Insights from Life Cycle Assessment to Inform Chemical Substitution, Alternatives Assessment and Safe and Sustainable-by-Design

Abstract

Designing chemicals and pharmaceuticals with optimal properties yet minimal negative effects on humans and the environment remains challenging. Chemical substitution enables us to address this challenge but requires significant methodological advancements and testing to become mature and more widely adopted worldwide. We use insights from life cycle assessment (LCA) to explore and synthesize synergies for advancing substitution approaches, including alternatives assessment and comparative safe and sustainable-by-design (SSbD). We screen boundary conditions for LCA and substitution and use an illustrative case study on comparing two sets of pest control options to discuss learnings for addressing challenges toward a mature chemical substitution field. For some aspects, LCA provides strong inspiration to inform developments of substitution approaches, such as using a common functional comparison basis, aligning assessment metrics and units, and considering full life cycles to avoid burden shifting. For other aspects, LCA is less informative and substitution requires additional developments. This includes systematic overviews of possible context-specific alternatives, mapping of chemical, material, and other functions, more detailed life cycle inventories, and rapid-screening assessment methods. Combining learnings from LCA with ways to address remaining challenges will help advance chemical substitution to support the objectives of UNEP’s Global Framework on Chemicals and other policy instruments for a broader, much needed sustainability transition.

1. Introduction

The chemical and pharmaceutical industry is a global driver of societal welfare, forming the basis for many other sectors, from health care and agrifood production to building/construction and consumer goods. Striking a careful balance between developing chemistries with beneficial properties while minimizing unintentional side effects on humans and the environment − is a key objective in chemical and drug design. However, designing chemicals and pharmaceuticals with optimal functional properties and minimal undesired impacts remains a key challenge to foster chemical innovation, which led to the establishment of fields like green chemistry and sustainable chemistry, the implementation of regulations like REACH in Europe and global policy initiatives like UNEP’s Global Framework on Chemicals, and the development of specific assessment frameworks like chemical substitution, alternatives assessment, and safe and sustainable-by-design (SSbD). −

Alternatives assessment is a tool that informs chemical substitution by identifying and evaluating a range of safer options to phase out and replace a harmful chemical in one or more product or process applications. SSbD, in contrast, is an emerging tool that aims at fostering innovation for designing new chemicals, materials, and products as part of the European Chemicals Strategy for Sustainability, , combining aspects of safety as well as sustainability. SSbD can essentially also be applied to compare and substitute harmful chemicals with new designs that are potentially safer and more sustainable alternatives or ultimately even striving for identifying alternatives that are actually safe and sustainable, when also considering relevant biophysical benchmarks, such as the planetary boundaries for certain environmental sustainability aspects. − However, all these frameworks face common methodological and practical challenges, which has led to ‘regrettable substitutions’ in the past, − hampering a broader adoption by industry and policy makers. Historically, ‘regrettable substitutions’ occurred where chemicals have been replaced by structurally similar alternatives with similar hazards or alternatives with different/unknown hazards, often based on missing or incomplete assessments of alternatives. Main reasons that can lead to ‘regrettable substitutions’ are data gaps for relevant hazard properties, where alternatives have either not yet been broadly tested and marketed or only tested for hazards that led to concerns for the chemicals that were substituted, − and a strong focus on replacing chemicals with similar chemistries that can fulfill the same desired function but by design often have similar hazard profiles.

Among the prevailing key challenges for sustainability-driven innovation in chemical substitution and alternatives assessment are the following: (a) Defining a common function at the relevant assessment scale. Identifying functionally equivalent alternatives is not straightforward for alternatives beyond drop-in chemicals, (e.g., multifunction chemistries, technologies, behavioral changes). (b) Considering the wider range of impacts associated with a chemical and related product or process life cycle. Impact burden can shift across different stages of chemical or technology life cycles and introduce bias in a substitution when left unaddressed. (c) Appropriately identifying relevant impact trade-offs. Trade-offs among alternatives often go beyond toxicity-related aspects and need to be considered in a fair assessment. (d) Including quantitative metrics for all relevant human and environmental health aspects. Adopting quantitative metrics allows for aggregation and benchmarking of impact performance results, which is relevant in comparative assessments. (e) Rapidly generating and aggregating assessment results. This facilitates an appropriate comparison of various potential alternatives to harmful chemicals. Most of these challenges are not new and have been extensively discussed and addressed in the development of more mature and widely applied decision-support frameworks, most prominently in environmental life cycle assessment (LCA).

LCA is globally standardized and aims at comparing the overall environmental performance of products and services over their entire life cycle from resource extraction via manufacturing and use to end-of-life treatments, such as disposal or recycling. , Several of the characteristics of LCA (e.g., being based on functional comparisons, considering various environmental impacts, and using quantitative metrics) render the learnings of developing and standardizing this framework potentially useful to inform substitution-related approaches to overcome challenges that are similar in LCA. In the present study, we hence provide an illustrative, real-world case study for substituting hazardous chemicals in a specific application context based on LCA. We use insights from this example to discuss some of the existing challenges in current frameworks focusing on chemical substitution and SSbD, while acknowledging the distinct purpose and boundary conditions of different assessment frameworks. As substitution context, we selected two distinct chemical pesticide active ingredients (hereafter referred to as ‘pesticides’) identified by the International Pesticide Action Network as Highly Hazardous Pesticides (HHP) to be phased out in agricultural uses worldwide. We summarize our findings in a set of specific recommendations in support of fostering the application of chemical substitution approaches and their uptake into policy instruments as an important component to promote sound and sustainable chemicals management.

2. Materials and Methods

2.1. Evaluating Assessment Boundary Conditions

While comparing two or more options is at the heart of both (comparative) LCA and substitution-related assessments, certain boundary conditions differ. For example, chemical substitution focuses on addressing harmful chemicals in materials, products, and processes and with that has a different assessment scope as compared to LCA typically evaluating life cycles at the level of overall products, technologies, or else. Both chemical substitution and alternatives assessment also include functional performance and economic and technical feasibility aspects that are not commonly considered in LCA. Finally, substitution and especially SSbD aiming at driving innovation at the early design stages across multiple potential options both require rather rapid-screening tools aligned with short industrial innovation cycles often in the range of some weeks, whereas LCA often requires several months for comparing typically two to three scenarios, product, or technology life cycles.

To facilitate the possible transfer of learnings from an LCA study to chemical substitution, alternatives assessment, and substitution-focused SSbD contexts, we focus in the present study on commonalities between LCA and substitution-related approaches. As a starting point, we analyzed a noncomprehensive set of relevant literature for boundary conditions on the one hand for LCA − ,− and on the other hand for chemical substitution, ,− alternatives assessment ,,− and SSbD (discussed in the context of chemical substitution or comparing alternatives). ,,− Identified boundary conditions were structured and contrasted according to (a) application or use context, (b) relevant assessment aspects, and (c) interpretation for decision support to discuss how addressing them in LCA can inform methodological ways forward in substitution and SSbD approaches.

2.2. Defining Viable Substitution Scenarios

Chemical substitution focuses either on replacing hazardous chemicals in industrial processes and consumer products or on replacing pesticides of concern in agricultural applications. Since several pesticides are known to be hazardous, we use the substitution of pesticides as an illustrative case study. Chemical pesticides are widely used in agriculture to control plant pathogens, insects, and weeds, but are also toxic-by-design and can therefore harm humans and nontarget ecosystems. ,,− Though chemical pesticides are regulated in many countries, there is potential for further reducing environmental impacts by substituting, especially, hazardous pesticides. This includes replacing harmful pesticides with less toxic/ecotoxic substances, with different types of functionally equivalent biopesticides or mechanical pest control solutions. Several studies already exist that aim to compare different (mostly xenobiotic, i.e., not naturally occurring) pesticides against each other based on specific performance indicators. − However, comparisons in such studies are usually narrow in scope, focus exclusively on chemical pesticides, and do not apply a consistent, broader set of impact performance indicators. To illustrate how different types of pest control options can be compared in a substitution context, we used this context for defining our case study.

We identified two distinct example pest control targets to define substitution scenarios (per target), namely, downy mildew as a plant pathogen controlled by fungicides in European grape/vine production, and the common waterhemp as an annual broadleaf weed controlled by herbicides in U.S. soybean production. For each target, we selected a base case marketed product containing a chemical pesticide listed as HHP, namely, the fungicide folpet and the herbicide glyphosate, respectively. Folpet is often accompanied by another pesticide; hence, we selected a product containing folpet and mandipropamid as the base case. Potential alternatives to the base case pesticides were screened based on controlling the same target pest, authorized for use in the same crop in the identified regions, and going beyond 1:1 substitution by another chemical pesticide. Examples of potential, functionally equivalent alternatives to the fungicide mix controlling downy mildew include the chemical fungicide oxathiapiprolin, the metal-based fungicide copper hydroxide, and the biopesticide laminarin (Table ). Potential alternatives to the chemical herbicide controlling common waterhemp include the bioherbicide pelargonic acid and mechanical weeding as a physical weed control measure (Table ). We assumed treatment during the same crop growth stage, equal target pest efficacy, same number of maximum treatments per crop cycle, and no pest resistance building, and we discuss these aspects further in our study limitations.

1. Overview of Preventive Plant Pathogen Control Scenarios Using Four Functionally Equivalent Alternatives for Managing Downy Mildew in Grapes and Vines in the European Union.

Scenario		Target pest	Pest type: Obligate parasitic microbe
			Pest species: Downy mildew (Plasmopara viticola)
		Crop	Grapes/vine (Vitis vinifera)
		Region/Country	European Union
		Pest control form	Preventive pest control
Alternatives	1. Chemical pesticide mixture (base case)	Product	Product name: Pergado (Syngenta Agro GmbH)
			Registration status: Approved until 15 Feb 2026
			Maximum applications per crop cycle: 4
			Application rate: 1.25 kg/ha
		Pesticide active ingredients	1.a Substance name: Folpet
			CAS RN: 133-07-3
			Substance group: Phthalimide fungicide
			Mode of action: Multisite inhibitor (blocks different enzymes that are essential in targeted pathogens)
			Content in product: 400 g/kg
			1.b Substance name: Mandipropamid
			CAS RN: 374726-62-2
			Substance group: Mandelamide fungicide
			Mode of action: Cellulose synthesis inhibitor (inhibits cell wall biosynthesis and spore germination in pathogens)
			Content in product: 50 g/kg
	2. Chemical pesticide	Product	Product name: Zorvec Zelavin (Corteva Germany GmbH)
			Registration status: Approved until 03 Aug 2028
			Maximum applications per crop cycle: 2
			Application rate: 0.16 l/ha
		Pesticide active ingredient	Substance name: Oxathiapiprolin
			CAS RN: 1003318-67-9
			Substance group: Piperidinyl-thiazole-isoxazoline fungicide
			Mode of action: Oxysterol-binding protein homologue inhibitor (OSBPI, inhibits all asexual life cycle stages of pathogens)
			Content in product: 100 g/L
	3. Metal-based pesticide	Product	Product name: Hycop (Sharda Cropchem Espana S.L.)
			Registration status: Approved until 31-Dec-2026
			Maximum applications per crop cycle: 4
			Application rate: 0.5 kg/ha
		Pesticide active ingredient	Substance name: Copper(II) hydroxide
			CAS RN: 20427-59-2
			Substance group: Inorganic fungicide
			Mode of action: Multisite cellular respiration and enzyme function inhibitor (disrupts the enzyme system of pathogens)
			Content in product: 767.9 g/kg
	4. Biopesticide	Product	Product name: Vacciplant (Stähler Suisse SA)
			Registration status: Approved until 28 Feb 2033
			Maximum applications per crop cycle: 4
			Application rate: 2 l/ha
		Pesticide active ingredient	Substance name: Laminarin
			CAS RN: 9008-22-4
			Substance group: 1,3-β-d-glucan polysaccharide fungicide (derived from brown algae Laminaria digitata)
			Mode of action: Systemic acquired resistance (SAR) activator (stimulates the natural defense system of crops)
			Content in product: 45 g/L

^aAuthorization of active ingredients lies with the European Commission, while registration of using individual market products on specific crops lies with and may vary across EU Member States.

^bProduct often not sufficiently effective, and may be complemented with, e.g., 3 kg/ha Armicarb (contains 850 g/kg potassium bicarbonate, CAS RN: 298-14-6).

2. Overview of Curative Weed Control Scenarios Using Three Functionally Equivalent Alternatives for Managing Common Waterhemp Representing a Wider Range of Weed Pests in Soybeans in the United States.

Scenario		Target pest	Pest type: Annual, biennial and perennial broadleaf/grass weeds
			Pest species: Common waterhemp (Amaranthus tuberculatus)
		Crop	Soybean (Glycine max), not genetically modified
		Region/Country	United States
		Pest control form	Curative pest control
Alternatives	1. Chemical pesticide (base case)	Product	Product name: Roundup PowerMAX (Bayer Crop Science US)
			Conditionally registered: No
			Maximum applications per crop cycle: 1
			Application rate: 3.2 l/ha
		Pesticide active ingredient	Substance name: Glyphosate
			CAS RN: 1071-83-6
			Substance group: Organophosphorous herbicide
			Mode of action: Enolpyruvyl shikimate phosphate synthase (EPSPS) inhibitor (inhibits the synthesis of aromatic amino acids involved in protein formation relevant for plant growth)
			Content in product: 660 g/L
	2. Biopesticide	Product	Product name: Scythe (Gowan Company LLC)
			Conditionally registered: No
			Maximum applications per crop cycle: not provided
			Application rate: 1.2 l/ha
		Pesticide active ingredient	Substance name: Pelargonic acid
			CAS RN: 112-05-0
			Substance group: Aliphatic acid herbicide
			Mode of action: Cuticular wax layer stripper/cell membrane disruptor (physically disrupts leaf surfaces and cell walls)
			Content in product: 500 g/L
	3. Mechanical pest control	Method	Pest control method: Mechanical weeding
			Weeder: Hatzenbichler Tine Weeder “Original Harrow”
			Technical aspects: Parallel linkage, hydraulic down pressure
			Harrow row working width: 15 m
			Field operations machinery: 0.75 kg/ha

^aRegistration has no expiration date, but registrants are obligated to pay annual maintenance fees for every registered product to ensure continuity of the registrations (https://npirs.org/ppis).

^bDifferent salts are commonly used that usually refer to the same toxicity and ecotoxicity effect test data for glyphosate, which is hence used as reference substance.

^cDetails provided in SI (Section S-1).

2.3. LCA-Based Impact Performance Assessment

Instead of performing a comparison of the substitution scenarios per defined target pest involving the various indicators and aspects laid out in substitution, alternatives assessment, and SSbD approaches, , we structured the assessment along the main LCA phaseslife cycle inventory (LCI) analysis and life cycle impact assessment (LCIA) , to keep the case study focused on adopting learnings from LCA in a chemical substitution context. The overall assessment workflow is outlined in Figure .

1.

Overall workflow for combining two distinct sets of functionally equivalent pest control substitution scenarios (a) with life cycle assessment (LCA) inventory (b) and quantitative impact assessment (c) to yield ranked scenarios per substitution context (d). *Common machinery: machinery (including fuel consumption, tire abrasion) that all scenarios have in common and that is hence excluded from the analyzed system. Net-added machinery: additional machinery used for mechanical pest control as compared to pesticide use due to differences in machinery utilization and lifetime.

As a functional basis, we compared scenarios at the level of pest control option (instead of comparisons at the level of individual chemical) to allow considering the full life cycle of potential alternatives and include options at the process rather than active ingredient level. , As a functional unit (FU) for comparing pest control scenarios, we hence used ‘1 ha of treated crop area’, where treatment refers to effectively controlling the defined target pest in the specified crop and geographical region.

Three main life cycle aspects were considered in the product system modeling (see Figure ), namely, resource use and emissions along the pesticide-based or mechanical pest control supply chains including related waste, field operations at the actual crop target site, and waste related to field operations. Scenarios of both pest control options were modeled in the LCA software SimaPro (version 9.6.0.1, simapro.com) using ecoinvent (version 3.10) as background inventory database for supply chain and waste operations. Attributional modeling was applied to derive life cycle emission flows using specific or market average data and allocation factors for coproduct allocation. Some pest control options target more than one pest. However, single-species control is often applied for our target pests, not requiring system expansion to address multifunctionality.

To model field operation (i.e., foreground) emissions, recommended product application rates and pesticide content were used (see Table and Table ) to derive pesticide application amount per hectare treated (Supporting Information, SI, Table S1) and to define 200 kg of related, mainly incinerated solid waste (i.e., pesticide container) per tonne of applied pesticide. For mechanical weeding, we only considered the manufacturing of excess machinery compared to pesticide application. Additional LCI modeling details, unit process descriptions, and a list of LCI flows are provided in SI (Section S-1).

For characterizing supply chain and waste (including field operations waste) flows in terms of environmental impacts, we selected the global state-of-the-art LCIA method LC-Impact, considering all included impact categories across three areas of protection (AoP)human health, ecosystem quality and natural resources (see Figure ). For quantifying human toxicity and ecotoxicity impacts associated with field operations, we combined the PestLCI Consensus model to estimate initial environmental emission fractions of field-applied pesticides − with the widely used impact model USEtox (consistent with the version used in LC-Impact) to estimate human toxicity and ecotoxicity impact potentials. ,, We included health impacts from exposure to pesticide residues in treated crops, often dominating human toxicity impacts from pesticide use. − Additional details on deriving impact characterization factors and a list of all included characterization factors are provided in the SI (Section S-2).

2.4. Performance Ranking and Sensitivity Analysis

Each LCI flow was multiplied with its respective LCIA impact characterization factor and summed to pest control scenario-specific impact scores for each impact category (e.g., climate change or ecotoxicity). Category-specific impacts were translated into damage per AoP, yielding the same unit across results from all contributing impact categories, such as population lifetime loss expressed in disability-adjusted life years (DALY) for human health. , With that, pest control scenarios for each target pest were ultimately compared at the level of summed damage per AoP and at the level of an overall single score across AoP, based on normalizing and weighing the three AoP following global LCIA recommendations.

Four general assumptions and choices were tested in a sensitivity analysis. First, we added the biopesticide product Armicarb (contains potassium bicarbonate) to the laminarin scenario, since the latter was found insufficiently effective on its own in some applications. Second, we tested alternative unit processes to derive pelargonic acid, for which no specific process is available in ecoinvent. Third, long-term supply chain emissions of toxic chemicals from landfills were excluded, dominating overall ecotoxicity impacts while currently facing methodological challenges in accounting for leaching dynamics, solid-phase precipitation, and allocation of the correct emission compartment in USEtox. , Fourth, we used ReCiPe 2016 (End point (H) version 1.09) as alternative LCIA method to characterize environmental impacts. Additional details for deriving impact scores and ranked single scores as well as on the sensitivity analysis are provided in the SI (Section S-3).

3. Results and Discussion

3.1. Boundary Conditions for LCA and Substitution

When performing an LCA on the selected base case pest control scenarios and their potentially viable alternatives, we complied with several boundary conditions, some of which are also relevant in other comparative assessment contextsmost importantly in the context of chemical substitution, alternatives assessment, and SSbD when used for comparing options. Our LCA case study had a specific focus on comparing relatively simple product systems (i.e., selected pest control systems related to a single agricultural pest) as compared to studies focusing on, for example, comparing entire pest control strategies, agricultural production systems, or diets. ,− Hence, we might miss boundary conditions that specifically apply to larger-scale systems or other study scopes (e.g., performing a hotspot analysis as compared to a system comparison). However, due to considering all life cycle stages in our case study, combining different characterization methods, aggregating impacts at the level of area of protection, and applying normalization and weighting ultimately allowing us to rank our pest control scenarios via a single score, we capture many boundary conditions that also apply to other LCA studies, while keeping the context strictly on comparing different systems as relevant for chemical substitution.

Several of the boundary conditions that we face in LCA studies are highly relevant also for different chemical substitution contexts and are related to both a study’s application and/or use context, different assessment aspects, and interpretation of study results for decision support. More specifically, some fundamental aspects of LCA also apply largely to chemical substitution. This includes specifying a clear study goal and scope, comparing alternatives based on a common function, considering different impact aspects and life cycle stages to avoid burden shifting and evaluate relevant trade-offs, and allowing comparisons at different scales. Learnings from addressing these aspects in LCA hence show a high potential for informing the development of chemical substitution methods and frameworks. Other conditions, however, are fundamentally relevant for chemical substitution but show low potential to apply learnings from LCA. These include, for example, systematically identifying a wider range of potential alternatives for each substitution context and applying rapid-screening methods in line with short industrial innovation cycles. Considering alternatives at different levels (e.g., chemical, material, technology) requires a systematic mapping of chemical, material, and product functions to wider societal services and to define the appropriate level for a function-based substitution. Input for developing rapid-screening methods is to develop criteria for systematically defining those impact categories to be considered for a given substitution and use context. Overall, LCA has undergone a long process to address various conditions relevant for a comparative, quantitative, and scalable assessment context, which offers a broad range of learnings that can inspire the further advancement of substitution methods. A full list of assessment boundary conditions, and details regarding their relevance for LCA as well as different substitution contexts, along with their respective synergy potentials indicating to what extent learnings from LCA can be transferred to inform the further development of methods for chemical substitution, alternatives assessment, and comparative SSbD is provided in Table .

3. Overview of Boundary Conditions for Comparative Assessments of Life Cycle Environmental Impacts of Different Functionally Equivalent Solutions, Their Relevance for Our Illustrative Pest Control Case Study and for Chemical Substitution Contexts, and Resulting Synergy Potentials Indicating the Level of Transferability of Learnings in Life Cycle Assessment (LCA) to Chemical Substitution Assessments for Each Boundary Condition.

Boundary condition	Relevance for LCA case study	Relevance for chemical substitution context	Synergy potential
A – Application or use context
A1 – Define a specific goal and scope of what should be compared	One of the main phases of LCA is the goal and scope definition to allow for considering decision context-specific aspects, such as intended use of the results, target audience, system boundaries and data representativeness	A ‘scope and problem formulation’ step is emphasized to be crucial for different chemical substitution contexts, while common criteria are currently not well-defined across decision contexts and frameworks	High
A2 – Identify potentially viable alternatives or scenarios	LCA usually starts from predefined scenarios or alternatives and does not require any systematic mapping or identification of potential alternatives to a given reference system in comparative decision contexts	Identifying a variety of (different types of) potential substitutes is the starting point for assessing alternatives to e.g. hazardous chemicals, but currently lacks methods to identify possible alternatives based on a set of agreed criteria that also consider context-specific aspects	Low
A3 – Allow to compare systems or products based on a common function	Definition of a common ‘functional unit’ to scale environmental performance of different systems across their life cycles is a fundamental characteristic of LCA	Knowledge on the chemical function in an application is fundamental; however, although desirable as a concept, , systematically defining a common function across alternatives beyond drop-in chemicals is currently not well established or widely used in substitution assessments	High
A4 – Map functions across scales to allow considering different types of alternatives	LCA mostly defines a single, systems-level functional unit to compare (a very limited number of) options or systems, but has over the years accumulated defined functions across different levels from process- to technology-level	Allowing to go beyond drop-in chemical-by-chemical solutions and consider material-, product-, process-, technology- or behavioral-level alternatives requires to map functions across these levels, , while related methods are currently broadly lacking	Medium
A5 – Consider the full life cycle to avoid burden shifting	LCA covers the entire life cycle of a given system from the extraction of natural resources to end-of-life waste handing to adequately address shifted burden from one life cycle stage to another, although specific studies may not cover all stages (e.g., cradle-to-factory gate studies)	In different frameworks relevant for chemical substitution (e.g., alternatives assessment and SSbD), considering life cycle impacts is an important component to avoid improved performance on one aspect at the expense of another, which can lead to ‘regrettable substitutions’ ,,	High
B – Assessment aspects
B1 – Assess all relevant impacts for comprehensive comparisons	LCA aims to comprehensively consider all environmental impacts on human health, ecosystem quality and natural resources that could be relevant for an assessed system, while many practical studies focus on a limited number or even single impact category (e.g., climate change)	While the wider set of impacts might not be a general priority in chemical substitution, certain sets of impact categories will be relevant for specific contexts (e.g., toxicity-related impacts for bioactive alternatives), with systematic methods to define which impacts to assess for what context still lacking	High
B2 – Evaluate trade-offs across impacts and life cycle stages	Trade-offs are a common aspect addressed and discussed in LCA studies, and can occur across life cycle stages, impact categories and regions	Trade-offs are potentially relevant in all substitution assessments considering multiple evaluation aspects, but are often only qualitatively considered across studies and approaches and mostly focus only on toxicity-related aspects ,	High
B3 – Align metrics and units to facilitate aggregation and comparison	Facilitating contribution analysis and comparison of results at different levels supports different decision elements and is an integral part of LCA studies, with known challenges of harmonizing metrics and units across methods	Consistent units and metrics across evaluated aspects will allow for aggregation and improved comparison across alternatives, while relevant methods are currently rarely adopted, either due to more qualitative approaches or aligning assessment steps with tools from different fields ,	High
B4 – Use central or representative assessment input data and assumptions	LCA aims at representing and comparing real-world practices and, hence, relies on ‘best estimates’ for data and assumptions representing such practices	Chemical substitution aims at representing real-world (current or future) practices; where it requires a comparative perspective (e.g., across alternatives), central or otherwise representative assumptions reflecting such practices should be adopted, while considering conservative assumptions as benchmarks where relevant	Medium
B5 – Apply streamlined methods for rapid-screening assessment	LCA is time-consuming for practitioners, with most time spent on system modeling and life cycle inventory analysis, despite LCA software tools that come with plenty of prebuilt functions and data	Rapid-screening methods are essential for chemical substitution approaches to comply with rapid innovation cycles in industry and allow comparison of wider ranges of distinct alternatives, while such methods currently are mainly qualitative or time-consuming, e.g. for human exposure assessment ,	Low
C – Interpretation for decision support
C1 – Compare alternatives against an internal reference	When applied in a comparative context (as compared to a hotspot analysis), the main strength of LCA is to compare the environmental performance of functionally equivalent systems with respect to multiple aspects (e.g., life cycle stages, impact categories)	Defining and comparing alternatives against an internal reference (e.g., a ‘chemical of concern’) is at the core of chemical substitution approaches that focus on comparisons across two or more functionally equivalent options, while comparative best-in-class assessments might not rely on a defined ‘reference’ option ,,	High
C2 – Chose appropriate aggregation level(s) to inform decisions	LCA facilitates reporting results at midpoint (i.e., physical metrics), damage (i.e., environmental issue of concern) and single-score levels, depending on the study scope and decision context	While different assessment aspects are often grouped (e.g., hazard aspects), results are commonly not aggregated in chemical substitution approaches, while sometimes adopting multicriteria decision analysis (MCDA) to facilitate aggregation with studies applying different underlying approaches ,	Medium
C3 – Adopt quantitative metrics to inform decisions at different scales	LCA is a quantitative sustainability assessment tool that allows for scaling functions and environmental performance results, for example according to the level of technological readiness of the studied system(s)	Chemical substitution approaches are currently rarely scalable beyond hazard aspects, partly due to the fact that mainly qualitative or semiquantitative metrics are used, with the justification of fulfilling conditions for a rapid-screening assessment ,	High
C4 – Use external benchmarks for performance evaluation	Methods are emerging to compare environmental performance of systems in LCA not only against each other but also against external, biophysical targets or conditions, but are currently ill-developed for toxic chemicals ,	External benchmarks or targets are currently not adopted in chemical substitution approaches, but discussed in the context of SSbD frameworks, and emphasized to be specifically relevant for chemical pollution as a key aspect of the substitution assessments that aim to measure their contribution to addressing reduction targets ,,	Medium

3.2. bEnvironmental Impact of Pest Control Substitution Scenarios

Life cycle impact performance results from our LCA study across pest control scenarios and impact categories are shown in Figure for preventive plant pathogen control scenarios and in Figure for curative weed control scenarios, disaggregated by main life cycle stages, namely, supply chain operations including resource extraction and manufacturing waste, field operations (i.e., pesticide application or mechanical weeding at the actual crop field), and field operations-related waste. This allows for effectively comparing performance across impact categories per area of protection (i.e., human health, ecosystem quality and natural resources) based on converting impact scores into a common unit (e.g., DALY/ha for damage on human health for all contributing impact categories) and identifying main contributing life cycle stages.

2.

Environmental impact profiles of substitution scenarios for preventive plant pathogen control across impact categories for the areas of protection ‘human health’ (population-level disability-adjusted life years per hectare treated, DALY/ha), ‘ecosystem quality’ (potentially disappeared fraction (PDF) of species over one year per hectare treated, PDF-yr/ha), and ‘natural resources’ (kilogram of ore resourced per hectare treated, kg/ha). Considered life cycle stages are supply chain operations including resource extraction and manufacturing waste (SC), field operations (Farm), and field operations-related waste (FW). Red columns represent impact category-specific performance of base case scenarios, and blue shaded columns represent performance of potentially suitable alternatives. Aggregated results across all life cycle stages are found in the SI (Figure S3).

3.

Environmental impact profiles of substitution scenarios for curative weed control across impact categories for the areas of protection ‘human health’ (population-level disability-adjusted life years per hectare treated, DALY/ha), ‘ecosystem quality’ (potentially disappeared fraction (PDF) of species over one year per hectare treated, PDF-yr/ha), and ‘natural resources’ (kilogram of ore resourced per hectare treated, kg/ha). Considered life cycle stages are supply chain operations including resource extraction and manufacturing waste (SC), field operations (Farm), and field operations-related waste (FW). Red columns represent impact category-specific performance of base case scenarios and blue shaded columns represent performance of potentially suitable alternatives. Aggregated results across all life cycle stages are found in SI ().

Comparing system level impact scores for the mix of chemical fungicides folpet + mandipropamid and their potential alternatives (chemical fungicide oxathiapiprolin, metal-based fungicide copper­(II) hydroxide, and biofungicide laminarin) to control downy mildew in European grape/vine production reveals various insights (Figure ). Across some categories contributing to human health (e.g., climate change for manufacturing and waste processes and noncancer toxicity impacts for field operations), the fungicide mix (base case) shows somewhat worse performance than its alternatives (typically within a factor 2–8). For other categories, the metal-based fungicide scenario performs worst, mainly manufacturing-related noncancer toxicity, freshwater ecotoxicity, and mineral resources scarcity as well as freshwater ecotoxicity from field operations (the latter being a factor >370 higher than the base case). Overall, the single chemical fungicide and the biofungicide scenarios perform best across most impact categories and life cycle stages, with slightly better performance of the former by about a factor of 5. Climate change impacts areas expectedhighest for base case manufacturing processes, since two distinct active ingredients are considered, which requires more energy for synthesizing two chemicals. Results stay consistent with these observations when scores across all life cycle stages per category are aggregated (SI, Figure S3a). The base case shows the highest human toxicity impact (being at least a factor of 10 higher for cancer and a factor of 70 for noncancer as compared to other options), while the metal-based fungicide has highest impacts for freshwater ecotoxicity (more than 1 order of magnitude worse than other options) and mineral resource scarcity (at least a factor of 30 worse than other options). In contrast, the chemical fungicide performs best across all impact categories except human noncancer toxicity, marine eutrophication, and water stress, where it performs very similar (i.e., within a factor of 2) as the biobased option, with the biofungicide ranking second. These results emphasize that both chemical and biobased fungicides can potentially be viable alternatives to improve overall environmental performance of specific pest control scenarios, with biggest improvement potential for human toxicity, particulate matter formation, and climate change driving human health impacts and ecotoxicity, acidification, and climate change driving ecosystem quality impacts. Furthermore, our results show that copper-based fungicides perform worse than the base case especially for ecotoxicity (as a common focus area to reduce pesticide-related impacts) and mineral resource scarcity, in line with such fungicides being labeled highly hazardous in international lists despite their wide use in organic farming.

Comparing system level impact scores for the chemical herbicide glyphosate (base case) and its potential alternatives (bioherbicide pelargonic acid and mechanical weeding) to control common waterhemp in U.S. soybean production yields a somewhat similar picture (Figure ). Base case impacts are somewhat higher than alternatives impacts for some categories contributing to ecosystem quality (e.g., freshwater eutrophication for manufacturing operations and ecotoxicity from field operations), for mineral resource scarcity impacts, and for all categories related to waste from field operations (on average, by a factor of 2–6). Mechanical weeding performs a factor 2.5 better than the base case for mineral resource scarcity from manufacturing processes and is with an average factor of 2 difference very similar to base case impacts, while showing slightly higher ozone formation, human toxicity cancer, and terrestrial ecotoxicity impacts, respectively, driven by nitrogen oxide (87%), hexavalent chromium (98%), and cobalt (85%) emissions from supply chain processes, such as coking. The bioherbicide performs best across categories and life cycle stages by a factor of 15 on average. The exception is human noncancer toxicity for field operations, where bioherbicide impacts are at least a factor of 25 higher than other options, driven by consumption of residues in harvested crop components. Although mechanical weeding does not contribute to field operations and related waste impacts, the overall picture remains the same when aggregating life cycle stages per impact category (SI, Figure S3b), with base case and mechanical weeding performing similar and the bioherbicide performing overall best. Our results emphasize that bioherbicides can be a promising option to improve overall environmental pest control performance, while certain trade-offs might arise, such as increased noncancer toxicity potential from field operations. Furthermore, replacing chemical herbicides by mechanical weeding may not necessarily lead to significant improvements due to trade-offs in environmental supply chain impacts. In our example, shifting to mechanical weeding reduces field-level impacts from glyphosate, while requiring additional machinery use per unit area (see SI, Section S-1), involving toxic emissions (elsewhere) from related energy use and other processes. However, when comparing such technology shifts, it is important to stress that mechanical weeding also comes with several benefits related to field-level soil health promoting crop growth and yield and enhancing ecosystem multifunctionality as compared to chemical pest control. , Moreover, toxicity impacts of glyphosate are widely discussed, yet still show considerable uncertainties, and do not consider toxicity impacts associated with coformulants, leading to underestimations of related impacts. Such aspects influence the comparison of scenarios and need to be considered by, for example, adapting or expanding the function that is used to compare these pest control options. In our example, this would considerably improve the overall performance of mechanical weeding in a practical substitution context as compared with glyphosate application.

3.3. Performance Ranking of Pest Control Substitution Scenarios

Comparing pest control scenarios at an overall system level requires to look beyond individual impact categories and life cycle stages and to bring all results to a level where they can be aggregated into a score expressed in a common unit. Such scores can then be used to rank scenarios, ultimately informing which identified potential alternatives may be feasible substitutes to the respectively defined pest control base case. Environmental performance ranks of pest control scenarios, their underlying scores, and contributions of impact categories and main life cycle stages are show in Figure .

4.

Environmental performance of substitution scenarios for preventive plant pathogen control (left) and curative weed control (right) in terms of their (a) overall ranks (lower value = better performance), (b) underlying aggregated single-score impacts (person-year equivalents per hectare treated), and (c) contribution of areas of protection (top), main life cycle stages (middle), and aggregated impact categories (bottom, e.g., climate change is aggregated across human health and ecosystem impacts) to single scores.

For preventive control of downy mildew in European grape/vine fields, the chemical fungicide oxathiapiprolin scenario ranks overall highest (i.e., best environmental life cycle performance), closely followed by the biofungicide laminarin (∼3-fold worse) and then by the chemical mix (base case) scenarios (∼16-fold worse), with the metal-based fungicide scenario ranking lowest (>700-fold worse) (Figure a,b left). This order would be even more pronounced when considering the difference in maximum treatments per crop cycle, which is n = 2 for the oxathiapiprolin scenario and n = 4 for all other scenarios (see Table ). Across scenarios, ecosystem quality impacts driver overall scores, closely followed by human health impacts (both very similar for chemical fungicide scenarios), with climate change, human toxicity, and ecotoxicity impacts as main contributors, largely from field operations, especially for the metal-based fungicide scenario, while supply chain impacts dominate for the biofungicide scenario (Figure b,c left). These results illustrate that overall biobased and chemical fungicides can be both relevant for substituting highly hazardous scenarios (given the same magnitude of the two best-performing scenarios), emphasizing the importance of a toxicity-oriented focus to improve environmental performance and acknowledging that copper-based pest control might not be a viable alternative even to some fungicide mixes.

For curative control of common waterhemp in U.S. soybean fields, the bioherbicide pelargonic acid scenario ranks overall slightly higher than the base case (glyphosate) scenario (∼6-fold difference), with mechanical weeding in between both (Figure a,b right). Given how close overall performance scores are across scenarios, none of them is different enough to be considered a viable alternative without further analysis. However, while the bioherbicide scenario is marginally worse than the base case for human health (∼3-fold), it outperforms the base case by ∼17-fold for ecosystem quality and ∼80-fold for natural resources. Furthermore, we assumed a single treatment per crop cycle (corresponds to the maximum treatments for glyphosate, see Table ). In cases, where alternative scenarios allow for additional treatments, the results would shift accordingly in favor of the base case. Human health and ecosystem quality again drive overall scores, with climate change and ecotoxicity from manufacturing operations as dominating contributors for the base case and mechanical weeding and with human toxicity but also climate change from manufacturing and ecotoxicity from field operations dominating for the bioherbicide scenario (Figure b,c right). These results demonstrate on the one hand that viable alternatives cannot easily be identified when performance scores are very similar, while on the other hand, they highlight that performance might differ substantially across scenarios for specific impact categories or life cycle stages, which can inform specific impact reduction measures even where overall scores are similar. Given that glyphosate as the base case scenario is applied to a wide variety of crops across the world, chemical substitution might require to look beyond a single crop, while also accounting for increased pest resistance-building potential especially for widely applied pesticides, while also addressing currently high uncertainty in toxicity-related impacts of glyphosate (and other pesticides) as well as considering possible benefits from technologies like mechanical weed control.

3.4. Sensitivity Analysis of Alternative Pest Control Solutions

The robustness of our results was tested and summarized as a difference in overall impact performance scores between scenarios where specific assumptions and choices were changed, and the reference scenario was used in our actual results as presented in Figure . Adding potassium bicarbonate to increase efficacy of laminarin would increase impacts of this biofungicide scenario by almost a factor of 10, driven by ecosystem quality and natural resources (Figure a), which would render it of similar performance as the related base case. This demonstrates that pest efficacy is an aspect that may vary across scenarios and hence should be explicitly included when comparing pest control options to correctly scale overall performance. Testing different unit processes to derive pelargonic acid (i.e., starting from different feedstocks) ranges from slightly improved performance of 19% (sunflower oil) to decreased performance of 140% (coconut oil), with impact changes driven by ecosystem quality and natural resources (Figure b). This indicates that feedstock choice can matter, while supporting the best-in-class rank of pelargonic acid. Excluding long-term toxic chemical emissions from landfills across scenarios leads to an expected decrease in overall impacts between 12% (pelargonic acid) and 79% (laminarin), driven by reduced ecotoxicity of different metal emissions contributing to ecosystem quality (Figure c). While not influencing the scenario rankings in our present study, this result emphasizes the need for improving the fate modeling of metals in USEtox and other multimedia models. ,, Finally, using an alternative LCIA method (i.e., using a different set of impact characterization factors) across scenarios yields slight impact reductions between 29% (pelargonic acid) and 88% (mechanical weeding) as the net result of reduced impacts on human health and ecosystem quality and increased impacts on natural resources that contributes less to overall scores (Figure d), while not changing our scenarios rankings. Overall, our results and rankings are robust, while offering insights for relevant components to consider in the future (e.g., pest efficacy) and for improving specific assessment components (e.g., environmental fate of metals) even in scenarios where they are not directly applied to agricultural fields as pesticides but part of the wider set of manufacturing- and waste-related emissions.

5.

Difference in single score performance of various scenario adaptations as compared to our reference scenarios: (a) adding another biofungicide product to laminarin, (b) testing three alternative biofeedstocks to derive pelargonic acid, (c) excluding impacts from long-term emissions, and (d) using an alternative life cycle impact assessment (LCIA) method to derive impact characterization factors. Key: HH, human health; EQ, ecosystem quality; NR, natural resources; Cu­(II) hydroxide, copper­(II) hydroxide; Mech. weeding, mechanical weeding. (1) ReCiPe: For the area of protection ‘Natural resources’, only mineral resource scarcity was used in the sensitivity analysis to be consistent with the impact categories covered in LC-Impact used for our reference scenarios. However, ReCiPe also includes fossil resource scarcitywhen adding this impact category in (d), overall NR-related differences in single score results across pest control scenarios increase between a factor of 3.5 for copper­(II) hydroxide and a factor of 350 for folpet + mandipropamid scenarios

3.5. Addressing Remaining Challenges for Advancing Substitution

Discussing the different boundary conditions for LCA versus substitution frameworks and conducting an illustrative study on assessing environmental life cycle impacts across two sets of functionally equivalent pest control options has demonstrated that LCA holds great potential to inform the further development of approaches in chemical substitution, alternatives assessment, and comparative SSbD. However, despite many methodological and conceptual overlaps, the different chemical substitution frameworks have each a specific focus on informing decisions that require aspects and assumptions that are not or less relevant for other frameworks, including LCA. Furthermore, there are several real-world aspects that are relevant in a substitution context but not addressed by LCA. This includes, for example, environmental justice trade-offs or differences in vulnerability among populations (e.g., children vs adults) to chemical exposure, although first attempts exist to account for the latter in LCA. , Hence, the biggest challenges for advancing chemical substitution approaches are found where learnings from LCA (and other assessment tools) are low (see Table ). Among the biggest remaining challenges are to (i) identify potentially viable alternatives, (ii) define scalable functions as comparison basis, (iii) generate highly detailed supply chain data, (iv) consider real-world conditions, and (v) develop rapid-screening assessment methods.

(i) In our case study, we compared different types of chemistries (e.g., organic, mixed organic, metal-based) and a different technology (mechanical weeding). Identifying those alternatives currently requires a deep knowledge of the application context, efficacy, and other information that is not generally available to practitioners. Building systematic inventories of possible alternatives and developing prediction tools that allow for considering the wider space of chemicals but also a wider scope beyond drop-in chemistries are a viable starting point to address this challenge. (ii) Bridging chemical, material, product, technology, and societal functions to allow considering a wide range of fundamentally different alternatives might require a way to map functions across scales and define for a given context a function to which all alternatives can be scaledonly then, substitution will be fair as in based on functionally equivalent comparisons. (iii) Supply chain data in LCA are often too generic to discriminate especially drop-in chemicals from the same class but also to be specific enough for other chemistries. Efforts to generate chemical-specific supply chain inventories for chemical emissions are a possible way forward but do currently not yet cover a wide range of substances and production conditions. (iv) In our specific case study, we did not consider the effects of pest resistance, that several pest control methods target multiple pests, technical feasibility across farmers and regions, and other real-world aspects. Such aspects need to be worked in as part of the actual substitution assessment and may change from one stakeholder or region to another. (v) Substitution requires rapid-screening methods to align assessments with short industrial innovation cycles of some weeks to a few months. Assessing in such a time frame a variety of alternatives will have to draw much more systemically on robust digitalization and automation methods. , Despite these challenges, the field of chemical substitution advances rapidly with synergies to harvest learnings from LCA and other assessment fields.

3.6. Policy Implications and Next Steps

The purpose of the present study was not to recommend specific substitutes for two illustrative pest control targetsthis would require considering additional aspects (e.g., pest resistance-building), which we omitted for simplicity. Instead, the purpose of our study was to understand how methodological aspects in LCA can inform to advance methods for chemical substitution, alternatives assessment, and comparative SSbD. We specifically identified boundary conditions that apply to both LCA and different substitution contexts, rendering LCA to be one of the tools whose development can inspire and cross-fertilize the advancement of substitution methods, while acknowledging remaining challenges where LCA may be less informative. Our learnings from the way of LCA to define a functional basis for comparing alternatives, align metrics and units to facilitate results aggregation, and consider the full life cycle to identify areas of burden shifting and relevant trade-offs provide a complementary element to the current substitution and SSbD literature. These learnings can be considered in new methodological developments for maturing the fields of chemical substitution, alternatives assessment, and SSbD, and avoiding ‘regrettable substitution’. Our identified learnings from LCA will help overcome limitations in substitution and SSbD related to functional considerations, scalability of methods, quantification of impacts, and results aggregation and interpretation. As a first step, developers of substitution and SSbD approaches could consider relevant aspects from LCA and modify them to align with their respective boundary conditions, while ranking results could be systematically tested in practical substitution and SSbD case studies, where different LCA components are considered to get an understanding of the potential influence of these components on the final decisions. Furthermore, addressing the remaining challenges specific to the different substitution contexts where LCA is less helpful will be key to truly advance the field and develop approaches that are both fit-for-purpose and operational. With that, chemical substitution, alternatives assessment, and comparative SSbD will each become valuable tools and key enablers for moving toward safer and more sustainable alternatives in the chemical, pharmaceutical, and other industries, supporting the objectives of UNEP’s Global Framework on Chemicals and other policy instruments for a broader and much needed sustainability transition worldwide.

Methods and additional results for life cycle inventories, impact characterization, impact/single scores and sensitivity analysis The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.est.5c12521.

• Methods and additional results for life cycle inventories, impact characterization, impact/single scores and sensitivity analysis (PDF)

• Tables S2, S3, S4, S5, S8, S9, and S10 (XLSX)

The authors declare no competing financial interest.

Published as part of Environmental Science & Technology special issue “Nobel Symposium 2025: The Future of Chemical Safety and Sustainable Materials Chemistry”.