Determination of Alkylphenols after Derivatization to

resulting esters are analyzed by gas chromatography with atomic emission detection (AED) in the iron-selective detection mode. This method utilizes th...
5 downloads 0 Views 137KB Size
Download PDF
Anal. Chem. 2001, 73, 3073-3082

Determination of Alkylphenols after Derivatization to Ferrocenecarboxylic Acid Esters with Gas Chromatography-Atomic Emission Detection Juergen Rolfes and Jan T. Andersson*

Department of Analytical Chemistry, Institute of Inorganic Chemistry, University of Muenster, Wilhelm Klemm Strasse 8, D-48149 Muenster, Germany

A method is described for the rapid determination of alkylphenols in nonpolar matrixes. The alkylphenols are derivatized with ferrocenecarboxylic acid chloride so that every phenol molecule is labeled with one iron atom. The resulting esters are analyzed by gas chromatography with atomic emission detection (AED) in the iron-selective detection mode. This method utilizes the AED’s low detection limit (0.05 pg/s) for iron and the high selectivity versus carbon (3.5 × 106) for the detection of the alkylphenols. Because the derivatization is performed before the first step of sample preparation, the risk of analyte loss by adsorption or volatilization is minimized. The total recoveries in the lower ppm concentration range vary between 79 and 125%. The quantification of 20 C0C3-alkylphenols in crude oils is demonstrated by analyzing a shale oil (SRM 1580) and a petroleum crude oil (SRM 1582). The complete workup is easily carried out in only 45 min/sample. Alkylphenols have various anthropogenic and natural sources and therefore occur in many different matrixes. Combustion of coal and oil leads to their release into the atmosphere and subsequently into water, soil, flora, and fauna.1 Cigarette smoke is the most important reason for phenol and cresol exposure to humans.2 Nonylphenols from nonionic surfactants are endocrinedisruptive compounds that can be found ubiquitously.3 Some alkylphenols are used as antioxidants in food.4 Beside that, coffee, tea, and wine naturally contain phenolic compounds.5 In urine, phenols occur as metabolic products of aromatic compounds and can be used for monitoring of the exposition to nonphenolic aromatic compounds.6 Because of their toxic potential several alkylphenols are regulated by the U.S. Environmental Protection Agency (EPA).7 * Corresponding author. Phone: +49-251-83 33141, Fax: +49-251-83 36013. E-mail: [email protected]. (1) Belloli, R.; Barletta, B.; Bolzacchini, E.; Meinardi, S.; Orlandi, M.; Rindone, B. J Chromatogr., A 1999, 846, 277-281. (2) Nanni, E. J.; Lovette, M. E.; Hicks, R. D.; Fowler, K. W.; Borgerding, M. F. J. Chromatogr. 1990, 505, 365-374. (3) Sonnenschein, C.; Soto, A. M. J. Steroid. Biochem. Mol. Biol. 1998, 65, 143-150. (4) Williams, G. M.; Iatropoulos, M. J.; Whysner, J. Food Chem. Toxicol. 1999, 37, 1027-1038. (5) Huang, M. T.; Ferraro, T. ACS Sym. Ser. 1992, No. 507, 8-34. (6) Bieniek, G. Int. Arch. Occup. Environ. Health 1997, 70, 334-340. 10.1021/ac001540h CCC: $20.00 Published on Web 05/12/2001

© 2001 American Chemical Society

Alkylphenols influence both the refining and the stability behavior of fuels. For example, insoluble rubbers formed from phenolic compounds can damage fuel lines and engines.8 In organic geochemistry, alkylphenols obtain increasing attention. Interactions with the polar surface of sediments during the migration processes are a consequence of their polarity with the result that they are partitioned between the crude oil and the water-saturated pores of the sediment rocks.9,10 The geochemical conclusions that can be drawn from the isomer distribution of those compounds are thus complementary to those drawn from the analysis of aliphatic and aromatic compounds. Recent experimental11 and numerical12 simulations have improved our understanding of such processes but require the determination of alkylphenols in crude oils for the interpretation of migration history and processes.13,14 There are several methods for the analysis of low molecular weight (C0-C3) alkylphenols in organic matrixes.10,15-25 For (7) Federal Register, EPA Method 604, Phenols, Part VIII, 40 CFR Part 136, Environmental Protection Agency, Washington, DC, current version from 7 to 1-99; URL: http://www.epa.gov/epacfr40/chapt-I.info/subch-D/ 40P0136.pdf (2000-11-09). (8) Power, A. J.; Mathys, G. I. Fuel 1992, 71, 903-908. (9) Taylor, P. Controls on the Occurrence of Phenols in Petroleums and Waters. Ph.D. Thesis, University of Newcastle upon Tyne, England, 1994. (10) Taylor, P.; Larter, S.; Jones, M.; Dale, J.; Horstad, I. Geochim. Cosmochim. Acta 1997, 61, 1899-1910. (11) Bennett, B.; Larter, S. R. Geochim. Cosmochim. Acta 1997, 61, 4393-4402. (12) van Duin, A. C. T.; Larter, S. R. Org. Geochem. 1998, 29, 1043-1050. (13) Larter, S. R.; Bowler, B. F. J.; Li. M.; Chen, M.; Brincat, D.; Bennett, B.; Noke, K.; Donohoe, P.; Simmons, D.; Kohnen, M.; Allan, J.; Telnaes, N.; Horstad, I. Nature 1996, 383, 593-597. (14) Larter, S. R.; Aplin, A. C. Reservoir Geochemistry: Methods, Applications and Opportunities. In The Geochemistry of Reservoirs; Cubitt, J. M., England, W. A., Eds. Geological Society Special Publication 86; Geological Society: London, 1995; pp 5-32. (15) Ioppolo, M.; Alexander, R.; Kagi, R. I. Org. Geochem. 1992, 18, 603-609. (16) Ioppolo-Armanios, M.; Alexander, R.; Kagi, R. I. Org. Geochem. 1993, 22, 815-823. (17) Hertz, H. S.; Brown, J. M.; Chesler, S. N.; Guenther, F. R.; Hilpert, L. R.; May, W. E.; Parris, R. M.; Wise, S. A. Anal. Chem. 1980, 52, 1650-1657. (18) MacCrehan, W. A.; Brown-Thomas, J. M. Anal. Chem. 1987, 59, 477-479. (19) Willsch, H.; Clegg, H.; Horsfield, B.; Radke, M.; Wilkes, H. Anal. Chem. 1997, 69, 4203-4209. (20) Rovere, C. E.; Crisp, P. T.; Ellis, J.; Korth, J. Fuel 1990, 69, 1099-1104. (21) Harvey, T. G.; Matheson, T. W.; Pratt, K. C. Anal. Chem. 1984, 56, 12771281. (22) Bennett, B.; Bowler, B. F. J.; Larter, S. R. Anal. Chem. 1996, 68, 36973702. (23) Green, J. B.; Yu, S. K.-T.; Vrana, R. P. J. High Resolut. Chromatogr. 1994, 17, 439-451.

Analytical Chemistry, Vol. 73, No. 13, July 1, 2001 3073

Figure 1. Esterification of alkylphenols to the corresponding esters with ferrocenecarboxylic acid chloride (FCC).

speciation of individual alkylphenols, the isolation of the phenol fraction is usually the first step, followed by high-performance liquid chromatography (HPLC) or gas chromatography (GC). Most isolation schemes employ a combination of aqueous alkaline extraction of the sample and, after acidification, back extraction into an organic solvent to prepare an extract suitable for GC or HPLC analysis.10,15-18 Some methods involve fractionation of the sample into compound classes by adsorption column chromatography,19,20 thin-layer chromatography,21 or solid-phase extraction (SPE)22 while others use anion chromatography to isolate the acidic fraction of the oil, followed by ion-pair chromatography to separate phenolic from other acidic compounds.23 Determination of the individual alkylphenols with HPLC and electrochemical detection is easy to perform, but the chromatographic resolution is poor so that only C0- and C1-alkylphenols can be analyzed without coelution.22 An acceptable resolution of the C2- and C3alkylphenols can only be achieved through the high separating power of GC techniques, but then a derivatization of the isolated alkylphenol fraction is usually required. The analytes can be transformed into trimethylsilyl ethers,22 acetates,24 or trifluoroacetates,23 followed by GC analysis with flame ionization detection or mass spectrometry (GC/MS). Despite many different methods published, alkylphenol analysis is still a challenging task and is therefore far from being a routine parameter in, for example, crude oil analysis. In this work, we describe a method that allows alkylphenols to be determined in a very sensitive and selective way after a rapid and simple workup that has the potential to be automatized for large sample numbers. This goal is achieved through labeling of the alkylphenols with an iron atom using ferrocenecarboxylic acid chloride (FCC) as shown in Figure 1. The iron atom in the ferrocene group can be detected very sensitively (0.05 pg/s) and selectively (3.5 × 106 versus carbon) in the iron-selective detection mode of the atomic emission detector (GC-AED).26-28 The method will be illustrated here using crude oils as samples but should be easily adaptable to other nonaqueous sample matrixes as well. EXPERIMENTAL SECTION Reagents. All alkylphenols, aliphatic alcohols, and fluoro alcohols were from commercial suppliers (Aldrich, Fluka, Merck, (24) Llompart, M. P.; Lorenzo, R. A.; Cela, R.; Pare´, J. R. J.; Be´langer, J. M. R.; Li, K. J. Chromatogr., A 1997, 757, 153-164. (25) Ioppolo-Armanios, I. The Occurrence and Origins of some Alkylphenols in Crude Oils. Ph.D. Thesis, Curtin University of Technology, Perth, Australia, 1995. (26) Quimby, B. D.; Sullivan, J. J. Anal. Chem. 1990, 62, 1027-1034. (27) Sullivan, J. J.; Quimby, B. D. Anal. Chem. 1990, 62, 1034-1043. (28) Uden, P. C. Element-Selective Gas Chromatographic Detection by Atomic Plasma Emission Spectroscopy. In Selective Detectors: Environmental, Industrial, and Biomedical Applications, 1st ed.; Sievers, R. E., Ed.; John Wiley & Sons: New York, 1995; pp 143-169.

3074

Analytical Chemistry, Vol. 73, No. 13, July 1, 2001

Lancaster, Riedel de Haen), and purities were usually higher than 97%. Ferrocene (98%) and 4-(dimethylamino)pyridine (>99%) were purchased from Aldrich, and oxalyl chloride (97%) was from Merck. Alumina (chromatography grade, 90 mesh, neutral, Fluka) was activated at 450 °C for 12 h before use. Benzene, toluene, pentane, and dichloromethane (all pesticide grade, Fluka) were purified by percolation through alumina. Silanized glass wool (pesticide grade) was purchased from Supelco. Preparation of Ferrocenecarboxylic Acid Chloride. The synthesis of FCC has been described.29 GC analysis showed detectable amounts only of three impurities: ferrocene (0.04%) and the ferrocenecarboxylic acid esters of methanol (0.02%) and ethanol (0.04%). Preparation of Reference Compounds. Twenty-two C0-C3alkylphenols were commercially available (Figure 2). Two C4alkylphenols, thymol and carvacrol, were also investigated. The ferrocenecarboxylic acid esters of 3,3,3-trifluoroethanol (F3EE) and 1,1,3,3-tetrafluoro-2-propanol (F4PE) were selected as quantification standards.29 Esters of all these phenols were synthesized according to the following procedure: In a 5-mL sample vial, 0.150 mmol (about 15-20 mg) of the alkylphenol was mixed with a solution of 0.160 mmol (40.9 mg) of FCC in 0.82 mL of dichloromethane and a solution of 0.221 mmol (27.4 mg) of 4-(dimethylamino)pyridine (DMAP) in 0.55 mL of dichloromethane. After 1 min, the DMAP and the excess of FCC were removed by separation on a minicolumn (50 mm × 5 mm i.d. of alumina). The esters were eluted with 5 mL of dichloromethane. The minicolumn was prepared from a Pasteur pipet by inserting a piece of glass wool into the pipet and then heating it in a lighter flame and extruding its tip to form a small capillary. For purification, the solvent was evaporated and the residue was fractionated on a second column (70 mm × 8 mm i.d. of alumina), after being dissolved in 200 µL of a mixture of dichloromethane/pentane (30:70 v/v, termed “30% DCM”). The nonpolar impurities were eluted with 4 mL of 30% DCM and the products with 1.5 mL of dichloromethane. The solvent was evaporated in a gentle stream of nitrogen at 50 °C. The purity of all esters was at least 98% and was checked by GCFID. For confirmation of the identity, the esters were analyzed by GC/MS. Instrumentation. A Hewlett-Packard (now Agilent, Avondale, PA) HP 5890 II GC with a split/splitless injector and an CTC A200 SE autosampler (Chromtech, Idstein, Germany) coupled to a HP 5921A AED was used. The microwave oven was a household Samsung Compact RE 2000 (Samsung Electronics, Schwalbach, Germany) with 500 W of power. GC Conditions. A DB-5-ms column (30 m × 250 µm × 0.25 µm, J+W Scientific Inc., Folsom, CA), equipped with a deactivated retention gap (3 m × 320 µm, J+W Scientific), was used. The injector was equipped with a Merlin high-pressure microseal injector (Alltech, Unterhaching, Germany) and a packed singletapered insert with 0.9-mL liner volume (Hewlett-Packard). Injector and detector block temperatures were set to 270 and 300 °C, respectively. Helium carrier gas was set to 172 kPa constant pressure, resulting in a linear velocity of 45 cm/s at 80 °C. Samples were injected with a 5-µL syringe (SGE, Darmstadt, Germany) in the split/splitless mode, with a splitless time of 66 s. (29) Rolfes, J.; Andersson, J. T. Anal. Commun. 1996, 33, 429-432.

Figure 2. Isomers of the 22 C0-C3-alkylphenols, thymol, and carvacrol analyzed in this work. Table 1. Temperature Programs Used for the Separation of the Alkylphenol Esters program I program II program III

80 °C for 1.5 min, 5 °C/min to 300 °C, hold for 15 min 80 °C for 1.5 min, 45 °C/min to 250 °C, hold for 1 min, 4 °C/min to 270 °C, hold for 0 min, 10 °C/min to 300 °C, hold for 15 min 80 °C for 1.5 min, 20 °C/min to 220 °C, hold for 0 min, 4 °C/min to 270 °C, hold for 0 min, 10 °C/min to 300 °C, hold for 15 min

Detector Optimization. A wavelength of 302 nm was used for iron-selective detection. The transfer line and the cavity block were hold at 300 °C. Helium supply pressure and cavity pressure were set at 200 (30 psi) and 10 kPa (1.5 psi), respectively To optimize the AED makeup gas flow, 1 µL of a solution of ferrocene in toluene (20 ng/µL) was injected at effective makeup flow rates between 45 and 215 mL/min (measured at the cavity vent, with window purge “off” and ferrule purge vent uncapped). For optimization, the prepressures of the scavenger gases hydrogen and oxygen were varied in the range from 100 to 520 kPa. Calibration and Detection Limit of the GC-AED System. A stock solution of the ferrocenecarboxylic acid ester of o-cresol in toluene with an iron concentration of 10.0 ng/µL (equal to a concentration of 57.3 ng/µL e ester) was diluted to concentrations in the range from 0.1 to 150 pg/µL iron. An volume of 2 µL was injected 5-fold from each solution. Optimization of the Temperature Program. To evaluate the chromatographic behavior of the ferrocenecarboxylic acid esters of the 24 alkylphenols, 2 µL of a standard mixture with all compounds (each about 10-30 pg/µL, calculated as iron) was used. The temperature programs used are listed in Table 1. General Derivatization Procedure. Figure 3 shows a flow diagram of the analytical procedure adopted for the derivatization of standard solutions and crude oil samples. In a 2-mL screwcapped sample vial, typically 25 mg of the crude oil was spiked with 50 µL of a mixture of F3EE and F4PE in toluene (1.500 ng/

Figure 3. Flowchart of the sample preparation process.

µL, calculated as iron) as internal standard. To this mixture a solution of 200 µL of FCC (21.1 mg/mL, 0.085 mol/L) and 300 µL of DMAP (20.1 mg/mL, 0.165 mol/L) in dichloromethane was added and the vial was sealed with two Teflon septa as tightly as possible. The vessel was heated twice for 30 s in a household microwave oven, with a 60-s break in between. The vials were allowed to cool, and after a few minutes, the reaction mixture was transferred by a Pasteur pipet to a minicolumn (23 mm × 5 mm i.d. of alumina). The resulting ferrocenecarboxylic acid esters of the alkylphenols were eluted with 4.5 mL of dichloromethane, while the DMAP, the excess FCC, and the more polar compounds of the crude oil remain on the column. After evaporation of the dichloromethane, the residue was redissolved in 0.2 mL of 30% DCM, transferred to a second minicolumn (23 mm × 5 mm i.d. of alumina), and the hydrocarbons from the crude oil matrix were eluted with 3.0 mL of 30% DCM. Subsequently, the alkylphenol esters were eluted with 4.5 mL of 100% DCM. To the eluate was added 0.3 mL of toluene as keeper and the dichloromethane was Analytical Chemistry, Vol. 73, No. 13, July 1, 2001

3075

Figure 4. Dependence of the peak area of the iron-selective response on the makeup gas flow. Injection of 1.0 µL of ferrocene standard solution (20 ng/µL, in toluene). Scavenger gases: O2 at 350 kPa, H2 at 480 kPa.

removed. Finally, the sample was diluted to 1.5 mL with toluene for GC analysis. Reproducibility and Recovery. The recovery of the alkylphenols throughout the derivatization procedure was tested for pure standard solutions and for a spiked crude oil containing no detectable amounts of alkylphenols. For this purpose, 25 mg of tetradecane or 25 mg of the crude oil was spiked with 100 µL of a mixture of 14 alkylphenols (each phenol 14.3 pmol/µL, equivalent to 1.34-1.95 ng/µL) and 50 µL of a solution of the internal standards F3EE and F4PE (each 26.8 pmol/µL, 1.5 pg/µL calculated as iron). Both samples were worked up and analyzed twice as described above. Derivatization of Oil Samples. Two standard reference materials (SRM) from the National Institute of Standards and Technology (NIST), in Gaithersburg, MD, were analyzed as described above. One is a shale oil (SRM 1580) and originates from the Mahogani Zone of the Colorado Green River Formation.17,30 The other sample (SRM 1582) is a Californian crude oil from the Wilmington Basin.18,31 Because the concentrations in the shale oil are much higher than in normal crude oils, the amount of quantification standards added to the sample had to be matched. Therefore, only 5 mg of the shale oil was spiked with 50 µL of a toluene solution of F3EE and F4PE (15.0 ng/µL, calculated as iron), and the resulting solution was diluted 10-fold before injection. RESULTS AND DISCUSSION This section will be divided into discussions on the detection characteristics, the most favorable separation conditions, and method validation including the analysis of the reference materials. Optimization of the Detector. The sensitivity of the AED in the iron-selective detection mode depends strongly on the gas flow parameters. Figure 4 shows a plot of the peak area versus the makeup gas flow rate (helium) for the injection of 20 ng of ferrocene at makeup gas flow rates between 45 and 220 mL/min. (30) Certificate of Analysis SRM 1580: Organics in Shale Oil; National Institute of Standards and Technology (NIST), Gaithersburg, MD, 1980 (URL: http:// ois.nist.gov/srmcatalog (2000-10-04)). (31) Certificate of Analysis SRM 1582: Petroleum Crude Oil; National Institute of Standards and Technology (NIST), Gaithersburg, MD, 1984 (URL:.http:// ois.nist.gov/srmcatalog (2000-10-04)).

3076 Analytical Chemistry, Vol. 73, No. 13, July 1, 2001

Figure 5. Dependence of the peak area in the iron-selective detection mode on the gauge pressure of the scavenger gases. Injection of 1.5 µL of standard solution of the ferrocenecarboxylic acid ester of o-cresol (2 ng/µL, in toluene). Helium makeup gas flow was 170 mL/min.

Figure 6. Calibration curve of the signal from the ferrocenecarboxylic acid ester of o-cresol in the iron-selective detection mode. The chromatogram shows the ester of o-cresol at the detection limit, corresponding to an injected amount of 0.2 pg of iron.

The response of the AED in this mode shows an optimum at 170 mL/min, so this flow was chosen for all further injections. In a similar way, the scavenger gases were optimized. Figure 5 displays the plot of the peak area of 3 ng of the ferrocenecarboxylic acid ester of o-cresol versus the scavenger gas pressure. The peak area of the o-cresol ester has an optimum at an oxygen pressure of 150 kPa. For hydrogen, the sensitivity increases up to the maximum pressure of 520 kPa. These pressures were used for all further measurements, including the calibration curve of the GC-AED system for the injection of 0.2-300 pg of Fe as the ferrocenecarboxylic acid ester of o-cresol, depicted in Figure 6. The detection limit is defined as the injection amount that results in a peak with a signal-to-noise ratio (S/N) ) 2 (long-term noise, measured over a period of 90 s). This is the case for the injection of 0.2 pg of Fe. For subsequent measurements with a second-generation Agilent G2350 GC-AED system, we found a detection limit of less than 20 fg of iron.32 The deviation from linearity becomes appreciable when more than (32) Rolfes, J. Bestimmung von Alkylphenolen in Rohoelen. Ph.D. Thesis, University of Muenster, Germany, 1995 (URL: http://www.uni-muenster.de/ Chemie/AC/anders/rolf/publications (2000-10-04)).

Table 2. Recovery and Repeatability of the Alkylphenols after Derivatization of Spiked Tetradecane Solutions and Spiked Crude Oilsa spiked tetradecane

spiked crude oil

compound

recovery (%)

RSD (%) (n ) 2)

recovery (%)

RSD (%) (n ) 2)

Phen o-Cres m-Cres p-Cres 2-EP 26-DMP 3-EP 35-DMP 2-n-ProP 3-i-ProP 34-DMP 4-i-ProP 236-TMP 235-TMP

97 107 99 106 88 94 93 95 98 92 79 89 75 77

1.4 0.0 3.9 1.9 2.3 2.3 0.2 1.7 0.1 0.1 0.9 1.4 1.1 0.8

110 121 107 122 101 116 103 107 110 102 99 107 108 107

1.1 1.1 3.6 2.1 3.5 1.8 1.6 2.9 3.0 1.0 1.9 1.9 3.8 2.9

average

92

1.3

109

2.3

a The recoveries were calculated according to eq 1 and refer to the whole derivatization and workup procedure.

Figure 7. Separation of the ferrocenecarboxylic acid esters of 24 selected alkylphenols at temperature program (a) I, (b) II, and (c) III. Peak numbers correspond to alkylphenol numbers in Table 3.

125 pg of iron is injected. For injected amounts of up to 125 pg of iron, the calibration function (linear regression) has a regression coefficient of R2 ) 0.9994. The standard deviation of the peak area for replicate injections (n ) 5) is lower than 3% between 2 and 300 pg of iron and increases to values of 16% at injection amounts of 0.2 pg of iron. Optimization of the Temperature Program. The ironselective chromatogram of the standard mixture of 24 alkylphenol esters with a standard temperature ramp of 5 °C/min is depicted in Figure 7a. With this program, 15 alkylphenol esters can be separated (resolution R g 1.5) in 41 min. m- and p-cresol cannot be separated without derivatization on nonpolar GC phases,15 but their ferrocenecarboxylic acid esters show excellent separation. Three ester peaks overlap (0.5 < R < 1.5) and three pairs of esters coelute (R e 0.5). It should be noted that the peaks of the alkylphenol esters do not elute as groups separated strictly according to their alkyl carbon number. For example, the esterified 2-isopropylphenol coelutes with the ester of 2,6-dimethylphenol, after the ester of 2-ethylphenol but ahead of all other C2-alkylphenol esters. With an optimized program (program II), utilizing a very steep initial temperature ramp of 45 °C/min, it is possible to separate 14 esters in only 14 min. Six compounds overlap and two pairs coelute. Finally, with a third temperature program (III) optimized to separate the two coeluting peak pairs from program II, 14 esters can be separated in 22 min. Now eight peaks overlap and only

one pair coelutes, but one different from that in program II. The chromatograms are depicted in Figure 7b and c. To permit the quantification of such peaks that overlap under all our GC conditions, the chromatograms were processed with a mathematical peak algorithm, the commercially available software Pro Analysi::Peaks.33 With this, the overlapping peaks can be deconvoluted and integrated. The improvements in resolution reached by the alternate temperature programs might be small but are essential for the deconvolution process and the successful quantification. The details will be described in a subsequent paper. A combination of temperature programs II and III and utilization of the peak algorithm allow all 24 alkylphenols in Figure 2 to be quantified. METHOD VALIDATION Reproducibility and Recovery. Recoveries for the derivatization of standard solutions and spiked crude oils were calculated according to eq 1 and are listed in Table 2.

recovery (%) )

A(AP)c(IS)M(AP) A(IS)c(AP)M(IS)

× 100

(1)

Here, A(AP) and A(IS) are the peak areas of the esters of the alkylphenol and the internal standard, c(AP) and c(IS) are the concentrations of the alkylphenol and the internal standard, and M(AP) and M(IS) are the molar masses of the alkylphenol and of the standard, respectively. Actually, the use of only one internal standard is sufficient, and the second one is employed for control purposes only. The recovery of individual alkylphenols from esterified standard solutions was between 79 and 108%. For spiked and esterified crude oils, the recovery varied between 99 and 125%. The average standard deviation is 1.2% for the spiked tetradecane and 2.5% for the spiked crude oil (n ) 2). (33) Software Pro Analysi::Peaks, current version 2.2, Joerg Baumgarten, D-51674 Wiehl (URL: http://www.proanalysi.de (2000-10-04)).

Analytical Chemistry, Vol. 73, No. 13, July 1, 2001

3077

Figure 8. Iron-selective chromatogram of the esterified shale oil SRM 1580 with temperature program II. The inset shows an expanded view of the retention region between 6.0 and 7.5 min.

Analysis of a Certified Standard Reference Material. To demonstrate that the method works with a real sample, we derivatized a shale oil SRM 1580 from the Colorado Green River Formation. This is the only Standard Reference Material from NIST with certified concentrations for three alkylphenols.17,30 Figure 8 gives an overview over all alkylphenols of the chromatogram of the derivatized shale oil, analyzed with temperature program II. In the early part of the chromatogram (see inset in Figure 8) appears a group of four peaks, consisting of the two internal standards F3EE and F4PE and of the ferrocenecarboxylic acid esters of methanol and ethanol. The last two probably result from impurities of the derivatization reagents. In the later part of the chromatogram, the esters of higher-alkylated phenols elute as an unresolved complex mixture. For the quantification of the individual alkylphenols, the chromatograms of the sample were processed with the peak algorithm. Figure 9 shows only that part of the chromatograms where the C0-C3-alkylphenol esters elute. The deconvolution resulted in ∼45 single peaks for each chromatogram. The peaks of the C0-C3-alkylphenols identified after deconvolution are printed as gray lines above the original chromatograms. For clarity, the unidentified peaks are omitted. The peak algorithm enables even very small overlapping peaks to be quantified. For instance, although the ester of 2-isopropylphenol is 30 times smaller than that of 2,6-dimethylphenol and their resolution is only R ) 0.75 (Figure 9), it is possible to integrate the deconvoluted peaks separately. However, for some alkylphenol esters, the determination is complicated by unidentified coeluting compounds from the sample, particularly for some C3-phenol esters. The strategy to achieve optimal results consists of independently quantifying the alkylphenols with both temperature programs. The data from the two chromatograms, obtained at different temperature programs, are averaged provided the peak areas are within 10% of each other. For higher deviations the lower concentration is assumed to be a better representation of the true value. Through this approach, it is possible to discover further unidentified and hidden peaks. For example, the peak of 2,4,6-trimethylphenol in the chromatogram analyzed with temperature program II coelutes with a hidden peak. Under the conditions of program III, there is still considerable overlap, but the deconvolution makes it possible to integrate the peaks separately. 3078

Analytical Chemistry, Vol. 73, No. 13, July 1, 2001

For 3-isopropylphenol, the situation is more complex. In program II, the ester of this compound coelutes with the ester of 4-ethylphenol. Although the two peaks can be separated with temperature program III, the sum of their concentrations exceeds the area of the coeluting peaks from program III by 15%, indicating that the ester of 3-isopropylphenol coelutes with a further unidentified peak which probably was hidden in program II by the ester of 2,3-dimethylphenol. This could be verified experimentally (shown elsewhere32) by analysis of the sample on a different stationary phase. On this column, with a slightly different selectivity, a fourth peak appears, separated from the other three. The limit of the method can be demonstrated for the quantification of carvacrol. In both temperature programs, the peak of the carvacrol ester is overlaid by a large unidentified peak that could not be deconvoluted by the peak algorithm. However, it can be roughly estimated that the concentration of the carvacrol ester can be no more than 10% of that of the overlapping peak. The situation is similar for the peaks eluting after the carvacrol ester. Even if the retention times of the deconvoluted peaks fit perfectly with that from the reference compounds, it cannot be excluded that they coelute with other unidentified alkylphenol esters. The concentrations of the alkylphenols were calculated according to eq 2. Here, A(AP) and A(IS) are the peak areas of the

concentration (ppm) )

A(AP)M(AP)c(IS)V(IS) A(IS)M(IS)m(oil)

× 106

(2)

esters of the alkylphenol and the internal standard, M(AP) and M(IS) are the molar masses of the alkylphenol and of the quantification standard, c(IS) and V(IS) are the concentration and the volume of the internal standard solution added to the crude oil, and m(oil) is the mass of the crude oil sample analyzed. The quantitative results are summarized in Table 3. Overall, it was possible to quantify 21 of the 24 alkylphenols monitored in this sample. The concentrations of individual alkylphenols in the shale oil vary from 5 to 660 ppm, considerably higher than for crude oils. It is noticeable that this oil contains large amounts of methylphenols, only moderate concentrations of ethyl- and n-propylphenols and only minor concentrations of alkylphenols with branched alkyl substituents such as isopropyl groups.

Figure 9. Iron-selective chromatograms of the esterified shale oil SRM 1580, displaying the ferrocenecarboxylic acid esters of the C0-C3alkylphenols at temperature program (a) II and (b) III.. Peak numbers correspond to alkylphenol numbers in Table 3. The upper chromatograms in both (a) and (b) were deconvoluted using the peak algorithm.

These results can be compared with the concentrations determined at NIST, where independent sample preparation procedures were used. The first method involves fractionation of the whole shale oil via HPLC and determination of the underivatized alkylphenol fraction by analytical HPLC-UV, GC-FID, and GC/MS. The second approach involves an alkaline extraction of the crude oil, followed by back extraction and analysis of the sample by GC/MS. Only the concentrations of phenol, o-cresol, and 2,6-dimethylphenol were certified; others are given solely for information purposes because they were obtained using only one method. For phenol, o-cresol, and 2,6-dimethylphenol, the concentrations determined by the present method differ by up to 17% from the certified values, but they all agree within their uncertainties. For the noncertified compounds m-cresol, p-cresol, 2,4dimethylphenol, 2,4,6-trimethylphenol, and 2,3,6-trimethylphenol, the deviations are below 25%, and for 2,5-dimethylphenol, the concentration is 38% below the value from NIST.

Two alternative procedures are reported in the literature for the determination of the alkylphenols in this shale oil. The first one uses the liquid chromatographic sample workup of NIST and analysis by HPLC with electrochemical detection (HPLC-ED). The second one utilizes a derivatization of the alkylphenols with trifluoroacetic anhydride (TFAA), followed by determination with GC/MS. For the latter method, the authors utilized not only the direct derivatization of the sample but also tried the alkaline extraction procedure of NIST. The samples were analyzed only once, and there are significant differences between both procedures, especially for some of the dimethylphenols. 2-Ethylphenol and 2,6-dimethylphenol were not found at all by this method. Altogether, 13 alkylphenols were identified by this method. For several compounds, the discrepancies between these and our results are larger than 100%. All results published for this shale oil are summarized in Table 3. Analytical Chemistry, Vol. 73, No. 13, July 1, 2001

3079

Table 3. Alkylphenol Concentration in Shale Oil SRM 1580a derivn with TFAA15

derivatization with FCC peak

alkylphenol

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

phenol o-cresol m-cresol p-cresol 2-ethylphenol 2-isopropylphenol 2,6-dimethylphenol 2,5-dimethylphenol 3-ethylphenol 3,5-dimethylphenol 2,4-dimethylphenol 2-n-propylphenol 4-ethylphenol 3-isopropylphenol 2,3-dimethylphenol thymol 3,4-dimethylphenol 2,4,6-trimethylphenol 4-isopropylphenol carvacrol 2,3,6-trimethylphenol 2,3,5-trimethylphenol 4-n-propylphenol 3,4,5-trimethylphenol

progr II (ppm) 348 ( 20 357 ( 26 262 ( 23 226 ( 12 69 ( 7 5(1 150 ( 11 204 ( 20 44 ( 4 265 ( 22 319 ( 26 23 ( 1 116 ( 10d 368 ( 23 (coele) [160 ( 9] [212 ( 15] 2(0 (coel) 371 ( 30 686 ( 51 8(1 36 ( 2

average (ppm)

NIST24 (ppm)

HPLC-ED11 (ppm)

direct (ppm) n ) 3

extracn (ppm) n ) 1

336 ( 20 345 ( 23 255 ( 17 220 ( 12 65 ( 5 5(1 150 ( 11 198 ( 15 42 ( 3 264 ( 13 308 ( 22 21 ( 3 102 ( 6 [