nor

aeF

ayse

lO

ses

sec

orP

gni

plu

P e

nilak

lA

fo

gnil

edo

M d

esa

B-a

ne

mo

neh

P y

tisr

evi

nU

otla

A

1202

smetsysoiB dna stcudorpoiB fo tnemtrapeD

desaB-anemonehPenilaklAfogniledoM

sessecorPgnipluP

noraeFayselO

LAROTCODSNOITATRESSID

seires noitacilbup ytisrevinU otlaASNOITATRESSID LAROTCOD 02 / 1202

fo gniledoM desaB-anemonehP sessecorP gnipluP enilaklA

noraeF ayselO

fo rotcoD fo eerged eht rof detelpmoc noitatressid larotcod A eht fo noissimrep eht htiw ,dednefed eb ot )ygolonhceT( ecneicS

cilbup a ta ,gnireenignE lacimehC fo loohcS ytisrevinU otlaA no loohcs eht fo )2oiB( oklaH moor gniteem eht ta dleh noitanimaxe

.00.21 ta 1202 hcraM 32

ytisrevinU otlaA gnireenignE lacimehC fo loohcS

smetsysoiB dna stcudorpoiB fo tnemtrapeD puorg yrtsimehC dooW

Printed matter4041-0619

NO

RDIC

SWAN ECOLABE

L

Printed matter1234 5678

rosseforp gnisivrepuS dnalniF ,ytisrevinU otlaA ,nenirouV inapaT rosseforP

rosivda sisehT

dnalniF ,ytisrevinU otlaA ,nenuttuuR itsöyK rotcoD

srenimaxe yranimilerP adanaC ,kciwsnurB weN fo ytisrevinU ,iN oahgnoY rosseforP

ytisrevinU etatS CN ,secruoseR larutaN fo egelloC ,leemaJ nasaH rosseforP

tnenoppO ecnarF ,arogaP -PNI elbonerG etisrevinU ,ahtroM.M draréG rosseforP

seires noitacilbup ytisrevinU otlaASNOITATRESSID LAROTCOD 02 / 1202

© 1202 noraeF ayselO

NBSI 7-2720-46-259-879 )detnirp( NBSI 4-3720-46-259-879 )fdp( NSSI 4394-9971 )detnirp( NSSI 2494-9971 )fdp(

:NBSI:NRU/if.nru//:ptth 4-3720-46-259-879

yO aifarginU iknisleH 1202

dnalniF

tcartsbA otlaA 67000-IF ,00011 xoB .O.P ,ytisrevinU otlaA if.otlaa.www

rohtuA noraeF ayselO

noitatressid larotcod eht fo emaN sessecorP gnipluP enilaklA fo gniledoM desaB-anemonehP

rehsilbuP gnireenignE lacimehC fo loohcS

tinU smetsysoiB dna stcudorpoiB fo tnemtrapeD

seireS seires noitacilbup ytisrevinU otlaA SNOITATRESSID LAROTCOD 02 / 1202

hcraeser fo dleiF yrtsimehC dooW

dettimbus tpircsunaM 0202 rebotcO 13 ecnefed eht fo etaD 1202 hcraM 32

)etad( detnarg ecnefed cilbup rof noissimreP 1202 yraunaJ 72 egaugnaL hsilgnE

hpargonoM noitatressid elcitrA noitatressid yassE

tcartsbAtfark rof ledom level-ralucelom desab-anemonehp a gnipoleved no sesucof siseht larotcod sihT ,scitenik noitcaer no egdelwonk latnemadnuf gnizilitu yb sessecorp gniplup enilakla rehto dna si deilppa hcaorppa latnemadnuf eht morf tfieneb rojam A .refsnart ssam dna scimanydomreht

eht fo gnilledoM .tnemnorivne noitcaer eht fo sseldrager dilav ,elpicnirp ni ,si ledom eht taht fo ecnatropmi evitaler eht enimaxe ot yaw wen a sedivorp level ralucelom ta ssecorp gniplup

wen gnitset dna gnipoleved rof mroftalp a edivorp dna seiroeht fo ytidilav eht ,anemonehp suoirav ot yaw tneicfife-tsoc erom a eb nac noitalumis ,hcraeser latnemirepxe htiw nosirapmoc nI .saedi

ot dna yldneirf yllatnemnorivne dna tneicfife erom meht ekam ot ,sessecorp knihter dna yduts .tsoc laminim htiw stcudorp ytilauq eht evorpmi

gnidulcni ,ledom yrtsimehc gniplup evisneherpmoc a etaerc ot saw krow siseht siht fo evitcejbo ehT

ehT .sessecorp gniplup enilakla rehto dna tfark ni noitacfiingiled dna noitadarged etardyhobrac htiw deweiver saw erutaretil ,tsriF .snrut ni devlos erew smelborp dna strap otni dedivid saw krow eht fo sretemarap citenik eht dna smsinahcem noitcaer eht no atad yrassecen eht dnfi ot tegrat a selihpoelcun gnorts fo scitenik noitcaer eht no egdelwonk gnissim eht ,yldnoceS .devlovni snoitcaer

hguorht yllatnemirepxe deniatbo saw ytilanoitcnuf lynobrac-α htiw ningil cilonehp-non htiw delledom erew setardyhobrac dna ningil ,rehtruF .enorelda dnuopmoc ledom ciremid eht gniylppa ehT .scitsiretcarahc lacimehc denimreted yllatnemirepxe htiw sdnuopmoc-oduesp ciremonom sa

.noitulossid ningil ni deredisnoc saw ningil fo tcepsa ralucelomorcam

dna lyxordyh fo snoitcaer eht fo yrarbil evisnetxe na fo demrof si depoleved ledom gniplup ehT eht gnitalumis nehW .serutcurts etardyhobrac dna ningil-oduesp htiw snoi edfilus negordyh

eht fo snoitartnecnoc tnedneped-emit fo stsisnoc emoctuo eht ,snoitidnoc nevig rednu gniplup hcus ,sretemarap gnireenigne emos ,yllanoitiddA .sdnuopmoc lacimehc laer dna serutcurts-oduesp

meht sevired taht ledom eht ni dedulcni erew ,ssenthgirb dna ytisocsiv cisnirtni ,rebmun appak sa sa atad lacitylana yb detadilav eb dluoc ledom eht ,yltneuqesnoC .noitisopmoc lacimehc eht morf

tsniaga derapmoc saw noitciderp ledom ehT .seulav sretemarap gnireenigne yb sa llew dedulcni tceffe nannoD ehT .noitalerroc doog a yllareneg detartsnomed dna stluser latnemirepxe .yltcerroc metsys eht ni dica cinorunexeh fo tnetnoc eht gnitciderp rof tnatropmi saw ledom eht ni ledom eht evorpmi ylbissop dluoc ningil fo tcepsa lacimehcoerets dna cimanydomreht fo noisulcnI

.rehtruf noitciderp

sdrowyeK msinahcem noitcaer ,scitenik noitcaer ,gniledom ,gniplup enilakla

)detnirp( NBSI 7-2720-46-259-879 )fdp( NBSI 4-3720-46-259-879

)detnirp( NSSI 4394-9971 )fdp( NSSI 2494-9971

rehsilbup fo noitacoL iknisleH gnitnirp fo noitacoL iknisleH raeY 1202

segaP 011 nru :NBSI:NRU/fi.nru//:ptth 4-3720-46-259-879

i

Acknowledgments

This thesis work was carried out at the Aalto University, School of Chemical En-

gineering, Department of Bioproducts and Biosystems. Funding for the research

came from Effibre VIC project and SEDEL project. Additionally, I would like to

thank Walter Ahlström Foundation and School of Chemical Engineering for

providing me grants.

My deepest gratitude goes to Professor Tapani Vuorinen for the privilege to

work as a doctoral candidate in his research group. Thank you, Tapani for eve-

rything you have taught me during these years. I would like to thank Dr Susanna

Kuitunen for introducing me to the modelling universe and her patience in guid-

ing, helping and advising me. I also want to say thanks to Professor Ville Alo-

paeus. I am grateful for the opportunity to work with Vesa Nykänen and Profes-

sor Raimo Alen.

I would like to thank Dr Markus Paananen for sharing his experimental data for

our model. I really appreciate Dr Kyösti Ruuttunen support and his advices at a

very important stage of my PhD journey.

I wish to warmly thank my opponent, Professor Gérard Mortha, for accepting

the invitation and using time evaluate my work. Big thank goes to my pre-ex-

aminers, Professor Yonghao Ni and Professor Hasan Jameel, for carefully ex-

amining my manuscript and helping me to improve the thesis.

I thank all my colleagues from the department whom I met during this long

journey. The page is not enough to mention all names of the wonderful people

whom I met during these years. Thank each of you, for our coffee breaks, thank

you for our quick chats, and thank you for the atmosphere you have been keep-

ing at the department. Christian, Naveen and Zhen - my dearest officemates -

ii

we are Fantastic Four! I am so happy that I have you around me for those years!

Sara, Iina and Jonna, I think words are unnecessary, you are my best! Special

thanks for our “what's up” group where you are supporting me through this

crazy corona time. I would like to thank all Puu staff for providing a friendly and

inspiring atmosphere. Special thanks go to Pentti Risku and Harri Koskinen for

helping me with computer issues and keeping my modelling safe.

Besides above-mentioned colleagues and friends, I would like to mention my

old friends Julia and Dasha. Thank you for who you are! Special thanks go to

the people who have been with me in the most difficult time of my life: Dasha,

Iliana&Martin, Julia, Jonna, Emmi and Naveen, simply thank you!

I express my heartfelt gratitude to my parents, especially to my mother. I am

infinitely thankful to my lovely kids. Daniel constantly broadens my horizons by

asking questions, for the answers to which I need to learn new things. Iiris

demonstrated to me that "impossible" is just a word. Finally, the deepest thanks

go to my husband Aleksei for understanding, encouragement and endless sup-

port during good and bad days.

Espoo, 7th of February 2021

Olesya Fearon

iii

Contents

1. Introduction ................................................................................... 1

2. Backround ..................................................................................... 5

2.2 Pulping models ........................................................................ 12

2.2.1 Phase based models Pulping models ................................... 13

2.2.2 Reaction mechanism based models ..................................... 14

2.2.3 Continuous distribution of reactivity models ...................... 15

2.2.4 Continuous digester models ................................................. 15

3. Experimental ............................................................................... 19

3.1 Chemicals ................................................................................. 19

3.2 Experiment procedure ............................................................. 19

3.3 Analytical Methods for publication 1 ...................................... 20

3.3.1 Gas Chromatography (GC) with Flame-Ionization Detection (FID) ............................................................................................. 20

3.3.2 MS instrument applied: ...................................................... 20

4. Methodology ................................................................................ 21

4.1 FLOWBAT ................................................................................ 21

4.2 Model ....................................................................................... 21

4.2.1 Lignin .................................................................................. 24

4.2.2 Carbohydrates ..................................................................... 26

4.2.3 Kraft Pulping model ............................................................ 26

5. Results and discussions ............................................................... 31

5.1 Reaction of hydroxide and hydrogen sulfide ions with adlerone and its kinetic parameters. (Publication 1) ................................................ 31

5.1.1 Delignification model (Publication 2) .................................... 36

5.1.2 Carbohydrate degradation model (Publication 3) .............. 42

5.1.3 Model applicability .............................................................. 46

6. Conclusions and Future Outlook ................................................ 49

References.............................................................................................. 53

iv

List of Publications

This doctoral dissertation consists of a summary and of the following publica-

tions which are referred to in the text by their numerals

1. Olesya Fearon; Susanna Kuitunen; Tapani Vuorinen (2016) Reaction kinet-

ics of strong nucleophiles with a dimetric non-phenolic lignin model com-

pound with α-carbonyl functionality (adlerone) in aqueous alkali solution.

Holzforschung, volume 70, issue 9, pages 811-818. DOI 10.1515/hf-2015-0236

2. Olesya Fearon; Susanna Kuitunen; Kyösti Ruuttunen; Ville Alopaeus; Ta-

pani Vuorinen (2020) Detailed Modeling of Kraft Pulping Chemistry. Deligni-

fication. Industrial&Engineering Chemistry Research, volume 59, issue29,

pages 12977-12985. Doi.org/10.1021/acs.iecr.0c02110

3. Olesya Fearon; Vesa Nykänen; Susanna Kuitunen; Kyösti Ruuttunen;

Raimo Alén; Ville Alopaeus; Tapani Vuorinen (2020) Detailed modeling of the

kraft pulping chemistry: carbohydrate reactions. AIChE Journal, volume 66,

issue 8, pages 1-9. Doi.org/10.1002/aic.16252

v

Author’s Contribution

Publication 1: Reaction kinetics of strong nucleophiles with a dimetric non-

phenolic lignin model compound with α-carbonyl functionality (adlerone) in

aqeous alkali solution.

OF was responsible for the research plan, experimental design and performing

analyses under the supervision of TV. Designing the model was conducted

with great support and supervision of SK. The manuscript was prepared by OF

while SK and TV reviewed and commented the manuscript.

Publication 2: Detailed modeling of kraft pulping chemistry. Delignification.

OF was responsible for the development of the reaction scheme and perform-

ing the parameter optimization with the great support of SK in modelling is-

sues and TV from chemistry point of view. OF analysed results and prepared

the manuscript that was reviewed by SK, VA, KR and TV.

Publication 3: Detailed modeling of kraft pulping chemistry: carbohydrate

reactions.

OF and VN developed reaction scheme and conduct parameter optimization

under great support of SK. OF has analysed results and prepared the manu-

script. SK, RA, VA, KR and TV critically reviewed the analysis and writing.

vi

vii

List of Abbreviations and Symbols

A frequency factor (parameter in Arrhenius rate equation)

AGU anhydroglucose unit

CM non-terminal groups in cellulose

CNR cellulose non-reducing end-group

COMA comanic acid

CR cellulose reducing end-group

Ea activation energy

EF effective alkali

FSP fiber saturation point

GGM galactoglucomannan

GGMR galactoglucomannan reducing end-group

GGMM non-terminal groups in galactoglucomannan

GGMNR galactoglucomannan non-reducing end-group

HexA hexenuronic acid

k reaction rate constant

K equilibrium constant

LCC lignin carbohydrate complex

L:W liquid to wood ratio

MeGlcA methylglucuronic acid

pKa negative logarithm of acid dissociation constant

QM quinone methide

SSE sum of squared errors of prediction

viii

t time

T temperature

XYL xylan

XYLR xylan reducing end-group

XYLNR xylan with non-reducing end-group

XYLM non-terminal groups in xylan

XyloIsoSaccA xyloisosaccharinic acid

XyloMetaSaccA xylometasaccharinic acid

1

1. Introduction

Wood and lignocellulosic materials have a significant role in our lives. People

have used wood throughout history – for example, in the building and heating

of houses, as well as producing furniture and household goods. Population

growth correlates with increasing usage of lignocellulose materials and wood.

With new developments in wood chemistry, wood will continue to play an ever-

increasing key role in fulfilling human needs.

Wood contains cellulose, hemicellulose, lignin and extractives. Lignin (about

18-35%) and carbohydrates (65-75%, whereof 35-55% is cellulose and 20-35%

are hemicelluloses) are polymeric materials contained within wood. 1 Extrac-

tives and inorganic materials (ash) account for 4-10% of wood dry matter. 1 The

elemental composition of wood is around 50% carbon, 6 % hydrogen and 44 %

oxygen, as well as trace amounts of several metal ions. 2 The elemental compo-

sition of wood tells us already then that it is a very promising source of carbon-

based products.

For many years, wood has been used for paper, tissue and cardboard produc-

tion. 3 Nowadays, pulp mills are becoming biorefineries, where beside the main

cellulose-based product, fuel, heat, chemicals and other valuable products are

manufactured. 4

The deeper the chemical composition of wood is studied, the more promising

the applications developed are. It is possible to say that wood has a direct effect

on the quality of human living. For example, cellulose is used in medicine as a

drug delivery material, in aggregates and hydrogels, as microcapsules and na-

noparticles and as membranes, as well as in electro active materials. 5 Bacterial

cellulose is used in skin repair treatment. 6 Nano-sized cellulose shows a supe-

rior adsorption capacity for heavy metals and dyes. 7 Lignin-carbohydrate com-

plexes (LCCs), formed in pulping process, also could be used for human needs,

2

with it having been reported that LCCs can be successfully utilized as a precur-

sor for the production of spherical biocarriers. 8

Lignin is also a key component of wood and has variety of applications. Various

industries have a high interest in the conversion of lignin into products includ-

ing: chelating agents, antioxidants, pesticides, vanillin and other phenols, ferti-

lizers, adhesives, and formaldehyde-free phenolic resin. 9-11 Lignin micro- and

nanoparticles synthesized from side-streams of the pulp industry could be uti-

lized as components in composite nanofillers, solid foams, emulsion stabilizers

or in UV protectors. 12 Hence it has been demonstrated that lignin is a very

promising source for different valuable products. Consequently, for future

ideas, it is important to know how different lignin structures are behaving.

The short overview presented above, about the components of wood and their

possible applications, shows that the right combination of research and practice

can lead to the production of novel materials and products, with many of these

applications aiming towards a green and sustainable future. However, as ra-

tional design is the key in any process, detailed understanding of the involved

process chemistry and the key parameters is imperative. Since the processes be-

come more complex, it is very complicated to carry out everything in the labor-

atory, and therefore there is a clear need for detailed process models. Simulation

tools are excellent and cheap choices for developing, scaling up, and optimizing

new ideas. 3

Previously, many different pulping models have been developed; however, most

of them do not cover the detailed chemistry of the process, especially for lignin.

Mainly, the existing models are based on empirical correlations to predict ki-

netic parameters in the kraft pulping process. 3,13-28 Those correlations are usu-

ally simple – it is fast and easy to use the models, however they rarely have phys-

ical meaning. Often, the parameter values are adjusted to a concrete experi-

mental setup and, consequently, the models are limited only to certain types of

pulps and/or experimental conditions. The kraft pulping process is a chemical

reaction system of wood chips and fibres, and these kind of reaction systems

could be modelled as phenomena-based models, which are grounded on chem-

ical reactions, mass transfer and thermodynamics. These types of models give

freedom of system composition. It is beneficial to have a kraft pulping model

based on deep knowledge of the reaction mechanisms that is independent from

external constraints. The phenomena-based model can reproduce experimental

3

data and predict the results of kraft pulping. In the current industrial situation,

where pulp is only one of the products, there is a clear need for a detailed mo-

lecular level-based physico-chemical model. These needs are the objective of

the research presented in this thesis; namely creating a comprehensive

model, which covers the whole chemistry of the pulping process, including car-

bohydrate degradation and delignification in kraft pulping.

Phenomenon-based modelling also helps in performing straightforward studies

of reaction mechanisms, as chemical reactions are modelled on a molecular

scale. The modelling approach, presented in the current research, has been ap-

plied earlier to the modelling of chlorine dioxide bleaching chemistry, as pre-

sented in dissertations by Ville Tarvo 29 and Tuula Lehtimaa 30, as well as in hot-

water extraction, as described in Susanna Kuitunen’s dissertation 31.

This work focuses on developing a phenomena-based molecular-level model for

kraft and other alkaline pulping processes. The model includes an extensive li-

brary of the reactions of hydroxide and sulfide ions with pseudo-lignin and car-

bohydrate structures. During this work, literature was examined, and the infor-

mation found was implemented in the model. If data was unavailable, inde-

pendent experimental work was performed. This thesis compromises of three

publications, attempting together to build the kraft pulping process model.

Publication 1 gives answers to the missing reaction mechanisms and kinetic

parameters in pulping, namely the reactions of adlerone or carbonyl containing

non-phenolic lignin units. Based on literature data, as well as the data from

Publication 1, a delignification model was developed and presented in Publi-

cation 2. Publication 3 covers the carbohydrate degradation model within

kraft pulping. Based on Publications 1, 2 and 3, a comprehensive kraft pulp-

ing model was developed.

4

5

2. Background

Wood is natural composite material containing lignin, cellulose, hemicelluloses

and extractives. Figure 1 represents a schematic structure of wood fibre. Cellu-

lose microfibrils form the skeleton; hemicelluloses are the matrix substance pre-

sent between the cellulose microfibrils and lignin, which acts to solidify the cell

wall via association with the matrix substances. 32 Softwood hemicelluloses con-

sist mainly of galactoglucomannan and arabino-4-O-methylglucoronarabinoxy-

lan.

Figure 1. Structure of wood fibre represented with cellulose, hemicellulose and

lignin. Adopted and modified from 33.

Lignin is the most abundant natural aromatic polymer and the second most

abundant (after cellulose) natural organic polymer on Earth. 34 The lignin mac-

romolecules contain different functional groups, including phenolic hydroxyl,

aliphatic hydroxyl, methoxyl, and carbonyl groups. Lignin is a polymeric natu-

ral product built from an enzyme-initiated dehydrogenative polymerization of

6

three primary precursor alcohols: trans-coniferyl, trans-sinapyl and trans-p-

coumaryl. 35 In Figure 2, the structure of lignin is presented. (Ralph et al. 2019.)

Figure 2. (a) Proposed lignin structure with several linkages; (b) the three

primary precursor alcohols of lignin. The figure is modified and reproduced

from 36.

The presented lignin precursors are interlinked by various bonds, forming the

highly complex structure of lignin. The exact structure of lignin is still unknown,

however it has been determined that around two thirds of the overall lignin link-

ages are ether type, with the rest being carbon-carbon bonds. (Hatakeyama and

Hatakeyama, 2009) A lot of research is still ongoing in order to determine the

structure of lignin and to find the best ways to treat it.

Cellulose is a linear polymer built from anhydro- -D-glucopyranosyl units

(AGU) and linked by (1→4)- glycosidic bonds. The cellulose polymer has the

formula (C6H10O5)n and the number of AGUs can be from hundreds to thou-

sands. 1 The detailed structure of cellulose is presented in Figure 3.

7

Figure 3. Structure of cellulose (degree of polymerization, DP = n).

Galactoglucomannan (GGM) is a linear molecule (Figure 4). The GGM back-

bone chain is built up from -D-glucopyranosyl and -D-mannopyranosyl resi-

dues, which are joined together by (1 → 4)-linkages. 37 The ratio of mannosyl to

glucosyl residues in the backbone (Man/Glc ratio) is usually around three. In its

native state, softwood galactoglucomannan is acetylated. 37

Figure 4. Structure of galactoglucomannan.

Arabino-4-O-methylglucuronoxylan is a linear polymer with a backbone

built up of -D-xylopyranosyl (a pentose) residues joined together by (1 → 4)-

linkages. 37 The arabino-4-O-methylglucuronoxylan has two type of side-chains:

4-O-methyl-D-glucuronic acid in the -pyranose ring form, (1 → 2)-linked to

the backbone xylose and L-arabinose in the -furanose ring form (1 → 3)-linked

to the backbone xylose (Figure 5). The ratio of 4-O-methylglucuronic acid to xy-

lose is about 1:5–10, and the ratio of arabinose to xylose varies substantially.

Figure 5. Structure of arabino-4-O-methylglucuronoxylan.

8

2.1 Kraft pulping

The pulp and paper industry represents two-thirds of the total production value

in forest industries.38 Consequently, the success of the pulp and paper industry

is important for the Finnish economy.

Within industry, a deep knowledge of pulping chemistry is needed, in order to

develop the processes further. The kraft pulping process is the globally domi-

nant pulping process nowadays 4, and it is therefore very important to take it

under consideration first.

The popularity of the kraft pulping process is due to the high strength properties

of the produced pulp, relatively easy recovery of the cooking chemicals and the

usage of dissolved wood material for energy production. 39,40 The disadvantages

of the kraft pulping process however include the dark colour of unbleached pulp,

the loss of pulp yield due to carbohydrate degradation and solubilization, as well

as the formation of odorous compounds. 40

The target of any pulping process is to dissolve lignin, while minimising cellu-

lose and hemicellulose degradation. In kraft pulping, wood fibres are treated

with white liquor solution which is a mix of sodium hydroxide and sodium sul-

fide, at a high temperature and pressure. Generally, the kraft pulping process is

divided into three phases: initial, bulk and residual delignification. Each phase

has a different selectivity with respect to carbohydrate losses. 41

Efficient and even transportation of the kraft pulping chemicals, hydrogen sul-

phide and hydroxide ions, to all parts of wood chips is recognized as one of key

factors of the initial phase and the whole pulping process. Mainly two phe-

nomena, i.e. liquor penetration and chemical diffusion, are responsible for the

transfer of HS- and HO- into the chips. Liquor penetration is the flow of the

chemical solution into the air-filled voids of wood chips under hydrostatic

pressure.40 The further transfer of the active chemicals from the lumina into

the cell wall occurs through diffusion. Wood density and morphology are fac-

tors that have influence on the rate of diffusion.40 The fast consumption of al-

kali in deacetylation of hemicelluloses emphasizes the need for effective pene-

tration and diffusion.

Time and temperature correlations have an effect on each phase, with an in-

creasing temperature resulting in faster pulping. Around 15-20% of the lignin

and 20-25% of the carbohydrates present in wood are dissolved at this stage. In

9

the initial phase, reactions of phenolic lignin structures take place, primarily in-

cluding the formation of quinone methide (QM) (Figure 6,a), followed by its

reaction with hydrogen sulfide ions (HS-), leading to the formation of thiol

structures (Figure 6, S8) and resulting in the cleavage of β-ether bonds. Addi-

tionally, competing reactions such as the ionization of γ-hydroxyl groups and

subsequent elimination of formaldehyde (intramolecular reaction) (Figure 6,

S5), or proton elimination by a hydroxyl ion (bimolecular reaction) (Figure 6,

S14), resulting in the formation of enol ether structures. 41 Moreover, the reac-

tions of lignin units containing carbonyl groups also proceed in the initial phase,

resulting in β-ether bond cleavage (Figure 6, c). 42 Formation of chromophores

also take place in the initial phase.

Figure 6. The most important lignin reactions during the delignification pro-

cess in kraft pulping, which are included in the kraft pulping model, where R,

R1, R2 denote alkyl group or aryl groups or a proton. a) Reaction of phenolic

lignin structures; b) non-phenolic lignin reactions with free -hydroxyl group;

c) reaction of lignin units containing carbonyl groups and d) demethylation re-

actions.

Carbohydrates undergo rapid deacetylation (of galactoglucomannan in soft-

wood and xylan in hardwood) during the initial phase. The deacetylation pro-

cess is characterized by a high consumption of alkali.1 As was illustrated in Fig-

ures 3-5, cellulose, galactoglucomannan (GGM) and xylan contain reducing and

non-reducing end groups. Reducing end groups are actively peeled off from the

10

polysaccharide chains, with peeling being either primary or secondary. Primary

peeling takes place already at the heating period in the initial stage of pulping.

Additionally however, new reducing end groups are forming, due to the random

cleavage of the glycosidic bonds in the middle of the polysaccharide chains via

alkaline hydrolysis. This reaction takes place mainly in the bulk and residual

delignification phases and the formed new reducing end groups enable second-

ary peeling. 1 The term peeling refers to an endwise degradation process where

a reducing end group is split off from the cellulose chain, resulting in soluble

degradation products, such as isosaccharinic acids, and liberation of a new re-

ducing end group (Figure 7). The cellulose molecule shortens when glucose

units continue to peeled off until a stopping reaction occurs. In this competing

reaction the reducing end group is converted into an alkali stable non-reducing

carboxylic acid end group (Figure 8). Additionally, a physical stopping may take

place when the reaction front reaches a crystalline region that is inaccessible for

alkali and chemical reaction. The amount of cellulose that is degraded in the

pulping process depends on the average amount of reducing end groups in a

cellulose molecule and its degree of polymerization. Originally, a cellulose mol-

ecule, containing n glucose units (degree of polymerization, DP = n), has one

reducing end group. Consequently, the amount of isosaccharinic acids formed

and the amount of cellulose degraded by the peeling reaction depends strongly

on the degree of polymerization of cellulose. 43

Figure 7. Mechanism for the peeling reaction in kraft and soda pulping. 43

11

Figure 8. Mechanism for the β-elimination in polysaccharide end groups, re-

sulting in stopping reactions and the formation of metasaccharinic acid. RO is

an arabinose unit. 43

The bulk phase of softwood kraft pulping is characterized by heating from 150

to 170 0C and cooking at 170 0C. The rate of delignification here is no longer

diffusion controlled, but is instead controlled by the chemical reactions them-

selves, and most of the lignin is removed in this phase. 1 Lignin degradation pro-

ceeds through the cleavage of ether bonds (Figure 6, b) and demethoxylation

(Figure 6, d). Condensation reactions, resulting in the formation of new carbon-

carbon linkages and chromophores, are generally undesirable lignin reactions

in the bulk phase. 42 In the bulk phase, polysaccharides undergo degradation

reactions in the form of alkali-catalysed glycoside hydrolysis (Figure 9) and sub-

sequent secondary peeling and stopping reactions (Figures 7 and 8). Because

the random hydrolytic chain scission lowers DP of cellulose significantly and

forms new reducing end groups accessible to secondary peeling, an undesired

yield loss of cellulose occurs. Formation of hexenuronic acid (HexA) groups

from the 4-O-methylglucuronic acid groups of xylan (Figure 10) by elimination

of methanol, as well as readsorption of xylan on fibres, also take place in the

bulk phase of kraft pulping. 44

12

Figure 9. Mechanism for the alkaline hydrolysis of glycosidic linkages via

epoxide formation. 43

Figure 10. The elimination of methanol from 4-0-methylglucuronic acid

groups in xylan during kraft and soda pulping. 43

The residual phase of kraft pulping is characterized by yield loss of carbohy-

drates due to alkaline hydrolysis (Figure 9) and peeling reactions (Figure 7), as

well as the dissolution of hemicelluloses. The low reactivity of residual lignin

could be explained by the presence of primarily carbon-carbon bonds, which are

hard to cleave. 41 Partly, the formation of lignin carbohydrate complexes (LCC)

could also explain the difficulties in the removal of residual lignin. 45,46

As described, the kraft pulping process includes many chemical reactions, which

determine the final outcome of the process. In order to predict the pulping pro-

cess results, researchers have therefore developed pulping models, some of

which are taken into consideration below.

2.2 Pulping models

Different models have been developed both for delignification and for carbohy-

drate degradation in the pulping process. The models are necessary in order to

analyse, develop, and control pulping processes. Initially, only pulp yield,

strength and brightness were of interest for the pulping industry. Consequently,

the first developed models were aiming to control lignin content – indicated by

the kappa number – and the carbohydrate yield. Several models for delignifica-

tion and carbohydrate degradation were developed. 3,13-20,22,24-28,47-51 A short

overview of some of these models is presented below. A detailed summary on

delignification models is presented in Publication 2. In work 52 the models for

carbohydrates were divided to three types: phase-based models, reaction mech-

13

anism-based models and continuous distribution of reactivity models. The divi-

sion is very logical and clear, and therefore it is applied in the current work to

describe different kraft pulping models (both lignin and carbohydrates).

2.2.1 Phase based models Pulping models

Phase models describe the joint effect of all factors in pulping on the kinetic

behaviour, but do not take into consideration mass transfer or the dissolution

of degraded products. 52 The pulping process is described by a few simple ki-

netic equations. The first kraft pulping model was presented by Vroom 28. He

implemented the concept “H-factor”, which combines the pulping temperature

and time as a single variable. Kleinert studied lignin and carbohydrate kinetics

as a function of temperature and alkali charge. 25 Later, Hatton 27 implement a

model that predicts the kappa number and residual lignin content, based on H-

factor and effective alkali (EF) charge. Vroom’s and Hatton’s models are im-

portant, because they are currently widely used in the industrial scale. However,

Hatton’s model has important drawbacks, namely that the effect of sulfidity and

the liquor-to-wood ratio are not included in the model. Additionally, while Hat-

ton’s model shows good predictive results for hardwood pulping; for softwood

pulping the results are not so reliable.

2.2.1.1 3-stage models LeMon&Teder51 were the first who presented the “3-stage model” approach to

the scientific community. In this model, the lignin and carbohydrate degrada-

tions were modelled based on pulping phases: initial, bulk and residual. Purdue

model 20 is the next chronological model. In this, wood chips were divided to

several components and each component was modelled individually. The Pur-

due model treats cellulose, galactoglucomannan and xylan separately. Addition-

ally, lignin is modelled as two phases: high and low reactivity lignin. The more

reactive lignin possesses β-ethers bonds, while the less reactive lignin contains

γ –ethers bonds. Later, the Purdue model was extended by Christensen 16. In

this study, the dependence of reactivity on hydroxide and hydrogen sulfide ions

was added, as well as unreactive wood components. Moreover, a special coeffi-

cient was used in order to adjust the kinetic parameters to different wood spe-

cies. Today, the Purdue model is the industrially utilized model, that can be used

in order to predict the effect of temperature, as well as to illustrate the mass

component profiles 53.

14

Another approach was introduced by Gustafson, who created a three-stage

model based on differential pulping periods. 18 In the developed model, the deg-

radation of carbohydrates depends on the delignification rate in each phase. The

three-stage model was developed further by Pu and Sarkanen, 1991, who mod-

elled cellulose and hemicellulose degradation separately. Another pulping

model was presented by Gustavson 44. It is based on carbohydrates degradation

as individual species (cellulose, galactoglucomannan and xylan). Carbohydrate

degradation was described by first order reactions, where the effects of hydrox-

ide and hydrogen sulfide ions were included. Additionally, the solubility of car-

bohydrates was included and expressed as a function of ionic strength.

A more recent attempt to improve Purdue model was performed by Andersson 54. In this work, degradation of all wood components is demonstrated by indi-

vidual parallel equations. However, model parameters do not distinguish car-

bohydrate species (cellulose, xylan, galactoglucomannan): all carbohydrates

have the same parameters in the model. Parameters vary only in different cook-

ing phases. The phase model presented by Johansson 55 is based on the idea of

how alkali and sodium concentration affect carbohydrates’ behaviour in kraft

pulping. However, the model was based on experimental data from later stage

of pulping, and consequently initial dissolution and primary peeling of carbo-

hydrates could not be included. Therefore, this model was focused on alkaline

hydrolysis instead of on the whole kraft pulping process.

2.2.2 Reaction mechanism-based models

Reaction mechanism-based models are models developed based on experi-

mental results from wood meal, in order to exclude the effect of mass transfer

on reaction kinetics. These models have been developed mainly for describing

carbohydrate degradation. 52

One of the first attempts to develop a reaction mechanism based model was by

Wigell 48, who described glucomannan degradation in alkaline pulping by a

power law equation. Galactoglucomannan was described by the amount of in-

soluble material, degradation through alkaline hydrolysis and primary peeling.

Later, Nieminen 19 presented a mathematical model where the actions of pri-

mary and secondary peeling, stopping and alkaline hydrolysis were all included.

In general, reaction mechanism-based models describe xylan and galactogluco-

mannan degradation similarly, which is inaccurate. Galactoglucomannan de-

15

grades as a result of endwise degradation through primary peeling, and second-

ary peeling following alkaline hydrolysis. 56,57 Xylan degradation through end-

wise peeling is however limited, due to stabilizing components such as arabi-

nose and 4-O-methylglucuronic acid present on the backbone of the polysac-

charide. 58 Additionally, xylan degradation also depends on dissolution of

longer polysaccharide chains. 59 Therefore, xylan degradation cannot be mod-

elled in the same way as galactoglucomannan, which means that a more com-

plex scheme for degradation and dissolution of xylan has to be developed. Con-

sequently, the existing reaction mechanism-based models cannot represent car-

bohydrate degradation details accurately.

2.2.3 Continuous distribution of reactivity models

Continuous distribution of reactivity models are models where the activation

energy of the degradation reaction is continuously distributing. 52 The degrada-

tion reactions are expressed through first order kinetics, with time-dependent

rate constants.60 So far, two models based on continuous distribution reactivity

model have been developed. They are presented in Montane 61 and Bogren 47

works. At the moment, Bogren 47 represents the most complex model, which

includes time-dependent rate constants, and pre-exponents dependencies on

the degree of delignification and temperature. However, the model is a hard to

use due to many adjustable “parameters”. For example, in order to predict xylan

removal in kraft pulping, 10 parameters are required. Additionally, it is hard to

predict how the model will behave at different reaction conditions, since the pa-

rameters are mathematically adjusted numbers in order to improve prediction

results. In general, this is the problem for all empirical models: they rarely have

a physical meaning, as they are based on mathematical fitting; adjusting simu-

lation results to experimental data.

2.2.4 Continuous digester models As well as these model types, Wisnewski 62 developed a continuous digester

model based on Purdue model 16, which was then later further developed and

integrated in dynamic process simulation software. 53 The earlier model was im-

proved by including wood chip compaction and chip and liquor level calcula-

tions, as well as including reaction kinetics presented by Gustafson 18.

Lo-SolidsTM is a simulation cooking model developed by Miyanishi 63. The model

is a steady-state simulation of continuous cooking systems. The kinetic model

utilized in the simulation is based on Purdue model 16,20. Besides the earlier pre-

16

sented components (high and slow reactive lignin, cellulose, galactoglucoman-

nan and xylan), the model also includes extractives. These are assumed to dis-

solve extremely quickly already in the beginning of cooking. The degradation of

the other five components is modelled based on initial, bulk and residual phases

of cooking. Additionally, the model is divided into three phases: wood compo-

nents, entrapped liquor, and free liquor.63

In general, mainly the modelling of softwood pulping has been considered. Only

a few studies16,64-67 have been previously performed for hardwood cooking pro-

cess.

It is clear that although a lot of models have been developed, there is no univer-

sal model with good prediction results. For example, xylan and galactogluco-

mannan cannot be described by the same equation, as was presented in the re-

action based models. Furthermore, the effect of pulping conditions has to be

reflected clearly in the model. A general model with an empirical kinetic equa-

tion may give a good fitting for individual cases; however, they rarely can give

reliable prediction results for a wide range of experimental setups due to the fact

that the equation, most often, has no physical or chemical meaning, and are just

representing mathematical fitting. This is clearly seen in Bogren’s model seem-

ingly presenting good prediction results, however with many coefficients having

to be applied (in xylan more than 10) for utilizing the model and obtaining reli-

able prediction results. Additionally, only few 19,68,69 studies take into consider-

ation the Donnan effect, which affects strongly the results, especially for carbo-

hydrates.

Looking into all mentioned models above, it is clear that a more fundamental

model of the kraft pulping process is needed. The model should not be limited

to certain experimental conditions such as wood species or white liquor compo-

sition. Moreover, the model should be able to explore a wide range of operating

conditions with confidence. In order to improve prediction of the models, the

amount of assumptions has to be limited. The model has to be built in such a

way that the modelled pulping processes could be easily expanded to accommo-

date different pulping additives. At the moment, no models based on detailed

physico-chemical phenomena of the kraft pulping process exists. Consequently,

there is a clear need for a model with several attributes:

is based on detailed, molecular-level reaction chemistry;

describes processes on physical and chemical bases;

17

predicts the behaviour of each type of carbohydrates (cellulose, xylan,

galactoglucomannan) separately;

predicts HexA formation and degradation;

describes delignification process in detail;

continuously monitors and reports the solid and solution (white liquor,

black liquor) compositions;

can be easily updated in the future;

is user-friendly.

18

19

3. Experimental

3.1 Chemicals

The model compound adlerone (Figure 12, structure 1) was synthesized accord-

ing to Adler 70, with a purity of 99% and dissolved in 40% 1,4-dioxane in distilled

H2O at 1 mM concentration. Stock solutions were prepared from the analytical

grade chemicals: NaOH and Na2S in distilled water at 1 M concentration.

3.2 Experiment procedure

The experimentation to find the adlerone reaction mechanism and its kinetic

parameters was a target for research in Publication 1. The degradation kinet-

ics of a non-phenolic lignin model compound with α-carbonyl functionality (ad-

lerone) was studied by varying the temperature and the concentrations of so-

dium hydroxide and sodium hydrogen sulfide. The kinetics of adlerone degra-

dation and the formation of its reaction products were monitored by UV-Vis

spectroscopy, and the product structures were analyzed by GC/MS. The two-

step degradation of adlerone was studied in two separate experimental setups.

In the first alkali catalyzed step, adlerone is converted to a β-elimination prod-

uct, that then reacts further in the second step with the hydrogen sulfide ion.

The reaction of the β-elimination product in the presence of [HS-] was studied

by UV-Vis spectroscopy by observing the absorbance decay at 350 nm. The Ar-

rhenius kinetic parameters were derived by the KinFit software 71.

20

3.3 Analytical Methods for Publication 1

3.3.1 Gas Chromatography (GC) with Flame-Ionization Detection (FID)

Trimethylsilylated samples were analysed on a Shimadzu Gas Chromatograph

GC-2010 PLUS (Shimadzu Corp., Kyoto, Japan), equipped with an HP-5 col-

umn (Agilent Technologies, USA) (25 m, 0.30 mm ID). Temperature program:

2 min at 50°C, 2°C min-1 to 280°C, and 2 min at 280°C. Approximately, 1 μl of

the sample was injected (split injection mode at 260°C). H2 was the carrier gas

(1 ml min-1); FID temperature: 290°C.

3.3.2 MS instrument applied:

A Thermo Scientific ISQ series single quadrupole mass spectrometer coupled

with a Trace 1300 GC instrument equipped with a TR-5MS column (Thermo

Fisher Scientific, MA, USA) (25 m, 0.25 mm ID) was employed, with carrier gas:

He (1.2 ml min-1). The temperature program was the same as for GC. The trans-

fer line and ion source temperature: 250°C and 200°C, respectively. A solvent

delay of 4.0 min was selected. In the full scan mode, EI mass spectra were rec-

orded in the range 50–700 m/z at 70 eV.

21

4. Methodology

4.1 FLOWBAT

FLOWBAT is a software that has been designed for the simulation of steady

state chemical processes, and can be applied for different process design phases.

For example, it can be utilized for describing reaction equilibria, as well as ma-

terial and energy balances, using simple and rigorous models. The software has

a large data bank of thermodynamic properties of pure compounds, which can

also be easily updated with new compounds. FLOWBAT also has the capability

to estimate various properties of pure compounds when the available data is not

complete. It has an extensive collection of methods to calculate and estimate

thermodynamic properties of mixtures.

One of the important features of FLOWBAT is its capability to carry out perfor-

mance calculations even with constraints set to the flowsheet. This means that

the user can define the objective to be minimized or maximized. Normally, this

includes the selection of variables to be optimised (FLOWBAT finds the best

values for them), writing constraints for the variables to be optimised, writing

constraints for dependent variables (FLOWBAT calculates them on the basis of

values determined for the variables to be optimised), and calculation of the ob-

jective function value.94

4.2 Modelling

Wood contains cellulose, hemicelluloses and lignin as its main components,

with additionally extractives and metals present in minor amounts. 1 In wood,

cellulose fibres are bound together by lignin. The target of any pulping process

is to dissolve lignin and separate cellulose fibres. In the kraft pulping process,

white liquor, containing water, sodium sulfide and sodium hydroxide at an ele-

vated temperature and pressure, dissolves lignin and separates the fibres. As a

result, a pulp suspension and black liquor are formed.

22

Wood chips, consisting mostly of fibres, when placed in an aqueous medium will

absorb liquid. Water bound to the fibre wall is named the Fibre-bound liquid

phase. Liquid external to the fibre wall is named the External liquid phase. The

volume of fibre bound liquid is defined by the fibre saturation point (FSP). In

the Handbook of Pulp 40,72, FSP for untreated wood is reported as 0.35 (0.3)

kgwater/kg fibres. The solid wood/fibres are treated in the model as an insolu-

ble phase. In contrast, the model compound studies, extracted mostly from lit-

erature, were modelled as liquid phase reactions. Division of the system to two

different phases gives a possibility to model variations in the chemical compo-

sitions through phase equilibria. Figure 11 presents a schematic picture of the

phenomena and components included in the pulping model. The model system

is described by chemical reactions (degradation and dissolution), equilibria be-

tween the fibre bound and the external liquid, and mass transfer (assumed to be

extremely fast in the current part of research).

Figure 11. A schematic picture of phenomena principles in the pulping model.

The picture represents chemical reactions, mass transfer and equilibria between

fibre bound and external liquid.

23

The model includes lignin, cellulose, hemicelluloses and inorganic compounds.

Lignin, cellulose and hemicelluloses were modelled as monomeric pseudo-com-

pounds. Certain monomeric compounds have their own properties and reactiv-

ity. Inorganic compounds were modelled by utilising their CAS numbers

(Chemical Abstract Services). For such components, properties needed for the

simulation were obtained from literature. Any compound included in the

model’s database has to match at least one of the following requirements:

it participates in one or more reactions included in the model;

it has an effect on pH of the solution;

it has an effect on ion exchange.

Pseudo units were determined based on analyses from wood and fibre samples

as well as experiments with model compounds. The self-performed experi-

mental work with the model compounds in order to determine the reaction

mechanisms and their kinetic parameters is presented in Paper 1. The model-

ling of delignification and all lignin pseudo-units are presented in Paper 2,

while modelling of carbohydrates and their pseudo-structures is reported in Pa-

per 3.

The cooking model consists of two liquid phases: fibre bound liquid and external

liquid. The concentrations of ionic species, including HO- and H3O+ (typically

expressed as pH), are different in the anionic fiber wall and in the external so-

lution. The anionic charge of the fiber wall originates mainly from dissociated

uronic acid groups (pKa ~3) in xylan and phenolic groups (pKa ~10) in lignin.

Towards the end of the pulping process the anionic charge of the cell wall de-

creases due to dissolution of lignin and xylan. Additionally, the elimination of

hexenuronic acid groups from xylan lowers the anionic charge of the cell wall.

The Donnan theory is used to define the uneven distribution of mobile ions be-

tween the two liquid phases and the real reaction conditions in the fiber wall. 68,73 Thus, the Donnan theory assists to make the model prediction more precise

and closer to the reality.

Reactions included in the model cover lignin degradation and dissolution, acid-

base equilibrium, peeling, stopping, and alkaline hydrolysis of cellulose, galac-

toglucomannan and xylan, as well as hexenuronic acid formation and degrada-

tion.

In the model, the equilibria between the fibre bound liquid phase and the exter-

nal liquid phase is included. Mass transfer take places between those two

24

phases. However, in the current research, mass transfer is assumed to be ex-

tremely fast, hence meaning that wood meal experiment were used for model

validation (in order to minimize the effect of mass transfer on reaction kinetics).

Mass transfer model of wood chips was developed by Kuitunen, unfortunately

the results are unpublished in scientific paper103. Reversible and irreversible re-

actions in both liquid phases are also included to the model. The idea is that

pseudo-compounds bound to the fibre wall react only with compounds in the

fibre wall liquid. Furthermore, reaction products of the fibre wall compounds

dissolve into the fibre bound liquid. Only from the fibre bound liquid can they

diffuse to the external liquid phase. A full list of incorporated chemical reactions

and rate parameters is given in Publication 2 (delignification) and in Publi-

cation 3 (carbohydrate degradation). The list does not contain all possible re-

actions taking place in pulping processes. However, they represent the most im-

portant reactions for the pulping process.

Several initial parameters have to be included in the models, such as:

wood composition - amount of lignin and carbohydrates (in order to cre-

ate an input file with a quantity of pseudo-compounds representing the

wood structure);

white liquor composition;

reaction mechanisms and kinetic parameters of all included reactions;

knowledge of pKa values;

process parameters (temperature, pressure).

In the cases when the reported reaction kinetic parameters were from own ex-

perimental work, or when the parameters obtained from literature were not

suitable, the reaction kinetic parameters were regressed with Kinfit software.71.

The objective function was minimized by the Levenberg-Marquardt method104

as presented by Equation 1: = ∑ ∑ ( , , ), (1)

where W is the weighting factor.

4.2.1 Lignin

The idea of this work was to obtain the kinetic parameters from lignin model

compound studies and to apply those parameters for developing a delignifica-

tion model. Lignin is a relatively complex compound (Figure 2), having a cross-

linked phenolic-type structure, which does not easily break down. 1 Therefore,

25

the structure of lignin was simplified in the model. Each pseudo-compound rep-

resents a group of real components with related structures and chemical prop-

erties. The full list of lignin structures used in the model is presented in Publi-

cation 2. In the current research, lignin-carbohydrate complexes (LCCs) and

lignin condensation reactions were not included in the model. The macromo-

lecular aspect of lignin was considered in lignin dissolution. Lignin dissolution

is a physical phenomenon, therefore it was modelled as a transfer process of

molecules from the insoluble fibre wall to the fibre-bound liquid through disso-

lution reactions in the same way is it is presented in Tarvo and Kuitunen works 74,75. All lignin fragmentation reactions were assumed to proceed similarly in the

fibre wall and in the liquor, so that the same kinetic parameters and reaction

routes apply to lignin pseudo-compounds in liquid phase (dissolved) and in the

fibre wall (insoluble). The stoichiometry and kinetics of the lignin fragmentation

reactions were based on the lignin model compound studies. The Arrhenius pa-

rameters were applied from the literature (frequency factor was calculated

based on published activation energies and reaction rates) or regressed against

experimental data presented in literature. Confidence limit was 93-97% in the

regressed values. Lignin has a macromolecular nature and this is taken into con-

sideration in its dissolution reactions. The main principles used in the lignin

dissolution scheme were:

the extent of dissolution is proportional to the amount of non-phenolic

units that are no longer attached to the next lignin unit from the β-car-

bon (meaning the extent of cleavage of -ether bonds in non-phenolic

units);

each dissolution reaction includes an ionized phenolic unit, promoting

the lignin fragment dissolution.

The main idea of the dissolution scheme is that the lignin ionized phenolic units

are attached to other lignin units, and hence they also force unreacted lignin to

dissolve. The rate of each dissolution reaction is proportional to the concentra-

tion of the ionized phenolic lignin derivative and the unreacted lignin unit. A

complete list of dissolution reactions is given in Publication 2. The dissolution

stoichiometry (ratio between units) and kinetic parameters were fitted using the

Kinfit software 71, with data from wood meal pulping studies 47,76-78

In Publication 2, a phenomena-based model for the delignification in kraft

pulping was described. Each reaction included in the model was first individu-

ally developed. In order to include a reaction in the model, its stoichiometry and

kinetic parameters should be known. Studies on the lignin model compounds

26

were used as source of lignin reaction mechanisms and kinetic parame-

ters.42,51,79-87 If detailed experimental data was available, regression of kinetic

parameters in KinFit71 was utilized. The modelling results were validated by ex-

perimental data until a satisfactory fit. The scheme of lignin reactions included

in the model is depicted in Figure 6. The list of all lignin reactions, included in

the model, and their kinetic parameters represented in the Publication 2.

When individual reactions of lignin degradation were included into the model,

a presentation of the macromolecular structure of lignin was developed. The

macromolecular aspect of lignin is described through a lignin dissolution model,

where lignin can dissolve from the fibre wall to the external liquid/black liquor.

Lignin dissolution model development was based on experimental work from

Bogren’s study 22 . Additionally, experimental results from Paananen’s research 77,78,88 were utilized.

4.2.2 Carbohydrates

Carbohydrates were modelled individually for each polysaccharide. Details

about the carbohydrate model are presented in Publication 3. Even though

each polysaccharide was modelled individually, the basic scheme for each was

similar. Galactoglucomannan, xylan and cellulose were represented as linear

polymers containing a reducing end group, a non-reducing end group and mon-

omeric units between these. The carbohydrate degradation model includes re-

versible acid-base equilibria, as well as peeling, stopping and alkaline hydrolysis

reactions for all carbohydrates. However, due to different chemistry in xylan

degradation, it was assumed that arabino(4-O-methylglucurono)xylan has two

kinds of side-groups. 4-O-Methylglucuronic acid (MeGlcA) side-groups were

treated as individual compounds, whereas arabinose side groups were consid-

ered as leaving groups that enhance the conversion of a reducing end group in

xylan into a stable metasaccharinic acid group through the stopping reaction.

Additionally, xylan dissolution as polymer 89 was included in the model. The

dissolved xylan was assumed to be degraded at the same rate as the fibre-bound

xylan. Possible readsorption of the dissolved xylan on the pulp surface was not

included in the model.

4.2.3 Kraft Pulping model

Based on the experimental results from Publication 1 and information availa-

ble in literature, the kraft pulping model was developed. For a detailed explana-

27

tion of the process, the model was particularly presented in two of the publica-

tions: Publication 2, about the delignification process, and Publication 3,

about the carbohydrate degradation. In both publications, the experimental

setup was identical and based on Bogren’s 22,47,76,90,91 and Paananen’s 77,78,88 ex-

perimental data. Pine wood composition, used in the model development, was

estimated by utilizing literature data from similar types of wood, as well as from

experimental results 77,78,88 (Table 3). The initial composition of the pulping liq-

uor as well as the reaction parameters, such as temperature, pressure and liq-

uid-to-wood ratio, were varied depending on the experimental setup. Extremely

fast mass transfer was assumed, by the fact that wood meal experimental studies 76,77,88 were utilized for the model validation and parameter optimization. Addi-

tionally, a high liquid-to-wood ratio was used in both studies in order to avoid

the effect of variable chemical concentrations on experimental results. Param-

eter regression with Kinfit software 71 using Levenberg- Marquardt optimization

algorithm was used to obtain the unknown model parameters.

The model follows the time evolution of the compounds presented in Table 3

during pulping, as well as the formation of new compounds. In the current

model the contents of Ca2+ and uronic acids have the strongest influence on the

Donnan effect, as was reported in Kuitunen work73.

Table 1. Composition of Scots pine (Pinus sylvestris L.) 1,77,88 used in the simu-

lations.

Component Content

% Mol(kg

wood)-1

Lignin 26.8

Nonphenolic (etherified) guaiacyl 1.07

Guaiacyl phenol 0.32

Nonphenolic guaiacyl with carbonyl group in α carbon 0.0081

Cellulose 41.8 %

Non-terminal groups in cellulose chain 2.466535

Reducing end group of cellulose chain 0.000223

Non-reducing end group of cellulose chain 0.000223

Glucomannnan 16.9 %

Non-terminal groups in glucomannan chain 1.02

Reducing end group in glucomannan 0.0102

Non-reducing end group in glucomannan 0.0102

28

Component Content

% Mol(kg

wood)-1

Arabinoxylan 8.4 %

Non-terminal groups Group in xylan chain 0.5784

Reducing end group of xylan chain 0.0056

Non-reducing end group of xylan chain 0.0056

4-O-methyl-D-glucuronic acid 0.08

Other carbohydrates 1.5 %

Arabinogalactan galactose non-reducing end-group 0.0106

Non-terminal groups in arabinogalactan 0.0160

Arabinogalactan arabinose non-reducing end group 0.0071

Methyl esterified galacturonan middle group 0.2308

Galacturonan middle group (non-esterified) 0.1539

Arabinose structure in hemicellulose 0.00056

Extractives 3.4 %

Ca++ 0.00021

K+ 0.00029

Mg++ 0.00007

Model Validation.

The comparison between experimental and modelling results regarding lignin

structures could not be performed on a molecular scale, because reliable analyt-

ical data for detailed characterization of the residual and dissolved lignin were

not available. Consequently, the simulation results were converted into stand-

ardized variables such as kappa number and lignin yield, in order to make the

evaluation. Tarvo et al.92 presented a detailed explanation of the kappa number

and lignin yield calculations in Flowbat software. The lignin content (weight

percent) was calculated as presented in Equation 2. The kappa number was

computed by taking into account only compounds bound to the solid fibre wall.

Equation 3 was utilized to convert the content of unsaturated structures in pulp

into the kappa number. The compound-specific oxidation equivalent values uti-

lized in the model are listed in Table 2.

2

29

3

where L = lignin content (wt %), Mw,i = molar weight of component i (g/mol),

ci = component concentration (mol/kg fiber), OxEqi = oxidation equivalent

value of component i, and OxEqarom = oxidation equivalent value of full aro-

matic ring.

Table 2. Oxidation equivalent values (OxEq) assigned for lignin.

Pseudo component OxEq

Full aromatic ring (phenolic and non-phe-

nolic lignin compounds)

16

Quinone 10

Muconic acid 12

Maleic acid 11

30

31

5. Results and discussions

5.1 Reaction of hydroxide and hydrogen sulfide ions with ad-lerone and its kinetic parameters. (Publication 1)

In Publication 1 the target was to verify the reaction mechanism of non-phe-

nolic lignin structures containing α-carbonyl functionality under alkaline con-

ditions in presence of hydrogen sulfide ions (Figure 12). A special area of inter-

est was the kinetic parameters of the reaction system to be included in the del-

ignification model. Although the content of α-carbonyl groups in natural lignin

is low, they may significantly enhance the speed of pulping process in presence

of common pulping nucleophiles.

The reaction of adlerone with NaOH was followed by UV-Vis spectroscopy by

monitoring the increase in absorbance (A) at 325-375 nm over time. Further-

more, in order to determine reaction products, GC-MS analyses were performed

at different reaction times.

In Figure 12 the reaction mechanisms of adlerone in different conditions are

presented. The scheme is based on earlier results from 42 and on own experi-

ments (Paper 1). The reaction between 1 and 2b is reversible, however the

equilibrium favors the formation of 2b, because the forward reaction is much

faster. The equilibrium between 1 and 2b and the ionization equilibrium of 1

were included in the kinetic model for calculation of the kinetic parameters for

the formation of 2a.

GC-MS analyses (Publication 1) showed the presence of a reversed aldol reac-

tion product that is formed through liberation of formaldehyde (compound 2a

ArCOCH2OAR’) and the β-elimination product 2b (ArCOC(OAr’)CH2). The rel-

ative amount of the products mainly depended on the hydroxyl ion concentra-

32

tion - degradation of adlerone was increasing linearly with increasing alkali con-

centration, which supports a 2nd order reaction between adlerone and HO- (Ta-

ble 1). The activation energy of the reaction was independent of [HO-] (Table 1).

The parameter optimization in Kinfit resulted in activation energy (Ea) of

69.1±4.3 kJmol-1, which is somewhat lower than the value published by Miksche

(73.6 kJ mol -1) 93

Figure 12. Reaction of adlerone (1) with sodium hydroxide leads to the for-

mation of formaldehyde elimination (2a) and β-elimination products (2b). Re-

action of the β-elimination product (2b) with sulfide ions leads to cleavage of β-

ether bond.

33

Table 3. Obtained kinetic parameters for the second order reaction of adlerone

with hydroxyl ions by fitting the experimental data with Kinfit software.

T (oC) [HO-] mol l-1 Ea (kJ mol-1) k1(at 25 oC) (M-1s-1)

25, 35, 45, 55 0.5 69.4±3.8 0.00312±0.00046

25, 35, 45, 65 0.05 68.1±2.6 0.00045±0.00012

25, 35, 45, 65 0.005 71.2±1.7 0.00005±0.00001

Model 0.5, 0.05, 0.005 69.1±4.3 0.03942±0.00203

GC-MS analyses showed that the β-aryl ether bond cleavage took place imme-

diately after hydrogen sulfide was added into the solution containing 2b. Addi-

tionally, guaiacol, ArCOCHCH2 (3b) and ArCOCH(OH)CH2OH were detected.

Moreover, traces of compounds 2a and 2b were observed in the GC-MS spectra.

The episulfide 3a was not detected, as it is a very reactive structure and hence

is rapidly converted, most probably through either nucleophilic addition of HO-

or elimination of sulphur. The activation energy of conversion of 2b to 3 de-

pended on [HO-] at constant [HS-] (Table 2).

Table 4. Obtained kinetic parameters by fitting in the Kinfit software experi-

mental data for 2b with HS- ion.

T (o C) [HO-] (mol l-1) [HS-]

(mol l-1)

Ea(kJ

mol-1)

k1(at 25 oC) (M-1s-1)

25,35,45,55 0.5 0.001 51.3±1.2 0.00087±0.00012

25, 45,65 0.05 0.001 49.9±2.1 0.00023±0.00005

25 ,45,65 0.005 0.001 43.4±1.4 0.00012±0.00002

Model 0.5,0.05,0.005 0.001 42.4±0.9 0.39416±0.02500

Based on an earlier study by Gierer 42 and experimental results in Publication

1, it was concluded that the formation and degradation of 2b in the presence of

OH- and SH- ions follows a multi-step reaction mechanism presented in Figure

12. The formation of 2b from adlerone is reversible, however the forward reac-

tion is significantly faster and so the equilibrium favours 2b ([2b] >  > [1]). The

extent of ionization of 1 is defined by its ionization constant Kion that equals to

the ratio of the acid dissociation constant Ka and the ion product of water (Kw):

Kion = [ 1i] /([ 1][ HO-])

(4)

The total concentration of adlerone, [1]tot, equals to the sum of the concentra-

tions of the undissociated (1) and ionized (1i) forms of adlerone:

34

[ 1]tot =[ 1]+[ 1i] (5)

[ 1] =[ 1]tot /( 1+Kion[HO-]) (6)

[ 1i] =[ 1]tot Kion[HO-] /( 1+Kion[HO-]) (7)

The overall rate of formation and degradation of 2b in aqueous mixtures of HO-

and HS- is expressed by Equation 5. This is because 2a and 2b are formed

through an intramolecular reaction of 1i, and an intermolecular reaction be-

tween 1 and [HO-], respectively, and since 2b is additionally degraded in an in-

termolecular reaction with [HS-] (Reaction 3, Figure 12).

d[2b]/dt=(k1-k2)Kion[HO-][1]tot/(1+Kion[HO-])-k3[HS-][2b] (8)

Although the degradation of 2b at low [HO-] could be described by the reaction

with HS- alone, the degradation of 1i was dominant at high [HO-] (Table 2).

The obtained experimental data was utilized in Kinfit software 71 in order to col-

lect precise kinetic parameters with rather narrow confidence limits (95%). The

chemical reaction scheme and kinetic parameters were included into the simu-

lation. The model shows good prediction results, presented in Figures 13 and 14.

Additionally, analysing the results from Table 2 shows that the simulation re-

sults are in line with the experimental results at low alkali concentration, when

product 2b mainly reacts with HS-. The experiments proved a linear depend-

ence of the reaction rate on [HS-] at a constant [HO-].

Figure 13. Dependence of the degradation rate of adlerone on time, simulation

versus experimental results, [HO-]=0.05M. Experiments are symbols and sim-

ulation are lines: ♦, ̶̶̶̶ ̶̶̶̶ ̶̶̶̶ ̶̶̶̶ 25 ºC; ●, ̶̶̶̶ ̶̶̶ ̶̶̶ ̶̶̶ ̶̶̶ ̶̶̶ ̶̶̶ 45ºC; ▲, . . . . . . 65ºC.

35

Figure 14. Dependence of the decomposition rate of 2b on time, experimental

versus simulation results, [HO-]= 0.05 M, [HS-]= 0.001 M. Experiment results

are presented as symbols and simulation are presented as lines: ♦, ̶̶̶̶ ̶̶̶ ̶̶̶ ̶̶̶ ̶̶̶ ̶̶̶ 25 ºC;

⁕, _ _ _ 35ºC; ●, . . . . . . 45ºC.

Kinetic parameters for adlerone reaction in kraft pulping conditions were ob-

tained from experimental work presented in Publication 1. Collected experi-

mental data and kinetic parameters were important for future work and they

were utilized for developing delignification model of kraft pulping. Additionally,

Publication 1 gave new insight on the cleavage kinetics of β-aryl ether bonds

in α-carbonyl containing non-phenolic lignin structures by hydrogen sulfide

ions.

36

5.2 Kraft Pulping Model

5.1.1 Delignification model (Publication 2)

Once lignin degradation and lignin dissolution were included in the model, the

overall delignification process could be modelled. In order to prove that the

model could represent real experimental data, a simulation based on Bogren’s

experimental setup was launched. In Figure 15 the model results versus experi-

mental data are presented.

Figure 15. Simulation (lines) and experimental results 76 (dots) experimental

and simulation conditions: [HO-]= 0.25M, [HS-]= 0.1 M (■, ̶̶̶̶ ̶̶̶ ̶̶̶̶ 168 ºC; ●, ̶̶̶̶ ̶̶̶ ̶̶̶ ̶̶̶ ̶̶̶ ̶̶̶

154 ºC; ▲, ̶̶̶ ̶̶̶̶ ̶̶̶ ̶̶̶̶ ̶̶̶ ̶̶̶̶ 139 ºC; ⁕, ̶̶̶̶ ̶̶̶ ̶̶̶̶ 123 ºC;♦, ̶̶̶̶ ̶̶̶ ̶̶̶̶ 108ºC)

As is visible in Figure 15, the model produces good correlation with the experi-

mental data. At lower temperatures (108-139 0C) the fitting is perfect; however,

at higher temperatures (154-168 ºC) the model slightly overestimates the extent

of delignification. As it is known, in delignification multitude of reactions pro-

ceed simultaneously 41; in the current model only the most important reactions

with the biggest effect are included (Figure 6). As a result of lignin degradation,

new linkages between carbohydrates and lignin are formed, as well as lignin

condensation products. 95 The amount of those products does not have a critical

effect on delignification; however, this could be one reason why in our model

delignification is slightly overestimated. In order to confirm that the model

37

could be used in different experimental setup, another study from Paananen et

al. 77 was used as a source of experimental parameters.

Figure 16. Simulation (lines) and experimental results 77 (dots) experimental

and simulation conditions: [HO-]= 1.55M, [HS-]= 0.31M (■, ̶̶̶̶ ̶̶̶ ̶̶̶̶ 160 ºC; ●, ̶̶̶̶ ̶̶̶ ̶̶̶ ̶̶̶ ̶̶̶ ̶̶̶

150ºC; ▲, ̶̶̶ ̶̶̶̶ ̶̶̶ ̶̶̶̶ ̶̶̶ ̶̶̶̶ 140ºC; ♦, ̶̶̶̶ ̶̶̶ ̶̶̶̶ 130ºC)

Figure 16 shows good agreement between the model and the experimental re-

sults. There is also noticeably again a slight overestimation at high temperature

(160 0C) in the initial and bulk phase of delignification. Additionally, it is visible

that at high temperature (160 0C) in the residual phase of pulping, delignifica-

tion is underestimated, which was also the case in Figure 15, in the delignifica-

tion at 168 0C. It is known, that about 20% of lignin degrades in the initial stage

of delignification. The degradation proceeds mainly by the reaction of phenolic

lignin units in several steps (formation of quinone methide, addition of HS-,

cleavage of β-ether bonds).96 In the bulk phase, reactions of non-phenolic lignin

take place (cleavage of β-aryl ether linkages through intramolecular nucleo-

philic substitution by ionized α- or γ-hydroxyl groups, demethylation reac-

tions). In the residual phase, cleavage of lignin carbon-carbon bonds, as well as

demethylation, take place; with both reactions being slow.1 Based on knowledge

about these reactions, it is probable to suggest that the underestimation of del-

ignification in the residual phase could result from the missing of lignin carbon-

carbon bonds cleavage reaction. Demethylation reactions are included in the

model, since it is known that demethylation can promote lignin dissolution by

formation of ionized HO groups 97 and play an important role in odorous com-

pounds formation 80.

38

The model is developed with the target to be a useful and practical tool for re-

search and development needs. Consequently, the model has to be able to pre-

dict cooking results depending on variations in chemical concentrations as well

as temperature and pressure. Figures 17 a and 17 b present results at tempera-

tures 168 0C and 139 0C, respectively, when hydroxide ion concentration is con-

stant and hydrogen sulfide concentration is varied.

Figure 17. imulation (lines) and experimental results (dots) , a) 168 0C and b)

139 0C. Experimental data points 76 represented by symbols, simulation results

expressed by lines. ●, ̶̶̶̶ ̶̶̶ ̶̶̶̶ ̶̶̶ ̶̶̶̶ [HO-]=0.25 M, [HS-]=0.25 M; ▲, ̶̶̶ ̶̶̶̶ ̶̶̶ ̶̶̶̶ ̶̶̶ ̶̶̶̶ [HO-]=0.25

M, [HS-]=0.50 M; ■, ̶̶̶ ̶̶̶̶ ̶̶̶ ̶̶̶̶ ̶̶̶ ̶̶̶̶ ̶̶ [HO-]=0.25 M, [HS-]=0.10 M

In Figure 17 a (168 0C), simulation results are in good correlation with the ex-

perimental data. However, at lower [HS-], the model overestimates delignifica-

tion in the initial and bulk phases. Explanation for this observation could be

hidden in the reactions included in the model. As it is known from different

studies 40,83,98,99, hydrogen sulfide ion is promoting lignin degradation by nucle-

ophilic substitution in β-O-4 linkages, and the ion is also involved in demethyl-

ation reactions. In industrial processes, when hydrogen sulfide concentration is

low, also competing reactions of lignin condensation are able to take place more

actively. Since the condensation reactions are not included in the model, the

effect of competing reactions on delignification are not available and our simu-

lation results slightly overestimate delignification. The modelling results with

higher hydrogen sulfide concentration support this suggestion: at 0.5 and 0.25

M [HS-] no significant overestimation is noticeable.

Figure 18 shows model prediction when hydroxide ion concentration is varied

while hydrogen sulfide ion concentration is kept constant.

a)

b)

39

Figure 18. Simulation (lines) and experimental results (dots), a) 168 0C and

b)139 0C. Experimental data points76 represented by symbols, simulation results

expressed by lines. ●, ̶̶̶ ̶̶̶ ̶̶̶̶ ̶̶̶ ̶̶̶ [HO-]=0.25 M, [HS-]=0.25 M; ▲, ̶̶̶̶ ̶̶̶ ̶̶̶̶ ̶̶̶ ̶̶̶̶ [HO-]=0.10 M,

[HS-]=0.25 M; ■, ̶̶̶ ̶̶̶̶ ̶̶̶ ̶̶̶̶ ̶̶̶ ̶̶̶̶ ̶̶ [HO-]=0.78 M, [HS-]=0.25 M

In Figure 18 a good correlation between the model and the experiment is shown.

At higher alkali concentrations ([HO-]=0.78 M) and higher temperature, the

model overestimates the delignification in the initial and bulk phases, while the

delignification is underestimated in the residual phase. As was discussed above,

the reason for this could lie in possible condensation reactions and carbon-car-

bon bond formation. Additional formation of lignin carbohydrate complexes

could also be having an effect. However, it is important to mention and take into

consideration the possibility that the dissolution scheme in our model has some

problems. As was described earlier, each dissolution reaction includes an ion-

ized phenolic unit, promoting the lignin fragment to dissolve. The free ionized

phenolic units are formed as a result of depolymerization reactions of lignin and

demethylation reactions. The optimization of reaction kinetic parameters and

stoichiometry coefficients in dissolution reactions was conducted with Kinfit

software71. The lignin dissolution is a physical phenomenon that was repre-

sented as a transfer process from the insoluble fibre wall into the fibre liquid by

chemical reactions, meaning it therefore contains a reliability risk in the mod-

elled dissolution scheme.

Another possible reason for the difficulties in lignin dissolution model is the fact

that links between lignin and cell wall polysaccharides could prevent the disso-

lution. These interunit linkages can be formed during condensation reactions.

Lignin dissolution is a complex process. It is possible that modelling only its

chemical aspect is not enough for the good prediction results. Such features as

a)

b)

40

thermodynamic and stereochemical aspects of lignin degradation and dissolu-

tion would be beneficial to include in the delignification model. Including lig-

nin-carbohydrate linkages to the model can either promote or prevent lignin

dissolution 100. Additionally, another possible feature is that lignin dissolution

should be considered based on solubility, rather than on chemical reactions.

Chemical reactions and chemical structures are important; however, the physi-

cal state of lignin as a polymer is also important for the solubility properties.

Research work on lignin model compounds 81,87,101 has shown that erythro and

threo isomers of β-O-4 structures of lignin have different kinetic parameters. It

was reported that threo isomers always have higher activation energy than their

erythro counterpart for all phases of delignification. 87

Already at the current stage of the model development, the simulation results

have reliable correlations with the experimental data. Figure 19 demonstrates the

linear regression between simulation results and experimental values76. Obtained

slopes are varying from 0.89 to 1.23 and r-square values are varying from 0.96

to 0.99. The lowest r-square values (0.96-0.98) belong to the 108 °C values, ex-

cept cooking at 108 °C and [HO-]=0.25 M, [HS-]=0.10 M. However only in these

cooking conditions [HO-]=0.25 M, [HS-]=0.10 M and at 139 °C - linear regres-

sion has the slope 0.89 with r-square (0.97). Others linear regression demon-

strate the slope 0.98-1.08 with r-square 0.98-0.99. From Figure 19 it is clearly

visible that deviations between the model and experimental values, mainly, take

place at the end of the cooking after approximately 0.2 of lignin yield. Addi-

tionally, deviations at the lower cooking temperatures are visible. This could be

improved by including thermodynamics in the delignification process, which

could improve both the prediction results and the scientific background of the

model.

In order to show that the model can be an important tool for industry, for exam-

ple, kappa number could be predicted by the model as presented in Figure 20.

Additionally, there are other engineering parameters that could be obtained from

the detailed chemical composition of the reaction system, including e.g intrinsic

viscosity, brightness and yield of the pulp, as well as active alkali, effective alkali

(Figure 21), sulfidity, and the higher heating value of the black liquor.

41

Figure 19. Linear regression between simulation (model, lignin yield) vs ex-

perimental (lignin yield) values 76

a) [HO-] =0.25 M, [HS-] =0.25 M, b) [HO-] =0.25 M, [HS-] =0.50 M,

c) [HO-] =0.25 M, [HS-] =0.10 M, d) [HO-]=0.10 M, [HS-]=0.25 M

Experimental and simulation conditions: ■, ̶̶̶̶ ̶̶̶ ̶̶̶̶ 168 ºC; ●, ̶̶̶̶ ̶̶̶ ̶̶̶ ̶̶̶ ̶̶̶ ̶̶̶ 154 ºC; ▲, ̶̶̶ ̶̶̶̶ ̶̶̶ ̶̶̶̶ ̶̶̶ ̶̶̶̶

139 ºC; ⁕, ̶̶̶̶ ̶̶̶ ̶̶̶̶ 123 ºC;♦, ̶̶̶̶ ̶̶̶ ̶̶̶̶ 108ºC)

Figure 20. Simulation (black symbols) and experimental results 77 (red sym-

bols) of kappa number [HO-]=1.55 M, [HS-]=0.31 M, (■ , ■ - 160 ºC; ●, ●

150ºC; ▲, ▲- 140ºC; ♦ , ♦ - 130ºC)

0 50 100 150 200 250 300 350 400 450 5000

20

40

60

80

100

Kapp

a, m

l/g

Time (min)

42

Figure 21 Simulations of alkali behaviour in pulping conditions: L:W (7:1),

[HO-] 0.25 M, [HS-] 0.25 M ---- 168 °C, ---- 154 °C and ---- 139 °C. [HO-] mol/kg

water versus time ----- 168 °C [HO-] 0.25 M, [HS-] 0.25 M and ---- 168 °C [HO-

] 0.25 M, [HS-] 0.1 M

Based on the results presented in Publication 2 and on the discussion pre-

sented here, the model gives reliable correlation results for delignification. At

higher temperature (168 0C) and low hydrogen sulfide concentration [HS-] (0.1

M), the model can slightly overestimate delignification. However, kappa num-

ber shows good correlation results between experimental and modelled results.

5.1.2 Carbohydrate degradation model (Publication 3)

The objective of Publication 3 was to model the carbohydrate degradation in

kraft pulping. The scheme of the carbohydrate degradation reactions in kraft

pulping included in the model are presented in Figure 21.

The experimental data from Paananen 77,88 was used for the model validation.

Paananen was using NaCl in order to obtain fixed ionic strength of the solution,

which was also included in the model. As is visible in the results (Publication

3), NaCl addition/fixed ionic strength has no effect on the carbohydrate degra-

dation, neither in the model nor in the performed experiments.

Temperature is a crucial parameter for carbohydrate degradation. In Figure 22,

the effect of temperature on carbohydrate degradation (cellulose, xylan, and

galactoglucomannan) in the same reaction conditions is plotted. It can be seen

that the model is highly successful in predicting carbohydrate degradation at

different temperatures. According to Figure 23, from all the modelled carbohy-

drates, the temperature has the strongest effect on xylan. The temperature also

has a significant effect on galactoglucomannan degradation; however the effect

is much lower than on xylan.

43

Figure 22. Carbohydrate reactions in kraft pulping, where R is a carbohydrate

chain. (1) Alkali-catalysed hydrolysis; (2) stopping reactions of reducing end

groups; (3) alkali-catalysed peeling of reducing end groups; (4) formation and

degradation of hexenuronic acid groups (HexA).

Figure 23. Simulation (lines) and experimental results (symbols) 77,88. Carbo-

hydrate simulation, [HO-]=1.55 M, [HS-]=0.31 M , [Cl]= 0.14 M. Red is cellulose,

green is galactoglucomannan (ggm), violet is xylan: ■, ̶̶̶ ̶̶̶ ̶̶̶̶ ̶̶̶ ̶̶̶ cellulose, 130 0C; ●, -

- - - cellulose, 160 0C; ▼, ̶̶̶̶ ̶̶̶ ̶̶̶̶ ̶̶̶ ̶̶̶̶ xylan 1300C; ♦, - - - - xylan, 160 0C; Δ, ̶̶̶ ̶̶̶ ̶̶̶̶ ̶̶̶ ̶̶̶ galactoglucomannan 130 0C; ▲, - - - - - galactoglucomannan 160 0C

44

Figure 24 presents the effect of alkali and hydrogen sulfide concentrations on

the carbohydrate degradation. The model results are in line with the experi-

mental results. Figure 24 proves that the developed model can predict the effect

of alkali and hydrogen sulfide concentration on carbohydrate degradation with

reliable prediction results.

Figure 24. Simulation (lines) and experimental results (symbols) 77,88. Carbo-

hydrate simulation, 160 0C; Green is cellulose, blue is xylan and orange is galac-

toglucomannan (ggm).

[HO-]=1.55 M, [HS-]=0.31 M , [Cl]= 0.14 M: ■, ̶̶̶ ̶̶̶ ̶̶̶̶ ̶̶̶ ̶̶̶ cellulose; ♦, ̶̶̶ ̶̶̶ ̶̶̶̶ ̶̶̶ ̶̶̶ ggm, ●, ̶̶̶ ̶̶̶ ̶̶̶̶ ̶̶̶ ̶̶̶ xylan; [HO-]=0.50 M, [HS-]=0.10 M , [Cl]= 0.14 M: ▲, _ _ _ _ cellulose; ▼, _ _ _

_ ggm; ⁕, _ _ _ _ xylan.

In Figure 25, simulation results of hexenuronic acid (HexA) formation and 4-

O-methylglucuronic acid degradation for Paananen’s experimental setup 77,88 is

presented. Paananen did not analyse the degradation of 4-O-methylglucuronic

acid or HexA formation and degradation, therefore the results in Figure 25 rep-

resent only the model prediction and could not be compared to real experi-

mental values.

45

Figure 25. HexA formation and methylglucoronic acid degradation simula-

tion, [HO-]=1.55, [HS-]=0.31 M , [Cl]= 0.14 M Continuous lines represent

methylglucuronic acid degradation, while the dashed lines represent HexA for-

mation. ___, ----- 130 0C; ___, ----- 140 0C; ___, ----- 150 0C; ___, ----- 160 0C.

Looking at results presented in Publication 1, 2 and 3, as well as results pre-

sented in the thesis, it is possible to conclude that the prediction results for kraft

pulping process are reliable. They are in line with the experimental data and

have good correlations. Through constructing the model, it is possible to

demonstrate the ability to predict and follow the trends of the main process var-

iables quite well, such as temperature and white liquor composition. It is very

important for the further development of the model to have more experimental

data from wood meal studies. Of special interest within the model development

are the black liquor and kraft lignin compositions. Moreover, wood meal exper-

iments with different species of wood could be beneficial in order to prove uni-

versality of the model. Due to the fact that the model is based on individual re-

actions and physical phenomena, wood composition should not have an effect

on the prediction results of the developed model, which is normally the case

with empirical models. However, it is very important that the correct wood spe-

cies’ chemical compositions are integrated into the model.

The developed model is based on the detailed chemistry of the lignin and car-

bohydrate reactions; therefore, it is a very complex, yet also very efficient model.

At this development stage, it is possible to improve or extend the model via the

addition of reaction stoichiometry and kinetic parameters to the simulator and

46

the database. The developed model predicts the composition of carbohydrates

very well. Unfortunately, it was not possible to prove how well the model can

predict kraft lignin and black liquor composition due to a lack of wood meal

studies with detailed analyses of these properties. Nevertheless, the model

shows good prediction results for delignification at different temperatures and

chemical compositions of white liquor. The model also shows a good match be-

tween experimental91 and simulation results for 4-O-methylglucuronic acid deg-

radation and HexA formation, as well as for HexA degradation. At the current

stage, when the model is being developed for the kraft pulping process, it is a

user-friendly model. In order to proceed with simulations, such input parame-

ters as wood and white liquor compositions are required, as well as the process

parameters; temperature and pressure. All reaction stoichiometry and kinetics

are included, therefore by filling only the input file with necessary parameters,

the user can obtain the output file with the prediction results. A graphical use

interface has been developed for the kraft pulping model, as well as for the

bleaching model. 102 If necessary, the models could be combined in order to sim-

ulate a whole sequence of pulping and bleaching, pre-extraction and pulping, or

a number of other sequences.

In general, the objectives of the current research, to develop a detailed model

of carbohydrate degradation and delignification in kraft pulping, has been

achieved. The model has been developed, and the simulations show good pre-

diction results for the experimental data.

5.1.3 Model applicability

The developed model is able to be utilized within kraft and alkaline pulping

simulations. Additionally, by some modifications and additions into the reac-

tion scheme, the model could be upgraded to polysulfide pulping.

In this thesis, it was demonstrated that the model could be an important tool

for the industry. It can predict different engineering parameters from the de-

tailed chemical composition of the reaction system, including for example the

kappa number, intrinsic viscosity, brightness and yield of the pulp, as well as

active alkali, effective alkali, sulfidity, and higher heating value of the black liq-

uor.

In the developed model, the impacts of the process changes could be reflected

with good accuracy even with moderate number of input parameters and pro-

cess variables. Therefore, the current model could be very useful in testing new

47

ideas or concepts, for example, it is possible to check the sensitivity of the for-

mation reactions of methyl mercaptan or dimethyl sulfide depending on time,

temperature, or any other parameters. As well as this, any other interesting

chemical compound concentrations or reaction sensitivities to changes in pa-

rameters could be predicted.

48

49

6. Conclusions and Future Outlook

The current dissertation summarizes the results of the research performed on

the physico-chemical modelling of alkaline pulping processes. As it was indi-

cated in the Introduction and Background sections, nowadays pulp mills are not

interested in pulp production only, but also in other valuable products. This

gives a clear need for a detailed pulping model, where the process is represented

by individual reactions combined with physical phenomena. The phenomena-

based models are difficult to develop, and a colossal amount of knowledge is

needed; however, the deeper the wood chemistry is studied, the more the input

data becomes available. The model could be improved by additional parameters,

reactions, and ideas at any moment, when the data is available. Additionally, by

dividing such a complex process as pulping into individual reactions, it is possi-

ble to reduce uncertainties and improve the quality of prediction.

The target of this work was to build a phenomena-based model. In order to

achieve this, the work was divided into three parts and the problems were solved

in turns. First, literature was researched in order to find the necessary data on

reaction mechanisms, kinetic parameters and experimental setup. The data for

lignin and carbohydrate model compounds, as well as the experiments with

wood meal, were needed. Furthermore, when available data was determined –

an experimental study on adlerone reaction mechanism and its kinetics param-

eters was realized. The obtained experimental data, as well as the literature in-

formation were utilized for delignification model developing. The model shows

generally good correlation results; however, at low hydrogen sulfide concentra-

tion the prediction loses precision.

The lignin dissolution scheme could then be revised for better prediction results.

In the model, lignin degradation reactions have the same kinetic parameters in

50

the fibre wall (immobilized components) and in external liquid (dissolved com-

ponents). However, in the real pulping, the kinetic parameters for lignin degra-

dation in fibre walls and in the liquid are different. Working on those parame-

ters could improve the quality of the model prediction.

Taking into consideration the thermodynamics and the stereochemical aspects

of lignin dissolution would be beneficial for the model. An important factor in

lignin dissolution is the effect of the dissolved solids on the diffusion rate. Dur-

ing pulping, the concentration of dissolved material is increasing with time, and

consequently, the viscosity of the external liquid is increasing, decreasing the

dissolution/diffusivity. In addition, including condensation reactions of lignin

in the model could be beneficial for better understanding of the relationship be-

tween fibre liberation points and rejected levels. Effect of dissolved solids and

condensation reactions is especially important to take in consideration when the

current model is combined with modelling the phenomena taking place in the

wood chips (chip level model) developed by Kuitunen103.

The carbohydrate model shows good prediction results. An interesting fact,

which came up during galactoglucomannan simulation is that in order to match

the experimental and the modelled results, the stopping reaction that describes

the possibility of the formation of metasaccharinic acid even at low alkali con-

centrations had to be added, otherwise simulation results were unsatisfactory.

Additionally, it is important to mention, that the Donnan effect included in the

model has a strong effect on the HexA predictions’ results. Overall, the carbo-

hydrate degradation model has very good correlation results.

Relating to this research, it is very important to mention here the chip level

model developed by Kuitunen 103. Combining the developed kraft pulping model

(chemistry part) with this chip level model, would enable simulating a real in-

dustrial process. Combining those models and the user interface 102 could offer

an exclusive model for the simulation of the kraft pulping process. Additionally,

by combining bleaching and pre-extraction models, whole sequences for some

processes and products could be modelled.

Overall, the objective of this work – to develop a kraft pulping model, which can

predict the delignification process in detail - was successfully met. More exper-

imental data, with wood meal especially, including detailed information on lig-

nin and black liquor composition would be beneficial, in order to regress and

validate the model parameters for better prediction. Moreover, wood meal from

51

different wood types could be also interesting to study and test in the frame of

the current model. The model can be further developed by introduction of sy-

ringyl units into the chemical compounds database. For this purpose reaction

of syringyl units in pulping conditions, have to be included in the chemical re-

action database, in order to get the best prediction results. Including the sy-

ringyl units in the model can improve prediction level of the model for hard-

wood and allow the possibility to predict effect of S/G ratio on pulping results.

Essentially, the presented modelling concept has been applied earlier for chlo-

rine dioxide bleaching as well as for hot water extraction. It has potential to be

developed further for other type of chemical processes, both for industry, and

for research needs. Additionally, the model could be used as a tool for testing

theories on reaction mechanisms and regressing kinetic parameters.

The presented model, as well as the models developed earlier, based on the real

physicochemical phenomena, have good capabilities for the prediction of pro-

cess behaviour. They can give a more detailed analysis of the effect of process

changes to the results than the traditional empirical/correlation models. Sim-

ulation tools are overall a cheap and efficient way to research, study and rethink

processes. Ultimately, accurate models can help to get more efficient, environ-

mentally friendly and quality products with minimal costs.

52

53

References

1. Sjöström E. Wood chemistry: Fundamentals and applications. 2nd ed. San

Diego: Academic Press, INC.; 1993.

2. Petterson RC. The chemical composition of wood. In: Comstock MJ, ed. The

chemistry of solid wood. American Chemical Society; 1984:57-126.

https://pubs.acs.org/isbn/9780841207967. 10.1021/ba-1984-0207.ch002.

3. Blanco A, Dahlquist E, Kappen J, Manninen J, Negro C, Ritala R. Use of

modelling and simulation in the pulp and paper industry. Mathematical and

Computer Modelling of Dynamical Systems. 2009;15(5):409-423.

4. Demuner IF, Olodette JL, Demuner AJ, Jardim CM. Biorefinery review:

Wide-reaching products through kraft lignin. BioResources. 2019;14(3):39.

5. Qiu X, Hu S. "Smart" materials based on cellulose: Review of the prepara-

tions, properties and applications. Materials. 2013;6(3):738-781.

6. Pichetha GF, Cleverton LP, Sierakowskia MR, et al. Bacterials cellulose in

biomedical applications: A review. International Journal of Biological Macro-

molecules. 2017;104(Part A):97-106.

7. Olivera S, Muralidhara HB, Venkatesh K, Guna K, Gopalakrishna K, Kumar

KY. Potential applications of cellulose and chitosan nanoparticles/composites

in wastewater treatment: A review. Carbohydrate Polymers. 2016;153:600-

618.

8. Tarasov D, Letch M, Fatehi P. Lignin-carbohydrate complexes: Properties,

applications, analyses and methods of extraction: A review. Biotechnology for

Biofuels. 2018;11.

54

9. Ooi Z, Harruddin N. Recovery of kraft lignin from pulping wastewater via

emulssion liquid membrane process. AiChEJournal. 2015;31(5):1305-1314.

10. Zhang Y, Yuan Z, Xu C. Engineering biomass into formaldehyde-free phe-

nolic resin for composite materials. AiChEJournal. 2015;61(4):1275-1283.

11. Yuan T, Xu F, Sun r. Role of lignin in a biorefinery: Separation characteri-

zation and valorization. J Chem Technol Biotechnol. 2013;88:346-352.

12. Abbati de Assis C, Greca LG, Ago M, et al. Techno-economic assesesment,

scalability , and applications of aerosol lignin micro-and nanoparticles. ACS

Sustainable Chemistry & Engineering. 2018;6(9):11853-11868.

13. Aguiar HC, Filho RM. Neutral network and hybrid model: A discussion

about different modeling techniques to predict pulping degree with industrial

data. Chemical Engineering Science. 2001;56(2):565-570.

14. Andersson N. Modelling of kraft cooking kinetics using near infrared

spectroscopy. [Diss]. Karlstad University, Karlstad, Sweden; 2003.

15. Burazin MA, McDonough TJ. Building a mechanistic model of kraft-anra-

quinone pulping kinetics. IPC Technical Paper Series. 1987:1-7.

16. Christensen T, Albright LF, Williams TJ. A kinetic mathematical model for

the kraft pulping of wood. Tappi Annual Meeting. 1983;239-243:239-243.

17. Clayton D., Easty D., Einsparhr D. et al. Kinetics and modeling of kraft

pulping. In: Grace TM, Malcolm EW, eds. Pulp and Paper Manufacture: Vol-

ume 5 Alkaline Pulping, Join Textbook Committee of the Paper Industry.

1989;45-73.

18. Gustafson RR, Slelcher CA, McKean WT, Finlayson BA. Theoretical model

of the kraft pulping process. Industrial & Engineering Chemistry Process De-

sign and Development. 1983;22(1):87-96.

19. Nieminen K, Paananen M, Sixta H. Kinetic model for carbohydrate degra-

dation and dissolution during kraft pulping. Industrial & Engineering Chemis-

try Research. 2014;53:11292-11302.

55

20. Smith CC, Williams TJ. Mathematical modeling, simulation and control

of the operation of a kamyr continuous digester for the kraft process. [Ph.D.

Thesis]. Purdue University, West Lafayette, IN, USA; 1974.

21. Santos RB, Hart P, Jameel H, Chang H. Wood based lignin reactions im-

portant to the biorefinery and pulp and paper industries. BioResources.

2013;8(1):1456-1477.

22. Bogren J, Brelid H, Theliander H. Towards a general kraft delignification

model. Nordic Pulp and Paper Research Journal. 2009;24(1):33-37.

23. Nieminen, K.,Kuitunen,S., Paananen M.,and Sixta, H.

Novel insight into lignin degradation during kraft cooking. Industrial & Engi-

neering Chemistry Research. 2014;53:2614-2624.

24. Saltin JF. A predictive dynamic model for continuous digester. TAPPI

Pulping conference. 1992:261.

25. Kleinert TN. Mechanism of alkaline delignification. 1. Overall reaction pat-

tern. Tappi. 1966;49(2):53-57.

26. Kerr A. The kinetics of kraft pulping - batch digester control. Tappi.

1976;59(5):89-91.

27. Hatton JV. Development of yield prediction equations in kraft pulping.

Tappi. 1973;56:97.

28. Vroom KE. The "H" factor: a means of expressing cooking times and tem-

peratures as a single variable. Pulp & Paper Magazine of Canada.

1957;58:228-231.

29. Tarvo V. Modeling chlorine dioxide bleaching of chemical pulp. [Diss.].

Aalto University, Espoo, Finland; 2012.

30. Lehtimaa T. Reactions of chlorine (III) and their kinetics in the chlorine

dioxide bleaching of kraft pulps. [Diss.]. Aalto University, Espoo, Finland;

2010.

56

31. Kuitunen S. Phase and reaction equilibria in the modelling of hot water

extraction, pulping and bleaching. [Diss.]. Aalto University, Espoo, Finland;

2014.

32. Fujita M, Harada H. Ultrastructure and formation of wood cell wall. In:

Hon DN-S, Shiraishi N, eds. Wood and cellulosic chemistry. 2nd ed. Marcel

Dekker; 2001:1-51.

33. Li T, Takkellapati S. The current and emerging source of technical lignins

and their applications: Source of technical lignins. Biofuels Bioproducts and

Biorefinery. 2018;12(5):756-787.

34. Christopher LP, Yao B, Ji Y. Lignin biodegradation with laccase-mediators

systems. Fronties in Energy Research: Bioenergy and Biofuels. 2014;2.

35. Sarkanen KV, Ludwig CH. Lignins occurence, formation, structure and re-

actions. USA: John Wiley&Sons, Inc; 1971.

36. Hatakeyama H, Hatakeyama T. Lignin structure, properties, and applica-

tions. In: Abe A, Dusek K, Kobayashi S, eds. Biopolymers. Advances in Poly-

mer Science, vol 232. Berlin, Heidelberg: Springer; 2009.

37. Reis JSG. Carbohydrate metabolism: Structural carbohydrates. In: Dey

PM, Harborne JB, eds. Plant biochemistry. Academic Press; 1997:205-236.

38. Finnish Forest Industries Federation. https://www.forestindustries.fi/sta-

tistics/pulp-and-paper-industry/. Updated 2019. Accessed 10/12, 2019.

39. Schönberg C, Oksanen T, Suurnäkki A, et al. The importance of xylan for

the strength properties of spruce kraft pulp fibres. Holzforschung.

2005;55(6):639-644.

40. Sixta H, Potthast A, Krotschek AW. Chemical pulping processes. In: Sixta

H, ed. Handbook of pulp. WILEY-VCHVerlag GmbH&Co; 2006:109.

41. Gierer J. Chemical aspects of kraft pulping. Wood Science and Technology.

1980;14(4):241-266.

57

42. Gierer J, Ljunggren S. The reactions of lignin during sulfate pulping, 18:

The significance of alfa-carbonyl groups for the cleavage of beta-aryl ether

structures. Svensk Papperstidning. 1980;3:75.

43. Ek M, Gellerstedt G, Henriksson G. Pulp and Paper Chemistry and Tech-

nology. Volume 2. Pulping Chemistry and Technology. Berlin: Walter de

Gruyter GmbH&Co; 2009.

44. Gustavson CA, Al-Dajani WW. The influence of cooking conditions on the

degradation of hexenuronic acid, xylan, glucomannan and cellulose during

kraft pulping of softwood. Nordic Pulp and Paper Research Journal.

2000;15(2):160-167.

45. Gellerstedt G, Lindfors E. On the structure and reactivity of residual lignin

in kraft pulp fibers. International Pulp Bleaching Conference. Stockholm,

Sweden. 1991. 1:73–88.

46. Gierer,J., Wännström, S. Formation of ether bonds between lignin and car-

bohydtrates during alkaline pulping process. Holzforschung. 1986;40:347.

47. Bogren, J. Brelid, H., Theliander, H. Assessment of reaction kinetic models

describing delignification fitted to well-defined kraft cooking data. Nordic

Pulp & Paper Research Journal. 2008;23(2):210-217.

48. Wigell A, Brelid H, Theliander H. Kinetic modelling of (galacto)glucoman-

nan degradation during alkaline cooking of softwood. Nordic Pulp & Paper

Research Journal. 2007;22(4):495-499.

49. Guan QQ, Zhou HJ, Peng LC, et al. Mathematical model for predicting the

dissolution behaviors of hemicelluloses during cold caustic extraction process.

AiChELetter: Biomolecular Engineering, Bioengineering, Biochemicals, Bio-

fuels, and Food. 2019;65(1):13-17.

50. Wilder HD, Daleski EJ. Kraft pulping kinetics. II. Delignification rate stud-

ies. Tappi. 1965;48(5):293-297.

51. LeMon S, Teder A. Kinetics of the delignification in kraft pulping. Svensk

Papperstidning. 1973;76(11):228.

58

52. Wetterling J. Modeling of hemicellulose degradation during softwood

kraft pulping. [Master Thesis]. Chalmers University of Technology, Gothen-

burg, Sweden; 2012.

53. Rantanen R. Modelling and control of cooking degree in conventional and

modified continuous pulping processes. [Dissertation]. Oulu University, Oulu,

Finland; 2006.

54. Andersson, N., Wilson, D. I., Germgård, U. An improved kinetic model

structure for softwood kraft pulping. Nordic Pulp & Paper Research Journal.

2003;18(2):200-209.

55. Johansson D, Germgård U. Carbohydrate degradation during softwood

kraft cooking – influence on cellulose viscosity, carbohydrate composition and

hexenuronic acid content. Nordic Pulp & Paper Research Journal.

2008;23(3):292-298.

56. Whistler RL, BeMiller JN. Alkaline degradation of polysacharides. Ad-

vances in Carbohydrate Chemistry. 1958;13:289-329.

57. Aurell R, Hartler N. Kraft pulping of pine. II. Influence of the charge of al-

kali on the yield, carbohydrate composition, and properties of the pulp. Svensk

Papperstidning. 1965.

58. Knill CJ, Kennedy JF. Degradation of cellulose under alkaline conditions.

Carbohydrate Polymers. 2003;51(3):281-300.

59. Norimoto M. Chemical modification of wood. In: Hon DN-S, Shiraishi N,

eds. Wood and cellulosic chemistry. 2nd ed. New York, Basel: Marcel Dekker,

Inc; 2001:573.

60. Montane D, Overend RP, Chornet E. Kinetic models for non-homogenous

complex systems with a time-dependent rate constant. Canadian Journal of

Chemical Engineering. 1998;76(1):58-68.

61. Montane D, Salvado J, Farriol X, Jollez P, Chornet E. Phenomenological ki-

netics of wood delignification: Application of a time dependent rate constant

59

and a generalized severity parameter to pulping and correlation of pulp prop-

erties. Wood Science and Technology. 1994;28(6):387-402.

62. Wisnewski PA, Doyle FJI, Kayihan F. Fundamental continuous-pulp-di-

gester model for simulation and control. AIChE Journal. 1997;43(12):3175–

3192.

63. Miyanishi T, Shimada H. Improvement of pulp strength and yield by com-

puter simulation on lo-solids kraft cooking (abstract). Tappi Journal.

2001;84(6):65.

64. Mortha G, Sarkanen K, Gustafson R. Alkaline pulping kinetics of short-ro-

tation, intensively cultured hybrid poplar. Tappi Journal. 1992;75(11):99-104.

65. Giudici R, Park SW. Kinetic model for kraft pulping of hardwood. Indus-

trial & Engineering Chemistry Research. 1996;35(3):856–863.

66. Saucedo VM, Josephson WE, Krishnagopalan GA. Dynamic modelling of

carbohydrates degradation in kraft pulping of hardwood. Appita Journal.

2002;55(5):398–403.

67. Kayihan F, Bills A, Hart PW. Mill application of a new continuous digester

model. AIChE Journal. 2005;51(9):2489–2494.

68. Towers M, Scallan AM. Predicting the ion-exchange of kraft pulp using

donnan theory. Journal of Pulp and Paper Science. 1996;22(9):332-337.

69. Pu, Q., Sarkanen, K. Donnan equilibria in wood-alkali interactions. Part II.

Effect of polysaccharide ionization at high alkalinities. Journal of Wood Chem-

istry and Technology. 1991;11(1):1-22.

70. Adler E, Lindgren BO, Saeden U. The beta-guaiacyl ether of alpha-veratryl-

glycerol as a lignin model. Svensk Papperstidning. 1952;7:245.

71. Jakobsson K, Aittamaa J. Kinfit parameter optimization software.

http://chemtech.aalto.fi/en/research/groups/chemical_engineering/soft-

ware/kinfit/. Accessed September, 29, 2015.

72. Sixta H. Ed. Handbook of pulp. Wiley-vch; 2006.

60

73. Kuitunen, S., Vuorinen, T., Alopaeus, V. The role of Donnan effect in kraft

liquour impregnation and hot water extraction. Holzforschung. 2013;67:511-

521.

74. Tarvo V, Kuitunen S, Aittamaa J, et al. Advanced approach on modelling

chemical pulp bleaching. PulPaper 2007 Conference: Innovative and Sustain-

able use of Forest Resources. 2007.

75. Kuitunen S, Tarvo V, Tamminen T, et al. Modeling of alkaline extraction

chemistry and kinetics of softwood kraft pulp. Holzforschung. 2014;10:68.

76. Bogren J, Brelid H, Theliander H. Reaction kinetics of softwood kraft delig-

nification - general considerations and experimental data. Nordic Pulp and

Paper Research Journal. 2007;22(2):177-183.

77. Paananen M,Liitiä T, Sixta, H. Further insight into carbohydrate degrada-

tion and dissolution behavior during kraft cooking under elevated alkalinity

without and in the presence of anthraquinone. Industrial & Engineering

Chemistry Research. 2013;52:12777-12784.

78. Paananen M, Rovio S, Liitiä T, Sixta H. Effect of hydroxide and sulfite ion

concentration in alkaline sulfite anthraquinone (ASA) pulping –

a comparative study Holzforschung. 2015;69(6):661-666.

79. Brunow G. A kinetic study on the mechanism of B-O-4 ether cleavage in

soda-antraquinone pulping. Paperi ja Puu. 1981;12:783-786.

80. William T, McKean J, Bjorn F, Sarkanen KV. Kinetic analysis of odor for-

mation in the kraft pulping process. Tappi. 1965;48(12):699-704.

81. Obst JR. Kinetics of alkaline cleavage of b-aryl ether bonds in lignin mod-

els: significance to delignification. Holzforschung. 1983;37(1):23-28.

82. Kondo R, Tsutsumi Y, Imamura H. Kinetics of beta-aryl ether cleavage of

phenolic syringyl type lignin model compounds in soda and kraft systems.

Holzforschung. 1987;41:83-88.

61

83. William T, McKean J, Bjorn F, Sarkanen KV. Kinetics of methyl mercap-

tane and dimethyl sulfide formation in kraft pulping. Tappi. 1968;51(12):564-

567.

84. Gierer J, Ljunggren S. The reaction of lignins during sulfate pulping, 16:

The kinetics of the cleavage of the beta-aryl ether linkages in structures con-

taining carbonyl groups. Svensk Papperstidning. 1979;3:71.

85. Gierer J, Ljunggren S. The reactions of lignin during sulfate pulping part,

17. Kinetic treatment of the formation and competing reactions of quinone

methide intermediates. Svensk Papperstidning. 1979;17:502.

86. Ljunggren S. The significance of aryl ether cleavage in kraft delignification

of softwood. Svensk Papperstidning. 1980;83(13):363-369.

87. Ahvazi BC, Argyropoulos DS. Thermodynamic parameters governing the

steroselective degradation of arylglycerol-b-aryl ether bonds in milled wood

lignin under kraft pulping conditions. Nordic Pulp & Paper Research Jour-

nal,. 1997;12:282-288.

88. Paananen M, Tamminen T, Nieminen K, Sixta H. Galactoglucomannan

stabilization during the initial kraft cooking of Scots pine. Holzforschung.

2010;64(6):683-692.

89. Danielsson S. Xylan reactions in kraft cooking: Process and product con-

siderations. [Doctoral thesis]. KTH, Stockholm, Sweden; ; 2007.

90. Bogren J, Brelid H, Theliander H. Effect of pulping conditions on the rates

of formation and degradation of hexenuroic acid in scots pine. Journal of Pulp

and Paper Science. 2008;34(1):23-29.

91. Bogren., J., Brelid ,H., Bialik, M., Theliander, H. Impact of dissolved so-

dium salts on kraft cooking reactions. Holzforschung. 2009;63:226-231.

92. Tarvo V, Lehtimaa T, Kuitunen S, Alopaeus V, Vuorinen T, Aittamaa J. A

model for chlorine dioxide delignification of chemical pulp. Journal of Wood

Chemistry and Technology. 2010;30(3):230-268.

62

93. Miksche GE. Lignin Reactions in Alkaline Pulping Processes (Rate processes in

Soda Pulping) Chemistry of delignification with oxygen, ozone and peroxides. 1st In-

ternational Symposium. 1975, 107.

94. Keskinen KI, Aittamaa J., eds., FLOWBAT, User’s Instruction manual,

Helsinki University of Technology, Laboratory of Chemical Engineering and

Plant Design, Espoo, 2004.

95. Gierer J. Chemistry of delignification - part 1: General concept and reac-

tions during pulping. Wood Science and Technology. 1985;19(4):289-312.

96. Gellerstedt G. Pulping chemistry. In: Hon DN-S, Shiraishi N, eds. Wood

and Cellulosic Chemistry. 2nd ed. Marcel Dekker, Inc.; 2001:859.

97. Calvo-Flores FG, Dobado JA. Lignin as renewable material. ChemSus-

Chem. 2010;3:1227-1235.

98. Turunen J. Demethylation and degradation of simple lignin model com-

pounds by pressure digestion with aqueous sodium hydrogen sulfide, sodium

methyl mercaptide and sodium hydroxide solutions. [Ph.D. Thesis]. Univer-

sity of Helsinki; 1963.

99. Brunow G, Miksche GE. Some reactions of lignin in kraft and polysulfide

pulping. . 1976;28:1155.

100. Jääskeläinen A, Willberg-Keyriläinen P, Liitiä T, Tamminen T. Carbohy-

drate-free and highly soluble softwood kraft. Nordic Pulp & Paper Research

Journal. 2017;32(4):483-490.

101. Miksche GE. Zum alkalischen abbau der p-alkoxy-arylglycerin-beta-ar-

ylätherstrukturen des lignins. *Versuche mit erythro-veratrylglycerin-beta-

guajacyläther. Acta Chemica Scandinavica. 1972;26:3275.

102. Kalliola A, Kangas P, Kuitunen S. Phenomena based simulation of chemi-

cal pulp production. 7 th International Colloquium on Eucalyptus Pulp. Vitó-

ria, Espirito Santo, Brazil. 2015.

103. Kuitunen S, Chip-scale mass transfer model. Unpublished results

63

104. Press WH, Flannery BP, Teukolsky SA, Vetterling WT, Numerical Recipes

- the Art of Scientific Computing (FORTRAN Version) Cambridge Universoty

Press, Cambridge, 1989

-otl

aA

DD

02

/ 1

20

2

+hchca

e*GMFTSH

9

NBSI 7-2720-46-259-879 )detnirp(

NBSI 4-3720-46-259-879 )fdp(

NSSI 4394-9971 )detnirp(

NSSI 2494-9971 )fdp(

ytisrevinU otlaA

gnireenignE lacimehC fo loohcS

smetsysoiB dna stcudorpoiB fo tnemtrapeD

fi.otlaa.www

+SSENISUBYMONOCE

+TRA+NGISED

ERUTCETIHCRA

+ECNEICSYGOLONHCET

REVOSSORC

LAROTCODSNOITATRESSID