calculate the volume at each point on_theGPC chromatogram.The sample molecular weights (an, M,) and molecular weight distribution ai-e then computed from the detector response/volume curve (6, 7) and the corresponding appropriate calibration curve ( 5 , 8 ) . Operator precision in sample injection and siphon counting is important. The injection must be made at a siphon dump to ensure that the correct volume is recorded. In our system the electronic counter and “GO” button for computer collection are ganged together for ease of operation. The sample used was NBS Polystyrene 706, M,, = 137 000 and Mu = 288 000. This sample was employed as the standard for calibration and was also used as the “unknown sample” in calculating molecular weights which were caused to vary by the experimentally perturbed flow rates. The NBS polystyrene was chromatographed at three different flow rates. This set of three experiments was then subjected to twelve separate MW calculations. In the top half of Table I, .“With Siphon Compensation”,the indicated flow rate was utilized. In the section of the Table designated “Without Siphon Compensation”, a flow rate of 2.0 ml/min was always used.
SUMMARY AND CONCLUSIONS I t was predicted ( I ) that gross molecular weight errors would result from poor pump performance, particularly with poor precision in pump resettability. These predictions are aptly verified by the data shown in Table I where uncompensated flow-rate variations of 5-10% lead to molecular weight errors of up to several hundred percent. Even these gross molecular weight errors from the large flow-rate variations are nearly completely eliminated by our siphon compensation method. Much smaller changes in flow rate which
may be encountered in practice also lead to significant errors ( I ) and are also compensated by this method, but become more difficult to illustrate experimentally. While not studied, the method is expected to be useful for compensating most MW errors caused by drift or changing flow rates occurring during an experiment.
LITERATURE CITED (1)D. D. Bly, H. J. Stoklosa, J. J. Kirkland, and W. W. Yau, Anal. Chem., 47, 1810-13 (1975). (2)D. D. Bly, H. J. Stoklosa, J. J. Kirkland, and W. W Yau, Anal. Chem., 47,2328 (1975). (3)W. W. Yau, H. L. Suchan, and C. P. Malone, J. Polym. Sci., Part A-2, 6, 1349-1355 (1968). (4) G. N. Patel, J. Appl. Polym. Sci., 18,3537-3542 (1974). (5) S.T. Balke, A. E. Hamielec, B. P. LeClair, and S. L. Pearce, lnd. Eng. Chem., Prod. Res. Dev., 8,54-7 (1969). (6) H. E. Pickett, M. J. R. Cantow, and J. F. Johnson, J. folym. Sci., PartC, 21, 67-81 (1968). (7)H. E. Pickett. M. J. R. Cantow, and J. F.Johnson, J. Polym. Sci., Part C, 21, 67-81 (1968). (8)T. D. Swartz, D. D. Bly, and A. S. Edwards, J. Appl. Polym. Sci., 16, 3353-3360 (1972).
D. D. Bly* W. W. Yau H. J. Stoklosa
E. I. du Pont de Nemours and Company Central Research and Development Department Wilmington, Del. 19898 RECEIVEDfor review February 12, 1976. Accepted April 6, 1976.
Petroleum Identification by Laser Raman Spectroscopy Sir: The fingerprinting of oil spills and the subsequent identification of the polluter is now carried out routinely by the U S . Coast Guard ( 1 ) .In most oil spill situations, it is desirable to have more than one analytical technique for matching a spilled oil to its suspected source. This was demonstrated in the recent spill by the Liberian Tanker Garhis which dumped -50 000 gallons of Light Arabian Crude Oil along the Florida Coast. Four different techniques, infrared spectroscopy, gas chromatography, fluorescence, and low temperature luminescence, were used to identify the Garbis as the spill source ( I ) . By using a combination of fingerprinting methods, the Coast Guard had a much stronger case against the Garhis than if they had only an infrared spectrum or a gas chromatogram as evidence. For the purpose of this multianalytical approach, we have investigated the possibility of using laser-Raman spectroscopy as an additional tool for the fingerprinting of oil spills (see Ref. 2-5 for details on fingerprinting petroleum by infrared spectroscopy). Laser-Raman is being used in more and more laboratories as a standard analytical instrument (6).However, as a tool in petroleum research, its role has been hindered by the high fluorescent background. This background is of such high intensity that it completely obscures any bands due to Raman scattering. We have tried several methods to eliminate this fluorescence background. We tried using column chromatography to remove the fluorescing compounds. An alumina over silica column was used to separate the oil into an aliphatic and aromatic fraction, and the fluorescence free aliphatic fraction was then analyzed by Raman. This method suffered from the disadvantages of requiring a new column for each oil, large amounts of solvent, and time consumed in constructing the column, eluting the oil, and evaporating off the solvent. Clearly, it was a very cumbersome method for routine analysis.
Theoretically, a longer wavelength of light for the laser exciting line should decrease sample fluorescence. Going from a green (5600 8,) to a red (6300 8,) exciting line using a tunable dye laser reduced the fluorescence, but not nearly enough. In addition, by using a longer exciting wavelength line, the Raman effect also decreased as did the output power of the laser. Thus, it was necessary to use high amplification which resulted in a very noisy background on top of the remaining fluorescence. The method which did work well was also the simplest. We found that the addition of powdered charcoal to a dilute solution of the oil removed the fluorescent compounds. Two ml of oil were added to 4 g of coconut charcoal and diluted with 4 ml of pentane (the solvent decreased the oil’s viscosity to form a solution which could be easily stirred and poured). The solution was stirred for 2 min and then filtered through Celulite Filter Aid to separate the pentane extract from the charcoal. The pentane was then allowed to evaporate at room temperature in an exhaust hood. (In addition to removing the solvent, this procedure removes the lower boiling hydrocarbons and serves to minimize differences in the degrees of weathering between spill and suspect spill source samples.) Whereas the oils before treatment were highly colored, they were now colorless or slightly yellow. The treated oil was then placed in a capillary tube and allowed to sit in the laser beam for 10 min to eliminate any remaining fluorescence before recording its spectrum (the mechanism for this reduction in fluorescence is not well understood, but the method is often used to reduce fluorescence). All the spectra were measured on a SPEX Model 1401 Double Spectrometer (spectral slit-width