Catalytic Hydrogenation of Carbon Dioxide to Methanol using Molecular Catalysts Von der Fakultät für Mathematik, Informatik und Naturwissenschaften der RWTH Aachen University zur Erlangung des akademischen Grades eines Doktors der Naturwissenschaften genehmigte Dissertation vorgelegt von Master of Science Sebastian Wesselbaum aus Recklinghausen Berichter: Univ.-Prof. Dr. rer. nat. Walter Leitner Univ.-Prof. Dr. rer. nat. Jürgen Klankermayer Univ.-Prof. Dr. rer. nat. Sonja Herres-Pawlis Tag der mündlichen Prüfung: 16.12.2016 Diese Dissertation ist auf den Internetseiten der Universitätsbibliothek online verfügbar.
154
Embed
Catalytic Hydrogenation of Carbon Dioxide to Methanol ...
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Catalytic Hydrogenation of Carbon Dioxide to Methanol using
Molecular Catalysts
Von der Fakultät für Mathematik, Informatik und Naturwissenschaften der RWTH Aachen
University zur Erlangung des akademischen Grades eines Doktors der Naturwissenschaften
genehmigte Dissertation
vorgelegt von
Master of Science
Sebastian Wesselbaum
aus Recklinghausen
Berichter: Univ.-Prof. Dr. rer. nat. Walter Leitner
Univ.-Prof. Dr. rer. nat. Jürgen Klankermayer
Univ.-Prof. Dr. rer. nat. Sonja Herres-Pawlis
Tag der mündlichen Prüfung: 16.12.2016
Diese Dissertation ist auf den Internetseiten der Universitätsbibliothek online verfügbar.
-II-
Für meine liebe Familie.
-III-
I hereby declare that this thesis is my own original work, and that it has not been submitted
anywhere else for any award. Wherever contributions of others are involved, every effort
was made to indicate this clearly with due reference to the literature and acknowledgement
of collaborative research.
-IV-
The experimental part of this thesis has been carried out at the chair for Technische Chemie
und Petrolchemie at the Institut für Technische und Makromolekulare Chemie (ITMC) of the
RWTH Aachen University between October 2011 and March 2015 under the supervision of
Prof. Dr. Walter Leitner and Prof. Dr. Jürgen Klankermayer.
Parts of this thesis have already been published in:
Hydrogenation of Carbon Dioxide to Methanol by Using a Homogeneous Ruthenium-
Phosphine Catalyst
S. Wesselbaum, T. vom Stein, J. Klankermayer, W. Leitner, Angew. Chem. Int. Ed.
2012, 51, 7499-7502.
Hydrogenation of carbon dioxide using a homogeneous ruthenium-Triphos catalyst:
from mechanistic investigations to multiphase catalysis
S. Wesselbaum, V. Moha, M. Meuresch, S. Brosinski, K. M. Thenert, J. Kothe, T. vom
Stein, U. Englert, M. Hölscher, J. Klankermayer, W. Leitner, Chem. Sci. 2015, 6, 693-
704.
Selective Catalytic Synthesis using the Combination of Carbon Dioxide and Hydrogen -
Catalytic Chess at the Interface of Energy and Chemistry
J. Klankermayer, S. Wesselbaum, K. Beydoun, W. Leitner, DOI:
10.1002/anie.201507458 and 10.1002/ange.201507458.
-V-
Acknowledgements
First of all, I would like to thank Prof. Dr. Walter Leitner for the challenging and very
interesting topic, the many fruitful scientific discussions, and the outstanding working
conditions. I also thank Prof. Dr. Jürgen Klankermayer for his guidance, his constant scientific
input, and for the many things he taught me.
I thank Prof. Dr. Sonja Herres-Pawlis for reviewing this thesis.
I would like to thank Dr. Markus Hölscher, Dr. Verena Moha, and Jens Kothe for the
successful cooperation and for performing the computational chemistry of this thesis.
I want to thank Dr. Giancarlo Franciò for scientific discussions and for educating me in
teaching students.
I thank Daniel Geier for his help with the continuous-flow equipment and for programming
the process control system.
I am grateful for the help of many colleagues, especially of Dr. Thorsten vom Stein, Markus
Meuresch, Dominik Limper, and Marcus Suberg.
I thank my former lab colleagues Dr. César A. Urbina-Blanco, Thomas Hermanns, and Dr.
Marcel Picard for teaching me practical skills in the lab, and for the nice working atmosphere
and fun we had. I also thank Dr. César A. Urbina-Blanco for proofreading this thesis.
I would also like to thank my skillful students who contributed to this work: Bernhard
Barwinski, Julian Kleemann, Dominik Schauenburg, and Katharina Thenert.
I thank Sandra Brosinski for her help with NMR measurements, and for her skillful help in the
lab.
Many thanks go to the staff of the analytical departments: Hannelore Eschmann, Julia
Wurlitzer, Elke Biener, and Wolfgang Falter of the GC/MS department, as well as Ines
Bachmann-Remy of the NMR department.
I thank Prof. Dr. Ulrich Englert for X-ray analysis.
I would like to thank Ralf Thelen and his crew from the mechanical workshop, as well as
Stefan Aey and Thomas Müller from the electrical workshop for taking care of the high-
-VI-
pressure equipment. I would also like to thank Günter Wirtz, Henning Kayser, Marcello
Padaro, Laurent Weisgerber, and Jendrik Wülbern for their help with computer problems.
I want to thank the whole Leitner group for the cooperativeness and wonderful working
atmosphere.
I want to thank all the partners of the SusChemSys project for financial support and for
giving me a lot of opportunities to train my softskills. The project “Sustainable Chemical
Synthesis (SusChemSys)” is co-financed by the European Regional Development Fund (ERDF)
and the state of North Rhine-Westphalia, Germany, under the Operational Programme
“Regional Competitiveness and Employment” 2007-2013.
This work was supported in part by the Cluster of Excellence “Tailor-Made Fuels from
Biomass”, which is funded by the Excellence Initiative by the German Federal and State
Governments to promote science and research at German universities.
Generous allocation of computer time by the Computation and Communication Centre of
RWTH Aachen University is gratefully acknowledged.
-VII-
Abstract
This thesis deals with the development and detailed investigation of the very first
organometallic catalyst for the selective hydrogenation of CO2 to methanol.
Organometallic catalysts were believed to allow the conversion of CO2 to methanol only via
1) the intermediate formation of CO via reverse watergas shift reaction (RWGS), leading to
an unselective raction, or via 2) the intermediate formation of alkyl formates as stable
Scheme 2: Reactions of CO2 with different numbers of H2 equivalents.
In Figure 1 the number of H2 equivalents used for CO2 reduction is displayed on the ordinate.
With an increasing number of H2 equivalents (vertical up on the ordinate) the energy
contents of the produced molecules increase: Neat formic acid has an energy density of
INTRODUCTION
-5-
2.086 kWh L-1 or ca. 6.2 MJ kg-1,[36] methanol has an energy density of 22.7 MJ kg-1,[1] and
methane has an energy density of 55.5 MJ kg-1.[1] Therefore, this strategy can be used to
store energy in chemical bonds. In the aforementioned reactions to CO, formic acid,
formaldehyde, methanol, and methane the carbon centre is reduced without increasing the
molecular complexity of the resulting molecule. On the “chessboard” shown in Figure 1 the
resulting molecules are placed on the fields with the coordinates (0/y).
To cover the whole range of conceivable CO2 transformation reactions not only reduction of
the carbon centre, but also bond forming reactions have to be considered.[37] This was
achieved in Figure 1 by counting the total number of newly formed C-C and C-hetero bonds
(not only the bonds formed to the CO2 carbon!) on the abscissa of the “chessboard”,
enabling a quantification of the molecular complexity. A typical example for increasing the
molecular complexity of CO2 without concomitant reduction of the carbon centre is the
synthesis of urea. In this case, zero H2 equivalents are needed as no reduction of CO2 takes
place and two new C-N bonds are formed, placing urea on the field with the coordinates
(2/0) on the “chessboard”.
INTRODUCTION
-6-
Figure 1: Classification of CO2 hydrogenation products according to the H2 equivalents needed for CO2 reduction (ordinate) and number of newly formed C-hetero or C-C bonds (abscissa) in a “catalytic chess game”.
By combining the reduction of CO2 to the “formic acid/CO stage” (1 H2 eq.), the
“formaldehyde stage” (2 H2 eq.), the “methanol stage” (3 H2 eq.), or the “methane stage”
(4 H2 eq.) (vertical up on the “chessboard”) with the formation of new bonds (rightward on
the “chessboard”) the scope of compounds available from CO2 becomes much larger.[37] This
strategy can be used to “harvest” renewable energy for the chemical value chain. Some
examples of molecules becoming available by this strategy are placed on the chessboard
shown in Figure 1 according to their classification:
Alkyl formates are obtained by a combination of CO2 reduction with one H2 equivalent to the
“formic acid/CO stage” and the formation of a new C-O bond, placing it on the field with the
coordinates (1/1) on the chessboard.[7, 10, 38-40] Carboxylic acids produced by
hydrocarboxylation require reduction of the CO2 with one H2 equivalent to CO as
intermediate, and successive formation of a new C-C bond,[41] placing it again on field (1/1).
Hydroaminomethylation for the production of tertiary amines from olefin, secondary amine,
INTRODUCTION
-7-
CO2 and H2 again proceeds via CO2 reduction with one H2 equivalent to CO as
intermediate.[42-43] As one new C-C and one new C-N bond are formed in this reaction the
product is placed on field (2/1). The additional two H2 equivalents needed for the
hydroformylation-step und hydrogenation of the carbon double bond are not counted in this
classification.
No combination of CO2 reduction to the “formaldehyde stage” with bond forming reactions
is known today,[44-48] showing that there is still a wide field for future research.
Only recently, combinations of CO2 reduction to the “methanol stage” with bond forming
reactions have been reported: The direct N-methylation of primary and secondary amines
with CO2 and H2 was demonstrated by Klankermayer and Leitner et al. in 2013 and shortly
after by Beller et al.[49-51] In Figure 1 these products are placed on field (1/3) as three H2
equivalents are used for CO2 reduction to the “methanol stage” and one new C-N bond is
formed. Even more interestingly, N-methylated tertiary amines could be produced in a one-
pot reaction as follows:[50] In the first step of this cascade reaction, a primary amine reacts
with an aldehyde to give an imine. This imine is hydrogenated to the corresponding
secondary amine with one H2 equivalent in the second step, and the resulting secondary
amine is methylated with CO2 and three H2 equivalents in the third step. On the
“chessboard” the pharmaceutical ingredient butenafine, which could be produced by this
route, is placed on field (2/3) as three H2 equivalents are used for CO2 reduction and two
new C-N bonds are formed. The fourth H2 equivalent needed for the intermediate
hydrogenation of the imine to the secondary amine is not counted in this classification. A
combination of CO2 reduction to the “methane stage” with bond forming reactions is as yet
unknown.[52-55]
If one sticks to the chessboard picture, one could regard the catalysts needed to allow the
transformations as the chess pieces needed to move from one field to another (Figure 1).
1.1.4 Sources for H2
For the production of one metric tonne of methanol 1.38 t CO2 and 0.19 t H2 are needed.[3]
Today, the production of methanol is the third major hydrogen consumer (9 % of global
consumption) after ammonia production and crude oil refining.[1] In order to close the
INTRODUCTION
-8-
carbon cycle when using CO2 as resource, the production of hydrogen gas must not be
accompanied by CO2 production or use of fossil fuels.[1, 3] Therefore, regenerative energy
sources (like wind, solar, geothermal energy, hydroelectric power, tide power) should be
used for water splitting into H2 and O2 (Scheme 3).[1, 3] Several electrochemical,
photochemical, and thermochemical water splitting methods have been developed,
however, electrolysis is the only method which is operating on an industrial scale at the
moment (up to 4 % of the hydrogen production in 2009).[1, 56] Existing technologies for the
coupling of photovoltaic with electrolysis have solar-to-hydrogen conversion efficiencies in
the range 5-20 % (around 20 % efficiency for photovoltaic and 80 % for electrolysis).[9] An
area of approximately 10-40 km² is needed for the production of 1 t H2 per day.[9]
Scheme 3: Water splitting into hydrogen and oxygen.
The advantage of the electrochemical methods is that they can be used in conjunction with
all kinds of regenerative energy sources. The classical method is alkaline electrolysis (AEL) of
a 30 % KOH solution at 80-90 °C.[1] Cheap electrodes based on nickel can be used for this
process.[1] However, for electrolysis using strongly fluctuating energies like wind and solar
power, the startup and shutdown behaviour of the electrolysis cell is of utmost
importance.[1] Here, the AEL has some disadvantages compared to newer methods, such as
proton-exchange membrane electrolysis (PEMEL).[1] Using PEMEL a compact setup with
higher power efficiency can be achieved, which, however, has higher investment costs.[1]
Importantly, the startup and shutdown behaviour is better suitable for fluctuating power.[1]
Many researchers focus on high-temperature electrolysis (HTEL) at temperatures >800 °C in
solid-oxide electrolysis cells (SOECs), because at these temperatures the decomposition
voltage of water and the overvoltage at the electrodes is decreased.[1, 57] Setups have been
constructed which use sunlight to generate the necessary heat as well as electricity at the
same time.[58]
In regions with much sunlight also photochemical and thermochemical water-splitting
methods can be of interest. Photochemical water splitting is in principle an interesting
alternative to electrolysis, as sunlight could be used directly.[1, 58] However, problems such as
conversion rates below 1 % and the fact that hydrogen and oxygen are not produced
separately still have to be solved in the future.[1, 58]
INTRODUCTION
-9-
Promising technologies are being developed in the field of thermochemical water splitting.[1,
58] To allow the separation of hydrogen and oxygen generation and to lower the splitting
temperature (>2500 °C for direct water splitting) processes using supporting reagents such
as the hybrid sulphur cycle (HyS) have been developed:[1, 58] Hydrogen is generated by
electrolysis of an aqueous SO2 solution, thereby generating sulphuric acid. The sulphuric acid
is decomposed at 800-1000 °C to restore SO2. Though not being used commercially today,
these solar thermal water splitting technologies use solar energy much more efficiently
compared to a combination of photovoltaic and electrolysis, making this an interesting
option for the future.[1, 58]
1.2 Methanol – A basic chemical
1.2.1 Background: Properties and applications of methanol
An impressive amount of 53 million tonnes of methanol were consumed in 2011, making it
one of the most important bulk chemicals of the chemical industry.[1] The largest producer of
this colourless liquid (mp = -97.6 °C, bp = 64.6 °C) is Methanex.[59]
Some important applications of methanol are shown in Scheme 4.[1, 59-60] In 2011, Methanol
was mainly used to produce formaldehyde (32 %), dimethyl ether (DME, 11 %), methyl
tertiary-butyl ether (MTBE) and tertiary-amylmethylether (TAME) (together 10 %), and acetic
acid (10 %).[1] Other important products are olefins (MTO and MTP processes, 6 %),
methylamines (4 %), methyl methacrylate (2 %), and chloromethane (1 %). Already in 1986,
Friedrich Asinger discussed the wide range of possible applications based on methanol in his
book “Methanol – Chemie und Energierohstoff” (translation: “Methanol – Chemical and
energy resource”).[61] Due to the raw-material situation in China today, his ideas are more
up-to-date than ever: In 2013, methanol based on coal derived synthesis gas was used to
produce around 1 million tonnes of propylene via the “methanol-to-propylene” process
(MTP), and the demand for propylene is still rising.[1] In 2006, George A. Olah published a
book extending Asinger’s vision and ever since coined the phrase “methanol economy”.[62]
INTRODUCTION
-10-
Scheme 4: Some important applications of methanol.[1, 59-60]
In addition to the use for chemical production, methanol can also serve as an energy vector.
Its very high energy density of 22.7 MJ/kg makes it suitable for energy storage and for the
use as fuel.[1] This energy density is much higher compared to the energy density of Li-ion
batteries (0.5-3.6 MJ/kg).[1] Because of its physico-chemical properties MeOH can be easily
stored and transported using existing technologies.[1] In contrast to oil, methanol is water
mixable and biodegradable.[1] Methanol can be used as a fuel additive or pure in modified
engines and direct methanol fuel cells.[1] The materials used in the fuel system have to be
resistant towards methanol. Moreover, methanol can be converted to conventional fuels by
the “methanol-to-gasoline” process (MTG).[1, 59] In 2011, already 11 % of the produced
methanol were used in gasoline/fuel applications, and this sector is growing fast.[1] A very
detailed, exhaustive discussion of methanol utilisation technologies and a methanol based
economy can be found in excellent books.[1, 61-62]
Another potential use of methanol might be the safe storage and transportation of hydrogen
and CO.[1] The gravimetric storage capacity in methanol is 87.4 wt.-% for CO and 12.5 wt.-%
for H2. Methanol can be catalytically split or reformed to different CO, H2, and CO2 gas-
mixtures.[1] Thus, methanol can serve as a liquid form of synthesis gas in the stoichiometric
ratio required for many industrial applications.[1] Large amounts of H2 are generated from
INTRODUCTION
-11-
methanol reforming (Scheme 5 b, back reaction). This process is typically catalyzed by
copper/zinc catalysts at 180-300 °C.[1] In 2013, it was shown that aqueous methanol can be
reformed to CO2 and H2 at much lower pressures and temperatures using homogeneous
organometallic catalysts (for details see 1.3.3).[63-66] Compared to formic acid (4.3 wt.-%),
methanol has a higher gravimetric storage capacity for H2 of 12.5 wt.-%. Interestingly, the
equimolar methanol/water mixture produced by CO2 hydrogenation to methanol (Scheme
5 b) has a similarly high H2 storage capacity of 12.0 wt.-%. Therefore, the separation of
methanol and water would be unnecessary for hydrogen storage applications.
1.2.2 Methanol production from conventional carbon sources
Today, methanol is produced on large scale by conversion of fossil-fuel derived synthesis gas
(CO/H2) in the presence of heterogeneous catalysts (e.g. Cu/Zn/Al-oxide) at elevated
pressures (50-250 bar) and temperatures (200-350 °C) (Scheme 5, a).[1, 59] For hydrogen-rich
synthesis gas mixtures (e.g. from methane steam reforming), CO2 is added as it consumes
more H2 than CO does (Scheme 5, b). Both reactions are tied through the WGS reaction
which is also catalysed by the typical heterogeneous methanol catalysts under reaction
conditions (Scheme 5, c).[1]
Scheme 5: Production of Methanol from synthesis gas.[1]
1.3 Methanol production from CO2 and H2
As envisioned by Asinger and Olah, it might be possible to build up an economy based on
methanol.[1, 61-62] Consequently, if CO2 and H2 could be efficiently converted to methanol, an
economy based on CO2 and H2 could be imagined. In fact, the conversion of CO2 and H2 to
methanol by heterogeneous catalysts is known from the classical methanol production
processes, in which CO2 is added to the synthesis gas stream (chapter 1.2.2).[1]
INTRODUCTION
-12-
1.3.1 Heterogeneous catalysis for CO2 hydrogenation to methanol
Heterogeneous catalysts are known to catalyse the hydrogenation of CO and CO2 to
methanol, as well as the WGS reaction (Scheme 5).[1] Especially Cu/Zn-oxide based catalysts
were investigated in this respect.[67-70] It was found that the addition of small amounts (up to
3 %) of CO2 to the synthesis gas enhances the yield of methanol.[1, 71] However, high amounts
of CO2 or pure CO2 led to the formation of water as byproduct which reduced the rate of
methanol formation.[1, 67-68]
Nevertheless, Lurgi demonstrated the conversion of CO2 to methanol using a Cu/Zn/Al-
catalyst by Süd-Chemie (now Clariant) in a pilot plant in 1994.[1, 70, 72-73] At 60 bar pressure
and ca. 260 °C methanol was obtained with per-pass conversions around 35-45 %. A slight
catalyst deactivation was observed. The selectivity was as high as 99.96 % (excluding water).
Starting in 1996, NIRE and RITE of Japan built a pilot plant (50 kg/day) based on a new
Cu/ZnO/ZrO2/Al2O3/SiO2 catalyst, which showed slow deactivation over time.[74-76] Another
interesting approach is the CAMERE process of the Korean Institute of Science and
Technology.[1, 77-79] In this process a rWGS reactor is coupled to a methanol formation
reactor. In the rWGS reactor, CO2/H2 is partly converted to CO and H2O. After water removal,
the resulting CO/CO2/H2 stream is fed to the methanol reactor. The production capacity of
the pilot plant is 100 kg methanol per day. Another pilot plant has been operated since 2009
by Mitsui Chemicals with a capacity of 100 tonnes per year.[1, 80-82] In this process, the
Cu/ZnO/ZrO2/Al2O3/SiO2 catalyst developed in a joint research with RITE (vide supra) is used.
CRI operates a plant in Iceland that produces methanol from geogenic CO2 and hydrogen
produced by water electrolysis.[83-85] The process is powered by geothermal energy,
rendering the process economic. The “George Olah Renewable Methanol Plant” has been
operated since 2011 and has a production capacity of 5 million litres per year. In December
2014, CRI announced collaboration with industrial partners, universities, and research
institutions to implement its technology in Germany, with the goal to recycle carbon-dioxide
emissions from a coal-fired powerplant.[83]
In 2012, Behrens et al. elucidated a detailed picture of the elementary steps and the role of
the multi-component catalyst material Cu/ZnO/Al2O3 in the hydrogenation of CO and CO2 to
methanol.[69] The seemingly simple overall transformation of CO or CO2 and H2 to methanol
proceeds through a complex series of bond cleavage and bond forming processes on the
INTRODUCTION
-13-
catalyst surface involving the intermediates HCOO, HCO, HCOOH, H2COOH, H2CO, and CH3O.
The active site consists of Cu steps with Zn substituted into the Cu steps. For a typical syngas
mixture (59.5 % H2, 8 % CO2, 6 % CO, rest inert) at 60 bar and 210 °C-250 °C a TOF of 75.6 h-1
(mol methanol/mol Cu sites) was calculated.[69] This was the benchmark activity at the time
the research for this thesis was started. There is strong evidence that CO2 is directly
converted on Cu/ZnO catalysts rather than being transformed to CO first.[69, 86]
While this thesis was in preparation, some advances in the field of heterogeneous catalysts
were made: In 2014, Graciani et al. demonstrated that the metal-oxide interface in
Cu/CeOx/TiO2 is highly active for CO2 conversion to methanol.[87] At 300 °C, they estimated
the TOF to be as high as 29160 h-1. However, one has to be careful when comparing this TOF
value for Cu/CeOx/TiO2 with the one reported by Behrens et al. for Cu/ZnO/Al2O3, as they
were calculated based on different models.[86] For the Cu/CeOx/TiO2 catalysts Graciani et al.
proposed a mechanism via consecutive rWGS and hydrogenation of CO to methanol.[87]
1.3.2 Homogeneous catalysis for CO2 hydrogenation to methanol
In contrast to heterogeneous catalysts, much less reports exist about homogeneous catalysts
for the CO2 hydrogenation to methanol. This seems to be somewhat surprising, given that
homogeneous organometallic catalysts have been known to activate CO2 for its
hydrogenation to formic acid since Inoue’s discovery in 1976.[88] In 2007, Philip G. Jessop
speculated that the reduction of CO2 beyond the formic acid level typically requires much
higher temperatures, and that only few catalysts are both kinetically capable and stable at
these reaction conditions.[7]
The first reports of methanol formation from CO2 and H2 in the presence of organometallic
catalysts stem from Tominaga et al. from the NIRE and RITE institutes in Japan.[53, 55] In 1993,
they reported the hydrogenation of CO2 to methane via successive formation of CO and
methanol as intermediates in the presence of a Ru3(CO)12/KI catalytic system under harsh
reaction conditions (240 °C, 90-140 bar).[53, 55] The homogeneous catalytic system was active
in the rWGS reaction, converting CO2 and H2 to CO and H2O, and in the successive
hydrogenation of CO to methanol. Methane formation was found to be mainly catalysed by
deposited ruthenium metal. Consequently, the selectivity could be shifted towards methanol
by addition of KI which prevented deposition of metallic ruthenium.[53, 55] In a typical
INTRODUCTION
-14-
experiment in the presence of KI, methanol was obtained with a TON (based on the number
of ruthenium atoms) of up to 32, besides CO with a TON of 11, methane with a TON of 8, and
traces of ethanol.[53, 55] Interestingly, Tominaga et al. showed that by using a similar
Ru3(CO)12/KI catalytic system in the presence of Co2(CO)8 as cocatalyst ethanol was produced
by homologation of methanol.[89-90] At 200 °C, CO2 could be converted to CO (TON = 8),
However, for almost another 20 years, no organometallic catalyst was found for the
selective hydrogenation of CO2 to methanol. It seemed that the key to selective methanol
formation at milder reaction conditions was to find a catalyst which catalyses the direct
hydrogenation of CO2 to methanol without catalysing the rWGS leading to the formation of
CO.
In 2010, the catalytic reduction of CO2 with stoichiometric amounts of boranes instead of
molecular H2 has been achieved using nickel-pincer complexes.[91-92] Metal-free catalytic
systems employing frustrated lewis pairs (FLP)[93-99] N-heterocyclic carbenes (NHC),[100] and
silyl-cations[101] have been shown to be active for the CO2 reduction to the methanol stage.
However, these systems are as yet limited by the use of stoichiometric amounts of boranes
and silanes as reduction agents, by the necessity to hydrolyse the formed intermediates with
H2O and/or NaOH to release the methanol product, and by the destruction of the FLP
systems upon the hydrolysis step. Metal-free catalytic systems for the reduction of CO2 were
discussed in detail in two comprehensive reviews.[102-103]
Due to the lack of organometallic catalysts being capable of transforming CO2 to methanol,
indirect routes from CO2 to methanol via CO2 derived intermediates were proposed by
Milsteins’s group in 2011.[104-106] Milstein et al. developed Ru-PNN pincer complexes
(Scheme 8, C) for the efficient hydrogenation of methyl formate,[104] dimethyl carbonate,[104]
methyl carbamates,[104] urea derivatives,[107] and formamides[108] to methanol. As those
substrates can be produced from CO2, two-step production processes were envisioned for
the indirect production of methanol from CO2 by organometallic catalysis (Scheme 6). Using
the Ru-PNN complex, dimethyl carbonate could be quantitatively hydrogenated to methanol
with a TOF up to 2500 h-1 (60 bar H2, 145 °C).[104] Methyl formate could be hydrogenated
using the same complex with a TOF up to 531 h-1 (50 bar H2, 110 °C).[104] Various alkyl and
INTRODUCTION
-15-
aryl urea derivatives, such as 1,3-dihexylurea, could be hydrogenated at 110 °C and 13.6 bar
H2 to yield methanol in 46-94 %.[107] Formylmorpholine hydrogenation at a H2 pressure of
10 bar and a temperature of 110 °C resulted in the formation of methanol in 97 % yield.[108]
Scheme 6: Indirect routes for methanol production from CO2.[104-108]
Ding’s group reported an interesting method for indirect methanol production as coproduct
by modification of the Shell omega process (Scheme 7).[109] In the omega process, ethylene
glycol (EG) is produced by hydrolysis of ethylene carbonate, which is produced from
ethylene oxide and CO2 in the first step of the reaction. Ding et al. proposed to replace the
ethylene carbonate hydrolysis by the ethylene carbonate hydrogenation to ethylene glycol
and methanol.[109] Using the Ru-PNP catalyst shown in Scheme 7, a TON of up to 87000 and a
TOF of up to 1200 h-1 could be obtained.[109]
Scheme 7: Methanol production by a modified omega process as proposed by Ding et al.[104, 109]
Milstein’s Ru-PNN complexes were not able to catalyse the hydrogenation of CO2 directly,
i.e. they could only catalyse the hydrogenation of the CO2 derived intermediates, making
INTRODUCTION
-16-
two-step procedures necessary to produce methanol. In 2011, Huff and Sanford tackled this
problem by setting up an integrated one-pot cascade reaction:[110] They used three different
catalysts (Scheme 8, A, B, C) in one reaction mixture to catalyse the three steps of the
cascade shown in Scheme 9, which were (a) hydrogenation of CO2 to formic acid, (b)
esterification of formic acid to methyl formate, and (c) hydrogenation of methyl formate to
methanol.[110] The combination of steps (a) and (b) of the cascade reaction had been known
to result in the formation of methyl formate.[7] In most cases, this reaction had been carried
out in the presence of a base (e.g. NEt3).[7] However, Huff and Sanford found that by adding
the Lewis acid Sc(OTf)3 (B) as co-catalyst to [RuCl(OAc)(PMe3)4] (A) in the absence of base,
the esterification step was significantly enhanced, giving a TON of 40 after 16 hours and a
TOF of 32 h-1 in the first hour.[110] The last step of the cascade, the hydrogenation of methyl
formate to methanol (c), had little precedent in the literature.[104, 110] Sanford and Huff used
the Ru-PNN complex (C) developed by Milstein et al.[104] to accomplish this reaction step
(Scheme 8).[110] Using a mixture of these three catalysts A, B, and C in CD3OH at 135 °C under
a CO2 pressure of 10 bar and a H2 pressure of 30 bar, a TON of 2.5 was achieved.[110] The
main factors hampering higher TONs were the inhibition/deactivation of the Ru-PNN catalyst
(C) by CO2 and by Sc(OTf)3 (B).[110] By spatial separation of catalysts A/B from catalyst C
inside one reactor a higher TON of up to 21 could be obtained.[110]
Scheme 8: Catalysts used for indirect hydrogenation of CO2 to methanol.[104, 110]
Scheme 9: Indirect hydrogenation of CO2 to methanol via methyl formate (lower pathway) as shown by Sanford/Huff.[110]
Transferred to the “catalytic chess” model introduced in chapter 1.1.3, Huff and Sanford
needed a combination of pawn, rook, and knight to “walk” from CO2 to methanol (Figure 2,
INTRODUCTION
-17-
blue pathway). Based on this eye-opening strategy the aim of the present work was to find
the “organometallic queen” which is able to walk directly from CO2 to methanol (green
pathway).
Figure 2: In a “catalytic chess” game, Huff and Sanford need three different chess pieces to walk from CO2 to MeOH.[104, 110]
The aim of the present work was to find the “organometallic queen” which can walk directly from CO2 to MeOH.
Only very recently, while this thesis was in the writing process, Huff/Sanford et al. described
a cascade reaction process for the hydrogenation of CO2 to methanol.[111] In contrast to their
previous report, methanol formation via intermediate formation of dimethylformamide
(DMF) was envisaged this time (Scheme 10).[111] The hydrogenation of CO2 in the presence of
HNMe2 has been known to lead to the formation of DMF with very high activities and
selectivities (steps (a) and (b) of the cascade shown in Scheme 10).[7, 10] Step (c) of the
cascade reaction, the hydrogenation of amides to methanol, had much less precedent in the
literature.[108, 112] Huff/Sanford et al. showed that the Ru-MACHO-BH4 complex shown in
Scheme 10 together with 50 equivalents of the base K3PO4 is capable of selectively
hydrogenating DMF to methanol at 50 bar H2 pressure and 155 °C. As catalyst deactivation
became apparent at 155 °C, a temperature ramp as well as a huge excess of H2 was used to
achieve high CO2 conversions in the overall cascade reaction to methanol of up to 96 %: CO2
hydrogenation to DMF was first carried out at 95 °C, and after 18 hours the temperature was
raised to 155 °C to achieve hydrogenation of the DMF intermediate to methanol. A TON of
INTRODUCTION
-18-
up to 550 could be obtained using this strategy. However, the selectivity to methanol was
only about 30 % with DMF and dimethylammonium formate (DMFA) as coproducts. Using
the same temperature ramp and the same catalytic system also dimethylammonium
dimethylcarbamate (DMC), which forms upon reaction of HNMe2 and CO2, could be used as
substrate instead of CO2. In this case, DMC initially reacts to CO2 and HNMe2 at 95 °C before
the cascade proceeds through DMF formation as described above. Based on this example,
combined capture of CO2 in the form of compounds like DMC and conversion to methanol
was proposed.
Scheme 10: Indirect hydrogenation of CO2 to methanol DMF (lower pathway) as shown by Sanford/Huff.[111]
An interesting method for combining low pressure CO2 capture with subsequent
hydrogenation to methanol was published by Milstein’s group shortly after (Scheme 11).[113]
In this approach CO2 is captured by an aminoalcohol at low pressures (1-3 bar) at 150 °C in
the presence of Cs2CO3 as catalyst, which leads to the formation of the corresponding
oxazolidone as intermediate.[113-114] The resulting oxazolidone solution could be
hydrogenated at 135 °C and 60 bar H2 to give methanol and to restore the aminoalcohol
after addition of the Ru-PNN pincer complex shown in Scheme 11 and tert-BuOK (10 mol-%).
Excess CO2 had to be removed after the CO2 capture step as the Ru-PNN pincer complex was
deactivated by the presence of CO2. Using valinol in DMSO as capture medium, CO2 could be
converted to methanol in about 50 % yield by this procedure. The allure of these combined
CO2 capture/conversion concepts is that energy costs associated with CO2 release from
capture solutions could be avoided.
INTRODUCTION
-19-
Scheme 11: CO2 capture and subsequent hydrogenation to MeOH as demonstrated by Milstein et al.[113]
Two years after first results of this thesis concerning the CO2 hydrogenation to methanol
using the catalytic system Ru/Triphos/acid (Triphos = 1,1,1-
tris(diphenylphosphinomethyl)ethane) had been published,[115] Cantat et al. showed that the
same catalytic system Ru/Triphos/acid could also be used for the disproportionation of
formic acid to methanol, CO2, and water.[116] This reaction was shown the first time by
Goldberg et al. in 2013.[117] Using [Ru(COD)(methylallyl)2]/Triphos/methylsulfonic acid at
150 °C, formic acid could be decomposed leading to the formation of methanol with a yield
of up to 50 % (TON = 83).[116] Cantat et al. found that this catalytic mixture catalysed the
decomposition of formic acid to CO2 and H2, as well as the transfer hydrogenation of formic
acid to methanol.[116] Together with the findings described in the excerpt from the present
thesis which had already been published at this time[115] a network of reactions leading to
methanol formation could be established (Scheme 12).[116]
Scheme 12: Proposed pathways for the disproportionation of formic acid to methanol according to [116]
.
The production of methanol by disproportionation of formic acid is an interesting variation,
however, as pure formic acid itself is a valuable chemical which cannot easily be prepared
from CO2 in pure form[7] the direct hydrogenation of CO2 to methanol still appears to be
much more promising for the production of methanol from alternative carbon sources.
INTRODUCTION
-20-
1.3.3 Homogeneous catalysis for methanol reforming
The back-reaction of the CO2 hydrogenation to methanol, the catalytic dehydrogenation of
aqueous methanol to CO2 and H2 (methanol reforming), was the first time reported using
homogeneous organometallic catalysts by Cole-Hamilton et al. in 1987.[118] They used
[Rh(bipy)2]Cl as catalyst in the presence of NaOH at 120 °C to decompose a 95/5 v/v solution
of methanol/H2O at a TOF of 7 h-1.
In 2013, while this thesis was in preparation, more efficient dehydrogenation of aqueous
methanol was reported by Grützmacher’s and Beller’s groups.[63-65, 119] Beller’s group showed
that with the catalyst [RuHCl(CO)PNP] (PNP = HN(CH2CH2PiPr2)2) methanol could be
decomposed to CO2 and three equivalents of H2 with a TOF (based on moles of methanol) of
up to 1573 h-1 in the presence of potassium hydroxide at 95 °C.[63] A 3/2 mixture of methanol
and water gave a TOF of up to 244 h-1 and a 9/1 MeOH/H2O mixture gave a TOF of 88 h-1.[63]
In a long-term stability test a 9/1 mixture of MeOH/H2O was decomposed over a period of
23 days giving a TON of >116667.[63]
Grützmacher’s group used [K(dme)2][Ru(H)(trop2dad)] (trop2dad = 1,4-bis(5H-
dibenzo[a,d]cyclohepten-5-yl)-1,4-diazabuta-1,3-diene) for the decomposition of a 1/1.3
MeOH/D2O mixture at 90 °C in the presence of THF solvent under base free conditions.[64]
After 10 hours, 78 % of the methanol was decomposed to CO2 and H2 (TON = 156). During
the catalytic cycle, the non-innocent azadiene ligands reversibly store molecular hydrogen.
Shortly after, Beller’s group reported the use of the iron pincer complex [FeH(BH4)(CO)(PNP)]
(PNP = HN(CH2CH2PiPr2)2) for methanol reforming.[119] Pure methanol was decomposed to
CO2 and H2 in the presence of KOH with a TOF of up to 245 h-1 at 91 °C. A mixture of 4/1
MeOH/H2O was decomposed with a TOF of 137 h-1 under the same conditions.
Beller et al. reported in 2014 that by using a mixture of Ru-MACHO-BH4 (shown in Scheme
10) and [Ru(H)2(dppe)2] (dppe = 1,2-bis(diphenylphosphino)ethane) aqueous methanol could
be dehydrogenated in the absence of base.[120] The catalysts were reported to operate in a
synergistic manner, i. e. their combined activity was higher than the sum of the activities of
each single catalyst. At 93.5 °C an average TOF of 93 h-1 was obtained over 7 hours and a
total TON of 4286 was obtained after 8 days.
INTRODUCTION
-21-
In 2014, Milstein’s group used the Ru-PNN pincer complex which was employed before for
the hydrogenation of alkyl formates to methanol (Scheme 8, C) for methanol reforming to
CO2 and H2 in the presence of NaOH or KOH and toluene solvent.[65] The released CO2 was
trapped as carbonate. Methanol was converted with 77 % within 9 days with an average TOF
of 14.3 h-1 based on methanol. Interestingly, the organic layer could be separated from the
aqueous layer and reused without addition of new catalyst. The procedure could be
repeated over a period of 1 month giving a TON of around 9667 based on methanol.
Interestingly, all reported catalytic systems active in the dehydrogenation of aqueous
methanol solutions possess multidentate ligands. None of these catalysts was, however,
reported to be active for the hydrogenation of CO2 to methanol.
AIM OF THE THESIS
-22-
2 Aim of the thesis
When the experimental work for this thesis was started, organometallic catalysts were
believed to allow the hydrogenation of CO2 to methanol only via formate esters as stable
intermediates,[104, 110] or via the intermediate formation of CO by rWGS (vide supra).[53, 55]
Whereas the cascade reaction via formate esters suffered from its complexity and partial
incompatibility of the three different catalysts used leading to limited turnover numbers, the
approach via rWGS suffered from its bad selectivity, leading to the formation of CO and
methane as side products.
The aim of the present thesis was the development of the very first catalytic system based
on a single organometallic complex for the selective hydrogenation of CO2 to methanol.
After identification of a suitable catalytic system, the mechanism of this system should be
investigated in detail. Reaction schemes should be developed and realised which allow for
easy recycling of the homogeneous catalyst in repetitive batch as well as in continuous-flow
mode.
RESULTS & DISCUSSION
-23-
3 Results & Discussion
3.1 Hydrogenation of CO2 to MeOH in the presence of alcohol additives
Parts of this chapter have been published in: S. Wesselbaum, T. vom Stein, J. Klankermayer,
W. Leitner, Angew. Chem. Int. Ed. 2012, 51, 7499-7502.[115]
3.1.1 Identifying a suitable catalytic system
As a starting point, a catalytic system should be identified which could integrate all three
reaction steps of the cascade reaction established by Huff and Sanford for the CO2
hydrogenation to methanol.[110] This reaction cascade consisted of three consecutive
reactions:[110] (a) the hydrogenation of CO2 to formic acid, (b) the esterification of formic acid
with methanol to methyl formate, and (c) the hydrogenation of methyl formate to methanol
(vide supra).[110] The hydrogenation of CO2 in the presence of methanol or ethanol using
organometallic catalysts was well known to lead to the formation of the corresponding alkyl
formates (combination of steps (a) and (b) of the cascade).[7] The reaction pathway is
believed to be CO2 hydrogenation to formic acid followed by thermal esterification of formic
acid with the present alcohol.[7] Several organometallic catalysts were known to catalyse this
reaction, with ruthenium-phosphine complexes being the most active ones.[7] Therefore, a
catalytic system based on a ruthenium phosphine complex seemed to be the obvious choice
to integrate all three reaction steps.
However, the last step of the cascade, the hydrogenation of alkyl formates to methanol (c),
had little precedent in the literature.[104, 110] Milstein et al. developed Ru-PNN pincer
complexes which made the efficient and selective hydrogenation of methyl formate to
methanol possible for the first time.[104] Due to a lack of alternatives, Huff and Sanford used
this Ru-PNN complex for step (c) of their cascade reaction, despite its deactivation by CO2
which led to limited turnover numbers (TON).[110]
A catalytic system that had been investigated already earlier by Klankermayer and Leitner et
al. and other groups was identified as promising candidate for the integration of steps (a),
RESULTS & DISCUSSION
-24-
(b), and (c) of the cascade.[121-127] This catalytic system was based on ruthenium as the
central metal and the tridentate ligand Triphos (Triphos = 1,1,1-
tris(diphenylphosphinomethyl)ethane). The active catalyst could be formed either in situ
from Ru(acac)3 and Triphos 1 or from the readily accessible ruthenium(II)-complex
[Ru(TMM)(Triphos)] (TMM = trimethylenemethane) 2 (Scheme 13),[128] and provided an
excellent catalyst for the hydrogenation of carboxylic esters, amides, lactones, and
carboxylic acids.[121-127] While this thesis was in progress, the scope of this catalytic system
was extended to the hydrogenation of carbonates, acid anhydrides, imides and ureas by
Klankermayer and Leitner et al.[129] Distinct reactivities of this catalytic system were
observed dependent on whether an acidic additive was present or not.[126-127, 129]
Mechanistic investigations by Klankermayer and Leitner et al. suggested that species of type
3, in which the Triphos ligand is coordinated in a facial fashion, facilitate hydride transfer and
protonolysis as key steps for the addition of hydrogen to the carboxylate functional group in
the presence of acidic additives (Scheme 13).[127]
Therefore, the Ru/Triphos catalytic system showed two characteristics which made it a very
promising candidate as catalyst for the whole cascade reaction: it was based on a Ru-
phosphine complex, which might favour CO2 activation by facile insertion into the Ru-H
hydride bond,[7, 130] and it proved to facilitate the hydrogenation of carboxylate groups.
Scheme 13: In situ catalytic system 1, preformed catalyst precursor 2, and active species 3 (S=solvent or substrate) for the hydrogenation of carboxylate functional groups.
RESULTS & DISCUSSION
-25-
However, the hydrogenation of formate esters, step (c) of the cascade reaction, had not yet
been demonstrated using this catalytic system. In a first set of experiments, both catalytic
systems 1 and 2 were tested in the hydrogenation of formate esters to methanol (Table 1).
Table 1: Catalytic hydrogenation of formate esters.[a]
Entry Cat. Acid
R pH2[b]
[bar]
TON[c]
1 1[d]
MSA Et 50 75
2 1[d]
MSA Me 50 74
3 1[d]
‒ Et 30 72
4 1[d]
MSA Et ‒ 0[e]
5 ‒ MSA Et 30 0[e]
6 2 ‒ Et 30 5
7 2 MSA Et 30 77
8[f]
1[d]
MSA Et 30 79
[a] Reaction conditions: Ru-complex (25 µmol), substrate (2.5 mmol), 2.0 mL THF, 140 °C, 24 h; [b] at room temperature; [c]
TON = mol MeOH/mol catalyst; [d] 50 µmol (2 eq.) Triphos; [e] no methanol but traces of ethanol (about 4 % yield) were
observed; [f] 12.5 mmol substrate.
Indeed, the hydrogenation of ethyl formate was possible at 140 °C and 50 bar H2 pressure
using the catalytic system 1 together with 1.5 eq. (equivalents) of methylsulfonic acid (MSA)
as acidic additive. Methanol was yielded in 75 % corresponding to a TON of 75 (Table 1,
entry 1). In an analogous manner, methyl formate could be hydrogenated giving only
methanol as product with a TON of 74 (Table 1, entry 2). Interestingly, omitting the acidic
additive had no significant influence on the reactivity in the case of catalytic system 1,
whereas it had a strong detrimental influence in the case of catalyst 2 (Table 1, entries 3 &
6). In the presence of 1.5 eq. MSA also catalyst 2 enabled the hydrogenation of ethyl
formate to methanol with a TON of 77 (Table 1, entry 7). The positive influence of the acid in
the case of catalyst 2 suggested that structures of type 3 which form in presence of acid play
an important role in this catalytic transformation.[127] The effect of the omitted acidic
additive might not be observable in the case of catalytic system 1, as the reaction solution
most probably contains the weak acid acetylacetone from hydrogenolysis of the
acetylacetonate ligands under H2 pressure.[131]
Control experiments in the absence of metal-precursor or hydrogen did not show the
formation of methanol (Table 1, entries 4 & 5). However, in both cases small amounts of
ethanol (around 4 % yield) were found in the reaction solution, which might stem from slow
RESULTS & DISCUSSION
-26-
decarbonylation of ethyl formate to CO and ethanol under reaction conditions, or from ethyl
formate hydrolysis.
Using higher concentrations of the substrate ethyl formate did not lead to increased TONs,
indicating that the catalyst was deactivated after around 80 turnovers (Table 1, entry 8).
Already after 6 hours reaction time the 31P{1H}-NMR spectrum of the reaction solution in d8-
THF showed exclusively a characteristic set of a doublet (δ = 18.2 ppm, JP-P = 28.7 Hz) and
triplet (δ = 6.4 ppm, JP-P = 28.7 Hz) (Figure 3). The 1H-NMR spectrum showed a characteristic
hydride signal (δ = -6.8 ppm, dt, JH-P = 63.9 Hz, JH-P = 15.3 Hz). Based on literature data, these
signals were ascribed to the cationic carbonyl complex [Ru(H)(CO)2(Triphos)]+ (4).[132] The
carbonyl ligands were most likely formed by decarbonylation of alkyl formate.[127, 133] Based
on this observation, a new synthesis route for [Ru(H)(CO)2(Triphos)]NTf2 (4NTf2) could be
established (Scheme 14): Stirring complex 2 together with 1 eq. of
bis(trifluoromethane)sulfonimide (HNTf2) in ethyl formate under 60 bar H2 pressure for 24
hours at 140 °C led to the exclusive formation of 4NTf2 which could be isolated in 97 % yield
by simply removing all volatiles in vacuo.
The results summarised in Table 1 proved that catalytic systems 1 and 2 were indeed able to
catalyse the hydrogenation of alkyl formates to methanol in the presence of an acidic
additive. This fact made these catalytic systems suitable candidates to be tested in the CO2
hydrogenation reaction.
RESULTS & DISCUSSION
-27-
Figure 3: 31
P{1H}-NMR spectrum (121 MHz, d8-THF, r.t.) of the reaction solution of an ethyl formate hydrogenation reaction
using catalyst 2 after 6 hours at 60 bar H2, 140 °C.
Scheme 14: Synthesis of the cationic carbonyl complex 4NTf2 by heating a solution of 2 in ethyl formate under H2 pressure.
3.1.2 Reaction cascade for the hydrogenation of CO2 to methanol
Encouraged by these results, the hydrogenation of CO2 to methanol with catalytic systems 1
and 2 was tested. Ethanol was added to the reaction solution in order to allow the formation
of ethyl formate as intermediate in the cascade reaction and to allow direct identification of
any methanol formed in solution by NMR and GC.
First, catalytic system 1 with 2 eq. Triphos was tested in ethanol/THF. After 24 hours at
140 °C, 10 bar CO2 and 30 bar H2 only very low quantities of methanol corresponding to a
RESULTS & DISCUSSION
-28-
TON of 2 were detected (Table 2, entry 1). However, in the presence of MSA (1.5 eq.) as
acidic additive, methanol was formed with a TON of 52, which was already the best TON
ever obtained using an organometallic complex in this transformation (Table 2, entry 2). An
even slightly higher TON of 63 could be achieved with complex 2. Like catalytic system 1,
also 2 displayed a much higher productivity in the presence of 1.5 eq. of MSA compared to
the absence of acid (Table 2, entries 3 & 4).
Table 2: Hydrogenation of carbon dioxide to methanol in the presence of alcohol additive.[a]
sulfonic acid; [c] at room temperature; [d] TON = mmol MeOH/mmol catalyst; [e] 50 µmol (2 eq.) Triphos.
GC-analysis of the gas phase showed no formation of CO, indicating that the reaction did not
proceed via rWGS.[53, 55] In the 1H-NMR spectrum of the reaction solution small amounts of
ethyl formate and traces of methyl formate could be detected (Figure 4). The methyl
formate traces could only be detected in the 1H-NMR spectrum by extensive zooming into
the formate region and were too small to be quantified. This supported the assumption that
methanol formation proceeded at least partly via a cascade reaction and alkyl formate
intermediates. GC-analysis of the liquid phase corroborated the formation of methanol and
ethyl formate (Figure 5). However, at this stage a stepwise reduction of CO2 to methanol via
the formate anion in the coordination sphere of the metal could not be ruled out. The
1H-NMR spectrum in Figure 4 also showed a small singlet at 8.8 ppm, which could stem from
traces of free formic acid as well as from formate coordinated to the ruthenium metal
centre.[134-136]
RESULTS & DISCUSSION
-29-
Figure 4: Representative 1H-NMR spectrum (300 MHz, d6-dmso, r.t.) of the reaction solution of a CO2 hydrogenation
reaction using catalytic system 1 in the presence of MSA (1.5 eq.) in ethanol/THF; mesitylene was added as internal
standard.
Figure 5: Representative gas chromatogram of the reaction solution of a CO2 hydrogenation reaction using catalytic system 1 in the presence of MSA (1.5 eq.) in ethanol/THF; heptane was added as internal standard.
RESULTS & DISCUSSION
-30-
Control experiments were performed to confirm the origin of the observed methanol from
homogeneously catalysed CO2 hydrogenation. As expected, in the presence of acid alone no
CO2 reduction products were formed (Table 2, entry 5). Also the reaction in the absence of
either CO2 or H2 did not lead to methanol formation (Table 2, entries 6 & 7). Reaction
solutions containing metal and ligand were yellow and clear after performing the reactions.
No metal deposition was visible to the naked eye. An experiment with ruthenium supported
on carbon (5 wt.-% Ru/C) as catalyst led to no detectable formation of methanol
(0.025 mmol Ru, 1.0 eq. HNTf2, 10 mmol EtOH, 1.5 mL THF, 10 bar CO2, 30 bar H2, 140 °C,
24 h). The molecular nature of the catalytically active metal species was further supported
by a mercury poisoning experiment. A CO2 hydrogenation reaction was performed
(0.025 mmol catalyst 2, 1.0 eq. HNTf2, 10 mmol EtOH, 1.5 mL THF, 20 bar CO2, 60 bar H2,
140 °C, 24 h) in the presence of a large excess of mercury (0.1 mL, 6.75 mmol).[137-138] The
amount of methanol determined in this experiment was about 60 % of the amount formed
in an experiment using the same reaction conditions in the absence of mercury. This
difference can be explained by losses of the reaction solution during the separation of the
reaction solution from the added mercury before analysis. The fact that mercury did not lead
to a shutdown of the catalysis supported the assumption of a molecular catalyst species.
Using the so far most productive complex 2, a first screening of reaction parameters was
carried out to identify key parameters for future optimisation (Table 3). By changing the acid
from MSA comprising a coordinating anion[125] to HNTf2 comprising a weakly coordinating
anion the TON could be increased from 63 (Table 2, entry 4) to 86 (Table 3, entry 1). Raising
the CO2 pressure to 20 bar and the H2 pressure to 60 bar further increased the TON to 221
(Table 3, entry 2). Variation of the amount of HNTf2 showed the best result for the 1 : 1
stoichiometric ratio (Table 3, entries 2-4). Interestingly, in the case of the sulfonic acids MSA
and p-toluenesulfonic acid monohydrate (p-TsOH) the highest TONs were observed in the
presence of a slight excess (1.5 eq.) of the acid (Table 3, entries 5-10). However, both acids
led to lower TONs as compared to HNTf2.
For a more detailed analysis of the influence of the acid the reader is referred to chapter
3.3.1, and for a more detailed analysis of other parameters like temperature and partial
pressures the reader is referred to chapter 3.3.2. The next chapter will focus on the
identification of the active catalyst species and the understanding of the reaction
mechanism as basis for further investigations concerning reaction parameters.
RESULTS & DISCUSSION
-31-
Table 3: Variation of some reaction parameters in the hydrogenation of carbon dioxide to methanol in the presence of alcohol additive.
[a]
Entry Cat. Acid (eq.)[b]
pH2[c]
[bar]
pCO2[c]
[bar]
TON[d]
1
2 HNTf2 (1.0) 30 10 86
2 2
HNTf2 (1.0) 60 20 221
3 2
HNTf2 (1.5) 60 20 201
4 2
HNTf2 (3.0) 60 20 186
5 2
MSA (1.0) 60 20 100
6 2
MSA (1.5) 60 20 149
7 2
MSA (3.0) 60 20 39
8 2
p-TsOH (1.0) 60 20 128
9 2
p-TsOH (1.5) 60 20 193
10 2
p-TsOH (3.0) 60 20 72
[a] Reaction conditions: catalyst (25 µmol), THF (1.5 mL), EtOH (10 mmol), 140 °C, 24 h; [b] equivalents to catalyst; [c] at
room temperature; [d] TON = mmol MeOH/mmol catalyst.
3.2 Investigating the mechanism of the CO2 hydrogenation to MeOH
Parts of this chapter have been published: S. Wesselbaum, V. Moha, M. Meuresch, S.
Brosinski, K. M. Thenert, J. Kothe, T. vom Stein, U. Englert, M. Hölscher, J. Klankermayer, W.
Leitner, Chem. Sci. 2015, 6, 693-704.[139]
The results shown in the previous chapter demonstrated the possibility to hydrogenate
carbon dioxide to methanol using a single homogeneous transition-metal catalyst under
relatively mild reaction conditions. This chapter focuses on the identification of the active
catalyst species and the detailed reaction mechanism.
3.2.1 Identification of the organometallic species in solution
NMR experiments were carried out in order to identify organometallic intermediates of the
catalytic reaction sequence with catalyst precursor 2 upon stepwise addition of the required
components. Therefore, an experiment was carried out in the absence of an alcohol
additive: A solution of catalyst precursor 2 and HNTf2 (1 eq.) in d8-THF was pressurised with
CO2 (20 bar at r.t.) and H2 (60 bar at r.t.), stirred for 1 h at 140 °C, and the resulting clear and
yellow reaction solution was analysed by NMR spectroscopy. The 1H-NMR spectrum showed
a sharp signal at 3.27 ppm which indicated the catalytic formation of MeOH with a TON of
35. This was surprising, as the reaction was carried out in the absence of any alcohol
additive. Thus, one of the species formed under these conditions must have been able to
RESULTS & DISCUSSION
-32-
serve as catalyst for the hydrogenation of CO2 to methanol without the intermediate
formation of the formate ester. The 31P{1H}-NMR spectra of the reaction solution recorded
at room temperature and at -40 °C are depicted in Figure 6. The [1H,31P]-HMBC-NMR
spectrum recorded at -40 °C is shown in Figure 7.
Figure 6: 31
P{1H}-NMR spectra (top: at r.t., bottom: at -40 °C) of the reaction solution of a CO2 hydrogenation to methanol
(20 bar CO2, 60 bar H2, 140 °C, 1 h) with catalyst 2 (50 μmol) and HNTf2 (1 eq.) in d8-THF (2.0 mL).
RESULTS & DISCUSSION
-33-
Figure 7: [1H,
31P]-HMBC (600 MHz, d8-THF, -40 °C) spectrum of the reaction solution after CO2 hydrogenation to methanol
(20 bar CO2, 60 bar H2, 140 °C, 1 h) with catalyst 2 (50 μmol) and HNTf2 (1 eq.) in d8-THF (2.0 mL).
The 31P{1H}-NMR spectrum showed a characteristic set of a doublet (δ = 18.2 ppm, JP-P =
28.7 Hz) and a triplet (δ = 6.4 ppm, JP-P = 28.7 Hz). This signal was correlated with the hydride
signal (δ = -6.7 ppm, dt, JH-P = 63.9 Hz, JH-P = 15.3 Hz) in the [1H,31P]-HMBC-NMR spectrum.
Based on literature data these signals could be assigned to the cationic carbonyl complex
[Ru(H)(CO)2(Triphos)]+ (4).[132] ESI-MS analysis (m/z = 783.1) further confirmed this
assignment. According to the integral ratios in the 31P{1H}-NMR spectrum, the content of
complex 4 in solution was about 4 %. The formation of complex 4 was well in line with the
assumption of cationic complexes of type 3 as catalytically active species. The carbonyl
ligands in complex 4 most likely formed by decarbonylation of aldehyde intermediates or
methanol.[127, 140-141] This hypothesis was supported by the fact that complex 4 was the only
complex observed during the studies concerning alkyl formate hydrogenation with complex
2 (vide supra). Formation of complex 4 was revealed as possible deactivation pathway by
testing its catalytic activity in the CO2 hydrogenation in the absence of alcohol (standard
and 6 (δ = 35.0 ppm, s) (Figure 8). The signals at 46.5 ppm (dd, JP-P = 39.4 Hz, JP-P = 16.9 Hz),
9.5 ppm (dd, JP-P = 39.4 Hz, JP-P = 31.9 Hz), and -2.2 ppm (m) could be assigned to
[RuCl(H)CO(Triphos)] (7) by comparison with literature data.[145] In the 1H-NMR spectrum the
corresponding hydride signal was observed at -5.8 ppm (ddd, JH-P = 95.0 Hz, JH-P = 19.7 Hz, JH-P
= 15.0 Hz). These results confirmed that Ru-Triphos complexes bearing chloro ligands are
inactive in the CO2 hydrogenation to methanol.
RESULTS & DISCUSSION
-35-
Figure 8: 31
P{1H}-NMR spectrum (121 MHz, d8-THF, r.t.) of the reaction solution after CO2 hydrogenation to methanol
(20 bar CO2, 60 bar H2, 140 °C, 1 h) with catalyst 2 (25 μmol) and HNTf2 (1 eq.) in d8-THF (2.0 mL) in the presence of 1-butyl-3-methylimidazolium chloride (3 eq.).
The main species in the 31P{1H}-NMR spectrum shown in Figure 6 accounting for over 85 % of
the total signal intensity gave rise to a broad singlet at 44.2 ppm. The shape of the signal
indicated fluxional behaviour of this species at room temperature. Therefore, a low-
temperature NMR measurement was performed at 233 K. The broad singlet split into a
doublet (46.3 ppm, 2P, JP-P = 42.5 Hz) and a triplet (43.9 ppm, 1P, JP-P = 42.5 Hz). In the
[1H,31P]-HMBC-NMR spectrum recorded at 233 K a coupling of these signals to a proton
signal at 8.7 ppm (br) was observed, which was well in the range of ruthenium coordinated
formate (Figure 7).[134-136] The corresponding [1H,13C]-HMBC-NMR experiment showed the
coupling of the proton signal at 8.7 ppm to a singlet at 178.8 ppm in the 13C{1H}-NMR,
further corroborating the formation of a formate complex.[110, 146-147] Hydride signals
corresponding to this species were not detected in the [1H,31P]-HMBC-NMR.
The same formate complex could be generated independently starting from complex 2 as
follows: HNTf2 (7.0 mg, 0.025 mmol) was dissolved in d8-THF (0.5 mL) and added to complex
RESULTS & DISCUSSION
-36-
2 (19.5 mg, 0.025 mmol, 1 eq.) in d8-THF (0.5 mL) at room temperature giving a deep red
coloured solution. HCO2H (0.9 μL, 0.025 mmol, 1 eq.) was then added via micro-syringe and
the solution turned orange. NMR analysis of the crude reaction mixture at 233 K showed the
generation of the same formate complex as observed in the reaction solution of a CO2
hydrogenation reaction (compare Figure 7 and Figure 9).
Figure 9: [1H,
31P]-HMBC spectrum (600 MHz, d8-THF, -40 °C) after addition of 1 eq. of HNTf2 and 1 eq. of HCO2H to complex
2 in d8-THF.
The 31P{1H}-NMR spectrum showed the characteristic set of doublet (δ = 47.1, JP-P = 42.4 Hz,
2P) and triplet (δ = 44.2, JP-P = 42.4 Hz, 1P) (Figure 9), the 1H-NMR spectrum showed the
formate signal at 8.8 ppm (Figure 9), and the 13C{1H}-NMR showed the formate signal at
178.4 ppm. According to the integrals in the 31P{1H}-NMR spectrum the formate complex
formed in about 70 % besides some other, yet unidentified phosphor containing species. The
yield of the formate complex could be increased to 86 % by adding HNTf2 and HCO2H more
carefully to complex 2 while vigorously stirring the solution. FT-IR analysis of the crude
solution at room temperature showed a νCO stretching mode at 1543 cm-1, which is a typical
value for 2-coordinated formate (Figure 10).[135-136, 146-147] Based on these data and on the
basis of literature precedence,[135] the structure of this complex was assigned as
[Ru(2-O2CH)(Triphos)(THF)]+ (8a) (Scheme 15). As a formate exchange between 2- and 1-
RESULTS & DISCUSSION
-37-
coordination could be excluded based on the IR-data,[135] the fluxional behaviour of complex
8a observed on the NMR time scale could be most plausibly explained as follows: The labile
THF ligand temporarily dissociates to form a five-coordinated intermediate. Subsequent
association of the THF ligand in another position, thereby changing places with a
coordinated formate-oxygen, leads to exchange of the axial and equatorial phosphorus
positions on the NMR time scale. Therefore, axial and equatorial positions cannot be
discriminated by NMR-spectroscopy. Cooling of the solution slows down the exchange, thus,
axial and equatorial positions become distinguishable.
Figure 10: FT-IR spectrum (transmission) after addition of 1 eq. of HNTf2 and 1 eq. of HCO2H to complex 2 in d8-THF.
RESULTS & DISCUSSION
-38-
Scheme 15: Formation of the same formate complex 8a from catalyst precursor 2 in either the presence of HNTf2 (1 eq.) and H2/CO2 under reaction conditions (upper pathway), or by addition of HNTf2 (1 eq.) and HCO2H (1 eq.) in THF (lower pathway).
This interpretation was supported by the formation of a non-fluxional formate complex
upon addition of the less labile ligand acetonitrile (0.1 mL) to the freshly prepared solution
of 8a in THF (0.5 mL) at room temperature: The signals due to 8a disappeared completely
and the 31P{1H}-NMR spectrum measured directly after addition of the acetonitrile showed
new signals due to the formation of complexes bearing the MeCN ligand (Figure 11, bottom).
The main newly formed species accounting for 74 % of the total signal intensity in the
31P{1H}-NMR spectrum gave rise to a doublet (42.8 ppm, 2P, JP-P = 42.2 Hz) and a triplet
(29.6 ppm, 1P, JP-P = 42.2 Hz). In the 1H-NMR spectrum the formate signal was still apparent
at 8.6 ppm (br) (Figure 12, bottom). FT-IR analysis of this solution at room temperature again
showed a νCO stretching mode at 1544 cm-1, consistent with the structure [Ru(2-
O2CH)(Triphos)(MeCN)]+ 8b. Interestingly, the signals of 8b decreased over a period of 5
hours at room temperature at the expense of a new doublet (47.6 ppm, 2P, JP-P = 20.6 Hz)
and triplet (5.5 ppm, 1P, JP-P = 20.6 Hz) (Figure 11, top). At the same time, the formate signal
at 8.6 ppm disappeared with concomitant formation of an upfield hydride signal (dt,
-5.5 ppm, JH-P = 105.0 Hz, JH-P = 19.3 Hz) in the 1H-NMR spectrum (Figure 12, top).
RESULTS & DISCUSSION
-39-
Figure 11: 31
P{1H}-NMR spectrum (121 MHz, d8-THF, r.t.) after addition of 1 eq. of HNTf2, 1 eq. of HCO2H, and 0.1 mL
acetonitrile to complex 2 in d8-THF measured directly (bottom) and again after 5 hours at room temperature (top).
Figure 12: 1H-NMR spectrum (300 MHz, d8-THF, r.t.) after addition of 1 eq. of HNTf2, 1 eq. of HCO2H, and 0.1 mL acetonitrile
to complex 2 in d8-THF measured directly (bottom) and again after 5 hours at room temperature (top).
RESULTS & DISCUSSION
-40-
These NMR data indicated formation of the literature known complex
[Ru(H)(MeCN)2(Triphos)]+ (9) by decarboxylation of 8b.[148] Therefore, the formation of
formate complex 8 from complex 2 could be most plausibly explained via reversible CO2-
insertion into the analogous solvent-coordinated cationic Ru-hydride complex (Scheme 16).
Scheme 16: Addition of acetonitrile to a solution of 8a in THF leads to the formation of the acetonitrile complex 8b; complex 9 is formed by decarboxylation of 8b at room temperature.
3.2.2 In situ NMR-spectroscopic investigations
Ruthenium-formate complexes are very common intermediates in the hydrogenation of CO2
to formic acid.[35, 149-150] Therefore, high-pressure NMR experiments were carried out to
probe if the formate complex 8a is a kinetically competent intermediate also in the
hydrogenation of CO2 to methanol: As described above, a solution of 8a was prepared by
adding HNTf2 (1 eq.) and HCO2H (1 eq.) to complex 2 in d8-THF. The orange solution (0.3 mL)
was transferred to a high-pressure NMR tube made of sapphire (inner volume = 0.93 mL),
and 1H-NMR and 31P{1H}-NMR spectra of this solution were recorded at room temperature.
Then, the NMR tube was pressurised with hydrogen (60 bar), and carefully heated at 140 °C
by dipping the lower 4 cm of the tube into an oil-bath. After 40 minutes at 140 °C, 1H-NMR
and 31P{1H}-NMR spectra were recorded at 80 °C in the spectrometer. Figure 13 shows both
1H-NMR spectra. The spectrum recorded directly after preparing the solution showed the
broad singlet of formate species 8a at 8.62 ppm (Figure 13, bottom). In the spectrum
recorded after heating the solution at 140 °C under H2 pressure, the formate signal had
nearly disappeared (Figure 13, top). Instead, a singlet at 3.27 ppm was observed, showing
the formation of methanol. Integration of the formate signal and the methanol signal
relative to the aromatic protons of Triphos revealed a nearly complete conversion (ca. 97 %)
of the formate-ligand to methanol. The corresponding 31P{1H}-NMR spectra are shown in
Figure 14. The spectrum recorded directly after preparing the solution showed the broad
singlet at 42.9 ppm in about 83 % of the total intensity corresponding to the formate
complex 8a (Figure 14, bottom). Besides this signal, a sharp singlet was observed at
RESULTS & DISCUSSION
-41-
57.0 ppm indicating the presence of a symmetric, yet unidentified phosphor containing
species (ca. 17 % of total intensity). No correlation of this signal to ligands other than Triphos
were observed in a [1H,31P]-HMBC-NMR measurement. One might speculate that this signal
belongs to a solvato complex in which three THF molecules are coordinated to the
ruthenium-Triphos fragment. In the spectrum recorded after heating the solution at 140 °C
under H2 pressure the signals due to 8a and the unknown complex had disappeared (Figure
14, top). Instead, the formation of the hydride dimer 5 in around 44 % yield was observed
besides some yet unidentified species (Figure 14, bottom).
Figure 13: 1H-NMR spectra (300 MHz, d8-THF) of the reaction mixture after addition of 1 eq. HNTf2 and 1 eq. HCO2H to
complex 2 in d8-THF measured in a high-pressure NMR tube at r.t. before pressurising with H2 (bottom), and at 80 °C after pressurising with 60 bar H2 and heating at 140 °C in an oil-bath (top).
RESULTS & DISCUSSION
-42-
Figure 14: 31
P{1H}-NMR spectra (121 MHz, d8-THF) of the reaction mixture after addition of 1 eq. HNTf2 and 1 eq. HCO2H to
complex 2 in d8-THF measured in a high-pressure NMR tube at r.t. before pressurising with H2 (bottom), and at 80 °C after pressurising with 60 bar H2 and heating at 140 °C in an oil-bath (top).
In summary, this experiment showed that methanol could be formed directly by
hydrogenation of the coordinated formate ligand in complex 8a (Scheme 17). No
intermediate stabilisation of the formate as alkyl formate was necessary.
Scheme 17: Hydrogenation of the formate complex 8a leads to the formation of methanol in high yield.
In situ high-pressure NMR studies were carried out to elucidate the behaviour of complex 2
directly under turnover conditions. Complex 2 (0.0125 mmol) was dissolved in d8-THF
(0.25 mL), and HNTf2 (1 eq.) dissolved in d8-THF (0.25 mL) was added to give a deep red
solution. 0.3 mL of this solution were transferred to a high-pressure NMR tube (inner volume
= 0.93 mL), and 1H-NMR and 31P{1H}-NMR spectra were recorded at 25 °C. After pressurising
with CO2 (20 bar) and H2 (60 bar) again spectra were recorded at 25 °C. The NMR tube was
heated at 80 °C by a hot air stream inside the NMR machine, and 1H-NMR spectra were
recorded directly, after 1 hour, after 2 hours, and after 4 hours. 31P{1H}-NMR spectra were
RESULTS & DISCUSSION
-43-
always recorded after the corresponding 1H-NMR spectra with a delay of 30 minutes due to
the time for measuring the 1H-NMR spectra and due to shimming. After cooling to room
temperature again a 1H-NMR spectrum was recorded.
The 31P{1H}-NMR spectra are depicted in Figure 15. The spectrum recorded at 0 bar
overpressure and 25 °C showed no signal due to the starting complex 2 and two new sharp
singlets at 48.6 ppm and 58.8 ppm. The sharp singlets indicated the formation of two
symmetric species. In an attempt to identify these species, a separate experiment was
carried out: Complex 2 (0.025 mmol) was dissolved in dichloromethane (DCM, 0.5 mL) and
HNTf2 (0.025 mmol, 1 eq.) in d8-THF was added at -78 °C. The deep red solution was
transferred to a NMR tube, and a [1H,31P]-HMBC-NMR experiment was carried out at -40 °C
(Figure 16). The 31P{1H}-NMR spectrum showed again mainly the formation of the two sharp
singlets at 49.4 ppm (74 % of total intensity in 31P{1H}-NMR) and 59.5 ppm (10 %).
Additionally, a singlet at 33.9 ppm (13 %) due to unreacted starting complex 2 was observed.
In the [1H,31P]-HMBC-NMR experiment no correlation of the singlet at 59.5 ppm to ligands
other than Triphos was apparent. As inferred from previous experiments (vide supra), one
might speculate that this signal belongs to a cationic solvato complex forming after complete
protonation of the TMM ligand to isobutene. The main signal at 49.4 ppm showed, besides
the coupling to the Triphos ligand (δ = 2.6 ppm, s, 6H, CH2; δ = 1.8 ppm, s, 3H, CH3), a
coupling to two further signals at 1.9 ppm (s) and 1.6 ppm (s). One might speculate that
these signals stem from a coordinated methylallyl ligand in [Ru(methylallyl)(Triphos)]+ (11),
which was reported to form upon protonation of the TMM ligand.[151] More details on this
reaction will be published in D. Limper’s PhD-Thesis.*
In the 31P{1H}-NMR spectrum recorded at 25 °C after pressurising with CO2/H2 the sharp
singlets at 48.6 ppm and 58.8 ppm had almost disappeared (Figure 15). Instead, a broad
singlet at 43.8 ppm with a shoulder was observed. After heating up to 80 °C solely this broad
singlet remained. Based on the previous investigations concerning formate complex 8a (vide
supra) this signal was assigned to 8a. In the 31P{1H}-NMR spectrum recorded after 90
minutes the formation of a sharp singlet at 43.2 ppm was observed, whose integral
increased upon longer reaction times. This signal was assigned to the dimeric complex 5
which was identified as a deactivation product earlier (vide supra). No formation of the
* Lehrstuhl für Technische Chemie und Petrolchemie, RWTH Aachen University, Germany.
RESULTS & DISCUSSION
-44-
second deactivation product [Ru(H)(CO)2(Triphos)] (4) was observed under these mild
conditions and short reaction time.
Figure 15: 31
P{1H}-NMR spectra (121 MHz, d8-THF) of a CO2 hydrogenation reaction carried out in a high-pressure NMR
tube. Reaction conditions: complex 2 (25 μmol/mL), HNTf2 (1 eq.), d8-THF (0.3 mL), p(CO2) = 20 bar at r.t., p(H2) = 60 bar at r.t.
RESULTS & DISCUSSION
-45-
Figure 16: [1H,
31P]-HMBC spectrum (600 MHz, d8-THF, -40 °C) after addition of HNTf2 (1 eq.) to complex 2 in d8-THF/DCM.
In Figure 17 magnifications of the formate areas of the recorded 1H-NMR spectra are
depicted. Directly after pressurising the solution with CO2/H2 at room temperature a broad
singlet at 8.7 ppm appeared. Together with the 31P{1H}-NMR data formation of a certain
amount of the formate complex 8a already at room temperature seemed to be likely. The
formate signal was strongly broadened at 80 °C indicating fluctuating formate species. At the
same time a singlet at 3.29 ppm (not shown) showed the formation of methanol. After 4
hours the NMR tube was cooled to room temperature, and again a 1H-NMR spectrum was
recorded which showed the signal due to formate complex 8a at 8.7 ppm.
Figure 18 shows magnifications of the hydride areas of the recorded 1H-NMR spectra. The
spectra recorded after 1 hour and after 2 hours at 80 °C showed a very small signal at
-6.7 ppm. However, this hydride species is unlikely to play a role in the catalytic
transformation of CO2 to methanol as the signal was not observed in the spectra recorded
thereafter. In the spectra recorded after 1, 2, and 4 hours a broad hydride signal was
observed at -8.8 ppm with increasing integrals indicating the formation of dimeric species 5.
RESULTS & DISCUSSION
-46-
Figure 17: 1H-NMR spectra (300 MHz, d8-THF) of a CO2 hydrogenation reaction carried out in a high-pressure NMR-tube.
Magnification of the formate area. Reaction conditions: complex 2 (25 μmol/mL), HNTf2 (1 eq.), d8-THF (0.3 mL), p(CO2) = 20 bar at r.t., p(H2) = 60 bar at r.t.
RESULTS & DISCUSSION
-47-
Figure 18: 1H-NMR spectra (300 MHz, d8-THF) of a CO2 hydrogenation reaction carried out in a high-pressure NMR-tube.
Magnification of the hydride area. Reaction conditions: complex 2 (25 μmol/mL), HNTf2 (1 eq.), d8-THF (0.3 mL), p(CO2) = 20 bar at r.t., p(H2) = 60 bar at r.t.
To assess the reversibility of the CO2 hydrogenation reaction to methanol the CO2/H2
pressure was released and the reaction solution was heated for 2.5 days at 80 °C. However,
no decrease of the methanol signal at 3.29 ppm could be observed in the 1H-NMR spectrum.
In summary, these NMR studies suggested that the formate complex 8a was formed directly
after pressurising a solution of complex 2 and HNTf2 with CO2/H2 at room temperature and
remained the main detectable species under reaction conditions. Throughout the reaction
no hydride species were detected which could be ascribed to active species, indicating that
the presumed cationic hydride intermediates are too short-lived to be observed on the NMR
time scale.[135] It is likely that the short-lived hydride species are rapidly and reversibly
converted to the observable formate complex 8a as resting state by CO2 insertion into the
ruthenium-hydride bond.[35, 136, 149-150]
RESULTS & DISCUSSION
-48-
3.2.3 Proposal for a catalytic cycle
The experimental results described above strongly suggested that CO2 can be reduced
stepwise via the formate anion in the coordination sphere of a cationic, homogeneous
organometallic complex based on Ru-Triphos. However, as the formate complex 8a was
identified as the resting state under turnover conditions, further spectroscopic insight into
the reaction mechanism could not be obtained directly. Therefore, DFT calculations were
performed in order to get a more detailed picture of this multistep transformation. A basic
catalytic cycle for the stepwise hydrogenation of CO2 to methanol through the formic acid
and formaldehyde stages could be formulated based on previous mechanistic investigations
on the Ru-Triphos system by Klankermayer and Leitner et al.,[127] and based on recent work
on hydrogenation of CO2/CO2-derivatives,[104, 110, 152-153] formic acid decomposition,[135, 154]
formaldehyde decomposition,[155] and methanol reforming (Scheme 18):[63] A plausible
starting point of this catalytic cycle is the cationic ruthenium-hydride species 3/I, which most
likely forms from complex 2 by protonation of the TMM ligand by the acidic additive and
hydrogenolysis. Migratory insertion of CO2 results in the formation of the spectroscopically
observed formate complex (8/V). Reaction with one equivalent H2 leads to the reduction
beyond the formic acid stage to Ru-hydroxymethanolate species (IX). Via intermediate
formation of formaldehyde and consumption of a second equivalent of hydrogen IX is
transformed to the Ru-methanolate complex (XVIII). The product methanol is finally
liberated and the cycle is closed by hydrogenolysis of the Ru-OMe unit with a third
equivalent of hydrogen.
RESULTS & DISCUSSION
-49-
Scheme 18: Proposed basic catalytic cycle for the hydrogenation of CO2 to methanol.
3.2.4 DFT-calculations
The DFT-calculations presented in this chapter have been performed in cooperation with
Verena Moha, Jens Kothe, and Markus Hölscher at the Institut für Technische und
Makromolekulare Chemie, RWTH Aachen University.
3.2.4.1 Calculation of the catalytic cycle
Generation of formate complex 8/V
The first step of the basic catalytic cycle is the insertion of CO2 into the Ru-H bond of
complex I to form the spectroscopically observed formate complex 8/V. This step was widely
investigated in the context of formic acid production. The system most closely resembling
the Ru-Triphos system was a ruthenium complex bearing three phosphine ligands, which
was investigated by Sakaki’s group.[152] Therefore, a similar reaction pathway was
investigated for the first step of the catalytic cycle (Figure 19, red pathway): The structure
RESULTS & DISCUSSION
-50-
with the lowest energy II (9.5 kcal mol-1) is obtained by replacing the THF molecule in I by
CO2. In the next step, the hydride ligand is transferred to the carbon atom of the CO2
molecule via transition state II-III (21.2 kcal mol-1). Rotation of the Ru-O and Ru-C bonds in III
leads to the formation of the stable 2-formate IV (3.3 kcal mol-1). Subsequent exchange of
the H2 in IV by THF leads to the even more stable resting state V/8a (-3.2 kcal mol-1). This
result was in agreement with the fact that only V/8a was observed during high-pressure
NMR investigations. Moreover, the small energetic span of 21.2 kcal mol-1 between I and
V/8a was in agreement with the observation of V/8a under CO2/H2 pressure already at room
temperature.
Figure 19: Generation of the spectroscopically observed formate complex 8/V and subsequent generation of Ru-hydroxymethanolate complex IX. The energetically most favourable pathway is shown in red. The Triphos ligand is omitted for clarity, S = THF.
Generation of Ru-hydroxymethanolate complex IX
The second step of the basic catalytic cycle is the reduction of the formate beyond the
formic acid stage resulting in the Ru-hydroxymethanolate species IX (Figure 19, red
pathway). Coordinated formic acid is generated from the coordinated formate in complex
V/8a by changing the formate coordination mode from bidentate to monodentate and
association of a H2 molecule (VI, 11.8 kcal mol-1), and by subsequent proton transfer to the
RESULTS & DISCUSSION
-51-
carbonyl C=O bond via a six-membered transition state (VI-VII, 13.6 kcal mol-1).[152] In the
resulting formic acid complex (VII, 12.5 kcal mol-1) the Ru-H unit is regenerated. Three other
pathways leading to VII were investigated (Figure 19, black pathways), including outer
sphere attack of CO2, but were much less energetically favourable and did not pass through
the spectroscopically observed formate complex V, making these pathways unlikely to play a
role. There was much less known about the reaction steps leading to reduction beyond the
formic acid stage.[63, 116-117, 155] For the Ru-Triphos system investigated here, formation of the
crucial Ru-hydroxymethanolate intermediate IX (16.1 kcal mol-1) could be rationalised by
exchange of the THF molecule in VII for a H2 molecule to give VIII (14.3 kcal mol-1) and
subsequent hydride transfer to the carbon atom of the formic acid via transition state VIII-IX
(29.8 kcal mol-1).
Generation of Ru-methanolate complex XVIII
The third step of the basic catalytic cycle is the reduction of the Ru-hydroxymethanolate
species IX to the Ru-methanolate complex XVIII via the formaldehyde stage. The
interconversion between free methanediol and formaldehyde has been studied.[155] Here,
the protonolysis within the coordination sphere through heterolytic cleavage of the
coordinated H2 molecule was investigated (Figure 20, the energetically most favourable
pathway is shown in red): Coordination of a THF molecule to IX results in the formation of
IXc (34.2 kcal mol-1). Via transition state IXc-XV (42.7 kcal mol-1), involving proton transfer
and C-O bond cleavage, formaldehyde complex XV (25.0 kcal mol-1) is formed directly. By
interaction with an additional THF molecule via hydrogen bonding the barrier is lowered to
40.5 kcal mol-1 (transition state IXc-XV(thf)). The other pathways shown in Figure 20 had
significantly higher energy barriers.
RESULTS & DISCUSSION
-52-
Figure 20: Generation of formaldehyde complex XV from methanediolate complex IX via intramolecular proton transfer. The energetically most favourable pathway is shown in red. The Triphos ligand is omitted for clarity, S = THF.
Pathways involving external medium-assisted proton transfer were also investigated (Figure
21): Carboxylate units as proton shuttles (like formic acid or acetic acid; here calculated for
the case of acetic acid, green pathway in Figure 21) may in principle lower the energy barrier
significantly to 27.1 kcal mol-1 (transition state Xa-XIa). A similar pathway using water as the
external proton shuttle gave a barrier of 41.2 kcal mol-1, which was comparable to the
barrier calculated for the intramolecular proton transfer pathways (Figure 21, blue pathway).
RESULTS & DISCUSSION
-53-
Figure 21: Generation of formaldehyde complex XV from methanediolate complex IX via medium-assisted proton transfer (water: blue; acetate: green). The Triphos ligand is omitted for clarity, S = THF.
The methanolate complex XVII (21.7 kcal mol-1) is formed by replacement of the THF
molecule in formaldehyde complex XV by a H2 molecule to give XVI (26.6 kcal mol-1) and
subsequent migratory hydride transfer (transition state XVI-XVII, 27.3 kcal mol-1) (Figure 22).
XVII is stabilised by an agostic C-H-Ru interaction. Association of a solvent molecule leads to
the formation of methanolate complex XVIII (22.3 kcal mol-1).
Formation of methanol and closing the catalytic cycle
The last step of the basic catalytic cycle is the formation of methanol and the closing of the
catalytic cycle by regeneration of hydride complex I (Figure 22): A proton is intramolecularly
transferred to the oxygen atom of the coordinated methanolate via the four-membered
transition state XVIII-XXIV (31.5 kcal mol-1) to give the methanol complex XXIV
(12.8 kcal mol-1). Dissociation of methanol and association of a H2 molecule regenerates
starting complex I’. I’ lies 14.1 kcal mol-1 above the reference point, indicating that the
reaction is endergonic under the boundary conditions of the calculation model. This was due
to the fact that all values shown here were computed for the gas phase, an approach
commonly applied because calculations with the inclusion of solvent effects are very time
RESULTS & DISCUSSION
-54-
consuming for huge organometallic complexes and gas phase calculations served sufficiently
well for the understanding of reaction mechanisms in many cases.[156] Recalculation with the
inclusion of solvent effects was therefore restricted to the net reaction, using the MN12-L
density functional in combination with the IEF-PCM (Grel(I’) = -1.9 kcal mol-1) and the CPCM
continuum model (Grel(I’) = -2.2 kcal mol-1), and the IEF-PCM additionally with a radii model
recently developed by Truhlar et al.[157] (Grel(I’) = -4.1 kcal mol-1). All results indicated the
reaction to be exergonic, in line with the experimental observations.
Figure 22: Formation of methanol and closing the catalytic cycle via methanolate complex XVIII. The Triphos ligand is omitted for clarity, S = THF.
In summary, a plausible catalytic cycle including all transition states could be found by DFT
calculations, indicating the possibility to reduce CO2 stepwise to methanol within the
coordination sphere of a single Ru-Triphos centre through a series of hydride transfer and
protonolysis steps. However, according to the energetic-span model introduced by Kozuch
and Shaik[156] complex V/8a (lying at -3.2 kcal mol-1 on the hypersurface; Figure 19)
represents the TOF determining intermediate (TDI), and IXc-XV(thf) (lying at 40.5 kcal mol-1
on the hypersurface; Figure 20) represents the TOF determining transition state (TDTS) of
the catalytic cycle. The resulting energetic span of 43.7 kcal mol-1 appeared to be high for a
reaction running at 140 °C.[158] The calculation method used (gas phase, M06-L/def2-TZVP)
RESULTS & DISCUSSION
-55-
was shown to give reliable results, at least for the hydrogenation of olefins.[158]
Nevertheless, recalculation of the catalytic cycle in the solvent phase might be interesting for
validation. As shown in Figure 21, external proton sources might significantly lower the
energetic span and therefore cannot be excluded to play a role in the catalytic cycle at this
stage.
3.2.4.2 The influence of the coordination geometry
As the Ru-Triphos system was the first organometallic catalyst which enabled the reduction
of CO2 to methanol within its coordination sphere, investigations concerning the influence of
the facial geometry of the Triphos ligand were carried out. A comparison of the facial
coordination with a meridional coordination could not be performed with the original
Triphos ligand because of its rigidity. Therefore, three P(Ph)2Me ligands instead of one
Triphos ligand were used to recalculate a crucial step in the catalytic cycle, the reduction of
formic acid complex VIII to methanediolate complex IX via transition state VIII-IX (Figure 23).
Complexes corresponding to the structures of VIII, IX and VIII-IX were constructed in facial
and meridional coordination mode. The meridional analogue of structure VIII was chosen as
reference point (0 kcal mol-1). The span for the facial case (Figure 23, red) was 11.7 kcal mol-1
and therefore in satisfying agreement with the barrier height of 15.5 kcal mol-1 of the
original Triphos complex, indicating that three P(Ph2)Me ligands were a suitable model for
the Triphos ligand. For the meridional case the span was 36.4 kcal mol-1 (Figure 23, black),
showing that the facial coordination of the Triphos ligand indeed was crucial for low energy
barriers.
RESULTS & DISCUSSION
-56-
Figure 23: Comparison of meridional with facial coordination by recalculation of the structures VIII, VIII-IX, and XI using the ligand P(Ph)2Me as model.
3.3 Hydrogenation of CO2 to MeOH in the absence of alcohol additives
Parts of this chapter have been published in: Chem. Sci. 2015, 6, 693-704.[139]
After demonstrating the possibility of CO2 reduction to methanol in the coordination sphere
of a single Ru-Triphos centre in the absence of an alcohol additive by NMR experiments and
DFT calculations further key parameters of the catalytic reaction were investigated, and the
results were analysed for consistency with the mechanistic proposal.
3.3.1 The role of the acidic additive - Development of a catalytic system with no need for
an acidic additive
Firstly, a more detailed analysis of the role of the acidic additive was carried out. In the
NMR-experiments described above the cationic formate complex 8 was identified as an
active intermediate in the catalytic transformation of CO2 to methanol. Additionally, the
DFT-calculations strongly supported that cationic species of type 3 play an important role in
the catalytic cycle. In summary, these results suggested that one role of the acid in the
catalytic system 2/HNTf2 is the generation of a cationic species as the active site upon
-2 0 2 4 6 8
10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40
DG
/ (
kcal
/mo
l)
TS
TS
fac
mer 0.0
5.6
36.4
17.3
30.7
10.5
RESULTS & DISCUSSION
-57-
reductive removal of the TMM ligand. To prove this assumption, the preparation of an
isolated cationic catalyst precursor was envisaged. Attempts to isolate formate complex 8
failed due to the instability of this complex. However, synthesis of the analogous cationic
acetate complex was successful (Scheme 19): In the first step, the literature known complex
[Ru(2-OAc)Cl(Triphos)] (12) was synthesized by stirring commercially available
[Ru(2-OAc)Cl(PPh3)3] (13) together with Triphos (1 eq.) in toluene at 110 °C for 3 hours.[159]
In the second step, isolated complex 12 was stirred together with AgNTf2 (1.03 eq.) in THF or
toluene at 60 °C for 3 hours. The greyish AgCl precipitate was filtered off by passing the
yellow solution over silica. Removal of the solvent in vacuo and drying for 24 hours in vacuo
at room temperature yielded a yellow-orange powder which was analysed by 1H-, 13C-, 31P-,
and 19F-NMR, FT-IR, and ESI-HRMS.
Scheme 19: Synthesis of the acetate complex 14 (S = THF, H2O or free coordination site, depending on if and which coordinating molecules are available).
The 31P{1H}-NMR spectrum in thoroughly dried d2-DCM (r.t.) showed only one sharp singlet
at 42.5 ppm, indicating the selective formation of either a symmetrical complex or formation
of a highly dynamic complex (Figure 24). In the 1H-NMR spectrum in thoroughly dried d2-
DCM no other signals than those due to the Triphos and acetate ligands were observed
(Figure 25). These data in combination with 13C- and 19F-NMR data suggested the presence
of the cation [Ru(2-OAc)(Triphos)]+ and the anion NTf2‒. This was supported by ESI-HRMS
analysis (m/z (+) = 785.14337). FT-IR analysis of the yellow-orange powder showed very
similar absorption bands between 1400 cm-1 and 1600 cm-1 as compared to the starting
complex 12 indicating a similar 2-binding mode of the acetate ligand.[160] Yellow single
crystals suitable for X-Ray diffraction were obtained by crystallisation from DCM layered
with pentane. The X-Ray diffraction data gave the structure of complex
[Ru(2-OAc)(Triphos)(H2O)]NTf2 (14a) in which the open coordination site was saturated with
H2O from adventitious traces of water (Figure 26). Consequently, 31P{1H}-NMR spectra of 14
measured in wet d2-DCM showed splitting of the singlet at 42.5 ppm into a doublet
(41.4 ppm, JP-P = 44.5 Hz, 2P) and a triplet (43.7 ppm, JP-P = 44.5 Hz, 1P) upon cooling the
RESULTS & DISCUSSION
-58-
solution to 223 K (Figure 27). However, the 1H-NMR spectrum of the dried complex in dry d2-
DCM showed no signal due to coordinated water (Figure 25), and thus the acetate complex
in solution could be formulated as [Ru(2-OAc)(Triphos)(S)]NTf2 (14) with S being either a
free coordination site or a weakly bound solvent molecule, depending on whether a
coordinating molecule is available or not.[161]
Figure 24: 31
P{1H}-NMR spectrum (243 MHz, d2-DCM, r.t.) of complex 14 in thoroughly dried d2-DCM.
RESULTS & DISCUSSION
-59-
Figure 25: 1H-NMR spectrum (400 MHz, d2-DCM, r.t.) of complex 14 in thoroughly dried d2-DCM.
Figure 26: Solid state structure of the cation 14a as derived from X-ray diffraction (hydrogen atoms are omitted for clarity). Some selected bond lengths (Å): Ru‒P1 = 2.245(9); Ru‒P2 = 2.255(3); Ru‒P3 = 2.253(0); Ru‒O1 = 2.171(2); Ru‒O2 = 2.208(6); Ru‒O3 = 2.204(7).
RESULTS & DISCUSSION
-60-
Figure 27: 31
P{1H}-NMR spectra (243 MHz, d2-DCM) of complex 14 measured in wet d2-DCM. Top: measured at room
temperature. Bottom: measured at 223 K. In the presence of water the coordination site in 14 is saturated by water giving
the complex [Ru(2-OAc)(Triphos)(H2O)]NTf2 (14a).
From NMR-spectroscopy and HRMS it was not possible to discriminate between monomeric
[Ru(2-OAc)(Triphos)(S)]NTf2 (14) and multinuclearic species [Ru(2-OAc)(Triphos)(S)]x[NTf2]x.
However, the discrimination was possible by using the two different Triphos derivatives
Triphos and Triphos-anisyl†[142] (1,1,1-tris{bis(4-methoxyphenyl)phosphinomethyl}ethan): An
equimolar mixture of [Ru(2-OAc)Cl(Triphos)] (12) (10.3 mg, 12.5 μmol, 1 eq.) and [Ru(2-
OAc)Cl(Triphos-anisyl)] (15) (12.5 mg, 12.5 μmol, 1 eq.) was stirred together with AgNTf2
(11.6 mg, 30 μmol, 2.4 eq.) in toluene (1.5 mL) at 60 °C for 5 hours. The toluene was
removed in vacuo and the residue dissolved in d2-DCM (0.5 mL). After passing the orange
solution through a syringe-filter to remove the AgCl precipitate the solution was analysed by
NMR-spectroscopy. The 31P{1H}-NMR spectrum showed two singlets at 43.5 and 45.3 ppm in
a ratio of nearly 1 : 1, indicating the formation of two complexes (Figure 28). The signals
were ascribed to [Ru(2-OAc)(Triphos)(S)]NTf2 (14) and [Ru(2-OAc)(Triphos-anisyl)(S)]NTf2
(16). The respective 1H-NMR spectrum confirmed the formation of 14 and 16 (Figure 29).
The absence of further signals in the 31P{1H}-NMR and 1H-NMR due to mixed multinuclearic
complex species like [Ru2(μ-OAc)2(Triphos)(Triphos-anisyl)][NTf2]2 corroborated the
monomeric structure of 14 in solution.[162]
† The ligand Triphos-anisyl was synthesised by Markus Meuresch, ITMC, RWTH Aachen University.
RESULTS & DISCUSSION
-61-
Figure 28: 31
P{1H}-NMR (243 MHz, d2-DCM, r.t.) of the reaction of an equimolar mixture (12.5 μmol) of
[Ru(2-OAc)Cl(Triphos)] (12) and [Ru(
2-OAc)Cl(Triphos-anisyl)] (15) with AgNTf2 (2.4 eq.).
RESULTS & DISCUSSION
-62-
Figure 29: 1H-NMR (600 MHz, d2-DCM, r.t.) of the reaction of an equimolar mixture (12.5 μmol) of [Ru(
2-OAc)Cl(Triphos)]
(12) and [Ru(2-OAc)Cl(Triphos-anisyl)] (15) with AgNTf2 (2.4 eq.).
A high-pressure NMR experiment was carried out to investigate the reactivity of acetate
complex 14 under CO2 hydrogenation conditions: Complex 14 (13.3 mg, 0.0125 mmol) was
dissolved in d8-THF (0.5 mL), and 0.3 mL of this yellow solution were transferred to a high-
pressure NMR tube (inner volume = 0.93 mL). 1H-NMR and 31P{1H}-NMR spectra were
recorded at 25 °C. After pressurising with CO2 (20 bar) and H2 (60 bar) again 1H-NMR and
31P{1H}-NMR spectra were measured at 25 °C. The NMR tube was heated at 80 °C by a hot air
stream inside the NMR machine, and 1H-NMR spectra were recorded directly, and again
after 1.5 hours. 31P{1H}-NMR spectra were always recorded after the corresponding 1H-NMR
spectra with a delay of 30 minutes due to the time for measuring the 1H-NMR and due to
shimming. The NMR tube was carefully heated for 1 hour at 140 °C in an external oil bath,
and after that again 1H-NMR and 31P{1H}-NMR spectra were measured at 80 °C and 25 °C.
Figure 30 shows the recorded 31P{1H}-NMR spectra. The 31P{1H}-NMR spectrum of the
catalyst solution at 25 °C showed the broad singlet at 42.3 ppm due to starting complex 14.
Directly after pressurising with CO2/H2 a small, broad singlet appeared at 42.9 ppm,
indicating formation of a little amount of formate complex 8a already at room temperature.
At 80 °C inside the NMR machine both signals overlapped and only one broad singlet at
RESULTS & DISCUSSION
-63-
around 42 ppm was observable. No changes were observed in the spectra measured at 80 °C
after heating the solution for 2 hours at 80 °C and for 1 hour at 140 °C. Only after cooling to
25 °C again both signals at 42.3 ppm and 42.9 ppm could be observed. The signal at
42.3 ppm (about 40 % of total intensity) was related to starting complex 14. The newly
formed signal at 42.9 ppm (about 60 % of total intensity) was assigned to formate complex
8a. The formation of 8a in small amounts directly after pressurising with CO2/H2 at room
temperature and in about 60 % (according to 31P{1H}-NMR) after heating for 2 hours at 80 °C
and 1 hour at 140 °C was corroborated by the observation of a formate signal at 8.6 ppm in
the corresponding 1H-NMR spectra measured at 25 °C (Figure 31). At the same time
formation of methanol was observed (s, 3.2 ppm). In the spectra measured at 80 °C only a
broad singlet around 8.1 ppm was observed in the formate region, whereas in the last
spectrum measured after cooling to room temperature the formate signal of 8a was
observed at 8.6 ppm besides a singlet at 7.6 ppm due to traces of methyl formate. Methyl
formate can form once methanol is present in the reaction solution. However, it does not
accumulate in the reaction mixture as methyl formate is also hydrogenated to methanol
under reaction conditions (3.1.1). In the 1H-NMR spectra measured directly after pressurising
with CO2/H2 and after heating at 80 °C a very small broad hump at 10.6 ppm (not shown in
the magnification) could be observed, indicating the release of a very small amount of free
carboxylic acid into the solution. In the spectra after heating at 140 °C this signal
disappeared. Ethanol was detected in the 1H-NMR (t, 1.1 ppm) from acetate/acetic acid
hydrogenation. No hydride species was observed in any of the recorded 1H-NMR spectra,
again indicating that the presumed cationic hydride intermediates are too short-lived to be
observed on the NMR time scale.[135]
In summary, these studies demonstrated that the same formate complex 8a as observed in
the studies with complex 2/HNTf2 was formed also from acetate complex 14 in the absence
of acid or alcohol additives. Methanol was observed in solution corresponding to a TON of 5
proving that complex 14 can be used as molecularly defined precursor for the hydrogenation
of CO2 to methanol without any acidic additive.
RESULTS & DISCUSSION
-64-
Figure 30: 31
P{1H}-NMR spectra (121 MHz, d8-THF) of a CO2 hydrogenation reaction carried out in a high-pressure NMR
tube. Reaction conditions: complex 14 (25 μmol/mL), d8-THF (0.3 mL), p(CO2) = 20 bar at r.t., p(H2) = 60 bar at r.t.
Figure 31: 1H-NMR spectra (300 MHz, d8-THF) of a CO2 hydrogenation reaction carried out in a high-pressure NMR tube.
Magnification of the formate area. Reaction conditions: complex 14 (25 μmol/mL), d8-THF (0.3 mL), p(CO2) = 20 bar at r.t., p(H2) = 60 bar at r.t.
Next, the performance of catalytic systems 2 and 14 in presence and absence of acid was
investigated in more detail. As suggested by the NMR studies and by DFT calculations the
hydrogenation of CO2 to methanol is possible in the absence of an alcohol additive.
Therefore, the following studies were carried out in pure THF (i.e. in the absence of any
RESULTS & DISCUSSION
-65-
alcohol additive). For a more detailed analysis of the influence of the alcohol additive see
chapter 3.3.4.
Catalytic systems 2 and 14 were compared using a standard set of reaction conditions in the
absence of alcohol additives (V(THF) = 2.1 mL, c(Ru) = 12 mmol L-1, p(CO2) = 20 bar at r.t.,
p(H2) = 60 bar at r.t., T = 140 °C, t = 24 h) (Table 4). Catalyst 2 gave only a very low TON in the
absence of an acid (Table 4, entry 1). However, in the presence of an equimolar amount of
the acid HNTf2 a greatly increased TON of 228 was observed (Table 4, entry 2). Using the
cationic catalyst precursor 14 in the absence of any acidic additive a high TON of 165 was
achieved, whereas the addition of HNTf2 (0.5 eq.) did not increase the catalyst performance
(Table 4, entries 3 and 4). These results were consistent with the formation of the cationic
complex [Ru(H)(H2)(S)(Triphos)]+ (3) as catalytically active species, as suggested by DFT
calculations (3.2.4).
The lower performance of catalyst 14 compared to catalytic system 2/HNTf2 could be
explained by the more difficult hydrogenation of the acetate ligand to give the common
intermediate 8a, as observed in the in situ NMR experiments (vide supra): Whereas catalytic
system 2/HNTf2 was converted completely to formate complex 8a upon heating to 80 °C
under CO2/H2 pressure, catalytic system 14 was converted to 8a in about 60 % after 2 hours
at 80 °C and 1 hour at 140 °C.
The more difficult hydrogenation of acetic acid/acetate compared to formic acid/formate
was also evident from experiments concerning the hydrogenation of the free acids. With
catalytic system 2/HNTf2 (c(Ru) = 12.5 mmol L-1, 1 eq. HNTf2) formic acid (100 eq.) could be
fully converted to methanol at a hydrogen pressure of 60 bar (at r.t.) and a reaction
temperature of 140 °C within 24 hours (2.0 mL THF), whereas for the hydrogenation of acetic
acid (100 eq.) a reaction temperature of 180 °C was necessary to achieve full conversion. The
efficient hydrogenation of formic acid under these conditions was in line with the proposed
mechanism shown in Scheme 18.
The picture could be completed by a successful hydrogenation of paraformaldehyde
(100 eq.) to methanol using the same catalytic system 2/HNTf2 (c(Ru) = 12.5 mmol L-1, 1 eq.
HNTf2, 60 bar H2 at r.t., 2.0 mL THF, 0.2 mL H2O, 140 °C, 24 h). Interestingly, the 31P{1H}-NMR
spectrum of this reaction solution measured at 233 K showed a doublet (δ = 46.3 ppm, 2P,
JP-P = 42.4 Hz) and a triplet (δ = 43.8 ppm, 1P, JP-P = 42.4 Hz) indicating the formation of
RESULTS & DISCUSSION
-66-
formate complex 8a. Formation of 8a was corroborated by the observation of a broad singlet
at 8.7 ppm in the 1H-NMR spectrum. The observation of 8a indicated reversibility of the
catalytic cycle.
Table 4: Hydrogenation of CO2 to methanol in the absence of alcohol additive: influence of the acid.[a]
Entry Cat. Acid (eq.)[b]
pH2[c]
[bar]
pCO2[c]
[bar]
TON[d]
1 2 ‒ 60 20 8
2 2 HNTf2 (1.0) 60 20 228
3 14 ‒ 60 20 165
4 14 HNTf2 (0.5) 60 20 156
5 2 HNTf2 (0.9) 60 20 155
6 2 HNTf2 (1.5) 60 20 196
7 2 HNTf2 (2.0) 60 20 181
8 2 p-TsOH (1.0) 60 20 135
9 2 p-TsOH (1.5) 60 20 152
10 2 p-TsOH (2.0) 60 20 115
11 2 MSA (1.0) 60 20 61
12 2 MSA (1.5) 60 20 68
13 2 MSA (2.0) 60 20 20
[a] Reaction conditions: catalyst (25 µmol), THF (2.1 mL), 140 °C, 24 h; [b] equivalents to catalyst; [c] at room temperature;
[d] TON = mmol MeOH/mmol catalyst.
The lack of activity with complex 2 in the absence of an acid was investigated in more detail:
A CO2 hydrogenation was performed in a batch reactor using complex 2 in the absence of
any acidic additive and terminated after 1 hour (V(d8-THF) = 1.0 mL, c(Ru) = 25 mmol L-1,
p(CO2) = 20 bar at r.t., p(H2) = 60 bar at r.t., T = 140 °C). The 31P{1H}-NMR spectrum of the
solution showed a triplet (δ = 33.6 ppm, 1P, JP-P = 31.9 Hz) and a doublet (δ = 25.2 ppm, 2P,
JP-P = 31.9 Hz) indicating the formation of the neutral literature-known complex
[Ru(H)2CO(Triphos)] (17) in about 94 % (based on the total intensity in the 31P{1H}-NMR
spectrum) (Figure 32).[127] The formation of complex 17 was corroborated by observation of
the corresponding doublet of doublets (δ = -7.3 ppm, JH-P = 50.6 Hz, JH-P = 18.1 Hz) in the
hydride region of the 1H-NMR spectrum. Interestingly, Zanobini et al. showed that
protonation of complex 17 with strong protic acids at low temperatures leads to the
formation of [Ru(H)(H2)CO(Triphos)]+ (18), a cationic structure resembling the active hydride
species 3 which was used as the starting point for the DFT calculations of the catalytic cycle
(3.2.4).[163] Therefore, the productivity of isolated complex 17 in the presence of the strong
acid HNTf2 (1 eq.) was assessed in a batch reaction (V(THF) = 2.1 mL, c(Ru) = 6 mmol L-1,
p(CO2) = 20 bar at r.t., p(H2) = 60 bar at r.t., T = 140 °C, t = 24 h). Indeed, an active catalyst
was formed under these reaction conditions, and a TON of 256 was obtained after 24 hours,
RESULTS & DISCUSSION
-67-
which was about 76 % of the TON obtained using catalyst 2/HNTf2 under the same reaction
conditions.
Figure 32: 31
P{1H}-NMR spectrum (121 MHz, d8-THF, r.t.) of the reaction solution of a CO2 hydrogenation reaction using
complex 2 in the absence of any acidic additive terminated after 1 hour. (V(d8-THF) = 1.0 mL, c(Ru) = 25 mmol L-1
, p(CO2) = 20 bar at r.t., p(H2) = 60 bar at r.t., T = 140 °C).
The reactivity of complex 17 was investigated by means of NMR spectroscopy: Addition of
HNTf2 (1 eq.) to a solution of 17 in d8-THF at room temperature led to the evolution of gas
indicating the loss of a hydride ligand by protonation to H2. In the 31P{1H}-NMR spectrum
three doublets of doublets appeared (δ = 51.4 ppm, 1P, JP-P = 41.0, JP-P = 20.9 Hz; δ =
the formation of a Ru-Triphos species with three more different ligands filling up the
coordination sphere (Figure 33). In the corresponding 1H-NMR spectra the hydride signal due
to 17 disappeared and a pseudo doublet of triplets appeared instead (δ = -5.5 ppm, JH-P =
88.0, JH-P = 17.7 Hz). These data suggested the formation of the cationic complex
[Ru(H)CO(Triphos)(S)]+ (19).
With this reaction solution a CO2 hydrogenation reaction was performed (V(d8-THF) =
1.0 mL, c(Ru) = 25 mmol L-1, p(CO2) = 20 bar at r.t., p(H2) = 60 bar at r.t., T = 140 °C, t = 1 h),
and the reaction solution was analysed by 1H- and 31P{1H}-NMR at 233 K. Once again,
formation of formate complex 8a was evident from the characteristic signals in the 31P{1H}-
NMR and 1H-NMR spectra, indicating that CO dissociated from complex 19. However,
RESULTS & DISCUSSION
-68-
association of two CO ligands led to formation of the inactive biscarbonyl complex 4, which
could not be reactivated (see chapter 3.3.3). The fact that already one equivalent of CO was
present in the reactor when carbonyl complex 17 was used as catalyst precursor explained
the 24 % decrease in TON, as catalyst deactivation by formation of 4 happened faster.
In summary, these studies supported that cationic species play a major role in the catalytic
transformation of CO2 to methanol. Using neutral catalyst precursor 2 led to formation of
the inactive neutral carbonyl complex 17, whereas using neutral catalyst precursor 2 in the
presence of HNTf2 (1 eq.) or using cationic catalyst precursor 14 in the absence of acid led to
the formation of the active intermediate 8a.
Figure 33: 31
P{1H}-NMR spectra (162 MHz, d8-THF, r.t.) of complex 17 in d8-THF before (top) and after (bottom) the addition
of 1 eq. of HNTf2 at room temperature (S = solvent).
Next, the influence of the amount of HNTf2 added to complex 2 was investigated (Table 4,
entries 2, 5-7). The maximum TON of 228 was observed at a 1 : 1 molar ratio, supporting the
fact that the proton was required in stoichiometric amounts for reductive removal of the
RESULTS & DISCUSSION
-69-
TMM ligand in 2 leading to 3. With only 0.9 equivalents of HNTf2 a lower TON of 155 was
achieved. Increasing the amount of HNTf2 to 1.5 and 2.0 equivalents led to lower TONs
(calculated from the amount of methanol formed) of 196 and 181, respectively. One
explanation could be formation of dimethyl ether (DME) from consecutive etherification of
methanol in the presence of excess acid. As DME is a gas at room temperature (boiling point
= -24 °C) it was not observable in the reaction solution after depressurising the reactor.
Therefore, high-pressure NMR experiments were carried out. However, one has to bear in
mind that also in these experiments formed products might partition between gas and liquid
phase and only the liquid phase was analysed by NMR spectroscopy.
In a first experiment, MeOH (81 mg, 2.5 mmol) and HNTf2 (3.5 mg, 0.0125 mmol) were
dissolved in d8-THF (1 mL), and 0.5 mL of this solution were transferred to a high-pressure
NMR tube (inner volume = 0.93 mL). The solution was heated at 140 °C in an external oil
bath for 5 hours, and afterwards quantitative 1H-NMR and 13C{1H}-NMR spectra were
recorded. In the 1H-NMR spectrum formation of DME was evident from a singlet at
3.19 ppm. This was supported by formation of a singlet at 58.4 ppm in the 13C{1H}-NMR
spectrum. As the singlet due to MeOH (3.24 ppm) overlapped with the singlet due to DME
(3.19 ppm) in the 1H-NMR spectrum quantification was done based on the integrals found in
the 13C{1H}-NMR spectrum. According to the integral ratio, conversion of methanol to DME
was about 37 %. This result proved that DME indeed easily formed under reaction conditions
in the presence of excess acid. A CO2 hydrogenation reaction using complex 2 in the
presence of 1 equivalent HNTf2 was performed inside the high-pressure NMR tube (V(d8-
THF) = 0.3 mL, c(Ru) = 25 mmol L-1, p(CO2) = 10 bar at r.t., p(H2) = 30 bar at r.t., T = 140 °C, t =
24 h). Again, quantitative 1H-NMR and 13C{1H}-NMR spectra were recorded. According to the
integral ratio in the 13C{1H}-NMR spectrum a small amount (about 3.7 %) of the formed
methanol had reacted to DME in a consecutive reaction (Figure 34, top). This was probably
caused by a very small excess of HNTf2 in the reaction solution. A second CO2 hydrogenation
experiment using complex 2 was performed in the presence of 2 equivalents HNTf2. In this
experiment about 46 % of the formed methanol had reacted to DME (according to the
integral ratio in the 13C{1H}-NMR spectrum) clearly showing that DME formation was
catalysed by excess acid (Figure 34, bottom).
RESULTS & DISCUSSION
-70-
Figure 34: 13
C{1H}-NMR spectra (75 MHz, d8-THF, r.t.) of reaction solutions of CO2 hydrogenation experiments inside a high-
pressure NMR tube using complex 2 in the presence of 1 eq. HNTf2 (top) and in the presence of 2 eq. HNTf2 (bottom) (V(d8-THF) = 0.3 mL, c(Ru) = 25 mmol L
-1, p(CO2) = 10 bar at r.t., p(H2) = 30 bar at r.t., T = 140 °C, t = 24 h).
The effect of other acidic additives, namely MSA and p-toluenesulfonic acid monohydrate (p-
TsOH), on the hydrogenation of CO2 with complex 2 was investigated (Table 4, entries 8-13).
The highest TONs were observed in the presence of a slight excess (1.5 eq.) of the respective
sulfonic acid. However, both acids led to lower TONs as compared to HNTf2. The obtained
maximum TONs were well in line with the expected coordination ability of the acid anions
according to their size and charge distribution (HNTf2 < p-TsOH < MSA) corroborating the
assumption of cationic complexes of type 3 as active catalytic species. In contrast to the
NTf2- anion, the anions of p-TsOH and MSA could coordinate to the ruthenium centre and
therefore interfere with the formation of active catalytic species of type 3 or 8a.[125]
RESULTS & DISCUSSION
-71-
To prove this assumption, a CO2 hydrogenation experiment using catalyst 2 and p-TsOH
(1.5 eq.) was carried out (V(d8-THF) = 1.0 mL, c(Ru) = 25 mmol L-1, p(CO2) = 20 bar at r.t.,
p(H2) = 60 bar at r.t., T = 140 °C). After a reaction time of one hour the reaction was
terminated, and the reaction solution was analysed by NMR at 233 K. The 31P{1H}-NMR
spectrum showed the formation of 5 phosphor-containing species (Figure 35). The major
component in solution was the common formate intermediate 8a (ca. 52 % based on the
integral ratios in the 31P{1H}-NMR spectrum). The already known deactivation product 4 was
also formed in about 15 %. Interestingly, no formation of the inactive hydride dimer 5 was
observed, indicating that formation of 5 was supressed by the presence of additional ligands
in solution. Compared to the spectrum recorded of a reaction using complex 2 together with
HNTf2 (Figure 6, chapter 3.2.1) some new signals were observed:
29.3 Hz, JP-P = 19.6 Hz, 1P). The signals due to a could be correlated to a doublet of doublets
of doublets in the hydride region (δ = -5.3 ppm, JH-P = 94.6 Hz, JH-P = 19.5 Hz, JH-P = 13.6 Hz) by
a [1H,31P]-HMBC-NMR experiment, indicating that one of the three ligands in a is a hydride.
In the same fashion the signals due to b could be correlated to a pseudo doublet of triplets
RESULTS & DISCUSSION
-72-
in the hydride region (δ = -5.6 ppm, JH-P = 88.1 Hz, JH-P = 17.1 Hz). Other ligands in a and b
might be CO, p-TsOH, THF, or H2O.
In summary, these experiments demonstrated that the presence of even weakly-
coordinating anions such as p-TsO- in the reaction mixture hampered the formation of the
formate complex 8a, explaining why HNTf2 was the preferred acidic additive.
Figure 35: 31
P{1H}-NMR spectrum (243 MHz, d8-THF, -40°C) of the reaction solution of a CO2 hydrogenation reaction using
complex 2 in the presence of p-TsOH (1.5 eq.) (V(d8-THF) = 1.0 mL, c(Ru) = 25 mmol L-1
, p(CO2) = 20 bar at r.t., p(H2) = 60 bar at r.t., T = 140 °C, t = 1 h).
RESULTS & DISCUSSION
-73-
Figure 36: 31
P{1H}-NMR spectra (162 MHz, d8-THF, r.t.) of the stepwise addition of p-TsOH to complex 2 in d8-THF at room
temperature.
3.3.2 Parameter variations
In the previous chapters catalytic system 2/HNTf2 (1 eq.) was identified as the most effective
catalyst precursor for the hydrogenation of CO2 to methanol. Next, the influence of key
parameters (like catalyst concentration, H2- and CO2-pressures, reaction temperature, and
reaction time) on the TON was investigated to gain an idea of suitable starting parameters
for later catalyst-recycling and continuous-flow experiments.
First, the catalyst concentration was varied using a standard set of reaction parameters
(V(THF) = 2.08 mL, p(CO2) = 20 bar at r.t., p(H2) = 60 bar at r.t., T = 140 °C, t = 24 h). By
varying the acid concentration in parallel a stoichiometric ratio of complex 2 to HNTf2 of 1 : 1
was maintained. Reducing the concentration of 2/HNTf2 from 12 mmol L-1 to 6 mmol L-1 and
3 mmol L-1 led to a huge increase in the obtained TONs from 228 to 335 and 442,
respectively (Figure 37). Decreasing the concentration to 1.5 mmol L-1 led to a further
smaller increase of the TON to 489. This observation was in line with the formation of the
inactive dimeric complex 5 which was identified as deactivation product in chapter 3.2.1,
and corroborated that the active catalyst species contains a single ruthenium centre.
Another effect greatly contributing to the increasing TON with decreasing catalyst
concentration resulted from the pressure drop throughout the reaction. As the reactions
were performed in a closed reaction system the pressure dropped significantly due to the
RESULTS & DISCUSSION
-74-
consumption of CO2 and H2 (e.g. from ca. 120 bar at 140 °C to about 40 bar at 140 °C in the
case of the highest concentration 12 mmol L-1). At a halved catalyst concentration only half
the amount of CO2 and H2 had to be converted to MeOH to obtain the same TON, i.e. the
pressure drop at a comparable TON was significantly lower. As lower pressures significantly
slowed down the reaction (vide infra) higher TONs were observed at lower catalyst
concentrations.
Figure 37: TON as a function of the concentration of complex 2/HNTf2 (1 : 1) (V(THF) = 2.1 mL, p(CO2) = 20 bar at r.t., p(H2) = 60 bar at r.t., T = 140 °C, t = 24 h).
Decreasing the reaction temperature to 120 °C, 100 °C, and 80 °C resulted in reduced TONs
of 169, 67, and 24, respectively (Figure 38). From thermodynamics a higher equilibrium
concentration of methanol would be expected at lower temperatures, showing that the
reaction was controlled rather by kinetics than by thermodynamics in the investigated
temperature range. According to Arrhenius a bigger increase in the obtained TONs with
increasing temperature would be expected. However, as deduced from high-pressure NMR
experiments (vide supra) the formation of the main catalyst deactivation product 4 also
increased with increasing reaction temperature, limiting the positive effect of higher
reaction temperatures.
0
100
200
300
400
500
12 6 3 1,5
228
335
442 489
TON
Concentration of complex 2/HNTf2 (1 : 1) / mmol L-1
RESULTS & DISCUSSION
-75-
Figure 38: TON as a function of the reaction temperature using complex 2/HNTf2 (1 : 1) (c(Ru) = 12 mmol L-1
, V(THF) = 2.1 mL, p(CO2) = 20 bar at r.t., p(H2) = 60 bar at r.t., t = 24 h).
Pressure variation at a constant stoichiometric ratio of p(CO2)/p(H2) = 1/3 from 40 bar to
80 bar, and to 120 bar (all pressures at r.t.) resulted in a strong increase of the resulting TON
from 78 to 228, and to 367, respectively (Figure 39). Higher pressures have a positive effect
on the equilibrium concentration of methanol as liquid products are formed from gaseous
reactants. However, temperature variation showed that the reaction was controlled rather
by kinetics than by thermodynamics (vide supra). Therefore, the higher TONs most probably
resulted from higher reaction rates, which in turn resulted from higher reactant
concentrations in the liquid catalyst phase at higher pressures.
Figure 39: TON as a function of the total pressure at a constant ratio of p(CO2)/p(H2) = 1/3 using complex 2/HNTf2 (1 : 1) (c(Ru) = 12 mmol L
-1, V(THF) = 2.1 mL, p(CO2) = 20 bar at r.t., p(H2) = 60 bar at r.t., T = 140 °C, t = 24 h).
0
50
100
150
200
250
80 100 120 140
24
67
169
228
TON
reaction temperature / °C
0
100
200
300
400
40 80 120
78
228
367
TON
total pressure at r.t. / bar
RESULTS & DISCUSSION
-76-
Next, the H2 partial pressure was varied at a constant CO2 pressure of 20 bar (at r.t.) (Figure
40). Increasing the H2 pressure from 60 bar (at r.t.) to 80 bar (at r.t.) led to an increase in
TON by 32 % to 301. A further increase in H2 pressure to 100 bar (at r.t.) led to an increase in
TON by 16 % to 348. In the latter case the conversion of CO2 to methanol was about 40 %, as
determined from the initially charged amount of CO2 (22.1 mmol, determined by weight)
and the amount of methanol formed (8.7 mmol). Interestingly, increasing the CO2 pressure
at a constant H2 pressure of 60 bar did not lead to increased TONs (Figure 41). Increasing the
CO2 pressure from 20 bar to 30 bar did not affect the TON, whereas increasing the pressure
further to 40 bar led to a slight decrease of the TON by about 10 % to 204.
Figure 40: TON as a function of the H2 pressure (at r.t.) at a constant CO2 pressure of 20 bar (at r.t.) using complex 2/HNTf2 (1 : 1) (c(Ru) = 12 mmol L
-1, V(THF) = 2.1 mL, T = 140 °C, t = 24 h).
Figure 41: TON as a function of the CO2 pressure (at r.t.) at a constant H2 pressure of 60 bar (at r.t.) using complex 2/HNTf2 (1 : 1) (c(Ru) = 12 mmol L
-1, V(THF) = 2.1 mL, T = 140 °C, t = 24 h).
0
100
200
300
400
60 80 100
228
301 348
TON
H2 pressure at r.t. / bar
0
50
100
150
200
250
20 30 40
228 226 204
TON
CO2 pressure at r.t. / bar
RESULTS & DISCUSSION
-77-
From the calculated catalytic cycle (see chapter 3.2.4) it became clear that the rate
determining states were the TOF determining intermediate (TDI) 8a/V and the TOF
determining transition state (TDTS) IXc-XV(thf). Only concentrations of the reactants
involved in the steps between the TDI and the TDTS accelerate or inhibit the reaction.[156] As
no CO2 is involved in these steps but H2 is, the increase of the CO2 pressure did not show an
accelerating effect whereas the increase of the H2 pressure did. This was also in line with the
observations during high-pressure NMR experiments (see chapter 3.2.2) in which the
formation of 8a already occurred at mild reaction conditions (25 °C, 20 bar CO2, 60 bar H2)
whereas the transformation of 8a to methanol involving three equivalents of H2 needed
higher temperatures of at least 80 °C. However, this could not explain the decrease in TON
at the highest CO2 pressure of 40 bar (at r.t.). This might be explained by the phenomenon of
gas expansion of liquids by CO2.[164-165] Many properties of the liquid, like dielectric constant,
diffusion rate, viscosity, and hydrogen solubility, are changed by dissolved CO2. Diffusion
rates and H2 solubility are increased by dissolved CO2 which should lead to higher TONs.
However, Jessop et al. investigated the influence of gas pressures on the hydrogenation of
CO2 to formic acid and found that expanding the liquid MeOH/NEt3 phase by adding the
unpolar inert gas ethane led to decreased reaction rates.[165] When they repeated the same
experiment using the more polar fluoroform instead of ethane no decrease in the reaction
rate was observed. The authors concluded that ethane rendered the liquid phase less polar,
thereby causing the reaction rate to drop. Similarly, dissolving the unpolar CO2 in most
organic liquids leads to decreased Kamlet-Taft parameters π*.[164, 166] For example, methanol
had a π*-value of 0.57 (40 °C) at 1 bar CO2 pressure which decreased to 0.47 (40 °C) upon
pressurising with 41.4 bar CO2.[167] The same trend was observed for acetone showing a π*-
value of 0.70 (40 °C) at 1 bar CO2 pressure and of 0.52 (40 °C) at 41.4 bar CO2 pressure. It is
likely that this trend is also valid for THF, giving a possible explanation for the lower
observed TON of 204 at a high CO2 pressure of 40 bar (at r.t.).
Indeed, comparison of solvents with different polarities (π*-values) and basicities (ß-values)
showed a detrimental effect of lower polarities on the observed TONs: The polarities
(π*-values) of the tested solvents ranked THF > 1,4-dioxane ≈ 2-MTHF > toluene.[168] The
4.09 mol L-1). This TON was twice as high as the TON obtained after the 72 hours reaction
time without intermediate repressurisation, showing that the catalyst was indeed much
more stable under nearly isobaric reaction conditions.
The absence of an induction period indicated that the formed methanol accumulating in
solution did not enhance the reaction rate under these reaction conditions. However, this
RESULTS & DISCUSSION
-80-
observation did not exclude the possibility that the reaction partly proceeded via a cascade
reaction involving methyl formate as intermediate. In fact, it is very likely that the reaction
proceeded in parallel partly via the cascade reaction, as traces of methyl formate were
found in high-pressure NMR experiments (3.3.1) which could easily be hydrogenated to
methanol with catalyst 2/HNTf2 (3.1.1). More details on this topic are given in chapter 3.3.4.
Figure 43: TON as a function of time using complex 2/HNTf2 (1 : 1) in THF (c(Ru) = 6 mmol L-1
, V(THF) = 2.1 mL, p(CO2) = 20 bar at r.t., p(H2) = 60 bar at r.t., T = 140 °C) as obtained from batch reactions terminated at the given reaction times. (■): Batch reactions without repressurisation. (▲): The autoclave was repressurised to the initial pressure with p(CO2)/p(H2) =
1/3 after 16 h. (●): The autoclave was re-pressurised after 16 h and again after 32 h. (♦): The autoclave was re-pressurised
after 16 h, after 32 h, and after 48 h.
0 10 20 30 40 50 60 70 800
100
200
300
400
500
600
700
800
900
1000
TON
reaction time / hours
RESULTS & DISCUSSION
-81-
Figure 44: 31
P{1H}-NMR spectra of the reaction solution of three different CO2 hydrogenation reactions using complex
2/HNTf2 (1 : 1) in d8-THF which were terminated after different reaction times without intermediate repressurisation
(p(CO2) = 20 bar at r.t., p(H2) = 60 bar at r.t., T = 140 °C). 4 = [Ru(H)(CO)2(Triphos)]+, 5 = [Ru2(-H)2(Triphos)2], 6 =
20 bar at r.t., p(H2) = 60 bar at r.t., T = 140 °C, t = 24 h).
§ This complex was synthesised by M. Meuresch, ITMC, RWTH Aachen University, Aachen.
** The value given here (393) is higher than the value stated in chapter 3.3.2 for the same reaction conditions.
Here, a hot plate stirrer equipped with an aluminium cone was used for heating of the autoclave instead of an oil bath, allowing faster heating to reaction temperature and more efficient heat transfer. To retain comparability, only experiments were compared throughout this thesis, which were performed using the same heating system. For clarification, it was always indicated if an aluminium cone was used.
RESULTS & DISCUSSION
-85-
3.3.4 Comparison of the reaction in presence of alcohol additive with the reaction in
absence of alcohol additive
In this chapter, the question if the observed formation of methanol proceeded via a cascade
reaction via alkyl formate intermediates or via direct CO2 hydrogenation within the
coordination sphere of the Ru-Triphos fragment is discussed.
There were several hints that the reaction proceeded partly via a cascade reaction as soon
as alcohol was present in the reaction solution: In the initial studies using catalyst 2/acid in a
mixture of ethanol (0.58 mL) and THF (1.5 mL) as solvent, clearly the formation of ethyl
formate was observed in small amounts besides traces of methyl formate (observable only
by extensive zooming into the formate region of the 1H-NMR spectrum) and considerably
formation of methanol (chapter 3.1.2). Moreover, catalyst 2/acid was shown to be an
efficient catalyst for the hydrogenation of methyl formate and ethyl formate to methanol
(chapter 3.1.1) making it very likely that part of the observed methanol in the CO2
hydrogenation reaction was generated via the observed alkyl formate as intermediate.
In the reactions performed in absence of alcohol additive methyl formate could be observed
in trace amounts by extensive zooming into the formate region of the 1H-NMR spectrum
(singlet at around 8.1 ppm), besides even smaller traces of formic acid (singlet at around
8.2 ppm) (Figure 46). The amounts of methyl formate were too small to be quantified by GC.
However, a rough estimation based on integration of the signal in the 1H-NMR spectrum
showed around 0.2 % selectivity (TON = 0.4) to methyl formate under standard reaction
conditions besides formation of methanol with a TON of 228 (complex 2/HNTf2, V(THF) =
2.1 mL, c(Ru) = 12 mmol L-1, HNTf2 (1 eq.), p(CO)2 = 20 bar at r.t., p(H2) = 60 bar at r.t., T =
140 °C, t = 24 h). At a lower temperature of 80 °C the formation of methanol was strongly
decreased to a TON of 24, and the TON found for methyl formate increased to roughly 1.7,
corresponding to a methyl formate selectivity of around 7 %. This indicated that the
formation of methyl formate proceeded already at lower temperatures,[7] whereas the
further hydrogenation of methyl formate to methanol and the direct transformation of CO2
to methanol within the coordination sphere of the catalyst was efficiently achieved only at
higher temperatures. Methyl formate did not accumulate in the reaction solution
throughout the reaction, as judged from the integrals of methyl formate in the 1H-NMR
spectra of reactions terminated after different reaction times. These observations strongly
RESULTS & DISCUSSION
-86-
suggested that the formation of methanol proceeded partly via methyl formate as
intermediate as soon as methanol was present in the reaction solution. A reaction sequence
involving CO as intermediate (e.g. formation of methyl formate decarbonylation of
methyl formate formation of methanol) was excluded as the presence of CO-gas in the
reaction mixture led to the exclusive formation of the inactive carbonyl complex 4 as judged
from the characteristic signals in the 31P{1H}-NMR spectrum and no methanol formation was
observed in this case (complex 2/HNTf2, c(Ru) = 25 mmol L-1, HNTf2 (1 eq.), V(d8-THF) = 1 mL,
p(CO)2 = 20 bar at r.t., p(CO) ≈ 1 bar at r.t., p(H2) = 60 bar at r.t., T = 140 °C, t = 24 h).
Consequently, an experiment employing only synthesis gas (CO/H2) also showed the
exclusive formation of 4 and no methanol formation (2/HNTf2, V(THF) = 2.0 mL, c(Ru) =
12.5 mmol L-1, HNTf2 (1 eq.), p(CO) = 10 bar at r.t., p(H2) = 30 bar at r.t., T = 140 °C, t = 24 h).
Figure 46: Formate region of the 1H-NMR spectrum (300 MHz, d6-dmso, r.t.) of a reaction solution of a CO2 hydrogenation
reaction performed at standard reaction conditions in the absence of alcohol additive (complex 2/HNTf2, V(d8-THF) = 2.1 mL, c(Ru) = 12 mmol L
-1, HNTf2 (1 eq.), p(CO)2 = 20 bar at r.t., p(H2) = 60 bar at r.t., T = 140 °C, t = 24 h).
There were also several hints that the formation of methanol proceeded directly by
transformation of CO2 within the coordination sphere of the Ru-Triphos fragment: In a high-
pressure NMR experiment it was shown that formate complex 8a (which could be generated
in situ from CO2/H2 or by the addition of HNTf2/HCO2H to complex 2) could be converted to
methanol in the presence of only H2 and in the absence of any alcohol additive (chapter
RESULTS & DISCUSSION
-87-
3.2.2). Moreover, the TON/time curve of the reaction in the absence of alcohol additive did
not show a pronounced induction period or an autocatalytic effect (chapter 3.3.2, Figure 43),
showing that CO2 could be directly converted to methanol in the absence of alcohol and that
the presence of alcohol had no enhancing effect on the rate of methanol formation.
Consequently, the TON obtained in a CO2 hydrogenation reaction in pure THF (V = 2.1 mL)
(TON = 228) was as high as the TON obtained in the presence of alcohol additive (V(THF) =
1.5 mL, V(EtOH) = 0.58 mL) (TON = 221) under otherwise same conditions (complex 2/HNTf2,
HNTf2 (1 eq.), c(Ru) = 12.5 mmol L-1, p(CO2) = 20 bar at r.t., p(H2) = 60 bar at r.t., T = 140 °C, t
= 24 h). DFT calculations further supported that a direct catalytic transformation of CO2 to
methanol can take place directly within the coordination sphere of the Ru-Triphos fragment
(chapter 3.2.4).
In summary, these results showed that CO2 was directly converted to methanol at the Ru-
centre. As soon as methanol was present in the reaction solution, formation of methyl
formate took place in parallel, and the methyl formate could subsequently be hydrogenated
to methanol. However, the absence of a rate enhancing effect of alcohol indicated that the
pathway via methyl formate was insignificant compared to the direct CO2 transformation.
3.3.5 Test of different catalyst precursors
All previous experiments and DFT calculations indicated that cationic species of the type
[Ru(H)(H2)(Triphos)(S)]+ (3) play an important role in the catalytic transformation.
Consequently, neutral catalyst precursors like [Ru(TMM)(Triphos)] (2) and
[Ru(H)2CO(Triphos)] (17) were found to be active in the presence of 1 eq. of acid, the
stoichiometric amount needed for the protonation of the TMM ligand or a hydride ligand.
However, 17 was found to be less productive than 2 due to the detrimental effect of CO in
the reaction solution (chapter 3.3.1). The cationic catalyst precursor [Ru(2-
OAc)(Triphos)(S)]NTf2 (14) was found to be active in the absence of acid, however, the
obtained TON was lower compared to the TON obtained with 2/HNTf2 due to the more
difficult activation by reductive removal of the acetate ligand (chapter 3.3.1).
Other catalyst precursors were evaluated and compared to the catalytic system
complex 2/HNTf2, which gave a TON of 393 (Table 5, entry 1). First, [Ru(OC6F5)2(Triphos)]
RESULTS & DISCUSSION
-88-
(22)†† was tested for its performance in the CO2 hydrogenation to methanol. In the absence
of acid a TON of 14 was obtained, whereas in the presence of 1 equivalent HNTf2 a TON of
403 was obtained, which was comparable to the TON obtained with complex 2/HNTf2 (Table
5, entries 2 & 3). The need for HNTf2 indicated that complex 22 had to be activated by
protonation of a -OC6F5-ligand to give a cationic complex. Using [Ru(OC6F2H3)2(Triphos)]
(23)††/HNTf2 (1 : 1) again gave a very similar TON of 407 (Table 5, entry 4), indicating that all
three precursors gave the same active catalyst species under reaction conditions.
Next, Ru-TMM complexes containing different Triphos-derivatives were compared. As
mentioned earlier, the complex [Ru(TMM)(Triphos-xylyl)] (21)‡‡/HNTf2 (1 : 1) did not show
an increased TON (385) (Table 5, entry 5). [Ru(TMM)(Triphos-tolyl)] (24)‡‡ (Triphos-tolyl =
1,1,1-tris(bis(3-methylphenyl)phosphinomethyl)ethane) gave an about 7 % increased TON of
419 in the presence of HNTf2 (1 eq.) (Table 5, entry 6), demonstrating that steric and/or
electronic modifications of the Triphos ligand can indeed lead to improved productivities.
Nevertheless, all subsequent experiments were carried out using complex 2/HNTf2 due to its
easier accessibility from commercially available starting materials.
Table 5: Hydrogenation of CO2 to methanol in the absence of alcohol additive: test of different catalyst precursors.[a]
Entry Cat. Acid (eq.)[b]
pH2[c]
[bar]
pCO2[c]
[bar]
TON[d]
1 2 HNTf2 (1.0) 60 20 393
2 22 ‒ 60 20 14
3 22 HNTf2 (1.0) 60 20 403
4 23 HNTf2 (1.0) 60 20 407
5 21 HNTf2 (1.0) 60 20 385
6 24 HNTf2 (1.0) 60 20 419
[a] Reaction conditions: in all reactions an aluminium cone was used for heating instead of an oil bath; catalyst (12.5 µmol),
THF (2.1 mL), 140 °C, 24 h; [b] equivalents to catalyst; [c] at room temperature; [d] TON = mmol MeOH/mmol catalyst.
3.4 Catalyst recycling and immobilisation
All NMR studies and the DFT studies were consistent with formate complex 3 being the
resting state of the active catalyst species. This chapter deals with recycling of the catalyst in
its resting state in repetitive batch experiments as well as with the development of a
continuous-flow process.
††
Synthesised by Dominik Limper, ITMC, RWTH Aachen University, Aachen. ‡‡
This complex was synthesised by M. Meuresch, ITMC, RWTH Aachen University, Aachen.
RESULTS & DISCUSSION
-89-
3.4.1 Catalyst recycling by distillation
An obvious possibility to recycle a homogeneous catalyst is separation of the volatile organic
solvents and products by distillation.[24] In continuous operation normally a part of the
bottom product of a column containing the concentrated catalyst is recycled back to the
reactor. However, in the case of the CO2 hydrogenation to methanol in THF the reaction
mixture contains THF, methanol, and water. For example, the reaction solution of the
reaction with the highest TON of 895 consisted of 76.8 wt.-% THF, 14.9 wt.-% MeOH, and
8.4 wt.-% H2O. The ternary system THF/MeOH/H2O forms two binary minimum azeotropes
under atmospheric pressure.[172] The THF/MeOH azeotrope contains 31 wt.-% methanol and
has a boiling point at 60.7 °C, the THF/H2O azeotrope contains 5 wt.-% H2O and has a boiling
point at 64.0 °C. Therefore, distillation of the reaction mixture would yield water as the
bottom product,[172] which is not suitable for recycling of the Ru-Triphos catalyst as it is
insoluble in water. Additionally, separation of the THF/MeOH/H2O mixture is tedious and
= 20 bar at r.t., p(H2) = 60 bar at r.t., T = 140 °C, t = 24 h), the reactor was cooled to 0 °C in an
ice bath, the remaining pressure released, and the reaction solution transferred to a
distillation apparatus via cannula under argon atmosphere. All volatiles were removed from
the reaction solution under reduced pressure at room temperature, and the condensate was
collected in a cooling trap, weighed, and analysed for its methanol content. The remaining
dry catalyst was redissolved in ethanol (0.58 mL)/THF (1.5 mL) and recycled to the reactor.
No new HNTf2 was added as all previous investigations showed that HNTf2 was only
necessary for initial catalyst activation. A batch reaction was performed under the same
conditions as before, and the catalyst was recycled a second time. In Figure 47 the TONs
obtained in each cycle as well as the summed up total TON are displayed. Already the TON
obtained in the first cycle (134) was much lower than the TON obtained in the analogous
batch reaction with direct analysis of the MeOH in the reaction solution (TON = 221). This
was due to incomplete transfer of the reaction solution to the distillation apparatus and due
to losses inside the distillation apparatus. In the second cycle, already a huge drop in
RESULTS & DISCUSSION
-90-
productivity to a TON of 44 was observed. In the third cycle the TON dropped further to 28.
This could partly be explained by catalyst losses during the transfer from the reactor to the
distillation apparatus and back. Moreover, from earlier attempts to isolate the active
formate intermediate 8a it was known that 8a is not stable in solid form and under reduced
pressures. Therefore, catalyst deactivation under these recycling conditions is very likely.
These results showed that the catalytic system cannot be recycled in its dry form. However,
the system still showed a remarkable stability, as in the second cycle still one third of the
initial productivity was obtained. Next, catalyst recycling in liquid solution was attempted.
Figure 47: Recycling of catalyst 2/HNTf2 (1 : 1) by removing all volatiles from the catalyst in vacuo at r.t. after a batch reaction (cycle 1), redissolving the remaining solid catalyst in ethanol/THF, and performing the next cycle (cycle 2). (c(Ru) = 12 mmol L
-1, HNTf2 (1 eq.), V(THF) = 1.5 mL, V(EtOH) = 0.58 mL, p(CO2) = 20 bar at r.t., p(H2) = 60 bar at r.t., T = 140 °C, t =
24 h).The TONs obtained per cycle are shown in dark grey, the total TONs summing up the cycles are shown in light grey.
3.4.2 Catalyst recycling in the biphasic system 2-MTHF/H2O
As shown in chapter 3.3.2, the CO2 hydrogenation to methanol could also be carried out in 2-
MTHF instead of THF, albeit with a somewhat lower TON of 186 instead of 228. Moreover,
the catalytic system was shown to be stable towards water (chapter 3.3.3). As 2-MTHF has a
miscibility gap with water, this opened up the possibility to realise a biphasic system, where
the catalyst is retained and recycled in an organic phase and methanol is removed in an
aqueous phase.[126, 174-175] The use of 2-MTHF as organic phase is desirable as 2-MTHF is
considered to be eco-friendly because it can be produced from renewable resources and is
degraded by sunlight and air.[174] Water is suitable as it has a miscibility gap with 2-MTHF and
is a by-product of the CO2 hydrogenation reaction.
0
50
100
150
200
250
1 2 3
134
44 28
134
178
206
TON
Cycle
cycle TON
total TON
RESULTS & DISCUSSION
-91-
The general partitioning of methanol in a biphasic 2-MTHF/H2O mixture was assessed by
dissolving methanol (0.158 g, 4.94 mmol, ca. 0.2 mL) in 2-MTHF (0.892 g, ca. 1.0 mL) and
adding deionised water (0.996 g, ca. 1.0 mL). After shaking the mixture for about 2 minutes,
the two clear phases were separated and analysed for their contents via 1H-NMR
(3.89 mmol, 124.7 mg), which was about 79 % of the total methanol, besides 13.4 wt.-%
2-MTHF (1.99 mmol, 171.8 mg) and 76.9 wt.-% H2O. The organic phase (0.765 g) contained
4.1 wt.-% methanol (0.98 mmol, 31.3 mg), which was about 20 % of the total methanol,
besides 8.5 wt.-% H2O and 87.4 wt.-% 2-MTHF. The mass balance based on methanol was
closed to 99 %. These results indicated that H2O was well suitable for extraction of methanol
from 2-MTHF for catalyst recycling purposes.
Recycling of catalyst 2/HNTf2 (1 : 1) using this extraction method in repetitive batch mode
was assessed as follows (Figure 48): A batch reaction was performed in 2-MTHF and
terminated after 16 hours by cooling in an ice/water bath and subsequent venting of the
reactor to release the remaining pressure (c(Ru) = 6.3 mmol L-1, HNTf2 (1 eq.), V(2-MTHF) =
2.0 mL, p(CO2) = 20 bar at r.t., p(H2) = 60 bar at r.t., T = 140 °C). The clear, orange reaction
solution was transferred to a Schlenk-tube, water (2.0 mL) was added, and the mixture was
stirred for about 20 seconds. After 10 minutes the phases were separated. The colourless
water phase was weighed and analysed for its contents by 1H-NMR spectroscopy (D1 = 10 s)
using mesitylene as internal standard in d6-acetone, and only the methanol content of the
aqueous phase was used for the calculation of the apparent TON. The orange 2-MTHF phase
was transferred back to the reactor and fresh 2-MTHF (0.25 mL) was added to compensate
the loss of 2-MTHF with the aqueous phase. No new HNTf2 was added as all previous
experiments indicated that the acid was only needed for initial formation of the active
catalyst species. The second, third, and fourth cycle were performed in the same way.
RESULTS & DISCUSSION
-92-
Figure 48: Procedure for repetitive batch experiments using catalyst 2/HNTf2 (1 : 1) in 2-MTHF and H2O to extract the methanol product.
The obtained TONs per cycle as well as the summed up total TONs are shown in Figure 49,
and the compositions of the aqueous phases from the extractions as well as of the 2-MTHF
phase after the last extraction in the fourth cycle are given in Table 6. Catalytic system
2/HNTf2 (1 : 1) could be recycled three times, giving a total TON of 769 after 4 cycles. After
the fourth cycle also the 2-MTHF phase was analysed by 1H-NMR spectroscopy for its
composition, and a methanol content corresponding to a TON of 12 was detected.
Moreover, washing of the autoclave with d6-DMSO (2.0 mL) gave an additional amount of
methanol corresponding to a TON of 25. In summary, a TON of 806 was obtained
corresponding to 323 mg methanol after a total reaction time of 64 hours. In the first cycle a
TON of 247 was obtained, which was around 84 % of the TON obtained in a batch reaction
conducted under the same reaction conditions but with direct analysis of the methanol
content in the 2-MTHF phase. This indicated that around 84 % of the produced methanol
were extracted by the water phase, which was in line with the observed partition of
methanol between water and 2-MTHF (vide supra). In cycle 2 still around 90 % of the initial
productivity was remained (TON = 222). This value decreased to 77 % in cycle 3 (TON = 191)
and to 45 % in cycle 4 (TON = 110). These results indicated that in principle the recycling of
complex 2/HNTf2 (1 : 1) in its active form was possible with the biphasic system
2-MTHF/H2O.
RESULTS & DISCUSSION
-93-
Figure 49: Recycling of catalyst 2/HNTf2 using 2-MTHF (2.0 mL) as the solvent and water (2.0 mL) for extraction of the produced methanol as shown in Figure 48. Each cycle was run for 16 hours (p(CO2) = 20 bar at r.t., p(H2) = 60 bar at r.t., T = 140 °C). The TONs obtained per cycle are shown in dark grey, the total TONs summing up the cycles are shown in light grey.
Table 6: Compositions of the phases obtained from a repetitive batch experiment using catalyst 2/HNTf2 in 2-MTHF as the solvent and water for extraction of the produced methanol as shown in Figure 48.
Mass [g] MeOH [wt.-%] 2-MTHF [wt.-%] H2O [wt.-%]
H2O phase cycle 1 2.222 4.5 13.4 82.1
H2O phase cycle 2 2.419 3.7 13.2 83.1
H2O phase cycle 3 2.422 3.2 13.9 83.0
H2O phase cycle 4 2.421 1.8 13.9 84.3
2-MTHF phase cycle 4 0.716 0.7 89.7 9.6
Using the system 2-MTHF/H2O a process scheme for continuous-flow operation is
conceivable in which all material streams can be recycled internally (Figure 50). However, as
shown in Table 6, a typical composition of the aqueous extraction phase was 4.5 wt.-%
MeOH, 13.4 wt.-% 2-MTHF, and 82.1 wt.-% H2O because of the solubility of 2-MTHF in water
of about 14 % at room temperature. The 2-MTHF content could be reduced to about 7 % by
performing the extraction at 60 °C,[175] but still tedious downstream processing would be
needed due to the formation of the azeotropes 2-MTHF/H2O and 2-MTHF/MeOH. Therefore,
intrinsic recycling of the catalyst by immobilisation inside the reactor was envisaged.
0
100
200
300
400
500
600
700
800
1 2 3 4
247 222 191
110
247
469
660
769
TON
Cycle
cycle TON
total TON
RESULTS & DISCUSSION
-94-
Figure 50: Process scheme for the CO2 hydrogenation to methanol using the system 2-MTHF/H2O for catalyst recycling.
3.4.3 Catalyst immobilisation for continuous-flow application
3.4.3.1 Identification of a suitable immobilisation medium
Ionic liquids
A proven concept for immobilising an organometallic catalyst in liquid phase inside a reactor
is the immobilisation in a non-volatile ionic liquid (IL).[24, 27, 29, 176-177] As the previous
experimental and theoretical mechanistic studies indicated ionic species of the type
[Ru(H)(H2)(Triphos)(S)]NTf2 (3) as active species and [Ru(2-O2CH)(Triphos)(THF)]NTf2 (8a) as
the resting state under reaction conditions, the catalytic system seemed to be perfectly
suitable for application in ionic liquids comprising the -NTf2-anion.
First experiments were carried out using the ionic liquid 1-ethyl-3-methylimidazolium
bis(trifluoromethylsulfonyl)imide ([EMIM][NTf2]) in high purity (HP) grade (Iolitec,
-1, HNTf2 (1 eq.), V(IL) = 2.0 mL, p(CO2) = 20 bar at r.t., p(H2) = 60 bar
at r.t., T = 140 °C, t = 1.5 h). Bottom: spectrum of the reaction solution with addition of d8-THF. Top: spectrum of the reaction solution with addition of d8-THF and acetonitrile (76 eq.). 4 = [Ru(H)(CO)2(Triphos)]
+; 5 = “Hydride-Dimer”; 8 =
[Ru(2-O2CH)(Triphos)(S)]
+; 9 = [Ru(H)(MeCN)2(Triphos)]
+; 10 = [Ru(MeCN)3(Triphos)]
2+.
In the course of his Master’s Thesis, Klügge used the same halide-free IL to gain more insight
into possible reasons for catalyst inhibition in the ILs used before, which gave poor TONs and
reproducibility.[178] One possible impurity in the ILs used before might have been traces of
N-methylimidazole from IL synthesis. Performing a reaction using complex 2/HNTf2 (1 : 1)
under standard conditions but in the presence of N-methylimidazole (25 eq. based on [Ru])
led to a complete deactivation of the catalyst, and no formation of methanol was
observed.[178] This indicated that basic impurities from the starting materials indeed are
detrimental to the catalyst activity and should be removed from the IL. Klügge synthesised
two batches of the structural similar IL [BMIM][NTf2] via a halide-free synthesis route
starting from N-methylimidazol and butyl-methanesulfonate.[178, 181] NMR-spectroscopy did
not show any organic impurities in both batches. Interestingly, performing a CO2
hydrogenation reaction in the IL from one batch gave a high TON of 186 comparable to the
TON obtained using the commercially available halide-free [EMIM][NTf2] (Table 7, entries 15-
RESULTS & DISCUSSION
-102-
21), whereas the same reaction in the IL from the other batch gave only a TON of 7
comparable to the TON in HP grade [EMIM][NTf2] (Table 7, entries 1-13). These results
stressed the fact that other impurities than halides were responsible for catalyst inhibition.
In summary, these results showed the possibility to perform the CO2 hydrogenation to
methanol in ionic liquids. Using the commercially available “super ultra pure” ionic liquid
liquid was used to explore the hydrogenation of CO2 to methanol in continuous-flow.
For continuous flow experiments the reaction rig described in chapter 5.4.3 was used (Figure
59). This setup was originally planned by Jens Theuerkauf and built by the mechanical
workshop of the ITMC, RWTH Aachen University.[185] Some modifications were made: The
dead volume of the system was reduced by exchanging the four two-way ball valves for
switching between “bypass” and “reactor”[185] with two three-way ball valves (TW 1 and
TW 2). To avoid condensation of extracted methanol and water inside the capillary between
the reactor outlet and the back pressure regulator this capillary was heated using heating
tape controlled by a PID controller (Eurotherm, model 91e, TIRC 1). A modified version of the
reactor used in all previous batch experiments served as reactor in the continuous-flow
setup. The reaction solution was stirred and heated using the same hot plate stirrer
equipped with an aluminium cone as was used for the well reproducible batch reactions in IL
(Table 7, entries 18-23). For an efficient gas transport into the IL phase and for efficient
stripping of the reaction products from the IL phase at reaction conditions (140 °C) the
reaction gases were introduced via a dip tube which immerged 0.5 cm into the reaction
solution when the glass liner was filled with 2 mL of IL. The depressurised product stream
was passed through a cooling trap (CT) at -72 °C filled with glass beads and THF. The cooling
RESULTS & DISCUSSION
-105-
trap was periodically exchanged and analysed for its methanol content by NMR and GC. A
sketch showing the principle of the continuous-flow setup is shown in Figure 53.
Figure 53: Sketch of the continuous-flow reaction setup.
In first experiments the applicability of the reaction setup for extraction of produced
methanol from the ionic liquid [EMIM][NTf2] was evaluated. Therefore, two independent
experiments were performed as follows: Methanol (400 mg) and [EMIM][NTf2] (2.0 mL) were
transferred to the reactor, the reactor was closed, pressurised with CO2/H2 (p(CO2)/p(H2) ≈
1/3) to 120 bar, and mounted in the reaction rig. Extraction was started at 140 °C by passing
the gas stream ( (H2) = 60 mLN min-1, (CO2) = 20 mLN min-1) through the reactor at 120 bar.
At these conditions, the density of the CO2/H2 phase was roughly one-tenth of the critical
density of CO2 (0.468 g cm-3),[186] indicating a low solubility strength of the gas phase.
Therefore, the extraction of the reaction products from the reactor will be mainly based on
product volatility. The cooling traps were periodically changed and the content was analysed
by GC using heptane as standard. Based on these results the evolution of the mass balance
(MeOH extracted [g]/MeOH initially charged [g]) with time was calculated (Figure 54, red
dots). Only a short induction period of 1 hour with an apparent average extraction rate of
11.2 mg h-1 was observed, showing a fast response time of the reaction setup. Between hour
1 and hour 5.8 an average extraction speed of 38.8 mg h-1 was obtained. After these 5.8
hours already 53 % of the initially charged MeOH had been extracted. With decreasing
amounts of MeOH remaining in the reactor, the extraction speed strongly decreased to an
RESULTS & DISCUSSION
-106-
average of 5.7 mg h-1 between hour 5.8 and 20.8 and further to 1.5 mg h-1 between hour 24
and 29. In the end the mass balance could be closed to about 80 %. The incomplete mass
balance was most probably due to losses of MeOH with the exhaust gases, especially when
the cooling trap remained unchanged for longer periods (over night). However, the same
continuous-flow extraction experiment was repeated and gave very similar results,
indicating a very good reproducibility using this setup (Figure 54, black squares). Therefore,
this setup was used to explore the hydrogenation of CO2 to methanol in continuous-flow.
Figure 54: Mass balance (MeOH extracted [g]/MeOH initially charged [g]) versus time of a continuous-flow extraction of
MeOH (400 mg) from [EMIM][NTf2] (2.0 mL) (T(reactor) = 140 °C, T(capillary) = 140 °C, p = 120 bar, (H2) = 60 mLN min-1
,
(CO2) = 20 mLN min-1
, stirring speed = 500 rpm). The red dots and the black squares indicate independent runs.
A CO2 hydrogenation using catalyst 2/HNTf2 (1 : 1) in [EMIM][NTf2] (c(Ru) = 12.5 mmol L-1,
HNTf2 (1 eq.), V(IL) = 2.0 mL) was performed in continuous-flow mode using the same setup
and the same reaction parameters as were applied in the extraction experiment described
before. The evolution of the summed up total TON (TTON) with the time on stream is
displayed in Figure 55. Average TOF values were calculated for the indicated periods.
After a short induction period of around 1 hour with an apparent TOF of only 2.9 h-1,
continuous methanol formation proceeded with an average TOF of 15.5 h-1 between hour 1
and 5, giving a TTON of 65 after 5 hours. From hour 5 to hour 19, methanol formation
continued smoothly with an average TOF of 13.4 h-1 giving a TTON of 253 after 19 hours on
stream. During the period from hour 19 to hour 29 the average TOF decreased to 7.3 h-1,
0 10 20 30 40 50 600
10
20
30
40
50
60
70
80
90
100
average extraction speed = 1.5 mg/h
ma
ss
ba
lan
ce
/ w
t.-%
time on stream / hours
total mass balance = 80 wt.-%
average extraction speed = 38.8 mg/h
RESULTS & DISCUSSION
-107-
indicating increasing catalyst deactivation after around 24 hours. Methanol formation
continued with a strongly decreased average TOF of 1.2 for the next 22 hours (hour 29-51)
and nearly stopped after 67 hours on stream (TOF = 0.2 h-1). After this time, a total TON of
365 was obtained. This experiment proved the feasibility to hydrogenate CO2 to methanol in
continuous-flow using a homogeneous catalyst for the first time.
Figure 55: Total TON (TTON) versus reaction time of a continuous-flow CO2 hydrogenation using complex 2/HNTf2 (1 : 1) in [EMIM][NTf2] (V(IL) = 2.0 mL, c(Ru) = 12.5 mmol L
-1, HNTf2 (1 eq.), T(reactor) = 140 °C, T(capillary) = 140 °C, p = 120 bar,
(H2) = 60 mLN min-1
, (CO2) = 20 mLN min-1
, stirring speed = 500 rpm).
As batch reactions with varied reaction pressures showed that much higher TONs could be
obtained at increased pressures, the next continuous-flow experiment was performed using
an increased pressure of 200 bar (Figure 56). At these conditions, the density of the CO2/H2
gas mixture was still one order of magnitude lower as compared to the critical density of
CO2.[186] Again, an induction period of 1 hour was observed with an apparent TOF of only
4.0 h-1. In the next four hours, methanol production was observed with an average TOF of
21.3 h-1 (hour 1-5). The TOF increased even further to an average value of around 30 h-1 for
the next 24 hours (hour 5-29). Maybe this further apparent increase was due to an increase
of the methanol concentration in the IL phase leading to a more efficient stripping of
methanol from the IL phase. After 29 hours on stream a total TON of 657 was obtained.
After this time, catalyst deactivation became apparent, and the average TOF between hour
29 and hour 44 decreased to 12.1 h-1 and further to an average TOF of 7.5 h-1 between hour
44 and 52. After 52 hours on stream a total TON of 1062 was obtained, which was the
0 10 20 30 40 50 60 70 800
50
100
150
200
250
300
350
400
average TOF = 0.2 h-1 (hour 66-67)
average TOF = 1.2 h-1 (hour 29-51)
average TOF = 7.3 h-1 (hour 19-29)
TTON
time on stream / hours
TTON = 365
average TOF = 15.5 h-1 (hour 1-5)
RESULTS & DISCUSSION
-108-
highest TON observed so far. Clearly, this was due to the high pressure of 200 bar and
isobaric conditions (high pressures and isobaric conditions led to strongly increased TONs
also in batch experiments, chapter 3.3.2). Compared to the continuous-flow experiment at
120 bar nearly doubled TOFs were observed. Whereas in the experiment at 120 bar catalyst
deactivation became apparent already in the period between hour 19 and hour 29, catalyst
deactivation became apparent not until the period between hour 29 and hour 44 in the
experiment at 200 bar. However, catalyst deactivation was still a problem. After the
continuous-flow experiment was stopped, the reactor was dismounted under remaining
pressure, cooled to room temperature, the pressure carefully released, and the clear, yellow
catalyst solution was transferred to a NMR tube under inert atmosphere. The 31P{1H}-NMR
spectrum showed the formation of the inactive carbonyl complex 4 in about 93 % (according
to the integral ratios in the spectrum) besides some phosphor containing species giving rise
to singlets at 42.5 ppm, 42.2 ppm, and 35.9 ppm. Comparison with the NMR spectrum of a
reaction solution of a batch reaction using the same catalyst in the same IL (Figure 52)
suggested that the singlet at 42.5 ppm was due to an active formate complex of type 8 and
the signal at 42.2 ppm was due to the hydride dimer 5. This indicated that catalyst
deactivation by formation of carbonyl complex 4 is the major challenge to be tackled in the
future.
In summary, it was demonstrated that the continuous-flow hydrogenation of CO2 to
methanol is possible by immobilising an organometallic catalyst in a high-boiling reaction
medium (estimated boiling point of [EMIM][NTf2] = 544 °C[187]), and by stripping the reaction
products from the reaction medium by excess reaction gases.
RESULTS & DISCUSSION
-109-
Figure 56: Total TON (TTON) versus reaction time of a continuous-flow CO2 hydrogenation using complex 2/HNTf2 (1 : 1) in [EMIM][NTf2] at an increased pressure of 200 bar (V(IL) = 2.0 mL, c(Ru) = 12.5 mmol L
-1, HNTf2 (1 eq.), T(reactor) = 140 °C,
T(capillary) = 140 °C, p = 200 bar, (H2) = 60 mLN min-1
, (CO2) = 20 mLN min-1
, stirring speed = 500 rpm).
Figure 57: 31
P{1H}-NMR spectrum (162 MHz, CD2Cl2, r.t.) of the reaction solution after stopping the continuous-flow
experiment using complex 2/HNTf2 (1 : 1) in [EMIM][NTf2] at an increased pressure of 200 bar (V(IL) = 2.0 mL, c(Ru) =
12.5 mmol L-1
, HNTf2 (1 eq.), T(reactor) = 140 °C, T(capillary) = 140 °C, p = 200 bar, (H2) = 60 mLN min-1
3.4.3.3 Continuous-flow hydrogenation of CO2 to methanol using SILP catalysis
After demonstrating that continuous-flow hydrogenation of CO2 to methanol is in principle
possible using complex 2/HNTf2 (1 : 1) in a bulk phase of the ionic liquid [EMIM][NTf2], the
possibility to immobilise the same combination of catalytic system and ionic liquid on a
porous support was examined. This concept, known as Supported Ionic Liquid Phase
(SILP),[188-189] was successfully applied for e.g. the aerobic oxidation of alcohols,[190] the
hydroformylation of 1-octene,[28] and the enantioselective hydrogenation of C=C bonds.[25-26]
As water is produced as byproduct in the CO2 hydrogenation to methanol, a hydrophobic
support material was chosen to avoid accumulation of water in the support material.
Perfluoro-alkyl [-Si(Me)2CH2CH2C6F13] functionalised silica (particle size 32-63 μm, mean pore
diameter 59 Å, BET surface area 207 m2 g-1, mesopore volume 0.8 mL g-1, SGFLUO) has been
proven to be suitable for this purpose by Hintermair and co-workers.[25, 191] For the synthesis
of the SILP catalyst a pore-filling degree of α = 0.5 was chosen (2.5 g of support per mL of IL),
as studies by Hintermair showed that this was the highest possible value at which the
catalyst was still a macroscopically dry powder and showed good mass transport
behaviour.[23] For better comparison with the bulk IL system the same catalyst concentration
in IL was chosen (c(Ru) = 12.5 μmol mL-1). The synthesis of the SILP catalyst based on this
support material, [EMIM][NTf2], and complex 2/HNTf2 (1 : 1) was performed following
literature-known procedures and is described in detail in chapter 5.5.6.[25-26, 28] However, THF
was used as the solvent for this procedure instead of the commonly used DCM, as
experience from catalyst synthesis and studies indicated that complex 2/HNTf2 (1 : 1) was
much more stable in the coordinating THF compared to weakly coordinating DCM. After
drying the SILP catalyst for 1 hour at room temperature in vacuo a pale yellow, dry powder
resulted.
To test this catalyst powder in the continuous-flow setup, a stainless steel tubular reactor
(inner diameter = 0.75 cm, length = 22.5 cm) equipped with two ball valves for closing was
used instead of the stirred batch reactor. The SILP catalyst (4 g SILP catalyst, containing
12.5 μmol of [Ru], 1 eq. HNTf2, and 1.0 mL [EMIM][NTf2]) was inserted into the tubular
reactor inside a glovebox giving a packed bed of 11 cm height, the reactor was mounted in
the reaction rig, and the reaction was started by heating the reactor at 140 °C and passing
the CO2/H2 flow through the reactor (p = 120 bar, (H2) = 60 mLN min-1, (CO2) = 20
RESULTS & DISCUSSION
-111-
mLN min-1). The evolution of the total TON with the reaction time is displayed in Figure 58.
After a short induction period of 1 hour with an apparent TOF of 5.7 h-1, methanol
production was observed with an average TOF of 15.8 h-1 for the next 5 hours. After 6 hours
on stream a TTON of 85 was obtained. However, catalyst deactivation was fast and the
average TOF decreased to 1.4 h-1 already in the period from hour 6 to hour 21, to 2.4 h-1 in
the period from hour 21 to hour 29, and finally to 0.5 h-1 in the period from hour 45 to hour
50.5. After 50.5 hours on stream a TTON of 130 was achieved. In comparison to the
experiment in bulk IL using a stirred tank reactor at 120 bar (Figure 55) nearly the same
average TOF around 15.5 h-1 was observed up to 5-6 hours on stream. However, catalyst
deactivation was much faster using the SILP system. The TTON of 130 obtained after 50.5
hours on stream using the SILP system was much lower compared to the TTON of 361
obtained after a similar time on stream (51 h) using the bulk IL. There were two main
differences between these two experiments: Firstly, the total amount of [Ru] present in the
reactor was only 12.5 μmol in the SILP system compared to 25 μmol in the bulk IL, and
secondly, in the SILP system support material was present.
The lower total amount of [Ru] present in the reactor at a constant stream of reaction gases
might lead to faster deactivation if impurities in the reaction gases were responsible for the
deactivation. In the continuous-flow experiments CO2 with the purity grade 4.5 (Westfalen,
99.995 vol.-%) was used, which contained < 5 vol.-ppm CO. Using the ideal gas law and
assuming the highest amount of CO (5 vol.-ppm) the molar CO stream was calculated to be
0.25 μmol h-1 for the employed CO2 stream of 20 mLN min-1. Under these circumstances, a
total amount of 12.6 μmol CO passed the reactor after 50.5 hours on stream, which was 1
equivalent to the amount of catalyst present. This might have been indeed enhancing the
formation of the inactive complex [Ru(H)(CO)2(Triphos)]+ (4). Complex 4 was observed as the
main deactivation product in the continuous-flow experiment using bulk IL (Figure 57).
However, the formation of 4 needs two equivalents of CO. Moreover, the batch experiments
with [Ru(H)2CO(Triphos)] (17)/HNTf2 showed that CO could dissociate from 17 as long as
there was only one CO coordinated to the complex (vide supra). As the CO which passes the
reactor with the gas stream does not accumulate inside the reactor, clearly the
decarbonylation of catalytic intermediates still played an important role in the formation of
4.
RESULTS & DISCUSSION
-112-
Figure 58: Total TON (TTON) versus reaction time of a continuous-flow CO2 hydrogenation using a SILP catalyst based on complex 2/HNTf2 (1 : 1), [EMIM][NTf2], and fluorinated silica support (n(Ru) = 12.5 μmol, HNTf2 (1 eq.), T(reactor) = 140 °C, T(capillary) = 140 °C, p = 120 bar, (H2) = 60 mLN min
-1, (CO2) = 20 mLN min
-1).
To explore the influence of the support material on the catalyst activity, a batch reaction was
performed with complex 2/HNTf2 (1 : 1) in THF in the presence of SGFLUO (100 mg) under
[EMIM] = 99.9 % (IC), [NTf2] = 99.9 % (IC)) was added and the mixture stirred at r.t.
Perfluoro-alkyl [-Si(Me)2CH2CH2C6F13] functionalised silica (particle size 32-63 μm, mean pore
diameter 59 Å, BET surface area 207 m2∙g-1, mesopore volume 0.8 mL g-1, SGFLUO)[25, 191] was
suspended in THF (10 mL). The catalyst/IL solution was added dropwise to this carefully
stirred silica suspension. After stirring this mixture for 1 h at r.t. THF was removed slowly in
EXPERIMENTAL
-132-
vacuo (over ca. 1 h). The remaining powder was dried in vacuo to give a pale yellow, dry
powder.
5.6 DFT-calculations
DFT-calculations were carried out by M. Hölscher, V. Moha, and J. Kothe (ITMC, RWTH
Aachen University) using the Gaussian09 program series (Revision C.01 and D.01).[195]
“Gas phase calculations
The M06-L density functional[196-200] and the def2-SVP basis set[201-205] with the associated
ECP[206-207] for ruthenium were used to calculate the optimised geometries of all structures
with no constraints or restraints. The automatic density fitting approximation was
activated.[208-209] Frequency calculations were carried out to assign structures as local minima
(i = 0) or transition states (i = 1). IRC calculations were performed for the most optimised
transition states to ensure connection of the transition state with the minima.
Thermochemical corrections were computed for standard state conditions. Single point
energies were additionally calculated using M06-L/def2-TZVP. Corrected values for the Gibbs
free energies were obtained by adding the thermochemical corrections from the lower-level
geometry optimisations to the electronic energies of the higher-level single-point
calculations. These values were used for discussion throughout this work.
Solvent phase calculations
Solvent phase calculations were performed for selected gas phase structures with no
constraints or restraints using the MN12-L density functional and the def2-TZVP basis set[201-
205] with the associated ECP[206-207] for ruthenium. The automatic density fitting
approximation was activated.[208-209] IEF-PCM and CPCM[210-211] formalisms were used to
consider solvent effects. Frequency calculations were carried out to assign structures as local
minima (i = 0) or transition states (i = 1). Thermochemical corrections were performed for
413.15 K, and entropy corrections in the condensed phase were considered by specifying a
pressure of 302 atm.[212-213]” [139]
REFERENCES
-133-
6 References
[1] M. Bertau, H. Offermanns, L. Plass, F. Schmidt, H.-J. Wernicke, (Eds.), Methanol: The Basic Chemical and Energy Feedstock of the Future, Springer, 2013.
[2] Bundesanstalt für Geowissenschaften und Rohstoffe (BGR), Reserven, Ressourcen und Verfügbarkeit von Energierohstoffen, Hannover, 2014.
[3] M. Peters, B. Kohler, W. Kuckshinrichs, W. Leitner, P. Markewitz, T. E. Müller, ChemSusChem 2011, 4, 1216-1240.
[4] Bundesanstalt für Geowissenschaften und Rohstoffe (BGR), Reserves, Resources and Availability of Energy Resources, Hannover, 2010.
[5] W. Leitner, E. Dinjus, F. Gaßner, in Aqueous Phase Organometallic Catalysis - Concepts and Applications (Eds.: B. Cornils, W. A. Herrmann), Wiley-VCH, Weinheim, 1998, pp. 486 - 498.
[6] W. Leitner, Angew. Chem. Int. Ed. 1995, 34, 2207-2221.
[7] P. G. Jessop, in The Handbook of Homogeneous Hydrogenation (Eds.: J. G. d. V ries, C. J. Elsevier), Wiley-VCH, Weinheim, 2007, pp. 489 - 511.
[8] M. Aresta, A. Dibenedetto, Dalton Trans. 2007, 2975-2992.
[9] M. Aresta, A. Dibenedetto, A. Angelini, Chem. Rev. 2014, 114, 1709-1742.
[10] W. Wang, S. Wang, X. Ma, J. Gong, Chem. Soc. Rev. 2011, 40, 3703-3727.
[11] International Energy Agency, CO2 Emissions from Fuel Combustion Highlights, Vol. 2014 Edition, 2014.
[12] G. Holmes, D. W. Keith, Philos. Trans. A. Math. Phys. Eng. Sci. 2012, 370, 4380-4403.
[13] M. Ricci, Recovery and Utilization of Carbon Dioxide (Ed.: M.Aresta), Kluwer Academic Publishers, The Netherlands, 2003.
[14] W. Leitner, Acc. Chem. Res. 2002, 35, 746-756.
[15] Chemical Synthesis Using Supercritical Fluids (Eds.: P. G. Jessop, W. Leitner, Wiley-VCH, Weinheim, 1999.
[16] B. Vega, A. R. Kovscek, Developments and Innovation in Carbon Dioxide (CO2) Capture and Storage Technology (Ed.: M. M. Maroto-Valer), Woodhead Publishing, Oxford, 2010, pp. 104-126.
[17] F. T. H. Chung, R. A. Jones, H. T. Nguyen, SPE Reservoir Engineering 2013, 3, 822-828.
[18] J. P. Hallett, T. Welton, Chem. Rev. 2011, 111, 3508-3576.
[19] F. Gozalpour, S. Ren, R., B. Tohidi, Oil & Gas Sc. and Techn. - Rev. IFP 2005, 60, 537-546.
[20] K. Zosel, Angew. Chem. Int. Ed. 1978, 17, 702-709.
[21] G. Anitescu, L. L. Tavlarides, J. Supercrit. Fluid. 2006, 38, 167-180.
[22] T. C. Stauffer, R. A. Venditti, R. D. Gilbert, J. F. Kadla, J. Appl. Polym. Sci. 2002, 83, 2699-2704.
[23] U. Hintermair, PhD thesis, RWTH Aachen (Aachen), 2011.
[24] U. Hintermair, G. Francio, W. Leitner, Chem. Commun. 2011, 47, 3691-3701.
[25] U. Hintermair, G. Francio, W. Leitner, Chem. Eur. J. 2013, 19, 4538-4547.
REFERENCES
-134-
[26] U. Hintermair, T. Höfener, T. Pullmann, G. Franciò, W. Leitner, ChemCatChem 2010, 2, 150-154.
[27] S. Wesselbaum, U. Hintermair, W. Leitner, Angew. Chem. Int. Ed. 2012, 51, 8585-8588.
[28] U. Hintermair, G. Zhao, C. C. Santini, M. J. Muldoon, D. J. Cole-Hamilton, Chem. Commun. 2007, 1462-1464.
[29] M. F. Sellin, P. B. Webb, D. J. Cole-Hamilton, Chem. Commun. 2001, 781-782.
[30] P. B. Webb, M. F. Sellin, T. E. Kunene, S. Williamson, A. M. Slawin, D. J. Cole-Hamilton, J. Am. Chem. Soc. 2003, 125, 15577-15588.
[31] P. B. Webb, T. E. Kunene, D. J. Cole-Hamilton, Green Chem. 2005, 7, 373-379.
[32] J. R. Hyde, P. Licence, D. Carter, M. Poliakoff, Appl. Catal. A – Gen. 2001, 222, 119-131.
[33] P. Licence, J. Ke, M. Sokolova, S. K. Ross, M. Poliakoff, Green Chemistry 2003, 5, 99-104.
[34] P. G. Jessop, T. Ikariya, R. Noyori, Chem. Rev. 1995, 95, 259-272.
[35] P. G. Jessop, F. Joo, C. C. Tai, Coord. Chem. Rev. 2004, 248, 2425-2442.
[36] Y. M. Zhu, S. Y. Ha, R. I. Masel, J. Power Sources 2004, 130, 8-14.
[37] C. Das Neves Gomes, O. Jacquet, C. Villiers, P. Thuery, M. Ephritikhine, T. Cantat, Angew. Chem. Int. Ed. 2012, 51, 187-190.
[38] C. Federsel, A. Boddien, R. Jackstell, R. Jennerjahn, P. J. Dyson, R. Scopelliti, G. Laurenczy, M. Beller, Angew. Chem. Int. Ed. 2010, 49, 9777-9780.
[39] C. Federsel, C. Ziebart, R. Jackstell, W. Baumann, M. Beller, Chem. Eur. J. 2012, 18, 72-75.
[40] C. Ziebart, C. Federsel, P. Anbarasan, R. Jackstell, W. Baumann, A. Spannenberg, M. Beller, J. Am. Chem. Soc. 2012, 134, 20701-20704.
[41] T. G. Ostapowicz, M. Schmitz, M. Krystof, J. Klankermayer, W. Leitner, Angew. Chem. Int. Ed. 2013, 52, 12119-12123.
[42] L. Wu, Q. Liu, R. Jackstell, M. Beller, Angew. Chem. Int. Ed. 2014, 53, 6310-6320.
[43] V. K. Srivastava, P. Eilbracht, Catal. Commun. 2009, 10, 1791-1795.
[44] M. M. T. Khan, S. B. Halligudi, S. Shukla, J. Mol. Catal. 1989, 57, 47-60.
[45] M. M. T. Khan, S. B. Halligudi, S. Shukla, J. Mol. Catal. 1989, 53, 305-313.
[46] M. M. T. Khan, S. B. Halligudi, N. N. Rao, S. Shukla, J. Mol. Catal. 1989, 51, 161-170.
[47] D. K. Lee, D. S. Kim, S. W. Kim, Appl. Organomet. Chem. 2001, 15, 148-150.
[48] P. Budininkas, G. A. Remus, Research and development study related to the synthesis of formaldehyde from CO2 and H2, Gen. American Transport. Corp., 1968, p. 72 pp.
[49] K. Beydoun, T. vom Stein, J. Klankermayer, W. Leitner, Angew. Chem. Int. Ed. 2013, 52, 9554-9557.
[50] K. Beydoun, G. Ghattas, K. Thenert, J. Klankermayer, W. Leitner, Angew. Chem. Int. Ed. 2014, 53, 11010-11014.
[51] Y. Li, I. Sorribes, T. Yan, K. Junge, M. Beller, Angew. Chem. Int. Ed. 2013, 52, 12156-12160.
[52] B. Denise, R. P. A. Sneeden, J. Organomet. Chem. 1981, 221, 111-116.
[53] K. Tominaga, Y. Sasaki, M. Kawai, T. Watanabe, M. Saito, Chem. Commun. 1993, 629-631.
[54] K. I. Tominaga, Y. Sasaki, K. Hagihara, T. Watanabe, M. Saito, Chem. Lett. 1994, 23, 1391-1394.
REFERENCES
-135-
[55] K. Tominaga, Y. Sasaki, T. Watanabe, M. Saito, Bull. Chem. Soc. Jpn. 1995, 68, 2837-2842.
[56] IHS, Chemical Economics Handbook Hydrogen, https://www.ihs.com/products/hydrogen-chemical-economics-handbook.html, accessed on 31.03.2015
[57] M. A. Laguna-Bercero, J. Power Sources 2012, 203, 4-16.
[58] Z. Wang, R. R. Roberts, G. F. Naterer, K. S. Gabriel, Int. J. Hydrogen Energ. 2012, 37, 16287-16301.
[60] M. Baerns, A. Behr, A. Brehm, J. Gmehling, H. Hofmann, U. Onken, A. Renken, Technische Chemie, WILEY-VCH, Weinheim, 2006.
[61] F. Asinger, Methanol - Chemie- und Energierohstoff, Springer-Verlag, Heidelberg, 1986.
[62] G. A. Olah, A. Goeppert, G. K. S. Prakash, Beyond Oil and Gas: The Methanol Economy, Wiley-VCH, Weinheim, 2006.
[63] M. Nielsen, E. Alberico, W. Baumann, H. J. Drexler, H. Junge, S. Gladiali, M. Beller, Nature 2013, 495, 85-89.
[64] R. E. Rodriguez-Lugo, M. Trincado, M. Vogt, F. Tewes, G. Santiso-Quinones, H. Grützmacher, Nat. Chem. 2013, 5, 342-347.
[65] P. Hu, Y. Diskin-Posner, Y. Ben-David, D. Milstein, ACS Catal. 2014, 4, 2649-2652.
[66] E. Alberico, M. Nielsen, Chem. Commun. 2015, 51, 6714-6725.
[67] M. Saito, T. Fujitani, M. Takeuchi, T. Watanabe, Appl. Catal. A – Gen. 1996, 138, 311-318.
[68] H. W. Lim, M. J. Park, S. H. Kang, H. J. Chae, J. W. Bae, K. W. Jun, Ind. Eng. Chem. Res. 2009, 48, 10448-10455.
[69] M. Behrens, F. Studt, I. Kasatkin, S. Kuhl, M. Havecker, F. Abild-Pedersen, S. Zander, F. Girgsdies, P. Kurr, B. L. Kniep, M. Tovar, R. W. Fischer, J. K. Norskov, R. Schlögl, Science 2012, 336, 893-897.
[70] H. Göhna, P. König, CHEMTECH 1994, 24, 36-39.
[71] P. Davies, F. Snowdon, G. W. Bridger, D. O. Hughes, P. W. Young, DE1241429B, 1963.
[72] D. Rotman, Chem. Week 1994, 154, 14.
[73] H. Göhna, P. König, US5631302, 1997.
[74] M. Saito, M. Takeuchi, T. Fujitani, J. Toyir, S. Luo, J. Wu, H. Mabuse, K. Ushikoshi, K. Mori, T. Watanabe, Appl. Organomet. Chem. 2000, 14, 763-772.
[75] K. Ushikoshi, K. Mori, T. Kubota, T. Watanabe, M. Saito, Appl. Organomet. Chem. 2000, 14, 819-825.
[76] M. Saito, Catalysis Surveys from Japan 1998, 2, 175-184.
[77] O.-S. Joo, K.-D. Jung, Y. Jung, in Studies in Surface Science and Catalysis, Vol. 153 (Eds.: J.-S. C. Sang-Eon Park, L. Kyu-Wan), Elsevier, 2004, pp. 67-72.
[78] O. S. Joo, K. D. Jung, I. Moon, A. Y. Rozovskii, G. I. Lin, S. H. Han, S. J. Uhm, Ind. Eng. Chem. Res. 1999, 38, 1808-1812.
[79] O.-S. Joo, CAMERE PROCESS FOR CARBON DIOXIDE HYDROGENATION TO FORM METHANOL, https://web.anl.gov/PCS/acsfuel/preprint%20archive/Files/45_4_WASHINGTON%20DC_08-00_0686.pdf, accessed on 28.02.2016.
[80] G. A. Olah, A. Goeppert, G. K. Prakash, J. Org. Chem. 2009, 74, 487-498.
REFERENCES
-136-
[81] V. Gronemann, W. Liebner, P. di Zanno, F. Pontzen, M. Rothaemel, Nitrogen + Syngas 2010, 308, 36-39.
[82] T. Matsushita, T. Haganuma, D. Fujita, Mitsui Chemicals Inc., WO2012067222A1, 2012.
[83] www.carbonrecycling.is, accessed on 17.12.2014.
[84] A. M. Shulenberger, F. R. Jonsson, O. Ingolfsson, K. Tran, S. Singh, O. F. Sigurbjornsson, WO2007108014 A1, 2007.
[85] A. M. Shulenberger, F. R. Jonsson, O. Ingolfsson, K.-C. Tran, WO2007108014 A1, 2007.
[86] M. Behrens, Angew. Chem. Int. Ed. 2014, 53, 12022-12024.
[87] J. Graciani, K. Mudiyanselage, F. Xu, A. E. Baber, J. Evans, S. D. Senanayake, D. J. Stacchiola, P. Liu, J. Hrbek, J. Fernandez Sanz, J. A. Rodriguez, Science 2014, 345, 546-550.
[88] Y. Inoue, H. Izumida, Y. Sasaki, H. Hashimoto, Chem. Lett. 1976, 863-864.
[89] K.-I. Tominaga, Y. Sasaki, M. Saito, K. Hagihara, T. Watanabe, J. Mol. Catal. 1994, 89, 51-55.
[90] K.-I. Tominaga, Y. Sasaki, T. Watanabe, M. Saito, in Studies in Surface Science and Catalysis, Volume 114 (Eds.: T. Inui, M. Anpo K. Izui S. Yanagida, T. Yamaguchi), Elsevier, 1998, pp. 495-498.
[91] S. Chakraborty, J. Zhang, J. A. Krause, H. Guan, J. Am. Chem. Soc. 2010, 132, 8872-8873.
[92] S. Chakraborty, Y. J. Patel, J. A. Krause, H. R. Guan, Polyhedron 2012, 32, 30-34.
[93] A. E. Ashley, A. L. Thompson, D. O'Hare, Angew. Chem. Int. Ed. 2009, 48, 9839-9843.
[94] M. J. Sgro, J. Domer, D. W. Stephan, Chem. Commun. 2012, 48, 7253-7255.
[95] G. Menard, D. W. Stephan, J. Am. Chem. Soc. 2010, 132, 1796-1797.
[96] G. Menard, D. W. Stephan, Dalton Trans. 2013, 42, 5447-5453.
[97] M. A. Courtemanche, J. Larouche, M. A. Legare, W. H. Bi, L. Maron, F. G. Fontaine, Organometallics 2013, 32, 6804-6811.
[98] S. Bontemps, S. Sabo-Etienne, Angew. Chem. Int. Ed. 2013, 52, 10253-10255.
[99] M. A. Courtemanche, M. A. Legare, L. Maron, F. G. Fontaine, J. Am. Chem. Soc. 2013, 135, 9326-9329.
[100] S. N. Riduan, Y. Zhang, J. Y. Ying, Angew. Chem. Int. Ed. 2009, 48, 3322-3325.
[101] A. Schäfer, W. Saak, D. Haase, T. Müller, Angew. Chem. Int. Ed. 2012, 51, 2981-2984.
[102] Y. N. Li, R. Ma, L. N. He, Z. F. Diao, Catal. Sci. Tech. 2014, 4, 1498-1512.
[103] F. G. Fontaine, M. A. Courtemanche, M. A. Legare, Chemistry 2014, 20, 2990-2996.
[104] E. Balaraman, C. Gunanathan, J. Zhang, L. J. Shimon, D. Milstein, Nat. Chem. 2011, 3, 609-614.
[105] P. H. Dixneuf, Nat. Chem. 2011, 3, 578-579.
[106] J. Choudhury, ChemCatChem 2012, 4, 609-611.
[107] E. Balaraman, Y. Ben-David, D. Milstein, Angew. Chem. Int. Ed. 2011, 50, 11702-11705.
[108] E. Balaraman, B. Gnanaprakasam, L. J. Shimon, D. Milstein, J. Am. Chem. Soc. 2010, 132, 16756-16758.
[109] Z. Han, L. Rong, J. Wu, L. Zhang, Z. Wang, K. Ding, Angew. Chem. Int. Ed. 2012, 51, 13041-13045.
REFERENCES
-137-
[110] C. A. Huff, M. S. Sanford, J. Am. Chem. Soc. 2011, 133, 18122-18125.
[111] N. M. Rezayee, C. A. Huff, M. S. Sanford, J. Am. Chem. Soc. 2015, 137, 1028-1031.
[112] N. J. Oldenhuis, V. M. Dong, Z. Guan, Tetrahedron 2014, 70, 4213-4218.
[113] J. R. Khusnutdinova, J. A. Garg, D. Milstein, ACS Catal. 2015, 5, 2416-2422.
[114] S. W. Foo, Y. Takada, Y. Yamazaki, S. Saito, Tetrahedron Lett. 2013, 54, 4717-4720.
[115] S. Wesselbaum, T. Vom Stein, J. Klankermayer, W. Leitner, Angew. Chem. Int. Ed. 2012, 51, 7499-7502.
[116] S. Savourey, G. Lefevre, J. C. Berthet, P. Thuery, C. Genre, T. Cantat, Angew. Chem. Int. Ed. 2014, 53, 10466-10470.
[117] A. J. Miller, D. M. Heinekey, J. M. Mayer, K. I. Goldberg, Angew. Chem. Int. Ed. 2013, 52, 3981-3984.
[118] D. Morton, D. J. Cole-Hamilton, Chem. Commun. 1987, 248.
[119] E. Alberico, P. Sponholz, C. Cordes, M. Nielsen, H. J. Drexler, W. Baumann, H. Junge, M. Beller, Angew. Chem. Int. Ed. 2013, 52, 14162-14166.
[120] A. Monney, E. Barsch, P. Sponholz, H. Junge, R. Ludwig, M. Beller, Chem. Commun. 2014, 50, 707-709.
[121] H. T. Teunissen, C. J. Elsevier, Chem. Commun. 1997, 667-668.
[122] H. T. Teunissen, C. J. Elsevier, Chem.Commun. 1998, 1367-1368.
[123] L. Rosi, M. Frediani, P. Frediani, J. Organomet. Chem. 2010, 695, 1314-1322.
[124] A. A. Nunez Magro, G. R. Eastham, D. J. Cole-Hamilton, Chem. Commun. 2007, 3154-3156.
[125] J. Coetzee, D. L. Dodds, J. Klankermayer, S. Brosinski, W. Leitner, A. M. Slawin, D. J. Cole-Hamilton, Chemistry 2013, 19, 11039-11050.
[126] F. M. A. Geilen, B. Engendahl, A. Harwardt, W. Marquardt, J. Klankermayer, W. Leitner, Angew. Chem. Int. Ed. 2010, 49, 5510-5514.
[127] F. M. Geilen, B. Engendahl, M. Hölscher, J. Klankermayer, W. Leitner, J. Am. Chem. Soc. 2011, 133, 14349-14358.
[128] T. vom Stein, T. Weigand, C. Merkens, J. Klankermayer, W. Leitner, ChemCatChem 2013, 5, 439-441.
[129] T. vom Stein, M. Meuresch, D. Limper, M. Schmitz, M. Hölscher, J. Coetzee, D. J. Cole-Hamilton, J. Klankermayer, W. Leitner, J. Am. Chem. Soc. 2014, 136, 13217-13225.
[130] Y. Y. Ohnishi, T. Matsunaga, Y. Nakao, H. Sato, S. Sakaki, J. Am. Chem. Soc. 2005, 127, 4021-4032.
[131] J. Stary, J. O. Liljenzin, in Pure Appl. Chem. 1982, 54, 2557.
[132] S. I. Hommeltoft, M. C. Baird, Organometallics 1986, 5, 190-195.
[133] G. Jenner, E. M. Nahmed, H. Leismann, Tetrahedron Lett. 1989, 30, 6501-6502.
[134] O. R. Allen, S. J. Dalgarno, L. D. Field, P. Jensen, A. C. Willis, Organometallics 2009, 28, 2385-2390.
[135] I. Mellone, M. Peruzzini, L. Rosi, D. Mellmann, H. Junge, M. Beller, L. Gonsalvi, Dalton Trans. 2013, 42, 2495-2501.
[136] M. K. Whittlesey, R. N. Perutz, M. H. Moore, Organometallics 1996, 15, 5166-5169.
REFERENCES
-138-
[137] J. A. Widegren, R. G. Finke, J. Mol. Catal. A – Chem. 2003, 198, 317-341.
[138] G. M. Whitesides, M. Hackett, R. L. Brainard, J. P. P. M. Lavalleye, A. F. Sowinski, A. N. Izumi, S. S. Moore, D. W. Brown, E. M. Staudt, Organometallics 1985, 4, 1819-1830.
[139] S. Wesselbaum, V. Moha, M. Meuresch, S. Brosinski, K. M. Thenert, J. Kothe, T. vom Stein, U. Englert, M. Hölscher, J. Klankermayer, W. Leitner, Chem. Sci. 2015, 6, 693-704.
[140] K. Tominaga, Y. Sasaki, J. Mol. Catal. A – Chem. 2004, 220, 159-165.
[141] N. Sieffert, R. Reocreux, P. Lorusso, D. J. Cole-Hamilton, M. Buhl, Chemistry 2014, 20, 4141-4155.
[142] M. Meuresch, PhD-Thesis, RWTH Aachen University (Aachen), 2016.
[143] M. Meuresch, S. Westhues, W. Leitner, J. Klankermayer, Angew. Chem. Int. Ed. 2016, 55, 1392-1395.
[144] B. D. Yeomans, D. G. Humphrey, G. A. Heath, Dalton Trans. 1997, 4153-4166.
[146] I. S. Kolomnikov, A. I. Gusev, G. G. Aleksandrov, T. S. Lobeeva, Y. T. Struchkov, M. E. Vol'pin, J. Organomet. Chem. 1973, 59, 349-351.
[147] M. G. Bradley, D. A. Roberts, G. L. Geoffroy, J. Am. Chem. Soc. 1981, 103, 379-384.
[148] C. Bianchini, A. Meli, S. Moneti, F. Vizza, Organometallics 1998, 17, 2636-2645.
[149] A. D. Getty, C. C. Tai, J. C. Linehan, P. G. Jessop, M. M. Olmstead, A. L. Rheingold, Organometallics 2009, 28, 5466-5477.
[150] A. Urakawa, F. Jutz, G. Laurenczy, A. Baiker, Chemistry 2007, 13, 3886-3899.
[151] P. Scherl, A. Kruckenberg, S. Mader, H. Wadepohl, L. H. Gade, Organometallics 2012, 31, 7024-7027.
[152] Y. Musashi, S. Sakaki, J. Am. Chem. Soc. 2000, 122, 3867-3877.
[153] X. Z. Yang, ACS Catal. 2011, 1, 849-854.
[154] C. Fellay, N. Yan, P. J. Dyson, G. Laurenczy, Chemistry 2009, 15, 3752-3760.
[155] L. E. Heim, N. E. Schlorer, J. H. Choi, M. H. Prechtl, Nat. Commun. 2014, 5, 3621.
[156] S. Kozuch, S. Shaik, Acc. Chem. Res. 2011, 44, 101-110.
[157] A. V. Marenich, C. J. Cramer, D. G. Truhlar, J. Phys. Chem. B 2009, 113, 6378-6396.
[158] K. Rohmann, M. Hölscher, W. Leitner, J. Am. Chem. Soc. 2016, 138, 433-443.
[159] A. B. Chaplin, P. J. Dyson, Inorg. Chem. 2008, 47, 381-390.
[160] G. B. Deacon, R. J. Phillips, Coord. Chem. Rev. 1980, 33, 227-250.
[161] M. Hirano, R. Fujimoto, K. Hatagami, N. Komine, S. Komiya, ChemCatChem 2013, 5, 1101-1115.
[162] J. Brown, I. Gridnev, J. Klankermayer, in Amplification of Chirality (Ed.: K. Soai), Springer, Berlin, Heidelberg, 2008, pp. 35-65.
[163] V. I. Bakhmutov, E. V. Bakhmutova, N. V. Belkova, C. Bianchini, L. M. Epstein, D. Masi, M. Peruzzini, E. S. Shubina, E. V. Vorontsov, F. Zanobini, Can. J. Chem. 2001, 79, 479-489.
[164] P. G. Jessop, B. Subramaniam, Chem. Rev. 2007, 107, 2666-2694.
[165] C. A. Thomas, R. J. Bonilla, Y. Huang, P. G. Jessop, Can. J. Chem. 2001, 79, 719-724.
REFERENCES
-139-
[166] M. J. Kamlet, J. L. Abboud, R. W. Taft, J. Am. Chem. Soc. 1977, 99, 6027-6038.
[167] V. T. Wyatt, D. Bush, J. Lu, J. P. Hallett, C. L. Liotta, C. A. Eckert, J. Supercrit. Fluid. 2005, 36, 16-22.
[168] P. G. Jessop, Green Chem. 2011, 13, 1391-1398.
[169] T. Schaub, R. A. Paciello, Angew. Chem. Int. Ed. 2011, 50, 7278-7282.
[170] J. Halpern, A. L. W. Kemp, J. Am. Chem. Soc. 1966, 88, 5147-5150.
[171] T. Ikarashi, Chem. Econ. Eng. Rev. 1980, 12, 31-34.
[172] J. Krug, C. Palm, BASF AG, DE4239243A1, 1994.
[173] W. H. E. Müller, M. Zölffel, Chemische Werke Huels AG, DE2927931A1, 1981.
[174] V. Pace, P. Hoyos, L. Castoldi, P. Dominguez de Maria, A. R. Alcantara, ChemSusChem 2012, 5, 1369-1379.
[175] D. F. Aycock, Org. Proc. Res. Dev. 2007, 11, 156-159.
[176] A. Bösmann, G. Franciò, E. Janssen, M. Solinas, W. Leitner, P. Wasserscheid, Angew. Chem. Int. Ed. 2001, 40, 2697-2699.
[177] D. J. Cole-Hamilton, Science 2003, 299, 1702-1706.
[178] M. F. Klügge, Master's Thesis, RWTH Aachen University (Aachen), 2014.
[179] M. J. Muldoon, C. M. Gordon, I. R. Dunkin, Perkin Trans. 2 2001, 433-435.
[180] L. Crowhurst, P. R. Mawdsley, J. M. Perez-Arlandis, P. A. Salter, T. Welton, Phys. Chem. Chem. Phys. 2003, 5, 2790-2794.
[181] C. C. Cassol, G. Ebeling, B. Ferrera, J. Dupont, Adv. Synth. Catal. 2006, 348, 243-248.
[182] D. J. Heldebrant, P. G. Jessop, J. Am. Chem. Soc. 2003, 125, 5600-5601.
[183] D. J. Heldebrant, H. N. Witt, S. M. Walsh, T. Ellis, J. Rauscher, P. G. Jessop, Green Chem. 2006, 8, 807-815.
[184] Z. Hou, N. Theyssen, A. Brinkmann, W. Leitner, Angew. Chem. Int. Ed. 2005, 44, 1346-1349.
[185] J. Theuerkauf, PhD thesis, RWTH Aachen University (Aachen), 2012.
[186] NIST Chemistry WebBook, NIST Standard Reference Database Number 69 (Eds.: P.J. Linstrom, W.G. Mallard), National Institute of Standards and Technology, Gaithersburg MD, 20899, http://webbook.nist.gov, (retrieved April 10, 2016).
[187] J. O. Valderrama, R. E. Rojas, Ind. Eng. Chem. Res. 2009, 48, 6890-6900.
[188] C. P. Mehnert, Chemistry 2004, 11, 50-56.
[189] A. Riisager, R. Fehrmann, M. Haumann, P. Wasserscheid, Eur. J. Inorg. Chem. 2006, 695-706.
[190] R. Ciriminna, P. Hesemann, J. J. Moreau, M. Carraro, S. Campestrini, M. Pagliaro, Chemistry 2006, 12, 5220-5224.
[191] S. Kainz, Z. Y. Luo, D. P. Curran, W. Leitner, Synthesis, 1998, 1425-1427.
[192] A. Muth, O. Walter, G. Huttner, A. Asam, L. Zsolnai, C. Emmerich, J. Organomet. Chem. 1994, 468, 149-163.
[193] U. Hintermair, C. Roosen, M. Kaever, H. Kronenberg, R. Thelen, S. Aey, W. Leitner, L. Greiner, Org. Proc. Res. Dev. 2011, 15, 1275-1280.
[194] U. Hintermair, PhD thesis, RWTH Aachen University (Aachen), 2011.
REFERENCES
-140-
[195] M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, G. Scalmani, V. Barone, B. Mennucci, G. A. Petersson, H. Nakatsuji, M. Caricato, X. Li, H. P. Hratchian, A. F. Izmaylov, J. Bloino, G. Zheng, J. L. Sonnenberg, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, T. Vreven, J. A. Montgomery Jr., J. E. Peralta, F. Ogliaro, M. J. Bearpark, J. Heyd, E. N. Brothers, K. N. Kudin, V. N. Staroverov, R. Kobayashi, J. Normand, K. Raghavachari, A. P. Rendell, J. C. Burant, S. S. Iyengar, J. Tomasi, M. Cossi, N. Rega, N. J. Millam, M. Klene, J. E. Knox, J. B. Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R. E. Stratmann, O. Yazyev, A. J. Austin, R. Cammi, C. Pomelli, J. W. Ochterski, R. L. Martin, K. Morokuma, V. G. Zakrzewski, G. A. Voth, P. Salvador, J. J. Dannenberg, S. Dapprich, A. D. Daniels, Ö. Farkas, J. B. Foresman, J. V. Ortiz, J. Cioslowski, D. J. Fox, Gaussian, Inc., Wallingford, CT, USA, 2009.
[196] Y. Zhao, D. G. Truhlar, J. Chem. Phys. 2006, 125, 194101.
[197] Y. Zhao, D. G. Truhlar, Theor. Chem. Acc. 2008, 120, 215-241.
[198] C. J. Cramer, D. G. Truhlar, Phys. Chem. Chem. Phys. 2009, 11, 10757-10816.
[199] B. B. Averkiev, Y. Zhao, D. G. Truhlar, J. Mol. Catal. A: Chem. 2010, 324, 80-88.
[200] A. D. Kulkarni, D. G. Truhlar, J. Chem. Theory. Comput. 2011, 7, 2325-2332.
[201] A. Sch fer, H. Horn, R. Ahlrichs, J. Chem. Phys. 1992, 97, 2571.
[202] A. Sch fer, C. Huber, R. Ahlrichs, J. Chem. Phys. 1994, 100, 5829.
[203] K. Eichkorn, F. Weigend, O. Treutler, R. Ahlrichs, Theor. Chem. Acc. 1997, 97, 119-124.
[204] F. Weigend, F. Furche, R. Ahlrichs, J. Chem. Phys. 2003, 119, 12753-12762.
[205] F. Weigend, R. Ahlrichs, Phys. Chem. Chem. Phys. 2005, 7, 3297-3305.
[206] D. Andrae, U. Häußermann, M. Dolg, H. Stoll, H. Preuß, Theoretica Chimica Acta 1990, 77, 123-141.
[207] B. Metz, H. Stoll, M. Dolg, J. Chem. Phys. 2000, 113, 2563-2569.
[208] B. I. Dunlap, J. Chem. Phys. 1983, 78, 3140-3142.
[209] B. I. Dunlap, J. Mol. Struc. – Theochem. 2000, 529, 37-40.
[210] J. Tomasi, B. Mennucci, R. Cammi, Chem. Rev. 2005, 105, 2999-3093.
[211] E. Cances, B. Mennucci, J. Tomasi, J. Chem. Phys. 1997, 107, 3032-3041.
[212] R. L. Martin, P. J. Hay, L. R. Pratt, J. Phys. Chem. A 1998, 102, 3565-3573.
[213] N. Sieffert, M. Buhl, Inorg. Chem. 2009, 48, 4622-4624.