《1. Introduction 》

1. Introduction

Ethylene is one of the most important organic materials; it is used as a building block to produce fibers, plastics, and other chemicals. Global ethylene production capacity was around 1.48×10t in 2014, representing a 32% increase over the past decade [1]. At present, ethylene is almost exclusively produced via the steam cracking of gaseous and liquid hydrocarbon feedstocks such as ethane, naphtha, and gas oil. Due to its high endothermicity and complex product-separation steps, steam cracking is one of the most energy-intensive processes in the chemical industry. Although ethane is the most common feedstock for cracking in the United States and Middle East, more than 80% of the ethylene produced in Europe and the Asian-Pacific region is from naphtha [2,3].

Naphtha is a mixture of hydrocarbons with a boiling point range of 30–200 ℃. In naphtha-based steam-cracking processes, naphtha is first fed into the convective section of the furnace for preheating and vaporization. At elevated temperatures in the radiant section of the furnace (750–900 ℃ or higher), naphtha is cracked into smaller molecules in the absence of catalysts. Gaseous light olefins are formed via reactions involving free radicals [4,5]. Steam cracking is highly endothermic and requires significant external heat input. Coke deposition, which is inevitable in the complex cracking reaction scheme, represents another significant challenge. Diluting the feedstock with steam lowers the propensity for coke formation. However, cracking furnaces still need to be periodically shut down and regenerated with air to avoid the accumulation of coke. The parasitic energy consumption due to steam usage and the need for periodic shutdowns result in an increase in capital and operating costs [6,7]. From an energy consumption standpoint, the pyrolysis section accounts for approximately 2/3 of the total process energy required in naphtha steam crackers, and leads to about 3/4 of the total exergy destruction [3]. Furthermore, large amounts of carbon dioxide (CO2) are generated in this process due to the need to combust carbonaceous fuels to satisfy the parasitic energy requirements. Therefore, novel technologies that address these shortcomings of steam cracking are highly desirable.

The oxidative cracking of naphtha using a heterogeneous catalyst is a promising option [8–10]. However, the high reactivity of the olefins poses a challenge in identifying a suitable catalyst with high selectivity [11]. Moreover, there is an additional difficulty of separating the oxygenate byproducts [6]. The flammability of the gaseous mixture, which contains oxygen (O2) and fuels, also presents safety concerns. Hot spots can result from over-oxidation of the hydrocarbons to carbon monoxide (CO) and CO2, lowering the product selectivity [3,12,13]. The current study investigates a two-step, redox oxy-cracking (ROC) process for naphtha conversion in an O2-free environment. The first step involves selective oxidation of the hydrogen (H2) produced from hexane cracking with active lattice oxygen in a mixed-oxide redox catalyst. The redox catalyst particles, which enter the reactor at high temperatures, provide the sensible heat to compensate for the endothermicity of the cracking reactions. Meanwhile, the catalyst is reduced during this step. The reduced redox catalyst is later oxidized, regenerated, and reheated in air to complete the redox loop. The absence of gas-phase oxygen in the naphtha conversion step has the potential to ① inhibit the non-selective reaction pathways, ② reduce the potential safety hazards [13–15], and ③ decrease parasitic energy losses by eliminating the air-separation unit and providing tighter temperature control [16]. The overall process also significantly decreases the energy consumption and, hence, the nitrogen oxide (NOx) and CO2 emissions. Despite its advantages over steam cracking and conventional oxy-cracking, this novel ROC approach has yet to be investigated in detail.

We recently reported a chemical looping-oxidative dehydrogenation approach for ethylene production from ethane using operational principles similar to those of ROC [17,18]. The redox catalyst development, performance, reaction mechanism, and overall process efficiency were investigated [17–19]. However, these studies were limited to ethane conversion. The oxidative cracking of n-hexane to olefins using lattice oxygen of VOx/Ce-Al2O3 has been investigated, showing up to 60% olefin selectivity at 30% n-hexane conversion [11]. The performance of sol-gel-synthesized Li/MgO in the oxidative cracking of hexane in the presence of gaseous oxygen, with 28% hexane conversion and 60% selectivity to light olefins, has also been shown [8]. The same authors also studied an integrated plasma-Li/MgO system to improve the oxidative conversion of hexane at the relatively low temperature of 500 ℃ [9]. Although the integration of plasma was effective to reduce the operating temperature, the light olefin yield was still limited to about 35%. Overall, existing oxy-cracking processes have demonstrated relatively low olefin yields (< 35%). Moreover, studies on the oxy-cracking of naphtha via the chemical looping scheme are limited.

In the current study, a detailed process simulation and analysis is conducted to investigate the performance of the ROC process. Experimental data obtained from a novel redox catalyst composed of alkali-doped mixed oxides were used in the simulation [20]. Single-pass hexane conversions of 79% with 61% light olefin selectivity were achieved in a laboratory-scale reactor. The experimental results were used in the process simulation, and the performance of the ROC process was compared with a state-of-the-art naphtha steam-cracking process. Using ASPEN Plus®, the effects of the reactor and process conditions were investigated (The flowsheets of ASPEN Plus® for hexane cracking and ROC process are given in Figs. S1 and S2 in the supplemental file, respectively). The potential advantages of the ROC process were validated using detailed process analyses. When compared with the steam-cracking process, the net energy demand of the upstream section of the ROC reference case (79% naphtha conversion) was shown to be 67% lower, and to require 25% less compression work. These decreases result in a 52% drop in the overall energy demand and a corresponding reduction in CO2 emissions, as compared with the conventional cracking process.

《2. Process descriptions》

2. Process descriptions

Naphtha can vary in composition depending on its source and refinery conditions. It is a complex mixture of hydrocarbons that ranges from straight-chain alkanes to aromatics. One of the major components of naphtha is n-hexane. Reaction (1) illustrates one of the representative cracking reactions of n-hexane to ethylene. In addition to its high endothermicity, the vaporization of naphtha can be energy consuming—the enthalpy of vaporization for nhexane at 298 K is 359 kJ·kg-1 . A simplified process flow diagram for the conventional naphtha steam-cracking process is shown in Fig. 1(a). As the first step, preheated naphtha is thermally cracked in the presence of steam. Since naphtha cracking is equilibrium limited and tends to form coke, steam is used as a diluent to enhance naphtha conversion and inhibit coke formation. In the cracking step, naphtha is passed through the tubes of the cracking furnace. The furnace is operated at high temperatures (~1000 ℃), so the heat for the endothermic cracking reactions is provided by the fuel combustion. The furnace consists of the radiant zone, where the cracking reaction occurs at > 750 ℃, and the convective zone, where the feedstock preheating takes place at up to 650 ℃. The product stream is at a temperature of approximately 800 ℃. Using transfer line heat exchangers, the stream is rapidly quenched to 300–400℃ . This avoids further cracking and preserves the product compositions. As the product contains significant amounts of heavy hydrocarbons, a primary fractionator is installed to remove tar and oily material (boiling point > 200 ℃) from the cracked gases. The product is further quenched to near-ambient temperature to remove the "heavies.” This is followed by a multistage compression system to raise the pressure of the lighter gaseous products to around 4 MPa. Drying and acid gas removal (AGR) units are present at the intermediate compressor stages, as they need to be operated at high pressures. An adsorption column of molecular sieves facilitates the drying, followed by a caustic wash to remove acid gases such as CO2. The formation of hydrates is prevented by having the drying unit before the caustic wash [2]


Fig. 1. (a) Simplified process scheme of naphtha steam-cracking process; (b) simplified schematic of the ROC process.

The light components from the compressed stream are sequentially separated from the heavy fractions using a separation train of distillation columns that ultimately yield high-purity ethylene and propylene. At the beginning of the sequence, methane (CH4) and lighter components (primarily H2) are selectively separated using a demethanizer (DM). A deethanizer (DE) column separates acetylene, ethane, and ethylene from the bottom product of the DM. The distillate of the DE is reacted with H2 to convert acetylene to ethylene, and then further distilled using a C2 splitter to generate the ethylene product. The C3+ components exit at the bottom of the DE, and are further purified in a depropanizer (DP) to separate the C4+ components. The debutanizer (DB) separates the C4 materials, raw pyrolysis gasoline, C5 materials, and aromatics from the bottom of the DP [2]. The C3 splitter follows the DP and recovers propylene, which is a valuable product.

Fig. 1(b) depicts the naphtha ROC process. A reactorregenerator scheme is used instead of the energy-intensive furnaces. An experimental product distribution for the ROC scheme, using an alkali-doped calcium-manganese (Ca-Mn)-based oxide similar to that reported in our previous work [20], is used in the current work. For the ASPEN Plus® simulations, an oxide mixture containing calcium oxide (CaO) and manganese oxide (MnOx) is used because the specific mixed-oxide is not available in the ASPEN Plus® database. Mn3O4 acts as the oxygen carrier phase participating in the oxidative dehydrogenation (ODH) reaction through lattice oxygen donation. Mn3O4 is reduced to MnO following Reaction (2). Naphtha is converted to a variety of products; however, for simplicity, Reaction (2) portrays n-hexane to ethylene as the model reaction, which involves the combustion of H2 to water. In a regenerator, the reduced redox catalyst (oxygen carrier) is replenished with air via Reaction (3) (which is highly exothermic). The ASPEN Plus® database values indicate that the surrogate catalyst exhibits similar heat of reaction and heat capacity values to those of the actual redox catalysts. In the current work, the reducer is operated at 775 ℃ and 1 atm (1 atm = 1.013× 105 Pa), while the regenerator is operated at 935 ℃ and 1 atm. The redox loop is completed by recirculating the re-oxidized particles to the reducer. The redox catalyst provides both the necessary lattice oxygen and the heat required for ROC reactions. The naphtha and the air are preheated to 650 ℃. The oxygen-depleted hot air from the regenerator is used to generate high-pressure steam to compensate for parasitic energy requirements. The gaseous output from the reducer passes through a series of downstream separation units similar to steam cracking. An amine scrubbing unit for CO2 removal is required for the ROC process, as this process produces more CO2 than traditional cracking. The heat requirements of the reducer and naphtha preheating are sufficiently provided by the heat stored in the re-oxidized solids. This is possible due to their high heat capacity and to the temperature difference (∆T) between the reducer and regenerator. The reducer is endothermic, whereas the regenerator is exothermic. The reducer-regenerator system is operated under adiabatic conditions. The cracking furnace constitutes a major share of the total energy requirement; since it is replaced by the ROC reactors, the energy demand is reduced. In the following sections, the ROC process is compared with naphtha cracking based on the overall energy requirements.

《3. Simulation assumptions》

3. Simulation assumptions

The performance of ROC was evaluated and compared with naphtha steam cracking using the ASPEN Plus® simulator [21]. Two kinds of naphtha feeds were used to model the cracking process: ① a complex naphtha consisting of a variety of hydrocarbons, based on literature data; and ② n-hexane. Since n-hexane is a major component of naphtha, the process model can be signifi- cantly simplified, provided that the results using the n-hexane feed provide a reasonable representation of those using the complex feed. The composition of the complex naphtha is shown in Fig. 2, which has been optimized to mimic "true” naphtha. For the simplicity of the model, pure n-hexane was used as a representative compound for naphtha in the ROC process. Table 1 summarizes the ASPEN Plus® modules, property methods, and physical property databanks, while Table 2 lists the carbon yields for the various processes. The cracking results are based on literature data [2,22], whereas the ROC results are experimental results that were published in our previous work [20]. A higher hexane conversion case was also considered, in which 97% of the hexane is assumed to be converted in the ROC process. Experiments on blank hexane cracking and the ROC case revealed an increase of roughly 30% in ethylene and propylene yield for the ROC. Keeping the COx yield similar to that in the base ROC case, the ethylene and propylene yields were proportionally increased for the high ROC case, with the overall hexane conversion being 97%. The other products were assumed to have yields similar to that of the base ROC. The product distributions of this hypothetical high ROC case are given in Table S1 in the supplemental file. The supplemental file also contains general information with respect to process configuration in ASPEN Plus®, materials flow, etc.


Fig. 2. Composition chosen to represent complex naphtha. (a) Based on Ref. [2]; (b) based on Ref. [22].

《Table 1 》

Table 1  Modules, property methods, and physical property databanks used in ASPEN Plus®.

《Table 2》

Table 2 Carbon yields for the cases investigated

a Equal conversions of all the feed components assumed are listed in Table 1.

b Conversion of n-hexane.

The drying unit and the primary fractionator were modeled as flash columns, in which almost all of the water is condensed along with the heavy hydrocarbons. To simulate the AGR unit, a separator (Sep) was used, with its energy consumption accounted for. This was placed at the exit of the compressor. The solids used in the simulation include Mn3O4 and MnO, with CaO added as an inert substance in order to closely represent the actual redox catalyst [20]. Table 3 lists the key operating conditions and simulation assumptions.

《Table 3 》

Table 3 Key operating conditions and simulation assumptions

In both the complex and n-hexane feed cases, the hydrocarbon feedstock is introduced to the cracking furnace after being preheated to 650 ℃ along with steam. A steam dilution of 0.5 kg per kilogram fresh feed is used [2]. The same preheating temperature is used in the ROC case, without steam dilution. In the process simulation, the Mn3O4-containing redox catalyst particles are initially fed into the reducer. They are reduced and flow to the regenerator, which is maintained at 935 ℃, with the temperature of the reducer at 160 ℃ lower. In the regenerator, 10% excess air preheated to 650 ℃ is used to completely oxidize the reduced oxygen carrier (Reaction (3)) prior to recirculation to the reducer for another redox loop. Based on the product distribution in Table 2, the ROC involves the combustion of 75% of the H2 to H2O.

To prevent further cracking, the product gases in all the cases are quenched to a temperature of 100 ℃. Heat released from the quenching is assumed to be utilized to generate steam and preheat the feed streams. The exothermicity of the reactor-regenerator system can also produce low-pressure steam, which can reduce the upstream energy load. The cooled gases are compressed to 2.5 MPa and flashed at 100 ℃, in order to remove the heavy hydrocarbons and water from the stream. In the present model, this compressor, with the flash column, functions as the primary fractionator. The light stream is further compressed to a pressure of 42 atm [2] using a four-stage isentropic compressor. The primary component of acid gas is assumed to be CO2, with no hydrogen sulfide (H2S) present. Negligible energy consumption is assumed for the caustic wash. For the ROC, which has substantially more CO2 in the product stream, an amine scrubbing process is assumed, with an energy consumption of 0.11 MWh per tonne CO2 [23]. Table 4 lists the operating conditions of the downstream columns for the cracking process [24]. The compressed stream is cooled to -100 ℃ prior to methane removal in the DM. This refrigeration step reduces the load on the DM by removing most of the H2 and CO. A single pass is assumed in all the cases. The reflux ratio for each column is fixed as twice that of the minimum reflux ratio. For a fair comparison, the same number of stages and recovery were assumed for the C2 splitters in both cases. A deacetylenizer is used after the DE, in order to hydrogenate the acetylene in the DE distillate (see Table S1 in the supplemental file). Purified H2 for this reaction is obtained through pressure swing adsorption (PSA), which operates with 80% H2 recovery, an exit purity of 99% H2, and a pressure drop of 0.1 MPa [25].

For process comparison, the different forms of energy (thermal, steam, and electricity) are converted into the same thermal basis. The higher heating value efficiencies used for different energy conversions are given in Table 3. The refrigeration power is calculated using the power-temperature graph provided in Ref. [26], and using the condenser and reboiler duties of the columns, as calculated by ASPEN Plus®. Table 4 [24] lists the refrigerants chosen for each unit.

《Table 4》

Table 4 Details of the distillation columns [24]

《4. Results and discussion 》

4. Results and discussion

《4.1. Effect of feed compositions 》

4.1. Effect of feed compositions

Using the product distribution shown in Table 2, both the complex naphtha and the pure n-hexane feed were simulated for the conventional steam-cracking process. Fig. 3 shows the similarities in the properties of the two feeds. Typically, a naphtha cracking unit has an overall energy demand in the range of 20–40 GJ per tonne ethylene [3]. Table 5 illustrates the overall and unit-wise energy distribution of the two cases with the complex and nhexane feeds. As can be seen, the results from the pure n-hexane feed are very similar to the results from the complex naphtha feed. This indicates that n-hexane can be used as a suitable model compound for naphtha, without compromising the accuracy of the simulation results. Fig. 3 breaks down the energy consumption of each section relative to the overall energy demand for both the pure n-hexane and the complex feeds. Again, close agreement was obtained.


Fig. 3. Section-wise energy distribution for naphtha cracking.

《Table 5 》

Table 5 Overall and unit-wise energy distribution for complex naphtha and n-hexane feeds (unit: GJ per tonne ethylene).

《4.2. Energy distributions in steam cracking 》

4.2. Energy distributions in steam cracking

Considering the similarities of the results from the complex and n-hexane feeds, n-hexane was used to represent naphtha in the rest of the studies. The ASPEN Plus® simulation of the naphtha cracking process indicates an energy consumption of 32.5 GJ per tonne ethylene. This finding is in agreement with the study of Ren et al. [3], which reports an energy consumption range of 20– 40 GJ per tonne ethylene. For the cracking process, 78% of the total energy is required in the upstream section of the process. Quenching of the hot gases generates high-pressure steam and recovers additional energy. The compression step corresponds to 57% of the overall downstream energy consumption, while the refrigeration for the DM feed and the C2 splitter consumes 17% and 16% of the downstream energy, respectively, with the C2 splitter being the most energy-intensive separation column.

《4.3. Process comparisons》

4.3. Process comparisons

As shown in Table 3, the conversions are comparable for the cracking and ROC base cases; however, the ROC process offers a 28% increase in the combined ethylene and propylene yield, which are the main products of the process. Fig. 4 shows the section-wise energy distributions of the naphtha cracking process and the ROC case. Table 6 lists the energy distributions for the two processes. The cracking process requires the energy-intensive furnace, which contributes about 27% of the overall energy demand. In the ROC case, the furnace is replaced by the reducer-regenerator system, which is operated based on cyclic redox reactions (Reaction (2) and Reaction (3)). The regenerator replenishes the O2 in the carrier via the highly exothermic Reaction (3). The nitrogen (N2)-rich hot air exiting the regenerator is used to generate high-pressure steam, and provides upstream energy in the ROC case. As the reactors are operated adiabatically, the upstream and overall energy requirements for the ROC case are reduced significantly. The absence of dilution steam further reduces the upstream (thermal) energy demand and decreases the upstream demand from 25.44 GJ per tonne ethylene for cracking to 8.32 GJ per tonne ethylene for the ROC case, as shown in Table 6. This upstream section shows the most significant reduction in energy demand in ROC compared with cracking.


Fig. 4. Section-wise energy distributions of the n-hexane cracking and the ROC. 

《Table 6 》

Table 6 Comparison of the energy demands (unit: GJ per tonne ethylene)

Note: The high-value products (HVPs) yield in n-hexane cracking is 2.5 kg per kilogram ethylene; the HVPs yield in ROC is 2.0 kg per kilogram ethylene.

The compression section provides the main energy savings downstream in the ROC, with a 25% reduction compared with cracking, due to a decrease in the total moles of the product gases, on a dry basis. The overall downstream energy requirement is roughly the same for the two cases. The refrigeration duty needed to cool the feed to the DM is reduced, as less mass is fed to the DM in the ROC case. Overall, 0.11 MWh of work per tonne of CO2 captured is needed for the ROC process [23]. Assuming 100% CO2 removal, this accounts for 0.12 GJ (electric) per tonne ethylene, contributing roughly 4% of the downstream energy demand. The PSA unit contributes an energy requirement of 2.2 kWh (electric) per kilogram H2 separated [25]. The ROC involves the separation of a lesser amount of H2, leading to reduced PSA energy requirements. In both cases, the deacetylenizer operates at high pressures (~2.5 MPa) and at temperatures of 50–70 ℃ [27]. The reactant gas from the top of a DE is at a very low temperature (~50 ℃). Higher ethylene to ethane and propylene to propane ratios lead to higher C2 and C3 splitter energy demands. Overall, the ROC case has the potential to provide a 52% reduction in the overall energy demand of the plant, compared with n-hexane cracking, as shown in Fig. 4. For process analysis purposes, we considered both the ethylene product and the high-value products (HVPs). Here we followed the definition of HVPs by Ren et al. [3], which considered all the olefins and C5+ hydrocarbons including aromatics as value-added products. Figs. 5 and 6 compare the two cases by normalizing the energy demand in terms of the HVPs.


Fig. 5. Comparison of the energy distributions in steam cracking and ROC processes. HVP: high-value product.


Fig. 6. Normalized energy demand (section-wise and overall) in steam cracking and ROC processes.

《4.4. Net demand and emission comparisons》

4.4. Net demand and emission comparisons

The cracking and the ROC process produce fuel gases (CO, CH4, and H2), which can be combusted to meet the energy requirements. Assuming that all the fuel gases are used, the n-hexane cracking shows a deficit of 5.2 GJ per tonne ethylene, which represents 16% of the total demand. In comparison, the ROC process requires no further fuel input, as calculated using the ∆Hreaction in Table 7 and the amount of available fuel based on ASPEN Plus®. In the steam-cracking case, a portion of the fuel oil can also be combusted to meet the energy demands. For simplicity, no fuel oil recycling was considered in this study, although fuel oil can be used as a feedstock to increase the ethylene and propylene yields.

《Table 7 》

Table 7 ∆Hreaction and molar flow rates of the byproduct fuels.

This section compares the two processes based on the net energy required to produce 1 t of ethylene. In this analysis, we used the energy values of the feedstock and products to determine the net energy consumptions for ethylene production. This permits an impartial comparison of the conventional and ROC processes since one could argue that ROC’s primary energy “savings” result from indirect combustion of H2 and hydrocarbons. As shown in Fig. 7, the entire process is divided into four relevant energy streams: ① the n-hexane feed (stream A); ② the overall thermal energy demand (stream B); ③ ethylene (the main product) (stream C); and ④ the byproduct fuel out (stream D). Lower heating values (LHVs) are used for the fuel streams and are normalized based on 1 t of ethylene produced.


Fig. 7. Schematic for net energy demand analysis.

As shown in Fig. 8 below, the ROC process requires 14% less nhexane per tonne of ethylene produced owing to higher ethylene yields, along with being less-energy intensive compared to traditional naphtha cracking. For every tonne of ethylene produced, the ROC process requires a net energy feed of 67 GJ, leading to 39% savings in overall process energy consumptions.

In the case of n-hexane cracking, a negligible amount of CO2 is present in the product gases. Here, it is assumed that methane is used to meet the energy requirements in both the cracking case and ROC. Based on the LHV of CH4 and the total energy demand of the two processes, as given in Table 7 and Fig. 5, 40.6 and 19.4 kmol CH4 are required to get 1 t ethylene for the cracking case and for ROC, respectively. This leads to an emission of 1.8 and 0.85 kg CO2 per kilogram ethylene, respectively. Ren et al. [3] report an emission of 1.8–2 kg CO2 per kilogram ethylene for naphtha cracking units. Therefore, ROC has the potential to result in a 52% reduction in CO2 emissions. The ROC process does not need hightemperature furnaces. Therefore, significantly reduced NOx emissions can also be anticipated [28]. The overall thermal energy demand for the high ROC case is 14 GJ per tonne ethylene, which provides an energy reduction of 10% and 57% compared with the base ROC and cracking cases, respectively, with proportional reduction in CO2 emissions. As shown in Fig. 8, although the net demand remains unchanged for the ROC cases, the ROC high conversion case provides a reduction of 12.8% in the n-hexane feed and 10% in the overall energy demand for every tonne of ethylene produced.


Fig. 8. Net (thermal) energy demand for ethylene production. (The net energy demand shown here follows the definition from Fig. 7 and is defined differently from the energy demand presented in Sections 4.1–4.3)

《5. Conclusion 》

5. Conclusion

The current study investigates an ROC process for light olefin production from naphtha. Based on our recently obtained experimental data using an Mn-based redox catalyst, the ROC process was simulated using ASPEN Plus® and compared with the traditional steam-cracking process. Compared with cracking, ROC replaces the energy-intensive cracking furnaces with an autothermal reducer-regenerator system. This results in significant energy savings. Moreover, the experimental results indicated significantly improved ethylene and propylene yields. For each tonne of ethylene product, the ROC process has the potential to reduce the overall thermal energy demand by 52% when compared with steam cracking. While the conventional naphtha cracking process is estimated to consume 32.5 GJ of energy in order to produce one tonne of ethylene, the ROC only requires a net thermal energy input of 15.5 GJ. The ROC can also provide a reduction of over 50% in CO2 emissions. Specifically, the upstream section of the ROC process consumes approximately 67% less energy while producing 28% more ethylene and propylene for every kilogram of naphtha feedstock. The absence of the cracking furnaces also results in a considerable reduction in NOx emissions for the ROC process. Overall, the current study indicates that the novel ROC concept has the potential to be an attractive and environmentally friendly option for ethylene and propylene production from naphtha.

《Acknowledgements 》


This work was supported by the US National Science Foundation (CBET-1604605) and the Kenan Institute for Engineering, Technology and Science at North Carolina State University.

《Compliance with ethics guidelines》

Compliance with ethics guidelines

Vasudev Pralhad Haribal, Yun Chen, Luke Neal, and Fanxing Li declare that they have no conflict of interest or financial conflicts to disclose.