Fully lignocellulose-based PET analogues for the circular economy

Polyethylene terephthalate is one of the most abundantly used polymers, but also a significant pollutant in oceans. Due to growing environmental concerns, polyethylene terephthalate alternatives are highly sought after. Here we present readily recyclable polyethylene terephthalate analogues, made entirely from woody biomass. Central to the concept is a two-step noble metal free catalytic sequence (Cu20-PMO catalyzed reductive catalytic fractionation and Raney Ni mediated catalytic funneling) that allows for obtaining a single aliphatic diol 4-(3-hydroxypropyl) cyclohexan-1-ol in high isolated yield (11.7 wt% on lignin basis), as well as other product streams that are converted to fuels, achieving a total carbon yield of 29.5%. The diol 4-(3-hydroxypropyl) cyclohexan-1-ol is co-polymerized with methyl esters of terephthalic acid and furan dicarboxylic acid, both of which can be derived from the cellulose residues, to obtain polyesters with competitive Mw and thermal properties (Tg of 70–90 °C). The polymers show excellent chemical recyclability in methanol and are thus promising candidates for the circular economy.

was accomplished at 40 °C using an SPD-M40 photoarray detector in series. The molecular weight determination were performed using polystyrene standards of known molecular weight distribution.
Differential Scanning Calorimetry (DSC) was conducted at the Graz University of Technology on a Perkin Elmer DSC 8500. In a typical procedure, the sample (5-10 mg) was weighed into a DSC aluminium pan and then capped with a lid. The sample was sealed and heated from 25 to 250 °C with a heating rate of 10 °C•min -1 . Then, it was cooled to 25 °C with a heating rate of 10 °C•min -1 . Subsequently, a second heating scan to 250 °C with the same heating rate was performed. All of the experiments were performed under N2 flow with a flow rate of 20 mL•min -1 .
Thermogravimetric analysis (TGA) was performed at the Graz University of Technology on a Netzsch Jupiter STA 449C thermogravimetric analyzer. Typically, the sample (1-3 mg) was weighed into a platinum pan. The sample was heated from 20 to 550 °C with a heating rate of 10 °C•min -1 under N2 flow with a flow rate of 20 mL•min -1 . The temperatures were recorded when 5 % weight loss (T5%) and 90% weight loss rate (T90 %) occurred.
Inductively coupled plasma mass spectrometry (ICP-MS) was performed at University of Graz on a Agilent 7900 ICP-MS. Typically, the samples were solubilized with 5 mL HNO3 in the MLS ultraclave and then heated to 250 °C for 30 mins before analysis by ICP-MS.

Preparation of Cu20-PMO catalyst
The Cu20-PMO catalyst was prepared according to our previously reported procedure. [1] In a typical procedure, a solution containing AlCl3·6H2O (12.07 g, 0.05 mol), Cu(NO3)2·2.5H2O (6.98 g, 0.03 mol) and MgCl2·6H2O (24.4 g, 0.12 mol) in deionized water (200 mL) was dropwise added to a solution containing Na2CO3 (5.30 g, 0.05 mol) in water (300 mL) at 60 °C under vigorous stirring. The pH value was always kept between 9 and 10 by addition of small portions of a 1 M solution of NaOH. The mixture was vigorously stirred at 60 o C for 72 h. After cooling to room temperature, the light blue solid was filtered and resuspended in a 2 M solution of Na2CO3 (300 mL) and stirred overnight at 40 °C. The catalyst precursor was filtered and washed with deionized water until chloride free. After drying the solid for 6 h at 100 °C followed by the calcination at 460 °C for 24 h in air, 9.5 g of Cu20-PMO was obtained.

Preparation of model compounds 1G, 1S, PC, PCcis, PCtrans and DMFD
Preparation of model compound 1G: 1G was synthesized according to previously developed procedure with slight modifications. [2,3] Typically, a 100 mL high pressure Parr autoclave was charged with 5 % Pd/C (0.1 g), 40 mL methanol and trans-ferulic acid (5 g, 25.74 mmol). The reactor was sealed, purged 3 times with H2 and then pressurized with H2 (40 bar) and stirred at room temperature overnight. After filtering through a Celite plug, the solvent was evaporated under reduced pressure to provide 5.02 g of 3-(4-hydroxy-3-methoxyphenyl)propanoic acid (25.61 mmol, 99.5 % yield), which was used without further purification. To a rapidly stirred suspension of LiAlH4 (1.45 g, 38.26 mmol) in 50 mL of THF was dropwise added a solution of 3-(4-hydroxy-3-methoxyphenyl)propanoic acid (5 g, 25.51 mmol) in 20 mL of THF at 0 °C. After the addition was completed, the reaction mixture was allowed to warm to RT and stirred overnight. The reaction mixture was quenched by ice water, neutralized by 1M HCl (10 mL) and then filtrated to remove the white solid. The remaining solution was extracted with EtOAc (3×50 mL). The combined organic extract was washed with saturated solution of NaHCO3 (50 mL) and brine (100 mL), dried over anhydrous MgSO4 and the solvent was evaporated under reduced pressure to provide 3.2 g of dihydroconiferyl alcohol (1G) (17.58 mmol, 68.9% yield).
Preparation of model compound 1S: 1S was synthesized according to previously developed procedure with slight modifications. [2,3] Typically, a 100 mL high pressure Parr autoclave was charged with 5 % Pd/C (0.1 g), 40 mL methanol, 3-(4-hydroxy-3,5-dimethoxyphenyl) acrylic acid (22.32 mmol, 5 g). The reactor was sealed, purged 3 times with H2 and then pressurized with H2 (40 bar) and stirred at RT overnight. After filtering through a Celite plug, the solvent was evaporated under reduced pressure to After the addition was completed, the reaction mixture was allowed to warm to RT and stirred overnight.
The reaction mixture was quenched by ice water, neutralized by 1M HCl and then filtrated to remove the white solid. The remaining solution was extracted by EtOAc (3×50 mL). The combined organic extract was washed with a saturated solution of NaHCO3 (50 mL), brine (50 mL), dried over anhydrous MgSO4; the solvent was evaporated under reduced pressure to provide 3.42 g of dihydrosinapyl alcohol (1S) (16.13 mmol, 63.2% yield).
Preparation of model compounds PCcis and PCtrans:The synthesis of PCcis and PCtrans was performed in three steps according to a reported procedure with slight modifications. [4] Step 1: Synthesis of para-coumaric acid ethyl ester. 5 g of para-coumaric acid (30.47  Step 2: Synthesis of 3-(4-hydroxycyclohexyl)propionic acid ethyl ester (cis and trans isomers). The autoclave was charged with 2 g Raney Ni catalyst, 5 g para-coumaric acid ethyl ester (26.04 mmol), and 20 mL isopropanol. The reactor was sealed and pressurized with H2 (40 bar) at RT. The reactor was heated to 160 °C and stirred at 400 rpm overnight. After completion of the reaction, the reactor was cooled to RT. The products 3-(4-hydroxycyclohexyl)propionic acid ethyl ester containing cis-isomer and trans-isomer with the ratio of 1:1 were isolated by silica gel column chromatography (gradient elution: Pentane : EtOAc 90:10 to 40:10) in 93 % yield (24.24 mmol, 4.8 g, cis:trans 1:1).
Step 3: Synthesis of 4-(3-hydroxypropyl)cyclohexanol (PC) diol (cis and trans isomers). For synthesis of PC (cis and trans), the same procedure as specified below was used. To a rapidly stirring suspension containing 30 mL THF and 0.4 g LiAlH4 as reducing agent, a solution of 3-(4-hydroxy-3-methoxyphenyl)propanoic acid (cis or trans, 7.57 mmol, 1.5 g) in 20 mL THF was dropwise added in a threeneck flask that was cooled in an ice-bath. After the addition was completed, the reaction mixture was allowed to warm to RT and stirred overnight. After completion of the reaction, the mixture was poured onto ice water to reduce the excess of LiAlH4, and 20 mL solution containing 1M HCl (10 mL ii. iv. For calculation of yield to hydrocarbon alkanes The quantification of hydrocarbon alkanes was performed using the response of the flame-ionization detector (FID) and the response factor was estimated by Effective Carbon Number (ECN) method. [6] F (R − wt) = Mw of products × ECN of dodecane Mw of dodecane × ECN of product Eq 7

ECN of dodecane = 12 (Carbon number)
ECN of products = Carbon number of hydrocarbons Eq 8 F (R − wt) = Peak area counts for dodecane × wt of products Peak area counts for products × wt of dodecane Eq 9

General experimental procedures
Reductive catalytic fractionation of lignocellulosic biomass (General procedure A): The mild depolymerization of pine, beech and poplar lignocellulose was carried out in a high-pressure Parr autoclave equipped with an overhead stirrer. Typically, the autoclave was charged with 0.4 g of Cu20-PMO catalyst, 2 g of lignocellulose (beech, pine or poplar) and methanol (20 mL) as a solvent. The reactor was sealed and pressurized with H2 (40 bar) at room temperature. The reactor was heated to 180 °C and stirred at 400 rpm for 18 h. After completion of the reaction, the reactor was cooled to room temperature. Then 0.1 mL solution was collected through a syringe and injected to GC-MS or GC-FID after filtration through a PTFE filter (0.45 µm). The solid was separated from the solution by centrifugation and subsequent decantation and additionally washed with methanol (3×20 mL). The methanol washings were combined in a round bottom flask and the solvent was removed in vacuo. The crude product was dried in a desiccator in vacuo overnight and was further used as specified below.
Fractionation procedure: To the obtained crude mixture, EtOAc (20 mL) was added and it was stirred overnight at room temperature, which resulted in precipitation of brownish colored solid. The suspension was then transferred into a 20 mL centrifuge tube. The solid was separated by centrifugation and decantation and additionally washed with EtOAc (2×20 mL) and dried in vacuo until constant weight . The EtOAc washings were combined in a separating funnel and were washed with small amount of saturated NaHCO3 (1×10 mL) and brine (2×10 mL) and the organic phase was dried over anhydrous MgSO4. After filtration, the solvent was transferred in a round bottom flask and the solvent was removed in vacuo for further use as specified below to give yellow brown crude product.
Demethoxylation/hydrogenation of model compound 1G to 4-n-propanolcyclohexanol (PC) (General procedure B): The demethoxylation/hydrogenation of 1G was carried out in 100 mL high-pressure Parr autoclave equipped with an overhead stirrer. Typically, the autoclave was charged with 1 g Raney Ni catalyst, 0.2 g (1.1 mmol) 1G, 15 mL isopropanol, 20 mg dodecane as internal standard. The reactor was sealed and pressurized with H2 (10 bar) at room temperature. The reactor was heated to 120 °C and stirred at 400 rpm for 2 h. After completion of the reaction, the reactor was cooled to room temperature.  Weight determination of cellulose for the production of poly (PC/TPA): The following calculations are relying on experimental data: RCF of 2 g beech lignocellulose (comprising 39.2 wt % cellulose) proceeds with cellulose retention up to 95.4 % in the investigated reaction conditions. Thus, the amount of usable cellulose is 747.9 mg (747.9 = 784 × 0.954 mg).
Cellulose is subjected to enzymatic hydrolysis to glucose in the yield of 79.6 % over cellulase at 50 °C for 72 h, following by isomerization and dehydration to 5-HMF (59.2 % yield) over AlCl3 catalyst at 120 °C for 4 h.
The following calculations are relying on selected literature routes/examples: The obtained 5-HMF was then subjected to hydrogenation/hydrogenolysis to DMF (99 % yield) over Pd/C catalyst [7] . DMF is further converted to p-xylene (90 %) by Diels-Alder cycloaddition with ethylene followed by dehydration over a H-Beta zeolite catalyst [8] . TPA can be obtained commercially by oxidation of pxylene (93 %) over cobalt-manganese-bromide catalyst (Amoco process) [9] , followed by methyl ester formation in a yield of 99 % (DMTA) [10] . Therefore, the available amount of cellulose definitely covers the need for the diacid building block from the same wood. The surplus of cellulose is 608.9 mg (608.9 mg = 747.9 -139 mg).
Calculating theoretical maximum PC yield and cellulose needed: RCF and 100 % yield for catalytic funneling, as well as 100% yield of polymer synthesis would provide the following PC amount: The theoretical maximum aromatic monomers yield (~ 34 % (0.34 = 0.58 × 0.58)) is determined by the β-O-4 content (~ 58 %) in beech wood [11] . Because S/G ratio in beech wood is 2.8 [1] , the theorectical For this, 113.5 mg DMTA and consequently 245 mg cellulose are needed, followed by the calculation shown above.
Thus, the surplus of cellulose in this case is (747.9-245 = 502.9 mg).
Option 1: determination of ethanol weight from excess cellulose.
Surplus of 608.9 mg cellulose, as calculated above can be transformed into bio-ethanol. One mol of glucose, yields two moles of ethanol. For a process with 74.1 efficiency (from cellulose to fructose) [12] and 95 % efficiency (from fructose to ethanol) [13] , the ethanol mass would be: Ethanol (mg) = 608.9 162.1 × 0.741 × 0.95 × 2 × 46 mg = 243.3 mg Eq12 Option 2: determination of ethylene glycol weight from excess cellulose.
Recently cellulose has been efficiently converted to ethylene glycol (EG) with 75 wt % yield using Ni-W2C/MC catalyst in water [14] . Thus following this procedure, complete conversion of the surplus cellulose (608.9 mg) would lead to 456.7 mg EG (456.7 mg = 608.9 × 0.75 mg).
Weight determination of hemicellulose for the production of other valuable products: RCF of 2 g beech lignocellulose (comprising 19.1 wt % cellulose) proceeds with hemi(cellulose) retention up to 86.5 % in the investigated reaction conditions. RCF of 2 g beech lignocellulose (comprising 19.1 wt % hemicellulose) proceeds with hemi(cellulose) retention up to 86.5 % in the investigated reaction conditions. The amount of hemicellulose after RCF of 2 g beech lignocellulose is: Hemicellulose has been demonstrated as interesting starting material towards the production of value added products (furfural and ethylene glycol). For fufural, with a yield of 87.8 %, as reported using Al2(SO4)3 catalyst in GVL/water [15]  reported from previous work [16] , thus ethylene glycol can be quantified (95.   Pressure Conversion [b] (%)  Table 3. Catalytic demethoxylation and hydrogenation of 1G to PC over various solvents [a] .

Proposed reaction network for defunctionalization of 1G
Supplementary Figure 27. Proposed reaction network for the catalytic demethoxylation and hydrogenation of 1G over Raney Ni catalyst using isopropanol as solvent.

Kinetic modeling of 1G defunctionalization and hydrogenation
To simplify the calculation, we assume a pseudo-homogeneous power-law model to obtain the apparent kinetic constants. Also, we assume a very high and constant hydrogen concentration in the liquid phase and on the catalyst surface [H] >> [any other compounds]. For the proposed reaction network for 1G transformation to diol PC, the system of ordinary deferential equations will be as following: The Parameters were estimated by least-square fit of the experimental data using the classic Levenberg-Marquardt algorithm implemented in DynaFit software. [17] 2.5 Fractionation, catalytic funneling and analysis of the crude mixture obtained from

RCF of beech lignocellulose
The reductive catalytic fractionation (RCF) of beech lignocellulose over Cu20-PMO catalyst was carried   were performed according to General procedure C using equal molar ratio of PC and cellulose-derived DMTA over Zn(OAc)2 catalyst [18] . Briefly, a 100 mL three-neck flask, equipped with a magnetic stirrer and reflux condenser, was charged with 0.    were distinguishable. According to literature data, the high field signals of C8 at 165.24 ppm (C8 cis) and 165.42 ppm (C8 trans) were assigned to H-H type [19] , the low field C8 signals (cis and trans overlapped) at 166.05 ppm to T-T type structure [20] , and the signals in-between (165. 28

Detailed structural analysis of poly (PC/FDCA)
To understand poly (PC/FDCA) dyad structure, similar experiments and characterizations were carried out as those described in the case of poly (PC/FDCA). According to literature data, the C8 in dicyclohexyl and dipropyl furandicarboxylate display a chemical shift at 157.6 ppm [21] and 158.2 ppm [22] .
Therefore, we proposed that the high field signals at 157. 55  In all cases, nearly no OH absorption band at around 3500 cm -1 is observable, which indicated successful copolymerization of PC with DMTA and DMFD by the formation of ester bonds. According to the literature reported FTIR data for poly (EG/TPA) [23] and poly (EG/FDCA) [24] , a clear signals assignments

A comprehensive biorefinery strategy for the production of gasoline, PET analogues and jet fuels from beech wood (10 g)
Supplementary Figure 75. A comprehensive and proposed catalytic protocol for complete utilization of beech wood.
Step 1: RCF of beech wood gave crude aromatic bio-oil over Cu20-PMO catalyst; Step 2: Catalytic funneling of EtOAc extracts gave crude aliphatic bio-oil. The crude aliphatic bio-oil was purified by distillation under 1mpa at (100 -120 °C) to deliver three Fractions A , B and C. HDO of Fraction A and Fraction C gave C7-C9 and C14-C17 cyclic alkanes, respectively.The hydrocarbons were quantified using the response of the flameionization detector (FID) and the response factors was estimated by Effective Carbon Number method (ECN); Step 3: Copolymerization of Fraction B with cellulose-derived DMFD to yield PET analogue poly (PC/1/FDCA).
Step 1: A large scale reductive catalytic fractionation (RCF) setup using beech wood was carried out according to General procedure A over our previously developed Cu20-PMO catalyst under the following reaction conditions: 10 g beech wood, 2 g catalyst, 180 °C, 120 mL methanol, 40  Step 2: The catalytic funneling of EtOAc extracts was carried out according to General procedure B in 100 mL high pressure Parr autoclave with an overhead stirrer. Typically, the autoclave was charged with 2 g Raney Ni catalyst, 1120 mg of EtOAc extracts, 20 mL isopropanol. The reactor was sealed and pressurized with H2 (30 bar). The reactor was heated and stirred at 150 °C for 10 h. After reaction, the reactor was cooled to room temperature and the solvent was removed in vacuo to deliver crude aliphatic bio-oil (PC, 229.5 mg, 1, 84.5 mg, 2, 37.9 mg, 3, 50.9 mg). Crude aliphatic bio-oil was subjected to distillation at temperature range between 100-120 °C using a Kugelrohr apparatus under observed after RCF of hardwood in literature [25,26] .

Depolymerization:
The methanolysis depolymerization [27,28] reaction was carried out according to the   In this work, we have first developed a straightforward protocol using beech wood sawdust with a size of more than 1 mm, taking advantage of sieve fractionation to liberate the spent Cu20-PMO catalyst from the carbohydrate pulp after RCF (Supplementary Figure 98). Typically, after RCF of beech wood, the spent Cu20-PMO catalyst was separated from the carbohydrate pulp through a mesh screening. Then, the solids were subjected to further ultrasonic treatment in water to get rid of the catalyst residues. After applying this method, the isolated reaction solids (mainly cellulose) were subjected to ICP analysis which showed minimal Cu contamination (1.45 mg Cu/ g carbohydrate) that confirmed the removal of Cu20-PMO catalyst. Next, the isolated carbohydrate fraction was subjected to three independent reaction steps (

(B)
Step 1: Mild enzymatic hydrolysis of solid residue The mild enzymatic hydrolysis of solid residue (0.2 g) (104 mg cellulose) was performed using cellulase (Cellic CTec2) (3 FPU) at sodium acetate buffer (10 mL, 5 mM, pH 4.8) in a glass vial equipped with a magnetic stirrer. The mixture was heated to 50 °C under 150 rpm for 72 h. After reaction, the solid residue was separated by filtration, the filtrate was collected and then analyzed by HPLC. Finally, 92 mg D-glucose was obtained in a mol yield of 79.6 %.
Step 2: Catalytic conversion of glucose to 5-HMF The obtained filtrate was first evaporated and then 10 mL DMSO and AlCl3 (0.02 g) was added. The reaction was proceeded in a round bottom flask at 120 °C for 4h. After reaction, AlCl3 catalyst was removed by filtration and the filtrate was subjected to HPLC analysis, confirming 43 mg HMF was achieved in a mol yield of 59.2 %. HPLC equipped with C18 column 3.5 μm, 4.6 x 100 mm and UV detector was used for analysis and it was done in a mixture of methanol in water (5/95, v/v) as a solvent gradient at elution flow rate of 1 mL/min. The column temperature was 303 K and the injection volume of the sample (filtered through 0.22 μm membrane) was 20 mL.
The reactor was purged with O2 thrice and pressurized with 2 bar O2. The reactor was then heated to 120 o C in heating block, with stirring speed of 500 rpm. After 10 h, the reaction was cooled down to room temperature and the white solid FDCA and Ru/C were separated by filtration. Then, the FDCA was resolublized in methanol (10 mL) under sonication. After filtration and wash twice with methanol, the white solid was obtained after removing the solvent under reduced pressure. Isolated yield is 72 %.

Supplementary Note 8. Preliminary techno-economic analysis (TEA)
On the basis of the experimental data, we performed the techno-economic assessment (TEA) of the process. The comprehensive evaluation includes the catalytic processing of beech lignocellulose by RCF, followed by the fractionation of the obtained bio-oil and the catalytic processing of the respective fractions to final products. This includes the Raney Ni mediated catalytic funneling of the monomers to PC diol, which is then converted to the respective fully bio-based polyesters; as well as the separation and conversion of the carbohydrate rich residues to FDCA, as well as furfural (based on best literature).
Due to the lab-scale level of the current process development the assessment was based on the experimental inputs and outputs. The required raw material inputs and catalysts were normalized to one dry ton of beech wood processed. Specific cost unit values were taken from other papers and industrial sources. Solvent recovery and catalyst recycling were estimated based on other papers. Basic assumptions have been made in line with the literatures [33,34] . Thus, we estimated fixed operating costs, utility costs and annualized capital cost as a relative share based on raw material costs [35] . Supplementary Table   12 provides an overview on the most important raw material costs. Beech wood as the main feedstock of the process is also the costliest raw material in process. Other catalysts -0.04 -[a]. Due to a lack of commercial availability, the cost of the Cu20-PMO catalyst was estimated based on the raw material costs. All catalysts were assumed to be sufficiently recyclable or maintainable (50% of the cost to 70% of the performance). Overall the process (shown in Supplementary Figure 99) converts beech wood into 1% of gasoline, 1% of jet fuel, 4% of PET and 4 % methanol, 11% of furfural and 10 % of FDCA on a mass basis (this being good efficiency with 80% of lignocellulose converted and deoxygenation taken place).

Supplementary
It is very encouraging, that with the currently achieved product yields, and with the assumptions made in line with literature data, the techno-economic evaluation shows a positive balance. More specifically, a 6.4 % rate of return can be achieved at 99 % solvent recovery. Returns are sensitive to methanol and isopropanol recovery at 96 % and 98 % respectively. Overall, our analysis indicates that catalyst and solvent costs are the main drivers of operating costs, which is not surprising considering the lab-scale development stage of the process.
FDCA and furfural form the most important revenue streams (see Supplementary Table 13) while fuels are neglectable in both volume and value. Hence, the profitability of the process is particularly depending on future FDCA price assumptions. Since no mature FDCA market is yet existing. Thus the revenue values estimated in previous papers have been used [45,46,47] .
It is clear that up-scaling would include a significant reduction of the solvent demand and consumption while optimizing its recovery. Another factor that future optimization may improve, is the total process yield, which is currently at 30 wt %. While this yield is already high considering the well-defined product streams obtained, we still see possibilities for improvement. For example, while the 12 wt % yield of PC is among the best in available literature for a lignin-based polymer building block, this value can be improved by optimizing the catalyst type and flow/vs batch operation of the RCF to maximize 1G/1S yield. For example, one of the highest yields to 1G/1S mixture in the literature is 44.8 wt % [49] , compared to 24.2 wt % in this work. Consequently, the PC yield (22.2 wt %) could achieve about double the amount (12 wt %) currently observed. In addition, the yield value to FDCA can also be dramatically optimized by selecting the best catalytic system. For example, starting from the cellulose, a combined yield of 82 % to FDCA can be obtained based on two-step metal catalytic sequence [RuCl3 catalyzed hydrolysis, dehydration and isomerization [50] and Pd/HT mediated catalytic oxidation [51] The amount of FDCA produced by this methodology can not only cover the required amount for the production of DMFD in a best yield of 99 % [52] , but also provides the huge surplus FDCA. In fact, such an increase in yields would enable a profitable (7% return) operation of the process even under the assumption of the lowest possible FDCA prices discussed in literature [43] .
Another important aspect is to carefully assess other benefits of bio-based products compared to fossilbased ones, especially in relation to carbon-neutrality and climate benefits. Our process is utilizing a relatively cheap raw material (Beech wood) and targets well defined and already existing products.
However, current prices of the substituted fossil-based products are too low considering they are made from rather cheap bulk petrochemicals. However, when assuming emission pricing in the range of 50-100 Euros per ton CO2 released would add between 2 and 4 % to the overall profitability.
Supplementary Figure 100. Rate of return as result of 60 different scenarios when varying raw material costs, chemical costs, recovery rates, CO2 prices, product prices, catalyst costs, always considering two different yield levels and with and without CO2-taxation..
It has to be noted, that we performed the process economics analysis based on currently obtainable experimental data and available product price data cited from the literature and commercial webpages. At the current stage of development this assessment is considered as a prospective, order of magnitude estimate, typical for TRL 3 to 4 [53] . The calculations were performed in Excel, and this approach has been proven successful in recent literature [54][55][56] . The Supplementary Figure 100 above shows the results of 60 different scenarios when varying raw material costs, chemical costs, recovery rates, CO2 prices, product prices, catalyst costs, always considering two different yield levels and with and without CO2-taxation. The results turn negative in case of lower recovery and lower furfural prices. If these scenarios can be avoided we can see that most runs (scenarios) provide rates of return between 5 and 13 %.