Ethylene (ethene) is one of the major building blocks of the chemical industry. The annual worldwide production of ethylene was approximately 170 Mt (2021) (Najari et al., 2021). Ethylene is mostly used for the production of PE which accounted for roughly 60% of the worldwide ethylene produced in 2000 (Zimmermann and Walzl, 2010). Other key products from ethylene include ethylene oxide, ethylene dichloride, ethylene glycol, and ethylbenzene (Zimmermann and Walzl, 2010).

Table 1 reviews the technical processes for ethylene production. Fossil-based processes with small market shares and low ethylene yields are not included. Such examples include the catalytic pyrolysis process (CPP), SUPERFLEX, propane dehydrogenation, Gas stream technologies, hydro-pyrolysis of naphtha, and Mobil Olefins Interconversion (MOI) (REN et al., 2006; Hsu and Robinson, 2017). Besides established and commercialized processes, the table shows emerging technologies which can be based on renewable feedstock.

The most common production route for ethylene production is the steam cracking process (Zimmermann and Walzl, 2010). The feedstock reacts at temperatures at around 750°C–875°C in a fired furnace (Zimmermann and Walzl, 2010). After cooling, the reaction products are separated. Product yields vary depending on process conditions and feedstock properties (Zimmermann and Walzl, 2010). Typical sizes of steam cracker plants are in the range of 1–3 Mt ethylene per year (Zimmermann and Walzl, 2010; Fan et al., 2012). Typical feedstock for steam cracking includes ethane, propane, butane, naphtha, gas oil, or hydrocracker residue (Zimmermann and Walzl, 2010). Fossil feedstock can be replaced by renewable alternatives like hydrodeoxygenated tall oil (Pyl et al., 2012), vegetable oil (Zámostný et al., 2012), bio-oil (Gayubo et al., 2010; Li et al., 2018), or bio-naphtha (Pyl et al., 2011).

Except for the biotechnological process routes and cracking processes, the carbon containing feedstock of the processes included in Table 1 exclusively consist of hydrocarbon fractions at shorter chain lengths, ranging from C₁ to C₃. Besides the carbon feedstock, some processes require auxiliary reactants like water, oxygen, or chlorine. The C₁ feedstocks (syngas, CO₂, CO, methane, and methanol) can all be derived from sustainable feedstock at low CO₂ emissions. The methanol to olefin process (MTO) is a commercial process available for the production of short olefins (Gogate, 2019). Ethanol and ethane are possible C₂ feedstock, while C₃ feedstock comprise of propene and glycerol. The process for ethanol dehydration to ethylene is commercialized and multiple plants are running worldwide (Fan et al., 2012; IRENA, 2013). C₆ sugars can be converted via the biotechnological route.

Table 2 reviews publications concerning process design, modelling, environmental assessment and economic assessment of one-step or two-step electrochemical CO₂ reduction processes. Only three studies consider the two-step process (CO as intermediate product) which is widely regarded as being beneficial in terms of efficiency and economics. The studies comparing the one-step with the two-step process report a better economic, environmental and technical performance for the two-step option (Kibria Nabil et al., 2021; Ramdin et al., 2021; Sisler et al., 2021). The energy demand can, according to (Kibria Nabil et al., 2021), be decreased by 27% for the two-step compared to the one-step process. Besides the oxygen evolution reaction on the anode, other reactions can be used to improve economics, as shown by Li et al. (2021) and Na et al. (2019).

General limitations of the reviewed studies include unrealistically high FE, low cell voltages and consideration of only one product from the reduction reaction. One important aspect in process design is also the product separation and purification which can lead to high cost and energy consumption. Greenblatt et al. (2018) elucidate on that topic in the context of electrochemical CO₂/CO reduction.

For both the combustion and gasification-based route, three distinct scenarios were investigated. Gasification based scenarios are tagged with “G” and combustion-based scenarios are tagged with “C” (see Table 3). For the electrochemical CO reduction, we consider different product yields to study the impact on the process performance. For the improved ethylene yield “-O” we are assuming less by-products and higher ethylene yield in COR. Scenario “−2” consider additional conversion of ethanol to ethylene.

For a comparative assessment of the scenarios, the levelized cost of ethylene (LCOE) is calculated based on Eq. 1. The LCOE comprises of the annualized capital expenditure (CAPEX), the yearly operating cost (OPEX) and the revenue from by-product sales, which are subtracted. This value is divided by the yearly production of ethylene to give the LCOE in €/t ethylene. L C O E = A n n u a l i z e d   C A P E X + O P E X − ∑ i = 1 4 m s i d e   p r o d u c t   i C s i d e   p r o d u c t   i y e a r l y   p r o d u c t i o n   e t h y l e n e (1)

The selling prices for the by-products ethanol (800 €/t), acetic acid (800 €/t), oxygen (10 €/t) and hydrogen (2000 €/t) are used to calculate the revenue. The operating costs include biomass (20 €/MWh), electricity (60 €/MWh), maintenance, labor, refrigeration, heat supply, deionized water for the anode reaction, and CO₂ separated from the flue gas. The base year for cost calculations is 2020. The annual full load hours are set at 8,400 h. For calculation of the annuity factor, the interest rate is set at 6%, and project lifetime at 20 years (40 years for power plant). The equipment cost is either derived from the Aspen Cost Estimator or scaled based on reference costs. The investment cost include the chemical plant, the gasification plant and the CHP plant. Details on the cost estimation are presented in the Supplementary Material.

Eqs 2–4 show the CE (η_CE) for three different system boundaries: overall process, CO production (gasification or combustion followed by CO₂ electrolysis), and electrochemical CO reduction. η C E , t o t a l = n ˙ C , E t h y l e n e + n ˙ C , E t h a n o l + n ˙ C , A c e t i c   a c i d n ˙ C , B i o m a s s (2) η C E , C O   p r o d u c t i o n = n ˙ C , C O n ˙ C , B i o m a s s (3) η C E , C O R = n ˙ C , E t h y l e n e + n ˙ C , E t h a n o l + n ˙ C , A c e t i c   a c i d n ˙ C , C O (4)

The energy efficiency of the process routes and scenarios are calculated based on the comparison of the product energy content (LHV-basis) to the major energy inputs: biomass and electricity. The energy efficiency is calculated for different balance boundaries or process stages, as shown in Eqs 5–8. η E E , t o t a l = m ˙ E t h y l e n e L H V E t h y l e n e + m ˙ A A L H V A A + m ˙ E t O H L H V E t O H + m ˙ H 2 L H V H 2 m ˙ B i o m a s s L H V B i o m a s s + P e l (5) η E E , C O   G a s i f i c a t i o n = m ˙ C O L H V C O + m ˙ H 2 L H V H 2 m ˙ B i o m a s s L H V B i o m a s s + P e l (6) η E E , C O 2   e l e c t r o l y s i s = m ˙ C O L H V C O P e l (7) η E E , C O R = m ˙ E t h y l e n e L H V E t h y l e n e + m ˙ A A L H V A A + m ˙ E t O H L H V E t O H + m ˙ H 2 L H V H 2 m ˙ C O L H V C O + P e l (8)

The specific electrical energy consumption (MJ/kg) per unit of all products is calculated according to Eqs 9, 10. p e l , t o t a l = P e l , t o t a l   f r o m   g r i d m ˙ E t h y l e n e + m ˙ E t h a n o l + m ˙ A c e t i c   a c i d + m ˙ H y d r o g e n + m ˙ O x y g e n (9) p e l , E t h y l e n e = P e l , t o t a l   f r o m   g r i d m ˙ E t h y l e n e (10)

A simplified calculation of direct and indirect CO₂ emissions was performed. The system boundary includes the production phase and not the use phase. Emissions from biomass and electricity generation are considered. The specific supply chain emissions of biomass are assumed at 11.9 kg CO₂/t dry biomass (Jäppinen et al., 2014). Based on prior experience, the emissions from electricity generation have the highest impact on the CO₂ balance of electrified processes. The electrical power P_el comprises the electricity needed from the grid. In scenarios “C,” a portion of the electricity is produced on-site and therefore not accounted for in the CO₂ balance. The emission factor of the grid electricity is varied to see the impact of varying carbon intensity of the used electricity mix. Emissions from additional heat is not accounted for in this calculation since it is assumed to be derived from biomass.

Here, the product-specific emissions are allocated based on the product price, following common practice in case of multiproduct processes. Oxygen is not included in the calculations. The product-specific emission factors are calculated using this equation: E F p r o d u c t = m ˙ b i o m a s s E F b i o m a s s + P e l E F e l m ˙ p r o d u c t C p r o d u c t ∑ i = 1 n m ˙ p r o d u c t , i C p r o d u c t , i m ˙ p r o d u c t (11)

In order to assess the heat that can be utilized in the processes, heat integration was performed using the pinch methodology. The global minimum temperature difference was set at 20°C, accounting for the large number of gaseous streams within the process. The required (minimum) heating and cooling duties can be derived. Purge streams from processes are combusted for fired heat within the processes (e.g., reformer or reactor preheating in ethanol to ethylene process). Surplus energy from purge streams is used to produce high pressure steam at 250°C to be utilized on-site.

Following Figure 1, the gasification route consists of biomass drying and gasifiction, followed by COR and product separation.

The combustion-based route consists of biomass combustion with CO₂ capture, conversion of CO₂ by electrolysis, and electrochemical conversion of CO followed by product separation.

In the electrochemical reactor, carbon monoxide is converted to mostly C₂ products. We assumed ethylene, ethanol, acetic acid, oxygen and hydrogen as products, neglecting other potential by-products. In the CO reduction model, these products are formed following the overall reactions described by Eqs 13–16 (anode and cathode reactions are shown in the Supplementary Material). 2 C O + 2 H 2 O → C 2 H 4 + 2 O 2 (13) 2 C O + 3 H 2 O → C H 3 C H 2 O H + 2 O 2 (14) 2 C O + 2 H 2 O → C H 3 C O O H + O 2 (15) 2 H 2 O → 2 H 2 + O 2 (16)

Figure 4 shows a flowsheet of the electrochemical reactor and downstream product separation. CO enters at the cathode which is a gas diffusion electrode (GDE). Gaseous products including ethylene and hydrogen are formed on the cathode and leave the compartment together with unreacted CO. The catholyte and anolyte (KOH solution) is constantly pumped through the catholyte and anolyte chamber. Ethanol accumulates in the catholyte chamber. Acetate, modelled as acetic acid, is found in the anolyte chamber. The dissociation to CO₂ was not considered. The concentration of each liquid product reaches an assumed maximum of 10 wt.% before the catholyte is sent to product separation which is in line with Ripatti et al. (2019) who showed the production of a 1.1 M acetate stream. The catholyte is separated from the anolyte by an anion exchange membrane (AEM).

In Aspen Plus, the electrochemical conversion was modeled using an RStoic reactor connected with two calculator blocks for the calculation of the energy demand, and the product yields based on FE. The electrochemical reactor is operated at 30°C and 1 bar. The FE are assumed at 60% for ethylene, 10% for acetic acid, 20% for ethanol and 10% for hydrogen, in line with recent publications (Ozden et al., 2021). The product distribution depends on many factors and can be adjusted (Ji et al., 2022). In order to evaluate the impact of product distribution, alternative scenarios with increased selectivity towards ethylene were considered. In the scenarios G1-O and C1-O, the following FE are assumed: 85% for ethylene, 5% for acetic acid, 5% for ethanol, and 5% for hydrogen.

75% single pass conversion of CO was assumed. For instance, Dinh et al. (2019) reported a single pass conversion of 68% in a flow cell. In another study, higher single pass conversions of up to 84% were reached but with less favorable product distributions (Ripatti et al., 2019). Since we are recycling back the unreacted CO after product gas separation, the incoming stream is not pure in CO which leads to hampered mass transfer of CO to the cathode. Therefore, very high single pass conversions seem unreasonable. The cell voltage was assumed at 3.5 V, which is slightly above recent experimental data (Ozden et al., 2021; Ramdin et al., 2021). The current density was set at 300 mA cm⁻² (Shin et al., 2021). In their experiments, Xia et al. (2021) showed that higher current densities require a higher overpotential, thus counterbalancing the tradeoff between lower investment cost and higher operating cost.

The outlet of the cathode side contains unreacted CO and the gaseous products ethylene and hydrogen. With two PSA units, the products are separated from CO. The gaseous mixture is compressed to 20 bar in a three-stage compressor. In the first PSA, ethylene is selectively adsorbed. Possible adsorbents for the selective adsorption of ethylene are for example activated carbon (Ramdin et al., 2021), 5A zeolite (van Zandvoort et al., 2019) or metal-organic frameworks (Bachman et al., 2018). The ethylene purity was assumed at 99.5 mol% with CO as the main impurity. The ethylene recovery was set at 80%. The subsequent hydrogen separation yields a 99.99 mol% hydrogen stream at hydrogen recovery of 80%.

For the PSA process the pressure drop along the bed was assumed to be 1 bar and the desorbed gas stream leaves the unit at 1 bar. Separating first ethylene and then hydrogen requires only a single compression step. The residual gas following hydrogen separation contains mainly CO and is recycled to the electrochemical cell. A purge of 5% was introduced to remove impurities like CO₂ or CH₄ originating from gasification or combustion.

The dehydration of ethanol to ethylene is a commercialized process with multiple plants in operation worldwide (Jamil et al., 2022). The selectivity towards ethylene is above 99% (Zimmermann and Walzl, 2010) and can reach 100% under lab conditions (Wu and Wu, 2017). Figure 5 shows the flowsheet of the modelled process.

The Peng-Robinson (PENG-ROB) equation of state model is used except for the washing column where the ELEC-NRTL model was used. The PENG-ROB method is applicable for gas processing (Aspen Technology, 2013). Th ELEC-NRTL model is recommended for processes with electrolytes involved. The treated ethanol stream from the COR (94.6 wt.%) is pumped to 5 bar. In a series of four adiabatic reactors (RX1 to RX4) with interstage heating, the conversion of ethanol to ethylene is accomplished. The reactors contain an alumina-based catalyst. Coke formation is low in presence of water and conversion is above 99% (Morschbacker, 2009; Andrade Coutinho et al., 2013). The inlet temperature to the reactors is 450°C (Morschbacker, 2009). The reactors are modelled with RStoic reactors with a fixed yield distribution (Arvidsson and Lundin, 2011). Besides the main reaction of ethanol dehydration (Eq. 17), seven side reactions are considered according to reaction Eqs 18–24. The conversion in the reactor is adjusted to keep the temperature change between inlet and outlet of the reactor at 100°C (Andrade Coutinho et al., 2013). In the last reactor the outlet temperature is at 377°C. The overall conversion to ethanol is above 99%. The selectivity towards ethanol ends up at 97%, within the reported range of 95%–99% (Morschbacker, 2009). C 2 H 5 O H → C 2 H 4 + H 2 O (17) 2   C 2 H 5 O H → C 2 H 5 2 O + H 2 O (18) C 2 H 5 O H → C 2 H 4 O + H 2 (19) C 2 H 5 O H + H 2 → C 2 H 6 + H 2 O (20) 3   C 2 H 5 O H → 2   C 3 H 6 + 3 H 2 O (21) C 2 H 5 O H → 0.5   C 4 H 6 + H 2 O (22) C 2 H 5 O H → C H 4 + C O + H 2 (23) C 2 H 5 O H + H 2 O → C O 2 + C H 4 + 2   H 2 (24)

After the last reactor, the stream is cooled to 40°C and washed in a column with water entering at 30°C. The column has five stages. Here, unreacted ethanol and other water soluble oxygenates are removed. The option to recycle ethanol to the reactor feed was not used due to the low concentration in the wash water. A caustic wash for the removal of carbon dioxide was not implemented since the carbon dioxide concentration in the washed gas stream is very low.

In the next step, the raw ethylene stream is compressed to 25 bar in a multistage compressor, with intercooling at 30°C and liquid knockout after each stage. The remaining water is removed by molecular sieve adsorption, modelled as an ideal split (Morschbacker, 2009). The stream has an ethylene purity of 98 mol% after this step. After precooling the ethylene stream to −10°C, it is further purified in a cryogenic rectification column with 20 stages (feed stage: 5), at distillate to feed ratio of 0.98 and reflux ratio of 2.44. The purity of ethylene is increased to 99.5 mol%. Ethylene and minor impurities from the top of the column are heated to 15°C to recover the low temperature energy. The bottom product, containing mostly 1-butene, is purged.

Table 4 summarizes the carbon and energy efficiencies, and specific electrical energy consumption for ethylene production. Since the upstream process for the production of CO is similar for the scenario categories “G” and “C,” their carbon efficiencies are equal. For the gasification-based scenarios (G) the efficiency is lower compared to the combustion scenarios (C). This is mainly due to the syngas stream from the hydrogen separation (PSA) which is sent to combustion. The carbon efficiency of the COR process is equal for all scenarios, at 97%. For reference, the experimental carbon efficiency (single pass) towards C₂₊ products has been reported at 84% (Xia et al., 2021). The total carbon efficiency is therefore mainly influenced by the upstream process for CO production. Scenarios “C” show distinctly higher efficiencies compared to scenarios “G.” The carbon efficiency for scenarios “C” could be further increased by increasing the capture rate of the MEA scrubber. The impact of carbon losses in the ethanol to ethylene process on the overall carbon efficiency is minor since only a small portion of the overall carbon is treated in the ethanol to ethylene process with a carbon efficiency of 96%.

In contrast to carbon efficiency, the total energy efficiency is significantly higher for scenarios “G” compared to scenarios “C.” The total energy efficiency is also governed by the upstream process for CO production; the energy efficiency of CO production via gasification shows a higher efficiency compared to combustion.

The specific electrical energy demand shows a similar behavior for scenarios “C” and “G.” The scenarios “1” show higher specific energy demand compared to “1-O” and “2.” The electrical energy demand for scenarios “G” is significantly lower than for scenarios “C.” The results indicate that in terms of specific electrical energy demand, the scenarios “2” are in the similar range than scenario “1-O.” When assessing the electrical energy demand for the total mass of products, the picture changes. The scenarios “G” show a higher specific energy demand compared to scenarios “C.” This is due to the production of oxygen in the CO₂ electrolysis which impacts immensely the total product mass flow.

The ethylene yield from biomass is generally higher for scenarios “C” compared to “G” due to the higher carbon efficiency of the process. Clearly, “1-O” scenarios, considering optimized ethylene yields from COR, show higher yields for ethylene compared to the other scenarios. The maximum yield of 0.44 t/t biomass is found in case “C1-O.” For reference, Li et al. (2022) report an ethylene yield of 0.30 t/t via first generation bio-ethanol. For second generation bio-ethanol the reported yield drops to 0.10 t/t and for gasification-based routes the yield ranges from 0.10 to 0.17 t/t (Li et al., 2022).

The mass balances for all process scenarios are given in Table 5. The biomass input is equal for all scenarios. The output of products is generally larger for the combustion-based route “C” compared to their equivalent gasification-based scenarios “G.” The annual ethylene production rate of ethylene ranges from 36.4 to 68.4 kt within the different scenarios.

The output of ash and unburnt carbon from combustion or gasification is equal in all scenarios. The scenarios “G” show a huge purge stream that is mostly responsible for the low carbon efficiency of the gasification process. The gasification process also releases a mass flow of 1 t/h H₂. In the combustion case, the power plant flue gas consists of 20.7 wt.% CO₂, 3.4 wt.% O₂, 12.1 wt.% H₂O, and 63.8 wt.% N₂. The power plant generates 94.2 t of high-pressure steam per hour.

In the CO₂ electrolysis, the CO₂ is converted to CO and O₂. The mass flow of CO in scenarios “G” is 14.9 t/h. In comparison, the CO flow for scenarios “C” is 33% higher, consequently leading to higher products mass flows. The CO stream (99.5 mol%) from CO₂ electrolysis contains CO₂ as the only impurity. The impurities for the CO stream from gasification (98.75 mol%) include CH₄, N₂, and CO₂.

Since the electrochemical conversion of CO to products is not complete, unreacted CO together with impurities and gaseous products are recycled. The recycle stream is approximately 50% of the incoming raw material stream. The concentration of CO in the recycle streams is approximately 60 mol% and 46 mol% for scenarios “C” and “G,” respectively. Major impurities in scenario “G” are CH₄ (22 mol%) and N₂ (12 mol%); for “C,” it is CO₂ with 18 mol%. The rest of the recycle stream consists of H₂ and ethylene. The accumulation of impurities in the recycle stream shows that high purity CO as raw material is important to avoid large recycle streams. Alternatively, a higher purge fraction would be able to remove impurities, leading to loss of CO. Clearly, high per-pass conversion in COR would also reduce the recycle. In scenarios “1-O,” the production of ethylene increases, and the by-product streams decrease due to higher assumed ethylene selectivity in COR. Additionally, these scenarios show slightly larger recycle and purge streams due to a larger gas stream leaving the COR cell.

The global ethylene balance shows that compared to the scenarios “1,” the ethylene output is increased by 42% and 31% for scenarios “1-O” and “2.” All scenarios produce a tremendous amount of O₂ from COR and CO₂ electrolysis. In case of the gasification-based scenarios, oxygen is needed internally for gasification. Nevertheless, scenarios “G” produce more than 10 t/h of surplus oxygen. The O₂ surplus for scenarios “C” is around 34 t/h. The amount of O₂ in scenarios “C” would be sufficient for oxyfuel combustion in the power plant, which would avoid energy-intensive post-combustion CO₂ capture. Hydrogen is produced in COR and the gasification process. The H₂ output from gasification is significant, resulting in a clearly higher H₂ output for scenarios “G.”

Table 6 shows the absolute values of energy flows for all scenarios. All cases are scaled to a biomass input of 100 MW_th. The electrochemical reduction of CO is the major electricity consumer. Scenarios “1-O” have a higher electricity demand in COR since the reaction towards ethylene consumes more electrons compared to reactions to acetic acid and hydrogen. The electrical energy input in scenarios “C” is increased by the significant electricity demand of CO₂ electrolysis.

The electricity output from the CHP plant varies due to extraction of steam to supply heat to CO₂ capture and other process stages. The CHP plant reaches an electrical energy efficiency of 26% in condensation mode, comparable to a reference range of 16%–36% (Lako et al., 2015). Steam extraction from a back pressure turbine results in a drop in the electrical efficiency. In scenarios “C1” and “C2,” all the steam after the first high pressure turbine is used for process heating. Nevertheless, additional process heat is needed for the overall process. The steam demand is lower for the “C1-O” case, resulting in a higher electrical energy output from the CHP plant. The higher electricity demand for compression and pumping in scenarios “G” arises from the gas compressors in the membrane and PSA gas separation steps. The ethanol to ethylene process included in scenarios “−2” requires refrigeration for the cryogenic distillation.

The purge streams from different process sections in scenarios “G” have a high thermal energy content, and are sufficient to supply heat to the reformer and the fired preheaters in the ethanol to ethylene process. The reformer thermal energy demand amounts to 9.3 MW. Surplus energy from purge streams is converted to steam at 250°C.

The CO streams supplied to the COR is higher in scenarios “C,” leading to increased electricity demand. In scenarios “G,” hydrogen production is increased by the hydrogen separated from the syngas stream. The decreased amount of by-products in cases “1-O” leads to a tremendous drop in heating demand for acetic acid and ethanol purification. This shows the importance of COR selectivity in terms of the overall process energy efficiency.

Pinch analysis was used to calculate the minimum external cooling and heating demand of the processes (Table 7). The corresponding composite curves are shown in the Supplementary Material. In order to reach the calculated targets, the heat exchanger networks would need to be designed and included, leading to additional capital costs; this was not considered in this paper. Additionally, the results represent the maximum heat integration which might be technically possible but probably not economically optimal. The heat for the MEA CO₂ capture (scenarios “C”) is not explicitly incorporated in the heat integration since the energy demand is directly satisfied by low pressure steam from the CHP plant.

In general, the cooling demand is tremendously higher than heating. Scenarios “1-O” show lower heating and cooling demands with decreased energy inputs to product separation. For these scenarios, no external heating is needed. 14 or 3 MW of low temperature heat for district heating (above 100°C) can be exported in scenarios “G1-O” and “C1-O,” respectively. The heating demand in scenarios “C1” and “C2” can essentially be covered by high pressure steam available from the first turbine stage of the CHP plant. In contrast, scenarios “G1” and “G2” require significant external heat inputs.

The ultimate goal of the suggested process routes is to reduce CO₂ emissions. The carbon intensity of the electricity used has the highest impact on the process CO₂ balance. Figure 6 shows the specific CO₂ emission as a function of carbon intensity of electricity input. The emission factors are presented for ethylene and key by-products, following emission allocation by product value. The horizontal lines show the CO₂ emission factor for fossil-equivalent product. The vertical lines show the reference emission factor for hydro power, wind power and photovoltaics (PV). The intersection points of the horizontal reference line with the linear curves (scenarios) represent the marginal emission factor for electricity. At this point, the specific emissions of the fossil equivalent and the suggested scenario are equal. It is noted that the reference value, with location- and technology-based variation, influences the marginal emission factor. Details on the reference values and data on overall CO₂ emissions for the scenarios are available in the Supplementary Material.

The marginal emission factors generally increase from ethylene, acetic acid, ethanol to hydrogen. For hydrogen it ranges from 213 to well above 300 g/kWh. In comparison, the emission factors for ethylene range from 45 to 79 g/kWh. The results show that the specific CO₂ emissions for ethylene cannot be reduced compared to the fossil-equivalent using electricity from PV. Depending on the scenario, PV can reduce CO₂ emissions (compared to fossil-equivalent) for acetic acid and ethanol. For hydrogen, the emission factor is reduced in all cases. Wind power can reduce CO₂ emissions for all products and scenarios.

The combustion-based routes have lower marginal emission factors compared to the gasification-based routes. Thus, the emission reduction potential is higher for scenarios “G”. Furthermore, scenarios “1-O” with optimized selectivity towards ethylene are inferior in terms of emissions, meaning that improvements in selectivity are not improving the sustainability of the process. The reason is the higher electricity demand for these processes. The difference of CO₂ emissions for scenarios 1 and 2 are marginal. External heat supply is not considered in the calculations. Fossil-based heat generation would diminish all benefits of the gasification-based route. In practice, the heat would likely be supplied by biomass combustion, which would result in a reduction in overall energy efficiency.

If wind power is exclusively used as power source, the emissions of ethylene can be decreased by 32%–60% compared to the reference emission factor. The highest reduction potential (55%–60%) is found for scenarios “G” (detailed values for all scenarios in the Supplementary Material). The highest reduction potential can be realized for hydrogen, at up to 92%. The lowest reduction potential for all products is found in the “1-O” cases.

The CO₂ balances exclude process related emissions from the combustion of purge streams as they are on the basis of their biogenic origin. At the same time, any potential credits for long-term carbon storage in the products are not considered. As seen, the carbon efficiency of scenarios “C” is higher than that of scenarios “G.” Scenarios “C” emit around 4 t CO₂/h, or 36 kt annually. The contribution of purge stream combustion is small compared to the CO₂ released with the flue gas (following CO₂ capture) from the CHP plant. For scenarios “G,” emissions are around 12 t CO₂/h, or 100 kt annually. The highest share of emissions is released from the combustion of the syngas purge.

The absolute numbers of the economic assessment are shown in the Supplementary Material. The LCOE is broken down in Figure 7, accounting for by-product sales as credits. The category “Others” includes labor cost, refrigeration, heating and deionized water. Generally, scenarios “C” show higher cost compared to scenarios “G.” Compared to current market prices, the LCOE is six to three-fold higher, respectively. The highest impacts arise from the annuity of investment and the electricity price. The impact of by-product revenue is the largest for scenarios “−1”. The impact of biomass, CO₂, and Others are marginal compared to the other cost categories.

The economics are dominated by the operating costs, accounting for 67% and 66% of the total annual production costs for scenarios “G” and “C,” respectively. The total fixed capital investment for combustion-based routes (approximately 1.4 billion €) are almost twice of those for the gasification-based routes. This is partly due to higher productivity of scenarios “C,” leading to larger equipment sizes. On the other hand, special cost drivers are CO₂ electrolysis (installed cost 221 M€) and the slightly higher cost of the CHP plant in comparison to the gasifier. The installed cost of the electrochemical CO reduction accounts for roughly 85% or 60% of the total capital costs in scenarios “G” and “C,” respectively. The share of CO₂ electrolysis is 28% in scenarios “C.” Therefore, cost reductions in COR and CO₂ electrolysis can have a significant impact on the CAPEX for the suggested process routes. Electricity accounts for 62%–65% of the operating cost. The second largest share is maintenance at roughly 25%. The impact of biomass cost is higher in scenarios “G” (11%) compared to “C” (6%).

“G1” and “C1” show the highest revenue from by-products. By-product revenues for scenarios “G” are higher compared to the equivalent scenarios “C” due to hydrogen output from the gasification process. Ethanol and acetic acid make up for 77%–87% of the total by-product revenue. Oxygen has a share of below 3% for scenarios “G” but is more significant (5%–13%) in “C” due to higher oxygen production and no internal oxygen demand. By-product revenues are obviously reduced in the cases “1-O” with improved ethylene yields. The LCOE between scenarios “C1-O” and “C2” (or, “G1-O” and “G2”) are similar, showing that improving the ethylene yield in the COR process is not clearly superior in terms of economics compared to the process routes with ethanol dehydration. The scenarios “G1” and “C1,” at lowest overall ethylene yields, show the highest LCOE. Thus, maximizing the ethylene output seems to be favorable in terms of the economics.

Since the electricity price has a big impact on the product price, Figure 7 also shows the total product price excluding cost for electricity. Even at free electricity, no scenario is able to reach the current market price of ethylene. An improved yield of the COR process or the conversion of ethanol to ethylene lower the levelized cost of ethylene for both scenarios “C” and “G.” Since the CO₂ electrolysis and COR process show lower technological maturity, cost savings are possible in future. For the CO₂ electrolysis lower investment cost are expected (see Supplementary Material). A lower operating voltage would decrease the electricity demand and consequently the operating cost of the COR process. Higher current densities and lower module cost (in €/m²) can lower the investment cost of the COR process.

The process design is based on a simplified product spectrum limited to ethylene, ethanol, acetic acid, hydrogen, and oxygen. Since the COR process is at an early development stage, the process is not fully understood yet and further development is needed to make the process technically feasible. The purpose of this study is therefore to stipulate future developments for efficiency and economic improvement, and to develop possible business cases for the technology.

The presence of more products, specifically at lower concentrations, will make the product purification more complex. Some studies (Table 2) consider ethylene as the only product which is very unrealistic and underestimates the effort for product purification. Besides capital expenditures, the heating and cooling demand of the purification processes should not be underestimated. To simplify downstream processing, increasing the selectivity of catalysts to selectively form the desired products will be of extreme importance for commercialization. For example, the presence of 1-propanol causes the issue of forming an azeotrope with water, further increasing the complexity of separation (Wu et al., 2020).

Comparing the CO₂ emissions to published estimates on CO2R/COR processes or other renewable routes (such as OCM or MTO) is difficult due to different boundaries and assumptions. Figure 6 showed that for ethylene, the indirect emissions from consumed electricity are very significant. In contrast, the other products can have higher emission factors for electricity to be competitive with the fossil-equivalent product. The CO₂ emissions for ethylene via CO2R/COR range from −17 to 16 t CO_2eq/t ethylene (Table 2). The wide range of variation underlines the point that relevant comparisons are difficult.

Layritz et al. (2021) investigated the electrified OCM process in comparison to conventional production routes. The results showed that emissions of 0.8 t CO₂/t ethylene (value read from diagram) can be achieved, at electricity emission factor of 0.1 t CO₂/MWh. Based on this study, at this emission factor, the specific emissions for ethylene double for the best case (Figure 6). Liptow et al. (2013) report fossil CO₂ emissions for ethylene from wood fermentation, sugarcane fermentation and gasification (via MTO) of 1.3, 1, and 0.6 t CO_2eq/t ethylene. Only the present scenarios “G1” and “G2” can compete with the estimate for gasification-based system, when using electricity from wind power.

The production cost estimates obtained here are well above the highest cost presented in other studies (Table 2). The lower reference values are largely due to more favorable economic assumptions, for instance lower cell voltages, no side product formation, and reduced electricity price (e.g., $20/MWh) (Sisler et al., 2021). The electricity cost assumed in this study (60 €/MWh) necessarily leads to high overall production costs due to high share of electricity in total cost.

In comparison to other upcoming technologies, reference estimates are in a similar range to our results. The OCM process was estimated to yield ethylene cost of roughly 5,000 €/t for the year 2020, decreasing to roughly 4,200 €/t in 2050 (Layritz et al., 2021). For the CO₂ and H₂ based process via methanol synthesis and MTO, the ethylene cost was estimated within the range of 2,160–2,910 €/t, depending on the electricity price (Ioannou et al., 2020). The gasification-based MTO was estimated to give an ethylene cost of 920 €/t (Li et al., 2022). For other processes, based on either fermentation or gasification coupled with ethanol dehydration, costs have been estimated from 712 to 1,876 €/t (Li et al., 2022).

The presented process in this paper is especially economically favorable for smaller plant sizes. Since the investment cost of the electrochemical processes (water electrolysis and COR) are scaled linearly with plant size, the process does not profit from economy of scale. Similarly, the OCM process and MTO process both involving water electrolysis and CO₂ hydrogenation suffer from the same constraint. However, the MTO process based on another sustainable source of methanol will be more competitive in larger scales due the economy of scale.

In order to further increase the ethylene yield, acetic acid can be hydrogenated to ethanol with subsequent conversion to ethylene. Hydrogenation of acetic acid has been demonstrated experimentally (Rakshit et al., 2020) and a simplified process model was developed in a previous publication (Melin et al., 2022). Based on the model results, ethylene production can be increased by 0.3–1.8 t/h (depending on the scenario) with acetic acid converted to ethylene via ethanol (see Supplementary Material for the calculations). Hydrogen for acetic acid hydrogenation could be provided internally within all of the scenarios, except for “C1” and “C2.” The base ethylene yield for the scenarios, in the range of 0.23–0.44 t ethylene per t biomass, could be improved to the range of 0.31–0.50 t/t (highest for “C2”) with the additional conversion of acetic acid.