Evaluating the Environmental Sustainability of Alternative Ways to Produce Benzene, Toluene, and Xylene

The petrochemical industry can reduce its environmental impacts by moving from fossil resources to alternative carbon feedstocks. Biomass and plastic waste-based production pathways have recently been developed for benzene, toluene, and xylene (BTX). This study evaluates the environmental impacts of these novel BTX pathways at a commercial and future (2050) scale, combining traditional life cycle assessment with absolute environmental sustainability assessment using the planetary boundary concept. We show that plastic waste-based BTX has lower environmental impacts than fossil BTX, including a 12% decrease in greenhouse gas (GHG) emissions. Biomass-based BTX shows greater GHG emission reductions (42%), but it causes increased freshwater consumption and eutrophication. Toward 2050, GHG emission reductions become 75 and 107% for plastic waste and biobased production, respectively, compared to current fossil-BTX production. When comparing alternative uses of plastic waste, BTX production has larger climate benefits than waste incineration with energy recovery with a GHG benefit of 1.1 kg CO2-equiv/kg plastic waste. For biomass (glycerol)-based BTX production, other uses of glycerol are favorable over BTX production. While alternative BTX production pathways can decrease environmental impacts, they still transgress multiple planetary boundaries. Further impact reduction efforts are thus required, such as using other types of (waste) biomass, increasing carbon recycling, and abatement of end-of-life emissions.


S1.1 BioBTX
The BTX production pathways (Figure 1 in the main text) from both mixed plastic waste and biomass is based on the Integrated Cascading Catalytic Pyrolysis process developed by BioBTX, a pilot-company located in the Netherlands.Firstly, the feedstock is the thermally cracked; secondly, the pyrolysis vapours are catalytically converted to BTX; thirdly, the BTX is separated and collected.BioBTX' primary data is based on a capacity of 48 kt feedstock/year and includes the core technology and downstream steps 1 .

S1.2 Carbon accounting
The system boundary was set to cradle-to-grave with an incineration end-of-life scenario based on the chemical structure of BTX.The fossil carbon content can therefore lead to an additional 3.36 kg CO 2 eq./kgBTX (Table S.1).In the case of bio-BTX, the additional CO 2 emissions were considered neutral.This was justified following the biogenic global warming potential (GWP bio ) approach by Cherubini et al. 3 and allocating temporary carbon storage in bio-products based on Guest et al., 5 , where biogenic carbon can be considered neutral, because both the crop rotation period and the storage period in the technosphere are short.This method accounts for the fate of the carbon embedded in the end products.

S1.4 Distillation
An additional distillation step is applied to the crude BTX output, based on a simplified distillation step.
Calculations are based on Piccinno et al., (2016), sum over each of the three chemicals is taken: where C p is the specific heat capacity, m mix and m dist the mass of the mixture and distillate, T boil the boiling point and T 0 the room temperature, ΔH vap the enthalpy of evaporation, R min the minimum reflux ratio (set to 1, due to lack of information), and ղ heat the heating efficiency is, heating losses are assumed to be 31% 6 .

S1.5 Evaluating emerging technologies
Here, we follow the framework of van der Hulst et al., 7 based on three successive steps to evaluate technology maturing up to industrial scale.The upscaling involves both technical aspects as well as adjustments to background processes.The steps are shown in Table S.3.

S1.6 Electricity use in fossil-BTX pathway
Each of the separate Ecoinvent 3.8 datasets on naphtha-based benzene, toluene and xylene are formatted as a unit process, therefore "electricity" as input from the technosphere is not indicated as such.For this reason, the electricity use could not be automatically adapted to the future electricity market dataset.Table S.4 includes the total electricity for fossil-BTX which was separately assessed.
Table S4: total electricity use in fossil-BTX process, to adapt to a electricity market of 2050.Data on primary energy and share of electricity is taken from the original PlasticsEurope data documentation 15 .

S1.7 Impact assessment PB-LCIA
In the following section the PB-LCIA impact assessment is discussed step by step, with a summary in Table S.5.The LCI elementary flows are converted into impacts on the Safe Operating Spaces (SOS) of the Planetary Boundaries (PB) by multiplying the flows with characterization factors (Equation S2) taken from Ryberg et al. 16 .For the missing CFs on biosphere integrity, the approach of Galán-Martín et al. 17 was adopted.
The transgression level is defined as the impact of the product in relation to the safe operating space of the PB (Equation S3).This involves downscaling, i.e. allocating, the safe operating space to the level of BTX production (Equation S4).This is done based on the sharing principle 'equal per capita', assuming it is fair to share equally among the population.To then translate one's personal share of SOS to the chemical BTX, allocation was based on economic value.This based on the assumption that a higher economic value promotes well-being and therefore should result in a larger share of the SOS.For more explanation on the transgression level definitions, read Tulus et al. 18 .
Furthermore, the life cycle impact assessment requires a continuous input, because PBs are expressed in annual threshold levels 19 .Therefore, the functional unit should include a time dimension (kg/year) 20 .In our case this is defined as total annual BTX production (in weight).However, this cancels out when combining equation S3 and S4.S.3).The characterization factors are taken from Ryberg et al. 16,21 , see Supplementaryfile S.2.The pop tot represents the total population and pop btx the population that benefits from BTX production, which was assumed equal in this case.The GVA BTX is the gross value added associated with BTX, GVA TOT is the total gross value added of the world.where IMP pb,BTX is the impact of the production of BTX (per FU) related to the planetary boundary (pb), where PB is the total of all the planetary boundaries, LCI eBTX is the life cycle inventory elementary flow e (expressed over time) associated with BTX production, CF epb is the characterization factor (CF) that maps the elementary flow e onto the control variable.

Equation transgression level
where IMP pb,BTX is the LCI impact mapped onto the control variable (CV) of planetary boundary pb.The GVA TOT is the total gross value added of the world -the global GVA is retrieved from The World Bank database.To be consistent, the basic prices (price c ) are taken from Tulus et

Mixed plastic waste
Currently, MPW is most commonly incinerated with energy recovery 25 .To model the energy recovery from incinerating mixed plastic waste, the PEF guidelines 26 are followed (S7).An average Dutch incineration efficiency is assumed: 33.3%, with 28% for electricity and 9.3% thermal efficiency, and 4% of electricity for self-consumption 27 .Only electricity is considered in our model.Nevertheless, the heat generated could be further used leading to additional benefits.To account for the avoided product, i.e. bio-oil, light fuel oil is used.The lower heating value from different sources are shown in Table S. 6.Based on the Product Environmental Footprint (PEF) the following energy recovery formula is used: Where E recovery is the energy recovery with credits for avoided primary energy (which is the overall GHG benefit/disadvantage); E ER the emissions and resources consumed for energy recovery (modelled with Ecoinvent 3.8: Waste plastic, mixture {CH}| treatment of, municipal incineration); E se,heat and E se,elec are the emissions and resources consumed that would have arisen from specific substituted energy source (as Ecoinvent 3.8: Electricity, medium voltage {NL}| market for Heat, district or industrial, natural gas {RER}| market group for); X ER,heat and X ER,elec is the efficiency of energy recovery (28% for electricity and 9.3% thermal efficiency); LHV is the lower heating value of the material used for energy recovery (Table S

Crude glycerol
Crude glycerol can be refined to provide the pharmaceutical, food and cosmetic industry with pure glycerol, which is valuable but only economically feasible for large producers 28 .Direct combustion of glycerol is challenging due to its high viscosity and low energy density, therefore, recent developments are made to convert glycerol into fuel, hydrogen, biogas and/or co-generate heat and power 29,30 .The main alternative uses of glycerol are therefore modelled as (Table S.7): purification of glycerol 31 and combustion of biogas, fermented from glycerol 32 to generate electricity and heat 12 .The counterfactual of pure glycerol, i.e. synthetic glycerol, is based on synthetization of propylene via epichlorohydrin and is taken from Ecoinvent (3.8).

Heat and power (direct combustion)
LHV glycerol: ~16 MJ/kg.Glycerol has a very high activation energy resulting in an auto-ignition temperature of 370 °C.Conclusion: Direct combustion of glycerol is challenging due to its high viscosity and low energy density.

S2.4 Sensitivity Analysis: electricity market scenarios for 2050
Alongside the baseline scenario for the SSP2 narrative on the electricity market of 2050, two more scenarios were tested: a representative concentration pathway of 1.9 W/m 2 (RCP1.9)as well as 4.5 W/m 2 (RCP4.5) in 2100 35 .

S2.6 Sensitivity Analysis: Glycerol production from other feedstock
Glycerol production has a big impact on the production of bio-BTX.The default of glycerol as a co-product of US soybean bio-diesel has the lowest impact in terms of GHG, acidification, land use and water consumption.Glycerol from rapeseed oil and soybean oil from Brazil have higher GHG emissions, resulting from large impacts from clear-cutting of primary forest to arable land.Likewise, palm oil is known for its large GHG impact from direct land-use changes 36 , but also cultivation of the palm fruit bunches themselves is very GHG intensive.For the USA produced glycerol, no LUC emissions were included in line with PAS2050 guidelines 37 .
If cultivation requires land use conversion from (tropical) forest to agricultural land, the resulting GHG emissions are high and in regard to the impact of biodiesel and glycerol, it often exceeds fossil-based emission levels 38 .In this regard, and based on the other agricultural relevant impact categories (Figure S.2), glycerol from soybean biodiesel production in the USA has the lowest impact out of the four options.
However, the use of other feedstock, e.g.lignin 39 , could further decrease bio-BTX's impact, this is discussed in the discussion section.

S2.7 Sensitivity Analysis: Allocation strategies
In the figure down below the results when applying mass, energy and economic allocation on the GHG emissions of the BTX pathways are shown.Other ways to handle multifunctionality were not applied.
Substitution was not applicable because glycerol is a by-product and not the main product, and the coproducts in the final step are not marginal -it would lead to skewed impacts.; m based on mass allocation of 0.25 45 of sugar from sugarcane production, Brazil.n the impact of soybean cultivation as allocated to glycerol production.

Figure S1 :
Figure S1: overview of points of allocation in both MPW-and bio-BTX pathways.
Equation S2IMP b, BTX = ∑  ,    Equation S3 TL b, BTX =  ,   ,  Equation S4 Based on equal per capita approach and gross value added:  ,        LCI e,BTX is a LCI elementary flow e in BTX production, CF eb is the characterization factor for the LCI elementary flow e linked to PB b (for an overview of the PBs, see Table

Figure S. 2
shows the environmental impact of different glycerol production routes.These are: glycerol from palm oil (Ecoinvent 3.8: Glycerine {MY} | esterification of palm oil), glycerol from rapeseed oil (Ecoinvent 3.8: Glycerine {Europe without Switzerland} | esterification of rape oil) and glycerol from soybean oil from Brazil (Ecoinvent 3.8: Glycerine {Br} | esterification of soybean oil).

Figure S2 :
Figure S2: Results of the production of 1 kg of glycerol made from palm oil (left bar), rapeseed oil (one to the right), soybean oil from Brazil (one from the right bar) and soybean oil USA (right bar) on (a) Climate change, (b) Terrestrial acidification, (c) Freshwater eutrophication, (d) Land use, (e) Water consumption, (f) Fine particulate matter formation.Green bar represents the base case glycerol from soybeans (USA).

Figure S3 :
Figure S3: Climate change results of MPW and bio-BTX applying mass, energy and economic allocation.The base scenario is indicated with the orange box.BTX = benzene-toluene-xylene; MPW = Mixed Plastic Waste; LFO = Light Fuel oil.

Table S1 :
data on carbon content BTX and related emissions

Chemical formula Molar weight (g/mol) CO 2 emissions related to carbon content Benzene
Figure S.1 indicates the points of allocation in the MPW-and bio-BTX pathways.Table S.2 is an overview of the different allocation factors including calculations.

Table S5 :
24,21methods and definitions, summary of Planetary Boundary concepts and the PB-LCIA method.CFs linking LCIA to PBs16,21.Additionally, for biosphere integrity, CF is based on the approach of Galán-Martín et al. (2021), build on work from Hanafiah et al.22and updated using Wilting et al.23and GLOBIO 3.5 MSA values24.
al. (2021): 2.55 (xylene), 0.73 (toluene) and 0.88 (benzene) in USD2018.The SOS pb is the total SOS for CV pb .For the steps to arrive at this equation, see Supplementary S.10 and the work of Tulus et al., 2021.

Table S6 :
.6). different data on LHV of plastics and mixed plastic waste.

Table S7 :
overview data on alternative uses for crude glycerol.
* Energy consumption per m 3 of biogas: 0.158 kWh of electricity and 3.470 MJ heat per m 3 biogas 32 .
. Total production of BTX per year was taken as 1.22E11 kg (ref).

Table S14 :
Global warming potential results for future (2050) BTX-pathways applying different recycling scenarios.

Table S15 :
Greenhouse gas (GHG) emissions, freshwater eutrophication, acidification and land use impact of biomass cultivation per kg of BTX, based on yields from literature.The percentage are the in-or decrease in impact compared to soybean cultivation impact.N.A. = not available.

Table S16 :
Data and calculations on plastic waste feedstock and it availability for BTX production.BTX = benzene-toluenexylene.