A Comprehensive Methodology to Support Decision-Making Towards Sustainability on Nanocomposite Materials in Additive Manufacturing Sector


 Innovative nanocomposite materials and resultant additive manufacturing products are necessary to be assessed for their carbon footprint towards top priorities of EU for plastics, including the European Green Deal principles and the Action Plan for Circular Economy. Life Cycle Assessment (LCA) is widely applied standardized methodology that aims to study potential environmental impacts of novel products. Nano-scale materials (NM) are usually dispersed in polymer to enhance their limited functional properties resulting in a spectrum of end-products for multiple applications. However, little information exists on their environmental impact. Within this context, this study presents a ‘cradle-to-gate plus end-of-life’ LCA approach, studying different types of 3D printing nanocomposite filaments across the supply chain. Three different types of polymer matrixes were examined: polyamide (PA), polypropylene (PP) and polylactic acid (PLA), additivated with three different types of nanomaterial additives: multiwall carbon nanotubes (MWCNTs), graphene oxide (GO) forms and graphene nanoplatelets (GNPs), considering lab-scale production. In addition, several different EoL scenarios have been examined for the materials. Finally, LCA findings are coupled with the performance (taken here as conductivity) of these new materials to assist the decision-making process for selecting efficient scenarios with the least environmental impact. The outputs of this examination enable identification of potential sustainability issues for novel nanocomposite materials at an early design stage, while also assisting in the definition of actions to mitigate such issues. Thus, LCA studies can generate knowledge on the environmental impacts of nano-enabled materials, while also serving as a valuable decision support tool towards optimizing material sustainability aspects.

Using LCA technique, the GHG emissions of CNT manufacturing production process was calculated, and the relevant hotspots were found supporting technological improvements.
Findings highlighted the signi cant impact that con guration adjustments display in the reduction of GWP, within setup parameters that include oxidative additive type (CO 2 or H 2 O), reactor growth modes (2D at-plate or 3D spherical), technique for catalyst deposition (sputtering or CVD), and purging gases selection (Ar or N 2 ). The highest potential and promise for achieving industrial scale of production of CNTs are considered to be displayed in continuous processes, such as uidized bed CVD, as discussed by Healy et al. (Healy ML, Dahlben LJ, Isaacs JA (2008) ), in a study that evaluated an unspeci ed catalytic CVD method. For Graphene, the synthesis technologies of which are still developing, there are limited literature-based information available on the study of potential environmental impacts within the context of its synthesis methods. The most widely applied methods for Graphene synthesis are exfoliation, CVD and epitaxial growth (Sivudu KS, Mahajan Y (2012) ).
In a review study conducted by Arvidsson (Arvidsson R (2017) ) concerning the environmental life cycle assessment evaluation of different graphene production routes, ve different routes were identi ed and taken into consideration: CVD, exfoliation (ultrasonication or thermal), chemical reduction of GO, and epitaxial growth. Signi cant challenges emerged in direct comparative analysis of the results, seeing that the functional unit considered in the reviewed studies was not the same. In another study, performed by Cossuta et al. (Cossutta M, McKechnie J and Pickering SJ (2017) ) a comparative cradle-to-gate LCA evaluation for three graphene production routes was conducted: CVD, electrochemical exfoliation, and chemical oxidation (followed by chemical or thermal reduction). The synthesis route displaying the lowest impacts for large quantity production of rGO was de ned as the chemical oxidation process followed by thermal reduction within the LCA study. The de nition of this synthesis route as the least impactful was con rmed by a complementary prospective LCA evaluation, in which impacts of a potential commercial scale were estimated.
Serrano-Lujan et al. (Serrano-Luján L, Víctor-Román S, Toledo C et al (2019)) studied the environmental impacts of the two techniques considered to present the highest e ciency towards producing rGO of high-performance properties, namely the Hummers and Marcano methods, studying in total, seven rGO production routes. Two functional units were proposed: In order to enable a direct comparative analysis of production routes, a functional unit of 1 kg of rGO was used, while a functional unit normalized by conductivity enabled an analysis speci c to the application studied. Hummer's method resulted in a decreased total energy consumption per kilogram of graphene produced. Pizza et al. (Pizza A, Renaud M, Mehrdad H, Jean-Louis B (2014) ) studied the LCA of high-quality nanocomposites made of thermally conductive GNPs, placing the focus on the study of the energy requirements for the stages of the nanomaterial production and nanocomposite manufacturing and excluding from the analysis the potential nanoparticle emissions and nano-waste generation. Based on the ndings, the fabrication of GNP ller was found to be an energy-intensive procedure (1,879 MJ/kg), and thermal conductivity values of 1 W/mK are achieved by incorporating 5.8% wt% ller in the manufacturing process for 1 kg of epoxy composite. Currently, in terms of EoL scenarios for plastic waste generated in Europe, it has been presented that 75% is recycled, considering 32% recycling & 43% energy recovery and the 25% is land lled (Crippa M, De Wilde B, Koopmans R, Leyssens J, Muncke J, Ritschkoff A-C, Van Doorsselaer K, Velis C, & Wagner M (2019)). Therefore, it is reasonable to consider the potential for nanocomposite recycling when designing the production of nano-enabled products. The market concerning 3D printing lament feedstock presents considerable growth potential, as it is projected to display a compound annual growth rate (CAGR) of 28. Oxide), By Region, And Segment Forecasts, 2020 -2027), respectively, up to 2027. Based on these market growth rates, and following the trend of nanomaterial incorporation for introduction of unique properties to the 3D printing lament materials, it can be projected that production volumes of nano-enabled laments will increase and the prospect of these products ending up in land lls and incineration plants should be avoided.
In this regard, LCA plays a critical role in the evaluation of the potential impacts of these new technologies and the different EoL treatment management options; in this manner LCA assists in the direction of research and innovation activities aimed at achieving environmentally compatible, sustainable products that could comply to the circular economy framework (Arvidsson R, Tillman AM, Sandén BA, Janssen M, Nordelöf A, Kushnir D, Molander S (2018) ).
The current study examines the environmental performance via LCA of multiple nanocomposite laments through melt-compounding (extrusion) suitable for FFF investigating three different thermoplastic matrices (PA, PP and PLA), three different carbon-based NMs (CNT, GNP and rGO) and alternative EoL treatment options (current and future projections EoL treatments). Moreover, the environmental impact of llers' loadings on different matrices has been coupled with the nanocomposite performance, taken as electrical conductivity, to assist the decision-making process by choosing the compounding with the required performance and the least environmental impact. As a result, the purpose of this study is to highlight key open di culties that must be solved with a solid holistic approach in order to support sustainable nano-enabled products in accordance with the European Green Deal and the circular economy era that it promotes.

Methodology
The current attributional LCA study employs the general framework provided in

Goal and scope
The goal of the study is to evaluate a variety of nano-composite laments in terms of their environmental impact and use the respective LCA results to strengthen decision-making in order to achieve an e cient and sustainable lament production. Three thermoplastics PA, PP and PLA in combination with different concentration of nano-llers (0.5-10wt%): multiwall carbon nanotubes (MWCNTs) synthesized by chemical vapour deposition (CVD), graphene oxide (GO) via Hummer's method and graphene nanoplatelets (GNPs) synthesized by graphite exfoliation method have been investigated. Since PLA, PA and PP are the most common used polymers in the AM sector and the most widely used polymers in 3D printing fabrication, they have been selected to be examined as the matrices, of the nano-composites laments, in combination with the most-used NM llers for the respective applications. Table 1 presents the different combinations-scenarios, among the polymeric matrixes, the ller types and their respective loadings in wt% of each ller as well as the EoL options that has been considered in the LCA study. For this study a 'cradle-to-gate plus EoL' approach has been considered. Thus, the system boundaries of this study include Materials: the production of raw materials (NMs and thermoplastics), Manufacturing: the melt-compounding (extrusion) process for the lament die manufacturing and the EoL management where different scenarios were examined; the use stage has been excluding ( Figure 1). As EoL scenarios incineration, land ll and recycling alternatives have been considered. The functional unit (FU) for all the assessed scenarios is mass-based at 1 kg of nanocomposite lament produced by melt-compounding. Concerning NMs they were modelled considering lab scale production and the following process routes per case: GNP: Expanded, exfoliated graphite pulverized, GO: Hummer's method, mild bath sonication and CNTs: Fluidized bed CVD, N 2 /20 bead cycles, lab scale production. by data from open access literature as detailed in sections below. In terms of electricity mix, the background data was updated for EU requirements. The present study did not take distribution and transportation into account. For the type of PA used in composites, nylon 6-6 was selected.

Production of Nanomaterials
CNT production process routes can vary in terms of their environmental impact in a product life cycle perspective; for the present LCA study a 'high quality' CNT synthesis methodology has been selected for the data inventory's development. Teah et. Al. (Teah HY, Sato T, Namiki K, Asaka M, Feng K, Noda S (2020) ) gave the high environmental impact of on-substrate CVD method, indicating its higher compatibility towards a smaller scale, in contrast with the uidized-bed methods of CVD, which offer increased suitability for larger scale applications or bulk uses. In this study, the electricity generation was based on Japan data, and the background system data were acquired from the Ecoinvent 3.4 database. Therefore, data were regenerated in ecoinvent 3.6.
Based on the step-based process that has been presented above, additional LCI information is provided.
Taking into consideration that CVD-based methods have been de ned as commercially viable technologies due to cost, uncomplicated operation, as well as ease of scalability, CVD has been selected as the most representative technology (Upadhyayula VKK, Meyer DE, Curran MA, Gonzalez MA (2012)).
Considering the limitations of the present literature that has evaluated the environmental aspects of CNTs, within the context of this study, the CVD method is de ned as the most promising technology for carbon nanotube synthesis, since substantial variations are seen in environmental impacts based on CNT synthesis method. The de nition of CVD as a synthesis method is coupled with the best-case scenario adapted from (Teah HY, Sato T, Namiki K, Asaka M, Feng K, Noda S (2020) ) for the laboratory scale manufacturing of CNTs, as well as the application of N 2 as a carrier gas and 20 times of substrates cycle.
The inventories of the Graphene oxide (GO) and the Graphene Nanoplatelet (GNP) production process have been developed based on data received by (Serrano-Luján L, Víctor-Román S, Toledo C et al (2019)) and (Pizza A, Renaud M, Mehrdad H, Jean-Louis B (2014) ), respectively.

Manufacture of the composite lament
Plastic lm extrusion as in ecoinvent 3.6, modi ed to include only the process (extrusion) resources was used as proxy to model the melt-compounding process. The process's productivity maintained at 98%. reduce land lling to a 10% maximum, within the context of the Circular Economy Package. As a result, incineration treatment will cover the remaining 35% max. In Table 2 all the relevant details concerning the materials examined and their different EoL options modelling in Ecoinvent 3.6 are presented.

Results
Both Life Cycle Impact Assessment and interpretation of the LCA results are presented in this section including: i) the environmental impact of the assessed materials, ii) a sensitivity analysis illustrating the effect of ller % on the Climate Change (CC) of the nanocomposites, (iii) the effect of different EoL scenarios in LCA results and (iv) the coupling of the nanocomposites' conductivity performance into the LCA ndings.

Environmental impacts of thermoplastics and NMs
When considering only the polymer matrix 'from cradle-to-gate', PLA has the lowest value 0.93 kg CO 2 eq/kg PLA, followed by PP with 1.84 kg CO 2 eq/kg and PA with 8.03 kg CO 2 eq/kg. PLA's low CC value is attributable to biogenic carbon absorbed from the atmosphere, as opposed to the other two petroleumbased thermoplastics PA and PP, which are made from fossil fuels.
However, PP presents the lowest CC value 1.57 kg CO 2 eq/kg, followed by PLA 2.20 kg CO 2 eq/kg and PA 4.36 kg CO 2 eq/kg, when the EoL (on a 'cradle-to-grave' approach for 1 kg of polymer) was considered.
Once a cradle-to-grave approach is applied, the biogenic carbon balances out since the amount of CO 2 absorbed in PLA is released to the atmosphere by thermal treatment during the EoL stage. The difference in CC values for the examined polymers indicate the importance of considering the EoL stage into a LCA, especially when accounting for biobased products.
When assessing only the llers in a 'cradle-to-gate' approach, CC shows the least impactful NM is GNP releasing 0.074 g CO 2 eq/g followed by CNT 0.246 kg CO 2 eq/g and GO 0.293 kg CO 2 eq/g. The selected nanomaterials' production pathways are energy-intensive procedures; hence electricity is the main contributor to the CC values. Results from all sixteen impact categories assessed are presented in Tables S1 (CNTs, GNPs and GO) and S2 (PA, PP and PLA) of the supplementary information.

Environmental impacts of nanocomposite laments
The Climate Change impacts of the examined nanocomposite laments at ller concentrations from 0.5% -10% are illustrated in Figure 2. In the Table S3 of the supplementary information, each scenario's speci c values are presented. Increase ller concentration -independent of the nano-ller type, results in higher environmental impacts. However, the CC impacts of thermoplastics reinforced with GNPs, as expected from the results in section 2.1, performs better in terms of environmental impact in CC category than the respective nanocomposites with CNTs and GO inclusions. (PA-CNT system), 8.51 (PP-CNT system) and 7.28 (PLA-CNT system) times more CO 2 emissions than the respective 0.5 wt% scenarios. Therefore, it becomes obvious when CNT or GO are used as llers, they dominate the CC even at low amounts (2-3%). The same does not count when GNP is used as it has the least CC compared to CNT and GO. Hence, the polymer-GNP scenarios release signi cant lower CO 2 eq amounts even at high ller percentages (i.e., less than 11.3 kgCO 2 eq at 10wt% NMs).
For the nine different polymer-nanomaterial scenarios it is also observed a similar trend in the rest of the impact categories. Tables S4, S5 and S6 (in the supplementary) are presented results for the PLA, PP and PA, respectively. The sixteen impact categories assessed were normalized and results are shown in Figure  3 for the system 1 kg PA-GO 3 wt%. As illustrated in Figure 3, the categories with the most signi cant impact are the freshwater ecotoxicity, the human toxicity both cancer and non-cancer, and the ionizing radiation. The hot spot analysis indicates that for the examined 3 wt% ller, the GO production process dominates almost all impact categories.

Environmental impact of alternative EoL treatment
Regarding the EoL stage, different EoL management routes have been considered for the composite laments, as shown in Table 2. The environmental impacts of the nine products are shown normalized by the worst-case scenario (PA-GO) in Figure 4, while the absolute values in kgCO 2 eq. are presented in Table   3 for the greenhouse gas (GHG) emissions excluding biogenic carbon impact category. The FU was 1kg of nano-enabled polymer lament in all cases. The llers seem to increase the total impacts of the composite lament, even though the polymers play a signi cantly smaller role. This applies for the nanocomposite materials incorporating CNT and GO. On the other hand, in the event of GNP additivation, the contribution of the polymer has a considerably higher in uence in the total impacts.
A positive environmental impact appears for PA and PP due to the recycling share that was considered at the EoL stage. This alternative EoL treatment gives credits to the overall GWP by avoiding the production and use of virgin raw material. PA with GO inclusions is presented as the lament with the highest contribution in comparison with the alternatives with an impact of 13.21 kg CO 2 eq. in climate change impact category. The PA-GO baseline scenario was modeled accordingly: 55% Recycling, 35% Incineration and 10% Land ll. Due to the fact that the EoL management alternatives, that were assessed in this work, are based on future estimations, different EoL scenarios were also examined and compared with the worst case (worst performing lament) scenario which was the PA-GO as shown in Figure 5 (normalized results) while the respective absolute values in kgCO 2 eq. are presented in Table 4. Table S7 of the supplementary information illustrates the impact on climate change of the composite laments produced by PA and GO as a ller in various loadings (wt%) and EoL management pathways. In 100% recycling scenario, the PA-GO lament shows a reduction of 30% in CC value due to the credits gained from the increased avoided materials than the baseline. A negative effect is observed in CC values for the 100% Incineration and 100% land ll scenarios. A respective increase of 37% and 33% was calculated.
An e cient, standardized and sustainable waste management approach for nanomaterials is hindered due to nano-enabled products not possessing speci c-labelling as of yet, leading to challenges in sorting and separation of the nanomaterial-containing waste streams. Hence, according to a study performed by

LCA results coupled with performance to support decision-making
Ranking environmental impacts of composite laments at ascending order, as shown in Figure 6, allows to compare impacts and offers a range of appropriate options to the end-user towards low-emission materials.
The LCA results presented in this gure are useful information considering the performance in terms of composite lament's environmental impact. PP-GNPs 4 wt% appears to have lower CC impact in comparison to PLA-GNPs 0.5 wt%. This is not an expected result considering that PLA has an impact of 0.93 kgCO 2 eq/kg while the PP presents a higher impact around 1.84 kgCO 2 eq/kg. However, including the EoL stage into the assessment we observe that it plays an important role to the overall lament's impact indicating the PP based composite lament as the least environmentally friendly option. These values in combination with the respective conductivities (or any relevant performance indicators), could assist the decision makers towards a more sustainable option based on their application needs.
The critical aspect for the transition from an insulating polymer to conductor is that amount of the ller above which no major differences in electrical conductivity of the composite exist, known as electrical percolation threshold (EPT) (Rahaman M, Aldalbahi A, Govindasami P, Khanam NP, Bhandari S, Feng P, Altalhi T (2017)). The goal of producing conductive nanocomposite laments is to break through the EPT and the same time retaining 3D printing rheological and processing parameters. Various factors such as the ller type, quality, dispersion, aspect ratio have proved to affect the percolation threshold of In Figure 7a decision-making map is presented that could support a greener AM process concerning the production of PLA nanocomposite laments. Conductivity values and percolation threshold for the PLA composite laments have been provided by literature, as presented in Table 5 and Table S8 of the supporting information. As shown, for PLA the options that surpass the percolation criterion per ascending CC values are: PLA-GNP 7.5wt% < PLA-rGO 2wt% < PLA-CNT 5wt% < PLA-CNT 7.5wt%. The ller concentration and the type of ller are strongly in uence lament's performance and this is illustrated in the decision-making map. PLA-GNP 7.5wt% would be the best option in terms of performance and environmental impact despite the fact that is under the almost red area. However, even at low concentration of rGO the electrical percolation threshold is achieved while for CNT and GNP higher concentrations are required. Therefore, PLA-rGO 2 wt% would be the second most preferable option.
Relevant stakeholders such as manufacturers could set a threshold regarding performance (e.g., Conductivity) above which could select among different sustainable options (green area). Thresholds can be speci ed based on the speci c requirements of nanocomposites' applications. laments. In this way, the concept of consumer safety ( lament purchaser/user in this case) is also introduced in the approach.
Crucially, a set of additional aspects referring to the processability of the various laments represented in the life cycle scenarios could further embellish the presented decision support methodology. It has been reported in the literature that introduction of nanomaterials in the polymer laments could signi cantly impact material printability (Dey A, Eagle INR, Yodo N (2021) ). Therefore, cross evaluation of the printability aspects with the effectiveness and sustainability criteria would enable identi cation of the allround more viable and advantageous alternative.
The development and re nement of a complementary approach involving the above-described elements (LCA, safety, performance, processability), could facilitate alignment and coordination between the technical work towards process/material optimization and the study of sustainability aspects of the respective technologies.

Conclusions
In this work, production and disposal of nanocomposite 3D printing laments are studied in terms of life cycle climate change impacts, on an aim to provide a decision-support tool towards sustainable development. End of life management of such materials was also discussed. Based on the results, the GWP of the nanocomposite laments varies based on polymer and nano ller type, ranging from 4.04 kg CO 2 eq. (PP-GNP) to 13.21 kg CO 2 eq. (PA-GO). GO and CNT llers have been identi ed as key hotspots within the nanocomposite lament life cycle. For the polymers, speci cally in the case of PA matrix, signi cant climate change impacts are displayed, due to comparatively more energy-intensive processes than the other thermoplastic alternatives. Considering the considerable climate change impact displayed for PA-based nanocomposites, the EoL treatment options present high signi cance towards reducing climate change impact potential across the material life cycle. The worst-performing lament (PA-GO) was examined in terms of alternative EoL treatment scenarios, and the results showed a ± 30% variation based on the applied treatment approach, while recycling presented the most favorable performance amongst the EoL alternatives in this case. Coupling LCA results with the respective conductivities (or any relevant performance indicators) offers an area of suitable development alternatives, which can support the selection of options balanced in terms of sustainability/performance, in accordance with applicationspeci c requirements. Additive manufacturing and nanotechnology are expected to gain signi cant market share among emerging technologies. Still, nanotechnology introduces potential unknown risks to human health and ecosystems. Thus, waste community need to anticipate recycling and EoL management opportunities and in close collaboration with LCA practitioners to achieve the zero waste and circularity targets considering that both sectors (AM and nanotechnology) are rapidly developing.

Authors' contributions
Foteini Petrakli (FP), Anastasia Gkika (AG), Anestis Vlysidis (AV), Panagiotis Karayannis (PK) and Elias Koumoulos (EK) were contributed to the development of this manuscript. AV and FP analyzed the data regarding LCA. FP and AG performed the literature review and were the major contributors in writing the manuscript. PK contributed to the nanosafety part. EK made the literature review concerning performance. All authors read and approved the nal manuscript.