MSc Chemistry
Track Molecular Sciences
Literature Thesis – 12 EC
Advances in efficient hydrogen storage in
liquid organic hydrogen carriers
by
Maartje van Rijn
11288779
29 January 2021
Research InstituteVan ‘t Hoff Institute for Molecular Sciences Research Group
Synthetic Organic Chemistry
Supervisor dr. Chris Slootweg
Second examiner prof. dr. Bas de Bruin
Daily supervisor dr. Kaj van Vliet
Abstract
This literature thesis discusses the process made in the development of catalytic systems for hydrogenation and dehydrogenation reactions of various liquid organic hydrogen carriers. Current problems associated with global warming are driving forces for the innovation of sustainable energy sources to supply the worlds energy needs. Molecular hydrogen is a promising alternative as the main energy carrier. It can be generated via water electrolysis and reversibly reacted with oxygen to only produce water as a product. The low volumetric storage density of hydrogen makes it difficult for storage and transport. Thus, the concept of liquid organic hydrogen carriers is a solution. Formic acid, methanol, N-ethylcarbazole and dibenzyltoluene are the first molecules considered and various homogeneously and heterogeneously catalyzed (de)hydrogenation reactions are discussed. Moreover, dehydrogenative coupling reactions of alcohols with alcohols and amines and their reverse are also considered. A techno-economic analysis of a full process chain are surveyed as well to illustrate the feasibility of liquid organic hydrogen carriers.
Table of contents
Abstract ... 2
1. Introduction: hydrogen storage ... 5
2. Formic acid ... 8
2.1 Industrial synthesis ... 8
2.2 Direct CO2 hydrogenation to formic acid ... 9
2.3 Dehydrogenation of formic acid ... 12
2.4 Single catalysts for reversible hydrogen storage in formic acid ... 13
3. Methanol ... 15
3.1 Industrial synthesis ... 15
3.2 Direct CO2 hydrogenation to methanol ... 16
3.3 Dehydrogenation of methanol ... 17
3.4 Reversible hydrogen storage in methanol ... 18
4. N-ethylcarbazole ... 20
4.1 Commercial synthesis of NEC ... 20
4.2 Hydrogenation of H0-NEC ... 20
4.3 Dehydrogenation of H12-NEC ... 21
4.4 Single catalysts for reversible hydrogen storage in NEC ... 22
5. Dibenzyltoluene ... 23
5.1 Commercial synthesis of DBT ... 23
5.2 Hydrogenation of H0-DBT ... 23
5.3 Dehydrogenation of H18-DBT ... 24
5.4 Reversible hydrogen storage in DBT ... 25
6. Alcohol reforming ... 26
6.1 Homogeneous catalysts for reversible alcohol reforming systems ... 26
6.2 Heterogeneous catalysts reversible alcohol reforming systems ... 29
7. Amine reforming ... 32
7.1 Homogeneous catalysts for amine reforming systems ... 32
7.2 Heterogeneous catalysis for amine reforming systems ... 35
8. Techno-economic evaluation ... 38
10. Conclusion & Future Research ... 44 11. References ... 45 Appendix ... 52
1. Introduction: hydrogen storage
To accommodate the ever-increasing energy demand of the world’s growing population, many ambitious and difficult goals have been set to reduce the uses of fossil fuels and transition to societies that can live from sustainable energy sources. In 2015, an estimated 19,3% of the global energy demand was provided by renewable energy sources.1 Investing in renewable energy power plants, such as solar energy or wind power, is being prioritized more and more by governments and companies around the world. It is of great importance to try and slow the increasing amounts of greenhouse gasses in the atmosphere and decrease environmental pollution, which both still continue to rise. The concentration of the most contributing greenhouse gas CO2 has risen to 412 ppm in 2020, where it was 318 ppm in around 1960.2
One well-known initiative is the Paris Agreement, which aims to keep the global temperature rise below 2 degrees Celsius.3 Another related set of initiatives is the European Green Deal, which has the overall goal of making Europe climate neutral by 2050.4,5 More specific aims include reducing greenhouse gas emission by 50% compared to 1990 levels by 2030. For all of this to be achievable, it is essential that a clean energy source can be used efficiently to meet the world’s energy consumption demand.
An increasingly interesting candidate is using molecular hydrogen as the main energy carrier.6–9 The possibilities and prospects in realizing hydrogen as the renewable fuel are being investigated extensively around the globe. According to BloombergNEF in their 2020 Hydrogen Energy Outlook, 150 billion US dollars of cumulative subsidies are necessary to make the costs of hydrogen competitive with current gas prices.10 The socio-economic driving forces for the renewable future are rising. For example, the Fuel Cells and Hydrogen Joint Undertaking FCH JU of the European commission has approved a 90 million euro project for the creation of the Hydrogen Valley in the north of the Netherlands.11 This is one of many examples.12
Hydrogen can be formed via the electrolysis of water, storing the electrical energy in the chemical bonding energy of hydrogen. The energy can be used later by reacting the hydrogen gas back with oxygen, regenerating water and making it a cyclic process. Hydrogen as an energy carrier is promising for several reasons. The efficiency of hydrogen fuel cells can be much greater than it is for conventional combustion engines running on gasoline. Moreover, the energy density of hydrogen is higher than for gasoline. The amount of energy 1 kilogram of hydrogen can deliver is the same as the energy that roughly 2.8 kilograms of gasoline can deliver.13 In other words, the gravimetric energy storage density of hydrogen is very high, meaning that one kilogram of liquid hydrogen contains about 120 MJ of energy compared to 43 MJ for conventional gasoline.9
However, the volumetric energy density of hydrogen is very low, so it takes up much more space for a certain amount of energy than gasoline. Because of this, hydrogen is often stored under very high pressure or in its liquified form in tanks. Figure 1 compares the volumetric and gravimetric density of several energy carriers. Thus, even though hydrogen has a high energy density, a problem arises for storage and transport. Additionally, because storage and transport are difficult, only 5% of the produced hydrogen is currently traded on the free market. The larger part of hydrogen is consumed in locations where it also produced, further illustrating that transport is not efficient.14
Figure 1 Comparison of various energy carriers in their volumetric density (left) and gravimetric density
(right).8
Research has focused much attention on the concept of hydrogen storage in other liquid compounds, as a solution for storage problems arising with pure hydrogen. These liquid organic hydrogen carriers (LOHCs) have the sole purpose of taking up and releasing hydrogen on command. The energy of hydrogen is stored as chemical energy in the hydrogenated form of the LOHC. This loaded compound is then more suitable for transportation. Eventually, the LOHC is dehydrogenated and the released hydrogen is used as fuel for mobility, energy, electricity et cetera. Advantages of using liquid organic compounds include the potential use of existing infrastructures, such as oil pipelines, and less risks in handling and storage.8 Figure 2 illustrates the general concept of liquid organic hydrogen carriers. As depicted, the process of hydrogenation is generally an exothermic reaction, whereas the dehydrogenation reactions are endothermic.15 The process is cyclic and virtually fossil fuel free. The energy required for hydrogen production is desired to come from renewable energy sources, e.g. solar or wind power. The unloaded LOHC is not consumed and can be reused to store hydrogen again.
Figure 2 Schematic representation of the LOHC principle.16
The hydrogenation and dehydrogenation reactions of the LOHC can be performed catalytically and the amount of research performed on (de)hydrogenation catalysts is extensive.7,8,15 Catalysis provides a different route for processes to occur in milder conditions that are otherwise performed in extremer environments. This literature thesis will discuss various different potential hydrogen carriers in the context of hydrogen storage and both heterogeneously or homogeneously catalyzed reactions are considered.
First, formic acid and methanol will be discussed, both as products of the combination of carbon dioxide and hydrogen, which have the advantage of capturing the greenhouse gas CO2. They are extensively researched compounds with desirable properties, such as non-toxicity and liquid at room temperature. Then N-ethylcarbazole (NEC) and dibenzyltoluene (DBT) are discussed. These molecules occur often in literature, because they are stable molecules, they have desirable hydrogen storage capacities and are readily available. Finally, a chapter will be dedicated to reactions which will be classified as alcohol and amine reforming. Those systems are based around the reversible dehydrogenative coupling of alcohols with alcohols or amines. This concept is relatively new in the context of hydrogen storage and is not based around the same principle as the ‘simple’ loading and unloading of a carrier molecule. The aim of this thesis is to provide an overview of the progress made in the catalytic hydrogenation and dehydrogenation of these liquid organic hydrogen carriers. The LOHCs will be discussed critically and the thesis will finish with concluding remarks.
2. Formic acid
The first LOHC to be considered is formic acid (FA). It is the smallest carboxylic acid and has a hydrogen storage capacity of 4.4 wt%. This percentage is below the target set by the U.S. Department of energy (DOE), which states that for light-duty vehicles the hydrogen storage has to be at least 5.5 wt% of hydrogen for 2025.17,18 One decomposition pathway of formic acid results in only gaseous products H2 and CO2, which can be performed catalytically (equation 1).19 The undesired decomposition pathway produces H2O and CO (equation 2), of which the latter is known to be very toxic.
HCOOH (l) ⟶ H!(g) + CO! (g) ∆𝐺 # = – 32.9 kJ/mol (1)
HCOOH (l) ⟶ H!O (l) + CO (g) Δ𝐺# = – 12.5 kJ/mol (2)
Besides the fact that CO is toxic for humans, it also harmful for many catalysts including hydrogenation catalysts.20,21 It often blocks active sites, thereby hindering the catalyst activity.22
2.1 Industrial synthesis
Formic acid is produced industrially in various processes. One large and well-known producer is BASF, which annually produces around 230,000 tons of formic acid.23 Generally speaking, formic acid can be produced via four processes,24 which all are indirectly based on refining fossil fuels:
1. Methyl formate hydrolysis 2. Oxidation of hydrocarbons 3. Hydrolysis of formamide
4. Preparation of free formic acids from formates
The first process is the biggest contributor for the FA production and it consists of two stages.25 The first step is carbonylation of methanol to methyl formate, followed by hydrolysis to formic acid and methanol (Figure 3), the latter of which is recycled back to the first stage. The required CO is produced by steam reforming of coal or natural gas. However, formic acid can also be produced from biomass. This has been reviewed in detail.26,27 Biomass based pathways can operate at much lower temperatures, but require multiple steps and are dependent on the growth of plants.
Figure 3 Generalized reaction for the methyl formate hydrolysis to formic acid.
CO CH3OH HCOOCH3 H2O HCOOH
2.2 Direct CO2 hydrogenation to formic acid
The direct hydrogenation of CO2 to formic acid seems very sustainable and beneficial, not only from the hydrogen storage perspective, but also because the greenhouse gas CO2 is directly used. However, the reaction is thermodynamically very restrained, even though the heat demand for this reaction is less than for the BASF process explained above.23 At room temperature the hydrogenation reaction of CO
2 has a DG° of 32.9 kJ/mol. This reaction can become slight exothermic at higher pressures and by introducing a base to remove FA from the reaction, shifting the equilibrium towards the products. The formed FA–base adducts can be separated by protonation to form the pure FA. However, the formed FA and pure base then also still need to be separated from each other.
Catalytic hydrogenation provides a means to lower the kinetic barrier. The homogeneously catalyzed hydrogenation of CO2 to formic acid has been reviewed extensively.28 Very high turnover frequencies (TOFs) have been reported, the highest TOF of 1,100,000 h-1 achieved by Pidko and co-workers with a Ru pincer catalyst (Figure 4).29 However, high turnover frequencies are not the only measure for good catalyst. Other factors, such as durability, recyclability, and ease of synthesis are also important. Hydrogenation catalysis often makes use of metal-pincer complexes, which will appear throughout this thesis. The general catalytic mechanism of such complexes has been studied as well.30
Figure 4 Ru-catalyst for hydrogenation of CO2 (DMF/DBU, 120 °C, 40 bar H2).
Homogeneous catalysis is not very suitable for large-scale industrial purposes. For scaled-up processes that are necessary for the hydrogen economy, the costs of homogeneous catalysts are too high and not profitable. This is largely due to difficult catalyst synthesis and multi-step separation processes. Heterogeneous catalysis has several advantages for industry, which are mostly associated with lower costs. The CO2 hydrogenation to formic acid and formates has also been reviewed.31,32
Bulushev and Ross conclude that for efficient catalysis adsorption or activation of CO2 and H2 occurs on support sites and on metal sites, respectively.23 Park et al. studied a Pd/g-C
3N4 system, from which they concluded that the reaction between CO2 and H2 takes place at the interface of the metal and the support.33 The hydrogen is adsorbed on the metal, while CO
2 is adsorbed on the support. This type of mechanism explains the importance of the metal particle size.23 Other CO
2 hydrogenation processes often show particle size effects, which is leading to the development of new catalysts. Incorporation of
Ru
CO
(
tBu)
2P
P(
tBu)
2N
nitrogen in the support is also improving catalytic performance, partially because the atomic dispersion of the metals are increased.34
To further illustrate the particle size effect, Figure 5 depicts the mechanism of a gold catalyst on a Al2O3 support (red arrows).35 In this mechanism, the authors also propose that hydrogen is adsorbed on the metal and the carbon dioxide on the support material. Thus, by varying the support material you can achieve higher CO2 adsorption and by varying the metal particle size, you can change the interface. Both of these factors can enhance the catalytic activity.
Figure 5 Proposed mechanism of reversible CO2 hydrogenation to formate over Au/Al2O3.35
Gunasekar et al. reported an Iridium(III)-N-heterocyclic carbene complex supported on a covalent triazine framework (CTF), which catalyzed the hydrogenation of CO2 to formate with an initial turnover frequency of 16,000 h-1 (Figure 6).36 The N-heterocyclic ligand is incorporated in the CTF support. Until then that was the highest reported TOF for heterogeneously catalyzed CO2 hydrogenation (0.68 mol% Ir, CO2:H2=1:1, 80 bar, 120 °C, 0.25h). CTF has gained attention, because they are stable over a broad range of pressures and temperatures and they are inert against attack by functional groups in basic and acidic media.37–39 This is beneficial for industrial purposes. As mentioned previously, the formate is trapped by a amine base, in this case triethylamine, to form the formate–amine adduct as a product. The authors statethat high activity could result from the chelating ligands being good s-donors and poor p-acceptors, resulting in high electron density on the Ir(III) centers. The authors state that previous theoretical studies have showed that this enhances the activity. The catalyst was recycled by filtration and reused for new hydrogenation cycles. The activity does decrease with passing cycles and this is likely due to the Ir-complexes are leaching into the solution.
Figure 6 Iridium(III)-N-heterocyclic carbene incorporated in CTF support.
More recently, the same group reported a similar heterogenized ruthenium catalyst suitable for a fixed bed reactor set-up, including a separation unit.40 By immobilizing the homogeneous catalyst covalently to a support, some advantages of both homogeneous and heterogeneous catalysis are combined, including rationally designed supports with better porosity and heteroatomic ligand sites, which can improve and diversify the catalysts.23,31,32 Most importantly, the facile separation of heterogeneous catalysts can be used. They reported a turnover number of 524,000 for 30 days, with no visible deactivation, and a conversion of 44.4% (triethylamine, 120 ºC, 120 bar). Again, due to improved electron donating ability of the metal center active sites, the activity was good.
The same paper also contains a proposed scheme for the production of formic acid with recycle pathways for water and triethylamine (Figure 7).40 Such a scheme illustrates the applicability of the process. The formic acid formation reactor is supplied with pure feed gas CO2/H2 and also via recycle pathways. The solvents triethylamine and water are supplied only via recycled pathways. The liquid formate adducts (Et3NH+HCOO-), formed in the reactor, move to the evaporator to increase the acid to amine ratio. The formate adducts are then combined with n-butyl imidazole (nBIMH+) and fed to the amine exchange column. There triethylamine is removed and the new formate adduct with n-butyl imidazole (nBIMH+HCOO-) is fed to the formic acid separation column. The authors explain that direct separation of the Et3NH+HCOO- is not achievable, because the proton binds to strongly to the amine.
N N N Ir Cl N N N N N N N N N N N N n n n Cl
Figure 7 Schematic representation of trickle-bed reactor for CO2 hydrogenation.40 2.3 Dehydrogenation of formic acid
The homogeneously catalyzed dehydrogenation reaction of formic acid has also been reviewed.28,41 Laurency and co-workers summarized many findings on the dehydrogenation of formic acid to CO2 and H2. A catalyst with a very high TON of 5,000,000is mentioned (Figure 8).42 In this work an iridium catalyst is used over a period of 3,5 months in a highly concentrated aqueous formic acid solution (40 vol%, 80 °C, 1 μmol of catalyst). The catalyst was recyclable over 10 times without losing activity. In addition, the catalyst solubility is very pH dependent, which results in easy separation by precipitation.
Figure 8 Iridium catalyst for formic acid dehydrogenation with TON of 5,000,000.40
Heterogeneous catalysts for dehydrogenation are also studied.43–45 Yamashita and co-workers summarized the work done with catalysts on carbon supported materials and conclude that the palladium based catalysts are among the most promising, because of their high tolerance to CO. They also show high conversion and selectivity under milder temperatures (ca. 25–100 ºC). In the category of monometallic Pd-catalyst, Zhang et al. reported that the use of the a formic acid–ammonium formate adduct enhanced the activity of the used Pd/C catalyst.46 They ascribed the increased activity to the dissolved formate and also to the better adsorption of ammonium and higher solubility in acidic
Ir Cp* N N OH OH H2O 2+
solutions. It appears that ammonium-formate adducts work better than the more classic sodium formate adducts for the formate dehydrogenation.
Some drawbacks of monometallic catalysts include the quick deactivation by toxic intermediates that adsorb on the catalyst surface, thereby deactivating them. This could be solved by using bimetallic catalysts, which creates different electronic structures that enhance activity and selectivity. However, bimetallic systems with increased TONs compared to monometallic formic acid dehydrogenation catalyst have not yet been reported.45
Yamashita generalized the mechanism for the dehydrogenation of formic acid over Pd/C catalysts (Figure 9).47 Kinetic isotope effect studies have been performed with varying sizes of the Pd nanoparticles. It was concluded that Step II of the mechanism is rate-determining and the rate does vary with different nanoparticle size.
Figure 9 Plausible mechanism for the dehydrogenation of formic acid or carbon supported Pd catalysts.47 2.4 Single catalysts for reversible hydrogen storage in formic acid
A future in which both hydrogenation and dehydrogenation can be performed with the same catalyst, can save considerable amounts of money. Especially for such large-scale reactions as are being envisioned in this thesis and for hydrogen storage in general, one catalyst would be convenient. A large part of papers based around the reversible storage by using a single catalyst usually involve homogeneous catalysts.19,29,48–52 For example, the previously mentioned Ru-PNP catalyst by Pidko and co-workers also proved to dehydrogenate formic acid with a TOF 257,000 h-1. So both the forward and backward reaction work well with this Ru catalyst (Figure 4).
Heterogeneous catalysts are rarely reported in this context. For the separate reactions, a lot of research has been done, of which only a few are mentioned in the above sections, but it is more difficult to find one catalyst used in for the interconversion between CO2 and formic acid. Similarities can be found in the catalysts used for the hydrogenation and dehydrogenation reactions. Pd based catalysts are well reported for both, thus it could be expected to find a catalyst able to do the interconversion.
Pd nanoparticles on mesoporous graphitic carbon nitride (mpg-C3N4) are reported by Lee et al. for CO2 mediated hydrogen storage in formic acid.53 The authors state that it was the first reported viable heterogeneous system for the LOHC formic acid. The nitrogen atoms in the carbon support enhanced the adsorption of the palladium nanoparticles, increasing the dispersion. The dehydrogenation of formic acid was performed under ambient conditions with 9.5 wt% Pd/mpg-C3N4 catalyst (water, 25°C, 3h). Increasing the formic acid concentration in the aqueous solution had a negative effect on the degree of dehydrogenation. Based on the dehydrogenation results and further DFT studies, it was concluded that the nitrogen functionalities in the support also enhanced the stability of the palladium nanoparticles. It provides basic sites for formic acid activation and CO2 activation. The CO2 hydrogenation to formic acid was tested and the highest yield was obtained at 150 °C (triethylamine, 30 bar H2, 24 h).
Masuda et al. report highly dispersed PdAg nanoparticles on phenylamine-functionalized mesoporous carbon (amine-MSC) for reversible hydrogen storage under relatively mild conditions.54 The bimetallic PdAg catalyst worked better compared to the single Pd. Smaller nanoparticle sizes also showed higher activity. For CO2 hydrogenation with the PdAg/amine-MSC in aqueous solution, a TON of 839 is achieved (CO2:H2=1:1, 20 bar, 100°C, 24h). Formic acid dehydrogenation resulted in a TOF of 5638 h-1 (0.91 wt%, 0.25 min, 75 °C). They continued to reuse the catalyst in three consecutive cycles of dehydrogenation and hydrogenation and the catalyst showed no loss of its activity. The authors explain that the addition of the amine to the carbon support enhanced the nanoparticle dispersion and the CO2 adsorption capacity, which was also shown for the previous Pd/mpg-C3N4 catalyst.
The two catalysts reported above are both palladium based and use carbon structures in their support. Generally, carbon supports are widely available and low in costs.55 The reactions are performed at relatively low temperatures (<200°C), which makes them suitable for application. Further mechanistic insight can aid in the improvement of these carbon supported materials.
3. Methanol
Methanol is also widely considered as a chemical hydrogen carrier.16,56–58 Especially compared to formic acid, it has a high hydrogen storage capacity of 12.6 wt%. Methanol is a colorless, odorless, flammable liquid with good miscibility with water. The dehydrogenation process of methanol can occur via three different pathways, depending on the products formed.28 It is either dehydrogenated to formaldehyde and one molecule of hydrogen, to CO and two molecules of hydrogen or to CO2 and three molecules of hydrogen (Table 1, equations 3–5 respectively). Equation 5 and its reverse reaction are to be discussed in the coming sections, but the equations 3 and 4 are often undesired side reactions that also occur. The research on methanol (de)hydrogenation catalysis is extensive and only a selection of systems is described below.
Table 1 Dehydrogenation reactions of methanol and their related Gibbs free energy.28,59,60 DG° (kJ mol-1)
CH$OH (l) → HCHO (g) + H! (g) + 63.5 (3) CH$OH (l) → CO (g) + 2 H!(g) + 29.0 (4) CH$OH (l) + H!O (l) → CO!(g) + 3 H!(g) + 0.6 (5)
Currently, industrial synthesis of methanol by hydrogenation of CO2 already exists, where heterogeneous catalysts are used, mainly of the Cu/ZnO type.28 These processes are still under development and optimization is necessary. The reverse reaction, the steam reforming of methanol where hydrogen is released, is also well known. However, this reaction takes place at very high temperatures (200 – 300 °C) and results in high levels of CO poisoning. This undesirable effect is one driving force for the development of catalysts that operate at low temperatures with negligible CO poisoning.
3.1 Industrial synthesis
Conventionally, methanol is produced from syngas, which is a mixture of CO and H2 that in turn has been formed by the steam reforming of natural gas.61–63 Methanol itself is used as a precursor for many chemicals, such as gasoline or formaldehyde, among many others. There are three existing industrial power plants where methanol is produced on ton scale in a direction conversion from CO2 and H2. One of these plants, the George Olah Renewable Methanol Plant by Carbon Recycling International (CRI) in Iceland, recycles 5500 tons of CO2 anually.64 George Olah was among the first researchers that published about the ‘methanol economy’, which eventually has resulted in a power plant named after him. Another plant in Japan, Mitsui Chemical, aims to produce 100 tons of renewable methanol per year.
3.2 Direct CO2 hydrogenation to methanol
Significant progress has already been made in the direct hydrogenation of CO2 to methanol with heterogeneous catalysis and is applied industrially in the above mentioned power plants. The homogeneously catalyzed hydrogenation has also been reviewed.28,60 The reaction from CO
2 to methanol can in a single step, but remains challenging because of the high activation barrier that requires high temperatures, pressures and often suffers from selectivity issues. Homogeneous catalysts operate less well in these conditions. To avoid this barrier, the indirect hydrogenation of CO2 derivatives, such as carbonate, formate, urea and esters, is also possible. Lastly, methanol can also be obtained by CO2 to formic acid hydrogenation and then subsequent hydrogenation of FA to methanol. The previous chapter already talked about CO2 hydrogenation to FA. The advantage of the indirect methanol synthesis from CO2 derivatives is the lower activation barrier for the separate reactions. Thus, the high energy demand for direct hydrogenation to methanol is avoided by performing two or more separate reactions with lower energy activation. However, this approach can be considered devious. The use of two separate consecutive reaction to afford the loaded LOHC is not efficient and undesirable for its application. In a scaled-up plant, this would lead to increasingly complex infrastructures.
Compared to the high TON’s and TOF’s achieved with homogeneous catalysts for the direct formic acid production, no such high numbers are reported for the direct CO2 to methanol reactions. Recently, an extensive review on the CO2 hydrogenation with heterogeneous catalysts was published.65 The catalysts discussed in the review are of four different types: (1) the conventional Cu-based, (2) precious metal based (Pd and Pt), (3) In2O3 and ZnO catalysts and (4) MOF/ZIF structures.
The authors conclude that for all catalyst types, higher methanol formation rates are achieved with compromised selectivity. The Cu-catalysts achieve highest yield (ca. 35 mol(CH3OH) kg-1 h-1) and at the lowest temperatures (230–250 °C, ca. 60% selective for methanol). Better selectivity, however, is often achieved with lower formation rates. In contrast, for the Pd and Pt based catalysts, the selectivity is not necessarily compromised by lower formation rates but they are generally lower than for Cu-type (ca. 20 mol(CH3OH)kg-1h-1, 230–270°C). In2O3 and ZnO show comparable formation rates to Cu-catalysts and with better selectivity (30 mol(CH3OH)kg-1h-1, 270–290°C, 80% selective). However, these reactions than require higher temperatures. The MOF/ZIF-derived catalysts show comparable selectivity to In2O3 and ZnO, but at lower temperatures (210–290 °C). Overall, the Cu-type catalyst work best.
These Cu-type catalysts, that are also used in syngas-to-methanol and as well as in CO2-to-methanol synthesis, have been investigated extensively. Even though they perform well, innovation is necessary, because they still suffer from short life-times and poor activity at lower temperatures.66 Mechanistically, the same principle applies for methanol formation as for formic acid formation. The H2 molecules adsorb
to the metal, forming hydrides, and the CO2 adsorb to the support material. This type of mechanism is illustrated in Figure 10.
Figure 10 Sketch of the catalysts surface and functionality of the various surface sites under CO2
hydrogenation conditions (160–200°C, 10 bar, CO2:H2:N2=3:9:1).67 3.3 Dehydrogenation of methanol
As seen in Table 1, there exists various decomposition pathways for methanol. In the aqueous reforming of methanol, the desired products are formed.60 The previous section also discussed about the fact that formic acid or formates are common intermediates in the steps for CO2 hydrogenation and for the reverse this is also true. The different dehydrogenative pathways often occur simultaneously and with increased temperature, the amount of carbon monoxide produced also increases.56 Furthermore, water has a great influence on the CO produced.68 Low-temperature dehydrogenation (< 100°C) can be achieved with Ru-pincer like catalysts and other homogeneous catalytic systems are also reported.69–71
Traditionally, catalysts used in hydrogen production from methanol are copper based. In this methanol steam reforming (MSR) process, which requires temperatures between 200 and 300 °C, the Cu/Zn/Al2O3 is used.60,72 The copper is the active metal and ZnO prevents the sintering of the copper at higher
temperatures. The alumina support provides high surface area for well dispersed metal particles, like in most heterogeneous catalysts. The Cu catalyst works well for methanol dehydrogenation but the high temperatures result in catalyst deactivation over time.73
In a recent review of catalytic conversion of C1 molecules, the authors conclude that the main challenges for methanol conversion are limiting the CO formation, which increases with increased temperature, and simultaneously achieving high activity at these lower temperatures.74 Tsang and co-workers have studied a CuZnGaOx catalyst that operates at milder temperatures (150–200 °C) without detectable CO levels.75 They found a correlation between the Cu particle size and the activity for the methanol dehydrogenation, where smaller particles give higher conversion (3.76 nm, 22.5% CH3OH conversion, 150 °C).
Cai et al. studied Zn modified Pt/MoC catalysts (Pt 2.0 wt%, Zn 0.5 wt%, 120 °C, 1 bar, 2.5 h).73 The addition of Zn, improved the Pt dispersion and the interaction between the active Pt sites and the a-MoC1–x surface. Stability was good and the CO selectivity remained low. Figure 11 depicts the mechanism of methanol decomposition over this catalyst. The first step involves the oxidation of methanol to formaldehyde, which is then attacked by nearby methoxy groups to form intermediate methyl formate species. These are hydrolyzed by water forming formic acid, after which decomposition to CO2 and H2 occurs. The authors mention that the direct nucleophilic attack of water on formaldehyde is another route for the formation of CO2 and H2. However, isotopic labeling showed that this did not occur.
Figure 11 Schematic representation of the mechanism of MSR over Zn-modified Pt/MoC catalyst.73 3.4 Reversible hydrogen storage in methanol
The research on single catalysts for both hydrogenation and dehydrogenation in methanol is scarce. There are industrial plants for both CO2 hydrogenation to methanol. As was mentioned in the section about reversible catalysts for formic acid, the same can be said for methanol. The research on the separate reactions is extensive, but much less is found on interconversion of CO2 and methanol for hydrogen storage. Methanol can also be used in other types of dehydrogenative coupling reactions, other than pure decomposition to CO2 and H2. These types of reactions will be covered in chapter 6.
The Cu/Zn/Al2O3 catalysts work for both reactions, thus it can be envisioned that these can be tested in comparable conditions. The reported conditions are often extreme, so it might be more useful to develop Cu-based catalyst that can operate at milder conditions.
4. N-ethylcarbazole
There exists a wide range of heteroaromatic compounds that could be potential LOHCs. Among these is the widely researched N-ethylcarbazole (NEC). It has a hydrogen storage capacity of 5.8 wt%, when H0-NEC is hydrogenated to H12-NEC (Figure 12).
Figure 12 Reversible hydrogen storage of N-ethylcarbazole.
Pez et al. from Air Products have conducted theoretical and experimental research, which showed that that the introduction of hetero-atoms in aromatics lowers the dehydrogenation enthalpy, because the aromaticity is decreased.76 When comparing the hydrogenated N-carbazole to N-ethylcarbazole, it was showed that the alkyl group decreases the adsorption strength of the dehydrogenated product H0-NEC on the catalyst surface, which prevents catalyst deactivation. Thus, NEC is more readily (de)hydrogenated compared to other aromatic or heteroaromatic compounds.77 However, H0-NEC is a solid compound at room temperature with a melting point of 64 °C,7 which is not necessarily desirable. Additionally, the N-alkyl bond is relatively labile and decomposition occurs at temperatures above 270 °C.78 But since practical applications rather require milder reaction conditions, this is not considered a major issue.
4.1 Commercial synthesis of NEC
H0-NEC can be synthesized via various methods.79,80 Most commonly it is prepared by alkylation of carbazole.65 These routes often produce large amounts of inorganic salts as waste, which are difficult to remove. An organic synthesis route also exists where carbazole reacts with diethyl carbonate.81 The synthesis of the precursor carbazole is also possible in many different ways,83 most of which indirectly or directly use refined chemicals that originate from fossil fuels. It is also worthwhile to mention the possibility to synthesize H0-NEC from biomass–derived compounds, such as cellulose or lignin. Biomass presents a more sustainable alternative to the synthesis of various common chemicals, even though the ‘carbon neutrality’ of biomass is under debate. Several suggestions have been done in which carbazole type compounds are produced with from biomass derived starting materials,84,85 but the relevant pathways are outside the scope of this study and also not yet explicitly for NEC synthesis.
4.2 Hydrogenation of H0-NEC
Tsang and co-workers reported a heterogeneous ruthenium catalyst on a Al2O3 support, for the full hydrogenation of H0-NEC with a catalytic activity of 8.2 moles g-1 s-1 and a selectivity of 98%.86 The
N
N + 6 H2
– 6 H2
activity was measured in the moles of H0-NEC converted per gram of metal per second. The Ru-catalyst was first chemically reduced by sodium borohydride. The authors state that mass transfer limitations may occur for this reaction because the H0-NEC is used directly as a molten reactant, instead of using it as a solvent. Further studies were conducted on the reaction and appeared that the hydrogenation occurs stepwise (Figure 13).87 After comparing various ruthenium on Alumina catalysts, it was shown that the smallest particle size gave the best conversion (100%) and selectivity (ca 95% at 130 °C, 70 bar H2, 3h) . This phenomenon was also seen for the heterogeneous catalyst for CO2 hydrogenation.
Figure 13 Simplified stepwise hydrogenation of H0-NEC.
Further experimental results and calculations lead to the conclusion that the side rings are hydrogenated on the terrace sites of the catalyst surface. Because of the increased steric hindrance experienced by the pyrrole after these first hydrogenation steps, the molecule desorbs of the terrace sites before they are adsorbed again in lower coordination on the ‘corners’ on the catalyst surface (Figure 14).
Figure 14 (a) Calculated structures of H0-NEC adsorbed on terrace site, (b) Kinetically favored desorption
of H8-NEC and (c) H0-NEC adsorbed on a ‘step’ site.88
These Ru-type catalyst are far the best performing catalyst and in many papers are assumed as the standard hydrogenation catalyst for NEC.89–91
4.3 Dehydrogenation of H12-NEC
As has been mentioned in the introduction, dehydrogenation is generally endothermic and thus more difficult. Yang et al. compared several commercial catalyst for the dehydrogenation on alumina and found that catalytic activity followed the order of Pd > Pt > Ru > Rh (5 wt% M/Al2O3, 180 °C, 5h, 1
N
N N
N N
bar).92 They continued to do a kinetic study on the palladium catalyst and it was found that the dehydrogenation also occurred stepwise.91 For their experiments, they first hydrogenated H0-NEC to H12-NEC with a Ru catalyst, after which they performed the dehydrogenation reaction with very low Pd catalyst loading to slow down the reaction. Assuming that diffusion and surface adsorption is faster and thus only the surface reactions determine reaction rates, they could distinguish three dehydrogenation steps. In each step two molecules of hydrogen are released and the last step is rate-limiting. As might be expected when looking at the reverse (Figure 13), the pyrrole ring is dehydrogenated first.
4.4 Single catalysts for reversible hydrogen storage in NEC
Considering that the hydrogenation reactions occurs stepwise and the dehydrogenation reactions occurs via the reverse stepwise mechanism, it can be imagined that one catalysts can perform both reactions depending on the applied conditions. Very recently, Yu et al. reported a LaNi alloy for the reversible hydrogen storage in NEC.88 They illustrated the mechanism of both reactions in Figure 15. Using techniques such as XRD, SEM and TEM they elucidated how both processes can occur on this catalyst.
Figure 15 Illustration of (a) LaNi5 nanoparticle, (b) H0-NEC hydrogenation and (c) 12H-NEC
dehydrogenation.88
In the experiment various different LaNi catalysts are compared. The alloy LaNi5+x (0 < x < 1) appeared to perform best for both hydrogen uptake and release (hydrogenation: 180 °C, 8h, 70 bar H2, dehydrogenation: 200 °C, 6h, 1 bar). The authors explain that by introducing more Ni to the catalyst, more hydrogen binding sites are provided. The H0-NEC/H12-NEC is then adsorbed on the Ni nanoparticles present on the surface and H2 adsorption takes place in the whole LaNi lattice, forming metal-hydrides. Because the La2O3 metal-oxide surface is amorphous, the hydrogen molecules can easily be adsorbed. They further compared the ratio of the two metals in LaNi5+x and concluded that LaNi5.5 was optimal.
5. Dibenzyltoluene
Industrially, dibenzyltoluene (DBT) is used as a heat transfer oil.56 It is marketed under tradenames Marlotherm SHÒ or Farolin WF0801Ó, with prices under 4 € per kg and produced on ton-scale.93–95 It has desirable properties, such as low-toxicity, absent carcinogenicity or mutagenicity and has a high stability. It exists in isomeric mixtures with a melting point at around –32 °C and a boiling point around 390 °C.96,97 It has a hydrogen storage capacity of 6.2 wt% and takes up nine molecules of hydrogen in a hydrogenation reaction (Figure 16).
Figure 16 Reversible hydrogen storage in dibenzyltoluene.
5.1 Commercial synthesis of DBT
Generally speaking, DBT is synthesized via Friedel-Crafts alkylation of toluene with benzyl chloride. Over the years this process has been changed and improved, where now multiple routes exist using different catalysts, different separation method or using different reaction conditions.98–102 Figure 17 depicts a general reaction equation for DBT synthesis.
Figure 17 Simplified reaction equation of the Friedel-Crafts alkylation of monobenzyltoluene to
dibenzyltoluene (>95% conversion, 1–10 mol% of Ti2O3, 100–130°C, 5–8h).99 5.2 Hydrogenation of H0-DBT
Much research has been done on the hydrogenation process of H0-DBT. This specific LOHC is often evaluated in reversible systems, where both hydrogenation and dehydrogenation reactions are performed. But these reactions are not necessarily performed with the same catalyst. Bruckner et al. compared the DBT as an LOHC with two other compounds. Hydrogenation was performed with a Ru-catalyst under 50 bar of H2 (150 °C, 0.25% mol Ru/Al2O3, 1h).103 Over 99% of hydrogen was bound after 1 hour and full loading was achieved after 1.5h. They further studied the hydrogenation process and concluded that the benzyl side rings are hydrogenated first and the middle toluene ring is hydrogenated last (Figure 18).104 It is even suggested that a different catalyst is used for hydrogenation of the last ring, by combining different ones in a catalyst bed.
+ 9 H2 – 9 H2 H0-DBT H18-DBT Cl + + HCl cat.
Figure 18 Simplified scheme of the hydrogenation of H0-DBT, where the tolyl-rings are hydrogenated
first.
A different approach has also been used by the group of Wasserscheid, where multicomponent hydrogen rich gas mixtures were used for hydrogenation of DBT, i.e. mixtures of hydrogen with CO2, CO and CH4.14 Conventional hydrogen production requires extensive purification steps to obtain pure hydrogen. By using gas mixtures, these steps are essentially replaced by the hydrogenation of DBT and the hydrogen is directly stored in the LOHC. They compared different catalyst on alumina support, including Ru, Rh, Pt and Pd of which only the Pd showed no deactivation when model gas mixtures were used. It was 99.5% selective for hydrogen-to-DBT and the unloading of H18-DBT resulted in 99.99% pure hydrogen gas.
This concept is interesting for selectivity purposes, for knowing if an LOHC or catalyst does not react with the other gases, but it is futile if we desire to get rid of steam reforming altogether. However, in the introduction it was mentioned that most of the hydrogen is consumed in places where it also produced. By implementing LOHC hydrogenation already in this way, current hydrogen transport can already be replaced by LOHC transport. A dehydrogenation reactor unit is then still required to eventually obtain the hydrogen.
5.3 Dehydrogenation of H18-DBT
For the dehydrogenation of dibenzyltoluene, heterogeneous platinum catalysts are frequently mentioned as suitable catalysts.16,105,106 Bessarabov et al. concluded that Pt performs better over similar Pd catalysts.107 Various reaction conditions were tested and it was concluded that 2 mol% Pt/Al
2O3 at a temperature of 320 °C gave the highest degree of dehydrogenation (96%). They further studied the mechanism with ab initio calculations. Surprisingly, the same order as hydrogenation is followed in dehydrogenation, where the side rings are dehydrogenated first and the middle ring last (Figure 18). Wasserscheid et al. also conclude that dehydrogenation works well with Pt but on a carbon support, achieving 97% degree of dehydrogenation at 310 °C with 1 mol% Pt/C.108
5.4 Reversible hydrogen storage in DBT
Several hydrogen storage system are designed where DBT is taken as the LOHC, that are on larger scale than lab experiments. The Fuel Cells and Joint Undertaking (FCH JU), mentioned in the introduction, are studying an LOHC system with dibenzyltoluene as a part of the HySTOC project. 109 In collaboration with LOHC Technologies Hydrogenious, storage and release are designed for actual application. Another company, Clariant©, supplies the catalysts for hydrogenation and dehydrogenation called the EleMax series, which are Pt and Pd on aluminum oxide supports.110,111
The Hydrogenious S-Series StoragePLANT 12tpd has a hydrogen storage capacity of 12 tons of hydrogen per day, or 500 kg per hour. This produces 12,000 litres of LOHC per hour. In a range of 30– 50 bar hydrogen in 99,99% purity is supplied. The S-Series ReleasePLANT 1.5tpd releases hydrogen at a rate of 1.5 tons per day, or 64 kg per hour. This difference is quite significant and accurately illustrates that dehydrogenation is energetically much less favorable and much harder.
6. Alcohol reforming
The following chapters will highlight a selection of more specific systems that propose a different take on liquid organic hydrogen carriers. Rather than simply loading and unloading one molecule with hydrogen, you can also react small molecules in dehydrogenative coupling reactions to form new compounds while releasing a certain amount of hydrogen. These reactions can be reversed in hydrogenative cleavage. Chapter 6 will cover reactions defined as alcohol reforming, in which alcohols couple to form esters. Chapter 7 will discuss amine reforming, where amide bonds are formed by combining amines with alcohols. These terms are analogous to the well-known and already mentioned steam reforming processes where water is reacted with natural gas to form syngas (CO + H2).
The principle of alcohol or amine reforming has several advantages compared to traditional steam reforming, in the context of sustainable hydrogen storage and transportation. The systems are desired to be reversible with the same single catalyst. The dehydrogenated products are soluble and do not have to recaptured and thus, can be used immediately for the hydrogen-loading. Lastly, pure hydrogen is formed and there is no need for gas separation.
6.1 Homogeneous catalysts for reversible alcohol reforming systems
The first system to be discussed has been proposed by Milstein et al, wherein ethylene glycol (EG) is used as the LOHC in a reversible system.112,113 The ethylene glycol is dehydrogenated to an ester-oligomer, catalyzed by a ruthenium pincer complex. EG is a widely available chemical, produced worldwide in over 34 million tons each year. It is used in a very large variety of products, such as antifreeze agents and coolants or as a precursor for polyester fibers, to name a few.114
Ethylene glycol is synthesized from fossil fuels mostly, but can also be made out of biomass. In industry it is produced via hydration of ethylene oxide, the result of the oxidation of petroleum-derived ethylene.115–117 This process was first proposed in 1937 by the Union Carbide Corporation and has been developed and improved with catalysts for higher selectivity.118 It is also produced from syngas.119 These routes are all fossil fuel based, but the bio-based synthesis is also intensively explored. Zhang et al. reported the one-pot conversion of cellulose to ethylene glycol, catalyzed by tungsten-catalysts in a cascade reaction.120 Chemistry technologies company Avantium produces ethylene glycol from plant-based sugar with their Ray TechnologyTM.121 It converts sugars, from sugar beet, sugar cane, wheat and corn, via hydrogenolysis to EG. Currently, a demonstration plant based in the Netherlands is in progress and it is planned to open a flagship plant in 2024.
For the dehydrogenative coupling of EG (Figure 19) several Ru-catalysts were screened by Milstein and co-workers.113 The best results were achieved with the complex depicted in Figure 20. The reaction was tested without added base in a 1:1 v/v ratio of toluene/DME with 1 mol% of catalyst. They measured 97% conversion and 61 mL of formed hydrogen, which is equal to a 64% yield (150 °C, 72h). After these encouraging results, they explored the recovery of ethylene glycol from the products obtained in the dehydrogenation. Under similar conditions, the reaction mixture was hydrogenated back to EG in 92% NMR yield. The reaction was performed under 40 bar of hydrogen over a period of 48 hours at 150 °C. The dehydrogenation was then also performed in neat EG and under a partial vacuum. This reaction also succeeded with 94% conversion after one week and the hydrogen yield was further increased to 75%. This experiment also showed a realized hydrogen storage capacity of 5.2 wt%, lower than the first estimated capacity of 6.5 wt%. The crude product mixture was then also tested for hydrogenation under the same conditions as the previous hydrogenation reaction, which worked and the oligomer mixture was fully hydrogenated.
Figure 19 Reversible dehydrogenative coupling reaction of ethylene glycol.
Figure 20 Ru-catalyst used in the reversible dehydrogenation reaction of ethylene glycol.87
Computational studies were conducted with this catalyst to elucidate and propose a mechanism (Figure
21) and calculate the related free energies of intermediates and transitions states. The first three steps of
the cycle consists of the coordination of EG to complex A, the dehydrogenation by protonation of the hydride, forming alkoxide species B, followed by a b-hydride elimination to C. In the transition state from B to C (TSBC) the glycolic OH interacts with N-heterocyclic pincer-ligand. This creates a vacant
site and allows for the correct geometry for b-hydride elimination. Complex C includes a new Ru–H bond is formed and an coordinated aldehyde. Generally, an alcohol coordinates more strongly than an aldehyde, but this is hindered by the ligand interactions. Another molecule of ethylene glycol comes in and attacks the carbonyl. Via a six-membered chair like transition state, the Ru–H is protonated and hydrogen evolves to form D, a hemiacetal like species. Another b-hydride elimination, resulting from
HO OH Ru, – H2 Ru, + H2 O O O O O O H H x y
Ru
OC
P
P
N
H
iPr
iPr
iPr
iPr
the similar interaction with the ligand backbone, and subsequent dissociation of the ester product, gives back the starting complex.
Figure 21 Proposed catalytic cycle for the Ru catalyst with related Gibbs free energies (kcal mol-1, with
respect to the starting material) at 423.15K.113
More recently, another paper was published where ethylene glycol is used in reversible hydrogen storage by combining it with ethanol (Figure 22).112 For this reaction, similar Ru-pincer complexes were tested, from which the complex in Figure 20 gave the best conversion, as was also the case in the previously described reactions. In a larger scale reaction, a 71% yield of hydrogen was achieved in the presence of 1 mol% Ru-catalyst (150 °C, 120h). The resulting ester mixture was hydrogenated back to a mixture of ethanol and EG under 50 bar of hydrogen pressure (1 mol% Ru, 150 °C, 48 h).
Figure 22 Dehydrogenative coupling reaction of ethanol with ethylene glycol to a mixture of linear esters.
HO OH Ru, – H2 Ru, + H2 OH O O O O O O O O O O + +
6.2 Heterogeneous catalysts reversible alcohol reforming systems
Dehydrogenation
The proposed by Milstein provides very promising insight into the future application of ethylene glycol as a LOHC. As mentioned by the authors themselves, however, the reaction is still far from practical application, even though the homogeneous catalysis on acceptorless alcohol dehydrogenation is extensively reviewed.122 In theory it should be possible to find a heterogeneous catalyst that can dehydrogenate ethylene glycol. Reviews exist on the acceptorless dehydrogenation (AD) of alcohols.123 Because AD reactions do not require sacrificial acceptors, hydrogen is formed, which can be used as fuel.
Shimizu et al. summarize that Cu-type catalysts have been reported for small alcohol to ester transformations. However, many of these reaction still require harsh conditions. Milder reactions (<200 °C) have been reported with Ru-pincer catalysts immobilized on a support, but these require the addition of a base and only works for specific alcohols.124 Pt on SnO
2 works for a variety of alcohols and the mechanism has been proposed as well. Figure 23 displays a proposed mechanism for the dehydrogenative coupling with a Pt supported catalyst by Shimizu (neat, 180 °C, 36 h).
Figure 23 Reaction scheme of the dehydrogenative coupling of primary alcohols to esters.125
The AD of diols is also discussed by Shimizu in reactions to form lactones or cyclic esters. The cyclisation of a single molecule of ethylene glycol is difficult because a strained three-membered acetolactone would form. However, much can be learned by looking at similar reactions. The same Pt/SnO2 catalysts have also been reported to cyclize such longer diols.126
The proposed mechanisms of the dehydrogenative coupling generally proceed first via the dehydrogenation of one alcohol to the aldehyde, which than coordinates to the support. Then the second alcohol comes in to form a hemiacetal species and eventually forms the ester, while releasing a second molecule of hydrogen.127,128 This mechanism might also work for ethylene glycol. Comparing it to the mechanism by of the Ru catalyst used for ethylene glycol coupling (Figure 21), similarities can be found. The first dehydrogenative steps of the pincer mechanism are also to form an aldehyde from an
alcohol. Next, a new molecule of EG comes in and a coordinative hemi-acetal species is formed while releasing a second molecule of hydrogen, followed by another b-hydrogen elimination resulting in the product. This is somewhat similar to the heterogeneous mechanism, where the second alcohol attacks the coordinated aldehyde, also forming an hemiacetal. The last step involves platinum hydride formation to form hydrogen and simultaneous dissociation of the product.
Hydrogenation
For the reverse reaction we have to look at hydrogenation reactions. Pidko et al. reviewed homogeneous and heterogeneous catalysis for hydrogenation of carboxylic acid derivatives,129 including reactions to produce ethylene glycol.130 Mechanistically, homogeneous catalysis for hydrogenation is well understood. For heterogeneous catalysts, the authors conclude that bimetallic catalysts work well, where one metal facilitates the hydrogenation, while the other metal activates the carbonylic groups. More can, however, be uncovered about the mechanism of hydrogenation.
Several attempts have been made in elucidating how ester hydrogenation can occur with heterogeneous catalysts. Narasimhan and co-workers performed hydrogenation of methyl oleate with Ru-Sn-B on alumina (80% conversion, 7h).131 The contact between the active Ru and boron resulted in higher electron density on the Ru surface, enhancing catalytic activity. The Sn promoted higher dispersion of Ru and the Lewis acidity of Sn polarizes the carbonyl bonds. Barrault et al. further studied the mechanism and concluded that, with the used substrates, it proceeds via a hemiacetal species (Figure
24).132
Figure 24 Mechanism of ester hydrogenation with Ru-Sn-B heterogeneous catalyst.129
The role of the solvent has also been studied. It was found that water facilitates heterolytic H2 cleavage by being a proton acceptor, while in organic solvent, the ester moiety itself acts as a base. Hydroxide groups on the catalyst surface, for example Al-OH, also promote cleavage through hydrogen bonding.
Hydrogenation of methyl propionate to 1-propanol could be achieved with Ru-Pt/AlOOH in 90% yield (180 °C, 50 bar H2).133
Cu-based catalysts have been studied for the hydrogenation reaction of diesters, including dimethyl oxalate to ethylene glycol. Conversion of 100% with > 90% selectivity for EG could be achieved with hexagonal mesoporous silica (HMS) as the support (5%wt Cu/HMS, 200°C, 25 bar H2).134 The yield strongly depended on which type of copper precursor was used. 98% ethylene glycol yield was achieved with Cu(NH3)4(NO3)2, due to high dispersion of active Cu particles.135
7. Amine reforming
The following section will cover similar types of reactions as described above with alcohols, but instead of alcohols, amines are used as nucleophiles to form amide bonds. This type of transformation will be termed amine reforming.136 In this case, different types of amines react with alcohols to form small amides or peptides and release hydrogen gas.137–139
7.1 Homogeneous catalysts for amine reforming systems
There are several papers in which the same principle is applied of an amine group dehydrogenatively coupling to an alcohol group. All of these reaction are catalysed by Ru-pincer complexes, that differ in the type of pincer ligands used. The group of Milstein has published many articles on reversible hydrogen storage, of which the alcohol reforming is described in the previous section. They have also published about the dehydrogenative amidation reaction. They are known for using pincer complexes that have appeared throughout this text. One of these publications describes the reaction of 2-aminoethanol with itself to either its cyclic product glycine anhydride (GA) or to linear peptides (LP) (Figure 25).138
Figure 25 Dehydrogenative coupling reaction of 2-aminoethanol to the products glycine anhydride or linear
peptides.
A conversion of 89% was achieved with 0.5 mol% of the pincer complex shown in Figure 26, resulting in a yield of 55% for GA (1.2 mol% KOtBu, dioxane, 12h, 105 °C). They continued to investigate the hydrogenation of GA back to the starting compound 2-aminoethanol and obtained a yield of 96% with 70 bar of hydrogen (0.5 mol% Ru, 1.2 mol% KOtBu, dioxane, 12h, 110 °C). The (de)hydrogenation
reactions were repeated without adding new catalyst and observed that 81% of 2-aminoethanol was still formed in the third cycle (i.e. the sixth reaction with the same catalyst). They added a new amount base KOtBu in each reaction to protect the catalyst from water that might be taken into the system during the
transfers. It is known that water reacts with alcohols to form carboxylic acids, in the presence of pincer PNN complexes.140
Figure 26 Ru-pincer catalyst for the reaction of 2-aminoethanol.
HO NH2 N H H N O O HO H N NH2 O + n Ru, – H2 Ru, + H2 Ru Cl N OC P HN H tBu tBu
The authors proposed a mechanism (Figure 27), which is very different from the mechanism for dehydrogenative coupling in Figure 21. The mechanistic studies were performed with a slightly different catalyst. The tBu-NH-arm of the pyridine is replaced with an Et
3N-group. Deprotonation by KOtBu results in the non-aromatic complex 2, where the mechanism starts.
First, 2-aminoethanol coordinates to complex 2 with simultaneous protonation of the basic ligand arm by the incoming alcohol group. The NNP-ligand is now aromatic. In the same step the diethyl-amine arm is replaced by the primary amine, which coordinates more strongly. Then, the first dehydrogenative step occurs from 6 to A, where the acidic proton of the pyridine arm is removed with a hydride from the alkoxy ligand. The newly formed coordinated aldehyde is susceptible for nucleophilic attack. The amine of a second molecule of 2-aminoethanol attacks, while simultaneously protonating the dearomatized pyridine arm again to form B. The second dehydrogenative step to C is very similar to the step from 6 to A. In the following step from C to D, the alcohol group protonates the pyridine arm to form a new alkoxy species, which dearomatizes to release hydrogen and forms a new coordinated aldehyde. The following steps can either be the amine attacking the aldehyde intramolecularly to give glycine anhydride (via F) or a new incoming 2-aminoethanol attacks the aldehyde to form the linear peptides. All the catalytic steps are based on the aromatization and dearomatization of the pyridinic NPP pincer ligand. This is evidently very different from the mechanism discussion in Figure 21, which was based on the interaction between -OH groups and the PNP pincer backbone.
In another paper by Milstein et al., hydrogen is generated by reacting ethylenediamine with ethanol (Figure 28), catalyzed by a different Ru-pincer catalyst.139 The reactions conditions for this reaction were optimized and 100% conversion for both ethylenediamine and ethanol was measured using the catalyst in Figure 29 (0.2 mol%, dioxane, 105 °C, 24h). Again, the reverse reaction was also performed and over 99% yield of ethylene diamine was recovered (0.4 mol%, dioxane, 70 bar H2, 115 °C). As for the previous reaction, catalytic amounts of KOtBu is added for in situ generation of the active
deprotonated complex.
Figure 28 Dehydrogenative coupling of ethylene diamine with ethanol to a mixture of amides.
Figure 29 Ru-catalyst for the dehydrogenation reaction of ethylenediamine and ethanol.
Olah et al. reported an efficient reversible hydrogen carrier system where methanol is reacted with primary or secondary diamines to form cyclic urea or diformamide, respectively (Figure 30(a)).137 A Ru-pincer catalyst was used for the first set of reactions (Figure 31). The primary amines did not result in high yields of urea compounds, so they moved to the secondary amine N,N’-dimethylethyleneadiamine. The theoretical hydrogen storage capacity is at 5.3 wt%. The highest yield of H2 was achieved with a 1:4 ratio of diamine:CH3OH in the presence of 5 mol% K3PO4, which is used as a base additive to enhance catalytic activity by deprotonating the catalyst.
Figure 30 Reversible dehydrogenative coupling of methanol with (a) primary ethylenediamine to cyclic urea
(6.6 wt% H2) or (b) secondary diamine to N,N-diformamide (5.3 wt% H2).
H2N NH2 OH H N O N H O H N O NH2 N NH2 + Ru, – H2 + + Ru, + H2
Ru
Cl
N
OC
P
N
H
tBu
tBu
CH3OH NH2 NH2 Ru, – H2 Ru, + H2 N H H N O + CH3OH NH NH Ru, – H2 Ru, + H2 N N + H O H O (a) (b)Figure 31 Structure of Ru-catalyst active for amine reforming with methanol.
The optimized conditions were then used to test variations of the Ru-catalyst. The catalyst in Figure 32 resulted in 86% hydrogen yield (toluene, 5 mol% K3PO4, 120 °C, 24h) and no CO was detected, which was the case for the previous catalyst (Figure 31). The authors further explain that the CO poisoning is often associated with a competing pathway, in which the first dehydrogenation product formaldehyde is directly dehydrogenated again to CO and H2, before nucleophilic attack by the amine can occur (Figure 33). The reverse reaction, the hydrogenation of N,N-diformamide, was also successfully performed under 60 bar of hydrogen pressure, resulting in 95% yield of the diamine (toluene, 120 °C, 24 h, K3PO4)
Figure 32 Selected Ru-catalyst in amine reforming with methanol.
Figure 33 Simplified pathway for dehydrogenative coupling of amine with methanol (Pathway 1) and the
undesired decomposition to CO (Pathway 2).137
7.2 Heterogeneous catalysis for amine reforming systems
Dehydrogenation
In the review by Shimizu, the coupling of amines and alcohols to amides is also briefly described.123 All of the reported reactions with heterogenous catalysts are performed with benzylic alcohols combined
Ru
H
N
P
H
P
CO
Ph
Ph
Ph
Ph
H
BH
3Ru
Cl
N
P
H
P
CO
iPr
iPr
iPr
iPr
H
with secondary amines, which originate from just two articles.141,142 Ag on alumina (4 mol%) is able to form amides via a dehydrogenation reaction in the presence of the weak base CsCO3 with a yield up to 93% (reflux, toluene, 24 h). For the mechanism it is suggested that the reaction proceeds via dehydrogenation of the alcohol to aldehyde, which then stay adsorbed to the surface. Reaction with the amines then result in hemiaminal species, followed by dehydrogenation to the amide. (Figure 34)
Figure 34 Simplified mechanism of amide formation from dehydrogenative coupling of an alcohol and an
amine.141
The same reactions were also possible with Au particles on hydrotalcite (HT) supports, but at lower temperatures (70–90 °C, Au/HT (1.9 mol%), KOtBu (120 mol%), o-xylene, 24–48h).142 This reaction worked only with cyclic secondary amines, while the primary and non-cyclic secondary amines afforded imines or amines as the major product. From control reactions it was concluded that for the Au catalyst the mechanism goes via the formation of an ester, after which the amide is formed. Dehydrogenative coupling of amines with alcohols can evidently also result in other types of compounds being formed than amides (e.g. imines, indoles, pyrroles). A more generalized mechanism for the coupling of small amines (such as aminoethane) with alcohols (such as methanol or alcohol) is yet to be reviewed for heterogeneous catalysts.
Hydrogenation
Recently, Beller and co-workers reviewed homogeneous and heterogeneous catalysis for amide hydrogenation.143 The first reported amide hydrogenation reactions occurred at temperatures over 200 °C and high hydrogen pressures (200–300 bar), which is not desired for industry. Additionally, the selectivity in amide hydrogenation proposed an issue. The hydrogenation of amides can occur via two routes.144 In the C–O bond cleavage pathway an imine is formed, which is hydrogenated to an amine. In the C–N cleavage pathway, an amine and aldehyde are formed, after which the aldehyde is hydrogenated to an alcohol. (Figure 35) Analogous to the transformations mentioned in the previous section, the C– N cleavage is the desired pathway.
