Anal. Chem. 2003, 75, 6394-6400
Electrospray-Mass Spectrometric Analysis of Reference Carboxylic Acids and Athabasca Oil Sands Naphthenic Acids Chun Chi Lo,† Brian G. Brownlee,‡ and Nigel J. Bunce*,†
Department of Chemistry and Biochemistry, University of Guelph, Guelph, Ontario, Canada N1G 2W1, and National Water Research Institute, Environment Canada, Burlington, Ontario, Canada L7R 4A6
Naphthenic acids (NAs) are complex mixtures of naturally occurring acyclic and cyclic aliphatic carboxylic acids that are responsible for the toxicity of the water in the tailings ponds associated with the recovery of bitumen from the Athabasca oil sands. NAs are difficult to analyze due to their complexity and the lack of commercially available NA standards. This paper describes the use of negative ion electrospray ionization mass spectrometry for the analysis of NAs. Model carboxylic acids, alone and in mixture, afforded mass spectral signal intensities that were highly dependent on extractor and cone voltages and on molecular structure. These effects were also observed for authentic NAs. Under conditions that were close to optimal for all the model compounds, their calibration sensitivities varied by a factor of 25% of Canada’s present oil production, a proportion that is expected to rise to >50% in 5 years.3 At present, about 0.1-0.2 m3 of contaminated * Corresponding author. Telephone: 519-824-4120, Ext 53962. Fax: 519-7661499. E-mail:
[email protected]. † University of Guelph. ‡ Environment Canada. (1) Department of Energy, Government of Alberta Indroduction to Oil Sands; 2001. http://www.energy.gov.ab.ca/com/Sands/Introduction/Oil+Sands.htm (accessed 06/2003). (2) FTFC (Fine Tailings Fundamentals Consortium); Advances in oil sands tailings research; Alberta Department of Energy, Oil Sands and Research Division, 1995. (3) Herman, D. C.; Fedorak, P. M.; MacKinnon, M. D.; Costerton, J. W. Can. J. Microbiol. 1994, 40, 467-77.
6394 Analytical Chemistry, Vol. 75, No. 23, December 1, 2003
water slurry (tailings) is accumulated for each tonne of oil sands processed.2 Environmental regulations require the tailings to be held in large tailing ponds to prevent release to the wider environment due to concerns about their toxicity.4 The tailings, whose total volume is expected to increase from the current level of 4 × 108 m3 to >109 m3 by the year 20205 comprise a slurry of sand, clay, water, and unrecovered bitumen in a stable aqueous suspension; they are characterized by alkaline pH and salinity, and the presence of organic contaminants such as polycyclic aromatic hydrocarbons and naphthenic acids (NAs).2 Naphthenic acids, which are normal constituents of crude oil,6 are released and concentrated from the oil sands in the alkaline hot water extraction process for bitumen. Their concentrations can be as high as 100 mg/L in the tailing pond surface water.7 NAs are complex mixtures of alkyl-substituted acyclic and cycloaliphatic carboxylic acids with the general formula CnH2n+ZO2, where n indicates the carbon number and Z is the hydrogen deficiency.4 For example, monocyclic, bicyclic, and tricyclic NAs belong to the Z ) -2, -4, and -6 families, respectively. Examples of the types of molecular structures thought to comprise naphthenic acids are shown in Figure 1, but because the mixtures are so complex, individual components have never been isolated. The surfactant properties of NAs are responsible for the acute toxicity of the tailings toward fish and amphibians, but no information is available concerning structure-toxicity relationships.3 Extensive fragmentation makes electron impact mass spectrometry (EI-MS) unsuitable for analysis of native NAs; high polarity and low volatility rule out the use of gas chromatography/ mass spectrometry (GC/MS),8,9 although success has been achieved with GC/EI-MS analysis of their tert-butyldimethylsilyl derivatives.7-10 No liquid chromatographic method has yet been reported for separating the components of NA mixtures, so (4) Lai, J. W. S.; Pinto, L. J.; Kiehlmann, E.; Bendell-Young, L. I.; Moore, M. M. Environ. Toxicol. Chem. 1996, 15, 1482-91. (5) Chalaturnyk, R. J.; Scott, J. D.; Ozum, B. Pet. Sci. Technol. 2002, 20, 102546. (6) Brient, J. A.; Wessner, P. J.; Doyle, M. N. In Encyclopedia of Chemical Technology, 4 ed.; Kirk-Othmer, ed.; Wiley: New York, 1996; Chapter, Naphthenic Acids. (7) Holowenko, F. M.; MacKinnon, M. D.; Fedorak, P. M. Water Res. 2002, 36, 2843-55. (8) Holowenko, F. M.; MacKinnon, M. D.; Fedorak, P. M. Water Res. 2001, 35, 2595-606. (9) Holowenko, F. M.; Fedorak, P. M. Evaluation of a gas chromatographyelectron impact mass spectrometry methods for characterizing naphthenic acids. Final Report. June 25, 2001, Syncrude Contract E3297. 10.1021/ac030093d CCC: $25.00
© 2003 American Chemical Society Published on Web 10/31/2003
Figure 1. Proposed structures of naphthenic acids in the Z ) 0, -2, -4, and -6 families, with both five- and six-carbon rings present. Table 1. Base Peaks That Are within the Definition of Naphthenic Acids between Masses of 100 and 600
n (C no.)
0
5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
101 115 129 143 157 171 185 199 213 227 241 255 269 283 297 311 325 339 353
-2
127 141 155 169 183 197 211 225 239 253 267 281 295 309 323 337 351
Z (no. of hydrogen loss due to ring structure) -4 -6 -8
167 181 195 209 223 237 251 265 279 293 307 321 335 349
193 207 221 235 249 263 277 291 305 319 333 347
219 233 247 261 275 289 303 317 331 345
-10
245 259 273 287 301 315 329 343
-12
271 285 299 313 327 341
“whole” NA mixtures must be analyzed by MS under nonfragmenting conditions. Fluoride ion negative ion chemical ionization mass spectrometry11 and negative ion fast atom bombardment mass spectrometry12 have been explored, as have soft ionization techniques such as electrospray ionization mass spectrometry (ESI-MS)13 and atmospheric pressure chemical ionization mass (10) St. John, W. P.; Green, S. A.; McGinnis, G. D. J. Chromatogr., A 1998, 807, 241-51. (11) Dzidic, I.; Somerville, A. C.; Raia, J. C.; Hart, H. V. Anal. Chem. 1988, 60, 1318-23. (12) Fan, T. P. Energy Fuels 1991, 5, 371-75. (13) Headley, J. V.; Peru, K. M.; McMartin, D. W.; Winkler, M. J.AOAC Int. 2002, 85, 182-87.
n (C no.) 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41
0
-2
367 381 395 409 423 437 451 465 479 493 507 521 535 549 563 577 591
365 379 393 407 421 435 449 463 477 491 505 519 533 547 561 575 589
Z (no. of hydrogen loss due to ring structure) -4 -6 -8 363 377 391 405 419 433 447 461 475 489 503 517 531 545 559 573 587
361 375 389 403 417 431 445 459 473 487 501 515 529 543 557 571 585 599
359 373 387 401 415 429 443 457 471 485 499 513 527 541 555 569 583 597
-10
-12
357 371 385 399 413 427 441 455 469 483 497 511 525 539 553 567 581 595
355 369 383 397 411 425 439 453 467 481 495 509 523 537 551 565 579 593
spectrometry.14 Even without fragmentation, NAs give unusually complex spectra, with significant intensity at every mass number, through a combination of homologues (spaced every 14 mass numbers), different Z values for the same n (spaced every two mass numbers), and 13C isotope peaks. These mass spectra afford congener profiles based on the relative intensities of the (pseudo)molecular ion peaks, with certain assumptions. Contributions from (M + 2) ions (two 13C atoms) are counted in with the next “Z” family, and any family Z ) -14 (7 rings), would overlap with the next “n” value if a low-resolution mass spectrometer were used. (14) Hsu, C. S.; Dechert, G. J.; Robbins, W. K.; Fukuda, E. K. Energy Fuels 2000, 14, 217-23.
Analytical Chemistry, Vol. 75, No. 23, December 1, 2003
6395
Figure 2. Mass spectrum of the 100 µM carboxylic acid mixture in ESI negative mode. The individual carboxylic acids each have a concentration of 16.7 µM. The sample is infused into 0.2 mL/min carrier solvent mixture prior to entrance into the ESI probe. m/z ) 102.14 is one of the base peaks with the addition of isopropylamine. Table 2. From the Six Equal Molar Carboxylic Acid Mixture Analysis, the Maximum Linear Range of Total Carboxylic Concentration of 100 µM, Which Corresponds to Individual Acid Concentrations of ∼16.7 µMa individual carboxylic acid analysis
cyclohexanecarboxylic acid (m/z ) 127) cyclohexanebutyric acid (m/z ) 169) decanoic acid (m/z ) 171) 1-pyrenebutyic acid (m/z ) 287) abietic acid (m/z ) 301) 5β-cholanic acid (m/z ) 359)
slope
peak intensity
45759 69063 54925 64436 68510 90417
8.40 × 1.13 × 106 8.38 × 105 1.07 × 106 1.04 × 106 1.41 × 106 105
mixture of six carboxylic acids analysis R2 1.0000 0.9998 0.9988 0.9983 0.9971 0.9984
slope
peak intensity
R2
63710 96812 76499 99052 90170 158687
1.13 × 1.62 × 106 1.29 × 106 1.65 × 106 1.50 × 106 2.64 × 106
0.9997 0.9999 0.9999 0.9983 0.9988 0.9991
106
a The slope of best-fit lines and the corresponding signal intensities of carboxylic acids at concentration 16.7 µM by individual carboxylic acid analysis and the mixture of six carboxylic acids analysis were compared. The slopes and interpolated intensities were generated by the Microsoft Excel program
Whether all congeners ionize with equal efficiency or whether different congeners interfere with each others’ ionization is not known. Finally, quantitative data cannot be obtained due to lack of standard solutions of naphthenic acids with known congener concentrations. Commercial NA mixtures originating from conventional crude oil refining are available, but their congener distributions and concentrations vary from lot to lot. In this work, we examine how the apparent congener composition changes with concentration and instrument settings, for both model NA carboxylic acids and an authentic sample of pond water acid fraction from the Mildred Lake tailing pond owned by Syncrude Canada Inc. Model compounds were studied both alone and in mixture, to examine interactive effects. The methodology 6396
Analytical Chemistry, Vol. 75, No. 23, December 1, 2003
involved negative ion ESI-MS without chromatographic preseparation. MATERIALS AND METHODS Chemicals. A commercial NA mixture was purchased from Fluka (Oakville, ON, Canada). Cyclohexanecarboxylic acid (98%), cyclohexanebutyric acid (99%), decanoic acid (99+%), 1-pyrenebutyric acid (97%), and abietic acid (70%) were purchased from Aldrich (Mississauga, ON, Canada), and 5β-cholanic acid (99%) was from Sigma (Mississauga, ON, Canada). Isopropylamine (99.5+%) was purchased from Aldrich. Although 1-pyrenebutyric acid and abietic acid were not included in the definition of NA, these acids were used to assess the signal variation of aromatic and alkene structure carboxylic acids under the electrospray
Figure 3. Total ion current spectrum obtained from the ESI-MS analysis of 125 µg/mL TPWNA under optimized voltage setting of cone voltage ) 25 V and extractor voltage ) 8 V. The m/z 102.14 peak was removed in order to give a representative naphthenic acids congener spectrum.
condition. HPLC-grade dichloromethane (DCM) and 2-propanol were purchased from Fisher Scientific (Toronto, ON, Canada). Water was purified with a Milli-Q UV plus purification system (Bedford, MA). Naphthenic Acid Extraction. Tailings pond water-derived naphthenic acids (TPWNA) were obtained by liquid-liquid extraction from fine tails pore water from the Syncrude tailing pond at Mildred Lake. The sample was adjusted to pH 12 with NaOH, and the basic and neutral organic compounds were removed by extraction with DCM. The aqueous solution pH was then adjusted to ∼2 by dropwise addition of 12 M H2SO4. This precipitated the majority of the NAs, whose removal was completed by 4-fold extraction with 50 mL of DCM. The DCM extracts and the precipitate were combined, filtered through anhydrous Na2SO4 into a round-bottom flask, and evaporated to dryness. Sample Preparation. A 1:1 v/v solution of 2-propanol and water containing 1% isopropylamine was degassed by ultrasonication for 20 min before use as the solvent mixture for sample preparation or as carrier phase for ESI-MS. Isopropylamine was present to enhance the ionization efficiency and increase the instrument response. Solutions of each of the six carboxylic acids and TPWNA (10 mM and 1 mg/mL, respectively) were prepared by diluting stock solutions prepared by mass, along with a solution containing all six carboxylic acids at equal molar concentration and a total carboxylic acid concentration of 10 mM. All samples were further diluted to working concentrations and stored in amber vials with Teflon-lined caps at room temperature in the dark. ESI-MS Analysis. Mass spectrometric analysis was performed by negative ion ESI, using a Waters single quadrupole
ZQ-LC/MS 2000 system equipped with a “pneumatically assisted” ESI probe and a Z-spray source. Solvent was delivered by a Waters 600 gradient pump system with a Waters 600 controller (Waters, Milford, MA/ Micromass, Manchester, U.K.). A Compaq Workstation with Masslynx v.3.5 software on Microsoft Windows NT platform was used for instrumental control and data acquisition. Mass spectral analysis of carboxylic acids was carried out in ESI negative mode with the capillary tip of the probe set to 3.2 kV to ionize the sample droplet. The distance between the electrospray capillary tip and the sampling orifice was 4 mm. Nitrogen (delivered by a Parker Balston model 75-72 nitrogen generator) was used as the heated desolvation gas (flow rate 300 L/min, desolvation temperature 220 °C) to assist in droplet formation and ionization and as cone gas (100 L/min) to prevent contamination on the ion source interface. The source temperature was set at 100 °C to avoid recondensation of sample droplets inside the ion source. The solvent mixture was used as carrier phase and was pumped continuously into the ESI probe at a rate of 0.2 mL/min. The carboxylic acids and the authentic NA sample were injected into the probe by the “flow infusion method” without chromatographic separation at a rate of 20 µL/min, using an on-board syringe pump and a Hamilton 500-µL syringe. The scan speed was set to 1 s/scan with 0.1-s interscan delay. Total ion current (TIC) spectra were obtained by integrating the first 40 scans in each acquisition. Due to signal intensity fluctuation, the instrument was calibrated daily with 100 µM cyclohexanecarboxylic acid before analysis, with the signal intensity of the peak adjusted to 4.6 × 106 by varying the multiplier voltage. All analyses were Analytical Chemistry, Vol. 75, No. 23, December 1, 2003
6397
Figure 4. Three-dimensional column graph of blank-corrected percentage intensity of congeners with carbon numbers (n) and Z family (Z) in TPWNA at a concentration of 125 µg/mL under (a) cone voltage 25 V, extractor voltage 8 V; (b) cone voltage 5 V, extractor voltage 5 V; and (c) cone voltage 45 V, extractor voltage 30 V.
carried out at least in duplicate. The MS was scanned from m/z 100 to 600, and peak heights were taken as the metric for concentration. The NA congeners in this mass range were identified using the modified m/z table developed by St. John et al.10 and edited by Holowenko et al.7 (Table 1). This shows 209 expected base peaks for NAs with 5 < n < 41 and 0 > Z > -12. RESULTS AND DISCUSSION Linear Range of Detection. Preliminary experiments using the mixture of six carboxylic acids and the commercial naphthenic acids mixture suggested a cone voltage of 25 V and extractor voltage of 8 V. Cone voltages of >30 V and extractor voltages of >10 V gave extensive collision-induced dissociation (CID), which was undesirable, because the analysis of the data depended on 6398 Analytical Chemistry, Vol. 75, No. 23, December 1, 2003
the presence of quasi-molecular ions only. Figure 2 shows the spectrum generated from 100 µM (total) carboxylic acid mixture. Despite considerable differences in molecular structure, the responses, on a molar basis, were similar (within ∼50%) and minimal CID was observed. Linearity of response was studied over the total concentration range from 5 µM to 1 mM for both individual carboxylic acids and their mixture; significant departure from linearity was seen when the total concentrations of carboxylic acids exceeded 100 µM (data not shown). Calibration sensitivities within the linear range varied by a factor of 75% of the blank-corrected signal when all Z families were aggregated (Table 3). Our observations were qualitatively similar to those obtained by Holowenko et al.7 by GC/EI-MS analysis of tert-butyldimethylsilyl derivatives of NA extracted from the oil sands; they found that ∑(Z) had a maximum at n ) 16. However, their results indicated a smaller proportion of congeners having n > 19 compared to our study. The differences may result from the decrease in ionization efficiencies or derivatization efficiencies in congeners with large molecular weight by GC/MS. The distributions from the Suncor site yielded higher proportions of “heavy” congeners and a tendency to double maximums within a “Z” family at C16 and C 25.7 In our experiments, the distribution of congeners within a “Z” family usually peaked near C15 (Figure 4a); e.g., Z ) 0 family
Figure 5. Standard addition curve of TPWNA spiked with cyclohexanecarboxylic acid.
peaked at C17; Z ) -4 and Z ) -6 families peaked at C14 - C15, but the Z ) -2 family showed a broader maximum at C13-C15. The Z ) -8 to Z ) -12 families had a much lower percentage TIC but still showed maximum ion current in the range C15-C18. However, as it is impossible for a C15 compound to have four or more medium-sized rings, it is likely that aromatic rings or double bonds are present. The hypothesis that NA mixtures might include aromatics has been suggested previously by Seifert and Teeter15 and Hsu et al.14 However, the difficulty of carrying out a clean extractive separation of surfactant NAs from basic and neutral aromatic compounds present in the oil sands makes it difficult to be certain whether NAs might contain aromatic constituents, and we are currently exploring this possibility further. In the present analytical context, an aromatic ring would contribute a -6 to Z, suggesting three more rings than would actually be present. Quantification by Standard Addition. The results with model compounds suggested that different NA congeners do not interfere with each other in terms of relative response in negative ion ESI-MS. However, the signal enhancement effect with mixtures makes it difficult to obtain the absolute concentration of a particular congener. Because the calibration sensitivities among different compounds varied less than 2-fold, we made a semiquantitative analysis using the standard addition method. Authentic pond water NA samples of concentration 125 mg/mL were spiked with various concentrations of cyclohexanebutyric acid such that the intensity of the peak at m/z 169 was increased by a factor in the range 2-20 to give a standard addition curve (Figure 5). The concentration of the highest intensity congener (m/z ) 223) in the 125 µg/mL authentic pond water NA sample was calculated as 28.7 µM. This suggests that if the concentration of NA in the tailing pond is 100 mg/L, the concentrations of individual congeners may be as high as 23.0 µM. Within the m/z range of our analyses, the average concentration was 4.4 µM for congeners with n e 20 and 0.27 µM for congeners with 21 e n e 41. Sensitivity of the Analysis to Variation of Cone and Extractor Voltage. The dependence of signal intensity on cone and extractor voltages was investigated for the model carboxylic acids in order to discover whether similar instrument settings would optimize the signal from the various structural types. Cone voltages were in the range 5-45 V and extractor voltages 5-30 (15) Seifert, W. K.; Teeter, R. M. Anal. Chem. 1970, 42, 750-58.
Analytical Chemistry, Vol. 75, No. 23, December 1, 2003
6399
Figure 6. Cyclohexanecarboxylic acid: signal intensities under various cone and extractor voltages.
V, with all other MS parameters the same as in the previous experiments. The model acids used in the analysis were as follows: decanoic acid, straight chain, Z ) 0; cyclohexanecarboxylic and cyclohexanebutyric acids, monocyclic, Z ) -2; abietic acid, tricyclic with alkene functionality, Z ) -10; 5β-cholanic acid and 1-pyrenebutyric acid, both tetracyclic, Z ) -8 and -24, respectively. The first three of these acids had signal intensities that were strongly dependent on cone and extractor voltages, maximizing when cone voltage was 20-25 V and extractor voltage was ∼5 V. The signal intensities at different cone voltages and extractor voltages for cyclohexanecarboxylic acid are shown in Figure 6. The signal intensities of the polycyclic abietic acid and 5β-cholanic acid both peaked when cone voltages were in the range 25-45 V, but the dependence of signal intensity on extractor voltage was less than in the preceding cases. Satisfactory signal intensities were obtained under a wider range of extractor voltages from 5 to 20 V, with peak at 10 V. 1-Pyrenebutyric acid gave maximum signal intensities at cone voltage 15-30 V and extractor voltage 5-10 V. However, fragmentation of the quasi-molecular ion was observed when cone voltages and extractor voltages exceeded 30 and 15 V, respectively. The significance of these observations is that instrument settings affect signal intensity in a structure-dependent manner. With our instrument, the best compromise was achieved for cone voltage near 20 V and extractor voltage
