Kinetic and Mechanistic Studies for the Direct
Conversion of Syngas to Ethanol
vorgelegt von
M.Sc.
Julia Ulrike Bauer
ORCID: 0000-0002-7610-6616
an der Fakultät II – Mathematik und Naturwissenschaften
der Technischen Universität Berlin
zur Erlangung des akademischen Grades
Doktor der Naturwissenschaften
– Dr.rer.nat. –
genehmigte Dissertation
Promotionsausschuss:
Vorsitzender: Prof. Dr. Matthias Drieß
Gutachter: Prof. Dr. Reinhard Schomäcker
Gutachter: Prof. Dr. Malte Behrens
Gutachter: Dr. Ing. Ralph Krähnert
Tag der wissenschaftlichen Aussprache: 03.12.2020
Berlin 2021
II
III
Selbstständigkeitserklärung
Hiermit erkläre ich, dass ich die vorliegende Arbeit selbstständig und eigenhändigsowie
ohne unerlaubte fremde Hilfe und ausschließlich unter Verwendung der aufgeführten
Quellen und Hilfsmittel angefertigt habe.
Berlin, den 28.07.2020
IV
V
Acknowledgement
„Alone we can do so little; together we can do so much.” – Helen Keller
Ich möchte an dieser Stelle die Gelegenheit nutzen, mich bei denjenigen zu bedanken, die
mich unterstützt und zum Gelingen dieser Arbeit maßgeblich beigetragen haben.
Zunächst gilt mein Dank Herrn Prof. Dr. Reinhard Schomäcker für die Übernahme des
Erstgutachtens und das fortwährende Interesse an meinen Arbeiten am BasCat. Herrn Prof.
Dr. Malte Behrens möchte ich herzlich für die Übernahme des Zweitgutachtens danken,
sowie Herrn Prof. Dr. Matthias Drieß für die Übernahme des Prüfungsvorsitzes.
Mein besonderer Dank gilt Dr. Frank Rosowski für die Möglichkeit, meine Promotion am
BasCat durchführen zu können. Im Laufe meiner verschiedenen Stationen am BasCat durfte
ich verschiedene Facetten der heterogenen Katalyse kennenlernen und die ganze Zeit über
das Gefühl erfahren, dass meine Arbeit sehr wertgeschätzt wurde. Vielen Dank dafür und
die vielen Diskussionen, die meine vorigen Arbeiten und insbesondere diese maßgeblich
beeinflusst haben.
Dr. Ing. Ralph Krähnert möchte ich für die Betreuung meiner wissenschaftlichen Arbeit
danken. Die zahllosen Diskussionen, detaillierten Nachfragen und Ideen haben die Arbeit zu
dem gemacht, was sie heute ist und ich habe sehr viel gelernt, sowohl wissenschaftlich als
auch methodisch.
Alle Probleme wiegen nur halb so schwer, wenn man sie mit einem großartigen Team teilen
kann. Und während ich über meine Jahre am BasCat viele großartige Menschen
kennenlernen, Freundschaften schließen und mir jederzeit Unterstützung gewiss sein
durfte, möchte ich ein paar Leute besonders hervorheben. Micha und Raoul, ihr seid das
Herz und der Motor des BasCat. Ohne eure Unterstützung, würde wohl keine Anlage laufen,
kein GC messen, kein Gas dosiert und kein Katalysator getestet werden. Ich danke euch
vielmals, für eure tatkräftige Unterstützung und eure Freundschaft. Ema, Du bist die Seele
hinter allem. Ich danke Dir, für organisatorische Kunststücke und stetes Interesse an
unserem Seelenheil. Phil und Kristian danke ich besonders für die vielen langen TEM-
Sessions, die ihr für mich absolviert habt. Auch hatte ich viel Unterstützung beim Betrieb
des Dinos. Danke an Jan für volle Gasflaschen, Schraubarbeiten im Downstreamofen und
gefüllten Ersatzteillagern, sowie an Daniel, Vincent und Gerrit für saubere Reaktoren und
gesiebte Katalysatorproben. Ein besonders großes Dankeschön schulde ich Felix und Martin
für eure engagierte und kollegiale Unterstützung und der großartigen Stimmung im Team
Dino – trotz seiner gelegentlichen Unarten. Benjamin möchte ich danken für seine Hilfe bei
VI
allen Fragen zur Kinetik und spontanen Gleichgewichtsrechnungen und Stephen für
Katalysatorsynthesen und die vielen kleinen Dinge.
Auch außerhalb meiner wissenschaftlichen Arbeit möchte ich dem BasCat-Team danken für
die schönen Zeiten, die Gesellschaft, erlebte Konferenzen und Aufnahme des täglichen
Gesprächsbedarfs; Christian besonders für die tolle Bürogemeinschaft, Verena und Miriam
für die Freundschaft, Anton, Rhea, Kristian, Piyush und Felix für die Begleitung über eine so
lange Zeit.
Also, I want to thank all other BasCats over the years for the great comradeship, especially
for “super awesome” salad group, beer meetings and discussions about life. I’m happy to
call you my friends.
Dr. Manuel Gentzen, Dr. Christiane Kuretschka und Sabine Borchers (BASF) danke ich für
die Bereitstellung der Katalysatorproben. Den restlichen Projektpartnern von BASF
(Dr. Lukasz Karwacki, Dr. Christian Almer, Dr. Ekkehard Schwab) und hte (Dr. Chiara
Boscagli, Dr. Ivana Jevtovikj, Dr. Harry Kaiser) danke ich für hilfreiche Diskussionen in den
verschiedenen Phasen der Projekte.
Für die Analytik zu den Katalysatorproben und Beantwortung aller Fragen dazu möchte ich
mich ebenfalls bedanken. Dr. Johannes Schmidt, Dr. Shuang Li und Meng-Yang Ye von der
Arbeitsgruppe Funktionsmaterialien der TU Berlin danke ich für XPS Messungen. Vom
Fritz-Haber-Institut danke ich Dr. Frank Girgsdies, Jasmin Allan und Maike Hashagen für
XRD und BET Messungen, außerdem Dr. Milivoj Plodinec für
Elektronenmikroskopieaufnahmen, sowie Dr. Thomas Lunkenbein und Dr. Walid Hetaba
für die Möglichkeit, im Rahmen des ChemieTEM Projektes das Elektronenmikroskop zu
nutzen.
Last but certainly not least: Ich möchte mich für die enorme Unterstützung durch meine
Familie und Freunde bedanken. Meine Bayreuther Mädels, ihr seid ein unumstößlicher
Baustein meines Chemiestudiums und noch wichtiger, allem Drumherum seit über 10
Jahren. Ich freue mich auf zahllose weitere Ausflüge mit euch, die einen stets etwas
erschöpft, aber umso glücklicher in den Alltag entlassen.
Wer meine Familie kennt, weiß wie glücklich ich mich schätzen kann. Vielen, vielen Dank
für den Halt und die Liebe. Hanny, dich zähle ich hier dazu.
Und Phil, danke für alles!
VII
Abstract
The direct conversion of syngas to ethanol (STE) is a promising route to ethanol from fossil
and non-fossil carbon resources. Rh-based catalysts offer the most promising results so far.
However, the accomplished yields and rates still must be improved before industrial
applications become viable. Owing to the complexity of the reaction network and the
catalyst's behavior under high pressure reaction conditions, the entire complexity of this
reaction at process relevant conditions has not been unraveled so far.
The current study develops a comprehensive picture of the STE reaction network and
reaction-induced changes of Rh/SiO2 catalysts at industrially relevant pressures,
temperature, time scale and catalyst complexity. Moreover, a reaction network is derived
that describes key influences of promoting elements Mn and Fe, including the dynamics of
surface restructuring.
The study investigates Rh/SiO2 catalysts in the absence and presence of Mn or Fe or both.
Reaction mechanism and kinetics were studied under high pressure conditions for
extended time scales with repeatedly tested standard conditions to assess catalyst aging.
The physicochemical bulk- and surface properties of all catalysts were comprehensively
analyzed prior to reaction as well as after different reaction times and atmospheres.
Catalytic measurements that included variations of reactant partial pressures, temperature,
contact time as well as cofeed and dropout experiments revealed a reaction network for
Rh/SiO2. Key features of the reaction network include a CO insertion mechanism as the only
plausible C-C-coupling step and the existence of two adsorbed CO* species with different
reactivity – one being hydrogenated and dissociated and the other performing CO insertion.
The necessity of two different CO* types implies the existence of two active sites that might
be different in nature. While hydrogenation reactions and C-O bond breaking were assigned
to metallic Rh, CO insertion possibly occurs on dynamically formed non-metallic Rh species
or clusters. A dynamic behavior of Rh metal particle surfaces is observed, which strongly
depends on CO partial pressure and time on stream.
The reaction network proposed for Rh/SiO2 proves to be generic in nature and describes
major aspects of the reaction network also for RhMn/SiO2, RhFe/SiO2, and RhMnFe/SiO2.
The following key features were identified for the promoting elements: (i) Mn stabilizes the
CHxCHO* intermediate resulting in enhanced ethanol, acetaldehyde, acetic acid yields,
(ii) Fe accelerates hydrogenation rates in general and methanol formation in particular, and
(iii) over RhMnFe/SiO2, the combination of both effects achieves the highest ethanol
selectivities among the tested catalysts.
VIII
The performed catalyst characterization also yields a comprehensive picture of the
catalyst's structure as well as reaction-induced reversible and irreversible changes.
Sintering and agglomeration strongly depend on reaction atmosphere, time on stream and
promoting elements. If present, Fe exists in close proximity to Rh and strongly modifies the
reactivity of Rh. In contrast, Mn appears to serve the role as oxidic support phase which
reduces Rh mobility and thus prevents sintering and agglomeration.
The derived insights provide a comprehensive picture of Rh/Fe/Mn based catalysts under
realistic operation conditions and pave the way to the development of improved catalysts
and reactor concepts.
IX
Kurzfassung
Die “Syngas to Ethanol (STE)”-Reaktion ist eine vielversprechende alternative Methode zur
Gewinnung von Ethanol aus fossilen und regenerativen Kohlenstoffquellen. Rh-basierte
Katalysatoren zeigten dabei bisher die vielversprechendsten Resultate. Für eine rentable
kommerzielle Nutzung müssen Reaktionsraten und Selektivitäten jedoch weiter verbessert
werden. Die gesamte Komplexität der Reaktion bei prozessrelevanten Bedingungen wurde
bisher nicht beschrieben, vor allem aufgrund des komplexen Reaktionsnetzwerkes sowie
des Verhaltens der Katalysatoren bei hohem Druck unter Reaktionsbedingungen.
Die vorliegende Arbeit bietet eine umfassende Darstellung der STE-Reaktion und zeigt
durch die Reaktion induzierte Veränderungen von Rh/SiO2 Katalysatoren bei
prozessrelevanten Drücken, Temperaturen und Zeiträumen auf. Es wird zudem ein
Reaktionsnetzwerk abgeleitet, welches sowohl die Betrachtung dynamischer
Oberflächenrestrukturierungen als auch die Identifikation der Schlüsseleffekte von Mn und
Fe als zusätzliche Elemente erlaubt.
In dieser Studie wurden Rh/SiO2 Katalysatorsysteme sowohl mit als auch ohne Mn
und/oder Fe untersucht. Diese katalytischen Untersuchungen umfassten neben kinetischen
und mechanistischen Experimenten auch lange Reaktionslaufzeiten mit wiederkehrenden
Messungen bei Referenzbedingungen. Dies ermöglichte eine fundierte Betrachtung der
Katalysatoren hinsichtlich ihrer Stabilität und Alterung. Eine umfängliche
Charakterisierung der Katalysatoren vor und nach den katalytischen Untersuchungen
diente der Bestimmung der relevanten, physikochemischen Eigenschaften der Proben und
ihrer Oberflächen unter dem Einfluss von verschiedenen Reaktionszeiten
und -atmosphären.
Das Reaktionsnetzwerk wurde aus katalytischen Daten von Rh/SiO2 abgeleitet, welche nicht
nur Partialdruck-, Temperatur- und Kontaktzeitvariationen, sondern auch die Zudosierung
von Reaktionsintermediaten und Drop-out-Experimente umfassten. Integraler Bestandteil
des Reaktionsnetzwerk ist zum einen die CO-Insertion als einziger C-C-
Kupplungsmechanismus und zum anderen die Existenz von zwei verschiedenen CO-
Adsorbaten welche sich hinsichtlich ihrer Reaktivitäten unterscheiden. Die eine wird
hydriert und dissoziiert, während die andere der CO-Insertion dient. Die Notwendigkeit von
zwei CO*-Spezies impliziert das Vorhandensein von zwei unterschiedlichen, aktiven
Zentren. Hydrierreaktionen und C-O-Bindungsspaltungen können der metallischen Rh-
Oberfläche zugeordnet werden, während die CO-Insertion wahrscheinlich auf
nichtmetallische Rh-Spezies oder Cluster zurückzuführen ist. Es wurde dabei ein
X
dynamisches Verhalten der Rh-Partikeloberfläche beobachtet, die eine starke Abhängigkeit
von CO Partialdruck und der Reaktionslaufzeit zeigt.
Das aufgestellte Reaktionsnetzwerk erwies sich als universell anwendbar auch für die drei
modifizierte Katalysatoren RhMn/SiO2, RhFe/SiO2, und RhMnFe/SiO2. Folgende
Oberflächenschritte wurden als elementar für das charakteristische Verhalten der
einzelnen Katalysatoren identifiziert: (i) Mn zeigt hohe Ausbeuten für Ethanol, Acetaldehyd
und Essigsäure durch die Stabilisierung des CHxCHO*-Intermediats, (ii) Fe erhöht die Raten
für Hydrierreaktionen im Allgemeinen und die Methanolbildung im Speziellen und (iii) an
RhMnFe/SiO2 führt die Kombination beider Effekte zu den höchsten erzielten
Ethanolselektivitäten unter den untersuchten Katalysatoren.
Die sorgfältige Charakterisierung der Katalysatoren liefert ein umfassendes Bild der
Katalysatorstrukturen sowie reversibler und irreversibler Änderungen durch die Reaktion.
Sintern und Agglomeration hängen dabei stark von der Reaktionsatmosphäre,
Reaktionslaufzeit und Katalysatorzusammensetzung ab. Wenn vorhanden, befindet sich Fe
in engem Kontakt zu Rh auf der Oberfläche und beeinflusst maßgeblich die Reaktivität des
Rh-Katalysators. Mn scheint dagegen als eine oxidische Trägerschicht zu agieren, durch die
die Rh-Mobilität reduziert und Partikelwachstum und Agglomeration verhindert werden
können.
Die abgeleiteten Erkenntnisse zeichnen ein umfassendes Bild von Rh/Fe/Mn-basierten
Katalysatoren unter realistischen Betriebsbedingungen und ebnen den Weg für die
Entwicklung von verbesserten Katalysatoren und Reaktorkonzepten.
XI
Content
Abstract.................................................................................................................................................VII
Kurzfassung...........................................................................................................................................IX
Content....................................................................................................................................................XI
SymbolsandAbbreviations..........................................................................................................XIII
1Scopeandoutlineofthesis.......................................................................................................1
2Stateoftheart...............................................................................................................................3
2.1
Demand and supply of ethanol .............................................................................................................................. 3
2.2
CO hydrogenation over transition metal catalysts ........................................................................................ 5
2.3
Syngas to ethanol (STE) ............................................................................................................................................ 8
Catalysts for STE .......................................................................................................................................................... 8
Reaction and thermodynamics ........................................................................................................................... 11
Reaction mechanisms ............................................................................................................................................. 12
Deduced research needs ........................................................................................................................................ 15
3Experimentaldetails................................................................................................................17
3.1
Catalyst synthesis ..................................................................................................................................................... 17
3.2
Gases for catalytic testing ...................................................................................................................................... 18
3.3
Catalytic test setup “Dino” ..................................................................................................................................... 19
3.4
Experimental procedures for catalytic testing ............................................................................................. 22
3.5
Experimental design of catalytic testing ......................................................................................................... 24
In-situ H
2
treatment ................................................................................................................................................. 24
Initial equilibration time at standard conditions ........................................................................................ 24
Parameter field test (DinoRun26) ..................................................................................................................... 24
Thermal stability test (DinoRun27) .................................................................................................................. 26
Cofeed and dropout experiments (DinoRun29) .......................................................................................... 26
Preparation of samples after CO dropout for characterization (DinoRun31) ................................ 28
Controlled cool-down of samples after reaction ......................................................................................... 28
3.6
Evaluation of catalytic data .................................................................................................................................. 29
Calculations ................................................................................................................................................................. 29
Statistical tree approach ........................................................................................................................................ 31
3.7
Catalyst characterization methods .................................................................................................................... 32
4MorphologyandsurfacepropertiesofRh/Mn/Fe/SiO2catalysts...........................35
4.1
Structural properties (XRD, XPS) ....................................................................................................................... 37
4.2
Morphology and elemental distribution (STEM-EDX) .............................................................................. 41
4.3
Conclusion ................................................................................................................................................................... 47
5Stabilityandreproducibilityofcatalyticdata................................................................49
5.1
Carbon balance .......................................................................................................................................................... 49
5.2
Catalytic stability ...................................................................................................................................................... 50
Thermal stability ....................................................................................................................................................... 50
Long-term stability ................................................................................................................................................... 52
Stability after cofeed and dropout experiments .......................................................................................... 55
5.3
Reproducibility .......................................................................................................................................................... 57
XII
6ReactionnetworkandkineticsforSTEoverRh/SiO2..................................................61
6.1
Product spectrum and selectivities ................................................................................................................... 61
6.2
Initial formation phase ........................................................................................................................................... 63
6.3
Influence of CO and H
2
partial pressures ........................................................................................................ 65
Consumption rates ................................................................................................................................................... 65
Product formation rates ......................................................................................................................................... 67
Apparent reaction orders ...................................................................................................................................... 70
6.4
Contact time variation ............................................................................................................................................ 74
6.5
Cofeed studies ............................................................................................................................................................ 77
Ethylene cofeed ......................................................................................................................................................... 79
Propylene cofeed ...................................................................................................................................................... 80
Acetaldehyde cofeed ................................................................................................................................................ 82
CO
2
cofeed .................................................................................................................................................................... 84
Mechanistic information derived from cofeeds ........................................................................................... 84
6.6
Dropout experiments .............................................................................................................................................. 86
6.7
Deduced reaction network ................................................................................................................................... 89
7ModificationofRh/SiO2withMnand/orFe.....................................................................95
7.1
Product spectrum and selectivities ................................................................................................................... 95
7.2
Initial formation phase ........................................................................................................................................... 98
7.3
Influence of CO and H
2
partial pressures ...................................................................................................... 100
7.4
Contact time variation .......................................................................................................................................... 111
7.5
Cofeed studies .......................................................................................................................................................... 113
7.6
Dropout experiments ............................................................................................................................................ 119
7.7
Effects of modification on reaction network ............................................................................................... 122
8GeneralDiscussion.................................................................................................................127
8.1
Reaction network for Rh-based catalysts ..................................................................................................... 127
8.2
Evidence for two CO* species with different reactivity .......................................................................... 130
8.3
Dynamics of changing catalytic properties under reaction conditions ........................................... 133
8.4
Influence of catalyst composition on structure and reactivity ............................................................ 135
8.5
Parameters for steering catalytic performance ......................................................................................... 137
8.6
Challenges for practical catalyst operation .................................................................................................. 139
9ConclusionandOutlook.......................................................................................................141
References.........................................................................................................................................143
Appendix............................................................................................................................................157
A
Experimental details ............................................................................................................................................................... 157
B
Morphology and surface properties of Rh/Mn/Fe/SiO
2
catalysts ....................................................................... 159
C
Stability and reproducibility of catalytic data .............................................................................................................. 179
D
Reaction network and kinetics for STE over Rh/SiO
2
............................................................................................... 185
E
Modification of Rh/SiO
2
with Mn and/or Fe ................................................................................................................. 191
XIII
Symbols and Abbreviations
Symbols
𝑛corrected mole fraction of compound i
𝑛, mole fraction of compound i as obtained from chromatogram
𝑛, mole fraction of Ar in reactor outlet
𝑛, mole fraction of Ar in bypass
𝑛, mole fraction of CO in the inlet gas
𝐶 carbon number of the product i
𝑉 total volume flow per reactor
𝑚 catalyst mass
𝑝 pressure at normal conditions (101.325 kPa)
𝑇 temperature at normal conditions (298 K)
𝑅 ideal gas constant (8.314 J K-1 mol-1)
𝑉 catalyst volume
𝑝 pressure inside the reactor
𝑇 temperature inside the reactor
𝛼chain growth probability
𝑑 mean size of ordered (crystalline) domains
𝐾 dimensionless shape factor
𝜆 X-ray wavelength
𝛽 line broadening at half the maximum intensity (FWHM)
𝜃 Bragg angle
𝑥, fraction of Rh on surface from XPS in wt-%
𝑥, Rh loading from ICP
𝑑 Rh particle size
𝑉 volume of one Rh atom
𝜎 cross-sectional area for one Rh atom
𝑘rate constant
𝑘frequency factor
𝐸,apparent activation energy
XIV
Abbreviations
ASFAnderson-Schulz-Flory distribution
BET Brunauer-Emmet-Teller
C1 Reaction products with a carbon number of 1
C2 Reaction products with a carbon number of 2
C2+ Reaction products with a carbon number of at least 2
C3 Reaction products with a carbon number of 3
DFT Density functional theory
EDX Energy-dispersive X-ray spectroscopy
EELS Electron energy loss spectroscopy
FID Flame ionization detector
GC Gas chromatography / Gas chromatograph
GHSV Gas hourly space velocity
HAADF High-Angle Annular Dark Field
ICP-OES Inductively coupled Plasma – Optical Emission Spectroscopy
m(CO) Reaction order of CO
MFC Mass flow controller
MS Mass spectrometry
n(H2) Reaction order of H2
oxy Oxygenates
STE Syngas to ethanol
(S)TEM (Scanning) transmission electron microscopy
TCD Thermal conductivity detector
TOS Time on stream
TPR Temperature-programmed reduction
XAS X-ray adsorption spectroscopy
XPS X-ray photoelectron spectroscopy
XRD X-ray diffractometry
1
1 Scope and outline of thesis
The direct conversion of syngas to ethanol (STE) is an approach for ethanol production
which is highly desired by industry. For commercially feasible operation, the development
of a high-performance catalyst is required. Extensive catalyst screening has provided an
overview of moderately active and selective catalysts. However, ethanol yields have to be
improved for industrial application.
The existing literature on STE provides many insights on specific questions regarding
reactivity of differently promoted Rh-based catalysts and the underlying reaction network.
However, simplification of the complex product spectrum and missing data targeting the
time on stream behavior, systematic evaluation of the reaction network by contact time
variation and cofeed experiments and application of reaction conditions in a limited range
prevented an understanding on the full complexity of this reaction so far.
Aim of this thesis is to provide a generalized reaction network with considerations on the
complex interplay of reaction conditions, catalyst structure, and thus key surface reactions
that crucially determine the characteristic catalytic performance of differently promoted
Rh/SiO2 catalysts. Therefore, detailed investigations were performed on the direct
conversion of syngas to ethanol over Rh-based catalysts targeting three aspects:
i) Understanding the interconnectivity of product formation over Rh/SiO2 to
provide a reaction network containing formation pathways to all reaction
products with information about their elementary steps and how they must be
enhanced or inhibited to achieve higher ethanol formation rates
ii) Insight into the effect of catalyst modification with the literature-known
promoters Mn and Fe to provide a basis how additional metals modify the
reaction network and why that leads to the observed performance
iii) Characterization of the catalyst structure depending on reaction conditions and
catalyst composition aiming for structural details that might define selectivities
for CO hydrogenation in general and for ethanol formation specifically and how
industrially relevant reaction conditions impact these structures
iv) Investigation of the dynamics of catalytic properties over of long time on stream
and different, realistic reaction conditions in a comparable manner for all four
considered catalyst systems in order to assess the complex interplay of reaction
conditions, catalyst structure and resulting performance
On the first front, an extensive kinetic dataset for Rh/SiO2 under variation of partial
pressures, temperatures and contact times was obtained and complemented by cofeed and
dropout experiments. Kinetic and mechanistic data were evaluated in detail providing a
solid basis for the development of a plausible reaction network over Rh/SiO2.
1Scopeandoutlineofthesis
2
The second target was approached simultaneously as the three modified catalysts
RhMn/SiO2, RhFe/SiO2, and RhMnFe/SiO2 were tested in parallel with Rh/SiO2 using a 4-
channel catalytic test setup. The obtained data were comparatively evaluated leading to the
identification of specific surface steps in the network that are affected by the addition of the
respective metal.
Thorough catalyst characterization in several different states is the basis to face the third
objective. Samples of all four catalysts were characterized with several analytical
techniques involving surface-sensitive and bulk methods. The obtained structural
information obtained for different catalyst compositions was related to catalytic effects of
the respective metal. Also, information about the impact of different reaction conditions on
the catalyst structure was obtained.
The fourth aspect was approached by careful evaluation of all obtained data and extraction
of the main factors that affect catalytic performance under comparison of the data sets
simultaneously obtained for the four catalyst systems.
The thesis is structured in nine chapters. In chapter2, the necessity of the development of
an alternative ethanol production route is explained as well as the current state-of-the-art
research regarding the direct conversion of syngas to ethanol. Chapter3 provides the
reader with all experimental details regarding catalytic testing and catalyst characterization
procedures. In chapter4, the results of catalyst characterization for all four catalyst
systems are provided with a focus on structural aspects such as particle size and dispersion
as well as the elemental distribution on the silica surface. Chapter5 evaluates the catalytic
stability over time on stream and temperature, and the reproducibility of catalytic results
four all catalysts. The systematic assessment of the obtained catalytic data for Rh/SiO2 and
detailed kinetic and mechanistic interpretation of the results is provided in chapter6. The
chapter closes with the proposal of a plausible reaction network und consideration of the
presented catalytic data. In chapter7, the obtained catalytic data for the modified catalysts
are presented and compared to the conclusions of the previous chapter. The effects of each
metal on the reaction network is demonstrated in the last part of this chapter. Chapter8
provides a general discussion of the presented data revealing six superordinate subjects
that could be extracted from the extensive dataset. Here, the insight into the catalyst
reactivity is related to the obtained structural and morphological information about the
catalysts. In chapter9, all insights are summarized in a short conclusion addressing the
questions targeted by this study as described above. Finally, a short outlook towards future
research and unanswered questions is given.
3
2 State of the art
Depletion of fossil resources and an increasing demand of a growing world population lead
to new challenges to provide industry and society with chemicals. In this chapter,
considerations on the demand and supply of ethanol as important base chemical will be
given and the state of the art of the direct synthesis of ethanol from syngas will be
elucidated.
2.1 Demand and supply of ethanol
The worldwide production of ethanol has constantly increased during the last years from
29 million m3 in 2000 to 128 million m3 in 2018.1 Brazil and the United States are the main
producers. A major driving force for the increasing demand is the use of ethanol as fuel or
fuel additive. In 2000, less than 60 % of the produced ethanol was used for fuel related
purposes. In 2018, the percentage increased to 85 %. Official incentives to use ethanol as a
fuel additive in China and the United States contributed to the continuing growth.2
Specifically, ethanol increases the octane number and combustion efficiency and will at least
partially replace methyl tert-butyl ether (MTBE) in today’s fuels.3,4
The use of ethanol as a hydrogen carrier is another field of current exploration. It might be
applied as a liquid hydrogen storage compound that is produced at a central site and then
transported to the point of use. It could then be used directly in a fuel cell5,6 or be reformed
via steam reforming7,8 or partial oxidation.9,10
Today, ethanol is mainly produced by two processes – the fermentation of sugars and
hydration of petroleum-based ethylene.11 However, the hydration of ethylene is
economically unfeasible in large scale because of the high costs of the resource. The
fermentation route generates most of the ethanol produced today. Sugars derived from corn
or sugar cane serve as the carbon source. In general, these resources are sustainable, but
their production directly competes with food production which causes ethical constraints.
Sugars derived from non-food biomass such as woody or inedible parts of a plant cannot be
used in the existing fermentation processes because of their composition.12 Also, the
production of fuel-grade ethanol from fermentation is inefficient and expensive due to
energy-intensive distillation steps.13,14
The growing demand and the outlined constraints of today’s ethanol routes make
alternative approaches for ethanol production highly desirable. Syngas as a carbon source
can be obtained from various sources such as coal, (un)conventional natural gas or all kinds
2Stateoftheart
4
of biomass. Syngas (short for synthesis gas) describes a mixture of CO and H2. Sometimes,
other compounds such as methane and CO2 are included under the same name. Syngas is
known in industry as important building block to produce a wide range of chemicals.15
Ethanol production from syngas is currently realized in very small scale viafermentation of
syngas by microorganisms.16,17 In China, a three step process involving methanol synthesis
from syngas, carbonylation of methanol to acetic acid and selective hydrogenation to
ethanol has been realized (Figure2.1).18,19
Figure2.1Schemeofthinkableethanolproductionpathwaysfeaturingprocessescommercializedinlargescale
(green),smallscale(blue)anduncommercializedreactions(white).Thisstudyfocusesonthedirectconversionof
syngastoethanolmarkedinred.AdoptedfromLuketal.17withpermissionfromTheRoyalSocietyofChemistry.
Other routes are proposed in literature, e.g. methanol homologation,20 coupling of methanol
with CO to form dimethyl oxalate and subsequent hydrogenation to ethanol,21,22 methanol
dehydrogenation to DME with subsequent carbonylation and hydrogenolysis of methyl
acetate23 and the direct conversion of syngas to alcohols. The latter is a particularly
promising route with low capital and operation costs of the single step process.17 This study
focuses on the direct formation of ethanol from syngas.
Direct
conversion
Methanol
synthesis
Biomass Fossil fuels
Syngas
Ethanol
Methanol and
CO coupling
DMO
hydrogenation
Methanol
carbonylation
Oxygenates
hydrogenation
Sugar
fermentation
Methanol
homologation Fermentation Ethylene
hydration
2.2COhydrogenationovertransitionmetalcatalysts
5
2.2 CO hydrogenation over transition metal catalysts
Choice of catalyst composition and reaction conditions have a crucial impact on reaction
pathways and selectivities. In general, hydrogenation of CO can yield a wide range of
reaction products such as methanol and methane, but also higher olefins, paraffins, alcohols,
aldehydes, and other oxygenates with varying carbon number.24 The reason for the large
pool of reaction products can be found in the characteristics of the reaction networks
featuring a large set of parallel and series reaction pathways (Figure2.2).25 Depending on
the material, the catalyzed surface steps include CO dissociation (direct or H-assisted),
hydrogenation of various surface fragments (CO, C, higher fragments) and C-C coupling
steps. The latter include coupling of a CHx fragment to other CxHy or CxHyO fragments
(carbide mechanism) and CO insertion into a metal-C bond. CO insertion can lead to chain
growth or termination.
Figure2.2SchemeofpossiblereactionpathwaysforCOhydrogenationcontainingallrelevantsurfacereactionsand
potentiallyformedproducts.25
In general, materials that catalyze CO hydrogenation must be able to adsorb H2 and CO.
While the dissociative adsorption of H2 is necessarily required, CO can be adsorbed in a
dissociative or non-dissociative way. The form of CO adsorption is thereby a key property
of the catalytic material to determine the selectivity and related to the position of a
transition metal in the periodic table (Figure2.3).
2Stateoftheart
6
Figure2.3DifferentmodesofCOadsorptionontransitionmetalsurfacesindependenceontheirpositioninthe
periodictable.Allmetalstotheleftoftheambienttemperatureline(grey)adsorbCOdissociativelyandmetalsto
therightadsorbCOmolecularlyatambienttemperatures.25,26Accordingly,thelinefortemperatureof200‐300°C
isgiven.Colorsrepresenttheabilityofmetalstoformstablecarbonylcompoundsatroomtemperature.27
The CO dissociation activity thereby is related to the Fermi level of metals and electron back
donation from the metal surface to the empty 2π* anti-bonding orbital of the adsorbed CO.
Van Santen and Neurock explained specific selectivity by C-O bond dissociation barriers
revealing similar results as expected from Fermi level assumptions.28 However, different
adsorption modes must be assumed to be present on the metal surface at the same time
under realistic conditions.29
The form of CO adsorption has fundamental impact on the product selectivity upon CO
hydrogenation. Fast C-O bond dissociation and formation of CHx species on the surface leads
to formation of methane and higher hydrocarbons known as the Fischer-Tropsch reaction.
This reaction has been studied intensely since it was first reported by Franz Fischer and
Hans Tropsch in 1923.26,30,31 Typical catalysts can be found close to the ambient
temperature line in the periodic system (Figure2.3), e.g. Co,32,33 Fe,34–36 and sometimes
Ru.37,38
The product distribution obtained from the Fischer-Tropsch reaction can be described as
the results of a surface polymerization reaction which proceeds via stepwise addition of a
monomer carbon species. Instead of presenting the selectivities of all observed products
individually, an empirical statistical product distribution model known as Anderson-Schulz-
Flory (ASF) distribution was developed for this purpose. It allows the calculation of a chain
growth probability α as a simple descriptor for the product distribution.30 For an ideal ASF
2.3Syngastoethanol(STE)
7
distribution, a chain growth mechanism via a C1 monomer, a constant ratio of the chain
growth rate relative to the rate of desorption, and only one product type must be assumed.
To account for more complex product mixtures, deviations of the observed distribution
from the ASF law must be considered. Often, a chain growth probability for each product
type is calculated.
Metals that are located to the right of the 200-300 °C line adsorb CO molecularly and
catalyze methanol formation. Methanol synthesis from syngas also has been heavily
investigated in industry and academic research. 39–43 Selectivity towards methanol has been
shown for Cu,44,45 Pt,46,47 Pd,48 and Ir.49 However, the most intensely investigated catalysts
are Cu/ZnO50,51 and Cu/ZnO/Al2O352,53 catalysts.
Transition metals between the two lines exhibit medium CO dissociation ability. Both
mechanisms take place in parallel. Higher oxygenate formation requires C-O bond scission
as well as molecular CO insertion.54 The right balance of different species is thereby crucial
for selective oxygenate formation.55 In particular, Rh has been shown to catalyze C2 – C4
oxygenate formation. While the Fischer-Tropsch reaction and methanol synthesis are
widely applied in industry, commercialization of C2 oxygenate formation form syngas could
not be realized yet despite extensive research efforts.
However, syngas is heavily applied in combination with Rh catalysts for the
hydroformylation of alkenes. The hydroformylation reaction was discovered by Otto
Roelen56 and is the largest homogeneously catalyzed process in industry.57 Originally, Co
catalysts58 were used until Wilkinson etal. invented Rh catalysts with improved
properties.59 The hydroformylation mechanism requires CO insertion into a metal-carbon
bond and is therefore closely related to the oxygenate formation pathway in CO
hydrogenation.
Besides the position in the periodic table, the ability to form carbonyl species might also
determine intrinsic product selectivity. Many transition metals can form stable carbonyl
complexes (compare Figure2.3). Some metal carbonyl complexes such as Ni(CO)4, Co2(CO)8
and Fe(CO)5 are formed directly from the metal in presence of CO, partly under high CO
pressures, e.g. Fe carbonylation at 150–200 °C and 50–200 bar CO pressure. Others are
rather accessible from metal halides or metal organic precursors like Mn2(CO)10 from Mn
acetate or Rh4(CO)12 from Rh chloride. For supported metal catalysts, the ability to form
surface (sub)carbonyls in CO or CO/H2 atmosphere is of particular interest. Such species
have been reported for Co,60,61 Ni,62 and Rh.63,64 Single Rh (sub)carbonyl sites have been
claimed to be active for CO insertion.25 Moreover, the mobility of metals on oxidic supports
and therefore sintering phenomena have recently been related to the formation of carbonyl
species on Co based catalysts.65,66 However, the role of (sub)carbonyls in syngas chemistry
is still to be investigated and subject of very recent research.
2Stateoftheart
8
2.3 Syngas to ethanol (STE)
The direct conversion of syngas to ethanol has first been reported in a patent filed by Ellgen
and Bhasin67 in 1976 for Union Carbide and further described by a number of articles and
patents shortly after.68–70 Luk etal. recently published an extensive review on the status and
prospects of higher alcohol synthesis from syngas.17 Earlier, Spivey and Egbebi71 and
Subramani and Gangwal11 contributed reviews on the characteristics of the syngas to
ethanol reaction. Ao etal. provided an overview of the currently discussed active sites for
STE.72
Catalysts for STE
A wide variety of catalysts has already been tested in the STE reaction. Four catalyst families
emerged that were found active for ethanol formation:
i) Rh-based catalysts
ii) Mo-based catalysts
iii) modified Fischer-Tropsch catalysts (mainly Co-based)
iv) modified methanol catalysts (mostly Cu-based)
The review by Luk etal.17 provides extensive information about the different catalyst
categories. A lot of research has been dedicated to a possibility to circumvent the use of Rh
due to cost and availability considerations. However, Rh-based materials still show the best
performance for ethanol production. Also, supported Rh catalysts are a particularly
attractive system for scientific evaluation since the product selectivities can be tuned
tremendously by choice of support73,74 and metal composition.72 Detailed investigation
offers the chance to learn about syngas chemistry on supported catalysts in general.
Therefore, this study focusses on silica-supported Rh catalysts.
Rh occupies an interesting position in the periodic table between CO dissociating and CO
non-dissociating elements (Figure2.3 in section 2.2). This position is often referred to as
reason for the intrinsic selectivity towards C2 oxygenates such as ethanol and acetaldehyde.
However, C2 oxygenates selectivities are limited. Depending on the choice of support,
promoters, and reaction conditions, Rh catalyzes the formation of methane, hydrocarbons,
alcohols, and other oxygenates in parallel.70,71 The formation of C2 oxygenates requires at
least two functionalities: CO dissociation and CO insertion. Density functional theory (DFT)
studies were used to assign different crystal facets to different reactivity. Results of Yang et
al. indicated that Rh(211) is around 6 times more active than Rh(111) but more selective
towards methane, suggesting a structure sensitivity requiring Rh(111) for high C2
oxygenate (mainly acetaldehyde) selectivities.75
2.3Syngastoethanol(STE)
9
Also, more complex structures were discussed as potential active sites for ethanol
formation. Adjacent Rh0-Rhn+ species have been made responsible for the bifunctionality.76
Metallic Rh0 has been assigned to CO dissociation and hydrogenation steps, while Rh+ might
be responsible for CO insertion.77,78 The relative proportion of Rh+/Rh0 would thereby be
crucial for product selectivities.79 Rh+ sites might be either formed by formation of Rh
(sub)carbonyl species under CO containing atmosphere or molecular adsorption of CO on
isolated Rh sites. Generally, Rh+ sites are described as Rh+(CO)2 species.64,80 However, the
presence of these gem-dicarbonyl species or other subcarbonyl species at reaction
conditions and their role as active sites are discussed controversially and are subject to very
recent research.81,82
For a more fundamental understanding of the underlying concepts, a detailed evaluation of
the simplest Rh/SiO2 system is crucial although it might not yield the highest performance
marks. The system has been investigated extensively in literature but substantially varying
results indicate how difficult the experimental acquisition of catalytic data can be. In this
case, catalytic evaluation is particularly challenging due to reaction conditions with
relatively high pressures for lab-scale setups and a broad product spectrum that must be
precisely analyzed regarding quality and quantity. A summary of experimental data from
literature shows that the detailed measurement of catalytic data for Rh/SiO2 in the own
setup is inevitable in order to compare results with more complex systems (Table2.1). For
example, the obtained ethanol selectivity varies between 0-16 % for similar catalysts and
reaction conditions.
Table2.1ExemplaryresultsforCOhydrogenationoverRh/SiO2catalysts.AdoptedfromthemasterthesisofF.
SchusterexecutedatBasCatin2018.
EntryCatalyst
/wt‐%Rha
T
/K
p
/bar
GHSV
/h‐1H2/COXCO
/%
SMethane
/%
SEthanol
/%
SC2oxya
/%
YC2oxyb
/%Ref.
12.5Rh/SiO257369‐ 13.1521743.11.370
21.5Rh/SiO2543301500020.80648.514.80.1683
31.5Rh/SiO254330400025.0708.214.30.7284
42.0Rh/SiO2543103300c22.1604.016.90.3685
52.0Rh/SiO2543208000c10.755100086
62.0Rh/SiO254824‐ 2104316292.987
73.0Rh/SiO255820‐ 24.33000088
85.5Rh/SiO252340‐ 11.6433.520.50.3389
aSelectivitytowardsC2oxygenates(ethanol,acetaldehyde,aceticacid)ifprovided
bYieldtowardsC2oxygenates(ethanol,acetaldehyde,aceticacid)ifprovided
cSpacevelocityincm3gcat‐1h‐1
2Stateoftheart
10
Although Rh is the most potent monometallic catalyst for the STE reaction, ethanol yields
are still far from industrial requirements. In particular, the achieved CO consumption rates
and ethanol selectivities are insufficient.17,72 However, in extensive studies a wide range of
metallic and oxidic promoters were tested. Mn, Fe and Li are among the most frequently
used metals for the improvement of catalytic performance (Table2.2).
Table2.2Top10promotedand/orsupportedRh‐basedcatalysts.ReproducedfromLuketal.17withpermission
fromTheRoyalSocietyofChemistry.
EntryCatalystaT
/K
p
/bar
GHSVb
/h‐1H2/COXCO
/%
SC2+oxy
/%
SHC
/%
YC2+oxy
/%Ref.
1RhMn/SiO255354170023954422190
2RhMnLi/Fe/SiO25733010000b22864341891
3RhMnLiFe/SiO25733010000b22858401692
4RhMn/SiO255820‐ 23643521688
5Rh/Ce0.8Zr0.2O2548242400b22744381287
6Rh/MnFe‐OMC573501200022646381293
7RhMnLi/SiO25733010000b21954421091
8RhMn/MSN543306600b22047471094
9RhMnLiFe/CMK‐959330120002146522995
10RhMn/SiO25433040002174652884
aCMK:cubic‐orderedmesoporouscarbon,OMC:orderedmesoporouscarbon,MSN:mesoporoussilicananoparticles.
bSpacevelocityincm3gcat‐1h‐1
Mn addition has shown high potential to increase C2 oxygenate yields.75,83,84,88,92,96–103 The
role of Mn has thereby been described as promoting CO dissociation104 and promoting CO
insertion.55,97 Experimental evidence has been contributed by Mao etal.that tilted adsorbed
CO species at Rh-Mn oxide interfaces cause increased CO dissociation rates that lead to
enhanced activity and selectivity.83 However, other studies contradict this hypothesis and
doubt the relevance of tilted adsorbed CO species for reactivity.105 Other studies suggest
that the contact with Mn oxide stabilizes Rh carbonyls and therefore ultimately promotes
CO insertion.96,103 As another hypothesis it was reported that the presence of Mn kinetically
suppresses methane formation by increasing the activation energy for methane formation
while it does not affect ethanol formation.84 The optimal molar Rh:Mn ratio is given in a
range of 0.5-1.87 over different supports.83,97,103,106
Besides Mn, Fe emerged as one of the most promising promoters for Rh catalysts on various
supports.107–111 The optimal Fe loading is dependent on the support and presence of other
2.3Syngastoethanol(STE)
11
metals. For RhFe/Al2O3 ethanol selectivity constantly increased for higher Fe loadings up to
10 wt-%.71 Other studies describe much smaller loadings as optimal.109–111 Some studies
assign increased ethanol selectivities and suppressed methane formation to RhFe alloy
formation under reaction conditions.86,112 Other studies suggest that the contact of Rh to Fe
oxide species are important to stabilize certain CO* species for C2 oxygenate
formation.108,109 The presence of Fe oxide was here proposed to support Rh+ formation and
therefore CO insertion. In contrast, Fe is also reported to work as reservoir for spill-over
hydrogen. In this case, its presence would increase hydrogenation rates and lead to CO
dissociation and hydrocarbon formation.113 The investigated systems are very different
from each other and therefore it is difficult to clearly assign structure-reactivity
relationships. Moreover, little information on their time on stream behavior, reaction
orders or consideration on the impact of Fe on the reaction network could be found.
Reaction and thermodynamics
The direct conversion of syngas to ethanol is highly exothermic. The reaction enthalpy ΔH
ranges between -260 and -270 kJ mol-1 in a temperature range of 100–400 °C.11
2 𝐶𝑂
4 𝐻
⎯
⎯
⎯
𝐶
𝐻
𝑂𝐻
𝐻
𝑂
∆𝐻° 254 𝑘𝐽 𝑚𝑜𝑙
∆𝐺° 123 𝑘𝐽 𝑚𝑜𝑙
(1)
The Gibbs free energy change ΔG shows that the reaction becomes unfavorable above 280 °C
and therefore would require elevated pressures to increase ethanol yield. Thermodynamic
analysis of the reaction under the assumption of a CO/H2 ratio of 1:3 and 54 bar total
pressure (standard conditions for experiments in this study) reveals decreasing ethanol and
water concentrations and increasing reactant concentrations with increasing temperature
(Figure2.4A).71 The results suggest that the reaction should be performed well below
350 °C under these conditions.
Methane formation is the main side reaction for the STE reaction. It is the
thermodynamically most significant product.114
𝐶𝑂
3 𝐻
⎯
⎯
⎯
𝐶𝐻
𝐻
𝑂
∆𝐻° 206 𝑘𝐽 𝑚𝑜𝑙
∆𝐺° 142 𝑘𝐽 𝑚𝑜𝑙
(2)
If methane is allowed as a product, ethanol concentration in the equilibrium composition is
negligible (Figure2.4B). Methane formation therefore must be kinetically limited in order
to allow significant ethanol formation.71
2Stateoftheart
12
Figure2.4EquilibriumcompositionandconcentrationprofilesofthehydrogenationofCOtoethanol(A,H
2
/CO=3,
54bar)andwithmethaneformationallowedunderthesameconditions(B).
Reaction mechanisms
A major part of research efforts on the reaction mechanism towards ethanol formation from
syngas over Rh has been contributed by theoretical studies in the last decade. In particular,
Rh single crystal surfaces were investigated including Rh(100),115 Rh(111),116 Rh(211),117–
119 and Rh(533).120 In addition, clusters such as Rh6,121 Rh4/Al2O3,122 and Rh/Mn/SiO299 have
been considered. All proposed mechanisms are composed of hydrogenation, C-O bond
breaking, and CO insertion steps. The considered products are usually methane, methanol,
and ethanol.
Choi and Liu found a high energy barrier for direct CO dissociation over Rh(111).116 The H-
assisted CO dissociation pathway, which requires hydrogenation of the molecularly
adsorbed CO* first, was therefore considered more plausible. The C-O bond cleavage step is
followed by insertion of molecularly adsorbed CO* under formation of a CH3CO* fragment
and subsequent hydrogenation to ethanol. Initial hydrogenation of CO* was found to be the
rate-limiting step. Kapur etal.119 and Wang etal.118 considered similar pathways as
favorable on a Rh(211) surface. However, no clear consensus on the degree of
hydrogenation of each intermediate was found. Filot etal. investigated the formation of a
larger variety of products including CO2, H2O, formaldehyde, and C2 hydrocarbons.117 The
preferred reaction products were found to be formaldehyde at temperatures below 330 °C,
ethanol at around 430 °C and methane at above 630 °C. However, no experimental data
supporting major formation of formaldehyde at these conditions could be found. Shetty et
al. investigated a Rh6 cluster instead of a single crystal surface.121 Similar to the results for
Rh(111), the rate-limiting step was CO* activation by initial hydrogenation. The authors
proposed a crucial dependence of product distribution on CO* and H* coverages. Methanol
and methane were preferred at H-rich conditions. High CO* coverages increased the barrier
for CO* hydrogenation and more acetaldehyde and acetic acid was predicted. Several
0 200 400 600 800
0.0
0.2
0.4
0.6
0.8
Mole fraction
Temperature / °C
Ethanol CO
H
2
Methane and H
2
O
B
0 200 400 600 800
0.0
0.2
0.4
0.6
0.8
Mole fraction
Temperature / °C
Ethanol
and H
2
O
CO
H
2
A
2.3Syngastoethanol(STE)
13
experimental studies support these calculations.83,85 This finding suggests that catalyst
structures other than single crystal surfaces and the structural impact of reaction conditions
must be closely investigated. However, it must be noted that all theoretical studies reported
so far had to rely on the assumption of Rh being present in a metallic state, vacuum
conditions, and the absence of reaction-induced surface reconstructions. It is thus not clear
how closely the obtained results can resemble the state of the catalytic surface under the
practically applied high pressure conditions, in particular high partial pressures of CO (see
Figure2.3). Moreover, selective STE catalysts feature one or several doping elements that
heavily contribute to control over the product selectivity. Such effects have been
incorporated into DFT models rarely and only under assumption of alloy formation but not
structural modification which is likely the case especially for oxidic dopant elements.
Apart from theoretical studies, experimental data have been reported in several kinetic
studies.83,98,123–125 Common reaction conditions were 200-310 °C and 1-30 bar using
different Rh-based materials. The typical total time on stream for such experiments are
several hours to a few days. Data on the time on stream behavior are very scarce and could
only be found for small time scales of some minutes to few hours reporting increasing
oxygenate and decreasing paraffin rates.125,126 However, considerations on the origin of this
behavior were not expressed.
The considered product spectrum is usually simplified towards the main products methane,
ethane, acetaldehyde, methanol, and acetic acid. Minor products such as acetates or higher
aldehydes and hydrocarbons are usually not reported. Also, there is often no discrimination
between olefins and paraffins85 or products are combined into groups such as “C2+
hydrocarbons” or “other oxygenates”.99,123 This simplification makes detailed comparison
of measured and reported catalytic results difficult and prevents a more detailed evaluation
of the very complex reaction network.
In the cases where the impact of reactant partial pressures was investigated, product
formation rates were reported to be highly dependent on CO and H2 partial pressures. For
most materials, activity increased with increasing H2 partial pressure while CO partial
pressure had detrimental effect.83,84,123 The impact of partial pressures on product
selectivities were different depending on the material. From Langmuir-Hinshelwood
expressions, rate-limiting steps were derived, again suggesting H-assisted CO* activation to
be an important rate-limiting step.123 However, the impact of partial pressures is usually
reported in a narrow range excluding data at e.g. very low CO pressures.
The formation pathway of ethanol has been discussed controversially. While cofeed
experiments with acetaldehyde suggested the consecutive hydrogenation of acetaldehyde
as the most plausible formation pathway.127 Other studies proposed separate formation
pathways for aldehyde and alcohol mostly based on isotope tracing experiments at very low
2Stateoftheart
14
pressures.78,128,129 Other cofeed experiments probing specific reaction pathways have not
been found in published studies so far.
Theoretical as well as experimental studies strongly focus on formation pathways towards
the target product ethanol and methane as major side product. Formation pathways
towards the large variety of other products such as higher olefins, paraffins, aldehydes, and
acids are usually not addressed. However, to fully understand syngas chemistry over
different materials, all reaction products must be considered to fully portray the complexity
of the reaction network including the formation of longer carbon chains as important step.
In STE research, a CHx coupling mechanism is commonly assumed the formation of products
with higher carbon numbers (Figure2.5).83,130 However, little theoretical or experimental
data have been provided to support this hypothesis.
Figure2.5Pathwaysfortheformationofmethane,methanolandC2+oxygenatesfromCOhydrogenationoverRh–
MnOx/SiO2catalysts.ReprintedfromMaoetal.83withpermissionfromElsevier.
In Fischer-Tropsch research, two plausible reaction pathways towards higher carbon
chains are mainly considered.131 The carbide mechanism involves coupling CHx* fragments
similar to what is usually assumed in STE reaction mechanisms (Figure2.6A).34 Those
fragments are generated by CO* dissociation and subsequent hydrogenation. Different
degrees of hydrogenation are thereby assumed, e.g. CH132 and CH2.133–135 The coupling of
those fragments leads to chain elongation and formation of higher hydrocarbons. CO
insertion would directly lead to chain termination under formation of a higher oxygenate.
Another mechanism sometimes assumed for Co catalysts is the CO insertion mechanism
(Figure2.6B).30,136 In contrast to the CHx coupling mechanism, CO* or CHO* insertion does
not lead to chain termination but is part of the chain elongation process. The insertion step
is followed by C-O bond cleavage and forms the next higher hydrocarbon fragment.
2.3Syngastoethanol(STE)
15
Figure2.6CH
x
coupling/carbide
34
(A)andCOinsertion
136
(B)mechanismsforFischer‐TropschSynthesis,where
dashedframesindicatechaininitiation(red),chainpropagation(green)andchaintermination(blue).Reproduced
fromLuketal.
17
withpermissionfromTheRoyalSocietyofChemistry.
The existing literature on STE provides many insights on specific questions regarding
reactivity of differently promoted Rh-based catalysts and the underlying reaction network.
However, simplification and missing data prevented the understanding of the full
complexity of this reaction so far. This complexity also applies to structural aspects of the
used catalysts which are rarely addressed in combination with extended catalytic testing.
Deduced research needs
Although major research has already been dedicated to the STE reaction with different
catalyst systems, many questions remain unanswered. For example, the development of
catalytic behavior over long time on stream has not been evaluated in detail so far.17
Typically, catalysts are only studied for short time on stream, although in this study it will
be shown that around 100 h time on stream are required for initial catalyst formation in
order to obtain reliable and reproducible catalytic results. The detailed characterization of
catalysts before and after reaction is often missing and therefore a discussion of the impact
of reaction conditions – exposition to high CO pressures in particular – on structural
properties has not been provided so far.
The impact of reactant partial pressures has been investigated only in a very small range
and the systematic analysis of the reaction network by contact time variation and cofeed
experiments has not been performed so far. Considerations on the reaction network are
mostly limited to the formation pathways of ethanol and methane as many studies appear
to not assess, quantify, or report the complete spectrum of formed reaction products.
Consequently, no generalized reaction network exists that describes the formation of all
products and the influence of common promoters on the reaction mechanism at industrially
2Stateoftheart
16
relevant conditions. To clearly define the impact of catalyst composition on the reaction
network, more precise and consistent catalytic data are necessary.
This study contributes mentioned missing data as well as a systematic evaluation of
catalytic data and structural features of the catalysts. Based on the detailed interpretation
of these results, a comprehensive picture of the STE reaction is provided contributing a
generalized reaction network with considerations on the complex interplay of reaction
conditions, catalyst structure, and thereby favored surface reactions leading to the
characteristic performance features of differently promoted Rh/SiO2 catalysts.
17
3 Experimental details
Experimental details, methods and procedures used for catalytic testing and catalyst
characterization are summarized in this chapter.
3.1 Catalyst synthesis
The support material “Davisil Grade 636” is a high-purity mesoporous silica with a specific
surface area of ~480 m2 g-1, a pore size of 60 Å and a particle size of 200-500 µm according
to Merck.137 The support was calcined without prior sieving at 550 °C for 6 h in a muffle
furnace.
Impregnation with metal nitrates follows an incipient wetness impregnation approach
based on the catalyst synthesis procedure of Hu et al.90 The specific amount of liquid
necessary to fully wet the support was previously determined experimentally. The
impregnation solution was prepared from an aqueous Rh(NO3)2 solution. If applicable, Mn
and Fe nitrates were dissolved in the Rh(NO3)2 solution under continuous stirring. The
solution was then diluted with water until the respective amount of liquid for the
impregnation process was reached. The metal nitrate solution was then mixed with the
calcined SiO2 support and the mixture was thoroughly homogenized with a spatula for at
least 5 min. Drying and calcination was performed in four steps under constant air flow
(Table3.1). After calcination, the pre-catalysts were sieved to receive the target particle
size for testing of 250-315 µm.
Table3.1Dryingandcalcinationprotocolforcatalystsynthesis
StepRamp/Kmin‐1Targettemperature/°CHoldingtime/min
158030
2510030
35120180
45350180
Four catalysts were prepared according to the described procedure. An overview of their
respective sample numbers after calcination, H2 treatment and different catalytic runs can
be found in TableA1 in Appendix A. Throughout this work the catalysts will be referred to
according to the simplified expressions “RhMn/SiO2”, “RhFe/SiO2”, and “RhMnFe/SiO2” due
3Experimentaldetails
18
to readability. This denotation does not contain information about the oxidation state of the
metals or their potential presence as metal oxides or alloys.
Table3.2Targetweight‐loadingsoftheas‐preparedcatalysts.
Simplified
catalystname
Catalystloading/wt‐%
RhMnFe
Rh/SiO22.8‐ ‐
RhMn/SiO22.81.5‐
RhFe/SiO22.8‐ 0.5
RhMnFe/SiO22.81.50.5
3.2 Gases for catalytic testing
All gases used in this study are listed below (Table3.3). With exception of a carbonyl trap
in the CO dosing line, no further purification or treatment was performed.
Table3.3Listofgasesusedforcatalysttesting.
GasMolecular
formular
Grade/
specification
Pressure/
bar
Retailer
CarbonmonoxideCO4.7200AirLiquide
HydrogenH25.0300AirLiquide
NitrogenN25.0300AirLiquide
ArgonAr5.0300AirLiquide
EthylenemixC2H4
H2
1000ppm
balance
200Linde
PropylenemixC3H6
H2
200ppm
balance
200Linde
AcetaldehydemixCH3CHO
H2
2045ppm
balance
200AirLiquide
CarbondioxidemixCO2
H2
balance
80%
200AirLiquide
3.3Catalytictestsetup“Dino”
19
3.3 Catalytic test setup “Dino”
The 4-channel parallel testing unit was planned and built by ILS – Integrated Lab Solutions.
The setup can be divided in three segments – feed section, reaction section, and separation
and analytical section (Figure3.1). In this section, each segment will be described in detail.
Figure3.1Frontviewofthe4‐channelparalleltestingunit“Dino”forCOhydrogenationreactions.
Feedsection
The feed section contains a gaseous and liquid feed section. Since the liquid feed option has
not been used for the tests presented in this study it will not be discussed in more detail.
The gases are supplied from gas cylinders stored in ventilated cabinets protected against
explosion. The inlet pressure supplied for the test setup is adjusted to 100 bar by pressure
reducers for CO, H2, Ar, high pressure N2 and an additional gas line for mixed gases. Low
pressure N2 is supplied by the house line with maximum 10 bar. For each gas type a dosing
unit is installed. Each dosing unit consists of a ball valve, a back pressure valve, a particle
filter, a pneumatic safety valve (with exception of Ar), a manometer to show the inlet
pressure, a ball valve and a mass flow controller (MFC) in this order. The used MFCs are of
the type EL-FLOW by Bronkhorst. The MFCs are calibrated for an inlet pressure of 95 bar
and 90 bar outlet pressure in the gas mixing section. They are chosen to realize GHSVs from
1500 – 7500 h-1 for a standard catalyst volume of 1 ml.
3Experimentaldetails
20
Pressurized CO in contact with steel can form volatile metal carbonyls, e.g. Ni or Fe
carbonyls. The CO feed line is equipped with a carbonyl trap to remove carbonyl impurities.
It consists of a U-shaped 1/2" stainless steel tube filled with Al2O3 and heated to 170 °C by
a heating sleeve. At this temperature Ni and Fe carbonyls decompose at the Al2O3 surface.
Behind the MFCs, all feed lines are combined in the mixing section. The pressure in the gas
mixing section is adjusted to 85-90 bar by a back pressure regulator (Equilibar, ZF Series,
SS 316L Body). The dome pressure is provided by the high pressure N2 supply with a
manually operated self-vented pressure reducing valve. The provided dome pressure and
the actual pressure of the mixing section can be controlled by two manometers. The mixed
gas feed is then equally distributed to four reactors and one bypass line. The equal
distribution is realized by five restriction capillaries.
Reactionsection
The four parallel reactors are made from ¼” stainless steel tubes with a length of 645 mm
which are connected to the reactor head with a VCR screw connection. The reactor head
provides connections for the gaseous feed, the liquid feed and the thermowell. The lower
end of the reactor is closed with a frit and sealed in a stainless-steel sleeve with a Kalrez O-
ring. The sleeves are located in a downstream oven usually heated to 180 °C.
The reactors are heated by a heating block made from an aluminum composite material.
The maximum temperature is 500 °C with an isothermal zone of 12 cm. The reactor tubes
have in inner diameter of 9.40 mm. With the thermowells inserted, the resulting effective
inner diameter is 6.22 mm. The thermowells enable the insertion of temperature sensors
into the reactor. The used sensors (Thermo-Sensor, Type K, 3 mm x 300 mm) offer three
thermocouples to measure the temperature along the catalyst bed in each reactor.
The pressure inside the reactors is controlled by five back pressure regulators equipped
with stainless steel membranes (Equilibar, ZF Series, SS 316L Body). The four back pressure
regulators for the reactors are located inside the downstream oven while the one for the
bypass line is located outside the oven. The N2 dome pressure is adjusted by an electronic
pressure regulator (Bronkhorst, El-Press) which can be controlled from the process control
system. The pressure regulator is connected to the five back pressure regulators by another
capillary system. The maximum pressure is 80 bar. The experiments for this study were
performed at a constant pressure of 54 bar.
Reaction conditions such as feed composition, total flow, reaction temperature and reaction
pressure cannot be adjusted individually but are always the same for all reactors.
3.3Catalytictestsetup“Dino”
21
Separationandanalyticalsection
The back pressure regulators are protected from contamination with liquid compounds by
coalescence filters. Inside those filters a 0.5 µm mesh filter element would lead to removal
of all compounds that condense at 180 °C inside the downstream oven. In the studies
discussed in this work, no or very small condensation was observed. To further protect the
analytic section, the reactor effluent is diluted with N2 downstream of the back pressure
regulators on the ambient pressure side.
A multiposition valve (6-channel common outlet) installed inside the downstream oven is
used for selecting one of the four reactors or the bypass line for sampling. All unselected
reactor effluents are purged into washing bottles filled with water outside the downstream
oven. All compounds that would condense at room temperature are separated. The
remaining gas flows are combined and led to the exhaust system.
Online monitoring of the gas composition is performed by gas chromatography. The gas
chromatograph (Agilent 7890B) is equipped with a thermal conductivity detector (TCD)
and a flame ionization detector (FID) with a Polyarc reactor. The TCD is used to quantify
permanent gases such as H2, Ar and N2. All carbon containing gases (CO, CO2, hydrocarbons,
oxygenates) are quantified by the FID. The measurement of one chromatogram took
~38 min and one round of chromatograms for all reactors and the bypass line took ~3.5 h.
A typical chromatogram from the FID is provided in FigureA1 in Appendix A. The
additionally installed Polyarc system facilitates the measurement of the complex product
spectrum including products that are difficult to calibrate such as acetic acid, acetaldehyde
or acetates. After separation in the GC columns, all carbon containing compounds are
oxidized to CO2 in the first catalytic reactor of the Polyarc system. In a second step, CO2 is
fully hydrogenated to methane. Therefore, only the response factor of methane must be
calibrated. In addition, compounds that usually cannot be detected by an FID, such as CO,
CO2, formaldehyde and formic acid become detectable.
To cut off the permanent gases of the full sample stream, a Hayesep Q (Hayesep Q 80/100,
1 m x 1/16”, SST) and a PoraBOND Q (PoraBOND Q, 50 m x 0.53 m x 10 µm) column was
used. The permanent gases were then separated in a mole sieve column (HP-Plot 5A, 30 m
x 0.53 mm x 50 µm) and detected in the TCD. C1 to C6 hydrocarbons and oxygenates were
separated in a HP-Plot Q column (HP-Plot Q, 30 m x 0.53 m x 40 µm), each compound
converted to methane in the Polyarc reactor and finally detected in the FID. A schematic
diagram of the GC setup is provided in FigureA2 in Appendix A.
3Experimentaldetails
22
Safetyconcept
Working with toxic and flammable gases requires a potent safety concept to reduce risks
and ensure a safe working environment. The setup is enclosed by a housing that is
connected to a central exhaust system. Several sensors are applied inside the housing: two
H2 sensors, one CO sensor and a fire detector are installed. Additionally, a flow sensor is
monitoring the ventilation system. In case of an alarm of one of the sensors, a power supply
failure or a significant deviation of the process values from their setpoints, an emergency
shut down procedure is activated. Depending on the cause, two different safe states are
approached. In the case of a ventilation failure, all gas flows are stopped, pressures are
maintained, and all ovens and furnaces are shut off. In all other emergency cases, pressure
is released from reactors and the system is purged with N2. All reactive gases and heat
sources are shut off. To prevent overpressure in the reactors during operation, safety valves
are installed before the reactor heads with a critical pressure of 100 bar.
3.4 Experimental procedures for catalytic testing
For this study, four catalytic runs were conducted. Each run was performed with all four
catalysts in parallel and was prepared by the same procedures to fill, install and pressurize
the reactors as described in this section.
Reactorfilling
The reactor tubes were connected to the reactor heads via VCR connectors. In a separate
pressure testing unit, the reactors were pressurized with up to 100 bar N2 and checked for
leakages. If no leakage was detected, the reactors were placed upside down to be filled from
the end of the reactor tube. Steatite with a particle size of 400-700 µm (CeramTec) was filled
into the reactor. The filling level was adjusted to the desired position inside the reactor
tubes in order to place the catalyst bed in the middle of the isothermal zone. The steatite
bed was then compressed by knocking against the tube from all sides.
The catalyst sample was diluted with a fraction of the support material SiO2 that was slightly
smaller than the catalyst particles (100-200 µm vs. 250-315 µm) to give a total volume of
2 ml for each sample. The mixture was then filled into the prepared reactors. In general, a
deionizer was used to ensure that all catalyst particles fall into the reactor despite the
electrostatic charge of the particles. The catalyst bed was again compressed by careful
knocking to avoid catalyst particles falling into the steatite bed due to their smaller size. The
reactors were then filled up with steatite to the top and closed with a stainless-steel frit.
3.4Experimentalproceduresforcatalytictesting
23
For this study, one specific reactor tube was used for each catalyst sample for all four
catalytic runs to avoid contamination. Each reactor tube was placed into the same reactor
slot for each catalytic run.
Start‐upprocedureforcatalytictests
The filled reactors were placed into the respective slot in the furnace of the testing setup
and immediately connected to a N2 flow. The furnace is generally heated to 100 °C before
the start of a reaction. At this point, the reactor sleeves inside the downstream oven were
checked for leakages for the first time with leak detector spray. If a leak was found, the
Kalrez O-ring inside the sleeves had to be replaced. If no leakage was found, the reactors
were pressurized to 20 bar with N2. The pressure increase inside the reactors was followed
and checked for even and steady behavior. At 20 bar, the next check for leakages was
performed according to the first one. If a leakage was detected, the pressure had to be
released, the O-ring exchanged, and the pressurization process was started over. If the test
passed at 20 bar, the pressure was increased to the target pressure of 54 bar and the same
procedure was followed. If no leakages were found and pressures were the same for all four
reactors, the catalytic run was started.
Catalystsievingandreactorcleaning
After reaction and cooling down to 100 °C, the reactors were taken out of the furnace. The
frit was removed, and the reactor was emptied carefully inside a fume hood. In a first step,
the steatite particles were removed from the catalyst/SiO2 mixture using a 315 µm sieve.
Then, the spent catalyst sample was separated from the SiO2 dilution with a 200 µm sieve,
labelled and stored for subsequent characterization. All sieving steps and transfers from
one container into another were performed using a deionizer.
The reactor tubes were removed from the reactor heads. The thermo wells which stay
connected to the head were cleaned thoroughly with water and ethanol. The reactor tubes
were first cleaned with water, then placed inside an acidic solution (1 % HCl, 10 % HNO3 in
water) for 30 min. Afterwards, the reactors were rinsed with water and scrubbed with a
plastic brush on a cleaning rod. A clean cloth soaked with ethanol was used to test if the
reactors were clean in addition to optical control. The cleaned reactors were left to dry
before being refilled.
3Experimentaldetails
24
3.5 Experimental design of catalytic testing
Four catalytic runs were performed for this study: a kinetic study (DinoRun26), a thermal
stability test (DinoRun27), a test including cofeeds and dropout experiments (DinoRun29)
and a short CO dropout run (DinoRun31). Reaction conditions were controlled fully
automized with the native process control system. Each run was started by the same in‐situ
H2 treatment and an equilibration phase.
In-situ H
2
treatment
After successful leak checks of the reactors at 54 bar under N2, the H2 treatment was started
according to the following procedure:
i) feed changed to 5 % H2 and 10 % Ar in N2 (100 ml min-1 total flow per reactor)
ii) temperature increased to 265 °C in ~30 min
iii) holding time of 1 h
iv) temperature decreased to 260 °C while reactors purged with 10 % Ar in N2
v) when 260 °C (starting reaction temperature) was reached, reaction was started
Initial equilibration time at standard conditions
For each catalytic run, equilibration of the catalytic behavior was allowed for at least 130 h.
During that time standard conditions were applied, and chromatograms were taken
continuously. Standard conditions are defined as a reaction temperature of 260 °C, feed
composition of H2:CO:Ar:N2 = 60:20:10:10 with a total flow of 41.7 ml min-1 per reactor and
a total pressure of 54 bar. The total pressure was held constant throughout the catalytic
runs for this study.
Parameter field test (DinoRun26)
For parameter field testing, all catalysts should be tested at a similar CO conversion level.
The amount necessary of each catalyst sample to yield ~5 % CO conversion has been
estimated from preliminary experiments (Figure3.2). The catalysts were diluted with SiO2
to give the same catalyst bed length.
3.5Experimentaldesignofcatalytictesting
25
Figure3.2Reactorfillingandreactionconditionsasappliedintheparameterfieldtest(DinoRun26).Repeatedly
testedstandardconditions(260°C,H2:CO:N2:Ar=60:20:10:10,41.6mlmin‐1totalflowperreactor)aremarkedby
greyareas.
After the initial equilibration phase, temperature was varied in the chronological order
260 °C – 250 °C – 243 °C – 250 °C – 260 °C with a feed composition of H2:CO:Ar:N2
60:20:10:10. Then, the reactant partial pressures were varied in a 3x6 matrix at constant
temperature of 260 °C and a total flow of 41.6 ml min-1 (Figure3.3). Partial pressure of CO
was varied from 1.6 bar to 10.8 bar which represent a molar CO content of 2.5 % to 20 %.
Partial pressure of H2 was varied from 16.2 bar to 32.4 bar which represents 30 % to 60 %.
Therefore, H2/CO ratios of 1.5 to 24 were realized. With varying CO and H2 contents, N2 was
used to balance total pressure and total flow. Ar as internal standard was kept constant
throughout the study. Finally, the total flow was varied in six steps from 8.3 ml min-1 to
58.4 ml min-1.
Figure3.3Reactionconditionsappliedduringpartialpressureandcontacttimevariation.Totalflowmeansthetotal
volumeflowofthereactionmixtureperreactor.Standardconditionsaremarkedinred.
H
2
, CO, Ar, N
2
Rh
500
mg
RhMn
100
mg
RhFe
50
mg
RhMnFe
165
mg
0 50 100 150 200 250 300 350 400 450 500 550 600
5
10
15
20
25
30
35
40
0
Time on stream / h
Contact time variation
Partial pressure variation
Volume flow / ml min
-1
Equilibration
Temperature variation
243 °C - 260 °C
Temperature
CO
230
235
240
245
250
255
260
265
270
Temperature / °C
H
2
3Experimentaldetails
26
Thermal stability test (DinoRun27)
As the results of DinoRun26 revealed that the activity of the RhFe/SiO2 catalyst has been
overestimated, its catalyst amount has been adjusted. All other catalysts were filled
according to DinoRun26 (Figure3.4).
Figure3.4Reactorfillingandreactionconditionsasappliedinthethermalstabilitytest(DinoRun27).Repeatedly
testedstandardconditions(260°C,H
2
:CO:N
2
:Ar=60:20:10:10,41.6mlmin
‐1
totalflowperreactor)aremarkedby
greyareas.
After the initial equilibration phase, temperatures were varied first between 270 °C and
243 °C for activation energy evaluation. Then, temperatures were increased up to 320 °C in
10 °C steps. Volume flow and partial pressures were kept constant.
Cofeed and dropout experiments (DinoRun29)
Reactors were filled according to DinoRun27. After the initial equilibration phase, H2
dropout experiments were performed. Dropout experiments were realized by replacing the
flow of one reactant by additional N2 to keep the total flow, GHSV, and concentrations of
other compounds constant. The dropout conditions were applied for at least 24 h before
standard reaction conditions were applied again. The first H2 dropout was performed at
265 °C as applied during the in‐situ H2 treatment before the start of reaction (Figure3.5).
The second H2 dropout was performed at standard temperature of 260 °C. After constant
catalytic behavior was measured at standard conditions, a CO dropout experiment was
performed accordingly.
H
2
, CO, Ar, N
2
Rh
500
mg
RhMn
100
mg
RhFe
150
mg
RhMnFe
165
mg
0 50 100 150 200 250 300 350 400
5
10
15
20
25
30
35
40
0
Time on stream / h
Volume flow / ml min-1
Equilibration
Temperature variation
243 °C - 320 °C
Temperature
CO
H
2
220
240
260
280
300
320
340
Temperature / °C
3.5Experimentaldesignofcatalytictesting
27
Figure3.5Reactorfillingandreactionconditionsasappliedduringthecofeedanddropoutexperiments
(DinoRun29).Repeatedlytestedstandardconditions(260°C,H2:CO:N2:Ar=60:20:10:10,41.6mlmin‐1totalflowper
reactor)aremarkedbygreyareas.
Cofeeding of potential intermediates was realized by usage of gas mixtures with a defined
concentration of the intermediate compound in H2 from extra gas cylinders (compare
section 3.2). The concentrations of ethylene, propylene and acetaldehyde were chosen to
enable cofeed concentrations in the range of the concentrations observed during reaction
at standard conditions. The concentration of CO2 was higher. For the introduction of the
cofeed compound into the gas stream, the H2 flow was fully or partially replaced by the
mixture keeping the H2 flow constant. The realized cofeed concentrations for each
compound are given in Table3.4.
Table3.4RealizedcofeedconcentrationsinDinoRun29.
Cofeedcompound
Concentration/vol‐%
LowHigh
Ethylene0.0300.057
Propylene0.00410.0078
Acetaldehyde0.070.14
CO21.773.48
H
2
, CO, Ar, N
2
Rh
500
mg
RhMn
100
mg
RhFe
150
mg
RhMnFe
165
mg
0 100 200 300 400 500 600 700 800 900 1000
5
10
15
20
25
30
35
40
0
Time on stream / h
Ethylene
cofeed
Volume flow / ml min
-1
Equilibration
H2
dropouts
CO dropout Acetaldehyde cofeed CO2 cofeed
Propylene
cofeed
Mix
230
235
240
245
250
255
260
265
270
Temperature / °C
H2
CO
Temperature
3Experimentaldetails
28
Preparation of samples after CO dropout for characterization (DinoRun31)
The fourth catalytic run was performed in order to prepare additional catalyst samples for
characterization. Reactors were filled according to DinoRun27 and DinoRun29. After the
equilibration phase at standard conditions, a CO dropout was performed as described in
section 3.5.5 but without application of standard conditions after the dropout. The reaction
was terminated with the CO dropout experiment.
Figure3.6ReactorfillingandreactionconditionsasappliedduringtheCOdropouttest(DinoRun31).Standard
conditions(260°C,H
2
:CO:N
2
:Ar=60:20:10:10,41.6mlmin
‐1
totalflowperreactor)aremarkedbygreyareas.
Controlled cool-down of samples after reaction
After the last experiment of DinoRun26, DinoRun27, and DinoRun29, the H2 flow was
replaced by N2 in order to stop the reaction. Then, the heating was stopped, and the catalyst
samples were cooled down in a 20 % CO in N2 flow (41.6 ml min-1 total flow per reactor).
When a temperature of <100 °C was reached, the reactors were purged with pure N2
(100 ml min-1 total flow per reactor) and the pressure was released. After ~30 min of
purging with N2 it was safe to open the connections and remove the reactors from the
furnace. A small flow of N2 was applied after removal of reactors in order to avoid oxygen
contamination inside the feed lines.
After termination of DinoRun31, the catalyst samples were cooled down in 60 % H2 in N2.
Except of the cool down atmosphere, the same protocol was applied as described above.
H
2
, CO, Ar, N
2
Rh
500
mg
RhMn
100
mg
RhFe
150
mg
RhMnFe
165
mg
0 20 40 60 80 100 120 140 160 180
5
10
15
20
25
30
35
40
0
Time on stream / h
Volume flow / ml min
-1
Equilibration CO dropout
Temperature
CO
H
2
230
235
240
245
250
255
260
265
270
Temperature / °C
3.6Evaluationofcatalyticdata
29
3.6 Evaluation of catalytic data
Chromatograms were integrated and evaluated with the software ChemStation by Agilent.
The resulting gas phase composition data were then matched which the respective process
data by the built-in data export function of the process control system. The data evaluation
was automized by a script written in the programming language R.
For comparison of effects of catalyst modification on the reaction network, a statistical tree
approach was used.
Calculations
The obtained concentrations of all compounds were corrected for volume changes due to
the reaction and the following N2 dilution. Therefore, the mole fraction of Ar was used as
inert internal standard according to equation (3).
𝑥
𝑥
,
∙ 𝑥
,
𝑥
,
(3)
𝑥
correctedmolefractionofcompoundi
𝑥
,
molefractionofcompoundi asobtainedfromchromatogram
𝑥
,
molefractionofArinreactoroutlet
𝑥
,
molefractionofArinbypass
Carbon monoxide conversion 𝑋
was calculated indirectly from the summation of carbon
numbers in all products rather than directly from the CO concentration as at small
conversions the quantification of CO was not precise enough (equation (4)).
𝑋
∑
𝑥
𝐶
𝑥
,
(4)
𝑥
,
molefractionofCOintheinletgas
𝐶
carbonnumberoftheproducti
The selectivity S for each product i was determined by equation (5).
𝑆
𝑥
𝐶
∑
𝑥
,
𝐶
,
(5)
𝑥
,
molefractionofproductiandallotherproductsj
𝐶
,
carbonnumberofproductiandallotherproductsj
3Experimentaldetails
30
The production rates towards a product 𝑖 were calculated according to equation (6).
𝑟𝑥 𝑉 𝑝
𝑅 𝑇 𝑚 (6)
𝑉 totalvolumeflowperreactor
𝑚 catalystmass
𝑝 pressureatnormalconditions(101.325kPa)
𝑇 temperatureatnormalconditions(273.15K)
𝑅 idealgasconstant(8.314JK‐1mol‐1)
Contact times were calculated according to equation (7).
𝜏𝑉
𝑉 𝑝 𝑇
𝑝 𝑇 (7)
𝑉 catalystvolume
𝑝 pressureinsidethereactor
𝑇 temperatureinsidethereactor
Evaluation of the chain growth probability according to the Anderson-Schulz-Flory
distribution was performed for each product group (paraffins, olefins, alcohols, …)
individually (equation (8)).
log 𝑥
𝐶𝐶
log𝛼 (8)
𝛼 chaingrowthprobability
If mole fractions of the products obey the laws of ASF distribution, a semi-logarithmic plot
of ni/Ci as function of the carbon number produces a straight line and the chain growth
probability 𝛼 can be obtained from the slope.
Reaction rates are evaluated according to a power-law equation (9). When temperature and
one reactant partial pressure are kept constant, a log-log-plot of the reaction rate against
the varied reactant partial pressure produces a straight line with the slope representing the
reaction order.
𝑟 𝑘∙𝑝
∙𝑝
(9)
𝑘 rateconstant
𝑝 COpartialpressure
𝑝 H2partialpressure
𝑚 reactionorderregardingCO
𝑛 reactionorderregardingH2
3.6Evaluationofcatalyticdata
31
Apparent activation energies were calculated from the Arrhenius equation (10). Under
consideration of equation (9), a straight line is obtained when the natural logarithm of the
reaction rate is plotted against T–1with the slope representing the negative apparent
activation energy divided by R.
𝑘 𝑘∙𝑒
,
(10)
𝑘
frequencyfactor
𝐸
,
apparentactivationenergy
Statistical tree approach
In a statistical tree approach, the probability of different events to occur from a common
starting point are evaluated. In this specific case, selectivities at standard reaction
conditions were evaluated following the reaction pathways in the proposed network. The
selectivity towards each reaction product thereby represents the product of probabilities
for each surface reaction step along the formation pathway (Figure3.7). Consequently, the
sum of probabilities for reaction pathways originating from the same surface species is
always 100 %.
Figure3.7Schematicrepresentationofthestatisticaltreeapproach.Selectivitiestowardscertainreactionproducts
resultfromtheproductofprobabilitiesforeachsurfacereactionstepalongtheformationpathway.
Surface
species 1
Surface
species 2
Product A
S = p
1a
Surface
species 3
Surface
species 4
Product B
S = p
1b
ꞏ p
2a
p
1a
p
1b
p
1c
p
2a
p
2b
p
4a
p
4b
p
3a
p
3b
3Experimentaldetails
32
3.7 Catalyst characterization methods
Catalyst characterization was performed for samples after calcination, after H2 treatment,
and after the parameter field test (DinoRun26), the thermal stability test (DinoRun27) and
the CO dropout test (DinoRun31). Respective sample numbers are given in TableA1 in
Appendix A.
X‐raydiffraction(XRD)
Powder X-ray diffraction (XRD) measurements were performed in Bragg-Brentano
geometry on a D8 Advance II theta/theta diffractometer (Bruker AXS), using Ni-filtered Cu
Kα1,2 radiation and a position sensitive energy dispersive LynxEye silicon strip detector.
The angle variation was performed from 6° to 140°, with a step size of 0.02°. The sample
powder was filled into the recess of a cup-shaped sample holder, the surface of the powder
bed being flush with the sample holder edge (front loading). XRD measurements were
performed by Dr. Frank Girgsdies at the Department for Inorganic Chemistry at the Fritz
Haber Institute of the Max Planck Society.
Data analysis was performed with the software STOE WinXPow by comparison with
database entries of the International Centre for Diffraction Data (ICDD).
For estimation of the crystallite sizes from the degree of reflection broadening the Scherrer
equation was used. The diffractograms of the catalyst samples were baseline corrected with
a reference diffractograms of the pure support material. The resulting reflections were
fitted in Origin. The obtained values for the line broadening at half the maximum intensity 𝛽
and the Bragg angle 𝜃 were then used in the Scherrer equation (Equation (11)).
𝑑 𝐾𝜆
𝛽𝑐𝑜𝑠𝜃 (11)
𝑑 meansizeofordered(crystalline)domains
𝐾 dimensionlessshapefactor
𝜆 X‐raywavelength
𝛽 linebroadeningathalfthemaximumintensity(FWHM)
𝜃 Braggangle
For the dimensionless shape factor K a value of 0.9 was chosen. This is a typical value
assumed for supported metal particles.
3.7Catalystcharacterizationmethods
33
X‐rayphotoelectronspectroscopy(XPS)
XPS was measured on K-Alpha™ + X-ray Photoelectron Spectrometer (XPS) System (Thermo
Scientific), with Hemispheric 180° dual-focus analyzer with 128-channel detector. X-ray
monochromator is Micro focused Al-Kα radiation. For the measurement, the as-prepared
samples were directly loaded on the sample holder for measurement. The data were
collected with X-ray spot size of 200 μm, 20 scans for survey, and 50 scans for regions. XPS
measurements were performed by Shuang Li and Dr. Johannes Schmidt in the group of
Prof. Dr. Arne Thomas at TU Berlin.
Data evaluation was performed with the software Avantage by Thermo Scientific. With
integrated response factors, the software provided the surface content of each element in
wt-% and atom-% from the respective signal areas.
Transmissionelectronmicroscopy(TEM)
Scanning and high-resolution transmission electron microscopy (STEM/HRTEM) were
conducted on a FEI Talos F200X microscope. The microscope was operated at an
acceleration voltage of 200 kV. Elemental mappings were recorded with a Super-X system
including four silicon drift energy-dispersive X-ray (EDX) detectors. All samples were
prepared on holey carbon-coated copper grids (Plano GmbH, 400 mesh). Particle size
distribution were determined by measuring at least 250 particles of three different
domains. Electron microscopy was performed by Phil Preikschas and Kristian Knemeyer
(BasCat) at the facilities of the Fritz Haber Institute of the Max Planck Society.
An EDX mapping takes in average 30 minutes. It was taken care that the sample has not
shifted during the measurement by comparison of STEM images before and after the EDX
process. The mappings shown in this study were evaluated by me with the software Velox
2.6 by ThermoFisher Scientific. Line scans were calculated from the information obtained
during the EDX mapping.
For measuring and counting particle diameters images taken with the bright field detector
at a magnification of 305.6 kx were used in the software imageJ.138 In defined areas all
observable particles were measured and counted before moving to the next area. A
minimum of 200 particles were evaluate for each sample.
3Experimentaldetails
34
N2physisorptionaccordingtoBrunauer‐Emmett‐Tellermethod(BET)
The surface area of the samples was analyzed by nitrogen physisorption at liquid nitrogen
temperature on a Quantachrome Autosorb AS-6B-MP analyzer. As pretreatment, the
samples were outgassed with a Quantachrome Autosorb Degasser at 200 °C for 2 h (ramp
2) in vacuum. In total 79 data points were collected (adsorption/desorption). The specific
surface area was calculated according to the Brunauer–Emmett–Teller method. Therefore,
adsorption data in the relative pressure p/p0 range of 0.05-0.3 were used. BET
measurements have been performed by Maike Hashagen at the Department for Inorganic
Chemistry at the Fritz Haber Institute of the Max Planck Society.
InductivelyCoupledPlasmaOpticalEmissionSpectrometry(ICP‐OES)
The elemental composition of the catalysts was determined via ICP-OES. The analysis was
performed by the contract laboratory Mikroanalytisches Labor Kolbe (www.mikro-lab.de).
35
4 Morphology and surface
properties of Rh/Mn/Fe/SiO2
catalysts
Four catalyst samples were used in this study. Beside the pure Rh/SiO2 catalysts, three
modified catalysts with added Mn and/or Fe were investigated. The catalyst loadings on the
calcined samples were verified via inductively coupled plasma optical emission
spectrometry (ICP-OES, Table4.1). It should be pointed out here, that throughout this work
the catalysts will be referred to with the simplified expressions “Rh/SiO2”, “RhMn/SiO2”,
“RhFe/SiO2”, and “RhMnFe/SiO2” due to readability. This denotation does not contain
information about the oxidation state of the metals or their potential presence as metal
oxides or alloys.
Table4.1Compositionandweight‐loadingsoftheusedcatalystsaftercalcinationdeterminedbyICP‐OES.Target
valuesaregiveninparenthesis.
CatalystnameCatalystloading/wt‐%
RhMnFe
Rh/SiO22.2(2.8)‐ ‐
RhMn/SiO22.4(2.8)1.6(1.5)‐
RhFe/SiO22.3(2.8)‐ 0.5(0.5)
RhMnFe/SiO22.3(2.8)1.5(1.5)0.5(0.5)
The catalyst samples were characterized after different treatments. For each catalytic run,
the calcined sample was the pre-catalyst filled into the reactors. All catalytic runs were
performed from the same batches. Before reaction conditions were applied, a standardized
in‐situ H2 treatment was applied. One set of samples was taken directly after that procedure
(Figure4.1). The sample obtained after the kinetic study has been exposed to various
reaction conditions but no other harsh impacts such as high temperatures or cofeeds.
However, it has seen the longest total time on stream. During the high temperature study,
temperatures up to 320 °C were applied at standard conditions (54 bar, H
2:CO:N2:Ar =
60:20:10:10). In order to investigate the influence of the absence of CO on the catalyst
structure, one set of samples was taken after a CO dropout experiment and cool down in
60 % H2/N2.
4MorphologyandsurfacepropertiesofRh/Mn/Fe/SiO2catalysts
36
Figure4.1Schematicoverviewofcharacterizedcatalystsamples.
1Reactionatmospherewasvaried:CO:H2:Ar:N2=2.5‐20:30‐60:10:balancein18steps.
2Standardreactionatmosphere:CO:H2:Ar:N2=20:60:10:10.
There are some considerations on the catalyst handling that should be noted. In its working
state, the catalyst is operating under high pressures of CO and H2. The characterization
methods shown here are applied before or after reaction. In order to keep the catalyst in a
state as close as possible to the working state, the cool down process after catalytic testing
usually was performed in a CO/N2 atmosphere except for the CO dropout sample. Dropout
studies have shown that the absence of CO has significant impact on the catalyst structure
as opposed to absence of H2. The dropout experiments are discussed in detail in chapters
6.6 and 7.6 for Rh/SiO2 and the modified catalysts, respectively.
Also, all catalyst samples were handled with contact to air which should be considered for
interpretation of all data. Mn and Fe are oxophilic and therefore easily oxidized at the
surface. It can be assumed that this effect does not have an impact on the location of the
metals on the support which means that EDX analyses are still meaningful. However, the
oxidation state cannot be determined with certainty. Structural features, such as alloy
formation under reaction conditions, should be considered possible despite non-metallic
oxidation states in spent samples.
Kinetic study
(DinoRun26)
Calcined
Stepwise
dried and
calcined at
maximum
350 °C in air
High temperature study
(DinoRun27)
CO dropout
(DinoRun31)
Maximum temperature
Total time on stream
Reaction atmosphere
Cool down atmosphere
260 °C
140 h
Standard
2
H
2
/N
2
Maximum temperature
Total time on stream
Reaction atmosphere
Cool down atmosphere
320 °C
380 h
Standard
2
CO/N
2
Maximum temperature
Total time on stream
Reaction atmosphere
Cool down atmosphere
260 °C
530 h
Various
1
CO/N
2
H
2
treated
Treated in reactor
according to
in-situ H
2
treatment
procedure
in-situ H
2
treatment
4.1Structuralproperties(XRD,XPS)
37
In order to gain information about their structural properties and the influence of different
treatments, the obtained sets of the four catalysts were characterized viaX-ray diffraction
(XRD), scanning transmission electron microscopy with energy-dispersive X-ray
spectroscopy (STEM, STEM-EDX), and X-ray photoelectron spectroscopy (XPS). The results
are discussed in the following sections.
4.1 Structural properties (XRD, XPS)
The support material used in this study is Davisil 636, a high surface, high purity silica
material. The specific surface area obtained from N2 adsorption is ~420–470 m2 g-1 for the
samples of all catalysts and different treatments. No clear impact of composition or
treatment on the surface area was observed (TableA2 in Appendix B).
XRD gives information on the crystallographic structure of the investigated sample. For
supported metal catalysts, XRD presents the challenge that small crystallite sizes lead to a
broadening of the peaks. For crystallite sizes below 2–2.5 nm, the obtained information on
specific phases is limited.139 However, the width of the peak is related to the crystallite size
via the Scherrer equation enabling the assessment of crystallite sizes from the
diffractograms (compare chapter 3.7 for details and equation).
After calcination, only vague indication of very small Rh2O3 crystallites could be identified
on Rh/SiO2 (Figure4.2A). In this case, the peak broadening is very distinct and the
estimation of crystallite sizes from the Scherrer equation might be unprecise. However, a
mean crystallite size of well below 2 nm can be assumed. After H2 treatment according to
the standard procedure prior to each catalytic run, a broad peak located at the typical
position of the most intense Rh(111) reflection of metallic Rh was observed, indicating that
Rh2O3 was reduced to metallic Rh crystallites. The crystallite size was calculated to be
approximately 2.5 nm after the H2 treatment. The CO dropout sample was treated under
reaction conditions for 140 h time on stream. No clear impact on the crystallite size was
observed. Even the application of temperatures up to 320 °C and a time on stream of 380 h
does not lead to a distinct crystallite growth compared to the H2 treated sample. However,
the Rh crystallites clearly have grown during the long-term catalytic testing over 530 h in
the kinetic study.
This behavior indicates that a long time on stream has more significant impact on crystallite
growth than other factors such as high temperatures. In addition, crystallite growth might
be favored under specific reaction conditions that were applied during the kinetic study.
The traditional idea of sintering would predict faster crystallite growth at higher
temperatures as the mobility of metal atoms on the support is usually increased.140 Since
this is not the case here, most likely other factors such as applied gas atmosphere have a
4MorphologyandsurfacepropertiesofRh/Mn/Fe/SiO2catalysts
38
more distinct influence on Rh particle sintering than temperature. Claeys etal. have shown
a surface carbonyl mediated sintering behavior of Co/Al2O3 dependent on reaction
conditions during Fischer-Tropsch catalysis.141 In particular, CO and H2O partial pressures
were shown to be relevant for particle growth. In the present study, crystallite growth was
rather observed after the catalytic test involving different gas phase compositions than after
exposition to high temperatures. The obtained results therefore suggest that also in this
case feed composition might have a more important impact than temperature.
Figure4.2ComparisonofX‐raydiffractogramsobtainedfromsamplesafterdifferenttreatmentsofRh/SiO2(A),
RhMn/SiO2(B),RhFe/SiO2(C),RhMnFe/SiO2(D).Catalyticrunsaresortedaccordingtoincreasingtotaltimeon
stream.PositionsformetallicRh(C5‐685),Rh2O3(C41‐541),MnO(00‐065‐0638),Fe2O3(C75‐449)andRhFealloy
(1:1,C25‐1408)reflectionsaredisplayedaccordingtorespectivereferencesintheICSDdatabase.
*Smallsharpreflectionsbelongtosteatite(MgSiO3,00‐011‐02)contaminationofsample.
30 40 50 60 70 80 90 30 40 50 60 70 80 90
30 40 50 60 70 80 90 30 40 50 60 70 80 90
Intensity / a.u.
2 / °
calcined
after H2 treatment
after kinetic study
after high temperatures
after CO dropout
2 / °
Intensity / a.u.
2 / °
calcined
after H2 treatment
after kinetic study
after high temperatures
after CO dropout
calcined
after H2 treatment
after kinetic study
after high temperatures
after CO dropout
calcined
after H2 treatment
after kinetic study
after high temperatures*
after CO dropout
2 / °
Rh2O3
metallic Rh
Rh2O3
metallic Rh
Rh2O3 Rh2O3
metallic Rh
metallic Rh
MnO
RhFe
Fe2O3
AB
CD
Rh/SiO
2
RhMn/SiO
2
RhFe/SiO
2
RhMnFe/SiO
2
4.1Structuralproperties(XRD,XPS)
39
Modification of the catalyst with Mn and Fe has little impact for the calcined and H2 treated
samples. For all of them, a phase transformation from Rh2O3 to metallic Rh was observed.
No clear indication for any other crystalline phases such as the respective metal oxides were
found. The calculated crystallite sizes are the same in the range of uncertainty (Table4.2).
After the kinetic study, all samples show reflections in the range of metallic Rh. The
reflection observed for the RhMn/SiO2 sample is slightly shifted and its peak shape suggests
overlapping reflections. However, no clear information on the crystalline phase could be
obtained from this diffractogram. A reference for MnO shows a major reflection where the
main peak is located but the absence of any other evaluable peaks does not allow a more
precise phase analysis. However, Mn addition has significant influence on the crystallite
sizes observed for the spent samples of the kinetic study (Figure4.2B,D). Mn containing
catalysts clearly show smaller crystallite sizes in the range of the H2 treated sample
indicating that Mn addition clearly stabilizes small Rh crystallites and hinders crystallite
growth. An exact evaluation of the crystallite diameter cannot be performed due to the
overlapping reflections and low signal intensity. Fe addition does not exhibit a similar effect.
Since no Fe related reflections, e.g. in the area of a RhFe alloy, were observed, the impact of
Fe on the catalyst properties cannot be defined via XRD. Fe oxides might be present but
could potentially be missed due to X-ray amorphous structures.
The results obtained for the spent samples after high temperatures and CO dropout
experiments are very similar to Rh/SiO2. Very broad reflections suggest small crystallite
sizes in the range of the H2 treated sample. Especially for RhMn/SiO2 the assignment of the
reflection to metallic Rh would be speculative. Since only the main reflection was observed,
a thorough phase analysis could not be performed.
Table4.2CrystallitesizesestimatedfromtheScherrerequation.Therespectivephasesaregiveninparenthesisifan
assignmentwaspossible.
Estimatedcrystallitesizesa/nm
Rh/SiO2RhMn/SiO2RhFe/SiO2RhMnFe/SiO2
Calcined1.3
(Rh2O3)
1.4
(Rh2O3)
1.5
(Rh2O3/Fe2O3)
1.6
(Rh2O3/Fe2O3)
H2treatment2.5(Rh0)2.0(Rh0)2.0(Rh0)2.5(Rh0)
COdropoutb2.4(Rh0)2.2(Rh0)2.4(Rh0)2.5(Rh0)
Hightemperaturesc2.3(Rh0)2.0(Rh0)2.3(Rh0)2.4(‐)
Kineticstudyd5.7(Rh0)2.8(‐)4.7(Rh0)2.5(‐)
aForallcrystallitesizesanuncertaintyof30%isassumed.
bSpentsampleofDinoRun31(260°Cmaximumreactiontemperature,140htimeonstream,cooldowninH2/N2)
cSpentsampleofDinoRun27(320°Cmaximumreactiontemperature,380htimeonstream,cooldowninCO/N2)
dSpentsampleofDinoRun26(260°Cmaximumreactiontemperature,530htimeonstream,cooldowninCO/N2)
4MorphologyandsurfacepropertiesofRh/Mn/Fe/SiO2catalysts
40
In conclusion, all investigated catalyst samples have some features in common. In their
calcined state, very small Rh2O3 crystallites give very broad reflections. After H2 treatment
according to the in‐situ procedure of catalytic runs, Rh2O3 is reduced to metallic Rh. The
phase transformation goes along with some crystallite growth by sintering. Reaction
conditions applied during the high temperature run and the CO dropout experiment do not
lead to significant increase of crystallite diameters. However, long time on stream and
different feed compositions cause crystallite growth on Rh/SiO2 and RhFe/SiO2. The
mechanism of crystallite growth is probably rather related to the gas phase composition
than to other factors like high temperatures. High concentrations of water or other reaction
products as well as specific CO:H2 ratios might contribute to enhanced crystallite growth.
XRD does not provide strong evidence for temperature induced sintering at STE conditions
up to 320 °C. Rather concepts such as the formation and disintegration of mobile carbonyl
species in dependence of the gas phase composition should be considered. As evidenced by
the four catalysts, the presence of Mn in both Rh and RhFe catalysts reliably hinders
sintering of the active metallic particles. One possible role of Mn thus appears to be a
stabilization of smaller active particles.
The change of oxidation states from Rh2O3 to metallic Rh upon H2 treatment could also be
confirmed with XPS. Also, the obtained Rh3d signals indicate that Rh is also in its metallic
state after reaction. For Mn and Fe, all XPS results suggest their presence as oxides.
However, their oxophilic nature might lead to quick oxidation on the surface when the
sample come in contact to air which was the case for all samples. Therefore, no clear
statement on the oxidation state of Mn and Fe under reaction conditions can be made. All
XPS spectra and a more detailed evaluation can be found in an additional section in
Appendix B.1. From the Si:metal ratio assumptions on the metal dispersion could be made.
Here, the evaluation showed a specifically pronounced effect of the reaction conditions
during the kinetic study indicating a high metal dispersion after the reaction. Electron
microscopy was used to further investigate the structural properties as a function of
different reaction conditions.
4.2Morphologyandelementaldistribution(STEM‐EDX)
41
4.2 Morphology and elemental distribution (STEM-EDX)
Electron microscopy was performed on the H2 treated and the three STE-treated samples of
each catalyst system. The calcined samples were not investigated due to the poor stability
of Rh2O3 nanoparticles upon exposition to the electron beam.142
General structural features were evaluated from images taken with the dark-field, bright-
field and high-angle annular dark-field (HAADF) detector. Particle size distributions were
obtained by systematic measuring and counting more than 200 particles per sample in
different areas from bright-field images. The obtained particle diameters are large in
accordance with the crystallite sizes obtained from XRD (Table4.3). In general, it was
observed that the particle sizes barely change after reaction on the Mn-containing catalysts
RhMn/SiO2 and RhMnFe/SiO2. In both cases, the small particles stay evenly distributed over
the support. For Rh/SiO2 and RhFe/SiO2, a slight particle growth was observed especially
after the kinetic study which is further consistent with XRD results. In addition, the
measured primary particles formed agglomerates after the kinetic study, further supporting
the idea of highly mobile Rh species under certain reaction conditions with an emphasis on
the composition of the atmosphere. For Rh/SiO2, agglomerates with a diameter of up to
200 nm were observed. On RhFe/SiO2, these structures were smaller with up to 100 nm.
Agglomeration instead of sintering might suggest non-crystalline particles or particles with
non-crystalline surface structures, e.g. induced by integration of adsorbed species into the
structure. For Co catalysts, a mechanism has been proposed that relates the mobility of Co
on a silica surface to the partial pressures of CO and water enabling the formation of highly
mobile (sub)carbonyl species.65,141 Such a process might as well be relevant for Rh-based
catalysts. To the best of our knowledge, this effect has not been reported in literature before.
Table4.3ParticlediametersretrievedfromTEMimages.Diametersofagglomeratesaregiveninparenthesesif
applicable.
Particlediametersa/nm
Rh/SiO2RhMn/SiO2RhFe/SiO2RhMnFe/SiO2
H2treatment2.9±0.92.4±0.72.9±0.82.2±0.5
COdropoutb2.4±0.52.9±0.83.0±0.83.2±0.7
Hightemperaturesc3.9±0.92.5±0.72.8±0.72.2±0.5
Kineticstudyd4.4±1.4(≤200)2.9±1.03.3±1.2(≤100)3.1±0.9
aParticlediameterswereobtainedfromBFimageswithamagnificationof305.6kx.
bSpentsampleofDinoRun31(260°Cmaximumreactiontemperature,140htimeonstream,cooldowninH2/N2)
cSpentsampleofDinoRun27(320°Cmaximumreactiontemperature,380htimeonstream,cooldowninCO/N2)
dSpentsampleofDinoRun26(260°Cmaximumreactiontemperature,530htimeonstream,cooldowninCO/N2)
4MorphologyandsurfacepropertiesofRh/Mn/Fe/SiO2catalysts
42
The mentioned structural features for each sample are evaluated and described in more
detail in an additional section Appendix 1.1.1.1B.2. This section also provides details about
the reproducibility of the STEM results and shows the very heterogeneous structure of the
Rh/SiO2 catalyst when different particles are investigated.
This chapter rather focusses on the distribution of the metals on the surface investigated
with the energy dispersive X-ray analysis (EDX) mapping technique which can be used to
gain locally resolved information about the elemental composition of materials. As
expected, the observed nanoparticles on Rh/SiO2 are formed from Rh indicated by the good
correlation of bright spots on the HAADF image and the red areas in the Rh EDX mapping
(Figure4.3). That applies for the H2 treated sample with small particles as well as the large
agglomerates in the spent sample of the kinetic study. Impurities like Fe could not be found
in the spectra of both samples. From the EDX mapping, no insight into the nature of
agglomerated structures could be obtained. Potential carbon containing structural features
are difficult to determine as the samples show common carbon contamination, carbon
containing grids were used and additional carbon is deposited on the sample surface during
the EDX scanning process.143
Figure4.3HAADF‐STEMimagesofaRh/SiO
2
sampleobtainedafterH
2
treatment(A)andthekineticstudy(C)with
therespectiveEDXmappingofRh(B,D).
4.2Morphologyandelementaldistribution(STEM‐EDX)
43
For the modified catalysts, EDX mapping gives information about how the metals are
located to each other. On the H2 treated sample of RhMn/SiO2, the particles indicated by the
white spots in the HAADF image clearly correlate with the red areas of the EDX scan for Rh
(Figure4.4A,B). Mn is well distributed over the support with some areas where a higher
concentration of Mn can be found (Figure4.4C). These areas do not correlate with the
location of Rh particles. In the spent sample of the kinetic study this effect becomes even
more pronounced. The areas with high Rh concentration do not exhibit increased Mn
concentration (Figure4.4D‐F).
Figure4.4HAADF‐STEMimagesofaRhMn/SiO
2
sampleobtainedafterH
2
treatment(A)andthekineticstudy(D)
withtherespectiveEDXmappingofRh(C,E)andMn(C,F).
From EDX scans, no information about the oxidation state of the metals can be received.
While DFT calculation suggest a RhMn alloy (Mn0),99 the spatial distribution of Rh and Mn
in the presented experiments contradicts potential alloy formation. In literature, electron
paramagnetic resonance experiments revealed a formal oxidation state of +2 for a reduced
RhMn/SiO2 catalyst.144 Another study suggests a formal oxidation state of +3 from X-ray
adsorption spectra (XAS) taken in air but lower valent Mn suboxide phases (+0.22) from
temperature-programmed reduction (TPR) experiments.82 The high oxophilicity of Mn
enables Mn oxide formation on silica even when Mn precursors with negative oxidation
states are used for impregnation on silica under protective atmosphere.145 It can be
assumed that the H2 treatment does not lead to reduction of Mn oxide to metallic Mn.
4MorphologyandsurfacepropertiesofRh/Mn/Fe/SiO2catalysts
44
Mn oxide is present all over the silica support building up a layer or matrix rather than
individual particles. Therefore, the most likely interpretation of the STEM-EDX results
features small Rh particles stabilized by Mn oxide surrounding the particles. The Mn oxide
addition to Rh/SiO2 therefore appears to rather influence the nanostructure of the catalyst
than the electronic structure. This interpretation agrees with earlier reports of Schwartz et
al.124 and Dimitrakopoulou etal.82 providing FTIR and EXAFS studies that do not indicate a
clear electronic effect induced by the presence of Mn on CO bond frequencies or alloy
formation. In contrast, Mei etal.99 claimed that a RhMn alloy is thermodynamically more
plausible than a mixed metal/metal-oxide under reaction conditions based on DFT studies
the alloy formation is often assumed in theoretical studies on the promoting effect of
Mn.96,115 In the present study, no evidence for such bimetallic alloy structures can be found.
Due to the small particle sizes in the H2 sample and the low Fe loading, assumption about
the interaction of Rh and Fe on RhFe/SiO2 are difficult (Figure4.5A‐C). No clear correlation
between Rh and Fe was observed. The opposite is the case for the spent sample after the
kinetic study. The presence of large agglomerates leads to a better signal-to-noise ratio. The
EDX scans for Rh and Fe clearly show that Fe is located in close proximity to Rh indicating a
strong interaction between the two metals (Figure4.5D‐F). Due to the poor signal-to-noise
ratio for the small particles in the H2 treated sample, it cannot clearly be stated if the close
interaction of Rh and Fe has been formed during synthesis or pretreatment of the catalyst
or if reaction conditions were necessary for the formation.
At this point it should be mentioned that the catalysts showing agglomerated particles
(Rh/SiO2 and RhFe/SiO2) do not show a higher degree of activity loss than the Mn-
containing catalysts without an observed agglomeration effect (compare chapter 5.2
Catalytic stability). This indicates that the surface of the primary particles building up the
agglomerates is still available for catalysis.
For RhFe/SiO2, particles were investigated regarding the elemental composition in different
domains yielding Rh:Fe ratios of 1.8–5.7. Also, the presence of pure Rh particles was
observed which is consistent with XRD results. A clear statement regarding the
stoichiometry of RhFe particles cannot be made. However, the formation of nanoalloyed
RhFe particles on SiO2 and TiO2 has been proposed from electron microscopy and H2
chemisorption before.86 XRD and pair distribution functions were used to quantify each
phase and provide information on local atomic structure under reaction conditions with a
total pressure of 2 bar.110 The authors proposed RhFe (1:1) alloying on the particle surface
and Rh cores. In contrast, the investigation of Rh and Fe line scans in this study rather
suggest a bulk alloy formation than core-shell particles. The detailed evaluation and
description of the observed RhFe particles are presented in an additional section in
Appendix 1.1.1.1B.3. Other studies suggest that the contact of Rh to Fe oxide species are
4.2Morphologyandelementaldistribution(STEM‐EDX)
45
important for certain selectivity features.108,109 In this study, no indication for the presence
of a Fe oxide layer in contact with Rh particles was found.
Figure4.5HAADF‐STEMimagesofaRhFe/SiO
2
sampleobtainedafterH
2
treatment(A)andthekineticstudy(D)
withtherespectiveEDXmappingofRh(B,E)andFe(C,F).
On RhMnFe/SiO2, EDX analysis for the H2 treated sample again shows a low signal-to-noise
ratio. A correlation of Rh and Fe scans might be assumed (Figure4.6B,D). Mn, however,
shows even distribution over the silica support (Figure4.6C). The slightly larger particles
after the kinetic study reveal a much clearer picture. Rh and Fe are clearly present together
in the particle as observed for RhFe/SiO2 (Figure4.6F,H). As already observed for
RhMn/SiO2, Mn oxide is evenly spread over the support and provides a matrix on which the
RhFe particles are anchored, preventing particle growth and agglomeration. The structural
investigation of the trimetallic catalyst therefore supports the hypothesis that Mn – or more
likely Mn oxide – has a stabilizing effect on the nanostructure of Rh or RhFe particles and Fe
has an electronic effect on Rh being in close contact or even under nanoalloy formation.
The results regarding Mn oxide matrix and RhFe alloy formation are consistent with
previous studies of the RhMn/SiO2 and RhMnFe/SiO2 catalysts used for this study.142 The
existence of a RhFe alloy phase and metallic Fe is supported by electron energy-loss
spectroscopy (EELS) and STEM-EELS mapping. CO adsorption experiments on the same
catalysts suggest the Mn oxide matrix to partially cover Rh or RhFe particles decreasing the
4MorphologyandsurfacepropertiesofRh/Mn/Fe/SiO2catalysts
46
specific metallic surface.82 However, these studies did not reveal the important impact of
reaction conditions on the catalyst morphology. The high mobility of Rh species leading to
agglomeration of Rh and RhFe particles could only be observed by characterization of
samples after different treatments. Also, the functionality of the Mn oxide matrix as anchor
for Rh and RhFe particles even under challenging reaction conditions could be highlighted
by the presented results.
Figure4.6HAADF‐STEMimagesofaRhMnFe/SiO
2
sampleobtainedafterH
2
treatment(A)andthekineticstudy(E)
withtherespectiveEDXmappingofRh(B,F),Mn(C,G)andFe(D,H).
4.3Conclusion
47
4.3 Conclusion
All four catalysts were investigated and compared regarding their composition and
treatment. The performed initial H2 treatment at the beginning of each catalytic run leads
to complete reduction of Rh2O3 to metallic Rh on all catalysts. Oxidation states of Mn and Fe
were difficult to assess due to low loadings and handling in air.
Reaction conditions featuring high temperatures or CO dropout have less structural impact
than the kinetic study involving the longest time on stream, different feed conditions and
contact times. Certain compositions of the surrounding atmosphere appear to have crucial
impact on the mobility of Rh on a silica surface. After the kinetic study, Rh and RhFe particles
have formed agglomerates of up to 200 nm and 50 nm, respectively (Figure4.7). Rh
mobility was observed to strongly depend on the gas phase composition which agrees to
literature findings reporting the formation of carbonyl species as important requirement
for particle growth. Therefore, it might be assumed that Rh carbonyl species play an
important role for the mobility of Rh on silica. However, the presence of a highly dispersed
Mn oxide matrix reduces the mobility of Rh and RhFe species significantly, not only leading
to smaller particle sizes but also preventing agglomeration.
Figure4.7SchematicrepresentationofthemorphologiesofRh/SiO2,RhMn/SiO2,RhFe/SiO2andRhMnFe/SiO2after
calcination,H2treatmentandafterreactioninthekineticstudy(DinoRun26)basedoninformationobtainedfrom
XRD,STEM‐EDXandXPSanalysis.
SiO
2
3 nm Rh particles
SiO
2
up to 200 nm agglomerates
5 nm Rh particles
SiO
2
4 nm RhFe particles
up to 50 nm agglomerates
Some Rh particles
SiO
2
1-2 nm Rh
2
O
3
particles
Rh/SiO
2
2.5 nm Rh particles
MnOx matrix
SiO
2
3 nm Rh particles
MnOx matrix
SiO
2
Very small FeOx particles
or FeOx matrix
1-2 nm Rh
2
O
3
particles
RhFe/SiO
2
SiO
2
3 nm Rh particles
Fe/FeOx particles or
FeOx matrix
SiO
2
2.5 nm Rh particles
(Mn,Fe)Ox matrix
SiO
2
SiO
2
(Mn,Fe)Ox matrix
RhMnFe/SiO
2
1-2 nm Rh
2
O
3
particles
H2treatment
500 h STE reaction (kinetic study)
1-2 nm Rh
2
O
3
particles
MnOx matrix
SiO
2
RhMn/SiO
2
(Mn,Fe)Ox matrix
SiO
2
3 nm RhFe particles
SiO
2
4MorphologyandsurfacepropertiesofRh/Mn/Fe/SiO2catalysts
48
Rh and Fe were found in close contact on the surface in both Fe containing catalysts. The
results suggest the formation of bulk RhFe phases rather than core-shell structures. A clear
statement on the RhFe alloy composition could not be made. For the RhFe/SiO2 catalyst,
pure Rh particles were observed next to RhFe particles.
The complexity and interconnectivity of catalyst structure and reaction conditions has not
been reported for this reaction so far. The following chapters will examine reactivity for the
four catalysts in detail. In chapter 8.4, the demonstrated structural features are further
discussed in relation to obtained reactivity features.
49
5 Stability and reproducibility of
catalytic data
For detailed mechanistic and kinetic investigation of catalytic data it is crucial to evaluate
stability as well as reproducibility. The time-on-stream behavior of the catalyst or
reproducibility issues must be known to decouple experimental phenomena from catalytic
and kinetic effects. In this chapter, all experimental effects on the catalytic performance of
the four considered catalyst systems are described. First, the quality of the data in terms of
the carbon balance is described. Furthermore, the catalytic stability upon thermal
treatment, long-term testing and special gas phase compositions stability are evaluated.
Finally, the reproducibility of catalytic data between different catalytic runs is shown.
5.1 Carbon balance
For evaluation of the carbon balances, the sum of concentrations of all carbon containing
reaction products weighted by their carbon number and unreacted CO at reactor outlet is
calculated, corrected by the dilution coefficient and compared to the CO concentration
measured at reactor inlet.
The carbon balances were determined to be 99 ± 2 % over a wide range of reaction
conditions for all four considered catalysts (Figure5.1). This indicates that all important
reaction products were detected, identified and quantified precisely. The complexity of the
product spectrum requires careful operation of the online GC. The addition of a Polyarc
reactor leads to an accurate quantification of hydrocarbons as well as oxygenates even in
small quantities. Especially the correct quantification of oxygenates usually is challenging
due to the difficult calibration of compounds with high vapor pressures. For the setup used
in this study, only methane had to be calibrated in an easy and precise manner leading to a
very high accuracy of the quantitative catalytic data.
However, it should be noted that CO conversion, CO consumption rates and selectivities are
calculated based on the product concentration rather than on feed concentrations. This
product-based approach leads to more reproducible and reliable results. Most of the results
were obtained at small CO conversions. Combined with the broad product spectrum, feed-
based calculations therefore are affected by a larger experimental error.
5Stabilityandreproducibilityofcatalyticdata
50
Figure5.1CarbonbalanceoverRh/SiO
2
(A),RhMn/SiO
2
(B),RhFe/SiO
2
(C)andRhMnFe/SiO
2
(D)asafunctionofCO
andH
2
partialpressures.(DinoRun26,54bar,260°C,H
2
:CO:N
2
:Ar=30‐60:2.5‐20:10‐57.5:10,41.7mlmin
‐1
,
Rh:2500h
‐1
,RhMn:12500h
‐1,
RhFe:25000h
‐1
,RhMnFe:7580h
‐1
)
5.2 Catalytic stability
Effects caused by changes of the catalyst structure that might be induced by thermal
treatment, long-term catalytic testing or unusual gas phase compositions need to be
separated from real kinetic effects. Therefore, reference reaction conditions were
repeatedly applied and obtained results were compared in order to evaluate the catalytic
stability.
Thermal stability
In a separate catalytic run the thermal stability of the four catalysts were evaluated. All
catalysts show an initial equilibration to reaction conditions up to 120 h time on stream and
a close to constant performance thereafter. The initial equilibration period will be discussed
in section 5.2.2. Once the catalysts showed equilibrated behavior, the temperature was
varied stepwise up to 320 °C at constant feed conditions. Each temperature was held
constant for approximately 24 h. Up to a temperature of 270 °C no loss of activity over a
time range of 24 h could be detected (Figure5.2, 130-220 h TOS). Starting from 280 °C
(230 h TOS) RhMn/SiO2 shows slight decrease of CO conversion over time on stream. At
290 °C (250 h TOS) a loss of activity over time is clearly observable for all catalyst.
5.2Catalyticstability
51
Figure5.2COconversionofRh/SiO2,RhMn/SiO2,RhFe/SiO2andRhMnFe/SiO2overtimeonstreamduringa
temperatureramp(DinoRun27,54bar,243‐320°C,H2:CO:N2:Ar=60:20:10:10,41.7mlmin‐1,Rh:2500h‐1,
RhMn:12500h‐1,RhFe:8330h‐1,RhMnFe:7580h‐1)
Comparison of the data obtained at 260 °C before high temperature treatment
(~180 h TOS) with data obtained after the application of high temperature shows a
permanent decrease of CO conversion. From this experiment, it cannot be determined if the
loss of activity is induced directly by high temperatures or by high CO conversion levels as
a result of increased temperatures. High conversion leads to high product concentrations in
the gas phase, e.g. water which could lead to water induced surface modifications. It should
be noted that no significant effect of the high temperature treatment on the S-vs-X behavior
could be observed. The effect of increasing temperatures on product selectivities are shown
in the Appendix for completion (FigureA15 and FigureA16 in Appendix C.1).
The extent of permanent activity loss differs between the catalyst systems. The CO
consumption rate decreases by 18 % from before to after the high temperature treatment
over Rh/SiO2. Over the modified catalysts RhMn/SiO2, RhFe/SiO2 and RhMnFe/SiO2 CO
consumption rates decrease by 37 %, 24 % and 36 %, respectively. The effect of high
temperatures is more severe on the modified catalysts compared to pure Rh/SiO2. This
could indicate structural changes that lead to a loss of the rate accelerating properties of the
modified catalysts. In particular, this seems to be important for Mn containing catalysts.
Two phenomena are thinkable. Either temperatures above 270 °C lead to sintering and
therefore a reduction of the Rh-Mn interface area or the high temperatures induce a higher
coverage of Rh particles with Mn oxide as a consequence of strong metal-metal oxide
interaction analogous to strong metal support interactions (SMSI). In this case, the higher
0 50 100 150 200 250 300 350 400
10
20
30
40
0
Time on stream / h
CO conversion / %
Temperature
Rh
RhMn
RhFe
RhMnFe
Temperature variationEquilibration
225
250
275
300
325
Temperature / °C
5Stabilityandreproducibilityofcatalyticdata
52
temperatures and higher concentrations of products such as water would increase the
mobility of Mn oxide leading to structural modifications induced by the reaction conditions.
The very small particle sizes found for Mn containing catalysts after the testing (compare
chapter 4.1) rather support the latter hypothesis.
The results of this experiment yielded important information about suitable reaction
conditions for kinetic experiments. Arrhenius plots over all temperatures (FigureA17-
FigureA20 in Appendix C.2) show linear trends for most products only up to 270 °C. Thus,
reaction induced surface modifications are likely above 270 °C. Hence, experiments
targeting reaction pathways and kinetics were performed only at temperature up to 260 °C
in this work.
STE performance data have been reported typically in a range from 250–320 °C (see
reference 11 and references therein). The resulting performance values differ between many
of those reports. Moreover, catalyst stability has not been addressed in most of those
reports. It is therefore likely that the variation in performance values reported for
temperatures above 270°C could be caused by a different extend of reaction induced surface
modification.
In conclusion, the obtained data have shown a common behavior for all considered catalysts
and revealed a critical temperature of 280/290 °C above which the loss of catalytic activity
is strongly accelerated. The activity loss is particularly severe for Mn containing catalysts
and likely it is caused by a loss of specific rate accelerating structures by poisoning with
reaction products or enhanced coverage of Rh or RhFe particles by Mn oxide (SMSI).
Long-term stability
In order to monitor the evolution of catalytic behavior over a long time under reaction
conditions, reference conditions were repeatedly applied throughout the kinetic study
(DinoRun26) over a total time on stream of 23 days. Throughout this run, conditions that
were assumed to cause lasting structural modification to the catalysts – such as high
temperatures, cofeeds, dropouts – were avoided.
All four catalysts show a common feature during the first 100 h time on stream. During this
initial period, changes in activity and selectivity were clearly observed (Figure5.3). The
application of the reaction mixture containing 20 vol-% CO and 60 vol-% H2 causes an initial
formation of the working catalyst. This behavior proves the importance of allowing enough
time for the catalyst equilibration before other catalytic experiments can start. The
observed time range of ~100 h suggests a slow formation of the catalyst structure rather
than an equilibration of adsorbed species. This period is discussed in detail in chapter 6.2
for Rh/SiO2 and chapter 7.2 for the modified catalysts.
5.2Catalyticstability
53
Figure5.3Long‐termstabilityofRh/SiO2(A),RhMn/SiO2(B),RhFe/SiO2(C)andRhMnFe/SiO2(D)plottedasCO
conversionatrepeatedlymeasuredreferenceconditionsovertimeonstream(blacksquares).Abovethefigures,
theexperimentsperformedbetweenreferenceconditionsaredenoted.Thecolumnsshowassociatedproduct
yieldsovertimeonstreamwiththeselectivityvaluesofthemajorproductsmarkedinwhite.YieldsofH2Oare
shownaswhitedots.(DinoRun26,54bar,260°C,H2:CO:N2:Ar=60:20:10:10,41.7mlmin‐1,Rh:2500h‐1,
RhMn:12500h‐1,RhFe:25000h‐1,RhMnFe:7580h‐1)
1
2
3
4
5
6
7
8
0
42 42 42 43 42 42 43
12
16 17 17 18 19 17
30
27 26 26 26 25 25
3
343335
0 50 100 150 200 250 300 350 400 450 500 550 600
Time on stream / h
CO
2
Ethyl acetate
Methyl acetate
Acetic acid
Acetaldehyde
n-Propanol
Ethanol
Methanol
Ethane
Methane
Methane
Methanol
Acetic acid
Ethanol
H
2
O yield
CO conversion
X
CO
and Y
Products
/ %
0.5
1.0
1.5
2.0
2.5
3.0
0.0
28 18 18 17 15 15 15
49
65 63 66 65 65 63
13
775656
2
444 556
CO
2
Methyl acetate
Acetic acid
Ethanol
Methanol
Ethane
Methane
Methane
Methanol
Acetic acid
Ethanol
H
2
O yield
CO conversion
X
CO
and Y
Products
/ %
1
2
3
4
5
6
7
8
9
0
35 33 32 32 32 32 31
4
4
11
11 12 13 12 12 11
14
15 16 16 17 18 17
17
21 21
20 22 22 24
Ethyl acetate
Methyl acetate
Acetic acid
Acetaldehyde
Ethanol
Methanol
Propylene
Ethylene
Propane
Ethane
Methane
Methane
Propylene
Ethane
Acetic acid
Acetaldehyde
Ethanol
H
2
O yield
CO conversion
X
CO
and Y
Products
/ %
Reaction network
analysis and kinetics
Reaction network
analysis and kinetics
Reaction network
analysis and kinetics
1
2
3
4
5
6
0
47 40 40 40 39 39 35
19 15 14 14 13 13
11
688777
6
610 11 11 12 12
13
6
10 10 10 12 12
17
CO
2
Ethyl acetate
Methyl acetate
Acetic acid
Propanal
Acetaldehyde
Ethanol
Methanol
Propylene
Ethylene
Propane
Ethane
Methane
Methane
Propylene
Ethane
Acetic acid
Acetaldehyde
H
2
O yield
CO conversion
Contact time variation
Partial pressure variation
X
CO
and Y
Products
/ %
Equilibration
Temperature variation
243 °C - 260 °C
A
RhMnB
RhFe
C
RhMnFeD
Rh
Reaction network
analysis and kinetics
5Stabilityandreproducibilityofcatalyticdata
54
After the initial formation process all catalysts show a slow but steady loss of catalytic
activity in terms of their CO consumption rate at reference conditions. The measured CO
consumption rate declines steadily over Rh/SiO2 with an activity loss rate
of -3·10-4 µmol (s gcat hTOS)-1 corresponding to a loss of 16% of activity in 14 days. After
contact time variation the CO consumption significantly drops. This might be a result of
previously achieved very high CO conversion levels but was not observed for any of the
modified catalysts.
The RhMn/SiO2 catalyst also loses activity after the initial phase. However, the rate is higher
than over pure Rh/SiO2 with a rate of -2·10-3 µmol (s gcat hTOS)-1. On the other hand, the
relative loss of activity is very similar to Rh/SiO2 with 19 % in 14 days. RhFe/SiO2 shows
very similar to RhMn/SiO2 with an activity loss rate and relative activity loss
of -1·10-3 µmol (s gcat hTOS)-1 and 16 % in 14 days, respectively. The doubly modified catalyst
RhMnFe/SiO2 shows slightly faster equilibration during the initial period than the other
catalysts. The steady activity loss after initial equilibration is slightly slower than over the
other modified catalysts with a rate of -8·10-4 µmol (s gcat hTOS)-1 and a relative loss of 15 %
in 14 days. Also, RhMnFe/SiO2 shows very stable selectivities from the beginning. This
behavior suggests a robust performance of RhMnFe/SiO2 and matches its superior
reproducibility discussed in chapter 5.3.
Considering product selectivities, all catalysts show a very stable behavior after the initial
equilibration phase throughout the kinetic study. That behavior suggest that the types of
active sites no longer change but rather their numbers decrease slowly over time on stream.
During an individual experiment between two reference points changes in CO conversion
are very small (<5 %). Combined with the high stability of product selectivities, it can be
concluded that the data obtained between 130 h and 550 h under reaction conditions are
sufficiently stable for kinetic evaluation for an extended period.
There is little information to be found in published data concerning the aging behavior of
Rh-based catalysts under industrially relevant CO hydrogenation conditions. In a recent
review by Luk etal., it is clearly stated that there is a major lack of information on the
catalytic stability so far.17 A kinetic study over RhMn catalysts describes an initial increase
of activity during the first 3 h before the catalysts shows a very stable behavior. However,
the maximum time on stream is 20 h. The maximum time on stream that was found is 45 h.85
In this study, Burch etal. describe an initial significant drop of CO consumption rate
followed by a slow further decrease. This observation is very similar to the data presented
in this chapter.
In conclusion, the time on stream behavior of all four catalysts must be divided in two
sections. During the initial equilibration and formation phase, the catalytic properties of the
5.2Catalyticstability
55
catalysts significantly evolve. During this time, kinetic data would be significantly
influenced by the unstable catalytic performance of the catalysts. Allowing at least 130 h for
equilibration is crucial for the acquisition of meaningful kinetic data. The reason for the
initial instability of the catalyst are likely to be found in structural rearrangement processes
as suggested by the amount of time necessary for stabilization. After equilibration, a much
slower loss of activity was observed. In contrast to the initial phase, selectivities are not
affected by this process. Here, the loss of activity might origin from a loss of active surface
area by sintering or poisoning. The rate of deactivation is too slow to have a significant
impact on kinetic data measured in this section.
Stability after cofeed and dropout experiments
For more understanding of the catalyst behavior after extreme reaction conditions and
different reaction pathways, dropout and cofeed experiments were performed in a separate
run (DinoRun29). In order to assess the data correctly, the impact of each experiment on
the catalyst performance must be checked.
At the end of the initial equilibration phase all catalysts reached a stable behavior (Figure
5.4). The first experiment performed in this run was a H2 dropout. The H2 flow was fully
replaced by N2. After 24 h reference conditions were applied again with 60 vol-% H2. This
experiment has very little impact on the catalyst performance. CO conversion over Rh and
RhMn are slightly reduced. The selectivities are not affected over any of the catalysts.
The same experiment was performed with CO as dropout component. In contrast to the H2
dropout, the CO dropout shows a strong impact on all catalysts. Generally, CO conversion is
enhanced and selectivities changed after returning to reference conditions. The catalysts
were allowed 60 h to equilibrate after CO dropout (TOS 410 – 470 h). RhFe reached a stable
state by end of this section. The equilibration of all other catalysts extends into the
beginning of the ethylene cofeed experiment. Therefore, for evaluation of the ethylene
cofeed experiment, reference data from after the cofeed should be used (TOS 590 – 630 h).
The following cofeed experiments with ethylene, acetaldehyde, propylene, and CO2 do not
change the catalytic behavior permanently. All results obtained at reference conditions
without cofeeds during this section are consistent for each catalyst. CO conversion is in
accordance with slow loss of activity over time on stream and the obtained selectivities in
the range of experimental accuracy.
5Stabilityandreproducibilityofcatalyticdata
56
Figure5.4StabilityofRh/SiO2(A),RhMn/SiO2(B),RhFe/SiO2(C)andRhMnFe/SiO2(D)duringdropoutandcofeed
experimentsplottedasCOconversionatrepeatedlymeasuredreferenceconditionsovertimeonstream(black
squares).Abovethefigures,theexperimentsperformedbetweenreferenceconditionsaredenoted.Thecolumns
showassociatedproductyieldsovertimeonstreamwiththeselectivitiesofthemajorproductsmarkedinwhite
numbers.(DinoRun29,54bar,260°C,H2:CO:N2:Ar=60:20:10:10,41.7mlmin‐1,Rh:2500h‐1,RhMn:12500h‐1,
RhFe:8330h‐1,RhMnFe:7580h‐1)
2
4
6
8
0
45 42 40 39 47 45 43 42 42 40
9
21 25 26
18 20 24 25 26 24
28
24 22 22
21 21 20 20 20 19
0 100 200 300 400 500 600 700 800 900 1000
Time on stream / h
CO2
Ethyl acetate
Methyl acetate
Acetic acid
Ethanol
Methanol
Ethane
Methane
Methane
Methanol
Acetic acid
Ethanol
CO conversion
X
CO
and Y
Products
/ %
2
4
6
8
0
53
41 40 39 46 41 39 38 38 35
12
766
766555
8
14 14 14
12 13 14 14 15 14
8
18 21 21
16 19 20 22 22 20
CO2
Ethyl acetate
Methyl acetate
Acetic acid
Propanal
Acetaldehyde
Ethanol
Methanol
Propylene
Ethylene
n-Butane
Propane
Ethane
Methane
Methane
Propylene
Ethane
Acetic acid
Acetaldehyde
CO conversion
1
st
and 2
nd
H
2
dropout
X
CO
and Y
Products
/ %
Equilibration
CO dropout
Ethylene
cofeed
Propylene
cofeed
CO
2
cofeedAcetaldehyde cofeed
2
4
6
8
10
12
0
37
32 33 33 36 33 32 32 31 30
6
333433222
9
10 10 11
988999
14
17 15 15
15 17 17 17 17 17
16
23 24 24
20 22 23 24 25 24
Ethyl acetate
Methyl acetate
Acetic acid
Propanal
Acetaldehyde
Ethanol
Propylene
Ethylene
n-Butane
Propane
Ethane
Methane
Methane
Propylene
Ethane
Acetic acid
Acet-
aldehyde
Ethanol
CO conversion
X
CO
and Y
Products
/ %
2
4
6
8
0
38 27 24 23 29 25 22 21 21 20
36
55 60 60 52 57 61 62 63 60
18
11 10 911 98887
CO
2
Ethyl acetate
Methyl acetate
Acetic acid
Ethanol
Methanol
Ethane
Methane
Methane
Methanol
Ethanol
CO conversion
X
CO
and Y
Products
/ %
A
RhMn
B
RhFe
C
RhMnFeD
Rh
5.3Reproducibility
57
5.3 Reproducibility
To gain insight on the catalytic behavior from different perspectives, a variety of catalytic
experiments were performed. Since some of them have an impact on the catalyst structure,
the experiments had to be contained in four separate catalytic runs. Evaluation of the
reproducibility therefore is important to have a measure of the comparability of the
different data sets.
During the equilibration phases of the four catalytic runs, Rh/SiO2 show very different
equilibration behaviors from a slight increase of CO conversion in Run 26 to a strong initial
drop of activity in Run 27 and 31 (Figure5.5A). However, after equilibration a CO
conversion close to the target conversion of ~5 % was reached in all cases. The observed
selectivities at the end of equilibration time differ significantly (Figure5.5B). The runs with
similar equilibration behavior of the CO conversion show similar selectivities as well.
There are several experimental and chemical reasons thinkable for this phenomenon.
Obvious experimental reasons can be ruled out. The used catalyst was always taken from
the same batch, the same reactor was used for the same catalyst and the in‐situ pretreatment
was the same throughout the runs. During the pretreatment of two runs (DinoRun27 and
DinoRun29) the pressurization routine had to be interrupted and restarted, but as these
two runs are not similar to each other the interruption does not seem to have a significant
effect. A possible chemical reason could be the very heterogeneous surface structure
described in chapter 4.2. The formation process of the structural feature might vary from
run to run.
The modification of Rh/SiO2with MnOx has revealed a stabilizing effect for small particles
leading to a very homogeneous surface structure. In agreement with the thoughts expressed
above, the uniform surface leads to very reproducible results RhMn/SiO2 (Figure5.5C,D).
The realized CO conversion is slightly above the target conversion of 5 % at reference
conditions. Due to the high activity of RhMn/SiO2 the amount of catalyst in the reactor is
very small. Weighing errors therefore are more severe and are likely to be the reason for
the slightly higher CO conversion in Run 29. The resulting selectivities however, are very
similar in each run with only the ratio of acetaldehyde and ethanol varying slightly.
The GHSV necessary to achieve the target conversion was estimated from tests with similar
catalyst samples prior to Run 26. For RhFe/SiO2 a too high activity was assumed. Therefore,
the loaded catalyst volume was too small and was corrected in the following runs.
5Stabilityandreproducibilityofcatalyticdata
58
Figure5.5Left:DevelopmentofCOconversionduringtheinitialequilibrationphasesmeasuredinfourcatalyticruns
overRh/SiO2(A),RhMn/SiO2(C),RhFe/SiO2(E)andRhMnFe/SiO2(G).Thedashedlinemarksthetargetconversion
of~5%.Right:ProductselectivitiesobtainedattheendofeachequilibrationphaseforRh/SiO2(B),RhMn/SiO2(D),
RhFe/SiO2(F)andRhMnFe/SiO2(H).Beloweachcolumn,COconversionandtotalCOconsumptionratearegiven.
(DinoRun26,27,29and31,54bar,260°C,H2:CO:N2:Ar=60:20:10:10,41.7mlmin‐1,Rh:2500h‐1,RhMn:12500h‐1,
RhFe:DinoRun26–25000h‐1;DinoRun27,29and31–8330h‐1,RhMnFe:7580h‐1)
0 25 50 75 100 125 150
0
2
4
6
8
10
12
14
16
18
CO conversion / %
Time on stream / h
RhMn
Run26 Run31
Run29
Run27
0 25 50 75 100 125 150
0
2
4
6
8
10
12
CO conversion / %
Time on stream / h
RhFe
Run26
Run31
Run29
Run27
Probably weighing
error
40 52 42 55
14 48
3
8251
11 14 13 13
10 14 17 15
0
20
40
60
80
100
5.04.04.04.9X
CO
/ % =
Product selectivities / %
CO
2
Ethyl acetate
Methyl acetate
Acetic acid
Propanal
Acetaldehyde
Ethanol
Propylene
Ethylene
Propane
Ethane
Methane
Run26 Run27 Run29 Run31
r / µmol s
-1
g
-1
=0.6 0.5 0.5 0.6
Acet-
aldehyde
Acetic acid
Propylene
Ethane
Methane
32 32 32 35
3233
3233
12 15 10 13
16 13 17 12
21 22 22 20
0
20
40
60
80
100
5.97.25.15.5XCO / % =
Product selectivities / %
CO
2
Ethyl acetate
Methyl acetate
Acetic acid
Propanal
Acetaldehyde
Ethanol
Propylene
Ethylene
Propane
Ethane
Methane
Run26 Run27 Run29 Run31
r / µmol s-1 g-1=3.5 3.2 4.6 3.8
Acet-
aldehyde
Acetic acid
Propylene
Ethane
Methane
Ethanol
18 29 27 32
63 50 53 49
612 11 12
5231
0
20
40
60
80
100
4.94.54.51.7XCO / % =
Product selectivities / %
CO
2
Ethyl acetate
Methyl acetate
Acetic acid
Ethanol
Methanol
Ethane
Methane
Run26 Run27 Run29 Run31
r / µmol s-1 g-1=2.1 1.9 1.9 2.1
Acetic acid
Methanol
Methane
Ethanol
42 39 42 42
17 21 20 21
26 27 24 24
4344
0
20
40
60
80
100
4.74.14.44.0XCO / % =
Product selectivities / %
CO
2
Ethyl acetate
Methyl acetate
Acetic acid
Acetaldehyde
Ethanol
Methanol
Ethane
Methane
Run26 Run27 Run29 Run31
r / µmol s-1 g-1=1.6 1.8 1.6 1.9
Acetic acid
Methanol
Methane
Ethanol
0 25 50 75 100 125 150
0
2
4
6
8
10
12
CO conversion / %
Time on stream / h
RhMnFe
Run26
Run31
Run29
Run27
0 25 50 75 100 125 150
0
2
4
6
8
10
12
14
16
18
CO conversion / %
Time on stream / h
Rh
Run26
Run31
Run29
Run27
AB
D
F
H
C
E
G
Different equilibration
behaviors and
selectivities
High reproducibility
High reproducibility
Catalyst amount too
small in first run
High reproducibility
5.3Reproducibility
59
For Run 27, Run 29 and Run31 all equilibration phases are very comparable (Figure5.5E).
After correction of the catalyst mass, the target CO conversion could be realized. The
observed product selectivities are very reproducible with only small deviations in methanol
and methane selectivity (Figure5.5F). Here, the reason for good reproducibility rather
seems to be a significant reduction of the product spectrum making the favored reaction
pathways much simpler as described in detail in chapter 7.7.
The data obtained for RhMnFe/SiO2 show very reproducible equilibration phases and
selectivities (Figure5.5G,H). It can be assumed that the combination of uniform surface
structure due to the stabilizing effect of Mn oxide and the less complex product spectrum
due to Fe addition leads to a very robust catalyst.
The obtained results cannot be compared to published data since typically no information
concerning the reproduction of experiments is given.
In conclusion, the modified catalysts RhMn/SiO2, RhFe/SiO2 and RhMnFe/SiO2 are highly
reproducible. This also proves the robust functionality of the catalytic test setup and
experimental procedures. However, Rh/SiO2 with its heterogeneous surface structure and
complex product spectrum does not yield particularly comparable results in different runs.
The smaller reproducibility of the catalyst can likely be related to structural changes that
are a function of time on stream and composition of the atmosphere. Therefore, this catalyst
is the most interesting sample for investigations targeting an understanding of dynamic
restructuring processes of Rh-based catalysts in general. Future experiments should also
assess the sensitivity of the performance towards changes in the catalyst synthesis
conditions such as different precursors, temperatures, and time ranges for calcination and
H2 treatment as well as the reaction start up procedure.
5Stabilityandreproducibilityofcatalyticdata
60
61
6 Reaction network and kinetics
for STE over Rh/SiO2
In this chapter, all catalytic and kinetic effects observed for the catalyst with the simplest
composition – Rh/SiO2 – are discussed in detail. First, product selectivities at reference
conditions are introduced. The catalytic behavior as function of different parameters such
as time on stream, reactant partial pressures, and contact time is discussed in detail. Cofeed
and dropout studies add necessary information to derive a plausible reaction network
which is proposed in the last section of this chapter. As the evaluation of apparent activation
energies did not provide vital information on the reaction network, they are shown and
discussed in Appendix D.1.
6.1 Product spectrum and selectivities
The very broad product spectrum is one of the main challenges to accurately investigate the
syngas to ethanol reaction. Besides the target product ethanol, other oxygenates such as
methanol, higher alcohols, aldehydes, and acids are formed. Also, methane and other
hydrocarbons add to a very complex mixture of compounds to be quantified. However, if
measured the broad diversity of products also offers the unique opportunity to unravel the
mechanistic details, and to decipher all the different classes of reactions that potentially
contribute to the overall catalyst performance.
Over the pure Rh/SiO2 catalyst, the main carbon containing product is methane with a
selectivity of 40 % at reference conditions and ~5 % CO conversion after equilibration of
the catalyst (Figure6.1A). It should be noted, however, that water was formed in much
higher quantities. In this case, for each mol of converted CO roughly 0.85 mol water were
produced. While water formation is not relevant for benchmarking the catalyst
performance, it is important to include water concentrations in considerations regarding
catalyst behavior and stability.
Ethane also is a major product. With decreasing selectivities, paraffins with carbon numbers
up to five were found in measurable amounts. It was tested if the paraffin formation follows
an Anderson-Schulz-Flory (ASF) distribution, which would imply a consecutive build up of
chain length with a chain growth probability being independent of the chain length. A linear
trend was found for the C3 to C5 paraffins providing a chain growth probability of 0.43
(Figure6.1B). Ethane slightly deviates from the distribution which already indicates the
6ReactionnetworkandkineticsforSTEoverRh/SiO2
62
high complexity of the STE mechanism featuring interactions of many different product
types. Since methane formation is not part of the chain growth mechanism, it typically
deviates from an ASF distribution and was not considered for the fitting.26,146 As for all other
product types, only linear paraffins are formed.
Figure6.1AProductselectivitiesobtainedoverRh/SiO2after170htimeonstreamatreferenceconditions.
BAnderson‐Schulz‐Floryfittingwithchaingrowthprobabilities(α)fordifferentproductgroupsoverRh/SiO2after
170hTOSatreferenceconditions.(DinoRun26,54bar,260°C,H2:CO:N2:Ar=60:20:10:10,41.7mlmin‐1,2500h‐1)
In the case of olefins, propylene shows the highest selectivity with 8 %. Here, the
concentration clearly deviates from a theoretical ASF distribution. Compared to n-butene
and n-pentene, propylene shows an increased and ethylene a reduced concentration. This
trend is comparable to what has been reported for other carbon chain building reactions
such as the Fischer-Tropsch reaction. It is usually attributed to a higher ability of ethylene
to form coordinative bonds to transition metal centers which might lead to further
consumption.29 To further elucidate this, ethylene and propylene were both included in the
cofeed studies discussed in section 6.5 of this chapter.
Acetaldehyde and acetic acid are the major products among oxygenates. Concentration of
higher aldehydes follow an ASF distribution with a very similar chain growth probability as
paraffins and olefins. Acetic acid is the only carbonic acid that was quantified. Methyl and
ethyl acetates were also found as minor products. Although much more ethanol was
produced than methanol, methyl and ethyl acetate selectivities are comparable. This
suggests an easier formation of methyl acetate. Under the applied reaction conditions, it
cannot be excluded that esterification reactions take place without contact to a catalytically
active surface.
0123456
-5.0
-4.5
-4.0
-3.5
-3.0
-2.5
-2.0
-1.5
-1.0
-0.5
0.0
0.5
log (c / n)
Carbon number
Paraffins
= 0.43
Aldehydes
= 0.46
Olefins
= 0.44
Alcohols
= 0.46
40
14
4
8
1
11
10
0
20
40
60
80
100
4.9XCO / % =
Methane
Propylene
Ethane
Acetic acid
Acetaldehyde
Ethanol
Product selectivities / %
CO2
Ethyl acetate
Methyl acetate
Acetic acid
Butanal
Propanal
Acetaldehyde
Propanol
Ethanol
Methanol
n-Pentene
n-Butene
Propylene
Ethylene
n-Pentane
n-Butane
Propane
Ethane
Methane
H
2
O
Rh
r / µmol s-1 g-1=0.6
Propane
Ethylene
Propanal
AB
Cha i ngrowth
probabilities
suggestcommon
chaingrowthsite
6.2Initialformationphase
63
Alcohol selectivities are very small. However, ethanol, n-propanol, and n-butanol only
roughly follow an ASF distribution while methanol clearly drops out. Methanol is not part
of the chain growth process and is therefore not part of the distribution.
The concentrations of all higher products follow ASF distributions with very similar chain
growth probabilities. This fact suggests that the chain growth step might be the same or
similar for all different product types. However, deviations from the ASF distributions for
the C2 compounds of all product types indicate the high complexity and interconnectivity
of the STE reaction network which cannot be solely described by an ASF mechanism.
Comparison to published data shows that some features described in this study agree with
literature reports. CO2 is not formed over Rh/SiO2 in considerable quantities.84 Methane
usually is described as the main product. More detailed comparison to literature is difficult.
Different results were published for similar systems as used in this study. For example,
ethanol selectivities vary from 0 %86,88 up to 16 %87 using catalysts with 1.5–3 wt-% Rh on
SiO2 at CO conversion levels of 2–10 % (compare Table2.1). The significant difference of
catalytic results underlines the important roles of catalyst preparation, treatment, reaction
conditions and online analytics. Moreover, published data are often not very specific
considering the large product spectrum. Minor products such as acetates or higher
aldehydes usually were not measured. And often there is no discrimination between olefins
and paraffins85 or products are combined into groups such as “C2+ hydrocarbons” or “other
oxygenates”.99,123
In conclusion, the obtained product spectrum and reaction mechanism is highly complex
and challenging to analyze experimentally. However, evaluation of minor products is crucial
for mechanistic considerations. Therefore, the detailed product analysis performed in this
study provides important information to complement the knowledge offered from
literature.
6.2 Initial formation phase
During the first 100 – 130 h time on stream, catalyst activity as well as product selectivities
strongly changed. While the total catalyst activity is shortly discussed in chapter 5.2.2, this
section focuses on the evolution of the individual reaction rates. Evaluating the trends in
product formation in the initial phase of catalyst equilibration contributes key information
on important reaction paths and their relations to each other.
Evaluation of the equilibration behavior for the different product formation rates gives
insight into related product formation pathways. All paraffins formation rates for example
6ReactionnetworkandkineticsforSTEoverRh/SiO2
64
were stable or slightly increasing over the initial period (Figure6.2). Since all paraffins
showed the same behavior, it seems that the chain growth probability did not change.
The concentrations of ethylene and propylene as examples for olefins behaved significantly
different. After 80 h time on stream, the olefin formation rates almost doubled, before they
stabilized. The same behavior was found for the aldehydes and acetic acid, suggesting
closely related pathways for the formation and/or the consumption of olefins, aldehydes,
and acids in secondary reactions.
Figure6.2EvolutionofnormalizedCOconsumptionandnormalizedproductconcentrationsduringinitial
equilibrationperiodoverRh/SiO2.(DinoRun26,54bar,260°C,H2:CO:N2:Ar=60:20:10:10,41.7mlmin‐1,2500h‐1)
Methanol and propanol concentrations show a maximum in the first 20 h time on stream,
while after the first increase ethanol concentration was rather stable. This behavior clearly
0.0
0.5
1.0
Acetic acid
Acetaldehyde
Normalized concentrations
Propanal
Acetaldehyde
Propanal
Acetic acid
0 20 40 60 80 100 120 140
0.0
0.5
1.0
Time on stream / h
H
2
OH
2
O
0.0
0.5
1.0
Ethanol
Propanol Methanol
0.0
0.5
1.0
Propylene
Ethylene
0.0
0.5
1.0
Methane Ethane Propane
0.0
0.5
1.0
-c
CO
Time on stream behavior suggests
related formation pathways
6.3InfluenceofCOandH2partialpressures
65
underlines the high complexity and interconnection of all compounds in the reaction
network.
In literature, there is only very limited information regarding the catalyst stability (compare
chapter 5.2). Gilhooley etal. observed decreasing paraffin yields and increasing oxygenate
yields before a steady-state was reached. The time scale was much shorter as only the first
4 h were investigated.126 Chuang etal. described a similar behavior but with a strong
increase of methanol formation rate over Rh/TiO2.147 Consideration regarding the origin of
this behavior could not be found.
In summary, the evaluation of the initial formation phase of the catalyst can give insight into
related product formation pathways. While paraffins and alcohols show separate trends,
the synchronous evolution of olefin, aldehyde and acid concentrations suggests a common
reaction pathway. Structural modification during the first hours under reaction conditions
might be crucial for this behavior. This context will be further elucidated in a discussion on
the impact of reactant partial pressures on formation rates in section 6.3.3 and finally used
in chapter 6.7 to derive a plausible reaction network.
6.3 Influence of CO and H2 partial pressures
To evaluate the catalyst behavior at different feed conditions, reactant partial pressures
were varied in a 3x6 matrix. Total flow, total pressure and temperature were kept constant.
In the following sections, it will be shown that H2 partial pressure control the total reactant
consumption rate, but CO partial pressure largely controls individual product formation
rates. The results reveal the CO insertion mechanism as plausible C-C-coupling mechanism
for all product types and two major effects of CO on reaction pathways: enhancing C-C
coupling via CO insertion and impeding the catalysts hydrogenation ability. It will be
discussed, how this effect might be related to a structural transition of the catalyst surface
induced by CO partial pressure.
Consumption rates
The reaction of CO and H2 consumption to varied feed conditions gives first evidence on the
adsorption behavior of CO and H2 under reaction conditions. All shown consumption rates
were calculated based on reaction products. The CO consumption rate increases almost
linearly with increasing H2 partial pressure, resulting in a reaction order of 0.9 (Figure
6.3A). The overall availability of adsorbed H* species therefore seems very important for
CO activation. On the other hand, CO partial pressure has very small influence on the total
CO consumption rate. The respective reaction order is 0.1. This suggests that the CO
6ReactionnetworkandkineticsforSTEoverRh/SiO2
66
coverage is high even at CO partial pressures as low as 1.6 bar. This finding is in accordance
with literature which reports full coverage at CO partial pressures of 200 Pa at 40 °C.82
An even more distinct behavior was observed for the H2 consumption rate (Figure6.3B).
H2 consumption is independent on CO pressure (reaction order 0). The obtained data does
not show evidence for site competition of CO and H2. In literature, FTIR studies show no
effect of H2 or H2O partial pressures on the CO coverage.148,149 Also DFT studies assume
different adsorption sites for CO and H2 due to the very different sizes of CO and H.116
At this point it is important to note that the total rate for syngas conversion of Rh/SiO2 is
solely controlled by the availability of H* species and therefore by H2 partial pressure. The
insensitivity of the rate towards CO partial pressure leads to a large range of CO conversion
levels ranging from 3 % for high CO and low H2 partial pressures to 26 % for the opposite
case (Figure6.3C). For this reaction, the catalyst activity should always be described by
specific rates rather than CO conversion since it is strongly dependent on reaction
conditions. The conversion of H2 on the other hand is independent from feed conditions and
calculated to be ~4 % over the whole range of tested partial pressures (Figure6.3D).
Figure6.3COandH
2
consumptionratesasfunctionofCOandH
2
partialpressures(A,B)andresultingconversionof
COandH
2
asfunctionofCOandH
2
partialpressuresoverRh/SiO
2
(C,D).(DinoRun26,54bar,260°C,H
2
:CO:N
2
:Ar=
30‐60:2.5‐20:10‐57.5:10,41.7mlmin
‐1
,0.5gcat,2500h
‐1
)
6.3InfluenceofCOandH2partialpressures
67
Product formation rates
Because of the diverse product spectrum, a detailed investigation of the product formation
rates is of highest importance. In this section the effects of partial pressures on individual
formation rates are discussed (Figure6.4 and Figure6.5). The respective figures showing
product selectivities over partial pressures can be found in the Appendix (FigureA22 and
FigureA23 in Appendix D.2).
The methane formation rate ranges from 0.1 to 0.4 µmol s-1 g-1 under different feed
conditions, resulting in methane selectivities of 30 – 80 %. H2 partial pressure has almost
linear impact as observed for the total CO consumption rate. Therefore, the impact of H2 on
the selectivity is small. On the other hand, increasing CO partial pressure leads to a strong
decrease of methane formation. High CO pressure clearly has an inhibiting effect on
methane formation. From a mechanistic point of view, this could either be the result of
competitive reaction preferred at high CO pressures or an inhibition of the hydrogenation
ability of the catalyst by high CO pressures.
The patterns obtained for higher paraffins look different. For ethane and propane
formation, the positive effect of H2 partial pressure is less pronounced at low CO partial
pressures. For butane and pentane this applies over the whole considered CO partial
pressure range. Ethane formation indicates a maximum at medium CO pressures. This
maximum exemplary visualizes the complex interplay of multiple effects further discussed
in the next section.
Olefin formation rates show very clear trends. H2 partial pressure has no or negative effect
which is related to consecutive hydrogenation of the unsaturated products. At the same
time, the strongly positive effect of CO partial pressure might result from a combination of
an increased C-C-coupling ability and an impeded hydrogenation ability.
Methanol formation shows the most positive reaction towards high H2 pressures. This
behavior supports the assumption of methanol formation taking place before C-O bond
cleavage. The full hydrogenation of molecularly adsorbed CO* seems only feasible at high
H2 and low CO partial pressures. For the formation of higher alcohols, the effect of H2 partial
pressure is much less distinct. Increasing CO partial pressures lead to decreased methanol
and ethanol formation but a maximum for propanol.
For more oxidized compounds such as aldehydes and acids, in all cases increasing CO partial
pressures increase formation rates. These products profit from CO insertion as well as
impeded hydrogenation. Formation rates for acetates are small over Rh/SiO2 which
corresponds to small alcohol concentrations. The behavior of the formation rates over feed
conditions might be interpreted as a combination of acetic acid and the respective alcohols.
6ReactionnetworkandkineticsforSTEoverRh/SiO2
68
Figure6.4ProductformationratesofC1andC2productsasafunctionofCOandH
2
partialpressuressorted
horizontallybycarbonnumberandverticallybyoxidationstateoftherespectivefunctionalgroup.(DinoRun26,
Rh/SiO
2
,54bar,260°C,H
2
:CO:N
2
:Ar=30‐60:2.5‐20:10‐57.5:10,41.7mlmin
‐1
,0.5gcat,2500h
‐1
)
6.3InfluenceofCOandH2partialpressures
69
Figure6.5ProductformationratesofC3toC5productsasafunctionofCOandH
2
partialpressuressorted
horizontallybycarbonnumberandverticallybyoxidationstateoftherespectivefunctionalgroup.(DinoRun26,
Rh/SiO
2
,54bar,260°C,H
2
:CO:N
2
:Ar=30‐60:2.5‐20:10‐57.5:10,41.7mlmin
‐1
,0.5gcat,2500h
‐1
)
6ReactionnetworkandkineticsforSTEoverRh/SiO2
70
Formation of CO2 is very slow over all feed conditions. Therefore, no meaningful
considerations regarding its origin can be made. Its formation is inversely related to H2
partial pressure which gives reason to suggest a water-gas-shift reaction. However, other
formation pathways such as steam reforming or decarboxylation pathways could lead to
similar results.
Apparent reaction orders
The calculation of apparent reaction orders gives a more quantitative understanding of the
product formation at varying feed conditions. Apparent reaction orders were calculated
according to a power law equation (equation (9) in chapter 3.6.1).
Evaluation of H2 reaction orders at different CO partial pressures shows a severe impact of
CO partial pressure on the obtained H2 reaction orders (Figure6.6). Generally, reaction
orders of H2 increase with increasing CO partial pressure. This finding is consistent with the
hypothesis that high CO pressures impede hydrogenation reactions. While for some
products, e.g. methane, this influence is rather small, for others it even includes a transition
from negative H2 reaction orders to positive reaction orders (higher paraffins, higher
aldehydes). Ethanol and propanol show a unique behavior. Their H2 reaction orders reach
a maximum at ~8 bar CO partial pressure, demonstrating the particularly complex
formation pathway for higher alcohols. This behavior complicates a detailed comparison
with literature where reaction orders were not obtained at the exact same reaction
conditions. Also, this complexity poses a challenge for kinetic modelling.
Figure6.6ReactionordersofH
2
forhydrocarbons(A)andoxygenates(B)asafunctionofCOpartialpressure.
(DinoRun26,54bar,260°C,H
2
:CO:N
2
:Ar=30‐60:2.5‐20:10‐57.5:10,41.7mlmin
‐1
,0.5gcat,2500h
‐1
)
0246810121416
-2
-1
0
1
2
reaction order n
H2
p
CO
/ bar
Methane
Ethane
Propane
Butane
Pentane
Propylene
Ethylene
Butene
AB
H
2
reactionorderforoxygenates
H
2
reactionorderforhydrocarbons
0246810121416
p
CO
/ bar
Propanol
Ethanol
Acet-
aldehyde
Propanal
Butanal
Acetic acid
high p
CO
low p
CO
high p
CO
low p
CO
6.3InfluenceofCOandH2partialpressures
71
As CO partial pressure poses a significant impact on obtained reaction orders of CO and H2,
two CO pressure regimes were defined and evaluated separately. CO pressures above 5 bar
are declared “high CO pressure” and below 5 bar “low CO pressure”. The subdivision was
defined according to specific features in the selectivity plots, e.g. the position of the
maximum ethane selectivity. H2 pressures only have small influence on the reaction orders.
A summary of all obtained reaction orders shows that all reaction orders of CO and H2
change for product formation reactions (Table6.1). However, for the reactant consumption
reaction no such an influence is observed.
Table6.1ApparentreactionordersofCOandH2calculatedonthebasistoCOhydrogenationoverRh/SiO2in
differentCOpartialpressureregimes.
GroupRate
lowpCO
pCO<5bar
highpCO
pCO>5bar
m(CO)a,bn(H2)a,cm(CO)a,dn(H2)a,e
COconsumption0.10.8 0.10.9
H2consumption0.00.9 0.00.9
H2group
(negativeCOand
medium/highH2order)
Methanol‐0.32.2 ‐0.43.9
Methane ‐0.11.0 ‐0.31.3
Ethanol‐0.20.7 ‐0.50.9
CO/H2optimumgroup
(mediumCOandH2
order)
n‐Propanol0.30.5 ‐0.21.0
Ethane0.50.3 0.11.1
n‐Propane0.60.0 0.30.8
COgroup
(highCOorderandlow
ornegativeH2order)
Acetaldehyde0.9 ‐0.1 0.40.7
n‐Butane0.9 ‐0.3 0.60.5
n‐Pentane1.2 ‐0.7 0.70.4
Butanal1.5 ‐0.6 0.90.6
Propanal1.7 ‐0.5 1.20.6
Propylene2.0 ‐1.5 1.20.1
1‐Butene2.1 ‐1.4 1.6 ‐0.3
Ethylene2.3 ‐1.5 1.80.0
Aceticacid2.6 ‐0.6 0.80.3
aReactionordersofCOandH2weredeterminedbyfittingapower‐lawrateexpression.
bReactionordersofCOwerecalculatedfromdataobtainedinthelowCOpartialpressureregimeof1.35–4.05bar.
cReactionordersofH2werecalculatedfromdataobtainedataconstantCOpartialpressureof1.35bar.
dReactionordersofCOwerecalculatedfromdataobtainedinthehighCOpartialpressureregimeof5.4–10.8bar.
eReactionordersofH2werecalculatedfromdataobtainedataconstantCOpartialpressureof10.8bar.
6ReactionnetworkandkineticsforSTEoverRh/SiO2
72
Regarding their reaction orders, groups can be defined containing all products that are
preferred at high H2 partial pressure (H2 group) or high CO partial pressures (CO group)
and one transition group with products that show maximum formation rates at medium CO
and H2 pressures.
The C1 products methanol and methane and the target product of the STE reaction belong
to the H2 group with negative CO reaction orders and strongly positive H2 reaction orders.
It is worth mentioning here that the similar behavior of methane and ethanol poses a major
challenge for optimizing reaction conditions towards a high ethanol yield. High ethanol
formation rates are always accompanied by increased formation of methane which is
usually highly unwanted for industrial operation. The next higher alcohol propanol already
belongs to the transition group, presumably because the chain elongation step requires
higher CO partial pressures.
With exception of methane, all other paraffins (blue in Table6.1) show positive CO reaction
orders. With growing carbon number, the reaction order of CO increases. The higher the
carbon number, the more clearly the product formation rates profit from high CO partial
pressures. This is counterintuitive if a formation of higher hydrocarbons from the coupling
of CHx* species was assumed. If this were the case, long chained paraffins would be expected
to profit from a high CHx* surface concentration and therefore behave very similarly to
methane. If a CO insertion mechanism is assumed, an increased CO partial pressure could
lead to more CO insertion and therefore promote chain growth. As methane formation does
not involve C-C-coupling, its reaction order is negative.
The apparent H2 reaction orders for paraffins decrease with increasing carbon number. For
low CO partial pressure, the values range from 1.0 for methane to -0.7 for pentane. At high
H2 partial pressure therefore short chain lengths are preferred, likely because of fast
hydrogenation and desorption before the next C-C-coupling step can take place. Ethane and
propane formation rates are maximized at medium CO and H2 pressures since for their
formation CO insertion as well as hydrogenation ability is required. If CO pressures are too
low, the formation of the necessary C2 intermediate by a C-C-coupling reaction – e.g. via CO
insertion – is slow and selectivity decreases in favor of methane formation. If CO pressures
are too high, the C2 intermediate is consumed for the formation of longer chains or the final
hydrogenation of the intermediate to form ethane and propane might be impeded.
The apparent CO reaction orders for olefins (green in Table6.1) in are strongly positive in
the high CO partial pressure regime and even higher at low CO pressures. H2 reaction orders
are strongly negative for low CO pressures and around 0 for high CO partial pressures.
These extreme values are likely a result of combining the two effects of CO. Especially the
impeded hydrogenation ability is likely important for olefin formation.
6.3InfluenceofCOandH2partialpressures
73
Aldehydes and acetic acid (orange and purple in Table6.1) also belong to the CO group. In
principal they behave similarly to olefins but with less negative or even positive reaction
orders of H2. Hydrogen availability might be more important for aldehydes because
desorption requires a hydrogenation step as opposed to olefins that presumably desorb
without hydrogenation.
Published reaction orders are generally in agreement with the finding in this study. For CO
consumption, positive values were reported for H2 reaction order and 0 or even small
negative values for CO reaction orders.83,123,125,149 The large range for the reaction orders for
paraffins depending on their carbon number was also reported by these studies. Mao etal.
published very similar results for olefins as well, confirming the considerations of this study
regarding olefins with more differential data.83 For oxygenated products, the values from
these studies scatter more broadly. Reaction orders of CO and H2 range between -1.52–-0.13
and 0.74–1.52, respectively. Our obtained results lie well inside the range provided by
literature. However, we could show the importance of evaluating apparent reaction orders
at different CO partial pressures and therefore provide a much more complete picture of the
influence of reactant partial pressures on product formation in this complex reaction.
Likely, the strong impact of reaction conditions is the reason for contradicting results in
literature.
The results discussed in this section allow further mechanistic considerations. For all
product types, the impact of CO partial pressure on their individual product formation rates
is much more distinct and diverse than the effect of H2 partial pressure. However, we have
shown that H2 partial pressure controls the overall CO consumption rate. Hence, surface H*
species seem mainly necessary to activate CO* which is abundantly available on the surface.
H-assisted C-O bond cleavage has been reported repeatedly as initial step for the formation
of CHx* species which is the key intermediate for the formation of a variety of other
products. The interpretation of our data is consistent with this hypothesis. However, while
H2 controls the total consumption rate, CO largely controls individual product formation
rates. Increasing CO partial pressure leads to distinct changes of the product spectrum. It
promotes C-C-coupling – likely via CO insertion mechanism – and impedes consecutive
hydrogenation of intermediates.
The massive impact of CO on product formation rates might be interpreted as a result of
structural reorganization of the surface depending on CO partial pressure. Almost full CO
coverage must be assumed already at very low CO pressures. The assignment of different
catalytic functionalities leading to oxygenate or hydrocarbon formation to different facet of
metallic Rh nanoparticles does not allow an explanation of such an behavior. Therefore, to
explain the tremendous impact of CO partial pressure on the reactivity, structural changes
of the catalyst dependent on CO pressure must be considered. Literature suggests processes
6ReactionnetworkandkineticsforSTEoverRh/SiO2
74
such as the formation of surface Rh (sub)carbonyl species under CO containing
atmospheres, although under different atmospheric conditions.64,150,151 Following this
hypothesis, the formation of Rh (sub)carbonyl species at elevated CO pressures could
promote CO insertion while hydrogenation reactions are rather catalyzed by metallic Rh
which dominates at high H2 and low CO partial pressures. Thereby, the finding that olefins
and aldehydes behave similarly over reactant partial pressures relates to their similar
behavior during the initial formation phase (see section 6.2). In this context, structural
adjustments as outlined above during the first hours under reaction conditions might cause
the shift in formation rates. The transition from metallic Rh to carbonyl structures would
lead to more olefin and aldehyde formation which is exactly what was observed in this time
range. The impact of reaction atmosphere on structure and reactivity is further discussed
and related to other results of this study in chapter 8.2 in the General Discussion.
In conclusion, the evaluation of consumption and formation rates over a large range of
reactant partial pressures revealed the major impact of CO partial pressure on catalytic
reactivity. High CO partial pressures have positive impact on chain elongation – most likely
via CO insertion mechanism – and impede hydrogenation reactions. This effect could be
related to a possible transition of the catalyst surface from a more metallic state in the low
CO pressure regime towards a more (sub)carbonylic species at high CO partial pressures.
6.4 Contact time variation
The evolution of product concentration over contact time gives insight about how rates
change over the course of reaction. The data were obtained by variation of the total flow in
six steps with constant gas composition. In this section, olefins, acetaldehyde and acetic acid
will be identified as potential reactive intermediates in the reaction network.
Total CO consumption – expressed here as the negative change of CO concentration –
follows an almost linear increase over contact time (Figure6.7A). The achieved CO
conversion is 3.2 – 17 %. The maximum H2 conversion thereby is 16 %. Up to these
conversion levels no limitation can be observed. Methane concentration increases in an
almost linear manner over contact time. This is a typical behavior for primary products that
are not consumed in secondary reactions.
All C2 products increase over contact time (Figure6.7B). Therefore, no clear indication of
consecutive reaction pathways can be determined. Consumed products would eventually
decrease. However, differences can be found between the products. Ethane formation rate
increases most distinctly and linearly. As stated before, ethane is potentially formed by
consecutive hydrogenation of ethylene. However, ethane can already be found at very low
conversions. Ethanol shows a comparable behavior to ethane but with much smaller rates.
6.4Contacttimevariation
75
Figure6.7TheconsumptionofCOexpressedasthenegativechangeofCOconcentrationandtheproduct
concentrationsofmethane(A),C2products(B),C3products(C),C4products(D),acetates(E),andCO2(F)as
functionofcontacttime.Fortherealizationofdifferentcontacttimes,GHSVwasvariedbyadjustingthetotalflows
intherangeof8.3–58.3mlmin‐1.(DinoRun26,54bar,260°C,H2:CO:N2:Ar=60:20:10:10,0.5gcat,500‐3500h‐1)
Acetaldehyde and acetic acid behave very similarly. At small CO conversions, their
concentrations are in the same range as ethane, but with increasing CO conversion the
formation rates decrease. This behavior could potentially be an indication of a consecutive
reaction but also a result of reaction kinetics. Ethylene concentration shows more distinct
behavior and is almost stable over the whole range of contact times.
The observed C3 and C4 products show a very comparable behavior to their C2 relatives
concerning the trends over contact time (Figure6.7C,D). That confirms the idea of a
common reaction network with continuously growing carbon chains. The concentrations of
different products are more similar than for the C2 products. Methyl and ethyl acetate
follow the increasing trends of their building blocks methanol and ethanol (Figure6.7E,F).
0 50 100 150 200 250
0.000
0.005
0.010
0.015
0.020
0 50 100 150 200 250
0.000
0.002
0.004
0.006
0.008
0.010
0 50 100 150 200 250
0.00
0.01
0.02
0.03
0.04
0 50 100 150 200 250
0
1
2
3
4
5
0 50 100 150 200 250
0.0
0.1
0.2
0.3
0.4
0.5
0 50 100 150 200 250
0.00
0.02
0.04
0.06
0.08
0.10
Product concentration / vol-%
Contact time / s
Butanal Ethyl acetate
Methyl acetate
CO2
Butane
Butene
Methanol
Product concentration / vol-%
Contact time / s
Methane
Ethane
Acetic acid
Acetaldehyde
Ethylene Ethanol Propanol
Propanal
Propane
Propylene
-cCO
DEF
C2products
COconsumption&methane C3products
Ethylene
showsplateau
ABC
Acetates
C4products Methanol&CO
2
6ReactionnetworkandkineticsforSTEoverRh/SiO2
76
All consumptions and product concentrations scale almost linear up to about 50 s contact
time. Hence, in this range the reactor can be assumed to operate in a differential regime,
supporting the results for apparent activation energies and reaction orders which were
evaluated with a contact time of approximately 39 s at standard flow conditions.
The depiction of concentrations against contact time does not allow easy estimation on the
development of selectivities. The interaction of different products is often more visible in
selectivity – conversion plots. As expected, methane selectivity is stable over the whole
range of CO conversion (Figure6.8A). Ethylene, acetaldehyde, and acetic acid selectivities
decrease (Figure6.8B) in favor of slightly increasing ethanol and strongly increasing
ethane selectivity. Plausible consecutive reactions would be the hydrogenation of ethylene
to ethane, CO insertion into ethylene or coupling with a CHx* fragment to form C3 products,
water addition to ethylene to form ethanol, hydrogenation of acetaldehyde and acetic acid
to form ethanol and/or ethane.
Figure6.8Productselectivitiesofmethane(A),C2products(B),andC3products(C),C4products(D),acetates(E),
andCO2(F)asfunctionofCOconversion.Fortherealizationofdifferentconversionlevels,GHSVwasvariedby
adjustingthetotalflowsintherangeof8.3–58.3mlmin‐1.(DinoRun26,54bar,260°C,H2:CO:N2:Ar=60:20:10:10,
0.5gcat,500‐3500h‐1)
0 5 10 15
0
10
20
30
40
50
0 5 10 15
0
5
10
15
20
25
30
0 5 10 15
0
1
2
3
4
5
0 5 10 15
0.0
0.5
1.0
1.5
2.0
0 5 10 15
0
2
4
6
8
10
0 5 10 15
0.0
0.5
1.0
1.5
2.0
Product selectivities / %
CO conversion / %
Methane
Ethane
Acetic acid
Acetaldehyde
Ethylene
Ethanol
Propanol
Propanal
Propane
Propylene
Product selectivities / %
Butanal
CO conversion / %
Ethyl acetate
Butane
Butene
Methyl acetate
CO
2
Methanol
ABC
C2products
Methane C3products
DEF
Acetates
C4products Methanol&CO
2
6.5Cofeedstudies
77
C3 products here show a more complex behavior. Propylene and propanal show a maximum
in selectivity at medium CO conversions of 7 – 10 % (Figure6.8C). This might be the result
of the formation of C3 products as secondary products from a C2 intermediate and
consecutive consumption in further reactions. Here, the analogous consecutive reactions to
their C2 relatives are thinkable.
The considerations for C3 products are analogously applicable for C4 products (Figure
6.8D). Selectivities of the acetates follow the alcohol selectivities (Figure6.8E). However,
the increase in acetate selectivity cannot fully account for the decrease of acetic acid
selectivity. Consecutive esterification therefore cannot be the only cause of the decreasing
acetic acid selectivity.
A systematic variation of the contact time could not be found in literature to compare with
the presented findings. Hence, the discussed data represents the first comprehensive set of
data for the influence of contact time on the formation of all products over long-term
equilibrated Rh/SiO2 recorded at high pressure.
In this section, potential intermediate products were identified from selectivity –
conversion plots. Ethylene, acetaldehyde, acetic acid, and their analogues might undergo
consecutive reactions. However, the concentration – contact time plots did not provide clear
indication on the consumption of the named compounds. Therefore, cofeed experiments are
necessary to probe the reaction network and will be discussed in the next section.
6.5 Cofeed studies
Cofeed studies were performed to probe reaction pathways with compounds that are likely
intermediates in the reaction network. The results in this section will suggest that chain
growth occurs only via CO insertion and hydrogenation of intermediates proceeds at a
separate hydrogenation site.
Cofeed studies were performed by replacing the H2 flow partially or fully with a flow from
a pre-mixed blend of cofeed compound and H2. To measure a baseline for the cofeed purity
a bypass measurement was performed for each cofeed compound. Blank reactivity on the
reactor walls and inert filling as well as the temporal response of the setup to stepwise feed
changes was examined in blank experiments. The blank reactor was filled with the catalyst
support Davisil (silica) and steatite which was used as the filling material up- and
downstream of the catalyst bed.
During the ethylene cofeed some ethane was observed in the outlet gas of the blank reactor
and the bypass line (Figure6.9A). The bypass line is a 1/8” stainless steel tube without
packing and is not heated. Therefore, it can be assumed that ethane was already present in
6ReactionnetworkandkineticsforSTEoverRh/SiO2
78
the gas cylinder containing the mixture. However, the ethane concentration only makes up
1 % of the nominal ethylene concentration and can be neglected in the evaluation process.
Also, the blank reactor does not show significant conversion of ethylene.
For propylene, the propane concentration present in the gas cylinder is much higher
(Figure6.9B). Around 30 % of the cofed C3 compound is propane instead of propylene. No
propylene conversion was observed in the blank reactor. The high propane concentration
however must be considered for evaluation of propylene cofeed experiments with catalyst.
Figure6.9Blankexperimentsforcofeedsofethylene(A),propylene(B),acetaldehyde(C)andCO2(D)inablank
reactorfilledwithsilicaandsteatite(filledsymbols)andthebypassline(emptysymbols)showingconcentrationsof
thecofeedcompoundsandtheirhydrogenationproductsasafunctionoftimeonstream.(DinoRun33,54bar,260
°C,H2:CO:N2:Ar=60:20:10:10+CO2cofeed,41.7mlmin‐1,0.5gDavisil)
160 180 200 220 240 260 280 300 320
0.00
0.02
0.04
0.06
0.08
0.10
400 420 440 460 480 500 520
0.000
0.002
0.004
0.006
0.008
0.010
280 300 320 340 360 380 400 420 440
0.00
0.05
0.10
0.15
0.20
500 510 520 530 540 550 560 570 580
0
1
2
3
4
5
Concentration / vol-%
Time on stream / h
Ethylene
Ethane
0.058 vol-%
cofeed
0.032 vol-%
cofeed
Propylene
Propane
0.006 vol-%
cofeed
0.011 vol-%
cofeed
Concentration / vol-%
Time on stream / h
Concentration / vol-%
Time on stream / h
0.07 vol-%
cofeed
0.14 vol-%
cofeed
Acetaldehyde
Ethanol
CO
2
Concentration / vol-%
Time on stream / h
1.8 vol-%
cofeed
3.6 vol-%
cofeed
Acetaldehyde cofeed – blank experiment
AB
C
Propylene cofeed – blank experiment
Ethylene cofeed – blank experiment
D
CO
2
cofeed – blank experiment
Propane
already
present in
gas cylinder
Acetaldehyde
missing in
blank rector
blank reactor
(silica + steatite)
bypass
6.5Cofeedstudies
79
In the acetaldehyde mixture no impurities were found in the bypass line. In the outlet gas of
the blank reactor less aldehyde was found than expected (Figure6.9D). The concentration
only slowly reached a steady-state at ~65 % of the cofeed concentration. Acetaldehyde
seems to adsorb on steatite and/or silica. Small amounts of ethanol were found, but not
enough to account for the missing acetaldehyde concentration and the ethanol
concentration appeared to be independent from acetaldehyde concentration. Ethanol might
be formed by hydrogenation in contact with silica and steatite or at the reactor wall.
However, only very small ethanol concentrations were found which do not scale with the
acetaldehyde cofeed concentration. Thus, the data obtained for acetaldehyde cofeed with
catalysts can be evaluated accordingly after a long equilibration time.
Evaluation of the CO2 cofeed experiments revealed the same CO2 concentrations in bypass
line and blank reactor. No impurities were found. CO2 is not converted on support or filling
material under the chosen reaction conditions.
Ethylene cofeed
Over Rh/SiO2, the cofed ethylene concentration is fully converted forming mostly ethane by
hydrogenation as well as propanal and propanol via CO insertion (Figure6.10).
Figure6.10TheconsumptionofCOexpressedasthenegativechangeofCOconcentrationandtheproduct
concentrationsofmethane(A),C2products(B),andC3products(C)asfunctionofethyleneinletconcentration.
NumbersinAdenotechronologicalorderofmeasureddatapoints.(DinoRun29,54bar,260°C,H
2
:CO:N
2
:Ar=
60:20:10:10+ethylenecofeed,0.5gcat,2500h
‐1
)
0.00 0.02 0.04 0.06
0.0
0.2
0.4
0.6
0.8
1.0
0.00 0.02 0.04 0.06
0.00
0.02
0.04
0.06
0.08
0.10
0.00 0.02 0.04 0.06
0.000
0.002
0.004
0.006
0.008
0.010
Concentration / vol-%
Ethylene inlet concentration / vol-%
Methane
Ethane
Acetic acid
Acetaldehyde
Ethylene
Ethanol Propanol
Propanal
Propane
Propylene
-c
CO
ABC
C2 products
CO consumption & methane C3 products
CO insertion
Hydrogenation
No CHx
coupling
CH
x
+ C
2
H
y
6ReactionnetworkandkineticsforSTEoverRh/SiO2
80
Ethylene cofeed has no clear effect on the CO consumption rate (Figure6.10A). The slight
decrease of CO consumption – represented by the negative change of CO concentration –
during the first part of the experiment results from slow equilibration process after the
previously performed CO dropout experiments (section 6.6).
Ethane concentration increases most distinctly upon ethylene cofeeding (Figure6.10B).
Hydrogenation of ethylene to form ethane therefore seems to be fast over Rh/SiO2. Ethanol
concentration does not change, indicating that water addition to ethylene to form ethanol
is not a relevant reaction pathway. Acetic acid shows a negative trend upon ethylene cofeed
concentration which might at least partly be a result of propyl acetate formation due to
increased propanol concentrations.
Among the C3 products, the increase of propanol concentration is most prominent (Figure
6.10C). The increasing propanol and propanal concentrations indicate that ethylene can
adsorb and undergo CO insertion. From this experiment, it cannot clearly be stated if
propanol is formed directly or via hydrogenation of propanal. Propane or propylene
concentrations do not increase upon ethylene cofeed. For similar experiments, increasing
C3 hydrocarbon concentrations have been reported for catalyst systems that are assumed
to catalyze C-C-coupling of olefins with CHx*, such as Ru,152 Fe and Co.29 Therefore, the
results of this study indicate that a coupling reaction of ethylene with a surface CHx* species
over Rh/SiO2 is unlikely. This finding contradicts common assumptions on the reaction
network.25,83 However, no clear experimental evidence has been found in published data
that support the CHx coupling mechanism over Rh-based catalysts.
Propylene cofeed
In contrast to ethylene, the cofed propylene is not fully converted (Figure6.11B). Propane
formed via hydrogenation is the main product. No direct chain growth to butane viaCHx
addition was observed. Again, water addition to the olefin under formation of propanol was
not observed.
The observed propylene conversion was calculated to be 40 % taking the cofed propane
into account. As stated above, the propylene mixture already contains high concentrations
of propane, which should be considered here. This behavior goes along with the deviation
of ethylene and propylene form an ASF distribution as discussed in section 6.1. The theory
behind AFS distribution includes equal reactivity independent from chain length. In this
case, ethylene shows a much higher reactivity than propylene leading to lower ethylene and
higher propylene concentration than expected.
In contrast to ethylene, propylene apparently does not undergo CO insertion to form the C4
aldehyde or alcohol (Figure6.11C). Butanol has been calibrated but concentrations were
6.5Cofeedstudies
81
always below the detection limit. Therefore, a small amount of butanol might be formed but
not in measurable quantities.
Figure6.11TheconsumptionofCOexpressedasthenegativechangeofCOconcentrationandtheproduct
concentrationsofmethane(A),C3products(B),andC4products(C)asfunctionofpropyleneinletconcentration.
NumbersinAdenotechronologicalorderofmeasureddatapoints.(DinoRun29,54bar,260°C,H2:CO:N2:Ar=
60:20:10:10+propylenecofeed,0.5gcat,2500h‐1)
The propylene cofeed demonstrated that propylene is less reactive than ethylene over
Rh/SiO2 which explains a non-ASF distribution of products. While the hydrogenation
reaction to propane was shown, chain elongation to form C4 products was not observed.
0.000 0.004 0.008 0.012
0.0
0.2
0.4
0.6
0.8
1.0
0.000 0.004 0.008 0.012
0.0000
0.0004
0.0008
0.0012
0.0016
0.000 0.004 0.008 0.012
0.000
0.004
0.008
0.012
0.016
Concentration / vol-%
Butanal
Butene
Butane
Propylene inlet concentration / vol-%
Methane
Propanol
Propanal
Propane
Propylene
-c
CO
12
3
4
ABC
C3 productsCO consumption & methane C4 products
CO insertion into
propylene not
observed
No CHx
coupling
CH
x
+ C
3
H
y
Hydrogenation
6ReactionnetworkandkineticsforSTEoverRh/SiO2
82
Acetaldehyde cofeed
Acetaldehyde is rapidly converted into ethanol (Figure6.12B). Consecutively enhanced
ethyl acetate formation consumes acetic acid. This leads to less methyl and probably also
less propyl acetate formation which in turn leads to higher methanol and propanol
concentrations escaping from the reactor.
Figure6.12TheconsumptionofCOexpressedasthenegativechangeofCOconcentrationandtheproduct
concentrationsofmethane(A),C2products(B),C3products(C),andacetates(D)asfunctionofacetaldehydeinlet
concentration.NumbersinAdenotechronologicalorderofmeasureddatapoints.(DinoRun29,54bar,260°C,
H
2
:CO:N
2
:Ar=60:20:10:10+acetaldehydecofeed,0.5gcat,2500h
‐1
)
The cofed acetaldehyde is almost fully hydrogenated to ethanol over Rh/SiO2. The observed
hydrogenation is faster by orders of magnitude compared to the blank experiment and can
0.00 0.04 0.08 0.12 0.16
0.0
0.2
0.4
0.6
0.8
1.0
0.00 0.04 0.08 0.12 0.16
0.00
0.03
0.06
0.09
0.12
0.15
0.00 0.04 0.08 0.12 0.16
0.000
0.002
0.004
0.006
0.008
0.010
0.00 0.04 0.08 0.12 0.16
0.000
0.002
0.004
0.006
0.008
0.010
Concentration / vol-%
Methane
-c
CO
1
2
3
4
Ethane
Acetic acid
Acetaldehyde
Ethylene
Ethanol
Concentration / vol-%
Propanol
Propanal
Propane
Propylene
Acetaldehyde inlet concentration / vol-%
Ethyl acetate
Methyl acetate
Methanol
AB
C
C2 products
CO consumption & methane
C3 products
D
Methanol & Acetates
Hydrogenation
Acetate
formation
6.5Cofeedstudies
83
therefore be considered a real catalytic effect. However, this finding is counterintuitive
considering the high acetaldehyde vs very low ethanol selectivities at reference conditions
without cofeed (section 6.1). That means that acetaldehyde produced inside the catalyst bed
is much less likely to be hydrogenated than acetaldehyde introduced into the reactor head.
Studies of the Fischer-Tropsch reaction reported similar findings that cofed intermediates
are much more likely to undergo secondary reactions than intermediates formed inside the
catalyst bed (e.g. ethylene).153 The authors suggest water or other reaction products to be
responsible for suppressing secondary reaction in lower parts of the catalyst bed.
Nonetheless, the experiment shows the ability of Rh/SiO2 to hydrogenate aldehydes and
therefore suggests alcohols being formed from secondary hydrogenation of aldehydes.
Some acetic acid is consumed by ethyl acetate formation due to the strongly increased
ethanol concentration demonstrating that acetates are formed as a function of the
concentrations of the respective alcohols and acetic acid (Figure6.12D). Over this acetate
formation mechanism, all alcohol concentrations are interconnected. Increasing the
concentration of one alcohol enhances the formation of its acetate which causes less acetic
acid being available for the formation of other acetates and therefore higher amounts of
other alcohols escaping from the reactor. In this case, increasing methanol and propanol
concentrations are related to a decreasing availability of acetic acid. Propyl acetate has not
been quantified. However, the increase of ethyl acetate concentration does not fully cover
the decrease of acetic acid concentration. In contrast, acetaldehyde cofeed could be
expected to promote acetic acid formation as it has been reported to be formed from
adsorbed acetaldehyde with a surface hydroxyl group.99,121 On the other hand, it is likely
that acetaldehyde cofeeding does not lead to the necessary surface intermediate perhaps
because of rapid hydrogenation at the beginning of the catalyst bed.
Acetaldehyde has a small negative impact on total CO consumption which might be a result
of slightly changing H2 concentrations during the cofeed (Figure6.12A). As for the olefin
cofeeds, methane formation is not affected.
Most of the C3 products show a decreasing trend (Figure6.12C). However, the diverse
interactions of other reaction products such as alcohols and acetates make it difficult to
extract clear reaction pathways from this information.
The acetaldehyde cofeed experiment confirms major proposed reaction pathways. Ethanol
can be formed efficiently by hydrogenation of acetaldehyde as a secondary product. Acetate
formation is strongly dependent of the respective alcohol and acetic acid concentrations.
Highly interconnected reaction pathways highlight the importance to quantify all reaction
products including acetates.
6ReactionnetworkandkineticsforSTEoverRh/SiO2
84
CO
2
cofeed
Industrially offered synthesis gas often contains significant amounts of CO2. Therefore, it is
important to investigate its impact on reactivity of catalysts although it is not a relevant
reaction product. Throughout the experiment CO2 does not adsorb or react over Rh/SiO2.
Total CO consumption is not affected and none of the product formation rates are changed
by CO2 cofeeding in the chosen concentration range. It should be noted that at much higher
CO2 concentrations or different reaction conditions this behavior might change.
Figure6.13TheconsumptionofCOexpressedasthenegativechangeofCOconcentrationandtheproduct
concentrationsofmethaneandCO
2
(A),C2products(B),andC3products(C)asfunctionofCO
2
inletconcentration.
NumbersinAdenotechronologicalorderofmeasureddatapoints.(DinoRun29,54bar,260°C,H
2
:CO:N
2
:Ar=
60:20:10:10+CO
2
cofeed,0.5gcat,2500h
‐1
)
Mechanistic information derived from cofeeds
In this study, cofeeding ethylene, propylene and acetaldehyde addressed the questions
derived from the selectivity – conversion behavior discussed in section 6.4. The performed
cofeeding experiments yield five major conclusions for Rh/SiO2:
i) Rapid hydrogenation of alkenes to alkanes and aldehydes to alcohols is possible.
ii) CO insertion is a step for chain growth.
iii) Chain growth via coupling of alkene with surface CHx-fragments is unlikely.
iv) Ethylene can desorb and adsorb to be further converted in the chain growth
process via CO insertion.
v) Propylene is much less reactive than ethylene.
01234
0
1
2
3
4
01234
0.00
0.02
0.04
0.06
0.08
01234
0.000
0.002
0.004
0.006
0.008
Concentration / vol-%
CO
2
inlet concentration / vol-%
Methane
Ethane
Acetic acid
Acetaldehyde
Ethylene
Ethanol
Propanol
Propanal
Propane
Propylene
-c
CO
1
2
3
4
CO
2
ABC
C2 products
CO
2
& CO cons. & methane C3 products
No CO
2
conversion
in this concentration
range
6.5Cofeedstudies
85
The combination of those findings suggest that hydrogenation and chain growth take place
at separate sites that are differently accessible for intermediates.
Cofeed experiments with ethylene and propylene were mainly reported in Fischer-Tropsch
or heterogeneous hydroformylation research. Some studies include ethylene cofeeds over
Rh catalysts and CO hydrogenation conditions. Chuang etal. observed hydrogenation of
ethylene to ethane and CO insertion to propanal/propanol.147 Incorporation of ethylene into
chain growth to higher hydrocarbons was not observed for Rh/SiO2 but for Rh/TiO2. These
findings are consistent with the present findings on Rh/SiO2.
In gas phase hydroformylation studies, the reaction conditions resemble CO hydrogenation
conditions with olefin cofeed at lower temperatures and higher olefin concentrations. It was
found that hydrogenation of ethylene largely dominates over unpromoted supported
metals (Pd, Ru, Rh).151,154 Huang etal. reported reactivity data from carbonyl complex
derived catalysts that suggest structure sensitivity for CO insertion and insensitivity for
hydrogenation reactions.155 A similar hypothesis was published by Hanaoka etal. proposing
large particles being responsible for hydrogenation and small particles for CO insertion.154
These claims are consistent with the fact that hydroformylation functionality is usually
observed in homogeneous catalysis where Rh complexes are the active species. The
attempts to apply heterogeneous catalysts therefore often target supported molecular Rh
species to imitate the homogeneous processes.156,157 In recent research, supported Rh
phosphides were applied and it was shown that catalysts with single Rh sites in a stable
Rh2P structure outperform samples on which the phosphide partly decomposed to metallic
Rh. 158 These described literature findings support the conclusions of the presented cofeed
experiments as well as the previously discussed hypothesis of the strongly structure
dependent reactivity of Rh/SiO2 at those reaction conditions (chapter 6.2 and 6.3).
However, a connection of those studies to reactivity in STE catalysis has not been described
so far.
The addition of acetaldehyde to H2 and different CO:H2 mixtures has been investigated by
Burch and Petch.127 They found that acetaldehyde hydrogenation is suppressed by presence
of CO. Acetic acid concentrations were not discussed. Significant amounts of hydrocarbon
formation upon acetaldehyde addition were reported as opposed to the results in the
present study. As a different experimental approach for cofeeding was used in their study,
this might also be an effect of slightly changing reaction conditions (e.g. CO:H2 ratio) during
the cofeeding.
C-C-coupling of olefins with CHx* surface groups was not observed contracting a common
assumption in STE related literature (compare chapter 2.3.3). However, this finding is
consistent with calculated surface coverages that reveal very low CHx* surface
concentrations.121 This would make the coupling event of two CHx* fragments or the
6ReactionnetworkandkineticsforSTEoverRh/SiO2
86
coupling of a CHx* fragment with a higher CxHy* species extremely unlikely. Also, in Fischer-
Tropsch research considerations have been published reasoning CO insertion as C-C-
coupling mechanism not only for oxygenate but also for hydrocarbon formation.136,159
The CO insertion mechanisms is also indicated by previously discussed findings like similar
chain growth probabilities for oxygenates and hydrocarbons (section 6.1) and common
trends for aldehydes and olefins upon varied CO partial pressure (section 6.3.2). Thus, it can
be concluded that the CO insertion mechanism is the only relevant C-C-coupling mechanism
over Rh/SiO2 at the presented industrially relevant reaction conditions.
6.6 Dropout experiments
Dropout experiments are suitable for investigating the long-term effect the presence or
absence of one of the reactants poses on the catalyst performance. It will be shown that the
absence of H2 does not affect reaction rates significantly. In contrast, a prolonged absence
of CO has a tremendous impact on reaction rates for several days further supporting the
idea of a CO-induced transformation of the catalyst structure under reaction conditions.
First, a H2 dropout experiment was performed. The H2 flow was stopped and replaced by
additional N2 flow to keep the GHSV constant. The catalyst was then kept under the CO/N2
atmosphere for 24 h before the reference syngas feed with 60 vol-% H2 was reapplied.
Although the used catalytic testing setup is not optimized for transient experiments, slow
dynamics of the reaction can be investigated. However, the following points should be
considered when discussing the data. Due to high pressures in the mixing section and the
size of the setup, complete feed equilibration is comparably slow. However, the atmosphere
is exchanged in one hour except of trace amounts that need up to 24 h to be purged out of
the setup. The reactant concentrations at reactor inlet over time on stream during the
dropout experiments are shown in FigureA24 in Appendix D. Another point to consider is
the slow sampling of an online GC. One chromatogram takes 40 min to be taken. With 5
sampling lines that means that each reactor is sampled once in three to four hours. Faster
processes cannot be resolved. To investigate faster processes, a setup equipped with
suitable switching valves and minimized dead volume in combination with time efficient
online analytics such as mass spectrometry should be used. However, the experiments
presented in this section give detailed information about slow non-steady-state processes.
When the H2 flow was stopped, the reaction stopped quickly (Figure6.14A). The immediate
drop of reaction rates shows that the catalyst is not able to store H* on the surface for a
longer time. Only small amounts of acetic acid have been detected for some hours, possibly
due to slow desorption from the surface or adsorption on the filling material or tubing.
6.6Dropoutexperiments
87
After 24 h, the H2 flow was reapplied and reference syngas conditions restored. Compared
to before the H2 dropout, formation rates only changed very little. Most rates were
immediately stable, only alcohols needed some hours to stabilize. CO conversion was
slightly reduced but selectivities did not change upon a longer absence of H2. This finding
suggests that the presence or absence of H2 in the atmosphere does not have an impact on
the structure of the working catalyst.
Figure6.14NormalizedconcentrationsofthemainproductsintheupperpartandCOconversion(blacksquares)
andproductyieldsinthelowerpartasfunctionoftimeonstreamduringaH2dropout(A)andCOdropout
experiment(B).Whitenumbersdenoteselectivitiesofthemainproducts.(DinoRun29,54bar,260°C,syngas
H2:CO:N2:Ar=60:20:10:10,fordropoutexperimentstherespectivereactantwasreplacedbyadditionalN2,0.5g
cat,2500h‐1)
Subsequently, a CO dropout experiment was performed following the same approach.
Directly after stopping the CO flow, ethanol and methane formation increased. It is likely,
that by this time small amounts of CO were still being fed into the reactor as explained
above. The reaction conditions therefore resemble very low CO partial pressures. Under
such conditions, methane and ethanol formation rates are strongly preferred as discussed
in section 6.3.2. However, a peak of methane and ethanol concentration during similar
experiments in more transient systems has been reported in literature as well.125 Ethanol
formation then stopped quickly, while methane and small amounts of ethane were further
produced over the whole 24 h.
AB
CO dropout
H
2
dropout
0.5
1.0
1.5
40 40 39
666
14 15 14
21 21 21
39 46 41
6
7
6
14
12
13
21
16
19
Normalized concentrations
Methane
Ethane
Ethanol
Acetaldehyde
Ethylene
syngas
CO + H2
CO/N2
syngas
CO + H2
Acetic
acid
300 320 340 360 380 400 420
0
1
2
3
4
CO conversion
and product yields / %
Time on stream / h
CO
2
Ethyl acetate
Methyl acetate
Acetic acid
Propanal
Acetaldehyde
Propanol
Ethanol
Methanol
1-Butene
Propylene
Ethylene
n-Butane
Propane
Ethane
Methane
CO conversion
Acet-
aldehyde
Ethane
Methane
Methane Ethane
Ethanol
Acetaldehyde
Ethylene
syngas
CO + H
2
H2/N2
syngas
CO + H2
Acetic acid
360 380 400 420 440 460 480 500 520340
Time on stream / h
6ReactionnetworkandkineticsforSTEoverRh/SiO2
88
After restoring reference conditions, some product formation rates were significantly
enhanced compared to before the CO dropout, e.g. methane, ethane, and ethanol. Other
products, such as acetic acid and acetaldehyde, showed comparable rates as before. In this
case, an effect of slow exchange of the gas atmosphere is unlikely, since small inaccuracies
of CO concentration would not have significant impact on reaction kinetics with ~20 vol-%
CO content in the atmosphere. Products with enhanced rates are all fully hydrogenated,
suggesting an enhanced hydrogenation ability of the catalyst after the CO dropout. All rates
showed the tendency to approach their original values over time on stream but after two
days the original catalyst performance was not restored. Combined with the slightly
increasing rate of ethylene, the behavior after CO dropout resembles the behavior during
the initial equilibration phase (section 6.2). The different behavior of the previously defined
three product groups therefore can be applied to the dropout experiment as well.
The enhanced formation rates result in an increase of CO conversion. On first sight, a CO
dropout or “additional H2 treatment” could be a strategy for in‐situ regeneration of the
catalytic activity. However, the yields of valuable oxygenates stays almost unchanged and
only the typically valueless products methane and ethane are enhanced. Therefore, such a
procedure is not economically feasible.
The prolonged absence of CO has a tremendous impact on the catalyst performance. The
time range necessary for equilibration suggests structural readjustments similar to the
formation process during the initial equilibration phase. This finding further supports
considerations on CO having a crucial impact on the Rh structure under reaction conditions
as previously proposed based on the influence of CO pressure on reactivity in section 6.3.2.
Burch and Petch have published switching experiments from reaction mixture to H2 and
back at relevant reaction conditions.125 Enhanced methane and ethanol formation after CO
dropout were interpreted to origin from hydrogenation of reactive surface intermediates.
After switching back to reaction mixture, paraffins were found to leave the reactor first
followed by increasing concentration of acetaldehyde. The time range includes 3 min after
switching without comparison to the steady-state before the switch. Therefore, no
interpretation in terms of catalyst structure is possible.
Basu etal. describe the reversible transformation from Rh(CO)2 carbonyls to metallic Rh on
Rh/SiO2 and Rh/Al2O3 catalysts.150,160 The structural changes were monitored by infrared
spectroscopy. Specific band for Rh carbonyls and hydroxyl groups were evaluated upon
changing gas atmospheres. Pure hydrogen led to formation of metallic particles while CO
containing atmosphere led to formation of Rh carbonyls. The observed dependency of
catalytic reactivity on gas phase composition in the present study therefore might be related
to a dynamic formation and destruction of carbonyl species in an equilibrium with metallic
particles. However, the reaction conditions for those spectroscopic studies were far from
6.7Deducedreactionnetwork
89
relevant conditions and catalysts with unusually high Rh loadings of 10 wt-% were used.
Similar results were obtained by Kruse etal. reporting CO-induced morphological changes
of Rh crystallites investigated via field ion microscopy.63
In conclusion, the presented dropout experiment demonstrates the dynamic structural
properties of Rh/SiO2. While presence or absence of H2 does not significantly affect catalyst
performance at reference reaction conditions, the prolonged absence of CO causes
enhanced formation rates over several days. Time range and similarity to the initial
formation phase of the catalyst indicate structural rearrangements, potentially involving a
dynamic equilibrium of metallic Rh particles and (sub)carbonyl surface species.
Based on literature evidence, we suggest that the reversible transformation of metallic Rh
and Rh carbonyl structures could be an integral part of high pressure STE. We propose that
the transformation happens during drop out, but also (gradually) during STE at high CO
pressures. The STE reaction therefore could be based on two coupled reaction networks,
one acting on a more metallic surface, the other on a more carbonyl containing surface.
6.7 Deduced reaction network
With the information obtained from the experiments discussed in the previous sections, a
plausible reaction network is proposed which reflects the observed trends and effects.
Two adsorbed CO* surface species with different reactivities dependent on CO partial
pressures are a key feature of the derived reaction network which becomes apparent from
mainly two observations. First, structural changes during the initial equilibration phase and
after a CO dropout experiment reveal the presence of CO in the atmosphere as a structure-
inducing compound for the working catalyst. Second, rising CO partial pressure has major
control on individual product formation rates without affecting total CO consumption rate.
In the proposed reaction network, the different types of CO are assigned as COA and COB
(Figure6.15). COA can be hydrogenated and undergo C-O bond cleavage to form more
complex surface intermediates. COB represents a surface species that is formed at high CO
partial pressures and long Time on stream and is crucial for the CO insertion step to form
an oxygenate intermediate. Although the precise assignment of specific structural motifs to
single surface steps is beyond the scope of this study, the presented results and literature
findings suggest Rh (sub)carbonyl structures to be formed which is related to surface steps
that require COB. Consequently, COA is rather related to metallic Rh nanoparticles.
Methanol formation occurs before C-O bond cleavage takes place. This feature has been
reported117,118 and is consistent with our data, e.g. an extraordinarily high H2 reaction order.
H* availability is crucial to enable full hydrogenation of molecular COA before C-O
6ReactionnetworkandkineticsforSTEoverRh/SiO2
90
dissociation can take place. This mechanism is also consistent with too low methanol
concentrations to fit into an ASF distribution of alcohols as it does not correspond to the
statistical laws related the ASF concept. Over Rh/SiO2 the direct hydrogenation of molecular
COA to form methanol is slow as H* availability is limited.
Figure6.15PlausiblereactionnetworkoverRh/SiO2.Arrowsmarkedinorangeweredirectlydeducedfromcofeed
studies.Differentpatternsindicatedifferentadsorptionsitesincludingthehydrogenadsorptionsite(fullgrey),a“low
COpressure”adsorptionsiteforCO*andotherC‐containingsurfacecompoundspotentiallyonmetallicRh
nanoparticles(stripes),a“highCOpressure”adsorptionsiteprovidingCO*forCOinsertion(dotted),andaseparate
hydrogenationsite(white).
H* availability is not only vital for methanol formation but also generally important for COA
activation in form of a H*-assisted C-O bond dissociation. This step reflects the linear
dependency of CO and H2 consumption rates on H2 partial pressure as demonstrated in
chapter 6.3 and is supported by experimental and theoretical studies in literature.75,116,118,148
The formed O* species is assumed to be hydrogenated to OH* and desorbs as. The CHxA
species on the other hand is an important key intermediate for all further reactions.
Generally, three thinkable reaction pathways originate from the CHxA intermediate. Results
of this study suggest that only two of them are relevant. First, further hydrogenation of CHxA
forms methane. As methane is the main product over this catalyst, this reaction step can be
considered comparably fast. However, with increasing CO partial pressure the formation
rate decreases. Potentially, this is a result of less H* and COA being available due to metallic
Rh nanoparticles gradually transforming into a COB rich carbonyl structure.
CO
A
CO
A
:highcoverage
alreadyatverysmallp
CO
CO
B
:onlyformed
atincreasedp
CO
H*
H
2(g)
CO
(g)
CO
B
Acetates
+Alcohols
(g)
CH
3
COOH
A
CH
3
COOH
(g)
Aceticacid
CH
3
CO
A
CH
3
CHO
(g)
Acetaldehyde
+CO
B
+OH*
CH
x
CH
xA
C
2
H
4(g)
Ethylene
+AcOH
(g)
H
x
CO
A
OH
A
H
2
O
(g)
CH
xA
+H*
CH
4(g)
Methane
CH
3
OH
(g)
Methanol
+CO
B
C
2
H
5
CHO
A
C
2
H
5
CHO
(g)
Propanal
C
2
H
x
CH
xA
C
3
H
6(g)
Propylene
+CO
B
+CH
xA
+H
*
+H*
Methylacetate
+H
*
+H*
+H* +H*
+CH
xA
‐H
2
O
‐H
2
O
+H*
+AcOH
(g)
Ethylacetate
+2H*
C
x
H
y
CH
2
OH
(g)
Alcohol
C
x
H
y
CHO
C
C
x
H
y
CHO
(g)
Aldehyde
C
x
H
y(g)
Olefin
C
x
H
y+2(g)
Paraffin
C
x
H
yC
Hydrogenationsite(s)
CH
3
CH
2
OH
(g)
Ethanol
CH
3
CHO
(g)
Acetaldehyde
+AcOH
(g)
Acetates
6.7Deducedreactionnetwork
91
The CHxA fragment can also undergo CO insertion of a “high pressure” COB species. This
pathway becomes more likely and more competitive with increasing CO partial pressure
explaining increasing selectivities for longer carbon chains and decreasing selectivities for
C1 products. The general ability of Rh/SiO2 to catalyze CO insertion was observed in
ethylene cofeed experiments.
The third thinkable reaction pathway from CHxA would be the coupling of two CHxA
fragments or a CHxA fragment with a higher hydrocarbon fragment increasing its chain
length. However, the results from ethylene and propylene cofeeds did not indicate a direct
C-C-coupling to higher hydrocarbons. The exclusion of this pathway to higher hydrocarbons
further agrees with the positive effect of CO partial pressure on their formation rates and
the increasing CO reaction orders with increasing carbon numbers. If CHxA coupling was the
relevant formation pathway, a high concentration of CHxA would be beneficial which most
likely is rather case at high H2 and low CO partial pressures. From theoretical studies, it was
reported that the surface coverage of CHx* is low due to high coverage of molecular CO* on
a metallic Rh surface117 or Rh6 cluster.121 Thus, a direct coupling of two CHx* intermediates
would be unlikely which supports our reaction network.
After successful CO insertion, the acetaldehyde intermediate CH3COA is formed. In this case,
three further reaction pathways were identified from the kinetic study and cofeed
experiments. First, hydrogenation of the fragment directly leads to acetaldehyde
desorption. Acetaldehyde cofeed experiments suggest that it does not readsorb on this
specific site from the gas phase.
Another proposed reaction pathway is the reaction with surface hydroxyls to form acetic
acid. This reaction step is rarely addressed in literature but has been suggested by
theoretical studies.99,121 We observed that the formation rate of acetic acid strongly
decreases with H2 partial pressure, likely because of faster removal of OH* species by water
formation. Based on our observations of acetate formation, we conclude that the presence
of acetic acid causes a complex interconnectivity of all alcohol concentrations due to their
common subsequent reaction that converts them into acetates.
The necessary step to form hydrocarbons via a CO insertion mechanism is the C-O bond
cleavage of the acetaldehyde intermediate to form ethylene. The mechanism is assumed to
work analogously to the first C-O dissociation forming the CHxA fragment. This step could
not be directly observed from cofeed studies. However, it is likely that the necessary
intermediate could not be formed by cofeeding acetaldehyde because it does not readsorb
or because it is rapidly hydrogenated to ethanol is the first part of the catalyst bed. However,
all indications discussed above point out that CO insertion must be assumed to account for
C-C-coupling and therefore chain growth. The transformation of the oxygenate intermediate
into the hydrocarbon fragment therefore is an inevitable part of the network. The closely
6ReactionnetworkandkineticsforSTEoverRh/SiO2
92
related formation pathway of aldehydes and olefins is consistent with the assignment of
both into the same product group after evaluation of the equilibration phase, partial
pressure evaluation and dropout experiments.
Analogous to the CHxA fragment, the adsorbed ethylene fragment undergoes two of three
thinkable pathways. Ethylene can desorb and readsorb at the same site. This ability reflects
its unique behavior, e.g. the flat profile of concentration over contact time and its too low
concentration to fit into an ASF distribution. Again, C-C-coupling can take place to form C3
products in another CO insertion – C-O bond cleavage cycle as observed in ethylene cofeed
studies. CHxA coupling does not take place as discussed above.
Alcohols and paraffins are proposed to form at an extra hydrogenation site. If it is the same
or several different sites cannot be determined from this study. The separation of the
hydrogenation site from the main “chain growth” pathways is mainly based on the ability of
the catalyst to hydrogenate cofed propylene while its reintroduction into the chain growth
cycle was not possible. However, hydrogenation might be possible on both sites for already
adsorbed intermediates. Alcohols and paraffins have both shown similar high apparent
activation energies and similar CO and H2 reaction orders. It is therefore likely that the final
hydrogenation to form these products is a limiting step for their formation over Rh/SiO2.
The enhanced formation rates after CO dropout might be a result of an increased abundance
of activated H* or suitable hydrogenation sites on a potentially metallic Rh surface.
Literature provides many studies addressing possible mechanisms for the formation of
different product types by CO hydrogenation. In particular, Rh catalysts have been
extensively investigated by theoretical studies. DFT calculations have been performed on
different metallic Rh surfaces (100),115 (111),116 (211),117,118 and (553)120 as well as
(un)supported Rh clusters.121,122,161 However, none of the computational methods allowed
for a possible transformation of metallic Rh into Rh (sub)carbonyl structures.
Some of the reaction paths in the proposed model are consistent with features that have
been reported in literature based on experimental and theoretical studies:
i) Methanol formation via direct hydrogenation of molecular CO has been
established by isotope scrambling experiments over Rh/TiO2 by Takeuchi etal.
and has been widely accepted since.162
ii) H-assisted C-O bond cleavage is commonly described as first step towards C2
oxygenate formation. The step is also proposed based on experimental data.148
iii) The CO insertion mechanism proposed by Xu etal. for oxygenate formation is
also widely accepted, however mostly described as final step of oxygenate
formation.163
6.7Deducedreactionnetwork
93
One feature that is subject to ongoing discussion is the relation of aldehydes and alcohols in
the network. The hydrogenation of aldehydes as intermediates for alcohol formation has
been reported based on cofeed studies127 and is often assumed from other catalytic data.83,98
However, the independent formation of aldehydes and alcohols from different
intermediates has also been proposed.78,128,129 These studies are mostly based on isotope
tracing experiments at very low pressure and might not correspond to relevant reaction
conditions. While it cannot be completely excluded that both mechanisms take place in
parallel, the presented data strongly suggests alcohols being formed by hydrogenation of
aldehydes.
Different from common understanding of STE in literature, we propose that:
i) The necessity of two different surface CO species potentially residing on two
structurally different Rh phases with different reactivity and dependency on
reaction conditions has not been described yet. The combination of long-term
testing, partial pressure variations and dropout experiments revealed a working
catalyst that is only formed under high pressures of CO and long time on stream.
It cannot be described as a metallic Rh surface covered by surface intermediates.
CO activation and CO insertion rates respond very differently to changing
reaction conditions. This interpretation would also explain many differences to
literature results that were obtained at low pressures or short time on stream
as well as discrepancies between experimental and computational studies.
ii) In our model, no CHx coupling mechanism is required to describe chain
elongation as all chain growth occurs via CO insertion. The CHxA coupling
pathway was found to be irrelevant for CO hydrogenation over Rh/SiO2. STE
related literature usually focuses on the formation of oxygenates and details
about hydrocarbon formation are rare but the CHxA coupling mechanism is
commonly assumed.54,72,84 This idea originates from early studies in Fischer
Tropsch research.26,133 Theoretical studies also support the CHXA coupling
mechanism as it is energetically more favorable than CO insertion.117 However,
DFT calculations of isolated species on empty metallic Rh surfaces miss the
crucial impact of dynamic catalyst structures and might favor misleading
conclusions in this case. Several shortcomings of this mechanism to describe
trends in Fischer-Tropsch over different catalysts led to a CO insertion
mechanism with “late” C-O bond breaking for Fe catalysts.30 The idea was
adopted for theoretical studies for Co,136 and further supported by chemical
transient experiments over Co/MgO.159 Our data and the presented detailed
conclusions strongly support the CO insertion with “late” C-O bond breaking
route also over Rh/SiO2.
6ReactionnetworkandkineticsforSTEoverRh/SiO2
94
Many studies address the active sites of supported Rh catalysts for CO hydrogenation. DFT
studies hold different facets of Rh single crystals responsible for different reactivity.75 A
recent review by Ao etal. focusses on the idea of adjacent Rh0-Rhn+ species that catalyze CO
dissociation as well as CO insertion.72 Structural concepts such as formation of surface
carbonyls64,150,151 or carbonyl hydrides130 are also proposed. The specific assignment of
active sites for different reactivity is beyond this work. However, the results clearly show
that the catalyst structure is dynamic. Simple considerations on single crystal facets and
metallic nanoparticles seem insufficient to describe the observed effects. The dynamic
formation and destruction of active species induced by reaction conditions and time on
stream must be considered. The STE reaction therefore appears to be a prime example of
the so-called "pressure and material gap" in heterogeneous catalysis, i.e. the concept that a
much higher complexity is required to describe a catalytic system under real operation
conditions compared to theoretical computations and high-vacuum experiments performed
with single crystal model systems of metals.164 For more applicable theoretical insight, more
complex catalyst structures must be investigated. From the experimental perspective,
operando spectroscopy is inevitable. However, the necessary reaction conditions with
pressures of 30- 50 bar and high CO content require sophisticated experimental
approaches. From the presented study, it can be speculated that the reversible formation of
Rh (sub)carbonyl species or carbonyl hydrides play an important role as the catalytic
properties are strongly dependent on CO partial pressure.
In conclusion, interpretation of the presented catalytic results yielded a plausible reaction
network that explains all observed trends and effects in a qualitative manner. Some features
of the network are consistent with widely accepted mechanisms. However, key features
such as two adsorbed CO* species with different reactivities and a common C-C-coupling
mechanism for oxygenates and hydrocarbons add important information to already
published networks as they emphasize the challenge to optimize reaction conditions
towards a high oxygenate yield. The difference of the obtained data to previously reported
data is the long time on stream, amount of tested reaction conditions, combination of
experiments including cofeed and dropout experiments, and precision of product
evaluation in terms of minor side products. Although a distinct assignment of reactive
surface sites was not possible, the crucial impact of CO partial pressure on catalyst structure
and reactivity was highlighted.
95
7 Modification of Rh/SiO2 with Mn
and/or Fe
All experiments described in the previous chapter were performed in parallel for Rh/SiO2
and the modified catalysts RhMn/SiO2, RhFe/SiO2, and RhMnFe/SiO2. In this chapter,
comparative evaluation of the obtained datasets for the different catalysts sheds light on the
effect of catalyst composition on catalytic performance and reaction kinetics. The combined
evidence suggests that syngas conversion on different Rh-based catalysts can be describe in
terms of a common unified reaction network and mechanism. As for Rh/SiO2, the evaluation
of apparent activation energies did not provide vital information on the reaction network
and are shown and discussed in Appendix E.1.
7.1 Product spectrum and selectivities
The modification of Rh/SiO2 with Mn and/or Fe has tremendous impact on product
selectivities and catalytic activity. For comparability reasons, the catalyst volumes were
adjusted to give a similar level of CO conversion of ~4–5% according to previous test
measurements. CO conversions obtained for Rh/SiO2, RhMn/SiO2, and RhMnFe/SiO2 were
close to the target value in the kinetic study (DinoRun26). RhFe/SiO2 was less active than
anticipated. Therefore, data from a later run (DinoRun27) with adjusted catalyst volume
was used for comparison of product selectivities (Figure7.1A).
The product spectrum of RhMn/SiO2 is similarly complex as observed for Rh/SiO2. However,
the product distribution changes. Selectivities towards oxygenates increase significantly.
Besides higher acetic acid and acetaldehyde selectivities, ethanol becomes a major product.
Methanol selectivity increases slightly but it remains a minor product. High acetic acid
concentrations lead to a more significant formation of acetates. Hydrocarbon selectivities
are generally reduced, selectivities for paraffins with higher carbon numbers in particular.
Comparison of formation rates show more directly how catalyst modification effects
reactivity. Mn addition leads to an increase of total CO consumption rate by a factor of ~6
(Figure7.1B). Here, it becomes obvious that some rates do not change upon Mn addition,
e.g. ethane formation rate with 0.04 µmol s-1 gcat-1 and 0.05 µmol s-1 gcat-1, respectively.
Methane formation rate increases by a factor of 4.5 which still leads to a smaller methane
selectivity. Formation rates of acetic acid and acetaldehyde increase by an order of
magnitude and the formation rate of ethanol even by a factor of 42. Among the tested
7ModificationofRh/SiO2withMnand/orFe
96
catalysts, RhMn/SiO2 is the most efficient catalyst when all C2 oxygenates are considered
valuable products. The similarity of the observed product spectrum is first indication for
common reaction pathways over Rh/SiO2 and RhMn/SiO2.
Figure7.1Comparisonofproductselectivitiesatiso‐conversion(A)andproductformationrates(B)amongthefour
catalystsystems.Beloweachcolumn,COconversion(A)andtotalCOconsumptionrate(B)aregiven.(54bar,
260°C,H2:CO:N2:Ar=60:20:10:10,41.7mlmin‐1,Rh/SiO2:DinoRun26,170hTOS,2500h‐1;RhMn/SiO2:DinoRun26,
170hTOS,12500h‐1;RhFe/SiO2:DinoRun27,190hTOS,8330h‐1;RhMnFe/SiO2:DinoRun26,170hTOS,7580h‐1)
In contrast to the effect of Mn addition, Fe addition leads to a significantly different product
spectrum compared to Rh/SiO2. Methanol becomes the main product with 50 % selectivity.
Only fully hydrogenated products were found. Olefins and aldehydes are missing in the
product spectrum. Selectivity towards acetic acid is significantly decreased but still
measurable also in methyl and ethyl acetates. However, no new products were formed. Also,
Fe addition leads to less chain growth. Ethanol is the only major product with a carbon
number of 2. Higher oxygenates or hydrocarbons were not observed. The total CO
consumption rate increases by a factor of 3 compared to Rh/SiO2. The ability of Fe to narrow
the product spectrum significantly might be beneficial for industrial application, e.g.
concerning separation issues. However, the ethanol formation rate is smaller than over
RhMn/SiO2. RhFe/SiO2 might be feasible in scenarios where methanol is considered a
valuable product, too. The simplified product spectrum does not directly yield information
if reaction pathways are the same or different over RhFe/SiO2.
The product spectrum observed for the trimetallic RhMnFe/SiO2 is similar to RhFe/SiO2.
However, selectivities to methanol is decreased in favor of higher methane and ethanol
selectivities. Mn addition again yields higher C2 oxygenate selectivity but due to the
presence of Fe only in its hydrogenated form ethanol. The obtained ethanol selectivity of
40 32 29
42
14
3
8
3
50 17
1
12
12
26
11
16
11
21
24
0
20
40
60
80
100
4.9XCO / % =
Methane
Propylene
Ethane
Acetic acid
Ethanol
Product selectivities / %
Methanol
Rh RhMnFeRhFeRhMn
Acet-
aldehyde
5.5 4.5 4.0
0.25
1.14
0.57 0.66
0.04
0.05 0.98
0.26
0.21
0.12
0.21
0.03
0.27
0.38
0.0
0.5
1.0
1.5
2.0
2.5
Methane
Ethane
Acetic acid
Acet-
aldehyde
Ethanol
Product formation rates / µmol s-1 g-1
CO
2
Ethyl acetate
Methyl acetate
Acetic acid
Butanal
Propanal
Acetaldehyde
n-Propanol
Ethanol
Methanol
1-Butene
Propylene
Ethylene
n-Butane
Propane
Ethane
Methane
Rh RhMnFeRhFeRhMn
Methanol
r / µmol
s
-1
g
-1
= 0.6 3.5 1.9 1.6
AB
Formation ratesSelectivities
7.2Initialformationphase
97
26 % is the highest among the tested catalysts at reference conditions. Catalyst activity in
terms of total CO consumption rate is the lowest among the modified catalysts. In
comparison to RhFe/SiO2, the lower total rate is predominantly caused by a reduction of
methanol formation rate. Ethanol formation rate is the same as for RhMn/SiO2. Methane
formation rate is thereby almost cut in half making RhMnFe/SiO2 the most suitable catalyst
if ethanol is the only desired product. As for RhFe/SiO2, the observed rates do not give clear
indication about related reaction pathways.
In literature, Mn and Fe addition to Rh-based catalysts have been tested intensively. Mn
addition is commonly described to cause significant increase of catalyst activity and C2
oxygenate selectivity.88,89,97,103 Mao etal. systematically varied Mn loadings to find an
optimal molar Rh:Mn ratio of 2 yielding 46 % C2+ oxygenate selectivity at 17 % CO
conversion.83 In terms of selectivity, these results are comparable to what was observed in
this study (49 % C2 oxygenate selectivity at 5.5 % CO conversion with a Rh:Mn ratio of 1).
Yu etal. propose an optimal Rh:Mn ratio of 1 when Li is additionally added reaching a C2+
oxygenate selectivity of 54 % at 19 % conversion.165 Fe addition to Rh catalysts has been
reported to facilitate H2 activation and increase H* availability,70,113,125,127 but different
effects on the catalytic reactivity were described. In some studies, an increase of ethanol
selectivity was observed but methanol remained a minor product.86,110 Others reported a
significant increase of methanol and ethanol selectivity similar to the results of this
study.70,108,166 However, reported Fe loadings vary from 0.01 wt-% to more than 10 wt-% or
Fe oxide is even used as support. It has been proposed that Fe loading, support material and
preparation methods have impact on Fe oxidation states and in turn on selectivities.113,167
In contrast to this study, catalytic data obtained for different catalysts are usually not
measured in direct comparison and no time for initial catalyst formation and equilibration
was allowed. This might have impact on the observed catalytic performance since the
different catalytic systems exhibit different equilibration behaviors as shown in the next
section. Conclusions on the effect of catalyst modification might therefore be substantially
influenced by the time on stream allowed for catalytic testing.
In sum we measured the most comprehensive product spectrum reported so far, where also
the influence of composition on the formation of minor products like acetates becomes
visible, which opens up the path to understand reaction pathways and mechanisms in more
detail.
7ModificationofRh/SiO2withMnand/orFe
98
7.2 Initial formation phase
The following section shows the initial development of reaction rates for all four tested
catalysts. It will be demonstrated that the reactivity for all catalysts likely is based on the
same reaction network as Rh/SiO2 and that characteristic features such as high C2
oxygenate selectivity for RhMn/SiO2 and high methanol selectivity for RhFe/SiO2 only
evolve during the first days under reaction conditions.
Over all modified catalysts, the CO consumption rate – expressed as negative change of CO
concentration – decreases by ~25 % during the first 130 h time on stream (Figure7.2).
Figure7.2NormalizedCOconsumptionandnormalizedproductconcentrationsovertimeonstreamininitial
equilibrationperiodoverRh/SiO2(A),RhMn/SiO2(B),RhFe/SiO2(C),andRhMnFe/SiO2(D).(DinoRun26,54bar,
260°C,H2:CO:N2:Ar=60:20:10:10,41.7mlmin‐1,Rh/SiO2:2500h‐1;RhMn/SiO2:12500h‐1;RhFe/SiO2:25000h‐1;
RhMnFe/SiO2:7580h‐1)
0.0
0.5
1.0
Normalized concentrations
Propanal
Acetaldehyde
Acetic acid
0 20 40 60 80 100120
0.0
0.5
1.0
H
2
O
Rh
0.0
0.5
1.0
Ethanol
Propanol
Methanol
0.0
0.5
1.0
Propylene
Ethylene
0.0
0.5
1.0
Methane
Ethane Propane
0.0
0.5
1.0
-c
CO
0 20 40 60 80 100120140
Propylene
Ethylene
Acetic acid
Acetaldehyde
Propanol Ethanol
Methanol
Ethane
Methane
Propane
-c
CO
RhMnFe
H
2
O
0 20 40 60 80 100120
Acetic acid
Acetaldehyde
Propanol
Ethanol
Methanol
Ethane
Methane
Propane
-c
CO
RhFe
H
2
O
0 20 40 60 80 100120
Propylene
Ethylene
Acetic acid
Acet-
aldehyde
Propanol
Ethanol
Methanol
Ethane
Methane
Propanal
Propane
-c
CO
RhMn
H
2
O
ABCD
Time on stream / h
7.2Initialformationphase
99
Over RhMn/SiO2, formation rates change very distinctly (Figure7.2B). In contrast to
Rh/SiO2, the trends not only depend on product types, but also on carbon numbers. C3
products decrease faster than C2 products with exception of ethane also following the
accelerated decrease. The catalyst appears to lose part of its chain growth ability.
Considering the reaction network proposed in chapter 6.7, it appears that the reaction step
from the acetaldehyde surface intermediate towards the ethylene intermediate which is the
key intermediate for ethane formation and chain elongation to C3 products becomes slower
during the formation period of the catalyst. This behavior is another indication that all C3
products are formed from the same intermediate.
Over RhFe/SiO2, the decrease of hydrocarbon formation in the first 50 h time on stream is
most prominent among the tested catalysts. (Figure7.2C). After that, concentrations still
decrease but in a more moderate manner. Only paraffins could be evaluated here since no
or very small olefin concentrations were found. Ethanol and propanol follow the severely
decreasing trend for paraffins with propanol concentration dropping below detection limit.
Acetic acid and acetaldehyde are rather stable, but the total concentrations are very low. In
opposition to all other products, methanol formation rate almost doubles during the first
20 h time on stream before it becomes more stable. The severe decrease of all other product
formation rates is likely a result of boosted methanol formation. The demand of H* for
hydrogenation of activated CHxO* species might compete with C-O bond breaking or
hydrogenation processes to form other products. The strongly affected formation rates
might be a result of RhFe alloy formation during the first hours at reaction conditions as
proposed in chapter 4. The alloy formation would then be accountable for the high
hydrogenation ability and thus, methanol selectivity.
Product formation rates over RhMnFe/SiO2 behave similarly to RhMn/SiO2 (Figure7.2D).
Paraffins decrease faster than olefins and C3 product formation decrease faster than C2
product formation. Ethanol is here an exception decreasing synchronously with
acetaldehyde. An initial boost of methanol formation was not observed over RhMnFe/SiO2.
The results obtained from the initial formation phases for the modified catalysts are largely
consistent with the network proposed for Rh/SiO2 in chapter 6.7. The trends over
RhMn/SiO2 suggest a common formation pathway for ethane and all C3 products. This
implies that the reaction pathway for C3 product formation over Rh/SiO2 involving only one
C-C-coupling step applies for RhMn/SiO2 as well. It also shows that the strongly enhanced
C2 oxygenate selectivities over RhMn/SiO2 might be a result of a structural modification
taking place during the first hours under reaction conditions. RhFe/SiO2 shows a boost of
methanol formation during the first hours which might be related to RhFe alloy formation.
The steep increase of methanol formation rate causes severe decrease of all other relevant
product formation rates. According to the proposed reaction network, methanol formation
7ModificationofRh/SiO2withMnand/orFe
100
takes place before the key intermediate for all other products CHx* is formed. Fast
hydrogenation of the activated CHxO* species might lead to lower concentrations of CHx*
and therefore slower formation rates of related products.
7.3 Influence of CO and H2 partial pressures
The influence of reactant partial pressures on consumption and formation rates yielded
important information for the development of the reaction network for Rh/SiO2 in the
previous chapter. In the following section, the trends of rates and reaction orders are
evaluated comparatively showing different effects of Mn and Fe addition. It will be
demonstrated that Mn addition leads to very similar results only with some differences for
acetaldehyde formation which might explain its superior C2 oxygenate productivity. Fe
addition causes more fundamental impact supporting its high hydrogenation ability which
leads to reaction orders that are less dependent on the CO partial pressure regime.
The general trend of total CO consumption over CO and H2 partial pressure is similar for all
four investigated catalysts. The rate increases rapidly with CO partial pressure before the
first measurement point and after that CO partial pressure has only small or no influence on
the rate (Figure7.3). However, small differences were observed. While over Rh/SiO2
increasing CO partial pressures still have a small positive effect, over RhMn/SiO2 a
maximum is observed for medium CO pressures Over both Fe containing catalysts, this
maximum appears to be at very low CO partial pressure before the first measurement point.
Two reasons are thinkable for the decrease of CO consumption rate at high CO partial
pressures. CO seems to adsorb very strongly on metallic Rh at low CO pressures which might
lead to CO or other intermediates to block important sites for product formation pathways.
The second reason might involve the gradual transformation of metallic Rh to a carbonyl
structure which offers less adsorption sites that are able to activate CO. Thereby, H2 partial
pressure has an almost linear effect for all four catalysts. In general, the trend observed for
Rh/SiO2 seem to be applicable for the modified catalysts as well although to different
extents concerning the inhibiting effect of high CO pressures on the total CO consumption
rate.
7.3InfluenceofCOandH2partialpressures
101
Figure7.3COconsumptionratesasafunctionofCOandH
2
partialpressuresoverRh/SiO
2
(A),RhMn/SiO2(B),
RhFe/SiO
2
(C)andRhMnFe/SiO
2
(D).(DinoRun26,54bar,260°CH
2
:CO:N
2
:Ar=30‐60:2.5‐20:10‐57.5:10,
41.7mlmin
‐1
,Rh:2500h
‐1
;RhMn:12500h
‐1
;RhFe:25000h
‐1
;RhMnFe:7580h
‐1
)
As previously discussed for Rh/SiO2, the effect of reactant partial pressures on product
formation rates give important indication for related formation pathways. Over RhMn/SiO2
the observed trends are very similar to what was observed and discussed for Rh/SiO2 in
chapter 6.3.2. Increasing CO partial pressures cause a rapidly decreasing methane
formation rate (Figure7.4). For ethane formation a maximum at medium CO pressures was
found while formation rates for higher paraffins still slowly increase with high CO pressures
(Figure7.5). Also, olefin selectivities show very similar patterns. High CO and low H2
pressures are strongly beneficial. The same trends – although slightly less pronounced –
were found for aldehyde and acetic acid selectivities. The assignment of olefins and
aldehydes into the same product formation pathway as assumed for Rh/SiO2 therefore
seems valid for RhMn/SiO2 as well. Alcohols show complex selectivity patterns. The
maximum methanol formation rate is achieved at very low CO and high H2 partial pressures.
The effect of H2 partial pressure on ethanol formation is less pronounced but the maximum
rates are also reached at low CO pressures. At these reaction conditions almost no aldehydes
and olefins are produced.
Overall, Mn addition leads to significant changes of the product selectivities, but selectivity
patterns remain almost identical. The observed high similarity indicates a common reaction
network for both catalysts where the presence of Mn oxide causes changes of specific
surface reaction rates that cause enhanced C2 oxygenate formation.
7ModificationofRh/SiO2withMnand/orFe
102
Figure7.4ProductformationratesofC1andC2productsoverRhMn/SiO
2
asafunctionofCOandH
2
partial
pressuressortedhorizontallybycarbonnumberandverticallybyoxidationstateofthecarbonatomofthe
respectivefunctionalgroup.(RhMn/SiO
2
,DinoRun26,54bar,260°C,H
2
:CO:N
2
:Ar=30‐60:2.5‐20:10‐57.5:10,
41.7mlmin
‐1
,12500h
‐1
)
7.3InfluenceofCOandH2partialpressures
103
Figure7.5ProductformationratesofC3toC5productsoverRhMn/SiO
2
asafunctionofCOandH
2
partialpressures
sortedhorizontallybycarbonnumberandverticallybyoxidationstateofthecarbonatomoftherespective
functionalgroup.(RhMn/SiO
2
,DinoRun26,54bar,260°C,H
2
:CO:N
2
:Ar=30‐60:2.5‐20:10‐57.5:10,41.7mlmin
‐1
,
12500h
‐1
)
7ModificationofRh/SiO2withMnand/orFe
104
Over RhFe/SiO2, reasonable data were obtained only for products of carbon numbers 1 and
2 (Figure7.6). Patterns for methane and ethane look different compared to Rh/SiO2 and
RhMn/SiO2. The negative impact of high CO partial pressures on methane formation is less
distinct. High H2 partial pressures have much less impact. The severe decrease of methane
formation rates with increasing CO partial pressure over Rh/SiO2 has been related to two
aspects: Competing reactions by CO insertion into a CHx* surface species and an inhibiting
effect of high CO pressures on hydrogenation reactions. Over RhFe/SiO2 one or even both
aspects seem to be of less importance for the formation rate of methane. Ethane formation
is independent from CO partial pressure and only slightly dependent on H2 pressures. Olefin
patterns could not be evaluated due to very low concentrations.
Methanol clearly benefits from high H2 partial pressures over RhFe/SiO2. Also, the negative
effect of high CO pressures is much less pronounced than on Rh/SiO2 and RhMn/SiO2. In
contrast to the Fe free catalysts, high CO partial pressure has a positive effect on ethanol
formation. This pattern is likely a result of fast hydrogenation of acetaldehyde since the
acetaldehyde formation rate does not show the previously observed rapid increase for
increasing CO pressures. For acetic acid, the positive impact of CO partial pressure remains
for RhFe/SiO2 as well. The acid appears to be more resistant against hydrogenation than the
aldehyde. RhFe/SiO2 shows significant CO2 formation with a maximum for low H2 and high
CO pressures.
As already observed for other features, the behavior for RhMnFe/SiO2 is a combination of
RhMn/SiO2 and RhFe/SiO2. The negative impact of CO on methane formation is less distinct
than over RhMn/SiO2 but more pronounced than over RhFe/SiO2 (Figure7.7). Similar
results were observed for ethane. High CO partial pressures lead to increasing ethane
formation, but the behavior is less complex than observed for RhMn/SiO2. Higher paraffins
show formation rate patterns similar to RhMn/SiO2 (Figure7.8). In contrast to RhFe/SiO2,
small amounts of olefins could be measured and evaluated. Their formation rate patterns
are identical to what was observed over the Fe free catalysts. Methanol formation is much
more similar to RhFe/SiO2 than to the Fe free catalysts. The same result was found for
ethanol. High CO partial pressure and therefore likely more CO insertion directly leads to
the alcohol formation. This trend was also observed for propanol. Acetaldehyde formation
rates are rather small and decreasing upon increasing H2 partial pressures further
supporting the fast hydrogenation of intermediates.
RhMnFe/SiO2 generally behaves like Rh/SiO2 and RhMn/SiO2 but some features are clearly
identical to RhFe/SiO2 most likely caused by the enhanced hydrogenation ability of Fe
containing catalysts.
7.3InfluenceofCOandH2partialpressures
105
Figure7.6ProductformationratesofC1andC2productsandmethylacetateoverRhFe/SiO
2
asafunctionofCO
andH
2
partialpressuressortedhorizontallybycarbonnumberandverticallybyoxidationstateofthecarbonatom
oftherespectivefunctionalgroup.(RhFe/SiO
2
,DinoRun26,54bar,260°C,H
2
:CO:N
2
:Ar=30‐60:2.5‐20:10‐57.5:10,
41.7mlmin
‐1
,25000h
‐1
)
7ModificationofRh/SiO2withMnand/orFe
106
Figure7.7ProductformationratesofC1andC2productsoverRhMnFe/SiO
2
asafunctionofCOandH
2
partial
pressuressortedhorizontallybycarbonnumberandverticallybyoxidationstateofthecarbonatomofthe
respectivefunctionalgroup.(RhMn/SiO
2
,DinoRun26,54bar,260°C,H
2
:CO:N
2
:Ar=30‐60:2.5‐20:10‐57.5:10,
41.7mlmin
‐1
,12500h
‐1
)
7.3InfluenceofCOandH2partialpressures
107
Figure7.8ProductformationratesofC3toC5productsoverRhMnFe/SiO
2
asafunctionofCOandH
2
partial
pressuressortedhorizontallybycarbonnumberandverticallybyoxidationstateofthecarbonatomofthe
respectivefunctionalgroup.(RhMn/SiO
2
,DinoRun26,54bar,260°C,H
2
:CO:N
2
:Ar=30‐60:2.5‐20:10‐57.5:10,
41.7mlmin
‐1
,8330h
‐1
)
7ModificationofRh/SiO2withMnand/orFe
108
Considering the reaction network obtained for Rh/SiO2, Mn addition does not seem to
change reaction pathways but rather affect specific surface reaction rates. Fe addition does
not only lead to a much simpler product spectrum, it also causes strongly altered selectivity
trends and patterns. It could be both, different reaction pathways due to the presence of Fe
as well as a similar reaction network for all catalysts with strongly altered rates for some
critical surface steps. Therefore, a common reaction network for all four catalysts is still
thinkable. More information must be obtained from cofeed and drop out experiments
(section 7.5 and 7.6). The trimetallic catalysts shows features of both bimetallic catalysts.
The high hydrogenation ability and tendency for methanol formation is typical for the Fe
containing catalyst. However, over the trimetallic catalyst all products could be evaluated,
and selectivity patterns allow an interpretation consistent with the presented reaction
network with highly accelerated hydrogenation rates.
For a more quantitative evaluation of the influence of CO and H2 partial pressures, apparent
reaction orders were investigated comparatively. As discussed in chapter 6.3.3 for Rh/SiO2,
reaction orders were determined for low (< 5 bar) and high (> 5 bar) CO partial pressures.
Reaction orders for CO and H2 consumption were found to be comparable over all four
catalysts. CO reaction orders are ~0 and H2 reaction orders ~0.9 for all catalysts with only
small impact of different CO pressure regimes (Figure7.9). All values are summarized in
TableA10 and TableA11 in Appendix E. The very similar dependencies suggest at least
similar mechanisms for CO and H2 activation and consumption involving high CO coverages
and H* availability as limiting resource on the surface.
Reaction orders were also evaluated for product formation rates. All catalysts show
negative CO reaction orders and positive H2 reaction orders for methane formation.
However, the investigated CO partial pressure regime has a significant impact over Rh/SiO2
and RhMn/SiO2. At increased CO pressure, more negative CO reaction orders and more
positive H2 reaction orders were observed suggesting that hydrogenation of CHx* fragments
is impeded. In chapter 6.3.3, it was proposed that this effect might be related to the
formation of Rh (sub)carbonyls at high CO pressures. This effect was not found for
RhFe/SiO2 and RhMnFe/SiO2. Modification with Fe plays an important role for increasing
H* availability which seems to make its hydrogenation ability more independent from CO
pressure. The catalytic data does not provide clear indication if carbonyl formation might
also be of importance for Mn and Fe. However, the respective surface carbonyls have not
been reported by spectroscopic studies to our knowledge. In general, Mn carbonyls are only
formed under harsh conditions from metal organic precursors such as Mn acetates.168 Fe
carbonyls are known to be formed under relatively mild conditions from metallic Fe in CO.
However, as described in chapter 4, Mn and Fe are most likely present as oxides on the SiO2
support which usually are not suitable precursors for carbonyl formation. If carbonyls can
be formed from RhFe alloys should be investigated spectroscopically in further research.
7.3InfluenceofCOandH2partialpressures
109
Figure7.9ReactionorderswithrespecttoCOandH2forCOconsumptionandproductformationreactionsover
Rh/SiO2,RhMn/SiO2,RhFe/SiO2,andRhMnFe/SiO2.Darkandlightbarsrepresentreactionorderscalculatedfrom
log‐log‐plotsofdataobtainedatlowCOpartialpressures(pCO<5bar)andhighCOpartialpressures(pCO>5bar),
respectively.(DinoRun26,54bar,260°C,H2:CO:N2:Ar=30‐60:2.5‐20:10‐57.5:10,41.7mlmin‐1,Rh/SiO2:2500h‐1;
RhMn/SiO2:12500h‐1;RhFe/SiO2:25000h‐1;RhMnFe/SiO2:7580h‐1)
-0.5
0.0
0.5
CO consumption
Rh RhMn RhFe RhMnFe Rh RhMn RhFe RhMnFe
CO reaction order H
2
reaction order
CO consumption
Methane
Ethylene
Propylene
-2
0
2
-0.5
0.0
0.5
-2
0
2
Ethane
Propane
Methane
Ethylene
Propylene
-1
0
1
Ethane
Propane
-2
0
2
-1
0
1
Acetic acid
Methanol
Ethanol
Propanol
Acetaldehyde
Propanal
-2
0
2
-3
0
3
-3
0
3
-3
0
3
-3
0
3
-1
0
1
-4
0
4
-1
0
1
-2
0
2
-1
0
1
-2
0
2
-3
0
3
-1
0
1
-3
0
3
Methanol
Ethanol
Propanol
Acetaldehyde
Propanal
-1
0
1
-4
0
4
-3
0
3
Acetic acid
low p
CO
< 5 bar
high p
CO
> 5 bar
7ModificationofRh/SiO2withMnand/orFe
110
Formation of higher paraffins shows similar trends. CO reaction orders are in a similar
range in the low CO regime but change severely in the high CO regime for Rh/SiO
2,
RhMn/SiO2 but not for RhMnFe/SiO2. High CO pressure does not impede hydrogenation of
CxHy* fragments when Fe is present. For olefin formation, the presence of the Mn oxide
matrix does not affect the reaction orders or trends over CO pressure. The strongly positive
CO reaction orders and strongly negative H2 reaction orders in the low CO regime are in
accordance with the hypothesis of the presence of metallic Rh that is active for
hydrogenation at low CO pressures and Rh (sub)carbonyls that are inactive for H2 activation
or hydrogenation at high CO pressures.
Methanol is a minor product over Fe free and major product over Fe containing catalysts.
The negative impact of CO partial pressure on methanol formation is much smaller for Fe
containing catalysts which is consistent with the considerations on hydrogenation reactions
stated above. For ethanol, CO reaction orders are diverse. Over Fe free catalysts, CO reaction
orders are ~0 at low CO pressures and negative in the high CO pressure regime. Again, these
results support the impeding effect of CO on final hydrogenation steps. Fe containing
catalysts show positive CO reaction orders for both CO pressure regimes. Results obtained
for propanol are largely in accordance with ethanol formation. Consecutive reaction such as
acetate formation, however, make detailed evaluation very difficult for oxygenates.
For acetaldehyde, Rh/SiO2 and RhMn/SiO2 show the similar trends which is consistent with
all other products. However, at low CO partial pressures, the CO reaction order is much
higher and H2 reaction order much lower over RhMn/SiO2. The formation of the key C2
oxygenate intermediate seems to follow slightly different rules which might be a reason for
the enhanced formation of C2 oxygenates over RhMn/SiO2. At high CO partial pressures,
reaction orders again are comparable. Considering the proposed effect of CO on the Rh
nanostructure, the stabilizing effect of the Mn oxide matrix for Rh nanoparticles is likely
related to that behavior. The results for propanal further support the described correlation.
Over Fe containing catalysts, CO reaction orders are smaller while H2 reaction order are
comparable to RhMn/SiO2. Acetaldehyde and ethanol are the only products found that show
a dependency of reaction orders on the CO pressure regime suggesting that the
hydrogenation of acetaldehyde to ethanol is actually impeded by high CO pressures while
this seems not to be the case for paraffins formation from olefins.
Overall, Rh/SiO2 and RhMn/SiO2 clearly show similar results suggesting a very similar
reaction network. In literature, the negligible impact of Mn addition on reaction orders has
been observed in less detailed experiments as well.83 In our study, we observed differences
for acetaldehyde formation which might be closely related to the stabilizing functionality of
Mn oxide and result in enhanced C2 oxygenate formation over the Mn modified catalyst. Fe
containing catalysts are similar to each other but different to Fe free catalysts. Fe clearly has
7.4Contacttimevariation
111
a more fundamental impact on reaction pathways than Mn. Effects of Fe addition on reaction
orders in literature were only partially consistent with the presented results.123 A reason
could be the strong dependence of reaction orders obtained for Rh/SiO2 on total CO
pressure that was not observed for Fe containing catalysts. Comparison of Rh/SiO2 and
RhFe/SiO2 in a limited pressure regime might therefore yield misleading or incomplete
information. In our presented results, it becomes apparent that an enhanced hydrogenation
ability leads to reaction orders that are much less dependent on the CO pressure regime.
This behavior could suggest that the RhFe alloy particles are less susceptible for structural
transformation into carbonyl structures and more metallic surface remains available for
hydrogenation steps.
7.4 Contact time variation
The concentration profiles over increasing contact times in this section will further support
the strong similarity of the reaction networks on Rh/SiO2 and RhMn/SiO2. Further, products
appearing over RhFe/SiO2 and RhMnFe/SiO2 at high contact times suggest a common
reaction network also for the Fe containing catalysts.
As for other features described in previous sections, RhMn/SiO2 shows a very similar
contact time behavior as Rh/SiO2 described in chapter 6.4. Over the considered contact time
range, CO consumption increases almost linearly with a maximum CO conversion of ~19 %
(Figure7.10A). Methane concentration follows the linear trend. Among the C2 products the
ratio is different with much higher oxygenate and lower hydrocarbon concentrations.
However, the trends over contact time are very similar. Ethylene concentration is low but
stable over the whole range. Ethane concentration increases linearly. Acetic acid and
acetaldehyde increase but formation rates slow down over contact time. Ethanol
concentration increases linearly and becomes most abundant C2 product at high contact
times. As already observed for Rh/SiO2, C3 products show slightly different behavior
compared to their C2 analogues. For example, propylene concentration is not stable but
increasing suggesting that readsorption and consumption rates are lower for propylene
than for ethylene. All minor products such as C4 products, acetates, methanol, and CO2
behave very similar in comparison to Rh/SiO2.
Over RhFe/SiO2 the product spectrum is reduced. Due to the high GHSV for this catalyst, CO
conversion stayed below 5 % for all contact times. All C1 products increase steadily. After
longer contact times, acetaldehyde and ethane are formed in measurable quantities, while
ethylene is not found at all. For ethanol and acetic acid, similar trends as over Rh/SiO2 were
found with linearly increasing ethanol concentration and decreasing acetic acid formation
rates. After even longer contact times, the C3 products propanol and propane are formed.
7ModificationofRh/SiO2withMnand/orFe
112
Figure7.10TheconsumptionofCOexpressedasthenegativechangeofCOconcentrationandtheproduct
concentrationofmethane,C2products,C3products,C4products,acetates,andmethanolandCO2asfunctionof
contacttimeoverRh/SiO2,RhMn/SiO2,RhFe/SiO2,RhMnFe/SiO2.Fortherealizationofdifferentcontacttimes,
GHSVwasvariedbyadjustingthetotalflowsintherangeof8.3–58.3mlmin‐1.(DinoRun26,54bar,260°C,
H2:CO:N2:Ar=60:20:10:10,Rh/SiO2:1ml,RhMn/SiO2:0.2ml,RhFe/SiO2:0.1ml,RhMnFe/SiO2:0.33ml)
0.000
0.001
0.002
0.003
0.004
0.005
0 5 10 15 20
0.0
0.1
0.2
0.3
0.4
0.0
0.2
0.4
0.6
0.8
1.0
0.00
0.01
0.02
0.03
0.04
0.000
0.001
0.002
0.003
0.004
Methyl acetate
CO
2
Methanol
Methane
Ethane
Acetic acid
Acetaldehyde
Ethanol
Propanol
Propane
-c
CO
0.000
0.001
0.002
0.003
0.004
0.00
0.01
0.02
0.03
0.04
0204060
0.0
0.1
0.2
0.3
0.4
0
1
2
3
4
5
0.0
0.1
0.2
0.3
0.4
0.000
0.002
0.004
0.006
0.008
Ethyl
acetate
Methyl acetate
CO
2
Butane
Methanol
Methane
Ethane
Acetic acid
Acetaldehyde
Ethanol
Propanol
Propane
Propylene
-c
CO
0.000
0.005
0.010
0.015
0.000
0.002
0.004
0.006
0.008
0 100 200
0.00
0.01
0.02
0.03
0.04
0
1
2
3
4
5
0.0
0.1
0.2
0.3
0.4
0.00
0.02
0.04
0.06
0.08
Butanal
Ethyl acetate
Methyl acetate
CO
2
Butane
Butene
Methanol
Product concentration / vol-%
Methane
Ethane
Acetic acid
Acetaldehyde
Ethylene
Propanol
Propanal
Propane
Propylene
-c
CO
0.000
0.005
0.010
0.015
0.00
0.02
0.04
0.06
0.08
010203040
0.00
0.01
0.02
0.03
0.04
0
1
2
3
4
5
0.0
0.1
0.2
0.3
0.4
0.00
0.01
0.02
0.03
0.04
Ethyl acetate
Methyl
acetate
CO
2
Butane
Butene
Methanol
Methane
Ethane
Acetic acid
Acetaldehyde
Ethylene
Ethanol
Propanol
Propanal
Propane
Propylene
-c
CO
Contact time / s
RhMn/SiO
2
Rh/SiO
2
RhFe/SiO
2
RhMnFe/SiO
2
CO consumption
& methane
C2 products
C3 products
C4 products
Acetates
Methanol & CO
2
Ethanol
7.5Cofeedstudies
113
The presence of acetaldehyde in the product spectrum of RhFe/SiO2 after longer contact
times further indicates that it is also part of the reaction network but does not escape from
the reactor at standard conditions due to its fast consumption in hydrogenation reactions.
The appearance of C3 products proves the general ability of RhFe/SiO2 for chain elongation.
This provides indication that the simplified product spectrum is not a result of altered
formation pathways but rather of altered kinetics. Severely accelerated hydrogenation and
therefore termination of chain growth in an early stage would lead to products with small
carbon numbers. The reaction pathway to ethanol as only major C2 product might therefore
not be conceptually different compared to the formation pathways to products with carbon
number of 2 or higher over Fe free catalysts.
The behavior of concentrations over RhMnFe/SiO2 again can be interpreted as combined
effects of Mn and Fe addition.
In summary, profiles of concentrations over contact times for RhMn/SiO2 further support
specific phenomena observed for Rh/SiO2 as reproducible features, e.g. the stable ethylene
concentration and decreasing acetaldehyde and acetic acid rates. Appearance of products
missing in the product spectrum of RhFe/SiO2 after longer contact times indicate that the
reaction network derived for Rh/SiO2 might also be applicable for RhFe/SiO2 and
RhMnFe/SiO2 with highly accelerated hydrogenation rates.
7.5 Cofeed studies
To further probe common or different product formation pathways, results of the cofeed
studies were evaluated comparatively. It will be shown, that all four catalysts respond to
the cofeeds in a very similar way further supporting a common reaction network for the
considered Rh-based catalysts.
In chapter 6.5.1 the effect of ethylene cofeeding on CO consumption and product formation
rates over Rh/SiO2 was discussed in detail. As for Rh/SiO2, the cofed ethylene is fully
converted over the modified catalysts as well (Figure7.11). RhMn/SiO2 shows a stable
ethylene concentration over different cofeed concentrations which is consistent with the
behavior over Rh/SiO2. For the latter, it was proposed that ethylene is mainly converted in
the first part of the catalyst bed whereas the presence of reaction products might impede
ethylene consumption in later parts of the bed (compare chapter 6.5.1 and references
therein). For the Fe containing catalysts, no ethylene has escaped from the reactors at
reference conditions with and without ethylene cofeed probably due to their high
hydrogenation ability.
7ModificationofRh/SiO2withMnand/orFe
114
Figure7.11COconsumptionexpressedasnegativechangeofCOconcentrationandproductconcentrationsof
methane,C2products,andC3productsasfunctionofethyleneinletconcentrationoverRh/SiO2,RhMn/SiO2,
RhFe/SiO2,andRhMnFe/SiO2.Numbersdenotechronologicalorderofdatapoints.(DinoRun29,54bar,260°C,
H2:CO:N2:Ar=60:20:10:10+cofeed,Rh:2500h‐1;RhMn:12500h‐1;RhFe:8330h‐1;RhMnFe:7580h‐1)
0.00 0.02 0.04 0.06
0.0
0.4
0.8
1.2
1.6
2.0
0.00 0.02 0.04 0.06
0.00
0.04
0.08
0.12
0.16
0.20
0.00 0.02 0.04 0.06
0.000
0.004
0.008
0.012
0.016
0.020
Concentration / vol-%
Methane
Ethane
Acetic acid
Acetaldehyde
Ethylene
Ethanol
Propanol
Propanal
Propane
Propylene
-cCO
0.00 0.02 0.04 0.06
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0.00 0.02 0.04 0.06
0.00
0.01
0.02
0.03
0.04
0.05
0.06
0.00 0.02 0.04 0.06
0.000
0.004
0.008
0.012
0.016
0.020
Concentration / vol-%
Methane
Ethane
Acetic acid
Acetaldehyde
Ethanol Propanol
Propanal
Propane
-c
CO
0.00 0.02 0.04 0.06
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0.00 0.02 0.04 0.06
0.00
0.02
0.04
0.06
0.08
0.10
0.12
0.00 0.02 0.04 0.06
0.000
0.004
0.008
0.012
0.016
0.020
Concentration / vol-%
Ethylene inlet concentration / vol-%
Methane Ethane
Acetic acid
Acetaldehyde
Ethanol Propanol
Propane
-cCO
0.00 0.02 0.04 0.06
0.0
0.2
0.4
0.6
0.8
1.0
0.00 0.02 0.04 0.06
0.00
0.02
0.04
0.06
0.08
0.10
0.00 0.02 0.04 0.06
0.000
0.002
0.004
0.006
0.008
0.010
Concentration / vol-%
Methane
Ethane
Acetic acid
Acetaldehyde
Ethylene Ethanol
Propanol
Propanal
Propane
Propylene
-c
CO
COconsumption&methane C2products C3products
COconsumption&methane C2products C3products
COconsumption&methane C2products C3products
Rh/SiO
2
RhMn/SiO
2
RhMnFe/SiO
2
1
2
3
4
1
2
3
4
1
2
3
4
RhFe/SiO
2
COconsumption&methane C2products C3products
1
2
3
4
7.5Cofeedstudies
115
Over all catalysts, ethylene is mainly hydrogenated to ethane with a selectivity of 65–70 %.
Conversion and selectivities for all compounds and catalysts are summarized in TableA12
and TableA13 in Appendix E.4. Other C2 products are not significantly affected by the
cofeed. The CO insertion step to form propanal and propanol from ethylene works over all
considered catalyst. As opposed to Rh/SiO2, none of the modified catalysts show significant
formation of propanol and/or propanal at reference conditions without cofeed.
Interestingly, Fe containing catalysts show a high CO insertion rate in the cofeed
experiments as well pointing towards similar reaction pathways over all four catalysts. As
for Rh/SiO2, no C3 hydrocarbons were formed from CHx* coupling over any catalyst.
A slight decrease of CO consumption rate during ethylene cofeeding was observed over all
catalyst. As described in detail in chapter 5.2.3, this is not a direct result of ethylene dosing
but ongoing activity loss over time on stream.
Cofed propylene is converted by 46 % over RhMn/SiO2 which is again consistent with the
result for Rh/SiO2 (48 %). Over RhFe/SiO2, propylene is almost fully converted emphasizing
the enhanced hydrogenation ability of Fe containing catalysts (Figure7.12). Propylene
conversion over the trimetallic catalyst is 66 % and therefore between the results found for
the bimetallic catalysts. Main product of propylene conversion is propane formation via
hydrogenation over all catalysts. CO insertion under C4 oxygenate formation was not
observed. The different behavior upon ethylene and propylene cofeed is consistent with the
results of cofeed experiments over Rh/SiO2. It supports that propylene does not or only to
very small extents undergo CO insertion after readsorption as discussed in more detail in
chapter 6.7. Propylene also does not affect CO consumption rates over any considered
catalyst.
Additional acetaldehyde is almost fully converted over all catalysts including Rh/SiO2
(Figure7.13). Especially for the Fe free catalysts this behavior indicates that the conversion
of acetaldehyde is significantly faster in the first part of the catalyst bed. The only product
directly formed from acetaldehyde is ethanol via hydrogenation. The decrease of acetic acid
partially or fully results at least partially from increased ethyl acetate formation with
increasing ethanol concentrations. CO consumption is again not affected.
None of the considered catalyst systems adsorbs or converts CO2. Neither CO consumption
rates nor any product formation rates are affected by CO2 cofeed in concentrations of up to
3.5 vol-% (Figure7.14). No significant CO2 hydrogenation or water-gas-shift activity was
observed for any catalyst at the applied reaction conditions. CO2 is produced in moderate
amounts on Fe containing catalysts. It might originate from reverse water-gas-shift but also
from methanol steam reforming, decarboxylation or similar. Concentration formed at
typical conditions are too low to gain more detailed insight. The cofed CO2 concentrations
are much higher than the product concentration at reference conditions.
7ModificationofRh/SiO2withMnand/orFe
116
Figure7.12COconsumptionexpressedasnegativechangeofCOconcentrationandproductconcentrationsof
methane,C3products,andC4productsasfunctionofpropyleneinletconcentrationoverRh/SiO2,RhMn/SiO2,
RhFe/SiO2,andRhMnFe/SiO2.Numbersdenotechronologicalorderofmeasureddatapoints.(DinoRun29,54bar,
260°C,H2:CO:N2:Ar=60:20:10:10+cofeed,Rh:2500h‐1;RhMn:12500h‐1;RhFe:8330h‐1;RhMnFe:7580h‐1)
0.000 0.004 0.008 0.012
0.0
0.4
0.8
1.2
1.6
2.0
0.000 0.004 0.008 0.012
0.000
0.001
0.002
0.003
0.004
0.000 0.004 0.008 0.012
0.000
0.004
0.008
0.012
0.016
0.020
Concentration / vol-%
Butanal
Butene
Butane
Methane
Propanol
Propanal
Propane
Propylene
-c
CO
1
2
3
4
0.000 0.004 0.008 0.012
0.0
0.2
0.4
0.6
0.8
1.0
0.000 0.004 0.008 0.012
0.0000
0.0004
0.0008
0.0012
0.0016
0.000 0.004 0.008 0.012
0.000
0.004
0.008
0.012
0.016
Concentration / vol-%
Butanal
Butene
Butane
Methane
Propanol
Propanal
Propane
Propylene
-c
CO
12
3
4
0.000 0.004 0.008 0.012
0.0
0.2
0.4
0.6
0.8
1.0
0.000 0.004 0.008 0.012
0.0000
0.0002
0.0004
0.0006
0.0008
0.0010
0.000 0.004 0.008 0.012
0.000
0.002
0.004
0.006
0.008
0.010
Concentration / vol-%
Butane
Methane
Propane
Propylene
-c
CO
1
2
3
4
0.000 0.004 0.008 0.012
0.0
0.2
0.4
0.6
0.8
1.0
0.000 0.004 0.008 0.012
0.0000
0.0002
0.0004
0.0006
0.0008
0.0010
0.000 0.004 0.008 0.012
0.000
0.002
0.004
0.006
0.008
0.010
Concentration / vol-%
Butane
Propylene inlet concentration / vol-%
Methane
Propanol
Propane
Propylene
-c
CO
12
3
4
CO consumption & methane C3 products C4 products
CO consumption & methane C3 products C4 products
RhFe/SiO
2
RhMnFe/SiO
2
CO consumption & methane C3 products C4 products
RhMn/SiO
2
CO consumption & methane C3 products C4 products
Rh/SiO
2
7.5Cofeedstudies
117
Figure7.13COconsumptionexpressedasnegativechangeofCOconcentrationandproductconcentrationsof
methane,C2products,andacetatesasfunctionofacetaldehydeinletconcentrationoverRh/SiO2,RhMn/SiO2,
RhFe/SiO2,andRhMnFe/SiO2.Numbersdenotechronologicalorderofmeasureddatapoints.(DinoRun29,54bar,
260°C,H2:CO:N2:Ar=60:20:10:10+cofeed,Rh:2500h‐1;RhMn:12500h‐1;RhFe:8330h‐1;RhMnFe:7580h‐1)
0.00 0.04 0.08 0.12 0.16
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0.00 0.04 0.08 0.12 0.16
0.00
0.04
0.08
0.12
0.16
0.20
0.24
0.00 0.04 0.08 0.12 0.16
0.000
0.004
0.008
0.012
0.016
0.020
Concentration / vol-%
Methane
-c
CO
12
3
4
Acetaldehyde inlet concentration / vol-%
Ethane
Acetic acid
Acetaldehyde
Ethanol
Ethyl acetate
Methyl acetate
0.00 0.04 0.08 0.12 0.16
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0.00 0.04 0.08 0.12 0.16
0.00
0.04
0.08
0.12
0.16
0.20
0.00 0.04 0.08 0.12 0.16
0.000
0.002
0.004
0.006
0.008
0.010
0.012
Concentration / vol-%
Methane
-
c
CO
1
2
3
4
Ethane
Acetic acid
Acetaldehyde
Ethanol
Ethyl acetate
Methyl
acetate
0.00 0.04 0.08 0.12 0.16
0.0
0.4
0.8
1.2
1.6
2.0
0.00 0.04 0.08 0.12 0.16
0.00
0.04
0.08
0.12
0.16
0.20
0.00 0.04 0.08 0.12 0.16
0.000
0.005
0.010
0.015
0.020
0.025
0.030
Concentration / vol-%
Methane
-c
CO
1
2
3
4
Ethane
Acetic acid
Acetaldehyde
Ethylene
Ethanol
Ethyl acetate
Methyl acetate
0.00 0.04 0.08 0.12 0.16
0.0
0.2
0.4
0.6
0.8
1.0
0.00 0.04 0.08 0.12 0.16
0.00
0.03
0.06
0.09
0.12
0.15
0.00 0.04 0.08 0.12 0.16
0.000
0.002
0.004
0.006
0.008
0.010
Concentration / vol-%
Methane
Ethane
Acetic acid
Acetaldehyde
Ethylene
Ethanol Ethyl acetate
-c
CO
1
2
3
4
Methyl acetate
CO consumption & methane C2 products Acetates
Rh/SiO
2
CO consumption & methane C2 products Acetates
RhMn/SiO
2
CO consumption & methane Ethanol C2 products & acetates
RhFe/SiO
2
CO consumption & methane Ethanol C2 products & acetates
RhMnFe/SiO
2
7ModificationofRh/SiO2withMnand/orFe
118
Figure7.14COconsumptionexpressedasnegativechangeofCOconcentrationandconcentrationsofmethaneand
CO2asfunctionofCO2inletconcentrationoverRhMn/SiO2,RhFe/SiO2,andRhMnFe/SiO2.Numbersdenote
chronologicalorderofmeasureddatapoints.(DinoRun29,54bar,260°C,H2:CO:N2:Ar=60:20:10:10+cofeed,Rh:
2500h‐1;RhMn:12500h‐1;RhFe:8330h‐1;RhMnFe:7580h‐1)
Chapter 6.5 provides information about cofeed experiments for STE or related reactions
provided by literature. By now, no detailed information about the effect of catalyst
modification on cofeed experiments could be found in reported data.
In summary, all catalysts show a high hydrogenation ability for olefins and aldehydes.
Results for the Fe free catalysts suggest an enhanced hydrogenation ability in the first part
of the bed. Fe containing catalysts show a generally enhanced hydrogenation ability in the
order RhFe/SiO2 > RhMnFe/SiO2 > RhMn/SiO2 ≈ Rh/SiO2. Ethylene is more easily converted
than propylene. It can undergo hydrogenation and CO insertion while propylene is only
hydrogenated. Ethanol can be formed from acetaldehyde efficiently over all four catalysts.
CO2 hydrogenation is not relevant under the applied STE reaction conditions. The very
similar behavior further supports a common reaction network for the four considered
catalysts.
01234
0
1
2
3
4
01234
0
1
2
3
4
01234
0
1
2
3
4
01234
0
1
2
3
4
CO
2
-c
CO
Methane
CO
2
inlet concentration / vol-%
CO
2
-c
CO
Methane
-c
CO
Methane
CO
2
-c
CO
Methane
CO
2
Concentration / vol-%
RhMn/SiO
2
RhFe/SiO
2
RhMnFe/SiO
2
Rh/SiO
2
7.6Dropoutexperiments
119
7.6 Dropout experiments
Dropout experiments over Rh/SiO2 revealed little effect of the absence of H2 while the
prolonged absence of CO poses a tremendous impact, indicating a CO induced restructuring
of the Rh surface similar to the initial equilibration phase of each experiment (compare
chapter 6.6). By comparing the drop-out response of the four different catalysts, it will be
shown that all studied Rh-based catalyst share a common reaction network, and that also
the dynamic transition of metallic Rh into carbonyl structures is likely to occur on all
catalysts.
Over RhMn/SiO2, the effect of H2 dropout is the same as previously observed for Rh/SiO2.
After the H2 was removed from the atmosphere, reaction quickly stopped (Figure7.15).
Right after the removal of H2, ethylene formation is enhanced which is probably caused by
a very low H2:CO ratio in the feed. After 24 h, reference syngas conditions were applied again
immediately restoring the original selectivities with a slightly reduced total activity.
Figure7.15NormalizedconcentrationsofthemainproductsintheupperpartandCOconversionandproduct
yieldsinthelowerpartasfunctionoftimeonstreamduringaH2dropoutoverRh/SiO2,RhMn/SiO2,RhFe/SiO2,
RhMnFe/SiO2.Whitenumbersdenoteselectivitiesofthemainproducts.(DinoRun29,54bar,260°C,
H2:CO:N2:Ar=60:20:10:10,forthedropoutexperimentsH2wasreplacedbyadditionalN2,Rh/SiO2:2500h‐1;
RhMn/SiO2:12500h‐1;RhFe/SiO2:8330h‐1;RhMnFe/SiO2:7580h‐1)
40 39 39
25 24 26
22 22 21
Ethylene
Acetaldehyde
syngas
CO + H
2
CO/
N
2
syngas
CO + H
2
350300 400
CO
2
Ethyl
acetate
Methyl
acetate
Acetic
acid
Propanal
Acet-
aldehyde
Propanol
Ethanol
Methanol
1-Butene
Propylene
Ethylene
Butane
Propane
Ethane
Methane
CO
conversion
32 33 33
333
10 10 11
15 15 15
25 25 24
Ethylene
syngas
CO + H
2
CO/
N
2
syngas
CO + H
2
350300 400
1
2
3
40 40 39
666
14 15 14
21 21 21
Normalized concentrations
syngas
CO + H
2
CO/
N
2
syngas
CO + H
2
350300 400
0
2
4
6
8
Product yields / %
24 23 23
59 60 61
10 10 9
Acetaldehyde
syngas
CO + H
2
CO/
N
2
syngas
CO + H
2
350300 400
RhMn/SiO
2
Rh/SiO
2
RhFe/SiO
2
RhMnFe/SiO
2
Time on stream / h
Methane
Acetic acid
Acet-
aldehyde
Ethanol
Methanol
Ethanol
Methane
7ModificationofRh/SiO2withMnand/orFe
120
Similar to the Fe free catalysts, product formation quickly stops after H2 was removed from
the feed over RhFe/SiO2. The normalized concentration of acetaldehyde is significantly
enhanced during the H2 dropout. The high value is mainly due to the very low acetaldehyde
concentrations at reference conditions which were used as base for normalization.
However, this behavior shows that acetaldehyde is formed over RhFe/SiO2 if only very little
H* is available further supporting that ethanol formation over RhFe/SiO2 proceeds via
acetaldehyde. This is further indication for a common reaction network of all Rh-based
catalysts.
As for most experiments discussed in this chapter, the results obtained from dropout
experiments over RhMnFe/SiO2 are a between the results for RhMn/SiO2 and RhFe/SiO2.
During the H2 dropout, acetaldehyde and ethylene concentration were enhanced. Both
products are rapidly hydrogenated at reference conditions but their appearance upon H2
removal supports their role as important intermediates. After the H2 dropout, the original
catalyst performance was immediately reestablished.
The removal of CO from the feed gas composition has clearly more diverse effects on the
catalysts’ behavior than H2 removal. Immediately after the dropout of CO, methane and
ethanol formation are enhanced over Rh/SiO2 and RhMn/SiO2 in consistence with
formation rates at very low CO:H2 ratios (Figure7.16). Ethanol formation then quickly
stops whereas methane is formed throughout the dropout duration from very small residual
CO concentrations in the feed. CO insertion does not take place anymore at these low CO
concentrations. After switching back to reference reaction conditions, CO conversion is
significantly enhanced compared to before the CO dropout. Also, selectivities changed to
more acetaldehyde and olefins and less acetic acid and ethanol. Until here, RhMn/SiO2
behaves very similar to Rh/SiO2. However, not all rates over the Mn modified catalysts show
a tendency to approach the original performance. Total CO consumption as well as
acetaldehyde and ethylene formation appear to be permanently enhanced. Likely, the
absence of CO caused an impact on the catalyst structure by modifying the interaction of Mn
oxide matrix with Rh particles and/or Rh (sub)carbonyls (compare chapter 4.3).
Over RhFe/SiO2, methane concentration after CO removal slowly increases instead of the
sharp increase over Fe free catalysts. This is consistent with reaction orders for methane
formation in the high CO pressure regime. CO reaction orders are negative for Rh/SiO2 and
RhMn/SiO2 and close to zero for RhFe/SiO2. The dropping CO concentration in the
atmosphere therefore has less impact on methane formation. After reapplying reference
reaction conditions, CO consumption rate is shortly enhanced mainly due to high methane
formation. However, product formation rates quickly approach their original value. The
original performance is reestablished after ~60 h. Enhanced acetaldehyde and acetic acid
rates are not significant due to their low total values. The time range for equilibration after
7.6Dropoutexperiments
121
CO dropout again suggests reversible structural rearrangements. In contrast to RhMn/SiO2,
no permanent change of the catalyst structure and performance was observed. Methanol
formation is immediately reestablished without the overshooting effect of the other rates.
This might be caused by the position of methanol formation in the reaction network.
Methanol is formed at the very beginning before the crucial CHx intermediate for all other
products is formed. This step might be less dependent on structural rearrangements and
therefore immediately reestablished as soon as CO is available again.
Figure7.16Normalizedconcentrationsofthemainproductsintheupperpartandproductyieldsinthelowerpart
asfunctionoftimeonstreamduringaH2dropout(A)andCOdropoutexperiment(B)overRhFe/SiO2.White
numbersdenoteselectivitiesofthemainproducts.(RhFe/SiO2,DinoRun29,54bar,260°C,H2:CO:N2:Ar=
60:20:10:10,fordropoutexperimentstherespectivereactantwasreplacedbyadditionalN2,8330h‐1)
Over RhMnFe/SiO2, methane formation during the CO dropout again corresponds to the
reaction orders for methane. It first slowly increases before it decreases with the residual
CO concentration in the feed. After the dropout, CO consumption rate is enhanced mainly
because of a higher methane formation rate. All rates show a tendency to approach their
original values at first, but many stabilize before the original state was reached. Similar as
over RhMn/SiO2, the CO dropout seems to cause an irreversible impact on the interaction
of the RhFe particles with the surrounding Mn oxide matrix. Methanol formation again does
23 32 25
60
51 57
10 11 9
Methane
Ethane
Ethanol
Acet-
aldehyde
syngas
CO + H
2
H
2
/
N
2
syngas
CO + H
2
Acetic
acid
Methanol
400300 500
39 47 45
26
18 20
21
21 21
Methane
Ethane
Ethanol
Acet-
aldehyde
syngas
CO + H
2
H
2
/
N
2
syngas
CO + H
2
Ethylene
Acetic
acid
Methanol
400300 500
CO
2
Ethyl
acetate
Methyl
acetate
Acetic
acid
Propanal
Acet-
aldehyde
Propanol
Ethanol
Methanol
1-Butene
Propylene
Ethylene
Butane
Propane
Ethane
Methane
CO
conversion
33 34 33
3
43
11
88
15
16 17
24
21 22
Methane
Ethane
Ethanol
Acetaldehyde
Ethylene
syngas
CO + H
2
H
2
/
N
2
syngas
CO + H
2
Acetic
acid
400300 500
1
2
39 46 41
6
76
14
12 13
21
16 19
Normalized concentrations
Methane
Ethane
Ethanol
syngas
CO + H
2
H
2
/
N
2
syngas
CO + H
2
400300 500
0
2
4
6
8
Product yields / %
RhMn/SiO
2
Rh/SiO
2
RhFe/SiO
2
RhMnFe/SiO
2
Time on stream / h
Methane
Acetic acid
Acet-
aldehyde
Ethanol
Methanol
Ethanol
Methane
7ModificationofRh/SiO2withMnand/orFe
122
not follow the overshooting concentrations of the other products, likely because of its
separate position in the reaction network.
Dropout experiments over the modified catalysts support the findings for Rh/SiO2. The
presence of CO has an important impact on the catalyst structure. Absence of CO in the gas
phase leads to reversible structural rearrangements on Rh/SiO2 and RhFe/SiO2 that cause
a catalytic behavior similar to their original state before an initial equilibration phase.
Dynamic formation and removal of Rh (sub)carbonyl species in dependence of CO pressure
could explain this observation. The restructuring effects on Mn containing catalysts are
partially permanent suggesting an irreversible impact on the interaction of the metal
particles or potentially Rh (sub)carbonyl species with the well dispersed Mn oxide matrix.
The similarity of the observed effects and appearance of intermediates over Fe containing
catalysts upon H2 removal provides further evidence that the underlying reaction
mechanisms are the same over all four Rh-based catalysts.
7.7 Effects of modification on reaction network
The presented results in the previous sections of this chapter – cofeed and dropout
experiments in particular – suggested a common reaction network for all four considered
Rh-based catalysts. Therefore, it can be assumed that the reaction network developed for
Rh/SiO2 is applicable for the modified catalysts as well. However, in order to reflect different
reactivities, modification of Rh/SiO2 with Mn and/or Fe must have significant impact on
specific surface reaction rates. A statistical tree diagram approach allows the calculation of
probabilities for each branch of the reaction network (compare chapter 3.6.2 for further
explanation). The results presented in this section were obtained using the rates obtained
at reference conditions and similar levels of CO conversion of ~5 % and reveal the impact
of Mn and Fe addition on specific steps in the reaction network.
The first part of the reaction network contains the initial CO activation by hydrogenation of
adsorbed COA necessary for the formation of all products (Figure7.17). Although rates are
likely to be different for this step, no respective probability can be calculated for this step.
The first branch for which a probability can be calculated is the formation of methanol from
the HxCOA fragment. Over the Fe free catalysts, this pathway is negligible. Instead, almost all
formed HxCOA intermediates undergo C-O bond cleavage. Clearly, over RhFe/SiO2 the
probability that a HxCOA intermediate is hydrogenated to methanol is highest. 52 % of the
intermediates are hydrogenated before C-O bond cleavage can take place. Addition of Fe
therefore already change probabilities at the very beginning of the reaction network due to
its increased hydrogenation ability. Over RhMnFe/SiO2 the probability of methanol
formation is between the bimetallic catalysts with 20 %.
7.7Effectsofmodificationonreactionnetwork
123
Figure7.17Part1ofthereactionnetworkcontainingtheinitialCOAactivationbyfirsthydrogenation,methanol
formationandH‐assistedC‐Obondcleavage.Numbersdenotetheprobabilityforthedifferentreactionpathways
originatingfromtheHxCOAsurfaceintermediateforRh/SiO2(blue),RhMn/SiO2(green),RhFe/SiO2(red)and
RhMnFe/SiO2(purple).Calculatedfromselectivitiesobtainedatstandardconditionsand~5%COconversion.
(54bar,260°C,H2:CO:N2:Ar=60:20:10:10,41.7mlmin‐1,Rh/SiO2:DinoRun26,2500h‐1;RhMn/SiO2:DinoRun26,
12500h‐1;RhFe/SiO2:DinoRun27,8330h‐1;RhMnFe/SiO2:DinoRun26,7580h‐1)
C-O cleavage in the first part of the network results in the CHxA fragment and a surface
hydroxyl group. The hydroxyl group can be hydrogenated to form water or be involved in
acetic acid formation later in the network. The CHxA fragment can be hydrogenated to form
methane which is assumed to easily desorb from the surface (Figure7.18). Alternatively,
the surface intermediate CHxCOA can be formed by CO insertion. For Rh/SiO2 and
RhMn/SiO2, CO insertion is more likely than hydrogenation to methane. The low availability
of H* on the Fe free catalysts is likely a reason. Nevertheless, methane is the main product
over both.
Figure7.18Part2ofthededucedreactionnetworkcontaininghydrogenationoftheCHxAfragmenttoform
methaneandCOBinsertionintoCHxAformtheCHxCOAsurfaceintermediate.Numbersdenotetheprobabilityfor
thedifferentreactionpathwaysoriginatingfromtheCHxAsurfaceintermediateintheorderRh/SiO2(blue),
RhMn/SiO2(green),RhFe/SiO2(red)andRhMnFe/SiO2(purple).Calculatedfromselectivitiesobtainedatstandard
conditionsand~5%COconversion.(54bar,260°C,H2:CO:N2:Ar=60:20:10:10,41.7mlmin‐1,Rh/SiO2:DinoRun26,
2500h‐1;RhMn/SiO2:DinoRun26,12500h‐1;RhFe/SiO2:DinoRun27,8330h‐1;RhMnFe/SiO2:DinoRun26,7580h‐1)
CO
A
CO
A
:highcoveragealreadyatverysmallp
CO
CO
B
:onlyformedatincreasedp
CO
H*
H
2(g)
CO
(g)
CO
B
H
x
CO
A
OH
A
CH
xA
CH
3
OH
(g)
Methanol
+H*
+H*
+H* +H*
0.4
4
52
20
99.6
96
48
80
Rh/SiO
2
RhMn/SiO
2
RhFe/SiO
2
RhMnFe/SiO
2
+2H*
CxHyCH2OH(g)
Alcohol
CxHyCHOC
CxHyCHO(g)
Aldehyde
CxHy(g)
Olefin
CxHy+2(g)
Paraffin
CxHyC
Hydrogenationsite(s)
COA
H*
H2(g) CO(g)
COB
CH3COOHA
CH3COOH(g)
Aceticacid
CH3COA
CH3CHO(g)
Acetaldehyde
CHxCHxA
C2H4(g)
Ethylene
HxCOAOH*
H2O(g)
CHxA
CH4(g)
Methane
CH3OH(g)
Methanol
C2H5CHOA
C2H5CHO(g)
Propanal
C2HxCHxA
C3H6(g)
Propylene
+COB
CH
3
CO
A
+CO
B
OH
A
H
2
O
(g)
CH
xA
+H*
CH
4(g)
Methane
+H*
40
34
61
52
60
66
39
48
+2H*
C
x
H
y
CH
2
OH
(g)
Alcohol
C
x
H
y
CHO
C
C
x
H
y
CHO
(g)
Aldehyde
C
x
H
y(g)
Olefin
C
x
H
y+2(g)
Paraffin
C
x
H
yC
Hydrogenationsite(s)
CO
A
H*
H
2(g)
CO
(g)
CO
B
CH
3
COOH
A
CH
3
COOH
(g)
Aceticacid
CH
3
CO
A
CH
3
CHO
(g)
Acetaldehyde
CH
x
CH
xA
C
2
H
4(g)
Ethylene
H
x
CO
A
OH*
H
2
O
(g)
CH
xA
CH
4(g)
Methane
CH
3
OH
(g)
Methanol
C
2
H
5
CHO
A
C
2
H
5
CHO
(g)
Propanal
C
2
H
x
CH
xA
C
3
H
6(g)
Propylene
+CO
B
Rh/SiO
2
RhMn/SiO
2
RhFe/SiO
2
RhMnFe/SiO
2
7ModificationofRh/SiO2withMnand/orFe
124
For RhFe/SiO2 the results are reversed and 61 % of the CHxA intermediates are
hydrogenated to form methane while only 39 % undergo CO insertion. This behavior is
consistent with previous indication for a high hydrogenation ability. The values found for
RhMnFe/SiO2 are again between RhFe/SiO2 and the Fe free catalysts.
The formed CHxCOA intermediate can be further converted in three different reactions
(Figure7.19). Hydrogenation and desorption yields acetaldehyde. For calculation of the
probability of acetaldehyde formation the sum of acetaldehyde, ethanol and ethyl acetate
was used since ethanol and ethyl acetate are assumed to be consecutive products from
acetaldehyde. The reaction of CHxCOA with a surface hydroxyl group forms acetic acid
(compare references in chapter 6.7). For calculations, the sum of acetic acid and all acetates
was used. In a third reaction, the C-O bond is dissociated yielding the intermediate for
ethylene and ethane formation. Over Rh/SiO2 the majority (57 %) of the formed CHxCOA
intermediate undergoes C-O bond cleavage. Acetaldehyde and acetic acid are formed with
22 % and 21 % probability, respectively. In this specific step, the results for RhMn/SiO2 are
clearly different. Only 7 % of CHxCOA species undergo C-O bond cleavage. The vast majority
desorbs as acetaldehyde and acetic acid with similar probabilities of 50 % and 43 %,
respectively. This is likely the key effect of Mn addition for higher C2 oxygenate selectivities.
The presence of the Mn oxide appears to be important for stabilization of CHxCOA and
therefore inhibiting C-O bond cleavage in favor of higher C2 oxygenate formation. Thereby,
the evaluation of the initial formation phase in chapter 7.2 suggests that this specific
functionality only evolves during the first days on stream in which rates of ethane and C3
products decrease much faster than products that are formed before the second C-O
cleavage step in the reaction network.
Figure7.19Part3ofthededucedreactionnetworkcontainingthehydrogenationofCHxCOAtoformacetaldehyde,
thereactionofCHxCOAwithsurfaceOH*toformaceticacidandtheC‐ObondcleavageofCHxCOAtoformtheC2
hydrocarbonsurfaceintermediate.Numbersdenotetheprobabilityforthedifferentreactionpathwaysoriginating
fromtheCHxCOAsurfaceintermediateintheorderRh/SiO2(blue),RhMn/SiO2(green),RhFe/SiO2(red)and
RhMnFe/SiO2(purple).Calculatedfromselectivitiesobtainedatstandardconditionsand~5%COconversion.
(54bar,260°C,H2:CO:N2:Ar=60:20:10:10,41.7mlmin‐1,Rh/SiO2:DinoRun26,2500h‐1;RhMn/SiO2:DinoRun26,
12500h‐1;RhFe/SiO2:DinoRun27,8330h‐1;RhMnFe/SiO2:DinoRun26,7580h‐1)
CH
3
COOH
A
CH
3
COOH
(g)
Aceticacid
CH
x
CO
A
CH
3
CHO
(g)
Acetaldehyde
+OH*
CH
x
CH
xA
+H* ‐H
2
O
+H*
57
7
11
1
22
50
71
76
21
43
18
23
+H*
C
2
H
4(g)
Ethylene
+2H*
C
x
H
y
CH
2
OH
(g)
Alcohol
C
x
H
y
CHO
C
C
x
H
y
CHO
(g)
Aldehyde
C
x
H
y(g)
Olefin
C
x
H
y+2(g)
Paraffin
C
x
H
yC
Hydrogenationsite(s)
CO
A
H*
H
2(g)
CO
(g)
CO
B
CH
3
COOH
A
CH
3
COOH
(g)
Aceticacid
CH
3
CO
A
CH
3
CHO
(g)
Acetaldehyde
CH
x
CH
xA
C
2
H
4(g)
Ethylene
H
x
CO
A
OH*
H
2
O
(g)
CH
xA
CH
4(g)
Methane
CH
3
OH
(g)
Methanol
C
2
H
5
CHO
A
C
2
H
5
CHO
(g)
Propanal
C
2
H
x
CH
xA
C
3
H
6(g)
Propylene
+CO
B
7.7Effectsofmodificationonreactionnetwork
125
As expected, hydrogenation towards acetaldehyde is the preferred pathway over RhFe/SiO2
with 71 % probability. In this part of the reaction network, RhMnFe/SiO2 acts very similar
to RhFe/SiO2. Acetic acid is formed with a somewhat lower probability. Faster removal of
OH* as water from the surface due to higher H* availability might be another reason for this
behavior. C-O bond cleavage to form the ethylene/ethane intermediate does not play a
significant role over both catalysts.
The reaction steps in the main network following the formation of CHxCHyA were not
evaluated with the tree probability approach. The adsorption/desorption equilibrium for
ethylene causes a more complex evaluation and concentrations for C3+ products are too
low for the modified catalysts to be further evaluated.
The final part of the reaction network is the consecutive hydrogenation of aldehydes and
olefins. Alcohols and paraffins are assumed to be solely formed by consecutive
hydrogenation of aldehydes and olefins. However, it cannot be fully excluded that
hydrogenation proceeds over different active sites. Rh/SiO2 and RhMn/SiO2 have similar
probabilities for hydrogenation of ethylene to ethane of 78 % and 74 %, respectively
(Figure7.20). The values for propylene hydrogenation are smaller but again very similar.
These findings are fully consistent with the results of the cofeed studies. Over RhFe/SiO2,
both olefins are fully hydrogenated. The trimetallic catalyst shows values between
RhFe/SiO2 and the Fe free catalysts. The obtained hydrogenation probability of propylene
of 59 % is consistent with the value of propylene conversion from the cofeed experiments
of 66 % (compare TableA12 in Appendix E.4). The catalyst behavior regarding
hydrogenation of olefins obtained from the tree probability approach at steady-state
conditions is therefore fully consistent with cofeed experiments indicating that the
underlying mechanism for paraffin formation is plausible.
The hydrogenation probability of acetaldehyde to form ethanol shows is only 11 % for
Rh/SiO2. Here, the obtained conversion from cofeed studies of 94 % clearly deviate. Over
RhMn/SiO2, the hydrogenation probability is increased to 37 % which is still a significantly
lower value than cofeed experiment suggest. Potential reasons are discussed in detail in
chapter 6.5. Propanal hydrogenation probabilities are the same as for acetaldehyde
hydrogenation indicating that chain length here does not have an impact. The Fe containing
catalysts again show superior hydrogenation ability with ~90 % for acetaldehyde
hydrogenation.
7ModificationofRh/SiO2withMnand/orFe
126
Figure7.20Consecutivehydrogenationreactionofintermediatescontainingthehydrogenationofethylene,
propylene,acetaldehydeandpropanal.Numbersdenotetheprobabilityofadesorbedintermediatetobe
hydrogenatedoverthedifferentcatalystsintheorderRh/SiO2(blue),RhMn/SiO2(green),RhFe/SiO2(red)and
RhMnFe/SiO2(purple).Calculatedfromselectivitiesobtainedatstandardconditionsand~5%COconversion.
(54bar,260°C,H2:CO:N2:Ar=60:20:10:10,41.7mlmin‐1,Rh:DinoRun26,170hTOS,2500h‐1;RhMn:DinoRun26,
170hTOS,12500h‐1;RhFe:DinoRun27,190hTOS,8330h‐1;RhMnFe:DinoRun26,170hTOS,7580h‐1)
Together with the experimental data presented in this chapter, the tree probability
evaluation in this section further supports a common reaction network with the catalyst
modification affecting specific surface reaction rates. From the evaluation of reference
conditions, two major effects of Mn and Fe modification on catalytic reactivity can be
suggested. Clearly, the addition of Fe enhances all surface steps that require H* species. It is
therefore likely, that over RhFe particles the surface coverage with H* is significantly
enhanced. As this state affects several surface reactions from the very beginning of the
reaction network, the observed selectivities and behavior upon partial pressure variation
varies significantly from the Fe free catalysts. The effect of Mn addition appears to be much
more specific. The obtained probabilities of reaction pathways differ from Rh/SiO2 only in
one part. The interaction of Rh with Mn oxide appears to have a stabilizing effect on the
CHxCOA intermediate against further C-O bond dissociation. As a result, the reaction does
not proceed and C2 oxygenates are formed with high yields. This very specific impact leads
to very similar behavior of Rh/SiO2 and RhMn/SiO2 upon partial pressure variation. The
tremendously higher CO consumption rates might be primarily a result of higher dispersion
of the Rh particles surrounded by the Mn oxide matrix. Both effects – hydrogenation ability
for Fe addition and stabilization of the C2 oxygenate intermediate for Mn – only evolve
during the first day under reaction conditions which further highlights the impact of the
reactive atmosphere at relevant pressures on the catalyst structure and the importance of
allowing a sufficiently long equilibration for catalytic investigations.
+2H*
C
2
H
4(g)
Ethylene
C
2
H
4C
C
2
H
6(g)
Ethane
Hydrogenationsite
78
74
100
94
+2H*
C
3
H
6(g)
Propylene
C
3
H
6C
C
3
H
8(g)
Propane
Hydrogenationsite
31
26
100
59
+2H*
C
2
H
5
CHO
(g)
Propanal
C
2
H
5
CHO
C
C
3
H
7
OH
(g)
Propanol
Hydrogenationsite
10
36
+2H*
CH
3
CHO
(g)
Acetaldehyde
CH
3
CHO
C
C
2
H
5
OH
(g)
Ethanol
Hydrogenationsite
11
37
91
90
+2H*
C
x
H
y
CH
2
OH
(g)
Alcohol
C
x
H
y
CHO
C
C
x
H
y
CHO
(g)
Aldehyde
C
x
H
y(g)
Olefin
C
x
H
y+2(g)
Paraffin
C
x
H
yC
Hydrogenationsite(s)
CO
A
H*
H
2(g)
CO
(g)
CO
B
CH
3
COOH
A
CH
3
COOH
(g)
Aceticacid
CH
3
CO
A
CH
3
CHO
(g)
Acetaldehyde
CH
x
CH
xA
C
2
H
4(g)
Ethylene
H
x
CO
A
OH*
H
2
O
(g)
CH
xA
CH
4(g)
Methane
CH
3
OH
(g)
Methanol
C
2
H
5
CHO
A
C
2
H
5
CHO
(g)
Propanal
C
2
H
x
CH
xA
C
3
H
6(g)
Propylene
+CO
B
127
8 General Discussion
Six superordinate themes emerged from the data presented in the previous chapters. First,
a general reaction network was derived, and the impact of the catalyst composition
deciphered. A key feature of the reaction network is the existence of two surface CO* species
that exhibit different reactivity. The provided evidence for their existence is illustrated in
the second section of this chapter. The third section demonstrates the effect of catalyst
composition on structure and reactivity. Subsequently, dynamic changes of the catalytic
properties under reaction conditions are assigned to different timescales. The two final
sections conclude on possibilities to steer the catalytic performance and the observed
challenges for practical catalyst operation of Rh-based catalysts for syngas conversion.
8.1 Reaction network for Rh-based catalysts
The derived reaction network for Rh/SiO2 and the impact of catalyst modification on
specific surface steps are described and explained in chapters 6.7 and 7.7, respectively. The
network contains 6 key features (Figure8.1A).
i) at least two different surface CO* types that show different dependency on
reaction conditions and different reactivity (COA and COB)
ii) H-assisted COA activation under formation of a CHxOA fragment
iii) C-O bond dissociation of CHxOA or its higher derivatives
iv) COB insertion as only type of C-C-coupling mechanism
v) Separate hydrogenation reactions that convert aldehydes to alcohols and olefins
to paraffins which might be heavily dependent on gas phase composition
While parts of the reaction network are in accordance with already published networks, the
proposed reaction network in this study is the first network containing all steps and
relevant surface species that describe the obtained data over a large range of reaction
conditions.
In general, the reaction network described in chapter 6.7 applies to all considered catalysts
as the results of cofeed and dropout experiments suggested. However, single surface steps
are considerably enhanced by addition of Mn and/or Fe to the Rh/SiO2 catalysts. The
evaluation of probabilities of individual reaction steps at standard conditions as introduced
in chapter 7.7 gives insight on the impact of the catalyst modification on a mechanistic level.
8GeneralDiscussion
128
Figure8.1Plausiblereactionnetworkforallfourcatalysts(A)andwitharrowsindicatingtheprobabilityofreaction
stepsoverRh/SiO2(B),RhMn/SiO2(C),RhFe/SiO2(D)andRhMnFe/SiO2(E).Mainreactionproductsareindicatedby
coloredfont,sideproductsinblackandproductsthatwereonlyobservedintraceamountsingrey.Calculatedfrom
selectivitiesobtainedatstandardconditionsand~5%COconversionaccordingtoatreeprobabilityapproach.(54
bar,260°C,H2:CO:N2:Ar=60:20:10:10,41.7mlmin‐1,Rh:DinoRun26,170hTOS,2500h‐1;RhMn:DinoRun26,170h
TOS,12500h‐1;RhFe:DinoRun27,190hTOS,8330h‐1;RhMnFe:DinoRun26,170hTOS,7580h‐1)
<10%
10–30%
31–54%
55–80%
>80%
Probabilities
+AcOH
(g)
Methylacetate
CO
A
CO
A
:highcoverage
alreadyatverysmallp
CO
CO
B
:onlyformed
atincreasedp
CO
H*
H
2(g)
CO
(g)
CO
B
Acetates
+Alcohols
(g)
CH
3
COOH
A
CH
3
COOH
(g)
Aceticacid
CH
3
CO
A
CH
3
CHO
(g)
Acetaldehyde
+OH*
CH
x
CH
xA
C
2
H
4(g)
Ethylene
H
x
CO
A
OH
*
H
2
O
(g)
CH
xA
+
H*
CH
4(g)
Methane
CH
3
OH
(g)
Methanol
+CO
B
C
2
H
5
CHO
A
C
2
H
5
CHO
(g)
Propanal
C
2
H
x
CH
xA
C
3
H
6(g)
Propylene
+CO
B
+H*
+
H*
+H* +H*
‐
OH*
+H*
‐
OH*
+CO
B
+H
*
+H*
+
H*
ProposednetworkforRh‐basedcatalysts
A
Propanal/
Propanol
C
2
H
x
CH
xA
CH
3
COOH
A
Aceticacid
C
2
H
5
CO
A
Propylene/
Propane
+H
*
CO
A
H
*
H
2(g)
CO
(g)
CO
B
CH
3
CO
A
Acetaldehyde /
Ethanol
+CO
B
+OH
*
CH
x
CH
xA
Ethylene/Ethane
H
x
CO
A
OH
A
H
2
O
(g)
CH
xA
+H
*
MethaneMethanol
+H
*
+H
*
+H
*
+H
*
+H
*
RhFe/SiO
2
D
+CO
B
RhMnFe/SiO
2
E
CO
A
H
*
H
2(g)
CO
(g)
CO
B
CH
3
CO
A
Acetaldehyde/
Ethanol
+CO
B
+OH
*
CH
x
CH
xA
Ethylene/Ethane
H
x
CO
A
OH
A
H
2
O
(g)
CH
xA
+H
*
MethaneMethanol
+H
*
+H
*
+H
*
+H
*
+H
*
Propanal/
Propanol
C
2
H
x
CH
xA
CH
3
COOH
A
Aceticacid
C
2
H
5
CO
A
Propylene/
Propane
+H
*
+CO
B
RhMn/SiO
2
C
Propanal/
Propanol
C
2
H
x
CH
xA
CO
A
H
*
H
2(g)
CO
(g)
CO
B
CH
3
CO
A
Acetaldehyde/
Ethanol
+CO
B
+OH
*
CH
x
CH
xA
Ethylene /Ethane
H
x
CO
A
OH
A
H
2
O
(g)
CH
xA
+H
*
MethaneMethanol
+H
*
+H
*
+H
*
+H
*
+H
*
CH
3
COOH
A
Aceticacid
C
2
H
5
CO
A
Propylene /
Propane
+H
*
+CO
B
Rh/SiO
2
B
CO
A
H
*
H
2(g)
CO
(g)
CO
B
CH
3
CO
A
Acetaldehyde/
Ethanol
+CO
B
+OH
*
CH
x
CH
xA
Ethylene/Ethane
H
x
CO
A
OH
A
H
2
O
(g)
CH
xA
+H
*
MethaneMethanol
+H
*
+H
*
+H
*
+H
*
+H
*
Propanal/
Propanol
C
2
H
x
CH
xA
CH
3
COOH
A
Aceticacid
C
2
H
5
CO
A
Propylene /
Propane
+H
*
+CO
B
+AcOH
(g)
Ethylacetate
+2H*
C
x
H
y
CH
2
OH
(g)
Alcohol
C
x
H
y
CHO
C
C
x
H
y
CHO
(g)
Aldehyde
C
x
H
y(g)
Olefin
C
x
H
y+2(g)
Paraffin
C
x
H
yC
Hydrogenationsite(s)
CH
3
CH
2
OH
(g)
Ethanol
CH
3
CHO
(g)
Acetaldehyde
+AcOH
(g)
Acetates
8.2EvidencefortwoCO*specieswithdifferentreactivity
129
For the monometallic Rh/SiO2 catalyst, methanol formation happening at the beginning of
the reaction path is negligible (Figure8.1B). The vast majority of formed CHxOA species
undergoes C-O bond cleavage. Although methane is the main product, insertion of a COB
species into CHxA is more likely than hydrogenation of CHxA to form methane. The resulting
CH3COA fragment rather undergoes another C-O bond dissociation process than forming C2
oxygenates by hydrogenation and desorption. This trend is similar to the preference of C-O
bond cleavage over methanol formation earlier in the network. The products observed over
Rh/SiO2 have the highest carbon numbers due to the high probabilities of C-O bond cleavage
and COB insertion building an effective chain elongation cycle.
Most prominently, Mn addition leads to acceleration of the overall CO consumption rate,
probably due to faster COA activation. The obtained probabilities for specific reaction
pathways over RhMn/SiO2 are very comparable to Rh/SiO2 in the first part (Figure8.1C).
These numbers reflect the high similarity of the two catalysts which is further supported by
almost identical reaction orders. However, on RhMn/SiO2 the important CH3COA
intermediate appears to be stabilized against subsequent C-O bond scission which leads to
breaking the chain elongation cycle. CH3COA is preferably converted to the C2 oxygenates
acetic acid and acetaldehyde which is the most important intermediate for the target
product ethanol. In contrast, literature often suggests enhanced CO insertion ability upon
Mn addition.96,97,103 However, the experimental data of this study do not support such an
effect.
The presence of Fe alters the preferred reaction pathway from an early stage of the reaction
network. All reaction pathways involving H* species are favored on RhFe/SiO2. The high
hydrogenation ability leads to desorption of early products of the network – methanol and
methane – before the formation of other products can take place (Figure8.1D). The results
of the probability evaluation at standard conditions therefore fully reflect the findings for
reaction orders and cofeed experiments (compare chapters 7.3 and 7.5) which significantly
differ from the results for Rh/SiO2.
The trimetallic catalyst RhMnFe/SiO2 shows a combination of effects observed for
RhMn/SiO2 and RhFe/SiO2 in most experiments. The probabilities of reaction pathways
fully reflect that behavior. The combination of CH3COA stabilization and high hydrogenation
ability leads to the highest observed ethanol selectivities (Figure8.1E).
8GeneralDiscussion
130
8.2 Evidence for two CO* species with different reactivity
The presence of at least two surface CO* types that behave differently is a key feature of the
presented reaction network. The precise assignment of the nature of these species is beyond
this work. However, the applied CO partial pressure and the time of exposure to CO are the
two main factors responsible for a selectivity shift from mainly C1 and hydrogenated
products to more C2 and unsaturated products.
High CO partial pressures lead to a selectivity shift from products formed in an early stage
of the reaction network – for Rh/SiO2 mainly methane – to products that are only formed
after successful CO insertion. These products include C2 oxygenates and hydrocarbons and
after another C-O bond scission – CO insertion cycle also C3 products (Figure8.2). The
proposed reaction network suggests that a high concentration of CO* species that perform
CO insertion (green COB in Figure8.1) is present at increased CO pressures. At the same
time, increasing CO pressures do not accelerate overall CO consumption rates.
Figure8.2Productselectivitiestowardsproductswithonecarbonatom(blue)andtwocarbonatoms(green)over
Rh/SiO2asafunctionofCOpartialpressure.SchemesillustratetheproposedpresenceofRhparticlesand/orRh
(sub)carbonylsdependingonCOpartialpressure.(54bar,260°C,H2:CO:N2:Ar=45:2.5‐20:25‐42.5:10,41.7mlmin‐1,
DinoRun26,170hTOS,2500h‐1)
In literature, several sites were considered to be responsible for the different reactivities of
adsorbed CO. For example, different facets of Rh crystals were investigated computationally
to find oxygenate formation preferred over Rh(111) compared to Rh(211).169 Other studies
based on spectroscopic data often assign CO dissociation to Rh0 sites while CO insertion is
assigned to Rhn+ where CO adsorbs in the geminal form.72,77 The dynamic catalytic behavior
presented in the preceding chapters of this study clearly favor the later approach. Moreover,
0123456789101112
0
20
40
60
80
100
Product selectivity / %
p
CO
/ bar
C2 products
C1 products
C3 products
CO
OC
OC
OC CO
CO
O
C
SiO
2
CO
CO
CO
CO
SiO
2
COCO
CO
CO
SiO
2
COCO COCO
CO
OC
OC
OC CO
CO
O
C
OC
OC COCO
OC
OC COCO
OC CO CO
8.2EvidencefortwoCO*specieswithdifferentreactivity
131
we could show that product selectivities – and therefore the relative concentrations of both
sites – are strongly dependent on reaction conditions and time on stream. Structural
readjustments are likely necessary to form Rh+ with geminal or Rh (sub)carbonyl species.
The disintegration of Rh (sub)carbonyls from metallic particles might lead to the reduction
of the number of sites suitable for CO activation. This process might be responsible for
changing selectivities at steady CO conversion with increasing CO partial pressures as
summarized above. Carbonyl formation and disintegration has been reported for Rh
nanoparticles64,66,150,170 and Co nanoparticles65,141 in CO containing atmosphere but not for
STE reaction conditions, so far. However, studies on the stability and abundance of such
species under relevant reaction conditions have yet to be provided. The described effect of
CO partial pressure on catalyst structure and reactivity appears to be universal for Rh-based
catalysts, at least within the range of compositions (Rh/Mn/Fe/SiO2) in this study.
The impact of the time component becomes particularly visible during the initial
equilibration phase at the beginning of each run. Rh/SiO2 initially shows a high
hydrogenation activity leading to a high methane selectivity. With time on stream the
hydrogenation ability decreases in favor of more olefin and oxygenate formation (Figure
8.3A). This might potentially be an effect on decreasing metallic Rh surface area by
formation of (sub)carbonyl species. The process is slow in a time range of several days. A
similar behavior has been observed for all four catalysts although the effect over Fe
containing catalysts was less distinct due to a generally enhanced hydrogenation ability. The
initial equilibration phases were evaluated in detail regarding the overall catalytic activity
in chapter 5.2 and regarding the evolution of product formation rates in chapter 6.2 and
chapter 7.2 for Rh/SiO2 and the modified catalysts, respectively.
8GeneralDiscussion
132
Figure8.3COconversionandproductyieldsoverRh/SiO2duringtheinitialequilibrationphaseofDinoRun29(A)
andtheCOdropoutexperimentperformedlaterinthesamerun(B).Whitenumbersdenoteselectivitiesofthe
mainproducts.SchemesillustratetheproposedpresenceofRhparticlesand/orRh(sub)carbonylsattherespective
timeonstream.(DinoRun29,54bar,260°C,H2:CO:N2:Ar=60:20:10:10,COwasreplacedbyadditionalN2forthe
dropoutexperiment,2500h‐1)
Reactant dropout experiments provided more information about the relevant gas phase
composition for catalyst restructuring and its reversibility. After prolonged absence of CO
in the gas phase, some of the initial reactivity can be restored (Figure8.3A). Considering
the availability of COB for CO insertion from a mechanistic view and presence of
(sub)carbonyl species from a structural perspective, this behavior suggests that the species
which formed as a function of CO pressure and time on stream can be removed from the
surface by removal of CO from the gas phase. This process is reversible since the addition of
CO to the gas phase causes the same restructuring process as observed for the initial
equilibration. Also, this experiments clearly proves that the restructuring is induced by gas
phase CO rather than any other gas phase components such as reaction products. The
removal of H2 from the gas phase also stops the reaction for several hours but no effect of
the absence of H2 and reaction products were observed as long as gas phase CO is provided.
Dropout experiments of both reactants were described in detail in chapter 6.6 for Rh/SiO2
and 7.6 for the modified catalysts.
1
2
3
4
5
6
7
8
0
53
41
12
7
1
3
8
14
8
18
0 50 100 150 200 250 300
Time on stream / h
Methane
Ethanol
Ethane
Acetic acid
Acetaldehyde
syngas
CO + H
2
X
CO
and Y
Products
/ %
39 46 41
6
7
6
3
3
3
14
12
13
21
16
19
380 400 420 440 460 480360
0
1
2
3
4
5
6
7
8
CO
2
Ethyl acetate
Methyl acetate
Acetic acid
Propanal
Acetaldehyde
Ethanol
Methanol
1-Butene
Propylene
Ethylene
n-Butane
Propane
Ethane
Methane
CO conversion
Time on stream / h
syngas
CO + H
2
H
2
/N
2
syngas
CO + H
2
Methane
Ethane
Ethanol
Acet-
aldehyde
Acetic acid
SiO
2
COCOCOCO
SiO
2
CO COCO
SiO
2
CO
SiO
2
COCO
SiO
2
AB
CO
OC
OC
OC CO
CO
O
C
CO
OC
OC
OC CO
CO
O
C
OC CO CO
OC
OC COCO
8.3Dynamicsofchangingcatalyticpropertiesunderreactionconditions
133
8.3 Dynamics of changing catalytic properties under
reaction conditions
The dynamics of the catalyst surface structure under reaction conditions have to be
evaluated in different time scales. There are very fast processes happening in the range of
seconds or few minutes such as the equilibration of the coverage. From literature it is known
that CO adsorbs strongly on Rh catalysts and already at low CO pressures high CO coverages
are reached.82,118,148 These processes are too fast to be resolved in the used test setup. Faster
analytical techniques such as mass spectrometry or IR spectroscopy would be required.
More specialized methods – e.g. steady-state isotopic transient kinetic analysis (SSITKA) –
might yield information on coverage dependent processes.171–173
Other processes take place in a time scale of several hours or days. The previously described
restructuring, presumably under formation of Rh (sub)carbonyls, belongs to this category
(Figure8.4). This process causes an initial activity loss in the first 50 – 130 h and goes along
with an increasing availability of COB species leading to more CO insertion activity (compare
Figure8.1). A shift of selectivities from methane to C2 and C3 products and more
oxygenated products takes place. Dropout experiments have shown that these restructuring
processes are reversible.
Apart from carbonyl formation, the detailed evaluation of the initial equilibration phase
suggested that other features such as the interaction of Rh and Mn oxide matrix are
established during that time. Especially the chain length of products indicate that
RhMn/SiO2 behaves similarly to Rh/SiO2 at the very beginning of the reaction. During the
equilibration phase, the formation rates of products with higher carbon numbers decrease
more rapidly until the characteristic stop of the chain growth cycle at the CHxCOA
intermediate leads to the high C2 oxygenate selectivity observed over RhMn/SiO2 at steady-
state conditions. In the case of RhFe/SiO2, the formation of a RhFe phase with increased
hydrogenation ability shifts selectivities from ethanol to methanol by faster hydrogenation
of the CHxOA intermediate. More elaborate description of these initial processes can be
found in chapter 6.2 and 7.2. Likely, the initial determination of metal interactions is not
reversible.
8GeneralDiscussion
134
Figure8.4SchematicrepresentationofchangesinthesurfacestructureofaRhcatalystexposedtorelevant
reactionconditionsasafunctionoftimeonstream(log)anditsimplicationsoncatalyticactivity.Coloredareas
indicateshiftingselectivities.
The gradual slow loss of activity without change of selectivities is a slow process over many
days or weeks. The most likely reason is a slow loss of active surface area or catalyst
poisoning. In contrast to the (sub)carbonyl formation, this process is irreversible. The
catalytic tests suggest that sintering and more generally the mobility of Rh on the surface is
rather dependent on the gas phase composition than on high temperatures (compare 4.3).
(Sub)carbonyl formation therefore might not only be relevant for activity and selectivity
aspects but also for Rh mobility. Particle growth in the presence of high CO and water
pressures has been described for Co catalysts.65,141 Generally, only little information on the
deactivation processes of Rh-based catalysts for CO hydrogenation is available. More
information would be necessary, e.g. in form of long-term studies with permanent in‐situ or
regular ex‐situ characterization of the catalyst structure. Also, the role of different reaction
products such as water and acetic acid should be investigated.
Activity
seconds hours / days days / weeks
SiO
2
Coverage
Restructuration / (sub)carbonyl formation
Sintering / Agglomeration
SiO
2
CO
CO
COCO
CO
CO
SiO
2
CO
CO
CO
CO
CO
CO
SiO
2
Time on stream
C2 products
Oxygenates
Methane
Hydrogenated products
CO
OC
OC
OC CO
CO
O
COC
OC COCO
OC CO CO
CO
OC
OC CO
CO
CO
OC
OC
OC CO
CO
O
C
O
C
OC
O
C
OC
OC
OC CO
CO
CO
CO
8.4Influenceofcatalystcompositiononstructureandreactivity
135
In this study, complex and interconnected effects of reaction conditions and time on stream
dynamics on catalyst structure and performance was demonstrated. Therefore, the
importance of careful interpretation of catalytic data under consideration of the entire
complexity was highlighted. Otherwise, incomplete interpretation might cause misleading
conclusions about structure-performance relationships and prevent an effective
knowledge-based catalyst development.
8.4 Influence of catalyst composition on structure and
reactivity
Catalyst characterization via XRD, electron microscopy, STEM-EDX, and XPS provided
information on catalyst structure before and after reaction. Particle sizes and the elemental
distribution of the metals over the silica support revealed an impact of catalyst composition.
A detailed presentation of the catalyst’s structures can be found in chapter 4.3.
The simplest Rh/SiO2 catalysts exhibits small metallic Rh particles evenly distributed over
the support after H2 treatment. However, after long time on stream very heterogeneous
surface structures featuring large agglomerates of a size up to 200 nm were observed
(Figure8.5). With around 5 nm the obtained mean particle size is the largest among the
tested catalysts. Although the catalyst composition is the simplest, the observed product
spectrum is the most complex including higher aldehydes, olefins and paraffins caused by
an effective C-O bond scission – CO insertion cycle. The high complexity of the product
spectrum and the surface structure causes the poorest reproducibility among the four
considered catalysts (compare 5.3).
Addition of Mn leads to the formation of a Mn oxide matrix surrounding Rh or RhFe
particles. The matrix has a stabilizing effect preventing particle growth and agglomeration
presumably due to reduced Rh mobility. The interaction of Rh with the Mn oxide matrix also
leads to the stabilization of the CHxCOA surface intermediate resulting in a strongly
enhanced C2 oxygenate productivity. The high catalytic activity can be ascribed to a higher
metal dispersion as well as a potentially facilitated CO activation.
The presence of Mn oxide rather has impact on the nanostructure of Rh particles than an
electronic impact. Also, the stabilization of small nanoparticles leads to a much higher
reproducibility when Mn oxide is present. Increased temperatures have a more pronounced
effect on Mn containing catalysts probably because the interaction of Rh or RhFe particles
with the Mn oxide matrix is affected by high temperatures. Also, after a CO dropout the
impact on product formation rates are more permanent than over Mn free catalysts.
8GeneralDiscussion
136
Possible reasons for both phenomena might be a migration of the Mn oxide matrix onto the
nanoparticles or other forms of strong metal support interaction.82
Figure8.5
SchematicrepresentationofthecatalyststructuresasobservedfromSTEM‐EDXanalysiswithcomments
onreproducibilityandreactivityasexemplaryshownbyproductselectivitiesatstandardconditionsand~5%CO
conversion.(54bar,260°C,H
2
:CO:N
2
:Ar=60:20:10:10,41.7mlmin
‐1
,Rh:DinoRun26,170hTOS,2500h
‐1
;RhMn:
DinoRun26,170hTOS,12500h
‐1
;RhFe:DinoRun27,190hTOS,8330h
‐1
;RhMnFe:DinoRun26,170hTOS,7580h
‐1
)
8.5Parametersforsteeringcatalyticperformance
137
The addition of Fe causes the formation of a RhFe phase. The RhFe particles show less
particle growth and a reduced tendency for agglomeration leading to a more uniform
surface structure than Rh/SiO2. The close contact of Fe and Rh suggest an electronic impact
strongly altering reactivity. An enhanced hydrogenation ability leads to a product spectrum
significantly reduced to early product of the reaction network. The more homogeneous
surface structure combined with simplified product spectrum also leads to improved
reproducibility.
The obtained data does not provide information if carbonyl formation might also be of
importance for Mn and Fe. However, no spectroscopic studies could be found describing the
presence of Mn and/or Fe surface carbonyls. Mn carbonyl formation generally requires
harsh conditions and has only been reported for metal organic precursors such as Mn
acetate.168 Therefore, we consider the presence of Mn carbonyls unlikely. Fe carbonyls can
be formed from metallic Fe under relatively mild conditions and can therefore not be fully
excluded.168 However, its oxidic nature when located on the silica support would probably
prevent the formation of carbonyls under the applied reaction conditions. The matter
becomes more complex when Fe is present in RhFe alloy structures. Spectroscopic studies
under reaction conditions would be necessary to provide more reliable information about
dynamic processes on the surface of such particles.
8.5 Parameters for steering catalytic performance
The results presented in the previous chapters have shown two major parameters to steer
the catalyst performance according to the target products. Changing the catalyst
composition can have tremendous impact on the product spectrum. Adjusting reaction
conditions leads to changes in activity and selectivity.
The underlying reaction network is the same for all four tested catalysts. Rh is the crucial
component to determine the syngas chemistry involved. However, the modification with Mn
and/or Fe leads to enhancement of specific surface reaction rates causing severe impact on
product formation rates and therefore selectivities. The presence of a Mn oxide leads to
enhanced C2 oxygenates formation. Considering the possibility to combine RhMn/SiO2 with
a selective hydrogenation catalyst, high ethanol yields might be possible. Further research
on such coupled systems and their interaction would be necessary.
Incorporation of Fe into Rh particles leads to a high combined methanol and ethanol
selectivity which might be beneficial if both alcohols are considered valuable products. The
trimetallic catalyst RhMnFe/SiO2 combines both effects leading to the highest ethanol
selectivity achieved with a single catalyst. A major challenge for further catalyst
8GeneralDiscussion
138
development will be the reduction of methane formation as methane is generally a highly
undesired product of CO hydrogenation.
Reaction conditions and in particular gas phase composition are equally important to
optimize the ethanol yield. Regarding the reaction temperature, the presented results
suggest a maximum temperature of 280 °C to avoid accelerated activity loss. Other than
that, the reaction temperature appears to be of minor importance. The reactant partial
pressures pose the highest impact on product selectivities. However, that matter is complex.
Over all catalysts high CO partial pressures are necessary to suppress methane formation.
For Fe containing catalysts, high CO partial pressures also lead to increased ethanol
selectivities. Over the Fe free catalysts, the contrary behavior was found. In that case, high
ethanol selectivities are always accompanied by high methane selectivities (Figure8.6).
Figure8.6CombinedselectivitytowardsC2oxygenatesincludingacetaldehyde,aceticacid,ethanol,andthe
contributionofacetates,selectivitytowardsethanolandmethaneoverRhMn/SiO2asfunctionsofthereactant
partialpressures.(RhMn/SiO2,DinoRun26,54bar,260°C,H2:CO:N2:Ar=30‐60:2.5‐20:10‐57.5:10,41.7mlmin‐1,
12500h‐1)
Here, another potential advantage of combining a Rh-based catalyst with a second
independent hydrogenation catalyst becomes apparent. Focusing on the formation of other
C2 oxygenates such as acetaldehyde, reaction conditions can be tuned towards high
oxygenate and low methane selectivities applying high CO partial pressures. High H2 partial
pressures ensure high reaction rates but have slightly negative impact on C2 oxygenate
selectivity. An optimal CO:H2 composition would have to be explored as well as a second
catalyst that works efficiently under the respective conditions with a high selectivity
towards the target product ethanol.
8.6Challengesforpracticalcatalystoperation
139
8.6 Challenges for practical catalyst operation
Apart from the discussed possibilities to improve the ethanol productivity by tuning catalyst
composition and reaction conditions, the results in this study also give insight into three
major challenges to transfer the STE reaction into a commercially feasible industrial
process.
i) The resulting ethanol yields in this study tend to be too low for reasonable
commercialization and therefore the development of a universally applicable
process is difficult. However, chances are that for specific business cases feasible
solutions can be developed, e.g. if other products such as methanol or acetic acid
are considered valuable products as well. The benchmarks for required product
yields vary significantly for specific business cases. A broad knowledge of the
syngas chemistry depending on catalyst composition and reaction conditions
are therefore highly important for flexible catalyst development according to
specific requirements.
ii) The derived underlying reaction network is highly complex. The formation
pathways to desirable and undesirable products are closely interconnected and
depend on each other. It must be assumed that the maximum achievable ethanol
selectivity is limited at some point. Optimizing catalyst composition and
reaction conditions requires comprehensive research and ideally a specific
assignment of catalyst properties to surface reaction rates.
iii) High and unstable Rh prices pose another challenge on required catalytic
properties. Excellent long-term stability and recycling options are crucial.
Despite extensive exploration of alternative materials for the STE reaction, Rh-
based catalysts are still the most selective. However, more fundamental
understanding of the dynamics of syngas chemistry on metal surfaces might lead
to new approaches for partial or full replacement of Rh in catalytic materials.
8GeneralDiscussion
140
141
9 Conclusion and Outlook
In the presented study, the examination of the complex interplay of reaction conditions,
catalyst properties and reaction network based on the detailed investigation of catalytic
performance and catalyst properties was described. We provided a comprehensive picture
on the reaction and put it into the context of existing literature.
Four Rh-based catalyst systems with increasingly complex composition were evaluated at
industrially relevant reaction conditions in four catalytic runs including various kinetic and
mechanistic experiments. Catalyst samples have been comprehensively characterized in
several states before and after reaction.
A plausible reaction network has been described for Rh/SiO2. The proposed reaction
network explains all features encountered in the presented catalytic data. Although many
surface species and reaction steps are literature-known, the combination and interplay of
species have not been considered in reported reaction networks so far. The existence of two
adsorbed CO* species with different reactivity is a key feature of the proposed reaction
network. Assignment of specific surface sites to reactivity is beyond the scope of this thesis.
However, strong dependency of the catalyst performance on CO partial pressure indicates
that reactivity might be related to the reversible formation and disintegration of Rh
(sub)carbonyl species. Following this hypothesis, C-O bond breaking and hydrogenation
reactions were related to metallic Rh whereas carbonyl species appeared to be necessary
for CO insertion. Product selectivities then depend on the ratio of metallic Rh to Rh carbonyl
sites. The dynamic formation and destruction of such structures dependent on CO partial
pressure and time on stream has been demonstrated based on catalytic data during the
initial formation phase, partial pressure variation and CO dropout experiments. Also, the
formation of carbonyls might be realted to high mobility of Rh on the silica surface leading
to formation of large agglomerates of Rh particles.
The reaction network developed for Rh/SiO2 has proved applicable for the modified
catalysts as well. No evidence for a fundamental change of the reaction network was found.
Rather, modification caused acceleration or inhibition of specific surface steps. Evaluation
of reaction pathways suggested that the presence of Mn oxide stabilizes the C2 oxygenate
intermediate which leads to significantly enhanced C2 oxygenate yields over RhMn/SiO2.
Increased overall rates were predominantly assigned to a high Rh dispersion due to
stabilized small Rh particles embedded in a Mn oxide matrix. Fe addition leads to highly
accelerated rates for hydrogenation steps and therefore affects many surface reactions over
RhFe/SiO2. Reactivity was related to the formation of RhFe phases with intimate contact of
9ConclusionandOutlook
142
both metals. Likely, Fe provides higher availability of H* species. The behavior of the
trimetallic RhMnFe/SiO2 catalyst behaves as a combination of the two bimetallic catalysts
in almost all cases, providing the highest ethanol selectivities among the tested catalysts.
The highest ethanol formation rates were observed over RhMnFe/SiO2 at high H2 and CO
partial pressures. However, possibilities to optimize reaction conditions for high ethanol
yields are limited as methane formation is promoted simultaneously due to highly
interconnected reaction pathways. The optimization of reaction conditions for high
acetaldehyde and acetic acid yields over RhMn/SiO2 and subsequent selective
hydrogenation to ethanol has therefore been proposed to show the highest potential as
basis for future commercialization efforts.
Throughout the study, the complexity of the tested catalyst systems regarding industrially
relevant reaction conditions, dynamic structural changes in different timescales and precise
evaluation of the catalyst performance including all reaction products was highlighted.
Contradicting results in literature often could be assigned to interpretation based on too
simplified assumptions.
To drive the presented research further, more research must be conducted in different
directions. From a molecular view, the formation, stability, and reactivity of carbonyl
species at industrially relevant reaction conditions must be investigated. Overall,
development of options for operando characterization will be crucial to further investigate
those systems which behave highly dependent on reaction conditions. Moreover, materials
should be evaluated regarding their carbonyl chemistry to potentially identify suitable
substitutes for Rh in this regard. Additional catalyst compositions based on the Rh/SiO2
catalyst should be tested to provide more information on the specific impact of each metal
on the reaction network. A library of such effects will be required for more knowledge-
based catalyst development. Finally, the development of a potential tandem catalyst system
for C2 oxygenate formation and subsequent hydrogenation involves detailed research. A
hydrogenation catalyst that is active and selective at respective reaction conditions must be
found and the interaction of the two catalyst systems must be explored in different reactor
and stacking geometries to optimize catalytic performance.
143
References
(1) StatistaGmbH.WeltweiteProduktionvonEthanolfürKraftstoffeindenJahren2000bis2018(in
MillionenKubikmeter).
https://de.statista.com/statistik/daten/studie/198606/umfrage/produktion‐von‐ethanol‐
weltweit‐seit‐2000/(accessedMay24,2020).
(2) Spath,P.L.;Dayton,D.C.PreliminaryScreening‐‐TechnicalandEconomicAssessmentof
SynthesisGastoFuelsandChemicalswithEmphasisonthePotentialforBiomass‐DerivedSyngas.
Natl.Renew.EnergyLab.2003,No.December,1–160.https://doi.org/10.2172/15006100.
(3) Thomas,V.;Kwong,A.EthanolasaLeadReplacement:PhasingoutLeadedGasolineinAfrica.
EnergyPolicy2001,29(13),1133–1143.https://doi.org/10.1016/S0301‐4215(01)00041‐6.
(4) Song,C.L.;Zhang,W.M.;Pei,Y.Q.;Fan,G.L.;Xu,G.P.ComparativeEffectsofMTBEandEthanol
AdditionsintoGasolineonExhaustEmissions.Atmos.Environ.2006,40(11),1957–1970.
https://doi.org/10.1016/j.atmosenv.2005.11.028.
(5) Ferreira‐Aparicio,P.;Benito,M.J.;Sanz,J.L.NewTrendsinReformingTechnologies:From
HydrogenIndustrialPlantstoMultifuelMicroreformers.Catal.Rev.‐Sci.Eng.2005,47(4),491–
588.https://doi.org/10.1080/01614940500364958.
(6) Akhairi,M.A.F.;Kamarudin,S.K.CatalystsinDirectEthanolFuelCell(DEFC):AnOverview.Int.J.
HydrogenEnergy2016,41(7),4214–4228.https://doi.org/10.1016/j.ijhydene.2015.12.145.
(7) Vaidya,P.D.;Rodrigues,A.E.InsightintoSteamReformingofEthanoltoProduceHydrogenfor
FuelCells.Chem.Eng.J.2006,117(1),39–49.https://doi.org/10.1016/j.cej.2005.12.008.
(8) Galvita,V.V.;Semin,G.L.;Belyaev,V.D.;Semikolenov,V.A.;Tsiakaras,P.;Sobyanin,V.A.
SynthesisGasProductionbySteamReformingofEthanol.Appl.Catal.AGen.2001,220(1–2),
123–127.https://doi.org/10.1016/S0926‐860X(01)00708‐6.
(9) Salge,J.R.;Deluga,G.A.;Schmidt,L.D.CatalyticPartialOxidationofEthanoloverNobleMetal
Catalysts.J.Catal.2005,235(1),69–78.https://doi.org/10.1016/j.jcat.2005.07.021.
(10) Wang,W.;Wang,Y.ThermodynamicAnalysisofHydrogenProductionviaPartialOxidationof
Ethanol.Int.J.HydrogenEnergy2008,33(19),5035–5044.
https://doi.org/10.1016/j.ijhydene.2008.07.086.
(11) Subramani,V.;Gangwal,S.K.AReviewofRecentLiteraturetoSearchforanEfficientCatalytic
ProcessfortheConversionofSyngastoEthanol.Energy&Fuels2008,22(2),814–839.
https://doi.org/10.1021/ef700411x.
(12) Gray,K.A.;Zhao,L.;Emptage,M.Bioethanol.Curr.Opin.Chem.Biol.2006,10(2),141–146.
https://doi.org/10.1016/j.cbpa.2006.02.035.
(13) Rostrup‐Nielsen,J.R.CHEMISTRY:MakingFuelsfromBiomass.Science(80‐.).2005,308(5727),
1421–1422.https://doi.org/10.1126/science.1113354.
(14) Bai,F.W.;Anderson,W.A.;Moo‐Young,M.EthanolFermentationTechnologiesfromSugarand
StarchFeedstocks.Biotechnol.Adv.2008,26(1),89–105.
https://doi.org/10.1016/j.biotechadv.2007.09.002.
144
(15) CatalysisinC1Chemistry;Keim,W.,Ed.;CatalysisbyMetalComplexes;SpringerNetherlands:
Dordrecht,1983;Vol.4.https://doi.org/10.1007/978‐94‐009‐7040‐3.
(16) Younesi,H.;Najafpour,G.;Mohamed,A.R.EthanolandAcetateProductionfromSynthesisGas
viaFermentationProcessesUsingAnaerobicBacterium,ClostridiumLjungdahlii.Biochem.Eng.J.
2005,27(2),110–119.https://doi.org/10.1016/j.bej.2005.08.015.
(17) Luk,H.T.;Mondelli,C.;Ferré,D.C.;Stewart,J.A.;Pérez‐Ramírez,J.StatusandProspectsin
HigherAlcoholsSynthesisfromSyngas.Chem.Soc.Rev.2017,46(5),1358–1426.
https://doi.org/10.1039/C6CS00324A.
(18) Horton,T.;Jevtic,R.;Johnston,V.J.;Pan,T.;Warner,R.J.;Weiner,H.;Bower,N.;Chapman,J.T.;
Grusendorf,G.PROCESSESFORMAXIMIZINGETHANOLFORMATIONINTHEHYDROGENATIONOF
ACETICACID.USPat.2010,US8,552,224B2.
(19) Johnston,V.J.;Orosco,A.;Sarager,L.;Scates,M.O.;Shaver,R.D.;Zink,J.H.PRODUCING
ETHANOLUSINGTWODIFFERENTSTREAMSFROMACETICACIDCARBONYLATIONPROCESS.US
Pat.2011,US8,809,598B2.
(20) Lin,J.‐J.;Knifton,J.F.PROCESSFORHOMOLOGATIONOFMETHONALTOETHANOL.USPat.1982,
US4,424,384.
(21) Gong,J.;Yue,H.;Zhao,Y.;Zhao,S.;Zhao,L.;Lv,J.;Wang,S.;Ma,X.SynthesisofEthanolvia
SyngasonCu/SiO2CatalystswithBalancedCu0‐Cu+Sites.J.Am.Chem.Soc.2012,134(34),
13922–13925.https://doi.org/10.1021/ja3034153.
(22) Yue,H.;Ma,X.;Gong,J.AnAlternativeSyntheticApproachforEfficientCatalyticConversionof
SyngastoEthanol.Acc.Chem.Res.2014,47(5),1483–1492.https://doi.org/10.1021/ar4002697.
(23) Liu,Y.;Murata,K.;Inaba,M.;Takahara,I.SynthesisofEthanolfromMethanolandSyngasthrough
anIndirectRouteContainingMethanolDehydrogenation,DMECarbonylation,andMethyl
AcetateHydrogenolysis.FuelProcess.Technol.2013,110,206–213.
https://doi.org/10.1016/j.fuproc.2012.12.016.
(24) Wender,I.ReactionsofSynthesisGas.FuelProcess.Technol.1996,48(3),189–297.
https://doi.org/10.1016/S0378‐3820(96)01048‐X.
(25) Chuang,S.S.C.;Zhang,L.ConversionofSyngastoFuelsandChemicals.InHandbookofClimate
ChangeMitigationandAdaptation;Chen,W.‐Y.,Suzuki,T.,Lackner,M.,Eds.;SpringerNewYork:
NewYork,NY,2015.https://doi.org/10.1007/978‐1‐4614‐6431‐0.
(26) Bell,A.T.CatalyticSynthesisofHydrocarbonsoverGroupVIIIMetals.ADiscussionofthe
ReactionMechanism.Catal.Rev.1981,23(1–2),203–232.
https://doi.org/10.1080/03602458108068076.
(27) Huheey,J.E.;Keiter,E.A.;Keiter,R.L.AnorganischeChemie‐PrinzipienvonStrukturUnd
Reaktivität;Steudel,R.,Ed.;DeGruyter:Berlin,Boston,2012.
https://doi.org/10.1515/9783110249088.
(28) VanSanten,R.A.;Neurock,M.ConceptsinTheoreticalHeterogeneousCatalyticReactivity.Catal.
Rev.1995,37(4),557–698.https://doi.org/10.1080/01614949508006451.
(29) VanDerLaan,G.P.;Beenackers,A.A.C.M.KineticsandSelectivityoftheFischer‐Tropsch
Synthesis:ALiteratureReview.Catal.Rev.‐Sci.Eng.1999,41(3–4),255–318.
https://doi.org/10.1081/CR‐100101170.
145
(30) Davis,B.H.Fischer‐TropschSynthesis:ReactionMechanismsforIronCatalysts.Catal.Today
2009,141(1–2),25–33.https://doi.org/10.1016/j.cattod.2008.03.005.
(31) Dry,M.E.HighQualityDieselviatheFischer‐TropschProcess‐AReview.J.Chem.Technol.
Biotechnol.2002,77(1),43–50.https://doi.org/10.1002/jctb.527.
(32) Ernst,B.;Libs,S.;Chaumette,P.;Kiennemann,A.PreparationandCharacterizationofFischer‐
TropschActiveCo/SiO2Catalysts.Appl.Catal.AGen.1999,186(1–2),145–168.
https://doi.org/10.1016/S0926‐860X(99)00170‐2.
(33) Prieto,G.;DeMello,M.I.S.;Concepción,P.;Murciano,R.;Pergher,S.B.C.;Martĺnez,A.Cobalt‐
CatalyzedFischer‐TropschSynthesis:ChemicalNatureoftheOxideSupportasaPerformance
Descriptor.ACSCatal.2015,5(6),3323–3335.https://doi.org/10.1021/acscatal.5b00057.
(34) DeSmit,E.;Weckhuysen,B.M.TheRenaissanceofIron‐BasedFischer‐TropschSynthesis:Onthe
MultifacetedCatalystDeactivationBehaviour.Chem.Soc.Rev.2008,37(12),2758–2781.
https://doi.org/10.1039/b805427d.
(35) Li,S.;Krishnamoorthy,S.;Li,A.;Meitzner,G.D.;Iglesia,E.PromotedIron‐BasedCatalystsforthe
Fischer‐TropschSynthesis:Design,Synthesis,SiteDensities,andCatalyticProperties.J.Catal.
2002,206(2),202–217.https://doi.org/10.1006/jcat.2001.3506.
(36) Bukur,D.B.;Mukesh,D.;Patel,S.A.PromoterEffectsonPrecipitatedIronCatalystsforFischer‐
TropschSynthesis.Ind.Eng.Chem.Res.1990,29(2),194–204.
https://doi.org/10.1021/ie00098a008.
(37) Shetty,S.;VanSanten,R.A.CODissociationonRuandCoSurfaces:TheInitialStepintheFischer‐
TropschSynthesis.Catal.Today2011,171(1),168–173.
https://doi.org/10.1016/j.cattod.2011.04.006.
(38) Ciobîcǎ,I.M.;Kramer,G.J.;Ge,Q.;Neurock,M.;VanSanten,R.A.MechanismsforChainGrowth
inFischer‐TropschSynthesisoverRu(0001).J.Catal.2002,212(2),136–144.
https://doi.org/10.1006/jcat.2002.3742.
(39) Kung,H.H.MethanolSynthesis.Catal.Rev.1980,22(2),235–259.
https://doi.org/10.1080/03602458008066535.
(40) Klier,K.MethanolSynthesis.InAdvancesinCatalysis;1982;Vol.31,pp243–313.
https://doi.org/10.1016/S0360‐0564(08)60455‐1.
(41) Waugh,K.C.MethanolSynthesis.Catal.Today1992,15(1),51–75.
https://doi.org/10.1016/0920‐5861(92)80122‐4.
(42) Lange,J.P.MethanolSynthesis:AShortReviewofTechnologyImprovements.Catal.Today2001,
64(1–2),3–8.https://doi.org/10.1016/S0920‐5861(00)00503‐4.
(43) Ali,K.A.;Abdullah,A.Z.;Mohamed,A.R.RecentDevelopmentinCatalyticTechnologiesfor
MethanolSynthesisfromRenewableSources:ACriticalReview.Renew.Sustain.EnergyRev.
2015,44,508–518.https://doi.org/10.1016/j.rser.2015.01.010.
(44) Robbins,J.L.;Iglesia,E.;Kelkar,C.P.;DeRites,B.MethanolSynthesisoverCu/SiO2Catalysts.
Catal.Letters1991,10(1–2),1–10.https://doi.org/10.1007/BF00764730.
146
(45) Yang,Y.;Mims,C.A.;Mei,D.H.;Peden,C.H.F.;Campbell,C.T.MechanisticStudiesofMethanol
SynthesisoverCufromCO/CO2/H2/H2OMixtures:TheSourceofCinMethanolandtheRoleof
Water.J.Catal.2013,298,10–17.https://doi.org/10.1016/j.jcat.2012.10.028.
(46) Nguyen,H.P.;Meriaudeau,P.;Naccache,C.MethanolSynthesisonSilica‐SupportedPlatinum
Catalysts:TheEffectofChromiumAdditive.Appl.Catal.1986,21(2),337–347.
https://doi.org/10.1016/S0166‐9834(00)81366‐4.
(47) Mériaudeau,P.;Albano,K.;Naccache,C.PromotionofPlatinum‐BasedCatalystsforMethanol
SynthesisfromSyngas.J.Chem.Soc.FaradayTrans.1Phys.Chem.Condens.Phases1987,83(7),
2113–2118.https://doi.org/10.1039/F19878302113.
(48) Xu,J.;Su,X.;Liu,X.;Pan,X.;Pei,G.;Huang,Y.;Wang,X.;Zhang,T.;Geng,H.MethanolSynthesis
fromCO2andH2overPd/ZnO/Al2O3:CatalystStructureDependenceofMethanolSelectivity.
Appl.Catal.AGen.2016,514,51–59.https://doi.org/10.1016/j.apcata.2016.01.006.
(49) Koningsberger,D.C.;Borgmans,C.P.J.H.;VanElderen,A.M.J.;Kip,B.J.;Niemantsverdriet,J.W.
MethanolfromSynthesisGasoverIron‐RichIron‐IridiumonSilicaCatalysts.J.Chem.Soc.Chem.
Commun.1987,No.12,892–894.https://doi.org/10.1039/C39870000892.
(50) Spencer,M.S.TheRoleofZincOxideinCu/ZnOCatalystsforMethanolSynthesisandtheWater‐
GasShiftReaction.Top.Catal.1999,8(3–4),259–266.
https://doi.org/10.1023/A:1019181715731.
(51) NaumannD’Alnoncourt,R.;Xia,X.;Strunk,J.;Löffler,E.;Hinrichsen,O.;Muhler,M.TheInfluence
ofStronglyReducingConditionsonStrongMetal‐SupportInteractionsinCu/ZnOCatalystsUsed
forMethanolSynthesis.Phys.Chem.Chem.Phys.2006,8(13),1525–1538.
https://doi.org/10.1039/b515487a.
(52) Herman,R.G.;Klier,K.;Simmons,G.W.;Finn,B.P.;Bulko,J.B.;Kobylinski,T.P.Catalytic
SynthesisofMethanolfromCO/H2.I.PhaseComposition,ElectronicPropertiesandActivityofthe
Cu/ZnO/M2O3Catalysts.Preprints1978,23(2),595–615.
(53) Behrens,M.;Studt,F.;Kasatkin,I.;Kuhl,S.;Havecker,M.;Abild‐Pedersen,F.;Zander,S.;Girgsdies,
F.;Kurr,P.;Kniep,B.‐L.;Tovar,M.;Fischer,R.W.;Norskov,J.K.;Schlogl,R.TheActiveSiteof
MethanolSynthesisoverCu/ZnO/Al2O3IndustrialCatalysts.Science(80‐.).2012,336(6083),
893–897.https://doi.org/10.1126/science.1219831.
(54) Chuang,S.S.C.;Stevens,R.W.;Khatri,R.MechanismofC2+OxygenateSynthesisonRhCatalysts.
Top.Catal.2005,32(3–4),225–232.https://doi.org/10.1007/s11244‐005‐2897‐2.
(55) Watson,P.R.;Somorjai,G.A.TheHydrogenationofCarbonMonoxideoverRhodiumOxide
Surfaces.LawrenceBerkeleyNatl.Lab.1981,LBL12750.
(56) Roelen,O.PRODUCTIONOFOXYGENATEDCARBONCOMPOUNDS.USPat.1939,US2,327,066.
(57) Franke,R.;Selent,D.;Börner,A.AppliedHydroformylation.Chem.Rev.2012,112(11),5675–
5732.https://doi.org/10.1021/cr3001803.
(58) Pospech,J.;Fleischer,I.;Franke,R.;Buchholz,S.;Beller,M.AlternativeMetalsforHomogeneous
CatalyzedHydroformylationReactions.Angew.Chemie‐Int.Ed.2013,52(10),2852–2872.
https://doi.org/10.1002/anie.201208330.
147
(59) Evans,D.;Osborn,J.A.;Wilkinson,G.HydroformylationofAlkenesbyUseofRhodiumComplex
Catalysts.J.Chem.Soc.AInorganic,Phys.Theor.1968,566(1966),3133–3142.
https://doi.org/10.1039/J19680003133.
(60) Wilson,J.;DeGroot,C.Atomic‐ScaleRestructuringinHigh‐PressureCatalysis.J.Phys.Chem.
1995,99(20),7860–7866.https://doi.org/10.1021/j100020a005.
(61) Bezemer,G.L.;Bitter,J.H.;Kuipers,H.P.C.E.;Oosterbeek,H.;Holewijn,J.E.;Xu,X.;Kapteijn,F.;
VanDiilen,A.J.;DeJong,K.P.CobaltParticleSizeEffectsintheFischer‐TropschReactionStudied
withCarbonNanofiberSupportedCatalysts.J.Am.Chem.Soc.2006,128(12),3956–3964.
https://doi.org/10.1021/ja058282w.
(62) Pereira,E.B.;Martin,G.A.AlcoholSynthesisfromSyngasoverNickelCatalysts:EffectofCopper
andLithiumAddition.Appl.Catal.A,Gen.1993,103(2),291–309.https://doi.org/10.1016/0926‐
860X(93)85059‐X.
(63) Kruse,N.;Gaussmann,A.CO‐InducedMorphologicalChangesofRhCrystallites:Mechanisms,
Kinetics,andReal‐SpaceImagingontheAtomicScale.J.Catal.1993,144,525–543.
(64) Suzuki,A.;Inada,Y.;Yamaguchi,A.;Chihara,T.;Yuasa,M.;Nomura,M.;Iwasawa,Y.TimeScale
andElementaryStepsofCO‐InducedDisintegrationofSurfaceRhodiumClusters.Angew.Chemie
‐Int.Ed.2003,42(39),4795–4799.https://doi.org/10.1002/anie.200352318.
(65) JanseVanRensburg,W.;VanHelden,P.;Moodley,D.J.;Claeys,M.;Petersen,M.A.;VanSteen,E.
RoleofTransientCo‐SubcarbonylsinOstwaldRipeningSinteringofCobaltSupportedonγ‐
AluminaSurfaces.J.Phys.Chem.C2017,121(31),16739–16753.
https://doi.org/10.1021/acs.jpcc.7b01907.
(66) Ouyang,R.;Liu,J.X.;Li,W.X.AtomisticTheoryofOstwaldRipeningandDisintegrationof
SupportedMetalParticlesunderReactionConditions.J.Am.Chem.Soc.2013,135(5),1760–
1771.https://doi.org/10.1021/ja3087054.
(67) Ellgen,P.C.;Bhasin,M.M.PROCESSFORPRODUCINGETHANOL,ACETICACIDAND/OR
ACETALDEHYDE,FROMSYNTHESISGAS.USPat.1976,US4,096,164.
(68) vanderLee,G.;Schuller,B.;Post,H.;Favre,T.L.F.;Ponec,V.OntheSelectivityofRhCatalystsin
theFormationofOxygenates.J.Catal.1986,98(2),522–529.https://doi.org/10.1016/0021‐
9517(86)90340‐4.
(69) Ichikawa,M.CatalyticSynthesisofEthanolfromCOandH2underAtmosphericPressureover
PyrolysedRhodiumCarbonylClustersonTiO2,ZrO2,andLa2O3.J.Chem.Soc.Chem.Commun.
1978,1500(13),566–567.https://doi.org/10.1039/C39780000566.
(70) Bhasin,M.M.;Bartley,W.J.;Ellgen,P.C.;Wilson,T.P.SynthesisGasConversionoverSupported
RhodiumandRhodium‐IronCatalysts.J.Catal.1978,54(2),120–128.
https://doi.org/10.1016/0021‐9517(78)90035‐0.
(71) Spivey,J.J.;Egbebi,A.HeterogeneousCatalyticSynthesisofEthanolfromBiomass‐Derived
Syngas.Chem.Soc.Rev.2007,36(9),1514.https://doi.org/10.1039/b414039g.
(72) Ao,M.;Pham,G.H.;Sunarso,J.;Tade,M.O.;Liu,S.ActiveCentersofCatalystsforHigherAlcohol
SynthesisfromSyngas:AReview.ACSCatal.2018,8(8),7025–7050.
https://doi.org/10.1021/acscatal.8b01391.
148
(73) Fan,Z.;Chen,W.;Pan,X.;Bao,X.CatalyticConversionofSyngasintoC2OxygenatesoverRh‐
BasedCatalysts‐EffectofCarbonSupports.Catal.Today2009,147(2),86–93.
https://doi.org/10.1016/j.cattod.2009.03.004.
(74) Solymosi,F.;Tombácz,I.;Kocsis,M.HydrogenationofCOonSupportedRhCatalysts.J.Catal.
1982,75(1),78–93.https://doi.org/10.1016/0021‐9517(82)90123‐3.
(75) Yang,N.;Yoo,J.S.;Schumann,J.;Bothra,P.;Singh,J.A.;Valle,E.;Abild‐Pedersen,F.;Nørskov,J.
K.;Bent,S.F.Rh‐MnOInterfaceSitesFormedbyAtomicLayerDepositionPromoteSyngas
ConversiontoHigherOxygenates.ACSCatal.2017,7(9),5746–5757.
https://doi.org/10.1021/acscatal.7b01851.
(76) Sheerin,E.;Reddy,G.K.;Smirniotis,P.EvaluationofRh/CexTi1‐XO2CatalystsforSynthesisof
OxygenatesfromSyngasUsingXPSandTPRTechniques.Catal.Today2016,263,75–83.
https://doi.org/10.1016/j.cattod.2015.07.050.
(77) Wang,Y.;Song,Z.;Ma,D.;Luo,H.;Liang,D.;Bao,X.CharacterizationofRh‐BasedCatalystswith
EPR,TPR,IRandXPS.J.Mol.Catal.AChem.1999,149(1–2),51–61.
https://doi.org/10.1016/S1381‐1169(99)00181‐8.
(78) Wang,Y.;Luo,H.;Liang,D.;Bao,X.DifferentMechanismsfortheFormationofAcetaldehydeand
EthanolontheRh‐BasedCatalysts.J.Catal.2000,196(1),46–55.
https://doi.org/10.1006/jcat.2000.3026.
(79) Abdelsayed,V.;Shekhawat,D.;Poston,J.A.;Spivey,J.J.Synthesis,Characterization,andCatalytic
ActivityofRh‐BasedLanthanumZirconatePyrochloresforHigherAlcoholSynthesis.Catal.Today
2013,207,65–73.https://doi.org/10.1016/j.cattod.2012.07.011.
(80) Duncan,T.M.StudiesofAdsorbateStructureandDynamicsonCatalyticSurfaceswithNMR
Spectroscopy:COonMetals.ColloidsandSurfaces1990,45(C),11–31.
https://doi.org/10.1016/0166‐6622(90)80011‐R.
(81) Chen,Y.;Zhang,H.;Ma,H.;Qian,W.;Jin,F.;Ying,W.DirectConversionofSyngastoEthanolover
Rh–Fe/Γ‐Al2O3Catalyst:PromotionEffectofLi.Catal.Letters2018,148(2),691–698.
https://doi.org/10.1007/s10562‐017‐2202‐6.
(82) Dimitrakopoulou,M.;Huang,X.;Kröhnert,J.;Teschner,D.;Praetz,S.;Schlesiger,C.;Malzer,W.;
Janke,C.;Schwab,E.;Rosowski,F.;Kaiser,H.;Schunk,S.;Schlögl,R.;Trunschke,A.Insightsinto
StructureandDynamicsof(Mn,Fe)O:X‐PromotedRhNanoparticles.FaradayDiscuss.2018,208,
207–225.https://doi.org/10.1039/c7fd00215g.
(83) Mao,W.;Su,J.;Zhang,Z.;Xu,X.C.;Fu,D.;Dai,W.;Xu,J.;Zhou,X.;Han,Y.F.AMechanisticBasis
fortheEffectsofMnLoadingonC2+OxygenatesSynthesisDirectlyfromSyngasoverRh‐
MnOx/SiO2Catalysts.Chem.Eng.Sci.2015,135,301–311.
https://doi.org/10.1016/j.ces.2015.03.023.
(84) Mao,W.;Su,J.;Zhang,Z.;Xu,X.C.;Dai,W.;Fu,D.;Xu,J.;Zhou,X.;Han,Y.F.KineticsStudyofC2+
OxygenatesSynthesisfromSyngasoverRh‐MnOx/SiO2Catalysts.Chem.Eng.Sci.2014,135,312–
322.https://doi.org/10.1016/j.ces.2015.02.035.
(85) Burch,R.;Petch,M.I.InvestigationoftheSynthesisofOxygenatesfromCarbon‐Monoxide
HydrogenMixturesonSupportedRhodiumCatalysts.Appl.Catal.A‐General1992,88(1),39–60.
149
(86) Haider,M.A.;Gogate,M.R.;Davis,R.J.Fe‐PromotionofSupportedRhCatalystsforDirect
ConversionofSyngastoEthanol.J.Catal.2009,261(1),9–16.
https://doi.org/10.1016/j.jcat.2008.10.013.
(87) Liu,Y.;Murata,K.;Inaba,M.;Takahara,I.;Okabe,K.SynthesisofEthanolfromSyngasover
Rh/Ce1‐XZrXO2Catalysts.Catal.Today2011,164(1),308–314.
https://doi.org/10.1016/j.cattod.2010.10.087.
(88) Liu,J.;Tao,R.;Guo,Z.;Regalbuto,J.R.;Marshall,C.L.;Klie,R.F.;Miller,J.T.;Meyer,R.J.Selective
AdsorptionofManganeseontoRhodiumforOptimizedMn/Rh/SiO2AlcoholSynthesisCatalysts.
ChemCatChem2013,5(12),3665–3672.https://doi.org/10.1002/cctc.201300479.
(89) Liu,Y.;Göeltl,F.;Ro,I.;Ball,M.R.;Sener,C.;Aragão,I.B.;Zanchet,D.;Huber,G.W.;Mavrikakis,
M.;Dumesic,J.A.SynthesisGasConversionoverRh‐BasedCatalystsPromotedbyFeandMn.ACS
Catal.2017,7(7),4550–4563.https://doi.org/10.1021/acscatal.7b01381.
(90) Hu,J.;Wang,Y.;Cao,C.;Elliott,D.C.;Stevens,D.J.;White,J.F.ConversionofBiomass‐Derived
SyngastoAlcoholsandC2OxygenatesUsingSupportedRhCatalystsinaMicrochannelReactor.
Catal.Today2007,120(1),90–95.https://doi.org/10.1016/j.cattod.2006.07.006.
(91) Yu,J.;Mao,D.;Han,L.;Guo,Q.;Lu,G.ConversionofSyngastoC2+OxygenatesoverRh‐
Based/SiO2Catalyst:ThePromotingEffectofFe.J.Ind.Eng.Chem.2013,19(3),806–812.
https://doi.org/10.1016/j.jiec.2012.10.021.
(92) Yu,J.;Mao,D.;Han,L.;Guo,Q.;Lu,G.TheEffectofFeontheCatalyticPerformanceofRh‐Mn‐
Li/SiO2Catalyst:ADRIFTSStudy.Catal.Commun.2012,27,1–4.
https://doi.org/10.1016/j.catcom.2012.06.010.
(93) Song,X.;Ding,Y.;Chen,W.;Dong,W.;Pei,Y.;Zang,J.;Yan,L.;Lu,Y.BimetalModifiedOrdered
MesoporousCarbonasaSupportofRhCatalystforEthanolSynthesisfromSyngas.Catal.
Commun.2012,19,100–104.https://doi.org/10.1016/j.catcom.2011.12.015.
(94) Huang,Y.;Deng,W.;Guo,E.;Chung,P.W.;Chen,S.;Trewyn,B.G.;Brown,R.C.;Lin,V.S.Y.
MesoporousSilicaNanoparticle‐StabilizedandManganese‐ModifiedRhodiumNanoparticlesas
CatalystsforHighlySelectiveSynthesisofEthanolandAcetaldehydefromSyngas.ChemCatChem
2012,4(5),674–680.https://doi.org/10.1002/cctc.201100460.
(95) Kim,T.W.;Kim,M.J.;Chae,H.J.;Ha,K.S.;Kim,C.U.OrderedMesoporousCarbonSupported
UniformRhodiumNanoparticlesasCatalystsforHigherAlcoholSynthesisfromSyngas.Fuel2015,
160(July),393–403.https://doi.org/10.1016/j.fuel.2015.07.062.
(96) Wang,S.R.;Guo,W.W.;Wang,H.X.;Zhu,L.J.;Qiu,K.Z.InfluenceofMnPromotiononCO
HydrogenationoverRh/CNTsCatalyst.Catal.Letters2014,144(7),1305–1312.
https://doi.org/10.1007/s10562‐014‐1248‐y.
(97) Ojeda,M.;Granados,M.L.;Rojas,S.;Terreros,P.;Garcıa‐Garcıa,F.J.;Fierro,J.L.G.Manganese‐
PromotedRh/Al2O3forC2‐OxygenatesSynthesisfromSyngas.Appl.Catal.AGen.2004,261(1),
47–55.https://doi.org/10.1016/j.apcata.2003.10.033.
(98) Egbebi,A.;Spivey,J.J.EffectofH2/CORatioandTemperatureonMethaneSelectivityinthe
SynthesisofEthanolonRh‐BasedCatalysts.Catal.Commun.2008,9(14),2308–2311.
https://doi.org/10.1016/j.catcom.2008.05.011.
150
(99) Mei,D.;Rousseau,R.;Kathmann,S.M.;Glezakou,V.‐A.;Engelhard,M.H.;Jiang,W.;Wang,C.;
Gerber,M.a.;White,J.F.;Stevens,D.J.EthanolSynthesisfromSyngasoverRh‐Based/SiO2
Catalysts:ACombinedExperimentalandTheoreticalModelingStudy.J.Catal.2010,271(2),325–
342.https://doi.org/10.1016/j.jcat.2010.02.020.
(100) Jiang,D.;Ding,Y.;Pan,Z.;Chen,W.;Luo,H.COHydrogenationtoC2‐OxygenatesoverRh‐Mn‐
Li/SiO2Catalyst:EffectsofSupportPretreatmentwithNC1‐C5Alcohols.Catal.Letters2008,121
(3–4),241–246.https://doi.org/10.1007/s10562‐007‐9322‐3.
(101) Han,L.;Mao,D.;Yu,J.;Guo,Q.;Lu,G.SynthesisofC2‐OxygenatesfromSyngasoverRh‐Based
CatalystSupportedonSiO2,TiO2andSiO2‐TiO2MixedOxide.Catal.Commun.2012,23,20–24.
https://doi.org/10.1016/j.catcom.2012.02.032.
(102) Han,L.;Mao,D.;Yu,J.;Guo,Q.;Lu,G.C2‐OxygenatesSynthesisthroughCOHydrogenationon
SiO2‐ZrO2SupportedRh‐BasedCatalyst:TheEffectofSupport.Appl.Catal.AGen.2013,454,81–
87.https://doi.org/10.1016/j.apcata.2013.01.008.
(103) Chen,G.;Zhang,X.;Guo,C.Y.;Yuan,G.Manganese‐PromotedRhSupportedonaModifiedSBA‐
15MolecularSieveforEthanolSynthesisfromSyngas.EffectofManganeseLoading.Comptes
RendusChim.2010,13(11),1384–1390.https://doi.org/10.1016/j.crci.2010.03.025.
(104) vandenBerg,F.G.A.;Glezer,J.H.E.;Sachtler,W.M.H.TheRoleofPromotersinCO/H2
Reactions:EffectsofMnOandMoO2inSilica‐SupportedRhodiumCatalysts.J.Catal.1985,93(2),
340–352.https://doi.org/10.1016/0021‐9517(85)90181‐2.
(105) Lisitsyn,A.S.;Stevenson,S.A.;Knözinger,H.CarbonMonoxideHydrogenationonSupportedRh‐
MnCatalysts.J.Mol.Catal.1990,63(2),201–211.https://doi.org/10.1016/0304‐5102(90)85144‐
7.
(106) Liu,J.;Guo,Z.;Childers,D.;Schweitzer,N.;Marshall,C.L.;Klie,R.F.;Miller,J.T.;Meyer,R.J.
CorrelatingtheDegreeofMetal‐PromoterInteractiontoEthanolSelectivityoverMnRh/CNTsCO
HydrogenationCatalysts.J.Catal.2014,313,149–158.
https://doi.org/10.1016/j.jcat.2014.03.002.
(107) Pan,X.;Fan,Z.;Chen,W.;Ding,Y.;Luo,H.;Bao,X.EnhancedEthanolProductioninsideCarbon‐
NanotubeReactorsContainingCatalyticParticles.Nat.Mater.2007,6(7),507–511.
https://doi.org/10.1038/nmat1916.
(108) Wang,J.;Zhang,Q.;Wang,Y.Rh‐CatalyzedSyngasConversiontoEthanol:Studiesonthe
PromotingEffectofFeOx.Catal.Today2011,171(1),257–265.
https://doi.org/10.1016/j.cattod.2011.03.023.
(109) Liu,W.;Wang,S.;Sun,T.;Wang,S.ThePromotingEffectofFeDopingonRh/CeO2forthe
EthanolSynthesis.Catal.Letters2015,145(9),1741–1749.https://doi.org/10.1007/s10562‐015‐
1577‐5.
(110) Palomino,R.M.;Magee,J.W.;Llorca,J.;Senanayake,S.D.;White,M.G.TheEffectofFe‐Rh
AlloyingonCOHydrogenationtoC2+Oxygenates.J.Catal.2015,329,87–94.
https://doi.org/10.1016/j.jcat.2015.04.033.
(111) Chen,G.;Guo,C.Y.;Huang,Z.;Yuan,G.SynthesisofEthanolfromSyngasoverIron‐PromotedRh
ImmobilizedonModifiedSBA‐15MolecularSieve:EffectofIronLoading.Chem.Eng.Res.Des.
2011,89(3),249–253.https://doi.org/10.1016/j.cherd.2010.07.014.
151
(112) Han,T.;Zhao,L.;Liu,G.;Ning,H.;Yue,Y.;Liu,Y.Rh‐FeAlloyDerivedfromYRh0.5Fe0.5O3/ZrO2
forHigherAlcoholsSynthesisfromSyngas.Catal.Today2017,298(November2016),69–76.
https://doi.org/10.1016/j.cattod.2017.05.057.
(113) Mo,X.;Gao,J.;Umnaijkaseam,N.;Goodwin,J.G.La,V,andFePromotionofRh/SiO2forCO
Hydrogenation:EffectonAdsorptionandReaction.J.Catal.2009,267,167–176.
https://doi.org/10.1016/j.jcat.2009.09.012.
(114) Mawson,S.;McCutchen,M.S.;Lim,P.K.;Roberts,G.W.ThermodynamicsofHigherAlcohol
Synthesis.EnergyandFuels1993,7(2),257–267.https://doi.org/10.1021/ef00038a016.
(115) Li,F.;Jiang,D.E.;Zeng,X.C.;Chen,Z.MnMonolayerModifiedRhforSyngas‐to‐Ethanol
Conversion:AFirst‐PrinciplesStudy.Nanoscale2012,4(4),1123–1129.
https://doi.org/10.1039/c1nr11121c.
(116) Choi,Y.M.;Liu,P.MechanismofEthanolSynthesisfromSyngasonRh(111).J.Am.Chem.Soc.
2009,131(36),13054–13061.https://doi.org/10.1021/ja903013x.
(117) Filot,I.A.W.;Broos,R.J.P.;VanRijn,J.P.M.;VanHeugten,G.J.H.A.;VanSanten,R.A.;Hensen,
E.J.M.First‐Principles‐BasedMicrokineticsSimulationsofSynthesisGasConversiononaStepped
RhodiumSurface.ACSCatal.2015,5(9),5453–5467.https://doi.org/10.1021/acscatal.5b01391.
(118) Wang,J.;Liu,Z.;Zhang,R.;Wang,B.EthanolSynthesisfromSyngasontheSteppedRh(211)
Surface:EffectofSurfaceStructureandComposition.J.Phys.Chem.C2014,118(39),22691–
22701.https://doi.org/10.1021/jp508885w.
(119) Kapur,N.;Hyun,J.;Shan,B.;Nicholas,J.B.;Cho,K.AbInitioStudyofCOHydrogenationto
OxygenatesonReducedRhTerracesandSteppedSurfaces.J.Phys.Chem.C2010,114(22),
10171–10182.https://doi.org/10.1021/jp911903u.
(120) Zhao,Y.H.;Yang,M.M.;Sun,D.;Su,H.Y.;Sun,K.;Ma,X.;Bao,X.;Li,W.X.Rh‐DecoratedCuAlloy
CatalystforImprovedC2OxygenateFormationfromSyngas.J.Phys.Chem.C2011,115(37),
18247–18256.https://doi.org/10.1021/jp204961g.
(121) Shetty,S.;vanSanten,R.A.;Stevens,P.A.;Raman,S.MolecularStepsfortheSyngasConversion
ontheRh6Cluster.J.Mol.Catal.AChem.2010,330(1–2),73–87.
https://doi.org/10.1016/j.molcata.2010.07.004.
(122) Zhang,R.;Duan,T.;Wang,B.;Ling,L.UnravelingtheRoleofSupportSurfaceHydroxylsandIts
EffectontheSelectivityofC2SpeciesoverRh/γ‐Al2O3CatalystinSyngasConversion:A
TheoreticalStudy.Appl.Surf.Sci.2016,379,384–394.
https://doi.org/10.1016/j.apsusc.2016.04.106.
(123) Gao,J.;Mo,X.;GoodwinJr.,J.G.La,V,andFePromotionofRh/SiO2forCOHydrogenation:
DetailedAnalysisofKineticsandMechanism.J.Catal.2009,268(1),142–149.
https://doi.org/10.1016/j.jcat.2009.09.012.
(124) Schwartz,V.;Campos,A.;Egbebi,A.;Spivey,J.J.;Overbury,S.H.EXAFSandFT‐IRCharacterization
ofMnandLiPromotedTitania‐SupportedRhCatalystsforCOHydrogenation.ACSCatal.2011,1
(10),1298–1306.https://doi.org/10.1021/cs200281g.
(125) Burch,R.;Fetch,M.I.KineticandTransientKineticInvestigationsoftheSynthesisofOxygenates
fromCarbonMonoxide/HydrogenMixturesonSupportedRhodiumCatalysts.Appl.Catal.AGen.
1992,88(1),77–99.https://doi.org/10.1016/0926‐860X(92)80197‐K.
152
(126) Gilhooley,K.;Jackson,S.D.;Rigby,S.Non‐Steady‐StateandTransientEffectsinMedium‐Pressure
HydrogenationofCarbonMonoxideoverRhodiumCatalysts.J.Chem.Soc.FaradayTrans.1Phys.
Chem.Condens.Phases1986,82(2),431.https://doi.org/10.1039/f19868200431.
(127) Burch,R.;Petch,M.I.InvestigationoftheReactionsofAcetaldehydeonPromotedRhodium
Catalysts.Appl.Catal.A,Gen.1992,88(1),61–76.https://doi.org/10.1016/0926‐860X(92)80196‐
J.
(128) Jackson,S.D.;Brandreth,B.J.;Winstanley,D.AMechanisticStudyofCarbonMonoxide
HydrogenationoverRhodiumCatalystsUsingIsotopicTracers.J.Catal.1987,106,464–470.
(129) Orita,H.;Naito,S.;Tamaru,K.MechanismofFormationofC2‐OxygenatedCompoundsfromCO+
H2ReactionoverSiO2‐SupportedRhCatalysts.J.Catal.1984,90(2),183–193.
https://doi.org/10.1016/0021‐9517(84)90247‐1.
(130) Chuang,S.S.C.;Pien,S.I.InfraredStudyoftheCOInsertionReactiononReduced,Oxidized,and
SulfidedRh/SiO2Catalysts.J.Catal.1992,135(2),618–634.https://doi.org/10.1016/0021‐
9517(92)90058‐P.
(131) VanSanten,R.A.;Markvoort,A.J.;Filot,I.A.W.;Ghouri,M.M.;Hensen,E.J.M.Mechanismand
MicrokineticsoftheFischer‐TropschReaction.Phys.Chem.Chem.Phys.2013,15(40),17038–
17063.https://doi.org/10.1039/c3cp52506f.
(132) Ciobica,I.M.;VanSanten,R.A.CarbonMonoxideDissociationonPlanarandSteppedRu(0001)
Surfaces.J.Phys.Chem.B2003,107(16),3808–3812.https://doi.org/10.1021/jp030010x.
(133) Brady,R.C.;Pettit,R.OntheMechanismoftheFischer‐TropschReaction.TheChainPropagation
Step.J.Am.Chem.Soc.1981,103(5),1287–1289.https://doi.org/10.1021/ja00395a081.
(134) Maitlis,P.M.;Long,H.C.;Quyoum,R.;Turner,M.L.;Wang,Z.Q.HeterogeneousCatalysisofC‐C
BondFormation:BlackArtorOrganometallicScience?Chem.Commun.1996,No.1,1–8.
https://doi.org/10.1039/CC9960000001.
(135) Gaube,J.;Klein,H.F.StudiesontheReactionMechanismoftheFischer‐TropschSynthesisonIron
andCobalt.J.Mol.Catal.AChem.2008,283(1–2),60–68.
https://doi.org/10.1016/j.molcata.2007.11.028.
(136) Zhuo,M.;Tan,K.F.;Borgna,A.;Saeys,M.DensityFunctionalTheoryStudyoftheCoInsertion
MechanismforFischer‐TropschSynthesisoverCoCatalysts.J.Phys.Chem.C2009,113(19),
8357–8365.https://doi.org/10.1021/jp900281h.
(137) Davisil636Merck
https://www.sigmaaldrich.com/catalog/product/sial/236802?lang=de®ion=DE(accessedMay
24,2020).
(138) Schindelin,J.;Arganda‐Carreras,I.;Frise,E.;Kaynig,V.;Longair,M.;Pietzsch,T.;Preibisch,S.;
Rueden,C.;Saalfeld,S.;Schmid,B.;Tinevez,J.‐Y.;White,D.J.;Hartenstein,V.;Eliceiri,K.;
Tomancak,P.;Cardona,A.Fiji:AnOpen‐SourcePlatformforBiological‐ImageAnalysis.Nat.
Methods2012,9(7),676–682.https://doi.org/10.1038/nmeth.2019.
(139) O’Connell,K.;Regalbuto,J.R.HighSensitivitySiliconSlitDetectorsfor1NmPowderXRDSize
DetectionLimit.Catal.Letters2015,145(3),777–783.https://doi.org/10.1007/s10562‐015‐1479‐
6.
153
(140) Moulijn,J.A.;vanDiepen,A.E.;Kapteijn,F.ChemInformAbstract:CatalystDeactivation:IsIt
Predictable?WhattoDo?ChemInform2010,32(27),no‐no.
https://doi.org/10.1002/chin.200127265.
(141) Claeys,M.;Dry,M.E.;VanSteen,E.;VanBerge,P.J.;Booyens,S.;Crous,R.;VanHelden,P.;
Labuschagne,J.;Moodley,D.J.;Saib,A.M.ImpactofProcessConditionsontheSintering
BehaviorofanAlumina‐SupportedCobaltFischer‐TropschCatalystStudiedwithaninSitu
Magnetometer.ACSCatal.2015,5(2),841–852.https://doi.org/10.1021/cs501810y.
(142) Huang,X.;Teschner,D.;Dimitrakopoulou,M.;Fedorov,A.;Frank,B.;Kraehnert,R.;Rosowski,F.;
Kaiser,H.;Schunk,S.;Kuretschka,C.;Schlögl,R.;Willinger,M.G.;Trunschke,A.Atomic‐Scale
ObservationoftheMetal–PromoterInteractioninRh‐BasedSyngas‐UpgradingCatalysts.Angew.
Chemie‐Int.Ed.2019,58(26),8709–8713.https://doi.org/10.1002/anie.201902750.
(143) Hettler,S.;Dries,M.;Hermann,P.;Obermair,M.;Gerthsen,D.;Malac,M.CarbonContamination
inScanningTransmissionElectronMicroscopyandItsImpactonPhase‐PlateApplications.Micron
2017,96,38–47.https://doi.org/10.1016/j.micron.2017.02.002.
(144) Wilson,T.P.;Kasai,P.H.;Ellgen,P.C.TheStateofManganesePromoterinRhodium‐SilicaGel
Catalysts.J.Catal.1981,69(1),193–201.https://doi.org/10.1016/0021‐9517(81)90141‐X.
(145) Preikschas,P.;Bauer,J.;Huang,X.;Yao,S.;Naumannd’Alnoncourt,R.;Kraehnert,R.;Trunschke,
A.;Rosowski,F.;Driess,M.FromaMolecularSingle‐SourcePrecursortoaSelectiveHigh‐
PerformanceRhMnOxCatalystfortheConversionofSyngastoEthanol.ChemCatChem2019,11
(2),885–892.https://doi.org/10.1002/cctc.201801978.
(146) Jahangiri,H.;Bennett,J.;Mahjoubi,P.;Wilson,K.;Gu,S.AReviewofAdvancedCatalyst
DevelopmentforFischer‐TropschSynthesisofHydrocarbonsfromBiomassDerivedSyn‐Gas.
Catal.Sci.Technol.2014,4(8),2210–2229.https://doi.org/10.1039/c4cy00327f.
(147) Chuang,S.C.;Goodwin,J.G.;Wender,I.TheEffectofAlkaliPromotiononCOHydrogenationover
RhTiO2.J.Catal.1985,95(2),435–446.https://doi.org/10.1016/0021‐9517(85)90121‐6.
(148) Escobar,M.;Gracia,F.;Karelovic,A.;Jiménez,R.KineticandinSituFTIRStudyofCOMethanation
onaRh/Al2O3Catalyst.Catal.Sci.Technol.2015,5(9),4532–4541.
https://doi.org/10.1039/c5cy00676g.
(149) Fisher,I.A.;Bell,A.T.AComparativeStudyofCOandCO2HydrogenationoverRh/SiO2.J.Catal.
1996,162(1),54–65.https://doi.org/10.1006/jcat.1996.0259.
(150) Basu,P.;Panayotov,D.;Yates,J.T.Rhodium‐CarbonMonoxideSurfaceChemistry:The
InvolvementofSurfaceHydroxylGroupsonA12O3andSiO2Supports.J.Am.Chem.Soc.1988,
110(7),2074–2081.https://doi.org/10.1021/ja00215a010.
(151) Lenarda,M.;Storaro,L.;Ganzerla,R.HydroformylationofSimpleOlefinsCatalyzedbyMetalsand
ClustersSupportedonUnfunctionalizedInorganicCarriers.J.Mol.Catal.AChem.1996,111(3),
203–237.https://doi.org/10.1016/1381‐1169(96)00211‐7.
(152) Iglesia,E.;Reyes,S.C.;Madon,R.J.Transport‐EnhancedAlpha‐OlefinReadsorptionPathwaysin
Ru‐CatalyzedHydrocarbonSynthesis.J.Catal.1991,129(1),238–256.
https://doi.org/10.1016/0021‐9517(91)90027‐2.
154
(153) Iglesia,E.;Reyes,S.C.;Madon,R.J.;Soled,S.L.SelectivityControlandCatalystDesigninthe
Fischer‐TropschSynthesis:Sites,Pellets,andReactors.Adv.Catal.1993,39(C),221–302.
https://doi.org/10.1016/S0360‐0564(08)60579‐9.
(154) Hanaoka,T.;Arakawa,H.;Matsuzaki,T.;Sugi,Y.;Kanno,K.;Abe,Y.EthyleneHydroformylation
andCarbonMonoxideHydrogenationoverModifiedandUnmodifiedSilicaSupportedRhodium
Catalysts.Catal.Today2000,58(4),271–280.https://doi.org/10.1016/S0920‐5861(00)00261‐3.
(155) Huang,L.;Xu,Y.;Guo,W.;Liu,A.;Li,D.;Guo,X.StudyonCatalysisbyCarbonylCluster‐Derived
SiO2‐SupportedRhodiumforEthyleneHydroformylation.Catal.Letters1995,32(1–2),61–81.
https://doi.org/10.1007/BF00806102.
(156) Kim,T.;Celik,F.E.;Hanna,D.G.;Shylesh,S.;Werner,S.;Bell,A.T.Gas‐PhaseHydroformylationof
PropeneoverSilica‐SupportedPPh3‐ModifiedRhodiumCatalysts.Top.Catal.2011,54(5–7),
299–307.https://doi.org/10.1007/s11244‐011‐9664‐3.
(157) Li,X.;Ding,Y.;Jiao,G.;Li,J.;Lin,R.;Gong,L.;Yan,L.;Zhu,H.ANewConceptofTetheredLigand‐
ModifiedRh/SiO2CatalystforHydroformylationwithHighStability.Appl.Catal.AGen.2009,353
(2),266–270.https://doi.org/10.1016/j.apcata.2008.10.052.
(158) AlvaradoRupflin,L.;Mormul,J.;Lejkowski,M.;Titlbach,S.;Papp,R.;Gläser,R.;Dimitrakopoulou,
M.;Huang,X.;Trunschke,A.;Willinger,M.G.;Schlögl,R.;Rosowski,F.;Schunk,S.A.Platinum
GroupMetalPhosphidesasHeterogeneousCatalystsfortheGas‐PhaseHydroformylationof
SmallOlefins.ACSCatal.2017,7(5),3584–3590.https://doi.org/10.1021/acscatal.7b00499.
(159) Schweicher,J.;Bundhoo,A.;Kruse,N.HydrocarbonChainLengtheninginCatalyticCO
Hydrogenation:EvidenceforaCO‐InsertionMechanism.J.Am.Chem.Soc.2012,134(39),
16135–16138.https://doi.org/10.1021/ja3068484.
(160) Basu,P.;Panayotov,D.;Yates,J.T.SpectroscopicEvidencefortheInvolvementofHydroxyl
GroupsintheFormationofDicarbonylrhodium(I)onMetalOxideSupports.J.Phys.Chem.1987,
91(12),3133–3136.https://doi.org/10.1021/j100296a007.
(161) Shetty,S.;Strych,S.;Jansen,A.P.J.;Santen,R.A.V.TheoreticalInvestigationofCOAdsorptionon
Rhn(n=3‐13)Clusters1.Can.J.Chem.2009,87(7),824–831.https://doi.org/10.1139/V09‐015.
(162) Takeuchi,A.;Katzer,J.R.MechanismofMethanolFormation.J.Phys.Chem.1981,85(8),937–
939.https://doi.org/10.1021/j150608a002.
(163) Xiaoding,X.;Doesburg,E.B.M.;Scholten,J.J.F.SynthesisofHigherAlcoholsfromSyngas‐
RecentlyPatentedCatalystsandTentativeIdeasontheMechanism.Catal.Today1987,2,125–
170.
(164) Gao,F.;Goodman,D.W.ModelCatalysts:SimulatingtheComplexitiesofHeterogeneous
Catalysts.Annu.Rev.Phys.Chem.2012,63(1),265–286.https://doi.org/10.1146/annurev‐
physchem‐032511‐143722.
(165) Yu,J.;Mao,D.;Ding,D.;Guo,X.;Lu,G.NewInsightsintotheEffectsofMnandLionthe
MechanisticPathwayforCOHydrogenationonRh‐Mn‐Li/SiO2Catalysts.J.Mol.Catal.AChem.
2016,423,151–159.https://doi.org/10.1016/j.molcata.2016.06.018.
(166) Magee,J.W.;Palomino,R.M.;White,M.G.InfraredSpectroscopyInvestigationofFe‐Promoted
RhCatalystsSupportedonTitaniaandCeriaforCOHydrogenation.Catal.Letters2016,146(9),
1771–1779.https://doi.org/10.1007/s10562‐016‐1801‐y.
155
(167) Chen,W.;Ding,Y.;Song,X.;Wang,T.;Luo,H.PromotionEffectofSupportCalcinationonEthanol
ProductionfromCOHydrogenationoverRh/Fe/Al2O3Catalysts.Appl.Catal.AGen.2011,407(1–
2),231–237.https://doi.org/10.1016/j.apcata.2011.08.044.
(168) Elschenbroich,C.Organometallchemie,6thed.;Vieweg+Teubner:Wiesbaden,2008.
https://doi.org/10.1007/978‐3‐8351‐9223‐2.
(169) Yang,N.;Medford,A.J.;Liu,X.;Studt,F.;Bligaard,T.;Bent,S.F.;Nørskov,J.K.IntrinsicSelectivity
andStructureSensitivityofRhodiumCatalystsforC2+OxygenateProduction.J.Am.Chem.Soc.
2016,138(11),3705–3714.https://doi.org/10.1021/jacs.5b12087.
(170) Goldsmith,B.R.;Sanderson,E.D.;Ouyang,R.;Li,W.X.CO‐andNO‐InducedDisintegrationand
RedispersionofThree‐WayCatalystsRhodium,Palladium,andPlatinum:AnAbInitio
ThermodynamicsStudy.J.Phys.Chem.C2014,118(18),9588–9597.
https://doi.org/10.1021/jp502201f.
(171) Frøseth,V.;Storsæter,S.;Borg,Ø.;Blekkan,E.A.;Rønning,M.;Holmen,A.SteadyStateIsotopic
TransientKineticAnalysis(SSITKA)ofCOHydrogenationonDifferentCoCatalysts.Appl.Catal.A
Gen.2005,289(1),10–15.https://doi.org/10.1016/j.apcata.2005.04.009.
(172) Lohitharn,N.;Goodwin,J.G.EffectofKPromotionofFeandFeMnFischer‐TropschSynthesis
Catalysts:AnalysisattheSiteLevelUsingSSITKA.J.Catal.2008,260(1),7–16.
https://doi.org/10.1016/j.jcat.2008.08.011.
(173) Schweicher,J.;Bundhoo,A.;Frennet,A.;Kruse,N.;Daly,H.;Meunier,F.C.DRIFTS/MSStudies
duringChemicalTransientsandSSITKAoftheCO/H2ReactionoverCo‐MgOCatalysts.J.Phys.
Chem.C2010,114(5),2248–2255.https://doi.org/10.1021/jp909754w.
(174) Nefedov,V.I.;Firsov,M.N.;Shaplygin,I.S.ElectronicStructuresofMRhO2,MRh2O4,RhMO4
andRh2MO6ontheBasisofX‐RaySpectroscopyandESCAData.J.ElectronSpectros.Relat.
Phenomena1982,26(1),65–78.https://doi.org/10.1016/0368‐2048(82)87006‐0.
(175) Fierro,J.L.G.;Palacios,J.M.;Tomas,F.AnAnalyticalSEMandXPSStudyofPlatinum‐Rhodium
GauzesUsedinHighPressureAmmoniaBurners.Surf.InterfaceAnal.1988,13(1),25–32.
https://doi.org/10.1002/sia.740130107.
(176) Dubecký,F.;Kindl,D.;Hubík,P.;Mičušík,M.;Dubecký,M.;Boháček,P.;Vanko,G.;Gombia,E.;
Nečas,V.;Mudroň,J.AComparativeStudyofMgandPtContactsonSemi‐InsulatingGaAs:
ElectricalandXPSCharacterization.Appl.Surf.Sci.2017,395,131–135.
https://doi.org/10.1016/j.apsusc.2016.04.176.
(177) Zhou,Y.;Peng,J.;Deng,C.;Zhu,M.TribochemicalBehaviorofPureMagnesiumduringSliding
Friction.Metals(Basel).2019,9(3).https://doi.org/10.3390/met9030311.
(178) XPSSimplifiedCarbonhttps://xpssimplified.com/elements/carbon.php(accessedApr8,2020).
(179) Powell,C.J.RecommendedAugerParametersfor42ElementalSolids.J.ElectronSpectros.Relat.
Phenomena2012,185(1–2),1–3.https://doi.org/10.1016/j.elspec.2011.12.001.
(180) Contour,J.P.;Mouvier,G.;Hoogewys,M.;Leclere,C.X‐RayPhotoelectronSpectroscopyand
ElectronMicroscopyofPt‐RhGauzesUsedforCatalyticOxidationofAmmonia.J.Catal.1977,48
(1–3),217–228.https://doi.org/10.1016/0021‐9517(77)90093‐8.
(181) NISTXPSDatabasehttps://srdata.nist.gov/xps/(accessedApr8,2020).
https://doi.org/10.18434/T4T88K.
156
(182) Tan,B.J.;Klabunde,K.J.;Sherwood,P.M.A.XPSStudiesofSolvatedMetalAtomDispersed
(SMAD)Catalysts.EvidenceforLayeredCobalt‐ManganeseParticlesonAluminaandSilica.J.Am.
Chem.Soc.1991,113(3),855–861.https://doi.org/10.1021/ja00003a019.
(183) XPSSimplifiedManganesehttps://xpssimplified.com/elements/manganese.php(accessedApr8,
2020).
(184) Schunemann,V.;Trevino,H.;Lei,G.D.;Tomczak,D.C.;Sachtler,W.M.H.;Fogash,K.;Dumesic,J.
A.FePromotedRh‐ClustersinZeoliteNaY:CharacterizationandCatalyticPerformanceinCO
Hydrogenation.J.Catal.1995,153(1),144–157.https://doi.org/10.1006/jcat.1995.1116.
(185) Nefedov,V.I.;Salyn,Y.V.;Leonhardt,G.;Scheibe,R.AComparisonofDifferentSpectrometers
andChargeCorrectionsUsedinX‐RayPhotoelectronSpectroscopy.J.ElectronSpectros.Relat.
Phenomena1977,10(2),121–124.https://doi.org/10.1016/0368‐2048(77)85010‐X.
157
Appendix
A Experimental details
OverviewBasCatsamplenumbers
TableA1
BasCatsamplenumbers(BCSN)forallinvestigatedcatalystsamplesafterdifferenttreatments.
BasCatsamplenumber
TreatmentCalcinedH2treatedKineticstudy
(DinoRun26)
Hightemp.study
(DinoRun27)
COdropout
(DinoRun31)
Rh/SiO224012447240824512614
RhMn/SiO224022448240924522615
RhFe/SiO224052449241024532616
RhMnFe/SiO224042450241124542617
Typicalchromatogram
FigureA1
TypicalchromatogramforRh‐basedcatalystsshowingCO,CO
2
,hydrocarbonsandoxygenatesdetectedin
theflameionizationdetector(FID1A).
158
Schemeofusedgaschromatograph
FigureA2
Setupoftheusedgaschromatographintheconfiguration“CocoChanel4.0”.
159
B Morphology and surface properties of Rh/Mn/Fe/SiO2
catalysts
Specificsurfaceareaofallcharacterizedsamples
TableA2SpecificsurfaceareaobtainedfromN2adsorptionaccordingtotheBrunauer‐Emmett‐Tellermethod(BET).
Surfacearea/m2g‐1
Rh/SiO2RhMn/SiO2RhFe/SiO2RhMnFe/SiO2
Calcined461466467462
H2treatment438433429426
Kineticstudya444380435423
COdropoutb456417442438
aSpentsampleofDinoRun26(260°Cmaximumreactiontemperature,530htimeonstream,cooldowninCO/N2)
bSpentsampleofDinoRun31(260°Cmaximumreactiontemperature,140htimeonstream,cooldowninH2/N2)
B.1 Analysis of surface properties and oxidation states via X-ray photoelectron
spectroscopy (XPS)
As the catalyst surface represents the interface where reaction with the gas phase takes
place, its structural properties are of great interest for heterogeneous catalysis. X-ray
photoelectron spectroscopy is a surface-sensitive characterization method primarily
providing information about oxidation states, but also about surface composition.
After calcination of the Rh/SiO2 catalyst, Rh is present as Rh2O3 with an oxidation state of +3
(FigureA3 A). The performed H2 treatment procedure at the beginning of each run leads to
full reduction of Rh2O3 to Rh0 on the surface.
160
FigureA3NormalizedXPSspectraintheRh3dregionofRh/SiO2samplesafterdifferenttreatments.Linesrepresent
thetypicallocationforsignalsofRh2O3174(green),metallicRh175(blue)andaMgKLL’augersignal176,177(orange)
overlappingwiththeRh3d5/2signalsandseparatedviadeconvolutioninAanddifferentchemicalstatesfor
carboninB.178
The metallic state of Rh is retained although the sample was handled in air after catalytic
testing. The samples were in contact with steatite in the reactor. Small Mg containing
steatite contamination of the sieved samples lead to a Mg KLL’ auger signal overlapping with
the Rh3d 5/2 peak. The area of the Mg KLL’ auger signal can thereby be related to the Mg1s
signal (FigureA4). Fitting of the Rh doublet and evaluation of the area was therefore based
on the Rh3d 3/2 signal. After reaction, Rh is still present in its metallic state. Different
reaction conditions do not appear to have different impact on the oxidation state of Rh. The
presence of metallic Rh on the surface is consistent with XRD reflections suggesting metallic
Rh in the nanoparticles. However, after the kinetic study the obtained binding energy is
slightly shifted to a lower value. This shift does not indicate a change in oxidation state since
both binding energies (307.4 eV and 307.1 eV, TableA3) are typical values found for
metallic Rh. Hence, the shift might be rather related to structural adjustments during the
kinetic run. This result supports the vital influence of long time on stream and application
of different reaction atmospheres on the structural properties of the catalyst. The observed
carbon signal is common for samples that have been in contact to air. Carbon in different
chemical states contribute to this adventitious carbon signal.
296300304308312316320324328 276280284288292296300
Normalized intensity / a.u.
Binding energy / eV
Mg KLL' auger
Rh
2
O
3
Rh
metal
CO dropout
Run31
H
2
treatment
Kinetic study
Run26
High temperatures
Run27
calcined
Normalized intensity / a.u.
Binding energy / eV
CO dropout
Run31
H
2
treatment
Kinetic study
Run26
calcined
C-O C-H
C=OO-C=O
High temperatures
Run27
Rh/SiO
2
–Rh3dscan Rh/SiO
2
–C1sscan
AB
161
FigureA4NormalizedXPSspectraintheRh3dregionofthespentRh/SiO2sampleafterkineticstudyandamixture
ofthesupportmaterialSiO2andtheMgcontainingreactorfillingmaterialsteatite.ComparisonshowstheMgKLL‘
augersignalcomingfromsteatiteimpuritiesandoverlappingwiththeRh3d5/2signal.
The areas of the respective peaks contain information about the surface composition. The
Si:O ratio is ~2:1 for all samples as it was to be expected. Si and O together make up
97 atom-% of the surface. For the calcined sample, a Rh content of 0.19 atom-% was
calculated (TableA3). In order to compare the surface composition of two samples, the Si
content can be regarded as constant and serve as a reference. The calcined sample has a
Si:Rh surface ratio of 1000:6. The carbon content is typical for samples which have been
prepared in contact with air.
TableA3SurfacecompositionandbindingenergiesofRh/SiO2afterdifferenttreatmentsfromXPS.
TreatmentSurfacecomposition/atom‐% Bindingenergya/eV
Si:Rhratio
SiORhC Rh3d5/2
Calcined31660.192.8 308.6(Rh2O3)1000:6
H2treatment31660.162.1 307.4(Rh0)1000:5
COdropoutb31660.162.8 307.4(Rh0)1000:5
Hightemperaturesc31650.163.7 307.4(Rh0)1000:5
Kineticstudyd29610.479.3 307.1(Rh0)179,1801000:16
aReferencesforbindingenergiesweretakenfromNISTX‐rayPhotoelectronSpectroscopyDatabase181
bSpentsampleofDinoRun31(260°Cmaximumreactiontemperature,140htimeonstream,cooldowninH2/N2)
cSpentsampleofDinoRun27(320°Cmaximumreactiontemperature,380htimeonstream,cooldowninCO/N2)
dSpentsampleofDinoRun26(260°Cmaximumreactiontemperature,530htimeonstream,cooldowninCO/N2)
After H2 treatment the Si:Rh ratio increases marginally which is in accordance with the
slightly increased mean particle size observed in XRD and STEM. The spent samples from
the CO dropout experiment and high temperature study give similar results as the H2
treated sample. However, the sample obtained after the kinetic study again differs clearly
296300304308312316320324328
Normalized intensity / a.u.
Binding energy / eV
Mg KLL' augerRh
metal
SiO
2
/ steatite
mixture
Kinetic study
Run26
162
from the others. The calculated Si:Rh ratio changes to 1000:16 which suggests a significantly
higher dispersion of Rh on the silica surface which might also account for the shift in binding
energy. The mean particle sizes obtained from XRD (5.7 nm) and STEM (4.4 nm) analysis do
not explain this observation. However, very small Rh clusters or single atoms might have
formed during the long-term reaction under various reaction conditions. These structures
would be invisible for STEM and XRD. The formation of Rh clusters or single atoms might
be crucial for the high mobility of Rh which leads to the formation of large agglomerates as
observed from the STEM studies. The Rh3d 5/2 binding energy is slightly shifted to a
smaller value. However, this shift is in the rage of the expected error of ±0.2 eV and might
be related to the much more intense signal. In addition, the carbon content is significantly
increased on the sample after kinetic study. This might be another indicator for a high metal
surface area adsorbing adventitious carbon from air.
Mn addition does not have an impact on Rh oxidation state for the different samples. Rh
occurs as Rh2O3 on the calcined sample and as metallic Rh after H2 treatment and reaction
(FigureA5 A). The Mn2p doublet seems to shift to higher binding energies upon H2
treatment and even more pronounced for the samples after STE reaction (FigureA5 B,
TableA4). Clearly, in none of the samples metallic Mn is present. However, for Mn oxides
the shifts of the signals corresponding to the different oxidation states of Mn are very small.
The satellite signal at 646 eV is typical for MnO suboxide. However, due to low Mn loading
and very small shifts, no reliable conclusion of the formal oxidation state was obtained. The
shift towards higher oxidation states after reaction is counterintuitive at first sight as
usually higher binding energies might be related to higher oxidation states which are not
expected to be formed in the reducing atmosphere of the STE reaction. However, other
structural transformations such as the formation of Mn acetates or carbonates during
reaction are thinkable and could have impact on the location of XPS signals. Respective
references could not be found in literature and therefore this shift should be investigated in
future research.
Evaluation of the surface composition on the calcined RhMn/SiO2 sample revealed a molar
Rh:Mn ratio of 1:1 corresponding to the ratio used for synthesis (TableA4). After H2
treatment the Si:Rh ratio is slightly increased like over Rh/SiO2. The Si:Mn ratio stays
constant. The decrease of Rh dispersion suggests sintering or a partial coverage of Rh
particles by Mn oxide. During reaction, the Rh and Mn content increases with time on stream
which might indicate an enhanced dispersion induced by reaction conditions. The
enrichment of Mn on the surface is particularly pronounced leading to a molar Rh:Mn ratio
of 1:2 for all spent catalysts further promoting the idea of strong metal support interaction
(SMSI). As for Rh/SiO2, the Si:Rh ratio is the significantly different for the sample after
kinetic study supporting the important impact of long time on stream and reaction
atmosphere.
163
FigureA5NormalizedXPSspectraintheRh3dregion(A),Mn2pregion(B),andC1sregion(C)ofRhMn/SiO2in
differentstates.LinesrepresentthetypicallocationforsignalsofRh2O3174(green),metallicRh175(blue)andaMg
KLL’augerlineoverlappingwiththeRhsignals176,177(orange)inA,MnO182(blue)andatypicalMnOsatellitepeak183
(green)inB,anddifferentchemicalstatesforcarboninC.178
630640650660670680
CO dropout
Run31
Kinetic study
Run26
Normalized intensity / a.u.
Binding energy / eV
MnOx
H2 treatment
High temperatures
Run27
calcined
MnOsat
296300304308312316320324328
Normalized intensity / a.u.
Binding energy / eV
Mg KLL' auger
Rh
2
O
3
Rh
metal
CO dropout
Run31
H
2
treatment
Kinetic study
Run26
High temperatures
Run27
calcined
6
276280284288292296300
Normalized intensity / a.u.
Binding energy / eV
CO dropout
Run31
H
2 treatment
Kinetic study
Run26
calcined
C-O C-H
C=OO-C=O
High temperatures
Run27
AB
RhMn/SiO
2
–Rh3dscan RhMn/SiO
2
–Mn2pscan
C
RhMn/SiO
2
–C1sscan
164
TableA4SurfacecompositionandbindingenergiesonRhMn/SiO2samplesafterdifferenttreatmentsfromXPS.
Treatment
Surfacecomposition/
atom‐%
Bindingenergya/
eVSi:Rh:Mn
ratio
SiORhMnC Rh3d5/2Mn2p3/2
Calcined30650.220.214.7 308.8(Rh2O3)641.6(MnOx)1000:7:7
H2treatment30630.150.217.5 306.9(Rh0)641.9(MnOx)1000:5:7
COdropoutb30650.150.314.0 307.4(Rh0)642.2(MnOxe)1000:5:10
Hightemperaturesc29630.170.367.3 307.3(Rh0)642.3(MnOxe)1000:6:12
Kineticstudyd27600.290.5812.7 307.3(Rh0)642.3(MnOxe)1000:11:22
aReferencesforbindingenergiesweretakenfromNISTX‐rayPhotoelectronSpectroscopyDatabase181
bSpentsampleofDinoRun31(260°Cmaximumreactiontemperature,140htimeonstream,cooldowninH2/N2)
cSpentsampleofDinoRun27(320°Cmaximumreactiontemperature,380htimeonstream,cooldowninCO/N2)
dSpentsampleofDinoRun26(260°Cmaximumreactiontemperature,530htimeonstream,cooldowninCO/N2)
eFormationofacetatesand/orcarbonatesshouldbeconsidered
The XPS results for RhFe/SiO2 provide comparable information as for RhMn/SiO2. The
reduction of Rh2O3 to metallic Rh after H2 treatment and the stability of metallic Rh during
reaction and handling in air was also observed (FigureA6 A). Assuming RhFe alloy
formation, Fe in its metallic state would be expected.89,99,110,166,184 The binding energies in
the Fe2p region rather show Fe oxide fitting to references for Fe2O3. (TableA5) However,
the evaluation of the Fe oxidation state is subject to the same limitations as for Mn. The
highly oxophilic character of Fe and ability to instantly oxidize in air might cause misleading
results (FigureA6 B). In contrast to the results for Mn oxide, no clear trend upon sample
treatment was observed.
165
FigureA6NormalizedXPSspectraintheRh3dregion(A),Fe2pregion(B),andC1sregion(C)ofRhFe/SiO2in
differentstates.LinesrepresentthetypicallocationforsignalsofRh2O3174(green),metallicRh175(blue)andaMg
KLL’augerlineoverlappingwiththeRhsignals176,177(orange)inA,Fe2O3oxide185(blue)inB,anddifferentchemical
statesforcarboninC.178
276280284288292296300
Normalized intensity / a.u.
Binding energy / eV
CO dropout
Run31
H
2
treatment
Kinetic study
Run26
calcined
C-O C-H
C=OO-C=O
High temperatures
Run27
690700710720730740750760
Kinetic study
Run26
CO dropout
Run31
Normalized intensity / a.u.
Binding energy / eV
Fe
2
O
3
H
2
treatment
High temperatures
Run27
calcined
296300304308312316320324328
Normalized intensity / a.u.
Binding energy / eV
Mg KLL' auger
Rh
2
O
3
Rh
metal
CO dropout
Run31
H
2
treatment
Kinetic study
Run26
High temperatures
Run27
calcined
AB
RhFe/SiO
2
–Rh3dscan RhFe/SiO
2
–Fe2pscan
C
RhFe/SiO
2
–C1sscan
166
After calcination, the Rh:Fe ratio on the surface is 7:8 (TableA5). Less Fe would be expected
from synthesis using a Rh:Fe ratio for impregnation of 3:1. The presence of additional Fe,
e.g. from the support, could be excluded. Catalysts without Fe addition and the support itself
do not show signals in the Fe2p region. After H2 treatment and catalytic reaction, the Fe
content on the surface was even further increased. That might suggest either the presence
of highly dispersed Fe on the silica surface not taking part in alloy formation or the
enrichment of Fe content on the RhFe particle surface. Finely distributed Fe might be
invisible in STEM-EDX evaluation and easily overlooked around the high Fe content in RhFe
particles. Line scans rather suggested uniform RhFe alloys than core-shell structures.
However, Fe oxidation might lead to dealloying of RhFe particles and transport of Fe oxide
on the particle surface. Moreover, XRD and STEM-EDX provided evidence for the presence
of larger Rh particles that might provide a comparably small Rh surface area. The presence
of pure Rh particles along with RhFe particles in combination with highly dispersed Fe or
Fe oxide on the silica surface therefore appears most plausible. Among the RhFe/SiO2
samples after reaction, the one from the kinetic study again shows highest dispersion.
However, in this case the H2 treated sample shows comparable Rh and Fe contents on the
surface suggesting that the redispersion effect observed over Rh/SiO2 is less pronounced
over RhFe/SiO2.
TableA5SurfacecompositionandbindingenergiesonRhFe/SiO2samplesafterdifferenttreatmentsfromXPS.
Treatment
Surfacecomposition/
atom‐%
Bindingenergya/
eVSi:Rh:Fe
ratio
SiORhFeC Rh3d5/2Fe2p3/2
Calcined31660.230.253.0 308.9(Rh2O3)710.7(FeOx)1000:7:8
H2treatment18450.180.2636 307.4(Rh0)710.3(FeOx)1000:9:14
COdropoutb31650.140.194.0 307.4(Rh0)710.2(FeOx)1000:4:6
Hightemperaturesc30640.130.275.4 307.4(Rh0)710.5(FeOx)1000:4:9
Kineticstudyd29620.240.388.5 307.3(Rh0)710.3(FeOx)1000:8:13
aReferencesforbindingenergiesweretakenfromNISTX‐rayPhotoelectronSpectroscopyDatabase181
bSpentsampleofDinoRun31(260°Cmaximumreactiontemperature,140htimeonstream,cooldowninH2/N2)
cSpentsampleofDinoRun27(320°Cmaximumreactiontemperature,380htimeonstream,cooldowninCO/N2)
dSpentsampleofDinoRun26(260°Cmaximumreactiontemperature,530htimeonstream,cooldowninCO/N2)
The evaluation of the trimetallic catalyst RhMnFe/SiO2 confirms the effects discussed for
the bimetallic catalysts above regarding the oxidation states for Rh, Mn and Fe (FigureA7).
The shift of Mn2p 3/2 binding energies towards higher values after STE reaction was
observed over both Mn-containing catalysts suggesting a systematic relevance rather than
individual event for a single sample. Here, the effect for the sample taken after CO dropout
167
and therefore additional H2 treatment the effect is less pronounced. Previous studies of the
used catalysts suggested mixed Mn-Fe oxides to be formed in the trimetallic catalyst.82,142
Changes for the Mn2p and Fe2p signals in comparison to their bimetallic counterparts were
not observed in this study.
Regarding the surface composition, similar results as for the bimetallic catalysts were found
as well. In the calcined state, the elemental composition Rh:Mn:Fe on the surface is 1:1:1.
This fits well for Rh and Mn which were used stoichiometrically for synthesis but the value
for Fe is tripled compared to what would be expected (TableA6).
168
FigureA7NormalizedXPSspectraintheRh3dregion(A),Mn2pregion(B),Fe2pregion(C),andC1sregion(C)of
RhMnFe/SiO2indifferentstates.LinesrepresentthetypicallocationforsignalsofRh2O3174(green),metallicRh175
(blue)andaMgKLL’augerlineoverlappingwiththeRhsignals176,177(orange)inA,MnO182(blue)andatypicalMnO
satellitepeak183(green)inB,Fe2O3oxide185(blue)inC,anddifferentchemicalstatesforcarboninC.178
6
276280284288292296300
Normalized intensity / a.u.
Binding energy / eV
CO dropout
Run31
H
2
treatment
Kinetic study
Run26
calcined
C-O C-H
C=OO-C=O
High temperatures
Run27
690700710720730740750760
CO dropout
Run31
Kinetic study
Run26
Normalized intensity / a.u.
Binding energy / eV
Fe
2
O
3
H
2
treatment
High temperatures
Run27
calcined
630640650660670680
Kinetic study
Run26
CO dropout
Run31
Normalized intensity / a.u.
Binding energy / eV
MnO
H
2
treatment
High temperatures
Run27
calcined
MnO
sat
296300304308312316320324328
Normalized intensity / a.u.
Binding energy / eV
Mg KLL' auger
Rh
2
O
3
Rh
metal
CO dropout
Run31
H
2
treatment
Kinetic study
Run26
High temperatures
Run27
calcined
A
RhMnFe/SiO
2
–Rh3dscan
C
RhMnFe/SiO
2
–Fe2pscan
D
RhMnFe/SiO
2
–C1sscan
B
RhMnFe/SiO
2
–Mn2pscan
169
After H2 treatment and reaction, Mn and Fe surface contents increase. Mn content is stable
at ~1.7 times the Rh content. This value is slightly higher compared to the result for the H2
treated sample of RhMn/SiO2 and slightly lower than the results for the samples of
RhMn/SiO2 after reaction. Still, the high Mn surface content implies the partial coverage of
Rh containing particles with Mn oxides. Rh:Fe ratios on RhMnFe/SiO2 are also comparable
to RhFe/SiO2 but show a wider range of 1.3–2.1. The trends regarding the reaction
conditions are thereby different. A low signal-to-noise ratio might be the reason for less
reproducible results. As for all other catalysts, the highest Rh dispersion is observed after
the kinetic study. The total value is comparable to the bimetallic catalysts and much lower
than for pure Rh/SiO2. The described findings are in good agreement with STEM-EDX
mappings implying RhFe particles embedded in a Mn oxide matrix. Due to the low Fe
content, a potential incorporation of some Fe oxide into the Mn oxide matrix might not be
visible. As previously discussed for RhFe/SiO2, the high surface content of Fe might either
imply the existence of larger Rh particles without Fe or the enrichment of Fe on the RhFe
particle surface.
TableA6SurfacecompositionandbindingenergiesonRhMnFe/SiO2samplesafterdifferenttreatmentsfromXPS.
Treatment
Surfacecomposition/atom‐%Bindingenergya/eV
Si:Rh:Mn:Feratio
SiORhMnFeCRh3d
5/2
Mn2p
3/2
Fe2p
3/2
Calcined29640.190.190.195.7308.9
(Rh2O3)
641.5
(MnOx)
710.6
(FeOx)
1000:6:6:6
H2treatment27590.150.250.2012.6307.4
(Rh0)
641.9
(MnOx)
710.4
(FeOx)
1000:5:9:7
COdropoutb30630.120.200.256.1307.3
(Rh0)
642.0
(MnOx)
710.7
(FeOx)
1000:4:7:8
High
temperaturesc30640.140.240.224.9307.0
(Rh0)
642.2
(MnOx)
710.5
(FeOx)
1000:5:8:7
Kineticstudyd28620.270.430.369.1307.4
(Rh0)
642.4
(MnOx)
710.4
(FeOx)
1000:9:16:13
aReferencesforbindingenergiesweretakenfromNISTX‐rayPhotoelectronSpectroscopyDatabase181
bSpentsampleofDinoRun31(260°Cmaximumreactiontemperature,140htimeonstream,cooldowninH2/N2)
cSpentsampleofDinoRun27(320°Cmaximumreactiontemperature,380htimeonstream,cooldowninCO/N2)
dSpentsampleofDinoRun26(260°Cmaximumreactiontemperature,530htimeonstream,cooldowninCO/N2)
170
B.2 STEM morphology analyses
General structural features were evaluated from images taken with the dark-field, bright-
field and high-angle annular dark-field (HAADF) detector. Particle size distributions were
obtained by systematic measuring and counting more than 200 particles per sample in
different areas from the bright-field images. The HAADF-STEM image of the H2 treated
Rh/SiO2 sample shows nanoparticles evenly distributed over the support (FigureA8 A).
The measured mean particles size is 2.9±0.9 nm which fits well to the crystallite size
estimated from XRD (2.5 ± 0.8 nm). After reaction, the samples look very different. After the
CO dropout experiment and high temperature run particles are still evenly distributed
(FigureA8 B,C). Particle sizes for the CO dropout sample are very small and comparable to
the state after H2 treatment which is consistent with XRD measurements. Particle sizes after
reaction at high temperature increased to 3.9 ± 0.9 nm. Such an particle growth has not
been observed in XRD studies.
The sample obtained after the kinetic study shows the biggest particles and broadest
particle size distribution. Individual particles are located closely together building up large
agglomerates (FigureA8 C). The measurement of particle sizes is challenging for this
sample since it is difficult to differentiate between single particles. This might be a reason
why the obtained particle size is slightly smaller than suggested by XRD. The formation of
agglomerates clearly indicates high mobility of Rh on the silica surface. However, the
mobility does not necessarily lead to sintering and formation of very large metallic particles.
171
FigureA8
ScanningtransmissionelectronmicroscopyanalysisoftheRh/SiO
2
catalystafterH
2
treatment(A),after
COdropout(B),afterreactionathightemperatures(C),andafterthekineticstudy(D).Theleftcolumnshows
HAADFimagesatamagnificationof152.8kx.ThemiddlecolumnshowsBFimageswithamagnificationof842.1kx.
Therightcolumnshowsparticlediameterdistributionsobtainedfromimageswithamagnificationof305.6kx.
Morethan200particleswereevaluatedforeachsample.
172
Agglomeration instead of sintering might suggest non-crystalline particles or particles with
non-crystalline surface structures, e.g. induced by integration of adsorbed species into the
structure. Also, the high mobility of Rh is again observed for the sample from the kinetic
study, supporting the hypothesis that reaction conditions – composition of atmosphere in
particular – has major impact on the mobility of Rh species. For Co catalysts, a mechanism
has been proposed that relates the mobility of Co on a silica surface to the partial pressures
of CO and water enabling the formation of highly mobile (sub)carbonyl species.65,141
The observed structural details are very heterogeneous when different domains are
compared for the kinetic study sample. On some silica particles, the majority of Rh particles
have gathered in agglomerates with a size of 50–100 nm (FigureA9 A). In other domains,
much smaller agglomerates (20–40 nm) have formed and some isolated particles are
present as well (FigureA9 B). The third domain features the extreme case with all Rh
particles observed in this image being agglomerated into one big structure (200 nm, Figure
A9 C).
FigureA9
ComparisonofthreeinvestigateddomainsofthesamesampleRh/SiO
2
obtainedafterthekineticstudy
(BC2408).Imagesweretakenatamagnificationof77.2kx.
The catalyst particles featuring the different domains might come from different positions
in the catalytic bed. The changing gas atmosphere along the catalytic bed might have crucial
impact on the mobility of Rh species and therefore on the tendency of Rh particles to
agglomerate. We observed the same agglomeration affect also on other Rh/SiO2 samples
after previous catalytic testing involving long time on stream and a partial pressure
variation (FigureA10). The effect is thus likely to be real and not an artefact of individual
measurements. To the best of our knowledge, this effect has not been reported in literature
before.
173
FigureA10
HAADF‐STEMimagesofaRh/SiO
2
sampleobtainedafterapreliminarykineticstudy(DinoRun15‐2,
BC1524).InthisexperimenttheRh/SiO2catalystwastestedwithfourdifferentcatalystvolumes.Reaction
conditionsweresimilartoDinoRun26(kineticstudy)discussedindetailinthiswork.
XRD results suggested that Mn addition has a stabilizing effect on small Rh particles. This
hypothesis is further supported by electron microscopy results. On the H2 treated
RhMn/SiO2 sample, particles are well distributed over the support and their mean size
agrees with XRD results (FigureA11). In contrast to the spent Rh/SiO2 samples, all
RhMn/SiO2 samples after reaction look very similar to the H2 treated sample before
reaction. Particle sizes on all samples are the same in the range of error and particles stay
evenly distributed. No agglomeration was observed indicating that Mn addition leads to
significantly less mobile metal species.
The same considerations were made for the RhFe/SiO2 catalysts. The H2 treated sample
shows slightly bigger particles than observed in XRD but still evenly distributed over the
support (FigureA12 A). Particle sizes obtained for the STE treated samples suggest no
significant particle growth during reaction. However, for the spent sample of the kinetic
study agglomerates have formed (FigureA12 D). The structures are similar to the
agglomerates observed for Rh/SiO2 but their sizes are smaller with 20–100 nm.
Fe addition does not have the same stabilizing effect as Mn addition. For the spent samples
from the high temperature and CO dropout experiments, agglomeration was not observed
which is consistent with the findings for Rh/SiO2.
The STEM and STEM-EDX analyses for RhMnFe/SiO2 combine the effects observed for the
bimetallic catalysts. Over all samples, metallic particles are evenly distributed without
tendency to build agglomerated structures similar to RhMn/SiO2 (FigureA13). Particle
sizes increased slightly after the CO dropout experiment and kinetic study but are still
comparable in the range of measurement error.
174
FigureA11
ScanningtransmissionelectronmicroscopyanalysisoftheRhMn/SiO
2
catalystafterH
2
treatment(A),
afterparameterfieldtesting(B),afterreactionathightemperatures(C),andafterCOdropout(D).Theleftcolumn
showsHAADFimagesatamagnificationof152.8kx.ThemiddlecolumnshowsBFimageswithamagnificationof
842.1kx.Therightcolumnshowsparticlediameterdistributionsobtainedfromimageswithamagnificationof
305.6kx.Morethan200particleswereevaluatedforeachsample.
175
FigureA12
ScanningtransmissionelectronmicroscopyanalysisoftheRhFe/SiO
2
catalystafterH
2
treatment(A),
afterCOdropout(B),afterreactionathightemperatures(C),andafterthekineticstudy(D).Theleftcolumnshows
HAADFimagesatamagnificationof152.8kx.ThemiddlecolumnshowsBFimageswithamagnificationof842.1kx.
Therightcolumnshowsparticlediameterdistributionsobtainedfromimageswithamagnificationof305.6kx.
Morethan200particleswereevaluatedforeachsample.
176
FigureA13
ScanningtransmissionelectronmicroscopyanalysisoftheRhMnFe/SiO
2
catalystafterH
2
treatment(A),
afterCOdropout(B),afterreactionathightemperatures(C),andafterthekineticstudy(D).Theleftcolumnshows
HAADFimagesatamagnificationof152.8kx.ThemiddlecolumnshowsBFimageswithamagnificationof842.1kx.
Therightcolumnshowsparticlediameterdistributionsobtainedfromimageswithamagnificationof305.6kx.
Morethan200particleswereevaluatedforeachsample.
177
B.3 Elemental composition and profiles of RhFe particles
For RhFe/SiO2, two domains were investigated for their elemental composition. The first
domain features small particles containing Rh and Fe along with some particles containing
only Rh (FigureA14). The presence of pure Rh crystallites is consistent with XRD results.
The elemental composition of the whole domain features a Rh:Fe ratio of 3 corresponding
to the ratio implied by synthesis. Investigation of the elemental composition of individual
RhFe particles yields Rh:Fe ratios of 1.8–5.7 (TableA7). A clear statement regarding the
stoichiometry of RhFe particles cannot be made. It is rather likely, that particles only partly
form nanoalloys or that particles with different Fe content overlap in the mapping. Although
there is a strong indication, the EDX scan does not finally prove alloy formation and more
experiments would be necessary, e.g. X-ray adsorption or Mößbauer spectroscopy.
However, the formation of nanoalloyed RhFe particles on SiO2 and TiO2 has been proposed
from electron microscopy and H2 chemisorption before.86 XRD and pair distribution
functions were used to quantify each phase and provide information on local atomic
structure under reaction conditions with a total pressure of 2 bar.110 The authors proposed
RhFe (1:1) alloying on the particle surface and Rh cores.
In order to investigate core-shell structures in this study, Rh and Fe line scans were
evaluated on bigger particles (FigureA14). Synchronous profiles for Rh and Fe along the
line scans rather suggest a bulk alloy formation than core-shell particles.
FigureA14
EDXmappingofRhandFeoftheRhFe/SiO
2
sampleobtainedafterthekineticstudyinDomain1and
Domain2.Yellowsquaresindicateareasforcompositionalevaluation,greenarrowsprofilesforlinescananalysis
andredarrowsindicateRhparticleswithlittleornoFecontent.
178
TableA7CompositionofRhandFecontainingparticlesandaggregatesontheRhFe/SiO2afterthekineticstudy
(DinoRun26)evaluatedfromEDXspectraindifferentareas.
AreaComposition/atom‐%Rh:Feratio
SiORhFe
Domain135±864±90.8±0.20.4±0.12.00
Area#128±653±611±27.7±1.31.42
Area#233±862±82.9±0.52.1±0.41.38
Area#329±760±76.1±1.14.2±0.71.45
Domain231±767±91.5±0.30.5±0.13.00
Area#119±448±728±54.9±0.95.71
Area#230±763±84.4±0.92.0±0.42.20
Area#329±766±83.0±0.61.6±0.31.88
Area#430±767±82.6±0.50.8±0.23.25
Area#530±765±83.5±0.71.4±0.32.50
Area#628±660±78.6±1.53.6±0.62.39
Area#723±552±719±36.0±1.03.17
179
C Stability and reproducibility of catalytic data
C.1 Effect of temperature on CO conversion and product selectivities
FigureA15COconversionoverRh/SiO2(A,blacksquares)andRhMn/SiO2(B)overtimeonstreamforatemperature
rampfrom243–320°C(grey).Dataobtainedatreferenceconditions(260°C)aremarkedbygreyboxes.The
associatedproductselectivitiesaredepictedincolumnsbelowtheconversiondata.(DinoRun27,54bar,
H2:CO:N2:Ar=60:20:10:10,41.7mlmin‐1,Rh:2500h‐1,RhMn:12500h‐1)
10
20
30
40
0
23 27 32 37 43 50 56 61 67
27
2223
3
3
333
22 22
2
12 13 15
16
18
19 18 18 16
15
15 14
13
11
10 7
12
33 28 22 17 11 8
30
120 140 160 180 200 220 240 260 280 300 320 340 360 380
0
20
40
60
80
Time on stream / h
Product selectivities / %
CO2
Ethyl acetate
Methyl acetate
Acetic acid
Propanal
Acetaldehyde
n-Propanol
Ethanol
Methanol
1-Butene
Propylene
Ethylene
n-Butane
Propane
Ethane
Methane
Methane
Propylene
Ethane
Acetic acid
Acetaldehyde
Ethanol
Temperature
CO conversion
-37%
RhMn
CO conversion / %
250
275
300
325
Temperature / °C
10
20
30
0
40
44 47 52 59 66 71 77 81 85
47
444
4
44444
3
322
1
445
555543
6
16 16 14 11 96
15
19 18 15 11 86
19
120 140 160 180 200 220 240 260 280 300 320 340 360 380
0
20
40
60
80
Time on stream / h
Product selectivities / %
CO2
Ethyl acetate
Methyl acetate
Acetic acid
Propanal
Acetaldehyde
n-Propanol
Ethanol
Methanol
1-Butene
Propylene
Ethylene
n-Butane
Propane
Ethane
Methane
Methane
Propylene
Ethane
Acetic acid
Acetaldehyde
Ethanol
Temperature
CO conversion
-16%
Rh
CO conversion / %
250
275
300
325
Temperature / °C
A
B
180
FigureA16COconversionoverRhFe/SiO2(A,blacksquares)andRhMnFe/SiO2(B)overtimeonstreamfora
temperaturerampfrom243–320°C(grey).Dataobtainedatreferenceconditions(260°C)aremarkedbygrey
boxes.Theassociatedproductselectivitiesaredepictedincolumnsbelowtheconversiondata.(DinoRun27,54bar,
H2:CO:N2:Ar=60:20:10:10,41.7mlmin‐1,RhFe:8330h‐1,RhMnFe:7580h‐1)
10
20
30
40
0
32 35 39 44 48 54 58 61 65
38
26 24 21 17 14 12 10 87
25
27 27 27 27 25 23 21 18 14 25
120 140 160 180 200 220 240 260 280 300 320 340 360 380
0
20
40
60
80
Time on stream / h
Product selectivities / %
CO
2
Ethyl acetate
Methyl acetate
Acetic acid
Propanal
Acetaldehyde
n-Propanol
Ethanol
Methanol
1-Butene
Propylene
Ethylene
n-Butane
Propane
Ethane
Methane
Methane
Acetic acid
Ethanol
Methanol
Temperature
CO conversion -36%
RhMnFe
CO conversion / %
250
275
300
325
Temperature / °C
10
20
30
40
0
23 25 29 33 37 42 48 53 59
27
55 54 50 46 44 40 35 29 23
53
11 12 12 13 13 13 11 10 710
120 140 160 180 200 220 240 260 280 300 320 340 360 380
0
20
40
60
80
Time on stream / h
Product selectivities / %
CO
2
Ethyl acetate
Methyl acetate
Acetic acid
Propanal
Acetaldehyde
n-Propanol
Ethanol
Methanol
1-Butene
Propylene
Ethylene
n-Butane
Propane
Ethane
Methane
Methane
Acetic acid
Ethanol
Methanol
Temperature
CO conversion -24%
RhFe
CO conversion / %
250
275
300
325
Temperature / °C
A
B
181
C.2 Arrhenius plots for Rh/Mn/Fe/SiO2 catalysts obtained at 243-320 °C
ArrheniusplotsRh/SiO2
FigureA17ArrheniusplotsofCOandH2consumption(A),paraffins(B),olefins(C),alcohols(D),oxygenates(E)and
CO2(F)overRh/SiO2rangingfrom243°Cto320°C.Linearfitsinthetemperaturerangefrom243°Cto270°Cwere
usedtodetermineapparentactivationenergies.(DinoRun27,54bar,H2:CO:N2:Ar=60:20:10:10,41.7mlmin‐1,
0.5gcat,2500h‐1)
1.65 1.70 1.75 1.80 1.85 1.90 1.95
-20
-10
0
10
20
30
1.65 1.70 1.75 1.80 1.85 1.90 1.95
-80
-60
-40
-20
0
20
1.65 1.70 1.75 1.80 1.85 1.90 1.95
-80
-70
-60
-50
-40
1.65 1.70 1.75 1.80 1.85 1.90 1.95
-80
-70
-60
-50
-40
-30
-20
1.65 1.70 1.75 1.80 1.85 1.90 1.95
-80
-70
-60
-50
-40
-30
-20
1.65 1.70 1.75 1.80 1.85 1.90 1.95
-60
-50
-40
-30
-20
ln(r)*R
1000 / T
CO and H
2
consumption
CO consumption
H
2
consumption
ln(r)*R
1000 / T
Paraffins
Methane
Ethane
Propane
Butane
Pentane
ln(r)*R
1000 / T
Olefins
Ethylene
Propylene
Butene
ln(r)*R
1000 / T
Alcohols
Propanol
Ethanol
Methanol
ln(r)*R
1000 / T
Oxygenates
Acetic acid
Methyl acetate
Acetaldehyde
Propanal
Butanal
Ethyl acetate
ln(r)*R
1000 / T
CO
2
CO
2
AB
CD
EF
182
ArrheniusplotsRhMn/SiO2
FigureA18ArrheniusplotsofCOandH2consumption(A),paraffins(B),olefins(C),alcohols(D),oxygenates(E)and
CO2(F)overRhMn/SiO2rangingfrom243°Cto320°C.Linearfitsinthetemperaturerangefrom243°Cto270°C
wereusedtodetermineapparentactivationenergies.(DinoRun27,54bar,H2:CO:N2:Ar=60:20:10:10,41.7mlmin‐1,
0.1gcat,12500h‐1)
1.65 1.70 1.75 1.80 1.85 1.90 1.95
0
10
20
30
40
1.65 1.70 1.75 1.80 1.85 1.90 1.95
-60
-40
-20
0
20
40
1.65 1.70 1.75 1.80 1.85 1.90 1.95
-60
-50
-40
-30
-20
-10
1.65 1.70 1.75 1.80 1.85 1.90 1.95
-60
-40
-20
0
20
1.65 1.70 1.75 1.80 1.85 1.90 1.95
-60
-50
-40
-30
-20
-10
0
1.65 1.70 1.75 1.80 1.85 1.90 1.95
-40
-30
-20
-10
0
10
ln(r)*R
1000 / T
CO and H
2
consumption
CO consumption
H
2
consumption
ln(r)*R
1000 / T
Paraffins
Methane
Ethane
Propane
Butane
ln(r)*R
1000 / T
Olefins
Ethylene
Propylene
Butene
ln(r)*R
1000 / T
Alcohols
Propanol
Ethanol
Methanol
ln(r)*R
1000 / T
Oxygenates
Acetic acid
Methyl acetate
Acetaldehyde
Ethyl acetate
Propanal
Butanal
ln(r)*R
1000 / T
CO
2
CO
2
AB
CD
EF
183
ArrheniusplotsRhFe/SiO2
FigureA19ArrheniusplotsofCOandH2consumption(A),paraffins(B),alcohols(C),oxygenatesandCO2(D)over
RhFe/SiO2rangingfrom243°Cto320°C.Linearfitsinthetemperaturerangefrom243°Cto270°Cwereusedto
determineapparentactivationenergies.(DinoRun27,54bar,H2:CO:N2:Ar=60:20:10:10,41.7mlmin‐1,8330h‐1)
1.65 1.70 1.75 1.80 1.85 1.90 1.95
-10
0
10
20
30
40
1.65 1.70 1.75 1.80 1.85 1.90 1.95
-80
-60
-40
-20
0
20
40
1.65 1.70 1.75 1.80 1.85 1.90 1.95
-40
-30
-20
-10
0
10
20
1.65 1.70 1.75 1.80 1.85 1.90 1.95
-80
-60
-40
-20
0
20
ln(r)*R
1000 / T
CO and H
2
consumption
CO consumption
H
2
consumption
ln(r)*R
1000 / T
Paraffins
Methane
Ethane
Propane
ln(r)*R
1000 / T
Alcohols
Ethanol
Methanol
ln(r)*R
1000 / T
Oxygenates and CO
2
Acetic acid
Methyl acetate
Acetaldehyde
Propanal
Ethyl acetate
CO
2
AB
CD
184
ArrheniusplotsRhMnFe/SiO2
FigureA20ArrheniusplotsofCOandH2consumption(A),paraffins(B),olefins(C),alcohols(D),oxygenates(E)and
CO2(F)overRhMnFe/SiO2rangingfrom243°Cto320°C.Linearfitsinthetemperaturerangefrom243°Cto270°C
wereusedtodetermineapparentactivationenergies.(DinoRun27,54bar,H2:CO:N2:Ar=60:20:10:10,41.7mlmin‐1,
7580h‐1)
1.65 1.70 1.75 1.80 1.85 1.90 1.95
-10
0
10
20
30
40
1.65 1.70 1.75 1.80 1.85 1.90 1.95
-80
-60
-40
-20
0
20
40
1.65 1.70 1.75 1.80 1.85 1.90 1.95
-80
-70
-60
-50
-40
1.65 1.70 1.75 1.80 1.85 1.90 1.95
-60
-40
-20
0
20
1.65 1.70 1.75 1.80 1.85 1.90 1.95
-60
-50
-40
-30
-20
1.65 1.70 1.75 1.80 1.85 1.90 1.95
-40
-30
-20
-10
0
10
20
ln(r)*R
1000 / T
CO and H
2
consumption
CO consumption
H
2
consumption
ln(r)*R
1000 / T
Paraffins
Methane
Ethane
Propane
Butane
ln(r)*R
1000 / T
Olefins
Ethylene
Propylene
ln(r)*R
1000 / T
Alcohols
Propanol
Ethanol
Methanol
ln(r)*R
1000 / T
Oxygenates
Acetic acid
Methyl acetate
Acet
-aldehyde
Ethyl acetate
ln(r)*R
1000 / T
CO
2
CO
2
AB
CD
EF
185
D Reaction network and kinetics for STE over Rh/SiO2
D.1 Apparent activation energies over Rh/SiO2
Apparent activation energies were calculated from the Arrhenius equation (equation (10)
in chapter 3.6.1).
For evaluation of apparent activation energies, data from Run 27 were considered at
temperatures from 243 °C – 270 °C. Above that temperature the catalytic behavior changed
probably due to structural impact of high temperatures.
The apparent activation energy for H2 consumption is slightly higher than for CO
consumption with 88 ± 6 kJ mol-1 and 76 ± 5 kJ mol-1, respectively (FigureA21). This is first
evidence towards H2 activation being an important elementary step for catalyst activity.
Apparent activation energies for alcohols and paraffins with small carbon numbers are in a
similar range of 90 – 110 kJ mol-1. For higher carbon number the activation energy appears
to decrease. Here, a similar activation energy suggests a common rate determining step
either at the beginning of the product formation process and/or for the final hydrogenation
of different intermediates to the hydrogenated products.
The obtained energies for aldehyde, acetic acid and acetate formation are unusually small.
It must be considered that the obtained data are not differential for those products as
already indicated by the nonlinear behavior in the Arrhenius plots (FigureA17 in Appendix
C.2). Consecutive reactions such as hydrogenation reactions lead to false small values. For
olefins, this effect was too pronounced to calculate apparent activation energies from those
plots. The assignment of both product types, aldehydes/acids and olefins, into the same
group can therefore also be associated with a common consecutive consumption reaction
in the network.
Two kinetic studies with comparable catalysts reported values for apparent activation
energies. Both used 1.5 wt-% Rh/SiO2 as opposed to 2.8 wt-% in this study. Regarding
reaction conditions, Gao etal. used similar gas phase composition and GHSV but lower total
pressure.123 Only few values for activation energies are given in literature. The reported
activation energies for CO consumption is higher (107 kJ mol-1) compared to the results of
this study (76 kJ mol-1). The activation energy for ethanol, however, is smaller than found
here (78 kJ mol-1 vs. 94 kJ mol-1). The different results might be a consequence of the
different catalyst loadings and/or the much shorter time allowed for catalyst equilibration
of 15 h before starting the kinetic measurements.
186
FigureA21ApparentactivationenergiescalculatedfromanArrheniusequationfortotalCOandH2consumptionas
wellasproductformationoverRh/SiO2.(DinoRun27Rh,54bar,243‐260°C,H2:CO:N2:Ar=60:20:10:10,2500h‐1)
In the second study, Mao etal. used similar conditions but with a much higher GHSV.84 The
results for paraffins and ethanol were reported to be slightly smaller than in this study but
still in good agreement. Even the decreasing activation energy for butane formation
compared to smaller paraffins is consistent with the present study. The high GHSV allows
to also obtain meaningful values for aldehydes and olefins. For olefins, the reported values
are ~10 – 20 kJ mol-1 smaller than for the respective paraffin with the same carbon number.
The apparent activation energy for acetaldehyde is 85.2 ± 3.5 kJ mol-1 and therefore the
same as for ethanol (84.3 ± 3.0 kJ mol-1) and similar to acetic acid (74.2 ± 3.0 kJ mol-1).
In conclusion, apparent activation energies were obtained for CO and H2 consumption as
well as alcohols and paraffins in this study. H2 consumption shows a slightly increased value
compared to CO consumption suggesting that H2 activation is an important step for overall
activity. Alcohols and paraffins show comparably high values of ~90 – 100 kJ mol-1. Those
similar values could indicate common rate determining steps. The results are largely in
accordance with a study of Mao etal. using similar conditions. In their study, supposedly
more accurate values for aldehydes, acetic acid and olefins were proposed. According to
that, olefins show slightly smaller values while oxygenates are comparable to alcohols.
0
20
40
60
80
100
120
140
160
Apparent activation energy / kJ mol
-1
Methane
Ethane Ethanol
Propanol
Acetaldehyde
Acetic acid
Propane
CO consumption
H
2
consumption
CO consumption
H
2
consumption
Methane
Ethane
Propane
n-Butane
n-Pentane
Methanol
Ethanol
n-Propanol
Acetaldehyde
Propanal
Butanal
Acetic acid
Methyl acetate
Ethyl acetate
Methanol
Acetates
Propanal
187
TableA8ApparentactivationenergiesaofCOandH2consumptionandproductformation.
Ea,app/kJmol‐1
OurdatabGaoetal.123,cMaoetal.83,d
COconsumption76±5107±11‐
H2consumption88±6‐ ‐
Methane103±7122±1287.2±1.1
Ethane90±9124±1284.4±2.2
n‐Propane99±11102±1082.6±4.4
n‐Butane78±11‐ 53.6±2.1
n‐Pentane68±10‐ ‐
Ethylene‐ ‐ 71.5±4.1
Propylene‐ ‐ 64.6±4.2
1‐Butene‐ ‐ 36.4±6.2
Methanol109±5‐ 52.5±3.3
Ethanol94±277±884.3±3.7
n‐Propanol99±2‐ 76.8±3.5
Acetaldehyde38±4‐ 85.2±5.0
Propanal28±7‐ ‐
Butanal39±10‐ ‐
Aceticacid13±7‐ 74.2±4.7
Methylacetate38±4‐ ‐
Ethylacetate50±3‐ ‐
aApparentactivationenergiesweredeterminedbyfittinganArrheniusexpression.
bDinoRun26,2.8wt‐%Rh/SiO2,54bar,243‐260°C,H2:CO:N2:Ar=60:20:10:10,41.7mlmin‐1,0.5gcat,2500h‐1
c1.5wt‐%Rh/SiO2,1.8bar,210‐270°C,H2:CO=2,0.3gcat,4500h‐1
d1.5wt‐%Rh/SiO2,30bar,230‐290°C,H2:CO:N2=60:30:10,0.1gcat,15000h‐1
188
D.2 Influence of CO and H
2
partial pressure on product selectivities
FormationratesofC1andC2productsasfunctionoffeedcomposition
FigureA22
ProductselectivitiesofC1andC2productsasafunctionofCOandH
2
partialpressuressorted
horizontallybycarbonnumberandverticallybyoxidationstateoftherespectivefunctionalgroup.(DinoRun26,
Rh/SiO
2
,54bar,260°C,H
2
:CO:N
2
:Ar=30‐60:2.5‐20:10‐57.5:10,41.7mlmin
‐1
,0.5gcat,2500h
‐1
)
189
FormationratesofC3toC5productsasfunctionoffeedcomposition
FigureA23
ProductselectivitiesofC3toC5productsasafunctionofCOandH
2
partialpressuressortedhorizontally
bycarbonnumberandverticallybyoxidationstateoftherespectivefunctionalgroup.(DinoRun26,Rh/SiO
2
,54bar,
260°C,H
2
:CO:N
2
:Ar=30‐60:2.5‐20:10‐57.5:10,41.7mlmin
‐1
,0.5gcat,2500h
‐1
)
190
D.3 Dropout experiments
Reactantconcentrationsduringdropoutexperiments
FigureA24ConcentrationsofCO(blacksquares)andH2(greydots)inthebypasslineovertimeonstreamduringa
H2dropout(A)andCOdropoutexperiment(B).(DinoRun29,54bar,260°C,syngasH2:CO:N2:Ar=60:20:10:10,for
dropoutexperimentstherespectivereactantwasreplacedbyadditionalN2,0.5gcat,2500h‐1)
300 320 340 360 380 400 420
10
20
30
40
50
60
0
360 380 400 420 440 460 480 500 520340
Reactant concentration / vol-%
Time on stream / h
syngas
CO + H
2
CO/N
2
syngas
CO + H
2
CO
syngas
CO + H
2
H
2
/N
2
syngas
CO + H
2
H
2
CO
H
2
COdropout
H2dropout
AB
Exchangeof
atmospherefast
comparedtoproduct
concentrationsin
dropoutexperiments
191
E Modification of Rh/SiO2 with Mn and/or Fe
E.1 Apparent activation energies for Rh/Mn/Fe/SiO2 catalysts
Apparent activation energies were calculated using an Arrhenius equation as described in
chapter 3.6.1. The corresponding Arrhenius plots and a summary of apparent activation
energies over all four catalysts can be found in FigureA17-FigureA20 in Appendix C.2 and
TableA9 at the end of this section, respectively.
Compared to Rh/SiO2, RhMn/SiO2 shows increased apparent activation energies of total CO
and H2 consumption (FigureA25A,B). The apparent activation energy of CO consumption
increased from 68.7 kJ mol-1 to 102.9 kJ mol-1. The values for RhFe/SiO2 are comparable to
Rh/SiO2 while the results for RhMnFe/SiO2 are more similar to RhMn/SiO2 (Figure
A25C,D). The presence of Mn oxide appears to increase the overall apparent activation
energy of syngas conversion. However, it is possible that the computed values for apparent
activation energies might also be influenced by temperature induced structural
modifications of the catalysts. As pointed out in the previous chapter, different Rh species
(metallic Rh and Rh (sub)carbonyls) might be responsible for specific reactivity. For
example, a thermal decomposition of the Rh (sub)carbonyl species might greatly affect
apparent activation energies. The catalyst composition might thereby have an impact on
presence and stability of such species. A comparative mechanistic interpretation of this data
therefore might be misleading. However, the data contains information about related
reaction pathways for each catalyst.
In contrast to Rh/SiO2, apparent activation energies for higher paraffins do not decrease
with carbon number over RhMn/SiO2. The effect of Fe addition is less pronounced. The
apparent activation energy for higher paraffins are rather comparable to Rh/SiO2 in the
range of error. The values obtained for RhMnFe/SiO2 were found to be between the results
for RhMn/SiO2 and RhFe/SiO2. Apparent activation energies for olefins could not be
obtained reliably due to the high tendency to undergo consecutive reactions.
While over Rh/SiO2 all values for alcohols were similar, Mn addition leads to an increase of
apparent activation energies for ethanol and propanol by ~40 kJ mol-1 while the apparent
activation energy for methanol decreases by 30 kJ mol-1. Fe addition causes an even more
severe decrease of methanol activation energy, but no significant impact on ethanol
activation energy and an increase of propanol activation energy. The reduced value for
methanol formation is consistent with the high methanol formation rate observed for
RhFe/SiO2. Over RhMnFe/SiO2 the same small value for methanol activation energy was
found. The results for ethanol and propanol are comparable to Rh/SiO2.
192
FigureA25ApparentactivationenergiescalculatedfromanArrheniusequationfortotalCOandH2consumptionas
wellasproductformationoverRh/SiO2(A),RhMn/SiO2(B),RhFe/SiO2(C)andRhMnFe/SiO2(D).(DinoRun27,54bar,
243–270°C,H2:CO:N2:Ar=60:20:10:10,41.7mlmin‐1,Rh:2500h‐1;RhMn:12500h‐1;RhFe:8330h‐1;
RhMnFe:7580h‐1)
0
20
40
60
80
100
120
140
160
180
200
Methane Ethane Ethanol
Propanol
Acetaldehyde
Acetic acid
Propane
CO consumption
H
2
consumption
CO consumption
H
2
consumption
Methane
Ethane
Propane
n-Butane
n-Pentane
Methanol
Ethanol
n-Propanol
Acetaldehyde
Propanal
Butanal
Acetic acid
Methyl acetate
Ethyl acetate
Methanol
Acetates
Propanal
CO consumption
H
2
consumption
Methane
Ethane
Propane
n-Butane
n-Pentane
Methanol
Ethanol
n-Propanol
Acetaldehyde
Propanal
Butanal
Acetic acid
Methyl acetate
Ethyl acetate
CO consumption
H
2
consumption
Methane
Ethane
Propane
n-Butane
Methanol
Ethanol
n-Propanol
Acetaldehyde
Acetic acid
Methyl acetate
Ethyl acetate
CO consumption
H
2
consumption
Methane
Ethane
Propane
n-Butane
n-Pentane
Methanol
Ethanol
n-Propanol
Acetaldehyde
Acetic acid
Methyl acetate
Ethyl acetate
0
20
40
60
80
100
120
140
160
180
200
Apparent activation energy / kJ mol
-1
Methane
Ethanol Propanol
Acetaldehyde
Propane
CO consumption
H
2
consumption
Methanol
Acetates
Propanal
Acetic acid
Ethane
0
20
40
60
80
100
120
140
160
180
200
Methane
Ethane
Ethanol Propanol
Propane
CO consumption
H
2
consumption
Methanol
Acetates
Acetaldehyde
0
20
40
60
80
100
120
140
160
180
200
Methane
Ethane
Ethanol
Propanol
Acetaldehyde
CO consumption
H
2
consumption
Methanol
Acetates
Acetic acid
Propane
A
B
Rh
RhMn
CRhFe
DRhMnFe
193
The evaluation of apparent activation energies does not directly explain for the reactivity in
section 7.1. Although RhMn/SiO2 shows enhanced formation rates, increased apparent
activation energies were observed. Mao etal. reported rapid increase of the apparent
methane activation energy but little change of the value for ethanol upon Mn addition.83 The
increase of the methane activation energy was made responsible for higher ethanol yields.
We did not observe such a behavior. The increase of reaction rates therefore rather seems
to be related to a higher number of active sites that might be a result of the stabilized small
Rh particles surrounded by a Mn oxide matrix as described in chapter 4.2.
Fe addition is reported to have a slightly decreasing effect on paraffin and ethanol activation
energies.123 The presented data of this study rather suggests a slight increase or no impact
of Fe on apparent activation energies with exception of methanol. The reduction of
methanol activation energy from 109.4 kJ mol-1 over Rh/SiO2 to 66.9 kJ mol-1 over
RhFe/SiO2 might at least partially explain the high methanol productivity of this catalyst.
TableA9Apparentactivationenergiesforreactantconsumptionandproductformationobtainedinatemperature
rangefrom243°Cto270°C.
Apparentactivationenergiesa,b/kJmol‐1
RhRhMnRhFeRhMnFe
COconsumption75.7±5.1103.8±4.183.1±2.5102.2±3.3
H2consumption87.6±5.9118.7±4.589.9±2.8106.4±3.2
Methane103.0±6.9143.8±4.2117.5±3.7128.1±3.5
Ethane90.3±8.7143.8±6.8105.2±4.2112.8±3.8
n‐Propane99.3±11.4148.4±8.9107.5±4.8124.8±6.2
n‐Butane78.3±10.7143.2±8.694.1±1.0116.8±7.7
n‐Pentane68.0±10.2136.5±8.8‐ ‐
Ethylene‐ 33.4±3.5‐ 37.0±7.9
Propylene‐113.4±6.7‐ 64.5±5.9
1‐Butene‐ 90.6±5.8‐ ‐
Methanol109.4±5.078.9±8.866.9±2.163.4±1.0
Ethanol93.7±2.3132.7±3.198.3±4.8101.1±4.2
n‐Propanol98.7±2.3135.4±6.3143.6±15.092.1±3.7
Acetaldehyde37.9±4.476.5±8.169.0±31.174.9±6.2
Propanal28.2±7.379.7±7.3‐ ‐
Butanal39.9±9.9117.9±7.5‐ ‐
Aceticacid12.5±7.043.6±3.2‐ 55.8±5.6
Methylacetate38.4±3.767.5±6.144.3±3.960.9±2.5
Ethylacetate50.1±3.1120.7±4.457.3±6.093.4±5.9
aApparentactivationenergiesweredeterminedbyfittinganArrheniusexpression.
bExperimentaldetails:DinoRun27,54bar,H2:CO:N2:Ar=60:20:10:10,41.7mlmin‐1,Rh:2500h‐1;RhMn:12500h‐1;
RhFe:8330h‐1;RhMnFe:7580h‐1.
194
E.2 Influence of reactant partial pressures on selectivities over Rh/Mn/Fe/SiO
2
Figure9.1
ProductselectivitiesofC1andC2productsoverRhMn/SiO
2
asafunctionofCOandH
2
partialpressures
sortedhorizontallybycarbonnumberandverticallybyoxidationstateofthecarbonatomoftherespective
functionalgroup.(RhMn/SiO
2
,DinoRun26,54bar,260°C,H
2
:CO:N
2
:Ar=30‐60:2.5‐20:10‐57.5:10,41.7mlmin
‐1
,
12500h
‐1
)
195
Figure9.2
ProductselectivitiesofC3toC5productsoverRhMn/SiO
2
asafunctionofCOandH
2
partialpressures
sortedhorizontallybycarbonnumberandverticallybyoxidationstateofthecarbonatomoftherespective
functionalgroup.(RhMn/SiO
2
,DinoRun26,54bar,260°C,H
2
:CO:N
2
:Ar=30‐60:2.5‐20:10‐57.5:10,41.7mlmin
‐1
,
12500h
‐1
)
196
Figure9.3
ProductselectivitiesofC1andC2productsandmethylacetateoverRhFe/SiO
2
asafunctionofCOandH
2
partialpressuressortedhorizontallybycarbonnumberandverticallybyoxidationstateofthecarbonatomofthe
respectivefunctionalgroup.(RhFe/SiO
2
,DinoRun26,54bar,260°C,H
2
:CO:N
2
:Ar=30‐60:2.5‐20:10‐57.5:10,
41.7mlmin
‐1
,25000h
‐1
)
197
Figure9.4
ProductselectivitiesofC1andC2productsoverRhMnFe/SiO
2
asafunctionofCOandH
2
partial
pressuressortedhorizontallybycarbonnumberandverticallybyoxidationstateofthecarbonatomofthe
respectivefunctionalgroup.(RhMn/SiO
2
,DinoRun26,54bar,260°C,H
2
:CO:N
2
:Ar=30‐60:2.5‐20:10‐57.5:10,
41.7mlmin
‐1
,7580h
‐1
)
198
Figure9.5
ProductselectivitiesofC3toC5productsoverRhMnFe/SiO
2
asafunctionofCOandH
2
partialpressures
sortedhorizontallybycarbonnumberandverticallybyoxidationstateofthecarbonatomoftherespective
functionalgroup.(RhMn/SiO
2
,DinoRun26,54bar,260°C,H
2
:CO:N
2
:Ar=30‐60:2.5‐20:10‐57.5:10,41.7mlmin
‐1
,
7580h
‐1
)
199
E.3 Summary of apparent reaction orders for Rh/Mn/Fe/SiO2 catalysts
ApparentCOreactionordersforallfourtestedcatalysts
TableA10COreactionordersforreactantconsumptionandproductformation.
COreactionordersa,b
RhRhMnRhFeRhMnFe
lowhighlowhighlowhighlowhigh
p
CO
p
CO
p
CO
p
CO
p
CO
p
CO
p
CO
p
CO
COconsumption0.10.10.2 ‐0.1 ‐0.1 ‐0.1 ‐0.10.0
H2consumption0.00.00.1 ‐0.2 ‐0.1 ‐0.1 ‐0.1 ‐0.1
Methane ‐0.1 ‐0.3 ‐0.1 ‐0.5 ‐0.1 ‐0.1 ‐0.2 ‐0.2
Ethane0.50.10.5 ‐0.20.00.10.30.3
n‐Propane0.60.30.5 ‐0.1 ‐0.1 ‐0.30.50.4
n‐Butane0.90.60.80.1‐ ‐ 0.80.8
n‐Pentane1.20.71.00.2‐ ‐ ‐ ‐
Ethylene2.31.82.61.2‐ ‐ ‐ ‐
Propylene2.01.21.80.6‐ ‐ ‐ ‐
1‐Butene2.11.62.11.0‐ ‐ ‐ ‐
Methanol ‐0.3 ‐0.4 ‐0.6 ‐0.1 ‐0.1 ‐0.2 ‐0.2 ‐0.2
Ethanol ‐0.2 ‐0.50.1 ‐0.60.40.30.30.1
n‐Propanol0.3 ‐0.20.4 ‐0.1‐ ‐ 0.60.7
Acetaldehyde0.90.41.90.50.40.20.41.2
Propanal1.71.22.41.1‐ ‐ ‐ ‐
Butanal1.50.9‐ ‐ ‐ ‐ ‐ ‐
Aceticacid2.60.83.20.6‐ 2.0‐ 1.5
aReactionordersweredeterminedbyfittinglog‐log‐plots.
bExperimentaldetails:DinoRun26,54bar,260°C,H2:CO:N2:Ar=30‐60:2.5‐20:10‐57.5:10,41.7mlmin‐1,Rh:2500h‐1;
RhMn:12500h‐1;RhFe:25000h‐1;RhMnFe:7580h‐1.
200
ApparentH2reactionordersforallfourtestedcatalysts
TableA11H2reactionordersforreactantconsumptionandproductformation.
H2reactionordersa,b
RhRhMnRhFeRhMnFe
lowhighlowhighlowhighlowhigh
p
CO
p
CO
p
CO
p
CO
p
CO
p
CO
p
CO
p
CO
COconsumption0.80.90.70.81.00.90.90.7
H2consumption0.90.90.71.01.01.00.90.8
Methane1.01.30.81.30.70.80.80.9
Ethane0.31.10.21.20.50.50.30.4
n‐Propane0.00.80.11.00.5‐ 00.2
n‐Butane ‐0.30.5 ‐0.30.8‐ ‐ ‐ 0
n‐Pentane ‐0.70.4 ‐0.30.8‐ ‐ ‐ ‐0.2
Ethylene ‐1.50.0 ‐2.40.0‐ ‐ ‐ ‐
Propylene ‐1.50.1 ‐1.50.2‐ ‐ ‐ ‐
1‐Butene ‐1.4 ‐0.3‐ 0.0‐ ‐ ‐ ‐
Methanol2.23.92.01.51.21.21.51.4
Ethanol0.70.90.61.20.30.70.40.6
n‐Propanol0.51.00.41.8‐ ‐ ‐0.20.5
Acetaldehyde ‐0.10.7 ‐0.80.6 ‐1.0 ‐0.1 ‐0.8 ‐0.5
Propanal ‐0.50.6‐ 0.7‐ ‐ ‐ ‐
Butanal ‐0.60.6‐ ‐ ‐ ‐ ‐ ‐
Aceticacid ‐0.60.3 ‐3.00.3‐ ‐ ‐ ‐0.3
aReactionordersweredeterminedbyfittinglog‐log‐plots.
bExperimentaldetails:DinoRun26,54bar,260°C,H2:CO:N2:Ar=30‐60:2.5‐20:10‐57.5:10,41.7mlmin‐1,Rh:2500h‐1;
RhMn:12500h‐1;RhFe:25000h‐1;RhMnFe:7580h‐1.
201
E.4 Summary of cofeed effects over Rh/Mn/Fe/SiO2 catalysts
Conversionofdifferentcofeedcompoundsoverallfourtestedcatalysts
TableA12ConversionofrespectivecofeedcompoundsoverRh/SiO2,RhMn/SiO2,RhFe/SiO2,andRhMnFe/SiO2.
Cofeed
compoundConcentration/vol‐%
Conversiona/%
RhRhMnRhFeRhMnFe
Ethylene0.05710097100100
0.0310097100100
Propylene0.007847479466
0.004149459667
Acetaldehyde0.14949796100
0.07979897100
CO21.770000
3.480000
aCalculatedfromdifferenceofconcentrationsofcofeedcompoundsinreactoroutletwithandwithoutcofeed.
SelectivitytowardshydrogenationandCOinsertionproductsofolefincofeedsoverallfour
testedcatalysts
TableA13SelectivitytowardshydrogenationandCOinsertionproductsofethyleneandpropyleneoverRh/SiO2,
RhMn/SiO2,RhFe/SiO2,andRhMnFe/SiO2.
Cofeed
compound
Concentration
/vol‐%
Selectivityhydrogenation
producta/%
SelectivityCOinsertion
producta/%
RhRhMnRhFeRhMnFeRhRhMnRhFeRhMnFe
Ethylene0.05768746569 18192521
0.0366696867 18192422
Propylene0.007838287170 0000
0.004135277077 0000
aCalculatedfromdifferenceofproductconcentrationsinreactoroutletwithandwithoutcofeed.
