Water, Air, and Soil Pollution

, Volume 194, Issue 1, pp 151–161

An Improved Gravimetric Method to Determine Total Petroleum Hydrocarbons in Contaminated Soils

Authors

    • Grupo de Bio-Geoquímica Ambiental, Laboratorio de FisicoQuímica Ambiental (LAFQA) Instituto de GeografíaUniversidad Nacional Autónoma de México (UNAM)
  • Ana Paulina Avila-Forcada
    • Dirección General del Centro Nacional de Investigación y Capacitación Ambiental (CENICA), Instituto Nacional de EcologíaSan Rafael Atlixco 186
  • Margarita Eugenia Gutierrez-Ruiz
    • Grupo de Bio-Geoquímica Ambiental, Laboratorio de FisicoQuímica Ambiental (LAFQA) Instituto de GeografíaUniversidad Nacional Autónoma de México (UNAM)
Article

DOI: 10.1007/s11270-008-9704-1

Cite this article as:
Villalobos, M., Avila-Forcada, A.P. & Gutierrez-Ruiz, M.E. Water Air Soil Pollut (2008) 194: 151. doi:10.1007/s11270-008-9704-1

Abstract

A gravimetric method to determine heavy fractions of total petroleum hydrocarbons (TPH) in soils is reported. The method was adapted and calibrated by modifying previous standard methods published, incorporating energy and cost savings where possible. Artificially contaminated soils with different organic matter content, and aged in stationary mode for a period of 8 months were used for calibration. Insufficient solvent evaporation was identified as the most prevalent and largest positive interference in the gravimetric detection. To overcome this, while minimizing the need for heating, a combination of three 10-min rotary evaporator steps and 30 min of vacuum in a desiccator were applied, for a total solvent volume of 60 ml. Hexane was chosen as the extraction solvent and a 40–60 min treatment in an ultrasound bath of 260 W was found suitable to extract 80–95% of TPH extracted by the Soxhlet method. Finally, the use of silica gel for cleanup of co-extracted natural organic matter was found unnecessary, because of the low amounts co-extracted for soils with up to 5% organic carbon, and because the chemical nature of the co-extracted organic matter prevents its selective adsorption to silica.

Keywords

Contaminated soilsGravimetric methodSilica gelSolvent evaporationTotal petroleum hydrocarbonsUltrasound extraction

1 Introduction

Soil pollution by petroleum hydrocarbons usually originates from spills or leaks of storage tanks during fuel supply and discharge operations. Most petroleum components are hazardous to human health and to the soil biota (MDEP 1994); therefore, measurement of their total concentration (total petroleum hydrocarbons—TPH) is the initial and most general indicator of hazard by hydrocarbons for environmental diagnostic purposes.

The US Environmental Protection Agency (USEPA) has published a series of methods for quantifying TPH in solid matrices (including soils; Table 1). The four categories of analytical methods described in Table 1 are based on the final detection method, i.e. gravimetry, infrared spectroscopy, gas chromatography, and immunoassays. Currently, Mexican legislation has implemented many of these as standard methods for diagnosis of contaminated soil sites, especially methods that employ various forms of gas chromatographic techniques, but also a gravimetric method has been recently adopted (NMX-AA-134-SCFI-2006 2006).
Table 1

USEPA methods to quantify TPH in various matrices (including soil; taken from Weisman, 1998)

Detection method

EPA method no. (year of most recent revision)

Extracting solvent and range of hydrocarbon species extracted

Approximate detection limit, mg/kg (soil)

Aprox. Cost/samplea

Advantages

Disadvantages

Gravimetric

9071 B (1998)

n-Hexane. Most compounds except volatile fractions

50

50

Simple, quick, and inexpensive

Lack of specificity, low sensibility, high loss of volatiles, prone to interferences; only useful for total quantification.

1664 A, for water samples (1999)

Infrared (IR)

8440 (1996)

Supercritical CO2. Most compounds except volatile and very heavy fractions

10

50–80

Simple, quick, and inexpensive

Lack of specificity, low sensitivity, high loss of volatiles, poor extraction of high molecular weight hydrocarbons, prone to interferences; only useful for total quantification

Gas chromatography (GC)

8015 C (2000)

Methanol, n-hexane, 2-propanol, ciclohexane, acetonitrile, acetone, methylene chloride, or mixtures of these. Normally between C6 and C25 or C36 (may be modified for a higher range)

10

100–150

Detects a broad range of hydrocarbons compounds; simple and sensitive; chemical speciation is possible.

Normally it cannot detect compounds below C6; does not detect polar hydrocarbons (alcohols, ethers, etc.); chlorinated solvents may be quantified as TPH

Immunoassay

4030 (1996)

The extractant depends on the commercial kit. Aromatic hydrocarbons (BTEX, PAHs)

10–500

13b

Simple, quick, inexpensive, and may be used in the field

Low sensitivity, prone to interferences, primarily measures aromatics; low accuracy and precision; must be used as a screening method; only useful for total quantification

64c

aUS dollars, MDEP (1994)

bCMECC (1996)

cVRAI (2007)

Chronologically, gravimetric methods were among the first ones developed, but were readily superseded with the advent of more specific spectroscopic and chromatographic methods that additionally allow for varying degrees of hydrocarbon chemical speciation. Presumably, for this reason, gravimetric methods show insufficient investigation pertaining proper method development, calibration, and validation that allow identification of the important variables that control interferences to the final gravimetric measurement. For example, it is not clear what conditions are required (and what their sensitivity is) to ensure complete evaporation of the solvent once the TPH extraction has been performed. Frequently, the procedures are vaguely stated to complete distillation in less than 30 minutes (EPA Method 9071 B 1998; NMX-AA-134-SCFI-2006), but no indication is given about the way to guarantee or to quantify such complete distillation. Thus, there is a high potential either for TPH overestimation, or for over expenditure of energy to ensure complete distillation, which increases the costs of analyses.

For laboratories with low resources, such as rural laboratories in developing countries, gravimetry is the method of choice due to the lower costs involved as compared to other methods. The most basic laboratory is equipped with an analytical balance as a basic tool, and thus, gravimetric methods require no additional expense for detection of hydrocarbons. Additionally, gravimetry allows for simplicity and low-cost analysis per sample (Table 1), which decreases considerably overall costs in situations where large numbers of samples are required for analysis, as may occur in conditions of high soil pollution heterogeneity.

Gravimetric detection is non-specific, although quite sensitive, and the potential for interferences is high. The present work focuses on developing/calibrating a gravimetric method for non-volatile TPH quantification in contaminated soils, in which cost-savings, simplicity, and speed have been defined as sought-after attributes for general applicability in low-resources laboratories. This investigation provided the procedures to recognize the important parameters for accurate quantification of TPH, identifying especially the highly sensitive steps that control the final precision and accuracy of the method.

The starting gravimetric method chosen in the present work is EPA Method 9071 B (1998), in combination with extraction EPA Method 3550 B (1996), which utilizes ultrasound treatment, both of which form the basis for the official Mexican norm NMX-AA-134-SCFI-2006. Ultrasound was chosen over Soxhlet extraction to further decrease costs. The latter method provides a highly quantitative extraction of TPH but is considerably more costly time- and energy-wise. Furthermore, as a variation from the published methods, the use of an ultrasound water bath instead of a probe horn was investigated, due to the higher availability of the former as basic lab equipment. Three principal factors were investigated in the overall method proposed: Extraction efficiency, quantitative evaporation of the extracting solvent, and co-extraction of natural organic matter. In this latter, the need for silica-gel cleanup of the extract, as recommended in some reference methods, was additionally investigated.

2 Materials and Methods

2.1 General Gravimetric Method

Soil samples were sieved (Sieve 10, ASTM—2 mm, followed by Sieve 200, ASTM—75 μm), and dried to 105°C for 12 h (Schwab et al. 1999; Xie et al. 1999). These were homogenized mechanically for several hours, and approximate accurately weighed 10 g subsamples were placed in round flasks previously dried (105°C) to constant weight. Ten grams of anhydrous Na2SO4 were added to create a free-flowing powder (EPA Method 3630 B 1996), and 35 ml n-hexane (here to forth referred to simply as hexane) were used for extraction in an ultrasound bath (EPA Method 3550 B 1996), the conditions of which were investigated for optimal extraction efficiency (cf. Section 3). The extracts were filtered through a column packed first with 0.6 g treated cotton or glass fibre, and optionally second and third with 5 g treated silica gel, and 1 g treated celite (EPA Method 3630 B 1996), respectively, and washed down with additional 25 ml hexane to complete 60 ml in the final liquid extract for analysis. The hexane was evaporated in a rotary evaporator, followed by drying of the flask outer walls with lint-free absorbent paper and evaporation of the remnant hexane under variable conditions (cf. Section 3). The residue was weighed in an analytical balance and designated as TPH.

The equipment and material used were: a semi-microanalytical Sartorious R 2000 balance; a Heidolph Laborota Digital 4003 rotary evaporator, equipped with Rotovac Senso vaccum system, and Rotacool cooling system (the conditions on the rotary evaporator were as follows: Water bath temperature = 40°C, Cooling water temperature = 14°C, Rotation speed = 70 rpm, automatic mode. Subsequently the flasks were dried with absorbent paper and placed in a desiccator for 2 h); two ultrasonic baths of 5.7 l: a Bransonic, Branson 3510 R-MTH of 100 watts, and a Fisher Scientific, FS60, of 260 watts; two Pyrex glass desiccators with silica gel desiccant: one of 7.5 l with plate diameter of 230 mm, and a second one of 2.2 l and plate diameter 140 mm connected to lab vacuum; a Felisa 243, Series 87013 oven of 1500 W; 250 ml Erlenmeyer Pyrex flasks; 50 ml flat-bottom round Pyrex flasks; and glass columns of 250 mm length × 10 mm internal diameter.

The reagents employed were Baker analytical reagents (n-hexane, acetone, silica gel 40–140, and anhydrous sodium sulfate). Celite (Química Meyer) and cotton (Protec) were industrial grade. The diesel used was from PEMEX (Mexican State Petroleum Company), and a Diesel/Lubricating Oil Standard (BAM 2000) was utilized. Hexane was chosen as the extractions solvent due to its lower toxicity, in contrast to other solvents, such as dichloromethane, which have been identified previously as ideal in mechanical shaking extractions (Schwab et al. 1999).

Cotton Treatment

Washed first with acetone, and then with hexane both in ultrasound bath for 25 min. Each solvent was decanted and the remnant evaporated in the hood after the washing procedure (EPA Method 3550 B 1996).

Silica Gel Treatment

Washed first with hexane in ultrasound bath for 25 min. Filtered, and washed again with fresh hexane, which was decanted. Left in the hood all night to ensure dryness. ‘Activation’ was performed at 150°C for at least 16 h (EPA Method 3630 B 1996).

Celite Treatment

Washed with distilled water in an ultrasound bath for 25 min, and filtered. Washed secondly with dicholoromethane for 25 min in ultrasound bath. Filtered and dried in hood all night. Heated in oven to 150°C for 4 h (Ricardo Alfaro, Chemistry Institute, UNAM, personal communication).

Soxhlet Extraction

Ten grams of soil samples were accurately weighed in duplicate in Whatman thimbles, and placed inside the glass tube of a Soxhlet apparatus with 180 ml hexane extractant. An empty thimble as blank was also processed. The extraction was performed for 8 h to ensure maximum recovery, after which the extract was concentrated to 40-ml total volume in the rotary evaporator and filtered through a cotton-packed column, washing with additional hexane to make a total volume of 60 ml. The procedure described in the general method was followed from this point onwards.

2.2 Preparation of Standard Soils Contaminated with TPH

Four non-contaminated soils were characterised (cf. Table 2) and artificially contaminated and aged in a stationary mode for 8 months, to be used in the final calibration of the method, especially pertaining optimization of the extraction procedure. Diesel was chosen as representative of relatively heavy TPH fractions, due to its homogeneous composition (well balanced between aliphatic and aromatic compounds in a C9 to C28 range; Schwab et al. 1999), and high boiling point (about 400°C).
Table 2

Main characteristics of non-contaminated soils used in the preparation of TPH-contaminated soil standards

Code

Soil descriptor

pH

Electric conductivity (mS/cm)

% Organic carbon (OC)a

F

Fertilized

7.7

0.730

1.72

S

Sodic saline

9.5

1.322

0.46

V

Vertisol

7.1

1.500

4.56

R

Red

7.2

1.224

0.57

Soils provided by courtesy of Dr. Arturo Aguirre Gómez, Department of Chemistry, FES Cuautitlán, UNAM.

aAnalysed in triplicate according to Aguirre et al. (2002)

For the artificial contamination (BAM 2002; Arce 2000), soils were sieved and dried as described above, and 100 g lots of each were weighed in round flasks of 1,000 ml. Mixtures of different amounts of diesel, accurately weighed, and dissolved in approximately 100 ml hexane were added to each lot, and homogenized by shaking for 6 h, followed by evaporation of the bulk hexane in the rotary evaporator. Soils were left overnight inside the hood (to achieve remnant hexane evaporation) covered with aluminium foil, and then kept stationary in the lab for 8 months. This provided conditions for diesel sorption to the solid minerals and natural organic matter (NOM) that were more representative of aged contamination scenarios, as compared to fresh contact. For each soil, five samples were contaminated with the diesel concentrations shown below, which represent a wide range of TPH soil contamination levels (from 0.1% to 10% in weight):

Soil number code

1

2

3

4

5

Applied diesel concentration (mg/kg)

1,018

2,135

20,103

100,110

0

In the following section the details of the procedures carried out and the parameters studied to calibrate the method are described in the order in which they were performed, providing proper sensitivity tests to these parameters.

2.3 Detection Limit (DL) and Quantification Limit (QL)

The DL definition adopted here is the one obtained from three standard deviations of the blank. In this particular case, because no standard soil with non-hexane-extractable NOM content was available that would serve as adequate blank, we limit this to a calculation of a theoretical DL, taken from the uncertainty of the analytical balance of 0.0005 g. The DL is thus 0.0015 g. If this figure comes from a 10 g soil sample processed by the method, the overall DL corresponds to 150 mg/kg. The quantification limit, defined as 10 times the uncertainty of the balance, i.e. 5 mg, translates to an overall QL of 500 mg/kg for a 10 g soil sample extraction. Analysis of detection limits will be discussed below for soils with hexane-extractable NOM.

2.4 Chromatographic Detection of Hexane-Extracted Soil NOM and TPH Standards

Gas chromatograms were generated from the soil NOM extracted by hexane, using GC-FID (Varian Star 3400 CX series) and compared to those generated under the same conditions for two TPH standards (commercial PEMEX Diesel used in the present work, and certified diesel/lubricating oil BAM standard mixture; BAM 2002), and for the extracts of an actual petroleum + oil − contaminated soil and several non-contaminated soils, including one adjacent to the contaminated one. The GC conditions applied for these experiments were: Column injection technique, DB-5 (30 m × 0.25 mm ID × 0.25 μm) column; nitrogen carrier gas (1 ml/min); flame ionization detector temperature, 320°C; furnace program, 70°C (isotherm, 0.5 min) to 125°C (rate, 8°C/min) to 245°C (rate, 6°C/min) to 300°C (rate, 4°C/min); total time, 49 min.

3 Results and Discussion

3.1 Evaporation of the Hexane Extractant

In gravimetric methods any mass that contributes to the final measured extract weight not originating from the analyte of interest generates a positive interference. These types of interferences are the most common ones in such methods when applied to measurements of heavy-fraction TPH, and it’s of utmost importance to minimize them. Principally, two steps in the procedure for evaporating the extractant potentially contribute to such positive interferences: (1) water sorbed to the glass of flasks placed on the rotary evaporator, and especially (2) insufficient evaporation of the solvent used, because of its high relative initial volume.

The mass contributed by (1) was found to be negligible for the glass flasks utilized, either from drying at 105°C flasks that were not previously oven-dried (error < 0.0007 g), or from previously oven-dried flasks at 105°C that were subjected to the conditions of the rotary evaporator procedure, simulating the solvent evaporation step, and placed in the desiccator for 2 h after drying externally with lint-free absorbent paper, (error < 0.0005 g).

Optimal conditions to achieve quantitative hexane evaporation were investigated in sets of duplicate flasks to which approximate amounts of 0.6 g diesel were accurately weighed in each, corresponding to hypothetical 60,000 mg/kg TPH extracted from 10 g of soil, and were mixed with 60 ml hexane. Two rotary evaporator time length regimes were investigated: (1) varying a first time step from 15 to 120 min while maintaining a second step constant at 10 min; and (2) applying three 10-min time-steps to achieve higher evaporation efficiencies.

Figure 1 shows the average results obtained for the set of experiments (1). Evaporation times above 60 min were necessary to achieve quantitative hexane evaporation, but losses of diesel were observed at higher times, producing negative errors.
https://static-content.springer.com/image/art%3A10.1007%2Fs11270-008-9704-1/MediaObjects/11270_2008_9704_Fig1_HTML.gif
Fig. 1

Weight of hexane as a function of time of the first step in the rotary evaporator procedure (programmed vacuum pressure = 352 mbar)

The practical application of this experiment to real samples has two limitations. One arises from the high sensitivity of the final evaporated mass to the evaporation time, in such a way that a very accurate time must be reproduced in every experiment to avoid the positive or negative interferences that are incurred around it; and two, and most important, is the long time required for complete evaporation, which elevates the energetic costs of the overall procedure.

In the set of experiments (2), other conditions were investigated to achieve the quantitative evaporation while reducing the total time in the rotary evaporator. After each 10-min time step the distilled hexane was removed from the receiving flask of the evaporator system to alleviate vapour saturation. Decreasing the programmed vacuum pressure from 352 mbar, recommended for hexane, down to 200 mbar, was found to achieve high evaporation yields (total “diesel recoveries” of 107–116%). However, quantitative evaporation was only accomplished after implementing an additional vacuum system in the final desiccation step (100.3% diesel recovery). The optimal conditions found for the applied vacuum were 30 min at 22.5 in Hg (=0.75 atm), but the procedure was not found to be too sensitive to the vacuum time, since increasing it to 60 min yielded only a 1.1% diesel loss (i.e., 98.9% diesel recovery).

Thus, an approximate 30 min total vacuum time is enough to achieve complete evaporation of hexane, but the procedure does not require a very strict control to adhere to this exact time. This makes it a very practical procedural step, while at the same time reducing rotary evaporator time costs to half of what was necessary in the previously found conditions [set of experiments (1)]. Also, it avoids the use of a drying oven as a final step (EPA Method 1664 A 1999; NMX-AA-134-SCFI-2006), which decreases energetic costs, and it optimizes the desiccation step by, simultaneously: cooling, desiccating external adhered water, and evaporating final hexane fractions. Finally, reproducibility was considerably improved with this final evaporation method [relative standard deviations (RSD) of 0.2–0.3%], as compared to all previous experiments (RSD of 2–7%).

3.2 Natural Organic Matter Co-extraction and Cleanup

The fraction and nature of natural organic matter (NOM) extracted from the standard soils by the hexane treatment was investigated. NOM is responsible for most of the sorption phenomena that petroleum hydrocarbons experience when equilibrating with soils presumably via a homogeneous distribution throughout the entire volume of NOM, in an analogous process to solvent partitioning (Essington 2004). Due to its low polarity, hexane is expected to extract a low fraction of this NOM, which contains a variety of polar functional groups (Essington 2004), however even low fractions in absolute terms may represent high relative gravimetric errors in the final determination of extracted TPH.

3.2.1 Quantification of NOM Extracted by Hexane

Quantification was performed on the clean soils described in Section 2.2, with NOM contents ranging from 0.5 to 4.6% organic carbon. The general extraction procedure was followed (cf. Section 2) at two different ultrasound power conditions (100 and 260 W), for a period of 25 min. The total hexane volume after extraction and filtration was evaporated according to the optimal conditions found in the previous section. Additionally, two soils from the Tabasco region in Southeast Mexico(courtesy of Dr. Rutilio Ortiz, LAFQA, Geography Institute, UNAM), with very high organic matter contents were included in the experiments: Soils T1 (pH 4.8, 40.1% organic carbon), and T2 (pH 3.6, 78.6% organic carbon). All analyses were performed in triplicate.

Table 3 shows that for soils with OC contents below 5%, extracted NOM was found below or barely at the detection limit of the method (150 mg/kg) (thus explaining the high relative errors found). From these experiments, the potential co-extraction of NOM from soils with hexane was not deemed interfering with the determination of contaminating TPH at a quantification limit of 500 mg/kg.
Table 3

Average concentrations (mg/kg) of NOM extracted by n-hexane from non-contaminated soils at two ultrasound power conditions, for 25 min

 

Soil

Ultrasound Power (W)

F

V

S

R

T1a

T2a

100

93

108

30

75

260

180

160

480

2,400

Relative SD (%)

30 and 32, respectively

34 and 33, respectively

33

23

aThese soil were obtained in a later phase of the experimental procedure, in which a higher ultrasound power was deemed necessary, therefore their extraction was performed only at the higher power.

However, in soils with very high NOM, extracted fractions are considerable and depend on the total NOM content. If no selective cleanup of these co-extracted fractions is performed, they will necessarily raise the effective TPH quantification limits on these soils. For example, for soil T1 the DL would be 1.44 g/kg and the QL 4.80 g/kg, while for T2 these would amount to 0.72% and 2.40% in mass, respectively (see DL and QL definitions in Section 2).

3.2.2 Extract Cleanup Using Silica-gel

Silica-gel, according to EPA methods, is utilized to eliminate interferences that are co-extracted with the analytes. It is not clear what the actual cleanup mechanism is since the hydroxyl groups at the silica gel surface tend to be deprotonated and thus negatively charged (Langmuir 1997) and would in principle show repulsion for predominantly negatively charged NOM functional groups (Essington 2004). Nonetheless, the use of silica-gel as a sorbent of interferent compounds is widespread in standard methods that quantify hydrocarbon contamination, and its efficacy will be investigated here.

According to results shown in the previous subsection, and given that NOM does not vary widely in composition across soils in this range (Essington 2004), this cleanup procedure would only be necessary for soils with total natural OC contents higher than 5%. The following experiments were designed to quantify the effectiveness and errors involved by introducing such a cleanup step in the overall gravimetric method.

Diesel Recovery After Eluting Through a Silica-Gel Packed Column

Accurately weighed 0.6 g of diesel were mixed with 35 ml of hexane. The mixture was eluted through a column packed with cotton, silica gel, and celite, according to the description in the general method, and subsequent hexane washings through the column were applied: one of 5 ml, and two of 10 ml (i.e., at the end, the extract contained a total of 60 ml hexane). The hexane was evaporated according to the optimal conditions established above. A 0.3% diesel loss was registered. This corresponds to a negligible concentration of 180 mg/kg loss from a hypothetical 10 g soil (i.e., barely above the detection limit of 150 mg/kg, and below the quantification limit of 500 mg/kg—cf. Section 2.3), and was deemed acceptable.

Efficiency of Silica Gel for Eliminating Co-Extracted Soil NOM

Accurately weighed 0.0025 g of NOM extracted from the Vertisol (V) soil in the experiment of the previous section were mixed with 0.0103 g of diesel (accurately weighed) in 35 ml of hexane. These proportions correspond to a hypothetical extraction of 10 g of soil containing 1 030 mg/kg TPH, from which 250 mg/kg NOM was co-extracted, which represents an amount intermediate from soils with total OC contents between 5% and 40%. The mixture was transferred through a column packed according to the general method, and washed with additional 25 ml of hexane. The total hexane was evaporated according to the optimal conditions found before.

A total mass recovery of 0.0120 g was registered from the total 0.0128 g added, i.e. a 0.0018 g loss. If we consider that this loss pertains exclusively to co-extracted soil NOM retained, it would correspond to (0.0008/0.0025 × 100=) 32% retention by the silica gel. This amount represents a very low retention efficiency, as was expected. To find out possible reasons for this low experimental efficiency of silica-gel as an adsorber of hexane-extracted soil NOM, a more theoretical investigative approach was followed by comparing the polar nature of this co-extracted fraction with that of TPH standards used, through gas chromatography with flame ionization detection (GC-FID).

Brief Chromatographic Investigation of the Nature of Hexane-Extracted Soil NOM

Figure 2 shows the corresponding gas chromatograms of soil extracts and standards. The non-contaminated soil extracts shows bands and peaks that appear in the 13 to 24 min time range. Figure 2a shows a chromatogram example of such extracts for soil F. This same pattern was confirmed on the extracts of other non-contaminated soil samples listed in Table 2 (chromatograms not shown). The bands and peaks observed overlap with those of the diesel (Fig. 2b) and the diesel/lubricating oil mixture standards (not shown), which appear in the same region. The nature of the compounds detected in this time range is of very low polarity or altogether non-polar, and the results suggest that at least a fraction of the extracted soil NOM is composed of compounds with similar chemical nature as the diesel compounds. This explains why these NOM compounds are co-extracted by hexane from soils with TPH, why a major fraction of co-extracted NOM is not retained by silica, and confirms their potential positive interference in the gravimetric determination of petroleum-contaminated soils, if high enough quantities are co-extracted (for example from soils with very high NOM).
https://static-content.springer.com/image/art%3A10.1007%2Fs11270-008-9704-1/MediaObjects/11270_2008_9704_Fig2_HTML.gif
Fig. 2

GC-FID chromatograms of hexane extracts from non-contaminated soils and of TPH standards [relative intensity (mV) vs. time (min)], non-contaminated F soil (a), PEMEX diesel standard (b)

From these results it was decided that silica gel cleanup was not efficient and could be avoided in the overall method. However, this poses limitations of the gravimetric method on the quantification limits for soils with very high NOM contents. For example, for soils with natural OC contents of 40%, in the absence of silica gel cleanup, the DL would be approximately one order of magnitude higher than the DL for more average soils with OC contents below 5%. For soils with near 80% OC, the DL would be approximately 50 times higher.

3.3 TPH Extraction Efficiency Using Ultrasound

Soxhlet extractions of artificially-contaminated standard soils was performed to serve as reference for the ultrasound extraction investigations, by computing the maximum extractable fractions in these. The procedure was performed for samples V1, V2, F1, and F2. After extraction, the optimized hexane evaporation procedure described in Section 3.1 [set of experiments (2)] was performed.

The results of these extractions (in duplicate) are reported in Table 4 after correction from the blank, but not corrected for possibly co-extracted NOM. Complete recovery of the diesel used as contaminant was obtained only for soil V1, the remaining soils showed recoveries close to 90%. Since no measurements, and thus no corrections, were made for co-extracted NOM, the real diesel extraction efficiencies are most probably somewhat lower than the values reported in Table 4.
Table 4

Recovery of diesel from artificially contaminated standard soils ‘V’ and ‘F’ by two different extraction procedures

Soil code

Contamination level (mg/kg)

% Recovery average

Relative difference (%)

Soxhlet extraction

 V1

1,018

103

3

 V2

2,135

87

2

 F1

1,018

90

3

 F2

2,135

89

2

Ultrasound extraction bath at 100 Watts for 25 min

SD (%)

 F1

1,018

72.1

0.2

 F2

2,135

68

1

 F3

20,103

85.7

0.3

 F4

100,112

90

3

These apparently incomplete recoveries may be a result of diesel losses during preparation of the contaminated standards, and/or presence of diesel fractions that have become non-extractable with aging. However, the reasons for incomplete recovery do not concern us here, since the Soxhlet method is not under investigation; these recovery figures will be used solely as references for the ultrasound extraction recoveries.

TPH extractions following the general method were performed using ultrasound baths at 100 W, and at 260 W, using soil F. The results for the 100-W bath are shown in Table 4 for the four levels of contamination, and for 25 min extractions in triplicate. Notice the low extraction efficiencies for F1 and F2, as compared to the Soxhlet extraction (Table 4). Figure 3 shows the results for both ultrasound powers, for soils F1 and F2, and as a function of extraction time, performed in triplicate. The results in this figure were normalized to those obtained for the Soxhlet method (Table 4), considering that the latter represent the maximum extractable amount of diesel in the conditions of the aged contaminated soils.
https://static-content.springer.com/image/art%3A10.1007%2Fs11270-008-9704-1/MediaObjects/11270_2008_9704_Fig3_HTML.gif
Fig. 3

Ultrasound extraction efficiencies of diesel-contaminated soils F and V as a function of time and ultrasound power, normalized to extraction data of an 8-h Soxhlet extraction

Figure 3 shows a considerable direct effect of the extraction time and ultrasound power on the total TPH extraction. However, an increase in the hexane extractant volume from 35 to 45 ml did not yield improved extraction efficiencies. Extraction times between 40 and 60 min were chosen as optimal; however, under these conditions, it must be noted that the recoveries obtained are somewhat lower by a range comprising 5–20% than those of the Soxhlet extraction.

In a final attempt to improve the ultrasound extraction efficiencies, the effect of sequential extractions was investigated, as suggested in standard methods. The above extraction method was repeated three consecutive times on the same flask containing the initial 10 g of soil sample (V1 and V2 only). After the first and second extractions, the major portion of the hexane + hydrocarbon extract was carefully decanted to the round-bottom flasks for evaporation, through a cotton-packed glass column, but ensuring that no soil particles were transferred to the columns. Additional 25 ml of hexane were added to clean the columns, making a total volume of 60 ml after each consecutive extraction, which were collected separately. After the third extraction, the whole soil was decanted to the column and both the original flask and the column were sequentially washed with 25 additional milliliters of hexane. The weights after evaporation of the three flasks for each sample were added together to compute total extraction efficiency.

No improvement was obtained upon three sequential extractions of the two soil samples, as compared to one single extraction, which means that the first extraction removed the whole ultrasound-extractable fraction of diesel sorbed to the soils, and the following two extractions did not contribute to any additional extracted analyte. This was not immediately obvious because of the nature of the experiment, where remnants of the first extraction contributed to the second and third extracted fractions. Thus, extraction efficiencies for these two soils ranged from 80 to 96% of the amounts extracted by the Soxhlet procedure using the high-power ultrasound bath (Fig. 3). These fractions may be regarded as representing those that may become potentially available and eventually hazardous to the biota and humans, in contrast to the non ultrasound-extractable fractions.

4 Conclusions

The procedure of the developed gravimetric method for TPH in soils in the present work may be summarized as follows: (1) Accurate weighing of approximately 10 g homogenized soil in round flasks, and addition of 10 g of anhydrous Na2SO4; (2) TPH extraction using 35 ml n-hexane in a 260-W ultrasound water bath for 60 min; (3) filtration of the extracts through a column packed with 0.6 g treated cotton or glass fibre, and down-washing with additional 25 ml hexane to complete 60 ml in the final liquid extract for analysis; (5) evaporation of the hexane in three 10-min steps in a rotary evaporator at 40°C, followed by quantitative evaporation of the remnant hexane in a desiccator under 22.5 in Hg (0.75 atm) vacuum for 30 min; and (6) weighing of the residue in an analytical balance.

The novel aspects of the method presented relate to cost- and time-saving steps. These are reflected in the extraction procedure by using a relatively low volume of hexane extractant and an ultrasound bath for 1 h, instead of the 4 to 8-h Soxhlet extraction recommended in standard methods; also, the overall use of a heating oven was minimized, since evaporation of the solvent (and of possible water adsorbed to the flask wall) is completed using vacuum (inside a desiccator). Method expediency was achieved by eliminating a silica-gel cleanup procedure, which was justified by independent experiments that showed that silica-gel does not selectively retain co-extracted natural organic matter. Insufficient solvent evaporation was identified as one of the most important positive interferences in the gravimetric detection in general, and it is highly recommended to carefully calibrate this step in any method modification, especially as a function of type and total extractant volume. Quantitative hexane evaporation via the desiccation under vacuum step investigated was not found to be highly sensitive to time and thus allowed for method convenience. However, the method was found to be positively sensitive to the extraction time and power of the ultrasound bath used, and these parameters should also be carefully calibrated in any method modification.

Acknowledgements

The authors are grateful to: Dr. Arturo Aguirre Gómez from the Department of Chemistry at FES-Cuautitlán, UNAM for supplying the soils of study, for the pH and electric conductivity data for these, and for his assistance in the organic carbon determinations; Dr. Rutilio Ortiz, Geography Institute, UNAM, for supplying the high organic matter content soils, and for the pH and organic carbon data for these; Rosaura Paez, Geography Institute, UNAM, Morelia Unit, for help in acquiring the gas chromatograms; Ricardo Alfaro, Chemistry Institute, UNAM, for his assistance in the initial methods development; and two anonymous reviewers who helped improve considerably the legibility of the manuscript.

Copyright information

© Springer Science+Business Media B.V. 2008