Investigations into the use of a novel graphene wool sampler for organic air pollutants

Abstract

With increasing concerns regarding the adverse health effects caused by trace levels of organic air pollutants, it has become ever more desirable to develop air samplers with superior capacities and sensitivities to target a wide range of gaseous trace analytes. Although carbon-based samplers are well known and widely used in the field of active air sampling, the novel GW sampler, presented in this study, has the potential to provide users with a reusable carbon-based sampler, which can be synthesized and assembled easily in-house, with a large sampling capacity due to the substantial surface area graphene has been reported to intrinsically possess. Selected fundamental laboratory-based studies were undertaken to investigate the strengths and weaknesses of a novel graphene wool (GW) sampler as an active air sampler for gaseous volatile organic compounds (VOCs) and semi-volatile organic compounds (SVOCs). In selected studies, the GW sampler was comparatively evaluated to a quartz wool (QW) sampler, made in a similar fashion to the GW sampler, to determine the impact that the graphene layers grown on the QW substrate had on various sampler properties. The GW sampler was also compared to a polydimethylsiloxane (PDMS) sampler, which has historically been used in a wide range of air monitoring studies. The hygroscopicity of a sorbent is pertinent to know before sampling, as water has shown to compete for active sites in hydrophilic sorbents, thereby decreasing the sampling capacity of a sampler and increasing risk of premature breakthrough of target analytes during sampling. It was shown that the GW adsorbed < 0.5% (m/m) water at 42 - 70% humidity, whilst at > 80% humidity ± 50% water was retained on the GW sampler by mass. The hygroscopicity of the GW sampler was compared to QW and PDMS samplers in humidity ranges > 90%, where it was found that the polar QW retained the most water whilst the non-polar PDMS sampler retained the least. Therefore, since the GW showed negligible adsorption of water at humidity ranges < 80%, it was shown that water retention will not be a problem when using the GW sampler under typical ambient conditions. The back-pressure incurred by the GW sampler was compared to that incurred by the QW and PDMS samplers at set flow rates. The back-pressures associated with the GW and QW samplers were found to be far higher and more unstable at flow rates > 400 mL.min-1, as compared to the back-pressures recorded for the PDMS samplers at equivalent flow rates. This may be primarily attributed to the GW and QW samplers having a packed fibrous sorbent structure resulting in a higher back-pressure, which would be more susceptible to instability as compared to the PDMS samplers, which consists of an open-tubular structure. The back-pressure incurred by the GW and QW samplers was observed to be stable at a flow rate of 400 mL.min-1, whilst the back-pressure incurred by the PDMS sampler was found to be stable at 500 mL.min-1. Although it is recommended to sample at a flow rate of 200 - 250 mL.min-1 by Manura (2019) and the International Organisation for Standardization (2001 and 1991), this result is positive as the GW sampler has shown low flow resistance. The application of both the plunger assisted solvent extraction (PASE) (Munyeza et al. 2018) and thermal desorption (TD) extraction methods were evaluated to determine the optimal extraction method for the GW sampler when targeting SVOC analytes such as polycyclic aromatic hydrocarbons (PAHs). PASE is cheaper than TD and allows for multiple analyses per sample, but has the disadvantage of diluting the sample which reduces sensitivity. The initial PASE study showed that the first extraction was ineffective in extracting the PAH analytes with hexane, as the bulk of the analytes were only extracted in the second and third successive extractions. An overestimation of the extracted PAHs was observed, which could be attributed to solvent losses. The second PASE study involved an 8 hr hexane soak of the GW in the sampler prior to PASE, which proved to be a promising procedure which should be investigated further, as the PAHs were effectively extracted in the first extract. TD of the GW sampler resulted in good extraction efficiencies for the lighter PAHs such as naphthalene (97% recovery) however, the extraction efficiency was observed to decrease as the molecular mass of the PAHs increased. An example of which is the 1.8% recovery of dibenz[a,h]anthracene from the GW sampler. This result indicates that the larger molecules are more susceptible to being irreversibly adsorbed onto the GW and therefore should not be targeted when using the GW sampler. It may thus be advantageous to optimise PASE, whilst exploring solvents used such as CS2 and acetonitrile, for PAH extraction from GW samplers. Intra- and inter- sampler variability studies involved the spiking of VOCs and SVOCs (hydrocarbons (HCs) and PAHs) onto fresh GW samplers with subsequent TD-GCxGC-TOF-MS analysis. It was found that VOCs with higher LogKow values had lower variability in the peak areas recorded as compared to compounds with lower LogKow values. This may be due to the more non-polar compounds adsorbing with greater reproducibly onto the GW sorbent. The C8 - C20 HCs showed increasing % RSDs (2.8 - 55%) with increasing molecular masses for the intra-sampler study, as expected from the previous TD study, whilst the inter-sampler study showed overall higher % RSDs (35 - 76%). The analysis of inter-sampler variability of thermally desorbing the PAHs from the GW sampler reflected % RSDs of approximately twice that of the intra-sampler analysis. This was likely due to non-uniformity of the graphene layers in the GW material, which was confirmed by transmission electron microscopy (TEM) images of the material, in which it was shown that fibres from the same GW batch did not possess uniform graphene layering growth on the QW substrate. The inconsistency in the uniformity of the graphene layers may thus be the primary cause of discrepancies observed between different GW samplers. Sampler retention volumes (RVs) were investigated for GW, PDMS and QW samplers for nine analytes with varying boiling points (BPs) and polarities using a GC-FID system, in which the GC column was replaced with the sampler of interest. This study was undertaken to better understand the RV trends exhibited by the GW sampler as opposed to the QW and PDMS samplers. It was found that GW had superior RVs for three of the nine analytes, all polar, as compared to the other two samplers. This was contrary to initial assumptions that the GW sampler would show a greater retention of non-polar compounds. This trend is likely attributed to exposed QW fibres in the GW, which would innately exhibit a higher affinity for polar compounds. The low RVs (ranging between 29 mL.g-1 sorbent for hexane to 563 mL.g-1 sorbent for cyclohexanone) for the GW sampler may be attributed to inconsistencies in packing of the GW resulting in channelling taking place with localised saturation of adsorbent sites. Therefore, optimisation of the packing of the GW sorbent is required. Following the laboratory-based studies, the application of the GW sampler was tested for sampling the combustion emissions from three different fuel sources using a combustion aerosol standard (CAST). Emissions were simultaneously sampled onto GW, PDMS and commercial activated charcoal tube samplers. The GW and PDMS samplers were directly thermally desorbed and analysed by GC-MS whilst analytes sampled onto the activated charcoal were extracted with CS2 prior to GC-MS analysis. Rapeseed oil methyl ester (RME) and gas-to-liquid (GTL) fuels were found to be better fuel alternatives to diesel (B0) due to lower VOC and SVOC emissions. In terms of the performance of the samplers during this sampling campaign, the activated charcoal was found to be ineffective as it showed high background noise and required liquid extraction, which diluted the sample to below the limit of detection (LOD) of the GC-MS. The results from the TD of the PDMS samples reflected relatively higher concentrations of the targeted VOC and SVOCs as compared to the GW sampler, however TD of the GW samplers showed that the target analytes were quantified with lower % RSDs than for the PDMS sampler, particularly for the HCs. Lighter HCs, such as octane and nonane, were only effectively sampled by means of the GW sampler. These results indicate that the GW sampler may be used to target trace concentrations of lighter non-polar analytes. Overall, it was found in this study that the GW sampler requires further optimisation prior to widespread application. However, the GW sampler is seen to be a promising candidate as an active air sampler for VOCs and lighter SVOCs.