Table of Contents:
Chemical Looping Reform for Selective Oxidation of Methane
There are two major routes to convert natural gas into higher value products: Indirect oxidation and direct oxidation. The indirect oxidation approach converts natural gas to an intermediate, syngas, which is then further converted to desired chemical products. Applying chemical looping in this indirect oxidation approach has been described in section 2. The direct CLR selective oxidation approach, however, converts natural gas directly into the desired value-added chemical products. By removing the intermediate syngas production process, the direct oxidation approach simplifies the overall process operations and, thus, can reduce the capital and operating costs. The major direct oxidation processes of methane to chemical products include selective oxidation to methanol, to formaldehyde (HCHO), and to ethylene and other olefins (Sinev et al., 2009; Wang et al., 1995; Anders, 2009). ARCO and DuPont have developed pilot scale CLR selective oxidation pilot plants. However, there are no commercial-scale CLR selective oxidation systems that exist to date. The highly stable methane molecule poses a challenge for industrial sized direct conversion methods for natural gas to chemicals production. For instance, the Oxidative Coupling of Methane (OCM) reaction must take place at a temperature of around 500-1000 °C due to the stability of the methane molecule (Gesser et al., 1985; Lunsford. 2000; Alvarez-Galvan et al., 2011; Ruiz-Martinez et al., 2016; Fleischer et al., 2016). With this high temperature, the commercialization of such a reaction system is also hindered by the high CH4 conversion with low selectivity issue.
Chemical Looping Oxidative Coupling of Methane (CLOCM)
The principle of the chemical looping scheme has been widely used in OCM, especially in pioneering OCM work. Historically, the CLOCM process has been referred to as the reducible catalytic metal oxide OCM approach or the OCM redox approach. The redox cycles and reducible catalytic metal oxides of CLOCM processes are conceptually analogous to redox cycles and oxygen carriers used in CLR processes. However, the controlling factors for the two differ, as the chemical reactions in the CLR processes are dictated by thermodynamics while the reactions in CLOCM are dictated by reaction kinetics. In this sense, the catalytic metal oxides in CLOCM behave more similarly to the functionality and mechanism of catalytic metal oxides in conventional OCM processes.
In typical OCM processes, methane and molecular oxygen are со-fed to a reactor filled with catalyst particles (Armor, 2014). Generally, the co-feed OCM takes place through a heterogeneous-homogeneous reaction pathway (Zavyalova et al., 2011). The desired homogeneous gas-pliase reactions compete with heterogeneous reactions which generate thermodynamically stable products of CO and CO,. As a result, hydrocarbon selectivity is normally inversely proportional to the methane conversion. The highly reactive C, intermediate may further take part in the unwanted yet thermodynamically favorable oxidation reactions (Tiemersma et ah, 2012). The key parameter that controls this relationship of co-feed OCM performance is the partial pressure of gaseous oxygen (Stangland, 2015). Several micro-kinetic models for co-feed OCM suggested that reactions responsible for CO/CO, generation are more dependent on oxygen partial pressure than those for producing C, hydrocarbons (Jenkins, 2012; Krylov, 2005).
Chemical looping uses the lattice oxygen from an oxygen earner that is obtained and transported from the gaseous O, from air (Fan et ah, 2015; Fan, 2010). This avoids direct contact between gaseous O, and the methane, thereby eliminating the need for an ASU, reducing the risk of highly flammable methane-oxygen mixtures, and allowing for inherent separation of CO,. The chemical looping concept with various oxygen earner particles and reactor designs has been extensively studied for combustion applications, as well as reforming applications, as described in section 2 (Fan et ah, 2015; Fan, 2010).
In 1982, to minimize non-catalyzed gas-phase methane oxidation reactions in studying catalytic oxidative coupling, Keller and Bhasin fed methane and air cyclically over catalytic metal oxides, with a short tune of purging with inert gas purge in between (Keller and Bhasin, 1982). Essentially, the catalytic metal oxides used in this OCM mode act as oxygen carriers that separate oxygen from air for methane oxidation by a way of redox reactions. The overall reaction scheme is similar to the catalytic process, except that lattice oxygen from an oxygen earner is used instead of gaseous oxygen from the air. Significant efforts to commercialize this technology were made by companies like Union Carbide and ARCO, with various reducible, catalytic metal oxides tested for producing higher hydrocarbons up to C7, such as toluene (Jones et ah, 1987; Jones et ah, 1984). However, due to the oil price crash in the 1980s, these technologies were abandoned before being scaled up to commercial olefin production or commercial liquid fuel synthesis (Keller and Bhasin. 1982).
4.1.1 CL OCM oxy’gen carriers
As with other chemical looping processes, the key for success and the efficiency of the CLOCM process is the oxygen carrier material (or so called “reducible catalytic metal oxides”), as they determine the selectivity to the desired products and reactant conversions. For the selective oxidation of methane, many possible reactions are competing with the desired methane to C, partial oxidation reaction. In addition to this, the desirable products generated from methane to C, partial oxidation reaction can take part in further oxidation reaction and be converted to uudesired CO or CO,.
Major efforts har e been made to identify and screen the oxygen earner candidates from a series of metal oxides (Jones et al., 1987). The screening results in these early tests are given in Table 11. Manganese-based catalytic metal oxides demonstrate high methane conversion, which has provided the basis for further material optimization. When pure Mn,03 was tested, C, selectivity did not go beyond 20%. When support materials, such as silica or alumina, were used, the selectivity could be improved by three times. It was observed that braunite, Mn?Si012, forms in the silica-supported manganese oxides, and this may be one of the reasons for the improved reaction performance. Furthermore, the addition of alkali components, such as sodium pyrophosphate, could further improve selectivity by about 10%. It was hypothesized that sodium promotes the formation of braunite. thus producing more selective and more active catalytic metal oxides. The metal oxide catalyst Mn/Na,WOy'SiO, has been observed, with methane conversions from 12-31% and C,_ selectivities from 60-80% (Wang et al., 1995; Salehoun et al., 2008; Li, 2001; Talebizadeh et al., 2009). Also, there is an indication that the support material should be low in acidity in order to obtain high selectivity. Silica is often used as the support material for this reason (Jones et al., 1987).
However, at a reaction temperature of 800 °C, even the best oxygen carrier with the most promising results, manganese oxides with support, is limited to a 20% of C, yield in a fluidized bed reactor, commonly
Table 11. Sample of CLOCM results with various catalytic metal oxides at 800 °C and 860 hr1 GHSV (Jones et al., 1987).
* Combined C,H4 and C',H5
used in the industry (Mleczko et al.. 1996; Jaso, 2011). This yield is short of the minimum required yield to make commercial OCM economically feasible (Zavyalova et al., 2011; Keller and Bliasin, 1982; Su et al., 2003). Thus, the redox mode may require innovative reactor design in order to maintain high yields (Mleczko et al., 1996; Jaso, 2011; Zavyalova et al.. 2011; Keller and Bliasin, 1982; Su et al., 2003). Previous research has compared the co-feed and chemical looping modes of manganese oxides, and demonstrated that, for the same conversion, higher C, selectivity occurs in the chemical looping redox mode than in the co-feed mode, as is shown in Figure 27 (Sofranko et al., 1988).
Under a CLOCM scheme, research has been focused on the selection of a suitable reducible metal oxide (Jones et al., 1987; Sofranko et al., 1987; Sofranko et al.. 1988). In addition, under a redox operating mode, because of the low oxygen canying capacity of the reducible catalytic metal oxides, the lattice oxygen is quickly depleted dining methane oxidation, rendering it difficult to improve yields. In general, methane conversion decreases and C, selectivity increases over time, and available lattice oxygen is depleted within 30 minutes. The CLOCM differs from the methane and oxygen co-feed approach since oxygen diffusion within the catalytic metal oxide is critical to its performance. Oxygen diffusion in the catalytic metal oxide could take place via various mechanisms, such as diffusion along the interstices, exchange of vacancies and ions, or simultaneous cyclic replacement of atoms (Kofstad, 1972). If the oxygen diffusion rate is too slow, the oxygen in the bulk phase will har e a minimal effect on the reaction. Various chemical composition and defective structures have been tested for their capacity to control the
Figure 27. Comparison of CLOCM (redox) to co-feed for Mn-based metal oxides (Sofranko et al., 1988).
change in oxygen mobility, and hence, the reaction rate of the OCM reaction (Greisli et al.. 2010; Mestl et al., 2001; Sung et al., 2010).
Chung et al. investigated the performance of Manganese-based reducible metal oxides in the co-feed and the chemical looping schemes. The CLOCM experiments were conducted in a fixed bed at 840 °C with alternating methane and oxygen flows at a gas hourly space velocity (GHSY) of 2400 In"1 of methane. The formation of higher hydrocarbons in the experiment was confirmed. A C,_ yield of 23.2% with C7+ selectivity of 63.24% at a methane conversion of 36.7% was reached (Chung et al., 2015, 2016). Hydrocarbons up to C7 were observed but the C5 and C7 compounds could not be analytically identified.
Il'chenko et al. investigated a series of modified perovskite catalysts based on SrCo03, where metals of Li, Na and К were added as solid oxidants. It was found that K0125Na01,5Sr075CoO3_!l had the highest activity and selectivity in addition to the catalytic stability on OCM without the presence of gaseous oxygen. The oxygen required for the reactions originated from the adopted oxygen on the surface of the catalyst and the lattice oxygen of the particles. The decrease of surface oxygen with time is partly compensated as a result of its diffusion from the volume of the catalyst (Il’chenko et al., 2000).
Novel OCM catalysts, in terms of far oring kinetics of C, production reactions and catalyst composition optimization, har e been developed and studied. Hedrzak and Michorczyk developed a Mn-Na,W04 catalyst using acrylic templates coupled with different methods of loading the active phases (Michorczyk and Hedrzak, 2017). Hou et al. investigated La-based oxides and proposed a La,0,C03 catalyst by way of hydro thermal and precipitation method with different morphologies (Hou et al., 2015). Elkins et al. investigated rare-earth oxides and discovered that using Li-TbO . MgO as a catalyst can achieve a high C, selectivity with a reaction temperature above 600 °C. however, the activity of the catalyst was not sustained (Elkins et al., 2016). Chung et al. identified a Mg-Mn composite oxygen carrier with a composition of Mg6MnOs for the CLOCM process. The oxygen carrier showed both stable reactivity and high oxygen carrying capacity duiing the test of multiple redox cycles (Chung et al., 2016).
Cheng et al. investigated the enhanced C, selectivity of Mg6MnOs using a Li dopant for the CLOCM process (Clieug et al.. 2018). Li was selected as the dopant because of its high catalytic function in OCM and its similar ionic radius to Mg (Ito and Lunsford, 1985). However, a high Li-dopaut concentration may modify the crystal phases of the oxygen earner which, in turn, leads to a decrease of the oxygen carrying capacity of the oxygen earner (Qin et al., 2017). Thus, the Li dopant concentration was controlled at a low value of around 1%. The study concluded that the Li-doped oxygen earner in CLOCM universally has a higher C, selectivity than the undoped Mg6MnOs oxygen carrier, with a maximum selectivity improvement of about 50%. Density functional theory calculations and redox experiments were conducted in order to reveal the reaction enhancement mechanism of the oxygen carrier. The way in which the doping-induced oxygen vacancy affects the selectivity of the oxygen earner was revealed, which, in turn, provides a dopant-screening strategy for identifying a high-performance catalytic oxygen earner for CLOCM.
4.1.2 CL OCM reaction kinetics
Oxygen diffusion in the catalytic oxygen carriers is an additional factor that increases the complexity of the CLOCM reaction kinetics, since diffusion of the lattice oxygen from the bulk to the surface must be considered. Reshetnikov et al. deni ed a relationship between oxygen diffusion and the OCM reaction for a 125Na0 n5Sr0 75CoO,_x perovskite (Reshetnikov et al., 2011). The oxygen carrier was identified as having the best perfonnance among a series of SrCo03-based perovskite catalysts by Il'chenko et al. (Il'chenko et al.. 2000). The model theoretically explained the effect of the mobility of the oxygen in a perovskite catalyst on the dynamics of the catalytic reaction of the OCM.
Duiing the OCM reaction process, the catalytic metal oxide perovskite was assumed to be comprised of catalytic metal oxide centers with adsorbed oxygen (ZO), catalytically active metal oxide centers (Z), and reduced inactive catalytic metal oxide centers (ZR). The rates of the reactions of the adsorbed oxygen on the catalytic metal oxide surface and methane, active catalytic metal oxide center and methane, and re-oxidation of the reduced catalytic metal oxide, are given as, rzo = kZ0C’6Z0, rz = kzC0z, and rZR = kZR6ZRas, respectively. In the expressions, r is the rate of reaction, к is the rate constant of the rate equation. C is the mole fraction of methane, 0 is the fraction of the ZO, Z and ZR. respectively, with 0ZO + 0Z + 0ZR = 1, and o__ is the oxidizability of the catalytic metal oxide surface.
To simplify the analysis, it was assumed that the reactor is a continuous stirred tank reactor (CSTR). At the surface of the catalytic metal oxide, the changing rates of methane concentr ation and fractions of the three metal oxide centers as a function of tune can be expressed as,
The oxidizability at the surface of the catalytic metal oxide is a function of oxygen diffusion from the ciystal lattice to the particle surface and the surface reduction rate due to reaction with methane, and is calculated by.
The left-hand side of the equation represents the changing rate of oxidizability on the catalytic metal oxide surface, and the right-hand side represents the oxygen diffusion rate within the lattice. The boundary conditions of the equation are
where cp2 = L2k5/D is the analog Thiele parameter, D is the effective coefficient of volumetric oxygen diffusion in the catalytic metal oxide, L is the characteristic dimension of the oxide crystallite, т = kZRt is the dimensionless time, and c = 1/L is a dimensionless coordinate in the crystallite.
The model was solved by numerical integration using the Ruuge-Kutta method with solution of the diffusion equation by the marching technique at each step of integration. The calculated coefficients obtained from the experiments with K01,5Na0 !};Sr0 ^CoO,^ perovskite as catalytic metal oxide were kZQ = 1 s'1, kz = l.le"2 s'1, and kZR = 1.2e"3 s'1 (Reshetuikov et al., 2011).
It was found that, in this model, the Thiele parameter was the most important factor to evaluate the effect of oxygen diffusivity in title catalytic metal oxide on the dynamics of the OCM reactions. When the Thiele parameter is small, the diffusion rate of lattice oxygen is high, lienee, the rate of surface oxygen regeneration is fast. With a small value of cp, the lattice oxygen concentration will be uniform within the crystallite. However, when the value of cp is large, a lattice oxygen gradient will occur in the crystallite since the oxygen diffusivity is insufficient to re-oxidize all the reduced catalytic metal oxide surface in time. Hence, the methane conversion decreases with time due to a decrease in oxidizability of the surface. Reshetuikov et al. found that cp has to be less than 7, a value coinciding with the one obtained fr om the dynamics of the reactions occurring in the presence of gaseous oxygen (Ostrovskii and Reshetnikov, 2005) in order to let the surface reaction not be limited by oxygen diffusion in the crystallite. Also, as the characteristic size L for mixed metal oxide crystallites falls in the range 2-30 nm, the possible value for D was found to be in the range of 10"ls to 10~16 enr/s, a range for typical value of the effective coefficient of oxygen diffusion in metal oxides (Kofstad, 1972).
4.1.3 CL OCM reactor design
A typical continuous CLOCM system consists of two reactors, where one reactor performs the OCM reaction (reducer) and the other regenerates the catalytic oxygen carrier (combustor), with the catalytic oxygen earner particles circulating between them. The reactors can be operated in either fluidized bed or moving bed modes. Another possible CLOCM can be made using a single reactor with a periodically operated mode. Instead of circulating between different reactors, the catalytic oxygen earner particles stay in the same reactor, while undergoing subsequent reduction and oxidation reactions at different times by means of either feeding methane or ah' to the reactor.
ARCO dual circulating fluidized beds
Circulating fluidized beds possess excellent heat and mass transfer properties, which is crucial to the success of the OCM redox approach. The hydrodynamics of circulating fluidized beds, which allow the gas and solid flows in the circulating systems to be well controlled, are also well known.
In 1985, ARCO patented a double circulating fluidized bed configuration for its CLOCM process (Jones et ah, 1985). It was claimed that improved results were obtained by employing a process wherein solids are continuously recirculated between two physically separated zones, a methane contact zone (reducer reactor) and an oxygen contact zone (combustor reactor). By maintaining fluidized beds of solids in the two reactors, the average solids residence time in each reactor was controlled. The mixing of the oxygen earner particles and methane was also ensured. One possible layout of ARCO’s process is exemplified in Figure 28 (Jones et ah, 1987). Using this system, ARCO tested itscatalytic metal oxide for 30,000 redox cycles over a period of six months. During the operation, the catalytic metal oxide maintained its reactivity, selectivity and fluidization properties (Sofranko and Jubin, 1989).
Fixed bed reactor design
The Lab-scaled reactors for CLOCM processes normally use the cyclic gas feeding to a single fixed bed reactor, as shown in Figure 29 (Fleischer et ah. 2016; Parishan et ah, 2018). It is based on the idea of dynamic experiments, where the feed is switched between different reactants. In the first step, oxygen is sent to the reactor to serve as the catalyst oxidation process. In the second step, methane is sent to the reactor for OCM reaction and reduction of the catalyst. Between the two steps, purging the reactor with inert gas is conducted so as to prevent oxygen and methane from mixing. A proper dosing strategy enables a continuous operation of the process.
Tecliuische Universitat Berlin realized such a system with two independently operated six-port pulse valves on the fixed bed reactor and tested N a, WO,/Mn/ S iO, catalyst material. 2 grams of catalyst were loaded to the reactor. Repetitive continuous simulated CLOCM was carried out for a total tune of 300 minutes and a total of 100 redox cycles, 50 cycles at 775 °C followed by another 50 cycles at 800 °C, with a CH4 flow rate of 25 ml/min.
Moving bed reactor design
A moving bed reactor has been applied for the experimental study on physical separation of hydrocarbon products and methane for co-feed OCM operation. Kruglow et al. demonstrated the operation of a countercurrent moving-bed chromatographic reactor (CMBCR) (Kruglov et al., 1996). In the operation
Figure 28. ARCO chemical looping dual circulating fluidized bed OCM process (Jones et al., 1987).
Figure 29. Schematic of the CLOCM experimental setup at Teclimsche Universitat Berlin.
of the unit, C, products were separated from unconverted reactants at low conversions and recycled back to the reactor. A 0.55 C, yield was reached in the experiment.
The moving bed reactor design can also be applied in a CLOCM process. It has been suggested that a moving bed reactor design would be beneficial for CLOCM applications (Iglesia, 2002; Sofranko et al.. 1987). However, a countercurrent moving bed reducer as used in CMBCR will put the desired products at the gas outlet in contact with the highest oxidation state of the catalytic metal oxide, resulting in undesired oxidation reactions to CO ., thus leading to a lower selectivity of chemicals. Therefore, for a CLOCM process, a cocurrent moving bed is more feasible. Such a cocurrent moving bed CLOCM reactor system, whose design shall be very similar to that of a moving bed CLR, as discussed in section 2, would har e many advantages over fluidized bed and fixed CLOCM reactors.
4.1.4 Process engineering of CLOCM
As of now, research into OCM processes is being focused on catalyst development, with little focus on reactor design. The development of the overall process, which considers all the steps, from feedstock to purified end products, is rarely investigated. However, the overall process development is a step that cannot be overlooked in the commercialization of the OCM process because heat integration, product separation and process design also play significant role on the final design of the system and the economic feasibilities of the process.
From a process standpoint, the CLOCM approach possesses several inherent advantages over the methane-oxygen co-feed approach. With the use of catalytic metal oxide instead of gaseous oxygen, the undesired gas-phase oxidative reactions can be effectively inhibited, the reactor can be operated beyond the flammability limit of methane, and the heat generated from the OCM reactions can be effectively managed. As there is a large amount of heat released from the OCM reactions, which, if not removed properly, will, in turn, increase the reaction temperature of the reactor and negatively impact the C, yields as the combustion reaction of methane and intermediate products to COx are thermodynamically favorable under high temperature. The heat management is very important for the continuous and reliable operation of the OCM reactor. Contrary to co-feed OCM processes, which use a single fixed bed reactor with the inherent defect of heat removal, the CLOCM process circulates catalytic metal oxide solids between different reactors, which provides multiple heat integration and heat removal options. When the two reactors are operated under fluidized bed or moving bed mode, an in-bed heat exchanger can be easily placed inside the reactor for heat removal.
Chung et al. conducted a simulation study of the CLOCM process with Aspen Plus®. The process flow diagram of the product generation section of the process is shown in Figure 30. Two reactors, two gas-solid separators, heat exchangers, a water pump, and an ah' compressor are used in the process. The reduced oxygen earner was pneumatically transported between the reducer and combustor while undergoing redox reactions. Gas-solids separators were placed after each reactor in order to continually cycle the solids between the two reactors. Heat exchangers are used to minimize the use of utilities, both hot and cold, through heat exchange with existing process streams and utilization of the heat of reaction. In the simulation, a CLOCM process that consumes 1.000 kmol/hr methane in the reducer was considered. Figure 30 shows the general process flow diagram and pror ides the temperatures of equipment and streams. While downstream processing, either separations or upgrading, which requires a high pressure OCM product, the gaseous stream from the reducer is always cooled to 40 °C. The depleted air is also cooled to 40 °C in order to maintain the temperature differential necessary for heat exchange. Based on a heat balance calculation, the OCM product generation section can generate excess heat for electricity generation or heating requirement from other sections of the OCM process.
From Figure 30, the simplest heat integration option would be to use heat of the O.-depleted ah' from the combustor outlet to heat the natural gas feedstock before it is sent to the reducer and use the heat of the OCM products from the reducer gas outlet to heat compressed ah for the combustor operation. In addition, the heat exchanger can be placed in the two reactors to remove the excess heat from the reactions, and the steam generated from the heat exchanger can produce 9.7 MWe of electricity. The electricity can be used to operate auxiliary equipment with excess electricity. Otherheatintegrationoptionsforthis CLOCM process can also be developed. The reactants can be partially heated either directly or indirectly using the exothermic heat of reaction. Combinations of reactant-product and reactant-heat of reaction can also be performed.
Figure 31 shows the general process flow diagram for ethylene separation from OCM. The numerous stages, temperature differences between inlet gas and purified products, and overall complexity of the separations process causes the separation section to account for approximately half of the capital cost
Figure 30. General process layout for CLOCM.
Figure 31. General set-up for ethylene purification.
of the entire OCM process. The product stream from the OCM reactor is cooled, compressed, rid of acid gases, dried, further compressed, and finally separated into individual components via ciyogenic distillation.
The complexity of the traditional ethylene separation process, combined with the unique composition exiting the OCM reactor, provides an opportunity to develop alternative separation schemes. The use of silver complexes to selectively bind ethylene is one such promising idea. Complexation is used at industrial scale, but so far, no large-scale process for ethylene separation is in operation. Membranes have been researched for separation of a species from a gas mixture and can be applied for ethylene separation, either from typical pyrolysis gas or from OCM product gas. Issues with cost, durability, and separation efficiency have limited the industrial practice of membranes for such separation tasks. One final idea is to directly consume or convert the hydrocarbons in the product stream as they are a rich blend of various hydrocarbons.
From a process design standpoint, both co-feed and CLOCM provide numerous advantages over traditional steam cracking for ethylene production. First, methane is directly upgraded to a high value product, ethylene, without intermediates. Second, the OCM reactions are highly exothermic, whereas steam cracking is endothermic, reducing energy requirements. CLOCM is advantageous over co-feed OCM in this aspect since heat integr ation options are more flexible and easily applied. Finally, separation equipment is reduced using OCM since the separations require only a demethanizer, deethanizer, and C, splitter. Additional pathways for direct product upgrading are also possible without the need for separations.