MSc Chemistry Literature Thesis
CO
2
and ethylene coupling towards acrylic acid and
related compounds
An impossible dream?
By
Klaas Willem Visscher
29 June 2020
Student Number
10748687
Research Institute
Van 't Hoff Institute for Molecular Sciences
Research Group
Homogeneous, Supramolecular and
Bio-Inspired Catalysis
Daily Supervisor
Dhr. R. (Raoul) Plessius MSc
Examiner
Dr. prof. Bas de Bruin
Second Examiner
Abstract
For the highly demanded acrylic acid and related compound an alternative pathway is being researched based on the coupling of ethylene and CO2. While ethylene and CO2 readily couple to
form a nickellalactone, no catalytic cycle has been completed to this date. In this thesis the exact nature of the two main bottlenecks, that are unfavourable thermodynamics and lack of β-hydride elimination, examined closely and possible solutions for these problems are discussed and compared. The strategies that indeed result in functioning catalytic cycles are those that target a different product than acrylic acid. The reason these strategies work is that by targeting a different product the reaction becomes exergonic. However, by reacting acrylate with an acid the main target can be produced. An in depth overview of using a Brønsted base as stochiometric reagents is given as this was the most successful strategy. The conclusions about the reaction conditions from different work were compared to each other and based of those an overlaying conclusion was attempted to be made. The main difficulty is that the base is susceptible for reaction with carbon dioxide and thus the reaction conditions should disfavour this as much as possible. A high pressure is necessary for this reaction, but there was no general consensus about the partial pressures. The only conclusion that could be applied to all research is that an aprotic, but polar solvent is preferred.
Contents
1 Introduction 1
1.1 Scope and Structure of the Thesis . . . 3
2 General Mechanism 4 3 Problems in Catalysis 4 4 Strategies 6 4.1 alkyl halides . . . 6 4.2 Brønsted Acids . . . 7 4.3 Brønsted Base . . . 9 4.4 Lewis Acids . . . 14
4.5 Early Transition Metals . . . 16
4.6 Polymerization . . . 17
5 Overview Thermodynamics 19
6 Case study: Acrylate synthesis 21
7 Conclusion 23
1
Introduction
Acrylic acid is used as precursor for a large variety of molecules in different fields, such as adhesives, coatings, elastomers, paints, plastic and textiles.1More than a million ton of this compound are produced annually, as the demand is high for this compound in the aforementioned industries.2 The merit of this
chemical becomes clear from simply looking at its chemical structure. The simplest unsaturated acid can readily by polymerized due to its vinyl group (Figure 2), while at the same time the acid group allows for interesting secondary modifications of said polymers, such as esterification. Furthermore, the β-carbon atom is polarized by the carbonyl group and is therefore an electrophile, thus favouring reacting with a large number of nucleophiles.
Figure 1: Polymerization of acrylic acid
Historically, this chemical was produced by the carboxylation of acetylene, using high pressures of carbonmonoxide and acetylene in combination with a heterogeneous nickel catalyst, which falls under the category of Reppe chemistry.2 The relatively high price of acetylene and the use of high pressure
carbon monoxide less than desirable, therefore this method was discarded for a more sustainable and safer one.
Figure 2: Historic method of producing acrylic acid
Nowadays, the industry produces most of the acrylic acid by a two-step oxidation of propylene using an early transition metal heterogeneous catalyst, which contains oxides of arsenic, niobium and molybdenum on a silica.3 Propylene is much cheaper than acetylene and is the byproduct from the
production of gasoline and ethylene.
Figure 3: Current route from bulk chemicals to acrylic acid
Instead of partially oxidizing propylene it is also possible to partially oxidize propane with the molecu-lar oxygen.4The benefit of this process is that propane is significantly cheaper than propylene. However,
a major problem with these reactions is that these processes require multiple distillation steps in order to remove aldehyde impurities.
The previous discussed chemical processes have in common that they oxidize reactants originating from fossil fuels, thus a more sustainable route is desirable. Like the Reppe process once was discarded for a safer and more sustainable process, the current process should also be replaced. The effects of climate change become more clear everyday, while at the same time the fossil fuels are depleting.
An example for such route is the dehydration of lactic acid catalyzed by a NaY zeolites.5While lactic acid can be produced sustainable by fermentation, the demand of acrylic acid is so high that this method production can not be feasible for such scale in the near foreseeable future.
Figure 4: Alternative pathway to acrylic acid from lactic acid.
A more interesting approach would be the coupling of CO2and ethylene directly. The great advantage
of this reaction is that it uses CO2 and thus combats the greenhouse directly. Therefore, CO2 capture
is a strategy which is often investigated in the fight against global warming. There are two types of carbon capture: there is carbon capture storage (CCS) and carbon capture utilization (CCU).6In CCS
the purpose is to capture carbon and to store it in order to reduce the amount of greenhouse gas in the atmosphere. In contrast, in CCU carbon dioxide is seen as another bulk ingredient that can be utilized to synthesize a large variety of chemicals. As seen in Figure 5, CO2 is a start product in this reaction,
thus it is carbon capture utilization as the purpose is to creation of useful chemicals.
Figure 5: Reaction equation of CO2coupling to ethylene.
Although it utilizes a CO2molecule, it would be in all likeliness not a good method to remove a CO2
permanently from the atmosphere as the chance is high that the final products made from acrylic acid would be burnt on the end of their life time. Yet, this still means that the total amount of CO2 in the
atmosphere will decrease, because it spends time captured as chemical. Thus it is still an interesting alternative to the present methods of producing acrylic acid.
Figure 6: The net increase of CO2 in the atmosphere per molecule of acrylic acid used based on the
chemicals consumed in the processes.
Moreover, the other bulk ingredient ethylene also has some advantages above propylene. It is like propylene produced by oil and gas winning, but ethylene is cheaper than propylene. However, ethylene is also a natural molecule as it functions as a hormone in plants.7 Furthermore, it can be synthesized
by the dehydration of ethanol, which can be produced sustainably.8 In a ideal situation, the ethylene
Table 1: Summary of the possible acrylic acid production
Reaction sustainable scalable
Furthermore, a new sustainable pathway is not only beneficial for the production of acrylic acid, but also to related chemicals. This new process gives the means to sustainable production of a whole class of chemicals related to acrylic acid, as the lessons learned from this reaction can be applied to related compounds. For example the monomer used for the production of acrylic glass, methyl methacrylate, could perhaps be produced in a similar fashion.
1.1
Scope and Structure of the Thesis
In this thesis, different strategies are discussed in order to couple ethylene and CO2 to acrylic acid and
related compounds. The two major obstacles in this reaction are the thermodynamics and the β-hydride elimination, these problems will be discussed in depth and how they can be possibly overcome.
The thesis is primarily focussed on the different strategies that can be implemented and why these do (not) work. While the sustainability of these strategies is taken into account in the discussion, it is not the primary goal of this thesis.
The thesis is divided in the following sections: The first section focusses on the general mechanism and the general problems. The following section shows the different strategies found in literature that are deployed in order to overcome the problems. After this section the different strategies are compared with each other and finally the most successful strategy will be discussed in more depth.
2
General Mechanism
The first close attempt of coupling ethylene and CO2 to acrylic acid was done by Hoberg’s group in
1983.9The use of a nickel catalyst with bis(dicyclohexylphosphino)ethane or 2,2’-bipyridin resulted in a nickellalactone. However, no β-hydide elimination occurred thus no acrylic acid was released. Treating the complex with acid resulted in the formation of propylic acid.
However, based on the results from Hoberg’s group a hypothetical mechanism of coupling CO2 and
ethylene can be crafted. The first step is the coordination of an ethylene molecule to the metal complex. This step is followed by the oxidative coupling or carboxylation of CO2, which could happen by either
an outer-sphere attack or an inner-sphere attack.10 Then a β-hydride elimination is executed and finally
the dissociation of the acrylic acid occurs.
Figure 7: Hypothetical catalytic cycle in which ethylene and CO2 are coupled to form acrylic acid.
3
Problems in Catalysis
There are two main reasons for the resistance against β-hydride elimination. Firstly, the thermodynamics are unfavourable; the total reaction is uphill with ∆Gr0 = +42.7 kJ/mol.11Thus difficulties in this cycle
were bound to happen sooner or later. Therefore, these reaction are always attempted under high pressures, but calculations indicate that this is not enough to circumvent this issue.
Secondly, the high activation barrier of β-hydride elimination is approximately ∆G = + 164 kJ/mol.12–14
The roots of this high activation barrier lie in the rigidity of the nickellalactone, which is caused by the ring strain and the high M-O bond dissociation energy. The hydrogen atom can simply not come close enough to the metal and thus the vacant dx2−y2 is not available due to the square planar ligands, as
.
Figure 8: d-orbital splitting of a nickellalactone in which the square planar orienation makes the vacant dx2−y2 unavailable for the β-hydrogen elimination
Thus in order to overcome difficulties in β-hydride elimination, the reaction conditions should favour the formation of acrylic acid via an alternative pathway through the potential energy surface must be created in order to favour β-hydride elimination.
4
Strategies
4.1
alkyl halides
Despite the initial promising results from Hoberg’s group in the eighties no full catalytic cycle has been realised, which caused to researchers to look for additional components to use in the reaction. One example of such compound is methyl iodide. Strictly speaking, using an alkyl halide does not result in acrylic acid, but in an ester. However, this reaction pathway is still a viable one as methyl acrylate for example is still a valuable building block. Most research is done on the use of methyl iodide to facilitate β-hydride elimination and pushes the thermodyanamics of the reaction. The general catalytic cycle of this reaction is given in Figure 9.
Figure 9: Hypothetical catalytic cycle in which ethylene and CO2 are coupled to form acrylic acid in
combination with MeI.
The use of methyl iodide changes the thermodynamics from ∆Gr0 + 42.7 kJ/mol to ∆Gr0 = +
21.0 kJ/mol.16 Although the thermodynamics are more preferable, the reaction remains endergonic.
However, the activation barrier of the β-hydride elimination is lowered considerably. The mechanistic reason behind this is that the free oxygen on the metallactone can attack the methyl iodine in a SN2
fashion.16
Figure 10: SN2 reaction of a metalloctone with methyl iodine.
This weakens the M-O bond significantly and causes the ligand to rotate in such fashion that β-hydride elimination becomes possible as shown in Figure 11a.16 Although the reaction barrier is lowered
by the use of this alkyl halide, there is a possible side-reaction resulting in a conformational change which is unable to do a β-hydride elimination. (Fortunately, this side reaction can be prevented by the use of bidentate ligands.
(a) Conformational change of the metal complex ren-dering the β-hydride elimination possible.
(b) Conformational change of the metal complex ren-dering the β-hydride elimination impossible.
Even after β-hydride elimination the reaction can still go astray. Hydrogenation of the α-hydrogen leads to the formation of a branched four membered nickellalactone. Whether this product is thermo-dynamically more stable than acrylic acid release is dependent on the ligands used, highlighting the importance of ligand design for this reaction. It is even thought that preventing this side-reaction should be the main concern when designing a ligand, opposed to lowering the barrier in the main productive pathway.16
Figure 12: Competitive product.
It has been shown that the use of a large excess MeI can cause the release of the acrylic acid. The exact nature of this phenomenon is still under debate. Deduced from the equilibrium constant, increasing methyl iodide concentration is expected to push the reaction more into the direction of methylacrylate.
K =[M ethylacrylate][HI] [CO2][ethylene][M eI]
(1)
Besides, this rather obvious effect on the equilibirum, MeI is also thought to disfavour the previous discussed side-reaction due to its low polarity and hence increase the formation of the desired product.16 Another possibility is that the excess is necessary to lower the barrier of β-hydride elimination.17However,
the high concentration of MeI has as disadvantage, namely it can do an oxidative addition to the Nickel 0 complex, reducing the amount of active catalysts.18
Despite all effort, no catalytic cycle up to date has been realized. Rieger and coworkers came very close, as they were able to release acrylic acid from the nickel complexes, alas no catalyst was recovered and thus no complete catalytic cycle was executed.18 Yet there is some consensus what characteristics
the ligand should posses. It seems that monodentate ligands are not preferred for this reaction as seen in 11b. The ligand should be flexible, as it should allow for the twisting of the ring. That’s why PNP pincer ligands are not effective due to their inability for adapting a different conformational form, preventing β-hydride elimination.19 Furthermore, the chelating bonding angle should not to be large as dppb did
not show any activity. Moreover, sp2 donor atoms hinder the reaction of nickellalactone with methyl iodide. Yet, it seems that the trans effect does not play an important role in the Ni-O cleavage. When C2H5 and CF3CH2I are used as alkylation reagents, the ring opening for tmeda system appears to be
faster than for the diphosphine ligands.
4.2
Brønsted Acids
In the previous discussed catalytic cycle (Figure 9), HI is expected to be reductively eliminated to regain the active metal centre. This could suggest that HI can also be used to cleave the MO bond and hence
opening the path for β-hydride elimination. HI is both consumed and liberated during this mechanism (Figure 13), meaning it is used catalytically, which is the great strength of using HI. However, this means there is no change in thermodynamics and thus the reaction remains strongly uphill.
Figure 13: Hypothetical catalytic cycle in which ethylene and CO2 are coupled to form acrylic acid in
combination with HX.
Despite the lack of improvement in thermodynamics, there has been research done in the use of HX for this reaction. Hoberg et all has conducted research in the coupling of CO2 with cyclooctene
or cyclopentene with an Brønsten acid as additional catalyst.20,21 Although, octene falls outside of the scope of this project, it is a nice proof of concept and could possibly applied to ethylene. While they were able to create the desired acids, it was not done in a catalytic cycle. Furthermore, when HCl is utilized a different reaction can occur, the saturated product is then formed.
Figure 14: Formation of saturated an unsaturated acid by utilizing HCl.
Moreover, it is interesting to note that there is some computational evidence that acrylic acid itself can catalyze β-hydride elimination. Multiple possible routes for this catalytic reaction were proposed in the work from Hofmann and coworkers. The first example of such pathway is intermolecular proton shuttling and is given in Figure 15.10
Figure 15: Acrylic acid catalyzing β-hydride elimination by proton shuttling.
Similar transition state energies were calculated for intramolecular pathways. In this pathway an additional acrylic acid binds to the complex and initiates ring opening and proton transfer happens intramolecular, which is visually clarified in Figure 16.
Figure 16: Acrylic acid catalyzing β-hydride elimination ring opening and intramolecular proton transfer
While these possible transition states show that the β-hydride elimination might be less cumbersome than initially thought, the endergonic nature of the reaction remains a problem, which is probably the reason why no catalytic cycle has been completed.
4.3
Brønsted Base
Another strategy to overcome inertness against β-hydride elimination is to skip this step. In the sense that the hydride is not directly transferred to the metal centre, but instead is accepted by an external base. The advantage of this approach is that difficulties caused by the steric hindrance due to the ring strain of the metallalactone are avoided, as the hydrogen is not required to have an agostic interaction with the metal, similar to the aforementioned alkyl halide strategy. The downside of this method is that instead of acrylic acid acrylate is produced and thus a stoichiometric amount of base is being used in this reaction.
However, since acrylate is the product of this reaction, there is a change in the thermodynamics. As seen in Figure 18 the product is lower in energy as the metallalactone. Furthermore, this method seems to be successful; Limbach and all reported in 2012 the first complete catalytic cycle with a turnover number greater than ten.22 The key to the success was having different stages in the process, in which the in the first step contained a high pressure of CO2 and in the later steps there pressure was reduced.
Figure 17: Step wise process that could be repeated in order to achieve the first catalytic cycle by changing the partial pressures.
Comparing different ligands in this study led to the hypothesis that the larger the bite angle, the greater the energy barrier. The C-C bond length of η2 bound ethylene ligands point to stronger Ni to
ethylene back bonding for the biphosphines with smaller bite angle. Yet, when dbpm or dppm was used the reaction failed, meaning that the bite angle can become too small. Furthermore, electron rich P stereogenic biphophines seem to increase the TON, but one should not discard the effects of solvents. How the base effects the transition state and the total energy of the reaction is given in Figure 18, the significant lowering of the energies explain why in this case a catalytic cycle could be completed. However, it seems that cations also play a role in this reaction, when a quaternary ammonium salt was used no yield was obtained, unless an external sodium agent was added. This hints at a possibility of Lewis acids playing a role in this reaction. Moreover, there are also difficulties in picking the correct base for this type reaction, due to the inability of bases to cooperate with CO2.22
2-F-PhONa as base.23Furthermore, this was a one pot synthesis, making this process far more practical
than the first catalytic cycle. In this study different phenoxides were investigated and the activity of the phenoxide appears to correlate with the pKa of the species. The base should be basic enough to react with nickellalactone, although the nucleophilicity should be low enough to keep neutralization reaction with CO2 at a minimum. Therefore, a huge excess of base (300 base/cat ratio) yielded the best result,
because than the likelihood increased for a nickellalactone to be deprotanated, which as seen in Figure 19 is an irreversible step. By using the correct base, their reaction conditions allowed for 20 bar CO2 and
10 bar C2H4 pressure. Furthermore, the addition of zinc powder had a beneficial effect on the TON,
the reason given in the paper for this phenomenon is that it can reduce the intermediate nickel species. While catalytic cycles were achieved without the addition of zinc, the TON was marginal in those cases, thus improvement is needed to achieve good yields without the additions of zinc using. Moreover, this setup allowed for the coupling of styrene to CO2 as well. Furthermore, they discovered that when the
temperature was increased to 120◦C much the reaction becomes sluggish due to the decarboxylation of the nickellalactone.
Figure 19: Suggested reaction mechanism for coupling CO2 and ethylene using BenzP* ligand and a
phenoxide as base in THF.
In a later study an even higher TON of 290 was achieved by optimizing the reaction conditions carefully.24 Ronchin and coworkers came to the conclusion that an aprotic solvent, but polar solvent
conclusion from the fist catalytic cycle, they came to the conclusion that a large bite angle, such as the dppp ligand has, is better for the reaction. Moreover, they concluded that NaOPh was more effective as base as than KOPh, which is due to the Lewis acidity of the involved cation. Furthermore, they used relatively high temperature of 90◦C and a pCO2 and pC2H4 of both 25 bar.
Although the usage of nickel catalyst provided promising results, it can be interesting to investigate other elements in the coupling of CO2 and ethylene to an acrylate. Palladium is for such purpose a
viable candidate as it is iso-electronic to nickel. Limbach et all, reported that palladium can indeed be used as catalyst as catalytic cycle was completed with a TON of 29 (Figure 20).25 Furthermore, in
the aforementioned research of Ronchin et all, they could complete a catalytic cycle with palladium, although the yield was much lower than for the respective nickel catalyst.
Figure 20: Coupling reaction using a Palladium catalyst
The general trend of nickel performing better than palladium, makes the work of Schaub et all impressive; their group reached a TON higher than 500 by using a palladium catalyst, see Figure 21.26
The reason for their success was the choice of using choosing amides as solvents. CO2has a high solubility
in amides and thus this favours the formation of palladiumalactones. Furthermore, the use of amides as solvents appear to stabilize the active species, while it also limits the decomposition of the catalyst. n-Cyclohexyl-2-pyrrolidon (CHP) as solvent resulted in the highest TON, partly due to the reluctance of forming catalyst clusters in this solvent.27–29Furthermore, this amide is an aprotic polar solvent, which fits the aforementioned conclusion from Ronchin et all. The pressure used in this process was 10 bar ethylene and 40 bar CO2 with NaOtBu as base.
Figure 21: Catalytic cycle of Pd catalyst in which the β-H is done externally by a base and dcpe is used as ligand.26
Moreover, the group also developed a continuous process in order to extract the product from the reaction, the foundation of this reaction lies on the poor miscibility of CHP and water. The process also includes recycling of the base and solvent, an overview of this process is given in Figure 22.
Figure 22: Continuous process of sodium acrylate production using a palladium catalyst and an amide solvent.26
A similar approach was carried out with nickel instead of palladium, but to no yield was obtained. Schaub et all suggests that this is due to the formation of inactive nickel carbonyl due to the reactivity of amides as carbonyling agents.26,30–32
As seen in Figure 23, there is much progress made in the coupling of ethylene and CO2 using
stoi-chiometric amounts of base. The steady increase of TON over the short amount of time gives prospect of applying this reaction in the nearby future in industry.
Figure 23: The achieved TON of the catalytic cycles over time with their respective metal
influence of a Lewis acid in the reaction. This hypothesis is supported by a computational study, as DFT calculations suggest that alkali metals can indeed lower the the energy barriers.33 (Figure 24)
Figure 24: The influence of alkali metals on the transition state.33
Dieter Vogt and coworkers confirmed the importance of a Lewis acid by completing a catalytic cycle with the use of LiI and a nickel catalyst.33 LiI was chosen as reacting agent as it contains a hard Lewis
acid and a soft Lewis base. A TON of 21 was reached when DCPP was used as ligand, NEt3 and a
high ethylene pressure was used. However, an additional Zn source was also necessary to reduce NiI2
back to the active catalyst. In Figure 25 the proposed catalytic cycle is given, the formation of LnNiI2
is excluded in this mechanism. It was also shown, that with the proper ligand the formation of LnNiI2
could also be avoided, although the TON was considerably lower in comparison with the experiment in which Zn was used as reducing agent. For this reaction it seemed that the increase of the chelate ring size and bite angle had a positive effect on the Ni-catalyzed reaction. As mentioned earlier DCPP gave the best results when Zn was added to the reaction mix, but DCPB gave the best result when Zn was excluded. Therefor it appears that the formation of the inactive [(ligand)NiI2] is suppressed by using
Figure 25: Proposed catalytic cycle by Dieter Vogt et all.33
Besides the use of an alkali metals, there is also research conducted on the use of a more ”classical” Lewis acid: BAr3. Although no catalytic cycle was completed, β-hydride elimination was promoted
rapidly.34The addition of this Lewis acid resulted in structure similar to the competitive product of the
reaction with MeI as seen in Figure 12. Due to a 2,1-insertion that proceeded the β-hydride elimination.
Figure 26: Product after adding a lewis acid to a metallactone.
The group of Bernskoetter proposed the use of an additional base to close to the cycle. However, despite the release of acrylic acid and the formation of the ethylene nickel adduct, the coupling of CO2
did not occur. The catalyst that was investigated was a (dppf)Ni species.
4.5
Early Transition Metals
All the work discussed so far have been inspired by the work of Hobergs’ group. However, another major player in the eighties conducting research on ethylene CO2 coupling reaction, was the group of
Carmona.35,36 Instead of using a nickel catalyst as Hoberg, they utilized the early transition metals
such as molybdenum and tungsten for this reaction. While no catalytic cycles were observed, a rather interesting phenomenon occurred. It appears that these catalysts did allow for β-hydride elimination. A molybdenum complex reacted as described in Figure 27.
Figure 27: Formation of dimer system in early transition metal complexes.
At first sight, it could be expected that the β-hydride elimination is enabled by the dimer system. However, upon closer investigation it can be concluded that this is not the origin of the facile β-hydride elimination. A highly similar reaction that employed tungsten instead of molybdenum yielded in the monomer species in which the β-hydride elimination occurred. Furthermore, the group attached chelating ligands on the molybdenum complex and thus sterically prevented the formation of the dimer system and this still resulted in hydride elimination. Rather, the H transfer is initiated due to the twisting of β-CH2group from the previously formed planar metallalactone in order to allow a Mo-H agostic interaction
and the formation of a hydrido acrylate species.37 The direct formation of the twisted metal complex
instead of first forming a planar metallalactone was considered unlikely due to the energy difference between the barriers and the products itself.38
The twisting of β-CH2is likely enabled due to the dissociation of the phosphine ligand. Herein we see
a clear distinction between nickellalactones and molybdenumalactone as the nickellalactone is a square planar d8complex, whereas the molybdenum complex has an octehedral geometry. This allows for easier
dissociation of ligands and therefor easier twisting of the ring for molybdenum, as the dissociation energy of the PMe3 ligand is only around 10 kcal/mol.38
Yet, while the use of these early transition metals enable facile β-Hydride elimination, no catalytic cycle has been completed by the cooperating these metals in a system. As these systems are restricted by the same unfortunate thermodynamics, difficulties are bound to happen in the system. In this case the problem lies in the ejection of acrylic acid of the metal complex, due to the electron rich nature of the metal centre the C=C bond is strongly attached to the metal by π-backbonding. Furthermore, due to the oxyphillic nature of early transition metals the oxygen is also tightly bound. In order to liberate the acrylic acid and to complete the catalytic cycle nBuLi can be used, which results in the
lithium acrylate.36 However, if the goal is to synthesise acrylic acid on a large sustainable scale, using
stoichiometric amounts ofnBuLi is not desirable.
In a different study methyliodide was used as liberation agent instead of n-BuLi on a tungsten complex.39 These tungsten acrylate hydrate species produce methyl acrylate in similar quantities as the aforementioned late transition metal systems. However, this reaction is not catalytic as the final reduction step to yield starting tungsten complex is missing.
Figure 28: Tungsten complex with phosphite ligands coupling CO2 and ethylene with methyliodine in
stochiometric amounts.
4.6
Polymerization
Since the only viable catalytic coupling of ethylene and CO2 discussed so far did not result in acrylic
acid, but in the acrylate, it may also be useful to look at other possible products. The reaction to these related compounds may have thermodynamics that are less harsh. A possible alternative is to polymerize
the product with ethylene. For every ethylene added to the chain the Gibbs free energy is lowered by 12 kcal/mol.40 An additional benefit of increasing the chain length is that the metalcyclic is enlarged, hence the ring strain is lessened, allowing the β-hydride elimination to take place.
This strategy was implemented by the group of Chirik. Interestingly, instead of a nickel catalyst they used an iron catalyst (Figure 29).40Usually iron is disregarded in this field of research due to its known affinity of forming iron carbonate complexes.41
Figure 29: The catalyst using in the polymerization reaction of Chirik.
No synthetic cycle was closed as the complex was unable to break the iron oxygen bond. An interesting feature of this setup is that by tuning the CO2 and ethylene pressure the length of the chain could be
controlled. Furthermore, the importance of ring strain becomes clear in this experiment as no acrylic acid is formed, but the longer the chain the more abundant the unsaturated product becomes in comparison to the saturated product. Besides, the saturated and unsaturated products, another termination of polymerization was the insertion of a second CO2 molecule, resulting in dicarboxylic acid. (Figure 31)
Figure 30: Possible products when using an iron complex in the coupling of CO2 and ethylene.
Due to the aforementioned tendency of iron complexes to form iron carbonates, a high ratio of ethylene/CO2was required in this experiment. Adding excess CO2resulted in a low yield. The low yield
is probably the result of the run away reaction of decarbonylation and iron carbonate formation is than possible.
Figure 31: Runaway reaction leading to iron carbonate by decarbonylation.
5
Overview Thermodynamics
Table 2 summarizes the strategies discussed so far; it illustrates which strategies do not need an additional stochimometric agent and those who do. Furthermore, the type of product that is produced is also shown.
Table 2: Summary of the strategies to overcome β-hydride elimination
Strategy working catalytic cycle no stochiometric additive product
Alkyl halides acrylic ester
Brønsted acid acrylic acid
Brønsted base acrylate
Lewis acid depends on LA
Early TM acrylic acid
Polymerization fatty acid
In the review from Wang and coworkers,42 they propose three additional strategies, besides ligand
design, to improve the catalytic process. Their first solution is to use an amphoteric compound containing both a Lewis acid and Brønsten base that can stabilize the carbonyl and induce ring contraction. Their reasoning is that this would eject acrylic acid and not the sodium acrylate.
Their second solution was to use dimethylcarbonate instead of methylidodine as liberating agent, as this is much more sustainable and could perhaps even be produced in the same process from CO2 and
methanol.
Their third solution was to create an alternative reaction mechanism, which lacks the metallalactone as intermediate. They envision this could be possible by direct C-H activation of ethylene and then directly inserting the vinyl group into the CO2 moiety. However, while these strategies are interesting
and potentially could offer some new insight to the sustainable production of acrylic acid, they gloss over the bigger underlying problem which no catalyst can solve and that is that this reaction is endergonic,13
which will be a main part of the discussion in this part.
In Table 3 the different types of coupling CO2 with ethylene are listed. It becomes clear why the
catalytic cycles so far only have been realized for constructing acrylates, as no catalyst can ever hope to overcome a positive ∆G. The ∆G is ultimately dependent on the reaction conditions in which the solvent and pressure can play an important role. In this case, increasing the pressure will benefit the thermodynamics of this coupling reaction. However, the pressure is already at such a high level at the aforementioned investigations, that the reaction can not be ’fixed’ simply by increasing the pressure.
Table 3: Thermodynamics of the different products formed in the coupling of ethylene and CO2 in gas phase Reaction ∆GR0 (kJ/mol) +57.413 +21.016 -5922 +57.4 - n x 5040
Due to the high stability of CO2, the unfavourable thermodynamics will always be a problem for the
fixation of CO2, therefore the only successful catalytic cycles we have seen so far is when another agent
is sacrificed in order to gain the necessary negative Gibbs free energy. As we have seen for the acrylate formation and the polymerization reaction. Yet, we know from nature that it is possible to perform a uphill reaction sustainable as life itself is full of examples of endergonic reactions.
Thus to synthesize acrylic acid or one of its derivatives there are two plausible options. The first option and usually the desired option to complete an endergonic reaction is to use energy directly in the form of photons as nature does with photosynthesis. Generally in chemistry, this greatly reduces the steps necessary to make a pathway sustainable and thus much more sustainable. However, usually a radical is present in some form in photochemistry, which means this option is not applicable for this particular reaction, due to the tendency of the product to polymerize (Figure 32). Although ethylene can also polymerize by free radicals, the pressure and temperature necessary for this reaction makes it less likely to occur under the conditions discussed so far.43Perhaps a combined system could provide an outcome, in which acrylic acid is produced from CO2 and ethylene by light, which is then polymerized.
The thermodynamics of that total reaction will be favourable.
Figure 32: Polymerization of ethylene to polyethene due to the initiation by a radical.
The second option is to use a sacrificial agent that can be produced sustainable, similar to how ATP is used in nature. As the first option to use light is highly complicated, it seems that the use of a sacrificial agent to tip the balance of the reaction in the favour of acrylic acid is the most probable strategy to sustainably produce acrylic acid from CO2. The sacrificial agents that have been used in this
project were bases, alkyl halides and ethylene itself. However, the use of these compounds led to different products than acrylic acid. If the goal is to produce acrylic acid, it should be taken into account that these products need to be converted. Therefore, this extra step should also be taken into consideration when discussing the sustainability of this reaction.
Of course to make the coupling reaction fully sustainable this extra step needs to be sustainable as well as the ethylene production itself. Fortunately, there is ongoing promising research being conducted regarding this subject.44 However, this is beyond the scope of this project to describe the sustainable
Brønsted base can be recycled from the produced alcohols in the work of Limbach or Schaubs, this would also offer perspective on a green production acrylate.
6
Case study: Acrylate synthesis
It appears that the production of acrylate is most likely candidate to be produced in the near future with the ethylene CO2 coupling reaction as we have already seen some catalytic cycles being completed
in this report. The general reaction is given in Figure 33
Figure 33: General reaction for acrylate production.
From this reaction the following equilibrium equation can be constructed, which can help to under-stand how to adjust the conditions to favour the formation of the desired product.
K = [HB][acrylate] [CO2][ethylene][B−]
(2)
This equilibrium shows that an high pressure CO2 and ethylene and a high base concentration are
desired to form acrylate. Yet, the reaction is more complex than this formula due to side reactions. A major problem in this process is the reactivity between base and CO2. Furthermore, the strength of
the base also has a large influence, the most important characteristics of the base is that the conjugated acid should be 6 units lower on the pKa scale as acrylic acid to make the reaction exergonic.45 Therefor,
there needs to be a balance between the strength of the base to push the reaction forwards, while it’s reactivity with CO2 should be kept at a minimum. In order to fulfil this requirement less nucleophillic
bases are preferred.
The ethylene/CO2partial pressure is thus important for the TON, based on the two aforementioned
equillibriums. According to the paper of Vogt and coworkers, it appears that the higher the ratio the better .46Although a direct reason for this observation was not given; it is possible that a high pressure in
the first place is necessary to prevent the decarbonyxlation reaction by pushing the equillibrium towards the metallalactone. Yet, at the same time the ratio is desired to prevent the reaction of CO2 with the
base, clarified in Figure 34. However, this is in contrast with the other catalytic cycles discussed so far, in those cycles there was a high ratio of CO2/ethylene partial pressure. Yet it should be noted that those
other studies did not study the influence of the partial pressure intensely. This could mean that possibly even higher TONs could be achieved in those systems by optimizing the partial pressures.
Figure 34: Both the CO2 and ethylene pressure push the desired first reaction forward, but only CO2
pushes the undesired second reaction forward.
Furthermore, in all the work discussed so far a high temperature was preferred in order to increase the yield. As long as no decarboxylation of the metallalactone occurs the temperature can be increased.23
Moreover, the reaction favours aprotic but polar solvents,24which is reflected in the working processes which we have seen so far (PhCl, DMSO, CHP, THF). The latter condition can be easily understood
by looking at the polarity of the reactants and the products, as the reactants are two apolar molecules ethylene and carbon dixoxide, whereas the product is the polar molecule acrylate, thus a polar solvent will favour the formation of product. Furthermore, a polar solvent can better stabilize the ionic intermediates. However, a protic solvent such as an alcohol can react with M(II)Cl to inactive M(II)OR species, therefore an aprotic solvent is preferred.24 A non-coordinating solvent is also preferred as this will strengthen the interaction between the metal and reagents.22 Moreover, as aforementioned the base should not have a high tendency to react with carbon dioxide. This is another reason why the choice of solvent is so important, as it has a major influence on this reaction46
Furthermore, it appears that a large concentration of base is beneficial to the TON, this is could be due to that the complex reacting with the base is the irreversible step or that a large portion of the base reacts with CO2.
While the fact that the reaction works is promising, there are some issues that would need to be solved before it is applied on large scale. First of all the TON should preferably be higher, which probably can be achieved by tweaking the previous discussed reaction conditions carefully. Furthermore, the catalyst plays a major role in improving the TON and its role should be studied in depth. Usually the metal center is varied in such studies, but in this case nickel appears to be highly effective for this reaction already, which is promising for industry as it is an inexpensive metal. However, palladium should not be discarded too easily as a TON > 500 has been achieved using this metal centre and an easy continuous process can be set up with this catalyst. Ligand design is the other obvious choice for improving upon the TON. However, what kind of ligand is necessary is dependent on the exact situation. In the aforementioned first catalytic cycle a small bite angle was desired, while other research suggests a higher bite angle.33 The reason for this discrepancy is that by using a base that can attack the complex outer-shell. Thus the rate-determining step is not necessary the β-hydride elimination and the most stable intermediate can differ based on the other reaction conditions.
7
Conclusion
The production of acrylic acid like chemicals by the coupling of CO2 to ethylene is a desired alternative
production method to the current production pathway that is the oxidation of propylene. The two big difficulties that need to be overcome for this novel reaction is the endergonic nature of this reaction and the β-hydride elimination step in the catalytic cycle.
An attractive method to overcome β-hydride elimination is by using an additional agent that can attack outer-sphere, thus the inherent rigidity of the square planar metallalactones is then no longer a problem. Possible agents have been investigated, such as alkyl halides, Brønsted acids and brønsted bases. Although acrylate has been produced by methylation, no catalytic cycle could be completed, which could be done when a Brønsted base was used as sacrificial agent and thus this method appears to be most likely candidate to be deployed on large scale. This method distinguishes itself from the other methods, by changing the thermodynamic nature of the reaction from endergonic to exergonic, dependent on the base used in this reaction. Furthermore, by reacting acrylate with an acid the main target, acrylic acid, can be produced.
Based on the current literature, the Brønsted base mediated CO2and ethylene coupling reaction that
will be deployed in industry probably possesses the following characteristics. First of all, the base should be just right; strong enough to push the reaction, but it should not have to much interference with CO2.
Secondly, an aprotic, but polar solvent will be used in the reactor. The process will have a high gas pressure possibly with a high ratio ethylene/CO2. Finally, a nickel catalyst containing a ligand with a
large bite angle will be deployed in this reactor. Yet, palladium should not be discarded to easily as the highest TON to date and a continuous process have been developed with this metal in conjunction with an amide solvent. Moreover, the reaction conditions in the ethylene CO2 are mutual dependent on each
other and thus can not be optimized individually. Only comprehensive studies covering multiple of those optimization will help to clarify these interrelationships and improve upon the current situation.
8
Outlook
Although, a sustainable production of acrylic acid like chemicals appears to be feasible in the near future, there are many things that should be considered before industrial plants can be set up. If in the future indeed Brønsted bases will be used in the production, it is important to remember that the cost of producing these bases in regard to environment also need to be taken into account. On a similar note, perhaps this future reaction appears to be sustainable on lab scale, but due to upscaling problems the industrial process might be more harmful to the environment than the current situation. However, the 290 TON catalytic cycle with a nickel catalyst and a TON of 514 with a palladium catalyst that have been completed is at least promising.
References
(1) Ohara, T.; Sato, T.; Shimizu, N.; Prescher, G.; Schwind, H.; Weiberg, O.; Marten, K.; Greim, H. In Ullmann’s Encyclopedia of Industrial Chemistry; American Cancer Society: 2011, pp C51–C53. (2) Ohara, T.; Sato, T.; Shimizu, N.; Prescher, G.; Schwind, H.; Weiberg, O.; Marten, K.; Greim, H.; Shaffer, T. D.; Nandi, P. In Ullmann’s Encyclopedia of Industrial Chemistry; American Cancer Society: 2020, pp 1–21.
(3) Campbell, W. E.; McDaniel, E. L.; Reece, W. H.; Williams, J. E.; Young, H. S. Product R&D 1970, 9, 325–334.
(4) Lin, M. M. Applied Catalysis A: General 2001, 207, 1–16.
(5) Zhang, J.; Zhao, Y.; Pan, M.; Feng, X.; Ji, W.; Au, C.-T. ACS Catalysis 2011, 1, 32–41. (6) Smit, B.; Park, A.-H. A.; Gadikota, G. Frontiers in Energy Research 2014, 2, 55.
(7) McManus, M. T., Annual plant reviews, the plant hormone ethylene; John Wiley & Sons: 2012; Vol. 44.
(8) Takahara, I.; Saito, M.; Inaba, M.; Murata, K. Catalysis Letters 2005, 105, 249–252. (9) Hoberg, H.; Schaefer, D. Journal of Organometallic Chemistry 1983, 251, c51–c53.
(10) Plessow, P. N.; Sch¨afer, A.; Limbach, M.; Hofmann, P. Organometallics 2014, 33, 3657–3668. (11) Limbach, M. In Advances in Organometallic Chemistry; Elsevier: 2015; Vol. 63, pp 175–202. (12) Plessow, P. N.; Weigel, L.; Lindner, R.; Sch¨afer, A.; Rominger, F.; Limbach, M.; Hofmann, P.
Organometallics 2013, 32, 3327–3338.
(13) Graham, D. C.; Mitchell, C.; Bruce, M. I.; Metha, G. F.; Bowie, J. H.; Buntine, M. A. Organometallics 2007, 26, 6784–6792.
(14) P´apai, I.; Schubert, G.; Mayer, I.; Besenyei, G.; Aresta, M. Organometallics 2004, 23, 5252–5259. (15) Jin, D.; Schmeier, T. J.; Williard, P. G.; Hazari, N.; Bernskoetter, W. H. Organometallics 2013,
32, 2152–2159.
(16) Guo, W.; Michel, C.; Schwiedernoch, R.; Wischert, R.; Xu, X.; Sautet, P. Organometallics 2014, 33, 6369–6380.
(17) Lee, S. Y. T.; Cokoja, M.; Drees, M.; Li, Y.; Mink, J.; Herrmann, W. A.; K¨uhn, F. E. ChemSusChem 2011, 4, 1275–1279.
(18) Bruckmeier, C.; Lehenmeier, M. W.; Reichardt, R.; Vagin, S.; Rieger, B. Organometallics 2010, 29, 2199–2202.
(19) Liang, L.-C.; Chien, P.-S.; Lee, P.-Y. Organometallics 2008, 27, 3082–3093.
(20) Hoberg, H.; Ballesteros, A.; Sigan, A. Journal of Organometallic Chemistry 1991, 403, C19–C22. (21) Hoberg, H.; Ballesteros, A.; Sigan, A.; Jegat, C.; Milchereit, A. Synthesis 1991, 1991, 395–398. (22) Lejkowski, M. L.; Lindner, R.; Kageyama, T.; B´odizs, G. ´E.; Plessow, P. N.; M¨uller, I. B.; Sch¨afer,
A.; Rominger, F.; Hofmann, P.; Futter, C.; Schunk, S. A.; Limbach, M. Chemistry - A European Journal 2012, 18, 14017–14025.
(23) Huguet, N.; Jevtovikj, I.; Gordillo, A.; Lejkowski, M. L.; Lindner, R.; Bru, M.; Khalimon, A. Y.; Rominger, F.; Schunk, S. A.; Hofmann, P.; Limbach, M. Chemistry - A European Journal 2014, 20, 16858–16862.
(24) Vavasori, A.; Calgaro, L.; Pietrobon, L.; Ronchin, L. Pure and Applied Chemistry 2018, 90, 315– 326.
(26) Manzini, S.; Cadu, A.; Schmidt, A.-C.; Huguet, N.; Trapp, O.; Paciello, R.; Schaub, T. Chem-CatChem 2017, 9, 2269–2274.
(27) Davila, M. J.; Aparicio, S.; Alcalde, R. Industrial & Engineering Chemistry Research 2009, 48, 10065–10076.
(28) Mooijman, F. R.; Engberts, J. B. The Journal of Organic Chemistry 1989, 54, 3993–3995. (29) Lawson, E.; Sadler, A.; Harmatz, D.; Brandau, D.; Micanovic, R.; MacElroy, R.; Middaugh, C.
Journal of Biological Chemistry 1984, 259, 2910–2912.
(30) Mistry, S.; Natarajan, S. Journal of Chemical Sciences 2014, 126, 1477–1491.
(31) Varshavsky, J. S.; Kiseleva, N.; Cherkasova, T.; Buzina, N. Journal of Organometallic Chemistry 1971, 31, 119–122.
(32) Serp, P.; Hernandez, M.; Richard, B.; Kalck, P. European Journal of Inorganic Chemistry, 2001, 2327–2336.
(33) Hendriksen, C.; Pidko, E. A.; Yang, G.; Sch¨affner, B.; Vogt, D. Chemistry - A European Journal 2014, 20, 12037–12040.
(34) Jin, D.; Schmeier, T. J.; Williard, P. G.; Hazari, N.; Bernskoetter, W. H. Organometallics 2013, 32, 2152–2159.
(35) Galindo, A.; Pastor, A.; Perez, P. J.; Carmona, E. Organometallics 1993, 12, 4443–4451.
(36) Alvarez, R.; Carmona, E.; Galindo, A.; Gutierrez, E.; Marin, J. M.; Monge, A.; Poveda, M. L.; Ruiz, C.; Savariault, J. M. Organometallics 1989, 8, 2430–2439.
(37) Hollering, M.; Dutta, B.; K¨uhn, F. E. Coordination Chemistry Reviews 2016, 309, 51–67. (38) Schubert, G.; P´apai, I. Journal of the American Chemical Society 2003, 125, PMID: 14640662,
14847–14858.
(39) Wolfe, J. M.; Bernskoetter, W. H. Dalton Transactions 2012, 41, 10763–10768.
(40) Rummelt, S. M.; Zhong, H.; Korobkov, I.; Chirik, P. J. Journal of the American Chemical Society 2018, 140, PMID: 30173506, 11589–11593.
(41) Karsch, H. H. Chemische Berichte 1977, 110, 2213–2221. (42) Wang, X.; Wang, H.; Sun, Y. Chem 2017, 3, 211–228.
(43) Malpass, D. B., Introduction to industrial polyethylene: properties, catalysts, and processes; John Wiley & Sons: 2010; Vol. 45.
(44) Dijkmans, T.; Pyl, S. P.; Reyniers, M.-F.; Abhari, R.; Van Geem, K. M.; Marin, G. B. Green Chemistry 2013, 15, 3064–3076.
(45) Perez, P., Advances in Organometallic Chemistry; ISSN; Elsevier Science: 2015.
(46) Huguet, N.; Jevtovikj, I.; Gordillo, A.; Lejkowski, M. L.; Lindner, R.; Bru, M.; Khalimon, A. Y.; Rominger, F.; Schunk, S. A.; Hofmann, P., et al. Chemistry–A European Journal 2014, 20, 16858– 16862.
