EUROLIGNIN). In-dustrial Crops and Products, 20(2):121–129, 2004.

[23] Prowledge Oy. Knowpulp - learning environment for chemicalpulping and automation. http://www.knowpulp.com, accessed onSeptember 13, 2012.

[24] F. Öhman. Precipitation and separation of lignin from kraft blackliquor. PhD thesis, 2006.

[25] M. R. Olsson, E. Axelsson, and T. Berntsson. Exporting lignin orpower from heat-integrated kraft pulp mills: A techno-economiccomparison using model mills. Nordic pulp & paper research jour-nal, 21(4):476–484, 2006.

[26] A.-S. Jönsson, A.-K. Nordin, and O. Wallberg. Concentration andpurification of lignin in hardwood kraft pulping liquor by ultra-filtration and nanofiltration. Chemical Engineering Research andDesign, 86(11):1271–1280, 2008.

[27] P. Tomani and P. Axegard. The ILI umbrella programme andother existing and new approaches in lignin research. In ILI 8thForum, pages 109–113. The International Lignin Institute, 2007.

[28] J. Ollila. Metso to deliver the first commercial LignoBoost plantto Domtar in North America, 2011. Press release on www.metso.

com on December 15, 2011.

[29] K. Toven and G. Gellerstedt. Structural changes of softwood kraftlignin in oxygen delignification and prebleaching. In 10th Inter-national Symposium on Wood and Pulping Chemistry, volume 2,pages 340–345, 1999.

[30] A. Jacobs and O. Dahlman. Absolute molar mass of lignins by sizeexclusion chromatography and MALDI-TOF mass spectroscopy.Nordic pulp & paper research journal, 15(2):121–127, 2000.

http://www.knowpulp.com

www.metso.com

www.metso.com

176 Bibliography

[31] N.-E. El Mansouri and J. Salvadó. Structural characterization oftechnical lignins for the production of adhesives: Application tolignosulfonate, kraft, soda-anthraquinone, organosolv and ethanolprocess lignins. Industrial Crops and Products, 24(1):8–16, 2006.

[32] J. D. Gargulak and S. E. Lebo. Commercial Use of Lignin-BasedMaterials, volume 742 of ACS Symposium Series, pages 304–320.American Chemical Society, 1999.

[33] E. K. Pye. Industrial lignin production and applications. InB. Kamm, P. Gruber, and M. Kamm, editors, Biorefineries-Industrial Processes and Products, pages 165–200. Wiley-VCH Ver-lag GmbH & Co. KGaA, Weinheim, 2006.

[34] R. J. A. Gosselink. Lignin as a renewable aromatic resource for thechemical industry. PhD Thesis - Proefschrift Wageningen, 2011.

[35] S. Fernando, S. Adhikari, C. Chandrapal, and N. Murali. Biore-fineries: Current status, challenges, and future direction. Energy& Fuels, 20(4):1727–1737, 2006.

[36] Y. H. Zhang. Reviving the carbohydrate economy via multi-product lignocellulose biorefineries. Journal of Industrial Microbi-ology & Biotechnology, 35(5):367–375, 2008.

[37] M. Galbe, G. Zacchi, and L. Olsson. Pretreatment of Lignocellu-losic Materials for Efficient Bioethanol Production Biofuels, vol-ume 108 of Advances in Biochemical Engineering/Biotechnology,pages 41–65. Springer Berlin / Heidelberg, 2007.

[38] Y. Sun and J. Cheng. Hydrolysis of lignocellulosic materials forethanol production: a review. Bioresource Technology, 83(1):1–11,2002.

[39] L. da Costa Sousa, S. P. S. Chundawat, V. Balan, and B. E. Dale.’cradle-to-grave’ assessment of existing lignocellulose pretreatmenttechnologies. Current Opinion in Biotechnology, 20(3):339–347,2009.

Bibliography 177

[40] IHS Chemical Sales. Lignosulfonates. Technical report,2009. http://chemical.ihs.com/nl/Public/2009Jan.pdf, ac-cessed on September 25, 2012.

[41] J. K. Fink. Cement additives. In Petroleum Engineer’s Guide toOil Field Chemicals and Fluids, pages 311–359. Gulf ProfessionalPublishing.

[42] J. E. Holladay, J. F. White, J. J. Bozell, and D. Johnson. Topvalue-added chemicals from biomass - Volume II – Results ofscreening for potential candidates from biorefinery lignin. Tech-nical report, 2007.

[43] R. A. Northey. Low-cost uses of lignin. In R. M. Rowell, T. P.Schultz, and R. Narayan, editors, Emerging Technologies for Mate-rials and Chemicals from Biomass, volume 476 of ACS SymposiumSeries, pages 146–175. American Chemical Society, WashingtonD.C., 1992.

[44] N. Eisenreich, W. Eckl, E. R. Inone, H. Nägele, and J. Pfitzer.Arboform R© - a thermoplastic made of renewable resources. InH. Reichl and H. Griese, editors, International Congress and Ex-hibition on Electronics Goes Green 2000+, pages 625–627. VDE -Verlag Gmbh.

[45] K. Freudenberg, W. Belz, and C. Niemann. Die aromatischeNatur des Lignins (10. Mitteilung über Lignin und Cellulose).Berichte der deutschen chemischen Gesellschaft (A and B Series),62(6):1554–1561, 1929.

[46] M. Singer. XXII. Beiträge zur näheren Kenntniss der Holzsubstanzund der verholzten Gewebe. Monatshefte für Chemie, 3(1):395–410, 1882.

[47] J. Harkin. Recent developments in lignin chemistry, volume 6 ofTopics in Current Chemistry, pages 101–158. Springer Berlin /Heidelberg, 1966.

[48] K. V. Sarkanen, C. H. Ludwig, and Editors. Lignins: Occurrenceand Formation, Structure, Chemical and Macromolecular Proper-ties, and Utilization. 1971.

http://chemical.ihs.com/nl/Public/2009Jan.pdf

178 Bibliography

[49] D. W. Goheen. Chemicals from lignin. In I. S. Goldstein, editor,Organic chemicals from Biomass, pages 143–162. CRC Press, BocaRaton, 1981.

[50] C. Amen-Chen, H. Pakdel, and C. Roy. Production of monomericphenols by thermochemical conversion of biomass: a review. Biore-source Technology, 79(3):277–299, 2001.

[51] F. J. Ruiz-Dueñas and A. T. Martínez. Microbial degradation oflignin: how a bulky recalcitrant polymer is efficiently recycled innature and how we can take advantage of this. Microbial Biotech-nology, 2(2):164–177, 2009.

[52] E. Dorrestijn, L. J. J. Laarhoven, I. W. C. E. Arends, and P. Mul-der. The occurrence and reactivity of phenoxyl linkages in ligninand low rank coal. Journal of Analytical and Applied Pyrolysis,54(1-2):153–192, 2000.

[53] Wahyudiono, M. Sasaki, and M. Goto. Conversion of biomassmodel compound under hydrothermal conditions using batch re-actor. Fuel, 88(9):1656–1664, 2009.

[54] D. Meier, R. Ante, and O. Faix. Catalytic hydropyrolysis of lignin:Influence of reaction conditions on the formation and compositionof liquid products. Bioresource Technology, 40(2):171–177, 1992.

[55] H.-R. Bjorsvik and F. Minisci. Fine chemicals from lignosulfonates.1. Synthesis of vanillin by oxidation of lignosulfonates. OrganicProcess Research & Development, 3(5):330–340, 1999.

[56] S. Ramachandra Rao and G. A. Ravishankar. Vanilla flavour: pro-duction by conventional and biotechnological routes. Journal of theScience of Food and Agriculture, 80(3):289–304, 2000.

[57] Triumph Venture Capital. Part 3-Aroma chemicals from petro-chemical feedstocks. Technical report, Fridge, Nedlac, SouthAfrica, 2004. http://www.nedlac.org.za/.

[58] D. Stewart. Lignin as a base material for materials applications:Chemistry, application and economics. Industrial Crops and Prod-ucts, 27(2):202–207, 2008.

http://www.nedlac.org.za/

Bibliography 179

[59] A. Ferraz, J. A. Souza, F. T. Silva, A. R. Goncalves, R. E. Bruns,A. R. Cotrim, and R. M. Wilkins. Controlled release of 2,4-Dfrom granule matrix formulations based on six lignins. Journal ofAgricultural and Food Chemistry, 45(3):1001–1005, 1997.

[60] S. H. Beis, S. Mukkamala, N. Hill, J. Joseph, C. Baker, B. Jensen,E. A. Stemmler, M. C. Wheeler, B. G. Frederick, A. v. Heiningen,A. G. Berg, and W. J. DeSisto. Fast pyrolysis of lignins. BioRe-sources, 5(3):1408–1424, 2010.

[61] F. Trachsel, C. Hutter, and P. Rudolf von Rohr. Transparent sili-con/glass microreactor for high-pressure and high-temperature re-actions. Chemical Engineering Journal, 135, Supplement 1(0):309–316, 2008.

[62] A. Günther and K. F. Jensen. Multiphase microfluidics: from flowcharacteristics to chemical and materials synthesis. Lab on a Chip,6(12):1487–1503, 2006.

[63] N. Assmann, S. Kaiser, and P. Rudolf von Rohr. Supercriticalextraction of vanillin in a microfluidic device. The Journal of Su-percritical Fluids, 67:149–154, 2012.

[64] C. Charcosset. Some perspectives. In Membrane Processes inBiotechnologies and Pharmaceutics, pages 295–321. Elsevier, Am-sterdam, 2012.

[65] E. Gibbins, M. D’ Antonio, D. Nair, L. S. White, L. M. Freitas dosSantos, I. F. J. Vankelecom, and A. G. Livingston. Observationson solvent flux and solute rejection across solvent resistant nanofil-tration membranes. Desalination, 147(1-3):307–313, 2002.

[66] S. Baumberger, A. Abaecherli, M. Fasching, G. Gellerstedt,R. Gosselink, B. Hortling, J. Li, B. Saake, and E. de Jong. Molarmass determination of lignins by size-exclusion chromatography:towards standardisation of the method. Holzforschung, 61(4):459–468, 2007.

[67] S. Y. Lin and C. W. Dence. Methods in lignin chemistry. Springerseries in wood chemistry. Springer-Verlag, Berlin, 1992.

180 Bibliography

[68] A. Gärtner, G. Gellerstedt, and T. Tamminen. Determination ofphenolic hydroxyl groups in residual lignin using a modified UV-method. Nordic Pulp and Paper Research Journal, 14(2):163–170,1999.

[69] N.-E. El Mansouri and J. Salvadó. Analytical methods for deter-mining functional groups in various technical lignins. IndustrialCrops and Products, 26(2):116–124, 2007.

[70] D. S. Argyropoulos. Quantitative 31P-NMR analysis of lignins, anew tool for the lignin chemist. Journal of Wood Chemistry andTechnology, 14(1):45–63, 1994.

[71] G. Zakis. Functional analysis of lignins and their derivatives.TAPPI Press, Atlanta, 1994.

[72] O. Faix. Fourier transform infrared spectroscopy of lignin in solidstate. In S. Y. Lin and C. W. Dence, editors, Methods in ligninchemistry, pages 83–109. Springer-Verlag, Berlin, 1992.

[73] O. Faix, B. Andersons, and G. Zakis. Determination of carbonylgroups of six round robin lignins by modified oximation and ftirspectroscopy. Holzforschung, 52(3):268–274, 1998.

[74] G. Vázquez, G. Antorrena, J. González, and S. Freire. FTIR, 1H-and 13C-NMR characterization of acetosolv-solubilized pine andeucalyptus lignins. Holzforschung, 51:158–166, 1997.

[75] T. Voitl and P. Rudolf von Rohr. Oxidation of lignin using aqueouspolyoxometalates in the presence of alcohols. ChemSusChem, 1(8-9):763–769, 2008.

[76] J. Gierer and N.-O. Nilvebrant. Studies on the degradation oflignins by oxygen in acidic media. Holzforschung, 48(1):51–58,1994.

[77] A. R. Gaspar, D. V. Evtuguin, and C. Pascoal Neto.Polyoxometalate-catalyzed oxygen delignification of kraft pulp: apilot-plant experience. Industrial & Engineering Chemistry Re-search, 43(24):7754–7761, 2004.

Bibliography 181

[78] I. A. Weinstock, K. E. Hammel, M. A. Moen, L. L. Lan-ducci, S. Ralph, C. E. Sullivan, and R. S. Reiner. Selectivetransition-metal catalysis of oxygen delignification using water-soluble salts of polyoxometalate (POM) anions. Part 2. Reactionsof α−[SiVW11O40]

5– with phenolic lignin-model compounds. Holz-forschung, 52(3):311–318, 1998.

[79] G. Wu, M. Heitz, and E. Chornet. Improved alkaline oxidation pro-cess for the production of aldehydes (vanillin and syringaldehyde)from steam-explosion hardwood lignin. Industrial & EngineeringChemistry Research, 33(3):718–723, 1994.

[80] W. Partenheimer. The aerobic oxidative cleavage of lignin toproduce hydroxyaromatic benzaldehydes and carboxylic acids viametal/bromide catalysts in acetic acid/water mixtures. AdvancedSynthesis & Catalysis, 351(3):456–466, 2009.

[81] A. R. Gaspar, J. A. F. Gamelas, D. V. Evtuguin, and C. Pas-coal Neto. Alternatives for lignocellulosic pulp delignification us-ing polyoxometalates and oxygen: a review. Green Chemistry,9(7):717–730, 2007.

[82] T. Voitl, M. V. Nagel, and P. Rudolf von Rohr. Analysis ofproducts from the oxidation of technical lignins by oxygen andH3PMo12O40 in water and aqueous methanol by size-exclusionchromatography. Holzforschung, 64(1):13–19, 2009.

[83] I. A. Pearl. Studies on lignin and related products. III. The ox-idation of sulfite spent liqour with cupric oxide under pressure.Journal of the American Chemical Society, 72(5):2309–2309, 1950.

[84] E. Profft and B. Smirnow. Zur Kenntnis des o-Vanillins. V. 5-Carboxyvanillin und Derivate. Journal für Praktische Chemie,28(5-6):225–235, 1965.

[85] B. Hjulström, S. Isaksson, and A. Hennius. Organic geochemicalevidence for pine tar production in middle eastern sweden duringthe roman iron age. Journal of Archaeological Science, 33(2):283–294, 2006.

182 Bibliography

[86] F. Shadkami and R. Helleur. Use of an injection portfor thermochemolysis-gas chromatography/mass spectrometry:Rapid profiling of biomaterials. Journal of Chromatography A,1216(31):5903–5910, 2009.

[87] I. Shulgin and E. Ruckenstein. Henry’s constant in mixed solventsfrom binary data. Industrial & Engineering Chemistry Research,41(6):1689–1694, 2002.

[88] R. M. Hann and G. C. Spencer. The preparation of chlorovanillinand some of its derivatives. Journal of the American ChemicalSociety, 49(2):535–537, 1927.

[89] K. F. Jensen. Microreaction engineering - is small better? Chem-ical Engineering Science, 56(2):293–303, 2001.

[90] S. Marre, A. Adamo, S. Basak, C. Aymonier, and K. F. Jensen.Design and packaging of microreactors for high pressure and hightemperature applications. Industrial & Engineering Chemistry Re-search, 49(22):11310–11320, 2010.

[91] F. G. Sales, L. C. A. Maranhão, N. M. Lima Filho, and C. A. M.Abreu. Experimental evaluation and continuous catalytic processfor fine aldehyde production from lignin. Chemical EngineeringScience, 62(18-20):5386–5391, 2007.

[92] J. Tokunaga. Solubilities of oxygen, nitrogen, and carbon dioxidein aqueous alcohol solutions. Journal of Chemical & EngineeringData, 20(1):41–46, 1975.

[93] T. Voitl and P. Rudolf von Rohr. Demonstration of a process forthe conversion of kraft lignin into vanillin and methyl vanillate byacidic oxidation in aqueous methanol. Industrial & EngineeringChemistry Research, 49(2):520–525, 2009.

[94] D. J. Gardner, T. P. Schultz, and G. D. McGinnis. The pyrolyticbehavior of selected lignin preparations. Journal of Wood Chem-istry and Technology, 5(1):85–110, 1985.

Bibliography 183

[95] V. Gomis, F. Ruiz, A. Marcilla, and M. d. C. Pascual. Equilibriumfor the ternary system water + sodium chloride + ethyl acetateat 30C. Journal of Chemical & Engineering Data, 38(4):589–590,1993.

[96] S. R. Patwardhan and D. P. Desphande. Extraction ofvanillin from its aqueous solution by some organic solvents.Petroleum&Chemical Industry Developments, 13(10):4, 1979.

[97] A. Klemola and J. Tuovinen. Method for the production of vanillin.(U.S. Patent 4,847,422), 1989.

[98] K. Stephan and F. Mayinger. Technische Thermodynamik - Grund-lagen und Anwendungen, volume 11. Springer, Berlin, 1977.

[99] J. H. Yoon, M. K. Chun, W. H. Hong, and H. Lee. High-pressurephase equilibria for carbon dioxide-methanol-water system: exper-imental data and critical evaluation of mixing rules. Industrial &Engineering Chemistry Research, 32(11):2881–2887, 1993.

[100] E. C. Makin. Purification of vanillin. (U.S. Patent 4,474,994),1984.

[101] N. Assmann, H. Werhan, A. Ladosz, and P. Rudolf von Rohr. Su-percritical extraction of lignin oxidation products in a microfluidicdevice. Chemical Engineering Science (in press), 2013.

[102] National Toxicology Program. Report on Carcinogens. U.S. De-partment of Health and Human Services, Research Triangle Park,NC, USA, 12 edition, 2011.

[103] A. Vigneault, D. K. Johnson, and E. Chornet. Base-catalyzed de-polymerization of lignin: Separation of monomers. The CanadianJournal of Chemical Engineering, 85(6):906–916, 2007.

[104] S. V. Satyanarayana, P. K. Bhattacharya, and S. De. Flux declineduring ultrafiltration of kraft black liquor using different flow mod-ules: a comparative study. Separation and Purification Technology,20(13):155–167, 2000.

184 Bibliography

[105] A. Toledano, A. García, I. Mondragon, and J. Labidi. Lignin sep-aration and fractionation by ultrafiltration. Separation and Purifi-cation Technology, 71(1):38–43, 2010.

[106] M. Zabková, E. A. Borges da Silva, and A. E. Rodrigues. Recoveryof vanillin from lignin/vanillin mixture by using tubular ceramicultrafiltration membranes. Journal of Membrane Science, 301(1-2):221–237, 2007.

[107] C. I. Koncsag and K. Kirwan. A membrane screening for the sep-aration/concentration of dilignols and trilignols from solvent ex-tracts. Separation and Purification Technology, 94:54–60, 2012.

[108] A. Boam and A. Nozari. Fine chemical: OSN - a lower energyalternative. Filtration & Separation, 43(3):46–48, 2006.

[109] P. Silva, L. G. Peeva, and A. G. Livingston. Nanofiltration in or-ganic solvents. Advanced Membrane Technology and Applications,pages 451–467, 2008.

[110] F. P. Cuperus and K. Ebert. Non-aqueous applications of nanofil-tration. In A. I. Schäfer, A. G. Fane, and T. D. Waite, editors,Nanofiltration - Principles and Applications, pages 521–536. Else-vier, Oxford, 2005.

[111] A. V. Volkov, G. A. Korneeva, and G. F. Tereshchenko. Organicsolvent nanofiltration: prospects and application. Russian Chem-ical Reviews, 77(11):983–993, 2008.

[112] A. G. Livingston, L. G. Peeva, and P. Silva. Organic solventnanofiltration. In S. P. Nunes and K.-V. Peinemann, editors, Mem-brane Technology, pages 203–228. Wiley-VCH Verlag GmbH & Co.KGaA, Weinheim, 2006.

[113] L. S. White and A. R. Nitsch. Solvent recovery from lube oil fil-trates with a polyimide membrane. Journal of Membrane Science,179(1-2):267–274, 2000.

[114] L. S. White. Transport properties of a polyimide solvent resistantnanofiltration membrane. Journal of Membrane Science, 205(1-2):191–202, 2002.

Bibliography 185

[115] J. G. Wijmans and R. W. Baker. The solution-diffusion model: areview. Journal of Membrane Science, 107(1-2):1–21, 1995.

[116] I. Sereewatthanawut, F. W. Lim, Y. S. Bhole, D. Ormerod, A. Hor-vath, A. T. Boam, and A. G. Livingston. Demonstration of molecu-lar purification in polar aprotic solvents by organic solvent nanofil-tration. Organic Process Research & Development, 14(3):600–611,2010.

[117] X. J. Yang, A. G. Livingston, and L. Freitas dos Santos. Experi-mental observations of nanofiltration with organic solvents. Jour-nal of Membrane Science, 190(1):45–55, 2001.

[118] D. Peshev, L. G. Peeva, G. Peev, I. I. R. Baptista, and A. T.Boam. Application of organic solvent nanofiltration for concentra-tion of antioxidant extracts of rosemary (rosmarinus officiallis l.).Chemical Engineering Research and Design, 89(3):318–327, 2011.

[119] M. Mulder, E. van Voorthuizen, and J. Peeters. Membrane charac-terization. In A. I. Schäfer, A. G. Fane, and T. D. Waite, editors,Nanofiltration - Principles and Applications, pages 89–118. Else-vier, Oxford, 2005.

[120] O. Töppel. Method of recovering vanillin. (U.S. Patent 2,745,796),1956.

[121] L. T. Sandborn, S. J. Richter, and H. G. Clemens. Process ofmaking vanillin. (U.S. Patent 2,057,117), 1936.

[122] L. T. Sandborn. Process of making vanillin. (U.S. Patent2,104,701), 1938.

[123] R. Servis. Process of making vanillin. (U.S. Patent 2,399,607),1946.

[124] C. C. Bryan. Vanillin purification by distillation. (U.S. Patent2,506,540), 1950.

[125] E. W. Schöffel. Vanillin purification. (U.S. Patent 3,049,566), 1962.

186 Bibliography

[126] K. G. Forss, E. T. Talka, and K. E. Fremer. Isolation of vanillinfrom alkaline oxidized spent sulfite liquor. Industrial & Engineer-ing Chemistry Product Research and Development, 25(1):103–108,1986.

[127] E. A. Borges da Silva, M. Zabková, J. D. Araújo, C. A. Cateto,M. F. Barreiro, M. N. Belgacem, and A. E. Rodrigues. An inte-grated process to produce vanillin and lignin-based polyurethanesfrom kraft lignin. Chemical Engineering Research and Design,87(9):1276–1292, 2009.

[128] Alibaba Group, 2012. www.alibaba.com, accessed on November16, 2012.

[129] National Aeronautics and Space Administration. Space shuttle useof propellants and fluids, 2001. http://www-pao.ksc.nasa.gov/kscpao/nasafact/pdf/ssp.pdf.

[130] G. D. Ulrich and P. T. Vasudevan. How to estimate utilitycosts, 2006. http://people.clarkson.edu/~wwilcox/Design/

utilcost.pdf.

[131] T. Leiknes. Wastewater treatment by membrane bioreactors. InE. Drioli and L. Giorno, editors, Membrane Operations - Innova-tive Separations and Transformations, pages 363–395. Wiley-VCHVerlag GmbH & Co. KGaA, Weinheim, 2009.

[132] G. Barbieri and F. Scura. Fundamental of chemical membranereactors. In E. Drioli and L. Giorno, editors, Membrane Opera-tions - Innovative Separations and Transformations, pages 287–308. Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim, 2009.

[133] F. Gallucci, L. Paturzo, and A. Basile. An experimental study ofCO2 hydrogenation into methanol involving a zeolite membranereactor. Chemical Engineering and Processing: Process Intensifi-cation, 43(8):1029–1036, 2004.

[134] J. G. Sanchez Marcano and T. T. Tsotsis. Economic and techni-cal feasibility issues of membrane reactor processes. In Catalytic

www.alibaba.com

http://www-pao.ksc.nasa.gov/kscpao/nasafact/pdf/ssp.pdf

http://www-pao.ksc.nasa.gov/kscpao/nasafact/pdf/ssp.pdf

http://people.clarkson.edu/~wwilcox/Design/utilcost.pdf

http://people.clarkson.edu/~wwilcox/Design/utilcost.pdf

Bibliography 187

Membranes and Membrane Reactors, pages 223–239. Wiley-VCHVerlag GmbH & Co. KGaA, Weinheim, 2004.

[135] N. Yan, C. Zhao, P. Dyson, C. Wang, L.-T. Liu, and Y. Kou.Selective degradation of wood lignin over noble-metal catalysts ina two-step process. ChemSusChem, 1(7):626–629, 2008.

[136] T. Yoshikawa, T. Yagi, S. Shinohara, T. Fukunaga, Y. Nakasaka,T. Tago, and T. Masuda. Production of phenols from lignin via de-polymerization and catalytic cracking. Fuel Processing Technology,2012.

[137] J. S. Shabtai, W. W. Zmierczak, and E. Chornet. Process forconversion of lignin to reformulated hydrocarbon gasoline. (U.S.Patent 5,959,167), 1999.

[138] C. Gasser, G. Hommes, A. Schäffer, and P. F.-X. Corvini. Multi-catalysis reactions: new prospects and challenges of biotechnol-ogy to valorize lignin. Applied Microbiology and Biotechnology,95(5):1115–1134, 2012.

[139] A. Hüttermann, C. Mai, and A. Kharazipour. Modification oflignin for the production of new compounded materials. AppliedMicrobiology and Biotechnology, 55(4):387–384, 2001.

[140] H. Morita. An alternative synthesis of 5-carboxyvanillin. CanadianJournal of Chemistry, 42(10):2362–2363, 1964.

List of publications 189

List of publications

Journal publications

1. J.R. Seay, H. Werhan, M.R. Eden, R.N. D’Alessandro, T. Thomas,H. Redlingshöfer, C. Weckbecker, K. Huthmacher. IntegratingLaboratory Experiments with Process Simulation for Reactor Op-timization. Computer Aided Chemical Engineering, 25:703-709,Paper 168 (CD Volume), 2008.

2. H. Werhan, J. Mora Mir, T. Voitl, and Ph. Rudolf von Rohr.Acidic oxidation of kraft lignin into aromatic monomers catalyzedby transition metal salts. Holzforschung, 65(5):703-709, 2011.

3. H. Werhan, A. Farshori, and Ph. Rudolf von Rohr. Separation oflignin oxidation products by organic solvent nanofiltration. Jour-nal of Membrane Science, 423-424:404-412, 2012.

4. N. Assmann, H. Werhan, A. Ładosz, and Ph. Rudolf von Rohr.Supercritical extraction of lignin oxidation products in a microflu-idic device. Chemical Engineering Science, in press, 2013.

Conference contributions

1. H. Werhan, T. Voitl, and Ph. Rudolf von Rohr. Catalytic oxida-tion of Kraft lignin by aqueous polyoxometalates. Italian meetingon lignocellulosic chemistry (ITALIC5), Varenna, 2009.

2. H. Werhan, T. Voitl, and Ph. Rudolf von Rohr. Depolymerizationof Kraft Lignin for the Production of Chemicals. R’09 Twin WorldCongress, Davos, 2009.

3. H. Werhan, J. Mora Mir, T. Pielhop, T. Voitl, and Ph. Rudolfvon Rohr. Acidic oxidation of kraft lignin to vanillin catalyzedby polyoxometalates and transition metal salts. 11th EuropeanWorkshop of Lignocellulosics and Pulp (EWLP 2010), Hamburg,2010.

4. H. Werhan, J. Mora Mir, T. Voitl, and Ph. Rudolf von Rohr.Production of vanillin and methyl vanillate by catalytic oxidation

190 List of publications

of kraft lignin. 7th European Congress of Chemical Engineering(ECCE-7), Prague, 2010.

5. H. Werhan, J. Mora Mir, and Ph. Rudolf von Rohr. Depolymer-ization of lignin to aromatic chemicals by acidic oxidation withmetal salt catalysts. 3rd Nordic Wood and Biorefinery Conference(NWBC 2011), Stockholm, 2011.

6. H. Werhan, and Ph. Rudolf von Rohr. Valorization of lignin byacidic oxidation with transition metal catalysts and fractionationof the products. 8th European Congress of Chemical Engineering(ECCE-8), Berlin, 2011.

7. H. Werhan, N. Assmann, and Ph. Rudolf von Rohr. Oxidationof Lignin in a continuous two-Phase Flow Silicon/Glass Microreac-tor. 12th International Conferences on Microreaction Technology(IMRET 12), Lyon, 2012.

8. H. Werhan, A. Farshori, and Ph. Rudolf von Rohr. Valorizationof Lignin by Fractionating its Oxidation Products using OrganicSolvent Nanofiltration (OSN). 12th European Workshop of Ligno-cellulosics and Pulp (EWLP 2012), Helsinki, 2012.

Curriculum Vitae 191

Curriculum Vitae

Holger Werhan

Date of birth: January 16, 1982

Place of birth: Stuttgart, Germany

Nationality: German

10/2008–04/2013 Doctoral studies at the Institute of Process Engineer-ing, ETH Zurich (Prof. Dr. Ph. Rudolf von Rohr)

10/2002–08/2008 Study of Process Engineering at theUniversity of Stuttgart(academic degree: Dipl. Ing.)

04/2007–09/2007 Internship at Degussa AG, Mobile, AL, USA

07/2002–08/2002 Internship at Eberspächer GmbH, Esslingen a.N.

07/2001–05/2002 Civil service

1992–2001 Otto-Hahn-Gymnasium, Ostfildern(academic degree: Abitur)

1988–1992 Primary school, Ostfildern

Zurich, April 2013

Rights / License: Research Collection In Copyright - Non ......The best wishes for Thomas Pielhop in...

Documents

Transcript of Rights / License: Research Collection In Copyright - Non ......The best wishes for Thomas Pielhop in...