Comparative life cycle assessment of activated carbon production from various raw materials

Activated carbon (AC) is an effective adsorbent in water treatment but its production method has significant emissions to the environment. This study aims to quantify the environmental impacts of various AC types and determine whether raw material selection could reduce the footprint of AC. A cradle-to-gate life cycle assessment (LCA) was conducted on coal, coconut shell, wood, peat, and reactivated coal ACs. The different types of raw materials were selected to reflect typical global and local availability in the selected location. Life cycle data was collected from the Ecoinvent database, scientific literature, and an industrial producer. Using CML 2001 as a characterization method, potential environmental impacts were calculated for 12 categories. The direct emissions of AC production and electricity production were the largest contributors to environmental impacts. Coal AC had the highest impact in ten out of the twelve categories. On the other hand, reactivated coal and coconut AC had the lowest impacts in three and five categories, respectively. The comparison in carbon footprints be-tween the AC types were found to be dependent on inclusion or exclusion of biogenic emissions: When including biogenic carbon emissions, the Global Warming Potential (GWP) of reactivated coal AC was 72 – 80% lower than for the virgin ACs. When biogenic carbon emissions were excluded, the GWPs of the residual biomass ACs (coconut shell and wood) were found to be about 50% lower than that of reactivated coal AC. The results demonstrate that raw material choice and production method significantly affect the environmental impact of AC. To minimize site-specific impacts of AC application, technical feasibility of AC and use phase emissions need to be assessed.


Introduction
Activated carbon (AC) is an effective adsorbent with high porosity, large surface area, and high surface reactivity.AC is commonly utilized in drinking water treatment, air filtration, flue gas treatment, municipal and industrial wastewater treatment, and purification in the food industry (Çeçen and Aktas ¸, 2012, p. 3).The global activated carbon market size in 2020 was USD 2.96 billion and the market is expected to grow to USD 4.50 billion in 2028 as growing environmental awareness leads to strict emissions guidelines (Fortuna Business Insights, 2021).Water treatment is estimated to contribute approximately 50% of the total market (Mordor Intelligence, 2021).In drinking water treatment, AC is used for the removal of organic substances (Matilainen et al., 2006), colorants (Tomaszewska et al., 2004), and trace substances such as chemicals (Konradt et al., 2021).An inherent contradiction in the use of AC is that while it provides environmental benefit through decontamination (Korotta-Gamage and Sathasivan, 2017), the extraction of raw material and activated carbon production, especially the activation phase, have significant environmental impacts (Bayer et al., 2005).
AC can theoretically be produced by activation of any carbonaceous raw material with low enough volatile content (Marsh and Rodríguez-Reinoso, 2006).Common AC raw materials include crude coal, hard woods, coconut shell, fruit stones, and peat (Heidarinejad et al., 2020).However, to minimize the environmental impacts of AC, alternative raw materials, such as biomass-based agricultural and industrial residuals, have been suggested.These materials include waste wood, biowaste, nutshells, and food industry by-products.However, biomass materials are not all equivalent in AC production and can differ significantly in emissions (Liao et al., 2020).In recent years, significant research has been dedicated to demonstrating the technical feasibility of a variety of local waste materials as AC raw materials (Ioannidou and Zabaniotou, 2007;Reza et al., 2020;Yahya et al., 2015).However, more research is needed to assess the economic feasibility of these waste-based ACs (Jaria et al., 2022).The environmental impacts of alternative ACs and conventional coal-based ACs have been compared using life cycle assessment (LCA) (Hjaila et al., 2013;Joseph et al., 2020;Kim et al., 2018;Loya-González et al., 2019;Sepúlveda-Cervantes et al., 2017).
LCA is an internationally recognized and standardized method for quantifying the possible environmental impacts that can occur during the life cycle of a product, service, or process.The LCA methodology has been used to determine the environmental impacts of AC and other water treatment methods in a variety of studies (Arena et al., 2016;Bayer et al., 2005;Gabarrell et al., 2012;Gu et al., 2018;Hjaila et al., 2013;Jeswani et al., 2015;Nowrouzi et al., 2021).Despite a number of studies evaluating the environmental impact of ACs, these studies have generally been limited to comparing the impacts of a single novel AC raw material or other adsorption method to industry standard coal-based AC.
To the best knowledge of the authors, no studies have compared the environmental impacts of AC production from a variety of both fossil and biomass raw materials.Because the use of AC is increasing, it is important to identify ways to reduce its impact.Choice of raw material presents a potentially important factor to consider when sourcing AC.Therefore, the aim of this study is to present a comparison of the environmental impacts of AC production from typical raw materials.Additionally, the goal is to find out if locally available raw materials produce a more environmentally friendly AC compared to global alternatives.A cradle-to-gate comparative LCA was conducted on ACs produced from coal and coconut shell (global raw materials), wood and peat (local raw materials), and reactivated AC from coal.This study was conducted from the perspective of a water treatment facility that uses coal-based granular AC (GAC) which is reactivated on average every four years in Belgium.Studies have already demonstrated that the emissions of reactivating coal GAC are smaller compared to those of activating virgin coal (Bayer et al. (2005) and Muñoz et al. (2007)), however, this study brings forth an added level of evaluation as reactivated coal GAC is compared to other GAC types.

LCA methodology
The ISO standards 14040 and 14044 on LCA were used as a methodological guideline in this study.This study follows the four steps of LCA: goal and scope definition, life cycle inventory (LCI), life cycle impact assessment (LCIA), and interpretation.However, this study is not fully compliant with the ISO standards because critical review by interested parties (which is required from a comparative LCA disclosed to the public) could not be conducted and complete equivalence of the products being compared could not be ensured within the scope of this study (ISO 14040, 2006;ISO 14044, 2006).

Goal and scope
The intended application of this study is comparison of the environmental impacts of AC (specifically GAC) production from various raw materials.The goal of this LCA study is to determine whether raw material precursor choice could contribute to reduced environmental impacts during AC production.The products compared were chosen to reflect a variety of typical AC options: virgin GACs made from coal, coconut shell, wood, peat, and reactivated coal GAC.The selection was based on global and local resources available in the assessment location of Finland, as well as renewable and fossil resources.
The system boundaries are limited to cradle to gate, in other words from raw material acquisition to transport to the water treatment facility for use.The use and end-of-life stages of the products are omitted from the system boundaries also because they are assumed to be similar for all products.Similarly to other LCAs on AC (Arena et al., 2016;Gu et al., 2018;Hjaila et al., 2013), an input-based functional unit, FU, of 1 kg of produced GAC was chosen as the functional unit.This allows for comparison of the GAC types beyond the end-application assumed in this study.However, because the use phase is excluded from system boundaries, a limitation of this study is omission of comparison of the technical performance between the GAC types.
LCA modeling software GaBi was used to compile the LCI and calculate the potential environmental impacts.Google Maps was used to calculate transport distances over land and Searates for overseas transport.Data for the LCI was collected from the Ecoinvent database and scientific literature.Additionally, data for peat GAC production was obtained froman industrial peat AC producer in Finland.

Assumptions, value choices, and limitations
The following assumptions were made for all the product systems collectively: • The cradle-to-gate study excludes the use phase from the system boundary.Properties such as purification performance or frequency of reactivation affect the use-phase environmental impacts.These are unique for each treatment plant depending on the application.• The ACs assessed are assumed to be GACs for use in a water treatment facility in Helsinki, Finland.• Truck transport in a 16-32 t truck is assumed for land transport.Euro 5 class is assumed for transport inside Finland and Euro 3 outside of Finland.Return journeys are excluded from the system boundaries.• Overseas transport is assumed to be sea freight transported directly to Finland with no stops in between.• Some AC production processes produce co-products such as heat and gypsum that are used in other product systems.To avoid allocation and remain consistent for all the GAC types, co-products intended for other applications are excluded from analysis.Not enough information is available on the exact quantities of these co-products equally for all the GAC types.• Transport of waste ash to waste treatment is considered negligible in impact and omitted from assessment.• Treatment of wastewater from the AC production process is excluded from the system boundaries as there is not enough information available about its creation.• Typically, in LCAs concerning biomass AC (or biomass combustion in general), biogenic carbon emissions are omitted from GWP because they are considered neutral to climate change (Arena et al., 2016;Gu et al., 2018;Kim et al., 2018).However, as Wiloso et al. (2016) point out, excluding biogenic carbon emissions can give a false impression of the true emissions.Therefore, to limit the role of value choices, GWP is reported both including and excluding biogenic carbon emissions in this study.The biogenic carbon storage of GAC was excluded from results as the lifetime of GAC is likely to be short enough that it will not serve as long-term carbon sink.

Life cycle inventory
Fig. 1 illustrates the product systems and system boundaries of the GACs being compared.Further details on compilation of the LCIs are described in the following subsections for each of the GAC types.

Coal GAC
China is home to some of the largest coal AC producers in the world.Therefore, mining and GAC production was assumed to take place in China.Little information was found on the coal extraction process for AC production, which Gabarrell et al. (2012) also noted in their study.Thus, a typical coal extraction process from the Ecoinvent database was used for modeling, similarly as was done by Gabarrell et al. (2012) and Muñoz et al. (2007).A market process from the Ecoinvent database for Chinese coal was used which consists of coal extraction and an average transport distance from the coal mine to the location of use.Table S1 lists the LCI for coal GAC production and transportation.
An Ecoinvent dataset on coal GAC production from Bayer et al. (2005) and Muñoz et al. (2007) was used for the material and energy balance and emission profile of GAC production from hard coal.This dataset has been frequently both used and compared to in the literature on AC sustainability (Gabarrell et al., 2012;Gu et al., 2018;Jeswani et al., 2015;Joseph et al., 2020).However, the dataset is premised on GAC production in Central Europe.Therefore, the processes used to model the flows were chosen to represent China while the quantities of the dataset were kept the same.To calculate transport distances, North-East China was assumed as the production location.
The inputs into the GAC production process are electricity, natural gas, water, and hard coal.During the process, coal is converted into GAC with a mass yield of 30%.The coal is first carbonized at 700 • C and then activated at 800-1000 • C. Natural gas is used to produce steam for activation.The outputs of the production process are GAC, ash (assumed to be landfilled), and emissions.The emissions reported by Bayer et al. (2005) are based on emissions of coal combustion for heat production.

Peat GAC production
Peat was chosen as a local AC raw material as it is commonly available in Finland.Peat GAC production was modeled after the process of a Finnish industrial peat AC producer, Novactor (Vapo Oy, 2019;Novactor, 2021).The material balance of peat activation was obtained from the industrial AC producer, missing energy balance from coal GAC production, and emissions adapted from peat energy use.South-Eastern Finland was assumed as the production location.Table S2 lists the LCI for peat GAC production and transportation.
An Ecoinvent dataset on peat extraction in the Nordic countries was used to model the raw material extraction process.Truck transport of 35 km was assumed from the peat lands to the production facility.After peat extraction, the peat is dried and pelletized (Vapo Oy, 2019).As Ecoinvent lacked data on electricity consumption of peat pelletization, data for wood pelletization was used.
The material balance of GAC production was compiled from a public environmental permit and private communication with the industrial producer.After pelletization, the peat is carbonized and activated in a multiple hearth furnace with steam (Vapo Oy, 2019).The exothermic reactions provide enough heat for drying of the peat during pelletization, upkeeping temperature during GAC production, and heating steam for activation.Additional fuel is only needed once a year during start-up, and this was omitted from the system boundaries.Therefore, electricity was assumed as the only external energy source.Because no data was publicly available on the electricity need of peat GAC production, the electricity consumption was assumed to be the same as for coal GAC from Bayer et al. (2005).
No data on the direct emissions of peat GAC production was publicly available, and therefore, the emissions of industrial peat combustion for energy were adapted to resemble GAC production.Other LCAs on AC production such as Bayer et al. (2005) and Gu et al. (2018) also recognize the lack of primary emission data and therefore use energy production emissions to model activation.A critical difference between combustion and activation is the carbon that is stored in GAC instead of being released as carbon emissions.Based on the wet yield of peat GAC production (10%), dry peat carbon content (56%), water content of peat (50%) and assumed carbon content of GAC (90%), one third of the carbon in peat was estimated to be stored in the GAC.Therefore, CO 2 , CH 4 , and CO emissions of peat combustion for energy were reduced by one third to represent the carbon that is stored in the GAC instead of being released in combustion.The other emissions were assumed to be the same as for electricity production from peat.

Coconut shell GAC production
Coconut shells differ from coal and peat because they are residual products of the coconut industry.Therefore, the environmental burden related to coconut shell acquisition for GAC production is excluded in this study.This judgment aligns with other LCAs on residual biomass ACs (Arena et al., 2016;Hjaila et al., 2013;Loya-González et al., 2019).Table S3 lists the LCI for coconut GAC production and transportation.
Data on coconut AC production was obtained from Arena et al. (2016).The studied process includes crushing, drying, carbonization, activation with steam, and final crushing.Similarly to peat GAC production, the process is self-sufficient in heating: the distillate products from carbonization are burned to provide heat for drying the coconut shells and the waste heat from activation is used to produce the steam required for activation.The only other energy source used in the process is electricity.Arena et al. (2016) modeled the production process using data from industrial producers and scientific literature.Carbonization is assumed to take place at a temperature of 500 • C and activation at 700-950 • C with steam.Arena et al. calculated the emissions of the process using data on the composition of the raw material used.
Coconut AC is typically produced in South-East Asia, and Southern India was assumed as the production location in this study.An average transportation distance of 100 km was assumed for the coconut shells to the AC production facility.Although Arena et al. (2016) assumed Indonesia as their production location, their data was judged to be appropriate also for India.

Wood GAC production
Wood is one of the most common locally available residual raw materials in many areas.Therefore, the wood was assumed to be sourced and produced into GAC in Finland.The raw material used for wood GAC production was assumed to be a residual from the wood industry such as bark or wood chips.Therefore, similarly to coconut GAC, no burden for raw material acquisition was attributed to wood GAC production.
The wood residual was assumed to be sourced in Finland 50 km away from the GAC production location.The same production location in Eastern Finland was assumed as for peat GAC production.The residual wood is first chipped into a uniform size and dried.Like for peat GAC production, the GAC production process is assumed to provide enough excess heat for drying, and therefore the only other necessary energy input is electricity.Gu et al. (2018) demonstrate the feasibility of this assumption.Table S4 lists the LCI for wood GAC production and transportation.
Data on the inputs and outputs of activation was collected from Gu et al. (2018).The data was based on measurement of the operation and emissions of a rotary kiln used for activating wood charcoal.The LCI of Gu et al. did not include waste ash, and therefore, the amount was assumed negligible.Although the same study included data on the carbonization stage, it was not applicable to this study since the primary product of carbonization was syngas with AC only as a by-product.Therefore, like for peat GAC, an Ecoinvent dataset for heat and electricity co-production from wood was used to model the emissions of carbonization.With the assumptions that the carbonization yield is 25% (Kim et al., 2018) and carbon content of GAC 90%, 43% of the carbon in the wood was calculated to be stored in the biochar.Therefore, the CO 2 , CO, and CH 4 emissions were reduced by 43% from the values for combustion.
In addition to drying the wood chips, the GAC production process was assumed to provide enough heat for steam activation and sustain heat in the carbonization and activation furnace.Therefore, the datasets for carbonization, and activation were modified to require only electricity as the external energy source.The electricity requirement was referenced from Gu et al. (2018).

Reactivated coal GAC
The raw materials for reactivated coal GAC are spent coal GAC and virgin coal GAC.Spent GAC is a waste product, and therefore has no environmental burden attributed to it.However, some virgin GAC is needed during reactivation to make up for losses, and therefore the burden of virgin coal GAC production is included in the system boundaries of reactivation.
Reactivation is assumed to take place in Feluy, Belgium as Finland does not have the required reactivation capacity.The electricity demand for pumping the spent GAC out of the filter into the transportation tanker was excluded from the system boundaries.The spent GAC is transported with an additional 25% mass due to water.The GAC is transported first by truck to the Helsinki port, then by sea freight to Antwerpen, Belgium, and again by truck to Feluy.
A coal GAC reactivation dataset from Ecoinvent by Bayer et al. (2005) and Muñoz et al. (2007) was used to model reactivation.Reactivation is assumed to be like activation, and therefore the Ecoinvent dataset is based on activation data modified to represent reactivation.The emission data for reactivation does not account for the emissions caused by gasification of the adsorbed pollutants in the GAC as these are different for each batch of spent GAC.
The Ecoinvent process assumes losses of 10% by mass of the GAC during reactivation.Instead, in this study, 15% losses were assumed, in line with the experiences of the water treatment facility studied.Therefore, the production of 15% make-up virgin GAC was included in the system boundaries.However, the inputs and emissions of the Ecoinvent process were not modified because the impact of the lower yield on emissions is uncertain.The LCI of reactivated coal GAC is listed in Table S5.

LCIA results
Fig. 2 presents the LCIA results for each of the environmental impact categories examined.The absolute impacts illustrated as bars are expressed per FU, 1 kg of GAC.Additionally, the impacts were normalized to the EU region using the method The CML 2001-January 2016 EU25 + 3, year 2000 (line in Fig. 2).This was done to identify the most significant environmental impact categories.The LCIA and normalized results are listed in tabular form in Tables S6 and S7.
When the impacts are normalized to EU person equivalents, the relative sizes of the different impacts can be compared.MAETP (i) stands out as the most significant impact for all the GAC types: its normalized impact is at least an order of magnitude higher than the rest of the categories.The main contributors to MAETP (i) are the direct emissions of GAC production and electricity production, particularly beryllium.HTP (h) and FAETP (e) are the next largest impact categories.A variety of air emissions from GAC and electricity production contribute to HTP (h) while for most of the GAC types electricity production emissions are the primary contributor to FAETP (e).For coconut and wood GAC, also GWP (f) is one of the highest normalized impacts.All the GAC types share the same smallest normalized impacts: ADP elements (a), ODP (j), and TETP (l).
Coal GAC shows the highest environmental impact in ten of the twelve impact categories.In particular, coal GAC has 2.5-6 times higher of an environmental impact in all three of the most significant categories identified in the normalized LCIA impacts: MAETP (i), HTP (h), and FAETP (e).This is explained by the emissions coal extraction and natural gas production, which do not occur in the life cycles of the other GAC types.The production of other virgin GAC types uses the heat and gases produced during activation.Similarly, coal and peat GAC show over 6 times higher results for ADP (fossil) (b) than the rest of the GAC types due to the use of fossil raw materials.
Coconut and reactivated coal GACs show the lowest overall environmental impacts: they have the lowest or second lowest impacts in nine and six of the impact categories, respectively.Still, in many categories their impacts are close to or even higher than those of peat and wood GAC.For MAETP (i), wood GAC shows the lowest and coconut GAC the second lowest impacts as only their electricity production emissions contribute towards the category.Reactivated coal GAC has a similar impact to coconut GAC due to its low emissions compared to virgin GAC production.For HTP (h), coconut GAC shows the lowest impact although this could be due to the use of a simplified emission profile from the literature.Peat and reactivated coal GAC show similar but slightly higher impacts.
The main contributors to acidification (c) and ozone creation (k) are the direct emissions of GAC production while both GAC and electricity production contribute to eutrophication (d) and terrestric ecotoxicity (l).For coal GAC, coal mining and natural gas production are also significant contributors to EP (d), TETP (l), POCP (k).
The GWP (f and g) impacts are highly dependent on whether biogenic carbon emissions are considered neutral towards climate change, and therefore excluded from GWP.When biogenic carbon emissions are included (f), reactivated GAC shows a 3-5 times lower impact than all the other GAC types.The carbon emissions of reactivation are lower than for activation.For the virgin GACs, peat GAC has the lowest GWP of 6.7 kg CO 2 -eq/kg GAC and coal GAC the highest of 9.5 kg CO 2 -eq/kg GAC.The results suggest that the direct carbon emissions of the fossil-based peat and coal GAC are lower than those of coconut and wood GAC.The fossil-based materials have a higher carbon content, and therefore require less burn-off to reach the high carbon content of AC.However, the GWP of coal GAC is still the highest due to emissions from coal mining and natural gas production.
When excluding biogenic carbon emissions (g), as is usually done, the GWPs of coconut and wood GAC decrease dramatically from 7.5 to 9.2 kg CO 2 -eq/kg GAC to 1.1 and 0.9 kg CO 2 -eq/kg GAC, respectively.In this case, their GWPs become about half of that of reactivated coal GAC.
Reactivation requires 15% by mass virgin GAC to make up for losses.However, the production of make-up GAC accounts for, on average, 65% of the impact in each impact category.The emissions of reactivation are much lower compared to activation which was also noted by Gabarrell et al. (2012), Jeswani et al. (2015), and Muñoz et al. (2007).
The environmental impact of transportation is low for all the GAC types.Transportation emissions are significant contributors only to the categories of ODP (j) and ADP elements (a), which were identified in the normalized impacts as the smallest impact types.The LCIA results show a similar contribution of transport to environmental impacts, regardless of the GAC production country.Even though coal, coconut, and reactivated coal GAC require long overseas transport, their transport does not have a significantly higher environmental impact because the emissions of sea freight are smaller than those of truck transport per tkm.
The comparison presented here could change depending on the unique use phase of the GAC application.However, this cradle-to-gate comparison serves as a generally-applicable starting point for further studies into the impacts of GAC in specific applications.

Comparison to the literature
LCA results vary widely depending on the data, methodology, and system boundaries used.Table 1 compares the results of this study to the literature.The studies compared use varying impact categories and therefore only the common category of GWP is included here.Many LCA studies on AC use different FUs, and therefore only studies with the same FU of 1 kg AC could be compared to.
For coal AC, results vary in the literature from 3.41 to 18.3 kg CO 2eq/kg AC.Joseph et al. (2020) explain their particularly low value by the fact that they only account for the emissions caused by removal of fossil carbon in the end-of-life incineration phase.Gu et al. (2018), on the other hand, report a particularly high value of 18.3 kg CO 2 -eq/kg AC.They mention they modeled their emission profile from coal combustion but their value may be too high if they have not accounted for the lower emissions of activation compared to combustion.The result of this study is closest to Bayer et al. (2005) but the comparison is not an independent one as the Ecoinvent data set used in this study was derived from Bayer et al.For reactivated coal GAC, the result of this study is again similar to Bayer et al. although the dataset used in this study was derived from Bayer et al.It is also similar to the carbon footprint reported by the manufacturer of reactivated coal GAC for the water treatment facility studied (1.5 kg CO 2 -eq/kg AC).
For biomass ACs, the impacts vary even more widely due to significant differences in the energy consumption and activation methods of the different processes.In general, the results of this study for coconut and wood AC are relatively similar to the range of biomass ACs reported in the literature.Gu et al.'s (2018) result is higher than that of this study due to the use of external fuel.This denotes the importance of efficient use of heat and gases during AC production.For both Hjaila et al. (2013) and Nowrouzi et al. (2021) the largest contributor to environmental impact is the production of activation chemicals.The physical activation process assumed in this study requires only steam for activation and therefore bypasses this source of emissions.

Sensitivity analysis
The emissions of GAC production and electricity production were identified as the most significant contributors to environmental impacts.Therefore, electricity usage and the amount of raw material needed to produce 1 kg of GAC were selected as parameters to be assessed in the sensitivity analysis.The conversion efficiency or yield of raw material into GAC directly effects the magnitude of the direct emissions of GAC production.The actual uncertainty in these two parameters was unknown, and therefore a conservative estimate of ± 20% was chosen to be assessed.The payload of transport was accordingly modified to reflect the changes in raw material quantity.For reactivated GAC, the production of virgin make-up GAC was identified to be the largest contributor to environmental impacts and therefore for in the sensitivity analysis the amount of virgin GAC was varied ± 20%.
For coconut GAC, an additional alternative scenario was modeled because it was recognized that in addition to full industrial production, open-pit carbonization is common for AC production in South-East Asia.Coconut shells are carbonized by regional charcoalers after which the charcoal is purchased by industries such as the AC production industry.Therefore, it was judged relevant to test how much the impacts of coconut GAC depend on the exact production method.As for the fully industrial coconut GAC production scenario, data from Arena et al. (2016) was used to compile the LCI.Arena et al. include a scenario in their study in which coconut shells are industrially carbonized but there is no abatement of emissions.The data from this scenario was used and modified to account for no combustion of the methane and carbon monoxide that would be the case in open pit carbonization.Therefore, methane and carbon monoxide were assumed to be emitted into the atmosphere and the carbon dioxide emissions of Arena et al. were decreased correspondingly.The LCI for open pit carbonization including emissions is listed in Table S8.There are different kinds of pit carbonization methods but for this assessment, a worst-case scenario was chosen where there is no abatement of emissions into the atmosphere, ground, or water.20 km of truck transport was assumed for the coconut shells to carbonization, and 100 km for the charcoal to the AC production facility.
The GWP results of the sensitivity analysis scenarios are illustrated in Fig. 3 and the rest of the categories in Fig. S1.The results are also given in Tables S9-S13.
Despite the uncertainty in the results, the production of coal GAC still has the overwhelmingly highest impacts in the toxicity impact categories.The only exception to this is open pit carbonized coconut GAC which has a higher HTP (S1f).The other GAC types have relatively similar impacts in the toxicity impact categories and therefore can't be placed in order of superiority with a high level of confidence.
When including biogenic emissions (3a), the GWP of reactivated coal GAC is significantly lower than that of any of the virgin GACs even when considering the sensitivity analysis.The GWPs of the virgin GACs are close enough to each other that confident comparison within the expected range of uncertainty cannot be done.However, when excluding biogenic emissions from GWP (3b), wood and coconut GAC have a lower impact than all the other GAC types including reactivated coal GAC.The exception, again, is open pit carbonized coconut GAC which has high methane emissions.With a high level of confidence, the GWP excl.bio. of peat and coal GAC are higher than for the other GAC types.
The environmental impact of open pit carbonized coconut GAC is higher than that of fully industrially produced in seven of the twelve impact categories.Notably, HTP (S3f), which was identified in the normalized impacts as one of the most important impact categories, is 11 times higher with open pit carbonization due to the emission of tar.The total carbon emissions of open pit carbonization (3a) are lower than those of full industrial production but the GWP excluding biogenic emissions (3b) is 2.5 times higher for the open pit carbonization scenario.This is explained by the global warming effect of methane emissions which are included in GWP even when excluding biogenic carbon emissions.
In the discussion of the results of this study, coconut and reactivated GAC had the lowest overall impacts.However, if coconut GAC is open pit carbonized, the potential environmental impact of its production becomes higher than of reactivated coal GAC in the categories of EP (S3d), FAETP (S3e), GWP (3a), GWP excl.bio (3b)., and HTP (S1f).Therefore, if open pit carbonization is assumed, the results seem to slightly favor reactivated coal GAC over coconut GAC.

Conclusions
This study compares the environmental impacts of GAC production for several typical raw materials, namely coal, peat, coconut, wood, and reactivated coal, within one cradle-to-gate LCA.The aim of the study is to explore raw material choice as an avenue for minimizing the environmental impact of AC.Normalized EU level impacts implicate that the most significant environmental impacts for all GAC types are Marine Aquatic Ecotoxicity Potential, Freshwater Aquatic Ecotoxicity Potential and Human Toxicity Potential.Direct emissions of AC production (i.e.air emissions of carbonization and activation) and electricity consumption are the major contributors to environmental impacts.
Coal GAC had the peak environmental impact in ten out of the twelve impact categories with the production methods assumed in this study.GACs from coconut, wood, and reactivated coal had the overall smallest impacts.The impacts of coconut GAC, especially toxicity impacts, are significantly increased if the open-pit carbonization method is utilized.When including biogenic carbon emissions, reactivated GAC had the smallest GWP of 2.1 kg CO 2 -eq/kg GAC.However, if biogenic emissions were excluded, as is generally the convention, the GWP of coconut and wood decrease to an even lower level than for reactivated coal at 0.9 and 1.1 kg CO 2 -eq/kg GAC, respectively.The large difference in GWP and GWP excluding biogenic emissions results in this study demonstrates the importance of understanding and transparently reporting both results in the context of biomass-based ACs.
The impact of transport was minor for all the GAC types: despite vast differences in transport distances, transport only accounted for 2-4% of the total GWP for the GAC types.Local raw materials did not show significantly smaller impacts from transportation.
The results demonstrate that the differences in emissions between raw materials and their production methods are significant and warrant further research to obtain data on actual emissions during industrial production.The cradle-to-gate study did not account for differences in the technical performance during the actual use stage of the GAC that could significantly change the results of the comparison.Due to the generic nature of the system boundaries of this study, results can be used as a starting point for application-specific cradle-to-grave assessment.The results of this study suggest residual biomass materials as a promising AC raw material, and therefore more research is needed on their large-scale economic feasibility and effectiveness in various applications.They also present an opportunity to prevent waste from entering the waste stream.As the emissions of reactivation for coal GAC were demonstrated to be lower than for activation, further research should be done on the feasibility and environmental impact of reactivating residual biomass GACs.

Fig. 1 .
Fig. 1.System boundaries and flow diagrams of GAC production.

Fig. 2 .
Fig. 2. Absolute (bar) and normalized (line) LCIA results for the different GAC types per FU, 1 kg GAC.Colours on the bars represent the different phases of GAC production.(For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)