Ammonium is a common contaminant found in the soils and groundwater at former gasworks, associated with historical gas production and the storage and disposal of by-product. However, it can also be present in groundwater at gasworks sites from a variety of natural and other anthropogenic sources. This study evaluates the use of nitrogen isotope analysis at eight former gasworks sites in the UK as a forensic tool to differentiate between ammonium from gasworks and non-gasworks sources. It also provides an understanding of how the parent coal, gas-making technology and by-product processing can influence the presence of ammonium on a former gasworks, and the importance of understanding the site layout when sampling.

Results of this study indicate that gasworks sources would be indicated typically by a δ15N of between −3.2 and +10.7‰, which correlate to published isotopic ranges specifically for coal and coal by-products. This broad range includes published values for the isotopic signature of parent coal (δ15N of −3.2 to +6.3‰,), coal pyrolysis residue/tar (δ15N of 4.2–10.7‰,), gas-purifier waste (δ15N of 2–5‰) and coking-works-derived ammonium sulfate (δ15N of −0.5‰). This suggests that gas-purification waste may have a distinct isotopic range compared to coal tar, a finding supported by results from Site A. Gasworks-sourced nitrogen typically had a lower δ15N value than non-gasworks sources, and predominantly in the δ15N(NH4) form. This study demonstrates nitrogen isotope analysis coupled with traditional hydrochemistry and a detailed site investigation is shown to have potential for use as part of the environmental forensic toolkit.

During the nineteenth century, the pyrolysis of coal became a major source of gas for industrial and domestic lighting, heating and cooking in Great Britain and the wider world. During the mid-twentieth century, coal-gas production declined due to the lack of quality gas-making coals and the introduction of new methods of gas production from petroleum feedstocks (e.g. naphtha) and, later, the widespread use of natural gas. Gas manufacture was eventually phased out in Great Britain between 1966 and 1976 (Thomas 2020). More than 3600 former gasworks sites have been recorded in England alone (Thomas 2020). These former gasworks varied in size from small private gasworks of no more than 50 m3 in area, which would have supplied a single country house, up to the largest city gasworks, which were over 200 ha in area. In addition to the variations in size, the production processes used on former gasworks evolved through the industry's long history (Thomas 2020). Many of these sites have already been redeveloped or will be redeveloped for other uses in the future.

Ammoniacal nitrogen (comprising dissolved NH3 and NH4+) is one of the common contaminants of concern associated with former gasworks sites. Ammonia (NH3) is an uncharged polar molecule, which is a strongly basic colourless gas that is readily soluble in water. The ammonium cation is a positively charged polyatomic ion with the chemical formula NH4+. It can be formed by the protonation of ammonia and exists as free ions in solution or as crystallized salt compounds. The major factor that determines the proportion of ammonia or ammonium in water is pH. The activity of ammonia is also influenced by temperature and ionic strength. This is important as unionized NH3 is most harmful to aquatic organisms. Both forms have the potential to impact groundwater and surface-water resources; however, ammonia is toxic to aquatic fauna even at very low concentrations and its allowable concentrations in the environment are controlled through several regulatory guidelines (Erskine 2000; Buss et al. 2004). Ammonium (NH4+) is listed in the EU Drinking Water Directive (2020/2184), with a threshold of 0.5 mg l−1 (EU 2020).

Objectives

The overall objective of this study is to develop the findings of Robinson et al. (2006), and to evaluate the use of nitrogen stable isotope analysis as a tool to discriminate between gasworks and non-gasworks sources of ammoniacal nitrogen to groundwater. The first objective of this study was to review published data that supported the hypothesis that sources of ammonium on former gasworks would have distinctive isotopic signatures that would distinguish them from non-gasworks sources. The second objective was to identify former gasworks sites where there were known to be elevated concentrations of NH4 in the groundwater, and where the existing monitoring infrastructure was suitable for further isotopic sampling. The third objective was to identify monitoring points within each site that may be impacted directly by the gasworks, and those monitoring points that were potentially impacted by non-gasworks sources of ammoniacal nitrogen. The final objective was to undertake sampling and analysis of the groundwater quality including δ15N(NH4), δ15N(NO3) or δ15N(total) and δ18O(NO3) isotopes, major ions, and different forms of dissolved nitrogen species.

The origin of nitrogen in coal

Coal is formed from the remains of dead plant material that accumulated as surficial deposits of peat and have been compacted, hardened, chemically altered and metamorphosed by the effects of pressure and sometimes heat over geological time. The plant matter forming the coal was predominantly from prehistoric forests and marshes but was subject to variation, ranging from algae to trees (Davidson 2004). As the plant matter is compressed it proceeds through the coalification process. During this process, the carbon content in the coal increases and the oxygen content decreases. As the coal progresses from the lowest rank form (lignite) to the highest rank form (anthracite) it becomes more ordered, forming a variety of large aromatic ring structures, its hardness increases and the proportion of volatile components decreases (Davidson 2004).

Stable isotopes had been applied to the study of coal, mostly focusing on carbon, but other studies have focused on nitrogen, oxygen and sulfur stable isotopes. The study of nitrogen isotopes has become important in evaluating its contribution to NOx emissions from coal combustion (Feng et al. 2020). Nitrogen has two isotopes: 14N (99.63% abundance) and 15N (0.37% abundance); due to the much higher abundance of 14N, measured isotopic values are reported as δ15N in parts per thousand (per mil: ‰). The isotopic composition of nitrogen in a sample is conventionally expressed as δ15N (δ15N (‰) = {(15N/14N) sample/(15N/14N) air − 1} × 1000).

Carbon (δ13C) and nitrogen (δ15N) stable isotopic signatures can be used to infer the composition of vegetation as well as changes that occur during early diageneses and coalification (Masood et al. 2022). Carbon stable isotope signatures (13C) are more typically studied for coal as they can be used to assess environmental changes that may occur during peat accumulation, such as air temperature, humidity and soil moisture. Nitrogen stable isotope signatures (15N) are less well studied for coal but can be used to infer the organic matter sources and the peat-forming vegetation (Peri et al. 2012; B. Liu et al. 2020; J. Liu et al. 2020; Masood et al. 2022). Nitrogen isotope profiles are now often used as proxies for depositional redox conditions, nitrogen cycling and nutrient uptake in modern and ancient marine systems (Quan and Adeboye 2021).

Observations of the potential correlation between coal rank and N isotope values vary. Stiehl and Lehmann (1980) found a correlation between the N isotope values and the rank of some European coals, with δ15N values increasing from 3.5‰ in the low-rank coals to 6.3‰ in anthracites. Work by Ader et al. (1998) suggested that as coal rank increases the loss of organic nitrogen from coal only occurs rapidly during anthracitization, the change of bituminous coal into anthracite by pressure or heat. This is independent of organic nitrogen isotopic composition, which does not change with rank increase. The same trend was observed by Masood et al. (2022) within Paleocene coals from the Salt Range in the Punjab. Rigby and Batts (1986) found the δ15N values in coal samples from the Beaver Lake Formation in Antarctic to be 3.0‰, coal samples from Australia to range from 0.3 to 3.7‰ and coal samples from the Stockton formation in New Zealand to be 1.3‰. The analysis of Canadian coals by Whiticar (1996) ranged from −3.2 to 1.4‰, and Ding et al. (2018) gave a range of −2.9 to +3.6‰ for various Chinese coals. Research by Feng et al. (2020) assessed δ15N, amongst other factors, in a range of coal types. They identified δ15N values between 1.86 and 4.35‰ in Russian coal, δ15N values between 2.46 and 3.4‰ in Australian coal, and δ15N values between 0.38 and 2.32‰ in Indonesian coal.

Ader et al. (1998) proposed that nitrogen isotope data may be significant indicators of the precursor plant material, the environment of deposition and the type and degree of alteration of the plant substances. Handley and Raven (1992) demonstrated that degradable stem xylem has more positive δ15N values than those found within the degradable leaves. Rimmer et al. (2006) identified a potential enrichment of δ15N in vitrinite over inertinite; this being due to the preferred release of δ14N due to bacterial activity during the peat-forming stage and woody tissue alteration during the early diagenic stage. The dominant factors responsible for the isotopic composition of organic nitrogen in coal are the original plant assemblages and the depositional environment (Handley and Raven 1992; Whiticar 1996; Rimmer et al. 2006; B. Liu et al. 2020; J. Liu et al. 2020).

The coal and the gas-making process

Coal formed the basis of gas production in the UK from the first decade of the nineteenth century until the 1950s. From the 1890s the Carburetted Water Gas (CWG) process was adopted in Great Britain, which used oil as a feedstock to enrich the gas. Isotopic analysis undertaken by Hoering and Moore (1958) on US crude oil (the source of most gas oils used in British gasworks) identified δ15N values between 1 and 6.7‰ within the wider range observed for coals. The CWG process was only used in larger gasworks at times of peak demand and formed only a small part of the overall gas production, so the nitrogenous substances within gasworks by-products were primarily sourced from the feedstock coal.

Only specific types of coal could be used for the manufacture of gas. In Britain, the coal used for manufacturing gas at gasworks was hard coal, which had a high volatile content with medium to strong caking properties, although slightly caking coals could be used in vertical retorts (Thomas 2020). These coals covered British National Coal Board coal types 401, 501, 601 and 701 (Speight 2012). Prior to the invention of the gas mantle, the incorporation of some cannel coal was preferred for gas production. This oil-rich coal produced a gas with a greater quantity of volatile organic compounds that gave better illuminating properties when burnt. Anthracite could also be used to form the fuel bed in producer gas or water gas plant, this seldomly occurred in Great Britain as coke was widely available as a by-product of the manufacture of gas using retorts.

The process of manufacturing gas from coal consisted of heating coal in a sealed long tubular vessel called a ‘retort’, in an oxygen-free environment the heat caused the coal to break down, releasing flammable gas and other by-products. The gas industry first used small benches of manually loaded horizontal cast-iron retorts. These were heated by the radiant heat from a furnace beneath the retorts, which burnt coke, and achieved temperatures within the retorts of c. 600°C. As technology developed, higher temperatures could be achieved through the development of regenerative heating of the incoming air; the application of gas producers that burnt the gas adjacent to the outside surface the retorts to maximize heating efficiency; the use of more heat-resistant materials (fireclay and silica) to manufacture the retorts; and greater mechanization. These improvements allowed temperatures for thermal decomposition within the retorts to eventually reach c. 1200°C (Thomas 2020).

As retort design evolved, inclined and vertical retorts developed from the original horizontal retort design, these designs were subject to different heating profiles, with lower sections of the retort being heated to higher temperatures than the upper sections (Thomas 2020). As technology progressed further, by-product coke ovens were developed by the coke-making industry with gas as a by-product. This development was mirrored by the gas industry in the form of smaller chamber ovens and, together with high-temperature horizontal retorts, represented the carbonization processes that exposed coal to the greatest degree of thermal decomposition.

When coal is thermally decomposed, substantial amounts of nitrogen oxides (NO and N2O) are formed from the coal-bound nitrogen, along with hydrogen cyanide (HCN) and ammonia (NH3) that act as reaction intermediates. The latter two are the predominant nitrogenous species within the released gases (Wojtowicz et al. 1995). Wojtowicz et al. (2001) stated that it was not clear whether NH3 and HCN are released independently or whether NH3 is released from the hydrogenation of HCN, as proposed by Baumann and Möller (1991) who observed HCN was nearly always formed at lower temperatures than NH3 and prior to NH3, and only after hydrogen was released from coal. The formation of ammonia as a primary product was only observed in low-rank coals and was interpreted as the direct cleavage of amino groups and amides (Wojtowicz et al. 2001). Such low-rank coals were not typically used in gas making, apart from processes such as the Lurgi process, which was developed in Germany for use with the local lignite coals (Thomas 2020). The impact of temperature on the by-products formed was demonstrated as far back as 1945 by Rhodes. Rhodes (1945) demonstrated that the higher the temperature of thermal decomposition, the greater the yield of gas and the smaller the yield of by-products such as coke and tar, as these were broken down into the gas phase. The effect being to release a greater portion of the nitrogen present within these by-products into the gas. Such compositional changes in the coal tar were observed by Gallacher et al. (2017a, b, c) in their analysis of coal tar from different gas-making processes. The temperature and type of process used to manufacture gas are therefore critical factors in the amount of ammonia formed.

As the gas left the retort the ammonia released from the coal was removed by cooling, which occurred in the hydraulic main, foul main and condensers, and then by washing the gas with water, which occurred in the washers and scrubbers. The ammonium-rich liquor formed from the cooling and washing processes was called ‘ammoniacal liquor’. It consisted of up to 1% ammonium, and a much lower concentration of phenol, ferrocyanide and thiocyanate. The ammoniacal liquor was transferred by pump or gravity to a below-ground tank, where it was stored. These tanks also stored the tar that was also removed from the gas by the cooling and washing processes (Thomas 2020).

Stiehl and Lehmann (1980), who had identified δ15N values in European coals ranging from 3.5‰ (low-rank coal) to 6.3‰ in anthracites, identified δ15N values that ranged from 4.2 to 10.7‰ in bituminous sediments (coal tar) from the pyrolysis of coal. These would be similar to coal tar by-products produced in gasworks. The effect of charring during pyrolysis experiments also demonstrated a positive shift in isotopic enrichment by 3.9–5.6‰ (Pyle et al. 2015). These observations suggest that coal tar may be isotopically enriched with regard to δ15N values.

The tar and liquor tanks tend to be viewed as the most significant source of ammonium on gasworks sites; however, their construction could vary, as could their likelihood to leak. These tanks were often built within the redundant circular tanks of former gasholders, which were typically brick built and lined externally with a thick layer of puddle clay to waterproof them; however, leaks could occur. Leaks and spillages of ammoniacal liquor could also occur from the pipework, and the washing and scrubbing plant, as well other plant that handled the ammoniacal liquor found on some gasworks. These include concentrated ammoniacal liquor (CAL) plants and sulfate of ammonia plants (Thomas 2020).

The CAL plant dehydrated the liquor to make it more cost effective to transport to the chemical works as a concentrate. The liquor was also used for the manufacture of sulfate of ammonia fertilizer. Such sulfate of ammonia plant were found in medium to large gasworks (Thomas 2020). Both plants could be an important source of NH4+ releases. Heaton (1986) identified that ammonium sulfate produced as a by-product of a South African coke oven plant (similar to a gasworks) had a δ15N value of −0.5‰, which is lower than the values obtained for coal or coal tar.

Small amounts of ammonium salts were also deposited in the gas purifiers (iron boxes filled with either lime or bog iron ore) as the gas was purified to remove hydrogen sulfide and hydrogen cyanide. The waste from this process was called ‘foul lime’ and ‘spent oxide’ (Thomas 2020), respectively. Trace amounts of ammonium were present in the purifier waste. Isotopic analysis of the cyanide compounds in German gas-purifier waste by Weihmann et al. (2007) demonstrated them to have δ15N values between 2 and 5‰. The cyanides present in the purifier waste can be biodegraded by certain microorganisms to form ammonium (Kao et al. 2003) and nitrate (Motzer 2006).

Some of the infrastructure used to undertake the removal or storage of by-products of the gas-manufacturing process are included in Figure 1 as sources.

As the ammoniacal liquor had little value, it was often disposed of to the sewer, surface waters or run to ground in soakaways. Some individual gasworks practices also included vaporizing the liquor in pans above the furnaces. The scale of gasworks, and the volume of gas and by-products made, varied significantly; the larger gasworks benefitted from a scale of economy to process the by-products produced. The smallest gasworks would employ a single worker and supply a less than 100 customers with an annual production of less than 10 000 m3 of gas per year, whereas the largest gasworks would have over 1000 employees with an annual production of greater than 150 000 000 m3 of gas per year and over 100 000 customers. It is therefore clear that the occurrence and fate of ammonium compounds on former gasworks was highly variable.

Coal gas manufacture used a large amount of water for raising steam and in the cooling and washing of the gas, and consequently gasworks sites were commonly located adjacent to rivers, canals or in seaports, which were also used to transport feedstocks (e.g. coal) and by-products (e.g. coal tar) to and from the gasworks. They were often in locations where there were, and have since been, many other potential sources of ammoniacal nitrogen. These include natural sources such as organic material (peat, coal, animal wastes and decomposing organic matter in sediments or marshland); intrusion of marine waters; anthropogenic sources such as leachate from landfills, artificial manure/fertilizer manufacture, leaking sewers and sewerage works, cemeteries (Hart 2005), spreading of slurry/manure, and fertilizer runoff (Wakida and Lerner 2005); and natural silicate minerals, which are the main carriers of nitrogen mainly in the form of ammonium (Mysen 2019), and minerals such as the sodium-nitrate-rich caliche (Ericksen 1983). In coastal areas, the ingress of saline water into the aquifer can increase ionic strength of the groundwater, causing surface active ions to be released to solution. This can increase the dissolved concentrations of desorbable ions such as ammonium and phosphate (Moore and Joye 2021).

To determine whether groundwater quality standards for ammonium will be attainable at a given gasworks site, and to ascertain how best to attain those standards, it is important to understand the proportion of ammoniacal nitrogen present in groundwater that is attributable to gasworks sources, as opposed to the proportion that is attributable to other anthropogenic and natural sources (i.e. background). Ascertaining the amount of anthropogenic ammoniacal nitrogen from specific sources is clearly also valuable in attributing liabilities associated with groundwater clean-up.

This study investigated the potential of using nitrogen isotopes to differentiate between sources of ammoniacal nitrogen in groundwater at former gasworks sites. This includes both those sources originating from onsite activities at the former gasworks and sources from the surrounding area. Although nitrogen isotopes have been used to evaluate the sources of nitrate to groundwater and large datasets have been collected on the isotopic composition of NO3 in groundwater (Lingle et al. 2017; Nikolenko et al. 2018; Zhang et al. 2020; Niu et al. 2021), less research has been undertaken with respect to NH4+ (Nikolenko et al. 2018); this study evaluates the use of nitrogen isotopes of nitrate and ammonium to evaluate sources to groundwater on a number of former gasworks sites of differing size and setting.

Previous studies using isotopic analysis as a tool in environmental forensics

Stable isotopes of nitrogen have been used to investigate sources of nitrate to groundwater since the 1970s but were initially controversial (Kohl et al. 1971; see Heaton 1986 for a review). A large body of forensic isotope work now exists relating to nitrates in groundwater but fewer have focused on ammoniacal nitrogen in groundwater, primarily due to the prevalence of nitrate as a threat to groundwater quality in the USA, where much of this research has been conducted. Groundwater nitrate δ15N enrichment has been used to identify human and animal sewage wastes, which are typically enriched in δ15N (Kreitler and Jones 1975; Mariotti et al. 1988; Jones et al. 2018), and to trace nitrogen inputs from inorganic fertilizers (Silva et al. 2002; Fuertes-Mendizábal et al. 2018).

Attribution of ammonia or ammonium to specific sources is generally attempted using multi-proxy approaches. This includes methods such as major ion chemistry to map groundwater mixing; tracer materials such as boron, caffeine or coliform bacteria to trace sewage; and chloride, strontium or bromide to trace seawater intrusion (Pastén-Zapata et al. 2014; Zhang et al. 2019). Robinson et al. (2006) was the first such multi-proxy study that incorporated use of nitrogen isotopes to evaluate sources of ammonium to groundwater. The gasworks-sourced groundwater displayed a δ15N fractionation from total nitrogen extraction of less than 5‰ and was successfully differentiated from groundwater impacted by ammonium release as a result of seawater intrusion.

Nitrogen-based compounds in groundwater exhibit δ15N enrichment or depletion based on groundwater inputs, modified by microbiological and physical processes occurring within the aquifer (Rivett et al. 2008). The processes governing the nitrogen cycle are described in detail by Rivett et al. (2008). In the context of the groundwater in a gasworks, important processes may include: (a) nitrification, where ammonium is converted to nitrate via the intermediate nitrite; (b) dissimilatory nitrate reduction to ammonium; (c) denitrification, where nitrate is converted through a series of intermediates to nitrogen gas; (d) anaerobic ammonium oxidation; and (e) uptake by microorganisms (Böhlke et al. 2006; Rivett et al. 2008).

Controls on δ15N in groundwater ammonium are summarized by Böhlke et al. (2006). Most notably, Böhlke et al. (2006) identified that nitrification of NH4+ yields 15N-depleted products (e.g. nitrate) and commonly results in a substantial increase in the δ15N value of the residualNH4+. Denitrification leads to the enrichment of the δ15N isotopes due to the fact that δ14N reacts faster than δ15N; a similar observation was noted for δ16O over δ18O (Böttcher et al. 1990). Clark et al. (2008) studied the fate of δ15N in the form of ammonium and nitrate to identify anaerobic ammonium oxidation (anammox) as a mechanism for strong attenuation of ammonium in groundwater from multiple industrial sources.

Whilst nitrate δ15N and ammonium δ15N values may not be directly comparable, they can be used together to understand nitrogen cycle processes such as nitrification, and can be combined to calculate total inorganic δ15N for comparison with total organic δ15N (Cravotta 2002).

Potential complications in the use of nitrogen isotopes in the attribution of ammonium to specific sources can include (but are not limited to):

  • Seasonal variations in assimilation by plants and microbiological processes, causing fluctuations in isotope enrichment or depletion (Ostrom et al. 1998);

  • mixing of nitrogen from multiple soil sources, and the impact of soil type on microbial processes and, therefore, on the δ15N signatures (Gormly and Spalding 1979);

  • predominance of soil nitrogen diluting potential signatures from anthropogenic nitrogen sources such as fertilizers; fractionation during sample preparation (Schmidt and Gleixner 2005); and

  • potential effects of different ammonia-oxidizing and ammonia-generating bacteria species on δ15N signatures (Casciotti et al. 2003).

Due to these complications, multi-proxy studies are generally considered more robust in identifying and discriminating between nitrogen sources to groundwater than those that rely on δ15N alone. For example, Curt et al. (2004) demonstrated difficulties in using only δ15N to differentiate between human and animal wastes. Moore et al. (2006) showed that geochemical data can be combined with isotopic tracers to distinguish between different kinds of agricultural and residential activity. Manning and Hutcheon (2004) provided an overview of the use of geochemical indicators, including salinity, potassium and sodium, to evaluate ammonium with a mineralogical provenance. Work has also used oxygen isotopes (δ18O) together with δ15N to evaluate nitrogen sources (Silva et al. 2002; Moore et al. 2006; Robinson et al. 2006; Spalding et al. 2019).

Site selection

A total of 200 former gasworks sites, owned by National Grid Property Holdings at the time of the project, were screened in a desktop review to identify suitable sites for further work. Sites were initially screened based on potential alternative sources of ammonium. They were then shortlisted based on several criteria, including whether: (i) previous studies allowed an understanding of the hydrogeological system and setting at the site; (ii) there were historical analyses of ammoniacal nitrogen, ammonia, nitrate, nitrite and Kjeldahl nitrogen in groundwater available; (iii) there were suitable groundwater monitoring wells on-site; and (iv) there were possible alternative sources of ammonium in the groundwater. Fourteen prospective sites were then prioritized and inspected to verify that the monitoring infrastructure was intact and suitable for collection of representative groundwater samples.

Table 1 summarizes key information including the site history, hydrogeology and monitoring network for the final sites selected for the study. This includes a site in SW England that has been studied for a longer period due to the installation of a groundwater remediation system.

Site conceptual models for multiple sources of NH4+

A generalized site conceptual model for the study is presented in Figure 1. Although all sites differ in the size and nature of the gasworks operation, groundwater configuration and proximity to other potential sources of ammoniacal nitrogen, the diagram seeks to illustrate some of the complexity in interpreting hydrochemical and isotopic data from industrial or rural areas. As demonstrated, a single gasworks may have multiple sources onsite and a mixture of anthropogenic and natural sources in the surrounding area.

Groundwater sampling

Groundwater samples were collected from groundwater monitoring wells at eight sites, referenced as A–H. Monitoring wells were initially gauged and the depth to water recorded. The wells were checked to ensure that no non-aqueous phase liquid (NAPL) was present, which would prevent isotope extraction from the samples in the laboratory. Monitoring wells were purged with a minimum of three well volumes of water prior to sampling using new disposable Waterra tubing and valves in each well.

Samples were field filtered and collected directly into laboratory-supplied sample bottles that contained sulfuric acid to preserve the samples at pH < 2. Samples were placed in cool boxes packed with ice pack to ensure that the samples were kept at 4°C or below. They were couriered to the laboratory at the end of each day.

Samples were submitted to the Environmental Laboratory Ltd in East Sussex, UK, for chemical analysis that included total alkalinity, Kjeldahl nitrogen, nitrite, nitrate, ammoniacal nitrogen and major ions (sodium, potassium, calcium, magnesium, chloride and sulfate).

Isotopic analysis

Isotopic ratios of N and O in ammonium, nitrate and/or total nitrogen were analysed by IsoAnalytical Ltd, a commercial laboratory based in Crewe, UK. The following analytical methods were used.

Ammonium was extracted from 1 l of each water sample (filtered) using the method described by Lehmann et al. (2001). In brief, the dissolved ammonium was stripped from the water by a cation exchange resin. The resin was washed with distilled water and then dried overnight in an oven. The dried resin was then directly analysed for nitrogen-15 as described below. For the samples from Site C, a control sample of ammonium sulfate (IA-R045) (5 mg of ammonium N l−1) was extracted alongside the samples.

Nitrate for nitrogen and oxygen isotopic analysis was extracted from 1 l of the water sample using the method outlined by Silva et al. (2000). In brief, dissolved nitrate was stripped from the water sample (pre-filtered) onto anion exchange resin. The nitrate was washed from the resin using hydrochloric acid and then converted to silver nitrate by adding silver oxide. After filtering the silver nitrate solution, barium chloride was added to precipitate dissolved sulfate. After further filtering, the silver nitrate solution was purified by passing it through a cation exchange resin. The silver nitrate solution was then neutralized by adding further silver oxide before filtering it for a third time. Dissolved organic carbon was then removed by adding activated carbon before filtering for a fourth and final time. The resulting silver nitrate solution was then freeze-dried and stored in amber vials prior to analysis.

Total-nitrogen samples were prepared by boiling down 80 ml of sample filtered through a 0.22 μm membrane. Nitrogen isotopic ratios were analysed using EA-IRMS (elemental analyser isotope ratio mass spectrometry). In brief, tin capsules containing appropriate amounts of samples or reference materials were dropped in sequence into a furnace held at 1000°C, where they were combusted in an oxygen-rich environment. The gases produced on combustion were swept in a helium stream over a combustion catalyst (Cr2O3), copper oxide wires (to oxidize hydrocarbons), and silver wool to remove sulfur and halides. The resultant gases – N2, NOx, H2O, O2 and CO2 – were swept through a reduction stage of pure copper wires held at 600°C. This step removed O2 and converted NOx species to N2. A magnesium perchlorate chemical trap was used to remove H2O. For nitrogen-15 analysis, CO2 was removed using a chemical trap (Carbosorb) and the nitrogen focused on a packed column gas chromatograph held at an isothermal temperature of 100°C. The resultant chromatographic peak of N2 entered the ion source of a Europa Scientific 20-20 IRMS where nitrogen species were separated and simultaneously measured using a Faraday cup collector array at masses 28, 29 and 30.

Oxygen isotope analysis was conducted by total conversion at 1080°C in a quartz reactor tube lined with a glassy carbon film, filled to a height of 170 mm with glassy carbon chips and topped with a layer of 50% nickelized carbon (10 mm deep). Carbon monoxide and nitrogen were separated on a gas chromatography column packed with molecular sieve 5A. The IRMS used was a Europa Scientific Geo 20-20 with a triple Faraday cup collector array to monitor the masses 28, 29 and 30.

The reference material used during the nitrogen analysis was IA-R045 ammonium sulfate with a δ15N value of −4.71 ‰ v. air; and for oxygen analysis, IAEA-NO-3 (potassium nitrate, δ18O = 25.6‰ v. V-SMOW). Reference standards IA-R045, IA-R046 (ammonium sulfate, δ15N = 22.04‰ v. air) and IAEA-NO-3 (potassium nitrate, δ15N = 4.7‰ v. air), were measured for quality control during analysis of the samples. IA-R045 and IA-R046 are traceable to IAEA-N-1 (ammonium sulfate, δ15N = 0.4‰ v. air). The International Atomic Energy Agency, Vienna, distributes IAEA-N-1 and IAEA-NO-3 as isotope reference standards.

Hydrochemistry

Major ion data for all samples are plotted on a Piper diagram in Figure 2. Major ions display a wide range in compositional types reflecting the hydrogeological setting of each site and the gasworks processes undertaken on those areas of the site sampled. Samples collected from sites A, B and C were characterized by or include Ca–HCO3-type waters, consistent with interactions with carbonate rocks known to underlie the latter two sites. At the time of the sampling, the land directly upgradient of Site A had been subject to the installation of cement piles. As this site had formerly been part of the gasworks that contained the purifiers and CAL plant, the alkaline cement used may have influenced the groundwater chemistry.

Samples obtained from sites G and H were brackish to saline and were dominated by Na–Cl ions, consistent with the site's location on former marshland and its proximity to the coast. Those samples that were most affected were those obtained from the area of the site adjacent to the tidally influenced harbour or river. Samples from sites, D, E and F displayed mixed-type signatures (Mg–Ca–SO4 to Na–Ca–Cl–HCO3). These sites are underlain by alluvial material and argillaceous to arenaceous bedrock. The first two sites were located within coastal areas, near to the sea, which would have influenced the groundwater composition.

In addition to Kjeldahl nitrogen, ammoniacal nitrogen, nitrite, nitrate and major ions, other compounds that are indicators of anthropogenic contamination were also considered. These included substances such as thiocyanate, cyanide and hydrocarbon compounds that are commonly associated with contamination from gasworks sites. Based on detections and concentrations in historical monitoring data (Table 2), monitoring wells were characterized as:

  • Potentially impacted by gasworks-derived compounds;

  • not impacted; or

  • potentially impacted by contaminant plumes from more than one source (mixed).

These classifications were then used to assist in interpreting the isotopic data. In addition to this classification, comments have also been provided on the sample locations and field observation, where useful.

Isotopic data

Isotopic data for all samples are summarized in frequency histograms for δ15N(total), δ15N(NH4), δ15N(NO3) and δ18O(NO3) in Figure 3. Samples have been classified as being potentially impacted by gasworks-related ammoniacal nitrogen, not impacted or potentially impacted by mixed sources of ammoniacal nitrogen, including off-site sources. The most striking feature of the isotopic data is the large isotopic range for all systems: the total ranges are 30.4‰ for δ15N(total), 22.7‰ for δ15N(NH4), 28.0‰ for δ15N(NO3) and 11.7‰ for δ18O(NO3). These ranges are wide compared with published data from a range of natural and anthropogenic sources of nitrogen, as demonstrated in Figure 4 (Zhang et al. 2019; Niu et al. 2021). The wide range could be accounted for by the mixture of gasworks and non-gasworks sources detected, and the differences in the sites studied. It is noted that nitrogen isotopic ratios can vary and correlate with other parameters such as salinity, aquifer conditions and monitoring bore location on a site scale, although not necessarily in a systematic way between sites.

Within a gasworks, the origin of the anthropogenic nitrogen would be sourced from the coal used in the gas-making process, as identified in the introduction to this paper. The range of δ15N isotopes in coal were found to be between −3.2 and +6.3‰, although for European coals this range was narrower (i.e. between 1.86 and 6.3‰). This isotopic range would be expected to be reflected within the by-products produced by the gas-making processes, unless these processes caused isotopic fractionation. The observations made by Stiehl and Lehmann (1980) and Pyle et al. (2015) suggest that the action of thermal decomposition by pyrolysis may cause a positive shift in isotopic enrichment for tar residues. This may not be the same for all by-products, as Heaton (1986) provided a value of −0.5‰ for ammonium sulfate fertilizer produced from coke ovens (a similar process to that used in gasworks). The work of Weihmann et al. (2007) demonstrated another by-product of gas making, gas-purifier waste that is rich in cyanide, to have a δ15N value of between 2 and 5‰. The ranges described above are plotted in Figure 4.

The results from Site A, a large former gasworks that has been studied in most detail, provide useful insights. Groundwater sample from the monitoring boreholes in the north of the site (samples 1-2, 2-2, 1-1, 2-1, 1-3, 2-3, 2-13 and 2-15) gave δ15N(NH4) and δ15N(NO3) values that range between 8.28 to 25.46‰. The north of the site had limited sources of gasworks-derived ammonium, it was formerly marshland and was hydraulically linked via a water intake to a tidally influenced river known to have received sewage; there were also potential leaking sewers in the roads upgradient of the site. The results obtained were similar to those observed for animal excreta or sewage (i.e. >10.00‰ δ15N: Motzer 2006; Niu et al. 2021).

Those groundwater samples taken from the area of the coal-tar NAPL plume within shallow alluvial valley (samples 1-9, 2-9, 1-11, 2-11, 2-14, 1-7 and 2-7) had δ15N(NH4) values that ranged from 6.54 to 10.73‰ (sample 1-7 also recorded 11.12‰ δ15N(NO3)). These samples were typically more enriched than those δ15N values obtained for European Coal but would be in the region expected for pyrolysis tars, which are more enriched than the parent coals. Biodegradation, which had previously been observed on this site (Gibert et al. 2007), can also enrich δ15N isotopes through the preferential use of δ14N in the groundwater.

The groundwater samples collected from the western area of the site (samples 1-4, 2-4, 1-6, 2-6, 1-8 and 2-16) had δ15N(NH4) values that are relatively depleted, ranging from 1.54 to 3.86‰ (sample 1-6 also gave a δ15N(NO3)value of 4.88‰). The land to the west of Site A had previously been part of the gasworks and was the location of the major sources of ammonium: the concentrated ammoniacal liquor plant and, most notably, the former gas-purification plant. The results obtained were very similar to the δ15N result obtained from gas-purifier waste by Weihmann et al. (2007) and for European coal. Due to groundwater abstraction onsite for contaminant recovery, all these potential sources were up the hydraulic gradient of the site. A distinct trend is noted: that those samples most impacted by gasworks by-products predominantly contain only δ15N(NH4) isotopes, and at concentrations between 1.54 and 10.73‰.

Site B was a former small town gasworks in an agricultural area, all samples at Site B showed evidence of gasworks by-products. Samples 1 and 4 were taken from two boreholes located on former allotments, and between 50 and 80 m down the hydraulic gradient from the location of the former purifiers and retort house. Samples 1 and 4 recorded δ15N values of −6.47 and −6.91‰ for δ15N(total), 9.78 and 5.33‰ for δ15N(NO3), and −4.26 and −7.07‰ for δ15N(NH4), respectively. Samples 2 and 4 were obtained from boreholes 90–110 m up the hydraulic gradient from samples 1 and 4, and before the former purifiers and retort house. Samples 2 and 4recorded higher δ15N values: 13.1 and 14.3‰ for δ15N(total), 20.8 and 21.16‰ for δ15N(NO3), and 10.76 and 11.64‰ for δ15N(NH4), respectively. This site is complex to interpret as another earlier gasworks had been built 50 m to the north of this site as well an electric light works that, along with the adjacent allotments and agricultural land, may have influenced the results.

Site C was a small town gasworks in an agricultural area. The sampling locations available for Site C were all within a similar area of the site. This was about 60–80 m down hydraulic gradient of the former retort house, oil tanks and purifiers in an area previously occupied by offices. The locations were 25 m upgradient of a river. Sample 3 (δ15N(total) of 21.6‰) had evidence of gasworks by-products and was obtained from a borehole located adjacent to a below-ground tank of unknown use; this was the only sample with detectable δ15N(NH4) (2.01‰), whose origin is likely to be associated with the contents of the adjacent tank. Samples C1 and C2 (14.6 and 7.2‰ δ15N(total), respectively) were also likely to have been influenced by sources from the upgradient allotments, playing fields and cemetery, giving a mixed signature.

Sample D was a large city gasworks built on reclaimed marshland. The samples from Site D were taken from within (sample 1) and outside (sample 2) a former below-ground gasholder tank. In an area of the gasworks up the hydraulic gradient from the location of sample 2 were the sites of a former condenser (50 m) and a former purifier house (100 m). Samples 1 and 2 had similar depleted δ15N(total) values (3.6 and 3.7‰, respectively), and sample 2 had a similar depleted δ15N(NH4)value (2.21‰). The perched groundwater sample (sample 2) from within the tank is likely to have been influenced by the diffusion of ammoniacal compounds from the tarry sludges typically found at the base of the gasholder tank. Sample D1 was likely to have been influenced by saline intrusion and organic material in the reclaimed marshland, as well as the mentioned gasworks sources.

Site E was a small town gasworks in a coastal resort. Three samples were analysed for nitrogen isotopes from Site E and the results provided useful insights. Sample 2 was taken from a borehole adjacent to both the original and later retort houses and near to the purifier house. This sample was rich in both δ15N(NH4) (15.64‰) and δ15N(NO3) (10.36‰). Sample 3 was obtained from a borehole located 35 m downgradient from sample 2; it was more visibly impacted and richer in ammonium but both δ15N(NH4) (−0.92) and δ15N(NO3) (8.05‰) were significantly depleted. Sample 1 was obtained from a borehole on the periphery of the site near to wetlands and a stream; only δ15N(NO3) (8.4‰) was detected. The borehole had no evidence of gasworks by-products and contained very little ammonium, and was likely to have been influenced by adjacent agricultural land or possible saline intrusions.

Site F was a former small town gasworks, built alongside a former colliery. Sample 4, which recorded the lowest δ15N(total) (+5‰), was located in the area of the site adjacent to the former purifier house. Samples 2 and 3 were obtained from boreholes within below-ground tank structures, the integrity of which was uncertain as the groundwater data did not suggest that the groundwater was perched. Sample 2 also contained 8.91‰ δ15N(NH4), the likely source being coal tar or tarry sludge from tank. Sample 1 had the most enriched value of δ15N(total) (23.5‰) and was at the northern boundary of the site adjacent to the former below-ground tanks; the sample also detected δ15N(NH4) (8.92‰). The result suggests that the sample was influenced by gasworks sources, as well as nitrogen sources from the adjacent colliery and coal measures, saline intrusions, and agricultural uses.

Site G was a former medium-sized town gasworks built on reclaimed estuarine mudflats within a harbour. Only δ15N(NO3) was detected within the samples recovered. Samples 2 (16.54‰), 3 (18.74‰) and 5 (14.08‰) were obtained from the original area of the gasworks site. Samples 1 (10.03‰) and 4 (9.08‰) were obtained from the area of the gasworks that had been reclaimed from the harbour. The groundwater regime at the harbour site was tidally influenced, with the impact of the main gasworks processes unclear as these were located off-site.

Site H was a very large city gasworks, located on the bank of an estuary. At Site H, nine samples were analysed for δ15N(total). The three samples with the highest concentrations of ammoniacal nitrogen (sample 1 (62.9 mg l−1), sample 3 (84 mg/l) and sample 8 (61.5 mg l−1)) had the most depleted δ15N(total) values (sample 1 (−0.045‰), sample 3 (−0.71‰) and sample 8 (−5.54‰)) and were three of the least saline. These depleted samples are inferred to have been impacted by gasworks-derived ammoniacal nitrogen, supported by their location being adjacent to the gasworks plant which would have been a major sources of nitrogen-containing gasworks by-products, including the tar and liquor tanks (samples 1 and 3), washers (sample 3), scrubbers (sample 8), and purifiers (samples 3 and 8). Sample 3 also had a high concentration of sulfate, which is indicative of gas-purifier waste. Sample 2 was obtained from a location adjacent the former producer gas and coke plant, and had a δ15N(total) value of 4.85‰, which is similar to that observed for coal. These gasworks were on the bank of a major estuary, with the sample locations (4, 5, 6 and 7) closest to the river coinciding with the highest salinity and the most enriched δ15N(total) values (8.2–20.19‰). Sample 7 was obtained from a location within a former coal-stocking area, which was adjacent to a large sewer and wastewater treatment plant in a former coal storage area that had a δ15N(total) of 4.72‰, which is similar to that observed for coal.

The aggregate data show an occurrence of δ15N(NH4) within those samples directly attributable to the presence of gasworks by-products or coal at sites A, C, D, E and F. This is not evident in Site G and less clear in Site B, where all samples contain δ15N(NH4), but the samples most impacted by the gasworks by-products are significantly more depleted than most of the other sites observed.

The aggregate data also appear to demonstrate bimodality in relation to δ15N(NO3), δ15N(NH4) and, to a lesser extent, δ18O(NO3). This bimodality is seen at the site scale at least at the two sites B and G, and possibly also at sites A and F, when the factors affecting the sampling locations are considered further. However, there is no consistent relationship between upgradient and downgradient or impacted and not impacted samples. In the case of Site G, the upgradient samples are enriched, and the depleted samples occur in the made ground in land reclaimed from the harbour, with the whole site having been greatly influenced by tides. These data suggest the influence of complex processes within a site but would require additional work to unravel the sources and pathways in detail.

Another feature of the aggregated data is a systematic enrichment in δ15N(NO3) relative to δ15N(NH4) of about 8–12‰, and a positive correlation between δ15N(NO3) and δ15N(NH4) in samples where both were analysed (Fig. 5). This enrichment is not consistent with oxidation of the ammonium to nitrate (nitrification), which would be expected to lead to an overall enrichment of δ15N in ammonium. Instead, the systematic enrichment in nitrate δ15N may indicate either distinct sources or enrichment of δ15N(NO3) due to denitrification. If the enrichment in δ15N(NO3) is related to denitrification, then we might also expect enrichment in δ18O(NO3) (Böttcher et al. 1990). In contrast, a bivariate plot of δ15N(NO3) v. δ18O(NO3) shows an apparent negative correlation between δ15N(NO3) and δ18O(NO3) in data from the three sites G, B and E. This would seem to rule out denitrification and, instead, possibly indicate multiple sources or pathways.

Initial work carried out by Robinson et al. (2006) indicated that gasworks-derived ammoniacal nitrogen might have more distinctive δ15N values than samples shown from other lines of evidence to be impacted by other (non-gasworks) contaminant sources. The review of the literature suggests that coals and coal by-products can each have their own potential isotope signatures, where δ15N isotopic signatures for coal and coal pyrolysis by-products may range between −3.2 and +10.7‰ (−3.2 to +6.3‰ for coal δ15N, 4.2–10.7‰ coal tar, 2–5‰ gas-purifier waste δ15N and −0.5‰ ammonium sulfate ).

Former gasworks vary significantly from one another in scale and the technology used. With gas-making processes varying in temperature employed from c. 600°C (small hand-charged horizontal retorts) to c. 1200°C (high-temperature horizontal retorts, and coke oven and chamber oven processes). The temperature and design of the plant affected the amount of ammonia formed within the gas and how much nitrogen remained within compounds in the coke or coal tar. That which entered the gas phase was later deposited in either ammoniacal liquor (ammonium) or gas-purifier waste (cyanide). The greater the temperature used the lower the proportion of nitrogen retained in the coke or coal tar.

Therefore, there are potentially multiple different isotope signatures to be expected on a gasworks, some of which overlap and require a thorough understanding of the site's historical layout when interpreting the data. The process of biodegradation potentially further enriches the isotopic values for δ15N through the preferential reaction of the lighter δ14N isotopes. In addition, this study has identified that there are many cases where there are multiple additional sources of non-gasworks ammonium. These isotopic signatures may be broad and overlap ranges with gasworks- or non-gasworks-derived sources. On this basis, gasworks should not always be considered as the only or primary source of ammonium in groundwater as there may be multiple sources and complex transformations and pathways occurring on-site or locally.

There are indications, particularly in the data from Site A, that ammonium derived from gasworks activities can have characteristic ranges in δ15N, similar to the parent coal, coal tar and purifier waste, and is predominantly found in the form of δ15N(NH4). Similarly for Site H, groundwater samples obtained from locations adjacent to former gas-processing or by-product storage infrastructure had δ15N isotopic signatures similar to those expected for coal or coal by-products, and more depleted than those samples where ammonium would have been sourced primarily from non-gasworks sources. The data from Site A also suggest that gas-purification waste may have a distinct δ15N isotope range (1.54–4.88‰) compared to coal tar (6.31–10.73‰). Sites A and H provide clearer data with more distinct sources visible due to the large size of the former gasworks studied.

Similar observations were also noted for sites C, D, E and F, with the trends less apparent at Site G and more extreme in Site B, where the depletion of the downgradient δ15N for δ15N(total) and δ15N(NH4) was more significant compared to the upgradient boreholes. Whilst the aggregated data (Table 2) from all the former gasworks sites show, in general, a wide range (+30.41‰ δ15N(total)) in δ15N compared with published ranges specifically for coal and coal by-products (−3.2 and +10.7‰, respectively), there is evidence for a range of nitrogen isotope values that are characteristic of gasworks or other contamination across all sites. The sampling undertaken was restricted to existing boreholes within the site owner's property and so with additional sampling this distinction is likely to be more defined.

The data suggest that, even within each site, evidence of gasworks processes is discernible, although the nitrogen cycling pathways can be complex. It is likely that the observed isotopic variation is potentially due to multiple process sources derived from complex and time-variable supply chains at the former gasworks potentially mixing with off-site or pre-existing ammonium contamination, and also complex microbially mediated transformations that can modify and broaden the initial δ15N ranges. Therefore, use of these data coupled with traditional hydrochemistry (Piper diagrams and bivariant plots) and a detailed knowledge of site operations, rather than just using isotopic data alone, has led to an enhanced understanding of the source of ammonium in the groundwater,.

This study illustrates the usefulness of combined chemical and isotopic analysis to identify multiple sources of ammonium in groundwater but also highlights the need for more detailed and targeted sampling and analysis at the site scale to identify and isolate these sources and their relative contributions to the nitrogen load. As an initial pilot study, the results indicate that chemical characterization is possible at former gasworks to show trends and differences in ammoniacal nitrogen sources. Further detailed isotopic analysis at sites A, B, E or G would be likely to yield useful insights into the origins of ammoniacal nitrogen in groundwater at former industrial sites.

Future research in this area would benefit from the following:

  • Sample and characterize gasworks by-products (coal tar and ammoniacal liquor), waste (foul lime and spent oxide) and feedstock (coal) to assess if they have specific isotopic signatures. This may require the development of specialized analytical techniques for heavily contaminated samples;

  • further sampling and characterization of potential non-gasworks sources; and

  • the investigation of further large gasworks sites with more extensive monitoring wells.

The authors would like to acknowledge National Grid Property Holdings for access to numerous sites for inspection and sampling. The support of WSP UK Ltd is also gratefully acknowledged.

RAPT: funding acquisition (equal), investigation (equal), writing – original draft (equal), writing – review & editing (lead), supervision (supporting), Investigation (supporting), visualization (equal); MJR: writing – original draft (equal), writing – review & editing (supporting); JDFR: writing – original draft (equal), conceptualization (equal), data curation (equal), formal analysis (lead), investigation (lead), methodology (equal), supervision (lead), visualization (equal); SJAB: conceptualization (equal), data curation (equal), methodology (equal), validation (lead), formal analysis (lead), visualization (equal), writing – original draft (equal), writing – review & editing (supporting); CT: funding acquisition (equal), writing – review & editing (supporting).

This work was funded by National Grid Property Holdings.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

The datasets generated during the current study beyond those provided in the tables and figures are not publicly available due to client confidentiality.

Scientific editing by Jonathan Smith; Christopher Swainston