Introduction

Because of their wide use, petroleum products are frequent pollutants of soil and groundwater. This situation results mainly from leakages of underground storage tanks or pipelines and from accidental spills (Rosenberg and Ron 1996; Head and Swannell 1999). In a context of environmental protection, laboratory-scale biotreatability studies are quite useful (Skladany and Baker 1994; Salanitro 2001). On the one hand, tests performed in soil microcosms provide data about the kinetics of pollutant degradation (Miles and Doucette 2001), the level of residual contaminant in the polluted zone and the possible toxic by-products or end-products which could accumulate. On the other hand, tests performed in liquid cultures provide crucial information about the intrinsic biodegradability of pollutants and the biodegradation capacity of the local native microflora.

The intrinsic biodegradability of petroleum has been extensively investigated in batch cultures. The substrates used are generally crude oil, hydrocarbon cuts obtained after distillation or individual hydrocarbons. Among the latter, the most representative low-molecular-weight molecules belonging to normal, branched, cyclic alkanes and aromatics were shown to be readily biodegraded by bacterial isolates (Rosenberg and Ron 1996; Di Lecce et al. 1997; Leahy and Olsen 1997; Paje et al. 1997; Meyer et al. 1999). Even peculiar types of molecules containing quaternary carbon atoms or consecutively methylated carbon chains were degraded by micro-organisms using specialised catabolic pathways (Fall et al. 1979; Bhattacharya et al. 2003) and/or by co-metabolism (Beam and Perry 1974). As a result of the high biodegradability of the low-molecular-weight hydrocarbons, gasoline, the lightest liquid oil cut from oil distillation was extensively biodegraded by a large number of various environmental microflorae (Solano-Serena et al. 1999).

In contrast to low-molecular-weight hydrocarbons, polycyclic aromatics and hydrocarbons included in the asphaltene fraction are usually considered as being only slightly biodegradable because of their insufficient availability to microbial attack (Gibson and Subramanian 1984; Cerniglia 1992; Kanaly and Harayama 2000). Among the pentacyclic triterpanes, hopanes (Ourisson et al. 1979) are so stable that they are commonly used as ubiquitous biomarkers for the assessment of biodegradation levels of crude oil. They were shown to be only slightly biodegraded by specialised microflorae under laboratory conditions (Frontera-Suau et al. 2002).

The biodegradation of diesel oil (DO) is more debatable because it is a middle distillate composed of hydrocarbons ranging from C11 to C25. Mineralisation of DO was reported to be complete in air-sparged liquid cultures (Geerdink et al. 1996) and incomplete in agitated flasks (Olson et al. 1999); and it was generally accepted that degradation of DO hydrocarbons in soil microcosms was incomplete (Chaineau et al. 1995; Gallego et al. 2001; Seklemova et al. 2001). This partial recalcitrance possibly resulted from the low intrinsic biodegradability of DO, a limitation in hydrocarbon transfer or an insufficient amount of available dioxygen. Furthermore, numerous degradation tests were performed in open systems where substantial amounts of substrate might disappear by volatilisation or irreversible adsorption to soil, as indicated by Song et al. (1990).

In fact, commercial DO is a complex hydrocarbon mixture of thousands of individual components. Its hydrocarbon composition depends on the origin of the crude oil used for the distillation process (straight-run DO), on the refining processes of crude oil and on the mixtures added by the refiner for final formulation. In the present study, the aerobic biodegradability of DOs with different hydrocarbon compositions was examined. Accordingly, the DOs used in the present work originate from different refinery processes, to investigate the intrinsic biodegradability of DO over the largest composition range and to address the suitability of these DOs to biodegradation in the environment. In biodegradation tests, we used two different microflorae. The first was an activated sludge from an urban wastewater treatment plant, as usually recommended by standard test procedures. The second was a hydrocarbon-polluted-soil microflora that possibly harboured acclimated species with specialised degradation capacities.

Materials and methods

Culture media and microflorae

The vitamin-supplemented mineral salt medium described by Bouchez et al. (1995) was used as a nutrient solution. Each DO type was added at 400 mg l−1 as a sole carbon source. Two microbial suspensions were used in the degradation tests. The first was a microflora from an urban wastewater treatment plant, obtained by centrifugation of an activated sludge at 15,000 g for 10 min (3 g l−1, dry weight). It was used after re-suspension into nutrient solution at a final concentration of 100 mg l−1 (dry weight). Sludge pellets were stored at −80°C for several months without significant loss of degradative capacity. The second suspension was a microflora from a DO-polluted clay-like gravel. The soil dry mass obtained by weight loss after heat treatment was 73%. The pH value determined according to the NF ISO 10390 standard was 6.05. The hydrocarbon content determined by gas chromatography (GC) with flame ionisation detector (FID) after cyclohexane:acetone (85:15, v/v) extraction was 10 g kg−1 dry soil. Microbial suspensions for biodegradation tests were prepared by dispersing 5 g l−1 of soil sample into the nutrient solution.

Diesel oil characteristics

The DOs involved in commercial DO formulation derive from refining processes. Straight-run DO was obtained from distillation units. Hydrocracking DO and light-cycle DO came from conversion units, hydrocracking and fluid catalytic cracking, respectively, converting heavy fractions to light ones. Fischer–Tropsch DO was a synthetic fuel produced by CO plus H2 condensation (Guibet 1997). Supplemented DO was a hydrocracking DO with a light aromatics cut.

Biodegradation tests

Biodegradation tests were generally performed for 4 weeks at 30°C in 120-ml agitated flasks closed with Teflon-coated stoppers and sealed with aluminium caps. Tests were started by the addition of 5 μl of DO to 10 ml of inoculated culture medium. The overall degradation kinetics were monitored at regular intervals with GC analysis of headspace CO2. Endogenous respiration was similarly monitored in control flasks without DO addition. Experiments were performed in five flasks and abiotic controls supplemented with HgCl2 were run under similar conditions. At the end of the incubation period, 10 ml of dichloromethane were introduced into the flasks, which were stored overnight at −20°C before extraction. The organic phase was evaporated to 1.5 ml and the residual DO was analysed by GC-FID, with dotriacontane (nC32) at 50 mg l−1 in dichloromethane as an internal standard. The final CO2 production was determined in the flask headspace by GC after acidification of the medium with 200 μl of 4 N HCl.

The DO degradation yield was calculated as the ratio of the amount of substrate degraded in test flasks to the amount of substrate recovered in abiotic controls. The net CO2 production was the difference between the final quantities of CO2 in the test flask and in the hydrocarbon-free flask. The mineralisation yield was the carbon ratio of the net CO2 produced to the DO consumed.

GC analyses

CO2 was measured using a GC equipped with a thermal conductivity detector and a Porapak Q column (80/100 mesh, 2 m), using an external standard method (Solano-Serena et al. 1999).

DOs were analysed with a 3400 GC (Varian, USA) equipped with a FID and a DB-5 column (60 m), using helium as the carrier gas. The detector temperature was 310°C. The column temperature was first set at 50°C for 10 min and then increased to 310°C at 2°C min−1. The injector temperature was initially 50°C for 0.2 min and then increased to 280°C at 180°C min−1. Quantification of DO was obtained using JMBS software (BORWIN) by integrating the global area of the separated peaks and unresolved-mass complex (UMC) to the baseline.

Fractions containing saturated and aromatic hydrocarbons of DOs were separated by liquid adsorption chromatography (LC). Mini-columns of Pasteur pipettes containing regenerated silica (1 g of SiO2, 60:70–230 mesh from Merck) were used. At the top of the column, 0.5 g of Na2SO4 was added and chromatography was carried out using 20 mg of DO. Saturated alkanes (linear, branched and cyclic alkanes) were eluted with 1.95 ml of hexane and 0.65 ml of a hexane:dichloromethane mixture (4:1, v/v). Aromatics were eluted with 3.25 ml of a hexane:dichloromethane mixture (1:1, v/v). Fractions were then quantified by GC-FID using a Varian 3400 GC and a DB-5 column (60 m), as previously described for DOs.

The structural hydrocarbon classes of DOs were determined by HPLC with refractometric identification (RI), using a Prostar system (Varian, USA) and a silica-NH2 column. Operating conditions were similar to the NF EN 12916 standard.

Results

Composition of DOs

The composition of the DOs used was determined using two methods. Saturated alkanes (fraction S) and aromatics (fraction A) were separated by LC and each fraction was quantified by GC-FID (Fig. 1). In fraction S (Fig. 1A), all the linear alkanes were identified. Branched hydrocarbons were not identifiable but they were globally quantified, assuming that their responses in the FID were identical to those of linear alkanes. The aromatics content present in fraction A was evaluated using the same method as that used for branched alkanes (Fig. 1B).

Fig. 1
figure 1

GC-FID pattern of the A saturated and B aromatic hydrocarbon fractions from light cycle DO. nCx correspond to linear alkanes with x atoms of carbon. Far, NPri, Pri and Phy indicate farnesane, nor-pristane, pristane and phytane, respectively

Full size image

DOs were also analysed using HPLC with RI detection (Fig. 2). Aliphatic hydrocarbons were first eluted at a retention time (RT) of 9 min. Mono-aromatics and tri-aromatics were eluted as single peaks at RTs of 12 min and 16 min, respectively. Di-aromatics were resolved into several peaks ranging from 17 min to 26 min RT, depending on the substituents of their aromatic rings.

Fig. 2
figure 2

HPLC pattern of light-cycle DO. Hydrocarbon classes are indicated: a saturated hydrocarbons, b mono-aromatics, c di-aromatics, d tri-aromatics. A silica-NH2 column was used as described in the Materials and methods. The heptane flow was 0.5 ml min−1

Full size image

Data from HPLC separation with RI detection was in agreement with those from LC separation with GC-FID quantification. Table 1 summarises the main characteristics of the DOs used. Fischer–Tropsch DO was mainly composed of linear hydrocarbons, while hydrocracking DO and light-cycle DO were principally constituted of aromatics. Straight-run DO and commercial DO contained large amounts of branched and cyclic alkanes.

Table 1 Hydrocarbon composition of the DOs used. The composition was determined by GC-FID after a pre-separation step by LC, as described in the Materials and methods
Full size table

Degradation of commercial DO by two different microflorae

The two microflorae tested were an activated sludge and the microflora from a polluted soil. Since no standard procedure was available for the use of soil microflorae in biodegradability tests, the experimental conditions had to be defined first. Preliminary experiments indicated that the best procedure for seeding was to use soil aliquots as inocula of test cultures. Figure 3A shows the effect of the initial amount of commercial DO on mineralisation kinetics. The final CO2 produced in test flasks increased linearly with the initial amount of DO introduced (Fig. 3B). Nevertheless, a substantial amount of CO2 was produced in the absence of DO, indicating that the soil amount used for seeding had to be reduced to minimise endogenous respiration. Endogenous respiration was adequately limited by using 5 g l−1 of soil. The kinetics of substrate mineralisation determined after deduction of endogenous CO2 were found to be similar using 5 g l−1 or 50 g l−1 of soil. However, using 1 g l−1 of soil as an inoculum, the mineralisation kinetics were substantially slowed down and did not ensure reproducible end-point values (data not shown).

Fig. 3
figure 3

Effect of the initial commercial DO concentration. A Effect on the biodegradation kinetics of CO2 production. B Correlation between final CO2 production and initial DO. Initial DO concentrations were: 0 mg l−1 (open triangles), 40 mg l−1 (open squares), 80 mg l−1 (open circles), 160 mg l−1 (filled triangles), 240 mg l−1 (filled circles), 400 mg l−1 (filled squares). The tests were performed in 120-ml flasks with 10 ml of medium and were inoculated using 5 g l−1 of soil. The incubation time was 30 days. L Litres

Full size image

At the end of the incubation period, residual DO hydrocarbons were analysed by GC-FID (Fig. 4). All linear alkanes were completely degraded within 3 days (data not shown) by the polluted-soil microflora. Some identifiable branched alkanes, such as farnesane, pristane and phytane, were also totally degraded. However, the UMC was only partially degraded and, in particular, the long-chain hydrocarbons of the UMC were still present. Activated sludge seemed to be less efficient than polluted soil. Even though n-alkanes were completely removed, some identifiable branched alkanes, such as farnesane, pristane and phytane, were detected by GC at the end of the incubation period (Fig. 4C). Compared with the residual hydrocarbons from the tests carried out with the soil microflora, the amount of UMC hydrocarbons was higher and hydrocarbons of lower-molecular-weight were still present.

Fig. 4
figure 4

GC-FID patterns of residual hydrocarbons from commercial DO after degradation: A abiotic control, B flask inoculated with polluted soil, C flask inoculated with activated sludge. The incubation time was 28 days

Full size image

As shown by the DO compositions determined by HPLC (Table 2), there was an increase in the relative abundance of saturated compounds (83% and 73% for the polluted soil and the activated sludge, respectively) compared with that of the initial content (64%), which accounted for the recalcitrance of branched and cyclic alkanes. Aromatics were more extensively degraded than branched and cyclic alkanes, especially with the soil microflora. Mono-aromatics were more abundant than di- and tri-aromatics in the final solution, suggesting the presence of highly alkylated mono-aromatics with low degradation properties.

Table 2 Variation in commercial DO composition (%) after incubation during 28 days with a polluted-soil or activated-sludge microflora. Values were determined by HPLC, as described in the Materials and methods
Full size table

Biodegradability of various DO types

The degradation performances of both microflorae were investigated for each DO type (Table 3). Abiotic recovery rates in all assays were higher than 75%, except for the light-cycle DO which contained many volatile compounds. Each DO type was biodegraded to a large extent by the polluted-soil microflora (88–95%). For all type of DO tested, this microflora was more efficient than the one from activated sludge. Therefore, the effect of the type of DO on biodegradation performances was much more pronounced using activated sludge, since its biodegradation rates varied from 45% to 82%. Supplementation of hydrocracking DO with light mono-aromatics increased the biodegradation rate from 61% to 82%, probably because the aromatics added were slightly water-soluble and thus easily biodegradable. Fischer–Tropsch DO contained a large amount of linear and slightly branched alkanes and was also extensively biodegraded. Similarly, the biodegradability of commercial DO was higher than that of straight-run DO because of a higher content in linear alkanes.

Table 3 Degradation of various DO types by microflorae from a polluted soil and an activated sludge. Each degradation test was performed in five flasks, except where indicated. The incubation time was 28 days
Full size table

The mineralisation yields of DOs by the microflorae are shown in Table 3. For all types of DO, the mineralisation yields were higher than 50%. Taking into account that a significant fraction of the carbon source is used for cell biomass, this shows that only limited amounts of metabolic intermediates accumulated during the degradation of DOs.

Discussion

In soils contaminated by petroleum compounds, the extent of natural attenuation mainly depends on the hydrocarbons present in the contaminated matrix (Husemann 1995). DO is a potential pollutant of soils and its composition may vary according to the processes used for its preparation. Hence, the biodegradability of various types of DO was studied in the present work.

The tests used for the determination of biodegradability of commercial products, such as those of OECD, are not suitable for petroleum products since they were designed for water-soluble compounds with low volatility properties. An easy-to-use test was developed for oil products by CONCAWE, i.e. the oil companies’ European organisation for the environment, health and safety (Battersby et al. 1999). Although highly reproducible, this test, based on CO2 production from low amounts of substrate, does not allow the measurement of residual hydrocarbons. The methodology used in the present study is based on both the determination of residual hydrocarbons and the production of CO2. Residual hydrocarbons are obtained by means of GC-FID, using an internal standard to prevent overestimation of the degree of biodegradation, as recommended for crude oil by Vinas et al. (2002). Our analytical procedure was quite satisfactory, as shown by the hydrocarbon recoveries measured in abiotic flasks (higher than 75% except for light cycle DO). Moreover, the mineralisation yield was estimated in order to know whether metabolic intermediates accumulated or not.

Using activated-sludge microflora, the biodegradation rates determined for each DO type were lower than those determined using polluted-soil microflora and showed more variability. This indicates that activated-sludge microflorae from wastewater treatment plants may be not appropriate for measuring the biodegradation of oil products and that adapted microflorae are probably more suitable, as indicated by Battersby et al. (1999). Moreover, the microflora from polluted soil used in the present study seems to be adapted to the degradation of recalcitrant hydrocarbons. In the environment, hydrocarbon-degrading micro-organisms are found ubiquitously (Jones and Edington 1968; Rosenberg 1992). The extended degradation capacities of polluted-soil microflora compared with those of the activated-sludge microflora account for the probable occurrence of a selective adaptation of the native microflora (Liu and Suflita 1993; Atlas 1995).

The relative recalcitrance of hydrocarbon types was illustrated in the tests performed with both microflorae. For each type of DO tested, linear alkanes were totally degraded within 1–2 days (data not shown). Recalcitrant hydrocarbons include mainly branched alkanes and aromatics (McKenna and Kallio 1971; Pirnik et al. 1974; Fall et al. 1979; Nakajima et al. 1985; Rontani et al. 1986). Moreover, mineralisation yields were higher than 50% in each case (Table 3). Considering that the mineralisation value of extensively degraded hydrocarbons such as n-hexadecane was around 70% (Battersby et al. 1999), it could be deduced that, apart from CO2 production, biomass formation may account for a large part of the carbon balance and that the accumulation of metabolic intermediates from DO was rather limited.

There is little data in the literature on the biodegradability of DO in closed liquid systems. Using an adapted bacterial consortium produced by subculturing, Richard and Vogel (1999) found that 90% of a commercial DO was biodegraded within 50 days. A similar extent of degradation was obtained by Marquez-Rocha et al. (2001) within 2 weeks. In the present study, the data obtained with commercial DO using a soil sample as an inoculum are in agreement with those results. Furthermore, the level of biodegradation (higher than 88%) obtained with all the DO types tested indicates that DO with its variable hydrocarbon composition is a highly biodegradable product. The biodegradation test developed in this work constitutes a supplementary and useful tool to study efficiency in pollutant attenuation.

Various methods have been proposed to characterise the hydrocarbon-degrading microflorae of soil (Theron and Cloete 2000; Widada et al. 2002; Margesin et al. 2003). Numerous studies have been used to investigate the composition of the microflorae involved in hydrocarbon degradation, such as culturing and molecular techniques based on 16S rRNA analysis or the characterisation of metabolic genes. In this study, we demonstrate an important difference in the biodegradation capacities of two microflorae (activated sludge, polluted soil); and experiments are in progress in our laboratory to determine the community structure of soil microflora.