IN
DEGREE PROJECT ENVIRONMENTAL ENGINEERING, SECOND CYCLE, 30 CREDITS
,
STOCKHOLM SWEDEN 2019
Sorption of perfluorinated and polyfluorinated alkylated
substances (PFASs) in the
subsurface of an industrial site in Sweden
CARL SKÖLD
KTH ROYAL INSTITUTE OF TECHNOLOGY
SCHOOL OF ARCHITECTURE AND THE BUILT ENVIRONMENT
TRITA TRITA-ABE-MBT-19519
www.kth.se
Abstract
Per- and polyfluoroalkyl substances (PFASs) are a group of emerging chemicals which have received increasing attention due to their toxicity, persistent properties, and global distribution.
In this study, sorption coefficients (Kd and KOC) of PFASs in an industrial site in Sweden were evaluated. Sorption is a measures of the mobility of a substance in the subsurface, and is a key factor in environmental risk assessments. Sorption coefficientswere calculated both from field samples processed in laboratory batch tests, and from a simplified approach involving the total concentrations in soil and groundwater (field-derived). Soil was sampled from two locations of the site; C8 and M6. Field-derived values were calculated based on concentrations which were historically measured. The aim was to compare the two methods, and to compare the values with literature values as well as guideline sorption values established by the Swedish Geotechnical Institute (SGI).
Sorption coefficients for PFHxA, PFHpA, PFOA, PFHxS, PFOS, 6:2 FTS and PFBS could be established. Results showed that laboratory-derived sorption coefficients were significantly higher than field-derived sorption coefficients. Laboratory-derived sorption values were also higher than to SGI’s preliminary sorption values. Comparing C8 and M6 KOC values to literature values, PFHpA, PFHxA, and PFBS exhibit values above literature values. PFOA, PFHxS, PFOS exhibit KOC values within the range of literature values. According to the sorption coefficients, predictive scenarios of leaching through the unsaturated zone were modelled, and it was concluded that leaching was higher in M6 compared to C8. The results also showed that an increase in precipitation increased the leaching.
Keywords
Sorption, soil-water partitioning, Kd, PFOS, PFOA, PFHxS,
Summary in Swedish
Per- och polyfluorerade alkylsubstanser (PFAS) är en grupp nyligen uppkomna kemikalier som har fått ökad uppmärksamhet pga. deras toxicitet, ihärdiga egenskaper och globala utbredning.
Detta examensarbete har studerat fördelningskoefficienter (Kd och KOC) för PFAS inom ett industriområde i Sverige. Fördelningskoefficienter är ett mått på mobilitet av en substans i underjorden, och det är en viktig komponent i riskbedömningar inom förorenad mark.
Fördelningskoefficienter beräknades dels utifrån jordprover som tagits i fält och analyserats i laboratorium med extraktionsmetoder, och dels utifrån ett förenklat tillvägagångssätt där beräkning skett med hjälp av tidigare uppmätta koncentrationer i jord och grundvatten.
Jordprover från två områden inom industriområdet; C8 och M6, togs och analyserades.
Fältbaserade fördelningskoefficienter beräknades utifrån koncentrationer som tidigare mätts vid brunnsinstallation och vid grundvattenövervakning. Målet med studien var att jämföra de två metoderna, och dessutom jämföra fördelningskoefficienterna med motsvarande i värden litteraturen samt riktvärden för fördelningskoefficienter som Statens Geotekniska Institutet (SGI) arbetet fram.
Fördelningskoefficienter för PFHxA, PFHpA, PFOA, PFHxS, PFOS, 6:2 FTS and PFBS kunde beräknas. Resultatet visade att laboratorie-baserade fördelningskoefficienter var betydligt högre än fältbaserade fördelningskoefficienter. Laboratorie-baserade fördelningskoefficienter var även högre än de preliminära riktvärden för fördelningskoefficienter som SGI etablerat. Vid jämförelse av fördelningskoefficienter för C8 och M6 kunde det konstateras att PFHpA, PFHxA, and PFBS hade högre värden än motsvarande i litteraturen. PFOA, PFHxS, PFOS visade på värden som var inom intervallet av värdena från litteraturen. Med hjälp av de beräknade fördelningskoefficienterna modellerades prediktiva utlaknings-scenarier. Utifrån resultatet sker utlakningen i större grad i M6 jämfört med C8. Utlakningen ökade även vid förhöjd nederbörd.
Nyckelord
Sorption, fördelningskoefficient, Kd, PFOS, PFOA, PFHxS,
Acknowledgments
I would like express my gratitude to several people that provided me with support to conduct this master thesis. First, I want to thank Marc Gath at AECOM for fundamental help in the initial stage of the study, and for continuous feedback along the way. I want to thank Professor Jon Petter Gustafsson for valuable feedback during the different stages of the study. A thank you to Hugo De Campos Pereira for allocating time to help me out in the laboratory, and for interesting discussions regarding the results. Furthermore, I want to thank Mattias Sörengård for the dialogue we had in the study’s initial phase, and for providing me with interesting articles for review. In addition, a thank you to Erik Gunnars for helping me with the SPE at SLU. I also want to thank my brother for motivating me throughout the study. Lastly I would like to thank Allmänna Idrottsklubben (AIK) for continuously providing a stress-free environment in otherwise stressful times.
Table of contents
1. Introduction 1
1.1 Problem formulation 1
1.2 Aim 2
2 Per-‐ and polyfluorinated alkylated substances 3
2.1 Classification and structure 3
2.2 Physiochemical and biological properties 4
2.3 Exposure and Toxicity 5
2.4 Legislation 5
2.5 Guideline values 6
2.6 Transport behavior 7
2.7 Sorption of PFAS 7
2.7.1 pH 7
2.7.2 Organic matter 8
2.7.3 Structure and chain length of PFAS 9 2.7.4 Literature value range of Kd and KOC 9
3. Site characteristics 11
3.1 Layout 11
3.2 Geology 11
3.2.1 Conceptual geological model 11
3.2.2 Bedrock 13
3.2.3 Soil characteristics 13
3.3 Hydrogeology 14
3.3.1 Conceptual model 14
3.3.2 Flow directions 14
3.3.3 Hydraulic conductivities 15
3.4 Prior investigations 15
4. Material and methods 16
4.1 Sampling procedure 16
4.1.1 Groundwater sampling 16
4.1.2 Soil sampling 17
4.2 Laboratory procedure 20
4.2.1 Direct injection 21
4.2.2 Solid phase extraction 21
4.3 Data processing and calculations 24
4.3.1 Field derived Kd based on historical dataset 24 4.3.2 Laboratory batch test Kd and KOC 25
4.4 1-‐D Modelling 25
4.5 Delimitations and assumptions 26
5. Results 27
5.1 Distribution of PFASs 27
5.2 Field-‐derived sorption coefficients 29
5.2.1 Range of Kd values 29
5.2.2 Kd versus pH 30
5.2.3 Kd versus chain length 31
5.3 Sorption coefficients calculated from direct injection 31
5.3.1 Range of Kd and KOC 31
5.4 Sorption coefficients calculated from direct-‐injection and SPE 33
5.5 1-‐D model 35
5 Discussion 36
6.1 Distribution of PFAS-‐11 36
6.2 Sorption coefficients 37
6.2.1 Variation within the site 37
6.2.2 Comparison between field-‐derived Kd and laboratory-‐derived Kd 37 6.2.3 Comparison with literature values 38 6.2.4 Effect of pH and chain length 38
6.3 1D model 39
6 Conclusions, environmental implications and future work 39
7 Uncertainties 40
8. References 41
Appendix A – PFAS concentrations historical data 47
Appendix B – Commercial lab results 49
Appendix C – Input data for model 56
Appendix D – SPE results 57
Appendix E – Kd and KOC vs pH and chain length 58
Abbreviations;
AFFF - Aqueous film forming foam B.g.l – Below ground level
C0 – Starting concentration Cs – Concentration in soil
Cw – Concentration in groundwater or leachate DOC – Dissolved organic carbon
FOSA – Perfluorooctane sulfonamide fOC – Fraction organic carbon
IF – Infiltration factor IS – Internal standard Kd – Sorption coefficient
KOC – Organic-carbon partition coefficient LC – Lethal concentration
LOI – Loss of ignition
LOQ – Limit of quantification POP – Persistent organic pollutant
PFAS – Perfluorinated and polyfluorinated alkylated substance PFASs – Perfluorinated and polyfluorinated alkylated substances PFOS – Perfluorooctane sulfonate
PFOA – Perfluorooctanoate PFBA – Perfluorobutanoate PFPeA – Perfluoropentanoate PFHxA – Perfluorohexanoate PFHpA – Perfluoroheptanoate PFNA – Perfluorononanoate PFDA – Perfluorodecanoate PFBS – Perfluorobutane sulfonate PFHxS – Perfluorohexane sulfonate pKa – Acid dissociation constant
SMHI – Swedish Meteorological and Hydrological Institute SPE – Solid phase extraction
TDI – Tolerable daily intake TOC – Total organic carbon TWI – Tolerable weekly intake
1. Introduction
There are currently over 85 000 sites in Sweden which are suspected or determined to be contaminated, and each year, more sites are identified (Swedish EPA, 2018). These sites are not exclusively contaminated due to former practices and operations when the environment was not of a significant concern, but also as a result of emerging chemicals synthesized during the last decades. Perfluoroalkyl and polyfluoroalkyl substances (PFASs) constitute one group of such contaminants. PFAS is a generic term for chemicals which contains one or more carbon of which the hydrogen atoms have been substituted by fluorine atoms (Buck et al., 2011).
PFAS is a human-made suite of chemicals, and they have been produced since the 1950s (Buck et al., 2011). Production is conducted by two main processes; electrochemical fluorination and telomerization (Buck et al., 2011). During the last decade, PFAS has been increasingly relevant as it was not until the early 21st century that their adverse effects were fully established (Giesy and Kannan, 2002). PFAS is widely distributed in the global environment, and due to their recalcitrant nature, many PFASs have accumulated in the environment. PFAS have been used for several purposes since the 1950s. Some common uses are; consumer products, manufacturing processes, precursor chemicals and firefighting either in response to fires or for training purposes. Certain sub-groups of PFAS, particularly perfluoroalkyl acids (PFAAs) which includes PFOA and PFOS, are persistent, bioaccumulative and mobile, in addition to being recalcitrant (U.S EPA 2003; ATSDR 2015; NTP 2016; Concawe 2016).
One of the sites which have had problems with the presence of PFAS is an industrial site in Sweden. Sampling for traditional contaminants (e.g hydrocarbons and volatile organic compounds) has been conducted since 2005, with the focus shifting to PFAS in 2016, and the magnitude of the environmental measuring and monitoring has been extensive. Sampling on the site has been conducted since the year 2005, and the magnitude of the environmental measuring and monitoring has been extensive. The site is a producer of base chemicals and equipment. Due to certain activities, an operational fire suppression system containing aqueous film forming foam (AFFF) is required, which is the source of PFAS on the site.
This master thesis study is conducted in collaboration with AECOM, Swedish University of Agricultural Sciences (SLU), Royal Institute of Technology (KTH) and the site operator.
AECOM is an international engineering consulting company that provide solutions in the built environment- and infrastructure sector. The company has been hired to further investigate the occurrence of PFAS on the site. The funding for this study was granted by the site owners.
1.1 Problem formulation
The behavior of PFAS in the subsurface depends on many aspects. Naturally, it is dependent on the composition of the compounds, such as chain length and ionic functional group (Vestergren, 2015). There are more than 4 700 different PFAS-compounds with different structures, functional groups and molecular weight (ITRC, 2018). These properties primarily affect PFAS interaction and transport with different soils. Furthermore, the behavior is also dependent on distribution and transport processes in the specific environment, e.g. if the medium is soil, water or sediment. In these media, it is dependent on environmental parameters like water flow, particle surface charges, phase interfaces, total organic carbon (TOC) and dissolved organic carbon (DOC). (Vestergren, 2015, ITRC, 2018). Therefore, the processes affecting PFAS are complex and dependent on the local environment, which is why it is crucial to established site-specific variables when assessing the behavior of PFAS on a site. It is, for
example, difficult to apply generic values to site-specific conditions with basic soil information and concentrations.
One key aspect of risk assessments in environmental science is sorption. Sorption is a term describing the sum of adsorption and absorption. To describe these processes, a sorption coefficient (also adsorption-desorption coefficient), hereafter Kd value, can be calculated. The Kd value is a measure of the mobility of a substance, commonly a contaminant, in the subsurface. Kd is an integral component of modeling contaminants transport and risk assessments (EPA, 2004). A high value means it is strongly adsorbed to the soil, and therefore, the contaminant has limited mobility. A low value means that the contaminant is less sorbed, occurs more in the dissolved phase and has a higher potential of desorbing and leaching.
Subsequently, it is therefore more mobile. The concentration over time depends on biological degradation, chemical degradation, losses due to evaporation and leaching(Mulder et al., 2001;
Volkering and Breure, 2003, Artola-Garicano et al., 2003; Hwang and Cutright, 2003;
Huesemann et al., 2004).
Batch tests are the most widespread method to acquire Kd values. In short, the experiment is done by spiking the sample with an internal standard containing a known concentration of the contaminant for a specific amount of time. The solution is, after this time, separated from the solid (i.e. the soil), and the amount of internal solution which is left in the solution is measured.
The contaminant concentration, which is sorbed to the solid, is obtained by calculating the difference between the initial and the final contaminant concentration. If the soil is clean of PFASs, the internal standard can be applied nevertheless to establish the potential of leaching.
A batch test can be done reasonably quickly. However, it is important to note that the method does not necessarily represent the field characteristics accurately sorption processes are complex.
A simplified approach to the calculation of field-derived Kd values is merely to divide the total concentration in soil with the total concentration in groundwater. The collection of water samples, including sampling period and sampling depth, is fundamental for PFAS analysis and can have a high impact on the results (Ahrens, 2011). The soil samples on this site have historically been taken in connection to monitoring well installation. Ideally when calculating field-derived Kd values, groundwater measurements should be taken in connection with the soil samples. A complication in calculating the Kd-values was that the groundwater samples were taken 1-3 weeks after the soil sampling. This approach is considered best-practice during environmental investigations to ensure that the hydrogeological setting is normalized after it has been disturbed due to drilling, however; it increases the uncertainty in calculating Kd. It can be of interest to assess the reliability of samples taken by this approach compared to Kd values established in the lab, and therefore, act as a bridge between the private sector and research.
The results could constitute part of more detailed risk assessments of the site, more specifically to assess the potential of PFAS migration, both vertically in the unsaturated zone towards the saturated zone, and ultimately horizontally with the groundwater. Kd values, together with other parameters, can be put in transport models of different scales, which can be of use in predictive scenarios of different timescales. This information can also aid in the process of evaluating remediation measures, e.g. what needs to be done to contain and remove PFAS and when it is necessary to initiate these actions.
1.2 Aim
The overall aim is to establish and analyze sorption coefficients of PFAS in the subsurface of
differ from results derived from laboratory processing. As sampling for this site has been conducted for several years, and in many different wells, the data can aid the process of acquiring a better understanding of the behavior of PFAS. The magnitude of the project as a whole gives insight on how to approach a case study concerning PFAS. Hence, part of this thesis was to evaluate the existing dataset and identify an appropriate approach to contribute to the knowledge of PFAS behavior. In the process of achieving the aim, several research questions are to be answered, presented below
• How do the field-derived sorption coefficients compare to laboratory-derived sorption coefficients?
• How do the field-derived sorption coefficients vary within the site?
• How do the sorption coefficients compare to literature values?
• How do the sorption coefficients vary with regards to pH and chain length? Is there a significant correlation?
• When put into a model, how does PFOS leach over time in the unsaturated zone, according to the calculated sorption coefficients?
2 Per- and polyfluorinated alkylated substances
2.1 Classification and structure
The term “Fluorinated substance” is a broad term that is defined as a large group of organic and inorganic substances which contains one or more fluorine atom (Banks et al., 2013).
“Perfluorinated” is applied when molecules in which fluorine atoms replace all the hydrogen atoms. The term “polyfluorinated” is applied when at least one, however not all, hydrogen atoms are replaced by fluorine (Buck et al., 2011). When these groups contain the perfluoroalkyl moiety CnF2n+1-S, they can subsequently be termed “Perfluoroalkyl and polyfluoroalkyl substance (PFAS)” (Buck et al., 2011). S corresponds to the charge functional group and n corresponds to the chain length. The functional group is a hydrophobic charged head, commonly a carboxylic or sulfonic acid, which defines its groups (Table 1). PFAS also contains a hydrophobic or oleophobic carbon tail (Figure 1).
The molecular structure of PFAS can be either linear or branched, which influences their physical and chemical properties.
Figure 1. Molecular structure of perfluorooctanesulfonic acid (PFOS), which is classified in the group PFSAs. PFASs have a hydrophobic (repellent of water) carbon tail/chain with multiple fluorine atoms and a hydrophilic (attracted to water) functional group/charged head. An organic group called a “spacer” links these components together. Edited from Wikimedia (2019).
The general approach of classifying PFASs is presented in Figure 2. There are several ways to divide and classify PFASs. PFASs can be divided into ionic, e.g. PFOS, and neutral PFASs, e.g. fluorotelomer alcohols (FTOHS). The terminology for each specific PFAS congener is derived from the number of carbons on the alkyl chain and the functional group.
Perfluoroalkane sulfonic acids (PFSAs) which have more than six carbon atoms and perfluorocarboxylic acids (PFCAs) with more than eight carbon atoms are considered long chained PFASs according to the Organization for Economic Co-operation and Development (OECD) (OECD., 2019).
Figure 2. Overview of the classifications of PFAS. The functional group of each PFAS defines its group.
Modified from Gesonline (2019).
2.2 Physiochemical and biological properties
Fluorine is the most electronegative element in the periodic table, and when it bonds with carbon, they form a strong and stable bond, with limiting interaction to other molecules (Järnberg et al., 2006). This bond is also the reason why PFASs do not quickly degrade (Swedish Chemical Agency, 2019). Commonly, there are several fluorine atoms bonded to each carbon atom, which further increases the strength. Therefore, the perfluoroalkyl moiety exhibits a high degree of thermal and chemical stability. These properties, in combination with the presence of both a hydrophobic and lipophilic component, make many PFASs applicable in surfactants, e.g. water repellant clothing, coatings and AFFFs. AFFFs are used to extinguish fires comprising flammable liquids (Kissa, 1994, Taylor, 1999, Kissa, 2001). These applications are made possible due to the aqueous surface tension lowering properties (Kissa, 1994, Taylor, 1999, Kissa, 2001). The AFFFs liquefies the foam and renders it more effective for dispersal (The Swedish Chemical Agency, 2019). AFFFs also increases the foams ability to withstand evaporation and heat radiation (The Swedish Chemical Agency, 2019).
The bioavailability of PFAS in soil decreases as the fluoric tail increases (Holmström, 2017).
Substances with more carbon atoms generally exhibit a higher degree of persistence, are more likely to be bioaccumulative and are overall more prevalent in the global environment (Wang et al., 2013). Long chain PFASs like PFOS and PFOA, with chain lengths of 8 carbon atoms, are two of the most unwieldy compounds. These two compounds are also of particular interest
since they are end-products of certain PFAS precursors. Efforts have been made to phase them out of production and use. Subsequently, the industry has moved towards the usage of short- chained PFASs. However, certain researchers, e.g. Neltner and Maffini (2019), argue that even though many of the long chained PFASs are replaced in favor of short-chained PFAS, the toxicity and biopersistence of these are not sufficiently researched (Neltner and Maffini, 2019).
2.3 Exposure and Toxicity
The most common exposure of PFAS is through contaminated groundwater, which is used for drinking water supply. Both PFOA and PFOS are prevalent in waste-, fresh- and groundwater systems. Water and wastewater treatment facilities have difficulties treating PFAS due to their persistence. One of the most common treatment approaches is via adsorption to activated carbon filters (Qu et al., 2009, Appleman et al., 2013). Other common exposure pathways are consumption of fish which originated from PFAS-contaminated water, using products which contain PFAS and consuming food with PFAS present in the packaging. Inhalation of contaminated soil and air are also potential pathways of exposure (EPA, 2018).
The science community is continuously investigating the PFAS effects on humans. However, the knowledge on the toxicity of PFAS is limited, and primarily based on acute exposure of PFOS and PFOA in animal and plant experiments (Karolinska Institutet, 2018). Studies have shown that PFOS and PFOA have adverse effects on the liver, the immune system, thyroid hormones and on the metabolism of fat (Karolinska Institutet, 2018). Newborns who have been exposed during the fetus-stage have recorded lower birthweights, delay in bone development, sexual maturity, as well as overall development (Karolinska Institutet, 2018).
Several toxicity tests have concluded ranges for acute toxicity. A study from Li (2009), four freshwater species and three plant species were exposed to PFOS and PFOA with the purpose to assess their acute toxicity. For PFOS, all species showed 48h LC50-values of between 27 to 233 mg/L. The 96h LC50 value for three of the species ranged from 10 mg/L to 178 mg/L. The equivalent values for PFOA were 181 to 732 mg/L and 337 to 672 mg/L (Li, 2009). However, more research on the long-term effects on flora and fauna, are needed to assess the ecological risks of PFOS and PFOA, as well as other PFASs (Li, 2009).
A difficulty in assessing the adverse effects of PFAS in the context of risk assessment is the process of comparing site concentration to guideline values. There is not a consensus within the scientific community regarding guideline values, and to put these values in a temporal context to assess the long-term effects are even more difficult. Another problem is, given the wide PFAS distribution and its persistence, there is a limitation in the capability to provide accurate and exact low-level quantification in laboratories.
2.4 Legislation
Research has generally been focused on long-chains PFASs, e.g. PFOA and PFOS, due to their high toxicity, persistence and overall higher prevalence in nature. This focus is reflected in the legislation and the different agencies. For example, PFOS has since 2009 been included in the Stockholm Convention list of persistent organic pollutants (POPs). The purpose of its inclusion is to restrict the production and usage of PFOS. PFOA, PFHxS and their salts are currently under review to be included in the list. As of March 2019, a decision on their status has not been made (POPs, 2019). There are, however, other institutions which have included them in their list.
PFOS is also included in the UN’s Long-Range Transboundary Air Pollution (LRTAP). In the EU, it is currently forbidden to use PFOS and precursors of PFOS in chemical production.
Firefighting foam with PFOS is not allowed to be marketed since 2007, and since 2011, the remaining inventory is not allowed to be used (KI, 2019). The EU also has, under its REACH initiative, classified certain substances of high concern (SVHC). These include PFOS, PFOA, PFNA, PFDA and PFHxS (European Chemicals Agency, 2019). PFOA will be banned in the EU during 2020 (Swedish EPA, 2019b).
Sweden and its Environmental Code (Miljöbalken) applies the Polluter Pays Principle (PPP).
However, there are few cases where groundwater pollution cases which have gone to trial, and even fewer so concerning PFAS pollution to groundwater (Banzhaf et al., 2016). Subsequently, there are limited legal precedents, and regulatory authorities have limited knowledge beyond what can be applied from the Environmental Code and EU directive, which both are open to ambiguous interpretation (Banzhaf et al, 2016). Complicating the process further, PFAS is a complex suite of contaminants. This complexity often results in a complex remediation process and therefore, a potentially high cost. Companies are likely to appeal court decisions, which opens the possibility for further directions in the form of legal precedents (Banzhaf et al., 2016).
2.5 Guideline values
PFOS is the only PFAS for which Swedish guideline values have successively been established (SGI, 2019). The Swedish geotechnical institute (SGI) is currently working on establishing guideline values for PFOA; however, lack of sufficient ecotoxicological data is delaying the process. For other PFASs, there is a general shortage of data, even though many compounds are prevalent in the Swedish environment. As for now, the current recommendation from Swedish National Food Agency is to add the sum of seven PFASs; PFBS, PFHxS, PFOS, PFPeA, PFHxA, PFHpA and PFOA, and compare it to the guideline value for PFOS in soil and groundwater (Swedish EPA, 2019a).
The standard analytical suite in Sweden established by Swedish National Food Agency is PFAS-11. Swedish National Food Agency is currently investigating to expand the list to include more PFASs. However, the laboratories limited capability to analyze these PFASs due to the low concentrations is a limitation.
SGI has established preliminary guideline values for PFOS in soil and groundwater. The values have not yet been set, since they are based on tolerable daily intake (TDI) values from 2008 and are currently under review (Swedish EPA, 2019b). These values are merely an indication that action needs to be taken. The soil value for sensitive land use was set to 3 µg/kg TS and 20 µg/kg TS for less sensitive land use. The preliminary guideline value PFOS-contaminated groundwater is set to 0.045 µg/L (Swedish EPA, 2019b). In 2018, The European Food Safety Authority (EFSA) set a tolerable weekly intake (TWI) of 0.013 µg/kg body weight for PFOS and 0.006 µg/kg body weight for PFOA.
Even though preliminary guideline values are set for PFOS, there is no threshold value in drinking water which is legally binding. There is, however, an action limit value, i.e. measures are needed to be taken if concentrations exceed this, with the purpose to eliminate the risks for human health. It was established by Swedish National Food Agency and is defined as 90 ng/l for the sum of 11 PFAS compounds (Swedish EPA, 2019a).
2.6 Transport behavior
PFASs have been observed in humans, fish and wildlife globally. They have even been identified in polar bears in the arctic circle due to atmospheric and aquatic transport. Certain PFAS are more widely distributed than other PFASs due to differences in physical and chemical properties. PFASs have hydrophobic and hydrophilic functionalities; hence, they behave differently than other common non-ionizable organic pollutants (Renner, 2001, Villagrasa et al., 2006). Ionizable PFASs have a lower vapor pressure and water solubility; hence they are not as prone to long-range atmospheric transport as neutral PFASs, e.g. FTOHs. The mobility of PFAS in soil and groundwater is a complex process (Ahrens, 2011). One key parameter is sorption processes, which will be further elaborated on in chapter 4.7.
2.7 Sorption of PFAS
Sorption processes govern the degree of removal, bioavailability, degradation, volatilization and transport of organic compounds in the environment (Weber and Miller, 1989). Sorption processes for synthetic chemicals like PFAS are complex in nature (Ahrens 2011).
Sorption processes are governed by the physiochemical properties of the PFASs, the properties of the solid and solution phases for which the PFAS interacts with, as well as the time of interaction between the chemical and sorbent phase (Li et al., 2018). Two of the major mechanisms in PFAS sorption processes is electrostatic interaction and hydrophobic interaction (Du et al., 2014). Electrostatic attraction is formed between the negatively charged functional head and the positively charged surfaces of the adsorbents. This attraction is made possible since in environmental pH conditions, PFASs commonly exists as anionic species due to their low acid dissociation constants (pKa) (Li et al., 2018). Oxides in the soil provide positively charged surfaces for electrostatic interaction. The mineral phase and organic content of the soil contribute to charged surfaces. Soils components often comprise of both positive and negative charged and can either be variable or permanent (Li et al., 2018).
A review of published data conducted by Li et al (2018) concluded that pH, OC or clay alone could sufficiently describe the sorption behavior of PFASs. Nevertheless, below research of certain factors which influences PFAS sorption is presented.
2.7.1 pH
PFASs are weakly acidic chemicals. As pH increases, the proportion of anionic molecules increases. This increase leads to a decrease in sorption (Calvet, 1989; Weber and Miller, 1989;
Lee et al., 1990; Kah and Brown, 2006). A change in pH can also have effects on the sorbents surface properties, and not only the charge of the PFAS molecule (Li et al., 2018). The sorbent’s surface average net charge on the mineral particles will be lowered as pH increases (Johnson et al., 2007). Due to the hydrophilicity of the sulfonate and carboxylate functional groups, influence from electrostatic interaction influences sorption (Higgins and Luthy, 2006, Du et al., 2014). PFAS pH dependency is mainly due to the particle surface properties. PFOS has a low pKa value; therefore, its charge is relatively stable.
Figure 3 shows compiled results from several datasets of the relationship between PFOS and pH by Li et al. (2018). It shows an overall decrease in Kd values as pH increases.
Figure 3 Relationship between Kd values for PFOS and pH (R2=0.06, n=27). As pH increases, the sorption decreases. Soils with high OC content are indicated, which overall increases Kd values. From Li et al. (2018).
2.7.2 Organic matter
The effect of organic content has been widely studied, and for long it was believed that the sorption of non-ionic chemicals was solely dependent on organic content, as it acts as a sorbent (Li et al., 2018). Consequently, the Kd value can be normalized according to the organic content, which is calculated according to the equation below
𝐾"# ='%&
() (1)
where foc is the portion of organic carbon in the soil sample and Koc the distribution of the chemical between the water phase and the organic carbon of the soil. The Koc value has been widely used for modeling purposes, including in the context of PFAS contamination, e.g.
Rosenqvist et al. (2017). SGI has suggested using the KOC value to calculate a guideline value for PFOS, and a generic value has been set to 500 L/kg based on the 90th percentile of literature values.
However, research has since concluded that it is not only the organic carbon that is vital for sorption processes but also the chemistry of the organic carbon. The Koc is a simplified approach in explaining the effect of organic carbon on sorption processes. Nevertheless, studies have a shown strong positive correlation between certain PFASs (e.g. PFOA, PFOS) and organic carbon, e.g. You et al. (2010), Ahrens et al., (2009), Ahrens (2011) Li et al., (2018).
Figure 4 Correlation between the fraction of organic carbon and Kd values. Compiled by Li et al. (2018).
2.7.3 Structure and chain length of PFAS
In a study conducted by Ahrens (2011), it was concluded that the chain length was the dominating internal factor affecting the degree of sorption. Short-chain perfluorocarboxylic acids (PFCAs) were solely found in the dissolved phases and long-chain PFCAs; e.g. PFOSA were more strongly bound to a particle (Ahrens, 2011). Thus, short chain PFASs have a higher potential for long-range aqueous transport (Ahrens, 2011). These short chained PFASs are commonly in the front end of a contamination plume. Several other studies, e.g. Higgins and Luthy (2006), Ahrens et al. (2009), Nordskog (2012), Johnson et al. (2007) have concluded that long-chained PFASs are more strongly adsorbed than short-chained PFASs. This increase in adsorption is due to the fact that with increased carbon chain, the compounds hydrophobic properties are also increased (Krafft and Riess, 2015). Long-chain PFASs are also more inclined to be found in the upper part of the stratigraphy, as more sorption takes place (Sepulvado et al, 2011).
The functional group in certain PFASs also affects sorption. Several studies have compared the degree of sorption for PFSAs and PFCAs. A study conducted by Higgins and Luthy (2006), concluded that PFSA was sorbed 1.7 times more strongly than PFCA. A study by Voogt and Saez (2006) concluded that the Koc value was two times higher for PFOS compared to PFOA.
2.7.4 Literature value range of Kd and KOC
In addition to the preliminary concentration guideline values, SGI also compiled literature values of Kd and KOC (Tables 1 and 2) (Swedish EPA., 2019a). These values are compiled solely from batch tests, and no field-derived data (Pettersson et al., 2015). As a precautionary principle, SGI suggested the 10th percentile value from the literature values as the KOC for PFOS (Table 1). This value, 460 L/kg, was ultimately rounded up to 500 L/kg. (Table 1).
Table 1 Kd values for PFOS. The data is compiled by SGI (2015) from Chen et al. (2013), Milinovic et al. (2015), Enevoldsen and Juhler (2010), Higgins and Luthy (2006) and Ahrens et al. (2011).
Medium n Mean
Kd [L/kg]
Median Kd [L/kg]
S.D [L/kg]
10th percentile [L/kg]
90th percentile [L/kg]
Soil 13 61 32 22 14 114
Soil &
sediment
20 48 18 15 7 111
Sediment 7 23 17 8 4 52
med
Furthermore, SGI calculated a preliminary guideline KOC value according to equation 1 by considering a soil of 2 % TOC, which was deemed representative of a standard Swedish soil sample. Kd subsequently equaled 10 L/kg (Table 2).
Table 2 KOC values for PFOS acquired by compilation of literature values. The data is compiled by SGI (2015) from Chen et al. (2013), Milinovic et al. (2015), Enevoldsen & Juhler (2010), Higgins and Luthy (2006) and Ahrens et al. (2011).
Medium n Mean [L/kg]
Median [L/kg]
S.D [L/kg]
10th percentile [L/kg]
90th percentile [L/kg]
Log KOC 10th percentile
Log KOC 90th percentile
Mean Log KOC
Soil 13 1935 987 680 685 3638 2.84 3.56 3.27
Soil &
sediment
20 2791 981 1136 460 6155 3.06 3.79 3.45
Sediment 7 4381 692 3053 268 12327 2.84 4.09 3.64
Results of additional sorption studies are presented in Table 3. These KOC values are relevant to compare with the PFASs which are quantified with the batch test.
Table 3 Compiled and modified from Campos Pereira et al (2018)
Compound Acronym Log KOC
Perfluorobutanoate PFBA 1.88f Perfluoropentanoate PFPeA 1.37f Perfluorohexanoate PFHxA 1.31f, 2.1a Perfluoroheptanoate PFHpA 1.63f, 2.1a Perfluorooctanoate PFOA 1.89-3.5a, c, d, e, f
Perfluorononanoate PFNA 2.36-4.0a, b, d, f Perfluorodecanoate PFDA 2.96-4.6a, b, d, f Perfluorobutane
sulfonate
PFBS 1.22e, 1.79f Perfluorohexane
sulfonate
PFHxS 2.05-3.7a, d, f Perfluorooctane
sulfonate
PFOS 2.6-3.8 a, b, c, d, e
Perfluorooctane sulfonamide
FOSA 4.2-4.5c, d
a Labadie and Chevreuil (2011).
b Higgins and Luthy (2006).
c Ahrens et al. (2011).
d Ahrens et al. (2010).
e Milinovic et al. (2015).
f Guelfo and Higgins (2013).
3. Site characteristics
3.1 Layout
The site is approximately 150 000 m2 (15 hectares) and located in an industrial area. Industrial activities have been continuous since the 1950s with numerous different owners and activities.
Before these activities, the site was primarily agricultural land. Certain parts were also used for clay extraction, and these areas were subsequently back-filled with clay and construction materials. The site is currently a producer of base chemicals and equipment. The overall industry area houses several industries with different purposes. Historically, there have been dry cleaning facilities, petrol stations and scrap metals yards in close vicinity to the site. The site is protected by fences all around, with public roads in the vicinity of it. The majority of the surface is covered with asphalt. Less than 10 % of the area have planted green areas, e.g. bushes and trees. The topography is generally flat.
The site has 110 monitoring wells installed, of which 76 remain active. The wells were installed based on current and historical activities and where hotspots of PFAS have been identified.
Figure 5 Partial overview of the site which incorporates relevant wells. Wells termed MW-3XX penetrates the deep groundwater zone. Well termed with grey text is no longer active. Overall, the site has 76 active monitoring wells. Road and building names have been removed.
3.2 Geology
3.2.1 Conceptual geological model
Three geological cross section has been established through borehole logs, striking N-S (Figure 6), E-W (Figure 6) and NE-SW (Figure 7). The major part of the area is characterized by fill material, with an overall thickness of 0-9 meters. It is thickest in areas where clay has been extracted. The filling material consists of heterogeneous material, with materials including demolition waste (e.g. bricks and wood), and general waste (e.g. nails, wires, plastic). The top
part of the layer consists of granular fill material. This top part of the layer has a thickness of approximately 1-4 meters. The lower part of the filling material is characterized by clay-rich materials, with a thickness of 1-6 meters. The permeability of the filling material decreases as the clay content increases. However, due to its overall heterogeneous content, the permeability of the material ranges to a vast degree.
Underlying the filling material is a layer of natural clay, with a thickness which varies between 0.5 – 11 meters (Figure 7). The upper part of the clay is postglacial clay, and underlying it is glacial clay. The composition of the clay varies from pure clay to silty clay to sandy clay. The clay layer is the thickest in the north part of the area (Figure 6).
Under the clay deposits is a layer of sandy glacial till, which at certain locations are gravel, silty sand or silty clay. The overall thickness of the soil deposits is approximately 20 meters. The maximum thickness of the glacial till is 8.1m, located at MW-109. The minimum thickness is 0,9 meters and located in MW-321. The moraine is generally thicker in the south and west parts of the site, with a shallower bedrock towards the east.
The site has one well for groundwater extraction from the deep groundwater zone. The water is used for non-contact cooling water, which after its use is processed in the site’s wastewater system and subsequently discharged to a municipal sewage treatment facility. A pumping permit allows a withdrawal of between 400-500 m3/day.
Figure 6 Geological cross-section along the western boundary, north (left) towards the south (right).
Figure 7 Geological cross-section across the site (east or northeast towards west or southwest)
3.2.2 Bedrock
The bedrock is generally located at a depth of between 10-20 m below ground level (b.g.l). It mainly consists of granodiorite (Figure 6 and 7). There is a small area of exposed bedrock in the outskirts of the site. The properties of the bedrock have not been thoroughly investigated.
However, earlier geological interpretations suggest it is not fractured. Due to the significant depth to the bedrock, its presence will not be discussed to a significant extent in this study. The occurrence of PFAS and its interaction to its surroundings is instead in the fill material and quaternary deposits.
3.2.3 Soil characteristics
16 soil samples have been analyzed for their particle size-distribution (Table 3). 13 of these samples were from the till, and 3 samples from the clay-rich filling material and natural clay.
The clay content varies with depth but is generally the highest in MW-319. The till samples generally exhibit a granular structure, with a slight content of fines. Three till samples, location in the western part of the site, did have a higher composition of finer material, with a fraction of 20 - 25 %.
Table 4 Overview of the results from the sieve analysis. The analysis was issued on 7 of the monitoring wells.
3.3 Hydrogeology 3.3.1 Conceptual model
The natural clay layer separates two groundwater zones on the site (Figure 6). The perched groundwater zone constitutes the filling material, including the clay-rich filling material. The groundwater level is located in the clay-rich fill material, with a depth which ranges from 1.47- 4.17 meters b.g.l. (October 2018). The variation is considerable and is due to several factors.
Due to the character of the filling material, there is a limited amount of groundwater stored in these deposits.
The deep groundwater zone, present within the moraine, is the primary groundwater zone which yields the most groundwater. Its piezometric level in September 2018 varied from 3.94 meters b.g.l to 8.06 meters b.g.l, which is in the lower part of the natural clay (October 2018). The properties of this clay, in addition to its thickness, enables the deposit to act as an aquiclude throughout the site. However, in certain areas where the deposit is thin, it could rather act as an aquitard. There is a possible interaction between these deposits. Furthermore, the subsurface has piles and pipes present, which could facilitate horizontal or vertical transport.
The deep groundwater zone has, with the exceptions of wells MW-307 and MW-311, lower groundwater levels than the levels for the perched groundwater zone. Hence, there is a downward head between these zones. The natural clay acting as an aquiclude, and the deep groundwater zone is consequently confined. Measurements in MW-308 during 2017 and 2018 were between 7.5 and 8.1 meters b.g.l., which is close to the top of the moraine layer, located at 8 meter b.g.l. These results indicated that the deep groundwater zone is only partially confined in the area, which could be a consequence of the thinner moraine and shallower depth to bedrock.
3.3.2 Flow directions
The groundwater levels of both perched and deep zones are relatively stable, as shown from measurements from 2005-2019. Fluctuations correlate well with each other and are assessed to be of natural origin. Through groundwater measurements done in September and October 2018, maps presenting groundwater levels and flow directions of the perched zone were established (Figure 8). The groundwater of the perched zone generally flows to the south, with minor local variations.
Figure 8 Groundwater contours for the perched groundwater zone. Road and building names have been removed. The direction of flow is generally to the south and southwest.
3.3.3 Hydraulic conductivities
The hydraulic conductivity (kcond) has been acquired by performing several slug tests. In the perched groundwater zone, the kcond is highly variable, from 0.015 m/d to 4.5 m/d. The estimated kcond in the natural clay is much lower, with values between 0.0006 m/d to 0.3 m/day, with a geometric mean of 0.01 m/d.
The kcond in the deep groundwater zone is >1 m/d in 15 out of 17 wells. In five of these wells, the kcond is >100 m/d, which is consistent with the granular material in the soil, i.e. gravel and cobbles. Overall it varies between 1-270 m/d. Two wells, located in the central part of the site, exhibits kcond of 0.012 m/d (MW320) and 0.025 m/d (MW306), which is likely due to a different geological setting.
3.4 Prior investigations
The monitoring and sampling on the site were initialized in 2005. Geotechnical investigations have also been conducted with construction purposes. The investigations with regards to PFAS were initiated in 2016. The areas which have been prioritized to investigate are areas with the presence of sprinklers, pipes for foam distribution, as well as areas of foam release incidents and where the foam concentrate has been stored. The levels of PFAS have continuously been compared to Swedish National Food Agencys values for a limit of action, as well as SGI’s guideline values with regards to PFOS. These values set by Swedish National Food Agency (Livsmedelsverket) and SGI will be presented in chapter 4.5 and 4.6.
The major part of the PFAS distribution is located in the filling material, both in the unsaturated and saturated zone. The aquiclude restricts the PFAS leakage downwards, however certain areas have slightly elevated levels of PFAS. The vertical leakage is, therefore, a possibility in certain areas where the clay is thin, for example, in the center part of the site (MW-311).
Elevated concentration values in soil (above less sensitive land use) have been discovered in the filling material for several heavy metals, including arsenic, mercury and zinc. The groundwater results have, however, continuously been low. It has earlier been determined that these elevated levels were due to the composition of the filling material.
In addition to the field sampling performed within the range of this study, AECOM has collected data from 2005. The data consist of previously collected soil and groundwater samples, as well as measurements of groundwater levels. This data serves as a complement to laboratory analyzes. Below information provides an overview of the data available.
• 509 soil samples analyzed for PFAS-11. A majority of these samples were taken in connection with installation of monitoring wells. A minority of the samples were taken of the top soil. Only soil data from well installation was processed in this study.
• 323 water samples analyzed for PFAS-11, with a majority of the samples being groundwater. These groundwater samples have been taken in wells at least 1-3 weeks after the well installation, and continuously onward as part the environmental measuring and monitoring.
• Certain of the soil and groundwater samples included analysis of metals, PCBs, PAHs, hydrocarbons, VOCs
• Groundwater levels (2005-2018)
• Basic soil characterization from drilling
• Particle size distributions
• Results from geotechnical investigations (undrained shear strengths, borehole logs, pore water pressure, coefficients of consolidation)
• Field parameters (pH, conductivity, DOC, temperature, oxidation-reduction potential [ORP])
• Depth intervals of soil samples
4. Material and methods
4.1 Sampling procedure
Field investigations consisting of sampling and measurement of groundwater levels was performed during October, November 2018 and in March 2019. The field investigations were conducted under AECOM supervision and in line with AECOM’s and the site’s health and safety procedures. The October- and November-2018 sampling was part of the ongoing monitoring of the site, and the sampling in March was conducted with the specific purpose of acquiring soil samples for commercial and laboratory processing.
4.1.1 Groundwater sampling
Sampling was conducted in order of increasing contaminant concentrations to reduce the potential for cross-contamination. The method of sampling was low flow purging, which involves purging a relatively low volume of water, corresponding to at least one well volume.
The purging was conducted at a low rate while monitoring wellhead parameters (pH, temperature, conductivity, dissolved oxygen). These parameters were measured with a YSI 556 multiprobe system and in-line flow-through cell (Figure 9). Monitoring of field parameters was performed every 5 minutes or every time the water in the flow cell had been exchanged. Purging was continued until (i) at least one well volume of water was removed and (ii) the parameter values are relatively stable (i.e. successive measurements show less than 10 % difference).
Sampling should then be performed using the same equipment while pumping at a similar or
lower flow rate. The groundwater level was continuously monitored during purging to assess its stability.
Figure 9 Overview of the setup for groundwater sampling. Approximately 20 wells were purged and sampled during October and November of 2018 and were the foundation for groundwater contour map (Figure 8).
The discharged groundwater was collected in a plastic container (Figure 9). When purging was completed, i.e. when the indicator parameters had stabilized, the discharge line for the peristaltic pump was detached from the flow-cell, and groundwater samples were collected from the discharge tubing straight to appropriate sample containers. The water gradually rinsed down of the tilted plastic container to evade aeration of the sample. The containers were filled to the top, with no bubbles. The wastewater collected in the plastic can was poured into a designated container for PFAS-contaminated water which was later treated through a water treatment system present on site.
The groundwater samples were collected and stored in clean plastic containers provided by the commercial laboratory. Water samples were transported under chain-of-custody to the commercial lab on the same day as the samples were taken. One duplicate sample was collected and analyzed. In addition, one field blank was also analyzed for quality assurance.
4.1.2 Soil sampling
6 soil samples were taken in two different areas of the site; C8 and M6, which are located approximately 200 meters between each other (Figure 10). Both areas have exhibited high concentrations of PFAS according to historical results, e.g. in wells MW-201 and MW-223.
These areas were mainly chosen since there was ongoing excavation works, which enabled an accessible collection of samples. Furthermore, the soils were of different compositions;
therefore, Kd values representative for more than 1 type of soil could be calculated, and thus the results could be applied more widely across the site. The excavations were conducted with the purpose to construct new buildings.
Figure 10 Map of the site, including sampling points M6 and C8 defined. Both areas are subjected to excavation due to future building plans.
In C8, soil samples were taken from a depth of approximately 3 meters, which is below the rest groundwater level. The ground level was approximately 8.2 meters and the bottom of the excavation approximately 4.9 meters. The upper part of the soil is filling material of clay, and its lower part is silty filling material with a heterogeneous character (Figure 11). The soil consisted of silty, heterogeneous filling material. Particles were overall granular, with fractions of stones with a diameter of 5 centimeters. In addition, there were fragments of bricks, metal scrap, wood and glass present.
In relation to the historical sampling, M6 is closest to wells MW-204, MW-216, MW-223, and MW16009. C8 is closest to wells MW-201 and MW-202 (figure 10). A focus will lie on the comparison between wells close to where the soil samples were taken.
Figure 11 Soil sampling from the excavation pit in C8. Digging was conducted to the left of the shovel, and soil was placed in the bags pictured. The soil was heterogeneous with elements of bricks, metal scrap, wood and glass present.
In M6, samples were taken from a depth of approximately 4 meters, which is below the groundwater level. The soil was pure clay and very wet. Water was present in the excavation pit and considering there had been no recent precipitation and the spot was under the groundwater level, the water was considered to be groundwater. Two bottles were sampled and transported to SLU. Pumps were controlling the groundwater in the excavation. These pumped the water out of the excavation for treatment and ultimately, disposal.
Two 5 liter bags were filled with soil from each location. The specific sampling point was the same for all three samples sets. To ensure the samples that were sent to a commercial lab and SLU were corresponding, the two bags were mixed in a joint bag. From this joint bag, containers of soil were filled and sent to the commercial lab. The remainder of the soil in the bags were transported to SLU. The soils were stored in a refrigerated room for approximately 10 days before laboratory processing was initiated.
Table 5 Overview of samples collected and the analysis conducted. *TOC were initially analyzed for TOC at commercial lab 1; however, the results were acquired according to the loss of ignition (LOI), which is a method with a considerable uncertainty. Subsequently, 2 samples were sent to commercial lab 2, for which a method which also takes the loss of carbon dioxide into account.
Sample Soil Time & Date sampled
SLU Laboratory analysis
Commercial lab analysis C8-1
Silty,
heterogeneous filling
09.43 2019.03.19
- PFAS-11, TOC, pH, DM
C8-2 09.45
2019.03.19
- PFAS-11, TOC, pH, DM
C8-3 09.47
2019.03.19
Batch test; Direct- injection and SPE
PFAS-11, 2x TOC*, pH, DM
M6-1
Clay filling
10.15 2019.03.19
- PFAS-11, TOC, pH, DM
M6-2 10:17
2019.03.19
- PFAS-11, TOC, pH, DM
M6-3 10:20
2019.03.19
Batch test; Direct- injection and SPE
PFAS-11, 2x TOC*, pH, DM
M6-
GWA Groundwater from
excavation pit
10:25 2019.03.19
Batch test; Direct- injection and SPE, pH
- M6-
GWB
10:25 2019.03.19
Batch test; Direct- injection and SPE, pH
-
4.2 Laboratory procedure
The batch test was conducted in SLU’s laboratory and following SLU’s safety procedures. The samples were analyzed with direct-injection and solid-phase extraction (SPE). Direct-injection was initially conducted to develop a better understanding of which PFASs are leaching, and SPE to acquire more in-depth results. The overall purpose of the batch test was to determine the desorption behavior of PFAS. Even though the samples are field samples which are processed in the laboratory, the term laboratory derived results are used.
Results from the commercial lab were received shortly after sampling of C8 and M6; therefore, the results could be interpreted before initiation of the laboratory procedure. Considering the results from each location, where the 3 samples of both M6 and C8 showed similar concentrations, 1 soil sample per area and a duplicate were deemed representative to analyze.
This approach also enabled analyzing the two groundwater samples, which were spontaneously sampled in the excavation pit. The pH of the groundwater samples was measured, as these samples were not sent to the commercial lab. A total of 4 samples and 4 duplicates were analyzed.
The soil samples were sieved to ensure that no particles larger than 4 mm were processed. The fraction dry matter was attained from commercial lab results. The liquid to solid ratio (L/S) was set to 10, and 200 mL was determined for the batch liquid. 20 g was divided by the fraction of dry matter for each sample to acquire the actual weight of the soil to process (Table 5).
Table 6 Overview of soil samples which were weighted for the batch test. B is the duplicate sample.
Dry matter [%] Calculated weight [g]
Sample weight A [g]
Sample weight B [g]
M6-3 57.4 34.84 34.90 34.94
C8-3 81.1 24.66 24.90 24.90