Characterisation of aquatic natural organic matter by micro-scale sealed vessel pyrolysis

Abstract

The analytical capacity of MSSV pyrolysis has been used to extend the structural characterisation of aquatic natural organic matter (NOM). NOM can contribute to various potable water issues and is present in high concentrations (> 5 mg L-1) in many Australian source supplies. NOM can also impede the filtration performance of ultrafiltration or other membranes used in the increasingly popular practices of desalination and wastewater treatment. Characterisation studies that provide a detailed understanding of the origins, structural features and reactivity of NOM in source waters will help predict its impact on potable supplies and allow targeted treatment.MSSV pyrolysis GC-MS analyses were conducted on XAD fractions of NOM from selected rivers, reservoirs, ground waters and biologically treated waste waters. The analytical sensitivity of the MSSV Py approach was demonstrated by the detection of high concentrations and complex distributions of pyrolysates. These included many additional products to those detected by corresponding flash pyrolysis GC-MS analysis, which is often limited by excessive degradation or poor chromatographic resolution of pyrolysates of high structural polarity. Nevertheless, flash pyrolysis did lead to several unique products from some samples, reflecting the complementary nature of the two methods.Despite the high product concentrations detected by MSSV pyrolysis of NOM, primary structural fragments are prone to further alteration due to the confined nature and extended (e.g. 72 hr) application of the moderate thermal conditions (e.g. 300°C). This approach has not been widely applied to the characterisation of recent or immature OM. Consequently, the mechanistic formation of many NOM pyrolysates is poorly understood, seriously limiting interpretation of their source and significance. As articulated in CHAPTER 1, these issues are specifically addressed by the present research, which aims to extend the application of MSSV pyrolysis to the characterisation of NOM and related environmental organic materials rich in intact biochemical inputs. To gain a better understanding of product formation pathways, several samples, including soil leachates, the organic foulant of ultra filtration membranes and a suite of standards representing potential biochemical precursors of NOM were separately analysed by MSSV Py. The effect of thermal conditions on product distributions was also addressed by analysis of a small sub-set of the samples at several different temperatures (260 – 330°C for 72 hours).The capacity of MSSV Py to convert functionalised biochemical precursors into hydrocarbon products more amenable to GC resolution was initially demonstrated by the conversion of bacterial hopanepolyols of several surface and ground water NOM fractions, a bacterial isolate and biomass growth from an ultrafiltration membrane into corresponding hopane biomarkers as described in CHAPTER 2. The significance and integrity of the hopane distribution of the MSSV data was assessed by analyses of the same samples by flash pyrolysis and the advanced analytical techniques of hydropyrolysis (HyPy) GC-MS and liquid chromatography (LC)-MS with atmospheric pressure chemical ionisation. Flash pyrolysis showed no evidence of hopanes.In comparison to the distributions of intact biohopanoids detected by LC-MS, the microbial hopane biomarker signatures detected by MSSV Py and HyPy were generally consistent, although HyPy did produce higher concentrations of ββ−diastereoisomers and higher MW fragments indicating a lower degree of structural alteration. Hopane products were detected in very low concentrations in the NOM samples, hence bacterial contribution may be more conveniently detected with biological methods (e.g., microbial arrays, bacterial counts). Nevertheless, MSSV pyrolysis represents a simple, low cost analytical method able to confirm the occurrence of diagnostic bacterial biomarkers in complex environmental settings, such as source waters and surrounding catchments, and may be a useful screening method prior to more involved characterisation possible with LC-MS. Moreover, this application represents an elegant demonstration of the capacity of MSSV pyrolysis to provide new information concerning functionalised biological precursors which have historically proved difficult to analyse by GC(MS).Additional source diagnostic molecular features detected by MSSV Py of the membrane biofoulant included sterane biomarkers of eukaryote triterpenoids (i.e. steroids), n-alkanes of fatty acids and C16-C19 phenylalkanes indicative of common surfactants used to clean the membranes. The vastly improved molecular characterisation of the polar lipid constituents of membrane foulants, including the identification of industrial chemicals used in cleaning processes, suggests that this analytical capacity might also be applicable to monitoring the fate of organic constituents through the entire potable water system, from source through treatment and distribution to tap.Unlike the established bacterial hopanoid source of hopane biomarkers, the origins of most of the major products from MSSV pyrolysis of the NOM samples are not clear. Subsequent chapters were separately dedicated to a detailed investigation of several of the major product classes.CHAPTER 3 focused on the alkyl aromatic pyrolysates of NOM. The multitude of potential precursors of alkyl substituted benzenes and polycyclic aromatics (e.g. naphthalenes, phenanthrenes) significantly limits their diagnostic potential, nevertheless they represented a major proportion (20 – 50 % of total GC amenable pyrolysate signal) of the MSSV, and to a lesser degree, the flash pyrolysates of the HPO fractions of several NOM samples. The more highly substituted alkyl aromatics (and heteroatom products) of potentially greater source diagnostic value were better preserved by MSSV Py. Distinctive distributions of alkyl aromatics were detected by MSSV Py of the HPO fractions of several surface waters and a lysimetric plate collected ground water. All samples showed high alkyl benzene (AB) concentrations, whilst the ground water showed higher alkyl naphthalene (AN) concentrations than the surface waters. Correlation of several isopropyl substituted benzenes indicative of plant resin terpenoids in a bark sample, suggested these may be a significant source of the alkyl aromatics products of NOM. Furthermore, several higher plant derived polycyclic aromatic terpenoid biomarkers (e.g. cadalene, eudalene, retene and dehydroabietins) were also identified in the NOM fractions. Allochthonous and autochthonous sourced terpenoids have been proposed to be significant precursors of aquatic NOM; however diagnostic flash pyrolysis information about these types of contributors is typically limited.HPO fractions of two waste waters also showed consistently high concentration of alkyl aromatics, reflecting the general recalcitrance of their precursors to biological treatment. The distributions of these products differed from the natural surface and ground waters. Resistant aliphatic biomolecules derived from algal and bacterial biomass, susceptible to cyclisation and aromatization during MSSV Py, were tentatively assigned as the source of these distinctive pyrolysates.MSSV pyrolysis proved particularly sensitive to detection of heteroatom containing products of the NOM samples, and O products (25 – 50 %) and S products (1 – 5 %) were the focus of CHAPTER 4. The alkyl (≤ C4) phenols (APs) of the HPO fractions of the humic rich Gartempe and Uruguay rivers accounted for ca. 40 % of the total product signal. Similarly high concentrations of APs were detected by MSSV Py of a lignin standard, demonstrating the laboratory simulated thermal transformation of methoxy phenolic structures into alkyl phenols. The high concentrations of APs and low concentrations of methoxyphenol biomarkers of lignin typically detected in NOM (e.g. by flash pyrolysis, 13C NMR) suggests that a similar structural change may also be diagenetically mediated. The detection of APs, therefore, may be a more sensitive indicator of lignin input than guiaicyl or syringyl based biomarkers. The polyphenol structural units of selected tannin standards did not survive MSSV Py treatment, so are not likely responsible for the AP MSSV pyrolysates of the NOM samples. The HPO fraction of the waste waters showed similarly high concentrations of APs and alkyl aromatics (as discussed in Chapter 3), suggesting these products are recalcitrant to biological treatment. The Naintré waste waters also contained additional higher MW C4-10 alkyl substituted phenols not detectable by flash pyrolysis. Several of these products were indicative of industrial chemicals of potential health concern.The TPI and COL fractions showed significant concentrations of alkyl furans which along with cyclic ketones were present in much lower abundance than APs in the HPO fractions. These products were attributed to carbohydrate sources following correlation with mono- and polysaccharide standards including glucose, cellulose and chitin. Trace or low relative abundances of these products in the biologically treated wastewaters reflects their vulnerability to mineralization. The HPO fractions of the NOM samples also showed low concentrations of alkyl benzofurans, similar distributions of which were detected in the SRFA standard, suggesting these are more stable polysaccharide metabolites, but still prone to further biodegradation as evident by only trace concentrations detected in the waste waters.Whilst MSSV provided increased access to several S-structural constituents of NOM, their relatively low concentrations and as yet undefined structural origins remain a challenge to source characterisation. Nevertheless, similar alkyl thiophene (AT) distributions were also generated from an S-containing amino acid standard. Notably higher concentrations of S-products in the waste waters may reflect additional anthropogenic sources (e.g. sewerage, industrial chemicals), which may also involve thermally catalysed reaction between H2S and humic substances, analogous to the interaction of inorganic S and functionalised lipids during sedimentary diagenesis.CHAPTER 5 was concerned with the notably high concentrations of N products (3 - 50 %) detected by MSSV Py GC-MS of the NOM samples. These products included a large range of alkyl- pyrroles, pyridines, pyrazines and pyridinamines, as well as amine substituted mono-aromatics and condensed N-heterocyclics. They were consistently detected over a broader range of MSSV Py conditions. Many of these products, particularly those with increased alkyl substitution were not detected by flash pyrolysis; leading to an historic underestimation of their contribution to NOM. Highest concentrations of N-products were detected in the COL fraction of NOM. Similarly high concentrations and distributions were also detected from the organic material prone to foul ultrafiltration membranes, confirming the colloid rich nature of this material. The distinctive low MW N-heterocyclic products of the COL fractions were correlated with the N-products of the amino sugar standard, and to a lesser extent the protein standards. The occurrence in high concentrations of low MW heterocyclics also provides potentially rare evidence for the environmental occurrence of Maillard reactions. The interaction of sugars and amino acids via Maillard processes may be an important contributor to humic substances, although there is much doubt about whether this process is supported by ambient or near surface temperatures. As these reactions are more favourable at high temperatures they may be artefacts of the MSSV Py process. However, MSSV Py of mixtures of carbohydrate and amino acid standards showed no evidence for the production of additional low MW heterocyclics. The waste waters showed relatively high concentrations of alkyl carbazoles and β-carbolines, potentially derived from alkaloid precursors of plants, algae and bacteria, which have been implicated in toxic N-DBPs from potable water treatment.To practically assess the analytical benefits of MSSV pyrolysis for NOM characterisation, it was used in combination with other established analytical methods to holistically characterise the NOM of the North Pine (NP) reservoir, a major source of the potable water supplies of Brisbane and SE Queensland. The NP water is of low colour and has moderate dissolved organic carbon (DOC; 5 mg L-1) levels, but is impacted by algae which periodically occur in bloom proportions. The hydrophobic (HPO; 65 % initial DOC) and transphilic (TPI; 12 %) fractions from XAD resin separation of the DOC both showed high (>1) H/C values, low UVabs characteristics and low aromatic-C measured by NMR, which are all indicative of a relatively low degree of aromaticity. However, MSSV Py of both fractions, in particular the HPO fraction, yielded prolific distributions of alkyl substituted aromatic hydrocarbon (i.e., benzenes, naphthalenes) and hydroaromatic (e.g. tetralins) products. These were attributed to aromatisation of aliphatic structural precursors, including terpenoids of algae and plants, which are usually difficult to detect by analytical pyrolysis. MSSV Py of both fractions also yielded high concentrations of alkyl phenols, which likely reflect contribution from non-methoxylated lignin units of catchment grasses, consistent with the vast forest cleared grassland regions of the NP catchment, but may also derive from algal biopolymers. None of the analytical methods used showed any significant evidence of dihydroxy or methoxy aromatic structures of wood lignin or tannin inputs.MSSVpy of the TPI fraction showed very high abundances of N-products (e.g., alkyl pyrroles, pyridines, indoles) reflecting the structural significance of diagenetically altered proteins, most likely derived from algal biomass. In contrast, much fewer Nproducts were detected by flash Py. This demonstrates the analytical capacity of MSSV to access the significant N content of this fraction, which was quantitatively indicated by low C/N ratio, measured by elemental analysis, and high amide and amine signals by 13C NMR and FTIR spectroscopy.Whilst MSSV generated much higher product concentrations than flash Py or TMAH thermochemolysis, the latter methods did include unique product information demonstrating the complementary nature of different pyrolysis methods. Overall, this case study demonstrates the significant contribution MSSV Py has made to characterising the structure and sources of the Brisbane source water, clearly distinguishing it from humic black waters such as the Gartempe, Arroyo Sanchez and Suwannee Rivers studied in preceding chapters.This PhD project represents the first detailed study of the potential of using MSSV Py to assist the organic speciation and molecular characterisation of biochemically rich NOM. Important additional pyrolysis information can be released with this analytical method which represents an obvious complement to conventional flash pyrolysis techniques where chromatographic resolution of polar biochemicals can be limited. The full realization of this approach, however, will need much further development as briefly alluded to in the closing comments of CHAPTER 7. It is hoped that the present project makes a significant early step in the realization of this potential.