Direct coupling of a liquid chromatograph to a continuous flow


High Performance Liquid Chromatography in Environmental Analysis: Present and Future Applications. Badar I. Afghan , Aaron W. Wolkoff. Journal of Liqu...
0 downloads 0 Views 558KB Size


1135

Ana/. Chem. 1980. 52. 1135-1140

CONCLUSIONS High precision, accurate metal isotope ratios can be obtained using volatile metal chelates. Alkaline earth metals are best studied using electron ionization, since under chemical ionization polymeric ion species that are unsuitable for isotope ratio studies predominate in t h e spectrum. Chemical ionization mass spectrometry is most useful for transition metal studies; i t provides simpler spectra with intense pseudomolecular ions. The ability to measure metal isotopic abundances rapidly, accurately, and with high precision using a conventional mass spectrometer will provide biomedical scientists with a powerful tool for exploring trace metal metabolism.

LITERATURE CITED Prasad, A. S.; Oberleas, D. "Trace Elements in Human Health and Disease"; Academic Press: New York. 1976; Volume 2 , Chapters 26-46. Moore, L. J.; Rosman, K. J. R. Abstracts of the 1978 Conference of the Federation of Analytical Chemistry and Spectroscopy Societies, Boston, Mass., Oct. 30-Nov. 3, 1978. No. 219. Rabinowitz, M. B.; Wetherill, G. W.: Kopple, J. D. Science 1973, 782, 725-727. Schwartz, R.; Giesecke, C. C. Clin. Chim. Acta 1979, 97, 1-8. Donohue, D. L.; Carter, J. A,; Franklin, J. C. Anal. Lett 1977, 70, 371-379. Schulten, H. R . ; Lehmann, W. D. I n "Quantitative Mass Spectrometry in Life Sciences 11", de Leenheer, A. P.,Roncucci, R. R . Von Peteghem, C., Eds.; Elsevier Scientific Publishing Company: Amsterdam, 1978; pp 63-82. Tsuge. S.; Leary, J. J.; Isenhour, T. L. Anal. Chem. 1974, 46, 106-110. Frew, N. M.; Leary, J. J.; Isenhour, T. L. Anal. Chem. 1972, 44, 665-67 1,

Hui, K. S.; Davis, B. A,: Boulton, A. A . Neurochem Res. 1977, 2 , 495-506. Miller, D. D.: Van Campen. D. Fed. Am. SOC.f x p . Biol. Proc., 1978, 3 7 , 488. Hileman, F. D.; Athin, C. L.; Lee, G. R.; Smith, D. L.; Hughes, 6 .M. I n "Proceedings of the 26th Annual Conference of Mass Spectrometry and Allied Topics", 1978, 336-338. Veillon, C.; Wolf, W. R.; Guthrie, 6 . li. Anal. Chem. 1979, 51, 1022-1024. Uden, P. C.; Henderson, D. E. Analyst(London), 1977. 102, 889-916. Klein, P. D.; Haumann, J. R.; Hachey, D. L. Clln. Chem. 1975, 21, 1253-1 257. Martell. A. E.; Belford, R . L.; Calvin, M. J . Inorg. Nucl. Chem. 1958, 5 , 170-181. Brauman, J. I.Anal. Chem. 1966, 38, 607-610. Beynon, J. H.; Williams, A. E. "Mass and Abundance lables for Use in Mass Spectrometry"; Eisevier Publishing Company: Amsterdam, 1963. Belcher, R.; Cranley, C. R.; Majer, J. R.; Stephen, W. I.; Uden, P. D. Anal. Chim. Acta 1972, 6 0 , 109-116. Joshi. K. C.; Pothak, V. N. Coord. Chem. Rev. 1977, 2 2 , 37-122. Prescott, S. R.; Campana, J. E.; Jurs, P. C.; Risby, T. t i , ; Yergey, A. L. Anal. Chem. 1976, 48, 829-832. Prescott, S. R.; Campana, J. E.; Risby, T. H. Anal. Chem. 1977, 49, 1501- 1504. Rozett, R . W. Anal. Chem. 1974, 4 6 , 2085-2089. Matthews, D. E . ; Hayes, J. M. Anal. Chem. 1976, 48, 1375-1382. McLaughiin, E.; Rozett, R. W. J . Organomet. Chem. 1973, 62, 261-268. Schoeller, D. A. Biomed. Mass Spectrom. 1978, 3 , 265-271

RECEIVED for review December 13, 1979. Accepted March 18, 1980. This work was supported by the U. S. Department of Energy under contract No. W-31-109-ENG-38. T h e authors wish to acknowledge financial support of the Institut d u Radium, Paris, France, given to J-C. Blais during his sabbatical visit to Argonne National Laboratory.

Direct Coupling of a Liquid Chromatograph to a Continuous Flow Hydrogen Nuclear Magnetic Resonance Detector for Analysis of Petroleum and Synthetic Fuels James F. H a w , T. E. Glass, D. W. Hausler, Edwin Motell,' and H. C. Dorn" Department of Chemistry, Virginia Polytechnic Institute, a n d State University, Blacksburg, Virginia 2406 1

Initial results obtained for a flow 'H nuclear magnetic resonance (NMR) detector directly coupled to a liquid chromatography unit are described. Results achieved for a model mixture and several jet fuel samples are discussed. Chromatographic separation of alkanes, alkylbenzenes, and substituted naphthalenes present in the jet fuel samples are easily identified with the 'H NMR detector. Results with our present flow 'H NMR insert indicate that 5-Hz linewidths are readily obtainable for typical chromatographic flow rates. The limitations and advantages of this liquid chromatography detector are compared with more commonly employed detectors (e.g., refractive index detectors).

T h e widespread applicability of high performance liquid chromatography (HPLC) for separation of complex mixtures is well recognized. Although a number of different detectors (e.g., refractive index, UV, etc.) are commonly employed in high performance liquid chromatography, most of these are nonselective for identification of discrete compounds. A possible exception is the Fourier Transform Infrared detector Visiting Professor, Department of Chemistry, San Francisco San Francisco, Calif.

State University.

0003-2700/80/0352-1135$01 .OO/O

for liquid chromatography ( I ) . Fourier transform nuclear magnetic resonance (FT-NMR) is a proven technique for spectral analysis of pure compounds a n d simple mixtures. In addition, the relatively high sensitivity of the 'H nuclide to detection by NMR is attractive for applications requiring limited sample and/or ''apailable time window" for spectroscopic examination. The latter requirements characterize the initial requirements for a continuous flow detector in high performance liquid chroinatography utilizing 'H NMR as the detector (LC-'HNMR). Nonchromatographic applications of flow NMR have previously been reported. Rapid irreversible chemical reactions as well as transient phenomena such as Chemically Induced Dynamic Nuclear Polarization (CIDNP) have been studied by stopped-flow NMR ( 2 ) . The theoretical formalism for the study of fast transient chemical reactions by FT-NMR has also been reported ( 3 ) . An apparatus for continuous-flow F'T-NMR has previously been described by Fyfe et al. (4). The effect of line broadening due to decreasing residence time with increasing flow rates was discussed. A value of T,(ossD, (observed spin-spin relaxation time) for varying flow rates is defined in t h e equation below (1/T2)(0BSD)

=

C 1980 American Chemical Society

(l/T?)IST-lT)

+

1136

ANALYTICAL CHEMISTRY, VOL. 52, NO. 7, JUNE 1980

'H

FLOW

NMR

INSERT TO

VACUUM

INLET

OUARTZ

REGION

TUBE

PYREX TUBING

5 m m OD SAMPLE

REGION 0

Flgure 1. Diagram of LC-" NMR insert. The sample is polarized in region A (within magnetic field) and free induction decays are monitored in region B

A

S

3C

60

3 :

20

I53

ISC

213

ELUTION VOLUME ( m l ) Figure 2. Refractive index trace of model mixture

Figure Y .

LL- H NMK

prorile Tor moaei mixture. tacn rile consisIs

T h e spin-spin relaxation time (T21STAT)) is the value for static samples and T is the residence time in the receiver coil (related to flow rate). An analogous expression can be written for t h e observed value for spin-lattice relaxation times. Another obvious consideration in LC-NMR is solvent selection, but this is also an important consideration with virtually every LC detector. Freon 113 (1,1,2-trifluorotrichloroethane), chloroform-d,, and carbon disulfide are possible solvent candidates which span a fairly wide range of polarity. Although perhaps too expensive to use neat (even with recycle), methanol-& could serve as a polar modifier a t a level of several percent. All separations involving deuterated solvents can be optimized with the cheaper protonated analogues prior to L,C-'HNMR analysis. Chemical shift references such as tetramethylsilane (TMS) may be added directly t o t h e mobile phase with negligible effect upon elution characteristics so long as the reference does not strongly interact with the stationary phase. Similarly, quantitative references may also be added. T h e extension of this technique to analysis of fluorinated derivatives by I9FF T - N M R is not restricted by solvent considerations. In general, the solvents that are suitable for LC-'HNMR have reasonable spectral windows for LC-FT-IR. T h e two techniques could easily provide complementary information. Obviously, one of the major advantages of the LC-'HNMR approach is the savings in sample manipulation and analysis time. This is particularly valid when compared with the tedious process of individual chromatographic fraction collection, solvent evaporation, and preparation for static NMR or other spectroscopic examination. Recently, a liquid chromatography effluent stream coupled to a 'H NMR spectrometer has been described for static examination of chromatographic fractions ( 5 ) . We believe that our work is the first description of a continuous flow LC-'HNMR system. In this paper, we report the design of our LC-'HNMR insert and present the results for a typical model mixture to demonstrate the advantages of this technique. A simple mixture of four different hydrocarbons was first examined. In addition, four experimental military jet fuels were studied, since they

OT i I U

scans (u.4-sacquisiriuri

iiriie,

L R Udld pUlIll5J UI

-03

PGI

IIIG

ANALYTICAL CHEMISTRY, VOL. 52, NO. 7, JUNE 1980

CH3

1137

CH3

CH3-C-CHp - CH-CH3 CH3

IO

8

6

4

2

0

PPM

IO

8

6 PPM

4

2

0

Figure 4. Selected spectra of each of the four model compounds near maximum model concentration on the LC-'H NMR protile

-

5

220

-1 a

80

-

70

-

P

z

605 0 -

40

-

>

30

-

a

20

-

IO

-

W

I=

-1

0 -

5 IO I5 2 0 2 5 3 0 3 5 4 0

Figure 5. Plot

FILE NUMBER of proton integrals vs. file number (protonogram) for

model mixture represent relatively complex mixtures amenable t o t h e LC' H N M R approach.

EXPERIMENTAL The jet fuel samples were supplied by the Air Force Aero Propulsion Laboratory (Wright-Patterson Air Force Base, Ohio) and the Naval Research Laboratory (Washington, D.C.). The four jet fuels used in this study were: routine JP-4, modified JP-4 (sample modified by adding xylenes and naphthalenes), Shale Paraho JP-5, and Coal Coed process JP-5. The last two samples are experimental fuels derived from shale and coal, respectively. The model mixture sample was prepared by mixing: 1.00 g isooctane, 1.00 g 2-hexene, 1.25 g benzene, and 1.00 g naphthalene. Trichlorotrifluoroethane (Miller-Stephenson Chemical Co.) was degassed prior to use as the chromatographic solvent. Enough tetramethylsilane (TMS)was added to the solvent after degassing to make it - 0 . 3 7 ~ by volume. A Merck Silica Gel 60 Size B (310 mm x 25 mm i.d.1 column was used after solvent equilibration. The pump was a Waters M-45 solvent delivery system. A Valco injection valve with a 500-rL sample loop was used throughout. All samples were injected neat. A Laboratory Data Control Model 1107 refractive index (RI) detector was used t o obtain classical chromatograms. The NMR flow cell was connected directly to the outlet of the

RI detector via a length of Teflon tubing (id. 1 mm). The time delay between the RI detector and the NMR receiver coil was determined to be 40 s at a flow rate of 2.5 mL/min. Effluent was removed from the NMR flow cell through a Teflon tube and into a cold trap by a vacuum pump. A Jeolco PS-100 nuclear magnetic resonance spectrometer was used to obtain 'H spectra at 100.0 MHz for the flon studies. The spectrometer was used in the Fourier Transform (FT) mode with a Digilab Data System. A fixed head 128K Alpha Data Disc allowed sufficient data storage for 42 (2048 point) spectral files and required a minimum of 110 ms for each spectral file transfer from the computer (Nova 1200). Figure 1 is a schematic diagram of the 'H NMR insert used in these LC-'H NMR studies. The receiver coil is wound about a 5-mm NMR tube which is connected at the bottom to the 1.5-mm Pyrex tubing which introduces the effluent stream. The spectrometer was operated with an external proton lock system. The comparison 'H NMR spectra of the jet fuels were obtained at 90.0 MHz in the CW mode utilizing a Varian EM-390 spectrometer. Hexamethyldisiloxane was used as the reference and lock signal.

RESULTS AND DISCUSSION T h e four components of the model mixture have a n elution order of isooctane, 2-hexene, benzene, and naphthalene, respectively, utilizing the chromatographic conditions described in t h e Experimental section. T h e normal Refractive Index (RI) trace obtained for this mixture is presented in Figure 2. Good resolution is indicated in spite of relatively high sample loading. This is also indicated in Figure 3 which is the LC-'H N M R profile obtained for this chromatographic run. Spectra with no peaks other than t h e reference TMS have been omitted. T h e growth and decay of compounds as one examines progressively later spectra is readily apparent. Figure 4 shows spectra selected from near the maxima of t h e four chromatographic peaks. Resolution arid sensitivity may be more readily evaluated in this figure t h a n in the stacked plot. Although line widths of 6-7 Hz were typically obtained, line widths of 5 Hz were the best t h a t could be obtained with our present insert. Since Fyfe ( 4 ) was able to d o considerably better with similar conditions, a considerable improvement in resolution is certainly feasible with improved receiver insert design. Given our 5-Hz limitation on magnetic inhomogeneity, a n acquisition time of 0.4 s was selected as a compromise between sensitivity and further degradation of line widths. Doubling the flow rate (2.5 t o 5 mL/min) did not substantially degrade resolution so the line width con-

ANALYTICAL CHEMISTRY, VOL. 52, NO. 7, JUNE 1980

1138

\

36

35

: ;

t l , ; 10

; : : <

I‘:

6

8

4

2

f i : : 10

0

8

<

: < fi: 4

6

PPM

2

(

0

PPM

Figure 6. LC-’H NMR profile for Paraho Shale Jet Fuel JP-5 (spectral conditions same as indicated in Figure 3) 0

7.6 0

7.2

X

6.4 n-C8HI8

IO

9

8

7

5

6

4

3

2

I

Figure 7. ’H FT NMR Spectrum for total Paraho Shale JP-5 mixture at 90 MHz. This sample contains tris(acetylacetonato)chromium(III) for quantitative measurements (resolution intentionally degraded for comparative purposes with LC-’H NMR profile) (Figure 6)

tribution from residence time was minor under these conditions. Since each file spectrum is collected under identical conditions, integrals from a given spectrum should be directly comparable with those obtained from other spectral files within the same chromatographic d a t a set. This is apparent in t h e relatively constant amplitude for t h e TMS peak in Figure 3 from file to file. Ideally, one could compute all integrals relative t o a quantitative standard (something less volatile t h a n T M S such as hexamethyldisiloxane would be preferable) present in the mobile phase. In any event, each spectrum in a set may be integrated and plotted vs. elution volume t o give a proton response chromatogram, a “protonogram”. T h e protonogram for t h e model mixture is presented in Figure 5 . This plot was constructed by integrating all peaks (exclusive of TMS) in each spectrum and plotting t h e total vs. file (spectrum) number. T h e protonogram shows resolution t h a t is very similar t o t h a t in the RI trace, demonstrating t h a t under these conditions t h e dead volume between t h e R I detector a n d t h e N M R receiver coil is negligible. I t should be noted that the sharp isooctane peak is underdetermined by the data points, whereas an excessive number of spectra have been taken across t h e later eluting, broader naphthalene peak. Ideally, spectra should be acquired very frequently early in the chromatogram (where higher concentration permits fewer transients to be taken t o obtain a signal-averaged spectrum for a given file). Spectra should

1 1

6.0 5.6

O

PPM

1 0

5.2 n-C7H‘6

4.8

n-C6HI4

4.4 4.0

t

t

0 I

I

1

1

5

6

7

8

FILE

1

1

I

91011

I

12

NUMBER

Figure 8. Correlation between average n-alkane chain length vs. file number for aliphatic peak of Paraho Shale Jet Fuel JP-5. The value x indicates the number of methylene units (-CH,-) in a linear alkane CH,(-CH,-),CH,

be taken infrequently across later eluting peaks where collecting more transients becomes important. Since peak width is (ideally) related linearly t o elution volume, the number of transients per spectrum could be increased. If one integrates a protonogram peak and corrects for t h e number of protons per molecule, a molar response is obtained. Based on t h e known composition of t h e model mixture, the area of t h e benzene peak should be 1.54 times t h a t of the naphthalene peak. I t should also be noted t h a t chromatographic peak widths for benzene a n d naphthalene are approximately equivalent (Figure 5 ) . A value of 1.50 was obtained from the protonogram. Given the line widths with the present instrumentation and the normal proton chemical shift range, there are approximately 150 spectral channels of information presently available. In principle, protonograms could be constructed for each channel, producing “chromatograms” very specific for certain structural features (e.g., aromatic hydrogen).

ANALYTICAL CHEMISTRY, VOL. 52, NO. 7, JUNE 1980

I

10

I

I

8

I

I

1

1

6

1

I

4

I

2

0

e:

IO

6

PPM

4

2

0

PPM

Figure 9. LC-'H NMR profile for Coal Coed Jet Fuel JP-5 (spectral conditions same as indicated in Figure

3)

I

IO

8

4

6

PPM

1139

2

0

1

0

8

6

4

2

File No

38

0

PPM

Figure 10. LC-'H NMR profile for Jet Fuel JP-4 (spectral conditions same as indicated in Figure 3)

Jet Fuels. The LC-'H NMR profile for the Paraho shale jet fuel (JP-5) is presented in Figure 6. T h e spectrum for t h e original jet fuel mixture yields considerably less information (Figure 7) with respect to the individual compounds present. T h e spectra for the alkanes (files 5-12) do not indicate t h e presence of branched hydrocarbons (no methine shoulder -1.5 ppm). I t is, therefore, reasonable to conclude t h a t all of the aliphatic compounds in this sample are linear alkanes. If one assumes that each compound eluting under this envelope has two methyl groups, the number of methylene groups may be calculated from t h e ratio of t h e methyl and methylene peak integrals. T h e resulting trend is presented in Figure 8. Files 'ithrough 12 suggest a steady progression from n-decane t o n-hexane. Elution volumes of standard samples run under identical conditions confirm this order. The methyl to methylene ratios for files 5 and 6 are anomolous but could be rationalized as frontal phenomena due to column overloading. T h e aromatic envelope in the Paraho shale oil sample contains 26% of t h e material by volume (ANSI/ASTM D 1319-77). The various compounds in this envelope elute over a broad range so the concentrations are relatively low. T h e signal-to-noise ratio for spectra of t h e aromatic envelope is,

therefore, considerably less than that for the aliphatic envelope. Several generalizations can be made regarding t h e compounds present in this portion of the profile. Simple alkylbenzenes with short alkyl groups predominate early in t h e peak with a progression through the xylenes toward mesitylene with possibly some tetra- and pentamethylbenzene (files 34-37). Again with coal COED JP-5, one can gather little information from t h e NMR spectrum prior t o separation. Figure 9 is t h e LC-'H NMR profile for this fuel. This coal-derived material contains highly branched aliphatic compounds. This is indicated by the high methyl/methylene peak ratio and obvious methine shoulder (-1.5 ppm). This high degree of branching is in sharp contrast to the linear alkane content found in the Paraho shale oil derived fuel. Files taken early on the aromatic peak of coal COED JP-5 show a preponderance of simple alkylbenzenes, xylenes, and Dossiblv mesitvlene. As more of the aromatic rbeak elutes. a progreision tobard more highly methylated benzenes and possibly tetralin is observed. T h e LC-'H NMR profile for the routine petroleum derived jet fuel (JP-4) is presented in Figure 10. For comparison, the LC-'H N M R profile for the modified (modified by addition

1140

ANALYTICAL CHEMISTRY, VOL. 52, NO. 7, JUNE 1980



File No

41

40

I

1

!

/

0

:

I

,

0

:

:

:

(

6 PPM



:

4

:

r

{

2

0

10

,

I

0

I

I

t

6

r

4

I

( , ’ f 2

0

PPM

Figure 11. LC-’H NMR profile for modified Jet Fuel JP-4 (spectral conditions same as indicated in Figure 3)

of aromatics, e.g., xylenes and naphthalene derivatives) JP-4 sample is presented in Figure 11. T h e spectra for the alkane region files 6-11, (Figures 10 a n d 11) of these fuels are very similar with increasing alkyl branching indicated a t later elution files (9-11). T h e aromatic region for t h e routine JP-4 (Figure 10) indicates the trend from simple alkyl aromatics (possibly ethyl a n d propyl) t o a preponderance of xylenes. By comparison, t h e modified JP-4 (Figure 11) clearly suggests t h e presence of 1- a n d 2-methylnaphthalene in the latter fractions. T h e RI trace also indicates a large peak at an elution volume typical for naphthalene.

CONCLUSION I t is clear from the LC-’H NMR results presented for the four different jet fuels that significant differences in sample composition are readily discernible by this approach. T h a t is, the LC-’H NMR technique provides unique “fingerprints” for each jet fuel. For example, the clear differentiation of branched alkane from linear alkane content is readily apparent in t h e LC-’H NMR results for these fuels whereas the RI detector indicates only one peak. Although greatly limited by sensitivity relative t o most LC detectors, flow LC-NMR provides a wealth of structural information quickly and conveniently. In cases where high resolution and sensitivity are necessary, flow I C - N M R could still be used to screen effluents t o identify fractions for conventional NMR analysis. In the extension of this technique t o chromatographic columns of higher efficiency, lower sample capacity will tend t o reduce sensitivity, but LC-’H NMR detection cells matched to smaller sample volume would minimize peak broadening. A second and perhaps more severe limitation of t h e LC-

N M R approach is the limited choice of chromatographic solvents. However, t h e use of fluorinated and deuterated solvents is a n obvious method of handling this limitation. In addition, saturation a n d / o r double resonance NMR techniques could be useful when hydrogen containing solvents cannot be avoided.

ACKNOWLEDGMENT We express our appreciation t o Edwige Denyszyn for technical assistance with figure and sample preparation. We are also indebted t o L. T. Taylor for helpful discussions on t h e chromatographic portion of this work. In addition, we wish to thank Robert Hazlett (KRL). Major Don Potter (USAF), and Ronald Butler (USAF) for providing the jet fuel samples and helpful discussions regarding properties of these fuels.

LITERATURE C I T E D (1) Vidrine, D. W.; Mattson, D. R. Appl. Spectrosc. 1978. 32, 502-506. (2) Grimaldi, J. J.; Baldo, J.; McMurray. C.; Sykes. B. J . Am. Chem. SOC. 1972. 94. 7641-7645. (3) Kuhne, R . ; Schaffhauser, T.; Wokaun, A . ; Ernst. R. J . Magn. Reson. 1979, 35, 39-67. (41 Fyfe, C. A , ; Cocivera, M.; Damji, S. W. H.; Hostetter, T. A , ; Sproat, D.; O’Brien, J. J . Magn. Reson. 1976, 23, 377-384. (5) Watanabe, N.; Niki, E. R o c . Jpn. Acad. 1978, 54, 194-199.

RECEIVED for review November 30, 1979. Accepted February 19, 1980. We gratefully acknowledge generous financial support for this work provided by the Naval Research Laboratory (Washington, D.C.) and the U.S. Air Force (WrightPatterson Air Force Base, Dayton, Ohio).