2. Methods

Air samples were collected using two types of active air samplers available at the sampling sites: i) high-volume air samplers (Digitel, flow rate 20–25 m³/hr) in Oslo, Reykjavik, Reykjanesbær and Birkenes, and ii) low-volume air samplers (Comde Derenda, flow rate 2.3 m³/hr) in Helsinki. All samples from the urban sites were collected on filters only, that means that only the particle phase in air was collected and analysed. This sampling approach was selected based on the available active air samplers at the sites and to obtain comparable data between the sites. The use of filters only most likely underestimates the real air concentrations of SCCPs and to some extent MCCPs that are semi volatile and present in both gas phase and particle phase in air. Estimation of expected gas-particle distribution suggests that the bulk concentration can be up to ten times higher than the particle concentrations alone. The filters for the high-volume samplers were glass fibre filters with a diameter of 150 mm while the filters for the low-volume sampler were Fluoropore Membrane Filters with a diameter of 47 mm. The sampling time was 48 hrs with the high-volume sampler and 24 hrs with the low-volume sampler. In order to obtain samples with similar sample times, the filters from Finland were bulked in in pairs before sample extraction. The samples from the background site were collected in accordance with the routine monitoring programme, that means with filter and Polyurethane foam (PUF) plugs collecting both particle and gas phase. This hampers the comparability to the urban sites but is even so included here.

Samples were collected in wintertime when studded tires are in use (January–April 2022) and in summertime when summer tires are obligatory (May–September 2022). The number of samples per season are presented in Table 2.

Prior to extraction, the samples had isotope labelled SCCP (1,5,5,6,6,10-hexachlorodecane) added for quantification purposes. The filters were extracted for 6 hrs in acetone:hexane (1:1) using Soxhlet extraction. The extracts underwent a clean-up procedure using concentrated sulphuric acid and a silica column to remove lipids and other interferences. Subsequently, solvent was changed to isooctane, and extracts were concentrated using a gentle nitrogen flow. A recovery standard of 1,2,3,4-tertachloronapthalene was added prior to analysis.

Analysis of chlorinated paraffins covered SCCPs, MCCPs, and LCCPs. SCCPs and MCCPs were analyzed using an Agilent 7890B GC, 7200 QToF (GC/HRMS). To achieve a necessary separation, an HP-5MS UI 15 m×0.25 mm id, fused silica capillary column was used with a constant Helium flow of 1.2 mL/min. PTV injection was applied in solvent vent mode. The GC temperature program for SCCPs and MCCPs was: 55 ˚C (2 mins), 70 ˚C/min to 200 ˚C (1 min), 10 ˚C/min to 280 ˚C (1 min), 5 ˚C/min to 300 ˚C (0 mins), 70 ˚C/min to 320 ˚C, 1 min. The MS was operated in ECNI mode using methane as moderating gas. A selection of [M-Cl]^- or [M-HCl]^- ions for both SCCPs and MCCPs were extracted and the quantification was performed based on the deconvolution method described in Bogdal et al. (2015). Briefly, C10 - C13 standards with 51%, 55% and 63% Cl (w/w), and single chain length standards of C10, C11, C12, and C13 (50% and 65% Cl) were analysed for short chain CPs (SCCPs), and for medium chain CPs (MCCPs), C14 - C17 standards of 42%, 52%, and 57% Cl, and single chain C14 with 52% Cl were analysed. The resulting congener group patterns were used to reconstruct the congener group patterns in the samples. To do this we used the Lawson-Hanson algorithm (nnls package in R Studio) to obtain non-negative least squares estimates for the contributions of the individual standards. Separate calibration curves with four concentration levels were prepared for each of the included standards. All integrated areas from samples and standards were normalized using isotope labelled internal standard. 

LCCPs were analyzed using an Agilent 1290 UHPLC, 6546 QToF (LC/HRMS). To achieve a necessary separation, an ACE Excel 5 Super C18, 75x2.1 mm column was used with a constant flow of 0.4 mL/min. The mobile phase was at start 70% water and 30% MeOH both containing 0.05mM of tetramethylammonium chloride. After 5 mins the mobile phase was adjusted to 100% of MeOH with 0.05mM of tetramethylammonium chloride. The MS was operated in negative ESI mode. A selection of [M+Cl]^- ions were extracted and the quantification was performed based on (Bogdal et al., 2015) as described above. For LCCPs, the availability of standards is more limited than for SCCPs and MCCPs. So, for LCCPs four technical mixtures of C18 – C30 with chlorination degree from 36% to 49% were used.

Lab blanks and field blanks were analysed along with the samples. All samples including field blanks were corrected for the levels in the lab blanks from the same extraction batch. The limits of detection (LoD) for all chlorinated paraffins were calculated using 3 times the standard deviation for all lab blanks.

When performing environmental screening studies for contaminants of emerging concern, all steps in the process, starting with study design, selection of the sampling sites, sampling frequency, time of sampling, performing the sampling, the transport and storage of samples, chemical analysis and data treatment, to some extent generate some degree of uncertainty (Thomas et al., 2014). To quantitatively estimate the contribution of all steps is challenging. However, we will discuss the relevance of the different contributions in a qualitative way.

Sampling and analysis of non- or recently regulated organic contaminants of emerging concern that are still in use (e.g., S/M/LCCPs) are associated with a bigger uncertainty than the well-established regulated legacy POPs. This is due to more diffuse sources in laboratories and sampling facilities (e.g., the use of CPs has increased again in a lot of different industrial, household products and consumer goods during the last years) that results in a larger risk for contamination. NILU is continuously taking actions to minimize this influence. Examples of such measures are improved pre-cleaning of analytical equipment, isolated work in clean cabinet facilities and removal of some materials and products where CPs are known to be present. However, samples cannot be sampled, stored, extracted and prepared for analysis without any physical contact with a lot of different materials and instruments. This raises the number of blank samples exceeding acceptable level, which in consequence raises the LOD for samples analysed in parallel with those blank samples. Therefore, all samples are corrected for the levels in the lab blanks. Further, samples are compared to the levels in the field blanks collected at corresponding locations to assess possibilities of contamination during sampling and transport.

Based on a 3-level categorization of quality and uncertainty in analysis of environmental samples, the quality and uncertainty of the analysis of the targeted S/M/LCCPs in this study fall into the lowest category:

Category 3: No or few reference materials or satisfying interlaboratory studies available. The methods are less reproducible, and the results have higher uncertainty.

The measurement uncertainty for the measurement of CPs is estimated to be much higher than for other POPs and probably higher than ± 40%.

There are international efforts to improve the analysis of CPs ongoing in which NILU is taking part. In 2021, NILU and nine other selected laboratories were invited by the European Commission to contribute to a certification and intercalibration study with the aim of certifying a reference material for SCCPs. The laboratories had to pass multiple quality assurance tests to be considered for the final evaluation of the material to be certified. NILU was one of five laboratories to pass the quality assurance tests, and thus providing data for the certification process of this reference material which became available in November 2022.

The uncertainty of the chemical analysis is governed by sample homogeneity, loss during extraction and clean-up, interference from other compounds, trueness of analytical standards, instrumental parameters, and contamination at all stages of sampling, transport, and laboratory work.

A normal approach to estimate and quantify these factors is the participation in a laboratory intercalibration. The uncertainty is expected to be larger for compounds which are analysed infrequently than for compounds which are analyzed commonly. For the CPs, where only a mixture analysis is possible the uncertainty is estimated to be higher than the normal 20%. SCCPs, MCCPs and LCCPs consists of a multitude of different congener groups. Furthermore, each congener group potentially consists of several thousand different isomers where complete chromatographic separation is impossible to achieve with present technology. Hence, single congener standards are not possible to isolate from a crude mixture and the ones available are very limited and rarely chemically representative. The individual instrumental response factors within the different congener groups will then not be possible to determine when the signals of the single congeners are overlapping. The responses will most likely depend greatly on the chlorine positions and are therefore subject to great variations. One of the most common approaches to quantification is to use average response factors from a number of standard mixtures with the closest resemblance to the samples as possible. Another approach which has been widely accepted during the last years, and has also recently been implemented at NILU, is the deconvolution method as first described by Bogdal et al. (2015). This method is based on the linear combination of several standard mixtures to determine response factors which is applied to the samples. The limitations are again the standard mixtures available, but most likely this approach is by now the most promising way to quantify CPs in environmental samples. Another addition to the great uncertainty in GC/MS analysis of CPs is increasing response factors with increasing concentrations. This means that not only is it important to incorporate standards or a combination of standards with pattern resemblance, but also incorporate standards with responses within each congener group corresponding to the samples. The latter is much more important in CP analysis than traditional single compound analysis even when the responses are within the instruments linear area.

We consider the analytical uncertainty as adequate for a screening study. However, as for all other available CPs data, caution should be applied if using these results for future comparisons and time trend studies.