17O Quantitative Nuclear Magnetic Resonance Spectroscopy of

reference material, at the levels currently used in retail gasolines (1-20%). In addition, the simultaneous detec- tion and quantification of the oxyg...
1 downloads 0 Views 101KB Size
Anal. Chem. 2003, 75, 4659-4666

17

O Quantitative Nuclear Magnetic Resonance Spectroscopy of Gasoline and Oxygenated Additives

David G. Lonnon* and James M. Hook*

NMR Facility and School of Chemical Sciences, University of New South Wales, Sydney, NSW, Australia 2052

17O

Nuclear magnetic resonance (NMR) spectroscopy allows exclusive detection and direct quantification of oxygenates in gasoline unaffected by its hydrocarbon content, using the internal standard quantitative NMR (QNMR) method. Chemical shifts of 24 oxygen-containing compounds as potential additives and contaminants have been measured in gasoline and corrected values of δO 18.1 and 3.9 determined for neat methyl tert-butyl ether (MTBE) and neat di-n-butyl ether, respectively. Quantification of ethanol in gasoline can be readily achieved by 17O QNMR with dimethyl sulfone as an internal standard reference material, at the levels currently used in retail gasolines (1-20%). In addition, the simultaneous detection and quantification of the oxygenates methanol, ethanol, 2-propanol, tert-butyl alcohol, and MTBE in gasoline has been established to further demonstrate the specificity of the method. 17O NMR has distinct advantages over 1H and 13C QNMR methods, and although it cannot reliably differentiate 1-propanol, 1-butanol, 1-pentanol, and isopentyl alcohol, 17O NMR does allow the rapid and unambiguous identification of unexpected oxygenates such as acetates and ketones found as contaminants in some retail gasoline. Oxygen-containing organic compounds, commonly known as oxygenates, are routinely added to motor gasoline to act as fuel extenders and octane improvers. Currently the most widely used oxygenates are aliphatic alcohols and methyl ethers, which contain one to six carbons, such as methanol, ethanol, and methyl tertbutyl ether (MTBE). Due to the increasingly widespread addition of oxygenates, as well as the rise in legislation relating to fuel additives and emissions standards, a method for their quantification in gasoline that is robust and routine while also being precise and accurate is in demand. A number of methods have been reported for oxygenate analysis in gasoline, most of which are based on gas chromatography (GC)1,2 or on spectroscopic techniques such as IR,3-5 near* Corresponding authors. E-mail: [email protected]. d.lonnon@ student.unsw.edu.au. (1) Diehl, J. W.; Finkbeiner, J. W.; Di Sanzo, F. P. J. High Resolut. Chromatogr. 1995, 18, 108-110. Lockwood, A. F.; Caddock, B. D. Chromatographia 1983, 17, 65-69. Alary, J.; Coeur, M. A. Bull. Trav. Soc. Pharm. Lyon 1971, 15, 13-22. Sevcik, J. J. High Resolut. Chromatogr. Chromatogr. Commun. 1980, 3, 166-168. Durand, J. P.; Petroff, N. Rev. Inst. Fr. Pet. 1982, 37, 575-578. 10.1021/ac034339q CCC: $25.00 Published on Web 07/18/2003

© 2003 American Chemical Society

IR,6 Raman,7 microwave,2,8,9 ultraviolet visible,10 and mass spectrometry.9,11 Methods that rely on GC have the limitation that oxygenates tend to coelute with the hydrocarbon constituents of gasoline, although several ways of overcoming this problem have been developed such as solvent extraction,12 multidimensional capillary column GC,13,14 and the use of selective detectors such as microwave-induced plasma atomic absorption,2 oxygen flame ionization detector,14 FT-IR,15 and oxygen-specific mass spectrometry.9 Generally these methods are accurate and reliable; however, they are often tedious and time-consuming, and while many of the spectroscopic methods are nondestructive, they are often hampered by the problem of signal overlap; for instance, methanol and ethanol give overlapping IR signals.4 Many spectroscopic methods also require extensive calibration to be carried out by the use of multivariate, statistical calibration techniques.5 Nuclear magnetic resonance (NMR) spectroscopy using 1H and 13C nuclei has also been effective for identifying and quantifying oxygenates in gasoline.16-21 In the 1H NMR spectra of gasoline, (2) Goode, S. R.; Thomas, C. L. J. Anal. At. Spectrom. 1994, 9, 73-78. (3) Battiste, D. R.; Fry, S. E.; White, F. T.; Scoggins, M. W.; McWilliams, T. B. Anal. Chem. 1981, 53, 1096-1099. Bessler, E. Cienc. Cult. 1977, 29, 928930. Fodor, G. E.; Kohl, K. B.; Mason, R. L. Anal. Chem. 1996, 68, 23-30. Lopez-Anreus, E,; Garrigues, S.; de la Guardia, M. Anal. Chim. Acta 1996, 333. 157-165. (4) Fry, S. E.; Fuller, M. P.; White, F. T.; Battiste, D. R. Anal. Chem. 1983, 55, 407-408. (5) Garcia, F. X.; De Lima, L.; Medina, J. C. Appl. Spectrosc. 1993, 47, 10361039. (6) Wong, J. L.; Jaselskis, B. Analyst 1982, 107, 1282-1285. (7) Choquette, S. J.; Chesler, S. N.; Duewer, D. L.; Wang, S.; O’Haver, T. C. Anal. Chem. 1996, 68, 3525-3533. (8) Aleksandrov, A. N.; Tysovskii, G. I. Neftepererabotka i Neftekhim 1966, 3739. (9) Orlando, R.; Munson, B. Anal. Chem. 1986, 58, 2788-2791. (10) Hubert, C.; Fichou, D.; Valat, P.; Garnier, F.; Villeret, B. Polymer 1995, 36, 2663-2666. (11) Shofstahl, J. H.; Hardy, J. K. Anal. Chem. 1986, 58, 2412-2414. Franke, G. Erdoel Kohle 1961, 14, 816-820. (12) Pauls, R. E.; McCoy, R. W. J. Chromatogr. Sci. 1981, 19, 558-561. (13) American Society for Testing Materials ASTM D 4815, 1990, 05.03. Luke, L; A., Ray, J. E. Analyst 1984, 109, 989-992. (14) Di Sanzo, F. P. J. Chromatogr. Sci. 1990, 28, 73-75. (15) Diehl, J. W.; Finkbeiner, J. W.; DiSanzo, F. P. Anal. Chem. 1992, 64, 32023205. (16) Louis, R. Erdoel Kohle 1966, 19, 281-287. Steinmetzer, H. C.; Baumeister, W. GIT Fachz. Lab. 1980, 24, 95-98, 102, Meusinger, R.; Moros, R. Fuel 2001, 80, 613-621. Skloss, T. W.; Kim, A. J.; Haw, J. F. Anal. Chem. 1994, 66, 536-542. (17) Renzoni, G. E.; Shankland, E. G.; Gaines, J. A.; Callis, J. B. Anal. Chem. 1985, 57, 2864-2867. (18) Kalsi, W. R.; Sarpal, A. S.; Jain, S. K.; Srivastava, S. P.; Bhatnagar, A. K. Energy Fuels 1995, 9, 574-579.

Analytical Chemistry, Vol. 75, No. 17, September 1, 2003 4659

the region from 3.0 to 4.6 ppm is typically free of peaks17 and is fortuitously the region in which peaks are observed for the protons attached to the R-carbons of alcohols. Renzoni et al.17 used this fact to identify and quantify alcohols in gasoline based on the construction of a calibration curve and achieved limits of detection for methanol of ∼0.007 vol % and for other C2-C5 alcohols of between 0.37 and 1.7 vol %. Subsequently, Sarpal et al.18 observed that such calibration curves are sensitive to field stability, sample temperature, and spectrometer frequency and so fresh curves may need to be constructed each time these variables alter. To avoid these shortcomings, Sarpal et al.18 employed a method based on 1H NMR and group molecular weight calculations and achieved limits of detection for some C1-C4 alcohols, MTBE, and tert-amyl methyl ether (TAME) of between 0.1 and 0.71 mol %. Sarpal’s group subsequently used 13C NMR20 in place of 1H NMR in order to minimize the unavoidable peak overlap that occurs in 1H NMR spectra of samples containing multiple oxygenates; however, acquisition of each spectrum is somewhat lengthy at greater than 4 h. The need for calibration curves can be avoided by the use of quantitative NMR (QNMR) with an internal standard reference material, which is either added directly to the analyte solution19,21-24 or added as a coaxial insert (the “insert method”).25,26 The QNMR internal standard method is a well-established primary ratio analytical method, widely used in organic chemical analysis.22-24 Meusinger has successfully employed this method for the analysis of oxygenates in gasoline19,21 using 1H NMR and dimethyl oxalate as the internal standard, with standard errors of estimate for analysis of some C1-C4 alcohols, MTBE, and TAME of between 0 and 0.4.21 However, as the internal standard method works most effectively with nonoverlapping peaks, the use of an NMR nucleus such as 17O that generally shows less peak overlap than 1H NMR would seem advantageous. Previously 17O NMR has been used for the determination of oxygenates in synthetic fuels;27 however, due to the limitations of the technology at the time, only rough percentage estimates were made that did not differentiate individual components. 17O NMR has also been employed by Alam and Alam28 in an attempt to quantify primary alcohol mixtures using multivariate analysis but without an internal standard. With this in mind, we report that the direct observation of the oxygen nuclei of oxygenates in gasoline with 17O NMR forms the basis of quantification of these oxygenates by QNMR with the (19) Meusinger, R. Fuel 1996, 75, 1235-1243. (20) Sarpal, A. S.; Kapur, G. S.; Mukherjee, S.; Jain, S. K. Energy Fuels 1997, 11, 662-667. (21) Meusinger, R. Anal. Chim. Acta 1999, 391, 277-288. (22) Saed Al Deen, T.; Brynn Hibbert, D.; Hook, J. M.; Wells, R. J. Anal. Chim. Acta 2002, 474, 125-135. Holzgrabe, U.; Diehl, B. W. K.; Wawer, I. J. Pharm. Biomed. Anal. 1998, 17, 557-616. Rabenstein, D. L.; Keire, D. A. Practical Spectrosc. 1991, 11, 323-369. Rackham, D. M. Talanta 1976, 23, 269-274. (23) Wells, R. J.; Hook, J. M.; Al-Deen, T. S.; Hibbert, D. B. J. Agric. Food Chem. 2002, 50, 3366-3374. (24) Maniara, G.; Rajamoorthi, K.; Rajan, S.; Stockton, G. W. Anal. Chem. 1998, 70, 4921-4928 and references therein. (25) Larive, C. K.; Jayawickrama, D.; Orfi, L. Appl. Spectrosc. 1997, 51, 15311536. (26) Henderson, T. J. Anal. Chem. 2002, 74, 191-198. (27) Grandy, D. W.; Petrakis, L.; Young, D. C.; Gates, B. C. Nature 1984, 308, 175-177. Grandy, D. W.; Petrakis, L.; Young, D. C.; Gates, B. C. NATO ASI Ser. C 1984, 124, 689-698. Petrakis, L., Allen, D., Eds.; NMR for Liquid Fossil Fuels; Elsevier Science: New York, 1987. (28) Alam, M. K., Alam, T. M. Spectrochim. Acta, Part A 2000, 56A, 729-738.

4660

Analytical Chemistry, Vol. 75, No. 17, September 1, 2003

internal standard method. This combination of techniques appears to have a number of distinct advantages. First, most short-chain (C1-C6) alcohols and ethers have well-separated 17O NMR peaks meaning that most oxygenates can be separately observed. Second, despite the low natural abundance of 17O (0.037%) and broad peaks encountered due to the quadrupolar nature of this nucleus (I ) 5/2), spin-lattice relaxation times (T1) are of the order of a few milliseconds and thus allow relatively rapid data acquisition. Third, as only the oxygen-containing compounds within blended gasoline are observed by 17O NMR, there is no interference from the hydrocarbon peaks and so the quantification of oxygenates is independent of gasoline composition. Last, the internal standard method allows the quantification of most oxygenates in a single measurement, thus avoiding the timeconsuming procedure of constructing calibration curves for each component being quantified. Both methods of QNMR analysis using an internal standard and calibration curves have been evaluated in this work. MATERIALS AND METHODS Chemicals. Dimethyl sulfone (DMSO2; 98%), 1,3,5-trioxane (99+%), 1-propanol, (98+%), di-n-butyl ether (99%), MTBE (98%), sec-butyl acetate (technical grade), TAME (97%), and deuterium oxide (99.9 atom % D) were used as supplied by Sigma-Aldrich Chemical Co. (Sydney, Australia). DMSO2 was determined to be 99.95% pure by differential scanning calorimetry (DSC) and gas chromatographic flame ionization detection (GCFID). 1,3,5-Trioxane was determined to be 99.9% pure by GCFID. Deuteriochloroform (99.8 atom % D) was a product of Cambridge Isotopes Laboratory (Andover, MA). The following reagents were used as supplied: 2-propanol (analytical reagent, AR; Asia Pacific Specialty Chemicals Ltd., Seven Hills, Australia); 1-butanol (May and Baker Ltd., Dagenham, England); 2-butanol (BioScientific Pty Ltd., Gymea, Australia); 2-methyl-2-propanol (tert-butyl alcohol) (laboratory reagent, LR), ethyl methyl ketone (LR), and acetone (AR) (Ajax, Auburn, Australia); 3-pentanol (Eastman Kodak Co.); 1-pentanol (LR), isopentyl alcohol (LR), and n-butyl acetate (LR) (British Drug House Ltd., Poole, England); ethyl acetate (AR) (Rhone-Poulenc Chemicals Pty Ltd.); di-n-pentyl ether and n-pentyl methyl ketone (Hopkins and Williams Ltd., Essex, England); tertbutyl acetate (LR) (Fluka). Ethanol (reagent grade) and methanol (99.8%), both from Aldrich, were distilled from magnesium turnings activated with iodine under a dinitrogen atmosphere immediately prior to use. Water was double distilled. Unleaded gasoline samples were used as supplied by the University Analytical Laboratories and are referred to as ULG (unleaded gasoline). All samples were measured in either 10-mm-o.d. 5135PP or 5-mm-o.d. 507pp, Wilmad tubes. Instrumentation. 1H, 13C, 13C DEPT135,29 1H COSY (correlation spectroscopy),29 1H TOCSY (total correlation spectroscopy),29 13C-1H HMQC(heteronuclear multiple quantum correlation spectroscopy),29 and 13C-1H HMBC (heteronuclear multiple bond correlation spectroscopy)29 NMR spectra were acquired with a Bruker Avance 300 NMR spectrometer operating at 300.13 MHz for 1H and at 75.49 MHz for13C, using a 5-mm probe. All data were processed using Bruker’s XWINNMR software. The following parameters were employed for acquisition of 1H NMR spectra: (29) MIT-NMR-Facility. MIT Department of Chemistry website, 2003.

spectral width, 22 ppm; acquisition time, 4.95 s; relaxation delay, 25 s; 90° pulse width, 7.3 µs; time domain, 64K complex points; 20 scans; temperature, 298 K. An exponential line-broadening window function of 0.5 Hz was used in the data processing. After Fourier transformation of the free induction decays, the spectra were phased, baseline corrected, and integrated in the appropriate region. The following parameters were employed for acquisition of 13C NMR spectra: spectral width, 80 ppm; acquisition time, 1.37 s; relaxation delay, 2.0 s; 30° pulse width, 11.50 µs; time domain, 64K complex points; 640 scans; temperature, 298 K. An exponential line-broadening window function of 1 Hz was used in the data processing. The following data for MBTE were obtained: 1H NMR (CDCl3) δ 3.20 (3H, OCH3), 1.19 (9H, CCH3) (this work), 3.2, and 1.3 ppm;30 13C NMR (CDCl3) δ 71.97 (1C, CCH3), 48.70 (1C, OCH3), 26.40 (3C, CCH3) (this work); 72, 49, and 26 ppm.30 17O NMR spectra were acquired on both Bruker Avance 300and 500-NMR spectrometers, operating at 40.69 and 67.80 MHz, respectively, using 10-mm broadband probes. All data were processed using Bruker’s XWINNMR software. Typically spectra were obtained unlocked with the following parameters: spectral width, 150 ppm; acquisition time, 37.8 ms; relaxation delay, 10 ms; 90° pulse width, 20 µs; time domain, 1K complex points; between 10 000 (identification, ∼7.5 min), 20 000 (calibration curve, ∼15 min), and 40 000 (internal standard-based quantification, ∼30 min) scans; temperature, 330 K. The inversion recovery method31 was used to determined spin-lattice relaxation times, T1, values of all 17O nuclei which were found to be less than 9 ms, and so typically spectra were acquired with a recycle time of 48.7 ms or ∼5T1. An exponential line-broadening window function of 50 Hz was used in the data processing, except for the spectra used in construction of the calibration curve, for which no line broadening was used. After Fourier transformation of the free induction decays, the spectra were phased, automatically baseline corrected, and integrated in the appropriate region. All spectra were integrated twice and the mean of integrated areas used in plotting the calibration curve and internal standard-based quantification calculations. The 17O chemical shifts were reproducible to at least (1 ppm and in order to reduce line widths were obtained at 57 °C. The shifts were referenced relative to H2O at 27 °C, δO(H2O) ) 0.0 ppm. Samples and standards were weighed on a Sartorious electronic balance to 0.01 mg. Internal Standard Method. Ethanol/ULG Independent Replicates. An ethanol/ULG stock solution was prepared by weighing ethanol (2.9554 g) directly into a known mass of an oxygenatefree sample of ULG (29.7229 g). Each sample was prepared by accurately weighing DMSO2 (∼30 mg) directly into a 10-mm NMR tube, adding CDCl3 (∼200 mg) to dissolve the DMSO2, and then adding an aliquot of stock solution (∼2.0 g), the mass of which was accurately determined. Each tube was sealed with Parafilm, weighed, immediately placed within the NMR magnet, and allowed to stand for 15 min in order to thermally equilibrate at 57 °C. Seven samples were prepared in total. Reweighing each tube after removal from the magnet confirmed that no measurable mass of solution had evaporated at the elevated temperatures of the spectrometer. (30) Sigma-Aldrich. Sigma-Aldrich website, 2003. (31) Sanders, J. K. M.; Hunter, B. K. Modern NMR Spectroscopy: A Guide for Chemists; Oxford University Press: Oxford, U.K., 1990; pp 61-65.

Oxygenate Mixture Replicates. Each sample was prepared by accurately weighing DMSO2 (∼30 mg) directly into a 10-mm NMR tube adding CDCl3 (∼200 mg) to dissolve the DMSO2 and then adding ethanol (∼80 mg), the weight of which was then accurately determined. Methanol, 2-propanol, tert-butyl alcohol, and MTBE were then added, with the weight of each oxygenate determined before the addition of the next oxygenate. Two separate samples were prepared, and both were treated as per the independent replicates. Repeatability. A single sample was prepared by weighing DMSO2 (52.5 mg) directly into a 10-mm NMR tube followed by CDCl3 (∼200 mg) to dissolve the DMSO2. A sample of ULG (2.3637 g) that contained an unknown percentage of ethanol was then weighed directly into the tube. Samples were then treated as per the independent replicates. Seven 17O NMR spectra were acquired at both 40.69 and 67.80 MHz. Calibration Curve Method. Nine separate samples were prepared by directly weighing an oxygenate-free sample of ULG (∼2.0 g) into a 10-mm NMR tube, into which ethanol was then directly weighed. The weight of ethanol added to each sample was such that ethanol/ULG (w/w) percentages of 1.4, 2.0, 4.7, 7.4, 12.1, 17.2, 25.1, 34.9, and 43.9% were obtained. Each sample was then treated as per the independent replicates. An 17O NMR spectrum of each sample was acquired at both 40.69 and 67.80 MHz, using acquisition parameters described above. RESULTS AND DISCUSSION Identification of Oxygenates. The quality of 17O NMR spectra typically obtained are displayed in Figure 1 showing 6 typical oxygenates in ULG while their 17O chemical shifts values, and those of 15 others, are collected in Table 1. The 17O chemical shift values in Table 1 were obtained from spectra acquired with solutions containing 5-10% (w/w) of each oxygenate in ULG, and other than those of methanol, di-n-butyl ether, and MTBE are generally in accord with previously reported values. The discrepancy between the methanol chemical shift values, δO(methanol)this work ) -33.3 and δO(methanol)32 ) -37.0, may be due to differences in the temperature and concentration at which the two samples were run. However, the discrepancy between the di-n-butyl ether values, δO(di-n-butyl ether)this work ) 3.9 and δO(di-n-butyl ether)33 ) -7.0, and the MTBE values, δO(MTBE)this work ) 17.7 and δO(MTBE)33 ) 8.5, seems hard to reconcile. In fact, when pure di-n-butyl ether and MTBE were measured separately at 25 °C, the chemical shift values are 3.8 and 18.1 ppm, respectively. This suggests that the initially reported values are incorrect and that the subsequent reports of these values34,35 have perpetuated the error. The identity of the MTBE used in this current work was confirmed by 1H and 13C NMR (see Materials and Methods). Table 1 also shows the 17O NMR peak line width (width of peak at half-height) of each oxygenate, as well as previously reported 17O chemical shift values. As expected from the quadrupolar nature of the 17O nucleus,28 the line width values follow the same general trend as the viscosity of the neat oxygenates,36 with (32) Crandall, J. K.; Centeno, M. A. J. Org. Chem. 1979, 44, 1183-1184. (33) Delseth, C.; Kintzinger, J. P. Helv. Chim. Acta 1978, 61, 1327-1334. (34) Beraldin, M. T.; Vauthier, E.; Fliszar, S. Can. J. Chem. 1982, 60, 106-110. (35) Martins, M. A. P.; Zanatta, N.; Bonacorso, H. G.; Siqueira, G. M.; Flores, A. F. C. Magn. Reson. Chem. 1999, 37, 852-855.

Analytical Chemistry, Vol. 75, No. 17, September 1, 2003

4661

Table 1. 17O NMR Chemical Shift Data of Water, Oxygenates, and Internal Standards in Gasoline δOa

sample

ω/2

δO (lit.)

ref

water

0.2b (-6.6)

116

0.0f

32

methanol ethanol 1-propanol 2-propanol 1-butanol tert-butyl alcohol 2-butanol 1-pentanol 3-pentanol isopentyl alcohol

Alcohols -33.3 (-31.8) 5.6 (7.1) -1.0 (-0.1) 36.9 (38.3) -0.5 (0.8) 60.2 -4.4 -0.2 22.3 -0.8

85 96 168 158 151 181 167 198 195 215

-37.0f 5.9f -0.5f 39.8f 0.0f 62.3f -2.0f -0.8f 22.8f -0.6f

32 32 32 32 32 32 32 32 32 32

di-n-butyl ether MTBE TAME di-n-pentyl ether

Ethers 3.9c 17.7d (21.4) 13.2e -3.3

93 87 105 205

-7.0g 8.5g

33 33

ethyl acetate

Acetates 364.6, 167.5

72, 72 363, 169h 353.3, 136.7i 363, 196h 348.3, 134.7i 93, 97 97, 96 91, 87 375, 207h 368.7, 204.4j 347.6, 131.1i

45 46 45 46

isopropyl acetate n-butyl acetate sec-butyl acetate tert-butyl acetate

364.9, 163.0 362.8, 189.0 376.3, 201.1

acetone

580.3

56

ethyl methyl ketone

566.8

65

45 47 46

Ketones

Figure 1. The 67.80-MHz 17O NMR spectra of various alcohols and MTBE prepared by the sequential addition of ethanol (containing water), methanol, MTBE, 1-butanol, 2-propanol, and tert-butyl alcohol to ULG at 57 °C.

n-pentyl methyl ketone 571.3 dimethyl sulfone 1,3,5-trioxane

569.0g 571i 558i 557.5g

34, 48 49 49 34, 48

163k 65l

43 50

136

Internal Standards 164.5 155 63.5 167

a Measured as 5-10% solutions in ULG at 57 °C. Numbers in parentheses are δO of sample 1 as shown in Figure 2. b Measured neat at 57 °C. c δO ) 4.7 when measured neat at 25 °C. d δO ) 18.1 when measured neat at 25 °C. e δO ) 11.1 when measured neat at 25 °C. f Measured neat at 65 °C. g Measured at 25 °C and although not stated presumed to be neat. h Measured neat at 34-36 °C. i Measured in acetonitrile at 75 °C. j Measured in CDCl3 (4 M) at 25 °C. k Measured in chloroform at 25 °C. l Measured at 34-36 °C.

an increase in line width correlating with an increase in viscosity. This is understood by considering that as the viscosity of the solution increases, molecular tumbling decreases and spin-spin relaxation times (T2) shorten, which in turn results in broader line widths.28 Once the 17O chemical shifts of various oxygenates in samples of gasoline had been established, it was then possible to determine relatively quickly (∼10 min) the presence of these oxygenates in samples of commercial ULG. The resolving power of 17O NMR is generally very good, although due to the broad line widths and small chemical shift differences (