Anders+T +Madsenم1

Downloaded from orbit.dtu.
dk on: Nov 07, 2022
Catalytic Production of Biodiesel
Madsen, Anders Theilgaard
Publication date:
2011
Document Version
Publisher's PDF, also known as Version of record
Link back to DTU Orbit
Citation (APA):
Madsen, A. T. (2011). Catalytic Production of Biodiesel. Technical University of Denmark.
General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright
owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.
 Users may download and print one copy of any publication from the public portal for the purpose of private study or research.
 You may not further distribute the material or use it for any profit-making activity or commercial gain
 You may freely distribute the URL identifying the publication in the public portal
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately
and investigate your claim.
___
__-_
O R 1 ,R 2,R 3
H 3CO R
___ R1 ,R2 ,R 3
OH R' O R''
R' R'' O
R' R''
Anders Theilgaard Madsen

Ph.D. Thesis
Technical University of Denmark
Department of Chemistry
June 2011
S
Centre for Catalysis and
Sustainable Chemistry
C C

Ph.D. Thesis
Technical University of Denmark
Department of Chemistry
June 2011
WASTE 2 VALUE
Ϯ͘ŐĞŶĞƌĂƚŝŽŶƐďŝŽĚŝĞƐĞů Danish National Research Foundation's
ƚŝůƚƌĂŶƐƉŽƌƚƐĞŬƚŽƌĞŶ Center for Sustainable and Green Chemistry
G
D
Preface
The present thesis is submitted in candidacy for obtaining the Ph.D.-degree from the Technical Uni-
versity of Denmark (DTU). The Ph.D.-project has been carried out from January 2008 to June 2011,
initially under the supervision of professor Claus Hviid Christensen at the Center for Sustainable and
Green Chemistry (CSG). When Claus sought new career opportunities professor Rasmus Fehrmann
took over supervision of the project in the current Centre for Catalysis and Sustainable Chemistry
(CSC).
The topic of the project has been the production of diesel oil from biomass, which may me pro-
duced via a number of different heterogeneous catalytic routes. The experimental work have focused
on two of these methods, both starting from fats and oils, namely the esterification and transes-
terification with methanol over acid and base catalysts to yield fatty acid methyl esters, and the
hydrodeoxygenation over supported transition-metal catalysts to yield straight-chain alkanes.
Financial support from the Danish Research and Innovation Council under the innovation con-
sortium “Waste-2-value” is gratefully acknowledged. Of the participants in Waste-2-value special
mention and thanks are due to Jakob S. Engbæk, Jesper Bøgelund, Sune D. Nygaard, and Jens Chris-
tiansen from the chemistry and materials groups at the Danish Technological Institute for spending
effort and time in preparing esterification catalysts and supports as well as for scientific and techni-
cal discussions. It has been inspiring and instructive to work together with Danish companies having
an practical interest in catalysis research.
I have had the pleasure of working together with a number of people to whom I wish to express
my sincere thanks: El Hadi Ahmed for performing many of the batch hydrodeoxygenation experi-
ments and synthesising a range of catalysts for that purpose; Olivier Nguyen van Buu for synthesising
a lot of the catalysts for the esterification and transesterification experiments; Lene Fjerbæk from the
University of Southern Denmark (SDU) for performing catalyst activity tests; Johannes Due-Hansen
and Uffe Vie Mentzel for an endless number of fruitful discussions on all technical matters arising
during the years as well as moral and scientific support. I wish to stress my thankfulness to all the
students and employees from both CSG and CSC for a helpful, social, and supportive working envi-
ronment.
I finally wish to express my utmost gratitude towards the Catalysis group at the Process Che-
mistry Centre, Åbo Akademi, Finland, for hosting my research stay there and helping to make it
indeed memorable. Especially professor Dmitry Murzin, associate professor Päivi Mäki-Arvela, and
Kari Eränen for their assistance and dedicated supervision. Special thanks go to Bartosz Rozmysłow-
izc, Toni Riittonen, and Atte Aho for good companionship, help, and fruitful discussions.
1
Abstract
The focus of this thesis is the catalytic production of diesel from biomass, especially emphasising
catalytic conversion of waste vegetable oils and fats.
In chapter 1 an introduction to biofuels and a review on different catalytic methods for diesel
production from biomass is given. Two of these methods have been used industrially for a number
of years already, namely the transesterification (and esterification) of oils and fats with methanol
to form fatty acid methyl esters (FAME), and the hydrodeoxygenation (HDO) of fats and oils to
form straight-chain alkanes. Other possible routes to diesel include upgrading and deoxygenation
of pyrolysis oils or aqueous sludge wastes, condensations and reductions of sugars in aqueous phase
(aqueous-phase reforming, APR) for monofunctional hydrocarbons, and gasification of any type of
biomass followed by Fischer-Tropsch-synthesis for alkane biofuels. These methods have not yet been
industrialised, but may be more promising due to the larger abundance of their potential feedstocks,
especially waste feedstocks.
Chapter 2 deals with formation of FAME from waste fats and oils. A range of acidic catalysts were
tested in a model fat mixture of methanol, lauric acid and trioctanoin. Sulphonic acid-functionalised
ionic liquids showed extremely fast convertion of lauric acid to methyl laurate, and trioctanoate was
converted to methyl octanoate within 24 h. A catalyst based on a sulphonated carbon-matrix made
by pyrolysing (or carbonising) carbohydrates, so-called sulphonated pyrolysed sucrose (SPS), was
optimised further. No systematic dependency on pyrolysis and sulphonation conditions could be
obtained, however, with respect to esterification activity, but high activity was obtained in the model
fat mixture. SPS impregnated on opel-cell Al2 O3 and microporous SiO2 (ISPS) was much less active
in the esterification than the orignal SPS powder due to low loading and thereby low number of
strongly acidic sites on the real catalyst. The ISPS-packed-bed catalysed conversion of rapeseed oil
revealed low activity but advantageous flow properties. A number of functionalised, organic bases
have been tested for their activity in transesterification of trioctanoate with methanol, especially
guanidines. The activity for trioctanoate conversion was promising, however, hygroscopic catalysts
be may lead to saponification of the triglyceride.
Hydrodeoxygenation is treated in chapter 3. The reaction routes and activity in batch hydrodeoxy-
genation are strongly dependent on the supported noble metal catalyst and temperature used for the
conversion to alkanes. Generally, Pt and Pd were the most active metals and have highest selectivity
for decarboxylation reaction. Stearic or oleic acid were converted much faster than tripalmitin. The
deoxygenation was performed in a continuous trickle-bed reactor over 2 wt% Pd/Sibunit at 300℃.
10 wt% stearic acid yielded almost complete and selective conversion to heptadecane in 5% H2 /Ar
at 20 bar, however pure Ar gas led to deactivation. A deactivation profile by coking builds up as
a function of the distance from the reactor inlet. A constant conversion of 12% was obtained with
neat stearic acid in 7 days time-on-stream of the spent catalyst. The activity for deoxygenation in
continuous mode decreased as stearic acid > ethyl stearate > tristearin under 5 % H2 in Ar, while
lack of H2 in the feed quickly led to complete deactivation of the catalysts in all feeds.
The work is concluded with a summary and an outlook in chapter 4.
2
Dansk Resumé
Hovedemnet i denne afhandling er katalytisk fremstilling af diesel fra biomasse, med særligt fokus
på katalytisk omdannelse af affaldsfedtstoffer som slagteriaffald, brugte stegeolier o.a.
Kapitel 1 giver en kort introduktion til biobrændstof og en redegørelse for publicerede metoder
til dieselfremstilling fra biomasse. Homogent katalyseret forestring og omestring af fedtstoffer med
methanol til såkaldt FAME og hydrodeoxygeneringen af fedtstoffer til alkaner er de to metoder, der
er udbredt i industriel skala til dato. Diesel fra biomasse kan dog også fremstilles ved opgradering
og deoxygenering af tjære og slam og pyrolyse-olier, ved raffinering af kulhydrater via kondensa-
tioner og reduktion i vandfase til monofunktionaliserede kulbrinter og ved komplet forgasning og
efterfølgende Fischer-Tropsch-syntese til alkaner. De sidste metoder er endnu ikke blevet opskaleret
til industriel skala, men kan blive betydende i fremtiden grundet billigere og lettere tilgængelige
råstoffer.
Kapitel 2 omhandler fremstillingen af FAME fra affaldsfedt. En række forskellige sure katalysa-
torer er blevet testet i en modelblanding af methanol, laurinsyre og trioctanoin. Et antal sulfonsyre-
funktionaliserede ioniske væsker viste nærmest omgående omsætning af laurinsyre til methyllaurat,
mens trioctanoat var fuldt omdannet til methylcaprylat efter et døgn. En type katalysatorer baseret
på sulfatiserede pyrolyserede kulhydrater (SPS), primært sukrose, blev i batch-forsøg fundet veleg-
nede baseret på høj aktivitet grundet høj surhed og er derfor blevet yderligere optimeret. Der er
ikke fundet nogen systematisk afhængighed på aktiviteten for forestring af pyrolyse- og sulfone-
ringsbetingelserne. SPS imprægneret på et åben-cellet Al2 O3 -ekstrudat og mikroporøse SiO2 -piller
(ISPS) var langt mindre aktiv i forestringen end SPS grundet et lavt antal stærke sure centre. ISPS
anvendt som pakket katalysatorleje i kontinuerte reaktorer for forestringen af methanol og laurin-
syre eller oliesyre viste tilsvarende moderat eller lav aktivitet. Forskellige organiske baser som f.eks.
guanidiner er blevet anvendt som katalysatorerer til omestringen af trioctanoin med methanol. Ak-
tiviteten for omdannelse til methyloctanoat er lovende for de stærke baser, men forsæbningen af
fedtstofferne er en uønsket sidereaktion, som kan finde sted ved brug af hygroskopiske baser.
Hydrodeoxygenering behandles i kapitel 3. Reaktionsveje og aktivitet i batch-reaktor er stærkt
afhængige af hvilket supporteret ædelmetal, der anvendes som katalysator, samt af reaktionstem-
peraturen. Pt og Pd er generelt de mest aktive ædelmetaller og har ligeledes højest selektivitet mod
decarboxylering. Stearinsyre og oliesyre blev i de udførte forsøg omdannet væsentlig hurtigere end
tripalmitin. Deoxygeneringen blev ligeledes undersøgt over et katalysatorleje af 2 wt% Pd/Sibunit i
en kontinuert rørreaktor med ved 300℃. En 10 wt% opløsning af stearinsyre gav næsten fuld om-
sætning og selektivitet til n-heptadecan i 20 bar 5% H2 /Ar, mens ren Ar resulterede i deaktivering.
En konstant omsætning på 12 % ren stearinsyre kunne efterfølgende opnås i 5% H2 /Ar. En deak-
tiveringsprofil opbyggedes i reaktoren som funktion af afstanden fra indløbet. I 5% H2 /Ar blev
stearinsyre omsat hurtigere end ethylstearat, der blev omsat hurtigere end tristearin, men skifte til
ren Ar gas førte til deaktivering af katalytsatoren indenfor kort tid med alle de tre reaktanter.
Afslutningsvis gives en perspektivering af arbejdet i kapitel 4.
3
Publications
List of published work and conference contributions during the Ph.D. project.
Scientific peer-reviewed articles in international journals
Anders Theilgaard Madsen, Bartosz Rozmysłowizc, Teuvo Kilpiö, Päivi Mäki-Arvela, Kari Eränen,
Dmitry Murzin, “Step changes and deactivation behaviour in the continuous decarboxylation of
stearic acid”, (2011), submitted
Anders Theilgaard Madsen, El Hadi Ahmed, Claus Hviid Christensen, Anders Riisager, Rasmus
Fehrmann, “Hydrodeoxygenation of Waste Fat for Diesel Production: Study on Model Feed with
Pt/alumina Catalyst”, Fuel, (2011), accepted
Anders Theilgaard Madsen, Helle Søndergaard, Anders Riisager, Rasmus Fehrmann, “Challenges
and Perspectives in the Production of Diesel from Biomass”, Biofuels, 2(4), (2011)
Esben Taarning, Anders Theilgaard Madsen, Jorge Mario Marchetti, Kresten Egeblad, Claus Hviid
Christensen, “Oxidation of Glycerol and Propanediols in Methanol over Heterogeneous Gold Cata-
lysts”, Green Chem. 10, 408-414 (2008)
Poster presentations at international conferences
Anders Theilgaard Madsen, El Hadi Ahmed, Claus Hviid Christensen, “Biodiesel via Hydrotreat-
ing of Fat”, 9th Netherlands’ Catalysis and Chemistry Conference, Noordwijkerhout, The Netherlands,
March 3rd - 5th (2008)
Anders Theilgaard Madsen, El Hadi Ahmed, Claus Hviid Christensen, “Investigation of a model
feed for hydrotreating of oils and fats”, INCA summerschool of Green Chemistry, Venice, Italy, October
12th - 18th (2008)
Anders Theilgaard Madsen, Anders Riisager, Rasmus Fehrmann, “Model feed for catalytic hydrodeoxy-
genation and fats for biodiesel production”, inGAP-NanoCat summerschool, Trondheim, Norway, June
21st - 26th (2009)
Anders Theilgaard Madsen, Anders Riisager, Rasmus Fehrmann, “Esterification of free fatty acids in
biodiesel production with sulphonated pyrolysed carbohydrate catalysts”, EuropaCat IX, Salamanca,
Spain, August 30th - September 4th (2009)
4
Anders Theilgaard Madsen, Anders Riisager, Rasmus Fehrmann, “Understanding hydrodeoxygena-
tion of oils and fats”, 2nd International Biodiesel Conference, Munich, Germany, November 15th - 17th
(2009)
Oral presentations at international conferences
Anders Theilgaard Madsen, El Hadi Ahmed, Claus Hviid Christensen, “Model Feed for the Hy-
drotreating of Fat for Biodiesel Production”, 2nd EuCheMS Chemistry Congress, Turin, Italy, Septem-
ber 16th - 20th (2008)
Popular scientific articles (in Danish)
Anders Theilgaard Madsen, Sune Dowler Nygaard, Hans Ove Hansen, Lene Bonde, “Bæredygtigt
Brændstof fra Affald”, Aktuel Naturvidenskab, 6, 30-33 (2008)
Anders Theilgaard Madsen, Esben Taarning, Claus Hviid Christensen, “Fremtidens Biodiesel: Kom
fedtaffald i tanken”, LMFK-bladet, 1, 26-30 (2009)
Anders Theilgaard Madsen, Anders Riisager, Rasmus Fehrmann, “Udfordringer og strategier ved
produktion af biodiesel”, Dansk Kemi, 90(8), 18-21 (2009)
5
Contents
1 Catalysis, Biofuels, and Diesel 9

1.1 Catalysis - Setting the scene . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.2 Diesel as a biofuel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.3 Fatty acid alkyl esters from fats and oils . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.4 Hydrodeoxygenation of fats and oils . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
1.5 Alternative diesel fuels from biomass . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
1.6 Diesel fuel properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
1.7 Comparison of diesel fuels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
1.8 Summary of the literature and outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
2 Catalytic Production of Fatty Acid Methyl Ester 43

2.1 Heterogeneous catalysts for production of FAME: Introduction . . . . . . . . . . . . . . 43
2.2 Heterogeneous catalysts for production of FAME: Experimental . . . . . . . . . . . . . . 48
2.3 Sulphonic acid-functionalised ionic liquids . . . . . . . . . . . . . . . . . . . . . . . . . 53
2.4 Sulphonic acid-functionalised pyrolysed carbohydrates . . . . . . . . . . . . . . . . . . 56
2.5 Immobilisation of SPS and continuous esterification . . . . . . . . . . . . . . . . . . . . 61
2.6 Basic catalysts for transesterification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
2.7 Heterogeneous catalysts for production of FAME: Conclusion . . . . . . . . . . . . . . . 75
3 Catalytic Hydrodeoxygenation of Fats and Oils 77

3.1 Hydrodeoxygenation of fats and oils: Introduction . . . . . . . . . . . . . . . . . . . . . 77
3.2 Batch hydrodeoxygenation: Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . 80
3.3 Batch hydrodeoxygenation: Results and discussion . . . . . . . . . . . . . . . . . . . . . 82
3.4 Continuous hydrodeoxygenation: Experimental . . . . . . . . . . . . . . . . . . . . . . . 90
3.5 Continuous hydrodeoxygenation: Long-term test & step changes . . . . . . . . . . . . . 95
3.6 Continuous hydrodeoxygenation: Stearic acid and derivatives . . . . . . . . . . . . . . . 110
3.7 Hydrodeoxygenation of fats and oils: Conclusion . . . . . . . . . . . . . . . . . . . . . . 117
4 Outlook and Concluding Remarks 119
5 References 121
6 Included Publications 131
7
Chapter 1
Catalysis, Biofuels, and Diesel 1

For most of the 20th century, mankind and especially the industrialised nations have relied heavily
on the cheap and abundant availability of fossil carbonaceous feedstock for transportation fuels,
chemicals and energy production (in the form of heat and electricity). However, towards the end
of the 20th century and the start of the 21st it has become increasingly evident that humanity faces
a number of unprecedented challenges in terms of future energy resources and consumption [1],
which may be sketched as:
The environmental challenge: The combustion of fossil oil, gas and coal releases the greenhouse
gas CO2 to the atmosphere from carbonaceous resources that have not been part of the global atmo-
sphere or biosphere for millions of years. The effect of greenhouse gases on the atmosphere are well
acknowledged both in terms of global and local climate changes and by a mean global temperature
increase [2]. But also the combustion-related emissions of nitrogen oxides, sulphur oxides and aro-
matic particulates from the consumption of fossil fuels have negative environmental effects as well
as being harmful to the health of human beings. The massive human use of consumable goods af
all kinds also leads to buildup of solid or liquid waste, which must be treated or disposed of in an
ecologically safe and sustainable manner. This recognition is not always fulfilled.
The social challenge: The global population is increasing, and the increasing populations especially
in the developing countries require a higher standard of living, thereby increasing the demand for
electricity, housing, fuel, sanitation, consumables, clean water, and food.
The resource challenge: The projected gradual depletion of easily accessible fossile resources of
oil, gas, and coal, especially mineral oil deposits, is leading to the search for and extraction of less
accessible carbonaceous resources like tar sands, deep-water oil deposits or deposits in ecologically
sensitive areas. These extraction methods will become more costly in terms of both energy, process-
ing and labour, and will take place in increasingly inhabitable areas of the world like the Arctic,
Canada, Siberia, or the like [1]. This happens while the global energy demand is increasing.
The ostensibly ever-increasing demand for fertiliser, plastics, consumables, electrical devices,
chemicals, fuels and energy put pressure on the existing energy supply and infrastructure, but also
spark search for sustainable alternatives and furthers the focus on energy and resource efficiency.
This is especially the case for fuel-consuming vehicles.
The most critical feedstock in the chemical industry and refineries is oil, and transportation fuels
constitute the largest part of the oil consumption [3]. This means that a global shift towards renew-
able feedstocks is necessary [4], and biomass is the only renewable carbonaceous material available
apart from CO2 . Its use as a base for sustainable fuels and chemicals in the future is therefore indis-
pensable.
9
1.1. Catalysis - Setting the scene
In the European Union (EU), the 2003 fuel directive 2003/30/EC mandated that a share of 5.75%
energy content should derive from biofuels in 2010, rising to 10% by 2020 [5]. This was enforced
by the EU in the 2009 directive for promotion of energy from renewable resources 2009/28/EC. At
the same time the EU has opted for a promotion of biofuel production from waste products and a
discouragement of “bad biofuels systems” in the Renewable Energy Road Map - i.e., an exclusion of
biofuels produced with a low or even negative displacement of CO2 from the atmosphere, or where
the production triggers deterioration of sensitive wild-life habitats and ecosystems [6, 7].
Furthermore, as transportation fuels based on fossil feedstocks are usually the cheapest type of
chemicals at all, the production processes for sustainable biofuels must also be economically feasible
and the feedstock as cheap as possible.
1.1 Catalysis - Setting the scene
Efficient, economical, and environmentally sound chemical production usually require the aid of
catalysis to run at acceptable rates. Most of the western standard of living may be said to be based on
the ability of the chemical industry to turn available raw materials efficiently into fuels, chemicals
and applicable energy.
It is useful to notice the definition of catalysis: A catalyst is a substance that takes part in and
enhances the rate of a chemical reaction without being consumed by the reaction. The science con-
cerned with the rate of reactions is chemical kinetics, which also embodies catalysis.
However, reactions have to be thermodynamically favourable to occur, i.e. the reaction must
be able to happen spontaneously - it must have a negative Gibbs-free energy. A catalyst does not
change the position of chemical equilibrium, but it accelerates the tendence of the reaction towards
equilibrium. Popularly coined thermodynamics defines what is possible, and catalysis and kinetics
how and how fast it is possible.
Industrial catalysis as a discipline, especially heterogeneous catalysis, has developed largely con-
currently with the modern methods of industrial petroleum refining and has steadily allowed more
sophisticated use of feedstocks for a growing range of and demand for petroleum products [4, 8]. For
instance, a large part of the current diesel and petrol is produced via respectively catalytic hydro-
cracking and fluid catalytic cracking of heavier fractions from the crude distillation; the hydrodesul-
phurisation of fuels to make sulphur-free fuels; or the production of hydrogen by steam reforming
of natural gas [9].
The shift towards biomass as a feedstock for fuels and chemicals requires development of new
catalytic chemistry [8]: Petroleum almost entirely lacks chemical functionalities that can be attacked
easily. This is suitable for fuels for internal combustion engines. This also means that most refinery
processes are run at high temperatures and with specific catalysts to activate the petroleum com-
pounds. However, biomass is from a petrochemical viewpoint over-functionalised especially with
respect to oxygen, and thus the conversion into applicable fuels or chemicals usually require re-
moval or defunctionalisation of most or all of these functionalities, depending on feedstock type.
Catalysis will likely gain an even higher importance as a discipline, and a profound knowledge
10
Chapter 1. Catalysis, Biofuels, and Diesel
of the science behind it, as well as harnessing it on a technical scale will be of utmost importance for
the future chemical industry.
1.2 Diesel as a biofuel
The production of biomass diesel is growing. The different possibilites for producing diesel oils
from biomass resources have undergone substantial research in the latter years, and a number of
catalytic methods have now been established: Fats and oils can either be transesterified with alcohol
to form fatty acid alkyl esters [10–15], deoxygenated with hydrogen to form n-alkanes [16–24], or
cracked at elevated temperatures to form a hydrocarbon mixture [22, 25–27]. Carbohydrates may be
converted by aqueous-phase-reforming to form unfunctionalised or monofunctional hydrocarbons
[28–31]. Biomass in general may be gasified to syn-gas and undergo Fischer-Tropsch-synthesis [32–
35], or bio-oils can be produced by flash pyrolysis and upgraded by cracking or hydrogenation [36–
41]. The possibilities are sketched in Figure 1.1.
Figure 1.1: An overview of feedstocks, conversion routes and products as plausible components in the production
of diesel from biomass.
The term “biodiesel” usually refers solely to as fatty acid alkyl esters, and again mostly the methyl
esters (FAME) are considered [42]. However others have suggested wider definitions and raised
points about too narrow focus regarding naming the fuels [43–45]. For the sake of clarity the term
11
1.2. Diesel as a biofuel
“biodiesel” will be avoided in the present thesis and the technical-chemical terms for each fuel will
be used instead. The market share of FAME is the largest of the diesel oils from biomass, but it is
nonetheless not the only one.
Pure vegetable oils or fats are not biodiesel fuels and cannot be used as such. Engine and fuel
system require modifications to be able to run on these fuels, and these changes are often illegal
for road vehicles and lead to annulment of engine warranties and specifications for emissions. The
combustion in the engine has a tendency to become incomplete, and both particle and NOx emissions
are reported to rise. Plant oils are also too viscous to be pumped in the fuel and injector systems
[13, 46].
12
1.3 Fatty acid alkyl esters from fats and oils
The most applied method for producing diesel from biomass as well as the one which has received
most attention the latter years is the alcoholysis or transesterification of fats and oils to yield fatty
acid alkyl esters (FAAE). Usually the alcohol is methanol due to lowest cost, easiest separation and
highest reaction rates obtained, whereby methyl esters of the fatty acids (FAME) result [47, 48].
FAME can be used directly in the engine without modification and can be mixed in all ratios with
traditional diesel fuel.
1.3.1 Industrial production of fatty acid methyl esters

Fats and oils consist primarily of triglycerides (TG). These, as well as mono- and diglycerides, can
be transesterified into FAMEs by methanol, see Figure 1.2. The sources of the TGs can be all types
of vegetable oils, animal fats, or waste greases [48, 49]. A successful transesterification leads to two
phases: a bottom phase containing the glycerol and an upper phase containing the FAME.
O O
O O R H3CO O R
OH
O O R' + 3 CH3OH OH + H3CO O R'
OH
O R'' H3CO R''
triglyceride methanol glycerol FAME
Figure 1.2: Transesterification of a triglyceride with methanol, yielding glycerol and fatty acid methyl esters.
1.3.1.1 Transesterification
In industrial practise, the reaction is performed at 60-70℃ and is catalysed by a few wt% of a strong
base - the hydroxides or methoxides of sodium or potassium are used as homogeneous catalyst, KOH
being the norm due to its low price [48, 50, 51]. One molecule of glycerol is obtained as a by-product
for every three FAME molecules produced.
The transesterification is equilibrium-driven and in fact reversible. However, due to use of 2-3
times molar excess of methanol and the immiscibility of FAME and alcohols, above 99% yield of
FAME can usually be obtained within a few hours of reaction [47, 48]. The oil- and FAME-phase
(non-polar) and the alcohol phase (polar) are not directly miscible with each other. This means that,
initially, the reaction is slow and dependent on vigorous mixing of the two phases (e.g. by stirring)
to get a large contact area between the two phases. During reaction, di- and mono-glycerides are
formed, and these compounds act as emulsifiers for the reaction mixture. Once the transesterifica-
tion is nearing completion, the two phases start to separate again (settling) [13]. Industrially the
transesterification is normally performed in a stirred batch-reactor, and as sufficient settling often
requires longer time than the transesterification reaction itself, the settling is often performed in a
separate settling-tank [52].
13
1.3. Fatty acid alkyl esters from fats and oils
1.3.1.2 Esterification
The homogeneous, base-catalysed transesterification is working efficiently, but it gives rise to a num-
ber of challenges. First and foremost oils and fats may contain free fatty acids (FFA). Most fresh
vegetable oils usually do not contain problematic amounts of free fatty acids, but especially waste
fats are challenging: 2 to 7 wt% FFA for used cooking oils, 5 to 30 wt% for waste animal fats and
abattoir waste, and some trap greases can contain over 50 wt% FFA [12, 13]. The FFAs will react
with the basic catalyst and form soap, and when the feedstock contains more than about 0.5-1 wt%
the reaction becomes unacceptably slow due to consumption of base otherwise intended for transes-
terification, and the formed soap emulsifies the phases and hinders the settling and separation after
reaction [48].
For this reason, the FFAs must first be reacted with methanol to form FAME. Esterifications are
acid-catalysed, and in the industry a few wt% sulphuric acid, H2 SO4 is often used for this purpose.
An acidic esterification step is therefore performed prior to the basic transesterification. Sulphuric
acid is the cheapest acid and has the advantage of being both strong and very hygroscopic. The
mixture of methanol and fat is esterified for a few hours at 60-70℃ , around the boiling point of
methanol, with stirring so as to bring the content of free fatty acids below 0.5 wt% (as is shown in
Figure 1.3) prior to the transesterification [48].
Thus, two consecutive reaction steps are needed. The esterification by homogeneous acid leads
to a number of other challenges: First of all the acid solution from esterification must be neutralised
and made alkaline with the addition of the base for the transesterification, and even by the end of
transesterification acid must again be added to neutralise the alkaline alcohol-phase. This naturally
yields large amounts of salts as by-products (usually K2 SO4 and KHSO4 ). This has no immediate use.
Furthermore it is necessary to use new homogeneous catalysts for each FAME batch. Another issue
with this is the purification of glycerol and excess methanol afterwards, which is often contaminated
with water and salt [52]. An immediate decantation removing most of the salt can however be made
after settling.
O O
HO R + CH3OH H2O + H3CO R

FFA methanol water FAME
Figure 1.3: Esterification of a Free Fatty Acid (FFA) with methanol to yield FAME and water.
In principle, the transesterification can be catalyzed by both bases and acids leaving room for
a one-step acid-catalysed process, but the base-catalysed reaction is 3-4 orders of magnitude faster
than the acid-catalysed reaction [10, 47, 48, 53]. Heating the batch to at least 150-200℃ must be per-
formed for completion of the transesterification with acid catalyst within an acceptable time-scale.
This requires working with pressurised equipment (due to the vapor pressure of methanol, which
has a normal boiling-point of 64.6℃ (STP)), and is therefore usually disregarded and considered
uneconomical at least with sulphuric acid.
It is of major interest to substitute the homogeneous catalysts with heterogeneous catalysts. This
14
would avoid the formation of inapplicable salt by-products, lower the expenses for buying catalysts
for each batch of FAME, and prevent the equipment corrosion by dissolved H2 SO4 in the methanol.
Water formed by the neutralisation of H2 SO4 with KOH will contribute to soap formation of FFAs
formed via basic hydrolysis of glycerides. Also, a solid catalyst would allow easier operation in a
liquid-liquid-solid plug-flow-reactor with static mixers, which may be much more desirable with
regards to labour demand, process control, -layout and -supervision, and optimisation of product
quality [48, 54].
1.3.2 Basic catalysts for transesterification
1.3.2.1 Inorganic metal-oxide bases

A wide range of basic oxides have been investigated as catalysts for transesterification of oils and
fats with methanol. Alkali or earth-alkali metal oxides or hydroxides are generally very basic and
thus active, but usually the monometallic metal oxides tend to dissolve in the methanol over time
[55]. Calcining MgO with ZrO2 , however, yielded a bimetallic oxide with high basicity almost un-
affected by dissolution [56, 57]. CuO, CoO, and MnO supported on Al2 O3 have been studied and
yielded up to 97% FAME at room temperature, dependent on the calcination temperature [58]. ZnO
and Al2 O3 −ZnO mixed oxide [59, 60] were investigated as well as rare-earth oxides [61], but both
required high temperatures (at least 200℃) for conversion of vegetable oils and methanol to FAME.
However, bimetallic oxides of calcium (Ca2 Fe2 O5 , CaMnO3 , CaCeO3 , CaTiO3 and CaZrO3 ) have
also been investigated and catalysed transesterification with methanol at 60℃, usually with good
reusability of the oxide [62].
Porous materials like zeolites BEA, USY, and FAU and mesoporous silicalites like KIT-6, ITQ-6,
SBA-15, and MCM-41 have been ion-exchanged with La 3+ , Mg 2+ , or K+ and tested [63–65], how-
ever, high activities for transesterification to FAME are usually only obtained at higher temperatures
(>150℃) with some ion-exchanged porous silicalites, and it is a question if too small pores in the
material is not a hindrance for reaction.
Basic hydrotalcites have been extensively tested. These are comprised of layered units of mixed
Al2 O3 −MgO, and due to their basicity and special porosity they are potential catalysts for transester-
ification to form FAME. Compared to other inorganic porous bases hydrotalcites often have superior
activities for transesterification [66, 67]. Several researchers have characterised their properties and
catalytic activity in relation to composition and calcination conditions [68–72]. Activity enhance-
ments of the hydrotalcites have as well been studied, for instance by doping with Cs, Ba, Sr, or La
[73], by substitution of Al with Fe [74], or substitution of part of the Mg with Co [75]. Embedding
hydrotalcite on a polymer-support was as well successful [76].
1.3.2.2 Organic amine bases

A number of different organic amines and derivatives functionalities have shown to be applicable for
the transesterification of oils and fats: Earlier a number of alkylguanidines or cyclic guanidines have
been suggested as very active catalysts for transesterification, and substituted tetramethyl guani-
dine (R-TMG) or substituted 1,5,7-triazabicyclo[4,4,0]dec-5-ene (R-TBD) nested on PS/PVB polymer
15
supports have been suggested as heterogenised analogoues - due to very strong Lewis-basic sites
[77, 78]. Up to 90% yield after 1 h methanlysis over 1 wt% of the homogeneous guanidines was
obtained, and almost the same could be achieved over the heterogenised analogues [77, 78]. Carbon
nanotubes doped with amines or gem-diamines have recently shown activity in the transesterifica-
tion of triglycerides, however, reaction times were longer even at slightly elevated temperatures (up
to 115℃) [79, 80].
Methyl-substituted phosphazenium catalysts (based on units of P−(N−(CH3 ) x )4 ) could be ap-
plied as a strongly basic catalyst, either unsupported or linked to silica support [81, 82]. The unsup-
ported catalysts yielded over 90% FAME in 15 min [81] at 60℃, while the supported phosphazeium
had much lower activity, dependent on the loading of active compound on SiO2 [82].
Transesterification reactions to achieve various alkyl mono- and diesters have been performed
with moderate success with amino-functionalised SBA-15, which however required 110℃ and 24 h
reaction time [83]. Base-functionalised metal-organic frameworks (MOF) and supported quarternary
substituted ammonium groups for methanolysis of various esters have been investigated [84, 85],
but the highest ester formation activity reported has been 60% after 4 h at 60℃ over 1 wt% of
quarternary substituted ammonium functionalities [85].
1.3.3 Acidic catalysts for esterification
1.3.3.1 Acidic inorganic oxides and derivatives

Metal-oxide supported and zirconium phosphate-supported tungstated catalysts (WO3 /M(P)Ox ) have
been investigated for the esterification of FFA, which readily takes place between 60 and 200℃ - at
the highest temperatures the transesterification reaction of the glycerides is also pronounced [86–
90]. Related to the solid tungstated oxides are the so-called heteropolyacids, based most regularly
on tungsten or molybdenum phosphate. They are strong acids in their protonated form, and may
be supported on metal oxide or even carbon [91]. By substituting some protons with a cation, for
instance Zr0.7 H0.2 PW12 O40 , the acids became even more acidic and catalysed both esterification and
transesterification at 65℃ within 4 to 8 h [92]. Completely protonated forms of tungstated and
molybdated phosphate and silicates (H3 PW12 O40 , H4 SiW12 O40 , H3 PMo12 O40 , and H4 SiMo12 O40 )
had different acidity, which however did correlate with transesterification activity: Molybdated sam-
ples where more active than tungstated ones [93]. This trend changed when supporting the cata-
lysts on silica support [94]. A range of experiments were conducted with 12-tungstophosphoric acid
(H3 PW12 O40 ) supported on various materials, which easily catalysed the esterification of FFA in
waste oils between 25℃ and 60℃ [95–97]. Also, a multifunctionalised mixture of H3 PW12 O40 sup-
ported on Nb2 O5 or Ta2 O5 (and optionally SiO2 ) was active for esterification at benign conditions
[98, 99], and tuning this by adding non-polar alkyl groups to the structure enhanced reaction rates
[100]. Some heteropolyacids may, however, suffer from leaching of the acidic species to methanol
[101].
Related again to the tungstated oxides are sulphated metal oxides. For instance, the sulphated ox-
ides of TiO2 , SnO2 and ZrO2 have been suggested as novel acidic catalysts for the esterification of FFA
with methanol to form FAME, as well as for the simultaneous transesterification of the triglycerides
16
[102–107]. The same oxides were as well evaluated supported on or mixed with high-surface-area
supports of Al2 O3 or SiO2 [102, 108–111]. It should be noted that some authors have reported leach-
ing of sulphate species to the methanol solution from SO42 – /MOx by hydrolysis due to formation of
water by esterification. This is likely dependent on calcination conditions [107].
1.3.3.2 Acidic micro- and mesoporous silicalites
Various Brøndsted-acidic zeolites have been shown to remove FFA via esterification at 60℃ [112].
Higher temperatures were, however, required for Brøndsted-acidic Al-MCM-41 to esterify FFA [113].
By functionalisation of MCM-41 and SBA-15 through impregnation and grafting with respectively
acidic WO3 , Keggin- & Preyssler-type heteropolyacids and sulphonic acid functionalities, esterifica-
tion of FFA with methanol could be achieved [114–116]. However, the grafted sulphonic acid had far
more superior activity to the other grafted acids on mesoporous supports, providing 55% ester yield
at 60℃ after 6 h over 0.4 wt% catalyst. MCM-41 functionalised with organic linkers and tin-triflate
functionalities have been tested for FAME synthesis from methanol and soybean oil at 65-70℃ with
ultrasonic or microwave activation and up to 74% yield at 70℃ after 2 h were achieved [53] due to
the use of microwaves.
1.3.3.3 Organic sulphonic acids
Amongst the acidic catalysts for esterification of FFA one type of organic acid has been studied
most extensively, namely sulphonic acids (R−SO3 H). Simple sulphonic acids soluble in methanol
include methanesulphonic acid (MeSO3 H) and p-toluenesulphonic acid, however solid sulphonic
acids can be prepared by sulphonation in sulphuric acid and thus grafting the functionality onto
polymeric supports; the resulting acid is often as strong as or stronger than sulphuric acid. Commer-
cial resin-type sulphonic acids can be purchased even in amounts applicable on a technical scale, for
instance ion-exchange resins [117], and some are specifically marketed as catalysts for esterification
of FFA. Esterification was performed over various Amberlyst-resins and Dowex HCR-W2-resins be-
tween 30℃ and 65℃ [118, 119], and satisfactory lowering of the FFA amount was achieved within
these conditions. Amberlyst BD20-resin was found superior to Amberlyst 15-resin, both with respect
to recycling of the catalyst and tolerance to water [120]. A study between the commercial cation-
exchange resins NKC-9, 001x7, and D61 proved the former to be superior with FFA and methanol,
yielding about 90% conversion of FFA at 62℃ after 2 h [121].
A number of polymer-based sulphonic-acids have been prepared for the esterification reaction
and studied in the literature, for instance starting from PVA-PS/PVB [122], PS [123, 124], sulphona-
ted PV and PS [125], or mesoporous silica with grafted functional groups [126], or polyanilines
deposited on carbon [127]. The yields of sulphonic acid groups are usually between 0.6 and 6.0
mmol/g of the total supported catalyst mass. In all cases, esterification of FFA may be performed
under normal operating conditions of 60℃ with 1-2 times molar excess of methanol.
Another approach for obtaining a sulphonic acid-containing polymer was first proposed by Toda
et al. [128]. Initially a carbohydrate source, eg. sucrose, glucose, starch, or cellulose is carbonised
(or pyrolysed) under inert atmosphere at around 400℃ for 4-24 h, then the resulting carbonated
17
carbohydrate (now consisting of small graphene layers [129]) is sulphonated with concentrated or
fuming sulphuric acid at elevated temperatures (around 150℃) for 4-24 h. Low surface areas are
obtained of the carbon material itself and its sulphonated derivative, which however increase upon
swelling in hydrophilic solvent. Nonetheless the catalyst is very active in the esterification of FFA
at 60℃ [130, 131]. The sulphonic acid obtained by this scheme had comparable strength to that
of H2 SO4 [132]. By following this procedure, a carbohydrate-based sulphonic acid was obtained
which was investigated for both esterification and transesterification in methanol at 150℃ and 17
bar. While FAME did result from both reactions, leaching of sulphonic acid functionalities occured
at these conditions [133]. If the carbohydrate source is first impregnated on mesoporous silicas,
porosity and hydrophobicity can be tailored to the reaction desired [134], and esterification can take
place at a faster rate due to an increased surface area induced by the support [135].
1.3.4 Alternative catalytic technologies for fatty acid methyl ester production
1.3.4.1 Ionic liquid (trans-)esterification
Brønsted-functionalised acidic ionic liquids (ILs) have recently been suggested as alternative esteri-
fication catalysts [136]. ILs are salts, usually organic, with a melting point below 100℃. Thus they
can be used as catalysts and/or solvents for numerous reactions. Even though ionic liquids have
been reported as a reaction medium for traditional catalysts [137–139], ILs can be functionalised by
incorporating a sulphonic acid functionality, yielding a very strong acid which at the same time can
be used as a liquid reaction medium or stirred/dissolved into the reaction mixture and afterwards
separated [140]. This makes ionic liquids suitable catalysts for the esterification and transesterifica-
tion reactions [136]. The cation may for instance be based on substituted imidazolium or pyridinium
ions functionalised with a sulphonic acid, and anions should preferably be very weak bases such as
bis-triflat or methyl sulfonate [136, 140, 141]. The melting point of the ILs must be below the reac-
tion temperature for FAME synthesis, as the activity is otherwise too low [140]. By functionalising
the IL with more sulphonic acid functionalities the IL may get an even higher catalytic activity [142].
1.3.4.2 Enzymatic (trans-)esterification
Besides chemical catalysts, lipases from different microorganisms can also be applied for esterifica-
tion and transesterification. The conditions are often mild, usually around 25-50℃. Furthermore
lipases may catalyse both the transesterification and the esterification [47]. For the lipases to be
applied in industrial FAME production they must be immobilised on a carrier material, preferably
porous to give a high surface area [143–146]. Unfortunately the catalytic activity is not to date re-
ported to be of the same magnitude as the basic and acidic catalysts [10, 50]. Activity can sometimes
be enhanced by using a co-colvent, though [147, 148]. Some lipases may as well be rendered inactive
by too high concentrations of methanol, and continuous addition of methanol, genetically altering
the lipases or switching alcohol can therefore be necessary [149, 150].
18
1.3.4.3 Supercritical (trans-)esterification

As an alternative to using catalysts, the esterification and transesterification may also be performed
with supercritical methanol (PCr = 81 bar, TCr = 234.5℃) [151]. Adding for instance carbon dioxide
or hexane as co-solvent or an inorganic base as catalyst to the supercritical methanol can furthermore
improve the yield of FAME [152–155]. However, the super-critical conditions require equipment
suitable for high pressures and temperatures, and the energy and economical balances for this may
not be advantageous.
1.3.5 Glycerol
The main by-product of the transesterification of fats and oils with alcohol is glycerol, as seen in
Figure 1.2 on page 13. Producing one ton of FAME yields about 100 kg of glycerol, and it has
previously been estimated (in 2006) that the production of glycerol would be 6 times more than
the demand in 2020 due to the production of fatty acid methyl esters [156]. While this figure may
be exaggerated today as the growth in production of FAME has not been as explosive as projected,
valorisation can still be an important contribution to the overall economy of the FAME process.
A few suggested reactions of glycerol into more valuable chemicals are seen in Figure 1.4, inclu-
ding reduction, dehydration, oxidation, or complete gasification into synthesis gas for production of
for instance methanol. A number of reviews on glycerol conversion to value-added chemicals are
available [157–159].
OH HO OH
OH
HO OH 1,3-Propanediol
OH
Glyceric acid 1,2-Propanediol
O OH
O HO OH O
HO OH Glycerol Acrolein
Dihydroxyacetone
OH
OH
O Acrylic acid
OH O
OH
Lactic acid
O Acetol
Figure 1.4: Overview of some of the reactions of glycerol.
1.3.5.1 Dehydration
Glycerol can be dehydrated into acrolein over acidic catalysts. Both acrolein and its oxidised deriva-
tives (acid and esters) are industrial polymer building blocks produced annually on megaton-scale
[160]. Various solid acids have been proposed as catalysts for this, eg. acidic zeolites in aqueous
or gaseous phase [161–163], over silica-alumina or vanadium phosphate acids [164], Nb2 O5 [165],
HPAs [166], or even near-critical water dehydration to acrolein. Acrylic acid may afterwards be ob-
19
tained by oxidation of acrolein [167]. The dehydration intermediate acetol (hydroxyacetone) may
also be employed. A scheme employing catalytic reactive distillation has been proposed to give high
acetol yields in the gas-phase [168].
1.3.5.2 Reduction
Selective reduction of glycerol with hydrogen can yield either 1,2-propanediol (1,2-PD, propylene
glycol) or 1,3-propanediol (1,3-PD), usually in a mixture dependent on the catalyst used while 1- or
2-pronanol may result as by-products. 1,3-PD can be used as a building block for polymers or as an
antifreeze, while 1,2-DP has use as solvent or antifreeze [157, 158]. High selectivity towards either
propanediol is challenging and difficult to obtain, while a resulting mixture of propanediols may
be difficult to separate. High or complete selectivity towards 1,2-PD have been achieved by using
Raney copper [169], Cu2 O or ZnO supported catalysts [170], or Ru/C possibly together with a cation
exchange resin [171–173]. A selectivity of 49% for 1,3-PD has been achieved over Rey Ir1-y Ox /SiO2
in aqueous solution [174], while also tungstic acid improved the selectivity towards 1,3-PD [170].
It has been suggested that adding Brønsted acidic catalyst enhances selectivity towards 1,3-PD, for
instance over Pt/WO3 /ZrO2 [175].
1.3.5.3 Oxidation
Being a triol glycerol can undergo oxidation resulting in a network of various products. Dihydrox-
yacetone (DHA) and glyceric acid, the primary intermediates in the glycerol oxidation network find
use in the pharmaceutical or food industry and must be obtained by selective oxidation - a chal-
lenging task for catalytic chemistry [157]. On an industrial scale the oxidant should preferably be
atmospheric oxygen. Especially supported Pt, Pd or preferably Au have been shown effective for the
aqueous conversion, often at high selectivities. In excess base and aqueous solvent 100% selectivity
towards the glycerate anion was achieved over 1% Au/C [176]. Without added base and over acidic
Au-Pt/H-MOR zeolite, 81% selectivity at full conversion has been achieved [177]. Glyceric acid can
also be synthesized by anodic oxidation of glycerol using an Ag2 O electrode [178]. Oxidation is se-
lective towards the DHA over supported Bi-Pt in acidic media [179, 180], while anodic oxidation
yielded only 25% selectivity as optimum [181]. The alkaline alcohol mixture separated out from
FAME production may also be oxidised directly over Au/TiO2 to form methyl glycerate, dimethyl
tartronate or dimethyl mesoxalate [182].
Polylactic acid has potential to become the green and biodegradable plastic of the future, and
glycerol can be converted directly to lactic acid by fermentation. Recently the lactate anion was
obtained by alkaline hydrothermal reaction of glycerol at 280℃, albeit by consuming equimolar
amounts of base [183]. Conversion via DHA or glyceraldehyde may be more efficient, for instance by
electrochemical oxidation and isomerisation [184], or as recently suggested via Lewis-acidic catalysts
[185, 186].
Microorganisms can selectively perform some of the above-mentioned reduction or oxidation
reactions by fermentation routes often with excellent selectivity [159], however these usually work
only at low temperatures, low aqueous concentrations and reaction rates, while from an industrial
20
point of view concentrated streams and high turnovers with minimised reactor volumes and more
tolerant chemical catalysts are preferred.
1.3.5.4 Derivatisation
Valorisation of glycerol could also be afforded by derivatisation of the molecule itself. Suggested re-
actions in the literature include etherification with iso-butenes to give octane enhancers for gasoline
[156, 187, 188], acetalisation with ketones and aldehydes [189], or condensation with carbonic acid
to afford monoglyceride carbonate as an oxygenated diesel additive [190]. The catalytic synthesis
of etherified oligomers and polymers of glycerol have as well been described for upgrading of the
compound [191].
1.3.5.5 Synthesis Gas

Glycerol may also be steam reformed into synthesis gas (syn-gas). This has been studied for in-
stance over transition-metal catalysts of Pt, Pt-Re, Rh, or Co supported on for instance C, Al2 O3 or
CeO2 −Al2 O3 [192–195]. Depending on conditions, by-production of tar-like compounds can inter-
fere and lead to clogging of systems and deactivation of reforming catalysts [196]. The syn-gas ob-
tained from steam reformed glycerol can furthermore be used for hydrogen production, for instance
for use in fuel cells, or it can be converted into methanol via the methanol synthesis or hydrocarbons
via the Fischer-Tropsch-synthesis (FTS) [197].
21
1.4. Hydrodeoxygenation of fats and oils
1.4 Hydrodeoxygenation of fats and oils
The hydrodeoxygenation of fats and oils is an alternative and a quite different approach for up-
grading fatty feedstock. It was first industrialised by Finnish Neste Oil at their refinery in Porvoo,
Finland, who refer to the product as “NExBTL” (Next generation Biomass-To-Liquids) as described
by Koskinen et al. [43] and Mikkonen [45]. Other names suggested are “green diesel” or “renew-
able diesel”, albeit the latter seem to be the preferred term [198]. The approach usually requires
hydrogen, and may be idealised as shown in Figure 1.5. Based on recent literature the catalysts for
upgrading fats and oils can be divided into three categories, namely 1) supported sulphided metals,
2) supported (noble) metals or 3) acid-base catalysts for cracking-type deoxygenation.
O
OO 13 H2 6 H2O
OO +
C3H8
O
Figure 1.5: Hydrogenation (full reduction) of a triglyceride to yield propane, water and long-chain alkanes.
From literature it is evident, however, that not only a full reduction of the feedstock is possible,
but other reactions compete as well, namely decarboxylation of the carboxylate functionality (ester
or carboxylic acid) to yield an n-alkane and CO2 , and decarbonylation to yield a 1-n-alkene, water,
and CO. The loss of the carboxylate functionality as CO or CO2 results in shortening the fatty acid
carbon chain by one carbon of the resulting alkane or 1-alkene, contrary to the full reduction which
conserves the number of carbon atoms in the carbon chains.
The latter two routes are illustrated in Figure 1.6. The preferential reaction routes are largely
determined by the conditions of the process and the specific catalyst used. Generelly noble-metal
catalysts have notable preference for the decarboxylation and decarbonylation pathways, while the
sulphided metal catalysts tend to favour the full reduction, although the picture is often more va-
ried. The deoxygenation over cracking-type catalysts does, due to higher temperature and often high
acidity, also yield cracking-type products (see Figure 1.8 on page 25).
O
H R= + H 2O + CO CO2 + H R
HO R
O
H R'= O O R' H R'
H R''= + "C3H2" + 3 H2O + 3 CO O O R'' 3 CO2 + "C3H2" + H R''
H R'''= O R''' H R'''
Figure 1.6: Decarbonylation (left-hand side) and decarboxylation (right-hand side) idealised reaction schemes for
deoxygenation of free fatty acids and triglycerides in oils and fats. R= denotes a terminal double-bond.
22
1.4.1 Supported sulphided metal catalysts
Kubička and co-workers evaluated the conversion of rapeseed oil over sulphided CoMo and NiMo
supported primarily on Al2 O3 , at 250-350℃ and 7 to 110 bar H2 . The activity of sulphided
NiMo/Al2 O3 superseeded that of separate Mo/Al2 O3 and Ni/Al2 O3 in liquid phase HDO [199], and
full conversion of rapeseed oil over three commercial sulphided NiMo/Al2 O3 at 70 bar H2 was only
achieved at >310℃, while lower temperatures left free fatty acids and triglycerides in the product
mixture [200]. At 20-110 bar H2 and 300-320℃ sulphided CoMo supported on mesoporous Al2 O3
and MCM-41 it was observed that higher yields of alkanes were achieved with lewis-acidic Al2 O3 as
support, but building Al into the framework of MCM-41 to make the support acidic improved the
yields [201, 202]. The primary reaction route of sulphided NiMo/Al2 O3 was towards full reduction,
which however decreased on behalf of decarbonylation/decarboxylation reaction with rising tempe-
rature and increasing conversion, while unsulphided Ni/Al2 O3 had the highest selectivity towards
decarbonylation/decarboxylation products [203].
The co-treating of sunflower oil and heavy vacuum oil (from petroleum distillation) over sul-
phided NiMo/Al2 O3 was studied by Huber et al at 300-450℃ at 50 bar in a tubular flow reactor,
which yielded a maximum carbon yield of 71% of C15 -C18 n-alkanes from the oil. Cracking reac-
tions rose with temperature, as did the isomerisation of resulting n-alkanes [18]. Co-treating was
also evaluated by Šimáček et al. with respectively 0% or 5% rapeseed oil in vacuum-gas oil (from
petroleum distillation) at 400-420℃, which yielded similar fuel properties [22].
Donnis and co-workers confirmed that both reduction and decarboxylation can take place in the
hydrotreatment over sulphided NiMo/γ-Al2 O3 , and selectivity to decarboxylation products from
rapeseed oil of as much as 64% at full conversion was obtained during deoxygenation in light gas oil
at 45 bar H2 and 350℃ [41]. The treating of pure palm oil was recently performed by Guzman et
al over NiMo/Al2 O3 at 40-90 bar H2 and 350℃. The selectivity to even-carbon n-alkanes (complete
reduction) was observed to increase with H2 pressure [21].
Krause et al studied the HDO of heptanoic acid, heptanol, methyl and ethyl heptanoate over
sulphided CoMo/Al2 O3 and NiMo/Al2 O3 [204, 205]. The treatment was performed at 250℃ and
15 or 75 bar H2 . NiMo, but not CoMo was very sensitive to the sulphidation conditions using either
H2 S and CS2 , while the unsulphided catalysts were neither very active nor selective to C6 and C7
alkane formation. Acid functionalities on the catalysts mediate esterification and dehydration of
some intermediates [205, 206].
Sulphided catalysts require that sulphur must be added as for instance H2 S or CS2 , or the feed-
stocks must be co-treated with sulphur-containing compounds like refinery gas oils - otherwise the
sulphided catalysts deactivate (desulphidise) by incorporation of sulphur in the product stream
[205]. The comsumption of hydrogen should optimally be minimised to save cost for the treatment,
so higher selectivity for decarboxylation or decarbonylation as described by Donnis et al. is advan-
tageous [41] - if methanation does not prove to be a problem. A great asset of this procedure is the
absence of formed aromatics.
23
1.4.2 Supported transition-metal catalysts

The deoxygenation of fatty feedstock over supported transition metal catalysts has been studied
mostly by the group of Murzin and Mäki-Arvela at Åbo Akademi. Usually conversion of saturated
feedstock like stearic acid, ethyl stearate, and tristearine to alkanes have been studied, normally
between 270-360℃ in inert gas or in a mixture with H2 at up to 40 bars.
In a screening of a range of active catalyst metals and supports Snåre et al. found that Pt or Pd
supported on carbon were most active and had the highest selectivities for decarboxylation of stearic
acid to n-heptadecane [17]. In connected works Pd/C was used as catalyst to convert tristearine,
ethyl stearate, and stearic acid to alkanes, which was modelled by Snåre, Kubičková and co-workers
[16, 207], as well as the kinetics for deoxygenation of other fatty acids [23, 208]. It could be concluded
that conversion of the carboxylic acid itself took place almost exclusively via decarboxylation, but
ester functionalities underwent more complicated mechanisms in the conversion to n-alkanes [16,
207].
In semibatch reaction-mode, Rosmysłowicz et al. studied deoxygenation of C18 tall oil fatty acid
over Pd on Sibunit carbon and found considerable impact on the conversion to C17 -hydrocarbons
from the concentration of hydrogen in the gasphase [20]. Crocker and co-authors studied conversion
of triglycerides over carbon-supported Ni, Pd and Pt at 350℃ without added hydrogen and ob-
served CO2 , CO, CH4 and small hydrocarbons in the gasphase, a range of liquid alkanes and heavier
paraffins, as well as free fatty acids as intermediates [209]. By studying the transformation of methyl
octanoate and methyl stearate over Pt/Al2 O3 at 330℃ in He and H2 gas it was found by Do et al. that
H2 suppresses formation of higher self-condensates of both compounds and that decarbonylation is
always the primary reaction route [210].
Palladium and platinum as catalysts have also been investigated on a number of nanoporous and
microporous supports. Lestari et al. investigated deoxygenation of stearic acid over Pd on SBA-15
[211], Ping and co-authors used Pd on and in a mesocellular SiO2 -foam also to deoxygenate stearic
acid [212], while Kikhtyanen and co-workers applied Pd/SAPO-31 for the conversion of sunflower
oil [24]. Hancsók and co-authors deoxygenated saturated vegetable oil over Pt/H-ZSM-22/Al2 O3 to
isomerised alkanes [213]. These studies indicated that acidic functionalities of the support lead to
modest isomerisation of the formed n-alkanes which improves the cold properties of the fuel, but
undesired cracking and deactivation may also take place if the temperature gets too high [213].
From literature it appears that carboxylic acids deoxygenate via decarboxylation (- CO2 ), while
esters undergo decarbonylation (- CO) - this can however be hard to determine in an experimen-
tal setup due to the Water-Gas Shift-equilibrium (WGS), as shown in Figure 1.7 a). The noble metals
used as catalysts usually have high selectivity towards decarboxylation or decarbonylation pathways,
which minimises the immediate use of hydrogen. Hydrogen is, however, needed to avoid deactiva-
tion by aromatisation or CO poisoning. This leads to another problem, namely the formation of
methane from CO or CO2 and 3 or 4 molecules of H2 , see Figure 1.7 b) & c), a highly undesirable
situation. The methane may be burned for process heat or steam reformed back to H2 and CO, but
the latter is costly and alone the separation of the gases may be tedious.
Deoxygenation over noble-metal catalysts of real feedstocks like vegetable oils or waste fats has
24
a) CO + H2 O CO2 + H2
b) CO + 3 H2 CH4 + H2 O
c) CO2 + 4 H2 CH4 + 2 H2 O
Figure 1.7: Side reactions taking place over a) Water-Gas Shift (WGS), b) & c) CO & CO2 methanation, respectively.
not been studied extensively and could be challenging. First of all they contain impurities like salts,
sterols, phospholipids etc., and secondly the triglycerides in real fatty feedstock are always more or
less unsaturated. Deactivation is suspected to form especially from unsaturated fatty acids, for in-
stance from cyclisation, dehydrogenations and Diels-Alder reactions [20, 208, 214]. This means that
hydrogen saturation of double-bonds in the fatty acids is also a necessary step in the deoxygenation.
1.4.3 Cracking-type catalysis

The deoxygenation of methyl octanoate over H-ZSM-5 at 500℃ resulted lighter hydrocarbon gases
and aromatisation, the latter of which proceeded through self-condensation products like tetrade-
cane, 8-pentadecanone and octyl octanoate. The aromatisation selectivity as well as the conversion
were much lower at 400℃ [215], Interestingly, in the cationic form of zeolite X, methyl octanoate
was converted to longer hydrocarbons (C14 −C16 ) over basic CsNaX, without forming any aromatics,
while the weakly acidic NaX resulted marked amounts of aromatic production before deactivating
the condensation pathways [216].
H-ZSM-5 [217], rare-earth-modified Y-zeolite [26], MCM-41, and mesoporous silica [25] were
used to crack palm oil at 400-500℃. Under these circumstances, the reaction mainly yielded gasoline-
range hydrocarbons, and the diesel yield was at best moderate. The selectivity to diesel-range hydro-
carbons rose with decreasing cracking temperature, however the conversion dropped as well.
Na et al. studied the decomposition of oleic acid in an autoclave at 400℃ over basic hydrotalcites,
yielding both the decarboxylation products heptadecane and heptadecene as well as cracking pro-
ducts like alkanes and minor carboxylic acids [27]. Quirino et al. studied soybeen oil pyrolytic
vapour decomposition over Al2 O3 -supported tin- and zincoxides between 350 and 400℃, which
yielded decarboxylation and cracking products of similar composition as Na et al. [218].
O
O O H2O, CO, CO2
O O + Light Hydrocarbons
Coke
O
Figure 1.8: The deoxygenation and cracking-type products (alkanes and olefins, light gases, aromatics) formed
during conversion of glycerides and fatty acids over acidic porous materials.
Generally the reported yields of hydrocarbons are low in most cases. The liquid yields are mostly
in the gasoline range and with less diesel and varying amounts of gases produced, as sketched in
Figure 1.8. The aromatisation inside and on the catalysts leads to deactivation by coking, however
this problem could be solved if the cracking reactions are performed in a fluid catalytic cracking
25
system where burning the coke in a regeneration zone is possible (provided that the catalyst can
endure this treatment). The aromatics in the product diesel pool are problematic as well and should
be minimised for environmental reasons and due to lowering of the cetane number of the diesel
[198]. The vapors of uncatalysed pyrolysis of palm, soybean, and castor oil at 350 to 400℃ led to
a catalytic H-ZSM-5 bed was reported to yield hydrocarbon condensates at over 98% yield without
aromatics, but containing shorter-chain fatty acids as a result of C-C scission [219].
26
1.5 Alternative diesel fuels from biomass
The potential supply of fats and oils with present technology is only sufficient to cover a minor frac-
tion of the present global diesel demand. Utilisation of more abundant biomass resources, namely
lignocelluloses, such as straw, wood, fibers etc. is therefore imperative. With regards to diesel three
overall strategies exists for this purpose, for instance as sketched by Lange [220]:
1. Upgrading of bio-oils or bio-tars from flash pyrolysis
2. Aqueous-phase reforming of carbohydrates
3. Gasification and Fischer-Tropsch-Synthesis (FTS)
1.5.1 Upgrading of bio-oils

One auspicious way of dealing with lignocelluloses relevant for diesel production is to convert the
biomass via so-called flash pyrolysis, i.e. fast heating of finely ground biomass to 400-650℃ at short
residence times, often under inert gas and with sand or other ceramics as a heat carrier. This protocol
produces a gas fraction containing various carbonaceous gases, a residual char fraction and various
amounts of condensable tars, which are called bio-oils or flash pyrolysis oils. Depending on the
pyrolysis conditions the yield of bio-oil may be over half of the biomass input mass.
Another way of producing bio-oils is by hydrothermal upgrading (HTU), for instance the pro-
prietary technology CatLiq, which takes place in aqueous solutions or suspensions at near-critical
conditions of water (TCr = 374℃, PCr = 221 bar). This treatment leads to dehydration, cyclisations
and rearrangements of for instance carbohydrates [221]. Lignocellulose networks are furthermore
degraded and the products as well undergo rearrangements, decarbonylations, and decarboxylations
[222, 223]. Catalysts, for instance acids, bases or metals can enhance the reactions [223–226]. This
may also be applied to aqueous sludge or organic waste streams for instance from food processing
plants [227].
Combusted / Gasified /
Steam reformed Sequestered
- CH4 -C + H2
Heat
CxHyOz CH4, CO2, H2O, C (coke), CxHyOz CxHy + H2O, CO, CO2
cat.
Biomass Separation Deoxygenated
Bio-oil
- CO2, H2O
Figure 1.9: Principle behind production and upgrading of bio-oils via flash pyrolysis.
Such bio-oils are comprised of hundreds of different chemical compounds, spanning ketones,
aldehydes, carboxylic acids, aromatic ethers, phenols, and alcohols, as well as poly-functional mole-
cules. The oils often contain an emulgated amount of water. They are therefore unsuited for di-
rect use as engine fuels, but they may be catalytically hydrodeoxygenated to various types of fuels:
Petrol, jetfuel or diesel - this depends on the distillation properties of the final hydrocarbons. This is
sketched in Figure 1.9
27
1.5. Alternative diesel fuels from biomass
Upgrading of bio-oils from flash pyrolysis bears some resemblance to HDO of fatty feedstocks,
and the catalysts can also be divided into the same categories, namely cracking-type, sulphided
metals and transition metals, with the notable inclusion that bio-oils may also be treated over a
catalyst during the pyrolysis itself - so-called catalytic pyrolysis. Necessary H2 for the reactions
should be obtained cheaply, possibly by reforming other biomass, natural gas, or the resulting light
gases in the pyrolysis in connection with the HDO upgrading. However the consumption of H2 is
potentially much higher than that for deoxygenation of triglycerides due to a much higher content of
both oxygen (high O/C-ratio) and aromatics as can be rationalised from Figure 1.10 [19, 228, 229].
CH2OH CH2OH CH2OH
OH OH
H3CO OCH3 OCH3 OCH3 OH OCH3 OH
OH OH OH CH3
Sinapyl alcohol p-Coumaryl alcohol Coniferyl alcohol Guaiacol Catechol Anisol Phenol p-methylacetophenone
Figure 1.10: Left-hand side: Monolignols, the primary monomers that constitute the lignin network in biomass;
Right-hand side: Different phenolic compounds used for studying upgrading of bio-oil [230].
Upgrading should remove part or all of the oxygenates (decrease O/C-ratio) and saturate aro-
matics, at least partly (increase H/C-ratio). The presence of aromatics in diesel fuels is limited by
fuel standards to some extent [45, 198]. The difference from the well-defined triglyceride molecules
largely complicates HDO of bio-oils. Removal of much of the oxygen in the bio-oils as well as wa-
ter can be a pretreatment to hydrocracking in conventional refinery operation, for instance together
with normal petroleum gas-oil feeds, justifying two-stage processes [231, 232]. Small amounts of
oxygen in alcohols or ethers may be tolerated in the fuel - in petrol this may enhance the fuel quality
in terms of octane number [229].
The conditions for maximal yield of respectively coke, tar, or bio-oils can be tuned via the reaction
conditions and retention time for the pyrolysis: Low temperature and long residence time optimises
the yield of char, high temperatures and long residence times optimise the yield of smaller gases,
while moderate temperature and short residence time is most optimal for bio-oil production - see for
instance Bridgwater, Huber or Lange [19, 47, 220, 233].
1.5.1.1 Cracking & catalytic pyrolysis

Aho and co-workers studied the pyrolysis of pine sawdust at 400℃ followed by upgrading at 450℃
over H-Y, H-BEA and ferrierite as their Fe-modified counterparts, yielding between 43 and 53 wt%
liquid [234]. The catalytic pyrolysis over quartz sand, zeolites H-BEA, H-Y, H-ZSM-5 and H-MOR
at 450℃ gave maximal liquid yield of 27 wt% over quartz sand but highest removal of oxygen over
H-ZSM-5 [235]. All materials could be regenerated by oxidation of formed coke [234, 235].
Valle et al. studied valorisation of bio-oil by heating to 400℃ for deposition of pyrolytic lignin
before cracking the bio-oil at 500℃ on H- and Ni-modified Y-zeolites, maintaining low liquid yields
and high yield of C2 −C4 and C5+ gases in the catalytic reaction - however CH4 formation is almost
completely avoided [236]. Co-treating of the residue from atmospheric petroleum distillation with
28
10 wt% pretreated bio-oil containing up to 28 wt% oxygen at 520℃ over a standard FCC catalyst
was found feasible and yielded normal ranges of light hydrocarbon gases, petrol and coke, likely due
to the preceding removal of the most reactive oxygen functionalites in the bio-oil [237].
H-ZSM-5 was more effective than CoO/MoO3 for the catalytic pyrolysis of lignin at 650℃ due to
its ability to crack the aliphatic lignin linkers to small hydrocarbon gases [238], while catalytic pyro-
lysis of glucose at 400-600℃ over H-ZSM-5 yielded mostly aromatics and coke - low coke formation
and high mono-aromatics formation was optimal at short residence time and high temperature [37].
For glucose and over sugars at 600℃, the type of zeolite catalyst is greatly affecting the yields of
aromatics, coke and gases [239].
1.5.1.2 Supported sulphided catalysts
Generally, the deoxygenation over sulphided metal catalysts has been studied at 200-400℃ and
10-200 bar of H2 . Delmon et al. investigated sulphided CoMo/C as a catalyst for the HDO of p-
methylphenol, p-methylacetophenone, guaiacol and di-ethyldecandioate. Presence and higher con-
centrations of H2 S in the feed was found to enhance most of the deoxygenation reactions, however
not the esters. Diaminopropane was found to inhibit some of the deoxygenation reactions, especially
acid sites, which were involved in the ester conversion. It was furthermore found that increasing
the loading of active phase on the support did not correlate linearly with an increase in the activity
due to larger size of CoMo particles [240–243]. It was found that carbonyl compounds exhibited 1.
order rate in reduction behaviour at 200-300℃ [244], while in a batch-reactor between at 350-390℃
the HDO experimental reaction rate order was found to 2.3 with respect to the total oxygen content,
while it was not very dependent on hydrogen pressure.
I batch systems it has been found that the hydrogen pressure only had minor influence on the re-
action rate over sulphided CoMo Deoxygenation of various oxygen-containing functionalities showed
different reaction orders [245]. This was also observed over sulphided NiMo/Al2 O3 , where French
and co-authors found that temperature controlled the reaction rate at up to 360℃ , high H2 pressures
and. At higher temperatures the loss of carbon by formation of carbonaceous gases (CO, CO2 , CH4 )
was considerable. A moderate to high degree of deoxygenation may be attractive, such as leaving a
few wt% of the bio-oil as oxygen in the form of ketones or ethers [232].
Sulphided CoMo/Al2 O3 was less active than sulphided NiMo/Al2 O3 for the HDO of guaiacol in
batch mode at 200-350℃, but the CoMo was more selective to HDO products [246]. Different phe-
nolic and aliphatic oxygenates were converted over supported sulphided CoMo and NiMo catalyst,
and reaction behaviour related to different mechanistic deoxygenation pathways of either pheno-
lic or aliphatic alcohol [247]. Dilchio Rocha and co-authors suggested using pressurised hydrogen
over FeS at 520℃ in the first stage of single- or two-stage so-called hydropyrolysis of biomass, with
supported sulphided NiMo as catalyst in the potential second bed. This procedure reduced the for-
mation of coke and gas and enhanced bio-oil formation with less oxygen functionalities compared to
zeolite catalysts [248].
29
1.5.1.3 Supported metal catalysts
The upgrading of bio-oils by supported metal catalysts is sparsely studied compared to sulphided
catalysts - maybe due to suspected deactivation by aromatics and coking - the former largely present
in the bio-oils and the latter likely formed herefrom during the HDO process. Elliot et al. studied
HDO over Pd/C followed by hydrocracking in a flow system and optimal HDO behaviour was found
to be 340℃ at 136 bar H2 [231]. Heeres and co-workers also contributed to the field: At 250 or 350℃
and pressures of H2 at 100 or 200 bar Ru/C was found to be superior to Pt/C and Pd/C as well as Ru
supported on SiO2 or Al2 O3 and more active than alumina-supported sulphided NiMo and CoMo.
The highest oil yield was 60 wt% corresponding to removal of 90% oxygen, with the formed liquid
being much less acidic, containing less water, and having a HHV of about 40 MJ/kg [38].
By employing a two-stage hydrodeoxygenation strategy also using Ru/C as a catalyst, the in-
termediate and produced oils were characterised by a range of analytical techniques to elucidate
content of different chemical functionalities. During 3-stage hydrodeoxygenation it was observed
that especially carboxylic acids, aldehydes, diols, phenolic groups were removed from the bio-oils
from hydrodeoxygenation [249]. By upgrading of bio-oil (O/C = 0.53, H/C = 1.55) in an autoclave it
was found that the recovery of carbon in the oil phase increased from 55% at 230℃ to 70% at 340℃,
the overall H/C ratio decreased from 1.71 to 1.6, and the O/C ratio decreased from 0.33 to 0.18 over
a 5 wt% Ru/C catalyst [237]. A number of different commercial and laboratory Ru/C catalysts have
been tested, and deactivation observed to take place by coke formation in the pore system. The deac-
tivation was dependent on fabrication procedure, possibly yielding different reactivities depending
on particle size of ruthenium on the surface [250].
1.5.2 Aqueous-phase reforming

An different approach to fuels and chemicals production has been pursued especially by the group
of James Dumesic and co-workers, who have studied the upgrading of sugars and polyols to hydro-
carbons by reforming biomass in water - aqueous-phase reforming (APR). As can ben seen in Figure
1.11, the principle behind this scheme is the simultaneous catalytic reduction and reforming of su-
gars and sugar derivatives. Polyols are very soluble in water, usually fully soluble, but once they
get sufficiently reduced for instance to monofunctional hydrocarbons, at the same time lowering
their density, they spontaneously separate to an upper, non-polar hydrocarbon-phase low in oxygen
content. A sweep-stream of hydrocarbons may be used to enhance this separation [30].
Reforming reactions Reduction reactions

O OH OH O OH OH
6 H2O + 7H2
6 CO2 + 12 H2 n-C6H14 + 6 H2O
C-C-bond cleavage Reduction with H2
C-O-bond cleavage OH OH OH OH OH OH
Reforming with H2O
Glucose Glucose Hexane
Water-gas shift
Figure 1.11: Principle behind aqueous-phase reforming of glucose.
30
1.5.2.1 Aqueous reforming of polyols

A range of catalysts have been suggested for the APR of sorbitol to optimise the hydrocarbon yields,
but Pt/SiO2 −Al2 O3 , Pt/ZrO2 or Pt-Re/C have been found advantageous [251, 252], and it is sug-
gested that a bifunctional catalyst containing a metal and an acid functionality is needed to facilitate
all reactions [251, 253]. Pt or Pt-Re alloys facilitate the dehydrogenation and C-C-scissions to liberate
CO species, which are then water-gas-shifted to CO2 and H2 [252]. Pt has a high activation barrier
for breaking the C-O bond meaning that methanation is supressed.
The APR is usually performed around 200-250℃ and at pressures of 18-40 bar. Due to the
fact that the initial reaction takes place in aqueous environment and at very mild conditions, many
known problems of reforming are strongly minimised. Aromatisation reactions are suppressed in
the aqueous environment. The water pushes the water-gas-shift equilibrium towards CO2 and H2 , so
deactivation by CO does not take place. The resulting organic liquid contains a mixture of different
monofunctional hydrocarbons such as ketones, carboxylic acids, alcohol, and more or less saturated
heterocycles in the range of C4 -C6 , while gas-phase-products constitute CO2 and minor amounts of
C1 -C6 alkanes - some of these can be further upgraded [252, 254].
1.5.2.2 Condensationens prior to reduction

Carbohydrates are built up from units of maximally 6 carbon atoms, meaning that longer hydro-
carbons with more than six atoms cannot be obtained without C-C-coupling-reactions. This may
be done prior to APR or reduction, as carbohydrates, being polyols with a carbonyl group, may un-
dergo crossed aldol condensation to couple carbonylic compounds, ketones and aldehydes, to other
carbonyls. 5-(hydroxymethyl)furfural (HMF), which can be made via dehydrations and isomerisa-
tions from fructose, glucose or cellulose [255], is suggested as a sugar derivative that may undergo
such reaction. For instance either one or two molecules of HMF may be reacted with acetone over
bi-metallic solid oxides, as seen in Figure 1.12, yielding either a C9 or C15 building block that can be
reduced with hydrogen [29, 256].
OH O
OH O O
OH O + OH O + OH O OH
+ H2
HO O O O O O
HO OH n-C15H34
OH
Glucose HMF F-3HB DF-PDO Pentadecane
Figure 1.12: Aldol crossed-condensations and reductions from carbohydrate and acetone. Dehydrations in all the
reactions are not shown. HMF: 5-(hydroxymethyl)furfural; D-3HB (furfurylalcohol)-3-hydroxybutene; DF-PDO:
1,5-(di-furfurylalcohol)-1,4-penta-dien-3-one.
1.5.2.3 Product condensation & reduction

A final step in the biomass conversion is the reduction of remaining monofunctional hydrocarbons
and unsaturated carbon-carbon-bonds. Further processing can then employ various types of catalytic
reactions. Cracking can be done over zeolites (eg. H-ZSM-5) to yield a mixture of light hydrocarbon
gases, iso-alkanes and aromatics. Light olefins or alcohols via dehydration could be upgraded to
larger olefins by oligomerisation, which is proposed as a route to yield diesel-length hydrocarbons
31
[252]. γ-valerolactone upgrading was investigated in two catalytic reactors in series first by aqueous
phase ring-opening and decarboxylation over SiO2 /Al2 O3 at 375℃ to butene, followed by oligomeri-
sation at 170-225℃ over for instance H-ZSM-5, while operating up to 36 bars. This gave up to over
75% total yields of C8+ hydrocarbons [257].
Ketonisation followed by reduction with hydrogen is proposed as a way to upgrade carboxylic
acids migrating to the hydrocarbon phase during the APR. The ketonisation merges two carboxylic
acids together, as seen in Figure 1.13. A temperature range of 175-350℃ at 5 bar over CeZrOx
catalysts has been investigated, and higher temperatures and low pressures of H2 O and CO2 were
found advantageous for pushing the reaction towards the ketone, as well as suppressing stepwise
ester ketonisation [258, 259]. Also acids in flash pyrolysis oils could be ketonised with reasonable
success [260].
O O O
+ + H 2O + CO2
R' OH R'' OH R' R''
Carboxylix acid Carboxylic Acid Ketone
Figure 1.13: Ketonisation of carboxylic acids and hydrogenation.
Two ketone molecules, directly from APR or from subsequent ketonisation may be reacted un-
der acidic conditions to form an enone as condensate (an unsaturated ketone) by splitting off wa-
ter. Saturation/reduction with hydrogen must be performed immediately of the resulting enone
during reaction to pull the equilibrium towards condensation products, as has been proposed over
Pd/Cex Zr1-x Oy at 325-350℃ [259, 261].
The mild temperatures, advantageous phase-separation behaviour, and internal production of
hydrogen are the main assets of this method. Sugars and polyols are however not as cheap nor
abundant as waste lignocellulosic biomass, so a greater feedstock tolerance would make this pro-
cess extremely viable, for instance such as the cascade-proces recently proposed by Serrano-Ruiz et
al. from cellulose via γ-valerolactone and ketonisation to 5-nonanone [262]. Larger water-soluble
biomass-derived molecules may pose problems to APR if reacted directly, as they can result heavy
coking [29].
1.5.3 Gasification and Fischer-Tropsch-synthesis

The Fischer-Tropsch-synthesis (FTS) was initially developed by German scientists Franz Fischer and
Hans Tropsch in the 1920s. By steam gasification of coal, steam reforming of natural gas or gasi-
fication of biomass, a synthesis gas (syn-gas) consisting of CO and H2 can be obtained after some
workup, which is used for the FTS yielding primarily straight-chained alkanes. At the time of Fis-
cher and Tropsch coal was the relevant raw material for producing syn-gas, providing a means of
turning inexpensive coal into liquid fuels - so-called coal-to-liquids (CTL). This has been employed
by Germany during World War II, and South African company Sasol has been using it for over 50
years - the plant in Sasolburg is the biggest point CO2 -emitter in the world [263]. The principle
behind this concept based on biomass is shown in Figure 1.14.
32
Today, the process is based on natural gas in other areas in the world - so-called gas-to-liquids
(GTL) - especially near large oil fields with a huge excess of cheap natural gas. However, biomass
may also be gasified with sufficient heat and converted to a syn-gas suitable for FTS (H2 /CO-ratio
of 2.1), and co-gasification-reforming schemes employing both biomass, natural gas and coal may be
advantageous to employ in the future [264]. Gasification may open up to production of other liquids
fuels than alkanes from FTS, including methanol, ethanol, DME, and hydrogen [35].
Burned for heat/

reformed
- CH4
Heat
CxHyOz CO2 , H2O , CH4 , CO , H2 x CO + (2x+1) H2 CxH2x+2 + H2O
Biomass cat.
Separation Gas cleaning Syn-gas Alkanes Water
- CO2, H2O Water-gas shift
Figure 1.14: Principle behind the biomass gasification and Fischer-Tropsch-synthesis pathway to diesel.
As illustrated in Figure 1.14 also steam and CO2 as well as small hydrocarbon gases result from
gasification. The WGS reaction (Figure 1.7 a) at page 25) can be used to get the desirable H2 /CO-ratio
of the syn-gas. Afterwards CO2 and H2 O is separated from the mixture.
A large part of the carbon from gasification ends up as CO2 . In terms of carbon capture and se-
questration (CCS) this procedure has the advantage of producing the CO2 concentrated and maybe at
elevated pressure. Thus having a net negative CO2 -emission from BTL-processes is possible, because
“burying” CO2 underground from biomass, being renewable, removes CO2 from the atmosphere, if
biomass is replanted to substitute the consumed amount [263].
Gasifiers are economy-of-scale equipment - the bigger, the cheaper and thus better [265]. Biomass
residues especially from agriculture on the other hand have a low energy density and then needs to
be transported over major distances, consuming fuel or energy, to centralised plants. Upgrading the
low carbon- and energy density of certain biomasses, especially straw, for instance via decentralised
small pyrolysis plants, may therefore be advantageous as described by Dinjus et al. [266].
1.5.3.1 Gasification
The gasification of biomass has been extensively studied, yet a range of technical challenges remain.
All types of biomass may in principle be gasificed and then converted, but different reactivity of
various raw materials will naturally determine the ease of gasification processing.
Usually, the biomass is heated to 700-1100℃ for instance in a simple fixed bed or more advanced
fluidised bed system [267], either in inert gas or with minor amounts of oxygen (resulting in some
combustion reactions to heat the reactions) and usually at low pressures, with steam and/or other
recycled process gases. The pyrolysis and gasification, however, are gradual processes in which the
biomass particles degas and volatilise step-wise [267]. Temperatures of at least 1000℃ may be nec-
essary for total carbon conversion to gas, depending on the biomass and conditions used, also due
to varying ash content, as alkali and earth-alkali ions affect the reactivity during gasification [268].
Ash may pose problems in terms of both disposal and operation, but can be reused as a heat carrier
in the gasification to ensure optimal regeneration of process heat and sufficient burnout of coke and
33
soot [269].
Catalytic pyrolysis at lower temperatures is a suggestion for reduction of gasification tempera-
ture, for instance of carbohydrates [192, 270]. Also catalytic gasification of biomass in super-critical
water has been proposed [271]. Catalytic techniques often lead to more complete gasification and
may even give the possibility of lowering the process temperature. Through careful optimisation it
was however found that at only 400℃ the non-catalytic gasification efficiency in air-steam mixtures
in a bubbling fluidised-bed could be as high as 60% [272].
1.5.3.2 Gas cleaning

Cooling of the gasification products must be performed to stop gas-phase reactions downstream.
However, some amounts of tar is produced while gasifying biomass [273], which can clog up pipes
downstream upon cooling and/or form coke on surfaces [274, 275]. Sulphur, phosphorous and nitro-
gen contained in the biomass form compounds that may poison the catalysts downstream - especially
the level of sulphurous gases must be brought down to ppm-level, as the active metals in FTS cata-
lysts iron or cobalt will otherwise form sulphides. Catalytic removal of tars, ammonia and sulphides
has been proposed for solving this problem [274, 276]. For instance nickel catalysts have been tested
in high-temperature cleaning of gas streams containing containing tars, NH3 , CH4 , and H2 S. Re-
duced nickel worked well for tar cracking at 700℃, however, 900℃ was necessary to crack NH3 and
CH4 , and H2 S inhibited reaction, especially at 700℃ [277].
1.5.3.3 Fischer-Tropsch-synthesis
An enormous body of literature is available within research on FTS catalysis [278], which is usually
performed by metal-oxide-supported metals at 150-330℃ and 50-200 bar. While supported ruthe-
nium has been shown to be the most active metal for FTS [197], iron and cobalt are normally used
industriallty, since the platinum group metals are usually too expensive to allow for commercial use
at this scale. The Co or Fe may then be doped with smaller amounts of other metals as promotors,
for instance Mn, Ni, Pt, Ru, K, or Ce. The formal reaction scheme is shown in Figure 1.14.
Apart from water (see Figure 1.14), straight-chain alkanes are achieved from FTS due to the
reaction mechanism of chain propagation, which is ideal for diesel fuels if they have the right carbon
number. Another important feature of the reaction is that for a certain carbon chain length i, the
weight fraction in the product mixture is given from the statistical probablility of chain growth, α,
by:
Wi = i · (1 − α)2 · α i−1
This equation is called the Anderson-Schultz-Flory-distribution (ASF) [279, 280]. The product
distribution as a function of the chain growth probability, α, is shown in Figure 1.15 for certain
groups of alkanes. Often the selectivity is expressed for instance as the yield of hydrocarbons with
a chain-length of at least 5 carbon atoms, C5+ . This is the group of hydrocarbons most relevant
in terms of refinery processing and fuel production. FTS-catalysts always form methane to some
degree, and a low selectivity to methane and as well the lighter hydrocarbon gases is usually desired
34
(mathematically this means a higher value of α in the ASF) [279, 280]. It follows mathematically
from the ASF that methane is always the product produced in the highest amount on a molar basis;
however the weight fraction is small due to the large molar weight of the heavier paraffins formed at
high values of α [279].
The primary catalysts for FTS Co and Fe usually have α ≈ 0.7−0.9, which can be optimised by ad-
dition of promotors to the primary metal. The largest fraction of diesel pool of C12 -C18 -hydrocarbons
is achieved with α ≈ 0.88, as is evident from Figure 1.15. Iron has been shown to be the more ac-
tive metal under the most severe conditions, i.e. at higher temperature, pressure and space velocity,
while contrarily cobalt is the more active catalyst in less forcing environments [279, 281].
1,0
C1 C2-C4
C5-C11 C12-C18
0,8
C19-C25 C26-C35
C35-C50 >C50
mass fraction
0,6
0,4
0,2
0,0
0,0 0,2 0,4 0,6 0,8 1,0
Figure 1.15: The Anderson-Schultz-Flory-distribution: Statistical mass fractions of product groups as a function
of the chain growth probability α.
In addition to straight-chain hydrocarbons, impurities of aldehydes, alcohols, and fatty acids

are also formed. Such oxygenated by-products should later be reduced by HDO if used for diesel
synthesis. Light isomerisation may be necessary to achieve sufficiently good cold properties [282].
The FTS catalysts are sensitive to poisons - especially sulphur, which form sulphides with the catalyst
metal if present over ppm-level.
The particle-sizes and structure of the catalyst metal, as well as the microscopic environment in
which they are used, affects the overall reaction behaviour. By supporting nanoparticles of Co, Fe
and their alloys inside carbon nanotubes showed that activity was enhanced, and the C5+ -selectivity
could be enhanced as well, which was attributed to a longer contact time with the catalysts particles
in the confined environment of the nanotubes [283, 284]. Co had the the highest C5+ -selectivity, and
Co particles up to the size of the carbon nanotube diameter of 12 nm could be selectively located
inside the nanotubes. Reduced sintering (irreversible deactivation) inside the nanotubes due to the
confinement and stronger support interaction enhanced the FTS catalyst lifetime [283]. Larger Co
particles had higher C5+ -selectivity but lower overall activity due to lower surface area of larger
particles [283, 284]. The optimal particle size is around 6-8 nm for Co as the C5+ -selectivity does not
increase above this particle size [285]. Oxidation of the Co crystallites takes place during reaction
35
and impedes activity, however, the oxidised Co may be reduced to regain activity [283]. The working
Co catalyst have been shown to be metallic [285]. The Co-Fe alloys had high selectivities towards
C2+ -alcohols as up to 26% of the carbon could be converted to alcohol [284]
Confinement of active sites inside micro- or mesoporous materials, such as zeolites, can dramati-
cally alter the behaviour of the FTS catalysts and thereby change the product distribution away from
ASF and towards a more selective range of products, for instance due to shape and size selectivity
[286]. Support acidity and microporosity may be used to directly isomerise products as well [287].
The pressure has a profound influence on the value of α as higher pressures usually resulting
higher chain-growth probabilities [279]. Promotion by alloying with transition metals and doping
with alkalis have also been investigted. For example, small amounts of K in a Co catalyst increased
the reduction temperature and decreased the activity, but enhanced C5+ -selectivity and the selecti-
vity towards olefins instead of paraffins. Potassium may, however, also enhance the activity, which is
dependent on temperature [279]. Ru decreased the reduction temperature on Co and increased ac-
tivity and C5+ -selectivity [288]. Mo addition to Fe reduced the deactivation by sintering of the active
iron particles and enhanced the formation of iron carbides [289] - part of the mechanism for FTS over
Fe involves formation of carbides [290, 291]. Ce and Mn promotors enhance the formation of active
carbide species for the chain growth over the Fe catalyst, while Ce is also a structural promotor for
Fe [290]. Iron catalysts, depending on promotion, can also catalyse the water-gas-shift equilibrium
during FTS [290].
Several researchers have tried to optimise the production of alcohols, aldehyde, or ketones [284,
292, 293]. The production of longer-chain alcohols or aldehydes may be another prospective deve-
lopment within FTS, since upgrading and drop-in blend for either petrol, jet fuel or diesel can now
be tailored, or even be steered towards the production of bulk chemicals via FTS. As was shown by
Mentzel and co-authors, higher alcohols may even be a preferable feedstock for producing petrol-
range fuel from the methanol-to-hydrocarbons reaction over acidic zeolites [294].
36
1.6 Diesel fuel properties
Diesel may be defined as anything that can be used as fuel in an unmodified diesel engine and has
fuel properties as a middle distillate (or middle-cut) from petroleum oil refining. The most important
properties for diesel oils are cetane number, heating value and filter blocking point, and these are
used to characterise and compare fuels - especially alternative fuels relative to the norm for diesel
fuel. Diesel oil constitutes the middle distillate (or middle-cut) from raw petroleum distillation, and
part of it is also produced via hydrocracking of vacuum gas oil. The number of carbon atoms in the
fuel ranges from 12 to 18 and primarily consists of alkanes.
The cetane number indicates the ease of ignition of the diesel oil in the engine. Since the diesel
is injected in a compression-ignition engine when the piston is approaching the cylinder top dead
center (near full compression) it must ignite immediately, contrary to petrol engines [295]. European
diesel must have a cetane number over 51 while the North American norm specifies a cetane number
of 47 [42]. Straight alkanes ignite easier than brached ones, and the higher the cetane number, the
easier the ignition takes place [296].
The heating value and density describes the energy density in the fuel and therefore also how
much work can be performed by the engine per litre of fuel and therefore also mileage. Usually this
heating value is 43 MJ/kg or 36 MJ/L for petrodiesel [43], however, it is not specified in European or
American diesel fuel standards [42]. The cold filter plugging point (CFPP) specifies the temperature
at which a designated test filter is plugged with diesel wax crystals and is the most common way to
report the cold properties of diesel. Additives are often used to lower the CFPP, which is necessary
during winter time in for instance North America, Russia, Central Europe or the Nordic countries.
On the contrary, longer paraffins can be added to summer diesel to enhance the cetane number.
Pour Point (PP) and Cloud Point (CP) are other measures for describing the cold properties. The
CP describes the temperature where, upon cooling, visible diesel crystals start to form as a cloud
wax, and this is usually a bit higher than CFPP. PP is measured by first cooling a diesel sample to
solid and then heating it up. The temperature within regular intervals at which it starts flowing is
the PP.
A number of other important features of diesel fuels are viscosity of the fuel, the lubricity as a
measure of the wear that is imposed on the engine, the storage and oxidation stability (especially in
the case of FAME), the distillation characteristics and the flash point, see Table 1.1. From an environ-
mental point of view especially the aromatics and sulphur contents are important, since sulphurous
oxides and aromatic particles emitted from the engine are harmful to biota in the immediate sur-
roundings.
These measures have relevance for the biofuels to be used as diesels as well, as any diesel oil
should be within the specification limits; for a more profound list of relevant biofuel properties
(specifically FAME) see for instance Knothe [42].
37
1.6. Diesel fuel properties
Table 1.1: Property description for diesel oils specifications [42, 43, 45, 198].
Property Unit Description

Density kg/L Density of the fuel.
LHV MJ/kg Lower Heating Value; energy content in the fuel.
CN % Cetane Number; measure of the ease of ignition.
Measured in a test engine as the percentage of cetane
(n−hexadecane, n−C16 H34 , CN = 100) in a mixture
with α-methylnaphthalene (1-methylnaphthalene,
C11 H10 , CN = 0) to be used to get the same ignition
properties as the test fuel [295]. In the latest decades,
however, iso-cetane (iso-hexadecane, 2,2,4,4,6,8,8-
heptamethylnonane, i−C16 H34 , CN = 15) has been
used instead of α-methylnaphthalene [296].
CFPP ℃ Cold Filter Plugging Point; The lowest temperature,
measured in steps on 1℃, where diesel can still flow
through a test filter.
PP ℃ Pour Point; The lowest temperature, measured in
steps of 3℃, at which diesel in a test jar still flows.
CP ℃ Cloud Point; The temperature at which the fuel gets
cloudy by diesel crystals.
Kinematic Viscosity mm2 /s Describes the ease of pumping and injecting the fuel.
Lubricity HFRR µm High Frequency Rotating Rig; The diameter of the
scar made by rubbing a steel ball againts a steel disc
submerged in the test fuel. The usual description of
the wear that the fuel causes to the engine.
Storage stability — Arbritrary evaluation of the difficulties of storing the
fuel over a longer time - relevant with storage tanks
containing impurities, especially water.
Aromatics wt% Content of aromatics in the fuel (can be split in
mono- and polyaromates).
Sulphur wt% Sulphur content in the fuel.
Distillation range ℃ -℃ Specifies distillation characteristics; the temperature
where 10 wt% respectively 90 wt% of the diesel is
distilled of. Relevant for vaporisation and ignition in
the engine.
Flash Point ℃ Safety measure; the temperature of the diesel fuel at
which a spark can ignite the vapors over it.
38
1.7 Comparison of diesel fuels
Much depend upon the processing and reaction conditions of each process, the feedstock used and
mechanistic details of the catalytic processes, especially from upgraded bio-oils or hydrocarbons
from aqueous-phase reformed biomass. However both the production of FAME, renewable diesel
from HDO of fats and BTL/GTL have reached such maturity that an actual comparison of the pro-
duced fuels is possible. Relevant properties have been compared in the literature with respect to
diesel fuels derived from biomass [43, 45, 198].
Table 1.2: Comparison of diesel fuels from biomass with EN 590 Summer petrodiesel. GTL is “gas-to-liquids”
from gasified natural gas and Fischer-Tropsch-synthesis; the FAME is from rapeseed oil. Table adapted from
S. Mikkonen of Neste Oil Oy [45].
unit EN 590 diesel Fat HDO Diesel GTL FAME (rape)

Density at 15℃ g/L 835 775 - 785 770 - 785 885
Viscosity at 40℃ mm2 /s 3.5 2.9 - 3.5 3.2 - 4.5 4.5
Cetane number 53 80 - 99 73 - 81 51
Distillation range ℃ 180 - 360 190 - 320 190 - 330 350 - 370
Cloud point ℃ -5 -5 to -251 0 to -25 -5
LHV MJ/kg 42.7 44 43 < 37.5
LHV MJ/L 35.7 34.4 34 < 33.2
Aromates, total wt% < 30 0 0 0
Polyaromates wt% <4 0 0 0
Oxygen content wt% 0 0 0 11
Sulphur content mg/kg < 10 < 10 < 10 < 10
Lubricity HFRR +60℃ µm < 4602 < 4602 < 4602 < 460
Storage stability Good Good Good Challenging
1 This requires a light isomerisation of the n-alkanes produced by the hydrotreatment.
2 By using normal additives. FAME blended with petrodiesel lubricate the engine and may not
require further additivation
Both the fat HDO and the Fischer-Tropsch-synthesised diesel oils have the advantage of produ-
cing long-chain alkanes as primary product with a cetane number nearing 100, depending on chain
length. Thus they may be used as cetane boosters/additives to diesels fuels not falling within the
specification on cetane number.
The cold properties of n-alkanes by HDO of oils and fats are problematic if they are not iso-
merised afterwards. Isomerisation betters cold properties but lowers the cetane number due to
branching. But also FAME can have troublesome cold properties, especially more saturated types
like those from animal waste fats. Saturated FAME, on the other hand, have higher cetane numbers
than unsaturated ones [15, 198]. The production of renewable diesel is usually not inhibited by FFA,
which deoxygenates more readily than ester compounds [16] - contrary to FAME where FFA usually
leads to addition of an esterification step in the production chain.
Removing sulphur from petrodiesel via hydrodesulphurisation has the negative effect that the
resulting fuel does not lubricate the engine sufficiently. Additivation of the fuel is therefore needed
(these additives may often derive from fats and oils). FAME fuels have excellent lubrication proper-
ties, and in mixtures down to as low as 2 wt% FAME no further lubrication additives in the fuel are
needed to lubricate the engine [15, 198]. On the other hand FAME is more prone to degradation
by microorganisms if the storage system contains water - this may be an issue with old, remote or
39
1.7. Comparison of diesel fuels
sparcely serviced tanks. The alkanes from HDO of fats and oil or FTS have, due to their chemical
resemblence with diesel fuel, neither of these properties.
Production of FAME from fats uses the mildest reaction conditions, the cheapest plant design and
may be decentralised, for instance in connection with abbatoir waste treatment or food production.
More forcing conditions are used for hydrodeoxygenation of fats and oils, and it is more suited for
traditional refinery infrastructure with possibilities of heat recovery and easy access to hydrogen
and isomerisation units. Gasification and Fischer-Tropsch-synthesis may in principle be applied for
all types of biomass, but the method requires the most forcing reaction conditions and the most
extensive plant design. Gasifiers are economy-of-scale plants and usually pose the biggest investment
in a BTL/GTL plant, so this procedure is only viable for very big plants and thus huge amounts of
available biomass [266].
At present and in the near future the diesel biofuels are being mixed in with petrodiesel in low
blends (no more than 5 or 10 wt% on a larger scale) and here most of the blend properties are not
very different from those of pure petrodiesel. Various diesel oils from biomass may thus be expected
to be blended into the diesel mix. In the EU, 5.75% of the energy in the fuel should come from
biomass sources by the end of 2010 [5], but for many present-day fuel systems only 5 wt% FAME in
the fuel is warrantied for the engine. Blending with alkanes from HDO of oils and fats or FTS thus
makes it possible to reach this without compromising engine warranty [45].
40
1.8 Summary of the literature and outlook
Currently both transesterification and hydrodeoxygenation of fats and oils have achieved industrial
maturity as methods for production of diesel from biomass. Feedstock price of the oil or fat remain
the largest cost in the production. In Europe, rapeseed oil is often used as a feedstock for FAME
while soy beans is more normal in North America and palm oil in the tropical regions of the world,
for instance in South-East Asia or Latin America.
The transesterification of fats and oils, and esterification of FFA, is the most widespread methods.
Benign production at ambient pressure at the boiling-point of co-reactant methanol yields FAME. In-
dustrially the reaction is based on homogeneous mineral acids and bases as catalysts which are used
batch-wise and discarded after use, and heterogeneous analogues have been vastly investigated. This
includes solid oxides of basic or acidic character, meso- or microporous materials, organic function-
alised resins or other solids. The application of reusable catalysts would cut costs and produce less
waste, and allow for a simpler and more economical process design.
The deoxygenation of fats and oils can be performed over three types of catalysts, namely sup-
ported sulphided transistion metals, supported transition metals in metallic form, and porous acidic
or basic materials for cracking. The sulphided metals requires sulphur and hydrogen in the feed to
remain active, so co-treating with hydrodesulphurisation feedstock in refineries is an option. The
metallic catalysts may formally run without hydrogen, but require it to avoid deactivating side-
reactions. Hydrodeoxygenation of fats and oils is still in need of unsulphided catalysts to avoid
adding sulphur to the feedstock, and of metal catalysts with a longer lifetime than those previously
reported. Deactivation phenomena needs to be studied and understood better to avoid them. Crack-
ing of fats and oils generally still gives too poor yields of primary fuels like gasoline, jet fuel, and
diesel due to high production of lighter gases and coke.
Fats and oils are sparse resources: The amount of waste oils from eg. abattoirs or restaurants
is limited compared to the global diesel consumption, and the land use associated with the growth
of vegetable oil resources cannot cope with demand for fuels either - only a minor percentage of
the global diesel demand can be covered with the current two technologies based on fats and oils
with the available resources in the near- and medium-term future [198]. The advent algae-based
triglycerides on a large scale could help to solve this feedstock issue and make it possible to produce
triglycerides for biofuels on an industrial scale in an environmentally acceptable manner; however
this is yet to be industrially demonstrated [297, 298].
The growing of crops for biofuel production on productive soil have several drawbacks from an
environmental point of view: Switching to growing agricultural crops for biofuels means that pro-
ductivity of the remaining agricultural production must either be increased or forest or wilderness
must be cleared to make room for new agricultural production with severe ecological implications
for the invaded ecosystems [298]. Emissions of nitrous oxide (N2 O, a greenhouse gas several orders
of magnitude stronger than CO2 [2]) from fertilisation is one effect of intensive agriculture, and fer-
tiliser production also emits CO2 . Farming in itself also requires fueling for tractors etc [2, 299, 300].
Thus, only lignocellulosic resources like straw, wood and waste appear to be sustainable feed-
41
1.8. Summary of the literature and outlook
stocks that are available in sufficient amount to cover a large fraction of the global fuel demand. Until
now, three catalytic strategies for conversion to fuels have been formulated for these feedstocks:
Flash pyrolysis of biomass or near-critical hydrothermal ugrading of biomass can yield a type
of bio-crude commonly referred to as bio-oil, containing up to 40 wt% oxygen. This oxygen may
removed by deoxygenation, which generally can use the same three types of catalysts that are used
when upgrading fats and oils, namely supported metals, sulphided metals or porous materials. The
hydrogen consumption and deactivation phenomena may, however, be even more severe and com-
plex than for HDO of fats and oils. The cracking-type catalysts like zeolites appear to be most prac-
tical for bio-oils, requiring no hydrogen and allowing for regeneration of zeolites or metal oxides by
oxidation by burning away coke formed as by-products.
Carbohydrates and hydrolysed cellulose can be converted by condensations, reductions and re-
forming reactions in the aqueous-phase as advocated first by the group of Dumesic [29]. The re-
forming reactions can take place at quite moderate temperatures and usually over supported Pt or
Pt-Re catalysts, and up to 95% energy efficiency to be contained in the fuel can be achieved for the
integrated process. Furthermore hydrogen is supplied internally by the reforming. Bio-oils may as
well be processed once condensation reactions and ketonisations possible from the multifunctional
bio-oil mixtures can be controlled.
The research into bio-oils production and upgrading and the aqueous phase reforming may there-
fore play an increasingly prominent role in the future, despite their more challenging nature. The
potential ability to use more parts of for instance wood or straw residues would make the aqueous-
phase upgrading even more promising, especially due to its high energy efficiency and carbon con-
tainment in the product.
A final method, in principle possible with all types of biomass, is the complete gasification of
the biomass into a producer gas, which can be cleaned into syn-gas (CO and H2 ) and then liquified
as alkanes via FTS. Gasification requires high temperatures and expensive process equipment. The
gasification products contains tars and impurities of sulphur and nitrogen from the biomass, which
must be removed prior to FTS. This gas cleaning is complex to integrate. The consecutive FTS is
usually catalysed by Co and Fe metals, however, the activities and selectivities can be vastly modified
by varying, promoting, and confining the active sites.
Major logistic challenges need to be resolved as well: Biomass is a local resource [301], while
gasifiers are usually huge equipment placed centrally. Dinjus et al. described the logistics around
a gasification fuel economy, and suggested that residues like straw or wood could be upgraded by
flash pyrolysis first in small decentralised plants [266]. Then the bio-oil and char fraction, now with a
much higher carbon- and energy density can be transported to and upgraded at central gasification-
FTS plants.
Primary and easy-to-upgrade feedstocks like plant oils are not and cannot be avaible in quantities
sufficient to replace the entire diesel oil demand, so waste products from food production, agricul-
ture and forestry hold greater perspective for use in the future. It is necessary that the processes
for future biofuel production have sufficient tolerance to use diverse raw materials, primarily in the
form of waste, as feedstocks.
42
Chapter 2
Catalytic Production of Fatty Acid Methyl Ester 2

The esterification and transesterification of fats and oils with methanol to yield fatty acid methyl
esters (FAME), according to Figure 1.2 and Figure 1.3 in section 1.3, are generally produced in in-
dustrial scale in a two-stage batch proces employing a strong mineral acid and a ditto base as homo-
geneous catalysts, respectively. The homogeneous catalysts are responsible for high yields with low
amount of catalyst used per batch, but they are not recovered during the production of the FAME
and end up as neutral salts that have to be disposed of.
Heterogeneous catalysts would, on the other hand, not be lost during inter-stage neutralisation,
workup and separation of the products. Furthermore, process simplification in terms of continuous
operation for instance in packed-bed reactors yielding a more even product quality would be pos-
sible, and the expenses for obtaining catalysts could be lowered. Therefore the substitution of the
homogeneous catalysts in FAME production represent an important goal for catalytic science and
techonology.
2.1 Heterogeneous catalysts for production of FAME: Introduction
A wide range of solid inorganic acids have been suggested as heterogeneous catalysts for respec-
tively esterification and, to a certain degree, transesterification. Tungstated or sulphated oxides like
WO3 /ZrO2 , SO42 – /ZrO2 , SO42 – /ZrO2 −TiO2 , SO42 – /SnO2 , SO42 – /SnO2 −SiO2 , SO42 – /SnO2 −AlO2 , he-
3– 4– 3–
teropolyacids like the protonated or partly protonated forms of PMo12 O40 , SiMo12 O40 , PW12 O40 ,
4–
SiW12 O40 , niobic or tantalic acid have been reported and recently reviewed by Sharma, Singh and
Korstad as potential catalysts for esterification [302].
Basic catalysts suggested for the transesterification include basic oxides like MgO, CaO, SrO,
BaO, ZnO, La2 O3 , either unsupported or supported on other oxides, mixed oxides like MgO · CaO,
MgO · Al2 O3 , MgO · La2 O3 , CaO · La2 O3 , Ca2 Fe2 O5 , CaMnO3 , CaCeO3 , CaTiO3 , and CaZrO3 , or sup-
ported bases like KF, Li2 O, K2 O or Na2 O [62, 303], or Mg-Al-hydrotalcites for instance substituted
or doped with other ions [73–75].
In some instances the inorganic acids or bases may, however, dissolve in the polar alcohol phase
(which due to esterification can also contain water, see Figure 1.3) over time resulting in leaching of
the catalyst. The flexibility of shaping the inorganic materials to the desired technical application
is not great. Often the porosities and solubilities are non-tunable. Many of the inorganic catalysts
are powder, which are unwanted for instance in for instance packed-bed reactors due to very high
pressure drops observed. Extrudating the powders, for instance with other ceramic binders and
carriers, may help shape powders to more suitable sizes and shapes for flow reactors.
43
2.1. Heterogeneous catalysts for production of FAME: Introduction
Organic acids and bases, however, offer different and often better possibilities for tailoring the
catalysts for the liquid-liquid-phase reactions. Supporting the catalysts for instance polymers, resins,
or porous inorganic materials leave options for synthetic chemistry. This allows pairing of strong
acidity or basicity with a suitable porous support material and a desirable macroscopic appearance
and making organic-inorganic hydrid catalysts.
Ionic liquids (ILs) have been suggested as an alternative reaction medium for the ester reactions
and it has been shown that a neutral IL can be mixed with a given mineral acid or base, for instance
H2 SO4 , K2 CO3 , or NaOMe, as catalyst to obtain an active and monophasic reaction medium [137,
304]. However, it may be much more prosperous to functionalise the ILs themselves as strong acids or
bases. Lewis-acidic ILs with metal chlorides as anions have proven to be active in transesterification
between 80 and 180℃ in a number of studies [139, 305, 306]. Sulphonic acid-functionalised ILs
have been used as strong acids for FFA esterification [307, 308]. Recently it has also appeared that at
mild temperatures around the boiling point of methanol, strong sulphonic acid-functionalised ILs
did even catalyse the transesterification at reasonable rates [141, 142, 309].
Other types of sulphonic acids are, however, available as well for formation of FAME. Okayasu
and co-workers polymerised vinyl-sulphonic acid and nested it to mesoporous silica or polystyrene
and achieved high loading of sulphonic acid and high activities for esterifications of smaller fatty
acids with ethanol at 50℃ [125, 310]. Grossi and co-authors prepared a sulphonic acid super-
absorbant based on polystyrene (PS) and demontrated higher activity for oleic acid esterification
than commercial acidic resins and even p-toluenesulphonic acid [123]. A hydrophilic polymer like
sulphonated polyvinylalcohol was reported to be more active for esterification than a sulphonated
PS at 60℃ by Caetano and co-workers [122]. Soldi et al. reported high yields even for transesterifi-
cation over a sulphonated PS at 60℃ [124]. On SBA-15 linked with arene-sulphonic acids, Melero et
al. achieved 90% yield of FAME from crude palm oil after 2 h at 160℃ [126].
SO3H
CH3SO3H H3C SO3H SO3H N N
n X-
methanesulphonic acid p-toluenesulphonic acid poly(styrenesulphonic acid) [BMIM-SO3H+][X-] (ionic liquid) _
Figure 2.1: A few different strong sulphonic acids. X – denotes a counter-anion.
Commercial sulphonic acid resins and ion-exchange resins are often available, however, no stu-
dies so far have reported yields comparable to those achieved by laboratory catalysts for transester-
ification with acidic resins [311]. At around 60℃, however, esterification of FFA around 90% yield
within an hour have been achieved on various sulphonic acid-resins [118–121]. Shibasaki-Kitakawa
et al. reported the continuous esterification of FFA over a sulphonic acid-resin followed by transes-
terification of glycerides over a basic resin in two packed-bed reactors in series [117].
Another starting material for solid sulphonic acids has been reported as catalyst for the esterifi-
cation in FAME production in the recent years. By carbonatisation (slow pyrolysis) of carbohydrates
and subsequent sulphonation by sulphuric acid, a layered graphene-like material can be obtained
with strong sulphonic acid functionalities connected to the graphene layers (sulphonated pyrolysed
44
Chapter 2. Catalytic Production of Fatty Acid Methyl Ester
sugars, SPS) [128, 132, 312, 313]. The suggested structure of SPS is shown in Figure 2.2.
Both esterification of oleic acid with ethanol at 80℃ and transesterification of triolein with
ethanol at 100℃ are catalysed with activities comparable to those of sulphuric acid, and up to
98% yield from transesterification of triolein with methanol after 5 h at 130℃ has been achieved
[312]. The high activities reported for the SPS are likely connected to the swelling of SPS on contact
with hydrophilic solvents like methanol because the dry catalyst has a surface area of just 1-2 m2 /g
[128, 130, 312, 313]. Ideally, most of the the oxygen should be removed to avoid interference and
side-reactions with the hot sulphuric acid during the sulphonation. However, some oxygen remain-
ing in the carbon material and moderate pyrolysis temperature of 400-450℃ has been found optimal
in terms of incorporation of a high amount of sulphonic acid functionalities and esterification acti-
vity [313].
OH
HO3S
SO3H
O
HOOC O
O
HO SO3H
HO
HO3S HO3S
O COOH
SO3H COOH
HO3S SO3H
HO3S OH
Figure 2.2: Structure of the sulphonated pyrolysis carbohydrates as suggested in literature [130, 132, 312, 313].
Also wood pyrolysis char has been sulphonated for esterification catalysis, and the resulting ma-
terial can have higher surface area by yielding microchannels in the resulting catalyst [131] - by
impregnating wood with ZnCl2 prior to pyrolysis, a surface area of the final sulphonated catalyst
up to 1560 m2 /g was achieved, which was several times more active in esterification reactions at 70
and 100℃ compared to the sulphonated pyrolysed wood not impregnated with ZnCl2 [314]. Lien
and co-workers observed over 90% transesterification of soybean oil after 2 h at 150℃ over 3 wt%
sulphonated pyrolysed glucose, but leaching of sulphonic acids was reported at this temperature
[133].
Peng et al. recently described the use of SBA-15 as rod-structure-directing agent for the SPS
synthesis, and the removal of SBA-15 by HF after sulphonation was highly advantageous in the es-
terification of oleic acid with ethanol at 80℃ [135]. The surface area of the resulting carbon material
was over 800 m2 /g, in stark contrast to the 2 m2 /g achieved for the “naked” carbohydrate carbonati-
sation and sulphonation [135, 313]. Mo and co-workers impregnated a polymer with carbohydrate
template followed by pyrolysis and sulphonation and achieved 20% esterification of palmitic acid
with metrhanol at 60℃ after 1 h, however gradual deactivation was also observed. Interestingly, the
polymer matrix swelled upon immersion in the methanol reactant [54].
In general, the SPS-approach has the advantage that water-soluble carbohydrates can be impreg-
45
2.1. Heterogeneous catalysts for production of FAME: Introduction
nated in other supports prior to pyrolysis and sulphonation. However, most SPS catalysts developed
in literature are powders, which are not desirable when to use when scaling up to technical-scale
synthesis, although extrudation may be a possibility.
Based on the literature reports, the goal of the present work have therefore been to develop
macrostructured, heterogeneous catalysts for FAME production. This was pursued by using a model
mixture of methanol, lauric acid and trioctanoate to investigate and optimise the synthesis condi-
tions for the sulphonated pyrolysed carbohydrates by esterification experiments and characterisa-
tion of acidic and porous properties. The active SPS-phase has been supported by nesting it on
various support materials, either by impregnation of sucrose on either open-cell ceramics or meso-
porous ceramic pellets prior to pyrolysis and sulphonation, or by suspension of the active SPS in a
polyurethane (PUR) matrix. Two of the resulting SPS impregnated on ceramics have been tested in
the continuous esterification of lauric acid with methanol.
Transesterification catalysis have been pursued by investigating and relating a number of organic
amidine and guanidine bases as well as amine-functionalised ionic liquids and carbonates. Also, a
combination of transesterification and esterification in one pot has been attempted over sulphonic
acid-functionalised ionic liquids.
Due to ease of comparison and uniformity of the performed reactions, the same model feed has
been employed for studying catalytic behaviour during esterification and transesterification - a mix-
ture of methanol, trioctanoate and lauric acid. Unlike FAME synthesis from methanol and oils and
fats the three reactants formed a single phase, making sampling during the course of the reaction
easier. The expected esterification and transesterification reactions are given in Figure 2.3. Thus, es-
terification of lauric acid leads to formation methyl laurate, while transesterification of trioctanoate
leads to formation of methyl octanoate. Therefore the esterification and transesterification activity
or potential hydrolysis reactions can be directly related to either reactant.
O O
a)
O C7H15 H3CO C7H15
O OH O
O O C7H15 + 3 CH3OH OH + H3CO O C7H15

H+/OH-
OH
O C7H15 H3CO C7H15
trioctanoate methanol glycerol methyl octanoate
b) O O
HO C11H23 + CH3OH H2O + H3CO C11H23

H+
lauric acid methanol water methyl laurate
Figure 2.3: Reactions of the model feed of methanol, trioctanoate and lauric acid: Transesterification (a)) and
esterification (b)).
Theoretically, in the presence of water, the acidic and basic catalysts may hydrolyse the ester
bonds to form FFA - in the case of the basic catalyst, the hydrolysis leads to soap formation (the salt
of the base and FFA). However, it is possible that the acidic transesterification may take place via
hydrolysis of the fat and consecutive esterification of the FFA as is shown in Figure 2.4.
46
a) O O
O C7H15 HO C7H15
O OH O
O O C7H15 + 3 H2O OH + HO O C7H15

H+/OH-
OH
O C7H15 HO C7H15
trioctanoate water glycerol octanoic acid
b) O O
HO C7H15 + CH3OH H2O + H3CO C7H15

H+
octanoic acid methanol water methyl octanoate
Figure 2.4: Potential hydrolysis (a)) and esterification (b)) of trioctanoate in the presence of water and methanol.
47
2.2. Heterogeneous catalysts for production of FAME: Experimental
2.2 Heterogeneous catalysts for production of FAME: Experimental
2.2.1 Activity measurement of catalysts

The reaction mixture was composed of the reactants methanol (>99.9%, Bie & Berntsen), lauric
acid (dodecanoic acid, >99%, Fluka), and trioctanoate (tricaprylin, >99%, Sigma), while dodecane
(>99%, Sigma-Aldrich) was used as an internal standard. Pyridine (>99.8%, Fluka) and N-methyl-
N-(trimethylsilyl)-trifluoroacetamide (MSTFA, >98.5%, Fluka) were used for sample preparation
for GC analysis. Methyl laurate (methyl dodecanoate, >99.5%, Aldrich), methyl caprylate (met-
hyl octanoate, >99% (GC), Fluka), octanoic acid (caprylic acid, >99%, Sigma), and glycerol (>99.0%,
Sigma-Aldrich) were used as reference standards for the GC analysis along with the reactants used.
The investigated catalysts were weighed and added each to a 25 ml two-necked round-bottomed
flask, placed in a heated oil bath with a magnatic stirring, and glass plugs were mounted in the
flask necks. The oil bath temperature was adjusted to a temperature of ca. 65℃, effectively giving
a temperature in the glass flasks of 60℃, which was measured by a thermometer. Typically 0.2 g of
catalyst (ionic liquid: 0.5 g) was added to each flask.
For testing the acidic catalysts, a monophasic reaction solution consisting of 0.724 g lauric acid,
1.700 g trioctanoate and 1.167 ml methanol (0.926 g) was mixed prior to reaction, giving an effective
molar ratio between methanol, trioctanoate and lauric acid of 8:1:1. Furthermore, 0.0703 g of n-
dodecane was added to the mixture as an internal standard. For the experimentation with basic
catalysts, a monophasic reaction mixture consisting of 2.023 g tripalmitin and 1.566 ml methanol
(1.240 g), yielding a molar ratio of methanol and trioctanoate of 9:1, while 0.0733 g of n-dodecane
was added as an internal standard.
Typically, 6 acidic or basic catalysts were tested in parallel using the same preparation of reaction
mixture. When the catalysts were heated, 4.00 ml (3.35 g) of the reactant solution was added to the
heated flasks containing the catalysts with stirring running at ca. 500 rev/min. All reactions were
performed as time series experiments by sampling over time between 0 and 96 h: 10.0 µL of each
flask were taken out as sample with a precision pipette after 0.25, 0.5, 1, 3, 7, 24, 48, and 96 h and
added to a GC glass vial, while a sample from the untreated reaction mixture was also taken out as
the 0 h sample. Each of the samples taken out were treated with 1.00 ml of pyridine and 100 µL of
MSTFA and silylated for at least 30 min at 60℃ and analysed on GC.
After the reaction was completed, the reaction solution from each flask with a solid catalyst was
filtered to remove the catalyst, and the filtrate was diluted with water and a concentrated solution of
barium nitrate (Ba(NO3 )2 ) was added to test the mixture for presence of sulphate species by precipi-
tation of any SO42 – as BaSO4 .
2.2.2 Liquid product GC analysis

The liquid samples obtained from reaction solutions with SPS as catalysts were quantified by an
Agilent Technologies 6890N gas chromatograph (GC) equipped with flame ionization detector (FID),
split/splitless injection system and a HP-5 capillary column (J&W Scientific, 30 m × 0.32 mm × 0.25
48
µm (5 mol% phenyl)-methylpolysiloxane). The liquid samples taken from reaction solutions with
ionic liquids, the supported and impregnated SPS in polyurethane or on ceramics, and the basic
catalysts tested were quantified by an Agilent Technologies 7890A GC equipped with FID, cool-on-
column injector and a high-temperature DB-5HT capillary column (J&W Scientific, 15 m × 0.32 mm
× 0.25 µm (5 mol% phenyl)-methylpolysiloxane) joined with a precolumn (or guard-column, J&W
Scientific, 1 m × 0.53 mm, of fused deactivated silica).
For the GC analysis He (>99.999%, AGA) was used as column, septum purge and make-up gas,
H2 (>99.999% or technical quality, AGA) was used as FID ionisation fuel, and dehumidified pres-
surised atmospheric air was used as oxidant for the FID. All the reference standards applied were
run on both of the GCs.
2.2.3 Sulphonic acid-functionalised ionic liquids

Four sulphonic acid-functionalised ionic liquids were tested as acidic catalysts for esterification and
transesterificatoin with the reaction model mixture described in section 2.2.1. Butyl(methylimidazo-
lium)sulphonate was used as cation with either mesylate ([BMIM−SO3 H+ ][MeSO3– ]), bis(triflamide)
([BMIM−SO3 H+ ][NTf2– ]), or hydrogen sulphate ([BMIM−SO3 H+ ][HSO4– ]) as anion, and butyl(pyridi-
nium)sulphonate hydrogensulphate ([BP−SO3 H+ ][HSO4– ]) was investigated as well. All ionic liquids
were in-house synthesised chemicals (>95% purity by NMR) by Olivier Nguyen van Buu.
2.2.4 Sulphonic acid-functionalised pyrolysed carbohydrates

The sulphonated were prepared on the basis of the method first proposed by Toda and co-authors
[128, 313]:
1. Initially, either sucrose (>99.5%, Sigma), starch (analytical grade, Riedel-de-Haën), glucose
(>99.5%, Sigma), or cellulose (powder, ca. 20µm, Aldrich) were weighed and about 2-15 g
carbohydrate per batch were pyrolysed for 1-15 h at 300-450℃ and a heating rate of 100℃/h
in Ar (>99.99%, AGA) flow. A glass tube or a tubular steel reactor was placed and isolated with
vermiculite in a programmed tubular furnace and connected with argon flow on the inlet side.
The carbohydrates were loaded into in a glass boat (1-1.5 g carbohydrate in each boat) in the
glass tube or loaded directly into the tubular steel reactor (up to 15 g), and the whole system
was tightened. After the pyrolysis the resulting carbon materials were weighed again.
2. The subsequent sulphonation was performed by immersing up to 5 g of carbon materials in

a 500 ml round flask with 250 ml of H2 SO4 , either concentrated (96 wt%, Sigma-Aldrich) or
fuming sulphuric acid of 15 or 30% oleum (based on 30% SO3 /H2 SO4 , Merck). The solution
was heated in an oil bath to 150℃, where it was left for 15 h with a flow of Ar of 5 ml/min to
renew the gas atmosphere.
3. After sulphonation the sulphuric acid solution and the carbon material was carefully diluted
a factor of 10 in demineralised water. The resulting black powder was filteted and washed
extensively with H2 O at 80℃ until no trace of SO42 – -ions could be found in the filtrate by
49
testing for BaSO4 with concentrated aqueous Ba(NO3 )2 . Afterwards the powder was dried in a
muffle oven for several hours at 110℃.
2.2.5 Immobilisation of SPS and continuous esterification

For the preparation of the SPS supported in polyurethane (PUR) foam, SPS powder was prepared
according to the procedure mentioned in section 2.2.4. The foaming was performed by the Danish
Technological Institute (TI) by Jesper Bøgelund. The powder was nested in the PUR foam matrix
by being mixed with the reactant during the synthesis of PUR foam, however, the conditions and
reactants of the foaming process are not known as the synthesis was proprietary information of TI.
The appearance of the resulting was as a soft, open-celled, dark foam. Before reaction, the foam
catalysts were cut in pieces of about 3 mm on each side.
For preparation of impregnated SPS supported on ceramics, two types of ceramic structures were
used: Open-celled ceramic were obtained from Drache GmbH in the form of 2 cm thick ceramic
plates of an open-cell ceramic foam of Al2 O3 and ZrO2 (cell diameter ca. 3-4 mm). Four different
types of ceramic pellets were delivered by Saint-Gobain Norpro in the form of SiO2 (surface area 251
m2 /g, average pore diameter 11.1 nm, pore volume of 0.93 mL/g), two types of TiO2 (#1: Surface
area 150 m2 /g, average pore diameter 15 nm; #2: Surface area 40 m2 /g, average pore diameter 28
nm), and ZrO2 (surface area 186 m2 /g, containing 4.6% SiO2 as a binder). The SiO2 -, TiO2 -, and
ZrO2 -pellets were used as received after drying.
Al2 O3 -cylinders were cut in sizes of 9.5, 15 and 20 mm diameter and lengths of 20 mm by Jakob
Engbæk at TI. The received Al2 O3 open-cell foam cylinders were fragile after pyrolysis & sulphona-
tion so they were re-calcined at 1200℃ for 24 h in air with a heating ramp of 100℃/h, yielding a
much harder ceramic not degraded by the sulphonation solution. The pelletised ceramics were used
as is after vaporisation of any water present in the mesopores at 110℃. All ceramics were weighed
and impregnated with a sucrose solution. The sucrose-impregnated ceramics were afterwards pyro-
lysed at 400℃ for 15 h with a 100℃/h heating ramp, sulphonated with sulphuric acid for 15 h at
150℃, washed with 80℃ hot water and dried according to the procedure given in section 2.2.4.
The activity of the impregnated SPS (ISPS) was assessed by batch-mode esterification and trans-
esterification as described in section 2.2.1. Furthermore, the pelletised ISPS/SiO2 were tested as
catalysts for continuous esterification of lauric acid and methanol in 1:30 molar ratio, while the
open-cell ceramic ISPS/Al2 O3 was exposed to a 1:1:8 molar mixture of trioctanoate, lauric acid and
methanol. n-Dodecane was used as internal standard in both cases.
The reactor setup was a tube of 9.75 mm in inner diameter and 18 cm length, insulated within
a tubular furnace, which was connected with fittings to a setup with steel and rubber tubing and a
peristaltic feed pump on the inlet side, while a condensing tube with an open-close sampling valve
situated on the outlet side. A thermocouple attached to the outlet side of the reactor tube was used to
monitor the reaction temperature. The flow rate of feed solution was varied between 0.010 ml/min
and 0.200 ml/min and the reactor temperature was varied between 25 and 72℃. Upon loading of
the catalyst, the ends of the bed were plugged with quartz wool and connected to the pump. The
reactor was placed in a horizontal furnace, a thermocouple was connected and tightened to the outer
50
part of the reactor inside the furnace, followed by insulation by aluminium foil. A flow of methanol
was pumped to the reactor to empty it of air before starting the experimentation.
The open-cell ceramic ISPS/Al2 O3 was also tested as catalyst for the esterification and transes-
terification of rapeseed oil (obtained from supermarket) and waste animal fat (supplied by Daka
Biodiesel) with methanol in a pilot reactor setup. The reactor was a steel tube of 75 cm and 2.0 cm
inner diameter, which was connected with with a pump, an emulsifier with feed inlets and product
outlet, as well as a recycle loop. The same ISPS/Al2 O3 open-cell cylinders were used as catalysts
for both feedstocks (rapeseed oil treatment first). The reactor could be run without recycling of the
product or with a desired recycle ratio - and the feeding of reactants and the extraction of product
can be shut off to operate the reactor like a packed-bed batch reactor. Thus, the reaction mixture
can in principle be recycled in the loop infinitely until a desired conversion is reached. This means
that the kinetic behaviour of the reaction will be like a batch-reactor, while the hydrodynamic be-
haviour will be like a packed-bed reactor. The experimentation with industrial feedstocks have been
performed by Lene Fjerbæk at the University of Southern Denmark (SDU). Analysis of these samples
were made by an HPLC-method, however, details are not available about the apparatus and eluents
used.
2.2.6 Characterisation of SPS and ISPS

The BET surface areas of SPS and of SPS-impregnated and unimpregnated SiO2 -pellets and SPS-
Al2 O3 open-cell cylinders were determined by N2 physisorption. The samples were degassed prior to
measurement at 150 ℃ for 4 h under vacuum. Adsorption- and desorption isotherms were measured
with liquid N2 at 77 K on a Micromeritics ASAP 2020 pore analyser.
Elemental CHNS-analysis of the PS, SPS, ISPS/SiO2 -pellets and ISPS/Al2 O3 open-cell cylinders
was performed by Theis Brock-Nannestad at the University of Copenhagen with a Flash EA 1112 NC
Analyser from CE Instruments, ThermoFisher Scientific equipped with a TCD detector. The content
of carbon, hydrogen, nitrogen, and sulphur can be detected by CHNS-analysis, however, the oxygen
content cannot.
Aqueous titrations of the PS, SPS, SPS/SiO2 -pellets and ISPS/Al2 O3 open-cell cylinders were
carried out with a Metrohm autotitrator with a pH working-electrode. The titration of ca 0.2 g of the
suspended catalysts was performed at 600 rev/min magnetic stirring in a beaker using a KOH (>85%,
Fluka, <1% Na, <0.5% other impurities, Fluka) solution of 0.0359 M as titrant, which had been
adjusted by a tripled manual titration with 0.1000 M solution of aqueous hydrochloric acid (Titrisol,
Merck). Prior to titration, ca. 5 g of KCl (>99,5%, Merck) was added to the beaker containing the
catalyst to make sure that the sulphonic acid functionalities were ion-exchanged with potassium to
minimise the delay in pH on addition of base.
2.2.7 Basic catalysts for transesterification

A number of different catalysts were tested for their transesterification activity as suspected bases.
The activity of the catalysts were tested according to the basic procedure mentioned in section 2.2.1.
Butyl(methylimidazolium)amine was used as a cation in three different ionic liquids, with either
51
bromide ([BMIM−NH+2 ][Br – ]), boron tetraflourate ([BMIM−NH+2 ][BF4– ]) or bis(triflamide)

([BMIM−NH+2 ][NTf2– ]) as anion. All the ionic liquids were in-house synthesised chemicals (>98%
purity on NMR) by Olivier Nguyen van Buu.
Cesium carbonate (Cs2 CO3 , >99%, Aldrich), guanidinium carbonate (Gua2 CO3 , 99%, Aldrich),
guanidinium chloride (GuaCl, 99%, Sigma) were obtained and employed as catalysts, while magne-
sium lanthanum mixed-oxide (MgO · La2 O3 ) was synthesised in-house by Yury Gorbanev according
to a procedure given in the literature [315, 316] and used as well.
Tetramethylguanidine (TMG, >99%, Aldrich), 1,8-Diazabicyclo[5.4.0]undec-7-ene (DBU, >99.0%,
Fluka), 1,5-Diazabicyclo[4.3.0]non-5-ene (DBN, >98.0% (GC), Aldrich), 1,5,7-triazabicyclo[4.4.0]dec-
5-ene (TBD, >99.0%, Fluka) were applied as guanidine and amidine strongly basic catalysts.
1,8-Bis(tetramethylguanidino)naphtalene (Bis-TMG-Naphthalene, >98.0% (NT), Fluka) was a com-
mercial chemical, while Merrifield’s peptide resin (70-90 mesh, 1.0-1.5 mmol/g Cl- loading, 1 %
PVB/PS, Aldrich) linked with TMG in-house (ca. 0.7 mmol/g TMG), Benzyl-TMG, Hexyl-TMG, and
Propyl-TMG were in-house synthesised chemicals (>99% purity by NMR after synthesis) by Olivier
Nguyen van Buu and used as catalysts as well.
52
2.3 Sulphonic acid-functionalised ionic liquids
The sulphonated cation [BMIM−SO3 H+ ] with [NTf2– ] or [MeSO3– ] as anions were applied as catalytic
reaction media for conversion of the model mixture of methanol, lauric acid and trioctanoate in 8:1:1
molar ratio to methyl laurate and methyl octanoate.
Apparently, at 60℃ and 0.5 g of IL, the conversion of lauric acid amounted to 98% with both
liquids after 0.5 h and the conversion of trioctanoate was respectively 98% and 99% using the IL
with [MeSO3– ]- and [NTf2– ]-anions after 24 h. The latter observation was surprising, as it was not
suspected that the acidic catalyst would be very active for transesterification. This prompted the
repetition of the two time series and the addition of two other time series experiments with different
ionic liquids as catalysts.
Thus, The ILs of [BMIM−SO3 H+ ] with [NTf2– ], [MeSO3– ], and [HSO4– ] and the IL of [BP−SO3 H+ ]
with [HSO4– ] were tested in the latter set of experiments. The values for the conversions and yields
of the four ILs are shown in Figure 2.5. From the experiments performed twice, the average values
from the two experients are used, as they were within ca. 2% deviation.
All the ILs showed complete turnover of the lauric acid within 3 h. The fastest conversion after 0.5
and 1 h was reached by [BMIM−SO3 H+ ][MeSO3– ], barely ahead of [BMIM−SO3 H+ ][NTf2– ], whereas
the turnover of the lauric acid over the ILs with [HSO4– ]-anions were somewhat slower. Some minor
deviation in the conversion of lauric acid can be observed in the time series over [BMIM−SO3 H+ ]-
[HSO4– ] and [BP−SO3 H+ ][HSO4– ] in the first samples.
However, the analysis of the resulting methyl laurate presented in Figure 2.5 appeared somewhat
peculiar, since some concentrations reached over 100% during the first hours, followed by levelling
off at about 80% yield. No other products containing the lauric acid chain have been observed. It
may be suspected that the missing methyl laurate has been hydrolysed during the sample preparation
before the GC analysis due to trace amounts of water in the chemicals.
1 1,2
Conversion of lauric acid
1
Yields of Me-laurate
0,8
0,8
0,6
0,6
0,4
0,4
0,2 0,2
0 0
0 0,5 1 1,5 2 2,5 3 0 12 24 36 48 60 72 84 96
Time (h) Time (h)
Figure 2.5: Conversion of lauric acid (left-hand side) and yield of methyl laurate (right-hand side) with sul-
phonic acid-functionalised ionic liquids as catalysing reaction medium. Legend: [BMIM−SO3 H+ ][-NTf2– ] ( ),
[BMIN−SO3 H+ ][MeSO3– ] ( ), [BMIM−SO3 H+ ][HSO4– ] ( N ), and [BP−SO3 H+ ][HSO4– ] ( • ). Conditions: 60℃,
4.3 mmol trioctanoate, 4.3 mmol lauric acid, 34.6 mmol methanol, dodecane as internal standard.
It should also be noted that the ionic liquids and the reactant solution formed two separate
phases, ionic liquid on the bottom and the organic reaction mixture on the top, however, it is not
known how much of the reactants and products are absorbed in the IL-phase and have not measured.
No visible expansion of the ionic liquid phase could not be observed. The stirring rate was sufficiently
53
2.3. Sulphonic acid-functionalised ionic liquids
fast to emulsify the ionic liquid-phase into small droplets suspended in the organic phase, so sam-
ples were taken out from the upper phase after a short period without stirring. Esters or acids may
have been absorbed .
The conversions of trioctanoate and the yields of methyl octanoate from the analyses of the
samples, as shown in Figure 2.6, reveals that the ionic liquids were also active for the conversion
of the triglycerides at 60℃. The more active catalysts for esterification of lauric acid were also
more active in the transesterification of trictanoate, in all probability due to higher acidity. In
fact, after 24 h, over 95% conversion of trioctanoate was observed with the [BMIN−SO3 H+ ][NTf2– ],
[BMIN−SO3 H+ ][MeSO3– ], and [BMIN−SO3 H+ ][HSO4– ].
1 1
Conversion of trioctanoate
Yields of Me-octanoate
0,8 0,8
0,6 0,6
0,4 0,4
0,2 0,2
0 0
0 12 24 36 48 60 72 84 96 0 12 24 36 48 60 72 84 96
Time (h) Time (h)
Figure 2.6: Conversion of trioctanoate (left-hand side) and yield of methyl octanoate (right-hand side) with sul-
phonic acid-functionalised ionic liquids as catalysing reaction medium. Legend: [BMIM−SO3 H+ ]-[NTf2 −] ( ),
BMIN−SO3 H+ -MeSO3– ( ), BMIM−SO3 H+ -HSO4– ( N ), and BP−SO3 H+ -HSO4– ( • ). Conditions: 60℃, 5
mmol trioctanoate, 45 mmol methanol, dodecane as internal standard.
The high conversion of >95% and the somewhat lower yields achieved at 60℃ are comparable
to the FAME yield of 83% and 88% achieved by Li et al. by using [BP−SO3 H+ ][CF3 SO3– ] at 80℃
after 6 and 8 h [141], respectively. This is furthermore comparable to the yields achieved by Liang
and Yang around 98% after 7 h at 70℃ using a tetra-(butyl-sulphonic acid) ionic liquid based on
hexamethylenetetramine, or Ghiaci and co-authors achieving 75, 77, and 82% yields after 6 h at
60℃ over various polysulphonic polyaromatic ionic liquids, considering the multifunctional nature
of the ionic liquids applied in both literature reports [142, 309].
The yields of methyl octanoate were somewhat lower than the conversion of trioctanoate would
predict. The gaps observed between the amounts of trioctanoate and methyl octanoate was found not
to be due to hydrolysis of octanoate esters to octanoic acid, as octanoic acid was found at maximally
1-2% of the samples of the reaction mixtures throughout the time series. The missing compounds
were, however, found to be larger amounts of the intermediates of glyceryl mono- and di-octanoate
from the trioctanoate, which had not been converted to the methyl octanoate. Some of the ester or the
corresponding acid may as well be absorbed in the ionic liquid-phase, which has not been analysed.
The production of methyl octanoate stagnated after 24 h, so it is suspected that the reaction rates
of the mono- and di-octanoates were slower than that of the trioctanoate. It is known that mono- and
di-glycerides act as emulsifiers, and during the end of the time series, a thickening of the reaction
mixtures could be observed (the solutions became “cloudy”). The increased viscosity of the mixture
could therefore have led to lower rates of reaction to to decrease of diffusion rate and poorer mixing.
Strong acids usually need higher temperatures to function as catalyst for the direct transesterifi-
cation, as the conversion is normally unacceptably slow around 60℃ [86, 93, 317, 318]. Therefore
54
it may be speculated that the reaction have taken place due to the formation of octanoic acid inter-
mediate through acidic hydrolysis of the glycerides to free fatty acid, which was then converted to
methyl octanoate, as given in Figure 2.4. However, very little octanoic acid was observed during the
reaction, possibly also deriving from the fact that esterification of the free fatty acids proceeded very
fast (as was the case of the esterification of lauric acid) in the ionic liquids.
Water could be added to the reaction mixture in varying amounts to investigate the role of water
in the mechanism and possibly elucidate this hypothesis. Liang and Yang found a slight depression
of the FAME yields with addition of up to 2 wt% water to the reaction mixture [142]. This indicates
that the reaction from trioctanoate taking place in the acidic medium was, as claimed, direct acidic
transesterification - although this property may well be dependent on the hydrophilicity of the ionic
liquid applied.
55
2.4. Sulphonic acid-functionalised pyrolysed carbohydrates
2.4 Sulphonic acid-functionalised pyrolysed carbohydrates
From the previous works in literature it is evident that SPS can have high sulphonic acid loading and
acidity, resulting in a high esterification rate at benign temperatures, but the preparation conditions
have a profound influence on the properties of the final catalyst [313].
It has been reported that the structure of the resulting material is an amorphous graphene layer
structure with its very few oxygen functionalities situated on the edges of the graphene layers (cf.
Figure 2.2) [312]. However, the removal of oxygen functionalities is not complete during the pyro-
lysis of the carbohydrates, especially at lower temperature, as given by Okamura et al. [313]. It may
be expected that the concentrated sulphuric acid at 150℃ during sulphonation will also react with
any phenolic groups present, creating water and a phenylic hydrogen sulfate group (this may later
hydrolyse during washing with hot H2 O).
The sulphonic acid-functionalities are as strong as sulphuric acid [132], so the loading and posi-
tion of the sulphonic acid groups determines the capacity for and the possible rate of esterification.
An investigation of the dependency of the carbohydrate source, the pyrolysis conditions and the
sulphonation medium have therefore been performed in the present work. A model mixture of lau-
ric acid, trioctanoate and methanol in a molar ratio of 1:1:8 has been used as reaction medium for
the activity tests. The yields reported are obtained by relating the obtained GC areas with those of
reference solutions.
A photograph of sucrose and starch before and after the pyrolysis at 400℃ for 15 h is depicted
in Figure 2.7. After sulphonation, washing and drying, a batch of the grainy black SPS powder
appeared as depicted on the photograph in Figure 2.8.
Figure 2.7: Photograph showing pyrolysed (upper left) and unpyrolysed (lower left) Figure 2.8: SPS,
starch and pyrolysed (upper right) and unpyrolysed (lower right) sucrose. washed and dried.
2.4.1 Variation of the carbohydrate source

The influence of the carbohydrate source on the SPS catalyst activity was investigated by synthesising
the SPS from sucrose, glucose, starch, and cellulose as carbohydrate precursors. The results of the
SPS-catalysed time series from 0 to 96 h are shown in Figure 2.9. It is clear that all of the tested SPS
feedstocks provided suitable catalysts for esterification as lauric acid and methanol yielded more
than 95% methyl laurate over the tested materials. It can be seen that the sucrose-based SPS show
a slightly higher than the cellulose- and glucose-based SPS. It is furthermore observed that some
percent of methyl octanoate resulted from the acidic transesterification of trioctanoate, at least with
the sulphonated pyrolysed cellulose and glucose. Initially it was observed that methyl octanoate is
56
also produced over the glucose-based SPS, albeit this is not found in the reaction mixture at later
instances in the experimentation. The tiny amounts observed are, however, almost in the range of
experimental inaccuracy.
From these results it may be suspected that the pyrolysis of the materials produce a material that
is not very dependent on the type of carbohydrate used for the SPS. It was expected that the cellu-
lose was more thermally stable and therefore a more difficult molecule to use to achieve sufficient
pyrolysis, but its activity was nonetheless on the same scale as the glucose- and sucrose-based SPS.
1 0,1
Yield of Me-laurate
Yield of Me-octanoate
0,8 0,08
0,6 0,06
0,4 0,04
0,2 0,02
0 0
0 4 8 12 16 20 24 0 12 24 36 48 60 72 84 96
Time (h) Time (h)
Figure 2.9: Yield of methyl laurate (left-hand side) and methyl octanoate (right-hand side) over sulphonated, py-
rolysed carbohydrates as a function of time by variation of the catalyst carbon source (pyrolysis: 400℃ for 15 h,
sulphonation 150℃ for 15 h with conc. H2 SO4 ). Legend: Glucose ( ), Cellulose ( ), Sucrose ( N ), Conditions:
60℃, 4.3 mmol trioctanoate, 4.3 mmol lauric acid, 34.6 mmol methanol, dodecane as internal standard.
2.4.2 Variation of the pyrolysis conditions

The conditions for the slow pyrolysis to achieve the carbonaceous catalyst material was varied while
the subsequent sulphonation, washing and drying conditions were kept constant. Thus, a pyrolysis
time of 1, 4, and 15 h at 400℃ and a pyrolysis temperature of 300, 350, 400, and 450℃ at 15 h of
pyrolysis were evaluated. The produced SPS catalysts were applied to react the model mixture of
methanol, lauric acid and trioctanoate at 60℃ with the results for esterification and transesterifica-
tion given in Figure 2.10.
1 0,05
Yield of Me-laurate
0,8 0,04
0,6 0,03
0,4 0,02
0,2 0,01
0 0
0 1 2 3 4 5 6 7 0 12 24 36 48 60 72 84 96
Time (h) Time (h)
Figure 2.10: Yield of methyl laurate (left-hand side) and methyl octanoate (right-hand side) over sulphonated, py-
rolysed sucrose (SPS) as a function of time by variation of the carbonatisation temperature and time for the SPS.
Legend: 300℃ for 15 h ( ), 350℃ for 15 h ( ), 450℃ for 15 h ( N ), 400℃ for 1 h ( ◦ ), 400℃ for 4 h
( ), 400℃ for 15 h ( • ). Reaction conditions: 60℃, 4.3 mmol trioctanoate, 4.3 mmol lauric acid, 34.6 mmol
methanol, dodecane as internal standard.
In all cases the sulphonated pyrolysed carbohydrates worked quite efficiently for the esterifici-
ation to methyl laurate and tended towards full conversion within a few hours, albeit at somewhat
57
different rates. Only vague variation tendencies as a function of the pyrolysis conditions during SPS
synthesis may be observed. After 3 h of reaction 100% yield of methyl laurate is observed in the
samples of SPS that have been pyrolysed for 15 h at 350, 400, and 450℃. This could indicate that
sufficient pyrolysis time is an advantage for this catalyst type. This is possibly due to the fact that
as much water as possible should be removed from the carbohydrates to avoid high dilution of the
sulphuric acid during sulphonation, thus yielding a higher amount of sulphonic acid functionali-
ties. 15 h at 300℃ was not sufficient to achieve the same yield in 3 h achieved at higher pyrolysis
temperatures. This tendence is, however, only clear from the 3 h samples. This observation does not
explain why there was not a bigger difference between the various pyrolysis temperatures; neither
does it explain why the SPS from 1 h pyrolysis at 400℃ was more active than that from 4 h pyrolysis.
It can safely be concluded that no significant transesterification activity was observed, as the yield
never climbed above 2% and no methyl octanoate were observed in the most samples taken after 7
h. The low amounts observed also lie on the border of GC noise and experimental uncertainty.
No systematic influence of the pyrolysis temperature of the SPS on the yield of methyl ester can
be conluded, although the pyrolysis time and temperature should at least be sufficient to remove as
much water as possible from the carbohydrates. 400℃ have been reported to be the optimal pyrolysis
temperature of glucose, while lower and higher temperatures yielded slightly lower conversion -
interestingly SPS from pyrolysis at temperatures of 550 and 600℃ were almost inactive and non-
acidic in methanol and water due to absence of swelling of the catalysts [313].
It may be speculated that the synthesis of the SPS could be simplified by performing the initial
dehydration of the carbohydrate in sulphuric acid instead of by pyrolysis, but it would most likely
require using bigger amounts of sulphuric acid for the same batch to gain the same sulphonation
degree. This has therefore not been attempted.
2.4.3 Variation of the sulphonation medium

The sulphonation medium of pyrolysed sucrose and starch were varied to evaluate if this influenced
the activity of the SPS catalyst during esterification and transesterification of lauric acid and trioc-
tanoate, respectively, with methanol. Respectively 30% and 15% fuming as well as concentrated sul-
phuric acid (96 wt%) were used for the sulphonation, and the SPS catalysts obtained were compared
also with the activity of unsulphonated pyrolysed sucrose in Figure 2.11. 30% fuming sulphuric acid
was on stock while 15% fuming sulphuric acid was made in-situ by mixing concentrated sulphuric
acid with the 30% fuming sulphuric acid.
The standard catalyst based on pyrolysed sucrose sulphonation with concentrated sulphuric was
more efficient compared with those based on the fuming sulphuric acid, as the yield of methyl laurate
grew most steeply with the standard catalyst. The catalysts based on fuming sulphuric acid were not
as efficient, however, some of the masses of catalysts used of these materials were also not as high as
0.21 g.
It was found that the fuming sulphuric acid, especially the 30% SO3 /H2 SO4 partly degraded
the macrostruture of the pyrolysed material. The sulphonation at 150℃ in 96% H2 SO4 resulted a
more grainy SPS consisting of coarse particles (as depicted in Figure 2.8). This meant that much of
58
the carbonaceous material was washed away during the filtration, and not enough catalyst could be
recoved to allow ca. 0.21 g of catalyst to be used in all the 4.00 ml reaction mixture. The yields
of methyl laurate after 0.25, 0.5, and 1 h have therefore been related to the catalyst mass via the
apparent catalyic reaction rates, which are given in Table 2.1. The reaction rates calculated over a
given catalyst should be equal at low conversions, but they should lower over time as the reaction
approaches equilibrium, which was also the case for most of the catalysts in the table. Some deviation
is found to this rule.
Interestingly, the yields of methyl octanoate grew over 10% after 96 h with some of the SPS ca-
talysts based on fuming H2 SO4 . Thus it is probable that the more severe sulphonation conditions
yield SPS with higher acidity, in turn leading to a more facile transesterification catalyst. Transesteri-
fication have been reported with SPS based on glucose recently, albeit at higher temperatures and
thus much higher rates - 96–98% yield of methyl oleate was achieved from triolein after 5 h at 130℃
[130].
1 0,2
Yield of Me-laurate
0,8
0,15
0,6
0,1
0,4
0,05
0,2
0 0
0 4 8 12 16 20 24 0 12 24 36 48 60 72 84 96
Time (h) Time (h)
Figure 2.11: Yield of methyl laurate (left-hand side) and methyl octanoate (right-hand side) over sulphonated, pyrol-
ysed sucrose as a function of time and variation of carbohydrate and percentage of SO3 in H2 SO4 during sulphona-
tion. Legend: Sucrose, 30% SO3 /H2 SO4 ( ), Starch, 30% SO3 /H2 SO4 ( ), Sucrose, 15% SO3 /H2 SO4 ( N ),
Starch, 15% SO3 /H2 SO4 ( ◦ ), Sucrose, 0% SO3 /H2 SO4 ( ), Pyrolysed sucrose (not sulphonated) ( • ). Re-
action conditions: 60℃, 4.3 mmol trioctanoate, 4.3 mmol lauric acid, 34.6 mmol methanol, dodecane as internal
standard.
Table 2.1: Reaction rates based on the yield of methyl laurate after 0.25, 0.5, and 1 h of reaction. Conditions: 60℃,
4.3 mmol trioctanoate, 4.3 mmol lauric acid, 34.6 mmol methanol, dodecane as internal standard.
Catalyst Catalyst mass [g] Rate [mmol/g/h)]

0.25 h 0.5 h 1h
S.P. sucrose, 30% SO3 /H2 SO4 0.0760 8.92 15.2 14.3
S.P. starch, 30% SO3 /H2 SO4 0.1709 19.3 18.1 13.6
S.P. sucrose, 15% SO3 /H2 SO4 0.2032 19.7 17.2 12.30
S.P. starch, 15% SO3 /H2 SO4 0.1606 3.01 6.85 7.00
S.P. sucrose, 96 wt% H2 SO4 0.2120 15.7 13.2 15.0
Pyrolysed sucrose 0.2800 0.52 0.068 0.13
It can be observed that the esterification rate over PS was very low, as about 5% and 7% yield of
methyl laurate was obtained after respectively 24 h and 96 h of reaction at 60℃. The rate was so low
that it may be speculated if the observed yield was due to self-catalysed esterification by the lauric
acid in the model reaction mixture. It can be stated that the pyrolysed sucrose (PS) was practically
inactive even though it may contain minor amounts of weak carboxylic acids linked to the graphene
sheets of the carbonised material [132, 312]. Some transesterification was indicated with this catalyst
59
as well, which should not be expected from weak acids.

Using fuming sulphuric acid for the catalyst sulphonation mostly enhanced the esterification of
lauric acid with methanol, but with the 15% SO3 /H2 SO4 the rate was only 40% of that of the sucrose
equivalent in Table 2.1, using the data after 0.5 h reaction. It is not clear why this material was
much less active than the other SPS catalysts having been exposed to fuming sulphuric acid. An
enhancement of the esterification rate of 60-95% by using 15% fuming sulphuric acid has previously
been found in literature by sulphonation of glucose pyrolysed at 400℃ [128, 313].
It can be speculated that the sulphonation in fuming sulphuric acid allowed removal of more
remaining oxygen in the form of water from the carbonaceous material, possibly also some binders
connecting the graphene layers. On the other hand the stronger sulphonation conditions also yielded
more sulphonic acid-functionalities on the resulting material. This has for instance been found by
Takagaki and co-authors [132].
Thus it can be concluded that using fuming sulphuric acid for the sulphonation enhanced the
formation of methyl ester during fatty acid esterification. However, the retention of the sulphonated
catalyst during filtration was much poorer with the sulphonated materials.
Reuse of the investigated catalysts has not been attempted so it cannot be established if any trend
in the activity as a function of the pyrolysis conditions would establish after several cycles of re-
use. However, the spent reaction mixtures were tested with an aqueous solution of Ba(NO3 )2 and no
precipitate of BaSO4 were oserved in any solution, thus it can be suspected that leaching of sulphate
did not take place and the observed differences in activity would be persistent over time.
´
60
2.5 Immobilisation of SPS and continuous esterification
The application of a strong sulphonic acid based on pyrolysed carbohydrates for esterification, as
mentioned in 2.4, could hold advantages to use in the esterification step of FAME production. How-
ever, the use of powders in liquid-phase large-scale chemical operation is usually not desirable, even
though forming powders into extrudates would be a suggestion with inorganic powders. Therefore
immobilisation of the active SPS phase (ISPS) on more suitable supports is necessary.
Impregnation of sucrose has been performed on various ceramic support materials followed by
pyrolysis, sulphonation, washing and drying. Furthermore, the acidic SPS powder has been sus-
pended in polyurethane foam. These two different types of immobilised or supported SPS catalysts
have been tested for esterification and transesterification of lauric acid and trioctanoate, respectively,
with methanol, analogously to the unsupported SPS described in section 2.4.
In addition, a pelletised ISPS/SiO2 and an open-cell ISPS/Al2 O3 catalyst have been tested for the
continuous esterification of a model feed containing methanol and lauric acid in a tubular packed-
bed reactor, and an open-cell ISPS/Al2 O3 have moreover been tested in transesterification and esteri-
fication of respectively rapeseed oil and abattoir waste fat with methanol.
2.5.1 Impregnation on ceramics and properties of SPS

The ceramic support materials were impregnated with varying amounts of sucrose via aqueous solu-
tions prior to pyrolysis at 400℃/15 h and sulphonation at 150℃/15 h. Table 2.2 shows the applied
catalysts and the data of their impregnation properties. The pelletised ceramics could be loaded
with up to 10% of their own mass as sucrose, however, the open-cell ceramic cylinders had a much
bigger hold-up of free volume in the cells and could potentially absorb many times more sugar than
the pellets. However, it is evident that the loosely bound pyrolysed and sulphonated impregnated
sucrose on the open-cell alumina is also more easily washed away during the sulphonation, and a
larger part of this organic matter is removed during pyrolysis as well.
During the sulphonation, the TiO2 and the ZrO2 pellets were somewhat degraded - the TiO2
crumbled into smaller black pieces while the ZrO2 degraded into small black flakes and black pow-
der. It was thus evident that the pellets could not endure the treatment with sulphuric acid at 150℃.
Titania and zirconia are known to be resistant even to strong acids, so it is suspected that the pellets
could have contained a basic binding matrix, for instance CaO. This has, however, not been reported
by the manufacturer. Nonetheless, the ISPS degraded pellet materials were applied as catalysts after
washing and drying, using 0.5 g of each.
The blackened SiO2 -pellets and the open-cell Al2 O3 materials formed by impregnation, pyroly-
sis and sulphonation - ISPS - are depicted in the photographs of Figure 2.12. Characterisation of the
powder have been performed by BET area estimation by N2 physisorption at 77 K, by CHNS elemen-
tal analysis, and by wet titration of the solids. This is given in Table 2.3. It can be observed that the
SPS had an almost negligible surface area. The BET technique is not that precise, and two repetitive
measurements showed no surface area at all for the SPS - however, this magnitude of SPS surface
area have been reported previously, and it is known that the surface area increasis by swelling in
61
2.5. Immobilisation of SPS and continuous esterification
liquid solvent [132, 313]. The surface areas of the impregnated ceramics are apparantly very depen-
dent on the original ceramic material. ISPS/alumina lost most of its interior surface area while the
ISPS/silica pellets did not and the alumina absorbed more sugar during impregnation, even though
the net gain in weight was smaller of the ISPS/alumina.
Table 2.2: Impregnation, carbonatisation and sulphonation.

Ø: Diameter, L: Length; OCC: Open-cell ceramic; IS: Impregnated sucrose; IPS:
Impregnated pyrolysed sucrose-impregnated; ISPS Impregnated sulphonated pyrol-
ysed sucrose.
weight gain of support

Batch Material Dimensions (Ø × L) IS IPS ISPS
[mm × mm] [wt%] [wt%] [wt%]
#1 ZrO2 -pellets 3 × 3-8 a 9.47% 3.8% 4.5%
#2 SiO2 -pellets 3 × 3-8 a,b 9.48% 3.8% 4.5%
#3 TiO2 -pellets 3 × 3-8 a 7.95% 3.19% 3.78%
#4 TiO2 -pellets 3 × 3-8 a 7.37% 2.96%
#5 Al2 O3 -OCC 14.5 × 20 a 39.59% 10.75% 1.12%
#6 Al2 O3 -OCC 20 × 20 a 19.08% 5.53% 3.87%
#7 Al2 O3 -OCC 9.5 × 20 b 7.36% 2.00% 1.40%
#8 Al2 O3 -OCC 20 × 20 c 32.74% 8.48% 5.73%
a tested in batch esterification of model feed
b tested in continuous esterification of model feed
c tested in continuous esterification and transesterification of industrial
feedstocks
Figure 2.12: Impregnated sulphonated pyrolysed sucrose on SiO2 -pellets (left-hand side) and Al2 O3 -open-cell ce-
ramic foam right-hand side.
From the gross formula proposed from the CHNS-analysis it appears that the original ratio of
H:O in the sucrose is more or less intact after pyrolysis. This means that one of the most important
processes of the pyrolysis is the formal removal of water, maybe some CO and CO2 , along with some
tar oils (which were observed as condensating products in the pyrolysis reactor outlet). However,
this H:O-ratio changed during sulphonation, where the amount of hydrogen was severely lowered
(about 60 % compared to the PS), while the amount of oxygen removed is only (28%) - it should be
expected that the sulphuric acid, being highly hygroscopic, would lead primarily to dehydration and
sulphonation of the pyrolysed sucrose.
The oxygen amount of the biomass cannot be estimated on the impregnated from the CHNS-
62
analysis and can only be estimated by a mole balance in PS and SPS. Importantly, the loading of
organic matter is higher on the ISPS open-cell alumina compared to the pelletised silica, and the
amount of sulphate functionalities in the organic matter itself is apparantly 65% percent higher
for the ISPS/alumina. It may therefore be expected that the ISPS/alumina is much more active
per mass than the ISPS/silica. It is also obvious that the content of hydrogen is much higher in
the ISPS/alumina than the ISPS/silica. Possibly the ISPS/alumina could have contained a lot of
water, whereby the analysis of H is affected, since the high amount of H in ISPS/alumina would
otherwise correspond almost to alkene hydrocarbons. This is also indicated by the weight gain of
the ISPS/alumina after synthesis, which was only 3.87% (while 10.3% of the sample during CHNS-
analysis was measurable compounds).
By titration of the PS, SPS and the ISPS/alumina it can be seen that a loading of almost 3 mmol
H+ /g can be achieved in the SPS powder, however, the open-cell ISPS/alumina did not contain as
many acidic sites when compared to the loading of organic matter impregnated on the ceramic, thus
the sulphonation degree cannot be directly transferred to the supported catalyst. The PS contained
very little acidity, possibly in the form of a few carboxylic acids. It was also attempted to titrate the
ISPS/silica, however, it did not show a lowering of pH by addition of a KCl-solution, and the titration
yielded negligible concentrations of H+ . This is not currently understood.
Table 2.3: Areas, composition and acidity of investigated PS, SPS, ISPS and support materials.
Batch Material BET area Aciditya Organic matterb Gross formulac

[m2 /g] [mmol H+ /g] [wt%] [mol/mol]
PS 0.053 100% CH1.20 O0.58
SPS 0.03 2.94 100% CH0.48 O0.42 (SO3 ) 0.016
#2 SiO2 pellets 233 0%
#2 ISPS/SiO2 pellets 177 0.00 6.1% CH0.61 Ox (SO3 ) 0.0096
#6 open-cell Al2 O3 39.2 0%
#6 open-cell ISPS/Al2 O3 3.5 0.174 10.3% CH1.89 Ox (SO3 ) 0.016
a Based on ion-exchanged titration.
b Known organic matter calculated from the CHNS-analysis, which means that non-sulphonate-
oxygen (eg. water) cannot be accounted for on the ISPS.
c Suggested from the CHNS-analysis on basis of the known organic matter.
2.5.2 Activity of impregnated SPS on ceramics

A series of batch experiments were conducted to compare the ceramic ISPS catalysts with the unsup-
ported SPS batch experiments. Figure 2.13 shows the yields of methyl laurate and methyl octanoate
during sampling from the catalysed esterification and transesterification over the ISPS on ceramic
pellets and open-cell ceramic foam. 0.5 g of each of the pelletised materials were used (batch # 1-4),
while respectively 7.0 g and 4.2 g were used of batch # 5 and 6. Batch # 1-5 were with 4.00 ml (3.35
g) of reactant mixture (1:1:8 molar ratio of trioctanoate, lauric acid and methanol), while 16 ml was
used for batch # 6.
The most active catalysts were the (partly crumbled) pelletised cylinders of TiO2 -cylinders. How-
ever, batch # 5 of ISPS/alumina was almost as active (not all of the catalysts could be covered with
reactant solution, so comparisons should be made with caution). Batch # 6, having a 6.7 times lower
63
catalyst-to-reactant ratio than batch # 5, was also less active than batch # 5, but not as much as
the difference in should imply due to the lack of reactant solution to completely cover the batch #
ISPS/Al2 O3 . The least active catalyst for the esterification was the ISPS/silica, reaching just over
80% yield of methyl laurate after 96 h.
1 0,15
0,8 0,12
Yield of Me-otanoate
Yield of Me-laurate
0,6 0,09
0,4 0,06
0,2 0,03
0 0
0 12 24 36 48 60 72 84 96 0 12 24 36 48 60 72 84 96
Time (h) Time (h)
Figure 2.13: Activity measurements of the impregnated sulphonated pyrolysed sucrose in batch-mode. Leg-
end: ISPS/ZrO2 -pellets, batch # 1 ( ), ISPS/SiO2 -pellets, batch # 2 ( ), ISPS/TiO2 -pellets, batch # 3 ( N ),
ISPS/TiO2 -pellets, batch # 4 ( • ), ISPS/Al2 O3 -open-cell, batch # 5 ( ), ISPS/Al2 O3 -open-cell, batch # 6 ( ◦ ).
Reaction conditions: 60℃, 4.3 mmol trioctanoate, 4.3 mmol lauric acid, 34.6 mmol methanol, dodecane as internal
standard.
In all of the samples with pelletised catalyst supports, the yield of methyl octanoate grew at the
highest rate for the first 7 h whereupon the formation settled on a somewhat slower pace for all
the tested ISPS ceramic pellets. The highest activity for transesterification to yield methyl octanoate
were found with the two ISPS/TiO2 -catalysts. The ISPS on SiO2 and ZrO2 only yielded about 35-40%
of the yields obtained for ISPS/TiO2 during the entire time series. The two open-cell ISPS/Al2 O3
catalysts showed different activity, and batch # 5 was also the most active during transesterification
after 96 h. The yields after 24 h and 48 h of batch # 5 were, however, too high compared to the
remaining samples taken out and should be disregarded in terms of activity comparison.
Compared with the SPS, the ISPS show over an order of magnitude lower activity, however, the
loading of active phase is only a fraction of that present in the unsupported SPS (merely 3 - 7%,
according to Table 2.3), which in fact make the supported and unsupported catalysts comparable.
2.5.3 Immobilisation and activity of SPS in polymer foam

Another strategy for supporting the SPS catalyst on a suitable carrier was building the SPS into
a polymer matrix with suitable porosity. 6 different PUR foam catalysts are employed with SPS-
loadings of either 16 wt% (3 different), 17 wt%, and 20 wt% loading of SPS and a blank sample of
0% SPS in the PUR foam matrix. However, no further information is available about the foaming
proces, apart form the fact that the method and chemicals used should produce an internal open
network of pores to allow reactants and products to diffuse in and out of the polymer matrix. It
is, however, not known if the polymers foamed together with SPS obtain this open-porous internal
network during foaming. The physical appearance of the SPS foam matrix samples was as a hard,
slightly plastic and visibly porous material, in which also the SPS powder particles were visible as
can be seen in Figure 2.14. Compared to the SPS-free PUR having a light yellowish colour, the PUR
64
Figure 2.14: Appearance of the polyurethane polymer matrices containing respectively 0% (left), 16% (middle), and
20% (right) SPS.
matrices containing SPS had a greyish colour due to the addition of the black powder.
Ca. 0.56 g of all the catalysts were exposed to 3.35 g of a 1:1:8 molar mixture of lauric acid, trioc-
tanoate and methanol. The yields of methyl laurate and methyl octanoate in the SPS-containing PUR
polymer matrices (SPS-PUR) as a function of time are shown in Figure 2.15. All the SPS-containing
foams show a rise in the methyl laurate concentration over the first hours, but the rises in the yields
started to stagnate after 7 or 24 h for all the SPS-PUR samples. After 96 h, only about 55% yield is
obtained in the best case, using the PUR foam containing 17 wt% SPS. The 20 wt% SPS-PUR yielded
about 40% methyl laurate, while the three 16 wt% SPS-PUR samples showed even lower yields be-
tween 15% and 35% after 96 h. For comparison, the concentration of methyl laurate over the blank
PUR foam containing no SPS rose steadily and yielded about 10% methyl laurate after 96 h.
1 0,3
0,25
0,8
Yield of Me-laurate
0,2
0,6
0,15
0,4
0,1
0,2
0,05
0 0
0 12 24 36 48 60 72 84 96 0 12 24 36 48 60 72 84 96
Time (h) Time (h)
Figure 2.15: Activity measurements of the impregnated sulphonated pyrolysed sucrose in batch-mode. Legend:
17% SPS-PUR ( ), 16% SPS-PUR # 1 ( ), 16% SPS-PUR # 2 ( N ), 20% SPS-PUR ( • ), 16% SPS-PUR # 3
( ), Blank PUR # 6 ( ◦ ). Reaction conditions: 60℃, 4.3 mmol trioctanoate, 4.3 mmol lauric acid, 34.6 mmol
methanol, dodecane as internal standard.
Minor amounts of methyl octanoate were also seen to be formed during the course of the experi-
mentation. A few of the values of methyl octanoate yield in the 24 h and 48 h samples are over 10%,
but show a drop towards the 96 h sample. Generally these outlying sample concentrations must be
considered erroneous outliers compared to the reaction progress observed in all of the other samples.
From Figure 2.15 it is observed that the yield of methyl octanoate over the blank sample rose faster
than any of the the SPS-PUR indicated a yield of 25% after 96 h, higher than in any of the other sam-
ples. This is peculiar and cannot be explained from acidic or basic properties of the PUR material,
as the polymer linkage in PUR is a neutral carbamate functionality. The chromatogram areas of the
samples were quite small in all of the reactants and products, so some systematic inaccuracy during
65
sampling and noise signals during GC-FID analysis may have led to this error.
It can be concluded that the SPS-PUR did not show high esterification activities, at least not
comparable to the pure SPS powder mentioned in section 2.4, which yielded up to around 90%
methyl laurate after 1 h reaction. This could be due addition of the solid acid to the PUR foaming
mixture, which was likely affected by acid. The foaming of PUR from isocyanate and alcohol is
catalysed by nucleophiles (usually bases like DABCO (1,4-diazabicyclo[2.2.2]octane) are used). Also
it is uncertain whether or not the pore structure of the claimed blank sample has resulted during
foaming of the SPS-PUR catalysts as well. The sulphonic acid functionalities of the SPS may be
completely covered in the polymer and not be accessible. Characterisation and investigation of the
foaming would be necessary and highly desirable to give an explanation for the low activity observed.
2.5.4 Continuous esterification of model feed

Two of the SPS-impregnated ceramics were loaded into a tubular steel reactor of 9,5 mm in inner
diameter × 15 cm in length and tested in the continuous esterification, namely the ISPS/SiO2 -pellets
and the open-cell ISPS/Al2 O3 . The reactor dimensions corresponded to a loading of 7 open-cell
ISPS/Al2 O3 cylinders of 4.354 g (batch # 7) and 4.582 g of ISPS/SiO2-pellets (batch # 2).
The ISPS/SiO2 -pellets were tested as catalysts in the esterification of a 30:1:0.2 mixture of the
molar ratios of respectively methanol, lauric acid, and dodecane as an internal standard. 3 samples
were taken at each steady-state condition and the average mole fraction calculated at each condition.
However, the sum of the GC yields of methyl laurate and remaining lauric acid varied between 80%
and 120%. To even out these fluctuations, the yield of methyl laurate and remaining lauric acid
was calculated by using the two GC areas with respect to their response-factors obtained from GC
reference standards of the pure compounds in methanol. This was justified as no side-reactions take
place during the esterification-hydrolysis equilibrium around 60℃ in dilute solutions of carboxylic
acid in alcohol. When doing this, the yields of methyl laurate and unconverted lauric acid obtained
were within 3-4% at each condition, but the internal standard was at the same time disregarded.
1
Yields during continuous esterification
0,8
0,6
0,4
Yield
0,2
Figure 2.16: Activity for continuous esterification of lauric acid with methanol at 1:30 molar ratio by variation of
temperature and flow rate over 4.582 g ISPS/SiO2 .
When the flow rate was increased from 10 to 200 µL/min (WHSV = 1.8-36 · 10−3 h−1 ) the amount
66
of converted lauric acid (yield of methyl laurate) goes down from 62% to 6.5%, as given in Figure
2.16. This is expected, as increased flow rate lowers the space time and therefore the contact time of
the feed with the catalyst, effectively lowering the yield. The sample at 40 µL/min was taken at 72℃,
however, the yield of methyl laurate seem to fit well in the row of samples with different flow rates
taken at 60℃. When using 45 and 30℃ at 20 µL/min for the esterification of lauric acid it is evident
that the conversion of lauric acid drops to 27% and 12%, respectively, compared to 47% conversion
at 60℃.
The open-cell ISPS/Al2 O3 of 9,5 mm × 20 mm (batch 7 #) were tested in a mixture with the
molar ratios 8:1:1:0.2 of respectively methanol, lauric acid, trioctanoin, and dodecane as an internal
standard (similar mixture to that mentioned in section 2.4). Unfortunately, stable operation was only
achieved at 3 different conditions (of 2 samples each) because the pump flow was unstable during
part of the experimentation. The yields obtained during stable operation are plotted in Table 2.4.
Table 2.4: Esterification yields in 1:8 molar ratio of lauric acid to methanol over 4.354 g open-cell ISPS/Al2 O3
ceramic cylinders.
Temperature Flow rate Yield methyl laurate Lauric acid remaining

25℃ 0.01 ml/min 0.022 0.978
60℃ 0.03 ml/min 0.140 0.860
60℃ 0.02 ml/min 0.207 0.793
At 25℃ and 10 µL/min liquid flow rate (WHSV = 1.9 · 10−3 h−1 ) a mere 2% conversion was
observed of the lauric acid. At 20 and 30 µL/min of liquid flow rate (WHSV = 3.8 and 5.8 · 10−3
h−1 ) and 60℃, the yields were respectively 21% and 14% of methyl laurate. Shorter residence times
(higher space velocities) led to lower conversion, while lower temperatures (room temperature) also
led to lower conversion.
When comparing the ISPS/SiO2 -pellets with the open-cell ISPS/Al2 O3 , the yield of methyl lau-
rate over the latter at 20 µL/min is 44% lower than over the former. This could be due to the differ-
ence in the loading of active phase of the ceramics and the intrinsic properties of the two different
catalysts. It may be suspected that higher intrinsic surface area of the SiO2 -pellets allowed a higher
deposition of active material and a higher degree of exposure to the alcohol-fatty acid solution. It
is also evident that the former catalysts had a much higher molar ratio of methanol to lauric acid
in the feed solution. From Table 2.2 it is evident that the SiO2 -pellets contained more than 3 times
active SPS phase compared to open-cell ISPS/Al2 O3 on a mass basis, and it can be concluded that the
active phase in open-cell ISPS/Al2 O3 was more active in the esterification than the ISPS/SiO2 when
considering the differences in active phase loading.
The open-cell ISPS/Al2 O3 probably has very advantageous flow properties for viscous oil emul-
sions which can only penetrate small pores slowly. Also, it could be visibly observed that some of
the SPS impregnated on the open-cell Al2 O3 was situated on the exterior surface of Al2 O3 as small
lumps and was thus exposed directly to the liquid flow, at least immediately after the sulphonation.
This is an advantage in terms of the low pore diffusivities of larger fat molecules. The viscosity of
dilute lauric acid in methanol is probably a few orders of magnitude lower than that of typical fats
and oils, thus the hydrodynamic properties of the feed have not been problematic during these ex-
periments. ISPS/SiO2 -pellets, despite having a reasonable bed porosity upon reactor loading and
67
therefore minimal pressure drop during reactant pumping, owe their high loading of active phase to
their micro- and mesoporosity. However, this may not be an advantage when using bigger molecules
like triglycerides and FFAs.
However, during these experiments it has been clearly demonstrated that continuous esterifica-
tion of the model feed was possible over both ISPS/SiO2 -pellets and ISPS/Al2 O3 -open-cell cylinders.
2.5.5 Continuous esterification of industrial fatty feedstocks

The SPS impregnated on open-cell Al2 O3 was subjected to further investigation in a larger flow-
reactor to evaluate the possibility of using the catalyst for esterification or transesterification on real
feeds and in a larger scale. 25 ISPSAl2 O3 -cylinders based on open-cell ceramic cylinders (batch # 8)
were stacked uniaxially after each other the reactor. The cylinders were 2.0 cm in diameter and 2.0
cm in length with an average cell diameter of 3 mm of the connected cavities in the open-cell Al2 O3 .
Cylindric static mixers were placed as spacers between each stack of 5 catalyst cylinders, effectively
dividing the bed in 5 equal-sized beds of 5 ISPS/Al2 O3 cylinders. The reactor was operated without
the continuous addition of reactants and extraction of products, i.e. as a batch reactor. A sketch of
the packed-bed reactor is given in Figure 2.17.
DĞƚŚĂŶŽů ZĞĂĐƚŽƌ
KŝůͬĨĂƚ WƌŽĚƵĐƚ
ZĞĐǇĐůĞ ůŽŽƉ
Figure 2.17: A schematic of the tubular packed-bed loop reactor. The red box is a mixer (emulsifier), the green
short cylinders constitute the static mixers between the black cylindrical catalysts of ISPS supported on open-cell-
Al2 O3 . The addition of reactants and the product stream (dotted lines) was shut off for an “infinite recycle ratio”,
thus converting the reactor into a packed-bed batch reactor.
0.5 kg of methanol and 2.0 kg of rapeseed oil were added as reactants to the continuous loop.
Then the extraction of product and the reactant inlets were closed and the emulsification by the stir-
ring mixer ensured maintenance of the emulsion of methanol and oil. The flow speed is unimportant
in this context, as the average contact time across the bed is independent of the pumping speed in a
closed loop. After 3 days of reaction at 60℃ the yield of FAME was at a marginal level of about 2%,
which must stem from transesterification reaction as the rapeseed oil did not contain any FFA.
As the produced amount of FAME from the rapeseed oil is low it may be speculated that the
formed FAME is simply produced due to the thermal transesterification effect at 60℃. However, the
amount of active catalytic sites is not high compared to the reactant amount. On the other hand, the
open-cell ceramic had excellent flow properties as the pressure drop across the entire catalyst bed
during pumping was minimal.
After the experimentation with rapeseed oil, 0.5 kg methanol and 2.0 kg abattoir waste fat from
Daka Biodiesel was tested according to the same procedure and using the same catalyst. Waste fats
usually contain anywhere between 5% and 30% fats, depending on the season, so it was expected
68
that the acidic catalyst would work efficiently in the esterification of the FFA to FAME. However,
after 22 h of reaction running no traces of FAME were observed during analysis.
Abattoir waste fats contain some amounts of alkali salts and these react with the Brønsted-acidic
catalyst and ion-exchanges it to the polar methanol phase. Also the waste fat was likely sufficiently
buffered by the minor presence of various salts (sulphates, phosphates, carbonates) to absorb the
proton to the liquid. Therefore it is easily suspected that the activity of the catalyst has been lost due
to the ion-exchange. This explains the absence of any FAME in the analysis of the organic products.
The ion-exchange was confirmed upon titration of some of the spent cylinders, which contained 0.00
mmol H+ /g compared to 0.174 mmol H+ /g of the fresh ISPS/Al2 O3 (batch # 6) and 2.94 mmol H+ /g
of the fresh SPS. Thus, it can be concluded that the catalyst deactivated by ion-exchange with alkali
ions in the abattoir waste fat.
69
2.6. Basic catalysts for transesterification
2.6 Basic catalysts for transesterification
A number of bases of different types have been screened as candidates for transesterification catalysts
that may be immobilised on a solid support for FAME synthesis. Despite the fact that sulphonic acid-
functionalised ionic liquids may be used for the transesterification as described in section 2.3, strong
bases usually work several orders of magnitude faster than strong acids. The role of the base in the
transesterification with methanol is to facilitate the formation of methoxide ions as the active species,
as transesterification takes place by attack of the alkoxide on the carbonyl C-atom [85]. For instance,
if the NaOH or KOH are added, they will form water and methoxide ion as showed in Figure 2.4 a),
however, the water formed is undesired as it can lead to basic hydrolysis of the esters. More generally
any base added will react with the methanol to form methoxide ion and correponding acid according
to the equilibrium in Figure 2.18 b).
a) OH- + CH3OH H 2O + CH3O-

hydroxide methanol water methoxide
b) [Base] + CH3OH [Base]H+ + CH3O-

strong base methanol weak acid methoxide
Figure 2.18: a) formation of methoxide ions and water by methanol and hydroxide during the industrial produc-
tion of FAME aided by KOH or NaOH, and b) the general role of any strong base added to the methanol-containing
solution to catalyse transesterification.
2.6.1 Amine-functionalised ionic liquids

Ionic liquids have advantageous phase properties for ester reactions, and the possibility of fine-
tuning hydrophilic and lipophilic properties can enhance the separation properties after reaction.
Thus, three amine-functionalised ionic liquids were therefore tested for transesterification, namely
the bromide ([Br – ]), boron flourate ([BF4– ]), and bis(triflamide) ([NTf2– ]) of amine-functionalised
butylmethylimidazolium ([BMIM−NH+2 ]). It was suspected that the amine functionality or the anion
of the ionic liquid could add basicity and take up protons to form methoxide from methanol, thus
facilitating transesterification. The yield of methyl octanoate and remaining trioctanoate observed
from the time series with these three ionic liquids are shown in Figure 2.19.
It is clear that only around 0.1% yield of methyl octanoate was obtained, possibly due to an
impurity in the trioctanoate or formed in the glass vials after sampling, as the amount of methyl
octanoate was not increasing with time. The amount of trioctanoate is generally around 100%, but
vary somewhat between some 90% and 125%. It is therefore evident that some inaccuracy during
the sampling, workup and analysis of the samples contributed to the variation. The inaccuracy
contributed more than what would usually be expected from GC analysis.
To conclude, the ionic liquids are completely inactive. Furthermore no octanoic acid is observed
in any of the samples, thus excluding loss of the reactant or product as FFA. This proved that despite
the suitable phase properties of ionic liquids a strong base (or strongly acidic ionic liquid, see section
2.3) is necessary to facilitate the transesterification reaction.
70
5% [BMIM-NH2+][Br-]
125%
Remaining trioctanoate
[BMIM-NH2+][BF4-]
Yield of Me.oct. (mol%)
4% [BMIM-NH2+][NTf2-]
[BMIM-NH2+][Br-] 100%
(mol%)
[BMIM-NH2+][BF4-]
3% 75%
[BMIM-NH2+][NTf2-]
2% 50%
1% 25%
0% 0%
1h 3h 24 h 96 h 1h 3h 24 h 96 h
reaction time Reaction time
Figure 2.19: Yield of methyl octanoate (left-hand side) and remaining trioctanoate (right-hand side) with amine-
functionalised ionic liquids as catalysing medium. The amine-functional cation was [BMIM−NH+ 2 ] using either
bromide ([Br−], ), borofluorate ([BF4 −], ), or bis(triflamide) ([NTf2 −], ) as cation. Conditions: 60℃, 5 mmol
trioctanoate, 45 mmol methanol, dodecane as internal standard.
2.6.2 Basic inorganic salts

Four organic and inorganic salts have been tested as catalysts for transesterification of the model mix-
ture of methanol and trioctanote. Guanidinium chloride (GuaCl), guanidinium carbonate (Gua2 CO3 ),
and cesium carbonate (Cs2 CO3 ) represented compounds that are all soluble in water and small alco-
hols. They were compared with the mixed oxide of MgO · La2 O3 , which was expected to be insoluble.
The role of guanidinium carbonate in the transesterification have previously been debated. It has
been claimed that carbonate vaporised from the solution as CO2 and that the remaining guanidine
acted as a strong base [319]. However, the carbonate could also be suspected to act as the base ac-
cording to Figure 2.18 when the salt dissolves into the carbonate and guanidinium ions in methanol.
It is at least not clarified by Peter & Weidner why the acidic CO2 vaporised from a basic solution of
methanol [319].
The results of the experimentation are given in Figure 2.20. It can be observed that Gua2 CO3 and
Cs2 CO3 showed formation of methyl octanoate at comparable rates, albeit the inorganic Cs-salt at a
slightly higher rate. Over 90% yield is established after 3 h of reaction. The guanidinium chloride
was completely inactive, being weakly acidic.
100% 120%
Remaining trioct. (mol%)
GuaCl
Yield of Me.oct. (mol%)
Gua2CO3 100%
80%
Cs2CO3
MgO · La2O3 80%
60%
60%
40%
40%
20%
20%
0% 0%
0.25 h 1h 3h 24 h 96 h 0.25 h 1h 3h 24 h 96 h
Reaction time Reaction time
Figure 2.20: Yield of methyl octanoate (left-hand side) and remaining trioctanoate (right-hand side) with vari-
ous salts used as catalysing medium. Legend: Guanidinium chloride (GuaCl, , 4 mmol), guanidinium car-
bonate (Gua2 CO3 , , 1.8 mmol), cesium carbonate (Cs2 CO3 , , 1.8 mmol), Magnesium lanthanum mixed oxide
(MgO · La2 O3 , , 0.22 mmol). Conditions: 60℃, 5 mmol trioctanoate, 45 mmol methanol, dodecane as internal
standard.
71
This clearly demonstrated that the guanidinium cation facilitated the carbonate dissolution in the
methanol. Guanidine is a strong base (pKa = 13.6), thus the corresponding acid, the guanidinium ion,
is weak and was most likely stable in the methanol solution. Therefore the dissolved carbonate ion
(CO32 – ) acted as the base forming methoxide ions according to Figure 2.18. However, the rates of
reaction measured over CO32 – -base were not as high as those reported (ref. [319]) and monitoring of
the CO2 -pressure over the solutions have not been performed in the present work, so any dependency
of the pressure on the turnover rate of methanolysis cannot be established [319].
The MgO · La2 O3 proved to be almost inactive compared with the carbonate, however, only 0.22
mmol MgO · La2 O3 was added compared to 1.8 mmol of the carbonates, and a mere 7% yield of
methyl octanoate is observed after 96 h.
Again it was found that the remaining trioctanoates amount were sometimes too low to complete
the mass balance due to inaccuracy of the trioctanoate analysis, albeit the remaining trioctanoate in
most samples of GuaCl and MgO · La2 O3 were between 90% and 100%. No or only minor traces of
octanoic acid were found during the experimentation with these catalysts, so the hydrolysis cannot
account for the missing reactant.
2.6.3 Amidines and guanidines

The posibility of using strong organic basis for the transesterification have been investigated. Organic
molecules often allow for the linkage to a polymeric support or a ceramic with functional groups
through organic synthesis, and this would make anchoring of the base to a given support easier and
possibly prevent leaching of the active basic sites over time.
Transesterifications over amines of varying basicity have previously been investigated. Villa and
co-workers nested various amines to carbon nanotubes and achieved up to 80% conversion of tri-
butyrate with methanol at 60℃ after 8 h. Differences in activity were ascribed to basicity, which
were higher for more substituted amines [79]. Conversions of tributyrate of equal magnitudes were
achieved at 115℃ using various di-amines [80]. Quarternary amine with OH – -counterions linked
to silica or a polymeric resin for transesterification of triacetin with methanol achieved up to 65%
yield in 4 h at 60℃ [85]. This model reaction was also studied over a number of simple amines,
but relatively high temperatures were necessary due to low basicity of the amines used [320]. Other
transesterification reactions have been studied as well. For instance, methyl esters have succesfully
been transtesterified with higher alcohols over MCM-41 functionalised with NH2 -linkers at 110℃
[83].
Initially, cyclic and uncyclic guanidine (TMG respectively TBD) and cyclic amidines (DBU and
DBN) have therefore been tested for transesterification of trioctanoate with methanol at 60℃ and
ambient pressure. The yields are given from Figure 2.21.
It is obvious that the reaction has run to completion almost instantaneously under the applied
conditions. 2 mmol catalyst was added to the solution in the case of DBU, DBN and TMG, compared
to 45 mmol MeOH and 5 mmol trioctanoate in the reaction mixture. The amounts of base used per
mass are ca. half those of the masses of the acidic ionic liquids used in section 2.3, but nonetheless
reaction was complete within minutes, not hours.
72
DBU DBN TMG TBD

100%
mole fraction (%)

80%
60%
40%
20%
0%
¼h 1h 3 h 24 h ¼h 1h 3 h 24 h ¼h 1h 3 h 24 h ¼h 1h 3 h 24 h
Reaction time
Figure 2.21: Yield of methyl octanoate () and by-product octanoic acid () during transesterification experiments
using DBU (1,8-Diazabicyclo[5.4.0]undec-7-ene, 2.01 mmol), DBN (1,5-Diazabicyclo[4.3.0]non-5-ene, 2.06 mmol),
TMG (1,1,3,3-tetramethylguanidine, 2.01 mmol), and TBD (1,5,7-triazabicyclo[4.4.0]dec-5-ene, 0.39 mmol) as cat-
alysts. Conditions in each experiment: 60℃, 5 mmol trioctanoate, 45 mmol methanol, dodecane as internal stan-
dard.
It is clear that over time - and in the case of TBD from the very beginning of the experimentation
- irreversible basic hydrolysis of trioctanoate or methyl octanoate have taken place. The TBD che-
mical may possibly have been impure and contained some water. Amidines and guanidines are very
hygroscopic and should generally be stored with a desiccant. Water may also have been present as
an impurity in the methanol and in the reactant solution (although anhydrous methanol was used).
Interestingly, only one fifth of TBD was added to the reaction solution compared to DBU, DBN
and TMG, however, TBD still led to full conversion of trioctanoate within 15 min (albeit to the fatty
acid and not the ester). It may be speculated that the lower amount of catalyst was the reason for
the high yield of the fatty acid instead of the methyl ester - impurities of water could be bound by
the hygroscopic organic bases, but only in the case that sufficient base was present in the reactant
solution. It has previously been stated that the strongly basic organic amines are hygroscopic and
that loss of activity may take place over time during transesterification [319].
To study the effect of substituting TMG with various linkers, a handful of substituted tetra-
methylguanidines were tested with the same model mixture. The results are given in Figure 2.22.
The substituted guanidines used were propyl-, hexyl-, and benzyl- tetramethylguanidines. 0.35 mol
was used for the propyl-TMG, while 0.26 mol was used for the latter two (compared with 5 mmol
trioctanoate and 45 mmol methanol in each reaction mixture).
From the left-hand side of Figure 2.22 it is observed that the conversion is completed within 1
h and the with saturated substituents (propyl- and hexyl-TMG) within 15 min. The difference in is
not clear, but the benzylic substituent attached to the TMG may contribute to a slight destabilisation
of the resonance structure of the benzyltetramethylguanidinium ion making the Benzyl-TMG less
basic.
However, compared with the experiments with the unsubstituted guanidines and amidines in
Figure 2.21, the hydrolysis was much more prominent, and methyl octanoate was the primary pro-
duct only in the 15 min sample. Hereafter about 80-100% octanoic acid is observed in the samples.
It is suspected that the hygroscopic guanidines were contaminated with large amounts of water (a
prerequisite of hydrolysis), that moisture was absorbed from the air, or that the reaction mixture
73
100%
Benzyl-TMG Propyl-TMG Hexyl-TMG 100%
1,8-(TMG)2 Naphth Merrifield-TMG
mole fraction (%)

mole fraction (%)
80% 80%
60% 60%
40% 40%
20% 20%
0% 0%
¼ h 1 h 3 h 24 h ¼ h 1 h 3 h 24 h ¼ h 1 h 3 h 24 h ¼h 1h 3 h 24 h ¼h 1h 3 h 24 h
Reaction time Reaction time
Figure 2.22: Yield of methyl octanoate () and by-product octanoic acid () during transesterification experiments
over substituted and nested guanidines. left-hand side: 2 mmol each of alkyl-substituted tetramethylguanidines
(TMGs); right-hand side: 1.99 mmol of 1,8-Bis(TMG)-Naphthalene and 0.51 mmol TMG on Merrifield resin. Con-
ditions in each experiment: 60℃, 5 mmol trioctanoate, 45 mmol methanol, dodecane as internal standard.
contained water.
For comparison with the unsubstituted guanidine and amidines ca. 2 mmol was used of the
1,8-bis(TMG)-napthalene “proton sponge” while a TMG-amount corresponding to ca. 0.5 mmol was
available with the TMG-functionalised Merrifield-resin. The results are shown in the right-hand
side of Figure 2.22. Full conversion was achieved with 1,8-bis(TMG)-napthalene within minutes. In
comparison with the substituted guanidines it is interesting, though, that very little hydrolysis of the
octanoate esters took place. The activity of the Merrifield-TMG seem to have stalled somewhat, since
over 30% methyl octanoate was observed at 15 min, but below 60% conversion was observed after 24
h. Contamination or deactivation cannot be excluded, perhaps due to a collapse of the TMG during
or after the linking to the support, by incomplete linkage or by a deactivation mechanism connected
to the support in parallel with that described by for instance by Liu and co-authors [85].
Schuchardt and co-workers studied substituted and polystyrene-linked guanidines, including
cyclic guanidines (R-TBD) and achieved up to 75% conversion during transesterification of trigly-
cerides using 5 mol% of various substituted guanidines nested on PS, but pentamethylguanidine
(PMG, the simplest completely substituted guanidine-analogue) yielded above 90% FAME within
1 h [78]. Using 1 mol% of various substituted guanidines, TBD and Methyl-TBD appeared to be
even more active than PMG [77]. Thus, the non-polymeric guanidines were always more active than
their nested counterparts [78]. However, leaching of the TMG was as well observed, and the nested
guanidines gradually lost activity for each reuse in the transesterification. Also the creation of a
“double-linkage” to TMG from the support will result an inactive alkyl-guanidinium cation [78].
74
2.7 Heterogeneous catalysts for production of FAME: Conclusion
The most widespread method of producing biomass-based diesel is the formation of FAME from fats
and oils and methanol. Esterification of FFAs in fats with methanol is catalysed by acids, while the
consecutive transesterification of the fatty acid glyceride esters are catalysed by bases - even though
very strong acids can be introduced, for instance as sulphonic acid-functionalised ionic liquids. In-
dustrially the reactions are usually performed with mineral acids and bases mixed in the methanol
phase, making the practical reuse of the homogeneous catalyst cumbersome.
Acid-functionalised ionic liquids as catalysts and reaction medium for both esterification and
transesterification may be another promising way to catalyse the production of FAME in a batch-
wise production process. The most acidic sulphonic acid-functionalised ILs investigated converted
the lauric acid with methanol to methyl laurate within an hour and catalysed the transesterification
of trioctanoate with methanol at about 99% conversion in 24 h. Interestingly, however, the mono-
or di-glycerides were not converted as fast as the triglycerides, possibly due to emulgation proper-
ties. The optimisation of catalytic activity, separation of products and reaction medium as well as
the leaching of the ionic liquid need thorough investigations before this method reaches sufficient
technical maturity.
It has been shown that a highly acidic sulphonic acid based on pyrolysed carbohydrates (SPS)
could be applied for the esterification of FFA for the production of FAME. Both starch, cellulose,
glucose, and sucrose have been applied as starting materials for SPS synthesis, however, the latter
was found to be slightly more active than cellulose and glucose for the esterification over SPS. No
specific influence of the pyrolysis temperature on the activity of the final SPS catalysts could be estab-
lished, however, sufficient pyrolysis time was necessary to ensure the efficiency of the sulphonation.
Sulphonation in 15% or 30% fuming sulphuric acid led to the degradation of the material into a very
fine powder and partial dissolution of the material in the wash water. The remaining sulphonated
material, however, was more active than the SPS obtained in concentrated sulphuric acid sulphona-
tion. Tests for sulphate leaching was performed with Ba(NO3 )2 , however no sulphate species were
found in the reaction solution. Thus, it is established that the sulphonated functionalities are stable
at 60℃ reaction temperature.
The solid catalysts being reported to this date are usually based on small pellets or powders,
however, in this work immobilisation of the active phase on an open-cell ceramic support has been
carried out, which is more suitable for continuous tubular reactors due to low pressure drop. The SPS
catalyst has been scaled up by supporting a suspension of SPS in polyerethane polymer, and impreg-
nated on porous or open-cell ceramics by impregnation prior to catalyst synthesis. The supported
SPS catalysts were less active than their unsupported analogues, due to over an order of magnitude
lower amount of active sites per unit mass.
The supported catalysts may be easier to integrate with continuous reactors for production of
FAME, for instance as tubular packed-bed reactors. The supported SPS catalysts was tested for con-
tinuous esterification of model compounds with methanol. The esterification activity was tested in
a tubular backed-bed flow reactor with ISPS/SiO2 -pellets and open-cell-ISPS/Al2 O3 . Both catalysts
75
2.7. Heterogeneous catalysts for production of FAME: Conclusion
were active albeit at low flow-rates of reactant solution. Furthermore, the open-cell-ISPS/Al2 O3
yielded a marginal conversion due to transesterification of an emulsion of rapeseed oil and methanol
at 60℃ after 72 h of reaction in a packed-bed batch reactor loop. However, no esterification or trans-
esterification at all was observed during the treatment of waste abattoir fats with methanol over the
same catalyst during 22 h batch loop reaction. This was due to ion-exchange of Brønsted-acidic sul-
phonic functionalities with alkali ions in the feedstock, which remains an overall challenge when
employing waste feedstocks, especially abattoir waste. The catalyst deactivates on the uptake of the
alkali ions and a regeneration strategy must be developed.
It was proven that easily soluble carbonates catalysed the transesterification, while the guani-
dinium ion from soluble salt, being a weak acid, in itself did not catalyse the transesterification.
Basic amine-functionalised ionic liquids were completely inactive. However, a number of homoge-
neous organic guanidines and amidines were tested and achieved full conversion within 0.25 - 1 h,
thus the reaction rate was at the same order of magnitude as the industrially used hydroxides or
methoxides.
The alkyl- and aryl-substituted and nested guanidines tested had somewhat lower activities for
the transesterification due to a lower loading of the active phase, but also a slightly lower basicity.
Thus, the strong organic bases investigated for transesterification can be linked to porous solids,
for instance ceramics or polymers, to obtain solid basic catalysts. This will make a continuous flow
packed-bed reactor system for production of FAME easier to realise. However, some of the inves-
tigated guanidines were sensitive to water, which was possibly absorbed and resulted hydrolysis of
the esters. This can make the guanidines difficult to apply with waste feedstock, for instance abattoir
wastes, which may contain several percent of water.
76
Chapter 3
Catalytic Hydrodeoxygenation of Fats and Oils 3

3.1 Hydrodeoxygenation of fats and oils: Introduction
As mentioned in section 1.4 three types of catalysts exists for upgrading of oils and fats to hydrocar-
bons. Cracking reactions over porous acidic or basic materials at elevated temperatures provides a
neat catalyst regeneration strategy (burning the coke formed), but the yields of diesel-range products
have not been impressive to date and the cracking often lead to petrol-range hydrocarbons instead.
The supported sulphided catalysts appear attractive from the point of view that they are already used
in refineries for removal of for instance sulphur from the crude, and oxygen may potentially be re-
moved as well with hydrogen. However, the use of supported transition-metal particles represents an
attrative alternative. First and foremost support metallic particles do not contain noxious elements
like sulphur that may potentially end up in the product or should be added continuously. Secondly
the reported selectivities of for instance Pt and Pd towards decarboxylation and decarbonylation to
alkanes are unrivaled [17], thus they may principally lower the consumption of H2 compared to their
sulphided equivalents according to Figure 1.6.
Naturally occuring oils and fats represents quite chemically varied fatty acid profiles, which
changes between plants, animals, and it varies considerably within a species. Normally rough fatty
acid ditributions are known [12, 214]. Simakova et al. showed, however, that the reaction rate was
independent of the fatty acid chain length (from C17 -C22 ) during deoxygenation of dilute dissolved
fatty acids over Pd/Sibunit at 300℃ [23]. This result is important, as fats and oils contain a distribu-
tion of various fatty acids, and as 3 fatty acids are bound to glycerol as a triester, the possible number
of different molecules found in oils and fats is huge. For this reason most authors have worked with
model compounds to ensure that the reactants under study were chemically well-defined.
A number of catalyst metals and supports have been tested by Snåre et al. who found Pd and Pt
to be the most active catalysts metals for deoxygenation, as well as the most selective towards de-
carboxylation and decarbonylation. The gas phase composition was addressed as well and contained
mixtures of CO and CO2 i varying ratios, and it was established that methanation is thermody-
namically favored and thus may be an unwanted side reaction [17] (Figure 1.7). In fact both the
reduction, decarbonylation and decarboxylation (Figure 1.5 and Figure 1.6) are thermodynamically
favored [17], so the properties of the catalyst are clearly important for the product distribution. Do
et al. reported formation primarily of CO over Pt/Al2 O3 during deoxygenation of methyl esters, but
almost no methane was formed during the experimentation even in H2 atmosphere. Immer and co-
authors claimed that solely CO2 was formed from decarboxylation of stearic acid and suspected that
H2 was in fact inhibiting the reaction [321].
Clearly, when the gas atmosphere over the catalysts contain both H2 and formed CO and CO2 ,
77
3.1. Hydrodeoxygenation of fats and oils: Introduction
the methanation reaction and WGS-equilibrium cannot be excluded. To date, however, no reports in
the literature have appeared which directly quantify the extent of methanation and water-gas shift
reactions, although these gas-phase side-reactions are of importance during deoxygenation and their
quantification would help to reveal mechanistic information. Berenblyum and co-workers recently
proposed that the deoxygenation could take place via a formic acid intermediate on the adsorbed sur-
face of palladium [322]. This can be difficult to assess as formic acid would decompose once formed
around 300-350℃, especially in the presence of a Pd catalyst [323], however, the decomposition may
either yield CO2 and H2 via dehydrogenation or CO and H2 O via dehydration, and the degree of
either reaction may as well be influenced by the catalytic properties of the surface. This is illustrated
in Figure 3.1. Thus, the influence of WGS equilibrium on the gas-phase composition and the amount
of CO and CO2 measured may in theory be minimal.
O
a) R HCO2H + CH2 R
HO
b) CH2 R + H2 CH3 R
O
c) H2O + CO CO2 + H2
HO H
Figure 3.1: Decarboxylation via formic acid mechanism proposed by Berenblyum [322]. a) Catalytic formation of
1-alkene and formic acid from fatty acid, b) saturation of the fatty acid by hydrogen, c) decomposition of the formic
acid via dehydration (left-hand side) or dehydrogenation (right-hand side).
Snåre and co-authors investigated the deoxygenation of unsaturated fatty acids and their methyl
esters and confirmed the rapid isomerisation and hydrogen-transfer reactions taking place on the
Pd/C active surface at 300℃ [324]. Rozmysłowicz and co-workers also studied conversion of unsat-
urated fatty acids via saturation/desaturation reactions and deoxygenation and found that extensive
amounts of coke and catalyst deactivation resulted in H2 -sparse conditions, especially with more
concentrated solutions of the unsaturated fatty acids. Likewise, stearic acid was the main product
from unsaturated fatty acids using little or no hydrogen, also confirming the saturation/desaturation
transfer of hydrogen between molecules [20]. Immer et al. confirmed the hydrogenation transfer
between feedstock and solvent and also found that saturation was necessary for obtaining a high
reaction rate [321].
As sketched, the formed products in both gas and liquid phase gives the impression of great com-
plexity in the network of reactions and the yields of products obtained. Usually the investigations
have been performed in batch- or semibatch mode and in diluted systems, but industrially more
concentrated reactant streams and continuous reactors are preferred.
The catalyst studies so far have concentrated on shorter reaction runs and the longest reaction
have been within a few days. It is necessary to demonstrate more extensive reaction times and in-
vestigate catalyst stability and/or deactivation during more the deoxygenation. Also, most authors
have worked with fatty acids and their methyl or ethyl esters, possibly due to the often cumbersome
chemical analysis for the direct quantification of triglycerides. From the point of view of studying
78
Chapter 3. Catalytic Hydrodeoxygenation of Fats and Oils
chemical reactivity and selectivity this may probably be too simplistic, however, as triglycerides are
usually the main component of the fats and oils.
Therefore, in the present work, the conversion over alumina-supported noble-metal catalysts of
model mixtures of tripalmitin and oleic acid resembling waste fats have been studied in batch mode.
The effect of pressure, temperature and active catalyst metal have been investigated. Stearic acid,
ethyl stearate, and tristearin have been deoxygenated over Pd/C catalysts in continuous mode to
relate the reactivities of different molecules. Furthermore deoxygenation of stearic acid over Pd/C
has been performed for 2 weeks time-on-stream for logging step-changes and steady-state behaviour.
79
3.2. Batch hydrodeoxygenation: Experimental
3.2 Batch hydrodeoxygenation: Experimental
3.2.1 Chemicals
The reactants n-tetradecane, oleic acid (>99.0%), n-docosane (>98.0%), tripalmitin (>99.0%), and
N-methyl-N-trimethylsilyl-trifluoroacetamide (MSTFA) (>97.0%) were obtained from Fluka. The
catalyst precursor salts H2 PtCl6 · 6 H2 O (>99.0%) and Pd(NO3 )2 · 2 H2 O (>99.8%) were obtained
from Fluka, while Ni(NO3 )2 · 6 H2 O (>99.0%), HAuCl4 · 3 H2 O (>99.9%), RuCl3 · 2 H2 O (>99.9%),
and Rh(NO3 )3 · H2 O (>99.0%) were purchased from Sigma-Aldrich
The support material γ-Al2 O3 was on stock in the laboratory. The gases H2 (>99.999%), 5%
CO/He (>99.999%), N2 (>99.999%), and Formier gas (10% H2 in N2 , >99.99%) were all purchased
from AGA. All chemicals were used as received.
3.2.2 Catalyst preparation

All applied catalysts were prepared with a metal loading of 5 wt% metal on the support. The sup-
ported metal catalysts were prepared by the incipient-wetness impregnation method described in
literature [325]: The active metal precursor salt was dissolved in demineralised water and added to
dry, fractionated (<180 µm) γ-alumina powder. The impregnated catalyst was dried in an oven at
110℃ in 2 h and afterwards calcined in atmospheric air by ramping with a 100℃/h to 400℃ where
the temperature was kept for 8 h. Any remaining chloride ions on the
3.2.3 Catalyst characterisation

The BET surface areas of the calcined catalysts and the γ-alumina support were determined by N2
physisorption. Before measurement the samples were degassed at 200℃ for 4 h under vacuum.
Adsorption- and desorption isotherm were measured with liquid N2 at 77K on a Micromeritics ASAP
2020 pore analyzer.
CO pulse-chemisorption was performed to determine the active metal dispersion and mean par-
ticle size. This was done on a Micromeritics Autochem II 2920 with a pulse loop size of 0.366 ml.
0.1 g catalyst sample was weighed, flushed with He, heated and reduced in 10% H2 /N2 at 150℃ for
2 h. The sample was then flushed with He and cooled, and CO pulse-chemisorption was performed
at 25℃ with 5 % CO/He in He carrier gas. CO-concentration in the effluent was continuously mea-
sured by a thermal conductivity detector (TCD) against reference carrier gas. For the Ni catalyst, the
reduction was attempted under mere severe conditions at up to 400°C for 12 h, which was needed
to entirely reduce the NiO. The stoichiometries for area calculation were Pt:CO = Ni:CO = 1:1 and
Pd:CO = 2:1, based on literature reports [326]
3.2.4 Hydrodeoxygenation reaction

Hydrodeoxygenation was performed in a 50 ml stainless steel autoclave (MicroClave from Autoclave
Engineers). A reaction mixture of totally ca. 9 g (taking up 11 ml of volume) was added to the auto-
80
clave, consisting of 0.806 g tripalmitin, 0.094 g oleic acid, 0.03 g of n-docosane as internal standard
and about 8.1 g n-tetradecane as solvent. Thus, the mixture consisted of 10 wt% reactant molecules
fat mixture - consisting of 75 mol% tripalmitin (90 mol% of the fatty acid chains present) and 25
mol% oleic acid (10 mol% of the fatty acid chains present). The remaining 90 wt% of the feed consist
of unreactive alkanes. A 0.15 ml reference sample was taken out and 0.20 g catalyst added to the
remaining liquid.
The autoclave was then sealed at room temperature, flushed several times first with pressurised
argon and then hydrogen, the relevant pressure of H2 was added and the autoclave was heated to the
desired reaction temperature and stirred at 900 rpm.
Samples of 0.15 ml were taken out from the reactor after quenching the reactor to 0℃ and de-
pressurising. After sampling, the reactor was tightened and the gas atmosphere filled and purges at
least 3 times first with pressurised Ar, then with pressurised H2 according to desired start pressure.
Then the reactor was re-heated to reaction temperature.
A standard hydrodeoxygenation experiment was performed at 325℃ over 5 wt% Pt/γ-alumina
as a catalyst under 20 bars of H2 , by sampling after 1, 2, 5, and 20 h. The amount of hydrogen in the
reactor after each filling corresponded to about 5 times the necessary amount of hydrogen compared
to complete reduction of the liquid reactants to alkanes.
By sampling in this manner after 1, 2, 5 and 20 h, temperature and hydrogen pressure variation
was performed to investigated dependency on the conversion of the reactants and the preference for
decarboxylation, decarbonylation or full reduction. The same was done to study a number of diffe-
rent transition-metals as active catalyst phase. Control experiments were furthermore performed by
reacting solely oleic acid or tripalmitin, by omitting respectively active metal, hydrogen or catalyst.
3.2.5 Product analysis

0.15 ml of the samples taken out from the liquid-phase was mixed with 20 µL of MSTFA, the vial
capped and silylation was completed at 60℃ for 30 min. After silylation, the samples were analysed
on GC. Silylation converts FFAs to trimethylsilyl esters to make them sufficiently volatile to allow
analysis with reduced peak tailing in gas chromatography [211].
The products were quantified with an Agilent Technologies 6890N split/splitless-injection gas
chromatograph with equipped with flame ionisation detector (GC-FID). Furthermore, qualitative
analysis of certain samples were performed by an Agilent Technologies 6850 split/splitless GC equip-
ped with a mass spectrometer (GC-MS).
A GC with split/splitless injector is not suitable to quantify the amount of tripalmitin in the
samples due to the low volatility of the large triglyceride moleculs. The yield of pentadecane and
hexadecane were therefore used to assess tripalmitin conversion according to Figure 3.2 in section
3.3, as alkanes of this size are measured quantitatively with good precision by split/splitless GC-FID.
This assumption was justified from the observation that n-pentadecane and n-hexadecane were the
only products of tripalmitin conversion.
81
3.3. Batch hydrodeoxygenation: Results and discussion
3.3 Batch hydrodeoxygenation: Results and discussion
A number of parameters were varied to study the hydrodeoxygenation over supported transition-
metal catalysts in batch reactor: The temperature was varied between 250-375℃ over Pt/γ-Al2 O3
while keeping the start pressure of H2 constant at 20 bar. In another series of experiments, the
pressure of H2 was varied between 0 to 40 bar in various experiments over Pt/γ-Al2 O3 at a constant
reaction temperature of 325℃. Ni, Ru, Rh, Pd, Pt, and Au were tested as supported metal catalysts on
γ-Al2 O3 in the autoclave at 325℃ at 20 bar H2 . Furthermore, a number of control experiments were
conducted to elucidate the role of the support material, the reactor, the presence of hydrogen, and
the composition of the reaction mixture. All deoxygenation experiments were conducted by making
time series of the conversions and yields after 1, 2, 5, and 20 h of reaction. The gas atmosphere was
renewed according to the initial reaction conditions after extracting the liquid samples of 1, 2, and 5
h reaction time.
The reactant molecules, oleic acid and tripalmitin, are respectively a C18 fatty acid and a 3 ×
C16 fatty acid glyceryl triester (triglyceride). Thus, the reaction products from both reactants can
be distinguished from each other: Oleic acid is suspected to form heptadecane from saturation and
decarbonylation or decarboxylation and form octadecane from the full reduction of the fatty acid
with hydrogen. Tripalmitin may form pentadecane via decarboxylation or decarbonylation and form
hexadecane from full reduction. This is depicted in Figure 3.2.
O
C17H36 + CO, CO2 C18H38 + H2O
HO
7 7
Oleic Acid
Decaboxylation, O Full
decarbonylation hydrogenation
O O C15H31
C15H32 + CO, CO2 O O C15H31 C16H34 + H2O

O C15H31
Tripalmitin
Figure 3.2: Deoxygenation products from decarbonylation and decarboxylation (left-hand side) and full reduction
(right-hand side). The resulting C3 -product from glycerol is omitted.
3.3.1 Deoxygenation time series over Pt/alumina

The yields and conversions of the deoxygenation of oleic acid and tripalmitin over 5 wt% Pt/γ-Al2 O3
at 325℃ and 20 bar H2 can be seen in Figure 3.3 as a function of reaction time.
From the time series it is evident that oleic acid was converted to alkanes at over twice the rate
than that of tripalmitin. 100% conversion of oleic acid was achieved after 5 h, but over 30% tri-
palmitin remained after 20 h reaction. This may be due to the larger size of the tripalmitin molecule
and a more hindered access to the ester functionalities compared to the carboxylic acid.
It is also evident that despite a moderate pressure of H2 in the gas atmosphere, the amounts of
heptadecane (C17 ) and pentadecane (C15 ), i.e. products obtained from loss of CO2 or CO, were both
82
more than an order of magnitude higher than those of octadecane (C18 ) and hexadecane (C16 ). This
result is in accordance with the observations made previously by Murzin et al. [17, 207, 208].
The saturation of the C-C-double-bond of oleic acid was observed to take place very fast and no
oleic acid was in fact observed after 1 h of the above-mentioned reaction - the remaining acid was
stearic acid. Saturation of the oleic acid double bond occured in all experiments involving hydrogen,
as it proceeded at much lower temperatures and at much faster rates than the deoxygenation of the
acid or ester functionalities. For the sake of clarity, the unconverted acid reactant is considered to be
the sum of stearic and oleic acids (termed “C18 acid”) and conversion is considered to be complete
conversion to alkanes.
1,0 1,0
Normalised concentration
Normalised concentration
0,8 0,8
0,6 0,6
0,4 0,4
0,2 0,2
0,0 0,0
0 5 10 15 20 0 5 10 15 20
Reaction hime (h) Reaction hime (h)
Figure 3.3: Hydrodeoxygenation time series at 325℃ and 20 bar H2 over 0.2 g of 5 wt% Pt/γ-Al2 O3 . Legend:
Tripalmitin (◦) and its products pentadecane ( C15 ) and hexadecane ( C16 ) (left-hand side), and C18 acid (◦) and
its products heptadecane (△ C17 ) and octadecane (N C18 ) (right-hand side).
It has been shown in literature that both CO2 and CO can be formed during deoxygenation,
by respectively decarboxylation and decarbonylation, and both were during the conversion of ethyl
stearate and stearic acid to alkanes [44, 207]. As Pt is likely active for the WGS equilibrium (Figure
1.7, page 25), the direct products from deoxygenation cannot necessarily be observed if the reactions
yield CO or CO2 exclusively since a mixture of CO, CO2 , H2 O, and H2 will always form. In fact it was
found by for instance Resasco et al. that methyl octanoate and methyl stearate decarbonylated over
1 wt% Pt/γ-Al2 O3 to yield CO, C7 - or C17 -alkene and, presumably, methanol, which was however
not observed [210].
Gas-phase analyses were performed as part of the experimentation before extracting the liquid
samples after 1, 2, and 5 h. A gas-valve was carefully opened to fill a gas bag with the gaseous
content of the cooled reactor, and via a septum the gas in the bag was transferred to a gas syringe
and injected into a split/splitless-injection GC equipped with TCD and FID. In all cases, the content
of CH4 constituted about 60% of the carbonaceous gases found and about 40% was a mixture of CO
and CO2 . Hydrogen could not be detected with the present GC setup, however. Pt is known not to
be very active in methanation reactions due to the high dissociation energy of CO [327], but clearly
methanation has taken place. At around 300℃ the methanation is thermodynamically favored [44].
However, also the reactor interior made of stainless stell could have contributed, as Fe is known to
be a good Fischer-Tropsch catalyst [327]. Clearly, the formation of methane is problematic as it is a
20 times more potent greenhouse gas than CO2 [2] and it consumes the hydrogen otherwise needed
83
to reduce the oxygen functionalities or prevent formation of unsaturated compounds.

No C3 -compounds like e.g. propane were found in the gas samples, which would be indicative of
the tripalmitin conversion. It may have been dissolved mostly in the tetradecane solvent, though.
3.3.2 Effect of reaction temperature

To test the dependence of the deoxygenation on temperature, a number of reaction runs were con-
ducted at temperatures between 250 and 375℃. 0.2 g of 5 wt% Pt/γ-Al2 O3 was used as catalyst to
deoxygenate a reaction mixture of 10 wt% fatty feedstock in tetradecane as solvent. The autoclave
was charged with H2 so that the pressure at the reaction temperature was 20 bar. The conversions
after 5 h are plotted for various temperatures in the left-hand side of Figure 3.4.
At 250 and 275℃ the conversions were below 10% after 5 h reaction. However, half of the C18
acid was converted at 300℃ while only 10% Tripalmitin were observed to be converted to alkanes.
At the investigated temperatures above 300℃ the C18 acid was fully converted to alkanes, while the
total yield of C15 and C16 alkanes rose in steps of 40, 55 and finally 100% at increasing temperature.
It is evident that the C18 acid reacted faster than the tripalmitin (except at the lowest temperatures
with very low conversions observed) as was also observed in section 3.3.1.
In the right-hand side of Figure 3.4 the towards full reduction to hexadecane and octadecane.
In all cases the selectivity is below 7% for producing even-carbon-numbered alkanes via reduction
compared to the odd-carbon-numbered alkanes of both C18 acid and tripalmitin over the Pt catalyst.
The preference of the Pt catalyst to split off the carboxylate functionality as CO2 or CO, however,
increased with temperature, making the selectivity for hydrogenation drop ca. 2-3%. It could also
be speculated that thermal decomposition was more pronounced at the higher temperatures.
Conversion, 5 h react. 5 h reaction

1 0,1
Full reduction selectivity
0,8 0,08
C18 acid C18 acid
Tripalmitin
Conversion
Tripalmitin
0,6 0,06
0,4 0,04
0,2 0,02
0 0
250°C 275°C 300°C 325°C 350°C 375°C 275°C 300°C 325°C 350°C 375°C
Temperature Temperature
Figure 3.4: Temperature variation over 5 wt% Pt/γ-Al2 O3 at 20 bar H2 pressure after 5 h. Left-hand side: Conver-
sion of respectively C18 acid () and tripalmitin (). Right-hand side: Ratio of decarboxylation/decarbonylation to
full reduction of respectively C18 acid ( C17 /C18 ) and tripalmitin ( C15 /C16 ).
3.3.3 Effect of hydrogen pressure

The effect of hydrogen pressure variation was investigated to assess if this had an influence on the
activity and selectivity of the Pt catalyst. Thus, hydrogen pressures of 0 to 40 bar were investigated
in a series of reaction performed at 325℃ using tripalmitin and oleic acid as reactants. From Figure
3.5 and Figure 3.7 the yield of C15 -C18 alkanes at increasing pressure can be seen.
84
1h alkane yield, various H2 pressures 1 h reaction

Yield of alkanes (mole fraction)
1 1

C15 C18 acid
0,8 C16 0,8
Tripalmitin
0,6 C17 0,6
C18
0,4 0,4
0,2 0,2
0 0
0 9 20 28 35 40 0 9 20 28 35 40
Pressure (bar) Pressure (bar)
Figure 3.5: Yields of alkanes from C18 acid and tri- Figure 3.6: RSelectivity towards full reduction at
palmitin at different pressures of H2 after 1 h reac- different deoxygenation pressures of H2 over 5
tion over Pt/γ-Al2 O3 at 325℃. Legend: Pentade- wt% Pt/γ-Al2 O3 at 325℃ after 1 h of respectively
cane ( C15 ), hexadecane ( C16 ), heptadecane ( C18 acid ( C17 /C18 ) and tripalmitin ( C15 /C16 ).
C17 ), octadecane ( C18 )
5h alkane yield, various H2 pressures 5 h reaction

1,0 1
C15
0,8 C16 0,8
C17 C18 acid
0,6 0,6
C18 Tripalmitin
0,4 0,4
0,2 0,2
0,0 0
0 9 20 28 35 40 0 9 20 28 35 40
Pressure (bar) Pressure (bar)
Figure 3.7: Yields of alkanes from C18 acid and tri- Figure 3.8: Selectivity towards full reduction at
palmitin at different pressures of H2 after 5 h reac- different deoxygenation pressures of H2 over 5
tion over Pt/γ-Al2 O3 at 325℃. Legend: Pentade- wt% Pt/γ-Al2 O3 at 325℃ after 5 h of respectively
cane ( C15 ), hexadecane ( C16 ), heptadecane ( C18 acid ( C17 /C18 ) and tripalmitin ( C15 /C16 ).
C17 ), octadecane ( C18 )
The highest turnover after 1 h is observed at the pressure of 9 bar H2 , yielding as high as 72%
heptadecane after 1 h reaction, while under 20 bar H2 the yield of heptadecane is only 58%. Inter-
estingly, after 5 h, the yield of heptadecane was higher under 20 bar than under 9 bar H2 . It may be
speculated that the catalyst may have deactivated more with only 9 bar H2 , likely due to coking - per-
haps by insufficient addition of hydrogen to keep the catalyst active. The yield of pentadecane under
9 bar of H2 compared to 20 bar showed comparable yields of about 13%, as well comparable with
the higher pressures of H2 , however, after 5 h about 72% yield is obtained under 9 bar H2 compared
to 44% under 20 bar. This may indicate that the deoxygenation of the ester is inhibited by hydrogen
and proceeds via a pathway were competitive adsorption or by-production can hinder the reaction.
The deoxygenation under 28, 35 and 40 bar hydrogen pressure more or less had the same product
yields, but were not as high as with the lower pressures. After 1 h about 15%, 10% pentadecane
and very small amounts of hexadecane and octadecane had been formed. After 5 h reaction a little
85
above 40% heptadecane and about 35% pentadecane had been formed and the products of complete
reduction of the ester and fatty acid were between 3% and 6%.
When the pressure of H2 is increased, it is found that the selectivity for the full reduction in-
creased after 1 and 5 h, as can be observed from Figure 3.6 and Figure 3.8, respectively. Interestingly,
during the course of the reaction, the selectivity to even-carbon-numbered alkanes increased (i.e.,
when more feedstock was converted). At the highest pressures of 35 and 40 bar H2 the reduction
selectivity almost doubled from 1 to 5 h.
Thus it can be stated that the preference for either complete reduction or decarboxylation and
decarbonylation is to some degree dependent on the pressure og hydrogen, but it was not the most
important feature for the selectivity under the pressures investigated here.
The experiment without added H2 show no alkane products. A small impurity of pentadecane is
visible, but otherwise no product alkanes are found. The catalysts were reduced by the addition of
H2 , stirring and heating before the reaction temperature was reached, and as no H2 was present the
catalyst have most likely not been activated.
3.3.4 Other alumina-supported transition metal catalysts

To assess the effect of other transition metals as catalysts, experiments were conducted with vari-
ous other γ-Al2 O3 -supported transitions metals for the deoxygenation reactions. Thus, 5 wt% of
respectively Pd, Ni, Ru, Rh, and Au were prepared supported on the γ-Al2 O3 by impregnation. The
reactions were conduted at 20 bar H2 and 325℃ with a mixture of tripalmitin and oleic acid in
tetradecane solvent.
As can be seen in from the plots of the alkane yields after 1 and 5 h in Figure 3.9 and Figure 3.11,
respectively, platinum was the most active catalyst metal followed by palladium in terms of C18
acid conversion. However, palladium was much more active than platinum to convert tripalmitin
to alkanes. This peculiar difference between Pd and Pt could further point to more fundamental
differences between the reaction pathways of respectively triglycerides and FFAs, but has not been
addressed in the present study.
From Figure 3.9 and Figure 3.11 the selectivities for full reduction tripalmitin and C18 acid after
1 and 5 h are given. Ru, Pd, and Pt had the lowest selectivities towards hexadecane and octadecane
after 1 h, however, Ru was almost inactive but showed a remarkable selectivity towards heptadecane
- it is possible that the metal is well suited just for the scission of the carboxylic acid functionality
from the free fatty acids. For Pd and Pt the selectivity towards even-carbon-number alkanes increase
slightly with increasing reaction time. Rhodium showed moderate activity, but also higher selectivity
to even-carbon-number alkanes than Pt and Pd. The Rh catalyst, however, mediated the formation
of the same amounts of even-carbon-number alkanes as Pt after both 1 and 5 h.
Ni had a rather high selectivity for the complete reduction of C18 acid, but it also appears to
be deactivated since the yields do not grow very much from 1 to 5 h of reaction (meaning that the
reduction selectivity is high, compared to the other catalysts). The gold catalyst was completely
inactive after 1 h, but minor yields below 10% of C15 , C17 and C18 were observed after 5 h reaction.
Care should be taken when comparing selectivities at very low yields, as impurities can give rise to
86
1h alkane yield over various catalysts 1 h reaction

0,6
1
C15

0,5 C16 C18 acid
0,8
0,4 C17 Tripalmitin
C18 0,6
0,3
0,4
0,2
0,1 0,2
0 0
Pt Ni Pd Ru Rh Au Pt Ni Pd Ru Rh Au
Catalyst active metal Catalyst active metal
Figure 3.9: Yields of alkanes from C18 acid and tri- Figure 3.10: Selectivity towards full reduction over
palmitin over various 5 wt% transition metals on γ- various 5 wt% transition metals on γ-Al2 O3 of re-
Al2 O3 at 325℃ and 20 bar H2 after 1 h reaction. spectively C18 acid ( C17 /C18 ) and tripalmitin (
Legend: Pentadecane ( C15 ), hexadecane ( C16 ), C15 /C16 ). 325℃ and 20 bar H2 after 1 h reaction.
heptadecane ( C17 ), octadecane ( C18 )
5h alkane yield over various catalysts 5 h reaction

1,0 1
C15 C18 acid

0,8 0,8
C16 Tripalmitin
0,6 C17 0,6
C18
0,4 0,4
0,2 0,2
0,0 0
Pt Ni Pd Ru Rh Au Pt Ni Pd Ru Rh Au
Catalyst active metal Catalyst active metal
Figure 3.11: Yields of alkanes from C18 acid and Figure 3.12: Selectivity towards full reduction over
tripalmitin over various 5 wt% transition metals on various 5 wt% transition metals on γ-Al2 O3 of re-
γ-Al2 O3 at 325℃ and 20 bar H2 after 5 h reaction. spectively C18 acid ( C17 /C18 ) and tripalmitin (
Legend: Pentadecane ( C15 ), hexadecane ( C16 ), C15 /C16 ). 325℃ and 20 bar H2 after 5 h reaction.
heptadecane ( C17 ), octadecane ( C18 )
significant errors.
The Ni, Pd, and Pt were subjected to further characterisation study to elucidate if their difference
in activity could be related to their intrinsic physical properties. The microscopic properties of the
catalysts used are shown in Table 3.1. It is evident that the higher the molar loading of the active
metal on the support, the more of the original surface areas of the γ-Al2 O3 of 255 m2 /g is lost (by
plugging and shrinking of pores by the active phase, as indicated by the shrinking pore volumes of
Table 3.1).
The Pt and Pd particles are of about the same size according to the CO pulse chemisorption ex-
periments in Table 3.1, but the lower atomic weight of Pd (and therefore higher molar metal loading
on the support) means that the active surface areas of the Pd is twice as big as that of Pt. Despite this
fact the Pt was still more active that Pd for deoxygenation of C18 acid. The nickel catalyst was much
less active compared to Pd and Pt - the slightly bigger size of the particles may have played a role in
its lower activity for deoxygenation reactions.
87
Table 3.1: 5 wt% Ni, Pd, and Pt on γ-Al2 O3 characterisation
Material mol% metal1 BET-area Pore volume Mean metal particle size Metal surf. area
m2 /g ml/g nm m2 /g
γ-Al2 O3 — 255 0.607 — —
5 wt% Pt/γ-Al2 O3 2.68 251 0.575 5.4 2.82
5 wt% Pd/γ-Al2 O3 4.80 236 0.539 4.6 5.69
5 wt% Ni/γ-Al2 O3 8.38 217 0.523 8.2 4.81
1 Calculated molar percentage of active metal on the γ-Al2 O3 support
3.3.5 Feed variation and diffusion effect

It was suspected that the application of bulky fatty acid and triglyceride molecules may have led to
diffusion limitations in the catalyst interior. Furthermore, some authors have reported that cracking-
type reactions may take place at elevated temperatures with various acidic catalysts [27, 218]. A
closer investigation of the diffusion properties and overall reaction pathways was necessary. There-
fore, control experiments using 5 wt% Pt/γ-Al2 O3 catalyst of 125<X<180 µm in size) at 20 bar H2
and 325℃ was performed and samples taken out after 1, 2, 5, and 20 h, exactly as the time series
experimentation mentioned in section 3.3.1.
A Mason-Boudart-test was conducted to assess if diffusion would be limiting the catalytic reac-
tion. Thus, the 5 wt% Pt/γ-Al2 O3 at the standard size of 125<X<180 µm was crushed down to <50
µm size by crushing in a mortar and sieving. Then a time series with 0.2 g of the finely crushed
catalyst with reaction sampling after 1, 2, 5, and 20 h were conducted. Within a few percent, this
produced the same yields of alkanes as those obtained with the larger catalyst particles.
It was then investigated if the C18 acid or tripalmitin underwent C-C-scission other places than
at the carboxylic acid/ester functionalities. This is in fact a prerequisite for using this type of model
feed to elucidate reactivities of each compound. First, a reaction only employing oleic acid as reactant
was conducted. Heptadecane was observed as the major product and octadecane as the minor one,
while no pentadecane nor hexadecane were observed as products in any of the extracted samples.
It is thus confirmed that no C16 or C15 products were formed from the C18 acid via cracking-type
reactions, just as in the time series experiment of section 3.3.1. Secondly, a deoxygenation experiment
employing solely tripalmitin as a reactant was conducted. Here again it was observed that prima-
rily pentadecane and minor amounts of hexadecane were seen in agreement with the observations
reported in section 3.3.1.
Interestingly, a few percent of palmitic acid were observed after 1, 2 and 5 h of reaction. This
indicated that the tripalmitin may have reacted at least partly via a mechanism involving hydrolysis
or partial hydrogenation to split off a fatty acid, which then further reacted in the same manner as
C18 acid. This is supported by the selectivities of full reduction for Pd and Pt, which are almost
identical for C18 acid and tripalmitin after 5 h, as seen in 3.12. However, the pressure variation
experiments mentioned in section 3.3.3 indicated that another reaction pathway may have existed for
the tripalmitin deoxygenation which was inhibited by too high pressures of H2 . Further investigation
of this phenomenon would be highly interesting, both in terms of finding the optimum reaction
conditions and for product selectivity.
A reference experiment solely with the solvent n-tetradecane and no oleic acid or tripalmitin at
88
325℃ and 20 bar H2 over 5 wt% Pt/γ-Al2 O3 was conducted. After 20 h about 1% of the tetradecane
had been converted to branched C14-alkane isomers (only methyltridecanes were observed). Thus it
is evident that minor isomerisation can take place, but this has not been observed with the reaction
products.
3.3.6 Control experiments and gas variation

One central step in the batch experimentation was to assess the influence of added H2 in the gas and
establish the necessity of the use of a catalyst- i.e. to exclude the influence of the reactor interior,
thermal reactions etc. Therefore experiments were made with this purpose at 325℃ and 20 bar gas
atmosphere (H2 or N2 ) either over no catalyst, over pure γ-Al2 O3 , or 5 wt% Pt/γ-Al2 O3 , and the
obtained results are summarised in Table 3.2.
Without the aid of added catalyst in 20 bar H2 at 325℃ atmosphere, the yields of alkanes were
only 3% from C18 acid and 4% from tripalmitin after 5 h. Thus the reactor interior or the thermal
degradation had minor influence on the deoxygenation activity. Then the support material γ-Al2 O3
was employed as catalytic material with either H2 or N2 in the gas atmosphere. The deoxygenation
yielded 7% and 6% under H2 from respectively C18 acid and tripalmitin, while only 2% and 5% were
found when using N2 in the gas atmosphere. It can therefore be stated that also the γ-Al2 O3 could
not function as a catalyst by itself, as it did not have a notable catalytic effect alone.
Finally it was investigated whether or not H2 would be necessary for the deoxygenation over the
Pt catalyst by performing the reaction in 20 bar N2 atmosphere at 325℃. The conversion of C18 acid
reached as high as 99% after 5 h reaction, however, only unsaturated compounds of C17 size were
observed under these reaction conditions. For comparison, 100% conversion to saturated products
were obtained in H2 -containing gas. It is however important to acknowledge that the conversion of
tripalmitin under N2 was only about 4%. This confirmed that triglycerides must have H2 present in
the feed to react.
Table 3.2: Conversion of C18 acid with supported catalyst, with support and without any catalyst in control exper-
iments. Conditions: 20 bar gas, 325°C, 5 h reaction, 0.2 g catalyst
Catalyst Gas atmosphere Oleic acid conversion (%) Tripalmitin conversion (%)
5 wt% Pt/γ-Al2 O3 H2 100 46
5 wt% Pt/γ-Al2 O3 N2 99 4
γ-Al2 O3 H2 7 6
γ-Al2 O3 N2 2 5
-none- H2 3 4
89
3.4. Continuous hydrodeoxygenation: Experimental
3.4 Continuous hydrodeoxygenation: Experimental
In recent years a number of reports have appeared on the noble-metal catalysed deoxygenation of
fatty acids, fatty esters or triglycerides. Most of these have focused on batch or semi-batch reactors,
i.e. autoclaves, where the semibatch-type system allow for renewal of the gas atmosphere while
maintaining the liquid phase in the reactor.
The continuous systems are more relevant in industrial chemistry on the scale of fuel produc-
tion. Therefore continuous deoxygenation experiments have been performed, primarily over 2 wt%
palladium supported on beads of a synthetic disordered mesoporous carbon type “Sibunit”.
Sibunit is a carbon matrix produced by reacting carbon black in a flow of light hydrocarbon
gases around 1000℃. This causes deposition of graphite-like carbonaceous material from the gases
as a binding material between the carbon black particles, so-called pyrolytic carbon. The resulting
material is then reacted and activated in a flow of hot steam, leaving mesopores in the resulting
material mostly by removal of the original carbon black in gaseous form [328]. The resulting porous
carbon material is then shaped to beads. The mesoporosity of the Sibunit beads makes them suitable
as catalyst supports, especially for larger reactant molecules like fatty acids and derivatives.
In the few studies that have appeared in literature on continuous deoxygenation over supported
noble-metal catalysts, the long-term reaction performance has not been investigated, although this
is a parameter of utmost interest from an industrial perspective, along with the deactivation mecha-
nisms and the operation limits for avoiding or minimising deactivation. This is therefore investigated
with deoxygenation of stearic acid in a ca. 300 h time-on-stream (TOS) experiment over 2 wt% Pd
on Sibunit.
To study the influence of the type of functional group in the feedstock, experiments have also
been made to compare the deoxygenation behaviour of stearic acid, ethyl stearate and tristearin over
Pd/Sibunit, with and without hydrogen present in the gas atmosphere.
3.4.1 Chemicals and materials

Sibunit beads of ca. 1.6 mm in diameter were obtained from Boreskov Institute of Catalysis, Novosi-
birk, Russia. PdCl2 (>99.9%), Stearic acid (>99%), myristic acid (>99%) and eicosane (>99%) were
supplied by Sigma-Aldrich. The silylation agent N,O-bis(trimethylsilyl)-trifluoroacetamide (BSTFA,
>99% purity) was delivered from Acros Organics, while n-dodecane (>99%) and pyridine (>99%)
were obtained from Fluka. Ar (>99.99%) and 5% H2 /Ar gases were delivered from AGA. All chemi-
cals were used as received.
3.4.2 Synthesis of Pd/C

The 2 wt% palladium on Sibunit (Pd/C) catalyst was prepared by the following method [329]:
Sibunit beads of about 1.6 mm in diameter were treated with 5 wt% HNO3 overnight at 25℃
and dried for several hours at 80℃. An aqueous solution of H2 PdCl4 was regulated to pH = 9 with
Na2 CO3 , and the dried Sibunit beads were added to this solution for deposition of the Pd(II)-hydro-
90
complexes. The deposition was continued for 6 h at 25℃; then, the beads were filtered and washed
with H2 O until no chloride ions could be detected in the filtrate. Then the catalyst was dried at 80℃
and calcined at 200℃ for 2 h.
3.4.3 Reactor configuration, loading and sampling

On the inlet side of the reactor the setup consisted of a specially designed heated liquid piston pump,
heated feed vessel and solvent vessel, as well as gas flow controllers and gas lines for Ar and H2 . The
liquid vessels were continuously flushed by bubbling N2 at 20 ml/min through the liquid and out of
the system to evacuate dissolved oxygen in the feed. On the outlet side of the reactor the setup was
equipped with heated lines, liquid sampling valve and collector for the remaining liquid effluent,
as well as gas flow controllers for regulating pressure and blending the effluent with carrier gas
before online analysis of CO and CO2 . The temperature, pressure and CO/CO2 could be logged via a
computer. During experimentation, the tubes, fittings, pump and vessel exposed to liquid fatty feed
were heated to 100℃ as the saturated fatty compounds used as reactants had melting points between
45 and 90℃. Also, stearic acid in dodecane was not sufficiently soluble at ambient temperature. The
configuration of the continuous setup is shown in Figure 3.14. The reactor itself was a tube of 18 cm
height and 1.58 cm inner diameter for a total volume of 35.1 ml. Before each run, the reactor was
loaded in the bottom in three steps:
• In the bottom a layer of quartz wool, then 6 ml of quartz sand 0.2-0.8 mm i diamter, and then
a layer of quartz wool;
• In the middle a catalyst bed of 10 g Sibunit or 10 g of 2 wt% Pd supported on Sibunit
• In the top a layer of quartz wool, then 4 ml of quartz sand 0.2-0.8 mm i diamter, and then a
layer of quartz wool.
Figure 3.13 shows the loading of the reactor. A thermocouple was inserted through the reactor
outlet to measure the temperature from the interior of the reactor.
During experimentation, liquid samples were taken out from the outlet of the reactor by two
valves. 2 minutes prior to sampling, the content of the sampling nozzle was purged. Samples were
collected in glas vials and weighed, followed by addition of external standard. This was diluted in 4
ml of dodecane and heated in an oven at 70℃ until the dodecane had dissolved the sample. Then, 10
µL of the dodecane-diluted sanple was added to a GC vial, along with 100 µL pyridine, 100 µL BSTFA
(silylation reagent) and 40 µL 0.013 M myristic acid in dodecane (silylation control compound). The
vial was capped and allowed to silylate in an oven at 70℃ for at least 30 min. before being analysed
by a gas chromatograph (GC).
The samples from the long-term experiments with stearic acid, the comparison experiments with
ethyl stearate and stearic acid and the step change experiments were analysed on a 5890N HP Agilent
split/splitless-injection GC with a flame-ionisation detector (FID). The gas-phase was analysed with
a Siemens Ultramat 6 IR-analyser, which was set up to measure the CO and CO2 content. The gas
was diluted before measurement on the Ultramat.
91
Liquid
inlet CO / CO2
quartz wool Analyser
Gas inlet
quartz sand
Reactor tube
quartz wool PID
Ar
FC
bed –
TC
catalyst Pump PC PID
or inert
Gas effluent
Collector
Sampling
TC: Temperature control
FC: Flow control
PC: Pressure control
quartz wool
PID: PID regulator
quartz sand FC FC FC
heated line
quartz wool Liquid
Wash Reactant effluent
H2 Ar N2
Figure 3.13: The packing Figure 3.14: Schematic representation of the continuous flow reactor
of the flow reactor before used in the experimentation, produced on the basis of Bernas et al.
each experiment. [330].
After reaction, the Sibunit beads (with or without Pd) were taken out of the reactor sequentially
and sorted into batches depending on their position in the catalyst bed.
3.4.4 Long-term performance of catalyst

The performance of deoxygenation has been evaluated over the same catalyst in both pure Ar and
5% H2 /Ar, and in both dilute and concentrated stearic acid liquid feed. During the entire course of
the experimentation, a liquid flow of 0.075 ml/min and a gas flow of 42 Nml/min was employed,
and the deoxygenation conditions were 300℃ and 20 bar. Liquid samples have been analysed by GC
and CO/CO2 in the gas phase have been measured by online NDIR-analysis.
Stearic acid, pure and in solution with dodecane, was used as a model reactant for the long-
term experiments at a liquid flow rate of 0.075 ml/min. The reactor was loaded with 10 g 2 wt%
Pd/Sibunit beads of 1.4-1.8 mm diameter, according to the procedure in Section 3.4.3. Prior to
reaction the catalyst was reduced for 0.5 h in 1 bar H2 at ambient temperature, then ramped at
10℃/min to 150℃, and then kept at this temperature for 1 h. Afterwards the reactor was ramped
at 10℃/min to 300℃ and the reaction started.
The entire experimentation programme is seen in Table 3.3. First, experiments with dilute stearic
acid in dodecane were performed for the first 3016 min TOS, first 300 min in 42 Nml/min Ar, then
with 42 Nml/min 5 wt% H2 /Ar at 20 bar until TOS = 1575 min, then Ar gas until the flow was
switched to dodecane at TOS = 3016 min. The reactor was quenched to 150℃ at TOS = 3300 min (in
15 min), while dodecane wash was continued until TOS = 4356 min. After the experimentation with
dilute stearic acid the reactor was quenched to 150℃, washed first with dodecane and then flushed
in pure Ar at 20 bar. The reactor was reheated to 300℃ for the next experimentation with pure
stearic acid. First, the deoxygenation was continued without H2 in the feed until TOS = 7056 min,
where the reactor was quenched to 150℃ and washed with dodecane and Ar overnight. Following
this the reactor was reheated to 300℃ and pure stearic acid was reacted with 5 wt% H2 /Ar until a
92
Table 3.3: Overview of the step changes applied during long-term deoxygenation.
Step change # TOS (min) Condition Changes applied
Diluted stearic acid in 5% H2 /Ar

-55 liquid Flow of dodecane started at 300℃
Step I) 0 liquid 10 mol% stearic acid in dodecane at 300℃
Step II) 300 Gas 5% H2 /Ar
Diluted stearic acid in Ar

Step III) 1575 Gas Pure Ar.
Step IV) 1816 Liquid 6.55 mol% stearic acid in dodecane
Step V) 3016 Liquid Dodecane
3300 Temperature Quenching to 150℃ in 15 min
4359 Temperature Cooling to 25℃ in 15 min
100% stearic acid in Ar experiment

4359 Temperature Heating startup from 25℃ to 300℃
4389 Reactor heated
Step VI) 4481 Liquid 100% stearic acid
Step VII) 7056 Liquid Dodecane
7056 Temperature Quenching from 300℃ to 150℃
100% stearic acid in 5% H2 /Ar

7056 Temperature Heating from 150℃ to 300℃ at 15℃ /min
7056 Gas 5% H2 /Ar
Step VIII) 7066 Liquid 100% stearic acid
Step IX) 18506 Liquid Dodecane
20021 Temperature Cooling reactor to 25℃
complete TOS of 19000 min (13.2 days).
3.4.5 Comparison of fatty acid functional groups

Stearic acid, ethyl stearate and tristearin were used as model compounds for these 3 experiments.
Prior to each experiment, a total of 10 g of 2 wt% Pd/Sibunit beads of 1.4-1.8 mm diameter were
loaded by the procedure mentioned in Section 3.4.3; however, a little quartz wool was used as spacers
to divide the catalyst bed into 5 equally large zones for easier separation afterwards. The reactant
was added to the heated reactant vessel. The catalyst was reduced for in 1 bar of H2 (i.e. 20 bar
of 5 % H2 /Ar at 42 Nml/min) for 0.5 h at ambient temperature, heated 10℃/min to 150℃ which
was kept for 1 h. Afterwards the reactor was heated 10℃/min to the reaction temperature at 300℃
where the reaction was started by switching on the feed pumping.
The gas atmosphere used in all cases was 20 bar 5% H2 /Ar at 42 Nml/min, however after 75
h (stearic acid: 96 h) TOS the gas flows were switched to pure Ar and the reaction continued for
another 75 h.
3.4.6 Catalyst characterisation

The distributions of particle diameter and the mean diameters of the catalysts used were calculated
based on the frequency of particles counted from images obtained by operating with acceleration
voltage of 120 kV on a LEO 912 OMEGA energy-filtered transmission electron microscope (TEM).
93
More than 100 particles were counted to obtain the average particle diameter of Pd. Spent and fresh
catalyst beads were crushed to a powder which was carefully mixed. Then each powder was glued
to an epoxy plate before obtaining the TEM micrographs.
The X-ray Diffraction (XRD) mean particle diameter based on volume of the particles were calcu-
lated with the Scherrer formula from powder diffractograms obtained with a Siemens D5000 X-ray
Powder Diffractometer.
Specific surface area and pore size distributions were obtained for all the catalysts tested. Fresh
and spent catalyst batches from the long-term experimentation with stearic acid deoxygenation were
analysed with a physisorption / chemisorption Sorptometer 1900 from Carlo Erba instruments with
liquid N2 at 77 K. The fresh and spent catalysts from the deoxygenation experiments with differ-
ent feedstocks were analysed with a Micromeritics ASAP 2020 Physisportion apparatus also using
liquid N2 at 77K. In both cases, specific surface areas were calculated using the Brunauer-Emmett-
Teller (BET) equation from the N2 adsorption-desorption isotherms. The pore size distributions were
obtained from the Dollimore-Heal correlation.
The distribution of active metal in fresh catalyst beads was investigated by laser ablation con-
nected to an inductively-coupled plasma mass-spectrometer (LA-ICP-MS), with laser ablation sys-
tem of New Wave UP-213 and Perkin-Elmer ICP-MS Sciex Elan 6100 DRC Plus. A number of catalyst
beads were cut in half and fixed in epoxy glue for the measurements.
Reactor effluent contents of Pd were measured by ICP-EOS with a Perkin-Elmer 5300 DV optical
emission spectrometer. 0.2 g of effluent was added to 5 ml of concentrated (65 wt%) HNO3 (aq) and 1
ml of 30 wt% H2 O2 (aq), heated in a microwave oven and then diluted to 100 ml (aq) before ICP-OES
analysis.
94
3.5 Continuous hydrodeoxygenation: Long-term test & step changes
3.5.1 Long-term deoxygenation of stearic acid
3.5.1.1 Dilute stearic acid in 5% hydrogen in Ar - step changes II)-III)

Steady-state decarboxylation of 10 mol% stearic acid diluted in dodecane was performed at 300℃
and 20 bar of 5% H2 /Ar, taking place at 450-1575 min TOS. The results from the online CO/CO2
analysis and the liquid samples taken after the reactor are shown in Figure 3.15.
The conversion of stearic acid to heptadecane was between 91 to 99% at 100% selectivity, how-
ever, in most samples the stearic acid molar fractions were about 2%. At TOS = 300-500 min about
1% monoaromatic C17 -products were observed. This was however a residual from the conversion in
pure Ar at TOS = 0-300 min, and after TOS = 500 min no unsaturated compounds were observed in
the liquid samples.
The CO2 and CO concentrations were also stable corresponding to about 40% and 20% conver-
sion of stearic acid compared to the feed rate, respectively. Thus, about 35-40% of the carbonaceous
gases were missing, most likely due to the thermodynamically favored methanation that consumed
hydrogen to form methane and water (see Figure 1.7 on page 25). However, it was not possible to
measure the methane by online analysis due to the present configuration of the Ultramat NDIR-
analyser, so this assumption could not be directly confirmed.
1.5
1,5
I) II) III) IV) V)
1.25
1,25
Gas: Gas: Gas
Ar 5% H2/Ar Ar
molar fraction
1.0
1
0.75
0,75
0.5
0,5
0.25
0,25
0.0
0
0 500 1000 1500 2000 2500 3000 3500
TOS (min)
Figure 3.15: Continuous deoxygenation of dilute stearic acid over 2 wt% Pd/Sibunit. Legend: Liquid molar sum
( ), Stearic acid ( • ), heptadecane ( N ), heptadecene ( ), undecylbenzene ( × ), CO (x), CO2 (•).
3.5.1.2 Dilute stearic acid in Ar - step changes IV)-V)

To investigate the effect of pure Ar on the deoxygenation during stable operation in dilute liquid
feed, the sweeping gas was switched to Ar at TOS = 1575 min.
However an unintended blocking of the Ar gas feed line had to be removed first, which gave rise
to the increased gas- and liquid concentrations seen after step change III) at TOS = 1575-1800 min in
Figure 3.15. At TOS = 1800 the liquid samples reached a stable level. The liquid feed was renewed
95
3.5. Continuous hydrodeoxygenation: Long-term test & step changes
with a dilute solution of 6.55 mol% stearic acid in dodecane - the molar fractions in Figure 3.15 are
calculated on basis of this number at TOS = 1816-4359 min.
Initially, at step change IV) (TOS = 1816 min), heptadecane is almost the exclusive product cor-
responding to a conversion of stearic acid at 75%. Neither stearic acid nor unsaturated side products
are detected at this position, but during the course of the experimental interval between step change
IV) and V) both unconverted stearic acid and unsaturated products were formed. Around TOS =
2800 min about 8-12% yield of 1-heptadecene and 4-5% yield of C17 -monoaromatics were obtained,
while the yield of heptadecane fell below 70% and declined afterwards. The mass balance only
reached 75% and increased to about 95% at TOS = 2800 min. This is an indication that products or
reactant were held up in the catalyst beads or that heavy condensation products were formed that
could not be detected by the GC analysis method applied. Thus it can be expected that some of the
unsaturated compounds have actually condensed to heavier products or simply started to form coke
by cyclisation and liberation of H2 .
The CO2 -level corresponded to 40% conversion in the start of step change IV) at TOS = 1816
min, which rose slightly to 48% at TOS = 2600 min. Interestingly, almost no CO was present in the
gas atmosphere at TOS = 1650, i.e. 1 h after the step change III) to Ar gas. At this point the CO
concentration rose steadily to around 28% at TOS = 2600 min. Surprisingly, at TOS = 2600-2800
the gas atmosphere reversed in composition, as the CO-level rose to above 50% while the CO2 -level
decreased to just 3%.
The heptadecene and C17 -monoaromatics formed strongly points to deactivation of the catalyst
by coking. A possible explanation of the gas-phase behaviour is that the unsaturation of the pro-
ducts formed have led to the liberation of more and more hydrogen used to partly methanate, partly
water-gas-shift the CO2 to CO (see Figure 1.7 on page 25). However, the unsaturated hydrocarbons
formed by the catalytic liberation of H2 result in C-C-coupling and build-up of carbonaceous by-
products in the pore system of the catalyst. The switch-over of resulting gas from CO2 to CO over
time has very recently been reported also by Immer and co-authors [331] who suspected that too
long experimentation in H2 -deficient atmosphere led to the shift in gas production.
Unsaturated compounds are often formed under conditions leading to deactivation [20, 211].
The liquid flow was switched from dilute stearic acid to dodecane flow at TOS = 3016 min (step
change V)), while the Ar gas flow of 42 Nml/min was maintained. Surprisingly, the CO-concentration
remained on levels corresponding to a conversion of about 55-65% for several hours, and it did not
start to decline until the reactor was cooled to 150℃ at TOS = 3300 min. The sum of all CO and
CO2 measured after this point amounted to 4.21 mmol, and it may be hypothesised that the CO was
covering the surface of the Pd parcticles. However, if applying a mean particle diameter of 12 nm
(measured by XRD, see section 3.5.3.2), asuming that Pd particles are spheres and that the Pd:CO
adsorption ratio is 2:1 [326], maximally 0.250 mmol can be adsorbed on the Pd particles. Thus, CO
and CO2 produced after the flow was switched to dodecane must primarily have come from reactant
still present in the reactor. It is suspected that a large amount of stearic acid was still situated in
the pore structure of the catalyst beads, as a large part of the catalysts beads were inert - namely all
but the outermost layer where the Pd particles are supported. The stearic acid continued to slowly
96
diffuse inside the catalyst beads and deoxygenates over time.

It has been suggested that the deoxygenation takes place via formic acid formation on the catalyst
surface, as depicted in Figure 3.1. Fatty acid decarbonylates to directly yield a 1-alkene and formic
acid. The alkene would be saturated by H2 in the gas feed, while the formic acid would disintegrate
to yield either CO2 and H2 (by dehydrogenation) or CO and H2 O (by dehydration) [322, 332]. As
formic acid would decompose over the Pd catalyst it was not possible to confirm this mechanism in
the present work. The measured liquid yield of up to 15% unsaturated products can be explained by
the decarbonylation pathway or by formic acid formation and dehydration. Lack of H2 means that
saturation of formed alkenes was slow and dependent on release of H2 by dehydrogenation.
3.5.1.3 Concentrated stearic acid in pure Ar - step changes VI)-VII)

After the reaction with diluted stearic acid the reactor was flushed in Ar at 150℃, then in dodecane
and then again in Ar overnight. Then the reactor was reheated to to 300℃ at 10℃/min, and once
the temperature of the reactor was stable the flow of pure stearic was started. Liquid samples and
online gas analysis were monitored during two days, at TOS = 4359-7056 min as shown in Figure
3.16.
1,2
1.2
Step change # VI) - VII) a
1
1.0
molar fraction
0,8
0.8
0,6
0.6
Gas: Ar
0,4
0.4 Liquid: 100% Stearic acid
0,2
0.2
0
0.0
4550 5050 5550 6050 6550 7050
TOS (min)
Figure 3.16: Deoxygenation of concentrated stearic acid in pure Ar flow over 2 wt% Pd/Sibunit. Legend: Liquid
mole balance ( ), Stearic acid ( • ), heptadecane ( N ), heptadecene ( ), undecylbenzene ( × ), CO (x), CO2
(•).
The first sample taken out had a mole fraction of stearic of almost 100%, but this declined in the
following samples to around 80%. Heptadecane yield declined gradually from 2% to 1% with the
exception of one sample. Likewise, the concentration of CO in the gas phase gradually declined from
about 3% to 1.5 % in the end of the experimentation interval. There was, however, no trace of CO2 .
Thus, there is a slightly better correspondance between the gas and liquid analysis offhand, and it
seems likely that the lack of H2 in the feed have minimised the extent of methanation.
However, about 20% of the stearic acid is missing in the liquid product analysis. It is suspected
that undetectable heavy condensation products have been formed. No 1-heptadecene is observed,
but two heptadecene molecules may undergo coupling-reaction to form tetratriacontene (a C34 -
alkene). Ketonisation from two molecules of stearic acid to form 18-oxopentatriacontane (a C35 -
97
ketone) is however not likely as ketonisations usually require a basic or acidic co-catalyst in addition
to a transition metal for reaction to take place [17, 228, 260]. If heavy condensation product are
formed it is likely that they cannot be detected by the current GC method. Heavy hydrocarbons
would be difficult to vaporise in the split/splitless injector and the column may not have been heated
sufficiently in the applied GC method to allow vaporisation of the species.
3.5.1.4 Pure stearic acid under 5% hydrogen in Ar - step changes VIII)-IX)

Following the experimentation the final assessment of steady-state catalytic behaviour was deoxy-
genation of neat stearic acid in hydrogen-containing atmosphere. After flushing the reactor with
dodecane followed by Ar at 150℃, the reactor was reheated in 5% H2 /Ar and stearic acid was led to
the reactor to obtain the steady-state deoxygenation as seen in Figure 3.17. Liquid samples confirm-
ing the stable behaviour were obtained at TOS = 8300-18500 min between step changes VIII) and
IX) in Table 3.3.
1.2
1,2
Step change # VIII) - IX) b
1.0
1
molar fraction
0.8
0,8
0.6
0,6
Gas: 5% H2/Ar
0.4
0,4 Liquid: 100% stearic acid
0.2
0,2
0.0
0
7800 9800 11800 13800 15800 17800
TOS (min)
Figure 3.17: Deoxygenation of concentrated stearic acid in 5% H2 /Ar over 2 wt% Pd/Sibunit. Legend: Liquid mole
balance ( ), Stearic acid ( • ), heptadecane ( N ), heptadecene ( ), undecylbenzene ( × ), CO (x), CO2 (•).
During this experimentation interval a constant yield of 12% heptadecane at 100% selectivity was
obtained, while the remaining liquid was unconverted stearic acid. The gas-phase yielded a level of
CO corresponding to ca. 7% conversion, steadily decreasing from 8% to 6% over the course of the
entire 10200 min TOS. No CO2 was observed, but the lack of part of the COx was again observed,
most likely due to methane formation with hydrogen.
The lack of CO2 could originate from the WGS equilibrium by consumption of H2 (Figure 1.7) to
yield H2 O and CO. This would, on the other hand, result in some CO2 left in the gas atmosphere,
but this is not the case. Immer and co-workers suggested that the deoxygenation proceedes via
decarbonylation over a catalyst already aged [331]. In this mechanism, as is suggested in Figure
1.6, a fatty acid decarbonylates to yield a 1-alkene while CO and H2 O are liberated, for instance
via formic acid as previously stated [321, 322, 332] (see Figure 3.1). The alkene is then quickly
saturated to an alkane by H2 in the gas. This would explain the complete absence of CO2 during the
deoxygenation of concentrated stearic acid, suggesting that WGS does not take place. Whether or not
98
decarboxylation and WGS or decarbonylation dominated, the net result was the same: One molecule
of heptadecane, of CO and of H2 O have been formed during the deoxygenation.
3.5.1.5 Overview of longterm experimentation

Table 3.4 shows the overview of the steady-state behaviour of deoxygenation at 300℃ and 42 Nml/min
sweeping gas at 20 bar of 0.075 ml/min solutions of stearic acid.
It is evident that the experimentation in pure Ar atmosphere both led to deactivation while re-
sulting molar balances of reactants and products were somewhat below 1. This point suggests that
deactivation took place by dehydrogenation and coking reactions in the pore system, and that the
products were either trapped in the pores or too heavy to be analysed in the GC-FID used to analyse
the samples. It should be noted that the experiments have been performed with the same catalyst
bed exposed to each condition consecutively, so deactivation is naturally cumulative of the previous
conditions applied.
Table 3.4: Yield of heptadecane on basis of stearic acid in the feed under stable conditions. 300℃, 20 bar sweeping
gas at 42 Nml/min, liquid feed 0.075 ml/min. Dilution is in dodecane
Sweeping gas atmo-

sphere
Liquid Feed: 5 % H2 in Ar Pure Ar
Diluted stearic acid 95% (15 mol%) 75% decreasing to
(mol% stearic acid in 55%; formation of
dodecane) unsaturated and
aromatic C17 com-
pounds (10 mol%)
Pure stearic acid 12 % 2% decreasing to 1%;
Possible formation of
heavier products
Lestari et al. observed initial deactivation in 20 bar pure Ar at 360℃ over 5 wt% Pd/Sibunit
during the continuous stearic acid deoxygenation [211]. However, at this temperature it was not
possible to regain the initial catalyst activity by switching gas flow to 5% H2 /Ar, but a stable con-
version of 15% was reached. This is comparable to the 12% obtained in the present work, albeit the
degree of deactivation of the catalysts may be different.
Importantly, part of the lost activity from deoxygenation in Ar could be regenerated as 12% hep-
tadecane yield was obtained steadily in over a week of experimentation. This further suggests that
the use of 5% H2 /Ar is sufficient to sustain the catalyst activity for an extended period of reaction.
However, as is evident from Figure 3.17, part of the COx formed will still be converted to methane
by methanation. It is not completely certain why decarbonylation and unsaturated products are not
seen during the experimentation with concentrated stearic acid under pure Ar gas flow.
3.5.2 Transients during deoxygenation step-changes

In industrial practice not only the steady-state behaviour is relevant for the performance of a cata-
lytic reactor, but also the behaviour when changing conditions and flows, starting and stopping the
reactor. In terms of catalyst evaluation in lab scale it is likewise relevant to investigate the start-up,
99
shut-down and step changes and the responses of the reactor setup taking place between the steady
states. This may reveal details about the flow behaviour and some of the pathways of the catalytic
reactions taking place in the catalyst bed. Thus, four step changes are treated here, namely:
• Startup in Ar and dilute stearic acid - step change I)
• Switch to 5% H2 /Ar in dilute stearic acid - step change II)
• Start-up of 100 stearic acid in 5% H2 /Ar - step change VIII)
• Shut-down of 100 stearic acid in 5% H2 /Ar - step change IX)
The step changes are shown in Figure 3.18 (I) and II)), Figure 3.19 (VIII)), and Figure 3.20 (IX)),
respectively.
3.5.2.1 Startup in Ar and dilute stearic acid - step change I)

After the activation of the catalyst by reduction in H2 the gas was changed to Ar at 42 ml/min to flush
the reactor overnight. Then, liquid flow of 0.075 ml/min dodecane was started to the reactor at TOS
= -55 min. At TOS = 0 min the liquid flow was switched to 0.075 ml/min of 10 mol% stearic acid
diluted in dodecane - step change I). The propagation of the reactant and products of deoxygenation
were monitored during the step change as seen in Figure 3.18.
The CO2 started to rise after about a 15 min lag. Then, it gradually rose and stabilised at a level
corresponding to 26% conversion in the course of the next 4-5 h. The CO concentration started to
rise at TOS = 90 min, and gradually stabilised at a level corresponding to 8 % at TOS = 300 min.
1,5
1.5
Gas: Ar Gas: 5% H2/Ar a
1.25 Liquid: 10 mol% stearic acid
1,25 Liquid: 10 mol% stearic acid
molar fraction
1
1.0
I) II)
0.75
0,75
0.5
0,5
0.25
0,25
0.0
0
-10 90 190 290 390 490
TOS (min)
Figure 3.18: Step changes of dodecane to dilute stearic acid under Ar (step change I)) and change to 5% H2 /Ar (Step
change II)), performed over 2 wt% Pd/Sibunit. Legend: Liquid mole balance ( ), Stearic acid ( • ), heptadecane
( N ), heptadecene ( ), undecylbenzene ( × ), CO (x), CO2 (•).
Minor amounts of stearic acid were found in the liquid samples taken out between 0 and 120 min,
which were probably due to impurities located in the sampling system from previous runs. However,
ca. 1½ -2 h after the step change, the concentration of heptadecane started to rise and developed into
a stable concentration corresponding to 37% conversion of stearic acid 5 h after step change I). A few
100
percent of unconverted stearic acid is still observed in this time interval. At TOS = 300 min, 1% yield
of a monoaromatic C17 -product was obtained. As previously stated this behaviour is connected with
deactivation of catalyst.
Importantly, the content of CO2 and CO in the gas phase accurately corresponded to the amount
of liquid products observed at TOS = 300 min. This indicates that no major side-reactions have taken
place in this time interval.
3.5.2.2 Switch to 5% hydrogen in Ar - step change II)
At 300 min TOS the gas flow was switched to 5% H2 /Ar while keeping the liquid flow constant. The
concentrations of CO2 and CO from the online gas analysis started to respond and rise after 15 min,
forming maxima of concentrations corresponding to 49% and 23% conversion after 1 h and 1/2 h
respectively, after which the concentrations of both CO2 and CO slowly declined and settled around
35% and 17% at steady-state. The liquid concentration started to respond to the step change after
1 h (at TOS = 360 min). Then the concentration of heptadecane increased and peaked at a molar
concentration around 110%, after which the concentration declined and settled at 90% after totally
2.5 h (TOS = 450 min). This is shown in Figure 3.18 after the dashed line marking step change II).
Naturally the molar balance should complete around 100% in steady-state behaviour, but the
step changes the concentration response may temporarily reach values over 100% due to the changed
reaction conditions. It is suspected that stearic acid absorbed in the porous catalyst beads from 0-300
min TOS is converted at a faster rate when H2 is present in the gas atmosphere and thus released as
heptadecane, giving a temporary mole balance of over 100%.
The low net flow of stearic acid to the reactor is responsible for the response-time of 1 h of the
reactant and products in the liquid feed. The liquid feedstock likely built up a trickling film that
wetted the catalyst bed. Stearic acid has a normal boiling point of 383℃ and thus passes the reactor
in the liquid phase. Heptadecane and 1-heptadecene have normal boiling points of 302℃ and could
thus be present in the gas phase at 1 bar, but due to the long response time of these molecules after
the step change they are expected also to propagate through the reactor in the liquid phase.
The normal boiling point of dodecane is 216℃, while the pressure of the saturated vapors at
300℃ is 6 bar. Although the reactor pressure was 20 bar, the gas flow was so high and the amount
of the dodecane solvent supplied so low that only about 4 bar of dodecane vapor could have been
formed if all the dodecane vaporised. Thus, it must be assumed that dodecane vaporised and passed
through the reactor in the gas-phase at 300℃. When the temperature cooled after the reactor outlet,
most of the dodecane condensed again.
Contrary to the case of deoxygenation in pure Ar at TOS = 300 min, the content of CO2 and
CO in the gas phase only corresponded to about 60% of the products formed with the amount of
product observed. This is likely due to the formation of methane from both CO and CO2 in the
hydrogen-containing atmosphere.
101
3.5.2.3 Start-up of 100% stearic acid in 5% hydrogen in Ar - step change VIII)

Following the deoxygenation of concentrated stearic acid under pure Ar gas (TOS = 4481-7056 min,
see Table 3.3), the reactor was flushed for several hours with pure dodecane at 150℃ and left at this
temperature overnight solely with Ar gas flow. Then the reactor was reheated to 300℃ and the gas
was switched to 42 ml/min of 5% H2 /Ar. Flow of concentrated stearic acid was started at TOS =
7066 min. This is seen in Figure 3.19.
The CO signal started to increase slowly after 5 minutes, but 3 h after step change IX) a concen-
tration corresponding to only 4-5% conversion was seen. Only insignificant amounts of CO2 were
observed at the same time. The concentration of both stearic acid and heptadecane started to in-
crease 1 h after step change IX) and 2-2½ h after the step change a relatively stable yield of 12%
heptadecane started to result.
The difference between the calculated conversion in respectively liquid and gas phase points
to side reactions in the gas phase, possibly methanation. Almost two thirds of the CO formed by
deoxygenation have reacted further to form methane.
Compared to the start-up step change I) in dilute stearic acid, the liquid product composition
from start-up in concentrated liquid feed respond faster to the step change. The higher net flow of
liquid simply reached the outlet of the reactor much faster, and the film pattern of trickling liquid
flow is established faster as well.
1.2
1,2
VIII)
Gas: 5% H2/Ar b
1.0
1
Liquid: Dodecane -> stearic acid
molar fraction
0.8
0,8
0.6
0,6
0.4
0,4
0.2
0,2
0.0
0
7060 7080 7100 7120 7140 7160 7180 7200 7220
TOS (min)
Figure 3.19: Step change of dodecane to concentrated stearic acid flow over 2 wt% Pd/Sibunit - step change VIII).
Legend: Liquid mole balance ( ), Stearic acid ( • ), heptadecane ( N ), heptadecene ( ), undecylbenzene
( × ), CO (x), CO2 (•).
3.5.2.4 Shut-down of 100% stearic acid in 5% hydrogen in Ar - step change IX)

The shutdown of the reactor after more than 10000 min steady-state deoxygenation of concentrated
stearic acid was monitored. The flow was switched to pure dodecane at TOS = 18506 min (step
change IX)) and the response after this step change is shown in Figure 3.20.
The CO/CO2 online gas analysis revealed that the CO produced decreased steadily from a level
corresponding to 6-7% conversion at the start of the step change to about 2% after 6 h. The rapid
102
responses observed when starting up the reactor in dilute stearic acid were not seen from the gas
analysis. Initially, the amount of CO present in the gas phase corresponded to 55% of the heptade-
cane present - thus, 45% of the CO have probably reacted to methane. No CO2 was observed.
1.2
IX) Gas: 5% H2/Ar c
1.0 Liquid: Stearic acid -> dodecane
molar fractoin
0.8
0.6
0.4
0.2
0.0
18500 18550 18600 18650 18700 18750 18800 18850 18900
TOS (min)
Figure 3.20: Step change of concentrated stearic acid flow to dodecane over 2 wt% Pd/Sibunit - step change IX).
Legend: Liquid mole balance ( ), Stearic acid ( • ), heptadecane ( N ), heptadecene ( ), undecylbenzene
( × ), CO (x), CO2 (•).
The response in stearic acid from the step change occured after about 1 h. The stearic acid molar
fraction did, however not drop to a molar concentration of 0. Instead, a molar fraction of 10%
stearic acid is obtained, after 2½ h (TOS = 18660 min), which was then constant for about an hour,
after which it decreased only slowly. The heptadecane production was about 12% initially, and
the liquid sample analysis showed a response after 1 h to the step change also from heptadecane.
However, instead of declining the molar fraction of heptadecane increased slightly to about 17%,
which persisted for over 2 h. After this the heptadecane molar fraction decreased slowly for the next
few hours.
A plausible interpretation of this step change result is that stearic acid was situated throughout
the pore system of the catalyst and was only released very slowly to the exterior to leave the reactor
- in fact most of the stearic acid were only released after having reacted to heptadecane, likewise
releasing CO or CO2 . The interior pore volume was considerable (0.35 ml/g) and worked as a re-
actant holdup, and most of the catalyst interior was inert due to the eggshell-impregnation of Pd
in the outer rim of the beads. Possibly the exterior liquid film on the catalyst beads of stearic acid
was gradually eliminated when the flow of stearic acid was changed to dodecane. This explanation
further supports the fact that CO concentration only diminished very slowly despite constant flow
of 42 Nml/min of Ar as observed from Figure 3.20.
In all cases the gas phase responsed to the step changes have taken place within a few minutes (5-
15 min), but steady-state gas-phase compositions took 2-4 h to develop, which was somewhat slower
that the change in liquid sample composition, which delevoped steady-states in 1-2 h.
103
3.5.3 Characterisation of catalysts
To investigate the causes of the observed deactivation during the long-term experimentation with
concentrated and diluted stearic acid, the catalyst beads were sorted in batches depending on their
axial location in the tubular reactor and analysed via BET, TEM, and XRD. Before unloading the
catalyst bed it was flushed in dodecane and Ar at 150℃, followed by flushing solely in Ar.
Catalyst deactivation may take place by a number of mechanisms. For instance, if the tempe-
rature is high enough, the surface area of the metal particles may decrease due to agglomeration or
sintering of the metal particles - the particles grow together via migration on the support. Impurities
in the feed or side products of the reaction may poison the catalysts, for instance species binding too
strongly to the active metal surface to desorp so that the active site is blocked. Furthermore the active
catalyst phase may also leach to the solvent or reactant passing the catalysts bed, thus washing the
active phase out of reactor.
To investigate the location of active Pd particles in the sibunit carbon beads, the fresh catalyst was
subjected to laser ablation (LA-ICP-MS) analysis (Figure 3.21). 2 particles cut in half were used for
the measurements. These confirmed that all of the palladium is impregnated in the outer rim of the
sibunit beads in an eggshell-like layer corresponding to about the outer fifth of the catalyst beads,
while the centre of the sibunit particle is completely free of Pd. This impregnation type minimises
the diffusion pathway for the otherwise bulky fat molecules.
Figure 3.21: Laser ablation of the catalyst beads coupled with ICP-MS to detect ions of palladium. The laser moved
at a constant rate over the catalyst diameter, blasting any species to the plasma phase to be detected by the MS.
The surface area, pore size distributions, free volume, and mean particle sizes of the fresh catalyst
beads and some of the the spent catalyst batches 3.5 were investigated via XRD and TEM. TEM
micrographs, the particle size distributions, and the XRD pattern for the fresh and spent catalyst
samples # 1, 4, and 6 (from Table 3.5) are shown in Figures 3.22, 3.23, 3.24, and 3.25, respectively.
104
Table 3.5: Characterisation of the spent catalyst depending on position in the catalyst bed
Total
Sample # Pore Tot. Pore size distribu- TEM Pd diam-
Sample Surface
(flow volume vol. tion based on pore eter (XRD in
mass area
direction: ↓) (BET) adsorb. diameter in nm parenthesis)
(BET)
>10 5-10 2-5 <2
[g]1 [m2 /g] [ml/g] [ml/g]2 % % % % [nm]3
1 1.22 101 0.321 327 1.3 7.4 76.2 15.2 7.7±3.5 (11.9)
2 1.17 118 0.430 395 1.3 8.5 77.8 12.3 —
3 1.13 130 0.377 361 1.1 8.7 71.4 18.8 —
4 2.69 137 0.383 345 1.5 9.1 76.3 13.2 7.1±4.5 (12.4)
5 2.65 157 0.502 410 1.4 10.5 70.4 17.7 —
6 3.87 166 0.474 383 1.3 10.0 67.0 21.8 6.0±4.5 (11.4)
Fresh cat. — 361 0.873 623 2.0 15.5 73.2 9.3 6.7±2.8 (12.8)
1 Masses of the spent catalysts samples taken out and sorted consecutively.
2 N2 at 298 K and 1 atm.
3 XRD mean Pd particle is based on volume calculated from the Scherrer equation; TEM is based on the
number of particles.
3.5.3.1 Coking in the catalyst pores

The summed mass of the spent catalyst beads was 12.72, however only 10 g of catalyst beads were
loaded into the reactor. It is suspected that this weight gain is due to the formation of carbon in
the pore system, although it could also be due to the presence of remaining fatty acids in the pores.
The weight gain corresponds to 0.227 mol of C, assuming 100% carbon. For comparison, the total
amount of stearic acid fed to the reactor amounts to 3.43 mol or 61.7 mol of carbon atoms.
The results from BET analysis of spent and fresh catalyst beads are shown in Table 3.5 as a func-
tion of their axial position in the tubular reactor.
Importantly, the catalyst beads near the inlet in the reactor (sample #1) had the lowest BET sur-
face area of 101 m2 /g, which increased monotonously to 166 m2 /g near the outlet (sample # 6).
However, the fresh catalyst has a surface area of 361 m2 /g. Thus, surface area decreased from 72%
near the inlet to 54% near the outlet. As the catalyst bed was left with dodecane and 5% H2 /Ar flow
for 19 h at 300℃ and stearic acid is assumed to be washed out of the catalysts beads, it is therefore
concluded that deactivation is caused by coking reactions, which have been reported in the literature
[214, 330].
The development of this deactivation profile is not fully understood. Possibly the catalyst was
more exposed to reactants and side reactions near the reactor inlet, thus forming coke and heavier
compounds leading to deactivation. If a temperature gradient had been built up downdraft in the
reactor, the reaction rates would likely have been different from each other, which could easily also
apply to side reactions leading to carbon build-up.
Compared to the fresh catalyst beads, the deactivated beads had a higher percentage of small
pores below 2 nm in diameter and a lower percentage of large pores above 5 nm in diameter. This
suggests that coking caused narrowing of the pores and that partial or complete occlusion of some
of the smaller pores have taken place. Previously it has been reported that deactivation during de-
oxygenation of stearic acid over 1% wt Pd/C takes place via dehydrogenation in hydrogen-deficient
atmosphere, leading to formation of alkenes and monoaromatics via cyclisations [20, 23, 331]. It can
105
therefore be concluded that the unsaturated liquid products observed when using pure Ar atmos-
phere were indicative of the deactivation taking place.
Deactivation and coke formed in the pores did not hinder conversion completely as 12% hep-
tadecane was formed selectively from stearic acid in the final 10500 min TOS (see Figure 3.17) under
5% H2 /Ar.
3.5.3.2 Pd particle sizes

From the X-ray diffractograms the mean particle diameter, <d>, can be calculated via the Scherrer
equation based on diffraction pattern intensity, which correlates with the mean volume of the Pd
particles:
Kλ
< d >= β cos θ
K is a shape factor usually around 0.9, λ is the wavelength of the X-ray source, β is the full
width at half maximum peak intensity, and θ is the Bragg angle. The surface area from the TEM
micrographs are however based on the frequency (i.e. the number) of particles. Thus, the bigger
particles contribute much more to the mean particle size in the X-ray diffractograms as compared
to the TEM micrographs, which is the reason behind the lower surface areas measured by TEM
compared to XRD.
The mean Pd particle diameters based on TEM micrographs varies from 7.7 nm in the top (near
the inlet) to 6.1 in the bottom. The fresh catalyst is however 6.7 nm. TEM only counts from the
small fraction of the sample that is exposed to the electron beam, thus it is somewhat dependent
on the specific sample part which is being analysed. From the figures 3.22, 3.23, 3.24, and 3.25 it
is evident that the TEM particle size distributions are quite similar, and that the variation may be
within experimental uncertainty.
The diffraction patterns of all of the spent catalysts as well as the fresh catalyst showed mean
diameters ranging from 11.4 nm to 12.8 nm (volume-based), as shown in Table 3.5. The variation
was not systemtic, and a slightly larger mean particle size of Pd was found on the fresh catalyst
compared to the spent ones, while the XRD mean particle sizes did not change monotonically as a
function of bed location.
It is concluded that the variations between the mean particle sizes of the catalysts are too small to
be significant. Furthermore it is excluded that sintering or agglomeration took place or contributed
to the deactivation of the Pd/C catalyst. This is in accordance with other reports in the literature on
deoxygenation over 1 wt% or 5 wt% Pd supported on carbon [23, 214, 330, 331].
3.5.3.3 Leaching of Pd
During the entire two-week experimentation, two consecutive batches of effluent from the collector
vessel were collected, each of ca. 0.5 kg. The two batches were melted and each of them thoroughly
homogenised by stirring, and samples of a few mg were analysed by ICP-EOS to assess the amount
of Pd. In the first batch the content of Pd was 1.67 mg/kg effluent, while in the second only 0.59
mg/kg effluent was found. This corresponded to 0.43% and 0.15%, respectively, of the total carbon-
supported Pd present in the catalyst bed. Thus, the degree of leaching over the two weeks was
106
Pd(111)
30 1500
<d> 6.7± 2.8 nm
Frequency (%)
1250
Intensity (count)
Pd(200)
20
1000
10
750
0 500
0 5 10 15 20 25 30 35 40 30 35 40 45 50
Particle size (nm) 2θ (degree)
Figure 3.22: TEM micrograph (left-hand side), particle size distribution (middle), and XRD-pattern (right-
hand side) of the fresh catalyst.
Pd(111)
30 1750
<d> 7.7± 3.5 nm
Intensity (count)
Frequency (%)
1500
20
Pd(200)
1250
10 1000
750
0
0 5 10 15 20 25 30 35 40 30 35 40 45 50
hand side) of spent catalyst sample # 1 (from the top of the catalyst bed).
Pd(111)
30 <d> 7.1± 4.5 nm 1500
Frequency (%)
1250
Pd(200)
Intensity (count)
20
1000
10 750
500
0
0 5 10 15 20 25 30 35 40 30 35 40 45 50
hand side) of spent catalyst sample # 5 (from the middle of the catalyst bed).
1500
Pd(111)
30 <d> 6.0± 4.5 nm

Intensity (count)
Frequency (%)
1250
Pd(200)
20
1000
10
750
0 500
0 5 10 15 20 25 30 35 40 30 35 40 45 50
2θ (degree)
Particle size (nm)
hand side) of spent catalyst sample # 9 (from the bottom of the catalyst bed).
107
minimal, corresponding to 0.58% of the Pd. It is also evident that the leaching degree lowered over
time. The leached palladium probably constituted a minimal fraction of loosely bound Pd particles
which were slowly washed out by the feed. Deactivation was therefore not due to leaching of the
active metal. Minimal leaching of Pd during deoxygenation over Pd/Sibunit have recently been
reported [23, 330].
3.5.4 Effect of Sibunit support and temperature

The step change experiments in Section 3.5.1 prompted the investigation of hydrodynamic properties
of the reactor without a catalyst to determine evaluate the flow behaviour of the stearic acid in the
reactor. The thermal influence as well as the influence of the reactor and the support material of
Sibunit beads. Thus, 10 g of Sibunit beads of the same sizes interval as the ones used in Section 3.5.1
catalyst was loaded into the reactor following the description in Section 3.4.3 and in Figure 3.13.
The reactor was flushed for 6 h in Ar at a flow of 42 Nml/min at 20 bar. Then dodecane flow at
0.075 ml/min was started and the reactor was left at these conditions overnight to reach steady-state.
At TOS = 0 min the liquid flow of stearic acid was started, and this was left for the next 480 min TOS
where the flow was switched back to dodecane. The results are shown in Figure 3.26.
Two samples at TOS = 0 min and 60 min with stearic acid content below 10% of were observed
during the first hour of the step change, however the response from the reactor to the liquid flow
change cannot have propagated so fast through the reactor, meaning that these constitute impurities
of stearic acid in the liquid sampling system. Two consecutive valves constituted the sampling system
and could not be dismantled, by which cleaning of the connecting fittings was difficult. At TOS =
90-210 min the CO-level increased from a level corresponding to 0% to 5 % conversion. CO2 was
however only observed at levels corresponding to under 0.5% conversion during the entire treatment.
The stearic acid concentration started to increase at TOS = 90 min - most steeply after 165 min
TOS. At TOS = 150 min the production of heptadecane increased from 0 to about 5% yield, thus
reaching the same level as the CO produced. When the production of heptadecanestarted to increase
also about 1% of monoaromatic C17 -compounds were found, but this only persisted for about an
hour. Between 0 and 1% of 1-heptadecene was also formed alongside with the heptadecane. At TOS
= 240 the molar balance of the liquid samples level reached a steady-state level of about 97%, 90%
hereof being unconverted stearic acid, thus 1.5 h was necessary for the liquid profile to develop in
the reactor. Thus, it took 4 h to reach stable operation after the step-change.
The step change back to dodecane liquid flow was initiated by switching the liquid feed at TOS
= 480 min. 1 h after the change, the concentration of stearic acid in the reactor started to drop quite
steeply for the next hour, but after TOS = 600 min the drop in stearic acid concentration flattened
off towards a molar fraction of stearic acid of 8% at TOS = 840 min. It is furthermore seen that the
heptadecane yield dropped to about 3% 2 h after the step change, but it increased to 4-5% towards
TOS = 840 min. Also, the CO-level remains quite stable corresponding to 5% conversion for the first
6 h after the step change (until TOS = 840 min), but after this point the CO-level decreased steadily
until TOS = 1100 min.
This behaviour of switching from stearic acid to dodecane is similar to that observed during
108
switch to dodecane in diluted stearic acid flow in Figure 3.15, section 3.5.1.2 or in concentrated
stearic acid flow in Figure 3.20, section 3.5.2.4. It is thus confirmed that stearic acid is simply still
present in the pore system of the Sibunit and continued to partly react and partly diffuse to the
exterior of the Sibunit to be removed by the flow of Ar and dodecane. This can explain why about 3-
4% of the liquid compounds were “missing” in the mole balance at TOS = 240-480 min: The stearic
acid was absorbed in the pores of the Sibunit beads as soon as it reached the Sibunit beads, and
this is naturally missing in the mole balance. It may be argued that as accumulation of reactant
is taking place in the reactor, the stable behaviour of TOS = 240-480 min cannot stricly be called
“steady-state”.
Apparantly, with the lack of catalyst and added H2 there is a good correspondance between the
liquid sample analysis and the CO-concentration in the gas atmosphere, especially at the conditions
approaching steady-state from TOS = 195 min until the step change started to affect the liquid profile
in the reactor at TOS = 540 min.
It may be speculated that the production of CO during the deoxygenation of stearic acid, as
observed during the entire run of section 3.5.1, were produced as the thermal effect and due to a
catalytic effect of the support while the Pd catalyst may have been deactivated completely. Further
investigations of these phenomena would be necessary to further elucidate the different reaction
routes from the fatty acids to alkanes.
1
Step change in Ar atmosphere:
0,9 dodecane -> stearic acid -> dodecane
0,8
0,7
Molar fraction (X)
0,6
0,5
0,4
0,3
0,2
0,1
0
0 200 400 600 800 1000 1200
TOS (min)
Figure 3.26: Step change behaviour of dodecane -> stearic acid -> dodecane. Pumping speed 0.075 ml/min. 300℃
and 20 bar Ar at 42 Nml/min. Legend: Liquid mole balance (), Stearic acid (•), heptadecane (N), heptadecene (),
undecylbenzene (×), CO (x), CO2 (•)
109
3.6. Continuous hydrodeoxygenation: Stearic acid and derivatives
3.6 Continuous hydrodeoxygenation: Stearic acid and derivatives
Both fatty acids, fatty acid alkyl esters and triglycerides have been suggested as feedstocks for de-
oxygenation [16, 208, 210, 324]. Both free fatty acids and glycerides are present in fats and oils, and
the former is especially present in waste feedstocks. It is therefore interesting to relate these feed-
stocks in terms of deoxygenation reactivity and reaction pathways. The aim of the present study is
therefore to compare the deoxygenation activity during reaction in both H2 -containing and inert gas
flow and by using a fresh catalyst for each feedstock. For comparison, stearic acid, ethyl stearate, and
tristearin (glyceryl-tristearate) have been used as model compounds. The treatment was performed
at 300℃ and at 20 bar pressure. During the first 4500 min TOS (stearic acid: 5760 min) a gas flow
of 42 Nml/min of 5% H2 /Ar was fed to the reactor. Then the gas atmosphere was changed to pure
Ar at 42 Nml/min.
The packing of the reactor was performed almost as shown in Figure 3.13, with the notable ex-
ception that the catalyst bed for each feed was loaded into the reactor by separating the catalyst
beads into five equally big beds of 2 g each, separated each by a layer of quartz wool. This made it
much easier to sort the spent beads into 5 equally sized batches after reaction for characterisation of
the spent catalysts.
After flushing with Ar, the catalysts were reduced prior to reaction in 5% H2 /Ar: The reactor was
heated at ca. 10℃/min to 150℃ where it was kept for 1 h. Then the temperature was increased to
the reaction temperature of 300℃ at 10℃/min where the reactor was kept for 0.5 h before the flow
of stearic acid was started. The CO/CO2 online analysis was unfortunately switched off during part
of the experimentation.
3.6.1 Stearic acid

Pumping of stearic acid feed to the reactor was started TOS = 0 min, and the liquid samples re-
sponded to the reactant flow after 180 min as visualised in Figure 3.27. Initially almost 100% hep-
tadecane was formed and only traces of stearic acid were observed. However, the yield of stearic
acid had decreased to 80% at TOS = 1560 min and 72% at TOS = 5760 min, thus showing a slow
and gradual deactivation during the deoxygenation in 5% H2 /Ar. Correspondingly, the amount of
unconverted stearic acid of 17-23% were observed at TOS = 1560-5760 min. These results clearly
show that the 12% yield of heptadecane obtained in section 3.5.1.4 was the product of an already
deactivated catalyst.
The CO and CO2 -levels started to increased 60 min after the pumping of stearic acid was started.
The online analyser was offline during the period of TOS = 360-2280 min. However, the online gas
analysis of the gas-phase clearly shows that CO2 corresponding to 30 % and CO corresponding to 5%
conversion was formed initially at TOS = 135 min, but the two gas concentrations steadily aproached
each other and met at a level correponding to 16% conversion of feedstock for each gas at TOS = 5760
min.
At TOS = 5760 the gas atmosphere was changed to pure Ar. This led to a drop in the CO-level
to 2% conversion and a peak increase in the CO2 -level in the gas; however, the CO2 -level dropped
110
again to a level of 0% conversion at TOS = 6040, while the CO-concentration rose to a peak of 5% an
declined thereafter.
After the change to Ar gas step change the heptadecane-concentration remained constant at a
yield of ca. 75% for 2 h, but then the concentration dropped steeply to 8% at 5 h after the change to Ar
gas. The heptadecane yield further decreased and leveled off at ca. 2%, following the concentration of
CO. The steep drop in heptadecane concentration is followed by an equally steep rise in stearic acid
concentration a few hours after the pure Ar-flow was started. The peak CO-concentration around
TOS = 6040 min is accompanied by a peak of 1-heptadecene of 2% yield and undecylbenzene of 1%
yield. Thus, catalyst deactivation in hydrogen-deficient gas flow was also in this case accompanied
by formation of unsaturated C17 -products. The remaining duration of the experiment, until TOS =
9000 min, 3% decreasing to 2% heptadecane yield was obtained, in accordance with behaviour of
the the deactivated catalyst described in Section 3.5.1.3
1,4
Stearic acid
1,2
1
Molar fraction (X)
0,8
1-heptadecene
Heptadecane
C17-heavy
0,6
Stearic acid
Molar balance
CO, conv., online
0,4
CO2, conv., online
0,2
0
0 1500 3000 4500 6000 7500 9000
TOS (min)
Figure 3.27: Conversion of concentrated stearic acid flow at 0.075 ml/min over 2 wt% Pd/Sibunit at 300℃ and 20
bar. TOS = 0-5760 min: 5% H2 /Ar; TOS = 5760-9000 min: Ar. Legend: Liquid mole balance (), (•), Stearic acid
(•), heptadecane (N), heptadecene (), undecylbenzene (×), CO (•), CO2 (•).
3.6.2 Ethyl stearate

The ethyl stearate liquid pumping was started at TOS = 0 min. The response in terms of heptadecane
was an initial yield of 90% after 3 h, decreasing quickly to 75% in the next few hours, as is shown
in Figure 3.28. In the beginning of the run about 2-3% unsaturated compounds in the form of 1-
heptadecene and undecylbenzene were formed from the ethyl stearate, although these decreased to
1% during the reaction in 5% H2 /Ar. The heptadecane formation stabilised at ca. 62% yield from
TOS = 800 min while the analysis of the ethyl stearate showed between 25% and 40% unconverted
ester - this varied considerably, as did the amount of stearic acid observed in 2% to 12% yield. This
111
led to a fluctuating mass balance between ca. 90% and 115%.

Initially, the gas analyser was offline for the first 2500 min TOS, so no information of gas compo-
sition is found for the first time interval. However, the gas analysis from TOS = 2500 min showed a
remarkably constant level of CO corresponding to 1.5% conversion, while it appeared that the CO2 -
concentration in the gas-phase leveled off at around 18-19% conversion level before TOS = 4500
min.
The gas atmosphere was changed from 5% H2 /Ar to pure Ar at TOS = 4500 min. and the gas
atmosphere was affected after about 20 min. A steep drop in CO2 concentration took place for
about an hour to a CO2 -level corresponding to practically 0% conversion. The CO-level fell more
moderately over a number of hours.
60 min after the change to Ar gas flow the heptadecane yield started to drop steeply as well, and
after another 60 min the yield reached 7% and after 120 min only 2%. 6 h after the gas change the
yield of heptadecane was below 1% and remained so for the remaining experimentation period. The
ethyl stearate responded to the change by increasing to a level over 100% within a few hours, and
between 110% and 130% was during the rest of the experiment. Mole balances over 100% are not
possible at steady-state, so this is clearly an inaccuracy in the sampling method, most likely from
using a too high response factor for the liquid concentrations from the GC-FID liquid analysis.
1,4
Ethyl stearate
1,2
1
1-heptadecene
Heptadecane
0,8
Molar fraction (X)
C17-heavy
Ethyl stearate
0,6
Stearic acid
Molar balance
0,4
CO conc. conv.
CO2 conc. conv.
0,2
0
0 1000 2000 3000 4000 5000 6000 7000 8000 9000
TOS (min)
Figure 3.28: Conversion of concentrated ethyl stearate flow at 0.075 ml/min over 2 wt% Pd/Sibunit at 300℃ and
20 bar. TOS = 0-4500 min: 5% H2 /Ar; TOS = 4500-9000 min: Ar. Legend: Liquid mole balance (), Ethyl stearate
(•), Stearic acid (•), heptadecane (N), heptadecene (), undecylbenzene (×), CO (•), CO2 (•).
Stearic acid was observed throughout the entire experimentation between 12% and 2% yield,
possibly as an intermediate of the conversion of ethyl stearate. Despite the catalyst being completely
inactive in the Ar atmosphere, about 4-5% stearic acid were still observed in the liquid samples from
the ractor. One explanation could be that the deoxygenation via stearic acid is faster than a potential
112
direct deoxygenation cleavage of the ester bond. This could happen through hydrolysis of the ester
with water vapor. However, there may be several reaction routes for the deoxygenation since stearic
acid was still formed (and not consumed to form C17 -hydrocarbons) during the period of pure Ar
atmosphere at TOS = 4500 - 9000 min.
3.6.3 Tristearin
Triglycerides are the major constituent of most fats and oils, however the analysis of the triglycerides
can be somewhat troublesome. Not many reports have appeared where an actual quantification
of the triglycerides themselves are performed. In the analysis performed here, the amounts from
the triglycerides calculated from the chromatograms by the cool-on-column GC-FID analysis were
too small to check to a molar balance of 100% and has therefore been omitted in the graphics of
Figure 3.29. Technical tristearin was used in the experimentation, which upon fatty acid analysis by
transesterification appeared to contain about 35% palmitic acid and 65% stearic acid as fatty acid
chains in the glycerides. Therefore, the products pentadecane (C15 ) and heptadecane (C17 ) can be
expected from decarbonylation and decarboxylation and hexadecane (C16 ) and octadecane (C18 ) via
complete reduction with H2 .
0,7
Technical Tristearine
0,6
C15
C16
0,5
C17
Mole fraction (X)
C18
0,4
CO, conv., online
CO2, conv., online
0,3
0,2
0,1
0
0 1500 3000 4500 6000 7500 9000
TOS (min)
Figure 3.29: Conversion of concentrated tristearin flow at 0.075 ml/min over 2 wt% Pd/Sibunit at 300℃ and
20 bar. TOS = 0-4500 min: 5% H2 /Ar; TOS = 4500-9240 min: Ar. Legend: Pentadecane (), hexadecane (×),
heptadecane (), octadecane (×), CO (•), CO2 (•).
After start-up of the flow to the reactor, the gas signal responded after 60 min TOS where the
CO2 -concentration peaked corresponding to ca. 21% yield and CO peaked at ca. 2.5%. The CO
level was almost constant for the remaining reaction i 5% H2 /Ar, while the CO2 -level dropped to 7%
conversion in the same time interval. The yields of heptadecane and pentadecane reached respec-
tively 60% and 25% after 150 min TOS, but both declined towards respectively 21% and 7% when
nearing TOS = 4500 min. Thus, the turnover rate of the catalyst bed had been reduced to 33% of the
113
immediate initial activity.

The gas flow was changed to pure Ar at TOS = 4500, and the CO/CO2 concentration responded
to this change immediately by dropping to 0, followed by a concentration peak after 40 min, and
a gradual decline over the next hours to a level below 1% conversion. The liquid production of
heptadecane and pentadecane declined after 120 min, leveling off and nearing zero in the follow-
ing hours. During the remaining experimentation until shutdown at TOS = 9240 min no further
conversion to alkanes was detected.
3.6.4 Comparison of reactants

It is evident that the shift to pure Ar atmosphere swiftly led to deactivation in all of the liquid feeds
applied. Also, in all of the deoxygenation reactions in hydrogen-containing atmosphere, the yields
of liquid hydrocarbons produced did not corresponnde to the CO- and CO2 -levels in the gas effluent
- methanation most likely consumed a large part of the COx formed. It appears that the reactor
still produced a few percent of heptadecane in the pure Ar gas flow with stearic acid as a reactant,
perhaps due to a thermal mechanism or catalysis by the support as described in section 3.5.4. This
behaviour is not seen with the ester compounds.
Interestingly, the gas atmosphere during the stearic acid experiment showed a gradual rise of
CO and decline of CO2 formation in H2 -containing gas flow. Ethyl stearate and tristearin yielded
primarily CO2 and a constant level of CO during reaction in H2 /Ar atmosphere, which is in contrast
to the results reported by Resasco et al. [210], who reacted methyl stearate over Pt/γ-Al2 O3 in a
semibatch reactor and primarily detected CO formation in the gas atmosphere during heptadecane
formation in both H2 and He atmosphere at 325℃ [210]. This difference may have to do with the
difference in catalyst metal or support or the fact that different reactor- and catalyst types were used.
The deoxygenation of stearic acid and tripalmitin have minor peaks of CO and CO2 after chang-
ing the gas atmosphere pure Ar gas, and a few hours “extra” catalyst lifetime with formation of
mostly saturated, but also unsatuated hydrocarbon products can be observed. It could be suspected
that part of the mechanism requires removal of H2 to the sweeping gas - such as the formic acid for-
mation/decomposition mechanism proposed by Berenblyum and co-authors [322], where the formic
acid must decompose to H2 and CO2 . Thus, H2 is both needed as a reactant to avoid formation of
unsaturated compounds and deactivation, but also inhibiting part of the mechanism or leading to
formation of CO, as observed by Immer et al. [321].
Ethyl stearate and tristearin yielded identical production levels of CO (despite different conver-
sions of the reactants), and it may be speculated that the same gas-producing reactions takes place.
Interestingly, varying minor amounts of stearic acid were obtained during the entire experimentation
with ethyl stearate as a possible part of the deoxygenation reaction mechanism, but during the run
in technical tristearin no palmitic or stearic acid were observed. Further careful studies to define the
reactions of the gases during deoxygenation would be highly desirable. The influence of gas pressure
and content is far from understood, as is the influence of methanation, water-gas-shift reactions and
the role of CO as a possible poison for the catalyst. This could for instance be tracked by measuring
methane, by mass spectroscopy using isotopic labeling of gases and reactants, or by dosing various
114
gases to the reactor with and without reactants to measure the influence on the liquid and gas phases.
Neither stearic acid nor ethyl stearate yielded any products from full reduction of the fatty acid
chain (i.e. octadecane). However, technical tristearin yielded 2-3% even-carbon-numbered alkanes
during the reaction in hydrogen-containing atmosphere, decreasing to about 1% during the period
of reaction with 5% H2 /Ar gas. It may be speculated that these derived impurities of C19 and C17
fatty acids, but they were not found during analysis by transesterification, thus the octadecane and
hexadecane observed must derive from stearic and palmitic acid. This is somewhat contrary to the
assumption that ethyl stearate and tristearin should have reacted via identical reaction routes, thus
the reaction network was more complex. No unsaturated compounds were observed during the en-
tire treatment of tristearin, but minor amounts of 1-heptadecene and monoaromatic C17 -compounds
were observed for both stearic acid and ethyl stearate, especially during start-up and gas atmosphere
change.
Both the stearic acid and ethyl stearate experimentation initially had a yield of around 100%
of formed alkanes, but the yield of the former was 75% at TOS = 4500 min in contrast to 62% of
the latter. However the tristearin initially yielded about 85% alkanes, which dropped to as low as
28% at 4500 min. All the molecules were supplied at the same liquid volume feeding rate, 0.075
ml/min at a pump and tubing temperature of 100℃, and as they have about the same densities the
feeding of carbon is also more of less the same. Thus it can be stated that the catalyst deactivated
faster during tristearin than ethyl stearate deoxygenation, which again deactivated faster than dur-
ing stearic acid deoxygenation. This supports a “the simpler, the better”-approach regarding the
feedstock molecules during the deoxygenation - the more complicated and hindered the molecule,
the faster the deactivation of the catalyst.
An assessment of the degree of deactivation was made by BET measurements of the batches of
the spent catalyst beads from the three different feeds by comparison with the fresh catalyst as seen
in Figure 3.30. Here it is evident that the catalyst beads used for ethyl stearate deoxygenation have
endured the least deactivation by loss of surface area compared to the fresh catalyst (36% surface
area lost on average), tristearin somewhat more (44% surface area lost) and stearic acid the most
(46% surface area lost). This order is different from the order of the degree of deactivation of the
catalysts mentioned above. The picture of deactivation must therefore be refined to include that var-
ious feedstocks may be more or less tolerant to coking of the catalyst during deoxygenation - or other
deactivation phenomena may have taken place at the same time, which have not been investigated.
This possibly had to do with different accessibility on active Pd sites for the two esters and the fatty
acid, especially on the partly deactivated catalysts. Coking or buildup of reactants or products is the
reason behind the shrinking of the internal pore surface area.
The surface area distribution of the spent catalyst beads in Figure 3.30 was also surprising and
somewhat different compared to the clear deactivation profile described in section 3.5.3 and Table
3.5. However, the variation of surface areas was bigger during long-term experimentation of stearic
acid compared to the areas given in Figure 3.30, which are also bigger. The stearic acid and tristearin
surface areas were biggest in the middle of the catalysts beds, so that deactivation by pore clogging
primarily took place in the top and bottom of the catalyst bed. The ethyl stearate conversion resulted
115
a deactivation profile where the areas decreased downwards in the bed, completely contrary to the
deactivation mentioned in Table 3.5. Generally this behaviour was unexpected, but possibly small
temperature differences depending on axial bed position have led to these surface area distributions.
400
Spent catalyst surface areas
350
1st batch
Surface area (m²/g) 300
2nd batch
250
3rd batch
200
4th batch
150
5th batch
100
50
0
Stearic acid Ethyl stearate Tristearin Fresh cat.
Figure 3.30: Surface area of the catalyst for stearic acid, ethyl stearate and tristearin deoxygenation. Measured by
N2 physisorption and calculated with the BET method. The batches are numbered by position downwards in the
bed, i.e. ascending order away from the inlet.
116
3.7 Hydrodeoxygenation of fats and oils: Conclusion
The supported transition-metal catalysed deoxygenation of fatty compounds have been investigated
in both batch and continuous reactor.
Tripalmitin and oleic acid dissolved in n-tetradecane were investigated for hydrodeoxygenation
over noble-metal catalysts in an autoclave as a batch reactor. Hydrogen pressure and temperature
was varied between 0 and 40 bar H2 and 250-375℃ over 5 wt% Pt/γ-Al2 O3 . Furthermore a number
of metals were tested as active materials for the deoxygenation. Part of the triglycerides are split to
FFA before deoxygeation.
Oleic acid was converted to alkanes faster than tripalmitin over 5 wt% Pt/γ-Al2 O3 , possibly due
to less sterical hindrance. The deoxygenation activity below 300℃ was low and resulted in yields
below 10% after 5 h reaction at 20 bar H2 , but all of the oleic was converted to alkane at 325℃
after 5 h - this was only reached for tripalmitin at 375℃. The optimum pressure of both tripalmitin
and oleic acid at 325℃ was around 10-20 bar of H2 - higher pressures of H2 resulted lower yields
of alkanes while no H2 in the gas atmosphere led to the formation of unsaturated hydrocarbons.
Higher H2 -pressures as well yielded more even-carbon-numbered alkanes by complete reduction of
the fatty acid chains.
The best catalysts amongst Ni, Rh, Ru, Pd, Pt, and Au supported on γ-Al2 O3 was found to be Pt
and Pd, as the others tested were not as active and usually not as selective towards decarboxylation
and decarbonylation as Pd and Pt. Pt was more active for the oleic acid conversion than Pd, but the
opposite was the case for tripalmitin. The surface area Pt on γ-Al2 O3 were only half of that of the
corresponding Pd. γ-Al2 O3 was not active for the deoxygenation alone. Catalytic methanation result
from the carbonaceous gases formed in the H2 -rich gas, and this reaction should be avoided in an
industrial system.
Stearic acid was deoxygenated in a continuous packed-bed reactor at 300℃ at 20 bar over 2
wt% Pd on Sibunit carbon during a catalytic run of almost 2 weeks, employing various step changes.
Almost complete conversion of dilute stearic acid to heptadecane was achieved in 5% H2 /Ar with the
fresh catalyst, but switching to Ar led to formation of unsaturated products, decreasing conversion
and a switch to formation of CO gas. Concentrated stearic acid was converted almost exclusively to
heptadecane in both 5% H2 /Ar, but the conversion was several times higher (12%) with hydrogen in
the gas compared to deoxygenation without hydrogen. Thus, H2 was necessary to avoid the gradual
deactivation of the catalyst in diluted stearic acid in dodecane and yielded higher conversion of the
fatty acid.
A gradual deactivation profile was observed in the catalyst bed as 72% of the original catalyst
surface area was lost near the reactor inlet, while only 54% was lost in the bottom. The deactivation
and coking reactions were more pronounced in the inlet zone of the reactor compared to the outlet.
Deactivation took place by coking while leaching, sintering and agglomeration could be excluded.
The response behaviour during step changes, especially the liberation of CO and the considerable
number of hours necessary before the liquid samples were free from stearic acid revealed that the
porous catalyst beads in the bed absorbed the stearic acid and released it over a number of hours in
117
3.7. Hydrodeoxygenation of fats and oils: Conclusion
inert liquid flow. Thus, stearic acid either diffused very slowly in the pores, perhaps due to capillary
forces, justifying the use of a catalyst with eggshell-impregnation of the active metal phase. The
treatment of stearic acid over Sibunit beads without Pd or added H2 yielded about 5-6% conversion
of the stearic acid primarily to heptadecane in Ar atmosphere. CO was produced as the main gaseous
product, while no or very little CO2 was observed from this experiment. This may hint that CO pro-
duction during transition-metal-catalysed deoxygenation is caused by a deactivated catalyst where
only a thermal or support effect is responsible for the conversion to alkanes.
The reactivities of FFA, ethyl and triester over 2 wt% Pd/Sibunit in continuous deoxygenation
were assessed under 5% H2 /Ar, and after 3 days TOS the conversion to alkanes was 75% with stearic
acid, 62% with ethyl stearate, and 27% with technical tristearin as the liquid feed molecules. Ethyl
stearate and tristearin conversion yielded primarily CO2 in the gas phase and a low and constant
amount of CO, while the gas composition during stearic acid deoxygenation steadily changed from
primarily CO2 as gaseous product to an equimolar mixture of CO and CO2 . When the gas flow was
switched to Ar, the catalyst deactivated fast with all of the three feedstocks. The stearic acid sustained
the slowest deactivation after switch to pure Ar gas flow compared to ethyl stearate and tristearin.
Technical tristearine yielded minor amounts of C18 - and C16 -alkane through full reduction of the
fatty acid chains, which was otherwise not observed with ethyl stearate and stearic acid.
The excellent selectivities of transition metals towards the decarboxylation or decarbonylation
pathways during deoxygenation of fatty acids and triglycerides make these catalysts promient for
this reaction. However, the hydrogen consumption, methane formation and lifetime of the sup-
ported noble-metal catalysts is still below that of sulphided hydrotreatment catalysts, and further
investigations and developments are required to increase catalyst lifetime and avoid side reactions
and deactivation for the noble-metal catalysts.
Although palladium and platinum have been found active in the present work, alternatives ought
to be found as Pt and Pd are scarce metals and expensive. However, rational and sparse use of the
metals could be justified for deoxygenation reactions, but this requires a profound comprehension of
the reaction involved, including mechanisms and selectivity for various liquid products, gas product
selectivities, side reactions, stable behaviour in long-term experimentation, and deactivation. Cata-
lysts used in industrial settings should be thoroughly investigated according to these parameters.
The current study may add to this.
118
Chapter 4
Outlook and Concluding Remarks 4

Although the deposits of easily accessible fossil fuels may last for longer than projected at present,
it is important from both an economic and environmental point of view to develop alternative fuel
technologies with a lower impact on the environment and less susceptible to fluctuating oil prices.
Solar, wind, wave, or geothermal power plants are installed in increasing numbers, but although
their usage is projected to rise dramatically in the coming years they are not projected to gain a sig-
nificantly higher share of the global energy consumption - as the growing global energy demand is
also being covered by fossil energy resources. Nuclear power may as well contribute to the produc-
tion of CO2 -neutral energy, but the nuclear waste problem is at best only partly solved.
Electrical energy cannot be stored economically in large amounts with the current technology.
And despite the promising perspective of a hydrogen economy no competitive system for using H2 as
a standard vehicle fuel exists yet, mostly due to the difficulties of storing the gas. Thus, the chemical
energy density remains the determining factor for the energy system in the transport sector, and it
is difficult to imagine this to change from hydrocarbons in the near future. And hydrocarbons is still
the only option for instance for aviation fuels and production of bulk chemicals.
The sinister forecasts of our addiction to oil, gas and coal puts an increasing pressure on us to
develop viable alternatives to the fossil resources and turn them into a green, sustainable industry.
This puts biomass in a central position in a shift away from fossil feedstocks, as biomass is the sole
applicable renewable carbonaceous resource at present. The production of biofuels, bioenergy and
biochemicals may help to solve the problems of increasing global resource consumption, warming
climate and deterioration of ecosystems around the globe.
The purpose of this thesis has been to contribute to the investigations and development of cata-
lysts for production of diesel fuel from biomass, especially with focus on methods employing waste
oils and fats as raw materials. A handful of technologies exist that potentially may yield diesel fuel,
however only fatty acid methyl esters (FAME) and alkanes via hydrodeoxygenation (HDO), both
from fats and oils, have yet been introduced on the diesel market in a larger scale. Catalysis will
most likely play a key role in enabling the conversion of any biomass to sustainable transportation
biofuels, and a profound basic and technical understanding of the catalytic reactions will be essential
for making the processes economical and efficient.
Biomass-based diesel and biofuels as such may contribute to meet the immense global challenge
of finding sustainable energy for the ever-increasing demand from the transportation sector. It is,
however, of equally immense importance that this biofuel is produced sustainably, for instance by
using waste agricultural and industrial products, and that the production does not foster bigger
ecological problems than it solves.
119
Chapter 5
References 5
[1] International Energy Agency. World Energy Outlook. International Energy Agency, Paris, France, (2010).
[2] Forster, P., Ramaswamy, V., Artaxo, P., Berntsen, T., Betts, R., Fahey, D., Haywood, J., Lean, J., Lowe,
D., Myhre, G., Nganga, J., Prinn, R., Raga, G., Schulz, M., and Van Dorland, R. Changes in Atmospheric
Constituents and in Radiative Forcing, chapter 2, 131–234. Cambridge University Press, Cambridge,
United Kingdom, 1 edition (2007).
[3] International Energy Agency (IEA). Key World Energy Statistics. International Energy Agency, IEA Pub-
lications, 9 rue de la Fédération, 75739 Paris Cedex 15, France, (2006).
[4] Thomas, J. M., Hernandez-Garrido, J. C., and Bell, R. G. Top. Catal. 52, 1630–1639 (2009).
[5] The European Union. J. Eur. Union 12, 1–20 (2003).
[6] Commission of the European Communities. Communication: “Renewable Energy Road Map. Renewable
energies in the 21st century: building a more sustainable future” COM(2006) 848 final (2007).
[7] The European Union. J. Eur. Union 140, 1–47 (2009).
[8] Nørskov, J. K., Bligaard, T., Rossmeisl, J., and Christensen, C. H. Nat. Chem. 1, 37–46 (2009).
[9] Thomas, J. M. and Thomas, W. J. Principles and practice of Heterogeneous Catalysis. VCH Verlagsge-
sellschaft mbH, Weinheim, Germany, 1st edition (1997).
[10] Ma, F. and Hanna, M. A. Bioresour. Technol. 70, 1–15 (1999).
[11] Van Gerpen, J. Fuel Process. Technol. 86, 1097–1107 (2005).
[12] Knothe, G., Van Gerpen, J., and Krahl, J. The Biodiesel Handbook. AOCS Press, Urbana, Illinois, USA, 1
edition, (2005).
[13] Meher, L., Vidya Sagar, D., and Naik, S. Renew. Sustain. Energy Rev. 10, 248–268 (2006).
[14] Mittelbach, M. and Remschmidt, C. Biodiesel: The Comprehensive Handbook. Martin Mittelbach, Graz,
Austria, 3 edition, (2006).
[15] Knothe, G. Top. Catal. 53, 714–720 (2010).
[16] Kubičková, I., Snåre, M., Eränen, K., Mäki-Arvela, P., and Murzin, D. Y. Catal. Today 106, 197–200
(2005).
[17] Snåre, M., Kubičková, I., Mäki-Arvela, P., Eränen, K., and Murzin, D. Y. Ind. Eng. Chem. Res. 45, 5708–
5715 (2006).
[18] Huber, G. W., OĆonnor, P., and Corma, A. Appl. Catal., A 329, 120–129 (2007).
[19] Huber, G. W. and Corma, A. Angew. Chem., Int. Ed. 46, 7184–7201 (2007).
[20] Rozmysłowicz, B., Mäki-Arvela, P., Lestari, S., Simakova, O. A., Eränen, K., Simakova, I., Murzin, D. Y.,
and Salmi, T. Top. Catal. 53, 1274–1277 (2010).
[21] Guzman, A., Torres, J. E., Prada, L. P., and Nuñez, M. L. Catal. Today 156, 38–43 (2010).
[22] Šimáček, P. and Kubička, D. Fuel 89, 1508–1513 (2010).
[23] Simakova, I., Simakova, O., Mäki-Arvela, P., and Murzin, D. Y. Catal. Today 150, 28–31 (2010).
[24] Kikhtyanin, O. V., Rubanov, A. E., Ayupov, A. B., and Echevsky, G. V. Fuel 89, 3085–3092 (2010).
[25] Twaiq, F. A., Zabidi, N. A. M., Mohameda, A. R., and Bhatia, S. Fuel Process. Technol. 84, 105–120 (2003).
[26] Tamunaidu, P. and Bhatia, S. Bioresour. Technol. 98, 3593–3601 (2007).
[27] Na, J.-G., Yi, B. E., Kim, J. N., Yi, K. B., Park, S.-Y., Park, J.-H., Kim, J.-N., and Ko, C. H. Catal. Today 156,
121
44–48 (2009).
[28] Simonetti, D. and Dumesic, J. Catal. Rev. - Sci. Eng. 51, 441–484 (2009).
[29] Huber, G. W., Chheda, J. N., Barrett, C. J., and Dumesic, J. A. Science 308, 1446–1450 (2005).
[30] Chheda, J. N., Huber, G. W., and Dumesic, J. A. Angew. Chem., Int. Ed. 46, 7164–7183 (2007).
[31] Serrano-Ruiz, J. C., Wang, D., and Dumesic, J. A. Green Chem. 12, 574–577 (2010).
[32] Rostrup-Nielsen, J. R. Catal. Rev. - Sci. Eng. 46, 247–270 (2004).
[33] Rostrup-Nielsen, J. R. Science 308, 1421–1422 (2005).
[34] Abatzoglou, N., Dalai, A., and Gitzhofer, F. In 3rd IASME/WSEAS international conference on energy,
environment, ecosystems and sustainable development, Greece, 223–232, (2007).
[35] Zhang, W. Fuel Process. Technol. 91, 866–876 (2010).
[36] Stöcker, M. Angew. Chem., Int. Ed. 47, 9200–9211 (2008).
[37] Carlson, T. R., Jae, J., Lin, Y.-C., Tompsett, G. A., and Huber, G. W. J. Catal. 270, 110–124 (2010).
[38] Wildschut, J., Mahfud, F. H., Venderbosch, R. H., and Heeres, H. J. Ind. Eng. Chem. Res. 48, 10324–10334
(2009).
[39] Laurent, E. and Delmon, B. J. Catal. 146, 281–291 (1994).
[40] Zhang, S., Yan, Y., Li, T., and Ren, Z. Bioresour. Technol. 96, 545–550 (2005).
[41] Donnis, B., Egeberg, R. G., Blom, P., and Knudsen, K. G. Top. Catal. 52, 229–240 (2009).
[42] Knothe, G. J. Am. Oil Chem. Soc. 83, 823–833 (2006).
[43] Koskinen, M., Sourander, M., and Nurminen, M. Hydrocarb. Process. 85, 81–86 (2006).
[44] Snåre, M. and Murzin, D. Y. Ind. Eng. Chem. Res. 45, 6875–6875 (2006).
[45] Mikkonen, S. Hydrocarb. Process. 87, 63–66 (2008).
[46] Srivastava, A. and Prasad, R. Renew. Sustain. Energy Rev. 4, 111–133 (2000).
[47] Huber, G. W., Iborra, S., and Corma, A. Chem. Rev. 106, 4044–4098 (2006).
[48] Lotero, E., Liu, Y., Lopez, D. E., Suwannakarn, K., Bruce, D. A., and Goodwin, J. G. Ind. Eng. Chem. Res.
44, 5353–5363 (2005).
[49] Jacobson, K., Gopinath, R., Meher, L. C., and Dalai, A. K. Appl. Catal., B 85, 86–91 (2008).
[50] Casanave, D., Duplan, J.-L., and Freund, E. Pure Appl. Chem. 79, 2071–2081 (2007).
[51] Sivasamy, A., Cheah, K., Fornasiero, P., Kemausuor, F., Zinoviev, S., and Miertus, S. ChemSusChem 2,
278–300 (2009).
[52] Demirbas, A. Prog. Energ. Combust. 31, 466–487 (2005).
[53] Verziu, M., El Haskouri, J., Beltran, D., Amoros, P., Macovei, D., Gheorghe, N. G., Teodorescu, C. M.,
Coman, S. M., and Parvulescu, V. I. Top. Catal. 53, 763–772 (2010).
[54] Mo, X., Lotero, E., Lu, C., Liu, Y., and Goodwin, J. G. Catal. Lett. 123, 1–6 (2008).
[55] Liu, X., He, H., Wang, Y., and Zhu, S. Catal. Commun. 8, 1107–1111 (2007).
[56] Sree, R., Seshubabu, N., Saiprasad, P., and Lingaiah, N. Fuel Process. Technol. 90, 152–157 (2009).
[57] Kozlowski, J. T., Aronson, M. T., and Davis, R. J. Appl. Catal., B 96, 508–515 (2010).
[58] Caland, L. B., Santos, L. S. S., Moura, C. V. R., and Moura, E. M. Catal. Lett. 128, 392–400 (2008).
[59] Bournay, L., Casanave, D., Delfort, B., Hillion, G., and Chodorge, J. Catal. Today 106, 190–192 (2005).
[60] Pugnet, V., Maury, S., Coupard, V., Dandeu, a., Quoineaud, A.-A., Bonneau, J.-L., and Tichit, D. Appl.
Catal., A 374, 71–78 (2010).
[61] Russbueldt, B. M. and Hoelderich, W. F. J. Catal. 271, 290–304 (2010).
[62] Kawashima, A., Matsubara, K., and Honda, K. Bioresour. Technol. 99, 3439–3443 (2008).
[63] Shu, Q., Yang, B., Yuan, H., Qing, S., and Zhu, G. Catal. Commun. 8, 2158–2164 (2007).
[64] Macario, A., Giordano, G., Onida, B., Cocina, D., Tagarelli, A., and Giuffrè, A. M. Appl. Catal., A 378,
160–168 (2010).
[65] Li, E. and Rudolph, V. Energy Fuels 22, 145–149 (2008).
122
Chapter 5. References
[66] Georgogianni, K. G., Katsoulidis, A. P., Pomonis, P. J., and Kontominas, M. G. Fuel Process. Technol. 90,
671–676 (2009).
[67] Georgogianni, K. G., Katsoulidis, A. K., Pomonis, P. J., Manos, G., and Kontominas, M. G. Fuel Process.
Technol. 90, 1016–1022 (2009).
[68] Zeng, H., Feng, Z., Deng, X., and Li, Y. Fuel 87, 3071–3076 (2008).
[69] Cantrell, D., Gillie, L., Lee, A., and Wilson, K. Appl. Catal., A 287, 183–190 (2005).
[70] Liu, Y., Lotero, E., Goodwin Jr, J. G., and Mo, X. Appl. Catal., A 331, 138–148 (2007).
[71] Brito, A., Borges, M. E., Garín, M., and Hernández, A. Energy Fuels 23, 2952–2958 (2009).
[72] Kim, M. J., Park, S. M., Chang, D. R., and Seo, G. Fuel Processing Technology 91, 618–624 (2010).
[73] Chuayplod, P. and Trakarnpruk, W. Ind. Eng. Chem. Res. 48, 4177–4183 (2009).
[74] Macala, G. S., Robertson, A. W., Johnson, C. L., Day, Z. B., Lewis, R. S., White, M. G., Iretskii, A. V., and
Ford, P. C. Catal. Lett. 122, 205–209 (2008).
[75] Li, E., Xu, Z. P., and Rudolph, V. Appl. Catal., B 88, 42–49 (2009).
[76] Guerreiro, L., Pereira, P., Fonseca, I., Martin-Aranda, R., Ramos, A., Dias, J., Oliveira, R., and Vital, J.
Catal. Today 156, 191–197 (2010).
[77] Schuchardt, U. J. Mol. Catal. A: Chem. 99, 65–70 (1995).
[78] Schuchardt, U. J. Mol. Catal. A: Chem. 109, 37–44 (1996).
[79] Villa, A., Tessonnier, J.-P., Majoulet, O., Su, D. S., and Schlögl, R. Chem. Commun. , 4405–4407 (2009).
[80] Cerro-Alarcón, M., Corma, A., Iborra, S., Martínez, C., and Sabater, M. J. Appl. Catal., A 382, 36–42
(2010).
[81] Cerro-Alarcón, M., Corma, A., Iborra, S., and Gómez, J. P. Appl. Catal., A 346, 52–57 (2008).
[82] Kim, M.-Y., Seo, G., Kwon, O. Z., and Chang, D. R. Chem. Commun. , 3110–3112 (2009).
[83] Saravanamurugan, S., Han, D., Koo, J., and Park, S. Catal. Commun. 9, 158–163 (2008).
[84] Savonnet, M., Aguado, S., Ravon, U., Bazer-Bachi, D., Lecocq, V., Bats, N., Pinel, C., and Farrusseng, D.
Green Chem. 11, 1729–1732 (2009).
[85] Liu, Y., Lotero, E., Goodwin Jr, J. G., and Lu, C. J. Catal. 246, 428–433 (2007).
[86] Sunita, G., Devassy, B. M., Vinu, A., Sawant, D. P., Balasubramanian, V., and Halligudi, S. Catal. Com-
mun. 9, 696–702 (2008).
[87] Park, Y.-M., Lee, J. Y., Chung, S.-H., Park, I. S., Lee, S.-Y., Kim, D.-K., Lee, J.-S., and Lee, K.-Y. Bioresour.
Technol. 101 Suppl, S59–61 (2010).
[88] Komintarachat, C. and Chuepeng, S. Ind. Eng. Chem. Res. 48, 9350–9353 (2009).
[89] Rao, K. N., Sridhar, A., Lee, A. F., Tavener, S. J., Young, N. A., and Wilson, K. Green Chem. 8, 790–797
(2006).
[90] Ngaosuwan, K., Mo, X., Goodwin Jr, J. G., and Praserthdam, P. Appl. Catal., A 380, 81–86 (2010).
[91] Kulkarni, M. G., Gopinath, R., Meher, L. C., and Dalai, A. K. Green Chem. 8, 1056 (2006).
[92] Zhang, X., Li, J., Chen, Y., Wang, J., Feng, L., Wang, X., and Cao, F. Energy Fuels 23, 4640–4646 (2009).
[93] Morin, P., Hamad, B., Sapaly, G., Carneiro Rocha, M., Pries de Oliveira, P., Gonzalez, W., Andrade Sales,
E., and Essayem, N. Appl. Catal., A 330, 69–76 (2007).
[94] Caetano, C., Fonseca, I., Ramos, A., Vital, J., and Castanheiro, J. Catal. Commun. 9, 1996–1999 (2008).
[95] Oliveira, C. F., Dezaneti, L. M., A.C. Garcia, F., de Macedo, J. L., Dias, J. A., Dias, S. C., and Alvim, K. S.
Appl. Catal., A 372, 153–161 (2010).
[96] Cardoso, A. L., Augusti, R., and Silva, M. J. J. Am. Oil Chem. Soc. 85, 555–560 (2008).
[97] Godói Silva, V. W., Laier, L. O., and Silva, M. J. D. Catal. Lett. 135, 207–211 (2010).
[98] Xu, L., Wang, Y., Yang, X., Yu, X., Guo, Y., and Clark, J. H. Green Chem. 10, 746–755 (2008).
[99] Srilatha, K., Lingaiah, N., Devi, B. P., Prasad, R., Venkateswar, S., and Prasad, P. S. Appl. Catal., A 365,
28–33 (2009).
123
[100] Xu, L., Wang, Y., Yang, X., Hu, J., Li, W., and Guo, Y. Green Chem. 11, 314–317 (2009).
[101] Alsalme, A., Kozhevnikova, E. F., and Kozhevnikov, I. V. Appl. Catal., A 349, 170–176 (2008).
[102] Lam, M. K., Lee, K. T., and Mohamed, A. R. Appl. Catal., B 93, 134–139 (2009).
[103] Suwannakarn, K., Lotero, E., Goodwin Jr, J. G., and Lu, C. J. Catal. 255, 279–286 (2008).
[104] Hu, X., Zhou, Z., Sun, D., Wang, Y., and Zhang, Z. Catal. Lett. 133, 90–96 (2009).
[105] Du, Y., Liu, S., Ji, Y., Zhang, Y., Wei, S., Liu, F., and Xiao, F.-S. Catal. Lett. 124, 133–138 (2008).
[106] López, D. E., Goodwin Jr, J. G., Bruce, D. A., and Furuta, S. Appl. Catal., A 339, 76–83 (2008).
[107] Rattanaphra, D., Harvey, A., and Srinophakun, P. Top. Catal. 53, 773–782 (2010).
[108] Yu, G., Zhou, X., Li, C., Chen, L., and Wang, J. Catal. Today 148, 169–173 (2009).
[109] de Almeida, R. M., Noda, L. K., Gonçalves, N. S., Meneghetti, S. M., and Meneghetti, M. R. Appl. Catal.,
A 347, 100–105 (2008).
[110] Juan, J. C., Zhang, J., and Yarmo, M. A. Appl. Catal., A 332, 209–215 (2007).
[111] Shi, W., He, B., Ding, J., Li, J., Yan, F., and Liang, X. Bioresour. Technol. 101, 1501–5 (2010).
[112] Chung, K.-H., Chang, D.-R., and Park, B.-G. Bioresour. Technol. 99, 7438–43 (2008).
[113] Carmo Jr, A., Desouza, L., Dacosta, C., Longo, E., Zamian, J., and da Rocha Filho, G. Fuel 88, 461–468
(2009).
[114] Jiménez-Morales, I., Santamaría-González, J., Maireles-Torres, P., and Jiménez-López, A. Appl. Catal., A
379, 61–68 (2010).
[115] Wu, S., Liu, P., Leng, Y., and Wang, J. Catal. Lett. 132, 500–505 (2009).
[116] Dhainaut, J., Dacquin, J.-P., Lee, A. F., and Wilson, K. Green Chem. 12, 296 (2010).
[117] Shibasaki-Kitakawa, N., Tsuji, T., Chida, K., Kubo, M., and Yonemoto, T. Energy Fuels 24, 3634–3638
(2010).
[118] Bianchi, C. L., Boffito, D. C., Pirola, C., and Ragaini, V. Catal. Lett. 134, 179–183 (2009).
[119] Özbay, N., Oktar, N., and Tapan, N. Fuel 87, 1789–1798 (2008).
[120] Park, J.-Y., Kim, D.-K., and Lee, J.-S. Bioresour. Technol. 101, S62–S65 (2010).
[121] Feng, Y., He, B., Cao, Y., Li, J., Liu, M., Yan, F., and Liang, X. Bioresour. Technol. 101, 1518–1521 (2010).
[122] Caetano, C., Guerreiro, L., Fonseca, I., Ramos, A., Vital, J., and Castanheiro, J. Appl. Catal., A 359, 41–46
(2009).
[123] Grossi, C. V., Jardim, E. D. O., Araújo, M. D., Lago, R. M., and Silva, M. J. D. Fuel 89, 257–259 (2010).
[124] Soldi, R. A., Oliveira, A. R., Ramos, L. P., and César-Oliveira, M. A. F. Appl. Catal., A 361, 42–48 (2009).
[125] Okayasu, T., Saito, K., Nishide, H., and Hearn, M. T. W. Chem. Commun. , 4708–4710 (2009).
[126] Melero, J. A., Bautista, L. F., Iglesias, J., Morales, G., Sánchez-Vázquez, R., and Suárez-Marcos, I. Top.
Catal. 53, 795–804 (2010).
[127] Zieba, A., Drelinkiewicz, A., Konyushenko, E., and Stejskal, J. Appl. Catal., A 383, 169–181 (2010).
[128] Toda, M., Takagaki, A., Okamura, M., Kondo, J. N., Hayashi, S., Domen, K., and Hara, M. Nature 438,
178 (2005).
[129] Suganuma, S., Nakajima, K., Kitano, M., Yamaguchi, D., Kato, H., Hayashi, S., and Hara, M. J. Am.
Chem. Soc. 130, 12787–12793 (2008).
[130] Hara, M. Top. Catal. 53, 805–810 (2010).
[131] Dehkhoda, A. M., West, A. H., and Ellis, N. Appl. Catal., A 382, 197–204 (2010).
[132] Takagaki, A., Toda, M., Okamura, M., Kondo, J. N., Hayashi, S., Domen, K., and Hara, M. Catal. Today
116, 157–161 (2006).
[133] Lien, Y.-S., Hsieh, L.-S., and Wu, J. C. S. Ind. Eng. Chem. Res. 49, 2118–2121 (2010).
[134] Nakajima, K., Okamura, M., Kondo, J. N., Domen, K., Tatsumi, T., Hayashi, S., and Hara, M. Chem.
Mater. 21, 186–193 (2009).
[135] Peng, L., Philippaerts, A., Ke, X., Van Noyen, J., De Clippel, F., Van Tendeloo, G., Jacobs, P. A., and Sels,
124
B. F. Catal. Today 150, 140–146 (2010).

[136] Wu, Q., Chen, H., Han, M., Wang, D., and Wang, J. Ind. Eng. Chem. Res. 46, 7955–7960 (2007).
[137] Lapis, A., de Oliveira, L., Neto, B., and Dupont, J. ChemSusChem 1, 759–762 (2008).
[138] Ha, S. H., Lan, M. N., and Koo, Y.-M. Enzyme Microb. Technol. 47, 6–10 (2010).
[139] DaSilveira Neto, B., Alves, M., Lapis, A., Nachtigall, F., Eberlin, M., Dupont, J., and Suarez, P. J. Catal.
249, 154–161 (2007).
[140] Han, M., Yi, W., Wu, Q., Liu, Y., Hong, Y., and Wang, D. Bioresour. Technol. 100, 2308–2310 (2009).
[141] Li, K.-X., Chen, L., Yan, Z.-C., and Wang, H.-L. Catal. Lett. 139, 151–156 (2010).
[142] Liang, X. and Yang, J. Green Chem. 12, 201–204 (2010).
[143] Nassreddine, S., Karout, A., Lorraine Christ, M., and Pierre, A. C. Appl. Catal., A 344, 70–77 (2008).
[144] Sim, J. H., Kamaruddin, A. H., and Bhatia, S. J. Am. Oil Chem. Soc. 87, 1027–1034 (2010).
[145] Shakeri, M. and Kawakami, K. Catal. Commun. 10, 165–168 (2008).
[146] Dizge, N., Aydiner, C., Imer, D. Y., Bayramoglu, M., Tanriseven, A., and Keskinler, B. Bioresour. Technol.
100, 1983–1991 (2009).
[147] Keng, P. S., Basri, M., Ariff, A. B., Abdul Rahman, M. B., Abdul Rahman, R. N. Z., and Salleh, A. B.
Bioresour. Technol. 99, 6097–6104 (2008).
[148] Fu, B. and Vasudevan, P. T. Energy Fuels 23, 4105–4111 (2009).
[149] Xiao, M., Mathew, S., and Obbard, J. P. GCB Bioenergy 1, 115–125 (2009).
[150] Hernández-Martín, E. and Otero, C. Bioresour. Technol. 99, 277–286 (2008).
[151] Warabi, Y. Bioresour. Technol. 91, 283–287 (2004).
[152] Yin, J., Xiao, M., and Song, J. Energy Convers. Manage. 49, 908–912 (2008).
[153] Hawash, S., Kamal, N., Zaher, F., Kenawi, O., and Diwani, G. Fuel 88, 579–582 (2009).
[154] Hegel, P., Mabe, G., Pereda, S., and Brignole, E. A. Ind. Eng. Chem. Res. 46, 6360–6365 (2007).
[155] Saka, S., Isayama, Y., Ilham, Z., and Jiayu, X. Fuel 89, 1442–1446 (2010).
[156] Christoph, R., Schmidt, B., Steinberner, U., Dilla, W., and Karinen, R. “Glycerol”. Ullmann’s Encyclopedia
of Industrial Chemistry. Wiley-VCH Verlag GmbH & Co, 7th edition (2006).
[157] Zhou, C.-H. C., Beltramini, J. N., Fan, Y.-X., and Lu, G. Q. M. Chem. Soc. Rev. 37, 527–549 (2008).
[158] Pagliaro, M., Ciriminna, R., Kimura, H., Rossi, M., and Della Pina, C. Angew. Chem., Int. Ed. 46, 4434–
4440 (2007).
[159] Johnson, D. and Taconi, K. Environ. Prog. 26, 338–348 (2007).
[160] Ohara, T., Sato, T., Shimizu, N., Prescher, G., Schwind, H., Weiberg, O., and Marten, K. “Acrylic acid and
derivatives”. Ullmann’s Encyclopedia of Industrial Chemistry. Wiley-CVH Verlag GmbH & Co, 7th edition
(2006).
[161] Corma, A., Huber, G. W., Sauvanaud, L., and O’Connor, P. J. Catal. 257, 163–171 (2008).
[162] Kim, Y. T., Jung, K.-D., and Park, E. D. Microporous Mesoporous Mater. 131, 28–36 (2010).
[163] Jia, C.-J., Liu, Y., Schmidt, W., Lu, A.-H., and Schüth, F. J. Catal. 269, 71–79 (2010).
[164] Suprun, W., Lutecki, M., Haber, T., and Papp, H. J. Mol. Catal. A: Chem. 309, 71–78 (2009).
[165] Chai, S., Wang, H., Liang, Y., and Xu, B. J. Catal. 250, 342–349 (2007).
[166] Tsukuda, E., Sato, S., Takahashi, R., and Sodesawa, T. Catal. Commun. 8, 1349–1353 (2007).
[167] Behr, A., Eilting, J., Irawadi, K., Leschinski, J., and Lindner, F. Green Chem. 10, 13–30 (2008).
[168] Chiu, C., Dasari, M., Suppes, G., and Sutterlin, W. AIChE J. 52, 3543–3548 (2006).
[169] Schmidt, S. R., Tanielyan, S. K., Marin, N., Alvez, G., and Augustine, R. L. Top. Catal. 53, 1214–1216
(2010).
[170] Chaminand, J., Djakovitch, L., Gallezot, P., Marion, P., Pinel, C., and Rosier, C. Green Chem. 6, 359–361
(2004).
[171] Kusunoki, Y., Miyazawa, T., Kunimori, K., and Tomishige, K. Catal. Commun. 6, 645–649 (2005).
125
[172] Miyazawa, T., Kusunoki, Y., Kunimori, K., and Tomishige, K. J. Catal. 240, 213–221 (2006).
[173] Miyazawa, T., Koso, S., Kunimori, K., and Tomishige, K. Appl. Catal., A 318, 244–251 (2007).
[174] Nakagawa, Y., Shinmi, Y., Koso, S., and Tomishige, K. J. Catal. 272, 191–194 (2010).
[175] Qin, L.-Z., Song, M.-J., and Chen, C.-L. Green Chem. 12, 1466–1472 (2010).
[176] Carrettin, S., McMorn, P., Johnston, P., Griffin, K., and Hutchings, G. J. Chem. Commun. , 696–697
(2002).
[177] Villa, A., Veith, G. M., and Prati, L. Angew. Chem., Int. Ed. 49, 4499–502 (2010).
[178] Kyriacou, D. and Tougas, T. P. J. Org. Chem. 52, 2318–2319 (1987).
[179] Garcia, R., Besson, M., and Gallezot, P. Appl. Catal., A 127, 165–176 (1995).
[180] Gallezot, P. Catal. Today 37, 405–418 (1997).
[181] Ciriminna, R., Palmisano, G., Pina, C. D., Rossi, M., and Pagliaro, M. Tetrahedron Lett. 47, 6993–6995
(2006).
[182] Taarning, E., Madsen, A. T., Marchetti, J. M., Egeblad, K., and Christensen, C. H. Green Chem. 10,
408–414 (2008).
[183] Ramírez-López, C. and Ochoa-Gómez, J. Ind. Eng. Chem. Res. 49, 6270–6278 (2010).
[184] Lux, S., Stehring, P., and Siebenhofer, M. Sep. Sci. Technol. 45, 1921–1927 (2010).
[185] Taarning, E., Saravanamurugan, S., Spangsberg Holm, M., Xiong, J., West, R. M., and Christensen, C. H.
ChemSusChem 2, 625–627 (2009).
[186] West, R. M., Holm, M. S., Saravanamurugan, S., Xiong, J., Beversdorf, Z., Taarning, E., and Christensen,
C. H. J. Catal. 269, 122–130 (2010).
[187] Karinen, R. and Krause, A. Appl. Catal., A 306, 128–133 (2006).
[188] Klepácová, K., Mravec, D., Kaszonyi, A., and Bajus, M. Appl. Catal., A 328, 1–13 (2007).
[189] Deutsch, J., Martin, A., and Lieske, H. J. Catal. 245, 428–435 (2007).
[190] Fabbri, D., Bevoni, V., Notari, M., and Rivetti, F. Fuel 86, 690–697 (2007).
[191] Clacens, J. Appl. Catal., A 227, 181–190 (2002).
[192] Kunkes, E., Simonetti, D., Dumesic, J., Pyrz, W., Murillo, L., Chen, J., and Buttrey, D. J. Catal. 260,
164–177 (2008).
[193] Simonetti, D., Kunkes, E., and Dumesic, J. J. Catal. 247, 298–306 (2007).
[194] Dauenhauer, P., Salge, J., and Schmidt, L. J. Catal. 244, 238–247 (2006).
[195] Cheng, C. K., Foo, S. Y., and Adesina, A. A. Catal. Commun. 12, 292–298 (2010).
[196] Sharma, P. O., Swami, S., Goud, S., and Abraham, M. A. Environ. Prog. Sustainable Energy 27, 22–29
(2008).
[197] Simonetti, D. A., Rass-Hansen, J., Kunkes, E. L., Soares, R. R., and Dumesic, J. A. Green Chem. 9, 1073–
1083 (2007).
[198] Knothe, G. Prog. Energ. Combust. 36, 364–373 (2010).
[199] Kubička, D. and Kaluža, L. Appl. Catal., A 372, 199–208 (2010).
[200] Šimáček, P., Kubička, D., Šebor, G., and Pospíšil, M. Fuel 88, 456–460 (2009).
[201] Kubička, D., Bejblová, M., and Vlk, J. Top. Catal. 53, 168–178 (2010).
[202] Kubička, D., Šimáček, P., and Žilková, N. Top. Catal. 52, 161–168 (2009).
[203] Kubička, D., Šimáček, P., Kolena, J., Lederer, J., and Šebor, G. In ICP2007 proceedings - 43rd International
Petroleum Conference, pp. 1–7 (, Bratislava, Slovak Republic, 2007).
[204] Şenol, O., Viljava, T.-R., and Krause, A. O. I. Appl. Catal., A 326, 236–244 (2007).
[205] Ryymin, E.-M., Honkela, M. L., Viljava, T.-R., and Krause, A. O. I. Appl. Catal., A 358, 42–48 (2009).
[206] Şenol, O. I., Ryymin, E. M., Viljava, T.-R., and Krause, A. O. I. J. Mol. Catal. A: Chem. 268, 1–8 (2007).
[207] Snåre, M., Kubičková, I., Mäki-Arvela, P., Eränen, K., Wärnå, J., and Murzin, D. Y. Chem. Eng. J. 134,
29–34 (2007).
126
[208] Mäki-Arvela, P., Holmbom, B., Salmi, T., and Murzin, D. Y. Catal. Rev. - Sci. Eng. 49, 197–340 (2007).
[209] Morgan, T., Grubb, D., Santillan-Jimenez, E., and Crocker, M. Top. Catal. 53, 820–829 (2010).
[210] Do, P. T., Chiappero, M., Lobban, L. L., and Resasco, D. E. Catal. Lett. 130, 9–18 (2009).
[211] Lestari, S., Mäki-Arvela, P., Eränen, K., Beltramini, J., Max Lu, G. Q., and Murzin, D. Y. Catal. Lett. 134,
250–257 (2009).
[212] Ping, E. W., Wallace, R., Pierson, J., Fuller, T. F., and Jones, C. W. Microporous Mesoporous Mater. 132,
174–180 (2010).
[213] Hancsók, J., Krár, M., Magyar, S., Boda, L., Holló, A., and Kalló, D. Microporous Mesoporous Mater. 101,
148–152 (2007).
[214] Lestari, S., Mäki-Arvela, P., Beltramini, J., Lu, G. Q. M., and Murzin, D. Y. ChemSusChem 2, 1109–19
(2009).
[215] Danuthai, T., Jongpatiwut, S., Rirksomboon, T., Osuwan, S., and Resasco, D. E. Appl. Catal., A 361,
99–105 (2009).
[216] Sooknoi, T., Danuthai, T., Lobban, L. L., Mallinson, R., and Resasco, D. E. J. Catal. 258, 199–209 (2008).
[217] Sang, O. Y. Energy Sources 25, 859–869 (2003).
[218] Quirino, R. L., Tavares, A. P., Peres, A. C., Rubim, J. C., and Suarez, P. A. Z. J. Am. Oil Chem. Soc. 86,
167–172 (2008).
[219] Lima, D. G., Soares, V. C., Riberio, E. B., Carvalho, D. A., Cardoso, E. C. V., Rassi, F. C., Mundim, K. C.,
Rubim, J. C., and Suarez, P. A. Z. J. Anal. Appl. Pyrolysis 71, 987–996 (2004).
[220] Lange, J.-P. Biofuels, Bioprod. Biorefin. 1, 39–48 (2007).
[221] Srokol, Z. W. and Rothenberg, G. Top. Catal. 53, 1258–1263 (2010).
[222] Hoekman, S. K., Broch, A., and Robbins, C. Energy Fuels (2011).
[223] Knežević, D., van Swaaij, W., and Kersten, S. Ind. Eng. Chem. Res. 49, 104–112 (2010).
[224] Karagöz, S., Bhaskar, T., Muto, A., and Sakata, Y. Bioresour. Technol. 97, 90–8 (2006).
[225] Bhaskar, T., Sera, A., Muto, A., and Sakata, Y. Fuel 87, 2236–2242 (2008).
[226] Becker, J., Toft, L. L., Aarup, D. F., Villadsen, S. R., Glasius, M., Iversen, S. B., and Iversen, B. B. Energy
Fuels 24, 2737–2746 (2010).
[227] Hammerschmidt, A., Boukis, N., Hauer, E., Galla, U., Dinjus, E., Hitzmann, B., Larsen, T., and Nygaard,
S. D. Fuel 90, 555–562 (2011).
[228] Mentzel, U. V. and Holm, M. S. Appl. Catal., A 396, 59–67 (2011).
[229] Barth, T. and Kleinert, M. Chem. Eng. Technol. 31, 773–781 (2008).
[230] Kleinert, M. and Barth, T. Chem. Eng. Technol. 31, 736–745 (2008).
[231] Elliott, D. C., Hart, T. R., Neuenschwander, G. G., Rotness, L. J., and Zacher, A. H. Environ. Prog.
Sustainable Energy 28, 441–449 (2009).
[232] French, R. J., Hrdlicka, J., and Baldwin, R. Environ. Prog. Sustainable Energy 29, 142–150 (2010).
[233] Bridgwater, A. V., Czernik, S., and Piskorz, J. The Status of Biomass Fast Pyrolysis, chapter 1, 1–22. CPL
Press, Birmingham, United Kingdom (2002).
[234] Aho, A., Kumar, N., Lashkul, A., Eränen, K., Ziolek, M., Decyk, P., Salmi, T., Holmbom, B., Hupa, M.,
and Murzin, D. Y. Fuel 89, 1992–2000 (2010).
[235] Aho, A., Kumar, N., Eränen, K., Salmi, T., Hupa, M., and Murzin, D. Y. Fuel 87, 2493–2501 (2008).
[236] Valle, B., Gayubo, A. G., Alonso, A., Aguayo, A. T., and Bilbao, J. Appl. Catal., B 100, 318–327 (2010).
[237] de Miguel Mercader, F., Groeneveld, M., Kersten, S., Way, N., Schaverien, C., and Hogendoorn, J. Appl.
Catal., B 96, 57–66 (2010).
[238] Mullen, C. A. and Boateng, A. A. Fuel Processing Technology 91, 1446–1458 (2010).
[239] Carlson, T. R., Tompsett, G. A., Conner, W. C., and Huber, G. W. Top. Catal. 52, 241–252 (2009).
[240] Ferrari, M., Maggi, R., Delmon, B., and Grange, P. J. Catal. 198, 47–55 (2001).
127
[241] Ferrari, M., Bosmans, S., Maggi, R., Delmon, B., and Grange, P. Catal. Today 65, 257–264 (2001).
[242] Ferrari, M., Delmon, B., and Grange, P. Carbon 40, 497–511 (2002).
[243] Ferrari, M., Delmon, B., and Grange, P. Microporous Mesoporous Mater. 56, 279–290 (2002).
[244] Laurent, E. and Delmon, B. Appl. Catal., A 109, 77–96 (1994).
[245] Su-Ping, Z. Energy Sources 25, 57–65 (2003).
[246] Gutierrez, A., Domine, M., and Solantausta, Y. In 15th European Biomass Conference & Exhibition, pp.
1–5 (, Berlin, Germany, 2007).
[247] Şenol, O., Ryymin, E.-M., Viljava, T.-R., and Krause, A. J. Mol. Catal. A: Chem. 277, 107–112 (2007).
[248] Dilcio Rocha, J. Org. Geochem. 30, 1527–1534 (1999).
[249] Oasmaa, A., Kuoppala, E., Ardiyanti, A., Venderbosch, R. H., and Heeres, H. J. Energy Fuels 24, 5264–
5272 (2010).
[250] Wildschut, J., Melián-Cabrera, I., and Heeres, H. Appl. Catal., B 99, 298–306 (2010).
[251] Huber, G. W., Cortright, R. D., and Dumesic, J. A. Angew. Chem., Int. Ed. 43, 1549–1551 (2004).
[252] Kunkes, E. L., Simonetti, D. A., West, R. M., Serrano-Ruiz, J. C., Gärtner, C. A., and Dumesic, J. A.
Science 322, 417–421 (2008).
[253] Li, N. and Huber, G. W. J. Catal. 270, 48–59 (2010).
[254] West, R. M., Kunkes, E. L., Simonetti, D. A., and Dumesic, J. A. Catal. Today 147, 115–125 (2009).
[255] Román-Leshkov, Y., Barrett, C. J., Liu, Z. Y., and Dumesic, J. A. Nature 447, 982–985 (2007).
[256] Chheda, J. and Dumesic, J. Catal. Today 123, 59–70 (2007).
[257] Bond, J., Alonso, D., Wang, D., West, R., and Dumesic, J. Science 327, 1110–1114 (2010).
[258] Gärtner, C. A., Serrano-Ruiz, J. C., Braden, D. J., and Dumesic, J. A. J. Catal. 266, 71–78 (2009).
[259] Gürbüz, E. I., Kunkes, E. L., and Dumesic, J. A. Green Chem. 12, 223–227 (2010).
[260] Gärtner, C. A., Serrano-Ruiz, J. C., Braden, D. J., and Dumesic, J. A. ChemSusChem 2, 1121–1124 (2009).
[261] Gürbüz, E. I., Kunkes, E. L., and Dumesic, J. A. Appl. Catal., B 94, 134–141 (2010).
[262] Serrano-Ruiz, J. C., Braden, D. J., West, R. M., and Dumesic, J. A. Appl. Catal., B 100, 184–189 (2010).
[263] Kintisch, E. Science 320, 306–308 (2008).
[264] Digman, B., Joo, H., and Kim, D. Environ. Prog. Sustainable Energy 28, 47–51 (2009).
[265] Bridgwater, T. J. Sci. Food Agric. 86, 1755–1768 (2006).
[266] Henrich, E., Dahmen, N., and Dinjus, E. Biofuels, Bioprod. Biorefin. 3, 28–41 (2009).
[267] McKendry, P. Bioresour. Technol. 83, 55–63 (2002).
[268] Moilanen, A., Nasrullah, M., and Kurkela, E. Environ. Prog. Sustainable Energy 28, 355–359 (2009).
[269] Gomez-Barea, A., Vilches, L., Leiva, C., Campoy, M., and Fernandez-Pereira, C. Chem. Eng. J. 146,
227–236 (2009).
[270] Rass-Hansen, J., Christensen, C. H., Sehested, J., Helveg, S., Rostrup-Nielsen, J. R., and Dahl, S. Green
Chem. 9, 1016–1021 (2007).
[271] Matsumura, Y., Minowa, T., Potic, B., Kersten, S., Prins, W., Vanswaaij, W., Vandebeld, B., Elliott, D.,
Neuenschwander, G., and Kruse, A. Biomass Bioenerg. 29, 269–292 (2005).
[272] Campoy, M., Gómez-Barea, A., Villanueva, A. L., and Ollero, P. Ind. Eng. Chem. Res. 47, 5957–5965
(2008).
[273] Wolfesberger, U., Aigner, I., and Hofbauer, H. Environ. Prog. Sustainable Energy 28, 372–379 (2009).
[274] Zwart, R., Van Der Drift, A., Bos, A., Visser, H., Cieplik, M., and Könemann, H. Environ. Prog. Sustainable
Energy 28, 324–335 (2009).
[275] Pope, C. J., Marrone, P. A., and Yeh, B. V. Environ. Prog. Sustainable Energy 29, 151–162 (2010).
[276] Torres, W., Pansare, S. S., and Goodwin Jr, J. G. Catal. Rev. - Sci. Eng. 49, 407–456 (2007).
[277] Rönkkönen, H., Klemkaitė, K., Khinsky, A., Baltušnikas, A., Simell, P., Reinikainen, M., Krause, O., and
Niemelä, M. Fuel 90, 1076–1089 (2011).
128
[278] Stranges, A. Fischer Tropsch Archive (http://www.fischer-tropsch.org). Accessed June 7th , 2011.
[279] Davis, B. H. Ind. Eng. Chem. Res. 46, 8938–8945 (2007).
[280] Kaneko, T., Derbyshire, F., Makino, E., Gray, D., and Tamura, M. (2005).
[281] Patzlaff, J., Liu, Y., Graffmann, C., and Gaube, J. Appl. Catal., A 186, 109–119 (1999).
[282] Gamba, S., Pellegrini, L. A., Calemma, V., and Gambaro, C. Catal. Today 156, 58–64 (2010).
[283] Tavasoli, A., Trépanier, M., Dalai, A. K., and Abatzoglou, N. J. Chem. Eng. Data 55, 2757–2763 (2010).
[284] Tavasoli, A., Trépanier, M., Malek Abbaslou, R. M., Dalai, A. K., and Abatzoglou, N. Fuel Process. Technol.
90, 1486–1494 (2009).
[285] Bezemer, G. L., Bitter, J. H., Kuipers, H. P. C. E., Oosterbeek, H., Holewijn, J. E., Xu, X., Kapteijn, F., van
Dillen, A. J., and de Jong, K. P. J. Am. Chem. Soc. 128, 3956–3964 (2006).
[286] Chen, W., Fan, Z., Pan, X., and Bao, X. J. Am. Chem. Soc. 130, 9414–9419 (2008).
[287] Liu, S., Gujar, A. C., Thomas, P., Toghiani, H., and White, M. G. Appl. Catal., A 357, 18–25 (2009).
[288] Trepanier, M., Tavasoli, A., Dalai, A., and Abatzoglou, N. Appl. Catal., A 353, 193–202 (2009).
[289] Ma, W., Kugler, E. L., Wright, J., and Dadyburjor, D. B. Energy Fuels 20, 2299–2307 (2006).
[290] Herranz, T., Rojas, S., Perezalonso, F., Ojeda, M., Terreros, P., and Fierro, J. J. Catal. 243, 199–211 (2006).
[291] Blanchard, J., Abatzoglou, N., Eslahpazir-Esfandabadi, R., and Gitzhofer, F. Ind. Eng. Chem. Res. 49,
6948–6955 (2010).
[292] Christensen, J. M., Mortensen, P., Trane, R., Jensen, P. A., and Jensen, A. D. Appl. Catal., A 366, 29–43
(2009).
[293] Durham, E., Zhang, S., and Roberts, C. Appl. Catal., A 386, 65–73 (2010).
[294] Mentzel, U. V., Shunmugavel, S., Hruby, S. L., Christensen, C. H., and Holm, M. S. J. Am. Chem. Soc.
131, 17009–17013 (2009).
[295] Dabelstein, W., Reglitzky, A., Schütze, A., and Reders, K. (2007).
[296] Knothe, G., Matheaus, A. C., and Ryan III, T. W. Fuel 82, 971–975 (2003).
[297] Dinh, L., Guo, Y., and Mannan, M. Environ. Prog. Sustainable Energy 28, 38–46 (2009).
[298] Vasudevan, P. and Briggs, M. J. Ind. Microbiol. Biotechnol. 35, 421–430 (2008).
[299] de Paula Gomes, M. S. and Muylaert de Araújo, M. S. Renew. Sustain. Energy Rev. 13, 2201–2204 (2009).
[300] Granda, C. B., Zhu, L., and Holtzapple, M. T. Environ. Prog. Sustainable Energy 26, 233–250 (2007).
[301] Willems, P. A. Science 325, 707–8 (2009).
[302] Sharma, Y. C., Singh, B., and Korstad, J. Biofuels, Bioprod. Biorefin. 5, 69–92 (2011).
[303] Refaat, A. International Journal of Environment Science and Technology 8, 203–221 (2010).
[304] Olivier-Bourbigou, H., Magna, L., and Morvan, D. Appl. Catal., A 373, 1–56 (2010).
[305] Guo, F., Fang, Z., Tian, X.-F., Long, Y.-D., and Jiang, L.-Q. Bioresour. Technol. 102, 6469–6472 (2011).
[306] Liang, X., Gong, G., Wu, H., and Yang, J. Fuel 88, 613–616 (2009).
[307] Zhang, L., Xian, M., He, Y., Li, L., Yang, J., Yu, S., and Xu, X. Bioresour. Technol. 100, 4368–4373 (2009).
[308] Fang, D., Yang, J., and Jiao, C. ACS Catal. 1, 42–47 (2011).
[309] Ghiaci, M., Aghabarari, B., Habibollahi, S., and Gil, A. Bioresour. Technol. 102, 1200–1204 (2011).
[310] Okayasu, T., Saito, K., Nishide, H., and Hearn, M. T. W. Green Chem. 12, 1981–1989 (2010).
[311] Melero, J. A., Iglesias, J., and Morales, G. Green Chem. 11, 1285 (2009).
[312] Hara, M. ChemSusChem 2, 129–135 (2009).
[313] Okamura, M., Takagaki, A., Toda, M., Kondo, J. N., Domen, K., Tatsumi, T., Hara, M., and Hayashi, S.
Chem. Mater. 18, 3039–3045 (2006).
[314] Kitano, M., Arai, K., Kodama, A., Kousaka, T., Nakajima, K., Hayashi, S., and Hara, M. Catal. Lett. 131,
242–249 (2009).
[315] Veldurthy, B., Clacens, J. M., and Figueras, F. Adv. Synth. Catal. 347, 767–771 (2005).
[316] Desmartin-Chomel, A., Hamad, B., Palomeque, J., Essayem, N., Bergeret, G., and Figueras, F. Catal.
129
Today 152, 110–114 (2010).
[317] Baig, A. and Ng, F. T. T. Energy Fuels 24, 4712–4720 (2010).
[318] Garcia, C. M., Teixeira, S., Marciniuk, L. L., and Schuchardt, U. Bioresour. Technol. 99, 6608–6613 (2008).
[319] Peter, S. and Weidner, E. Eur. J. Lipid Sci. Technol. 109, 11–16 (2007).
[320] Zhang, L., Guo, W., Liu, D., Yao, J., Ji, L., Xu, N., and Min, E. Energy Fuels 22, 1353–1357 (2008).
[321] Immer, J. G., Kelly, M. J., and Lamb, H. H. Appl. Catal., A 375, 134–139 (2010).
[322] Berenblyum, A. S., Danyushevsky, V. Y., Katsman, E. A., Podoplelova, T. A., and Flid, V. R. Pet. Chem.
50, 305–311 (2010).
[323] Bulushev, D. A., Beloshapkin, S., and Ross, J. R. Catal. Today 154, 7–12 (2010).
[324] Snåre, M., Kubickova, I., Mäki-Arvela, P., Chichova, D., Eränen, K., and Murzin, D. Y. Fuel 87, 933–945
(2008).
[325] Pinna, F. Catal. Today 41, 129–137 (1998).
[326] Dulaurent, O., Chandes, K., Bouly, C., and Bianchi, D. J. Catal. 188, 237–251 (1999).
[327] Andersson, M., Bligaard, T., Kustov, A., Larsen, K., Greeley, J., Johannessen, T., Christensen, C., and
Nørskov, J. K. J. Catal. 239, 501–506 (2006).
[328] Surovikin, V. F., Plaxin, G. V., Semikolenov, V. A., Likholobov, V. A., and Tiunova, I. J. Porous carbona-
ceous material. United States Patent No. US4978649 (1990).
[329] Simonov, P. A., Troitskii, S. Y., and Likholobov, V. A. Kinet. Catal. 41, 255–269 (2000).
[330] Bernas, H., Eränen, K., Simakova, I., Leino, A.-R., Kordás, K., Myllyoja, J., Mäki-Arvela, P., Salmi, T.,
and Murzin, D. Y. Fuel 89, 2033–2039 (2010).
[331] Immer, J. G. and Lamb, H. H. Energy Fuels 24, 5291–5299 (2010).
[332] Boda, L., Onyestyák, G., Solt, H., Lónyi, F., Valyon, J., and Thernesz, A. Appl. Catal., A 374, 158–169
(2010).
130
Chapter 6
Included Publications 6
131
Step changes and deactivation behaviour in the continuous decarboxylation of stearic acid
a,b b b,c b d
Anders Theilgaard Madsen , Bartosz Rozmysłowicz , Irina L. Simakova , Teuvo Kilpiö , Anne-Riikka Leino , Krisztián
d b b b,
Kordás , Kari Eränen , Päivi Mäki-Arvela , Dmitry Yu. Murzin *
a
: Centre for Catalysis and Sustainable Chemistry, Technical University of Denmark, DK-2800 Kgs. Lyngby, Denmark
b
: Process Chemistry Center, Åbo Akademi University, Biskopsgatan 8, FI-20500 Turku/Åbo, Finland
c
:Boreskov Institute of Catalysis. Novosibirsk, Russia
d
: Laboratory of Microelectronics and Materials Physics, University of Oulu, PL 4500, 90014 Oulu
* To whom correspondence should be addressed: [email protected]
Abstract
Deoxygenation of diluted and concentrated stearic acid over 2% Pd/C beads was performed in a continuous reactor at
300°C and 20 bar pressure of Ar or 5% H2/Ar. Stable operation was obtained in 5% H2 atmosphere, with 95% conversion of
10 mol% diluted stearic acid in dodecane and 12% conversion of pure stearic acid. Deactivation took place in H2-deficient
gas atmosphere, probably due to formation of unsaturated products and coking in the pore system. Transient experiments
with step changes were performed: 1 h was required for the step change to be visible in liquid sampling, while steady-
states were achieved after totally 2.5 - 3 h. Post-reaction analysis of the spent catalyst revealed that a deactivation profile
was formed downwards over the catalyst bed.
1. Introduction (reactions (1)-(3)), in all cases yielding n-alkanes upon

The diminishing availability of cheap fossil fuels and saturation with hydrogen. The primary reaction route
their impact on the global environment have led the and the side reactions taking place are strongly
struggle towards higher fuel efficiency and development dependent on the catalyst used – especially reactions of
of alternative fuels for road vehicles. The EU had hydrogen and product gases may interfere and prompt
mandated that at least 5.75% of the energy content in water-gas-shift (WGS) or methanation (reactions (4)-(6)).
diesel and petrol should be based on biofuels by the end
1
of 2010, rising to 10% in 2020. Lately, in their 2010
report, the International Energy Agency (IEA) described
the urgent need for the swift transition towards more
sustainable and environmentally sound energy and fuel
2
production.
Fats and oils are generally the types of biomass
resource that, from a chemical point of view, most
closely resembles diesel oil and is easiest to upgrade into
3,4
a diesel fuel. Diesel fuel can be made from fats and oils
via (trans-)esterification with methanol to form fatty
acid methyl esters (FAME), which have previously been
extensively investigated. Another way could be through A number of reports have appeared on catalytic
the deoxygenation of fats and oils potentially using cracking-type reactions of various vegetable oils without
hydrogen (so-called HDO, hydrodeoxygenation) to form added hydrogen. These are usually performed at 400-
8-13
straight-chain alkanes, which is advantageous under 500°C without H2 in the reaction media. FAME have
5 14, 15
some conditions. also been used as feedstocks. Porous Brønsted
The HDO process was first introduced on an acidic materials tested include H-MCM-41 or zeolites H-
8-11,15
industrial scale in 2007 by Neste Oil in Porvoo, Finland ZSM-5 or H-Y. Porous basic materials tested
6,7
based primarily on palm oil, and other companies are comprise zeolite (Cs,Na)-X, hydrotalcites or porous SnO-
12-14
following the same route. The reports on understanding ZnO-Al2O3. Generally, the procedures result in a
the deoxygenation processes have however not been mixture of hydrocarbons due to cracking reactions, as
extensive to this date. Generally, three types of catalysts well as light gases and coke – petrol can often be the
have been described for deoxygenation of fatty product of the highest yield. Saponification of resulting
feedstocks, namely supported sulphided metals free fatty acids can also occur over some basic catalysts
12, 14
catalysts, supported metallic catalysts, and porous acidic like MgO.
or basic catalysts. Deoxygenation has been described to The sulphided NiMo/Al2O3 and CoMo/Al2O3 at 250°C
occur via three different routes: decarboxylation, under 15 or 75 bars of H2 and various sulphidation
decarbonylation or full reduction with hydrogen conditions were studied as HDO catalysts. The
-1-
conversion of heptanoic alcohol, acid and esters and the process (for making cellulose fibres from wood), and
38
ratio between C6 and C7 hydrocarbons varied depending their deoxygenation was investigated over Pd/Sibunit.
on the catalysts type and added sulphur compounds in Deactivation was especially pronounced with
the feed. Unsulphided NiMo/Al2O3 or CoMo/Al2O3 unsaturated fatty acids, therefore it is suspected to
catalysts were neither very active nor selective towards occur via cyclisations from C-C-double-bonds in the
linear hydrocarbons, however, the sulphided catalysts feedstock followed by dehydrogenation over Pd catalyst
35, 38, 39
had high activities and yielded higher ratios of C6/C7 thus forming C17-aromatic hydrocarbons.
16, 17
hydrocarbon products. The conversion of rapeseed Continuous decarboxylation of stearic acid, a typical
oil over various sulphided NiMo/Al2O3 have been fatty acid found especially in saturated fats was recently
investigated: employing temperatures in the range of demonstrated in continuous mode over Pd/C catalysts at
35,40
260-360°C and pressures of 70-150 bar H2 . The effects 300-360°C. Continuous reactors are more industrially
of pressure, temperature, sulphidation conditions, and interesting than batch or semibatch reactors, and it is
18-20
Ni and Mo composition of the catalyst were studied. necessary to obtain information on long-term
Co-hydrocracking of 5% rapeseed oil in vacuum gas oil performance and transients during continuous
over sulphided NiMo/Al2O3 at higher temperatures (400- decarboxylation to describe the catalyst involved. In this
21
420°C) was also investigated. Supporting sulphided work the continuous decarboxylation of stearic acid at
CoMo on inorganic mesoporous materials showed that 300°C is investigated with the aim of describing step
organised mesoporous alumina had better performance changes during start-up and shutdown, steady-state
than the reference industrial γ-alumina as a support, conversion and deactivation phenomena.
which in turn exhibited a better performance than
MCM-41 mesoporous alumina-silicates. Incorporating 2. Experimental
more Al into the Si-framework of MCM-41 increased the 2.1. Chemicals. Stearic acid (>99% purity) was
22, 23
hydrocarbon yield. The HDO of sunflower and palm supplied by Sigma-Aldrich. N,O-bis(trimethylsilyl)-
oil was also studied over sulphided NiMo/Al2O3 in trifluoroacetamide (or BSTFA, of >99% purity) as a
24, 25
mixture with vacuum gas oil or neat. silylation agent was delivered from Acros Organics. n-
Decarboxylation is generally favoured by higher dodecane (>99% purity) and pyridine (>99% purity) were
temperatures and low pressures of H2 over sulphided obtained from Fluka. Ar (>99.99%) and 5% H2/Ar gases
catalysts. were delivered from AGA . All chemicals were used as
Deoxygenation of triglycerides with supported received.
26-31
transition-metals has been scarcely studied: Typically 2.2 Catalyst preparation & characterisation. 2 wt%
high-surface-area oxide or carbon materials are used for Pd on Sibunit (synthetic mesoporous carbon) was
26,27
supporting Pd or Pt metals, and isomerisation may prepared by the following method as previously
be performed simultaneously by using strongly acidic 41
28-30
published: The Sibunit was treated with 5% HNO3
porous materials as support or co-support. Also overnight at 25°C and dried at 80°C. H2PdCl4 was
FAME deoxygenation had successfully been performed hydrolysed in an aqueous solution, Na2CO3 was added
over Pt-based catalyst at 300-350°C in 6.9 bar of H2 or until a pH = 9 was reached, and Sibunit was added to
31
He. this solution for deposition of the Pd(II)-hydro-
Several waste feedstocks such as abattoir waste, complexes. The solution was left for 6 h at 25°C, then
used cooking oils or greases contain considerable filtered and washed with H2O until no chloride ions were
amounts of free fatty acids (FFAs). It was shown that detected in the wash water. The catalyst was finally
FFAs are also intermediates in the deoxygenation of dried and subsequently calcined for 2 h at 200°C.
27,32
corresponding esters. The study of FFA can therefore The particle diameter distributions and mean Pd
be considered crucial in deoxygenation. Palladium or diameters were calculated based on the frequency of
platinum metals have previously been found to be the particles from images obtained by a LEO 912 OMEGA
most selective and active catalysts for decarboxylation energy-filtered transmission electron microscope (TEM)
33
of FFAs. Various decarboxylation parameters have operating with acceleration voltage of 120 kV. At least
previously been investigated in semibatch-mode (a 100 particles were counted for obtaining average
stirred autoclave with liquid reactants and continuous particle diameter of Pd. The fresh and spent catalyst
renewal of only the gas phase), primarily with stearic beads were crushed to a powder, mixed thoroughly and
acid as a feedstock over commercial Pd catalysts, Pd on glued to an epoxy plate.
SBA-15 or Pd on Sibunit carbon (a mesoporous and The X-ray Diffraction (XRD) mean particle diameter
34-36
temperature-stable carbon composite). It was found based on volume of the particles were calculated with
that the deoxygenation rate was independent on the the Scherrer formula from powder diffractograms
37
fatty acid chain length in the range C18 to C23. Tall Oil
Fatty Acids (TOFA) are by-products from the Kraft
-2-
obtained with a Siemens D5000 X-ray Powder 1. During experiments liquid samples of ca. 0.1 g were
Diffractometer taken out in glass vials downstream the reactor, while
The pore volume and specific surface area remaining liquid effluent was led to a heated collector
measurements were conducted with a physisorption / vessel to be purged at set intervals. Gases were purged
chemisorption Sorptometer 1900 from Carlo Erba continuously by a pressure reduction controller, and
instruments with liquid N2 at 77 K. Specific surface areas part of the gas stream was diluted in He and analysed
were calculated using the Brunauer-Emmett-Teller (BET) for CO and CO2 on a Siemens ULTRAMAT 6 online IR
equation from the N2 adsorption-desorption isotherms. analyser. The glass vials with liquid samples were
The pore size distribution was obtained from the weighed and 4 ml n-dodecane was added to the glass
Dollimore-Heal correlation. vial as a solvent. Of this, an aliquot of 10.0 μL of was
The distribution of active metal in fresh catalyst transferred to a GC vial, followed by addition of 100 μL
beads was investigated by laser ablation connected to dodecane, 100 μL pyridine and 50 μL BSTFA for
an inductively-coupled plasma mass-spectrometer (LA- derivatisation and left to stand at 70°C for 60 min. An
ICP-MS), with laser ablation system of New Wave UP- overview of the reactor changes is given in Table 1.
213 and Perkin-Elmer ICP-MS Sciex Elan 6100 DRC Plus.
A number of catalyst beads were cut in half and fixed in Table 1: Changes during continuous decarboxylation with
epoxy glue for the measurements. diluted and pure stearic acid over 2 wt% Pd/Sibunit
Reactor effluent contents of Pd were measured by Step
TOS Con-
ICP-EOS with a Perkin-Elmer 5300 DV optical emission change Changes applied
(min) dition
spectrometer. 0.2 g of effluent was added to 5 ml of 65% #
HNO3 (aq) and 1 ml of 30 % H2O2 (aq), heated in a
Diluted stearic acid in 5% H2/Ar
microwave oven and then diluted to 100 ml (aq) before
-55 liquid Started flow of dodecane
ICP-OES analysis.
I) 0 liquid 10 mol% stearic acid in dodecane
2.3 Reactor loading, catalyst activation and II) 300 Gas 5% H2/Ar
deoxygenation. A tubular reactor of 18 cm height and
1.58 cm inner diameter (volume 35.1 ml) was loaded Diluted stearic acid in 5% H2/Ar
from the bottom up with a small layer of quartz wool, III) 1575 Gas pure Ar
then 6 ml of quartz sand of 0.2-0.8 mm in diameter and 6.55 mol% stearic acid in
IV) 1816 Liquid
a layer of quartz wool. Afterwards the catalyst bed of 10 dodecane
g (volume uptake 19.0 ml) of Pd/Sibunit was loaded into V) 3016 Liquid dodecane
the reactor. Thereafter a layer of quartz wool, 4 ml of 4389 Temp Quenching to 150°C started
quartz sand and a layer of quartz wool were introduced.
The quartz sand and quartz wool mixed the inlet flow in 100% stearic acid in Ar
the reactor as well as the outlet to make liquid flow experiment
Heating startup from 150°C to
behaviour and sampling uniform. 4359 Temp
300°C
A thermocouple was fixed at the centre of the
4389 Reactor heated
catalyst bed. Then the reactor was mounted and leak- VI) 4481 Liquid 100% stearic acid
tested with Ar at room temperature and afterwards VII) 7056 Liquid dodecane
purged for 16 h at 20 bars with Ar also at room 7056 Temp Quenching from 300°C to 150°C
temperature. An aluminium heating element was fitted
around the reactor. 100% stearic acid in 5% H2/Ar
The catalyst was then activated in 5% H2/Ar at 20 bar Heating from 150°C to 300°C at
7056 Temp
by heating at 10°C/min to 150°C, kept at 150°C for 1 h, 15°C/min
followed by heating at 10°C/min to the reaction 7056 Gas 5% H2/Ar
temperature of 300°C. Reactions were performed in VIII) 7066 Liquid 100% stearic acid
either pure Ar or 5% H2/Ar at 42 Nml/min and with a IX) 18506 Liquid Dodecane
flow of either pure stearic acid or 10 wt% in n-dodecane, 18881 Gas Pure Ar
18881 Liquid Stopped flow
both at 0.075 ml/min.
18881 Temp Cooling reactor to 25°C
Since stearic acid melts at 70°C, it was kept in a
heated storage vessel at 90°C, while the pump and inlet
piping were kept at 100°C during experimentation with
pure stearic acid. Trace gas impurities (e.g. molecular
oxygen) were expelled from the storage vessel by
continuous purging of 20 ml/min N2 gas through the
stearic acid. The flow reactor setup is sketched in Figure
-3-
1.5
1,5
I) II) III) IV) V)
1.25
1,25
Gas: Gas: Gas
Ar 5% H2/Ar Ar
molar fraction
1.0
1
0.75
0,75
0.5
0,5
0.25
0,25
0.0
0
0 500 1000 1500 2000 2500 3000 3500
TOS (min)
Figure 2: Decarboxylation in liquid flow of stearic acid over

Pd/C at 300°C diluted in dodecane at 0.075 ml/min, 5 % H2/Ar
or pure Ar gas at 20 bar at 42 Nml/min. Changes in the
operational parameters: I) Gas: Ar, II) Gas: 5% H2/Ar, III)
Erroneous liquid sampling, IV) Gas: Ar, V) Liquid: Dodecane.
Legend: Liquid mole balance (-♦-), Stearic acid (-●-),
heptadecane (-▲-), heptadecene (-■-), undecylbenzene (-x-),
CO (x), CO2 (♦).The Roman numbers refer to Table 1.
Initially a steep increase in gas and liquid

concentrations is seen due to an unintended blocking of
the gas feed line (Figure 2, section III), which was then
Figure 1: Flow reactor schematic for continuous deoxygenation
of stearic acid. unblocked shortly thereafter. Stable operation was
approached and correctly sampled liquid obtained at
After reaction, the reactor was washed for several TOS = 1816 min for pure Ar sweeping gas. Initially only
hours with dodecane and Ar gas. Thereafter the system heptadecane and CO2 were formed and no CO. The CO2
was cooled and the pressure carefully reduced. The concentration corresponded to 50 mol% of the
reactor was disassembled and the spent catalyst beads concentration of heptadecane. However, the CO
sorted into sample glasses depending on axial position in concentration exiting the reactor increased over time,
the reactor bed. and around TOS = 2650-2850 min the CO2 concentration
dropped to practically zero while concentration of CO
rose even more before settling at a concentration
3. Results & discussion
corresponding to around 55% conversion at 2800 min
3.1. Steady state decarboxylation of stearic acid.
TOS. The liquid samples also were found to contain high
3.1.1. Stearic acid diluted in dodecane – step changes
levels of heptadecene (C17 alkene) of 8-12% and undecyl-
II)-III). At 450-1575 min of time-on-stream (TOS) the
benzene (C17-monoaromate) of 4-5%, as well as
reaction was performed in 5% H2/Ar. The reactor was
unconverted stearic acid (20%). The production of
kept at steady state conditions while liquid samples
heptadecane (C17 alkane) started to drop as well. The
were extracted. The online analysis of CO2 and CO
formation of unsaturated compounds are likely caused
concentrations showed that a conversion around 60%
by the lack of added H2, leading to cyclisation and
would have been achieved, however from the liquid
dehydrogenation of the products.
samples it is clear that the conversion of stearic acid to
The initial lack of CO may be due to the lack of H2
heptadecane is close to 95 % at full selectivity. The
affording reverse water-gas-shift (WGS) reaction,
discrepancy between CO/CO2 and liquid phase analysis
however the constant and gradual rise in CO
could likely be due to formation of methane via
concentration followed (with a delay of 1-2 h) the
methanation of either CO2, CO or both due to the
formation of heptadecane and liberation of H2 via
hydrogen content in the sweeping gas. Methane formed
aromatisation to undecylbenzene. H2O and H2 may have
could not be measured on the Ultramat with the present
been formed via these side reactions.
configuration.
The formation of linear hydrocarbons from FFAs has
3.1.2. Effect of Ar – step changes IV)-V). To see
been proposed to take place by a mechanism going
whether or not this behaviour would persist in
through formic acid: First, the fatty acid splits into formic
hydrogen-free atmosphere the sweeping gas was
acid and 1-alkene. Then formic acid rapidly decomposes,
changed back to Ar at TOS = 1575 min. The
either via dehydrogenation to CO2 and H2, or by
concentration in the feed vessel was also lowered from
dehydration to CO and H2O (overall reaction (1) and (2)).
10 mol% (ca. 15 wt%) to 6.5 mol% (10 wt%) stearic acid.
Hydrogen from the gas-phase afterwards saturated the
42,43
formed alkene. This can explain the enhanced
formation of unsaturated compounds during reaction in
-4-
pure Ar. Formic acid, however, quickly decomposes 1.2
1,2
a
Step change # VI) - VII)
under the present conditions and therefore cannot be 1.0
1
44
observed in liquid or gas phase.
molar fraction
0.8
0,8
Heptadecene and aromatic C17-compounds are 0.6

0,6
Gas: Ar
formed during reaction where catalyst deactivation is 0.4
0,4
Liquid: 100% Stearic acid
36,38
often fast. Therefore the flow was switched to 0.2
0,2
dodecane at TOS = 3120 min, and the Ar flow was
0.0
0
maintained to wash the reactor (Figure 2, step V). The 4550 5050 5550 6050 6550 7050
TOS (min)
CO and CO2 concentration did not start to decline until
1.2
1,2
TOS = 3300 min when the reactor was cooled to 150°C. Step change # VIII) - IX) b
1.0
1
The sum of molar amounts of CO and CO2 detected in
molar fraction
0.8
0,8
the IR analyser after the liquid flow was switched to
dodecane corresponds to 4.21 mmol, while maximally 0.6
0,6
Gas: 5% H2/Ar
Liquid: 100% stearic acid
0.250 mmol of CO can be situated on the surface of the 0.4
0,4
Pd particles (calculated from XRD mean diameter of 12 0.2

0,2
nm as reported in section 3.3, taking particles as spheres 0.0

0
45 7800 9800 11800 13800 15800 17800
and Pd:CO ratio to 2). Thus, the CO and CO2 measured TOS (min)
in dodecane liquid flow came from reactant holdup in Figure 3: Mole fraction based on stearic acid, steady state
the catalyst bed – the stearic acid situated in the pores decarboxylation of 100 % stearic acid flow. a) Pure Ar gas at 20
of the catalyst particles continued to react. bar from step change VI) to VII). b) 5 % H2/Ar gas at 20 bar
3.1.3. Effect of concentrated stearic acid under Ar – from step VIII) to IX). Legend: Liquid mole balance (-♦-), Stearic
step change VI)-VII). Following the reaction with 10 wt% acid (-●-), heptadecane (-▲-), heptadecene (-■-),
undecylbenzene (-x-), CO (x), CO2 (+).The Roman numbers refer
stearic acid in dodecane, the reactor was flushed
to Table 1.
overnight with Ar and cooled to ambient temperature,
and reheated to 150°C, flushed for several hours with
3.1.4. Effect of concentrated stearic acid under 5%
dodecane and argon and afterwards solely with argon.
H2/Ar – step change VIII)-IX). After the above-mentioned
Then the reactor was reheated to 300°C at 10°C/min,
Ar-experiment (Section 3.1.3) the next step was to
and a flow of pure stearic acid was fed to the reactor to
assess the effect of 5% H2/Ar as gas atmosphere. The
determine the steady-state deoxygenation of
reactor was again cooled to 150°C, flushed with
concentrated stearic acid in Ar during the following time
dodecane to wash out residual stearic acid and products
interval of two days (TOS = 4481 - 7060 min). This is seen and reheated to 300°C to make a new attempt to
in Figure 3a.
analyse the start-up transient and steady-state
The mass balance of the liquid samples performing
conversion in 5% H2/Ar atmosphere (Figure 3b).
the deoxygenation of pure stearic acid under Ar was
From 8300 to 18500 min TOS the production of
only ca. 80%. At the same time about 2% heptadecane heptadecane was constant around 12% (atmosphere 5%
was formed in the start of the period, decreasing with
H2/Ar) while no other products were detected in the
time to about 1% at the end. The online IR-active
liquid phase. In the gas-phase only CO was detected,
analysis showed CO corresponding to about 3%
corresponding to a conversion of between 8 % and 6 %,
conversion, decreasing over time to about 1.5%. No 1-
decreasing with time. CO may have been formed from
heptadecene is detected, therefore methanation is not
CO2 due to the WGS equilibrium in the H2-containing
supposed to take place. It is suspected that coupling 42,43
atmosphere or from deoxygenation via formic acid.
reactions with unsaturated compounds, reactants or The missing CO or CO2 was most likely converted to
both may take place, for instance coupling of formed
compounds not detectable in the IR-active online
alkenes to tetratriacontene (C34-alkene). 18-
analysis – methane being a possible product, or coke via
Oxopentatriacontane (C35-ketone) formed via
the Boudouard reaction. CO2 formed from Boudouard
ketonisation is however less likely to have been formed reaction can again react with H2 to CO via WGS or more
33,46
as it usually requires a basic metal oxide as support.
H2 to form methane.
The heavy components formed would not have
3.1.5. Comparison of the catalyst performance in
vaporised easily in split/splitless injection in GC analysis,
different gas and liquid feeds. The steady-state
and might as well have been too heavy to be seen with
deoxygenation experiments with stearic acid are
the applied GC method. It may be concluded that the
compared in Table 2. It is evident that the lack of H2 in
lack of added H2 in the gas feed led to low activity and
the feed led to deactivation, while stable catalyst activity
gradual deactivation of the catalyst during the ca. 2600 can be upheld in 5% H2/Ar. Initial deactivation during the
min TOS mentioned.
deoxygenation of neat stearic acid was previously
-5-
observed using 20 bar Ar instead of 5% H2/Ar as 1,5
1.5
Gas: Ar Gas: 5% H2/Ar a
Liquid: 10 mol% stearic acid Liquid: 10 mol% stearic acid
sweeping gas over 5% Pd/Sibunit at 360°C, but it was not 1.25
1,25
found that the initial catalyst activity could be regained
molar fraction
1.0
1
I) II)
at this temperature - a stable conversion of about 15% 0.75
0,75
neat stearic acid was obtained under 5% H2/Ar at 0.5

0,5
35
360°C, comparable with 12% conversion of neat stearic 0.25
0,25
acid in 5% H2/Ar at 300°C obtained in the present study. 0.0

0
-10 90 190 290 390 490
TOS (min)
Table 2: Comparison of the steady state-yields of heptadecane 1,2
1.2
VIII)
Gas: 5% H2/Ar b
using different stearic acid concentrations in the feed. 1
Liquid: Dodecane -> stearic acid
1.0
Conditions: 300°C, 20 bar sweeping gas at 42 Nml/min, liquid
molar fraction
0.8
0,8
feed 0.075 ml/min. Dilution is in dodecane
Yields of heptadecane (mol%) 0.6
0,6
Sweeping gas atmosphere: 0.4

0,4
Liquid Feed: 5 % H2 in Pure Ar 0.2

0,2
Ar 0.0
0
Diluted stearic acid 95% (15 75% decreasing to 55 %; 7060 7080 7100 7120 7140 7160 7180 7200 7220
TOS (min)
(mol% stearic acid in mol%) formation of unsaturated
1.2
dodecane) and aromatic C17 IX) Gas: 5% H2/Ar c
Liquid: Stearic acid -> dodecane
compounds (10 mol%) 1.0
Pure stearic acid 12 % 2% decreasing to 1%;
molar fractoin
0.8
Possible Formation of 0.6

heavier products 0.4
0.2
3.2. Transients during step-changes. Three
0.0
transients in the catalytic deoxygenation over Pd/Sibunit 18500 18550 18600 18650 18700 18750 18800 18850 18900
are emphasised here, namely the flow pattern during TOS (min)
Figure 4: Mole balances based on stearic acid feed content for

start-up for 10% stearic acid in n-dodecane as well as the
step change. a) start-up of 10 mol% stearic acid in dodecane
start-up and shut-down of 100% stearic acid under 5%
flow – step change I)+II). b) start-up of 100 % stearic acid flow
H2/Ar, respectively. The liquid pumping speeds are the – step change VIII). c) shut-down of 100 % stearic acid flow –
same for the three conditions. The transients can be step change IX). Legend: Liquid mole balance (-♦-), Stearic acid
seen in Figure 4 a-c. (-●-), heptadecane (-▲-), heptadecene (-■-), undecylbenzene (-
There was a considerable hold-up time when x-), CO (x), CO2 (+).The Roman numbers refer to Table 1.
switching the flow to or from dodecane until the
sampling on the outlet side of the reactor was affected – After about 15 min the CO2-concentration from the
in all experiments approximately 1 h, as seen from online CO2/CO analysis started to rise, meaning that
Figure 4a-c. Once the concentration in the liquid decarboxylation of the reactant was taking place.
samples had started to change, the transient lasted However, the first traces of heptadecane in the liquid
another 1.5 to 2 h before the concentration of the phase occur only over 2 h after the step change. The
steady-state was reached and the reactor stabilised. production of CO2 and CO as well as the heptadecane
During shutdown, the concentration of reactant and seems to be stable at TOS = 300 min, with both gas-
products did not fall to zero, as an additional number of phase and liquid phase corresponding to 35-40%
hours were needed to completely remove the last conversion.
compounds probably located in the porous catalyst Thereafter, the gas atmosphere was changed to 5%
beads and in cavities connected to the reactor. H2/Ar at TOS = 300 min and the liquid flow kept
3.2.1. Start-up of stearic acid diluted in dodecane – constant. The CO2 and CO concentrations as determined
step change I) and II). After initial leak-testing of the by the online CO/CO2 analysis start to respond to this
reactor, the liquid flow was switched to Ar-flushing change after about 15 min, however, the corresponding
overnight and the catalyst was reduced in 20 bar of 5% response in heptadecane concentration is not visible in
H2/Ar: First, the reactor was heated at 10°C/min to the liquid phase analysis before 1 h (around TOS = 360
150°C, where it was kept for 1 h, and the finally heated min), and a steady state was only reached 2.5 h after the
at 10°C/min to the reaction temperature of 300°C. Then step change was induced (around TOS = 450 min). At
the flow was switched to pure Ar and the pumping of 10 TOS = 475 min the steady state concentration profile
mol% stearic acid in dodecane begun (at TOS = 0 min), had developed, and now the mass balance compute as
depicted in Figure 4a. 90-95% heptadecane compared to theoretical possible
value as well as 5 % stearic acid.
-6-
It is suspected that the liquid feed built up a trickling reactor for several hours prior to switch to stearic acid
film wetting the beads of the catalyst bed and quartz flow. Therefore minute amounts of stearic acid may
material, and that the slow response in stearic acid appear in the samples already before flow of stearic acid
concentration was due to the low overall flow of the is introduced in the reactor. It is presumed that both
reactant. The solvent n-dodecane has a normal boiling stearic acid and products was still situated in the
point of 216°C (vapour pressure is 6 bar at 300°C), the microporous catalysts beads, due to capillary forces, a
products heptadecane and 1-heptadecene have normal long time after stopping the liquid flow and from here
boiling points of ca. 302°C (vapour pressure around 1 they are only slowly mixed into the reactor after the
bar at 300°C), and as these compounds are not supplied switch to dodecane. This could explain the constant
or produced in amounts sufficient to saturate the gas production of heptadecane despite the abrupt decrease
phase they will vaporise and quickly leave the reactor in stearic acid concentration.
with the sweeping gas once formed. Stearic acid is not 3.3. Post-reaction analyses – catalyst deactivation.
suspected to have vaporised (boiling point is 383°C), and A thorough post-reaction analysis of the spent catalyst
will have propagated through the reactor in liquid form. was performed. The reactor was flushed first in n-
The low feed rate of stearic acid means that a trickling dodecane and then flushed extensively in Ar gas at
liquid film hereof will propagate through the reactor 150°C. The reactor was then dismantled and the catalyst
very slowly. taken out of the reactor in a stepwise manner by sorting
3.2.2. Start-up of 100 % stearic acid feed in 5% H2/Ar the catalyst beads into batches depending on their axial
– step change VIII). Following reaction with liquid feed of position in the bed.
100 % stearic acid under pure Ar atmosphere, the The total weight of the dried catalyst bed extracted
reactor was flushed several hours with liquid dodecane from the reactor summed to 12.72 g versus fresh weight
followed by Ar gas at 150°C. Finally the gas atmosphere of 10 g, thus the uptake of carbonaceous species is 2.72
was switched to 5% H2/Ar and flow of stearic acid to the g or 0.227 mol of C. In comparison, the total molar
reactor was started at TOS = 7060 min. The CO signal amount of stearic acid passing through the reactor is
increased slightly after 5 minutes (no CO2 formation was 3.43 mol stearic acid or 61.7 mol of C, giving a ratio of
observed), but an increase in the concentration of coke-to-stearic acid percentage of 6.6%. The white and
stearic acid was not seen before 60 to 75 min. The transparent quartz wool and quartz sand near the
concentration of stearic acid and product heptadecane reactor inlet and the outlet were not coloured by the
rose over an additional 1 h period before approaching a process. This indicates that coking can be attributed to
steady state concentration with up to 12 % conversion the catalyst itself.
(TOS = 7190 min). This is depicted in Figure 4 b. LA-ICP-MS confirmed that palladium is impregnated
3.2.3. Shut-down of 100 % stearic acid feed in 5% in an “eggshell-like” layer in the outer rim of the carbon
H2/Ar – step change IX). After more than 10000 min on beads, while the centre of the particle is inert, thus
stream in stable behaviour, the flow was switched from minimising the diffusion pathway. It is not expected that
pure stearic acid to pure dodecane to initiate reactor the majority of the palladium particles were mobile
shut down (at TOS = 18500 min), see Figure 4 c. The CO under the conditions used.
signal from the online analysis was affected after a few 3.3.1. Coking in the pore system: The results from the
minutes and started to decline gradually over a period of analysis of the fresh and spent catalyst beads, sorted by
8 h. In the liquid sampling the response to this step position in the catalyst bed are shown in Table 3. The
change is seen after ca. 1 h, and the concentration of BET area and pore volume decreased downwards in the
stearic acid declined to 10 % at 2.5 h after the step bed, and a deactivation profile formed downwards in
change. Peculiarly, the concentration of heptadecane the catalyst bed in a way that the largest amount of coke
did not decrease in 3-4 h after the step change. After was formed in the catalyst beads near the reactor feed
these few hours the concentrations decreased only very inlet and it decreased down the catalyst bed.
slowly. This is ascribed to the presence of small Furthermore it is evident that the catalyst beads near
reservoirs of liquid after the reactor – in the the reactor inlet have a lower percentage of pores in the
thermocouple fittings or quartz material after the 5-10 nm range and below 2 nm compared to the beads
catalyst bed. This also means that the apparatus may not near the reactor outlet, apparently due to a higher
be completely liberated of stearic acid despite the percentage of pores in the 2-5 nm range.
purging with dodecane at 300°C or 150°C through the
-7-
Table 3: Characterisation of the spent catalyst depending on position in the catalyst bed
Sample # Catalyst Total Surface area Relative Pore Pore size distribution based TEM Pd diameter
(flow bed (BET) decrease in volume on pore diameter in nm (XRD in parenthesis)
direction: position surf. Area (BET)
↓)
>10 10-5 5-2 <2
2 *
[mm] [m /g] % [ml/g] % % % % [nm]
1 0.0 - 9.6 101 72 0.321 1.3 7.4 76.2 15.2 7.7±3.5 (11.9)
2 9.6 - 19.0 118 67 0.430 1.3 8.5 77.8 12.3 ---
3 19.0 - 27.8 130 64 0.377 1.1 8.7 71.4 18.8 ---
4 27.8 - 49.2 137 62 0.383 1.5 9.1 76.3 13.2 7.1±4.5 (12.4)
5 49.2 - 69.8 157 57 0.502 1.4 10.5 70.4 17.7 ---
6 69.8 - 100 166 54 0.474 1.3 10.0 67.0 21.8 6.0±4.5 (11.4)
Fresh cat. --- 361 0.873 2.0 15.5 73.2 9.3 6.7±2.8 (12.8)
***
Sibunit --- 504 1.23 66 10 14 10 ---
*
XRD mean Pd particle is based on volume calculated from the Scherrer formula; TEM is based on the frequency of particles counted in
the micrographs.
** 35
Based on ref.
on the spent catalyst was 7.7 nm near the inlet, whereas
The coking profile suggests that the larger pores have it was 6.1 nm near the outlet of the reactor.
became narrower due to the gradual coking taking Furthermore, the average Pd particle size in the fresh
place, and that some of the smaller pores have been catalyst was 6.7 nm. As TEM only analyses a tiny part of
occluded partially or completely. The by-products the catalyst bead, these differences are within
formed in the pores do not completely prevent experimental uncertainty. The mean particle diameters
deoxygenation reaction, since a stable conversion of obtained via analysis of scattered X-rays were within a
12% was obtained constantly in the final 10500 min of narrower interval around 12 nm. It can be therefore
TOS with neat stearic acid under 5% H2/Ar. Further concluded that deactivation is not caused by sintering.
exploration of this catalyst deactivation profile as a This is furthermore in accordance with previous reports
function of bed length could be a topic of a subsequent on fresh and spent Pd/Sibunit catalysts for
study. It has been reported that deactivation takes place decarboxylation, both of 1 wt% or 5 wt% Pd loading.
by cyclisation of unsaturated compounds in hydrogen- Spent catalysts reported in the literature have active
37,38,47
deficient atmosphere over 1 wt% Pd/Sibunit. Thus metal dispersions and particle sizes either marginally
35,37,40,47
it can be concluded that catalyst deactivation is caused bigger or of the same sizes as the fresh catalysts.
by coking, since the specific BET surface area decreased
by 72% to 54% as a function of catalyst bed length 1 cm
and 9 cm from the inlet, respectively. This has also been
35,40
the case in previous studies.
3.3.2. Palladium particle sizes. The fresh catalyst
beads as well as catalyst samples 1, 4, and 6 in Table 3
were selected for TEM and XRD analysis to obtain
average particle sizes of the palladium and particle size
distributions. The results are shown in Table 3, and a
high-resolution TEM images and calculated particle
diameter distribution of the fresh catalyst and sample #
1 are shown in Figure 5. 30 <d> 6.7± 2.8 nm
30
<d> 7.7± 3.5 nm
The differences observed between XRD and TEM
Frequency (%)
Frequency (%)
20 20
mean Pd particle sizes in Table 3 is due to the different
way of counting particles – XRD mean particle diameter 10 10
is based on volume of the particles while TEM mean

0 0
particle diameter is based on the number of particles in 0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40
Particle size (nm)
Particle size (nm)
a small area. According to Figure 5, there is a percentage
of particles above 10 nm diameter which will contribute Figure 5: TEM micrographs and particle diameter distribution
much more to the average particle size based on volume of a fresh catalyst bead (left-hand side) and a spent catalyst
(XRD) in contrast to the average Pd diameter based on bead from the top of the bed (catalyst sample # 1) (right-hand
the frequency of counted particles (TEM). The TEM side)
micrographs indicated an average diameter of palladium
-8-
3.3.3. Leaching. The effluent from the reactor was References
collected in two batches of about 0.5 kg each and (1) The European Union. Directive 2009/28/EC on the
analysed with ICP-EOS. The total mass of Pd in the promotion of the use of renewable energy . Off. J. EU. 2009,
effluent was 1.15 mg, indicating that 0.58% of the Pd 140, 1-47.
present in the catalyst bed (200 mg) had leached to the (2) International Energy Agency World Energy Outlook;
effluent during the course of the entire time-on-stream International Energy Agency: Paris, France, 2010; p. 700.
(3) Knothe, G.; Van Gerpen, J.; Krahl, J. The Biodiesel
period over 14 days at 300°C. The leaching in the latter 7
Handbook; Knothe, G.; Van Gerpen, J.; Krahl, J., Eds.; 1st ed.;
days was only about one third of the leaching in the first
AOCS Press: Urbana, Illinois, USA, 2005.
7 days of reaction (0.15% versus 0.43%). It is therefore (4) Mittelbach, M.; Remschmidt, C. Biodiesel: The
expected that the leaching was due to a few loosely Comprehensive Handbook; Mittelbach, M., Ed.; 3rd ed.; Martin
bound palladium particles on the catalyst surface and Mittelbach: Graz, Austria, 2006.
that leaching ceased with time as these were removed. (5) Kalnes, T. N.; Koers, K. P.; Marker, T.; Shonnard, D. R. A
Deactivation is, therefore, not due to leaching, which is technoeconomic and environmental life cycle comparison of
in accordance with results from previous studies green diesel to biodiesel and syndiesel. Environ. Prog.
concluding that no or below 1% leaching of palladium Sustainable Energy, 2009, 28, 111-120.
(6) Koskinen, M.; Sourander, M.; Nurminen, M. Apply a
takes place during deoxygenation over 1 wt%
37,40 comprehensive approach to biofuels. Hydrocarb. Process.
Pd/Sibunit.
2006, 85, 81-86.
(7) Mikkonen, S. Second-generation renewable diesel
4. Conclusion offers advantages. Hydrocarb. Process. 2008, 87, 63-66.
Transient and steady-state behaviour of continuous (8) Sang, O. Y. Biofuel Production from Catalytic Cracking of
stearic acid decarboxylation was studied over 2 wt% Palm Oil. Energy Sources, 2003, 25, 859-869.
Pd/Sibunit-carbon at 300°C at 20 bars. Pure Ar or 5% H2 (9) Twaiq, F. A.; Zabidi, N. A. M.; Mohameda, A. R.; Bhatia,
S. Catalytic conversion of palm oil over mesoporous
in Ar were used as sweeping gases, and both pure and
aluminosilicate MCM-41 for the production of liquid
dodecane-diluted feeds of stearic acid were employed as
hydrocarbon fuels. Fuel Process. Technol. 2003, 84, 105-120.
reactants. (10) Lima, D. G.; Soares, V. C. D.; Riberio, E. B.; Carvalho, D.
Diluted stearic acid was almost fully converted to A.; Cardoso, É. C. V.; Rassi, F. C.; Mundim, K. C.; Rubim, J. C.;
heptadecane in hydrogen-containing gas, while Suarez, P. A. Z. Diesel-like fuel obtained by pyrolysis of
deactivation via formation of unsaturated compounds vegetable oils. J. Anal. Appl. Pyrolysis. 2004, 71, 987-996.
took place in pure argon gas. Deactivation irreversibly (11) Tamunaidu, P.; Bhatia, S. Catalytic cracking of palm oil
changed the catalyst to produce CO instead of CO2. Pure for the production of biofuels: optimization studies. Bioresour.
stearic acid selectively yielded around 12% heptadecane technol. 2007, 98, 3593-601.
(12) Na, J.-G.; Yi, B. E.; Kim, J. N.; Yi, K. B.; Park, S.-Y.; Park,
in hydrogen-containing gas, while only 2% heptadecane
J.-H.; Kim, J.-N.; Ko, C. H. Hydrocarbon production from
and likely formation of heavier compounds resulted
decarboxylation of fatty acid without hydrogen. Catal. Today,
from hydrogen-free atmosphere. 2009, 156, 44-48.
The same catalyst was used during all of the (13) Quirino, R. L.; Tavares, A. P.; Peres, A. C.; Rubim, J. C.;
experimentation. A deactivation gradient formed during Suarez, P. A. Z. Studying the Influence of Alumina Catalysts
reaction as a function of the distance from the liquid Doped with Tin and Zinc Oxides in the Soybean Oil Pyrolysis
feed inlet – the most deactivated catalyst beads were Reaction. J. Am. Oil Chem. Soc. 2008, 86, 167-172.
those closest to the inlet. Deactivation took place (14) Sooknoi, T.; Danuthai, T.; Lobban, L. L.; Mallinson, R.;
mainly by coke formation in the pores and on the Resasco, D. E. Deoxygenation of methylesters over CsNaX. J.
Catal. 2008, 258, 199-209.
surface, but not by agglomeration, leaching or sintering.
(15) Danuthai, T.; Jongpatiwut, S.; Rirksomboon, T.;
Osuwan, S.; Resasco, D. E. Conversion of methylesters to
Acknowledgements hydrocarbons over an H-ZSM5 zeolite catalyst. Appl. Catal., A.
A. T. Madsen acknowledges support from the Danish 2009, 361, 99-105.
Research and Innovation Agency through the (16) Senol, O. I.; Viljava, T.-R.; Krause, A. O. I. Effect of
consortium “Waste-2-value”, and is furthermore grateful sulphiding agents on the hydrodeoxygenation of aliphatic
to the Laboratory of Industrial Chemistry and Reaction esters on sulphided catalysts, Appl. Catal., A. 2007, 326, 236-
244.
Engineering at Åbo Akademi University for hosting his
(17) Ryymin, E.-M.; Honkela, M. L.; Viljava, T.-R.; Krause, A.
stay in the group during 2010. The authors wish to thank
O. I. Insight to sulfur species in the hydrodeoxygenation of
Paul Ek from the Process Chemistry Centre, Åbo aliphatic esters over sulfided NiMo/γ-Al2O3 catalyst. Appl.
Akademi University for performing ICP-MS and ICP-EOS Catal., A. 2009, 358, 42-48.
measurements. (18) Šimáček, P.; Kubička, D.; Šebor, G.; Pospíšil, M.
Hydroprocessed rapeseed oil as a source of hydrocarbon-
based biodiesel. Fuel, 2009, 88, 456-460.
-9-
(19) Šimáček, P.; Kubička, D.; Šebor, G.; Pospíšil, M. Fuel Deoxygenation of Stearic Acid over Supported Pd/C Catalyst.
properties of hydroprocessed rapeseed oil. Fuel, 2010, 89, 611- Catal. Lett. 2008, 122, 247-251.
615. (35) Lestari, S.; Mäki-Arvela, P.; Bernas, H.; Simakova, O.;
(20) Kubička, D.; Kaluža, L. Deoxygenation of vegetable oils Sjöholm, R.; Beltramini, J.; Lu, G. Q. M.; Myllyoja, J.; Simakova,
over sulfided Ni, Mo and NiMo catalysts. Appl. Catal., A. 2010, I.; Murzin, D. Y. Catalytic Deoxygenation of Stearic Acid in a
372, 199-208. Continuous Reactor over a Mesoporous Carbon-Supported Pd
(21) Šimáček, P.; Kubička, D. Hydrocracking of petroleum Catalyst. Energy Fuels, 2009, 23, 3842-3845.
vacuum distillate containing rapeseed oil: Evaluation of diesel (36) Lestari, S.; Mäki-Arvela, P.; Eränen, K.; Beltramini, J.;
fuel. Fuel, 2010, 89, 1508–1513. Max Lu, G. Q.; Murzin, D. Y. Diesel-like Hydrocarbons from
(22) Kubička, D.; Šimáček, P.; Žilková, N. Transformation of Catalytic Deoxygenation of Stearic Acid over Supported Pd
Vegetable Oils into Hydrocarbons over Mesoporous-Alumina- Nanoparticles on SBA-15 Catalysts. Catal. Lett. 2009, 134, 250-
Supported CoMo Catalysts. Top. Catal. 2009, 52, 161-168. 257.
(23) Kubička, D.; Bejblová, M.; Vlk, J. Conversion of (37) Simakova, I.; Simakova, O.; Mäki-Arvela, P.; Murzin, D.
Vegetable Oils into Hydrocarbons over CoMo/MCM-41 Y. Decarboxylation of fatty acids over Pd supported on
Catalysts. Top. Catal. 2010, 53, 168–178. mesoporous carbon. Catal. Today, 2010, 150, 28-31.
(24) Huber, G. W.; O’Connor, P.; Corma, A. Processing (38) Rozmysłowicz, B.; Mäki-Arvela, P.; Lestari, S.;
biomass in conventional oil refineries: Production of high Simakova, O. A.; Eränen, K.; Simakova, I.; Murzin, D. Y.; Salmi,
quality diesel by hydrotreating vegetable oils in heavy vacuum T. Catalytic Deoxygenation of Tall Oil Fatty Acids Over a
oil mixtures. Appl. Catal., A. 2007, 329, 120–129. Palladium-Mesoporous Carbon Catalyst: A New Source of
(25) Guzman, A.; Torres, J. E.; Prada, L. P.; Nuñez, M. L. Biofuels. Top. Catal. 2010, 53, 1274-1277.
Hydroprocessing of crude palm oil at pilot plant scale. Catal. (39) Mäki-Arvela, P.; Kubičková, I.; Snåre, M.; Eränen, K.;
Today, 2010, 156, 38-43. Murzin, D. Y. Catalytic Deoxygenation of Fatty Acids and Their
(26) Morgan, T.; Grubb, D.; Santillan-Jimenez, E.; Crocker, Derivatives. Energy Fuels, 2007, 21, 30-41.
M. Conversion of Triglycerides to Hydrocarbons Over (40) Bernas, H.; Eränen, K.; Simakova, I.; Leino, A.-R.;
Supported Metal Catalysts. Top. Catal. 2010, 53, 820-829. Kordás, K.; Myllyoja, J.; Mäki-Arvela, P.; Salmi, T.; Murzin, D. Y.
(27) Kubičková, I.; Snåre, M.; Eränen, K.; Mäki-Arvela, P.; Deoxygenation of dodecanoic acid under inert atmosphere.
Murzin, D. Y. Hydrocarbons for diesel fuel via decarboxylation Fuel, 2010, 89, 2033-2039.
of vegetable oils. Catal. Today, 2005, 106, 197-200. (41) Simonov, P. A.; Troitskii, S. Y.; Likholobov, V. A.
(28) Hancsók, J.; Krár, M.; Magyar, S.; Boda, L.; Holló, A.; Preparation of the Pd/C catalysts: A molecular-level study of
Kalló, D. Investigation of the production of high cetane number active site formation. Kinet. Catal. 2000, 41, 255-269.
bio gas oil from pre-hydrogenated vegetable oils over (42) Berenblyum, A S.; Danyushevsky, V. Ya.; Katsman, E. A;
Pt/HZSM-22/Al2O3. Microporous Mesoporous Mater. 2007, Podoplelovau, T. A; Flid, V. R. Production of Engine Fuels from
101, 148-152. Inedible Vegetable Oils and Fats. Petroleum Chemistry. 2010,
(29) Kikhtyanin, O. V.; Rubanov, A. E.; Ayupov, A. B.; 50, 305-311.
Echevsky, G. V. Hydroconversion of sunflower oil on Pd/SAPO- (43) Boda, L.; Onyestyák, G.; Solt, H.; Lónyi, F.; Valyon, J.;
31 catalyst. Fuel, 2010, 89, 3085-3092. Thernesz, A. Catalytic hydroconversion of tricaprylin and
(30) Ping, E. W.; Wallace, R.; Pierson, J.; Fuller, T. F.; Jones, caprylic acid as model reaction for biofuel production from
C. W. Highly dispersed palladium nanoparticles on ultra-porous triglycerides. Appl. Catal., A. 2010, 374, 158-169.
silica mesocellular foam for the catalytic decarboxylation of (44) Bulushev, D. A.; Beloshapkin, S.; Ross, J. R. H.
stearic acid. Microporous Mesoporous Mater. 2010, 132, 174- Hydrogen from formic acid decomposition over Pd and Au
180. catalysts. Catal. Today, 2010, 154, 7-12.
(31) Do, P. T.; Chiappero, M.; Lobban, L. L.; Resasco, D. E. (45) Dulaurent, O.; Chandes, K.; Bouly, C.; Bianchi, D. Heat
Catalytic Deoxygenation of Methyl-Octanoate and Methyl- of Adsorption of Carbon Monoxide on a Pd/Al2O3 Solid Using
Stearate on Pt/Al2O3. Catal. Lett. 2009, 130, 9-18. Infrared Spectroscopy at High Temperatures. J Catal. 1999,
(32) Snåre, M.; Kubičková, I.; Mäki-Arvela, P.; Eränen, K.; 188, 237-251.
Wärnå, J.; Murzin, D. Y. Production of diesel fuel from (46) Gärtner, C. A.; Serrano-Ruiz, J. C.; Braden, D. J.;
renewable feeds: Kinetics of ethyl stearate decarboxylation. Dumesic, J. A. Catalytic coupling of carboxylic acids by
Chem. Eng. J. 2007, 134, 29-34. ketonization as a processing step in biomass conversion. J.
(33) Snåre, M.; Kubičková, I.; Mäki-Arvela, P.; Eränen, K.; Catal. 2009, 266, 71-78.
Murzin, D. Y. Heterogeneous Catalytic Deoxygenation of (47) Immer, J. G.; Lamb, H. H. Fed-Batch Catalytic
Stearic Acid for Production of Biodiesel. Ind. Eng. Chem. Res. Deoxygenation of Free Fatty Acids. Energy Fuels, 2010, 24,
2006, 45, 5708-5715. 5291-5299.
(34) Lestari, S.; Simakova, I.; Tokarev, A.; Mäki-Arvela, P.;
Eränen, K.; Murzin, D. Y. Synthesis of Biodiesel via
- 10 -
JFUE 5350 No. of Pages 7, Model 5G
16 June 2011
Fuel xxx (2011) xxx–xxx

1
Contents lists available at ScienceDirect
Fuel
journal homepage: www.elsevier.com/locate/fuel
2 Hydrodeoxygenation of waste fat for diesel production: Study on model feed

3 with Pt/alumina catalyst
4 Anders Theilgaard Madsen, El Hadi Ahmed, Claus Hviid Christensen, Rasmus Fehrmann, Anders Riisager ⇑
5 Centre for Catalysis and Sustainable Chemistry, Department of Chemistry, Technical University of Denmark, Building 207, DK-2800 Kgs. Lyngby, Denmark
6
7
a r t i c l e i n f o a b s t r a c t
9
2 2
10 Article history: Hydrodeoxygenation of waste fats and oils is a viable method for producing renewable diesel oil. In this 23
11 Received 12 April 2011 study a model feed consisting of oleic acid and tripalmitin in molar ratio 1:3 was hydrotreated at 325 °C 24
12 Received in revised form 31 May 2011 with 20 bars H2 in a stirred batch autoclave with a 5 wt% Pt/c-Al2O3 catalyst, and samples were extracted 25
13 Accepted 1 June 2011
periodically and analyzed on GC. Despite the significant hydrogen pressure hydrogenation of both reac- 26
14 Available online xxxx
tants were limited and decarboxylation or decarbonylation of the ester and carboxylic acid functionalities 27
were highly favored, yielding carbon chain lengths of odd numbers. Moreover, Pd/c-Al2O3 was observed 28
15 Keywords:
to be slightly more active than Pt/c-Al2O3 and had a higher ratio of decarboxylation and decarbonylation 29
16 Alkanes
17 Biodiesel
to hydrogenation, while Ni/c-Al2O3 was substantially less active than Pt and also showed a markedly 30
18 Fats and oils lower ratio of decarboxylation and decarbonylation to hydrogenation. Variation of the temperature 31
19 Hydrogenation showed that triglycerides as well as free fatty acids were converted at all investigated temperatures, 32
20 Decarboxylation but the conversion of oleic acid increased significant from 6% to 100% when the temperature was 33
21 increased from 250 °C to 325 °C. The tripalmitin reacted via a palmitic acid intermediate, and its conver- 34
sion was limited by formation of this free fatty acid. 35
Ó 2011 Elsevier Ltd. All rights reserved. 36
37
38
39 1. Introduction for 2000, potentially amounting to 4% of the road vehicle diesel de- 61
mand if converted to an equal amount of diesel [4]. 62
40 The worldwide production of biofuels is growing. This growth is In the past decade another production strategy for diesel oil 63
41 amongst other factors driven by capital interests, environmental from fats has been shown possible [9–18]. By reacting fatty feed- 64
42 concerns, and the desire to make national fuel supplies more inde- stock at elevated temperatures over a heterogeneous catalyst in 65
43 pendent of the global petroleum supply and price. The European the presence of hydrogen, oxygen functionalities can be split off 66
44 Union (EU) fuel directive 2009/28/EC stated a target of 5.75% as H2O, CO2 or CO, resulting in long-chain alkanes by hydrodeoxy- 67
45 energy content from biofuels in 2010 and a rise to 10% by 2020 genation (HDO), as shown in Fig. 1a–d. Compared to transesterifi- 68
46 [1]. The most developed and widespread industrial method for pro- cation not many studies have appeared on HDO of fats and oils, 69
47 ducing diesel oils from biomass is the base-catalyzed transesterifi- although it may be especially suited for large scale integration with 70
48 cation of oils and fats with alcohol, preferably methanol. The petroleum refining. The HDO process has first been industrialized 71
49 product is referred to as fatty acid methyl ester (FAME). Though by Neste Oil in Porvoo, Finland who are using primarily vegetable 72
50 most authors specifically refer to FAME as biodiesel, others suggest oils as feedstock [16,17]. 73
51 wider definitions [2,3]. Refined plant oils are usually well suited for Krause and co-workers studied HDO of sulfided CoMo/Al2O3 74
52 transesterification as they have low contents of impurities, espe- and NiMo/Al2O3 on heptanoic acid, heptanol and heptanoate 75
53 cially free fatty acids (FFA) that form soap on addition of base methyl and ethyl esters. Sulfidation by addition of H2S instead of 76
54 [4]. However, inexpensive feedstocks like used cooking oils, abat- CS2 shifted the selectivity from C7 to C6 hydrocarbons. It was found 77
55 toir wastes and trap greases that would be preferable to use that NiMo, but not CoMo was very sensitive to the sulfidation con- 78
56 contain high amounts of FFA [5,6]; 2–7 wt% for used cooking oils, ditions and that the unsulfided catalysts were neither very active 79
57 5–30 wt% for waste animal fats and trap greases even higher nor selective to C6 and C7 alkane formation [19,20]. 80
58 [7,8]. Accordingly, these cheap feedstocks represent some chal- Kubička and co-workers evaluated the conversion of rapeseed 81
59 lenges due to more tedious workup. For instance in the USA a pro- oil over sulfided catalysts in a number of studies. When the activity 82
60 duction of 5.08 million m3 of waste fats and greases was reported and selectivity of sulfided NiMo/Al2O3 catalyst was compared to 83
that of unsulfided Ni/Al2O3 at 270–350 °C, the sulfided catalyst 84
was found to be much more active. However, the unsulfided cata- 85
⇑ Corresponding author. Fax: +45 4588 3136. lysts primary yielded odd number hydrocarbons (decarboxylation/ 86
E-mail address: [email protected] (A. Riisager). decarbonylation) while the sulfided yielded primarily even num- 87
0016-2361/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.fuel.2011.06.005
Please cite this article in press as: Madsen AT et al. Hydrodeoxygenation of waste fat for diesel production: Study on model feed with Pt/alumina catalyst.
Fuel (2011), doi:10.1016/j.fuel.2011.06.005
16 June 2011
2 A.T. Madsen et al. / Fuel xxx (2011) xxx–xxx
O
(a) R'-H
C
O O R' R'-CH3 (b)
H2 C H2
3 CO2 or H2O+CO + C3H8 + R''-H O O R'' R''-CH3 + C3H8 + 6 H2O
cat. cat.
R'''-H C
O R''' R'''-CH3
Triglyceride
Free fatty acid

(c) O (d)
H2 C H2
CO2 or CO + R-H HO R R-CH3 + 2 H2O
cat. cat.
Fig. 1. Decarbonylation, decarboxylation (left) and hydrogenation (right) of triglycerides and free fatty acids.
88 bered (complete reduction) [21]. The activity of sulfided NiMo/ 2. Experimental 137
89 Al2O3 superseded that of Mo/Al2O3 and Ni/Al2O3 separately in li-
90 quid phase HDO below 300 °C in H2 atmosphere [22]. Rapeseed 2.1. Catalyst preparation 138
91 oil HDO over commercial sulfided NiMo/Al2O3 at 70 bar H2
92 achieved full conversion above 300 °C, while higher temperatures A catalyst metal loading of 5 wt% on alumina was chosen as cat- 139
93 yielded higher selectivity towards odd-number hydrocarbons alyst to ensure activity and similitude to industrial practice. 5 wt% 140
94 [23]. Supporting sulfided CoMo on mesoporous Al2O3 rather than Pt/c-Al2O3 catalyst was prepared by the incipient-wetness impreg- 141
95 MCM-41 resulted in higher yields of alkanes, but incorporating of nation method [35], where the required amount of the active metal 142
96 Al into the framework of MCM-41 did improve yields due to sup- precursor H2PtCl66H2O (P99.0%, Fluka) was dissolved in deminer- 143
97 port acidity [24,25]. alized water and added to dried and fractionated (<180 lm) c-alu- 144
98 Resasco et al. concluded that HDO of methyl octanoate and mina powder (brand, BET area 194 m2/g). The wet catalyst was left 145
99 methyl stearate over Pt/Al2O3 at 330 °C proceeds in inert atmo- at room temperature for 2 h, then dried in an oven at 110 °C for 2 h 146
100 sphere, but addition of H2 suppresses formation of higher self- and finally calcined in air at 400 °C for 8 h with a 100 °C/h ramp. 147
101 condensates of both compounds. Decarbonylation was found al- 5 wt% Pd/c-Al2O3 and 5 wt% Ni/c-Al2O3 catalysts were prepared 148
102 ways to be the primary reaction route [26]. The deoxygenation in an analogous way by dissolving respectively Pd(NO3)22H2O 149
103 of methyl octanoate over H-ZSM-5 at 500 °C resulted in lighter (P99.8%, Fluka) or Ni(NO3)26H2O (P99.0%, Aldrich) in water and 150
104 hydrocarbon gasses and aromatization, the latter of which pro- impregnating alumina, as described above. 151
105 ceeded through self-condensation products (tetradecane, 8-penta-
106 decanone, octyl octanoate) [27]. Methyl octanoate was converted
107 to C6–C8 alkenes and condensated hydrocarbons (C14–C16) over 2.2. Catalyst characterization 152
108 basic CsNaX without forming aromatics, while weakly acidic

109 NaX resulted in marked production of aromatics before deactivat- The BET surface areas of the prepared catalysts as well as the c- 153
110 ing [28]. alumina support were determined by nitrogen physisorption by 154
111 Murzin and co-workers have performed seminal studies of the measuring adsorption and desorption isotherm at liquid nitrogen 155
112 deoxygenation of fatty acids and their derivatives [12–15,29–34]. temperature on a Micromeritics ASAP 2020 pore analyzer. Before 156
113 A range of supported metal catalysts were tested and noble met- measurement the samples were degassed at 200 °C under vacuum 157
114 als on carbon were found most beneficial for the decarboxylation for 4 h. 158
115 of stearic acid [14]. The conversions of stearic acid, ethyl stearate CO pulse-chemisorption was performed to determine the active 159
116 and tristearine to alkanes were studied in semibatch-reactor be- metal area, dispersion and particle size. This was done on a 160
117 tween 270 and 360 °C, either in gas-mixture with H2 or inert Micromeritics Autochem II 2920 with a loop size of 0.366 ml. Ca. 161
118 gas at up to 40 bars over Pd/C. The products in all cases were al- 100 mg catalyst sample was flushed with He and then heated 162
119 most exclusively n-heptadecane [12,13]. and it was shown that and reduced at 150 °C for 2 h in 10% H2/N2. Then the sample was 163
120 stearic acid is an intermediate in the conversion of the ester reac- flushed with He and cooled, and CO pulse-chemisorption was per- 164
121 tants to n-heptadecane [15]. It was furthermore shown that deox- formed at 25 °C with 5% CO/He using He carrier gas while the CO- 165
122 ygenation kinetics of other fatty acids proceeds with the same concentration in the effluent was continuously measured by a 166
123 rate [29]. thermal conductivity detector (TCD). For the Ni catalyst, the reduc- 167
124 Also in semibatch reaction-mode, Rozmysłowicz et al. studied tion was also attempted under varying conditions at up to 400 °C 168
125 tall oil fatty acid hydrogenation and decarboxylation [30], while for up to 12 h. The stoichiometries for area calculation were 169
126 Lestari et al. investigated deoxygenation of stearic acid over Pd Pt:CO = Ni:Co = 1:1 and Pd:CO = 2:1. 170
127 on SBA-15 [31] as well as stearic and palmitic acid over Pd on Sib-
128 unit carbon [32,33]. The deoxygenation of stearic acid was even 2.3. Hydrodeoxygenation 171
129 performed in a continuous reactor setup [34].
130 In this work we have focused on linking products to triglyceride The hydrogenation was carried out in a 50 ml stainless steel 172
131 and FFA in an idealized model mixture of fat in batch-mode reac- autoclave (MicroClave from Autoclave Engineers) by filling the 173
132 tion, and thus investigated the hydrodeoxygenation of a feed with autoclave with 8.1 g n-tetradecane (P99.0%, Fluka) as solvent, 174
133 a molar ratio of 1:3 between oleic acid and tripalmitin. This ap- 0.806 g tripalmitin (P99.0%, Fluka), 0.094 g oleic acid (P99.0%, 175
134 proach has made it possible to investigate reactivity and product Fluka) and 0.03 g of n-docosane (P98.0%, Fluka) as internal stan- 176
135 selectivity of different substrate molecules and product molecules dard. A 0.15 ml reference sample was taken out and 0.20 g 5 wt% 177
136 may be related distinctively to molecules in the model mixture. Pt/c-Al2O3 catalyst added to the remaining 10 wt% model fat mix- 178
Fuel (2011), doi:10.1016/j.fuel.2011.06.005
16 June 2011
A.T. Madsen et al. / Fuel xxx (2011) xxx–xxx 3
179 ture (consisting of 90 mol% tripalmitin and 10 mol% oleic acid) in obtained for the standard reactions conducted at 325 °C and 240
180 n-tetradecane. 20 bars H2 with 0.2 g 5 wt% Pt/c-Al2O3 are shown in Fig. 2. 241
181 The autoclave was then sealed at room temperature, charged As seen in Fig. 2, full conversion of the oleic acid to C17 and C18 242
182 with 14 bar hydrogen (P99.999%, Air Liquide), heated to 325 °C alkanes were achieved within 5 h. In comparison the tripalmitin 243
183 and mechanically stirred at 900 rpm. This procedure effectively re- was converted rather slowly, which is probably related to the lar- 244
184 sulted in a pressure-rise in the autoclave to 20 bars and also re- ger size of the molecule and more hindered access to the functional 245
185 sulted in activation of the catalyst. After 1 h of reaction the group to be reacted for the triester. It was observed that the yields 246
186 autoclave was rapidly cooled to below room temperature in an of pentadecane (C15) and heptadecane (C17) were more than an or- 247
187 ice-water bath, opened and a sample of about 0.15 ml withdrawn der of magnitude higher than those of hexadecane (C16) and octa- 248
188 for analysis. After closing the autoclave again it was purged and re- decane (C18), respectively. Hence, the pressure of H2 was 20 bars, 249
189 charged with hydrogen and reheated to 325 °C and the procedure the liberation of the carboxylate functionalities from the carbon 250
190 repeated to obtain samples after 2, 5 and 20 h of reaction. chains of the feed (decarboxylation and decarbonylation) domi- 251
191 In another experimental set the amount of tripalmitin and oleic nated the reactions for both the oleic acid and tripalmitin, in accor- 252
192 acid were varied while keeping the n-tetradecane amount fixed. dance with the findings of Murzin and co-workers [13,15]. 253
193 Variation of the reaction temperature was also performed to study Notably, the hydrogenation and decarboxylation reactions of 254
194 the dependence on conversion of the reactants, and the influence stearic acid have almost identical Gibbs free energies 255
195 of hydrogen pressure to examine the preference for decarboxyl- (DG 85 kJ/mol) at about 300 °C, while the Gibbs free energy 256
196 ation, decarbonylation or complete reduction. Furthermore, two of decarbonylation is less negative (DG 68 kJ/mol) [14]. This 257
197 other 5 wt% Pd/c-Al2O3 and 5 wt% Ni/c-Al2O3 catalysts were tested strongly support that the product distribution may be effected by 258
198 for selectivity towards the hydrogenation and decarboxylation catalyst. 259
199 reactions. It was observed during the experiments that the double bond in 260
oleic acid was rapidly saturated to give stearic acid. This proceeds 261
200 2.4. Product analysis
much faster and at lower temperatures than the ester/acid decar- 262
boxylation and the complete saturation of the fatty acid chain 263
201 The liquid reaction mixtures were quantified by an Agilent
was observed in all the experiments. 264
202 Technologies 6890N GC equipped with flame ionization detector
203 (FID), split/splitless injection system and a HP-5 capillary column
204 (J&W Scientific, 30 m 0.32 mm 0.25 lm, 5 mol% phenylmethyl
205 polysiloxane). Qualitative analysis was performed by GC–MS with 3.2. Variation and characterization of catalysts 265
206 an Agilent Technologies 6850 GC–MS with a HP-5MS column (J&W
207 Scientific, 30 m 0.25 mm 0.25 lm, 5 mol% phenylmethyl poly- Comparative screening of catalysts with 5 wt% Pd or Ni on 266
208 siloxane). A few gas samples of selected runs were taken out from c-Al2O3 revealed that these metals were also active for the deoxy- 267
209 the autoclave in a gas bag and injected with a gas syringe on an genation. Actually it could be observed that the Pd catalyst was 268
210 Agilent Technologies 6890N GC equipped with thermal conductiv- more active than the Pt catalyst, and also more selective to the 269
211 ity detector (TCD) and flame ionization detector (FID), split/split- C17 and C15 products (Table 1). The higher activity is probably re- 270
212 less injection system and a DB-1 capillary column (J&W lated to the higher molar loading of Pd (4.8 mol%) compared to 271
213 Scientific, 50 m 0.32 mm 0.5 lm of dimethylpolysiloxane). Pt (2.7 mol%) on the support where palladium metal area is double 272
214 All samples were silylated by addition of 20 ll N-methyl-N-tri- that of platinum, as measured by the CO pulse-chemisorption 273
215 methylsilyl-trifluoroacetamide (MSTFA) (P97.0%, Fluka) to each experiments (Table 2). Nickel, however, was less active than Pt 274
216 0.15 ml sample, followed by heating to 60 °C for 30 min and cool- and Pd as well as less selective to decarboxylation reactions, but 275
217 ing to room temperature before running the analysis on GC and the catalyst particles were also larger than Pd and Pt particles on 276
218 GC–MS. Free fatty acids (FFAs) were hereby converted to trimeth- average (Table 2) and for this reason may be less reactive. The re- 277
219 ylsilyl esters which were sufficiently volatile and thermally stable sults obtained after 5 h can be seen in Table 1. The strong tendency 278
220 to allow analysis with reduced peak tailing in the gas chromato- towards decarboxylation and decarbonylation for Pd and Pt cata- 279
221 gram [31]. It should be noted that a split/splitless injector cannot lysts as well as the relative tendency in catalyst activity has also 280
222 itself be used in directly quantifying the amount of tripalmitin in been observed by Murzin et al. [13–15], although it was indicated 281
223 a given hydrocarbon sample. Discrimination will result when mo- that Ni should favor split-off of CO or CO2 to a somewhat higher de- 282
224 lar masses of the compounds gets too high, which may also make gree. This was also presented by Kubička et al. [21], though a 283
225 split/splitless GC analyses of fats cumbersome. Instead we used the
226 yield of pentadecane and hexadecane to calculate the tripalmitin
227 conversion, as these alkanes are measured quantitatively with
228 good precision by split/splitless GC injection and FID detection.
229 This assumption could be justified since no other compounds were
230 observed above impurity threshold.
231 During the experimentation stearic acid was observed as an
232 intermediate by hydrogenation of oleic acid. The sum of oleic acid
233 and stearic acid concentrations calculated from the GC-FID chro-
234 matograms has been used to denote the amount of unconverted
235 fatty acid.
236 3. Results and discussion
237 3.1. Time series

Fig. 2. Hydrodeoxygenation time series conducted at 325 °C and 20 bar H2 with
0.2 g 5 wt% Pt/c-Al2O3. Legend: Tripalmitin (– –) and its products pentadecane (–
238 Results of conversion and yield were plotted against the time of j–) and hexadecane (–h–), oleic acid (– + –) and its products heptadecane (–N–)
239 sampling to illustrate time series of each catalytic run. The results and octadecane (–D–).
Fuel (2011), doi:10.1016/j.fuel.2011.06.005
16 June 2011
Table 1 4% after 5 h, thus confirming that a catalyst is necessary to perform 316

Q1 Catalysts for hydrodeoxygenation at 20 bar H2, 325 °C, 5 h reaction over 0.2 g catalyst. the sought reaction (Table 3, entry 5). 317
Entry Catalyst Tripalmitin Oleic acid As the support material c-alumina itself could play a role in the 318
(CO2 + CO)/ Conversion (CO2 + CO)/ Conversion reaction, it was also important to establish if such an influence was 319
H2b (%) H2b (%) present. By using only the c-Al2O3 as additive under H2 atmo- 320
1 5 wt% Pt/c- 21.9 46.2 23.3 100

sphere a total yield of alkanes of 7% was obtained from oleic acid 321
Al2O3a and 6% from tripalmitin after 5 h, while respectively 2% and 5% 322
2 5 wt% Ni/c- 4.8 10.4 1.4 76.8 were obtained under N2 atmosphere (Table 3, entries 3 and 4). 323
Al2O3a When using Pt/Al2O3 as catalyst under N2 pressure the conversion 324
3 5 wt% Pd/ 28.6 71.1 28.5 100
of oleic acid after 5 h reached 99% (Table 3, entry 2). Thus hydrogen 325
c-Al2O3a
was not directly necessary to convert oleic acid to hydrocarbons 326
b
Ratio of decarboxylation + decarbonylation to hydrogenation. with Pt catalyst, but unsaturated compounds were produced in- 327
stead of linear alkanes. Under these conditions the conversion of 328
tripalmitin was only 4% after 5 h. 329
Table 2
Catalyst and support characterization.
Entry Catalyst mol% BET- Pore Mean metal Metal 3.4. Feed variation 330
metala area volume particle size surf. area
(m2/g) (ml/g) (nm) (m2/g)
The free fatty acids reacted faster to hydrocarbons than triglyc- 331
1 c-Al2O3 — 255 0.607 — —
erides, as shown above. However, the ratio of decarboxylation/ 332
2 5 wt% 2.68 251 0.575 5.4 2.82
Pt/c- decarbonylation to hydrogenation was almost identical for both 333
Al2O3 triglyceride and free fatty acid. This implied that triglycerides re- 334
3 5 wt% 4.80 236 0.539 4.6 5.69 acted via an intermediate, which was likely palmitic acid formed 335
Pd/c- by hydrolysis or partial hydrogenation of the triglyceride 336
Al2O3
4 5 wt% 8.38 217 0.523 8.2 4.81
molecules. 337
Ni/c- A catalytic run with only oleic acid as reactant was then per- 338
Al2O3 formed at 325 °C and 20 bar H2 using 5 wt% Pt/c-Al2O3 catalyst. 339
Here only heptadecane (major product) and octadecane (minor 340
a
Calculated molar percentage of active metal on the c-Al2O3 support.
product) was observed as products during the first 5 h of reaction. 341
This observation confirmed that formation of n-heptadecane and 342
284 poorer conversion of carboxylic acids over Ni/Al2O3 than over sup- n-octadecane resulted selectively from linear monoacids of 18 C- 343
285 ported Pd catalysts has previously been reported [36]. atoms. No palmitic acid was observed. 344
The analogous experiment with only tripalmitin was also con- 345
ducted. Here it was confirmed that the treatment of tripalmitin re- 346
286 3.3. Control experiments sulted in the formation of pentadecane (major product) and 347
hexadecane (minor product), and palmitic acid furthermore ap- 348
287 A key step in the experimentation was to establish if deoxygen- peared during the reaction. These results implied that palmitic acid 349
288 ation of the model feed actually resulted in production of linear- was at least one of the intermediates formed during tripalmitin 350
289 chain alkanes only by the mentioned overall schemes in Fig. 1a– conversion. 351
290 d. To examine this, we therefore performed control experiments A likely reaction for formation of the intermediate is partial 352
291 to elucidate the influence of the support material, the gas atmo- hydrogenation, as shown in Fig. 3. Since the degree of hydrogena- 353
292 sphere used, the temperature and the method of preparation of tion was less pronounced compared to decarboxylation, the partial 354
293 the Pt catalyst. pressure of water was quite low. On the other hand the hydrogen 355
294 First a Mason–Boudart-test was made to investigate if diffusion pressure was high, so it is likely that propandiyl-dipalmitate and 356
295 resistance played a role in the catalytic tests. The 5 wt% Pt/c-Al2O3 palmitic acid are formed (Fig. 3). This behavior and mechanism 357
296 of <180 lm size was crushed down and sieved to a size of <50 lm have previously been described in the literature [13,14]. 358
297 and used for deoxygenation of a mixture of tripalmitin and oleic The hydrogenated di-ester may then have been further hydro- 359
298 acid. This produced within a few percent the same yields of result- genated to the mono-ester and palmitic acid, and again to finally 360
299 ing C15–C18 alkanes (not shown) after 1, 2, 5 and 20 h, so diffusion have released propane and the last acid molecule, while the pal- 361
300 limitation was deemed unimportant. mitic acid, parallel to oleic or stearic acid, may have underwent 362
301 Then it was investigated whether or not formed alkanes and the decarboxylation, decarbonylation or hydrogenation to pentadec- 363
302 substrates underwent C–C scission at positions other than the es- ane or hexadecane. 364
303 ter/acid functionality or isomerization, by performing a reference

304 experiment with Pt/Al2O3 at 20 bar hydrogen and 325 °C without Table 3
305 reactants. From this experiment it was confirmed that less than Conversion of oleic acid with supported catalyst, with support and without any
306 1% of n-tetradecane was isomerized after treatment for 20 h. It is catalyst in control experiments at 20 bar gas 325 °C, 5 h reaction over 0.2 g catalyst.
307 known that alkanes may undergo C–C-scission or isomerization Entry Catalyst Gas Oleic acid Tripalmitin
308 under these circumstances, and though the degree of isomerization atmosphere conversion (%) conversion (%)
309 was found to be quite low, it must be noted that alkanes are not 1 5 wt% Pt/c- H2 100 46
310 completely unaltered by Pt catalysts at these temperatures, which Al2O3
311 has also been reported previously for similar experiments [14,15]. 2 5 wt% Pt/c- N2 99 4
312 To evaluate the alteration of the reactants by the temperature a Al2O3
3 c-Al2O3 H2 7 6
313 reaction run was also performed with reactants at 325 °C and 4 c-Al2O3 N2 2 5
314 20 bar H2 without added catalyst. This experiment resulted in a 5 None H2 3 4
315 conversion of oleic acid of about 3% and of tripalmitin of about
Fuel (2011), doi:10.1016/j.fuel.2011.06.005
16 June 2011
A.T. Madsen et al. / Fuel xxx (2011) xxx–xxx 5
365 3.5. Variation of the temperature
366 Experiments were further performed at different temperatures

367 to evaluate the performance of the Pt catalyst with respect to the
368 model feed. The conversions found for oleic acid and tripalmitin
369 after 5 h of reaction are plotted against temperature in Fig. 4.
370 The conversions of both reactants were below 5% after 5 h at
371 250 °C and 275 °C. However, at 300 °C about half of the oleic acid
372 was converted while only around one tenth of the tripalmitin
373 was converted to alkanes. At 325 °C the conversion of the oleic acid
374 to alkanes reached 100%, and the conversion of tripalmitin reached
375 40%. The tripalmitin only reached full conversion after 5 h when
376 using a temperature of 375 °C, clearly showing that the triglyceride
377 was more reluctant to react under the applied conditions. Fig. 4. Conversion as a function of temperature between 250 and 375 °C with
20 bar H2 over 0.2 g Pt/c-Al2O3 after 5 h. Legend: Oleic acid (light gray) and
378 Murzin and co-workers have showed that not just CO2 but also
tripalmitin (dark gray).
379 CO was formed in the mixture during the reaction of stearic acid
380 and ethyl stearate to alkanes [14,15]. It can thus be suspected that
381 decarbonylation also takes place during the reaction studied here. Table 4
382 It should be noted that if either CO or CO2 are formed exclusively Thermodynamics of water–gas shift and methanation reactions at 300 °C [14].
383 during the course of the reaction it may not necessarily be ob- Reaction DH573 K (kJ/mol) DG573 K (m2/g)
384 served, because the catalysts could be active for the water–gas
(a) CO + H2O M H2 + CO2 39.2 17.6
385 shift (WGS) equilibrium in presence of hydrogen and water vapor (b) CO + 3H2 M CH4 + H2O 216.4 78.8
386 (Table 4a) so that a mixture of CO and CO2 will always form. (c) CO2 + 4H2 M CH4 + 2H2O 117.2 61.2
387 According to thermodynamics also methanation of both CO and
388 CO2 with H2 (Table 4b and c, respectively) may occur around
389 300 °C, as also suggested by Murzin et al. [14]. Although platinum
towards full reduction is not improved by adding more H2, and the 416
390 is known not to be very active in reactions involving CO due to high
oleic acid conversion declined. This behavior is unexpected and 417
391 dissociation energy of CO [37], methanation does play a role in the
may call for a more elaborated study, but possibly the dissolved 418
392 studied gas–solid reactions. Hence, gas-phase analysis of reaction
H2 in the alkane solvent reached saturation while the higher 419
393 mixture from tripalmitin and oleic acid deoxygenation over
hydrogen pressure pushed the water–gas shift equilibrium to- 420
394 5 wt% Pt/c-Al2O3 at 325 °C and 20 bar showed that in all cases
wards increased CO formation from CO2. CO is known to preferen- 421
395 about 60% CH4 was formed and about 40% as a mixture of CO
tially adsorb on surfaces and may thus deactivate the surface. 422
396 and CO2. However, here the reactor interior (i.e., stainless steel)
Previously it has been observed that stearic acid react via decar- 423
397 may also have contributed since iron is a rather good Fischer–Trop-
boxylation over supported Pd, while esters may react via both 424
398 sch catalyst [37]. Some N2 and Ar was present as well due to leak-
decarboxylation or decarbonylation over Pd and Pt depending on 425
399 age of air. No C3-species (e.g. propane) from the tripalmitin
the process parameters [12–15] Also competitive adsorption of 426
400 conversion were detected, though. These may be mixed in the li-
hydrogen may have hindered the acid adsorption or the hydrogen 427
401 quid tetradecane phase at elevated pressures.
may have inhibited the decarboxylation of oleic or stearic acid. 428
402 It must be noted that methanation is problematic in this process
As previously mentioned (Section 3.3) H2 may be necessary to 429
403 because methane is not useful as a fuel in the resulting low concen-
avoid deactivation of the catalyst by scavenging the surface from 430
404 trations, it is a much worse greenhouse gas than CO2, and metha-
deactivating species [13,30,38]. Such deactivation is reported to 431
405 nation consume more hydrogen than full reduction of the fatty
take place over a Pd/C catalyst by formation of aromatics from 432
406 acids. If methanation cannot be avoided in an industrial process,
unsaturated compounds [30,38], and thus likely also over platinum 433
407 it is more attractive to fully reduce the fatty acid chains to alkanes
catalysts. Formation of alkenes was observed over a sulphided 434
408 as shown in Fig. 1b and d.
NiMo-catalyst as part of a deoxygenation mechanism from alde- 435
hydes and alcohols in hydrogen [18]. C17-alkenes and heavier ke- 436
409 3.6. Variation of hydrogen pressure tones have previously been observed to form from methyl 437
410 The pressure variation experiments are summarized in Fig. 5.

411 Highest yield of heptadecane after 1 h reaction at 325 °C was ob-
412 tained with 9 bar H2. When the pressure of H2 was increased above
413 9 bars the amount of heptadecane (C17H36) formed from oleic acid
414 decreased, while the formed amounts of hexadecane (C16H34) and
415 octadecane (C18H38) remained low. This implied that the selectivity
O O
O O C15H31 O O C15H31
H2
O O C15H31 C15H31COOH + O C15H31
cat.
O C15H31 CH3 Fig. 5. Yields of C15–C18 alkanes after 1 h reaction at 325 °C with 0.2 g 5 wt% Pt/c-
Al2O3. Legend: Pentadecane (–j–), hexadecane (–h–), heptadecane (–N–), octade-
Fig. 3. Partial hydrogenation of tripalmitin to yield intermediates palmitic acid and cane (–D–). Dotted lines represent products of tripalmitin conversion and solid
propandiyl-dipalmitate. lines are products of oleic acid conversion.
Fuel (2011), doi:10.1016/j.fuel.2011.06.005
16 June 2011
438 stearate over Pt/Al2O3 in He atmosphere, which was suspected to [6] Meher LC, Vidya Sagar D, Naik SN. Technical aspects of biodiesel production by 499
transesterification – a review. Renew Sust Energy Rev 2006;10:248–68. 500
439 lead to deactivation [39]. Accordingly, it is likely that the high 501
[7] Van Gerpen J. Biodiesel processing and production. Fuel Process Technol
440 activity for conversion of oleic acid in inert atmosphere (Section 2005;86:1097–107. 502
441 3.3) yielding unsaturated compounds in turn may cause catalyst [8] Mittelbach M, Remschmidt C. Biodiesel: The comprehensive handbook. 3rd ed. 503
Graz, (Austria), Martin Mittelbach; 2006. 504
442 deactivation. 505
[9] Huber GW, Iborra S, Corma A. Synthesis of transportation fuels from biomass:
443 The pentadecane (C15H32) formation was more or less constant chemistry, catalysts, and engineering. Chem Rev 2006;106:4044–98. 506
444 at 12% when the pressure of H2 increased from 9 to 40 bars of [10] Stumborg M, Wong A, Hogan E. Hydroprocessed vegetable oils for diesel fuel 507
improvement. Bioresource Technol 1996;56:13–8. 508
445 hydrogen. H2 may not be the rate-limiting component in the con-
[11] Huber GW, O’Connor P, Corma A. Processing biomass in conventional oil 509
446 version of tripalmitin, but some H2 was necessary for the conver- refineries: production of high quality diesel by hydrotreating vegetable oils in 510
447 sion as no H2 resulted in a yield of pentadecane of 3%. This could heavy vacuum oil mixtures. Appl Catal A 2007;329:120–9. 511
448 be explained by the mechanistic suggestion in Section 3.4, assum- [12] Kubičková I, Snåre M, Eränen K, Mäki-Arvela P, Murzin DY. Hydrocarbons for 512
diesel fuel via decarboxylation of vegetable oils. Catal Today 513
449 ing that hydrogen is needed to split off the fatty acids from the gly- 2005;106:197–200. 514
450 cerides for further reaction [13]. Thus it appeared that the [13] Mäki-Arvela P, Kubičková I, Snåre M, Eränen K, Murzin DY. Catalytic 515
451 optimum hydrogen pressure under the examined reaction condi- deoxygenation of fatty acids and their derivatives. Energy Fuels 516
2007;21:30–41. 517
452 tions was around 10 bars in terms of the yields of pentadecane [14] Snåre M, Kubičková I, Mäki-Arvela P, Eränen K, Murzin DY. Heterogeneous 518
453 and heptadecane. catalytic deoxygenation of stearic acid for production of biodiesel. Ind Eng 519
Chem Res 2006;45:5708–15. 520
[15] Snåre M, Kubičková I, Mäki-Arvela P, Eränen K, Wärnå J, Murzin DY. Production 521
454 4. Conclusion of diesel fuel from renewable feeds: kinetics of ethyl stearate decarboxylation. 522
Chem Eng J 2007;134:29–34. 523
[16] Koskinen M, Sourander M, Nurminen M. Apply a comprehensive approach to 524
455 Hydrodeoxygenation of a model fat mixture composed of oleic
biofuels. Hydrocarb Process 2006;85:81–6. 525
456 acid and tripalmitin has been studied with an alumina-supported [17] Mikkonen S. Catalytic upgrading of biorefinery oil from micro-algae. 526
457 platinum catalyst. The conversion of the free fatty acids took place Hydrocarb Process 2008;87:63–6. 527
458 faster than that of triglycerides, a property that may be ascribed to [18] Donnis B, Egeberg RG, Blom P, Knudsen KG. Hydroprocessing of bio-oils and 528
oxygenates to hydrocarbons. Understanding the reaction routes. Top Catal 529
459 the difference in metal particle sizes and diffusivity of the reac- 2009;52:229–40. 530
460 tants, besides the intrinsic difference in reactivity for esters and [19] Senol O, Viljava T, Krause A. Hydrodeoxygenation of aliphatic esters on 531
461 carboxylic acids. By excluding respectively oleic acid and tripalmi- sulphided NiMo/c-AlO and CoMo/c-AlO catalyst: the effect of water. Catal 532
Today 2005;106:186–9. 533
462 tin in comparative experiments it was demonstrated that triglycer- 534
[20] Senol OI, Ryymin EM, Viljava T-R, Krause AOI. Reactions of methyl heptanoate
463 ides reacted to alkanes via their corresponding free fatty acids, and hydrodeoxygenation on sulphided catalysts. J Mol Catal A 2007;268:1–8. 535
464 that this reaction determined the overall reaction rate under these [21] Kubička D, Šimáček P, Kolena J, Lederer J, Šebor G. Catalytic transformations of 536
vegetable oils into hydrocarbons. In: ICP2007 proceedings – 43rd international 537
465 conditions. 538
petroleum conference, Bratislava, Slovakia; 2007. p. 1–7.
466 At reaction temperatures below 300 °C the conversions of both [22] Kubička D, Kaluža L. Deoxygenation of vegetable oils over sulfided Ni, Mo and 539
467 triglycerides and free fatty acids were low, the latter being faster. NiMo catalysts. Appl Catal A 2010;372:199–208. 540
[23] Šimáček P, Kubička D, Šebor G, Pospíšil M. Hydroprocessed rapeseed oil as a 541
468 At 325 °C and above the reaction proceeded fast for both triglycer-
source of hydrocarbon-based biodiesel. Fuel 2009;88:456–60. 542
469 ide and free fatty acid. Gas-phase analysis revealed that methana- [24] Kubička D, Bejblová M, Vlk J. Conversion of vegetable oils into hydrocarbons 543
470 tion and probably also water–gas-shift contribute a role in altering over CoMo/MCM-41 catalysts. Top Catal 2010;53:168–78. 544
[25] Kubička D, Šimáček P, Žilková N. Transformation of vegetable oils into 545
471 the atmosphere during reaction progression, maybe from interfer-
hydrocarbons over mesoporous-alumina-supported CoMo catalysts. Top 546
472 ence with the reactor interior. Formation of unsaturated hydrocar- Catal 2009;52:161–8. 547
473 bons dominated in absence of hydrogen, while a hydrogen [26] Do PT, Chiappero M, Lobban LL, Resasco DE. Catalytic deoxygenation of 548
474 pressure between 9 and 20 bars was found to be optimal for the methyl-octanoate and methyl-stearate on Pt/Al2O3. Catal Lett 2009;130:9–18. 549
[27] Danuthai T, Jongpatiwut S, Rirksomboon T, Osuwan S, Resasco DE. Conversion 550
475 reaction, depending on the reactant being either free fatty acid or of methylesters to hydrocarbons over an H-ZSM5 zeolite catalyst. Appl Catal A 551
476 triglyceride. 2009;361:99–105. 552
477 The hydrodeoxygenation of oils and fats is a promising and [28] Sooknoi T, Danuthai T, Lobban LL, Mallinson R, Resasco DE. Deoxygenation of 553
methylesters over CsNaX. J Catal 2008;258:199–209. 554
478 flexible way of providing biofuels, and examination on model feeds 555
[29] Simakova I, Simakova O, Mäki-Arvela P, Murzin DY. Decarboxylation of fatty
479 – such as a mixture of tripalmitin and oleic acid applied in the acids over Pd supported on mesoporous carbon. Catal Today 2010;150:28–31. 556
480 present work – is a suitable approach to get detailed insight into [30] Rozmysłowicz B, Mäki-Arvela P, Lestari S, Simakova OA, Eränen K, Simakova I, 557
et al. Catalytic deoxygenation of tall oil fatty acids over a palladium- 558
481 the factors determining the production of renewable long-chain 559
mesoporous carbon catalyst: a new source of biofuels. Top Catal
482 hydrocarbons. 2010;53:1274–7. 560
[31] Lestari S, Mäki-Arvela P, Eränen K, Beltramini J, Lu GQM, Murzin DY. Diesel- 561
like hydrocarbons from catalytic deoxygenation of stearic acid over supported 562
483 Acknowledgement
Pd nanoparticles on SBA-15 catalysts. Catal Lett 2009;134:250–7. 563
[32] Lestari S, Simakova I, Tokarev A, Mäki-Arvela P, Eränen K, Murzin DY. Synthesis 564
484 A.T.M is grateful for the funding of his PhD stipend in the of biodiesel via deoxygenation of stearic acid over supported Pd/C catalyst. 565
Catal Lett 2008;122:247–51. 566
485 Waste-2-Value innovation consortium through the Danish Agency
[33] Lestari S, Mäki-Arvela P, Simakova I, Beltramini J, Lu GQM, Murzin DY. 567
486 for Science, Technology and Innovation. Catalytic deoxygenation of stearic acid and palmitic acid in semibatch mode. 568
Catal Lett 2009;130:48–51. 569
487 References [34] Lestari S, Mäki-Arvela P, Beltramini J, Lu GQM, Murzin DY. Transforming 570
triglycerides and fatty acids into biofuels. ChemSusChem 2009;2:1109–19. 571
[35] Pinna F. Supported metal catalysts preparation. Catal Today 1998;41:129–37. 572
488 [1] The European Union Directive. 2009/28/EC, on the promotion of the use of
[36] Maier WF, Roth W, Thies I, Schleyer PVR. Chem Ber 1982;115:808–12. 573
489 energy from renewable sources. Offic J EU 2003;12:1–20.
[37] Andersson M, Bligaard T, Kustov A, Larsen K, Greeley J, Johannessen T, et al. 574
490 [2] Snåre M, Murzin DY. Reply to ‘‘comment on ‘heterogeneous catalytic
Toward computational screening in heterogeneous catalysis: Pareto-optimal 575
491 deoxygenation of stearic acid for production of biodiesel’’’. Ind Eng Chem Res
methanation catalysts. J Catal 2006;239:501–6. 576
492 2006;45:6875.
[38] Lestari S, Mäki-Arvela P, Bernas H, Simakova O, Sjöholm R, Beltramini J, et al. 577
493 [3] Ma F, Hanna MA. Biodiesel production: a review. Bioresource Technol
Catalytic deoxygenation of stearic acid in a continuous reactor over a 578
494 1999;70:1–15.
mesoporous carbon-supported Pd catalyst. Energy Fuels 2009;23:3842–5. 579
495 [4] Knothe G, Van Gerpen J, Krahl J. The biodiesel handbook. (Urbana, Illinois,
[39] Do PT, Chiappero M, Lobban LL, Resasco DE. Catalytic deoxygenation of 580
496 USA): AOCS Press; 2005.
methyl-octanoate and methyl-stearate over Pt/Al2O3. Catal Lett 581
497 [5] Kulkarni MG, Dalai AK. Waste cooking oil – an economical source for biodiesel:
2009;130:9–18. 582
498 a review. Ind Eng Chem Res 2006;45:2901–13.
583
Fuel (2011), doi:10.1016/j.fuel.2011.06.005
Review
Challenges and perspectives for catalysis in production of

diesel from biomass
Biofuels (2011) 2(4), xxx–xxx
Anders Theilgaard Madsen1, Helle Søndergaard1, Rasmus Fehrmann1 & Anders Riisager†1
The production of biofuels is expected to increase in the future due to environmental concerns, accelerating
oil prices and the desire to achieve independence from mineral oil sources. Of the proposed methods for
diesel production from biomass, the esterification and transesterification of plant oils or waste fats with
methanol is the most prominent and has been applied industrially for a decade. Homogeneous acid and
base catalysis is normally used, but solid acids, solid bases, ionic liquids and lipases are being developed as
replacements. Hydrodeoxygenation of vegetable oils has likewise been commercialized. Diesel from biomass
may also be produced by catalytic upgrading of bio-oils from flash pyrolysis, by aqueous-phase reforming
of carbohydrates into non- or mono-functionalized hydrocarbons via consecutive reduction-condensation
reactions, or by gasification of biomass to synthesis gas of CO and H2 and liquefaction to alkanes via Fischer–
Tropsch synthesis. Here, the current challenges and perspectives regarding catalysis and raw materials for
diesel production from biomass are surveyed.
Diesel are usually also the most environmentally benign to

Biomass raw materials & conversion technologies employ, the use of such feedstock is highly desirable.
The production of diesel from biomass is growing Utilization of biomass to produce diesel oils has under-
worldwide. This growth is driven by, amongst other gone substantial research in recent years and a number
factors, capital interests, environmental concerns and of catalytic methods exists; fats and oils can either be
the desire to make national fuel supplies more indepen- transesterified with alcohol to form FAMEs [3–5] , deoxy-
dent of global petroleum supply and price. The EU fuel genated at moderate temperatures with hydrogen to form
directive 2003/30/EC mandated in 2003 that a share n-alkanes [6–10] or cracked at elevated temperatures to
of 5.75% energy content should derive from biofuels form a hydrocarbon blend [11–13] . Carbohydrates may be
in 2010, rising to 10% by 2020 [1] . At the same time, converted by aqueous-phase reforming (APR) to form
the EU has opted for a promotion of biofuel produc- unfunctionalized or monofunctional hydrocarbons
tion from waste products and a discouragement of ‘bad [14–16] . Biomass in general may be gasified to syn-gas (a
biofuels systems’, that is, exclusion of biofuels produced mixture of carbon monoxide and hydrogen) and lique-
with a low or even negative displacement of carbon diox- fied via Fischer–Tropsch synthesis (FTS) [17,18] . Thermal
ide from the atmosphere or biofuels production trigger- depolymerization either by flash pyrolysis or hydrother-
ing deterioration of sensitive wildlife habitats and eco- mal upgrading can yield a bio-oil that can be further
systems [2] . Furthermore, as transportation fuels based upgraded by cracking or hydrogenation [19–22] . A graphi-
on fossil feedstocks are usually the cheapest petroleum cal overview of these techniques can be seen in Figure 1.
products, the production processes for sustainable fuels The term ‘biodiesel’ usually refers solely to fatty acid
must also be economically feasible and the feedstock as alkyl esters and in particular the fatty acid methyl esters
cheap as possible. As waste products (second-generation (FAMEs) [23] . The market share of FAMEs is the largest of
resources) represent the cheapest feedstock available and the available diesel oils from biomass, but it is not the only
Centre for Catalysis and Sustainable Chemistry, Department of Chemistry, Building 207, Technical University of Denmark, DK 2800 Kgs. Lyngby, Denmark
1
Author for correspondence: Tel.: +45 4525 2233; Fax: +45 4588 3136; E-mail: E-mail: [email protected]
†
future science group 10.4155/BFS.11.115 © 2011 Future Science Ltd ISSN 1759-7269 1
Review Madsen, Søndergaard, Fehrmann & Riisager
the various catalytic technologies

Natural Wood, in terms of fuel properties, biomass
Fats and oils or straw, Bio-sludge Carbohydrates
bio-gas waste
resources and process intensities.
Diesel properties
HTU/CatLiq™
Fla
+H2O
Diesel may be defined as liquid fuels
sh
-CO2
Aqueous-phase reforming (APR)

py
that can be used in an unmodified
Gasific
Gasification
+H2O
rol
-gas
(hyd
-CO2 diesel engine and has fuel properties
y
-char
sis
(trans-)esterification
+H2O +H2O similar to those of a middle distil-

a
ro)-d
tion
-CO2 -CO2
Bio-oil late (or so-called middle-cut) from
eox y
s ific (CxHyOz) petroleum oil refining, from which

Ga
gena
+CH3OH most of the world’s diesel fuel is pro-

-glycerol duced. Another part of it is produced
tion
Syn-gas
Ke
Deoxygenation via hydrocracking of vacuum gas oil
(CO + 2H2)
ton
+H2
from the bottom of the distillation
isa
+H2 -H2O
tio
-H2O -CO -H O column. The number of carbon
n
2 2
-CO2
FTS
-CO2 +H2O atoms in the fuel ranges from 12 to

18 and consists primarily of alkanes.
The most important properties for
Alcohols, diesel oil as well as for characterizing
Aromatics
FAME n-alkanes ketones,
cycl. alkanes
furans
alternative diesel fuel relative to the
norm for petroleum diesel fuel are the
cetane number describing the ignition
Figure 1. Overview of biomass conversion technologies leading to diesel oils. properties, lower heating value (con-
nected with the density) describing
the energy content and the cold flow
Key terms share. Accordingly, suggestions for properties described by the cold filter
Bio-oil: Tar fraction from biomass wider definitions regarding naming plugging point. Other relevant measures for describing
pyrolysis. Contains numerous of fuels have been suggested [24–26] . the cold properties are pour point and cloud point.
compounds: ketones, aldehydes,
carboxylic acids, aromatic ethers, For the sake of clarity, we will refer The contents of unwanted sulfur and aromatics are
phenols, alcohols and polyfunctional to the technical–chemical terms for a measure of the environmental quality of the fuels.
molecules in emulsified water. It is each fuel in this review. Importantly, The long-term storage stability is relevant for refiners
unsuited for fuelling engines, but may biodiesel is not pure vegetable oil or and end-users. The flash point is a measure of safe han-
be catalytically hydrodeoxygenated to
usable fuels. fat. Engine and fuel system require dling temperature, while viscosity and lubricity describes
modifications to be able to run on behavior in the fuel system and engine.
Cetane number: Indication of ‘ease of
ignition’ of the diesel measured in a test such fuels and these changes are
engine. Diesel is injected in a often illegal for road vehicles and Transesterification & esterification of fats & oils
compression-ignition engine when the lead to annulment of engine warran- The most applied and studied method for producing
piston is approaching the cylinder top
dead center (near full compression) and
ties and specifications for emissions. diesel from biomass is the alcoholysis or transesterifi-
it must ignite immediately. Specifies the Combustion of vegetable oil in the cation of fats and oils to yield fatty acid alkyl esters.
percentage of cetane (n-hexadecane, engine often becomes incomplete Usually the alcohol is methanol due to lowest cost, ease
n-C16H34) in a mixture with iso´-cetane and both particle and NOx emissions of separation and high reaction rate, whereby FAMEs
(2,2,4,4,6,8,8-heptamethylnonane,
i-C16H34) that gives similar engine
are reported to rise. Plant oils are also are formed [33] . FAME can be used directly in the engine
performance as the test fuel. The cetane often too viscous to be pumped in without modification and be mixed in all ratios with pet-
number must be > 51 in Europe and the fuel and injector systems [3] . rochemical diesel fuel. The global production of FAME
> 47 in North America. Linear chained A considerable number of in 2008 reached 11.1 Mton – of these, the EU countries
alkanes ignite easier than branched
ones and high cetane number gives reviews on production of diesel produced 7.76 Mton and the USA 2.33 Mton [5,201,202] .
fast ignition. from biomass have been published
Lower heating value: Describes the in recent years, especially regard- Industrial production of FAME
energy density in the fuel. Usually ing FAME [27–30] , but also on more Fats and oils consist primarily of triglycerides (TGs).
around 43 MJ/kg or 36 MJ/l for advanced fuel production [31,32] . These, as well as mono- and di-glycerides can be trans-
diesel oils.
Here, however, we highlight the esterified into FAMEs by methanol (Figure 2A) . The
catalytic technologies necessary for the conversion of sources of the TGs are all types of vegetable oils, animal
various biomasses into diesel fuel, and we compare fats or waste greases [34] . A successful transesterification
2 Biofuels (2011) 2(4) future science group

Challenges & perspectives for catalysis in production of diesel from biomass Review
leads to two phases: glycerol and excess methanol A O O

separated from FAME. Notably, equimolar glycerol is
obtained as a byproduct for every three FAME mol-
O O R H3CO O R
OH
ecules produced. The oil and FAME phase (nonpolar) O O R' + 3 CH3OH OH + H3CO O R'
base
and the alcohol phase (polar) are not directly miscible OH
O R'' H3CO R''
with each other. In industrial practice, the reaction is
normally performed at approximately 60–70°C and B O O
with a few wt% of homogeneous alkali metal hydrox- HO R + CH3OH H2O + H3CO
acid R
ides or methoxides as base catalyst. Typically potas-
sium hydroxide is preferred as it is nonhygroscopic and
cheap [35,36] . Figure 2. Production of fatty acid methyl esters. (A)
The transesterification is equilibrium driven and Base-catalyzed transesterification of a triglyceride with
in fact reversible. However, due to use of two-three methanol to form glycerol and fatty acids methyl esters.
times molar excess of methanol and the immiscibility (B) Acid-catalyzed esterification of free fatty acid with
of FAME and alcohols, the transesterification of oils methanol to form water and fatty acid methyl esters.
typically reaches over 99% yield after a few hours of
reaction [33] .
Initially the reaction is slow and dependent on vigor- a strong acid and very hygroscopic. Key terms
ous mixing of the two phases (e.g., by stirring) to get a The mixture of methanol and fat is
Cloud point: Temperature where, upon
large interfacial area between the two phases. During esterified at 60–70°C with stirring cooling, visible diesel crystals start to
reaction, di- and mono-glycerides are formed and these for a few hours to bring the content of form as a cloud wax; cloud point is
compounds act as emulsifiers for the reaction mixture. FFAs under 0.5 wt% (as is shown in usually a bit higher than cold filter
plugging point.
Once the transesterification is nearing completion, the Figure 2B) prior to transesterification.
two phases start to separate again (settling). Industrially In principle, the transesterifica- Storage stability: Arbitrary evaluation
of the ease of long-term storage of the
this reaction is normally done in a stirred batch-reaction can be catalyzed by both bases diesel fuel.
tor with a separate settling-tank, since the settling is and acids, leaving room for a one-
Flash point: Safety measure indicating
often slower than the transesterification reaction [37,38] . step acid-catalyzed process, but the the temperature of the liquid fuel at
However, continuous reactors for FAME synthesis have base-catalyzed reaction is several which a flash can ignite vapors over it.
appeared over the last couple of years using homoge- orders of magnitude faster than Viscosity: Measure (in mm2/s) of the
neous catalysts. This is advantageous in terms of process the acid-catalyzed reaction [33] . ‘thickness’ of the liquid fuel. Too high
operation and control, labor demand, supervision and Temperatures of at least 150–200°C viscosity makes pumping difficult.
optimization of product quality. are needed to perform transesteri- Lubricity: Lubricating properties of the
The homogeneous, base-catalyzed transesterification fication with sulfuric acid to get a diesel, measured as the ‘wear’ imposed
when rubbing a steel ball against a steel
works efficiently, but oils and fats may contain free fatty reasonable space time yield, thus disk immersed in the fuel.
acids (FFAs). Waste fats have especially high amounts requiring pressurized equipment
Cold filter plugging point:
of FFAs: 2–7 wt% for used cooking oils, 5–30 wt% (due to the vapor pressure of metha- Temperature at which a test filter is
for waste animal fats and abattoir waste and some trap nol, bp.: 64.6°C) and therefore it is plugged with diesel wax crystals; the
greases often above 50 wt% [3] . When the feedstock usually considered undesired. The most common way to report cold
contains more than approximately 0.5–1 wt% of FFAs, esterification by homogeneous acid properties of diesel. Can be lowered by
additives, which is generally necessary
these will react with the basic catalyst and form soap also leads to other challenges: in cold climates.
(i.e., FFA salt); whereby: Neutralization and basification Pour point: Measured by first cooling a
Base otherwise intended for transesterification will before the transesterification and diesel sample to solid and then heating
be consumed; neutralization before final separa- it up; the temperature within regular
intervals at which the diesel starts
tion of the alcohol phases must be flowing is the pour point.
The formed soap emulsifies the phases and hinders
performed. This yields large
the settling and separation after reaction;
amounts of alkali salts as byproducts (usually K 2SO4
Water formed by saponification lead to basic hydro- and KHSO4) with no immediate use;
lysis of the glycerides yielding more soap.
Water formed during esterification and neutraliza-
For these reasons, the FFAs must first be esterified tion lead to saponification in the consecutive basic
with methanol to form FAMEs. Esterifications are acid transesterification reaction;
catalyzed and in the biodiesel industry a few wt% sul- Homogeneous catalysts must be added constantly,
furic acid (H2SO4) is normally used for this purpose. which can be costly;
Sulfuric acid has the advantage of being both cheap,
future science group www.future-science.com 3

The formed salts ends up partly dissolved in the glyc- linked to silica support, or mesoporous MCM-41 func-
erol and methanol phase, which must be purified by tionalized with tin–triflate or amines have been tested
decantation and distillation after reaction and for transesterification with promising results [59–62] . Even
settling [38] . base-functionalized metal–organic frameworks and sup-
ported quaternary substituted ammonium groups were
It is of immense interest to substitute the homoge- also shown to be promising organic materials for basic
neous catalysts in FAME production with heterogeneous methanolysis of oils and fats [63,64] .
catalysts. This would fit well into a flow reactor with solid
catalyst bed, for instance by using statis mixers and it Acidic inorganic oxides & derivatives
would greatly ease downstream separation, prevent salt Supported tungstated catalysts (WO3/M(P)Ox) have been
byproducts, lower the catalyst inventory cost and prevent investigated for the esterification of FFA with methanol
corrosion by the acidic H2SO4 [39] . An extensive num- between 60 and 200°C and at the highest temperatures
ber of organic and inorganic bases and acids have been full conversion of FFA is easily obtained while transesteri-
proposed in literature as catalysts for the two reactions. fication of TGs also takes place [65–68] . Related to the solid
tungstated oxides are the so-called heteropolyacids based
Inorganic metal-oxide bases on tungsten or molybdenum phosphate. These are strong
A range of different inorganic bases have been tested acids in their protonated form and can be supported on
in the transesterification reaction; alkali or earth-alkali metal oxide or even carbon [69] . Partial substitution of pro-
metal oxides or hydroxides are generally very basic and tons with other cations, for instance as Zr0.7H0.2PW12O40,
thus active, but normally dissolve in methanol [40] . ZnO increased the acidity and allowed both esterification
or Al2O3-ZnO mixed oxide were investigated as well as and transesterification at 65°C within 4–8 h [70] . The
rare-earth oxides, but both required high temperatures completely protonated heteropolyacids H3PW12O40,
(at least 200°C) for conversion of vegetable oils with H4SiW12O40, H3PMo12O40 and H4SiMo12O40 had dif-
methanol to FAME [41,42] . Bimetallic oxides of calcium ferent acidities, which correlated with their transesteri-
(Ca 2Fe2O5, CaMnO3, CaCeO3, CaTiO3 and CaZrO3) fication activities [71] . H3PW12O40, supported on various
were found to catalyze transesterification with metha- materials easily catalyzed the esterification of FFA in waste
nol at 60°C with good reusability of the catalysts [43] . oils at 25–60°C [72,73] . A multifunctionalized composite
Porous silicates such as zeolite BEA, USY and FAU, and of H3PW12O40 supporting Nb2O5 or Ta2O5 was active for
mesoporous silicalites such as KIT-6, ITQ-6, SBA-15 esterification at benign conditions, and activity could be
and MCM-41, have been ion-exchanged with La 3+, increased by adding nonpolar alkyl functionalities [74,75] ,
Mg2+ or K+ions and tested in FAME production as well, although dissolution of the acid in methanol can occur.
however high transesterification activity often requires Sulfated metal oxides are related to their tungstated
higher temperatures (>150°C) with these porous materi- analogues and sulfated TiO2, SnO2 and ZrO2 have been
als [44–46] . Hydrotalcites have shown superior transesteri- suggested as acidic catalysts for the esterification to form
fication activity to some inorganic porous bases depen- FAME as well as for the simultaneous transesterifica-
dent on composition and calcinations conditions [47–51] . tion of the TGs [76–80] . The same oxides were further
Activity has also been enhanced by doping with Cs, Ba, evaluated supported on or mixed with high surface-area
Sr or La, by substituting Al with Fe or Mg with Co, or supports of Al2O3 or SiO2 [76,81] . Leaching of sulfate to
by embedding hydrotalcite on a polymer support [52–55] . the methanol from SO42-/MOx is, however, a common
problem [80] .
Basic organic amines Various Brønsted-acidic zeolites have been shown
A number of different organic amines and derivatives to esterify FFA at 60 °C, whereas higher temperature
have been shown to be applicable for the transesteri- was required for Brønsted-acidic Al-MCM-41 [82,83] .
fication of oils and fats. At an early stage, a number of Functionalization of MCM-41 or SBA-15 via anchor-
substituted alkylguanidines or cyclic guanidines have ing with acidic WO3, heteropolyacids and sulfonic acid
been suggested as very active catalysts for transesterifica- functionalities also facilitated esterification of FFA with
tion and substituted tetramethyl guanidine (R-TMG) or methanol [84,85] .
substituted 1,5,7-triazabicyclo[4,4,0]dec-5-ene (R-TBD)
nested on PS/PVB polymer supported analogoues [56] . Organic sulfonic acids
Carbon nanotubes doped with amines or gem-diamines Amongst the acidic catalysts for esterification one type
have recently shown activity in the transesterification of organic acid has been studied extensively, namely
of TG at conditions near normal industrial operating sulfonic acids (R-SO3H). Simple sulfonic acids include
temperatures [57,58] . Synthetic solid bases such as methyl- MeSO3H and p-toluenesulfonic acid, while solid sul-
substituted phosphazenium, either unsupported or fonic acids can be prepared by anchoring the sulfonic

functionality onto polymeric supports by treatment sulfonate [99,102,103] . The melting point of the ILs must
with sulfuric acid; the resulting acid is often as strong be below the reaction temperature for FAME synthesis,
as or stronger than sulfuric acid. otherwise the activity is too low [102] . By functionalizing
Resin-type sulfonic acid can be bought on a technical the IL with more sulfonic acid functionalities the IL
scale, for instance ion-exchange resins [86] . Esterification exhibits an even higher catalytic activity [104] .
can be performed over various Amberlyst-resins and
Dowex HCR-W2 between 30 and 65°C, which yielded Enzymatic (trans-)esterification
a satisfactory esterification of FFA [87] . A comparative In addition to chemical catalysts, lipases from differ-
study between the sulfonic resins NKC-9, 001x7 and ent microorganisms can also be applied in FAME syn-
D61 proved the former to be superior in the reaction thesis. The conditions are often mild, usually around
of FFA and methanol at 60°C [88] . ambient temperature (i.e., 25–50°C), and both trans-
A number of polymer-based sulfonic-acids have been esterification and esterification are catalyzed [33] . For
prepared for esterification and studied, for instance the lipases to be applied in industrial FAME produc-
starting from polymers such as PVA-PS/PVB, PS, tion they must be immobilized on a porous support
sulfonated PV and PS, mesoporous silica or polyani- [105–107] . To date the catalytic activities reported are
lines deposited on carbon [89–93] . The incorporation of not of the same magnitude as the chemical catalysts,
sulfonic acid groups into the matrixes can be tailored although the activity can sometimes be enhanced by
and is usually between 0.6 to 6.0 mmol/g of the total using a co-solvent [35,108,109] . Some lipases may also be
supported catalyst mass. Esterification of FFA may be rendered inactive by too high concentrations of metha-
performed under normal operating conditions of 60°C nol and continuous addition of methanol, genetically
with one–two times molar excess of methanol. altering the lipases or switching alcohol can therefore
Another approach for obtaining a sulfonic acid-con- be necessary [110,111] .
taining polymer was first proposed by Toda et al. [94] . A
carbohydrate source, for example, sucrose or cellulose, Super-critical (trans-)esterification
is carbonized (or pyrolyzed) under inert atmosphere of As an alternative to using catalysts, the esterification and
around 400°C for 4–24 h. The resulting coke consisting transesterification may be performed using supercritical
of small graphene layers is then sulfonated with sulfuric methanol (PC = 81 bar, TC = 234.5°C) at very short reac-
acid at elevated temperatures (~150°C) also for 4–24 h. tion times. Adding a co-solvent, such as carbon dioxide
Despite low surface areas of the material itself, the cata- or hexane, or a base to the supercritical methanol can
lyst was found to be very active in the esterification of further improve the reaction rates [112–114] . However,
FFA at 60°C and had acid strength comparable to that the harsh supercritical conditions require high-pressure
of H2SO4 [95,96] . At 150°C and 17 bars a carbohydrate- equipment and handling, and the energy and economical
based sulfonic acid was also active for transesterification, balance for this may not be advantageous.
but leaching of sulfonic acid functionalities resulted [97] .
Impregnation upon mesoporous silicas allowed tailoring Glycerol
porosity and hydrophobicity and thereby enhancing the The main byproduct of the transesterification is glycerol
esterification rate [98] . (Figure 2A) . The increasing amount of FAME produced
accordingly yields increasing amounts of glycerol, since
Ionic liquid catalyzed (trans-)esterification approximately 100 kg of glycerol is formed per tonne
Within the last few years sulfonic acid functionalized of FAME made. Valorization of glycerol can therefore
acidic ionic liquids (ILs) have been suggested as alter- be an important contribution to the overall economy of
native Brønsted acid catalysts [99] . ILs are salts, usu- the FAME process. A few suggested reactions of glyc-
ally organic, with a melting point below 100°C and erol into more valuable chemicals include reduction,
can be used as catalysts and/or solvents for numerous dehydration, oxidation, or complete gasification into
reactions. Although ILs have been reported as reaction synthesis gas for production, for example, methanol.
media for homogeneous catalysts for transesterification, Some of these products are shown in Figure 3, but we
ILs may further be functionalized by incorporating sul- direct readers to the reviews on glycerol conversion by
fonic acid functionality, yielding both a liquid reaction Zhou and co-authors [115] or Johnson and Taconi [116]
medium and a very strong acid as catalyst [100–102] . This for more details.
approach is attractive for the esterification and trans-
esterification reactions [99] . The cation can be based Overview of catalysts for FAME production
on substituted imidazolium or pyridinium ions func- An overview of the catalysts for FAME production via
tionalized with sulfonic acid, and anions should prefer- esterification and transesterification of fats and oils with
ably be very weak bases, for example, triflate or methyl alcohol are found in Table 1.

Sulfided metal catalysts

HO OH
OH
Kubička and co-workers evaluated the HDO of rapeseed
1,3-Propanediol oil over sulfided CoMo and NiMo supported primar-
OH ily on Al2O3 in a number of studies at 250–350°C and
1,2-Propanediol 7–110 bars H2 [117–120] . Activity of sulfided NiMo/Al2O3
superseded that of separate MoS2/Al2O3 and NiS/Al2O3
O OH in liquid phase HDO and full conversion in rapeseed
HO OH HO OH O oil conversion over three commercial sulfided NiMo/
Al2O3 catalysts at 70 bar hydrogen was only achieved
Dihydroxyacetone Glycerol Acrolein
at > 310°C, while lower temperatures left FFAs and
OH
TGs in the product. Using a porous, Brønsted-acidic
OH
support of Al-MCM-41 for sulfided CoMo instead of
HO OH
OH Lewis-acidic Al2O3 did not alter activity, but MCM-41-
O support without Al in the framework was less active at
OH
O
Glyceric acid
Acrylic acid 20–110 bar H2 and 300–320°C.
Huber et al. co-treated sunflower oil and heavy vac-
O
Lactic acid uum oil over sulfided NiMo/Al2O3 at 300–450°C and
50 bars [121] . Maximal carbon yield of 71% of C15-C18
Figure 3. Useful products from oxidation (with molecular O2 or H2O2 over n-alkanes were obtained from the mixture, and crack-
supported noble metal catalysts), reduction (with H2 over supported ing reactions and isomerization of products became
transition metal catalysts) and dehydration of glycerol (via acid catalysis). more pronounced at higher temperatures. Šimáček
and co-authors observed that a co-desulfurization-
Hydrodeoxygenation of fats & oils deoxygenation of 5% rapeseed oil in vacuum gas oil at
The hydrodeoxygenation (HDO) of fats and oils is an 400–420°C yielded similar fuel product properties as
alternative approach for upgrading fatty feedstock. It pure vacuum gas oil desulfurization [13] .
was first industrialized by the Finnish company Neste Donnis and co-workers confirmed that both reduction
Oil at their refinery in Porvoo, Finland, as described by and decarboxylation can take place in the HDO over
Koskinen et al. [24] and Mikkonen [26] . The approach sulfided NiMo/Al2O3 and obtained selectivity to decar-
usually requires hydrogen and may be visualized as boxylation products from rapeseed oil of up to 64% at
shown in Figure 4. Based on recent literature the cata- full deoxygenation in light gas oil at 45 bar H2 and 350°C
lysts for deoxygenation of fats and oils can be divided [22] . Guzman et al. performed HDO of pure palm oil over
into three categories: supported sulfided metals, sup- sulfided NiMo/Al2O3 at 40–90 bar H2 and 350°C and
ported (noble) metals and micro- or meso-porous reported that selectivity to n-alkanes with even carbon
acid–base catalysts for cracking-type reactions. numbers (complete reduction) increased with pressure [8] .
Krause et al. studied the HDO
Table 1. Respectively basic, acidic and alternative esterification and transesterification of heptanoic acid, heptanol, methyl
catalysts for use in the production of fatty acid methyl esters. and ethyl heptanoate over sulfided
Catalyst Advantages Challenges CoMo/Al 2O3 and NiMo/Al 2O3 at
250°C and 15 or 75 bar H2 [122,123] .
Organic amines Strong bases Water and methanol tolerance
Porosity via supporting
NiMo, but not CoMo was very sen-
Basic inorganic oxides Temperature stable Possibility of dissolution
sitive to the sulfidation conditions.
Organic sulfonic acids Very acidic Temperature stability
Unsulfided catalysts were neither
Porosity via supporting very active nor selective towards C6
Acidic oxides and Very acidic Dissolution in methanol or C7 alkane formation.
heteropolyacids Tailoring of surface properties The sulfided catalysts pose the
Sulfated metal oxides Very acidic Sulfate leaching inherent challenge that they deacti-
Ionic liquids Reaction medium and catalyst Separation problems vate (desulfidize) over time if sulfur
Properties can be tailored Possible eco-toxicity issues is not added periodically, for instance
Anchoring possible as H2S or CS2 or by co-treating with
Can catalyze transesterification sulfur-containing compounds such
Immobilized enzymes Benign reaction conditions Low activity as refinery gas oils. The consump-
Ease of reuse Tolerance to methanol tion of hydrogen should optimally
Both esterification and be minimized to save costs, so higher
transesterification selectivities for decarboxylation or

decarbonylation, as described by
A O B
Donnis et al., should be aimed for
if methanation can be avoided [22] . O R
O
A great asset of this procedure is the H R H3C R
absence of formed aromatics. H R' + C3H8 + H2O, CO, CO2
H2 12 H2
6 H2O + C3H8 +
O O R' H3C R'
cat. cat.
H R'' H3C R''
Transition metal catalysts O R''
Most studies of deoxygenation of
fatty acids and their derivatives
O
over metal catalysts have been per-
formed by Murzin and co-workers, H R + CO2 HO R
3 H2
2 H2O + H3C R
although others have contributed as cat. cat.
well. Especially conversion of stearic
acid, ethyl stearate and tristearine to Figure 4. Hydrodeoxygenation of triglycerides and free fatty acids. (A) Decarboxylation and
alkanes have been extensively studied decarbonylation of feedstock, which is the prominent route with supported transition metals.
and the reaction routes elucidated in (B) Complete reduction with H2 of the feedstock, which is usually the dominant route over many
semibatch-reactors between 270 and sulphided catalysts.
360°C, either in gas mixture with
H2 or inert gas at up to 40 bars. By screening a range of while ester undergo decarbonylation (-CO). This selec-
active catalyst metals and supports, Snåre et al. found tivity is an advantage, avoiding the immediate use of
that platinum or palladium supported on carbon were hydrogen. However, hydrogen may be required to avoid
most active and had the highest selectivities for decar- deactivation by aromatization or CO poisoning. This
boxylation of stearic acid to n-heptadecane [6] . leads to a problem of methanization of CO or CO2 with
In connected work, Pd/C was used as a catalyst to H2, a highly undesirable situation. The methane may be
establish kinetic models for deoxygenation of tristea- burned for process heat or steam reformed back to H2
rine, ethyl stearate, stearic acid and of other fatty acids to and CO, but this is costly and the separation of the gases
alkanes [9,124,125] . Deoxygenation of the carboxylic acid alone may be tedious. Deoxygenation of real feedstocks
itself took place almost exclusively via decarboxylation, such as vegetable oils or waste fats is sparsely studied
but ester functionalities underwent more complicated and more challenging to convert over noble-metal cata-
mechanisms [124] . Deoxygenation of methyl octanoate lysts. They contain impurities such as salts, sterols and
and methyl stearate over Pt/Al2O3 at 330°C in He and H2 phospholipids. Furthermore, the TGs are always more or
gas revealed that H2 suppressed formation of higher self- less unsaturated, which is believed to cause deactivation
condensates of the esters and that decarbonylation was by coking via cyclization, dehydrogenations and Diels–
always the primary reaction route [126] . Dehydrogenation Alder reactions. More research is needed to unravel the
of C18 tall oil fatty acids and the pressure of hydrogen mechanisms behind deactivation by coking and how to
had a considerable impact on deoxygenation to C17- avoid it, but saturation of the double-bonds with H2 prior
hydrocarbons over Pd/Sibunit [127] . The deoxygenation of to deoxygenation may be a solution [125,127,132] .
TGs over carbon-supported Ni, Pd and Pt at 350°C with-
out added hydrogen yielded CO2, CO, CH4 and lighter Cracking-type deoxygenation
hydrocarbons in the gas phase, a range of liquid alkanes Resasco et al. studied deoxygenation of methyl octano-
and heavier paraffins, and FFAs as intermediates [128] . ate over various zeolites. Using H-ZSM-5 at 500°C,
Palladium and platinum as catalysts have also been lighter hydrocarbon gases and aromatic products were
investigated on a number of nanoporous and micropo- formed, the latter of which proceeded through self-con-
rous supports. Pd catalysts were investigated supported densation products such as tetradecane, 8-pentadecan-
on SBA-15, mesocellular SiO2-foam or SAPO-31 for the one and octyl octanoate. The aromatization selectivity,
deoxygenation of stearic acid or sunflower oil [10,129,130] . as well as the conversion, was much lower at 400°C.
Hancsók and co-authors deoxygenated saturated vegetable Interestingly, with zeolite X, longer hydrocarbons (C14-
oil over Pt/H-ZSM-22/Al2O3 to isomerized alkanes [131] . C16) and no aromatics were formed over basic CsNaX,
Acid functionalities of the support lead to modest isomeri- while the weakly acidic NaX resulted in aromatics before
zation of the formed n-alkanes to improve the cold proper- deactivating [133,134] .
ties of the fuel, but undesired cracking and deactivation H-ZSM-5, rare-earth-modified Y-zeolite and MCM-
also took place if the temperatures were too high. 41 have been used to crack palm oil at 400–500°C.
From literature it appears that deoxygenation of car- The products were mainly gasoline- and diesel-range
boxylic acids deoxygenate via decarboxylation (-CO2), hydrocarbons, and the selectivity to the latter rose

with decreasing cracking temperature and thereby also produces a gas fraction containing various carbonaceous
decreasing conversion [11,135] . Heptadecane, heptadecene gases, a residual char fraction and varying amounts of
and cracking products, such as alkanes and minor condensable, multifunctional tar called flash pyrolysis-
amounts of carboxylic acids, were found during oleic oil or just bio-oil. Another less-explored way is through
acid decomposition at 400°C over basic hydrotalcites the conversion of aqueous sludge or dissolved or sus-
[12] . Analogous products were formed in soybean oil pended biomass in near-critical water with inorganic cat-
pyrolysis over Al2O3-supported tin and zinc oxides at alysts; so-called hydrothermal upgrading of biomass or
350–400°C [136] . the proprietary CatLiqTM process. This harsh treatment
Both diesel- and gasoline-range products and varying also yields a bio-oil fraction after decantation [137,138] .
production of gas result from the cracking of fats and A few of the typical aromatic bio-oil compounds are
oils. Aromatics are also produced, which are undesired shown in Figure 5. Depending on the pyrolysis condi-
in diesel fuel [23] . The aromatization inside micro- and tions the output of bio-oils may be more than half of the
meso-porous catalysts leads to deactivation by coking, biomass input weight.
especially with acidic porous materials. Regeneration Upgrading of bio-oils from flash pyrolysis bears some
could, however, be performed in a technical FCC system resemblance to HDO of fatty feedstocks, and the cata-
by burning off the coke in a regeneration zone (provided lysts can also be divided into the same categories, namely
that the catalyst can endure this treatment). cracking-type, supported sulfided metals and supported
transition metals, with the notable inclusion that bio-oils
Overview of catalysts for HDO of fats may also be treated over a catalyst during the pyrolysis
A comparison of the catalysts for HDO of fats and oils itself – this is known as catalytic pyrolysis. The difference
mentioned in the HDO of fats and oils section is given from the well-defined TG molecules largely complicates
in Table 2. the HDO of bio-oils. These processes will remove a part
or all of the oxygen from the oils (lower the O/C-ratio)
Alternative technologies for biomass diesel and saturate double-bonds (increase the H/C ratio).
The potential supply of fats and oils at present can only Removal of much of the oxygen in the bio-oils, as well
cover a minor fraction of the diesel demand, although as water, can be a pretreatment to hydrocracking in con-
algae farming may provide a supply of more oils on the ventional refinery operation, for instance together with
long term. Utilization of the most abundant biomass normal petroleum gas-oil feeds, justifying two-stage pro-
resource, namely lignocellulose (e.g., straw, wood and cesses [139] . Supported metal and sulfided metal catalysts
fibers) is therefore imperative. Three potential and have the advantage of working at lower temperatures than
complementary strategies are proposed for this, namely the cracking of acidic or basic catalyst, but they do require
thermal depolymerization of biomass and upgrading of hydrogen to function as well as sulfurous compounds
the resulting bio-oils, APR of carbohydrate biomass and in the feed for the catalysts to remain active. However,
gasification of biomass followed by FTS. aromatics in the deoxygenated bio-oils are generally
undesired for diesel fuel [23] .
Upgrading of bio-oils
One auspicious way of dealing with lignocelluloses rel- Cracking-type upgrading
evant for diesel production is upgrading bio-oils. These Zeolites, especially acidic, are suited for cracking bio-oils
are usually produced through flash pyrolysis, that is, fast
since they are crystalline and therefore more robust for the
heating to 400–650°C of finely ground of wood or straw harsh conditions in cracking reactions with water vapor
at short residence time, often under inert gas and with present. They can also be easily regenerated by burning
sand or other ceramics as a heat carrier. This protocol away coke. Aho and co-workers studied pyrolysis and
cracking, as well as catalytic pyrolysis
Table 2. Comparison of sulfided metals, transition metals and porous ceramics as catalysts of pine sawdust at 400–450°C over
for hydrodeoxygenation. quartz sand and zeolites H-BEA,
Catalyst Advantages Challenges H-FAU, H-MFI or H-MOR and
Supported sulfided metal Not easily deactivated Needs sulfur addition some of their Fe-modified counter-
Co-treating with petrodiesel H2 consumption parts. The highest removal of oxygen
Supported transition metals High selectivity towards H2 consumed by side reactions was achieved with H-MFI [140,141] .
decarboxylation or Deactivation Cracking of bio-oil was studied
decarbonylation Methanation at 500°C on H-FAU and Ni-FAU
Micro/mesoporous ceramics No hydrogen needed Coking zeolites, giving low liquid yields
Easy regeneration after Diverse product range; and high yield of C2-C 4 and C5+
deactivation low diesel selectivity gases. Notably, methane formation

was avoided [142] . Co-cracking of a OH OH

distillation residue with 10 wt% of
bio-oil at 520°C over a commercial
FCC catalyst was found feasible and OCH3 OH OCH3 OH
Guaiacol Catechol Anisole Phenol
yielded normal coking behavior due
to a low total oxygen concentration
in the feed [143] . At temperatures
CH2OH CH2OH CH2OH
up to 650°C, the catalytic pyrolysis
of lignin and glucose over H-MFI O
and other zeolites resulted mostly

in cracking to aromatics, coke and
light gases [20,144,145] .
H3CO OCH3 OCH3
Sulfided metal catalysts
OH OH OH CH 3
The HDO of bio-oils over sup- Sinapyl alcohol p-Coumaryl alcohol Coniferyl alcohol p-methylacetophenone
ported sulfided catalysts were stud-
ied over a decade ago, for instance by Figure 5. Aromatic monomers in lignin networks and also found as a major part of many
Delmon et al. [146] . Sulfided CoMo/ bio-oils.
Al 2O3 and NiMo/Al 2O3 are usu-
ally the catalysts of choice. More recently, the sulfided Polyols are very soluble in water, but once they get suf-
CoMo/Al2O3 was shown to be less active than sulfided ficiently reduced, for instance to monofunctional hydro-
NiMo/Al2O3 for the HDO of guaiacol in batch mode at carbons, they spontaneously separate and form a nonpo-
200–350°C, but the CoMo was more selective to HDO lar hydrocarbon-phase. This phase contains a mixture of
products [147] . Different phenolic and aliphatic oxygen- different monofunctional hydrocarbons – ketones, car-
ates were converted over supported sulfided CoMo and boxylic acids, alcohols and more or less saturated furanic
NiMo catalyst with alcohols, aldehydes and carboxylic heterocycles with a carbon range of C4-C6. Gas-phase
acids as intermediates. The acidic g-Al2O3 support also products are CO2 and minor amounts of C1-C6 alkanes
mediated esterification, dehydration and hydrolysis reac- [150,151] . A sweep-stream of hydrocarbons can enhance
tions [148] . Bio-oil HDO has been studied in two-step the separation [15] .
semibatch systems over sulfided NiMo/Al2O3. Leaving
a few wt% of oxygen as ketones or ethers may be attrac- Aqueous reforming of polyols
tive with respect to hydrogen consumption and product A range of catalysts have been suggested for the APR of
properties [149] . sorbitol, but Pt and Pt-Re catalysts supported on ceram-
ics or carbon have been found advantageous and it is
Transition metal catalysts suggested that a bifunctional catalyst containing a metal
A few reports are available on HDO of bio-oils over and an acid functionality are needed to facilitate all reac-
supported metal catalysts. The activity of Pd, Pt and tions [150,152] . They facilitate both dehydrogenation and
Ru with different metal particle sizes supported on C-C-scissions to liberate adsorbed CO species, which
for instance carbon, SiO2 or Al2O3 have been tested as are then water-gas-shifted to CO2 and hydrogen [150] .
catalysts in the HDO of various model oil compounds APR is usually performed at approximately 200–250°C
between 250°C and 350°C and up to 200 bar H2. An oil and at pressures of 18–40 bar, and in an aqueous phase.
yield of 60 wt% with 90 mol% oxygen removal has been Aromatization, methanization and CO production are
achieved from bio-oil, yielding oil with a higher heating for these reasons minimized.
value of approximately 40 MJ/kg. Catalysts are reported
to deactivate by coking in the pore system [21,139] . Condensations prior to reduction
Carbohydrates are built by units of maximally six carbon
Aqueous-phase reforming atoms, meaning that hydrocarbons with more than six
An auspicious approach to fuel and chemicals produc- atoms require C-C-coupling reactions. This may be done
tion have been pursued especially by Dumesic and co- prior to reduction, as carbohydrates may undergo aldol
workers, who have studied the upgrading of sugars and condensation to couple ketones and aldehydes to other
polyols to hydrocarbons by reforming biomass in water carbonyls. This can take place via 5-(hydroxymethyl)
– APR. As shown in Figure 6A , the principle behind this furfural by dehydrations and isomerizations from fruc-
approach is the simultaneous catalytic reduction and tose, glucose and cellulose [153] . One or two molecules of
reforming of sugars and sugar derivatives. 5-(hydroxymethyl)furfural may for instance be reacted

Sugars and polyols are not as cheap as waste lignocel-

A
+ 6H2 lulosic biomass, so a greater feedstock tolerance would
OH OH OH make this process extremely viable, for instance, such as
6 CO 2 + 13 H 2 n- C 6H 14
the cascade-process recently proposed by Serrano-Ruiz
C-C-bond cleavage metal et al. from cellulose via g-valerolactone and ketoniza-
C-O-bond cleavage
reforming with H2O
OH OH OH tion to 5-nonanone [160] . Larger water-soluble biomass-
Water-gas shift Sorbitol Hexane derived molecules may pose problems to APR, as they
can result in coking.
B
OH O
OH O + OH OH
- H 2O + H2
HO
HO
O
OH
O O
n- C 9H 20 Gasification & FTS
Glucose
OH aci d
HMF
base
butene-3-hydroxy-HMF
metal
Nonane
By gasification of coal or biomass or steam reforming
of natural gas, a synthesis gas (syn-gas) consisting of
C
CO and H2 can be obtained, which can then be used
O O O
for the FTS yielding primarily linear-chained alkanes.
R' OH
+
R' ' OH base R' R' '
+ H 2O + CO 2 The principle behind this approach is shown in Figure 7.
Carboxylic acid Carboxylic Acid Ketone
Figure 6. Reforming, reduction and condensations of carbohydrates. Gasification

(A) Principle in aqueous-phase reforming of carbohydrates and derivatives The gasification of biomass has been extensively studied,
over dual-function acid and metal catalyst. (B) Acid-catalyzed isomerization yet a range of technical challenges remain. All types of
and dehydration, base-catalyzed aldol condensation (C-C coupling), and biomass may in principle be gasified and then converted,
reduction of carbohydrates over supported metal catalysts. but differing reactivities of various raw materials dom-
(C) Base-catalyzed ketonization of carboxylic acids. inate the processes. Usually, the biomass is heated to
700–1100°C in a fixed- or fluidized-bed system, either in
with acetone over basic catalysts, as seen in Figure 6B, inert gas or with minor amounts of oxygen (resulting in
yielding either a C9 or C15 building block that can be combustion reactions to heat the reactions) and usually
reduced [154] . at low pressures, with steam and/or other recycled pro-
cess gases. The pyrolysis and gasification is a gradual pro-
Product condensation/reduction reactions cess in which the biomass particles degases and volatilizes
A final step in the biomass conversion is the reduction step-wise [161] . Often temperatures of at least 1000°C are
of remaining single-oxygen-containing compounds and necessary for total carbon conversion to gas, depending
unsaturated carbon-carbon bonds. This processing can on the feedstock and the conditions used, as alkali and
involve various types of catalytic reactions. earth-alkali ions affect the gasification reactivity [162] .
Cracking can be done over acidic zeolites to yield Catalytic gasification or gasification in super-critical
a mixture of light hydrocarbon gases, iso-alkanes and water may allow conversion at lower temperatures requir-
aromatics [150] . Light olefins or alcohols obtained via ing less expensive equipment [163–165] . Through optimi-
dehydration could be upgraded to larger olefins by zation, however, it was found that only at 400°C the
oligomerization yielding for instance diesel-length noncatalytic gasification efficiency in air-steam mixtures
hydrocarbons; g-valerolactone upgrading was investi- in a bubbling fluidized bed could be as high as 60% [166] .
gated by aqueous-phase ring-opening and decarboxyl- Quenching must be performed to stop gas-phase reac-
ation to butene over silica-alumina catalyst at 375°C, tions downstream. However, some tar is still produced
followed by oligomerization over an acidic zeolite. This during gasification [167] , which can clog up pipes down-
yielded over 75% C8+ hydrocarbons [155] . stream upon cooling and form coke on surfaces [168,169] .
Ketonization can upgrade carboxylic acids migrating to Sulfur-, phosphorous- and nitrogen-containing com-
the hydrocarbon phase during the APR. The ketonization pounds or tars contained in the producer gas may also
merges two carboxylic acids together as seen in Figure 6C, poison the catalysts downstream, and catalytic removal
for instance over an acidic CeZrOx catalyst and esterifica- of these compounds has been reviewed [168,170] .
tion with alcohols present in the organic phase also takes
place [156,157] . Ketonizations have even been achieved with Fischer–Tropsch synthesis
acids in flash pyrolysis oils with reasonable success [158] . A vast amount of research literature is available about
Two ketone molecules, directly from APR or from sub- FTS catalysis [203] , which is usually performed by metal
sequent ketonization, may react under acidic conditions oxide-supported metals at 150–330°C and 50–200 bar.
to form an enone as condensate (an unsaturated ketone) Supported ruthenium has been shown to be the most
by splitting off water and then be reduced, as has been active metal for FTS, but supported iron and cobalt are
proposed over Pd/CexZr1-xOy above 300°C [157,159] . used industrially due to lower price [171] . The Co or Fe

may be doped with smaller amounts of other metals, for Burned for heat /
instance Mn, Ni, Pt, Ru, K or Ce. reformed
Water and straight-chain alkanes (Figure 7) are the Heat
- CH4
C xH yO z CO2, H 2O, CH4, CO, H2 CO, H 2 C xH 2x+2 + H 2O
main FTS products, the latter being ideal for diesel metal
Biomass Separation Syn-gas Alkanes Water
fuels. An important feature of the reaction is that
- CO2, H2O
if diffusivities and steric effects are disregarded, the
weight fraction of a certain carbon chain length (i) is Figure 7. Gasification of biomass to producer gas, consecutive gas
given from the statistical probability of chain growth, cleaning and adjustment, and Fischer–Tropsch synthesis catalyzed by
a, by the Anderson–Schultz–Flory-distribution: supported Fe or Co catalysts.
Wi = i · (1-a)2 · ai-1.
Often the selectivity is expressed as the yield of the
most desired hydrocarbons, namely those with a chain- are the work-intensive batch-mode reaction and the
length of at least five carbon atoms, C5+. FTS-catalysts byproduction of salts, which arises due to the use of
always form methane to some degree, and the higher a homogeneous catalysts. A range of both organic or
is, the less methane is formed. Impurities of aldehydes, inorganic solid acids and bases have been suggested
alcohols and fatty acids are also formed however, which in literature as heterogeneous catalysts and the most
must be reduced if FTS is used for diesel production. prominent ones are expected to be industrially imple-
Light isomerization might also be necessary to achieve mented, especially with fresh vegetable oil feedstock.
low CFPP [172] . More advanced continuous flow systems and reusable
The FTS catalysts are sensitive to poisons, especially heterogeneous catalysts for optimized plant economy
sulfur, which form sulfides with the catalyst metal if should also be demonstrated on technical scale. The
present above ppm level. The particle sizes and struc- downstream valorization of glycerol is expected to be
ture of the catalyst metal, as well as the microscopic a larger contributor to the overall transesterification
environment in which they are used, strongly affect economy as new glycerol-based commodity chemicals
activity as well as C5+ selectivity [173,174] . The effect on are introduced.
selectivity, activity and catalyst stability of alloying
metals or promoting with alkalis are often complex HDO of fats & oils
[175–177] . Confinement of active sites can dramatically HDO of fats and oils is performed industrially with
alter the selectivity [178] , while support acidity and sulfided metal catalysts, and these require the addition
microporosity may be used to directly isomerize prod- and presence of sulfur in the feedstock to function.
ucts [179] . Optimization towards production of longer It is therefore relevant to co-treat fats and oils with a
chain alcohols, aldehydes or ketones is another strategy, sulfur-containing petroleum feedstock to be desulfur-
since upgrading and drop-in blend for petrol, jet fuel ized. Nonsulfided catalysts are suitable for treating
or diesel can then be tailored – or steered towards the fatty feedstocks that only contain oxygen function-
production of bulk chemicals via FTS [180,181] . alities, but their lifetimes should be optimized and
A large part of the carbon from gasification and deactivation phenomena studied before implementa-
FTS ends up as CO2. In terms of carbon capture and tion as they are difficult to regenerate. Cracking of fats
storage, this procedure has the advantage of produc- and oils over acidic or basic porous materials, which
ing concentrated CO2 – maybe at elevated pressure – a are easier to regenerate, generally yields byproduction
net negative CO2 emission from gasification and FTS of lighter, less valuable gases and poorer selectivities
is therefore possible by storing CO2 from the biomass to either of the primary fuels such as gasoline, jet fuel
that was taken up during the growth of the flora [182] . and diesel.
Gasifiers are economy-of-scale equipment – the big- Fats and oils are, however, a sparse resource. The
ger, the cheaper. On the other hand, biomass from amounts of waste fats or greases from, for example,
agriculture has a low energy density and needs to be abattoirs and restaurants are limited and the land use
transported a long way to centralized plants, so upgrad- associated with growing vegetable oil resources can-
ing the energy density via decentralized pyrolysis plants not cope with demand for fuels either. In the com-
may be advantageous [183] . ing years, research into routes for biofuels from more
abundant biomass waste will become much more
Future perspective important and can be expected to be introduced at an
Transesterification of fats & oils industrial scale as well. The research into flash pyroly-
The transesterification of oils and fats with alcohol is sis bio-oils production and aqueous phase reforming
applied on industrial level today, but the process has and their consecutive upgrading steps can thus become
a number of drawbacks. The main two disadvantages increasingly prominent.

Upgrading of bio-oils gas-phase reactions in the produced gas. Technical

Bio-oils can be considered ‘bio-crude oils’ due to their solutions to these issues, as well as furnishing a closer
chemically complex nature and high oxygen content, process and heat integration to optimize energy effi-
which pose challenges in terms of catalytic process- ciency, should be accomplished. Research into FTS
ing and upgrading. In-depth studies and progress in will seek to further optimize selectivity by structural
understanding the catalytic deoxygenation over both and electronic promotion, encasing or encapsulating
sulfided and unsulfided transition metals and porous active sites; use of meso- and micro-porous materials
acids or bases are needed to integrate them in refining. for synthesis of more advanced and selective catalysts,
Coupling reactions need to be furnished to yield longer so as to avoid methane formation and steer selectivity
hydrocarbons suitable for different fuel pools, espe- towards a desired range of hydrocarbons. Major logistic
cially for diesel, and maximization of the bio-oil yield challenges also need to be resolved regarding gasifica-
must be further studied and implemented. Hydrogen tion-based biofuels. Dinjus et al. described the logis-
consumption is an issue to be resolved and is depen- tics around collection of biomass for central process-
dent on the type of upgrading and the potential ability ing and suggested that residues such as straw or wood
to reform some of the bio-oil or biomass into hydro- are upgraded by flash pyrolysis in small decentralized
gen. The extraction of functional aromatic chemicals plants, so that the bio-oil and char fraction, now with
instead may be valuable as well. The solid char from a much higher carbon and energy density can be trans-
the pyrolysis can be gasified or burned for heat, but it ported to and upgraded at central gasification–FTS
can also be used for enhancing or remediating fertility plants [183] .
of soils (as ‘Terra Preta’) and sequestering carbon via
the biosphere. Conclusion
In terms of fuel quality, much depends upon the pro-
Aqueous-phase reforming cessing and reaction conditions of each process, the
Aqueous-phase reforming is a more benign approach to feedstock used and mechanistic details of the catalysts,
biomass refining and its ability to reform water-soluble especially from upgraded bio-oils or hydrocarbons from
biomass, even cellulose and hemi-cellulose, through aqueous-phase reformed biomass. However, both the
acidic hydrolysis is a major asset. The process integra- production of FAME or alkanes from oils and fats and
tion, mild temperatures and pressures makes it possible the gasification and FTS fuel reached such maturity that
to conserve a very high energy content of the biomass the fuels can be compared. This has been attempted by
in the products and facilitate building smaller plants Mikkonen [26] and Knothe [184] .
in regional farming districts avoiding long-distance Production of FAME from fats uses the most benign
transport of biomass. The methodology circumvents reaction conditions, the cheapest plant design and may
distillation of dilute aqueous solutions of fuel grade be decentralized, for instance in connection with abat-
compounds (e.g., ethanol). The integrated production toir waste-treatment or food-production plants. More
and consumption balance described for hydrogen is yet forcing conditions are used for HDO of fats and oils,
another advantage. It may be expected that multifunc- and it is more suited for traditional refinery infrastruc-
tional catalysis gives rise to more selective processes ture with possibilities of heat recovery and easy access
in APR, in turn yielding higher selectivity of targeted to hydrogen and isomerization units.
classes of hydrocarbons. This is achieved from carbo- Gasification and FTS may, in principle, be applied
hydrates by catalytic hydrolysis, dehydration, selective to all types of biomass, but the method requires the
condensations and hydrogenations in integrated pro- most forcing reaction conditions and the most extensive
cesses. The catalysts for these reactions are solid acids plant design. Gasifiers are economy-of-scale plants and
and bases as well as functional microporous materi- usually pose the largest investment in a gasification and
als, supported transition-metals and their alloys, and FTS plant, so this procedure is only viable for very big
multifunctional catalysts. plants and thus huge amounts of available biomass.
However, the necessary reactors and separators for The APR and the pyrolysis followed by deoxygen-
different treatments may end up in a vastly complex ation are intermediate technologies compared with
process layout, which can be complicated to erect in the former. They are much more complex than trans-
smaller systems. esterification or HDO of fats and oils, but they enable
cost reduction via process integration, beneficial logis-
Gasification & FTS tics, upgrading energy density and using more abun-
Gasification may be implemented with all types of dant feedstocks. In addition, they can also be used in
biomass, but it faces a number of challenges especially conjunction with existing refining infrastructure or
from tar clogging and coking in pipes and unwanted gasification and FTS.

Primary and easily upgradeable feedstock such as Financial & competing interests disclosure
plant oils are not and cannot be available in quanti- Anders Theilgaard Madsen is thankful to the Danish Agency for
ties sufficient to replace the entire diesel oil demand, Research and Innovation for financing his PhD-stipend through
so waste products from agriculture and forestry hold the innovation consortium ‘Waste-to-Value’. The authors have no
greater perspective for use in the future. It is necessary other relevant affiliations or financial involvement with any
that in the future biorefineries have sufficient toler- organization or entity with a financial interest in or financial
ance to use diverse raw materials as feedstock, primar- conflict with the subject matter or materials discussed in the
ily in the form of waste. New biorefineries employing manuscript apart from those disclosed.
these raw materials for biochemicals and biofuels will No writing assistance was utilized in the production of
come online over the coming decade. this manuscript. Executive summary
Executive summary
Transesterification of fats & oils
Transesterification is the largest source of diesel from biomass today.
Active heterogeneous catalysts are required for ease of purification, process economy, lowering environmental impact and designing
efficient continuous flow systems.
Already suggested and auspicious heterogeneous catalysts are both organic and inorganic solid acids and bases, ionic liquids, or
immobilized enzymes.
Byproduct glycerol has found a few new industrial uses as raw material.
Deoxygenation of fats & oils
Deoxygenation can be performed over traditional sulfided metal catalysts, supported noble-metal catalysts and porous acids or bases.
The sulfided metals are employed industrially today.
Noble metal catalysts are selective and consume less hydrogen but deactivate fast.
Acidic or basic catalysts yield a mix of hydrocarbons and coke via cracking, but they are easy to regenerate.
Side reactions and deactivation behavior must be understood and avoided with most catalysts, especially when using unsaturated
fatty feedstock.
Deoxygenation of bio-oils
Deoxygenation is more demanding than that of fats due to the multifunctional and aromatic nature of bio-oils; however the same catalysts
have been studied.
Catalytic pyrolysis may provide process simplification and higher energy efficiency.
Net hydrogen consumption must be lowered, for instance by steam reforming of the feedstock, and the production of aromatics must be
lowered.
Primary deoxygenation of bio-oils is necessary before potential co-processing in petroleum refineries.
Catalysts less prone to coking and deactivation should be developed or a durable regeneration strategy integrated.
Deoxygenation of bio-oils and contingent hydrogen production should be integrated with the gaseous products from flash pyrolysis for
higher process efficiency.
Aqueous-phase reforming, aldol-condensations & ketonization
Aqueous-phase reforming is a two-phase process for reforming water-soluble carbohydrates into hydrocarbons. Also cellulose and
hemicelluloses may be used as raw materials. The preferred catalyst is usually Pt-Re/C.
High energy efficiency due to integrated heat- and hydrogen use, selective processes and mild reaction conditions.
Pre- and post-reactions (hydrolysis, aldol-condensations, ketonization) of the aqueous-phase reforming can target the plant layout
towards the desired hydrocarbon product range; the yield of diesel-range hydrocarbons should be optimized.
Gasification & Fischer–Tropsch synthesis
All biomass can be gasified into synthesis gas and ‘reconstructed’ as hydrocarbons via Fischer–Tropsch synthesis (FTS).
Gasifiers are economy-of-scale equipment, and logistics of biomass collection and energy efficiency trade-offs are important. Plant and
process economy should be optimized for smaller, decentralized units.
Gasification yield should be optimized for temperatures as low as possible, for instance via catalysis.
Gas cleaning and cooling must be performed to avoid coking, clogging, corrosion and side-reactions, as well as FTS catalyst deactivation.
The FTS catalysts are improved by tailoring the active site structure for higher activity and selectivity via promotion, bi-functional catalysts
and use of porous materials.
The FTS catalyst tolerance to impurities of N-, P- or S-containing compounds or alkali salts should be addressed.
Future perspective
Different types of diesel fuels from biomass will be used together in the near future.
Waste feedstocks will get a more prominent role due to ecological considerations and their low price. Food versus fuel dilemma will be
increasingly controversial. Technologies to convert agricultural and forestry waste of all kinds to fuels will be increasingly important.

Bibliography 13 Šimáček P, Kubička D. Hydrocracking of diesel offers advantages. Hydrocarb. Process.

Papers of special note have been highlighted as: petroleum vacuum distillate containing 87(Feb), 63–66 (2008).
n
of interest rapeseed oil: evaluation of diesel fuel. Fuel 27 Sharma YC, Singh B, Korstad J.
of considerable interest
n n
89(7), 1508–1513 (2010). Advancements in solid acid catalysts for
1 Directive 2003/30/EC of the European 14 Huber GW, Chheda JN, Barrett CJ, Dumesic ecofriendly and economically viable synthesis
Parliament and of the Council. Offic. JA. Production of liquid alkanes by of biodiesel. Biofpr. 5(1), 69–92 (2011).
J. European Union 12(2), 1–20 (2003). aqueous-phase processing of biomass-derived 28 Yan S, DiMaggio C, Mohan S, Kim M, Salley
carbohydrates. Science 308(5727), 1446–1450 SO, Ng KYS. Advancements in heterogeneous
2 The European Commission. Renewable (2005).
Energy Road Map. Renewable Energies in the catalysis for biodiesel synthesis. Top. Catal.
21st Century: Building a More Sustainable 15 Chheda JN, Huber GW, Dumesic JA. 53(11–12), 721–736 (2010).
Future. Working Paper: COM(2006) 848 Liquid-phase catalytic processing of 29 Melero JA, Iglesias J, Morales G.
Final. The European Commission, Brussels, biomass-derived oxygenated hydrocarbons to Heterogeneous acid catalysts for biodiesel
Belgium (2007). fuels and chemicals. Angew. Chem. Int. Ed. production: current status and future
46(38), 7164–7183 (2007). challenges. Green Chem. 11(9), 1285–1308
3 Meher LC, Vidya Sagar D, Naik SN.
Technical aspects of biodiesel production by 16 Serrano-Ruiz JC, Wang D, Dumesic JA. (2009).
transesterification – a review. Renew. Sust. Catalytic upgrading of levulinic acid to 30 Semwal S, Arora AK, Badoni RP, Tuli DK.
Energy Rev. 10(3), 248–268 (2006). 5-nonanone. Green Chem. 12(4), 574–577 Biodiesel production using heterogeneous
(2010). catalysts. Bioresour. Technol. 102(3),
4 Mittelbach M, Remschmidt C. Biodiesel: the
Comprehensive Handbook. Mittelbach M 17 Abatzoglou N, Dalai A, Gitzhofer F. Green 2151–2161 (2010).
(Ed.). Martin Mittelbach, Graz, Austria diesel from Fischer–Tropsch synthesis: 31 Choudhary TV, Phillips CB. Renewable fuels
(2006). challenges and hurdles. Presented at: 3rd via catalytic hydrodeoxygenation. Appl. Catal.
IASME/WSEAS international conference on A 397(1–2), 1–12 (2011)
5 Knothe G. Biodiesel: current trends and energy, environment, ecosystems and sustainable
properties. Top. Catal. 53(11–12), 714–720 32 Bulushev DA, Ross JHR. Catalysis for
development. Greece, 24–26 July 2007.
(2010). conversion of biomass to fuels via pyrolysis
18 Zhang W. Automotive fuels from biomass via and gasification: a review. Catal. Today DOI:
6 Snåre M, Kubicková I, Mäki-Arvela P, Eränen gasification. Fuel Process. Technol. 91(8),
K, Murzin DY. Heterogeneous catalytic 10.1016/j.cattod.2011.02.005 (2011) (Epub
866–876 (2010). ahead of print).
deoxygenation of stearic acid for production
of biodiesel. Ind. Eng. Chem. Res. 45(16), 19 Stöcker M. Biofuels and biomass-to-liquid 33 Huber GW, Iborra S, Corma A. Synthesis of
5708–5715 (2006). fuels in the biorefinery: catalytic conversion of transportation fuels from biomass: chemistry,
lignocellulosic biomass using porous catalysts, and engineering. Chem. Rev.
7 Huber GW, Corma A. Synergies between materials. Angew. Chem., Int. Ed. 47(48),
bio- and oil refineries for the production of 106(9), 4044–4098 (2006).
9200–9211 (2008).
fuels from biomass. Angew. Chem. Int. Ed. n
Description and comparison of feedstocks
20 Carlson TR, Jae J, Lin Y-C, Tompsett GA,
46(38), 7184–7201 (2007). and methods for the production of biofuel,
Huber GW. Catalytic fast pyrolysis of glucose
Review on the possibilites for producing especially emphasizing the catalytic
with HZSM-5: the combined homogeneous
n n
biofuels in existing refinery infrastructure, chemistry involved.

and heterogeneous reactions. J. Catal. 270(1),
with emphasis especially on catalytic 110–124 (2010). 34 Jacobson K, Gopinath R, Meher LC, Dalai
cracking and hydrotreatment AK. Solid acid catalyzed biodiesel production
21 Wildschut J, Mahfud FH, Venderbosch RH,
8 Guzman A, Torres JE, Prada LP, Nuñez ML. from waste cooking oil. Appl. Catal. B
Heeres HJ. Hydrotreatment of fast pyrolysis
Hydroprocessing of crude palm oil at pilot 85(1–2), 86–91 (2008).
oil using heterogeneous noble-metal catalysts.
plant scale. Catal. Today. 156(1–2), 38–43 Ind. Eng. Chem. Res. 48(23), 10324–10334 35 Casanave D, Duplan J-L, Freund E. Diesel
(2010). (2009). fuels from biomass. Pure Appl. Chem. 79(11),
9 Simakova I, Simakova O, Mäki-Arvela P, 2071–2081 (2007).
22 Donnis B, Egeberg RG, Blom P, Knudsen
Murzin DY. Decarboxylation of fatty acids KG. Hydroprocessing of bio-oils and 36 Sivasamy A, Cheah KY, Fornasiero P,
over Pd supported on mesoporous carbon. oxygenates to hydrocarbons. Understanding Kemausuor F, Zinoviev S, Miertus S. catalytic
Catal. Today. 150(1–2), 28–31 (2010). the reaction routes. Top. Catal. 52(3), applications in the production of biodiesel
10 Kikhtyanin OV, Rubanov AE, Ayupov AB, 229–240 (2009). from vegetable oils. ChemSusChem 2(4),
Echevsky GV. Hydroconversion of sunflower 278–300 (2009).
23 Knothe G. Analyzing biodiesel: standards and
oil on Pd/SAPO-31 catalyst. Fuel 89(10), other methods. J. Am. Oil Chem. Soc. 83(10), 37 Lotero E, Liu Y, Lopez DE, Suwannakarn K,
3085–3092 (2010). 823–833 (2006 Bruce DA, Goodwin Jr JG. Synthesis of
11 Tamunaidu P, Bhatia S. Catalytic cracking of biodiesel via acid catalysis. Ind. Eng. Chem.
24 Koskinen M, Sourander M, Nurminen M.
palm oil for the production of biofuels: Res. 44(14), 5353–5363 (2005).
Apply a comprehensive approach to biofuels.
optimization studies. Bioresour. Technol. Hydrocarb. Process. 85(2), 81–86 (2006 38 Demirbas A. Biodiesel production from
98(18), 3593–3601 (2007). vegetable oils via catalytic and non-catalytic
25 Snåre M, Murzin DY. Reply to ‘Comment on
12 Na J-G, Yi BE, Kim JN et al. Hydrocarbon
supercritical methanol transesterification
heterogeneous catalytic deoxygenation of
production from decarboxylation of fatty acid methods. Progress Energ. Combust. Science
stearic acid for production of biodiesel’. Ind.
without hydrogen. Catal. Today. 156(1–2), 31(5–6), 466–487 (2005).
Eng. Chem. Res. 45(20), 6875–6875 (2006).
44–48 (2009). 39 Mo X, Lotero E, Lu C, Liu Y, Goodwin JG.
26 Mikkonen S. Second-generation renewable

A novel sulfonated carbon composite solid Transesterification of triacetin, tributyrin, (2008).

acid catalyst for biodiesel synthesis. Catal. and soybean oil with methanol over 63 Savonnet M, Aguado S, Ravon U et al.
Lett. 123(1–2), 1–6 (2008). hydrotalcites with different water contents. Solvent free base catalysis and
40 Liu X, He H, Wang Y, Zhu S. Fuel Process. Technol. 91(6), 618–624 (2010). transesterification over basic functionalised
Transesterification of soybean oil to biodiesel 52 Chuayplod P, Trakarnpruk W. metal–organic frameworks. Green Chem.
using SrO as a solid base catalyst. Catal. Transesterification of rice bran oil with 11(11), 1729–1732 (2009).
Comm. 8(7), 1107–1111 (2007). methanol catalyzed by Mg(Al)La 64 Liu Y, Lotero E, Goodwinjr J, Lu C.
41 Pugnet V, Maury S, Coupard V et al. hydrotalcites and metal/MgAl oxides. Ind. Transesterification of triacetin using solid
Stability, activity and selectivity study of a Eng. Chem. Res. 48(9), 4177–4183 (2009). Brønsted bases. J. Catal. 246(2), 428–433
zinc aluminate heterogeneous catalyst for the 53 Macala GS, Robertson AW, Johnson CL et al. (2007).
transesterification of vegetable oil in batch Transesterification catalysts from iron doped 65 Sunita G, Devassy BM, Vinu A, Sawant DP,
reactor. Appl. Catal. A 374(1–2), 71–78 hydrotalcite-like precursors: solid bases for Balasubramanian VV, Halligudi SB.
(2010). biodiesel production. Catal. Lett. 122(3), Synthesis of biodiesel over zirconia-supported
42 Russbueldt BME, Hoelderich WF. New rare 205–209 (2008). isopoly and heteropoly tungstate catalysts.
earth oxide catalysts for the transesterification 54 Li E, Xu ZP, Rudolph V. MgCoAl-LDH Catal. Comm. 9(5), 696–702 (2008).
of triglycerides with methanol resulting in derived heterogeneous catalysts for the 66 Park Y-M, Lee JY, Chung S-H et al.
biodiesel and pure glycerol. J. Catal. 271(2), ethanol transesterification of canola oil to Esterification of used vegetable oils using the
290–304 (2010). biodiesel. Appl. Catal. B 88(1–2), 42–49 heterogeneous WO3/ZrO2 catalyst for
43 Kawashima A, Matsubara K, Honda K. (2009). production of biodiesel. Bioresour. Technol.
Development of heterogeneous base catalysts 55 Guerreiro L, Pereira PM, Fonseca IM et al. 101(Suppl. 1), S59–S61 (2010).
for biodiesel production. Bioresour. Technol. PVA embedded hydrotalcite membranes as 67 Komintarachat C, Chuepeng S. Solid acid
99(9), 3439–3443 (2008). basic catalysts for biodiesel synthesis by catalyst for biodiesel production from waste
44 Shu Q, Yang B, Yuan H, Qing S, Zhu G. soybean oil methanolysis. Catal. Today used cooking oils. Ind. Eng. Chem. Res.
Synthesis of biodiesel from soybean oil and 156(3–4), 191–197 (2010). 48(20), 9350–9353 (2009).
methanol catalyzed by zeolite beta modified 56 Schuchardt U. Transesterification of soybean 68 Ngaosuwan K, Mo X, Goodwin Jr JG,
with La 3+. Catal. Comm. 8(12), 2158–2164 oil catalyzed by alkylguanidines Praserthdam P. Effect of solvent on hydrolysis
(2007). heterogenized on different substituted and transesterification reactions on tungstated
45 Macario A, Giordano G, Onida B, Cocina D, polystyrenes. J. Mol. Catal. A: Chem. 109(1), zirconia. Appl. Catal. A 380(1–2), 81–86
Tagarelli A, Giuffrè AM. Biodiesel 37–44 (1996). (2010).
production process by homogeneous/ 57 Villa A, Tessonnier J-P, Majoulet O, Su DS, 69 Kulkarni MG, Gopinath R, Meher LC, Dalai
heterogeneous catalytic system using an Schlögl R. Amino-functionalized carbon AK. Solid acid catalyzed biodiesel production
acid–base catalyst. Appl. Catal. A 378(2), nanotubes as solid basic catalysts for the by simultaneous esterification and
160–168 (2010). transesterification of triglycerides. Chem. transesterification. Green Chem. 8(12),
46 Li E, Rudolph V. Transesterification of Comm. (29), 4405–4407 (2009). 1056–1062 (2006).
vegetable oil to biodiesel over mgo- 58 Cerro-Alarcón M, Corma A, Iborra S, 70 Zhang X, Li J, Chen Y et al. Heteropolyacid
functionalized mesoporous catalysts. Energy Martínez C, Sabater MJ. Methanolysis of nanoreactor with double acid sites as a highly
Fuels 22(1), 145–149 (2008). sunflower oil using gem-diamines as active efficient and reusable catalyst for the
47 Georgogianni KG, Katsoulidis AP, Pomonis organocatalysts for biodiesel production. transesterification of waste cooking oil.
PJ, Kontominas MG. Transesterification of Appl. Catal. A 382(1), 36–42 (2010). Energy Fuels. 23(9), 4640–4646 (2009).
soybean frying oil to biodiesel using 59 Cerro-Alarcón M, Corma A, Iborra S, Gómez 71 Morin P, Hamad B, Sapaly G et al.
heterogeneous catalysts. Fuel Process. Technol. JP. Biomass to fuels: a water-free process for Transesterification of rapeseed oil with
90(5), 671–676 (2009). biodiesel production with phosphazene ethanol. I. Catalysis with homogeneous
48 Georgogianni KG, Katsoulidis AK, Pomonis catalysts. Appl. Catal. A 346(1–2), 52–57 Keggin heteropolyacids. Appl. Catal. A 330,
PJ, Manos G, Kontominas MG. (2008). 69–76 (2007).
Transesterification of rapeseed oil for the 60 Kim M-Y, Seo G, Kwon OZ, Chang DR. The 72 Oliveira CF, Dezaneti LM, Garcia FAC et al.
production of biodiesel using homogeneous exceptional activity of a phosphazenium Esterification of oleic acid with ethanol by
and heterogeneous catalysis. Fuel Process. hydroxide catalyst incorporated onto silica in 12-tungstophosphoric acid supported on
Technol. 90(7–8), 1016–1022 (2009). the transesterification of tributyrin with zirconia. Appl. Catal, A. 372(2), 153–161
49 Liu Y, Lotero E, Goodwin Jr JG, Mo X. methanol. Chem. Comm. (21), 3110–3112 (2010).
Transesterification of poultry fat with (2009).
73 Cardoso AL, Augusti R, Silva
methanol using Mg-Al hydrotalcite derived 61 Verziu M, El Haskouri J, Beltran D et al. MJ. Investigation on the esterification of fatty
catalysts. Appl. Catal. A 331(), 138–148 Mesoporous tin–triflate based catalysts for acids catalyzed by the H3PW12O40
(2007). transesterification of sunflower oil. Top. Catal. heteropolyacid. J. Am. Oil Chem. Soc. 85(6),
50 Brito A, Borges ME, Garín M, Hernández A. 53(11–12), 763–772 (2010). 555–560 (2008).
Biodiesel production from waste oil using 62 Saravanamurugan S, Han D, Koo J, Park S. 74 Srilatha K, Lingaiah N, Devi BLAP, Prasad
Mg-Al layered double hydroxide catalysts. Transesterification reactions over morphology RBN, Venkateswar S, Prasad PSS.
Energy Fuels. 23(6), 2952–2958 (2009). controlled amino-functionalized SBA-15 Esterification of free fatty acids for biodiesel
51 Kim MJ, Park SM, Chang DR, Seo G. catalysts. Catal. Comm. 9(1), 158–163 production over heteropoly tungstate

supported on niobia catalysts. Appl. Catal. A 86 Shibasaki-Kitakawa N, Tsuji T, Chida K, 2118–2121 (2010).
365(1), 28–33 (2009). Kubo M, Yonemoto T. Simple Continuous 98 Nakajima K, Okamura M, Kondo JN et al.
75 Xu L, Wang Y, Yang X, Hu J, Li W, Guo Y. production process of biodiesel fuel from oil Amorphous carbon bearing sulfonic acid
Simultaneous esterification and with high content of free fatty acid using groups in mesoporous silica as a selective
transesterification of soybean oil with ion-exchange resin catalysts. Energy Fuels. catalyst. Chem. Mater. 21(1), 186–193
methanol catalyzed by mesoporous Ta 2O5/ 24(6), 3634–3638 (2010). (2009).
SiO2–[H3PW12O40 /R] (R = Me or Ph) hybrid 87 Özbay N, Oktar N, Tapan NA. Esterification 99 Wu Q, Chen H, Han M, Wang D, Wang J.
catalysts. Green Chem. 11(3), 314–317 of free fatty acids in waste cooking oils Transesterification of cottonseed oil catalyzed
(2009). (WCO): Role of ion-exchange resins. Fuel by brønsted acidic ionic liquids. Ind. Eng.
76 Lam MK, Lee KT, Mohamed AR. Sulfated 87(10–11), 1789–1798 (2008). Chem. Res. 46(24), 7955–7960 (2007).
tin oxide as solid superacid catalyst for 88 Feng Y, He B, Cao Y et al. Biodiesel 100 Lapis AAM, de Oliveira LF, Neto BAD,
transesterification of waste cooking oil: an production using cation-exchange resin as Dupont J. Ionic liquid supported acid/
optimization study. Appl. Catal. B 93(1–2), heterogeneous catalyst. Bioresour. Technol. base-catalyzed production of biodiesel.
134–139 (2009). 101(5), 1518–1521 (2010). ChemSusChem 1(8–9), 759–762 (2008).
77 Suwannakarn K, Lotero E, Goodwin Jr JG, 89 Caetano CS, Guerreiro L, Fonseca IM, 101 DaSilveira Neto B, Alves M, Lapis A et al.
Lu C. Stability of sulfated zirconia and the Ramos AM, Vital J, Castanheiro JE. 1-n-butyl-3-methylimidazolium tetrachloro-
nature of the catalytically active species in the Esterification of fatty acids to biodiesel over indate (BMI.InCl4) as a media for the
transesterification of triglycerides. J. Catal. polymers with sulfonic acid groups. Appl. synthesis of biodiesel from vegetable oils.
255(2), 279–286 (2008). Catal. A 359(1–2), 41–46 (2009). J. Catal. 249(2), 154–161 (2007).
78 Du Y, Liu S, Ji Y et al. Synthesis of sulfated 90 Soldi RA, Oliveira ARS, Ramos LP, 102 Han M, Yi W, Wu Q, Liu Y, Hong Y, Wang
silica-doped tin oxides and their high César-Oliveira MAF. Soybean oil and beef D. Preparation of biodiesel from waste oils
activities in transesterification. Catal. Lett. tallow alcoholysis by acid heterogeneous catalyzed by a Brønsted acidic ionic liquid.
124(1–2), 133–138 (2008). catalysis. Appl. Catal. A 361(1–2), 42–48 Bioresour. Technol. 100(7), 2308–2310
79 López DE, Goodwin Jr JG, Bruce DA, Furuta (2009). (2009).
S. Esterification and transesterification using 91 Okayasu T, Saito K, Nishide H, Hearn 103 Li K-X, Chen L, Yan Z-C, Wang H-L.
modified-zirconia catalysts. Appl. Catal. A MTW. Preparation of a novel Application of pyridinium ionic liquid as a
339(1), 76–83 (2008). poly(vinylsulfonic acid)-grafted solid phase recyclable catalyst for acid-catalyzed
80 Rattanaphra D, Harvey A, Srinophakun P. acid catalyst and its use in esterification transesterification of jatropha oil. Catal. Lett.
simultaneous conversion of triglyceride/free reactions. Chem. Comm. (31), 4708–4710 139(3–4), 151–156 (2010).
fatty acid mixtures into biodiesel using (2009).
104 Liang X, Yang J. Synthesis of a novel
sulfated zirconia. Top. Catal. 53(), 773–782 92 Melero JA, Bautista LF, Iglesias J, Morales G, multi-SO3H functionalized ionic liquid and
(2010). Sánchez-Vázquez R, Suárez-Marcos I. its catalytic activities for biodiesel synthesis.
81 JC, Zhang J, Yarmo MA. Structure and Biodiesel production over arenesulfonic Green Chem. 12(2), 201–204 (2010).
reactivity of silica-supported zirconium acid-modified mesostructured catalysts:
105 Nassreddine S, Karout A, Lorraine Christ M,
sulfate for esterification of fatty acid under optimization of reaction parameters using
Pierre AC. Transesterification of a vegetal oil
solvent-free condition. Appl. Catal. A 332(2), response surface methodology. Top. Catal.
with methanol catalyzed by a silica fibre
209–215 (2007). 53(11–12), 795–804 (2010).
reinforced aerogel encapsulated lipase. Appl.
82 Chung K-H, Chang D-R, Park B-G. Removal 93 Zięba A, Drelinkiewicz A, Konyushenko E, Catal. A 344(1–2), 70–77 (2008).
of free fatty acid in waste frying oil by Stejskal J. Activity and stability of polyaniline
106 Shakeri M, Kawakami K. Effect of the
esterification with methanol on zeolite sulfate-based solid acid catalysts for the
structural chemical composition of
catalysts. Bioresour. Technol. 99(16), transesterification of triglycerides and
mesoporous materials on the adsorption and
7438–7443 (2008). esterification of fatty acids with methanol.
activation of the Rhizopus oryzae lipase-
Appl. Catal. A 383(1–2), 169–181 (2010).
83 Carmo Jr A, Desouza L, Dacosta C, Longo E, catalyzed trans-esterification reaction in
Zamian J, Darochafilho G. Production of 94 Toda M, Takagaki A, Okamura M et al. organic solvent. Catal. Comm. 10(2), 165–168
biodiesel by esterification of palmitic acid over Biodiesel made with sugar catalyst. Nature (2008).
mesoporous aluminosilicate Al-MCM-41. 438(7065), 178 (2005).
107 Dizge N, Aydiner C, Imer DY, Bayramoglu
Fuel 88(3), 461–468 (2009). 95 Hara M. Biodiesel production by amorphous M, Tanriseven A, Keskinler B. Biodiesel
84 Jiménez-Morales I, Santamaría-González J, carbon bearing SO3H, COOH and phenolic production from sunflower, soybean, and
Maireles-Torres P, Jiménez-López A. OH groups, a solid Brønsted acid catalyst. waste cooking oils by transesterification using
Zirconium doped MCM-41 supported WO3 Top. Catal. 53(11–12), 805–810 (2010). lipase immobilized onto a novel microporous
solid acid catalysts for the esterification of 96 Dehkhoda AM, West AH, Ellis N. Biochar polymer. Bioresour. Technol. 100(6),
oleic acid with methanol. Appl. Catal. A based solid acid catalyst for biodiesel 1983–1991 (2009).
379(1–2), 61–68 (2010). production. Appl. Catal. A 382(2), 197–204 108 Keng PS, Basri M, Ariff AB, Abdul Rahman
85 Dhainaut J.; Dacquin J-P, Lee AF, Wilson K. (2010). MB, Abdul Rahman RNZ, Salleh AB.
Hierarchical macroporous–mesoporous 97 Lien Y-S, Hsieh L-S, Wu JCS. Biodiesel Scale-up synthesis of lipase-catalyzed palm
SBA-15 sulfonic acidcatalysts for biodiesel synthesis by simultaneous esterification and esters in stirred-tank reactor. Bioresour.
synthesis. Green Chem. 12(2), 296–303 transesterification using oleophilic acid Technol. 99(14), 6097–6104 (2008).
(2010). catalyst. Ind. Eng. Chem. Res. 49(5), 109 Fu B, Vasudevan PT. Effect of organic

solvents on enzyme-catalyzed synthesis of of aliphatic esters on sulphided catalysts. Appl. Mallinson R, Resasco DE. Deoxygenation of
biodiesel. Energy Fuels 23(8), 4105–4111 Catal. A 326(2), 236–244 (2007). methylesters over CsNaX. J. Catal. 258(1),
(2009). 123 Ryymin E-M, Honkela ML, Viljava T-R, 199–209 (2008).
110 Xiao M, Mathew S, Obbard JP. Biodiesel fuel Krause AOI. Insight to sulfur species in the 135 Sang OY. Biofuel production from catalytic
production via transesterification of oils using hydrodeoxygenation of aliphatic esters over cracking of palm oil. Energy Sourc. A. 25(9),
lipase biocatalyst. GCB Bioenergy 1(2), sulfided NiMo/g-Al 2O3 catalyst. Appl. Catal. 859–869 (2003).
115–125 (2009). A 358(1), 42–48 (2009). 136 Quirino RL, Tavares AP, Peres AC, Rubim
111 Hernández-Martín E, Otero C. Different 124 Snåre M, Kubičková I, Mäki-Arvela P, Eränen JC, Suarez PAZ. Studying the influence of
enzyme requirements for the synthesis of K, Wärnå J, Murzin DY. Production of diesel alumina catalysts doped with tin and zinc
biodiesel: novozym 435 and lipozyme TL IM. fuel from renewable feeds: Kinetics of ethyl oxides in the soybean oil pyrolysis reaction.
Bioresour. Technol. 99(2), 277–286 (2008). stearate decarboxylation. Chem. Eng. J. Am. Oil Chem. Soc. 86(2), 167–172 (2008).
112 Yin J, Xiao M, Song J. Biodiesel from soybean J. 134(1–3), 29–34 (2007). 137 Knežević D, Schmiedl D, Meier D, Kersten S,
oil in supercritical methanol with co-solvent. 125 Mäki-Arvela P, Kubičková I, Snåre M, Eränen van Swaaij W. High-throughput screening
Energ. Convers. Manage. 49(5), 908–912 K, Murzin DY. Catalytic deoxygenation of technique for conversion in hot compressed
(2008). fatty acids and their derivatives. Energy Fuels water: quantification and characterization of
113 Hegel P, Mabe G, Pereda S, Brignole E. Phase 21(1), 30–41 (2007). liquid and solid products. Ind. Eng. Chem.
transitions in a biodiesel reactor using 126 Do PT, Chiappero M, Lobban LL, Resasco Res. 46(6), 1810–1817 (2007).
supercritical methanol. Ind. Eng. Chem. Res. DE. catalytic deoxygenation of methyl- 138 Hammerschmidt A, Boukis N, Hauer E et al.
46(19), 6360–6365 (2007). octanoate and methyl-stearate on Pt/Al 2O3. Catalytic conversion of waste biomass by
114 Saka S, Isayama Y, Ilham Z, Jiayu X. New Catal. Lett. 130(1–2), 9–18 (2009). hydrothermal treatment. Fuel 90(2), 555–562
process for catalyst-free biodiesel production 127 Rozmysłowicz B, Mäki-Arvela P, Lestari S (2011).
using subcritical acetic acid and supercritical et al. Catalytic deoxygenation of tall oil fatty 139 Elliott DC, Hart TR, Neuenschwander GG,
methanol. Fuel 89(7), 1442–1446 (2010). acids over a palladium-mesoporous carbon Rotness LJ, Zacher AH. Catalytic
115 Zhou C-HC, Beltramini JN, Fan Y-X, Lu catalyst: a new source of biofuels. Top. Catal. hydroprocessing of biomass fast pyrolysis
GQM. Chemoselective catalytic conversion of 53(15–18), 1274–1277 (2010). bio-oil to produce hydrocarbon products.
glycerol as a biorenewable source to valuable 128 Morgan T, Grubb D, Santillan-Jimenez E, Environ. Prog. Sust. Energ. 28(3), 441–449
commodity chemicals. Chem. Soc. Rev. 37(3), Crocker M. Conversion of triglycerides to (2009).
527–549 (2008). hydrocarbons over supported metal catalysts. n
Meticulous investigation of various
116 Johnson DT, Taconi KA. The glycerin glut: Top. Catal. 53(11–12), 820–829 (2010). upgrading procedures for bio-oils, especially
options for the value-added conversion of 129 Lestari S, Mäki-Arvela P, Eränen K, concerning the resulting liquid composition
crude glycerol resulting from biodiesel Beltramini J, Max Lu GQ, Murzin DY. of upgraded fast pyrolysis oils from
production. Environm. Progress. 26(4), diesel-like hydrocarbons from catalytic various sources.
338–348 (2007). deoxygenation of stearic acid over supported 140 Aho A, Kumar N, Lashkul AV et al. Catalytic
117 Kubička D, Kaluža L. Deoxygenation of Pd nanoparticles on SBA-15 catalysts. Catal. upgrading of woody biomass derived pyrolysis
vegetable oils over sulfided Ni, Mo and NiMo Lett. 134(3–4), 250–257 (2009). vapours over iron modified zeolites in a
catalysts. Appl. Catal. A 372(2), 199–208 130 Ping EW, Wallace R, Pierson J, Fuller TF, dual-fluidized bed reactor. Fuel 89(8),
(2010). Jones CW. Highly dispersed palladium 1992–2000 (2010).
118 Šimáček P, Kubička D, Šebor G, Pospíšil M. nanoparticles on ultra-porous silica 141 Aho A, Kumar N, Eränen K, Salmi T, Hupa
Hydroprocessed rapeseed oil as a source of mesocellular foam for the catalytic M, Murzin DY. Catalytic pyrolysis of woody
hydrocarbon-based biodiesel. Fuel 88(3), decarboxylation of stearic acid. Microporous biomass in a fluidized bed reactor: influence
456–460 (2009). Mesoporous Mater. 132(1–2), 174–180 (2010). of the zeolite structure. Fuel 87(12),
119 Kubička D, Bejblová M, Vlk J. Conversion of 131 Hancsók J, Krár M, Magyar S, Boda L, Holló 2493–2501 (2008).
vegetable oils into hydrocarbons over CoMo/ A, Kalló D. Investigation of the production of 142 Valle B, Gayubo AG, Alonso A, Aguayo AT,
MCM-41 catalysts. Top. Catal. 53(3–4), high cetane number bio gas oil from Bilbao J. Hydrothermally stable HZSM-5
168–178 (2010). pre-hydrogenated vegetable oils over Pt/ zeolite catalysts for the transformation of
HZSM-22/Al 2O3. Microporous Mesoporous crude bio-oil into hydrocarbons. Appl. Catal.
120 Kubička D, Šimáček P, Žilková N.
Mater. 101(1–2), 148–152 (2007). A 100(1–2), 318–327 (2010).
Transformation of vegetable oils into
hydrocarbons over mesoporous-alumina- 132 Lestari S, Mäki-Arvela P, Bernas H et al. 143 de MiguelMercader F, Groeneveld MJ,
supported CoMo catalysts. Top. Catal. catalytic deoxygenation of stearic acid in a Kersten SRA, Way NWJ, Schaverien CJ,
52(1–2), 161–168 (2009). continuous reactor over a mesoporous Hogendoorn J. Production of advanced
carbon-supported Pd catalyst. Energy Fuels biofuels: co-processing of upgraded pyrolysis
121 Huber GW, O’Connor P, Corma A.
23(8), 3842–3845 (2009). oil in standard refinery units. Appl. Catal. B
Processing biomass in conventional oil
refineries: production of high quality diesel by 133 Danuthai T, Jongpatiwut S, Rirksomboon T, 96(1–2), 57–66 (2010).
hydrotreating vegetable oils in heavy vacuum Osuwan S, Resasco DE. Conversion of 144 Mullen CA, Boateng AA. Catalytic
oil mixtures. Appl. Catal. A 329, 120–129 methylesters to hydrocarbons over an pyrolysis-GC/MS of lignin from several
(2007). H-ZSM5 zeolite catalyst. Appl. Catal. A sources. Fuel Process. Technol. 91(11),
361(1–2), 99–105 (2009). 1446–1458 (2010).
122 Şenol O, Viljava T-R, Krause AOI. Effect of
sulphiding agents on the hydrodeoxygenation 134 Sooknoi T, Danuthai T, Lobban LL, 145 Carlson TR, Tompsett GA, Conner WC,

Huber GW. Aromatic production from Dumesic JA. Integrated catalytic conversion content and composition in producer gas of
catalytic fast pyrolysis of biomass-derived of g-valerolactone to liquid alkenes for fluidized bed gasification of wood-Influence
feedstocks. Top. Catal. 52(3), 241–252 transportation fuels. Science 327(5969), of temperature and pressure. Environ. Prog.
(2009). 1110–1114 (2010). Sust. Energ. 28(3), 372–379 (2009).
146 Ferrari M, Maggi R, Delmon B, Grange P. 156 Gärtner CA, Serrano-Ruiz JC, Braden DJ, 168 Zwart RWR, Van der Drift A, Bos A, Visser
Influences of the hydrogen sulfide partial Dumesic JA. Catalytic coupling of carboxylic HJM, Cieplik MK, Könemann HWJ. Oil-
pressure and of a nitrogen compound on the acids by ketonization as a processing step in based gas washing-flexible tar removal for
hydrodeoxygenation activity of a CoMo/ biomass conversion. J. Catal. 266(1), 71–78 high-efficient production of clean heat and
carbon catalyst. J. Catal. 198(1), 47–55 (2009). power as well as sustainable fuels and
(2001). 157 Gürbüz EI, Kunkes EL, Dumesic JA. chemicals. Environ. Prog. Sust. Energ. 28(3),
147 Gutierrez A, Domine M, Solantausta Y. Dual-bed catalyst system for C-C coupling of 324–335 (2009).
Co-processing of upgraded bio-liquids in biomass-derived oxygenated hydrocarbons to 169 Pope CJ, Marrone PA, Yeh BV.
standard refinery units – fundamentals. fuel-grade compounds. Green Chem. 12(2), Thermodynamic driving forces for
Presented at: 15th European Biomass 223–227 (2010). postgasification carbon deposition. Environ.
Conference & Exhibition. Berlin, Germany, 158 Gärtner CA, Serrano-Ruiz JC, Braden DJ, Prog. Sus. Energ. 29(2), 151–162 (2010).
7–11 May 2007. Dumesic J. Catalytic upgrading of bio-oils by 170 Torres W, Pansare S, Goodwin Jr JG. Hot gas
148 Şenol O, Ryymin E-M, Viljava T-R, Krause ketonization. ChemSusChem 2(12), removal of tars, ammonia, and hydrogen
AOI. Effect of hydrogen sulphide on the 1121–1124 (2009). sulfide from biomass gasification gas. Catal.
hydrodeoxygenation of aromatic and aliphatic 159 Gürbüz EI, Kunkes EL, Dumesic Rev. 49(4), 407–456 (2007).
oxygenates on sulphided catalysts. J. Mol. J. Integration of C-C coupling reactions of 171 Simonetti DA, Rass-Hansen J, Kunkes EL,
Catal. A Chem. 277(1–2), 107–112 (2007). biomass-derived oxygenates to fuel-grade Soares RR, Dumesic JA. Coupling of glycerol
149 French RJ, Hrdlicka J, Baldwin R. Mild compounds. Appl. Catal. B 94(1–2), 134–141 processing with Fischer–Tropsch synthesis for
hydrotreating of biomass pyrolysis oils to (2010). production of liquid fuels. Green Chem.
produce a suitable refinery feedstock. Environ. 160 Serrano-Ruiz JC, Braden DJ, West RM, 9(10), 1073–1083 (2007).
Prog. Sustainable Energy. 29(2), 142–150 Dumesic J. Conversion of cellulose to 172 Gamba S, Pellegrini La, Calemma V,
(2010). hydrocarbon fuels by progressive removal of Gambaro C. Liquid fuels from Fischer–
150 Kunkes EL, Simonetti DA, West RM, oxygen. Appl. Catal. B 100(1–2), 184–189 Tropsch wax hydrocracking: isomer
Serrano-Ruiz JC, Gärtner CA, Dumesic JA. (2010). distribution. Catal. Today. 156(1–2), 58–64
Catalytic conversion of biomass to 161 McKendry P. Energy production from (2010).
monofunctional hydrocarbons and targeted biomass (part 3): gasification technologies. 173 Tavasoli A, Trépanier M, Dalai AK,
liquid-fuel classes. Science 322(5900), Bioresour. Technol. 83(1), 55–63 (2002). Abatzoglou N. Effects of confinement in
417–421 (2008). carbon nanotubes on the activity, selectivity,
162 Moilanen A, Nasrullah M, Kurkela E. The
151 West RM, Kunkes EL, Simonetti DA, effect of biomass feedstock type and process and lifetime of Fischer–Tropsch Co/carbon
Dumesic J. Catalytic conversion of biomass- parameters on achieving the total carbon nanotube catalysts. J. Chem. Eng. Data.
derived carbohydrates to fuels and chemicals conversion in the large scale fluidized bed 55(8), 2757–2763 (2011).
by formation and upgrading of mono- gasification of biomass. Environ. Prog. Sust. 174 Tavasoli A, Trépanier M, MalekAbbaslou
functional hydrocarbon intermediates. Catal. Energ. 28(3), 355–359 (2009). RM, Dalai AK, Abatzoglou N. Fischer–
Today. 147(2), 115–125 (2009). Tropsch synthesis on mono- and bimetallic
163 Kunkes E, Simonetti D, Dumesic J et al. The
152 Li N, Huber GW. Aqueous-phase role of rhenium in the conversion of glycerol Co and Fe catalysts supported on carbon
hydrodeoxygenation of sorbitol with Pt/ to synthesis gas over carbon supported nanotubes. Fuel Process. Technol. 90(12),
SiO2-Al2O3: Identification of reaction platinum-rhenium catalysts. J. Catal. 260(1), 1486–1494 (2009).
intermediates. J. Catal. 270(1), 48–59 (2010). 164–177 (2008). 175 Ma W, Kugler EL, Wright J, Dadyburjor DB.
153 Román-Leshkov Y, Barrett CJ, Liu ZY, 164 Rass-Hansen J, Christensen CH, Sehested J, Mo- Fe catalysts supported on activated
Dumesic J. Production of dimethylfuran for Helveg S, Rostrup-Nielsen JR, Dahl S. carbon for synthesis of liquid fuels by the
liquid fuels from biomass-derived Renewable hydrogen: carbon formation on Ni Fischer–Tropsch process: effect of mo
carbohydrates. Nature 447(7147), 982–985 and Ru catalysts during ethanol steam- addition on reducibility, activity, and
(2007). reforming. Green Chem. 9(9), 1016–1021 hydrocarbon selectivity. Energy Fuels 20(6),
154 Chheda J, Dumesic J. An overview of (2007). 2299–2307 (2006).
dehydration, aldol-condensation and 165 Matsumura Y, Minowa T, Potic B et al. 176 Herranz T, Rojas S, Perezalonso F, Ojeda M,
hydrogenation processes for production of Biomass gasification in near- and super- Terreros P, Fierro J. Genesis of iron carbides
liquid alkanes from biomass-derived critical water: Status and prospects. Biomass and their role in the synthesis of hydrocarbons
carbohydrates. Catal. Today 123(1–4), 59–70 Bioenergy 29(4), 269–292 (2005). from synthesis gas. J. Catal. 243(1), 199–211
(2007). (2006).
166 Campoy M, Gómez-Barea A, Villanueva AL,
n
Outline of aqueous-phase reforming Ollero P. Air-steam gasification of biomass in 177 Trepanier M, Tavasoli A, Dalai A, Abatzoglou
reactions to biofuels from sugars, with a fluidized bed under simulated autothermal N. Co, Ru and K loadings effects on the
emphasis on carbon yields, catalysts and adiabatic conditions. Ind. Eng. Chem. Res. activity and selectivity of carbon nanotubes
and selectivities. 47(16), 5957–5965 (2008). supported cobalt catalyst in Fischer–Tropsch
synthesis. Appl. Catal. A 353(2), 193–202
155 Bond JQ, Alonso DM, Wang D, West RM, 167 Wolfesberger U, Aigner I, Hofbauer H. Tar
(2009).

178 Chen W, Fan Z, Pan X, Bao X. Effect of 182 Kintisch E. The greening of synfuels. Science Websites
confinement in carbon nanotubes on the 320(5874), 306–308 (2008).
201 Emerging Markets Online.
activity of Fischer–Tropsch iron catalyst. 183 Henrich E, Dahmen N, Dinjus E. Cost www.emerging-markets.com/PDF/
J. Am. Chem. Soc. 130(29), 9414–9419 estimate for biosynfuel production via Biodiesel2020Study.pdf
(2008). biosyncrude gasification. Biofuels Bioprod. (Accessed 11 May 2011)
179 Liu S, Gujar AC, Thomas P, Toghiani H, Biorefin. 3(1), 28–41 (2009).
202 European Biodiesel Board.
White MG. Synthesis of gasoline-range Evaluation of costs, energy balance and
n
www.ebb-eu.org/prev_stats_production.php
hydrocarbons over Mo/HZSM-5 catalysts.
logistic challenges with Germany as a case (Accessed 11 May 2011)
Appl. Catal. A 357(1), 18–25 (2009).
for synthetic fuel production via decentral
203 Fischer–Tropsch archive.
180 Christensen JM, Mortensen PM, Trane R, pyrolysis and central gasification and www.fischer-tropsch.org
Jensen PA, Jensen AD. Effects of H2S and Fischer–Tropsch synthesis. (Accessed 11 May 2011)
process conditions in the synthesis of mixed
184 Knothe G. Biodiesel and renewable diesel: a
alcohols from syngas over alkali promoted
comparison. Progress Energ. Combust. Science
cobalt-molybdenum sulfide. Appl. Catal. A
36(3), 36, 364–373 (2010).
366(1), 29–43 (2009).
181 Durham E, Zhang S, Roberts C. Diesel-
n n
Thorough comparison of Renewable Diesel
length aldehydes and ketones via supercritical and FAME with respect to production, fuel
Fischer Tropsch synthesis on an iron catalyst. properties and engine performance.
Appl. Catal. A 386(1–2), 65–73 (2010).

Anders+T +Madsenم1

Uploaded by

Copyright:

Available Formats

Anders+T +Madsenم1

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Anders+T +Madsenم1

Uploaded by

Copyright:

Available Formats

Downloaded from orbit.dtu.

dk on: Nov 07, 2022

Catalytic Production of Biodiesel

Madsen, Anders Theilgaard

Link back to DTU Orbit

Anders Theilgaard Madsen

Catalytic Production of Biodiesel

Anders Theilgaard Madsen

Anders Theilgaard Madsen

Scientific peer-reviewed articles in international journals

Poster presentations at international conferences

Oral presentations at international conferences

Popular scientific articles (in Danish)

1 Catalysis, Biofuels, and Diesel 9

2 Catalytic Production of Fatty Acid Methyl Ester 43

3 Catalytic Hydrodeoxygenation of Fats and Oils 77

4 Outlook and Concluding Remarks 119

6 Included Publications 131

Catalysis, Biofuels, and Diesel 1

1.1 Catalysis - Setting the scene

1.2 Diesel as a biofuel

1.3 Fatty acid alkyl esters from fats and oils

1.3.1 Industrial production of fatty acid methyl esters

HO R + CH3OH H2O + H3CO R

1.3.2 Basic catalysts for transesterification

1.3.2.1 Inorganic metal-oxide bases

1.3.2.2 Organic amine bases

1.3.3 Acidic catalysts for esterification

1.3.3.1 Acidic inorganic oxides and derivatives

1.3.3.2 Acidic micro- and mesoporous silicalites

1.3.3.3 Organic sulphonic acids

1.3.4.1 Ionic liquid (trans-)esterification

1.3.4.2 Enzymatic (trans-)esterification

1.3.4.3 Supercritical (trans-)esterification

Figure 1.4: Overview of some of the reactions of glycerol.

1.3.5.5 Synthesis Gas

1.4 Hydrodeoxygenation of fats and oils

H R''= + "C3H2" + 3 H2O + 3 CO O O R'' 3 CO2 + "C3H2" + H R''

H R'''= O R''' H R'''

1.4.1 Supported sulphided metal catalysts

1.4.2 Supported transition-metal catalysts

1.4.3 Cracking-type catalysis

1.5 Alternative diesel fuels from biomass

1. Upgrading of bio-oils or bio-tars from flash pyrolysis

2. Aqueous-phase reforming of carbohydrates

3. Gasification and Fischer-Tropsch-Synthesis (FTS)

1.5.1 Upgrading of bio-oils

H3CO OCH3 OCH3 OCH3 OH OCH3 OH

1.5.1.1 Cracking & catalytic pyrolysis

1.5.1.2 Supported sulphided catalysts

1.5.1.3 Supported metal catalysts

1.5.2 Aqueous-phase reforming

Reforming reactions Reduction reactions

Figure 1.11: Principle behind aqueous-phase reforming of glucose.

1.5.2.1 Aqueous reforming of polyols

1.5.2.2 Condensationens prior to reduction

1.5.2.3 Product condensation & reduction

Carboxylix acid Carboxylic Acid Ketone

Figure 1.13: Ketonisation of carboxylic acids and hydrogenation.