A tailored multi-functional catalyst for ultra-efficient styrene production under a cyclic redox scheme

Styrene is an important commodity chemical that is highly energy and CO2 intensive to produce. We report a redox oxidative dehydrogenation (redox-ODH) strategy to efficiently produce styrene. Facilitated by a multifunctional (Ca/Mn)1−xO@KFeO2 core-shell redox catalyst which acts as (i) a heterogeneous catalyst, (ii) an oxygen separation agent, and (iii) a selective hydrogen combustion material, redox-ODH auto-thermally converts ethylbenzene to styrene with up to 97% single-pass conversion and >94% selectivity. This represents a 72% yield increase compared to commercial dehydrogenation on a relative basis, leading to 82% energy savings and 79% CO2 emission reduction. The redox catalyst is composed of a catalytically active KFeO2 shell and a (Ca/Mn)1−xO core for reversible lattice oxygen storage and donation. The lattice oxygen donation from (Ca/Mn)1−xO sacrificially stabilizes Fe3+ in the shell to maintain high catalytic activity and coke resistance. From a practical standpoint, the redox catalyst exhibits excellent long-term performance under industrially compatible conditions.

Role of lattice oxygen in the redox catalyst was probed by comparing oxidative dehydrogenation behaviors over the (Ca/Mn) 1-x O@KFeO 2 redox catalysts with 0 wt.% and 90 wt.% available oxygen storage capacity (OSC), as shown in Supplementary Fig. 3. The catalyst with 0 wt.% av. OSC was obtained by using hydrogen to reduce the sample until no water formed at 600 °C. As shown in Supplementary Fig. 3a, the hydrogen conversion is very low (~5%) due to the lack of active lattice oxygen. As a result, the ethylbenzene conversion is limited to 50%. As a comparison, a nearly 100% conversion of hydrogen to water was achieved in redox catalyst with 90 wt.% av. OSC and the corresponding ethylbenzene conversion is as high as 93.2% ( Supplementary Fig. 3b). The overall styrene yield is twice comparing (Ca/Mn) 1-x O@KFeO 2 redox catalyst with no active lattice oxygen remaining (0 wt.% OSC). The results reveals that active lattice oxygen species is essential for Redox-ODH.
(Ca/Mn) 1-x O@KFeO 2 redox catalyst as shown in Supplementary Fig. 4a to S4c respectively. It was observed that there are some low quality peaks between 17 and 30 o . These are mostly sub-peaks of the Ca 2 Fe 2 O 5 phase and the low XRD signal is due to the nature of our measurement, which were set at a relatively fast scanning rate ( For the fitting of Fe 2p 3/2 , we adopted the method by Grosvenor et al, 2 and the results are shown in Supplementary Fig. 6 (a-1) to (a-4). 4 multiplet peaks were used to fit for Fe 3+ , and 3 multiplet peaks were used to fit for Fe 2+ . It was determined that, oxidized and partially reduced redox oxide contains average Fe 3+ on the surface, where deeply reduced redox catalyst contains average Fe 2+ on the surface. For the fitting of Mn 2p 3/2 , we adopted the methods by Ilton et al. and Biesinger et al, 3,4 and used 4 multiplet peaks to fit for Mn 4+ . As shown in Supplementary Fig. 6 (b-1) to (b-4), oxidized redox catalyst was mainly composed of Mn 4+ , whereas reduced and deeply-reduced redox catalysts composed much larger Mn 3+  Based on literatures, the multiplet peaks of O 1s peak can be assigned to lattice oxygen peak, carbonate oxygen peak and hydroxyl oxygen peak. 5 It was determined in Supplementary Fig. 8 that oxidized redox catalyst was comprised with a large fraction of lattice oxygen peak (64%) and a small fraction of carbonate oxygen peak (22%), whereas partially reduced redox catalyst was composed of a much smaller lattice oxygen peak (17%) and a much larger carbonate oxygen peak (80%). This result is consistent with the in-situ FTIR results in Fig. 5 And then the redox catalyst was cooled down to 100 ºC. It was observed that ethylbenzene is activated even at as low as 100 ºC, as indicated by the exchanged C-H peak bands observed between 2600 and 3000 cm -1 ( Supplementary Fig. 14). It is noted that only exchanged bands assigned for ethyl branch C-H were observed, whereas C-H bands assigned for the benzene ring are absent. This indicates that only C-H at the ethyl branch are activated at lower temperatures. On the contrary, another TPSR-FTIR experiment was done by absorbing ethylbenzene onto D-hydroxylated redox catalyst. The D-hydroxylated redox catalyst was prepared by treating the redox catalyst with D 2 /O 2 redox cycles under 600 ºC for two cycles and then cooled to 100 ºC. Minimal peaks assigned to C-H could be detected. This is likely to be due to the limited dissociation rate of C-D due to kinetic isotope effects as described in the main text (Fig. 6b).  To prove the importance of water cofeed and the role of (Ca/Mn) 1-x O, ethylbenzene Redox-ODH was conducted on standalone KFeO 2 without water cofeed. As shown in Supplementary Fig. 15 for the on-line product distributions, styrene formation quickly decayed, indicating that ethylbenzene conversion decreased quickly. The overall ethylbenzene conversion (44.3%) was much lower than that on (Ca/Mn) 1-x O@KFeO 2 .
Additionally, there is significant amount of H 2 signal detected by the QMS, indicating that the H 2 by-product was not effectively oxidized to H 2 O. These results further confirmed that (Ca/Mn) 1-x O is important in oxidizing the H 2 by-product to H 2 O and to increase ethylbenzene conversion.

Aspen Plus simulation:
Three cases have been analyzed to explore the impact of increased EB-to-styrene yield on the overall process energy savings. · Case 1: Industrial catalytic dehydrogenation of EB to styrene · Case 2: Redox-ODH of EB to styrene (10% EB in steam) · Case 3: Redox-ODH of EB to styrene (product yields as in Case 2; no steam addition) All the cases have been simulated using AspenPlus™ software. The product distribution for the three cases are given below: Supplementary

Case 1: Industrial catalytic dehydrogenation of EB to styrene
Catalytic dehydrogenation of ethylbenzene is the major industrial route for styrene production 7 . The endothermic reaction is carried out in the vapor phase with steam, over a catalyst consisting primarily of iron oxide. Based on the existing literature [6][7][8][9] , a series of two adiabatic reactors is assumed for EB conversion, as shown in Supplementary Fig. 16. The feed to each reactor is in 650-675 o C. Since the adiabatic reaction drops the temperature after the first reactor, the stream is heated back up to 650 o C before entering the second reactor. EB concentration to first reactor is maintained at 5% at the inlet with the remaining primarily being steam. The feed also contains the recycled unreacted EB stream. Dilution steam prevents formation of coke from EB. A single-pass EB conversion of 40% is assumed in each reactor, making the combined single-pass conversion to be 64%. An operating pressure of 2 bar is assumed 6,8,9 . The reactor effluent is fed through a heat recovery system to minimize energy consumption, condensed, and separated into vent gas, a crude styrene hydrocarbon stream, and a steam condensate stream. The crude styrene goes to a distillation system to separate into 99.9% pure styrene and unreacted EB, which is recycled. Ethylbenzene and styrene, having similar boiling points, require 70 -100 trays for their separation. The 82-stage column operates under vacuum with a reflux-drum pressure of 10 kPa to give pure styrene. The second column for recycle EB has 38 stages and operates slightly above atmospheric pressure. Separation scheme also produces an aromatic stream rich in benzene and toluene, along with vent streams containing mainly CO 2 and H 2 .   Supplementary Fig. 17 shows the reaction and separation scheme for Redox-ODH of EB to styrene. MnO 2 ↔MnO transition is used in the model, to mimic the change in Mn-oxidation state in the original redox material. Product distribution is listed in Supplementary Table 1 above. To accurately reproduce the single pass yield and heat of reactions based on experimental results, the reducer reactor was modeled in ASPEN with two RStoic reactors. The first RStoic step converts EB whereas the second step combusts hydrogen with MnO 2 (redox catalyst). 100% H 2 -to-H 2 O combustion is assumed, based on experimental results. EB: steam ratio entering the first reactor is 1:9. The two reactors in series (reducer) together are adiabatic, first one being endothermic and the second exothermic. The reduced MnO is oxidized in a following reactor (oxidizer) using 20% excess air. The regenerated MnO 2 is recycled back to the reducer reactors. Hot, O 2 -deficient air is utilized in generating the feed steam. The downstream separation scheme is kept similar to Case 1, to perform a consistent basis for comparison.

Case 3: Redox-ODH of EB to styrene (product yields as in Case 2; no steam addition)
Case 3 represents a case where no steam addition is required ( Supplementary Fig. 18). Here, EB conversion and product yields are as used in Case 2 (Supplementary Table 1). The rest of the system is as described in  Variables · Amount of steam mixed with feed EB and recycle, before the stream has exchanged heat with the reactor outlet stream (X s ) · Temperature of the stream (fresh EB + Xs amount of steam + recycle) before mixing with the remaining steam · Temperature of the oxidizer (temperature of the reduced solids entering the reducer) [(600 ± 50) o C] Both the reducer reactors in Case 2 and Case 3 are operated at 600 o C. The above variables are used to ensure the total Q of both the reactors equals zero (±0.5%).
The optimized conditions result in all of the steam being mixed with EB (fresh and recycled) before heat exchange, with the output of the stream at 590 o C. The oxidizer temperature is set at 630 o C, which allows for adiabatic operation of the two reactors, with an outlet temperature of 600 o C.

Aspen Plus Simulation Results:
Supplementary Table 5 lists the heat duty required for the different sections of the process, for all three cases. The values are normalized to per tonne of styrene produced. These results are obtained based on the assumptions in Supplementary Table 4 and models developed in Aspen Plus. Case 1 yields an overall heat duty demand of 16.7 GJ/tonne styrene, which is in close agreement with the value reported in literature 10 . Reports in literature show various optimized process plant configurations for Case 1, but for this study, the most widely used base case is adopted as Case 1. Redox ODH Case 2 depicts the experimental results on the redox catalyst which is the focus of this study. This provides an overall styrene yield of 91.4%, which is 38.4% higher than the reported value for Case 1 (53%).
Case 2 leads to a decrease in the heat duty requirement by roughly 50% compared to Case 1. Major savings arise from the substantial drop in steam requirement for Redox ODH schemes, which can operate with low coke and high styrene selectivity at 0.1 atm EB partial pressure. In our experimental studies, we used 0.1 atm EB balance inert Ar. Case 2 conservatively assumed that a 1:9 EB to steam molar ratio is used in redox-ODH (steam is used in the place of Ar). This results in >50% reduction in steam usage compared to the commercial DH process. Case 3 covers a more optimistic case, which assumes that "dry" EB can be used without steam to achieve product yields and selectivity identical to those obtained from our experimental redox-ODH unit with 0.1 atm EB balance Ar. We note that our experimentally obtained EB conversion and styrene yield are significantly higher than those in conventional DH. This increased EB yield also contributes to lower energy requirements in the EB separation column. Comparison of the styrene separation columns in Case 1 and Cases 2/3 are shown in Supplementary Table 6. Overall, the redox ODH process, as analyzed in Cases 2/3, allows for highly efficient styrene production from EB, reaching 91.4% yields, with a heat duty demand half of what the industrial EB dehydrogenation route demands. Redox ODH process can operate at high EB: steam ratios, leading to significant savings in steam requirement, which is the prime source of energy demand.
Supplementary # Heat released from product cooling is of very low grade (low pressure steam condensation) and is therefore not recoverable