Detection of fluorescent compounds in the environment ... - Springer Link

Original article

Detection of fluorescent compounds in the environment using granular activated charcoal detectors Chris Smart Æ Brad Simpson

Abstract Granular activated charcoal (GAC) acts as a general adsorptive medium for organic molecules, and is widely used to capture fluorescent tracer dyes. To investigate the capability of GAC in contaminant screening and tracer adsorption, a number of detectors were deployed for 2–6 days in a range of urban surface waters, and the adsorbed compounds eluted in an alkaline alcohol solution. Simultaneous water samples showed a diverse range of fluorescence environments, ranging from relatively clean, steady groundwater discharge, to highly concentrated and variable treated municipal sewage. A wide variety of organic compounds and dyes were found in the waters, as chronic and acute contaminants. The relationship between charcoal and water spectra depended on exposure time and loading. Short exposure times emphasized short wavelengths, longer exposure times emphasized longer wavelengths at the expense of shorter wavelengths. The magnitude of the effect depended on loading, being greater in enriched waters. In general, charcoal eluent shows a significant gain in fluorescence intensity over water. However, there may be an apparent loss at shorter wavelengths for samples with long exposure times and high loading. A similar bias was also discovered with the elution time of activated charcoal. Short wavelength fluorescence intensity peaked after a few minutes of elution; longer wavelength fluorescence increased over many days. These results show that charcoal is a reasonably effective material for adsorption of longer wavelength compounds. However, the ubiquity of many fluorescent dyes in the environment, and the complex relationship between

Received: 17 September 2001 / Accepted: 16 October 2001 Published online: 17 April 2002 ª Springer-Verlag 2002 C. Smart (&) Æ B. Simpson Department of Geography, University of Western Ontario, London, Ontario, N6A 5C2, Canada E-mail: [email protected] Tel.: +1-519-6192111 (ext. 85007) Fax: +1-519-6613150

538

Environmental Geology (2002) 42:538–545

the waters and the eluent spectrum, suggest that considerable care is required in the interpretation of eluent spectra. Keywords Activated charcoal Æ Environmental contaminants Æ Fluorescence spectroscopy Æ Natural organic matter Æ Tracer dyes

Introduction The pathways followed by underground waters can be best identified by using fluorescent tracer dyes to label water and make it identifiable further downstream. Resolution of underground drainage trajectory is important, not only in defining the catchment areas of karst springs, but also in resolving the source and destination of contaminants. Good dye tracing practice excludes the use of ‘‘excessive coloration’’, as this is offensive and possibly harmful. Instead, remarkably small quantities of dye can be used, and detected using sensitive fluorometers. Unfortunately, it is seldom possible to define exactly when and where a tracer will appear. So rather than install fluorometers or water samplers at every possible site, discrete packages of granular activated charcoal are used to scavenge any dye that should pass by them. Unfortunately, charcoal detectors also pick up contaminants and dyes regardless of their origin, and these may compromise the interpretation of a dye trace, either by obscuring the tracer dye or creating a false impression of its presence. Both eventualities might be significant in misdirecting litigation and management decisions. Granular activated charcoal (GAC) is widely used as a semi-quantitative sampling medium in fluorometric dye tracing, generally deployed in streams and springs in discrete packages known as ‘‘bugs’’ (e.g., Alexander and Quinlan 1992). The advantage of GAC bugs lies in their efficiency; they are considered to scavenge dye from water to give amplification of the ambient signal (10 times, e.g., Matisova and Skrabakova 1995); they are integrative, trapping transient dye breakthrough which discrete water samples might fail to capture; they are relatively immune to disturbance and vandalism, and they are inexpensive, most of the cost being associated with their distribution, collection, and analysis.

DOI 10.1007/s00254-001-0517-4

Original article

GAC bugs are generally eluted in an alkaline alcohol mixture, which is subjected to spectrofluorometric analysis (e.g., Alexander and Quinlan 1992). The resulting spectra are amenable to quantitative analysis to permit objective identification of tracers with a known spectral signature (e.g., Tucker and Crawford 1999). However, it is generally accepted that differences in the adsorption efficiency of various dyes, uncontrollable flow in the field, and imprecision in most elution procedures make the absolute concentration of tracer less reliably ascertained. Nevertheless, relative concentration is indicated by spectral peak height, and this is commonly used as an indicator of quality assurance. Granular activated charcoal is a highly adsorptive medium, prepared by roasting of coarse, high-grade (cocoanut) charcoal. Fresh GAC contains a wide range of adsorption sites that vary in their strength and accessibility. Older charcoal tends to be less adsorptive because of denaturing of the more energetic adsorption sites, and presumably capture of organic molecules from the atmosphere. When deployed in the environment, GAC captures a broad range of organic molecules, including fluorescent dyes. A complex hierarchy of adsorption occurs based on the range of adsorptive sites, their accessibility, and the loading (composition and duration of the flow). Similarly, the desorption process targets more weakly bound, accessible species, with somewhat different results with different eluents (e.g., Cheremisnoff and Cheremisnoff 1993). These remarks might indicate a comprehensive grasp of the adsorption–desorption process in fluorometric dye tracing. However, these processes are of such complexity in the environment that most practitioners have adopted a conventional ‘‘best practice’’, which appears to work well and is generally accepted as reliable. While accepted, such practice is not well-founded on either theoretical understanding, or controlled experiments.

Problem definition The characteristics of GAC noted above suggest that it may be used in combination with the fluorescence spectrum as an integrative, low cost screening agent for organic contaminants. Vesper and others (1999) employed a proprietary adsorption system (Gore-sorber), which they subjected to GC-MS analysis for organic contaminants in karst springs. Many aromatic organic contaminants are fluorescent molecules (e.g., Pharr and others 1992), which are adsorbed by GAC. Although fluorescence spectroscopy is analytically indiscriminate, it provides a very inexpensive, broadly based screening tool, suitable for efficient targeting of more costly discriminatory techniques such as HPLC and GC-MS. The supposed ability of detectors to act as integrative sentries is particularly appealing in providing scope for capture of acute (i.e., transient) contamination episodes that might be missed by itinerant sampling. In general, we can report that fluorescence spectroscopy and charcoal are very effective in their respective screening and sentry roles. However, we were concerned with the

apparent inconsistency of the results obtained with GAC detectors. The organic molecules contributing to the fluorescence signature are those that contribute to fluorescence background, so our work is also relevant to tracing with fluorescent dyes in urban environments. Once the basic fluorescent characteristics of site waters were assessed from water samples, three questions were addressed. First, what was the relationship between the spectra of waters and associated GAC-eluent spectra. Is the accepted ‘‘gain’’ of 10· correct? The second question addressed the influence of exposure time and contaminant concentration (‘‘loading’’) on the resulting spectra. The third question concerned the influence of elution time on the resulting spectrum. This involved slightly more controlled elution of a standard charcoal sample.

Methods The city of London, Ontario, Canada, occupies the confluence of the north and south forks of the Thames River, which drains 1,800 km2 and has a mean annual discharge of 37.6 m3 s–1. The catchment is composed of glacial, fluvioglacial and glaciolacustrine sediments and drains a mid-west corn-belt-style catchment. London (population 300,000) is fairly typical of rejuvenated ‘‘rust-belt’’ cities, and the river is flanked by a number of former industrial sites, capped landfill sites, and a major sewage-treatment plant (‘‘Greenway Pollution Control Plant’’ sic). Groundwater emerges from sand and gravel aquifers adjoining the river. Coal tar is known to be leaking into the river bed and tributary creeks are contaminated with PCBs. Five grams of fresh granular activated charcoal were placed in nylon screen packages, which were deployed just below water surface at a range of sites for between 2 and 6 days. Detectors were placed at the outlets of storm drains, former industrial sites, and miscellaneous seeps along the south branch of the Thames River and downstream of its confluence with the North Branch. A number of detectors were placed in the river itself. Subsurface water samples were collected during deployment and retrieval of detectors. Detectors were rinsed in deionized water and about 2.5 g charcoal eluted using 15 ml ‘‘Smart’’ solution (1-propanol, deionized water and 30% NH4OH in the ratio 5:3:2; Smart 1972; Alexander and Quinlan 1992). Elution time ranged from 40 min to 5 h within batches. In one case, a series of analyses were made drawing from a bulk preparation of eluent from 1 min to several days. Waters and eluents were decanted into quartz (UV-transparent) 3-ml fluorometric cuvettes, which were analyzed on a PTI QM-1 spectrofluorometer. The instrument was calibrated using a Rhodamine B photon source and a NBS traceable lamp, and excitation and emission corrections were applied. Fluorometric dye standards were used to ensure continuity and replicate analyses performed to ensure adequate precision. Field (handled, but not deployed) and laboratory blanks (detectors soaked in deionized water or eluted without soaking) were run to test for contamination.


539

Original article

consistent signature comparable with deionized or tap water. Greenway sewage treatment plant showed much greater fluorescence and variation, with the appearance of miscellaneous peaks, presumably reflecting the composition of the feed water at the time. The composition of the Thames River was relatively consistent at a number of sampling sites, reflecting the dilution effect of the river. A representative sample is used. The magnitude of single exclusive peaks in fluorescence spectra can be used as indicators of concentration of a compound. However, unconstrained environmental samples consist of a mix of fluorescence spectra. In dilute samples, fluorescence is generally additive. However, at higher concentrations, there is significant interference between the absorption, excitation, and emission of various compounds. This means that a particular peak height is not simply indicative of concentration of the respective compound. In addition, there is great variation in the fluorescence of different compounds; fluorescence spectra are dominated by the most powerful fluorophores, such as Waters uranine. A diverse subset of a range of contaminated sites is drawn In Fig. 1, the ubiquitous peak at 300 nm (Dk=20) is on here to characterize water and charcoal eluent signa- ambiguous. It is generally regarded as indicating miscellaneous hydrocarbon fuels. Controlled extractions of vartures (Fig. 1). Groundwater springs discharging from ious fuels into cyclohexane are reported to give exquisite a sand and gravel aquifer provided a relatively clean,

Analysis of variance experiments suggested that the environmental signal was generally ‘‘machine noise’’ handling and sampling artefacts (Smart and Karunarantne 2001). A deionized water blank was used as a standard spectrum for comparative purposes. Synchronous scan spectra were run from excitation wavelengths (kex)=250–600 nm with wavelength offsets (Dk)=20 and 90 nm with 2-nm slit settings. Scan rates were 600 nm min–1 at 1-nm intervals. These Dk provide identification of common fluorescent dyes (Ka¨ ss 1998), petroleum compounds (Pharr and others 1992), polycyclic aromatic hydrocarbons (PACs; Kumke and others 1995), and natural organic matter (Mobed and others 1996). Data were smoothed with a triple pass Hanning filter in Excel spreadsheets, and visualization was developed in Surfer and Grapher (Golden Software 2000).

Fig. 1 Fluorescence spectra from three contrasting environments. Groundwater discharge from a spring-fed pond, discharge from a sewage treatment plant, and a large surface river. The thin line represents spectra from granular activated charcoal with 2-day exposure (4 days for the Thames River). The thick line represents 4-day exposure (6 days for the Thames River). The dashed line is an arbitrary water sample during the period of sampling; the dotted line is the spectrum for standard deionized water

540


Original article

differentiation at astonishingly small Dk (Pharr and others 1992). In our experience, the complex composition of fuels (Sigsby and others 1987), fate in the environment (Baehr 1987), Rayleigh scattering arising from the small Dk, and the coincidence of the Raman peak for water make such characterization difficult, to say the least. In general, volatilization and degradation of hydrocarbon contaminants appear to result in displacement of the hydrocarbon peak(s) to higher Dk and longer emission wavelengths. Many peaks in this region of the spectrum are seen occasionally in water samples, indicating acute (i.e., shortlived) contamination, typically in response to runoff events. Other peaks are repeatedly associated with particular sites and indicate a characteristic chronic contamination problem. However, such intriguing features are not of primary interest in this paper. The broad peak in waters (Fig. 1) at around 400 nm (Dk=20 and 90 nm) is characteristic of natural organic matter(e.g., Mobed and others 1996). This peak arises from a complex of humic and fulvic acids, which shift in composition and expression as the complex degrades or shifts in composition. This ubiquitous peak compromises identification of many other compounds in this region of the spectrum, by masking and absorption (Mobed and others 1996; Roch 1997). The 300-nm peak is often inversely correlated with the magnitude of 400-nm peak. One reason for this is that the natural organic matter tends to be opaque in the ultraviolet and is suppressing the 300-nm peak (Mobed and others 1996). Waters frequently show a peak at 512 nm (Dk=20), which is characteristic of uranine (i.e., sodium fluorescein, a common tracer dye), presumably indicating the presence of tinted ethylene glycol antifreeze. The peak is acute in rivers (where it is indicative of road runoff, or a vehicle accident), but chronic at locations such as landfill sites, former wrecking yards, and vehicle-maintenance compounds. It should be no surprise that many tracer compounds are commonly encountered in the environment; they are readily available, widely used, and freely disposed of. The critical question is whether their presence significantly compromises fluorescent tracing.

Water and eluent spectra Any comparison of water and the eluent spectra presupposes that the loading of the latter is adequately characterized by the former. This may be the case in the laboratory, or for groundwater and large river sites, which can be accepted as having a reasonably consistent chemical composition. However, our data suggest that spectra obtained from creeks, sewage outfalls, storm drains, and intermittent flows are highly variable. Such sites, therefore, are poorly characterized by occasional water samples. The nature of our investigation left this factor inadequately controlled. Figure 1 shows the GAC eluent and water spectra for the three selected sites. In general, the charcoal eluent

provides clear amplification of the water spectrum, and reasonable correspondence of spectral form . As anticipated, the sewage treatment plant shows the poorest match, reflecting the idiosyncratic loading of the detectors. A ‘‘Rhodamine’’ dye shoulder is observed in sewage discharge (kem580 nm, Dk=20) charcoal, but is absent from water samples, suggesting that the red dye is more intermittent than uranine seen in both water (512 nm at Dk=20) and eluent (520 nm at Dk=20). A slight red shift of the eluent uranine peak arises from the elevated pH of the eluent solution (Ka¨ ss 1998). To what extent does charcoal eluent amplify the spectrum seen in water? Linear gain of the GAC-eluent can be expressed as a spectral ratio of GAC fluorescence intensity divided by water fluorescence intensity (Fig. 2). This shows performance approaches and exceeds the order of magnitude (10·) expected over much of the spectrum. However, the gain spectrum is far from flat, with significant gains only at higher wavelengths, and significant losses at shorter wavelengths, especially for longer exposure times. Furthermore, the magnitude of the gain appears to reflect the composition of the water, with the strongest gains occurring in the most heavily loaded sites at peaks in concentration, such as the uranine peak. This is of concern as it suggests that dominant fluorophores appear to be preferentially amplified. The effect of exposure time is evaluated in the next section. The gain from charcoal is important where fluorophores are at the detection limit. Thus, subtle peaks become more marked in eluent spectra. Note the use of a logarithmic intensity scale in Fig. 2 to show that this is not simply a scale effect arising from differences in peak height. A result of the variable gain spectrum is that subtle and equivocal peaks in water are clearly defined in charcoal, e.g., the appearance of ‘‘Rhodamine’’(580 nm), and 455-nm peaks in eluents, but not waters from sewage and the river (Fig. 1). However, we reiterate that the water sample does not adequately characterize the loading environment, especially at the sewage outfall. Of course, such a gain in sensitivity will amplify tracers regardless of their origin, which implies serious concerns for uranine as a tracer compound.

Exposure time In dye tracing, control detectors are commonly placed in sites unlikely to encounter tracer, in order to provide quality assurance concerning the validity of a sample spectrum with respect to contamination (i.e., accidental or incidental presence of tracer dye) and background (i.e., non-tracer compounds contributing to the fluorescence spectrum). Replicate detectors are not commonly used, but would provide an indication of at-a-site sampling precision. Replicate sample spectra are expected to posses a similar form, although differences in strength (fluorescence intensity) are anticipated from contrasts in environmental exposure time, eluent volume, GAC mass ratio,


541

Original article

Fig. 2 Gain in fluorescence intensity for activated charcoal detectors over water. The thin line represents 2-day exposure (4 days for the Thames River), and the thick line represents 4-day exposure (6 days for the Thames River). Dashed lines indicate a reference gain of unity

and elution time. Our replicates exhibited startling differences in spectral form, arising from at least two factors: environmental exposure time, and elution time. The former is considered here. Figure 1 demonstrates that 2- and 4-day replicate spectra are very similar in groundwater samples, but radically different in sewage discharge. In part, this reflects the inherent instability of composition of the sewage. However, many other samples (e.g., Thames River samples in Fig. 1) indicate a similar systematic bias, with longer exposure times reducing shorter wavelength intensity and enriching longer wavelength intensity. Unfortunately, the magnitude of this bias depends on the site; highly enriched waters (e.g., sewage discharge, Fig. 1) show a more marked transformation. The more volatile, shorter wavelength compounds appear to be initially scavenged very efficiently by the charcoal. Longer wavelength compounds are adsorbed less rapidly, but cumulatively. Are short wavelength compounds truly ‘‘lost’’ from the charcoal, or are they progressively bound more tightly? Or is this simply an adsorptive interference? Undoubtedly, absorptive quenching of the 300-nm peak does occur, but the magnitude of increase in the 400-nm peak is generally much less than the order of loss in the 300-nm peak (e.g., Fig. 1: groundwater discharge Dk=20 nm). The 542


nature of activated charcoal provides a range of binding energies, with a range of accessibility. The tracer Rhodamine WT provides an example of a compound so strongly bound that conventional elution releases only a small fraction of the adsorbed dye. However, the sensitivity to loading suggests an exchange process (Smart 1972). These results significantly compromise the use of activated charcoal for fluorescence screening. From the dye-tracing perspective, the strongest effects appear to occur over the initial few days of exposure; typical weekly deployment may reach a quasi equilibrium. (A limited set of tests suggest that uranine eluent concentrations may decline at exposure times approaching 1 week.) The high initial variation in charcoal scavenging performance suggests the possibility of artificially pre-loading detectors in order to provide adsorption kinetics as flat as possible. At tracer wavelengths (450–600 nm), the exposure time effect may peak at a few days. However, it is not clear how broad wavelength-based generalization might be; the behavior of all tracers may not follow this trend. Further experiments are needed to follow up on exposure time effects. Meanwhile, a further constraint is placed on quantitative analysis of activated charcoal eluent spectra: the form and intensity of a tracer peak depends on the adsorption environment and history.

Original article

better behaved; the kinetics appear to be of similar order to that of uranine. Approximately 1 day of elution achieves A large GAC sample was obtained from the ‘‘Brickpits’’, a about 80% of the 10-day concentration. These tests were performed with the objective of assessing known contaminated site with various hydrocarbon, lubricant, and degraded products present. A single char- the potential of GAC for contaminant screening using fluorescence spectra. In this context, fresh GAC is incoal–eluent mixture was established and aliquots were drawn at intervals of 1.7 and 17 min, 2.7, 8.4, and 27.8 h, transigent, possibly unusable. The situation appears more and 5 and 13 days. Spectra were computed at Dk=10, 20, reasonable for fluorescent tracer dyes. However, it is 50, and 90 nm on each aliquot. For each Dk, the spectra clearly important to establish the elution kinetics of contemporary tracer dyes, and to design elution protocol were projected against log time, gridded, and displayed. The resulting three-dimensional plots (Fig. 3) show how accordingly, if any quasi quantitative analysis is to be the spectrum evolves with time of elution. The grid figures attained. are visually appealing, but can not be scrutinized in detail. Contour plots (Fig. 4) allow more precise evaluation of relative gains and losses at particular times and wavelengths. Replicate experiments were performed. Conclusions The eluent spectra evolution depended on Dk. At Dk=50 and 100 nm, the peak attributed to natural organic matter The work we have undertaken to date has been exploratory rises monotonically over 10 days. There is slight decline at and empirical in character, and not sufficiently well conthe lower wavelengths of the spectrum after 1 h. At trolled to allow categorical conclusions. The complexity of shorter Dk, the spectral evolution is more complex. Very the fluorophore-organic environment makes controlled short wavelength peaks, presumably representing hydro- experiments desirable, but difficult to design for general carbon fuels (kem.290 nm), rise rapidly to peak between 10 applicability. The type of organic loading occurring in the Thames River, however, is probably typical of that and 60 min, after which they decline. Less volatile hyencountered in many rivers and tracer tests in North drocarbons rise monotonically to give 380- and 390-nm peaks. Similarly, uranine (520 nm) increases monotoni- America and Europe. In general, granular activated charcoal provides amplifically across the 10-day period. There seems to be little cation of the fluorescence signature of waters. However, loading at the wavelengths of Rhodamine (580 nm); presumably there is none of this tracer present. However, the gain is non-linear and conditioned by the loading environment, and exposure time. Note that with GAC we the fluorescence intensity at this wavelength seems to decline over a period of hours and then increase slightly found uranine peaks in all the environments sampled. Only a mains water leak and controls remained clean! The over days. The results displayed in Figs. 3 and 4 are conditional on former serendipitous result suggests an invaluable field the loading on the charcoal. We selected a reasonably rich diagnostic for leaking water mains. Rhodamine-type peaks, along with many others, appeared frequently spectrum for the evaluation. Generally, more volatile enough to make the issue of background definition and compounds appear to be very rapidly eluted, but are stability a serious concern (Smart and Karunaratne 2001). subsequently lost from the analysis, in part through absorption, but evaporation into free air may also play a The gain of GAC detectors is sustained and improved at longer wavelengths with exposure time, whereas shorter role. Natural organic matter desorption (400 nm) is

Elution time

Fig. 3 Three-dimensional portrayal of the evolution of synchronous scan spectra for Dk=10 nm with elution time


543

Original article

Fig. 4 The evolution of spectra at four different Dk with elution time. Note that although the wavelength and time scales are identical in all four plots, the intensity contour intervals are different (104 for Dk=10 and 20 nm, and 2·104 in the lower two plots)

wavelength fluorescence intensity is reduced. The reasons for this are unclear, and the extent of absorptive quenching needs to be investigated. However, general volatility of many of these compounds may render them weakly adsorbed, and readily displaced as higher concentrations of larger molecules, fluorescing at longer wavelengths, are accumulated. Organic adsorption is also conditioned by the overall loading environment, with transitions from gain to loss occurring much more rapidly in enriched sites. Things seems to be more settled after a number of days. This may have led to the general practice of deploying GAC detectors for around 1 week (Alexander and Quinlan 1992) It may be possible to stabilize the condition by preloading GAC detectors, perhaps with organic matter, prior to their field deployment. It is important to match the elution period to the compounds of interest. The rate of elution of organic materials from GAC and the loss of more volatile compounds appears to follow reproducible forms. First, it is important to determine the elution rate constants for principle tracers. Second, it may be possible to strategically design elution protocols that allow preferred extraction of tracers and suppressed elution of background. The distinct problem of tracer contamination remains unresolved. No investigation of particular tracer dyes was undertaken. However, some tracers, or close analogs, were present in the environment. The relationship between compound, concentration, loading history, and adsorption are not well understood, but sensitivity appears to decline with exposure time. Thus, a spike of dye early on in deployment will be adsorbed more than an equivalent later spike. Overall, GAC detectors appear reasonably well suited to qualitative dye tracer applications, but should be used with a clear understanding of the non-linearities and artifacts inherent in their use. 544


Acknowledgements The Natural Sciences and Engineering Research Council of Canada generously funded this work through equipment and research grants. K. Karunaratne assisted in the laboratory and field work.

References Alexander C, Quinlan JF (1992) Practical tracing of groundwater with emphasis on karst terranes. A short course manual. Geological Society of America, Boulder Baehr AL (1987) Selective transport of hydrocarbons in the unsaturated zone due to aqueous and vapor phase partitioning. Water Resour Res 23:1926–38 Cheremisnoff NP, Cheremisnoff PN (1993) Carbon adsorption for pollution control. PTR Prentice Hall, Toronto Golden Software (2000) Surfer and Grapher. Golden, Colorado Ka¨ ss W (1998) Tracing technique in geohydrology. Balkema, Rotterdam Kumke MU, Lohmannsroben HG, Roch T (1995) Fluorescence spectroscopy of polynuclear aromatic compounds in environmental monitoring. J Fluor 5:139–153 Matisova E, Skrabakova S (1995) Carbon solvents and their utilization for the preconcentration of organic pollutants in environmental sampling. J Chromatogr A 707:145–179 Mobed JJ, Hemmingsen SL, Autry JL, McGowan LB (1996) Fluorescence characterization of IHSS humic substances: total luminescence spectra with absorbance correction. Environ Sci Technol 30:3061–3065 Pharr DY, McKenzie JK, Hickman AB (1992) Fingerprinting petroleum contamination using synchronous scanning fluorescence spectroscopy. Groundwater 30:484–489 Roch T (1997) Evaluation of total luminescence data with chemometrical methods: a tool for environmental monitoring. Analyt Chem Acta 356:61–74 Sigsby JE, Tejada S, Ray W (1987) Volatile organic compound emissions from 46 in-use passenger cars. Environ Sci Technol 21:466–475

Original article

tracer dyes etc. In: Beck BF, Petit AJ, Herring JG. (eds) Smart CC, Karunarantne KC (2001) Statistical characterisation of Hydrogeology and engineering geology of sinkholes and karst – natural background fluorescence as an aid to tracer test design. 1999. Balkema, Rotterdam, pp 307–314 In: Beck BF, Herring JG (eds) Geotechnical and environmental Vesper DJ, Rice JE, Fenstermacher RF (1999) Use of GOREapplications of karst geology and hydrology. Balkema, SORBER modules to screen for organic contaminants in karst Rotterdam, pp 271–276 springs. In: Beck BF, Petit AJ, Herring JG (eds) Hydrogeology Smart PL (1972) A laboratory evaluation of the use of activated and engineering geology of sinkholes and karst – 1999. Balkema, carbon for the detection of tracer dye Rhodamine WT. MSc Rotterdam, pp 351–355 Thesis, University of Alberta Tucker RB, Crawford NC (1999) Non-linear curve fitting analysis as a tool for identifying and quantifying multiple fluorescence


545