Techno-economic assessment and Carbon footprint of processes for the large-scale production of Oxymethylene Dimethyl Ethers from Carbon Dioxide and Hydrogen

Poly(oxymethylene) dimethyl ethers (OME) show promising fuel properties enabling drop-in into the existing infrastructure, especially as an alternative or additive to diesel fuel leading to a significant reduction in local...


Introduction
The CO 2 emission reduction targets of the German federal government in the mobility sector were achieved in 2020 according to the German Environment Agency (UBA) report with a 19 million ton reduction to reach an overall 146 million ton of CO 2 -eq. emissions. Nevertheless, this was correlated to the COVID-19 pandemic, where the mobility sector has been behind in its CO 2 emission reduction targets in the last two decades. 1 This sector is required to signicantly reduce its global (mainly CO 2 ) and local emissions specially under the EURO 6 (NO x < 0.08 g km À1 , PM < 0.0045 g km À1 ) and the upcoming stricter EURO 7 emission reduction standards. 2 Diesel engines are currently one of the major internal combustion engines used in the mobility sector, 3,4 with their share expected to grow in various mobility sector modes (passenger vehicles, road freight, ships, etc.). An expected increase in oil demand by passenger vehicles from 20 million barrel per day in 2020 to 27.5 million barrel per day in 2030 was reported by the International Energy Agency (IEA) in 2020. 5 Accordingly, besides combustion engine optimization and the already complex exhaust gas system improvements, altering the active combustion substance (fuel) is one of the most effective strategies to achieve these emission reduction targets.
Poly(oxymethylene)dimethyl ethers with the chemical formula H 3 C-(O-CH 2 ) n -O-CH 3 where n $ 2 (denoted hereon as OME) can be blended with diesel fuel or applied as neat fuel in diesel combustion engines, resulting in signicant reductions in local emission (i.e., soot and NO x ). This is due to the chemical structure of OME, which has no direct C-C bond, their high intermolecular oxygen content and favourable fuel properties. This also enables the use of existing infrastructure for the transportation, storage, and distribution of OME blends with diesel. [6][7][8][9] The well-to-wheel (WtW) CO 2 emissions can be signicantly reduced compared to fossil-based fuels if OME are produced from renewable carbon sources and low-carbon H 2 . 10 Several studies investigated different blends of OME with diesel fuel to conform to the EN590 standard, which showed a signicant reduction in local emissions for both heavy duty and passenger cars. [11][12][13][14][15][16][17][18][19][20][21][22][23][24] Under certain conditions, even a mixture of 10 vol% OME with diesel can lead to a signicant reduction in NO x and soot emissions. 12 A stoichiometric evaluation showed that already 10 vol% OME blend with diesel will correspond to ca. 441 billion litres (diesel equivalent) per year OME production capacity worldwide and ca. 7.58 billion litres per year for Germany, which emphasizes the need for large-scale production plants in blending cases. Various blending rates were investigated in heavy duty and passenger car engines by several researchers, showing signicant potential to increase the engine efficiency by increasing the exhaust gas recirculation (EGR) as a consequence of overcoming NO x and soot trade-off. [25][26][27] Importantly, life cycle assessment (LCA) studies showed the potential of CO 2 reduction based on neat OME or blends with fossil or Fischer-Tropsch diesel. For a certain case study using neat OME 3-5 , Hank et al. evaluated that the WtW greenhouse gas emissions (GHGE) can be reduced by 86%, corresponding to 29 g(CO 2 -eq.) km À1 (OME 3-5 -fuel) compared to 209 g(CO 2 -eq.) km À1 (diesel fuel). 10 Deutz et al. investigated the WtW LCA for methylal (OME 1 ) and concluded that it has the potential to serve as an almost carbon-neutral blending component, where replacing 24 wt% diesel with OME 1 could reduce the global warming impact by 22% and the emissions of NO x and soot by 43% and 75%, respectively. 28 However, special sealing materials such as ethylene propylene diene rubber (EPDM) 26,29 and fuel injection system modications are needed to achieve certain blending rates of OME with diesel fuel. 9 Neat OME applications are discussed using dedicated engines for niche markets (e.g., agricultural engines, stationary engines, and hand-used machinery) and for captive eets (public buses, trains, etc.). 30,31 Moreover, a recent study by Frontier Economics 32 concluded that Power-to-X (PtX) fueled internal combustion engine vehicles (ICEVs) possess advantages over battery electrical vehicles (BEVs) considering the complete value chain efficiency and importantly; including the renewable generator yield efficiency and the uctuating nature of renewable electricity production. The study concluded that the overall efficiency factor in a holistic consideration between BEVs and ICEVs is between 1.1 and 1.6. Considering that PtX synthetic fuels can be produced wherever there is high renewable energy (RE) resource penetration and dense fuels can be easily imported to the utilization point, PtX fuels can be added to existing infrastructure, offering an efficient, fast and reliable solution. Particularly, for transport modes that indeed require high density fuels (aviation, ships, trucks, etc.), there are no feasible means to achieve defossilization targets without PtX fuels. OME production processes have been investigated intensively since the early work by DuPont in the middle of the 20 th century on the production of longer chain OME. 33 Since the 1990s, short chain OME have been recognized as interesting diesel blends or substitutes. 34 Subsequently, intensive research efforts have been focused on engine testing on one hand, and production processes on the other hand, led by Ford motor company and Eni SpA. 35,36 At the beginning of the 21 st century, fundamental developments led by BASF and BP established the production processes for OME on research and pilot scales. Most of the following contributions on the process side were led by Chinese research and industrial groups, especially the important work by China Petroleum & Chemical Corporation SINOPEC. 33,37 An overview of the publications and the patents, together with the research activities in Germany and worldwide considering OME is given in the review work by Hackbarth et al., 38 elaborating the intensity of research in this eld. Currently, some OME plants are in operation or under construction as reported in China with production capacities of 10-400 kt per annum mostly based on the commercial fossilbased feedstock OME 1 or methanol and paraformaldehyde (pFA) or formalin. 38 The product quality, reproducibility, and long-term production capacities of these plants are not dened due to their varying operational strategies. Additionally, the complexity of these plants to achieve pure OME products could limits the scalability of these technologies to the desired largescale, which is needed in the context of usage as a fuel.
Considering the value chain starting from H 2 and CO 2 towards OME production or partially considering the production of OME based on the feedstock methanol, several technoeconomic assessments (TEA) and process simulations for different synthetic routes were pursued. The studies highlighted insights regarding the economic potential of OME production and hurdles regarding the demonstration and technology realization. Prominently, the work by Burger et al., 39 Schemme et al., 40 Bongartz et al., 41 Ouda et al., 42 and Held et al. 43 were considered. However, the underlying assumptions, considered process routes, scope of the evaluation, modelling approaches and boundary conditions vary among these contributions, which lead to different conclusions and complicates the comparison between different process routes. Additionally, a complete and detailed process description identifying the material and energy integration concepts of the sub-processes in a complete process chain, where engineering data and TEA data can be extracted are lacking. The purication of a highly non-ideal and reactive product mixture containing several heterogeneous azeotropes, complex vapor-liquid-liquid equilibria (VLLE), especially the separation of water and formaldehyde from the target OME product, and challenges regarding solidication are cumbersome. This leads to simplication strategies and assumptions, particularly on the product purication side, which in some cases lead to misleading results concerning the OME product composition and purity. Additionally, material and energy balances can deviate signicantly from experimental cases.
In the light of the previous discussion, the introduction of OME to the market at large-scale requires next to the technological developments, a detailed comparison and evaluation of material and energy integrated, efficient, economically feasible and scalable industrial process routes for the production of OME with a certain chain length.

Objectives
Herein, as part of the project Sustainable Mobility through Synthetic Fuels (NAMOSYN), the most promising OME production process routes are evaluated using a standardized simulation methodology with the objective to evaluate and compare the promising process routes and obtain the design parameters for the rst European large-scale OME production process.
In addition to the OME production process concepts, the validation of the employed models using numerous published experimental data will be described. Furthermore, based on the modelling results, the material and energy integrated process, energy efficiency, carbon footprint and the TRL of the investigated process routes will be introduced and discussed. Correspondingly, the net production costs of OME will be presented together with a sensitivity analysis covering the cost impact for H 2 and CO 2 , showing the potential for the sustainable production of OME [3][4][5] . Finally, the understanding from this work can highlight the critical process components in the OME value chain for further R&D endeavours.

Theory and background
For the synthesis of OME, methyl-capping groups such as methanol (H 3 C-OH, MeOH), methylal (H 3 C-O-(CH 2 O) 1 -CH 3 , OME 1 ), and dimethyl ether (H 3 C-O-CH 3 , DME) need to react over acid catalysts with a source of formaldehyde group such as formalin, paraformaldehyde (HO-(CH 2 O) n -H with n ¼ 8-100, pFA), trioxane (C 3 H 6 O 3 , TRI), and anhydrous formaldehyde (H 2 C-O, FA). The reaction proceeds through an initiation, growth, and termination mechanism, as described by Baranowski et al., 44 Schmitz et al., [45][46][47][48] and Oestreich et al. 49 This leads to several simultaneous catalyzed and non-catalyzed reactions and the formation of undesirable side-products such as poly-(oxymethylene)hemiformals (HF), poly-(oxymethylene) glycols (MG), water, and others, as shown in Fig. 1 and extended in the ESI. † The formation of the unstable HF and MG was experimentally investigated and quantied using NMR techniques in a series of fundamental studies, mainly by the group of Hasse et al., [50][51][52][53] where the concepts of the "overall composition" and the "true composition" were introduced. The "true composition" is not measurable using standard analytical techniques and represents all the reaction mixture components, namely, MeOH, H 2 O, FA, OME 1Àn , HF n , MG n and possible side products. The "overall composition" represents the decomposition of HF and MG to their original reactants of MeOH, H 2 O and FA. This is a deciding aspect in process modelling and simulation, especially considering reaction and separation process design. [45][46][47][48] Estimations of heating and cooling duties, power loads and size of unit operation depend on thermodynamic and physical properties of the pure components such as heat capacity, enthalpy of vaporization and vapor pressure to account for real phase behaviour and phase equilibria. Therefore, these results are inuenced signicantly by the true or the overall composition. Besides, to date, components such as OME, HF and MG are not included in commercial owsheet simulation soware. The thermodynamic properties of these intermediates are described in the original literature but the right property implementation and consideration in a process simulation require fundamental knowhow about this reactive system. The work by Maurer, Hasse, Burger and Schmitz et al. [45][46][47][48]54 offers a concrete basis for the validation of the simulation models. Furthermore, the methodology of the implementation of these reaction and phase behaviour considerations along with the whole owsheet simulation was introduced by Bongartz et al. 41 employing tool boxes from the electrochemical eld, namely, the chemistry section in ASPEN Plus® and is further modied in this work. There have been several signicant modelling and simulation efforts to describe this complex system behaviour, where the works by Burger et al., 39 Schmitz et al., [45][46][47][48] Bongartz et al., 41 and Ouda et al. 42 are acknowledged, and the simulation results generated in this work are a progression on their previous work.

Process description
As described previously, OME can be produced starting from different feedstocks. Based on this, several process routes can be allocated to produce a desired mixture of OME 3-5 . An overview of the four selected promising process routes investigated in this work is shown in Fig. 2. The selection of the processes focuses on the feedstock availability on a large-scale, technological maturity of the considered processes and availability of experimental data for consistent process evaluation.
All four process routes (denoted here as P1-P4) start with the synthesis and purication of MeOH from the H 2 and CO 2 feedstocks. MeOH is converted into intermediate products, namely, FA and OME 1 , which are further used to synthesize a mixture containing OME n with n ¼ 1-10. This mixture is then fractionated to obtain a nal product mixture containing OME [3][4][5] . Consequently, the conversion of MeOH to OME 3-5 goes through different subprocesses, which in turn inuence the total process energy efficiency and the product yields. These subprocesses are MeOH synthesis, FA (aqueous) synthesis, FA (anhydrous) synthesis, OME 1 synthesis and OME n synthesis. The combinations thereof lead to the aforementioned production routes of OME. In the following section, each of these subprocesses are briey explained.

MeOH synthesis
MeOH synthesis is one of the oldest thermochemical processes with the highest production capacities and is mainly based on fossil feedstocks. The development of a Cu-based process enabled a signicant reduction in the synthesis conditions to temperatures as low as 200-280 C and pressure of 50-125 bar. Advanced catalyst development allows MeOH synthesis based on a CO 2 -rich feed with enhanced catalyst water tolerance. 55,56 The process conditions for the synthesis of MeOH are based on the work by Otto 57 and Bongartz et al. 58 The reaction network considered in this work is introduced in eqn (1)- (3). The reaction equilibrium and kinetic relations are implemented based on the work by Nestler et al. 59 The MeOH synthesis process takes place at 250 C and 70 bar in a plug ow reactor and the downstream purication of MeOH from non-reacted gases such as CO 2 , CO, H 2 and H 2 O goes through a cascade of ash drums with intermediate cooling, operating at different pressure levels, then followed by a distillation column. The light gases with the non-reactants are recycled back to the reactor to increase the product yield. A simplied owsheet of this subprocess is shown in Fig. 3.
Main reaction network:

FA (aqueous) synthesis
The FA (aqueous) subprocess comprises the conversion of MeOH to FA. Formalin or pFA is synthesized commercially from MeOH. The former is produced either via the silver catalystbased process or the FORMOX process. In these processes, MeOH is partially oxidized over an Ag-based catalyst or metal oxide-based catalyst to selectively produce formalin aqueous solution (FA concentration 37-55 wt%). In the Ag-based process, MeOH is mixed with an air stream and fed to  This journal is © The Royal Society of Chemistry 2022 Sustainable Energy Fuels, 2022, 6, 528-549 | 531 a reactor to be converted to FA via partial oxidation and dehydrogenation reactions, as shown in eqn (4)- (6). Main reaction network: The FA (aqueous) synthesis subprocess takes place at T > 650 C and near ambient pressure in a kinetically controlled regime. H 2 O is formed as a by-product, and thus this subprocess is denoted heron as FA (aqueous). The process concept of this sub-process was presented by Franz et al., 60 which considers the separation of FA from volatile gases in an absorber column using H 2 O as a washing liquid. Providing a FA product stream containing about 55 wt% FA and 45 wt% H 2 O, this stream should be concentrated to be further used for the synthesis of longer chain OMEs. Therefore, this stream is fed to a cascade of two evaporators, which split it into two output streams. The target product from this subprocess stream with about 85 wt% of FA is further used for the synthesis of longer chain OMEs, whereas the evaporator side product stream with about 10 wt% of FA is partially used as a washing liquid for the aforementioned absorber column and partially leaves the subprocess as a wastewater stream.

FA (anhydrous) synthesis
There is no commercial anhydrous FA synthesis based on the endothermic dissociation of MeOH to monomeric FA and valuable H 2 (eqn (7)), although this route has been investigated since 1960 to identify selective catalysts for the anhydrous synthesis of FA. 61 The lack of direct application of the highly reactive monomeric FA product hinders the market establishment of this production route. In the case of OME synthesis, this valuable monomeric FA product is important, and thus this subprocess is considered the "dream reaction" for the OME value chain. The reaction occurs at high temperatures > 650 C, which requires 85 kJ mol À1 FA to be produced. Due to the high reactivity of FA, the retention time is very short to avoid the formation of the thermodynamically unfavoured CO, as shown in eqn (8). The challenge in this reaction system is to reach high MeOH conversions at high FA selectivity without deactivating the catalysts in this strongly reducting H 2 environment, which is an aspect that has been intensively experimentally investigated in the scientic community. For the implementation in the simulation platform, the process described by Sauer et al. 62 combined with the process concept published by Ouda et al. 42 is adapted. MeOH is saturated in a carrier gas and further dissociated at 900 C to FA and H 2 over an Na-based catalyst, following the complete MeOH conversion and selectivity experimentally investigated by Sauer et al. 62 CO is formed as an undesired side-product. For the separation of the monomeric FA from the reaction products, absorber columns using mainly MeOH or recycled OME fractions as washing liquids are used.

OME 1 synthesis
Methylal is available commercially based on MeOH and FA feedstock. The OME 1 subprocess comprises the conversion of MeOH and FA to OME 1 . For the implementation of this subprocess, the process concept was adapted from Drunsel. 63 The synthesis takes place via a heterogenous catalytic reaction at 60 C and 2 bar over an acidic catalyst, e.g., Amberlyst® 15, in a plug ow reactor. Several reactions are possible, as shown in the reaction network eqn (9)- (13). The reaction product purication takes place in a downstream reactive distillation column. The column is used to overcome the reaction equilibrium restrictions and convert the rest of FA almost completely to OME 1 , while separating H 2 O and MeOH from the azeotropic mixture of OME 1 and MeOH. In a consecutive distillation column operating at a higher-pressure level, OME 1 is separated from the azeotropic mixture of OME 1 and MeOH and leaves the distillation column as the bottom product. The distillate product is recycled to the reactive distillation column. Main reaction network:

Higher OME synthesis and purication
For the implementation in the simulation platform, the higher OME subprocess comprises the conversion of the aforementioned intermediates selectively to longer-chain OME and the purication of the target OME 3-5 product. The process concept adapted here is suitable for different feedstocks and was proposed by Schmitz et al. 39,46 OME n synthesis takes place via a heterogenous catalytic reaction at 80 C and 2 bar in a plug ow reactor in the presence of an acidic catalyst, e.g., Amberlyst® 46. The reaction network taking place here is presented in eqn (9)-(15) and summarized in Fig. 1. The reaction product purication takes place in a cascade of two distillation columns and a membrane unit for the selective separation of H 2 O aer the rst column, as shown in Fig. 3. The main product stream containing OME 3-5 exits the subprocess and the streams containing OME <3 and OME >5 are recycled to the synthesis reactor. Main reaction network: Combustion A combustion subprocess was implemented to use the energy of the purge streams to produce process steam, which was utilized in the subprocesses. For the simulation of the combustion reactions, an adiabatic Gibbs reactor was applied, and excess air was added accordingly to achieve complete combustion and keep the adiabatic temperature rise below 800 C. The stoichiometric amount of O 2 required for a complete combustion can be estimated using eqn (16). Main reaction network: Total process routes P1-P4 The combinations thereof lead to production routes P1-P4 of OME. In the following section, each of these total processes are briey explained. Process route P1 consists of the subprocesses MeOH synthesis from H 2 and CO 2 , FA (aqueous) synthesis and the combination of MeOH and aqueous FA to higher OME synthesis and purication.
Process route P2 consists of the subprocesses MeOH synthesis from H 2 and CO 2 and FA (anhydrous) synthesis, in which H 2 is produced as a by-product and separated and recycled to the MeOH subprocess. Downstream MeOH and anhydrous FA are synthesized to higher OMEs and puried. Almost no H 2 O enters the OME 3-5 subprocess, which reduces the formation of side-products and improves the process energy efficiency associated with smaller recycle streams. However, H 2 O is still formed as a by-product in the synthesis of longer chain OMEs and needs to be separated from the recycle stream using a membrane unit.
Process route P3 consists of the subprocesses MeOH synthesis from H 2 and CO 2 , FA (aqueous) synthesis and the combination of MeOH and aqueous FA for OME 1 synthesis. OME 1 is further introduced with aqueous FA to the higher OME synthesis and purication. The advantage of P3 is the conversion of MeOH towards OME 1 prior to the synthesis of longer chain OMEs, which allows the separation of a large portion of the H 2 O formed before the OME n synthesis subprocess. In addition, fewer side-products are formed during the synthesis of longer chain OMEs, thereby reducing the energy demand during product purication towards OME 3-5 and consequently enhancing the process energy efficiency.
Process route P4 consists of the subprocesses MeOH synthesis from H 2 and CO 2 , FA (anhydrous) synthesis and the combination of MeOH and anhydrous FA for OME 1 synthesis. OME 1 is further introduced with anhydrous FA to the higher OME synthesis and purication. Hence, this affords the synthesis of longer chain OMEs from H 2 O free reactants. This prevents the formation of the side-product H 2 O, and thus various other side-products are not formed, which is a very important advantage for the purication of the OME n reaction product. This is very advantageous given that the recycle of the H 2 side-product of the FA (anhydrous) synthesis leads to signicant material and process energy efficiency improvements, in addition to the process energy efficiency improvement potential due to the signicantly less purication energy demand in the OME 3-5 subprocess. More details on process routes P1 to P4 are presented in the ESI. †

Methodology
This chapter addresses the general assumptions, system boundaries and methodology of process modelling and simulation, followed by process techno-economic and carbon footprint evaluation methodologies. This is then followed by the description of the comparison criteria.

General assumptions and system boundaries
The system boundaries were set for the evaluation on the simulation level from the feedstock H 2 and CO 2 , followed by the synthesis of intermediates up to the desired product OME [3][4][5] . It is assumed that the production plant is integrated in a chemical park where the necessary infrastructure for the provision of utilities such as steam and cooling water is available at market prices. CO 2 and H 2 are purchased as waste products or raw materials from renewable non-fossil sources. The waste streams of the processes are wastewater and exhaust gases. The wastewater is treated at market prices, while the exhaust gases are assumed to be released without further treatment. The production of OME 3-5 consists of several subprocesses, which are altogether material and heat integrated. The obtained distribution of OME chain lengths in the nal product OME 3-5 differs slightly between the investigated process routes. Nevertheless, it is assumed that in all cases, the specication range is fullled without further processing and that the heating value from Held et al. 43 represents the actual heating value as a good approximation. The system boundaries are shown in Fig. 4, and further details on the assumptions are given in the ESI. †

Process modelling and simulation
Steady-state simulations for P1 to P4 were implemented using Aspen Plus® soware V11 from Aspen Technology Inc. Aspen Energy Analyzer V11 and Aspen Process Economic Analyzer V11 were used for heat integration and unit operation dimensioning.
The components considered in the simulations were H 2 , CO 2 , CO, N 2 , O 2 , FA, MeOH, H 2 O, OME 1-10 , HF 1-10 and MG [1][2][3][4][5][6][7][8][9][10] . Most of these components such as the acetals, glycols and OME are not included in the standard Aspen database, and thus new components were added in the property analysis environment. An overview of the pure component thermodynamic and thermophysical properties used in the simulations can be found in the ESI. † To simulate purication processes, in particular thermal separations, interaction parameters are required, which describe the real behaviour of the gas and the liquid phases. For mixtures that contain FA, a UNIFAC-based model was introduced by Maurer et al. 54 This model was extended in the following decades by adding new components and by adapting the interaction parameters to new experimental data. Schmitz et al. 47 published a new version of the model considering OME n which was adapted and implemented. An overview of the model and its validation is presented in the ESI. † Adequate model parameter and implementation of the thermodynamic model is crucial for a realistic simulation of this special reactive mixture. A variety of reaction models describing the MeOH, FA (aqueous), FA (anhydrous), OME 1 and OME n syntheses were used to assess the product compositions exiting the reactors. The models implemented in the simulation environment can be found in the process description of the subprocesses. The synthesis of FA was described based on the conversion and yields from literature values, while the other syntheses were described according to published kinetic models. An overview of the employed reaction models is presented in the ESI. † Initially, the subprocesses were implemented separately in the simulation platform. Aerwards, the material integration interconnecting these subprocesses to describe process routes P1 to P4 was implemented. The production capacity was adjusted to 100 kt OME [3][4][5] per year, and the heat integration was conducted to improve the overall process energy efficiency. Consequently, the heat exchanger network was designed. Aer the network design, heat exchangers were interconnected in the process simulation to transfer heat from hot streams to cold streams and steam utilities were implemented to consider the steam supply by single process unit operations and the steam demand of other process unit operations. Subsequently, the process components were dimensioned. Based on the simulation material and heat balance, the utilities demand and main operational cost parameters were extracted. Since a complete process route is a combination of several subprocesses, more recycle loops that are interconnected should be converged. Standard numerical solvers in Aspen Plus® were used, while the complex loop convergence of the total process routes was achieved stepwise starting with connecting the subprocesses. Aer the material integrated loop convergence, heat integration of the total process routes was performed stepwise, followed by convergence of the energy integrated loop. The equipment sizing of unit operations followed the convergence and integration steps using Aspen Process Economic Analyzer V11. For the parameters categorized as non-conventional unit operations, namely, the membranes, heat exchangers, reactors, and the column dimensions for the FA (anhydrous) absorption considered as ideal separation units, literature-based sizing methods were adapted. Membrane areas were evaluated according to the methods described by Schmitz et al. 47 and Baker, 64 while the column dimensions were estimated by correlation with the absorption column in the aqueous FA subprocess. A detailed description of the procedure for dimensioning the process components and the heat integration procedure applied for the process routes can be found in the ESI. †

Process evaluation and comparison criteria
The implemented processes P1 to P4 were evaluated using various key performance indicators (KPIs). These KPIs were translated into process evaluation criteria used to compare the process routes, given that they provide a concise summary of the different process routes in terms of mass and energy balance. The process route utility demands and process energy efficiencies were evaluated based on the total mass and energy balance from P1 to P4.
In this work, the energy efficiency of the overall process, h energy , is dened in eqn (17).
where _ m denotes the mass ow rate of the reactants i and the OME 3-5 containing product stream. LHV is the lower heating value at 298 K, while _ Q k and _ W l represent the externally supplied heat uxes and electric power demand, respectively.
In addition to the energy efficiency of the process routes, the material balance was assessed, and the performance was indicated by two parameters. The rst parameter, h C , reecting the carbon efficiency, i.e., the ratio of carbon atoms, C, in the feedstock and the carbon atoms in the OME 3-5 product stream, is dened in eqn (18). The second parameter, h mass , considers the mass ow rates, i.e., the ratio of the OME 3-5 product mass ow rate with respect to the feedstock mass ow rate, as dened in eqn (19).
where _ m denotes the mass ow rate of the reactants i.

Economic evaluation
The methodological approach for production cost estimation is a factorial method described by Peters et al. 65 based on the equipment data and material and energy streams obtained from the process simulations. The calculations were carried out with the DLR-inhouse soware Techno-Economic Process Evaluation Tool (TEPET) in an automated and standardized manner. 66 The targeted parameters of the economic analysis are the investment and operating costs as well as the net production costs related to a production unit. An overview of the most important assumptions for the economic evaluation is given in Table 1 (detailed list is given in the ESI †).
The net production costs of fuels from renewable H 2 and CO 2 are typically highly sensitive to the raw material price of the educts. 67 Hence, the future price development exhibits high uncertainty. Consequently, a wide range of CO 2 and H 2 costs was examined to evaluate the inuence of cost variations.

Capital expenditures (CAPEX)
CAPEX are calculated based on the purchased equipment costs of the main process equipment. The equipment was categorized as conventional and non-conventional and the sizing data was obtained based on the simulation results explained in the previous section and further explained in the ESI. † For conventional unit operations, packed columns and shelland-tube heat exchangers are assumed as equipment types for all applied columns and heat exchangers, respectively. For H 2 compression reciprocating compressors were assumed, while all other compressors were taken as centrifugal-rotary ones. Knockout drums were considered as storage vessels for the cost calculation. 73 The costs of multistage compressor cooling, absorber cooling and column reboilers and condensers were considered by applying shell-and-tube heat exchanger costs. Reux pumps were considered separately in the column cost calculation, while the costs for column insulation and connections were neglected. Also, 316 stainless steel was assumed as a standard material due to the presence of FA in the process, 74 except for the burner consisting of carbon steel. For non-conventional unit operations, H 2 O membrane plate-and-frame modules were assumed, while the membranes for H 2 separation were taken as hollow-ber modules in accordance to Baker. 64 The use of one uniform data base is preferable in the selection of the specic cost for the different types of equipment. Hence, cost functions from Peters et al. 65 were applied as appropriate. Specic costs from references other than Peters et al. 65 were only chosen in the case of multitube reactors, xed-bed reactors, thin lm evaporators and membrane modules due to the inadequate capacity range or lack of data given in Peters et al. 65 A detailed explanation of the deviations from the standard reference and the related assumptions is given in the ESI. † When the capacity limit of the chosen cost function was exceeded, the equipment was separated formally into the minimum number of equally sized components to t the given range.
Lang factors for uid processing plants are assumed for the calculation of the additional direct and indirect CAPEX from the purchased equipment cost. 65 Reduced Lang factors for main compressors were assumed considering a reduced share of additional CAPEX in the case of compressors larger than 1000 kW of nominal power. An overall cost factor of 2.27 (compared to 5.93 for all other equipment) was applied. 68 Fixed capital investment (FCI) is calculated from the total direct and indirect cost by assuming an additional share of contractor's fee and contingencies. 65 The total capital investment (TCI) includes a working capital of 15% of the TCI. 65 Annual capital cost (ACC) is determined from the TCI by the annuity method 65 assuming an interest rate of 5%. More details of the methodological approach are described by Albrecht et al. 66 All the applied Lang factors are given in the ESI. †

Operational expenditures (OPEX)
For the determination of raw material and utility costs (OPEX R&U ) all mass and energy input streams and wastewater output streams and surplus steam provided by the simulation results were considered. H 2 and CO 2 costs in the base case were taken from generic values, 68 which were calculated following the same assumption basis as in this work. The lower limit of the considered CO 2 cost range was taken from Naims. 69 In this reference, the CO 2 costs for carbon capture from various point sources were examined and a value of approx. 65 V per t CO 2 was estimated for a largescale carbon capture and utilization (CCU) scenario. For the upper limit, a value of 720 V per t CO 2 was applied according to House et al. 70 for direct air capture (DAC). To the best of our knowledge, this represents the highest CO 2 cost from DAC published to date. 75,76 The H 2 cost ranges were taken from the Hydrogen Europe report 71 considering steam methane reforming (SMR) costs as the lower limit, which are assumed to be the target value for green H 2 costs that could possibly be reached using the proposed optimization strategies. The conservative estimations of electrolysis cost from wind power were taken as the upper limit. 71 The assumed H 2 and CO 2 cost ranges are given in Table 1.
The requirement of high temperature heat (>250 C) was covered by electrical power, assuming a power-to-heat efficiency of 95% for energy conversion. 77 All catalysts were considered as raw material streams calculated from the catalyst lifetime. The costs were estimated from the trading price in the case of silver and by extrapolation of low quantity prices on bulk prices in the case of all other catalysts. All other specic costs for raw material and utilities as well as the assumed catalyst lifetimes are given in the ESI. † Employee-hours per year (h labor ) were estimated from the characteristic plant capacity and the number of production steps according to the procedure proposed by Peters et al. 65 Specic labor costs for Germany were assumed. 78 Additional direct and indirect operational expenditures (OPEX dir/ind ) were calculated with the factorial method described by Peters et al. 65 and Albrecht et al. 66 Given that only the manufacturing costs were determined, no costs for distribution and selling as well research and development were considered. All applied cost factors are given in the ESI. †

Net production costs (NPC)
Specic net production costs (NPC) were calculated from ACC, total OPEX and the product output and related to a litre diesel equivalent (l DE ) according to: with the output OME 3-5 mass ow _ m OME 3-5 , the lower heating value LHV OME 3-5 of OME 3-5 , 43 and the lower heating value LHV diesel and density r diesel of diesel. 72 CO 2 footprint evaluation LCA is a methodology for evaluating the environmental impacts of product systems along the entire life cycle. LCA is standardized in the ISO 14040/14044 standards 79,80 and considers all environmental impacts of the material and energy ows that are exchanged with the environment.
The carbon footprints of the four process routes P1-P4 for OME 3-5 were compared by applying a well-to-tank approach. The system boundaries of all routes with feedstock and energy supply are described above and shown in Fig. 4. Note that the CO 2 emissions from the use-phase and end-of-life phase of OME 3-5 were neglected given that both are identical for all four routes, and thus neutralize each other in a comparison. The construction of chemical plants was neglected in this analysis due to the lack of data.
Comparing technologies consistently requires a common basis. In LCA, this common basis is the so-called "functional" unit. 77 For the well-to-tank approach, we choose "the provision of 1 MJ of enthalpy of combustion" as the functional unit and additionally present the results per litre diesel equivalent. For this study, we focused on evaluating the carbon footprint in the life cycle impact assessment (LCIA) of the four alternative process routes. The environmental impacts were assessed in accordance with the LCIA methodology "ReCiPe Midpoint (H) V1.13 no long-term (LT)" following the methodological guidelines for LCA of synthetic fuels of the research center for energy economics (FfE). 81 Although other impact categories are also relevant, they are beyond the scope of this study.
For the life cycle inventory (LCI), we considered generic process data of the FfE for H 2 , CO 2 , and electricity supply, as shown in Table 2. 68 For the electricity supply, we considered today's electricity grid mix in Germany given that it is modelled based on the guideline of FfE. For process heat, we assumed that heat below 250 C is supplied by steam that is produced as an energy carrier in the chemical industry, 82 while for heat above 250 C, we assumed an electrode boiler with a power-to-heat efficiency of 95%. 77 Note that processes routes P1, P2, and P4 export steam, for which we consider an environmental credit, i.e., avoided burden, for steam production as energy carrier in the chemical industry. 82 For the H 2 supply, we considered the average carbon footprints of alkaline electrolysis (AEL), solid oxide electrolyser cell (SOEC), and polymer electrolyte membrane (PEM) electrolysis. However, it should be noted that conventional SMR would result in a lower carbon footprint of H 2 supply than electrolysis combined with today's electricity grid mix in Germany. Specically, the carbon footprint of today's German electricity grid mix, as considered in our base case year of 2018, is considerably high with 24.93 kg CO 2eq. per kg H 2 . Thus, due to this reason, we analyzed the inuence of electricity supply on the carbon footprint in a sensitivity analysis, as shown in Fig. 13. CO 2 is supplied by the average of DAC and mono-ethanol amine (MEA) scrubbing in the cement industry. The environmental credit, i.e., avoided burden, for CO 2 utilization is credited to the production of OME [3][4][5] .
For additional process data for the supply of utilities such as nitrogen, compressed air, cooling water, and wastewater treatment, we considered LCA datasets of the ecoinvent database. 82 The ESI † presents more details on the used LCA datasets.

Technology readiness level (TRL)
The TRL is dened as a criterion for evaluating the development status of new technologies based on a systematic analysis, which indicates on a scale of 1 to 9 how advanced a technology is. The TRL of the ve individual subprocesses and the total process routes P1-P4 was assessed based on the published TRL scale in the Energy Research Program of the German Federal Government addressing the innovations for the energy sector. 83 The TRL scale is presented in the ESI. † For the purpose of assigning the TRL of the total process routes P1-P4, not only the subprocess with the least TRL was considered to determine the TRL of the total process route, rather we also assigned a mean TRL value for the subprocesses. Thereby, a better comparison of the TRL of the total process routes is possible. Given that all process routes are limited by individual process components such as the FA synthesis reactor in the FA (anhydrous) subprocess and the H 2 O separation unit in the OME 3-5 subprocess, limiting the TRL of the total process routes to the aforementioned process components would reduce the TRL of P1 to P4 to TRL 3-4, while P1 and P3 would benet from the high TRL of the FA (aqueous) subprocess. Thus, the mean TRL value allows a realistic presentation of the signicant research and development work, which is still to be done to realize the process routes to any considerable extent.

Results and discussion
Process simulation and evaluation

Energy and mass balance
Aer implementing the complete processes in the simulation platform and applying the described methodologies, the results were used for evaluation and comparison. Fig. 5 presents a summary of the results of the mass and energy balance of P1-P4 in the form of e-Sankey diagrams. In addition, the input and output streams are listed in Table 3 with respect to the main product stream OME 3-5 , accounting for the results of the mass balance at a production of 100 kt per annum OME [3][4][5] . The mass balance evaluation of P1-P4 showed that P1 and P3 require more H 2 and less CO 2 feedstock in comparison to P2 and P4 to produce the targeted 100 kt per annum OME [3][4][5] . In fact, this is the outcome of the two different process design concepts for the production of FA relying on aqueous or anhydrous FA. Considering the aqueous routes, they are characterized with a higher production of H 2 O, which is the by-product of the acetalization reaction and exits the process as wastewater streams in the case of P1 and P3. Moreover, P1 and P3 have smaller exhaust gas ows due to the use of O 2 as oxidizing agent for the FA (aqueous) subprocess. In contrast, in the FA (anhydrous) subprocess considered in P2 and P4, N 2 is used as a carrier for the feedstock MeOH, which should be introduced at certain dilution to the FA reactor. As a result of purging a portion of the carrier gas to prevent the accumulation of the side product CO, the FA (anhydrous) subprocess has a higher exhaust gas ow. In addition, the side product CO of the FA (anhydrous) subprocess leads to a higher demand of CO 2 for P2 and P4. Alternatively, the H 2 side-product of the endothermic MeOH dissociation reaction in the anhydrous FA synthesiswhich is recycled to the MeOH synthesislowers the demand for the total process H 2 feedstock in comparison to P1 and P3. Consequently, this results in higher input and output mass ows for P2 and P4. In addition, the OME 3 , OME 4 and OME 5 compositions reveal small differences between the process routes. However, the study here focused on similar product compositions rather than minimal recycle ows to dene the feedstock composition of the OME 3-5 subprocess. This approach is based on the assumption that product composition is of greater importance to the application than the process energy efficiency of the production process. Table 4 shows the energy ows of the input and output streams relative to the energy ow of the product stream OME 3-5 for P1-P4. The energy ow is expressed as the product of LHV and mass ow. In addition, the energy demands in terms of electricity, steam, cooling water and heat utilities above 250 C are shown relative to the OME 3-5 product stream.
The energy balance of the considered process routes shows that P2 and P4 require more electricity, which is mainly due to the higher energy demand at the high temperature level > 250 C for the synthesis of FA (anhydrous). Additionally, the higher dilution rate required for the successful conversion of MeOH to FA and H 2 increases the demand for the compression recycling of the carrier gas stream.
Moreover, P1, P2 and P4 show negative steam demands in the case of 4 or 20 bar, which is mainly due to the fact that the Table 3 Results of the mass balance for P1 to P4 for a production of 100 kt per annum OME [3][4][5]  a Air used for the FA (aqueous) synthesis and for the combustion of purge streams, while the generated heat was utilized in the processes, as shown in Fig. 3.  steam generated in the MeOH synthesis reactor is used to supply the heat to the reboilers in the OME 3-5 subprocess. This effect was also elaborated by Schemme et al. 40 In addition, different feedstocks for the OME 3-5 subprocess result in different mass ows being fractionated in the distillation columns due to the formation of side-products, which in turn leads to different reboiler and condenser duties. The absorber of the FA (aqueous) subprocess requires cooling water with maximum temperature rise of up to 20 C. Overall, P1 and P3 show a lower electricity demand and higher steam and cooling water demand but no demand for heat above 250 C.

Process efficiencies
Based on the methodology discussed in Section 3, different process energy and material efficiencies were evaluated and the results are listed in Table 5 and presented in Fig. 6. Process routes P2 and P4 comprising the anhydrous FA synthesis subprocess exhibit the highest energetic efficiencies due to the recycling of the valuable side product H 2 as a feedstock. In contrast, P2 and P4 exhibit lower carbon efficiencies principally due to the side reaction in the synthesis of FA (anhydrous) to CO. Evidently, as shown in Table 4, the lower energetic efficiencies of P1 and P3 arise principally from the higher H 2 demand, which is not fully compensated for by the heat required at above 250 C in P2 and P4. The lower overall material efficiency, h mass , of the feedstock being converted to OME 3-5 for P1 and P3 is a result of the production of large amounts of the side product H 2 O in the synthesis of FA (aqueous) due to the MeOH partial oxidation reaction. This generated H 2 O is separated with large effort downstream to the FA (aqueous) subprocess and leaves the process in the form of wastewater.
Held et al. 43 investigated different scenarios to produce OME 3-5 based on stoichiometric material balances together with different heat integration strategies within the subprocesses and carbon capture scenarios for the feedstock CO 2 . Particularly, one scenario allows for heat integration between all subprocesses in combination with CO 2 from point sources (CPS) assuming CO 2 is available for zero incremental energy costs. This scenario is consistent with our CO 2 feedstock assumptions, which consider purchasing already prepared CO 2 without extending the system boundaries to include the separation and preparation of CO 2 . Under this scenario, a process energy efficiency of 59-60% was estimated, which is slightly higher than the process energy efficiency estimated in this work. The difference is particularly a result of the different level of detail considered for the process simulation in both studies. Schemme et al. 40 and Burre et al. 41 also investigated different routes to produce OME 3-5 based on H 2 feedstock. However, in these studies, a process energy efficiency of 31-40% was estimated, which is signicantly lower than that estimated in this work, being closer to the results from the scenario reported Held et al., 43 in which heat integration is only considered within the subprocesses themselves rather than within the entire process chain. Hence, this highlights the impact and importance of heat integration on process energy efficiency with respect to the entire process route. Specically, the effect of using the surplus heat from the MeOH subprocess throughout the entire process heat integration has a positive impact on the process energy efficiency.

Economic evaluation CAPEX base case
An overview of the calculated CAPEX is given in Table 6. The obtained purchased equipment cost (EC) differs signicantly, ranging from 28 MioV in the case of P3 up to 49 MioV in the case of P2. Fig. 6 Results of the process efficiencies for the process routes P1 to P4. The contribution of the different process steps and the different types of equipment to the overall equipment cost is shown in Fig. 7. The main contributors are the MeOH and FA (aqueous/anhydrous) subprocesses in all routes. With respect to the equipment type, compressors and heat exchangers exhibit generally the highest share of equipment costs. The marginal contribution of the pump costs amounts to less than 1% for all routes.
The elevated EC of routes P2 and P4 result mainly from the FA (anhydrous) subprocess, while the EC of all the other process steps only differs slightly between the different routes. Remarkably, the additional OME 1 subprocess of routes P3 and P4 does not lead to a signicant increase in EC compared to routes P1 and P2 without an OME 1 step. The high EC of the FA (anhydrous) subprocess can be explained by the elevated reaction temperature (900 C) compared to the FA (aqueous) subprocess (650 C) and the additional separation and recycle of H 2 to the MeOH subprocess involving compression from 2 to 29 bar. This leads to larger heat exchanger areas and a higher demand of compression work in comparison to the FA (aqueous) subprocess, and therefore higher costs for heat exchangers and compressors. A detailed breakdown of the calculated EC is given in the ESI. † In total, the calculated FCI ranges from around 130 MioV in the case of P1 and P3 to 215 MioV for P2, resulting in an ACC between 11 MioV per annum and 19 MioV per annum. Table 7 gives an overview of the calculated OPEX. The main contributors to the overall operational costs are OPEX R&U , while labor costs and other direct and indirect OPEX only contribute to around 10% in total.

OPEX base case
A breakdown of OPEX R&U is shown in Fig. 8. As expected, the CO 2 and H 2 costs have the greatest impact on OPEX R&U , which range from 60 MioV per annum (P1/P3) to 68 MioV per annum (P4) for CO 2 and 89 MioV per annum (P2) to 113 MioV per annum (P1) for H 2 in the base case. The lower H 2 costs in P2 and P4 compared to P1 and P3 result from the FA (anhydrous) subprocess, which allows a higher H 2 efficiency given that less or no H 2 O is generated as a side product. However, the savings are diminished by the somewhat higher CO 2 demand in P2 and P4 due to the considerable amount of CO formed as by-product, which leads to higher purge rates in the synthesis of FA (anhydrous).
The residual OPEX R&U amounts to around 15 MioV per annum for all the routes. The main contributors are steam, compressed air, and electricity, as well as the silver catalyst in the case of P1 and P3. It should be noted that silver catalysts can be regenerated electrolytically with negligible material and activity loss. 60,84 This may lead to a reduction in the silver catalyst costs. However, due to its low overall impact, the consideration of silver regeneration was beyond the scope of the present work. Selling of excess heat in P1, P2 and P4, which cannot be integrated into the processes, only leads to minor revenue compared to the overall utility cost. A detailed breakdown of the calculated OPEX R&U is given in the ESI. †

NPC base case
A breakdown of the NPC is shown in Fig. 9. The lower CAPEX of P1 and P3 is compensated by the somewhat higher raw material cost. In general, raw material costs have a much higher impact on the NPC than CAPEX.
The calculated NPC amounts to 4.14 V 2018 per l DE for P1, 4.05 V 2018 per l DE for P2, 4.18 V 2018 per l DE for P3 and

Variation of CO 2 and H 2 price
The NPC was calculated for a wide range of CO 2 and H 2 costs to examine the inuence of the variability of future price developments. The results are shown in Fig. 10 and 11 for a variation in the CO 2 and H 2 cost, respectively, xing the other raw material costs as in the base case. This is extended in Fig. 11 (right), where the H 2 cost is varied against different CO 2 cost scenarios, covering a wide range of production locations worldwide and business cases. The high dependency of the NPC on the CO 2 and H 2 price is illustrated. In the case of a CO 2 cost variation, an NPC of between 3.0 V per l DE and 5.8 V per l DE is obtained, while a variation in the H 2 price leads to an NPC of 2.8 to 5.4 V per l DE .
The maximum discrepancy between the different routes is observed at the lower limit of the CO 2 cost range between P3 (3.3 V per l DE ) and P4 (3.0 V per l DE ) and at the higher limit of the H 2 cost range between P2/P4 (5.0 V per l DE ) and P3 (5.4 V per l DE ). However, these discrepancies with a maximum of 10% are still small considering the uncertainties and the rational assumptions made. Additionally, Fig. 11 (right) represents P4 for the variation in the CO 2 cost with the variable H 2 cost. Hence, the NPC ranges from 1.7 V per l DE and 6.7 V per l DE . In contrast, the dashed green line represents the highest NPC of P4 with variable H 2 cost and a DAC price of 720 V per t CO 2 , 70 where the price range for carbon capture with DAC technology is currently subject to great uncertainties, ranging from 100 to 800 V per t CO 2 . 85

Discussion of economic results
CAPEX, OPEX and NPC were calculated for the different production routes presented in this work. The main differences in the cost structure were observed for the different approaches for the synthesis of FA (aqueous/anhydrous). Firstly, the higher investment cost for heat exchangers in the FA (anhydrous) subprocess results from its high reaction temperatures. This cost could be diminished by milder process conditions which is a matter of further R&D efforts. Secondly, differences arise from the fact that in the FA (anhydrous) subprocess, H 2 is formed and recycled back to the MeOH subprocess. The recycling leads to somewhat higher investment costs for compressors, but higher H-efficiency, and consequently lower H 2 costs. These savings in OPEX, which have a much higher impact on the NPC, exceed the additional CAPEX signicantly. Nevertheless, no signicant differences in the overall NPC of the different routes were obtained over the whole considered parameter ranges. Consequently, no route is clearly preferable from an economic point of view. A preference can be obtained from weighing the different relations of CAPEX and OPEX resulting from the different FA subprocesses.
To the best of our knowledge, there is one peer-reviewed study in the literature that conducted a techno-economic assessment of OME 3-5 production from H 2 and CO 2 as feedstock. 40 Other publications deal with OME production from biomass 86,87 or MeOH 88,89 as feedstock, and thus are not directly comparable to this work.
In the study by Schemme et al., 40 the production routes of different oxygenated alternative fuels were studied analyzing the process efficiencies and production cost. The methodology for the economic evaluation and system boundaries are similar to the approach applied in this work, however the rigorosity and level of detail of the process simulation is extended in this work. Hence, a comparison between the economic results from Schemme et al. 40 and this work was carried out. To achieve a comparable data basis, the framework assumptions from Schemme et al. 40 were applied as much as possible to the model in this work. This was mainly the case for the base year, interest rate and LHV OME 3-5 as well as the CO 2 , H 2 , steam and electricity costs.
Primarily, no adaptions could be made for the plant capacity, CO 2 feed conditions, cooling water and steam conditions, compressor and pump efficiencies and pressure losses. A detailed comparison of the assumptions of Schemme et al. 40 and this work and the results of the comparative calculations are given in the ESI. † Applying the assumptions made by Schemme et al. 40 to the model from this work results in an NPC of 3.67 V per l DE for route P1 compared to 3.46 V per l DE obtained by Schemme et al. 40 ("Route A" in ref. 40), which corresponds to a deviation of 6%. Hence, an agreement of the NPC of route P1 can be stated in the error range of the calculations. Moreover, it can be concluded that the novel OME production routes P2 to P4 studied in this work are competitive to the TRI routes ("Route B" and "Route C") considered by Schemme et al. 40 Carbon footprint evaluation

Contribution analysis base case
In the base case year of 2018, the carbon footprints of process routes P1 and P3 are around 0.35 kg CO 2 -eq. per MJ, and thus signicantly higher than the carbon footprints of P2 and P4 with about 0.30 kg CO 2 -eq. per MJ (Fig. 12 and the ESI †). Therefore, we compared the hotspots in the OME 3-5 supply chain using P3 and P4 (Fig. 12) as examples in the proceeding contribution analysis. Fig. 12 Contribution analysis of the carbon footprint of the OME 3-5 routes P3 (left) and P4 (right) for year 2018. The left y-axis shows the carbon footprint in kg CO 2 -eq. per MJ, while the right y-axis additionally indicates results in kg CO 2 -eq. per l DE . The end-of-life of waste streams considers CO 2 in the exhaust gas and carbon carriers in the wastewater. Minor emissions due to wastewater treatment and the supply of cooling water, nitrogen, and compressed air are summarized as "other".
The carbon footprints of OME 3-5 production are mostly driven by H 2 and CO 2 supply for both routes P3 and P4, as shown in Fig. 12. Most emissions arise from H 2 supply, which is produced by electrolysis via today's electricity mix in the base case. Note that in P4, the H 2 supply contributes less to the carbon footprint than in P3 (Fig. 12) given that P4 requires 20% less H 2 per kg OME 3-5 than P3. The environmental credit, i.e., avoided burden, for utilized CO 2 strongly reduces the carbon footprint of both P3 and P4. However, in P4, the environmental credit for CO 2 utilization is higher than that of P3, given that P4 requires 13% more CO 2 per kg of produced OME 3-5 due to its lower carbon selectivity towards OME [3][4][5] in comparison to P3. Thus, the higher amount of utilized CO 2 in P4 is converted into more carbon carriers in waste streams, i.e., wastewater and exhaust gas (Fig. 12).
Although P4 is credited for its steam export, the heat supply in P4 contributes roughly twice as much to the carbon footprint as in P3 (Fig. 12). In contrast to P3, P4 additionally requires high-temperature heat above 250 C, which is supplied electricity-based via an electrode boiler.
Overall, the carbon footprint depends strongly on the carbon footprint of electricity supply if electricity-based H 2 and hightemperature heat are used. Therefore, we also investigated the inuence of the carbon footprint of electricity supply on the carbon footprint of the OME 3-5 product for all four process routes in a sensitivity analysis.

Sensitivity analysis for carbon footprint of electricity supply
In this sensitivity analysis, the carbon footprint of OME 3-5 is minimized in a supply chain optimization as a function of the carbon footprint of electricity supply, as shown in Fig. 13. For the supply chain optimization, conventional, fossil-based H 2 and high-temperature heat supply via steam methane reforming 90 and a natural gas boiler, 82 respectively, are included as alternatives to the PtX technology electrolysis and electrode boiler of the base case. The tipping point between a fossil-based and PtX supply chain is reached at a carbon footprint of electricity supply of 220 g CO 2 -eq. per kW h el . Above 220 g CO 2 -eq. per kW h el , the supply chain optimization results in fossil-based H 2 and hightemperature heat supply given that a PtX supply chain would result in a much larger carbon footprint (Fig. 13, grey dotted lines).
In contrast, H 2 and high-temperature heat are supplied by PtX technologies below 220 g CO 2 -eq. per kW h el . With an electricity carbon footprint below 56 to 68 g CO 2 -eq. per kW h el , the production of OME 3-5 is carbon negative. Please note that this negative carbon footprint only considers the system boundaries of OME 3-5 production, where the combustion of OME 3-5 releases CO 2 , making the entire life cycle of OME 3-5 carbon neutral in a complete WtW (cradle-to-grave) scope. When wind power is used for electricity supply, the carbon footprints of the OME [3][4][5] production are about À36 g CO 2 -eq. per MJ, which is in good agreement with the results of Hank et al. 10 for low-carbon electricity supply. Thus, with today's power grid mix, a fossil-based OME 3-5 supply chain would be more environmentally friendly in terms of the carbon footprint. Overall, the OME 3-5 routes P2 and P4 have the lowest carbon footprint independent of the carbon footprint of electricity supply.

Sensitivity analysis for carbon footprint of CO 2 and H 2 supply
The sensitivity of the OME 3-5 carbon footprint towards the supply of both raw materials CO 2 and H 2 was also investigated. In the case of the sensitivity analysis for CO 2 supply (Fig. 14), DAC (light grey) and MEA cement (dark grey) were analysed for the years 2018 and 2050, in accordance with the generic process data of the FfE. 81 Processes P2 and P4 yield lower carbon footprints than P1 and P3, irrespective of the CO 2 supply.
In case of the sensitivity analysis for H 2 supply (Fig. 15), generic process data for H 2 supply via electrolysis of the FfE is With the generic process data for 2018 (light grey), processes P2 and P4 yield the lowest carbon footprints. In contrast, when the generic process data for 2050 (dark grey) or SMR is considered, processes P1 and P3 have lower carbon footprints than P2 and P4. With this clean H 2 supply in 2050, the higher heat consumption of P2 and P4 becomes more decisive and leads to higher carbon footprints in both processes in this case.
In the ESI, † we additionally present the sensitivity analyses for the carbon footprint of high-temperature heat supply as well as the environmental credit, i.e., avoided burden, for steam exports. The additional sensitivity analyses show that process routes P2 and P4 yield lower carbon footprints compared to process routes P1 and P3 irrespective of the considered ranges of high-temperature heat supply and the environmental credit for steam exports.

Technology readiness level (TRL)
The TRLs assigned to the ve individual subprocesses and the entire process routes P1 to P4 are shown in Table 8.
The MeOH subprocess based on H 2 and CO 2 feedstocks exists on a large scale. 91 Hence, it has a TRL of 9. Identical assumptions were made by Schemme et al., 40 and Bardow et al. 92 also cites a high TRL for this technology.
On the one hand, the FA (aqueous) subprocess also exists on a large scale. 60 Hence, it has a TRL of 9. On the other hand, the FA (anhydrous) subprocess requires more research and demonstration steps to achieve the same TRL. In fact, the proof of concept was validated in several laboratory experiments. 42,62,93 However, besides these studies, to the best of our knowledge, long-term experiments and scale-up plants have not been implemented to date. Hence, a TRL of 3-4 was assigned to this subprocess.
The OME 1 subprocess was investigated by Drunsel et al., working on a laboratory-scale distillation column. 63,94 In addition, a plant that produces OME 1 on a large scale was commissioned following the process concept of Drunsel et al.
Hence, it has a TRL of 9. 95,96 Schemme et al. 40 also investigated the subprocess assuming a TRL of at least 5.
Different feedstocks are used for the OME 3-5 subprocess, resulting in the need for adjustments, particularly for the reactor, distillation columns and the H 2 O separating process unit. However, previous studies have been conducted with regard to the reaction, distillation and H 2 O separation for specic mixtures, some of which are very similar to the mixtures assessed in this work, and a demonstration plant 97 was built and commissioned in the scope of the NAMOSYN project. Therefore, the TRL was assumed to be 3-4. If the demonstration plan shows promising results during long-term test runs regarding the performance of the critical process components, the TRL can be assigned higher values.
The TRL for P1-P4 is presented in two ways. The rst number indicates the lowest TRL of the considered subprocesses, and therefore is equal to 3 for all four process routes. Given that this only shows the main hurdle of the entire process chain, but does not show the advantages of the individual subprocesses, the second number presents the mean TRL of the subprocesses. This is much higher for all four process routes, where a difference between P1 and P3 with a TRL 7 and 8 is observed in comparison to P2 and P4 with a TRL 5 and 6. This is a result of the low TRL of the FA (anhydrous) subprocess in comparison to the FA (aqueous) subprocess. Due to the benecial results of P2 and P4 regarding process efficiency and carbon footprint, further investigations to improve the TRL of FA (anhydrous) should be carried out.
An overview of the process evaluation criteria for the considered processes in web diagrams for the years 2018, 2030 and 2050 is available in ESI Fig. S12-S14. †

Conclusions
Based on a standardized and validated modelling and simulation methodology implemented in Aspen Plus®, four different process routes for the production of OME 3-5 were evaluated considering techno-economic assessment and carbon footprint. The evaluated processes are based on scalable technologies, which have the potential to achieve a feasible large-scale OME Fig. 15 Carbon footprint of all four OME 3-5 process routes as a function of the carbon footprint of H 2 supply. The left y-axis shows the carbon footprint in kg CO 2 -eq. per MJ, while the right y-axis additionally indicates the results in kg CO 2 -eq. per l DE . The shaded areas indicate the carbon footprint ranges of H 2 supply for the years 2018 (light grey) and 2050 (dark grey). 68 SMR: steam methane reforming. . No signicant differences in the overall NPC of the different routes were obtained over the whole considered parameter ranges. This is due to the trade-off between the CAPEX and OPEX, which occurred due to the lower cost of the feedstock for the anhydrous FA-based processes P2 and P4, while having a higher CAPEX mainly due to the high temperature endothermic methanol dissociation reaction at low MeOH concentrations. The high dependency of the NPC on the CO 2 and H 2 costs was illustrated using a sensitivity analysis. In the case of CO 2 cost variation (at H 2 base case cost), the NPC of OME between 3.0 V per l DE and 5.8 V per l DE was obtained, while a variation in the H 2 price (at CO 2 base case cost) led to an NPC 2.8 to 5.4 V per l DE . The combination of OME production based on the CO 2 point source feedstock cost and H 2 cost provisioned future production using cheap renewable electricity (at H 2 price of #2438 V per t from AEL in 2050 and CO 2 price of #62 V per t from MEA, cement) can signicantly reduce the costs for sustainable OME production at #2.33 V per l DE . In the case of CO 2 feedstock available onsite at no price, the OME production cost under 2 V per l DE can be achieved with green H 2 produced under favored conditions at #2000 V per t H 2 . The carbon footprints of OME 3-5 production are mostly driven by H 2 and CO 2 supply for both anhydrous (P4)-and aqueous FA (P3)-based routes. In the base case in the year 2018, the carbon footprint of process routes P1 and P3 is around 0.35 kg CO 2 -eq. per MJ, and thus signicantly higher than the carbon footprints of P2 and P4 with about 0.30 kg CO 2 -eq. per MJ. This is a result of the lower H 2 demand, including the anhydrous FA production. The environmental credit, i.e., avoided burden, for CO 2 supply strongly reduces the carbon footprint of all process routes, especially P4 given that P4 requires 20% less H 2 per kg OME 3-5 than P3. However, in P4, the environmental credit for CO 2 utilization is higher than that of P3, given that P4 requires 13% more CO 2 per kg of produced OME 3-5 due to the lower C-selectivity towards OME 3-5 in comparison to P3. For the carbon footprint of the electricity supply of 220 g CO 2 -eq. per kW h el , the PtX supply chain has a smaller carbon footprint than a fossilbased supply. If wind power is used for the electricity supply, the carbon footprint of the OME 3-5 production is about À36 g CO 2 -eq. per MJ, which is in good agreement with the literature results 10 for low-carbon future electricity supply. However, the current German power grid mix 2018 is not suitable to improve the carbon footprint of the OME 3-5 production, and thus high penetration of renewable energy generators is indeed for sustainable production.
Overall, routes P2 and P4 have the lowest carbon footprint independent of the carbon footprint of the electricity supply. Based on the sensitivity analysis with a future scenario for the German electricity mix, the carbon footprint of promising OME production routes such as P4 can reach 0.29 kg CO 2 -eq. per l DE , showing the opportunity for the sustainable production of OME. The TRL of the considered processes was evaluated in a conservative and optimistic manner at 3-7 based on experimental developments for critical process components of the OME value chain.
Considering the previous results, generally, there is no signicant economic difference for the analysed OME production routes considered in this work. However, the overall energy efficiency of the considered processes shows a positive tendency for the anhydrous FA-based routes, which are still under research and development. An important lever to enhance the energy efficiency of the processes is to reduce the energy losses by valorising low temperature excess heat through usage in external processes. A big share of the excess heat (>43%) is still at useful temperature levels considering the usage of heat pump technology, an approach that is under investigation. This strategy is crucial in the context of PtX processes, where the production will probably take place where cheap renewable electricity is abundant. This will reduce the supply of external utility streams and besides the overall process efficiency enhacement, this can reect positively on the NPC.
Given that the main driver of the production costs is the feedstock (H 2 and CO 2 ) with almost 74% of the NPC, the measures to create an economical frame for sustainable OME as neat fuel or a blend is signicantly dependent on the H 2 production in areas with low levelized costs of electricity and with low costs for CO 2 . The carbon footprint evaluation shows the potential of the environmental credit of OME production processes, which can reect lower production costs when a WtW system boundary is considered. Monetary frames such as H 2 Global ‡ from the federal German government and modications in REDII § on the European level to include ‡ H 2 Global is a funding concept by the federal German Government for short-term economic market entry into international PtX projects and achieving the goals adopted in the German National Hydrogen Strategy in connection with the production of green H 2 and its import. 91 § Renewable Energy Directive -Recast to 2030 (REDII) 2018/2001/EU. 92 In REDII, the overall EU target for renewable energy sources consumption by 2030 has been raised to 32%. Member states must ensure that a minimum of 14% of the nal energy consumption in the transport sector is provided by fuel suppliers from renewable sources. awarding systems and enhance the market introduction for sustainable fuels that lead to CO 2 avoidance are one of the main levers for the realization of the large-scale production of these fuels. The techno-economic and carbon footprint potentials of OME large-scale production processes in Germany and worldwide based on sensitivity analysis presented in this work emphasize the potential of OME as important constituents in a sustainable future mobility sector.

Author contributions
Paper conception, the process modelling, simulation, the subsequent technical evaluation and workow coordination was done by the Fraunhofer Institute for Solar Energy Systems ISE. KIT supported this work by evaluating the technological concepts and comparing it to the published state of the art. The cost estimation and subsequent economic evaluation was done by the DLR. The carbon footprint estimation and subsequent evaluation was done by the LTT of the RWTH Aachen University. TUM contributed to the conception, data analysis, critical revision for important content.

Conflicts of interest
The authors declare no competing nancial interest.