Region-Specific Sourcing of Lignocellulose Residues as Renewable Feedstocks for a Net-Zero Chemical Industry

Biobased chemicals, crucial for the net-zero chemical industry, rely on lignocellulose residues as a major feedstock. However, its availability and environmental impacts vary greatly across regions. By 2050, we estimate that 3.0–5.2 Gt of these residues will be available from the global forest and agricultural sectors, with key contributions from Brazil, China, India, and the United States. This supply satisfies the growing global feedstock demands for plastics when used efficiently. Forest residues have 84% lower climate change impacts than agricultural residues on average globally but double the land-use-related biodiversity loss. Biobased plastics may reduce climate change impacts relative to fossil-based alternatives but are insufficient to fulfill net-zero targets. In addition, they pose greater challenges in terms of biodiversity loss and water stress. Avoiding feedstock sourcing from biodiversity-rich areas could halve lignocellulose residues-related biodiversity loss without significantly compromising availability. Improvements in region-specific feedstock sourcing, agricultural management and biomass utilization technologies are warranted for transitioning toward a sustainable chemical industry.


S1.1.2. GLOBIOM model
The Global Biosphere Management Model (GLOBIOM) is a partial equilibrium economic model that focuses on the agriculture, forest, and bioenergy sectors. 1,2 t computes market equilibrium by optimizing total consumer and producer surpluses through the allocation of land use within these sectors. 1,2 LOBIOM is spatially explicit, operating at a spatial resolution of 200 km × 200 km for modeling land use and biomass production, while it models biomass demand, e.g. for food or as a building material, at the regional level. 3Supplementary Table 2 presents a full list of the GLOBIOM outputs used in this study for the assessments of availability and associated environmental impact.

S1.2.1. Residue-to-production ratios
The residue-to-production ratios were expressed as empirical functions of crop yields (y, in tonne dry mass/ha), which are spatially explicit outputs of the GLOBIOM model, and commonly take linear, logarithmic, or exponential forms (Table S3). 4 Constant RPRs (Table S4) are used to calculate process residue availability (rice husks and sugarcane bagasse) and harvest residues when empirical RPR functions are not available (Sugarcane tops and leaves).

S1.2.3. Summary of different potentials considered
Table S5.Different potentials considered for lignocellulose residues and key assumptions

S1.4. Life-cycle inventories of lignocellulose residues
The environmental impacts associated with lignocellulose residues were assessed using prospective LCAs, implemented with the Brightway2 framework, 14 as outlined in Figure S4.Differentiating the land-use intensities of croplands and managed forests enables a more precise impact assessment of land-use-related biodiversity loss.In GLOBIOM, the land-use intensity of croplands was differentiated with the endogenous parameter of crop management systems.
Subsistence farming was associated with minimal land use, low-input rainfed systems with light land use, and high-input rainfed / irrigated systems with intense land use.
The assessment of different land-use intensities in the forest sector, on the other hand, was enabled through GLOBIOM-forest, a submodel of GLOBIOM that provides a more comprehensive and detailed representation of the forest sector.It encompasses three distinct forest types: primary forests, secondary forests, and managed forests.The "managed forests" category is further subdivided into three distinct management intensity levels-low intensity, multifunctional, and high intensity-thereby providing a detailed spectrum of forest management practices. 15This level of detail enables a more comprehensive assessment of the land-use-related impacts of forestry products.
Allocation of the land-use changes for cropland was performed following the PAS 2050-1 Guidelines. 16 A similar allocation process was developed for managed forests (Figure S5).It is important to note that the expansion of secondary forests and afforestation are categorized independently and do not contribute to the impacts or credits associated with wood products harvested from managed forests.

S1.4.2. Life-cycle inventories of agricultural residues
Table S7 provides a detailed description of the procedures used to create the inputs and emissions.
Agri-footprint 6 incorporates ecoinvent 3.8 for its background processes such as fuel, electricity, and transportation.These processes were relinked with regionalized and prospective premisegenerated background LCIs.In this study, we extended our analysis with more countries for each crop than is covered by Agri-footprint 6.For the inputs and emissions not specifically modeled in our study, we calculated a global average for each crop.We did this by taking the production amount from countries included in Agri-footprint 6 and using them to weigh these averages.We then merged these global averages with the region-specific data we modeled, applying them to the countries that fell outside the scope of Agri-footprint 6.These calculated values were directly applied to all the countries that were not included.The impacts from the reference flow of one hectare of cropland were allocated to crops and their corresponding ecological potential of harvest residues, according to their economic values.However, as agricultural residues were not included in the GLOBIOM model as products, it was assumed that per kilogram agricultural residues shared the same price as forest residues.The impacts of agricultural process residues (rice husks and sugarcane bagasse) were not modeled.To avoid unreasonably high application, the highest possible application rate of potassium fertilizer was assumed to be the 95 th percentile of the application rate in the FUBC database.
To match the aggregated NPK fertilizer application with each fertilizer product, the average agricultural use of each fertilizer product from 2018 to 2020 on the country level from the Food and Agriculture Organization of the United Nations (FAO) database was used. 19The matching procedure followed Agri-footprint 6 methodology report. 20nd use change emissions The considered GHG emissions from land use change include CO2 emissions from net carbon change in biomass, and net soil organic carbon (SOC) change in mineral soils.The calculation steps followed Agri-footprint 6 methodology report 20 and the IPCC guideline, 21 which below specified data sources: The above ground and below ground carbon stock in forest by country reported in Global Forest Resources Assessments by FAO 22 was used in this study.The default of carbon stock in annual and perennial croplands (4 and 20 tonne C/ha respectively) were used according to PAS 2050-1 guideline 16 .The carbon stock in grassland varies according to climate and soil types, with default values in each category from the IPCC guideline. 21The climate zone maps 21 and the harmonized world soil database v1.2 23 were used to derive the share of each climate-soil combination on the country level.

Fertilizer emissions
The updated field emissions as a result of fertilizer application include CO2 from urea, direct and indirect N2O, NH3 and NO3 --N emissions from nitrogen fertilizers.IPCC tier 1 emission factors and constants 24 were used in the calculation.

S18
Crop residue emissions Due to a lack of data on the fraction of crop residues being incinerated on the field for future scenarios, it was assumed that the entire unharvested crop residue is left in the field, and thus lead to direct and indirect N2O emissions.IPCC tier 1 emission factors and constants 24 were used in the calculation.

S1.4.4. Land use mapping
The mapping between the background LCI (ecoinvent 3.8) and the LCIA method follows Scherer et al. 25 Table S9

S1.5. Process simulation of biomass fractionation
The process simulation of the biomass fractionation process was developed in Aspen Plus v12 and it is based on the work of Talebi Amiri et al. 26 More specifically, on steps 35 to 83 of the propionaldehyde biomass fractionation protocol to afford cellulose, stabilized lignin and dipropylxylose (DPX) (Figure S6).We assumed an initial biomass composition of birch wood extracted from the same work, excluding the minor sugars, acid-soluble lignin and acetyl (Table S10).
The components xylan, arabinan, lignin, stabilized lignin, glucan and DPX were created as new Aspen components.We estimated the missing property data of the new components with the UNIFAC method built in Aspen Plus v12.
Table S11 collects the initial data used for the component property estimation.
The process simulation mass flows were adapted from the original laboratory protocol to reflect a potential industrial-scale capacity.For example, the initial components entering the process (biomass, propionaldehyde, 1,4-dioxane and 37% wt.hydrochloric acid) were linearly scaled from 4.5 g, 4.8 mL, 25 mL and 0.85 mL to 45 kg/h, 48 L/h, 25 L/h and 8.5 L/h, respectively.Operating temperatures and pressures were faithfully adapted from the protocol.
Due to the lack of specific data about the separations and the solid/solvent interactions, extractions and washes were assumed to behave ideally (i.e., the totality of the desired fraction to be removed with the solvent, and the solvent itself, is successfully recovered).Additionally, on top of the already existing vacuum flash drums described in the original protocol, several distillation columns were added to the process to recover and recycle most of the solvents used in these extraction and wash steps.Table S13 shows the specifications of these columns.Furthermore, a purge of 0.1% of each recycle stream was assumed to consider potential losses.
Due to the lack of a vacuum pump unit in Aspen Plus, electricity consumption associated with vacuum generation was approximated to the work required to recompress the subatmospheric pressure streams after the flash drum separation back to 1 bar.
The main reactor is modeled as an isothermal conversion reactor, while the neutralization reactors and the cellulose hydrolysis reactor are considered adiabatic.The chemical reactions taking place and their respective conversion are displayed in Table S13.
In the furnace, all unseparated solvents and unretrieved biomass are combusted using air and considering 100% conversion.The energy of the process cold and hot streams is integrated considering the targets provided by Aspen Energy Analyzer.** assuming all carbon from the lignocellulose residue that is not embedded in the methanol is released as biogenic CO2.It includes fuel emissions as well as process emissions.

Table S16. Life-cycle inventory of propylene produced from methanol-to-olefin process
The inventory data is based on Hoppe et al.

S2.7. Contribution analysis of climate change impacts of biobased platform chemical
To contextualize our results in a transition towards a sustainable chemical industry, a case study is performed to assess the impacts associated with production of biobased platform chemicals (glucose, xylose, and lignin), considering the RCP1.9 scenario for 2050.The climate change impacts are highlighted in Figure S18, showcasing a range from 1.4 to 2.8 kg CO2-eq/kg chemical.
As expected, the choice of biomass feedstock plays a key role in determining the climate change impacts of biobased chemical production.While feedstocks from forest residues contribute negligible climate change impacts on global average (scenarios S1, S2, S4, and S5), the use of agricultural residues may account for up to 20% of the overall impact (scenarios S3 and S6).
Low-impact feedstock alone does not guarantee overall low climate change impacts of biobased chemicals.Electricity consumption is another major determinant, where 4.9 kWh is required per kg of chemical products, particularly for solvent recycling.The impacts of electricity vary depending on the carbon intensity of the region's energy mix, typically ranging from 0.10 to 0.97 kg CO2-eq/kg chemical.
In comparison with chemicals produced with existing technologies, glucose from maize starch has a carbon footprint of 0.69 kg CO2-eq/kg when the energy mix is updated with premise to represent the RCP1.9 scenario in 2050, as per ecoinvent 3.8 database 31 .However, maize starch is not a focus of this study due to concerns about potential competition with food.
This case study is based on process simulations of a laboratory-scale protocol, emphasizing the need for process optimization prior to industrialization.
In addition, biogenic carbon storage, achieved with durable product design and proper end-of-life management, offers opportunities to reduce the climate change impacts throughout the lifecycle of biobased chemicals.and biomass feedstock scenarios (min: 0.004 kg CO2-eq/kg DM, wood chips in China; max: 0.315 kg CO2-eq/kg DM, rapeseed straw in Bangladesh).

S3.1.1. Residues price
We assume that agricultural residues share the same price as forest residues, because they are considered as perfect substitutes for each other in biobased chemicals.However, agricultural residues may be used as animal food and bedding in addition, which may increase its price.If this were the case, more impacts would be allocated to agricultural residues than in this study.
Due to the higher demand, biomass residues have higher price in 2050 under RCP1.9 than under RCPref.On the other hand, the price of the main products (crops and wood products) does not show a big gap between the two RCP scenarios.As a result of economic allocation, more impact is allocated to biomass residues and less impact is allocated to the main product.This partially explains why the climate change impacts of crop residues under RCP1.9 in some countries show a growing trend in the future.We therefore did a sensitivity analysis of the climate change impacts of aggregated crop residues in the major biomass producing countries and assumed the price of biomass residues under RCP1.9 is the same as under RCPref (Figure S19).Under this assumption, the climate change impacts of crop residues under RCP1.9 is lower than RCPref because of the cleaner energy systems.

S3.1.2. GLOBIOM model with endogenous supply of crop residues
Crop residues are not endogenously included in the GLOBIOM model as biomass resource.Instead, GLOBIOM assumes that a constant of 31 EJ biomass is supplied by other biomass (including crop residues) than wood and energy crops.However, when more crop residues are used to satisfy the biomass demand depicted by SSP and RCP scenarios, less energy crops would be needed and hence less land would be transformed for plantation.To understand the impact of crop residues on land use change, a sensitivity analysis was conducted to include crop residues endogenously in the GLOBIOM model under some simplified assumptions:  50% of the total crop residues are left on the field for ecological reasons.

S3.2.1. Product demand
The uncertainty of the available potential of lignocellulose biomass residues is partially addressed by the choice of different residue-to-crop ratio (RPR) empirical functions to capture the possible range of residue yield.However, the available potential of lignocellulose biomass residues is also affected by the production and consumption of crop and wood products, which are highly uncertain for future scenarios.In this study, we examine the future scenarios under the narrative of the Shared Socioeconomic Pathway 2 (SSP2), which represents a moderate development scenario 33 .Yet we do acknowledge that the uncertainty of consumption behavior are depicted in other SSP scenarios.
Daioglou et al. 34 projected a global availability of agricultural and forest residues in 2050 based on the IMAGE IAM model 35 .They found that the different SSP scenarios do not have a strong impact on the available potential.

S3.2.2. End-of-life biogenic CO2 emissions
Currently, there is a lack of consensus on the methodology for assessing the climate change impacts of biogenic CO2 emissions.The uncertainties arise when considering the varied rotation periods and management practices of biomass sources, as well as carbon storage time before CO2 release.
For agricultural residues, with typically short rotation periods, the assumption is that the release of biogenic CO2 has negligible impact on global warming.In contrast, the impact of biogenic CO2 emissions from forest residues is more uncertain due to the influence of longer rotation periods and diverse forest management practices.The extent to which these factors affect the net climate impact of biogenic CO2 emissions is not well established.Furthermore, the climate change impacts of different end-of-life scenarios for biobased products, such as incineration or recycling, add another layer of uncertainty.With recycling, the carbon storage in the biobased products is longer, and the climate change impacts of biogenic CO2 emissions can be reduced.

S3.3.3. Climate change impacts of intensified forest management
More forests will be under intensified management to satisfy the growing biomass demand under RCP1.9, which may lead to a decrease in carbon stocks in the forest.This decrease can be attributed to several key factors.Firstly, the shift towards monoculture and the prioritization of fast-growing species for economic gain often leads to a homogenization of the forest ecosystem, which can result in lower biomass per area as these species may not store as much carbon as a diverse array of native species.Secondly, intensified practices such as frequent harvesting and soil disturbance from mechanical operations disrupt soil carbon pools, leading to direct releases of carbon dioxide into the atmosphere.Additionally, shorter rotation periods prevent forests from reaching their full carbon storage potential, as mature forests typically hold more carbon in their biomass and soil than young forests.These factors collectively contribute to a reduction in the forest's ability to act as a carbon sink, thus posing the challenge of climate change mitigation.Therefore, forest residues from intensified forests might have higher climate change impacts than from forests under minimal or light management levels.However, the management levels are not differentiated the IPCC guidelines 21 , and hence, the climate change impacts of intensified forest management are not quantified in this study.

S3.3.4. Additional environmental impacts caused by lignocellulose residue removal
Our analysis incorporates ecological constraints to ensure that a significant proportion of residues (28-43%) remains on the field, compared to the current global average of 42-49% of cereal residues left in situ. 368][39] However, globally, up to 26% of all cereal residues (considered within our study's "sustainable potential") are currently used for purposes such as domestic fuel or are burnt on the field. 36In developing countries, this number is even higher. 36Redirecting these residues from such uses does not inherently result in additional SOC loss.Therefore, applying a global emission factor from the literature, which assumes no current residue harvest, would overestimate the climate change impacts associated with SOC changes from lignocellulose residues.In addition, the effects of residue removal on SOC are complex and vary widely by local climate, soil type, and agricultural and forestry practices, making it a challenging impact to quantify accurately on a global scale.
The biodiversity impacts of removing residues, particularly from forests, have not been directly quantified in this study.Residues play a critical role in maintaining ecological functions and biodiversity by providing habitats.Although our analysis maps biodiversity impacts to different types of land use and land use change, it does not specifically distinguish impacts based on the presence or absence of residue removal, which may lead to underestimations of the biodiversity impacts associated with lignocellulose residue-based products.
To better reflect the complexity and regional variability of environmental impacts due to residue removal, future studies should develop more region-specific guidelines for quantifying SOC changes and biodiversity loss impacts due to residue removal, similar to existing IPCC guidelines for land use change. 21Such guidelines would enhance the accuracy of environmental impact assessments and support more sustainable residue management practices.

S3.3.5. Other uncertainties
The life-cycle assessment in this study relies on background databases such as ecoinvent 3.8 and Agri-footprint 6.These databases include the production of fertilizers and other chemical feedstocks under current technology, which is mostly fossil-based production pathways.As there is no information regarding the future low-carbon production pathways of fertilizers and other chemical feedstocks, these background databases with fossil-based pathways were utilized with updated energy mixes for the future scenarios.This may lead to an overestimation of the climate change impacts, especially for 2050 under RCP1.9 scenario.
The cradle-to-gate climate change impacts of forest residues are mainly determined by the energy use during harvest.We acknowledge the regional differences in wood harvesting methods, however, as such information is not available for most countries, the harvest activities in Switzerland in ecoinvent 3.8 are used for all other countries.Wood harvest in Switzerland is highly mechanized, with even 1.1-3.2%wood being harvested by helicopter 40 .In other countries with less mechanization of wood harvest, this impact would be smaller.Despite the potential over-estimation of the cradle-to-gate climate change impacts, forest residues still present lower climate change impact than agricultural residues.

S1. 3 .
Global plastics production and embedded carbon

Figure S5 .
Figure S5.Land use change allocation process for managed forest

Figure S6 .
Figure S6.Simplified process scheme of the biomass fractionation process.

Figure S11 .S2. 5 .
Figure S11.Projected water stress and land-use-related biodiversity loss impacts of lignocellulose residues in 2050 under the RCPref scenario.

Figure S12 .S2. 6 .
Figure S12.Land use and its related impacts of lignocellulose residues from 2020 to 2050 in Brazil.

Figure S17 .
Figure S17.Climate change, water stress, and biodiversity loss merit-order curves of lignocellulose residues inBrazil, China, India, and the United States under the RCP1.9 scenario in 2050.

Figure S18 .
Figure S18.Climate change impacts of biobased platform chemicals based on propionaldehyde fractionation in 2050 under the RCP1.9 scenario.

 3 
No trade in crop residues Heat values and densities are same for all crops: 16 GJ/tonne dry mass and 0.45 tonne dry mass/m Constant crop residues yields are assumed according to Holmatov et al.32

Figure S19 .
Figure S19.Sensitivity analysis: the impact of biomass residue price on climate change impacts of aggregated crop residues.

Table S1 .
List of lignocellulose residues included in the study ** Only from managed forests.Forest plantations are excluded from the study scope.

Table S2 .
List of GLOBIOM outputs * crop technology in GLOBIOM corresponds to different management intensity levels, see TableS9** see FigureS2* A carbon price in Integrated Assessment Models is a theoretical or modeled societal cost assigned to greenhouse gas emissions to simulate the economic impacts of carbon emissions and evaluate the effectiveness and costs of mitigation strategies.

Table S7
18BC) survey data published by the International Fertilizer Association was used instead18.Since there is future projections in FUBC, it was assumed that the application rate of potassium fertilizer would increase with the same ratio as nitrogen fertilizer in future scenarios following Eq.S6.
16List of updated inputs and emissions for LCIs of crop production(reference flow: one hectare of cropland)Inputs and emissionsDescriptionLand occupation 1 ha of cropland, as the reference flow.The share of each cropland use intensity was categorized for country-crop combinations based on the share of harvest area with the corresponding crop technology, with details in TableS9.Land transformationFollowing PAS 2050-1 guideline,16the area of each land use type was assessed for the reference year and 20 years before that.Land use change was assumed to distribute equally over the 20 years.The model structure for the detailed calculation steps was updated based on PAS 2050-1 guideline by taking primary forest into account.FertilizerNitrogen and phosphorus fertilizer application rate [kg/ha] is embedded in the crop technology parameter from the GLOBIOM model at the 200 km × 200 km resolution gird level for each crop.As potassium fertilizer is not included in GLOBIOM model, the data from the fertilizer use S17 by crop ( ,,2020 − [kg/ha]: nitrogen fertilizer application rate for crop  in country  reported by FUBC.

Table S8 .
List of regionalized ecoinvent activities for forest residues production

Harvesting activities: regionalized on the country level Cleft
Sawlog and veneer log, hardwood, measured as solid wood under bark | hardwood forestry, mixed species, timber, measured as dry mass | hardwood forestry, mixed species, sustainable forest management (logging residue, hardwood) Cleft timber, measured as dry mass | softwood forestry, mixed species, sustainable forest management (logging residue, softwood) * Considered forest residues in this study are marked bold.

Table S10 .
Composition of the lignocellulose residues

Table S11 .
Bibliographic data used for the property estimation of the new Aspen components.

Table S13 .
Design specifications for the solvent recovery distillation columns.

. Life-cycle assessment of biobased plastics
Table S14 to Table S16 are life-cycle inventories of glucose, methanol, and propylene used for the biobased plastics.The life-cycle inventories of lactic acid, polypropylene and polylactic acid are based on the Process Economics Program Yearbook 28 that require an additional license from IHS Markit.

Table S14 .
Life-cycle inventory of glucose produced from biomass fractionation.

Table S15 .
Life-cycle inventory of methanol produced from biomass gasification.